text
stringlengths 254
1.16M
|
|---|
---
title: 'Association of bone mineral density with trichlorophenol: a population-based
study'
authors:
- Zijian Yan
- Xianmei Xiong
- Jiasheng Tao
- Sheng Wang
journal: BMC Musculoskeletal Disorders
year: 2023
pmcid: PMC10022061
doi: 10.1186/s12891-023-06323-y
license: CC BY 4.0
---
# Association of bone mineral density with trichlorophenol: a population-based study
## Abstract
### Background
Trichlorophenols (TCPs) are metabolites of several organochlorine chemicals, including chlorobenzene, hexachlorocyclohexane, and chlorophenoxy acid, present in air, surface water, soil, and sediment. Many studies have shown that endocrine disruptors (EDs)may contribute to decreased bone mass and the increased risk of osteoporosis. However, the relationship between TCP and bone mineral density (BMD) has not been studied yet.
### Methods
We conducted a cross-sectional study by using data from the 2005–2010 National Health and Nutrition Examination Survey (NHANES). TCP levels were measured in urine samples from 3385 participants and bone mineral density was obtained by dual X-ray absorptiometry (DXA) lumbar spine and femur scanning. Multiple regression analysis, stratified analysis, curve fitting analysis, and trend tests were used to assess the relationship between TCP and BMD.
### Result
After adjusting for confounding factors, the results of multiple regression analysis only showed that ln-2,4,5-TCP was negatively associated with BMD of lumbar spine. In stratified analyses, Male, Menstruating Female and Menopausal Female were divided into three groups for analysis. The results showed that ln-2,4,5-TCP and ln-2,4,6-TCP were not statistically associated with BMD in total femur, femoral neck, femoral tuberosity, intertrochanteric femur and lumbar spine, which was also confirmed by curve fitting analyses and trend tests.
### Conclusion
This study demonstrated that 2,4,5-TCP and 2,4,6-TCP in urine samples were not significantly associated with BMD in the US population. Therefore, 2,4,5-TCP and 2,4,6-TCP may not be detrimental to BMD.
## Introduction
The chlorophenols group comprises a total of five different types of chlorophenols, corresponding to their degree of chlorination: monochlorophenols (MCPs), dichlorophenols (DCPs), trichlorophenols (TCPs), tetrachlorophenols (TeCPs) and pentachlorophenols (PCP). TCPs including 2,4,5-trichlorophenol (2,4,5-TCP) and 2,4,6-trichlorophenol (2,4,6-TCP) are formed when three chlorine atoms are joined to one phenol molecule. TCPs have been widely used in herbicides, wood preservatives and fungicides, and as intermediates in various chemically synthesised pesticides, dyes and other products, and are widely present as environmental pollutants [1, 2]. TCPs are also metabolites of several organochlorine chemicals, including chlorobenzene, hexachlorocyclohexane, and chlorophenoxy acid, present in air, surface water, soil, and sediment. Therefore, the general population may be exposed to TCPs through ingestion of food and water or inhalation of air contaminated with TCPs or other organochlorine chemicals [3, 4]. Biomonitoring surveys of toxic substances are essential to determine average exposure levels of populations, to identify at-risk groups and to prevent further adverse effects [5]. National biomonitoring surveys for environmental contaminants such as TCP have been conducted in a number of countries, including the USA and some countries in Germany.
Osteoporosis is a common systemic bone disease worldwide that can lead to an increased risk of fracture [6]. Based on data from the United States Centers for Disease Control National Health and Nutrition Examination Survey (NHANES; 2005–2010), the National Osteoporosis Foundation estimates that more than 9.9 million Americans have osteoporosis and an additional 43.1 million have low bone mineral density (BMD) [7].Current treatment for osteoporosis is clinically well-established and the corresponding therapeutic effects are clear, with definite results obtained with bone-building drugs [8]. However, a clearer understanding of the risks that lead to osteoporosis and further prevention of osteoporosis could better improve quality of life and have important social benefits.
Epidemiological studies indicated that 2,4-dichlorophenol concentrations are associated with lower bone mineral density, higher osteopenia, and higher prevalence of osteoporosis in men [9]. However, the relationship between TCPs and bone mineral density has not been studied yet. Therefore, we used 2005–2010 National Health and Nutrition Examination Survey (NHANES) data to assess the association between urinary TCPs and BMD in the US population.
## Study pupulation
The NHANES is a major program of the National Center for Health Statistics (NCHS), which is part of the Centers for Disease Control and Prevention [10]. A representative sample of non-institutional US populations is selected through a complex stratified multi-stage sampling design. The NHANES programme was approved by the Ethical Review Board of the National Center for Health Statistics of the CDC, and written informed consent was provided to all participants during the survey [11, 12]. In the three cycles of 2005–2010, 10,537, 10,149 and 10,348 people participated in the NHANES program, a total of 31,034 people as shown in Fig. 1. Our analysis was applicable to participants older than 20 years, but only those who had urine TCP and BMD information tested were combined and used for the analysis. Therefore, we first excluded participants with missing TCPs data ($$n = 23$$,133). The participants with missing BMD information was then further excluded ($$n = 2437$$). Participants younger than 20 years of age were excluded ($$n = 2079$$). Finally, this study included 3385 participants, including 1737 males and 1648 females.
Fig. 1Flow chart of the screening process for the selection of eligible participants in NHANES.
## Assessment of BMD and T scores
The lumbar spine and femur scanning were acquired on Hologic Discovery model A densitometers, using software version Apex 3.2. The radiation exposure from DXA scans is extremely low at less than 20 uSv. All scans were analyzed with Hologic APEX version 4.0 software with NHANES BCA option. A high level of quality control was maintained throughout the DXA data collection and scan analysis, including a rigorous phantom scanning schedule. DXA scanning was used for eligible survey participants over the age of 8. Participants who meet one of the following criteria are prohibited from having a DXA scanning: [1] Pregnancy (positive urine pregnancy test and/or self-reported at the time of DXA examination [2]. Self-reported of radiographic contrast agent (barium) use within the past 7 days [3]. Self-reported weight over 450 pounds (209 kg) or height over 6’5” (1.96 m) [13]. The BMD of total femur, femoral neck, femoral tuberosity, intertrochanteric femur and lumbar spine were selected in this study [14]. L1-4 lumbar spine are used for lumbar spine BMD [15–17].
## Assessment of TCP
Information on urinary TCPs was obtained from the NHANES data [18]. Urine specimens are processed, stored, and shipped to the Division of Environmental Health Laboratory Sciences, National Center for Environmental Health, Centers for Disease Control and Prevention for analysis. Detailed specimen collection and processing instructions are discussed in the NHANES Laboratory/Medical Technologists Procedures Manual (LPM). According to the LPM [9], “urine samples were collected from participants with standard urine collection cups. Samples were refrigerated as soon as possible and transferred to specimen vials within 4 h of collection. At least 5 ml of urine were collected and stored frozen in borosilicate glass, polypropylene vials or specimen cups.“ Vials are stored under appropriate frozen (-20℃) conditions until they are shipped to National Center for Environmental Health for testing. A Sciex API 4000 mass spectrometer was used in negative ion APCI mode. The negative fragment ions were used for quantification of the urinary concentrations of 2,4,5-TCP and 2,4,6-TCP [19]. Creatinine-corrected urine concentrations of 2,4,5-TCP and 2,4,6-TCP were further calculated and analyzed.
## Covariates
In the NHANES database, there is a column for demographic data. In this column we collected information on the age, race, sex, body mass index (BMI), household income poverty ratio, education level, alcohol consumption, serum cotinine, milk intake, moderate physical activity, vigorous physical activity, serum calcium, serum phosphorus, serum lead and diabetes. Furthermore, we grouped women according to menstruation in 12 months, reasons for no menstruation, hysterectomy performed, and ever use female hormones as premenopausal women and postmenopausal women, respectively. Interviewer-administered questionnaires are used to record demographic information such as age, sex, race/ethnicity, household income poverty ratio, and educational attainment. Race self-reported as Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, Other. BMI was calculated by dividing body weight (in metric tons) by body height (in meters squared). Serum cotinine, calcium, lead and phosphorus were serum specimens from participants that were transported to a collaborating laboratory service for analysis. Whether participants had diabetes was considered based on diabetes self-report or fasting blood glucose or two-hour postprandial blood glucose. Vigorous physical activity was identified from the questionnaire (did you do any vigorous activities for at least 10 min that caused heavy sweating, or large increases in breathing or heart rate such as running, lap swimming, aerobics classes or fast bicycling). Moderate physical activity was determined from the questionnaire (did you do moderate activities for at least 10 min that cause only light sweating or a slight to moderate increase in breathing or heart rate such as brisk walking, bicycling for pleasure, golf, and dancing). Drinking milk was determined from the questionnaire (regular milk use 5 times per week).
## Statistical analysis
All statistical results were calculated based on NHANES sample weights. Model 1 did not adjust for any confounding factors. Model 2 was adjusted for sex, age, BMI and race. Model 3 also adjusts for sex, age, race, BMI, household income poverty ratio, education level, milk consumption, alcohol consumption, serum cotinine, serum lead, calcium concentration, phosphorus concentration, diabetes, moderate physical activity and vigorous physical activity. After adjusting for potential confounders, weighted multiple regression analysis was used to estimate the independent relationship between TCPs and BMD. Stratified analyses based on sex and menopausal state in women were then conducted to examine the association between TCPs and BMD at different sites. Finally, we proceed to the smoothing curve fitting and trend tests were performed to resolve the linear or curvilinear relationship in the subgroup analyses.
## Characteristics of the study population
The proportion of men in the study population was $51\%$ and the proportion of women was $49\%$. The mean age of the study population was 46.21 years for males and 46.78 years for females, and the mean BMI was 27.86 kg/m^2 for males and 28.00 kg/m^2 for females, with no statistically significant differences in BMI. Similarly, there were no significant differences between males and females in terms of education level. The average serum calcium and serum lead in males were higher than those in females, but the serum phosphorus was lower than that in females. The percentages of smoking, alcohol consumption and milk consumption were all statistically greater for males than females. There was no statistically significant difference between males and females in moderate physical activity. However, in vigorous physical activity, the percentage of males was significantly higher than that of females. There was no significant difference in the proportion of men and women with diabetes. In total femur, femoral neck, femoral tuberosity, intertrochanteric femur and lumbar spine, males had higher mean BMD than females (Table 1).
Table 1Characteristics of the study populationVariableMaleFemaleP-valueN17371648Age(years)46.21 ± 17.0146.78 ± 16.93< 0.01Race (%)0.23Mexican American351 (20.21)308 (18.69)Other Hispanic137 (7.89)156 (9.47)Non-Hispanic White838 (48.24)772 (46.84)Non-Hispanic Black326 (18.77)339 (20.57)Other Race85 (4.89)73 (4.43)BMI (kg/m^2)27.86 ± 4.9728.00 ± 6.090.44Education(%)0.63High School Grad or Below468 (26.94)432 (26.21)College Grad or Above1269 (73.06)1216 (73.79)Income poverty ratio(%)0.02≤ 0.99325 (18.71)360 (21.84)> 11412 (81.29)1288 (78.16)Serum calcium(mg/dL)9.49 ± 0.359.42 ± 0.37< 0.01Serum phosphorus(mg/dL)3.65 ± 0.573.84 ± 0.56< 0.01Serum lead(ug/dL)2.14 ± 2.141.46 ± 1.12< 0.01Serum cotinine(ng/mL)72.61 ± 139.9854.85 ± 124.15< 0.01Drink wine(%)< 0.01Yes1459 (84.00)1043 (63.29)No278 (16.00)605 (36.71)Drink milk(%)< 0.01Yes1383 (79.62)1206 (73.18)No354 (20.38)442 (26.82)Moderate physical activity(%)0.18Yes802 (46.17)723 (43.87)No935 (53.83)925 (56.13)Vigorous physical activity(%)< 0.01Yes598 (34.43)351 (21.30)No1139 (65.57)1297 (78.70)Diabetes(%)0.1Yes244 (14.05)200 (12.14)No1493 (85.95)1448 (87.86)Ln-2,4,5-TCP(ug/L)-7.19 ± 0.73-6.81 ± 0.80< 0.01Ln-2,4,6-TCP(ug/L)-5.59 ± 0.71-5.25 ± 0.75< 0.01BMDTotal femur(g/cm2)1.04 ± 0.150.92 ± 0.15< 0.01Femoral neck(g/cm2)0.88 ± 0.140.82 ± 0.15< 0.01Femoral tuberosity(g/cm2)0.78 ± 0.130.69 ± 0.12< 0.01Intertrochanteric femur(g/cm2)1.22 ± 0.171.09 ± 0.17< 0.01Lumbar spine(g/cm2)1.06 ± 0.141.01 ± 0.15< 0.01
## Multiple regression analysis
The results of the multiple regression analysis are shown in Table 2. In the unadjusted model, both ln-2,4,5-TCP and ln-2,4,6-TCP were negatively correlated with BMD of total femur, femoral neck, femoral tuberosity, intertrochanteric femur and lumbar spine ($P \leq 0.01$). After adjusting for sex, age, BMI and race confounders, there was no significant linear relationship between ln-2,4,5-TCP and ln-2,4,6-TCP and BMD of total femur, femoral neck, femoral tuberosity, intertrochanteric femur. However, ln-2,4,5-TCP were still negatively correlated with BMD of lumbar spine ($$P \leq 0.02$$). In Model 3, ln-2,4,5-TCP were still negatively correlated with BMD of lumbar spine ($$P \leq 0.04$$), and not significantly correlated with BMD in other parts.
Table 2Multiple regression analysis of TCP and BMDBMDModel 1 β ($95\%$ CI) P-valueModel 2 β ($95\%$ CI) P-valueModel 3 β ($95\%$ CI) P-valueTotal femurln-2,4,5-TCP-0.056 (-0.063, -0.050) < 0.01-0.003 (-0.009, 0.002) 0.26-0.003 (-0.009, 0.003) 0.27ln-2,4,6-TCP-0.049 (-0.056, -0.042) < 0.01-0.004 (-0.010, 0.001) 0.14-0.005 (-0.011, 0.001) 0.11Femoral neckln-2,4,5-TCP-0.053 (-0.059, -0.047) < 0.01-0.001 (-0.007, 0.004) 0.70-0.001 (-0.007, 0.004) 0.67ln-2,4,6-TCP-0.046 (-0.052, -0.039) < 0.01-0.004 (-0.010, 0.001) 0.12-0.005 (-0.010, 0.001) 0.08Femoral tuberosityln-2,4,5-TCP-0.041 (-0.046, -0.035) < 0.01-0.003 (-0.008, 0.002) 0.29-0.003 (-0.008, 0.003) 0.32ln-2,4,6-TCP-0.037 (-0.042, -0.031) < 0.01-0.004 (-0.009, 0.002) 0.20-0.004 (-0.009, 0.002) 0.17Intertrochanteric femurln-2,4,5-TCP-0.064 (-0.072, -0.057) < 0.01-0.004 (-0.011, 0.003) 0.25-0.004 (-0.011, 0.003) 0.26ln-2,4,6-TCP-0.056 (-0.064, -0.047) < 0.01-0.004 (-0.011, 0.003) 0.23-0.005 (-0.012, 0.002) 0.20Lumbar spineln-2,4,5-TCP-0.033 (-0.039, -0.027) < 0.01-0.008 (-0.014, -0.001) 0.02 -0.007 (-0.013, -0.000) 0.04 ln-2,4,6-TCP-0.029 (-0.036, -0.023) < 0.01-0.006 (-0.013, 0.000) 0.06-0.006 (-0.013, 0.000) 0.06Trichlorophenol: TCP; Bone mineral density: BMD; Model 1 did not adjust for any confounding factors. Model 2 was adjusted for sex, age, BMI and race. Model 3 also adjusts for sex, age, race, BMI, household income poverty ratio, education level, milk consumption, alcohol consumption, serum cotinine, serum lead, calcium concentration, phosphorus concentration, diabetes, moderate physical activity and vigorous physical activity
## Stratified analysis
In Table 3, a stratified study on the relationship between ln-2,4,5-TCP, ln-2,4,6-TCP and BMD of five study sites in men and women, as well as before and after menopause. In model 1, the results of stratified analysis showed that TCP was negatively correlated with BMD in all parts. In model 3, after adjusting for the included covariates, stratified analysis showed that ln-2,4,5-TCP and ln-2,4,6-TCP had no statistically significant relationship with BMD.
Table 3Stratified analysis of TCP and BMD.BMDModel 1 β ($95\%$ CI) P-valueModel 2 β ($95\%$ CI) P-valueModel 3 β ($95\%$ CI) P-valueMaleTotal femurln-2,4,5-TCP-0.039 (-0.048, -0.029) < 0.01-0.007 (-0.015, 0.002) 0.13-0.005 (-0.013, 0.004) 0.27ln-2,4,6-TCP-0.028 (-0.038, -0.019) < 0.01-0.006 (-0.014, 0.002) 0.16-0.005 (-0.014, 0.003) 0.21Femoral neckln-2,4,5-TCP-0.043 (-0.052, -0.034) < 0.01-0.003 (-0.011, 0.005) 0.41-0.002 (-0.010, 0.006) 0.62ln-2,4,6-TCP-0.032 (-0.042, -0.023) < 0.01-0.005 (-0.013, 0.003) 0.19-0.005 (-0.013, 0.003) 0.22Femoral tuberosityln-2,4,5-TCP-0.026 (-0.034, -0.018) < 0.01-0.006 (-0.014, 0.002) 0.13-0.004 (-0.012, 0.003) 0.26ln-2,4,6-TCP-0.019 (-0.027, -0.011) < 0.01-0.005 (-0.013, 0.003) 0.21-0.004 (-0.012, 0.003) 0.29Intertrochanteric femurln-2,4,5-TCP-0.045 (-0.056, -0.034) < 0.01-0.008 (-0.018, 0.002) 0.12-0.006 (-0.016, 0.004) 0.24ln-2,4,6-TCP-0.032 (-0.043, -0.021) < 0.01-0.007 (-0.017, 0.003) 0.19-0.006 (-0.016, 0.004) 0.24Lumbar spineln-2,4,5-TCP-0.016 (-0.025, -0.007) < 0.01-0.009 (-0.019, 0.000) 0.05-0.008 (-0.017, 0.001) 0.09ln-2,4,6-TCP-0.013 (-0.023, -0.004) < 0.01-0.007 (-0.016, 0.003) 0.17-0.006 (-0.015, 0.003) 0.19BMD Menstruating Female Total femurln-2,4,5-TCP-0.028 (-0.038, -0.017) < 0.01-0.005 (-0.015, 0.005) 0.34-0.007 (-0.017, 0.003) 0.19ln-2,4,6-TCP-0.031 (-0.041, -0.020) < 0.01-0.008 (-0.019, 0.002) 0.11-0.010 (-0.020, 0.001) 0.06Femoral neckln-2,4,5-TCP-0.033 (-0.044, -0.023) < 0.01-0.004 (-0.013, 0.006) 0.44-0.005 (-0.015, 0.004) 0.28ln-2,4,6-TCP-0.036 (-0.047, -0.025) < 0.01-0.007 (-0.017, 0.003) 0.15-0.009 (-0.019, 0.001) 0.09Femoral tuberosityln-2,4,5-TCP-0.018 (-0.027, -0.009) < 0.01-0.004 (-0.013, 0.005) 0.39-0.005 (-0.015, 0.004) 0.24ln-2,4,6-TCP-0.021 (-0.030, -0.011) < 0.01-0.006 (-0.015, 0.003) 0.19-0.007 (-0.017, 0.002) 0.13Intertrochanteric femurln-2,4,5-TCP-0.030 (-0.042, -0.018) < 0.01-0.005 (-0.017, 0.007) 0.43-0.007 (-0.019, 0.005) 0.25ln-2,4,6-TCP-0.032 (-0.045, -0.020) < 0.01-0.008 (-0.021, 0.004) 0.19-0.010 (-0.022, 0.003) 0.12Lumbar spineln-2,4,5-TCP-0.021 (-0.031, -0.010) < 0.01-0.006 (-0.016, 0.005) 0.29-0.004 (-0.015, 0.006) 0.43ln-2,4,6-TCP-0.024 (-0.034, -0.013) < 0.01-0.007 (-0.018, 0.004) 0.20-0.005 (-0.016, 0.006) 0.38BMD Menopausal Female Total femurln-2,4,5-TCP-0.034 (-0.048, -0.020) < 0.010.004 (-0.009, 0.017) 0.530.005 (-0.007, 0.018) 0.41ln-2,4,6-TCP-0.028 (-0.044, -0.013) < 0.01-0.000 (-0.014, 0.013) 0.960.003 (-0.010, 0.017) 0.60Femoral neckln-2,4,5-TCP-0.032 (-0.045, -0.019) < 0.010.006 (-0.006, 0.018) 0.340.007 (-0.005, 0.019) 0.24ln-2,4,6-TCP-0.028 (-0.043, -0.014) < 0.01-0.001 (-0.013, 0.012) 0.890.002 (-0.010, 0.015) 0.72Femoral tuberosityln-2,4,5-TCP-0.022 (-0.034, -0.011) < 0.010.004 (-0.007, 0.016) 0.440.005 (-0.006, 0.017) 0.36ln-2,4,6-TCP-0.020 (-0.033, -0.007) < 0.01-0.000 (-0.012, 0.012) 0.970.003 (-0.009, 0.014) 0.64Intertrochanteric femurln-2,4,5-TCP-0.042 (-0.058, -0.025) < 0.010.003 (-0.012, 0.018) 0.710.004 (-0.011, 0.019) 0.58ln-2,4,6-TCP-0.033 (-0.051, -0.015) < 0.01-0.000 (-0.016, 0.016) 0.970.004 (-0.012, 0.020) 0.60Lumbar spineln-2,4,5-TCP-0.028 (-0.042, -0.014) < 0.01-0.002 (-0.016, 0.012) 0.78-0.002 (-0.015, 0.012) 0.83ln-2,4,6-TCP-0.024 (-0.039, -0.009) < 0.01-0.004 (-0.018, 0.010) 0.58-0.001 (-0.015, 0.013) 0.90Stratified analyses based on sex and menopausal state in women
## Trend testing and curve fitting
ln-2,4,5-TCP was divided into quartiles. A curve fit analysis of BMD in the lumbar spine with ln-2,4,5-TCP is shown in Fig. 2. We found that the three lines representing male, menstruating female and menopausal female have no obvious linear relationship, so we performed a trend test to verify (Table 4). Trend test results showed that ln-2,4,5-TCP was not linearly correlated with lumbar spine BMD ($P \leq 0.05$).
Fig. 2Smooth curve fit of lumbar spine BMD and ln-2,4,5-TCP. Red: Male; Green: Menstruating Female; Blue: Menopausal Female Table 4Trend test of lumbar spine BMD and ln-2,4,5-TCP.BMDMale (β ($95\%$ CI) P-value)Menstruating Female(β ($95\%$ CI) P-value)Menopausal Female(β ($95\%$ CI) P-value)Q1---Q20.012 (-0.005, 0.029) 0.16-0.017 (-0.039, 0.005) 0.140.001 (-0.040, 0.041) 0.98Q30.002 (-0.016, 0.020) 0.85-0.019 (-0.042, 0.004) 0.100.004 (-0.034, 0.042) 0.85Q4-0.012 (-0.032, 0.007) 0.22-0.013 (-0.037, 0.011) 0.280.003 (-0.035, 0.041) 0.88
## Discussion
This is, to the best of our knowledge, the first and largest population-based study to examine the relationship between BMD and 2,4,5-TCP and 2,4,6-TCP in a nationally representative sample.
TCPs are organochlorine compounds that are found all over the environment and are known to cause cancer. Carcinogenesis bioassays were conducted by giving 2,4,6-TCP in feed to groups of 50 male and female Fischer rats and male B6C3F1 mice for two years. Dietary concentrations were 0 [20/group], 5000 [$0.5\%$], or 10,000 [$1\%$] ppm. It was found that TCP caused leukemia/lymphoma and liver tumor in rats [20]. 2,4,6-TCP poses a risk of human cancer, according to a number of national and international agencies. Therefore, for primary cancer prevention, exposures to 2,4,6-TCP should be minimized or eliminated, as is the case with all carcinogens that can cause cancer in animals or humans [21–25]. However, little is known about their other adverse effects on humans. Due to the numerous exposure pathways, it is difficult to measure individual exposure to TCPs. Exposure to TCPs or organochlorine chemicals that are metabolized and excreted as TCPs is indicated by TCPs excreted in the urine [2]. Urinary TCP levels can be used to accurately estimate individual exposure [26]. Organochlorine pesticides, which include 2,4,5-TCP and 2,4,6-TCP, are a large class of multipurpose chlorinated hydrocarbon chemicals that accumulate in fatty tissue and slowly degrade in the environment. Xu found that exposure to TCP may increase the risk of behavioural impairment in children in a 1999–2004 NHANES study of 2546 children [27]. Several biomonitoring studies on phenolic compounds have shown that the determination of the actual urinary excretion would provide a reliable estimation of individual exposure and that the levels of chemicals excreted from the body vary significantly depending on the population studied, reflecting differences in race and ethnicity [28–30]. In comparison to concentrations found in the United States and Canada, TCP exposure is still fairly common, but exposure levels among children and adolescents in Germany were generally low [31]. Geometric mean urinary 2,4,5-TCP concentrations decreased with age and increased with education level and income. Age remained significantly related to urinary 2,4,5-TCP concentration after the adjustment of covariates [32]. Therefore, in this study, race, age, education level, and income were considered for adjustment as covariates. A Korean study of 165 girls aged 7 to 8 years found that chlorophenol exposure was positively associated with central obesity in Korean girls [33]. While it is common knowledge that weight gain builds bone, several obesity-related mechanisms make bone more fragile. These include an increase in inflammatory cytokines that activates bone-resorbing osteoclasts, mutations in the FTO gene, increased osteoblast senescence caused by obesity, and an increased production of bone marrow fat cells at the expense of bone-forming osteoblasts [34]. Therefore, the covariate also includes BMI.
Although the aforementioned metabolites are considered to be endocrine disruptors (EDs), only a small number of population-based studies have examined the relationships between these metabolites and osteoporosis, which is another important endocrine disorder. One recent study primarily on personal care products found that in men and premenopausal women, higher paraben concentrations, particularly ethyl-, methyl-, and propylparabens, were linked to higher BMD in the femoral neck, intertrochanter, and lumbar spine. Men had a higher prevalence of osteopenia/osteoporosis and a lower BMD when exposed to 2,4-dichlorophenol. Postmenopausal women were found to have a higher prevalence of osteopenia and osteoporosis in the lumbar spine when exposed to bisphenol A. Men and premenopausal women tended to have a higher BMD of the femur when exposed to benzophenone-3 [9]. However, to date, no studies have shown whether TCP affects bone health in humans. Only one animal study showed that development of zebrafish head cartilages was seriously affected by exposure of triclosan, 2,4-dichlorophenol and 2,4,6-trichlorophenol [35].
A total of 3385 participants were included in this study, including 1737 males and 1648 females. The mean ln-2,4,5-TCP and ln-2,4,6-TCP for males were − 7.19 ug/L and − 5.59 ug/L, respectively. The mean ln-2,4,5-TCP and ln-2,4,6-TCP for females were − 6.81 ug/L and − 5.25 ug/L. Therefore, males have less ln-TCP on average than females. This may be related to the difference in the metabolic capacity of men and women. In the unadjusted model, both ln-2,4,5-TCP and ln-2,4,6-TCP were negatively correlated with BMD of five study sites. However, after adjusting for confounders such as sex, age, race, BMI, household income poverty ratio, education level, milk consumption, alcohol consumption, serum cotinine, serum lead, calcium concentration, phosphorus concentration, diabetes, moderate physical activity and vigorous physical activity, only ln-2,4,5-TCP were still negatively correlated with BMD of lumbar spine. The results of stratified analysis showed that there was no significant linear relationship between ln-2, 4, 5-TCP and BMD of lumbar spine in the three stratifications of male, menstruating female and menopausal female. Curve fitting analysis and trend test also verified this conclusion. Therefore, ln-2,4,5-TCP and ln-2,4,6-TCP levels in the US population may not adversely affect BMD. However, this result is only applicable to the American population. Due to the differences in the production, lifestyle and ethnicity of each country, more research may be needed to evaluate the harm of TCP.
This study had a number of limitations. First, this study was a cross-sectional one. As a result, it is impossible to fully verify causal relationships or coincidental phenomena. Second, it is impossible to completely rule out residual and unmeasured confounding factors in this observational study. Thirdly, additional research is required to ascertain the underlying mechanism and direction of the associations’ causality. Last but not least, since this paper is the first study to study the relationship between TCP and human bone health, there are not many literatures that can be cited or referenced. However, this is also a major innovation of this study.
## Conclusion
This study demonstrated that 2,4,5-TCP and 2,4,6-TCP in urine samples were not significantly associated with BMD in the US population. Therefore, 2,4,5-TCP and 2,4,6-TCP may not be detrimental to BMD.
## References
1. Kauppinen T, Kogevinas M, Johnson E, Becher H, Bertazzi PA, Bueno de Mesquita HB. **Chemical exposure in manufacture of phenoxy herbicides and chlorophenols and in spraying of phenoxy herbicides**. *Am J Ind Med* (1993.0) **23** 903-20. DOI: 10.1002/ajim.4700230607
2. Nichkova M, Marco MP. **Biomonitoring human exposure to organohalogenated substances by measuring urinary chlorophenols using a high-throughput screening (HTS) immunochemical method**. *Environ Sci Technol* (2006.0) **40** 2469-77. DOI: 10.1021/es0518629
3. Lechner ML, Somogyi MA, Biro ML. **Detection and determination of chlorophenols and chlorophenoxyacetic acids by instrumental analysis**. *Talanta* (1966.0) **13** 581-7. DOI: 10.1016/0039-9140(66)80268-0
4. Macholz RM, Knoll R, Lewerenz HJ, Petrzika M, Engst R. **Metabolism of alpha-hexachlorocyclohexane. Free metabolites in urine and organs of rats**. *Xenobiotica* (1982.0) **12** 227-31. DOI: 10.3109/00498258209052460
5. Draper WM. **Biological monitoring: exquisite research probes, risk assessment, and routine exposure measurement**. *Anal Chem* (2001.0) **73** 2745-60. DOI: 10.1021/ac010394s
6. Ensrud KE, Crandall CJ. *Ann Intern Med* (2017.0) **167** ITC17-ITC32. DOI: 10.7326/AITC201708010
7. Berman NK, Honig S, Cronstein BN, Pillinger MH. **The effects of caffeine on bone mineral density and fracture risk**. *Osteoporos Int* (2022.0). DOI: 10.1007/s00198-021-05972-w
8. Cotts KG, Cifu AS. **Treatment of osteoporosis**. *JAMA* (2018.0) **319** 1040-1. DOI: 10.1001/jama.2017.21995
9. Wang N, Wang Y, Zhang H, Guo Y, Chen C, Zhang W. **Association of bone mineral density with nine urinary personal care and consumer product chemicals and metabolites: a national-representative, population-based study**. *Environ Int* (2020.0) **142** 105865. DOI: 10.1016/j.envint.2020.105865
10. 10.National Health and Nutrition Examination Survey, Questionnaires D, Related Documentation. and. Available online: https://wwwn.cdc.gov/nchs/ doi
11. Fu YP, Chen WY, Guo LQ, Zhu YQ, Yuan JS, Liu YH. **The association between hearing threshold and urinary personal care and consumer product metabolites in middle-aged and elderly people from the USA**. *Environ Sci Pollut Res Int* (2022.0). DOI: 10.1007/s11356-022-21459-5
12. Wen X, Wang M, Xu X, Li T. **Exposure to Per- and polyfluoroalkyl substances and mortality in U.S. adults: a Population-Based Cohort Study**. *Environ Health Perspect* (2022.0) **130** 67007. DOI: 10.1289/EHP10393
13. 13.Centers for Disease Control and Prevention. NHaNES-dd, codebook, and frequencies: dual energy X-ray absorptiometry-femur (DXXFEM_F). https://wwwn.cdc.gov/Nchs/Nhanes/2009-2010/DXXFEM_F.htm.doi
14. Looker AC, Melton LJ, Harris T, Borrud L, Shepherd J, McGowan J. **Age, sex, and race/ethnic differences in total body and subregional bone density**. *Osteoporos Int* (2009.0) **20** 1141-9. DOI: 10.1007/s00198-008-0809-6
15. Kanis JA. **Assessment of fracture risk and its application to screening for postmenopausal osteoporosis: synopsis of a WHO report. WHO Study Group**. *Osteoporos Int* (1994.0) **4** 368-81. DOI: 10.1007/BF01622200
16. Looker AC, Borrud LG, Hughes JP, Fan B, Shepherd JA, Melton LJ. **Lumbar spine and proximal femur bone mineral density, bone mineral content, and bone area: United States, 2005–2008**. *Vital Health Stat* (2012.0) **11** 1-132
17. Looker AC, Johnston CC, Wahner HW, Dunn WL, Calvo MS, Harris TB. **Prevalence of low femoral bone density in older U.S. women from NHANES III**. *J Bone Miner Res* (1995.0) **10** 796-802. DOI: 10.1002/jbmr.5650100517
18. 18.https://wwwn.cdc.gov/Nchs/Nhanes/2009-2010/PP_F.htm NHaNESP-E-U.doi
19. Ye X, Kuklenyik Z, Needham LL, Calafat AM. **Automated on-line column-switching HPLC-MS/MS method with peak focusing for the determination of nine environmental phenols in urine**. *Anal Chem* (2005.0) **77** 5407-13. DOI: 10.1021/ac050390d
20. Huff J. **Long-term toxicology and carcinogenicity of 2,4,6-trichlorophenol**. *Chemosphere* (2012.0) **89** 521-5. DOI: 10.1016/j.chemosphere.2012.05.015
21. Tomatis L. **Primary prevention of cancer in relation to science, sociocultural trends and economic pressures**. *Scand J Work Environ Health* (2005.0) **31** 227-32. DOI: 10.5271/sjweh.874
22. Tomatis L. **Identification of carcinogenic agents and primary prevention of cancer**. *Ann N Y Acad Sci* (2006.0) **1076** 1-14. DOI: 10.1196/annals.1371.039
23. Tomatis L. **Role of experimental and epidemiological evidence of carcinogenicity in the primary prevention of cancer**. *Ann Ist Super Sanita* (2006.0) **42** 113-7. PMID: 17033130
24. Huff J. **Primary prevention of cancer**. *Science* (2011.0) **332** 916-7. DOI: 10.1126/science.332.6032.916-b
25. Huff J. **Predicting chemicals causing cancer in animals as human carcinogens**. *Occup Environ Med* (2010.0) **67** 720. DOI: 10.1136/oem.2009.054569
26. Angerer J, Maass R, Heinrich R. **Occupational exposure to hexachlorocyclohexane. VI. Metabolism of gamma-hexachlorocyclohexane in man**. *Int Arch Occup Environ Health* (1983.0) **52** 59-67. DOI: 10.1007/BF00380608
27. Xu X, Nembhard WN, Kan H, Kearney G, Zhang ZJ, Talbott EO. **Urinary trichlorophenol levels and increased risk of attention deficit hyperactivity disorder among US school-aged children**. *Occup Environ Med* (2011.0) **68** 557-61. DOI: 10.1136/oem.2010.063859
28. Calafat AM, Ye X, Wong LY, Reidy JA, Needham LL. **Exposure of the U.S. population to bisphenol A and 4-tertiary-octylphenol: 2003–2004**. *Environ Health Perspect* (2008.0) **116** 39-44. DOI: 10.1289/ehp.10753
29. Kim K, Park H, Yang W, Lee JH. **Urinary concentrations of bisphenol A and triclosan and associations with demographic factors in the korean population**. *Environ Res* (2011.0) **111** 1280-5. DOI: 10.1016/j.envres.2011.09.003
30. Vandenberg LN, Chahoud I, Heindel JJ, Padmanabhan V, Paumgartten FJ, Schoenfelder G. **Urinary, circulating, and tissue biomonitoring studies indicate widespread exposure to bisphenol A**. *Cien Saude Colet* (2012.0) **17** 407-34. DOI: 10.1590/s1413-81232012000200015
31. Schmied-Tobies MIH, Murawski A, Schmidt L, Rucic E, Schwedler G, Apel P. **Pentachlorophenol and nine other chlorophenols in urine of children and adolescents in Germany - Human biomonitoring results of the German Environmental Survey 2014–2017 (GerES V)**. *Environ Res* (2021.0) **196** 110958. DOI: 10.1016/j.envres.2021.110958
32. Kim K, Park H, Lee JH. **Urinary concentrations of trichlorophenols in the korean adult population: results of the National Human Biomonitoring Survey 2009**. *Environ Sci Pollut Res Int* (2014.0) **21** 2479-85. DOI: 10.1007/s11356-013-2180-1
33. Seo MY, Choi MH, Hong Y, Kim SH, Park MJ. **Association of urinary chlorophenols with central obesity in korean girls**. *Environ Sci Pollut Res Int* (2021.0) **28** 1966-72. DOI: 10.1007/s11356-020-10628-z
34. Proietto J. *F1000Res* (2020.0). DOI: 10.12688/f1000research.20875.1
35. Zhang Y, Liu M, Liu J, Wang X, Wang C, Ai W. **Combined toxicity of triclosan, 2,4-dichlorophenol and 2,4,6-trichlorophenol to zebrafish (Danio rerio)**. *Environ Toxicol Pharmacol* (2018.0) **57** 9-18. DOI: 10.1016/j.etap.2017.11.006
|
---
title: 'Knowledge of modifiable cardiovascular diseases risk factors and its primary
prevention practices among diabetic patients at Jimma University Medical Centre:
A cross-sectional study'
authors:
- Abdata Workina
- Asaminew Habtamu
- Tujuba Diribsa
- Fikadu Abebe
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022065
doi: 10.1371/journal.pgph.0000575
license: CC BY 4.0
---
# Knowledge of modifiable cardiovascular diseases risk factors and its primary prevention practices among diabetic patients at Jimma University Medical Centre: A cross-sectional study
## Abstract
Cardiovascular diseases (CVDs) are the most common cause of mortality and morbidity globally. This is due to the increasing prevalence of modifiable CVDs risk factors. Hence, the study was aimed to identify knowledge and unhealthy behaviors that contribute to CVD among diabetes mellitus patients. An institutional-based cross-sectional study design was employed among diabetes mellitus patients. A close-ended questionnaire developed from up-to-date similar literature was pretested and face-to-face interview techniques were used to collect data. Checked data were entered into the Epidata 4.1 versions. Then, descriptive and bivariate logistic regression was done using SPSS 25 versions. Of the study participants included in the study, 318, more than half of them, 167($52.5\%$), were aged ≥45 years and 187($58.8\%$) of them were females. Among the study participants, more than half, 198($62.3\%$), had good Knowledge of modifiable CVDs risk factors. Most of the study participants identified consuming foods rich in fats instead of vegetables and fruits 198($62.3\%$), followed by physical inactivity 196($61.6\%$) as a risk factor for CVD. Regarding CVDs prevention practice, 175($55.0\%$) of the patients had a good practice. More than three-fourths, 267($84.0\%$), practice avoiding foods rich in fats and, sugar, and cigarette smoking 250($78.6\%$). Predictors like educational status, college and above (AOR 2.68; $95\%$ CI 1.14–6.27), and urban residence area (AOR 1.94; $95\%$ CI 1.09–3.15) were associated with knowledge of CVDs risk factors. While sex, marital status monthly income, and age of the participants had no association with knowledge and prevention practice of modifiable CVDs risk factors. The study participants’ knowledge and prevention practice of modifiable cardiovascular disease risk factors was satisfactory, even though continuous awareness creation is required to lower CVD mortality and morbidity burdens. Educational status and residence are of the study participants affect the knowledge and prevention practice modifiable of cardiovascular disease risk factors.
## Introduction
CVDs are a group of disorders of the heart and blood vessels including coronary heart disease, cerebrovascular disease(stroke), peripheral arterial disease, rheumatic heart disease, heart defects, hypertension, deep vein thrombosis, and pulmonary embolism [1,2]. Coronary heart disease and stroke were the top two leading causes of CVD health loss in each world region [3].
Cardiovascular diseases are the most common cause of mortality and morbidity globally [4–6]. About 17.9 million people die annually from cardiovascular diseases (CVDs). This represents $32\%$ of all deaths worldwide annually. Of these deaths >$75\%$ happen in middle- and low-income countries [7]. More than 22.2 million people will die annually by 2030 due to CVDs unless intervened [8].
The modifiable CVD risk factors include smoking tobacco, hypercholesterolemia, diabetes, sedentary lifestyle, overweight/obesity, high-fat content diet, and excessive alcohol consumption [3,6,9]. According to, 2017 global burdens of disease report, high blood pressure, and smoking were the leading global risk factors causing early death and disability for all age groups [10].
In Ethiopia, dyslipidemia ($90.6\%$), physical inactivity ($76\%$), and hypertension ($62.7\%$) were the most common prevalent cardiovascular disease risk factors [11].
The effects of behavioral risk factors that contribute to the development of cardiovascular diseases and its complication may show up in individuals as raised blood pressure, raised blood glucose, raised blood lipids, and obesity. Furthermore, economic and cultural change, globalization, urbanization, population aging, poverty, and hereditary factors were the underlying determinants of CVDs [12,13].
Most cardiovascular diseases can be prevented by cessation of tobacco use, reduction of salt in the diet, eating more fruit and vegetables instead of eating high-fat content foods, regular physical activity, and avoiding excessive alcohol consumption have been shown to reduce the risk of cardiovascular disease. Creating conducive health policy environments that making healthy choices affordable and available are essential for motivating people to adopt and sustain healthy behaviors [3,12–14].
Additionally, early detection and treatment adherence of hypertension, diabetes and high blood lipids are a cost-effective approach to reduce cardiovascular disease burdens both in high and low-income countries [15].
Different studies were conducted regarding knowledge and prevention practice of modifiable cardiovascular disease worldwide, even though the global pandemic of modifiable cardiovascular disease risk factors were as usual especially in low-income countries [5,6,15–19]. However, to date, no information was found in Ethiopia regarding knowledge and prevention practice of modifiable cardiovascular disease. Hence, this study was aimed to identify knowledge and unhealthy behaviors that contributed to CVDs and factors associated with it.
## Ethics statement
The ethical clearance approved by Jimma university institute of health IRB with Ref No: IHRPGD/$\frac{469}{2020}$ was obtained and given to Jimma University’s medical centre administrator. Verbal informed consent was obtained from the hospital administrator to collect the data and this study was conducted in accordance with the Helsinki declaration. Written informed consent was taken from each study participant. Data collectors explained the objectives, their right to refuse and discontinue the data collection. Participant’s name was not recorded as well as their confidentiality was highly secured through this study process.
## Study design and setting
An institutional-based cross-sectional study design was conducted at Jimma university medical center (JUMC) from January 8, 2021, to February 24, 2021. Jimma university medical center was located in the southwest of Ethiopia, providing teaching and research services in addition to medical services. The medical center had provided services for more than 20 million patients with 800 beds from the southwest of Ethiopia [20,21].
## Study population
Patients with diabetes mellitus visited Jimma university medical center for follow-up at the diabetic clinic during the data collection period (January 8, 2021, to February 24, 2021).
## Study variables
The predictor study variables were sociodemographic characteristics (age, sex, marital status, occupation, educational status, residence area, monthly income and types of DM and dependent variables were knowledge and prevention practice of modifiable CVDs risk factors.
## Eligibility criteria
Diabetic Mellitus patients whose age was 18 years old and above who visited the study site for follow-up were included in the study while diabetic mellitus patients who have already developed cardiovascular disease and mentally ill were excluded from the study.
## Sample size determination and sampling procedure
The required sample size was determined using single population formula with the assumption of confidence interval $95\%$, the margin of error $5\%$, and since no study was conducted in Ethiopia on knowledge of cardiovascular disease risk factors and its primary prevention practice proportion of diabetic patients who had good knowledge of cardiovascular disease risk factors (p) $50\%$ was considered; initial sample size (ni) = (Zα/2)2p (1-p)/d2 = 384. Since the sample size was taken from a population of < 10, 000 the initial sample size was adjusted using the correction formula; final sample size (nf) = ni (N)/ ni+$$n = 294$$, where $$n = 1260$$ number of diabetic mellitus patients who were on follow up at JUMC. Finally, after adding a $10\%$ non-response rate a total of 323 diabetic patients were targeted for the study. The list of patient’s record orders on the appointment chart was used as a sampling frame. To select each study unit, the systematic sampling technique was employed using the sampling fraction (k) = N/n; $\frac{1260}{323}$ = 4. Then each study participant was selected by adding a sample fraction until reaching the total sample size targeted for this study and the first study participants were selected using lottery methods.
## Data collection tool and procedure
A closed-ended questionnaire was developed to update similar literature which contains socio-demographic characteristics, knowledge of modifiable cardiovascular disease risk factors, and cardiovascular disease risk factors prevention measuring variables [18,19,22–25]. Knowledge of modifiable CVDs risk factors and its prevention practices were dichotomized into good or poor knowledge of cardiovascular disease risk factors or prevention practices of cardiovascular disease. Good knowledge of cardiovascular disease risk factors or primary prevention practices of cardiovascular diseases were computed from the mean score and those participants who had knowledge of CVDs risk factors or its prevention practices above mean score were considered to have good knowledge or good CVDs prevention practice [16,22,26]. Data were collected by 5 BSc nurses through face-to-face interview techniques.
## Data quality assurance
The questionnaire was pretested on $5\%$ of the sample size on patients with diabetes mellitus who were on follow-up at Shenen Gibe hospital. During the pre-test, each variable of the questionnaire was assessed for its understand-ability, sensitivity, and reliability statistics was computed with Cronbach’s alpha of 0.83. The questionnaire was translated from the English version to Afan Oromo and Amharic languages by language experts for data collection then, back to English during data analysis. Data collectors obtained training three days before data collection and the supervisor provided on-site close supervision and checked the completeness of the questionnaires’ during data collection.
## Data processing and analysis
Checked data were entered into Epi-data entry client 4.6 versions. Then, cleaned data were exported and analyzed using SPSS 25.0 versions. Descriptive statistics were used to summarize categorical variables of patients’ socio-demographic characteristics, knowledge of risk factors, and prevention practice of CVD. The knowledge of CVD risk factors and its prevention practice was dichotomized into good/poor knowledge of CVD risk factors or CVD prevention practice then, it was analyzed using binary logistic regression. The fit of the model was checked by the Hosmer-Lemeshow goodness of fit test. In the binary logistic regression predictors with p-value < 0.25 at $95\%$ CI were candidates for multivariate logistic regression. In the multivariate logistic regression independent variables with a p-value of < 0.05 were considered statistically significant association.
## Socio-demographic characteristics
Of the study participants included in the study, 318, more than half of them, 167($52.5\%$), were aged ≥45 years and 187($58.8\%$) of them were females. Regarding educational status, most of them, 95($29.9\%$) were educated up preparatory school followed by college and above 81($25.5\%$) whereas, 66($20.8\%$) of the study participants were illiterate. Concerning residential areas, 210($66.0\%$) of the study participants reside in urban areas. Among the study participants, 249($78.3\%$) had type 2 diabetes mellitus, while the rest had type 1 diabetic mellitus (Table 1).
**Table 1**
| Variables(n = 318) | Variables(n = 318).1 | Frequency | Percent (%) |
| --- | --- | --- | --- |
| Age (in yrs.) | <45 | 151 | 47.5 |
| Age (in yrs.) | ≥45 | 167 | 52.5 |
| Sex | Male | 131 | 41.2 |
| Sex | Female | 187 | 58.8 |
| Marital status | Married | 219 | 68.9 |
| Marital status | Single | 60 | 18.9 |
| Marital status | Divorced | 25 | 7.9 |
| Marital status | Widowed | 14 | 4.4 |
| Occupation | Farmer | 60 | 18.9 |
| Occupation | Merchant | 68 | 21.4 |
| Occupation | Government employee | 51 | 16.0 |
| Occupation | Other* | 56 | 17.6 |
| Occupation | Housewife | 83 | 26.1 |
| Educational status | Elementary | 76 | 23.9 |
| Educational status | High school/preparatory | 95 | 29.9 |
| Educational status | College and above | 81 | 25.5 |
| Educational status | Illiterate | 66 | 20.8 |
| Residence area | Urban | 210 | 66.0 |
| Residence area | Rural | 108 | 34.0 |
| Monthly income (in ETB) | = >5000 | 38 | 11.9 |
| Monthly income (in ETB) | <5000 | 280 | 88.1 |
| Types of DM patients had | Type 2 DM | 249 | 78.3 |
| Types of DM patients had | Type 1 DM | 69 | 21.7 |
## Knowledge of modifiable cardiovascular disease risk factors
Amongst the study participants, 198($62.3\%$) of them were identified consuming foods rich in fats instead of vegetables and fruits might cause CVD, followed by physical inactivity 196($61.6\%$) while, only less than half, 152($47.8\%$), the patients knew that cigarette smoking was a risk factor of cardiovascular diseases. More than half, 198($62.3\%$), of the study participants had good Knowledge of modifiable CVDs risk factors (Table 2).
**Table 2**
| Variables | Variables.1 | Frequency | Percent (%) |
| --- | --- | --- | --- |
| Is high blood pressure a risk factor for CVD? | Yes | 174 | 54.7 |
| Is high blood pressure a risk factor for CVD? | No | 54 | 17.0 |
| Is high blood pressure a risk factor for CVD? | I don’t know | 90 | 28.3 |
| Is overweight a risk factor for CVD? | Yes | 181 | 56.9 |
| Is overweight a risk factor for CVD? | No | 71 | 22.3 |
| Is overweight a risk factor for CVD? | I don’t know | 66 | 20.8 |
| Is excessive alcohol intake a risk factor for CVD? | Yes | 169 | 53.1 |
| Is excessive alcohol intake a risk factor for CVD? | No | 76 | 23.9 |
| Is excessive alcohol intake a risk factor for CVD? | I don’t know | 73 | 23.0 |
| Does DM be a risk factor for CVD? | Yes | 153 | 48.1 |
| Does DM be a risk factor for CVD? | No | 71 | 22.3 |
| Does DM be a risk factor for CVD? | I don’t know | 94 | 29.6 |
| Is physical inactivity a risk factor for CVD? | Yes | 196 | 61.6 |
| Is physical inactivity a risk factor for CVD? | No | 42 | 13.2 |
| Is physical inactivity a risk factor for CVD? | I don’t know | 80 | 25.2 |
| Is cigarette smoking a risk factor for CVD? | Yes | 152 | 47.8 |
| Is cigarette smoking a risk factor for CVD? | No | 71 | 22.3 |
| Is cigarette smoking a risk factor for CVD? | I don’t know | 95 | 29.9 |
| Is consuming foods rich in fats instead of vegetables and fruit a risk factor for CVD? | Yes | 198 | 62.3 |
| Is consuming foods rich in fats instead of vegetables and fruit a risk factor for CVD? | No | 72 | 22.6 |
| Is consuming foods rich in fats instead of vegetables and fruit a risk factor for CVD? | I don’t know | 48 | 15.1 |
| Knowledge of modifiable CVDs risk factors | Poor knowledge | 120 | 37.7 |
| Knowledge of modifiable CVDs risk factors | Good knowledge | 198 | 62.3 |
## Primary prevention practices of cardiovascular diseases
Among the study participants more than three fourth, 267($84.0\%$), practice avoiding foods rich in fats, sugar, and salt, followed by avoiding cigarette smoking 250($78.6\%$), then adherence to DM treatment protocol 237($74.5\%$) although, only 137($43.1\%$) of them practice reduce alcohol consumption. Concerning cardiovascular diseases prevention practice status, 175($55.0\%$) of the patients had a good practice (Table 3).
**Table 3**
| Variables (n = 318) | Variables (n = 318).1 | Frequency | Percent (%) |
| --- | --- | --- | --- |
| Avoiding foods rich in fats, sugar, and salt | Yes | 267 | 84.0 |
| Avoiding foods rich in fats, sugar, and salt | No | 51 | 16.0 |
| Regular physical activity | Yes | 198 | 62.3 |
| Regular physical activity | No | 120 | 37.7 |
| Adherence to DM treatment protocol | Yes | 237 | 74.5 |
| Adherence to DM treatment protocol | No | 81 | 25.5 |
| Avoid cigarette smoking | Yes | 250 | 78.6 |
| Avoid cigarette smoking | No | 68 | 21.4 |
| Reduce alcohol consumption | Yes | 137 | 43.1 |
| Reduce alcohol consumption | No | 181 | 56.9 |
| Screening for high blood pressure | Yes | 142 | 44.7 |
| Screening for high blood pressure | No | 176 | 55.3 |
| CVDs prevention practice status | Poor practice | 143 | 45.0 |
| CVDs prevention practice status | Good practice | 175 | 55.0 |
## Factors associated with knowledge of modifiable cardiovascular diseases risk factors and its primary prevention practices
To identify factors associated with knowledge of modifiable cardiovascular disease risk factors and its prevention practices, knowledge of modifiable CVDs and its prevention practices was dichotomized into good or poor knowledge of risk factors and good or poor practice of CVD prevention. Then a cross tab was computed with predictor variables to identify whether cells were sufficient to perform logistic regression. Model fit was checked by using Hosmer and Lemeshow’s goodness of fit test. In the bivariate logistic regression predictors with p-value < 0.25 were candidates for multivariate logistic regression.
In the bivariate logistic regression, age ≥ 45 years (COR 1.31; $95\%$ CI.83–2.06), occupation of the patient, merchant (COR 2.93; $95\%$ CI 1.42–6.08), government employee, (COR 5.70; $95\%$ CI 2.36–13.77), housewife (COR 1.14; $95\%$ CI.58–2.21), elementary (COR 2.14; $95\%$ CI 1.15–3.95), high school/preparatory (COR 6.58; $95\%$ CI 3.22–13.48), college and above (COR 5.07; $95\%$ CI 2.44–10.50) and urban residence (COR 2.77; $95\%$ CI 1.71–4.48) were factors associated with knowledge of modifiable CVDs risk factors, while sex and marital status of the participants had no association with knowledge of CVDs risk factors (Table 4).
**Table 4**
| Variables | Variables.1 | Good knowledge of CVDs risk factors | Good knowledge of CVDs risk factors.1 | Good knowledge of CVDs risk factors.2 | Good practice of CVDs prevention | Good practice of CVDs prevention.1 | Good practice of CVDs prevention.2 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Variables | Variables | P-value | COR (95% C.I) | AOR (95% C.I) | P-value | COR (95% C.I) | AOR (95% C.I) |
| Age in yr. | ≥45 | .249 | 1.31(.83–2.06) | | .045 | 1.58(1.01–2.47) | |
| Age in yr. | <45 | .249 | 1.00 | | | 1.00 | |
| Sex | Male | .298 | 1.28(.80–2.04) | | .261 | .77(.49–1.21) | |
| Sex | Female | .298 | 1.00 | | | 1.00 | |
| Marital status | Married | .511 | 2.32(.78–6.92) | | .007* | .59(.19–1.82) | .10(.03-.39) |
| Marital status | Single | .511 | 2.15(.66–6.98) | | .007* | .78(.23–2.60) | .08(.02-.36) |
| Marital status | Divorced | .511 | 2.37(.62–9.03) | | .007* | 1.43(.35–5.79) | .37(.08-.84) |
| Marital status | Widowed | .511 | 1.00 | | | 1.00 | |
| Occupation | Farmer | 0.004 | 1.00 | | 0.011 | 1.00 | |
| Occupation | Merchant | 0.004 | 2.93(1.42–6.08) | | 0.011 | 2.41(1.18–4.91) | |
| Occupation | Government employee | 0.004 | 5.70(2.36–13.77) | | 0.011 | 6.53(2.70–15.82) | |
| Occupation | Housewife | 0.004 | 1.14(.58–2.21) | | 0.011 | 1.02(.52–2.00) | |
| Educational status | Illiterate | .001* | 1.00 | | | 1.00 | |
| Educational status | Elementary | .001* | 2.14(1.15–3.95) | 1.73(.90–3.32) | .000* | 4.71(2.36–9.44) | 7.82(3.38–18.09) |
| Educational status | High school/preparatory | .001* | 6.58(3.22–13.48) | 4.05(1.82–9.01) | .000* | 9.11(4.37–19.01) | 16.98(7.03–41.02) |
| Educational status | College and above | .001* | 5.07(2.44–10.50) | 2.68(1.14–6.27) | .000* | 16.58(7.24–37.98) | 30.28(11.61–78.98) |
| Residence area | Urban | 0.010* | 2.77(1.71–4.48) | 1.94(1.09–3.15) | .045 | 1.61(1.01–2.57) | |
| Residence area | Rural | 0.010* | 1.00 | | | 1.00 | |
| Monthly income | ≥5000 ETB | .236 | .64(.31–1.34) | | .016 | .40(.19-.84) | |
| Monthly income | <5000 ETB | .236 | 1.00 | | | 1.00 | |
Based on multivariate logistic regression those participants who had education status of college and above were 2.7 times more likely to have good knowledge of CVDs risk factors than illiterate participants (AOR 2.68; $95\%$ CI 1.14–6.27) and additionally, urban residence area (AOR 1.94; $95\%$ CI 1.09–3.15) were associated with knowledge of CVDs risk factors (Table 4).
With regard to CVDs prevention practices, marital status of married (AOR.10; $95\%$ CI.03-.39), single (AOR.08; $95\%$ CI (.02-.36), and divorced (AOR.37; $95\%$ CI.08-.84) and educational status of; elementary (AOR 7.82; $95\%$ CI 3.38–18.09), high school/preparatory (AOR 16.98; $95\%$ CI 7.03–41.02), college and above (AOR 30.28; $95\%$ CI 11.61–78.98) were predictors of good CVDs prevention practices (Table 4).
## Discussion
CVDs are the leading cause of premature death and disability worldwide [6,13,13,27]. Prevention of modifiable cardiovascular disease (CVD) risk factors have significantly reduced CVD mortality and morbidity [28]. Since magnitude of CVDs risk factors in the populations change over time and need continuous awareness creation and other intervention, this study was aimed to show gap on knowledge and primary prevention practice of modifiable cardiovascular disease risk factors and associated factors and might be used as secondary source of data for researchers who want to research on the same inquiry.
This study shows that more than half, 198($62.3\%$), of the study participants had good Knowledge of modifiable CVDs risk factors and most of them were identified consuming foods rich in fats instead of vegetables and fruits 198($62.3\%$), physical inactivity 196($61.6\%$) and hypertension 174 ($54.7\%$), while only less than half of the study participants knew, cigarette smoking 152 ($47.8\%$), and DM 153($48.1\%$) was a risk factor of CVDs. This study finding was relatively consistent with a study conducted in Malaysia which shows, $58\%$ of the participants know hypertension is a risk of CVD, [26] and as well with the study conducted in Egypt, which shows 88($44\%$) of the participants mentioned high blood pressure is a risk factor for heart diseases [25]. This finding can robust on evidence of key knowledge of CVDs risk factors similarity over various countries.
Whereas, it was lower than a study conducted in Nigeria among health care workers which revealed $86.2\%$ of the participants had good knowledge of chronic heart disease risk factors [29]. Again another study conducted in Nigeria among university students shows that smoking 342($85.1\%$), and hypertension 334($83.1\%$), was a risk for developing heart disease [24]. This discrepancy could be related to a study conducted in Nigeria, in which the study participants were health workers and university students with higher knowledge regarding CVDs. However, our study finding was higher than the study conducted in Tanzania revealed, only $25.4\%$ of the participants had good knowledge of CVD risk factors [30]. This inconsistency might be due to the researchers were used open-ended questions to assess the knowledge of CVD risk factors and the participants included in the study were only rural residential. Similarly, this study finding was higher than another study finding that revealed, most participants had unsatisfactory knowledge of risk factors ($64.6\%$) [31].
This study shows that amongst study participants 175($55.0\%$) of the patients had a good CVD prevention practice. More than three fourth, 267($84.0\%$), of them, practice avoiding foods rich in fats, sugar, and salt, followed by avoiding cigarette smoking 250($78.6\%$), then adherence to DM treatment protocol 237($74.5\%$) while, only 137($43.1\%$) of them practiced reducing of alcohol consumption. This study finding was relatively consistent with the study conducted in Ibadan, $54.8\%$ of the participants had good preventive behaviour against CVDs, [24] and Nigeria, which shows, 114 ($75.0\%$) of the participants eat fruits and vegetables regularly and 101 ($66.4\%$) of them ensure appropriate treatment of diabetes mellitus [29]. Moreover, our study finding was consistent with another study conducted in Nigeria that spectacles, 81 ($48.2\%$) had good practice of primary CVD prevention [15]. This shows that there almost related behavioural practices of modifiable cardiovascular disease risk factors.
However, our study finding was higher than the study conducted in Nepal and Kenya that spectacle, 113 ($31.7\%$) and 27($9.0\%$) of the respondents screened regularly for high blood pressure respectively [18,23]. This disparity could be due to differences in healthcare availability and study participants across studies. On the other hand, our study was lower than other undertaken in Iraq that shows, $82.5\%$ of participants reported that they had their blood pressure checked [19].
Based on multivariate logistic regression educational status and residence area had a strong association with knowledge modifiable CVDs risk factors, college and above (AOR 2.68; $95\%$ CI 1.14–6.27), and (AOR 1.94; $95\%$ CI 1.09–3.15), respectively. While there was no association with sex, age, income, and marital status of study participants. This study finding was similar to the study conducted in Cameroon, which shows a high level of education (AOR = 2.26 (1.69–3.02), $$p \leq 0.0001$$) was associated with good knowledge [32]. Furthermore, our study was consistent with studies conducted in Nigeria, Iraq, and Kenya that show there was no significant difference in knowledge scores among gender of participants, no significant association was found between the participants’ gender, marital status, and knowledge level ($P \leq 0.05$), and higher education was a strong predictor of CVD risk factor knowledge (6.72, $95\%$ CI 1.98–22.84, $P \leq 0.0001$) respectively [18,33,34]. This consistency shows that participants who had higher educational level might research out about their health. So, creating awareness could be tackled unhealthy behaviors of cardiovascular diseases.
On the other hand, our study was in contrast with another two studies undertaken in Nigeria and Palestine which reveals, sex and age, and gender, age-group ($$p \leq 0.039$$), and gender ($$p \leq 0.0.19$$) of participants were associated with knowledge of CVDs risk factors respectively [24,35,36]. This discrepancy might be due to sociodemographic characteristics difference among countries of studies.
## Strengths and limitations of the study
The study’s shortcoming was because it was focused on a single institution, generalization as a whole may not have been considered and since the duration of data collection was short it might affect the finding of the result.
The strengths of this study were that there is no information regarding knowledge and primary prevention practice of modifiable CVDs risk factors which makes the study more significant. Furthermore, the study participants were even at risk of developing the disease. So, the study can identify whether the study site institution provided information about unhealthy behaviors of CVDs and continuously monitored it.
## Conclusion
The study participants’ knowledge of modifiable cardiovascular disease risk factors was satisfactory, even though continuous awareness creation is required to lower CVD mortality and morbidity burdens. Primary cardiovascular diseases prevention practice among the study participants was good, but needs to be improved more.
Those patients who had higher educational level and residing in the urban area were more knowledgeable on modifiable CVDs risk factors and its prevention practice, whereas age, sex, marital status and income level of the participant had no effect on knowledge and prevention practice of modifiable CVDs risk factors.
## References
1. Hajar R.. **Framingham Contribution to Cardiovascular Disease**. *Heart Views Off J Gulf Heart Assoc* (2016.0) **17** 78-81. DOI: 10.4103/1995-705X.185130
2. Hajar R.. **Risk factors for coronary artery disease: Historical perspectives**. *Heart Views* (2017.0) **18** 109. DOI: 10.4103/HEARTVIEWS.HEARTVIEWS_106_17
3. Benjamin EJ, Blaha MJ, Chiuve SE, Cushman M, Das SR, Deo R. **Heart Disease and Stroke Statistics—2017 Update: A Report From the American Heart Association**. *Circulation* (2017.0) **135**. PMID: 28122885
4. Jauch EC, Saver JL, Adams HP, Bruno A, Demaerschalk BM, Khatri P. **A Guideline for Healthcare Professionals from the American Heart Association/American Stroke Association**. (2013.0) 870-94. DOI: 10.1161/STR.0b013e318284056a
5. Mazalin Protulipac J, Sonicki Z, Reiner Ž. **Cardiovascular disease (CVD) risk factors in older adults–Perception and reality**. *Arch Gerontol Geriatr* (2015.0) **61** 88-92. DOI: 10.1016/j.archger.2015.04.001
6. Mazloomy SS, Baghianimoghadam MH, Ehrampoush MH, Baghianimoghadam B, Mazidi M, Mozayan MR. **A Study of the Knowledge, Attitudes, and Practices (KAP) of the Women Referred to Health Centers for Cardiovascular Disease (CVDs) and Their Risk Factors.**. *Health Care Women Int* (2014.0) **35** 50-9. DOI: 10.1080/07399332.2012.755980
7. Kaptoge S, Pennells L, De Bacquer D, Cooney MT, Kavousi M, Stevens G. **World Health Organization cardiovascular disease risk charts: revised models to estimate risk in 21 global regions**. *Lancet Glob Health* (2019.0) **7** e1332-45. DOI: 10.1016/S2214-109X(19)30318-3
8. Ruan Y, Guo Y, Zheng Y, Huang Z, Sun S, Kowal P. **Cardiovascular disease (CVD) and associated risk factors among older adults in six low-and middle-income countries: results from SAGE Wave 1**. *BMC Public Health* (2018.0) **18** 778. DOI: 10.1186/s12889-018-5653-9
9. Angaw DA, Ali R, Tadele A, Shumet S. **The prevalence of cardiovascular disease in Ethiopia: a systematic review and meta-analysis of institutional and community-based studies**. *BMC Cardiovasc Disord* (2021.0) **21** 37. DOI: 10.1186/s12872-020-01828-z
10. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM. **Global Burden of Cardiovascular Diseases and Risk Factors, 1990–2019: Update From the GBD 2019 Study**. *J Am Coll Cardiol* (2020.0) **76** 2982-3021. DOI: 10.1016/j.jacc.2020.11.010
11. Abdosh T, Weldegebreal F, Teklemariam Z, Mitiku H. **Cardiovascular diseases risk factors among adult diabetic patients in eastern Ethiopia**. *JRSM Cardiovasc Dis* (2019.0) **8** 1-7. PMID: 31523424
12. Bebu I, Schade D, Braffett B, Kosiborod M, Lopes-Virella M, Soliman EZ. **Risk Factors for First and Subsequent CVD Events in Type 1 Diabetes: The DCCT/EDIC Study**. *Diabetes Care* (2020.0) **43** 867-74. DOI: 10.2337/dc19-2292
13. Arnett DK, Blumenthal RS, Albert MA, Buroker AB, Goldberger ZD, Hahn EJ. **2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease**. *J Am Coll Cardiol* (2019.0) **74** e177-232. DOI: 10.1016/j.jacc.2019.03.010
14. Kilkenny MF, Dunstan L, Busingye D, Purvis T, Reyneke M, Orgill M. **Knowledge of risk factors for diabetes or cardiovascular disease (CVD) is poor among individuals with risk factors for CVD**. *PloS One* (2017.0) **12** e0172941. DOI: 10.1371/journal.pone.0172941
15. Amadi CE, Lawal FO, Mbakwem AC, Ajuluchukwu JN, Oke DA. **Knowledge of cardiovascular disease riskfactors and practice of primary prevention of cardiovascular disease by CommunityPharmacists in Nigeria: a cross-sectional study**. *Int J Clin Pharm* (2018.0) **40** 1587-95. DOI: 10.1007/s11096-018-0744-3
16. Yeluri S, Gara H, Vanamali DR. **Assessment of Knowledge with Regard to Cardiovascular Disease Risk Factors among College Students Using Heart Disease Fact Questionnaire**. *J Evol Med Dent Sci* (2021.0) **10** 347-51
17. Maharani A, Sujarwoto, Praveen D, Oceandy D, Tampubolon G, Patel A, Musinguzi G. **Cardiovascular disease risk factor prevalence and estimated 10-year cardiovascular risk scores in Indonesia: The SMARThealth Extend study**. *PLOS ONE* (2019.0) **14** e0215219. DOI: 10.1371/journal.pone.0215219
18. Temu TM, Kirui N, Wanjalla C, Ndungu AM, Kamano JH, Inui TS. **Cardiovascular health knowledge and preventive practices in people living with HIV in Kenya**. *BMC Infect Dis* (2015.0) **15** 421. DOI: 10.1186/s12879-015-1157-8
19. Molan J, Ajeel N. **Knowledge and practice of adults attending primary health care centers regarding major coronary heart disease risk factors in Basrah city**. *MJBU* (2020.0) **38** 89-99
20. 20New Medical Center Inaugurated at Jimma University [Internet]. Ethiopian News. 2018 [cited 2021 Apr 20]. Available from: https://ethio.news/2018/12/08/new-medical-center-inaugurated-at-jimma-university/.
21. Demissie WR. **Prevalence of Cardiac Arrhythmias among Chronic Obstructive Pulmonary Disease Patients Admitted to Jimma University Medical Center**. *Biomed J Sci Tech Res* (2018.0) **10**
22. Yasin SM, Isa MR, Ismail N, Suddin LS, Saman MSA, Ismail Z. **The relationship between socioeconomic status and knowledge on cardiovascular risk factors and its preventive practices among urban dwellers in Selangor, Malaysia**. *Malays J Public Health Med* (2020.0) **20** 292-8
23. Bhattarai SP, Neupane P. **Knowledge and practices on prevention of coronary artery diseases in nepalese community: A cross-sectional from five different cities in Nepal**. *Nepal Heart J* (2018.0) **15** 33-6
24. Oyewole OE, Solabi A. **Knowledge of modifiable risk factors for cardiovascular diseases among university undergraduates in Ibadan, Nigeria**. *Eur J Clin Exp Med* (2020.0) **18** 195-205
25. Mohamed HS. **Knowledge of Coronary Heart Disease and Risk Factors Among Type 2 Diabetes Mellitus at Outpatient Department of Kasr Alainy Hospital**. *Egypt J Hosp Med* (2021.0) **82** 785-91
26. Ibrahim M, Rahman N, Rahman N, Haque M. **Knowledge, Attitude and Practice of Malaysian Public University Students on Risk Factors for Cardiovascular Diseases**. *J Appl Pharm Sci* (2016.0) **6** 056-63
27. Saeed A, Gulati M, Wong ND, Amsterdam EA, Toth PP. *ASPC Manual of Preventive Cardiology* (2021.0) 441-61. DOI: 10.1007/978-3-030-56279-3_19
28. Truthmann J, Busch MA, Scheidt-Nave C, Mensink GBM, Gößwald A, Endres M. **Modifiable cardiovascular risk factors in adults aged 40–79 years in Germany with and without prior coronary heart disease or stroke**. *BMC Public Health* (2015.0) **15** 701. DOI: 10.1186/s12889-015-1929-5
29. Awosan K, Sabir A, Ibrahim M, Ibrahim T, Bello-Baura M, Ibrahim Z. **Knowledge of Coronary Heart Disease and Practices on Detection and Management of Its Risk Factors among Primary Health Care Workers in Sokoto, Nigeria**. *J Adv Med Med Res* (2017.0) **24** 1-10
30. Muhihi AJ, Anaeli A, Mpembeni RNM, Sunguya BF, Leyna G, Kakoko D. **Public knowledge of risk factors and warning signs for cardiovascular disease among young and middle-aged adults in rural Tanzania**. *BMC Public Health* (2020.0) **20** 1832. DOI: 10.1186/s12889-020-09956-z
31. Omoronyia O, Ayuk A, Nwafor K, Legogie A. **Hypertensive patients’ knowledge of cardiovascular disease in Calabar, Nigeria**. *J Egypt Public Health Assoc* (2020.0) **95** 16. DOI: 10.1186/s42506-020-00045-y
32. Aminde LN, Takah N, Ngwasiri C, Noubiap JJ, Tindong M, Dzudie A. **Population awareness of cardiovascular disease and its risk factors in Buea, Cameroon**. *BMC Public Health* (2017.0) **17** 545. DOI: 10.1186/s12889-017-4477-3
33. Molan J, Ajeel N. **Knowledge and practice of adults attending primary health care centers regarding major coronary heart disease risk factors in Basrah city**. *MJBU* (2020.0) **38** 89-99
34. Odunaiya NA, Adesanya TB, Okoye EC, Oguntibeju OO. **Towards cardiovascular disease prevention in Nigeria: A mixed method study of how adolescents and young adults in a university setting perceive cardiovascular disease and risk factors**. *Afr J Prim Health Care Fam Med* (2021.0) **13**
35. Dele-Ojo B, Raimi T, Fadare J, Ajayi E, Ajayi D, Ojo O. **Knowledge and Prevalence of Heart Disease Risk Factors Among Staff of a Tertiary Institution in Nigeria**. *Int Q Community Health Educ* (2021.0) **41** 387-94. PMID: 33167795
36. Amro NR, Rasheed O, Khdour M, Qraqra D, Ghrayeb FA. *Knowledge of Cardiovascular Disease among Undergraduate University Students in Palestine* (2017.0) **7** 7
|
---
title: 'Relationships between circulating metabolites and facial skin aging: a Mendelian
randomization study'
authors:
- Zhengye Liu
- Jiarui Mi
- Huiling Wu
journal: Human Genomics
year: 2023
pmcid: PMC10022075
doi: 10.1186/s40246-023-00470-y
license: CC BY 4.0
---
# Relationships between circulating metabolites and facial skin aging: a Mendelian randomization study
## Abstract
### Background
Blood metabolites are important to various aspects of our health. However, currently, there is little evidence about the role of circulating metabolites in the process of skin aging.
### Objectives
To examine the potential effects of circulating metabolites on the process of skin aging.
### Method
In the primary analyses, we applied several MR methods to study the associations between 249 metabolites and facial skin aging risk. In the secondary analyses, we replicated the analyses with another array of datasets including 123 metabolites. MR Bayesian model averaging (MR-BMA) method was further used to prioritize the metabolites for the identification of predominant metabolites that are associated with skin aging.
### Results
In the primary analyses, only the unsaturation degree of fatty acids was found significantly associated with skin aging with the IVW method after multiple testing (odds ratio = 1.084, $95\%$ confidence interval = 1.049–1.120, $$p \leq 1.737$$ × 10−06). Additionally, 11 out of 17 unsaturation-related biomarkers showed a significant or suggestively significant causal effect [$p \leq 0.05$ and > 2 × 10−4 ($\frac{0.05}{249}$ metabolites)]. In the secondary analyses, seven metabolic biomarkers were found significantly associated with skin aging [$p \leq 4$ × 10−4 ($\frac{0.05}{123}$)], while six of them were related to the unsaturation degree. MR-BMA method validated that the unsaturation degree of fatty acids plays a dominant role in facial skin aging.
### Conclusions
Our study used systemic MR analyses and provided a comprehensive atlas for the associations between circulating metabolites and the risk of facial skin aging. Genetically proxied unsaturation degree of fatty acids was highlighted as a dominant factor correlated with the risk of facial skin aging.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s40246-023-00470-y.
## Introduction
The skin is the largest organ, covering the entire surface of the human body [1]. Exploring the mechanisms involved in skin aging, especially facial aging, has been an area of interest, not only for esthetic purposes but also because they may provide mechanistic insights into diseases with similar mechanisms [2, 3]. The process of skin aging is affected by circulating metabolites which are involved in a variety of cellular processes including cellular organization, post-translational modification as well as epigenetic changes [4, 5]. Some metabolic biomarkers, such as the levels of unsaturated lipids and polyunsaturated fatty acids (PUFAs), have been reported to play an important role in skin aging [6–10]. However, there are conflicting opinions from different studies [11]. For example, a recent study based on mouse models found that PUFA supplementation can protect mice from photoaging [9]. However, another study reported that PUFA supplementation may induce inflammation and extracellular matrix degradation [10]. There are several reasons that may lead to the contradictory results. Firstly, some studies have been performed on cultured cells or on mouse models, where the conditions may differ from the real-life skin aging process [9, 10]. Furthermore, randomized controlled trials are not an option due to ethical considerations and feasibility. Therefore, most of the studies are based on observational designs, in which the influence of confounding factors or reverse causality cannot be completely ruled out [12]. Finally, circulating metabolites can be divided into multiple subclasses, while different subclasses of metabolites may have distinct effects. Although various types of dietary intake have been studied in different aspects of tissue aging, the overall effect of the complex dietary components is difficult to imply the net effect of individual supplements [13]. Therefore, novel methodologies, which are unbiased by confounders, are needed for further exploration. Due to the increasing popularity of long-term oral supplementation of micronutrients in modern society, it appears as an urgent need for systematic epidemiological evaluation demonstrating the overall impact of metabolites on skin aging.
Mendelian randomization (MR) is an epidemiological method that employs genetic variants as instrumental variables to proxy an exposure variable of interest and study the effect of the exposure on a certain outcome [14]. Since single-nucleotide polymorphisms (SNPs) are assigned randomly at conception, they are unlikely to be affected by confounding factors [15]. The bias to reverse causality is also diminished as genetic variants cannot be affected by the development of the outcome traits [15]. Furthermore, the genetic instrumental variables reflect lifetime exposure, making MR an ideal tool to study aging-related topics. Nowadays, with the advent of high-throughput metabolomics, the levels of hundreds of circulating metabolites can be measured simultaneously. Several genome-wide association studies (GWASs) investigating associations between circulating metabolic biomarkers and SNPs were recently published [16–18]. In this study, by integrating the largest human genomic datasets to date, we employed several MR methods to 1) estimate the causal effects of circulating metabolites on the risk of skin aging and 2) prioritize the metabolites that promote skin aging after adjusting for the effects of similar ones.
## Study design
In this study, we have used instrumental variables obtained from two different metabolomics quantitative trait loci studies on circulating metabolites for primary and secondary analyses, respectively, to study the roles of plasma metabolites on skin aging. For the primary analyses, the summary-level GWAS datasets of 249 circulating metabolites that were divided into nine major categories were obtained from UK Biobank (unpublished, accessible via MRC IEU OpenGWAS database). Skin aging-related GWAS datasets were obtained from a recent publication by Roberts V. et al. [ 3]. For the secondary analyses, summary-level statistics on 123 circulating metabolites were obtained from Kettunen et al. [ 18] and two-sample MR analyses were performed to further validate our findings. Considering that the metabolites in the same subcategory were highly correlated, we performed an MR Bayesian model averaging (MR-BMA) analysis to prioritize the effect of major metabolites [19]. Only individuals of European ancestry were included in the analyses. Written informed consent and approval from the local ethical committee were obtained by all included studies.
## Metabolic profile for primary analyses
Summary-level datasets on 249 circulating metabolites used in primary analysis were obtained from Nightingale Health Metabolic Biomarkers Phase 1 release study in UK Biobank (June 2019–April 2020) (Table 1). This study included 115,078 randomly selected participants. Metabolic biomarkers were measured with non-fasting baseline EDTA plasma samples by high-throughput nuclear magnetic resonance (NMR) (https://biobank.ndph.ox.ac.uk/ukb/label.cgi?id=220). The biomarkers include 168 absolute metabolites (unit, mmol/L) and 81 metabolite ratios spanning multiple metabolic pathways such as lipoproteins, fatty acids, amino acids, and ketone bodies. The details of sample collection and NMR profiling have been depicted in previous publications [20–22].Table 1Detailed information of included data sourcesTraitsSample sizeYearPopulationPubMed IDWeb source249 Circulating metabolites (primary analyses)115,0782020EuropeanNAhttps://www.ukbiobank.ac.uk/123 Circulating metabolites (secondary analyses)24,9252016European27005,778http://www.computationalmedicine.fi/data/NMR_GWAS/Facial skin aging (perceived age)423,992 (8,630 reported looking older than their biological age, 103,300 reported looking about their age, and 312,062 reported looking younger)2020European32339537https://doi.org/10.5523/bris.21crwsnj4xwjm2g4qi8chathha BOLT-LMM (linear mixed model) was used to account for population structure, with further adjustment for age, sex, fasting status, and genotyping chips. Over 12.3 million SNPs were included for further analyses after adjusting for covariates and quality control.
## Metabolic profile for secondary analyses
Summary-level datasets on 123 circulating metabolites used in the secondary analysis were obtained from a previous publication by Kuttunen et al. [ 18] (Table 1). Metabolite concentrations were quantified with high-throughput NMR spectroscopy from 10 studies including 24,925 individuals of European ancestry. Datasets from different cohorts were analyzed separately with an additive model and then pooled together by a fixed-effect meta-analysis, with up to 12,133,295 SNPs included. All metabolite concentrations were adjusted for age, sex, time from the last meal, and ten first principal components.
## IV selection
SNPs associated with metabolite biomarkers were selected with a conventional genome-wide association significance threshold ($p \leq 5$ × 10−8). Linkage disequilibrium (LD) clumping was used to identify and exclude SNPs that were in LD (R2 > 0.001 or within ± 10,000 kilobase (kb) distance 1000 Genomes European-ancestry Reference Panel). Mean F-statistics were calculated to test for weak instruments as previously described [23].
## Facial skin aging
Summary statistics of skin aging were obtained from a previous GWAS based on UK Biobank [3] (Table 1). Eligible participants identified from health records in National Health Service were invited to participate in baseline assessments including questionnaires, physical measurement, biological samples collection, and follow-ups. The participants were asked the following questions in the questionnaires: “Do people say that you look…?” The possible answers were “Younger than you are,” “Older than you are,” “About your age,” “Do not know,” or “Prefer not to answer.” Participants that did not respond were excluded from subsequent analyses. After imputation and quality control, genome-wide analysis was performed with a linear mixed model using BOLT-LMM. Only individuals of European ancestry were included in the GWAS.
## Mendelian randomization
The inverse-variance weighted (IVW) was used as the main method for causal estimation. Wald ratios of individual SNPs’ effects on the outcome were combined with a fixed-effect IVW when IVs ≤ 3 or a random effect IVW when over 3 IVs were included. Heterogeneities of the IVW analyses were estimated with Cochran’s Q values, I2, and the H-statistics [24, 25]. We further performed MR-Egger, weighted median as sensitivity analyses [26–28]. MR-*Egger is* a method that can give valid causal estimates even with the existence of pleiotropy (p for intercept < 0.05), as it detects and corrects for potential horizontal pleiotropy [26]. The weighted median is a method that can be used to strengthen the causal estimates when up to fifty percent of the weight in the MR analyses came from invalid instrument variables [27]. Multivariable Mendelian randomization (MVMR) is a method that estimates the direct effect of different exposures on the outcome after adjusting for the effects of other exposures. In this study, we have also used the MVMR method to estimate the causal associations of candidate metabolites/ratio index on the risk of facial skin aging adjusting for several common risk factors of aging including BMI, smoking behavior (cigarettes per day), and alcohol drinking (alcoholic drinks per week) [29].
## Colocalization analysis
We have also performed a colocalization analysis between the degree of unsaturation and facial skin aging with HyprColoc (R package hyprcoloc: https://rdrr.io/github/jrs95/hyprcoloc/) [30]. The default prior probability that an SNP is causal to one trait was 1 × 10−4. If the posterior probability of one SNP being shared between the two traits in one region was greater than 0.8, we regarded it as a signal of colocalization.
## MR Bayesian model averaging (MR-BMA)
As many metabolic traits involved in the study are highly correlated based on sharing a substantial number of SNPs, it appears necessary to correct for the effects of “measured pleiotropy.” Here, we used the MR-BMA to discover the metabolic biomarkers that play predominant roles in the causal associations with skin aging, from a group of related factors [19]. Compared with conventional multivariable MR methods, the MR-BMA method is useful in disentangling the correlated metabolic biomarkers which may act via the same causal pathway [31]. In this study, we followed up the results from primary analyses, by assessing the causal effects of unsaturation-related biomarkers on skin aging with MR-BMA. SNPs associated with all selected biomarkers were pooled and then strictly clumped to exclude SNPs in LD (R2 < 0.001 in 10,000 kb distance in 1000 Genomes European-ancestry Reference Panel). Posterior probability (PP) was calculated for all specific models (i.e., one biomarker or a combination of multiple biomarkers). The marginal inclusion probability (MIP) for each biomarker, which is the sum of the PP over all models where the biomarker is present, was used to rank the causal associations of the traits with the outcome. We also calculated model-averaged causal effects (MACE), which demonstrates the direct causal effect of a biomarker on skin aging averaged across all related models. Cook’s distance was used to identify outliers in the MR-BMA analyses.
## Statistical analyses
All statistical analyses in this study are two-sided. For primary analyses, a p-value < 2 × 10−4 ($\frac{0.05}{249}$, Bonferroni adjusted) was considered statistically significant, and a p-value between 0.05 and 2 × 10−4 was considered suggestively significant. For secondary analyses, a p-value < 4 × 10−4 ($\frac{0.05}{123}$, Bonferroni adjusted) was considered statistically significant, and a p-value between 0.05 and 4 × 10−4 was considered suggestively significant. All the analyses were performed on R platform (version 4.1.0), with “TwoSampleMR” (0.5.5), “Mendelian randomization” (0.5.0), “MVMR,” “HyprColoc,” and “ggplot2” packages [28–30, 32, 33].
## Primary analyses
By assessing the 249 metabolic biomarkers’ effect on skin aging with univariable MR analyses, only the unsaturation degree of fatty acids was observed to have a significant causal effect on skin aging after Bonferroni adjustment (odds ratio [OR] = 1.084, $95\%$ confidence interval [CI] = 1.049–1.120, $$p \leq 1.737$$ × 10−06) (Fig. 1, Additional file 2: Table S1). The causal estimate remained consistent with sensitivity analyses (Additional file 2: Tables S2, S3). No horizontal pleiotropy was identified with the MR-Egger method (Additional file 2: Table S4).Fig. 1Volcano plot showing the causal estimates of 249 metabolic biomarkers on facial skin aging in the primary analyses with IVW method. IVW, inverse-variance weighted; VLDL, very-low-density lipoprotein; HDL, High-density lipoprotein Besides, 65 biomarkers were shown to have a suggestively causal effect on skin aging (Additional file 2: Table S1). To understand the relationships between different kinds of metabolic biomarkers and skin aging, we classified the 249 metabolic traits into nine major groups (Fig. 2, Additional file 1: Figs. S1–S8). Among different groups of biomarkers, we surprisingly found that unsaturation-related biomarkers showed a consistent association with skin aging, with 11 out of 17 unsaturation-related biomarkers showing a significant or suggestively significant causal effect (Fig. 2). Among them, the ratio of PUFA to total fatty acids (OR = 1.084, $95\%$ CI 1.022–1.151, $$p \leq 0.008$$), PUFA to monounsaturated fatty acids (MUFA) (OR = 1.072, $95\%$ CI 1.025–1.121, $$p \leq 0.002$$), n-3 PUFA to fatty acids (OR = 1.054, $95\%$ CI 1.014–1.100, $$p \leq 0.008$$), docosahexaenoic acid (DHA) to total fatty acids (OR = 1.086, $95\%$ CI 1.039–1.135, $$p \leq 2.679$$ × 10−04), n-3 PUFA (OR = 1.052, $95\%$ CI 1.017–1.088, $$p \leq 0.003$$), and DHA levels (OR = 1.066, $95\%$ CI 1.027–1.106, $$p \leq 8.103$$ × 10−04), as biomarkers known to be associated higher unsaturation degree, increased the risk of skin aging (Fig. 2). But the ratio of n-6 to n-3 PUFA (OR = 0.945, $95\%$ CI 0.912–0.979, $$p \leq 0.002$$), MUFA to total fatty acids (OR = 0.932, $95\%$ CI 0.895–0.970, $$p \leq 5.280$$), linoleic acid to total fatty acids (OR = 0.917, $95\%$ CI 0.869–0.968, $$p \leq 0.002$$), and MUFA levels (OR = 0.945, $95\%$ CI 0.908–0.984, $$p \leq 0.006$$) were negatively associated with the susceptibility to skin aging (Fig. 2).Fig. 2Heatmap showing the causal estimates of unsaturation-related traits on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. MR, Mendelian randomization; IVW, inverse-variance weighted Heatmaps of the causal associations are shown in Additional file 1: Figs. S1–S8. Notably, multiple triglyceride-related biomarkers showed an overall tendency to reduce the risk of skin aging; however, none of them remained significant in the sensitivity analyses (Additional file 1: Fig. S1).
Mean F-statistics of all metabolites were higher than 10, indicating a low risk of weak instrument bias. Heterogeneity and horizontal pleiotropy for all the analyses are presented in Additional file 2: Tables S4 and S5. Detailed information on used SNPs is provided in Additional file 2: Table S6.
To rule out the possibility that the skin aging process changes the levels of candidate metabolite or saturation degree, we also performed a reverse MR assessing the causal effects of facial skin aging on the 249 metabolic biomarkers. We observed no significant effects of facial skin aging on any of the included biomarkers with the IVW method (Additional file 2: Table S15).
We further performed a colocalization analysis to test whether the degree of unsaturation colocalizes with facial skin aging, and we identified potential colocalization of the two traits at two regions. One candidate causal SNP rs13107325 is in region Chr4:102688709-103688709, in gene SLC39A8, with a posterior probability of 0.887 and regional probability of 0.897. The other candidate causal SNP rs174564 is in region Chr11:60953822-61953822, with a posterior probability of 0.7938 and regional probability of 1. Interestingly, rs174564 is in a protein-encoding gene FADS2 (fatty acid desaturase 2) which encodes an enzyme that regulates unsaturation of fatty acids by introducing double bonds between defined carbons of the fatty acyl chain.
We further used the MVMR method to estimate the causal associations of the degree of unsaturation on the risk of facial skin aging adjusting for common risk factors of aging including BMI, smoking behavior (cigarettes per day), and alcohol drinking (alcoholic drinks per week). All four exposures remained significantly causally associated with facial skin aging after adjusting for other factors (Additional file 2: Table S16).
## Secondary analyses
In the secondary analyses, we estimated the causal effects of 123 circulating metabolic biomarkers on the risk of skin aging with two-sample MR. Seven out of 123 biomarkers demonstrated statistical significance after adjusting for multiple testing (Fig. 3A). Interestingly, six of these seven biomarkers were associated with unsaturation degree of fatty acids (Fig. 3A). Specifically, biomarkers indicative of a higher unsaturation degree, including average number of double bonds in a fatty acid chain (OR = 1.073, $95\%$ CI 1.042–1.105, $$p \leq 2.99$$ × 10−06), the ratio of bis-allylic groups to double bonds (OR = 1.073, $95\%$ CI 1.042–1.105, $$p \leq 2.99$$ × 10−06), ratio of bis-allylic groups to total fatty acids (OR = 1.078, $95\%$ CI 1.042–1.115, $$p \leq 1.12$$ × 10−05), and other polyunsaturated fatty acids than 18:2 (OR = 1.053, $95\%$ CI 1.025–1.082, $$p \leq 1.82$$ × 10−04), significantly increased the risk of facial skin aging (Fig. 3A, B). On the contrary, biomarkers that lead to a reduced level of unsaturation, including the average number of methylene groups per double bond (OR = 0.916, $95\%$ CI 0.892–0.941, $$p \leq 1.77$$ × 10−10) and the average number of methylene groups in a fatty acid chain (OR = 0.888, $95\%$ CI 0.847–0.929, $$p \leq 4.13$$ × 10−07), were inversely correlated with skin aging predisposition (Fig. 3a, b). The significance of the unsaturation-related traits remained consistent in all the sensitivity analyses (Additional file 1: Fig. S9). Causal estimates from sensitivity analyses, horizontal pleiotropy, and heterogeneity are shown in Additional file 2: Tables S7–S11. Detailed information on all included IVs is presented in Additional file 2: Table S12.Fig. 3Causal effects of circulating metabolome on facial skin aging in the secondary analyses. A Volcano plot showing the causal estimates of 123 metabolic traits on facial skin aging in the secondary analyses with IVW method; B forest plots showing the causal estimates of seven metabolic traits that are significantly associated with facial skin aging in the secondary analyses with IVW, MR-Egger, and weighted median methods. MR, Mendelian randomization; IVW, inverse-variance weighted, No., number; SNP, single-nucleotide polymorphism; CI, confidence interval
## MR Bayesian model averaging
We further performed an MR Bayesian model averaging analysis with 17 unsaturation-related traits from the primary analyses. A total of 463 SNPs were identified as associated with the 17 biomarkers after removing duplicate SNPs. We then removed SNPs in LD and retained 214 SNPs for downstream analyses (Additional file 2: Table S13).
In the MR-BMA analyses, we selected the best models with the highest posterior probability (Additional file 2: Table S14). After that, the MIPs of all included metabolite biomarkers were calculated and used to rank the biomarkers for their causal associations with skin aging risk (Table 2). Degree of unsaturation was identified as the top-ranked biomarker that increases the risk of skin aging (MIP = 0.654, average effect = 0.056, $$p \leq 0.010$$). Besides, the ratio of PUFA to MUFA (MIP = 0.106, average effect = 0.008, $$p \leq 0.040$$) and MUFA percentage (MIP = 0.093, average effect = -0.006, $$p \leq 0.040$$) were also identified to be independently associated with skin aging (Table 2). No outliers were identified in the analyses by using Cook’s distance (Additional file 1: Figures S10–S12). We have used the Q-statistics for identifying outliers in MR-BMA; after removing outliers, degree of unsaturation remained the top-ranked biomarker associated with the risk of skin aging (MIP = 0.852, average effect = 0.073, $$p \leq 0.010$$) (Additional file 1: Fig. S13).Table 2Ranking of unsaturation-related metabolic biomarkers for the risk of skin aging using MR-BMAMetabolite biomarkersMR-base IDRanking by MIPMIPAverage effectp valueDegree of unsaturationmet-d-Unsaturation10.6540.0560.009901Ratio of polyunsaturated fatty acids to monounsaturated fatty acidsmet-d-PUFA_by_MUFA20.1060.0080.039604Ratio of monounsaturated fatty acids to total fatty acidsmet-d-MUFA_pct30.093− 0.0060.039604Ratio of polyunsaturated fatty acids to total fatty acidsmet-d-PUFA_pct40.0880.0070.09901Ratio of docosahexaenoic acid to total fatty acidsmet-d-DHA_pct50.0670.0020.366337Docosahexaenoic acidmet-d-DHA60.025− 0.0010.930693Ratio of omega-3 fatty acids to total fatty acidsmet-d-Omega_3_pct70.020.0020.930693Ratio of linoleic acid to total fatty acidsmet-d-LA_pct80.018− 0.0010.960396Ratio of omega-6 fatty acids to total fatty acidsmet-d-Omega_6_pct90.0170.0010.970297Ratio of omega-6 fatty acids to omega-3 fatty acidsmet-d-Omega_6_by_Omega_3100.0160.0010.950495Linoleic acidmet-d-LA110.013− 0.0020.980198Monounsaturated fatty acidsmet-d-MUFA120.012− 0.0010.039604Omega-6 fatty acidsmet-d-Omega_6130.0110.0010.980198Omega-3 fatty acidsmet-d-Omega_3140.01100.970297Saturated fatty acidsmet-d-SFA150.00800.990099Polyunsaturated fatty acidsmet-d-PUFA160.00700.990099Ratio of saturated fatty acids to total fatty acidsmet-d-SFA_pct170.00501MIP marginal inclusion probability, MR Mendelian randomization, MR-BMA MR based on Bayesian model averaging, PUFA polyunsaturated fatty acids, MUFA monounsaturated fatty acids, DHA docosahexaenoic acid, LA linoleic acid, SFA saturated fatty acids
## Discussion
The aging process may be influenced by various factors including intrinsic aging, environment, and lifestyle habits [34–37]. Nutritional factors have been shown to play an important role in maintaining the normal function of the skin [38]. However, the association between nutritional status and changes in skin appearance remains unclear. Our study, for the first time, comprehensively studied the individual causal effects of a broad range of circulating metabolic traits on the predisposition of skin aging. Our results highlighted the effect of the degree of unsaturation and several unsaturation-related metabolites on the risk of skin aging. We also observed that multiple triglyceride-related biomarkers showed a trend toward reduced skin aging risk. This further confirmed the robustness of our analyses as triglycerides are a major component of sebum, which is known to be important for moisturizing and protecting human skin [39].
There are limited publications on unsaturated fatty acids in skin aging. Some studies found an improved photoprotection with dietary supplementation of PUFAs [12, 40–42]. However, most of these studies were based on observational designs or short-term oral supplementation and only assessed the interaction of unsaturation with environmental risk factors for skin aging such as sunlight exposure. The long-term effects of lipid unsaturation degree on skin aging process are still not fully established. Besides, oral supplementation commonly contains several categories of unsaturated fatty acids, while fatty acids with different degrees of unsaturation may generate diverse effects on skin aging [9, 42]. Even though shown to be photoprotective, fat unsaturation has also been reported to be associated with aging by several studies [9, 43, 44]. A higher degree of fat unsaturation in tissue membrane promotes the aging process through free radical production and oxidative stress [45, 46]. This is consistent with our observations in this study that genetically proxied degree of unsaturation was positively correlated with the risk of facial skin aging as an independent risk factor (Fig. 1). We also observed that increased ratios of total PUFA, n-3 PUFA, and DHA tend to contribute to skin aging, while a higher ratio of n-6 to n-3 PUFA, and the ratio of linoleic acid, reduced the risk of skin aging (Fig. 1). Intriguingly, neither the absolute level nor the percentage of saturated fatty acids were associated with facial skin aging in any of the methods of analysis, suggesting that changes in the proportions of different kinds of unsaturated fatty acids with different degrees of unsaturation were more important factors for skin aging (Figs. 1, 2).
These results were validated in our secondary analyses with an independent dataset for circulating metabolites. As n-6 PUFAs have fewer double bonds than n-3 PUFAs (2–4 compared to 3–6 double bonds), we observed that a higher number of double bonds in fatty acids increased facial skin aging risk (Fig. 3). Furthermore, we also found that higher numbers and ratios of bis-allylic groups, which have been reported to determine cells’ susceptibility to free radical-mediated peroxidative events, added to the risk of skin aging (Fig. 3) [47]. DHA is a kind of n-3 PUFA that contains five bis-allylic positions. It is highly sensitive to radical oxidation and may lead to deleterious advanced lipid peroxidation end products (ALEs) [48, 49]. When ALEs are formed at toxic levels, they may disrupt the cellular membrane and cause DNA damage [50, 51]. In this study, we also observed that genetically proxied higher DHA percentage and absolute level had a causal effect on skin aging (Fig. 2). Metabolite biomarkers analyzed in this study include both absolute levels and their percentage in total fatty acids or ratios to other metabolites. Intriguingly, it appears that the percentages of these unsaturation-related metabolites tend to generate a more significant effect on skin aging than their absolute concentrations (Fig. 1, Table 2). Previous publications also suggested that dietary intake of PUFA should be below a ceiling percentage of total energy [52]. The evidence suggests that maintaining a rational proportion of dietary unsaturated fatty acids might be of great importance to prevent their adverse effects on human skin.
The effects of PUFA on aging and obesity have also been intensively studied in mouse models. It has been shown that linoleic acid, compared with saturated fat, is more prone to induce obesity and insulin resistance and reduce motility [14962692, 22334255, 27886622]. To rule out the potential confounding factors and mediation effects, we also performed the MVMR to adjust for BMI and common lifestyle habits that could lead to metabolic dysregulation. Notably, our MVMR results further confirmed our MR findings, indicating that the unsaturation degree of fatty acid may be an independent risk factor in inducing facial aging.
There are several strengths and limitations of our study. To our knowledge, this is the first study employing an MR approach to assess the effects of individual circulating metabolites on skin aging. This is particularly important as results from observational studies or experimental mouse models are based on the aggregate effect of different dietary intake components [53]. Our results highlighted the importance of unsaturation degree in facial skin aging and provided a good reference for future studies. Besides, by using genetic variants as instrumental variables for the metabolic biomarkers, we minimized the bias from confounding factors and reverse causality. Also, the results remained consistent across the primary and replication analyses, which guaranteed the robustness of the findings. Lastly, the study population were refined to individuals of European ancestry to minimize bias from population stratification. However, this also restricted the generalizability of the conclusions, and it is necessary to validate the findings in other populations. Another restriction is that using genetic variants as proxies mimics a lifetime exposure, while oral supplementation for short period may generate a different effect. Finally, the metabolic biomarkers were measured in a non-fasting population, which can lead to inaccurate measurements. Nevertheless, the GWASs of the metabolites were adjusted for fasting time and found the alterations in the estimates were neglectable. Lastly, our MR study is unable to explore the cellular and molecular mechanisms underlying the effects of metabolites on facial aging. *The* gene GPR120 has been identified as the natural receptor of PUFA; we believed that further investigations using mutant mouse models (such as Gpr120 mutant mice), skin cell in vitro culture, single-cell RNA-seq, and proteomics experiments are warranted to reveal the detailed molecular events in skin tissue after the supplementation of PUFA.
In conclusion, our study provided evidence suggesting the unsaturation degree of circulating fatty acids as the predominant trait that is involved in the development of facial skin aging. Further studies are needed to investigate the role of long-term supplementation of unsaturated fatty acids in facial skin aging.
## Supplementary Information
Additional file1: Figure S1. Heatmap showing the causal estimates of triglycerides related traits on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S2. Heatmap showing the causal estimates of amino acids on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S3. Heatmap showing the causal estimates of cholesterol ester on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S4. Heatmap showing the causal estimates of free cholesterol on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S5. Heatmap showing the causal estimates of lipoprotein cholesterol on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S6: Heatmap showing the causal estimates of small metabolites on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S7. Heatmap showing the causal estimates of phospholipids on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S8. Heatmap showing the causal estimates of total lipids on facial skin aging in the primary analyses with IVW, MR-Egger, and weighted median methods. Figure S9. Heatmap showing the causal estimates of 123 metabolic traits on facial skin aging in the secondary analyses with IVW, MR-Egger, and weighted median methods. Figure S10. Dot plot of Cook’s distance for the causal effects of degree of unsaturation on facial skin aging with MR-BMA method. Figure S11. Dot plot of Cook’s distance for the causal effects of MUFA on facial skin aging with MR-BMA method. Figure S12. Dot plot of Cook’s distance for the causal effects of PUFA to MUFA ratio on facial skin aging with MR-BMA method. Figure S13. Dot plot of Q-statistics for the causal effects of degree of unsaturation on facial skin aging with MR-BMA method. Additional file 2: Table S1. Results with IVW method from primary analyses. Table S2. Results with MR-Egger method from primary analyses. Table S3. Results with weighted median method from primary analyses. Table S4. Measurements of horizontal pleiotropy in primary analyses. Table S5. Measurements of heterogeneity in primary analyses. Table S6. Detailed information of all SNPs in primary analyses. Table S7. Results with Wald ratios and IVW method from secondary analyses. Table S8. Results with MR-Egger method from secondary analyses. Table S9. Results with weighted median method from secondary analyses. Table S10. Measurements of horizontal pleiotropy in secondary analyses. Table S11. Measurements of heterogeneity in secondary analyses. Table S12: Detailed information of all SNPs in secondary analyses. Table S13. Detailed information of all SNPs in MR-BMA analyses. Table S14. Best models and corresponding posterior probabilities in MR-BMA analyses. Table S15. Causal effects of facial skin aging on metabolic traits from primary analyses with IVW method. Table S16. Direct effects of common risk factors on facial skin aging with Multivariable MR analyses.
## References
1. Lee SH, Jeong SK, Ahn SK. **An update of the defensive barrier function of skin**. *Yonsei Med J* (2006) **47** 293-306. DOI: 10.3349/ymj.2006.47.3.293
2. Williams HC. **Clinical practice**. *Atopic dermatitis N Engl J Med* (2005) **352** 2314-2324. DOI: 10.1056/NEJMcp042803
3. Roberts V, Main B, Timpson NJ, Haworth S. **Genome-wide association study identifies genetic associations with perceived age**. *J Invest Dermatol* (2020) **140** 2380-2385. DOI: 10.1016/j.jid.2020.03.970
4. Schmauck-Medina T, Moliere A, Lautrup S, Zhang J, Chlopicki S, Madsen HB. **New hallmarks of ageing: a 2022 Copenhagen ageing meeting summary**. *Aging* (2022) **14** 6829-6839. DOI: 10.18632/aging.204248
5. Lopez-Otin C, Blasco MA, Partridge L, Serrano M, Kroemer G. **The hallmarks of aging**. *Cell* (2013) **153** 1194-1217. DOI: 10.1016/j.cell.2013.05.039
6. Muzumdar S, Ferenczi K. **Nutrition and youthful skin**. *Clin Dermatol* (2021) **39** 796-808. DOI: 10.1016/j.clindermatol.2021.05.007
7. Rhodes LE, Durham BH, Fraser WD, Friedmann PS. **Dietary fish oil reduces basal and ultraviolet B-generated PGE2 levels in skin and increases the threshold to provocation of polymorphic light eruption**. *J Invest Dermatol* (1995) **105** 532-535. DOI: 10.1111/1523-1747.ep12323389
8. Kim HH, Cho S, Lee S, Kim KH, Cho KH, Eun HC. **Photoprotective and anti-skin-aging effects of eicosapentaenoic acid in human skin in vivo**. *J Lipid Res* (2006) **47** 921-930. DOI: 10.1194/jlr.M500420-JLR200
9. Wang P, Yan G, Xue H, Shen S, Cao Y, Zhang G. **Proteomics and lipidomics reveal the protective mechanism of dietary n-3 PUFA supplementation for photoaging**. *Food Funct* (2021) **12** 7883-7896. DOI: 10.1039/D0FO03228J
10. Gruber F, Ornelas CM, Karner S, Narzt MS, Nagelreiter IM, Gschwandtner M. **Nrf2 deficiency causes lipid oxidation, inflammation, and matrix-protease expression in DHA-supplemented and UVA-irradiated skin fibroblasts**. *Free Radic Biol Med* (2015) **88** 439-451. DOI: 10.1016/j.freeradbiomed.2015.05.006
11. Helal NA, Eassa HA, Amer AM, Eltokhy MA, Edafiogho I, Nounou MI. **Nutraceuticals' novel formulations: the good, the bad, the unknown and patents involved**. *Recent Pat Drug Deliv Formul* (2019) **13** 105-156. DOI: 10.2174/1872211313666190503112040
12. Latreille J, Kesse-Guyot E, Malvy D, Andreeva V, Galan P, Tschachler E. **Association between dietary intake of n-3 polyunsaturated fatty acids and severity of skin photoaging in a middle-aged Caucasian population**. *J Dermatol Sci* (2013) **72** 233-239. DOI: 10.1016/j.jdermsci.2013.07.006
13. Curtis R, Geesaman BJ, DiStefano PS. **Ageing and metabolism: drug discovery opportunities**. *Nat Rev Drug Discov* (2005) **4** 569-580. DOI: 10.1038/nrd1777
14. Davey Smith G, Hemani G. **Mendelian randomization: genetic anchors for causal inference in epidemiological studies**. *Hum Mol Genet* (2014) **23** R89-98. DOI: 10.1093/hmg/ddu328
15. Sanderson E, Smith GD, Windmeijer F, Bowden J. **Corrigendum to: an examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings**. *Int J Epidemiol* (2020) **49** 1057. DOI: 10.1093/ije/dyaa101
16. Shin SY, Fauman EB, Petersen AK, Krumsiek J, Santos R, Huang J. **An atlas of genetic influences on human blood metabolites**. *Nat Genet* (2014) **46** 543-550. DOI: 10.1038/ng.2982
17. Roederer M, Quaye L, Mangino M, Beddall MH, Mahnke Y, Chattopadhyay P. **The genetic architecture of the human immune system: a bioresource for autoimmunity and disease pathogenesis**. *Cell* (2015) **161** 387-403. DOI: 10.1016/j.cell.2015.02.046
18. Kettunen J, Demirkan A, Wurtz P, Draisma HH, Haller T, Rawal R. **Genome-wide study for circulating metabolites identifies 62 loci and reveals novel systemic effects of LPA**. *Nat Commun* (2016) **7** 11122. DOI: 10.1038/ncomms11122
19. Zuber V, Colijn JM, Klaver C, Burgess S. **Selecting likely causal risk factors from high-throughput experiments using multivariable Mendelian randomization**. *Nat Commun* (2020) **11** 29. DOI: 10.1038/s41467-019-13870-3
20. Mi J, Jiang L, Liu Z, Wu X, Zhao N, Wang Y. **Identification of blood metabolites linked to the risk of cholelithiasis: a comprehensive Mendelian randomization study**. *Hepatol Int* (2022). DOI: 10.1007/s12072-022-10360-5
21. Ahola-Olli AV, Mustelin L, Kalimeri M, Kettunen J, Jokelainen J, Auvinen J. **Circulating metabolites and the risk of type 2 diabetes: a prospective study of 11,896 young adults from four Finnish cohorts**. *Diabetologia* (2019) **62** 2298-2309. DOI: 10.1007/s00125-019-05001-w
22. Soininen P, Kangas AJ, Wurtz P, Suna T, Ala-Korpela M. **Quantitative serum nuclear magnetic resonance metabolomics in cardiovascular epidemiology and genetics**. *Circ Cardiovasc Genet* (2015) **8** 192-206. DOI: 10.1161/CIRCGENETICS.114.000216
23. Zhou J, Mi J, Peng Y, Han H, Liu Z. **Causal associations of obesity with the intervertebral degeneration, low back pain, and sciatica: a two-sample Mendelian randomization study**. *Front Endocrinol* (2021) **12** 740200. DOI: 10.3389/fendo.2021.740200
24. Higgins JP, Thompson SG, Deeks JJ, Altman DG. **Measuring inconsistency in meta-analyses**. *BMJ* (2003) **327** 557-560. DOI: 10.1136/bmj.327.7414.557
25. Bowden J, Del Greco MF, Minelli C, Davey Smith G, Sheehan NA, Thompson JR. **Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic**. *Int J Epidemiol* (2016) **45** 1961-1974. PMID: 27616674
26. Burgess S, Thompson SG. **Interpreting findings from Mendelian randomization using the MR-Egger method**. *Eur J Epidemiol* (2017) **32** 377-389. DOI: 10.1007/s10654-017-0255-x
27. Bowden J, Davey Smith G, Haycock PC, Burgess S. **Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator**. *Genet Epidemiol* (2016) **40** 304-314. DOI: 10.1002/gepi.21965
28. Verbanck M, Chen CY, Neale B, Do R. **Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases**. *Nat Genet* (2018) **50** 693-698. DOI: 10.1038/s41588-018-0099-7
29. Sanderson E, Spiller W, Bowden J. **Testing and correcting for weak and pleiotropic instruments in two-sample multivariable Mendelian randomization**. *Stat Med* (2021) **40** 5434-5452. DOI: 10.1002/sim.9133
30. Foley CN, Staley JR, Breen PG, Sun BB, Kirk PDW, Burgess S. **A fast and efficient colocalization algorithm for identifying shared genetic risk factors across multiple traits**. *Nat Commun* (2021) **12** 764. DOI: 10.1038/s41467-020-20885-8
31. Lord J, Jermy B, Green R, Wong A, Xu J, Legido-Quigley C. **Mendelian randomization identifies blood metabolites previously linked to midlife cognition as causal candidates in Alzheimer's disease**. *Proc Natl Acad Sci USA* (2021). DOI: 10.1073/pnas.2009808118
32. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D. **The MR-Base platform supports systematic causal inference across the human phenome**. *Elife* (2018). DOI: 10.7554/eLife.34408
33. Yavorska OO, Burgess S. **MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data**. *Int J Epidemiol* (2017) **46** 1734-1739. DOI: 10.1093/ije/dyx034
34. Gilchrest BA. **Skin aging and photoaging: an overview**. *J Am Acad Dermatol* (1989) **21** 610-613. DOI: 10.1016/S0190-9622(89)70227-9
35. Gomez EC, Berman B. **The aging skin**. *Clin Geriatr Med* (1985) **1** 285-305. DOI: 10.1016/S0749-0690(18)30968-6
36. Cosgrove MC, Franco OH, Granger SP, Murray PG, Mayes AE. **Dietary nutrient intakes and skin-aging appearance among middle-aged American women**. *Am J Clin Nutr* (2007) **86** 1225-1231. DOI: 10.1093/ajcn/86.4.1225
37. Kennedy C, Bastiaens MT, Bajdik CD, Willemze R, Westendorp RG, Bouwes Bavinck JN. **Effect of smoking and sun on the aging skin**. *J Invest Dermatol* (2003) **120** 548-554. DOI: 10.1046/j.1523-1747.2003.12092.x
38. Dalgard F, Svensson A, Holm JO, Sundby J. **Self-reported skin morbidity among adults: associations with quality of life and general health in a Norwegian survey**. *J Investig Dermatol Symp Proc* (2004) **9** 120-125. DOI: 10.1046/j.1087-0024.2003.09111.x
39. Thody AJ, Shuster S. **Control and function of sebaceous glands**. *Physiol Rev* (1989) **69** 383-416. DOI: 10.1152/physrev.1989.69.2.383
40. Boelsma E, Hendriks HF, Roza L. **Nutritional skin care: health effects of micronutrients and fatty acids**. *Am J Clin Nutr* (2001) **73** 853-864. DOI: 10.1093/ajcn/73.5.853
41. Sies H, Stahl W. **Nutritional protection against skin damage from sunlight**. *Annu Rev Nutr* (2004) **24** 173-200. DOI: 10.1146/annurev.nutr.24.012003.132320
42. Pilkington SM, Watson RE, Nicolaou A, Rhodes LE. **Omega-3 polyunsaturated fatty acids: photoprotective macronutrients**. *Exp Dermatol* (2011) **20** 537-543. DOI: 10.1111/j.1600-0625.2011.01294.x
43. Hulbert AJ, Faulks SC, Buffenstein R. **Oxidation-resistant membrane phospholipids can explain longevity differences among the longest-living rodents and similarly-sized mice**. *J Gerontol A Biol Sci Med Sci* (2006) **61** 1009-1018. DOI: 10.1093/gerona/61.10.1009
44. Calhoon EA, Ro J, Williams JB. **Perspectives on the membrane fatty acid unsaturation/pacemaker hypotheses of metabolism and aging**. *Chem Phys Lipids* (2015) **191** 48-60. DOI: 10.1016/j.chemphyslip.2015.08.008
45. Barja G. **The mitochondrial free radical theory of aging**. *Prog Mol Biol Transl Sci* (2014) **127** 1-27. DOI: 10.1016/B978-0-12-394625-6.00001-5
46. Barja G. **Free radicals and aging**. *Trends Neurosci* (2004) **27** 595-600. DOI: 10.1016/j.tins.2004.07.005
47. Wagner BA, Buettner GR, Burns CP. **Free radical-mediated lipid peroxidation in cells: oxidizability is a function of cell lipid bis-allylic hydrogen content**. *Biochemistry* (1994) **33** 4449-4453. DOI: 10.1021/bi00181a003
48. Rosell M, Giera M, Brabet P, Shchepinov MS, Guichardant M, Durand T. **Bis-allylic deuterated DHA alleviates oxidative stress in retinal epithelial cells**. *Antioxidants* (2019) **8** 447. DOI: 10.3390/antiox8100447
49. Vistoli G, De Maddis D, Cipak A, Zarkovic N, Carini M, Aldini G. **Advanced glycoxidation and lipoxidation end products (AGEs and ALEs): an overview of their mechanisms of formation**. *Free Radic Res* (2013) **47** 3-27. DOI: 10.3109/10715762.2013.815348
50. Semchyshyn HM. **Reactive carbonyl species in vivo: generation and dual biological effects**. *Sci World J* (2014) **2014** 417842. DOI: 10.1155/2014/417842
51. Yin H, Xu L, Porter NA. **Free radical lipid peroxidation: mechanisms and analysis**. *Chem Rev* (2011) **111** 5944-5972. DOI: 10.1021/cr200084z
52. Eritsland J. **Safety considerations of polyunsaturated fatty acids**. *Am J Clin Nutr* (2000) **71** 197S-201S. DOI: 10.1093/ajcn/71.1.197S
53. Shannon OM, Ashor AW, Scialo F, Saretzki G, Martin-Ruiz C, Lara J. **Mediterranean diet and the hallmarks of ageing**. *Eur J Clin Nutr* (2021) **75** 1176-1192. DOI: 10.1038/s41430-020-00841-x
|
---
title: 'Oral status and affecting factors in Iranian ICU patients: a cross-sectional
study'
authors:
- Mostafa Arkia
- Jahangir Rezaei
- Nader Salari
- Siavash Vaziri
- Alireza Abdi
journal: BMC Oral Health
year: 2023
pmcid: PMC10022095
doi: 10.1186/s12903-023-02867-6
license: CC BY 4.0
---
# Oral status and affecting factors in Iranian ICU patients: a cross-sectional study
## Abstract
### Background
Oral care is crucial in intensive care units (ICUs). Meanwhile, this action is not well-performed, therefore, mouth cavity-associated disorders cause serious outcomes, e.g. ventilator-dependent pneumonia. Considering a lack of studies in Iran on this subject, this study aimed to determine the oral status and affected factors in ICU patients in Iran.
### Methods
In a cross-sectional study in 2019, we assessed the oral status of 138 patients admitted to the ICUs in the Kermanshah and Ilam provinces by census method. The tools were a demographic and clinical characteristics checklist, and Beck's oral status assessment scale (BOAS). The researcher investigated the condition of the patient's mouth, and their records. Data were analyzed using descriptive and inferential statistics.
### Results
In this study, the prevalence of moderate and severe disorders of the lips, gums and oral mucosa, tongue, teeth, and saliva were 14.4, 26.1, 16.6, 49.3, and 34.8 percent, respectively. Six percent of patients had a normal oral condition. Oral status had a significant relationship with education level, age, marital status, brushing teeth, NG tube, and consciousness level.
### Conclusion
Compared to other studies, the prevalence of oral cavity disorders in ICU patients of the Ilam and Kermanshah provinces was high. It mandates paying crucial attention to nurses' continued education, using standard guidelines, and applying new facilities. Moreover, it is mandated for periodical visits of patients by a dentist in ICUs.
## Background
In the intensive care unit (ICU), critically ill patients are continuously monitored and cared for while providing specialized services [1]. Statistics show that $62.2\%$ of ICU patients undergo complications during hospitalization [2]. They are suffering from oral problems caused by various factors e.g. malnutrition, the presence of endotracheal and stomach tubes, and reduced fluid intake. Studies have shown that inadequate oral care causes further difficulties in dryness of the oral mucosa, decrease in saliva flow, inflammation of the oral mucosa, formation of dental plaque, inflammation of the gums, and accumulation of pathogenic bacteria in the mouth and throat [3]. Bacteria present in dental plaques create heart infections, joint diseases, and ventilator-dependent pneumonia (VAP). As the most common and dangerous hospital infection in ICUs, VAP causes $50\%$ mortality [4].
The indecent condition of the mouth induces the natural flora of the mouth to change in favor of negative microorganisms. Therefore, in these patients, oral fibronectin decreases. Likewise, increasing the strength of bacteria to attach to the tongue and teeth, then creates oral lesions [5]. It has been proved that proper care of the mouth of patients hospitalized in ICUs can significantly reduce the number of dental plaques, gum diseases, and the occurrence of ventilator-dependent pneumonia [6], however, despite the importance of this topic, researchers have reported that oral care is not performed well in ICUs, procedures are not properly recorded, nursing knowledge in this field is not up-to-date, oral care training for nursing students in colleges is considered less important and there is no standard protocol in this field [3].
In Iran, oral care in patients hospitalized in ICUs is considered ancillary care, so in the study of Ranjber et al., it was found that $55.7\%$ of nurses believed that oral care is the work of the patient's assistant and the patient's family, $9\%$ of nurses believed in not acquiring enough training in the oral care field. Additionally, $69\%$ did not have enough time for oral care, $40\%$ considered oral care unpleasant, and $83.8\%$ of nurses wanted more training about oral care [6]. In Masoumi et al. 's study on the prevalence of oral disorders in patients admitted to the ICUs of a hospital in Zanjan-Iran, the prevalence of oral cavity disorder was $79.7\%$ on the third day and $90.54\%$ on the fourth day of admission [7]. In another study in ICUs of Birjand hospitals in Iran, $84.9\%$ of nurses were trained in oral care, however, oral care was ranked as the 10th priority of nursing care [8]. In the study of Rafael et al., which was conducted under the title of the prevalence of oral disorders in patients hospitalized in ICUs in Brazil, the prevalence of normal structures of the lips, tongue, gums, cheeks, floor of the mouth, and palate was 8.56, 85.4, 85.2, 97.2, 100 and $98.3\%$, respectively [9].
In developed countries, there are special evaluation systems, guidelines, and protocols for oral cavity evaluation, and it is considered basic care [10, 11]. Though, in Iran and other developing countries this measure is done arbitrarily, without any determined regulations [6].
The prevalence of oral disorders in ICU patients is different in various contexts, based on disparate factors [9]. Due to the lack of studies in Iran regarding oral cavity disorders in patients hospitalized in ICUs, the present study aimed for determining the rate of oral cavity disorders and affecting factors in ICU patients in the Kermanshah and Ilam provinces, west of Iran.
## Methods
We conducted a descriptive-analytical study from January to June 2019 in ICU wards of educational hospitals affiliated with Kermanshah and Ilam Universities of Medical Sciences. The sampling method was census. Herewith, we recruited all patients (with any condition) who met the inclusion criteria, 138 patients in total. People aged over 18 years, hospitalized in ICUs for at least 48 h, and absent of oral lesions or trauma once admitted were enrolled. We exclude the patients with incomplete information in their records and no consent to participate in the study. The tool had three elements; the demographic characteristics form including gender, job, educational level, marital status, age, income, and height; the clinical checklist consists of the person in charge of mouth care, using a toothbrush, materials used for mouth washing, using ointment, GCS, total duration of intubation, number of mouth wash per day, diet method, received food type, and having NG-tube; and the standard scale for examining oral cavity status entitled Beck Oral Assessment Scale (BOAS) [7].
Beck Oral Assessment Scale has five sub-scales that evaluate lips, mucous and gums, teeth, saliva, and tongue, and each is graded in four points from 1 to 4. The overall score is between 5 and 20. The highest the score, the more severe of oral disorder. Likewise, a score of 5 indicates no disorder, 6–10 mild disorder, 11–15 moderate disorder, and 16–20 severe disorder [7]. It was designed and validated by Beck in 1974 [12]. Figure 1 shows the components of BOAS, which is brought from the study of Nguh [2016] [12].Fig. 1The components of Beck Oral Assessment Scale Safarabadi et al. [ 2017] confirmed the validity and reliability of the BOAS tool, in which content validity was supported by three anesthesiologists, two neurosurgeons, and five expert nurses working in ICUs. The test–retest method has been used to determine its reliability by evaluating twenty ICU patients by two observers separately, and a correlation coefficient of 0.92 was obtained [13]. For the demographic form and the clinical checklist, ten academic members and experts in the ICU area approved the content validity of the checklists, and their opinions were considered.
Data collection was begun after obtaining permission from the vice-chancellor of research and technology of Ilam and Kermanshah Universities of Medical Sciences and presenting it to the officials of the selected hospitals, which were in Kermanshah (Imam Reza, Imam Ali, Taleghani, Imam Khomeini and Farabi hospitals) and Ilam (Imam Khomeini and Shahid Mustafa Khomeini hospitals) cities. After introducing and commenting on the research objectives and obtaining informed consent from the research samples or their legal guardians, the first researcher collected the data. In this way, at first, the ICU clothes were worn, and after entering the ward, the patient's file was checked, and if the patient met the inclusion criteria, clinical and demographic information was recorded. After that, the researcher checked the oral condition after washing his hands for 30 s with soap and water and wearing disposable gloves, a mask, and glasses. Herewith, first, the lips were observed and then the patient's mouth opened slowly. With the use of a tongue depressor stick and a flashlight, the patient's mouth status was observed and recorded in the questionnaire.
Data were analyzed using descriptive statistics and non-parametric tests such as Mann–Whitney U and Kruskal–Wallis and Spearman's correlation by SPSS-25 software. The significant level of all tests was less than 0.05.
## Results
In this study, 138 ICU patients were recruited, $67.4\%$ of them were male, $20\%$ were unemployed, $25.4\%$ illiterate, and $85.5\%$ lived in the city (Table 1). The mean and SD of age, income, and height were 57.68 ± 18.44 years, 15.50 ± 12.30 million Rial (Iran currency), and 169.18 ± 8.24 cm, respectively. In six percent of patients, the general status of the mouth was normal and $94\%$ had at least one problem in one of the components. Moderate and severe disorders of the lip, gum, and oral mucosa structures, teeth, and saliva were 14.4, 26.1, 16.6, 49.3 and $34.8\%$, respectively (Table 2).Table 1Frequency and frequency percentage of demographic variablesVariableFrequency(%)JobSelf-employment43(%31) Employed6(%4.3) Retired29(%21) Unemployed28(%20) Other32(%23.2) Total138(%100)Location Rural20(%14.5) Urban118(%85.5) Total138(%100)Gender Male93(%67.4) Female45(%32.6) Total138(%100)*Marital status* Single15(%10.9) Married110(%79.7) Widow/divorced13(%9.4) Total138(%100)Education level Illiterate35(%25.4) Under diploma61(%44.2) Diploma37(%26.8) Academic5(%3.6) Total138(%100)Person in charge of mouth care Nurse120 (87.0) Nurse assistance4 (2.9) Other14 (10.1)Tooth brushing Yes23 (16.7) No115 (83.3)Material used for mouth washing Chlorhexidine111 (80.5) Normal saline18 [13] Other9 (6.5)Use of emollient ointment Yes79 (57.2) No59 (42.8)Table 2Descriptive data on the prevalence of oral statusVariableLevel of disorderFrequency(%)Lipsnormal64(%46.4)mild54(%39.1)Moderate10(%7.2)sever10(%7.2)total138(%100)Gum and oral mucosanormal49(%35.5)mild53(%38.4)Moderate31(%32.5)sever5(%3.6)total138(%100)Languagenormal54(%39.1)mild61(%44.2)Moderate21(%15.2)sever2(%1.4)total138(%100)Teethnormal19(%13.8)mild41(%29.7)Moderate57(%41.3)sever11(%8)total138(%100)Salivanormal63(%45.7)mild27(%19.6)Moderate44(%31.9)sever4(%2.9)total138(%100)Overall oral statusnormal8(%5.8)mild64(%46.4)Moderate50(%36.2)sever7(%5.1)total138(%100) Inferential statistics revealed that people who either had a university education ($p \leq 0.01$), or married patients ($$p \leq 0.03$$), had better oral condition, however, the problem was worse in older patients ($$p \leq 0.009$$). Brushing teeth ($p \leq 0.001$) in nursing care was accompanied by a lower BOAS score. By reducing the level of consciousness the oral status would be more compromised ($$p \leq 0.039$$). No significant relationship was found between the number of daily mouth washing, gender, duration of intubation, type of mouthwash, and the person in charge of mouth washing with the overall score of mouth status (Tables 3, 4 and 5). The overall oral status score was higher in patients with a Naso-gastro tube (NG tube)(Tables 6 and 7). Table 3Relationship of oral status variables with qualitative demographic and clinical variablesOral status variablesSaliva status scoretongue status scoreTeeth statusDemographic and clinical variablesMean(SD)Mean rankMean(SD)Mean rankMean(SD)Mean rankGenderMale1.94(0.95)70.411.84(0.75)71.902.29(1.03)69.21female1.86(0.91)67.421.69(0.73)64.542.29(1.05)70.1Statistical analysis(Mann–Whitney U test)Z = -0.41p = 0.68Z = -1.09p = 0.27Z = -0.12P = 0.89JobSelf-employment1.88(0.85)68.852.3(1.103)80.092.3(1.1)69.81employed2.16(0.98)79.752(0.89)56.252(0.89)56.33retired2.1($\frac{1}{11}$)75.222.55(0.73)66.122.55(0.79)78.36unemployed1.53(0.83)54.072.18(1.18)67.682.18(1.18)65.3other2.09(0.93)76.772.19(1.09)62.412.19(1.09)67.19Statistical analysis(Kruskal–Wallis test)K2 = 7.19P = 0.12K2 = 5.82p = 0.21K2 = 5.82p = 0.21Locationrural2(0.85)73.681.75(0.73)68.152.2(1.15)68.03urban1.9(0.96)68.791.8(0.75)69.732.31(1.02)69.75Statistical analysis(Mann–Whitney U test)Z = -0.54p = 0.58Z = -0.17p = 0.85Z = -0.18p = 0.85Marital statussingle1.4(0.63)49.371.93(1.28)58.81.93(1.28)54.09married1.95(0.96)70.742.3[1]71.752.3[1]70.02Widow/divorced2.23(0.92)82.272.62(1.04)61.622.62(1.04)81.96Statistical analysis(Kruskal–Wallis test)K2 = 6.05p = 0.04*K2 = 2.04p = 0.36K2 = 3.64p = 0.16Level of educationilliterate1.97(0.92)71.912.29(1.15)69.972.29(1.15)71.78Under diploma2.11(0.95)77.22.51(1.04)79.842.51(1.4)78.5diploma1.54(0.86)54.011.92(0.89)53.031.92(0.89)52.41Academic2[1]73.22.4(0.54)622.4(0.54)69.6Statistical analysis(Kruskal–Wallis test)K2 = 9.23p = 0.02*K2 = 12.39p = 0.006*K2 = 11.08p = 0.01*Person in charge of mouth carenurse1.93(0.83)68.641.76(0.85)70.519.94(3.27)66.01Nurse assistance1.9(0.73)69.11.3(0.48)50.208.5(3.34)49.25other2.11(1.02)751.94(1.11)74.0610.12(4.33)64.38Statistical analysis(Kruskal–Wallis test)K2 = 0.44p = 0.8K2 = 3.13p = 0.2K2 = 1.88p = 0.39Tooth brushingyes1.43(0.66)46.391.17(0.49)41.597.56(2.77)38.39no2(0.85)74.121.87(0.89)75.0810.35(3.35)70.22Statistical analysis(Mann–Whitney U test)Z = -3.22P < 0.001*Z = -4P < 0.001*Z = -3.78P < 0.001*Material used for mouth washingchlorhexidine1.95(0.88)32.691.86(0.91)73.9610.04(3.49)66.61Normal saline1.78(0.64)63.671.28(0.46)48.499(2.93)56.64other2.22(0.83)83.331.44(0.72)55.729.33(3.57)58.83Statistical analysis(Kruskal–Wallis test)K2 = 1.64p = 0.43K2 = 8.61p = -0.01*K2 = 1.33p = 0.52Use of emollient ointmentyes1.95(0.9)69.431.73(0.84)69.239.72(3.59)63.81no1.93(0.78)69.531.78(0.93)69.869.92(3.3)65.44Statistical analysis(Mann–Whitney U test)Z = -0.56p = 0.57Z = -0.09p = 0.99Z = -0.24p = 0.8* is significantTable 4Relationship of oral status variables with demographic and clinical variablesOral status variableGum and oral mucosa status scoreLips statusOverall oral statusDemographic and clinical variableMean(SD)Mean rankMean(SD)Mean rankMean(SD)Mean rankGenderMale1.95(0.85)69.771.66(0.78)66.199.84(3.29)64.66female1.93(0.86)68.931.96(1.02)76.449.85(3.69)64.18Statistical analysis(Mann–Whitney U test)Z = -0.12p = 0.9Z = -1.52p = 0.12Z = -0.06p = 0.94JobSelf-employee2.07(0.82)75.511.77(0.89)69.9710.28(3.33)69.49employee1.5(0.83)491.83(0.4)81.679(2.75)55.5retired2(0.92)71.761.59(0.78)62.6010.10(3.48)68.61unemployed1.93(0.97)67.431.68(1.02)62.469.14[4]54.59other1.81(0.69)65.031.94(0.87)799.85(3.05)65.93Statistical analysis(Kruskal–Wallis test)K2 = 3.51p = 0.47K2 = 4.88p = 0.3K2 = 3.25p = 0.51Locationrural2(0.85)72.102(0.91)20.8110.58(3.6)72.12urban1.93(0.85)69.061.71(0.86)$\frac{53}{679.73}$(3.39)62.33Statistical analysis(Mann–Whitney U test)Z = -0.33p = 0.73Z = -1.54p = 0.12Z = -0.91p = 0.36Marital statussingle1.8(0.94)62.80.91(0.21)57.108(3.21)43.07married1.96(0.84)70.570.83(0.08)69.058.96(3.37)65.65Widow/divorced1.92(0.86)68.151.09(0.03)87.5811.08(3.44)79.75Statistical analysis(Kruskal–Wallis test)K2 = 0.58p = 0.74K2 = 4.9p = 0.08K2 = 6.87p = 0.03*Level of educationilliterate2.06(0.87)74.761.77(0.8)73.1310.46(2.83)73.15Under diploma2.11(0.87)77.131.93(0.99)75.8110.83(3.75)74.4diploma1.54(0.67)51.841.49(0.69)58.427.91(2.56)44.88academic2[1]72.401.4(0.54)56.109.4(2.96)61.7Statistical analysis(Kruskal–Wallis test)K2 = 11.27p = 0.01*K2 = 6.04p = 0.1K2 = 18.42P < 0.03*Person in charge of mouth carenurse1.93(0.83)68.641.84(0.85)70.519.94(3.27)66.01Nurse assistance1.9(0.73)69.101.3(0.48)50.209.94(3.27)44.25other2.11(1.02)751.94(1.11)74.0610.12(4.33)64.38Statistical analysis(Kruskal–Wallis test)K2 = 0.44p = 0.8K2 = 3.13p = 0.2K2 = 1.88p = 0.39Tooth brushingyes1.43(0.66)46.391.17(0.49)41.597.56(2.77)38.39no2(0.85)74.121.87(0.89)75.0810.35(3.35)70.22Statistical analysis(Mann–Whitney U test)Z = -3.22p = 0.001*Z = -4P < 0.001*Z = -3.78P < 0.001*Material used for mouth washingchlorhexidine1.95(0.88)32.691.86(0.91)73.9610.04(3.49)66.41Normal saline1.78(0.64)63.671.28(0.46)48.499(2.93)56.64other2.22(0.83)83.131.44(0.72)55.729.33(3.57)58.83Statistical analysis(Kruskal–Wallis test)K2 = 1.64p = 0.43K2 = 8.61p = 0.01*K2 = 1.3p = $\frac{0}{52}$Use of emollient ointmentyes1.95(0.9)69.431.73(0.84)69.239.79(3.52)63.81no1.93(0.78)69.531.78(0.93)69.869.92(3.3)65.44Statistical analysis(Mann–Whitney U test)Z = -0.56p = 0.57Z = -0.09p = 0.99Z = -0.24p = 0.8* is significantTable 5Correlation between oral variables with quantitative demographic and clinical variables by Spearman's correlationOral position variableSaliva statusTeeth statustongue statusGum and oral mucosa statusLips statusOverall oral statuseDemographic and clinical variableAge(years)Mean (57.68)$r = 0.184$$p \leq 0.031$*$r = 0.079$$p \leq 0.0355$r = 0.08p = 0.352r = 0.059p = 0.488r = 0.143p = 0.095r = 0.229p = 0.009*Income(million toman)mean(1.55 toman)$r = 0.106$$p \leq 0.216$r = 0.074p = 0.375r = -0.005p = 0.953r = 0.006p = 0.948r = -0.076p = 0.374r = 0.059p = 0.507Height(centimeter)Mean (169.18)$r = 0.109$$p \leq 0.203$r = 0.029p = 0.736r = 0.025p = 0.769r = -0.01p = 0.907r = -0.083p = 0.336r = 0.003p = 0.927GCS(The average of the last 72 horses)Mean (8.76)r = -0.258P = 0.002*$r = 0.118$*$$P \leq 0.169$$r = -0.176P = 0.039*r = -0.138P = 0.106r = -0.243P = 0.004*$r = 0.183$$P \leq 0.039$*The number of mouthwashes per dayMean (2.71)$r = 0.084$$P \leq 0.33$r = -0.002P = 0.984r = -0.049P = 0.569r = -0.061P = 0.478r = 0.091P = 0.29r = 0.013P = 0.882Total duration of intubation(day)Mean (7.06)$r = 0.019$$P \leq 0.822$r = -0.068P = 0.426r = 0.025P = 0.768r = -0.02P = 0.819r = -0.16P = 0.062r = 0.094P = 0.291* is significantTable 6Relationship of oral status variables with nutritional variablesOral status variablesSaliva statustongue statusTeeth statusClinical variablesMean(SD)Mean rankMean(SD)Mean rankMean(SD)Mean rankDiet Methodvein2.16(0.93)79.751.83(0.75)72.6711.83[1]57.13gavage1.93(0.91)70.491.84[73]72.032.27[1]68.73oral1.42(0.85)49.251.64[73]60.932.21(0.69)63.21other2(1.03)71.981.73[73]65.922.55(1.06)78.5Statistical analysis(Kruskal–Wallis test)K2 = 5.26p = 0.15K2 = 1.52p = 0.47K2 = 3.55P = 0.31Food typeIntralipid1.75(0.96)62.581.67(0.65)64.4622.25(0.96)68.29Homemade food1.66(1.07)57.881.75(0.74)67.881.92(1.16)58Food package1.98(0.89)72.521.91(0.8)75.042.29(0.99)70.02Hospital food1.5(0.84)52.61.5(0.7)54.62(0.94)54.2NPO2.08(1.01)75.411.78(0.75)69.092.51(1.12)77.78other1.91(0.9)70.081.67(0.65)64.462.25(1.05)67.04Statistical analysis(Kruskal–Wallis test)K2 = 4.95P = 0.42K2 = 3.35p = 0.64K2 = 4.56p = 0.47NG-Tubeyes2.01(0.44)73.411.82(0.74)70.942.23(1.07)71.23No1.27(0.66)43.441.62(0.74)59.852.06(0.8)58Statistical analysis(Mann–Whitney U test)Z = -3.18*$$p \leq 0.001$$Z = -1.18p = 0.23Z = -1.38p = 0.16* is significantTable 7Relationship of oral status variables with nutritional variablesOral status variablesGum and oral mucosa statusLips statusOverall oral statusClinical variablesMean(SD)Mean rankMean(SD)Mean rankMean(SD)Mean rankDiet Methodvein2(0.95)72.252.08(0.79)87.839.88(2.84)65.44gavage1.94(0.86)68.941.85(0.93)73.0710.13(3.55)67.72oral1.93(0.61)71.071.21(0.42)45.148.35(2.56)84.96other1.94(0.89)69.181.64(0.82)64.629.84(3.53)63.69Statistical analysis(Kruskal–Wallis test)K2 = 0.1p = 0.99K2 = 10.52p = 0.01*K2 = 3.06P = 0.38Food typeIntralipid11.83(0.83)65.251.58(0.51)66.929.36(2.97)59.91Homemade food1.83(0.57)66.751.58(0.66)64.679.5(3.53)60.90Food package1.98(0.82)71.581.93(0.97)75.9710.37(3.43)70.81Hospital food1.7(0.94)56.51.2(0.42)44.307.9(3.1)41.70NPO2.03(0.92)73.081.74(0.89)69.539.97(3.55)65.29other1.92(0.99)66.751.75(0.96)68.179.63(3.5)61.32Statistical analysis(Kruskal–Wallis test)K2 = 1.97P = 0.85K2 = 6.73p = 0.24K2 = 5.67p = 0.33NG-Tubeyes1.97(0.87)70.381.82(0.87)72.7310.15(3.44)67.77No1.78(0.64)63.671.33(0.86)47.948(2.7)44.50Statistical analysis(Mann–Whitney U test)Z = -0.7p = 0.48Z = 2.67p = 0.007*Z = -2.48p = 0.001** is significant
## Discussion
The results of this study showed that $14.4\%$ of ICU patients had moderate and severe lip disorder, which is in line with the study conducted by Dakrose et al. in 2014, in which 72 h after admission, $17\%$ of ICU patients had lip ulcers [14]. Lip disorders in ICU patients are affected by the presence of a tracheal tube, tracheal tube fixers, decreasing consciousness level, fever, and dehydration [14, 7]. Moreover, insufficient attention to lip care, improper fixing of the tracheal tube, and inappropriate nursing checklist to assess the lip condition are other reasons.
According to the results of this study, $26.1\%$ of patients had moderate and severe gums and oral mucosa disorders. The results were consistent with the study of Kima et al. in 2019 in South Korea, in which the prevalence of mucosal ulcers in the lower, middle, and upper parts of the mouth was $36.3\%$, $11.5\%$, and $7.1\%$, respectively [15]. The almost high prevalence of gum and oral mucosa lesions in these patients is due to the presence of underlying conditions such as diabetes, endotracheal tube, tracheal tube stabilizers, airway, decreased consciousness, use of sedative drugs, the level of hemoglobin, hematocrit, and low plasma proteins [16]. In addition to the above-mentioned notes, this prevalence could be associated with the continuous opening of the patients' mouths and the remaining dryness of oral mucosa.
According to the results of this study, $16.6\%$ of these patients had moderate and severe tongue disorders. In D'Cruz et al. study in 2014, 72 h after admission to the ICU, $82\%$ of patients had a visible coating on more than $70\%$ of their tongue [14]. Furthermore, this event was intense in non-ICU patients revealed in a descriptive-analytical study by Molania et al. [ 2018] by examining oral problems in patients referred to behavioral disease counseling centers in Sari city of Iran, and oral lesions prevalence was seen in $96\%$, and the most common lesions were related to the tongue ($80\%$) [17]. This variation may be related to the difference in medications and nursing care.
In this study, $49.3\%$ of patients had moderate and severe teeth disorders. Masoumi et al. [ 2015] found the number of oral lesions on the fourth day was $90.54\%$ in ICU patients, and the number of teeth disorders was high and had a direct relationship with VAP [7]. Teeth disorders in ICU patients are disturbances of normal oral flora in these patients. Within 48 h after a person's admission to the hospital, flora changes in favor of Gram-negative organisms with greater pathogenicity. These changes cause the accumulation of bacteria and the proliferation of opportunistic pathogens in the oral cavity and cause local and general complications e.g. stomatitis, tooth decay, infection of the tissues around the tooth, followed by the systematic spread of infection, bacteremia, and respiratory infections such as pneumonia. The infection also affects the joints and the heart [14–20]. In addition, the prevalence of tooth disorders in patients hospitalized in ICUs can be due to nurses' fear of tracheal tubes getting stuck while brushing, and the lack of nurses' training to brush these patients' teeth, which needs more studies.
According to the results of this study, $34.8\%$ of patients had moderate and severe saliva disorders. The causes of salivary disorders include atrophy of salivary glands, use of drugs such as antidepressants, underlying medical conditions, intubation, and the advanced age of most ICU patients [20]. In addition to the above, failure to use mouthwash in the standard way and the lack of proper use of nebulizers for susceptible patients would be considered other sources of saliva disorders.
According to the results of this study, only $5.8\%$ of these patients had normal oral conditions. Inconsistent with the study of Rafael et al. who examined the mouths of patients hospitalized in the ICU department, the results of the oral and dental evaluation showed that the normality of different classes of mouth and palate was $100\%$ and $98.3\%$, respectively, and $82\%$ of patients did not have any bleeding from the gums [9]. However, other studies showed that patients hospitalized in the ICU have disturbed oral conditions [15, 16]. The high normal condition may be due to the presence of a dental surgeon in the ICUs [9]. Omer et al. [ 2015] checked the awareness of ICU nurses about the oral care of these patients, it showed that $97.4\%$ of them considered this care important [21]. While in a study conducted in Iran, $55.7\%$ of ICU nurses believed that taking care of the mouth is the job of the nursing assistants and the patient's family [6]. In addition to a lack of knowledge and insufficient training, the high prevalence of oral cavity disorders can be due to the lack of oral and dental specialists in ICUs.
Based on the results of this study, the oral condition was significantly worse in widow patients. The results were inconsistent with the study of Rashidi et al. [ 2013], which was no significant relationship between oral condition and marital status [22]. Akbari et al. [ 2014] investigated the oral and dental treatment needs of drug abusers, and the condition of the mouth of divorced people was worse than single or married people [23].
Based on our results, higher education has a positive connection with oral conditions among ICU patients. Ardakani et al. [ 2013] found education level would play an essential role in predicting the oral and dental health of people. They emphasized that these variables can be changed through training [24]. The possible reason for the worsening of the oral condition with the decrease in the level of education in ICU patients can be connected to less desire and awareness of patients for cooperating with nurses during mouth washing and hygiene.
In this study, older patients had worse mouth conditions. According to the results of Kosari et al. 's study, there was no significant relationship between the indicators related to the condition of the mouth with age [25]. The disparity could be attributed to the difference in population. *In* general, elderly patients are exposed to various diseases due to physiological and pathological changes; weakness of the immune system, chronic diseases, and the use of many drugs. Moreover, the use of artificial teeth is another factor that affects the oral mucosa in these people [26, 27]. In addition to the mentioned reasons, the prevalence of oral disorders in elderly people in the ICUs can be due to a lack of education, and scientific knowledge of the nurses on specific care of elderly people.
According to the results of this study, the condition of the mouth was better in the patients who brushed their teeth. Shafipour et al. [ 2015] in a review study declared brushing teeth significantly reduces dental plaques, and the accumulation of oral microorganisms [28], which is in line with our study. Dental plaques are an important source for the growth of microorganisms that cause oral lesions, and brushing is a good way to remove these plaques [28].
According to the results of this study, the condition of the mouth was worse in patients with a low level of consciousness. Masoumi et al. showed there was a significant correlation between the decrease in the level of consciousness and tongue lesions in patients hospitalized in ICUs [7], which was the same as our study. In patients with a decreased level of consciousness due to lack of swallowing and lack of jaw movement, salivary secretions declined, and as a result, the tongue becomes dry.
In the patients of this study who used chlorhexidine to rinse their mouths, their lips were worse than those who applied normal saline. Consistent with our study, Rezaei et al., compared the effect of a mouthwash containing brush wood extract, aloe vera, and chlorhexidine on the gingival index in intubated patients. According to the results, chlorhexidine has side effects such as mucus color change and dryness of the mouth and lips [29].
Our study had some limitation, the first researcher needs to be an expert in assessing oral care by the BOAS scale for data collection, therefore, he was educated in ICU by the fourth and fifth authors in two sessions. Some records had incomplete information, we tried to take the information from the staff or families.
## Conclusion
The results of this study showed that the prevalence of oral cavity disorders in patients hospitalized in special care units in the Ilam and Kermanshah provinces was high, and only $5.8\%$ of these patients had the normal oral condition. In patients with less education, older age, unmarried patients, lower levels of consciousness, Ng-tube, and patients who have not brushed their teeth, the overall condition of the mouth was significantly worse. The results of this research can be used by practitioners, nursing education designers, and administrators to adjust the training program for nursing students. Moreover, it is crucial for periodical visits of patients by a dentist in ICUs. The nursing students also should be familiar with the concepts of oral care and disorders in ICUs. It is recommended to use the results of the present research for the prevention, treatment, and reduction of oral complications in ICUs.
## References
1. Kim S, Chan C, Oliner M, Escobar G. **ICU admission control. An empirical study of capacity allocation and its implication for patient outcomes**. *Manage Sci* (2014) **61** 19-38. DOI: 10.1287/mnsc.2014.2057
2. Blair E, Langdon K, McIntyre S, Lawrence D, Watson L. **Survival and mortality in cerebral palsy: observations to the sixth decade from a data linkage study of a total population register and National Death Index**. *BMC Neurol* (2019) **19** 111. DOI: 10.1186/s12883-019-1343-1
3. Adib-Hajbaghery M, Ansari A, Azizi-Fini I. **Oral care in ICU patients: a review of research evidence**. *Feyz* (2011) **15** 280
4. Shafiei Z, Babaee S, Nazari A. **Relationship between mood state and quality of life in patients undergoing coronary artery bypass graft surgery**. *Iran J Nurs* (2013) **26** 57-67
5. Abidia FR. **Oral care in the intensive care unit: a review**. *J Contemp Dent Pract* (2007) **8** 76-82. DOI: 10.5005/jcdp-8-1-76
6. Ranjbar H, Arab M, Abbasszade A, Ranjbar A. **Affective factors on oral care and its documentation in intensive care units**. *Kerman Med Sci Univ* (2011) **4** 45-52
7. Masomi M, Hanifi N, Jamshidi MR, Faghihzadeh S. **Investigation of the relationship between oral lesions and early pneumonia associated with mechanical ventilation in patients undergoing mechanical ventilation in intensive care unit**. *Sciences* (2016) **3** 107-114
8. Javadinia S, Kuchi Z, Saadatju A, Tabasi M, Adib-Hajbaghery M. **Oral care in trauma, patients admitted to the ICU. Viewpoints of ICU Nurses**. *Trauma, Mon* (2014) **2** e15110
9. Raphael C, Lecidamia C, Larycia V, Raphael C, JohnysBerton M, Eugênia L, Ana M. **Factors associated with oral biofilm in ICU patients with infectious diseases**. *Rev Odontol UNESP* (2017) **46** 343-350. DOI: 10.1590/1807-2577.05117
10. Prendergast V, Kleimanb C, Mary K. **The Bedside Oral Exam and the Barrow Oral Care Protocol: Translating evidence-based oral care into PracticeIntensive and Critical**. *Care Nursing* (2013) **29** 282-290
11. Johnstone L, Spence D, Koziol J. **Oral hygiene care in the pediatric intensive care unit: practice recommendation**. *Pediatr Nurs* (2010) **36** 85-96. PMID: 20476510
12. Nguh J. **Oral care practice guidelines for the care-dependent hospitalized adult outside of the intensive care unit setting**. *J Interprofessional Educ Pract* (2016) **1** 59-67. DOI: 10.1016/j.xjep.2016.05.004
13. Safarabadi M, Ghaznavi-Rad E, Pakniyat A, Rezaie K, Jadidi A. **Comparing the effect of echinacea and chlorhexidine mouthwash on the microbial flora of intubated patients admitted to the intensive care unit**. *Iran J Nurs Midwifery Res* (2017) **22** 481. DOI: 10.4103/ijnmr.IJNMR_92_16
14. da Cruz M, de Morais T, Trevisani D. **Clinical assessment of the oral cavity of patients hospitalized in an intensive care unit of an emergency hospital**. *Rev Bras Ter Intensiva Oct-Dec* (2014) **26** 379-383
15. Kima CH, Kimb M, Parkd N. **Oral mucosa pressure ulcers in intensive care unit patients: a preliminary observational study of incidence and risk factors**. *J Tissue Viability* (2019) **28** 27-34. DOI: 10.1016/j.jtv.2018.11.002
16. Atashi V, Yazdannik A, Mahjobipoor H, Yousefi H, Bekhradi R. **The impact of systematic oral care on oral health status in patients of intensive care units**. *J Health Care* (2018) **3** 244-251
17. Molania T, Mohsen A, Mohsen A, Mohsen A, Jamshid Y, Mohammad Reza P. **Frequency of Oral Lesions in Patients with HIV/AIDS in Sari, Iran**. *J Mazandaran Univ Med Sci* (2019) **30** 92-97
18. Jahanirinejad M, Arian Kia AN, Shahbazian HBB, Kabi N. **Community Periodontal Index of Treatment Needs Assessment of Periodontal Disease in Type 2 Diabetic Patients Compared to Nondiabetic Patients Referred to Ahvaz Golestan Hospital in 2016–2017**. *Jundishapur Sci Med J* (2018) **17** 295-300
19. Khatibi M, Najafizadeh SR, Shamshyrgaran S, Moezzi Ghadim N. **The prevalence of oral mucosal lesions and associated factors in 385 patients with rheumatoid arthritis in rheumatology clinics of Tehran University for a period of one year**. *Pejouhandeh* (2014) **19** 112-118
20. Yogini J, Naghsh N, Hadadi H. **evaluation some oral health indexes Mechanically ventilated patients Admitted into Intensive Care Unit of public Hospitalise in Isfahan in 2015**. *J Isfahan Dent Sch* (2017) **13** 250-263
21. Omer O, Mohamed Ibrahim S, Mohamed MA. **Nurses’ knowledge, attitude and practice of oral care for intensive care unit patients**. *Open J Stomatol* (2015) **5** 179-186. DOI: 10.4236/ojst.2015.57023
22. Rashidi M, Haerian A, Zarabadi P, Pourbaferani H. **Evaluation of Oral Health of Elderly Patients Referring to Khatam ol Anbia Clinic in Yazd in 2014**. *J Health Spring Suppl* (2016) **7** 227-235
23. Akbari M, Kazemian A, Afshari R. **Evaluation of oral health status and dental need assessment in narcotic drug abusers**. *J Mash Dent Sch* (2015) **39** 191-200
24. Ardakani A, Sharifabad M, Rezapour Y, Ardakani A. **Investigation of the relationship of oral health literacy and oral hygiene self-efficacy with DMFT and Gingival index in students of Ardakan University**. *Payesh* (2015) **14** 351-362
25. Kosari M, Hoseini Z, Golriz F. **The investigation of oral and dental status and its related factors in the elderly in Torbat Heydariyeh in 2015**. *J Health Chimes* (2016) **4** 39-44
26. Owlia F, Ahadian H, RustaeiZadeh Z, Bahadori L. **The prevalence of oral lesion among elderly in Yazd Province Nursing Homes in 2014**. *J Health* (2017) **8** 321-327
27. Maleki Z, GhaemMaghami A, Lesan S. **Comparison of soft tissue lesions (OSTLs) prevalence in the elderly institutionalized population who are resided in private and governmental institutes in Tehran Iran 2001–2003**. *J Dent Sch Shahid Beheshti* (2005) **23** 663-911
28. Shafipour V, Haghighi A, BagheriNesami M. **The impact of oral care in the prevention of ventilator-associated pneumonia: a narrative review**. *IJNR* (2018) **12** 25-36. DOI: 10.21859/ijnr-12064
29. Rezaei S, Rezaei K. **Comparison the effect of Miswak/Aloe Vera & chlorhexidine mouthwashes on gingival index in intubated patients**. *CMJA* (2015) **5** 1155-1166
|
---
title: 'Childhood overweight and obesity at the start of primary school: External
validation of pregnancy and early-life prediction models'
authors:
- Nida Ziauddeen
- Paul J. Roderick
- Gillian Santorelli
- John Wright
- Nisreen A. Alwan
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022097
doi: 10.1371/journal.pgph.0000258
license: CC BY 4.0
---
# Childhood overweight and obesity at the start of primary school: External validation of pregnancy and early-life prediction models
## Abstract
Tackling the childhood obesity epidemic can potentially be facilitated by risk-stratifying families at an early-stage to receive prevention interventions and extra support. Using data from the Born in Bradford (BiB) cohort, this analysis aimed to externally validate prediction models for childhood overweight and obesity developed as part of the Studying Lifecourse Obesity PrEdictors (SLOPE) study in Hampshire. BiB is a longitudinal multi-ethnic birth cohort study which recruited women at around 28 weeks gestation between 2007 and 2010 in Bradford. The outcome was body mass index (BMI) ≥91st centile for overweight/obesity at 4–5 years. Discrimination was assessed using the area under the receiver operating curve (AUC). Calibration was assessed for each tenth of predicted risk by calculating the ratio of predicted to observed risk and plotting observed proportions versus predicted probabilities. Data were available for 8003 children. The AUC on external validation was comparable to that on development at all stages (early pregnancy, birth, ~1 year and ~2 years). The AUC on external validation ranged between 0.64 ($95\%$ confidence interval (CI) 0.62 to 0.66) at early pregnancy and 0.82 ($95\%$ CI 0.81 to 0.84) at ~2 years compared to 0.66 ($95\%$ CI 0.65 to 0.67) and 0.83 ($95\%$ CI 0.82 to 0.84) on model development in SLOPE. Calibration was better in the later model stages (early life ~1 year and ~2 years). The SLOPE models developed for predicting childhood overweight and obesity risk performed well on external validation in a UK birth cohort with a different geographical location and ethnic composition.
## Introduction
Childhood overweight and obesity has been identified as one of the most serious public health challenges of the 21st century with 38 million children aged under 5 years and over 340 million aged 5 to 19 years overweight or obese globally [1]. The first ‘1000’ days, the time from conception to age 2 years, is recognised to be a critical period of development and is also a period of more intensive contact with health care professionals. Utilising this close contact to risk-stratify families at an early-stage to receive prevention interventions and extra support could be one approach to tackling the childhood obesity epidemic. Weight can be a sensitive topic to raise [2] and our consultation work with practitioners suggests that health professionals would like an ‘objective’ way to stratify risk rather than individualised clinical judgement which feels subjective. A meta-analysis of 48 studies concluded that there was a high degree of BMI tracking over time and a low probability of weight change without weight loss treatment [3]. A five year longitudinal study of 5863 students aged 11–12 years at baseline recruited from 36 London schools found little evidence of new cases of overweight and obesity emerging over adolescence but few overweight and obese adolescents reduced to a healthy weight [4] supporting the case for targeting obesity prevention strategies in early years. National guidance for the clinical management of overweight and obesity in both adults and children in the UK recommends multicomponent interventions which include behaviour change strategies to increase physical activity levels/decrease inactivity, improve eating behaviour and diet quality, and reduce energy intake [5]. Existing guidance on management of overweight and obesity could be adapted to target prevention in high-risk groups.
Prediction models are used to estimate the probability of developing a particular disease or outcome. Prediction models can provide more accurate risk estimates compared to more subjective predictions [6] and can augment clinical judgement [7] to enable intervention at an early stage before the development of the disease or outcome under consideration. Prediction models generated using routinely collected data are easier to apply in clinical practice as they only utilise data that is already being collected. A systematic review identified eight prediction models for the risk of childhood overweight and obesity [8]. The age at which outcome was predicted in the models varied from 1 to 13 years making it difficult to combine or compare models against each other. Only four of the eight prediction scores were externally validated. Inappropriate handling of missing data and discarding information through categorisation of continuous variable during model development can introduce bias. Two studies reported carrying out multiple imputation to handle missing data, the other studies either carried out complete case analysis or did not report presence/handling of missing data. Four studies retained all continuous predictors as continuous whereas the other studies categorised or dichotomised some or all continuous variables. Maternal pre-pregnancy BMI, infant gender and birthweight were the most common predictors but no single risk factor was included in all the prediction models. Model discrimination using area under the curve (AUC) ranged from 0.64 to 0.91 but only two could be applied in routine healthcare in the UK as predictors related to the father (such as paternal BMI or employment) or household (such as parental education, smoking in the household, number of siblings, income) are not routinely collected and may be complex to measure routinely. The two models which could be applied in routine healthcare both incorporated weight z-scores calculated using UK 1990 reference values [9].
As part of the Studying Lifecourse Obesity PrEdictors (SLOPE) study, we utilised anonymised routinely-collected antenatal and birth records linked to child health records for births registered at University Hospital Southampton (UHS) in the South of England between 2003 and 2018 to develop childhood overweight and obesity prediction models. UHS provides maternity care to residents in the city of Southampton and the surrounding areas of Hampshire. All maternal and birth variables in SLOPE were collected as part of healthcare for pregnant women in the study region. As part of England’s National Child Measurement Programme (NCMP), childhood weight and height were measured in 30,958 children at 4–5 years which was defined as the sample for the developing the prediction model. The models were developed in stages, incorporating data collected at first antenatal appointment (booking), later pregnancy/birth, and early-life predictors (~1 and ~2 years). Logistic regression with backward stepwise elimination and fractional polynomials were used to develop the models. Models were internally validated using bootstrapping (1000 repetitions). Models were well calibrated and area under the curve improved from 0.64 for the model only incorporating maternal predictors to 0.82 when incorporating all predictors up to child weight at ~2 years [10].
As a model usually performs better in the data used for its development, external validation in data that were not used to develop the model is needed to quantify model performance and to assess generalizability before application in practice. Model performance is evaluated using discrimination and calibration. This analysis aimed to externally validate the SLOPE prediction models using outcome data from children aged 4–5 years from the Born in Bradford (BiB) cohort [11].
## Methods
Data from the Born in Bradford (BiB) cohort was used for the external validation of the SLOPE models. BiB is a longitudinal multi-ethnic birth cohort study which recruited 12,453 women comprising 13,776 pregnancies at around 28 weeks gestation between 2007 and 2010 in Bradford, located in the North of England. Written informed consent for the data collection and linkage to routine data was provided by the child’s caregiver. Ethics approval was granted by the National Health Service Health Research Authority Yorkshire and the Humber (Bradford Leeds) Research Ethics Committee (reference: 16/YH/0320).
## Outcome
As part of the National Child Measurement Programme (NCMP) [12], height and weight are measured in all children attending state schools in England at 4–5 and 10–11 years. This coincides with the first and final years of primary school in England. Parental consent for linkage to the BiB cohort was provided at recruitment, but parents can opt-out of consenting to NCMP measurements. BMI was converted to age- and sex-adjusted BMI z-scores according to the UK 1990 growth reference charts [9, 13]. The outcome was overweight/obesity at 4–5 years, and was defined as BMI ≥91st centile based on the UK clinical cut-off [5, 14] used to develop the SLOPE models.
## Predictors from the SLOPE models
Maternal height (cm), self-reported ethnicity, education and smoking during pregnancy were obtained from an administered baseline questionnaire which was completed at recruitment at approximately 26–28 weeks gestation. Detailed self-reported ethnicity was condensed to White, Mixed, Asian, Black/African/Caribbean and Other. Highest maternal educational qualification was categorised as secondary (GCSE) and under, college (A levels), and university degree or above. Smoking was categorised as current smoker, ex-smoker, or non-smoker. Maternal weight at first antenatal booking appointment (approximately 12 weeks gestation), gestational age at birth, birthweight and child sex were obtained from electronic maternity records. Maternal BMI was calculated using weight measured at pregnancy booking and height from the baseline questionnaire. With regards to child BMI in early life, weight and height was measured at 12 and 24 months of age in a sub-cohort of the total BiB sample (BiB 1000) [15]. Participants in BiB 1000 were recruited as part of an intensive follow-up during infancy to study the patterns and aetiology of childhood obesity. All measurements were taken by researchers who received training in anthropometric measurements at the beginning of BiB using a study measurement protocol/standard. Where infant anthropometric measurements were missing, they were supplemented using linked data collected by health visitors as part of routine NHS care in the UK.
The four model stages were first antenatal appointment (booking), birth, early life ~1 year and early life ~2 years. Maternal BMI at booking, smoking status, ethnicity, intake of folic acid supplements and partnership status were predictors at every stage. Additional predictors at each model stage were [1] maternal age, first language and parity (booking); [2] maternal age, educational attainment, first language, parity, birthweight and gestational age at birth (birth); [3] maternal age, educational attainment, birthweight, gestational age at birth, infant gender and weight at ~1 year (early life ~1 year); and [4] educational attainment, birthweight, gestational age at birth, infant gender and weight ~2 years (early life ~2 years).
## Statistical analysis
All analysis was conducted using Stata 15 [16]. The sample was restricted to singleton children with a valid height and weight measurement at 4–5 years (to calculate the outcome of interest). Estimated risk of overweight and obesity at 4–5 years was calculated using the SLOPE CORE tool. Missing predictor values were imputed using multiple imputation by chained equations (MICE) with truncated regression for continuous variables and predictive mean matching for categorical variables. The percentage of missing data was highest for early life weight at ~1 and ~2 years and this was used to decide the number of imputations (55 imputations with 10 iterations per imputed dataset) based on the recommendation that the number of imputations equals the percentage of missing data in the dataset [17]. The results from analyses of each of the imputed datasets were combined to produce estimates and confidence intervals that incorporate the uncertainty of imputed values.
Predictive performance was assessed by examining discrimination and calibration. Discrimination is a measure of how well the model differentiates between individuals and was quantified by calculating the area under the receiver operating curve (AUC). An AUC value of 0.5 represents no discrimination capacity, with discrimination improving up to the value of 1 which represents perfect discrimination. Calibration measures how well the predicted outcome of the model agrees with the observed outcome on average [18]. This was assessed for each tenth of predicted risk, ensuring 10 equally sized groups, by calculating the ratio of predicted to observed risk and plotting observed proportions versus predicted probabilities.
A risk prediction tool is used to identify high risk groups by classifying individuals into high and low risk groups using a pre-defined risk threshold. A risk threshold of $20\%$ was reached for the SLOPE CORE (Childhood Obesity Risk Estimation) tool which was guided by prediction performance, local stakeholder consultation and prediction tools for other outcomes used in UK healthcare [10]. Thus, we compared the proportion identified using risk thresholds of $20\%$, $25\%$ and $30\%$ to that subsequently developed overweight and obesity at 4–5 years.
## Results
The BiB cohort included 8003 eligible children (Fig 1). Complete data for all predictors was available for $22.3\%$ of women/children ($$n = 1824$$). Most participants were missing data on none ($$n = 1824$$, $22.8\%$) or one ($$n = 2889$$, $36.1\%$) predictor. The percentage of missing data was highest for early life weight at ~1 ($54.4\%$) and ~2 years ($41.4\%$). The percentage of missing data for other predictors ranged from 17.8 to $21.4\%$.
**Fig 1:** *Flow diagram showing the eligible sample (n = 8003) for this analysis after excluding children without a valid weight and/or height measurement and twins/triplets.*
Cohort characteristics are presented in Table 1. Mean maternal age at booking was 27.3 years (standard deviation (SD) 5.5). Mean maternal BMI at booking was 26.0 kg/m2 (SD 5.6). Nearly three-quarters of women reported being never smokers. Nineteen percent of the women had a university degree or a higher qualification. The cohort was predominantly of Asian ($60.3\%$) or White ($36.0\%$) ethnicity. Two-thirds of women reported English as a second language. Fifteen percent of mothers reported being a lone parent at the booking appointment. The prevalence of overweight and obesity at 4–5 years was $14.5\%$. The prevalence of overweight and obesity was higher in children of mothers who smoked during pregnancy ($17.2\%$) than children of mothers who were never smokers ($14.0\%$).
**Table 1**
| Predictors | Mean ± SD | % of children ≥91st centile BMI |
| --- | --- | --- |
| N | 8003 | |
| Maternal age at booking, years | 27.3 ± 5.6 | - |
| Maternal BMI at booking, kg/m2 | 26.0 ± 5.6 | - |
| Birthweight, kg | 3.2 ± 0.6 | - |
| Gestational age at birth, days | 276 ± 13 | - |
| Child weight at ~1 year, kg | 9.1 ± 1.2 | - |
| Child weight at ~2 years, kg | 12.6 ± 1.6 | - |
| | % (95% CI) | |
| Maternal smoking status | | |
| Never smoked | 73.4 | 14.0 |
| Never smoked | 72.4 to 74.4 | 14.0 |
| Ex-smoker | 12.9 | 14.3 |
| Ex-smoker | 12.1 to 13.6 | 14.3 |
| Current smoker | 13.7 | 17.2 |
| Current smoker | 12.9 to 14.5 | 17.2 |
| Maternal educational attainment | | |
| University or above | 19.2 | 12.3 |
| University or above | 18.4 to 20.1 | 12.3 |
| College (A levels) | 11.3 | 13.1 |
| College (A levels) | 10.6 to 12 | 13.1 |
| Secondary (GCSE) or lower | 69.4 | 15.3 |
| Secondary (GCSE) or lower | 68.4 to 70.5 | 15.3 |
| Maternal ethnicity | | |
| White | 36.0 | 15.5 |
| White | 34.9 to 37.1 | 15.5 |
| Mixed | 1.7 | 13.5 |
| Mixed | 1.4 to 2.0 | 13.5 |
| Asian | 60.3 | 13.7 |
| Asian | 59.1 to 61.4 | 13.7 |
| Black/African/Caribbean | 1.4 | 19.6 |
| Black/African/Caribbean | 1.1 to 1.7 | 19.6 |
| Other | 0.7 | 17.8 |
| Other | 0.5 to 0.8 | 17.8 |
| Maternal intake of folic acid supplements | | |
| Started taking once pregnant | 78.8 | 14.5 |
| Started taking once pregnant | 77.8 to 79.8 | 14.5 |
| Not taking supplement | 21.2 | 14.3 |
| Not taking supplement | 20.2 to 22.2 | 14.3 |
| Maternal first language English | | |
| No | 33.9 | 14.1 |
| No | 32.9 to 34.9 | 14.1 |
| Yes | 66.1 | 14.7 |
| Yes | 65.1 to 67.1 | 14.7 |
| Partnership status at booking | | |
| Partnered | 85.3 | 14.2 |
| Partnered | 84.4 to 86.1 | 14.2 |
| Single | 14.7 | 15.6 |
| Single | 13.9 to 15.6 | 15.6 |
| Parity at booking | | |
| 0 | 36.8 | 14.4 |
| 0 | 35.7 to 37.8 | 14.4 |
| 1 | 27.9 | 14.1 |
| 1 | 26.9 to 28.8 | 14.1 |
| 2 | 17.3 | 13.3 |
| 2 | 16.5 to 18.1 | 13.3 |
| ≥3 | 18.0 | 16.2 |
| ≥3 | 17.2 to 18.9 | 16.2 |
| Child sex | | |
| Male | 51.3 | 15.0 |
| Male | 50.2 to 52.4 | 15.0 |
| Female | 48.7 | 13.9 |
| Female | 47.6 to 49.8 | 13.9 |
| Overweight/obese at 4–5 years | | |
| No | 85.5 | - |
| No | 84.8 to 86.3 | - |
| Yes | 14.5 | - |
| Yes | 13.7 to 15.2 | - |
The AUC on external validation was comparable to that at model development in SLOPE at all stages (booking, birth, ~1 year and ~2 years) (Table 2). The AUC at development was 0.66 ($95\%$ confidence intervals (CI) 0.65 to 0.67) at booking compared to 0.64 ($95\%$ CI 0.62 to 0.66) on external validation. Similarly, the AUC was 0.83 ($95\%$ CI 0.82 to 0.84) at ~2 years at development and 0.82 ($95\%$ CI 0.81 to 0.84) on external validation.
**Table 2**
| Unnamed: 0 | Booking | Birth | Early life (~1 year) | Early life (~2 years) |
| --- | --- | --- | --- | --- |
| Discrimination (AUC, 95% CI) | | | | |
| SLOPE development | 0.66 | 0.69 | 0.78 | 0.83 |
| SLOPE development | 0.65 to 0.67 | 0.68 to 0.70 | 0.77 to 0.79 | 0.82 to 0.84 |
| BiB external validation | 0.64 | 0.65 | 0.75 | 0.82 |
| BiB external validation | 0.62 to 0.66 | 0.64 to 0.67 | 0.74 to 0.77 | 0.81 to 0.84 |
Fig 2 shows the agreement between mean observed and mean predicted risk grouped by tenths of predicted risk for the four model stages. The SLOPE model at later stages (~1 year and ~2 years) gave a more accurate estimate of predicted risk. There was less agreement between observed and predicted risk grouped by tenth of risk at the early stages (booking and birth) in individuals with higher risk indicative of overestimation of risk.
**Fig 2:** *Predicted versus observed risk of overweight and obesity at 4–5 years by model stage for the SLOPE CORE tool by tenth of risk.*
Using a $20\%$ risk threshold, $36.9\%$ were identified as high risk at booking and captured $53.3\%$ of overweight and obesity events at 4–5 years (Table 3). The proportion identified as high risk decreased to $21.2\%$ at ~2 years and captured $59.6\%$ of events. Using higher risk thresholds ($25\%$ and $30\%$) identified fewer children as high risk but also captured fewer events. This indicates that the optimal threshold for determining high risk using these models is $20\%$.
**Table 3**
| Unnamed: 0 | Proportion identified as high risk | Low risk | Low risk.1 | High risk | High risk.1 |
| --- | --- | --- | --- | --- | --- |
| | Proportion identified as high risk | Not overweight and obese | Overweight and obese | Not overweight and obese | Overweight and obese |
| 20% risk threshold | | | | | |
| Booking | 36.9 | 65.9 | 46.7 | 34.1 | 53.3 |
| Booking | 36.9 | 64.6 to 67.2 | 43.5 to 50.0 | 32.8 to 35.4 | 50.0 to 56.5 |
| Birth | 23.0 | 79.6 | 61.3 | 20.4 | 38.7 |
| Birth | 23.0 | 78.6 to 80.7 | 58.2 to 64.5 | 19.3 to 21.5 | 35.5 to 41.8 |
| Early life (~1 year) | 22.7 | 82.2 | 48.0 | 17.8 | 52.0 |
| Early life (~1 year) | 22.7 | 81.0 to 83.4 | 4.5 to 51.4 | 16.6 to 19.0 | 48.6 to 55.5 |
| Early life (~2 years) | 21.2 | 85.3 | 40.4 | 14.7 | 59.6 |
| Early life (~2 years) | 21.2 | 84.3 to 86.3 | 36.9 to 43.9 | 13.7 to 15.7 | 56.1 to 63.1 |
| 25% risk threshold | | | | | |
| Booking | 20.9 | 81.5 | 65.1 | 18.5 | 34.9 |
| Booking | 20.9 | 80.4 to 82.6 | 62.0 to 68.3 | 17.4 to 19.6 | 31.7 to 38.0 |
| Birth | 11.9 | 90.1 | 76.4 | 9.9 | 23.6 |
| Birth | 11.9 | 89.3 to 90.9 | 73.7 to 79.1 | 9.1 to 10.7 | 20.9 to 26.3 |
| Early life (~1 year) | 15.8 | 88.5 | 59.2 | 11.5 | 40.8 |
| Early life (~1 year) | 15.8 | 87.5 to 89.5 | 55.9 to 62.5 | 10.5 to 12.5 | 37.5 to 44.1 |
| Early life (~2 years) | 16.3 | 89.6 | 49.0 | 10.4 | 50.1 |
| Early life (~2 years) | 16.3 | 88.7 to 90.4 | 45.5 to 52.6 | 9.6 to 11.3 | 47.4 to 54.5 |
| 30% risk threshold | | | | | |
| Booking | 10.0 | 91.6 | 80.9 | 8.4 | 19.1 |
| Booking | 10.0 | 90.7 to 92.4 | 78.3 to 83.4 | 7.6 to 9.3 | 16.6 to 21.7 |
| Birth | 5.9 | 95.6 | 85.9 | 4.4 | 14.1 |
| Birth | 5.9 | 95.0 to 96.1 | 83.7 to 88.0 | 3.9 to 5.0 | 12.0 to 16.3 |
| Early life (~1 year) | 10.7 | 92.9 | 68.1 | 7.1 | 31.9 |
| Early life (~1 year) | 10.7 | 92.1 to 93.7 | 64.8 to 71.4 | 6.3 to 7.9 | 28.6 to 35.2 |
| Early life (~2 years) | 12.8 | 92.4 | 56.2 | 7.6 | 43.8 |
| Early life (~2 years) | 12.8 | 91.6 to 93.2 | 52.8 to 59.7 | 6.8 to 8.4 | 40.3 to 47.2 |
## Discussion
We evaluated the performance of the SLOPE models using data from the South of England for predicting the risk of overweight and obesity at 4–5 years in the BiB cohort in the north of England. Discrimination of the SLOPE equations in this cohort was comparable to that in the development cohort at all model stages. Risk was over-predicted (predicted risk was higher than observed risk) in higher risk groups using the early pregnancy and birth equations but calibration was good in the later model stages (early life ~1 year and ~2 years). Using a $20\%$ risk threshold, $36.9\%$ were identified as high risk at booking and captured $53.3\%$ of overweight and obesity events at 4–5 years.
Of eight prediction models for childhood overweight and obesity previously identified in a systematic review, four have been externally validated [8]. Prediction models for childhood obesity developed using data from the 1986 Northern Finland Birth Cohort were externally validated in a retrospective cohort in Italy and a prospective birth cohort (Project Viva) in the US. The AUC was slightly lower in the US external validation cohort than the Finnish development cohort (0.73 vs 0.78) but calibration was not satisfactory. A modified version of the model excluding two predictors was externally validated in the Italian cohort and the AUC was 0.73 on development and 0.70 on external validation with adequate calibration [19].
In the UK, a risk prediction algorithm developed using data from the UK Millennium Cohort Study (MCS) [20] was externally validated in a $10\%$ sample of the Avon Longitudinal Study of Parents and Children (ALSPAC) known as the Children in Focus (CiF). Of the 1432 children in this sub-sample, data on child weight was available for 980 children and this was the final sample for external validation. The AUC on developing the model in MCS was 0.72. Applying the model in the ALSPAC validation sample resulted in an AUC of 0.67. The AUC increased to 0.70 on model recalibration [21].
The prediction model developed using data from the BiB cohort was also validated using the CiF subsample of ALSPAC. Models were developed for use at three stages (6±1.5 months, 9±1.5 months and 12±1.5 months). The AUC ranged from 0.86 to 0.91 on model development, and from 0.85 to 0.89 on external validation [22].
In most externally validated models, AUCs on external validation were slightly lower than in development models, and this in line with our study. An exception is a risk index which was developed using data from a retrospective cross-sectional school based cohort collected in 2007 in Greece which had a similar AUC of 0.64 on development and external validation in a cross-sectional school based cohort collected in 2012–13 [23].
In terms of assessing calibration, one published study reported model calibration [19] but used a different measure (Hosmer-Lemeshow test) to the measure used in our study. It reported unsatisfactory calibration on external validation in one of the two cohorts they used, while our models remained well calibrated on external validation.
## Strengths and limitations
The SLOPE models were developed in a population-based cohort in the South of England using routinely collected healthcare data. The prevalence of the outcome was comparable in both the development ($14.8\%$) and external validation ($14.5\%$) cohorts and was measured as part of the NCMP thus limiting variation in measurement practices. Key differences between the development and external validation cohorts were the ethnic composition where the development cohort was predominantly White ($90\%$) whereas the external validation cohort was $60\%$ South Asian and $31\%$ White. This is reflected in the slightly lower average birthweight [24] in the overall sample and a higher proportion reporting English as a second language. There was also a higher proportion of lone mothers, mothers with lower educational attainment and higher order pregnancies in BiB.
Bradford ranked 19th and Southampton 54th out of 317 local authorities in England (1 is most deprived) in 2015 so are both relatively deprived cities [25]. The use of data from more deprived areas is a strength of this analysis given the higher prevalence of overweight and obesity in more deprived areas [26]. The BiB cohort is representative of the local population in Bradford and has similarities with other UK cities with high levels of ethnic minority groups but is not representative of the rest of the country [11].
The differences between the cohorts are a strength of the analysis as it assesses model performance in a population with different characteristics and therefore the findings would be more generalisable. The performance of the SLOPE model on external validation in BiB was comparable to model performance on development and supports its use to predict risk of childhood overweight and obesity within a wider UK setting.
There was a high level of missing data for early life predictors (weight at ~1 year and ~2 years) in both the development and validation cohort. Multiple imputation was used to address this. Measuring health and weight in children at these ages is part of statutory care and so the missing data could reflect an issue in how these measurements are recorded. It is possible that the child’s handheld record is updated but not their electronic record or that the measurement is entered in open text boxes in the electronic records rather in designated response boxes which makes it difficult to access for research purposes. Recording of key variables in electronic records have improved over time and considering that the model performance was highest for age groups which also had the highest percentage of missing data, the implementation of this tool in practice may have implications for both data recording and utilisation of the tool.
BMI is a useful population-level measure of overweight and obesity as it is easy to measure and the same regardless of gender or age (once adulthood is attained). BMI is the most commonly used marker of overweight and obesity [27] and is the marker used in the NCMP in England after adjusting for age and sex. However, BMI does not account for differences in body composition and thus may not be the best marker of body fat [28] or cardio-metabolic risk in South Asian children [29].
This study also shows that the optimal threshold for identifying high risk was $20\%$ which is what was suggested based on the sensitivity, specificity, positive predictive values and negative predictive value of the models at development [10]. At the $20\%$ risk threshold, $52\%$ and $60\%$ of cases are identified using the models at ~1 year and ~2 years respectively. This reduces to 32–$50\%$ at these stages if higher risk thresholds are used.
As the model incorporates several equations requiring background calculations, we developed a website so that the tool can be used easily. The user is required to enter the values for the continuous predictors and choose the appropriate option from the dropdown for the categorical variables. Information on all predictors is required at each stage to calculate the risk score. The risk score is categorised as low risk (0–$20\%$, green), medium risk (20–$30\%$, yellow) and high risk (≥$30\%$, orange). The feasibility and acceptability of using the tool in practice by midwives and health visitors, including the risk cut-offs and risk communication is being explored in Wessex [30]. We are testing the usability and feasibility of the prediction tool (SLOPE CORE (Childhood Obesity Risk Estimation)) as an aid to healthcare professionals to guide delivery of an intervention rather than just as a screening test. This is based on public involvement consultations in which mothers expressed interest in early identification of risk but stressed the need for support and advice to modify risk. The tool is envisaged to provide a prompt for the health professional to introduce the topic at an early stage and to help target extra support in resource limited setting. Additionally, we envisage health professionals will use their professional judgement and continue to provide support to groups they identify as needing additional support even if they are categorised as low risk using the tool. Next steps also include developing and validating prediction models for childhood overweight and obesity using the routinely measured outcome of BMI at age 10–11 years as this captures key development of overweight and obesity.
The SLOPE models developed for predicting childhood overweight and obesity risk demonstrated good model performance on external validation in a birth cohort with a different geographical location and ethnic composition.
## References
1. **Obesity and overweight**. (2020.0)
2. Michie S.. **Talking to primary care patients about weight: a study of GPs and practice nurses in the UK**. *Psychol Health Med.* (2007.0) **12** 521-5. DOI: 10.1080/13548500701203441
3. Bayer O, Krüger H, Von Kries R, Toschke AM. **Factors Associated With Tracking of BMI: A Meta-Regression Analysis on BMI Tracking***. *Obesity* (2011.0) **19** 1069-76. DOI: 10.1038/oby.2010.250
4. Wardle J, Brodersen NH, Cole TJ, Jarvis MJ, Boniface DR. **Development of adiposity in adolescence: five year longitudinal study of an ethnically and socioeconomically diverse sample of young people in Britain**. *BMJ* (2006.0) **332** 1130-5. DOI: 10.1136/bmj.38807.594792.AE
5. 5National Institute of Heath and Care Excellence. Obesity: identification, assessment and management [Internet]. 2014
Nov. Available from: https://www.nice.org.uk/guidance/cg189. *Obesity: identification, assessment and management [Internet]* (2014.0)
6. Kattan MW, Yu C, Stephenson AJ, Sartor O, Tombal B. **Clinicians Versus Nomogram: Predicting Future Technetium-99m Bone Scan Positivity in Patients With Rising Prostate-specific Antigen After Radical Prostatectomy for Prostate Cancer**. *Urology* (2013.0) **81** 956-61. DOI: 10.1016/j.urology.2012.12.010
7. Wong DJN, Harris S, Sahni A, Bedford JR, Cortes L, Shawyer R, Menon D. **Developing and validating subjective and objective risk-assessment measures for predicting mortality after major surgery: An international prospective cohort study**. *PLOS Med.* (2020.0) **17** e1003253. DOI: 10.1371/journal.pmed.1003253
8. Ziauddeen N, Roderick PJ, Macklon NS, Alwan NA. **Predicting childhood overweight and obesity using maternal and early life risk factors: a systematic review**. *Obes Rev Off J Int Assoc Study Obes* (2018.0) **19** 302-12. DOI: 10.1111/obr.12640
9. Freeman JV, Cole TJ, Chinn S, Jones PR, White EM, Preece MA. **Cross sectional stature and weight reference curves for the UK, 1990**. *Arch Dis Child* (1995.0) **73** 17-24. DOI: 10.1136/adc.73.1.17
10. Ziauddeen N, Wilding S, Roderick PJ, Macklon NS, Smith D, Chase D. **Predicting the risk of childhood overweight and obesity at 4–5 years using population-level pregnancy and early-life healthcare data**. *BMC Med.* (2020.0) **18** 105. DOI: 10.1186/s12916-020-01568-z
11. Wright J, Small N, Raynor P, Tuffnell D, Bhopal R, Cameron N. **Cohort Profile: The Born in Bradford multi-ethnic family cohort study**. *Int J Epidemiol.* (2013.0) **42** 978-91. DOI: 10.1093/ije/dys112
12. 12NHS Digital. National Child Measurement Programme [Internet]. [cited 2020 Aug 25]. Available from: https://digital.nhs.uk/services/national-child-measurement-programme/. *National Child Measurement Programme [Internet]*
13. Vidmar S, Carlin J, Hesketh K, Cole T. **Standardizing Anthropometric Measures in Children and Adolescents with New Functions for Egen**. *Stata J Promot Commun Stat Stata.* (2004.0) **4** 50-5
14. 14Scientific Advisory Committee on Nutrition (SACN), Royal College of Paediatrics and Child Health (RCPCH). Consideration of issues around the use of BMI centile thresholds for defining underweight, overweight and obesity in children aged 2–18 years in the UK [Internet]. 2012. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/339411/SACN_RCPCH_defining_child_underweight__overweight_and_obesity_in_the_UK_2012.pdf. *Consideration of issues around the use of BMI centile thresholds for defining underweight, overweight and obesity in children aged 2–18 years in the UK [Internet]* (2012.0)
15. Bryant M, Santorelli G, Fairley L, West J, Lawlor DA, Bhopal R. **Design and characteristics of a new birth cohort, to study the early origins and ethnic variation of childhood obesity: the BiB1000 study**. *Longitud Life Course Stud* (2013.0) **4** 119-35
16. 16Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC; 2017.. *Stata Statistical Software: Release 15* (2017.0)
17. White IR, Royston P, Wood AM. **Multiple imputation using chained equations: Issues and guidance for practice**. *Stat Med.* (2011.0) **30** 377-99. DOI: 10.1002/sim.4067
18. Harrell FE, Lee KL, Mark DB. **Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors**. *Stat Med.* (1996.0) **15** 361-87. DOI: 10.1002/(SICI)1097-0258(19960229)15:4<361::AID-SIM168>3.0.CO;2-4
19. Morandi A, Meyre D, Lobbens S, Kleinman K, Kaakinen M, Rifas-Shiman SL, Manco M. **Estimation of Newborn Risk for Child or Adolescent Obesity: Lessons from Longitudinal Birth Cohorts**. *PLoS ONE* (2012.0) **7** e49919. DOI: 10.1371/journal.pone.0049919
20. Weng SF, Redsell SA, Nathan D, Swift JA, Yang M, Glazebrook C. **Estimating Overweight Risk in Childhood From Predictors During Infancy**. *PEDIATRICS* (2013.0) **132** e414-21. DOI: 10.1542/peds.2012-3858
21. Redsell SA, Weng S, Swift JA, Nathan D, Glazebrook C. **Validation, Optimal Threshold Determination, and Clinical Utility of the Infant Risk of Overweight Checklist for Early Prevention of Child Overweight**. *Child Obes* (2016.0) **12** 202-9. DOI: 10.1089/chi.2015.0246
22. Santorelli G, Petherick ES, Wright J, Wilson B, Samiei H, Cameron N. **Developing prediction equations and a mobile phone application to identify infants at risk of obesity**. *PloS One* (2013.0) **8** e71183. DOI: 10.1371/journal.pone.0071183
23. Manios Y, Vlachopapadopoulou E, Moschonis G, Karachaliou F, Psaltopoulou T, Koutsouki D. **Utility and applicability of the “Childhood Obesity Risk Evaluation” (CORE)-index in predicting obesity in childhood and adolescence in Greece from early life: the “National Action Plan for Public Health.”**. *Eur J Pediatr* (2016.0) **175** 1989-96. DOI: 10.1007/s00431-016-2799-2
24. West J, Lawlor DA, Fairley L, Bhopal R, Cameron N, McKinney PA. **UK-born Pakistani-origin infants are relatively more adipose than white British infants: findings from 8704 mother-offspring pairs in the Born-in-Bradford prospective birth cohort**. *J Epidemiol Community Health* (2013.0) **67** 544-51. DOI: 10.1136/jech-2012-201891
25. 25Department for Communities and Local Government. The English Indices of Deprivation 2015. 2015.. *The English Indices of Deprivation 2015* (2015.0)
26. 26NHS Digital. National Child Measurement Programme, England 2018/19 School Year [NS] [Internet]. 2019
Oct [cited 2020 Mar 18]. Available from: https://digital.nhs.uk/data-and-information/publications/statistical/national-child-measurement-programme/2018-19-school-year#. *National Child Measurement Programme, England 2018/19 School Year [NS] [Internet]* (2019.0)
27. Dinsdale H, Ridler C, Ells L. *A simple guide to classifying body mass index in children* (2011.0)
28. Nightingale CM, Rudnicka AR, Owen CG, Cook DG, Whincup PH. **Patterns of body size and adiposity among UK children of South Asian, black African-Caribbean and white European origin: Child Heart And health Study in England (CHASE Study)**. *Int J Epidemiol.* (2011.0) **40** 33-44. DOI: 10.1093/ije/dyq180
29. Santorelli G, West J, Yang T, Wright J, Bryant M, Lawlor DA. **Differences in total and regional body fat and their association with BMI in UK-born White and South Asian children: findings from the Born in Bradford birth cohort**. *Wellcome Open Res.* (2021.0) **6** 65
30. 30The Wessex FRIEND Toolbox (Family Risk IdEntificatioN and Decision) [Internet]. [cited 2021 Mar 29]. Available from: https://www.arc-wx.nihr.ac.uk/research-areas/healthy-communities/the-friend-tool-to-predict-childhood-obesity-and-provide-local-measures-of-poverty/
|
---
title: The impact of inotersen on Neuropathy Impairment Score in patients with hereditary
transthyretin amyloidosis with polyneuropathy
authors:
- Aaron Yarlas
- Andrew Lovley
- Duncan Brown
- Montserrat Vera-Llonch
- Sami Khella
- Chafic Karam
journal: BMC Neurology
year: 2023
pmcid: PMC10022100
doi: 10.1186/s12883-023-03116-7
license: CC BY 4.0
---
# The impact of inotersen on Neuropathy Impairment Score in patients with hereditary transthyretin amyloidosis with polyneuropathy
## Abstract
### Background
Patients with hereditary transthyretin amyloidosis (ATTRv) frequently experience symptoms of polyneuropathy (PN) that worsen over time and impair daily functioning. Previous analyses supported efficacy of inotersen, an antisense oligonucleotide, to slow neuropathic progression in patients with ATTRv-PN, as indicated by larger mean changes, relative to placebo, in total score and several subscales of the Neuropathy Impairment Score (NIS), and for the subset of NIS items specific to lower limbs (NIS-LL) for the overall study sample. A key objective of the current study was to evaluate efficacy of inotersen for slowing neuropathic progression in NIS/NIS-LL within key clinical subgroups of patients with ATTRv-PN. Additionally, for this study, responder definition (RD) thresholds were estimated for NIS/NIS-LL total and subscale scores, for the purpose of evaluating clinically meaningful benefit of inotersen at the individual patient-level.
### Methods
Post hoc analyses used data from the NEURO-TTR phase 3 trial of inotersen in patients with ATTRv-PN (NCT01737398). Treatment differences in mean changes on NIS/NIS-LL total and subscale scores from baseline to week 65 were examined within patient subgroups defined by clinical characteristics. Anchor- and distribution-based approaches estimated RDs for NIS/NIS-LL scores, with responders defined as patients who did not experience clinically meaningful neuropathic progression. Responder analyses compared the proportion of patients classified as responders for each NIS/NIS-LL score between treatment arms.
### Results
Within each patient subgroup, mean increases in NIS/NIS-LL total and muscle weakness subscales were significantly smaller after 65 weeks of treatment with inotersen compared to placebo. Similar patterns were observed for some, but not all, subgroups on NIS/NIS-LL reflex subscale scores. Recommended RDs were 8.1 points for NIS total and 4.7 points for NIS-LL total. Patients receiving inotersen for 65 weeks were significantly less likely than those receiving placebo to exhibit clinically meaningful increases on NIS/NIS-LL total, muscle weakness, and sensation subscales.
### Conclusions
This study supports previous evidence for efficacy of inotersen in this patient population and provides interpretation guidelines for clinically meaningful changes in NIS/NIS-LL scores.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12883-023-03116-7.
## Background
Hereditary transthyretin amyloidosis (ATTRv) is a rare, systemic, progressive, and life-threatening disease caused by the misfolding of the transthyretin (TTR) protein and consequent formation of amyloid fibrils, which deposit in organs and tissues throughout the body and disrupt their ability to function [1, 2]. The accumulation of TTR amyloid in nervous tissue often leads to polyneuropathy (PN), manifesting as sensorimotor impairment and autonomic dysfunction that worsens rapidly over time without treatment [3]. Common symptoms such as numbness, fatigue, and weakness in the limbs can increasingly limit patients’ independence and ability to carry out daily activities, with substantial impact on their quality of life [4–6].
*Current* gene-silencing treatments for patients with ATTRv-PN aim to slow or halt further damage to organs and tissues, and worsening symptoms, by limiting the production of new TTR amyloid [7]. In the phase 3 NEURO-TTR trial [8], patients with ATTRv-PN receiving inotersen, an antisense oligonucleotide, exhibited slower progression of neuropathic symptoms, relative to patients receiving placebo, after 65 weeks of treatment, as measured by statistically significant treatment differences in change on the muscle weakness, sensation loss, and reflex subscales of the clinician-reported Neuropathy Impairment Score (NIS) and the NIS-Lower Limbs (NIS-LL), with the latter comprised of a subset of NIS items specific to the lower limbs [9]. However, while treatment differences were observed for the total sample, the benefit of inotersen on changes in NIS and NIS-LL scores within subgroups of patients with ATTRv-PN defined by key clinical characteristics has not yet been evaluated. ATTRv-PN is a highly heterogeneous disease, with differences in genetic mutation, organ involvement, stage of disease progression, and symptoms, contributing to a diversity of patient experiences. Evaluating the efficacy of treatment among patient subgroups defined by these clinical characteristics could show where treatment benefit is greatest, as well as help identify areas of unmet need.
Further, while the benefit of inotersen on NIS and NIS-LL was shown to be statistically significant at the group level, whether this benefit was clinically meaningful at the level of the individual patient has not yet been investigated. Responder definition (RD) thresholds, also referred to as the minimal clinically important difference (MCID), have been defined as the smallest difference in score that patients would consider a benefit and would warrant a change in their treatment [10]. While some researchers have proposed RD thresholds in NIS or NIS-LL scores, these values are based on a misinterpretation of the literature. For example, researchers have stated that the minimal clinically meaningful change is a 2-point increase (i.e., worsening) on the NIS-LL total score [11], which has a score range of 88 points, while others have stated that this 2-point threshold applies to the NIS plus 7 nerve tests (NIS + 7) [12], which has a score range of 240 points, or the modified NIS + 7 (mNIS+ 7) [8], which has a score range of 369 points, despite the fact that all three cite the same source for these values [13]. Further, the cited source actually refers to a 2-point change on the NIS total score, which has a range of 244 points, as indicating meaningful change, although no empirical evidence is provided to support this value, and this threshold was not estimated with respect to patients with ATTRv-PN [13]. A 2-point change represents a change of $0.5\%$ on the mNIS+ 7, a change of $0.8\%$ on the NIS or NIS + 7, and a change of $2.3\%$ on the NIS-LL, which are all far lower than what is typically observed for the magnitude of an RD threshold. As such, there are currently no established, empirically supported RD thresholds that represent meaningful change in NIS or NIS-LL scores, limiting the degree to which a treatment benefit can be evaluated as providing a clinically meaningful benefit to patients with ATTRv-PN.
Based on the evidential gaps described here, this study had three objectives. The first objective was to compare, within patient subgroups defined by key clinical characteristics, mean changes in NIS and NIS-LL total and subscale scores after 65 weeks between patients with ATTRv-PN receiving inotersen or placebo in the NEURO-TTR trial. The second objective was to estimate RD thresholds for NIS and NIS-LL total and subscale scores within this patient sample. The third objective was to examine the efficacy of inotersen for clinically meaningful slowing of neuropathic progression at the level of the individual patient by classifying responders using estimated RD thresholds, and then comparing the proportions of responders at week 65 between patients receiving inotersen and those receiving placebo.
## Data source
Data for this study are from the NEURO-TTR trial, a phase 3, multinational, multicenter, randomized, placebo-controlled, double-blinded study of inotersen for the treatment of ATTRv-PN (ClinicalTrials.gov ID: NCT01737398) [8]. Adult patients with ATTRv-PN were randomized in a 2:1 ratio to receive 300 mg subcutaneous inotersen sodium or matching placebo once weekly for 65 weeks. A total of 172 patients (inotersen: 112; placebo: 60) were enrolled in the safety set, having received at least one dose of the study drug. Patients with amyloidosis confirmed by biopsy, a TTR variant confirmed by genotyping, and a NIS total score between 10 and 130 (inclusive) were eligible to participate; patients confined to wheelchairs or bedridden were not eligible to participate in the study.
## Ethical standards
The NEURO-TTR study protocol was approved by the relevant institutional review boards or local ethics committees and regulatory authorities. The study was conducted in accordance with Good Clinical Practice guidelines of the International Conference on Harmonization and the principles of the Declaration of Helsinki. All patients provided written informed consent to participate in the study.
## Neuropathy Impairment Score (NIS)
The NIS is a clinician-rated measure of neuropathic progression that involves 37 bilateral assessments of the cranial nerves and limbs for muscle weakness, sensation loss, and decreased reflexes [13]. Assessments are conducted by a trained clinician who rates the degree of neuropathy at each site on scales ranging from 0 (normal nerve function) to 4 (paralysis) for cranial nerve and muscle weakness, and from 0 (normal) to 2 (absent) for reflex and sensation tests. Ratings are summed to calculate a composite total score that ranges from 0 to 244. Subscale scores for the NIS include cranial nerves (range: 0 to 40), muscle weakness (range: 0 to 152), sensation loss (range: 0 to 32), and decreased reflexes (range: 0 to 20). Higher NIS total and subscale scores reflect greater neuropathic impairment.
## NIS-lower limbs (NIS-LL)
The NIS-LL is a subset of 14 NIS assessments specific to neuropathy in the lower limbs. A composite NIS-LL total score ranges from 0 to 88, with subscale scores for muscle weakness (range: 0 to 64), sensation loss (range: 0 to 16), and decreased reflexes (range: 0 to 8). Higher NIS-LL scores indicate greater neuropathic impairment of the lower limbs.
The NIS (and thus the NIS-LL item subset) was administered at baseline and week 65 visits. At each of these visits, the NIS was administered twice, with the two assessments recommended to occur on consecutive days. The two assessments at each visit were averaged. If only one assessment was conducted, then values from the single assessment were used. When possible, each patient was assessed by the same neurologist for all visits. Neurologists trained to administer the NIS/NIS-LL used standard procedures and equipment (e.g., cotton wool, pins, tuning fork, reflex hammer) and were instructed to consider abnormal nerve function in the context of the patient’s age, sex, weight, height, and overall physical fitness.
## Statistical analyses
This analysis was exploratory and conducted post hoc. The study population was the full analysis set (FAS), which included all randomized patients who had at least one dose of the study drug and at least one post-baseline efficacy assessment ($$n = 165$$). NIS and NIS-LL data from baseline and week 65 visits were used for this analysis. Due to very few patients in the study showing signs of cranial nerve impairment, this subscale of the NIS was not analyzed [9].
## Evaluation of treatment benefit for Inotersen within patient subgroups
Treatment differences in least-squares (LS) mean change in NIS and NIS-LL scores from baseline to week 65 were examined for the FAS, as well as within patient subgroups defined by the following clinical characteristics: genetic mutation (V30M, non-V30M), familial amyloid polyneuropathy (FAP) disease stage (Stage 1 [ambulatory without assistance], Stage 2 [ambulatory with assistance of cane or walker]) [14], previous treatment status with tafamidis and/or diflunisal (pretreatment, no pretreatment), cardiomyopathy (CM) status (CM, no CM), and age of symptom onset (early [< 50 years], late [≥50 years]). Treatment differences in mean changes were analyzed using mixed-effects models for repeated measures (MMRM). Specifications of MMRM for analyses of the FAS included fixed categorical effects for treatment, time, randomization stratification factors (i.e., presence/absence of previous treatment with tafamidis and/or diflunisal; FAP Stage 1 or Stage 2; and V30M or non-V30M mutation), and treatment-by-time interaction, with fixed covariates for the baseline value and baseline-by-time interaction. Model specifications for analyses of subgroups included fixed categorical effects for treatment, time, randomization stratification factors, treatment-by-time interaction, treatment-by-subgroup interaction, and treatment-by-time-by-subgroup interaction, with fixed covariates for the baseline value and baseline-by-time interaction. Note that when subgroups included a stratification factor, that stratification factor was not included in the model. For example, when examining treatment differences within V30M and non-V30M subgroups, this factor was not included as a fixed categorical effect in the model.
The magnitude of treatment effects was also assessed using effect sizes for standardized mean differences, expressed as Cohen’s d, and interpreted according to Cohen’s published guidelines ($d = 0.2$, small effect; $d = 0.5$, medium effect; $d = 0.8$, large effect) [15].
## Estimation of responder definition thresholds
Because ATTRv-PN is a progressive disease with a treatment goal of slowing or stabilizing neuropathy, rather than reversing it, as in previous studies estimating or applying RD thresholds for measures of neuropathic impairment in these patients [11, 16], responders in this study were defined as patients who did not exhibit clinically meaningful progression of neuropathic impairment, as measured by increases in NIS and NIS-LL scores after 65 weeks of treatment. RD thresholds for NIS-LL total and muscle weakness subscale scores were estimated using both anchor-based and distribution-based methods. RD thresholds for NIS muscle weakness and reflex subscales, as well as for the NIS-LL reflex subscale, were estimated using distribution-based methods only, as there were no appropriate anchor measures available for these outcomes. Table 1 provides a schematic for approaches and methods used to estimate RD thresholds for NIS and NIS-LL total and subscale scores. Table 1Summary of methods used to estimate responder definition (RD) thresholds Anchor-based Approaches Target Measures ➣ Mean Change Mean change in NIS/NIS-LL scores for patients who did not exhibit meaningful worsening on the anchor ➣ Linear Regression Linear regression models with change in NIS/NIS-LL scores as the outcome and change in the anchor (LLF/PND) as the predictor ➣ Receiver Operating Characteristic Curve Receiver operating characteristic curves to identify the optimal cut-off point on NIS/NIS-LL scores for classifying patients showing meaningful worsening or not based on the anchor measure Anchor: Polyneuropathy Disability (PND) score • NIS Total• NIS Sensation• NIS-LL Sensation Anchor: Lower Limb Function (LLF) test • NIS-LL Total• NIS-LL Muscle Weakness Distribution-based Approaches Target Measures ➣ Effect Size Group difference or change over time relative to the standard deviation at baseline ➣ Standard Error of Measurement Measurement error of a scale based on the standard deviation of baseline scores and the scale’s intra-rater reliability ➣ Standardized Response Mean Group difference or change over time relative to the standard deviation of change scores • All NIS and NIS-LL total and subscales Abbreviations: NIS Neuropathy Impairment Score, NIS-LL Neuropathy Impairment Score – Lower Limbs
## Anchor-based approaches
Anchor-based approaches estimate RD thresholds based on the correspondence between changes in the target measure and in an anchor measure. An anchor measure is an independent criterion measure for which there are clearly defined indicators for interpreting change in a patient’s clinical health. Appropriate anchor measures assess similar constructs as those captured by the target measure, and changes in the anchor should have at least a moderate statistical association with the target; a correlation ≥|0.30| between changes in the target measure and any anchor measure is recommended [17].
One measure was identified as an appropriate anchor for NIS total score and sensation subscales for both the NIS and NIS-LL: the Polyneuropathy Disability score (PND). The PND is a clinician-rated classification of patients into one of five stages of ambulatory disability: Stage I, indicating sensory disturbances in limbs without motor impairment; Stage II, indicating difficulty walking without the need of a walking aid; Stage IIIa, for which one stick or one crutch is required for walking; Stage IIIb, for which two sticks or two crutches are required for walking; and Stage IV, for patients who are confined to a wheelchair or bedridden [18]. An increase of one point on the PND can be interpreted as a clinically meaningful change. The PND was administered at baseline and week 65 visits. Spearman rank-order correlations between the changes from baseline to week 65 in PND score and changes in NIS total score and the NIS and NIS-LL sensation subscales were 0.30, 0.31, and 0.31, respectively, all of which were statistically significant ($p \leq 0.001$), supporting the use of the PND as an anchor measure for all three of these measures. Correlations between PND scores and all other NIS and NIS-LL measures were < 0.30.
A second measure was identified as an appropriate anchor for the NIS-LL total score and muscle weakness subscale: the lower limb function test (LLF). The LLF is a 3-item clinician assessment of a patient’s ability to walk on their toes, walk on their heels, and stand from a kneeling position [19]. Each item is assessed as normal (coded as 0) or abnormal (coded as 1), and are assessed bilaterally, yielding an LLF score ranging from 0 to 6, with higher scores indicating greater neuropathic impairment. An increase of two points on the LLF, representing bilateral change, can be interpreted as clinically meaningful. The LLF was administered at baseline and week 65 visits. Spearman rank-order correlations between the changes from baseline to week 65 in LLF score and changes in NIS-LL total and NIS-LL muscle weakness were 0.35 and 0.32, respectively, all of which were statistically significant ($p \leq 0.001$), supporting the use of the LLF as an anchor measure for both. Correlations between LLF scores and all other NIS and NIS-LL measures were < 0.30.
No appropriate anchor measure from the NEURO-TTR trial was identified for the NIS muscles weakness subscale or for NIS or NIS-LL reflex subscales, as no other clinician-rated assessment of neuropathic impairment that was conceptually related to these outcomes had straightforward interpretation of what would indicate clinically meaningful improvement.
Three anchor-based methods were used to estimate RD thresholds for the NIS/NIS-LL measures from corresponding anchors. First, the mean change in NIS/NIS-LL scores for patients who did not exhibit meaningful worsening on the anchor (i.e., < 2-point increase on the LLF, or < 1-point increase on the PND) was subtracted from mean change in these scores for patients with a ≥ 2-point increase on the LLF/≥1-point increase on the PND [17, 20, 21]. Second, linear regression models were conducted, with change in NIS/NIS-LL scores as the outcome and change in LLF/PND as the predictor [22]. The β-coefficient from each model represents the change in NIS/NIS-LL score corresponding to a 2-point increase in LLF/1-point increase in PND. Third, receiver operating characteristic (ROC) curves were used to identify the optimal cut-off point on NIS/NIS-LL scores for classifying patients showing meaningful worsening or not based on the anchor measure (i.e., ≥2-point increase vs. < 2 increase on LLF/≥1-point increase vs. < 1 increase on PND) [20, 23–25]. The optimal cut-off point was defined using the Index of Union method, which identifies the point at which the sensitivity and specificity values are simultaneously closest to the value of the area under the curve [26].
## Distribution-based approaches
Distribution-based approaches estimate RD thresholds based on statistics that describe the variation and precision of scores, such as a scale’s standard deviation (SD) and reliability, to assess the amount of difference or change on a measure that cannot be explained by measurement error and is considered to reflect a clinically meaningful treatment effect. Three distribution-based statistics were used in the estimation of RD thresholds on all NIS and NIS-LL scores: effect size (ES), standardized response mean (SRM), and standard error of measurement (SEM). The mean of these estimates was then used as the recommended RD threshold.
The ES has long been used to interpret the magnitude of difference between groups or change over time in education, psychology, and health outcomes research [15, 27]. Group difference or change over time is measured against the standard deviation at baseline (SDBaseline). For this analysis, the ES was set to 0.5, which is considered to indicate a medium-sized effect and has been shown to closely align with estimates of RD thresholds for other clinical outcome assessments (COAs) used across many health conditions [28, 29]. This value is then multiplied by the SDBaseline.
The SRM is another statistic used to interpret group differences and change over time, this time measured against the standard deviation of change from baseline (SDChange). The SRM was set to 0.5 for this analysis, which was then multiplied by the SDChange.
The SEM captures measurement error of a scale based upon variability of scores (SDBaseline) and the scale’s reliability. Researchers have observed that the SEM of a measure had a magnitude similar to RD thresholds estimated using anchor-based approaches [30, 31]. Intra-rater reliability was used in this analysis, assessed across the two administrations of the NIS/NIS-LL conducted by the same rater (recommended to occur on consecutive days) during the baseline visit, calculated using intraclass correlation coefficient (ICC). ICC was calculated using Shrout and Fleiss’ [1, 2] model [32], a two-way random effects model appropriate for capturing intra-rater reliability when a single rater performs two assessments of the same target, when assuming that scores from raters are generalizable to the population of raters [33]. The SEM is then calculated by multiplying SDBaseline by the square root of one minus the ICC.
## Recommended RD threshold based on triangulation
Triangulation across multiple estimates, which is generally considered best practice [17, 34–36], was used to establish a recommended RD threshold for each scale. The recommended RD threshold was calculated as the mean across all estimates.
## Responder analysis
Responder analysis was conducted to evaluate the efficacy of inotersen for slowing neuropathic progression at the patient level. For each NIS and NIS-LL measure, the proportion of patients classified as responders (i.e., patients whose score increased by less than the recommended RD threshold after 65 weeks of treatment) were compared between treatment groups using odds ratios (OR) with $95\%$ confidence intervals (CI), and Fisher’s exact tests (two-tailed α) for statistical significance. Because these analyses were exploratory, no adjustments were made to the familywise Type 1 error rate for multiple comparisons.
To be consistent with the efficacy analysis of NIS/NIS-LL in the NEURO-TTR study, no imputation of missing values was performed in the primary responder analysis. As such, the responder analysis used a complete-case analysis, in which only patients with non-missing scores at the week 65 visit were included.
## Empirical cumulative distribution function (eCDF) curves
Empirical cumulative distribution function (eCDF) curves were plotted to visually represent the percentages of patients with changes in NIS or NIS-LL total score below each observed change score from baseline to week 65. Treatment differences in the percentage of patients at each threshold of change were explored by plotting separate curves for treatment and placebo groups and examining the distance between the curves on the y-axis at each point of the x-axis.
## Patient characteristics
Baseline patient characteristics and NIS/NIS-LL total and subscale scores by treatment arm are reported in Table 2. Patients in each treatment group were very similar in age and sex distribution, as well as on other key clinical characteristics such as mutation type, FAP stage, previous treatment status, cardiomyopathy, and age of symptom onset. Differences between treatment groups were not statistically significant on any patient characteristics or NIS/NIS-LL scores. Table 2Baseline patient characteristics in the NEURO-TTR trial, full analysis set ($$n = 165$$)Inotersen ($$n = 106$$)Placebo ($$n = 59$$)Age, mean (SD)59.6 (12.4)59.4 (14.1)Sex, N (%) Male75 [71]41 [70] Female31 [29]18 [30]Mutation Type, N (%) V30M54 [51]33 [56] Non-V30M52 [49]26 [44]FAP stage, N (%) Stage 1 (ambulatory without assistance)71 [67]42 [71] Stage 2 (ambulatory with assistance)35 [33]17 [29]Previous treatment statusa, N (%) Pretreatment62 [59]35 [59] No pretreatment44 [41]24 [41]Cardiomyopathy status, N (%) Cardiomyopathy70 [66]32 [54] No cardiomyopathy36 [34]27 [46]Age of symptom onset, N (%) Early onset31 [29]20 [34] Late onset75 [71]39 [66]NIS, mean (SD) Total46.6 (25.7)43.4 (24.7) Muscle Weakness21.2 (17.5)20.0 (16.1) Sensation14.4 (6.3)13.3 (6.9) Reflexes10.9 (6.0)10.1 (6.4)NIS-LL, mean (SD) Total30.1 (15.5)28.7 (16.0) Muscle Weakness13.9 (11.3)13.4 (11.0) Sensation10.2 (4.0)9.8 (4.5) Reflexes6.0 (2.3)5.6 (2.7) Abbreviations: FAP familial amyloid polyneuropathy, NIS Neuropathy Impairment Score, NIS-LL Neuropathy Impairment Score – Lower Limb, SD standard deviation aPrevious treatment with tafamidis and/or diflunisal
## Treatment differences in mean change for NIS and NIS-LL scores for clinical subgroups
Treatment differences in LS mean changes on NIS and NIS-LL total and subscale scores from baseline to 65 weeks for key clinical subgroups are presented in Figs. 1, 2, 3 and 4.Fig. 1Treatment Differences in NIS (a) and NIS-LL (b) Total Mean Change Scores from Baseline to Week 65. Note: Means in purple ink are statistically significantly different from 0 ($p \leq 0.05$). Abbreviations: CI, confidence interval; CM, cardiomyopathy; FAP, familial amyloid polyneuropathy; FAS, full analysis set; LS, least-squares; NIS, Neuropathy Impairment Score; NIS-LL, Neuropathy Impairment Score – Lower LimbFig. 2Treatment Differences in NIS (a) and NIS-LL (b) Muscle Weakness Domain Change Scores from Baseline to Week 65. Note: Means in purple ink are statistically significantly different from 0 ($p \leq 0.05$). Abbreviations: CI, confidence interval; CM, cardiomyopathy; FAP, familial amyloid polyneuropathy; FAS, full analysis set; LS, least-squares; NIS, Neuropathy Impairment Score; NIS-LL, Neuropathy Impairment Score – Lower LimbFig. 3Treatment Differences in NIS (a) and NIS-LL (b) Sensation Domain Change Scores from Baseline to Week 65. Note: Means in purple ink are statistically significantly different from 0 ($p \leq 0.05$). Abbreviations: CI, confidence interval; CM, cardiomyopathy; FAP, familial amyloid polyneuropathy; FAS, full analysis set; LS, least-squares; NIS, Neuropathy Impairment Score; NIS-LL, Neuropathy Impairment Score – Lower LimbFig. 4Treatment Differences in NIS (a) and NIS-LL (b) Reflexes Domain Change Scores from Baseline to Week 65. Note: Means in purple ink are statistically significantly different from 0 ($p \leq 0.05$). Abbreviations: CI, confidence interval; CM, cardiomyopathy; FAP, familial amyloid polyneuropathy; FAS, full analysis set; LS, least-squares; NIS, Neuropathy Impairment Score; NIS-LL, Neuropathy Impairment Score – Lower Limb Statistically significant mean differences for NIS total scores (Fig. 1a) ranged from − 11.7 (FAP Stage 1) to − 17.2 points (FAP Stage 2), with large treatment effects within all subgroups (ds ranged from − 0.89 to − 1.30). Similar findings were observed for the NIS-LL total score (Fig. 1b), with statistically significant mean differences ranging from − 6.3 (V30M, FAP Stage 1, and No Pretreatment) to − 8.3 (FAP Stage 2) and large treatment effects (ds ranged from − 0.87 to − 1.12).
For the NIS muscle weakness subscale (Fig. 2a), statistically significant mean differences ranged from − 6.3 (No CM) to − 13.5 points (FAP Stage 2), with medium-to-large effects (ds ranged from − 0.66 to − 1.35). Statistically significant mean differences on the NIS-LL muscle weakness subscale (Fig. 2b) ranged from − 3.9 (FAP Stage 1) to − 7.5 (FAP Stage 2), also with medium-to-large treatment effects (ds ranged from − 0.65 to − 1.23).
Overall, LS mean change from baseline on the NIS sensation subscale (Fig. 3a) after 65 weeks of treatment followed the same pattern as the NIS total score and muscle weakness subscale. Statistically significant differences in mean change on NIS sensation subscale scores ranged from − 2.2 (FAP Stage 1) to − 4.5 points (FAP Stage 2), with medium-to large treatment effects (ds ranged from − 0.55 to − 1.11). Treatment differences in LS mean change for the NIS-LL sensation subscale (Fig. 3b) were statistically significant for Non-V30M, FAP Stage 1, Pretreatment, No CM, and Late Onset subgroups, ranging from − 1.1 (Late Onset) to − 2.0 points (No CM), with small-to-large treatment effects (ds ranged from 0.43 to − 0.87).
Statistically significant differences in mean change on the NIS reflexes subscale (Fig. 4a) were observed for V30M, FAP Stage 1, and Early Onset subgroups, ranging from − 1.7 (V30M) to − 2.7 points (Early Onset) and with medium-to-large treatment effects (ds ranged from − 0.51 to − 0.84). For the NIS-LL reflexes subscale (Fig. 4b), statistically significant differences in mean change were observed for the V30M, FAP Stage 1, Pretreatment, CM, and Early Onset subgroups, ranging from − 0.6 (CM) to − 1.0 points (V30M), with small-to-large effects (ds ranged from − 0.44 to − 0.82).
## Responder definition estimates for NIS and NIS-LL scores
Scale properties (SDbaseline, SDchange, and ICCs for reliability) and RD threshold estimates for NIS and NIS-LL total and subscales, including the mean of the estimates (i.e., the recommended RD threshold) are presented in Table 3. ICCs for intra-rater reliability at baseline ranged from 0.88 (NIS-LL reflexes) to 0.99 (NIS total and muscle weakness). Means of RD thresholds estimated using anchor-based and distribution-based methods were similar for each measure for which both were estimated: 8.6 vs. 7.7, respectively, for NIS total; 2.4 vs 2.5 for NIS sensation; 4.8 vs. 4.6 for NIS-LL total; 3.5 vs. 3.4 for NIS-LL muscle weakness; and 1.1 vs. 1.5 for NIS-LL sensation. Among anchor-based methods, RD threshold estimates were larger for mean change methods than for linear regression and ROC curve methods. For distribution-based methods, RD threshold estimates based on ES were consistently the largest, followed by estimates based on SRM, and generally smallest for SEM. Relative magnitude of variation among RD threshold estimates was particularly evident for the NIS muscle weakness subscale, with a range of 6.6 points and a relatively large coefficient of variation (CV) of $63\%$, and for NIS-LL total and muscle weakness scores, with ranges of 5.8 and 4.2 and CVs of 52 and $53\%$, respectively. CVs were smallest for NIS sensation ($29\%$) and NIS-LL reflex subscales ($25\%$).Table 3Estimates of responder definition thresholds for NIS and NIS-LL total and subscale scoresInstrument/ScoreScale PropertiesAnchor-based estimatesDistribution-based estimatesRecommended RD thresholda ReliabilitySDbaseline SDchangeMean changeLinear regressionROCESSEMSRM NIS Total0.9925.314.611.8b 7.0b 6.9b 12.73.07.38.1 Muscle weakness0.9917.010.7–––8.51.95.45.3 Sensation0.926.54.33.3b 2.0b 1.8b 3.31.92.22.4 Reflex0.956.13.5–––3.11.31.72.0 NIS-LL Total0.9815.77.87.0c 2.0c 5.3c 7.82.23.94.7 Muscle weakness0.9811.26.35.2c 1.4c 3.8c 5.61.43.13.4 Sensation0.914.22.51.8b 1.2b 0.3b 2.11.31.21.3 Reflex0.882.41.6–––1.20.80.80.9 Abbreviations: ES effect size, LLF Lower Limb Function Test, NIS Neuropathy Impairment Score, NIS-LL Neuropathy Impairment Score – Lower Limb, PND Polyneuropathy Disability score, RD responder definition, ROC receiver operating characteristic, SD standard deviation, SEM standard error of measurement, SRM standardized response mean aRecommended RD threshold was calculated as the mean of RD estimates bRD threshold estimate was calculated using PND as the anchor measure cRD threshold estimate was calculated using LLF as the anchor measure
## Responder analysis for NIS and NIS-LL scores
Results from the responder analysis for all NIS and NIS-LL scores are shown in Figs. 5 and 6, respectively. For each NIS score, the proportion of responders (i.e., patients whose scores increased by less than the RD threshold) at week 65 was larger among patients treated with inotersen than patients who received placebo. Among patients treated with inotersen, the proportion of responders on NIS total and subscale scores ranged from 64 to $74\%$, compared to 37 to $52\%$ among patients who received placebo, with statistically significant treatment differences for total (OR = 4.4), muscle weakness (3.8), and sensation (2.7), all $p \leq 0.01$, but not for reflexes (1.9, $$p \leq 0.11$$).Fig. 5Proportion of Responders at Week 65 by Treatment Arm for NIS Total and Subscale Scores. Abbreviations: CI, confidence interval; NIS, Neuropathy Impairment Score; OR, odds ratioFig. 6Proportion of Responders at Week 65 by Treatment Arm for NIS-LL Total and Subscale Scores. Abbreviations: CI, confidence interval; NIS-LL, Neuropathy Impairment Score – Lower Limb; OR, odds ratio For each NIS-LL score, the proportion of responders at week 65 was larger among patients receiving inotersen (range: 71–$78\%$) than patients receiving placebo (38–$60\%$). Statistically significant treatment differences in proportion of responders were observed for NIS-LL total (OR = 5.6) and muscle weakness (3.6), $p \leq 0.001$ for both, but not for sensation (1.9, $$p \leq 0.09$$) or reflexes (1.9, $$p \leq 0.10$$).
## eCDF curves
The eCDF curves for changes in NIS and NIS-LL total scores from baseline to week 65 within each treatment arm are presented in Supplementary Figs. 1 and 2, respectively. A description of these results, which supports treatment benefit on both measures across the entire range of change, is reported in the Supplementary Appendix.
## Discussion
Treatment differences on mean change and response on the NIS and NIS-LL demonstrate the efficacy of inotersen for slowing neuropathic progression in patients with ATTRv-PN. Treatment benefits of inotersen were observed for NIS and NIS-LL total and muscle weakness scores, as well as NIS sensation scores, across patient subgroups based on genetic mutation, FAP disease stage, previous treatment, cardiomyopathy status, and age of symptom onset. Responder analysis demonstrated that treatment benefits on the NIS and NIS-LL were experienced by a majority of patients with ATTRv-PN treated with inotersen (64–$78\%$), which would indicate that treatment differences in mean changes at the group level were not due to outliers in either treatment arm. Patients treated with inotersen were significantly less likely to experience clinically meaningful progression of neuropathic symptoms, including muscle weakness and sensation, than patients who received placebo. For example, 72 and $78\%$ of patients treated with inotersen did not experience clinically meaningful worsening on NIS and NIS-LL total scores, respectively, compared to 37 and $38\%$ of patients who received placebo. Visual inspection of eCDF curves for change in NIS and NIS-LL total scores after 65 weeks reinforced these findings, revealing treatment differences favoring inotersen across the range of observed change scores on both measures.
Treatment benefits of inotersen on the NIS-LL sensation subscale, and for both NIS and NIS-LL reflexes subscales, were observed in some, but not all, patient subgroups. For some of these subgroups where treatment differences were not statistically significant, the magnitudes of mean differences and ESs were similar to those observed in the FAS. Thus, in some of these cases, the lack of significant treatment differences could have been at least partially an effect of insufficient statistical power to detect a true treatment effect due to small sample sizes in some subgroups. Additionally, while patients treated with inotersen showed little-to-no progression on these subscales, patients who received placebo also experienced less progression on these subscales relative to the others, as shown from analyses at both the group and patient level.
Patients with ATTRv-PN in FAP Stage 2 experienced the largest benefit from treatment with inotersen on the NIS and NIS-LL total scores and muscle weakness subscales. As would be expected, all NIS/NIS-LL total and domain scores are substantially worse at baseline for patients in FAP Stage 2 than for those in FAP Stage 1. Patients in FAP Stage 2 also showed the most progression in most of these measures, as evidenced by the larger increases in score compared to patients in FAP Stage 1 for the placebo arm on NIS total score (mean increase of 24.1 points for FAP Stage 2 vs. 15.5 points for FAP Stage 1, $$p \leq 0.031$$, $d = 0.67$), muscle weakness (18.2 vs. 9.6, $$p \leq 0.005$$, $d = 0.88$), and sensation (4.8 vs. 2.0, $$p \leq 0.017$$, $d = 0.74$), as well as for NIS-LL total score (11.6 vs. 8.6, $$p \leq 0.152$$, $d = 0.44$) and muscle weakness (10.2 vs. 5.9, $$p \leq 0.014$$, $d = 0.76$). At the same time, patients in FAP Stage 2 showed the smallest treatment benefit on the NIS and NIS-LL reflexes domain. Again, patients with FAP Stage 2 disease had substantial reflex deficits at baseline compared to patients with FAP Stage 1 disease: for those in the placebo arm, baseline mean scores were 15.2 vs 8.0 on NIS reflexes, and 7.7 vs. 4.4 on NIS-LL reflexes, both $p \leq 0.001$, both d > 1.2. However, unlike for other domains, only minimal further progression was observed at 65 weeks for patients in this subgroup receiving placebo, with a mean increase of only 1.1 points on the NIS reflexes domain (compared to 3.3 points for FAP Stage 1 patients, $$p \leq 0.040$$, $d = 0.63$) and 0.3 points on the NIS-LL reflexes domain (compared to 1.3 points for FAP Stage 1 patients, $$p \leq 0.013$$, $d = 0.78$) *These data* may indicate that patients with FAP Stage 2 disease had experienced full loss of reflexes prior to treatment, and as such received no treatment benefit. The case for earlier treatment of these patients then is quite clear.
When interpreting treatment differences across subgroups, it is important to take into account the fact that these groups are not independent, but rather some of these patient characteristics covary, leading to substantial overlap in patients across some subgroups. A post hoc analysis, using χ2 tests of independence, found statistically significant associations among several patient subgroups, particularly among those based on genetic mutation, FAP stage, cardiomyopathy status, and age of symptom onset. For example, $75\%$ of patients with V30M had early onset of symptoms, while the majority of patients with non-V30M ($57\%$) had late onset of symptoms (perhaps surprisingly, there was no statistical association between patients’ CM status and whether they had received pretreatment with tafamidis and/or diflunisal). These overlaps among patient subgroups may account for similarities in treatment effects, such as finding significant treatment benefit on reflexes for both V30M and early onset patients, but not for non-V30M and late onset patients.
The ability to evaluate the clinical significance of treatment differences on COA measures such as the NIS/NIS-LL is limited without the availability of RD thresholds. This study contributes interpretation guidelines for change in NIS and NIS-LL scores by providing empirically based estimates of RD thresholds that represent clinically meaningful progression of neuropathic symptoms. The RD thresholds recommended here are derived from multiple estimates, including three distribution-based estimates for all scores, and three anchor-based estimates for NIS total and sensation scores, and NIS-LL total, muscle weakness, and sensation subscale scores. The similarities in magnitudes of estimated values across the distribution-based and anchor-based estimates for scales in which both methods were used would seem to support that the distribution-based estimates for the remaining scales are similar to estimates that would be derived if appropriate anchors for those scales were available.
The recommended RD thresholds for the NIS and NIS-LL total scores (8.1 and 4.7 points, respectively), are much larger than the 2-point RD threshold considered in some previous studies [11–13] and represent a greater change relative to the range of these scales ($3.3\%$ versus $0.8\%$ on the NIS, and $5.3\%$ versus $2.3\%$ on the NIS-LL). The establishment of thresholds indicating meaningful change at the level of the individual patient could be used in clinical practice, as it would enable clinicians to track progression of neuropathic impairments more accurately among their patients with ATTRv-PN, including evaluation of the effectiveness of treatments for stabilization. The establishment of these thresholds will also aid in the design and interpretation of data from clinical trials of patients with ATTRv-PN; these values could be used to determine the minimum sample size for a trial to be adequately powered to detect treatment differences in trials for which the NIS or NIS-LL is a key endpoint and would allow for conducting responder analyses that would inform evaluation of treatment benefit at the patient level.
This study had some limitations that should be noted. Due to the lack of suitable anchor measures for the NIS muscle weakness scores and for NIS and NIS-LL reflexes subscale scores, RD threshold estimates for these scales were derived entirely from distribution-based methods, which are typically considered to be inferior or secondary to anchor-based methods for estimating clinically meaningful change [17, 37–39]. Another limitation concerns the amount of variability across distribution-based estimates for RD thresholds. Distribution-based estimates for RD thresholds for NIS/NIS-LL total and muscle weakness subscale scores varied considerably, with estimates based on the ES up to four times larger than estimates based on the SEM.
An additional limitation of this study is that the threshold representing a meaningful change on a measure may vary as a function of a patient’s baseline severity, particularly for scales that are non-linear like those examined here [21, 40]. In the current sample, two-thirds of patients at baseline were able to walk unassisted (FAP Stage 1), and the remainder required the assistance of walkers or canes (FAP Stage 2). Patients at earlier and less severe stages of the disease have the potential to still experience more significant decline than patients whose disease has already progressed to later stages. The experience of decline itself may also carry different meaning for patients at earlier or later stages of disease progression. Due to this, these estimates for RD thresholds cannot be assumed to represent clinically important change for patients at all levels of disease severity and progression.
## Conclusion
In conclusion, this is the first research to examine the efficacy of inotersen on NIS and NIS-LL scores among subgroups of patients with ATTRv-PN based on key clinical characteristics, such as genetic mutation, FAP stage, pretreatment status, presence of cardiomyopathy, and age of onset. Additionally, RD thresholds for NIS and NIS-LL scores in patients with ATTRv-PN were estimated using anchor-based and/or distribution-based methods, allowing for improved evaluation of treatment benefit for slowing neuropathic impairment. The mean of these estimates was given as the recommended RD threshold for the total and subscale scores of the NIS (total: 8.1 points; muscle weakness: 5.3; sensation: 2.4; and reflexes: 2.0) and NIS-LL (total: 4.7 points; muscle weakness: 3.4; sensation: 1.3; and reflexes: 0.9). Patient-level responder analyses using these RD thresholds showed that inotersen provides a clinically meaningful benefit for limiting the progression of neuropathic impairment in patients with ATTRv-PN, particularly on measures of muscle weakness and sensation loss. These results support previous evidence demonstrating the efficacy of inotersen in this patient population.
## Supplementary Information
Additional file 1.Additional file 2: Sup Fig. 1. Empirical distribution function curve for change in NIS total score from baseline to week 65 by treatment arm. Abbreviation: NIS, neuropathy impairment score. Additional file 3: Sup Fig. 2. Empirical distribution function curve for change in NIS total score from baseline to week 65 by treatment arm. Abbreviation: NIS-LL, neuropathy impairment score – lower limb.
## References
1. Gertz MA, Benson MD, Dyck PJ, Grogan M, Coelho T, Cruz M. **Diagnosis, prognosis, and therapy of transthyretin amyloidosis**. *J Am Coll Cardiol* (2015.0) **66** 2451-2466. DOI: 10.1016/j.jacc.2015.09.075
2. Ando Y, Coelho T, Berk JL, Cruz MW, Ericzon B-G, Ikeda S. **Guideline of transthyretin-related hereditary amyloidosis for clinicians**. *Orphanet J Rare Dis* (2013.0) **8** 1-18. DOI: 10.1186/1750-1172-8-31
3. Lin X, Yarlas A, Vera-Llonch M, Baranwal N, Biber J, Brown D. **Rate of neuropathic progression in hereditary transthyretin amyloidosis with polyneuropathy and other peripheral neuropathies: a systematic review and meta-analysis**. *BMC Neurol* (2021.0) **21** 70. DOI: 10.1186/s12883-021-02094-y
4. Lovley A, Raymond K, Guthrie SD, Pollock M, Sanchorawala V, White MK. **Patient-reported burden of hereditary transthyretin amyloidosis on functioning and well-being**. *J Patient Rep Outcomes* (2021.0) **5** 3. DOI: 10.1186/s41687-020-00273-y
5. Yarlas A, Gertz MA, Dasgupta NR, Obici L, Pollock M, Ackermann EJ. **Burden of hereditary transthyretin amyloidosis on quality of life**. *Muscle Nerve* (2019.0) **60** 169-175. DOI: 10.1002/mus.26515
6. Stewart M, Shaffer S, Murphy B, Loftus J, Alvir J, Cicchetti M. **Characterizing the high disease burden of transthyretin amyloidosis for patients and caregivers**. *Neurol Ther* (2018.0) **7** 349-364. DOI: 10.1007/s40120-018-0106-z
7. Luigetti M, Servidei S. **Patisiran in hereditary transthyretin-mediated amyloidosis**. *Lancet Neurol* (2021.0) **20** 21-23. DOI: 10.1016/S1474-4422(20)30397-5
8. Benson MD, Waddington-Cruz M, Berk JL, Polydefkis M, Dyck PJ, Wang AK. **Inotersen treatment for patients with hereditary transthyretin amyloidosis**. *N Engl J Med* (2018.0) **379** 22-31. DOI: 10.1056/NEJMoa1716793
9. Dyck PJB, Kincaid JC, Wiesman JF, Polydefkis M, Litchy WJ, Mauermann ML. **mNIS+7 and lower limb function in inotersen treatment of hereditary transthyretin-mediated amyloidosis**. *Muscle Nerve* (2020.0) **62** 502-508. DOI: 10.1002/mus.27022
10. Jaeschke R, Singer J, Guyatt GH. **Measurement of health status. Ascertaining the minimal clinically important difference**. *Control Clin Trials* (1989.0) **10** 407-415. DOI: 10.1016/0197-2456(89)90005-6
11. Coelho T, Maia LF, Da Martins SA, Waddington Cruz M, Planté-Bordeneuve V, Lozeron P. **Tafamidis for transthyretin familial amyloid polyneuropathy: a randomized, controlled trial**. *Neurology* (2012.0) **79** 785-792. DOI: 10.1212/WNL.0b013e3182661eb1
12. Berk JL, Suhr OB, Obici L, Sekijima Y, Zeldenrust SR, Yamashita T. **Repurposing diflunisal for familial amyloid polyneuropathy: a randomized clinical trial**. *JAMA* (2013.0) **310** 2658-2667. DOI: 10.1001/jama.2013.283815
13. Dyck PJ, Kratz KM, Lehman KA, Karnes JL, Melton LJ, O'Brien PC. **The Rochester diabetic neuropathy study: design, criteria for types of neuropathy, selection bias, and reproducibility of neuropathic tests**. *Neurology* (1991.0) **41** 799-807. DOI: 10.1212/WNL.41.6.799
14. Coutinho P, Da Martins SA, Lopes Lima J, Resende BA, Glenner G, Costa P, de Freitas A. **Forty years of experience with type I amyloid neuropathy: review of 483 cases**. *Amyloid and amyloidosis* (1980.0) 88-98
15. Cohen J. *Statistical power analysis for the behavioral sciences* (2013.0)
16. 16.Yarlas A, Lovley A, Brown D, Kosinski M, Vera-Llonch M. Responder analysis for neuropathic impairment and quality-of-life assessment in patients with hereditary transthyretin amyloidosis with polyneuropathy in the NEURO-TTR study. J Neurol. 2022;269(1):323–35.
17. Revicki D, Hays RD, Cella D, Sloan J. **Recommended methods for determining responsiveness and minimally important differences for patient-reported outcomes**. *J Clin Epidemiol* (2008.0) **61** 102-109. DOI: 10.1016/j.jclinepi.2007.03.012
18. Suhr O, Danielsson A, Rydh A, Nyhlin N, Hietala SO, Steen L. **Impact of gastrointestinal dysfunction on survival after liver transplantation for familial amyloidotic polyneuropathy**. *Dig Dis Sci* (1996.0) **41** 1909-1914. DOI: 10.1007/BF02093589
19. Dyck PJ, Turner DW, Davies JL, O'Brien PC, Dyck PJB, Rask CA. **Electronic case-report forms of symptoms and impairments of peripheral neuropathy**. *Can J Neurol Sci* (2002.0) **29** 258-266. DOI: 10.1017/S0317167100002043
20. Copay AG, Subach BR, Glassman SD, Polly DW, Schuler TC. **Understanding the minimum clinically important difference: a review of concepts and methods**. *Spine J* (2007.0) **7** 541-546. DOI: 10.1016/j.spinee.2007.01.008
21. Crosby RD, Kolotkin RL, Williams G. **Defining clinically meaningful change in health-related quality of life**. *J Clin Epidemiol* (2003.0) **56** 395-407. DOI: 10.1016/S0895-4356(03)00044-1
22. Brant R, Sutherland L, Hilsden R. **Examining the minimum important difference**. *Stat Med* (1999.0) **18** 2593-2603. DOI: 10.1002/(SICI)1097-0258(19991015)18:19<2593::AID-SIM392>3.0.CO;2-T
23. Deyo RA, Centor RM. **Assessing the responsiveness of functional scales to clinical change: an analogy to diagnostic test performance**. *J Chronic Dis* (1986.0) **39** 897-906. DOI: 10.1016/0021-9681(86)90038-X
24. Terwee CB, Peipert JD, Chapman R, Lai J-S, Terluin B, Cella D. **Minimal important change (MIC): a conceptual clarification and systematic review of MIC estimates of PROMIS measures**. *Qual Life Res* (2021.0) **30** 2729-2754. DOI: 10.1007/s11136-021-02925-y
25. Turner D, Schünemann HJ, Griffith LE, Beaton DE, Griffiths AM, Critch JN. **Using the entire cohort in the receiver operating characteristic analysis maximizes precision of the minimal important difference**. *J Clin Epidemiol* (2009.0) **62** 374-379. DOI: 10.1016/j.jclinepi.2008.07.009
26. Unal I. **Defining an optimal cut-point value in ROC analysis: an alternative approach**. *Comput Math Methods Med* (2017.0) **2017** 3762651. DOI: 10.1155/2017/3762651
27. Kazis LE, Anderson JJ, Meenan RF. **Effect sizes for interpreting changes in health status**. *Med Care* (1989.0) **27** S178-S189. DOI: 10.1097/00005650-198903001-00015
28. Norman GR, Sloan JA, Wyrwich KW. **Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation**. *Med Care* (2003.0) **41** 582-592. DOI: 10.1097/01.MLR.0000062554.74615.4C
29. Norman GR, Sloan JA, Wyrwich KW. **The truly remarkable universality of half a standard deviation: confirmation through another look**. *Expert Rev Pharmacoeconomics Outcomes Res* (2004.0) **4** 581-585. DOI: 10.1586/14737167.4.5.581
30. Wyrwich KW, Nienaber NA, Tierney WM, Wolinsky FD. **Linking clinical relevance and statistical significance in evaluating intra-individual changes in health-related quality of life**. *Med Care* (1999.0) **37** 469-478. DOI: 10.1097/00005650-199905000-00006
31. Wyrwich KW, Tierney WM, Wolinsky FD. **Further evidence supporting an SEM-based criterion for identifying meaningful intra-individual changes in health-related quality of life**. *J Clin Epidemiol* (1999.0) **52** 861-873. DOI: 10.1016/S0895-4356(99)00071-2
32. Shrout PE, Fleiss JL. **Intraclass correlations: uses in assessing rater reliability**. *Psychol Bull* (1979.0) **86** 420-428. DOI: 10.1037/0033-2909.86.2.420
33. Koo TK, Li MY. **A guideline of selecting and reporting intraclass correlation coefficients for reliability research**. *J Chiropr Med* (2016.0) **15** 155-163. DOI: 10.1016/j.jcm.2016.02.012
34. Yost KJ, Eton DT. **Combining distribution- and anchor-based approaches to determine minimally important differences: the FACIT experience**. *Eval Health Prof* (2005.0) **28** 172-191. DOI: 10.1177/0163278705275340
35. Leidy NK, Wyrwich KW. **Bridging the gap: using triangulation methodology to estimate minimal clinically important differences (MCIDs)**. *COPD* (2005.0) **2** 157-165. DOI: 10.1081/COPD-200050508
36. Revicki DA, Cella D, Hays RD, Sloan JA, Lenderking WR, Aaronson NK. **Responsiveness and minimal important differences for patient reported outcomes**. *Health Qual Life Outcomes* (2006.0) **4** 70. DOI: 10.1186/1477-7525-4-70
37. Devji T, Carrasco-Labra A, Qasim A, Phillips M, Johnston BC, Devasenapathy N. **Evaluating the credibility of anchor based estimates of minimal important differences for patient reported outcomes: instrument development and reliability study**. *BMJ* (2020.0) **369** m1714. DOI: 10.1136/bmj.m1714
38. Guyatt GH, Osoba D, Wu AW, Wyrwich KW, Norman GR. **Methods to explain the clinical significance of health status measures**. *Mayo Clin Proc* (2002.0) **77** 371-383. DOI: 10.4065/77.4.371
39. 39.U.S. Department of Health and Human Services, Food and Drug AdministrationGuidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims2009. *Guidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims* (2009.0)
40. Hays RD, Woolley JM. **The concept of clinically meaningful difference in health-related quality-of-life research. How meaningful is it?**. *Pharmacoeconomics* (2000.0) **18** 419-423. DOI: 10.2165/00019053-200018050-00001
|
---
title: 'Integrating interferon-gamma release assay testing into provision of tuberculosis
preventive therapy is feasible in a tuberculosis high burden resource-limited setting:
A mixed methods study'
authors:
- Simon Muchuro
- Rita Makabayi-Mugabe
- Joseph Musaazi
- Jonathan Mayito
- Stella Zawedde-Muyanja
- Mabel Nakawooya
- Didas Tugumisirize
- Patrick Semanda
- Steve Wandiga
- Susan Nabada-Ndidde
- Abel Nkolo
- Stavia Turyahabwe
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022101
doi: 10.1371/journal.pgph.0000197
license: CC BY 4.0
---
# Integrating interferon-gamma release assay testing into provision of tuberculosis preventive therapy is feasible in a tuberculosis high burden resource-limited setting: A mixed methods study
## Abstract
The World Health Organization recommends the scale-up of tuberculosis preventive therapy (TPT) for persons at risk of developing active tuberculosis (TB) as a key component to end the global TB epidemic. We sought to determine the feasibility of integrating testing for latent TB infection (LTBI) using interferon-gamma release assays (IGRAs) into the provision of TPT in a resource-limited high TB burden setting. We conducted a parallel convergent mixed methods study at four tertiary referral hospitals. We abstracted details of patients with bacteriologically confirmed pulmonary tuberculosis (PBC TB). We line-listed household contacts (HHCs) of these patients and carried out home visits where we collected demographic data from HHCs, and tested them for both HIV and LTBI. We performed multi-level Poisson regression with robust standard errors to determine the associations between the presence of LTBI and characteristics of HHCs. Qualitative data was collected from health workers and analyzed using inductive thematic analysis. From February to December 2020 we identified 355 HHCs of 86 index TB patients. Among these HHCs, uptake for the IGRA test was $\frac{352}{355}$ ($99\%$) while acceptability was $\frac{337}{352}$ ($95.7\%$). Of the 352 HHCs that were tested with IGRA, the median age was 18 years (IQR 10–32), 191 ($54\%$) were female and 11 ($3\%$) were HIV positive. A total of $\frac{115}{352}$ ($32.7\%$) had a positive IGRA result. Among HHCs who tested negative on IGRA at the initial visit, 146 were retested after 9 months and 5 ($3.4\%$) of these tested positive for LTBI. At multivariable analysis, being aged ≥ 45 years [PR 2.28 ($95\%$ CI 1.02, 5.08)], being employed as a casual labourer [PR 1.38 ($95\%$ CI 1.19, 1.61)], spending time with the index TB patient every day [PR 2.14 ($95\%$ CI 1.51, 3.04)], being a parent/sibling to the index TB patients [PR 1.39 ($95\%$ CI 1.21, 1.60)] and sharing the same room with the index TB patients [PR 1.98 ($95\%$ CI 1.52, 2.58)] were associated with LTBI. Implementation challenges included high levels of TB stigma and difficulties in following strict protocols for blood sample storage and transportation. Integrating home-based IGRA testing for LTBI into provision of TB preventive therapy in routine care settings was feasible and resulted in high uptake and acceptability of IGRA tests.
## Background
Tuberculosis (TB) is among the top ten causes of morbidity and mortality. In 2019, an estimated 10.0 million people fell ill with TB, and approximately 1.5 million people died from the disease in the same year [1]. Further, about a quarter of the world (approximately 2 billion persons) is infected with latent TB [2]. Among these, 10–$15\%$ will progress to active disease in their lifetime, usually within two years following exposure [1]. The risk of disease progression is increased by certain conditions e.g., age, immunosuppressive states like HIV, diabetes, cancer, and malnutrition [3, 4]. Consequently, the World Health Organization (WHO) outlined provision of TB preventive therapy (TPT) for persons at risk of developing active TB as one of the key components in its strategy to end the global TB epidemic by 2035 [5]. In line with this provision, the WHO updated its guidelines for programmatic management of latent TB infection (LTBI) to recommend TPT for HIV negative household contacts (HHCs) older than 5 years in whom active TB has been ruled out. The guidelines also recommend testing for LTBI using the interferon-gamma release assays (IGRAs), where feasible, to identify individuals who would benefit most from TPT [6].
In Uganda, the WHO symptom screen remains the main stay for ruling out latent TB among persons in close contact with patients with confirmed TB. Although immunological tests e.g., as the Tuberculin Skin Test (TST) have better sensitivity and specificity than the WHO symptom screen, their wide-spread use is limited by the need for cold chain maintenance, inter-reader variability and low specificity due to cross-reactivity with the Bacille Calmette-Guérin (BCG) vaccine and other non-tuberculous mycobacteria. The interferon-gamma release assays (IGRAs) is an alternative immunological test for the presence of LTBI which uses whole blood. This test has several advantages over the TST because its interpretation is not user dependent and the test does not cross react with BCG vaccine resulting in higher specificity [7]. We aimed to explore the feasibility of incorporating LTBI screening using an IGRA test (QuantiFERON-TB Gold Plus test (QFT-Plus) into the national algorithm for management of LTBI among HHCs older than 5 years in Uganda.
## Study setting
Between February and December 2020, we conducted a parallel convergent mixed methods study at four tertiary referral hospitals. To get a fair representation of the urban and rural settings, we selected one national referral hospital based in the capital city Kampala (Mulago national referral hospital) and three tertiary referral hospitals (RRH) based in the East (Soroti regional referral hospital), Northwest (Arua regional referral hospital), and West (Hoima regional referral hospital) of the country (Fig 1).
**Fig 1:** *A map of Uganda showing regional distribution of tertiary referral hospitals during study implementation.Link; https://data.humdata.org/dataset/uganda-administrative-boundaries-admin-1-admin-3.*
## Sample size determination
The estimated sample size required to estimate prevalence of IGRA positivity among HHCs was 385, assuming prevalence of IGRA positivity of $49\%$[8], a $95\%$ confidence (standard normal deviate, $Z = 1.96$) and margin of error of $5\%$. Inflating the sample size to account for $10\%$ of the HHCs that we assumed would have a positive symptom screen for active TB disease, the required sample size was 424 HHCs. We assumed that each index TB patient would have 4 household members [8], and thus based on the sample size of 424, 106 index TB patients were needed to accrue this sample size. However, we attained 352 HHCs from 86 index TB patients due to the limited availability of test kits.
## Selection of index TB patients
Three months prior to study commencement in February 2020, 619 patients were diagnosed with TB across study sites. Of these, 299 patients were PBCs, of whom 263 were eligible for selection of the 86 required for the study. Those excluded from the study had drug resistant TB or were children under 15 years. HHCs of drug resistant TB patients were not eligible for TPT according to the Ugandan LTBI treatment guidelines at the time of study implementation. However, WHO recommends TPT use in selected high-risk HHCs of patients with drug resistant TB based on individualized risk assessment and clinical justification [9]. We then used sampling proportionate to size to determine the number of patients to be selected from each hospital. For each hospital, we used systematic random sampling to select the required number of index TB patients. One index TB patient declined study participation due to non-disclosure to a new partner and was replaced with the next consecutive eligible index TB patient at the specific study site. Consequently, 86 index TB patients’ homes were visited maintaining sampling proportionate to size for each of the participating hospitals. All HHCs who were eligible for the study and provided informed consent were included in the study.
## Selection of household contacts
Data collection among HHCs was carried out between February and December 2020 after obtaining permission from index TB patients to visit their homes to carry out household contact tracing. The study team consisted of qualified health workers who underwent 4 days training on the study protocol and procedures prior to implementation. A team comprising of a clinician (either a nurse, clinical officer, or doctor), counselor, and laboratory technician/phlebotomist conversant with the local dialect visited the index TB patient’s home on a scheduled day and requested HHCs to consent to participate in the study.
Detailed information about the study was provided and consent or ascent for screening and enrolment into the study was sought. The study team line listed all HHCs who consented to study participation excluding those who were <5 years, with a history of TPT within the past two years or currently on TB treatment. We screened HHCs using the WHO symptom screen. This involved asking the study participants if they had cough of any duration, weight loss, fevers, and night sweats. For all HHCs without signs and symptoms of TB, we collected socio-demographic data, home-based blood sample collection for LTBI testing using QFT-Plus (manufactured by QIAGEN QIAGEN-Gruppe Germany) and HIV counselling and testing (if HIV status was reported as negative or unknown). Study team phlebotomist collected five milliliters (mls) of whole blood: four mls for the IGRA test and 1 ml for HIV 1 & 2 testing using the national testing algorithm. Blood samples collected from the capital city (Kampala) were transported in QFT-plus blood-collection-tubes within the recommended 16 hours to the central laboratory, while blood samples collected from distant study sites were kept at room temperature for utmost three hours and transported in lithium heparin tubes in ice-cold boxes maintained at 2–8 0C within 48hours to the central public health laboratory. In addition, we collected information on duration and nature of contact with the index patient. Data was collected electronically using the open data toolkit (ODK).
All asymptomatic HHCs who tested positive on IGRA test were initiated on six months of isoniazid preventive therapy (IPT) while those who tested negative on the initial IGRA test had a second home-based IGRA test performed after nine months. The repeat IGRA test was initially planned to be done at 6 months to rule out LTBI. As a result of Covid-19 travel restrictions, it was performed at 9 months. Those found to be positive on the second test were initiated on TPT.
## Qualitative data
Qualitative data was collected from four tertiary hospitals through focus group discussions (FGDs) and key informant interviews (KIIs). Two focus FDGs were held for the participants from Mulago hospital because they had large teams that would meet the criteria for holding an FGD. KII were conducted across other RRHs. The days for the FDGs were specially arranged and the participants were informed on the agenda, date, and approximate duration of the meeting prior to the meeting. The FGDs were conducted in English. All discussions were audiotaped and transcribed. Participant identifiers were not used, but individual participants provided written informed consent and were assigned codes, e.g., five group members will be assigned 01–05. Individual responses in each group were coded by item. Using a phenomenological approach, we explored the experiences of health workers focusing on their experiences during IGRA study implementation. We purposively sampled health workers who had been involved in contact tracing & implementation of IGRA. Sampling was based on purposeful maximum variation that involved distinct categories of participants like nurses, clinicians, and laboratory technicians. The majority were laboratory staff this being a predominantly lab-based test, involving home-based blood draws, packaging, and transportation of blood samples to the central laboratory. Similarly, both females and males were included in the study. The interview guide consisted of five open ended questions with probes (S1 Text) and follow up questions to create additional depth. Interview questions were developed based on additional information required, the questions were kept sufficiently broad to encourage new concepts to emerge and minimize interviewer bias. Data collection and analysis was led by an independent senior behavioural scientist (AT) who was assisted by members of the research team. We interviewed respondents until saturation was achieved.
Study definitions. For this study, we defined a bacteriologically confirmed TB patient as one with a positive Xpert MTB Rif test or positive sputum smear [10], an index case of TB as the initially identified case of new or recurrent TB in a person of any age with bacteriologically confirmed TB diagnosis, and a HHC as a person who shared the same enclosed living space as the index case for one or more nights or for frequent or extended daytime periods during the three months before the start of current treatment [6]. Finally, we defined LTBI as the presence of a positive IGRA test either on the date of first testing or on the date of second testing nine months later.
Qualitative interviews were coded using an inductive approach with descriptive thematic coding. Interview transcripts, recordings and notes were reviewed for content related to the research question and a coding frame developed with flexibility to accommodate emergent new themes as coding evolved. Using the framework, each transcript was read and reread for recurrent ideas. Codes were assigned to relevant segments of the text; similar codes were aggregated to form themes that were then used to address the research questions and develop coherent narratives [11]. The initial coding framework was developed by a senior behavioral scientist (AT) experienced in qualitative research after reviewing $5\%$ of the transcripts. Subsequent analyses of transcripts were carried out by two members of the research team (RMM and SM) who then compared and discussed their findings. Discrepancies were resolved by mutual agreement. To ensure trustworthiness, transcripts were coded independently, compared, discussed [12].
## Quantitative data
We analyzed the data in Stata version 16.1 Special Edition (StataCorp, College Station, Texas, USA). We summarized the characteristics of study participants using frequencies and percentages for categorical variables, and medians with interquartile ranges for continuous variables like age. Study outcomes: IGRA test uptake, acceptability, and IGRA test positivity was summarized as frequencies, proportions, and compared across participants’ characteristics using Chi-square test or Fisher’s Exact if expected counts are less than 5. IGRA uptake was determined as the proportion of household contacts who took the IGRA test out of all contacts screened and were eligible to take the test. A multivariable multi-level Poisson regression model with exchangeable covariance matrix was used to examine factors associated with LTBI. Robust standard errors were used to correct for overdispersion. Variables were entered into the multivariable regression analysis if they had a p-value of <0.2 at unadjusted analysis. We used variance inflation factors (VIFs) to evaluate multicollinearity in fitted models, where in VIFs >10 were indicative of severe multicollinearity. Analyses were not corrected for multiplicity given the exploratory nature of the study.
## Ethics statement
The study protocol was approved by the Mengo Hospital Research & Ethics Committee (MHREC $\frac{57}{5}$–2019) and the Uganda National Council of Science and Technology (UNCST HS 2721). All HHCs provided written informed consent and assent (for participants younger than 18 years) before undergoing any study related procedures. Similarly, written informed consent, including consent to audio-record interviews was obtained from healthcare workers who participated in the qualitative interviews.
## Results
Between February and December 2020, we visited 86 households of index TB patients and identified 355 HHCs, of whom 352 ($99.2\%$) accepted IGRA test. The median number of contacts per index TB patient were six and inter-quartile range of three and seven contacts. The proportion of indeterminate IGRA test results were $1\%$ and $11\%$ at baseline and at repeat testing on follow-up respectively. Fig 2 below shows the flow of study participants through the study.
**Fig 2:** *A consort diagram showing participants’ flow during study implementation.*
Of the 352 HHCs on whom IGRA test was done, $54\%$ were female with a median age of 18 years (IQR 10–32), $61\%$ had no employment of whom $64\%$ ($\frac{138}{214}$) were children of school going age (5 to 14 years), the majority had at-least attained primary level education (>$80\%$), while $73\%$ were HIV negative (Table 1).
**Table 1**
| Characteristics | Number (%), N = 352 |
| --- | --- |
| Gender | |
| Female | 191(54.3) |
| Male | 161(45.7) |
| Age in years, median (IQR) | 18(10–32) |
| Age groups | |
| 5–14 | 138(39.2) |
| 15–24 | 84(23.9) |
| 25–44 | 91(25.8) |
| ≥ 45 | 39(11.1) |
| Educational level | |
| | 50(14.2) |
| Primary | 206(58.5) |
| Post primary | 96(27.3) |
| Employment type¶ | |
| | 46(25.0) |
| Formal employment | 15(8.1) |
| Business | 50(.27.2) |
| Casual laborer | 21(11.4) |
| Agriculture | 52(28.3) |
| Smoking status | |
| Never | 328(93.2) |
| Ex-smoker | 5(1.4) |
| Current smoker | 19(5.4) |
| Live with a smoker (Yes) | 121(34.4) |
| HIV status | |
| Negative | 257(73.0) |
| Positive | 11(3.1) |
| Unknown* | 84(23.9) |
## Uptake and acceptability of IGRA test
IGRA test uptake was $99.2\%$ ($\frac{352}{355}$) (Fig 2). Of 352 that offered a blood sample for the IGRA test, $95.7\%$ said their phlebotomy experience was good or excellent. The $4.3\%$ that reported a bad phlebotomy experience were mainly among the younger age group, notably due to pain. Older age (P-value <0.01), level of education (P-value = 0.02) and health facility (P-value <0.01) were significantly associated with acceptability of IGRA test among HHCs (Table 2, P values unadjusted)
**Table 2**
| Factor | IGRA acceptability level | IGRA acceptability level.1 | Unnamed: 3 | Unnamed: 4 |
| --- | --- | --- | --- | --- |
| | Poor n (%) | Good or Excellent n (%) | Total participants (N) | P-value * |
| Overall | 15 (4.3) | 337 (95.7) | 352 | |
| Gender | | | | |
| Female | 8 (4.2) | 183 (95.8) | 191 | 0.94 |
| Male | 7 (4.3) | 154 (95.7) | 161 | 0.94 |
| Age group in years | | | | |
| 5–14 | 14 (10.1) | 124 (89.9) | 138 | <0.01 |
| 15–24 | 1 (1.2) | 83 (98.8) | 84 | <0.01 |
| 25–44 | 0 | 91 (100.0) | 91 | <0.01 |
| ≥ 45 | 0 | 39 (100.0) | 39 | <0.01 |
| Educational level | | | | |
| | 0 | 50 (100) | 50 | 0.02 |
| Primary | 14 (6.8) | 192 (93.2) | 206 | 0.02 |
| Post primary | 1 (1.0) | 95 (99.0) | 96 | 0.02 |
| Health facility | | | | |
| Mulago NRH | 0 | 140 (100.0) | 140 | <0.01** |
| Hoima RRH | 15 (21.1) | 56 (78.9) | 71 | <0.01** |
| Soroti RRH | 0 | 71 (100.0) | 71 | <0.01** |
| Arua RRH | 0 | 70 (100.0) | 70 | <0.01** |
## Prevalence of latent TB infection
Of 352 household contacts on whom IGRA test was done, 115 ($32.7\%$) had LTBI on the first IGRA test. Among the 231 who did not have LTBI, 146 ($63.2\%$) received repeat IGRA testing at nine months, of whom 5($3.4\%$) had LTBI. Therefore, the total number of HHCs with LTBI in this study was $\frac{120}{352}$ ($34.1\%$) (Fig 1).
## Factors associated with a positive IGRA test
At multivariable analysis, being aged ≥ 45 years compared to age 5–14 years [Prevalence ratio (PR) 2.28 ($95\%$ CI 1.02, 5.08)]; being employed as a causal labourer compared to no employment [PR 1.38 ($95\%$ CI 1.19, 1.61)]; spending time with the index TB patient everyday compared to not every day [PR 2.14 ($95\%$ CI 1.51, 3.04)]; sleeping in the same room with the index TB patient compared to sleeping in different houses [PR 1.98 ($95\%$ CI 1.52, 2.58)]; and being a parent/sibling to the index TB patients compared other relationship with index [PR 1.39 ($95\%$ CI 1.21, 1.60)] were significantly associated with having LTBI (Table 3). No severe multicollinearity was noted, all VIFs were below 10.
**Table 3**
| Factor | Number | IGRA positive, n(%) † | Unadjusted | Unadjusted.1 | Adjusted ‡ | Adjusted ‡.1 |
| --- | --- | --- | --- | --- | --- | --- |
| Factor | Number | IGRA positive, n(%) † | PR (95%CI) | P-value | PR (95%CI) | P-value |
| Overall | 352 | 115 (32.7) | | | | |
| Site location | | | | | | |
| Rural | 212 | 61 (28.8) | 1 | | 1 | |
| Urban | 140 | 54 (38.6) | 1.27 (1.08, 1.49) | <0.01 | 1.04 (0.88, 1.23) | 0.66 |
| Gender | | | | | | |
| Female | 191 | 64 (33.5) | Reference | | | |
| Male | 161 | 51 (31.7) | 1.00 (0.96, 1.05) | 0.92 | - | - |
| Age groups | | | | | | |
| 5–14 | 138 | 28 (20.3) | Reference | | Reference | |
| 15–24 | 84 | 24 (28.6) | 1.14 (0.86, 1.50) | 0.36 | 1.05 (0.73, 1.49) | 0.80 |
| 25–44 | 91 | 38 (41.8) | 1.76 (0.99, 3.13) | 0.05 | 1.38 (0.65, 2.92) | 0.40 |
| ≥ 45 | 39 | 25 (64.1) | 2.56 (1.47, 4.46) | <0.01 | 2.28 (1.02, 5.08) | 0.04 |
| Educational level | | | | | | |
| | 50 | 17 (34.0) | Reference | | | |
| Primary | 206 | 60 (29.1) | 0.90 (0.77, 1.07) | 0.23 | - | - |
| Post primary | 96 | 38 (39.6) | 1.10 (0.70, 1.72) | 0.69 | - | - |
| Employment type | | | | | | |
| | 214 | 53 (24.8) | Reference | | Reference | |
| Formal employment | 15 | 7 (46.7) | 1.61 (0.87, 2.95) | 0.13 | 1.33 (0.92, 1.91) | 0.13 |
| Business | 50 | 21 (42.0) | 1.54 (1.13, 2.10) | 0.01 | 1.21 (0.99, 1.46) | 0.05 |
| Casual laborer | 21 | 11 (52.4) | 1.81 (1.19, 2.74) | 0.01 | 1.38 (1.19, 1.61) | <0.01 |
| Peasant farming | 52 | 23 (44.2) | 1.66 (1.46, 1.88) | <0.01 | 1.19 (0.78, 1.82) | 0.42 |
| Smoking status | | | | | | |
| Never | 328 | 105(32.0) | Reference | | | |
| Ex-smoker | 5 | 3(60.0) | 1.68 (0.97, 2.90) | 0.06 | - | - |
| Current smoker | 19 | 7(36.8) | 1.01 (0.62, 1.65) | 0.97 | - | - |
| Live with a smoker | | | | | | |
| No | 231 | 73(31.6) | Reference | | | |
| Yes | 121 | 42(34.7) | 1.05 (0.74, 1.48) | 0.79 | - | - |
| HIV status | | | | | | |
| Negative | 257 | 85(33.1) | Reference | | | |
| Positive | 11 | 5(45.4) | 1.23 (0.88, 1.74) | 0.21 | - | - |
| Unknown | 84 | 25(29.8) | 0.97 (0.82, 1.16) | 0.77 | - | - |
| BCG scar present | | | | | | |
| No | 35 | 14(40.0) | Reference | | | |
| Yes | 317 | 101(31.9) | 0.85 (0.68, 1.06) | 0.15 | - | - |
| Time spent with TB index | | | | | | |
| Not every day | 15 | 2 (13.3) | Reference | | Reference | |
| Everyday | 337 | 113 (33.5) | 2.84 (1.74, 4.66) | <0.01 | 2.14 (1.51, 3.04) | <0.01 |
| Contact proximity with TB index Sleeps in | | | | | | |
| #Different house | 197 | 46(23.3) | Reference | | Reference | |
| Same house / different room | 86 | 28(32.6) | 1.27 (0.94, 1.71) | 0.12 | 1.11 (0.76, 1.61) | 0.60 |
| Same room | 69 | 41(59.4) | 2.26 (1.78, 2.86) | <0.01 | 1.98 (1.52, 2.58) | <0.01 |
| Relationship with index TB case | | | | | | |
| Others | 112 | 24(21.4) | Reference | | Reference | |
| Parent/Sibling | 221 | 78 (35.3) | 1.53 (1.16, 2.00) | <0.01 | 1.39 (1.21, 1.60) | <0.01 |
| Spouse | 19 | 13 (68.4) | 2.68 (1.57, 4.56) | <0.01 | 1.28 (0.74, 2.23) | 0.38 |
## Index TB patients’ information
Information from 53 out of 86 index TB patients was accessed at the health facilities, of whom, majority were male ($73.6\%$) with a median age of 32, $98\%$ were new cases of bacteriologically confirmed pulmonary tuberculosis (Table 4).
**Table 4**
| Unnamed: 0 | Number (%) N = 53* |
| --- | --- |
| Sex | |
| Female | 14(26.4) |
| Male | 39(73.6) |
| Age | |
| Median age in years (interquartile range) | 32 (26–46) |
| Age group (years) | |
| 15–17 | 3(5.6) |
| 18–24 | 9(17.0) |
| 25–34 | 17(32.1) |
| 35–44 | 10(18.9) |
| ≥45 | 14(26.4) |
| TB disease classification | |
| Bacteriologically confirmed pulmonary TB | 53(98.1) |
| TB patient type | |
| New | 46(86.8) |
| Relapse | 4(7.6) |
| Re-treatment after failure | 3(5.7) |
| HIV status | |
| Negative | 47(88.7) |
| Positive | 6(11.3) |
## Characteristics of the qualitative arm participants
In March 2020, we carried out two focus group discussions (FGDs) each with five participants, and 14 key informant interviews (KII) giving a total of 24 healthcare worker participants in this study. Thirteen of these ($54\%$) were male. There were seven laboratory technicians, five nurses, two counsellors, three community healthcare workers, two clinical officers, two doctors, one laboratory scientist, one quantitative economist and one physician.
Several key themes emerged from the data regarding the health workers experiences, challenges, and barriers to implementation of LTBI screening using IGRA.
## Positive health worker experience during implementation of LTBI screening using IGRA
Multi-disciplinary teams coupled with the eagerness and self-motivation of the health workers to find out the burden of LTBI among HHCs of index TB patients enabled the smooth implementation of IGRA testing in the community.
Importance of IGRA and its usefulness versus symptom screen as the current standard of care. The healthcare workers said that LTBI screening using IGRA helped them better appreciate the importance of TB preventive therapy. The exercise also helped them realize the importance of testing before treating for LTBI so as to target the limited supplies of TPT to those who need it most and lessen the chances of toxicities.
## Barriers to implementation of LTBI screening using IGRA
The healthcare workers reported some challenges with homebased screening with IGRA. These included access, poor household ventilation, lack of privacy, stigma, sample storage and transportation to the central laboratory for testing.
## Stigma
Whereas index patients were welcoming and comfortable with the visiting study teams, some of their household contacts were concerned about the neighbors’ perceptions as to why the study teams were visiting those particular homes in the villages. Thus, the study teams were invited to sit inside some poorly ventilated houses of the index patients houses. This was to prevent the neighbors from seeing what was going on which could have resulted into stigma.
## Fear of injection
Most people were fearing the injection. They thought it was taking off sputum. They were like, “but for us we know TB is tested through sputum, and now you people are coming with injections…” (KII_NURSING OFICER_Hoima_12)
## Poor ventilation
Due to stigma, all activities had to be carried out inside the houses of the index TB patient. Majority of which had poor ventilation with no open widows. History of recurrent TB disease was noted in some of the homes.
## Sample storage and transportation
The test had to be transported to one central laboratory in the capital city. This limited time flexibility between sample collection, incubation, and analysis. Moreover, those processes had to be done under stringent conditions to ensure accurate results. The long distance increased the turn-around time & cost of the test. A dedicated team was required to ensure these timelines are met.
The participants also reported difficulty in transportation of samples from recipients’ homes to the laboratory.
## Community response to the IGRA test
The healthcare workers found that the community was very accepting of IGRA testing. Community members who were contacts of confirmed cases were anxious to know if they were infected with TB while even those who were not contacts of the index case requested to be tested.
Even among child participants, IGRA uptake was very high. The community was receptive of needle pricks.
Preferred approach to LTBI screening using IGRA. The acceptability of the test was due, in part, to the fact that a homebased screening approach was employed which such that no transport costs were incurred in the process of receiving care.
Willingness to pay for IGRA test. The data shows that clients were not willing to pay for the test.
## Discussion
Using a parallel convergent mixed methods design, we determined the prevalence of latent TB and health worker experiences in using IGRA home-based screening for LTBI. We found a LTBI prevalence of $32.7\%$. The risk factors associated with latent TB included being aged ≥ 45, being in formal employment or casual laborer, longer time spent with the index case, more intimate relationship with index case (parents or siblings) and sharing the same bedroom as the index case. The uptake and acceptability of the IGRA test among HHC of index TB patients was high at $99\%$ and $95.7\%$ respectively. Further, the test was viewed as useful by the health workers in detecting LTBI and bringing to light its true burden in our setting. Our study used a door-to-door approach, which provided the perspectives at the community level. It was found that the communities were receptive to the intervention. However, the challenges noted during IGRA implementation included difficult access to homes due to the poor state of roads in the slum dwellings, stigma, fear of injection, poor ventilation, challenges with sample storage and sample transportation, and delay in sample delivery.
The uptake and acceptability of the IGRA test in this study was generally high, however those who decline were mainly children aged 5–14 years. Refusal was uncommon ($1\%$), similar to another study done amongst immigrants [13]. The main reason cited for refusal to take the test was pain from the needle prick.
The home-based approach to LTBI testing using IGRA could explain the high acceptability rates observed in our study. Similar home-based approaches in TB HHC investigation using other techniques like portable molecular diagnostics (portable GeneXpert-Instrument) [14] and home-based sputum collection [15] have showed that home-based approaches are convenient, trustworthy and help to overcome barriers to clinic-based testing like waiting time, distance and transportation costs.
The prevalence of LTBI determined in this study was lower than that reported by other studies in Uganda which reported prevalence that ranged from $51\%$ to $65\%$ [8, 16, 17]. Several reasons could explain the observed difference. Previous studies were carried out in urban or peri-urban setting which tend to have more crowding and poor ventilation which encourage transmission of TB infection. In addition, a study by Kizza et al, used TST rather than IGRA [8]. TST has a lower specificity than IGRA due to cross reactivity with BCG antigen and environmental non-tuberculous mycobacteria. Furthermore, our study population was a predominantly young population with the majority being between the (5–14) year age bracket compared to other studies where HHCs were older [8].
The factors associated with latent TB identified in our study were similar to those reported elsewhere. In India and China, LTBI was associated with increasing age and being in close contact to a case of tuberculosis [18, 19]. In addition, our study found that being employed as a casual labor was associated with a higher risk for LTBI positivity [19]. Older age increases the cumulative lifetime exposure to Mycobacterium tuberculosis, while being employed increases the risk for latent TB infection acquisition outside the household setting [19].
Similar to our study findings, other studies found that proximity of contact to a TB index case was associated with an LTBI positivity [20, 21]. In addition, our study showed that IGRA positivity was associated with increasing the time spent with the index TB patient which is similar to what was found in India [22]. Presence of a BCG scar was not found to be statistically significant in our study, however, previous other studies have found BCG to be a protective factor against LTBl [23, 24]. This could be due to differences in the prevalence of TB between the study settings.
Despite the challenges experienced, IGRA based latent TB screening was well received by the community largely because it was free, and it was delivered at home. Free home-based latent TB screening overcame two of the major barriers to IGRA testing that includes transport costs and the need to pay for the test. A study carried out in Uganda to assess barriers to TPT uptake found that having to attend clinic refill visits and the need to pay for the service decreased participants willingness to initiate TPT [25].
Similar to findings elsewhere in the Netherlands [26] and Brazil [27], TB stigma was a major barrier to LTBI services. Increased knowledge and awareness of LTBI led to an increase in expressed stigma [27]. This was also the case in our setting were HHCs did not want the health worker teams to carry out any procedures from outside the house as they expressed fear of stigma from neighbors. To overcome these challenges, there is need to develop strategies that address stigma at the community level to help those affected to resist TB related stigma through counselling, creating TB support clubs, and community dialogues [28]. Strategies to decentralize laboratory testing capability will help address challenges of sample storage and transportation. Further, due to the current COVID-19 pandemic, only $63.2\%$ received repeat IGRA testing at 9 months. This period was characterised by Index TB patients and their HHCs moving away from urban residences to rural areas for socio-economic reasons. Innovative patient-centred approaches need to be developed and evaluated as these will become increasingly relevant [1].
The study had several strengths and some limitations. We had regional representation from different parts of the country and so the findings are likely to be representative of different settings across the country. The study combined both qualitative and quantitative methods of data collection, which elucidated different perspectives of the study variables e.g., acceptability of the IGRA test, associated risk factors and barriers to implementation. Further it enabled triangulation of methods, data sources as well as researchers that enabled better understanding of the research questions.
One limitation of our study was the that the study population was heavily skewed towards children, given that children constituted the majority of HHCs in the study setting. The low sensitivity of IGRA in extremes of age [29] was mitigated by retesting at 9 months of follow up. In addition, this enabled identification of those who seroconvert later to be prioritized for LTBI treatment. Furthermore, our study had no age specific measures for acceptability, future studies should consider age-specific measures of acceptability to assess any differences in acceptability of the IGRA test among different age groups. Another study limitation is that perspectives in the qualitative research analysis were those of health personnel. More studies that explore the perspectives of TB household contacts need to be explored. Further, during the second study home visit, high rates of indeterminate IGRA results were reported as compared to first home visit. This may have been due to blood sample transportation delays due to political riots during pre-election campaigns.
Finally, the accrued sample size fell short of the estimated sample size due limited availability of test kits. However, our study’s sample size was larger than for any prior study in Uganda and the study was spread across the country. Therefore, our study still gives the best estimate of the prevalence of LTBI in Uganda.
## Conclusion
Integrating home-based IGRA screening for LTBI into provision of TPT in routine care settings, resulted in high uptake and acceptability and was therefore feasible in a resource-limited setting. Addressing challenges identified will be critical to scaling up IGRA based LTBI screening.
## References
1. 1World Health Organisation (WHO). Global tuberculosis report. 2020 [Available from: https://apps.who.int/iris/bitstream/handle/10665/336069/9789240013131-eng.pdf
2. 2World Health Organisation (WHO). Use of high burden country lists for TB by WHO in the post-2015 era. 2015 [Available from: https://www.who.int/tb/publications/global_report/high_tb_burdencountrylists2016-2020.pdf.
3. Houben RM, Dodd PJ. **The Global Burden of Latent Tuberculosis Infection: A Re-estimation Using Mathematical Modelling**. *PLoS medicine* (2016.0) **13** e1002152. DOI: 10.1371/journal.pmed.1002152
4. Workneh MH, Bjune GA, Yimer SA. **Prevalence and associated factors of tuberculosis and diabetes mellitus comorbidity: A systematic review**. *PloS one* (2017.0) **12** e0175925. DOI: 10.1371/journal.pone.0175925
5. 5World Health Organisation (WHO). The END TB strategy 2015 [Available from: https://www.who.int/tb/strategy/End_TB_Strategy.pdf.
6. 6World Health Organisation (WHO). Latent tuberculosis infection: Updated and consolidated guidelines for programmatic management. 2018 [Available from: https://apps.who.int/iris/bitstream/handle/10665/260233/9789241550239-eng.pdf.
7. Andersen P, Munk ME, Pollock JM, Doherty TM. **Specific immune-based diagnosis of tuberculosis**. *Lancet (London, England)* (2000.0) **356** 1099-104. DOI: 10.1016/s0140-6736(00)02742-2
8. Kizza FN, List J, Nkwata AK, Okwera A, Ezeamama AE, Whalen CC. **Prevalence of latent tuberculosis infection and associated risk factors in an urban African setting**. *BMC infectious diseases* (2015.0) **15** 165. DOI: 10.1186/s12879-015-0904-1
9. 9World Health Organisation (WHO). WHO consolidated guidelines on tuberculosis: module 1: prevention: tuberculosis preventive treatment 2020 [Available from: https://www.who.int/publications/i/item/9789240001503.
10. 10Ministry of Health. Uganda National Tuberculosis and Leprosy Programme. Manual for management and control of tuberculosis. 2017 [3rd Edition:[Available from: https://www.health.go.ug/cause/manual-for-management-and-control-of-tuberculosis-and-leprosy-3rd-edition-march-2017/.
11. Braun V, Clarke V. **Using thematic analysis in psychology**. *Qualitative Research in Psychology* (2006.0) **3** 77-101
12. Pamela B, Jack S. **Qualitative Case Study Methodology: Study Design and Implementation for Novice Researchers**. *Qual Rep* (2008.0) **13** 544-59
13. Dewan PK, Grinsdale J, Liska S, Wong E, Fallstad R, Kawamura LM. **Feasibility, acceptability, and cost of tuberculosis testing by whole-blood interferon-gamma assay**. *BMC infectious diseases* (2006.0) **6** 47. DOI: 10.1186/1471-2334-6-47
14. Medina-Marino A, de Vos L, Bezuidenhout D, Denkinger CM, Schumacher SG, Shin SS. **"I got tested at home, the help came to me": acceptability and feasibility of home-based TB testing of household contacts using portable molecular diagnostics in South Africa**. *Tropical medicine & international health: TM & IH* (2021.0) **26** 343-54. DOI: 10.1111/tmi.13533
15. Armstrong-Hough M, Ggita J, Turimumahoro P, Meyer AJ, Ochom E, Dowdy D. **’Something so hard’: a mixed-methods study of home sputum collection for tuberculosis contact investigation in Uganda**. *The international journal of tuberculosis and lung disease: the official journal of the International Union against Tuberculosis and Lung Disease* (2018.0) **22** 1152-9. DOI: 10.5588/ijtld.18.0129
16. Biraro IA, Kimuda S, Egesa M, Cose S, Webb EL, Joloba M. **The Use of Interferon Gamma Inducible Protein 10 as a Potential Biomarker in the Diagnosis of Latent Tuberculosis Infection in Uganda**. *PloS one* (2016.0) **11** e0146098. DOI: 10.1371/journal.pone.0146098
17. Odera S, Mureithi M, Aballa A, Onyango N, Anzala O, Oyugi J. **Latent tuberculosis among household contacts of pulmonary tuberculosis cases in Nairobi, Kenya**. *The Pan African medical journal* (2020.0) **37** 87. DOI: 10.11604/pamj.2020.37.87.21102
18. Dolla CK, Padmapriyadarsini C, Thiruvengadam K, Lokhande R, Kinikar A, Paradkar M. **Age-specific prevalence of TB infection among household contacts of pulmonary TB: Is it time for TB preventive therapy?**. *Transactions of the Royal Society of Tropical Medicine and Hygiene* (2019.0) **113** 632-40. DOI: 10.1093/trstmh/trz049
19. Chen C, Zhu T, Wang Z, Peng H, Kong W, Zhou Y. **High Latent TB Infection Rate and Associated Risk Factors in the Eastern China of Low TB Incidence.**. *PloS one* (2015.0) **10** e0141511. DOI: 10.1371/journal.pone.0141511
20. Muyindike WR, Fatch R, Cheng DM, Emenyonu NI, Ngabirano C, Adong J. **Tuberculin skin test positivity among HIV-infected alcohol drinkers on antiretrovirals in south-western Uganda**. *PloS one* (2020.0) **15** e0235261. DOI: 10.1371/journal.pone.0235261
21. Campbell JR, Krot J, Elwood K, Cook V, Marra F. **A systematic review on TST and IGRA tests used for diagnosis of LTBI in immigrants**. *Molecular diagnosis & therapy* (2015.0) **19** 9-24. DOI: 10.1007/s40291-014-0125-0
22. Mumpe-Mwanja D, Verver S, Yeka A, Etwom A, Waako J, Ssengooba W. **Prevalence and risk factors of latent Tuberculosis among adolescents in rural Eastern Uganda**. *African health sciences* (2015.0) **15** 851-60. DOI: 10.4314/ahs.v15i3.20
23. Oni T, Gideon HP, Bangani N, Tsekela R, Seldon R, Wood K. **Risk factors associated with indeterminate gamma interferon responses in the assessment of latent tuberculosis infection in a high-incidence environment**. *Clinical and vaccine immunology: CVI* (2012.0) **19** 1243-7. DOI: 10.1128/CVI.00166-12
24. Soysal A, Millington KA, Bakir M, Dosanjh D, Aslan Y, Deeks JJ. **Effect of BCG vaccination on risk of Mycobacterium tuberculosis infection in children with household tuberculosis contact: a prospective community-based study**. *Lancet (London, England)* (2005.0) **366** 1443-51. DOI: 10.1016/S0140-6736(05)67534-4
25. Kadota JL, Katamba A, Musinguzi A, Welishe F, Nabunje J, Ssemata JL. **Willingness to accept reimbursement for visits to an HIV clinic for tuberculosis preventive therapy**. *The international journal of tuberculosis and lung disease: the official journal of the International Union against Tuberculosis and Lung Disease* (2020.0) **24** 729-31. PMID: 32718409
26. Spruijt I, Haile DT, van den Hof S, Fiekert K, Jansen N, Jerene D. **Knowledge, attitudes, beliefs, and stigma related to latent tuberculosis infection: a qualitative study among Eritreans in the Netherlands**. *BMC public health* (2020.0) **20** 1602. DOI: 10.1186/s12889-020-09697-z
27. Rebeiro PF, Cohen MJ, Ewing HM, Figueiredo MC, Peetluk LS, Andrade KB. **Knowledge and stigma of latent tuberculosis infection in Brazil: implications for tuberculosis prevention strategies**. *BMC public health* (2020.0) **20** 897. DOI: 10.1186/s12889-020-09053-1
28. Courtwright A, Turner AN. **Tuberculosis and stigmatization: pathways and interventions**. *Public health reports (Washington, DC: 1974)* (2010.0) **125** 34-42. DOI: 10.1177/00333549101250S407
29. Tebruegge M, de Graaf H, Sukhtankar P, Elkington P, Marshall B, Schuster H. **Extremes of age are associated with indeterminate QuantiFERON-TB gold assay results**. *Journal of clinical microbiology* (2014.0) **52** 2694-7. DOI: 10.1128/JCM.00814-14
|
---
title: Aging and hypertension among the global poor—Panel data evidence from Malawi
authors:
- Iliana V. Kohler
- Nikkil Sudharsanan
- Chiwoza Bandawe
- Hans-Peter Kohler
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022104
doi: 10.1371/journal.pgph.0000600
license: CC BY 4.0
---
# Aging and hypertension among the global poor—Panel data evidence from Malawi
## Abstract
Hypertension is a rapidly growing disease burden among older persons in low-income countries (LICs) that is often inadequately diagnosed and treated. Yet, most LIC research on hypertension is based on cross-sectional data that does not allow inferences about the onset or persistence of hypertension, its correlates, and changes in hypertension as individuals become older. The Mature Adults Cohort of the Malawi Longitudinal Study of Families and Health (MLSFH-MAC) is used to provide among the first panel analyses of hypertension for older individuals in a sub-Saharan LIC using blood pressure measurements obtained in 2013 and 2017. We find that high blood pressure is very common among mature adults aged 45+, and hypertension is more prevalent among older as compared to middle-aged respondents. Yet, in panel analyses for 2013–17, we find no increase in the prevalence of hypertension as individuals become older. Hypertension often persists over time, and the onset of hypertension is predicted by factors such as being overweight/obese, or being in poor physical health. Otherwise, however, hypertension has few socioeconomic predictors. There is also no gender differences in the level, onset or persistence in hypertension. While hypertension is associated with several negative health or socioeconomic consequences in longitudinal analyses, cascade-of-care analyses document significant gaps in the diagnosis and treatment of hypertension. Overall, our findings indicate that hypertension and related high cardiovascular risks are widespread, persistent, and often not diagnosed or treated in this rural sub-Saharan population of older individuals. Prevalence, onset and persistence of hypertension are common across all subgroups—including, importantly, both women and men. While age is an important predictor of hypertension risk, even in middle ages 45–55 years, hypertension is already widespread. Hypertension among adults aged 45+ in *Malawi is* thus more similar to a “generalized epidemic” than in high-income countries where cardiovascular risk has strong socioeconomic gradients.
## Introduction
Hypertension is not only widespread in higher-income countries. To the contrary, levels of hypertension are actually often higher in sub-Saharan African (SSA) and low- and middle-income countries (LMICs) as compared to high-income countries (HICs) (Fig 1A). Among adults over the age of 18 in SSA as a whole, despite an overall young age structure, more than $20\%$ of men and women have hypertension (systolic BP ≥ 140 mmHg or diastolic BP ≥ 90 mmHg) [1, 2]. However, the diagnosis and treatment of this widespread hypertension is often inadequate [3]. Moreover, the high prevalence of hypertension in SSA and/or LMIC contrasts with an often low prevalence of “classic” risk factors (i.e., high rates of obesity, unhealthy life styles and reduced physical activity) [1, 2, 4], indicating that hypertension in LIC has partially different causes than hypertension in high-income contexts. In global comparison, for example, women in SSA have among the highest prevalence of hypertension, while they also are among the leanest in the world (Fig 1A+1B).
**Fig 1:** *Prevalence of high blood pressure and obesity by sex across different regions in the world (2015, in %, age-standardized).Source: Own calculations based on NCD-RisC [1]. High blood pressure is defined by NCD-RisC as systolic BP ≥ 140 mmHg or diastolic BP ≥ 90 mmHg. Obesity is defined as BMI ≥ 30. Prevalence of high blood pressure and obesity are age-standardized so that they are better comparable across countries.*
This prevalence and number of individuals in need of hypertension care is expected to grow substantially in LMICs [2, 5–7]. This rise in hypertension is related to a broader shift in the burden of disease in LMICs towards non-communicable diseases (NCDs) that have become leading causes of adult morbidity and mortality in sub-Saharan African (SSA) countries [8]. The shifting disease burden towards NCDs, along with a gradual aging of global populations [9], has been described as a frontier in global health research [10] that is likely to regain focus once the Covid-19 pandemic has receded [7, 11].
Despite the importance of understanding and addressing hypertension in SSA, an important limitation of existing hypertension research is that the findings are largely based on cross-sectional data. Specifically, cross-sectional data are not informative about the incidence of hypertension as individuals age, and they cannot differentiate persistent hypertension from temporary episodes of elevated blood pressure. For example, age-gradients of hypertension observed in cross-sectional data can arise from cohort differences, and such age gradients thus may not reflect changes in hypertension prevalence as individuals become older. Hence, if age-based increases in hypertension are used to create thresholds for targeting care efforts, cross-sectional data may misguide clinical decisions. Cross-sectional studies are also limited in their ability to identify the extent to which enhancing awareness and knowledge about blood pressure will improve blood pressure levels over time [12, 13]. Ultimately, only analyses of longitudinal data can provide a better understanding of hypertension as a dynamic process that is interrelated with individual aging and socioeconomic changes. Yet, very few such studies have been conducted in SSA, or more broadly in low-income countries (LICs).
The purpose of this paper is to provide among the first longitudinal analyses of widespread hypertension among aging global poor using panel data from a low-income country. The individuals that we study had lifecourse experiences that are typical for the aging global poor: They lived most of their lives with per capita incomes of less than $1/day [14]. Fertility was high in these cohorts [15], the status of women was generally low, and income generation through subsistence agriculture and small scale agriculture entailed relatively high levels of physical activity [16]. Older study participants were born when under-5 mortality was almost one out of three [5], and period life expectancies at birth were around 35 years in the 1950s, and remained below 46 years until the early 2000s [5]. All members of our study population have survived sustained poverty, frequent undernutrition, and epidemics. Health at older ages is often poor, and individuals experience accelerated aging with an early onset of morbidity across multiple domains [8, 17–19].
Due to the lack of longitudinal studies of hypertension in such low income populations, large gaps remain the literature about some basic aspects of hypertension risk during the life-course. Key research questions guiding our analyses for instance include: *In this* LIC population, do blood pressure and the prevalence of hypertension increases as individuals get older over time? What are socioeconomic predictors of the high blood pressure, or the onset of hypertension? Do patterns of hypertension and their predictors differ between men and women? What are possibly life-course consequences in terms of mortality or physical/mental health of hypertension? Do individuals in our LIC study population have knowledge about the health risk associated with hypertension? Finally, what can analyses of the cascades of care tell us about the diagnosis and treatment (or lack thereof) for individuals affected by hypertension?
While our paper does not directly test hypotheses related to these research questions, our results are useful for targeting hypertension prevention and care efforts to the populations with the most need and towards risk factors with the greatest potential benefit. For example, our results inform decisions about whether hypertension care should be targeted by age and sex, and they help clarify the degree to which intervention efforts should focus on addressing behavioral risk factors or improved medical treatment of hypertension.
## Data
Data for this study come from the Mature Adults Cohort of the Malawi Longitudinal Study of Families and Health (MLSFH-MAC), which to our best knowledge is currently the only population-based cohort study of adults in a SSA LICs that collects longitudinal information on blood pressure along with data on hypertension prevalence, incidence, treatment, awareness and knowledge [16, 20]. The MLSFH is based in Malawi, one of the poorest countries in the world that is ranked 174 out of 189 countries in terms of the Human Development Index in 2019 (HDI2019 =.485) [21] and has a per-capita GDP equal to about $5\%$ of the global average [22]. In rural areas, where our study is based and most Malawians ($85\%$) live, about $60\%$ of the rural population was considered poor in $\frac{2016}{17}$, thus having a total consumption that does not provide 2,400 calories per day per person [23].
The MLSFH-MAC study population consists of adults aged 45+ years living in rural areas in three districts in Malawi (Mchinji, Balaka, Rumphi). Baseline enrollment in 2012 was 1,266 individuals, and the cohort was followed up in 2013, 2017, and 2018, when each time additional eligible MLSFH respondents reaching age 45 years were enrolled in MLSFH-MAC. An MLSFH-MAC Cohort Profile provides detailed information about sampling procedures, study instruments, attrition/follow-up rates [16], and summary statistics for the study population are provided in S1 and S2 Tables. Life expectancy at birth is currently ≈63 years, and life expectancy at age 45 is currently ≈28 years [5]. Healthy life expectancy at age 45 is estimated around 22 years [8], and older adults can expect to live a large proportion of their later years subject to physical limitations on their activities [24].
We focus on MLSFH-MAC respondents with longitudinal measurements of blood pressure in 2013 (baseline) and the 2017 follow-up. The mean age is about 60 years in 2017, with $37\%$ of respondents being age 45–54 years and $14\%$ age 75+ years. $60\%$ of the study population is female, and one quarter is Muslim. The study population is characterized by low levels of formal schooling ($30\%$ of respondents have no formal education, and $64\%$ have only primary education). $58\%$ of female and $94\%$ of male respondents were married in 2017, and about half of the respondents resided in a house with a metal/tiled roof. The HIV prevalence for the MLSFH-MAC cohort was $8\%$ in 2017, with HIV+ individuals concentrated at ages 45–49 years. The majority of respondents ($66\%$) have body mass index (BMI) that is within the normal range (18 ≤ BMI ≤ 25) or are underweight ($17\%$ have BMI below 18). While only a small fraction of the study population is overweight ($12\%$) or obese ($4.7\%$), women are more likely to fall into these high risk BMI categories.
## Measurement of blood pressure in MLSFH-MAC
Blood pressure (BP) in MLSFH-MAC was measured in 2013 and 2017 following the protocol developed by the Health and Retirement Study (HRS) [25]. Three BP measurements on the left arm were taken about 1 minute apart in a sitting position in a chair, using an upper arm automated BP monitor (Omron HEM-780N). BP data is available for 1,229 respondents aged 45+ in 2013 and 1,516 in 2017. Longitudinal BP measurements are available for 1,065 MLSFH-MAC participants. Respondents who measured BP above 160 mmHg systolic and/or 110 mmHg diastolic were given a referral for follow up with a doctor or medical professional, but were not provided any other information about blood pressure. Virtually all respondents (e.g., $99\%$ of participants in 2017) agreed to the BP measurement. Prior to the BP measurement, respondents were asked if they have been diagnosed with hypertension by a doctor or medical personnel in the last two years before the survey, and if they are currently taking medication for reducing BP. Respondents’ knowledge about the risks associated with high BP was elicited by asking why it is important for individuals to know their blood pressure, and the importance that respondents attributed to knowing their BP was assessed through a question asking how far (long) they would be willing to walk to a clinic to obtain a (free) BP measurement. Additionally, the 2017 survey included questions on knowledge about BP and associated risk factors, and a set of questions focused on the ability of respondents to recognize symptoms associated with high BP.
## Approach
Individuals are defined as pre-hypertensive if 120 ≤ BPsys < 140 or 80 ≤ BPdia < 90, as hypertensive stage 1 if 140 ≤ BPsys < 160 or 90 ≤ BPdia < 100 and hypertensive stage 2 if BPsys ≥ 160 or BPdia ≥ 100, where BPsys denotes systolic and BPdia diastolic blood pressure. Onset of hypertension is defined as having normal blood pressure or pre-hypertension at baseline [2013], and having Stage 1 or 2 hypertension at follow-up in 2017. In $85\%$ of cases, systolic BP determines the hypertension classification. Persistent hypertension is defined as being Stage 1 or Stage 2 hypertensive at both baseline [2013] and follow-up [2017]. Persistent Stage 2 hypertension is defined as being Stage 2 hypertensive in both 2013 and 2017.
The “cascade of care” concept [26, 27] is used to measure linkage to care, health seeking behavior, and effective treatment for individuals with hypertension. The cascade represents the proportion of individuals who reach each separate stage of care, conditional on being included in the previous stage, and it is useful to identify the stages of the disease care cascade where the management of a disease fails. For hypertension care, the following cascade stages are defined: [1] being hypertensive ($47\%$ and $46\%$ of all respondents in 2013 and 2017 respectively), either based on measured BP ($90\%$ and $84\%$ of hypertensive cases in 2013 and 2017 respectively) and/or a reported diagnosis ($22\%$ or $39\%$ of hypertensive cases in 2013 and 2017 respectively); [2] diagnosis of hypertension by a doctor or medical personnel within 2 years prior to the survey; [3] treatment of high BP with medication, and [4] treatment for high BP and controlled BP (= measured BP does not indicate hypertension at time of survey).
We employ standard regression analysis and focus on the results from the longitudinal analyses of hypertension during 2013–17, which are enabled by the longitudinal cohort information available in MLSFH-MAC. Related cross-sectional or complementary results are reported in the Supplemental Materials.
## Ethics approval
This research based on the MLSFH-MAC was approved by the IRB at the University of Protocol #815016) and the College of Medicine (now Kamuzu University of Health Sciences) in Malawi (COMREC Protocol #P$\frac{.01}{12}$/1165).
## Prevalence of hypertension, hypertension incidence and persistent hypertension
At baseline in 2013, $42.5\%$ of mature adults were hypertensive ($23.5\%$ Stage 1, $19\%$ Stage 2; (Table 1, Column 1). In multivariate analyses, prevalence of hypertension is higher at older ages, is similar between women and men, varies across MLSFH study regions and is not associated with having obtained formal schooling (S3 Table).
**Table 1**
| Unnamed: 0 | 2013 Prevalence of Hypertension | 2013 Prevalence of Hypertension.1 | Change in Hypertension 2013–17 (Longitudinal Sample) | Change in Hypertension 2013–17 (Longitudinal Sample).1 | Change in Hypertension 2013–17 (Longitudinal Sample).2 |
| --- | --- | --- | --- | --- | --- |
| | Baseline Samplea (1) | Longitud. Sampleb (2) | Onset Hypertension (3) | Persistent Hypertension (4) | Persistent Stage 2 Hypertens. (5) |
| Normal blood pressure | 25.3% | 26.0% | 10.5% | – | – |
| Pre-hypertension | 32.1% | 33.0% | 28.5% | – | – |
| Stage 1 hypertension | 23.5% | 23.2% | – | 59.1% | – |
| State 2 Hypertension | 19.0% | 17.8% | – | 83.1% | 58.4% |
| Obs (N) | 1229 | 1065 | | | |
Hypertension prevalence is similar between the longitudinal sample (Table 1, Column 2), which is used for our cohort analyses of changes in hypertension 2013–17, and the cross-sectional sample at baseline in 2013 (Column 1), except for a lower prevalence of Stage 1 and Stage 2 hypertension that is due to selective survival (hypertension predicts mortality, but not other attrition, see below).
Cohort analyses (Table 1, Columns 3–5) show that $10.5\%$ respondents with normal BP at baseline were hypertensive (Stage 1 or 2) by 2017; this proportion increases to $28.5\%$ among those who were pre-hypertensive at baseline. $59\%$ and $83\%$ of respondents with Stage 1 or Stage 2 hypertension at baseline respectively were still hypertensive (Stage 1 or 2) by 2017. $58\%$ of those with Stage 2 hypertension were still Stage 2 hypertensive by 2017. These patterns of prevalence, onset and persistence of hypertension among adults aged 45+ are very similar for men and women, and with one exception (onset of hypertension among respondents classified as pre-hypertensive at baseline), not statistically different (S4 Table).
Several factors are associated with the onset or persistence of hypertension among adults aged 45+ during 2013–17 (Table 2). There is a weak indication that women are more likely to have an onset of hypertension, but otherwise no clear gender difference in the persistence of hypertension over time. Being overweight or obese increases the odds of hypertension onset and the persistence of hypertension, and poor physical health (low SF12 physical score) and nutritional stress (no consumption of fish or meat in last seven days) predict the onset of hypertension. The same factors also predict changes in systolic BP among respondents not hypertensive at baseline, and in some cases also for the overall study population or other subgroups. Being HIV-positive or Muslim protects against increases in systolic BP. Being married is associated with increases in blood pressure and the persistence of hypertension, contrary to common findings that marriage often found to be protective for cardiovascular health [28]. Notably wealth (various indicators), schooling and grip strength are not predictors of the onset/persistence of hypertension or changes in blood pressure (S5 Table).
**Table 2**
| Unnamed: 0 | Among 2013 respondents who are | Among 2013 respondents who are.1 | Among 2013 respondents who are.2 | Among 2013 respondents who are.3 | Among 2013 respondents who are.4 | Among 2013 respondents who are.5 | All respondents Combined | All respondents Combined.1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | Not hypertensive | Not hypertensive | Hypertensive Stage 1 or 2 | Hypertensive Stage 1 or 2 | Hypertensive Stage 2 | Hypertensive Stage 2 | All respondents Combined | All respondents Combined |
| | Onset of hypertension 2013–17 (Odds Ratio) | Change in systolic BP 2013–17 (OLS Coef) | Persistence of hypertension 2013–17 (Odds Ratio) | Change in systolic BP 2013–17 (OLS Coef) | Persistence of Stage 2 hypertension 2013–17 (Odds Ratio) | Change in systolic BP 2013–17 (OLS Coef) | Change in systolic BP 2013–17 (OLS Coef) | Change in diastolic BP 2013–17 (OLS Coef) |
| Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) | Gender (Ref.: Male) |
| Female | 1.45+ | 0.99 | 1.28 | 1.43 | 1.57 | 5.31+ | 1.23 | 1.24* |
| Female | [0.93,2.27] | [-1.44,3.42] | [0.86,1.91] | [-2.19,5.06] | [0.86,2.85] | [-0.27,10.9] | [-0.83,3.28] | [0.20,2.29] |
| Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) | Body mass index (BMI), Categorical (Ref.: Normal weight or underweight) |
| Overweight/obese | 2.49** | 4.72** | 1.82+ | 6.61** | 2.88** | 10.0* | 4.55** | 0.55 |
| Overweight/obese | [1.55,4.01] | [1.31,8.14] | [0.96,3.45] | [1.99,11.2] | [1.42,5.85] | [2.28,17.8] | [1.40,7.71] | [-1.06,2.16] |
| Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) | Marital status (Ref.: Not currently married) |
| Currently married | 1.19 | 1.49 | 1.88* | 6.38* | 2.86* | 6.55 | 3.74* | 1.45+ |
| Currently married | [0.70,2.02] | [-1.68,4.66] | [1.06,3.33] | [1.16,11.6] | [1.15,7.08] | [-1.64,14.7] | [0.70,6.78] | [-0.031,2.92] |
| HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) | HIV status (Ref.: HIV-negative) |
| HIV-positive | 0.24+ | -5.62** | 0.63 | -4.04 | 0.2 | 4.54 | -4.61* | -3.79** |
| HIV-positive | [0.054,1.03] | [-9.65,-1.58] | [0.26,1.56] | [-12.4,4.36] | [0.019,2.11] | [-6.85,15.9] | [-8.82,-0.39] | [-6.43,-1.16] |
| Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) | Religion (Ref.: Christian) |
| Muslim | 0.70 | -5.39** | 1.02 | -5.59* | 0.67 | -8.34 | -5.57** | -1.25 |
| Muslim | [0.31,1.59] | [-9.09,-1.69] | [0.47,2.23] | [-11.1,-0.065] | [0.29,1.56] | [-18.8,2.14] | [-8.65,-2.50] | [-3.53,1.02] |
| Other/none | 0.73 | 1.30 | 0.58 | -5.50 | 0.62 | -4.50 | -2.07 | -0.41 |
| Other/none | [0.25,2.10] | [-5.13,7.72] | [0.24,1.39] | [-15.3,4.33] | [0.18,2.19] | [-21.5,12.5] | [-6.74,2.59] | [-2.95,2.14] |
| Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) | Low protein consumption (Ref.: Ate fish or meat at least once in last 7 days) |
| Ate no fish or meat in last 7 days | 2.13** | 5.56** | 1.01 | 0.055 | 0.55 | 1.22 | 3.02+ | 2.46** |
| Ate no fish or meat in last 7 days | [1.36,3.34] | [2.42,8.70] | [0.52,1.99] | [-5.51,5.62] | [0.24,1.28] | [-7.05,9.49] | [-0.29,6.33] | [0.72,4.20] |
| SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) | SF12 Physical health score (z-score) |
| SF12 Physical health z-score | 0.77* | -2.34** | 1.31* | 0.34 | 1.15 | -0.47 | -1.48+ | -0.65 |
| SF12 Physical health z-score | [0.62,0.95] | [-4.00,-0.67] | [1.02,1.68] | [-2.45,3.14] | [0.79,1.68] | [-5.55,4.60] | [-3.03,0.074] | [-1.51,0.21] |
| Observations | ≈625 | ≈620 | ≈435 | ≈410 | ≈190 | ≈175 | ≈1030 | ≈1030 |
## Blood pressure and aging
Table 3 shows only modest within-person increases in systolic blood pressure during 2013–17: For women aged 45–54 years there is an annual increase in systolic BP of.94 base points per year of age during 2013–17 (Column 1). This increase is similar to the age gradient in Western, but more than that in some other low-income populations [29]. There is no systematic age-related increase for men, or women aged 55+ (Columns 2–4). Diastolic blood pressure declines on average as respondents age. Identical conclusions are obtained comparing the adjacent box plots in Fig 2 that show 2013 and 2017 BP among longitudinally followed individuals: there is only minimal change in the distribution of systolic BP as cohort members aged during 2013–17, while diastolic BP declined as the cohort got older.
**Fig 2:** *Systolic blood pressure (mmHg) among mature adults in 2013 and 2017, by age in 2013.Notes: Analyses include all MLSFH-MAC respondents with both 2013 and 2017 BP measurement. Age group is assigned based on age in 2013. Systolic BP is top-coded at 200 mmHg. The corresponding graph for diastolic BP is reported in S1 Fig. S2 and S3 Figs provide analyses separate for men and women.* TABLE_PLACEHOLDER:Table 3 In combination, Table 3 and Fig 2 indicate that there is no increase in average hypertension risk in the MLSFH-MAC cohort as cohort members aged four years during 2013–17. While hypertension is highly prevalent and often persistent among MLSFH mature adults (Table 1), overall hypertension was not worsening in the MLSFH-MAC cohort: The distribution of hypertension categories in 2017 is indistinguishable from that in 2013 (p(χ2) =.94) for cohort members who survived and had BP measured at both waves (S6 Table).
## Contrasting cohort and cross-sectional patterns in hypertension
In contrast to the relatively “flat” hypertension profile as the cohort got older during 2013–17, there is a strong cross-sectional increase in systolic BP and hypertension with age (Fig 2). For instance, $54.3\%$ adults aged 45–54 had systolic BP of 120 or higher in 2013, and only $22\%$ had systolic BP of 140 or higher. At ages 75+, the respective factions are $78.1\%$ ($p \leq .01$) and $52.5\%$ ($p \leq .01$). The cross-sectional age gradient for systolic BP is.73 (CI:.95–.87) per year of age in 2013, and.077 (CI:.0062–.13) for diastolic BP (S3 Table). There is also a corresponding cross-sectional increase in hypertension with age: $30\%$ of adults aged 45–54 were hypertensive (Stage 1 or 2) in 2013, increasing to $54\%$ ($p \leq .01$) at ages 75+. A 10 year higher age increases the odds of being hypertensive in the cross-section by about $42\%$ (CI: 32–$54\%$), with similar increases also pertaining to of stage 2 hypertension (S3 Table). This cross-sectional patterns remains stable in 2017 (S3 Table).
The strong cross-sectional age gradient in BP is in contrast to the flat BP profile documented in our earlier longitudinal analyses (Table 3). Specifically, this divergence of the BP age-gradient between cohort and cross-sectional analyses is suggestive of potential cohort influences on BP and hypertension, that is, the possibility that the higher BP and hypertension prevalence among older adults is due to higher levels of life-course adversity experience by older persons. Such cohort effects are plausible because worse early-life conditions among older cohorts has been linked to worse health outcomes at older ages (including worse cardiovascular health) [30, 31]. This aspect is especially important in countries having experienced socioeconomic development or improvements in health (and thus more pronounced cohort differences in early-life environments) during the 20th century [32, 33]. For example, height, a common indicator of early-life nutritional status that was measured in 2017, declines strongly in our study population with age (age-gradient = -.15cm, $p \leq .01$, with controls for region and gender).
## Blood pressure, hypertension prevalence and body mass index
Our earlier analyses have shown that being overweight/obese or having high BMI is a risk factor factor for the onset and persistence of hypertension during 2013–17 (Table 2), as well as for having elevated blood pressure or being hypertensive (S2 Table). Yet, the overall high prevalence of hypertension among MLSFH Mature Adults contrasts sharply with an overall low prevalence of adiposity (Fig 3). Only a small fraction of respondents are overweight ($12\%$ of the total sample in 2017), and an even smaller fraction ($4.7\%$) is obese. Above age 75, where the hypertension prevalence peaks, more than $25\%$ of the respondents are underweight, and only few are obese ($2\%$). There is neither a cross-sectional increase of BMI with age, nor a substantive within-person change in BMI during 2012–17. Age-related increases in BMI are not a factor driving the higher prevalence of hypertension at older ages, further confirming that other factors—such as worse early-life conditions or higher levels of life-course adversity—are contributing to the higher levels of hypertension among older persons.
**Fig 3:** *Body Mass Index (BMI) of MLSFH-MAC respondents aged 2012–13.Notes: Based on measured height and weight in 2012 and 2017. Red line denotes the cut-off for obesity (BMI≥30), and the blue line denotes the cut-off for overweight (25≤BMI≥30). Age group is assigned based on age in 2017. Source: Kohler et al. [16].*
## Consequences of being hypertensive
Hypertension predicts mortality in the MLSFH-MAC cohort, with hypertensive individuals facing about $60\%$ higher odds of dying during 2013–18 (Table 4, Column 1). Importantly for our cohort analyses, hypertension does not predict survey attrition for reasons other than mortality (Column 2). Among survivors, hypertension at baseline is also associated with several adverse outcomes. For example, hypertension predicts declines in subjective health and SF12 physical health score relative to non-hypertensive cohort members during 2013–17, and it predicts (relative) increases in PHQ9 depression score and GAD7 anxiety scores (Columns 3–6). There is also evidence that hypertension reduces effort in common moderately-intensive work activities (e.g., time spent in animal care, firewood preparation, planting, etc.) ( Table 4, Column 7). Baseline hypertension does not predict changes in physical health indicators such as grip strength or BMI during 2013–17, nor does it predict changes in cognitive scores (Columns 8–10).
**Table 4**
| Hypertension in 2013 (baseline) | Mortality 2013–18 (Odds Ratios) (1) | Non-mortality attrition (Odds Ratios) (2) | 2013–17 Change in: | 2013–17 Change in:.1 | 2013–17 Change in:.2 |
| --- | --- | --- | --- | --- | --- |
| Hypertension in 2013 (baseline) | Mortality 2013–18 (Odds Ratios) (1) | Non-mortality attrition (Odds Ratios) (2) | Subjective health (OLS Coef.) (3) | SF12 Physical health score (OLS Coef.) (4) | PHQ9 Depression score (OLS Coef.) (5) |
| Stage 1 Hypertension | 1.24 | 1.28 | 0.013 | -0.56 | 0.18 |
| Stage 1 Hypertension | [0.78,1.97] | [0.73,2.23] | [-0.17,0.20] | [-1.94,0.82] | [-0.52,0.88] |
| Stage 2 Hypertension | 1.89** | 1.03 | -0.22* | -2.51** | 1.04* |
| Stage 2 Hypertension | [1.19,2.98] | [0.60,1.78] | [-0.40,-0.034] | [-4.20,-0.83] | [0.15,1.94] |
| Observations | 1225 | 1227 | 1108 | 1104 | 1108 |
| | 2013–17 Change in: | 2013–17 Change in: | 2013–17 Change in: | 2013–17 Change in: | 2013–17 Change in: |
| | GAD7 Anxiety score (OLS Coef.) (6) | Work efforts (hours/wk) (OLS Coef.) (7) | Grip Strength (OLS Coef.) (8) | Body Mass Index (BMI) (OLS Coef.) (9) | ICA Cognition score (OLS Coef.) (10) |
| Stage 1 Hypertension | 0.32 | -1.57 | -0.18 | -0.025 | 0.13 |
| Stage 1 Hypertension | [-0.27,0.91] | [-4.25,1.11] | [-0.92,0.56] | [-0.40,0.35] | [-0.44,0.71] |
| Stage 2 Hypertension | 0.96** | -6.37** | -0.13 | 0.11 | -0.069 |
| Stage 2 Hypertension | [0.30,1.62] | [-9.19,-3.55] | [-0.94,0.68] | [-0.28,0.50] | [-0.81,0.68] |
| Observations | 1108 | 1080 | 976 | 1046 | 1030 |
## Awareness of hypertension and knowledge of associated health risks
Respondents attribute significant value to knowing their BP, and around $60\%$ report in 2013 a willingness to walk up to 1 hour to obtain a BP measurement, and $\frac{1}{3}$ reports a willingness to walk more than 2 hours (Table 5). These reports are consistent with the almost universal consent to BP measurement as part of our study. This willingness does not increase among respondents who are hypertensive, but it is consistent with respondents associating BP with several dimensions of their overall health. However, not all of these associations are accurate. For instance, about $37\%$ of respondents in 2013 ($33\%$ in 2017) think that knowing one’s BP may indicate being HIV-positive (Table 5), an association that is not generally established [34, 35]. More than $70\%$ of baseline respondent correctly state that BP knowledge can be informative about the with risk of stroke, heart disease, and diabetes. About $80\%$ of respondents think BP is informative about their ability to work and being productive in daily life, in sharp contrast and the mostly asymptomatic nature of high BP (albeit there is some evidence in our study population that stage 2 hypertension leads to reductions in work effort; Table 4). About $46\%$ of baseline respondents associate BP with malaria, an association that is empirically tenuous [36]. Notably, these assessments about the importance of knowing ones BP do not differ by respondent’s own hypertension status: hypertensive individuals do not state more accurate reasons of why it is important to know BP, nor do they give more importance to BP knowledge. At follow-up in 2017, respondents report similar associations of blood pressure with other health outcomes as in 2013, albeit overall there is a declining perception that knowledge about BP can be indicative of other health issues.
**Table 5**
| Unnamed: 0 | Respondent is | Respondent is.1 | Respondent is.2 |
| --- | --- | --- | --- |
| | Not hypertensive (1) | Hypertensive (2) | Combined (3) |
| If a government clinic offering blood pressure measurement were available nearby, would you get your blood pressure measured if you had to walk: | | | |
| 1 hour | 63% | 59% | 62% |
| 2 hours | 45% | 45% | 45% |
| More than 2 hours | 36% | 34% | 35% |
| Why do you think that it might be important for individuals to know their blood pressure? It may tell them… | | | |
| 2013: | | | |
| …if they have HIV | 37% | 35% | 36% |
| …about their risk of stroke | 76% | 70% | 73% |
| …something about how much they can work | 78% | 79% | 79% |
| …if they have heart problems | 87% | 88% | 87% |
| …if they have diabetes | 70% | 74% | 72% |
| …if they have malaria | 45% | 47% | 46% |
| 2017: | | | |
| …if they have HIV | 33% | 29% | 31% |
| …about their risk of stroke | 64% | 65% | 64% |
| …something about how much they can work | 71% | 74% | 72% |
| …if they have heart problems | 73% | 77% | 75% |
| …if they have diabetes | 46% | 44% | 45% |
| …if they have malaria | 34% | 33% | 34% |
## Cascade of care for hypertension
There remain substantial gaps in the diagnosis and treatment of hypertension among adults aged 45+ in rural Malawi (Fig 4A). Among all respondents who were identified as hypertensive in 2013, only $22\%$ have been diagnosed in the last 2 years by a doctor or medical professional as hypertensive (red bar). Among those who were diagnosed, less then $9\%$ reported in 2013 to be taking treatment for hypertension (green bar); that is for about $61\%$ of respondents their medical needs for disease treatment were not met in 2013. Among those who were on treatment, only $11\%$ had BP levels “under control,” i.e., within the normal range. The cascade for 2017 among cohort respondents shows an overall improved cascade (Fig 4B), but there remain substantial gaps in care: only $58\%$ of respondents with measured high levels of BP were not diagnosed in the past as hypertensive, and $54\%$ of the diagnosed do not report taking treatment. However, only $17\%$ of hypertensive respondents have attained a measured BP in the normal range. In a related study we have has documented that prior testing—and specifically the receipt of referral letters to health care providers—helped reduce blood pressure, increase diagnoses and update of medication [12] (similar findings have also been shown in South Africa [13]). The blood pressure screening that was conducted as part of the MLSFH-MAC may therefore have been instrumental in achieving improvements in case between 2013 and 2017 (Fig 4A+4B).
**Fig 4:** *Cascade of care for hypertension in rural Malawi, cohort respondents age 45+ yrs. 2013–17.Notes: The cascade represents the proportion of individuals who reach each separate stage of care, conditional on being included in the previous stage. Analyses are based on respondents who participated in both the 2013 and 2017 MLSFH-MAC to ensure that variations in the study populations do not affect the findings; however, similar patterns are identified based on analyses of all 2013 or 2017 respondents.*
## Summary and discussion
Cardiovascular diseases (CVDs) are rapidly becoming a major source of morbidity and mortality in many Sub-Saharan African (SSA) countries, while disease burdens attributable to communicable diseases are decreasing [8]. Addressing the implications of this shifting disease burden, which occurs together with a rapid aging of SSA populations, is critical for achieving the “grand convergence” in health that has been proposed as an achievable global health goal by 2035 [37]. Importantly, a renewed focus on NCDs—including hypertension as a top priority—will be critical as the Covid-19 pandemic has further revealed the incredible challenges of addressing a dual burden of communicable and non-communicable diseases in LICs [38], as well as the health and social vulnerabilities in LIC population that are caused by an often inadequately treated burden of noncommunicable diseases [39].
The rural subsistence agriculture population that is the focus of this study has low levels of modernization, very limited exposure to “classic” hypertension risk factors such as fast food consumption/unhealthy life styles, and is characterized by high levels of physical activity. Hypertension in such low-income populations has rarely been investigated using longitudinal data, and this is especially the case for older persons who are most at risk of cardiovascular diseases. Drawing on MLSFH-MAC panel data, this study is able to fill an important niche in the literature by highlighting cohort patterns of hypertension and highlighting the dynamics of onset and persistence of hypertension as individuals age during 2013–17.
Key findings of our analyses of blood pressure in this aging poor SSA population pertain to several domains of hypertension research: [1] Prevalence of hypertension and its predictors: *Hypertension is* highly prevalent among individuals aged 45+, with more than $40\%$ being hypertensive and more than $30\%$ being pre-hypertensive. These high rates of hypertension affect both men and women age 45+, in contrast to strong gender differences that have been documented in higher-income populations (Fig 1A). In our study population, the prevalence of hypertension is much higher than that of being overweight or obese, and while being overweight/obese is a predictor of hypertension, the higher prevalence of hypertension at older ages is not associated with higher levels of adiposity at older ages.
[2] Hypertension and aging—longitudinal changes in blood pressure: (a) While there is a strong cross-sectional age gradients in blood pressure and hypertension, with older individuals having worse cardiovascular health, our longitudinal analyses using MLSFH-MAC data 2013–17 that on average a relatively flat profile of blood pressure or hypertension. As cohort members aged during 2012–17, there is no discernible increase in average blood pressure or the prevalence of hypertension: among longitudinally followed MLSFH-MAC respondents, the distribution of hypertension categories in 2017 is indistinguishable from that in 2013. The aging of the cohort during 2013–17 did therefore not increase average blood-pressure-related cardiovascular risk in this poor study population. ( b) This divergence of cross-sectional and longitudinal age-gradient in BP and hypertension is likely due to cohort influences: earlier cohorts had worse early-life environments that, consistent with the Barker hypotheses [40], translated into worse cardiovascular health at older ages. ( c) Despite the relatively stable distribution of blood pressure as the MLSFH-MAC cohort aged during 2021–17, blood-pressure-related cardiovascular risk is dynamic on the individual level. Among respondents with normal BP in 2013, about $10\%$ experienced an onset of hypertension by 2017; among pre-hypertensive respondents in 2013, this rate almost triples and $28.5\%$ experienced an onset of hypertension. ( d) Stage 1 or 2 Hypertension, however, are often persistent: $59\%$ of respondents with Stage 1 hypertension in 2013 were still hypertensive in 2017, and this is the case for $84\%$ of respondents with Stage 2 hypertension at baseline in 2013. ( e) The overall stable hypertension profile in the MLSFH-MAC cohort during 2013–17 arises because the onset of hypertension is compensated by who experience a decline in blood pressure or a transition from being hypertensive to normal blood pressure or pre-hypertension.
[3] Predictors of onset of hypertension over time: (a) Being overweight/obese is a strong predictor of the onset and persistence of hypertension during 2012–17, and poor physical health (low S12 physical score) and facing nutritional stress (not having eaten fish or meat in last 7 days at baseline) predicts the onset of hypertension. Being HIV-positive or Muslim implies lower odds of experiencing increases in systolic BP. Interestingly, marriage may not be protective for cardiovascular health, in contrast to a common finding in other populations [28]. Being currently married in our study population is associated with increasing blood pressure over time and the persistence of hypertension. ( b) Nevertheless, there are few systematic socioeconomic predictors of the onset or persistence of hypertension in this study population aged 45+ during 2013–17; notably neither schooling, wealth, nor grip strength (all measured in 2013) predict the onset/persistence of hypertension during 2013–17 or changes in systolic/diastolic blood pressure (albeit these factors predict cross-sectional level differences in BP or hypertension prevalence in the expected direction).
[4] Life-course consequences of hypertension: Importantly, hypertension is has measurable consequences among adults aged 45+ in Malawi. Hypertension for example predicts mortality in the MLSFH-MAC population during 2013–17, with hypertensive individuals facing about $60\%$ higher odds of dying. Yet, hypertension is not associated with non-mortality attrition in the MLSFH-MAC. Hypertension at baseline also associated with declines in subjective health, SF12 physical health scores and work efforts during 2013–17 and it is associated with increases in depression and anxiety. These findings thus reaffirm that hypertension is an important public health concern among adults aged 45+ in rural Malawi that is associated with more rapid declines in health as individuals get older.
[5] Cascade of hypertension care: While diagnoses and treatment of hypertension are improving, our analyses of the cascade of care document that there remain substantial gaps in the diagnosis and treatment of hypertension among adults aged 45+ in rural Malawi. In 2013, for example, only $22\%$ have been diagnosed in the last two years by a doctor or medical professional as hypertensive, and among those who were diagnosed, less then $9\%$ reported in 2013 to be taking treatment for hypertension. Yet, this cascade of care substantially improved by 2017. $43\%$ of hypertensive individuals were diagnosed, and $46\%$ of the diagnosed were taking treatments. Yet, despite these improvements in the cascade of care, only $17\%$ of hypertensive respondents have attained a measured BP in the normal range. These gaps in the cascade of care are disconcerting as high blood pressure is one of the main modifiable causes of cardiovascular disease risk with a large body of evidence showing that lowering blood pressure through low-cost and widely available medications significantly reduces CVD mortality [41]. The findings of our cascade-of-care analyses are also important as they help inform possible primary health care interventions that can increase management and treatment of hypertension and related chronic non-communicable diseases. Importantly, prior research [42, 43] indicates that the chronicity of these diseases necessitates a restructuring of healthcare to address the multidisciplinary, sustained care including psychosocial support and development of self-management skills, with primary healthcare providers offering promising opportunities for targeted interventions, including for instance by increasing contact of at-risk populations with the health care system (including through screening interventions [7, 12, 13]), improvements of awareness, equipment (e.g., blood pressure monitors) and access to medications among health care providers [44], updates/enhancements of the primary care guidelines to better reflect the needs of older persons [43] or other at-risk individuals (e.g., by integrating hypertension and HIV management [45]), improved linkage to care using cell phone and related technologies [46], and other improvements in the longitudinal control of hypertension [47, 48].
Overall, our findings in this paper indicate that hypertension and related high cardiovascular risks are widespread, persistent and often not diagnosed or treated in this rural sub-Saharan population aged 45+. The prevalence, onset and persistence of hypertension cuts across all subgroups in this population—including, importantly, both women and men. While age is an important predictor of hypertension risk, as has been documented in other LMIC contexts [7], hypertension is already widespread in in middle ages 45–55 years. Hence, hypertension among adults 45+ in Malawi seems to be more similar to a “generalized epidemic” than in high-income countries where cardiovascular risk has strong socioeconomic gradients and untreated hypertension particularly prevalent in vulnerable subsets of older persons [49].
Yet, our findings do not provide a bleak picture about hypertension and cardiovascular disease risk in poor aging populations in SSA or elsewhere. In contrast, a recent body of research has started to provide evidence that relatively inexpensive screenings and referrals for hypertension and are effective approach to reduce the gaps in the cascade of care for hypertension [12, 13]. Our findings highlight the urgency of building on this recent evidence and expand information about cardiovascular risk, screening for hypertension and available treatments for elevated blood pressure to the global poor. There are also primary healthcare intervention that can be scaled up at modest costs [42, 43]. Such screening efforts or primary healthcare enhancements are likely to have large returns in terms of improving population health among older persons [37].
Some limitations of our study are noteworthy. While the MLSFH-MAC is based on a large, population-based sample of adults 45+ years old living in a rural SSA LIC context, it is not a nationally representative sample. This concern is somewhat ameliorated by the fact that the MLSFH-MAC cohort characteristics closely match that of cross-sectional nationally-representative samples [16] and the fact that roughly $85\%$ of the Malawian population resides in rural areas [50]. Hence, our findings can likely be generalized to other rural areas in Malawi and similar low-income populations in southeastern SSA. The age range covered in our study (age 45+ years) is also comparable to other aging studies in SSA LICs [51], and given life expectancy trends in Malawi and SSA more generally, the study represents adequately the experience of the older population in the region. As in any longitudinal study, attrition is a concern for the MLSFH. However, non-mortality-related attrition has been very low in the MLSFH-MAC as a result of migration tracking and related fieldwork procedures; during 2012–18 the MLSFH-MAC retained $97\%$ of surviving respondents [16]; importantly, non-mortality-related attrition 2013–17 is not associated with hypertension. Detailed analyses of attrition as part of the MLSFH cohort profiles [16, 20] have also shown that, even though respondent characteristics often differ significantly between those who were lost to follow-up and those who were re-interviewed, the coefficient estimates for standard family background variables in regressions and probit equations for many health outcomes are not significantly affected by attrition.
Population-based blood pressure measurement in rural older and poor populations in challenging, and measurements of blood pressure on a single session—as is conducted in most population-based health surveys, including the MLSFH—does not provide a clinical diagnoses of hypertension. Yet the MLSFH-MAC adopted the protocol for Health and Retirement Study [25], and through extensive training and carefully-developed fieldwork logistics, the MLSFH-MAC not only achieved very high rates of consent to blood pressure measurements, and data quality of the blood pressure measures is generally high [16]. However, the MLSFH-MAC data lack details about how individuals are affected by hypertension, and among those who are treatment, what specific type of hypertension they receive and/or how frequently they are followed up by the health system.
Given the scarcity of longitudinal data on blood pressure and hypertension in SSA low-income populations, the limitations of this study are relatively minor. The MLSFH-MAC is one of the rare cohort studies in a SSA LICs that provides detailed longitudinal information about blood pressure, hypertension knowledge and treatment, and our analyses are among the first that document the cohort patterns of hypertension in an aging SSA LIC population.
## References
1. **Worldwide trends in blood pressure from 1975 to 2015: A pooled analysis of 1479 population-based measurement studies with 19.1 million participants**. *Lancet* (2017.0) **389** 37-55. DOI: 10.1016/S0140-6736(16)31919-5
2. NCD Risk Factor Collaboration (NCD-RisC). **Worldwide trends in hypertension prevalence and progress in treatment and control from 1990 to 2019: a pooled analysis of 1201 population-representative studies with 104 million participants**. *Lancet* (2021.0) **398** 957-980. DOI: 10.1016/S0140-6736(21)01330-1
3. Geldsetzer P, Manne-Goehler J, Marcus ME, Ebert C, Zhumadilov Z, Wesseh CS. **The state of hypertension care in 44 low-income and middle-income countries: a cross-sectional study of nationally representative individual-level data from 1.1 million adults**. *Lancet* (2019.0) **394** 652-662. DOI: 10.1016/S0140-6736(19)30955-9
4. 4The Economist. Daily chart: The curious case of high blood pressure around the world. Economist. 2017. Available from: https://www.economist.com/graphic-detail/2017/01/13/the-curious-case-of-high-blood-pressure-around-the-world
5. 5United Nations Population Division. World Population Prospects, the 2019 Revision; 2019. Available from: https://population.un.org/wpp/.
6. Sudharsanan N, Geldsetzer P. **Impact of Coming Demographic Changes on the Number of Adults in Need of Care for Hypertension in Brazil, China, India, Indonesia, Mexico, and South Africa**. *Hypertension* (2019.0) **73** 770-776. DOI: 10.1161/HYPERTENSIONAHA.118.12337
7. Kirschbaum TK, Theilmann M, Sudharsanan N, Manne-Goehler J, Lemp JM, De Neve JW. **Targeting Hypertension Screening in Low- and Middle-Income Countries: A Cross-Sectional Analysis of 1.2 Million Adults in 56 Countries**. *Journal of the American Heart Association* (2021.0) **10** e021063. DOI: 10.1161/JAHA.121.021063
8. **Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019**. *Lancet* (2020.0) **396** 1204-1222. DOI: 10.1016/S0140-6736(20)30925-9
9. Kämpfen F, Wijemunige N, Evangelista B. **Aging, non-communicable diseases, and old-age disability in low- and middle-income countries: a challenge for global health**. *International Journal of Public Health* (2018.0) **63** 1011-1012. DOI: 10.1007/s00038-018-1137-z
10. Greenberg H, Raymond SU, Leeder SR. **The Prevention of Global Chronic Disease: Academic Public Health’s New Frontier**. *American Journal of Public Health* (2011.0) **101** 1386-1390. DOI: 10.2105/AJPH.2011.300147
11. Herrick C, Reubi D. **The future of the global noncommunicable disease agenda after Covid-19**. *Health & Place* (2021.0) **71** 102672. DOI: 10.1016/j.healthplace.2021.102672
12. Ciancio A, Kämpfen F, Kohler HP, Kohler IV. **Health Screening for Emerging Disease Burdens Among the Global Poor: Evidence from Sub-Saharan Africa**. *Journal of Health Economics* (2021.0) **75**. DOI: 10.1016/j.jhealeco.2020.102388
13. Sudharsanan N, Chen S, Garber M, Bärnighausen T, Geldsetzer P. **The Effect Of Home-Based Hypertension Screening On Blood Pressure Change Over Time In South Africa**. *Health Affairs* (2020.0) **39** 124-132. DOI: 10.1377/hlthaff.2019.00585
14. 14Malawi National Statistical Office. Malawi Poverty Report 2018. Lilongwe, Malawi: Government of Malawi, National Statistical Office (NSO); 2018. Available from: http://www.nsomalawi.mw/images/stories/data_on_line/economics/poverty/Malawi%20Poverty%20Report%20-%202019%20.pdf.
15. Kohler IV, Kohler HP, Anglewicz P, Behrman JR. **Intergenerational Transfers in the Era of HIV/AIDS: Evidence from Rural Malawi**. *Demographic Research* (2012.0) **27** 775-834. DOI: 10.4054/demres.2012.27.27
16. Kohler IV, Bandawe C, Ciancio A, Kämpfen F, Payne C, Mwera J. **Cohort Profile: The Mature Adults Cohort of the Malawi Longitudinal Study of Families and Health (MLSFH-MAC)**. *BMJ Open* (2020.0) **10** e038232. DOI: 10.1136/bmjopen-2020-038232
17. Koyanagi A, Lara E, Stubbs B, Carvalho AF, Oh H, Stickley A. **Chronic Physical Conditions, Multimorbidity, and Mild Cognitive Impairment in Low- and Middle-Income Countries**. *Journal of the American Geriatrics Society* (2018.0) **66** 721-727. DOI: 10.1111/jgs.15288
18. Bollyky TJ, Templin T, Cohen M, Dieleman JL. **Lower-Income Countries That Face The Most Rapid Shift In Noncommunicable Disease Burden Are Also The Least Prepared**. *Health Affairs* (2017.0) **36** 1866-1875. DOI: 10.1377/hlthaff.2017.0708
19. 19WHO. World Report on Ageing and Health. Geneva, Switzerland: World Health Organization (WHO); 2015. Available from: http://www.who.int/ageing/publications/world-report-2015/en/.
20. Kohler HP, Watkins SC, Behrman JR, Anglewicz P, Kohler IV, Thornton RL. **Cohort Profile: The Malawi Longitudinal Study of Families and Health (MLSFH)**. *International Journal of Epidemiology* (2015.0) **44** 394-404. DOI: 10.1093/ije/dyu049
21. 21United Nations Development Programme. Human Development Report 2020: Briefing Notes for Malawi; 2020. Available from: http://hdr.undp.org/sites/all/themes/hdr_theme/country-notes/MWI.pdf.
22. 22World Bank. World Bank Open Data: GNI per capita, Atlas method (current US$); 2020. Available from: https://data.worldbank.org/indicator/NY.GNP.PCAP.CD?locations=MW-1W&view=chart.
23. 23Malawi National Statistical Office. Preliminary Report: 2018 Malawi Population and Housing Census: Be Counted. Leave No One Behind. Lilongwe, Malawi: Government of Malawi, National Statistical Office (NSO); 2018. Available from: http://www.nsomalawi.mw/images/stories/data_on_line/demography/census_2018/2018%20Population%20and%20Housing%20Census%20Preliminary%20Report.pdf
24. Payne C, Mkandawire J, Kohler HP. **Disability Transitions and Health Expectancies Among Adults 45 Years and Older Mature in Malawi: A Cohort Modeling Approach**. *PLOS Medicine* (2013.0) **10** e1001435. DOI: 10.1371/journal.pmed.1001435
25. 25Health and Retirement Study (HRS). HRS 2010 Study Materials (CD for training interviewers and data collectors); 2010.
26. Stokes A, Berry KM, Mchiza Z, Parker W, Labadarios D, Chola L. **Prevalence and unmet need for diabetes care across the care continuum in a national sample of South African adults: Evidence from the SANHANES-1, 2011-2012**. *PLOS ONE* (2017.0) **12** e0184264. DOI: 10.1371/journal.pone.0184264
27. Ali MK, Bullard KM, Gregg EW, Del Rio C. **A cascade of care for diabetes in the United States: Visualizing the gaps**. *Annals of Internal Medicine* (2014.0) **161** 681-689. DOI: 10.7326/M14-0019
28. Wong CW, Kwok CS, Narain A, Gulati M, Mihalidou AS, Wu P. **Marital status and risk of cardiovascular diseases: a systematic review and meta-analysis**. *Heart* (2018.0) **104** 1937. DOI: 10.1136/heartjnl-2018-313005
29. Gurven M, Blackwell AD, Rodriguez DE, Stieglitz J, Kaplan H. **Does Blood Pressure Inevitably Rise With Age?**. *Hypertension* (2012.0) **60** 25-33. DOI: 10.1161/HYPERTENSIONAHA.111.189100
30. Barker DJP. **Fetal origins of coronary heart disease**. *British Medical Journal* (1995.0) **311** 171. DOI: 10.1136/bmj.311.6998.171
31. Hoffman DJ, Reynolds RM, Hardy DB. **Developmental origins of health and disease: current knowledge and potential mechanisms**. *Nutritrion Reviews* (2017.0) **75** 951-970. DOI: 10.1093/nutrit/nux053
32. Engle PL, Black MM, Behrman JR, de Mello MC, Gertler Kapiriri L. **Strategies to avoid the loss of developmental potential in more than 200 million children in the developing world**. *Lancet* (2007.0) **369** 229-242. DOI: 10.1016/S0140-6736(07)60112-3
33. Grantham-McGregor S, Cheung YB, Cueto S, Glewwe P, Richter L, Strupp B. **Developmental potential in the first 5 years for children in developing countries**. *Lancet* (2007.0) **369** 60-70. DOI: 10.1016/S0140-6736(07)60032-4
34. Jericó C, Knobel H, Montero M, Sorli ML, Guelar A, Gimeno JL. **Hypertension in HIV-Infected Patients: Prevalence and Related Factors**. *Amerian Journal of Hypertension* (2005.0) **18** 1396-1401. DOI: 10.1016/j.amjhyper.2005.05.016
35. Dillon DG, Gurdasani D, Riha J, Ekoru K, Asiki G, Mayanja BN. **Association of HIV and ART with cardiometabolic traits in sub-Saharan Africa: a systematic review and meta-analysis**. *International Journal of Epidemiology* (2014.0) **42** 1754-1771. DOI: 10.1093/ije/dyt198
36. Eze IC, Bassa FK, Essé C, Koné S, Acka F, Laubhouet-Koffi V. **Epidemiological links between malaria parasitaemia and hypertension: findings from a population-based survey in rural Côte d’Ivoire**. *Journal of Hypertension* (2019.0) **37** 1384-1392. DOI: 10.1097/HJH.0000000000002071
37. Jamison DT, Summers LH, Alleyne G, Arrow KJ, Berkley S, Binagwaho A. **Global health 2035: A world converging within a generation**. *Lancet* (2013.0) **382** 1898-1955. DOI: 10.1016/S0140-6736(13)62105-4
38. Collins T, Tello J, Van Hilten M, Mahy L, Banatvala N, Fones G. **Addressing the double burden of the COVID-19 and noncommunicable disease pandemics: a new global governance challenge**. *International Journal of Health Governance* (2021.0) **26** 199-212. DOI: 10.1108/IJHG-09-2020-0100
39. 39WHO. The impact of the COVID-19 pandemic on noncommunicable disease resources and services: results of a rapid assessment. Geneva, Switzerland: World Health Organization WHO; 2020. Available from: https://www.who.int/teams/noncommunicable-diseases/covid-19.
40. Barker D, Eriksson J, Forsen T, Osmond C. **Fetal origins of adult disease: strength of effects and biological basis**. *International Journal of Epidemiology* (2002.0) **31** 1235-1239. DOI: 10.1093/ije/31.6.1235
41. **Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks for 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017**. *Lancet* (2018.0) **392** 1923-1994. PMID: 30496105
42. Macinko J, Leventhal DGP, Lima-Costa MF. **Primary Care and the Hypertension Care Continuum in Brazil**. *Journal of Ambulatory Care Management* (2018.0) **41** 34-46. DOI: 10.1097/JAC.0000000000000222
43. Lall D, Prabhakaran D. **Organization of primary health care for diabetes and hypertension in high, low and middle income countries**. *Expert Review of Cardiovascular Therapy* (2014.0) **12** 987-995. DOI: 10.1586/14779072.2014.928591
44. Wood R, Van Der Merwe L, Viljoen V, Mash R. **Quality of care for patients with non-communicable diseases in the Dedza district, Malawi**. *African Journal of Primary Health Care and Family Medicine* (2015.0) **7** 1-8. DOI: 10.4102/phcfm.v7i1.838
45. Patel P, Speight C, Maida A, Loustalot F, Giles D, Phiri S. **Integrating HIV and hypertension management in low-resource settings: Lessons from Malawi**. *PLOS Medicine* (2018.0) **15** e1002523. DOI: 10.1371/journal.pmed.1002523
46. Piette JD, Datwani H, Gaudioso S, Foster SM, Westphal J, Perry W. **Hypertension Management Using Mobile Technology and Home Blood Pressure Monitoring: Results of a Randomized Trial in Two Low/Middle-Income Countries**. *Telemedicine and e-Health* (2012.0) **18** 613-620. DOI: 10.1089/tmj.2011.0271
47. 47Mauer N, Geldsetzer P, Manne-Goehler J, Davies J, Stokes AC, McConnell M, et al. Treatment discontinuation and loss of hypertension control: Longitudinal evidence from population-based cohorts in four middle-income countries; 2022.
48. Ali MK, Chwastiak L, Poongothai S, Emmert-Fees KMF, Patel SA, Anjana RM. **Effect of a collaborative care Model on Depressive Symptoms and glycated Hemoglobin, Blood Pressure, and Serum Cholesterol Among Patients With Depression and Diabetes in India: The INDEPENDENT Randomized Clinical Trial**. *JAMA* (2020.0) **324** 651-662. DOI: 10.1001/jama.2020.11747
49. Leng B, Jin Y, Li G, Chen L, Jin N. **Socioeconomic status and hypertension: a meta-analysis**. *Journal of Hypertension* (2015.0) **33** 221-229. DOI: 10.1097/HJH.0000000000000428
50. 50Malawi National Statistical Office. 2018 Malawi Population and Housing Census: Main Report. Zomba, Malawi: Malawi National Statistical Office (Malawi NSO); 2019. Available from: http://www.nsomalawi.mw/images/stories/data_on_line/demography/census_2018/2018%20Malawi%20Population%20and%20Housing%20Census%20Main%20Report.pdf.
51. Brinkmann B, Payne CF, Kohler I, Harling G, Davies J, Witham M. **Depressive symptoms and cardiovascular disease: a population-based study of older adults in rural Burkina Faso**. *BMJ Open* (2020.0) **10** e038199. DOI: 10.1136/bmjopen-2020-038199
|
---
title: Post-COVID-19 syndrome risk factors and further use of health services in East
England
authors:
- Maciej Debski
- Vasiliki Tsampasian
- Shawn Haney
- Katy Blakely
- Samantha Weston
- Eleana Ntatsaki
- Mark Lim
- Susan Madden
- Aris Perperoglou
- Vassilios S. Vassiliou
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022108
doi: 10.1371/journal.pgph.0001188
license: CC BY 4.0
---
# Post-COVID-19 syndrome risk factors and further use of health services in East England
## Abstract
Post-COVID syndrome, defined as symptoms persisting for more than twelve weeks after the diagnosis of COVID-19, has been recognised as a new clinical entity in the context of SARS-CoV-2 infection. This study was conducted to characterise the burden and predictors for post-COVID-19 syndrome in the local population. It was a community-based web-survey study conducted in Norfolk, East England, UK. We sent the survey to patients with confirmed COVID-19 infection by real-time polymerase chain reaction by December 6th, 2020. Questions related to the pre-COVID and post-COVID level of symptoms and further healthcare use. Baseline characteristics were collected from the primary care records. Logistic regression analysis was conducted to establish predictors for post-COVID-19 syndrome and further healthcare utilisation. Of 6,318 patients, survey responses were obtained from 1,487 participants ($23.5\%$). Post-COVID-19 syndrome symptoms were experienced by 774 ($52.1\%$) respondents. Male sex compared to female sex was a factor protective of post-COVID symptoms; relative risk (RR) 0.748, $95\%$ confidence interval (CI), 0.605–0.924. Body mass index was associated with a greater risk of developing post-COVID-19 symptoms (RR 1.031, $95\%$ CI, 1.016–1.047, for 1 kg/m2). A total of 378 ($25.4\%$) people used further health services after their index COVID-19 infection, of whom 277 ($73.2\%$) had post-COVID symptoms. Male sex was negatively associated with the use of further health services (RR 0.618, $95\%$ CI, 0.464–0.818) whereas BMI was positively associated (RR 1.027, $95\%$ CI, 1.009–1.046). Overall, post-COVID-19 symptoms increased the probability of using health services with RR 3.280, $95\%$ CI, 2.540–4.262. This survey of a large number of people previously diagnosed with COVID-19 across East England shows a high prevalence of self-reported post-COVID-19 syndrome. Female sex and BMI were associated with an increased risk of post-COVID-19 syndrome and further utilisation of healthcare.
## Introduction
Post-COVID-19 syndrome, also referred to as Long COVID syndrome, is a term used to describe a condition faced by patients presenting with signs and symptoms that develop during or after infection with COVID-19, continue for more than 12 weeks and are not explained by an alternative diagnosis. When signs and symptoms of COVID-19 only persist from 4 to 12 weeks, the National Institute for Health and Care Excellence (NICE) defined it as ongoing symptomatic COVID-19, whilst when exceeding 12 weeks this was defined as Post-COVID-19 syndrome [1]. The complex condition affects people in different ways, with breathlessness, cough, heart palpitations, headaches, and severe fatigue being among the most prevalent symptoms. However, the symptoms commonly include also chest pain or tightness, problems with memory and concentration ("brain fog"), difficulty sleeping (insomnia), dizziness, pins and needles, joint pain, depression and anxiety, tinnitus, earaches, nausea, diarrhoea, stomach aches, loss of appetite, a high temperature, headaches, sore throat, changes to sense of smell or taste and rashes [1]. This is a new condition and due to the lack of validated tools to effectively assess and manage, primary care and hospital outpatient clinics are being overwhelmed due to the sheer numbers of patients seeking support [1,2].
Given the global scale of this pandemic, there has been a rapid effort to characterise the burden and find predictors for post-COVID-19 syndrome. As of March 24th 2022, the World Health Organisation estimates there have been 472,816,657 confirmed cases of COVID-19, including 6,099,380 deaths worldwide, whereas in the UK, there have been 20.75 million people testing positive and 164.734 deaths [3,4]. The Office for National Statistics, UK (ONS) estimates that as of 2 January 2022, 1.3 million people living in private households in the UK ($2.1\%$ of the population) were experiencing self-reported symptoms persisting for more than four weeks after the first suspected coronavirus (COVID-19) infection that were not explained by something else [5]. Ongoing symptomatic COVID-19 or post-COVID-19 syndrome is estimated to be adversely affecting the day-to-day activities of 836,000 people in the United Kingdom according to the ONS report with 244,000 saying their ability to undertake day-to-day activities had been "limited a lot".
There is growing evidence that having the vaccine reduces the risk of developing post-COVID-19 syndrome [6]. However, the disease mechanisms causing post-COVID-19 syndrome are unknown, and there are no evidence-based treatment options to alleviate the symptoms or reduce the duration [7].
This study was conducted to characterise the burden and predictors for post-COVID-19 syndrome in Norfolk, East England, United Kingdom and to build a statistical model that can inform regional post-COVID-19 syndrome healthcare service planning.
## Study design
This study was a community-based cross-sectional study conducted in Norfolk, East England, UK as a part of Protect Norfolk and Waveney post-COVID-19 syndrome Project (NoW). The project was established to understand the prevalence and ongoing needs of the local population, who are experiencing the longer-term effects of post-COVID-19 syndrome. A web-based survey was developed based on: COVID-19 Yorkshire Rehab Screening Tool (C19-YRS) [8], COVID-19 rapid guideline: managing the long-term effects of COVID-19 (NG188) [1], and Medical Research Council (MRC) Dyspnoea Scale [9]. C19-YRS questionnaire is a clinically validated outcome measure tool recommended by the National Health Service (NHS) England and the National Institute for Health and Care Excellence (NICE) to routinely capture the severity of symptoms that persist longer than four or more weeks after contracting COVID-19 [1,10].
Patients were sent an invitation in February 2021 via a letter to complete the online survey consisting of a mixture of 39 single-choice yes/no or Likert scale questions (Supplementary Material). Questions related to pre-COVID and post-COVID breathlessness, use of any health services in relation to post-COVID-19 syndrome, chest pain, loss of sense of taste or smell, difficulty eating or drinking, weight loss or nutritional concerns, problems with walking, fatigue, short-term memory problems, new bowel concerns, any pain, severity of current anxiety and depression compared to pre-COVID levels, presence of dreams related to COVID illness or hospital admission, problems with communication. Patients were deemed to suffer from post-COVID-19 syndrome when they selected one or more new self-reported symptoms from the questionnaire. Multiple participation was not possible. The survey design did not require the individuals to answer all questions. Approval for this work was granted by the Control Of Patient Information (COPI) notice, approved by the Secretary of State for Health and Social Care for COVID19 research, and Institutional Approval received by the NHS Norfolk and Waveney Integrated Care Board. Participants received an invite by post which included consent information explaining the purpose of the survey. The consent was implied when participants logged in and provided their responses. The anonymity of participants was ensured by full anonymisation before data extraction.
The patient-identifiable data was processed under the control of patient information (COPI) notice. COPI was changed in March 2020 under the COVID-19 Public Health Directions to support various Covid-19 purposes. Under those regulations patient information was to be shared and used for the purposes of informing health services planning and supporting vital research on the cause, effects, treatment and outcomes for patients with the virus. Protect NoW Project was approved and signed off with our Population Health Clinical Team. We reported the results in accordance with the Checklist for Reporting Of Survey Studies (CROSS) [11].
## Patient and public involvement
No patient and public involvement group input was obtained for this study. However, follow-up letters have been issued to signpost participants to further support.
## Participants
Study inclusion criteria were: confirmed COVID-19 infection by real-time polymerase chain reaction (PCR) between the start of COVID-19 pandemic and December 6th 2020. Exclusion criteria included residents of care or nursing homes and South Norfolk residence.
## Variables
Demographics, comorbidities, medications, smoking status and deprivation index were collected from clinical National Health Service Digital databases. Deprivation was expressed as a deprivation decile and was measured using the Index of Multiple Deprivation (IMD), which provides an overall relative measure of deprivation for each lower layer super output area (LSOA) [12]. An LSOA is a small area with an average population of 1,500 people. The overall IMD scores are ranked for all LSOAs within a country and can be divided into 10 groups (deciles) where decile 1 represents the most deprived LSOAs and decile 10 represents the least deprived LSOAs. The IMD is a score based on the area as a whole and not everyone within an LSOA necessarily experiences the same level or type of deprivation. Body mass index (BMI) was calculated based on the most current General Practice records. Medications were categorised into the following categories: beta-blockers, calcium channel blockers, diuretics, oral steroids, angiotensin-converting enzyme inhibitors (ACEI) or angiotensin receptor blocker (ARB), antiplatelets, statins, antidepressants, insulin, oral diabetic medications, anticoagulants, proton pump inhibitors. Previous administration of COVID-19 vaccine was collected from NHS records.
## Statistical analysis
The data were analysed by an independent professional statistician who had access to all data underpinning this study. For all analyses, a 2-sided $p \leq 0.05$ was considered statistically significant. All data were processed using R statistical package. Standard descriptive statistics methods were used to describe data, means with standard deviations (SD) for continuous variables and frequency tables for categorical. Missing values were reported for BMI (608 out of 1487) and deprivation index (114 out of 1487). For those we used multiple imputation with 30 iterations in MICE (R). We used a multivariable logistic regression model to predict the probability of post-COVID-19 syndrome. A full model with all covariates was assessed. We also used backward variable selection to identify the most important variables in the model. Risk profiles were based on the model with variable selection.
## Results
Of 6,318 patients, survey responses were obtained from 1,487 participants ($23.5\%$). Females accounted for $61\%$ of the cohort ($$n = 907$$). Mean age of participants was 50, SD 18.1. BMI was available for 879 participants, mean 28.4, SD 6.9 kg/m2. Current smoking declared 81 participants, 395 were ex-smokers, 704 never smoked and information was not available for 307 participants. Mean deprivation decile was 5.3, SD 2.4. Comorbidities and baseline medications are presented in Tables 1 and 2. Eleven people had received their first dose of COVID mRNA Vaccine BNT162b2.
There were 774 ($52.1\%$) respondents who had experienced post-COVID-19 syndrome symptoms. Incidence of post-COVID-19 syndrome symptoms in women was $55.9\%$ and in men $46.0\%$ ($p \leq 0.001$).
In regression analyses (Figs 1 and 2), male sex compared to female sex was a factor protective of post-COVID symptoms; relative risk (RR) 0.748, $95\%$ confidence interval (CI), 0.605–0.924. Body mass index was associated with greater risk of developing post-COVID19 symptoms, with patients having 1.031 times higher chance of suffering with these symptoms for each 1 kg/m2 increase their BMI (RR 1.031, $95\%$ CI, 1.016–1.047). Age per 1-year increase was not found to be an independent risk factor for higher incidence of post-COVID19 symptoms (RR 1.003, $95\%$ CI, 0.998–1.009). Based on the above, it can be estimated that for a man to have the same probability as female of same age with normal BMI they have to have a BMI over 35 kg/m2.
**Fig 1:** *Multivariable regression model analyzing potential risk factors for post-COVID-19 syndrome.* **Fig 2:** *Regression model for covariates with p<0.1.Sex and BMI were statistically significant whilst age showed a trend towards significance.*
A total of 378 people used further health services after their index COVID-19 infection, of whom 277 ($73.2\%$) had post-COVID symptoms. Patients suffering with post-COVID-19 syndrome were significantly more likely to seek further help and advice from health services compared to those who did not have post-COVID-19 syndrome (RR 3.28, $95\%$ CI, 2.54–4.26). Additionally, it was noted that men were much less likely than women to use further health services (RR 0.750, $95\%$ CI, 0.580–0.966) (Fig 3). In keeping with this, in a further detailed analysis of 277 patients with post-COVID19 syndrome, it was demonstrated that only $14.5\%$ ($\frac{84}{580}$) of men used further health services compared to $21.3\%$ ($\frac{193}{907}$) of women. Regression analysis showed that male sex was negatively associated with the use of further health services, (RR 0.618, $95\%$ CI, 0.464–0.818) while BMI was a positive predictive indicator of this (RR 1.027 for each 1 kg/m2 increase, $95\%$ CI 1.009–1.046) (Fig 4). Overall, in the whole population, post-COVID-19 symptoms increased the probability of using health services with RR 3.280, $95\%$ CI 2.540–4.262.
**Fig 3:** *Analysis of 378 patients that used further health services.Suffering from Post-COVID-19 syndrome was the leading factor requiring any further health services input.* **Fig 4:** *Predictors of using health services when having post-COVID-19 syndrome symptoms (analysis for 277 patients with post-COVID-19 syndrome and further use of health services).Higher BMI associated with increased use, whilst being a man associated with less use of services.*
## Discussion
The results of this study provide a snapshot of self-reported symptoms after recent confirmed COVID-19 infection in the East of England. It was found that female sex and high BMI are associated with higher likelihood of developing post-COVID19 syndrome. Those two factors have a significant predictive value in the use of further health services among those diagnosed with post-COVID19 syndrome.
The reported incidence of post-COVID-19 syndrome in the present study amounted to $52\%$ and was concordant with a meta-analysis based on studies published before January 2021 showing that approximately half of the individuals ($53\%$, $95\%$ CI: 41–$65\%$) reported persistence or presence of one or more symptoms over 12 weeks after COVID-19 diagnosis [13]. The study notes that the most prevalent symptoms were: anxiety ($32\%$), general pain or discomfort ($28\%$), fatigue ($25\%$), insomnia ($22\%$) and cognitive impairment ($20\%$). Another meta-analysis of studies published before 1st January 2021 showed that $80\%$ ($95\%$ CI 65–92) of the patients with COVID-19 have long-term symptoms [14]. In that study, the 5 most common manifestations were fatigue ($58\%$), headache ($44\%$), attention disorder ($27\%$), hair loss ($25\%$), dyspnoea ($24\%$).
A study conducted on data of all people registered with a primary care general practice (GP) in November 2020 encompassing $96\%$ of the English population between 1 February 2020 and 25 April 2021 assessed the use of post-COVID-19 syndrome codes in GP Practices [15]. Up to 25 April 2021, there were 23,273 ($0.04\%$) patients with a recorded code indicative of a post-COVID-19 diagnosis. Interestingly, the rate of documented post-COVID19 diagnosis varied substantially between regions, from a minimum of 20.3 per 100 000 people in the East of England ($95\%$ CI,19.3–21.4) to a maximum of 55.6 in London ($95\%$ CI, 54.1–57.1). To date, this study provides the most comprehensive insights into the post-COVID-19 syndrome prevalence and risk factors in a demographically distinct population of East of England–one of the eldest populations in the UK [15,16].
Our study showed that female sex and high BMI are important predictive factors for the development of post-COVID19 syndrome. This is in keeping with a previous study conducted in England, which showed that females had a higher rate of recorded post-COVID-19 syndrome codes than males (52.1 versus 28.1 per 100,000 people) [15]. According to ONS, the prevalence of any post-COVID-19 syndrome symptoms is higher in women compared with men ($23.6\%$ versus $20.7\%$), while the age group estimated to be most significantly affected by post-COVID-19 syndrome symptoms is 35–49 years ($26.8\%$), followed by 50–69 years ($26.1\%$), and the ≥70 years group ($18\%$) [5]. Obesity has been associated with an increased need for diagnostic tests starting from 30 days after a positive SARS-CoV-2 test with a hazard ratio of 1.39 ($95\%$ CI 1.13 to 1.71) for severe obesity (BMI≥40) versus normal BMI (18–24 kg/m2), and an HR of 1.25 ($95\%$ CI 1.02 to 1.53) for moderate obesity (BMI 35–39) versus normal BMI [17]. The authors noted that there was a higher rate of cardiac, vascular, pulmonary, gastrointestinal, and mental health related testing among obese patients than those with normal BMI.
A Coronavirus (COVID-19) Infection Survey (CIS) performed by ONS in the UK, showed that among participants with COVID-19 who were exactly one-to-one matched to control participants on a number of factors, $5.0\%$ out of 12,611 participants reported symptoms at 12 to 16 weeks which was statistically significantly higher than in the control group ($3.4\%$) [18]. The difference in prevalence remained statistically significant at 20 to 24 weeks. The results of the CIS study suggest that the prevalence of symptoms comprising fever, headache, muscle ache, weakness/tiredness, nausea/vomiting, abdominal pain, diarrhoea, sore throat, cough, shortness of breath, loss of taste, and loss of smell following COVID-19 infection is greater than the background prevalence of these symptoms in the population.
As a response to challenges posed by the post-COVID-19 syndrome on healthcare needs and patients’ ability to work along with lack of established treatments for people living with post-COVID-19 syndrome several trials has been launched in the UK looking at improving home monitoring and self-management of symptoms and identifying effective treatments [19]. ReDIRECT trial is testing the hypothesis of whether a well-established weight management programme, delivered and supported remotely, can improve symptoms for people with post-COVID-19 syndrome and overweight/obesity [20]. STIMULATE-ICP, the largest long COVID trial to date, recruiting more than 4,500 people with the condition, will test the effectiveness of existing drugs to treat post-COVID-19 syndrome and the impact on patients’ symptoms, mental health and outcomes such as returning to work [21].
## Limitations
The survey responses were received around three months after the first Covid vaccination was administered in the UK on the 8th of December 2020. By the time of the first response, many people received their first dose COVID-19 vaccine in the UK. Therefore, this survey represents the post-COVID-19 syndrome picture in the largely pre-vaccination era and the findings cannot be generalised to populations with a high percentage of vaccinated people. Also, as the new SARS-CoV-2 have emerged, the results of this study relate to pre-Omicron variants. Having said this, whilst the vaccines have reduced the number of individuals suffering from post-COVID-19 syndrome, it is unlikely that they would have altered the risk factors for this. Likewise, the new variants are unlikely to have changed the risk factors either.
The response rate was $23.5\%$ which although appropriate for a survey, could raise the possibility of non-response bias if the responders differed from the non-responder in terms of baseline characteristics and prevalence of post-COVID-19 syndrome symptoms which could not be evaluated in this study. The selection bias was minimised as the survey was sent to all patients with confirmed COVID-19 diagnosis in the regions of interest. As our study involved self-reporting symptoms and use of further health services by participants there is a considerable risk of response bias. We did not have a SARS-CoV-2 negative control group and did not explore if other conditions could explain the reported symptoms. In addition, we did not report the meantime from COVID-19 diagnosis to survey completion which may have contributed to the underestimation of the reported prevalence of post-COVID-19 syndrome.
## Conclusion
The present survey of a large number of people previously diagnosed with COVID-19 across East of England shows a high prevalence of post-COVID-19 syndrome. The results suggest females and increased BMI were associated with an increased risk of post-COVID-19 syndrome and further utilisation of health care, with increasing age showing a trend towards significance. The survey results provide valuable insights to help plan the local, national and international integrated referral pathway and assessment centres.
## References
1. **National Institute for Health and Care Excellence: Clinical Guidelines**. *London* (2020)
2. Goyal DK, Mansab F, Naasan AP, Iqbal A, Millar C, Franklin G. **Restricted access to the NHS during the COVID-19 pandemic: Is it time to move away from the rationed clinical response?**. *Lancet Reg Health Eur* (2021) **8** 100201. DOI: 10.1016/j.lanepe.2021.100201
3. 3World Health Organisation. WHO Coronavirus (COVID-19) Dashboard. 2022 [cited 2022 March 30]. Available from: https://covid19.who.int/.. *WHO Coronavirus (COVID-19) Dashboard* (2022)
4. 4Coronavirus (COVID-19) in the UK. 2022 [cited 2022 March 30]. Available from: https://coronavirus.data.gov.uk/.. *Coronavirus (COVID-19) in the UK* (2022)
5. 5Office for National Statistics. Prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 6
January
2022 [cited 2022 March 30]. Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/prevalenceofongoingsymptomsfollowingcoronaviruscovid19infectionintheuk/6january2022.. *Prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK* (2022)
6. 6The effectiveness of vaccination against long COVID. A rapid evidence briefing. 2022 [cited 2022 March 30]. Available from: https://ukhsa.koha-ptfs.co.uk/cgi-bin/koha/opac-retrieve-file.pl?id=fe4f10cd3cd509fe045ad4f72ae0dfff.. *A rapid evidence briefing* (2022)
7. Brown K, Yahyouche A, Haroon S, Camaradou J, Turner G. **Long COVID and self-management**. *Lancet* (2022) **399** 355. DOI: 10.1016/S0140-6736(21)02798-7
8. 8Covid 19 Yorkshire Rehab Screen (C19-YRS). 2022 [cited 2022 March 30]. Available from: https://c19-yrs.com/wp-content/uploads/2022/06/Self-report-C19-YRS-latest-version-April-2021.pdf.. *Covid 19 Yorkshire Rehab Screen (C19-YRS)* (2022)
9. 9Medical Research Council. The Questionnaire on Respiratory Symptoms. 1986 [cited 2022 Jan 14]. Available from: https://mrc.ukri.org/research/facilities-and-resources-for-researchers/mrc-scales/mrc-dyspnoea-scale-mrc-breathlessness-scale/.. *The Questionnaire on Respiratory Symptoms* (1986)
10. 10COVID-19 Yorkshire Rehabilitation Scale. 2020 [cited 2022 March 30]. Available from: https://c19-yrs.com/about/the-system/.. *COVID-19 Yorkshire Rehabilitation Scale* (2020)
11. Sharma A, Minh Duc NT, Luu Lam Thang T, Nam NH, Ng SJ, Abbas KS. **A Consensus-Based Checklist for Reporting of Survey Studies (CROSS)**. *J Gen Intern Med* (2021) **36** 3179-87. DOI: 10.1007/s11606-021-06737-1
12. 12Office for National Statistics. Socioeconomic inequalities in avoidable mortality, England and Wales: 2001 to 2016. 2018 [cited 2022 March 30]. Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/causesofdeath/articles/measuringsocioeconomicinequalitiesinavoidablemortalityinenglandandwales/2001to2016.. *Socioeconomic inequalities in avoidable mortality, England and Wales: 2001 to 2016* (2018)
13. Domingo FR, Waddell LA, Cheung AM, Cooper CL, Belcourt VJ, Zuckermann AME. **Prevalence of long-term effects in individuals diagnosed with COVID-19: a living systematic review**. *medRxiv* (2021). DOI: 10.1101/2021.06.03.21258317
14. Lopez-Leon S, Wegman-Ostrosky T, Perelman C, Sepulveda R, Rebolledo PA, Cuapio A. **More than 50 long-term effects of COVID-19: a systematic review and meta-analysis**. *Sci Rep* (2021) **11** 16144. DOI: 10.1038/s41598-021-95565-8
15. Walker AJ, MacKenna B, Inglesby P, Tomlinson L, Rentsch CT, Curtis HJ. **Clinical coding of long COVID in English primary care: a federated analysis of 58 million patient records in situ using OpenSAFELY**. *Br J Gen Pract* (2021) **71** e806-e14. DOI: 10.3399/BJGP.2021.0301
16. 16Office for National Statistics. 2022 [cited 2022 March 30]. Available from: https://www.ons.gov.uk/.. *Office for National Statistics* (2022)
17. Aminian A, Bena J, Pantalone KM, Burguera B. **Association of obesity with postacute sequelae of COVID-19**. *Diabetes Obes Metab* (2021) **23** 2183-8. DOI: 10.1111/dom.14454
18. 18Office for National Statistics. Technical article: Updated estimates of the prevalence of post-acute symptoms among people with coronavirus (COVID-19) in the UK: 26 April 2020 to
1
August
2021. 2021 [cited 2022 March 30]. Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/articles/technicalarticleupdatedestimatesoftheprevalenceofpostacutesymptomsamongpeoplewithcoronaviruscovid19intheuk/26april2020to1august2021.. *Technical article: Updated estimates of the prevalence of post-acute symptoms among people with coronavirus (COVID-19) in the UK: 26 April 2020 to* (2021) **2021**
19. 19Department of Health and Social Care. New research into treatment and diagnosis of long COVID. 2021 [cited 2022 March 30]. Available from: https://www.gov.uk/government/news/new-research-into-treatment-and-diagnosis-of-long-covid.. *New research into treatment and diagnosis of long COVID* (2021)
20. 20The Remote Diet Intervention to REduce long Covid symptoms Trial (ReDIRECT). 2021 [cited 2022 March 30]. Available from: https://redirectstudy.co.uk/.. *The Remote Diet Intervention to REduce long Covid symptoms Trial (ReDIRECT)* (2021)
21. Symptoms Trajectory. *Inequalities and Management: Understanding Long-COVID to Address and Transform Existing Integrated Care Pathways (STIMULATE-ICP)* (2022)
|
---
title: Prevalence and factors associated with stroke risk factors in an urban community
of Parakou, Northern Benin, 2016
authors:
- Yessito Corine Nadège Houehanou
- Mendinatou Agbetou
- Oyéné Kossi
- Maurice Agonnoudé
- Hospice Hountada
- Thierry Adoukonou
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022112
doi: 10.1371/journal.pgph.0000667
license: CC BY 4.0
---
# Prevalence and factors associated with stroke risk factors in an urban community of Parakou, Northern Benin, 2016
## Abstract
Sub-Saharan Africa faces a heavy burden of stroke due to the growth of its risk factors. We aimed to estimate the prevalence of stroke risk factors and identify the factors associated with metabolic risk factors in the district of Titirou, in Parakou (northern Benin) in 2016. A cross-sectional study was conducted. It included people aged at least 15 years, living in Titirou for at least 6 months, and who had given their written consent to participate in the study. A door-to-door survey was performed from 15 march to 15 July 2016 in each neighborhood until the pre-determined number was reached. Sociodemographic data, medical histories, anthropometric and blood pressure measures were recorded using the WHO STEPS approach. The prevalence of stroke risk factors was calculated, and a multivariable logistic regression was done to identify the factors associated with metabolic risk factors for stroke. A total of 4671 participants were included with a mean age of 27.7±12.9 years and a sex ratio of 0.98. Concerning the behavioral risk factors for stroke, $17.2\%$ were alcohol consumers, $3.5\%$ were smokers, $21.5\%$ had low fruit and vegetable intake, and $51.1\%$ had low physical activity practice. The prevalence of metabolic risk factors for stroke was respectively of $8.7\%$ for obesity, $7.1\%$ for high blood pressure, $1.7\%$ for self-reported diabetes, and $2.2\%$ for dyslipidemia. Age ($p \leq 0.001$), sex ($p \leq 0.001$), marital status ($p \leq 0.001$) and professional occupation ($$p \leq 0.010$$) were associated with obesity. Age was also associated with high blood pressure ($p \leq 0.001$) and diabetes ($p \leq 0.001$). Dyslipidemia varied according to smoking ($$p \leq 0.033$$) and low physical activity practice ($$p \leq 0.003$$). The study revealed a significant prevalence of some stroke risk factors. Targeted local interventions for primary prevention of stroke should be promoted in this community.
## Background
Cardiovascular diseases are a public health problem in the world. In 2016, the World Health Organization (WHO) estimated that around 17.9 million people died of cardiovascular disease, meaning one-third of global mortality [1]. Most of these deaths were due to vascular diseases such as heart attack and stroke and occurred in low-and middle-income countries. A study published in 2019 showed that while the incidence and mortality of stroke are declining in high-income countries, they are increasing in low-income countries [2].
Progresses made in the control of infectious diseases in Sub-Saharan Africa (SSA) are threatened by the burden of non-communicable diseases. In fact, many cross-sectional studies realized in Benin between 2012 and 2020 showed a prevalence of stroke that varies from $0.2\%$ to $1.5\%$ [3–5]. Moreover, a study conducted in Parakou from 2012 to 2018 found a 5-year stroke mortality among stroke survivors patients of $23.5\%$ [6]. The cost of stroke treatment is quite expensive. In 2011, the direct hospital cost of stroke treatment in Parakou University hospital was estimated at around 620 USD [7].
Despite the heavy burden of stroke, it can be reduced by carrying out interventions on risk factors. In fact, the promotion of a healthy diet and regular physical activity, smoking cessation, and reducing alcohol consumption are the keys to the primary prevention by controlling metabolic risk factors such as: obesity, high blood pressure, raised blood sugar, and raised blood cholesterol [1].
The changes in food habits and growing sedentary lifestyles are leading to rising frequencies of stroke risk factors in SSA proved by results of STEPS surveys among adults [8]. In Benin, the STEPS survey conducted in 2015 among adults aged of 18 to 69 years old showed that $5\%$ of adults used tobacco and $26.5\%$ of them consumed alcohol. Moreover, about $93.1\%$ of the participants had insufficient fruit and vegetable intake and $15.9\%$ had low physical activity practice with almost a tenth ($7.4\%$) that had obesity, $25.9\%$ high blood pressure, $12.4\%$ raised blood sugar and $4.4\%$ raised blood cholesterol [9]. The survey also showed disparities between regions, areas of residence (rural or urban), and social categories, highlighting the value of comprehensive studies on targeted subpopulations.
For an effective fight against stroke in SSA countries and particularly in Benin, it is necessary to strengthen the primary prevention measures accessible to the local health system and communities. Considering epidemiological data and periodically update them are important for the implementation of targeted interventions.
## Aim, design, and frame
We aimed to estimate the prevalence of stroke risk factors and identify factors associated with metabolic risk factors in the district of Titirou, in Parakou (Benin) in 2016. A cross-sectional study was conducted in Titirou, the second most populated district of Parakou. According to the 2013 census, its population was estimated to 25,530 with 12,816 people aged at least 15 years [10]. It is subdivided into seven neighborhoods and is located at 1 km far from the University Hospital.
## Participants and sampling
The study included people aged at least 15 years, resident in Titirou and present at the date of the survey, and who had given their written consent to participate. Those unable to answer questions or those who were missed over three visits were not included.
A door-to-door survey was conducted. In each neighborhood, the researchers randomly determined a direction from the center and randomly chose one side in the selected direction. The houses were visited based on a close-to-close approach. All persons living in households and who met the inclusion criteria were interviewed until the expected number was reached. If in a neighborhood this number was not reached, the investigators returned to the center and the operation was repeated in another direction until the number was finally got.
The sample size was initially calculated for a survey focused on the stroke prevalence with an expected value of 4.6 per 1000, accuracy of 0.002 and an alpha risk of 0.05, giving a minimal number of subjects of 4600. The size by neighborhood was proportional to the number of residents aged at least 15 years.
For our study, this sample size corresponded to an accuracy of $1.5\%$ for the estimation of the prevalence of stroke risk factors, considering a theoretical value of $50\%$, and an alpha risk of 0.05.
## Variables
Metabolic risk factors for stroke (“yes” or “no”) were the dependent variables. They were defined according to WHO STEPS surveys manual [11]. [ 1] Obesity was defined by a body mass index (BMI) ≥30 kg/m2 that was calculated by dividing the weight (kg) by the square of the size (m); [2] high blood pressure by a systolic blood pressure ≥140 mm Hg and/or a diastolic blood pressure ≥ 90 mm Hg during the survey; [3] diabetes by self-report history of diabetes (at least two results of fast venous blood glucose tests ≥ 1,26 g/l or using diabetes drugs during the survey); [4] dyslipidemia by self-report of history raised total blood cholesterol (at least one result ≥ 2,40 g/l or using statins during the survey).
Sociodemographic and behavioral risk factors for stroke were the independent variables. Behavioral risk factors for stroke (“yes” or “no”) were defined based on the self-report information [11]. [ 1] “Low fruit and vegetable intake” was defined by the consumption of less than 5 servings of fruit and vegetable per day during the last 12 months; [2] “current smoking” by the consumption of tobacco during the last 12 months; [3] “alcohol consumption” by the consumption of alcoholic beverages during the 30 last days; [4] “low physical activity practice” by the practice of less than 150 minutes per / week of moderate physical activity or the equivalent of vigorous physical activity or a combination of moderate and vigorous activities, during the last 12 months.
## Data collection
Two medical students at the end of their training, helped by 8 students in public health school at the University of Parakou, collected data from 15 March to 15 July 2016. They were trained on data collection tools and supervised by 2 physicians. The data collection form was written in French from the WHO STEPS instrument [11]. It comprised data on sociodemographic factors (age, sex, marital status, occupations, religion, and ethnicity), medical histories (family history of stroke, hypertension, and diabetes), behavioural risk factors (smoking, fruit and vegetable intake, physical activity practice, alcohol consumption), height, weight, and blood pressure. The data collection tools were tested before the survey in another district (“Bannikani”) in Parakou.
At the participant’s home, an individual structured interview was performed in French or the local language. Then, the weight was taken using a weight scale (SECA, United Kingdom) with a precision of 100g. The height was measured using a height rod with a precision of 1 cm (SECA, United Kingdom). After a 5-minute rest, blood pressure was measured with an electronic device (OMRON M3, Japon). The measurement was done in a seated position, on the left arm with the hand resting on a support. Three consecutive measures were taken at 3-minute intervals (between two readings). The mean value of the last two measurements was considered [11].
## Statistical analysis
Statistical analyses were performed using Epi-Info version 7.1.3.10. Software (Epi info, CDC Atlanta, USA). The categorical variables were described by using numbers and percentages and the continuous variables, by using mean ± standard deviation. The Chi2 test or exact Fisher test was used to compare percentages between two groups. The prevalence of stroke risk factors was estimated with a $95\%$ confidence interval. The association between metabolic risk factors for stroke, sociodemographic and behavioral variables was explored through a logistic regression. All the variables with a p-value of 0.20 or less were simultaneously introduced in multivariate analysis by using a step-by-step background approach. The crude and adjusted odds ratios (cOR, aOR) and their confidence intervals at $95\%$ were determined. P-value under 0.05 was considered as significant.
## Ethical considerations
The administrative authorization of the local authorities was obtained before the survey. The local ethical committee of biomedical research of the University of Parakou approved the research (Reference: 029/CLERB-UP/P/SP/R/SA). Each participant approved the written consent form before inclusion. In addition, for persons under the age of 18, written consent from a parent or guardian was obtained prior to their inclusion. The data were managed with confidentiality.
## General characteristic of participants
A total number of 4671 participants were included with a mean age of 27.6 ± 12.9 years and a sex ratio of 0.98. Among the participants, $68.5\%$ were under 30 years old. Almost a fifth of them lived in couples ($16.7\%$) and had no formal education ($16.7\%$); resellers were most represented ($20.4\%$) (Table 1).
**Table 1**
| Unnamed: 0 | N (%) | Obesity | Obesity.1 | Obesity.2 | High blood pressure | High blood pressure.1 | High blood pressure.2 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | N (%) | N (%) | Crude OR[95%CI] | p | N (%) | Crude OR [95%CI] | p |
| Sample | 4671 (100) | 407 (8.7) | | | 332 (7.1) | | |
| Age (years) | | | | <0.001 | | | <0.001 |
| 15–29 | 3198 (68.5) | 134 (4.2) | 1 | | 131 (4.1) | 1 | |
| 30–44 | 927 (19.8) | 182 (19.6) | 5.6 [4.4–7.1] | | 89 (9.6) | 2.5 [1.9–3.1] | |
| 44–59 | 362 (7.8) | 73 (20.2) | 5.8 [4.2–7.9] | | 79 (21.8) | 6.5 [4.2–7.9] | |
| ≥60 | 184 (3.9) | 18 (9.8) | 2.5 [1.5–4.2] | | 33 (17.9) | 5.1 [1.5–4.2] | |
| Sex | | | | <0.001 | | | 0.653 |
| Female | 2365 (50.6) | 315 (13.3) | 1 | | 172 (7.3) | 1 | |
| Male | 2306 (49.4) | 92 (4.0) | 0.3 [0.2–0.4] | | 160 (6.9) | 1.0 [0.8–1.2] | |
| School education level | | | | <0.001 | | | <0.001 |
| | 807 (17.3) | 135 (16.7) | 1 | | 89 (11.0) | 1 | |
| Primary | 875 (18.7) | 132 (15.1) | 0.9 [0.7–1.1] | | 78 (8.9) | 0.8 [0.6–1.1] | |
| Secondary | 2581 (55.3) | 124 (4.8) | 0.3 [0.2–0.3] | | 129 (5.0) | 0.3 [0.3–0.6] | |
| Universitary | 408 (8.7) | 16 (3.9) | 0.2 [0.1–0.3] | | 36 (8.8) | 0.8 [0.5–1.2] | |
| Marital status | | | | <0.001 | | | <0.001 |
| Couple | 2022 (43.3) | 337 (16.7) | 1 | | 233 (11.5) | 1 | |
| Alone | 2649 (56.7) | 70 (2.6) | 0.2 [0.1–0.3] | | 99 (3.7) | 3.4 [2.6–4.3] | |
| Professional activity | | | | <0.001 | | | <0.001 |
| craftsman | 1859 (39.8) | 31 (1.7) | 1 | | 61 (3.3) | 1 | |
| Worker/farmer | 1320 (28.3) | 170 (12.9) | 8.7 [5.9–12.9] | | 112 (8.5) | 2.7 [2.0–3.8] | |
| Resellers | 759 (16.3) | 155 (20.4) | 15.1 [10.2–22.5] | | 80 (10.5) | 3.5 [2.5–4.9] | |
| No economic activity/others | 733 (15.6) | 51 (7.0) | 4.4 [2.8–6.9] | | 79 (10.8) | 3.6 [2.5–5.0] | |
| Alcohol consumption | | | | 0.204 | | | 0.292 |
| No | 3867 (82.8) | 346 (8.9) | 1 | | 282 (7.3) | 1 | |
| Yes | 801 (17.2) | 61 (7.6) | 0.8 [0.6–1.1] | | 50 (6.2) | 0.9 [0.6–1.2] | |
| Low fruit and vegetable intake | | | | 0.406 | | | 0.303 |
| No | 990 (21.3) | 80 (8.0) | 1 | | 78 (7.9) | 1 | |
| Yes | 3667 (78.7) | 327 (8.9) | 1.1 [0.9–1.4] | | 254 (6.9) | 0.9 [0.7–1.1] | |
| Low physical activity practice | | | | 0.001 | | | 0.484 |
| No | 2380 (51.0) | 238 (10.0) | 1 | | 175 (7.4) | 1 | |
| Yes | 2284 (49.0) | 169 (7.3) | 0.7 [0.6–0.9] | | 156 (6.8) | 0.9 [0.7–1.2] | |
| Smoking | | | | 0.264 | | | 0.156 |
| No | 4510 (96.5) | 397 (8.8) | 1 | | 16 (9.9) | 1 | |
| Yes | 161 (3.5) | 10 (6.2) | 0.7 [0.4–1.3] | | 316 (7.0) | 0.9 [0.9–2.5] | |
## Prevalence of stroke risk factors
The prevalence of behavioral risk factors was estimated at: $17.2\%$ ($95\%$ confidence interval (CI) [16.1–18.3]) for alcohol consumption, $21.5\%$ ($95\%$CI [20.1–22.5]) for low fruit and vegetable intake, $51.1\%$ ($95\%$CI [49.6–52.5]) for low physical activity practice and $3.5\%$ ($95\%$CI [2.9–4.0]) for smoking. Concerning the metabolic risk factors for stroke, the prevalence was estimated at: $8.7\%$ ($95\%$CI [7.9–9.6]) for obesity and $7.1\%$ ($95\%$CI [6.4–7.9]) for high blood pressure. The prevalence of self-reported diabetes and dyslipidemia was respectively estimated at $1.7\%$ ($95\%$CI [1.0–1.6]) and $2.2\%$ ($95\%$CI [1.9–2.7]).
## Factors associated with metabolic risk factors for stroke
Data on univariate analysis are displayed in Tables 1 and 2. All the sociodemographic variables were associated with obesity and the high blood pressure, excepted for the variable sex ($$p \leq 0.653$$) that was not associated with high blood pressure (Table 1).
**Table 2**
| Unnamed: 0 | N (%) | Diabetes | Diabetes.1 | Diabetes.2 | Dyslipidemia | Dyslipidemia.1 | Dyslipidemia.2 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | N (%) | N (%) | Crude OR [95%CI] | p | N (%) | Crude OR [95%CI] | p |
| Sample | 4671 (100) | 59 (1.3) | | | 104 (2.2) | | |
| Age (years) | | | | <0.001 | | | <0.001 |
| 15–29 | 3198 (68.5) | 32 (1.0) | 1 | | 67 (2.1) | 1 | |
| 30–44 | 927 (19.8) | 7 (0.8) | 0.8 [0.3–1.7] | | 18 (1.9) | 0.9 [0.6–1.6] | |
| 44–59 | 362 (7.8) | 9 (2.5) | 2.5 [1.2–5.3] | | 11 (3.1) | 1.5 [0.8–2.8] | |
| ≥60 | 184 (3.9) | 11 (6.0) | 6.2 [3.1–12.7] | | 8 (4.4) | 2.1 [1.0–4.5] | |
| Sex | | | | 0.179 | | | 0.457 |
| Female | 2365 (50.6) | 35 (1.5) | 1 | | 49 (2.1) | 1 | |
| Male | 2306 (49.4) | 24 (1.0) | 0.7 [0.4–1.2] | | 55 (2.4) | 1.2 [0.8–1.7] | |
| School education level | | | | 0.721 | | | 0.115 |
| | 807 (17.3) | 11 (1.4) | 1 | | 17 (2.1) | 1 | |
| Primary | 875 (18.7) | 13 (1.5) | 1.1 [0.5–2.5] | | 19 (3.3) | 1.6 [0.9–2.9] | |
| Secondary | 2581 (55.3) | 32 (1.2) | 0.9 [0.5–1.8] | | 50 (2.0) | 0.9 [0.5–1.6] | |
| Universitary | 408 (8.7) | 3 (0.7) | 0.5 [0.2–1.9] | | 8 (2.0) | 0.9 [0.4–2.2] | |
| Marital status | | | | 0.241 | | | 0.572 |
| Couple | 2022 (43.3) | 30 (1.5) | 1 | | 48 (2.4) | 1 | |
| Alone* | 2649 (56.7) | 29 (1.1) | 0.7 [0.4–1.2] | | 56 (2.1) | 0.9 [0.6–1.3] | |
| Professional activity | | | | 0.116 | | | <0.001 |
| Craftsman | 1859 (39.8) | 19 (1.0) | 1 | | 38 (2.1) | 1 | |
| Worker/farmer | 1320 (28.3) | 13 (1.0) | 1.0 [0.5–1.9] | | 27 (2.1) | 1.0 [0.6–1.6] | |
| Resellers | 759 (16.3) | 12 (1.6) | 1.6 [0.8–3.2] | | 24 (3.2) | 1.6 [0.9–2.6] | |
| No economic activity/others | 733 (15.6) | 15 (2.1) | 2.0 [1.0–4.0] | | 15 (2.1) | 1.0 [0.5–1.8] | |
| Alcohol consumption | | | | 0.178 | | | 0.756 |
| No | 3867 (82.8) | 45 (1.2) | 1 | | 85 (2.2) | 1 | |
| Yes | 801 (17.2) | 14 (1.8) | 1.5 [0.8–2.8] | | 19 (2.4) | 1.1 [0.7–1.8] | |
| Low fruit and vegetable intake | | | | 0.716 | | | 0.246 |
| No | 1004 (21.5) | 11 (1.1) | 1 | | 28 (2.8) | 1 | |
| Yes | 3667 (78.5) | 46 (1.3) | 1.1 [0.6–2.2] | | 76 (2.1) | 0.8 [0.5–1.2] | |
| Low physical activity practice | | | | 0.223 | | | 0.033 |
| No | 2291 (49.0) | 33 (1.5) | 1 | | 61 (2.7) | 1 | |
| Yes | 2380 (51.0) | 25 (1.1) | 1.4 [0.8–2.3] | | 43 (1.8) | 0.7 [0.4–0.9] | |
| Smoking | | | | 0.034 | | | 0.003 |
| No | 4510 (96.5) | 54 (1.2) | 1 | | 95 (2.1) | 1 | |
| Yes | 161 (3.5) | 5 (3.1) | 2.6 [1.1–6.7] | | 9 (5.6) | 2.7 [1.4–5.5] | |
In contrary, no behavioral factors were neither associated with obesity nor with high blood pressure, excepted for the low physical activity practice ($$p \leq 0.002$$) that was significantly associated with obesity (Table 1). As for the self-reported diabetes mellitus, only the variables age ($p \leq 0.001$) and smoking ($$p \leq 0.034$$) were significantly associated (Table 2). Concerning the dyslipidemia, the association with age ($p \leq 0.001$), professional activity ($p \leq 0.001$), low physical activity practice ($$p \leq 0.035$$) and smoking ($$p \leq 0.003$$) was significant (Table 2).
Data on multivariate analysis are displayed in Tables 3 and 4. Age ($p \leq 0.001$), sex ($p \leq 0.001$), marital status ($p \leq 0.001$) and professional occupation ($$p \leq 0.010$$) were associated with obesity (Table 3). The obesity was less prevalent in men (adjusted prevalence-ratio (aOR) = 0.4; $95\%$ CI [0.3–3.3]) compared to women and in contrary, more prevalent in older participants compared to those aged 15 to 29 years. It was also more prevalent in participants living in couple (aOR = 2.2; $95\%$ CI [1.6–3.1]) than those living alone. Concerning the professional occupation, the resellers had the highest prevalence (aOR = 3.5; $95\%$ CI [2.1–5.6]) compared to participants who had other occupations.
Only age ($p \leq 0.001$) and marital status ($p \leq 0.001$) were significantly associated with high blood pressure after adjustment (Table 3). The prevalence of high blood pressure was higher in older participants compared to “15–29” years old: “30–44” years old (aOR = 1.7; $95\%$ CI [1.2–2.4]), “45–59” years old (aOR = 4.4; $95\%$ CI [3.1–6.2]), and “≥60” years old (aOR = 3.5; $95\%$ CI [2.2–5.4]). This prevalence was also higher in participants living in couple (aOR = 1.7; $95\%$ CI [1.2–2.5]) compared to those living alone.
The prevalence of diabetes increased with the age (Table 4) with the highest prevalence observed among the group of participants aged “≥60” years old (aOR = 5.9; $95\%$ CI [2.9–12.1]). As for dyslipidemia, it was more prevalent among smokers (aOR = 2.3; $95\%$ CI [1.1–4.8]); a lower prevalence was observed among participants who had a low practice physical activity (aOR = 0.7; $95\%$ CI [0.4–0.9]) compared to with those physically active (Table 4).
## Discussion
This study showed the magnitude of stroke risk factors in a sample of relatively young people (mean age: 27.6 ± 12.9 years), in Titirou, in 2016.
The slight female predominance and the high proportion of people with none school education level are in line with national demographic data [10].
The prevalence of alcohol consumption ($21.5\%$) is in the range reported during the STEPS surveys conducted in SSA from 2013 to 2016 (1.4–$40.7\%$) [8]. Smoking ($3.5\%$) was less prevalent compared to STEPS survey results (4.2–$13.3\%$), probably due to the young age of participants. The same observation was made for low fruit and vegetable intake ($21.5\%$) for which the prevalence varied between 67.9–$97.6\%$ [8]. Fruit and vegetable intake depends on their availability and accessibility. Our result may be linked to the greater availability of fruits and vegetables in Parakou compared to other towns in Benin. Indeed, Parakou has many gardening areas and is surrounded by fields. On the contrary, low physical activity practice prevalence ($51.1\%$) was higher compared to STEPS survey data (4.3–$17.7\%$) [8].
The prevalence of obesity ($8.7\%$) was in line with the STEPS survey results (1.2–$20.5\%$) while that of high blood pressure ($7.1\%$) was lower. For instance, considering the same definition, $17.6\%$ was reported in Burkina-Faso in 2013, $23.1\%$ in Uganda in 2014, $25.2\%$ in Benin in 2015 [8,9]. Our result could be explained by the fact that the sample comprised younger people while in the STEPS studies, median ages were higher. Concerning the diabetes mellitus and dyslipidemia variables, lower self-reported diabetes mellitus ($0.7\%$) and raised total blood cholesterol ($0.4\%$) were previously noted in the STEPS survey in Benin. Our higher results could be explained by the fact that our study took place in the department of Borgou that has been demonstrated as an area of high prevalence of diabetes [12,13]. Further studies more focused in this area and based on the blood glucose measurement could allow comparisons.
Age and gender were significantly associated with obesity. The results are consistent with the literature data. The increase of obesity with age as observed in our study was previously described [8]. In addition, a positive association between obesity and female gender was noted during several cross-sectional studies in SSA, Brazil, and China [8,14–18]; these results were explained by social and cultural factors. However, a contrary result (higher prevalence of obesity in men than women) was reported by Boua et al. during a study in demographic and health surveillance site in Burkina Faso [19].
The prevalence of obesity was higher among participants living in couples than those living alone and could be explained by the fact that couples might be more sedentary than those living alone. Dagne et al. in Ethiopia had also noted that being married increased the risk of obesity [20]. Resellers seemed more obese than people who practice other activities, probably because of differences in lifestyle.
High blood pressure prevalence increased with the age, but the classic linear association [21–23] was not observed. Contrary to literature data, no association was noted neither between obesity and behavioral factors, nor between high blood pressure and behavioral factors. A larger study could show these associations.
The link between age and diabetes was confirmed in this study. In fact, diabetes increased linearly with age as previously reported in some steps surveys in SSA [8]. As one might expect, a positive association was observed between dyslipidemia and smoking. On the contrary, dyslipidemia was inversely linked to physical activity practice. One explanation to this trend could be that people who knew their status concerning cholesterol rate may be more sensitized to regular physical activity practice than the others.
## Strengths and limitations of the study
This study used a methodology that allowed extrapolating the results to Titirou District. Otherwise, the sample comprised young people aged 15 to 17 years that was not considered in the STEPS surveys. The results provided data on stroke risk factors that could be used for the implementation of targeted interventions in Titirou. For scientific communities, these data showed that hypertension, diabetes, and raised total blood cholesterol are not, yet the main metabolic risk factors in young population in Benin. The main one is rather obesity according to the results obtained in our study, calling then for actions against obesity by tackling physical inactivity. The data can also be used in an advocacy for strengthening of early detection and management of metabolic risk factors in the peripheral health centers of Parakou.
Except for obesity and high blood pressure, stroke risk factors were assessed based on self-reported information which could have introduced information bias due to a wrong or not worthy declaration. In addition, considering the lack of measurements for diabetes and dyslipidemia diagnosis in this study, the prevalence of diabetes and dyslipidemia may have probably been underestimated.
The sampling could be considered by some experts as non-random despite the door-to-door approach in random directions of the neighborhoods. Another limitation is that the population of *Titirou is* not representative of the population of Parakou what does not allow us to extrapolate our results to the entire population of Parakou.
## Conclusion
The study revealed that a significant part of the residents of Titirou, aged at least 15 years, lived with stroke risk factors. For primary prevention of stroke in this community, adapted actions, including the physical activity promotion should be implemented. The actions should take into account socio-cultural realities.
## References
1. 1World Health Organisation. Preventing chronic diseases: a vital investment: WHO global report.
WHO
2016 [Cited 1 Jul 2021]. Available from://apps.who.int/iris/handle/10665/43314.. (2016.0)
2. **Global, regional, and national burden of stroke, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016**. *Lancet Neurol* (2019.0) **18** 439-58. DOI: 10.1016/S1474-4422(19)30034-1
3. Cossi M-J, Gobron C, Preux P-M, Niama D, Chabriat H, Houinato D. **Stroke: prevalence and disability in Cotonou, Benin**. *Cerebrovasc Dis* (2012.0) **33** 166-72. DOI: 10.1159/000334195
4. Gnonlonfoun D, Adjien KC, Adoukonou T, Goudjinou G, Weke N, Houinato D. **Prévalence des principales affections neurologiques chez les sujets de 15 ans et plus en population générale à Cotonou, Bénin.**. *Rev Neurol* (2017.0) **173** S180
5. Adoukonou T, Yahouédéou B, Agbétou M, Hountada H, Choki B, Kossi O. **Prevalence of stroke survivors in Parakou in northern Benin: A door-to-door community survey.**. *Rev Neurol* (2020.0) **176** 839-45. DOI: 10.1016/j.neurol.2020.02.005
6. Adoukonou T, Agbétou M, Bangbotché R, Kossi O, Fotso Mefo P, Magne J. **Long-Term Mortality of Stroke Survivors in Parakou: 5-Year Follow-Up**. *J Stroke Cerebrovasc Dis* (2020.0) **29** 104785. DOI: 10.1016/j.jstrokecerebrovasdis.2020.104785
7. Adoukonou T, Kouna-Ndouongo P, Codjia J-M, Covi R, Tognon-Tchegnonsi F, Preux P-M. **Direct hospital cost of stroke in Parakou in northern Benin**. *Pan Afr Med J* (2013.0) **16** 121. DOI: 10.11604/pamj.2013.16.121.2790
8. 8World Health Organization. WHO STEPS Country Reports. WHO [Cited 1 Jul 2021]. Available from: https://www.who.int/teams/noncommunicable-diseases/surveillance/systems-tools/steps.. *WHO STEPS Country Reports*
9. 9Ministry of Health of Benin. Final report of Benin STEPS survey in 2015. Ministry of Health
2017 [Cited 1 Jul 2021]. Available from: https://www.who.int/teams/noncommunicable-diseases/surveillance/data/benin.. (2017.0)
10. 10Institut national de la statistique et de l’analyse économique (INSAE BENIN). Recensement Général de la Population et de l’Habitation. INSAE 2016 [Cited 1 Feb 2021]. Available from: http://www.insae-bj.org/recensement-population.html?file=files/enquetes-recensements/rgph/Resultats_provisoires_RGPH4_2103.pdf.
11. 11World Health Organization. WHO STEPS Surveillance Manual. WHO [Cited 1 Jul 2021] Available from: https://www.who.int/teams/noncommunicable-diseases/surveillance/systems-tools/steps/manual.. *WHO STEPS Surveillance Manual*
12. Azandjeme CS, Delisle H, Fayomi B, Ayotte P, Djrolo F, Houinato D. **High serum organochlorine pesticide concentrations in diabetics of a cotton producing area of the Benin Republic (West Africa).**. *Environ Int* (2014.0) **69** 1-8. DOI: 10.1016/j.envint.2014.04.002
13. Djrolo F, Adoukonou T, Houehanou C, Houinato JD, Houinato D. **Diabetes in Borgou Department in Benin: Prevalence and Associated Factors.**. *J Diabetes Mellit* (2015.0) **05** 90
14. Li YQ, Wright SCD. **Risk factors for cardiovascular disease in the Ga-Rankuwa community**. *Curationis* (2007.0) **30** 79-87. DOI: 10.4102/curationis.v30i4.1120
15. Rhazi KE, Nejjari C, Zidouh A, Bakkali R, Berraho M, Gateau PB. **Prevalence of obesity and associated sociodemographic and lifestyle factors in Morocco.**. *Public Health Nutr* (2011.0) **14** 160-7. DOI: 10.1017/S1368980010001825
16. Oguoma VM, Nwose EU, Skinner TC, Digban KA, Onyia IC, Richards RS. **Prevalence of cardiovascular disease risk factors among a Nigerian adult population: relationship with income level and accessibility to CVD risks screening**. *BMC Public Health* (2015.0) **15** 397. DOI: 10.1186/s12889-015-1709-2
17. Abadio Finco FDB, Finco MVA, Graeve L. **Prevalence and associated factors of overweight in adults of rural Brazilian Amazonia**. *NFS J* (2017.0) **9** 1-7
18. Song N, Liu F, Han M, Zhao Q, Zhao Q, Zhai H. **Prevalence of overweight and obesity and associated risk factors among adult residents of northwest China: a cross-sectional study**. *BMJ Open* (2019.0) **9** e028131. DOI: 10.1136/bmjopen-2018-028131
19. Boua RP, Sorgho H, Rouamba T, Nakanabo Diallo S, Bognini JD, Konkobo SZ. **Gender differences in sociodemographic and behavioural factors associated with BMI in an adult population in rural Burkina Faso—an AWI-Gen sub-study**. *Glob Health Action* (2018.0) **11** 1527557. DOI: 10.1080/16549716.2018.1527557
20. Dagne S, Gelaw YA, Abebe Z, Wassie MM. **Factors associated with overweight and obesity among adults in northeast Ethiopia: a cross-sectional study**. *Diabetes Metab Syndr Obes Targets Ther* (2019.0) **12** 391-9. DOI: 10.2147/DMSO.S179699
21. Guwatudde D, Nankya-Mutyoba J, Kalyesubula R, Laurence C, Adebamowo C, Ajayi I. **The burden of hypertension in sub-Saharan Africa: a four-country cross sectional study**. *BMC Public Health* (2015.0) **15** 1211. DOI: 10.1186/s12889-015-2546-z
22. Beaney T, Schutte AE, Tomaszewski M, Ariti C, Burrell LM, Castillo RR. **May Measurement Month 2017: an analysis of blood pressure screening results worldwide**. *Lancet Glob Health* (2018.0) **6** e736-43. DOI: 10.1016/S2214-109X(18)30259-6
23. Desormais I, Amidou SA, Houehanou YC, Houinato SD, Gbagouidi GN, Preux PM. **The prevalence, awareness, management and control of hypertension in men and women in Benin, West Africa: the TAHES study**. *BMC Cardiovasc Disord* (2019.0). DOI: 10.1186/s12872-019-01273-7
|
---
title: Exploring preconception signatures of metabolites in mothers with gestational
diabetes mellitus using a non-targeted approach
authors:
- Ling-Jun Li
- Ximeng Wang
- Yap Seng Chong
- Jerry Kok Yen Chan
- Kok Hian Tan
- Johan G. Eriksson
- Zhongwei Huang
- Mohammad L. Rahman
- Liang Cui
- Cuilin Zhang
journal: BMC Medicine
year: 2023
pmcid: PMC10022116
doi: 10.1186/s12916-023-02819-5
license: CC BY 4.0
---
# Exploring preconception signatures of metabolites in mothers with gestational diabetes mellitus using a non-targeted approach
## Abstract
### Background
Metabolomic changes during pregnancy have been suggested to underlie the etiology of gestational diabetes mellitus (GDM). However, research on metabolites during preconception is lacking. Therefore, this study aimed to investigate distinctive metabolites during the preconception phase between GDM and non-GDM controls in a nested case–control study in Singapore.
### Methods
Within a Singapore preconception cohort, we included 33 Chinese pregnant women diagnosed with GDM according to the IADPSG criteria between 24 and 28 weeks of gestation. We then matched them with 33 non-GDM Chinese women by age and pre-pregnancy body mass index (ppBMI) within the same cohort. We performed a non-targeted metabolomics approach using fasting serum samples collected within 12 months prior to conception. We used generalized linear mixed model to identify metabolites associated with GDM at preconception after adjusting for maternal age and ppBMI. After annotation and multiple testing, we explored the additional predictive value of novel signatures of preconception metabolites in terms of GDM diagnosis.
### Results
A total of 57 metabolites were significantly associated with GDM, and eight phosphatidylethanolamines were annotated using HMDB. After multiple testing corrections and sensitivity analysis, phosphatidylethanolamines 36:4 (mean difference β: 0.07; $95\%$ CI: 0.02, 0.11) and 38:6 (β: 0.06; 0.004, 0.11) remained significantly higher in GDM subjects, compared with non-GDM controls. With all preconception signals of phosphatidylethanolamines in addition to traditional risk factors (e.g., maternal age and ppBMI), the predictive value measured by area under the curve (AUC) increased from 0.620 to 0.843.
### Conclusions
Our data identified distinctive signatures of GDM-associated preconception phosphatidylethanolamines, which is of potential value to understand the etiology of GDM as early as in the preconception phase. Future studies with larger sample sizes among alternative populations are warranted to validate the associations of these signatures of metabolites and their predictive value in GDM.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12916-023-02819-5.
## Background
Metabolic alterations in a healthy pregnancy include a $30\%$ increment in basal endogenous glucose production by late pregnancy, primarily by hepatic function [1]. In order to maintain euglycemia, circulating fasting glucose concentrations decrease during pregnancy mainly due to an increase in plasma volume in early pregnancy and an increase in glucose utilization in later gestation by the fetoplacental unit [2, 3]. Women with disruption during such metabolic adaptation during any time of pregnancy, such as reduced peripheral insulin sensitivity (e.g., reduced glucose uptake in skeletal muscle and adipose tissue, adverse amino acid, and lipid metabolism) [1] and diminished pancreatic β-cell reserve [4], could develop hyperglycemia also known as gestational diabetes mellitus (GDM).
GDM affects from $1\%$ to > $30\%$ of pregnancies worldwide and is exceptionally prevalent in Asian populations such as Saudi Arabia, India, and Singapore [5]. GDM is also important for public health awareness for two reasons. To begin with, GDM increases the risk of pregnancy complications for both mothers and offspring [3]. Subsequently, a GDM diagnosis identifies populations at risk (i.e., women and their offspring) of obesity, diabetes, and premature cardiovascular disease in the long run [3, 6, 7]. The etiology of GDM remains unclear, even though its involvement with maternal obesity, inflammation, and oxidative stress-mediated insulin resistance has been widely suggested [3, 8].
Metabolomics approach, defined as the study of global metabolite profiles in bio-samples under a given set of conditions, has been beneficial for assessing cardiometabolic conditions [9]. Emerging evidence rooted in epidemiological studies has also investigated plasma, serum, and even urine metabolite profiles during pregnancy that significantly differed between women with and without GDM [8, 10–15]. However, the findings in metabolite panels were inconsistent, and later assessment in pregnancy was subjective to reverse causality [9, 16]. In addition, metabolic disruption as early as the preconception phase could help in understanding the etiology of GDM, which has been lacking in current literature. Therefore, we aimed to investigate distinctive metabolites during the preconception phase between GDM and non-GDM controls using a non-targeted approach, in a nested case–control study among homogenous Chinese pregnant subjects in Singapore.
## Study population and study design
Within the Singapore PREconception Study of long-Term maternal and child Outcomes (SPRESTO), a preconception and pregnancy cohort conducted from February 2015 to October 2017, we recruited Chinese, Malay, and Indian women without type 1 or type 2 diabetes, aged 18–45 years, and planning to conceive within 12 months. We published the cohort profile earlier and described the study objective and protocol in detail [17]. The study was conducted according to the guidelines under the Declaration of Helsinki and approved by the SingHealth Centralized Institute Review Board ($\frac{2014}{629}$/D).
## GDM diagnosis
Among 376 singleton live births, we ascertained 33 Chinese pregnant women with GDM according to the International Association of Diabetes and Pregnancy Study (IADPSG) criteria between 24 and 28 weeks of gestation [18]. We then matched them at a 1:1 ratio with 33 non-GDM Chinese women by age (± 2 years) and categories of pre-pregnancy body mass index (ppBMI) (underweight [< 18.5 kg/m2], normal weight [18.5–22.9 kg/m2], overweight [23.0–27.4 kg/m2], and obese [≥ 27.5 kg/m2]) [17]. We performed blood collection among all participants during the preconception phase (within 12 months prior to conception).
## Non-targeted metabolomics approach
Non-targeted metabolomics analysis was performed using liquid chromatography-mass spectrometry (LC–MS), as described in accordance with standard procedures [19, 20]. Upon enrolment during the preconception phase, we collected fasting serum samples from all participants and stored biospecimens at – 80 °C until thawing them immediately before assay. Briefly, fasting serum samples were processed with methanol or butanol/methanol precipitation for polar or non-polar metabolite analysis and were run in the same batch. Quality control samples were prepared by mixing equal amounts from all the samples and analyzed after every ten samples to monitor the stability of the system. We then reported the intra-coefficient of variation (CV) accordingly. LC–MS analysis was conducted using Agilent 1290 ultrahigh pressure liquid chromatography system equipped with a 6550 quadrupole time-of-flight mass detector managed by a MassHunter workstation. Mass data were collected between m/z 100 and 1000 Da at a rate of two scans per second. The electrospray ionization mass spectra were acquired in both positive and negative ion mode. Two reference masses were continuously infused to the system to allow constant mass correction during the run: m/z 121.0509 (C5H4N4) and m/z 922.0098 (C18H18O6N3P3F24). Raw spectrometric data were analyzed by the Agilent MassHunter Qualitative Analysis software. The molecular features characterized by retention time, chromatographic peak intensity and accurate mass were obtained using Agilent MassHunter Mass Profiler Professional software. A total of 5263 features extracted from both positive and negative ion modes were used for subsequent statistical analyses. All these features have an intensity ≥ 20,000 counts (approximately three times the detection limit of our LC–MS instrument) and were found in at least $80\%$ of the samples in either GDM or non-GDM groups.
## Covariates
During recruitment, trained research coordinators conducted in-person interviews and measured the participants’ weight, height, and blood pressure at the study clinic. Covariates collected via questionnaires at study entry included socio-demographic factors, health history, menstrual characteristics, and lifestyle behaviors. In detail, they were maternal age, parity (nulliparous vs. parous), pre-pregnancy BMI, family history of diabetes, smoking (never vs. past or current), alcohol intake (never vs. past or current), micronutrient supplements intake in the past 3 months (yes vs. no). In addition, a 2-h 75 g two-time point oral glucose tolerance test (OGTT) was performed at the clinic during the first preconception visit. Preconception prediabetes was diagnosed based on fasting glucose concentration at ≥ 6.1 mmol/L and/or 2-h glucose concentration at ≥ 7.8 mmol/L, according to World Health Organization (WHO) guidelines [21].
## Statistical analyses
We performed the following steps for our statistical analyses. Step 1, we compared baseline characteristics between GDM and non-GDM controls using either generalized linear mixed model (GLMM) or generalized estimation equation (GEE) to account for 1:1 ratio matching effect. Step 2, we applied Student’s t-test to identify a pool of metabolite candidates, from which we further used GLMM to identify signatures of metabolites with adjustment of maternal age and ppBMI and accounting for the case–control matching effect. Step 3, upon GLMM identification, we performed annotation via Human Metabolite Database (HMDB) and pathway analysis via the KEGG database and assessed the correlation among all signatures of metabolites using the spearman rank correlation. Step 4, we compared the means and standard errors of all annotated metabolites after log-transformation their peak area between GDM and non-GDM controls using GLMM model and indicated regression estimate in mean difference between GDM and non-GDM subjects for each signature of metabolite identified. Due to the exploratory nature of our study, we corrected multiple testing using the false discovery rate (FDR) approach [22] and conducted sensitivity analysis by additionally adjusting for preconception diabetes according to WHO 2015 criteria [21], family history of diabetes, and parity. Step 5, all annotated metabolites from the preconception phase were presented in box plots, including median and inter-quartile range (IQR). Step 6, we performed the receiver operating characteristic (ROC) curve to calculate the area under the curve (AUC) value for GDM using annotated metabolites and traditional risks (e.g., maternal age, ppBMI) at the preconception phase.
In the descriptive table, we expressed all maternal characteristics data as median with IQR or mean with standard deviation (SD) when appropriate. We conducted all statistical analyses using the SIMCA 13.2 (Umetrics, Umea, Sweden), MetaboAnalyst (Version 4.0), and R Software (Version 3.5.0). We reported regression estimates in mean difference with $95\%$ confidence interval (CI) after log-transformation of all signatures of metabolites and deemed significance at p-value (2-sided) less than 0.05.
## Results
No significant difference in baseline maternal characteristics were identified between GDM cases and matched controls (Table 1). A total of 57 metabolites were significantly associated with GDM. Nine were further successfully annotated. Since two were duplicated, a total of eight phosphatidylethanolamines were identified using HMDB, including 34:1, 34:2, 36:2, 36:4, 38:4, 38:5, 38:6, and 40:6. All of them were normalized after log-transformation in peak area and highly correlated with each other using the spearman rank correlation (Additional file 1: Tab. S1).Table 1Preconception maternal characteristics between GDM and non-GDM matching controls by maternal age, ethnicity, and pre-pregnancy body mass index categoryMaternal characteristics before pregnancyGDM subjects ($$n = 33$$)Non-GDM subjects ($$n = 33$$)p*Mean (SD) or N (%)Mean (SD) or N (%)Age, years30.09 (2.52)30.59 (2.39)0.50Parity0.55 023 (67.65)22 (68.75) 19 (26.47)10 (31.25) 22 (5.88)0 [0]BMI, kg/m223.5* (4.13)23.77* (4.24)0.44BMI category,0.86 Underweight, < 18.5 kg/m21 (3.0)3 (9.1) Normal weight, 18.5–22.9 kg/m217 (51.5)17 (51.5) Overweight, 23.0–27.4 kg/m27 (21.2)6 (18.2) Obese, > = 27.5 kg/m28 (24.2)7 (21.2)Waist to hip ratio0.85 (0.06)0.86 (0.04)0.60HbA1c, mmol33.2 (2.96)32.5 (3.09)0.33Prior GDM, yes0 [0]0 [0]–OGTT Fasting glucose, mmol/L4.91 (0.5)4.8 (0.26)0.16 2-h glucose, mmol/L6.1 (1.61)5.48 (1.37)0.13 Prediabetes, yes1 (3.23)2 (6.25)0.10Family history of DM, yes12 (36.36)7 (21.88)0.27Time-to-pregnancy, months2.45 (7.74)3.84 (5.22)0.86Abbreviations: GDM Gestational diabetes mellitus, IQR Inter-quartile range, BMI Body mass index, HbA1c Glycated hemoglobin, OGTT Oral glucose tolerance test, DM Diabetes*Accounted for matching with generalized linear mixed model or generalized estimating equation, if applicable These eight phosphatidylethanolamines expressed significantly higher signals in the GDM than the non-GDM controls (β range of mean difference: 0.04–0.07, all $p \leq 0.05$). The intra-CV in these eight phosphatidylethanolamines ranged between 2.86 and $4.14\%$ (Additional file 2: Tab. S2). After FDR correction and sensitivity analysis, phosphatidylethanolamines 36:4 (mean difference β: 0.07; $95\%$ CI: 0.02, 0.11, FDR-corrected p-value: 0.0084) and 38:6 (mean difference β: 0.06; $95\%$ CI: 0.004, 0.11; FDR-corrected p-value: 0.0032) remained significantly different between GDM and non-GDM controls (Table 2) during preconception phase. We applied scatterplot overlaying box plots to showcase the distribution and signal mean differences between GDM and non-GDM controls according to eight phosphatidylethanolamines (Additional file 3: Fig. S1) and highlighted 36:4 and 38:6 in Fig. 1. Sensitivity analysis by additionally adjusting for family history of diabetes, parity, prediabetes during the preconception phase, and prior GDM history did not significantly attenuate our findings on 36:4 and 38:6 (Additional file 4: Tab. S3).Table 2Comparison of eight annotated metabolites between GDM and non-GDM subjects using GLMM during preconception phase within 12 months prior to conceptionPhosphatidylethanolaminesGDM vs. non-GDM (reference)Mean differenceβ ($95\%$ CI)p valueFDR34:10.04 (− 0.01, 0.09)0.040.049834:20.06 (− 0.01, 0.12)0.030.049836:20.05 (− 0.004, 0.11)0.0490.049836:40.07 (0.02, 0.11)0.0010.008438:40.05 (0.01, 0.09)0.030.049838:50.04 (− 0.003, 0.09)0.040.049838:60.06 (0.004, 0.11)0.010.033240:60.06 (0.0002, 0.11)0.0490.0498GLMM was adjusted for maternal age and pre-pregnancy body mass index (pp-BMI) at the preconception phaseAbbreviations: GDM Gestational diabetes mellitus, GLMM Generalized linear mixed model, FDR False discovery rateFig. 1Scatter plots and box plots of phosphatidylethanolamines 36:4 and 38:6 at preconception (within 12 months prior to conception) phase between GDM and non-GDM controls in the nested case–control study embedded in SPRESTO study Furthermore, no pathway was identified among eight annotated metabolites, whereas 12 pathways were identified among unannotated metabolites via KEGG pathway analysis (Additional file 5: Tab. S4).
Since there were eight metabolites identified while two remained significant after FDR correction, we performed the ROC curve for all candidate models based on different sets of metabolites. With 36:4 and 38:6 signals in addition to traditional risk factors such as maternal age, ppBMI, family history of diabetes, prior history of GDM, preconception prediabetes, and parity, the AUC increased from 0.620 to 0.773 and R2 increased from 0.048 to 0.236 (Table 3 and Fig. 2). With all eight metabolites’ signals in addition to the same set of traditional risk factors mentioned above, the AUC increased from 0.620 to 0.843 and R2 increased from 0.048 to 0.377 (Table 3 and Fig. 2). Due to the multiple adjustments within a relatively small sample size, all comparisons did not reach statistical significance. Table 3Predictive value for GDM using traditional and novel biomarkers identified in our cohortModelsR2AUCP valueRef (model 1)Ref (model 2)Ref (model 3)Metabolites prediction model Model 1, adjusting for 36:4, 38:60.1270.694N/AN/AN/A Model 2, adjusting for eight metabolites (34:1, 34:2, 36:2, 36:4, 38:4, 38:5, 38:6, and 40:6)0.1940.7540.554N/AN/ATraditional risks prediction model Model 3, adjusting for maternal age, ppBMI, family history of GDM, prior history of GDM, preconception prediabetes, parity0.0480.6200.3820.612N/AMetabolites and traditional risks combined prediction model Model 4, model 1 + model 30.2360.7730.3600.7380.480 Model 5, model 2 + model 30.3770.8430.4330.2570.300Abbreviations: GDM Gestational diabetes mellitus, ppBMI pre-pregnancy body mass index, AUC Area under the curveFig. 2Receiver operating characteristic (ROC) curve admissions of the predictive models on GDM using identified metabolites and traditional risks. The red line represents the ROC curve of model 3: GDM ~ all traditional maternal risk factors including maternal age, ppBMI, family history of diabetes, prior history of GDM, preconception prediabetes and parity at the preconception phase (R2 = 0.048, AUC = 0.620). The yellow line represents the ROC curve of model 5: GDM ~ all eight metabolites identified at preconception in addition to model 3 (R2 = 0.377, AUC = 0.843)
## Discussion
In our longitudinal study of 33 pairs of Chinese GDM and non-GDM controls on non-targeted metabolomics signature at the preconception phase within 12 months prior to conception, fifty-seven metabolites were significantly related to GDM. Among them, eight phosphatidylethanolamines were successfully annotated with a range of fatty acid chain lengths. After FDR correction in multiple testing, only phosphatidylethanolamines 36:4 and 38:6 remained significant in association with GDM. Compared with non-GDM controls, these two glycerophospholipids were related to adverse cardiometabolic profiles and exhibited significantly higher signals during the preconception phase in GDM subjects.
Emerging evidence has shown that serum, plasma, and even urine metabolites (e.g., lipids, fatty acids, amino acids, acylcarnitines, dopamine) were associated with incident GDM in either early or mid-pregnancy [8–11, 23, 24]. However, we are unaware of studies on metabolomics measurements before pregnancy. Women at risk of GDM already exert differences in fat deposition and glycose tolerance even before their pregnancy [25]. The signatures of metabolites in preconception are of potential value to understand the pathophysiology of GDM.
Our study is the first to investigate the missing link of changes in metabolites as early as in the preconception phase. Among the eight metabolites identified to differentiate GDM from non-GDM controls, all were phosphatidylethanolamines—a class of glycerophospholipids that is made in the endoplasmic reticulum (ER) via the cytidine diphosphate-diacylglycerol-ethanolamine pathway [26]. Since phosphatidylethanolamine is one of the most abundant glycerophospholipids in mammalian cells and is easy to obtain from human blood and small biopsy tissues, clinical studies in the past decade have widely investigated its association with insulin sensitivity. Emerging evidence has demonstrated the key role of phosphatidylethanolamine in the insulin signaling pathway, and it was suggested that increased phosphatidylcholine/phosphatidylethanolamines ratio was associated with reduced insulin sensitivity [27] and elevated among patients with type 2 diabetes [28]. In mice models, accumulation of phosphatidylethanolamine production in mitochondria is suggested to modulate glucose [29] and increase diacylglycerol [30], the latter of which is known for causing insulin resistance in cells [31].
Our study showed a significantly higher signal of phosphatidylethanolamines 36:4 and 38:6 in GDM cases than in non-GDM controls. Emerging evidence also showed elevated levels of phosphatidylethanolamines (e.g., 18:1, 22:2, 36:1, 36:4, and 38:6) in both mid- and late pregnancy [11, 32], among women with different racial backgrounds. Even though their biological functions underlying the pathogenesis of GDM are largely unknown, we postulated that phosphatidylethanolamines could adversely impact cellular activity and glucose and fatty acid metabolism [27], since lipid metabolism disorders often accompany glucose metabolism disorders in diabetes, and the complex relationship between metabolism and numerous lipid metabolites needs further elucidation. A recent longitudinal study also reported that glycerophospholipids could predict the transition from GDM to type 2 diabetes in the early postpartum period, which was a superior indicator to clinical parameters [33]. Our findings and others might provide strong evidence to pinpoint the consistent and distinctive values of certain types of phosphatidylethanolamines related to GDM, from preconception to postpartum phases. And the identification of these might help classify and prevent GDM and even postpartum type 2 diabetes among women at risk. Future studies should explore such metabolites and pathways underlying the GDM etiology in alternative populations and with larger sample sizes.
The strength of this study lies in the preconception blood sample within 12 months prior to conception, and the comprehensive measures of metabolomics based on an untargeted approach, from a group of homogenous Chinese women. The study is not without limitations. Firstly, the relatively small sample size may limit the statistical power of the study. We are not able to validate our results in a subset of samples within our cohort. However, even with 33 pairs of GDM cases and controls, we robustly annotated two metabolites that could distinctively differentiate GDM subjects from non-GDM controls after FDR correction. Secondly, the study design of one-time measurement of metabolites within 12 months prior to conception might not capture the dynamic trajectories of metabolic profiles. Even though it may be practically much more challenging, future studies with longitudinal measures before pregnancy are warranted. Considering our subjects were more motivated to maintain a relatively healthy lifestyle and physique to achieve a successful pregnancy than the general population [34], and those who entered pregnancy with livebirth outcomes had a more healthful plant-based eating dietary pattern [35], the impact of identifying metabolites due to dynamic trajectories in our study is speculated less significant than the general population. Lastly, even though we developed the prediction model from a nested case–control study, the under-sampling of non-outcomes might potentially overestimate the AUC performance in our study.
## Conclusions
Our data identified distinctive signatures of metabolites of GDM, specifically in preconception fasting serum (i.e., phosphatidylethanolamines 36:4 and 38:6), which is of potential value to understand in depth on the etiology of GDM as early as in the preconception phase. Future studies with larger sample sizes in a multiracial prospective study setting with external validation and multiple time points of metabolites testing are warranted to validate the association of these signatures with GDM and even evaluate the predictive value of such metabolites.
## Supplementary Information
Additional file 1: Tab. S1. Spearman rank correlation among all annotated phosphatidylethanolamines ($$n = 8$$).Additional file 2: Tab. S2. The intra-coefficient of variation (CV) among all identified phosphatidylethanolamines in our study after eight attempts of quality control. Additional file 3: Fig. S1. Scatter plots and box plots of eight annotated phosphatidylethanolamines at preconception (within 12 months prior to conception) phase between GDM and non-GDM controls in the nested case–control study embedded in SPRESTO study. Additional file 4: Tab S3. Sensitivity analysis. Additional file 5: Tab. S4. KEGG pathway analyses of all GLMM-identified and yet unannotated metabolites ($$n = 48$$).
## References
1. Catalano PM. **Trying to understand gestational diabetes**. *Diabet Med* (2014) **31** 273-281. DOI: 10.1111/dme.12381
2. Zhu Y, Zhang C. **Prevalence of**. *Curr Diab Rep* (2016) **16** 7. DOI: 10.1007/s11892-015-0699-x
3. McIntyre HD, Catalano P, Zhang C, Desoye G, Mathiesen ER, Damm P. **Gestational diabetes mellitus**. *Nat Rev Dis Primers* (2019) **5** 47. DOI: 10.1038/s41572-019-0098-8
4. Buchanan TA. **Pancreatic B-cell defects in gestational diabetes: implications for the pathogenesis and prevention of type 2 diabetes**. *J Clin Endocrinol Metab* (2001) **86** 989-993. DOI: 10.1210/jcem.86.3.7339
5. Li LJ, Huang L, Tobias DK, Zhang C. **Gestational**. *Front Endocrinol (Lausanne)* (2022) **13** 840331. DOI: 10.3389/fendo.2022.840331
6. Daly B, Toulis KA, Thomas N, Gokhale K, Martin J, Webber J, Keerthy D, Jolly K, Saravanan P, Nirantharakumar K. **Increased risk of ischemic heart disease, hypertension, and type 2 diabetes in women with previous gestational diabetes mellitus, a target group in general practice for preventive interventions: a population-based cohort study**. *PLoS Med* (2018) **15** e1002488. DOI: 10.1371/journal.pmed.1002488
7. Clausen TD, Mathiesen ER, Hansen T, Pedersen O, Jensen DM, Lauenborg J, Damm P. **High prevalence of type 2 diabetes and pre-diabetes in adult offspring of women with gestational diabetes mellitus or type 1 diabetes: the role of intrauterine hyperglycemia**. *Diabetes Care* (2008) **31** 340-346. DOI: 10.2337/dc07-1596
8. Koos BJ, Gornbein JA. **Early pregnancy metabolites predict gestational diabetes mellitus: implications for fetal programming**. *Am J Obstet Gynecol* (2021) **224** 215 e211-215 e217. DOI: 10.1016/j.ajog.2020.07.050
9. Alesi S, Ghelani D, Rassie K, Mousa A. **Metabolomic biomarkers in gestational diabetes mellitus: a review of the evidence**. *Int J Mol Sci* (2021) **22** 5512. DOI: 10.3390/ijms22115512
10. Zhu Y, Tsai MY, Sun Q, Hinkle SN, Rawal S, Mendola P, Ferrara A, Albert PS, Zhang C. **A prospective and longitudinal study of plasma phospholipid saturated fatty acid profile in relation to cardiometabolic biomarkers and the risk of gestational diabetes**. *Am J Clin Nutr* (2018) **107** 1017-1026. DOI: 10.1093/ajcn/nqy051
11. Rahman ML, Feng YA, Fiehn O, Albert PS, Tsai MY, Zhu Y, Wang X, Tekola-Ayele F, Liang L, Zhang C. **Plasma lipidomics profile in pregnancy and gestational diabetes risk: a prospective study in a multiracial/ethnic cohort**. *BMJ Open Diabetes Res Care* (2021) **9** e001551. DOI: 10.1136/bmjdrc-2020-001551
12. Chen X, Scholl TO, Leskiw M, Savaille J, Stein TP. **Differences in maternal circulating fatty acid composition and dietary fat intake in women with gestational diabetes mellitus or mild gestational hyperglycemia**. *Diabetes Care* (2010) **33** 2049-2054. DOI: 10.2337/dc10-0693
13. Zhu Y, Li M, Rahman ML, Hinkle SN, Wu J, Weir NL, Lin Y, Yang H, Tsai MY, Ferrara A. **Plasma phospholipid n-3 and n-6 polyunsaturated fatty acids in relation to cardiometabolic markers and gestational diabetes: a longitudinal study within the prospective NICHD Fetal Growth Studies**. *PLoS Med* (2019) **16** e1002910. DOI: 10.1371/journal.pmed.1002910
14. Lu W, Luo M, Fang X, Zhang R, Li S, Tang M, Yu X, Hu C. **Discovery of metabolic biomarkers for gestational diabetes mellitus in a Chinese population**. *Nutr Metab (Lond)* (2021) **18** 79. DOI: 10.1186/s12986-021-00606-8
15. Sikorski C, Azab S, de Souza RJ, Shanmuganathan M, Desai D, Teo K, Atkinson SA, Morrison K, Gupta M, Britz-McKibbin P. **Serum metabolomic signatures of gestational diabetes in South Asian and white European women**. *BMJ Open Diabetes Res Care* (2022) **10** e002733. DOI: 10.1136/bmjdrc-2021-002733
16. Zhang M, Yang H. **Perspectives from metabolomics in the early diagnosis and prognosis of gestational diabetes mellitus**. *Front Endocrinol (Lausanne)* (2022) **13** 967191. DOI: 10.3389/fendo.2022.967191
17. Loo EXL, Soh SE, Loy SL, Ng S, Tint MT, Chan SY, Huang JY, Yap F, Tan KH, Chern BSM. **Cohort profile: Singapore Preconception Study of Long-Term Maternal and Child Outcomes (S-PRESTO)**. *Eur J Epidemiol* (2021) **36** 129-142. DOI: 10.1007/s10654-020-00697-2
18. Metzger BE, Gabbe SG, Persson B, Buchanan TA, Catalano PA, Damm P, Dyer AR, Leiva A. **International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy**. *Diabetes Care* (2010) **33** 676-682. DOI: 10.2337/dc09-1848
19. Cui L, Lee YH, Thein TL, Fang J, Pang J, Ooi EE, Leo YS, Ong CN, Tannenbaum SR. **Serum metabolomics reveals serotonin as a predictor of severe dengue in the early phase of dengue fever**. *PLoS Negl Trop Dis* (2016) **10** e0004607. DOI: 10.1371/journal.pntd.0004607
20. Alshehry ZH, Barlow CK, Weir JM, Zhou Y, McConville MJ, Meikle PJ. **An efficient single phase method for the extraction of plasma lipids**. *Metabolites* (2015) **5** 389-403. DOI: 10.3390/metabo5020389
21. Bansal N. **Prediabetes diagnosis and treatment: a review**. *World J Diabetes* (2015) **6** 296-303. DOI: 10.4239/wjd.v6.i2.296
22. Wang XM, Gao Y, Eriksson JG, Chen W, Chong YS, Tan KH, Zhang C, Zhou L, Li LJ. **Metabolic signatures in the conversion from gestational diabetes mellitus to postpartum abnormal glucose metabolism: a pilot study in Asian women**. *Sci Rep* (2021) **11** 16435. DOI: 10.1038/s41598-021-95903-w
23. Lin Y, Wu J, Zhu Y, Hinkle SN, Rawal S, Liang L, Weir NL, Tsai MY, Zhang C. **A longitudinal study of plasma acylcarnitines throughout pregnancy and associations with risk of gestational diabetes mellitus**. *Clin Nutr* (2021) **40** 4863-4870. DOI: 10.1016/j.clnu.2021.07.008
24. Zhao L, Wang M, Li J, Bi Y, Li M, Yang J. **Association of circulating branched-chain amino acids with gestational diabetes mellitus: a meta-analysis**. *Int J Endocrinol Metab* (2019) **17** e85413. DOI: 10.5812/ijem.85413
25. Plows JF, Stanley JL, Baker PN, Reynolds CM, Vickers MH. **The pathophysiology of gestational diabetes mellitus**. *Int J Mol Sci* (2018) **19** 3342. DOI: 10.3390/ijms19113342
26. Chang W, Hatch GM, Wang Y, Yu F, Wang M. **The relationship between phospholipids and insulin resistance: from clinical to experimental studies**. *J Cell Mol Med* (2019) **23** 702-710. DOI: 10.1111/jcmm.13984
27. Funai K, Lodhi IJ, Spears LD, Yin L, Song H, Klein S, Semenkovich CF. **Skeletal muscle phospholipid metabolism regulates insulin sensitivity and contractile function**. *Diabetes* (2016) **65** 358-370. DOI: 10.2337/db15-0659
28. Newsom SA, Brozinick JT, Kiseljak-Vassiliades K, Strauss AN, Bacon SD, Kerege AA, Bui HH, Sanders P, Siddall P, Wei T. **Skeletal muscle phosphatidylcholine and phosphatidylethanolamine are related to insulin sensitivity and respond to acute exercise in humans**. *J Appl Physiol (1985)* (2016) **120** 1355-1363. DOI: 10.1152/japplphysiol.00664.2015
29. van der Veen JN, Lingrell S, da Silva RP, Jacobs RL, Vance DE. **The concentration of phosphatidylethanolamine in mitochondria can modulate ATP production and glucose metabolism in mice**. *Diabetes* (2014) **63** 2620-2630. DOI: 10.2337/db13-0993
30. Chang W, Chen L, Hatch GM. **Berberine treatment attenuates the palmitate-mediated inhibition of glucose uptake and consumption through increased 1,2,3-triacyl-sn-glycerol synthesis and accumulation in H9c2 cardiomyocytes**. *Biochim Biophys Acta* (2016) **1861** 352-362. DOI: 10.1016/j.bbalip.2015.12.017
31. Selathurai A, Kowalski GM, Burch ML, Sepulveda P, Risis S, Lee-Young RS, Lamon S, Meikle PJ, Genders AJ, McGee SL. **The CDP-ethanolamine pathway regulates skeletal muscle diacylglycerol content and mitochondrial biogenesis without altering insulin sensitivity**. *Cell Metab* (2015) **21** 718-730. DOI: 10.1016/j.cmet.2015.04.001
32. Zhan Y, Wang J, He X, Huang M, Yang X, He L, Qiu Y, Lou Y. **Plasma metabolites, especially lipid metabolites, are altered in pregnant women with gestational diabetes mellitus**. *Clin Chim Acta* (2021) **517** 139-148. DOI: 10.1016/j.cca.2021.02.023
33. Lai M, Liu Y, Ronnett GV, Wu A, Cox BJ, Dai FF, Rost HL, Gunderson EP, Wheeler MB. **Amino acid and lipid metabolism in post-gestational diabetes and progression to type 2 diabetes: A metabolic profiling study**. *PLoS Med* (2020) **17** e1003112. DOI: 10.1371/journal.pmed.1003112
34. Stephenson J, Heslehurst N, Hall J, Schoenaker D, Hutchinson J, Cade JE, Poston L, Barrett G, Crozier SR, Barker M. **Before the beginning: nutrition and lifestyle in the preconception period and its importance for future health**. *Lancet* (2018) **391** 1830-1841. DOI: 10.1016/S0140-6736(18)30311-8
35. Lim SX, Loy SL, Colega MT, Lai JS, Godfrey KM, Lee YS, Tan KH, Yap F, Shek LP, Chong YS. **Prepregnancy adherence to plant-based diet indices and exploratory dietary patterns in relation to fecundability**. *Am J Clin Nutr* (2022) **115** 559-569. DOI: 10.1093/ajcn/nqab344
|
---
title: 'Aligning policymaking in decentralized health systems: Evaluation of strategies
to prevent and control non-communicable diseases in Nigeria'
authors:
- Whenayon Simeon Ajisegiri
- Seye Abimbola
- Azeb Gebresilassie Tesema
- Olumuyiwa O. Odusanya
- Dike B. Ojji
- David Peiris
- Rohina Joshi
journal: PLOS Global Public Health
year: 2021
pmcid: PMC10022121
doi: 10.1371/journal.pgph.0000050
license: CC BY 4.0
---
# Aligning policymaking in decentralized health systems: Evaluation of strategies to prevent and control non-communicable diseases in Nigeria
## Abstract
Noncommunicable diseases (NCDs) are leading causes of death globally and in Nigeria they account for $29\%$ of total deaths. Nigeria’s health system is decentralized. Fragmentation in governance in federalised countries with decentralised health systems is a well-recognised challenge to coherent national health policymaking. The policy response to the rising NCD burden therefore requires strategic intent by national and sub-national governments. This study aimed to understand the implementation of NCD policies in Nigeria, the role of decentralisation of those policies, and to consider the implications for achieving national NCD targets. We conducted a policy analysis combined with key informant interviews to determine to what extent NCD policies and strategies align with Nigeria’s decentralised health system; and the structure and process within which implementation occurs across the various tiers of government. Four inter-related findings emerged: NCD national policies are ‘top down’ in focus and lack attention to decentralisation to subnational and frontline care delivery levels of the health system; there are defective coordination mechanisms for NCD programmes which are underpinned by weak regional organisational structures; financing for NCDs are administratively burdensome and fragmented; and frontline NCD service delivery for NCDs are not effectively being integrated with other essential PHC services. Despite considerable progress being made with development of national NCD policies, greater attention on their implementation at subnational levels is needed to achieve more effective service delivery and progress against national NCD targets. We recommend strengthening subnational coordination mechanisms, greater accountability frameworks, increased and more efficient funding, and greater attention to integrated PHC service delivery models. The use of an effective bottom-up approach, with consideration for decentralization, should also be engaged at all stages of policy formulation.
## National burden of non-communicable diseases
Noncommunicable diseases (NCDs) are leading causes of death globally with associated large economic, social and health impacts [1]. The burden of NCDs is highest in low-income and middle-income countries (LMIC) [2]. Most populations with limited access to services and conditions, such as adequate education and routine screening, that will enhance the prevention, early detection and prompt treatment of NCDs experience a disproportionate share of the disease burden compared to those with adequate access to essential services [1].
The current (year 2020) NCD progress monitor report reveals a rising NCD burden in Nigeria with 617,300 NCD related deaths, accounting for $29\%$ of total deaths, of which, $22\%$ occurred among those aged between 30–70 years (referred to as premature deaths) [3]. Cardiovascular diseases account for $11\%$ of these deaths, $4\%$ are due to cancers, $2\%$ are due to chronic respiratory diseases, $1\%$ diabetes and other NCDs account for the remaining $11\%$ [1]. The country’s NCD burden was generated using mortality estimated from 2016 WHO Global Health Estimates and the most recent United Nations Population Division World Population Prospects. The likelihood of dying between ages 30–70 years from the four main NCDs were calculated from age-specific death rates and proportional mortality for NCDs [1]. In addition to these four leading NCDs, sickle cell diseases (SCDs) are also significant NCDs in Nigeria. Nigeria is estimated to be the highest SCD burden globally [4] and contributes about $30\%$ of the global burden of children born with sickle cell anaemia annually [5].
## Nigeria’s health system and national policy response to NCDs
Nigeria has a three-tier government structure (federal, state, and local government), and consequently, the health system is decentralized. In practice, health system decentralization is “the transfer of authority and power from higher to lower levels of government or from national to subnational levels of government” [6]. This decentralized system places health on the concurrent legislative list [7, 8], and this implies that the health system operates with shared authority across each tier of government, [9] such that delivery, management, and financing of health services is the responsibility of all three tiers of government [10]. The constitution does not delineate the responsibilities of each tier of government with regards to health [8]. As each possesses a high level of autonomy, significant authority is exercised by each tier with regards to resource allocation and utilization [11]. The federal government is responsible for development of national health policies and issuing guidelines for their implementation at the state and local government level [11, 12]. Every state has an elected governor who is the head of the executive council, and a legislative body–the house of assembly. Local governments are managed by an elected executive chairperson along with legislative councillors from political wards. Each state has a Ministry of Health, and each local government has a department of health.
The private health sector plays a significant role in the health system. It constitutes about $30\%$ of the country’s health facilities across all levels of healthcare system and (along with ‘informal’ healthcare providers such as traditional medicine providers, patent and proprietary medicine vendors, drug shops and complementary and alternative health practitioners) delivers about $60\%$ of the country’s healthcare services [10].
Health system fragmentation in federalised countries with decentralised governance structures is a well-recognised risk to coherent national health policymaking [13]. Consequently, the NCD policy response in Nigeria requires strategic intent by all levels of government. Until 2020, the National Strategic Plan of Action on Prevention and Control of NCDs was the overarching policy document for NCDs prevention and control in Nigeria. First launched in 2013, it was updated in 2015 to span the period 2016–2020 [4]. It provided a framework for using a multisectoral approach to strengthen the health system for the prevention and control of NCDs. In 2019, the National Multi-Sectoral Action Plan for the Prevention and Control of Non-Communicable Diseases (2019–2025) was launched. This action plan supersedes the previous policy and is currently the main guiding document for a national, multi-sectoral response to NCDs [14].
Previous NCD policy analysis have evaluated the importance of a multisectoral approach and implementation of NCD ‘best buys’–well-evidenced interventions that are feasible, low-cost and appropriate to implement within the constraints of the local health system [15]. One such study analysed NCD policies across multiple stakeholder organizations in Nigeria. *It* generated evidence on the use of a multisectoral approach in formulating policies for NCD ‘best buys’ implementation as well as assessed its barriers and facilitators. Nigeria’s WHO membership, leading to government commitment to a series of resolutions, was found to be the most important facilitator, while over-dependence on donor funding, lower political priority and poor understanding of how to implement multisectoral plans were cited as barriers [16].
Studies that explored NCD risk factors found comprehensive tobacco related policies [17] and some alcohol-related policies [18]. However, both areas had weak multisectoral approaches, and some did not adhere to the principles of ‘best buys’. Multi-country studies that have analysed NCD prevention policies through a multisectoral lens found that the policies are influenced by several global and local factors such as political will, available resources and locally generated data. These studies established the existence of policy implementation gaps that require mechanisms to attain better policy outcomes with a particular focus on contextual factors such as political support and adequate resource allocation [19–22].
The extent to which Nigeria’s decentralized health system governance may be an important driver of NCD policy implementation has been inadequately appraised to date. In this study, we aimed to understand the implementation of NCD policies in Nigeria, and the implications of decentralisation for achieving national NCD targets.
## Material and methods
We conducted an analysis of NCD policies and combined this with interviews of key informants in the public sector, focussing on the structures and mechanisms by which the NCD policies are implemented.
## NCD policy analysis
We analyzed national NCD guidelines, policy and strategy documents by the Federal Ministry of Health over the period 2009–2019. This was supplemented by empirical studies and country reports on the implementation of NCD programmes to explore the context of the implementation of NCD policies and strategies in Nigeria. We retrieved publicly available NCDs policy documents developed by the ministry of health from government’s and World Health Organization’s websites. Key individuals were contacted through email or direct call and the Federal and State Ministries of Health were also visited to obtain other relevant documents that were not accessible online.
Using guidelines on decentralization [23], we examined how these policies aligned with the multiple dimensions of decentralization as it applies to both unitary and federal countries. The OECD guidelines on decentralization was developed multi-level governance studies series and applied to some countries. It outlines ten domains for decentralization that are necessary for local and regional development [23]. It also provides the rationale for each domain, suggested practical guidance, stated drawbacks to avoid, listed good practices and included a checklist for action. Five of the ten domains were chosen because they bear direct relevance to the aim of our study. The other five domains are beyond the scope of this study as they focus broadly on legislative and fiscal structures. The five domains considered were: [1] clear roles and responsibilities of different government levels; [2] sufficient funds for all responsibilities; [3] support subnational capacity building; [4] adequate coordination mechanisms among levels of government; and [5] accountability framework and performance monitoring system.
## Stakeholder perspectives on implementation of NCD policies
To understand the structure and process of implementation of NCD policies across the various level of the health system, qualitative data were collected from August 2019 to September 2019 and guided by the consolidated criteria for reporting qualitative research guidelines for qualitative research [24]. Interviews with key informant NCD stakeholders were conducted by the lead author (WSA), a male public health researcher who has worked with the Nigerian government at various level of the country’s health system. He was supported by two other data collectors who were trained to become familiar with the aims of the study, interview questions and the use of field notes. All recruited participants were interviewed face-to-face except for one who provided a written response. All interviews were audio recorded, conducted in locations conducive and appropriate for the participants’ privacy such as personal office space with only the researchers and participants present. Interview duration ranged from 30–60 minutes.
At the national level, we interviewed staff in the Departments of Public Health (NCD Division) and Hospital Services (Cancer Control Unit) on the structure and process of implementation of the overarching NCD policies. At the sub-national level, we interviewed staff in four states, two in each of the Southern and Northern regions. This is because each region has varying health indices profiles [11]. These states were selected on the basis of varying socio-economic profiles, health indices and similarity in health intervention programmes being implemented. Purposive sampling was used to select the policy actors based on their roles, relevance, or expertise in the NCD prevention policies and strategies. This was to ensure a maximum variation across all relevant units. We also took a ‘snowballing’ approach to identify additional respondents during interviews with the initial key informants. Interviews focussed on the structure, resources and mechanisms through which the Nigeria National Policy and Strategy on NCDs 2015–2020 was delivered [25] (see S1 File for the interview guide).
Interviews were transcribed, coded, and analyzed thematically. Themes were both derived from the data and also guided by the five domains from the OECD guidelines on decentralization described above [23].
Altogether, eleven policies, guidelines, and strategic plans for NCDs in Nigeria were examined and twenty-two key informant interviews conducted.
## Ethical considerations
Ethical approval was granted by the National Health Research Ethics Committee of Nigeria (Approval no: NHREC/$\frac{01}{01}$/2007) and the University of New South Wales Human Research Ethics Committee (HC: 190051). Informed written consent was obtained from all participants before conducting the interview. Anonymity and confidentiality of all respondents were maintained throughout the process. Participants names were also replaced with codes during data analysis (Table 1).
**Table 1**
| Participant’s Code | Sex | Organization/Location | Designation |
| --- | --- | --- | --- |
| KII 1 | Male | Federal Ministry of Health | Assistant Director |
| KII2 | Male | Federal Ministry of Health | Deputy Director |
| KII3 | Male | Federal Ministry of Health | Deputy Director |
| KII4 | Male | National Primary Health Care Development Agency | Chief Medical Officer |
| KII5 | Male | State Ministry of Health (North) | Senior Medical Officer |
| KII6 | Male | State Ministry of Health (North) | Director, Public Health/ NCD Coordinator |
| KII7 | Female | State Ministry of Health (South) | NCD Coordinator |
| KII8 | Female | State Ministry of Health (South) | NCD Coordinator |
| KII9 | Female | World Health Organization | Consultant, NCD Unit |
| KII10 | Male | SPHCDA/PHC (North) | Head of Facility (Community Health Extension Worker) |
| KII11 | Female | SPHCDA/PHC (North) | Head of Facility (Chief Nursing Officer) |
| KII12 | Female | SPHCDA/PHC (North) | Head of Facility (Chief Nursing Officer) |
| KII13 | Female | SPHCDA/PHC (North) | Head of Facility (Community Health Extension Worker) |
| KII14 | Female | SPHCDA/PHC (North) | Head of Facility (Community Health Extension Worker) |
| KII15 | Female | SPHCDA/PHC (North) | Head of Facility (Community Health Extension Worker) |
| KII16 | Female | SPHCDA/PHC (North) | Head of Facility (Community Health Extension Worker) |
| KII17 | Female | SPHCDA/PHC (South) | Head of Facility (Chief Nursing Officer) |
| KII18 | Female | SPHCDA/PHC (South) | Head of Facility (Chief Nursing Officer) |
| KII19 | Female | SPHCDA/PHC (South) | Head of Facility (Chief Nursing Officer) |
| KII20 | Female | SPHCDA/PHC (South) | Head of Facility (Chief Nursing Officer) |
| KII21 | Female | SPHCDA/PHC (South) | Head of Facility (Chief Nursing Officer) |
| KII22 | Female | SPHCDA/PHC (South) | Head of Facility (Chief Nursing Officer) |
## Results
The finding from this study are broadly divided into sections A and B. Section A revealed the findings from NCD documents analysis (Table 2) while Section B presents the four themes (I–IV) that emerged from the interviews (Table 3).
The themes are:
## A. NCD policy documents and decentralization
Eleven national documents were reviewed—four on cancers, one on sickle cell disease, one on tobacco control, two on diet related NCDs, and two on multiple NCDs. There were no government-led national guidelines on the management of hypertension, diabetes, and respiratory diseases. ( Table 2) Although there are some clinical guidelines developed by national health professional associations for diabetes and hypertension, these were not developed under the auspices of any national government body. The two key findings from the document analysis and interviews was that there was inadequate consideration for decentralisation and delayed implementation of the National Strategic Plan.
## I. Inadequate consideration for decentralization in NCD policies and strategy documents
Five of the seven policy and strategic documents, such as the National Strategic Plan of Action on Prevention and Control of Non-Communicable Diseases [2015] have clearly assigned roles for each of the levels of government. However, there was evidence of overlapping roles and responsibilities between levels of government in some of the documents.
Five documents discussed or proposed a source of funding for the activities/roles assigned to the various levels of government. Most of these were dependent on budgetary allocation by each level of government for service provision and management of policies. All strategic documents proposed capacity building at various levels. Some discussed support for subnational capacity building by the federal government.
Four documents mentioned coordination mechanisms. Among these, only the Health Sector Component of National Food and Nutrition Policy National Strategic Plan of Action for Nutrition (2014–2019) discussed vertical coordination mechanisms among all levels of government while others were mainly directed toward stakeholders at national levels. With regards to accountability, four documents propose a monitoring and evaluation framework (including indicators) as a proxy for accountability, however, no other accountability framework were mentioned in any of the other documents. Similarly, no form of sanction, discipline or performance reward system was mentioned.
## I. Slow implementation of the National Strategic Plan of Action on Prevention and Control of NCDs
The overarching NCD policy and strategic document has been operational for the last four years and is due for review in the year 2020. According to a federal level participant mentioned looking forward to this review: “…the policy is supposed to be reviewed this year, so we are waiting for WHO, to support that. We are going to look at it in line with the current trend, so we can see if *Nigeria is* on the right path or not” (KII2). Despite this national document, there was little evidence of sub-national implementation. The non-implementation of the policy documents could possibly be due to lack of awareness or weak coordination mechanisms as perceived by state level participants. “ Unfortunately, again, the policy document was drafted before we came in, so I personally was not involved, and I don’t think my predecessors were involved in crafting that” (KII6). Over the period of four years, there appears to be a disconnect with regards to the policy documents between national and sub-national levels as one state level participant mentioned: “The guideline is still a draft and yet to be disseminated to the States for full adherence and implementation…our state was not involved in its formulation” (KII7) and another said: “I have been in this department for several years and I am not aware of the existence of the document…” (KII8)
## II. Poor political will (macro-level view) and inadequate resources for NCD programmes and policy implementation
Political will was considered to be important for NCD programme implementation. A national level respondent believed that the chronic nature of NCDs and lack of proper understanding of NCD burden could underpin low levels of political engagement in NCDs: “The political commitment may be lacking; I often combine political commitment with action…. You can also add that the lack of understanding by policy makers, for them to see the need that these issues should be given high priority …..they are indifferent if it comes to the issue of support for NCDs…they are chronic in nature, but they don’t seem to see it in that sense to act on time” (KII1) The lack of political will can be seen across every level of government and appears to relate to capacity to deliver. Although this affects different aspect of health sectors, NCDs may be one of the worst hit as mentioned by one of the participants: “The allocation to health at both the national and state level is inadequate, the political will is lacking in the area of investment in health, and it impacts negatively on whatever plans and interventions you may have …there are other sectors that also need attention, and all these things affect implementation of policies and programs. …. And as you all are aware, NCDs are usually neglected, and the interest in it… is quite poor. ( KII2) Inadequate funding was also considered a barrier to NCD programme and policy implementation across all tiers of government. This is also associated with budgeting most of the funds for infectious diseases and limited donor support for NCDs as mentioned by one national level participant: “funding challenges could be another, and that transcends from the federal level to the state to the local level government, so funding has been a serious issue and it is quite sad that most activities are driven by donor support. And this funding is traditionally marked to fight infectious diseases. ( KII1) Due to lack of state-wide data to demonstrate the burden of NCDs, the government and donors seem to not be convinced that NCDs are a problem and consequently funding of NCD programmes is not prioritized. This concern was expressed by another state level participant: “…when we do find out what the problem is or the bottom (extent) of the problem and we plan a strategy for the implementation, we will have the challenge of funding from government, and because we have not done anything yet, even attracting partners to support us is also a challenge” (KII5,)
## III. Weak governance structure and defective coordination mechanisms for NCD programme and policy implementation
The Division of NCDs is domiciled within the Department of Public Health, Federal Ministry of Health, and a counterpart NCD unit/division also exists at state levels. However, the vertical support and coordination between these two was considered weak. At the state level, the State Ministry of Health (SMOH) is in charge of NCD activities at the state level while the operations of the PHCs are under the State Primary Health Care Development Agency/Board. While this fragmented structure appears to currently work for maternal and child health, it may not be the case for NCDs—in the words of one participant (Federal level) “…for essential basic support, like things that have to do with maternal and child health, the Federal Government has a mechanism of deploying such services down to the primary health care level, through the National Primary Health Development Agency, but unfortunately this agency is not in charge of the essential services for non-communicable diseases … for non-communicable disease there is a gap” (KII1) There has been some albeit slow progress in creating stale-level structures for NCDs–as described by a participant (national level): “… last year we prepared a council memo from the Federal Ministry of Health, and took it to the national council on health…we wanted states to establish an NCD desk in their States Ministries of Health (NCH), (because) we didn’t have anything going on at the state level. But now I can tell you that every state now has an NCD desk, because it was the decision of the national council on health. ( KII2) *Despite this* progress, there have been limitations as to how well these structures currently function, as described by a participant (national level): “…we are yet to fully engage the NCD focal points at the state level (SMOH), …it is not enough to have a functional desk, you need to also engage them. So, we are hoping [for] that this year. We’ve been able to secure it in the 2019 budget.” ( KII1) The accountability framework and performance monitoring system for NCD programme appears to be weak. The lack of any penalty or consequence at the sub-national level for non-performance or non-implementation of NCD policies was mentioned by a federal level participant: “it is difficult to punish a state that refuses or defers implementation, we cannot hold that state responsible, we cannot say that there are punitive measures [even] if we narrow it to non-communicable diseases…” (KII1). One of the reasons attributed to this was the lack of funding of sub-national levels for NCD activities by the national government according to another national level participant: “…if we are funding, then if they do not comply we can withdraw funding. Rather there is hardly any consequence, in a federation kind of system that we have…everybody has their kind of independence. So, it’s more like a begging system” (KII4). While there seems to be a coordination or multisectoral approach with regards to NCD activities and policy implementation, there was little evidence of vertical coordination between what happens at the national and sub-national levels. According to one national level participant: “…we can say we have established a defined federal non-communicable disease coordination mechanism at the federal level, and that mechanism is working but there is a disconnect between the federal mechanism and the state mechanism.” ( KII1)
## IV. Limited alignment of NCDs with the delivery and reporting of other PHC services
Although Primary Health Care workers (mostly non-physicians) deliver various forms of NCD services, the majority have never received in-service training on NCD management. This contrasts with the provision of supportive supervision and capacity building for infectious diseases, maternal, child and reproductive health services: “There’s none (training for NCDs). It’s only for immunization we’ve been getting…and one on resuscitation of babies… apart from that anything on hypertension, diabetes, there is no training on that. ( KII17).”.
PHC staff are also responsible for data collection on NCDs as a part of Nigeria’s IDSR (Integrated Diseases Surveillance and Response) guidelines. These data are transmitted to the national level (Nigeria Centre for Disease Control) through the State Ministry of Health. At the national level, the surveillance office is domiciled within the Nigeria Centre for Disease Control (NCDC), an agency under the Federal Ministry of Health. However, because the NCDC’s mandate is only for infectious diseases, there is minimal use of NCD data for improving PHC performance management or for capacity building of health workers. Consequently, PHC workers only get feedback on communicable diseases. This seems to have influenced the subnational level HCWs as a national level respondent said: “So when they are doing the monthly reporting, they tend to report better the component for the infectious diseases, than the component for the non-infectious diseases. The report is inconsistent, it is haphazard and sometimes they do not even report it, because they believe that even if they report it no action will come out of it. So, the IDSR reporting for infectious diseases is better compared to the one that has non-communicable diseases…..” (KII2)
## Discussion
Four inter-related findings emerged from the study: [1] current NCD national policies are evolving and provide minimal consideration for effective decentralisation to regional and frontline care delivery levels of the health system; [2] current financing for NCDs is limited, administratively burdensome and fragmented; [3] regional organisational structures are weak leading to defective coordination mechanisms for NCD programmes; and [4] frontline service delivery for NCDs is not being effectively aligned alongside other essential PHC services. We discuss each of these in more detail below.
## Evolving NCD policies with limited consideration for decentralization
All policy documents examined primarily originated from the national level of government, but sub-national governments are chiefly responsible for the implementation. Mapping these documents against domains of decentralization suggests they are insufficiently developed to support sub-national implementation.
Lack of clearly assigned roles for each of the levels of government is known to be associated with inefficient service provision and may result into failure to effectively address critical needs [23] such as the rising burden of NCDs. Findings from this study is similar to the report of the National Health Policy which revealed that the national constitution and the 2014 National Health Act has failed to address the clear roles and responsibilities of each tier of government towards health [8, 12]. Greater role clarity and articulation of shared responsibilities for NCD prevention and control could ensure that duplication is avoided, and accountability for implementation is enhanced. It is therefore important that more attention be paid, in these policies, to adoption, scale up, and quality implementation at subnational levels.
This needs to be accompanied by adequate funding for the activities and roles assigned to the various level of governments accompanied by effective coordination mechanisms (discussed below). Well designed and implemented decentralization policies could deliver multiple benefits including enhanced frontline service delivery for NCD programmes, efficient resource allocation and ultimately a positive impact on health indices [23].
## Limited political will and inadequate financial resources for NCD programmes and policy implementation
Successful implementation of policies and programmes requires strong political will. For translation of intent to action, political will must be also be accompanied by political capacity [26]. This implies the creation of an enabling governance environment and structure to drive the process. Such enabling socio-political and bureaucratic environments can lead to increased availability and accessibility to necessary human and financial resources [26]. The lack of recent, reliable state-wide and national data on the burden of NCDs could reflect the lack of such political will and capacity. While there are regular surveys on infectious diseases such as HIV/AIDS and tuberculosis, the most recent national NCD survey is almost three decades old [14]. It is therefore important that contemporary and robust data are generated to advocate for increased political commitment to support NCD policy implementation.
Nigeria can draw lessons from the role that enhanced political will and capacity played in the elimination of poliomyelitis. During the era of polio scourge in Nigeria, all Presidents were outspoken in their commitment to elimination of the virus. A presidential taskforce was formed to directly supervise and coordinate national campaigns, and to monitor progress of each states to ensure accountability, including sanctions for poor performances [27]. National level campaigns were launched alongside community level engagement that included religious leaders, community leaders, opinion leaders and local government chairmen [28]. State Governors were also required to sign a commitment to polio eradication, provide additional funding for the implementation and report regularly to federal government. The presidential task force also tracked progress at local government levels and made all data publicly accessible [27]. This high level of political will and capacity across all governance levels was a major factor in polio eradication and the approaches taken to optimise Nigeria’s decentralized health system are prescient for addressing NCDs.
Though political commitment is a necessary condition for adapting and implementing policies and strategies, it is not sufficient in itself [29]. It needs to be accompanied by other factors including robust governance structures and adequate resource allocation. Currently, the national government provides minimal financial support to sub-national level governments (especially for NCD programmes). This combined with inadequate locally generated revenues, inevitably leads to NCDs being placed lower on the policy agenda at the subnational level [30]. There is a pressing need for financing reforms to foster the appropriate environment for effective NCD policy implementation. First, the main revenue source currently for NCD programmes is via the annual budget appropriation. While this is commendable, the bureaucracy involved results in a minority of the appropriated budget being disbursed downstream to those responsible for implementing these policies. This complex bureaucratic process associated with NCD budgets also reflected Botswana’s experience even in the face of political commitment [31]. Second, although multisectoral approaches to improve NCD programme implementation are essential, there needs to be a less bureaucratic processes for leveraging resource contributions and allocations from different government departments. Third, there is limited consideration given to NCDs in the National Health Insurance Schemes (NHIS) and the Basic Health Care Provision Funds (BHCPF) [14]. Fifty per cent of the BHCPF is disbursed through NHIS for a basic minimum package of health services (BMPHS). Of the nine BMHPS interventions, only one relates to NCDs (urinalysis and blood pressure check to screen for diabetes and hypertension) with the remaining interventions dedicated to infectious diseases, maternal and child health [32]. Fourth, NCD programmes attract little international donor support relative to infectious diseases and maternal and child health programmes. This places further constraints on the fiscal space needed to expand NCD policy implementation, especially at the sub-national level [33]. The persistence of vertical program funding and the lack of a health system strengthening approach from international donors are major barriers to addressing NCD prevention and care. [ 34, 35].
In order to reduce the burden of NCDs, make progress in achieving national targets as well as reduced out-of-pocket spending associated with NCDs care, Nigeria needs to increase and prioritize funding for NCDs through multiple sources and at all levels of care [36]. Increasing the current national budget allocation to health from $3\%$ to $15\%$ according to the Abuja Declaration [10] could generate increased funding of NCD programme. Like Uganda, Nigeria can also develop a costed NCD strategy using locally generated data to determine the country’s scope of NCD services. Uganda has quarantined funds to NCDs programmes annually and this is increased proportionally when there is a need for implementation of special programmes [37].
## Weak regional structures and defective coordination mechanisms
The setting and structure in which a policy is delivered influences both implementation and outcomes [25]. The presence of many actors at different governing levels of the system make coordination very challenging as some of these structures may have intersecting or competing roles in the delivery of NCD services [34]. Improved coordination mechanisms have potential to harmonize the engagement and activities of all relevant stakeholders [38].
The findings from the study are not unique to Nigeria. Previous studies show that in sub-Saharan Africa, the implementation of NCD prevention and control programmes is poorly coordinated [39] and often relies on non-governmental organizations to compensate for weak governance structures [21]. In Ghana, the coordination of NCD programmes between the policy (national) and service delivery (subnational) arms of the health sector was described as poor [40]. Achieving a successful sub-national implementation plan and efficient use of available resources in Nigeria will therefore depends largely on improved coordination between national and subnational levels [41]. Additionally, enforceable and practical accountability frameworks such as sanctioning of non-performance states and rewarding performance should characterise policy design and implementation [30].
For effective implementation of disease programmes such as NCDs, where ‘last mile’ connectivity for service delivery is required within the community, these coordination mechanisms needs to include the lowest level governance units [42]. Leveraging existing coordination structures is one of the most expedient way to achieve this, reducing the time, effort and costs needed to establish new structures and processes. Thailand implemented an effective coordination structure for coordinating alcohol and tobacco programmes and this has since been expanded to accommodate other NCDs [43]. Cambodia [44] and Kenya [45] also leveraged on existing HIV platform for care and coordination of NCD care delivery on a pilot scale and achieved a successful outcome. While the task of scaling this up on a larger scale may be complex, there are no doubts about its potential benefits.
## Integrated PHC service delivery and the way forward
The primary health facility, the lowest and most important level of the formal health care system for health programme implementation, is worst hit by the effect of the poor political will and inadequate resources for NCD programmes [10]. Despite constrained human resources, infectious disease and maternal and child health services are far better supported than NCD services. These programmes are better coordinated, augmented by task shifting policies (e.g. the midwives service schemes [46]), attract numerous donors supports for community outreach [10]. This starkly contrasts with the situation for NCDs. PHC facilities have limited management guidelines and minimal accountability frameworks for NCDs.
There is need to build the capacity of PHC staff, most of which have not received any form of in-service training for NCD management and prevention but are currently providing various forms of NCD services. In sub-Saharan Africa, only around one third of countries reported having trained PHC health workers for NCD management or have a national strategy with such plan [47]. It is therefore important that regular in-service training for PHC staff should be prioritized for successful integration of NCD care into routine service delivery. This will also ensure achievement of the planned task-shifting between the primary health care team [48] and equip them for effective service delivery [49]. Nigeria can learn from models such as those implemented by eSwatini to support health workers to achieve long term goals of NCD care implementation in decentralized health systems [50]. More so, all NCD services, including data management of the DHIS2 and IDSR, need to be done in a coordinated and integrated approach along with other essential services at the PHCs.
PHC capacity to integrate NCD services into frontline care delivery is a challenge in most countries in the African region. A recent study found that no African country met all the recommended indicators for integrating NCDs services into PHC [47]. To effectively align NCD prevention and control strategies with the country’s decentralized system, similar service delivery strategies deployed for chronic infectious diseases such as HIV and poliomyelitis should be considered for NCD integration. This may include pooling human resources, technical and financial support in conjunction with sub-national advocacy efforts to drive NCD programme implementation with a focus on achieving national targets. NCD-HIV service integration in rural Malawi has demonstrated high patient retention rates and statistically significant improvements in clinical outcomes for patients with NCDs [51].
## Study limitations
The current study was conducted at the national level, four states and a few selected local government areas. The findings for these states and local governments are therefore not generalizable as each sub-national entity has varying socio-economic and political profiles. However, the findings are transferable, when contextualised to the local circumstances in each state and local government area. More so, issues such as budgetary allocation and accessible funds for NCDs were self-reported and was not verified by the research teams as they were not publicly available. Other limitations of our study are that, the study focused mainly on the public sector of the health system and did not also include the non-health sector stakeholders. To address some of these potential limitations, we selected two states each from the Northern and Southern parts of the country and recruited relevant key participants to ensure robust data collection. However, future studies should not only explore the structures and processes within which policy implementation occurs (the focus of our study) but also conduct a detailed evaluation of the actual implementation of these strategies. This should be extended to the private sector of the health system as it plays a central role in the delivery of PHC services in the country. It is important that future research explore how the non-health ministries, departments and public agencies approach decentralization and its impact on multisectoral collaboration with the health sector for NCDs prevention and control.
## Conclusion
Despite considerable progress with regards to national NCD policies in Nigeria, this study found much work is needed particular at the state and local government levels to support their implementation. This may be achieved by ensuring national policy documents actively consider Nigeria’s decentralized health system in their formulation and accompanying implementation plans. Engaging a bottom-up approach at all stages of policy formulation is key to achieving this. Substantial institution strengthening, particularly at the subnational and service delivery levels, is needed to support achieving national NCD targets. This requires strengthened regional coordination mechanisms, enhanced accountability frameworks, increased and more efficient use of funding which in turn requires enhanced political will, and greater attention to integrated PHC service delivery models.
## References
1. 1World Health Organization. Non-communicable Diseases Country Profiles 2018 [Internet]. Geneva: World Health Organization; 2018 [cited 2020 Jun 9] p. 224. Available from: https://www.who.int/nmh/publications/ncd-profiles-2018/en/.. *Non-communicable Diseases Country Profiles 2018* (2018.0) 224
2. Olawuyi A, Adeoye I. **The prevalence and associated factors of non-communicable disease risk factors among civil servants in Ibadan, Nigeria.**. *PLOS ONE* (2018.0) **13** e0203587. DOI: 10.1371/journal.pone.0203587
3. 3World Health Organization. Noncommunicable Diseases Progress Monitor, 2020. [Internet]. Geneva; 2020 [cited 2020 Jun 9] p. 236. Available from: https://www.who.int/publications/i/item/ncd-progress-monitor-2020.
4. 4Federal Ministry of Health, Nigeria. National Strategic Plan of Action on Prevention and Control of Non-Communicable Diseases.
Federal Ministry of Health Nigeria; 2015.. *National Strategic Plan of Action on Prevention and Control of Non-Communicable Diseases.* (2015.0)
5. Piel F, Hay S, Gupta S, Weatherall D, Williams T. **Global Burden of Sickle Cell Anaemia in Children under Five, 2010–2050: Modelling Based on Demographics, Excess Mortality, and Interventions.**. *PLoS Med*
6. 6World Health Organization. Health Laws and Universal Health Coverage: Decentralization [Internet]. WHO. World Health Organization; 2002 [cited 2020 Sep 21]. Available from: http://www.who.int/health-laws/topics/governance-decentralisation/en/.. *Health Laws and Universal Health Coverage: Decentralization* (2002.0)
7. 7Federal Government of Nigeria. Constitution of the Federal Republic of Nigeria 1999. 1999.
8. 8Federal Ministry of Health. National Health Policy 2016. 2016.
9. Eboreime E, Abimbola S, Obi F, Ebirim O, Olubajo O, Eyles J. **Evaluating the sub-national fidelity of national Initiatives in decentralized health systems: Integrated Primary Health Care Governance in Nigeria.**. *BMC Health Serv Res* (2017.0). DOI: 10.1186/s12913-017-2179-2
10. 10Federal Ministry of Health, Nigeria. Second National Strategic Health Development Plan 2018–2022 [Internet]. Abuja, Nigeria: Federal Ministry of Health Nigeria; 2018 [cited 2020 Jun 25]. 150 p. Available from: https://health.gov.ng/doc/NSHDP_II_ME_Plan.pdf.
11. 11World Health Organization. Primary Health Care Systems (PRIMASYS): Case study from Nigeria. [Internet]. Geneva: World Health Organization; 2017 [cited 2020 Jul 15] p. 44. Available from: https://www.who.int/alliance-hpsr/projects/alliancehpsr_nigeriaprimasys.pdf?ua=1.. *Primary Health Care Systems (PRIMASYS): Case study from Nigeria.* (2017.0) 44
12. 12Federal Government of Nigeria. National Health Act, 2014. 2014.
13. Abimbola S, Baatiema L, Bigdeli M. **The impacts of decentralization on health system equity, efficiency and resilience: a realist synthesis of the evidence.**. *Health Policy Plan* (2019.0) **34** 605-17. DOI: 10.1093/heapol/czz055
14. 14Federal Ministry of Health, Nigeria. National Multi-Sectoral Action Plan for the Prevention and Control of Non-communicable Diseases (2019–2025) [Internet]. Abuja, Nigeria: Federal Ministry of Health Nigeria; 2019 [cited 2020 Jun 17]. Available from: https://www.health.gov.ng/doc/NCDs_Multisectoral_Action_Plan.pdf.. *National Multi-Sectoral Action Plan for the Prevention and Control of Non-communicable Diseases (2019–2025)* (2019.0)
15. 15World Health Organization. From burden to “best buys”: Reducing the economic impact of NCDs in low- and middle-income countries [Internet]. WHO. World Health Organization; 2011 [cited 2020 Nov 25]. Available from: https://www.who.int/nmh/publications/best_buys_summary.pdf.. *From burden to “best buys”: Reducing the economic impact of NCDs in low- and middle-income countries* (2011.0)
16. Oladepo O, Oluwasanu M, Abiona O. *Analysis of Non-communicable Diseases Prevention Policies in Nigeria.* (2017.0) 100
17. Oladepo O, Oluwasanu M, Abiona O. **Analysis of tobacco control policies in Nigeria: historical development and application of multi-sectoral action.**. *BMC Public Health* (2018.0). DOI: 10.1186/s12889-018-5831-9
18. Abiona O, Oluwasanu M, Oladepo O. **Analysis of alcohol policy in Nigeria: multi-sectoral action and the integration of the WHO “best-buy” interventions.**. *BMC Public Health* (2019.0) **19** 810. DOI: 10.1186/s12889-019-7139-9
19. Juma P, Mohamed S, Wisdom J, Kyobutungi C, Oti S. **Analysis of Non-communicable disease prevention policies in five Sub-Saharan African countries: Study protocol.**. *Arch Public Health* (2016.0). DOI: 10.1186/s13690-016-0137-9
20. Juma K, Juma P, Mohamed S, Owuor J, Wanyoike A, Mulabi D. **First Africa non-communicable disease research conference 2017: sharing evidence and identifying research priorities.**. *Journal of Global Health* **8** 020301. DOI: 10.7189/jogh.09.010201
21. Juma P, Mapa-tassou C, Mohamed S, Matanje-Mwagomba B, Ndinda C, Oluwasanu M. **Multi-sectoral action in non-communicable disease prevention policy development in five African countries**. *BMC Public Health* (2018.0) **18** 953. DOI: 10.1186/s12889-018-5826-6
22. Juma P, Mohamed S, Matanje-Mwagomba B, Ndinda C, Mapa-tassou C, Oluwasanu M. **Non-communicable disease prevention policy process in five African countries.**. *BMC Public Health* (2018.0) **18**
23. 23OECD. Making decentralisation work: A handbook for policy-makers. 2019 Mar 19 [cited 2020 Sep 21];135–87. Available from: https://www-oecd-ilibrary-org.wwwproxy1.library.unsw.edu.au/urban-rural-and-regional-development/making-decentralisation-work_dd49116c-en.
24. Tong A, Sainsbury P, Craig J. **Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups.**. *Int J Qual Health Care.* (2007.0) **19** 349-57. DOI: 10.1093/intqhc/mzm042
25. Moore G, Audrey S, Barker M, Bond L, Bonell C, Hardeman W. **Process evaluation of complex interventions: Medical Research Council guidance**. *BMJ* (2015.0) 350. DOI: 10.1136/bmj.h1258
26. 26Brinkerhoff D. Unpacking the concept of political will to confront corruption [Internet]. U4 Partner Agencies; 2010 [cited 2021 Mar 10]. Available from: https://www.cmi.no/publications/file/3699-unpacking-the-concept-of-political-will-to.pdf.
27. Cooke J, Tahir F. *Polio in Nigeria—The Race to Eradication* (2012.0) 22
28. Cochi S, Hegg L, Kaur A, Pandak C, Jafari H. **The Global Polio Eradication Initiative: Progress, Lessons Learned, And Polio Legacy Transition Planning.**. *Health Affairs* **35** 277-83. PMID: 26858381
29. Tesema A, Joshi R, Abimbola S, Ajisegiri WS, Narasimhan P, Peiris D. **Addressing barriers to primary health-care services for noncommunicable diseases in the African Region**. *Bull World Health Organ* (2020.0) **98** 906-8. DOI: 10.2471/BLT.20.271239
30. Ajisegiri W, Tesema A, Joshi R, Peiris D, Odusanya O, Abimbola S. **Is decentralisation effective for prevention and control of NCDs in Nigeria**. *African Health Journal* (2020.0) 16-8
31. Tapela N, Tshisimogo G, Shatera B, Letsatsi V, Gaborone M, Madidimalo T. **Integrating noncommunicable disease services into primary health care, Botswana**. *Bulletin of the World Health Organization* (2019.0) **97** 142-53. DOI: 10.2471/BLT.18.221424
32. 32Federal Ministry of Health, National Health Insurance Scheme and, the National Primary Health Care Development Agency. Harmonized Guidelines for the Administration, Disbursement, Monitoring and Fund Management of the Basic Healthcare Provision Fund [Internet]. 2018 [cited 2020 Nov 25]. Available from: https://www.health.gov.ng/doc/OM%20of%20BHCPF.pdf.
33. Nugent R, Feigl A. *Where Have All the Donors Gone? Scarce Donor Funding for Non-Communicable Diseases* (2010.0) 38
34. 34World Health Organization. WHO global coordination mechanism on the prevention and control of noncommunicable diseases: final report: WHO GCM/NCD working group on the alignment of international cooperation with national NCD plans (Working group 3.2, 2016–2017). 2018 [cited 2021 Mar 10]; Available from: https://apps.who.int/iris/handle/10665/312273.
35. Alliance NCD. *Protecting Everyone: Integration of Noncommunicable Diseases into Universal Health Coverage in the era of COVID-19* (2020.0) 29
36. Allen L. **Financing national non-communicable disease responses**. *Glob Health Action* (2017.0) **10**. DOI: 10.1080/16549716.2017.1326687
37. Schwartz J, Guwatudde D, Nugent R, Kiiza C. **Looking at non-communicable diseases in Uganda through a local lens: an analysis using locally derived data.**. *Globalization and Health* (2014.0) **10** 77. DOI: 10.1186/s12992-014-0077-5
38. Genay-Diliautas S, Malgouires C. **NCDs, time to deliver the role of the WHO Global Coordination Mechanisme on NCDs.**. *Med Sante Trop.* (2018.0) **28** 368-70. DOI: 10.1684/mst.2018.0836
39. Juma K, Juma P, Mohamed S, Owuor J, Wanyoike A, Mulabi D. **Participants for the first Africa NCD research conference 2017 in Nairobi, Kenya. First Africa non-communicable disease research conference 2017: sharing evidence and identifying research priorities.**. *J Glob Health.* (2019.0) **8** 020301. DOI: 10.7189/jogh.09.010201
40. Nyaaba G, Stronks K, Masana L, Larrea- Killinger C, Agyemang C. **Implementing a national non-communicable disease policy in sub-Saharan Africa: Experiences of key stakeholders in Ghana.**. *Health Policy OPEN* (2020.0) **1** 100009
41. Rottingen J, Ottersen T, Ablo A, Arhin-Tenkorang D, Benn C, Elovainio R. *Shared responsibilities for health: a coherent global framework for health financing; final report of the Centre on Global Health Security Working Group on Health Financing* (2014.0)
42. 42World Health Organization. Approaches to Establishing Country-Level Multisectoral Coordination Mechanisms for the Prevention and Control of Noncommunicable Diseases. 2015 p. 31.
43. 43World Health Organization. Multisectoral coordination mechanisms and responses to noncommunicable diseases in South-East Asia: Where are we in 2018? [Internet]. Geneva; 2019 [cited 2021 May 19] p. 123. Available from: https://www.ecoi.net/en/file/local/2018073/B5455.pdf.
44. Janssens B, Van Damme W, Raleigh B, Gupta J, Khem S, Soy Ty K. **Offering integrated care for HIV/AIDS, diabetes and hypertension within chronic disease clinics in Cambodia**. *Bull World Health Organ* (2007.0) **85** 880-5. DOI: 10.2471/blt.06.036574
45. Lamptey P, Dirks R. **Building on the AIDS Response to Tackle Noncommunicable Disease.**. *Global Heart* (2012.0) **7** 67-71. DOI: 10.1016/j.gheart.2012.01.010
46. Abimbola S, Okoli U, Olubajo O, Abdullahi MJ, Pate MA. **The Midwives Service Scheme in Nigeria.**. *PLOS Medicine* (2012.0) **9** e1001211. DOI: 10.1371/journal.pmed.1001211
47. Tesema A, Ajisegiri W, Abimbola S, Balane C, Kengne AP, Shiferaw F. **How well are non-communicable disease services being integrated into primary health care in Africa: A review of progress against World Health Organization’s African regional targets.**. *PLOS ONE* (2020.0) **15** e0240984. DOI: 10.1371/journal.pone.0240984
48. 48Federal Ministry of Health. Task-Shifting and Task-Sharing Policy for Essential Health Care Services in Nigeria. 2014.
49. Joshi R, Thrift A, Smith C, Praveen D, Vedanthan R, Gyamfi J. **Task-shifting for cardiovascular risk factor management: lessons from the Global Alliance for Chronic Diseases**. *BMJ Global Health* (2018.0) **3** e001092. DOI: 10.1136/bmjgh-2018-001092
50. Greve M, Lichtenberg S, Diederich S, Brendel A. *Supporting Non-Communicable Disease Prevention through a mHealth Application in Decentralized Healthcare Systems: Action Design Research in Eswatini* (2020.0)
51. Wroe E, Kalanga N, Dunbar E, Nazimera L, Price N, Shah A. **Expanding access to non-communicable disease care in rural Malawi: outcomes from a retrospective cohort in an integrated NCD–HIV model**. *BMJ Open* (2020.0) **10** e036836. DOI: 10.1136/bmjopen-2020-036836
|
---
title: 'Nexus between constructs of social cognitive theory model and diabetes self-management
among Ghanaian diabetic patients: A mediation modelling approach'
authors:
- Yaa Obirikorang
- Emmanuel Acheampong
- Enoch Odame Anto
- Ebenezer Afrifa-Yamoah
- Eric Adua
- John Taylor
- Linda Ahenkorah Fondjo
- Sylvester Yao Lokpo
- Evans Asamoah Adu
- Bernard Adutwum
- Enoch Ofori Antwi
- Emmanuella Nsenbah Acheampong
- Michael Adu Gyamfi
- Freeman Aidoo
- Eddie-Williams Owiredu
- Christian Obirikorang
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022127
doi: 10.1371/journal.pgph.0000736
license: CC BY 4.0
---
# Nexus between constructs of social cognitive theory model and diabetes self-management among Ghanaian diabetic patients: A mediation modelling approach
## Abstract
The promotion of Diabetes Self-Management (DSM) practices, education, and support is vital to improving the care and wellbeing of diabetic patients. Identifying factors that affect DSM behaviours may be useful to promote healthy living among these patients. The study assessed the determinants of DSM practices among Type 2 diabetes mellitus (T2DM) patients using a model-based social cognitive theory (SCT). This cross-sectional study comprised 420 (T2DM) patients who visited the Diabetic Clinic of the Komfo Anokye Teaching Hospital (KATH), Kumasi-Ghana. Data was collected using self-structured questionnaires to obtain socio-demographic characteristics, T2DM-related knowledge, DSM practices, SCT constructs; beliefs in treatment effectiveness, level of self-efficacy, perceived family support, and healthcare provider-patient communication. Path analysis was used to determine direct and indirect effects of T2DM-related knowledge, perceived family support, and healthcare provider service on DSM practices with level of self-efficacy mediating the relationships, and beliefs in treatment effectiveness as moderators. The mean age of the participants was 53.1(SD = 11.4) years and the average disease duration of T2DM was 10 years. Most of the participants ($65.5\%$) had high (>6.1mmol/L) fasting blood glucose (FBG) with an average of 6.93(SD = 2.41). The path analysis model revealed that age ($$p \leq 0.176$$), gender ($$p \leq 0.901$$), and duration of T2DM ($$p \leq 0.119$$) did not confound the relationships between the SCT constructs and DSM specified in the model. A significant direct positive effect of family and friends’ support (Critical ratio (CR) = 5.279, $p \leq 0.001$) on DSM was observed. Self-efficacy was a significant mediator in this relationship (CR = 4.833, $p \leq 0.001$). There were significant conditional indirect effects (CIE) for knowledge of T2DM and family and friends’ support at medium and high levels of belief in treatment effectiveness ($p \leq 0.05$) via level of self-efficacy on DSM practices. However, no evidence of moderated-mediation was observed for the exogenous variables on DSM. Diabetes-related knowledge of T2DM, family and friends’ support, level of self-efficacy, and belief in treatment effectiveness are crucial in DSM practices among Ghanaian T2DM patients. It is incumbent to consider these factors when designing interventions to improve DSM adherence.
## Background
Diabetes mellitus (DM) is a chronic health condition with devastating consequences on patients and public health. It is characterized by high blood glucose levels that are caused by defects in insulin release or action [1]. Type II *Diabetes mellitus* (T2DM), constitute over half of all known diabetes cases, affecting an estimated 463 million adults globally [2, 3]. Meanwhile, the International Diabetes Federation (IDF) has extrapolated that 700 million people will have the disease by 2045 [2, 3]. The most disturbing aspect is that individuals in sub-Saharan Africa are the most affected, with a recorded prevalence ranging between 7–$20\%$ [4]. Currently, Ghana’s diabetes prevalence is $6.4\%$ [5] in adults and if measures to confront the disease are not revamped, the prevalence is likely to increase due to the country’s already under-resourced and outstretched healthcare system and lifestyle behaviours.
Research has shown that T2DM progression is due to poor lifestyle choices and hence, promoting self-management practice through the modification of health behaviours, is the surest way to delaying diabetes-related mortality and morbidity [6, 7]. Basic diabetes self-management (DSM) practices include medication adherence, daily glucose monitoring, maintaining a healthy diet, regular exercising, and daily foot examinations [8–10]. Accumulating evidence indicates that compliance and implementation of DSM practices improve the health of T2DM patients, whereas T2DM patients with little self-management skills are prone to the negative consequences of the disease [11, 12].
To this end, researchers and health professionals have developed theories and cognitive models in pursuit of health promotion behaviours and interventions [13, 14]. However, most of the health promotion and theories only predict health behaviour and cannot explain the interactions between the constructs of the models [13, 15]. Social cognitive theory (SCT) has emerged as the most suitable model for examining health-related behaviours among individuals with chronic diseases [12, 16]. SCT proposes that cognitive processes could develop someone’s behaviour. The theory explains health-associated behaviours based on a three-way mutual interaction between environmental factors, personal factors, and behaviours [12, 16, 17].
Several studies have reported on the SCT model in forecasting self-care among T2DM patients and demonstrated that the actions of patients with T2DM could be influenced by their self-care regime [14, 18]. SCT has been reported to have several constructs such as knowledge of diabetes, self-efficacy, and belief in treatment effectiveness family and friends’ support, healthcare-patient communication [19–21]. These constructs interact with each other to allow patients to retain the influence of their disease [22]. Therefore, the SCT model provides opportunities to examine and explore the interaction between personal and environmental factors that influence DSM behaviours [12].
In Ghana, self-management practices of T2DM include healthy eating, being active and doing regular aerobic exercise, regular blood glucose monitoring, medical adherence, and knowledge about general complications of uncontrolled diabetes [23, 24]. Therefore, studies conducted among T2DM patients in Ghana have focused on medical compliance [25], factors that affect patients’ compliance to self-care activities [26], and a combination of both medication adherence and self-care behaviours [24]. While findings from studies have been impactful, they also revealed a lack of studies of the psychological aspect of self-management among Ghanaian T2DM patients. Moreover, these studies used correlation analyses for data collection that do not provide adequate information about the interactivity between independent and dependent variables.
SCT model is very effective in predicting and explaining DSM, however, most of the studies have been done in advanced countries Hence, the present study employed the SCT model to determine predictors for proper DSM practices among T2DM patients in a Ghanaian population.
## Study design/area
The study was a hospital-based cross-sectional study design undertaken from November 2018 to April 2019 at the Diabetes Clinic of Komfo Anokye Teaching Hospital (KATH) in the Ashanti Region of Ghana, Kumasi. KATH is in Kumasi, the Regional Capital of the Ashanti Region with a total projected population of 4,780,380 (2010 Ghana Population Census). KATH is the second-largest hospital in Ghana and Diabetes *Clinic is* part of the medical directorate that provides services to about 250 patients from Monday to Friday per week with an average of 10,706 out-patient attendance per year. The geographical location of the thousand two hundred- [1200-] bed capacity, the road network of the country, and commercial nature of Kumasi makes the hospital accessible to all the areas that share the boundaries with Ashanti Region and others that are further away. KATH takes direct referrals from 12 out of the 16 administrative regions in Ghana. These are the Ashanti, Bono, Bono East, Ahafo, Western North, Savannah, Northern, Northeast, Upper East, Upper West, and some parts of the Central and Eastern regions of Ghana. It also receives patients from neighbouring countries such as Ivory Coast and Burkina Faso. The diabetic centre of the KATH is situated beneath the medicine block (D block) just between the chest clinic and diagnostic centre and behind the emergency unit of the hospital.
## Study population and subject sampling strategy
A consecutive sampling approach was used to recruit a total of 420 T2DM participants aged 30 years and above who visited the KATH Diabetic Clinic for routine check-ups and treatment. Participants were consecutively recruited until the calculated sample size was achieved. This study was conducted in consultation with clinicians and qualified health professionals. T2DM was diagnosed by clinicians at KATH, and it was established based on the international classification of disease (ICD-10-CM Diagnosis Code E11.9). Each patient was carefully examined, and their medical records were thoroughly reviewed. As a result, we excluded all those individuals who were suffering from cancer, arthritis, infectious diseases, cardiovascular disease, thyroid disorders, pituitary disorders, and adrenal disorders. The study did not include pregnant and lactating mothers. Since T2DM is largely a disease of ageing, the study recruited only individuals who were 30 years and above. Furthermore, to limit potential confounding and the likelihood of recruiting participants with type 1 diabetes, we excluded participants on insulin injections. Patients who were physically or mentally challenged and those who had less than 6 months duration of diabetes were excluded.
## Sample size justification
Using the standard normal variate for significance (Z) of 1.96, $5\%$ margin of error (d), and an assumed adherence rate (p) of $39.2\%$ to self-care behaviours among adult T2DM patients [27], the recommended minimum sample size (n) for the study was 366 using the formula n = [p(1-p)]×z2/d2. However, to adjust for random error and strong statistical power, a maximum sample of 420 was used for the study.
## Ethical consideration
Ethical approval for the study was obtained from the Committee on Human Research, Publication, and Ethics of the School of Medicine and Dentistry, Kwame Nkrumah University of Science and Technology (KNUST), and KATH Ethical Committee (ref: CHRPE/AP/$\frac{084}{17}$). Participation was voluntary and written informed consent was obtained from each participant according to the Helsinki Declaration.
## Conceptual framework and hypotheses
The objective of self-management strategies is to assist patients with chronic conditions to perform therapeutic and preventive health-associated activities. Self-management behaviour is the mainstay of diabetes care. As a process-oriented approach, SCT epitomises direct and indirect analysis of the relationship between personal and environmental factors, and DSM. Personal factors involve diabetes-related knowledge, belief in treatment effectiveness, self-efficacy while environmental factors include family and friends support, and healthcare providers’ support [18].
Knowledge of the basic physiology of diabetes, medication, diet, testing, and monitoring of blood sugar, general diabetes are important elements of diabetes management [28]. Knowledge of diabetes has been shown to be associated with DSM but exerted influence on DSM among T2DM individuals through belief in treatment effectiveness and self-efficacy [29]. It is evident in the literature that belief in treatment effectiveness is associated with positive DSM among T2DM patients [30–32]. Interestingly, belief in treatment effectiveness seems to moderate relationships between diabetic knowledge, provider-patient communication, family support, and DSM among T2DM patients.
Self-efficacy implies a person’s confidence to carry out healthy behaviour. It is one of the principal concepts of SCT that is fundamental to behavioural accomplishment [33, 34]. Besides the exploration of the association between personal factors and DSM among T2DM individuals, studies have shown that family and friend support exerts its effect on DSM through belief in treatment effectiveness and self-efficacy [34–36]. These pieces of evidence show that family and friends’ support could directly or indirectly impact DSM among T2DM. Communication between a healthcare provider and a patient is critical in the management of diabetes. Studies have reported that the absence of advice from healthcare providers negatively affects DSM among T2DM [11, 34, 37].
Therefore, we hypothesized that self-efficacy mediates the relationship between personal and environmental factors and DSM with belief in treatment effectiveness moderating this association [Fig 1].
**Fig 1:** *Hypothesized model of determinants influencing DSM among T2DM in a Ghanaian population.Fig 1 shows modified model functions of predictors of DSM. The level of self-efficacy mediates the relationship of personal and environmental factors with DSM. The dependent variables include diabetes-related knowledge, belief in treatment effectiveness, self-efficacy, family and friends’ support, and healthcare provider service. The dependent variable is diabetes self-management.*
## Measurement instrument
A self-reported questionnaire and validated questionnaires were used for the data collection on study variables. The first part of the questionnaire consisted of 28 questions divided into three sections including section A: demographic characteristics, section B: health profile, and section C: knowledge of T2DM. The second part consisted of 39 validated questions that measured DSM practices and SCT constructs. The DSM practices and SCT questionnaires have been validated [11, 12, 37, 38] respectively. In addition, we assessed the reliability of the questionnaires in our setting as described in the sections below. The questionnaire was completed by each patient via a face-to-face interview approach by ethically recognised researchers and in consultation with clinicians. All interviews and data were collected at the Diabetic Clinic of the KATH. Interviews were conducted during the morning hours for the regular diabetic Clinic at the KATH.
## Measurement of DSM
DSM practices were assessed using the summarised version of Diabetes Self-Care Activities (DSCA) [37, 38]. The study utilised a revised DSCA constituting 10 items that determine medical adherence (2 items), healthy eating habits (2 items), physical activities (2 items) foot care (2 items), and FBG testing (2 items) [38]. Each item had a scale response that ranged from 1 to 7 indicating diabetes self-care practices over one week. The total score on the scale ranged between 0 and 70. Internal consistency and reliability of the scale were assessed to be α = 0.744.
## Level of self-efficacy
Self-efficacy for diabetes scale was used to measure perceived self-efficacy [33, 34]. Seven items were used to measure the respondents’ perceived self-efficacy. On a Likert scale of 0 (definitely, yes) to 4 (definitely not) where higher scores indicate poorly perceived self-efficacy, each item was reversed coded to indicate higher scores for higher perceived self-efficacy. Internal consistency and reliability of the scale were assessed to be α = 0.618.
## Belief in treatment effectiveness
To measure belief in treatment effectiveness, we adapted a questionnaire from Xu, [37]. The scale of the questionnaire was defined as the perceived relevance of self-care in managing diabetes. Eight items were used to assess the perceived benefits. Each item had a Likert scale that ranged from 0 (Not important) to 4 (extremely important) to measure the grades of perceived benefits. Internal consistency and reliability of the scale were assessed to be α = 0.746.
## Family and friends’ support
The scale used to measure family support was adapted from Xu et al [11]. The scale consisted of 7 items for measuring family supportive behaviours on a Likert scale of “never = 0” to “always = 4”. The reliability and validity of the family support scale were determined to be α = 0.827.
## Healthcare provider service
Seven questions were used to measure healthcare provider-patient communication. Item scale ranged from “never = 0” to “always = 4”. Internal consistency and reliability of the scale were assessed to be α = 0.496. The scale used to measure healthcare provider-patient communication was adapted from Xu et al [11].
## Diabetes-related knowledge
The questionnaire for general diabetes knowledge was employed to determine the knowledge level of diabetes among participants [12]. Eleven questions about diabetes were used to measure diabetes-related knowledge. Each question item had a scale response that ranged from 0 to 2 indicating “No”, “Yes” or “Don’t know”. The total score on the scale ranged between 2 and 22. Internal consistency and reliability of the scale were assessed to be Cronbach’s alpha (α) = 0.697.
## Data analysis
Data were analyzed using R version 4.0.2 and IBM SPSS AMOS version 25. The normality of the distribution of numeric data was determined using the Kolmogorov–Smirnov. Categorical data were presented as frequencies. The Mann-Whitney test was used to compare skewed data. Data that were distributed normally were presented as mean ± standard deviation and a T-test was used to compare between groups. The correlation between diabetes-related knowledge, SCT constructs, and DSM was conducted with Spearman’s correlation test. Structural equation modelling (SEM) was used to establish how the sample data closely fit the theory-driven model, by describing the relations of the dependency between the latent variables, which are usually accepted to have cause-and-effect outcomes [15]. A path analysis was conducted to describe the nature of the relationship between the HBM constructs and DSM, controlling for age, gender, and duration of disease. A p-value of less than 0.05 was deemed statistically significant.
## Results
The mean age of the study participants was 52.4 years (SD = 12.9) and a higher proportion of them were above 61 years ($35.2\%$). There were more females than males ($53.3\%$ vs. $46.7\%$). Also, the male participants were significantly older compared to the females (54.2 ± 12.2 vs. 50.3 ± 13.3, $$p \leq 0.018$$). The average duration of DM was 9.9 years (SD = 6.9) and most of the participants have had the condition for 6–10 years ($40.5\%$). Majority of the participants were self-employed ($50.0\%$), married ($72.4\%$), and had completed high school education ($39.5\%$). Of the total participants, $73.1\%$ ($\frac{307}{420}$) had T2DM with co-morbid hypertension, $65.5\%$ ($\frac{275}{420}$) had uncontrolled FBG levels (> 6.1mmol/L) and $41.7\%$ ($\frac{175}{420}$) had diabetes-related microvascular complications [Table 1].
**Table 1**
| Variable | Total (n = 420) | Female (n = 194) | Male (n = 226) | P-value |
| --- | --- | --- | --- | --- |
| Age (years) mean±SD | 52.9 ±13.4 | 51.1±13.7 | 54.2±12.9 | 0.018 |
| Age groups (years)^ | | | | |
| 30–40 | 107 (25.5) | 57 (29.4) | 50 (22.1) | 0.093 |
| 41–50 | 82 (19.5) | 42 (21.6) | 40(17.7) | 0.325 |
| 51–60 | 83 (19.8) | 29 (14.9) | 54(23.9) | 0.02 |
| ≥ 61 | 148 (35.2) | 66 (34.0) | 82 (36.3) | 0.61 |
| Duration of T2DM (years) mean±SD | 9.9±6.9 | 9.6±6.9 | 10.3±6.9 | 0.343 |
| Duration of T2DM (years)^ | | | | |
| ≤ 5 | 133 (31.7) | 65(33.5) | 68 (30.1) | 0.529 |
| 6–10 | 170 (40.5) | 80 (41.2) | 90 (39.8) | 0.823 |
| >10 | 117 (27.8) | 49 (25.3) | 68 (30.1) | 0.276 |
| Occupational Status | | | | |
| Self-employed | 210 (50.0) | 102 (52.6) | 108 (47.8) | 0.435 |
| Government employee | 116(27.8) | 51 (26.3) | 65 (28.8) | 0.585 |
| Pensioners | 44(10.5) | 21(10.8) | 23(10.2) | 0.874 |
| Not working | 50 (11.9) | 20(10.3) | 30 (13.3) | 0.367 |
| Marital status | | | | |
| Single | 17 (4.1) | 11 (5.7) | 6 (2.7) | 0.142 |
| Married | 304 (72.4) | 144 (74.2) | 160(70.8) | 0.587 |
| Divorced | 19(4.5) | 7(3.6) | 12(5.3) | 4831.0 |
| Widowed | 80 (19.0) | 32 (16.5) | 48 (21.2) | 0.215 |
| Educational level | | | | |
| Basic School | 103 (24.5) | 46 (23.7) | 57 (25.2) | 0.734 |
| High School | 166(39.5) | 77 (29.7) | 89 (39.4) | 0.999 |
| Tertiary | 151 (36.0) | 71 (36.6) | 80 (35.4) | 0.919 |
| Regular source of income | | | | |
| No | 92(21.9) | 45(23.2) | 47(20.8) | 0.553 |
| Yes | 328(78.1) | 149(76.8) | 179(79.2) | 0.553 |
| With hypertension comorbidity | | | | |
| No | 113 (26.9) | 59 (30.4) | 54 (23.9) | 0.456 |
| Yes | 307 (73.1) | 135 (44.0) | 172 (76.1) | 0.456 |
| Microvascular complications | | | | |
| No | 245 (58.3) | 121 (62.4) | 124 (54.9) | 0.12 |
| Yes | 175 (41.7) | 73 (37.6) | 102 (45.1) | 0.12 |
| Current (FBG) (mmol/L) | 6.93±2.41 | 6.98±2.53 | 6.88±2.30 | 0.786 |
| Controlled (≤ 6.1mmol/L) | 145 (34.5) | 67 (34.5) | 78 (34.5) | 0.996 |
| Uncontrolled (> 6.1mmol/L) | 275 (65.5) | 127 (65.5) | 148 (65.5) | 0.996 |
## Strength and direction of association between study variables
The path analysis model revealed that age ($$p \leq 0.176$$), gender ($$p \leq 0.901$$), and duration of T2DM ($$p \leq 0.119$$) did not confound the relationships between the HPM constructs and DSM specified in the model. With respect to the associations among the control variables: significant positive associations between age and duration of T2DM (Critical Ratio (CR) = 7.531, $p \leq 0.001$) and between age and gender (CR = 2.340, $$p \leq 0.019$$). Among the exogenous variables, a significant positive association was found between knowledge of diabetes mellitus and perceived family support (CR = 5.429, $p \leq 0.001$). The associations among the endogenous variables revealed that level of self-efficacy had a significant positive relationship with diabetes self-management (CR = 14.009, $p \leq 0.001$). In assessing direct significant associations between exogenous and endogenous variables, we found a significant negative association between perceived family support and level of self-efficacy (CR = -2.545, $p \leq 0.011$), and DSM practices (CR = 5.279, $p \leq 0.001$), supporting H3 [Fig 2]. No significant direct effect of knowledge of T2DM and healthcare provider service on DSM practices, implying that H1 and H2 were not supported. A significant association between the level of self-efficacy and the interaction between belief in the treatment and exogenous variables was observed [Table 2].
**Fig 2:** *Standardized estimates of path analysis model illustrating the perceived association between key determinants of diabetes self-management whilst controlling for age, gender, and duration of T2DM.* TABLE_PLACEHOLDER:Table 2
## Testing the fit of the conceptual model and evidence of mediated effect
The fit statistics indicate that good model fit was achieved (GFI = 0.976 > 0.95, AGFI = 0.943 > 0.9), with excellent parsimony-adjusted indexes (RMSEA < 0.024, $95\%$ CI: [0.012, 0.043], PCLOSE = 0.642). The difference between the residuals of the sample covariance matrix and the hypothesized model indicates a good fit (SRMR = 0.044 < 0.080). With respect to the variance of the endogenous variables explained by exogenous variables, $31\%$ of the variability in the level of self-efficacy was explained and $40\%$ of the variability in DSM was explained.
In testing for the evidence of mediated-moderation, the simple slopes for the exogenous variables were tested on the level of self-efficacy at three different levels of belief in treatment effectiveness using the standard pick-a-point approach. We evaluated the significance of estimates of total effects (which evaluates the combined indirect and direct effects,) and that of specific paths through the mediation variable (level of self-efficacy), moderated by belief of treatment effectiveness, based on 2000 bootstrap estimates from the bias-corrected percentile method [Table 3]. This allowed for the construction of confidence bounds around the estimates obtained for conditional indirect and direct effects. There were significant differences in simple slopes (SS) for knowledge of T2DM on level of self-efficacy at the medium (SS = 0.527, $95\%$ CI: [0.175, 0.919], $$p \leq 0.005$$) and high levels of belief in treatment effectiveness (SS = 0.158, $95\%$ CI: [-0.004, 0.323], $$p \leq 0.055$$). Similar results were found for perceived family support on level of self-efficacy at medium (SS = 0.090, $95\%$ CI: [0.025, 0.164], $$p \leq 0.007$$) and high levels of belief in treatment efficacy (SS = 0.109, $95\%$ CI: [0.022, 0.201], $$p \leq 0.018$$). No significant difference was found in how healthcare provider service affects the level of self-efficacy irrespective of the belief in treatment effectiveness.
**Table 3**
| Exogenous variables | Parameters | Simple slopes for heat exposure on adaptation strategies (SS) | Simple slopes for heat exposure on adaptation strategies (SS).1 | Simple slopes for heat exposure on adaptation strategies (SS).2 | Conditional indirect effects (CIE) | Conditional indirect effects (CIE).1 | Conditional indirect effects (CIE).2 | Index of moderation mediation (IMM) | Index of moderation mediation (IMM).1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Exogenous variables | Parameters | Belief in treatment effectiveness | Belief in treatment effectiveness | Belief in treatment effectiveness | Belief inf treatment effectiveness | Belief inf treatment effectiveness | Belief inf treatment effectiveness | Moderated-mediation (MM) | Moderated-mediation (MM) |
| Exogenous variables | Parameters | Low | Medium | High | Low | Medium | High | Index of MM | Remarks |
| Knowledge of mellitus diabetes | Estimate | 0.460 | 0.527 | 0.594 | 0.481 | 0.551 | 0.621 | -0.092 | No moderated-mediation effect |
| Knowledge of mellitus diabetes | 95% CI | [-0.124, 1.145] | [0.175, 0.919] | [0.191, 0.982] | [-0.114, 1.280] | [0.178, 1.014] | [0.198, 1.075] | [-3.746, 4.086] | No moderated-mediation effect |
| Knowledge of mellitus diabetes | p-value | 0.126 | 0.005 | 0.004 | 0.118 | 0.004 | 0.004 | 0.984 | No moderated-mediation effect |
| Perceived Family support | Estimate | 0.072 | 0.090 | 0.109 | 0.075 | 0.095 | 0.114 | -0.082 | No moderated-mediation effect |
| Perceived Family support | 95% CI | [-0.024, 0.160] | [0.025, 0.164] | [0.022, 0.201] | [-0.024, 0.173] | [0.026, 0.177] | [0.025, 0.219] | [-0.716, 0.447] | No moderated-mediation effect |
| Perceived Family support | p-value | 0.150 | 0.007 | 0.018 | 0.153 | 0.006 | 0.016 | 0.719 | No moderated-mediation effect |
| Healthcare Provider service | Estimate | -0.108 | -0.148 | -0.188 | -0.113 | -0.155 | -0.197 | 0.229 | No moderated-mediation effect |
| Healthcare Provider service | 95% CI | [-0.525, 0.159] | [-0.432, 0.103] | [-0.504. 0.149] | [-0.559, 0.169] | [-0.452, 0.107] | [-0.512, 0.166] | [-2.021, 1.913] | No moderated-mediation effect |
| Healthcare Provider service | p-value | 0.439 | 0.253 | 0.255 | 0.442 | 0.250 | 0.261 | 0.787 | No moderated-mediation effect |
There were significant conditional indirect effects (CIE) for knowledge of T2DM at medium and high levels of belief in treatment effectiveness ($p \leq 0.05$). However, the indexes of mediated-moderation (IMM). indicated that there was no evidence of moderated-mediation for knowledge of diabetes (IMM = -0.092, $95\%$ CI: [-3.746, 4.086], $$p \leq 0.984$$) on diabetes self-management [Table 3]. We also found significant conditional indirect effects for perceived family support at medium and high levels of belief in treatment effectiveness ($p \leq 0.05$), but no moderated-mediation effect was observed (IMM = -0.082, $95\%$ CI: [-0.716, 0.447], $$p \leq 0.719$$). No significant conditional indirect effect and moderated-mediated effect was observed for healthcare provider service ($p \leq 0.05$) [Table 3].
## Discussion
The purpose of this study was to test a hypothesized model describing the effects of personal and environmental factors on DSM. We employed the path analysis model to allow for the identification of mediation influences (direct and indirect effects of variables) or factors and to demonstrate associations between the SCT constructs (exogenous and endogenous variables) and DSM. Based on our conceptual model, the present study assessed self-efficacy as a mediating influence in the association between personal and environmental traits, and DSM, and how belief in treatment effectiveness moderates this relationship. Altogether, the path analysis model demonstrated a moderate positive relationship between the SCT constructs and DSM. Our results showed that perceived family support had a significant direct effect on DSM practices and the combined path from perceived family support and level of self-efficacy had a significant effect on DSM.
The result of the current study also revealed that $31\%$ of the variability in the level of self-efficacy by the exogenous variables and their interaction with the moderating variables. Level of self-efficacy expresses people’s self-belief in their abilities to perform specific behaviours under a particular situation [39, 40] and in this present study, the conditional indirect effect of perceived family support and knowledge of T2DM via a level of self-efficacy was significant for patient DSM practices. This finding is in line with reports from a previous cross-sectional study by Didarloo et al., among Iranian women with T2DM [35]. In that study, they identified self-efficacy as the strongest predictor of self-care behaviours [35]. Another cross-sectional study by Tol et al [41] further reported that self-efficacy also had a greater influence on DSM practices among T2DM patients in a Thailand population. Systematic review investigations on the level of self-efficacy in patients with diabetes indicate that self-efficacy can positively influence health care behaviours [42]. Per previous reports and findings, medical researchers and health professionals have suggested that diabetes is a self-management disease and hence, it is the duty of the patients in part to take care of themselves [43, 44]. Recent pieces of evidence across multiple behaviour domains have shown that increased self-efficacy is associated with improved health outcomes, especially in T2DM patients [31, 45, 46]. As such, healthcare professionals should design better intervention strategies focussed on normalizing patients’ experiences and validating their subjective experiences, ultimately promoting their confidence, and reinforcing patient self-efficacy.
Our study did not find any statistically significant direct effect of healthcare provider-patient communication on DSM practices. This result is contrary to findings of many studies that have shown that belief in treatment effectiveness and healthcare provider support have a positive relationship with DSM practices [10, 11, 32, 40, 47]. Healthcare providers can encourage patients to self-manage with the supply of compassionate, practical, and individualized support. Nevertheless, belief in treatment effectiveness played a role in the indirect influence of healthcare provider services via self-efficacy on DSM practices. Like other studies [30, 32, 48], beliefs in treatment effectiveness may have influenced DSM practices through a change in self-efficacy. This relationship has been well substantiated based on the findings that T2DM patients effectively follow diet plans and self-monitor blood sugar levels when they believe in the benefits of undertaking DSM behaviours appropriately [48]. We found that the higher belief in treatment effectiveness observed a high slope for the exogenous variables’ score on level of self-efficacy.
Public health education is a significant component of T2DM management. Like in most studies, there has been a statistically significant association between educational level and proper health behavior practices [11, 24, 29, 49]. In Particular, T2DM patients with some prior level of diabetes knowledge were more likely to undertake DSM practices [24, 28, 50]. These documented findings were confirmed with a report from a randomized single-blind controlled study that assessed the educational effect on self-management among T2DM [51]. The authors found that a two-week follow-up after a diabetes education program significantly improved self-management among T2DM patients [51]. A notable finding of this current study was that the direct path from knowledge of T2DM to DSM was nonsignificant. Instead, knowledge affected DSM indirectly through belief in treatment effectiveness and self-efficacy. These findings concord with reports from previous studies that have shown that knowledge did not lead to behaviour change directly [11, 29] but affected DSM indirectly through endogenous mediating variables [11]. In our analysis, knowledge of diabetes was significant for and affected both endogenous mediating variables. Adequate knowledge is important to improve DSM practices, but an individual’s belief in treatment effectiveness and self-efficacy might also be involved. It seems that knowledge is necessary but not sufficient alone for behaviour changes in DSM among diabetic patients.
In this present study, we also observed family support as a significant positive factor for DSM practices. It has been well documented in the literature, that modification of behaviours and management of oneself are burdensome for people living with T2DM [47, 52]. One way to relieve this burden and promote better health outcomes is by providing social support via families and friends [53], which is positively associated with greater psychological and physiological well-being and reduced risk of morbidity and mortality in many chronic diseases [54]. Our findings of a direct influence are consistent with reports from previous cross-sectional and family-based intervention studies [29, 55, 56] which found family support to be a predictor for patient compliance with DSM practices. Patients’ confidence level is built by strong support from family, which results in more efficient self-management and improved disease management [11]. Appraisal and information effect can be obtained through social support and this offers coping strategies designed to assist patients in managing diabetes-related stress and changed daily routines [36, 57]. A cross-sectional study by Mayberry and Osborn [36] reported that relatives who demonstrated better self-care behaviours were the ones that were better informed about diabetes and had better social support. Thus, it is essential to improve a better understanding of disease and the relationships between patients and their family members in clinical settings in such a way as to foster positive and supportive behaviours.
Overall, the cumulative effects of knowledge of T2DM, self-efficacy, and belief in treatment effectiveness influenced diabetes self-management practices and this relationship was well supported by our model. Perceived family support also showed a significant direct effect on DSM practices in this population. A limitation of this study is the cross-sectional nature of the study limits the capacity to unveil the causal relationship between predictors and DSM practices. Nevertheless, this is the first study among T2DM in a Ghanaian population, where a plenary description of factors associated with DSM practices has been extensively explored. The application of the SCT in this study has allowed for the identification of predictors of DSM and analysis of various moderating factors in DSM practices among diabetic patients by controlling for cofounders such as age, gender, and duration of disease. Additionally, the current findings add substantially to our understanding of the significant role SCT constructs play in DSM practices among Ghanaian T2DM patients. Considering the very devastating rate of morbidity and mortality associated with T2DM, more efforts will be required to augment mainstream clinical management approaches by appropriately developing behavioural change interventions with a focus on mediating factors that will positively influence patient self-care management behaviours and practices.
## Conclusion
Overall, our results provide a detailed appreciation of the interaction between personal and environmental factors and their effect on proper DSM practices among T2DM patients. Self-efficacy, belief in treatment effectiveness, and prior diabetes knowledge emerged as the most significant facilitating factors for proper DSM practices among T2DM patients. In this relationship, self-efficacy served as a significant mediating variable. Perceived family support also showed a significant direct effect on DSM practices in our studied population. Iinterventions targeted to address mediator roles such as patient self-efficacy, belief in treatment effectiveness, and family support may be required to improve proper DSM performance among T2DM patients. Other approaches that target behavioural factors for proper DSM practices in the wider population should be encouraged.
## References
1. Vilsbøll T, Knop F, Krarup T, Johansen A, Madsbad S, Larsen S. **The pathophysiology of diabetes involves a defective amplification of the late-phase insulin response to glucose by glucose-dependent insulinotropic polypeptide—regardless of etiology and phenotype**. (2003.0) **88** 4897-4903. DOI: 10.1210/jc.2003-030738
2. 2Diabetes facts & figures [https://idf.org/aboutdiabetes/what-is-diabetes/facts-figures.html]
3. Williams R, Karuranga S, Malanda B, Saeedi P, Basit A, Besançon S. **Global and regional estimates and projections of diabetes-related health expenditure: Results from the International Diabetes Federation Diabetes Atlas**. (2020.0) **162** 108072. DOI: 10.1016/j.diabres.2020.108072
4. Mercer T, Chang AC, Fischer L, Gardner A, Kerubo I, Tran DN. **Mitigating the burden of diabetes in sub-saharan Africa through an integrated diagonal health systems approach**. (2019.0) **12** 2261. DOI: 10.2147/DMSO.S207427
5. Asamoah-Boaheng M, Sarfo-Kantanka O, Tuffour AB, Eghan B, Mbanya JC. **Prevalence and risk factors for diabetes mellitus among adults in Ghana: a systematic review and meta-analysis**. (2019.0) **11** 83-92. DOI: 10.1093/inthealth/ihy067
6. Shrivastava SR, Shrivastava PS, Ramasamy J. **Role of self-care in management of diabetes mellitus**. (2013.0) **12** 1-5. DOI: 10.1186/2251-6581-12-14
7. Chrvala CA, Sherr D, Lipman RD. **Diabetes self-management education for adults with type 2 diabetes mellitus: a systematic review of the effect on glycemic control**. (2016.0) **99** 926-943. DOI: 10.1016/j.pec.2015.11.003
8. Baquedano IR, dos Santos MA, Martins TA, Zanetti ML. **Self-care of patients with diabetes mellitus cared for at an emergency service in Mexico**. (2010.0) **18** 1195-1202. DOI: 10.1590/s0104-11692010000600021
9. Rhee MK, Slocum W, Ziemer DC, Culler SD, Cook CB, Elkebbi IM. **Patient adherence improves glycemic control**. (2005.0) **31** 240. DOI: 10.1177/0145721705274927
10. Shigaki C, Kruse RL, Mehr D, Sheldon KM, Ge B, Moore C. **Motivation and diabetes self-management**. (2010.0) **6** 202-214. DOI: 10.1177/1742395310375630
11. Xu Y, Savage C, Toobert D, Pan W, Whitmer K. **Adaptation and testing of instruments to measure diabetes self-management in people with type 2 diabetes in mainland China**. (2008.0) **19** 234-242. DOI: 10.1177/1043659608319239
12. Wattanakul B. (2012.0)
13. Amico KR, Mugavero M, Krousel-Wood MA, Bosworth HB, Merlin JS. **Advantages to using social-behavioral models of medication adherence in research and practice**. (2018.0) **33** 207-215. DOI: 10.1007/s11606-017-4197-5
14. Ghoreishi M-S, Vahedian-Shahroodi M, Jafari A, Tehranid H. **Self-care behaviors in patients with type 2 diabetes: Education intervention base on social cognitive theory**. (2019.0) **13** 2049-2056. DOI: 10.1016/j.dsx.2019.04.045
15. Boonroungrut C, Fei H. **The Theory of Planned Behavior and Transtheoretical Model of Change: a systematic review on combining two behavioral change theories in research**. (2018.0) **16** 75-87
16. Bandura A. (2009.0) 110-140
17. Luszczynska A, Schwarzer R. **Social cognitive theory**. (2015.0) 225-251
18. Borhaninejad V, Iranpour A, Shati M, Tahami AN, Yousefzadeh G, Fadayevatan R. **Predictors of self-care among the elderly with diabetes type 2: using social cognitive theory**. (2017.0) **11** 163-166. DOI: 10.1016/j.dsx.2016.08.017
19. Wendling S, Beadle V. **The relationship between self-efficacy and diabetic foot self-care**. (2015.0) **2** 37-41. DOI: 10.1016/j.jcte.2015.01.001
20. Smith Y, Garcia-Torres R, Coughlin SS, Ling J, Marin T, Su S. **Effectiveness of Social Cognitive Theory–Based Interventions for Glycemic Control in Adults With Type 2 Diabetes Mellitus: Protocol for a Systematic Review and Meta-Analysis.**. (2020.0) **9** e17148. DOI: 10.2196/17148
21. Fitzgerald JT, Funnell MM, Anderson RM, Nwankwo R, Stansfield RB, Piatt GA. **Validation of the revised brief diabetes knowledge test (DKT2)**. (2016.0) **42** 178-187. DOI: 10.1177/0145721715624968
22. Mahadi N, Abdullah SN, Baskaran S. **A Review of Social Cognitive Theory and Self-care for Type 2 Diabetes Mellitus Patients**. (2020.0) **11** 148-159
23. Stephani V, Opoku D, Beran D. **Self-management of diabetes in Sub-Saharan Africa: a systematic review**. (2018.0) **18** 1-11. DOI: 10.1186/s12889-018-6050-0
24. Afaya RA, Bam V, Azongo TB, Afaya A, Kusi-Amponsah A, Ajusiyine JM. **Medication adherence and self-care behaviours among patients with type 2 diabetes mellitus in Ghana.**. (2020.0) **15** e0237710. DOI: 10.1371/journal.pone.0237710
25. Bruce SP, Acheampong F, Kretchy I. **Adherence to oral anti-diabetic drugs among patients attending a Ghanaian teaching hospital**. (2015.0) **13**. DOI: 10.18549/pharmpract.2015.01.533
26. Mogre V, Abanga ZO, Tzelepis F, Johnson NA, Paul C. **Adherence to and factors associated with self-care behaviours in type 2 diabetes patients in Ghana**. (2017.0) **17** 1-8. PMID: 28183304
27. Ayele K, Tesfa B, Abebe L, Tilahun T, Girma E. **Self care behavior among patients with diabetes in Harari, Eastern Ethiopia: the health belief model perspective**. (2012.0) **7** e35515. DOI: 10.1371/journal.pone.0035515
28. Taha H. **Assessment of knowledge about the role of diet regime and physical exercise to control diabetes among type 2 Diabetic Patients in Khartoum State, Sudan**. (2014.0) **3** 916-921
29. Tahmasebi R, Noroozi A, Tavafian SS. **Determinants of self-management among diabetic patients: A path analysis**. (2015.0) **27** NP524-NP534. DOI: 10.1177/1010539513475652
30. Lee King PA, Cederbaum JA, Kurzban S, Norton T, Palmer SC, Coyne JC. **Role of patient treatment beliefs and provider characteristics in establishing patient–provider relationships**. (2015.0) **32** 224-231. DOI: 10.1093/fampra/cmu085
31. Adu MD, Malabu UH, Malau-Aduli AE, Malau-Aduli BS. **Enablers and barriers to effective diabetes self-management: A multi-national investigation**. (2019.0) **14** e0217771. DOI: 10.1371/journal.pone.0217771
32. Dao‐Tran TH, Anderson D, Chang A, Seib C, Hurst C. **Factors associated with self‐management among Vietnamese adults with type 2 diabetes**. (2018.0) **5** 507-516. DOI: 10.1002/nop2.158
33. 33Self-efficacy for diabetes [http://patienteducation.stanford.edu]
34. Kurnia AD, Amatayakul A, Karuncharernpanit S. **Predictors of diabetes self-management among type 2 diabetics in Indonesia: Application theory of the health promotion model**. (2017.0) **4** 260-265. DOI: 10.1016/j.ijnss.2017.06.010
35. Didarloo A, Shojaeizadeh D, Asl RG, Habibzadeh H, Niknami S, Pourali R. **Prediction of self-management behavior among Iranian women with type 2 diabetes: application of the theory of reasoned action along with self-efficacy (etra).**. (2012.0) **14** 86. PMID: 22737561
36. Mayberry LS, Osborn CY. **Family Support, Medication Adherence, and Glycemic Control Among Adults With Type 2 Diabetes.**. (2012.0) **35** 1239-1245. DOI: 10.2337/dc11-2103
37. Xu Y. *Understanding the factors influencing diabetes self-management in chinese people with type 2 diabetes using structural equation modeling* (2005.0)
38. Toobert DJ, Hampson SE, Glasgow RE. **The summary of diabetes self-care activities measure: results from 7 studies and a revised scale**. (2000.0) **23** 943-950. DOI: 10.2337/diacare.23.7.943
39. Pender NJ, Murdaugh CL, Parsons MA. *Health promotion in nursing practice* (2006.0)
40. Oftedal B, Karlsen B, Bru E. **Life values and self‐regulation behaviours among adults with type 2 diabetes**. (2010.0) **19** 2548-2556. DOI: 10.1111/j.1365-2702.2010.03243.x
41. Tol A, Rad GS, Eslami AA, Alhani F. **Factors Influencing Self-Management Behavior in Type-2 Diabetes Patients: A Strategy Proposed to be Adopted When Planning Theory/Model Based Interventions**. (2012.0)
42. Mohebi S, Azadbakht L, Feizi A, Sharifirad G, Kargar M. **Review the key role of self-efficacy in diabetes care**. (2013.0) **2**. DOI: 10.4103/2277-9531.115827
43. Van Smoorenburg AN, Hertroijs DF, Dekkers T, Elissen AM, Melles M. **Patients’ perspective on self-management: type 2 diabetes in daily life**. (2019.0) **19** 1-8. PMID: 30606168
44. Lambrinou E, Hansen TB, Beulens JW. **Lifestyle factors, self-management and patient empowerment in diabetes care**. (2019.0) **26** 55-63. DOI: 10.1177/2047487319885455
45. Nyunt SW, Howteerakul N, Suwannapong N, Rajatanun T. **Self-efficacy, self-care behaviors and glycemic control among type-2 diabetes patients attending two private clinics in Yangon, Myanmar**. (2010.0). PMID: 21073070
46. Krichbaum K, Aarestad V, Buethe M. **Exploring the connection between self-efficacy and effective diabetes self-f management**. (2003.0) **29** 653-662. DOI: 10.1177/014572170302900411
47. Norris SL, Nichols PJ, Caspersen CJ, Glasgow RE, Engelgau MM, Jack L. **Increasing diabetes self-management education in community settings: a systematic review**. (2002.0) **22** 39-66. DOI: 10.1016/s0749-3797(02)00424-5
48. von Arx L-BW, Gydesen H, Skovlund S. **Treatment beliefs, health behaviors and their association with treatment outcome in type 2 diabetes**. (2016.0) **4** e000166. DOI: 10.1136/bmjdrc-2015-000166
49. Glanz K, Rimer BK, Viswanath K. (2008.0)
50. Al-Qazaz HK, Sulaiman SA, Hassali MA, Shafie AA, Sundram S. **Diabetes knowledge and control of glycaemia among type 2 diabetes patients in Penang, Malaysia**. (2012.0) **3** 49-55
51. Atak N, Gurkan T, Kose K. **The effect of education on knowledge, self management behaviours and self efficacy of patients with type 2 diabetes**. (2008.0) **26** 66
52. Weijman I, Ros WJ, Rutten GE, Schaufeli WB, Schabracq MJ, Winnubst JA. **The role of work-related and personal factors in diabetes self-management**. (2005.0) **59** 87. DOI: 10.1016/j.pec.2004.10.004
53. Barrera M, Sandler IN, Ramsay TB. **Preliminary development of a scale of social support: Studies of college students**. *American Journal of Community Psychology* (1981.0) **9** 435-447
54. Uchino BN. **Social support and health: a review of physiological processes potentially underlying links to disease outcomes**. (2006.0) **29** 377-387. DOI: 10.1007/s10865-006-9056-5
55. Mastura I, Chew BH, Lee PY, Cheong AT, Sazlina SG, Jamaiyah H. **Control and treatment profiles of 70,889 adult type 2 diabetes mellitus patients in Malaysia-a cross sectional survey in 2009**. (2011.0) **3**
56. Withidpanyawong U, Lerkiatbundit S, Saengcharoen W. **Family-based intervention by pharmacists for type 2 diabetes: A randomised controlled trial**. (2019.0) **102** 85-92. DOI: 10.1016/j.pec.2018.08.015
57. Kadirvelu A, Sadasivan S, Ng SH. **Social support in type II diabetes care: a case of too little, too late.**. (2012.0) **5** 407. DOI: 10.2147/DMSO.S37183
|
---
title: 'Lived experiences of women with low birth weight infants in the Solomon Islands:
A descriptive qualitative study'
authors:
- Lydia S. Kaforau
- Gizachew A. Tessema
- Hugo Bugoro
- Gavin Pereira
- Jonine Jancey
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022132
doi: 10.1371/journal.pgph.0001008
license: CC BY 4.0
---
# Lived experiences of women with low birth weight infants in the Solomon Islands: A descriptive qualitative study
## Abstract
Every year, around 20 million women worldwide give birth to low birth weight (LBW) infants, with majority of these births occurring in low-and middle-income countries, including the Solomon Islands. Few studies have explored the pregnancy lived experience of women who deliver LBW infants. The aim of the study is to understand the lived experience of women in the Solomon Islands who gave birth to LBW infants by exploring their personal (socio-demographic and health), behavioural, social and environmental contexts. We used a qualitative descriptive approach and purposely selected 18 postnatal women with LBW infants in the Solomon Islands for an in-depth interview. All data were analysed using thematic analysis in NVivo. We identified six themes reported as being related to LBW: health issues, diet and nutrition, substance use, domestic violence, environmental conditions and antenatal care. Our findings suggest that women in the Solomon Islands are exposed to various personal, behavioural, social and environmental risk factors during pregnancy that can impact birth outcomes, particularly LBW. We recommend further research should be redirected to look at the factors/themes identified in the interviews.
## Introduction
It is estimated that 140 million women are pregnant worldwide every year, with most of these pregnancies occurring in low-and middle-income countries (LMICs) [1]. Although pregnancy can be a rewarding experience and an exciting journey, this may not be the case for many women in LMICs due to their living conditions and exposure to risk factors that may adversely affect their pregnancies and birth outcomes [2–7]. Women’s health during pregnancy in LMICs is largely influenced by socio-demographics, health, behavioural and environmental factors. Socio-demographic factors include maternal age, household income and education levels, and health factors include malaria infections and anaemia [8, 9]. Behavioural risk factors include substance use such as tobacco, alcohol and betel nut [10, 11], and environmental risks include the lack of access to antenatal care and poor sanitation [12, 13].
The Solomon *Islands is* located in the South Pacific region, an island country with a fragile subsistence-based economy that relies on foreign aid for health services and socio-economic development [14]. There are high rates of unemployment, illiteracy, and poverty [14], with more than $74\%$ of the population living in rural areas as subsistence farmers [15]. Despite government efforts to ensure that health facility access is within an hour’s reach, health service access and quality remain a problem, particularly for rural women [16]. Community health services and programs are often lacking in many rural communities due to geographical isolation and dispersed communities across the 992 scattered islands [16]. The number of health workers per population density is lower than that of most countries in the Asia Pacific region. Reports showed there were approximately 19 doctors, 145 nurses, and midwives per 100 000 of the population [17, 18]. Clean water supplies and sanitation are non-existent in more than half of the rural communities [19, 20], and substance use, especially alcohol, marijuana, and kwaso (illegally distilled alcohol), are prevalent among the population [21, 22]. The Solomon *Islands is* also a patriarchal society comprising various indigenous cultures, where men, as head of the family, own the land and possess inherited wealth, predisposing women to oppression and violence [23].
Low birth weight (LBW), defined as birth weight below 2500 grams is the most widely used perinatal benchmark for adverse birth outcomes globally [24]. LBW comprise $20\%$ of global births (approximately 20 million per year), with $95\%$ of these births occurring in LMICs [24]. The Solomon Islands has a prevalence for LBW of $10\%$, which places it within the minimal range compared to South-East Asian countries and Melanesian countries of the Pacific such as PNG and Vanuatu (between 10–$17\%$ prevalence) [2, 25–27]. However, this rate is elevated compared to some countries in the Polynesia region, with Tonga, Samoa and Tuvalu reporting LBW of well below $7\%$ [2, 27]. In addition, LBW in the Solomon Islands was predominantly preterm births, with a recent hospital-based study reporting LBW accounted for $77\%$ of neonatal deaths [28–30].
LBW can result from women’s exposure to the various risk factors experienced during pregnancy as quantified by numerous studies in LMICs [2–7, 25, 27, 31–34]; however, limited studies have quantified the associated risk factors for LBW locally. Furthermore, no studies have been conducted in the Solomon Islands on women’s lived experiences during pregnancy. This study aimed to understand the lived experience during the most recent pregnancy of women in the Solomon Islands who gave birth to LBW infants by exploring and describing their personal (socio-demographic and health), behavioural, social and environmental context during their pregnancy. The enquiry into the women’s personal accounts will illuminate their unique experiences in their respective communities.
## Study design and sampling
We employed a descriptive qualitative study using purposive sampling to recruit women with LBW infants to participate in one-on-one in-depth interviews. This methodology was deemed appropriate to explore women’s personal stories and experiences during their recent pregnancies resulting in LBW babies [35, 36]. The consolidated criteria for reporting qualitative research (COREQ) checklist was used to ensure high-quality research reporting [36] (S1 File).
## Study setting and recruitment
Eighteen postnatal women aged 15 to 39 years were recruited from the special care nursery of the National Referral Hospital (NRH) in the Solomon Islands. NRH receives referral from all provinces. However, since it is located in Honiara on the Island of Guadalcanal, most of the women were either referred from Honiara or rural Guadalcanal. The women were required to have given birth to LBW infants weighing less than 2500 grams, irrespective of gestational age, and within the last week before recruitment. It was also a requirement that the health of the mother and infant was stable. Women were required to be fluent in Pidgin English, the Solomon *Islands lingua* franca. The recruitment of participants was halted when data saturation was reflected in no new information emerging.
## Data collection
The lead author (LSK) conducted in-depth one on one semi-structured interviews with women within one week of delivery and before the infants were discharged from hospital. Due to travel restrictions resulting from the global COVID-19 pandemic, although it was originally planned to undertake face-to face interviews, all interviews were conducted via telephone. The processes were assisted by two local research assistants (RA) trained in ethical research processes. JT (RA1), a clinical nurse instructor in the paediatric and neonatal ward of the National Referral Hospital in the Solomon Islands, identified women meeting the inclusion criteria and invited them to participate in the telephone interviews after explaining the aims of the study, participant information sheet, and interview procedures and ensuring provision of secured private room for the women to sit during the interviews. LK (RA2), an academic at the University of South Pacific, Solomon Island Campus, supported the study by arranging and scheduling the telephone call processes and incentives for the women. Women received a pair of infant booties in appreciation of their time.
The interview schedule was assessed for face and content validity by an expert panel who have PhDs and expertise in clinical and public health research. It was then trialled with four Solomon Island women living in Australia to ensure the content was understandable and informed the research aim. We asked the women about their health, health behaviours, social and environmental conditions, access to antenatal care and their understanding of risk factors during pregnancy that may have contributed to having a LBW infant. Demographic data (age, residence, ethnicity, education level) were also collected (S1 Table). Prior to the interviews, a further three interviews were conducted with women of LBW infants in the Solomon Islands, to assess the interview schedule and the feasibility of the telephone interview method. Bracketing was undertaken through dialogue between the researchers to identify potential preconceived notions that might influence the data collection and analysis [37]. From this, a list of preconceived notions that might influence the data collection and analysis was written on a memo as a checklist guide to refer to during the data analysis.
The women who provided informed consent and permission for recording of the interview were placed in a secure private room for the interview. Confidentiality and privacy were strictly maintained for all participants throughout the interview process. RA1 (JT) escorted each participant into the room and assisted them in answering the telephone (if they were unfamiliar with the handset). Once comfortable, RA1 left and closed the door before the telephone interview commenced. The interviews ranged from 30 to 70 minutes, with an average time of 54 minutes.
## Data management and analysis
Data were de-identified by removing the participant names and replacing them with unique codes, that were uploaded to a password-protected database. All interviews were conducted in Pidgin English. Interviews were audio taped and field notes were taken. The lead author (LSK) transcribed all the interviews from Pidgin transcriptions directly into English, as she is fluent in both languages. Also, given that the Solomon Islands Pidgin creole is a derivative of the English language combined with local languages, direct transcribing and translation were considered optimal as both languages are somewhat analogous in most terms. To ensure validity in the translation process, three translated scripts were reviewed by three colleagues fluent in Pidgin and English. The translated and transcribed data were uploaded into NVivo and then analysed using thematic analysis [38, 39]. This involved repeatedly reading through each interview to become immersed and familiar with its content and annotating meanings arising from the data; generating succinct labels (codes) to identify essential features of the data; collating data to identify significant broader patterns of meaning and potential themes; reviewing themes against the dataset to ensure they told the data’s story; defining and naming themes; and weaving together the analytic narrative with quotes to illuminate themes [38, 39]. For validity and reliability, co-authors (JJ and GAT) who are experienced in qualitative research provided guidance and reviewed the generated themes with the lead author (LSK), leading to a consensus on the final themes.
## Ethical statement
Ethics approval was obtained from the Solomon Islands Health Research and Ethics Board of the Ministry of Health and Medical Services with ethics approval number HRE$\frac{039}{19.}$ Reciprocal ethics was also granted by Curtin University Human Research Ethics Committee (HREC) with approval number HRE2020-0530. Informed consent was taken from all participants involved in this study. Written consent was taken from parents and guardians of mothers of infants of 18 years and below.
## Demographic characteristics of participants
A total of 18 women aged 15 to 39 years were recruited, $80\%$ ($$n = 15$$) were in a union (permanent relationship), and $60\%$ ($$n = 11$$) were from rural areas (S1 Table).
## Themes
Six major themes were identified: 1) health issues, 2) substance use, 3) diet and nutrition, 4) domestic violence, 5) environmental conditions, and 6) antenatal care.
## Health issues
Most women ($$n = 11$$) reported experiencing a range of physical health issues during their pregnancies, which they believed impacted their pregnancy outcomes, such as early birth and LBW infant. For example, a woman stated: “I was bleeding the entire pregnancy, and that was why I had a small baby. ”[P8, 23-year-old] Illnesses experienced during pregnancy included malaria and non-specific infections, antepartum haemorrhage (APH) and chest pain as attested by two of the women. “ At 4 months, I developed fever and shivering, got tested at the clinic and confirmed malaria positive. I believe I gave birth early due to the malaria infection. ”[P10,18-year-old]“I had prolonged chest pain [chronic] the entire pregnancy without any diagnosis of my illness. I delivered my baby at seven months. ”[P2,39-year-old] Women who experienced health problems during their pregnancy did not have the confidence to talk about these issues with the health staff, especially when it came to their sexual health. This was particularly so for young women aged less than 20 years who experienced symptoms suggestive of sexually transmitted infections (STIs) or urinary tract infections (UTIs) during their pregnancy. These young women did not inform the health staff, thereby no screening or treatment was provided. This was reflected by two 19-year-old women as follows. “ I had experienced abnormal white foul pus coming out from my private parts with painful urination. I did not seek help from the clinic nurse. ”[P15] “I had yellow foul vaginal discharge during pregnancy. I did not tell the nurses. They [nurses] did not check me [vaginal swab] and I did not receive any treatments. ”[P14] Women from both rural and urban areas tended to engage in hard physical work due to their subsistence lifestyle. Five women reported physical stress, falls and accidents during their pregnancy, which they believed resulted in early labour and birth. For example, two of them described their experiences as follows. “ I fell twice at five months of pregnancy on my way to the garden and slipped over the slippery bush track while carrying a heavy load of garden produce to sell at the market. I barely had rest, from gardening and selling vegetable. From then on, I started to develop abdominal pains that led me to an early birth. ”[P17,35-year-old, urban] “The nurse told me the bleeding was because of hard labour of carrying a heavy load without much rest. ”[P13,25-year-old, rural] Conversely five other women claimed they did not experience any major health problems during their entire pregnancy as reflected by a19-year-old. “ I was well during pregnancy except the regular morning sickness. ”[P14]
## Substance use
Substance use was widespread among this population, especially betel nut, tobacco, marijuana and alcohol. Most women ($$n = 11$$) reported chewing betel nut and smoking locally grown or manufactured tobacco during pregnancy, despite knowing that these substances could have adverse health consequences. Some women ($$n = 4$$) even reported being heavy users, or not being able to stop using betel nut or tobacco during pregnancy. “ I am a heavy betel nut chewer; I chewed up to 7, 6, 5 nuts per day before and during my last pregnancy. I also smoked tobacco and savusavu roll [homegrown dried tobacco leaves]of 5 to 6 rolls per day for 3 years. Savusavu is powerful, can cause dizziness and bad feeling. I had to shower to feel better; then, I took one more roll. I think I had my baby very early because of this. ”[P13,25-year-old] Some women ($$n = 3$$) denied or were unsure of betel nuts’ potential harm to their fetus, especially among the rural women as highlighted by two of the women as follows: “I don’t think betel nut will affect my baby. ”[P9, 25-year-old] “I am not sure if betel nut is harmful for my unborn baby. ”[P3,16-year-old] A few women ($$n = 3$$) who were heavy chewers (3–6 betel nuts per day) believed betel nut helped them during their pregnancy, as it reduced the bad taste and gave them a good feeling and more energy, as expressed by two of the women: "I could not stop chewing betel nut because it improved my taste and made me feel better. "[P6,20-year-old] “I could not stop because betel nut gives me good feelings, energy to do things and keeps me awake. ”[P18,18-year-old] Some women ($$n = 6$$) also reported witnessing pregnant women consuming other substances, including kwaso (home-made distilled alcohol), marijuana and beer during pregnancy. “ I have heard and seen other women have taken kwaso, marijuana and beer during pregnancy. ”[P11,34-year-old] One participant expressed how women use the substances during pregnancy to cope with stress. “ I knew of a neighbour who took too much alcohol and smoked marijuana during pregnancy and gave birth to a sick baby. She said she took those due to stress. ”[P7,30-year-old] While substance use was quite prevalent among the women, several women ($$n = 7$$) abstained from these substances during pregnancy due to religious beliefs and knowledge of their harmful effects. “ As seventh day Adventists, we are not allowed to use betel nut and tobacco as it can cause problems with blood [anaemia].”[P12,18-year-old] These women demonstrated some level of knowledge of the impact of substance use during pregnancy. “ I don’t take any of these during pregnancy, I am too scared of them. Betel nut, tobacco and marijuana can lead to having a small baby. ”[P7,30-year-old]
## Diet and nutrition
Knowledge about a nutritious diet was at times limited, especially among the less educated women (primary school or less), and those from rural areas. “ I don’t know what a healthy diet is like. ”[P13,25-year-old] Most participants ($$n = 14$$) reported eating at least three meals a day. “ I ate a roasted green banana [plantain] for breakfast, tea for lunch and sweet potatoes for dinner [3 meals]. I did not like to eat sausage. ”[P12,18-year-old] However, their reported daily dietary intake during pregnancy tended to lack variety, often comprising starchy vegetables and carbohydrates (e.g., potatoes, rice) with limited quantities of protein. “ I usually eat potatoes, cassava, rice, cabbage like Amau [local cabbage], slippery cabbage [Ibika], and pakchoi [Chinese cabbage] and beans, tomato and eggplant. I think chicken is not healthy to eat during pregnancy. ”[P13,25-year-old] Although more than half of the participants ($$n = 10$$) reported local home-grown food to be a healthier option, some also believed that some meat and fish were not healthy and so these food types were not eaten during pregnancy. “ Sausage, chicken and meat are not healthy. ”[P12,18-year-old] Many ($$n = 12$$) of the participants reported cultural food taboos, food restrictions, or personal preference for certain foods which contributed to their limited protein intake as reflected by two of the women. “ We are not allowed to eat crab, pig meat [pork] and the fish with big mouths during pregnancy. ”[P15,19-year-old] “Pregnant women are not allowed to eat ura [prawns], reef fish, shellfish and megapod eggs; they will cause problems during labour. ”[P14,19-year-old] For some women ($$n = 6$$) food security was impacted by environmental calamities, such as heavy rains and floods, which destroyed crops. “ Heavy rains and flood destroyed our food garden, and all our banana plants died which affected our food supply. ”[P12,18-year-old] Also limited finances reduced women’s ability to purchase nutritious foods. “ The biggest challenge was to buy enough nutritious food and share with my household of extended family members. ”[P11,34-year-old] Conversely, five women reported that they experienced no food restrictions during their pregnancy and had an adequate food supply. “ I planted my own vegetables, so I have enough food with good nutrition. ”[P4,24-year-old]
## Domestic violence
Almost half of the participants ($$n = 8$$) reported experiencing both physical and emotional violence during their pregnancy from their partners or significant others. Many of them ($$n = 7$$) believed their exposure to violence had contributed to their child’s early birth as expressed by two of the women. “ Every weekend, he would return drunk and beat me up. He beat me and kicked me on my back twice during pregnancy which led to the premature birth. ”[P15,19-year-old] “During the last pregnancy of my twin babies, he [husband] bashed me. It was a terrifying experience. As a result of this, I delivered my twins prematurely. ”[P7,30-year-old] Although ten of the women did not experience domestic violence, most women understood the negative consequences of violence during pregnancy. “ Experiencing domestic violence during pregnancy can cause premature labour. ”[P15,19-year-old]
## Environmental conditions
Some rural women ($$n = 5$$) described their dwellings and traditionally built homes as placing them at increased exposure to insects, particularly mosquitoes. “ We all lived in thatched sago palm-built house with one big open space where my family of seven lives [overcrowded]. It is not safe from pests and insects like mosquitoes. ”[P13,25-year-old] Many of the women ($$n = 7$$) reported living in large extended families and believed that overcrowding increased their risk of infection. We lived in a two-bed room house and built a small extension in the vicinity for 14 of us including relatives. It is very crowded with a high risk for spreading infection. ”[P11,34-year-old] Half of the women, living in rural and urban areas ($$n = 5$$), walked long distances to access water for drinking and domestic use. The water sources were often unsafe during rainy and drought seasons due to contamination. “ We used rainwater, stream and river nearby for drinking, cooking and domestic use. We must go downhill to reach the water sources. The stream and river get murky during rains or dries up during drought. ”[P14,19-year-old] Two of the women claimed they had premature births due to stress from carrying heavy water containers. “ I think I had premature labour because I worked so hard every day, carrying large containers of water for household use from the river. ”[P16,32-year-old] “I carried water containers from the stream to the house uphill. One day I suddenly felt labour pains and gave birth early.[24 weeks]”[P10,18-year-old] Poor sanitation was pervasive, with a lack of proper toilet facilities throughout urban and rural areas. Some participants did not have toilet facilities and used bushes, creeks and the seaside to defecate. “ We do not have a proper toilet; we use the nearby bushes[chuckled]. In Wagina, we would use the seaside. We have not received any information on good toilets and sanitation. ”[P17,35-year-old] “We do not have a proper toilet like others in the community. We used a nearby creek as a toilet. ”[P15,19-year-old] However, a few participants ($$n = 3$$) living in urban areas had inbuilt water and sanitary systems. “ I lived in a townhouse with an inbuilt modern toilet and water system. ”[P7,30-year-old]
## Antenatal care
Women described several challenges related to their antenatal care. These included limited access to health services, limited provision of health information by health professionals, and their own inability to seek out health information. The ability to obtain antenatal health care was frequently impeded by access issues, such as distances to antenatal services, which was more common in rural areas, as well as geographical barriers (e.g., mountainous areas), extreme weather conditions (e.g., tropical rain), and unaffordable transportation costs as expressed by three of the women. “ Our clinic is very far from the village without road access [where we live] and took two hours by foot. I walked to get there, climbing hills and walking down the creek footpath which was very difficult. ”[P5,38-year-old] “My antenatal attendance depends on the weather. I would not go when the weather was bad. ”[P4,24-year-old]“The biggest challenge [to antenatal care] was the distance and travel costs. I would not go If I do not have the money for bus fare. ”[P12,18-year-old] Participants also nominated a range of barriers to obtaining optimal antenatal health care. These included the poor condition of health facilities, medicine shortages, no or inadequate health screening, long waiting times, and lack of professionalism among nursing staff. This was illustrated by participants who described the conditions women often experienced. “ Health clinic services [provision] should be improved; women should be checked properly and must be given malaria medicine [prophylaxis] during antenatal [care].”[P5,38-year-old]“The clinic needs proper clean beds for pregnant women to lay on. Also, the nurses always scold us, are slow in their performance, resulting in long waiting hours and some mothers just left. ”[P7,30-year-old] Conversely, some participants ($$n = 7$$) reported better access to health facilities and appreciation of the antenatal cares service provided. “ My closest clinic is 50 meters away. I am satisfied with the services provided, and the information was adequate. The nurse checked my baby’s position, gave health advice did a urine test and supplied the red medicine for blood [ferrous sulphate tablets].”[P1,20-year-old]
## Summary of findings
Our study is the first to explore women’s lived experiences during pregnancy in the Solomon Islands, providing an understanding of the perceived risks these women are exposed to during pregnancy that led to an early birth and LBW. Risk factors identified stemmed from the women’s knowledge and experiences, which were described as personal (socio-demographic and health issues), behavioural (diet and nutrition and substance use) social (domestic violence) and environmental conditions. Although some women were conscious of the potential impact of these risk factors on pregnancy outcomes, many were unaware.
## Overview of findings
Commonly reported personal health issues experienced by the women during pregnancy included malaria, APH, STIs and UTIs, which are recognised as causes of early births and LBW [27, 40–42]. Malaria is endemic in the Solomon Islands [26], a known risk for LBW, affecting 125 million pregnant women globally [41, 43–45]. APH was another reported health issue, which can be triggered by infection, obstetric causes, or physical trauma [40]. Some women claimed that having bleeding during pregnancy was due to strenuous physical work. Others reported experiences of falls and accidents and perceived that these prompted the early onset of their labour, leading to early births. Similar findings have been reported in LMICs, where physical stress, falls, and trauma during pregnancy were recognised as causes of APH and early births [27, 42]. The lack of screening and treatment for STIs or UTIs was also a nominated issue, with women being unaware of the risks to their pregnancies. STIs and UTIs are well-established risk factors for LBW [46, 47], with a previous local study confirming that UTIs affected $59\%$ of pregnant women and were linked to preterm births [30]. The women also reported not seeking help from health care providers for these STIs symptoms, a negative attitude to sexual health problems in the Solomon Islands, which has also been observed in a previous study [26]. Furthermore, STI screening, including vaginal swabbing is out of the scope of the current Solomon Island antenatal protocol, which is a shortfall in antenatal care [48, 49].
Our findings indicate that the interviewed women had limited knowledge of nutrition, reporting a diet high in starchy vegetables and active avoidance of meat protein (e.g., chicken and sausage) during their pregnancy. Local food taboos led to the further avoidance of protein, as certain fish were viewed as causing illness in newborns, a sickness referred to as ‘fis’ [50]. Although there is limited local research on food taboos, studies in other LMICs have found them to be associated with suboptimal nutrition during pregnancy and adverse birth outcomes [51–53]. In the Solomon Island $41\%$ of women of reproductive age experience anaemia, a condition mostly affecting women of low educational status and those living in rural areas, which further increases their risk of having preterm and LBW infants [26].
The women’s diet was also impacted by food availability due to environmental factors, such as heavy rains and floods causing the destruction of crops and contributing to food insecurity [50, 54–56]. This is despite the introduction of the National Food Security, Food Safety and Nutrition Policy 2019–23 by the Ministry of Health and Medical Services (MHMS) [57]. Priority policy areas included better nutrition for vulnerable populations (e.g., women and children), strengthening of the food supply chain, awareness of safe and healthy food choices and promotion of health and nutrition. To date, it remains uncertain whether this has been well implemented at a consumer level.
Substance use (e.g., betel nut, tobacco, alcohol and marijuana) during pregnancy was also reported by our participants, with betel nut use found to be particularly frequent. Although many of the women were aware of the impacts of these substances on pregnancy, there was a lack of knowledge and awareness, especially regarding betel nut use. There were also reports of betel nut addiction among some women who claimed to use betel nuts to increase energy levels and reduce bad tastes during pregnancy, which has also been reported in a Papua New Guinea (PNG) study [11]. Betel nut has been found to be associated with adverse birth outcomes, especially LBW and preterm birth, in several studies in PNG and Southeast Asia [10, 11, 13, 27, 58, 59]. Conversely, the women’s use of tobacco, alcohol and marijuana during pregnancy was limited, with the women reporting some level of awareness of their harmful effects. The use of tobacco and alcohol during pregnancy and their effect on pregnancy and birth has been well documented [58, 60, 61]. Although evidence on marijuana use in pregnancy in LMICs is limited, studies in the United States have shown its use to be common among younger women and a potential risk factor for LBW and other adverse birth outcomes [61–63]. There were also reports that women tended to use these drugs to relieve stress, which has been reported in a previous Solomon Island study [23].
Domestic violence inflicted by an intimate partner or significant other was the social risk women reported. Almost half of the women in this study reported experiencing physical or emotional violence during their pregnancy. This is consistent with previous Solomon Islands studies, which have reported that up to $66\%$ of women experience domestic violence from their intimate partner during pregnancy [23, 64]. Some of the women attested that being bashed or kicked in the back, led to early births, a serious consequence of violence during pregnancy [23, 64–67]. Despite the legislation of the Family Protection Act to protect families and promote safety and community education by the Solomon Islands government [64, 68], many of the women expressed a lack of social and community support, particularly in rural communities.
Environmental risks that women experienced included poor access to health facilities and lack of proper sanitation. The MHMS Role Delineation Policy underpins and safeguards affordable and accessible health care for its citizens, yet access to health care services remains a challenge [69]. It was evident that appropriate antenatal care was challenged by poor access, under-resourced health facilities, lack of staff professionalism and limited health information. Antenatal access was hampered by distances, geographical isolation, and extreme weather conditions, all governmental challenges in servicing this dispersed population [26]. Poor access to antenatal care impacts attendance, as reflected by previous local studies, which showed $85\%$ of women have late antenatal bookings (trimesters 2 and 3), and $25\%$ or less have only four antenatal visits [26, 70]. Poor access is a contextual risk for adverse pregnancy outcomes, which have been reported by other LMICs [12, 71]. Limiting health care, which is required during pregnancy [44, 45, 72], diminishes the ability to detect health risks and promote health services for better pregnancy outcomes. Women also expressed dissatisfaction with antenatal care due to not being treated with respect, and poor health facilities, also confirmed by one study from LMIC [71].
Poor water and sanitation remain pervasive in rural and urban areas of the Solomon Islands, even though the National Water and Sanitation Policy 12-year implementation plan was introduced in 2017 [73]. Poor water and sanitation poses a risk of vector and water-borne infections, as supported by a local study [26]. Although studies on sanitation and birth outcomes are limited, poor sanitation can threaten pregnancy due to infection [74, 75]. Furthermore, many of the women expressed that information on proper water and sanitation is limited in their community which may contribute to the lack of awareness of these risks.
## Study limitations and strength
This study provided an opportunity to talk with Solomon Island women from urban and rural areas, providing insights into their exposure to risk factors during their pregnancy. However, due to the COVID-19 travel restrictions, the interviews were conducted by telephone, which may have inhibited the discussion and reduced the opportunity for the researcher to observe the women’s body language. The women were mainly from Guadalcanal province and Honiara, and the findings were limited to these ethnic groups. Using a highly trained health professional as the interviewer can also be seen as a limitation as they may presume a greater level of participant knowledge than non-health professionals. Health professionals may themselves be influenced by aspects of care based on their prior experience, which governs the dialogue and interpretation. Participants may be less well-included to open up to trained health professionals who maintain their professional identity throughout the interview process compared to non-health professionals who are perceived as peers. On the other hand, the strength of the study was that the primary investigator is a female health professional, familiar with this type of participant, the clinical setting from which the women came, knowledgeable in the subject area, the local language of the participants (pidgin English) and translation processes.
## Conclusion
The study contributes to our understanding of the personal (socio-demographic and health), behavioural, social and environmental risk factors women who gave birth to LBW infants in the Solomon Islands experienced during pregnancy. The study identified women’s knowledge and experience of potential risk factors which included health issues, diet and nutrition, substance use, domestic violence, inadequate antenatal care, and environmental conditions. We recommend further research should be redirected to look at the factors/themes identified in the interviews.
## References
1. Lassi Z.S.. **Effects of Preconception Care and Periconception Interventions on Maternal Nutritional Status and Birth Outcomes in Low- and Middle-Income Countries: A Systematic Review.**. *Nutrients* (2020.0) **12** 606. DOI: 10.3390/nu12030606
2. Blencowe H.. **National, regional, and worldwide estimates of low birthweight in 2015, with trends from 2000: a systematic analysis**. *The Lancet Global Health* (2019.0) **7** e849-e860. DOI: 10.1016/S2214-109X(18)30565-5
3. Das Gupta R.. **Factors associated with low birth weight in Afghanistan: a cross-sectional analysis of the demographic and health survey 2015**. *BMJ Open* (2019.0) **9** e025715. DOI: 10.1136/bmjopen-2018-025715
4. Endalamaw A.. **Low birth weight and its associated factors in Ethiopia: a systematic review and meta-analysis**. *Italian Journal of Pediatrics* (2018.0) **44** 141. DOI: 10.1186/s13052-018-0586-6
5. Falcão I.R.. **Factors associated with low birth weight at term: a population-based linkage study of the 100 million Brazilian cohort.**. *BMC Pregnancy and Childbirth* (2020.0) **20** 536. DOI: 10.1186/s12884-020-03226-x
6. Khan N., Mozumdar A., Kaur S.. **Determinants of low birth weight in India: An investigation from the National Family Health Survey**. *American Journal of Human Biology* (2020.0) **32** e23355. DOI: 10.1002/ajhb.23355
7. Singh U., Ueranantasun A., Kuning M.. **Factors associated with low birth weight in Nepal using multiple imputation.**. *BMC Pregnancy and Childbirth* (2017.0) **17** 67. DOI: 10.1186/s12884-017-1252-5
8. Tessema Z.T.. **Prevalence of low birth weight and its associated factor at birth in Sub-Saharan Africa: A generalized linear mixed model.**. *PloS one* (2021.0) **16** e0248417-e0248417. DOI: 10.1371/journal.pone.0248417
9. Kassaw M.W.. **Trends of proximate low birth weight and associations among children under-five years of age: Evidence from the 2016 Ethiopian demographic and health survey data.**. *PLOS ONE* (2021.0) **16** e0246587. DOI: 10.1371/journal.pone.0246587
10. Ome-Kaius M.. **Determining effects of areca (betel) nut chewing in a prospective cohort of pregnant women in Madang Province, Papua New Guinea.**. *BMC Pregnancy Childbirth* (2015.0) **15** 177. PMID: 26286026
11. Senn M.. **Betel nut chewing during pregnancy, Madang province, Papua New Guinea.**. *Drug and Alcohol Dependence* (2009.0) **105** 126-131. DOI: 10.1016/j.drugalcdep.2009.06.021
12. Peters H., Vince J., Friesen H.. **Low birthweight at a Papua New Guinea highlands hospital**. *Journal of Tropical Pediatrics* (2001.0) **47** 17-23. DOI: 10.1093/tropej/47.1.17
13. De Silva M.. **Systematic review of areca (betel nut) use and adverse pregnancy outcomes.**. *Int J Gynaecol Obstet* (2019.0) **147** 292-300. DOI: 10.1002/ijgo.12971
14. Gibson J., Simler K., Carnahan M.. **, Solomon Islands poverty profile based on the 2012/13 household income and expenditure survey.**. *2015 Solomon Islands National Statistic Office Honiara* 56
15. 15SINSO, 2019
National National Housing and Census Project 2020, Solomon Islands National Statistic Office Honiara p. 9. *National National Housing and Census Project 2020* (2019.0) 9
16. 16UNFPA, Reproductive, maternal, newborn, child and adolescent health workforce. 2019
UNFPA: UNFPA. p. 148.. *Reproductive, maternal, newborn, child and adolescent health workforce* (2019.0) 148
17. Westcott M.. **Critical care resources in the Solomon Islands: a cross-sectional survey**. *BMC Int Health Hum Rights* (2012.0) **12** 1. DOI: 10.1186/1472-698X-12-1
18. 18Oecd and W.H. Organization, Health at a Glance. Health at a Glance: Asia/Pacific. 2020, Paris: OECD Publishing.. *Health at a Glance. Health at a Glance: Asia/Pacific* (2020.0)
19. 19MHMS, Descriptive Health Report, Solomon Islands 2018 MHMS, Honiara, Solomon Islands
20. 20WASH, Water Aid, Water Sector Analysis, Solomon Islands in Solomon Islands WASH analysis 2016 Ausralian Aid or DFAT Honiara p. 53.
21. Pratt S.. **The challenge of betel nut consumption to economic development: a case of Honiara, Solomon Islands.**. *Asia-Pacific Development Journal* (2014.0) 103
22. Tovosia S.. **Prevalence and Associated Factors of Betel Quid Use in the Solomon Islands: A Hyperendemic Area for Oral and Pharyngeal Cancer**. *The American Journal of Tropical Medicine and Hygiene* (2007.0) **77** 586-590. PMID: 17827384
23. 23SPC, Solomon Islands Family Health and Safety Study: A study on violence against women and children. 2009, Secretariat of the Pacific Community: Noumea p. 252.. *Solomon Islands Family Health and Safety Study: A study on violence against women and children* (2009.0) 252
24. Cutland C.L.. **Low birth weight: Case definition & guidelines for data collection, analysis, and presentation of maternal immunization safety data**. *Vaccine* (2017.0) **35** 6492-6500. PMID: 29150054
25. Dahlui M.. **Risk factors for low birth weight in Nigeria: evidence from the 2013 Nigeria Demographic and Health Survey.**. *Global Health Action* (2016.0) **9** 28822. DOI: 10.3402/gha.v9.28822
26. 26SINSO, Solomon Islands Demographic and Health Survey 2015 2017, SPC’s Noumea HeadquartersB.P. D5,: SPC’s Noumea Headquarters B.P. D5, 98848 Noumea Cedex, New Caledonia.
27. Kaforau L.S.K.. **Prevalence and risk factors of adverse birth outcomes in the Pacific Island region: A scoping review**. *The Lancet Regional Health—Western Pacific* (2022.0) **21** 100402. DOI: 10.1016/j.lanwpc.2022.100402
28. Tosif S.. **Assessment of the quality of neonatal care in the Solomon Islands.**. *J Paediatr Child Health* (2018.0) **54** 165-171. DOI: 10.1111/jpc.13686
29. Tosif S.. **Cause-specific neonatal morbidity and mortality in the Solomon Islands: An assessment of data from four hospitals over a three-year period.**. *J Paediatr Child Health* (2020.0) **56** 607-614. DOI: 10.1111/jpc.14699
30. Cafaro J.. **An assessment of current antenatal care practices and identification of modifiable risk factors for prematurity and low birth weight infants in pregnancy in Solomon Islands.**. *Rural and remote health,* (2015.0) **15** 369-371. PMID: 26363609
31. Gebregzabiherher Y.. **The Prevalence and Risk Factors for Low Birth Weight among Term Newborns in Adwa General Hospital, Northern Ethiopia.**. *Obstetrics and Gynecology International* (2017.0) 2149156. DOI: 10.1155/2017/2149156
32. Girma S.. **Factors associated with low birthweight among newborns delivered at public health facilities of Nekemte town, West Ethiopia: a case control study.**. *BMC Pregnancy and Childbirth* (2019.0) **19** 220. DOI: 10.1186/s12884-019-2372-x
33. 33WHO, Global Nutrition Targets 2025 Low Birth Weight Policy Brief. 2014
World Health Organisation World Health Organisation p. 1–8. *Global Nutrition Targets 2025 Low Birth Weight Policy Brief* (2014.0) 1-8
34. Chhea C., Ir P., Sopheab H.. **, Low birth weight of institutional births in Cambodia: Analysis of the Demographic and Health Surveys 2010–2014.**. *PLOS ONE* (2018.0) **13** e0207021. DOI: 10.1371/journal.pone.0207021
35. Jacobsen K.H.. *Introduction to health research methods: A practical guide* (2017.0) 5
36. Tong A., Sainsbury P., Craig J.. **, Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups.**. *International journal for quality in health care* (2007.0) **19** 349-357. DOI: 10.1093/intqhc/mzm042
37. Tufford L., Newman P.. *Bracketing in Qualitative Research* (2010.0) **11** 80-96
38. Braun V., Clarke V.. **Using thematic analysis in psychology.**. *Qualitative Research in Psychology* (2006.0) **3** 77-101
39. Maguire M., Delahunt B.. **Doing a Thematic Analysis: A Practical, Step-by-Step Guide for Learning and Teaching Scholars**. *AISHE-J* (2017.0) 3351-33514
40. Battula S.P., Mohammed N.H.A., Datta S.. **Antepartum haemorrhage. Obstetrics**. *Gynaecology & Reproductive Medicine* (2021.0) **31** 117-123
41. Bauserman M.. **An overview of malaria in pregnancy**. *Semin Perinatol* (2019.0) **43** 282-290. DOI: 10.1053/j.semperi.2019.03.018
42. Gupta N., Gandhi A.. *Principles of Critical Care in Obstetrics* (2016.0) **I** 281-301
43. Stanisic D.I.. **Risk factors for malaria and adverse birth outcomes in a prospective cohort of pregnant women resident in a high malaria transmission area of Papua New Guinea**. *Trans R Soc Trop Med Hyg* (2015.0) **109** 313-24. DOI: 10.1093/trstmh/trv019
44. Unger H.W.. **Microscopic and submicroscopic Plasmodium falciparum infection, maternal anaemia and adverse pregnancy outcomes in Papua New Guinea: a cohort study**. *Malaria Journal* (2019.0) **18** 302. DOI: 10.1186/s12936-019-2931-7
45. Guyatt Helen L., Snow Robert W.. **Impact of Malaria during Pregnancy on Low Birth Weight in Sub-Saharan Africa**. *Clinical Microbiology Reviews* (2004.0) **17** 760-769. DOI: 10.1128/CMR.17.4.760-769.2004
46. Johnson H.L.. **Sexually Transmitted Infections and Adverse Pregnancy Outcomes Among Women Attending Inner City Public Sexually Transmitted Diseases Clinics.**. *Sexually Transmitted Diseases* (2011.0) **38** 167-171. DOI: 10.1097/OLQ.0b013e3181f2e85f
47. Kalinderi K.. **Urinary tract infection during pregnancy: current concepts on a common multifaceted problem**. *Journal of Obstetrics and Gynaecology* (2018.0) **38** 448-453. DOI: 10.1080/01443615.2017.1370579
48. Marks M.. **Prevalence of sexually transmitted infections in female clinic attendees in Honiara**. *Solomon Islands. BMJ Open* (2015.0) **5** e007276
49. 49MHMS, Standard Treatment Manual Obstetrics & Gynaecology. 2018
Ministry of Health and Medical Service, Solomon Islands Honiara p. 220. *Ministry of Health and Medical Service* (2018.0) 220
50. 50Jones, C., et al., Foods and diets of communities involved in inland aquaculture in Malaita Province, Solomon Islands. Penang, Malaysia:CGIAR Research Program on Aquatic Agricultural Systems. Program Report: AAS-2014-30. 2014 WorldFish, P.O. Box 438, Honiara, Solomon Islands: WorldFish, P.O. Box 438, Honiara, Solomon Islands. p. 36.
51. Iradukunda F.. **Food taboos during pregnancy**. *Health Care Women Int* (2020.0) **41** 159-168. DOI: 10.1080/07399332.2019.1574799
52. Tsegaye D., Tamiru D., Belachew T.. **Food-related taboos and misconceptions during pregnancy among rural communities of Illu Aba Bor zone, Southwest Ethiopia. A community based qualitative cross-sectional study.**. *BMC Pregnancy and Childbirth* (2021.0) **21** 309. DOI: 10.1186/s12884-021-03778-6
53. Jerzy K.. **Food taboos and traditional customs among pregnant women in Papua New Guinea: Missed opportunity for education in antenatal clinics.**. *Contemporary PNG Studies* (2019.0)
54. Horsey B.. **Dietary Diversity of an Adult Solomon Islands Population.**. *Nutrients* (2019.0) **11** 1622. DOI: 10.3390/nu11071622
55. Albert J.. **Malnutrition in rural Solomon Islands: An analysis of the problem and its drivers.**. *Wiley-Maternal and Child Nutrition Malnutrition in rural Solomon Islands: An analysis of the problem and its drivers* 12. DOI: 10.1111/mcn.12921
56. Vogliano C.. **Assessing Diet Quality of Indigenous Food Systems in Three Geographically Distinct Solomon Islands Sites (Melanesia, Pacific Islands).**. *Nutrients* (2021.0) **13**
57. 57SIG, National Food Security, Food Safety and Nutrition Policy 2019–2023, M.o.A.a.L.D. Ministry of Health and Medical Services, Ministry of Fisheries and Marine Resources and Ministry of Education and Human Resources Development, Editor. 2019 Honiara, SolomoN Island Government: Honiara, SolomoN Island Government. p. 55.
58. Berger K.E.. **The Effects of Chewing Betel Nut with Tobacco and Pre-pregnancy Obesity on Adverse Birth Outcomes Among Palauan Women**. *Matern Child Health J* (2016.0) **20** 1696-703. DOI: 10.1007/s10995-016-1972-6
59. Hossain M.F.. **Adverse effects on health posed by consumption of Areca nut (Areca catechu L., family: Palmaceae).**. *Int J Community Med Public Health* (2015.0) **2** 357-60
60. Yang M.J.. **Betel quid chewing and risk of adverse birth outcomes among aborigines in eastern Taiwan**. *J Toxicol Environ Health A* (2001.0) **64** 465-72. DOI: 10.1080/152873901753215920
61. Brown H.L., Graves C.R.. **Smoking and Marijuana Use in Pregnancy**. *Clinical Obstetrics and Gynecology* (2013.0) **56**. DOI: 10.1097/GRF.0b013e318282377d
62. Thompson R., DeJong K., Lo J.. **Marijuana Use in Pregnancy: A Review**. *Obstetrical & gynecological survey* (2019.0) **74** 415-428. DOI: 10.1097/OGX.0000000000000685
63. Warshak C.R.. **Association between marijuana use and adverse obstetrical and neonatal outcomes**. *Journal of Perinatology* (2015.0) **35** 991-995. DOI: 10.1038/jp.2015.120
64. Koete B.. **Prevalence and Characteristics of Intimate Partner Violence (IPV) Among Pregnant Women Seeking Antenatal Care, Solomon Islands (2016).**. *Journal of Community Medicine & Health Education* (2017.0) **7** 1000558
65. Nongrum R.. **Domestic violence as a risk factor for maternal depression and neonatal outcomes: a hospital-based cohort study**. *Indian J Psychol Med* (2014.0) **36** 179-181. DOI: 10.4103/0253-7176.130989
66. 66WHO, Violence against women prevalence estimates, 2018: global, regional and national prevalence estimates for intimate partner violence against women and global and regional prevalence estimates for non-partner sexual violence against women. 2021.
67. Sigalla G.N.. **Intimate partner violence during pregnancy and its association with preterm birth and low birth weight in Tanzania: A prospective cohort study.**. *PLOS ONE* (2017.0) **12** e0172540. DOI: 10.1371/journal.pone.0172540
68. 68SIG, Family protection bill 2014. 2014. p. 72.. *Family protection bill 2014* (2014.0) 72
69. 69SIG, Role delineation policy for Solomon Islands, in Infrustructure, human resecources and equipment for improved health services provision towards a happy, health and reproductive Solomon Islands 2016 Ministry of Health and Medical Services Ministry of Health and Medical Services p. 54
70. Kluckow H.. **Socio-demographic predictors of unintended pregnancy and late antenatal booking in Honiara, Solomon Islands.**. *Australian and New Zealand Journal of Obstetrics and Gynaecology* (2018.0) **58** 349-357. DOI: 10.1111/ajo.12782
71. Finlayson K., Downe S.. **Why do women not use antenatal services in Low- and Middle-Income Countries? A Meta-Synthesis of qualitative studies**. *PLOS Medicine* (2013.0) **10** e1001373. DOI: 10.1371/journal.pmed.1001373
72. Aderoba A.K.. **Prevalence of helminth infestation during pregnancy and its association with maternal anemia and low birth weight**. *International Journal of Gynecology & Obstetrics* (2015.0) **129** 199-202. DOI: 10.1016/j.ijgo.2014.12.002
73. 73SI-NWSIP, Solomon Islands National Water and Sanitation Implementation Plan 2017–2033 An integrated national water resource and sanitation management plan to implement the goals and objectives of the National Water and Sanitation Policy, E.a.R.E. National Intersectoral Water Coordination Committee Ministry of Mines, Editor. 2017 National Intersectoral Water Coordination Committee Ministry of Mines, Energy and Rural Electrification: National Intersectoral Water Coordination Committee Ministry of Mines, Energy and Rural Electrification. p. 40.
74. Campbell O.M.R.. **Getting the basic rights–the role of water, sanitation and hygiene in maternal and reproductive health: a conceptual framework**. *Tropical Medicine & International Health* (2015.0) **20** 252-267. DOI: 10.1111/tmi.12439
75. Bradbury R.. **Ancylostoma ceylanicum Hookworm in the Solomon Islands**. *Emerging Infectious Diseases* (2017.0) **23**. DOI: 10.3201/eid2302.160822
|
---
title: 'Anti-Helicobacter pylori antibody status is associated with cancer mortality:
A longitudinal analysis from the Japanese DAIKO prospective cohort study'
authors:
- Satoshi S. Nishizuka
- Masahiro Nakatochi
- Yuka Koizumi
- Asahi Hishida
- Rieko Okada
- Sayo Kawai
- Yoichi Sutoh
- Keisuke Koeda
- Atsushi Shimizu
- Mariko Naito
- Kenji Wakai
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022139
doi: 10.1371/journal.pgph.0001125
license: CC BY 4.0
---
# Anti-Helicobacter pylori antibody status is associated with cancer mortality: A longitudinal analysis from the Japanese DAIKO prospective cohort study
## Abstract
Paradoxically, patients with advanced stomach cancer who are Helicobacter pylori-positive (HP+) have a higher survival rate than those who are HP-. This finding suggests that HP infection has beneficial effects for cancer treatment. The present study examines whether HP+ individuals have a lower likelihood of death from cancer than those who are HP-. Prospective cohort data ($$n = 4$$,982 subjects enrolled in the DAIKO study between 2008–2010) were used to assess whether anti-HP antibody status was associated with cancer incidence. The median age in the primary registry was 53 years-old (range 35–69 years-old). Over the 8-year observation period there were 234 ($4.7\%$) cancer cases in the cohort and 88 ($1.8\%$) all-cause deaths. Urine anti-HP antibody data was available for all but one participant ($$n = 4$$,981; $99.98\%$). The number of HP+ and HP- individuals was 1,825 ($37\%$) and 3,156 ($63\%$), respectively. Anti-HP antibody distribution per birth year revealed that earlier birth year was associated with higher HP+ rates. With a birth year-matched cohort ($$n = 3$$,376), all-cancer incidence was significantly higher in HP+ individuals than those who were HP- ($$p \leq 0.00328$$), whereas there was no significant difference in the cancer death rate between HP+ and HP- individuals ($$p \leq 0.888$$). Cox regression analysis for prognostic factors revealed that the hazards ratio of HP+ was 1.59-fold ($95\%$CI 1.17–2.26) higher than HP- in all-cancer incidence. Potential systemic effects of HP+ status may contribute to reduced likelihood of death for patients after an initial diagnosis of cancer.
## Introduction
Helicobacter pylori (HP) infection is considered to be an etiologic carcinogen of the stomach [1]. Paradoxically, however, investigators have recognized that the survival rate of patients with advanced gastric cancer who are HP+ is higher than that for HP- patients from studies carried out in a variety of countries that have varying HP genotype and diet as well as different therapeutic strategies for advanced gastric cancer [2–8]. Regarding HP infection in the context of post-operative adjuvant chemotherapy for patients with Stage II/III gastric cancer [9,10], our previous study demonstrated that those patients who were HP+ had a clearly higher survival rate over the 10-year observation period [11]. In particular, patients who were HP+ and received adjuvant chemotherapy with S-1 showed >$20\%$ better overall survival than did those who were HP-, which was further confirmed by propensity score matching analysis. Stratified Cox proportional hazards regression analysis revealed that all 10 subgroups had better survival in the respective HP+ groups [11]. More recently, our separate analysis revealed that the better survival rate in HP+ patients was further pronounced in patients with PD-L1- as well as S-1 adjuvant chemotherapy [12]. These results collectively suggest that HP infection sustains potential anti-tumor host immunity, which may contribute to prolonged survival.
Alternatively, HP infection could trigger non-specific enhancement of innate immune system activity to create a de facto heterologous immunological memory that has been termed trained immunity [13]. Such triggered immunity that has some anticancer effects has been shown in patients with high-risk non-muscular invasive bladder cancer for which intravesical instillation of Bacillus Calmette-Guérin (BCG) has been the gold-standard treatment [14]. For gastric cancer, immunotherapy using OK-432, a freeze-dried product prepared by incubating the low-virulence Su strain of group A *Streptococcus pyogenes* with penicillin [15], has been actively investigated [16]. Although not the gold-standard for gastric cancer, therapies like OK-432 that promote immune responses appear to suppress relapse in both humans and mouse models of gastric cancer [17–20]. If immune activation indeed plays a role in suppressing either the development or progression of cancer, then sustained HP infection should also show some beneficial effect in terms of cancer incidence and all-cause or cancer death in a longitudinal cohort.
The subject patients of advanced gastric cancer, including ours, had undergone gastrectomy, meaning that HP could no longer colonize the stomach. Thus, the observed effect of HP on prolonged survival could be systemic and perhaps lifelong, and thus may have an antitumor potential. Using data from the DAIKO prospective cohort in Nagoya, Japan, the present study aimed to understand whether the putative effect associated with HP infection has an antitumor effect with respect to reduction of incidence of all-cancer as well as breast, colorectal, lung, prostate, and stomach cancers. The effects of HP on all-cancer deaths were also assessed. Although the observation period of the DAIKO cohort to date is eight years, the beneficial effect of HP infection can be determined if the all-cancer death rate in HP+ individuals is lower than the all-cancer incidence in those who are HP+. This putative effect suggests that HP+ cancer patients have a lower likelihood of death compared to HP- cancer patients.
## Ethics statement
Ethical approval for the parent Japan Multi-Institutional Collaborative Cohort Study (J-MICC) study was granted by Nagoya University Graduate School of Medicine Institutional Review Board [21]. Based on the IRB approval by the Nagoya University Graduate School of Medicine, a separate ethical approval for the current analysis using the DAIKO study was granted by both Nagoya University Graduate School of Medicine IRB [2008-0618-2] and Iwate Medical University School of Medicine IRB (MH2018-019). The study was conducted according to Helsinki Declaration principles.
## Source cohort
The DAIKO study includes 5,165 participants aged 35–69 years-old who were residents of Nagoya and were enrolled between June 2008 and May 2010 at the Daiko Medical Center of Nagoya University, Nagoya, Japan. The J-MICC *Study is* the parental cohort of the DAIKO study, which includes 13 institutes across Japan with enrollment of >100,000. Follow-up surveys have been conducted since 2005 [21]; and the end of follow-up of the current study was Dec 31, 2016. The main objective of the J-MICC study is to investigate gene-environmental interactions with lifestyle-related diseases, mainly cancer. Recruitment for the DAIKO study was conducted as a part of the J-MICC study. The subject sources for J-MICC were as follows: 1) volunteers residing in the areas defined by local governmental administration; 2) health checkup examinees run by local governments; 3) visitors of health checkup facilities; and 4) visitors of a cancer hospital [21], whereas only the subject source of 1) was applied for the DAIKO study. In addition, the following criteria were applied for the DAIKO subject: 1) age between 35–69 years for males or females; and 2) residents of Nagoya city. The participants were invited by distribution of leaflets to the residents in Nagoya city. The DAIKO study is not a patient cohort, but may have included cancer patient at baseline during registration. The cohorts have also provided opportunities to perform cross-sectional studies on lifestyle factors, biomarkers, and genotypes [22–26]. In the present study, we analyzed DAIKO study data in terms of cancer incidence and survival to verify whether these events were associated with anti-HP antibody status that indicates the immune status of an individual with respect to past/present HP infection. Although the current analysis of this paper focuses on cancer incidence and deaths, due to the nature of the non-patient individual cohort, we do not have access to detailed patient information, such as staging, treatment, and pathological and genomic status.
## Study population
Participants in the present study were a subset of the DAIKO study, from whom written informed consent was obtained upon enrollment in studies outside of the J-MICC-assigned program. Baseline data were obtained from 5,165 participants, and 183 participants were excluded for the following reasons: lack of consent to be a part of research studies except for the parent J-MICC study ($$n = 170$$); ineligibility ($$n = 12$$); and revocation of informed consent at baseline ($$n = 1$$). Overall, 4,982 ($96.5\%$ of the J-MICC study participants) individuals were enrolled in the present study.
## Baseline variables
Baseline covariates of the present study included age, birth year, gender, body mass index, blood pressure, laboratory blood data such as total cholesterol, high-density lipoprotein cholesterol, aspartate aminotransferase, history of cancer, history of medication for HP, and urine anti-HP antibody status determined using an immunochromatography kit (RAPIRAN, Otsuka Pharmaceutical, Tokushima, Japan) [27–29]. For the baseline laboratory blood data analysis, peripheral blood was drawn in the morning using three 7-ml vacuum tubes after overnight fasting. Details of the procedures were described in our previous study [30]. The RAPIRAN is a de facto standard kit for anti-HP antibody detection in urine [31]. It has been reported that the sensitivity, specificity, and consistency of the RAPIRAN kit against an independent serum enzyme immunoassay E-plate (Eiken, Tokyo, Japan) was $86.2\%$, $93.7\%$, and $90.1\%$, respectively [31,32]. Moreover, an apparent birth-year effect to HP infection rate can be assessed in order to evaluate if the RAPIRAN kit has introduced a bias for anti-HP antibody detection in the current cohort.
## Outcome measures
The primary end point was all-cancer incidence, and all-cause death. The all-cancer incidence was further categorized into cancer types when appropriate. Analysis of individual cancer types included those of asynchronous secondary cancer. The classification of cancer types was defined using the International Statistical Classification of Diseases for Oncology, Third Edition (ICD-O-III). The vital and residential status of participants was determined by using the population registry. For logistical reasons, we ended the follow-up of subjects who had moved out of the study area (Nagoya city). Causes of death were confirmed based on the vital statistics data provided by the Ministry of Health, Labor and Welfare, Japan. Cancer incidence data was obtained from the Aichi Cancer Registry.
## Statistical analysis
In principle, all variables were compared based on the anti-HP antibody status defined in a binary manner (i.e., presence of antibody, HP-positive, HP+; absence of antibody, HP-negative, HP-). Categorical and continuous variables were analyzed using Fisher’s exact test and Wilcoxon’s rank sum test, respectively. Subject matching was performed to correct for birth year.
Crude rates of cancer incidence and death are shown as the number of cancer incident cases and deaths per 1,000 person-years, respectively. The crude rates were compared between HP+ and HP- using the Wald test.
To minimize the effects of a lack of background uniformity when comparing cancer incidence/death and anti-HP antibody status, the following procedure was performed when necessary: (a) Exclusion of participants who had a case history of any type of cancer; and (b) frequency matching according to birth year by gender. The proportion of event-free survival was estimated using the Kaplan-Meier method. A log-rank test was used to examine the null hypothesis whereby there was no difference in the proportion of event-free survival between the populations in terms of the probability of an event at any time point. A Cox proportional hazards model was used to estimate the hazards ratio (HR). The follow-up period was computed from the baseline survey to cancer incidence (in the analysis for cancer incidence), death of any cause, moving out of the study area, and the end of follow-up (Dec 31, 2016). The model included age, sex (1: male, 2: female), smoking status (1: current or ever, 0: never), and drinking habits (1: current or ever, 0: never). P-value <0.05 was considered statistically significant. All statistical analyses were performed using R.
## Participant characteristics
The purpose of the parental cohort, the J-MICC study, was to confirm and detect gene-environment interactions of lifestyle-related diseases, mainly of cancer, through the cohort analysis. It included a cross-sectional analysis of lifetime factors, biomarkers, and genotypes as well as confirmation/screening of new biomarkers that can be used for early diagnosis of cancer [21]. In the present study, however, we exclusively focused on investigating whether HP infection status was associated with an all-cancer death rate. The population of the DAIKO cohort study that was eligible for the present study ($$n = 4$$,982) included 1,416 men and 3,566 women, with a median age of 53. Results of urine anti-HP antibody tests were available for 4,981 out of 4,982 ($99.98\%$) participants. There were 1,825 ($37\%$) HP+ individuals and 3,156 ($63\%$) HP- individuals. The summary of participant characteristics is shown in Table 1. Individuals who were HP+ tended to be male and older relative to HP- individuals. HP+ status was also associated with higher BMI and history of smoking. Although the absolute difference was small, all clinical and laboratory continuous variables tested exhibited higher risk factors for major chronic diseases for HP+ compared to HP- individuals (S1 Table) [33,34].
**Table 1**
| Unnamed: 0 | Unnamed: 1 | HP+ (n = 1,825) | HP+ (n = 1,825).1 | HP- (n = 3,156) | HP- (n = 3,156).1 | Unnamed: 6 |
| --- | --- | --- | --- | --- | --- | --- |
| Variable | Value | N | % | n | % | P value |
| Sex | Male | 562 | 30.8 | 854 | 27.1 | 0.00506 |
| Sex | Female | 1263 | 69.2 | 2302 | 72.9 | 0.00506 |
| Age | Median(1st-3rd quartile) | 58.6 (49.3, 64.7) | 58.6 (49.3, 64.7) | 50.3 (42.4, 60.5) | 50.3 (42.4, 60.5) | 3.47 x 10−60 |
| BMIa | Median(1st-3rd quartile) | 21.5 (19.7, 23.8) | 21.5 (19.7, 23.8) | 21.1 (19.4, 23.3) | 21.1 (19.4, 23.3) | 0.0000874 |
| Smokingb | Yes | 609 | 33.4 | 962 | 30.5 | 0.034 |
| Smokingb | No | 1215 | 66.6 | 2194 | 69.5 | 0.034 |
| Drinking | Yes | 1009 | 55.3 | 1795 | 56.9 | 0.273 |
| Drinking | No | 816 | 44.7 | 1360 | 43.1 | 0.273 |
## HP eradication
HP is classified as a Group 1 carcinogen by the World Health Organization. Symptoms frequently involved in HP infection include nausea, epigastralgia, and heartburn. These symptoms may ultimately be considered to be triggers for peptic ulcers and gastric cancer. To reduce these symptoms and risk, HP eradication therapy has been conducted in Japan [35]. The chronological change in anti-HP antibodies has been considered to be a potential predictor of successful HP eradication therapy, although the immunological significance of antibody titer remains to be elucidated [36]. A recent study from a large-scale cohort study in Japan suggested that the majority of patients who received eradication therapy had reduced anti-HR antibody levels within one year [36]. In the present study, we compared the measured urine anti-HP antibody levels and self-reported HP eradication history. Nearly $90\%$ of participants had never been assessed for HP prior to enrolling in the DAIKO study. Therefore, the majority of participants reflect the current HP infection status. Among the 472 ($9.5\%$) participants who had undergone an HP examination (method unknown), 314 ($66.5\%$) were positive, and 133 ($28.2\%$) were negative. The status of 25 individuals ($5.3\%$) was unknown or no credible information available. Among the previously-diagnosed HP+ participants ($$n = 314$$), 152 ($48.4\%$) were anti-HP antibody positive (HP+) while 162 ($51.6\%$) were HP- at the DAIKO baseline. Among the previously-diagnosed HP- participants ($$n = 133$$), 26 ($19.5\%$) were HP+ while 107 ($80.5\%$) were HP- at DAIKO baseline. Thus, the consistency rate was $57.7\%$ among any previous HP exam and the present urine anti-HP antibody test. It should be noted that HP- participants at the DAIKO baseline could include those for whom HP was successfully eradicated. In fact, among the previously-HP+ participants, 162 ($51.6\%$) had successful eradication. The self-claimed “successful eradication” resulted in 53 ($33.1\%$) anti-HP antibody positive subjects, whereas 109 ($66.9\%$) were negative at baseline. No definitive time point for eradication was available for the present cohort. Of note, if the eradication therapy was conducted within one year, one to five years, and more than 6 years before the cohort enrollment, it was reported that anti-HP antibody titer was decreased by $76.8\%$, $88.2\%$, and $91.5\%$, respectively [36]. Furthermore, information about the antibiotics used for the eradication was not available.
## Birth year effect
In addition to biological/epidemiological information, the birth-year effect on HP infection rate may confirm that the reliability of the HP infection detection kit (i.e., RAPIRAN) was reasonable. It is known that the HP infection rate in the general population has been decreasing since the 1940s [37]. We thought the birth-year HP infection rate bias may be substantial, and therefore it should be used for statistical matching. The number of HP+ individuals categorized by birth year was 76 ($1.5\%$ of all HP+ cases) for 1938–1939, 876 ($17.6\%$) for 1940–1944, 793 ($15.9\%$) for 1945–1949, 617 ($12.4\%$) for 1950–1954, 645 ($12.9\%$) for 1955–1959, 706 ($14.2\%$) for 1960–1964, 629 ($12.6\%$) for 1965–1969, 630 ($12.6\%$) for 1970–1974, and 9 ($0.18\%$) for 1975. There was a clear trend that lower birth year was associated with higher rate of HP+ (Fig 1). The mean HP+ fraction by birth year 1938–1944 was 0.58, whereas that for 1970–1975 was 0.21. The fraction of females with HP+ and birth year 1938–1944 was 0.45, whereas that for 1970–1975 was 0.15. The HP+ fraction for men was higher than that for women across the birth years. The present data seem consistent with previous data from the Japanese population [37]. Overall, we conclude that information obtained from the RAPIRAN kit was reasonable with minimal bias.
**Fig 1:** *Anti-HP antibody status by birth year.The HP+ rate is shown with five-year bins for male participants (upper panel, n = 1,416: 1938–1944, n = 307; 1945–1949, n = 245; 1950–1954, n = 187; 1955–1959, n = 175; 1960–1964, n = 180; 1965–1969, n = 157; 1970–1975, n = 165) and female participants (bottom panel, n = 3,565; 1938–1944, n = 645; 1945–1949, n = 548; 1950–1954, n = 430; 1955–1959, n = 470; 1960–1964, n = 526; 1965–1969, n = 472; 1970–1975, n = 474) according to birth year. Black areas in each bar represent the HP+ fraction.*
## Matching
Matching is a technique used to avoid confounding in a study design. The HP+ participants were frequency-matched to HP- respondents for five-year bins of birth year and gender to minimize confounding factors. Those subjects who had a previous history of cancer were excluded. With birth year matching, the categorical patient characteristics showed no significant difference between HP+ and HP- individuals (Table 2). Moreover, after matching, almost all continuous variables showed no difference between HP+ and HP- individuals except for the HDL cholesterol level for which the estimated mean difference was 3 mg/dl (S2 Table).
**Table 2**
| Unnamed: 0 | Unnamed: 1 | HP+ (n = 1,688) | HP+ (n = 1,688).1 | HP- (n = 1,688) | HP- (n = 1,688).1 | Unnamed: 6 |
| --- | --- | --- | --- | --- | --- | --- |
| Variable | Value | n | % | n | % | P value |
| Sex | Male | 489 | 29.0 | 489 | 29.0 | 1.00 |
| Sex | Female | 1199 | 71.0 | 1199 | 71.0 | 1.00 |
| Age (yrs) | Median(1st-3rd quartile) | 57.9(48.6, 64.1) | 57.9(48.6, 64.1) | 57.8(48.8, 63.9) | 57.8(48.8, 63.9) | 0.866 |
| BMIa | Median(1st-3rd quartile) | 21.5(19.7, 23.8) | 21.5(19.7, 23.8) | 21.4(19.7, 23.5) | 21.4(19.7, 23.5) | 0.37 |
| Smokingb | Yes | 545 | 32.3 | 522 | 30.9 | 0.395 |
| Smokingb | No | 1142 | 67.7 | 1166 | 69.1 | 0.395 |
| Drinking | Yes | 920 | 54.5 | 968 | 57.3 | 0.103 |
| Drinking | No | 768 | 45.5 | 720 | 42.7 | 0.103 |
## Cancer incidence
Using the data generated by matching, we first analyzed cancer incidence tabulated with HP infection status. It is clear that gastric cancer has the highest impact from HP infection, so we analyzed individuals with gastric and non-gastric cancer in the present cohort. The non-gastric cancer types were chosen based on the number of incidents that are relatively large among available data sets. Overall cancer incidence was 105 and 67 in HP+ ($$n = 1688$$) and HP- ($$n = 1688$$) individuals, respectively ($$p \leq 0.003$$, Wald test). The individual cancer types that had significantly higher incidence in the HP+ population were both gastric (20 and 5 in HP+ and HP- individuals, respectively) and non-gastric (87 and 62 in HP+ and HP- individuals, respectively). Other cancer types including uterine, lung, prostate, colon, and breast showed no significant difference in cancer incidence in terms of HP infection status, although the number of cases was too small to reach a definitive conclusion about the effect of HP status. The all-cause death incidence in terms of HP status also showed no significant difference. The cancer incidence after matching is shown in S3 Table. These results suggest that HP infection may increase the risk for a wide range of cancer, although the risk for gastric cancer is more pronounced than other cancer types.
## Survival analysis
We next analyzed the cancer incidence and death rate using the Kaplan-Meier method. The crude cumulative all-cancer incidence rate was higher for HP+ individuals than that for HP- ($$n = 3$$,376, $$P \leq 0.00328$$; Fig 2).
**Fig 2:** *All cancer incidence.Kaplan-Meier estimation of all cancer incidence detected in this study (i.e., gastric, uterine, lung, prostate, colon, and breast cancer) stratified by anti-HP antibody status. The black line indicates HP+ and the gray line indicates HP-. Panel B is a partial enlarged view of panel A. A, vertical axis range 0.00–1.00; B, vertical axis range 0.90–1.00.*
As expected, the incidence of gastric cancer was significantly higher in HP+ individuals than those who were HP- ($$p \leq 0.0027$$; Fig 3A and 3B). Interestingly, however, the incidence of non-gastric cancer was also significantly higher in HP+ individuals than those who were HP- ($$p \leq 0.039$$; Fig 3C and 3D). Consistent with the above tabulated analysis, this result suggests that HP infection may increase the risk for a wide range of cancers, although the risk for gastric cancer is more pronounced than other cancer types. The cumulative cancer incidence was also assessed for lung, colorectal, rectum, prostate, and breast cancer, although the number was limited (S1 Fig).
**Fig 3:** *Gastric and non-gastric cancer incidence.Kaplan-Meier estimation of gastric (A, B) and non-gastric (C, D) cancer incidence stratified by anti-HP antibody status. Non-gastric cancer includes all those who started with ICD-O-III (i.e., not limited to uterine, lung, prostate, colon, and breast cancers). The black line indicates HP+ and the gray line indicates HP-. Panels B and D is a partially enlarged view of panels A and C, respectively. A and C, vertical axis range 0.00–1.00; B and D, vertical axis range 0.90–1.00.*
The cumulative all-cause death rate was not significantly different between HP+ and HP- individuals ($$p \leq 0.972$$; Fig 4A and 4B). It is important to note that the cancer incidence was higher in the HP+ population than the HP- population for both gastric and non-gastric cancer. Therefore, the finding that there was no significant difference between the HP+ and HP- populations in overall survival suggests that the HP+ population has a lower risk for death than that of HP- population. In addition, death related to cancer almost always occurs in advanced stages, whereas cancer incidence is generally measured at early stages. Therefore, the deceased population should have received some treatment procedures. One possible hypothesis is that potential systemic effects of HP infection may contribute to a reduced likelihood of death in the context of advanced cancer treatment. Although a limited number of deaths were recorded ($$n = 33$$ out of 3,376 matching cases), cancer-specific deaths did not differ between HP+ and HP- individuals ($$n = 3$$,376, $$p \leq 0.888$$; Fig 4C and 4D).
**Fig 4:** *All-cause and cancer-specific deaths.Kaplan-Meier estimation of all-cause deaths (A, B) and deaths due to cancer (C, D) stratified by anti-HP antibody status. Cancer type includes gastric, uterine, lung, prostate, colon, and breast. The black line indicates HP+ and the gray line indicates HP-. Panels B and D are partially enlarged views of panels A and C, respectively. A and C, vertical axis range 0.00–1.00; B and D, vertical axis range 0.90–1.00.*
To identify independent predictive variables for the risk of incidence for all-cancer, gastric, and non-gastric cancer, a Cox proportional-hazards model was applied with variables including age, sex, drinking, smoking, and HP infection. The hazards ratio of all-cancer incidence by HP was 1.59 ($$p \leq 0.00297$$), which was lower than the hazard ratio for smoking (1.97; $$p \leq 0.000422$$) but higher than that for alcohol consumption (1.24; $$p \leq 0.216$$, S4 Table). The hazards ratio for the gastric cancer incidence was 3.93 ($$p \leq 0.00631$$, S5 Table), whereas that for non-gastric cancer incidence in HP+ over HP- was 1.42 ($$p \leq 0.0356$$, S6 Table). However, the number of gastric cancer cases was 20 and 5 in HP+ ($$n = 1$$,688) and HP- ($$n = 1$$,688) individuals, respectively, which was too small to allow additional modeling with respect to other cancer types in the current study observation period.
## Discussion
In our study cohort from the DAIKO study, we demonstrate that the cancer incidence of the HP+ population was significantly higher than the HP- population. However, the all-cancer death rate for the HP+ population was similar to that for the HP- population. Two possible hypotheses can be drawn from these observations. First, the HP+ population appears to be more susceptible to cancer. Second, HP+ populations may have higher response or sensitivity to treatment for advanced cancer. Important premises can explain these hypotheses including establishment of HP+ infection early in life [38–41] and therapy for advanced cancer in *Japan is* almost always attempted before death as part of coverage by the Japanese universal health insurance system, kaihoken [42]. Therefore, the present results suggest that HP infection may have triggered acquisition of innate immunity that is beneficial for advanced cancer therapy [13].
However, the notion that innate immunity is ultimately relevant to death and plays a role in advanced cancer treatments has been poorly presented to date. In the present study, non-gastric cancer types, including (but not limited to) uterine, lung, prostate, colon, and breast were studied to evaluate whether HP infection was associated with cancer incidence as well as death rate. In fact, there has not been definitive evidence directly linking HP infection with cancer death [43–47]. Due to limitations on data access in this study, we only showed that the HP+ population had a higher cancer incidence rate than the HP- population, whereas there was no difference in the death rate. These results suggest that HP+ population has a lower risk for death after the initial diagnosis of cancer. Importantly, almost all cancer patients are treated after diagnosis, and therefore the HP+ population may have better treatment outcomes. Although it remains to be elucidated if HP infection may influence treatment efficacy, it may be of a practical hallmark to stratify treatment indications.
After recognition of a pathogen, one of the ultimate purposes of disease control is to reduce deaths caused by the pathogen. For gastric cancer, eradication of HP has been considered to reduce a variety of gastric diseases including gastritis, peptic ulcer, and potentially gastric cancer [48–50]. However, the role of HP in gastric carcinogenesis may be limited to the development of “early” gastric cancer, in which normal mucosal cells transform to a malignant phenotype [51,52]. More advanced-stage gastric cancer cases have substantial invasion beyond the submucosae where cancer cell division is no longer regulated by the malignant signal transduction induced by HP, which has been well presented in the “hit-and-run” model [53].
In Japan, the 5-year disease-free survival of pT1N0M0 gastric cancer is $99\%$, and that for pT1N1M0/pT2N0M0 is $94\%$ and $92\%$, respectively [54]. Thus, early gastric cancer may not be prioritized in terms of efforts to reduce gastric cancer-deaths. It is clear that nearly all gastric cancer-deaths are caused by advanced stage or recurrent gastric cancer. Therefore, the beneficial effect of HP in these gastric cancer patients should be considered differently from the simple malignant transformation of gastric mucosal cells. Here we suggest that stratification of treatment for these advanced, including non-gastric, cancer patients by HP infection status in which promoting therapy or maintaining alternative regimens for the HP- population might be a prompt practical option.
The prevalence of HP in some regions of the world remains high [55]. The eradication of HP appears to be associated with a reduced incidence of gastric cancer [56]. It is, however, important to consider the fact that gastric cancer death rate has been continuously decreasing in many ethnic groups and countries including Japan since 1970s, without a concerted nationwide eradication program [37,57–60]. In the present study, we demonstrated that anti-HP antibody status was associated with birth year in that those having later birth year had a lower HP+ rate. These findings suggest that the role of HP in cancer treatment is one of the treatment stratification indices for advanced gastric cancer patients in most countries, whereas the number of gastric cancer deaths will likely continue to decrease naturally without efforts for HP eradication.
The present study has some limitations. The observational period is eight years, and the mean age at baseline was 58. A longer observation period would increase the cancer incidence and number of deaths. In fact, the number of all-cancer deaths may be too small to be conclusively analyzed as an independent parameter. We would have more evaluable cases in terms of tumor-type specific HP effect other than gastric cancer. The HP eradication history was reported by individual participants as personal medical records were not fully accessible for the present cohort. The HP infection was evaluated by only testing for anti-HP antibody in the urine. The anti-HP antibody status may be considered to reflect past or present HP infection, but no information about whether urine antibody levels had naturally attenuated was available. Therefore, this parameter might be less robust than extracting positive results from multiple HP examinations. However, past reports for HP infection rate that took birth-year into account were consistent with our birth-year effect observations (i.e., earlier birth year was associated with higher HP+ rates). Although our current HP infection status definition could have been affected by our detection kit (i.e., RAPIRAN kit), this consistency supports that the anti-HP antibody detection method using the RAPIRAN kit may have minimized the possible bias to the actual HP infection status. The HP strain was not assessed, although most HP strains in Japan are CagA+ [61]. Depending on the practicality and objectives, detailed characterization of HP may be warranted in future studies. Moreover, the cohort subject is based on non-patient individuals, not a patient cohort. Hence, information such as treatment details, cancer stage, and molecular characteristics were not available.
Although details on the participants (i.e., performance status) were not accessible, the present finding has been well supported by our recent study demonstrating that HP+ post-gastrectomy patients whose Programmed Death-Ligand1 (PD-L1) status was negative had significantly better survival than those who were HP- after adjuvant chemotherapy using S-1 [12]. That study was limited to advanced gastric cancer patients, but a systemic effect may have been involved in the treatment response. Therefore, our current results showing that the HP+ population had better overall survival may be reasonably applied to multiple cancer types.
HP infection is associated with susceptibility to diverse types of cancer. However, HP infection has a favorable survival effect for patients with advanced cancer. With previous reports showing a better prognosis of advanced gastric cancer patients who are HP+ than those for HP- patients from various regions of the world [2–8,11] where the therapeutic regimens are diverse, HP infection may be a predominant factor for survival of not only patients with gastric cancer, but also for those with other types of cancer. In conclusion, our present results may suggest that HP infection may reduce the risk for death after an initial diagnosis of cancer.
## References
1. **Schistosomes, liver flukes and Helicobacter pylori. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Lyon, 7–14 June 1994**. *IARC Monogr Eval Carcinog Risks Hum* (1994) **61** 1-241. PMID: 7715068
2. Choi IK, Sung HJ, Lee JH, Kim JS, Seo JH. **The relationship between Helicobacter pylori infection and the effects of chemotherapy in patients with advanced or metastatic gastric cancer**. *Cancer Chemother Pharmacol* (2012) **70** 555-8. DOI: 10.1007/s00280-012-1944-5
3. Kang SY, Han JH, Ahn MS, Lee HW, Jeong SH, Park JS. **Helicobacter pylori infection as an independent prognostic factor for locally advanced gastric cancer patients treated with adjuvant chemotherapy after curative resection**. *Int J Cancer* (2012) **130** 948-58. DOI: 10.1002/ijc.26081
4. Lee WJ, Lin JT, Shun CT, Lee WC, Yu SC, Lee PH. **Comparison between resectable gastric adenocarcinomas seropositive and seronegative for Helicobacter pylori**. *Br J Surg* (1995) **82** 802-5. DOI: 10.1002/bjs.1800820627
5. Marrelli D, Pedrazzani C, Berardi A, Corso G, Neri A, Garosi L. **Negative Helicobacter pylori status is associated with poor prognosis in patients with gastric cancer**. *Cancer* (2009) **115** 2071-80. DOI: 10.1002/cncr.24253
6. Meimarakis G, Winter H, Assmann I, Kopp R, Lehn N, Kist M. **Helicobacter pylori as a prognostic indicator after curative resection of gastric carcinoma: a prospective study**. *Lancet Oncol* (2006) **7** 211-22. DOI: 10.1016/S1470-2045(06)70586-1
7. Postlewait LM, Squires MH, Kooby DA, Poultsides GA, Weber SM, Bloomston M. **Preoperative Helicobacter pylori Infection is Associated with Increased Survival After Resection of Gastric Adenocarcinoma**. *Ann Surg Oncol* (2016) **23** 1225-33. DOI: 10.1245/s10434-015-4953-x
8. Yoon H, Kim N, Lee HS, Shin CM, Park YS, Lee DH. **Helicobacter pylori-negative gastric cancer in South Korea: incidence and clinicopathologic characteristics**. *Helicobacter* (2011) **16** 382-8. DOI: 10.1111/j.1523-5378.2011.00859.x
9. Sakuramoto S, Sasako M, Yamaguchi T, Kinoshita T, Fujii M, Nashimoto A. **Adjuvant chemotherapy for gastric cancer with S-1, an oral fluoropyrimidine**. *N Engl J Med* (2007) **357** 1810-20. DOI: 10.1056/NEJMoa072252
10. Sasako M, Sakuramoto S, Katai H, Kinoshita T, Furukawa H, Yamaguchi T. **Five-year outcomes of a randomized phase III trial comparing adjuvant chemotherapy with S-1 versus surgery alone in stage II or III gastric cancer**. *J Clin Oncol* (2011) **29** 4387-93. DOI: 10.1200/JCO.2011.36.5908
11. Nishizuka SS, Tamura G, Nakatochi M, Fukushima N, Ohmori Y, Sumida C. **Helicobacter pylori infection is associated with favorable outcome in advanced gastric cancer patients treated with S-1 adjuvant chemotherapy**. *J Surg Oncol* (2018) **117** 947-56. DOI: 10.1002/jso.24977
12. Koizumi Y, Ahmad S, Ikeda M, Yashima-Abo A, Espina G, Sugimoto R. **Helicobacter pylori Modulated Host Immunity in Gastric Cancer Patients With S-1 Adjuvant Chemotherapy**. *J Natl Cancer Inst* (2022) **114** 1149-58. DOI: 10.1093/jnci/djac085
13. Mantovani A, Netea MG. **Trained Innate Immunity, Epigenetics, and Covid-19**. *N Engl J Med* (2020) **383** 1078-80. DOI: 10.1056/NEJMcibr2011679
14. van Puffelen JH, Keating ST, Oosterwijk E, van der Heijden AG, Netea MG, Joosten LAB. **Trained immunity as a molecular mechanism for BCG immunotherapy in bladder cancer**. *Nat Rev Urol* (2020) **17** 513-25. DOI: 10.1038/s41585-020-0346-4
15. Okamoto H, Hayaishi O. **Flavin adenine dinucleotide requirement for kynurenine hydroxylase of rat liver mitochondria**. *Biochem Biophys Res Commun* (1967) **29** 394-9. DOI: 10.1016/0006-291x(67)90469-x
16. Oba MS, Teramukai S, Ohashi Y, Ogawa K, Maehara Y, Sakamoto J. **The efficacy of adjuvant immunochemotherapy with OK-432 after curative resection of gastric cancer: an individual patient data meta-analysis of randomized controlled trials**. *Gastric Cancer* (2016) **19** 616-24. DOI: 10.1007/s10120-015-0489-9
17. Ananth AA, Tai LH, Lansdell C, Alkayyal AA, Baxter KE, Angka L. **Surgical Stress Abrogates Pre-Existing Protective T Cell Mediated Anti-Tumor Immunity Leading to Postoperative Cancer Recurrence.**. *PLoS One* (2016) **11** e0155947. DOI: 10.1371/journal.pone.0155947
18. Block MS, Dietz AB, Gustafson MP, Kalli KR, Erskine CL, Youssef B. **Th17-inducing autologous dendritic cell vaccination promotes antigen-specific cellular and humoral immunity in ovarian cancer patients**. *Nat Commun* (2020) **11** 5173. DOI: 10.1038/s41467-020-18962-z
19. Krall JA, Reinhardt F, Mercury OA, Pattabiraman DR, Brooks MW, Dougan M. **The systemic response to surgery triggers the outgrowth of distant immune-controlled tumors in mouse models of dormancy**. *Sci Transl Med* (2018) **10**. DOI: 10.1126/scitranslmed.aan3464
20. Ott PA, Hu Z, Keskin DB, Shukla SA, Sun J, Bozym DJ. **An immunogenic personal neoantigen vaccine for patients with melanoma**. *Nature* (2017) **547** 217-21. DOI: 10.1038/nature22991
21. Hamajima N, Group JMS. **The Japan Multi-Institutional Collaborative Cohort Study (J-MICC Study) to detect gene-environment interactions for cancer**. *Asian Pac J Cancer Prev* (2007) **8** 317-23. PMID: 17696755
22. Hishida A, Nakatochi M, Akiyama M, Kamatani Y, Nishiyama T, Ito H. **Genome-Wide Association Study of Renal Function Traits: Results from the Japan Multi-Institutional Collaborative Cohort Study**. *Am J Nephrol* (2018) **47** 304-16. DOI: 10.1159/000488946
23. Naito M, Eguchi H, Okada R, Ishida Y, Nishio K, Hishida A. **Controls for monitoring the deterioration of stored blood samples in the Japan Multi-Institutional Collaborative Cohort Study (J-MICC Study).**. *Nagoya J Med Sci* (2008) **70** 107-15. PMID: 18954029
24. Okada R, Kawai S, Naito M, Hishida A, Hamajima N, Shinchi K. **Matrix metalloproteinase-9 gene polymorphisms and chronic kidney disease.**. *Am J Nephrol* (2012) **36** 444-50. DOI: 10.1159/000343742
25. Wakai K, Hamajima N, Okada R, Naito M, Morita E, Hishida A. **Profile of participants and genotype distributions of 108 polymorphisms in a cross-sectional study of associations of genotypes with lifestyle and clinical factors: a project in the Japan Multi-Institutional Collaborative Cohort (J-MICC) Study.**. *J Epidemiol* (2011) **21** 223-35. DOI: 10.2188/jea.je20100139
26. Yamaguchi M, Uemura H, Arisawa K, Katsuura-Kamano S, Hamajima N, Hishida A. **Association between brain-muscle-ARNT-like protein-2 (BMAL2) gene polymorphism and type 2 diabetes mellitus in obese Japanese individuals: A cross-sectional analysis of the Japan Multi-institutional Collaborative Cohort Study**. *Diabetes Res Clin Pract* (2015) **110** 301-8. DOI: 10.1016/j.diabres.2015.10.009
27. Yamamoto S, Uemura N, Okamoto S, Yamaguchi S, Mashiba H, Tachikawa T. **A new rapid test for detecting anti-Helicobacter pylori antibody excreted into urine**. *Helicobacter* (2000) **5** 160-4. DOI: 10.1046/j.1523-5378.2000.00025.x
28. Graham DY, Reddy S. **Rapid detection of anti-Helicobacter pylori IgG in urine using immunochromatography**. *Aliment Pharmacol Ther* (2001) **15** 699-702. DOI: 10.1046/j.1365-2036.2001.00968.x
29. Fujisawa T, Kaneko T, Kumagai T, Akamatsu T, Katsuyama T, Kiyosawa K. **Evaluation of urinary rapid test for Helicobacter pylori in general practice**. *J Clin Lab Anal* (2001) **15** 154-9. DOI: 10.1002/jcla.1019
30. Morita E, Hamajima N, Hishida A, Aoyama K, Okada R, Kawai S. **Study profile on baseline survey of Daiko Study in the Japan Multi-Institutional Collaborative Cohort Study (J-MICC Study).**. *Nagoya J Med Sci* (2011) **73** 187-95. PMID: 21928700
31. Yamamoto T, Ishii T, Kawakami T, Sase Y, Horikawa C, Aoki N. **Reliability of urinary tests for antibody to Helicobacter pylori in patients with pulmonary tuberculosis**. *World J Gastroenterol* (2005) **11** 412-4. DOI: 10.3748/wjg.v11.i3.412
32. Kawai T, Kawakami K, Kudo T, Ogiahara S, Handa Y, Moriyasu F. **A new serum antibody test kit (E plate) for evaluation of Helicobacter pylori eradication**. *Intern Med* (2002) **41** 780-3. DOI: 10.2169/internalmedicine.41.780
33. Khera AV, Emdin CA, Drake I, Natarajan P, Bick AG, Cook NR. **Genetic Risk, Adherence to a Healthy Lifestyle, and Coronary Disease**. *N Engl J Med* (2016) **375** 2349-58. DOI: 10.1056/NEJMoa1605086
34. Li Y, Sperrin M, Ashcroft DM, van Staa TP. **Consistency of variety of machine learning and statistical models in predicting clinical risks of individual patients: longitudinal cohort study using cardiovascular disease as exemplar**. *BMJ* (2020) **371** m3919. DOI: 10.1136/bmj.m3919
35. Asaka M, Kato M, Sakamoto N. **Roadmap to eliminate gastric cancer with Helicobacter pylori eradication and consecutive surveillance in Japan**. *J Gastroenterol* (2014) **49** 1-8. DOI: 10.1007/s00535-013-0897-8
36. Tanaka S, Goto A, Yamagishi K, Iwasaki M, Yamaji T, Shimazu T. **Long-term response of Helicobacter pylori antibody titer after eradication treatment in middle-aged Japanese: JPHC-NEXT Study.**. *J Epidemiol* (2021). DOI: 10.2188/jea.JE20200618
37. Inoue M.. **Changing epidemiology of Helicobacter pylori**. *Japan. Gastric Cancer* (2017) **20** 3-7. DOI: 10.1007/s10120-016-0658-5
38. Malaty HM, El-Kasabany A, Graham DY, Miller CC, Reddy SG, Srinivasan SR. **Age at acquisition of Helicobacter pylori infection: a follow-up study from infancy to adulthood**. *Lancet* (2002) **359** 931-5. DOI: 10.1016/S0140-6736(02)08025-X
39. Okuda M, Miyashiro E, Booka M, Tsuji T, Nakazawa T. **Helicobacter pylori colonization in the first 3 years of life in Japanese children**. *Helicobacter* (2007) **12** 324-7. DOI: 10.1111/j.1523-5378.2007.00510.x
40. Rothenbacher D, Inceoglu J, Bode G, Brenner H. **Acquisition of Helicobacter pylori infection in a high-risk population occurs within the first 2 years of life**. *J Pediatr* (2000) **136** 744-8. PMID: 10839870
41. Thomas JE, Dale A, Harding M, Coward WA, Cole TJ, Weaver LT. **Helicobacter pylori colonization in early life**. *Pediatr Res* (1999) **45** 218-23. DOI: 10.1203/00006450-199902000-00010
42. Reich MR, Ikegami N, Shibuya K, Takemi K. **50 years of pursuing a healthy society in Japan**. *Lancet* (2011) **378** 1051-3. DOI: 10.1016/S0140-6736(11)60274-2
43. Butt J, Epplein M. **Helicobacter pylori and colorectal cancer-A bacterium going abroad?**. *PLoS Pathog* (2019) **15** e1007861. DOI: 10.1371/journal.ppat.1007861
44. Deng B, Li Y, Zhang Y, Bai L, Yang P. **Helicobacter pylori infection and lung cancer: a review of an emerging hypothesis**. *Carcinogenesis* (2013) **34** 1189-95. DOI: 10.1093/carcin/bgt114
45. Fang CW, Chen CH, Muo CH, Wu SC. **Risk of subsequent prostate cancer in peptic ulcer patients who received helicobacter pylori eradication therapy: an Asian population-based cohort study**. *BMC Urol* (2020) **20** 135. DOI: 10.1186/s12894-020-00706-2
46. Franceschi F, Zuccala G, Roccarina D, Gasbarrini A. **Clinical effects of Helicobacter pylori outside the stomach**. *Nat Rev Gastroenterol Hepatol* (2014) **11** 234-42. DOI: 10.1038/nrgastro.2013.243
47. O’Connor A, O’Morain CA, Ford AC. **Population screening and treatment of Helicobacter pylori infection**. *Nat Rev Gastroenterol Hepatol* (2017) **14** 230-40. DOI: 10.1038/nrgastro.2016.195
48. Amieva M, Peek RM. **Pathobiology of Helicobacter pylori-Induced Gastric Cancer**. *Gastroenterology* (2016) **150** 64-78. DOI: 10.1053/j.gastro.2015.09.004
49. Graham DY. **History of Helicobacter pylori, duodenal ulcer, gastric ulcer and gastric cancer**. *World J Gastroenterol* (2014) **20** 5191-204. DOI: 10.3748/wjg.v20.i18.5191
50. Wu JY, Lee YC, Graham DY. **The eradication of Helicobacter pylori to prevent gastric cancer: a critical appraisal**. *Expert Rev Gastroenterol Hepatol* (2019) **13** 17-24. DOI: 10.1080/17474124.2019.1542299
51. Marusawa H, Chiba T. **Helicobacter pylori-induced activation-induced cytidine deaminase expression and carcinogenesis**. *Curr Opin Immunol* (2010) **22** 442-7. DOI: 10.1016/j.coi.2010.06.001
52. Matsumoto Y, Marusawa H, Kinoshita K, Endo Y, Kou T, Morisawa T. **Helicobacter pylori infection triggers aberrant expression of activation-induced cytidine deaminase in gastric epithelium**. *Nat Med* (2007) **13** 470-6. DOI: 10.1038/nm1566
53. Hatakeyama M.. **Helicobacter pylori CagA and gastric cancer: a paradigm for hit-and-run carcinogenesis**. *Cell Host Microbe* (2014) **15** 306-16. DOI: 10.1016/j.chom.2014.02.008
54. Katai H, Ishikawa T, Akazawa K, Isobe Y, Miyashiro I, Oda I. **Five-year survival analysis of surgically resected gastric cancer cases in Japan: a retrospective analysis of more than 100,000 patients from the nationwide registry of the Japanese Gastric Cancer Association (2001–2007).**. *Gastric Cancer* (2018) **21** 144-54. DOI: 10.1007/s10120-017-0716-7
55. Eusebi LH, Zagari RM, Bazzoli F. **Epidemiology of Helicobacter pylori infection**. *Helicobacter* (2014) **19** 1-5. DOI: 10.1111/hel.12165
56. Lee YC, Chiang TH, Chou CK, Tu YK, Liao WC, Wu MS. **Association Between Helicobacter pylori Eradication and Gastric Cancer Incidence: A Systematic Review and Meta-analysis**. *Gastroenterology* (2016) **150** 1113-24. DOI: 10.1053/j.gastro.2016.01.028
57. Fitzmaurice C, Allen C, Barber RM, Barregard L, Bhutta ZA. **Global, Regional, and National Cancer Incidence, Mortality, Years of Life Lost, Years Lived With Disability, and Disability-Adjusted Life-years for 32 Cancer Groups, 1990 to 2015: A Systematic Analysis for the Global Burden of Disease Study**. *JAMA Oncol* (2017) **3** 524-48. DOI: 10.1001/jamaoncol.2016.5688
58. Hallowell BD, Endeshaw M, Senkomago V, Razzaghi H, McKenna MT, Saraiya M. **Gastric cancer mortality rates among US and foreign-born persons: United States 2005–2014.**. *Gastric Cancer* (2019) **22** 1081-5. DOI: 10.1007/s10120-019-00944-w
59. Herbreteau E, Jooste V, Hamza S, Lepage C, Faivre J, Bouvier AM. **Trends in the management of gastric cancer over a 32-year period: a French population-based study**. *Gastric Cancer* (2015) **18** 129-37. DOI: 10.1007/s10120-014-0342-6
60. Lim D, Ha M, Song I. **Trends in major cancer mortality in Korea, 1983–2012, with a joinpoint analysis.**. *Cancer Epidemiol* (2015) **39** 939-46. DOI: 10.1016/j.canep.2015.10.023
61. Suzuki H, Mori H. **Different Pathophysiology of Gastritis between East and West? An Asian Perspective.**. *Inflamm Intest Dis* (2016) **1** 123-8. DOI: 10.1159/000446301
|
---
title: Non-pharmacologic hypertension management barriers and recommendations by hypertensive
patients at Pentecost Hospital, Madina
authors:
- Evans Osei Appiah
- Susana Boateng Agyekum
- Amertil P. Ninon
- Cyndi Appiah
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022142
doi: 10.1371/journal.pgph.0000085
license: CC BY 4.0
---
# Non-pharmacologic hypertension management barriers and recommendations by hypertensive patients at Pentecost Hospital, Madina
## Abstract
The number of hypertension cases keeps rising worldwide. Africa is not exempted from the prevalence of hypertension. The Sub-Saharan region over the years has been recording high numbers of hypertension cases due to low consciousness, poor management and lack of control of urbanization. However, it has been established that hypertension as a condition can be managed by controlling familiar risk factors such as alcohol consumption, tobacco use, physical inactivity and intake of an unhealthy diet. The researchers, therefore, intend to explore the non-pharmacologic hypertension management barriers and recommendations by hypertensive patients at Pentecost Hospital, Madina. The researchers employed the qualitative exploratory-descriptive design using a purposive sampling technique to select 20 participants between the ages of 35–65, who met the inclusion criteria. Using a semi-structured interview guide, participants were engaged in 30–60 minutes of face-to-face interviews. The demography of the participants revealed that $60\%$ [12] were females, and $40\%$ [8] were also males. Participants reported that they visit the clinic once a week with a budget of hundred Ghana Cedis to five hundred Ghana Cedis (100–500 GHS). Two main themes and 7 subthemes emerged from the study analysis. The barriers identified include financial constraints, difficulty adjusting to lifestyle changes, personal factors (laziness, forgetfulness, stress), lack of motivation, and busy work schedules and limited time. Recommendations were also made to overcome the barriers which include follow ups by health care professionals, and advice to hypertensive and non-hypertensive patients. In conclusion, the study found that adherence to non-pharmacologic management of hypertension is greatly influenced by one’s finances, some personal factors and external influences. Hence, it is necessary address these factors and also to ensure effective follow-ups and reminders in order to improve adherence to the non-pharmacologic management of hypertension. Further studies can also be conducted to address other obstacles to non-pharmacologic hypertension management.
## Introduction
Recently, the number of hypertension cases has alarmingly risen worldwide. For instance, the rate of hypertension incidence in England, USA, and Canada was $30\%$, $29\%$, and $19.5\%$ respectively [1]. In this respect, it was noted that the average systolic blood pressure in England was higher than in the USA and Canada across all gender. Moreover, it was discovered in China that hypertension was on the rise as a total of $26.6\%$ of adults were found to be with the condition [2]. However, this condition can be avoided by controlling familiar risk factors such as alcohol consumption, tobacco use, physical inactivity and intake of an unhealthy diet [3–5].
A study conducted in Africa revealed that there was a high prevalence of hypertension. The pervasiveness of the condition was attributed to factors such as low consciousness, poor management and control in the urban population [6]. In Egypt, it was estimated that a population of about 15 million people will be affected with hypertension by the year 2014. This was as a result of defective health care practitioners and deficient health care systems which served as obstacles in the management of hypertension [7]. He added that $60\%$ of people with hypertension develop complications. He however confirmed that adequate provision of education by the physicians and workable healthcare facilities provided good channels in the management of hypertension. Nevertheless, it was established that only $25.8\%$ of countries in Africa have developed guidelines for the effective management of hypertension.
In Ghana, it was discovered that hypertension incidence varied between $2.4\%$ to $32.9\%$ among people who were 15–19 years and 45–49 years [8]. It was however predicted that there will be a $60\%$ increase in the year 2025 by 1.56 billion people in Ghana as compared to the year 2000 in which people with hypertension consisted of 972 million people [9]. Despite the increasing rate of hypertension in Ghana, poor control of hypertension was a familiar result [8, 10]. The poor control of hypertension in these studies was attributed to the deficiency of proper knowledge of hypertension, lack of skills and resources to manage and treat hypertension by some health systems, inadequate anti-hypertensive drugs, and long distances to the hospitals.
Patient’s reluctance to adhere to lifestyle modification was identified as the major barrier to non-pharmacologic management of hypertension [11]. In Nepal, a study discovered that people consider lifestyle modification (diet) difficult despite their awareness of the numerous benefits associated with it [12]. The frequently reported barrier by the participants in this research was a food-related issue; the taste and desire for a certain food. It was also noted that dietary habits formed over a lifetime were difficult to break from [13]. Hence, the objective of this study was to assess non-pharmacologic hypertension management barriers to help recommend ways of addressing these barriers in order to improve adherence of hypertension patients to the non-pharmaclogic hypertensive management and to reduce hypertension burden locally and internationally.
## Methods
The researchers employed the qualitative exploratory-descriptive design to aid in exploring participants views on the perceived barriers, and make recommendations of non-pharmacological management of hypertension among hypertensive patients. This method and design also allowed participants to share their views on obstacles preventing them from adhering to the non-pharmacolgic hypertension management. A semi-structured interview guide was the data collection tool formulated by the researchers based on the study objectives and the literature review to provide insight into problems and broaden understanding regarding non-pharmacological hypertension management. See details of the interview guide in S1 File attached.
Hypertensive patients of Pentecost Hospital, Madina were the target population for the study. The hospital is located in the Greater Accra Region of Ghana, and its Out-Patient Department (OPD) runs four days a week, from 8 a.m to 2 p.m. However, the hypertensive clinic is organized on Thursdays and Fridays. The criteria for inclusion were; patients who had been diagnosed with hypertension for at least three [3] months. This group of hypertensive patients were considered based on shared common management methods of hypertension. In addition, the sampled population included hypertensive patients who were 35 years or more and were fluent speakers of English and Twi languages as these were languages spoken by the researchers. This was to ensure smooth communication between the researchers and the participants and also to present accurate information as the result of the study. Hypertensive patients who were seriously ill, unconscious and with other co-morbidities affecting their ability to speak were excluded from the study.
Participants were purposively selected with the study objective in mind to ensure that only participants who meets the inclusion criteria as stated above are selected and in other to ensure the richness of the data collected. The sample size was based on data saturation. Saturation is frequently used for determination of sample size qualitative studies [14]. Saturation is a sample determination size in qualitative research where responses are repeated by participants and data becomes redundant. The researchers kept recruiting and interviewing participants until the stage where no new data/information was retrieved (when participants started repeating the same ideas). This was reached on the 20th participant.
The data which were audio recordings of the interview sessions were transcribed and later analysed using a content analysis framework. In order to have total control over the data to facilitate coding and searching of the meaning and patterns, the transcript was read and reread. Codes that were accumulated were tabulated and grouped as themes and subthemes.
The methodological rigor of this study was maintain by ensuring that the data was credible and dependable. In addition, the confirmability and transferability were also maintained. This was ensured in the study by being cognizance of ethical issues, conducting a pilot study using 3 participants, describing the study design, sampling methods and technique. The interview guide was also provided and the whole study was done in accordance to the objectives of the study. This was done to ensure the trustworthiness of the study. Also, there was due diligence on the part of the researchers to guarantee the safety of the participants during the interview process. Privacy was maintained by interviewing respondents independently in a private office. Pseudonyms were accorded to each respondent to maintain confidentiality throughout the data collection. The respondents were also given a detailed explanation of the study, its methodology, its sample, data analysis and the procedural guidelines.
## Ethical consideration
The researchers obtained an ethical clearance and permission letter from the Dodowa Health Research Centre Institutional Review Board (DHRCIRB/$\frac{35}{03}$/20). The ethical clearance obtained was submitted to the leadership of Pentecost Hospital, Madina to prove the authenticity of the study. Following the approval from the authorities of the hospital, Participants were contacted after permission was sought from the ward in-charges. The participants were briefed on the purpose, as well as their role in the study. The sampled population was informed about the purpose of the study. Written consent forms were given to the participants to sign before the data collection procedure started. The participants were assured of confidentiality and were also given the privilege to ask any questions about the study. The recorded data were saved on the researchers’ pen drive and laptop secured with a security code known to the researchers alone.
## Socio-demographic data
Twenty [20] participants with hypertension were interviewed. The participants were between the ages of thirty-five [35] and sixty-five [65]. Thirteen ($65\%$) were married, three ($15\%$) were widows, and four ($20\%$) were single. The English language was used for the interview. Ten ($50\%$) participants were government workers and ten ($50\%$) were self-employed (hairdresser, seamstress, shop-keeping, tailor). In terms of religion, the majority of the participants, 17 ($85\%$) were from Christian denominations with three ($15\%$) being Muslims. All the participants had some form of formal education:12 ($60\%$) people have tertiary education, and 5 ($25\%$) have only primary education. Participants resided in different locations in the Greater Accra Region of Ghana (Haatso, Adenta, Madina, Oyibi, Nungua, Amrahia, Legon, Frafraha, Shiashi, Amanfrom, and Okponglo). The demography of the participants revealed that females were 12, representing $60\%$, and the remaining$40\%$ [8] were males. The majority of the participants,$75\%$ reported that they visit the clinic once a week, with a budget ranged between one hundred and five hundred Ghana Cedis (100–500 GHS). The details of the socio-demographic characteristics of the participants are listed in Table 1.
**Table 1**
| Variable | Frequency (n = 20) | Per cent (%) |
| --- | --- | --- |
| Age group | | |
| 18-35-40 | 5.0 | 25.0 |
| 21-41-50 | 2.0 | 10.0 |
| 22-51-59 | 3.0 | 15.0 |
| 26-60-65 | 10.0 | 50.0 |
| Religion | | |
| Christian | 17.0 | 85.0 |
| Muslim | 3.0 | 15.0 |
| Traditionalist | 0.0 | 0.0 |
| Occupation | | |
| Government worker | 10.0 | 50.0 |
| Self-employed | 10.0 | 50.0 |
| Not working | 0.0 | 0.0 |
| Marital Status | | |
| Single | 4.0 | 20.0 |
| Married | 13.0 | 65.0 |
| widow | 3.0 | 15.0 |
| Educational status | | |
| Primary | 5.0 | 25.0 |
| Secondary | 3.0 | 15.0 |
| Tertiary | 12.0 | 60.0 |
| Gender | | |
| Females | 12.0 | 60.0 |
| Males | 8.0 | 40.0 |
| Others | 0.0 | 0.0 |
| Visit the clinic | | |
| Once weekly | 15.0 | 75.0 |
| Twice weekly | 5.0 | 25.0 |
| Monthly | 0.0 | 0.0 |
| Weekly budget | | |
| 100–199 | 10.0 | 50.0 |
| 200–300 | 5.0 | 25.0 |
| 301–500 | 5.0 | 25.0 |
| 500 and above | 0.0 | 0.0 |
## Themes and sub-themes
Two main themes and 7 sub-themes emerged from the study analysis. The two themes were perceived barriers of non-pharmacological management of hypertension, and recommendations by hypertensive patients. The themes and subthemes are shown in Table 2.
**Table 2**
| THEMES | SUBTHEMES | CODES |
| --- | --- | --- |
| Barriers | 1. financial constraints | Bar |
| Barriers | 2. difficulty adjusting to lifestyle changes | Bar |
| Barriers | 3. personal factors (laziness, forgetfulness, stress) | Bar |
| Barriers | 4. lack of motivation, and | Bar |
| Barriers | 5. busy work schedules and limited time | Bar |
| Recommendations to overcome barriers | 6. follow up by healthcare professionals | Rec |
| Recommendations to overcome barriers | 7. advice to hypertensive and non-hypertensive patients | Rec |
## Perceived barriers of non-pharmacological management of hypertension
There are five [5] sub-themes that emerged under this theme. These sub-themes—financial constraints, difficulty adjusting to lifestyle changes, personal factors (laziness, forgetfulness, stress), lack of motivation and busy work schedules and limited time–, our findings revealed, served as hindrances to the participants in their bid to follow a lifestyle modification regimen.
In the present study, the results indicated that one of the key factors inhibiting changes in lifestyle behaviour towards management of hypertension was monetary consideration. The participants in this study assert that fruits and vegetables in the Ghanaian markets are costly nowadays. This in effect prevents them from adding fruits and vegetables to their diet daily. They indicated that even though they are encouraged to change their diet, their income levels serve as an impediment. This corresponds and affirms a study conducted in Eritrea which revealed that the pricey fruits are not patronized though they are necessities in the diet of hypertensive patients [15]. Admittedly, the study discovered that some fruits are low-priced, yet the participants claimed they are unable to afford them due to meagre salaries and financial constraints. This is of concern since non-pharmacological approach such as dietary modification was discovered to play a significant role in controlling blood pressure [16].
Moreover, it was realized from the study that making healthy living a lifestyle is a major problem for most people. Most of the participants in the study confirmed their uncomfortable experiences whenever they try to integrate the various lifestyle changes into their daily lives. Adapted to the habit of living a sedentary life, it was difficult trying to overturn their lives to live healthy overnight. Some also conferred that certain lifestyles–meat intake and intake of Ghanaian salted fish–could not be altered easily due to addiction. Any attempt they make to adopt a healthy lifestyle proves futile, hence their resolution to accept their current state. For instance, this finding resonates with a similar study in Barbados, which reported dietary habits formed over a period of time is very difficult to break from [16]. Thus, people have always encountered difficulties in trying to adopt lifestyle modifications that can help control, and manage hypertensive conditions. However, adhering to these lifestyle practices is of major benefits since some authors have established significant relationship between hypertension and the unhealthy lifestyle practices [3–5].
Personal factors such as non-adherence to an exercise plan, stress, and lack of discipline are accounted as barriers to lifestyle modification by some participants in the current study. Failure to exercise was the major personal factor that militates against successful non-pharmacological management of hypertension. This assertion is affirmed by the responses of the participants that they feel lazy to exercise due to body pains experienced from the previous day’s exercises. There was also the claim that they sustained various injuries from the exercise they do, which discourages them from continuing [17]. Others also avow marathon of household chores render them exhausted, and make them forget to exercise. Furthermore, stress is another confirmed factor that prevents people from exercising. Stressed from the day’s activities, participants mentioned that they are unable to exercise or eat healthily. According to a study conducted among African-Americans, stress is identified as a common barrier in managing hypertension [18]. This was mainly due to the busy nature of their daily activities. This was mainly due to the busy nature of their daily activities. Nevertheless, the importance of exercise in managing hypertension has been stressed on by several authors [19].
Another important factor in adhering to the lifestyle modification strategies of hypertension is motivation from health personnel or family. A couple of the participants unveiled that they were usually not motivated because they have no one to check up on them to ascertain their progress in adhering to the various health strategies. Importantly, they mentioned that apart from the weekly reviews they go for, they want the health personnel to check up on them and encourage them regularly to adhere to a healthy lifestyle regimen. The participants, therefore, maintained that lack of the needed encouragement prevented them from modifying their lifestyle to manage hypertension. Deficient health care systems and inadequate motivation by the health practitioners discouraged participants in a study in Egypt as they stated that the channels the health facilities provided were not adequate in the management of hypertension [7]. Due to this, the health professionals failed to get in touch with their patients more often. Similarly, some researchers have revealed that hypertension in *Ghana is* poorly controlled [8, 10].
Time factor plays a significant role as a barrier to successful non-pharmacological hypertension management. Most of the participants of the study conferred that one major problem they have concerning hypertension management is time constraints. They posited that time constraints result in late and unhealthy eating. This also affects the time to exercise. To most of the participants, work has occupied all their time to the extent that they have limited time for their personal needs, including exercising. Research shows that thirty-nine per cent ($39\%$) of clients in an interview pointed out that time constraints hinder their exercise schedules [19]. Other concerns also reveal that cooking consumes a lot of time and energy, and as a result that participants prefer eating from eateries to cooking their own food. Participants in the current study expressed this concern. Accumulation of these factors resulted in the failure of participants to modify their lifestyles in the management of hypertension. The results were also not surprising since a study in Nepal revealed that adherence to lifestyle modifications is a difficult behavior [12].
## Financial constraints
One of the barriers to following guidelines related to lifestyle recommendations was financial constraints usually accustomed to healthy foods. Below are excerpts from the interviews with the participants.
It was also found that one’s budget tends to increase especially when you have dependents.
## Difficulty adjusting to lifestyle changes
Lifestyle habits formed over a lifetime were difficult to change. Participants shared their challenging experiences in adopting new healthy lifestyle. They avowed that due to the difficult nature of adopting a new healthy lifestyle, they have always given up whenever they have made the attempt to engage in activities of such nature. Some of them shared their experiences as: The study findings also revealed that hypertensive patients find difficulty in reducing or eliminating their meat intake because they have been taking meat throughout their lifetime. Below are the excerpts from such participants:
## Personal factors
From the interview conducted, participants mentioned that some personal factors that militate against successful non-pharmacological management of hypertension. Such personal factors include laziness to exercise, forgetfulness and stress due to busy work schedules and lack of discipline. They volitionally admitted that these issues are real factors obstructing a successful lifestyle modification. A few of the participants noted that they forget and usually feel lazy exercising all the time. Some responses from the participants are indicated below: Forgetfulness, stress and lack of discipline were also reckoned as barriers to lifestyle modification.
## Lack of motivation
Motivation plays an important role in the treatment of hypertension. A few of the participants claimed that they feel relaxed when they are not motivated and hence, do not feel the need to engage in any lifestyle change.
Lack of support was also identified as one of the de-motivating factors to non-pharmacologic hypertension management.
## Busy work schedules and limited time
For some participants, the nature of their work serves as an obstacle to a successful lifestyle modification. Thus, they are unable to engage in activities of healthy living. In other words, their busy schedules do not permit them to embark on lifestyle changes models.
It was indicated that cooking consumes time and energy, as a result of that, participants resorted to buying food from outside. Consequently, some had little or no control over their diet and therefore made it difficult to engage in healthy lifestyle activities. The participant’s assertion is indicated below:
## Recommendations
As a way of disabling the barriers to non-pharmacological management of hypertension, the participants recommended that there should be a follow-up service for hypertension patients by healthcare professionals, and also, there should be intensive public education on hypertension and non-pharmacological management of hypertension.
Regular follow-up by health professionals serves as motivation for some participants. However, it is unfortunate that no conversations ensue between hypertensive patients and the health personnel in-between the review periods. Therefore, the participants suggest that the health personnel should frequently contact their patients in-between the review periods to serve as a follow-up service. Other participants also stated that the education they receive from the doctors were inadequate to help them manage hypertension, consequently leading to ineffective management of hypertension. It was also recommended that primary care providers should utilize a patient-centered approach in the care of patients with hypertension as their patients proved not to be familiar with the standard guidelines for managing hypertension, be it pharmacological or non-pharmacological [16]. A researcher also concluded that clinicians should spend adequate time in providing relevant information on the value of lifestyle modifications in controlling their blood pressure [20]. Thus, there should be intensive public education on non-pharmacological hypertension management.
The participants used in the current study seized the opportunity to reach out to both hypertensive patients and non-hypertensive patients, advising them to adhere to the various lifestyle strategies. They advised that people should, as a matter of importance, ensure preventive rather than curative methods in hypertension management. They reiterated that people should go for regular checkups in order to check their affordances. To the hypertensive patients, they encouraged them to have a normal life even though they may not resume their previous lifestyle. They added that hypertensive patients should try to adjust their lives to suit their current conditions in order to live to appreciate the life they have. It was however realized that the general public, unfortunately, has chosen not to heed any healthy lifestyle strategy until they are diagnosed. Some also have decided not to change their behaviors to suit their conditions, but rather, continued with their old unhealthy lifestyles [21, 22]. Similar to other findings, education, and provision of information has been identified as effective means in non-pharmacologic hypertension management adherence [6, 7].
## Follow-up service by healthcare professionals
Even though some participants stated that they were often checked up by healthcare personnel, it was unfortunate that was not so for most of the participants. They, therefore, recommended that healthcare professionals should make it a point to frequently check up on them. They remarked as follows:
## Intensive public education on hypertension and non-pharmacological management of hypertension
Some participants also used this opportunity to advise their fellow hypertensive and non-hypertensive patients on preventive methods they could use in order to stay free from the condition. Below are some excerpts: Non hypertensive patients were also encouraged by the study participants to read a lot about ways to prevent hypertension.
## Conclusion
The study found that adherence to non-pharmacologic management of hypertension is greatly influenced by one’s finances, some personal factors and external influences. Hence, it is necessary address these factors and also to ensure effective follow-ups and reminders in order to improve adherence to the non-pharmacologic management of hypertension. Further studies can also be conducted to address other obstacles to non-pharmacologic hypertension management.
## References
1. Joffres M, Falaschetti E, Gillespie C, Robitaille C, Loustalot F, Poulter N. **Hypertension prevalence, awareness, treatment and control in national surveys from England, the USA and Canada, and correlation with stroke and ischaemic heart disease mortality: a cross-sectional study**. *BMJ Open* (2013) **3**
2. Gao Y, Chen G, Tian H, Lin L, Lu J, Weng J. **Prevalence of hypertension in China: a cross-sectional study**. *PloS one* (2013) **8** e65938. DOI: 10.1371/journal.pone.0065938
3. Awosan KJ, Ibrahim MT, Essien E, Yusuf AA, Okolo AC. **Dietary pattern, lifestyle, nutrition status and prevalence of hypertension among traders in Sokoto Central market, Sokoto, Nigeria**. *International Journal of nutrition and metabolism* (2014) **6** 9-17
4. Abubakar S, Muhammad LU, Ahmed A, Idris F. **Knowledge, attitude, and adherence to nonpharmacological therapy among patients with hypertension and diabetes attending hypertension and diabetes clinics at Tertiary Hospitals in Kano, Nigeria**. *Sahel Medical Journal.* (2017) **20** 102
5. Okubo Y, Sairenchi T, Irie F, Yamagishi K, Iso H, Watanabe H. **Association of alcohol consumption with incident hypertension among middle-aged and older Japanese population: the Ibarakai Prefectural Health Study (IPHS).**. *Hypertension* (2014) **63** 41-7. DOI: 10.1161/HYPERTENSIONAHA.113.01585
6. Dzudie A, Kengne AP, Muna WF, Ba H, Menanga A, Kouam CK. **Prevalence, awareness, treatment and control of hypertension in a self-selected sub-Saharan African urban population: a cross-sectional study**. *BMJ Open* (2012) **2**
7. Ibrahim MM. **Hypertension in developing countries: a major challenge for the future**. *Current hypertension reports* (2018) **20** 1-0. PMID: 29349522
8. Sanuade OA, Boatemaa S, Kushitor MK. **Hypertension prevalence, awareness, treatment and control in Ghanaian population: Evidence from the Ghana demographic and health survey**. *PloS one* (2018) **13** e0205985. DOI: 10.1371/journal.pone.0205985
9. Anowie F, Darkwa S. **The knowledge, attitudes and lifestyle practices of hypertensive patients in the cape coast metropolis-Ghana**. *Journal of scientific research and reports* (2015) **6** 1-5
10. Sanuade OA, Awuah RB, Kushitor M. **Hypertension awareness, treatment and control in Ghana: a cross-sectional study**. *Ethnicity & health* (2020) **25** 702-16. DOI: 10.1080/13557858.2018.1439898
11. Nicoll Rachel, Henein Michael Y.. **Hypertension and lifestyle modification: how useful are the guidelines?**. (2010) 879-880. DOI: 10.3399/bjgp10X544014
12. Mehata S, Shrestha N, Mehta R, Vaidya A, Rawal LB, Bhattarai N. **Prevalence, awareness, treatment and control of hypertension in Nepal: data from a nationally representative population-based cross-sectional study**. *Journal of hypertension* (2018) **36** 1680-8. DOI: 10.1097/HJH.0000000000001745
13. Mothiba TM, Malema RN, Lekhuleni ME. **The experiences of the newly diagnosed hypertensive patients admitted into tertiary hospital campus in Limpopo Province, South Africa**. *African Journal for Physical Health Education, Recreation and Dance* (2013) **19** 213-21
14. Zenari V, Barbour R. **Doing Focus Groups**. *The Canadian Journal of Action Research* (2014) **15** 65-6
15. Gebrezgi MT, Trepka MJ, Kidane EA. **Barriers to and facilitators of hypertension management in Asmara, Eritrea: patients’ perspectives**. *Journal of Health, Population and Nutrition* (2017) **36** 1-7. DOI: 10.1186/s41043-017-0090-4
16. Adams OP, Carter AO. **Knowledge, attitudes, practices, and barriers reported by patients receiving diabetes and hypertension primary health care in Barbados: a focus group study.**. *BMC family practice* (2011) **12** 1-7. PMID: 21223592
17. Anthony H, Valinsky L, Inbar Z, Gabriel C, Varda S. **Perceptions of hypertension treatment among patients with and without diabetes**. *BMC family practice* (2012) **13** 1-7. DOI: 10.1186/1471-2296-13-24
18. Flynn SJ, Ameling JM, Hill-Briggs F, Wolff JL, Bone LR, Levine DM. **Facilitators and barriers to hypertension self-management in urban African Americans: perspectives of patients and family members**. *Patient preference and adherence* (2013) **7** 741. DOI: 10.2147/PPA.S46517
19. Cornelissen VA, Buys R, Smart NA. **Endurance exercise beneficially affects ambulatory blood pressure: a systematic review and meta-analysis**. *Journal of hypertension* (2013) **31** 639-48. DOI: 10.1097/HJH.0b013e32835ca964
20. Tesfaye B., Alebel A., Gebrie A., Zegeye A., Leshargie C. T., Ferede A.. **Diabetes mellitus and its association with hypertension in Ethiopia: a systematic review and meta-analysis**. (2019) **156** 107838. DOI: 10.1016/j.diabres.2019.107838
21. Heymann AD, Gross R, Tabenkin H, Porter B, Porath A. **Factors associated with hypertensive patients’ compliance with recommended lifestyle behaviours**. *IMAJ-Israel Medical Association Journal* (2011) **13** 553
22. Khan MS, Bawany FI, Mirza A, Hussain M, Khan A, Lashari MN. **Frequency and predictors of non-compliance to dietary recommendations among hypertensive patients**. *Journal of community health* (2014) **39** 732-6. DOI: 10.1007/s10900-014-9819-9
|
---
title: Visual prognosis and surgical timing of Ahmed glaucoma valve implantation for
neovascular glaucoma secondary to diabetic vitrectomy
authors:
- Jong Suk Lee
- Young Bok Lee
- Tae-Woo Kim
- Kyu Hyung Park
journal: BMC Ophthalmology
year: 2023
pmcid: PMC10022148
doi: 10.1186/s12886-023-02846-z
license: CC BY 4.0
---
# Visual prognosis and surgical timing of Ahmed glaucoma valve implantation for neovascular glaucoma secondary to diabetic vitrectomy
## Abstract
### Background
Evaluate the visual outcomes of *Ahmed glaucoma* valve implantation (AGVI) in patients with neovascular glaucoma (NVG) who underwent diabetic vitrectomy and suggest appropriate AGVI timing.
### Methods
Medical records of patients who underwent AGVI due to NVG after diabetic vitrectomy were reviewed. Successful intraocular pressure (IOP) control was defined as an IOP between 6 and 21 mmHg. Visual outcome was compared before NVG diagnosis and after AGVI, and the “favorable” visual outcome was defined as a postoperative deterioration in BCVA of less than 0.3 logMAR units compared to those before the development of NVG. Various factors including surgical timing were evaluated to identify the risk factors associated with unfavorable visual outcome.
### Results
A total of 35 eyes were enrolled and divided into group 1(medically uncontrolled NVG group, IOP more than 30mmHg, 16 eyes) and group 2(NVG group responded well to the initial non-surgical treatment but eventually required AGVI, 19 eyes). Despite the favorable rate of normalization of post-AGVI IOP ($85.7\%$), $43.8\%$ in Group 1 and $26.3\%$ in Group 2 showed unfavorable visual outcomes. In group 1, delayed surgical timing more than 1 week from the NVG diagnosis showed a significant association with unfavorable visual outcomes ($$P \leq 0.041$$). In group 2, poor patient compliance (follow up loss, refuse surgery) was the main factor of unfavorable visual outcomes.
### Conclusion
When NVG occurs in patients with proliferative diabetic retinopathy after vitrectomy, physicians should be cautious not to delay the surgical intervention, especially in patients with IOP of 30 or more despite non-surgical treatment. Early AGVI within six days might be necessary to preserve useful vision in these patients.
## Background
Neovascular glaucoma (NVG) is a severe type of secondary glaucoma characterized by the proliferation of fibrovascular membranes in the anterior segment of the eye, usually with poor treatment response and poor visual outcomes [1, 2]. With the increasing morbidities of ischemic vascular disease, including diabetes, the prevalence of NVG has also increased and now accounts for over $30\%$ of refractory glaucoma [3, 4]. The management of NVG includes both the reduction of ischemic drive with pan-retinal photocoagulation (PRP) or intravitreal anti-vascular endothelial growth factor (VEGF) administration and a decrease in intraocular pressure (IOP) with medical therapy and/or surgery.
If IOP-lowering medical therapy is insufficient, proper IOP management can be achieved with surgical options, including cyclodestructive procedures, filtering surgery, and glaucoma drainage device implantation [1, 5, 6]. When physicians can expect the possibility of visual preservation, the Ahmed valve, a flow-restrictive type glaucoma drainage device, has been used as the first choice of surgical intervention in NVG [1, 3, 7, 8].
However, few studies have been conducted regarding the surgical success rate of *Ahmed glaucoma* valve implantation (AGVI) in NVG. In previous studies, IOP after AGVI was the main criterion for surgical success [3, 8–12]. Because the underlying conditions in patients with NVG are heterogeneous and vision is already poor in many cases before NVG occurs, other factors such as visual acuity and glaucomatous optic nerve damage were not sufficiently considered. In addition, there is no consistent standard for defining the surgical success or failure of NVG, and above all, it is difficult to collect an appropriate cohort [1, 10]. As a result, there is no criterion for the golden time for favorable visual prognosis until AGVI is decided.
Proliferative diabetic retinopathy (PDR), along with ischemic central retinal vein occlusion (CRVO) and ocular ischemic syndrome (OIS), is one of the most common causative conditions, accounting for $33\%$ of NVG [2, 13]. In particular, after pars plana vitrectomy (PPV) for PDR, NVG tends to be more likely to occur [14–17]. In these cases, the subsequent NVG is devastating because PPV was performed to resolve the vision-threatening conditions in the PDR. Moreover, despite the normalization of IOP after AGVI, some cases have unexplained, irreversible central vision loss. It is clinically important to preserve the central vision because PDR patients after successful diabetic vitrectomy usually have better central vision before NVG than severe ischemic CRVO or OIS. This cohort is also appropriate for evaluating the visual prognosis and associated factors of AGVI for NVG.
In this study, we analyzed the effect of AGVI in NVG that occurred after receiving PPV due to PDR. This study is focused on evaluating the effects of the degree and duration of IIOP on visual prognosis in these patients. The therapeutic effect of AGVI was evaluated mainly by focusing on visual outcomes, and the surgical success rate based on IOP was also presented. Other factors that may affect visual prognosis were also evaluated. This article suggests the appropriate surgical timing for AGVI to preserve useful central vision and provides a hypothesis on the possible mechanism of central visual deterioration.
## Subjects
The medical records of patients who underwent AGVI for NVG after PPV for PDR at Seoul National University Bundang Hospital (SNUBH) between January 2005 and January 2019 were reviewed. NVG was diagnosed by neovascularization of the iris (NVI) and/or iridocorneal angle (NVA), with an IOP elevation of 22 mmHg or more by Goldmann applanation tonometry. Among them, patients with a follow-up period of less than one postoperative (AGVI) year and had been diagnosed with any type of glaucoma (normal-tension glaucoma, primary open-angle glaucoma, preexisting NVG, angle-closure glaucoma including post-PPV angle-closure event) were excluded. Furthermore, patients with other ocular diseases (age-related macular degeneration, retinal vein occlusion, uveitis, and pathologic myopia) or a history of interventions (cataract operation, corneal transplantation, and YAG laser posterior capsulotomy) that can affect visual acuity during the post-AGVI follow-up period were excluded from the analysis. Patients with visual acuity less than hand motion before the diagnosis of NVG were also excluded to prevent a floor effect (Fig. 1). This study was approved by the institutional review board (IRB) of SNUBH, Korea (IRB No. B$\frac{2003}{601}$ − 104) and conducted in accordance with the tenets of the Declaration of Helsinki. The IRB of SNUBH allowed the waiver of informed consent for individual patients due to the retrospective nature of the study and the analysis used anonymous clinical data.
Fig. 1Flow chart for eligible subjects with neovascular glaucoma. Abbreviations: SNUBH, Seoul National University Bundang hospital; AGVI, *Ahmed glaucoma* valve implantation; NVG, neovascular glaucoma; PPV, pars plana vitrectomy; NTG, normal-tension glaucoma; POAG, primary open-angle glaucoma; ACG, angle-closure glaucoma; NLP, no light perception; LP, light perception; AMD, age-related macular degeneration; RVO, retinal vein occlusion; YAG, Yttrium-Aluminum-Garnet The degree and duration of increased IOP are considered major factors that can affect the final visual prognosis. To evaluate the effect of pre-AGVI IOP on the visual prognosis, the group with consistently high IOP despite non-surgical treatment (MTMT, anti-VEGF injection, and additional PRP) and the group with a good response to initial non-surgical treatment should be classified separately. We classified the patients with sustained high IOP who did not obtain a pre-AGVI IOP less than 30 mmHg after initial non-surgical treatment into group [1] Patients who obtained an IOP less than 30 mmHg at two or more consecutive outpatient visits after initial non-surgical treatment were classified into group [2] In both group, AGVI was administered at the decision of glaucoma specialists.
## Data collection
The following preoperative baseline data, including demographic features, were collected: age, sex, best-corrected visual acuity (BCVA), IOP, preoperative anterior chamber paracentesis, intravitreal anti-VEGF injection, and the number of IOP-lowering medications during the first week after the diagnosis of NVG. Preoperative BCVA was measured in a sufficiently stable state after the recovery from diabetic vitrectomy. Preoperative IOP includes the IOP measured at the time of the NVG diagnosis and the last IOP within the first week after the NVG diagnosis. During the first week, the change in IOP was also calculated to evaluate the treatment response to initial medical treatment.
The following postoperative data were collected: BCVA, IOP, and surgical complications related to AGVI. Postoperative BCVA was selected as the best result of the BCVA measurements during the one year follow-up period after AGVI. The main outcome measures were preoperative and postoperative BCVA and IOP. Comparing preoperative BCVA and postoperative BCVA, a favorable visual outcome was defined as a BCVA that deteriorated to less than 0.3 logMAR unit after AGVI. Surgical success based on IOP was defined as an IOP between 6 and 21 mmHg with or without anti-glaucoma medication. Surgical failure was defined as an IOP > 21 mmHg on maximally tolerated medication at two consecutive visits, additional glaucoma surgery for IOP control, and other devastating complications such as endophthalmitis and hemophthalmos requiring additional operations. For statistical analysis, we used a LogMAR value of 2.3, 2.6, and 2.9 to represent hand motion, light perception, and no light perception, respectively [18]. If both eyes of a single patient were eligible, both eyes were included.
## Statistical analysis
All data were presented as numbers with percentages for categorical variables or the mean ± standard deviation for continuous variables. A comparative analysis was performed between groups 1 and 2 and between the group with favorable visual outcomes and unfavorable visual outcomes in group 1. Comparisons between the groups were performed using the chi-square test, Fisher’s exact test, independent t-test, and Mann Whitney test. The statistical analysis was performed using SPSS version 25.0 (SPSS Inc., Chicago, IL, USA), and P values of < 0.05 were considered significant.
## Results
Thirty-five eyes (of 32 patients) were included in this study. The time interval between diabetic vitrectomy and NVG diagnosis was 259.0 ± 313.7 days (Group 1, 303.3 ± 393.5 days and Group 2, 221.7 ± 231.7). Sixteen eyes with sustained high IOP despite initial non-surgical treatment served as group [1] The remaining 19 eyes served as group [2] In group 2, AGVI was performed after an average of six months (176.2 ± 173.2 days) after the diagnosis of NVG, and the reasons for performing the AGVI varied. In this group, AGVI was administered at the decision of glaucoma specialists when glaucomatous optic neuropathy (GON) progressed despite IOP in the early 20s, or when IOP rose again despite sustained non-surgical treatment. In addition to the above criteria, the patient’s preferences, contralateral blindness, and individual social and environmental factors were also considered.
The demographic features and preoperative information of groups 1 and 2 are summarized in Table 1. There were no significant differences in age or sex. Among the initial non-surgical treatments for NVG, the number of IOP-lowering medications was significantly lower in group 1. This difference was due to the patients who received immediate AGVI at the time of NVG. They usually used less than three IOP-lowering eye drops before AGVI. In three patients, emergency AGVI was performed on the same day that NVG was diagnosed because the involved eye was the last eye, or the patient strongly wanted immediate AGVI. Although there were no significant differences in IOPs at the time of NVG diagnosis between the two groups, the last IOP ‘within the first week’ after the diagnosis of NVG was significantly lower in group 2. Intravitreal anti-VEGF injection showed no statistically significant effect on the acute phase IOP reduction.
Table 1Preoperative data from groups 1 and 2 during the first week after neovascular glaucoma diagnosisTotal ($$n = 35$$)Group 1 ($$n = 16$$)Group 2 ($$n = 19$$)P-value Baseline characteristics Age (years)58.4 ± 10.358.4 ± 10.351.9 ± 10.20.246†Sex (M/F)$\frac{22}{138}$/$\frac{814}{50.179}$‡Laterality (R/L)$\frac{18}{177}$/$\frac{911}{80.505}$‡VA (LogMAR)BCVA before NVG diagnosis0.85 ± 0.620.88 ± 0.500.82 ± 0.710.483†VA at the time of NVG diagnosis1.67 ± 0.792.13 ± 0.501.28 ± 0.710.008†IOP at the time of NVG diagnosis (mmHg)40.2 ± 8.341.3 ± 8.339.3 ± 8.20.905†*Lens status* (Phakic/Pseudophakic)$\frac{5}{303}$/$\frac{132}{170.642}$‡ Non-surgical treatment and therapeutic response during the first week after NVG diagnosis No. of IOP-lowering medications2.51 ± 0.692.13 ± 0.782.84 ± 0.360.005†AC paracentesis (%)14 (40.0)6 (37.5)8 (42.1)1.000‡Anti-VEGF injections (%)27 (77.1)12 (75.0)15 (78.9)1.000‡Add PRP5 (14.3)1 (6.3)4 (21.1)0.347‡Last IOP within the first week (mmHg)30.2 ± 11.3 *38.7 ± 8.3 **22.3 ± 7.2 ***0.000†Delta IOP (mmHg)(IOP at the day of NVG diagnosis - IOP at the last day within the first week)11.3 ± 10.8 *3.0 ± 7.7 **19.0 ± 6.8 ***0.000†Delta IOP > 10 mmHg (%)16 (45.7)3 (23.1)13 (68.4)0.000‡* Except for eight cases who received immediate AGVI or did not visit within one week from NVG diagnosis** Except for three cases who received immediate AGVI*** Except for five cases who did not visit within one week from NVG diagnosis† Calculated using an independent t-test and Mann Whitney test‡ Calculated using a Chi-Square test and Fisher’s Exact testAbbreviations: NVG, neovascular glaucoma; M, male; F, female; R, right; L, left; No., number; IOP, intraocular pressure; AC, anterior chamber; AGVI, *Ahmed glaucoma* valve implantation; Add PRP, additional pan-retinal photocoagulationGroup 1, Group of NVG patients who maintained an IOP of 30 mmHg or more before AGVI despite initial non-surgical treatment including maximum tolerable medical treatment; Group 2, Group of NVG patients who responded well to initial non-surgical treatment but eventually required AGVI.
Postoperative information are summarized in Table 2. The surgical success rate based on the IOP one year after AGVI was $81.3\%$ and $89.5\%$ in groups 1 and 2, respectively. However, although the overall surgical success rate of AGVI based on IOP was $85.7\%$, the final visual outcomes were unfavorable in $30\%$ of all patients ($43.8\%$ and $26.3\%$ in groups 1 and 2, respectively).
Table 2Postoperative data from groups 1 and 2 during the first year from *Ahmed glaucoma* valve implantationTotal($$n = 35$$)Group 1($$n = 16$$)Group 2($$n = 19$$)P-value Post-AGVI BCVA 1.22 ± 0.791.21 ± 0.691.25 ± 0.860.961† Unfavorable visual outcome (%) 12 (34.3)7 (43.8)5 (26.3)0.311‡ IOP at the time of AGVI (mmHg) 36.8 ± 9.3641.3 ± 7.433.1 ± 9.20.003† Postoperative IOP 3 months after AGVI17.9 ± 7.117.8 ± 9.718.0 ± 3.6*0.108†6 months after AGVI17.6 ± 6.1**16.3 ± 3.7*18.7 ± 7.3*0.513†12 months after AGVI15.2 ± 4.0**14.5 ± 3.7*15.7 ± 4.2*0.327† Surgical success at one year based on IOP (%) 30 (85.7)13 (81.3)17 (89.5)0.642‡ Complications (%) Anterior chamber hyphema6 (17.1)3 (18.8)3 (15.8)1.000‡Tube obstruction1 (2.9)1 (6.3)0 [0]0.457‡Vitreous hemorrhage8 (22.9)4 (25.0)4 (21.1)1.000‡Early hypotony with choroidal detachmentor collapsed anterior chamber1 (2.9)1 (6.3)0 [0]0.441‡Endophthalmitis0 [0]0 [0]0 [0]1.000‡*Except for one case who underwent re-AGVI**Except for two cases who underwent re-AGVI† Calculated using an independent t-test and Mann Whitney test‡ Calculated using a Chi-Square test and Fisher’s Exact testAbbreviations: AGVI, *Ahmed glaucoma* valve implantation; BCVA, best-corrected visual acuity; IOP, intraocular pressureGroup 1, Group of NVG patients who maintained an IOP of 30 mmHg or more before AGVI despite initial non-surgical treatment including maximum tolerable medical treatment; Group 2, Group of NVG patients who responded well to initial non-surgical treatment but eventually required AGVI.
Figure 2 shows the overall visual prognosis according to the time interval between NVG diagnosis and AGVI in group 1 and 2. In group 1, it reveals that the patients with unfavorable visual outcomes are concentrated in the intervals between NVG diagnosis and AGVI for a week or more. On the other hand, no remarkable tendency was observed between visual outcome and AGVI timing in group 2.
In group 1, delayed surgical timing of AGVI for more than one week showed a statistically significant association with unfavorable visual outcomes ($$P \leq 0.041$$, Table 3). There were no significant differences in pre-AGVI IOP, baseline BCVA before the diagnosis of NVG, and initial non-surgical treatment options including anti-VEGF injection between the favorable and unfavorable visual outcome groups.
In group 2, the final visual outcome was unfavorable in five patients, but there was no tendency related to AGVI timing. Two of these patients had long-term or multiple follow-up loss events after NVG diagnosis and before AGVI. The other two patients recommended AGVI for an IOP of 40 or more, but the patient’s rejection delayed the AGVI. In the other patient, the IOP decreased to 27 mmHg within one week after the non-surgical treatment, but since the visual acuity had already decreased significantly, AGVI was performed. It is estimated that the period from the outpatient visit to the hospital was delayed after the IOP event occurred. Of the 14 patients with a favorable visual outcome, five patients within one week and 11 patients within three weeks obtained normal IOP with non-surgical treatment. The remaining three patients underwent AGVI due to maintaining IOP in the mid-20s. These 14 patients eventually.
underwent AGVI due to the progression of GON or delayed increase in IOP, but central vision was preserved.
Fig. 2Scatter plot showing the relationship between the timing of AGVI (*Ahmed glaucoma* valve implantation) and visual outcomes. Groups 1 and 2 were categorized according to the treatment response of initial non-surgical treatment for increased IOP. In group 1, the patients with an unfavorable visual outcome are concentrated in the intervals between NVG diagnosis and AGVI for a week or more (dotted box). In group 2, no clear association was observed between vision prognosis and timing of AGVI Table 3Factors affecting central vision loss after *Ahmed glaucoma* valve implantation in Group 1Favorable visual outcome* ($$n = 9$$)Unfavorable visual outcome ($$n = 7$$)P-value Age (years) 58.7 ± 11.458.0 ± 8.70.916† Sex (M/F) $\frac{5}{43}$/41.000‡ IOP (mmHg) At the time of NVG diagnosis41.0 ± 8.641.6 ± 7.80.874†At the time of AGVI41.4 ± 7.441.0 ± 7.50.791† BCVA (LogMAR) Before the diagnosis of NVG0.74 ± 0.421.05 ± 0.550.238†At the time of NVG diagnosis2.23 ± 0.192.00 ± 0.230.032† Early AGVI timing (%) (less than one week from NVG diagnosis) 7 (77.8)1 (14.3)0.041‡ No. of IOP-lowering medications 2.0 ± 0.942.3 ± 0.450.678† AC paracentesis (%) 6 (66.7)6 (42.9)1.000‡ Anti-VEGF injections (%) 5 (62.5)5 (86.7)0.585‡ Add PRP (%) 0 [0]1 (14.3)0.438‡*Favorable visual outcome: Comparing preoperative BCVA(sufficiently stable BCVA after the recovery from diabetic vitrectomy) and postoperative BCVA (the best result of the BCVA measurements during the one year follow-up period after AGVI), a favorable visual outcome was defined as a BCVA that deteriorated to less than 0.3 logMAR unit after AGVI.† Calculated using an independent t-test and Mann Whitney test‡ Calculated using a Chi-Square test and Fisher’s Exact testAbbreviations: M, male; F, female; IOP, intraocular pressure; BCVA, best-corrected visual acuity; AGVI, *Ahmed glaucoma* valve implantation; NVG, neovascular glaucoma; AC, anterior chamber; Add PRP, additional pan-retinal photocoagulationGroup 1, Group of NVG patients who maintained an IOP of 30 mmHg or more before AGVI despite initial non-surgical treatment, including maximum tolerable medical treatment.
## Discussion
The prevalence of NVG after diabetic vitrectomy was reported in 4.6-$23.6\%$ of cases [14–16, 19]. Concerning changes in intraocular anti-VEGF levels after diabetic vitrectomy, it has been suggested that a high VEGF level could be maintained in the vitreous cavity after vitrectomy for PDR [20–22]. Moreover, vitrectomy with or without a cataract operation can disrupt the structures considered barriers to the diffusion of angiogenic factors to the anterior segment from the posterior segment [23]. Although the effect of vitrectomy (removing the vitreous) or phacovitrectomy itself on the intraocular vasoproliferative substances (VEGF and inflammatory cytokines) and development of NVG has not been established, it is clear that NVG developed after diabetic vitrectomy is clinically important [24–28].
Few studies have been conducted on the prognosis of surgical treatment (AGVI) for NVG after diabetic vitrectomy [17, 29]. They suggested that AGVI is a safe and effective procedure that enables successful IOP control in patients with NVG associated with diabetic vitrectomy. Other studies related to the treatment of NVG also positively evaluated the therapeutic effect of AGVI [3, 8–12]. However, previous studies included cases with heterogeneous underlying ocular conditions and only focused on the normalization of IOP in evaluating the surgical success of AGVI. Detailed visual outcome has not been sufficiently considered due to generally poor visual outcomes and only the proportion of vision less after AGVI was calculated. Because NVG patients after diabetic vitrectomy tend to have better baseline visual acuity before the diagnosis of NVG compared to other conditions, surgical success rate based on visual acuity could be evaluated. Especially, this study focused on patients with unfavorable visual outcomes, even though IOP was well controlled after AGVI.
In addition to normalization of IOP, this study compared the baseline BCVA “before” NVG diagnosis and after AGVI, and the factors affecting visual prognosis were also considered. The overall surgical success rate based on IOP was comparable to a previous study that reported percentages of $83.8\%$ at one year in the NVG after diabetic vitrectomy group [17]. $43.8\%$ and $26.3\%$ had unfavorable visual outcomes in groups 1 and 2, respectively. It is estimated that consistently high IOP may be associated with poor visual prognosis. Especially in group 1, early surgical timing within a week was significantly related to vision preservation (favorable visual outcome). This suggests that close observation should be performed during the first week after NVG diagnosis, and AGVI should be considered without delay in cases with insufficient response to initial non-surgical treatment. Based on our study’s results, we suggest that the IOP change during the first week after NVG diagnosis could be used as a criterion for distinguishing group 1 from group 2. In particular, the change in IOP over the first week can be used as a useful indicator when the standard was set to 10 mmHg (Table 1).
Park et al. suggested that visual deterioration after NVG in patients with PDR was attributed to the progression of diabetic retinopathy or GON [17]. In addition to these factors, considering the relationship between the period of IOP increase and deterioration of central vision and the representative case (Fig. 3), retinal ischemia due to blood flow restriction during severe IOP rise seems to be a reasonable additional cause of visual deterioration. The probable mechanisms for retinal ischemia due to increased IOP are as follows: [1] decreased blood flow in the optic nerve head (ONH), [2] failure of blood flow autoregulation in the ONH and [3] decreased blood flow in the arterioles and capillaries of the inner retina (Fig. 4). Perfusion pressure decrease and retinal blood flow autoregulation failure due to severe IOP rise inhibit blood flow in ONH [30, 31]. In addition, there have been many reports of retinal artery occlusion due to IOP rise [18, 32]. When perfusion from the optic nerve head is non-physiologically reduced, inner retinal blood circulation may be more directly affected by increased IOP. Moreover, diabetic peripheral neuropathy, the aging process, and accompanying hypertension or arteriosclerosis/atherosclerosis are additional factors that can accelerate retinal ischemia due to increased IOP [33, 34].
Figure 3 shows serial optical coherence tomography (OCT) photographs, which reveal inner retinal edema with hyper-reflectivity and subsequent progressive retinal thinning. These findings are similar to the pattern of incomplete CRAO rather than complete CRAO and support our conclusion that early AGVI can prevent vision loss. It can be explained because the mechanism of retinal ischemia is not due to embolism, but because the blood flow is limited by the decrease in perfusion pressure and blood flow autoregulation failure. Because the mechanism of retinal ischemia is different, unlike in embolic CRAO, where the damage starts at the middle retinal layer, ischemic change was observed mainly in the inner layer adjacent to the eyeball cavity.
Fig. 3Optical coherence tomographic findings (vertical sections) of a representative case showing ischemic retinal damage due to sustained high IOP during the NVG attack. This patient had undergone AGVI seven days after NVG diagnosis, and their final visual outcome was unfavorable. A is an OCT finding taken one year after diabetic PPV (three months before NVG diagnosis). B and C are OCT findings at the time of NVG diagnosis and three days after AGVI, respectively. Figures B and C shows inner retinal edema and hyper-reflectivity, suggesting retinal ischemia. D is an OCT photograph one year after the occurrence of NVG, and prominent thinning of inner retinal layers is confirmed. A-1 and B-1 are magnified images of A and B, respectively. The inner retinal layer includes the nerve fiber layer, ganglion cell layer, inner plexiform layer, and inner nuclear layer. The outer retinal layer includes the outer plexiform layer and photoreceptor layer. Measurements of the inner retinal layer (bold double-sided arrows) and outer retinal layer (double-sided arrows) were made 1 mm from the fovea. Among the inner retinal layers, progressive thinning was observed in the nerve fiber layer, and the ganglion cell layer (arrows) and the inner nuclear layer was relatively preserved (arrowheads). E and F show the average retinal thickness values (um) of A and D, respectively. The average retinal thickness values were obtained from the 1, 3, 6 mm thickness map of each quadrant (OCT, Spectralis OCT; Heidelberg Engineering, Heidelberg, Germany). The retinal thickness of the whole macula was markedly decreased one year after the NVG diagnosis Cascades from the increased IOP mentioned above are presumed to be the contributors to vision loss, especially in group 1 (Fig. 4). The reason for delayed AGVI was not only due to patient compliance, including rejection of surgical treatment, personal circumstances, or difficulties undergoing emergency operations but also delayed consulting process from retina specialists who diagnosed NVG to glaucoma specialists who did not have much experience in managing NVG after diabetic vitrectomy. It has been difficult to specify IOP, which causes impaired blood flow in the optic nerve head and retinal circulation. In hyphema patients (otherwise, healthy eye), it is accepted that surgical management should be performed on the second day for 60 mmHg, the fifth day for 50 mmHg, and the seventh day for 35 mmHg [35]. Based on our study results, in NVG patients with advanced PDR, the maximum tolerable period of the retina against ischemia due to medically uncontrolled high IOP of more than 30 mmHg is estimated to be less than one week. The consulting process with glaucoma specialists, medical treatment attempts, and the evaluation of its effectiveness should be completed within one week without delay.
Fig. 4All possible mechanisms related to acute central vision loss after IOP rise due to NVG in patients with PDR. Diabetes mellitus and several other systemic conditions are factors that make NVG patients more susceptible to retinal ischemia. IOP rise due to NVG can accelerate the progression of glaucomatous optic neuropathy and diabetic retinopathy. In addition, especially in group 1, sustained extremely increased IOP may restrict the retinal blood flow through ONH and inner retinal circulation. Since the blood flow autoregulation is effective over only a narrow critical range normally, under the high IOP, the retinal blood flow restriction may not be compensated. The factors mentioned above can affect vision loss in a complex fashion. The Fig. 4 was created by Jong Suk Lee who is one of the co-authors Limitations of this study include its retrospective nature, small sample size, and limited follow-up period. However, it is difficult to conduct a well-designed case-controlled study or prospective randomized trials because NVG after vitrectomy is rare, conducted under different systemic and ocular conditions, and the threat of vision. Our study applied specific exclusion criteria despite having an insufficient number of cases to evaluate visual outcomes and its associated factors. Although long-term outcomes over one year of AGVI were not included in this study’s results, we aimed to investigate the effect of temporary IOP rise on the retina and visual function of PDR patients, so that the appropriate timing of AGVI could be determined. Furthermore, the time point of NVG diagnosis may not be exactly the time point of the IOP increase. However, this study analyzed the effect of the degree and duration of IOP elevation on visual acuity clinically rather than a detailed review of the pathophysiology, ideal treatment options for each stage, and course of NVG. Theoretically, in many cases of group 2, the open-angle served as a potential reservoir for medical treatment. Future research will be required to provide strong evidence for predicting responsiveness to initial non-surgical treatment through the iridocorneal angle evaluation or alternative methods in NVG. In addition, although there was no difference in IOP at the time of NVG, preoperative BCVA showed difference between group 1 and group 2. Media opacity (corneal edema, hyphema, and vitreous hemorrhage), diabetic optic neuropathy or microvascular infarction of macula may be possible factors. A well-designed, large scale study is needed to determine whether these factors can affect the timing and prognosis of AGVI.
In conclusion, AGVI was a safe and effective procedure that enabled successful IOP control in patients with NVG associated with diabetic vitrectomy. However, to improve the visual outcomes of AGVI, physicians should be cautious not to delay surgical intervention, especially in patients who do not respond to initial non-surgical treatment. If IOP is not adequately controlled (< 30 mmHg) within a week from the time of NVG diagnosis, early AGVI within six days might be necessary to preserve useful vision.
## References
1. Shchomak Z, Cordeiro Sousa D, Leal I, Abegão Pinto L. **Surgical treatment of neovascular glaucoma: a systematic review and meta-analysis**. *Graefes Arch Clin Exp Ophthalmol* (2019.0) **257** 1079-89. DOI: 10.1007/s00417-019-04256-8
2. Havens SJ, Gulati V. **Neovascular Glaucoma**. *Dev Ophthalmol* (2016.0) **55** 196-204. DOI: 10.1159/000431196
3. He Y, Tian Y, Song W, Su T, Jiang H, Xia X. **Clinical efficacy analysis of Ahmed glaucoma valve implantation in neovascular glaucoma and influencing factors: a STROBE-compliant article**. *Medicine* (2017.0) **96** e8350. DOI: 10.1097/MD.0000000000008350
4. Hayreh SS. **Neovascular glaucoma**. *Prog Retin Eye Res* (2007.0) **26** 470-85. DOI: 10.1016/j.preteyeres.2007.06.001
5. Yap-Veloso MI, Simmons RB, Echelman DA, Gonzales TK, Veira WJ, Simmons RJ. **Intraocular pressure control after contact transscleral diode cyclophotocoagulation in eyes with intractable glaucoma**. *J Glaucoma* (1998.0) **7** 319-28. DOI: 10.1097/00061198-199810000-00006
6. Allen RC, Bellows AR, Hutchinson BT, Murphy SD. **Filtration surgery in the treatment of neovascular glaucoma**. *Ophthalmology* (1982.0) **89** 1181-7. DOI: 10.1016/S0161-6420(82)34672-2
7. Hong CH, Arosemena A, Zurakowski D, Ayyala RS. **Glaucoma drainage devices: a systematic literature review and current controversies**. *Surv Ophthalmol* (2005.0) **50** 48-60. DOI: 10.1016/j.survophthal.2004.10.006
8. Yalvac IS, Eksioglu U, Satana B, Duman S. **Long-term results of Ahmed glaucoma valve and Molteno implant in neovascular glaucoma**. *Eye* (2007.0) **21** 65-70. DOI: 10.1038/sj.eye.6702125
9. Zhang HT, Yang YX, Xu YY, Yang RM, Wang BJ, Hu JX. **Intravitreal bevacizumab and Ahmed glaucoma valve implantation in patients with neovascular glaucoma**. *Int J Ophthalmol* (2014.0) **7** 837-42. PMID: 25349803
10. Zhou M, Xu X, Zhang X, Sun X. **Clinical outcomes of Ahmed Glaucoma valve implantation with or without Intravitreal Bevacizumab pretreatment for neovascular Glaucoma: a systematic review and Meta-analysis**. *J Glaucoma* (2016.0) **25** 551-7. DOI: 10.1097/IJG.0000000000000241
11. Tang M, Fu Y, Wang Y, Zheng Z, Fan Y, Sun X. **Efficacy of intravitreal ranibizumab combined with Ahmed glaucoma valve implantation for the treatment of neovascular glaucoma**. *BMC Ophthalmol* (2016.0) **16** 7. DOI: 10.1186/s12886-016-0183-7
12. Liu L, Xu Y, Huang Z, Wang X. **Intravitreal ranibizumab injection combined trabeculectomy versus Ahmed valve surgery in the treatment of neovascular glaucoma: assessment of efficacy and complications**. *BMC Ophthalmol* (2016.0) **16** 65. DOI: 10.1186/s12886-016-0248-7
13. Rodrigues GB, Abe RY, Zangalli C, Sodre SL, Donini FA, Costa DC. **Neovascular glaucoma: a review**. *Int J Retina Vitreous* (2016.0) **2** 26. DOI: 10.1186/s40942-016-0051-x
14. 14.Sakamoto M, Hashimoto R, Yoshida I, Ubuka M, Maeno T. Risk factors for neovascular glaucoma after vitrectomy in eyes with proliferative diabetic retinopathy. Clinical ophthalmology (Auckland, NZ). 2018;12:2323–9.
15. Goto A, Inatani M, Inoue T, Awai-Kasaoka N, Takihara Y, Ito Y. **Frequency and risk factors for neovascular glaucoma after vitrectomy in eyes with proliferative diabetic retinopathy**. *J Glaucoma* (2013.0) **22** 572-6. DOI: 10.1097/IJG.0b013e31824d514a
16. Kwon JW, Jee D, La TY. **Neovascular glaucoma after vitrectomy in patients with proliferative diabetic retinopathy**. *Medicine* (2017.0) **96** e6263. DOI: 10.1097/MD.0000000000006263
17. Park UC, Park KH, Kim DM, Yu HG. **Ahmed glaucoma valve implantation for neovascular glaucoma after vitrectomy for proliferative diabetic retinopathy**. *J Glaucoma* (2011.0) **20** 433-8. DOI: 10.1097/IJG.0b013e3181f3eb06
18. 18.Lee JS, Kim JY, Jung C, Woo SJ. Iatrogenic ophthalmic artery occlusion and retinal artery occlusion.Progress in retinal and eye research. 2020:100848.
19. 19.Liang X, Zhang Y, Li YP, Huang WR, Wang JX, Li X. Frequency and Risk Factors for Neovascular Glaucoma After Vitrectomy in Eyes with Diabetic Retinopathy: An Observational Study. Diabetes therapy: research, treatment and education of diabetes and related disorders. 2019;10(5):1801–9.
20. Itakura H, Kishi S, Kotajima N, Murakami M. **Persistent secretion of vascular endothelial growth factor into the vitreous cavity in proliferative diabetic retinopathy after vitrectomy**. *Ophthalmology* (2004.0) **111** 1880-4. DOI: 10.1016/j.ophtha.2004.03.035
21. Wakabayashi Y, Usui Y, Okunuki Y, Ueda S, Kimura K, Muramatsu D. **Intraocular VEGF level as a risk factor for postoperative complications after vitrectomy for proliferative diabetic retinopathy**. *Investig Ophthalmol Vis Sci* (2012.0) **53** 6403-10. DOI: 10.1167/iovs.12-10367
22. Chen HJ, Ma ZZ, Li Y, Wang CG. **Change of vascular endothelial growth factor levels following vitrectomy in eyes with proliferative Diabetic Retinopathy**. *J Ophthalmol* (2019.0) **2019** 6764932. DOI: 10.1155/2019/6764932
23. 23.Weinreb RN, Wasserstrom JP, Parker W. Neovascular glaucoma following neodymium-YAG laser posterior capsulotomy. Archives of ophthalmology (Chicago, Ill: 1960). 1986;104(5):730-1.
24. Senn P, Schipper I, Perren B. **Combined pars plana vitrectomy, phacoemulsification, and intraocular lens implantation in the capsular bag: a comparison to vitrectomy and subsequent cataract surgery as a two-step procedure**. *Ophthalmic Surg Lasers* (1995.0) **26** 420-8. DOI: 10.3928/1542-8877-19950901-09
25. 25.Stefansson E, Landers MB 3rd, Wolbarsht ML. Increased retinal oxygen supply following pan-retinal photocoagulation and vitrectomy and lensectomy. Trans Am Ophthalmol Soc. 1981;79:307–34.
26. Chung TY, Chung H, Lee JH. **Combined surgery and sequential surgery comprising phacoemulsification, pars plana vitrectomy, and intraocular lens implantation: comparison of clinical outcomes**. *J Cataract Refract Surg* (2002.0) **28** 2001-5. DOI: 10.1016/S0886-3350(02)01354-8
27. Lahey JM, Francis RR, Kearney JJ. **Combining phacoemulsification with pars plana vitrectomy in patients with proliferative diabetic retinopathy: a series of 223 cases**. *Ophthalmology* (2003.0) **110** 1335-9. DOI: 10.1016/S0161-6420(03)00454-8
28. Tseng HY, Wu WC, Hsu SY. **Comparison of vitrectomy alone and combined vitrectomy, phacoemulsification and intraocular lens implantation for proliferative diabetic retinopathy**. *Kaohsiung J Med Sci* (2007.0) **23** 339-43. DOI: 10.1016/S1607-551X(09)70419-X
29. Cheng Y, Liu XH, Shen X, Zhong YS. **Ahmed valve implantation for neovascular glaucoma after 23-gauge vitrectomy in eyes with proliferative diabetic retinopathy**. *Int J Ophthalmol* (2013.0) **6** 316-20. PMID: 23826525
30. Hayreh SS. **Blood flow in the optic nerve head and factors that may influence it**. *Prog Retin Eye Res* (2001.0) **20** 595-624. DOI: 10.1016/S1350-9462(01)00005-2
31. Hayreh SS. **Ischemic optic neuropathy**. *Prog Retin Eye Res* (2009.0) **28** 34-62. DOI: 10.1016/j.preteyeres.2008.11.002
32. 32.Rooney DM, Oltmanns MH, Sapp MR, Morris RE, NECROSIS OF THE RETINA AND UVEAL TRACT SECONDARY TO EXPANDING GAS TAMPONADE.Retinal cases & brief reports. 2018.
33. 33.Hayreh SS, Bill A, Sperber GO. Effects of high intraocular pressure on the glucose metabolism in the retina and optic nerve in old atherosclerotic monkeys. Graefe’s archive for clinical and experimental ophthalmology = Albrecht von Graefes Archiv fur klinische und experimentelle Ophthalmologie. 1994;232(12):745–52.
34. Haefliger IO, Meyer P, Flammer J, Lüscher TF. **The vascular endothelium as a regulator of the ocular circulation: a new concept in ophthalmology?**. *Surv Ophthalmol* (1994.0) **39** 123-32. DOI: 10.1016/0039-6257(94)90157-0
35. Walton W, Von Hagen S, Grigorian R, Zarbin M. **Management of traumatic hyphema**. *Surv Ophthalmol* (2002.0) **47** 297-334. DOI: 10.1016/S0039-6257(02)00317-X
|
---
title: Seven-year outcomes following intensive anti-vascular endothelial growth factor
therapy in patients with exudative age-related macular degeneration
authors:
- Regina Lukacs
- Miklos Schneider
- Zoltan Zsolt Nagy
- Gabor Laszlo Sandor
- Kinga Kaan
- Antonia Asztalos
- Lajos Enyedi
- Gyorgy Pek
- Gyorgy Barcsay
- Antal Szabo
- Agnes Borbandy
- Illes Kovacs
- Miklos Denes Resch
- Andras Papp
journal: BMC Ophthalmology
year: 2023
pmcid: PMC10022151
doi: 10.1186/s12886-023-02843-2
license: CC BY 4.0
---
# Seven-year outcomes following intensive anti-vascular endothelial growth factor therapy in patients with exudative age-related macular degeneration
## Abstract
### Background
Anti-vascular endothelial growth factor (VEGF) therapy is currently the most effective therapy of exudative age-related macular degeneration (AMD). The aim of this study was to assess long-term benefits of intensive aflibercept and ranibizumab anti-VEGF therapy in patients with exudative AMD.
### Methods
Two clinical trial sites recruited their original subjects for a re-evaluation 7 years after the baseline visit of the phase-3 Vascular Endothelial Growth Factor (VEGF) Trap-Eye: Investigation of Efficacy and Safety in Wet Age-Related Macular Degeneration (VIEW 2) trial.
Forty-seven eyes of 47 patients with AMD originally treated with ranibizumab (14 eyes) or aflibercept (33 eyes) were included.
### Results
Mean number of injections was 17.8 ± 3.0 during participation in the VIEW 2 trial. Fourteen of 47 ($30\%$) eyes were given additional injections with a mean number of 5.7 ± 4.5 after the trial. At a mean follow-up time of 82 ± 5 months best corrected visual acuity (BCVA) remained stable or improved (≤ 10 letters lost) in $55\%$ of patients in the entire study population, in $43\%$ in the ranibizumab group and in $60\%$ in the aflibercept group. In both groups combined mean BCVA was 54 ± 13 letters at baseline, 65 ± 17 letters at the end of the intensive phase and 45 ± 25 letters at the end of follow-up. There was no statistically significant difference in BCVA between the two groups at baseline ($$p \leq 0.88$$) and at the end of follow-up ($$p \leq 0.40$$). Macular atrophy was observed in $96\%$ of eyes, average area was 7.22 ± 6.31 mm2 with no statistically significant difference between groups ($$p \leq 0.47$$). Correlation between BCVA at end-of-follow-up and the area of atrophy was significant ($p \leq 0.001$). At the end of follow-up, fluid was detected in 7 of 47 eyes ($15\%$) indicating disease activity.
### Conclusion
Long-term efficacy of aflibercept and ranibizumab was largely consistent. Following a two-year intensive therapy with as-needed regimen, BCVA was maintained or improved in almost half of the patients and in the ranibizumab group and more than half of the patients in the aflibercept group with very few injections. In a remarkable proportion of eyes, BCVA declined severely which underlines the need for long-term follow-ups and may indicate a more prolonged intensive therapy.
### Trial registrations
VIEW 2 study: ClinicalTrials.gov ID: NCT00637377, date of registration: March 18, 2008.
Long-term follow-up: IRB nr.: SE RKEB $\frac{168}{2022}$, ClinicalTrials.gov ID: NCT05678517, date of registration: December 28, 2022, retrospectively registered.
## Background
Anti-vascular endothelial growth factor (anti-VEGF) therapy has been the most effective therapy of exudative (wet) age-related macular degeneration (AMD) for over a decade and is currently considered standard-of-care. Aflibercept has a more prolonged and potent anti-VEGF effect compared to prior treatments (bevacizumab and ranibizumab) [1]. The reason for this enhanced effect is that aflibercept binds to VEGF-B and placental growth factor (PGF) isoforms as well in addition to isoforms of VEGF-A, whereas bevacizumab and ranibizumab only binds to the latter [2]. Additionally, compared to ranibizumab, aflibercept has a longer intravitreal half-life that allows a less frequent treatment regimen [2–4]. Aflibercept has also been proven to be effective in other diseases such as diabetic macular edema [5] and central retinal vein occlusion [6, 7] where VEGF plays an important role in the pathogenesis.
Long-term outcomes of anti-VEGF treatments on wet AMD are less frequently documented in clinical trials as the length of those are usually limited to one to two years. However, in the recent past, several papers, many of which based on real-life data, were published presenting the long-term efficacy of intravitreal anti-VEGF agents and the number of given injections with variable results [8–22].
This study aims to assess long-term benefits of intensive anti-VEGF therapy followed by a very low number of injections in patients with exudative AMD.
## Study design and patient recruitment
This study was conducted at the Department of Ophthalmology at Semmelweis University and the Bajcsy-Zsilinszky Hospital in Budapest, Hungary. Study participants were patients with neovascular AMD, subjects of two sites in Budapest, Hungary of the “Vascular Endothelial Growth Factor (VEGF) Trap-Eye: Investigation of Efficacy and Safety in Wet Age-Related Macular Degeneration (VIEW 2)” phase-3 multicenter, prospective, randomized, double blind clinical trial (ClinicalTrials.gov ID: NCT00637377) [23]. The trial was carried out in accordance with the tenets of the Declaration of Helsinki, with approval of respective institutional review boards of the participating centers. For the long-term follow-up after the VIEW 2 study, institutional review board approval was obtained for a retrospective analysis (IRB approval nr: SE RKB $\frac{168}{2022}$, ClinicalTrials.gov ID: NCT05678517). Written informed consent was obtained from each participant before enrollment and at the end of follow-up for the retrospective analysis.
Patients were treated with intravitreal anti-VEGF injections between 2008 and 2017. During the first 2 years in the framework of the VIEW 2 study (intensive phase), participants were randomized per protocol to intravitreal aflibercept (aflibercept group) or ranibizumab treatment arms (ranibizumab group). In the first year, patients received aflibercept injections every 4 or 8 weeks or ranibizumab every 4 weeks. During the second year, patients received injections every 4–12 weeks with a minimum of dosing every 12 weeks and interim as-needed monthly injections.
After finishing the intensive phase, patients were followed under regular clinical care in real-life conditions (post-intensive phase) and both groups were treated with predominantly ranibizumab anti-VEGF injections (3 patients received aflibercept occasionally when ranibizumab was not available). Treatment was administered pro re nata (PRN, as needed). Retreatment criteria in the post-intensive phase were based on visual acuity (more than 5 letters worsening from the previous visit), signs of activity on optical coherence tomography (OCT) (+ 50 µm increase in central retinal thickness from the previous visit) or indirect slit lamp biomicroscopy (new hemorrhage at the lesion). The decision to retreat ultimately rested on the judgement of the patient’s treating physician, taking the availability of the medicine into account.
Results at the end of the follow-up were retrospectively analyzed.
## Study objectives
Primary outcomes of the study were change in best corrected visual acuity (BCVA), change in size of geographic atrophy on fundus autofluorescence (FAF) and presence or absence of intra- or subretinal fluid on spectral-domain optical coherence tomography (OCT) at the end of follow-up. BCVA was measured with standardized Early Treatment Diabetic Retinopathy Study (ETDRS) visual acuity charts. Visual acuity was defined as stable when less than 10 letters were gained or lost, improved when at least 10 letters were gained and worsened when at least 10 letters were lost on the ETDRS chart. FAF and OCT examinations were performed on the Spectralis OCT (Heidelberg Engineering GmbH, Heidelberg, Germany) device. Area of geographic atrophy was calculated using the built-in software (Heidelberg Eye Explorer, version: 1.10.2.0) of the Spectralis OCT device following manual marking.
Secondary endpoints included number of injections and potential adverse events.
## Statistical analysis
Mann–Whitney test was used to compare differences in mean visual acuity and geographic atrophy size between the start and end of follow-up in the two treatment groups. Wilcoxon test was used to assess changes in visual acuity in groups together and separately. The correlation between BCVA at the end of follow-up and the area of atrophy was evaluated with Spearman test. For rounding percentage values, the largest remainder method was used.
Statistica 8.0 (Statsof, Tulsa OK, USA) software was used for the analysis. Statistically significant difference was considered as $p \leq 0.05.$
## Patient characteristics and follow-up time
Forty-seven eyes of 47 patients were included in this study. Among the participants there were 16 men and 31 women. Average age ± standard deviation of the study patients was 71 ± 8 [range: 53–84] years. Mean follow-up time was 82 ± 5 months.
## Number of injections
Fourteen of the 47 eyes received intravitreal ranibizumab and 33 eyes received intravitreal aflibercept injections. Mean number of injections was 17.8 ± 3.0 during the intensive phase and 19.5 ± 5.0 during the entire follow-up. During the post-intensive phase, 14 of 47 eyes ($30\%$) were given additional injections with a mean number of 5.7 ± 4.5 [minimum: 1, maximum: 13] per patient, while the remaining $70\%$ received no injections. Of those who did not receive injections during the post-intensive phase, 8 patients ($17\%$) had lasting stability with no need for treatment, 13 patients ($28\%$) had poor visual prognosis due to fibrosis, atrophy or subretinal bleeding, and the remaining 12 ($25\%$) did not receive injections due to compliance problems or budgetary restrictions.
## Visual acuity
In both groups combined, mean BCVA was 54 ± 13 letters at baseline, 65 ± 17 letters at the end of the intensive phase and 45 ± 25 letters at the end of follow-up (Fig. 1). Visual acuity at the end of follow-up was stable in 14 of 47 eyes ($30\%$), improved in 12 eyes ($25\%$) and worsened in 21 eyes ($45\%$) (Fig. 2).Fig. 1Box and Whisker plot graph showing changes in best corrected visual acuity during follow-up: baseline, exit of study and end of follow-up. Abbreviation: BCVA: Best corrected visual acuity. Single asterisk (*) indicates p ≤ 0.05, double asterisk (**) indicates p ≤ 0.01, triple asterisk (***) indicates p ≤ 0.001. Abbreviation: EOF: end of follow-upFig. 2Bar graph showing distribution of eyes grouped by number of lost or gained ETDRS letters at the end of follow-up. Abbreviation: BCVA: Best corrected visual acuity Difference in visual acuity in all eyes at the end of follow-up was statistically significant compared to baseline values ($$p \leq 0.02$$). Mean change in ETDRS letters was statistically significant at the end of the intensive phase compared to baseline (+ 10 ± 14 letters, $p \leq 0.001$), and at the end of follow-up compared to the end of intensive phase (-19 ± 23 letters, $p \leq 0.001$).
In the ranibizumab group, mean BCVA score was 54 ± 13 letters at baseline, 64 ± 18 letters at the end of intensive phase, and 41 ± 25 letters at the end of follow-up (Fig. 1). At the end of follow-up 5 of 14 eyes ($36\%$) had stable visual acuity, 1 eye ($7\%$) improved and 8 eyes ($57\%$) got worse (Fig. 2).
In the aflibercept group, mean BCVA was 55 ± 13 letters at baseline, 65 ± 17 letters at the end of intensive phase, then decreased to 47 ± 25 letters by the end of follow-up (Fig. 1). At the end of follow-up 11 of 33 eyes ($33\%$) had stable visual acuity, 9 eyes ($27\%$) improved, and 13 eyes ($40\%$) worsened (Fig. 2).
Statistical results in separate groups and the total study population were largely consistent. Visual acuity changes were significant at the end of intensive phase compared to baseline ($$p \leq 0.004$$ and $$p \leq 0.001$$ in the ranibizumab and aflibercept-treated group, respectively) and at the end of follow-up compared to the end of intensive phase ($$p \leq 0.001$$ and $p \leq 0.001$ in the ranibizumab and aflibercept-treated group, respectively). Visual acuity decrease at the end of follow-up was significant compared to baseline values in the ranibizumab ($$p \leq 0.03$$) but was not significant in the aflibercept-treated group ($$p \leq 0.22$$). There was no statistically significant difference in BCVA between the two treatment groups at baseline ($$p \leq 0.88$$) and at the end of follow-up ($$p \leq 0.40$$) (Fig. 1).
## Area of atrophy and disease activity
At the end of follow-up, we observed macular atrophy in $96\%$ of the study eyes. Average area of macular atrophy measured on FAF was 7.22 ± 6.31 mm2. Average area of atrophy was 8.00 ± 6.48 mm2 in the ranibizumab group and 6.89 ± 6.31 mm2 in the aflibercept group. Difference between the treatment groups was not statistically significant ($$p \leq 0.47$$) (Fig. 3). The correlation between BCVA at the end of follow-up and the area of atrophy was statistically significant (Spearman R = -0.50, $p \leq 0.001$) (Fig. 4).Fig. 3Graph showing difference in area of macular atrophy (mean ± SD) between ranibizumab and aflibercept treatment groups at the end of follow-up. Difference between groups was statistically not significant ($$p \leq 0$$,4708)Fig. 4Graph showing the relationship between best corrected visual acuity and area of macular atrophy at the end of follow-up. Correlation was statistically significant ($$p \leq 0$$,0002) At the end of follow-up, fluid was detected on OCT images in 7 of 47 eyes ($15\%$) indicating disease activity, 3 of which ($6\%$) were patients not receiving treatment in the post-intensive phase.
## Adverse events
During follow-up, we observed no serious ocular adverse events. Eight of 47 eyes ($17\%$) underwent cataract surgery during the trial. Serious systemic adverse events included stroke, which occurred in two patients during the entire follow-up period.
## Discussion
Short-term benefits of aflibercept and ranibizumab treatment in exudative AMD were extensively published previously [24–28]. In the recent past, some clinical trials had extended follow-up periods [8, 10, 12, 16, 29], a handful papers with long-term real-life data [9, 13, 15, 17, 18, 21, 30–32] and a few systematic reviews of long-term results were published [14, 19].
Our study reports on the 7-year long-term outcomes of a 2-year-long intensive anti-VEGF treatment followed by a 5-year period with a very low number of injections and compares long-term results between groups initially treated with aflibercept or ranibizumab.
In the first two years, our patients were included in the VIEW 2 clinical trial. The VIEW 1 and VIEW 2 clinical trials were two multicenter, phase-3, randomized studies aiming to compare monthly or bimonthly administered intravitreal aflibercept, with monthly administered ranibizumab in patients with exudative AMD [23]. At week 52, proportion of patients losing < 15 letters on ETDRS chart was similar in all treatment groups ranging from $94.4\%$ to $96.1\%$. Anatomic improvements were also comparable between treatment groups. The studies concluded that aflibercept treatment given monthly or bimonthly after 3 initial monthly injections was proven noninferior to monthly ranibizumab. Additionally, occurrence of ocular or systematic adverse events in aflibercept groups was not more frequent. In the second year of the VIEW 1 and 2 studies the observed anatomic and visual improvement was maintained using an as-needed regimen with fixed quarterly dosing [33]. The proportion of patients maintaining BCVA was $91.5\%$ to $92.4\%$, results were similar between treatment groups and aflibercept was noninferior to ranibizumab treatment at week 96.
Results of our 47 patients at the end of year 2 were similar to those of the VIEW 1 and 2 studies, which is not surprising since they were treated according to the same protocol. Visual acuity of our patients improved significantly, mean gain was 11 ETDRS letters achieved by an average of 18 injections. In the VIEW 1 and 2 studies an average improvement of 6.6 to 7.9 letters was seen at week 96 requiring 11.2 to 16.5 injections [33]. In the separate treatment groups results were similar, both ranibizumab and aflibercept treatment resulted in significant visual gain by the end of the second year, showing that aflibercept and ranibizumab treatments were equally effective.
A retrospective study by Eleftheriadou et al. analyzing 2-year treatment results of aflibercept in neovascular AMD had comparable promising results. With a mean of 11.4 injections they observed an average 5.1-letter improvement [34]. Similarly to the former ones, most clinical trials and real-life studies agree on the fact that short-term results of intravitreal anti-VEGF treatment are very positive [10, 11, 21].
A relevant question remains: can this favorable effect be maintained on the long term? In our study we tried to find an answer to that question by observing our patients for an average of 82 months. At the end of our follow-up, mean visual acuity declined in the entire study population compared with both baseline and end-of-intensive-phase values. Our results were similar in the ranibizumab and aflibercept groups. Comparing visual acuity at end of follow-up with the end of intensive phase, the decline was more remarkable and statistically significant, mean decrease was 20 letters in all groups together and results were similar in both treatment groups. Our results are comparable with the SEVEN-UP study [8] which aimed to assess long-term results of ranibizumab treatment in patients with exudative AMD. The SEVEN-UP study included patients who completed 24 months in the ANCHOR or the MARINA trial and then also completed another 24 months in the HORIZON study in the ranibizumab arm. Seven years after the initial entry, patients were called back for a single reassessment visit, when a mean decrease of 8.6 letters ($p \leq 0.005$) was observed from baseline despite continuous treatment in years 2–7 [8].
After the second year, the number of injections given to our patients decreased remarkably. Similarly, in the SEVEN-UP study the number of injections given notably decreased after the exit of the initial ANCHOR or MARINA and HORIZON trials. During the period between the HORIZON exit and end of follow-up (3.4 years on average) mean number of injections given was 6.8 per eye, yet it meant far more treatments per eye compared to our trial (5.7 injections on average during the 4.8 years of the post-intensive phase in those treated, while $70\%$ of our patients did not receive treatment). Such a low number of injections can potentially explain the long-term vision decrease in ranibizumab and aflibercept-treated patients. In a study with a 10-year follow-up Garweg et al. had similar results, 7.3 to 11.9 letters of visual decline in years 3–10 with a mean of 2.8 yearly injections [13]. A meta-analysis by Gerding concluded that the cumulative number of injections correlates to the gained and maintained visual improvement [14].
At the end of follow-up, macular atrophy was observed in $96\%$ of our patients, and the area of atrophy was associated with decreased visual acuity, their relationship was statistically significant. Similarly, in the SEVEN-UP study $98\%$ of eyes had atrophy, the average area of atrophy was 9.4 ± 7.4 mm2, correlation between declined BCVA and the extent of macular atrophy was statistically significant [8]. Despite the treatment, macular atrophy developed in most patients eventually which can be another potential explanation for the long-term visual acuity decline [8]. In the second report of the SEVEN-UP study, authors concluded that macular atrophy had a growth rate of 0.28 mm.2/year, its progression was significantly associated with visual decline and confirmed that final macular atrophy size was significantly related to final visual acuity [22]. Another study by Munk et al. focused on the prevalence and progression of macular atrophy and found that the majority ($73.5\%$) of their patients showed atrophy after long-term anti-VEGF treatment [17]. Berg et al. also concluded that macular atrophic development is the most likely explanation for long-term visual decline [11].
The SEVEN-UP study also showed that disease activity can be observed even long-term in a considerable proportion of patients (exudation was proven by OCT evidence in $68\%$ of the eyes examined) [8]. At the end of our trial we observed disease activity in only $15\%$ of eyes. These data highlight that patients with exudative AMD need long-term follow-up and treatment can be necessary even after several years of disease detection.
We studied the proportion of patients with maintained or improved visual acuity during follow-up as well. BCVA remained stable or improved in $55\%$ of patients in the entire study population, in $43\%$ in the ranibizumab group and in $60\%$ in the aflibercept group.
We observed long-term vision improvement in a remarkable proportion of patients, especially in the aflibercept group despite the very low number of injections after the second year. In these patients, visual improvement could be maintained even after the two years of intensive treatment for a prolonged period with only a few injections or without any. In the SEVEN-UP study, results are similar, $43\%$ of eyes had stable or improved BCVA compared with baseline values [8]. However, the other portion of our patients experienced visual decline, $45\%$ in the whole patient pool, $57\%$ in the ranibizumab group and $39\%$ in the aflibercept group.
In another trial, Peden et al. [ 18] also studied the long-term effects of anti-VEGF (primarily ranibizumab) therapy in exudative AMD for at least 5 years. In contrast to the SEVEN-UP, in this study patients were treated with fixed-interval injections administered in every 4 to 8 weeks. At year 7, mean visual improvement was 12.1 letters, a vastly different result than that of the SEVEN-UP study and what we observed in our trial. These outcomes support the superiority of long-term fixed-interval intensive treatment in exudative AMD.
Another interesting observation regarding long-term treatment by Sagiv et al. was that good final visual acuity was associated with good initial visual acuity, highlighting the benefits of early detection and treatment [20].
Our study has several limitations, first being its retrospective nature, second, its low number of participants, and third, the non-standardized treatment regimen in the post-intensive phase. Patients after exiting the VIEW 2 study returned to real-life conditions and were treated by their physicians on an as-needed basis, which was sometimes severely limited by budgetary constraints resulting in undertreatments. Additionally, patient compliance got worse over the course of the follow-up, not all patients showed up for their check-ups or treatment sessions. After the VIEW 2 study, patients could not continue receiving their original trial medication, since at that time the treatment of choice was ranibizumab, aflibercept was only used later when ranibizumab was not available. This happened in a total of 3 patients on 3, 5 and 5 occasions, respectively.
In conclusion, long-term efficacy of aflibercept and ranibizumab was similar in our patients with exudative AMD. During our nearly 7 year-long follow-up, both drugs proved safe and well tolerated with no serious ocular adverse events. Following a two-year intensive therapy with an as needed regimen, visual acuity was maintained or improved in almost half of the patients in the ranibizumab group, and more than half of the patients in the aflibercept group, with very few injections. In a remarkable proportion of eyes, visual acuity declined severely during the post-intensive period which underlines the need for long-term follow-ups and may indicate a more prolonged intensive therapy.
## References
1. Stewart MW. **Clinical and differential utility of VEGF inhibitors in wet age-related macular degeneration: focus on aflibercept**. *Clin Ophthalmol* (2012.0) **6** 1175-1186. DOI: 10.2147/OPTH.S33372
2. Papadopoulos N, Martin J, Ruan Q, Rafique A, Rosconi MP, Shi E. **Binding and neutralization of vascular endothelial growth factor (VEGF) and related ligands by VEGF Trap, ranibizumab and bevacizumab**. *Angiogenesis* (2012.0) **15** 171-185. DOI: 10.1007/s10456-011-9249-6
3. Garcia-Layana A, Figueroa MS, Araiz J, Ruiz-Moreno JM, Gomez-Ulla F, Arias-Barquet L. **Treatment of exudative age-related macular degeneration: focus on aflibercept**. *Drugs Aging* (2015.0) **32** 797-807. DOI: 10.1007/s40266-015-0300-y
4. Semeraro F, Morescalchi F, Duse S, Parmeggiani F, Gambicorti E, Costagliola C. **Aflibercept in wet AMD: specific role and optimal use**. *Drug Des Devel Ther* (2013.0) **7** 711-722. DOI: 10.2147/DDDT.S40215
5. Do DV, Schmidt-Erfurth U, Gonzalez VH, Gordon CM, Tolentino M, Berliner AJ. **The DA VINCI study: phase 2 primary results of VEGF trap-eye in patients with diabetic macular edema**. *Ophthalmology* (2011.0) **118** 1819-1826. DOI: 10.1016/j.ophtha.2011.02.018
6. Brown DM, Heier JS, Clark WL, Boyer DS, Vitti R, Berliner AJ. **Intravitreal aflibercept injection for macular edema secondary to central retinal vein occlusion: 1-year results from the phase 3 COPERNICUS study**. *Am J Ophthalmol* (2013.0) **155** 429-37 e7. DOI: 10.1016/j.ajo.2012.09.026
7. Holz FG, Roider J, Ogura Y, Korobelnik JF, Simader C, Groetzbach G. **VEGF trap-eye for macular oedema secondary to central retinal vein occlusion: 6-month results of the phase III GALILEO study**. *Br J Ophthalmol* (2013.0) **97** 278-284. DOI: 10.1136/bjophthalmol-2012-301504
8. Rofagha S, Bhisitkul RB, Boyer DS, Sadda SR, Zhang K. **Seven-year outcomes in ranibizumab-treated patients in ANCHOR, MARINA, and HORIZON: a multicenter cohort study (SEVEN-UP)**. *Ophthalmology* (2013.0) **120** 2292-2299. DOI: 10.1016/j.ophtha.2013.03.046
9. Adrean SD, Chaili S, Ramkumar H, Pirouz A, Grant S. **Consistent long-term therapy of neovascular age-related macular degeneration managed by 50 or more anti-VEGF injections using a treat-extend-stop protocol**. *Ophthalmology* (2018.0) **125** 1047-1053. DOI: 10.1016/j.ophtha.2018.01.012
10. Amstutz CA, Fleischhauer J, Zweifel S, Barthelmes D. **Long-term outcome in patients with intravitreal anti-VEGF Therapy for Exudative AMD**. *Klin Monbl Augenheilkd* (2015.0) **232** 533-537. DOI: 10.1055/s-0035-1545673
11. Berg K, Roald AB, Navaratnam J, Bragadottir R. **An 8-year follow-up of anti-vascular endothelial growth factor treatment with a treat-and-extend modality for neovascular age-related macular degeneration**. *Acta Ophthalmol* (2017.0) **95** 796-802. DOI: 10.1111/aos.13522
12. Maguire MG, Martin DF, Ying GS, Jaffe GJ, Daniel E. **Five-year outcomes with anti-vascular endothelial growth factor treatment of neovascular age-related macular degeneration: the comparison of age-related macular degeneration treatments trials**. *Ophthalmology* (2016.0) **123** 1751-1761. DOI: 10.1016/j.ophtha.2016.03.045
13. Garweg JG, Zirpel JJ, Gerhardt C, Pfister IB. **The fate of eyes with wet AMD beyond four years of anti-VEGF therapy**. *Graefes Arch Clin Exp Ophthalmol* (2018.0) **256** 823-831. DOI: 10.1007/s00417-018-3907-y
14. Gerding H. **Long-term results of intravitreal anti-VEGF injections in wet AMD: a meta-analysis**. *Klin Monbl Augenheilkd* (2016.0) **233** 471-474. DOI: 10.1055/s-0041-111835
15. Gillies MC, Campain A, Barthelmes D, Simpson JM, Arnold JJ, Guymer RH. **Long-term outcomes of treatment of neovascular age-related macular degeneration: data from an observational study**. *Ophthalmology* (2015.0) **122** 1837-1845. DOI: 10.1016/j.ophtha.2015.05.010
16. Keenan TD, Vitale S, Agron E, Domalpally A, Antoszyk AN, Elman MJ. **Visual acuity outcomes after anti-vascular endothelial growth factor treatment for neovascular age-related macular degeneration: age-related eye disease study 2 report number 19**. *Ophthalmol Retina* (2020.0) **4** 3-12. DOI: 10.1016/j.oret.2019.06.001
17. Munk MR, Ceklic L, Ebneter A, Huf W, Wolf S, Zinkernagel MS. **Macular atrophy in patients with long-term anti-VEGF treatment for neovascular age-related macular degeneration**. *Acta Ophthalmol* (2016.0) **94** e757-e764. DOI: 10.1111/aos.13157
18. Peden MC, Suner IJ, Hammer ME, Grizzard WS. **Long-term outcomes in eyes receiving fixed-interval dosing of anti-vascular endothelial growth factor agents for wet age-related macular degeneration**. *Ophthalmology* (2015.0) **122** 803-808. DOI: 10.1016/j.ophtha.2014.11.018
19. Rasmussen A, Sander B. **Long-term longitudinal study of patients treated with ranibizumab for neovascular age-related macular degeneration**. *Curr Opin Ophthalmol* (2014.0) **25** 158-163. DOI: 10.1097/ICU.0000000000000050
20. Sagiv O, Zloto O, Moroz I, Moisseiev J. **Different clinical courses on long-term follow-up of age-related macular degeneration patients treated with intravitreal anti-vascular endothelial growth factor injections**. *Ophthalmologica* (2017.0) **238** 217-225. DOI: 10.1159/000479437
21. Wecker T, Ehlken C, Buhler A, Lange C, Agostini H, Bohringer D. **Five-year visual acuity outcomes and injection patterns in patients with pro-re-nata treatments for AMD, DME, RVO and myopic CNV**. *Br J Ophthalmol* (2017.0) **101** 353-359. PMID: 27215744
22. Bhisitkul RB, Mendes TS, Rofagha S, Enanoria W, Boyer DS, Sadda SR. **Macular atrophy progression and 7-year vision outcomes in subjects from the ANCHOR, MARINA, and HORIZON studies: the SEVEN-UP study**. *Am J Ophthalmol* (2015.0) **159** 915-24 e2. DOI: 10.1016/j.ajo.2015.01.032
23. Heier JS, Brown DM, Chong V, Korobelnik JF, Kaiser PK, Nguyen QD. **Intravitreal aflibercept (VEGF trap-eye) in wet age-related macular degeneration**. *Ophthalmology* (2012.0) **119** 2537-2548. DOI: 10.1016/j.ophtha.2012.09.006
24. Brown DM, Kaiser PK, Michels M, Soubrane G, Heier JS, Kim RY. **Ranibizumab versus verteporfin for neovascular age-related macular degeneration**. *N Engl J Med* (2006.0) **355** 1432-1444. DOI: 10.1056/NEJMoa062655
25. Brown DM, Michels M, Kaiser PK, Heier JS, Sy JP, Ianchulev T. **Ranibizumab versus verteporfin photodynamic therapy for neovascular age-related macular degeneration: two-year results of the ANCHOR study**. *Ophthalmology* (2009.0) **116** 57-65 e5. DOI: 10.1016/j.ophtha.2008.10.018
26. Heier JS, Boyer D, Nguyen QD, Marcus D, Roth DB, Yancopoulos G. **The 1-year results of CLEAR-IT 2, a phase 2 study of vascular endothelial growth factor trap-eye dosed as-needed after 12-week fixed dosing**. *Ophthalmology* (2011.0) **118** 1098-1106. DOI: 10.1016/j.ophtha.2011.03.020
27. Nguyen QD, Shah SM, Browning DJ, Hudson H, Sonkin P, Hariprasad SM. **A phase I study of intravitreal vascular endothelial growth factor trap-eye in patients with neovascular age-related macular degeneration**. *Ophthalmology* (2009.0) **116** 2141-2148. DOI: 10.1016/j.ophtha.2009.04.030
28. Rosenfeld PJ, Brown DM, Heier JS, Boyer DS, Kaiser PK, Chung CY. **Ranibizumab for neovascular age-related macular degeneration**. *N Engl J Med* (2006.0) **355** 1419-1431. DOI: 10.1056/NEJMoa054481
29. Evans RN, Reeves BC, Phillips D, Muldrew KA, Rogers C, Harding SP. **Long-term visual outcomes after release from protocol in patients who participated in the Inhibition of VEGF in Age-related Choroidal Neovascularisation (IVAN) Trial**. *Ophthalmology* (2020.0) **127** 1191-1200. DOI: 10.1016/j.ophtha.2020.03.020
30. 30.Arpa C, Khalid H, Chandra S, Wagner S, Fasler K, Faes L, et al. Ten-year survival trends of neovascular age-related macular degeneration at first presentation. Br J Ophthalmol. 2021;105(12):1688–95.
31. Finger RP, Puth MT, Schmid M, Barthelmes D, Guymer RH, Gillies M. **Lifetime Outcomes of anti-vascular endothelial growth factor treatment for neovascular age-related macular degeneration**. *JAMA Ophthalmol* (2020.0) **138** 1234-1240. DOI: 10.1001/jamaophthalmol.2020.3989
32. Fu DJ, Keenan TD, Faes L, Lim E, Wagner SK, Moraes G. **Insights from survival analyses during 12 years of anti-vascular endothelial growth factor therapy for neovascular age-related macular degeneration**. *JAMA Ophthalmol* (2021.0) **139** 57-67. DOI: 10.1001/jamaophthalmol.2020.5044
33. Schmidt-Erfurth U, Kaiser PK, Korobelnik JF, Brown DM, Chong V, Nguyen QD. **Intravitreal aflibercept injection for neovascular age-related macular degeneration: ninety-six-week results of the VIEW studies**. *Ophthalmology* (2014.0) **121** 193-201. DOI: 10.1016/j.ophtha.2013.08.011
34. Eleftheriadou M, Vazquez-Alfageme C, Citu CM, Crosby-Nwaobi R, Sivaprasad S, Hykin P. **Long-term outcomes of aflibercept treatment for neovascular age-related macular degeneration in a clinical setting**. *Am J Ophthalmol* (2017.0) **174** 160-168. DOI: 10.1016/j.ajo.2016.09.038
|
---
title: 'Determinants of blood pressure and blood glucose control in patients with
co-morbid hypertension and type 2 diabetes mellitus in Ghana: A hospital-based cross-sectional
study'
authors:
- Yakubu Alhassan
- Adwoa Oforiwaa Kwakye
- Andrews K. Dwomoh
- Emmanuella Baah-Nyarkoh
- Vincent Jessey Ganu
- Bernard Appiah
- Irene A. Kretchy
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022155
doi: 10.1371/journal.pgph.0001342
license: CC BY 4.0
---
# Determinants of blood pressure and blood glucose control in patients with co-morbid hypertension and type 2 diabetes mellitus in Ghana: A hospital-based cross-sectional study
## Abstract
Hypertension and diabetes are major risk factors for cardiovascular diseases and optimal control of blood pressure (BP) and blood glucose are associated with reduced cardiovascular disease events. This study, therefore, sought to estimate the prevalence and associated factors of controlled BP and blood glucose levels among patients diagnosed with both hypertension and Type 2- diabetes mellitus (T2DM). A quantitative cross-sectional study was conducted in a primary health setting in Ghana among patients 18 years and older diagnosed with both hypertension and T2DM. Pearson’s chi-square was used to assess the association between BP and blood glucose levels and the independent variables. The multivariable binary logistic regression model was used to assess the adjusted odds of controlled BP and blood glucose levels. Among the 329 participants diagnosed with both hypertension and T2DM, $41.3\%$ ($95\%$ CI: 36.1–$46.8\%$) had controlled BP, $57.1\%$ ($95\%$ CI: 51.7–$62.4\%$) had controlled blood glucose whilst $21.8\%$ ($95\%$ CI: 17.7–$26.7\%$) had both controlled BP and blood glucose levels. Increased age, non-formal education, non-married, employed, single-dose anti-hypertensives or anti-diabetic medications, and hyperlipidaemia or stroke co-morbidities were positively associated with controlled BP levels. Being female, married, taking 2 or more anti-hypertensive medications, and moderate to high medication-related burden were positively associated with controlled blood glucose levels. In terms of both controlled BP and blood glucose levels, being employed, reduced income level, being registered with national health insurance, single anti-diabetes or anti-hypertensive medications, hyperlipidaemia or stroke co-morbidities, and moderate to high medication-related burden were positively associated with having both controlled BP and blood glucose levels. One in five patients with hypertension and T2DM had both BP and blood glucose levels under control. The benefits and risks of blood pressure and blood glucose targets should thus be factored into the management of patients with hypertension and T2DM.
## Introduction
Hypertension and diabetes remain major public health threats worldwide [1,2]. When these cardio-metabolic conditions co-exist in an individual, there is a worsening of both glycaemic and cardiovascular endpoints [3,4]. Both diseases have similarities in risk factors including lifestyle, dyslipidaemia, familial, and racial as well as complications [5,6]. These complications can be categorized into micro and macrovascular complications. Myocardial infarction, stroke, peripheral vascular disease, coronary artery disease, and congestive heart failure are examples of macrovascular problems, while retinopathy, nephropathy, and neuropathy are examples of microvascular complications [7,8].
A significant contributor to the onset and development of diabetes-related illnesses and complications is inadequate and poor glycaemic control. This can significantly increase medical costs, diminish the quality of life, and reduce life expectancy [4]. Research has shown that improving glycaemic control can help patients live longer, have an improved quality of life, and delay the development and progression of diabetic complications [9]. Also, improving glycaemic control significantly lowers expenditures associated with the management of diabetes [10]. Blood glucose levels must be tightly controlled to delay the onset of diabetes and its related complications.
Clinical recommendations in diabetes care support have varied blood pressure (BP) values; nonetheless, most advocate lower levels compared to people without diabetes [11,12]. The lowering of BP in patients with diabetes results in a significant reduction in cardiovascular problems [5]. The target BP for diabetic patients should be less than $\frac{140}{90}$ mmHg, according to the Joint National Committee’s eighth report (JNC 8), and most patients will need to take two or more antihypertensive drugs to do so [12]. Increased patient and healthcare provider understanding about the illness, access to care, suitable lifestyle changes, evidence-based treatment, high levels of medication adherence, and thorough follow-up are all components of high-quality BP control [13–15]. In addition, studies have shown that old age, chronic renal diseases, longer duration of hypertension, and uncontrolled diabetes mellitus are significant risk factors for poor BP control [16,17].
According to a Ghana Ministry of Health report, diabetes and hypertension are among the top fifteen causes of outpatient visits [18,19]. Effective management of these patient groups requires substantive knowledge of the patient characteristics and other factors affecting their BP control. Even though there have been studies on factors associated with blood pressure in diabetes [20,21]. Literature on factors associated with regulated BP as well as controlled blood glucose levels among persons with co-morbid hypertension and diabetes is limited. This study, therefore, assessed the factors associated with BP and blood glucose control among patients with co-morbid hypertension and type 2 diabetes mellitus (T2DM) at an outpatient department of a lower-middle-income country hospital.
## Study design and context
A hospital-based cross-sectional study was conducted at the Adabraka Polyclinic which is a public primary health facility in Ghana. The clinic provides all essential general healthcare services including assessment and management of chronic conditions such as hypertension and diabetes mellitus. All the commonly prescribed antihypertensive and oral hypoglycaemic agents are included in the national essential drugs list and covered by the National Health Insurance Scheme (NHIS) [22]. This is in line with the national policy on non-communicable diseases (NCDs) which provides guidelines on the primary prevention (e.g., health promotion), secondary and tertiary prevention (e.g., screening, and early detection, clinical care/case management) of NCDs [19]. The Outpatient Department (OPD) holds twice-weekly clinics for patients with hypertension and T2DM and provides services for an average of eighty patients daily.
## Study participants
The study engaged and recruited adults aged at least 18 years known to have both hypertension and T2DM with documented evidence of these clinical diagnoses in their clinical files. Patients who were on anti-hypertension and anti-diabetes medications for at least six months prior to the study were included in the study. Participants excluded from the study include those with Type 1 diabetes mellitus, gestational diabetes, and evidence of impaired cognitive function. Participants provided informed consent prior to their inclusion in the study.
This study is part of a previous study among people with co-morbid hypertension and T2DM, with a sample size of 326 estimated, which was based on a $74.05\%$ prevalence rate of medication adherence [4], $95\%$ confidence interval, $5\%$ error margin and a $10\%$ non-response rate. The study participants were recruited systematically between October 2021 and November 2021.
Patients’ medical records were reviewed to validate the diagnosis of hypertension and T2DM diagnosis.
## Measures
Data were collected using a validated questionnaire that included general information on socio-demographic characteristics (e.g., age, sex, marital status, religion, education level, occupation, monthly income, payment method for drugs, monthly expenditure), clinical characteristics (number of medications, frequency of daily dose of medication, presence of co-morbidities, blood glucose level, systolic blood pressure (SBP), diastolic blood pressure (DBP), duration since diagnosis and frequency of follow-up visits, and family history of hypertension and T2DM and medication burden.
The patient’s BP was measured using a manual mercury sphygmomanometer. Patients were instructed to rest for 3 to 5 minutes before having their BP measured. They sat back in a chair, their back supported. The nurses ensured that the appropriate cuff size was used for BP measurement based on the patient’s arm diameter. The readings were taken three times and the average was taken. In this study, participants with SBP below 140mmHg and DBP below 90mmHg were considered to have controlled BP level.
The fasting blood sugar was measured using a certified automated glucometer (GOLD-ACCU). The finger surface was first cleaned with $75\%$ alcohol, and then the sterilised needle was used to prick it. The very first drop of blood was discarded. After 5 seconds, the reading appeared after the blood sample was placed on the strip. These tasks were carried out by well-trained nurses. In this study participants with blood glucose level below 7.0mmol/L were considered to have controlled blood glucose level.
Pill burden was assessed with the 41-item Living with Medicines Questionnaire (LMQ-3) with the overall score ranging from 41 to 205 [23,24]. Pill burden was categorized according to score range as no burden at all (scores in range 41–73); minimum burden (scores in range 74–106); moderate burden (scores in range 107–139); high burden (scores in range 140–172); and Extremely high burden (scores in range 173–205). The LMQ-3 was dichotomized for analysis in this study as moderate/high burden against minimum burden. The scale is reliable with a Cronbach’s alpha score 0.9208 computed in this study.
The 5-item medication adherence report scale (MARS-5) measured adherence behaviour [25]. Each item is rated from 1 = always to 5 = never with the composite score ranging from 5 to 25. This study reports a Cronbach’s alpha score of 0.8568 for the MARS-5.
## Statistical analysis
Stata IC version 16 (StataCorp, College Station, TX, US) was used for analysis. Descriptive statistics were presented using frequency and percentages for categorical variables, the mean and standard deviation for normally distributed continuous variables, and median and interquartile range (IQR) for non-normally distributed continuous variables. The distribution of SBP, DBP, and blood glucose concentration was described using the box and whiskers plots. The percentage of participants with controlled and uncontrolled levels of BP, blood glucose level and both were presented and corresponding $95\%$ confidence interval were estimated using the binomial exact estimation approach.
The Pearson’s chi-square test was used to assess the bivariate association between the socio-demographic, clinical and medication-related characteristics observed in the study and the BP and blood glucose levels among the participants. The multiple binary logistic regression model with robust standard errors was used to assess the adjusted odds of controlled BP, controlled blood glucose and controlled levels of both BP and blood glucose among participants across the various characteristics.
Multicollinearity between the variables was assessed using the variance inflation factor (VIF) which recorded a mean VIF of 3.01 (range: 1.72 to 6.98) which is within the acceptable range of less than 10 all three models. The area under the receiver operating characteristics curve was 0.8552 ($95\%$ CI: 0.8130 to 0.8975) for the BP level model, 0.7884 ($95\%$ CI: 0.7390 to 0.8376) for the blood glucose model and 0.8537 ($95\%$ CI: 0.8041 to 0.9033) for both controlled BP and blood glucose model which were all above $70\%$. The Hosmer-Lemeshow goodness of fit test was insignificant for both the BP level (p-value = 0.073), blood glucose level (p-value = 0.268) and both BP and blood glucose (p-value = 0.636) model indicating models were appropriately fitted. All statistical analysis in this study were considered significant with p-values less than 0.05.
## Ethics statement
The study was approved by Ghana Health Service Ethical Review Committee (GHS-ERC $\frac{043}{09}$/21). Participants were chosen for the study depending on their willingness to participate. Informed consent for participation was also obtained from each patient.
## Background characteristics of study participants
The mean age of the 329 study participants was 57.5 ± 13.2 years. More than half ($56.2\%$) of them were female and $58.4\%$ were married. Less than a fifth ($17.9\%$) had no formal education whilst $21.3\%$ had a tertiary level of education. The median monthly expenditure on medication was 50.0 cedis (thus, 8.00 USD) with a fifth ($20.7\%$) paying for all their medication using the health insurance. Family history of hypertension and diabetes were $68.1\%$ and $49.8\%$ respectively. Most participants ($94.2\%$) of the participants were on amlodipine as an anti-hypertensive medication. The median number of antihypertensive medications taken was 2 (IQR: 1 to 3) with $70.2\%$ of the participants taking at least 2 different medications. The median number of anti-diabetic medications taken was 2 (IQR: 1 to 2) with $51.1\%$ on at least 2 medications (Table 1).
**Table 1**
| Characteristics | Frequency (%) |
| --- | --- |
| Overall | 329 (100.0) |
| SOCIO-DEMOGRAPHICS | |
| Sex | |
| Male | 144 (43.8) |
| Female | 185 (56.2) |
| Age, Mean [±SD] | 57.5 [±13.2] |
| Age group | |
| <50 | 80 (24.3) |
| 50–59 years | 108 (32.8) |
| 60–69 | 78 (23.7) |
| 70+ | 63 (19.1) |
| Marital status | |
| Single | 84 (25.5) |
| Married | 192 (58.4) |
| Divorced | 25 (7.6) |
| Others | 28 (8.5) |
| Highest education | |
| No formal education | 59 (17.9) |
| Primary | 72 (21.9) |
| Secondary | 128 (38.9) |
| Tertiary | 70 (21.3) |
| Occupation | |
| Unemployed | 41 (12.5) |
| Trader/artisan | 176 (53.5) |
| Professional | 53 (16.1) |
| Retired | 48 (14.6) |
| Others | 11 (3.3) |
| Monthly income | |
| 0–500 cedis (0–80 USD) | 150 (45.6) |
| 501–1000 cedis (81.0–160 USD) | 121 (36.8) |
| >1000 (>160 USD) | 58 (17.6) |
| CLINICAL AND MEDICATION RELATED | |
| Monthly expenditure on drugs, Median (IQR) | 50.0 (30.0, 100.0) |
| Monthly expenditure on drugs | |
| None/ Health insurance | 68 (20.7) |
| <50 cedis (<8.0 USD) | 101 (30.7) |
| 51–100 cedis (8.1–16.0 USD) | 88 (26.7) |
| >100 cedis (>16.0 USD) | 72 (21.9) |
| Number of anti-hypertensive medications, Median (IQR) | 2 (1, 3) |
| Number of anti-hypertensive medications | |
| <2 medicine | 98 (29.8) |
| 2+ medicines | 231 (70.2) |
| Number of anti-diabetes medications, Median (IQR) | 2.0 (1.0, 2.0) |
| Number of anti-diabetes medications | |
| <2 medicine | 161 (48.9) |
| 2+ medicines | 168 (51.1) |
| Other medications | |
| Dyslipidaemia medications | |
| No | 264 (80.2) |
| Yes | 65 (19.8) |
| Soluble aspirin | |
| No | 314 (95.4) |
| Yes | 15 (4.6) |
| Frequency of daily dose of medication | |
| Once | 145 (44.1) |
| Twice | 181 (55.0) |
| Three times | 3 (0.9) |
| Have co-morbidities | |
| No | 269 (81.8) |
| Hyperlipidaemia | 57 (17.3) |
| Stroke | 3 (0.9) |
| Family history of hypertension | |
| Yes | 224 (68.1) |
| No | 105 (31.9) |
| Family history of diabetes mellitus | |
| Yes | 164 (49.8) |
| No | 165 (50.2) |
| Duration since diagnosis of hypertension in years, Median (IQR) | 5.0 (2.0, 6.0) |
| Duration since diagnosis of hypertension | |
| <5 years | 151 (45.9) |
| 5+ years | 178 (54.1) |
| Duration since diagnosis of Diabetes Mellitus in years, Median (IQR) | 3.0 (1.0, 5.0) |
| Duration since diagnosis of Diabetes Mellitus | |
| <5 years | 243 (73.9) |
| 5+ years | 86 (26.1) |
| Frequency of follow-up | |
| Every 2 weeks | 22 (6.7) |
| Monthly | 86 (26.1) |
| Every 2 months | 221 (67.2) |
| Medication related burden | |
| Minimum burden | 228 (69.3) |
| Moderate/high burden | 101 (30.7) |
| Medication adherence | |
| Non-adherence | 208 (63.2) |
| Adherence | 121 (36.8) |
Hyperlipidaemia and/or stroke co-morbidities were prevalent among $17.3\%$ and $0.9\%$ of the participants respectively. A medication-related burden was moderate/high among $30.7\%$ whilst adherence to medication was $36.8\%$ (Table 1).
## Blood pressure and blood glucose levels among participants
The median SBP was 140 mmHg (IQR: 128 to 157mmHg), DBP was 85 mmHg (IQR: 77 to 92mmHg) and blood glucose was 6.5 mmol/L (IQR: 5.6 to 8.3 mmol/L). ( Fig 1).
**Fig 1:** *Systolic blood pressure, diastolic blood pressure, and blood glucose levels among study participants.*
Less than half ($41.3\%$) of the participants had controlled BP levels (SBP<140mmHg & DBP<90mmHg) ($95\%$ CI: $36.0\%$ - $46.8\%$) whilst most ($57.1\%$) had controlled blood glucose levels (<7.0 mmol/l) ($95\%$ CI: $51.7\%$ - $62.4\%$). About a fifth ($23.4\%$) had uncontrolled BP and blood glucose levels, $35.3\%$ had uncontrolled blood pressure but controlled blood glucose levels, $19.5\%$ had controlled BP levels but uncontrolled blood glucose levels and $21.8\%$ had controlled BP and blood glucose levels (Table 2).
**Table 2**
| Outcomes | Frequency (N = 329) | Percentage | 95% CI |
| --- | --- | --- | --- |
| Blood pressure level | | | |
| Uncontrolled (SBP≥140mmHg or DBP ≥90mmHg) | 193.0 | 58.7 | [53.2, 63.9] |
| Controlled (SBP<140mmHg & DBP<90mmHg) | 136.0 | 41.3 | [36.1, 46.8] |
| Blood glucose level | | | |
| Uncontrolled blood glucose (7.0+ mmol/L) | 141.0 | 42.9 | [37.6, 48.3] |
| Controlled blood glucose (<7.0 mmol/L) | 188.0 | 57.1 | [51.7, 62.4] |
| Blood pressure and blood glucose level | | | |
| Uncontrolled BP & uncontrolled blood glucose | 77.0 | 23.4 | [19.1, 28.3] |
| Uncontrolled BP but controlled blood glucose | 116.0 | 35.3 | [30.3, 40.6] |
| Controlled BP but uncontrolled blood glucose | 64.0 | 19.5 | [15.5, 24.1] |
| Controlled BP & controlled blood glucose | 72.0 | 21.8 | [17.7, 26.7] |
## Blood pressure control levels only
The bivariate analysis showed that the statistically significant socio-demographic factors associated with the controlled levels of BP among the participants included age group ($$p \leq 0.024$$), marital status ($$p \leq 0.001$$) and the highest level of education ($$p \leq 0.001$$). The statistically significant clinical and medication-related factors associated with BP levels included expenditure on drugs ($$p \leq 0.014$$), number of anti-hypertensive medications ($p \leq 0.001$), number of anti-diabetic medications ($p \leq 0.001$), frequency of daily dose of medication ($p \leq 0.001$), duration of hypertension diagnosis ($$p \leq 0.031$$), and frequency of follow-up visits to clinic ($$p \leq 0.021$$) (Table 3).
**Table 3**
| Unnamed: 0 | Total | Controlled BP only | Controlled BP only.1 | Controlled blood glucose only | Controlled blood glucose only.1 | Controlled BP and blood glucose level | Controlled BP and blood glucose level.1 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Variables & categories | N | n (%) | P-value | n (%) | P-value | n (%) | P-value |
| Overall | 329 | 136 (41.3) | | 188 (57.1) | | 72 (21.8) | |
| SOCIO-DEMOGRAPHICS | | | | | | | |
| Sex | | | 0.910 | | 0.001 | | 0.140 |
| Male | 144 | 60 (41.7) | | 68 (47.2) | | 26 (18.1) | |
| Female | 185 | 76 (41.1) | | 120 (64.9) | | 46 (24.9) | |
| Age group | | | 0.024 | | 0.023 | | 0.240 |
| <50 | 80 | 22 (27.5) | | 51 (63.7) | | 13 (16.3) | |
| 50–59 years | 108 | 46 (42.6) | | 67 (62.0) | | 29 (26.9) | |
| 60–69 | 78 | 36 (46.2) | | 33 (42.3) | | 14 (17.9) | |
| 70+ | 63 | 32 (50.8) | | 37 (58.7) | | 16 (25.4) | |
| Marital status | | | 0.001 | | 0.820 | | 0.140 |
| Single | 84 | 48 (57.1) | | 50 (59.5) | | 24 (28.6) | |
| Married | 192 | 65 (33.9) | | 107 (55.7) | | 35 (18.2) | |
| Divorced/widowed/separated | 53 | 23 (43.4) | | 31 (58.5) | | 13 (24.5) | |
| Highest education | | | 0.001 | | 0.006 | | <0.001 |
| No formal education | 59 | 33 (55.9) | | 36 (61.0) | | 20 (33.9) | |
| Primary | 72 | 23 (31.9) | | 40 (55.6) | | 13 (18.1) | |
| Secondary | 128 | 61 (47.7) | | 84 (65.6) | | 36 (28.1) | |
| Tertiary | 70 | 19 (27.1) | | 28 (40.0) | | 3 (4.3) | |
| Employment status | | | 0.340 | | 0.049 | | 0.004 |
| Unemployed/retired | 89 | 33 (37.1) | | 43 (48.3) | | 10 (11.2) | |
| Employed | 240 | 103 (42.9) | | 145 (60.4) | | 62 (25.8) | |
| Monthly income | | | 0.310 | | 0.011 | | 0.280 |
| 0–500 cedis (0–80 USD) | 150 | 61 (40.7) | | 93 (62.0) | | 38 (25.3) | |
| 501–1000 cedis (81.0–160 USD) | 121 | 46 (38.0) | | 72 (59.5) | | 25 (20.7) | |
| >1000 (>160 USD) | 58 | 29 (50.0) | | 23 (39.7) | | 9 (15.5) | |
| CLINICAL AND MEDICATION RELATED | | | | | | | |
| Source of expenditure on drugs | | | 0.014 | | 0.160 | | 0.092 |
| Health insurance only | 68 | 37 (54.4) | | 44 (64.7) | | 20 (29.4) | |
| Out-of-pocket | 261 | 99 (37.9) | | 144 (55.2) | | 52 (19.9) | |
| Number of hypertensive medications | | | <0.001 | | 0.015 | | <0.001 |
| <2 medicine | 98 | 63 (64.3) | | 46 (46.9) | | 34 (34.7) | |
| 2+ medicines | 231 | 73 (31.6) | | 142 (61.5) | | 38 (16.5) | |
| Number of diabetes medications | | | <0.001 | | 0.660 | | <0.001 |
| <2 medicine | 161 | 85 (52.8) | | 94 (58.4) | | 51 (31.7) | |
| 2+ medicines | 168 | 51 (30.4) | | 94 (56.0) | | 21 (12.5) | |
| dyslipidaemia medications | | | 0.970 | | 0.280 | | 0.023 |
| No | 264 | 109 (41.3) | | 147 (55.7) | | 51 (19.3) | |
| Yes | 65 | 27 (41.5) | | 41 (63.1) | | 21 (32.3) | |
| Soluble Aspirin | | | 0.240 | | 0.400 | | 0.036 |
| Does not take medication | 314 | 132 (42.0) | | 181 (57.6) | | 72 (22.9) | |
| Takes medication | 15 | 4 (26.7) | | 7 (46.7) | | 0 (0.0) | |
| Frequency of daily dose of medication | | | <0.001 | | 0.002 | | 0.460 |
| Once | 145 | 43 (29.7) | | 97 (66.9) | | 29 (20.0) | |
| Twice/thrice | 184 | 93 (50.5) | | 91 (49.5) | | 43 (23.4) | |
| Have co-morbidities | | | 0.350 | | 0.026 | | 0.002 |
| | 269 | 108 (40.1) | | 146 (54.3) | | 50 (18.6) | |
| Hyperlipidaemia /Stroke | 60 | 28 (46.7) | | 42 (70.0) | | 22 (36.7) | |
| Family history of hypertension | | | 0.530 | | 0.056 | | 0.570 |
| Yes | 224 | 90 (40.2) | | 136 (60.7) | | 51 (22.8) | |
| No | 105 | 46 (43.8) | | 52 (49.5) | | 21 (20.0) | |
| Family history of diabetes mellitus | | | 0.620 | | 0.950 | | 0.810 |
| Yes | 164 | 70 (42.7) | | 94 (57.3) | | 35 (21.3) | |
| No | 165 | 66 (40.0) | | 94 (57.0) | | 37 (22.4) | |
| Duration since diagnosis of hypertension | | | 0.031 | | 0.950 | | 0.430 |
| <5 years | 151 | 72 (47.7) | | 86 (57.0) | | 36 (23.8) | |
| 5+ years | 178 | 64 (36.0) | | 102 (57.3) | | 36 (20.2) | |
| Duration since diagnosis of T2DM | | | 0.095 | | 0.120 | | 0.960 |
| <5 years | 243 | 107 (44.0) | | 145 (59.7) | | 53 (21.8) | |
| 5+ years | 86 | 29 (33.7) | | 43 (50.0) | | 19 (22.1) | |
| Frequency of follow-up | | | 0.021 | | 0.007 | | 0.060 |
| 2 weeks/monthly | 108 | 35 (32.4) | | 73 (67.6) | | 17 (15.7) | |
| Every 2 month | 221 | 101 (45.7) | | 115 (52.0) | | 55 (24.9) | |
| Medication related burden | | | 0.200 | | 0.430 | | 0.046 |
| Minimum burden | 228 | 89 (39.0) | | 127 (55.7) | | 43 (18.9) | |
| Moderate/high burden | 101 | 47 (46.5) | | 61 (60.4) | | 29 (28.7) | |
| Adherence to medication | | | 0.810 | | 0.180 | | 0.210 |
| Non-adherence | 208 | 87 (41.8) | | 113 (54.3) | | 41 (19.7) | |
| Adherence | 121 | 49 (40.5) | | 75 (62.0) | | 31 (25.6) | |
## Blood glucose control levels only
The bivariate analysis showed that sex ($$p \leq 0.001$$), age group ($$p \leq 0.023$$), highest education ($$p \leq 0.006$$), employment status ($$p \leq 0.049$$) and monthly income ($$p \leq 0.011$$) were the statistically significant socio-demographic factors associated with the blood glucose levels of participants. The statistically significant clinical and medication related factors associated with blood glucose level included number of anti-hypertensive medications ($$p \leq 0.015$$), frequency of daily dose of medication ($$p \leq 0.002$$), presence of hyperlipidaemia or stroke co-morbidity ($$p \leq 0.026$$) and frequency of follow-up visits to clinic ($$p \leq 0.007$$) (Table 3).
## Blood pressure and blood glucose control levels
The bivariate analysis showed that highest level of education ($p \leq 0.001$) and employment status ($$p \leq 0.004$$) were the only socio-demographic characteristics associated with controlled BP and blood glucose levels. Number of anti-hypertensive medications ($p \leq 0.001$), number of anti-diabetic medication ($p \leq 0.001$), dyslipidaemia medication ($$p \leq 0.023$$), soluble aspirin ($$p \leq 0.036$$), presence of hyperlipidaemia/stroke comorbidities ($$p \leq 0.002$$), and medication related burden ($$p \leq 0.046$$) were the clinical- and medication- related factors associated with controlled BP and blood glucose level (Table 3).
## Blood pressure control levels
In the multivariable binary logistic regression model, the adjusted odds of controlled BP compared to those less than 50 years old was over 5 times high among the 60–69 years old (AOR: 5.16, $95\%$ CI: 1.77–15.02, $$p \leq 0.003$$) and over 9 times higher among the 70 years and older (AOR: 9.44, $95\%$ CI: 3.06–29.16, $p \leq 0.001$). Compared to the married, the odds of controlled BP were over 5 times high among the single (AOR: 5.70, $95\%$ CI: 2.45–13.26, $$p \leq 0.003$$). Also, compared to those with no formal education, the odds of controlled BP was $86\%$ less among those with primary school education (AOR: 0.14, $95\%$ CI: 0.05–0.35, $p \leq 0.001$) and $68\%$ less among those with tertiary education (AOR: 0.32, $95\%$ CI: 0.12–0.82, $$p \leq 0.018$$). The adjusted odds of controlled BP was over 5 times high among the employed (AOR: 5.47, $95\%$ CI: 2.30–12.97, $p \leq 0.001$) (Table 4).
**Table 4**
| Unnamed: 0 | Multiple binary logistic regression model | Multiple binary logistic regression model.1 | Multiple binary logistic regression model.3 | Multiple binary logistic regression model.4 | Multiple binary logistic regression model.6 | Multiple binary logistic regression model.7 |
| --- | --- | --- | --- | --- | --- | --- |
| | Controlled blood pressure level | Controlled blood pressure level | High blood glucose level | High blood glucose level | Controlled Blood pressure and blood glucose level | Controlled Blood pressure and blood glucose level |
| Variables and categories | AOR [95% CI] | P-value | AOR [95% CI] | P-value | AOR [95% CI] | P-value |
| SOCIO-DEMOGRAPHICS | | | | | | |
| Sex | | | | | | |
| Male | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Female | 0.93 [0.47, 1.84] | 0.828 | 2.04 [1.04, 4.00] | 0.038 | 1.31 [0.49, 3.48] | 0.588 |
| Age group | | | | | | |
| <50 | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| 50–59 years | 0.97 [0.37, 2.56] | 0.952 | 1.38 [0.65, 2.92] | 0.401 | 1.09 [0.36, 3.35] | 0.877 |
| 60–69 | 5.16 [1.77, 15.02] | 0.003 | 0.65 [0.27, 1.55] | 0.331 | 1.25 [0.31, 5.00] | 0.753 |
| 70+ | 9.44 [3.06, 29.16] | <0.001 | 1.22 [0.37, 3.99] | 0.743 | 2.33 [0.66, 8.20] | 0.188 |
| Marital status | | | | | | |
| Married | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Single | 5.70 [2.45, 13.26] | <0.001 | 0.50 [0.27, 0.96] | 0.037 | 0.88 [0.41, 1.88] | 0.736 |
| Divorced/widowed/separated | 2.25 [0.96, 5.29] | 0.063 | 0.91 [0.35, 2.33] | 0.837 | 1.97 [0.78, 4.98] | 0.151 |
| Highest education | | | | | | |
| No formal education | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Primary | 0.14 [0.05, 0.35] | <0.001 | 0.48 [0.21, 1.09] | 0.081 | 0.16 [0.05, 0.49] | 0.001 |
| Secondary | 0.68 [0.30, 1.56] | 0.361 | 1.81 [0.88, 3.71] | 0.104 | 0.82 [0.28, 2.34] | 0.705 |
| Tertiary | 0.32 [0.12, 0.82] | 0.018 | 0.58 [0.24, 1.40] | 0.226 | 0.11 [0.03, 0.44] | 0.002 |
| Employment status | | | | | | |
| Unemployed/retired | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Employed | 5.47 [2.30, 12.97] | <0.001 | 1.88 [0.79, 4.50] | 0.156 | 10.12 [3.28, 31.25] | <0.001 |
| Monthly income | | | | | | |
| 0–500 cedis (0–80 USD) | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| 501–1000 cedis (81.0–160 USD) | 0.64 [0.32, 1.28] | 0.208 | 0.83 [0.43, 1.59] | 0.573 | 0.25 [0.10, 0.60] | 0.002 |
| >1000 (>160 USD) | 0.75 [0.28, 2.01] | 0.564 | 0.34 [0.14, 0.78] | 0.012 | 0.07 [0.02, 0.25] | <0.001 |
| CLINICAL AND MEDICATION RELATED | | | | | | |
| Source of expenditure on drugs | | | | | | |
| Health insurance only | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Out-of-pocket | 0.68 [0.33, 1.41] | 0.302 | 0.56 [0.28, 1.11] | 0.095 | 0.44 [0.20, 0.98] | 0.046 |
| Number of hypertensive medications | | | | | | |
| <2 medicine | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| 2+ medicines | 0.16 [0.07, 0.37] | <0.001 | 2.52 [1.34, 4.74] | 0.004 | 0.40 [0.17, 0.94] | 0.036 |
| Number of diabetes medications | | | | | | |
| <2 medicine | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| 2+ medicines | 0.30 [0.15, 0.59] | 0.001 | 0.79 [0.45, 1.39] | 0.419 | 0.27 [0.13, 0.56] | <0.001 |
| Frequency of daily dose of medication | | | | | | |
| Once | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Twice/thrice | 7.00 [2.88, 17.01] | <0.001 | 0.52 [0.29, 0.93] | 0.027 | 2.40 [0.92, 6.25] | 0.073 |
| Have co-morbidities | | | | | | |
| | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Hyperlipidaemia /Stroke | 1.06 [0.38, 2.92] | 0.911 | 2.40 [0.96, 5.99] | 0.061 | 3.35 [1.50, 7.48] | 0.003 |
| Duration since diagnosis of hypertension | | | | | | |
| <5 years | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| 5+ years | 0.46 [0.23, 0.91] | 0.027 | 1.70 [0.92, 3.12] | 0.089 | 1.29 [0.64, 2.58] | 0.474 |
| Duration since diagnosis of Diabetes Mellitus | | | | | | |
| <5 years | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| 5+ years | 0.42 [0.16, 1.07] | 0.069 | 0.75 [0.37, 1.52] | 0.424 | 1.75 [0.78, 3.95] | 0.177 |
| Frequency of follow-up | | | | | | |
| 2 weeks/monthly | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Every 2 month | 1.94 [0.99, 3.79] | 0.054 | 0.47 [0.26, 0.85] | 0.012 | 1.86 [0.83, 4.19] | 0.132 |
| Medication related burden | | | | | | |
| Minimum burden | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Moderate/high burden | 1.37 [0.73, 2.57] | 0.328 | 2.77 [1.59, 4.85] | <0.001 | 2.42 [1.17, 5.05] | 0.018 |
| Adherence to medications | | | | | | |
| Non-adherence | 1.00 [reference] | | 1.00 [reference] | | 1.00 [reference] | |
| Adherence | 0.99 [0.56, 1.75] | 0.983 | 1.65 [0.96, 2.87] | 0.071 | 2.12 [1.03, 4.34] | 0.040 |
The odds of controlled BP was $84\%$ less among those on 2 or more anti-hypertensive medications (AOR: 0.16, $95\%$ CI: 0.07–0.37, $p \leq 0.001$) and $70\%$ less among those on 2 or more anti-diabetic medications (AOR: 0.30, $95\%$ CI: 0.15–0.59, $$p \leq 0.001$$). Participants who take two or three daily doses of medications had increased odds of having controlled BP compared to those taking a single dose (AOR: 7.00, $95\%$ CI:2.88–17.01, $p \leq 0.001$). The odds of controlled BP were $54\%$ less among those who had been diagnosed with hypertension for at least 5 years (AOR: 0.46, $95\%$ CI: 0.23–0.91, $$p \leq 0.027$$) (Table 4).
## Blood Glucose control levels
In the multivariable binary logistic regression model, the odds of controlled blood glucose levels were 2 times higher among females compared to males (AOR: 2.04, $95\%$ CI: 1.04–4.00, $$p \leq 0.038$$). The odds of controlled blood glucose levels were $50\%$ less among the single compared to the married (AOR: 0.50, $95\%$ CI: 0.27–0.96, $$p \leq 0.037$$). Also, participants with the highest monthly income levels (>1000 cedis) had $66\%$ reduced odds of controlled blood glucose level (AOR: 0.34, $95\%$ CI: 0.14–0.78, $$p \leq 0.012$$) (Table 4).
The odd of controlled blood glucose level was 2 times among those on 2 or more anti-hypertensive medications (AOR: 2.52, $95\%$ CI: 1.34–4.74, $$p \leq 0.004$$). Participants who took two or three daily doses of medications had reduced odds of having controlled blood glucose levels compared to those taking a single dose (AOR: 0.52, $95\%$ CI: 0.29–0.93, $p \leq 0.027$). The odds of controlled blood glucose levels were $53\%$ less among those with follow up visits every 2 months (AOR: 0.47, $95\%$ CI: 0.26–0.85, $$p \leq 0.012$$) (Table 4).
## Both controlled blood pressure and blood glucose levels
In the multivariable binary logistic regression model, the adjusted odds of controlled BP and blood glucose levels compared to those with no formal education was $84\%$ less among primary school holders (AOR: 0.16, $95\%$ CI: 0.05–0.49, $$p \leq 0.001$$) and $89\%$ less among the tertiary school educated (AOR: 0.11, $95\%$ CI: 0.03–0.44, $$p \leq 0.002$$). The adjusted odds of controlled BP and blood glucose were 10 times high among the employed (AOR: 10.12, $95\%$ CI: 3.28–31.25, $p \leq 0.001$). Increased monthly income level was associated with decreased odds of controlled BP and blood glucose level, thus compared to those with 0–500 cedis of monthly income, the odds of controlled BP and blood glucose were $75\%$ less for those with 501–1000 cedis monthly income (AOR: 0.25, $95\%$ CI: 0.10–0.60, $$p \leq 0.002$$) and $93\%$ less among those earning more than 1000.00 cedis (AOR: 0.07, $95\%$ CI: 0.02–0.25, $p \leq 0.001$) (Table 4).
The odds of controlled BP and blood glucose level was $56\%$ less among those whose expenditure on medications were out-of-pocket (AOR: 0.44, $95\%$ CI: 0.20–0.98, $$p \leq 0.046$$). Compared to those on single medications, the odds of controlled BP and blood glucose was $60\%$ less among those taking 2 or more antihypertensive medications (AOR: 0.40, $95\%$ CI: 0.17–0.94, $$p \leq 0.036$$) and $73\%$ less among those taking 2 or more anti-diabetic medications (AOR: 0.27, $95\%$ CI: 0.13–0.56, $p \leq 0.001$). The odds of controlled BP and blood glucose were over 3 times high among those having co-morbid hyperlipidaemia/ stroke (AOR: 3.35, $95\%$ CI: 1.50–7.48, $$p \leq 0.003$$). Controlled BP and blood glucose levels were significantly high among participants with moderate or high medication related burden (AOR: 2.42, $95\%$ CI: 1.17–5.05, $$p \leq 0.018$$). Adherence to medication was associated with over 2 times higher odds of having both controlled BP and blood glucose level (AOR: 2.12, $95\%$ CI:1.03, 4.34, $$p \leq 0.040$$) (Table 4).
## Discussion
The current study sought to assess the BP and blood glucose levels among persons with co-morbid hypertension and T2DM, and the factors associated with these outcomes. The research observed that less than half ($41.3\%$) of patients with hypertension and diabetes comorbidity had their BP under control. These findings are similar to findings a from South African study ($42\%$) [26] and the Jimma University Medical Center ($43.51\%$) [4]. Though BP control levels from this research is better compared to other studies conducted in Addis Ababa ($19.4\%$) [27], and Malaysia ($23.5\%$) [28], this was relatively lower than a study conducted in the Ho municipality in Ghana which reported BP control of $58.7\%$ among people living with diabetes. Furthermore, this study showed the median SBP and DBP were 140 mmHg (IQR: 128–157) and 85 mmHg (IQR: 77–92) respectively. The SBP and DBP estimates from this study were higher than the 135.4 mmHg and 83.3 mmHg estimates from a study among out-patients in two diabetic clinics in Ghana [29]. However, recommendations for managing BP, such as the JNC 8, typically call for lower systolic and diastolic levels in diabetics [12]. Maintaining sufficient BP control is the primary therapeutic goal; hence these findings point to the need for greater attention to the optimum care of patients with co-morbid hypertension and T2DM. Also, in line with the objectives of the Ghana NCD policy, which aims to strengthen early detection and management of NCDs including co-morbid hypertension and T2DM, maintaining BP and blood glucose control will lead to a reduction in morbidity and mortality from NCDs [19,24].
Compared with some patients with diabetes in Ghana, the blood glucose levels in this study was higher [30–32]. Generally, in low-resourced settings like Ghana, and especially in public health facilities where this study was conducted, glycaemic control was assessed using the fasting blood sugar. This is due to the high costs associated with more robust measures like the glycated haemoglobin (HbA1c) which measures average glycaemia over three months.
Also, a fifth of the participants had uncontrolled BP and blood glucose levels whilst another one-fifth had controlled levels for both BP and blood glucose. Without effort from patients, achieving target BP and fasting blood glucose (FBG) will be a difficult challenge. However, when patients ensure optimal adherence to recommended treatment, which can also be based on patients’ adherence to dietary and lifestyle changes. In cases when the initial aims are not met, it is advised to alter the medicines and to regularly evaluate patients [33]. Additionally, it has been demonstrated that following clinical guidelines when prescribing drugs for T2DM and hypertension improves clinical outcomes [34].
The findings from this study also showed that controlled BP levels were high among the older age groups which is inconsistent with the findings from a similar study conducted in Ethiopia where patients who fell among the older age groups had two times uncontrolled BP compared to patients among the younger age groups [4] This observed result may be due to the fear in the prevalence of worse clinical outcomes in the elderly who present with comorbid hypertension and diabetes and hence a better rate of adherence to treatment. Although other studies did not find any significant association between marital status and control of BP levels [4,35], this study showed that those who were not currently married had controlled BP levels. Patients taking two or three doses of medications daily, had controlled BP levels which is inconsistent with other studies which have showed that multiple daily dosing of medications leads to a higher rate of non-adherence and consequently unfavourable clinical outcomes [36–38]. The differences in results may be due to a better understanding of the co-morbid nature of their disease condition and therefore a better rate of adherence to their treatment. This study further revealed that those who were employed had controlled BP levels which disagrees with a similar study in Addis Ababa where patients who were employed had uncontrolled BP levels [35]. This may be due to the strains of their jobs and therefore a recognition of the need to adhere to treatment to prevent a worsened clinical outcome. On the other hand, the study findings showed that controlled BP levels were low among those with some form of formal education. This disagrees with studies that have shown formal education to have a positive association with BP levels [4,35]. The differences in observed results may be due to the differences in health seeking behaviours. Participants taking 2 or more different anti-hypertensive and anti-diabetic medications had controlled BP levels which is inconsistent with other studies which have shown that patients who take two or more different medications for their chronic diseases feel burdened and therefore do not adhere to their treatment [36–38]. The observed results may be due to an understanding of the need to adhere to treatment. This study also revealed that those who have been diagnosed with hypertension for at least 5 years had controlled BP levels which is inconsistent with studies which have shown that patients who have been diagnosed with hypertension for over 5 years have uncontrolled BP levels due to a reduction in their health-seeking behaviour [35].
In terms of blood glucose control, the study findings agree with other studies which have shown that two or more medications, and moderate to high medication related burden were significantly associated poor blood glucose control levels [39,40]. Patients taking two or more medications may feel burdened by the number of pills they have to take which consequently leads to non-adherence to treatment and poor blood glucose control outcome [36]. On the other hand, although other studies disagree with the study findings which showed that two or three daily doses of medications were associated better blood glucose control levels, they agree that bi-monthly clinic follow-up periods were implicated blood glucose control levels [36,39,40]. The observed results may be due to an understanding of the disease nature and therefore an increase in health seeking behaviour. Though this study showed that being married and being female has a significant association with increased odds of blood glucose control levels and increased monthly income has significant association with decreased odds of blood glucose control levels, other studies found no association with blood control levels [39,40].
In terms of combined BP and blood glucose control, this study further found that being employed, having hyperlipidaemia or stroke co-morbidity, and moderate to high medication related burden were associated with high controlled BP and blood glucose levels. This is inconsistent with other studies which have shown that being employed, having other co-morbidities and moderate to high medication related burden are associated with uncontrolled BP and blood glucose levels [35,39,40]. In order to prevent unfavourable outcomes due to the presence of other co-morbidities and the strains of their jobs, these patients may have adhered to their treatment regimen which is probably the reason for the observed results. On the other hand, this study showed that having some form of formal education is associated with decreased chances of having both BP and blood glucose levels controlled, which is inconsistent with studies which have shown significant association between formal education and controlled levels of BP and blood glucose [4,35]. This may be due to a decrease in health seeking behaviour. Consistent with other studies [36–38], this study has also shown two or more anti-hypertensive and anti-diabetic medications to be associated with decreased chances of having both BP and blood glucose levels controlled. This is due to the pill burden these patients feel which may lead to non-adherence to treatment and a deterioration of their condition. Socio-behavioural interventions may be needed for encouraging such patients to adhere to their medications. Although other studies found no association between income and BP and blood glucose, this study showed that increased monthly income level and out-of-pocket payment for medications were associated with decreased chances of having both BP and blood glucose control levels and these have implications for medicines availability, affordability and adherence behaviour [4,35].
## Study limitations
This study is a cross-sectional study hence cannot establish causal inferences but rather associations. Interpretation of findings should therefore be done cautiously. Since this was a hospital-based study, information on the estimates of BP levels and blood glucose were facility specific and may not reflect what pertains within the general community exhibit. Also, because this study measured the BP and blood glucose level of participants at one time point during the interview, it is unable to determine the fluctuations at different time points to establish consistencies in these parameters. Nonetheless, this study provides some estimates on the extent of BP and blood glucose controls among patients with hypertension and T2DM to inform the development of interventions. Again, the rigorous nature of the analytical procedure also reduces the potential biases that are likely to occur. The sample size of 339 is also large enough to draw valid conclusions from a single centre study.
## Conclusion
In this study, two in five patients with hypertension and T2DM had controlled BP levels, three in five had controlled blood glucose levels and one in five had both BP and blood glucose levels under control with factors such as employment status, hyperlipidaemia or stroke co-morbidity, adherence to medication, and moderate to high medication related burden being associated with the control. Identifying patients at risk of poor BP and blood glucose control can lead to targeted interventions in line with the Ghana NCD policy on strengthening management especially among high-risk groups to reduce morbidity and mortality from NCDs. In addition, policy makers should institute measures to ensure that health promotion and education on long-term diseases such as hypertension and T2DM form part of the routine management and care practices. Potential drawbacks to poor control of BP and blood glucose parameters should also be included in these policy documents to guide clinical practice and ensure better health outcomes for patients with co-morbid hypertension and T2DM.
## References
1. Alsaadon H, Afroz A, Karim A, Habib SH, Alramadan MJ, Billah B. **Hypertension and its related factors among patients with type 2 diabetes mellitus–a multi-hospital study in Bangladesh**. *BMC Public Health* (2022) **22** 1-10. DOI: 10.1186/S12889-022-12509-1/TABLES/2
2. Correia JC, Lachat S, Lagger G, Chappuis F, Golay A, Beran D. **Interventions targeting hypertension and diabetes mellitus at community and primary healthcare level in low- and middle-income countries:a scoping review**. *BMC Public Health 2019 19:1* (2019) **19** 1-20. DOI: 10.1186/s12889-019-7842-6
3. Salameh AB, Hyassat D, Suhail A, Makahleh Z, Khader Y, EL-Khateeb M. **The prevalence of hypertension and its progression among patients with type 2 diabetes in Jordan**. *Annals of Medicine and Surgery* (2022) 73. DOI: 10.1016/j.amsu.2021.103162
4. Muleta S, Melaku T, Chelkeba L, Assefa D. **Blood pressure control and its determinants among diabetes mellitus co-morbid hypertensive patients at Jimma University medical center, South West Ethiopia**. *Clin Hypertens* (2017) **23**. DOI: 10.1186/s40885-017-0085-x
5. Petrie JR, Guzik TJ, Touyz RM. **Diabetes, Hypertension, and Cardiovascular Disease: Clinical Insights and Vascular Mechanisms**. *Canadian Journal of Cardiology* (2018) **34** 575-584. DOI: 10.1016/j.cjca.2017.12.005
6. Kim MJ, Lim NK, Choi SJ, Park HY. **Hypertension is an independent risk factor for type 2 diabetes: the Korean genome and epidemiology study**. *Hypertension Research* (2015) **38** 783. DOI: 10.1038/hr.2015.72
7. Yamazaki D, Hitomi H, Nishiyama A. **Hypertension with diabetes mellitus complications**. *Hypertens Res* (2018) **41** 147-156. DOI: 10.1038/s41440-017-0008-y
8. Perreault L, Pan Q, Aroda VR, Barrett-Connor E, Dabelea D, Dagogo-Jack S. **Exploring residual risk for diabetes and microvascular disease in the Diabetes Prevention Program Outcomes Study (DPPOS).**. *Diabet Med* (2017) **34** 1747-1755. DOI: 10.1111/dme.13453
9. Rodríguez-Gutiérrez R, Montori VM. **Glycemic Control for Patients With Type 2 Diabetes Mellitus: Our Evolving Faith in the Face of Evidence**. *Circ Cardiovasc Qual Outcomes* (2016) **9** 504-512. DOI: 10.1161/CIRCOUTCOMES.116.002901
10. Banerji MA, Dunn JD. **Impact of Glycemic Control on Healthcare Resource Utilization and Costs of Type 2 Diabetes: Current and Future Pharmacologic Approaches to Improving Outcomes.**. *Am Health Drug Benefits* (2013) **6**. PMID: 24991370
11. **Introduction: Standards of Medical Care in Diabetes—2021**. *Diabetes Care* (2021) **44** S1-S2. DOI: 10.2337/dc21-Sint
12. Armstrong C.. **JNC8 guidelines for the management of hypertension in adults**. *Am Fam Physician* (2014) **90** 503-504. PMID: 25369633
13. Bakris GL, Williams M, Dworkin L, Elliott WJ, Epstein M, Toto R. **Preserving renal function in adults with hypertension and diabetes: A consensus approach**. *American Journal of Kidney Diseases* (2000) **36** 646-661. DOI: 10.1053/ajkd.2000.16225
14. Kwakye AO, Buabeng KO, Opare-Addo NAM, Owusu-Dabo E. **Clinical pharmacists education and counselling in patients with co-morbid hypertension and diabetes in a Municipal hospital in Ghana**. *Afr J Pharm Pharmacol* (2021) **15** 183-190. DOI: 10.5897/ajpp2021.5267
15. Carey RM, Whelton PK, Aronow WS, Casey DE, Collins KJ, Himmelfarb CD. **Prevention, detection, evaluation, and management of high blood pressure in adults: Synopsis of the 2017 American College of Cardiology/American Heart Association Hypertension Guideline**. *Ann Intern Med* (2018) **168** 351-358. DOI: 10.7326/M17-3203
16. Dedefo MG, Gemechu DB, Fekadu G, Dibessa TT. **Blood Pressure Control among Hypertensive Diabetic Patients on Follow-Up at Chronic Clinic of Nekemte Referral Hospital in West Ethiopia**. *Int J Hypertens* (2020). DOI: 10.1155/2020/7526257
17. Baynouna LM, Nagelkerke NJD, Al Ameri TA, Zein Al Deen SM, Ali HI. **Determinants of Diabetes and Hypertension Control in Ambulatory Healthcare in Al Ain, United Arab Emirates**. *Oman Med J* (2014) **29** 234. DOI: 10.5001/omj.2014.58
18. 18Ghana Health Service. Ghana Health Service 2016 Annual Report. 2017.. *Ghana Health Service 2016 Annual Report* (2017)
19. 19Ministry of Health (MOH). Republic of Ghana National Policy for the Prevention and Control of Chronic Non-Communicable Diseases in Ghana. 2012
Aug.. *Republic of Ghana National Policy for the Prevention and Control of Chronic Non-Communicable Diseases in Ghana* (2012)
20. Serra N, Oliveras A, Bergoñon S, Sans L, Cobos A, Martínez P. **Factors associated with blood pressure control in diabetic patients treated in nephrology units. PRESDIAB Study**. *Nefrologia* (2011) **31** 313-321. DOI: 10.3265/Nefrologia.pre2011.Apr.10556
21. Abougalambou SSI, Abougalambou AS, Sulaiman SAS, Hassali MA. **Prevalence of hypertension, control of blood pressure and treatment in hypertensive with type 2 diabetes in Hospital University Sains Malaysia**. *Diabetes and Metabolic Syndrome: Clinical Research and Reviews* (2011) **5** 115-119. DOI: 10.1016/J.DSX.2012.03.001
22. 22National Health Insurance Scheme (NHIS)—Ghana. Medicines List.
2022 [cited 14 Oct 2022]. Available: https://www.nhis.gov.gh/medlist.. *Medicines List.* (2022)
23. Awad A, Alhadab A, Albassam A. **Medication-Related Burden and Medication Adherence Among Geriatric Patients in Kuwait: A Cross-Sectional Study**. *Front Pharmacol* (2020) **11**. DOI: 10.3389/fphar.2020.01296
24. Zidan A, Awisu A, El-Hajj MS, Al-Abdulla SA, Figueroa DCR, Kheir N. **Medication-Related Burden among Patients with Chronic Disease Conditions: Perspectives of Patients Attending Non-Communicable Disease Clinics in a Primary Healthcare Setting in Qatar**. *Pharmacy* (2018) **6** 85. DOI: 10.3390/pharmacy6030085
25. Horne R, Weinman J. **Patients’ belief about prescribed medicines and their role in adherence to treatment in chronic physical illness**. *J Psychosom Res* (1999)
26. Adebolu FA, Naidoo M, Adebolu F. **Blood pressure control amongst patients living with hypertension presenting to an urban district hospital outpatient clinic in Kwazulu-Natal**. (2014). DOI: 10.4102/phcfm.v6i1.572
27. Shimels T, Abebaw M, Bilal AI, Tesfaye T. **Treatment Pattern and Factors Associated with Blood Pressure and Fasting Plasma Glucose Control among Patients with Type 2 Diabetes Mellitus in Police Referral Hospital in Ethiopia**. (2018). DOI: 10.4314/ejhs.v28i4.12
28. Chew BH, Mastura I, Shariff-Ghazali S, Lee PY, Cheong AT, Ahmad Z. **Determinants of uncontrolled hypertension in adult type 2 diabetes mellitus: an analysis of the Malaysian diabetes registry 2009**. *Cardiovasc Diabetol* (2012) **11** 1-8. DOI: 10.1186/1475-2840-11-54/TABLES/2
29. Apini R, Annan R, Apprey C, Asamoah-Boakye O. **Predictors of Glycaemic Control among Ghanaian Type 2 Diabetic Patients Using Diabetes Self-management Approach**. *Int J Biochem Res Rev* (2018) **23** 1-18. DOI: 10.9734/IJBCRR/2018/42976
30. Afaya RA, Bam V, Azongo TB, Afaya A, Kusi-Amponsah A, Ajusiyine JM. **Medication adherence and self-care behaviours among patients with type 2 diabetes mellitus in Ghana**. *PLoS One* (2020) **15** e0237710. DOI: 10.1371/journal.pone.0237710
31. Djonor SK, Ako-Nnubeng IT, Owusu EA, Akuffo KO, Nortey P, Agyei-Manu E. **Determinants of blood glucose control among people with Type 2 diabetes in a regional hospital in Ghana**. *PLoS One* (2021) **16** e0261455. DOI: 10.1371/journal.pone.0261455
32. Osei-Yeboah J, Lokpo SY, Owiredu WKBA, Johnson BB, Orish VN, Botchway F. **Medication Adherence and its Association with Glycaemic Control, Blood Pressure Control, Glycosuria and Proteinuria Among People Living With Diabetes (PLWD) in the Ho Municipality, Ghana**. *Open Public Health J* (2019) **11** 552-561. DOI: 10.2174/1874944501811010552
33. Davies MJ D, ’Alessio DA, Fradkin J, Kernan WN, Mathieu C, Mingrone G. **Management of Hyperglycemia in Type 2 Diabetes, 2018. A Consensus Report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD)**. *Diabetes Care* (2018) **41** 2669-2701. DOI: 10.2337/dci18-0033
34. De Boer IH, Bangalore S, Benetos A, Davis AM, Michos ED, Muntner P. **Diabetes and Hypertension: A Position Statement by the American Diabetes Association**. *Diabetes Care* (2017) **40** 1273-1284. DOI: 10.2337/dci17-0026
35. Amare F, Nedi T, Berhe DF. **Blood pressure control practice and determinants among ambulatory hypertensive patients attending primary health care facilities in Addis Ababa**. *SAGE Open Med* (2020) **8** 2050312120946521. DOI: 10.1177/2050312120946521
36. Bernhard J, Maibach R, Thürlimann B, Sessa C, Aapro MS. **Patients’ estimation of overall treatment burden: why not ask the obvious?**. *J Clin Oncol* (2002) **20** 65-72. DOI: 10.1200/JCO.2002.20.1.65
37. Gallacher K, May CR, Montori VM, Mair FS. **Understanding patients’ experiences of treatment burden in chronic heart failure using normalization process theory**. *Ann Fam Med* (2011) **9** 235-243. DOI: 10.1370/afm.1249
38. Krska J, Katusiime B, Corlett SA. **Patient experiences of the burden of using medicines for long-term conditions and factors affecting burden: A cross-sectional survey**. *Health Soc Care Community* (2018) **26** 946-959. DOI: 10.1111/hsc.12624
39. Mariye T, Girmay A, Tasew H, Teklay G, Ayele E, Gerensea H. **Determinants of hypertension among diabetic patients in public hospitals of the central zone, Tigray, Ethiopia 2018: Unmatched case-control study**. *Pan African Medical Journal* (2019) **33**. DOI: 10.11604/pamj.2019.33.100.17094
40. Yimam M, Desse TA, Hebo HJ. **Glycemic control among ambulatory type 2 diabetes patients with hypertension Co-morbidity in a developing country: A cross sectional study**. *Heliyon* (2020) **6**. DOI: 10.1016/J.HELIYON.2020.E05671
|
---
title: Effect of somatic antigens of Dirofilaria repens adult worms on angiogenesis,
cell proliferation and migration and pseudo-capillary formation in human endothelial
cells
authors:
- María del Pilar Pérez Rodríguez
- Claudia Alarcón-Torrecillas
- Miguel Pericacho
- Iván Rodríguez-Escolar
- Elena Carretón
- Rodrigo Morchón
journal: Parasites & Vectors
year: 2023
pmcid: PMC10022164
doi: 10.1186/s13071-023-05726-z
license: CC BY 4.0
---
# Effect of somatic antigens of Dirofilaria repens adult worms on angiogenesis, cell proliferation and migration and pseudo-capillary formation in human endothelial cells
## Abstract
### Background
Angiogenesis is defined as the formation of new vessels by sprouting of endothelial cells from pre-existing vessels in response to stimuli, such as hypoxia or inflammation. Subcutaneous dirofilariasis, caused by Dirofilaria repens, is a zoonotic disease characterized by the formation of subcutaneous nodules with the presence of at least one encapsulated worm, showing perivascular vascularization around it. The aim of this study is to analyze whether the somatic antigen of adult D. repens worms interacts with and modulates the angiogenic mechanism, cell proliferation and migration, and formation of pseudo-capillaries.
### Methods
The expression of VEGF-A, VEGFR-1/sFlt, VEGFR-2, mEnd and sEnd in cultures of human vascular endothelial cells stimulated with somatic antigen of adult worms of D. repens (DrSA), vascular endothelial growth factor (VEGF) and DrSA + VEGF were evaluated by using ELISA commercial kits. Cellular viability was analyzed by live cell count, cytotoxicity assays by using a commercial kit, cell proliferation by MTT-based assay, cell migration by wound-healing assay carried out by scratching wounds and capacity of formation of pseudo-capillaries analyzing cell connections and cell groups in Matrigel cell cultures. In all cases unstimulated cultures were used as controls.
### Results
DrSA + VEGF significantly increased the expression of VEGF-A, VEGFR-2 and mEndoglin compared to other groups and unstimulated cultures. Moreover, DrSA + VEGF produced cell proliferation and migration and increased the formation of pseudo-capillaries.
### Conclusions
Somatic antigen of adult D. repens worms activated the proangiogenic mechanism, cell proliferation and cell migration as well as formation of pseudo-capillaries in this in vitro human endothelial cell model. These processes could be related to the survival of adult D. repens in subcutaneous nodules in infected hosts.
## Introduction
Angiogenesis is defined as the formation of new vessels by sprouting of endothelial cells from pre-existing vessels in response to stimuli such as hypoxia or inflammation [1–3]. A series of morphogenetic changes occur, consisting of endothelial cell activation, extracellular matrix degradation, endothelial cell proliferation and migration, vascular lumen formation, and vessel stabilization and maturation [4]. Endothelial cells produce a series of factors in response to these processes, including vascular endothelial growth factor (VEGF), which stimulates endothelial cells in adjacent vessels to grow and form new vessels [2, 5]. Increased levels of VEGF-A are detected by endothelial cells through binding to its tyrosine kinase-like receptor VEGFR-2, at which point a conformational change occurs that results in receptor dimerization and, via endothelial cells, triggers the release of nitric oxide and increased vascular permeability [5, 6]. However, both VEGFR-1 and its soluble form (sFlt1) exert negative regulation of signaling through VEGFR-2, as they act by sequestering the ligand and preventing it from binding to the receptor [7]. Endoglin is a vascular protein that plays a fundamental role in endothelial and vascular physiology, highlighting processes such as angiogenesis and vascular remodeling [8–10]. Endoglin expression increases in areas where vascular injury and active angiogenesis are taking place, in both tumor and non-tumor cells [11–15]. High concentrations of soluble endoglin (sEndoglin) have been described in patients with cancer, pre-eclampsia or cardiac conditions; in addition, antiangiogenic properties have been attributed to it, as it can prevent the correct development of angiogenesis in vivo and in vitro [16].
Subcutaneous dirofilariasis is a zoonotic disease, caused by Dirofilaria repens, which mainly affects canine reservoirs, both domestic and wild, and humans. In addition, it is a vector-borne disease that mainly affects Old World countries [17]. Like other filarial species, D. repens harbors intracellular symbiont bacteria of the genus Wolbachia whose contribution to inflammatory processes is key [18, 19]. Human subcutaneous dirofilariasis usually presents as a local inflammation at the subcutaneous level, which causes a nodule to form where the worm is encapsulated and destroyed [20]. In patients with subcutaneous nodules, ultrasound and Doppler techniques have shown that a clear peripheral vascularization develops around these nodules [21].
There are no studies analyzing the angiogenic character of D. repens but there are studies on other nematodes such as *Trichinella spiralis* and Dirofilaria immitis. In the first case, it has been shown that encapsulated larvae initiate angiogenesis and attract a set of highly permeable blood vessels to the surface of their collagenous capsule present in the musculature for nutrient acquisition and waste elimination, thus maintaining a long-term host-parasite relationship [22, 23]. Regarding D. immitis, Zueva et al. [ 24, 25] observed a proangiogenic effect of somatic antigens of D. immitis adults and an antiangiogenic effect of Wolbachia spp. In addition, in other diseases caused by lymphatic nematodes, it is suggested that microfilariae and adult filariae induce lymphangiogenesis and in vitro remodeling of lymphatic channels [26].
Against this background, the aim of our study was to determine whether D. repens is involved in the stimulation of the angiogenic process and in the cell proliferation and migration and the formation of pseudo-capillaries from adult worms located within subcutaneous nodules using an in vitro model of human endothelial cells.
## Cell culture
Human umbilical vein endothelial cells (HUVECs) were grown in Endothelial Basal Medium 2 (Lonza, Walkersville, MD, USA) supplemented with SingleQuots® (Lonza): $20\%$ fetal bovine serum, heparin (22.5 µg/ml), VEGF (0.5 ng/ml), ascorbic acid (1 µg /ml), hFGF-B (10 ng/ml), hydrocortisone (0.2 µg /ml), hEGF (5 ng/ml), gentamicin (30 mg/ml), amphotericin B (15 µg/ml) and R3-IGF-3 (20 ng/ml). Plates were pre-coated with $0.1\%$ pig gelatin (Sigma-Aldrich, Saint Louis, MO, USA), $0.01\%$ fibronectin (Sigma-Aldrich) and $0.01\%$ collagen (Corning). Cells were cultured at 37 °C in a humidified atmosphere in the presence of $5\%$ CO2/$95\%$ air. The medium was changed every 3 days. Expansion was carried out by trypsinizing the cells (Trypsin/EDTA, Lonza) and replating them when the proliferating cells had reached a sufficient density. Passaging was performed at the ratio of 1:3. Cell counts were performed using a Countess® Automated Cell Counter (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions.
## Reagents and stimulation of endothelial cells, cytotoxity and cellular viability
Adult D. repens somatic antigens (DrSA) were prepared as previously described [27] and stored at −80 °C until use. In brief, D. repens adult worms [5] from a human skin nodule [28] was washed, macerated and sonicated in PBS, pH 7.2. The homogenate was centrifuged at 10,000g/30 min and the sediment discarded. The supernatant was the somatic antigenic extract employed for stimulations. Protein concentration was measured by DC protein assay commercial kit (Bio-Rad).
HUVECs were treated as previously described by Morchón et al. [ 27]. In brief, endothelial cells (106 cells/plate) were plated on 60-mm culture plates and were grown for 4 days to obtain confluent cultures and treated with three different stimuli: 1 μg/ml of DrSA or Vascular Endothelial Growth Factor (VEGF) (R&D SYSTEMS) and 1 μg/ml of DrSA plus 1 μg/ml of VEGF (DrSA + VEGF). Unstimulated cells were used as controls in the same conditions. Stimulated and unstimulated cell cultures were carried out in triplicate. Finally, the supernatant of the cell cultures was collected, and HUVECs were lysed in ice-cold lysis buffer [20 mM Tris–HCl (pH 7.5); 140 mM NaCl; 10 mM ethylendiaminetetraacetic acid; $10\%$ glycerol; $1\%$ Igepal CA-630; aprotinin, pepstatin and leupeptin at 1 μg/ml each; 1 mM phenylmethylsulfonyl fluoride and 1 mM sodium orthovanadate].
Cytotoxicity was assessed in the supernatant of the stimulated and control cell cultures using the Toxilight BioAssay Kit (Cambrex, Verviers, Belgium) following the commercial instructions. This commercial kit quantitatively measures the release of adenylate kinase from damaged cells. Cellular viability was analyzed by cell counts using the Countess® Automated Cell Counter (Invitrogen) following the manufacturer’s instructions. The results are presented as the mean ± SEM of three experiments performed in duplicate.
## Angiogenic factors assays
VEGF-A, VEGFR-1/sFlt, VEGFR-2 and sEndoglin concentrations in the endothelial cells culture medium were measured by ELISA using a Human VEGF-A Quantikine ELISA kit (R&D Systems, Minneapolis, MN, USA), Human VEGFR-1/sFlt Quantikine ELISA kit (R&D Systems), Human VEGFR-2 Quantikine ELISA kit (R&D Systems) and Human Endoglin Quantikine ELISA kit (R&D Systems), respectively, and membrane Endoglin (mEndoglin) concentration in the lysed endothelial cells was measured by Human Endoglin Quantikine ELISA kit (R&D Systems) following the manufacturers’ instructions. The results are presented as the mean ± SEM of three experiments performed in duplicate.
## Proliferation assays
Proliferation assays were assessed as previously described [29], with some modifications. In brief, 1000 cells were seeded on a 96-well plate and stimulated in complete HUVEC medium with 1 μg/ml DrSA, Vascular Endothelial Grown Factor (VEGF) (RRD SYSTEMS), 1 μg/ml DrSA plus 1 μg/ml VEGF and 1 μg/ml *Cut plus* 1 μg/ml of VEGF for 10 days. Unstimulated cells were used as controls in the same conditions. Proliferation at different days (every 2 days) of culture was determined by incubating cell cultures with 0.5 mg/ml 3-[4.5-dimethylthiazol-2-yl]-2.5-diphenyl tetrazolium bromide (MTT) (Sigma-Aldrich, St. Louis, MO, USA) for 4 h. Then, $10\%$ SDS in 0.01 M HCl was added at a 1:1 (v/v) ratio and left overnight at 37 °C. Finally, absorbance was measured at 570 nm. The results are presented as the mean ± SEM of three experiments performed in triplicate.
## Migration assays
Wound-healing assays were assessed as previously described by González-Miguel et al. [ 30] with some modifications. In brief, in vitro scratched wounds were created by scraping confluent cell monolayers in 60-mm sterile plates with a sterile disposable pipette tip. The remaining cells were washed with sterile PBS buffer, incubated with the endothelial supplemented medium and stimulated with five different stimuli up to 6 h. Unstimulated cells were used as controls in the same conditions. Endothelial cell migration into the denuded area was monitored by photographing the plates every 30 min. The results are presented as the mean ± SEM of three experiments performed in duplicate.
## Endothelial cell tube formation assay
Endothelial cell tube formation was assessed as previously described by Jerkic et al. [ 31] with some modifications. In brief, a total of 8000 HUVECs per well were plated on Matrigel® precoated µ-Slide Angiogenesis® plates (Ibidi, Gräfelfing, Germany) in complete HUVEC medium with DrSA, VEGF and DrSA + VEGF (1:10 dilution). After seeding on Matrigel®, cells spread and aligned with each other to develop hollow tube-like structures. The cells and intercellular junctions were observed every 30 min for 5 h of incubation, and the morphological changes were photographed at 3 h using a phase contrast inverted Leica microscope (Leica, Wetzlar, Germany). Subsequently, the intercellular junctions were divided between the cell bodies to calculate the relationship between them (endothelial cell tube formation = cellular connections/cellular bodies). Unstimulated cells were used as controls in the same conditions. Each experiment was performed in triplicate.
## Statistical analysis
GraphPad Prism v.7 was used for all data analyses. Analyses were performed by ANOVA and corrected for repeated measurements when appropriate. If ANOVA revealed overall significant differences, individual means were evaluated post hoc using Tukey’s test. All results were expressed as the mean ± SEM. In all experiments, a significant difference was defined as a p value < 0.05.
## Effect of DrSA on cell viability, cytotoxicity and angiogenic factors
To determine whether D. repens is able to modify the production of some angiogenic factors, we analyzed the production of VEFG-A, VEGFR-1/sFLt, VEGFR-2, mEndoglin and sEndoglin in in vitro cultures of endothelial cells stimulated with DrSA, VEGF and DrSA + VEGF, where unstimulated cultures were used as controls.
No differences were found in cell viability and cytotoxicity of stimulated cultures with DrSA, VEGF and DrSA + VEGF compared to unstimulated cell cultures (data not shown).
The stimulation of cell cultures with DrSA + VEGF significantly increased the expression of VEGF-A compared to cell cultures stimulated with DrSA ($t = 63.70$, df = 4, $P \leq 0.0001$), VEGF ($t = 40.28$, df = 4, $P \leq 0.0001$) and unstimulated cultures ($t = 63.82$, df = 4, $P \leq 0.0001$). In addition, VEGF-stimulated cell cultures showed a significant increase in VEGF-A production compared to DrSA ($t = 21.76$, df = 4, $P \leq 0.0001$) and unstimulated cultures ($t = 18.97$, df = 4, $P \leq 0.0001$) (Fig. 1A). In addition, there were no significant differences for VEGFR-1/sFlt between stimulated and unstimulated cell cultures (Fig. 1B), and only VEGFR-2 was detected in DrSA + VEGF stimulated cell cultures. In brief, DrSA + VEGF stimulated cell cultures showed a significant increase compared with DrSA ($t = 8.802$, df = 2, $$P \leq 0.0127$$), VEGF stimulated cultures ($t = 5.364$, df = 2, $$P \leq 0.033$$) and unstimulated cultures ($t = 6.484$, df = 2, $$P \leq 0.023$$) (Fig. 1C).Fig. 1Effects of DrSA and Cut antigens on VEGF (A), VEGFR-1/sFlt1 (B) and VEGFR-2 (C) in unstimulated cultures () and cultures stimulated with VEGF (), DrSA () and DrSA + VEGF (). Results are expressed as the mean ± SEM of three independent experiments. The asterisk or plus sign (*/+) indicates significant differences ($p \leq 0.05$): DrSA + VEGF vs. control, VEGF and DrSA (*) and VEGF vs. control and DrSA (+) The stimulation of cell cultures with DrSA + VEGF only significantly increased the expression of mEndoglin when compared to cell cultures stimulated with DrSA ($t = 6.46$, df = 2, $$P \leq 0.0231$$), VEGF ($t = 4.559$, df = 2, $$P \leq 0.0449$$) and unstimulated cultures ($t = 5.112$, df = 2, $$P \leq 0.0362$$). However, when we analyzed the expression of sEndoglin, no significant differences were observed between stimulated and unstimulated cultures (Fig. 2).Fig. 2Effects of DrSA and Cut antigens on sEndoglin and mEndoglin in unstimulated cultures () and cultures stimulated with VEGF (), DrSA () and DrSA + VEGF (). Results are expressed as the mean ± SEM of three independent experiments. Significant differences (*) in comparisons with the other groups are indicated ($p \leq 0.05$)
## DrSA produces cell proliferation
The effect of DrSA on the proliferation of endothelial cells was quantified using the MTT technique in a 10-day period (Fig. 3). All cultures showed typical cell growth curves in all experimental groups with a progressive growth between days 0 and 6 or 8 post-stimulation, experiencing a decrease of viable cells from there until day 10 post-stimulation. MTT technique showed a significant increase in the number of viable cells on day 6 post-stimulation in cultures stimulated with DrSA + VEGF compared with cultures stimulated with DrSA ($t = 5.346$, df = 4, $$P \leq 0.0059$$), VEGF ($t = 3.139$, df = 4, $$P \leq 0.0349$$) and unstimulated cultures ($t = 3.45$, df = 4, $$P \leq 0.0251$$) on day 8 post-stimulation in cultures stimulated with DrSA + VEGF compared with cultures stimulated with DrSA ($t = 7.051$, df = 4, $$P \leq 0.0021$$), VEGF ($t = 5.68$, df = 4, $$P \leq 0.0047$$) and unstimulated cultures ($t = 4.711$, df = 4, $$P \leq 0.0092$$) and on day 10 post-stimulation in cultures stimulated with DrSA + VEGF compared with cultures stimulated with DrSA ($t = 5.914$, df = 4, $$P \leq 0.0041$$), VEGF ($t = 2.878$, df = 4, $$P \leq 0.0451$$) and unstimulated cultures ($t = 3.424$, df = 4, $$P \leq 0.0267$$).Fig. 3Effects of DrSA and Cut antigens on cell proliferation in unstimulated cultures () and cultures stimulated with VEGF (), DrSA () and DrSA + VEGF (). Results are expressed as the mean ± SEM of three independent experiments. Significant differences (*) in comparisons with the other groups are indicated ($p \leq 0.05$)
## DrSA produces cell migration
A wound-healing assay was performed to assess migration of endothelial cells (Fig. 4). The quantification was carried out by measuring the distance of migration compared with negative control (untreated cells) to 6 h post-stimulation. A significant decrease of distance migration after stimulation with DrSA + VEGF with respect to DrSA ($t = 12.5$, df = 2, $$P \leq 0.002$$) and VEGF ($t = 4.853$, df = 2, $$P \leq 0.0083$$) stimulated and unstimulated cultures ($t = 10.84$, df = 2, $$P \leq 0.0004$$).Fig. 4Effects of DrSA and Cut antigens on cell migration distance in unstimulated cultures () and cultures stimulated with VEGF (), DrSA () and DrSA + VEGF (). Results are expressed as the mean ± SEM of three independent experiments. Significant differences (*) compared with the other groups are indicated ($p \leq 0.05$)
## Effect of DrSA on pseudo-capillary formation
The capacity for pseudo-capillary formation was evaluated by analyzing the cell junctions (connections) and the cellular set that emerged in stimulated and unstimulated cell cultures (Fig. 5). The formation of pseudo-capillaries and the connections/joint relationship in cultures stimulated with DrSA + VEGF showed a significant increase compared to cell cultures stimulated with DrSA ($t = 7.74$, df = 2, $$P \leq 0.0163$$), VEGF ($t = 7.159$, df = 2, $$P \leq 0.019$$) and unstimulated cultures ($t = 6.514$, df = 2, $$P \leq 0.0228$$).Fig. 5Effects of DrSA and Cut antigens on connections and cellular set in unstimulated cultures () and cultures stimulated with VEGF (), DrSA () and DrSA + VEGF (). Results are expressed as the mean ± SEM of three independent experiments. Significant differences (*) compared with the other groups are indicated ($p \leq 0.05$)
## Discussion
Subcutaneous dirofilariasis (D. repens) is a vector-borne zoonotic disease mainly affecting canids and humans, which causes the formation of subcutaneous nodules in most cases [32]. Dirofilaria repens has been shown to be able to develop mechanisms that allow it to lengthen the survival of the parasite in the host, including the formation of subcutaneous nodules and modulation of the immune response, among others [19].
There are no studies analyzing the angiogenic character of D. repens but there are studies on other nematodes such as T. spiralis, in which larvae initiate angiogenesis and attract a set of highly permeable blood vessels to the surface of the collagenous capsule present in the musculature to achieve nutrient acquisition, waste elimination and thus maintain a long-term host-parasite relationship [22, 23]. The role of D. immitis and Wolbachia in the angiogenic process has also been studied. In fact, the somatic antigen of D. immitis promotes the production of angiogenic molecules, while Wolbachia and adult D. immitis worms from dogs treated with doxycycline are able to stimulate anti-angiogenic molecules and decrease pseudo-capillary formation [24, 25]. In other lymphoid nematodes, it is suggested that microfilariae and adult filariae induce lymphangiogenesis and in vitro remodeling of lymphatic channels, which would demonstrate that the parasites stimulate mechanisms to promote vascular supply in damaged tissues [26]. In patients with subcutaneous nodules caused by D. repens, ultrasound and Doppler techniques have shown that peripheral vascularization is evident around these nodules [21].
The aim of this study was to determine whether adult D. repens worms could stimulate the angiogenic process (formation of new blood vessels from pre-existing vessels) at the endothelial level. To recreate the conditions under which the angiogenic process is initiated by endothelial cells after an obstructive or hypoxic process, among others, human endothelial cells were stimulated with VEFG, the first factor that occurs in the angiogenic process [33], and DrSA.
First, neither DrSA nor VEGF produced a cytotoxic effect or altered endothelial cell viability. Second, DrSA + VEGF significantly stimulated VEGF-A and VEGFR-2 production compared to VEGF-produced stimulations and in unstimulated cells. Both molecules are potent proangiogenic mediators that have mitogenic and anti-apoptotic effects on endothelial cells and are able to inhibit the host immune response, among other functions [26, 34]. A similar effect occurred in macrophage and mast cell culture stimulated with antigens of encapsulated larvae of T. spiralis [35, 36] and in endothelial cells stimulated with somatic antigen of adult D. immitis [18, 24, 27]. In addition, some authors suggested that VEGF is a key factor for the formation of new vessels around nurse cells in parasitic nematodes [37]. However, the levels of VEGFR-1/sFlt-1 were not modified, similar to studies carried out by Zueva et al. [ 24, 25], where their production was analyzed in canine endothelial cells stimulated with somatic extracts of D. immitis derived from dogs untreated and treated with doxycycline (with lesser amounts of Wolbachia) and recombinant Wolbachia Surface Protein. These results may indicate that VEGFR-1/sFlt-1 does not participate in the angiogenic process for at least the first 24 h.
Third, only DrSA + VEGF increased mEndoglin expression without altering sEndoglin expression compared the other stimulated and unstimulated cultures. mEndoglin is the cell membrane-bound form of endoglin, which causes a proangiogenic effect, and its expression has been observed to increase under physiological conditions during tissue vascularization as well as in pathological conditions including angiogenesis [24]. In other studies, mEndoglin production decreased when endothelial cell cultures were stimulated by Wolbachia [25], while sEndoglin production (related to anti-angiogenic processes [3]) increased. Although adult D. repens worms contain the endosymbiont *Wolbachia bacteria* [38], Wolbachia does not appear to be a determinant when D. repens proteins are in the majority, as in the case of D. immitis [24], but is a determinant when it is in the majority [25] or when it has previously been eliminated [24] and anti-angiogenic mechanisms are stimulated.
Fourth, the present study analyzed whether cell proliferation and migration processes were affected when stimulation with DrSA and VEGF was performed in our HUVEC model, and the results showed that both processes were affected, with increased cell proliferation and migration observed in endothelial cell cultures stimulated with DrSA + VEGF and unaffected by the other stimuli. VEGF production seemed able to promote cell proliferation and migration and to inhibit the host immune response [3, 26, 31, 39], which are closely related to vasculogenesis and angiogenesis. In studies carried out in other canine endothelial cell models, D. immitis seemed to increase cell proliferation and migration within the fibrinolytic process, which is related to angiogenesis [30]. Therefore, these results confirmed previous findings that the proangiogenic process was favored when the endothelial cell culture was stimulated with DrSA + VEGF.
Finally, the effect of DrSA on the formation of vascular pseudo-capillaries was analyzed. These structures form on a Matrigel matrix [40], which simulates the formation of immature vessels that form during angiogenesis. In the present human endothelial cell model, only DrSA + VEGF produced a significant increase in the formation of pseudo-capillaries, which is similar to previous results. However, other studies have shown that the presence of Wolbachia significantly decreased pseudo-capillary formation in canine endothelial cells [25], which is related to anti-angiogenic processes. In this study, the effect of somatic antigen from adult D. repens worms, which contains Wolbachia [38], has been shown to be contrary to this fact, so that the amount of Wolbachia used alone or as a minority part of the protein load of adult D. repens worms in the host may condition the drift of the angiogenic process.
## Conclusions
The results obtained in the present study provide the first data on the angiogenic effect produced by adult D. repens worms together with VEGF in human endothelial cells. This effect favors the production of proangiogenic molecules, cell proliferation and migration as well as the formation of pseudo-capillaries, which could facilitate parasite survival by favoring the formation of new vessels surrounding subcutaneous nodules. Further studies are needed to investigate the effect of these antigens on the angiogenic process and on other mechanisms in which a direct parasite-host interaction is established, aiming to facilitate the survival of the parasite in the hosts.
## References
1. Geudens I, Gerhardt H. **Coordinating cell behaviour during blood vessel formation**. *Development* (2011) **138** 4569-4583. DOI: 10.1242/dev.062323
2. Núñez-Gómez E, Pericacho M, Ollauri-Ibáñez C, Bernabéu C, López-Novoa JM. **The role of endoglin in post-ischemic revascularization**. *Angiogenesis* (2017) **20** 1-24. DOI: 10.1007/s10456-016-9535-4
3. Ollauri-Ibáñez C, López-Novoa JM, Pericacho M. **Endoglin-based biological therapy in the treatment of angiogenesis-dependent pathologies**. *Expert Opin Biol Ther* (2017) **17** 1053-1063. DOI: 10.1080/14712598.2017.1346607
4. Senger DR, Davis GE. **Angiogenesis**. *Cold Spring Harb Perspect Biol* (2011) **3** a005090. DOI: 10.1101/cshperspect.a005090
5. Koch S, Tugues S, Li X, Gualandi L, Claesson-Welsh L. **Signal transduction by vascular endothelial growth factor receptors**. *Biochem J* (2011) **437** 169-183. DOI: 10.1042/BJ20110301
6. Shoeibi S, Mozdziak P, Mohammadi S. **Important signals regulating coronary artery angiogenesis**. *Microvasc Res* (2018) **117** 1-9. DOI: 10.1016/j.mvr.2017.12.002
7. Hazarika S, Dokun AO, Li Y, Popel AS, Kontos CD, Annex BH. **Impaired angiogenesis after hindlimb ischemia in type 2 diabetes mellitus: differential regulation of vascular endothelial growth factor receptor 1 and soluble vascular endothelial growth factor receptor 1**. *Circ Res* (2007) **101** 948-956. DOI: 10.1161/CIRCRESAHA.107.160630
8. Bernabeu C, Conley BA, Vary CP. **Novel biochemical pathways of endoglin in vascular cell physiology**. *J Cell Biochem* (2007) **102** 1375-1388. DOI: 10.1002/jcb.21594
9. López-Novoa JM, Bernabeu C. **The physiological role of endoglin in the cardiovascular system**. *Am J Physiol Heart Circ Physiol* (2010) **299** H959-H974. DOI: 10.1152/ajpheart.01251.2009
10. Kapur NK, Morine KJ, Letarte M. **Endoglin: a critical mediator of cardiovascular health**. *Vasc Health Risk Manag* (2013) **9** 195-206. DOI: 10.2147/VHRM.S29144
11. Kuiper P, Hawinkels LJ, de Jonge-Muller ES, Biemond I, Lamers CB, Verspaget HW. **Angiogenic markers endoglin and vascular endothelial growth factor in gastroenteropancreatic neuroendocrine tumors**. *World J Gastroenterol* (2011) **17** 219-225. DOI: 10.3748/wjg.v17.i2.219
12. Muñoz M, Martinez-Armesto J, Coveñas R. **NK-1 receptor antagonists as antitumor drugs: a survey of the literature from 2000 to 2011**. *Expert Opin Ther Pat* (2012) **22** 735-746. DOI: 10.1517/13543776.2012.697153
13. Anderberg C, Cunha SI, Zhai Z. **Deficiency for endoglin in tumor vasculature weakens the endothelial barrier to metastatic dissemination**. *J Exp Med* (2013) **210** 563-579. DOI: 10.1084/jem.20120662
14. Park S, Dimaio TA, Liu W, Wang S, Sorenson CM, Sheibani N. **Endoglin regulates the activation and quiescence of endothelium by participating in canonical and non-canonical TGF-β signaling pathways**. *J Cell Sci* (2013) **126** 1392-1405. PMID: 23418351
15. Barnett JM, Suarez S, McCollum GW, Penn JS. **Endoglin promotes angiogenesis in cell- and animal-based models of retinal neovascularization**. *Invest Ophthalmol Vis Sci* (2014) **55** 6490-6498. DOI: 10.1167/iovs.14-14945
16. Ruiz-Remolina L, Ollauri-Ibáñez C, Pérez-Roque L. **Circulating soluble endoglin modifies the inflammatory response in mice**. *PLoS ONE* (2017) **12** e0188204. DOI: 10.1371/journal.pone.0188204
17. Simón F, Diosdado A, Siles-Lucas M, Kartashev V, González-Miguel J. **Human dirofilariosis in the 21st century: a scoping review of clinical cases reported in the literature**. *Transbound Emerg Dis* (2022) **69** 2424-2439. DOI: 10.1111/tbed.14210
18. Simón F, Morchón R, Rodríguez-Barbero A, López-Belmonte J, Grandi G, Genchi C. *Vet Parasitol* (2008) **158** 223-231. DOI: 10.1016/j.vetpar.2008.09.010
19. Simón F, Siles-Lucas M, Morchón R, González-Miguel J, Mellado I, Carretón E. **Human and animal dirofilariasis: the emergence of a zoonotic mosaic**. *Clin Microbiol Rev* (2012) **25** 507-544. DOI: 10.1128/CMR.00012-12
20. Pampiglione S, Rivasi F. **Human dirofilariasis due to**. *Parassitologia* (2000) **42** 231-254. PMID: 11686084
21. Ilyasov B, Kartashev V, Bastrikov N. **Thirty cases of human subcutaneous dirofilariasis reported in Rostov-on-Don (Southwestern Russian Federation)**. *Enferm Infecc Microbiol Clin* (2015) **33** 233-237. DOI: 10.1016/j.eimc.2014.04.002
22. Ock MS, Cha HJ, Choi YH. **Verifiable hypotheses for thymosin β4-dependent and -independent angiogenic induction of**. *Int J Mol Sci* (2013) **14** 23492-23498. DOI: 10.3390/ijms141223492
23. Rayia DA, Othman A, Harras S, Helal D, Dawood L, Soliman S. **Bevacizumab: a new take on therapy of muscle phase of**. *Acta Trop* (2022) **230** 106409. DOI: 10.1016/j.actatropica.2022.106409
24. Zueva T, Morchón R, Carretón E. **Angiogenic response in an in vitro model of dog microvascular endothelial cells stimulated with antigenic extracts from**. *Parasit Vectors* (2019) **12** 315. DOI: 10.1186/s13071-019-3570-0
25. Zueva T, Morchón R, Carretón E. **Angiogenesis in cardiopulmonary dirofilariosis: does the Wolbachia surface protein have a pro- or anti-angiogenic effect?**. *J Helminthol* (2020) **94** e162. DOI: 10.1017/S0022149X20000450
26. Wysmołek ME, Długosz E, Wiśniewski M. **The immunological role of vascular and lymphatic endothelial cells in filarial infections**. *Animals* (2022) **12** 426. DOI: 10.3390/ani12040426
27. Morchón R, Rodríguez-Barbero A, Velasco S, López-Belmonte J, Simón F. **Vascular endothelial cell activation by adult**. *Parasitol Int* (2008) **57** 441-446. DOI: 10.1016/j.parint.2008.05.004
28. Kramer LH, Kartashev VV, Grandi G. **Human subcutaneous dirofilariasis**. *Russia Emerg Infect Dis* (2007) **13** 150-152. DOI: 10.3201/eid1301.060920
29. Pérez-Roque L, Núñez-Gómez E, Rodríguez-Barbero A, Bernabéu C, López-Novoa JM, Pericacho M. **Pregnancy-induced high plasma levels of soluble endoglin in mice lead to preeclampsia symptoms and placental abnormalities**. *Int J Mol Sci* (2020) **22** 165. DOI: 10.3390/ijms22010165
30. González-Miguel J, Morchón R, Carretón E, Montoya-Alonso JA, Simón F. **Can the activation of plasminogen/plasmin system of the host by metabolic products of**. *Parasit Vectors* (2015) **8** 194. DOI: 10.1186/s13071-015-0799-0
31. Jerckic M, Rivas-Elena JV, Santibáñez JF, Prieto M, Rodríguez-Barbero A, Pérez-Barriocanal F. **Endoglin regulates cyclooxygenase-2 expres- sion and activity**. *Circ Res* (2006) **99** 248-256. DOI: 10.1161/01.RES.0000236755.98627.69
32. Sgroi G, Buono F, Iatta R. **Vector-borne pathogens of zoonotic concern in hunting dogs of southern Italy**. *Acta Trop* (2022) **232** 106502. DOI: 10.1016/j.actatropica.2022.106502
33. Peach CJ, Mignone VW, Arruda MA. **Molecular pharmacology of VEGF-A isoforms: binding and signalling at VEGFR2**. *Int J Mol Sci* (2018) **19** 1264. DOI: 10.3390/ijms19041264
34. Melincovici CS, Boşca AB, Şuşman S. **Vascular endothelial growth factor (VEGF)—key factor in normal and pathological angiogenesis**. *Rom J Morphol Embryol* (2018) **59** 455-467. PMID: 30173249
35. Shariati F, Pérez-Arellano JL, López-Abán J, Arefi M, Martínez-Fernández AR, Muro A. *Exp Parasitol* (2009) **123** 347-353. DOI: 10.1016/j.exppara.2009.08.016
36. El-Dardiry MA, Abdel-Aal AA, Abdeltawab MSA. **Effect of mast cell stabilization on angiogenesis in primary and secondary experimental**. *Parasit Vectors* (2021) **14** 567. DOI: 10.1186/s13071-021-05075-9
37. Despommier DD. **How does**. *Parasitol Today* (1998) **14** 318-323. DOI: 10.1016/S0169-4758(98)01287-3
38. Grandi G, Morchón R, Kramer L, Kartashev V, Simón F. **Wolbachia in Dirofilaria repens, an agent causing human subcutaneous dirofilariasis**. *J Parasitol* (2008) **94** 1421-1423. DOI: 10.1645/GE-1575.1
39. Alexandrakis MG, Passam FH, Pappa CA. **Relation between bone marrow angiogenesis and serum levels of angiogenin in patients with myelodysplastic syndromes**. *Leuk Res* (2005) **29** 41-46. DOI: 10.1016/j.leukres.2004.05.002
40. Francescone RA, Faibish M, Shao R. **A Matrigel-based tube formation assay to assess the vasculogenic activity of tumor cells**. *J Vis Exp* (2011) **55** 3040
|
---
title: 'Effects of disorganization of retinal inner layers for Idiopathic epiretinal
membrane surgery: the surgical status and prognosis'
authors:
- Huanhuan Li
- Conghui Zhang
- Hui Li
- Shuai Yang
- Yao Liu
- Fang Wang
journal: BMC Ophthalmology
year: 2023
pmcid: PMC10022165
doi: 10.1186/s12886-023-02856-x
license: CC BY 4.0
---
# Effects of disorganization of retinal inner layers for Idiopathic epiretinal membrane surgery: the surgical status and prognosis
## Abstract
### Background
To compare the surgical status in idiopathic epiretinal membrane (IERM) patients with or without disorganization of retinal inner layers (DRIL) and to correlate with optical coherence tomography angiography (OCTA) and clinical data.
### Methods
In 74 eyes from 74 patients with IERM treated by surgery with 12-month follow-up. According to the superficial hemorrhage, the patients were divided into group A (no macular bleeding), group B (macular parafoveal bleeding) and group C (macular foveal bleeding). Optical coherence tomography (OCT) were evaluated for presence of DRIL,central retina thickness and integrity of the inner/outer segment layer recorded at baseline and at 1, 3, 6, and 12 months postoperatively and best-corrected visual acuity (BCVA) was recorded simultaneously. OCTA was conducted at 12 months postoperatively. Main outcome measures is correlation between DRIL and superficial hemorrhage in membrane peeling,and BCVA and OCTA outcomes postoperatively.
### Results
The rate of DRIL and BCVA had statistically significant differences between the three groups at the time points(baseline and 1, 3, 6, and 12 months after surgery), respectively ($P \leq 0.001$ for all). FD-300 value ($$P \leq 0.001$$)and DCP in all parafoveal regions (superior: $$P \leq 0.001$$; inferior: $$P \leq 0.002$$;Nasal: $$P \leq 0.014$$;Tempo: $$P \leq 0.004$$) in eyes with DRIL were lower than those without DRIL.There was a linear regression relationship between FD-300 and postoperative BCVA ($$P \leq 0.011$$).
### Conclusion
IERM Patients with DRIL have more intraoperative adverse events and limited benefits from surgery which should be considered in the decision whether to perform mebrane peeling. OCT-A provides more detailed vascular information that extends our understanding of persistent DRIL.
## Background
Idiopathic epiretinal membrane (IERM) is a fibrous hyperplastic membrane formed in front of the macula, which can cause severe vision loss and metamorphopsia of patients. The latest researches suggested that the incidence of IERM in the people over 70 years old is as high as $15.1\%$ [1–3].*Pars plana* vitrectomy (PPV) remains the predominant treatment for the IERM patients, and most physicians tend to peeling the internal limiting membrane (ILM) simultaneously [4] to avoid the recurrence of IERM [5]. Although the technique of membrane stripping is more and more stable, the intraoperative situations during membrane peeling and the recovery of postoperative visual function are quite different. A large number of clinical studies have revealed that there are no clearly indicators to accurately indicate the surgical status and prognosis of patients [3].
In the previous study, we have documented the intraoperative situations and explored the relationship between them and postoperative visual recovery. The results showed that superficial hemorrhage was the most important risk factor for postoperative poor visual recovery [6].However, the reason is still not clear yet. In 2014, Sun and associates firstly discovered the disorganization of retinal inner layers (DRIL) in optical coherence tomography (OCT) imaging [7]. DRIL was identified as a new marker for the prognosis of macular edema in various diseases, as diabetic retinopathy (DR) [7–10], and retinal vein occlusion (RVO) [11–13]. Moreover, studies about the correlation between DRIL and the prognosis of IERM surgery have attract wide attention [14].
The present study retrospectively analyzed the OCT of 74 eyes from 74 patients with IERM, which combined with the optical coherence tomography angiography (OCTA), to investigate the correlation between the structural status of the inner retinal layer and the abnormal intraoperative situations during IERM and ILM peeling,with the purpose of clarify the reasons.
## Patient information
A total of 135 patients with epiretinal membrane peeling surgery for IERM in Shanghai 10th People's Hospital Affiliated to Tongji University (Shanghai,China) from January 2017 to January 2020.The inclusion criteria were as follows: Confirmation of ERM using funduscopic examination under slit-lamp microscopy and OCT; visual impairment; metamorphopsia;Amsler Grid Test-positive; intact outer layers of the retina at 12 months postoperative; the thickness of the central macular retina with a diameter of 1 mm (CRT-1 mm) ranged from 219 to 363 μm at 12 months postoperative (The lowest CRT-1 mm was obtained from a large sample study by using 4000 SD-OCT and the highest was the mean CRT-1 mm of 135 patients at 12 months postoperative).The exclusion criteria were as follows: history of other ocular diseases;surgery and trauma;myopia ≤ -6 diopters or hyperopia ≥ + 6 diopters;axial length ≥ 26 mm or ≤ 22 mm; occurrence of diabetes, renal dysfunction and other systemic diseases, which may have interfered with the measurements. Finally, 74 patients (74 eyes) met the criteria for inclusion and were performed a complete 12-months follow-up period. The present study followed the Declaration of Helsinki and was approved by the Institutional Ethical Review Board of Shanghai 10th People's Hospital affiliated to Tongji University (Shanghai, China).Written informed consent was obtained from all patients or their guardians at the beginning of the study.
## Surgical methods
The standard 23G three-port vitrectomy was performed under indirect panretinoscopy. In brief, triamcinolone acetonide (0.05 ml, 10 mg/ml) was injected above the optic disc using an ultra-wide-angle lens in order to identify the posterior vitreous cortex before the disc and macula, prior to performing vitrectomy. The complete detachment of posterior vitreous cortex and the posterior retinal pole was validated. Subsequently, the intraocular perfusion pressure was decreased and staining with $0.025\%$ indocyanine green was performed for 30 s. The ILM and ERM within the macular area were peeled. All affected eyes were received combined cataract surgery and all surgeries were performed by the experienced physician.
The entire surgical process was video-recorded. The patients were classified according to superficial hemorrhage during membrane peeling. If the patients were not bleeding in the surgery, they would be put into group A. If the patients were bleeding in the area that outside the center of the macula with a diameter of 3 mm, they would be classified into group B.And, the patients with bleeding in the area that within the center of the macula with a diameter of 3 mm were classified as group C.
## Preoperative and postoperative examinations
In addition to the patients' medical history, the following information was recorded preoperatively: Sex, age, disease duration (the time of distorted vision complained by patients), best-corrected visual acuity (BCVA), Amsler grid, fundus photography (Model CR-2, Canon, Inc.) and OCT (Zeiss Cirrus HD-OCT 400). All patients were returned for follow-up at 1, 3, 6 and 12 months postoperatively. During each follow-up, the preoperative examinations were repeated. OCTA (Optovue RTVue XR Avanti, Optovue, Inc.) was performed at 12 months postoperatively.
DRIL was defined as disorganizations of the inner retinal layers, more precisely, as the inability to identify that by SD-OCT the well-known-delineated boundaries between the ganglion cell-inner plexiform layer, inner nuclear layer and outer plexiform layer within the central 1500 μm region by SD-OCT [8, 9, 15].All OCT images were interpretated by two experienced technicians. In case of different opinions, the third technician was referred to interpret the results.
OCT-A scanning was conducted under Angio Retina mode (6 × 6 mm). After acquisition of the image, the foveal region was defined as the circular region with a central diameter of 1 mm, and the parafoveal region (upper, lower, temporal, and nasal) was defined as the circular region with an outer diameter of 3 mm and an inner diameter of 1mm [9]. The software automatically segmented the tissue into 4 layers, and two of these layers were used in the following measurements. The superficial retinal layer starts from the inner limiting membrane (with an offset of 0 μm) to the inner plexiform layer with an offset of − 9 μm. The deep retinal layer starts from inner plexiform layer with an offset of − 9 μm to the outer plexiform layer with an offset of 9 um. The vessel density in the superficial capillary plexus (SCP) and the deep capillary plexus (DCP) was quantified automatically by AngioVue Analytics (RTVue-XR version 2017.1.0.155 software) [16]. The area, acircularity index (AI) and perimeter (PERIM) of fovea avascular zoon(FAZ), and the vessel density of the full retina in a width of 300 μm around the FAZ (FD-300) were also obtained by the software in the FAZ mode. The motion arteacts of eyes were decreased by eye tracking mode, and were removed by motion correction technology. The cutoff value of the signal strength index was set at ≥ 40.
## Statistical analysis
Statistical analysis was performed using SPSS software (version 21.0; IBM Corp.). Values were expressed as the mean ± standard deviation. BCVA was converted to Logarithm of the Minimum Angle of Resolution scoring. LSD-t test was used to compare the multiple means of visual acuity among the three groups. Differences in vessel density, FAZ, and AI between eyes with DRIL and without DRIL were tested by two independent-samples t test. Fisher’s exact test was used to test the difference between qualitative variables. The Spearman correlation coefficient and the multiple linear regression analysis were used to assess correlations. A value of $P \leq 0.05$ was accepted as statistically significant.
## Patient characteristics
74 eyes from 74 patients(age: 65.88 ± 5.66, 35 males and 39 females) which diagnosed with IERM were defined as eligible for this study. According to the superficial hemorrhage during the stripping operation, 35 patients in group A (no bleeding), 22 patients in group B (parafoveal bleeding) and 17 patients (macular fovea bleeding) in group C.There were no significant differences in age, course of disease, axial length, follow-up time, and CRT-1 mm at the time points of before and 12 months after surgery among the three groups (Table 1).Table 1Baseline characteristics of 74 eyes from 74 patients with IERMGroup AGroup BGroup CFPPatients352217Age (years)64.80 ± 5.5067.53 ± 5.0265.81 ± 6.272.2580.110Disease duration (months)5.83 ± 2.446.81 ± 3.136.84 ± 2.771.7680.176Axial length of eye (mm)24.14 ± 0.9023.96 ± 0.9723.85 ± 0.950.9280.399Pre-op CRT-1 mm (um)425.60 ± 22.22417.97 ± 18.18421.26 ± 22.641.2640.287Post-op CRT-1 mm (um)345.45 ± 11.51347.53 ± 9.91348.58 ± 11.250.8250.441IERM idiopathic epiretinal membrane, BCVA best-corrected visual acuity, Pre-op pre-operative, Post-op post-operative, CRT central macular thickness
## Preoperative and postoperative BCVA
There were significant differences in the mean postoperative BCVA (logMAR) at 1, 3, 6 and 12 months among the three groups ($P \leq 0.001$). Specifically, in group C, the postoperative BCVA (logMAR) at 1, 3, 6 and 12 months was significantly higher than both Group A and Group B ($P \leq 0.001$). Moreover, the difference of post-operative BCVA (logMAR) between group A and group B was only appeared at 3 months after surgery ($$p \leq 0.004$$). The preoperative BCVA (logMAR) showed no difference among the three groups ($F = 1.339$, $$P \leq 0.267$$) (Table 2). Spearman correlation indicated that superficial hemorrhage was significantly correlated with poor postoperative BCVA ($r = 0.698$, $$P \leq 0.000$$).Table 2Comparison of pre- and post operative BCVA in the three groups of 74 patients with IERMPre-op1 m Post-op3 m Post-op6 m Post-op12 m Post-opGroup A0.60 ± 0.160.40 ± 0.090.24 ± 0.080.16 ± 0.080.11 ± 0.09Group B0.59 ± 0.170.41 ± 0.100.31 ± 0.070.20 ± 0.090.14 ± 0.12Group C0.65 ± 0.160.79 ± 0.150.72 ± 0.130.66 ± 0.140.58 ± 0.17F Value1.339131.838262.623258.310147.316PP = 0.267P < 0.001P < 0.001P < 0.001P < 0.001Group A&BP > 0.05P > 0.05P = 0.004P > 0.05P > 0.05Group B&CP > 0.05P < 0.001P < 0.001P < 0.001P < 0.001Group A&CP > 0.05P < 0.001P < 0.001P < 0.001P < 0.001
## DRIL measurements
Among the 36 patients with DRIL by preoperative OCT, 24 patients had intraoperative superficial retinal hemorrhage (parafoveal: 10 cases, macular foveal: 14 cases) and that occurred in 15 of 38 patients (parafoveal: 12cases, macular foveal: 3 cases) without DRIL ($r = 0.328$, $$P \leq 0.000$$).
4 patients exhibited improved DRIL 1 month after operation. Furthermore, 13 additional patients demonstrated improved DRIL at 6 months post-operatively and 5 additional patients exhibited improvement at 12 months post-operatively. The mean pre-operative BCVA (logMAR) of patients without DRIL and with DRIL were (0.57 ± 0.21) and (0.14 ± 0.14), respectively ($r = 0.692$, $$P \leq 0.000$$) (Table 3).Table 3Comparison of intraoperative status and postoperative BCVA in 74 IERM patients with or without DRILPre-op12 m Post-opWith DRILWithout DRILWith DRILWithout DRILGroup A1223134Group B1012220Group C143116r0.3280.547P0.0000.000BCVA0.62 ± 0.160.58 ± 0.160.57 ± 0.210.14 ± 0.14r0.1390.692P0.1490.000 In group A, there were 12 cases with DRIL before operation, among which 9 cases had clear and identifiable inner retinal layer structure at 6 months after operation, and only 1 case had DRIL left at 12 months after operation. In group B, there were 10 cases with DRIL before operation, only 3 cases of inner retinal structure failed to recover at 6 months after operation, and only 2 cases with DRIL at 12 months after operation. In group C, 14 cases with DRIL before operation, 13 cases at 6 months after operation, and 11 cases left at 12 months after operation (Fig. 1, Fig. 2). The DRIL rates before surgery, and 1, 3, 6, and 12 months after surgery of the three groups were statistically significan ($r = 0.328$, $$P \leq 0.000$$) ($r = 0.375$, $$P \leq 0.000$$) ($r = 0.457$, $$P \leq 0.000$$) ($r = 0.570$, $$P \leq 0.000$$) ($r = 0.589$, $$P \leq 0.000$$).Fig. 1Pre- and post-operative changes in DRIL rate of IERM patients. Group A (35 cases). There were 12 cases with DRIL before operation, 10 cases had DRIL at post-op 1 m,7 cases at post-op 3 m, 3cases at post-op 6 m and only 1 case had DRIL left at post-op 12 m. Group B (22 cases). 10 cases with DRIL at pre-op, 8 cases at post-op 1 m, 5 cases at post-op 3 m,3 cases at post-op 6 m and only 2 cases at post-op 12 m. Group C (17 cases). 14 cases with DRIL at pre-op, 1 case showed improvement at post-op 6 m and 11 cases had DRIL left at post-op 12 mFig. 2Intra-operative photos and OCT at pre-operative stage and 12 months post-operatively of patients with IERM. A. The patient with no hemorrhage during membrane peeling from Group A (male, 66 years old), which showed clear and identifiable inner retinal layer structure at pre-op and 12 m post-op. B. The patient with parafoveal bleeding during membrane peeling from Group B (female, 72 years old), which displayed DRIL at pre-op and improved at 12 m post-op. C. The patient with macular fovea bleeding during membrane peeling from Group C (female, 65 years old), which exhibited DRIL at pre-op and failed to recover at 12 m post-op
## OCTA findings
OCTA was evaluated in 36 patients at 12 months post-operatively, including 14 eyes with DRIL and 22 eyes without DRIL. In all cases, OCT-A images showed a typical ischemic appearance in both SCP and DCP, deficient capillary network (Fig. 3). Quantitatively, OCT-A imaging revealed a significant decrease in vessel density in DCP in all parafoveal regions (superior: $t = 3.795$, $$P \leq 0.001$$; inferior: $t = 3.390$, $$P \leq 0.002$$;Nasal: $t = 2.601$, $$P \leq 0.014$$;Tempo: $t = 3.106$, $$P \leq 0.004$$) and FD-300 areas ($t = 3.581$, $$P \leq 0.001$$) in DRIL (+) eyes compared with those in DRIL [-] eyes.(Table 4) However, there were no difference of FAZ,VD of SCP and DCP in macular fovea and SCP in parafovea in DRIL (+) eyes compared with those in DRIL [-] eyes (Tables 5 and 6). The results of multiple linear regression indicated that only FD-300 was significantly correlated with postoperative BCVA (t = -2.807, $$P \leq 0.011$$).Fig. 3Post-operative OCT and OCTA images of control and IERM patients. A.OCT and OCTA images of a control patient (female, 53 years old), demonstrated normal reference structure of retinal, FAZ, SCP and DCP. B. OCT and OCTA images of an IERM patient (female, 65 years old) without DRIL at post-op12 months, had clear and identifiable inner retinal layer structure and a reduced area of FAZ with irregular morphology and tortuous vessel. C.OCT and OCTA images of an IERM patient (male, 61 years old) with DRIL at post-op12 months, showed inability to identify boundaries between the ganglion cell-inner plexiform layer, inner nuclear layer and outer plexiform layer and a typical ischemic appearance in both FD-300 (the area around the actual FAZ within 300 μm width marked yellow) and DCP, capillary networkTable 4Comparison of FAZ measurements in 36 patients with or without DRILFAZ area (mm2)PERIM (mm)AIFD (%)Without DRIL0.11 ± 0.041.42 ± 0.281.21 ± 0.0545.08 ± 3.71With DRIL0.10 ± 0.031.32 ± 0.101.18 ± 0.0640.27 ± 4.26t0.0711.4721.2723.581P0.9440.1520.2120.001FAZ fovea avascular zone, PERIM FAZ perimeter, AI A-circularity index, FD foveal vessel density in a 300 μm wide region around FAZ, DRIL disorganization of retinal inner layersTable 5Comparison of superficial capillary plexus (SCP) measurements in 36 patients with or without DRILFovea VD (%)Tempo VD (%)Superior VD (%)Nasal VD (%)Inferior VD (%)Without DRIL28.63 ± 5.7239.01 ± 3.8739.50 ± 4.0940.25 ± 3.8140.13 ± 4.84WithDRIL25.97 ± 4.3739.10 ± 2.3139.13 ± 2.7338.70 ± 2.3239.21 ± 2.83t1.484-0.0750.2961.3640.637P0.1470.9410.7690.1820.528Table 6Comparison of deep capillary plexus (DCP) measurements in 36 patients without or with DRILFovea VD (%)Tempo VD (%)Superior VD (%)Nasal VD (%)Inferior VD (%)Without DRIL43.15 ± 5.8949.40 ± 3.4650.10 ± 3.0749.53 ± 3.4949.84 ± 2.42With DRIL41.43 ± 3.6444.99 ± 5.0845.01 ± 5.0046.36 ± 3.6646.76 ± 2.99t0.9783.1063.7952.6013.390P0.3350.0040.0010.0140.002
## Discussion
Pars plana vitrectomy (PPV) with membrane peeling is the standard treatment to release the forces and restore the normal structure of the macula of IERM eyes. Some surgeons consider ILM peeling as an important aid in the removal of residual ERM [4], because of a lower recurrence rate [5]. On the other hand, ILM peeling has been shown to cause structural damages of retina [17]. The effects of IERM operations on postoperative visual acuity were investigated, including the difficulty of stripping, indocyanine green staining and intraoperative bleeding. Multivariate Logistic regression analysis showed that foveal retinal surface bleeding during stripping was an important factor leading to postoperative poor visual acuity [6]. However, the reason is still unclear.
To our best knowledge, this is the first study to investigate that superficial hemorrhage after membrane peeling is associated with DRIL based on OCT data of 74 IERM patients, and the corresponding changes of macular morphology and blood flow density in these patients were observed by OCTA to quantitatively analyze the retinal blood vessels in patients with persistent DRIL after IERM, so as to clarify the reason why the foveal hemorrhage during IERM surgery seriously affects the postoperative visual recovery of patients.
Although a large number of studies have focused on changes in microstructure of fovea due to IERM to identify the anatomic changes associated with the vision recovery after surgery, the predictors of postoperative visual outcome are still under investigated [18].DRIL was first detected in OCT imaging in 2014 [7]. It is believed that the DRIL indicates retinal vascular dysfunction and the damage of retinal microcirculation that will lead to destruction of neural structure [15]. Therefore, DRIL can predict the collapse of anatomical structures in the visual transmission pathway [19]. Nicholson suggested that DRIL could be used as an OCT biomarker to predict BCVA in DME patients [15]. Radwan and associates studied the correlation between the length of DRIL and visual changes [9]. Similarly, several studies have shown that the inner retinal layer is important for assessing visual prognosis in IERM patients [14, 20, 21]. Accordingly, DRIL has been gradually recognized as a "marker" of visual prognosis in patients with macular disease [2–9].
As the first study to correlate retinal nerve fiber layer bleeding during ILM peeling with DRIL,it was found that the probability of intraoperative bleeding in patients with preoperative DRIL was significantly higher than that in patients without preoperative DRIL especially in macular foveal. It was further found that the recovery rate of DRIL in patients with macular foveal bleeding was only $21.43\%$ ($\frac{3}{14}$), which was significantly lower than that in patients without bleeding ($91.67\%$, $\frac{11}{12}$). Therefore, we concluded that DRIL is one of the causes of retinal bleeding in IERM surgeries. The possible pathological mechanism is that the mechanical traction force of IERM not only destroys the normal macular capillary distribution and also is accompanied by various nerve cell damage, macular capillary ischemia and microcirculation destruction [22], resulting in abnormal distribution and exudation of blood vessels, extracellular fluid accumulation in the inner retinal space, and finally the inner retinal layer structure disorder. Macular vascular abnormalities cause bleeding during IERM sugery, and hemorrhagic injury further aggravate the destruction of macular microcirculation, which is responsible for the persistence of DRIL and poor postoperative visual recovery. Therefore, preoperative evaluation of DRIL in IERM patients will be helpful to predict their postoperative visual acuity.
At present, there are limited studies on DRIL and impaired retinal microcirculation [23]. Some studies have found that the vascular density in the foveal region of DRIL patients is significantly decreased in diabetic retinopathy, suggesting that DRIL and the abnormal blood perfusion in the macular foveal region interact with each other [24]. OCTA were frequently used to study the changes in macular microvascular characteristics, and FD-300 is a foveal vessel density automatically identified by OCTA around the actual FAZ area within 300 μm width. Recent studies have reported that FD-300 index can accurately monitor early diabetic retinopathy [25], and can determine the prognosis and treatment response of DME [25, 26]. In this study, retinal microvascular quantitative analysis was performed in 36 IERM patients with (14 eyes) or without (22 eyes) DRIL. It was found that the patients with DRIL had no significant difference in the VD of DCP and SCP in macular fovea, but FD-300 was significantly decreased. Compared with foveal DCP and SCP, FD-300 can minimize the bias caused by FAZ morphology and retinal stratification in the evaluation of foveal microvessel density [25]. Previous studies suggested that the integrity of microvasculature around the FAZ may play a critical role in the homeostasis of fluid [27]. In this study, FD-300 was used to evaluate the changes of macular foveal vessel density in IERM patients for the first time. Lower FD-300 showed more severe damage in juxtafoveal capillaries and worse microangiopathy. Multiple linear regression analysis showed that FD-300 can predict the post-operative visual acuity of IERM patients. Therefore, FD-300 deserves further attention as a reference index for postoperative visual acuity in patients with IERM.
In the analysis of parafoveal VD, the difference between the patients with and without DRIL after IERM surgery was mainly found in VD of DCP. This result is consistent with the previous study, which found that focal perfusion defects identified by FFA corresponded to the locations of absent or low flow signals in DCP, however, no sign of non-perfusion was observed in these locations in distorted SCP vessels [28]. Lin and associates suggested that mechanical stress caused by ERM more profoundly affects the DCP than the SCP in regard to decrease in blood flow [29]. In addition, multistage capillary non-perfusion may be one of the important pathogenesis of DRIL [30]. The reason why DCP is more prone to damage may be related to its anatomical tissue characteristics. In addition, while the recovery of defects were observed in some eyes, the damage of some others could not be reversible due to focal non-perfused areas affecting the function of the inner retinal layer. Therefore, we speculate that the abnormal deep capillaries in IERM patients hinder the repair of postoperative DRIL, and the persistent DRIL is an important indicator for postsurgical visual prognosis.
In order to eliminate the interference, patient selection was as strictly as possible in this study. Patients with diabetes mellitus, high myopia, uveitis, systemic diseases, incomplete outer retinal structure and severe macular edema (CFT-1 mm > 363um) on OCT were excluded, to focus on the inner retina solely. In addition, considering the average age of the 74 patients (65.55 years old) and the impact of the operation on the lens, all patients were perfomed cataract extraction combined with intraocular lens implantation. The baseline visual acuity of 74 patients with IERM was not correlated with preoperative DRIL, and preoperative lens status may be one reason.
There are limitations to our study. First, since this study was retrospective, some of the results should be treated with caution, and future prospective studies are needed. Second, not all patients conducted OCT-A because of some patients with good visual recovery lost to follow-up. In addition, because of the complexity of its pathogenesis and the diversity of its pathogenic course, it is difficult to accurately predict the prognosis of IERM.
## Conclusions
Our study preliminary established that the DRIL predicted superficial hemorrhage during membrane peeling, and the bleeding was a reason for the problematic recovery of the DRIL in turn. The persistence of DRIL directly affects the postoperative vision of patients with IERM. Moreover, our results indicated that OCT-A could precisely quantify the reduction in capillary plexus perfusion in IERM patients with DRIL. Therefore, DRIL is a novel valuable index for evaluating intraoperative situations and postoperative vision in IERM patients. This should be considered in the decision process whether to perform ERM peeling in this patient group.
## References
1. Lombardo M, Scarinci F, Giannini D, Pileri M, Ripandelli G, Stirpe M, Lombardo G, Serrao S. **High-resolution multimodal imaging after idiopathic epiretinal membrane surgery**. *Retina* (2016) **36** 171-180. DOI: 10.1097/IAE.0000000000000679
2. Lai CC, Wang NK, Wu WC, Yeung L, Hwang YS, Chen KJ, Chen TL, Chuang LH. **The long-term anatomical and visual effect of intravitreal triamcinolone injection during vitrectomy for the treatment of idiopathic macular epiretinal membrane**. *Cutan Ocul Toxicol* (2011) **30** 292-297. DOI: 10.3109/15569527.2011.568031
3. Machado LM, Furlani BA, Navarro RM, Farah ME, Maia A, Magalhaes O, Rodrigues EB, Moraes N, Maia M. **Preoperative and intraoperative prognostic factors of epiretinal membranes using chromovitrectomy and internal limiting membrane peeling**. *Ophthalmic Surg Lasers Imaging Retina* (2015) **46** 457-462. DOI: 10.3928/23258160-20150422-09
4. Sorcinelli R. **Surgical management of epiretinal membrane with indocyanine-green-assisted peeling**. *Ophthalmologica* (2003) **217** 107-110. DOI: 10.1159/000068556
5. Park DW, Dugel PU, Garda J, Sipperley JO, Thach A, Sneed SR, Blaisdell J. **Macular pucker removal with and without internal limiting membrane peeling: pilot study**. *Ophthalmology* (2003) **110** 62-64. DOI: 10.1016/S0161-6420(02)01440-9
6. Li HHLX, Xie CL, Gao XR, Wang H, Wang F. **Intraoperative risk factors associated with visual acuity outcomes of pars plana vitrectomy in idiopathic epiretinal membrane**. *Chin J Ophthalmol* (2017) **53** 344-351
7. Sun JK, Lin MM, Lammer J, Prager S, Sarangi R, Silva PS, Aiello LP. **Disorganization of the retinal inner layers as a predictor of visual acuity in eyes with center-involved diabetic macular edema**. *JAMA Ophthalmol* (2014) **132** 1309-1316. DOI: 10.1001/jamaophthalmol.2014.2350
8. Sun JK, Radwan SH, Soliman AZ, Lammer J, Lin MM, Prager SG, Silva PS, Aiello LB, Aiello LP. **Neural retinal disorganization as a robust marker of visual acuity in current and resolved diabetic macular edema**. *Diabetes* (2015) **64** 2560-2570. DOI: 10.2337/db14-0782
9. Radwan SH, Soliman AZ, Tokarev J, Zhang L, van Kuijk FJ, Koozekanani DD. **Association of disorganization of retinal inner layers with vision after resolution of center-involved diabetic macular edema**. *JAMA Ophthalmol* (2015) **133** 820-825. DOI: 10.1001/jamaophthalmol.2015.0972
10. Xu YFY, Xie C. **Correlation between disorganization of retinal inner layers and visual prognosis after vitrectomy in proliferative diabetic retinopathy patients**. *Chin J Ocular Fundus Dis* (2020) **36** 867-873
11. Mimouni M, Segev O, Dori D, Geffen N, Flores V, Segal O. **Disorganization of the retinal inner layers as a predictor of visual acuity in eyes with macular edema secondary to vein occlusion**. *Am J Ophthalmol* (2017) **182** 160-167. DOI: 10.1016/j.ajo.2017.08.005
12. Babiuch AS, Han M, Conti FF, Wai K, Silva FQ, Singh RP. **Association of disorganization of retinal inner layers with visual acuity response to anti-vascular endothelial growth factor therapy for macular edema secondary to retinal vein occlusion**. *JAMA Ophthalmol* (2019) **137** 38-46. DOI: 10.1001/jamaophthalmol.2018.4484
13. Chan EW, Eldeeb M, Sun V, Thomas D, Omar A, Kapusta MA, Galic IJ, Chen JC. **Disorganization of retinal inner layers and ellipsoid zone disruption predict visual outcomes in central retinal vein occlusion**. *Ophthalmol Retina* (2019) **3** 83-92. DOI: 10.1016/j.oret.2018.07.008
14. Zur D, Iglicki M, Feldinger L, Schwartz S, Goldstein M, Loewenstein A, Barak A. **Disorganization of retinal inner layers as a biomarker for idiopathic epiretinal membrane after macular surgery-The DREAM study**. *Am J Ophthalmol* (2018) **196** 129-135. DOI: 10.1016/j.ajo.2018.08.037
15. Nicholson L, Ramu J, Triantafyllopoulou I, Patrao NV, Comyn O, Hykin P, Sivaprasad S. **Diagnostic accuracy of disorganization of the retinal inner layers in detecting macular capillary non-perfusion in diabetic retinopathy**. *Clin Exp Ophthalmol* (2015) **43** 735-741. DOI: 10.1111/ceo.12557
16. Samara WA, Shahlaee A, Sridhar J, Khan MA, Ho AC, Hsu J. **Quantitative optical coherence tomography angiography features and visual function in eyes with branch retinal vein occlusion**. *Am J Ophthalmol* (2016) **166** 76-83. DOI: 10.1016/j.ajo.2016.03.033
17. Clark A, Balducci N, Pichi F, Veronese C, Morara M, Torrazza C, Ciardella AP. **Swelling of the arcuate nerve fiber layer after internal limiting membrane peeling**. *Retina* (2012) **32** 1608-1613. DOI: 10.1097/IAE.0b013e3182437e86
18. Scheerlinck LM, van der Valk R, van Leeuwen R. **Predictive factors for postoperative visual acuity in idiopathic epiretinal membrane: a systematic review**. *Acta Ophthalmol* (2015) **93** 203-212. DOI: 10.1111/aos.12537
19. Saxena S, Srivastav K, Cheung CM, Ng JY, Lai TY. **Photoreceptor inner segment ellipsoid band integrity on spectral domain optical coherence tomography**. *Clin Ophthalmol* (2014) **8** 2507-2522. PMID: 25525329
20. Cho KH, Park SJ, Cho JH, Woo SJ, Park KH. **Inner-retinal irregularity index predicts postoperative visual prognosis in idiopathic epiretinal membrane**. *Am J Ophthalmol* (2016) **168** 139-149. DOI: 10.1016/j.ajo.2016.05.011
21. Govetto A, Lalane RA, Sarraf D, Figueroa MS, Hubschman JP. **Insights into epiretinal membranes: presence of ectopic inner foveal layers and a new optical coherence tomography staging scheme**. *Am J Ophthalmol* (2017) **175** 99-113. DOI: 10.1016/j.ajo.2016.12.006
22. Moein HR, Novais EA, Rebhun CB, Cole ED, Louzada RN, Witkin AJ, Baumal CR, Duker JS, Waheed NK. **Optical coherence tomography angiography to detect macular capillary ischemia in patients with inner retinal changes after resolved diabetic macular edema**. *Retina* (2018) **38** 2277-2284. DOI: 10.1097/IAE.0000000000001902
23. Dodo Y, Murakami T, Suzuma K, Yoshitake S, Yoshitake T, Ishihara K, Fujimoto M, Miwa Y, Tsujikawa A. **Diabetic neuroglial changes in the superficial and deep nonperfused areas on optical coherence tomography angiography**. *Invest Ophthalmol Vis Sci* (2017) **58** 5870-5879. DOI: 10.1167/iovs.17-22156
24. Cennamo G, Montorio D, Fossataro F, Fossataro C, Tranfa F. **Evaluation of vessel density in disorganization of retinal inner layers after resolved diabetic macular edema using optical coherence tomography angiography**. *PLoS ONE* (2021) **16** e0244789. DOI: 10.1371/journal.pone.0244789
25. Huang WH, Lai CC, Chuang LH, Huang JC, Wu CH, Lin YT, Yeung L. **Foveal microvascular integrity association with anti-VEGF treatment response for diabetic macular edema**. *Invest Ophthalmol Vis Sci* (2021) **62** 41. DOI: 10.1167/iovs.62.9.41
26. Han Y, Wang X, Sun G, Luo J, Cao X, Yin P, Yu R, He S, Yang F, Myers FL. **Quantitative evaluation of retinal microvascular abnormalities in patients with type 2 diabetes mellitus without clinical sign of diabetic retinopathy**. *Transl Vis Sci Technol* (2022) **11** 20. DOI: 10.1167/tvst.11.4.20
27. Daruich A, Matet A, Moulin A, Kowalczuk L, Nicolas M, Sellam A, Rothschild PR, Omri S, Gelize E, Jonet L. **Mechanisms of macular edema: beyond the surface**. *Prog Retin Eye Res* (2018) **63** 20-68. DOI: 10.1016/j.preteyeres.2017.10.006
28. dell'Omo R, Viggiano D, Giorgio D, Filippelli M, Di Iorio R, Calo R, Cardone M, Rinaldi M, dell'Omo E, Costagliola C. **Restoration of foveal thickness and architecture after macula-off retinal detachment repair**. *Invest Ophthalmol Vis Sci* (2015) **56** 1040-1050. DOI: 10.1167/iovs.14-15633
29. Lin TC, Chung YC, Lin CY, Lee FL, Chen SJ. **Focal nonperfusion of deep retinal capillary plexus in eyes with epiretinal membranes revealed by optical coherence tomography angiography**. *Ophthalmic Surg Lasers Imaging Retina* (2016) **47** 404-409. DOI: 10.3928/23258160-20160419-02
30. Onishi AC, Ashraf M, Soetikno BT, Fawzi AA. **Multilevel ischemia in disorganization of the retinal inner layers on projection-resolved optical coherence tomography angiography**. *Retina* (2019) **39** 1588-1594. DOI: 10.1097/IAE.0000000000002179
|
---
title: Health professional’s readiness and factors associated with telemedicine implementation
and use in selected health facilities in Ghana
authors:
- Nathan Kumasenu Mensah
- Godwin Adzakpah
- Jonathan Kissi
- Richard Okyere Boadu
- Obed Uwumbornyi Lasim
- Martha Khainde Oyenike
- Abigail Bart-Plange
- Maxwell Ayindenaba Dalaba
- Felix Sukums
journal: Heliyon
year: 2023
pmcid: PMC10022178
doi: 10.1016/j.heliyon.2023.e14501
license: CC BY 4.0
---
# Health professional’s readiness and factors associated with telemedicine implementation and use in selected health facilities in Ghana
## Abstract
### Background
Telemedicine, which is the practice of medicine using technology to deliver health care remotely, has a low adoption rate in low- and middle-income countries (LMICs). However, the advent of coronavirus disease 2019 (COVID-19) has forced healthcare systems in these settings to begin implementing telemedicine programs. It is unknown how prepared health professionals and the healthcare system are to adopt this technology. Therefore, this study aimed to assess the readiness of health professionals and explore factors associated with telemedicine implementation in Ghana.
### Methods
A cross-sectional study was conducted in six health facilities between March and August 2021. Convenience sampling was used to select the six health facilities, and the participants were selected randomly for the study. Questionnaires were self-completed by participants. Data was exported into STATA 15.0 for analysis, and appropriate statistical methods were employed. All statistical tests were performed at a significance level of $p \leq 0.05.$
### Results
Of the 613 health professionals involved in the study, about 579 ($94.5\%$) were comfortable using computers, and the majority, 503 ($82.1\%$) of them, had access to computers at the workplace. Health professionals agreed that the measures outlined by the health facilities supported their readiness to use telemedicine for healthcare services. Analysis revealed a statistically significant positive relationship between health facilities’ core readiness and health professionals’ readiness, with a correlation coefficient (r) of 0.5484 and a p-value<0.0001. Of the factors associated with health professionals’ readiness towards telemedicine implementation, facility core readiness, engagement readiness, staff knowledge and attitude readiness showed a statistically significant relationship with health professionals’ readiness.
### Conclusion
The study revealed that health professionals are ready to adopt telemedicine. There was a statistically significant relationship between health facilities’ core readiness, engagement readiness, staff knowledge and attitude readiness, and health professionals’ readiness. The study identified factors facilitating telemedicine adoption.
## Introduction
There has been considerable advocacy towards exploiting electronic health (eHealth) services' potential to enhance healthcare quality and safety [1]. With the advent of the Coronavirus disease 2019 (COVID-19), this advocacy has become more relevant, and healthcare systems have begun to use more eHealth services due to the imposition of social isolation restrictions [2,3]. Also, innovations and technological advancements have made internet communication cheaper, providing a unique opportunity for integrating telemedicine into healthcare practices, especially in low- and middle-income countries (LMICs) [[4], [5], [6]]. However, many LMICs face challenges, such as lack of an appropriate policy framework to guide how telemedicine should be successfully implemented [7,8], especially in settings where acceptability and adoption rates of similar technologies have been notoriously low [1,9]. The health needs and challenges in many LMICs are also different [10]. Several persons living in rural areas in most LMICs have challenges accessing high-quality healthcare [11].
In Ghana, the mode of primary health care delivery is rooted in Community-based health planning and services (CHPS), health posts and health centres. This method of conventional primary health services delivery comes with several challenges. As a result of inadequate funding sources, health facilities struggle to employ adequately trained health professionals to man these facilities amid the high volume of patients in need of care. According to the World Bank, about $42\%$ of Ghana’s population in 2021 live in rural areas [12]. Unfortunately, this section of the population is the most disadvantaged by the unequal allocation of healthcare resources [11]. In 2017, Ghana’s doctor-to-population ratio was 1:7374 and nurse to population ratio was 1:505 [13]. These ratios are even more prevalent in rural healthcare delivery.
Also, the distribution of health infrastructures such as pharmacies, clinics, and hospitals across the country is concentrated in urban areas, often with fewer or no such health facilities in the rural areas. Due to these challenges, patients are more likely to travel long distances to access improved healthcare services in urban areas [14]. Additionally, the majority of specialist physicians are stationed in urban health facilities. This means that cases that cannot be handled by doctors in rural areas must be referred to urban health facilities.
There is a need for a system that will reduce costs while ensuring that patients receive satisfactory healthcare irrespective of their geographical locations. Telemedicine offers solution to these issues as a result of rapid developments in Information and Communication Technologies (ICT). Unfortunately, large-scale investments in ICT infrastructure across Africa have been inadequate, in part due to inadequate or lack of funding in some regions. In addition, the likelihood of providing telemedicine is extremely low in rural areas, where it will be most beneficial to the poor because of the enormous infrastructure requirements and connectivity costs [15]. Outside of the urban area, internet connectivity is also poor [16].
Telemedicine has been defined as using computers and related accessories such as cameras, speakers, high-speed internet and other ICT tools to administer and facilitate healthcare delivery over distance as an alternative to face-to-face interactions between clinicians and patients [17,18].
Telemedicine has a remote capability that has been exploited to offer speciality care, for rapid assessment and treatments. In high-income countries, it has been used to address health issues such as the health disparity between health seekers’ needs and health service availability [19], to improve obstetric care outcomes [5,20] and to provide solutions to issues of shortage of health professionals, especially specialist in underserved communities [6]. With the ageing population and age-related chronic diseases increasing worldwide, telemedicine has been an avenue for accessing healthcare among the aged and to reduced frequent visits to health facilities [6]. Evidence shows that the use of telemedicine for home healthcare services has resulted in reduced mortality, better medication compliance and improved safety [21], albeit, most of these studies were conducted in high-income countries. The occurrence of the COVID-19 pandemic has demonstrated that an immense burden can be placed upon both health professionals and health facilities. Hospitals have been forced to make rapid preparation towards transitioning to telemedicine to alleviate this burden [5]. According to Scott and others, telemedicine can improve the capabilities of health professionals while also reducing the likelihood of crowd contact and virus transmission [22]. Kronfield and colleagues stated that telemedicine services played a crucial role in halting the progression of adverse diseases during any pandemic and that health information technology provides us with a more suitable means of monitoring cancers and chronic diseases [23]. The concept of telemedicine enables healthcare professionals to enrol in e-learning programs to build their continuous professional education [24].
The transition towards telemedicine is rapidly growing globally, however, only a few telemedicine projects initiated in LMICs have reached maturity level. The slow uptake of telemedicine in many LMICs relates to factors such as inadequate infrastructure, slow acceptance, inadequate technological equipment, limited financial resources, and inadequate skilled human resources [[25], [26], [27], [28], [29]]. Other reasons accounting for the slow uptake of LMICs are failure to assess health facility readiness before the implementation [30]. Studies have shown that the availability of telemedicine is not sufficient condition for its usage [31] but depends on quality-oriented culture, the self-sufficiency of the hospital, and how flexible the hospital functions [32]. The readiness to implement and use telemedicine may vary with the type of health facility [31,32]. Therefore, health facilities and professionals’ readiness must be assessed before the implementation of telemedicine [7,8,18,[33], [34], [35]]. This study assessed the readiness of health professionals and explored factors associated with the implementation and use of telemedicine in six selected study sites in Ghana.
## Study design
An institutional-based cross-sectional study was carried out in six selected health facilities in Ghana between March and August 2021. All six health facilities were selected on the basis that they used an Electronic Medical Record system (EMR) to provide care, which could serve as a proxy for integrating telemedicine services.
## Study area
The study was conducted in six purposively selected health facilities in Ghana. Two tertiary facilities, namely; the University of Ghana Medical Centre (UGMC) and Cape Coast Teaching Hospital (CCTH), and four other secondary health facilities, namely; Essikado District Hospital, Eastern Regional Hospital, Baptist Medical Centre and Mab International Hospital.
## Study population
The study targeted health professionals employed in the selected health facilities. They comprised of nurses/midwives, physicians, pharmacists, radiologists/sonographers, physiotherapists, and dietetics/nutritionists.
## Sample size, sampling and sampling procedure
The minimum sample size for this study was based on a priori power calculation [36]. Thus, we sampled a minimum of 194 respondents, as this would provide enough statistical power (0.80) to detect small-sized correlation coefficients (0.20) [36]. This made room for a larger sample size, as it would increase the statistical power for detecting smaller effects and strengthen the robustness of the findings. Convenience sampling was used to select respondents who were available and willing to participate in the study. Table 1 below shows the distribution of selected respondents by health facility. Table 1Distribution of sample by health facility. Table 1Name of Health facilitySelected samplePer centBaptist Medical Centre, Nalerugu15024.47Cape Coast Teaching Hospital, Cape Coast9916.15Eastern Regional Hospital, Koforidua11719.09Essikado District Hospital, Essikado8614.03Mab International Hospital, Accra6310.28University Ghana Medical Centre, Accra9815.99Total613100Source: Authors' analysis
## Operational definitions
Core readiness was defined as realising the need for telemedicine and having an appropriate plan for its implementation drawn by policymakers and user groups, which includes budget and identifying resources needed to integrate telemedicine with current services.
Engagement readiness was defined as correctly identifying prioritized needs in the facility that telemedicine will address and express dissatisfaction with the current way of working and awareness of the role of telemedicine among staff.
Knowledge and attitude of health professionals’ readiness were defined as having the proper knowledge and positive attitude towards telemedicine.
Health professionals' readiness was defined as staff's general comfort and willingness to use telemedicine [37].
## Inclusion and exclusion criteria
All 613 health professionals working in the six selected facilities and participating in the telemedicine survey were included in the study for higher precision and accuracy. These health professionals include nurses/midwives, physicians, pharmacists, radiologists/sonographers, physiotherapists, and dietetics/nutritionists. Excluded from the study were health professionals who were less than six months in their current position. Health professionals such as health information officers and laboratory staff were also excluded from the survey.
## Data collection instruments and procedures
Health professionals' responses on their readiness for Telemedicine adoption were gathered using a structured and pretested self-administered questionnaire modified from Ref. [ 38] “readiness assessment instrument”. Other resource-constrained contexts have validated and used this tool [39,40]. In a South African study, the *Cronbach alpha* was 0.864 and 0.910 for core and engagement readiness [41]. This is over the cut-off of 0.7, suggesting good instrument reliability. Since the instrument was created initially to gauge provider and patient enthusiasm for the adoption of eHealth, the original device was modified to fit the purpose of this study Health professionals’ readiness and factors associated with telemedicine implementation and use. The questionnaires were distributed to respondents who were required to self-complete them. The questionnaires were collected later so as not to disrupt the working schedules of the survey participants. Six data collectors and two supervisors participated in the data collection. A data collector was stationed in each facility, and each supervisor was responsible for three facilities. A one-day training was provided to the data collectors and the supervisors. The training focused on the objectives of the study and the data collection process.
The survey questionnaire had a total of 31 questions in 5 thematic sections aiding the objectives of the research. Section A comprised socio-demographic; section B was on the core readiness toward implementation of telemedicine systems; section C comprised engagement readiness toward implementation of telemedicine; section D was on behavioural (Knowledge and Attitude) of health professionals' readiness towards implementation of the telemedicine system, and Section E assessed health professionals’ readiness towards implementation of telemedicine systems at the facility.
## Data processing and management
All the questionnaires were checked for accuracy, completeness and legibility before being entered into an electronic data-capturing tool developed using EpiData 3.1 software. The data screens had in-build checks to minimize data entry errors. Each completed questionnaire was assigned a unique number for quality control and easy recall purposes.
## Data analysis
The data was exported and converted into STATA Version 15 for analysis. Descriptive statistics (such as frequencies, means and standard deviations) were used to describe the data. To achieve the test of normality, Shapiro-Wilk and Bartlett's tests were used to assess the symmetry of all continuous data. The reliability of the variables in the datasets was evaluated by examining Cronbach’s alpha coefficient. The measure of sampling adequacy (MSA) using Kaiser–Meyer–Olkin (KMO) test and Bartlett's test of sphericity were also calculated. The KMO overall measure of sampling adequacy (MSA) was 0.910, which falls within the acceptable level, and was significant at $p \leq 0.0001.$ The Bartlett's test of sphericity (degree of freedom = 465) was 8839.944 and p-value<0.0001, indicating a highly significant correlation among the survey questions. Participants' readiness was assessed by calculating the overall readiness score for each respondent. Spearman’s rank correlation coefficient was performed to assess the relationship between health professionals' readiness and other readiness factors. Unadjusted regression was conducted to explore factors influencing health professionals' readiness for telemedicine implementation. All statistical tests were performed at a significance level of $p \leq 0.05.$
## Ethical approval
The study received ethical approval from the University Cape Coast Institutional Review Board (UCCIRB/CHAS/$\frac{2021}{62}$) and the Cape Coast Teaching Hospital Ethical Review Committee (CCTHERC/EC/$\frac{2021}{062}$). Written permission was obtained from each administrative head of the health facilities where the study was conducted. All participants consented after being informed that participation in the survey was voluntary and all aspects of the study, including the objectives and participants' responsibilities and rights in the study, were explained. Participants’ privacy, confidentiality and anonymity were protected.
## Socio-demographic characteristics
A total of 613 health professionals were involved in the study, of whom 306 ($49.9\%$) were female and 307 ($50.1\%$) were male representing the majority. About forty-three percent, 266 ($43.4\%$) of the respondents were below 30 years old, and 248 ($40.5\%$) were between 30 and 39 years old. Regarding educational level, 243 ($39.6\%$) of the health professionals had completed a bachelor’s degree, and 80 ($13.1\%$) had obtained a master’s degree. The majority of the health professionals 302 ($49.4\%$) were single, while 297 ($48.5\%$) were married. More than three-quarters of the staff 487 ($78.0\%$) worked full-time, while the remaining 42 ($6.9\%$) were part-timers and 93 ($15.2\%$) were temporary or casual workers. The majority of staff 282 ($46.0\%$) had spent between 1 and 5 years working in the current facilities. The majority of the staff 503 ($82.1\%$) had access to computers at their workplace and 359 ($58.6\%$) used the computers several times per day, while 72 ($11.8\%$) used them a few times a week to perform their routine duties. The majority of the health professionals were either very comfortable 320 ($52.2\%$) or comfortable 259 ($42.3\%$) using computers. Only 34 ($5.5\%$) were not comfortable using computers (Table 2).Table 2Demographic characteristics of health professionals from the six selected study sites in Ghana. Table 2CharacteristicsNumber of Respondents [$$n = 613$$]PercentAge group (years) Below 3026643.4 30–3924840.5 40–497712.6 50–59213.4 60 plus10.2Gender Male30750.1 Female30649.9Educational level Masters8013.1 BSc24339.6 Higher National Diploma284.6 Diploma18430.0 Senior Secondary School91.5 Others6911.3Marital status Married29748.5 Single30249.4 Divorced71.1 Widow/Widower61.0Profession Physicians/Doctors12220.0 Nurses/Midwives33655.0 Pharmacy Technicians345.6 Dietetics/Nutritionist101.6 Radiologist132.1 Physiotherapist61.0 Others9014.7Employment status Full-time48778.0 Part-time426.9 Temporal/Casual9315.2Total years of service Less than 116827.4 1–528246.0 6–109715.8 11–15315.1 15 plus355.7Total years of practice Less than 120934.1 1–531050.6 6–106310.3 11–15101.6 15 plus213.4Comfortable using Computers Very comfortable32052.2 Comfortable25942.3 Not comfortable345.5Access to computers at workplace Yes50382.1 No11017.9Access to computers at home Yes47377.2 No14022.8How often do you use computers at workplace? Not at all9916.0 About once each month91.5 A few times a month182.9 About once each week142.3 A few times a week7211.8 5 to 6 times a week203.3 *Once a* day213.4 Several times a day35958.6 Others20.3Do you share computers with colleagues? Yes47978.1 No13421.9Data are presented as frequencies and percentages; N - total number of participants. Source: Authors' analysis
## Descriptive statistics, scale and Item reliability test
The overall mean analysis of health professionals' perceived telemedicine readiness with their subscales is presented in Table 3. The overall mean score for the facility’s core readiness, staff engagement readiness, staff knowledge and attitude readiness, and health professionals' readiness was above 3. This shows that a favourable mean score above 3 (neutral) indicates health professionals agree that the measures outlined support their readiness to use telemedicine to provide services to clients. The Cronbach’s alpha item reliability test showed that the scale is internally reliable [42]. For each study construct, Cronbach’s alpha coefficient ranges from 0.78 to 0.87. The Content Validity Index (CVI) was $87.1\%$, and the Content Validity Ratio (CVR) was $93.5\%$, respectively. Table 3Descriptive statistics, scale and item reliability test of study constructs. Table 3ConstructNo of ItemsMean responseStandard DeviationMean Values under $95\%$ Confidence IntervalCronbach’s AlphaCore Readiness (CR)113.460.50[3.42–3.50]0.83Engagement Readiness (ER)63.400.55[3.35–3.44]0.78Knowledge and Attitude Readiness (KAR)103.690.56[3.64–3.73]0.82Health Professionals Readiness (HPR)43.630.72[3.58–3.69]0.87The Kaiser–Meyer–Olkin measure of sampling adequacy = 0.910.Bartlett's test of sphericity, chi-square = 8839.944, Degree of freedom (DF) = 465, significance <0.0001.Source: Authors' analysis
## Relationship between core, engagement, knowledge and attitude and health professionals’ readiness
Fig. 1, Fig. 2, Fig. 3 show spearman’s rank correlation coefficient (r) between core, engagement, knowledge and attitude, and health professionals' readiness. Analysis revealed a statistically significant positive relationship between health facilities' core readiness and health professionals' readiness with $r = 0.5484$ and p-value<0.0001 (Fig. 1). The study further revealed strong evidence of a statistically significant relationship between staff engagement readiness and health professionals' readiness ($r = 0.4882$; $p \leq 0.0001$) (Fig. 2). Furthermore, staff knowledge and attitude readiness also showed strong evidence of a statistically significant positive relationship with health professionals' readiness ($r = 0.6701$; $p \leq 0.0001$) (Fig. 3).Fig. 1Scatterplot showing the relationship between facility core readiness and health prefessionals’readiness.spearman Rank correlation (r) between facilities core readiness and health professionals readiness [$r = 0.5484$; $p \leq 0.0001$].Source: Authors' analysisFig. 1Fig. 2Scatterplot showing the relationship between staff engagement readiness and health prefessionals’readiness.spearman Rank Correlation (r) between staff engagement readiness and health professionals' readiness [$r = 0.4882$; $p \leq 0.0001$].Source: Authors' analysisFig. 2Fig. 3Scatterplot showing the relationship between staff knowledge and attitude readiness and health prefessionals’readiness.spearman Rank correlation (r) between knowledge and attitude readiness and health professionals' readiness [$r = 0.6701$; $p \leq 0.0001$].Source: Authors' analysisFig. 3
## Exploring factors influencing health professionals’ readiness to implement and use telemedicine
Table 4 shows the unadjusted logistics regression of factors influencing health professionals' readiness towards telemedicine implementation in selected health facilities in Ghana. The outcome from the unadjusted model showed that educational level, total years of service, comfortable using computers, access to computers at the workplace, and frequency of using computers at work had a statistically significant relationship with health professionals' readiness. The analysis further revealed that facility core readiness, staff engagement readiness, and staff knowledge and attitude readiness also had a statistically significant relationship with health professionals' readiness. Unadjusted logistic regression showed that there was a significant association between healthcare professional readiness and core readiness. The evidence revealed that there was a 0.77 ($p \leq 0.001$) mean score increase in health professionals' readiness due to facility core readiness. Further analysis confirmed that there was a significant association between healthcare professional readiness and staff engagement readiness. There was a 0.61 ($p \leq 0.001$) increase in the mean score of health professionals' readiness due to staff engagement readiness. Finally, the analysis showed that there was a significant association between healthcare professional readiness, and staff knowledge and attitude readiness. The unadjusted model showed that there was a 0.83 ($p \leq 0.001$) mean score increase in health professionals’ readiness due to staff knowledge and attitude readiness (Table 4).Table 4Exploring factors affecting health professionals’ readiness towards telemedicine implementation in six selected health facilities in Ghana. Table 4CharacteristicsUnadjustedCoefficient$95\%$ CIp-valueAge group (years) Below 30ref 30–39−0.09[-0.22–0.04]0.130 40–49−0.02[-0.20–0.17]0.525 50–59−0.21[-0.54–0.11]0.031 60 plus0.32[-1.10–1.74]0.669Gender Femaleref Male0.06[-0.05–0.18]0.273Educational level BScref Masters−0.05[-0.23–0.13]0.592 Higher National Diploma−0.50[-0.78 to −0.22]0.001** Diploma−0.19[-0.33 to −0.06]0.006* Senior Secondary School−0.10[-0.58–0.37]0.668 Others0.06[-0.13–0.25]0.560Marital status Singleref Married0.07[-0.05–0.18]0.260 Divorced0.33[-0.21–0.87]0.229 Widow/Widower0.24[-0.35–0.82]0.425Profession Nurses/Midwivesref Physicians/Doctors0.10[-0.05–0.25]0.212 Pharmacy Technicians0.14[-0.11–0.40]0.278 Dietetics/Nutritionist−0.23[-0.68–0.23]0.331 Radiologist0.01[-0.39–0.42]0.945 Physiotherapist0.23[-0.35–0.82]0.437Employment status Full-timeref Part-time−0.06[-0.29–0.16]0.584 Temporal/Casual−0.03[-0.19–0.13]0.730Total years of service 1–5ref Less than 1−0.05[-0.19–0.09]0.482 6–10−0.01[-0.16–0.17]0.929 11–15−0.37[-0.63 to −0.10]0.007* 15 plus−0.17[-0.42–0.09]0.199Total years of practice 1–5ref Less than 10.08[-0.05–0.21]0.218 6–10−0.12[-0.31–0.08]0.248 11–150.13[-0.32–0.59]0.562 15 plus−0.01[-0.33–0.31]0.960Comfortable using Computers Very comfortableref Comfortable−0.10[-0.22–0.01]0.085 Not comfortable−0.44[-0.70 to −0.19]0.001*Access to computers at work place Yesref No−0.17[-0.32 to −0.02]0.025*Access to computers at home Yesref No−0.11[-0.24–0.03]0.124Frequency of use of computers at work place Several times a dayref Not at all−0.31[-0.47 to −0.16]<0.001** About once each month−0.54[-1.00 to −0.08]0.021* A few times a month0.01[-0.33–0.33]0.990 About once each week−0.46[-0.83 to −0.09]0.015* A few times a week−0.56[-0.73 to −0.38]<0.001** 5 to 6 times a week−0.12[-0.43–0.20]0.468 *Once a* day0.13[-0.18–0.43]0.417 Others−0.02[-0.98–0.95]0.974Do you share computers with colleagues Yesref No−0.06[-0.20–0.07]0.361Core Readiness (CR)0.77[0.67–0.87]<0.001**Engagement Readiness (ER)0.61[0.52–0.70]<0.001**Knowledge and Attitude Readiness (KAR)0.83[0.75–0.90]<0.001**Notes: *$p \leq 0.05$, **$p \leq 0.001$ were considered statistically significant. CI – confidence interval; ref-reference group. Source: Authors' analysis
## Discussion
This study assessed the readiness of health professionals in six selected health facilities towards the implementation of telemedicine. Most of the health professionals in this study had access to computers, one of the principal technological tools involved in the practice of telemedicine. This is a positive finding, since lack of access to computers, especially in many LMICs, has been reported as a major barrier to the adoption of similar technologies in workplaces [26,43]. However, we need to be cautious not to overemphasise this point, since other technological gaps may exist despite the positive correlation [30,44].
The current study also revealed that the majority of health professionals were comfortable or very comfortable using computers. This is not surprising since the majority of the respondents in the study were in the younger age groups. Studies [38,43] have shown that computer or technology acceptance is positively correlated with younger age groups. This is because younger people have a natural tendency and interest, to adopt new technology as compared to their older counterparts, and therefore show better readiness for new technologies.
Another important finding is that most health professionals use computers often to perform routine tasks, an indication that they have the skills and positive attitude towards the use of computers. The current study demonstrated that access to computers at the workplace and the frequency of use of these computers at the workplace were statistically significantly associated with health professional readiness. Having prior knowledge of computers breeds a positive attitude towards their use [43,45]. This puts health professionals in a ready position to transition to telemedicine. Studies [31,38] have shown that awareness and use of similar technology systems may be translated into core clinical and learning readiness. It is encouraging that health professionals who were computer literate, use computers and/or had computers at work were likely to be eager to use similar technology. A growing body of literature has shown that poor computer skills or an unwillingness to use computers are related to poor readiness [31,38]. The current study, therefore, showed that health professionals were ready to adopt telemedicine.
In our assessment the overall mean scores for health facility’s core readiness, engagement readiness, knowledge and attitude readiness, and health professionals' readiness towards implementation of telemedicine were above 3.4, an indication that there were favourable responses from health professionals that the selected health facilities had some measures put in place for the use of telemedicine to provide health services to their clients. In this current study, facility core readiness has shown a statistically significant relationship with health professionals' readiness. This shows that once healthcare providers can put in place measures to facilitate telehealth, health professionals are likely to respond positively knowing that these tools would facilitate their ability to provide quality and safe healthcare to their clients remotely at a lower cost. These studies [27,43,[46], [47], [48]] in LMICs have shown that health professionals are ready to embrace new technologies with the support of their employees. The study showed that health professionals believe that there is enough plan in place supported by adequate technical infrastructure to implement telemedicine. This is evidenced by the 0.76 mean score increase in health professionals' readiness due to facility core readiness in the unadjusted model. The availability of the technical infrastructure in the facility can change the poor perception health professionals might have, building their confidence that when they transition to telemedicine it will be successful. This result is comparable to Ref. [ 38], where the presence of good technical equipment, access to computers, and frequent use resulted in an increased readiness of the health professionals.
The study further revealed that facility engagement readiness has a significant relationship with health professionals’ readiness. This shows that healthcare providers can engage and create awareness among staff on the role of telemedicine in providing remote quality health care to their clients. This also echoes the fact that the involvement of stakeholders in planning and decision-making is an important component of technology adoption which can influence their belief in success and must therefore not be ignored.
Again, the study also showed that having the right knowledge and attitude is an important factor that can influence a change of perception. When health professionals have higher educational levels and are comfortable using computers at work, they turn out to be more willing to adopt new technologies. This was confirmed in our study where the level of education of health professionals and health professionals' knowledge and attitude readiness were statistically significantly associated with health professionals’ readiness. On the contrary, poor computer skills of health professionals have been associated with poor readiness of health professionals for technological systems [38,43,45,48]. This result also echoes the need for adequate training so that health professionals gain the right knowledge and attitude. Several studies have shown that many eHealth technologies failed to take up because health professionals were not interested in change and want to keep the status quo [9,10,49,50].
## Limitations
One limitation of the study was the use of the convenience sampling method in the selection of the respondents. This may have introduced a selection bias and limited the generalizability of the findings. Another limitation is that respondents self-completed the questionnaires, and their responses may be based on perceptions [51]. Future studies must endeavour to include semi-structured interviews to explore for details or validate the findings.
## Conclusion
The majority of health professionals in the selected health facilities are ready to adopt and use telemedicine. Telemedicine implementation success and use in LMICs hinges on the availability of several factors, such as policy framework, trust and awareness of the health professionals in the technology and reliable ICT infrastructure. Many eHealth projects in LMICs are largely initiated and funded by central governments and are often left unsustained when funding withers off. We highlighted the factors which can help sustain the implementation and use of telemedicine. Awareness can be raised through the training and education of health professionals. This will encourage health facilities and professionals to prioritise their needs and prepare towards telemedicine adoption.
## Author contribution statement
Nathan Kumasenu Mensah; Godwin Adzakpah: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Richard Okyere Boadu; Jonathan Kissi; Obed Uwumbornyi Lasim; Martha Khainde Oyenike; Abigail Bart-Plange: Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Maxwell Ayindenaba Dalaba; Felix Sukums: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper.
## Funding statement
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
## Data availability statement
Data will be made available on request.
## References
1. Ossebaard H.C., Van Gemert-Pijnen L.. **eHealth and quality in health care: implementation time**. *Int. J. Qual. Health Care* (2016) **28** 415-419. PMID: 27029590
2. da Fonseca M.H., Kovaleski F., Picinin C.T., Pedroso B., Rubbo P.. **E-Health practices and technologies: a systematic review from 2014 to 2019**. *Healthcare (Basel)* (2021) **9**
3. Kruse C., Heinemann K.. **Facilitators and barriers to the adoption of telemedicine during the first year of COVID-19: systematic review**. *J. Med. Internet Res.* (2022) **24**
4. Balestra M.. **Telehealth and legal implications for nurse practitioners**. *J. Nurse Pract.* (2018) **14** 33-39
5. Bhaskar S., Bradley S., Chattu V.K., Adisesh A., Nurtazina A., Kyrykbayeva S.. **Telemedicine across the globe-position paper from the COVID-19 pandemic health system resilience PROGRAM (REPROGRAM) international consortium (Part 1)**. *Front. Public Health* (2020) **8**
6. O'Connell P.. **Advantages and challenges to using telehealth medicine**. *Global J. Med. Res.* (2015) **15** 19-22
7. Mauco K.L., Scott R.E., Mars M.. **Critical analysis of e-health readiness assessment frameworks: suitability for application in developing countries**. *J. Telemed. Telecare* (2018) **24** 110-117. PMID: 28008790
8. Mauco K.L., Scott R.E., Mars M.. **Validation of an e-health readiness assessment framework for developing countries**. *BMC Health Serv. Res.* (2020) **20** 575. PMID: 32576174
9. Kim E., Gellis Z.D., Bradway C., Kenaley B.. **Key determinants to using telehealth technology to serve medically ill and depressed homebound older adults**. *J. Gerontol. Soc. Work* (2019) **62** 451-474. PMID: 30040598
10. Scott R.E., Mars M.. **Telehealth in the developing world: current status and future prospects**. *Smart Homecare Technol. TeleHealth* (2015) **3** 25-37
11. Tchao E., Acquah I., Kotey S., Aggor C., Kponyo J.. **On telemedicine implementations in Ghana**. *Int. J. Adv. Comput. Sci. Appl.* (2019) **10**
12. **Ghana - rural population**. (2022)
13. Adam A., Fusheini A., Kipo-Sunyehzi D.D.. (2021)
14. Carney C.. **Delivering healthcare to rural Ghana: telmedx and Mahiri Mobile serve patients in remote areas**. *Telehealth Med. Today* (2016) **1**
15. Mars M.. **Telemedicine and advances in urban and rural healthcare delivery in Africa**. *Prog. Cardiovasc. Dis.* (2013) **56** 326-335. PMID: 24267440
16. Tchao E.T., Diawuo K., Ofosu W.K.. **Mobile telemedicine implementation with WiMAX technology: a case study of Ghana**. *J. Med. Syst.* (2017) **41** 1-10. PMID: 27817129
17. Yallah A.. (2014)
18. Yu-Tong T., Yan Z., Zhen L., Bing X., Qing-Yun C.. **Telehealth readiness and its influencing factors among Chinese clinical nurses: a cross-sectional study**. *Nurse Educ. Pract.* (2022) **58**
19. Lesher A.P., Shah S.R.. *Seminars in Pediatric Surgery* (2018)
20. Reisinger‐Kindle K., Qasba N., Cayton C., Niakan S., Knee A., Goff S.L.. **Evaluation of rapid telehealth implementation for prenatal and postpartum care visits during the COVID‐19 pandemic in an academic clinic in Springfield, Massachusetts, United States of America**. *Health Sci. Rep.* (2021) **4** e455. PMID: 34938899
21. Hoffman G.J., Webster N.J., Bynum J.P.. **A framework for aging-friendly services and supports in the age of COVID-19**. *J. Aging Soc. Pol.* (2020) **32** 450-459
22. Scott Benjamin K., Miller Geoffrey T., Fonda Stephanie J., Ronald E., Yeaw J., Gaudaen ames C., Pavliscsak Holly H.. **Advanced digital health technologies for COVID-19 and future emergencies**. *Telemed. e-Health* (2020) **26** 1226-1233
23. Kronenfeld J.P., Penedo F.J.. **Novel Coronavirus (COVID-19): telemedicine and remote care delivery in a time of medical crisis, implementation, and challenges**. *Translat. Behav. Med.* (2020) **11** 659-663
24. Dzando G., Akpeke H., Kumah A., Agada E., Lartey A.A., Nortu J.. **Telemedicine in Ghana: insight into the past and present, a narrative review of literature amidst the Coronavirus pandemic**. *J. Publ. Health Afr.* (2022) **13**
25. Holland J.J., Hatcher W., Meares W.L.. **Understanding the implementation of telemental health in rural Mississippi**. *J. Health Hum. Serv. Adm.* (2018) **41** 52-86
26. Mensah N.K., Boadu R.O., Adzakpah G., Lasim O.U., Amuakwa R.D., Taylor-Abdulai H.B.. **Electronic health records post-implementation challenges in selected hospitals: a qualitative study in the Central Region of southern Ghana**. *Health Info. Manag. J.* (2022)
27. Muigg D., Kastner P., Duftschmid G., Modre-Osprian R., Haluza D.. **Readiness to use telemonitoring in diabetes care: a cross-sectional study among Austrian practitioners**. *BMC Med. Inf. Decis. Making* (2019) **19** 26
28. Scott Kruse C., Karem P., Shifflett K., Vegi L., Ravi K., Brooks M.. **Evaluating barriers to adopting telemedicine worldwide: a systematic review**. *J. Telemed. Telecare* (2018) **24** 4-12. PMID: 29320966
29. Van Dyk L.. **A review of telehealth service implementation frameworks**. *Int. J. Environ. Res. Publ. Health* (2014) **11** 1279-1298
30. Kiberu V.M., Scott R.E., Mars M.. **Assessment of health provider readiness for telemedicine services in Uganda**. *Health Inf. Manag.* (2019) **48** 33-41. PMID: 29359588
31. Kiberu V.M., Scott R.E., Mars M.. **Assessing core, e-learning, clinical and technology readiness to integrate telemedicine at public health facilities in Uganda: a health facility - based survey**. *BMC Health Serv. Res.* (2019) **19** 266. PMID: 31035976
32. Fardazar F.E., Safari H., Habibi F., Haghighi F.A., Rezapour A.. **Hospitals' readiness to implement clinical governance**. *Int. J. Health Pol. Manag.* (2015) **4** 69
33. Abera A.A., Mengesha G.H., Musa P.F.. **Assessment of Ethiopian health facilities readiness for implementation of telemedicine**. *Commun. Assoc. Inf. Syst.* (2014) **34** 67
34. Guerrero K.S., Puls S.E., Andrew D.A.. **Transition of care and the impact on the environment of care**. *J. Nurs. Educ. Pract.* (2014) **4**
35. Mauco K.L., Scott R.E., Mars M.. **Development of an eHealth readiness assessment framework for Botswana and other developing countries: interview study**. *JMIR Med Inform* (2019) **7**
36. Bujang M.A., Baharum N.. **Sample size guideline for correlation analysis**. *World* (2016) **3** 37-46
37. Coleman A., Coleman M.F.. **Activity Theory Framework: a basis for e-health readiness assessment in health institutions**. *J. Commun.* (2013) **4** 95-100
38. Biruk S., Yilma T., Andualem M., Tilahun B.. **Health Professionals' readiness to implement electronic medical record system at three hospitals in Ethiopia: a cross sectional study**. *BMC Med. Inf. Decis. Making* (2014) **14** 1-8
39. Ajami S., Ketabi S., Isfahani S., Heidari A.. (2011)
40. Kgasi M.R., Kalema B.M.. **Assessment E-health readiness for rural South African areas**. *J Ind Intell Inf* (2014) **2** 131-135
41. Snyder R.H.. **Indicators of organizational readiness for clinical information technology/systems innovation: a Delphi study**. *Int J Med Inform* (2001) **63** 179-204. PMID: 11502432
42. DeVellis R.. (2012) 104-110
43. Sukums F., Mensah N., Mpembeni R., Kaltschmidt J., Haefeli W.E., Blank A.. **Health workers' knowledge of and attitudes towards computer applications in rural African health facilities**. *Glob. Health Action* (2014) **7**
44. Muigg D., Kastner P., Duftschmid G., Modre-Osprian R., Haluza D.. **Readiness to use telemonitoring in diabetes care: a cross-sectional study among Austrian practitioners**. *BMC Med. Inf. Decis. Making* (2019) **19** 1-10
45. Saleh S., Khodor R., Alameddine M., Baroud M.. **Readiness of healthcare providers for eHealth: the case from primary healthcare centers in Lebanon**. *BMC Health Serv. Res.* (2016) **16** 644. PMID: 27832788
46. Awol S.M., Birhanu A.Y., Mekonnen Z.A., Gashu K.D., Shiferaw A.M., Endehabtu B.F.. **Health professionals' readiness and its associated factors to implement electronic medical record system in four selected primary hospitals in Ethiopia**. *Adv. Med. Educ. Pract.* (2020) **11** 147-154. PMID: 32110135
47. Currie M., Philip L.J., Roberts A.. **Attitudes towards the use and acceptance of eHealth technologies: a case study of older adults living with chronic pain and implications for rural healthcare**. *BMC Health Serv. Res.* (2015) **15** 1-12. PMID: 25603697
48. Mensah N., Sukums F., Awine T., Meid A., Williams J., Akweongo P.. **Impact of an electronic clinical decision support system on workflow in antenatal care: the QUALMAT eCDSS in rural health care facilities in Ghana and Tanzania**. *Glob. Health Action* (2015) **8**
49. Kiberu V.M., Scott R.E., Mars M.. **Assessment of health provider readiness for telemedicine services in Uganda**. *Health Info. Manag. J.* (2019) **48** 33-41
50. Saleh S., Khodor R., Alameddine M., Baroud M.. **Readiness of healthcare providers for eHealth: the case from primary healthcare centers in Lebanon**. *BMC Health Serv. Res.* (2016) **16** 1-11. PMID: 26728278
51. Simba D., Sukums F., Kumalija C., Asiimwe S.E., Pothepragada S.K., Githendu P.W.. **Perceived usefulness, competency, and associated factors in using District health information system data among District health managers in Tanzania: cross-sectional study**. *JMIR Form Res* (2022) **6**
|
---
title: 'The association between diabetes and abdominal aortic aneurysms in men: results
of two Danish screening studies, a systematic review, and a meta-analysis of population-based
screening studies'
authors:
- Katrine Lawaetz Larsen
- Egle Kavaliunaite
- Lars Melholt Rasmussen
- Jesper Hallas
- Axel Diederichsen
- Flemming Hald Steffensen
- Martin Busk
- Lars Frost
- Grazina Urbonaviciene
- Jess Lambrechtsen
- Kenneth Egstrup
- Jes Sanddal Lindholt
journal: BMC Cardiovascular Disorders
year: 2023
pmcid: PMC10022183
doi: 10.1186/s12872-023-03160-8
license: CC BY 4.0
---
# The association between diabetes and abdominal aortic aneurysms in men: results of two Danish screening studies, a systematic review, and a meta-analysis of population-based screening studies
## Abstract
### Background
A paradoxical protective effect of diabetes on the development and progression of abdominal aortic aneurysms (AAA) has been known for years. This study aimed to investigate whether the protective role of diabetes on AAAs has evolved over the years.
### Methods
A cross-sectional study, a systematic review and meta-analysis. This study was based on two large, population-based, randomised screening trials of men aged 65–74; VIVA (2008–2011) and DANCAVAS (2014–2018), including measurement of the abdominal aorta by ultrasound or CT, respectively. Analyses were performed using multiple logistic regressions to estimate the odds ratios (ORs) for AAAs in men with diabetes compared to those not having diabetes. Moreover, a systematic review and meta-analysis of population-based screening studies of AAAs to visualise a potential change of the association between diabetes and AAAs. Studies reporting only on women or Asian populations were excluded.
### Results
In VIVA, the prevalence of AAA was $3.3\%$, crude OR for AAA in men with diabetes 1.04 ($95\%$ confidence interval, CI, 0.80-1.34), and adjusted OR 0.64 (CI 0.48-0.84). In DANCAVAS, the prevalence of AAA was $4.2\%$, crude OR 1.44 (CI 1.11-1.87), and adjusted OR 0.78 (CI 0.59-1.04). Twenty-three studies were identified for the meta-analysis ($$n = 224$$ 766). The overall crude OR was 0.90 (CI 0.77-1.05) before 2000 and 1.16 (CI 1.03-1.30) after 1999. The overall adjusted OR was 0.63 (CI 0.59-0.69) before 2000 and 0.69 (CI 0.57-0.84) after 1999.
### Conclusion
Both the crude and adjusted OR showed a statistically non-significant trend towards an increased risk of AAA by the presence of diabetes. If this represents an actual trend, it could be due to a change in the diabetes population.
### Trial registration
DANCAVAS: Current Controlled Trials: ISRCTN12157806. VIVA: ClinicalTrials.gov NCT00662480.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12872-023-03160-8.
## Background
Traditionally, abdominal aortic aneurysms (AAAs) have been seen as a manifestation of atherosclerosis in the abdominal aorta, and the common risk factors for atherosclerosis, including diabetes, were assumed to apply [1, 2]. However, numerous studies have undermined the perception of AAAs as a manifestation of atherosclerosis and found that diabetes reduces the risk of AAA and its rate of progression [3].
The mechanism underlying the protective effect of diabetes has not been established. One study found an inverse association between fasting glucose and aortic diameter [4], while another did not find an association between fasting glucose and the presence of AAA [5].
We have previously shown that elevated glycated haemoglobin is associated with reduced AAA growth [6]. Studies found that antidiabetics in general and metformin are inversely associated with the presence and growth of AAAs [7–10]. On the other hand, we found no effect of long-term use of metformin regarding the risk of ruptured AAAs in a register-based study [11].
The definition of diabetes, the limits for fasting blood sugar, and the diagnostic tests have changed over the years [12]. Similarly, there is a change in the antidiabetic treatment [12, 13] and the prevention and treatment of other comorbidities, such as atherosclerosis [13]. Thus, the population with diabetes has changed over time, and the prevalence of type 2 diabetes has increased [13–15].
As mentioned, research has shown that elevated blood sugar could be an essential driver behind impaired growth and development of AAAs [4, 6]. Doctors are more aware of the consequences of poorly regulated diabetes, and glycated haemoglobin, reflecting average blood sugar over three months, might be lower on average now than 10–20 years ago [13]. People with diabetes benefit from this focus on reducing blood sugar, but it may influence the protective effect of diabetes on the development and progression of AAAs.
The present study aimed to investigate whether the protective role of diabetes on AAAs is consistent over time. Our objectives were to compare the prevalence of diabetes in men with and without AAAs in two large, Danish, population-based, randomised screening trials. Furthermore, to conduct a meta-analysis of screening studies measuring the abdominal aorta to compare odds ratios (ORs) for the association between AAA and diabetes. Thus, if aneurysmal growth is impaired by higher levels of glycated haemoglobin [6], we might see an effect on aneurysmal growth if people with diabetes have lower level of glycated haemoglobin now than 10–20 years ago [13]. The hypothesis was to demonstrate a shift from an inverse association between diabetes and AAAs to no association—or even an increased risk of AAA by the presence of diabetes.
## Methods
This cross-sectional study is based on two population-based, randomised, clinically controlled screening trials of men aged 65–74 years in the Central Denmark Region and the Region of Southern Denmark: VIVA [16, 17] and DANCAVAS [18, 19], respectively. Also, a meta-analysis of AAA screening studies preferably in men > 60 years.
## Data source
The Danish national healthcare system is tax-supported and provides the entire Danish population (5.8 million in 2019) with free, unrestricted access to public health services and partial reimbursement for most prescribed drugs. All Danish residents are assigned a unique 10-digit civil registration number at birth or immigration. Using the civil registration system, we randomly selected the participants in our screening trials based on sex, age, and the municipality of residence. The trial participation was free of charge.
## The VIVA trial
The Viborg Vascular (VIVA) Screening *Trial is* a population-based, randomised, clinically controlled screening trial. Some 50 170 Danish men aged 65–74 years living in the Central Denmark Region were enrolled from 2008 to 2011 [16]. Briefly; study participants were randomly assigned 1:1 to triple screening or no systematic screening. Lifestyle parameters, medical history, and use of medication were self-reported. The triple screening included evaluation of AAA, hypertension, and peripheral arterial disease (PAD) with specially trained nurses performing ultrasound scans of the infrarenal abdominal aorta measuring the maximal systolic inner-to-inner anterior–posterior diameter. A maximal infrarenal aortic diameter ≥ 30 mm was defined as AAA. Dilatations ≥ 50 mm were referred for computed tomography (CT) followed by a consultation with a vascular surgeon if the aorta was ≥ 55 mm.
## The DANCAVAS trial
The Danish Cardiovascular Screening Trial (DANCAVAS) is a population-based, randomised, clinically controlled screening trial with enrolment from 2014 to 2018 in the Region of Southern Denmark [18]. A total of 47 322 Danish men aged 65–74 years were randomised 1:2 to a seven-faceted cardiovascular screening or no systematic screening. The screening included a low-dose non-contrast CT to detect aortic/iliac aneurysms and coronary artery calcification. Furthermore, an evaluation of hypertension, PAD, and telemetric assessment of the heart rhythm. Lifestyle parameters, medical history, and use of medication were self-reported. The maximal outer-to-outer anterior–posterior diameter of the infrarenal aorta was measured, and aorta ≥ 30 mm was defined AAA. Dilatations ≥ 55 were referred for contrast-enhanced CT, followed by a consultation with a vascular surgeon.
## Covariates
Based on the questionnaires, we used self-reported diabetes or the use of antidiabetics in our analyses. Oral antidiabetics were not divided into subgroups. Body mass index (BMI) was classified as follows: < 18.5, underweight; 18.5–24.9, normal; 25–29.9, overweight; and ≥ 30, obese. In our analyses, we used measurements of the anterior–posterior diameter of the aorta.
## Inclusion
We defined our study population based on the source population. For both VIVA and DANCAVAS, the inclusion criterion was a measurement of the abdominal aorta.
## Analyses
We used baseline data of the VIVA and DANCAVAS populations to investigate whether the presence of AAA was associated with diabetes. Both simple and multiple logistic regression estimated the ORs for AAA and diabetes for both populations.
Group comparisons were made using either parametric or non-parametric tests. $P \leq 0.05$ was considered significant. Crude and adjusted ORs were calculated using simple and multiple logistic regression. The $95\%$ confidence intervals (CIs) are given with the corresponding OR.
ORs with CIs from the individual studies were combined visually in a Forest plot sorted by data sampling time. If the year of inclusion was not given, we used the year of publication ($$n = 4$$). If an OR was not available, we calculated crude ORs based on the prevalence of diabetes in people with and without an AAA ($$n = 18$$). We performed a subgroup analysis by stratifying before and after the millennium. The measure of consistency, I [2], was reported.
All analyses were performed using Stata® Release 15.1 (StataCorp, College Station, TX, USA).
## Potential confounders
Questionnaires provided information on the participants' medical histories, smoking status, and current medications. In both populations, current medications and comorbidities were self-reported. PAD was defined as an ankle-brachial index < 0.9 or > 1.4 at the screening. In VIVA, we extracted previous acute myocardial infarction (AMI) and chronic obstructive pulmonary disease from the National Patient Register using the civil registration number. Medications were grouped according to the Anatomical Therapeutic Chemical (ATC) classification system developed by the World Health Organization [20]. Medications included oral blood-glucose-lowering medications (A10B), insulin (A10A), statins (C10AA), angiotensin-converting enzyme inhibitors (C09AA, C09BA, C09BB, C02EA, and C02LM), angiotensin II receptor antagonists (C09C), beta-blockers (C07), calcium antagonists (C08), low dose acetylsalicylic acid (B01AC06), non-steroidal anti-inflammatory drugs (M01AB), oral corticosteroids (H02), and inhaled corticosteroids (R03BA). All medications were categorised in groups, as mentioned above, and not specified by subgroup. Antidiabetic therapy was categorised in oral antidiabetics or insulins.
Potential confounders were selected based on an automated empirical procedure. We selected a wide range of variables and calculated the OR for having AAA and diabetes with or without the potential confounder included. If the OR for diabetes changed > $5\%$ in either direction by adding the potential confounder, we added the variable in our multiple analysis. Thus, the following factors that empirically behaved as confounders were included in our model: a) age, b) smoking status, c) grouped BMI, d) presence of PAD, e) previous AMI, f) self-reported hypertension, g) use of statins, h) acetylsalicylic acid, i) beta-blockers, and j) angiotensin-converting enzyme inhibitors grouped with angiotensin II receptor antagonists.
## Literature extraction
Literature searches were performed in Embase, Medline, and Cochrane databases. KL conducted the search, and the last search was performed on November 16, 2018. The search terms were abdominal aortic aneurysm and screening, using both Medical Subject Headings and free text searches (see the supplementary material for the search strategy). These citations were searched manually to identify studies comprising abdominal aortic measurements with the prevalence of diabetes. In addition, the bibliographies from the citations as well as reviews were searched to identify additional studies.
## Eligibility
We limited the articles to those written in English and included studies with original data regarding AAA screening and the presence of diabetes. Studies solely examining women were excluded, as the prevalence of AAA is different in men and women [2, 21]. Furthermore, studies with only Asian populations were excluded because the prevalence of both diabetes and AAA is reversed. The prevalence of AAA in Asian populations is up to tenfold lower [22], and the prevalence of diabetes is up to fourfold higher than Caucasians [23]. We excluded studies based on a select group of participants (e.g., a specific diagnosis) or a referral for outpatient evaluation (e.g., carotid or cardiac) concomitant with screening for AAAs. Furthermore, studies with self-referred participants or people participating due to some select membership and studies with information about AAAs based on hospital or discharge records were excluded. Thus, we only included population-based screenings to minimise the risk of bias. The outcome of interest was the OR with corresponding CI for having AAA and diabetes.
The eligibility assessment was performed independently by two authors, KL and EK, and a consensus was resolved. Data extraction was performed by KL and verified by EK. Extracted data included the year of study sampling, country, inclusion and exclusion criteria, number of participants, ethnicity, sex, age, the prevalence of AAA and diabetes, and the OR (crude and adjusted) for having AAA and diabetes with the corresponding CI. If possible, in studies with both men and women, we extracted data only for men. KL contacted eight authors of original studies for missing information; one responded but provided no additional information. A review protocol has not been published. PRISMA guidelines were used for the preparation and reporting of the meta-analysis [24].
## Association between AAA and diabetes, VIVA
The VIVA population comprised 18 697 men with a median age of 69 years (Table 1). The participation rate was $74.7\%$ [17]. The median aorta was 17.8 mm for men with diabetes and 18.3 mm for men without diabetes ($p \leq 0.001$). The prevalence of AAA was $3.3\%$.Table 1Characteristics of the VIVA and DANCAVAS populationCharacteristicsVIVA + DMVIVA – DMDANCAVAS + DMDANCAVAS–DMMen($$n = 2027$$)($$n = 16$$ 632)($$n = 1245$$)($$n = 9223$$)Age69 (67–71)69 (67–71)69 (67–71)69 (67–71)Family history of AAA64 ($3.2\%$)538 ($3.2\%$)56 ($4.5\%$)432 ($4.7\%$)BMI < 25365 ($18.0\%$)5709 ($34.3\%$)142 ($11.4\%$)2332 ($25.3\%$) ≥ 25—< 30935 ($46.1\%$)8159 ($49.1\%$)505 ($40.6\%$)4658 ($50.5\%$) ≥ 30700 ($34.5\%$)2535 ($15.2\%$)596 ($47.9\%$)2222 ($24.1\%$)Smoking Current378 ($18.6\%$)3550 ($21.3\%$)197 ($15.8\%$)1385 ($15.0\%$) Former1141 ($56.3\%$)8109 ($48.8\%$)718 ($57.7\%$)4804 ($52.1\%$) Never505 ($24.9\%$)4963 ($29.8\%$)328 ($26.3\%$)3002 ($32.5\%$)Abdominal aortic measurement17 (16–19)18 (16–20)19 (17–21)19 (18–21)AAA69 ($3.4\%$)546 ($3.3\%$)71 ($5.7\%$)372 ($4.0\%$)Self-reported comorbidity Diabetes mellitus2027 ($100\%$)-1245 ($100\%$)- Previous AMI82 ($4.0\%$)411 ($2.5\%$)145 ($11.6\%$)548 ($5.9\%$) Hypertension1424 ($70.3\%$)6529 ($39.3\%$)950 ($76.3\%$)3837 ($41.6\%$) COPD70 ($3.5\%$)394 ($2.4\%$)104 ($8.4\%$)605 ($6.6\%$) PADa479 ($23.6\%$)2281 ($13.7\%$)278 ($22.3\%$)881 ($9.6\%$)Drugs Antidiabetics, oral953 ($47.0\%$)-895 ($71.9\%$)- Insulin452 ($22.3\%$)-306 ($24.6\%$)- Statins1507 ($74.3\%$)5540 ($33.3\%$)949 ($76.2\%$)2787 ($30.2\%$) ACE inhibitors and ATII antagonists1326 ($65.4\%$)4780 ($28.7\%$)852 ($68.4\%$)2827 ($30.7\%$) Beta-blockers710 ($35.0\%$)3354 ($20.2\%$)393 ($31.6\%$)1343 ($14.6\%$) Acetylsalicylic acid1259 ($62.1\%$)5192 ($31.2\%$)425 ($34.1\%$)1311 ($14.2\%$)Data are given as n (%) or median (interquartile range) unless otherwise notedDM Diabetes mellitus, AAA Abdominal aortic aneurysm, BMI Body mass index, AMI Acute myocardial infarction, COPD Chronic obstructive pulmonary disease, PAD Peripheral arterial disease, ACE Angiotensin-converting enzyme, ATII Angiotensin II receptoraDiagnosed by ankle-brachial index at the screening Using logistic regression, we found a crude OR of 1.04 ($95\%$ CI 0.80–1.34) for having an AAA and diabetes compared to no diabetes. When adjusting for confounders, the OR reduced to 0.64 (CI 0.48–0.84). Age, BMI, smoking (both former and present), presence of hypertension and PAD, and the use of statins and acetylsalicylic acid were significant factors (Supplementary Table S1). Adjusting for the potential variables one at a time, we found that the use of statins shifted the OR towards a positive association and away from the crude OR more than the other potential confounders (Supplementary Table S2).
## Association between AAA and diabetes, DANCAVAS
The DANCAVAS population comprised 10 468 men with a median age of 69 years (Table 1). The participation rate was $62.4\%$ [19]. The median aorta was 19.6 mm for men with diabetes and 19.8 mm for men without diabetes ($$p \leq 0.034$$). The prevalence of AAA was $4.2\%$.
Using logistic regression, we found a positive association between having AAA and diabetes compared to no diabetes (crude OR 1.44, CI 1.11–1.87). However, when adjusting for confounders, the OR reduced to 0.78 (CI 0.59–1.04). Age, BMI, smoking (both former and present), presence of hypertension and PAD, previous AMI, and the use of statins and acetylsalicylic acid were significant factors (Supplementary Table S1). Again, we found that adjusting for the use of statins distinctly shifted the OR compared to other potential confounders (Supplementary Table S2).
## Meta-analysis
In Embase, Medline, and Cochrane databases, we identified 2758 articles and abstracts, 21 of which were relevant and assessed in detail [2, 4, 21, 25–42]. Moreover, we found two eligible studies reporting unique data by manual search [1, 43] (Fig. 1). Thus, 23 population-based studies were included in the review (Table 2 and Supplementary Table S3).Fig. 1Prisma flow chart of studies in the meta-analysisTable 2Studies included in the meta-analysis; 23 studies and the results from our Danish studiesStudyAgeMen %AAA/no AAAAAA + DM/ AAA-DMCrude OR (CI)Adjusted OR (CI)RefBrazila 1987–93 ≥ 55100 h$\frac{17}{9951}$/160.79 (0.10–6.04)- [27]Italy 1991–9465–$\frac{7546.370}{15319}$/560.98 (0.48–2.02)- [37]USA 1992–93 ≥ 6541.3 h$\frac{252}{170140}$/2120.84 (0.59–1.20)- [30]USAb 1992–9550–$\frac{7997.22335}{70085}$NA-0.68 (0.60–0.77) [33]USAc 1992–9550–$\frac{7997.21031}{70085}$NA-0.54 (0.44–0.65) [33]England 199365–75100 h$\frac{219}{237811}$/2080.83 (0.44–1.55)- [39]Norwayd 1994–9555–7448.0 h$\frac{251}{23357}$/2440.69 (0.32–1.51)- [43]Belgium 1995–9665&75100 h$\frac{33}{6947}$/262.13 (0.89–5.06)- [42]USAb 1995–9750–$\frac{7997.41304}{50828}$NA-0.60 (0.50–0.71) [21]USAc 1995–9750–$\frac{7997.4613}{50828}$NA-0.50 (0.39–0.65) [21]Netherlandsd 1995 ≥ 5542.0 h$\frac{91}{21267}$/840.72 (0.33–1.57)- [1]Australia 1996–9965–79100 h$\frac{933}{11270103}$/8300.89 (0.72–1.11)0.79 (0.63–0.98) [4]England 199665–8043.4 h$\frac{178}{216310}$/1681.08 (0.56–2.10)0.80 (0.41–1.58) [2]Scotland 2001–0465–74100 h$\frac{414}{773243}$/3710.93 (0.68–1.29)- [29]Brazile 2002–03 ≥ $\frac{6034.321}{8065}$/161.63 (0.59–4.51)- [26]Swedenf 2006–1065100 h$\frac{233}{1437824}$/2090.83 (0.54–1.26)- [41]Sweden 2007–0765–75100 h$\frac{168}{1408110}$/1581.31 (0.69–2.50)- [40]Italy 2007–09 ≥ $\frac{6552.6512}{772266}$/4011.41 (1.07–1.84)- [35]Spain 2007–1065–74100 h$\frac{15}{6363}$/120.77 (0.21–2.76)- [36]Spaing 2008–0965100 h$\frac{37}{739}$NA0.28 (0.08–0.90)0.38 (0.11–1.06) [25]Sweden 2008–1070100 h$\frac{107}{460819}$/881.19 (0.72–1.96)- [32]Denmark 2008–1164–75100 h$\frac{617}{1808069}$/5461.04 (0.80–1.34)0.64 (0.48–0.84)Italy 2010–1360–8548.6 h$\frac{19}{7351}$/180.35 (0.05–2.64)- [28]Spain 2013–14 ≥ 60100 h$\frac{11}{9983}$/81.00 (0.26–3.80)- [38]Italy 2013–1650–7563.7 h$\frac{56}{233511}$/451.51 (0.77–2.95)- [31]Belgium 2014–1465–8565.6 h$\frac{35}{6877}$/281.06 (0.45–2.48)- [34]Denmark 2014–1865–74100 h$\frac{443}{1002571}$/3721.44 (1.11–1.87)0.78 (0.59–1.04)AAA Abdominal aortic aneurysm, DM Diabetes, OR Odds ratio, CI Confidence interval, Ref Reference numberaA study of three groups, the reference group based on the general population is depictedbAorta 30–39 mm compared to < 30 mmcAorta ≥ 40 mm compared to < 30 mmdExtra by manual search. Both studies defined AAA as aorta ≥ 35 mmeDifferent number in their table (total diabetes) compared to the total number of 834fOnly data on 14,611 despite another number givengUnable to measure aorta in 5hPrevalence and OR in men only The studies were conducted between 1987 and 2018. The majority of the studies did not report crude or adjusted ORs for AAA and diabetes, but we calculated crude ORs by extracting the prevalence. Moreover, 11 studies ($47.8\%$) included only men, but crude ORs were calculated for men in 15 studies ($65.2\%$) based on the reported numbers.
In the meta-analysis, we found a change in the crude OR over time, as visualised in Fig. 2. However, substantial confounding was corrected, and the adjusted OR remained almost unchanged over time (Fig. 3).Fig. 2Forest plot of the crude OR of having an AAA and diabetes compared to no diabetes sorted by year of inclusion and a subgroup analysis before and after the millennium. * A study of three groups, the reference group based on the general population is depictedFig. 3Forest plot of the adjusted OR of having an AAA and diabetes compared to no diabetes sorted by year of inclusion and a subgroup analysis before and after the millennium. * The authors compared aorta 30–39 mm with no aneurysm; †The authors compared AAA ≥ 40 mm with no aneurysm
## Discussion
In this observational study of 29 165 men aged 65–74 years in whom the association between AAA and diabetes was estimated, we found a significant inverse association in the VIVA population (2008–2011), but not in the later DANCAVAS population (2014–2018).
In VIVA, after adjusting for potential confounders, we found that the presence of diabetes significantly reduced the OR for AAA by $38\%$. In DANCAVAS, in the crude analysis, men with diabetes had almost $50\%$ higher odds of having an AAA compared to no diabetes. When adjusting for potential confounders, the inverse association was not significant. Overall, the use of statins affected the OR towards a positive association, which could be due to the use of statins per se or an indirect adjustment for high cardiovascular risk. Previous studies have found conflicting results regarding the use of statins, finding either no effect on growth rate [7, 44] or a reduced growth rate of AAA [45]. Furthermore, studies have observed an increased risk of AAA with the presence of atherosclerosis, including claudication [1] and coronary artery disease [41].
The results of our two Danish screening trials could indicate a shift in the positive association between diabetes and AAA. Over time, there has been a change in the prevalence of AAA-associated risk factors (Table 1). An independent positive association seems to persist after adjusting for these dispositions in VIVA, but not DANCAVAS. The meta-analysis found a similar shift over time towards a change in the risk of having an AAA and diabetes. However, the independent protective effect of diabetes on AAAs remained unchanged. When we compared the adjusted analyses of the studies included in the meta-analysis, we found that none was adjusted for statins. However, they were adjusted to some form of atherosclerosis (Table S3). It is important to emphasise that the definition of diabetes, diagnostic tests, and antidiabetic therapy have improved over the years [12–15]. The prevalence of people with diabetes has increased [14, 46], and the management, resulting in an increased prevalence of people receiving antidiabetic therapy such as metformin [13, 15]. Consequently, there could be an earlier diagnosis and improved glycaemic and risk profiles resulting in less severe and better-managed diabetes in the later studies than in the earlier studies.
Studies have found a decreased prevalence of AAAs [47] with the use of metformin and a reduced growth rate [8–10]. It is impossible to deduce anything regarding metformin from this study since we do not have information about the subgroups of oral antidiabetic therapy.
## Strengths and limitations
Our study has some strengths. First, the data are based on two large, population-based, randomised, clinically controlled screening trials with 29 165 male participants. One trial comprises ultrasound-verified AAAs and the other non-contrast CT-verified AAAs. As the participants were randomly selected according to their unique civil registration number and participation rates were high, the risk of selection bias was minimal. On the other hand, we cannot reject selection bias, as people with severe diabetes, or any severe disease in general, may be unwilling to attend voluntary screenings [48, 49]. In our study, the prevalence of diabetes was $11.0\%$ for VIVA (launched in 2008). In a study based on the Danish National Diabetes Register, the prevalence of diabetes was approximately $15\%$ for men aged 70 years [50]. Thus, we have an underrepresentation of men with diabetes, which could shift the OR towards or above 1. However, suppose we have an underrepresentation of men with severe diabetes. In that case, other population-based studies may as well, and, therefore, our results are comparable when we look at the shift over time. Our meta-analysis was comprehensive with only two search terms, abdominal aortic aneurysm and screening. We found several studies not mentioning diabetes at all except in the methods and a table. We assumed that the risk of measurement error across studies was minor, as screening for AAAs with ultrasound is reasonably reproducible in skilled hands. We took precautions regarding the meta-analysis and applied several exclusion criteria to minimise the risk of bias across studies. We excluded numerous studies based on a selected group of participants because we wanted a representative cross-section of the general population. There are several AAA screening studies in patient populations undergoing, for example, coronary artery bypass grafting. We excluded studies of Asian populations because Asian and Caucasian populations are not comparable in both AAA and diabetes [22, 23]. Lastly, we excluded studies with self-referred people because this may increase the risk of bias, as, for example, self-referred people are likely better educated or wealthier, and people with relatives with an AAA may be more prone to attend screenings for AAA. First degree relatives are well-known as having a higher risk of AAA [4, 21].
Our study has several potential limitations. Our data are based on self-reported information regarding lifestyle, medication, and previous illnesses and could, therefore, contain inaccuracies and recall bias, leading to misclassifications. We did not combine the two datasets in our analyses. The studies used different types of imaging for diagnosing AAA, and the prevalence of AAAs may differ between VIVA and DANCAVAS due to the nature of the screenings. The DANCAVAS screening was more comprehensive than the VIVA screening, and people may have attended despite a known diagnosis of AAA. *Regarding* generalizability, we only included men in our Danish screening trials, and the interpretation must be restricted to men. Furthermore, although the data originated from a clinical trial, the data should be regarded as observational, entailing a risk of confounding as addressed in this discussion. We tried to eliminate bias by doing the systematic review with broad search terms and not including "diabetes". Studies with positive findings of diabetes are reported in the abstract, but the broad search made it possible to include studies that focused elsewhere. Given that most of the studies included in our meta-analysis investigated the prevalence of AAAs, we assume the risk of publication bias or selective reporting is minimal. However, some screening studies do not report the prevalence of diabetes in detail. Another potential limitation is the definition of AAA. Most studies defined AAA as aorta ≥ 30 mm, some with a ratio ≥ 1.5 (infrarenal/suprarenal), whereas some defined it as aorta ≥ 35–40 mm. There is a risk of information bias since we sorted the studies by publication if the year of inclusion was missing. Our meta-analysis comprised population-based screening studies, which have the advantage of low expenses and are often faster to complete. However, they cannot exclude potential confounding factors, and we cannot conclude anything about causality. On the other hand, we focused solely on population-based studies based on the general population, i.e., every citizen could be included if he or she were the right age at the given time. Therefore, our results are based on a representative cross-section of the population, both the meta-analysis and the screening trials. We found some heterogeneity in our meta-analysis. However, we have included several studies in the analysis, and we mostly have overlap in the CI’s and with some CI’s rather wide. By including adjusted estimates in our model, we tried to eliminate the effect of potential confounders. However, there is a risk of residual confounding. The comparison of adjusted ORs in our meta-analysis carries some uncertainty. The studies do not adjust for precisely the same potential confounders, and data are based on a small number of studies. However, we added the potential confounders included in the studies in Table S1. Lastly, we estimated the association between AAA and diabetes, but in this study, it is impossible to know which appeared first; the AAA or diabetes.
## Conclusion
Both the crude and adjusted OR showed a statistically non-significant trend towards an increased risk of AAA by the presence of diabetes. If this represents an actual trend, it could be due to a change in the diabetes population.
## Supplementary Information
Additional file 1: Table S1. Adjusted odds ratio of the association between diabetes and abdominal aortic aneurysms with $95\%$ confidence intervals and the potential confounders. Table S2. Crude and adjusted odds ratios (ORs) of the association between diabetes and abdominal aortic aneurysms with $95\%$ confidence intervals. Table S3.Studies included in the meta-analysis; 23 studies identified in databases and the results from our Danish studies.
## References
1. 1.Pleumeekers HJ, Hoes AW, van der Does E, van Urk H, Hofman A, de Jong PT, et al. Aneurysms of the abdominal aorta in older adults. The Rotterdam Study Am J Epidemiol. 1995;142(12):1291–9.
2. Kanagasabay R, Gajraj H, Pointon L, Scott RA. **Co-morbidity in patients with abdominal aortic aneurysm**. *J Med Screen* (1996.0) **3** 208-210. DOI: 10.1177/096914139600300410
3. Xiong J, Wu Z, Chen C, Wei Y, Guo W. **Association between diabetes and prevalence and growth rate of abdominal aortic aneurysms: A meta-analysis**. *Int J Cardiol* (2016.0) **221** 484-495. DOI: 10.1016/j.ijcard.2016.07.016
4. Le MT, Jamrozik K, Davis TM, Norman PE. **Negative association between infra-renal aortic diameter and glycaemia: the Health in Men Study**. *Eur J Vasc Endovasc Surg* (2007.0) **33** 599-604. DOI: 10.1016/j.ejvs.2006.12.017
5. Bengtsson H, Bergqvist D, Ekberg O, Janzon L. **A population based screening of abdominal aortic aneurysms (AAA)**. *Eur J Vasc Surg* (1991.0) **5** 53-57. DOI: 10.1016/S0950-821X(05)80927-2
6. Kristensen KL, Dahl M, Rasmussen LM, Lindholt JS. **Glycated Hemoglobin Is Associated With the Growth Rate of Abdominal Aortic Aneurysms: A Substudy From the VIVA (Viborg Vascular) Randomized Screening Trial**. *Arterioscler Thromb Vasc Biol* (2017.0) **37** 730-736. DOI: 10.1161/ATVBAHA.116.308874
7. Thompson A, Cooper JA, Fabricius M, Humphries SE, Ashton HA, Hafez H. **An analysis of drug modulation of abdominal aortic aneurysm growth through 25 years of surveillance**. *J Vasc Surg* (2010.0) **52** 55-61. DOI: 10.1016/j.jvs.2010.02.012
8. Fujimura N, Xiong J, Kettler EB, Xuan H, Glover KJ, Mell MW. **Metformin treatment status and abdominal aortic aneurysm disease progression**. *J Vasc Surg* (2016.0) **64** 46-54. DOI: 10.1016/j.jvs.2016.02.020
9. Golledge J, Moxon J, Pinchbeck J, Anderson G, Rowbotham S, Jenkins J. **Association between metformin prescription and growth rates of abdominal aortic aneurysms**. *Br J Surg* (2017.0) **104** 1486-1493. DOI: 10.1002/bjs.10587
10. Itoga NK, Rothenberg KA, Suarez P, Ho TV, Mell MW, Xu B. **Metformin prescription status and abdominal aortic aneurysm disease progression in the U.S. veteran population**. *J Vasc Surg* (2019.0) **69** 710-6. DOI: 10.1016/j.jvs.2018.06.194
11. Kristensen KL, Pottegard A, Hallas J, Rasmussen LM, Lindholt JS. **Metformin treatment does not affect the risk of ruptured abdominal aortic aneurysms**. *J Vasc Surg* (2017.0) **66** 768-774. DOI: 10.1016/j.jvs.2017.01.070
12. Christensen DH, Rungby J, Thomsen RW. **Nationwide trends in glucose-lowering drug use, Denmark, 1999–2014**. *Clin Epidemiol* (2016.0) **8** 381-387. DOI: 10.2147/CLEP.S113211
13. Thomsen RW, Baggesen LM, Svensson E, Pedersen L, Norrelund H, Buhl ES. **Early glycaemic control among patients with type 2 diabetes and initial glucose-lowering treatment: a 13-year population-based cohort study**. *Diabetes Obes Metab* (2015.0) **17** 771-780. DOI: 10.1111/dom.12484
14. 14.Fang M. Trends in the Prevalence of Diabetes Among U.S. Adults: 1999–2016. Am J Prev Med. 2018;55(4):497–505.
15. 15.Fang M. Trends in Diabetes Management Among US Adults: 1999–2016. J Gen Intern Med. 2020;35(5):1427–34. 10.1007/s11606-019-05587-2.
16. Grondal N, Sogaard R, Henneberg EW, Lindholt JS. **The Viborg Vascular (VIVA) screening trial of 65–74 year old men in the central region of Denmark: study protocol**. *Trials* (2010.0) **11** 67. DOI: 10.1186/1745-6215-11-67
17. Grondal N, Sogaard R, Lindholt JS. **Baseline prevalence of abdominal aortic aneurysm, peripheral arterial disease and hypertension in men aged 65–74 years from a population screening study (VIVA trial)**. *Br J Surg* (2015.0) **102** 902-906. DOI: 10.1002/bjs.9825
18. Diederichsen AC, Rasmussen LM, Sogaard R, Lambrechtsen J, Steffensen FH, Frost L. **The Danish Cardiovascular Screening Trial (DANCAVAS): study protocol for a randomized controlled trial**. *Trials* (2015.0) **16** 554. DOI: 10.1186/s13063-015-1082-6
19. Lindholt JS, Rasmussen LM, Sogaard R, Lambrechtsen J, Steffensen FH, Frost L. **Baseline findings of the population-based, randomized, multifaceted Danish cardiovascular screening trial (DANCAVAS) of men aged 65–74 years**. *Br J Surg* (2019.0) **106** 862-871. DOI: 10.1002/bjs.11135
20. Kildemoes HW, Sorensen HT, Hallas J. **The Danish National Prescription Registry**. *Scand J Public Health* (2011.0) **39** 38-41. DOI: 10.1177/1403494810394717
21. Lederle FA, Johnson GR, Wilson SE, Chute EP, Hye RJ, Makaroun MS. **The aneurysm detection and management study screening program: validation cohort and final results. Aneurysm Detection and Management Veterans Affairs Cooperative Study Investigators**. *Arch Intern Med* (2000.0) **160** 1425-30. DOI: 10.1001/archinte.160.10.1425
22. Salem MK, Rayt HS, Hussey G, Rafelt S, Nelson CP, Sayers RD. **Should Asian men be included in abdominal aortic aneurysm screening programmes?**. *Eur J Vasc Endovasc Surg* (2009.0) **38** 748-749. DOI: 10.1016/j.ejvs.2009.07.012
23. Moody A, Cowley G, Ng Fat L, Mindell JS. **Social inequalities in prevalence of diagnosed and undiagnosed diabetes and impaired glucose regulation in participants in the Health Surveys for England series**. *BMJ Open* (2016.0) **6** e010155. DOI: 10.1136/bmjopen-2015-010155
24. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP. **The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration**. *J Clin Epidemiol* (2009.0) **62** e1-34. DOI: 10.1016/j.jclinepi.2009.06.006
25. Barba A, Vega de Ceniga M, Estallo L, de la Fuente N, Viviens B, Izagirre M. **Prevalence of abdominal aortic aneurysm is still high in certain areas of southern Europe**. *Ann Vasc Surg* (2013.0) **27** 1068-73. DOI: 10.1016/j.avsg.2013.01.017
26. Barros F, Pontes S, Taylor M, Roelke L, Sandri J, De Melo JC. **Screening for abdominal aortic aneurysm in the population of the city of Vitoria, ES**. *Brazil J Vasc Bras* (2005.0) **4** 59-65
27. Bonamigo TP, Siqueira I. **Screening for abdominal aortic aneurysms**. *Rev Hosp Clin Fac Med Sao Paulo* (2003.0) **58** 63-68. DOI: 10.1590/S0041-87812003000200002
28. Corrado G, Durante A, Genchi V, Trabattoni L, Beretta S, Rovelli E. **Prevalence of previously undiagnosed abdominal aortic aneurysms in the area of Como: the ComoCuore "looking for AAA" ultrasonography screening**. *Int J Cardiovasc Imaging* (2016.0) **32** 1213-1217. DOI: 10.1007/s10554-016-0911-3
29. Duncan JL, Harrild KA, Iversen L, Lee AJ, Godden DJ. **Long term outcomes in men screened for abdominal aortic aneurysm: prospective cohort study**. *BMJ* (2012.0) **344** e2958. DOI: 10.1136/bmj.e2958
30. Freiberg MS, Arnold AM, Newman AB, Edwards MS, Kraemer KL, Kuller LH. **Abdominal aortic aneurysms, increasing infrarenal aortic diameter, and risk of total mortality and incident cardiovascular disease events: 10-year follow-up data from the Cardiovascular Health Study**. *Circulation* (2008.0) **117** 1010-1017. DOI: 10.1161/CIRCULATIONAHA.107.720219
31. Gianfagna F, Veronesi G, Tozzi M, Tarallo A, Borchini R, Ferrario MM. **Prevalence of Abdominal Aortic Aneurysms in the General Population and in Subgroups at High Cardiovascular Risk in Italy. Results of the RoCAV Population Based Study**. *Eur J Vasc Endovasc Surg* (2018.0) **55** 633-9. DOI: 10.1016/j.ejvs.2018.01.008
32. Hager J, Lanne T, Carlsson P, Lundgren F. **Lower prevalence than expected when screening 70-year-old men for abdominal aortic aneurysm**. *Eur J Vasc Endovasc Surg* (2013.0) **46** 453-459. DOI: 10.1016/j.ejvs.2013.07.014
33. Lederle FA, Johnson GR, Wilson SE, Chute EP, Littooy FN, Bandyk D. **Prevalence and associations of abdominal aortic aneurysm detected through screening. Aneurysm Detection and Management (ADAM) Veterans Affairs Cooperative Study Group**. *Ann Intern Med* (1997.0) **126** 441-9. DOI: 10.7326/0003-4819-126-6-199703150-00004
34. Makrygiannis G, Labalue P, Erpicum M, Schlitz M, Seidel L, El Hachemi M. **Extending Abdominal Aortic Aneurysm Detection to Older Age Groups: Preliminary Results from the Liege Screening Programme**. *Ann Vasc Surg* (2016.0) **36** 55-63. DOI: 10.1016/j.avsg.2016.02.034
35. Palombo D, Lucertini G, Pane B, Mazzei R, Spinella G, Brasesco PC. **District-based abdominal aortic aneurysm screening in population aged 65 years and older**. *J Cardiovasc Surg* (2010.0) **51** 777-782. PMID: 21124273
36. Salvador-Gonzalez B, Martin-Baranera M, Borque-Ortega A, Saez-Saez RM, de Albert-Delas VM, Carreno-Garcia E. **Prevalence of Abdominal Aortic Aneurysm in Men Aged 65–74 Years in a Metropolitan Area in North-East Spain**. *Eur J Vasc Endovasc Surg* (2016.0) **52** 75-81. DOI: 10.1016/j.ejvs.2016.04.005
37. Simoni G, Pastorino C, Perrone R, Ardia A, Gianrossi R, De Cian F. **Screening for abdominal aortic aneurysms and associated risk factors in a general population**. *Eur J Vasc Endovasc Surg* (1995.0) **10** 207-210. DOI: 10.1016/S1078-5884(05)80113-3
38. Siso-Almirall A, Kostov B, Navarro Gonzalez M, Cararach Salami D, Perez Jimenez A, Gilabert Sole R. **Abdominal aortic aneurysm screening program using hand-held ultrasound in primary healthcare**. *PLoS ONE* (2017.0) **12** e0176877. DOI: 10.1371/journal.pone.0176877
39. Smith FC, Grimshaw GM, Paterson IS, Shearman CP, Hamer JD. **Ultrasonographic screening for abdominal aortic aneurysm in an urban community**. *Br J Surg* (1993.0) **80** 1406-1409. DOI: 10.1002/bjs.1800801117
40. Stackelberg O, Wolk A, Eliasson K, Hellberg A, Bersztel A, Larsson SC. **Lifestyle and Risk of Screening-Detected Abdominal Aortic Aneurysm in Men**. *J Am Heart Assoc* (2017.0) **6** e004725. DOI: 10.1161/JAHA.116.004725
41. Svensjo S, Bjorck M, Gurtelschmid M, Djavani Gidlund K, Hellberg A, Wanhainen A. **Low prevalence of abdominal aortic aneurysm among 65-year-old Swedish men indicates a change in the epidemiology of the disease**. *Circulation* (2011.0) **124** 1118-1123. DOI: 10.1161/CIRCULATIONAHA.111.030379
42. Vazquez C, Sakalihasan N, D'Harcour JB, Limet R. **Routine ultrasound screening for abdominal aortic aneurysm among 65- and 75-year-old men in a city of 200,000 inhabitants**. *Ann Vasc Surg* (1998.0) **12** 544-549. DOI: 10.1007/s100169900198
43. Jorgensen L, Singh K, Berntsen GK, Jacobsen BK. **A population-based study of the prevalence of abdominal aortic aneurysms in relation to bone mineral density: the Tromso study**. *Am J Epidemiol* (2004.0) **159** 945-949. DOI: 10.1093/aje/kwh132
44. Ferguson CD, Clancy P, Bourke B, Walker PJ, Dear A, Buckenham T. **Association of statin prescription with small abdominal aortic aneurysm progression**. *Am Heart J* (2010.0) **159** 307-313. DOI: 10.1016/j.ahj.2009.11.016
45. Karlsson L, Gnarpe J, Bergqvist D, Lindback J, Parsson H. **The effect of azithromycin and Chlamydophilia pneumonia infection on expansion of small abdominal aortic aneurysms–a prospective randomized double-blind trial**. *J Vasc Surg* (2009.0) **50** 23-29. DOI: 10.1016/j.jvs.2008.12.048
46. 46.World Health Organization. Global report on diabetes. 2016. ISBN 9789241565257.
47. Hsu CY, Su YW, Chen YT, Tsai SH, Chang CC, Li SY. **Association between use of oral-antidiabetic drugs and the risk of aortic aneurysm: a nested case-control analysis**. *Cardiovasc Diabetol* (2016.0) **15** 125. DOI: 10.1186/s12933-016-0447-9
48. **Abdominal Aortic Aneurysm Screening Reduces All-Cause Mortality: Make Screening Great Again**. *Angiology* (2018.0) **69** 205-211. DOI: 10.1177/0003319717693107
49. Hansen TB, Lindholt JS, Sogaard R. **Role of Experience With Preventive Medication and Personal Risk Attitude in Non-Attendance at Triple Vascular Screening**. *Eur J Vasc Endovasc Surg* (2018.0) **56** 282-290. DOI: 10.1016/j.ejvs.2018.04.016
50. Carstensen B, Borch-Johnsen K. **Register-based studies of diabetes**. *Scand J Public Health* (2011.0) **39** 175-179. DOI: 10.1177/1403494811404279
|
---
title: CT-based Hounsfield unit values reflect the degree of steatohepatitis in patients
with low-grade fatty liver disease
authors:
- Ha Neul Kim
- Hong Jae Jeon
- Hei Gwon Choi
- In Sun Kwon
- Woo Sun Rou
- Jeong Eun Lee
- Tae Hee Lee
- Seok Hyun Kim
- Byung Seok Lee
- Kyung Sook Shin
- Hyun Jung Lee
- Hyuk Soo Eun
journal: BMC Gastroenterology
year: 2023
pmcid: PMC10022198
doi: 10.1186/s12876-023-02717-3
license: CC BY 4.0
---
# CT-based Hounsfield unit values reflect the degree of steatohepatitis in patients with low-grade fatty liver disease
## Abstract
### Background/Aims
Nonalcoholic fatty liver disease (NAFLD) is the most common liver disease worldwide. Ultrasound, the most used tool for diagnosing NAFLD, is operator-dependent and shows suboptimal performance in patients with mild steatosis. However, few studies have been conducted on whether alternative noninvasive methods are useful for diagnosing mild hepatic steatosis. Also, little is known about whether noninvasive tests are useful for grading the severity of hepatic steatosis or the degree of intrahepatic inflammation. Therefore, we aimed to evaluate whether the HSI, the FLI and HU values in CT could be used to discriminate mild hepatic steatosis and to evaluate the severity of hepatic steatosis or the degree of intrahepatic inflammation in patients with low-grade fatty liver disease using liver biopsy as a reference standard.
### Methods
Demographic, laboratory, CT imaging, and histological data of patients who underwent liver resection or biopsy were analyzed. The performance of the HSI, HU values and the FLI for diagnosing mild hepatic steatosis was evaluated by calculating the area under the receiver operating characteristic curve. Whether the degree of hepatic steatosis and intrahepatic inflammation could be predicted using the HSI, HU values or the FLI was also analyzed. Moreover, we validate the results using magnetic resonance imaging proton density fat fraction as an another reference standard.
### Results
The AUROC for diagnosing mild hepatic steatosis was 0.810 ($p \leq 0.001$) for the HSI, 0.732 ($p \leq 0.001$) for liver HU value, 0.802 ($p \leq 0.001$) for the difference between liver and spleen HU value (L-S HU value) and 0.813 ($p \leq 0.001$) for the FLI. Liver HU and L-S HU values were negatively correlated with the percentage of hepatic steatosis and NAFLD activity score (NAS) and significantly different between steatosis grades and between NAS grades. The L–S HU value was demonstrated the good performance for grading the severity of hepatic steatosis and the degree of intrahepatic inflammation.
### Conclusions
The HU values on CT are feasible for stratifying hepatic fat content and evaluating the degree of intrahepatic inflammation, and the HSI and the FLI demonstrated good performance with high sensitivity and specificity in diagnosing mild hepatic steatosis.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12876-023-02717-3.
## Introduction
Nonalcoholic fatty liver disease (NAFLD) is the most common liver disease worldwide, characterized by an excessive accumulation of intrahepatic fat associated with insulin resistance [1]. Globally, the prevalence of NAFLD diagnosed by imaging tests is approximately $25.24\%$ [2–4]. Patients diagnosed with nonalcoholic steatohepatitis (NASH), a progressive form of NAFLD which indicates hepatic inflammation and steatosis on histology, have an increased risk of progression of fibrosis, liver cirrhosis, and hepatocellular carcinoma [5–8].
Liver biopsy is the gold standard for diagnosing NAFLD. Based on the percentage of hepatocytes that contain fat vacuoles, steatosis is classified as normal or grade 0 if steatotic hepatocytes are < $5\%$; mild or grade 1 if $5\%$–$33\%$; moderate or grade 2 if $34\%$–$66\%$; and severe or grade 3 if > $66\%$ [9–11]. However, liver biopsy is an invasive technique with potentially fatal complications [12]. Therefore, noninvasive methods for diagnosing fatty liver are preferred in various clinical settings. In particular, ultrasound is the most commonly used method. However, it is operator-dependent and has suboptimal performance in diagnosing mild steatosis and grading the severity of hepatic steatosis [13–16]. Therefore, CT is often used as an initial evaluation for patients with elevated liver enzyme levels and suspected hepatic steatosis, especially patients with obesity with poor sonic window on ultrasound examination. In many studies, both the liver HU value and the difference between the HU value of the liver and spleen (L–S HU value) have proven to be useful tools for diagnosing fatty liver disease [17–20]. Moreover, some guidelines recommend that scores using serum biomarkers could provide an alternative mean for diagnosing NAFLD [21]. Considering the under-diagnosis of and the lack of adequate care for NAFLD in the primary care setting, introducing an effective but simple steatosis scoring system that can be easily used by primary care providers is necessary [22, 23]. Among various steatosis scores, the hepatic steatosis index (HSI) and the fatty liver index (FLI) are ones of the simplest and consist of easily obtainable information [24]. HSI consists of sex, the presence of type 2 diabetes mellitus (DM), body mass index (BMI), and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels, while fatty liver index consists of BMI, waist circumference, triglycerides (TG) level, and gamma-glutamyl transferase (GGT) level. However, studies which aimed to evaluate the usefulness of noninvasive methods, including CT, the HSI, and the FLI, in diagnosing mild hepatic steatosis, which is relatively common in clinical practice, are few. Additionally, whether the HU value, the HSI, or the FLI correlates with the histological severity of hepatic steatosis remains unclear.
Meanwhile, several guidelines have stated that these noninvasive methods have limitations in diagnosing steatohepatitis [2, 25]. The presence of steatohepatitis is known to be the most important factor in the progression of fibrosis, while the severity of fibrosis is the most important histologic marker associated with the incidence of liver-related complications and mortality in patients with NAFLD [26–29]. In addition, since improvement of intrahepatic inflammation is known to be associated with improvement of fibrosis, the improvement of intrahepatic inflammation has been used as surrogate endpoints in various clinical trials [30]. NAFLD activity score (NAS) is a widely used scoring system to evaluate the degree of steatohepatitis in patients with NAFLD [11]. NAS is based on histological findings and calculated by scoring the degree of steatosis, hepatocyte ballooning, and inflammation and summing these values. The usefulness of NAS has been confirmed in several studies and is recommended as a method for evaluating changes in liver histology in patients with NASH. However, since liver biopsy is invasive and carries the risk of complications, it has limitations in being used as a method for evaluating the improvement of steatohepatitis in a general clinical practice. While several guidelines have stated that noninvasive tests are not acceptable alternative to biopsy for the diagnosis of NASH, few studies have been conducted on whether noninvasive tests are useful for evaluating the severity of intrahepatic inflammation in patients with biopsy-proven NAFLD [2, 25]. If there is a noninvasive method that can evaluate the severity of intrahepatic inflammation in patients with biopsy-proven NAFLD, it will be useful for determining the effectiveness of treatment in clinical situations.
Therefore, we aimed to evaluate whether HU values, the HSI, and the FLI could be helpful in diagnosing mild hepatic steatosis, stratifying the severity of hepatic steatosis and predicting inflammatory activity in patients with low-grade hepatic steatosis.
## Study design and participants
Patients aged between 18 and 75 years who underwent histological examination of the liver at Chungnam National University Hospital between January 2008 and December 2022 were enrolled in this study. Patients who underwent CT within 3 months prior to liver biopsy were included, and their electronic medical records were retrospectively reviewed. We excluded patients who had steatosis in > $33\%$ of the hepatocytes on liver biopsy, therefore, only patients with mild or grade 1 hepatic steatosis were enrolled. We also enrolled patients with a magnetic resonance imaging proton density fat fraction (MRI-PDFF) of less than $33\%$ among patients who underwent MRI-PDFF and CT within 6 months of MRI-PDFF at Chungnam National University Hospital. Patients with other liver diseases, such as chronic viral hepatitis B, chronic viral hepatitis C, autoimmune hepatitis, and primary biliary cholangitis, and those with excessive alcohol consumption (≥ 30 g/d of alcohol consumption for men and ≥ 20 g/d for women) were excluded. Additionally, patients with liver cirrhosis and a history of hepatocellular carcinoma or other liver-related malignancies within 5 years were also excluded.
## Data collection, calculation of the HSI and the FLI, and measurement of the HU value
We reviewed the histological data of the enrolled patients and collected demographic, laboratory, CT imaging, and MRI-PDFF data by investigating electronic medical records. From the data collected, the HSI and the FLI were calculated. The HSI was calculated as ‘8 × (ALT/AST ratio) + BMI (+ 2, if female; + 2, if with DM)’ [31], while the FLI was calculated as '(e0.953 × ln(TG) + 0.139 × BMI + 0.718 × ln(GGT) + 0.053 × WC −15.745/ 1 + e0.953 × ln(TG) + 0.139 × BMI + 0.718 × ln(GGT) + 0.053 × WC −15.754) × 100'. We also used only pre-contrast CT images to measure the HU values of the liver and spleen. Specifically, the HU values of ten randomly selected parts of the liver and spleen were measured, and the averages of HU values were calculated. The area measured at each time was set to 2.5–3 cm2. We defined the average of HU values of the liver as the liver HU value. The L–S HU value was calculated by subtracting the average HU value of the spleen from the average HU value of the liver.
## Definition of hepatic steatosis, low-grade hepatic steatosis, mild steatosis grade, and NAS grade
Hepatic steatosis was defined as the accumulation of fat vacuoles in > $5\%$ of hepatocytes. In our study, low-grade hepatic steatosis was defined as the presence of steatosis in < $33\%$ of hepatocytes, and these patients were classified again according to the percentage of hepatic steatosis as follows: steatosis < $5\%$, the mild steatosis grade 0 (mild G0 or mG0) group; steatosis $5\%$–$19\%$, the mild steatosis grade 1 (mild G1 or mG1) group; and steatosis $20\%$–$33\%$, the mild steatosis grade 2 (mild G2 or mG2) group. Similarly, patients who underwent MRI-PDFF were classified according to MRI-PDFF as follows: MRI-PDFF < $5\%$, the mild steatosis grade 0 (mild G0 or mG0) group; MRI-PDFF $5\%$–$19\%$, the mild steatosis grade 1 (mild G1 or mG1) group; and MRI-PDFF $20\%$–$33\%$, the mild steatosis grade 2 (mild G2 or mG2) group. In our study, each NAS grade was defined as follows. NAS < 3, the grade 1 (G1); NAS 3–4, grade 2 (G2); and NAS > 4, grade 3 (G3).
## Statistical analyses
The Student’s t-test for continuous data and chi-squared test for categorical data were used to compare the baseline characteristics between patients with and without hepatic steatosis. We evaluated the performance of each method to diagnose hepatic steatosis by calculating the area under the receiver operating characteristic curve (AUROC). Correlations between variables were determined using Pearson correlation coefficient. Logistic regression analyses were performed to identify independent predictive factors of hepatic steatosis. All factors with a $p \leq 0.05$ in the univariate analysis were included in the multivariate analysis, with the exception of multivariate analysis to assess whether the HSI or the FLI are an independent predictive factor for hepatic steatosis. In that exceptional case, DM, BMI, and the AST and ALT levels were excluded for calculating the HSI and BMI, waist circumference, TG level, and GGT level were excluded for calculating the FLI due to potential multicollinearity. Student’s t-test was used to compare the HSI, liver HU value, L-S HU value and the FLI between the two steatosis grade groups or the two NAS grade groups. All statistical analyses were performed using SPSS (version 26.0; IBM Corp., Armonk, NY, USA).
## Results
Altogether, the data of 2,031 patients aged between 18 and 75 years, who underwent liver biopsy or hepatic resection at Chungnam National University Hospital between January 2008 and December 2022, were reviewed. Among them, 1,746 patients underwent CT within 3 months prior to liver biopsy or hepatic resection, and the rest 285 patients were excluded. And among these 1,746 patients, 1,604 patients who did not meet the enroll criteria were sequentially excluded. Finally, of the patients who underwent liver biopsy or liver resection, 142 patients were enrolled in our study. Of the 142 patients analyzed, 44 had hepatic steatosis ≥ $5\%$, and 98 patients had hepatic steatosis < $5\%$ or did not have clinically significant hepatic steatosis (Fig. 1).Fig. 1Flow chart showing enrollment of patients who underwent liver biopsy or liver resection
## Comparison of the baseline characteristics of patients with and without hepatic steatosis
The baseline characteristics of the patients are summarized in Table 1. The number of patients classified into the mild steatosis group was 98 (mG0), 33 (mG1), 11 (mG2). And the number of patients classified into the steatohepatitis group using NAFLD activity score (NAS) was 23 (G1), 29 (G2), 6 (G3). BMI and serum AST, ALT, triglyceride (TG), albumin levels, and waist circumference (WC) were significantly higher in patients with hepatic steatosis than in those without hepatic steatosis. The mean HSI value was higher in patients with hepatic steatosis (37.37, $95\%$ confidence interval [CI]: 35.60–39.31) than in those without hepatic steatosis (31.54, $95\%$ CI: 30.62–32.88) ($p \leq 0.001$), while mean liver HU value and mean L–S HU value were lower in patients with hepatic steatosis (liver HU: 46.56, $95\%$ CI: 45.85–51.75/L–S HU: -1.401, $95\%$ CI: -2.586–3.051) than in those without hepatic steatosis (liver HU: 54.38 $95\%$ CI: 53.74–56.63/L–S HU: 7.813, $95\%$ CI: 6.884–9.936) (both $p \leq 0.001$). The mean FLI value was higher in patients with hepatic steatosis (61.83, $95\%$ CI: 54.07–69.55) than in those without hepatic steatosis (33.06, $95\%$ CI: 27.55–38.89) ($p \leq 0.001$).Table 1A comparison of characteristics between participants with and without hepatic steatosisBaseline characteristicsVariablesSteatosis (–)($$n = 98$$)Steatosis (+)($$n = 44$$)p valueDemographic variables Age (years)54.53 ± 13.8544.93 ± 14.66 < 0.001 Gender (M/F)$\frac{34}{6422}$/22 Body mass index (kg/m2)23.05 ± 3.24326.26 ± 3.493 < 0.001Comorbidities Diabetes mellitus11 ($11.2\%$)10 ($22.7\%$) Hypertension11 ($11.2\%$)5 ($11.4\%$) Dyslipidemia13 ($13.3\%$)3 ($6.82\%$)Biochemical parameters Aspartate aminotransferase (IU/L)28.56 ± 19.7241.55 ± 33.440.004 Alanine aminotransferase (IU/L)26.47 ± 23.4053.02 ± 65.720.001 Triglycerides (mg/dL)103.6 ± 62.79202.4 ± 124.9 < 0.001 Total cholesterol (mg/dL)184.2 ± 46.02200.8 ± 47.920.068 Total bilirubin (mg/dL)0.970 ± 0.9850.932 ± 0.9730.830 Gamma-glutamyl transpeptidase (IU/L)119.9 ± 158.0111.8 ± 133.80.7720 Serum glucose (mg/dL)105.9 ± 36.69119.3 ± 45.890.064 Serum albumin (g/dL)3.812 ± 0.5804.161 ± 0.500 < 0.001 Platelet count (103/uL)247.9 ± 83.54259.9 ± 81.190.419 Waist circumference (cm)81.65 ± 9.24991.64 ± 9.737 < 0.001Liver histology Steatosis grade S0/S1/S2/S$\frac{398}{0}$/$\frac{0}{08}$/$\frac{36}{0}$/0 *Mild steatosis* grade mG0/mG1/mG$\frac{298}{0}$/$\frac{00}{33}$/11 METAVIR score F0/F1/F2/F3/F$\frac{476}{15}$/$\frac{6}{1}$/$\frac{028}{7}$/$\frac{6}{3}$/0 NAFLD activity score grade G1/G2/G$\frac{316}{0}$/$\frac{07}{29}$/6 *Hepatic steatosis* index (mean ± SD)31.54 ± 4.09037.37 ± 5.729 < 0.001 Liver HU (mean ± SD)54.38 ± 6.12546.56 ± 9.911 < 0.001 Liver HU-Spleen HU (mean ± SD)7.813 ± 6.198-1.401 ± 8.988 < 0.001 Fatty liver index (mean ± SD)33.06 ± 22.9761.83 ± 21.74 < 0.001
## Comparison of performance of the HSI, liver HU value, L–S HU value and the FLI for diagnosing mild hepatic steatosis
The HSI had the highest AUROC for diagnosing hepatic steatosis (AUROC 0.810), followed by L–S HU value (AUROC 0.802), liver HU value (AUROC 0.732) and the FLI (AUROC 0.813) (Fig. 2). The HSI, with a low cut-off value of 30 and a high cut-off value of 36, diagnosed hepatic steatosis with $87\%$ sensitivity and $74\%$ specificity. Additionally, the L–S HU value with a cut-off value of 3 diagnosed hepatic steatosis with $70\%$ sensitivity and $82\%$ specificity, while the liver HU value with a cut-off value of 47 diagnosed hepatic steatosis with $54\%$ sensitivity and $89\%$ specificity. The FLI, with a low cut-off value of 30 and a high cut-off value of 60, diagnosed hepatic steatosis with $85\%$ sensitivity and $77\%$ specificity. Fig. 2ROC curves and diagnostic performance of hepatic steatosis index, liver HU value, liver HU value-spleen HU value and fatty liver index for diagnosing mild hepatic steatosis
## Factors associated with hepatic steatosis
The univariate analysis revealed that age, BMI, serum AST, ALT, TG, albumin levels, WC, the HSI, liver HU value, L–S HU value and the FLI were associated with hepatic steatosis. In the multivariate analysis, HSI, L–S HU value and the FLI remained as independent diagnostic factors for hepatic steatosis (Tables 2, 3 and 4). In patients with hepatic steatosis, the liver HU value was negatively correlated with BMI, AST, ALT, TG and glucose level. The L–S HU value was also negatively correlated with BMI, AST, ALT, TG and glucose level in patients with hepatic steatosis (data not shown).Table 2Univariate and multivariate analyses using the *Hepatic steatosis* index for patients with and without hepatic steatosisUnivariate analysisMultivariate analysisVariablesOR ($95\%$ CI)p valueOR ($95\%$ CI)p valueAge0.96(0.93–0.98)0.0010.99(0.94–1.04)0.636Hypertension0.90(0.29–2.80)0.851Hyperlipidemia2.28(0.62–8.37)0.215HSI1.31(1.18–1.45) < 0.0011.32(1.09–1.61)0.005TG (mg/dL)1.01(1.01–1.02) < 0.0011.01(1.00–1.02)0.037TC (mg/dL)1.01(1.00–1.02)0.067TB (mg/dL)0.96(0.65–1.41)0.830GGT (U/L)1.00(1.00–1.00)0.781Glucose (mg/dL)1.01(1.00–1.02)0.076Albumin (g/dL)4.79(1.89–12.1)0.00114.3(1.79–113.3)0.012Platelets(103/μL)1.00(1.00–1.01)0.422Waist circumference (cm)1.12(1.07–1.17) < 0.0011.00(0.93–1.08)0.933Multivariate analysis adjusted model: Diabetes, BMI Body mass index, AST Aspartate aminotransferase, ALT Alanine aminotransferase, were excluded because they were correlated with HSICI Confidence interval, HU Hounsfield unitTable 3Univariate and multivariate analyses using the liver-spleen Hounsfield unit for patients with and without hepatic steatosisUnivariate analysisMultivariate analysisVariablesOR ($95\%$ CI)p valueOR ($95\%$ CI)p valueAge0.96(0.93–0.98)0.0010.99(0.94–1.04)0.712Diabetes0.44(0.17–1.12)0.084Hypertension0.90(0.29–2.80)0.851Hyperlipidemia2.28(0.62–8.37)0.215L-S HU0.84(0.78–0.90) < 0.0010.84(0.74–0.96)0.011BMI1.33(1.17–1.50) < 0.0011.07(0.77–1.47)0.690AST (U/L)1.02(1.01–1.04)0.0100.99(0.95–1.04)0.779ALT (U/L)1.02(1.01–1.04)0.0031.02(0.98–1.07)0.333TG (mg/dL)1.01(1.01–1.02) < 0.0011.01(1.00–1.02)0.041TC (mg/dL)1.01(1.00–1.02)0.067TB (mg/dL)0.96(0.65–1.41)0.830GGT (U/L)1.00(1.00–1.00)0.781Glucose (mg/dL)1.01(1.00–1.02)0.076Albumin (g/dL)4.79(1.89–12.1)0.00117.2(2.07–142.1)0.008Platelets (103/μL)1.00(1.00–1.01)0.422Waist circumference (cm)1.12(1.07–1.17) < 0.0011.01(0.91–1.13)0.844CI Confidence interval, HU Hounsfield unitTable 4Univariate and multivariate analyses using the Fatty liver index for patients with and without hepatic steatosisUnivariate analysisMultivariate analysisVariablesOR ($95\%$ CI)p valueOR ($95\%$ CI)p valueAge0.96(0.93–0.98)0.0010.96(0.91–1.02)0.209Diabetes0.44(0.17–1.12)0.084Hypertension0.90(0.29–2.80)0.851Hyperlipidemia2.28(0.62–8.37)0.215FLI1.05(1.03–1.08) < 0.0011.08(1.03–1.13)0.002AST (U/L)1.02(1.01–1.04)0.0101.00(0.95–1.05)0.911ALT (U/L)1.02(1.01–1.04)0.0031.02(0.98–1.07)0.334TC (mg/dL)1.01(1.00–1.02)0.067TB (mg/dL)0.96(0.65–1.41)0.830Glucose (mg/dL)1.01(1.00–1.02)0.076Albumin (g/dL)4.79(1.89–12.1)0.00197.5(6.80–1397.3)0.001Platelets (103/μL)1.00(1.00–1.01)0.422Waist circumference (cm)1.12(1.07–1.17) < 0.0010.96(0.87–1.06)0.447CI Confidence interval, HU Hounsfield unitMultivariate analysis adjusted model: BMI Body mass index, TG Triglycerides, GGT Gamma-glutamyl transpeptidase, were excluded because they were correlated with FLI
## Distribution and performance of the HSI, liver HU value, L–S HU value and FLI according to steatosis grade group
The percentage of hepatic steatosis was positively correlated with the HSI ($r = 0.5391$) ($p \leq 0.0001$) or the FLI ($r = 0.4512$) ($p \leq 0.0001$), and negatively correlated with liver HU value (r = − 0.3152) ($$p \leq 0.0001$$) or L–S HU value (r = − 0.4018) ($p \leq 0.0001$) (Fig. 3). The mean liver HU value for patients in mG0, mG1, and mG2 was 55.2 ($95\%$ CI: 55.15–55.25), 49.49 ($95\%$ CI: 49.38–49.60), and 46.93 ($95\%$ CI: 46.78–47.08), respectively. Liver HU value was significantly different between mG0 and mG1($p \leq 0.001$) or mG2 ($p \leq 0.001$) and between mG1 and mG2 ($$p \leq 0.04$$) (Fig. 4). Moreover, the mean L–S HU values for patients in mG0, mG1, and mG2 were 8.497 ($95\%$ CI: 8.449–8.545), 1.292 ($95\%$ CI: 1.190–1.394), and -3.024 ($95\%$ CI: -3.166–-2.883), respectively. The L–S HU value was also significantly different between mG0 and mG1($p \leq 0.001$) or mG2 ($p \leq 0.001$) and between mG1 and mG2 ($$p \leq 0.01$$). Although the HSI was also significantly different between mG0 and mG1 ($p \leq 0.001$) or mG2 ($p \leq 0.001$), the differences between mG1 and mG2 were not statistically significant ($$p \leq 0.47$$). The FLI was significantly different between mG0 and mG1 ($p \leq 0.001$) or mG2 ($$p \leq 0.016$$), the differences between mG1 and mG2 were not statistically significant ($$p \leq 0.43$$).Fig. 3The correlation between each index and the percentage of hepatic steatosis. A Scatter plots showing the positive correlation between the hepatic steatosis index and the percentage of hepatic steatosis (B) Scatter plots showing the negative correlation between liver HU value and the percentage of hepatic steatosis (C) Scatter plots showing the negative correlation between liver HU value-spleen HU value and the percentage of hepatic steatosis (D) Scatter plots showing the positive correlation between the fatty liver index and the percentage of hepatic steatosis (E) Scatter plots showing the positive correlation between the hepatic steatosis index and the NAFLD activity score (F) Scatter plots showing the negative correlation between liver HU value and the NAFLD activity score (G) Scatter plots showing the negative correlation between liver HU value-spleen HU value and the NAFLD activity score (H) Scatter plots showing the positive correlation between the fatty liver index the NAFLD activity scoreFig. 4The comparison of each index according to steatosis grade group and performance of each index in grading the severity of hepatic steatosis. A The comparison of hepatic steatosis index according to mild steatosis grade group. B The comparison of liver HU value according to mild steatosis grade group. C The comparison of liver HU value-spleen HU value according to mild steatosis grade group. D The comparison of fatty liver index according to mild steatosis grade group. Performance of hepatic steatosis index, liver HU value, liver HU value-spleen HU value and fatty liver index in grading the severity of hepatic steatosis was also shown. Mild G0 = Group consisting of patients with the percentage of hepatic steatosis < $5\%$; mild G1 = Group consisting of patients with the percentage of hepatic steatosis of ≥ $5\%$ and < $20\%$; mild G2 = Group consisting of patients with the percentage of hepatic steatosis ≥ $20\%$ and < $33\%$
## Performance of the HSI, liver HU value, L-S HU value and the FLI in grading the severity of hepatic steatosis
Figure 4 shows the AUROCs of the HSI, liver HU value, L–S HU value and the FLI for grading the severity of hepatic steatosis. The L–S HU value demonstrated the best performance in grading the severity of low-grade hepatic steatosis. The optimal cut-off L–S HU values were 3 HU for ≥ mG1, and -3 HU for ≥ mG2.
## Distribution and performance of the HSI, liver HU value, L–S HU value and the FLI according to NAS grade group
The NAS was positively correlated with the HSI ($r = 0.5074$) ($p \leq 0.0001$) or FLI ($r = 0.3556$) ($p \leq 0.0001$), and negatively correlated with liver HU value (r = − 0.4117) ($$p \leq 0.0013$$) or L–S HU value (r = − 0.4876) ($$p \leq 0.0001$$) (Fig. 3). The mean liver HU value for patients in G1, G2, and G3 was 52.49 ($95\%$ CI: 49.25–55.73), 47.13 ($95\%$ CI: 43.06–51.19), and 40.04 ($95\%$ CI: 32.56–47.51), respectively. Liver HU value was significantly different between G1 and G2 ($$p \leq 0.02$$) or G3 ($p \leq 0.001$) (Fig. 5). Moreover, the mean L–S HU values for patients in G1, G2, and G3 were 5.767 ($95\%$ CI: 2.632–8.902), -1.060 ($95\%$ CI: -4.714–2.593), and -7.35 ($95\%$ CI: -15.48–0.781), respectively. The L–S HU value was also significantly different between G1 and G2 ($$p \leq 0.003$$) or G3 ($p \leq 0.001$). The HSI was significantly different between G1 and G2($$p \leq 0.004$$) or G3 ($p \leq 0.001$), and between G2 and G3 ($p \leq 0.001$). The FLI was significantly different between G1 and G3 ($$p \leq 0.02$$).Fig. 5The comparison of each index according to NAFLD activity score group and performance of each index in grading the severity of steatohepatitis. A The comparison of hepatic steatosis index according to NAFLD activity score group. B The comparison of liver HU value according to NAFLD activity score group. C The comparison of liver HU value-spleen HU value according to NAFLD activity score group. D The comparison of fatty liver index according to NAFLD activity score group. Performance of hepatic steatosis index, liver HU value, liver HU value-spleen HU value and fatty liver index in grading the severity of steatohepatitis was also shown. G1 = Group consisting of patients with the NAFLD activity score < 3; G2 = Group consisting of patients with the NAFLD activity score 3–4; G3 = Group consisting of patients with the NAFLD activity score > 4
## Performance of the HSI, liver HU value, L-S HU value and the FLI in evaluating the degree of steatohepatitis
Figure 5 shows the AUROCs of the HSI, liver HU value, L–S HU value and the FLI for evaluating the degree of steatohepatitis. The L–S HU value, with a cut-off value of -3, predicted whether NAS was 3 or higher or not with $71\%$ sensitivity and $71\%$ specificity. And the L–S HU value, with a cut-off value of -1, predicted whether NAS was 5 or higher or not with $100\%$ sensitivity and $71\%$ specificity. Additionally, the HSI and the FLI had high AUROC for predicting NAS of 3 or more and NAS of 5 or more.
## Comparison of NAS between patients with and without metabolic syndrome
Among 2,031 patients aged between 18 and 75 years, who underwent liver biopsy or hepatic resection at Chungnam National University Hospital during study period, 285 patients who didn’t underwent CT within 3 months prior to liver biopsy or hepatic resection were excluded (Fig. S1). And among these 1,746 patients, 1,604 patients who did not meet the enroll criteria were sequentially excluded. Metabolic syndrome could not be evaluated in 21 patients of 142 patients due to missing variables. Among 77 patients without metabolic syndrome, hepatic steatosis was observed in 17 patients ($22.1\%$), and among 44 patients with metabolic syndrome, hepatic steatosis was observed in 21 patients ($47.8\%$) ($$p \leq 0.03$$). NAS was evaluated in 52 patients with steatotic hepatocytes on liver biopsy and there was no difference in NAS between the group with and without metabolic syndrome ($$p \leq 0.351$$).
## Distribution of the liver HU value and L–S HU value according to MRI-PDFF
During the study period, 152 patients underwent MRI-PDFF at Chungnam National University Hospital, and 88 of them underwent CT within 6 months. Of these 88 patients, 64 did not meet the enroll criteria. Therefore, of the patients who underwent MRI-PDFF, 22 patients were enrolled in our study. Of the 22 patients analyzed, 13 had MRI-PDFF ≥ $5\%$, and 9 patients had MRI-PDFF < $5\%$ (Fig. S2). The percentage of hepatic steatosis was positively correlated with the HSI ($r = 0.6794$) ($$p \leq 0.0007$$) or the FLI ($r = 0.6720$) ($$p \leq 0.0030$$), and negatively correlated with liver HU value (r = − 0.7638) ($p \leq 0.0001$) or L–S HU value (r = − 0.5781) ($$p \leq 0.0024$$) (Fig. S3). The mean liver HU value for patients in mG0, mG1, and mG2 was 53.57 ($95\%$ CI: 47.27–59.86), 42.77 ($95\%$ CI: 36.55–48.99), and 29.80 ($95\%$ CI: 26.05–33.55), respectively. Liver HU value was significantly different between mG0 and mG1 ($$p \leq 0.006$$) or mG2 ($p \leq 0.001$) and between mG1 and mG2 ($$p \leq 0.014$$) (Fig. S4). Moreover, the mean L–S HU values for patients in mG0, mG1, and mG2 were 5.47 ($95\%$ CI: -0.62–11.6), -2.66 ($95\%$ CI: -8.75–3.43), and -10.8 ($95\%$ CI: -51.4–29.7), respectively. The L–S HU value was also significantly different between mG0 and mG1 ($$p \leq 0.02$$) and between mG0 and mG2 ($$p \leq 0.02$$). Although the HSI was significantly different between mG2 and mG0 ($p \leq 0.001$) or mG1 ($p \leq 0.001$). The FLI was also significantly different between mG0 and mG1 ($$p \leq 0.03$$) and between mG0 and mG2 ($$p \leq 0.03$$).
## Discussion
In our study, the HU values on CT were useful in quantifying and stratifying liver fat contents in patients with low-grade hepatic steatosis, and the HSI and the FLI was demonstrated good performance with high sensitivity and specificity in diagnosing mild hepatic steatosis. In addition, the HU values were useful in evaluating the degree of intrahepatic inflammation in patients with low-grade hepatic steatosis. Transabdominal ultrasound, which is a commonly used diagnostic test for fatty liver disease in clinical field, has various limitations, such as a poor sonic window in patients with obesity and subjectivity according to the operator, resulting in low accuracy in diagnosing mild hepatic steatosis and evaluating the severity of hepatic steatosis. In these cases, it is possible to diagnose and evaluate fatty liver disease by using a serologic marker using blood test results or by imaging an abdominal CT scan. In this regard, our study is the first study to present the usefulness of the HSI and the FLI in company with HU value on CT to overcome the limitations of liver ultrasound for the diagnosis and severity assessment of mild fatty liver disease based on the results of histological evaluation of hepatic steatosis. In addition, we first demonstrated that the HU values on CT could be useful in evaluating the degree of steatohepatitis in patients who have already been diagnosed with NAFLD through histological examination.
In patients with NAFLD, lifestyle modification and pharmacological intervention can improve liver histology, and thereby prognosis. Therefore, an accurate diagnosis of fatty liver is important in patients with suspected NAFLD. Many guidelines recommend ultrasound as the first-line tool for diagnosing NAFLD [2, 21]. However, ultrasound exhibits suboptimal performance in diagnosing mild hepatic steatosis. For example, Ahn et al. evaluated hepatic steatosis in living liver donors without evidence of fatty liver on ultrasonography, and have reported a high prevalence of mild hepatic steatosis of $39.6\%$ in ultrasound-negative patients, suggesting that ultrasound cannot exclude mild hepatic steatosis [13]. Moreover, Tanaka et al. have reported that mild hepatic steatosis was diagnosed by biopsy in $28\%$ of patients with elevation of serum ALT levels and normal hepatic ultrasound image [14]. In our study, in contrast to low diagnostic accuracy of ultrasound for diagnosing mild hepatic steatosis reported in the literature, the HSI and the FLI demonstrated high performance with AUROC of 0.810 and AUROC of 0.813, respectively, in diagnosing mild hepatic steatosis. The HSI is a non-invasive and non-imaging screening tool devised based on the Korean health check-up data [31]. When low and high cut-off values of the HSI were used to discriminate the presence or absence of NAFLD in patients included in the validation set of the original paper, a sensitivity of $93.1\%$ and specificity of $93.1\%$ were achieved. The FLI was devised based on Italian study which enrolled 280 persons with normal liver and 216 persons with hepatic disease. When high cut-off value of the FLI was used to discriminate the presence of NAFLD in patients, a positive predictive value of $99\%$ and negative predictive value of $15\%$ were achieved. The performance of the HSI and the FLI in diagnosing NAFLD has been validated in various studies. Lee et al. evaluated the performance of several screening scores for diagnosing NAFLD in patients who underwent a health checkup, and the HSI indicated a high AUROC of 0.86 [32]. Murayama et al. also evaluated the performance of the HSI, Zhejiang university index, and fatty liver index using ultrasound-diagnosed fatty liver as a reference standard, and the HSI and the FLI demonstrated good predictive ability with AUROC of 0.874 and 0.884, respertively [33]. However, many previous studies validated the performance of the HSI in diagnosing NAFLD using ultrasound as a reference standard. As mentioned above, because diagnosing mild hepatic steatosis using ultrasound may be inaccurate, in our judgment, these studies have some limitations because they are not based on histological evaluation. Our study was conducted defining fatty liver histologically as the presence of steatosis in > $5\%$ of hepatocytes and in particular, we enrolled only patients with mild hepatic steatosis (steatosis in < $33\%$ of hepatocytes). Therefore, we confirm good performance of the HSI and the FLI in diagnosing mild hepatic steatosis more objectively and strictly in our study than in previous studies. Considering the high sensitivity and specificity of HSI and the FLI for diagnosing mild hepatic steatosis observed here, additional tests to exclude mild hepatic steatosis might be beneficial for suspected NAFLD in patients with negative US findings but have the HSI of ≥ 36 or the FLI of ≥ 60.
Indeed, the liver HU value showed a low sensitivity in diagnosing mild hepatic steatosis in our study. The low sensitivity of CT in diagnosing mild hepatic steatosis has been reported in several previous studies [34, 35]. In particular, since the liver HU value may be affected by the reconstruction algorithm or the vendor of the CT scanner, the L–S HU value using spleen as an internal control is more commonly used for diagnosing fatty liver disease [35]. Therefore, when compared with the liver HU value in our study, the L–S HU value demonstrated significantly higher AUROC value and sensitivity in diagnosing mild hepatic steatosis, suggesting that the L–S HU value has advantage over liver HU value for detecting mild hepatic steatosis.
In addition to the limited diagnostic accuracy of ultrasound in detecting mild hepatic steatosis, the suboptimal performance to evaluate the degree of fatty liver, which may be due to the qualitative and subjective nature that causes inter-observer variability, is another disadvantage of ultrasound. Strauss et al. have reported that the inter-observer agreement for grading the severity of fatty liver using ultrasound was 47.0–$63.7\%$ [15]. Qayyum et al. have also reported that the correlation of ultrasound score with histological hepatic steatosis was low due to low inter-observer agreement for ultrasound [16]. In contrast to ultrasound, CT scan is not dependent on the operator. Our study revealed that the L–S HU value on CT could be better than ultrasound for quantifying and stratifying liver fat content, based on the results of histological evaluation as well as MRI-PDFF. Although proton magnetic resonance spectroscopy (1H-MRS) was recommended in clinical trials and experimental studies for the quantitative estimation of hepatic steatosis, it was not recommended in common clinical settings because of its high cost [21]. Recently, chemical-shift-encoded MRI (CSE-MRI) method has shown promising results, but accessibility to MRI is limited in various clinical settings, including primary care. Moreover, CT is almost routinely used clinically in patients with poor sonic view due to obesity or anatomical characteristics, although it is accompanied by elevated liver enzyme levels. Therefore, CT could have advantage over MRI for quantifying liver fat content in various particular clinical conditions, such as routine surveillance or opportunistic detection. Kramer et al. evaluated the diagnostic accuracy of various imaging methods in the quantification of hepatic steatosis using 1H-MRS as the reference standard, and reported an excellent correlation between HU value and 1H-MRS [36]. Another study demonstrated that CT-based liver fat quantification exhibited good correlation with MRI-PDFF measured by CSE-MRI [37]. Our results support the results of previous studies, using histological hepatic steatosis, as well as MRI-PDFF, as a reference standard. Consistent with the results of previous studies, our results demonstrate that CT-based liver fat quantification validated as a result of histological examination is a useful alternative to MRI-based liver fat quantification. Compared with previous studies, our study enrolled only patients with low-grade hepatic steatosis. Evaluation of the severity of fatty liver in these low-grade hepatic steatosis patients is of great clinical importance in terms of metabolic diseases. Several studies have emphasized that the degree of hepatic steatosis should be classified, even though in patients with mild hepatic steatosis. Li et al. have reported that patients with liver fat content > $10\%$ had higher odds ratios of impaired glucose regulation than those with liver fat content < $10\%$ [38]. Ducluzeau et al. have also reported that hepatic fat fraction between 5 and $10\%$ confers the same risk of having the metabolic syndrome and that hepatic steatosis > $10\%$ is associated with a very high probability of having the metabolic syndrome [39]. Therefore, if the L-S HU value presented by our study is used, it is expected to be helpful in classifying the severity of low-grade steatosis patients and to suggest a differentiated treatment strategy to the patients. In the future, further studies will be needed on whether the liver fat content measured using CT could predict the prognosis in patients with low-grade hepatic steatosis.
Our study also demonstrated that the HU values were useful in evaluating the degree of intrahepatic inflammation in patients with biopsy-proven NAFLD. Hepatocyte injury and inflammation has been known to be the most important factors in the progression of fibrosis, and reducing intrahepatic inflammation has been known to be associated with improvement of fibrosis [26–29]. In particular, Brunt et al. found that an improvement in NAS of 2 points or more as well as resolution of NASH was most strongly associated with fibrosis improvement [30]. However, because biopsy is invasive, there is a limit to repeatedly performing liver biopsy to evaluating the improvement of NASH in patients who have been diagnosed with NASH through histological examination [40, 41]. Therefore, if there is a noninvasive method to evaluate the improvement of intrahepatic inflammation, it would be helpful to evaluate the effectiveness of treatment and to determine whether to continue current treatment or change to another treatment option. In this study, L-S HU values discriminated patients with NAS of 3 or higher with a sensitivity of $92\%$ and specificity of $90\%$, and patients with a NAS of 5 or higher with a sensitivity of $100\%$ and specificity of $82\%$ in patients with low-grade liver steatosis. Considering the high accuracy of the L-S HU value in evaluating the degree of intrahepatic inflammation observed in this study, we think that L-S HU value could be used in determining the improvement of intrahepatic inflammation in patients with biopsy-proven NAFLD.
This study had several limitations. First, the number of enrolled patients was relatively small, which may have reduced the generalizability of the results. Second, the performance of t HU values, the HSI, and the FLI for diagnosing mild hepatic steatosis or grading the severity of hepatic steatosis could not be directly compared with that of ultrasound. Multicenter, large cohort prospective studies are required to overcome these limitations.
In conclusion, the HU values are feasible for quantifying and stratifying hepatic fat content and for evaluating the degree of intrahepatic inflammation, and the HSI and the FLI are also useful tools for diagnosing mild hepatic steatosis.
## Supplementary Information
Additional file 1: Figure S1. Comparison of the proportion of hepatic steatosis and NAS between patients with and without metabolic syndrome. ( A) Flow chart showing enrollment of patients (B) Comparison of the proportion of patients with hepatic steatosis between patients with and without metabolic syndrome (c) Comparison of NAS between patients with and without metabolic syndrome. Additional file 2: Figure S2. Flow chart showing enrollment of patients who underwent MRI-PDFF.Additional file 3: Figure S3. The correlation between each index and MRI-PDFF. ( A) Scatter plots showing the positive correlation between the hepatic steatosis index and MRI-PDFF (B) Scatter plots showing the negative correlation between liver HU value and MRI-PDFF (C) Scatter plots showing the negative correlation between liver HU value-spleen HU value and MRI-PDFF (D) Scatter plots showing the negative correlation between fatty liver index and MRI-PDFF.Additional file 4: Figure S4. The comparison of each index according to steatosis grade group evaluated by MRI-PDFF and performance of each index in grading the severity of hepatic steatosis. ( A) The comparison of hepatic steatosis index according to mild steatosis grade group. ( B) The comparison of liver HU value according to mild steatosis grade group. ( C) The comparison of liver HU value-spleen HU value according to mild steatosis grade group. ( D) The comparison of fatty liver index according to mild steatosis grade group. Performance of hepatic steatosis index, liver HU value, liver HU value-spleen HU value and fatty liver index in grading the severity of hepatic steatosis was also shown. Mild G0 = Group consisting of patients with MRI-PDFF < $5\%$; mild G1 = Group consisting of patients with MRI-PDFF ≥ $5\%$ and < $20\%$; mild G2 = Group consisting of patients with MRI-PDFF ≥ $20\%$ and < $33\%$.
## References
1. Puri P, Sanyal AJ. **Nonalcoholic fatty liver disease: Definitions, risk factors, and workup**. *Clin Liver Dis (Hoboken)* (2012) **1** 99-103. DOI: 10.1002/cld.81
2. Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, Harrison SA, Brunt EM, Sanyal AJ. **The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the Study of Liver Diseases**. *Hepatology* (2018) **67** 328-357. DOI: 10.1002/hep.29367
3. Younossi ZM, Koenig AB, Abdelatif D, Fazel Y, Henry L, Wymer M. **Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes**. *Hepatology* (2016) **64** 73-84. DOI: 10.1002/hep.28431
4. Williams VF, Taubman SB, Stahlman S. **Non-alcoholic fatty liver disease (NAFLD), active component, U.S. Armed Forces, 2000–2017**. *Msmr* (2019) **26** 2-11. PMID: 31657583
5. Ekstedt M, Hagstrom H, Nasr P, Fredrikson M, Stal P, Kechagias S, Hultcrantz R. **Fibrosis stage is the strongest predictor for disease-specific mortality in NAFLD after up to 33 years of follow-up**. *Hepatology* (2015) **61** 1547-1554. DOI: 10.1002/hep.27368
6. Ekstedt M, Nasr P, Kechagias S. **Natural History of NAFLD/NASH**. *Curr Hepatol Rep* (2017) **16** 391-397. DOI: 10.1007/s11901-017-0378-2
7. Rinella ME. **Nonalcoholic fatty liver disease: a systematic review**. *JAMA* (2015) **313** 2263-2273. DOI: 10.1001/jama.2015.5370
8. Petrick JL, Thistle JE, Zeleniuch-Jacquotte A, Zhang X, Wactawski-Wende J, Van Dyke AL, Stampfer MJ, Sinha R, Sesso HD, Schairer C. **Body mass index, diabetes and intrahepatic cholangiocarcinoma risk: the liver cancer pooling project and meta-analysis**. *Am J Gastroenterol* (2018) **113** 1494. DOI: 10.1038/s41395-018-0207-4
9. Onyekwere CA, Ogbera AO, Samaila AA, Balogun BO, Abdulkareem FB. **Nonalcoholic fatty liver disease: Synopsis of current developments**. *Niger J Clin Pract* (2015) **18** 703-712. DOI: 10.4103/1119-3077.163288
10. Jung ES, Lee K, Yu E, Kang YK, Cho MY, Kim JM, Moon WS, Jeong JS, Park CK, Park JB. **Interobserver Agreement on Pathologic Features of Liver Biopsy Tissue in Patients with Nonalcoholic Fatty Liver Disease**. *J Pathol Transl Med* (2016) **50** 190-196. DOI: 10.4132/jptm.2016.03.01
11. Kleiner DE, Brunt EM, Van Natta M, Behling C, Contos MJ, Cummings OW, Ferrell LD, Liu YC, Torbenson MS, Unalp-Arida A. **Design and validation of a histological scoring system for nonalcoholic fatty liver disease**. *Hepatology* (2005) **41** 1313-1321. DOI: 10.1002/hep.20701
12. Sumida Y, Nakajima A, Itoh Y. **Limitations of liver biopsy and non-invasive diagnostic tests for the diagnosis of nonalcoholic fatty liver disease/nonalcoholic steatohepatitis**. *World J Gastroenterol* (2014) **20** 475-485. DOI: 10.3748/wjg.v20.i2.475
13. Ahn JS, Sinn DH, Gwak GY, Kim JM, Kwon CH, Joh JW, Paik YH, Choi MS, Lee JH, Koh KC. **Steatosis among living liver donors without evidence of fatty liver on ultrasonography: potential implications for preoperative liver biopsy**. *Transplantation* (2013) **95** 1404-1409. DOI: 10.1097/TP.0b013e31828d1588
14. Tanaka N, Tanaka E, Sheena Y, Komatsu M, Okiyama W, Misawa N, Muto H, Umemura T, Ichijo T, Matsumoto A. **Useful parameters for distinguishing nonalcoholic steatohepatitis with mild steatosis from cryptogenic chronic hepatitis in the Japanese population**. *Liver Int* (2006) **26** 956-963. DOI: 10.1111/j.1478-3231.2006.01338.x
15. Strauss S, Gavish E, Gottlieb P, Katsnelson L. **Interobserver and intraobserver variability in the sonographic assessment of fatty liver**. *AJR Am J Roentgenol* (2007) **189** W320-323. DOI: 10.2214/AJR.07.2123
16. Qayyum A, Chen DM, Breiman RS, Westphalen AC, Yeh BM, Jones KD, Lu Y, Coakley FV, Callen PW. **Evaluation of diffuse liver steatosis by ultrasound, computed tomography, and magnetic resonance imaging: which modality is best?**. *Clin Imaging* (2009) **33** 110-115. DOI: 10.1016/j.clinimag.2008.06.036
17. Esterson YB, Grimaldi GM. **Radiologic Imaging in Nonalcoholic Fatty Liver Disease and Nonalcoholic Steatohepatitis**. *Clin Liver Dis* (2018) **22** 93-108. DOI: 10.1016/j.cld.2017.08.005
18. Chartampilas E. **Imaging of nonalcoholic fatty liver disease and its clinical utility**. *Hormones (Athens)* (2018) **17** 69-81. DOI: 10.1007/s42000-018-0012-x
19. Byun J, Lee SS, Sung YS, Shin Y, Yun J, Kim HS. **Yu Es, Lee S-G, Lee M-g: CT indices for the diagnosis of hepatic steatosis using non-enhanced CT images: development and validation of diagnostic cut-off values in a large cohort with pathological reference standard**. *Eur Radiol* (2018) **29** 4427-4435. DOI: 10.1007/s00330-018-5905-1
20. Iwasaki M, Takada Y, Hayashi M, Minamiguchi S, Haga H, Maetani Y, Fujii K, Kiuchi T, Tanaka K. **Noninvasive evaluation of graft steatosis in living donor liver transplantation**. *Transplantation* (2004) **78** 1501-1505. DOI: 10.1097/01.TP.0000140499.23683.0D
21. **EASL-EASD-EASO Clinical Practice Guidelines for the management of non-alcoholic fatty liver disease**. *J Hepatol* (2016) **64** 1388-1402. DOI: 10.1016/j.jhep.2015.11.004
22. Polanco-Briceno S, Glass D, Stuntz M, Caze A. **Awareness of nonalcoholic steatohepatitis and associated practice patterns of primary care physicians and specialists**. *BMC Res Notes* (2016) **9** 157. DOI: 10.1186/s13104-016-1946-1
23. Kallman JB, Arsalla A, Park V, Dhungel S, Bhatia P, Haddad D, Wheeler A, Younossi ZM. **Screening for hepatitis B, C and non-alcoholic fatty liver disease: a survey of community-based physicians**. *Aliment Pharmacol Ther* (2009) **29** 1019-1024. DOI: 10.1111/j.1365-2036.2009.03961.x
24. Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, Castiglione A, Tiribelli C. **The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population**. *BMC Gastroenterol* (2006) **6** 1-7. DOI: 10.1186/1471-230X-6-33
25. Kang SH, Lee HW, Yoo J-J, Cho Y, Kim SU, Lee TH, Jang BK, Kim SG, Ahn SB, Kim H. **KASL clinical practice guidelines: Management of nonalcoholic fatty liver disease**. *Clin Mol Hepatol* (2021) **27** 363. DOI: 10.3350/cmh.2021.0178
26. Angulo P, Kleiner DE, Dam-Larsen S, Adams LA, Bjornsson ES, Charatcharoenwitthaya P, Mills PR, Keach JC, Lafferty HD, Stahler A. **Liver fibrosis, but no other histologic features, is associated with long-term outcomes of patients with nonalcoholic fatty liver disease**. *Gastroenterology* (2015) **149** 389-397. e310. DOI: 10.1053/j.gastro.2015.04.043
27. Hagström H, Nasr P, Ekstedt M, Hammar U, Stål P, Hultcrantz R, Kechagias S. **Fibrosis stage but not NASH predicts mortality and time to development of severe liver disease in biopsy-proven NAFLD**. *J Hepatol* (2017) **67** 1265-1273. DOI: 10.1016/j.jhep.2017.07.027
28. Dulai PS, Singh S, Patel J, Soni M, Prokop LJ, Younossi Z, Sebastiani G, Ekstedt M, Hagstrom H, Nasr P. **Increased risk of mortality by fibrosis stage in nonalcoholic fatty liver disease: systematic review and meta-analysis**. *Hepatology* (2017) **65** 1557-1565. DOI: 10.1002/hep.29085
29. Taylor RS, Taylor RJ, Bayliss S, Hagström H, Nasr P, Schattenberg JM, Ishigami M, Toyoda H. **Wong VW-S, Peleg N: Association between fibrosis stage and outcomes of patients with nonalcoholic fatty liver disease: a systematic review and meta-analysis**. *Gastroenterology* (2020) **158** 1611-1625. e1612. DOI: 10.1053/j.gastro.2020.01.043
30. Brunt EM, Kleiner DE, Wilson LA, Sanyal AJ, Neuschwander-Tetri BA, Network NSCR. **Improvements in histologic features and diagnosis associated with improvement in fibrosis in nonalcoholic steatohepatitis: results from the nonalcoholic steatohepatitis clinical research network treatment trials**. *Hepatology* (2019) **70** 522-531. DOI: 10.1002/hep.30418
31. Lee YH, Bang H, Park YM, Bae JC, Lee BW, Kang ES, Cha BS, Lee HC, Balkau B, Lee WY. **Non-laboratory-based self-assessment screening score for non-alcoholic fatty liver disease: development, validation and comparison with other scores**. *PLoS ONE* (2014) **9** e107584. DOI: 10.1371/journal.pone.0107584
32. Lee JH, Kim D, Kim HJ, Lee CH, Yang JI, Kim W, Kim YJ, Yoon JH, Cho SH, Sung MW. **Hepatic steatosis index: a simple screening tool reflecting nonalcoholic fatty liver disease**. *Dig Liver Dis* (2010) **42** 503-508. DOI: 10.1016/j.dld.2009.08.002
33. Murayama K, Okada M, Tanaka K, Inadomi C, Yoshioka W, Kubotsu Y, Yada T, Isoda H, Kuwashiro T, Oeda S. **Prediction of Nonalcoholic Fatty Liver Disease Using Noninvasive and Non-Imaging Procedures in Japanese Health Checkup Examinees**. *Diagnostics (Basel)* (2021) **11** 132. DOI: 10.3390/diagnostics11010132
34. Lee SS, Park SH. **Radiologic evaluation of nonalcoholic fatty liver disease**. *World J Gastroenterol* (2014) **20** 7392-7402. DOI: 10.3748/wjg.v20.i23.7392
35. Lee DH. **Imaging evaluation of non-alcoholic fatty liver disease: focused on quantification**. *Clin Mol Hepatol* (2017) **23** 290-301. DOI: 10.3350/cmh.2017.0042
36. Kramer H, Pickhardt PJ, Kliewer MA, Hernando D, Chen GH, Zagzebski JA, Reeder SB. **Accuracy of Liver Fat Quantification With Advanced CT, MRI, and Ultrasound Techniques: Prospective Comparison With MR Spectroscopy**. *AJR Am J Roentgenol* (2017) **208** 92-100. DOI: 10.2214/AJR.16.16565
37. Guo Z, Blake GM, Li K, Liang W, Zhang W, Zhang Y, Xu L, Wang L, Brown JK, Cheng X. **Liver Fat Content Measurement with Quantitative CT Validated against MRI Proton Density Fat Fraction: A Prospective Study of 400 Healthy Volunteers**. *Radiology* (2020) **294** 89-97. DOI: 10.1148/radiol.2019190467
38. Li X, Xia M, Ma H, Hu Y, Yan H, He W, Lin H, Zhao N, Gao J, Gao X. **Liver fat content, evaluated through semi-quantitative ultrasound measurement, is associated with impaired glucose profiles: a community-based study in Chinese**. *PLoS ONE* (2013) **8** e65210. DOI: 10.1371/journal.pone.0065210
39. Ducluzeau PH, Boursier J, Bertrais S, Dubois S, Gauthier A, Rohmer V, Gagnadoux F, Leftheriotis G, Cales P, Andriantsitohaina R. **MRI measurement of liver fat content predicts the metabolic syndrome**. *Diabetes Metab* (2013) **39** 314-321. DOI: 10.1016/j.diabet.2013.01.007
40. Angulo P, Hui JM, Marchesini G, Bugianesi E, George J, Farrell GC, Enders F, Saksena S, Burt AD, Bida JP. **The NAFLD fibrosis score: a noninvasive system that identifies liver fibrosis in patients with NAFLD**. *Hepatology* (2007) **45** 846-854. DOI: 10.1002/hep.21496
41. Chan W-K, Treeprasertsuk S, Goh GB-B, Fan J-G, Song MJ, Charatcharoenwitthaya P, Duseja A, Dan Y-Y, Imajo K, Nakajima A. **Optimizing use of nonalcoholic fatty liver disease fibrosis score, fibrosis-4 score, and liver stiffness measurement to identify patients with advanced fibrosis**. *Clin Gastroenterol Hepatol* (2019) **17** 2570-25802537. DOI: 10.1016/j.cgh.2019.03.006
|
---
title: 'Different clinical features in Malawian outpatients presenting with COVID-19
prior to and during Omicron variant dominance: A prospective observational study'
authors:
- Marah G. Chibwana
- Herbert W. Thole
- Cat Anscombe
- Philip M. Ashton
- Edward Green
- Kayla G. Barnes
- Jen Cornick
- Ann Turner
- Desiree Witte
- Sharon Nthala
- Chikondi Thom
- Felistas Kanyandula
- Anna Ainani
- Natasha Mtike
- Hope Tambala
- Veronica N’goma
- Dorah Mwafulirwa
- Erick Asima
- Ben Morton
- Markus Gmeiner
- Zaziwe Gundah
- Gift Kawalazira
- Neil French
- Nicholas Feasey
- Robert S. Heyderman
- Todd D. Swarthout
- Kondwani C. Jambo
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022204
doi: 10.1371/journal.pgph.0001575
license: CC BY 4.0
---
# Different clinical features in Malawian outpatients presenting with COVID-19 prior to and during Omicron variant dominance: A prospective observational study
## Abstract
The SARS-CoV-2 Omicron variant has resulted in a high number of cases, but a relatively low incidence of severe disease and deaths, compared to the pre-Omicron variants. Therefore, we assessed the differences in symptom prevalence between Omicron and pre-Omicron infections in a sub-Saharan African population. We collected data from outpatients presenting at two primary healthcare facilities in Blantyre, Malawi, from November 2020 to March 2022. Eligible participants were aged >1month old, with signs suggestive of COVID-19, and those not suspected of COVID-19, from whom we collected nasopharyngeal swabs for SARS-CoV-2 PCR testing, and sequenced positive samples to identify infecting-variants. In addition, we calculated the risk of presenting with a given symptom in individuals testing SARS-CoV-2 PCR positive before and during the Omicron variant-dominated period. Among 5176 participants, $6.4\%$ were under 5, and $77\%$ were aged 18 to 50 years. SARS-CoV-2 infection prevalence peaked in January 2021 (Beta), July 2021 (Delta), and December 2021 (Omicron). We found that cough (risk ratio (RR), 1.50; $95\%$ confidence interval (CI), 1.00 to 2.30), fatigue (RR 2.27; $95\%$ CI, 1.29 to 3.86) and headache (RR 1.64; $95\%$ CI, 1.15 to 2.34) were associated with a high risk of SARS-CoV-2 infection during the pre-Omicron period. In comparison, only headache (RR 1.41; $95\%$ CI, 1.07 to 1.86) did associate with a high risk of SARS-CoV-2 infection during the Omicron-dominated period. In conclusion, clinical symptoms associated with Omicron infection differed from prior variants and were harder to identify clinically with current symptom guidelines. Our findings encourage regular review of case definitions and testing policies to ensure case ascertainment.
## Introduction
As of June 2022, the COVID-19 pandemic has resulted in 543 million cases and 6.3 million deaths globally [1]. In sub-Saharan Africa, the pandemic has however been associated with a lower rate of hospitalisation and deaths than in Europe and the Americas [1], despite widespread SARS-CoV-2 community transmission [2], and low COVID-19 vaccine coverage [3]. Due to the continued emergence of SARS-CoV-2 variants of concern, surveillance is essential for monitoring the pandemic and informing public health interventions, however the optimal approach to surveillance in low-income, resource-poor settings is unclear [4].
By June 2022, Malawi had experienced four epidemic waves peaking in July 2020, January 2021, July 2021, and December 2021. There were 86,348 confirmed SARS-CoV-2 cases nationally, with 2,645 COVID-19-associated deaths [1]. However, due to the availability of testing there is considerable case under ascertainment, as evidence by the high seroprevalence of >$65\%$ observed in Malawi as of July 2021 [5]. Blantyre has had the highest number of reported COVID-19 cases in Malawi, with $28.6\%$ of the national cases [6]. Recently, the Omicron variant has resulted in less hospitalisations and mortality in Malawi compared to the Delta variant [1], which has coincided with high seroprevalence of SARS-CoV-2 antibodies in Malawi and across sub-Saharan Africa [5, 7]. Further, Malawi introduced COVID-19 vaccines in March 2021 [8]. The COVID-19 vaccination coverage for *Malawi is* $4.5\%$, including the AstraZeneca, Janssen and Pfizer vaccines, with AstraZeneca vaccine constituting most of the doses [8].
Using data from early in the pandemic, a standardised case definition for COVID-19 was developed by the World Health Organisation (WHO) [9] and United States Centre for Disease Control (US CDC) [10], and these have allowed targeted SARS-CoV-2 testing. However, in sub-Saharan Africa, there is a high burden of other febrile illnesses such as malaria, pneumonia, TB and salmonellosis that have clinical features that overlap with COVID-19 [11, 12]. Further, there is limited data on the differences in clinical presentation between infections caused by different variants of concern (VOC), especially amongst non-hospitalised patients. To address these gaps, our study measured the prevalence of PCR-confirmed SARS-CoV-2 infection among outpatients presenting with medical conditions at primary healthcare facilities and compared the symptom profiles between Omicron and pre-Omicron infections.
## Study design and population
From November 2020 to March 2022, we conducted a SARS-CoV-2 prevalence study in primary healthcare facilities in the city of Blantyre, southern Malawi. Blantyre is Malawi’s commercial city with a population of 800,264 (pop density, 3334/km2). Adults and children were recruited voluntarily from two government-owned primary healthcare facilities, Ndirande Health Centre and Limbe Health Centre, both overseen by the Blantyre District Health Office. Census data shows Ndirande HC serves a catchment area of 135,736, while Limbe HC serves a catchment area of 145,604, but the actual catchment is likely much higher.
From November 2020 to July 2021, individuals with medical conditions were screened at the facility’s outpatient services department. Following assessment by a clinician and a review of the individual’s health passport (patient retained medical record), a nasopharyngeal swab was collected for SARS-CoV-2 screening by SARS-CoV-2 by RT-PCR from patients suspected of COVID-19 according to the WHO case definition [9]. From August 2021, following a protocol amendment to include capturing a more detailed clinical history using the ISARIC symptom list [13–15], participants included both those suspected and those not suspected of COVID-19 according to the WHO case definition, with a 2:1 numerical bias towards those with suspected COVID-19. Patients with suspected COVID-19 were recruited as and when they presented to the facility, while those with not clinically suspected of COVID-19 were selected by approaching every third patient in health facility’s triage area.
## Data and specimen collection
Following informed consent and assent (for children) from August 2021, we used an abridged International Severe Acute Respiratory and Emerging Infection Consortium (ISARIC) Clinical Characterisation Protocol (CCP) electronic case report form (eCRF) [13–15] to collect demographic and clinical data from all participants. Study nurses collected nasopharyngeal swabs in Universal Transport Medium (UTM) (Copan, Brescia, Italy) from all participants. Samples were initially stored and transported to the Malawi-Liverpool-Wellcome Programme (MLW) laboratory on ice and processed within 48 hours.
## Laboratory testing
Nasopharyngeal swabs were tested for SARS-CoV-2 RNA using the CDC 2019-nCoV RNA RT-PCR diagnostic panel (Integrated DNA Technologies, Iowa, USA). A cycle threshold (Ct) value of <40 was considered positive for SARS-CoV-2 using QuantStudio Real-Time PCR software v1.3 (Applied Biosystems, UK). Ribonuclease protein was used as an internal control to identify presence of human RNA. A negative extraction control and a PCR no-template control were also performed with every test. The results of patients with positive PCR tests were shared with the Blantyre District Health Office for further follow up and patient management.
## Genomic sequencing and analysis
Samples were extracted using the Qiasymphony-DSP mini kit 200 (Qiagen, UK) with offboard lysis. Samples were then tested using the CDC N1 assay to confirm the Ct values before sequencing. Samples with a Ct value <27 were sequenced. The following sequencing protocols were used; ARTICv2 and v3 was used from November 2020 from July 2021 to July 2021 [16] and UNZA [17] from August 2021 on wards. Initially two primer pools were used, however a third pool was made for primer pairs that commonly had lower depth compared to the average [15]. PCR cycling conditions were adapted to the new sequencing primers, with annealing temperature changed to 60°C. Sequencing was carried out with the Oxford Nanopore Technologies MinION sequencer. Samples that had poor coverage (<$70\%$) with the ARTIC primer set were repeated with the UNZA primer set.
For analysis of sequencing data, the lineage of each consensus genome was identified using pangolin with the following versions; pangolin v3.1.17, pangolearn 2021-12-06, constellations v0.1.1, scorpio v0.3.16, pango-designation used by pangoLEARN/Usher v1.2.105, pango-designation aliases v1.2.122 [18]. Samples were re-analysed when the Pangolin database was updated. The run was repeated if there was contamination in the negative control.
## Statistical analysis
We performed statistical analyses and graphical presentation using R statistical package, version 4.1.0. Categorical variables were summarized using frequency distributions and compared using Pearson’s Chi-squared test and Fisher’s exact test. The continuous variables were presented as median with interquartile range.
We employed multivariable logistic regression models, as implemented in the R package stats (version 3.6.2), to investigate odds of presenting with particular symptoms in Omicron compared pre-Omicron phases, adjusting for age and sex. A multivariable logistic regression model adjusting for age, sex, vaccination status, vaccination doses and days since last vaccine dose was also employed to investigate the impact of COVID-19 vaccination on PCR-confirmed SARS-CoV-2. P-values <0.05 were considered significant.
## Ethics approval
The study was approved by the College of Medicine Research and Ethics Committee (COMREC P$\frac{.08}{20}$/3099) and Liverpool School of Tropical Medicine Research Ethics Committee (LSTMREC 21–058). Written informed consent was obtained from the parent/guardian of each participant under 18 years of age, and from individual adult participants.
## Demographic and clinical characteristics of participants
From November 2020 through March 2022, 6147 (Ndirande Health Centre, $$n = 2899$$; Limbe Health Centre, $$n = 3248$$) individuals were approached, and 5188 participants were enrolled but 12 were excluded as they did not have sex recorded. Refusals were mostly from parents or guardians who did not consent for their children to undergo nasopharyngeal swabbing, and this did not change throughout the study period. Overall, the participants’ median age was 28 years (IQR 21–38), of which $6.4\%$ ($\frac{331}{5176}$) were under 5 years, $9.2\%$ ($\frac{331}{5176}$) were 6 to 17 years ($\frac{479}{5176}$), $77\%$ ($\frac{4000}{5176}$) were 18 to 50 years, and $7.1\%$ ($\frac{368}{5176}$) were above 50 years old. Of the total $50\%$ ($\frac{2596}{5176}$) were female (Table 1).
**Table 1**
| Characteristic | N = 5,1761 |
| --- | --- |
| Sex | |
| Female | 2,596 (50%) |
| Male | 2,580 (50%) |
| Age | 28 (21, 38) |
| Age group | |
| 0–5 years | 331 (6.4%) |
| 6–12 years | 228 (4.4%) |
| 13–17 years | 249 (4.8%) |
| 18–50 years | 4,000 (77%) |
| 51+ years | 368 (7.1%) |
| Sars-coV-2 PCR positivity | |
| Negative | 3,992 (77%) |
| Positive | 1,184 (23%) |
## Prevalence of SARS-CoV-2 infection and genomic surveillance
The overall prevalence of PCR-confirmed SARS-CoV-2 infection was $23\%$ ($\frac{1187}{5176}$) (Table 1). SARS-CoV-2 prevalence varied over time, with three distinct peaks over the study period, namely January 2021, July 2021, and December 2021 (Fig 1A). SARS-CoV-2 prevalence was lowest in those under 5 years of age ($5.74\%$ [CI 3.49–8.82]) compared to all other age groups (6-12yrs, $16.7\%$ [12.1–22.2]; 13-17yrs, $25.3\%$ [20.0–31.2]; 18-50yrs, $24.5\%$ [23.2–25.9]; 50+yrs, $22.8\%$ [18.6–27.5]) (Fig 1B). The three prevalence peaks corresponded with emergence of variants of concern (VOC), including Beta (B.1.351; January 2021), Delta (1.617.2; July 2021) and Omicron (BA$\frac{.1}{2}$; December 2021) (Fig 1C). Only Omicron (BA.1) was detectable in all age groups (Fig 1D).
**Fig 1:** *Prevalence of SARS-CoV-2 infections.A) Prevalence of SARS-CoV-2 infections across age groups. B) Prevalence of SARS-CoV-2 infections over time. Grey area represents confidence intervals. C) SARS-CoV-2 variants of concern across the three pandemic waves. D) SARS-CoV-2 variants of concern across age groups. (n = 402).*
## Symptoms associated with SARS-CoV-2 infection, pre- and during the Omicron-dominated phase
Forty-nine percent ($\frac{2520}{5176}$) of the total participants were recruited from August 2021 to March 2022, hence had detailed symptom and medical history (S1 Fig). Out of these 2509 had complete symptomology data and were used in the subsequent analysis. Using a multivariable analysis, cough ($71\%$ vs $68\%$, risk ratio, 1.50; $95\%$ CI, 1.00 to 2.30, $$p \leq 0.056$$), fatigue ($14\%$ vs $6.3\%$, risk ratio, 2.27; $95\%$ CI, 1.29 to 3.86, $$p \leq 0.003$$) and headache ($49\%$ vs $37\%$, risk ratio, 1.64; $95\%$ CI, 1.15 to 2.34, $$p \leq 0.007$$) were associated with a high risk of PCR-confirmed SARS-CoV-2 infection during the pre-Omicron period (Table 2 and S1 Table). While, during the Omicron-dominated period, only headache ($39\%$ vs $30\%$, risk ratio, 1.41; $95\%$ CI, 1.07 to 1.86, $$p \leq 0.015$$) was associated with a high risk of PCR-confirmed SARS-CoV-2 infection.
**Table 2**
| Characteristic | Univariable (Pre-Omicron) | Univariable (Pre-Omicron).1 | Univariable (Pre-Omicron).2 | Univariable (Pre-Omicron).3 | Multivariable (Pre-Omicron) | Multivariable (Pre-Omicron).1 | Multivariable (Pre-Omicron).2 | Univariable (Omicron) | Univariable (Omicron).1 | Univariable (Omicron).2 | Univariable (Omicron).3 | Multivariable (Omicron) | Multivariable (Omicron).1 | Multivariable (Omicron).2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Characteristic | N | OR1 | 95% CI1 | p-value | OR1 | 95% CI1 | p-value | N | OR1 | 95% CI1 | p-value | OR1 | 95% CI1 | p-value |
| Fever | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 0.76 | 0.46, 1.21 | 0.3 | 0.65 | 0.39, 1.05 | 0.093 | | 1.30 | 0.94, 1.78 | 0.11 | 1.23 | 0.88, 1.71 | 0.2 |
| Cough | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.15 | 0.80, 1.68 | 0.5 | 1.50 | 1.00, 2.30 | 0.056 | | 0.97 | 0.75, 1.25 | 0.8 | 1.04 | 0.79, 1.38 | 0.8 |
| Fatigue | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 2.34 | 1.37, 3.87 | 0.001 | 2.27 | 1.29, 3.86 | 0.003 | | 1.40 | 0.65, 2.89 | 0.4 | 1.24 | 0.55, 2.64 | 0.6 |
| Loss of smell | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.29 | 0.59, 2.55 | 0.5 | 0.89 | 0.34, 2.16 | 0.8 | | 1.51 | 0.63, 3.38 | 0.3 | 0.70 | 0.22, 2.06 | 0.5 |
| Loss of taste | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.80 | 0.99, 3.11 | 0.044 | 1.79 | 0.83, 3.65 | 0.12 | | 2.07 | 1.05, 4.01 | 0.031 | 2.04 | 0.81, 5.13 | 0.12 |
| Sore throat | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.14 | 0.67, 1.85 | 0.6 | 1.03 | 0.59, 1.72 | >0.9 | | 1.36 | 0.95, 1.93 | 0.091 | 1.35 | 0.93, 1.93 | 0.11 |
| Headache | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.63 | 1.16, 2.28 | 0.005 | 1.64 | 1.15, 2.34 | 0.007 | | 1.49 | 1.14, 1.95 | 0.003 | 1.41 | 1.07, 1.86 | 0.015 |
| Muscle ache | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.25 | 0.83, 1.85 | 0.3 | 1.15 | 0.72, 1.80 | 0.6 | | 1.33 | 1.01, 1.74 | 0.044 | 1.28 | 0.94, 1.73 | 0.12 |
| Joint pain | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.39 | 0.77, 2.38 | 0.2 | 1.07 | 0.54, 2.01 | 0.8 | | 1.44 | 0.87, 2.34 | 0.15 | 1.00 | 0.57, 1.73 | >0.9 |
| Abdominal pain | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.24 | 0.75, 1.98 | 0.4 | 1.33 | 0.76, 2.23 | 0.3 | | 1.10 | 0.69, 1.72 | 0.7 | 1.29 | 0.79, 2.08 | 0.3 |
| Diarrhoea | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 0.80 | 0.39, 1.51 | 0.5 | 0.89 | 0.41, 1.78 | 0.8 | | 0.79 | 0.45, 1.31 | 0.4 | 0.87 | 0.49, 1.49 | 0.6 |
| Shortness of breath | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 0.58 | 0.25, 1.14 | 0.15 | 0.55 | 0.23, 1.15 | 0.14 | | 0.79 | 0.46, 1.30 | 0.4 | 0.80 | 0.41, 1.52 | 0.5 |
| Chest pain | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 0.92 | 0.58, 1.41 | 0.7 | 0.83 | 0.51, 1.32 | 0.5 | | 1.21 | 0.83, 1.74 | 0.3 | 1.32 | 0.90, 1.93 | 0.2 |
| Runny nose | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 0.84 | 0.58, 1.21 | 0.4 | 0.78 | 0.52, 1.16 | 0.2 | | 0.96 | 0.70, 1.31 | 0.8 | 0.96 | 0.69, 1.32 | 0.8 |
| Pneumonia | 1311 | | | | | | | 1198 | | | | | | |
| No | | Ref. | — | | Ref. | — | | | Ref. | — | | Ref. | — | |
| Yes | | 1.08 | 0.51, 2.05 | 0.8 | 1.51 | 0.67, 3.13 | 0.3 | | 0.68 | 0.37, 1.19 | 0.2 | 0.78 | 0.37, 1.58 | 0.5 |
## Impact of COVID-19 vaccination on the risk of PCR-confirmed SARS-CoV-2 infection
Eighty percent ($\frac{2009}{2520}$) of the participants with detailed symptomology were eligible for vaccination (18 years and above) and had complete vaccination history. We, therefore, used these individuals to determine whether the risk of PCR-confirmed SARS-CoV-2 infection was different between vaccinated compared to unvaccinated adults. Using a multivariable analysis, adjusting for days since last vaccine dose, sex, age and recruitment period; COVID-19 vaccination was not associated with a reduced risk of PCR-confirmed SARS-CoV-2 infection (1 dose, OR 1.10[CI 0.39–2.66]; 2 doses, OR 1.11[CI 0.40–2.57]; <91 days, OR 1.93[CI 0.74–5.68]; 91+ days, OR 1.54[CI 0.63–4.34]) (Table 3). However, the Omicron recruitment period (December 2021 to March 2022), was associated with a threefold increase in the risk of PCR-confirmed SARS-CoV-2 infection than the pre-Omicron period (August 2021 to November 2021) (OR 2.56 [CI 2.02–3.26]) (Table 3).
**Table 3**
| Characteristic | Overall, N = 4471 | Omicron phase, N = 3021 | Pre-omicron phase, N = 1451 | p-value2 |
| --- | --- | --- | --- | --- |
| Sex | | | | 0.2 |
| Female | 232 (52%) | 151 (50%) | 81 (56%) | |
| Male | 215 (48%) | 151 (50%) | 64 (44%) | |
| Age | 31 (24, 39) | 31 (25, 38) | 32 (23, 39) | 0.8 |
| HIV status | | | | 0.4 |
| Negative | 277 (84%) | 173 (83%) | 104 (87%) | |
| Positive | 51 (16%) | 35 (17%) | 16 (13%) | |
| Unknown | 119 | 94 | 25 | |
| PLHIV on ART | 49 (11%) | 34 (11%) | 15 (10%) | 0.8 |
| Hypertension | 7 (1.6%) | 5 (1.7%) | 2 (1.4%) | >0.9 |
| Diabetes | 4 (0.9%) | 0 (0%) | 4 (2.8%) | 0.011 |
| Asthma | 15 (3.4%) | 8 (2.6%) | 7 (4.8%) | 0.3 |
| COVID-19 vaccination (doses) | | | | 0.025 |
| 0 | 310 (69%) | 206 (68%) | 104 (72%) | |
| 1 | 63 (14%) | 37 (12%) | 26 (18%) | |
| 2 | 74 (17%) | 59 (20%) | 15 (10%) | |
| Days since last vaccine dose | 137 (51, 187) | 173 (141, 208) | 78 (40, 137) | <0.001 |
| Unknown | 375 | 263 | 112 | |
## Symptoms associated with PCR-confirmed SARS-CoV-2 infection change with variants
Based on the genomic surveillance data (Fig 1C), we assigned all individuals infected within the period from August 2021 to November 2021 as pre-Omicron infections (most commonly Delta infections), and those from December 2021 to March 2022 as Omicron infections. Twenty-two percent ($\frac{447}{2009}$) of vaccine-eligible patients who had provided the date since the last vaccine dose had PCR-confirmed SARS-CoV-2 infection (Table 4). Thirty-five out of 228 patients were less than 18 years of age and had PCR-confirmed SARS-CoV-2 infection (S2 Table). As such subsequent analyses only focused on the adult population (≥18 years). Fifty percent ($\frac{151}{302}$) of these patients in Omicron phase were female, while $56\%$ ($\frac{81}{145}$) were female in the pre-Omicron phase. The median age of patients from the two phases was similar (31 years [IQR 25–38] vs. 32[IQR 23–39], $$p \leq 0.8$$) (Table 4). Among those with known self-reported HIV status, the HIV prevalence was $16\%$, which was similar between the two phases ($17\%$ vs $13\%$, $$p \leq 0.4$$). Of those People Living with HIV (PLHIV), $97\%$ ($\frac{34}{35}$) in the Omicron phase and $94\%$ ($\frac{15}{16}$) in the pre-Omicron phase were on antiretroviral therapy. Other chronic illnesses were uncommon, with only $1.6\%$ ($\frac{7}{447}$), $0.9\%$ ($\frac{4}{447}$) and $3.4\%$ ($\frac{15}{447}$) self-reported to have hypertension, diabetes, and asthma, respectively, although hypertension and diabetes, in particular, are widely underdiagnosed in Malawi [19, 20]. Thirty-one percent ($\frac{137}{447}$) of the participants reported being vaccinated with at least one dose of the COVID-19 vaccine (AstraZeneca vaccine ($80\%$ ($\frac{109}{137}$)) or Janssen vaccine ($20\%$ ($\frac{28}{137}$)). Of which, $54\%$ ($\frac{74}{447}$) had self-reported to have received at least two doses of the AstraZeneca vaccine, with more patients in the Omicron phase than the pre-Omicron phase ($20\%$ ($\frac{59}{302}$) vs. $10\%$ ($\frac{15}{145}$), $$p \leq 0.025$$). Moreover, the days since last vaccine dose were longer in the Omicron phase than pre-Omicron phase (173[IQR 141–208] vs. 78[IQR 40–137], $p \leq 0.001$).
**Table 4**
| Characteristic | PCR positivity | PCR positivity.1 | PCR positivity.2 | PCR positivity.3 | Univariable | Univariable.1 | Univariable.2 | Univariable.3 | Multivariable | Multivariable.1 | Multivariable.2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Characteristic | Overall, N = 2,0091 | Negative, N = 1,6271 | Positive, N = 3821 | p-value2 | N | OR3 | 95% CI3 | p-value | OR3 | 95% CI3 | p-value |
| COVID-19 vaccine doses | | | | 0.027 | 2009 | | | | | | |
| 0 | 1,718 (100%) | 1,408 (82%) | 310 (18%) | | | Ref. | — | | Ref. | — | |
| 1 | 142 (100%) | 107 (75%) | 35 (25%) | | | 1.49 | 0.98, 2.20 | 0.053 | 1.10 | 0.39, 2.66 | 0.8 |
| 2 | 149 (100%) | 112 (75%) | 37 (25%) | | | 1.50 | 1.00, 2.20 | 0.042 | 1.11 | 0.40, 2.57 | 0.8 |
| Days since last vaccine dose | | | | 0.003 | 2009 | | | | | | |
| 0 | 1,763 (100%) | 1,447 (82%) | 316 (18%) | | | Ref. | — | | Ref. | — | |
| <91 days | 90 (100%) | 68 (76%) | 22 (24%) | | | 1.48 | 0.88, 2.39 | 0.12 | 1.93 | 0.74, 5.68 | 0.2 |
| 91+ days | 156 (100%) | 112 (72%) | 44 (28%) | | | 1.80 | 1.23, 2.58 | 0.002 | 1.54 | 0.63, 4.34 | 0.4 |
| Sex | | | | 0.4 | 2009 | | | | | | |
| Female | 1,001 (100%) | 803 (80%) | 198 (20%) | | | Ref. | — | | Ref. | — | |
| Male | 1,008 (100%) | 824 (82%) | 184 (18%) | | | 0.91 | 0.72, 1.13 | 0.4 | 0.84 | 0.67, 1.06 | 0.15 |
| Age group | | | | 0.13 | 2009 | | | | | | |
| 18–50 years | 1,904 (100%) | 1,536 (81%) | 368 (19%) | | | Ref. | — | | Ref. | — | |
| 51+ years | 105 (100%) | 91 (87%) | 14 (13%) | | | 0.64 | 0.35, 1.10 | 0.13 | 0.62 | 0.33, 1.09 | 0.12 |
| Recruitment period | | | | <0.001 | 2009 | | | | | | |
| Pre-omicron phase | 1,080 (100%) | 943 (87%) | 137 (13%) | | | Ref. | — | | Ref. | — | |
| Omicron phase | 929 (100%) | 684 (74%) | 245 (26%) | | | 2.47 | 1.96, 3.11 | <0.001 | 2.56 | 2.02, 3.26 | <0.001 |
Clinical symptoms associated with PCR-confirmed SARS-CoV-2 infections were different during the Omicron and pre-Omicron phases (S3 Table). Cough ($70\%$ vs. $54\%$, $p \leq 0.001$), fatigue ($14\%$ vs $3.6\%$, $p \leq 0.001$), loss of taste ($10\%$ vs. $5.0\%$, $$p \leq 0.033$$), headache ($50\%$ vs. $39\%$, $$p \leq 0.034$$), and abdominal pain ($14\%$ vs. $8.3\%$, $$p \leq 0.043$$) were more frequent in patients from the pre-Omicron phase compared to the Omicron phase. In contrast, muscle ache was more common in the Omicron phase than pre-Omicron phase ($35\%$ vs $25\%$, $$p \leq 0.029$$).
Controlling for age, sex and COVID-19 vaccination status in a multivariable analysis, cough (OR 0.37 [CI 0.22–0.61], $p \leq 0.001$), fatigue (OR 0.20 [CI 0.08–0.48], $p \leq 0.001$), headache (OR 0.47 [CI 0.29–0.79], $$p \leq 0.001$$), and abdominal pain (0R 0.38 [CI 0.18–0.78]) were independently associated with pre-Omicron than Omicron SARS-CoV-2 PCR positive diagnosis (Fig 2 and Table 5). Conversely, fever was independently associated with Omicron than pre-Omicron SARS-CoV-2 PCR positive diagnosis (OR 2.46 [CI 1.29–4.97]) (Fig 2 and Table 5).
**Fig 2:** *Clinical symptoms associated with PCR-confirmed SARS-CoV-2 infection.Differences in clinical symptoms of PCR-confirmed SARS-CoV-2 infected adults during the Omicron and Pre-Omicron phases ($$n = 447$$).* TABLE_PLACEHOLDER:Table 5
## Discussion
In a setting where there is a high burden of presentations with infectious disease, we found that the symptoms associated with SARS-CoV-2 Omicron infection have become considerably less distinct, differing significantly from those infections with pre-Omicron variants (predominantly Delta). Indeed fever, which is common to many infectious presentations [11, 12], was most prevalent among presumed Omicron infected patients, while headache, cough, fatigue, and abdominal pain were significantly more prevalent among pre-Omicron cases.
Our data showing a different symptom profile associated with Omicron infection is consistent with studies elsewhere [21] with headache being prominent in three other studies from the UK [21–23]. However, we observed high odds for presenting with fever in presumed Omicron-infected patients than pre-Omicron patients, in contrast with the two studies in the UK [21, 23]. The main differences between the Malawi study and the UK studies are age and prevalence of Omicron sub-lineages, with Malawi cohort being a younger population and having predominantly BA.1 at time of sampling. BA.1 is associated with a different symptom profile than BA.2 [22]. Together, our findings and those of others suggest that the clinical case definition of COVID-19 used for testing and surveillance may need to be revised to maintain case ascertainment.
Data from Malawi and elsewhere has shown that the Omicron variant has presented with less severe disease, hospitalisation and deaths, than the pre-Omicron VOCs [1, 24]. The Omicron variant have been shown to be less capable of transition from the upper to lower respiratory tract infection [25], and this could potentially contribute to the low incidence of severe disease. However, data in non-immunised populations in Hong Kong indicate that *Omicron is* not intrinsically mild [26, 27], suggesting that immune response or past exposure could be an important determinant of this low severity. Data from South Africa has shown that high SARS-CoV-2 seroprevalence has been associated with low number of deaths and hospitalisation attributed to the Omicron variant [24]. In Malawi, seroprevalence data has shown that more than $70\%$ of the population had anti-SARS-CoV-2 receptor binding domain (RBD) antibodies before the Omicron variant pandemic wave [5]. In line with previous findings [28], COVID-19 vaccination was not associated with a reduced risk of PCR-confirmed SARS-CoV-2 infection, especially during the Omicron wave. It is therefore plausible that the altered clinical presentation observed in our study could also be attributed to pre-existing immunity from prior SARS-CoV-2 exposure.
Furthermore, our findings align with the temporal dynamics of the COVID-19 pandemic waves in Malawi and the region [1, 5, 24]. A high SARS-CoV-2 prevalence of 30–$50\%$ among patients presenting to primary healthcare at the peak of the three pandemic waves, is consistent with high reported national COVID-19 cases during the same period [1]. Furthermore, consistent with genomic surveillance [15, 29], our sentinel surveillance correctly identified the VOCs driving the local pandemic waves. Due to the consistency in our sampling over time, we were able to provide real-time data to aid public health response in Malawi, especially on the identification of VOCs driving community transmission. Collectively, this indicates that sentinel surveillance backed up by diagnostics and genomics data could be an early warning system for national pandemic response in resource-limited settings, considering that by the time hospitalisations are rising it is already too late to intervene effectively.
Our study had several limitations. Firstly, the study was conducted in urban Blantyre and findings may not be generalisable to rural settings. Secondly, our study cohort (median age 28 years (IQR 21–38)) was not fully representative of the population structure in Malawi (median age 17.5 years) [30]. Thirdly, since our genomic surveillance was limited to a subset of samples with low PCR CT values, this approach biases our identification of variants to those causing high viral burden infections at time of recruitment. Lastly, our analysis of the impact of COVID-19 vaccination did not adjust for immunity induced following previous exposure to SARS-CoV-2, as it has been previously shown to be protective against symptomatic COVID-19 [31, 32]. The SARS-CoV-2 seroprevalence has been reported to be very high in Malawi [5], despite a low vaccination coverage [33].
In conclusion, our study demonstrates changes in clinical symptoms overtime, aligned to infecting variant, indicating that case definitions of COVID-19 need constant monitoring and revision to match SARS-CoV-2 evolution to maintain its relevance for institutional and national testing policies. This study also highlights the importance and utility of sentinel surveillance in low-resourced settings to aid timely public health response against the COVID-19 pandemic and future pandemics.
## References
1. 1JHU. COVID-19 Case Tracker
John Hopkins University
2020. Available from: https://coronavirus.jhu.edu/map.html.. *COVID-19 Case Tracker* (2020.0)
2. Lewis H, Ware H, Whelan M, Subissi L, Li Z, Ma X. **SARS-CoV-2 infection in Africa: A systematic review and meta-analysis of standardised seroprevalence studies, from January 2020 to December 2021**. *medRxiv* (2022.0). DOI: 10.1136/bmjgh-2022-008793
3. 3Organization WH. Africa COVID-19 vaccination dashboard:
WHO; 2022. Available from: https://app.powerbi.com/view?r=eyJrIjoiY2ViYzIyZjItYzhkMi00ZWVkLTgyM2ItZTk1ZTJmODRjMTkxIiwidCI6ImY2MTBjMGI3LWJkMjQtNGIzOS04MTBiLTNkYzI4MGFmYjU5MCIsImMiOjh9.. *Africa COVID-19 vaccination dashboard:* (2022.0)
4. 4WHO. End-to-end integration of SARS-CoV-2 and influenza sentinel surveillance. Global influenza programme [Internet]. 2022.
5. Mandolo J, Msefula J, Henrion MYR, Brown C, Moyo B, Samon A. **SARS-CoV-2 exposure in Malawian blood donors: an analysis of seroprevalence and variant dynamics between January 2020 and July 2021**. *BMC Med* (2021.0) **19** 303. DOI: 10.1186/s12916-021-02187-y
6. 6PHIM. Daily COVID–19 Situation Report.
Lilongwe:
Public Health Institute of Malawi, Ministry of Health, 2020 1st July 2020. Report No.. *Daily COVID–19 Situation Report.* (2020.0)
7. **Estimating global, regional, and national daily and cumulative infections with SARS-CoV-2 through Nov 14, 2021: a statistical analysis**. *Lancet* (2022.0). DOI: 10.1016/S0140-6736(22)00484-6
8. 8WHO. COVID-19 vaccine tracker and landscape: WHO; 2021. Available from: https://www.who.int/publications/m/item/draft-landscape-of-covid-19-candidate-vaccines.. *COVID-19 vaccine tracker and landscape* (2021.0)
9. 9WHO. Coronavirus disease (COVID-19) 2022. Available from: https://www.who.int/health-topics/coronavirus#tab=tab_3.
10. 10CDC. Symptoms of COVID-19: CDC; 2022. Available from: https://www.cdc.gov/coronavirus/2019-ncov/symptoms-testing/symptoms.html.
11. Maze MJ, Bassat Q, Feasey NA, Mandomando I, Musicha P, Crump JA. **The epidemiology of febrile illness in sub-Saharan Africa: implications for diagnosis and management**. *Clin Microbiol Infect* (2018.0) **24** 808-14. DOI: 10.1016/j.cmi.2018.02.011
12. Elven J, Dahal P, Ashley EA, Thomas NV, Shrestha P, Stepniewska K. **Non-malarial febrile illness: a systematic review of published aetiological studies and case reports from Africa, 1980–2015**. *BMC Med* (2020.0) **18** 279. DOI: 10.1186/s12916-020-01744-1
13. Morton B, Barnes KG, Anscombe C, Jere K, Matambo P, Mandolo J. **Distinct clinical and immunological profiles of patients with evidence of SARS-CoV-2 infection in sub-Saharan Africa**. *Nat Commun* (2021.0) **12** 3554. DOI: 10.1038/s41467-021-23267-w
14. 14ISARIC. COVID-19 CRF 2022. Available from: https://isaric.org/research/covid-19-clinical-research-resources/covid-19-crf/.
15. Anscombe C, Lissauer S, Thole H, Rylance J, Dula D, Menyere M. **Genomically informed clinical comparison of three epidemic waves of COVID-19 in Malawi**. *medRxiv* (2022.0). DOI: 10.1101/2022.02.17.22269742
16. 16ARTIC field bioinformatics pipeline:
GitHub; 2022. Available from: https://github.com/artic-network/fieldbioinformatics.. *ARTIC field bioinformatics pipeline:* (2022.0)
17. Simulundu E, Mupeta F, Chanda-Kapata P, Saasa N, Changula K, Muleya W. **First COVID-19 case in Zambia—Comparative phylogenomic analyses of SARS-CoV-2 detected in African countries**. *International journal of infectious diseases: IJID: official publication of the International Society for Infectious Diseases* (2021.0) **102** 455-9. DOI: 10.1016/j.ijid.2020.09.1480
18. O’Toole Á, Scher E, Underwood A, Jackson B, Hill V, McCrone JT. **Assignment of epidemiological lineages in an emerging pandemic using the pangolin tool**. *Virus Evol* (2021.0) **7** veab064. DOI: 10.1093/ve/veab064
19. Amberbir A, Lin SH, Berman J, Muula A, Jacoby D, Wroe E. **Systematic Review of Hypertension and Diabetes Burden, Risk Factors, and Interventions for Prevention and Control in Malawi: The NCD BRITE Consortium**. *Glob Heart* (2019.0) **14** 109-18. DOI: 10.1016/j.gheart.2019.05.001
20. Price AJ, Crampin AC, Amberbir A, Kayuni-Chihana N, Musicha C, Tafatatha T. **Prevalence of obesity, hypertension, and diabetes, and cascade of care in sub-Saharan Africa: a cross-sectional, population-based study in rural and urban Malawi**. *Lancet Diabetes Endocrinol* (2018.0) **6** 208-22. DOI: 10.1016/S2213-8587(17)30432-1
21. Menni C, Valdes AM, Polidori L, Antonelli M, Penamakuri S, Nogal A. **Symptom prevalence, duration, and risk of hospital admission in individuals infected with SARS-CoV-2 during periods of omicron and delta variant dominance: a prospective observational study from the ZOE COVID Study**. *Lancet* (2022.0) **399** 1618-24. DOI: 10.1016/S0140-6736(22)00327-0
22. Whitaker M, Elliott J, Bodinier B, Barclay W, Ward H, Cooke G. **Variant-specific symptoms of COVID-19 among 1,542,510 people in England**. *medRxiv* (2022.0). DOI: 10.1101/2022.05.21.22275368
23. Vihta K-D, Pouwels KB, Peto TE, Pritchard E, House T, Studley R. **Omicron-associated changes in SARS-CoV-2 symptoms in the United Kingdom**. *medRxiv* (2022.0). DOI: 10.1093/cid/ciac613
24. Madhi SA, Kwatra G, Myers JE, Jassat W, Dhar N, Mukendi CK. **Population Immunity and Covid-19 Severity with Omicron Variant in South Africa**. *N Engl J Med* (2022.0). DOI: 10.1056/NEJMoa2119658
25. Hui KPY, Ho JCW, Cheung MC, Ng KC, Ching RHH, Lai KL. **SARS-CoV-2 Omicron variant replication in human bronchus and lung ex vivo**. *Nature* (2022.0) **603** 715-20. DOI: 10.1038/s41586-022-04479-6
26. Lewis D.. **Will Omicron finally overpower China’s COVID defences?**. *Nature* (2022.0). DOI: 10.1038/d41586-022-00884-z
27. Tso WWY, Kwan M, Wang LY, Leung LK, Leung D, Chua GT. **Intrinsic Severity of SARS-CoV-2 Omicron BA.2 in Uninfected, Unvaccinated Children: A Population-Based, Case-Control Study on Hospital Complications**. *SSRN* (2022.0). DOI: 10.2139/ssrn.4063036
28. Feikin DR, Higdon MM, Abu-Raddad LJ, Andrews N, Araos R, Goldberg Y. **Duration of effectiveness of vaccines against SARS-CoV-2 infection and COVID-19 disease: results of a systematic review and meta-regression**. *Lancet* (2022.0) **399** 924-44. DOI: 10.1016/S0140-6736(22)00152-0
29. Wilkinson E, Giovanetti M, Tegally H, San JE, Lessells R, Cuadros D. **A year of genomic surveillance reveals how the SARS-CoV-2 pandemic unfolded in Africa**. *Science* eabj4336. DOI: 10.1126/science.abj4336
30. 30NSO. Malawi Population and Housing Census.
Zomba:
National Statistical Office, Malawi, 2018.. *Malawi Population and Housing Census.* (2018.0)
31. Wei J, Pouwels KB, Stoesser N, Matthews PC, Diamond I, Studley R. **Antibody responses and correlates of protection in the general population after two doses of the ChAdOx1 or BNT162b2 vaccines**. *Nature Medicine* (2022.0) **28** 1072-82. DOI: 10.1038/s41591-022-01721-6
32. Chemaitelly H, Ayoub HH, AlMukdad S, Coyle P, Tang P, Yassine HM. **Protection from previous natural infection compared with mRNA vaccination against SARS-CoV-2 infection and severe COVID-19 in Qatar: a retrospective cohort study**. *The Lancet Microbe* (2022.0). DOI: 10.1016/S2666-5247(22)00287-7
33. Mathieu E, Ritchie H, Rodés-Guirao L, Appel C, Giattino C, Hasell J. **Coronavirus Pandemic (COVID-19)**. (2020.0)
|
---
title: 'The global burden of high fasting plasma glucose associated with zinc deficiency:
Results of a systematic review and meta-analysis'
authors:
- James P. Wirth
- Wu Zeng
- Nicolai Petry
- Fabian Rohner
- Scott Glenn
- William E. S. Donkor
- Rita Wegmüller
- Erick Boy
- Keith Lividini
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022216
doi: 10.1371/journal.pgph.0001353
license: CC BY 4.0
---
# The global burden of high fasting plasma glucose associated with zinc deficiency: Results of a systematic review and meta-analysis
## Abstract
Non-communicable diseases (NCDs) account for the largest share of the global disease burden, and increasing evidence shows that zinc deficiency (ZD) contributes to NCDs by inducing oxidative stress, insulin resistance, and impaired lipid metabolism. A systematic review and meta-analysis was conducted to determine whether ZD was associated with fasting plasma glucose (FPG), a key risk factor for NCDs. A random effects meta-analysis was conducted to determine the strength of the association in the form of an odds ratio (OR) and subsequently the population attributable risk (PAR) with population prevalences of high FPG. The disease burden from high FPG attributable to ZD was expressed as disability adjusted life years (DALYS). Data from seven studies were obtained as part of the systematic review. The meta-analysis shows a significant ($p \leq 0.01$) inverse relationship between ZD and high FPG (OR = 2.34; $95\%$ CI: 1.16, 4.72). Globally, the PAR of ZD’s contribution to high FPG is $6.7\%$, with approximately 8.2 million high FPG DALYs attributable to ZD. Cardiovascular diseases, diabetes, and chronic kidney diseases account for more than $90\%$ of the total DALYs. Total DALYs attributable to ZD are largest in the “Southeast Asia, East Asia, and Oceania” and “High Income” Super Regions. While the disease burden is highest among populous countries (e.g., China, India, USA), the population-standardized DALYs are highest among island nations, particularly island nations in the South Pacific and Caribbean. While ZD accounts for a small share of the high FPG disease burden, the total number of DALYs far surpasses other estimates of the disease burden attributable to ZD, which focus on diarrheal diseases in childhood. Zinc interventions are urgently needed to help address the increasing disease burden from NCDs, and the double burden of malnutrition.
## Introduction
Non-communicable diseases (NCD; e.g., cardiovascular diseases, diabetes, cancers, chronic respiratory diseases) are a leading contributor to the burden of disease globally, and account for a majority of the global deaths [1]. The burden of disease per capita related to NCD is highest in high-income countries, but is rising in low- and middle-income countries [2]. A key metric of disease burden is the Disability-Adjusted Life Years (DALY), which standardizes the loss of life from death and disability and enables a direct comparison of disparate disease states. Globally, DALYs from NCDs have almost doubled between 1990 and 2017 [3], and most countries are not on track to meet the 2030 Sustainable Development Goals’ targets [1].
Obesity, hypertension, and insulin resistance are well-documented risk factors for NCDs, and there is increasing evidence that low circulating zinc levels contribute to NCDs by inducing oxidative stress [4], insulin resistance [5], and by altering lipid metabolism [6]. Multiple randomized controlled trials (RCTs) have investigated the effect of zinc supplementation on risk factors for diabetes and cardiovascular disease such as total cholesterol, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, and triglycerides, or direct measures of the disease such as fasting plasma glucose (FPG) or glycated hemoglobin (HbA1c) [5,7–11]. Several meta-analyses have systematically analyzed the results of these RCTs [5,7,9,10]. Although the inclusion criteria of the different meta-analyses varied, almost all analyses revealed that zinc administration reduced FPG, HbA1c, and insulin resistance. Pompano & Boy [7] included RCTs that administered zinc supplements at doses below 100mg/ day to healthy and unhealthy patients, subgrouping the interventions by dose and duration. They found larger effects for almost all indicators when the zinc dose administered was <25mg/d and for trials with a duration of 12 weeks or more. Similar results have been reported from a meta-analysis by Capdor et al. [ 10], which found a significant reduction in FPG in the intervention groups receiving zinc supplements. In sub-group analyses by health status, the effect was stronger in the diabetic sub-group and ceased to exist in the healthy subgroup [10]. In contrast, Wang et al. [ 5] found no difference in the magnitude of the effect between trials using different doses or duration.
These studies suggest that zinc supplementation has a therapeutic effect among diabetic patients and may delay the progression of zinc-associated chronic diseases. Specifically, there is mounting evidence that poor zinc status can increase levels of FPG, a key contributor of type-2 diabetes, cardiovascular diseases, and certain cancers [12], and that increasing dietary zinc could reduce the risk of developing type-2 diabetes [13] and the risk factors of certain cardiovascular diseases [14]. However, the strength of the association between zinc supplementation and elevated FPG has not yet been fully quantified among individuals who are zinc deficient compared with zinc-replete individuals.
Historically, zinc deficiency (ZD) and its contribution to the burden of disease has been linked to under-nutrition related conditions in children, such as diarrhea [15], pneumonia [16], and growth faltering [17]; and the disease burden from ZD has been quantified on the same basis for more than a decade [18,19] for both characterization and cost-effectiveness analysis of zinc interventions [20,21]. NCDs have not been included in these analyses; however, due to the growing body of evidence showing ZD’s association with insulin resistance and type-2 diabetes, quantification of the zinc-related disease burden from NCD in all population groups is warranted.
The objective of this study was to estimate the global disease burden of NCDs attributable to ZD. Our study had four aims, which were all achieved. The first aim was to conduct a systematic review of studies that determined elevated FPG among those with and without ZD. The second aim was to characterize the odds ratio (OR) for elevated FPG from ZD. The third aim was to determine the population attributable risk (PAR) for all countries using the OR and the estimated prevalence of elevated FPG for each country. The final aim was to estimate the total number of DALYs from elevated FPG attributable to ZD using each country’s calculated PAR and its estimated burden of disease due to elevated FPG determined from the Global Burden of Disease (GBD) project.
By estimating the OR based on a systematic review of the literature, this study makes an empirical contribution to literature examining the association between the ZD and high FPG. Furthermore, this analysis makes a methodological contribution by utilizing the results from the meta-analysis and data from the GBD project to estimate the disease burden from high FPG that is associated with ZD.
## Experimental design
The aims and experimental design of this study are shown in Fig 1. The study commenced with a systematic review of literature databases (Aim 1) to identify studies from multiple population groups from several countries and locations that specifically included information on both zinc status (deficient vs. sufficient) and FPG status (elevated vs. normal). The systematic review also included an author outreach step, whereby authors of studies that contained zinc status and FPG indicators were contacted and requested to re-analyze their data if the article did not already present the results in a usable format. Subsequently, a random effects meta-analysis was conducted to determine the strength of the association in the form of an OR (Aim 2) that could be used for the estimation of the PAR of ZD inducing high FPG (Aim 3). Consequently, the disease burden attributable to ZD was calculated as the product of the PAR and the disease burden from high FPG—presented in DALYs and presented separately for each condition (Aim 4). A detailed description of the procedures undertaken for the meta-analysis and estimation of the PARs and disease burden attributable to ZD are presented in the Statistical Analysis section below.
**Fig 1:** *Analytical phases of the research study.*
At the outset of the study, the research team also conducted a systematic review and meta-analysis of the dichotomous relationship between ZD and high LDL cholesterol. This meta-analysis yielded few studies and did not show an association between the two factors. As such, the meta-analysis results were not used to estimate the proportion of the disease burden from high LDL cholesterol that was attributable to ZD (see Fig A and Fig B in S1 Text for a summary of these findings).
## Search strategy
A literature search was conducted in PUBMED, Scopus, SciElo, and Academic Search Premier databases, and the Cochrane Review Library (last search in all databases November 2021) to identify articles that examined the association between zinc status and fasting blood glucose. The search term used was as follows: (zinc OR zinc* OR zn) AND ("deficiency" OR "status" OR “level” OR “intake” OR “insufficiency”) AND ("fasting glycemia" OR "fasting glucose" OR "glucose" OR "glycemic" OR "glycemia" OR “fasting blood glucose” OR “blood glucose”) AND (epidemiology* OR "epidemiological studies" OR "case control" OR “case-control” OR “retrospective study” OR "cohort study" OR “study” OR “trial”). Prior to screening, any duplicates were removed and only studies written in English, Spanish, or Portuguese were retained.
## Inclusion and exclusion criteria of title and abstract screenings
Two authors (JPW and FR) independently screened the titles of all articles. Studies were included if A) the title was suggestive, or B) the title indicated that the study was conducted on humans or the subjects were not specified in the title. If human population was determined by criteria B, the study was retained if the human population was not a group currently undergoing intense medical interventions, such as chemotherapy or dialysis. Article titles that indicated that human subjects had a medical condition (e.g., diabetes, cancer) were retained during the title screening.
JPW and FR then also conducted an independent abstract screening. Articles were retained if the abstract did not specify the study population, or when the abstract noted the subjects were humans that were not currently undergoing intense medical interventions or there was no mention of medical interventions. Articles were also retained if the human population did not have a condition that adversely affects zinc status, such as pancreatic insufficiency or inflammatory bowel diseases [22]. Studies that were conducted only with diabetic patients (without non-diabetic control group) were excluded at this stage since studies of diabetic patients only would not contain sufficient variation in glucose levels to estimate the OR between ZD and high FPG, as there would be no subjects with normal FPG levels. To ensure that the results of the systematic review and meta-analysis would be analogous to the DALYs produced by the Institute of Health Metrics and Evaluation (IHME), studies were retained only if glucose (fasting or non-fasting) was measured; studies that only measured other biomarkers related to FPG (e.g., HbA1c) were not retained. No exclusion was made based on the zinc indicator noted in the abstract, and studies measuring zinc status (e.g., serum/plasma zinc, toenail zinc, hair zinc) and dietary zinc intake were all retained.
## Full text reviews and data extraction
As part of the full text reviews, all relevant data were extracted into a Microsoft Excel worksheet. The full text reviews and data extraction were undertaken by two researchers (FR & WESD). A third researcher (JPW) extracted the data from a random subsample for $20\%$ of the studies. A full double data extraction was planned if the kappa value of the inter-observer agreement was below 0.6 [23], but this was not required as the kappa value was 0.85.
As part of the data extraction, general information was recorded about the design of the studies and the population groups included. Regarding zinc status, information related to zinc biomarkers was collected, such as the unit of measurement, deficiency cutoff used, and ranges of zinc if presented as quantiles. Information about zinc intake was also collected, including the measurement method and cutoff for insufficient intake. Basic information related to glucose was also recorded, including the glucose indicator, measurement unit and cutoffs used. The researchers noted if any association between zinc and glucose was presented in the article, and if so, noted the metric that was presented (e.g., OR, risk ratio, hazard ratio) and if it was calculated using bivariate or multivariate models. When multivariate models were used, the control variables were also recorded.
During the data extraction process, the research team also noted if a study did not report the outcomes in a useable manner, but contained variables of interest (e.g., plasma zinc and FPG). For these articles, the research team contacted the respective corresponding authors by email and requested that the data be reanalyzed to produce the required ORs or for the raw data to be provided. When sending the requests, the research team asked for authors to conduct an unweighted logistic regression with selected control variables (e.g., age, sex, body-mass index, smoking status) if the study had a cross-sectional, case-control, or cohort design. For RCTs, the research team requested that a conditional logistic regression be used to generate the required ORs.
## Publicly available data and primary data analysis
An ancillary activity of the systematic review was an online search for publicly available data that could be analyzed for the purposes of this study. The authors were able to identify and obtain datafiles from the US NHANES and the South Korea NHANES (KNHANES). Regarding the US-NHANES, three survey rounds contained both zinc and glucose data: 2011–12, 2013–14, and 2015–16. As the systematic review already identified an article including data from the 2011–12 and 2013–14 rounds [24], only the 2015–2016 US-NHANES data were utilized in addition. Regarding the KNHANES, data from 2010 contained both zinc and glucose measures [25]. The 2010 KNHANES dataset was not, however, included in the meta-analysis as the regression results were unstable because only four of the 2354 subjects with both zinc and glucose data were zinc deficient.
## Risk of bias
Risk of bias was assessed for all studies identified through the systematic review and all primary databases obtained. To assess the risk for bias, we used a modified version of the Newcastle Ottawa Scale (NOS) for evaluating the quality of non-randomized studies in meta-analyses [26]. The modification consisted mainly in expanding the NOS to include cross-sectional studies and specific features of our analysis. Specific characteristics that were assessed included representativeness, randomness, ascertainment of exposure (i.e., indicator used to assess ZD), and ascertainment of outcomes (i.e., analytical method used to measure FPG concentration). The quality of the studies included were rated good, fair, or poor depending on the score given to each characteristic following the guidelines of the NOS. Two researchers (WZ & JPW) independently reviewed the risk of bias of each study to be included in the meta-analyses.
## Ethical considerations
As primary data collection was not required for this study, no ethical approval was sought. Prior to conducting the study, the study protocol was registered with the international registry of systematic reviews, PROSPERO (CRD42021264234).
## Laboratory analyses
Zinc concentrations were measured using atomic absorption spectrometry by Gonoodi et al. [ 27], Bo et al. [ 28], and Obeid et al. [ 29], and using inductively coupled plasma mass spectrometry by Shan et al. [ 30], Bulka et al. [ 24], Li et al. [ 31], and the 2015–16 NHANES [32]. Bulka et al. [ 24] and the 2015–16 NHANES [33] measured FPG concentrations using the “gold standard” hexokinase enzymatic method. Bo et al. [ 28] measured FPG concentrations using the glucose oxidase method, whereas all other studies use automatic biochemical analyzers or commercial kits for measuring FPG.
## Indicator parameters
ZD was pre-defined as serum/plasma zinc ≤70 ug/dL [34] when raw data were available or when authors re-analyzed data for this study. When data were not presented in a dichotomous format, a conversion was made to indirectly estimate the dichotomous association between ZD and high FPG with a modelled threshold of ≤70 ug/dL (see details below). High FPG glucose concentration was defined as ≥ 126 mg/dl based on the WHO diagnostic criteria for diabetes [35].
## Primary data analysis of 2015–16 NHANES
Primary data analysis of the 2015–16 NHANES survey was done using a similar approach as used by Bulka et al. [ 24], who used data from the NHANES’ 2011–12 and 2013–14 rounds and whose results were included in the meta-analysis. In brief, we developed a logistic regression model with high FPG as the dependent variable. The independent variables in the model included ZD, age, BMI, education level, total caloric intake, number of alcoholic drinks per day in the past year, smoking status, and physical activity status. The regression model accounted for stratification and clustering, but survey weights were not used for the regression as the NHANES does not recommend weighting when key indicators were measured in different sub-samples, as was the case for zinc and glucose [36].
## Conversion of categorical results to dichotomous results
All the studies used the dichotomous outcome high FPG. ORs and $95\%$ confidence intervals were estimated for each study. A conversion step was performed when zinc status was not measured dichotomously. If zinc status was measured in percentiles, a weight was calculated and assigned to each OR obtained from the study. The weights were calculated as 1/(number of groups -1). For example, if zinc status was grouped in terciles (three categories), two ORs and the associated $95\%$ confidence intervals should be obtained from the study, then a weight of 0.5 was attached to each of the ORs. To achieve normality of the distribution, we then converted each OR to the logarithmic form. Using the total sample size in the study, we calculated the number of observations in the sample in each category, and then re-sampled from the transformed distributions by randomly drawing with replacement the number of log ORs equal to the sample size for each category. A weighted log OR and associated $95\%$ CIs were then calculated as the aggregated log OR if the zinc status were measured dichotomously. The OR was then estimated by taking the antilog of log OR.
When zinc status was measured as a continuous variable, the study result was interpreted as the change in log odds with each 1-unit increase in zinc. In contrast, we used a dichotomous ZD variable (0 = not deficient, 1 = deficient) in the analysis of NHANES 2015–16 data to calculate the change in log odds of elevated FPG when comparing those with ZD to those who are not deficient. The value was determined to be 23.79 μg/dl. The OR for dichotomous measure was then estimated through exponentiation, as exp(ln(1/OR*23.79)).
## Meta analyses
A total of seven studies were included in the meta-analysis (see Fig 1). For the meta-analysis, we estimated the weighted pooled effect size (OR) and its $95\%$ confidence interval (CI) using a random-effects model. The selection of the model was determined by the degree of heterogeneity, which was evaluated using the I2 statistic. The random-effects model was favored in the presence of heterogeneity. To explore heterogeneity among studies, we reran the meta-analyses after we removed each study one at a time to examine its impact on the heterogeneity and summary of effect size. Additionally, raw and standardized residuals of the fitted random-effects model were examined. A forest plot was generated to present the summary of effects size. Funnel plot asymmetry was evaluated, and a significant publication bias was considered if the P value was less than 0.05. All meta-analyses were performed using the Metafor package in R statistical software (version 4.1.2; R Foundation, Vienna, Austria). All tests were 2-tailed, and $P \leq .05$ was considered statistically significant.
## Estimating PAR
The OR produced by the meta-analysis was used in conjunction with estimates of the prevalence of high FPG at the country- or GBD Super Region-level to calculate PARs. The estimated prevalence of high FPG was calculated for all countries and GBD Super Regions using the estimated density curves for FPG that were previously calculated by the 2019 GBD project [37]. In the 2019 GBD project, for each country or region, a lower bound of 126 mg/dl was set, and the area under the curve above the lower bound was determined and used to calculate the proportion of each population with FPG ≥ 126 mg/dl or the area above the lower bound.
## Estimation of NCD burden attributable to ZD
Prior to estimating the disease burden attributable to ZD, we modified the methodology used by IHME to estimate the disease burden from high FPG by excluding the burden of disease for the estimated proportion of the population with FPG concentrations < 126 mg/dl. This was done so that the DALYs from high FPG would match the dichotomous nature of the meta-analysis, whereby the OR expressed the association between two dichotomous variables, ZD and high FPG, where the null value of the high FPG variables presumes no risk. This is in contrast to the publicly-available DALYs attributable to high FPG available on the Global Health Data Exchange [38] that use a continuous risk curve to estimate how the burden of disease increases as glucose concentrations rise.
DALY estimates were calculated for all disease outcomes based on IHME’s hierarchy of four aggregation levels of disease burden. Movement down the hierarchy (i.e., from level 1 to level 4) results in greater specificity in disease groupings at each subsequent level. The total disease burden at lower levels aggregates to the next higher level in the hierarchy. For our purposes, IHME’s “level 3” diseases provide a suitable level of specificity for estimating the zinc-related disease burden. Specifically, “level 3” outcomes of high FPG included 15 separate conditions: 1) alzheimer’s disease and other dementias, 2) bladder cancer, 3) blindness and vision loss, 4) breast cancer, 5) chronic kidney disease, 6) colon and rectum cancer, 7) diabetes mellitus, 8) ischemic heart disease, 9) liver cancer, 10) ovarian cancer, 11) pancreatic cancer, 12) peripheral artery disease, 13) stroke, 14) tracheal, bronchus, and lung cancer, and 15) tuberculosis.
To estimate the FPG disease burden associated with ZD, the PAR for each country was multiplied by the DALYs associated with the 15 aforementioned disease outcomes for that country. The resulting product was used as the point estimate for the DALYs attributable to ZD. To estimate confidence intervals around these point estimates, Monte Carlo simulation was performed. We conducted 10,000 random draws of data on the disease-specific burden and OR for high FBG from ZD, respectively, and estimated the disease specific burden attributable to ZD from the 10,000 samples. Values at the 2.5 and 97.5 percentile were then estimated as the $95\%$ conference interval of the point estimate.
The high FPG disease burden associated with ZD was calculated and presented both as total DALYs and population-standardized DALYs (i.e., DALYs per 100,000 people). The latter is the quotient of the total DALYs (numerator) and the IHME’s 2019 population estimate (denominator) for countries and Super Regions [39].
## Geographic aggregation and visualization
This manuscript presents disease burden results at the global and Super Region levels. Country-level results are presented in Table A in S1 Text. The GBD project developed Super Regions as a way of grouping countries based on the geographical proximity and epidemiological similarity, resulting in seven groupings: 1) Sub-Saharan Africa, 2) SAEAO, 3) CEEECA, 4) Latin America and Caribbean, 5) North Africa and Middle East, 6) South Asia, and High-Income countries [40]. A list of countries assigned to each Super Region are provided in Table B in S1 Text.
To visualize the results of the study, all maps categories were created by creating quintiles of each variable being mapped, and the cutoffs of each group were selected to have roughly equal numbers of countries in each quintile.
## Patient and public involvement
Our study used secondary and publicly-available data exclusively, and did not require the collection of data from patients or any study participants. As no patients or study participants were involved in the study, these groups will not be included at the time the results are released.
## Literature search
We conducted a literature search to identify studies that specifically included information on both zinc status (deficient vs. sufficient) and FPG status (elevated vs. normal). After removing duplicate references found during the searches of the literature databases, there were 498 unique references included in the systematic review. As shown in Fig 2, the title screening eliminated 386 references, and the subsequent abstract screening eliminated an additional 17 references. As part of the full text review of 95 articles, 42 references were excluded as they did not meet the inclusion criteria. Following the full text review, three articles contained results (e.g., ORs between zinc status quantiles and high FPG) that could be converted to dichotomous associations. One of the three articles (Bulka et al. [ 24]) analyzed data from the US National Health and Nutrition Examination survey (NHANES) rounds from 2011–12 and 2013–14. Using the publicly available NHANES data from a subsequent survey round, JPW and WZ replicated the multivariable model constructed by Bulka et al. and conducted primary data analysis using the 2015–16 NHANES data (see details in the "Statistical Analysis–*Primary data* analysis of 2015–16 NHANES” below). These results constitute an additional data point for the meta-analysis.
**Fig 2:** *PRISMA flow chart for study selection.*
As part of the author outreach, another three studies containing zinc and glucose status were re-analyzed and subsequently included in the analysis. Raw data was also obtained for a study by Vidović et al., [ 41] but due to the study’s small sample size ($$n = 60$$) and minimal overlap of between ZD and high FPG ($$n = 1$$), regression results were unstable and could not be included in the meta-analysis.
## Description of included studies
We conducted a meta-analysis to calculate a pooled estimate of the dichotomous association between high FPG and ZD. Table 1 describes the basic information of the studies included in the meta-analysis. The included studies collected data on subjects in China, Iran, Italy, Lebanon, and the United States of America (USA). Only one included study used a case-control design [30], whereas the other studies were cross-sectional studies. Only the Gonoodi et al. [ 27] study was conducted on adolescents girls, whereas other studies were conducted on adults. Apart from the study by Bo et al. [ 28], which was conducted in pregnant women, all other studies contained non-pregnant women and men. All included studies used serum or plasma zinc as the zinc status indicator, and measured plasma glucose concentrations in individuals that were fasting. In total, the included studies contained 5,858 subjects.
**Table 1**
| Study (ref), publication year | Country of data collection | Design | Participants, n | Age range or mean age in years | Sex | Covariates included in logistic regression | Conversion required (Yes/No) |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Gonoodi et al. [27], 2018 | Iran | Cross-sectional | Adolescent girls, 408 | 12–18 | Female | Age and body-mass index categories | No |
| Shan et al. [30], 2014 | China | Case-control | Adults, 1796 | ≥25 | Both | Age, sex, body-mass index, family history of diabetes, and hypertension status | Yes |
| Bo et al. [28], 2005 | Italy | Cross-sectional | Pregnant women, 194 | 33.0 ± 4.9 | Female | Age, gestational age, body-mass index before pregnancy, and familial diabetes | Yes |
| Obeid et al. [29], 2008 | Lebanon | Cross-sectional | Adults, 398 | 18–65 | Both | Sex, age, body-mass index | No |
| Bulka et al. [24], 2019 | USA | Cross-sectional | Adults, 1088 | >20–80 | Both | Age, sex, body-mass index, race/ethnicity, family income: poverty ratio, total caloric intake, educational attainment, smoking status, average number of drinks per day in past year, physical activity status | Yes |
| Li et al. [31], 2019 | China | Cross-sectional | Adults, 1478 | 21–80 | Both | Age, sex, area (Shimen or Huayuan), body mass index, education, smoking status, alcohol consumption status, and physical activity | No |
| NHANES microdata [42], 2015–2016 | USA | Cross-sectional | Adults, 496 | 20–80 | Both | Age, sex, body-mass index, race/ethnicity, family income: poverty ratio, total caloric intake, educational attainment, smoking status, average number of drinks per day in past year, physical activity status | No |
The risk of bias of the identified studies is presented in Table 2. Three of the seven studies were classified as low risk, and the remaining four studies were classified as medium risk. No studies were excluded based on the risk of bias assessment. Only two studies were based on nationally-representative survey data, and two studies were not representative as subjects were recruited at medical facilities, and one study [31] did not specify how participants were recruited. Simple random selection of subjects was only undertaken by three studies, and the randomness of other studies was mixed. While included studies measured serum zinc using similar spectrometry methods, only two studies measured FPG concentrations with the gold standard hexokinase enzymatic method. ORs were directly estimated using logistic regression in four studies, whereas ORs from three studies were converted.
**Table 2**
| Domain | Scoring criteria | Gonoodi et al. 2018 | Shan et al. 2014 | Bo et al. 2004 | Bulka et al. 2019 | Li et al. 2019 | Obeid et al. 2008 | NHANES 2015–16 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Representativeness | 0 = nationally representative, 1 = representative at lower administrative level, 2 = not representative | 1 | 2 | 2 | 0 | 2 | 1 | 0 |
| Randomness of sample | 0 = Fully random, 1 = partially random, 2 = not random | 0 | 1 | 1 | 0 | 2 | 2 | 0 |
| Biomarker of exposure (i.e., zinc status) | 0 = Serum/Plasma Zinc, 1 = Other zinc biomarker, 2 = dietary zinc | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Measurement method of FPG | 0 = hexokinase enzymatic method, 1 = glucose oxidase or automatic biochemical analyzers or commercial kits | 1 | 1 | 1 | 0 | 1 | 1 | 0 |
| Measurement of association | 0 = Odds ratio directly calculated 1 = Dichotomous odds ratio converted from continuous or quantile odd/risk/hazard ratios | 0 | 1 | 1 | 1 | 0 | 0 | 0 |
| Risk score | | 2 | 5 | 5 | 1 | 5 | 4 | 0 |
| Risk category | 0–3 = low risk 4–6 = medium risk 7–9 = high risk | Low risk | Medium risk | Medium risk | Low risk | Medium risk | Medium risk | Low risk |
## Meta analyses on the relationship between ZD and FPG
Meta-analysis results are presented in Fig 3, and show a significant ($p \leq 0.01$, random effects model) inverse relationship between ZD and high FPG. This association was observed in five of the seven studies included in the meta-analysis. The inclusion of a data point based on primary data analysis (i.e., NHANES 2015–16 data) did not substantially impact the results; the model OR based only on the six studies obtained via the systematic review was similar (OR = 2.34; $95\%$ CI: 1.16, 4.72). Further, an examination of the standardized residuals of the fitted random-effects model showed that none of the included studies had statistically significant high residuals.
**Fig 3:** *Meta-analysis of the association between zinc deficiency and high fasting plasma glucose.*
## Population proportion of high FPG attributable to ZD
Using the calculated OR from the meta-analysis and estimated prevalence of high FPG, we determined each country’s PAR of high FPG to ZD. At the global level, the PAR of ZD’s contribution to high FPG was $6.7\%$. For GBD Super Regions, the lowest PARs were found in Sub-Saharan Africa ($2.4\%$). Higher PAR levels were found in South Asia ($6.4\%$), North Africa and Middle East ($6.5\%$), Southeast Asia, East Asia, and Oceania (SAEAO) ($6.7\%$), and Central Europe, Eastern Europe, and Central Asia (CEEECA) ($6.9\%$), and the highest PARs were found in Latin America and Caribbean ($7.7\%$), and the GBD High-Income Super Region ($10.5\%$). The PAR for all Super Regions and countries is provided in Table A in S1 Text.
## DALYs attributable to ZD
As shown in Table 3, high FPG contributes to the disease burden of 15 separate “level 3” causes by the GBD categorization. The largest share of the FPG disease burden is from “cardiovascular diseases” and “diabetes and chronic kidney diseases”, and these conditions account for more than $90\%$ of the total FPG DALYs. Accordingly, these conditions also account for the greatest share of the disease burden attributable to ZD.
**Table 3**
| Level 2 Causes | Level 3 Causes | 2019 DALYs† | 2019 DALYs†.1 | DALYs attributable to zinc deficiency | DALYs attributable to zinc deficiency.1 | DALYs attributable to zinc deficiency, per 100,000 | DALYs attributable to zinc deficiency, per 100,000.1 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | | DALYS | 95% CI | DALYS | 95% CI | DALYS | 95% CI |
| Neoplasms | | 8575244 | (2,355,428, 17,557,275) | 678242 | (102,065, 1,609,825) | 8.77 | (1.32, 20.81) |
| | Bladder cancer | 399383 | (81,547, 865,320) | 32388 | (4,074, 83,815) | 0.42 | (0.05, 1.08) |
| | Breast cancer | 1238619 | (237,680, 2,785,554) | 93679 | (13,147, 239,169) | 1.21 | (0.17, 3.09) |
| | Colon and rectum cancer | 1902208 | (453,351, 4,167,499) | 151246 | (20,987, 393,517) | 1.95 | (0.27, 5.09) |
| | Liver cancer | 99249 | (23,846, 217,971) | 7427 | (1,007, 19,474) | 0.10 | (0.01, .25) |
| | Ovarian cancer | 353538 | (68,469, 823,891) | 27483 | (3,704, 71,335) | 0.36 | (0.05, .92) |
| | Pancreatic cancer | 943775 | (221,091, 2,037,675) | 77165 | (10,155, 199,975) | 1.00 | (0.13, 2.58) |
| | Tracheal, bronchus, and lung cancer | 3638472 | (855,701, 8,008,314) | 288852 | (36,558, 739,907) | 3.73 | (0.47, 9.56) |
| Sense organ diseases | Sense organ diseases | 672727 | (159,423, 1,564,433) | 45527 | (5,924, 119,163) | 0.59 | (0.08, 1.54) |
| | Blindness and vision loss | 672727 | (159,423, 1,564,433) | 45527 | (5,924, 119,163) | 0.59 | (0.08, 1.54) |
| Neurological disorders | Neurological disorders | 2445832 | (413,738, 7,861,236) | 195891 | (8,788, 570,701) | 2.53 | (0.11, 7.38) |
| | Alzheimer’s disease & other dementias | 2445832 | (413,738, 7,861,236) | 195891 | (8,788, 570,701) | 2.53 | (0.11, 7.38) |
| Cardiovascular diseases | Cardiovascular diseases | 19002527 | (13,200,142, 27,561,958) | 1364478 | (211,992, 3,262,716) | 17.63 | (2.74, 42.17) |
| | Ischemic heart disease | 11103141 | (6,413,307, 18,357,813) | 807031 | (124,350, 1,956,200) | 10.43 | (1.61, 25.28) |
| | Peripheral artery disease | 429156 | (274,718, 675,697) | 35801 | (5,545, 88,248) | 0.46 | (0.07, 1.14) |
| | Stroke | 7470230 | (4,608,885, 12,049,312) | 521645 | (79,503, 1,266,180) | 6.74 | (1.03, 16.36) |
| Respiratory infections and tuberculosis | Respiratory infections and tuberculosis | 3800995 | (2,412,798, 5,199,038) | 212670 | (31,639, 525,132) | 2.75 | (0.41, 6.79) |
| | Tuberculosis | 3800995 | (2,412,798, 5,199,038) | 212670 | (31,639, 525,132) | 2.75 | (0.41, 6.79) |
| Diabetes and kidney diseases | Diabetes and kidney diseases | 82087721 | (70,100,149, 95,923,086) | 5738229 | (906,207, 13,461,173) | 74.16 | (11.71, 173.97) |
| | Diabetes mellitus | 69230727 | (58,236,244, 82,305,643) | 4847319 | (764,491, 11,388,030) | 62.65 | (9.88, 147.18) |
| | Chronic kidney disease | 12856993 | (10,875,354, 14,984,438) | 890910 | (142,148, 2,092,086) | 11.51 | (1.84, 27.04) |
| Total | | 116585045 | (97,195,911, 139,180,788) | 8235036 | (1,295,611, 19,182,407) | 106.43 | (16.74, 247.92) |
As shown in Table 4, there is considerable regional variation in the scale of the disease burden from high FPG, as well as the share of the burden attributable to ZD. South Asia, SAEAO, and High-Income countries account for more than $70\%$ of the high FPG disease burden and zinc-attributable disease burden globally. The population-standardized disease burden (i.e., DALYs per 100,000 people) in South Asia and SAEAO is lower than all regions apart from Sub-Saharan Africa, indicating that the large population size of these regions is a key driver of the total disease burden. The population-standardized disease burden is largest in High Income countries, Latin America and the Caribbean and in CEEECA.
**Table 4**
| GBD Super Regions | 2019 Population | DALYs from high FPG | DALYs from high FPG.1 | DALYs attributable to ZD | DALYs attributable to ZD.1 | DALYs attributable to ZD, per 100,000 | DALYs attributable to ZD, per 100,000.1 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | 2019 Population | DALYs | 95% CI | DALYs | 95% CI | DALYs | 95% CI |
| Central Europe, Eastern Europe, and Central Asia | 417725139 | 7139995 | (5,629,602, 9,005,307) | 496097 | (77,541, 1,182,410) | 118.76 | (18.56, 283.06) |
| High-Income Countries | 1083976063 | 20792148 | (16,007,450, 26,446,113) | 2174228 | (337,965, 4,947,321) | 200.58 | (31.18, 456.41) |
| Latin America and Caribbean | 584378201 | 11869848 | (10,084,837, 13,881,480) | 917273 | (145,722, 2,175,099) | 156.97 | (24.94, 372.21) |
| North Africa and Middle East | 608713645 | 8650929 | (7,491,033, 10,160,736) | 561053 | (88,773, 1,347,994) | 92.17 | (14.58, 221.45) |
| South Asia | 1805200296 | 26821314 | (22,672,025, 31,485,652) | 1711731 | (260,223, 4,087,413) | 94.82 | (14.42, 226.42) |
| Southeast Asia, East Asia, and Oceania | 2159261972 | 32630669 | (27,034,554, 39,566,269) | 2170135 | (339,165, 5,169,313) | 100.50 | (31.46, 479.44) |
| Sub-Saharan Africa | 1078209307 | 8680141 | (7,083,279, 10,374,583) | 204519 | (31,782, 520,484) | 18.97 | (2.95, 48.27) |
| Grand Total | 7737464623 | 116585045 | (97,195,911, 139,180,788) | 8235036 | (1,295,611, 19,182,407) | 106.43 | (16.74, 247.92) |
Fig 4 illustrates the geographic distribution of the disease burden from high FPG that is attributable to ZD at the country level. The three countries with the largest total number of DALYs attributable to ZD (Panel A) are India (~1.5 million), China (~1.3 million), and the United States of America (USA; ~1.0 million). Countries with high population-standardized DALYS (Panel B; DALYS/100000 people) were largely found in island nations or states: Niue [1368], Palau [1231], Fiji [1072], Mauritius [1021], Cook Islands [954], American Samoa [843], Trinidad & Tobago [811], Puerto Rico [724], Marshall Islands [674], the US Virgin Islands [634], and others. Among the three countries with the highest total DALYs, the USA had the highest number of DALYs per 100,000 [309], followed by India [107] and China [89]. The proportion of the high FPG disease burden attributable to ZD (Panel C) was highest among the island nations of Niue ($19.5\%$), American Samoa ($18.2\%$), Palau ($17.4\%$), Cook Islands ($16.4\%$), Puerto Rico ($16.3\%$), and others. ZD also accounted for a relatively high proportion of the disease burden in the USA and many Western European countries. ZD accounted for the lowest proportion of the disease burden in most countries in Sub-Saharan Africa, with the lowest PAR found in Niger ($1.3\%$).
**Fig 4:** *Country level non-communicable disease burden attributable to zinc deficiency: (A) total DALYs, (B) DALYs per 100,000 people, (C) Percent of high FPG disease burden attributable to zinc deficiency. (Map base layer: https://hub.arcgis.com/datasets/UIA::uia-world-countries-boundaries/about).*
## Disease burden attributable to ZD
To the authors’ knowledge, this is the first study to estimate the NCD burden attributable to ZD. Prior to our analysis, ZD’s contribution to the global disease burden, as estimated by the GBD project in 2019, has been estimated as a portion of the disease burden from diarrhea in children 1–4 years of age [43]. The GBD project previously included ZD as a risk factor of lower respiratory infection (LRI) [44], but ZD was not considered a risk factor for LRI in 2019 as multiple studies found that reductions in ZD only generated small decrease in LRI mortality [45]. Other studies also previously developed models to estimate the zinc-attributable disease burden from diarrhea, pneumonia, and stunting in children <5 years of age [19], demonstrating the evolution in knowledge of the association as more data become available.
Our results find that ZD accounts for a modest proportion of less than $10\%$ of the NCD burden from high FPG, globally. Nonetheless, due to the large total disease burden from NCDs caused by high FPG, the “FPG DALYs” attributable to ZD are considerably higher than the diarrhea-related DALYs attributable to ZD estimated by the GBD project [43]. To illustrate, the GBD project’s 2019 disease burden estimates show that globally, ZD accounts for $0.32\%$ of the DALYs from diarrheal diseases (diarrheal disease burden attributable to ZD = 258,813 DALYs; total diarrheal disease burden = 80,917,779) [38]. Our estimates of the FPG DALYs attributable to ZD are approximately 32 times greater. While a number of approaches can be used to value a DALY, one approach uses a standardized value of USD 1000 to present the DALY burden in terms of an economic cost that can be considered in cost-benefit analyses considering various interventions [19]. Using this standard value, the economic cost of the ZD-associated NCD burden would be currently estimated at USD 8.2 billion.
More than $90\%$ of the FPG DALYs—and subsequently the FPG DALYs attributable to ZD—stem from four disease outcomes: diabetes mellitus, ischemic heart disease, stroke, and chronic kidney disease. The global disease burden from each of these disease outcomes has increased over the past 30 years, however, these disease outcomes have not increased in all regions. To illustrate, in industrialized countries, the disease burden from type 2 diabetes has steadily increased since 1990 and is predicted to continue increasing into the future [46]. In contrast, the disease burden from ischemic heart disease has declined in High-Income countries since 1990, and global increases were driven by an increasing disease burden in other regions, with the biggest increases found in South Asia and in SAEAO [38], which are the two most populous super-regions and have experienced a significant rise in the mortality rate from ischemic heart disease [47]. The disease burdens due to stroke and chronic kidney disease have also increased since 1990, with the majority of the increase in the burden from stroke observed in SAEAO [38], and similar increases in the total burden from chronic kidney disease in both South Asia and SAEAO. The only decreases in the four outcomes between 1990 and 2019 occurred in High-Income countries, where the disease burdens from ischemic heart disease and stroke decreased, and in CEEECA, where the stroke disease burden decreased.
While the disease burden is highest among populous countries (e.g., China, India, USA), the population-standardized DALYs are highest among island nations, particularly island nations in the South Pacific and Caribbean. High levels of NCDs in these regions have been previously identified. A pooled analysis of 751 population-based studies observed that Micronesia and Polynesia experienced the largest increase in the prevalence of diabetes between 1980 and 2014, with the highest prevalence in 2014 found in American Samoa [48]. Increases in the prevalence of diabetes have also been observed in Caribbean nations [49]. While these nations, which have small populations, do not contribute substantially to the total disease burden, the intra-country economic effects of the high prevalence of NCDs are well documented. The costs of treatment, hospitalization, permanent disability, premature death, and other factors all add to the economic toll caused by a high prevalence of NCDs [50].
## Zinc and the “double burden” of malnutrition
ZD alongside the sizable disease burden from NCDs denotes a “double burden” of malnutrition, where under- and over-nutrition occurs in the same communities and/or individuals [51]. However, while a “double burden” simply implies co-occurrence of measures of under- and over-nutrition [52], ZD appears to directly contribute to the disease outcomes (e.g., diabetes, heart disease) typically associated with over-nutrition outcomes, such as overweight and obesity. Thus, a “double burden” of malnutrition when ZD is the indicator of under-nutrition characterizes a situation whereby a portion of the over-nutrition disease burden could be reduced by programs that increase the intake of zinc and decrease the prevalence of ZD.
The features of this “double burden” vary by geographic region due to industrialization and dietary factors. Using data from national food balance sheets, Wessells and Brown [53] estimated that $17\%$ of the global population was at risk of insufficient zinc intake, with prevalences <$15\%$ in most countries in North America and Western Europe, and prevalences $15\%$-$25\%$ or >$25\%$ in most countries on the African continent and in South and Southeast Asia. These estimates suggest that zinc interventions could be implemented in nearly all regions, including low- and middle-income countries. Though the disease burden from high FPG is lowest in Sub-Saharan Africa, many countries in this region are undergoing a “nutrition transition” where there is an increasing prevalence of conditions associated with over-nutrition and/or poor diets [2]. A similar transition is taking place in South Asia and SAEAO, and large populations in tandem with sizable FPG disease burden per capita suggest that zinc interventions would be appropriate. In addition, zinc interventions are also appropriate in High-Income countries, as the even modest reductions in ZD prevalence could translate to substantial reductions in the disease burden from high FPG.
## Zinc-related public health programs
The three most-common public health measures used to increase zinc intake are supplementation, industrial fortification, and biofortification. Zinc supplementation has been found to both reduce the incidence of diarrhea in children <5 years of age [54] and improve the indicators related to diabetes and cardiovascular diseases (e.g., FPG, HbA1c, insulin resistance, and LDL cholesterol) [7]. Zinc supplementation of pregnant women with low zinc status has also been shown to significantly increase the birth weight and head circumference of the offspring [55]. At present, zinc supplements are frequently administered to manage acute diarrhea in children, and the WHO recommends that children receive zinc supplements (<6 months, 10mg; 6–59 months, 20mg) for 10–14 days after a diarrheal episode [56]. Mass zinc supplementation of adults is currently undertaken as part of national prenatal programs that deliver multiple micronutrient supplements to pregnant women. Some countries have replaced iron-folic acid supplements targeted to pregnant women with multiple micronutrient supplements, as multiple micronutrient supplements have been found to lower the risk of certain birth outcomes, such as low birth weight, small-for-gestational age, and pre-term birth [57].
The addition of zinc to industrially milled cereal grains (e.g., wheat, maize, rice) is mandated in 34 countries located in Sub-Saharan Africa ($$n = 16$$), SAEAO ($$n = 6$$), Latin America and Caribbean ($$n = 6$$), CEEECA ($$n = 3$$), and North Africa and Middle East ($$n = 3$$) [58]. A recent systematic review and meta-analysis of efficacy and effectiveness studies found that that fortification with zinc and other micronutrients was associated with improved plasma zinc concentrations and increased weight gain in children [59]. Though relatively few countries fortify with zinc, there are an additional 57 countries that mandate the fortification of wheat, maize, or rice, but do not mandate the inclusion of zinc. Changes to the fortification standards in these countries could be a possible approach for governments to reduce ZD and the disease burden from NCDs at relatively low additional cost. While these findings suggest that mass fortification may be a feasible approach to delivering zinc, studies examining the coverage of fortified foods have identified issues related to compliance, resulting in a lower-than-anticipated coverage [60,61].
Zinc-biofortified crops offer another option to increase the intake of zinc in populations experiencing a high prevalence of deficiency. There are several studies with women and children that demonstrate higher total absorbed zinc from biofortified crops (e.g., pearl millet, rice, wheat, maize). However, to date there are only two published intervention trials examining the efficacy of a zinc biofortified crop. Sazawal et al. [ 62] and Jongstra et al. [ 63] conducted a double masked randomized, controlled trials in India and Bangladesh, respectively. Sazawal et al. [ 62] examined the impact of six-month consumption of biofortified wheat flour on children and mothers, and found no significant treatment effects on mean plasma zinc at endline or changes from baseline to endline. Both mothers and children consuming zinc-biofortified wheat did, however, report significantly less morbidity; mothers reported significantly fewer days with fever, and children reported significantly fewer days with pneumonia and vomiting. Jongstra et al. [ 63] examined the impact of nine-months of zinc-biofortified rice consumption in children. The researchers also found no significant changes in plasma zinc concentration, but in contrast to Sazawal et al. [ 62], found that children consuming biofortified rice had higher levels of morbidity from respiratory infections while also experiencing higher linear growth during the intervention period.
Due to the differing programmatic landscapes in industrialized and developing countries, policy makers in each country must determine the type of program(s) that best suit their population’s consumption patterns. As the total and population-standardized FPG disease burden attributable to ZD is highest among high-income countries, it is warranted to examine the feasibility of zinc interventions for these countries, particularly among elderly populations where the prevalence of ZD has been shown to be higher than the general population [64].
As the proportion of the NCD burden attributable to ZD is relatively small, programs that increase zinc intake should be implemented as part of national public health policies and programs that aim to improve a populations diet and reduce other risk factors, such as obesity, sedentarism, and hypertension.
## Strengths and limitations
The approach taken by the research team has notable strengths. As this approach utilizes a modified version of the FPG disease burden estimated annually by the GBD project, the results are truly global and would enable policymakers to identify countries where zinc interventions could be implemented to reduce the FPG disease burden attributable to ZD. Furthermore, as more data become available, the meta-analysis could be readily updated and new estimates produced.
The meta-analysis component of our study was limited by the identification of a small number of data points. This was due to the fact that few studies that measure both zinc and FPG concentrations calculated the dichotomous association required for our meta-analysis or calculated a metric that could be converted to a dichotomous OR. Furthermore, no data included in the meta-analysis were present from countries in Latin America, Sub-Saharan Africa, and Central Asia, or low-income countries in any region. While the results of our study suggest a biological association between zinc and glucose status, data from a wider array of countries would be more ideal since the meta-analysis’ OR is multiplied by national and regional estimates of the proportion of high FPG to calculate the PAR. The meta-analysis included several studies with medium risk, which would affect the accuracy of the estimation of the disease burden attributable to ZD. An obvious concern that most of the studies included in the meta-analysis were from non-representative samples with non-randomly selected participants. Future studies on this topic should be representative and should randomly select participants to provide a more accurate estimation of the strength of the relationship between ZD and high FPG. The meta-analysis portion of our study is also limited by the variation in the covariates used by each study’s regression model. Our study is also limited by the poor accuracy and precision of serum/plasma zinc. While there are distinct challenges to measuring ZD, population-level ZD is often defined as serum/plasma zinc concentrations that are below set thresholds, and serum/plasma zinc is often measured as part of population-based micronutrient and nutrition surveys [34]. Until improved biomarkers of zinc status are developed and verified, the associations between zinc status defined by serum/ plasma zinc levels and risk factors of key NCDs will have to be interpreted with caution.
## Conclusion
Via a systematic review and meta-analysis, our study found a significant association between ZD and high FPG. Calculations using the OR from the meta-regression and estimates of the disease burden from high FPG show that ZD accounts for a modest proportion of the DALYs from high FPG. While ZD accounts for a small share of the high FPG disease burden, the total number of DALYs far surpasses other estimates of the disease burden attributable to ZD, which focus on diarrheal diseases in childhood. Our results show that the disease burden from ZD is largest in High-Income countries, but smaller countries, particularly island nations, can have the highest per-capita disease burden.
## References
1. **NCD Countdown 2030: worldwide trends in non-communicable disease mortality and progress towards Sustainable Development Goal target 3.4**. *Lancet (London, England).* (2018.0) **392** 1072-1088. DOI: 10.1016/S0140-6736(18)31992-5
2. Gouda HN, Charlson F, Sorsdahl K, Ahmadzada S, Ferrari AJ, Erskine H. **Burden of non-communicable diseases in sub-Saharan Africa, 1990–2017: results from the Global Burden of Disease Study 2017**. *Lancet Glob Heal.* (2019.0). DOI: 10.1016/S2214-109X(19)30374-2
3. **Global Burden of Disease Project 2017**. *In: GBD Compare Viz Hub [Internet]* (2020.0)
4. Olechnowicz J, Tinkov A, Skalny A, Suliburska J. **Zinc status is associated with inflammation, oxidative stress, lipid, and glucose metabolism**. *J Physiol Sci* (2018.0) **68** 19-31. DOI: 10.1007/s12576-017-0571-7
5. Wang X, Wu W, Zheng W, Fang X, Chen L, Rink L. **Zinc supplementation improves glycemic control for diabetes prevention and management: a systematic review and meta-analysis of randomized controlled trials**. *Am J Clin Nutr* (2019.0). DOI: 10.1093/ajcn/nqz041
6. Ranasinghe P, Wathurapatha WS, Ishara MH, Jayawardana R, Galappatthy P, Katulanda P. **Effects of Zinc supplementation on serum lipids: A systematic review and meta-analysis.**. *Nutrition and Metabolism* (2015.0). DOI: 10.1186/s12986-015-0023-4
7. Pompano LM, Boy E. **Effects of Dose and Duration of Zinc Interventions on Risk Factors for Type 2 Diabetes and Cardiovascular Disease: A Systematic Review and Meta-Analysis.**. *Adv Nutr* (2020.0). DOI: 10.1093/advances/nmaa087
8. Jafarnejad S, Mahboobi S, McFarland L V. **Meta-analysis: Effects of zinc supplementation alone or with multi-nutrients, on glucose control and lipid levels in patients with type 2 diabetes**. *Prev Nutr Food Sci* (2019.0). DOI: 10.3746/pnf.2019.24.1.8
9. Khazdouz M, Djalalinia S, Sarrafi Zadeh S, Hasani M, Shidfar F, Ataie-Jafari A. **Effects of Zinc Supplementation on Cardiometabolic Risk Factors: a Systematic Review and Meta-analysis of Randomized Controlled Trials**. *Biol Trace Elem Res* (2020.0). DOI: 10.1007/s12011-019-01870-9
10. Capdor J, Foster M, Petocz P, Samman S. **Zinc and glycemic control: A meta-analysis of randomised placebo controlled supplementation trials in humans**. *J Trace Elem Med Biol* (2013.0). DOI: 10.1016/j.jtemb.2012.08.001
11. Jayawardena R, Ranasinghe P, Galappatthy P, Malkanthi RLDK, Constantine GR, Katulanda P. **Effects of zinc supplementation on diabetes mellitus: A systematic review and meta-analysis**. *Diabetology and Metabolic Syndrome* (2012.0). DOI: 10.1186/1758-5996-4-13
12. Wu J, He H, Zhang Q, Zhang Y. **Fasting blood glucose was linearly associated with colorectal cancer risk in the population without self-reported diabetes mellitus history**. *Medicine (Baltimore).* (2021.0) **100** e26974. DOI: 10.1097/MD.0000000000026974
13. Fernández-Cao JC, Warthon-Medina M, Moran VH, Arija V, Doepking C, Serra-Majem L. **Zinc intake and status and risk of type 2 diabetes mellitus: A systematic review and meta-analysis**. *Nutrients* (2019.0). DOI: 10.3390/nu11051027
14. Yu X, Huang L, Zhao J, Wang Z, Yao W, Wu X. **The Relationship between Serum Zinc Level and Heart Failure: A Meta-Analysis.**. *BioMed Research International* (2018.0). DOI: 10.1155/2018/2739014
15. Hambidge KM. **Zinc and diarrhea**. *Acta Paediatr Suppl* (1992.0) **381** 82-6. DOI: 10.1111/j.1651-2227.1992.tb12377.x
16. Fischer Walker C, Black RE. **Zinc and the risk for infectious disease**. *Annu Rev Nutr* (2004.0) **24** 255-75. DOI: 10.1146/annurev.nutr.23.011702.073054
17. Ploysangam A, Falciglia GA, Brehm BJ. **Effect of marginal zinc deficiency on human growth and development**. *J Trop Pediatr* (1997.0) **43** 192-8. DOI: 10.1093/tropej/43.4.192
18. Fischer Walker CL, Ezzati M, Black RE. **Global and regional child mortality and burden of disease attributable to zinc deficiency**. *Eur J Clin Nutr* (2009.0). DOI: 10.1038/ejcn.2008.9
19. Stein AJ, Meenakshi JV, Qaim M, Nestel P, Sachdev HPS, Bhutta ZA. **Analyzing the health benefits of biofortified staple crops by means of the Disability-Adjusted Life Years approach: a handbook focusing on iron, zinc and vitamin A.**. *HarvestPlus Technical Monograph* (2005.0)
20. Stein AJ, Nestel P, Meenakshi JV, Qaim M, Sachdev HPS, Bhutta ZA. **Plant breeding to control zinc deficiency in India: how cost-effective is biofortification?**. *Public Health Nutr* (2007.0) **10** 492-501. DOI: 10.1017/S1368980007223857
21. Brown KH, Hess SY, Vosti SA, Baker SK. **Comparison of the estimated cost-effectiveness of preventive and therapeutic zinc supplementation strategies for reducing child morbidity and mortality in sub-Saharan Africa**. *Food Nutr Bull* (2013.0) **34** 199-214. DOI: 10.1177/156482651303400209
22. King JC, Brown KH, Gibson RS, Krebs NF, Lowe NM, Siekmann JH. **Biomarkers of nutrition for development (BOND)-Zinc Review.**. *Journal of Nutrition* (2016.0). DOI: 10.3945/jn.115.220079
23. Viera AJ, Garrett JM. **Understanding Interobserver Agreement: The Kappa Statistic.**. *Fam Med* (2005.0) **37** 36-3. PMID: 15619154
24. Bulka CM, Persky VW, Daviglus ML, Durazo-Arvizu RA, Argos M. **Multiple metal exposures and metabolic syndrome: A cross-sectional analysis of the National Health and Nutrition Examination Survey 2011–2014**. *Environ Res* (2019.0) **168** 397-405. DOI: 10.1016/j.envres.2018.10.022
25. Kweon S, Kim Y, Jang M, Kim Y, Kim K, Choi S. **Data resource profile: the Korea National Health and Nutrition Examination Survey (KNHANES).**. *Int J Epidemiol* (2014.0) **43** 69-77. DOI: 10.1093/ije/dyt228
26. Wells GA, Shea B, O’Connell D al, Peterson J, Welch V, Losos M. **The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses**. *Oxford* (2000.0)
27. Gonoodi K, Moslem A, Darroudi S, Ahmadnezhad M, Mazloum Z, Tayefi M. **Serum and dietary zinc and copper in Iranian girls**. *Clin Biochem. United States* (2018.0) **54** 25-31. DOI: 10.1016/j.clinbiochem.2018.02.006
28. Bo S, Lezo A, Menato G, Gallo M-L, Bardelli C, Signorile A. **Gestational hyperglycemia, zinc, selenium, and antioxidant vitamins.**. *Nutrition. United States* (2005.0) **21** 186-191. DOI: 10.1016/j.nut.2004.05.022
29. Obeid O, Elfakhani M, Hlais S, Iskandar M, Batal M, Mouneimne Y. **Plasma copper, zinc, and selenium levels and correlates with metabolic syndrome components of lebanese adults**. *Biol Trace Elem Res* (2008.0) **123** 58-65. DOI: 10.1007/s12011-008-8112-0
30. Shan Z, Bao W, Zhang Y, Rong Y, Wang X, Jin Y. **Interactions between zinc transporter-8 gene (SLC30A8) and plasma zinc concentrations for impaired glucose regulation and type 2 diabetes**. *Diabetes. United States* (2014.0) **63** 1796-1803. DOI: 10.2337/db13-0606
31. Li Z, Xu Y, Huang Z, Wei Y, Hou J, Long T. **Association between exposure to arsenic, nickel, cadmium, selenium, and zinc and fasting blood glucose levels**. *Environ Pollut. England* (2019.0) **255** 113325. DOI: 10.1016/j.envpol.2019.113325
32. 32NHANES. Copper, Selenium & Zinc—Serum (CUSEZN_I). In: 2015–2016 Data Documentation, Codebook, and Frequencies [Internet]. 2018 [cited 15 Jun 2021]. Available: https://wwwn.cdc.gov/Nchs/Nhanes/2015-2016/CUSEZN_I.htm.
33. 33NHANES. Plasma Fasting Glucose (GLU_I). In: 2015–2016 Data Documentation, Codebook, and Frequencies [Internet]. 2018 [cited 15 Jun 2021]. Available: https://wwwn.cdc.gov/Nchs/Nhanes/2015-2016/GLU_I.htm.
34. 34IZiNCG. Assessing population zinc status with serum zinc concentration. 2007.. *Assessing population zinc status with serum zinc concentration* (2007.0)
35. 35World Health Organization. Definition and diagnosis of diabetes mellitus and intermediate hyperglycaemia: report of a WHO/IDF consultation. Geneva, Switzerland; 2006.. *Definition and diagnosis of diabetes mellitus and intermediate hyperglycaemia: report of a WHO/IDF consultation.* (2006.0)
36. Mirel LB, Mohadjer LK, Dohrmann SM, Clark J, Burt VL, Johnson CL. **National Health and Nutrition Examination Survey: estimation procedures, 2007–2010.**. *Vital Health Stat 2* (2013.0) 1-17
37. Murray CJL, Aravkin AY, Zheng P, Abbafati C, Abbas KM, Abbasi-Kangevari M. **Global burden of 87 risk factors in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019**. *Lancet. Elsevier* (2020.0) **396** 1223-1249. DOI: 10.1016/S0140-6736(20)30752-2
38. **Global Health Data Exchange [Internet].**. (2021.0)
39. **Global Burden of Disease Study 2019 (GBD 2019) Population Estimates 1950–2019 [Internet]. Seattle, United States of America**. (2020.0). DOI: 10.6069/7EGY-0354
40. 40IHME. GBD 2019 Codebook. In: Global Health Data Exchange (GHDx) [Internet]. 2019 [cited 28 Nov 2021]. Available: http://ghdx.healthdata.org/sites/default/files/ihme_query_tool/IHME_GBD_2019_CODEBOOK.zip.
41. Vidović B, Dorđević B, Milovanović S, Škrivanj S, Pavlović Z, Stefanović A. **Selenium, zinc, and copper plasma levels in patients with schizophrenia: relationship with metabolic risk factors**. *Biol Trace Elem Res* (2013.0) **156** 22-28. DOI: 10.1007/s12011-013-9842-1
42. 42U.S. Centers for Disease Control and Prevention. NHANES 2015–2016 Laboratory Data. In: Website—National Health and Nutrition Examination Survey [Internet]. 2020 [cited 2 Dec 2020]. Available: https://wwwn.cdc.gov/Nchs/Nhanes/Search/DataPage.aspx?Component=Laboratory&CycleBeginYear=2015.
43. 43IHME. Zinc deficiency—Level 3 risk. In: Global Burden of Disease 2019 [Internet]. 2021 [cited 15 Nov 2021]. Available: http://www.healthdata.org/results/gbd_summaries/2019/zinc-deficiency-level-3-risk.
44. **Estimates of the global, regional, and national morbidity, mortality, and aetiologies of lower respiratory infections in 195 countries, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016**. *Lancet Infect Dis* (2018.0) **18** 1191-1210. DOI: 10.1016/S1473-3099(18)30310-4
45. Brown N, Kukka AJ, Mårtensson A. **Efficacy of zinc as adjunctive pneumonia treatment in children aged 2 to 60 months in low-income and middle-income countries: a systematic review and meta-analysis**. *BMJ Paediatr open* (2020.0) **4** e000662. DOI: 10.1136/bmjpo-2020-000662
46. Khan MAB, Hashim MJ, King JK, Govender RD, Mustafa H, Al Kaabi J. **Epidemiology of type 2 diabetes–global burden of disease and forecasted trends.**. *J Epidemiol Glob Health.* (2020.0) **10** 107. DOI: 10.2991/jegh.k.191028.001
47. Moran AE, Forouzanfar MH, Roth GA, Mensah GA, Ezzati M, Murray CJL. **Temporal trends in ischemic heart disease mortality in 21 world regions, 1980 to 2010: the Global Burden of Disease 2010 study**. *Circulation* (2014.0) **129** 1483-92. DOI: 10.1161/CIRCULATIONAHA.113.004042
48. **Worldwide trends in diabetes since 1980: a pooled analysis of 751 population-based studies with 4.4 million participants**. *Lancet (London, England).* (2016.0) **387** 1513-1530. DOI: 10.1016/S0140-6736(16)00618-8
49. Gallardo-Rincón H, Cantoral A, Arrieta A, Espinal C, Magnus MH, Palacios C. **Review: Type 2 diabetes in Latin America and the Caribbean: Regional and country comparison on prevalence, trends, costs and expanded prevention**. *Prim Care Diabetes* (2021.0) **15** 352-359. DOI: 10.1016/j.pcd.2020.10.001
50. Barcelo A, Arredondo A, Gordillo-Tobar A, Segovia J, Qiang A. **The cost of diabetes in Latin America and the Caribbean in 2015: Evidence for decision and policy makers**. *J Glob Health* (2017.0) **7** 020410. DOI: 10.7189/jogh.07.020410
51. Prentice AM. **The emerging epidemic of obesity in developing countries**. *International Journal of Epidemiology* (2006.0) 93-99. DOI: 10.1093/ije/dyi272
52. 52World Health Organization. The double burden of malnutrition. Policy brief. World Heal Organ. 2017; 1–12. Available: http://apps.who.int/iris/bitstream/10665/255413/1/WHO-NMH-NHD-17.3-eng.pdf?ua=1%0Ahttp://www.who.int/nutrition/publications/doubleburdenmalnutrition-policybrief/en/.. *The double burden of malnutrition. Policy brief. World Heal Organ* (2017.0) 1-12
53. Wessells KR, Brown KH. **Estimating the Global Prevalence of Zinc Deficiency: Results Based on Zinc Availability in National Food Supplies and the Prevalence of Stunting**. *PLoS One* (2012.0) 7. DOI: 10.1371/journal.pone.0050568
54. Tam E, Keats EC, Rind F, Das JK, Bhutta AZA. **Micronutrient Supplementation and Fortification Interventions on Health and Development Outcomes among Children Under-Five in Low- and Middle-Income Countries: A Systematic Review and Meta-Analysis.**. *Nutrients* (2020.0) 12. DOI: 10.3390/nu12020289
55. Goldenberg RL, Tamura T, Neggers Y, Copper RL, Johnston KE, DuBard MB. **The effect of zinc supplementation on pregnancy outcome**. *JAMA* (1995.0) **274** 463-8. DOI: 10.1001/jama.1995.03530060037030
56. 56WHO. Zinc supplementation in the management of diarrhoea. In: e-Library of Evidence for Nutrition Actions (eLENA) [Internet]. 2019 [cited 1 Nov 2021]. Available: https://www.who.int/elena/titles/zinc_diarrhoea/en/.
57. Keats EC, Haider BA, Tam E, Bhutta ZA. **Multiple-micronutrient supplementation for women during pregnancy.**. *Cochrane database Syst Rev* (2019.0) **3** CD004905. DOI: 10.1002/14651858.CD004905.pub6
58. 58Global Fortification Data Exchange: GFDx. Map nutrient levels in fortification standards. In: GFDx Website [Internet]. 2022 [cited 9 Feb 2022]. Available: https://fortificationdata.org/map-nutrient-levels-in-fortification-standards/.
59. Tsang BL, Holsted E, McDonald CM, Brown KH, Black R, Mbuya MNN. **Effects of Foods Fortified with Zinc, Alone or Cofortified with Multiple Micronutrients, on Health and Functional Outcomes: A Systematic Review and Meta-Analysis**. *Adv Nutr* (2021.0) **12** 1821-1837. DOI: 10.1093/advances/nmab065
60. Aaron GJ, Friesen VM, Jungjohann S, Garrett GS, Neufeld LM, Myatt M. **Coverage of Large-Scale Food Fortification of Edible Oil, Wheat Flour, and Maize Flour Varies Greatly by Vehicle and Country but Is Consistently Lower among the Most Vulnerable: Results from Coverage Surveys in 8 Countries.**. *J Nutr* (2017.0) **147** 984S-994S. DOI: 10.3945/jn.116.245753
61. Friesen VM, Aaron GJ, Myatt M, Neufeld LM. **Assessing Coverage of Population-Based and Targeted Fortification Programs with the Use of the Fortification Assessment Coverage Toolkit (FACT): Background, Toolkit Development, and Supplement Overview**. *J Nutr* (2017.0) **147** 981S-983S. DOI: 10.3945/jn.116.242842
62. Sazawal S, Dhingra U, Dhingra P, Dutta A, Deb S, Kumar J. **Efficacy of high zinc biofortified wheat in improvement of micronutrient status, and prevention of morbidity among preschool children and women—A double masked, randomized, controlled trial 11 Medical and Health Sciences 1117 Public Health and Health Se.**. *Nutr J* (2018.0). DOI: 10.1186/s12937-018-0391-5
63. Jongstra R, Hossain MM, Galetti V, Hall AG, Holt RR, Cercamondi CI. **The effect of zinc-biofortified rice on zinc status of Bangladeshi pre-school children: a randomized, double-masked, household-based controlled trial**. *Am J Clin Nutr* (2021.0). DOI: 10.1093/ajcn/nqab379
64. Yasuda H, Tsutsui T. **Infants and elderlies are susceptible to zinc deficiency**. *Sci Rep.* (2016.0) **6** 21850. DOI: 10.1038/srep21850
65. García OP, Ronquillo D, del Carmen Caamaño M, Martínez G, Camacho M, López V. **Zinc, iron and vitamins A, C and E are associated with obesity, inflammation, lipid profile and insulin resistance in Mexican school-aged children.**. *Nutrients* (2013.0) **5** 5012-5030. DOI: 10.3390/nu5125012
66. Onat T, Demir Caltekin M, Turksoy VA, Baser E, Aydogan Kirmizi D, Kara M. **The Relationship Between Heavy Metal Exposure, Trace Element Level, and Monocyte to HDL Cholesterol Ratio with Gestational Diabetes Mellitus**. *Biol Trace Elem Res* (2021.0) **199** 1306-1315. DOI: 10.1007/s12011-020-02499-9
|
---
title: 'Addressing TB multimorbidity in policy and practice: An exploratory survey
of TB providers in 27 high-TB burden countries'
authors:
- Alexander Jarde
- Noemia Siqueira
- Saima Afaq
- Farah Naz
- Muhammad Irfan
- Pervaiz Tufail
- Faiza Aslam
- Olamide Todowede
- Shagoofa Rakhshanda
- Humaira Khalid
- Yan Lin
- Olivia Bierman
- Asma Elsony
- Helen Elsey
- Najma Siddiqi
- Kamran Siddiqi
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022227
doi: 10.1371/journal.pgph.0001205
license: CC BY 4.0
---
# Addressing TB multimorbidity in policy and practice: An exploratory survey of TB providers in 27 high-TB burden countries
## Abstract
In people with TB, co-existence of long-term conditions (e.g., depression, diabetes and HIV) and risk factors (e.g.,alcohol misuse, malnutrition, and smoking) are associated with increased mortality and poor treatment outcomes including delayed recovery, TB treatment failure and relapse. However, it is unclear as to what extent these comorbidities are addressed in TB policy and practice. Between August and October 2021, we conducted an online cross-sectional survey in high-TB burden countries. We recruited a purposive sample of TB health workers, managers, policy makers, advisors and advocates from these countries. The survey enquired about the extent to which various comorbid conditions are: (a) mentioned in TB policies, plans, and guidelines; (b) screened, diagnosed, treated or referred to specialist services by TB healthcare workers. We summarised using descriptive analysis. Of the 1100 potential respondents contacted in 33 countries, 543 responded but only 446 ($41\%$) from 27 countries provided sufficient data for inclusion in the study. We found no notable differences between these providing insufficient data and those completing the survey. HIV, diabetes mellitus, depression and tobacco and alcohol use disorders were identified as the most common and concerning comorbid conditions in TB. HIV was screened for and managed by TB services in most countries. Screening for diabetes and/or tobacco and alcohol use disorders was offered by almost half of all TB services but only a few offered relevant treatments. Depression was rarely screened for, almost never treated, and only infrequently referred to specialist services. Most respondents felt confident in screening/diagnosing these comorbid conditions but not in treating these conditions. With the exception of HIV, chronic comorbid conditions are only partially screened for and rarely managed within TB services. Mental health conditions are for the most part neglected. Given their adverse impact on TB outcomes, integrating screening and management of these comorbidities within TB programmes offers a significant opportunity to meet TB targets, address non-communicable diseases and improve patient well-being.
## Introduction
In 2020, tuberculosis (TB) caused more deaths worldwide (1.5 million) than any other infectious disease [1]. A range of other long-term conditions (e.g., depression, diabetes and HIV) and risk factors (e.g., alcohol misuse, malnutrition, and smoking) that commonly co-occur with TB are associated with increased mortality amongst the TB patients and these conditions and risk factors may also worsen its clinical course, delay recovery and lead to TB treatment failure and relapse [2]. TB multimorbidity, defined as the co-occurrence of TB and one or more chronic conditions in a single individual at one point in time, is also a growing global concern [2]. There is some (albeit limited) information available on the patterns and risks of multimorbidity compared to the general population, and on the burden of disease associated with TB multimorbidity. According to an analysis of the World Health Survey in 48 low- and middle-Income countries (LMIC), the prevalence of one or more non-communicable diseases was almost twice as high among those with TB than those without ($68.8\%$ vs. $34.4\%$) [3]. A number of meta-analyses have estimated the prevalence of various comorbidities; about half of TB patients in LMIC have depression [4], about one third in Sub-Saharan Africa have HIV [5], and worldwide, around $15\%$ have diabetes [6]. Furthermore, TB multimorbidity is associated with increased symptom complexity, healthcare use and costs, and significantly worse treatment outcomes (odds ratio of 1.98 [$95\%$CI, 1.56–2.52]), including increased mortality (relative risk upto 4 [$95\%$CI, 3.3–4.6]), compared with TB alone [3, 7–11]. An analysis of the world health survey in 48 LMICs showed that at least one third of the the YLDs in those who had TB multimorbidity were attributable to non communicable diseases [3]. There is a growing need to address TB multimorbidity, and to offer patient-centred, integrated care for multiple conditions [12, 13]. However, it is unclear to what extent current TB services identify and manage these other mental and physical comorbidities in patients with presumptive and/or confirmed TB. While the WHO and STOP TB partnership recognises the significant impact of comorbid conditions on TB outcomes [14], the extent to which these are detected and managed is not reported even for high TB-burden countries. Our aim was to understand whether TB multimorbidity is currently considered and addressed in high-TB burden countries [15]. Our key objectives were to, overall: 1) understand the extent to which TB multimorbidity was considered in policy, management and services within TB healthcare; 2) assess which specific comorbidities were being considered; and 3) identify potential gaps in service provision for TB multimorbidity.
## Methods
We conducted a cross-sectional multi-country online survey between August and October 2021. This study was approved by the Research Governance Committee of the University of York.
## Inclusion and exclusion criteria
The survey was conducted in 33 countries; 30 were classified as high-TB burden countries by the WHO for 2021–2025, while the other three (Cambodia, Russia and Zimbabwe) were part of the previous list but removed in this update [15]. Our key informants were TB health workers, managers, policy makers, advisors and employees of TB advocacy organisations in these countries. For TB health workers, we included people based in primary, secondary and tertiary care TB services, in either the public or the private (for profit or not for profit) sector, in a high-TB burden country. Respondents could be in a management, clinical or advisory/advocacy role. In addition, TB programme managers, supervisors and coordinators at the national, provincial, regional and district level were also included.
## Sampling strategy
Given the exploratory nature of this study, a formal sample size was not estimated; instead, we aimed to recruit as many respondents as possible from as many countries as possible in the time period available for data collection (August-October 2021).
We used a variety of strategies to identify the widest range of potential respondents including using targeted emails to people identified through online searching e.g., Stop TB Partnership Partners’ Directory (http://www.stoptb.org/partners/). We asked members of the Tuberculosis Multimorbidity Network (https://www.impactsouthasia.com/tbmm/), representatives of The Union and the World Health Organization, researchers in the field and TB programme managers for potential key informants. We also included social platforms (Twitter) to publicise the survey widely.
We recruited a purposive sample of key respondents in management, advisory, and clinical positions at different levels of the public and private systems offering TB services in each high-burden country.
## Informed consent
An Informed consent was obtained from all participants.
## Ethics statement
Ethical Approval was obtained from Health Sciences Research Governance Committee, Department of Health Sciences, University of York (YO10 5DD), United Kingdom having approval number: HSRGC/$\frac{2021}{458}$/A.
## Data collection
We invited potential respondents to participate in the survey via email. In addition to the link to the survey, the invite included information about the survey, rights of the participants and arrangements for data confidentiality and security. Engaging with the survey was taken as consent for participation. A reminder email was sent halfway through the data collection period.
The online survey (developed using Qualtrics) included the following sections: [1] respondent’s characteristics, including their country, organisation and role (but no personal information such as age, gender, or socioeconomic status); [2] relevant policies, strategic plans, and clinical guidelines in their country (e.g., whether such documents existed, what comorbidities were mentioned in them and what were clinicians asked to do); [3] how comorbid conditions were considered in respective TB services/clinics (e.g., which comorbidities were diagnosed, screened for, and treatment offered); and [4] what were the three most common and concerning comorbidities in their experience and how capable they felt in diagnosing or treating them (see S1 Appendix- ‘TBMM survey preview’ in the online supplementary materials). The survey was available in English, French and Portuguese.
## Data analysis
We excluded any incomplete surveys where no responses were provided for sections 2 to 4 (i.e., only characteristics of the respondent were provided). We summarised quantitative and qualitative responses using descriptive analyses. Given the heterogeneity in the number and type of respondents for each country, the non-random, non-representative sampling procedure, and the small sample size, we decided against pooling responses from all countries together. The main analysis only included countries with ≥10 responses to avoid findings being skewed due to too few responses. Results including all countries are provided in S2 Appendix—‘TBMM survey all country results’ in online supplementary materials [1].
For items referring to policy documents and clinical guidelines relevant to wider geographical regions (provinces or countries), we made the following assumption: If >$50\%$ of respondents of a country stated that policy documents and clinical guidelines in that country contained certain information (e.g., requirement of clinicians to screen for a certain comorbidity), we projected this to the whole country.
In order to facilitate the interpretation of the results, we collapsed certain items together: Responses to diagnosis and screening questions were collated in a single item, diagnosis/screening; a positive response to either one of these two items was considered positive for the new item. The same strategy was used to combine items asking about the start and maintenance of care (start/maintain care) and items asking about referral to, or liaison with a specialised service for a comorbidity (referral/liaison).
Further details can be found in the online supplement (Report).
## Results
We invited 1,100 prospective respondents to participate in the survey. 543 responded, of whom $18\%$ [97] did not engage beyond section 1 (respondent’s characteristics) and their data were excluded. A further $16.4\%$ [89] and $3.7\%$ [20] did not answer any questions beyond sections 2 and 3, respectively; this left $62.1\%$ [337] respondents who completed the full survey (Fig 1). When we compared the characteristics of excluded and included respondents, we found no notable differences (S3 Appendix—’respondents’ and non-respondents’ characteristics’ in the online supplementary materials).
**Fig 1:** *Number of respondents completing sections of the survey by country.Only the number of included respondents are plotted (value label). Participants who stopped answering in the first section of the survey (regarding participants characteristics) were excluded. Countries that are also in the TB/HIV high burden list are indicated with an asterisk [*].*
## Respondents’ characteristics
We obtained responses from 27 countries out of the 33 countries, with two-thirds of all respondents from four countries (Brazil: 116, India: 77, the Philippines: 62, and China: 39) (Table 1). We obtained more than 10 responses from three other countries (Nigeria: 37, Uganda: 20 and Mozambique: 13), whilst the remaining 20 countries (Angola, Pakistan, Bangladesh, Kenya, Papua New Guinea, Sierra Leone, Democratic Republic of Congo, Zambia, Zimbabwe, South Africa, Lesotho, United Republic of Tanzania, Congo-Brazzaville, Ethiopia, Myanmar, Central African Republic, Indonesia, Liberia, Namibia, Russian Federation) offered 10 or fewer responses; no response was received from six countries (Cambodia, Democratic People’s Republic of Korea, Gabon, Mongolia, Thailand, Vietnam) (Fig 1).
**Table 1**
| Country | frequency | % |
| --- | --- | --- |
| Brazil | 90 | 26.7 |
| India | 54 | 16.0 |
| Philippines | 48 | 14.2 |
| China | 35 | 10. 4 |
| Nigeria | 29 | 8.6 |
| Uganda | 18 | 5.3 |
| Other | 27 | 18.7 |
| Position | | |
| TBP manager | 158 | 35.4 |
| Consultant | 61 | 13.7 |
| 1y.health care | 71 | 15.9 |
| 2y.health care | 25 | 5.6 |
| 3y.health care | 64 | 14.3 |
| Other | 67 | 15.0 |
| Do you work at a DOTS clinic? (health care workers only) | Do you work at a DOTS clinic? (health care workers only) | Do you work at a DOTS clinic? (health care workers only) |
| Yes | 83 | 51.9 |
| No | 71 | 44.4 |
| . | 6 | 3.6 |
| Main role in the organisation | Main role in the organisation | |
| Service manager | 105 | 24.0% |
| Clinician/ TB healthc | 162 | 36.0% |
| Advocacy / advisory | 78 | 17.0% |
| Other (please specify | 93 | 21.0% |
| NR | 8 | 2.0% |
| Type of service provider | Type of service provider | |
| public | 117 | 52.0% |
| private (not-for-prof | 28 | 12.0% |
| private (for-profit) | 7 | 3.0% |
| Other | 5 | 2.0% |
| NR | 70 | 31.0% |
| At what level are you working? (TBP managers and Consultants only) | At what level are you working? (TBP managers and Consultants only) | At what level are you working? (TBP managers and Consultants only) |
| International | 17 | 5.8 |
| National | 64 | 21.8 |
| Provincial | 43 | 14.7 |
| State | 56 | 19.1 |
| District | 77 | 26.3 |
| Subdistrict | 36 | 12.3 |
| Work in a TB clinic? | | |
| Yes, I do work in a TB clinic | 182 | 52.2 |
| No, I do not work in a TB clinic | 167 | 47.8 |
Thirty-five percent of respondents described themselves as ‘TB programme manager/ supervisor/ coordinator’ (referred henceforth as National TB Programme [NTP] managers), and a similar proportion ($36\%$) described themselves as working directly with people with TB in primary, secondary, or tertiary health care. Half of the remaining respondents ($14\%$) were ‘the UNION/ WHO/ other non-governmental organisation consultants/ advocates/ advisors’ referred henceforth as Consultants (Fig 2). When asked specifically about their role in the organisation or TB service, 162 ($36\%$) described it as ‘clinician/ TB healthcare professional’, 105 ($24\%$) as ‘service manager’ and 78 ($17\%$) as ‘advocacy or advisory’ (Table 1). This indicates that not all health care workers had a clinical role, and not all NTP managers described their role as a service manager (Table C in S3 Appendix—’respondents and non-respondents’ characteristics’ in the online supplementary materials).
**Fig 2:** *Positions of respondents.*
HIV, diabetes mellitus, depression, tobacco use, and alcohol misuse disorders were consistently mentioned as the most common and concerning comorbid conditions or risk factors in people with TB. Most TB health professionals recalled multimorbidity mentioned in TB clinical guidelines and a slightly fewer proportion thought it was mentioned in TB policies (Fig 3). TB programme managers were more likely to refer to policy documents and strategic plans as opposed to clinicians who mostly remembered it being mentioned in clinical guidelines.
**Fig 3:** *Proportion of TB health professionals who thought that multimorbidity is mentioned in TB policies and guidelines.*
## HIV
Most [22] TB high-burden countries included in our survey were also listed under TB/HIV high-burden countries. In countries with >10 responses, survey respondents reported that HIV was mentioned in TB policy documents and clinical guidelines (Table 2); and that in practice, HIV was screened for in people with TB. A high proportion of respondents in these countries also stated that care for HIV was started or maintained, and TB services referred to, or liaised with specialist HIV services, or did both (Table 3). Most of the respondents felt capable of diagnosing HIV and slightly fewer felt capable of treating it (Table 4). The above pattern was repeated across all countries included in the survey with most respondents recalling integration of HIV diagnosis and management within TB policy and practice (Tables 2–4).
## Diabetes mellitus
In countries with >10 responses, diabetes was mentioned in TB policy documents and clinical guidelines except for Nigeria (Table 2). In practice, most TB service providers reported diagnosing/screening for diabetes in people with TB but fewer reported starting/maintaining care or referring to or liaising with specialist diabetes services. Only in Mozambique, most respondents reported people with TB and diabetes were referred to specialist services (Table 3). While most respondents felt capable of diagnosing diabetes in people with TB, again, fewer felt capable of treating this condition (Table 4).
When considering all countries with at least one response, diabetes was considered in practice in about half of the countries (Table 3). A slightly higher proportion of respondents felt capable of diagnosing and treating TB (Table 3).
## Depression
In countries with >10 responses, depression was rarely mentioned in TB policy documents and clinical guidelines (with the exception of Mozambique where clinicians were asked to prevent, screen and treat depression in TB services) (Table 2). Consequently, in practice, depression was neither screened for nor diagnosed in any of these countries according to most respondents. In Mozambique, little over half of the respondents ($55\%$) reported screening or diagnosis of depression in TB services. Likewise, most respondents reported that depression was not treated in TB services in any of these countries. However, in Brazil, China, Mozambique and Philippines, most patients were at least referred to or offered liaison with mental health services. While approximately half of the respondents (service providers) felt capable of diagnosing depression, most were not confident in treating depression (Table 4). The above picture did not change by including all high-TB burden countries with >1 response in the sensitivity analysis (Tables 2–4) except that the vast majority of responding service providers felt incapable of diagnosing or treating depression.
## Tobacco use
Tobacco was mentioned in TB policy and clinical guidelines in all countries with 10 responses except Nigeria. However, prevention of tobacco, amongst those diagnosed with TB, use was the focus in most of these guidelines, with less attention on screening and cessation advice (Table 2). This was reflected in practice, with only about half of the respondents in these countries (fewer in Nigeria) reporting that patients were routinely asked about tobacco use. Even fewer reported that cessation advice/treatment or referral to cessation specialists was offered (Table 3). Only in four countries with 10 responses, respondents reported that they felt capable of asking about tobacco use; and less than half felt capable of treating tobacco addiction in people with TB (Table 4).
In only half of the countries with at least one response, people with TB were asked or advised about tobacco use and only in a few countries were patients treated for tobacco addiction or referred to specialist services (Tables 2–4). Similarly, only in about half of the countries a majority of respondents felt capable of screening for tobacco use, and in very few countries [3] did respondents feel capable of treating tobacco addiction (Tables 2–4).
## Alcohol use
The response pattern for alcohol use was very similar to that for tobacco use. Except for Nigeria, in all countries with 10 responses, clinicians were encouraged in policy documents and clinical guidelines, to diagnose and screen for alcohol use in TB patients. They reported less often that these documents included requirements to start/maintain care or to involve other specialist services for alcohol misuse (Table 2). In practice, alcohol use was diagnosed/screened for in about half of the services, with considerably fewer starting/maintaining care or referring/liaising with other specialist services (Table 3). Regarding how capable respondents felt about diagnosing alcohol use disorder in people with TB, the percentage ranged from $46\%$ to $81\%$ in the countries with 10 responses. However, most respondents did not feel capable of treating alcohol use disorder (Table 4).
In countries with at least one response, most clinicians reported screening for alcohol in TB patients, but people from only a small number of countries were treated (2 countries) or referred (5 countries) for alcohol use (Tables 2–4). In about one half of the countries, most respondents felt capable of screening for alcohol use disorder in TB patients. ( Tables 2–4
## Discussion
The survey highlighted that HIV, diabetes, tobacco use, alcohol use disorders, and depression are the five most common and concerning comorbidities in people with TB. While the prevention, screening and treatment of HIV was considered within TB policies and guidelines across most countries, tobacco, alcohol, diabetes and depression were mentioned less often. At the level of service provision, most of the high-TB burden countries screened, diagnosed and offered referrals for HIV in people with TB. About half of the participating countries offered similar services or some levels of liaison with specialist services for comorbid diabetes. On the other hand, screening for tobacco and alcohol use disorders were conducted only in a minority of countries; very few countries initiated, maintained care or liaised with specialist services. Depression was consistently absent from policies and service provision in almost all countries.
The impact of HIV on TB, and its implications for TB and HIV control, have long been recognised as a global public health challenge. This is reflected in the level of integration of care for HIV in TB services observed in our survey. The WHO TB/HIV policy [13], published over a decade ago, was aimed at establishing and strengthening collaboration between HIV and TB control programmes. The goal was to reduce the burden of HIV in TB patients by providing HIV prevention, diagnosis and treatment through integrated TB and HIV services. In line with this policy, most high-TB/HIV burden countries have put policies and guidelines in place for the screening and management of HIV in TB patients. Our findings are in line with other literature highlighting that the TB-HIV collaborative care has been integrated at various levels; these range from referral-based approaches to a more integrated person-centred care model where TB-HIV services are provided under one roof [16, 17].
The WHO, in 2011, declared TB-diabetes collaborative care as one of its TB eradication strategies and included it as an essential part of the efforts to achieve Sustainable Development Goals (SDGs) [18]. Our findings suggest that most health care systems are yet to fully integrate diabetes care within TB care; more countries need to get involved in collaborative TB-diabetes management. Other researchers have also observed varying levels of collaborative care: for example, In Pakistan, the uptake of diabetes testing among TB cases was reported as $21.8\%$ while the uptake of TB testing among individuals with pre-diabetes and diabetes has been reported to be as low as $4.7\%$ [19]. This lack of TB-DM co-management or integrated care may be due to a lack of programmatic leadership, investment, training and workforce in diabetes care. However, other studies [20, 21] reveal that given adequate capacity-building programmes, identification and mitigation of operational challenges and provision of logistic supplies, there is good potential to increase awareness about TB-diabetes collaborative care, and provide TB-diabetes bi-directional screening, management and reporting.
Despite being recognised in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5-TR) [22] tobacco and alcohol use disorders are rarely considered for further advice or treatment by clinicians offering TB care, as shown in our survey. Tobacco and alcohol use are important determinants of unfavourable TB treatment outcomes [23]. In 2007, WHO and the International Union Against TB and Lung Diseases, highlighted the dangerous interaction between TB and tobacco smoking and emphasised that tobacco control activities needs implementing as an integral part of TB management [24]. This interaction between TB and tobacco epidemics may account for over 15–$20\%$ of TB-related deaths [25]. Findings of the ASSIST trial [26], show that offering cessation support intervention is highly effective when delivered by TB DOTS (Directly Observed Treatment Short-course) facilitators. The role of TB DOTS facilitators in delivery of smoking cessation support effectively in a TB clinic has been acknowledged by other studies also [27]. and attempts to scale up tobacco cessation with national TB programmes are gaining ground [28]. Similarly, the relationship between TB and alcohol use disorder has been firmly established [29, 30]. Imtiaz et al estimated that in 2014 about 170 000 deaths due to TB were attributable to alcohol consumption worldwide, and as many as $17\%$ of incident cases of TB and $15\%$ of deaths due to TB could be prevented by eliminating the harmful use of alcohol [31]. Given the advantages of offering tobacco and alcohol cessation support to TB patients, approaches should be implemented for systematic screening and early identification and effective management, including referral to specialists for support.
Our findings suggest a lack of integrated depression services within the TB programme; these findings are in line with a recent [2019] global survey of national TB programme directors by Sweetland et al [32]. This semi-structured survey of national TB programme directors from 26 countries of all income levels reported that only two out of the twenty six national TB programmes included screening for any mental disorder. Similar to our findings, the survey reported that national TB programmes currently do not address mental disorders as part of routine practice. Nevertheless, receptivity was high, creating an opportunity to integrate the management of depression and other mental health problems within TB programmes.
## Strengths and limitations
Our sampling strategy resulted in over 400 responses from 27 of 33 high-burden countries but most responses (two-thirds) came from only four countries (Brazil, India, Philippines and China). Therefore, our findings are most relevant to these four countries and cannot be generalised across all high-TB burden countries. Furthermore, a lack of probability sampling meant that even the responses from each individual country may not necessarily be representative of that country’s policy and practice. In addition, $85\%$ of the respondents represent government or UN agencies, which limits the possibility of triangulation between governmental and non-governmental sources. Since the survey was anonymous, we were unable to track who did and who did not answer our invitation and can therefore not assess the characteristics of non-respondents. However, we obtained responses from a wide range of TB programme managers, primary, secondary and tertiary health care workers, and consultants. We, therefore, believe that our results provide valuable initial insight into this topic, particularly for the countries with multiple respondents.
Despite the aforementioned limitations, this survey provides a useful snapshot of how conditions comorbid with TB are prioritised and addressed both in policy and guidelines, and in practice in high-TB-burden countries. We found little evidence of routine screening and reporting of comorbid conditions by TB services. Our findings highlight an important surveillance gap in the existing TB recording and reporting system. Although TB-HIV collaborative services have been embedded at various levels of the TB programme, there is a lack of non-communicable disease (NCD) integration within TB programmes (and limited evidence on evidence-based integrated care). Integrating NCD services within TB programmes may offer a viable solution to reduce TB and NCD-related disease burden. Global organisations, such as the WHO and The Union could help educate policy makers and programme managers through international conferences and through the dissemination of educational materials. Furthermore, the integration of NCD services into TB programmes also requires strong capacity-building efforts among primary care workers. An initial step for the TB programmes across the globe may be to provide and scale up routine screening of various NCDs. We also recommend development and evaluation of TB specific interventions targeting common comorbidities to help fill the knowledge gap on how to achieve integrated care and its effectiveness for people (with TB) with comorbidities.
## References
1. Programme GT. *Global tuberculosis report* (2021.0)
2. Siddiqi K, Stubbs B, Lin Y, Elsey H, Siddiqi N. **TB multimorbidity: a global health challenge demanding urgent attention**. *Int J Tuberc Lung Dis* (2021.0) **25** 87-90. DOI: 10.5588/ijtld.20.0751
3. Stubbs B, Siddiqi K, Elsey H, Siddiqi N, Ma R, Romano E. **Tuberculosis and Non-Communicable Disease Multimorbidity: An Analysis of the World Health Survey in 48 Low- and Middle-Income Countries**. (2020.0). DOI: 10.2139/ssrn.3667847
4. Duko B, Gebeyehu A, Ayano G. **Prevalence and correlates of depression and anxiety among patients with tuberculosis at WolaitaSodo University Hospital and Sodo Health Center, WolaitaSodo, South Ethiopia, Cross sectional study**. *BMC Psychiatry* (2015.0). DOI: 10.1186/s12888-015-0598-3
5. Gelaw YA, Williams G, Soares Magalhães RJ, Gilks CF, Assefa Y. **HIV Prevalence Among Tuberculosis Patients in Sub-Saharan Africa: A Systematic Review and Meta-analysis**. *AIDS Behav* (2019.0) **23** 1561-1575. DOI: 10.1007/s10461-018-02386-4
6. Noubiap JJ, Nansseu JR, Nyaga UF, Nkeck JR, Endomba FT, Kaze AD. **Global prevalence of diabetes in active tuberculosis: a systematic review and meta-analysis of data from 2·3 million patients with tuberculosis**. *The Lancet Global Health* (2019.0) e448-e460. DOI: 10.1016/s2214-109x(18)30487-x
7. Soares LN, Spagnolo LM de L, Tomberg JO, Zanatti CL de M, Cardozo-Gonzales RI. **Relationship between multimorbidity and the outcome of the treatment for pulmonary tuberculosis**. *Rev Gaucha Enferm* (2020.0) **41** e20190373. DOI: 10.1590/1983-1447.2020.20190373
8. Ruiz-Grosso P, Cachay R, de la Flor A, Schwalb A, Ugarte-Gil C. **Association between tuberculosis and depression on negative outcomes of tuberculosis treatment: A systematic review and meta-analysis**. *PLOS ONE* (2020.0) e0227472. DOI: 10.1371/journal.pone.0227472
9. Huddart S, Svadzian A, Nafade V, Satyanarayana S, Pai M. **Tuberculosis case fatality in India: a systematic review and meta-analysis**. *BMJ Glob Health* (2020.0) **5** e002080. DOI: 10.1136/bmjgh-2019-002080
10. Eshetie S, Gizachew M, Alebel A, van Soolingen D. **Tuberculosis treatment outcomes in Ethiopia from 2003 to 2016, and impact of HIV co-infection and prior drug exposure: A systematic review and meta-analysis**. *PLOS ONE* (2018.0) e0194675. DOI: 10.1371/journal.pone.0194675
11. Gautam S, Shrestha N, Mahato S, Nguyen TPA, Mishra SR, Berg-Beckhoff G. **Diabetes among tuberculosis patients and its impact on tuberculosis treatment in South Asia: a systematic review and meta-analysis**. *Sci Rep* (2021.0) **11** 2113. DOI: 10.1038/s41598-021-81057-2
12. Smith SM, Wallace E, O’Dowd T, Fortin M. **Interventions for improving outcomes in patients with multimorbidity in primary care and community settings**. *Cochrane Database of Systematic Reviews* (2016.0). DOI: 10.1002/14651858.CD006560.pub3
13. Sculier D, Gétāhun H. **WHO Policy on Collaborative TB/HIV Activities: Guidelines for National Programmes and Other Stakeholders**. *World Health Organization* (2012.0)
14. 14Co-morbidities. [cited 25 Apr 2022]. Available: https://www.who.int/southeastasia/activities/co-morbidities-tb
15. 15WHO releases new global lists of high-burden countries for TB, HIV-associated TB and drug-resistant TB. [cited 25 Apr 2022]. Available: https://www.who.int/news/item/17-06-2021-who-releases-new-global-lists-of-high-burden-countries-for-tb-hiv-associated-tb-and-drug-resistant-tb
16. Legido-Quigley H, Montgomery CM, Khan P, Atun R, Fakoya A, Getahun H. **Integrating tuberculosis and HIV services in low- and middle-income countries: a systematic review**. *Trop Med Int Health* (2013.0) **18** 199-211. DOI: 10.1111/tmi.12029
17. Uyei J, Coetzee D, Macinko J, Guttmacher S. **Integrated delivery of HIV and tuberculosis services in sub-Saharan Africa: a systematic review**. *Lancet Infect Dis* (2011.0) **11** 855-867. DOI: 10.1016/S1473-3099(11)70145-1
18. 18World Health Organization. Collaborative Framework for Care and Control of Tuberculosis and Diabetes. World Health Organization; 2011.. *Collaborative Framework for Care and Control of Tuberculosis and Diabetes* (2011.0)
19. Basir MS, Habib SS, Zaidi SMA, Khowaja S, Hussain H, Ferrand RA. **Operationalization of bi-directional screening for tuberculosis and diabetes in private sector healthcare clinics in Karachi, Pakistan**. *BMC Health Serv Res* (2019.0) **19** 147. DOI: 10.1186/s12913-019-3975-7
20. Arini M, Sugiyo D, Permana I. **Challenges, opportunities, and potential roles of the private primary care providers in tuberculosis and diabetes mellitus collaborative care and control: a qualitative study**. *BMC Health Serv Res* (2022.0) **22** 215. DOI: 10.1186/s12913-022-07612-3
21. 21Enhanced, integrated diabetes tuberculosis care, WDF14-906. In: World diabetes foundation [Internet]. 28 May 2015 [cited 25 Apr 2022]. Available: https://www.worlddiabetesfoundation.org/projects/pakistan-wdf14-906
22. 22American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. American Psychiatric Association Publishing; 2022.. *Diagnostic and Statistical Manual of Mental Disorders* (2022.0)
23. Marais BJ, Lönnroth K, Lawn SD, Migliori GB, Mwaba P, Glaziou P. **Tuberculosis comorbidity with communicable and non-communicable diseases: integrating health services and control efforts**. *Lancet Infect Dis* (2013.0) **13** 436-448. DOI: 10.1016/S1473-3099(13)70015-X
24. 24WHO/UNION. Association between exposure to tobacco smoke and tuberculosis: a qualitative systematic review. 2007. Available: http://www.who.int/tobacco/resources/publications/tb_tobac_monograph.pdf. *Association between exposure to tobacco smoke and tuberculosis: a qualitative systematic review* (2007.0)
25. Siddiqi K, Dogar O. **TB and Tobacco—An Unholy Alliance**. *Pak J Chest Med* (2018.0) **24** 01-03
26. Siddiqi K, Khan A, Ahmad M, Dogar O, Kanaan M, Newell J. **Action to Stop Smoking In Suspected Tuberculosis (ASSIST) in Pakistan: a Cluster Randomized, Controlled Trial**. *Ann Intern Med* (2013.0) **158** 667-675. DOI: 10.7326/0003-4819-158-9-201305070-00006
27. Sereno AB, Soares ECC, Lapa E Silva JR, Nápoles AM, Bialous SA, Costa E Silva VL da. **Feasibility study of a smoking cessation intervention in Directly Observed Therapy Short-Course tuberculosis treatment clinics in Rio de Janeiro, Brazil**. *Rev Panam Salud Publica* (2012.0) **32** 451-456. PMID: 23370189
28. Elsey H, Al Azdi Z, Regmi S, Baral S, Fatima R, Fieroze F. **Scaling up tobacco cessation within TB programmes: findings from a multi-country, mixed-methods implementation study**. *Health Res Policy Syst* (2022.0) **20** 43. DOI: 10.1186/s12961-022-00842-1
29. Lönnroth K, Williams BG, Stadlin S, Jaramillo E, Dye C. **Alcohol use as a risk factor for tuberculosis—a systematic review**. *BMC Public Health* (2008.0) **8** 289. DOI: 10.1186/1471-2458-8-289
30. Rehm J, Samokhvalov AV, Neuman MG, Room R, Parry C, Lönnroth K. **The association between alcohol use, alcohol use disorders and tuberculosis (TB)**. *A systematic review. BMC Public Health* (2009.0) **9** 450. PMID: 19961618
31. Imtiaz S, Shield KD, Roerecke M, Samokhvalov AV, Lönnroth K, Rehm J. **Alcohol consumption as a risk factor for tuberculosis: meta-analyses and burden of disease**. *Eur Respir J* (2017.0) 50. DOI: 10.1183/13993003.00216-2017
32. Sweetland AC, Galea J, Shin SS, Driver C, Dlodlo RA, Karpati A. **Integrating tuberculosis and mental health services: global receptivity of national tuberculosis program directors**. *Int J Tuberc Lung Dis* (2019.0) **23** 600-605. DOI: 10.5588/ijtld.18.0530
|
---
title: 'County-level societal predictors of COVID-19 cases and deaths changed through
time in the United States: A longitudinal ecological study'
authors:
- Philip J. Bergmann
- Nathan A. Ahlgren
- Rosalie A. Torres Stone
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022229
doi: 10.1371/journal.pgph.0001282
license: CC BY 4.0
---
# County-level societal predictors of COVID-19 cases and deaths changed through time in the United States: A longitudinal ecological study
## Abstract
People of different racial/ethnic backgrounds, demographics, health, and socioeconomic characteristics have experienced disproportionate rates of infection and death due to COVID-19. This study tests if and how county-level rates of infection and death have changed in relation to societal county characteristics through time as the pandemic progressed. This longitudinal study sampled monthly county-level COVID-19 case and death data per 100,000 residents from April 2020 to March 2022, and studied the relationships of these variables with racial/ethnic, demographic, health, and socioeconomic characteristics for 3125 or $97.0\%$ of U.S. counties, accounting for $96.4\%$ of the U.S. population. The association of all county-level characteristics with COVID-19 case and death rates changed significantly through time, and showed different patterns. For example, counties with higher population proportions of Black, Native American, foreign-born non-citizen, elderly residents, households in poverty, or higher income inequality suffered disproportionately higher COVID-19 case and death rates at the beginning of the pandemic, followed by reversed, attenuated or fluctuating patterns, depending on the variable. Patterns for counties with higher *White versus* Black population proportions showed somewhat inverse patterns. Counties with higher female population proportions initially had lower case rates but higher death rates, and case and death rates become more coupled and fluctuated later in the pandemic. Counties with higher population densities had fluctuating case and death rates, with peaks coinciding with new variants of COVID-19. Counties with a greater proportion of university-educated residents had lower case and death rates throughout the pandemic, although the strength of this relationship fluctuated through time. This research clearly shows that how different segments of society are affected by a pandemic changes through time. Therefore, targeted policies and interventions that change as a pandemic unfolds are necessary to mitigate its disproportionate effects on vulnerable populations, particularly during the first six months of a pandemic.
## Introduction
The SARS-CoV-2 corona virus causes the disease COVID-19. It emerged in late 2019 and has spread throughout the world. As of May 1, 2022, the United States led the world in infections and mortality, with over 84 million cases and over 1 million deaths. Since the beginning of the pandemic, patterns of unequal infection and mortality have emerged across countries and communities based on factors such as behavior, race/ethnicity, demography, health, and socioeconomics, and these are often interconnected [1–5].
For example, in the U.S., Black and Hispanic communities have experienced disproportionately high COVID-19 infection and mortality rates [6–11]. Socioeconomic and health disparities resulting from structural racism have been implicated in these patterns [12–17]. The U.S. Black and Hispanic populations disproportionately reside in densely populated areas, and have lower incomes [18,19], higher unemployment, or are employed in frontline, essential occupations [4]. These communities often mistrust health care institutions [20,21], have lower access to health care due to income and lack of health insurance [22], and have a higher incidence of pre-existing conditions such as diabetes [23] than their White counterparts.
How these various factors affect rates of COVID-19 infections and death in the general population also differ. For example, high-density housing, along with frontline, essential work are likely to lead to higher rates of infection due to lack of ability to socially distance in these settings [24–26]. Meanwhile, lack of access to healthcare and higher incidence of pre-existing conditions are likely to lead to higher mortality [22,23,27]. In contrast, high unemployment may coincide with lower rates of infection due to lower exposure to other people.
County-level demographic and COVID-19 data have been valuable in understanding the disproportionate burden across communities [13,14]. However, most studies have only taken snapshots of the pandemic’s effect by using the most recent cumulative data available at the time of analysis [18,19,28]. This provides an understanding of cumulative effects since the beginning of the pandemic, but, as a result, we lack an understanding of how the pandemic has progressed through time in relation to race/ethnicity, demographic, health, and socioeconomic factors. For example, we know that Black communities show higher rates of infection and mortality [9,29], but we do not know whether the effect on these communities has worsened, remained constant, or improved through time. It is likely that the effects of the pandemic are not temporally static. In particular, the elderly comprised a high proportion of infection and mortality early in the pandemic, but as the pandemic progressed, the demographics of infection, but not mortality, shifted to younger groups [5,30–32].
Here we present an analysis of how COVID-19 infections and mortality relate to a range of racial/ethnic, demographic, health, and socioeconomic factors through time for 3125 or $97.0\%$ of U.S. counties, accounting for $96.4\%$ of the U.S. population. We specifically address how these relationships have changed on a monthly basis from April 2020 to March 2022. Although community-level data have important limitations, including being subject to the ecological fallacy, they provide the best means for a comprehensive understanding of patterns over time across the entire United States. Differences in data reporting across jurisdictions precludes use of more precise data that also cover the entire country [33,34].
## Variables and data sources
We conducted a longitudinal ecological study using county-level population data and sampling COVID-19 case and death data on a monthly basis. County-level demographic and health data were downloaded and compiled from the United States Census Bureau’s 2014–2018 American Community Survey for 3220 counties (www.tigerweb.geo.census.gov/tigerwebma), and from the County Health Rankings & Roadmaps (CHRR) database (www.countyhealthrankings.org). We calculated population density as the total population for each county divided by its land area, using U.S. Census data. Although we included all explanatory variables in our models, for ease of presentation, we divided them into racial/ethnic, demographic, health-related, and socioeconomic variables. We selected explanatory variables known to be predictors of COVID-19 outcomes. For racial/ethnic variables, we included the proportion of the county population that was non-Hispanic White, Black, Hispanic, Asian, and Native American. For demographic variables, we included population density, and the proportion of the population that was foreign-born non-citizen, female, and that lives in a rural setting. For health-related variables, we included the proportion of the population that was elderly, disabled, obese, and lacked health insurance. Finally, for socioeconomic variables, we included median household income, the proportion of households under the poverty line, the proportion of the population that was unemployed, or had a university degree, and the Gini coefficient, which is a measure of income inequality (zero being complete income equality, and one signifying that a single person has all the income). The specifics for each explanatory variable are listed in S1 Table. We acknowledge that many other societal characteristics and variables have been identified as related to COVID-19 cases and deaths, but wished to focus on a manageable set that is well represented in the literature.
County-level COVID-19 case and death data were compiled from The New York Times COVID-19 repository (www.github.com/nytimes/covid-19-data) [35] and normalized per 100,000 residents (we refer to these as case and death rates). We retrieved cumulative COVID-19 case and death data for the first day of each month from April 2020 until March 2022, and calculated the number of cases and deaths per 100,000 residents that occurred during each month by subtraction. These monthly case and death rates served as the response variables in our analyses. The interval that we selected coincided with the first month in which COVID-19 had spread in parts of the U.S. until the outbreak of the Omicron BA.1 variant subsided and testing was still widely available. We ended data analysis with March 2022 because as 2022 progressed, at-home testing for COVID-19 increased, often with no mechanism for patients to report test results, leading to increased underreporting of cases [36,37].
The S1 Data includes all data that we compiled and used. All data are publicly available and county-level, so their use did not require approval by the Institutional Review Board.
## Statistical analysis
We used negative binomial regression models weighted by county population to quantify the relationships between COVID-19 case and death rates, and all of the racial/ethnic, demographic, health, and socioeconomic variables that we studied. The partial slopes of these models provide estimates of relationship between each explanatory variable with the response, while taking into account all other included explanatory variables [38]. Negative binomial models are flexible in accounting for differing levels of overdispersion and zeros in the data [13,39,40], such as when many counties have no deaths during a particular month. We weighted our analyses by county population [28,41], which considerably increased the variance in COVID-19 case and death rates that our analyses explained (S2 Table).
For each month, we fitted a negative binomial regression model with the glm.nb function in the MASS package [42] using R v4.0.3 [43]. We ensured that collinearity did not compromise the analyses by calculating the tolerance of each explanatory variable. All of our variables had tolerances >0.1, which was deemed acceptable [13,38], except percent non-Hispanic White (S3 Table). To address this, we excluded this variable from our analyses, and then repeated analyses including percent non-Hispanic White and all demographic, health, and socioeconomic variables, but excluding the other racial/ethnic variables. We present results as partial slopes from the models that included all variables except percent non-Hispanic White, plus the partial slope for percent non-Hispanic White from the additional analyses. We then plotted partial slopes with their $95\%$ confidence intervals through time. All statistical results, including p-values are presented in S4 and S5 Tables. R2 values for all models are presented in S2 Table. We repeated these analyses without weighting by county population (i.e., treating each county the same), and obtained mostly qualitatively similar patterns, except for the proportion of county population that lived in rural settings or were elderly (S1 Fig). We do not discuss the unweighted analyses further, but the similarity of observed patterns suggests that our results are robust to analytical choices.
## Results
We found dramatic changes through time in how racial/ethnic, demographic, health, and socioeconomic characteristics of U.S. counties related to per capita COVID-19 infections and deaths. The changes that we present account for all other county characteristics included in our analyses, representing their independent effects. The context for these patterns are the national COVID-19 case and death rates, which have also fluctuated (Fig 1). Additionally, a number of important events happened during the pandemic that might impact how particular groups have been affected, including the spread of the alpha (from October 2020), delta (from June 2021) and omicron (from December 2021) variants [44,45], and the widespread availability of vaccination (from February 2021) and boosters (from September 2021) (Fig 1). Over this timeframe, the models that we fit also fluctuated in the amount of variance in case (15–$62\%$) and death (9–$59\%$) rates that they explained (S2 Table). Many of the lower R2 values corresponded with months when overall COVID-19 case and death rates were low across the U.S.
**Fig 1:** *Graph of COVID-19 case and death rates for the United States through time.Vertical dashed lines indicate the approximate time when the alpha (α), delta (δ), and omicron (o) variants of SARS-CoV-2 started to spread in the U.S. Dot-dashed lines indicate when vaccination (Vax) and boosting (Boost) became widespread in the U.S. Blue lines represent case rate and red lines represent death rate.*
## How racial/ethnic characteristics relate to COVID-19 infection and mortality
The patterns of relationship between COVID-19 and the proportions of White and Black residents were approximately inverses of one another (Fig 2A and 2B). Counties with greater proportions of Black and lower proportions of White residents had higher COVID-19 case and death rates for the first seven months of the pandemic. These patterns reversed around the holidays of Thanksgiving and Christmas 2020, before disappearing round February 2021 when vaccination became widely available (Fig 2A and 2B). After that time, COVID-19 case and death rates were weakly related to proportion of Black residents, becoming negatively related towards the end of 2021, after vaccine boosters became available (Fig 2B). In contrast, after vaccination became available, counties with higher proportions of White residents tended to have higher case and death rates, but this fluctuated through time (Fig 2A). Relationships for White residence were strongly decoupled for case (positive relationship) and death (negative relationship) rates just after vaccinations became available.
**Fig 2:** *Relationships through time between either COVID-19 case (blue) or death (red) rates and proportion of U.S. county populations comprised of non-Hispanic White (A), Black (B), Hispanic (C), Asian (D), and Native American (E) residents. Partial slopes through time represent the strength of relationship for each variable. Error bars for each month are 95% confidence intervals. Filled circles are significantly different from zero, and open circles are not. Dashed lines indicate the approximate time when the alpha (α), delta (δ), and omicron (o) variants of SARS-CoV-2 started to spread in the U.S. Dot-dashed lines indicate when vaccination (Vax) and boosting (Boost) became widespread in the U.S. Models including non-Hispanic White did not converge for May 2020, resulting in missing partial slopes for that variable and that month.*
We also found that counties with a higher proportion of Hispanic residents had lower case and death rates for the first four months, followed by higher case and death rates in August and September 2020, January and February 2021, and July and August 2021, with periods of weakly negative or no relationships between these dates (Fig 2C). The relationship of COVID-19 rates with proportion of residents of Asian descent were strongly negative until August 2020, followed by fluctuations between positive and negative (Fig 2D). After vaccination became widespread, relationships for case and death rates with Asian population proportion decoupled somewhat with strong negative relationships for cases and weak relationships for deaths (Fig 2D). Relationships of COVID-19 case and death rates with proportion of county residents that were Native American fluctuated through the pandemic, but were often positive, especially at the beginning of the pandemic (Fig 2E). Counties with higher population proportions of Native Americans also had much higher death rats in May and June 2021 (Fig 2E). Death rates for this group tended to be lower after vaccine boosters became available (Fig 2E).
## How demographic characteristics relate to COVID-19
The relationships of COVID-19 case and death rates with the proportion of county residents that were foreign-born non-citizens were closely coupled and strongly positively related during April to July 2020 (Fig 3A). From August 2020 until February 2021, the relationships between COVID-19 and this demographic were either not different from zero or slightly negative. Counties with higher foreign-born non-citizen populations had lower case and death rates in July to October 2021, coinciding with the spread of the delta variant, before relationships became weak or non-existent again (Fig 3A). Counties with higher proportion of female residents had lower case rates but higher death rates from May until October 2020, followed by fluctuating patterns that were more coupled between case and death rates, and mostly positive (Fig 3B). The proportion of county population living in a rural setting was strongly negatively related to both case and death rates for the first six months of the pandemic, followed by weakly negative relationships (Fig 3C). However, in August and September 2021, more rural counties had higher death rates (Fig 3C). Finally, county population density was strongly positively related to case and death rates during the first few months of the pandemic, then negatively related in August to October 2020, followed by fluctuations between positive and negative relationships (Fig 3D). The relationship between population density and case and death rates was negative while the delta variant dominated in the U.S., and then became more positive when the omicron variant spread (Fig 3D).
**Fig 3:** *Relationships through time between either COVID-19 case (blue) or death (red) rates and proportion of U.S. county populations comprised of foreign-born non-citizens (A), females (B), and residents living in rural settings (C), as well as the county population density (D). Partial slopes through time represent the strength of relationship for each variable. Error bars for each month are 95% confidence intervals. Filled circles are significantly different from zero, and open circles are not. Dashed lines indicate the approximate time when the alpha (α), delta (δ), and omicron (o) variants of SARS-CoV-2 started to spread in the U.S. Dot-dashed lines indicate when vaccination (Vax) and boosting (Boost) became widespread in the U.S.*
## How health-related characteristics relate to COVID-19
Counties with a higher population proportion that was elderly had higher COVID-19 case rates for the first three months of the pandemic and higher death rates for the first year (Fig 4A). For most of the pandemic, case rates were negatively related to elderly population proportion. Death rates were positively related until they became strongly negatively related in July to September 2021, and then remained weak after boosters were widespread (Fig 4A).
**Fig 4:** *Relationships through time between either COVID-19 case (blue) or death (red) rates and proportion of U.S. county populations comprised of elderly (A), disabled (B), health uninsured (C), and obese (D) residents. Partial slopes through time represent the strength of relationship for each variable. Error bars for each month are 95% confidence intervals. Filled circles are significantly different from zero, and open circles are not. Dashed lines indicate the approximate time when the alpha (α), delta (δ), and omicron (o) variants of SARS-CoV-2 started to spread in the U.S. Dot-dashed lines indicate when vaccination (Vax) and boosting (Boost) became widespread in the U.S.*
The proportion of county population that was disabled or lacked health insurance fluctuated with case and death rates in similar ways (Fig 4B and 4C). The relationships to death rates lagged slightly in time relative to case rates for both variables. There was a negative relationship between COVID-19 and both of these variables for the first four months of the pandemic, followed by fluctuations that became stronger from about July 2021, when the delta variant spread (Fig 4B and 4C). Specifically, proportions of disabled and uninsured residents were positively related to COVID-19 case and death rates from about July to about November 2021, followed by weaker or negative relationships afterward (Fig 4B and 4C).
The relationship between proportion of county population that was obese with both COVID-19 case and death rates was strongly negative in April 2020, then quicly becoming positive and fluctuating until about April 2021, when relationships to case and death rates became more decoupled (Fig 4D). From May to September 2021, counties with more obese populations had higher death rates but lower or unrelated case rates. After September 2021, case and death rates fluctuated weakly between positive and negative (Fig 4D).
## How socioeconomic characteristics relate to COVID-19
Median county income, the Gini index (a measure of income inequality), and the proportion of households in poverty had similar temporal relationships with COVID-19 case and death rates (Fig 5A, 5B and 5C). For the first four to six months of the pandemic, counties with higher incomes, income inequality, and higher poverty had higher case and death rates. Subsequently, all three variable fluctuated through time in their relationship with COVID-19, with fluctuations become less pronounced and lower for median income and poverty (Fig 5A and 5C), while being more sizeable for income inequality (Fig 5B). From July 2021 until March 2022 (the end of our sampling), counties with higher incomes and poverty had either no relationship with COVID-19 case or death rates, or negative relationships. In contrast, counties with higher income inequality had higher case and death rates January to May, as well as August to September 2021, with intervening periods of negative relationship or no relationship (Fig 5B).
**Fig 5:** *Relationships through time between either COVID-19 case (blue) or death (red) rates and U.S. county median income (A), proportion of households below the poverty line (B), proportion of residents that unemployed (C), and proportion of residents with a university degree (D). Partial slopes through time represent the strength of relationship for each variable. Error bars for each month are 95% confidence intervals. Filled circles are significantly different from zero, and open circles are not. Dashed lines indicate the approximate time when the alpha (α), delta (δ), and omicron (o) variants of SARS-CoV-2 started to spread in the U.S. Dot-dashed lines indicate when vaccination (Vax) and boosting (Boost) became widespread in the U.S.*
The proportion of county populations that were unemployed had negative relationships with COVID-19 case rates, and positive or not significant relationships with death rates through September 2020 (Fig 5D). Subsequently, the relationship between unemployment and death rate became mostly negative and coupled with patterns for case rates. From about July 2021 until March 2022, case and death rates fluctuated between being positively and negatively related to unemployment rates (Fig 5D).
Finally, the proportion of county populations with a university degree was the most consistent predictor of COVID-19 case and death rates. Throughout much of the pandemic, counties with higher proportions of university-educated residents had lower death rates and mostly lower case rates (Fig 5E). However, in May, June and December 2021, as well as March 2022 counties with more highly educated populations had higher case rates, although death rates remained lower (Fig 5E).
## Discussion
In our study, we tested whether and how the relationships between COVID-19 and a range of racial/ethnic, demographic, health, and socioeconomic factors changed through time across the U.S. Most importantly, we found that not only do these relationships change through time, but they differ between COVID-19 case and death rates, and between societal factors at the county level. Our findings provide an important step in understanding how a pandemic affects different segments of society as it progresses, and has important implications, particularly that policies and practice for mitigating the effects of a pandemic must also change through time.
That some racial and ethnic groups have been disproportionately affected by COVID-19 is well established [13,16,26,27,46], but how this has changed temporally was not well understood. We found that counties with a higher proportion of the population that is Black had higher case and death rates for the first six months of the pandemic, before this pattern reversed, and then fluctuated through time (Fig 1B). This shift closely coincided with the holidays of Thanksgiving, Christmas, and the New Year of 2020, and repeated for 2021, when travel across the country peaked despite warnings from public health officials [1,47]. It seems probable that widespread travel indiscriminately spread COVID-19, and that this served to either obfuscate racial disparities observed at the beginning of the pandemic, or reflect racial differences in holiday travel. That the relationships between COVID-19 with proportions of White and Black individuals were somewhat inverse in temporal pattern supports this (Fig 1A and 1B). However, this inverse pattern must be interpreted cautiously because the proportion White population variable was highly collinear with all other racial categories (S3 Table). On the other hand, the inverse pattern was only observed between Black and White groups, not other racial groups. Nevertheless, the strong effect of COVID-19 on counties with high proportions of Black residents were likely due to the additive effects of systemic racism [12,24,48], but appear to have been mitigated to some degree as the pandemic progressed. The disproportionate impact on U.S. Black populations is even more striking given a higher biological susceptibility to COVID-19 of populations of European descent than those of African descent due to a genomic segment, inherited from Neanderthals, that is prevalent in the former and virtually absent in the latter [49,50]. However, there are also regional differences in these patterns in the U.S. [15,51,52] that we did not consider here.
In contrast, the temporal patterns of how county proportions of Hispanic, Asian, and Native American residents did not change around the year-end holidays, but had strong patterns especially at the beginning of the pandemic. There are important cultural and socioeconomic differences within both the Hispanic and Asian populations, so ethnic subgroups within these categories may be differentially affected by the pandemic [53–55]. Counties with higher population proportions of foreign-born individuals that were not U.S. citizens also had higher case and death rates early in the pandemic, but subsequently, this demographic was related to mildly lower case and death rates (Fig 2A).
The patterns that we documented in how median income, income inequality (Gini index), and household poverty were related to COVID-19 case and death rates were similar (Fig 5A–5C). In particular, counties with higher incomes, income inequality, and poverty suffered higher COVID-19 case and death rates for the first seven months of the pandemic, followed by fluctuating relationships. A positive relationship between income inequality and COVID-19 has been documented internationally and in the U.S. and is a proxy for socioeconomic disadvantage [56–58]. Similarly, poor households also tend to be more crowded with smaller living spaces [59], and poverty corresponds with lower access to healthcare in the U.S. [29], likely leading to worse outcomes. The pandemic also produced disproportionate job loss and food and medical insecurity among low-wage subpopulations [4]. The matching temporal pattern between median income, income inequality, and poverty is counterintuitive, but may be due to cost of living differences across counties [60].
The results discussed above showed that the first six months of the pandemic had many of our explanatory variables strongly related to COVID-19 case and death rates. After this time, relationships changed, likely because society responded more effectively to the pandemic. For example, during that initial stage of the pandemic, counties with higher population proportions of Black, Native American, foreign-born, elderly, and obese residents suffered higher case and death rates. Additionally, counties with higher density populations, high income inequality, and high poverty rates suffered higher case and death rates, and these counties also tended to have higher median incomes. These patterns reveal that segments of society were particularly vulnerable to the pandemic when it began, but also suggest some success of society in responding to these vulnerabilities, as relationships became less pronounced after that period.
The contrasting temporal patterns of how population density and proportion of the population living in rural settings related to COVID-19 infections and mortality (Fig 3C and 3D) suggest that the independent effects of these variables measure different things. That population density related strongly and positively to mortality, both at the beginning of the pandemic and after the alpha and omicron variants became widespread, may correspond with factors such as ability to isolate effectively when infected in high-density settings. It also may represent differences between urban, low-density housing such as the suburbs versus urban high-density housing [18].
We found evidence of human behavior affecting COVID-19 progression for some segments of the population. For example, except for the first three months of the pandemic, case rates were largely either unrelated to the elderly county population, or negatively related (Fig 4A). It is likely that elderly people took considerable precautions, such as mask wearing and social distancing [61] to avoid infection and that public health practices improved as the pandemic progressed. Similarly, mortality in counties with higher elderly populations improved after vaccination became widespread (Fig 4A), consistent with the observed high uptake of vaccines by this demographic [62]. However, older age remains an important predictor of COVID-19 outcomes [32,63,64]. Counties with a greater proportion of the population with a university degree consistently had lower COVID-19 case and death rates (Fig 5E), with behavior again being the most likely explanation. In particular, people with a university degree were more likely to be employed in situations where they could work from home to facilitate social distancing [65], and more likely to accept vaccination [62].
The observed weak relationships between unemployed residents and COVID-19 case and death rates (Fig 5D) may be rationalized in that unemployed people had few workplace interactions, and likely engaged in less travel for work, recreation, and shopping [66]. During the first six months of the pandemic, proportion of county populations that were disabled was also either unrelated or weakly negatively related to COVID-19 case and death rates (Fig 4B). Similarly, this may be explained in that persons with disabilities may have lower mobility, resulting in lower potential for exposure and infection [29]. Subsequently, relationships between COVID-19 case/death rates fluctuated dramatically and similarly with population proportions of disabled and residents that lacked health insurance (Fig 4B and 4C). These similar patterns are harder to explain. However, the strongest positive relationships between these groups and COVID-19 coincided with the autumn 2021 peaks in COVID-19 case and death rates (Fig 1), and this suggests an inability of these groups to avoid COVID-19 during at least that particular surge [67].
## Limitations
County-level data have important limitations, including being subject to the ecological fallacy, but provide the best means for a comprehensive understanding of temporal patterns across the entire U.S. County-level data are an amalgamation of populations, limiting the ability to make definitive conclusions about mechanisms. However, differences in data reporting across jurisdictions precludes use of more precise data that also cover the entire country [33,34]. Some of the county-level social variables (S1 Table: percentage of the population that lacks health insurance, is obese, or is unemployed) that we used are modeled, and data for counties with low populations or low response rates to surveys in particular are estimates based on those models [68]. Therefore, these estimates may be less accurate than direct data for the other variables we included [68], which are based directly on data. These estimated variables may be collinear with other variables partly as a result of being estimated using other variables.
Another limitation is differential COVID-19 data quality across jurisdictions. U.S. states have maintained different data reporting standards, including for counting cases and deaths, and the frequency of reporting. Differences in COVID-19 testing availability exacerbated this. Therefore, deaths and especially cases are undercounted [69]. How this might bias the results is unknown and likely complex. Weighting our analyses by county population should address some of these biases by increasing the weight of counties with higher populations and more resources for counting cases and deaths. Weighted analyses explained more variance in COVID-19 case and death rates than unweighted analyses (S2 Table). Finally, we note that the explanatory power of our models changes substantially through time, with less variation in COVID-19 case rates being explained between March and July 2021, and June to August 2021 for death rates (S2 Table). This period of time mostly coincided with low case and death rates nationally (Fig 1), which may explain the low explanatory power.
Given that the COVID-19 case and death data are temporal in nature, a time series analysis is another option for analysis. We did not use a time series analysis because COVID-19 is an emerging and quickly evolving disease and monthly sampling for only two years (24 time points) is insufficient for a robust analysis [70,71]. This might be remedied with a shorter time interval (e.g., weekly), but then the number of cases and particularly deaths would be zero for most counties during most time points, also weakening the analysis.
## Conclusions
We showed that relationships between racial/ethnic, demographic, health, and socioeconomic factors with COVID-19 case and death rates changed through time in the U.S. Temporal changes and differences in how particular population segments are infected and die from COVID-19 are critical to informing policy and practice behind mitigation efforts, especially in resource-limited scenarios such as a pandemic. This could include prioritizing efforts to mitigate spread versus enhancing access to health care. For example, we found that counties with higher Black, Native American, foreign-born, elderly, high density, and impoverished populations were particularly susceptible to infection and mortality from COVID-19 early in the pandemic. Efforts to address factors leading to the spread of the virus and higher mortality of these vulnerable groups are particularly important at the onset of a pandemic. A health equity lens to mitigate COVID-19 disparities is key. For example, enhancing access to testing in places where these groups are more likely to receive care [72], adopting more racially equitable triage in racially diverse areas [73], and addressing implicit biases in medical treatment [74] would all help address these disparities. The first six months of a pandemic appear to be critical in addressing these issues, and so learning from the COVID-19 pandemic should be applied to any future pandemics. Additionally, the consistently negative relationship between university education and COVID-19 case and death rates highlights a well-established positive impact of education on health outcomes and mitigating health disparities [75]. Finally, our results suggest that efforts to decrease the spread of COVID-19 among more elderly populations were somewhat successful, especially later in the pandemic, but efforts to decrease mortality in this demographic took much longer to establish.
## References
1. Boccia S.. **Christmas Festivities and COVID-19: A Foreseeable Risk to Anticipate**. *Frontiers in Public Health* (2021.0) **8** 639647. DOI: 10.3389/fpubh.2020.639647
2. Mena GE, Martinez PP, Mahmud AS, Marquet PA, Buckee CO, Santillana M. **Socioeconomic status determines COVID-19 incidence and related mortality in Santiago**. *Chile. Science* (2021.0) **372** eabg5298. DOI: 10.1126/science.abg5298
3. Pathangey G, Fadadu PP, Hospodar AR, Abbas AE. **Angiotensin-converting enzyme 2 and COVID-19: patients, comorbidities, and therapies**. *American Journal of Physiology* (2020.0) **320** L301-L30. DOI: 10.1152/ajplung.00259.2020
4. Selden TM, Berdahl TA. **COVID-19 And Racial/Ethnic Disparities In Health Risk, Employment, And Household Composition: Study examines potential explanations for racial-ethnic disparities in COVID-19 hospitalizations and mortality**. *Health Affairs* (2020.0). DOI: 10.1377/hlthaff.2020.00897
5. Sudharsanan N, Didzun O, Bärnighausen T, Geldsetzer P. **The Contribution of the Age Distribution of Cases to COVID-19 Case Fatality Across Countries**. *Annals of Internal Medicine* (2020.0) **173** 714-20. DOI: 10.7326/M20-2973
6. Egede LE, Walker RJ, Garacci E, Raymond JR. **Racial/Ethnic Differences In COVID-19 Screening, Hospitalization, And Mortality In Southeast Wisconsin: Study examines racial/ethnic differences in COVID-19 screening, symptom presentation, hospitalization, and mortality among 31,549 adults tested for COVID-19 in Wisconsin**. *Health Affairs* (2020.0) **39** 1926-34. PMID: 33136498
7. Gross CP, Essien UR, Pasha S, Gross JR, Wang S-Y, Nunez-Smith M. **Racial and Ethnic Disparities in Population Level Covid-19 Mortality**. *Journal of General Internal Medicine* (2020.0) **35** 3097-9. DOI: 10.1007/s11606-020-06081-w
8. Hatcher SM, Agnew-Brune C, Anderson M, Zambrano LD, Rose CE, Jim MA. **COVID-19 Among American Indian and Alaska Native Persons—23 States, January 31–July 3, 2020**. *MMWR Morbidity & Mortality Weekly Report* (2020.0) **69** 1166-9. DOI: 10.15585/mmwr.mm6934e1
9. Hooper MW, Nápoles AM, Pérez-Stable EJ. **COVID-19 and Racial/Ethnic Disparities**. *JAMA* (2020.0) **323** 2466-7. DOI: 10.1001/jama.2020.8598
10. Kirby T.. **Evidence mounts on the disproportionate effect of COVID-19 on ethnic minorities**. *The Lancet Respiratory Medicine* (2020.0) **8** 547-8. DOI: 10.1016/S2213-2600(20)30228-9
11. Yancy CW. **COVID-19 and African Americans**. *JAMA* (2020.0) **323** 1891-2. DOI: 10.1001/jama.2020.6548
12. Khanijahani A, Tomassoni L. **Socioeconomic and Racial Segregation and COVID-19: Concentrated Disadvantage and Black Concentration in Association with COVID-19 Deaths in the USA**. *Journal of Racial and Ethnic Health Disparities* (2021.0). DOI: 10.1007/s40615-021-00965-1
13. McLaughlin JM, Khan F, Pugh S, Angulo FJ, Schmitt H-J, Isturiz RE. **County-level Predictors of Coronavirus Disease 2019 (COVID-19) Cases and Deaths in the United States: What Happened, and Where Do We Go from Here?**. *Clinical Infectious Diseases* (2020.0). DOI: 10.1093/cid/ciaa1729
14. Torrats-Espinosa G.. **Using machine learning to estimate the effect of racial segregation on COVID-19 mortality in the United States**. *Proceedings of the National Academy of Sciences of the USA* (2021.0) **118** e2015577118. DOI: 10.1073/pnas.2015577118
15. Torres Stone RA, Ahlgren NA, Bergmann PJ. **Multiple Measures of Structural Racism as Predictors of County-Level COVID-19 Cases and Deaths Across Counties in the United States**. *Ethnic & Racial Studies* (2022.0). DOI: 10.1080/01419870.2022.2105655
16. Yang T-C, Choi S-wE, Sun F. **COVID-19 cases in US counties: roles of racial/ethnic density and residential segregation**. *Ethnicity & Health* (2020.0) **26** 11-21. DOI: 10.1080/13557858.2020.1830036
17. Yu Q, Salvador CE, Melani I, Berg MK, Neblett EW, Kitayama S. **Racial residential segregation and economic disparity jointly exacerbate COVID‐19 fatality in large American cities**. *Annals of the New York Academy of Sciences* (2021.0). DOI: 10.1111/nyas.14567
18. Adhikari S, Pantaleo NP, Feldman JM, Ogedegbe O, Thorpe L, Troxel AB. **infections and deaths in large US metropolitan areas**. *JAMA network open* (2020.0) **3** e2016938-e. PMID: 32721027
19. Snowden LR, Graaf G. **COVID-19, Social Determinants Past, Present, and Future, and African Americans’ Health**. *Journal of Racial and Ethnic Health Disparities* (2020.0) **8** 12-20. DOI: 10.1007/s40615-020-00923-3
20. Corbie-Smith G, Thomas SB. **George DMMS. Distrust, race, and research**. *Archives of Internal Medicine* (2002.0) **162** 2458-63. PMID: 12437405
21. Jones JH, Emanuel EJ, Grady CC, Crouch RA, Lie RK, Miller FG, Wendler DD. *The Oxford Textbook of Clinical Research Ethics* (2008.0) 86-96
22. Venkatesh AK, Chou SC, Li SX, Choi J, Ross JS, D’Onofrio G. **Association between insurance status and access to hospital care in emergency department disposition**. *JAMA Internal Medicine* (2019.0) **179** 687-93. DOI: 10.1001/jamainternmed.2019.0037
23. Gupta R, Hussain A, Misra A. **Diabetes and COVID-19: evidence, current status and unanswered research questions**. *European Journal of Clinical Nutrition* (2020.0) **74** 864-70. DOI: 10.1038/s41430-020-0652-1
24. Gould E, Wilson V. **Black workers face two of the most lethal preexisting conditions for coronavirus—racism and economic inequality**. *Economic Policy Institute Report* (2020.0)
25. Kinder M, Ross M. **Reopening America: Low-wage workers have suffered badly from COVID-19 so policymakers should focus on equity**. *Brookings Institute Report* (2020.0)
26. Poulson M, Geary A, Annesi C, Allee L, Kenzik K, Sanchez S. **National Disparities in COVID-19 Outcomes between Black and White Americans**. *Journal of the National Medical Association* (2021.0) **113** 125-32. DOI: 10.1016/j.jnma.2020.07.009
27. Price-Haywood EG, Burton J, Fort D, Seoane L. **Hospitalization and mortality among black patients and white patients with Covid-19**. *New England Journal of Medicine* (2020.0) **382** 2534-43. DOI: 10.1056/NEJMsa2011686
28. Figueroa JF, Wadhera RK, Lee D, Yeh RW, Sommers BD. **Community-Level Factors Associated With Racial And Ethnic Disparities In COVID-19 Rates In Massachusetts: Study examines community-level factors associated with racial and ethnic disparities in COVID-19 rates in Massachusetts**. *Health Affairs* (2020.0) **39** 1984-92. PMID: 32853056
29. Abedi V, Olulana O, Avula V, Chaudhary D, Khan A, Shahjouei S. **Racial, economic, and health inequality and COVID-19 infection in the United States**. *Journal of Racial and Ethnic Health Disparities* (2020.0) **8** 732-42. DOI: 10.1007/s40615-020-00833-4
30. Boehmer TK, DeVies J, Caruso E, van Santen KL, Tang S, Black CL. **Changing Age Distribution of the COVID-19 Pandemic—United States, May–August 2020**. *MMWR Morbidity & Mortality Weekly Report* (2020.0) **69** 1404-9. DOI: 10.15585/mmwr.mm6939e1
31. 31CDC. COVID-NET: COVID-19-Associated Hospitalization Surveillance Network.: Centers for Disease Control and Prevention; 2021 [May 15, 2021]. Available from: https://covid.cdc.gov/covid-data-tracker/#covidnet-hospitalization-network.. *COVID-NET: COVID-19-Associated Hospitalization Surveillance Network.* (2021.0)
32. Ioannidis JPA, Axfors C, Contopoulos-Ioannidis DG. **Second versus first wave of COVID-19 deaths: Shifts in age distribution and in nursing home fatalities**. *Environmental Research* (2021.0) **195** 110856. DOI: 10.1016/j.envres.2021.110856
33. Hasell J, Mathieu E, Beltekian D, Macdonald B, Giattino C, Ortiz-Ospina E. **A cross-country database of COVID-19 testing**. *Scientific Data* (2020.0) **7** 345. DOI: 10.1038/s41597-020-00688-8
34. Sáez C, Romero N, Conejero JA, García-Gómez JM. **Potential limitations in COVID-19 machine learning due to data source variability: A case study in the nCov2019 dataset**. *Journal of the American Medical Informatics Association* (2021.0) **28** 360-4. DOI: 10.1093/jamia/ocaa258
35. 35The New York Times. Coronavirus (Covid-19) Data in the United States. Accessed: 3/21/20212021 [updated 3/21/20213/21/2021]. Available from: https://github.com/nytimes/covid-19-data.
36. Rader B, Gertz A, Iuliano AD, Gilmer M, Wronski L, Astley CM. **Use of At-Home COVID-19 Tests—United States, August 23, 2021-March 12, 2022.**. *Morbidity and Mortality Weekly Report* (2022.0) **71** 489-94. DOI: 10.15585/mmwr.mm7113e1
37. Rubin R.. **COVID-19 testing moves out of the clinic and into the home**. *JAMA* (2021.0) **326** 1362-4. DOI: 10.1001/jama.2021.15679
38. Quinn GP, Keough MJ. *Experimental Design and Data Analysis for Biologists* (2002.0) xvii+537
39. Crawley MJ. *The R Book* (2012.0) 1076
40. Tian T, Zhang J, Hu L, Jiang Y, Duan C, Li Z. **Risk factors associated with mortality of COVID-19 in 3125 counties of the United States**. *Infectious Diseases of Poverty* (2021.0) 10. DOI: 10.1186/s40249-020-00786-0
41. Kazanchi R, Beiter ER, Gondi S, Beckman AL, Bilinski A, Ganguli I. **County-level association of social vulnerability with COVID-19 cases and deaths in the USA**. *Journal of General Internal Medicine* (2020.0) **35** 2784-7. DOI: 10.1007/s11606-020-05882-3
42. Venables WN, Ripley BD. *Modern Applied Statistics with S* (2002.0)
43. 43R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria
2021.. *R: A Language and Environment for Statistical Computing* (2021.0)
44. Hadfield J, Megill C, Bell SM, Huddleston J, Potter B, Callender C. **Nextstrain: real-time tracking of pathogen evolution**. *Bioinformatics* (2018.0) **34** 4121-3. DOI: 10.1093/bioinformatics/bty407
45. Hadfield J, Megill C, Bell SM, Huddleston J, Potter B, Callender C. *Genomic epidemiology of SARS-CoV-2 with subsampling focused globally since pandemic start*
46. Muñoz-Price LS, Nattinger AB, Rivera F, Hanson R, Gmehlin CG, Perez A. **Racial disparities in incidence and outcomes among patients with COVID-19**. *JAMA Network Open* (2020.0) **3** e2021892. DOI: 10.1001/jamanetworkopen.2020.21892
47. Tanne JH. **Covid-19: US faces “surge upon a surge”**. *BMJ* (2020.0). DOI: 10.1136/bmj.m4693
48. Bailey ZD, Krieger N, Agénor M, Graves J, Linos N, Bassett MT. **Structural racism and health inequities in the USA: evidence and interventions**. *The Lancet* (2017.0) **389** 1453-63. DOI: 10.1016/S0140-6736(17)30569-X
49. **Genomewide association study of severe Covid-19 with respiratory failure**. *New England Journal of Medicine* (2020.0) **383** 1522-34. DOI: 10.1056/NEJMoa2020283
50. Zeberg H, Pääbo S. **The major genetic risk factor for severe COVID-19 is inherited from Neanderthals**. *Nature* (2020.0) **587** 610-2. DOI: 10.1038/s41586-020-2818-3
51. Desmet K, Wacziarg R. **JUE Insight: Understanding spatial variation in COVID-19 across the United States**. *Journal of Urban Economics* (2022.0) **127** 103332. DOI: 10.1016/j.jue.2021.103332
52. Zhu D, Ye X, Manson S. **Revealing the spatial shifting pattern of COVID-19 pandemic in the United States**. *Scientific Reports* (2021.0) **11** 8396. DOI: 10.1038/s41598-021-87902-8
53. Budiman A, Ruiz NG. **Key facts about Asian origin groups in the U.S. American Community Survey 1-yr estimates**. *Pew Research Center* (2021.0)
54. Noe-Bustamante L.. **Key facts about U.S. Hispanics and their diverse heritage**. *Fact Tank, Pew Reserach Center* (2019.0)
55. Torres Stone RA, McQuillan J. **Beyond Hispanic/Latino: The importance of gender/ethnicity-specific earnings analyses**. *Social Science Research* (2007.0) **36** 175-200. DOI: 10.1016/j.ssresearch.2005.11.003
56. Oronce CI, Scannell CA, Kawachi I, Tsugawa Y. **Association Between State-Level Income Inequality and COVID-19 Cases and Mortality in the USA**. *Journal of General Internal Medicine* (2020.0) **35** 2791-3. DOI: 10.1007/s11606-020-05971-3
57. Tan AX, Hinman JA, Abdel Magid HS, Nelson LM, Odden MC. **Association Between Income Inequality and County-Level COVID-19 Cases and Deaths in the US**. *JAMA Network Open* (2021.0) **4** e218799. DOI: 10.1001/jamanetworkopen.2021.8799
58. Wildman J.. **COVID-19 and income inequality in OECD countries**. *European Journal of Health Economics* (2021.0) **22** 455-62. DOI: 10.1007/s10198-021-01266-4
59. Krieger J, Higgins DL. **Housing and health: time again for public health action**. *American Journal of Public Health* (2002.0) **92** 758-68. DOI: 10.2105/ajph.92.5.758
60. Curran LB, Wolman H, Hill EW, Furdell K. **Economic Wellbeing and Where We Live: Accounting for Geographical Cost-of-living Differences in the US**. *Urban Studies* (2006.0) **43** 2443-66. DOI: 10.1080/00420980600970698
61. Haischer MH, Beilfuss R, Hart MR, Opielinski L, Wrucke D, Zirgaitis G. **Who is wearing a mask? Gender-, age-, and location-related differences during the COVID-19 pandemic**. *PLOS One* (2020.0) **15** e0240785. DOI: 10.1371/journal.pone.0240785
62. Mondal P, Sinharoy A, Su L. **Sociodemographic predictors of COVID-19 vaccine acceptance: a nationwide US-based survey study**. *Public Health* (2021.0) **198** 252-9. DOI: 10.1016/j.puhe.2021.07.028
63. Nanda A, Vura NVRK, Gravenstein S. **COVID-19 in older adults**. *Aging Clinical and Experimental Research* (2020.0) **32** 1199-202. DOI: 10.1007/s40520-020-01581-5
64. Shahid Z, Kalayanamitra R, McClafferty B, Kepko D, Ramgobin D, Patel R. **COVID‐19 and older adults: what we know**. *Journal of the American Geriatrics Society* (2020.0) **68** 926-9. DOI: 10.1111/jgs.16472
65. Messacar D, Morissett R, Deng Z. *Inequality in the feasibility of working from home during and after COVID-19. Statistics Canada* (2020.0)
66. Llorente A, Garcia-Herranz M, Moro E. **Social Media Fingerprints of Unemployment**. *PLOS One* (2015.0) **10** e0128692. DOI: 10.1371/journal.pone.0128692
67. Kavanagh A, Dickinson H, Carey G, Llewellyn G, Emerson E, Disney G. **Improving health care for disabled people in COVID-19 and beyond: Lessons from Australia and England**. *Disability Health Journal* (2021.0) **14** 101050. DOI: 10.1016/j.dhjo.2020.101050
68. Remington PL, Catlin BB, Gennuso KP. **The County Health Rankings: rationale and methods**. *Population Health Metrics* (2015.0) **13** 11. DOI: 10.1186/s12963-015-0044-2
69. Dubrow JK. **Local data and upstream reporting as sources of error in the administrative data undercount of Covid 19**. *International Journal of Social Research Methodology* (2021.0). DOI: 10.1080/13645579.2021.1909337
70. Jebb AT, Tay L, Wang W, Huang Q. **Time series analysis for psychological research: Examining and forecasting change**. *Frontiers in Psychology* (2015.0) **6** 1-24. PMID: 25688217
71. Warner RM. *Spectral analysis of time-series data* (1998.0)
72. Peek ME, Simons RA, Parker WF, Ansell DA, Rogers SO, Edmonds BT. **COVID-19 among African Americans: an action plan for mitigating disparities**. *American Journal of Public Health* (2021.0) **111** 286-92. DOI: 10.2105/AJPH.2020.305990
73. Bruce L, Tallman R. **Promoting racial equity in COVID-19 resource allocation**. *Journal of Medical Ethics* (2021.0) **47** 208-12. DOI: 10.1136/medethics-2020-106794
74. Milam AJ, Furr-Holden D, Edwards-Johnson J, Webb B, Patton JW, Ezekwemba NC. **Are clinicians contributing to excess African American COVID-19 deaths?**. *Unbeknownst to them, they may be. Health Equity* (2020.0) **4** 139-41. PMID: 32368712
75. Zajacova A, Lawrence EM. **The relationship between education and health: reducing disparities through a contextual approach**. *Annual Review of Public Health* (2018.0) **39** 273-89. DOI: 10.1146/annurev-publhealth-031816-044628
|
---
title: Statin-use and perceptions of high cholesterol as predictors of healthy lifestyle
behaviours in Nigerians
authors:
- Joyce F. Coker
- Kate M. Hill
- Akaninyene A. Otu
- Allan House
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022232
doi: 10.1371/journal.pgph.0000190
license: CC BY 4.0
---
# Statin-use and perceptions of high cholesterol as predictors of healthy lifestyle behaviours in Nigerians
## Abstract
It is unclear how statin-use influences the adoption of healthy lifestyle choices. It is important to understand the nature of this relationship as this could facilitate targeted public health interventions which could help promote a healthy lifestyle, curb the rise of non-communicable diseases, and facilitate overall health. This study aimed to explore whether statin-use influenced the adoption of healthy lifestyle choices by changing the way urban and semi-urban Nigerians thought about their high cholesterol and their future risk of cardiovascular disease. Structured questionnaires were used to compare the lifestyle behaviours, perceptions of high cholesterol and future risk of cardiovascular disease of statin users and non-statin users recruited in urban and a semi-urban Nigeria. In-depth, face-to-face interviews were used to further explore the relationship between statin-use and the adoption of healthy lifestyle choices, and explore the influence of personal and social factors on this relationship. The odds of adopting a low-fat diet increased as perceived statin-effectiveness increased (OR = 2.33, $p \leq 0.05$), demonstrating a synergistic relationship between statin-use and the adoption of healthy of lifestyle choices. In addition to this synergistic association, at interview, two other relationships were found between statin use and the adoption of healthy lifestyle choices: an antagonistic relationship fuelled by a strong perception of statin effectiveness and a perceived inability to make healthy lifestyle changes, which favoured statin-use, and an antagonistic relationship fuelled by congruous cause-control beliefs and concerns about medication-use which favoured the adoption of healthy lifestyle choices. The odds of adopting a low-fat diet was 5 times greater in urban dwellers than in semi-urban dwellers ($p \leq 0.01$). Statin-use influenced the adoption of healthy lifestyle choices in three different ways, which require exploration at clinical consultation. Gender, social obligations, and physical environment also influenced statin-use and the adoption of healthy lifestyle choices.
## Introduction
Developing countries, particularly those in sub-Saharan Africa face a double burden of non-communicable and communicable diseases. The latter is still the major cause of adult mortality in the region. However, non-communicable diseases pose a substantial burden and appear to be on the rise, further straining already over-burdened and fragile health care systems [1–3]. In Nigeria, non-communicable diseases account for $29\%$ of adult deaths, $11\%$ of which are due to cardiovascular disease [4]. Almost half of these deaths occurred in the prime of life [5].
The rise of cardiovascular disease and other non-communicable disease in sub-Saharan Africa has been attributed to an aging population and urbanisation [6–10]. A key strategy for the primary prevention of cardiovascular disease is the lowering of lipid levels [2, 11]. This can be achieved by making lifestyle modifications such as eating a low-fat diet, regular physical activity and smoking cessation. Where needed, lipid-lowering medications are prescribed for use alongside lifestyle modifications [11–13]. Due to their documented effectiveness in lipid modification, statins have become the lipid-lowering medication of choice and one of the most prescribed medications in the world [14, 15]. However, there remains some doubt about their benefit for the primary prevention of cardiovascular disease where even relatively low rates of side effects may not be justified by their effect on preventing uncommon events [16–18].
There is sparse and conflicting data available on the relationship between statin-use and the adoption of healthy lifestyle choices [19]. Some researchers claim that statin-use provides a false sense of security, enabling people to neglect lifestyle modifications and continue to make poor lifestyle choices [20–22]. Others have either found no difference or found that statin users consumed less dietary fat than non-statin users [19, 23]. As statin-use increases, it is important to assess whether there is an interaction between statin-use and the adoption of healthy lifestyle choices for a variety of reasons. Firstly, whilst research confirms that statins produce a moderate reduction in cardiovascular disease risk, their effect is enhanced when combined with lifestyle modifications [24–26]. Secondly, these lifestyle modifications are beneficial not only for the control of lipid levels, but also for the control of other cardiovascular disease risk factors, non-communicable diseases, and promote overall health and wellbeing [5, 8, 27]. Thirdly, understanding factors that influence modifiable health-related behaviours such as the adoption of healthy lifestyle choices may facilitate the development and implementation of targeted public health interventions [2, 28–30]. This could potentially help curb the rise of cardiovascular disease risk factors in sub-Saharan Africa.
We have not identified any studies that investigate the relation between statin-use and the adoption of healthy lifestyle choices in sub-Saharan Africa where the burden of cardiovascular disease risk factors is on the increase. Consequently, this study aimed to explore (i) whether statin-use influenced the adoption of healthy dietary and exercise choices and (ii) whether it did so, by changing the way urban and semi-urban Nigerians perceived their high cholesterol and their future risk of cardiovascular disease. To fulfil these research aims, the following 4 research questions were explored: The Common Sense Model of self-regulation, and the Health Belief Model were used as frameworks to understand how illness perceptions and health beliefs influence health-related behaviours [28, 31–33]. The Common Sense Model posits that when faced with a health threat, individuals produce a cognitive representation of the health threat which informs their choice of coping strategies [33]. The Health Belief Model was originally developed to explain or predict preventive health behaviours in healthy individuals [32]. Both models acknowledge the modifying role of socio-cultural factors on illness perceptions.
## Mixed methods approach
This study employed an explanatory sequential mixed methods approach. This approach involved the collection and analysis of data in 2 distinct phases. Phase 1 involved the collection and analysis of quantitative data. The findings of phase 1 were used to inform the choice of participants purposefully sampled to take part in phase 2—a qualitative study. Both studies were analysed separately and integrated in the discussion section [34].
## Quantitative research methods
A structured questionnaire was used to obtain information about participants’ demographics, cardiovascular risk factors, dietary and exercise behaviours, perceptions of their high cholesterol, and future risk of cardiovascular diseases.
## Qualitative research methods
In-depth interviews carried out in-person were used to explore the narratives voiced by statin users and non-statin users about their dietary and exercise behaviours, perceptions of high cholesterol, and perceived future risk of cardiovascular diseases. These interviews were also used to explore participants’ accounts of how personal and social factors influenced their aforementioned behaviours and perceptions.
## Study setting
This study was conducted across two sites in Nigeria. The first site was the Nigeria National Petroleum Corporation (NNPC) medical services located in the cosmopolitan Maitama district in Abuja, Nigeria. NNPC is a national oil and gas company and provides free healthcare for its staff and their immediate family. Abuja is the capital of Nigeria and has been exposed to a lot of Western influences. This site will be referred to as the urban research site. The second site was the University of Calabar Teaching Hospital (UCTH) located in Calabar, the capital of Cross-River state in the oil-rich *Niger delta* region. Tourism has become an avenue of wealth creation for Cross-River state [35]. Consequently, Cross River state is rapidly undergoing urbanisation [36]. This site will be referred to as the semi-urban research site. Healthcare at this site was not provided free of charge.
## Ethical approval and informed consent
Ethical approval was obtained from the Nigerian Institute for Medical Research (IRB/$\frac{13}{216}$), from the local research ethics committees of both research sites (GGM/$\frac{11}{12}$ & UCTH/LR/DM/16), and from the Leeds Institute of Health Sciences and Leeds Institute of Genetics, Health and Therapeutics and Leeds Institute of Molecular Medicine joint ethics committee (HSLTLM/$\frac{12}{063}$ R).
Prior to completion of the questionnaire, the researcher verbally went through and provided all participants with an information sheet which detailed the voluntary nature and purpose of the study, details of participation, and freedom to and process of withdrawal from the study. Written consent was not requested for questionnaire completion. Rather, the return of self-completed questionnaires, or the act of following the researcher into a vacant consultation room, and verbally completing researcher-administered questionnaires were deemed as implied consent. Before the commencement of all interviews, the researcher verbally went through the information sheet again with participants and written consent was sought before the recorder was switched on and interviews begun.
## Study sample
Adults aged 18-years and older who had a recorded diagnosis of hyperlipidaemia and attended a hospital appointment at either research site between August to October 2013 were invited to take part in the study. People were excluded from the study if they: (i) did not speak English; (ii) had experienced a diagnosed cardiovascular event—in accordance with our focus on the primary prevention of cardiovascular disease and; (iii) had a diagnosis of familial hypercholesterolemia or type 1 diabetes due to the significant role of genetics in these conditions.
The respondents’ age for statin and non-statin users were 52.7±10.5 and 53.2±11.9 years, respectively. Participants were predominantly females ($57\%$), had hypertension ($60\%$), and $24\%$ had hypertension and diabetes (Table 1). Over half ($55\%$) of participants were recruited from the semi-urban site. Urban dwellers were significantly better educated than semi urban dwellers (tertiary education = $79.4\%$ urban dwellers vs $38.3\%$ semi urban dwellers, Fisher’s exact = 41.14, $p \leq 0.001$), and were more likely to have lived outside of Nigeria in the 10-years prior to this study than semi-urban dwellers (OR = 7.5, X2 = 12.72, $p \leq 0.001$).
**Table 1**
| Variable | Statin users n = 78 (52.7%) | Non-statin users n = 70 (47.3%) | Overall n = 148 (100%) |
| --- | --- | --- | --- |
| Mean age (years) (SD) | 52.68 (10.45) | 53.16 (11.88) | 52.91 (11.11) |
| 20–29 | 0 (0) | 2 (2.9) | 2 (1.4) |
| 30–39 | 9 (11.5) | 4 (5.7) | 13 (8.8) |
| 40–49 | 21 (26.9) | 20 (28.6) | 41 (27.7) |
| 50–59 | 27 (34.6) | 25 (35.7) | 52 (35.1) |
| 60–69 | 19 (24.4) | 13 (18.6) | 32 (21.6) |
| 70–79 | 1 (1.3) | 4 (5.7) | 5 (3.4) |
| 80–89 | 1 (1.3) | 2 (1.4) | 3 (2.0) |
| Gender | | | |
| Male | 35 (44.9) | 29 (41.4) | 64 (43.2) |
| Female | 43 (55.1) | 41 (58.6) | 84 (56.8) |
| Site of recruitment | | | |
| Semi-urban | 39 (50.0) | 43 (61.4) | 82 (55.4) |
| Urban | 39 (50.0) | 27 (38.6) | 66 (44.6) |
| Marital status | | | |
| Single | 6 (7.7) | 6 (8.6) | 12 (8.1)* |
| Married | 62 (79.5) | 53 (75.7) | 115 (77.7)* |
| Separated/ divorced | 7 (9.0) | 1 (1.4) | 8 (5.4)* |
| Widowed | 3 (3.8) | 9 (12.9) | 12 (8.1)* |
| Unknown | 0 (0) | 1 (1.4) | 1 (0.7) |
| Education | | | |
| Primary or less | 21 (26.9) | 26 (37.1) | 47 (31.8) |
| Secondary level | 11 (14.1) | 5 (7.1) | 16 (10.8) |
| Tertiary level | 44 (56.4) | 37 (52.9) | 81 (54.7) |
| Unknown | 2 (2.6) | 2 (2.9) | 4 (2.7) |
| Lived outside Nigeria in past 10 years | | | |
| Yes | 11 (14.1) | 7 (10.0) | 18 (12.2) |
| No | 67 (85.9) | 62 (88.6) | 129 (87.2) |
| Unknown | 0 (0) | 1 (1.4) | 1 (0.7) |
| Ethnicity | | | |
| Cross Rivers | 35 (44.9) | 42 (60.0) | 77 (52.0)* |
| Hausa and Fulani | 8 (10.3) | 0 (0) | 8 (5.4)* |
| Igbo | 12 (15.4) | 11 (15.7) | 23 (15.5)* |
| Ijaw | 6 (7.7) | 2 (2.9) | 8 (5.4)* |
| Yoruba | 10 (12.8) | 5 (7.1) | 15 (10.1)* |
| Other | 6 (7.7) | 9 (12.9) | 15 (10.1)* |
| Unknown | 1 (1.3) | 1 (1.4) | 2 (1.4) |
| Hypertension | | | |
| Yes | 49 (62.8) | 39 (55.7) | 88 (59.5) |
| No | 18 (23.1) | 18 (25.7) | 36 (24.3) |
| Unknown | 11 (14.1) | 13 (18.6) | 24 (16.2) |
| Diabetes | | | |
| Yes | 31 (39.7) | 23 (32.9) | 54 (36.5) |
| No | 35 (44.9) | 32 (45.7) | 67 (45.3) |
| Unknown | 12 (15.4) | 15 (21.4) | 27 (18.2) |
| Hypertension and diabetes | | | |
| Yes | 19 (24.4) | 17 (24.3) | 36 (24.3) |
| No | 47 (60.3) | 38 (54.3) | 85 (57.4) |
| Unknown | 12 (15.4) | 15 (21.4) | 27 (18.2) |
## Recruitment strategy
Eligible participants were approached by the researcher in the waiting room. The researcher verbally went through the information sheet and obtained verbal consent prior to the administration of the questionnaires. In the semi-urban site, all questionnaires were researcher-administered. In the urban research site, participants were given the option to self-administer their questionnaires and return them to the researcher or have their questionnaires administered by the researcher.
A purposive sample of participants who completed the questionnaires were invited to take part in in-depth interviews and written consent was sought. The intention was to recruit a minimum of 4 participants from each of the 4 categories below: All interviews took place in a vacant consultation room where only the interviewee and the researcher were present. All interviews were audio-recorded and subsequently transcribed verbatim.
## Statin status (explanatory variable)
Participants who were taking a statin at the time of the study were classified as statin users. Non-statin users, included participants who had not been prescribed a statin, and participants who had previously been prescribed a statin but were not taking a statin at the time of the study.
## Demographics and cardiovascular disease risk factors (explanatory variables)
A structured questionnaire was used to elicit demographic information from participants. A medical diagnosis of diabetes, hypertension and statin-use at the time of the study were confirmed by participants prior to the administration of the questionnaire.
## Perception of high cholesterol and future risk of cardiovascular disease (explanatory variables)
The Revised Illness Perception Questionnaire (IPQ-R) was used to assess participants’ perceptions of their high cholesterol [37]. The identity or symptom sub-scale was not included in this study because high cholesterol is asymptomatic [31]. Two items “spiritual causes” and “fate/destiny” were added to the cause subscale of the IPQ-R because religion plays a major role in the way Nigerians perceive the world around them and make sense of illness, death and suffering [38].
Champion’s Health Belief Model Scale was used to assess the participants’ beliefs about their future risk of cardiovascular disease [39]. The health motivation sub-scale was not included in this study as it is not one of the 4 core components of the health belief model [32, 40].
## Adoption of a low fat diet and healthy exercise choices (primary outcome variables)
Prochaska and DiClemente’s model of change was used to assess adoption of a low-fat diet and adoption of healthy exercise choices [41]. Participants classified by this model as belonging to either the pre-contemplation, contemplation or decision stages of change for either health behaviour were regarded as non-adopters of the relevant health behaviour (low fat diet, healthy exercise choices). Participants in the action or maintenance stages of change were regarded as adopters of the relevant health behaviour [41].
## Dietary and exercise behaviours (secondary outcome variables)
Questions from the European Prospective Investigation of Cancer and Nutrition food frequency questionnaire were used to elicit information about the dietary fat consumption [42, 43]. The options of cooking oil in the study questionnaire were altered to reflect the Nigerian diet.
Questions from the UK Department of Health’s General Practice Physical Activity Questionnaire were used to assess participants’ level of physical activity and classify them as either: inactive, moderately inactive, moderately active or active [44].
To ensure that the questionnaire was understandable and culturally acceptable, it was piloted on 10 Nigerians aged between 27-65years.
## Quantitative data analysis
Data from the questionnaires was entered into and analysed using the Statistical Package for Social Sciences version 20. Frequencies were calculated for categorical variables and mean scores were calculated for each of the sub-scales in the Revised Illness Perception Questionnaire and Champion’s Health Belief Model Scale [37, 39, 45]. In accordance with the instructions for the Revised Illness Perception Questionnaire, the total score and not mean score of each cause subscale was calculated [37]. Chi-square tests or Fisher’s exact tests were used to assess between group differences in categorical variables. Independent t-tests or Mann-Whitney tests were used to assess between group differences in continuous variables.
Logistic regression models were used to identify variables that were associated with the adoption of a low-fat diet and the adoption of healthy exercise choices. The minimum sample size required for this study was calculated on the premise that 15 participants were required for each predictor variable included in the regression model [46, 47]. Therefore, the minimum sample size required to perform a reliable logistic regression with 5 predictor variables was 75 participants. An increase in the number of participants recruited was matched with an increase in the number of variables included in the regression model. For all statistical tests, $p \leq 0.05$ was taken as the level of statistical significance.
## Qualitative data collection
The researcher conducted face-to-face, semi-structured interviews with participants in a vacant consultation room. Existing literature was used to develop an interview guide which helped elicit information from interviewees about the influence of statin-use on: (i) their adoption of a low-fat diet and/or healthy exercise behaviours; (ii) their perception of high cholesterol; and (iii) their perception of the future risk of cardiovascular disease. The interview guide was piloted on a small sample of Nigerians aged between 27-65years. All interviews were audio recoded, lasted between 30–90 minutes and were anonymised and transcribed verbatim by the researcher (JC) who is familiar with the use and interpretation of Pidgin English and Nigerian slang words.
## Qualitative data analysis
All interview transcripts were coded using the NVivo 10 software package. Transcripts were re-read to ensure transcription was verbatim and to familiarise researchers with the interview data. Braun and Clark’s 6 phases of thematic analysis were used to guide the analysis of interview data and ensure a rigorous thematic analysis was conducted [48].
The initial coding process was inductive and focused on identifying themes that were emerging from the data. The research questions were not at the forefront of the coding process. However, an awareness of the various components of the Common Sense Model guided the choice of theme names. To ensure that the coding process was consistent, the emerging codes were used to generate a codebook. Codes were agreed upon, and collated into themes, which in turn were categorised into clusters by JC, KH and AH.
## Dietary patterns
Most participants ($76\%$) reported that they ate fried foods at home on a weekly basis, while only $35\%$ reported doing so away from their homes. As shown in Table 2, $69\%$ of participants reported that they had adopted a low-fat diet. Over half of these participants ($59\%$) had been doing so for a least a year prior to the study. Statin and non-statin users did not significantly differ in their reported adoption of a low-fat diet or any other reported dietary behaviours.
**Table 2**
| Variable | Statin users n = 78 (52.7%) | Non-statin users n = 70 (47.3%) | Overall n = 148(100%) |
| --- | --- | --- | --- |
| Weekly fried food consumption at home | | | |
| Yes | 59 (75.6) | 54 (77.1) | 113 (76.4) |
| No | 18 (23.1) | 16 (22.9) | 34 (23.0) |
| Unknown | 1 (1.3) | 0 (0) | 1(0.7) |
| Weekly fried food consumption outside the home | | | |
| Yes | 30 (38.5) | 22 (31.4) | 52 (35.1) |
| No | 48 (61.5) | 47 (67.1) | 95 (64.2) |
| Unknown | 0 (0) | 1 (1.4) | 1(0.7) |
| Think current diet is low fat? | | | |
| Yes | 51 (65.4) | 38 (54.3) | 89 (60.1) |
| No | 20 (28.2) | 23 (32.9) | 43 (29.1) |
| Unknown | 7 (9.0) | 9 (12.9) | 16 (10.8) |
| Ever decreased fat in diet? | | | |
| Yes | 62 (79.5) | 53 (75.7) | 115 (77.7) |
| No | 16 (20.5) | 17 (24.3) | 33 (22.3) |
| Currently decreasing fat in diet? | | | |
| Yes (Adopters) | 57 (73.1) | 45 (64.3) | 102 (68.9) |
| No (Non-adopters) | 20 (25.6) | 25 (35.7) | 45 (30.4) |
| Unknown | 1 (1.3) | 0 (0) | 1 (0.7) |
| Adopters only | | | |
| Duration of decreasing fat in diet | | | |
| <30 days | 4 (7.0) | 3 (6.7) | 7 (6.9) |
| 1–6 months | 12 (21.1) | 11 (24.4) | 23 (22.5) |
| 7–12 months | 7 (12.3) | 2 (4.4) | 9 (8.8) |
| >1 year | 32 (56.1) | 28 (62.2) | 60 (58.8) |
| Unknown | 2 (3.5) | 1 (2.2) | 3(2.9) |
| Non-Adopters only | | | |
| Considered decreasing dietary fat in the past month? | | | |
| Yes | 0 (0) | 3 (12.0) | 3 (6.7) |
| No | 19 (95.0) | 17 (68.0) | 36 (80.0) |
| Unknown | 1 (5.0) | 5 (20.0) | 6 (13.3) |
Urban dwellers were significantly more likely to report that they ate fried foods on a weekly basis than semi-urban dwellers ($65\%$ vs $91\%$, X2 = 15.61, $p \leq 0.001$). The odds of reporting the consumption of fried foods outside the home was 16 times higher in urban dwellers than in semi-urban dwellers ($11\%$ vs $65\%$, Fisher’s exact = 51.41, $p \leq 0.001$). Nevertheless, the odds of reportedly adopting a low-fat diet was 6 times higher in urban dwellers than in semi-urban dwellers ($55\%$ vs $86\%$, X2 = 18.30, $p \leq 0.001$). They were also significantly more likely to reportedly think they were eating a low-fat diet than semi-urban dwellers ($44\%$ vs $80\%$, X2 = 21.90, $p \leq 0.001$).
## Exercise patterns
As shown in Table 3, Majority of participants ($80\%$) were classified as physically inactive/moderately inactive, reported that they had never increased the frequency/intensity of their exercise ($78\%$), and were classified as non-adopters of healthy exercise behaviours ($85\%$). Statin users were significantly less inactive/moderately inactive than non-statin users ($86\%$ vs $74\%$, X2 = 3.852, $$p \leq 0.05$$).
**Table 3**
| Variable | Statin users n = 78 (52.7%) | Non-statin users n = 70 (47.3%) | Overall n = 148(100%) |
| --- | --- | --- | --- |
| Physical activity level | | | |
| Inactive/moderately inactive | 67 (85.9) | 52 (74.3) | 119 (80.4)* |
| Moderately active/active | 10 (12.8) | 18 (25.7) | 28 (18.9) |
| Unknown | 1 (1.3) | 0 (0) | 1 (0.7) |
| Ever increased frequency/intensity of exercise? | | | |
| Yes | 18 (23.1) | 14 (20.0) | 32 (21.6) |
| No | 60 (76.9) | 56 (80.0) | 116 (78.4) |
| Currently doing more exercise? | | | |
| Yes (adopters) | 13 (16.7) | 11 (15.7) | 24 (16.2) |
| No (non-adopters) | 65 (83.3) | 59 (84.3) | 124 (83.8) |
| Adopters only | | | |
| Duration of increasing frequency/intensity of exercise | | | |
| <30 days | 6 (46.2) | 1 (9.1) | 7(29.2) |
| 1–6 months | 3 (23.1) | 4 (36.4) | 7 (29.2) |
| 7–12 months | 0 (0) | 1 (9.1) | 1 (4.2) |
| >1 year | 4 (30.8) | 5 (45.5) | 9 (37.5) |
| Non-adopters only | | | |
| Considered increasing frequency/intensity of exercise in the past month? | | | |
| Yes | 13 (20.0) | 16 (27.1) | 29 (23.4) |
| No | 52 (80.0) | 40 (71.2) | 94 (75.8) |
| Unknown | 0 (0) | 1 (1.7) | 1 (0.8) |
Urban and semi-urban dwellers did not significantly differ in their levels of physical activity or reported adoption of healthy exercise behaviours. However, significantly more urban dwellers reported that they had at some point in time increased their frequency/intensity of exercise (OR = 2.5, $30\%$ vs $15\%$, X2 = 5.30, $$p \leq 0.027$$). Urban dwelling non-adopters were also 15 times more likely to report that they had considered adopting healthy exercise behaviours in the previous month than semi-urban dwelling non-adopters ($48\%$ vs $6\%$, X2 = 30.00, $p \leq 0.001$).
## Perception of high cholesterol
As shown in Table 4, participants reportedly believed their high cholesterol could be controlled by their own behaviours (x¯ = 3.9, SD = 0.62) and by taking a statin (x¯ = 3.7, SD = 0.64). However, perceptions of the behavioural control were stronger than perceptions of statin control. Correspondingly, the most important cause of high cholesterol reported by participants was lifestyle causes, followed by biomedical causes. The low score obtained on the timeline acute/chronic subscale, indicates that participants perceived their high cholesterol as an acute condition.
**Table 4**
| Variables | Statin | Non-statin | Overall |
| --- | --- | --- | --- |
| | users | users | |
| | n = 75 (SD) | n = 64 (SD) | n = 139 (SD) |
| Mean perceived timeline acute/chronic | 2.12 (0.67) | 2.31 (0.70) | 2.21 (0.66) |
| Mean perceived timeline cyclical | 3.16 (0.84) | 3.18 (0.85) | 3.17 (0.84) |
| Mean perceived consequences | 2.94 (0.68) | 2.94 (0.65) | 2.94 (0.66) |
| Mean perceived personal control | 3.87 (0.68) | 3.84 (0.58) | 3.85 (0.62) |
| Mean perceived statin control | 3.78 (0.66) | 3.57 (0.59) | 3.68 (0.64) ** |
| Mean perceived emotional response | 2.95 (0.95) | 3.03 (0.93) | 2.99 (0.94) |
| Mean perceived illness coherence | 3.24 (1.13) | 3.19 (1.13) | 3.22 (1.12) |
| Total perceived biomedical cause£ | 13.99 (2.90) | 14.06 (3.91) | 14.02 (3.39) |
| Total perceived spiritual cause€ | 7.08 (3.37) | 6.55 (2.98) | 6.83 (3.20) |
| Total perceived lifestyle cause$ | 23.51 (4.35) | 23.97 (4.86) | 23.72 (4.58) |
Statin users and non-statin users reported stronger perceptions of personal control than statin control of high cholesterol. This indicates that both groups reportedly believed that their behaviours could control their high cholesterol better than statin-use. However, statin users reported significantly stronger statin control perceptions than non-statin users (x¯ = 3.8 vs x¯ = 3.6, $U = 1721.500$, $$p \leq 0.003$$).
Urban dwellers reportedly thought that they had a poor understanding of high cholesterol compared to semi-urban dwellers (x¯ = 3.6 vs x¯ = 2.9 $U = 1497.000$, $p \leq 0.001$). They also reported significantly stronger perceptions that high cholesterol is an acute condition (x¯ = 2.0 vs x¯ = 2.4 $U = 1587.500$, $p \leq 0.001$), but they perceived it to be less predictable i.e. less cyclical than semi-urban dwellers (x¯ = 2.9 vs x¯ = 3.4, $U = 1626.500$, $$p \leq 0.002$$).
## Perception of future risk of cardiovascular disease
As shown in Table 5, participants reportedly perceived healthy lifestyle choices (x¯ = 4.00, SD = 0.65) to be more beneficial than statin-use for the prevention of cardiovascular disease (x¯ = 3.8, SD = 0.70). However, they thought there were more barriers to making healthy lifestyle choices than to taking a statin. The low score obtained on the perceived susceptibility sub-scale indicates that participants did not perceive themselves to be at risk of cardiovascular disease.
**Table 5**
| Variable | Statin | Non-statin | Overall |
| --- | --- | --- | --- |
| | users | users | |
| | n = 71 (SD) | n = 62 (SD) | n = 133 (SD) |
| Perceived susceptibility | 2.44 (0.84) | 2.25 (0.80) | 2.35 (0.83) |
| Perceived severity | 3.10 (0.68) | 2.86 (0.75) | 2.99 (0.72)* |
| Perceived benefits of statins | 3.88 (0.67) | 3.72 (0.73) | 3.80 (0.70) |
| Perceived benefits of healthy lifestyle choices | 4.00 (0.65) | 4.01 (0.66) | 4.00 (0.65) |
| Perceived barriers to statins | 2.31 (0.72) | 2.56 (0.79) | 2.43 (0.76)* |
| Perceived barriers to healthy lifestyle choices | 2.76 (0.76) | 2.69 (0.74) | 2.73 (0.75) |
Statin users reportedly perceived significantly fewer barriers to statin-use for cardiovascular disease prevention (x¯ = 2.3 vs x¯ = 2.6, $U = 1715.000$, $$p \leq 0.027$$), and perceived cardiovascular disease to be significantly more severe than non-statin users (x¯ = 3.1 vs x¯ = 2.9, $U = 1745.500$, $$p \leq 0.040$$).
Urban dwellers reportedly perceived cardiovascular disease to be significantly more severe than semi-urban dwellers (x¯ = 3.2 vs x¯ = 2.8, $U = 1707.500$, $$p \leq 0.027$$). They also thought they were fewer barriers to (x¯ = 2.0 vs x¯ = 2.8, $U = 1815.000$, $$p \leq 0.840$$) and significantly more benefits of the adoption of healthy lifestyle for cardiovascular disease prevention than semi-urban dwellers (x¯ = 4.1 vs x¯ = 3.9, $U = 1635.500$, $$p \leq 0.007$$).
## Factors associated with the adoption of a low-fat diet
Of the 148 participants recruited, $86\%$ completed all sections of the questionnaire that were entered into a logistic regression model where the dependent variable was adoption of a low-fat diet and independent variables were: statin-status, gender, research site, physical activity level, perceived statin control of high cholesterol, perceived barriers to statin-use for cardiovascular disease prevention, and perceived severity of cardiovascular disease.
The logistic regression model was statistically significant (X2 = 25.822, $$p \leq 0.001$$) and correctly classified $73\%$ of cases. The model explained between $18.3\%$ (Cox and Snell R2) and $26.0\%$ (Nagelkerke R2) of variance in the adoption of a low-fat diet, this highlights the importance of exploring the role of other factors using qualitative research methods. The Hosmer and Lemeshow test of fit indicated that the model was a good fit to the data (X2 = 6.00, $$p \leq 0.647$$).
As shown in Table 6, only research site and perceived statin control of high cholesterol made statistically significant contributions to the model. The odds of adopting a low-fat diet was 5 times greater in participants recruited from the urban research site and increased as reported perceived statin control of high cholesterol increased.
**Table 6**
| Variables | B | S.E | Wald | P value | OR | 95%CI |
| --- | --- | --- | --- | --- | --- | --- |
| Statin status (statin-use) | -0.2 | 0.46 | 0.2 | 0.66 | 0.82 | 0.34–1.96 |
| Gender (Male) | 0.45 | 0.48 | 0.87 | 0.35 | 1.56 | 0.61–4.00 |
| Research site(semi-urban) | -1.51 | 0.52 | 8.9 | 0.003** | 0.21$ | 0.08–0.59$ |
| Physical activity level (inactive/moderately inactive) | -0.2 | 0.6 | 0.11 | 0.74 | 0.82 | 0.25–2.66 |
| Statin control of high cholesterol | 0.85 | 0.36 | 5.6 | 0.018** | 2.33 | 1.16–4.69 |
| Barriers to statins-use to prevent CVD | -0.33 | 0.3 | 1.24 | 0.27 | 0.72 | 0.40–1.29 |
| Perceived severity of CVD | 0.41 | 0.38 | 1.19 | 0.28 | 0.28 | 0.72–3.16 |
The model assessing factors associated with the adoption of healthy exercise behaviours was not statistically significant (X2 = 9.674, $$p \leq 0.378$$) and will not be presented in this article.
## Qualitative results
Eight participants were interviewed; 4 statin users (3 classified as non-adopters in quantitative study) and 4 non-statin users (1 non-adopter). The characteristics of interviewees are presented in Table 7.
**Table 7**
| Participants | Recruitment site | Healthy lifestyle adoption classification from quantitative study |
| --- | --- | --- |
| Statin users | Statin users | Statin users |
| Male 1 | Semi-urban site | Non-adopter |
| Male 2 | Semi-urban site | Diet and exercise adopter |
| Female 1 | Semi-urban site | Non-adopter |
| Female 5 | Semi-urban site | Non-adopter |
| Non-statin users | Non-statin users | Non-statin users |
| Female 2 | Semi-urban site | Diet only adopter |
| Female 3 | Semi-urban site | Non-adopter |
| Female 4 | Semi-urban site | Diet only adopter |
| Female 6 | Urban site | Diet and exercise adopter |
Five main themes emerged from the analysis of participant interviews, namely: (i) consequences, (ii) cause, (iii) control, (iv) “for your own good”, and (v) “the whole world will talk”. As shown in Fig 1, these themes clustered into 2 areas of discussion, namely: discussions about the medical world of high cholesterol, and discussions about the participant as an individual in his/her social world. Themes regarding the medical world of high cholesterol were named according the aspect of the high cholesterol under discussion. Themes regarding the individual in his/her social world were named using a phrase participants’ themselves used. Each theme and subsequent subthemes are described below. Relevant interview extracts are presented in Table 8.
**Fig 1:** *Thematic map.* TABLE_PLACEHOLDER:Table 8
## Themes 1: Consequences
Both male interviewees identified heart attacks, strokes and death as potential consequences of high cholesterol. These perceived consequences appeared to encourage statin-use and the adoption of healthy lifestyle choices as illustrated by the quote from Male 2.
## Themes 2: Cause
Interviewees attributed their high cholesterol to 2 factors, namely: (i) medical factors and (ii) lifestyle factors.
## Sub-theme 1: Medical factors
Two interviewees (Male 2, Female 3) attributed their high cholesterol to genetics factors and recounted their family history to the interviewer without being prompted. They both cited statin-use and the adoption of healthy lifestyle choices simultaneously as a suitable strategy to manage their high cholesterol. This demonstrates a synergistic relationship between statin-use and the adoption of healthy lifestyle choices. It should be noted that Male 2 was classified as a statin-user and adopted of both a low fat diet and healthy exercise behaviours whilst Female 3 was classified as a non-statin user and non-adopter of both a low fat diet and healthy exercise behaviours.
## Sub-theme 2: Lifestyle factors
Lifestyle factors, predominantly diet, were the most common cause of high cholesterol discussed by interviewees. Most of these participants also cited lifestyle as the best way to control their high cholesterol, demonstrating a congruous cause-control relationship. No female interviewee cited physical inactivity as a cause of their high cholesterol. Rather, they cited the consumption of sugary foods and carbohydrates, consumption of fatty foods, eating late, and increased alcohol consumption as a result of stress as the cause of their high cholesterol.
## Themes 3: Control
The control beliefs and descriptions of the changes interviewees discussed making or not making in an attempt to control their high cholesterol fall in 4 subthemes, namely: (i) no control; (ii) statin-use; (iii) lifestyle changes; and (iv) “spiritual something”
## Sub-theme 1: No control
Only 2 interviewees mentioned that they made no effort to control their high cholesterol. Female 6 cited work stress as the cause of her increased alcohol consumption and poor dietary habits. She explained that she previously had no control over her busy schedule and thus was initially unable to make healthy lifestyle changes. Her work circumstances had however changed by the time of the study. Whilst she felt unable to make changes due to her perceived lack of agency, Male 1 refused to make changes because of his presence of agency. He explained that he had not made any lifestyle changes to control his high cholesterol because he was happy as he was and felt no-one should tell him what to do with his body.
## Sub-theme 2: Statin-use
Interviewees expressed mixed statin control beliefs and demonstrated 3 different relationships between statin-use and the adoption of healthy lifestyle choices. Some interviewees voiced positive statin control perceptions, and also had very positive beliefs about regular hospital visits and medical checks. Most of these interviewees were female statin users who had not adopted a low-fat diet or healthy exercise behaviours. They also voiced weak lifestyle control perceptions because they perceived themselves as unable or struggling to adopt healthy lifestyle choices. These interviewees demonstrate an antagonistic relationship between medical control and lifestyle control of high cholesterol, in a manner that facilitated statin-use, but hindered the adoption of healthy lifestyle choices.
Similarly, most of the interviewees who expressed weak statin control perceptions, voiced strong lifestyle control perceptions, were non-statin users, and were classified as adopters of a low-fat diet and/or healthy exercise behaviours. These interviewees demonstrate an antagonistic relationship between medical control and lifestyle control of high cholesterol in a manner that hindered statin-use but facilitated the adoption of healthy lifestyle choices. Their negative statin control beliefs were fuelled by concerns about the side-effects of statin use, dislike of lifelong medication use, and the belief that their high cholesterol was caused by lifestyle factors, and therefore should be controlled by lifestyle factors (congruous cause-control perceptions).
Only 1 interviewee voiced positive statin control and lifestyle control perceptions simultaneously and demonstrated a synergistic relationship between statin use and the adoption of healthy lifestyle choices. He (Male 2) attributed his high cholesterol to genetic factors, believed statin-use and lifestyle changes could control his high cholesterol, and was classified as a statin user, and an adopter of both a low-fat diet and healthy exercise behaviours.
## Sub-theme 3: Lifestyle changes
Most interviewees described attempting to make a variety of dietary changes, such as: reducing their consumption of fatty foods, fast foods, sugary foods, fizzy drinks, alcohol, salt and carbohydrates. They also discussed changing their meal times and improving their food purchasing habits. Only a few interviewees discussed attempting to adopt healthy exercise behaviours. Those who did, mentioned that they tried to walk or jog more. There were also mentions of swimming, skipping and using a treadmill.
Male interviewees believed that they could make any lifestyle changes they chose to make. Female interviewees however, described several barriers to adopting healthy lifestyle choices and recounted more barriers to make healthy exercise changes than to adopting healthy dietary choices. They cited the pleasure they derived from eating, and the time demands of work life and family obligations, as factors that hindered their adoption of healthy dietary choices. These time demands were also said to hinder their adoption of healthy exercise behaviour, as well as: the body aches and pains that accompany exercise, physical limitations as a result of other comorbidities, not feeling like exercising, the constant commitment required to keep the weight off, and the belief that exercise is in itself stressful and could have adverse health effects as illustrated by the quote from Female 3 in Table 8.
## Sub-theme 4: “Spiritual something.”
Mutual to all interviewees was the belief that forces greater than themselves had the ability to impact their lives. They all either discussed the role of prayer, God and/or expressed fatalistic beliefs. Several interviewees described using prayer alongside medical and/or lifestyle control strategies. A few of whom stated that prayer, whilst important to them, should not usurp statin-use and/or the adoption of healthy lifestyle choices. Fatalistic beliefs however, appeared to hinder statin-use and the adoption of healthy lifestyle choices. The belief that life is unpredictable or predestined, appeared to allow interviewees to externalise control of their high cholesterol. It also diminished their perceptions of the consequences of high cholesterol potentially hindering statin-use and/or the adoption healthy lifestyle choices.
## Themes 4: “For your own good”
This theme discusses the personal factors interviewees cited as facilitators of, or barriers to their statin-use and adoption of healthy lifestyle choices. These factors appear to differ by gender. Majority of female interviewees described themselves as fat, heavy or healthy and expressed their desire to “trim down” or maintain their current weight and fitness levels. However, they emphasised that they did not want to become slim or too slim. These body image concerns and concerns for their health appeared to be the main personal factors which both encouraged and hindered their adoption of healthy lifestyle choices.
Male interviewees expressed the importance of: (i) their choice/decision, and (ii) their capability to adopt healthy lifestyle choices. They both emphasised that they were capable of making lifestyle and medication changes, if it was something they desired, or had chosen to do. They narrated instances of openly disagreeing with advice from medical professionals and refusing to adopt medication or lifestyle changes which they did not deem necessary or beneficial for themselves. In contrast, only a few female interviewees described themselves as drivers of their dietary behaviours and medication-use. Females in this study mainly discussed feeling compelled or obligated to take their statins because it had been prescribed by a doctor and was for their own good.
## Themes 5: “The whole world will talk”
This theme discusses the social pressures exerted by, and the support received from the members of the social world of interviewees which facilitated or hindered their statin-use and/or adoption of healthy choices. This theme consists of 2 themes, namely: (i) “I don’t want to expose myself”; and (ii) “you can’t do it alone”.
## Sub-theme 1: “I don’t want to expose myself.”
Most participants narrated accounts of tailoring their behaviours to suit the expectations of members of their social world and maintain what they perceived to be a favourable public image. Male interviewees discussed hiding their health issues and adoption of healthy lifestyle, or simply not adopting healthy lifestyle choices all together in a bid to protect their public image. Female interviewees explained that losing weight may make people think they were sick, had HIV, or were experiencing life stress. They appeared to perceive being “slim” negatively. Rather, they said they wanted to “just trim down”, “just to reduce”, “just to maintain shape” and “just want to feel healthy”. Their choice of words seem to indicate that they only wanted to make small or minor changes and avoided losing too much weight. This illustrates how the perceptions of the ideal body image held by oneself and one’s social world hinders the adoption of healthy lifestyle choices.
## Sub-theme 2: “You can’t do it alone.”
A few interviewees mentioned that they received social support from their parents in the form of advice on healthy living, and financial support to aid access to medical advice and the purchasing of medications.
Male interviewees described the practical and emotional support they received from their wives as a facilitator of their adoption of healthy lifestyle choices. Female interviewees cited the body image preferences of their husbands, and the time constraints of their marital and childcare obligations as factors that hindered their adoption of healthy lifestyle choices. They did however describe their desire to teach their children healthy lifestyle behaviours and ensure their family remained healthy as a facilitator of their adoption of healthy lifestyle choices.
Male and female interviewees also differed in their discussions of the role group membership/friendships played in their adoption of healthy lifestyle choices. Male interviewees described their friendship groups as a hindrance. Male 2 explained that he knew his friends were not a good influence on his healthy lifestyle, but they made him happy and he enjoyed their companionship. Female interviewees described their friendship groups as supportive. Their friendships appeared to offer them information about locally available healthy options and opportunities, as well as companionship in their journeys to adopt healthy lifestyle choices.
## Discussion
The primary aim of this study was to explore whether statin-use influenced the adoption of healthy dietary and exercise choices. The quantitative study found that statin-use in itself did not influence the adoption of a low-fat diet. However, statin-users were more physically inactive/moderately inactive than non-statin users.
## Statin-use and diet
Our quantitative study and three American studies found that statin-use was not independently associated with the adoption of a low-fat diet [19, 20, 49]. In contrast, a Swedish study found that statin-users had better dietary and exercise behaviours than non-statin users [23]. However, the statin users in the Swedish study had significantly higher cardiovascular risk profiles, had experienced more cardiovascular events than their non-statin using comparators, and were recruited from a pharmacy [23]. Consequently, they have may have received more lifestyle advice, may have stronger perceptions of the consequences of high cholesterol, and may represent an adherent population.
## Statin-use and physical activity
Our quantitative study found that significantly more statin-users were classified as physically inactive/moderately inactive than non-statin users. This contrasts with the findings of 2 American studies which found no difference in the physical activity levels of participants using lipid-lowering medications versus those not taking lipid-lowering medications [19, 49]. This contrast may have occurred because participants in this study, as identified in our qualitative study, attributed their high cholesterol to dietary factors therefore prioritised making dietary changes (congruous cause-control perceptions). Consequently, statin users may have felt that statin-use alongside the adoption of healthy dietary behaviours were adequate strategies to control their high cholesterol. This may have allowing them to neglect the adoption of healthy exercise behaviours.
It should be noted that participants in this study (statin users and non statin statin users) demonstrated a preference for making dietary changes than making exercise changes. Our quantitative study found that although majority of our participants had adopted a low-fat diet, and had done so for at least a year prior to this study, majority of participants were classified as physically inactive/moderately inactive, had never adopted healthy exercise behaviours, and had not considered doing so in the month prior to this study. Similarly, our qualitative study revealed that most interviewees discussed making healthy dietary changes but only a few recounting making any exercise changes. Research on the interplay between dietary behaviours and physical activity behaviours is inconsistent, inconclusive and limited [50, 51]. However, there is some evidence of an asymmetrical relationship between dietary and exercise behaviours, where exercise is used to compensate for, or offset poor dietary behaviours, although this is not always the case [51, 52]. Dietary behaviours therefore may be seen as the major, whilst exercise the minor behaviour. An asymmetrical diet and exercise relationship observed in this study may have co-existed or may have been fuelled by the attribution of high cholesterol to dietary factors resulting in the preferences for making healthy dietary changes than healthy exercise changes.
## Interactions between statin-use and the adoption of healthy lifestyle choices
This study also aimed to explore whether statin-use influenced the adoption of healthy lifestyle choices by changing the way urban and semi-urban Nigerians thought about their high cholesterol and their future risk of cardiovascular disease. Our quantitative study found that statin-use in itself did not significantly influence the adoption of a low-fat diet. However, our qualitative study found 3 different types of interactions between statin-use and adoption of healthy lifestyle choices: (i) a synergistic relationship between statin-use and the adoption of health lifestyle choices, (ii) an antagonist relationship between statin-use and the adoption of health lifestyle choices which favoured the statin-use, and (iii) an antagonist relationship between statin-use and the adoption of health lifestyle choices which favoured the adoption of healthy lifestyle choices. These findings and the conflicting data from other studies [19–23] suggest that the relationship between statin-use and the adoption of healthy lifestyle choices is complex and may vary depending on the context in which it occurs.
Statin-use in itself did not significantly influence the adoption of a low-fat diet in our quantitative study. However, the odds of adopting a low-fat diet increased as perceived statin control of high cholesterol increased. This is consistent with the synergistic relationship between statin-use and the adoption of health lifestyle choices identified in our qualitative study. In our qualitative study, this synergistic relationship occurred in a context where high cholesterol was not solely attributed to dietary factors. Research also offers 3 possible explanations for why some participants thought statin-use worked in unison with the adoption of healthy lifestyle choices to manage their high cholesterol. The first explanation may be that medical advice encouraging the adoption of healthy lifestyle choices occurred at the same time statins were prescribed [19]. Secondly, the prescription of a statin may increase risk perception and serves as a wake-up call which facilitates the adoption of healthy lifestyle choices [19, 23]. Thirdly, a dislike for medication-use or concern about long-term medication-use may encourage both statin-use and the adoption of healthy lifestyle choices in an attempt to lower cholesterol cease medication-use and return to optimal health [19].
The second relationship we found between statin-use and the adoption of healthy lifestyle choices was an antagonistic relationship which favoured the adoption of healthy lifestyle choices. This relationship occurred in the context where interviewees disliked long-term medication-use, were concerned about the side-effects of statin-use, and believed that high cholesterol was caused by and thus best controlled by lifestyle factors (congruous cause-control beliefs). This relationship was mainly voiced by interviewees who were non-statin users and is consistent with the quantitative finding that non-statins users had significantly weaker statin control beliefs than statin users, and perceived significantly more barriers to statin-use for the prevention of cardiovascular disease than non-statin users. It is well-documented that statin-use is frequently declined or discontinued by patients around the world because of concerns about side-effects [53, 54]. Existing studies which explored perceptions of statin-use have also found that a dislike of general medication-use not just statin-use specifically fuels a preference for adopting healthy lifestyle choices or other non-pharmaceutical methods such as the use of herbal remedies and supplements over statin-use [55–60].
The third relationship we found between statin-use and the adoption of healthy lifestyle choices was an antagonistic relationship which facilitated statin-use and hindered the adoption of healthy lifestyle choices. This relationship occurred in a context where interviewees perceived themselves as unable to adopt healthy lifestyle choices due to a weak sense of agency and/or competing priorities. This is consistent with our quantitative finding that despite being more perceived as beneficial for the prevention cardiovascular disease, the adoption of healthy lifestyle choices was thought to have more barriers than statin-use. Similarly, Mann et al. reported that some participants begun statin-use because of their perceived inability to adopt healthy lifestyle choices [20]. Statin-use thus appears to be viewed by some as an easier alternative to the adoption of healthy lifestyle choices rather than as a complementary strategy [20, 49, 61]. Another possible explanation for the preference for statin-use over the adoption of healthy lifestyle choices may be that improvements in lipid profile following the commencement of statin-use may foster/encourage the perception that cholesterol levels have/are being adequately managed by statin-use and there is no need to employ additional strategies such as the adoption of healthy lifestyle choices [49]. This is problematic as the effects of statins are enhanced by the adoption of healthy lifestyle choices [62], and healthy lifestyle behaviours are beneficial for the control non-communicable disease and overall health [5, 8, 27].
## Gender and social membership
This study also found that the factors which influenced the adoption of healthy lifestyle choices differed by gender. Female interviewees described how their desire to be trim but not too slim (body image concerns) infleunced their adoption of healthy lifestyle choices. Jemisenia et al. describe this as the desire to look plump i.e. not too fat, but not too slim [63]. Existing research evidence demonstrates a preference for plumper body sizes among females in Nigeria and many other developing countries [63–67]. It should be noted that being fat and flabby is not perceived as desirable. Rather, beauty, fertility, wealth and health are equated with being shapely [63, 65–67]. The preference for plumpness appears to hold even in females who have high educational achievements and are aware of the link between excess body weight and poor health outcomes [64]. This suggests and is consistent with the narrative among female interviewees in our study, that the cultural preference for plumpness may override the need to lose a significant amount of weight for health benefits and hinder the adoption of healthy lifestyle choices.
Female interviewees also worried that a significant amount of weight loss could taint ones public image by portraying them as unwell, having HIV/AIDS or experiencing difficult life circumstances and this hindered their adoption of healthy lifestyle choices. Indeed it is documented that in in Nigeria, plumpness is often associated with abundance, wealth and health while slimness denotes hunger, poor health, weakness, and HIV/AIDS [65, 66]. HIV/AIDS was sometimes referred to as the slim disease because of its association with significant weight-loss [68]. Male interviewees also demonstrated a desire to to protect or maintain a public image albeit slightly differently. They hid or completely disregarded health behaviours that may portray them as less manly to protect or maintain their public image. This is consistent with literature on hegemonic masculinity which cites a desire for good health, power and dominance as key features [69].
Other features of hegemonic masculinity which emerged from the male narrative as factors that influenced their adoption of healthy lifestyle choices were the ability to make decisions, control, protect and conquer [69, 70]. Hegemonic masculinity has been directly linked to health behaviours in men such as medication-use and dietary behaviours. Many men aspire to this form of masculinity because it is associated with power, assertiveness, success and health [69].
Family and partner support also differed by gender. Whilst family support facilitated of their adoption of healthy lifestyle choices in the male narrative—consistent with existing evidence of the positive influence wives can play on the lifestyle behaviours of men [71, 72]. Family in the female narrative, characterised busy schedules of tasks, obligations, and the body image preferences of their husbands was described as a hinderance to the adoption of healthy lifestyle choices. It is well-documented that in many instances females still play a leading role in childcare and home management, often alongside paid work. This limits their leisure time and ability to adopt healthy lifestyle choices particularly exercise [38, 73]. Nevertheless, the role females played in ensuring that their families were healthy was cited as a facilitator of the adoption of healthy dietary choices.
## Urban and semi-urban living and lifestyle
The factor which contributed the most to the multivariant model assessing the adoption of a low-fat diet, was the site from which participants were recruited. Indeed, the largest differences observed in the adoption of healthy lifestyle choices, and perception of high cholesterol and future cardiovascular disease risk occurred between urban dwellers and semi-urban dwellers.
The quantitative study findings reveal that urban dwellers consumed more fried foods, ate out more often, and were more likely than semi-urban dwellers to have lived outside of Nigeria in the 10-years prior to this study. This demonstrates that the urban dwellers in this study were more exposed to Western diets and had acquired more unhealthy lifestyle behaviours than semi-urban dwellers. This is consistent with research conducted in low-middle income countries which has found that the increased exposure to western diets (rich in salt, sugar and saturated fat) and the built environment of urban areas i.e. the relatively high availability of fast food outlets, and sedentary lifestyles facilitates the adoption unhealthy lifestyle choices and the development of cardiovascular disease and its risk factors [74, 75].
Nevertheless, urban dwellers were more likely than non-urban dwellers to have adopted a low-fat diet, and to consider adopting healthy exercise behaviours. This, alongside the findings that urban dwellers perceived their future cardiovascular disease risk to be more severe, and the adoption of healthy lifestyle choices to be more beneficial for the prevention of cardiovascular disease than the semi-urban dwellers, indicates that urban dwellers were more aware of the negative consequences of a westernised lifestyle and were actively trying to make healthy lifestyle choices. Urban dwellers may also have had more facilities and resources that facilitate the adoption of healthy lifestyle choices available to them than semi-urban dwellers [76].
## Study limitations
Illness perceptions often develop and change with time, and the nature of the relationship between illness perceptions and health behaviour is complex [77]. The former may influence the latter, but the latter may also feedback and influence the former [28, 78]. Collecting data at a single point in time using a quantitative questionnaire simplifies the complex relationship between illness perceptions and health behaviours. Furthermore reliance on the recall of interviewees influences the accuracy of the information obtained [79]. Consequently, no claims can be made about whether certain illness perceptions caused certain health behaviours and vice versa.
Although self-reported measures of lifestyle behaviours are commonly used, they are thought to overestimate adherence to health behaviours [80, 81]. Furthermore, recruiting participants from hospitals, conducting interviews, and collecting data in hospital using a researcher administered questionnaire, introduces social desirability in an adherent population. This could lead to increased discussions centred around the medical world of high cholesterol and an overestimation of adherence to healthy lifestyle choices.
We interviewed a very limited number of men and urban dwellers. This was because many of them had limited time to spend because they had other priorities such as picking up their prescriptions and returning to their daily routines. Their narrative may have enriched our data collection and enabled us to paint a more complete picture of the interaction between statin-use and the adoption of healthy lifestyle choices.
## Conclusion
This study found that the relationship between statin-use and the adoption of healthy lifestyle choices is complex and may vary depending on the context in which it occurs. It identified 3 relationships between statin-use and the adoption of healthy lifestyle choices and the contexts in which they occur. We found that in the context where high cholesterol was not solely attributed to dietary factors, some participants believed that statin-use and the adoption of healthy lifestyle choices worked in unison to manage their high cholesterol. In the context where there is concern about side-effects and a dislike for long-term medication some participants favoured the adoption of healthy lifestyle choices over statin-use. While in a context where participants felt unable or struggled to make healthy lifestyle choices, statin-use was perceived as an easier alternative. The latter poses a problem as the adoption of healthy lifestyle choices is beneficial not just for the management of high cholesterol, but also for the prevention and management of all non-communicable disease and overall health. Efforts should be made to identify people who may hold the latter set of beliefs and attempts should be made to help identify and tackle their perceived barriers to the adoption of healthy lifestyle choices and address their reliance on statin control for the management of their high cholesterol.
We also found that the adoption of healthy exercise choices among study participants was uncommon and the prevalence of physical inactivity was high. This highlights the need for public health interventions aimed at increasing physical activity for the prevention of non-communicable disease and improvement of overall health. Such interventions should take into account the influences of gender and physical environment on the adoption of healthy lifestyle choices.
Finally, this study found that urban dwellers were more likely to have adopted and thought about adopting healthy lifestyle choices than semi-urban dwellers. This demonstrates a growing awareness of the negative future impacts of Westernised lifestyle behaviours in urban areas yet highlights the need for public health interventions in semi-urban areas in order to curtail the adoption of unhealthy lifestyle choices and diminish the rise of cardiovascular disease in Nigeria.
## References
1. Maher D, Smeeth L, Sekajugo J. **Health transition in Africa: practical policy proposals for primary care**. *Bulletin of the World Health Organization* (2010.0) **88** 943-8. DOI: 10.2471/BLT.10.077891
2. BeLue R, Okoror TA, Iwelunmor J, Taylor KD, Degboe AN, Agyemang C. **An overview of cardiovascular risk factor burden in sub-Saharan African countries: a socio-cultural perspective.**. *Globalization and Health* (2009.0) **5** 10. DOI: 10.1186/1744-8603-5-10
3. Ekpenyong C, NE U, Akpan E, Samson T. **Double burden, non-communicable diseases and risk factors evaluation in sub-Saharan Africa: The Nigerian experience**. *European Journal of Sustainable Development* (2012.0) **1** 249-70
4. 4World Health Organisation. Noncommunicable diseases (NCD) Country profiles, Nigeria. 2018 [cited 2019 16 Jul 2019]. Available from: https://www.who.int/nmh/countries/nga_en.pdf?ua=1.
5. Akinroye K, Olukosi Y, Atinmo T, Omueti O, Babasola C, Idigbe O. **Consensus summit: lipids and cardiovascular health in the Nigerian population**. *European Journal of Nutrition & Food Safety* (2017.0) 109-19
6. Awosan J.. *Currents of Thought in African Sociology and the Global Community: How to Understand Research Findings in the Context of Sociological Perspectives* (2009.0)
7. Essien OE, Andy J, Ansa V, Otu AA, Udoh A. **Coronary artery disease and the profile of cardiovascular risk factors in South South Nigeria: a clinical and autopsy study**. *Cardiology Research and Practice* (2014.0) 2014. DOI: 10.1155/2014/804751
8. Mendis S, Puska P, Norrving B. *Global atlas on cardiovascular disease prevention and control:* (2011.0)
9. Okpechi IG, Chukwuonye II, Tiffin N, Madukwe OO, Onyeonoro UU, Umeizudike TI. **Blood pressure gradients and cardiovascular risk factors in urban and rural populations in Abia State South Eastern Nigeria using the WHO STEPwise approach**. *PLOS ONE* (2013.0) **8** e73403. DOI: 10.1371/journal.pone.0073403
10. Sani MU, Wahab KW, Yusuf BO, Gbadamosi M, Johnson OV, Gbadamosi A. **Modifiable cardiovascular risk factors among apparently healthy adult Nigerian population-a cross sectional study.**. *BMC Research Notes* (2010.0) **3** 11. DOI: 10.1186/1756-0500-3-11
11. **National heart, lung and blood institute Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatmentof High Blood Cholesterol in Adults (Adult Treatment Panel III) Final Report.**. *Circulation* (2002.0) **106** 3143-421. PMID: 12485966
12. 12National Institute for Health and Care Excellence. Lipid Modification: Cardiovascular Risk Assessment and the Modification of Blood Lipids for the Primary and Secondary Prevention of Cardiovascular Disease.
Clinical guideline [CG67] London: National Institute for Health and Care Excellence (UK). Copyright (c) National Clinical Guideline Centre, 2014.; 2014 [16 Jul 2019]. Available from: https://www.nice.org.uk/guidance/cg67.. *Lipid Modification: Cardiovascular Risk Assessment and the Modification of Blood Lipids for the Primary and Secondary Prevention of Cardiovascular Disease.* (2014.0) **2014**
13. 13World Health Organisation. Prevention of cardiovascular disease: World Health Organization; 2007 [16 Jul 2019]. Available from: https://www.who.int/cardiovascular_diseases/guidelines/Full%20text.pdf.. *Prevention of cardiovascular disease* (2007.0)
14. Ganjali S, Bianconi V, Penson PE, Pirro M, Banach M, Watts GF. **Commentary: Statins, COVID-19, and coronary artery disease: killing two birds with one stone**. *Metabolism-Clinical and Experimental* (2020.0) 113
15. Johnston T, Korolenko T, Pirro M, Sahebkar A. **Preventing cardiovascular heart disease: Promising nutraceutical and non-nutraceutical treatments for cholesterol management**. *Pharmacological research* (2017.0) **120** 219-25. DOI: 10.1016/j.phrs.2017.04.008
16. Kajinami K, Takekoshi N, Saito Y. **Pitavastatin: efficacy and safety profiles of a novel synthetic HMG‐CoA reductase inhibitor.**. *Cardiovascular Drug Reviews* (2003.0) **21** 199-215. DOI: 10.1111/j.1527-3466.2003.tb00116.x
17. Kastelein JJ. **Dyslipidaemia in perspective**. *The Lancet* (2014.0) **384** 566-8
18. Opie LH, Gersh BJ. *Drugs for the Heart E-Book.* (2011.0)
19. Mann DM, Natarajan S. **Inverse relationship between lipid-lowering drugs and saturated fat intake in US adults.**. *Cardiovascular Drugs and Therapy* (2007.0) **21** 109-15. DOI: 10.1007/s10557-007-6013-7
20. Mann DM, Allegrante JP, Natarajan S, Montori VM, Halm EA, Charlson M. **Dietary indiscretion and statin use**. *Mayo Clinic Proceedings* (2007.0) **82** 951-3. DOI: 10.4065/82.8.951
21. Redberg RF. **Statins and weight gain**. *JAMA internal medicine* (2014.0) **174** 1046. DOI: 10.1001/jamainternmed.2014.1994
22. Sugiyama T, Tsugawa Y, Tseng C-H, Kobayashi Y, Shapiro MF. **Different time trends of caloric and fat intake between statin users and nonusers among US adults: gluttony in the time of statins?**. *JAMA Internal Medicine* (2014.0) **174** 1038-45. DOI: 10.1001/jamainternmed.2014.1927
23. Lytsy P, Burell G, Westerling R. **Cardiovascular risk factor assessments and health behaviours in patients using statins compared to a non-treated population**. *International Journal of Behavioral Medicine* (2012.0) **19** 134-42. DOI: 10.1007/s12529-011-9157-6
24. Chiuve SE, McCullough ML, Sacks FM, Rimm EB. **Healthy lifestyle factors in the primary prevention of coronary heart disease among men: benefits among users and nonusers of lipid-lowering and antihypertensive medications**. *Circulation* (2006.0) **114** 160-7. DOI: 10.1161/CIRCULATIONAHA.106.621417
25. Cobb MM, Teitelbaum HS, Breslow JL. **Lovastatin efficacy in reducing low-density lipoprotein cholesterol levels on high-vs low-fat diets**. *JAMA* (1991.0) **265** 997-1001. PMID: 1992214
26. Hunninghake DB, Stein EA, Dujovne CA, Harris WS, Feldman EB, Miller VT. **The efficacy of intensive dietary therapy alone or combined with lovastatin in outpatients with hypercholesterolemia**. *New England Journal of Medicine* (1993.0) **328** 1213-9. DOI: 10.1056/NEJM199304293281701
27. Spring B, Moller AC, Coons MJ. **Multiple health behaviours: overview and implications**. *Journal of public health* (2012.0) **34** i3-i10. DOI: 10.1093/pubmed/fdr111
28. Pickett S, Allen W, Franklin M, Peters RM. **Illness beliefs in African Americans with hypertension**. *Western Journal of Nursing Research* (2014.0) **36** 152-70. DOI: 10.1177/0193945913491837
29. Coutu M-F, Dupuis G, D’Antono B, Rochon-Goyer L. **Illness representation and change in dietary habits in hypercholesterolemic patients**. *Journal of Behavioral Medicine* (2003.0) **26** 133-52. DOI: 10.1023/a:1023082622228
30. Diefenbach MA, Leventhal H. **The common-sense model of illness representation: Theoretical and practical considerations**. *Journal of Social Distress and the Homeless* (1996.0) **5** 11-38
31. Ross S, Walker A, MacLeod MJ. **Patient compliance in hypertension: role of illness perceptions and treatment beliefs**. *Journal of Human Hypertension* (2004.0) **18** 607-13. DOI: 10.1038/sj.jhh.1001721
32. Rosenstock IM. **Historical origins of the health belief model.**. *Health Education Monographs* (1974.0) **2** 328-35
33. Leventhal H, Brissette I, Leventhal EA, Cameron LD, Leventhal H. *The Self-Regulation of Health and Behaviour.* (2003.0) 42-65
34. Creswell JW, Creswell JD. *Research design: Qualitative, quantitative, and mixed methods approaches* (2017.0)
35. Anosike F, Mammah C, Ojo J, Ugbede N, Egwuda E, Azuwuike M. *The best of Nigeria. The Nigerian Country Book: New windows of opportunities* (2009.0)
36. Ansa V, Oyo-Ita A, Essien O. **Perception of ischaemic heart diease, knowledge of and attitude to reduction of its risk factors**. *East African Medical Journal* (2007.0) **84** 318-23. DOI: 10.4314/eamj.v84i7.9586
37. Moss-Morris R, Weinman J, Petrie K, Horne R, Cameron L, Buick D. **The revised illness perception questionnaire (IPQ-R).**. *Psychology & Health* (2002.0) **17** 1-16
38. Falola T. *Culture and customs of Nigeria* (2001.0)
39. Champion VL. **Instrument development for health belief model constructs**. *Advances in Nursing Science* (1984.0) **6** 73-85. DOI: 10.1097/00012272-198404000-00011
40. Abraham C, Sheeran P, CONNER M, NORMAN P. *Predicting Health Behaviour* (2005.0) 28-80
41. Curry SJ, Kristal AR, Bowen DJ. **An application of the stage model of behavior change to dietary fat reduction**. *Health Education Research* (1992.0) **7** 97-105. DOI: 10.1093/her/7.1.97
42. Bingham SA, Welch AA, McTaggart A, Mulligan AA, Runswick SA, Luben R. **Nutritional methods in the European prospective investigation of cancer in Norfolk**. *Public Health Nutrition* (2001.0) **4** 847-58. DOI: 10.1079/phn2000102
43. 43University of Cambridge, European Prospective Investigation of Cancer (EPIC). EPIC-Norfolk: nutritional methods- Food Frequency Questionnaire download. 2014 [cited 2019 16th May 2019]. Available from: http://www.srl.cam.ac.uk/epic/nutmethod/FFQii.shtml.
44. 44Department of Health and Social Care. General Practive Physical Activity Questionnaire 2013 [16th May 2019]. Available from: https://www.gov.uk/government/publications/general-practice-physical-activity-questionnaire-gppaq.
45. Hsiao C-Y, Chang C, Chen C-D. **An investigation on illness perception and adherence among hypertensive patients**. *The Kaohsiung Journal of Medical Sciences* (2012.0) **28** 442-7. DOI: 10.1016/j.kjms.2012.02.015
46. Field A.. *Discovering statistics using SPSS: and sex drugs and rock ’n’ roll.* (2009.0)
47. Pallant J.. *SPSS Survival Manual* (2010.0)
48. Braun V, Clarke V. **Using thematic analysis in psychology.**. *Qualitative Research in Psychology* (2006.0) **3** 77-101. DOI: 10.1191/1478088706qp063oa
49. Lofgren I, Greene G, Schembre S, Delmonico MJ, Riebe D, Clark P. **Comparison of diet quality, physical activity and biochemical values of older adults either reporting or not reporting use of lipid-lowering medication.**. *The Journal of Nutrition, Health & Aging* (2010.0) **14** 168-72. DOI: 10.1007/s12603-010-0030-0
50. Wilcox S, King AC, Castro C, Bortz W. **Do changes in physical activity lead to dietary changes in middle and old age?**. *American journal of preventive medicine* (2000.0) **18** 276-83. DOI: 10.1016/s0749-3797(00)00117-3
51. Malpass A, Andrews R, Turner KM. **Patients with type 2 diabetes experiences of making multiple lifestyle changes: a qualitative study**. *Patient education and counseling* (2009.0) **74** 258-63. DOI: 10.1016/j.pec.2008.08.018
52. Savoca MR, Miller CK, Quandt SA. **Profiles of people with type 2 diabetes mellitus: the extremes of glycemic control**. *Social science & medicine* (2004.0) **58** 2655-66. DOI: 10.1016/j.socscimed.2003.09.031
53. Nelson AJ, Puri R, Nissen SE. **Statins in a distorted mirror of media**. *Current atherosclerosis reports* (2020.0) **22** 1-8
54. Moon J, Cohen Sedgh R, Jackevicius C. **Examining the Nocebo Effect of Statins Through Statin Adverse Events Reported in the FDA Adverse Event Reporting System (FAERS).**. *Circulation: Cardiovascular Quality and Outcomes* (2020.0)
55. Chee YJ, Chan HHV, Tan NC. **Understanding patients’ perspective of statin therapy: can we design a better approach to the management of dyslipidaemia? A literature review**. *Singapore medical journal* (2014.0) **55** 416. DOI: 10.11622/smedj.2014099
56. Harrison TN, Derose SF, Cheetham TC, Chiu V, Vansomphone SS, Green K. **Primary nonadherence to statin therapy: patients’ perceptions.**. *The American journal of managed care* (2013.0) **19** e133-9. PMID: 23725451
57. Casula M, Tragni E, Catapano AL. **Adherence to lipid-lowering treatment: the patient perspective.**. *Patient preference and adherence.* (2012.0) **6** 805. DOI: 10.2147/PPA.S29092
58. Vicki F, Sinclair F, Wang H, Dailey D, Hsu J, Shaber R. **Patients’ perspectives on nonadherence to statin therapy: a focus-group study**. *The Permanente Journal* (2010.0) **14** 4. DOI: 10.7812/tpp/09-090
59. Fung V, Graetz I, Reed M, Jaffe MG. **Patient-reported adherence to statin therapy, barriers to adherence, and perceptions of cardiovascular risk.**. *PLoS One* (2018.0) **13** e0191817. DOI: 10.1371/journal.pone.0191817
60. Mann DM, Allegrante JP, Natarajan S, Halm EA, Charlson M. **Predictors of adherence to statins for primary prevention.**. *Cardiovascular Drugs and Therapy* (2007.0) **21** 311-6. DOI: 10.1007/s10557-007-6040-4
61. Tolmie EP, Lindsay GM, Kerr SM, Brown MR, Ford I, Gaw A. **Patients’ perspectives on statin therapy for treatment of hypercholesterolaemia: a qualitative study**. *European Journal of Cardiovascular Nursing* (2003.0) **2** 141-9. DOI: 10.1016/S1474-5151(03)00028-8
62. Athyros VG, Mikhailidis DP. **Therapy: Caloric and fat intake in statin users**. *Nature Reviews Endocrinology* (2014.0) **10** 450. DOI: 10.1038/nrendo.2014.82
63. Jemisenia JO, Ugochukwu SA, Adejoh A, Oniwon GO. **“The Way She Makes Me Feel”: Examining The Effects of Celebrity Instagram Images on Body Satisfaction and self-esteem in Young Nigerian Women.**. (2019.0)
64. Okoro E, Oyejola B, Etebu E, Sholagberu H, Kolo P, Chijioke A. **Body size preference among Yoruba in three Nigerian communities.**. *Eating and Weight Disorders-Studies on Anorexia, Bulimia and Obesity.* (2014.0) **19** 77-88. DOI: 10.1007/s40519-013-0060-9
65. Helman C. *Culture, health and illness* (2007.0)
66. Ofosu HB, LaFreniere KD, Senn CY. **Body image perception among women of African descent: A normative context?**. *Feminism & Psychology* (1998.0) **8** 303-23
67. Toriola AL, Dolan BM, Evans C, Adetimole O. **Weight Satisfaction of Nigerian Women in Nigeria and Britain: Inter‐Generational and Cross‐Cultural Influences.**. *European Eating Disorders Review: The Professional Journal of the Eating Disorders Association.* (1996.0) **4** 84-94
68. De Cock KM, Selik RM, Soro B, Gayle H, Colebunders RL. **For debate. AIDS surveillance in Africa: a reappraisal of case definitions.**. *BMJ: British Medical Journal* (1991.0) **303** 1185. DOI: 10.1136/bmj.303.6811.1185
69. Barry A-M, Yuill C. *Understanding the sociology of health: An introduction:* (2011.0)
70. Uchendu E. **Masculinity and Nigerian youths.**. *Nordic Journal of African Studies* (2007.0) **16**
71. Tripathee S, Sweeting H, Chambers S, Maclean A. **How men receive and utilise partner support when trying to change their diet and physical activity within a men’s weight management programme.**. *BMC public health* (2020.0) **20** 199. DOI: 10.1186/s12889-020-8213-z
72. Ragonese C, Barker G. **Understanding masculinities to improve men’s health**. *The Lancet* (2019.0) **394** 198-9. DOI: 10.1016/S0140-6736(19)31609-5
73. 73The Lancet Public Health. Time to tackle the physical activity gender gap. 2019.. *Time to tackle the physical activity gender gap* (2019.0)
74. de Groot R, van den Hurk K, Schoonmade LJ, de Kort WL, Brug J, Lakerveld J. **Urban-rural differences in the association between blood lipids and characteristics of the built environment: a systematic review and meta-analysis.**. *BMJ global health* (2019.0) **4**. DOI: 10.1136/bmjgh-2018-001017
75. Adegboye OR, Smith C, Anang D, Musa H. **Comparing and contrasting three cultural food customs from Nigeria and analyzing the nutrient content of diets from these cultures with the aim of proffering nutritional intervention**. *Critical reviews in food science and nutrition* (2016.0) **56** 2483-94. DOI: 10.1080/10408398.2013.862201
76. Ejim E, Okafor C, Emehel A, Mbah A, Onyia U, Egwuonwu T. **Prevalence of cardiovascular risk factors in the middle-aged and elderly population of a Nigerian rural community.**. *Journal of Tropical Medicine* (2011.0) 2011. DOI: 10.1155/2011/308687
77. Stack RJ, Bundy C, Elliott RA, New JP, Gibson JM, Noyce PR. **Patient perceptions of treatment and illness when prescribed multiple medicines for co-morbid type 2 diabetes.**. *Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy.* (2011.0) **4** 127-35. DOI: 10.2147/DMSO.S17444
78. Salovey P, Rothman AJ. *Social Psychology of Health: Key Readings* (2003.0)
79. Walter M.. *Social Research Methods.* (2012.0)
80. Hekler EB, Lambert J, Leventhal E, Leventhal H, Jahn E, Contrada RJ. **Commonsense illness beliefs, adherence behaviors, and hypertension control among African Americans**. *Journal of Behavioral Medicine* (2008.0) **31** 391-400. DOI: 10.1007/s10865-008-9165-4
81. Dunbar-Jacob J, Schlenk EA, Caruthers D, Christensen AJ, Antoni MH. *Chronic Physical Disorders: Behavioral Medicine’s Perspective* (2002.0) 69-82
|
---
title: 'Light at night and cause-specific mortality risk in Mainland China: a nationwide
observational study'
authors:
- Yao Lu
- Peng Yin
- Jie Wang
- Yiping Yang
- Fei Li
- Hong Yuan
- Shenxin Li
- Zheng Long
- Maigeng Zhou
journal: BMC Medicine
year: 2023
pmcid: PMC10022237
doi: 10.1186/s12916-023-02822-w
license: CC BY 4.0
---
# Light at night and cause-specific mortality risk in Mainland China: a nationwide observational study
## Abstract
### Background
While epidemiological studies have found correlations between light at night (LAN) and health effects, none has so far investigated the impacts of LAN on population mortality yet. We aimed to estimate the relative risk for mortality from exposure to LAN in Mainland China.
### Methods
This time-stratified case-crossover nationwide study used NPP-VIIRS to obtain daily LAN data of Mainland China between 2015 and 2019. The daily mortality data were obtained from the Disease Surveillance Point System in China. Conditional Poisson regression models were applied to examine the relative risk (RR) for mortality along daily LAN in each county, then meta-analysis was performed to combine the county-specific estimates at the national or regional level.
### Results
A total of 579 counties with an average daily LAN of 4.39 (range: 1.02–35.46) were included in the main analysis. The overall RRs per 100 nW/cm2/sr increases in daily LAN were 1.08 ($95\%$CI: 1.05–1.11) for all-cause mortality and 1.08 ($95\%$CI: 1.05–1.11) for natural-cause mortality. A positive association between LAN and all natural cause-specific mortality was observed, of which the strongest effect was observed on mortality caused by neuron system disease (RR = 1.32, $95\%$CI: 1.14–1.52). The results were robust in both younger and old, as well as in males and females. The more pronounced effect of LAN was observed in median LAN-level regions. Combined with an exposure–response curve, our study suggests a non-linear association between LAN and mortality in China.
### Conclusions
Our study shows LAN is associated with mortality in China, particularly for neuron system disease-related mortality. These findings have important implications for public health policy establishment to minimize the health consequences of light pollution.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12916-023-02822-w.
## Background
Light at night (LAN) pollution has been a long-established man-made disturbance and a prominent environmental issue, along with the context of urbanization and industrialization. It is estimated more than $80\%$ of the population in the world may currently live under light pollution (> 14 μcd/m2, $8\%$ above natural nighttime brightness), and nearly $90\%$ of the population is exposed to light pollution in China [1]. LAN is becoming a globally widespread environmental pollutant and continuing to expand both in spatial extent and intensity without intervention. Besides, the advent and widespread adoption of electric lighting over the past century has profoundly affected the behavior of many individuals; electric lighting in homes, work environments, and public areas has extended daytime activities into the evening, thus increasing the intensity and duration of night-time exposure to light. The number of studies documenting chronic exposure to LAN impacts on human health has grown dramatically in the last decade [2–7]. LAN may represent an emerging health risk factor, and investigation is urgent to clarify the associated health effects, particularly in highly polluted regions. This information is essential for planning suitable public health policies and for public education about the effects of night light pollution.
Exposure to LAN has been linked to a variety of health disorders in people through the circadian disruption mechanism as circadian rhythm is an important regulator of physiology and disease, coordinating the behavior and physiology of all organs for whole-body homeostasis [8, 9]. A circadian rhythm is endogenously generated and can be modulated by external cues. Circadian disruption due to LAN is common in modern life and contributes to a wide range of human diseases, including coronary heart disease [3], diabetes [4], obesity [5], and cancers [6, 7]. The evidence for the effect of LAN is generally restricted to the incidence risk of disease at the individual level, and no studies have evaluated the effects of LAN on mortality or broader categories of specific causes of mortality. Besides, the LAN data of the previous study was limited by the inferences that can be drawn from satellite images (Defence Meteorological Satellite Programme Operational Line Scan) with an insufficient spatial resolution (5 km) [10]. Therefore, an analysis based on a higher-resolution image is needed to explore the effects of LAN on mortality or disease burden.
Here, we used two established nationwide datasets including 579 main Chinese counties to perform a national assessment of the effects of LAN on mortality. We aimed to examine and compare the associations of LAN with daily all-cause and cause-specific mortality at the national and regional levels, as well as to explore the effects of LAN at different LAN intensity levels. Our study provides insight into light pollution that can improve the planning and implementation of public policy aimed at reducing the environmental and mortality burden.
## Data collection
This study was based on two national databases on LAN and cause-specific mortality in 579 counties of Mainland China from 2015 to 2019. These counties were selected according to the death registry of China’s Disease Surveillance. To ensure adequate representation at national levels, surveillance points were randomly selected by an iterative method involving multistage stratification that took into account the sociodemographic characteristics of the Chinese population [11]. In brief, the multistage stratification process involves the following steps: first, counties and districts in each province are divided into four strata based on median urbanization index (i.e., high or low urbanization) and median population size (i.e., high or low population size); then, counties and districts in these four strata are further subdivided into two strata based on the median total mortality rate in each of these four strata in each province. The detailed information of this death registry has been previously described [12]. In total, almost all cities at or above the prefecture level were included in the Disease Surveillance Points System, which included 605 districts and counties (equal to the number of administrative districts in China). Of these surveillance points, 579 counties were included because they had LAN information and more than 3 deaths per day. The daily mortality data from 2015 to 2019 were extracted from China’s Disease Surveillance Points system.
Causes of death were coded by the International Classification of Disease 10th (ICD-10): all causes (codes A00–Z99), accident (referred to as “external cause” in the present study; codes S00–Z99), self-harm (codes X60–X84), non-accident (referred to as “natural cause” in the present study; codes A00–R99), nervous system disease (codes G00–G99), digestive system disease (codes K00–K93), urinary system disease (codes N00–N39), renal failure (codes N17–N19), cardiovascular disease (CVD, codes I00–I99), heart failure (codes I50), stroke (codes I60–I69), coronary heart disease (CHD, codes I20–I25), myocardial infarction (MI, codes I21–I23), respiratory system diseases (codes J00–J99), asthma (codes J45–J46), chronic obstructive pulmonary disease (COPD, codes J41–J44), and cancer (codes C00–C97). “ Urinary system disease” included renal failure; “cardiovascular disease” included heart failure, stroke, CHD, and MI; and “respiratory disease” included COPD and asthma. Based on tumor location, we then divided “cancer” into subtypes. “ Cancers” included skin cancer, lip/oral/pharyngeal cancer, reproductive system cancer (male or female), hematologic cancer, lung cancer, gastrointestinal cancer, and breast cancer (female).
## Measurement of LAN and other environmental risks
Data on daily LAN for each county from China were derived from the Visible Infrared Imaging Radiometer Suite (VIIRS) carried by the Suomi National Polar-Orbiting Partnership (Suomi NPP) satellite (NPP-VIIRS nighttime light data, https://ngdc.noaa.gov) at a spatial resolution of 750 m × 750 m. The data were started in April 2012, with wavelengths covering from 500 to 900 nm. VIIRS observes the LAN of the earth every 24 h, and the transit time of the NPP satellite is 1:30 a.m. at local time. LAN has a specified dynamic range of approximately 7 orders of magnitude from 3 × 10−9 to 0.02 W/cm2/sr [13, 14]. Therefore, we estimated an objective measure of LAN in units of radiance (nW/cm2/sr). The daily composite NPP-VIIRS LAN data from 2015 to 2019 was obtained from the website. Then, we estimated the objective measure of daily LAN for each included county from 2015 to 2019 based on its geocoded address from the NPP-VIIRS. The daily LAN at the county scale uses the tool “Raster Calculator” in ArcGIS 10.2. The collected LAN records are mainly outdoor artificial LAN. Due to inclement weather conditions, LAN data cannot be collected by artificial satellites on some days (Additional file 1: Fig. S1). Our analyses are based on available data.
We also derived daily particular matter 2.5 (PM2.5) concentrations from 2015 to 2019 were derived from the China High Air Pollutants (CHAP) dataset (https://weijing-rs.github.io/product.html), with a spatial resolution of 1 × 1 km2. The CHAP dataset has been generated from MODIS/Terra + Aqua MAIAC AOD products together with other auxiliary data (e.g., ground-based measurements, satellite remote sensing products, atmospheric reanalysis, and model simulations) using artificial intelligence by considering the spatiotemporal heterogeneity of air pollution [15, 16]. The estimated annual PM2.5 concentrations from the dataset are highly correlated with ground-based measurements (R2 = 0.94) [16]. These daily PM2.5 concentrations were assigned to each county based on the longitude and latitude of the address, through the Baidu Web service API of “Geocoder” with Python version 3.8.3 [17]. The daily temperature, dewpoint temperature, and pressure were calculated using Python based on hourly data, which were collected from the ERA5-Land dataset (https://cds.climate.copernicus.eu/) with a spatial resolution of 10 × 10 km. ERA5-Land has been produced by replaying the land component of the ECMWF ERA5 climate reanalysis. The daily humidity was estimated by combing temperature, dewpoint temperature, and pressure. The unit of daily PM2.5, daily temperature, and daily humidity were µg/m3, °C, and %, respectively.
## Statistical analysis
In this nationwide observational study, we performed a two-stage analysis to calculate the association between daily LAN and daily mortality in China. In stage 1, we first examined the association between daily LAN and daily cause-specific mortality in each county by using a time-stratified case-crossover design, which has been widely used to assess the effects of environmental risk on adverse health events. In this design, all days during which mortality was reported were selected as the case days, while the control days were selected from the same month and year and matched by day of the week to the case days. To account for the relative risk (RR) of cause-specific mortality associated with an increase in daily LAN, this stage analysis applied a conditional Poisson regression model. To control for confounding effects of weather conditions, we also adjusted for daily temperature, daily humidity, and daily PM2.5. In stage 2, we used random-effects models to pool the county-specific estimates for all counties or counties in different groups.
To test the reliability of the main results, we did a series of sensitivity analyses. We modified the modeling choices in stage 1: model 1, adjusted by PM2.5 and temperature; model 2, adjusted by PM2.5 and humidity; and model 3, adjusted by temperature and humidity. To investigate the different lag structures in the model, we performed conditional quasi-Poisson regression to capture the delayed effects of LAN and PM2.5 for the lag 0, lag 1, lag 2, lag 3, and lag 0–1 (average of the present and previous day), respectively. We explored the potential overall lag effects of daily LAN and non-linear association with daily mortality using conditional quasi-Poisson regression combined with distributed lag non-linear model. In this model, a natural cubic smooth function with 7 degrees freedom (df) per year was introduced to account for trends in mortality and seasonality. To control potentially non-linear effects of weather conditions, natural spline functions with 6 df for temperature and 3 df for humidity were also introduced into the model.
To further identify the potential effect modifiers, we also conducted a series of stratified analyses. Firstly, to assess whether the association between LAN and mortality differed across subpopulations, we conducted analyses stratified by age (younger means < 65; old means ≥ 65 years) and sex to confirm the robustness of our results. Secondly, all counties in Mainland China were categorized into 3 levels based on a 5-year average per capita gross domestic product (GDP) from 2015 to 2019, which were derived from the yearbook of statistics in China at provincial levels. Above the national 5-year average per GDP (unit of renminbi (RMB): 59,781.8) was regarded as a high economic level and less than 45,000 RMB as a low economic level. Low, median, and high economic level groups include 172, 198, and 209 counties, respectively. We pooled the county-specific estimates for counties in different economic groups. Thirdly, according to tertiles of the 5-year average of daily LAN estimates from 2015 to 2019, we divide all counties into three levels which correspond to low, medium, and high LAN levels, respectively (first tertile means low LAN level: < 2.67 nW/cm2/sr; the second tertile means median LAN level 2.67–3.63 nW/cm2/sr; the third tertile means high LAN level: > 3.63 nW/cm2/sr). Low, median, and high LAN groups include 193 counties, respectively. We also carried out the same analysis for different LAN levels to further identify the effect of LAN. The post hoc comparisons with the Kruskal–Wallis test were performed to make sure whether the effects of LAN were similar in different subgroups.
The analyses were performed with the R software (version 3.4.2, R Foundation for Statistical Computing). Two-tailed P values less than 0.05 were considered statistically significant for all statistical tests.
## Descriptive statistics
A total of 579 counties in Mainland China with available data on LAN and mortality were included in the present study during 2015–2019. The spatial patterns of average daily LAN estimations during the study period for each included county in Mainland China were presented in Fig. 1. Compared with most western counties in China, LAN levels were much higher in most eastern counties in China, especially in Shanghai and Beijing. Table 1 provides descriptive statistics on the average number of daily deaths during the study period and the averages of daily LAN and other environmental risks included in the analysis from 2015 to 2019. Large variations in the average LAN intensity of Mainland China were observed with a daily mean LAN of 4.39 nW/cm2/sr. Daily average LAN (blue line) and daily number of death (red line) in Mainland China are described in Additional file 1: Fig. S1. The number of all-cause deaths increased along with the daily average LAN during the study period, and similar temporal change trends of two variables were observed. During the study period, we recorded a daily average of 4015 total deaths in 579 counties in Mainland China. There were 291 average daily deaths from external causes and 3724 from natural causes. The number of deaths for different age and sex populations is shown in Additional file 1: Table S1.Fig. 1Distribution of the counties with a 5-year average of daily LAN level in Mainland China during the study period (2015–2019). The dot indicates the location of the included county, while the color of the dot indicates the level of LAN, with red denoting a higher LAN levelTable 1Summary descriptive statistics on the average number of daily deaths and light at night levels in 579 Chinese counties, 2015–2019Average daily deathMean (SD)RangeMedian (IQR)All-cause death4015 [719]22–71453848 [788]External291 [39]2–456287 [42]Natural3724 [695]20–66973558 [758]Neuron system51 [13]2–9849 [16]Digestive system92 [17]1–16290 [20]Cancer971 [112]3–1450962 [121]Urinary system45 [11]0–9044 [13]Cardiovascular1819 [398]5–34691725 [448]Respiratory system466 [137]5–974422 [157]Average daily LAN (nW/cm2/sr) National4.39 (3.63)1.02–35.462.92 (2.88) Low-economic level region3.49 (3.42)0.71–31.591.85 (2.91) Median-economic level region3.67 (3.56)0.70–30.202.15 (2.61) High-economic level region6.38 (4.04)1.34–49.905.20 (3.33) Low LAN level2.23 (2.98)0.30–30.300.75 (2.29) Median LAN level3.15 (4.22)0.42–36.161.10 (3.01) High LAN level8.15 (3.84)2.24–44.067.17 (3.58)Average daily PM2.5(µg/m3)35.18 (13.97)15.13–104.7431.21 (16.64)Average daily temperature (°C)16 [9] − 8–2918 [17]Average daily humidity (%)61.20 (7.73)34.05–79.2662.12 (11.72)LAN light at night, PM2.5 particular matter 2.5, SD standard deviation, IQR interquartile range
## Association between LAN and mortality
At the national level, we estimated that every 100 nW/cm2/sr increase in daily LAN was associated with all-cause mortality (RR = 1.08, $95\%$ confidence interval [CI], 1.05–1.11) and natural-cause morality (RR = 1.08, $95\%$CI = 1.05–1.11), while mortality from external causes was not related to LAN (RR = 1.04, $95\%$CI = 0.97–1.13) (Fig. 2). In the natural cause-specific mortality analysis, the highest effects of LAN were observed on mortality caused by neuron system disease (RR = 1.32, $95\%$CI = 1.14–1.52), followed by digestive system disease, urinary system disease, cancer, cardiovascular disease, and respiratory system disease. Fig. 2Pooled relative risks of daily specific mortality associated with an increase of 100 nW/cm2/sr of daily LAN in 579 Chinese counties, 2015–2019. The results were adjusted by temperature, humidity, and PM2.5 In the analysis for specific disease-related mortality, the consistent effects of LAN were also observed in coronary heart disease, stroke, heart failure, asthma, and renal failure. The effects of daily LAN were higher on heart failure, asthma, and renal failure-related death. However, no significant daily LAN-mortality associations were found for mortality caused by MI, COPD, and self-harm (Fig. 2).
In our sensitivity analyses, different adjustments for other environmental risks did not change the results, indicating that our main findings are solid under a series of parameter changes during modeling (Additional file 1: Table S2). Using lag 0, lag 1, lag 2, and lag 3 for LAN and PM2.5 did not substantially change the effect of the association between LAN and all-cause and natural cause mortality (Additional file 1: Table S3). In addition, the overall lag effects of LAN, as well as the dose–response relationship between LAN and morality, are shown in Additional file 1: Fig. S2. The effects of LAN on mortality decline as the lag time increases. In Additional file 1: Fig. S2, we have further observed the non-linear association between LAN and morality. The exposure–response curve for LAN did not increase monotonically, with a slight decrease at the high level of LAN exposure, but an upward-sloping at the highest level (Additional file 1: Fig. S2).
## Stratified analysis according to age, sex, and region
When we repeated our analysis within different subgroup populations, our results did not change after stratified by age and sex. Similar to the nationwide estimates, the all-cause and natural cause mortality burden caused by LAN were still robust in various subgroups, while external cause mortality was still statistically insignificant (Table 2). Estimations of respiratory system disease and urinary system disease mortality relative risks were statistically insignificant in younger, and respiratory system diseases were also statistically insignificant in males. Through post hoc comparison, we found the effects of LAN did not differ between the younger group and the old group ($$P \leq 0.51$$), nor between males and females ($$P \leq 0.89$$).Table 2Relative risk of daily mortality associated with an increase of 100 nW/cm2/sr of daily LAN in 579 Chinese counties, 2015–2019, by subgroupDeath causeRR ($95\%$CI)YoungerOldFemaleMaleAll-cause1.05 (1.01, 1.10)1.09 (1.06, 1.13)1.07 (1.03, 1.11)1.09 (1.06, 1.12)External1.07 (0.97, 1.18)1.05 (0.95, 1.16)1.10 (0.96, 1.25)1.07 (0.98, 1.16)Natural1.06 (1.02, 1.10)1.09 (1.06, 1.13)1.07 (1.03, 1.11)1.09 (1.06, 1.13)Neuron system disease1.82 (1.36, 2.45)1.32 (1.12, 1.56)1.50 (1.22, 1.85)1.41 (1.15, 1.73)Digestive system disease1.39 (1.16, 1.66)1.24 (1.08, 1.42)1.21 (1.02, 1.44)1.33 (1.15, 1.55)Urinary system disease1.26 (0.95, 1.66)1.44 (1.19, 1.74)1.35 (1.06, 1.72)1.37 (1.12, 1.69)Cancer1.13 (1.07, 1.18)1.10 (1.04, 1.17)1.14 (1.07, 1.21)1.11 (1.05, 1.17)Cardiovascular disease1.08 (1.01, 1.15)1.08 (1.04, 1.12)1.06 (1.01, 1.11)1.10 (1.05, 1.15)Respiratory system disease1.01 (0.86, 1.20)1.09 (1.03, 1.15)1.12 (1.02, 1.22)1.06 (0.98, 1.13)P for post hoc test0.510.89The results adjusted by temperature, humidity, and PM2.5RR relative risk, CI incidence intervals When dividing 579 counties into 3 regions by different economical levels, different effects of daily LAN on mortality were not observed according to the results of post hoc analysis ($$P \leq 0.42$$). The relationship between daily LAN and mortality was consistent across different levels of economic development (Table 3). These results reflect that the economic level of the regions did not affect the association between daily LAN and mortality. However, when dividing all 579 counties into 3 regions by 5-year average daily LAN level, different effects of daily LAN on mortality were found according to the results of post hoc analysis ($$P \leq 0.02$$). Within the median LAN level regions, there is evidence for more significant associations between daily LAN and mortality (Table 3).Table 3Relative risk of daily mortality associated with an increase of 100 nW/cm2/sr of daily LAN in 579 Chinese counties, 2015–2019, by economic development levels and LAN levelsDeath causeRR ($95\%$CI)High development levelMedian development levelLow development levelAll-cause1.06 (1.03, 1.10)1.10 (1.04, 1.17)1.07 (1.02, 1.12)External1.03 (0.93, 1.13)1.04 (0.88, 1.23)1.06 (0.93, 1.22)Natural1.07 (1.03, 1.10)1.11 (1.04, 1.18)1.07 (1.02, 1.12)Neuron system disease1.27 (1.05, 1.54)1.61 (1.20, 2.15)1.61 (1.20, 2.15)Digestive system disease1.36 (1.12, 1.65)1.26 (0.99, 1.60)1.10 (0.91, 1.33)Urinary system disease1.19 (0.93, 1.52)1.20 (0.90, 1.61)1.26 (0.95, 1.67)Cancer1.07 (1.01, 1.13)1.19 (1.09, 1.30)1.13 (0.81, 1.58)Cardiovascular disease1.07 (1.02, 1.12)1.09 (1.004, 1.17)1.05 (0.99, 1.12)Respiratory system disease1.02 (0.95, 1.09)1.09 (0.98, 1.21)1.13 (1.01, 1.27)P for post hoc test0.42High LAN levelMedian LAN levelLow LAN levelAll-cause1.05 (1.01, 1.09)1.10 (1.05, 1.15)1.10 (1.02, 1.18)External0.97 (0.88, 1.07)1.09 (0.97, 1.23)1.09 (0.88, 1.35)Natural1.05 (1.01, 1.09)1.10 (1.06, 1.15)1.10 (1.02, 1.19)Neuron system disease1.22 (0.99, 1.49)1.37 (1.08, 1.74)1.68 (1.08, 2.62)Digestive system disease1.12 (0.93, 1.34)1.38 (1.16, 1.64)1.28 (0.90, 1.81)Urinary system disease1.08 (0.86, 1.37)1.34 (1.06, 1.69)1.30 (0.82, 2.07)Cancer1.07 (1.01, 1.13)1.11 (1.03, 1.19)1.27 (1.13, 1.42)Cardiovascular disease1.04 (0.99, 1.10)1.11 (1.05, 1.18)1.06 (0.96, 1.16)Respiratory system disease1.03 (0.96, 1.11)1.06 (0.98, 1.16)1.19 (0.98, 1.43)P for post hoc test0.02The results adjusted by temperature, humidity, and PM2.5RR relative risk, CI incidence intervals, LAN light at night
## Discussion
To our knowledge, this is the first study to examine the association between daily LAN and the risk of natural-cause mortality as well as external-cause mortality. In this study, exposure to excessive LAN was associated with an increased risk of all-cause mortality. We used nationwide survey data and a dataset of LAN levels in Mainland China, which allowed for a more convincing analysis of the relationship between LAN and mortality. During the last decades, both outdoor LAN and indoor LAN have increased, mainly because of urbanization, industrialization, lights at home turned on during the night, and new sources of exposure such as smartphones. Collectively, our results suggest that exposure to LAN is a novel and important environmental risk factor that cannot be ignored.
LAN exerts profound effects on disease and mortality by directly and indirectly entraining circadian rhythms, melatonin secretion, and sleep deprivation, all of which are crucial for the fitness and survival of species [18]. Given the primordial dominance of the circadian rhythm over physiological processes, it is apparent that the circadian system is crucial for maintaining synchrony between internal physiology, behavior, and the cues deriving from the external environment [18–20]. When the intrinsic circadian clock is disrupted, substantial serious biological disorders will emerge to affect several systems. Habitually exposure to LAN could not only suppress the secretion of melatonin, a hormone released by the pineal gland that regulates sleep–wake cycles, but also causes inflammatory responses and detrimentally affects the immune system [21]. Thus, LAN could lead to a higher risk of mortality by disrupting our intrinsic circadian rhythms and increasing sympathetic tone.
In the cause-specific mortality analysis, the strongest effect of LAN was found in neuron system disease-related mortality. Disruption of the endogenous local circadian rhythm caused by LAN is one of the biologically plausible mechanisms that potentially explain the strongest association between LAN and mortality due to neuron system disease. The central circadian rhythm generator is located in the hypothalamus suprachiasmatic nucleus (SCN), which contains pacemaker neurons that drive rhythm [22]. Indeed, preclinical and clinical studies have already correlated circadian disruption with the accumulation of neurotoxic proteins and neurodegeneration disease [23–25], including Huntington’s disease, Parkinson’s disease, and Alzheimer’s disease. The mechanisms by which LAN exposure modulates neuron function are multifactorial. LAN-induced neuron dysfunction may also be mediated by direct synaptic inputs from the circadian center in the brain without disturbing the circadian rhythm [26]. Evidence from animal studies shows acute LAN displayed increased proinflammation cytokine expression within the hippocampus [27]. Chronic LAN has also been shown to exacerbate LPS-induced proinflammation cytokine expression within the hippocampus [28]. Imaging studies have also shown that light exposure can influence cortical and subcortical networks involved in cognitive processes, both directly and indirectly [29–31]. The latest published epidemiological study provides direct evidence that LAN exposure is associated with a higher risk of mild cognitive impairment in Chinese [32]. Our findings (RR = 1.32, $95\%$CI = 1.14–1.52, for neuron system disease-related mortality) also further suggest that excessive LAN exposure in *China is* associated with a significant neuron system disease burden. The impact of chronic LAN exposure on the neuron system needs further investigation to clarify why LAN has the strongest impact on death from neuron system disease.
In 2007, a statement from The World Health Organization classified shift work that disrupted human circadian rhythms as a probable human carcinogen [33]. Recently, epidemiologic and experimental studies also indicated that cancer is associated with circadian rhythms caused by LAN. Clinical studies have demonstrated a potential association between LAN and the incidence risk of thyroid cancer [34], pancreatic cancer [7], and prostate cancer [35], especially breast cancer [6, 35–37]. The latest meta-analysis [37] showed a positive association between exposure to LAN and the incidence risk of breast cancer, particularly in premenopausal women. Our study further proved a close relationship between LAN and cancer mortality, especially skin cancer. *In* general, there is a bidirectional relationship between the circadian clock and the cell cycle, and the dysregulation of the shared regulatory and coupling connections between the two pathways can be both necessary and sufficient for tumorigenesis [38].
Our study found that LAN exposure also affects cardiovascular mortality, especially heart failure. This result supports previous findings showing that LAN was linked to a higher risk of cardiovascular disease [3, 39]. A previous cohort study based on 58,692 participants aged 65 years and older followed for a median of 11 years in Hong Kong also found outdoor LAN at the residential address was associated with a higher risk of CHD incidence and mortality [3]. Besides, studies of night shift workers, who tend to be exposed to higher levels of LAN, provide similar insights [40, 41]. A meta-analysis [42] of 34 studies reported that night shift work was associated with vascular events including myocardial infarction, stroke, and coronary events. Another latest meta-analysis [43] also discovered the important role of shift work in cardiovascular disease. The effects of LAN and shift work depend on circadian rhythms, which play a key role in the etiology of CVD [44]. In addition to affecting vascular function, disruption of circadian rhythms may secondarily contribute to cardiometabolic disorders through the deregulation of clock function within adipose tissue, liver, and muscle [45].
In the sub-analysis of specific diseases, the results also showed a stronger association between LAN and renal failure. Consistent with our results, a Chinese cross-sectional study reported that night light index (NLI) at a 5-year moving average was significantly associated with an increased risk of chronic kidney disease prevalence [46]. Okuliarova et al. [ 47] also revealed that chronic exposure to dim LAN disturbed renal immune and redox homeostasis in animal models. In addition to renal failure, LAN also showed a closer association with asthma. A recent study of Chinese college students similarly showed the strongest effect of LAN on asthma [48]. Further studies are needed to evaluate the corresponding renal failure and asthma disease burden attributable to LAN.
This study had several major strengths. First, to our knowledge, it is the first nationwide investigation of the adverse impact on the mortality of LAN. Compared with previous studies confined to individual cities, our study included 579 counties in China, which provided robust evidence with reliable representatives for our findings. Our findings provide an ecological association between LAN and mortality based on data at the county level, which will be quite important for subsequent policy at the national level, to deal with the abuse of LAN. Secondly, compared with the traditional Defense Meteorological Satellite Program’s Operational Linescan System (DMSP-OLS) nighttime light data, the NPP-VIIRS light data at night have a higher spatial resolution, higher data quality, and no over-saturation problem [49]. Higher spatial resolution (0.75 km) is able to differentiate exposure to LAN from other factors that covary across city districts at fine spatial scales. Thirdly, we evaluated the association between LAN and mortality with analyses of a range of detailed causes of death. Our study provided robust evidence of the mortality risk from both natural and external causes. Finally, our analysis accounted for several potential covarying explanatory factors, including atmospheric pollution, temperature, and humidity explicitly.
There were several limitations to consider. First, we used the nighttime light imagery to evaluate LAN at the county level as surrogate data of individuals’ LAN exposure, which may lead to misclassification of LAN exposure. There are possible variations within the same county based on subject residences. However, previous epidemiological studies regarding other environmental risk factors have used the same approaches to estimate exposure levels. Secondly, we cannot adjust for some potential residual confounding effects at individual levels, including indoor light intensity, wearing an eyepatch, and sleep patterns, all of which may interact with LAN. Thirdly, although we have adjusted for atmospheric pollution and climate, we cannot entirely rule out that LAN exposure may be confounded by noise or other unrecognized environmental factors, which may interact with LAN to influence mortality.
## Conclusions
In conclusion, our study provides comprehensive evidence of positive associations between daily LAN and daily mortality caused by all natural causes, especially for neuron system disease-related mortality. By capturing the population exposed to a wide range of LAN levels in China and including objective measures of other environmental risk factors, our study adds to evidence that LAN is having profound impacts on public health, which substantially extent the existing knowledge. The findings on a national level can help improve public health practices to reduce the disease burden associated with LAN and call for policies on strategies for abating them.
## Supplementary Information
Additional file 1: Table S1. Summary descriptive statistics on average number of daily deaths in 579 Chinese counties stratified by sex and age, 2015-19. Table S2. RR and $95\%$CI of daily mortality associated with a 100 nanoWatts/cm2/sr increase of daily light at night levels for three models. Table S3. RR and $95\%$ CI for association between daily mortality and a 100 nanoWatts/cm2/sr increase of daily light at night levels for different lag structure. Fig. S1. Basic information about average daily light at night (nanoWatts/cm2/sr) and daily number of deaths. Fig. S2. RRs and $95\%$ CIs of mortality associated with daily light at night level on different lag days during 2015-2019.
## References
1. Falchi F, Cinzano P, Duriscoe D, Kyba CC, Elvidge CD, Baugh K, Portnov BA, Rybnikova NA, Furgoni R. **The new world atlas of artificial night sky brightness**. *Sci Adv* (2016) **2** e1600377. DOI: 10.1126/sciadv.1600377
2. Davies TW, Smyth T. **Why artificial light at night should be a focus for global change research in the 21st century**. *Glob Chang Biol* (2018) **24** 872-882. DOI: 10.1111/gcb.13927
3. Sun S, Cao W, Ge Y, Ran J, Sun F, Zeng Q, Guo M, Huang J, Lee RS, Tian L. **Outdoor light at night and risk of coronary heart disease among older adults: a prospective cohort study**. *Eur Heart J* (2021) **42** 822-830. DOI: 10.1093/eurheartj/ehaa846
4. Mason IC, Grimaldi D, Reid KJ, Warlick CD, Malkani RG, Abbott SM, Zee PC. **Light exposure during sleep impairs cardiometabolic function**. *Proc Natl Acad Sci U S A* (2022) **119** e2113290119. DOI: 10.1073/pnas.2113290119
5. Lai KY, Sarkar C, Ni MY, Gallacher J, Webster C. **Exposure to light at night (LAN) and risk of obesity: a systematic review and meta-analysis of observational studies**. *Environ Res* (2020) **187** 109637. DOI: 10.1016/j.envres.2020.109637
6. Lai KY, Sarkar C, Ni MY, Cheung LWT, Gallacher J, Webster C. **Exposure to light at night (LAN) and risk of breast cancer: a systematic review and meta-analysis**. *Sci Total Environ* (2021) **762** 143159. DOI: 10.1016/j.scitotenv.2020.143159
7. Xiao Q, Jones RR, James P, Stolzenberg-Solomon RZ. **Light at night and risk of pancreatic cancer in the NIH-AARP diet and health study**. *Cancer Res* (2021) **81** 1616-1622. DOI: 10.1158/0008-5472.CAN-20-2256
8. Crnko S, Du Pré BC, Sluijter JPG, Van Laake LW. **Circadian rhythms and the molecular clock in cardiovascular biology and disease**. *Nat Rev Cardiol* (2019) **16** 437-447. DOI: 10.1038/s41569-019-0167-4
9. Zhou L, Zhang Z, Nice E, Huang C, Zhang W, Tang Y. **Circadian rhythms and cancers: the intrinsic links and therapeutic potentials**. *J Hematol Oncol* (2022) **15** 21. DOI: 10.1186/s13045-022-01238-y
10. Kyba CC. **Defense Meteorological Satellite Program data should no longer be used for epidemiological studies**. *Chronobiol Int* (2016) **33** 943-945. DOI: 10.1080/07420528.2016.1189432
11. Liu S, Wu X, Lopez AD, Wang L, Cai Y, Page A, Yin P, Liu Y, Li Y, Liu J. **An integrated national mortality surveillance system for death registration and mortality surveillance, China**. *Bull World Health Organ* (2016) **94** 46-57. DOI: 10.2471/BLT.15.153148
12. Chen R, Yin P, Wang L, Liu C, Niu Y, Wang W, Jiang Y, Liu Y, Liu J, Qi J. **Association between ambient temperature and mortality risk and burden: time series study in 272 main Chinese cities**. *BMJ* (2018) **363** k4306. DOI: 10.1136/bmj.k4306
13. Zhao M, Zhou YY, Li XC, Cao WT, He CY, Yu BL, Li X, Elvidge CD, Cheng WM, Zhou CH. **Applications of satellite remote sensing of nighttime light observations: advances, challenges, and perspectives**. *Remote Sensing* (2019) **11** 35. DOI: 10.3390/rs11171971
14. Liao LB, Weiss S, Mills S, Hauss B. **Suomi NPP VIIRS day-night band on-orbit performance**. *J Geophys Res-Atmos* (2013) **118** 12705-12718. DOI: 10.1002/2013JD020475
15. Wei J, Li Z, Cribb M, Huang W, Xue W, Sun L, Guo J, Peng Y, Li J, Lyapustin A. **Improved 1-km-resolution PM2.5 estimates across China using enhanced space–time extremely randomized trees**. *Atmos Chem Phys* (2020) **20** 3273-3289. DOI: 10.5194/acp-20-3273-2020
16. Wei J, Li Z, Lyapustin A, Sun L, Peng Y, Xue W, Su T, Cribb M. **Reconstructing 1-km-resolution high-quality PM2.5 data records from 2000 to 2018 in China: spatiotemporal variations and policy implications**. *Remote Sensing Environ* (2021) **252** 112136. DOI: 10.1016/j.rse.2020.112136
17. Wang L, Chen G, Pan Y, Xia J, Chen L, Zhang X, Silang Y, Chen J, Xu H, Zeng C. **Association of long-term exposure to ambient air pollutants with blood lipids in Chinese adults: the China Multi-Ethnic Cohort study**. *Environ Res* (2021) **197** 111174. DOI: 10.1016/j.envres.2021.111174
18. Vitaterna MH, Takahashi JS, Turek FW. **Overview of circadian rhythms**. *Alcohol Res Health* (2001) **25** 85-93. PMID: 11584554
19. Peek CB, Levine DC, Cedernaes J, Taguchi A, Kobayashi Y, Tsai SJ, Bonar NA, McNulty MR, Ramsey KM, Bass J. **Circadian clock interaction with HIF1α mediates oxygenic metabolism and anaerobic glycolysis in skeletal muscle**. *Cell Metab* (2017) **25** 86-92. DOI: 10.1016/j.cmet.2016.09.010
20. Lu C, Thompson CB. **Metabolic regulation of epigenetics**. *Cell Metab* (2012) **16** 9-17. DOI: 10.1016/j.cmet.2012.06.001
21. Bedrosian TA, Fonken LK, Walton JC, Nelson RJ. **Chronic exposure to dim light at night suppresses immune responses in Siberian hamsters**. *Biol Lett* (2011) **7** 468-471. DOI: 10.1098/rsbl.2010.1108
22. Ralph MR, Foster RG, Davis FC, Menaker M. **Transplanted suprachiasmatic nucleus determines circadian period**. *Science* (1990) **247** 975-978. DOI: 10.1126/science.2305266
23. Leng Y, Musiek ES, Hu K, Cappuccio FP, Yaffe K. **Association between circadian rhythms and neurodegenerative diseases**. *Lancet Neurol* (2019) **18** 307-318. DOI: 10.1016/S1474-4422(18)30461-7
24. Xu F, Kula-Eversole E, Iwanaszko M, Hutchison AL, Dinner A, Allada R. **Circadian clocks function in concert with heat shock organizing protein to modulate mutant huntingtin aggregation and toxicity**. *Cell Rep* (2019) **27** 59-70.e54. DOI: 10.1016/j.celrep.2019.03.015
25. Kim M, Subramanian M, Cho YH, Kim GH, Lee E, Park JJ. **Short-term exposure to dim light at night disrupts rhythmic behaviors and causes neurodegeneration in fly models of tauopathy and Alzheimer’s disease**. *Biochem Biophys Res Commun* (2018) **495** 1722-1729. DOI: 10.1016/j.bbrc.2017.12.021
26. An K, Zhao H, Miao Y, Xu Q, Li YF, Ma YQ, Shi YM, Shen JW, Meng JJ, Yao YG. **A circadian rhythm-gated subcortical pathway for nighttime-light-induced depressive-like behaviors in mice**. *Nat Neurosci* (2020) **23** 869-880. DOI: 10.1038/s41593-020-0640-8
27. Walker WH, Borniger JC, Gaudier-Diaz MM, HecmarieMeléndez-Fernández O, Pascoe JL, Courtney DeVries A, Nelson RJ. **Acute exposure to low-level light at night is sufficient to induce neurological changes and depressive-like behavior**. *Mol Psychiatry* (2020) **25** 1080-1093. DOI: 10.1038/s41380-019-0430-4
28. Chen R, Weitzner AS, McKennon LA, Fonken LK. **Light at night during development in mice has modest effects on adulthood behavior and neuroimmune activation**. *Behav Brain Res* (2021) **405** 113171. DOI: 10.1016/j.bbr.2021.113171
29. Vandewalle G, Maquet P, Dijk DJ. **Light as a modulator of cognitive brain function**. *Trends Cogn Sci* (2009) **13** 429-438. DOI: 10.1016/j.tics.2009.07.004
30. Vandewalle G, Archer SN, Wuillaume C, Balteau E, Degueldre C, Luxen A, Dijk DJ, Maquet P. **Effects of light on cognitive brain responses depend on circadian phase and sleep homeostasis**. *J Biol Rhythms* (2011) **26** 249-259. DOI: 10.1177/0748730411401736
31. Chellappa SL, Ly JQ, Meyer C, Balteau E, Degueldre C, Luxen A, Phillips C, Cooper HM, Vandewalle G. **Photic memory for executive brain responses**. *Proc Natl Acad Sci U S A* (2014) **111** 6087-6091. DOI: 10.1073/pnas.1320005111
32. Chen Y, Tan J, Liu Y, Dong GH, Yang BY, Li N, Wang L, Chen G, Li S, Guo Y. **Long-term exposure to outdoor light at night and mild cognitive impairment: a nationwide study in Chinese veterans**. *Sci Total Environ* (2022) **847** 157441. DOI: 10.1016/j.scitotenv.2022.157441
33. **Painting, firefighting, and shiftwork**. *IARC Monogr Eval Carcinog Risks Hum* (2010) **98** 9-764. PMID: 21381544
34. Zhang D, Jones RR, James P, Kitahara CM, Xiao Q. **Associations between artificial light at night and risk for thyroid cancer: a large US cohort study**. *Cancer* (2021) **127** 1448-1458. DOI: 10.1002/cncr.33392
35. Garcia-Saenz A, Sánchez de Miguel A, Espinosa A, Valentin A, Aragonés N, Llorca J, Amiano P, Martín Sánchez V, Guevara M, Capelo R. **Evaluating the association between artificial light-at-night exposure and breast and prostate cancer risk in Spain (MCC-Spain Study)**. *Environ Health Perspect* (2018) **126** 047011. DOI: 10.1289/EHP1837
36. Xiao Q, James P, Breheny P, Jia P, Park Y, Zhang D, Fisher JA, Ward MH, Jones RR. **Outdoor light at night and postmenopausal breast cancer risk in the NIH-AARP diet and health study**. *Int J Cancer* (2020) **147** 2363-2372. DOI: 10.1002/ijc.33016
37. Urbano T, Vinceti M, Wise LA, Filippini T. **Light at night and risk of breast cancer: a systematic review and dose-response meta-analysis**. *Int J Health Geogr* (2021) **20** 44. DOI: 10.1186/s12942-021-00297-7
38. Walker WH, Bumgarner JR, Walton JC, Liu JA, Meléndez-Fernández OH, Nelson RJ, DeVries AC. **Light pollution and cancer**. *Int J Mol Sci* (2020) **21** 9360. DOI: 10.3390/ijms21249360
39. Obayashi K, Saeki K, Kurumatani N. **Light exposure at night is associated with subclinical carotid atherosclerosis in the general elderly population: the HEIJO-KYO cohort**. *Chronobiol Int* (2015) **32** 310-317. DOI: 10.3109/07420528.2014.974809
40. Vetter C, Devore EE, Wegrzyn LR, Massa J, Speizer FE, Kawachi I, Rosner B, Stampfer MJ, Schernhammer ES. **Association between rotating night shift work and risk of coronary heart disease among women**. *JAMA* (2016) **315** 1726-1734. DOI: 10.1001/jama.2016.4454
41. Wang N, Sun Y, Zhang H, Wang B, Chen C, Wang Y, Chen J, Tan X, Zhang J, Xia F. **Long-term night shift work is associated with the risk of atrial fibrillation and coronary heart disease**. *Eur Heart J* (2021) **42** 4180-4188. DOI: 10.1093/eurheartj/ehab505
42. Vyas MV, Garg AX, Iansavichus AV, Costella J, Donner A, Laugsand LE, Janszky I, Mrkobrada M, Parraga G, Hackam DG. **Shift work and vascular events: systematic review and meta-analysis**. *BMJ* (2012) **345** e4800. DOI: 10.1136/bmj.e4800
43. Torquati L, Mielke GI, Brown WJ, Kolbe-Alexander T. **Shift work and the risk of cardiovascular disease. A systematic review and meta-analysis including dose-response relationship**. *Scand J Work Environ Health* (2018) **44** 229-238. DOI: 10.5271/sjweh.3700
44. Thosar SS, Butler MP, Shea SA. **Role of the circadian system in cardiovascular disease**. *J Clin Invest* (2018) **128** 2157-2167. DOI: 10.1172/JCI80590
45. Bass J, Lazar MA. **Circadian time signatures of fitness and disease**. *Science* (2016) **354** 994-999. DOI: 10.1126/science.aah4965
46. Liang Z, Wang W, Wang Y, Ma L, Liang C, Li P, Yang C, Wei F, Li S, Zhang L. **Urbanization, ambient air pollution, and prevalence of chronic kidney disease: a nationwide cross-sectional study**. *Environ Int* (2021) **156** 106752. DOI: 10.1016/j.envint.2021.106752
47. Okuliarova M, Mazgutova N, Majzunova M, Rumanova VS, Zeman M. **Dim light at night impairs daily variation of circulating immune cells and renal immune homeostasis**. *Front Immunol* (2020) **11** 614960. DOI: 10.3389/fimmu.2020.614960
48. Tang Z, Li S, Shen M, Xiao Y, Su J, Tao J, Wang X, Shan S, Kang X, Wu B. **Association of exposure to artificial light at night with atopic diseases: a cross-sectional study in college students**. *Int J Hyg Environ Health* (2022) **241** 113932. DOI: 10.1016/j.ijheh.2022.113932
49. Shi K, Yu B, Huang Y, Hu Y, Yin B, Chen Z, Chen L, Wu J. **Evaluating the ability of NPP-VIIRS nighttime light data to estimate the gross domestic product and the electric power consumption of China at multiple scales: a comparison with DMSP-OLS data**. *Remote Sensing* (2014) **6** 1705-1724. DOI: 10.3390/rs6021705
|
---
title: Conceptualizing multi-level determinants of infant and young child nutrition
in the Republic of Marshall Islands–a socio-ecological perspective
authors:
- Stephen R. Kodish
- Maryam Matean
- Kelsey Grey
- Uma Palaniappan
- Martina Northrup-Lyons
- Akshata Yalvigi
- Stanley Gwavuya
- Judy Mclean
- Wendy Erasmus
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022247
doi: 10.1371/journal.pgph.0001343
license: CC BY 4.0
---
# Conceptualizing multi-level determinants of infant and young child nutrition in the Republic of Marshall Islands–a socio-ecological perspective
## Abstract
The East and Pacific region includes 14 Pacific Island Countries where, between 2000 and 2016, indicators of stunting, wasting, and micronutrient deficiencies have plateaued or worsened, while rates of overweight, obesity, and associated disease have risen. The Republic of Marshall Islands (RMI) is no exception: maternal and child nutrition indicators have not improved in decades. A study of the contemporary factors shaping the RMI nutrition situation was needed for informing policy and tailoring interventions. This formative study had an iterative design utilizing qualitative methods. An exploratory Phase 1 included 59 semi-structured interviews with community members, 86 free lists with caregivers, 8 participatory workshops, and 20 meal observations (round 1). Findings were synthesized to inform a confirmatory Phase 2 where 13 focus groups, 81 pile sorts, 15 meal observations (round 2), and 2 seasonal food availability workshops were conducted. Textual data were analyzed thematically using NVivo while cultural domain analysis was conducted in Anthropac. RMI faces interrelated challenges that contribute to a lack of nutritious and local food availability, which is compounded by high food costs relative to household incomes. A decades-long cultural transition from local to processed diets has resulted in infant and young child diets now characterized by morning meals of doughnuts, bread, and ramen with tea, coffee, or Kool-Aid and afternoon meals that include rice with canned meats (e.g., store-bought chicken, hot dogs). Individual preferences for processed food imports have increased their supply. Low maternal risk perception toward nutrition-related illnesses may further explain sub-optimal diets. Improving the RMI food environment will require approaches that align with the multi-level determinants of sub-optimal diets found in this study. As the ten-year 2013 RMI Food Security Policy soon ends, study findings may be used to inform new policy development and investments for improving the nutrition situation.
## Introduction
The East Asia and Pacific region is comprised of 37 countries where economies have grown, health care access has improved, and child mortality has reduced [1]. Despite these gains, the regional nutrition situation remains serious: nearly all forms of malnutrition co-exist among most segments of the population, but especially young children [2]. Stunting (13 million), wasting (4.5 million) and overweight (9.7 million) represent the largest burden of malnutrition among children under 5 years of age in this region [3]. This phenomenon is not unique to the region: globally, countries developing economically are moving from traditional, local diets to food systems that are delocalized and characterized by high-energy, nutrition-poor food choices [4].
A sub-set of this region includes 14 Pacific Island Countries (PIC), where the nutrition situation is just as serious. Between 2000 and 2016, indicators of stunting, wasting, and micronutrient deficiencies have plateaued or only marginally worsened, while rates of overweight, obesity, and associated diseases have sharply risen [2, 5]. Among children, for instance, there has been a $4\%$ increase in stunting and an $88\%$ increase in overweight during this period. At any given time, a third of children in PICs suffer from micronutrient deficiencies. Poverty and inequity, poor maternal nutrition, and sub-optimal health and nutrition practices are leading drivers of child malnutrition in this region; however, their contributions vary by country where the burden of malnutrition also differs [2].
The Republic of the Marshall Islands (RMI) is a sovereign PIC where an approximate 55,000 people live on 29 atolls and five coral islands with 180 square kilometers of land mass [6]. Historically, malnutrition has been a documented challenge for a Marshallese population that lost livelihoods and faced hunger due to food shortages after forced relocation to outer islands due to United States (U.S.) military actions during World War II [7–9]. Since then, then RMI nutrition situation has dramatically changed, and indicators of population hunger are now low, with child wasting at $4.0\%$ [3]. However, the co-occurrence of micronutrient deficiencies, especially clinical vitamin A and iron deficiency anemia, and overweight and obesity have been well documented in RMI [10–14].
Indicators of the maternal and child nutrition situation of RMI have not improved in decades. In 2003, two thirds of adult women were overweight ($29\%$) or obese ($31\%$), while $35.5\%$ of children under 5 years were stunted. By 2020, $70.5\%$ of women were overweight ($28.4\%$) or obese ($42.1\%$) and $37.1\%$ of children under 5 years had stunted growth, a ‘very high’ prevalence classification by the WHO [10]. Also, $25\%$ of households with a child under 5 years included both a stunted child and an overweight or obese mother (i.e. mother-child double burden (MCDB)) [15].
Upstream drivers of malnutrition in RMI are consistent with countries moving through the nutrition transition globally: poverty and inequity, changing demographics, delocalized food systems, and resulting sub-optimal diets reliant on highly processed, nutrient-dense food imports [4]. Specific risk factors for maternal and child malnutrition have recently been investigated in RMI. For instance, short maternal height (<160cm) is a leading predictor of MCDB as well as a risk factor of child stunting in RMI where over a quarter ($27.5\%$) of adult women are shorter than 150cm [10]. But individual-level determinants of nutritional status fall short of telling the full story of RMI whose complex colonial history lends itself to understanding health and disease through more than mere biomedical modelling [16].
Within the RMI food system, individuals make dietary and feeding-related decisions multiple times, every day. Understanding the drivers of those food and feeding choices at multiple, interrelated behavioral levels has not been completed since in-depth work by Gittelsohn and colleagues between 1998–2001 [17, 18]. Therefore, in collaboration with local stakeholders, we carried out a mixed-methods, qualitative study to explore contemporary, community-level drivers of infant and young child feeding practices in urban and rural RMI. This community-based, participatory study had the following aims: 1) understand how household food security may influence diets; 2) contextualize infant and young child feeding practices; 3) explore local conceptions of illness in relation to feeding behaviors; and 4) generate community recommendations for improving the nutrition situation.
## Study setting
The RMI has a hot and humid tropical climate that averages 27 degrees Celsius and experiences both a wet season (May–October) and a dry season (November–April) annually [19]. Its societal kinship structure is matrilineal, with a primarily indigenous Marshallese population that speaks both Marshallese and English. Nearly $75\%$ of the RMI population lives in two urban centers, Majuro (Majuro Atoll) and Ebeye (Kwajalein Atoll) [20]. All other RMI atolls are classified as rural.
This study was conducted in both Majuro [urban] and Arno [rural] communities. Majuro is the RMI capital and the economic and political center, where approximately $52\%$ (27,797 residents) of the population lives [20]. Urban fieldwork was conducted in Djarrit town, which is located on the northern end of Majuro atoll and is home to nearly 5,000 people who live on just 0.42 square kilometers of land. Rural fieldwork was completed in Arno and Ine villages on Arno atoll. Arno is the geographically closest atoll to Majuro with a population of nearly 2,000 people. In Majuro, the high population density has put pressure on the delivery of basic social and health services, whereas in Arno the logistics and costs of reaching rural communities only accessible by plane or boat are challenges to service delivery [20].
## Study design and data collection methods
This multi-phase, formative study had an iterative design utilizing qualitative methods. A local data collection team was hired based on previous experience conducting health-related fieldwork, secondary school education level, computer literacy, and language proficiency. Data collection occurred from August until October 2018.
## Exploratory phase 1
During Phase 1, a total of 59 semi-structured interviews were conducted with female caregivers ($$n = 26$$), male caregivers ($$n = 13$$), professional health workers ($$n = 10$$), senior-level health and agricultural staff ($$n = 3$$), and community leaders ($$n = 7$$). Participants were asked semi-structured questions covering key domains including but not limited to community characteristics, health behaviors, food security, feeding and hygiene practices, household member roles and responsibilities, and communication channels. Free lists ($$n = 86$$) were conducted with caregivers of young children to explore food and illness cognitive domains. Community members joined participatory workshops ($$n = 8$$) to identify and vote on the top barriers and solutions to optimal nutrition. Direct observations ($$n = 35$$) of infant and young child meals were conducted in phase 1 and repeated in the same households during phase 2 to reduce reactivity [21].
## Confirmatory phase 2
Phase 2 was designed to clarify and corroborate findings from Phase 1. Focus group discussions ($$n = 13$$) were conducted with approximately 6–10 caregivers per discussion to investigate social norms and clarify findings from phase 1 interviews. Seasonal food availability workshops ($$n = 2$$) generated data to create food availability calendars across seasons. Food availability was specified by symbols indicating foods that are 1) not available, 2) low availability, 3) medium availability, and 4) high availability. Phase 1 free list findings generated items for pile sorts ($$n = 81$$) to explore local conceptions of food and illness.
## Sampling procedures
Local health workers with knowledge of the community assisted with participant recruitment during both study phases. Health workers escorted research team members to selected communities during fieldwork activities, assisting in courtesy meetings with local leaders and identification of prospective participants. Study participants were purposively sampled at multiple behavioral levels of the socio-ecological model, which served as the theoretical foundation of this study [22–24] (Table 1).
**Table 1**
| Level of influence | Participant types |
| --- | --- |
| Policy | Senior-level health staff from RMI Ministry of Health and Human Services |
| Organizational | Professional and traditional health workers |
| Community | Community leaders |
| Interpersonal | Male caregivers, grandparents |
| Individual | Female caregivers |
Sample sizes were calculated based on achieving data saturation for qualitative methods (i.e., interviews, focus groups, observations) and validity for ethnographic methods generating cultural domain data (i.e., free lists and pile sorts) [25, 26].
## Textual analysis
Textual analysis of interviews and focus group data took an inductive approach drawing from Grounded Theory [27]. To do so, an initial codebook was developed using the contents of the data collection instruments. Codes were then used to assign meaning to units of text until themes and sub-themes relevant to the guiding study aims emerged across transcripts. During the coding process using NVivo software, the codebook was continually refined using a team-based coding approach [28, 29]. Textual field notes from direct observations were thematically analyzed by infant or child age range (6–11 mo.; 12–23 mo.). Findings across qualitative methods were synthesized to draw conclusions pertinent to the study aims.
## Cultural domain analysis
Using Anthropac software, free list data were tabulated to generate a salience statistic for each item based on both frequency of mention and relative rank across lists [30, 31]. Those items with a salience ≥0.30 were used to generate pile sort terms which were analyzing using multi-dimensional scaling procedures. Findings were displayed using maps that fit items of each food and illness domain into two-dimensional figures that considered goodness-of-fit through measurement of a Stress statistic (0 (worst possible fit)– 1 (best possible fit)) [31]. Field notes and interview data were then synthesized to contextualize cultural domain results and create an ethnomedical model of illness.
## Numerical analysis
*Data* generated from seasonal food availability calendar and participatory workshops were tallied and ranked based participant votes. Descriptive field notes were reviewed to help explain numerical values.
## Ethical considerations
Ethics approval for the study was granted by the RMI Ministry of Health and Human Services, which approved the use of verbal informed consent due to the minimal risks associated with the study. Verbal informed consent was obtained from participants and recorded with digital recorders prior to data collection. In cases when digital recorders were not used for collecting a specific type of data (e.g., cultural domain data) then written informed consent was obtained.
## Food availability
Underlying nutrition-related challenges in RMI are the lack of nutritious food availability and differential levels of nutritious food access for community members in both Majuro [urban] and Arno [rural]. While there are important seasonal influences on local food availability in RMI, the food system relies on processed food imports, which are more readily available than local, fresh foods across seasons. ‘ Energy foods’ (e.g., rice, bread) and ‘body-building foods’ (e.g., fresh fish, canned meat) have year-round availability. The majority of ‘protective foods’ (e.g., pandanus, coconut, papaya) have medium availability throughout the year. Local, fresh foods (e.g., breadfruit, taro, makmok, pandanus, coconut, papaya) have variable availability across seasons, especially in Majuro [urban] markets (Tables 2 and 3).
The primary challenges to growing food in Majuro [urban] included insufficient space for gardening and refusal of land owner permission to plant. Challenges to home agricultural production in Arno [rural] included not having proper training, lacking the necessary gardening tools and seeds, and disruption by foraging animals which ruins gardens. In addition, copra production, which receives government subsidy, is now more profitable for landowners and, as a result, contributing to reduced agricultural production.
## Food access
Food choice on RMI is also governed largely by what food items are available and affordable on any given day, as imported foods are inconsistently stocked, and household incomes are not always aligned with available supplies.
In Majuro [urban], the key barriers identified were financial as both low household income and high food prices were commonly mentioned. Urban households rely on store-bought foods as lack of space makes homestead production difficult. Fresh fruits and vegetables are often unaffordable so many families purchase cheaper, processed foods to feed large families.
Structured pile sorting of foods based on affordability corroborated interview and workshop data indicating high affordability of the least nutritious food options (e.g., candy, chips, ramen) relative to others typically provided to infants and young children (Fig 1).
**Fig 1:** *Multi-dimensional scaling map of infant and young child foods by affordability in urban Majuro.S = 0.11, Eigenvalue: 11.03, Eigenratio: 7.64.*
Family size also influences food purchasing patterns, favoring more affordable food items in larger quantities that will be adequate to sustain everyone; household sizes in this study ranged from 4 to 14 including both nuclear and extended family. Food quantity is perceived to be more highly valued than food quality among most study participants whose primary concern is household nourishment.
To the contrary, Arno [rural] community members did not identify financial barriers to food access as the primary challenge for optimal nutrition but instead explained that the lack of agricultural production creates a greater reliance on important food options. Barriers to homestead production of fresh fruits and vegetables, specifically, included poor access to seeds and other supplies, reliance on copra production for household income, and limited knowledge of farming techniques.
Pile sorting of foods based on reported affordability in Arno [rural] supported interview and workshop data that locally available foods (e.g., papaya, pandanus, coconut) are typically more affordable than imported options (Fig 2).
**Fig 2:** *Multi-dimensional scaling map of infant and young child foods by affordability in rural Arno.S = 0.06, Eigenvalue: 20.74, Eigenratio: 8.51.*
Further, interview data suggest that over time individual preferences for the convenience of processed foods (higher demand) have increased their supply. Eating processed foods is now an accepted social norm in both urban and rural RMI for adults and children alike. Food purchasing is now considered to be just “more convenient” and “faster” than planting or fishing in both rural and urban settings.
## Breastfeeding
Caregivers and health workers both reported that early initiation of breastfeeding is now commonly practiced in RMI, a key behavior promoted by professional health workers. In the past, colostrum was not perceived to be healthful, but perceptions toward it are positively changing thanks to health workers who now promote it. Barriers to exclusive breastfeeding 0–6 months do exist: firstly, it is a socially-accepted norm that breastmilk stores are insufficient for infant nutrient needs, particularly in rural areas where health workers reinforce this notion.
Secondly, traditional medicines, including herbal medicines made with untreated water for kijonkan (infant jaundice), are common treatments for infants in early life. There is a positive influence of healthcare professionals (doctors, nurses) who do promote exclusive and continued breastfeeding practices which can persist well after a child’s second birthday in some households. Direct observations revealed that continued breastfeeding occurred in $38\%$ (urban) and $56\%$ (rural) of study households with children aged 6–23 months, however. Both individual and community perceptions contribute to breastfeeding cessation before two years.
## Feeding of infants and children aged 6–23 months
Caregivers reported feeding infants and young children aged 6–23 months three meals per day: morning, midday, and evening meals. Meal observations found that typical household meals consisted disproportionately of store-bought foods in both Majuro [urban] and Arno [rural] RMI. Morning food items included bread, pancake, doughnut, ramen, and either tea, coffee, or Kool-Aid. Mid-day and evening meals consisted of rice, canned meat, and fish in Arno [rural], with the addition of store-bought chicken and hot dogs in Majuro [urban].
Intra- and inter-household food sharing was observed during most meals and was identified as a core cultural value reflective of an underlying interdependent cultural context. The Marshallese phrase, jake jebol eo (sharing is caring), which was identified during interviews, was explicitly expressed through mealtime food sharing.
Common first foods for infants included pandanus juice and pandanus paste, both locally available. In Majuro [urban], store-bought baby foods and cereals were also commonly fed to infants. Imported rice, the primary staple of RMI, contributed to the largest proportion of energy intake among infants and young children. Infants and young children commonly ate foods comprised of processed starches and refined carbohydrates such as ramen, bread, and store-bought biscuits during meals. Fresh fruits and vegetables were said to be nutritious and important for good health during interviews, but direct observations revealed a reality of very limited consumption among infants and young children, particularly in rural households. Caregivers largely used food-based incentives (e.g., juices, biscuits) to soothe and to encourage eating during meals.
Observations of infants aged 6–11 months, specifically, revealed very little consumption of animal-source proteins and instead a reliance on a locally-made watery porridge called likobla, comprised of flour, coconut milk, and sugar. Among young children aged 12–23 months, approximately half of observed households provided meals to children with at least one animal source, including canned tuna or mackerel, corned beef, Vienna sausage, ham, or hotdogs. All household meals in Arno [rural] and Majuro [urban] included at least 3 sources of refined carbohydrates, typically white rice and flour-based foods such as bread or donuts.
While snacking was not commonly observed among infants aged 6–11 months, it was very common among young children aged 12–23 months and largely consisted of chips, cookies, sweets, and chocolate food items. Interview data suggest consistently high knowledge among caregivers about the importance of dietary diversity for infant and young child nutrition, citing the healthfulness of fresh fish and a wide range of seasonal fruits and vegetables. In practice, however, locally-available, nutritious foods were not central to infant and young child diets during meal observations. Tea and coffee consumption by infants and young children were reported in interviews but rarely observed; sugary drink consumption (e.g., Kool-Aid) was commonly observed though.
## Local conceptions of illness
Local conceptions of infant and young child illness may help to explain the gap between knowledge and practice in RMI. The most salient infant and young child illnesses in RMI were fever, cough, and diarrhea. Nutrition-related illness terms, such as malnutrition, micronutrient deficiencies, underweight, overweight/obesity, wasting, and stunting, did not emerge among the top ten most salient illnesses to caregivers in either Majuro [urban] or Arno [rural], suggesting lower perceived importance relative to other illnesses (Table 4).
**Table 4**
| Rank | Illness term(Marshallese) | Illness term(English) | Frequency(%) | AverageRank | Salience(S) |
| --- | --- | --- | --- | --- | --- |
| 1 | Bwil | Fever | 93.5 | 2.16 | 0.74 |
| 2 | Bok bok | Cough | 80.4 | 2.19 | 0.61 |
| 3 | Bidrodro | Diarrhea | 69.6 | 3.28 | 0.39 |
| 4 | Pilo | Pink eye | 32.6 | 4.0 | 0.16 |
| 5 | Kajjinok | AsthmaShortness of breath | 21.7 | 2.8 | 0.14 |
| 6 | Metak lojeen | Stomach ache | 17.4 | 3.38 | 0.1 |
| 7 | Uwor | Runny nose | 17.4 | 4.63 | 0.09 |
| 8 | Molanlon | Nausea | 15.2 | 4.71 | 0.07 |
| 9 | Metak boran | Headache | 10.9 | 4.0 | 0.07 |
| 10 | Kor kori | Skin rash | 23.9 | 6.36 | 0.06 |
In Majuro [urban], 28 unique illnesses were reported during free listing. Local terms for jabwe on (lacking nutrients/malnutrition) and jabwe botoktok (anemia) were ranked at 17th and 27th, respectively (Table 5).
**Table 5**
| Rank | Illness term(Marshallese) | Illness term(English) | Frequency(%) | AverageRank | Salience(S) |
| --- | --- | --- | --- | --- | --- |
| 1 | Bwil | Fever | 97.5 | 1.51 | 0.87 |
| 2 | Bokbok | Cough | 80.0 | 2.06 | 0.65 |
| 3 | Bidrodro | Diarrhea | 67.5 | 3.44 | 0.39 |
| 4 | Kor kori | Skin rash | 40.0 | 4.94 | 0.19 |
| 5 | Molanlon | Nausea | 37.5 | 4.87 | 0.17 |
| 6 | Pilo | Pink eye | 27.5 | 4.36 | 0.15 |
| 7 | Uwor | Runny nose | 27.5 | 4.45 | 0.14 |
| 8 | Ebboj lojeen | Stomach bump | 22.5 | 5.56 | 0.09 |
| 9 | Metak boran | Headache | 22.5 | 5.11 | 0.11 |
| 10 | Wot | Boil | 17.5 | 6.29 | 0.07 |
In Arno [rural], 26 unique illnesses were reported, with nutrition-related illnesses ranking 13th and 14th for jabwe on (lacking nutrients/malnutrition) and pilo in bon (night blindness), respectively. Interviews further explored conceptions of local illnesses. Jabwe on (lacking nutrients/malnutrition), jabwe botoktok (anemia), and pilo in bon (night blindness) were ascribed to “missing key foods in the diet” suggesting an understanding of the connection between illness and diet. Free list and interview findings were synthesized to create an ethnomedical model of illness using the most salient illness terms (Fig 3).
**Fig 3:** *Ethnomedical model of salient infant and young child illnesses.Each perceived cause and prevention/treatment strategy represents a salient theme from interview or free list data.*
Locally specific conceptions of illness etiology, such as ebboj (stomach bump caused from a fall), were found to be important perceived causes of childhood illnesses. Fever was said to be caused by exposure to cold air, cold water, or ebboj. Ebboj can only be cured through traditional medicines, as it is not a clinical diagnosis and is not formally treated.
The ethnomedical model suggests an underlying medical belief system that considers a combination of biomedical and traditional explanations for disease. Preventative strategies and remedies for bwil (fever), bok bok (cough), and bidrodro (diarrhea) included a combination of both traditional (e.g. ‘using a wet cloth to suck out a fever’) and clinical approaches (e.g. ‘giving Tylenol’).
Decisions of when and how to seek care depends on type of illness (e.g., ebboj (stomach bumps) are only curable through traditional medicine) as well as consideration of the associated costs, such as those for transportation and clinic fees.
## Community recommendations for improving the nutrition situation
During participatory community workshops, the top voted solutions to improve infant and young child nutrition differed between [Majuro] urban and Arno [rural], given their context-specific challenges. The top-voted solutions to overcome low financial access in Majuro [urban] communities included: 1) improving food affordability by decreasing prices of healthy foods, 2) increasing production of traditional Marshallese foods, 3) increasing household salaries, and 4) government-funded food programs for school-aged children.
The top voted solutions to help overcome low agricultural production in Arno [rural] communities included: 1) making farming more accessible by providing seeds and gardening tools, 2) implementing programming to teach effective farming methods, 3) improving the nutrition of young children through school lunch programs, and 4) increasing availability of technology to help with food storage.
## Discussion
The RMI nutrition situation is as serious as it is complex. Studies to understand these drivers have illustrated that it is not one but a combination of enabling, underlying, and immediate factors that, in combination, influence population nutrition [32]. In RMI, we observed meals where young child diets were predominantly comprised of nutrient-dense, processed food imports. Seasonal food availability calendars and interviews with agricultural experts, leaders, and caregivers illustrated how those diets are a result of an unreliable food system for ensuring nutritious diets. We found a food environment where nutritious food availability is limited and inconsistent across seasons, coupled with low food access due to high prices relative to incomes. Losses of traditional livelihoods of fishing and agriculture have contributed to a food environment where it is easier to find and feed energy-dense, non-nutritious young child foods than it is locally available nutritious diets. Understanding the present-day food environment of RMI benefits from a historical perspective that weaves together a complex narrative involving colonialism, trade liberalization, aid dependence, and present-day policies.
RMI was occupied by Japan during World War I and the U.S. during World War II when its outer atolls were used as a testing site for nuclear weapons [9]. In addition to the direct consequences on Marshallese health, these occupations have had far-reaching consequences for the livelihoods, health, and nutrition of Marshallese [7, 8, 33]. Historical evidence implicates this foreign influence on changes in food consumption patterns, from diets consisting of locally available foods to those reliant on processed imports [34, 35]. Rising rates of overweight and obesity coincided with the advent of U.S. subsidies in the 1960s and continue today [36]. Nowadays, the United States provides economic assistance to RMI under the Compact of Free Association, an agreement that, coupled with changes in global food trade patterns, helps ensure a foreign food import dependency [34, 37].
Between 1987 and 2013, international food trade increased dietary diversity for much of the world, but at the expense of a heighted dependency on imports [38]. In fact, more than $80\%$ of the countries in the world are now reliant on food imports to meet population demands. RMI is no exception: its food system is delocalized, whereby dietary requirements are met using food items shipped from distant places through commercial channels [39]. Dietary delocalization reflects trade liberalization policies which can result in national economic gains at the expense of population health. There is a positive relationship between trade liberalization and increased processed food imports in the Pacific Island Countries where they amounted to $9,264 million USD, continuing to increase from 2003 until 2013 [37]. The top imported food items included high sodium meat products (e.g., canned meats and fishes), as well as nutrient dense, non-nutritious cereals (e.g., rice). In RMI, at least $90\%$ of the food supply is imported, contributing to a food environment where more than one third of all households are food insecure [36, 40].
RMI communities are now characterized by a cultural context where traditional beliefs and practices persist but have been adulterated through increasing modernization. Caregivers told us that infant and young child feeding is not the sole responsibility of primary caregivers, but an effort shared by health professionals, neighbors, friends, and family. This collectivism is a core cultural value of not only Marshallese communities but also the Pacific region at large [41]. Collectivism is important for understanding food choices that are made within family structures that average 6.8 persons/household [20]. In the modern food environment where nutritious foods are expensive and not always available, food purchasing decisions consider the ability of food items to feed the entire family, resulting in preferences for foods that favor energy adequacy (e.g., rice) to nutrient quality (e.g., locally caught fish).
Modernization also means that caregivers face more competing demands than they did in the past. For families that engage in copra production, a subsidized cash crop in the RMI economy, time for homestead food production, traditional meal preparation, and optimal feeding practices were said to be “too time consuming.” Thus, individual food choices on RMI have been slowly shaped by the increased convenience and low cost of processed food imports in an economy where much time is now spent outside of the house working [42]. As a result, personal taste preferences now favor energy-dense, processed food options to traditional staples. Present-day food choice and feeding behaviors are a consequence of a rapidly changing socio-cultural landscape where economic liberalization and modernization have resulted in both high supply and demand for processed food imports among adults, adolescents, and children alike [18, 34, 37, 38, 43].
We designed this study to also generate recommendations to improve the nutrition situation. Among urban participants who primarily source food from stores on Majuro, we heard that improving food accessibility should be a priority. In 2003, Gittelsohn and colleagues implemented a 10-week store-based intervention that aimed to do so by pairing nutritious food stocking with tailored communications (e.g., recipes, cooking demonstrations, point-of-sale prompts) using mass media and in-store promotions throughout Majuro [42]. This trial increased consumer knowledge of diabetes and food label reading, as well as resulted in increased purchasing of promoted foods [44]. Importantly, the intervention was tailored to the cultural context of RMI based on in-depth formative research, one likely reason why consumer exposure to the intervention was high and positive impacts on cognitive and behavioral factors were seen [45]. Working with vendors within food stores of RMI to improve nutritious food accessibility may hold promise.
Nowadays, interventions tailored to RMI may benefit from embracing both the old and the new: designing behavior change communication strategies that value the longstanding oral traditions of storytelling and power of face-to-face communications yet harness the utility of social media and phone usage has potential for both impact and coverage [46, 47]. Increasing modernization is a reality in RMI and thus interventions aiming to reach all population segments may consider using such tailored approaches to blend the old with the new. We found that traditional medicine is an important aspect of Marshallese culture still today. It is not uncommon for infant jaundice (kinjonkan) to be treated using herbs and water, a practice reflective of an underlying medical belief system that ascribes illness to both internalizing and externalizing forces [48]. In 2003, a third of RMI survey participants agreed that diabetes is sometimes caused by ‘black magic’ [18]. While much traditional knowledge of agricultural practices and fishing has been lost during over a century of colonialization, our rural study participants suggested agricultural training and provision of farming resources to engage in homestead food production as strategies to improve the food environment. Although a three-year (2011–2014) horticulture project on RMI saw only modest positive effects on fruit and vegetable consumption through provision of resources and training, lessons learned may be applied to designing similar interventions across the food value chain of RMI [36, 49].
Tailored intervention approaches, embodied by biomedical principles that acknowledge traditional medical belief systems, may be appropriate, acceptable, and effective in RMI. For instance, investments in health worker capacity strengthening at both facility and community levels may improve service delivery. Conducting tailored trainings using formative findings that enable health assistants to become qualified nurses within facilities, as well as dually doing so to enhance capacities among the current cadre of RMI health workers to more effectively deliver preventative services, such as nutrition promotion, at community level, may help improve the full continuum of care. To be sure, investments in community health workers has been successful in similar contexts where facility-based approaches fell short of reaching more vulnerable population segments [50–53]. In RMI, community health workers may help to fill health access challenges, especially on the outer islands, where coverage of services is lower than on Majuro [54]. Additionally, Marshallese community health workers would be well positioned to listen and understand community challenges, while problem solving nutrition problems using locally available strategies and indigenous knowledge backed by evidence-informed solutions [55, 56]. Promoting nutrition from Marshallese health workers may be an appropriate, effective, and sustainable interpersonal approach to behavior change.
Improving the RMI food environment will require multi-level approaches that align with the multi-level determinants found in this study. Doing so will also benefit from multi-sectoral synergies that consider not only food and nutrition, but also, for example, water, sanitation, and hygiene, given the importance of infection for infant and young child nutritional status [57]. Regardless of the planned approaches, political will and ample time for sustainable and successful implementation will be foundational. Now is an opportune time to evaluate the effectiveness of the 2013 RMI Food Security Policy, which will soon be 10 years old and may be coupled with a 2023 health financing transition with implications for public health service delivery [58, 59]. Critically reviewing the policy contents and their implementation, collating the lessons learned from previous interventions, and incorporating research findings, both old and new, may help to shape a forward-thinking ‘do no harm’ policy inclusive of the present-day food environment with consideration to new challenges such as COVID-19 and climate change [19, 60–62].
## Conclusion
Creating a healthy food environment where caregivers of infants and young children find accessing nutritious foods to be convenient, preferable, and affordable is important yet complex. Doing so may require an interdisciplinary, systems-based, and multi-sectoral approach backed by political will, evidence, innovation, and collaboration [12]. Contemporary nutrition challenges on RMI have been shaped by historical events and geopolitical forces that are not easily undone. Without action, the food environment of RMI will remain a place where caregivers of infants and young children struggle to find nutritious, affordable diets for healthy growth and development.
## References
1. You D, Hug L, Ejdemyr S, Idele P, Hogan D, Mathers C. **Global, regional, and national levels and trends in under-5 mortality between 1990 and 2015, with scenario-based projections to 2030: A systematic analysis by the UN Inter-agency Group for Child Mortality Estimation**. *The Lancet* (2015.0) **386** 2275-86
2. Blankenship JL, Rudert C, Aguayo VM. **Triple trouble: Understanding the burden of child undernutrition, micronutrient deficiencies, and overweight in East Asia and the Pacific.**. *Matern Child Nutr.* (2020.0) **16** e12950. DOI: 10.1111/mcn.12950
3. 3UNICEF. The state of the world’s children 2019.
New York:
UNICEF; 2019.. *The state of the world’s children 2019.* (2019.0)
4. Popkin BM, Corvalan C, Grummer-Strawn LM. **Dynamics of the double burden of malnutrition and the changing nutrition reality**. *The Lancet* (2020.0) **395** 65-74. DOI: 10.1016/S0140-6736(19)32497-3
5. Ravuvu A, Waqa G. **Childhood obesity in the Pacific: Challenges and opportunities.**. *Curr Obes Rep* (2020.0) **9** 462-9. DOI: 10.1007/s13679-020-00404-y
6. 6Department of Foreign Affairs and Trade. Republic of the Marshall Islands country brief: Australian Government;
2022 [Available from: https://www.dfat.gov.au/geo/republic-of-marshall-islands/republic-of-the-marshall-islands-country-brief#:~:text=RMI%20is%20an%20archipelago%20of,RMI’s%20capital%20is%20Majuro].. *Republic of the Marshall Islands country brief: Australian Government;* (2022.0)
7. Niedenthal J.. **A history of the people of Bikini following nuclear weapons testing in the Marshall Islands: With recollections and views of elders of Bikini Atoll**. *Health Phys* (1997.0) **73** 28-36. DOI: 10.1097/00004032-199707000-00003
8. Palafox NA. **Health consequences of the Pacific U.S. Nuclear Weapons Testing Program in the Marshall Islands: Inequity in protection, health care access, policy, regulation**. *Rev Environ Health* (2010.0) **25** 81-5. DOI: 10.1515/reveh.2010.25.1.81
9. Simon SL. **A brief history of people and events related to atomic weapons testing in the Marshall Islands**. *Health Phys* (1997.0) **73** 5-20. DOI: 10.1097/00004032-199707000-00001
10. Blankenship JL, Gwavuya S, Palaniappan U, Alfred J, deBrum F, Erasmus W. **High double burden of child stunting and maternal overweight in the Republic of the Marshall Islands.**. *Matern Child Nutr* (2020.0) **16** e12832. DOI: 10.1111/mcn.12832
11. Gamble MV, Palafox NA, Dancheck B, Ricks MO, Briand K, Semba RD. **Carotenoid status among preschool children with vitamin A deficiency in the Republic of the Marshall Islands.**. *Asia Pac J Clin Nutr* (2004.0) **13** 336-40. PMID: 15563437
12. Ichiho HM, deBrum I, Kedi S, Langidrik J, Aitaoto N. **An assessment of non-communicable diseases, diabetes, and related risk factors in the Republic of the Marshall Islands, Majuro Atoll: A systems perspective.**. *Hawaii J Med Public Health.* (2013.0) **72** 87-97. PMID: 23901367
13. Palafox NA, Gamble MV, Dancheck B, Ricks MO, Briand K, Semba RD. **Vitamin A deficiency, iron deficiency, and anemia among preschool children in the Republic of the Marshall Islands.**. *Nutrition* (2003.0) **19** 405-8. DOI: 10.1016/s0899-9007(02)01104-8
14. Passmore E, Smith T. **Dual burden of stunting and obesity among elementary school children on Majuro, Republic of Marshall Islands.**. *Hawaii J Health Soc Welf.* (2019.0) **78** 262-6. PMID: 31463476
15. **Republic of the Marshall Islands Integrated Child Health and Nutrition Survey 2017, Final Report.**. *Majuro, Republic of the Marshall Islands* (2017.0)
16. Yamada S, Palafox N. **On the biopsychosocial model: The example of political economic causes of diabetes in the Marshall Islands**. *Fam Med* (2001.0) **33** 702-4. PMID: 11665910
17. Cortes LM, Gittelsohn J, Alfred J, Palafox NA. **Formative research to inform intervention development for diabetes prevention in the Republic of the Marshall Islands**. *Health Educ Behav* (2001.0) **28** 696-715. DOI: 10.1177/109019810102800604
18. Gittelsohn J, Haberle H, Vastine AE, Dyckman W, Palafox NA. **Macro- and microlevel processes affect food choice and nutritional status in the republic of the marshall islands**. *J Nutr* (2003.0) **133** 310s-3s. DOI: 10.1093/jn/133.1.310S
19. **Adaptation communication report—Republic of the Marshall Islands.**. *Majuro, Republic of the Marshall Islands: Climate Change Directorate* (2020.0)
20. **The RMI 2011 census of housing and population summary and highlights.**. *Majuro, Republic of the Marshall Islands: Government of the Republic of the Marshall Islands* (2011.0)
21. Gittelsohn J, Shankar AV, West KP, Ram RM, Gnywali T. **Estimating reactivity in direct observation studies of health behaviors.**. *Human Organization* (1997.0) **56** 182-9
22. Etikan I, Abubakar Musa S, Sunusi Alkassim R. **Comparison of convenience sampling and purposive sampling**. *American Journal of Theoretical and Applied Statistics* (2016.0) **5**
23. Gentles S, Charles C, Ploeg J, McKibbon K. **Sampling in qualitative research: Insights from an overview of the methods literature.**. *The Qualitative Report.* (2015.0) **20** 1772-89
24. Owen N, Fisher E, Glanz K, Rimer B, Viswanath K. **Ecological models of health behavior.**. *Health behavior and health education: theory, research, and practice San Francisco: Jossey-Bass;* (2008.0)
25. Borgatti S, Halgin DS. **Elicitation techniques for cultural domain analysis.**. *The Ethnographer’s Toolkit* 115-51
26. Fusch P, Ness L. **Are we there yet? Data saturation in qualitative research**. *Qualitative Report.* (2015.0) **20** 1408-16
27. Charmaz K.. *Constructing grounded theory.* (2014.0)
28. AlYahmady HH, Alabri SS. **Using Nvivo for data analysis in qualitative research**. *International Interdisciplinary Journal of Education* (2013.0) **2** 181-6
29. MacQueen KM, McLellan E, Kay K, Milstein B. **Codebook development for team-based qualitative analysis**. *CAM Journal* (1998.0) **10** 31-6
30. Technologies Analytical. *Anthropac software.* (1992.0)
31. Weller S, Romney AK. *Systematic data collection.* (1998.0)
32. 32UNICEF. UNICEF conceptual framework on maternal and child nutrition, 2020: A framework for the prevention of malnutrition in all its forms. New York, USA; 2021.
33. Duke MR. **Neocolonialism and health care access among Marshall Islanders in the United States.**. *Med Anthropol Q* (2017.0) **31** 422-39. DOI: 10.1111/maq.12376
34. Cassels S.. **Overweight in the Pacific: Links between foreign dependence, global food trade, and obesity in the Federated States of Micronesia.**. *Globalization and Health.* (2006.0) **2** 10. DOI: 10.1186/1744-8603-2-10
35. Englberger L, Marks GC, Fitzgerald MH. **Insights on food and nutrition in the Federated States of Micronesia: A review of the literature.**. *Public Health Nutr.* (2003.0) **6** 5-17. DOI: 10.1079/PHN2002364
36. Cheng Y-T, Tseng Y-C, Iwaki Y, Huang MC. **Sustainable food security in Small Island Developing States (SIDS): A case of horticulture project in Marshall Islands.**. *Marine Policy.* (2021.0) **128** 104378
37. Ofa SV, Gani A. **Trade policy and health implication for Pacific island countries**. *International Journal of Social Economics* (2017.0) **44** 816-30
38. Kummu M, Kinnunen P, Lehikoinen E, Porkka M, Queiroz C, Röös E. **Interplay of trade and food system resilience: Gains on supply diversity over time at the cost of trade independency.**. *Global Food Security* (2020.0) **24** 100360
39. Pelto GH, Pelto PJ. **Diet and delocalization: Dietary changes since 1750**. *The Journal of Interdisciplinary History* (1983.0) **14** 507-28. PMID: 11617350
40. Troubat NaS M.K.. **Food consumption in the Marshall Islands–Based on analysis of the 2019/20 Household Income and Expenditure Survey**. *Majuro, Republic of Marshall Islands: FAO and SPC* (2021.0)
41. Matenga-Ikihele A, McCool J, Dobson R, Fa’alau F, Whittaker R. **The characteristics of behaviour change interventions used among Pacific people: a systematic search and narrative synthesis**. *BMC Public Health* (2021.0) **21** 435. DOI: 10.1186/s12889-021-10420-9
42. Gittelsohn J, Dyckman W, Tan ML, Boggs MK, Frick KD, Alfred J. **Development and implementation of a food store-based intervention to improve diet in the Republic of the Marshall Islands**. *Health Promot Pract* (2006.0) **7** 396-405. DOI: 10.1177/1524839905278620
43. Kuhnlein HV, Receveur O. **Dietary change and traditional food systems of indigenous peoples**. *Annu Rev Nutr* (1996.0) **16** 417-42. DOI: 10.1146/annurev.nu.16.070196.002221
44. Gittelsohn J, Dyckman W, Frick KD, Boggs MK, Haberle H, Alfred J. **A pilot food store intervention in the Republic of the Marshall Islands.**. *Pac Health Dialog* (2007.0) **14** 43-53. PMID: 19588606
45. Bentley ME, Johnson SL, Wasser H, Creed-Kanashiro H, Shroff M, Fernandez Rao S. **Formative research methods for designing culturally appropriate, integrated child nutrition and development interventions: an overview**. *Ann N Y Acad Sci* (2014.0) **1308** 54-67. DOI: 10.1111/nyas.12290
46. Kernbach S.. *Storytelling Canvas: A Visual Framework for Developing and Delivering Resonating Stories* (2018.0) 390-5
47. Sy A, Tannis C, McIntosh S, Demment M, Tomeing T, Marriott J. **An Assessment of E-health Resources and Readiness in the Republic of the Marshall Islands: Implications for Non-communicable Disease Intervention Development.**. *Hawaii J Health Soc Welf* (2020.0) **79** 52-7. DOI: 10.1089/tmj.2006.0064
48. Young A.. **Internalizingand externalizing medical belief systems: an Ethiopian example**. *Soc Sci Med (1967)* (1976.0) **10** 147-56. DOI: 10.1016/0037-7856(76)90041-x
49. Bammann H.. *Participatory value chain analysis for improved farmer incomes, employment opportunities and food security* (2019.0)
50. Jeet G, Thakur JS, Prinja S, Singh M. **Community health workers for non-communicable diseases prevention and control in developing countries: Evidence and implications.**. *PLoS One.* (2017.0) **12** e0180640. DOI: 10.1371/journal.pone.0180640
51. Kane SS, Gerretsen B, Scherpbier R, Dal Poz M, Dieleman M. **A realist synthesis of randomised control trials involving use of community health workers for delivering child health interventions in low and middle income countries.**. *BMC Health Services Research* (2010.0) **10** 286. DOI: 10.1186/1472-6963-10-286
52. Khetan AK, Purushothaman R, Chami T, Hejjaji V, Madan Mohan SK, Josephson RA. **The effectiveness of community health workers for CVD prevention in LMIC.**. *Glob Heart* (2017.0) **12** 233-43. DOI: 10.1016/j.gheart.2016.07.001
53. Schneider H, Okello D, Lehmann U. **The global pendulum swing towards community health workers in low- and middle-income countries: a scoping review of trends, geographical distribution and programmatic orientations, 2005 to 2014.**. *Hum Resour Health* (2016.0) **14** 65. DOI: 10.1186/s12960-016-0163-2
54. Keni BH. **Training competent and effective primary health care workers to fill a void in the outer islands health service delivery of the Marshall Islands of Micronesia.**. *Human Resources for Health* (2006.0) **4** 27. DOI: 10.1186/1478-4491-4-27
55. Englberger L, Lorennij R, Taylor M, Tuia VS, Aalbersberg W, Dolodolotawake U. **Carotenoid content and traditional knowledge of breadfruit cultivars of the Republic of the Marshall Islands**. *Journal of Food Composition and Analysis* (2014.0) **34** 192-9
56. Purvis RS, Bing WI, Jacob CJ, Lang S, Mamis S, Ritok M. **Community health warriors: Marshallese community health workers’ perceptions and experiences with CBPR and community engagement.**. *Prog Community Health Partnersh* (2017.0) **11** 315-20. DOI: 10.1353/cpr.2017.0037
57. Freeman MC, Garn JV, Sclar GD, Boisson S, Medlicott K, Alexander KT. **The impact of sanitation on infectious disease and nutritional status: A systematic review and meta-analysis**. *Int J Hyg Environ Health* (2017.0) **220** 928-49. DOI: 10.1016/j.ijheh.2017.05.007
58. **Food Security Policy 2013.**. *Majuro, Republic of the Marshall Islands: Government of the Republic of the Marshall Islands* (2013.0)
59. Wheatley A, Banerji S. **Health engagement challenges and strategic perspectives for the 2023 health financing transition in the Federated States of Micronesia and the Republic of the Marshall Islands.**. *Pacific Journal of Health* (2019.0) **2**
60. Popkin BM. **To assist the large number of countries facing the double burden of malnutrition we must understand its causes and recognize the need for policies that do no harm**. *The American Journal of Clinical Nutrition* (2021.0) **113** 765-6. DOI: 10.1093/ajcn/nqaa419
61. Savage A, Schubert L, McIver L. *The nexus of climate change, food and nutrition security and diet-related non-communicable diseases in Pacific Island Countries and Territories* (2017.0)
62. Schubert L, Foley W, Savage A, Muriuki G. **Small Island Developing States, climate change, and food and nutrition security**. *Advanced Science Letters* (2017.0) **23**
|
---
title: 'Most common reasons for primary care visits in low- and middle-income countries:
A systematic review'
authors:
- Jacob Bigio
- Emily MacLean
- Nathaly Aguilera Vasquez
- Lavanya Huria
- Mikashmi Kohli
- Genevieve Gore
- Emma Hannay
- Madhukar Pai
- Pierrick Adam
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022248
doi: 10.1371/journal.pgph.0000196
license: CC BY 4.0
---
# Most common reasons for primary care visits in low- and middle-income countries: A systematic review
## Abstract
With the Covid-19 pandemic and the introduction of the WHO’s Essential Diagnostics List (EDL), increasing global attention is focused on the crucial role of diagnostics in achieving universal health coverage. To create national EDLs and to aid health system planning, it is vital to understand the most common conditions with which people present at primary care health facilities. We undertook a systematic review of the most common reasons for primary care visits in low- and middle-income countries. Six databases were searched for articles published between January 2009 and December 2019, with the search updated on MEDLINE to January 2021. Data on the most common patient reasons for encounter (RFEs) and provider diagnoses were collected. 17 of 22,279 screened articles were included. Most studies used unvalidated diagnostic classification systems or presented provider diagnosis data grouped by organ system, rather than presenting specific diagnoses. No studies included data from low-income countries. Only four studies (from Brazil, India, Nigeria and South Africa) using the ICPC-2 classification system contained RFE and provider diagnosis data and could be pooled. The top five RFEs from the four studies were headache, fever, back or low back symptom, cough and pain general/multiple sites. The top five diagnoses were uncomplicated hypertension, upper respiratory tract infection, type 2 diabetes, malaria and health maintenance/prevention. No psychological symptoms were among the top 10 pooled RFEs. There was more variation in top diagnoses between studies than top RFEs, showing the importance of creating location-specific lists of essential diagnostics for primary care. Future studies should aim to sample primary care facilities from across their country of study and use ICPC-3 to report both patient RFEs and provider diagnoses.
## Introduction
Primary health care (PHC) is a major point of entry into healthcare systems for people seeking care. Defined by the World Health Organization (WHO) as a whole-of-society approach to health that focuses on people’s needs as early as possible along the continuum of health and as close as feasible to their everyday environment [1], PHC is recognised as a cornerstone of achieving universal health coverage (UHC) and meeting the health-related Sustainable Development Goals [1, 2]. According to the WHO, scaling up PHC interventions across low- and middle-income countries (LMICs) could save 60 million lives and increase average life expectancy by 3.7 years by 2030 [2].
In recent years and especially with the ongoing Covid-19 pandemic, increasing global attention has been focused on the crucial role of diagnostics in high-quality healthcare systems, including PHC, with the introduction of the WHO’s annual Essential Diagnostics List (EDL) [3] and the formation of the Lancet Commission on Diagnostics [4]. Poor access to diagnostics, particularly in LMICs, can lead to lack of trust in health services and under-utilization of services, patients being started on presumptive or empiric treatment, which can lead to poor health outcomes, waste resources and contribute to antimicrobial resistance in the case of infectious diseases [4, 5]. Lack of diagnostics is also a major concern for managing common non-communicable diseases.
The Covid-19 pandemic has further highlighted the importance of diagnostics in curbing transmission of the virus. There has been unprecedented global collaboration through mechanisms such as the Access to Covid-19 Tools (ACT)-Accelerator Diagnostics Pillar, which is co-convened by the Foundation for Innovative New Diagnostics (FIND) and the Global Fund and aims to accelerate development, equitable allocation and delivery of diagnostic tests for Covid-19 worldwide [6].
The WHO EDL acts as a policy tool for countries to create their own national EDLs based on local contexts and needs, a process so far undertaken by Bangladesh, India, Nigeria and Pakistan [7, 8]. To create such a national EDL, and to aid health system planning, resource allocation and the training of healthcare workers, it is vital to understand both the most common symptoms with which patients present to primary care, often known as patient reasons for encounter (RFEs), and the most common provider diagnoses. Along with other sources of data such as the major causes of death and disability in a country, this information will guide the range of diagnostics required at the primary care level. Knowing the most common reasons for primary care will allow WHO, FIND and country governments to develop a package of essential tests for primary care.
The WHO is currently in discussions with other LMICs, mostly in Africa, to create their own national EDLs [7]. However, little published information is available on the reasons for primary care visits in LMICs. To our knowledge, only one systematic review, published in 2018, has summarized data on reasons for primary care visits globally [9]. It included data from only three LMICs (India, Serbia and South Africa) and pooled all studies together so the most common reasons for PHC visits in LMICs could not be distinguished from the global data which mostly focused on high-income countries (HICs). Our systematic review provides an updated summary of the reasons for primary care visits, focusing exclusively on LMICs.
## Methods
The protocol for this review was registered at the International Prospective Register of Systematic Reviews (PROSPERO), identifier CRD42020159469 [10]. Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.
## Search strategy
In our initial searches, MEDLINE, EMBASE, Global Health, Web of Science Core Collection, CINAHL, LILACS were searched for papers published between 1st January, 2009 and 12th December, 2019. Subsequently, we updated the search on MEDLINE until 1st January 2021. The search strategy was developed in consultation with a librarian (GG) and based on terms relating to “primary care” and “conditions” or “reasons” for the visit [S1 Appendix]. No restrictions on language were applied to the search. We also reviewed the papers included in the previous systematic review on this topic [9].
## Study selection
Four reviewers (JB, NAV, LH and PA) conducted the title/abstract screening, with each title/abstract independently screened by a combination of two of the four reviewers. Full text screening of included titles/abstracts was independently conducted by two reviewers (JB and PA). Articles were assessed using pre-defined inclusion and exclusion criteria, with conflicts resolved through discussion between reviewers.
Quantitative observational studies and mixed-methods studies with a quantitative observational component were included. Qualitative studies, modelling studies, economic evaluations, interventional studies and case-control studies were excluded. Studies conducted in primary care settings in LMICs (defined using the World Bank classification system [11]) reporting a minimum of five distinct RFEs or provider diagnoses were included. Studies with a data collection period of at least three months were included, though data collection did not have to occur continuously throughout the three months. Studies focusing only on specific types of visits (e.g. follow-up visits for acute conditions, referred visits, routine examinations) were excluded. Studies that selected the population based on particular symptoms (e.g. patients with fever, children with respiratory symptoms) were excluded. Studies that selected the population based on particular morbidities (e.g. HIV-positive patients, diabetic patients) or were conducted exclusively in specialized care settings (e.g. sexually transmitted infection clinics, specialized medical departments) were excluded. Studies in which data collection occurred before 1st January, 2009 were excluded. Studies not published in English or French were excluded. Editorials, commentaries, conference abstracts and grey literature were excluded.
After deduplication, 22,279 records were identified. A total of 21,938 records were excluded after title and abstract screening. Of the remaining 341 studies, three full-texts could not be retrieved and a further 321 were excluded after full-text review. The top three reasons for full-text exclusion were data collection that took place before January 1st, 2009, studies reporting fewer than five diagnoses or reasons for visits and studies being of the wrong design (e.g. case-control studies) [Fig 1]. 17 studies were included in the review. Detailed study characteristics are shown in Table 1.
**Fig 1:** *PRISMA flow diagram.* TABLE_PLACEHOLDER:Table 1 Included studies represented seven LMICs: Bangladesh ($$n = 1$$), Brazil ($$n = 3$$), Cambodia ($$n = 1$$), India ($$n = 7$$), Malaysia ($$n = 1$$), Nigeria ($$n = 2$$) and South Africa ($$n = 2$$). Each of these countries is a middle-income country (MIC)—no studies had data from low-income countries (LICs).
The number of healthcare facilities per study ranged from one to 204 (median one, interquartile range (IQR) 1–4). The number of visits reported in each study ranged from 310 to 73,236 (median 3,294, IQR 487–17,672) and was not reported for five studies. The number of diagnoses reported in each study ranged from 310 to 107,016 (median 5,692, IQR 546–31,451) and was not reported for two studies. The sources of data were medical records for 10 studies [15–19, 21, 23, 25, 26, 28], provider questionnaire for five studies [13, 20, 22, 27, 29], patient interview for one study [14] and unclear for one study [24].
## Data extraction
Data were extracted using a standardised extraction form in Google Forms, which was piloted beforehand. Three reviewers (JB, EM and PA) conducted the data extraction, with the data from each paper independently extracted by a combination of two of the three reviewers. Extracted data were compared and any discrepancies were resolved through discussion between the reviewers. Extracted data included: country of study, data collection period, type of healthcare facility, type of healthcare providers, healthcare sector, patient demographics, source of outcome data, classification system used, total number of visits or patients, number of conditions reported, frequency of different conditions.
## Quality assessment
A quality assessment tool was adapted from the work of Hoy et al [12] for use in this study. The tool has seven domains, assessing: whether the study’s target population was a close representation of the national population; whether the PHC facilities sampled were a close representation of the PHC facilities in the target area; how the sample of patient visits was chosen; how the sample of healthcare workers was chosen; whether the same mode of data collection was used for all subjects; whether the outcome measures were consistently and validly recorded; and whether the numerators and denominators were appropriate. [ S2 Appendix].
Quality assessment was conducted independently by two reviewers (JB and PA) for all included studies. Disagreements were resolved through discussion between the reviewers. All assessed studies were included, regardless of the quality assessment results.
Table 2 shows the quality assessments for the included studies. Risk of bias was high for 16 of 17 studies in the national population domain due to studies being conducted in only one province or region of a country, and so not being a close representation of the national population. Risk of bias was high for 12 of the 17 studies in the facility sampling domain due to sampled healthcare facilities not being a close representation of the province or region from which they were drawn. Risk of bias was low for 13 studies in the outcome measures domain, 14 studies in the visit sampling domain, 16 studies in the healthcare worker sampling domain and for all studies in the data collection and numerator and denominator domains.
**Table 2**
| Study | Study population representative of national population | PHC facilities in study representative of PHC facilities in area | Visit sampling | Healthcare worker sampling | Data collection | Outcome measures | Numerator and denominator |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Begum 2017 | High | High | High | Unclear | Low | High | Low |
| Chueiri 2020 | Low | Low | Low | Low | Low | Low | Low |
| Doyle 2019 | High | High | Low | Low | Low | High | Low |
| Enato 2012 | High | High | Low | Low | Low | Low | Low |
| Gupta 2014 | High | High | High | Low | Low | Low | Low |
| Gupta 2015 | High | High | High | Low | Low | Low | Low |
| Kamarudin 2012 | High | High | Low | Low | Low | Low | Low |
| Kshirsagar 2019 | High | High | Low | Low | Low | High | Low |
| Kumar 2018 | High | High | Low | Low | Low | Low | Low |
| Mash 2012 | High | Low | Low | Low | Low | Low | Low |
| Merali 2014 | High | Low | Low | Low | Low | High | Low |
| Mohan 2014 | High | High | Low | Low | Low | Low | Low |
| Olagundoye 2016 | High | High | Low | Low | Low | Low | Low |
| Prabhune 2017 | High | High | Low | Low | Low | Low | Low |
| Silva 2014 | High | Low | Low | Low | Low | Low | Low |
| Swain 2017 | High | Low | Low | Low | Low | Low | Low |
| Torres 2015 | High | High | Low | Low | Low | Low | Low |
## Data analysis
Different studies used different disease classification systems, including the International Classification of Primary Care version 2 (ICPC-2), the International Classification of Diseases 10th revision (ICD-10) and in-house classification systems. ICD-10 classifies diseases by provider diagnosis, such as essential (primary) hypertension (I10). ICPC-2 classifies diseases by patient RFE, such as fever (A03), and by provider diagnosis, such as hypertension, uncomplicated (K86).
Both ICD-10 and ICPC-2 are organised into chapters based on body systems, such as Chapter IX: Diseases of the circulatory system, and some studies presented summed totals of the number of provider diagnoses in each chapter. Such summed totals by chapter were not considered useful for the purposes of this review, as they do not give sufficient detail about the precise provider diagnoses or RFEs to help inform health system planning or the creation of national EDLs, and so they were not pooled.
Studies were pooled if at least three studies using the same classification system presented data on diagnoses or RFEs. Data from studies using different classification systems were not pooled, as it is not possible to reliably convert between different classification systems without access to individual patient data. Studies using in-house classification systems were not pooled.
Data were pooled via a rank sum system. For each study, provider diagnoses and patient RFEs were ranked from most to least common. The number of provider diagnoses ranked was determined by the study that reported the lowest number of provider diagnoses. For example, of the studies that used ICPC-2, the lowest number of provider diagnoses reported was 10 so the top 10 provider diagnoses in each study were ranked. The most common provider diagnosis in each study was assigned rank 10, the second most common chapter was assigned rank 9 and so on. Rankings from each study were combined and mean ranks were determined. The same process was followed for ranking RFEs.
## Data pooling
The classification systems used were ICPC-2 for six studies [14, 22, 25, 27–29], ICD-10 for four studies (17–19, 24), in-house or unclear for six studies [13, 15, 16, 20, 21, 23] and Medical Dictionary for Regulatory Activities System Organ Class for one study [26]. Of the four studies which used ICD-10, three presented summed totals of the number of provider diagnoses in each chapter and only one included specific provider diagnoses. Unpooled results from studies which included specific provider diagnoses (one using ICD-10 [19] and three using in-house classification systems [16, 23, 24]), and from one study which included patient RFEs using an in-house classification system [29] are shown in S3 Appendix.
Four studies (Chueiri 2020 [14], Mash 2012 [22], Olagundoye 2016 [25] and Swain 2017 [28]) using ICPC-2 contained data on both RFEs and provider diagnoses in adults and were pooled. The top nine RFEs were ranked. The five most common RFEs were headache (N01), fever (A03), back symptom/low back symptom (L02, L03), cough (R05) and pain general/multiple sites (A01).
The top 10 provider diagnoses were ranked. The five most common provider diagnoses were hypertension, uncomplicated (K86), upper respiratory tract infection (R74),type 2 diabetes (T90), malaria (A73) and health maintenance/prevention (A98). [ Table 3]. Raw data from the pooled studies are presented in S3 Appendix for reference.
**Table 3**
| Reasons for encounter | Rank score | Provider diagnoses* | Rank score.1 |
| --- | --- | --- | --- |
| Headache (N01) | 29.0 | Hypertension, uncomplicated (K86) | 37 |
| Fever (A03) | 27.0 | Upper respiratory tract infection (R74) | 23 |
| Back symptom/low back symptom (L02, L03) | 22.0 | Type 2 diabetes (T90) | 18 |
| Cough (R05) | 20.0 | Malaria (A73) | 10 |
| Pain general/multiple sites (A01) | 16.0 | Health maintenance/prevention (A98) | 10 |
| Abdominal pain/cramps general (D01) | 13.0 | Allergic rhinitis (R97) | 9 |
| Vertigo/Dizziness (N17) | 11.0 | Pregnancy (W78) | 9 |
| Heart burn (D03) | 9.0 | HIV/AIDS (B90) | 8 |
| Leg/thigh symptom/complaint (L14) | 8.0 | Visual disturbance other (F05) | 8 |
| | | Acute bronchitis/bronchiolitis (R78) | 7 |
| | | Gastroenteritis/diarrhoea (D73, D11) | 7 |
| | | Peptic ulcer (D86) | 7 |
## Discussion
Understanding patient RFEs and provider diagnoses in primary care in LMICs is of vital global health importance, as this information will guide the range of diagnostics required at the primary care level of each country. For example, malaria is often over-diagnosed in patients with febrile illness in settings in LMICs which lack access to appropriate diagnostics and presumptive treatment is given, leading to poor outcomes and the development of antimicrobial resistance [30–33]. Additionally, collecting such information could help to identify the diagnostic requirements of primary care healthcare providers in different settings and so guide the focus of future research and development in diagnostic technologies. In addition to guiding the choice of essential diagnostics, RFEs have a range of other uses. They can help to guide the choice of essential medicines, to understand care-seeking patterns in different settings, and they can be a valuable input for quality improvement efforts, as they are suggestive of the main competencies a health system needs to have to meet patient needs. The top five pooled RFEs in adults were headache, fever, back or low back symptoms, cough and pain general/multiple sites. The top five pooled provider diagnoses were uncomplicated hypertension, upper respiratory tract infection, type 2 diabetes, malaria and health maintenance/prevention. It is notable that the top 10 diagnoses varied between the studies substantially more than the top 10 RFEs. The diagnoses of HIV/AIDS and tuberculosis appeared only in Mash 2012 [22] from South Africa whereas malaria appeared only in Olagundoye 2016 [25] from Nigeria, in which it was the most common diagnosis. Pregnancy appeared only in Chueiri 2020 [14], though the prevalence of pregnancy in primary care depends on the demographics of the region or country and the organisation of maternal health services.
These findings show the importance of creating country-specific lists of essential diagnostics for primary care, as patients presenting with similar symptoms in different parts of the world may have substantially different disease patterns which require different diagnostic tools. Disease patterns may also vary substantially within countries, based on national disease burden. The development of multiplex or multi-disease molecular or point-of-care tests for fever-causing pathogens would be highly valuable but present a range of technical challenges [34].
A previous systematic review, mostly comprising data from HICs, had cough, back or spinal pain, unspecified abdominal condition, pharyngitis and dermatitis as the top five RFEs, with fever and headache only the sixth and seventh most common RFEs, respectively [9]. Upper respiratory tract infection and hypertension appeared in the top three provider diagnoses of the previous review, showing the overlap in diseases between LMICs and HICs, although the pathogens causing upper respiratory tract infections are likely to vary between locations.
Two of the three top provider diagnoses in this review are non-communicable diseases (NCDs), reflecting the increasing burden of NCDs in LMICs [35]. Uncomplicated hypertension causes no symptoms, as does type 2 diabetes in many cases, showing the importance of screening for diseases unrelated to patient RFEs. Diagnostics for hypertension and type 2 diabetes are simple and should be available at all PHCs, but diagnosis remains the weakest link in the cascade of care for these common NCDs in LMICs [36, 37].
The lack of psychological symptoms or disorders in the pooled lists of RFEs is striking. Despite mental and addictive disorders causing an estimated $7\%$ of the entire global burden of disease [38], none of the top 9 pooled RFEs were psychological symptoms. Mash 2012 [22] reported the top 56 RFEs in their study, none of which were psychological symptoms, and concluded that “providers appears to be failing to recognise and treat mental health problems such as depression and anxiety disorders” in the four provinces they studied in South Africa. Swain 2017 [28] suggested that the lack of psychological RFEs among the 17 reported in their study may be due to stigma and culturally influenced health-seeking behaviour among the communities in their context in India. Chueiri 2020 [14] noted the unexpectedly low rate of psychological RFEs in their national survey of Brazil when compared to data from the Global Burden of Disease Study for Brazil [39] but did not suggest a reason for this finding.
The four studies which reported RFEs and provider diagnoses using ICPC-2 were at low risk of bias in most domains. However, three of the four were at high risk of bias in the national population domain, making comparison of their results problematic in a global perspective as they cannot be said to be representative even of their own national populations.
Where possible, future studies should aim to randomly sample primary care facilities from across their country of study, or to provide countrywide data from a national primary care database, if available. Ideally, studies should use ICPC-3 as it is designed for classifying primary care encounters, it allows reporting of both RFEs and specific diagnoses and unlike in-house classification systems, it is easily comparable between studies. ICPC-3 codes should be included with RFEs and specific diagnoses, with the numbers of included patients, visits, diagnoses and RFEs clearly identified. Data for each code should be presented separately, with no combining of codes. An electronic version of ICPC-3 is available for free on the ICPC-3 website [40]. If ICPC-3 cannot be used, another validated classification system such as ICD-11 should be used instead.
Strengths of this systemic review include a comprehensive literature search, detailed data on the characteristics of each study and an analysis of the strengths and limitations of different classification systems for recording reasons for primary care visits in LMICs. Additionally, using a non-parametric approach to pooling data, involving ranking RFEs and provider diagnoses and summing the ranks, enabled comparison of highly heterogenous data.
Limitations are that the variations in the classification systems used between studies made it difficult to directly compare most of the studies with each other. Only three studies could be pooled but even within these three, problems emerged with pooling. For example, in the RFEs Mash 2012 [23] provided combined data for back symptom (L02) and low back symptom (L03) whereas Olagundoye 2016 [18] and Swain 2017 [20] presented the two categories separately. Similarly, for provider diagnoses Mash 2012 combined the gastroenteritis presumed (D73) and diarrhoea (D11) categories whereas Olagundoye 2016 presented gastroenteritis presumed (D73) separately. It is unclear why Mash 2012 combined codes in a few select instances but presented most codes separately. Of the four pooled studies, one reported data from a three-month period so may not account for seasonal variation in disease presentation [25]. Additionally, we restricted our search to articles in English and French and did not include grey literature.
Given the substantial heterogeneity among the most common reasons for primary care visits both between and within countries, a country-by-country or subnational analysis may be preferable to a global review for informing country-specific EDLs. Additionally, national or subnational information may be stored in national health information systems, rather than published in peer-reviewed journals. However, quality of routinely collected data from health facilities is a concern, while research studies might offer more reliable data. Also, findings from this study show that such information is published in peer-reviewed journals, often with data from tens of thousands of patient encounters, and systematic reviews such as this are valuable for compiling and summarising the available data.
This systematic review found 17 studies from seven LMICs, none of them LICs, reporting data on RFEs and provider diagnoses in primary care. However, data from most studies could not be pooled. Headache, fever, back or low back symptoms, cough and pain general/multiple sites were the most common pooled RFEs but the diseases causing these symptoms varied substantially between settings, showing the importance of creating location-specific lists of essential diagnostics for primary care. Future studies should aim to sample primary care facilities from across their country of study and use ICPC-3 to report both patient RFEs and provider diagnoses. In particular, studies on the reasons for primary care visits in LICs are required.
## References
1. **A vision for primary health care in the 21st century: towards universal health coverage and the Sustainable Development Goals**. (2018.0)
2. 2World Health Organization. Primary health care 2021. Available from: https://www.who.int/news-room/fact-sheets/detail/primary-health-care.
3. 3World Health Organization. The selection and use of essential in vitro diagnostics: report of the third meeting of the WHO Strategic Advisory Group of Experts on In Vitro Diagnostics, 2020 (including the third WHO model list of essential in vitro diagnostics). Geneva: World Health Organization, 2021.
4. Wilson ML, Atun R, DeStigter K, Flanigan J, Fleming KA, Horton S. **The Lancet Commission on diagnostics: advancing equitable access to diagnostics**. *The Lancet* (2019.0) **393** 2018-20. DOI: 10.1016/S0140-6736(19)31052-9
5. Kohli M, Walia K, Mazumdar S, Boehme CC, Katz Z, Pai M. **Availability of essential diagnostics in primary care in India**. *The Lancet Infectious Diseases* (2018.0) **18** 1064-5. DOI: 10.1016/S1473-3099(18)30539-5
6. 6Foundation for Innovative New Diagnostics. About the ACT-Accelerator 2021. Available from: https://www.finddx.org/covid-19/act-accelerator/.
7. 7World Health Organization. WHO publishes new Essential Diagnostics List and urges countries to prioritize investments in testing
2021 [08/07/21]. Available from: https://www.who.int/news/item/29-01-2021-who-publishes-new-essential-diagnostics-list-and-urges-countries-to-prioritize-investments-in-testing.. *WHO publishes new Essential Diagnostics List and urges countries to prioritize investments in testing* (2021.0)
8. 8Indian Council of Medical Research. National Essential Diagnostics List. New Delhi: 2019.
9. Finley CR, Chan DS, Garrison S, Korownyk C, Kolber MR, Campbell S. **What are the most common conditions in primary care? Systematic review**. *Can Fam Physician* (2018.0) **64** 832-40. PMID: 30429181
10. Adam PK, Mikashmi Jacob, Bigio Madhukar. *Most common reasons for primary care visits in low-and-middle income countries: a systematic review* (2020.0) **2020** CRD42020159469
11. 11World Bank. World Bank Country and Lending Groups 2021. Available from: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups.
12. Hoy D, Brooks P, Woolf A, Blyth F, March L, Bain C. **Assessing risk of bias in prevalence studies: modification of an existing tool and evidence of interrater agreement**. *J Clin Epidemiol* (2012.0) **65** 934-9. DOI: 10.1016/j.jclinepi.2011.11.014
13. Begum M, Khan MK, Hossain MT, Ali MY, Alam FS, Dipu SS. **Disease Pattern among Children attending Pediatric Outpatient Department in Community Based Medical College Hospital, Bangladesh**. *Mymensingh Med J* (2017.0) **26** 863-7. PMID: 29208876
14. Chueiri PS, Gonçalves MR, Hauser L, Wollmann L, Mengue SS, Roman R. **Reasons for encounter in primary health care in Brazil**. *Fam Pract* (2020.0) **37** 648-54. DOI: 10.1093/fampra/cmaa029
15. Doyle AM, McHunu L, Koole O, Mthembu S, Dlamini S, Ngwenya N. **Primary healthcare and school health service utilisation by adolescents and young adults in KwaZulu-Natal, South Africa**. *BMC Health Services Research* (2019.0) **19** 905. DOI: 10.1186/s12913-019-4559-2
16. Enato E, Sounyo A, Madadi P. **Assessment of disease profiles and drug prescribing patterns of health care facilities in Edo State, Nigeria**. *Journal of Public Health in Africa* (2012.0) **3** 101-6. DOI: 10.4081/jphia.2012.e25
17. Gupta A, Chellaiyan V, Lohiya A, Suliankatchi R, Upadhyay R, Chinnakali P. **Morbidity profile of out-patients attending a Primary Health Centre in rural Puducherry, south India**. *National Journal of Community Medicine* (2014.0) **5** 424-7
18. Gupta A, Reddy BV, Nagar MK, Chandel A, Bali S. **Portfolio of Outpatients Attending Centre for Urban Health, Madhya Pradesh, Central India**. *Health Serv Res Manag Epidemiol* (2015.0) **2**. DOI: 10.1177/2333392815598291
19. Kamarudin MF, Noh KM, Jaafar S. **Morbidity profiles at three primary care clinics in Perlis, Malaysia**. *Med J Malaysia* (2012.0) **67** 363-8. PMID: 23082442
20. Kshirsagar VD, Rajderkar S, Aldar S, Wetam S. **Socio-Demographic Profile and Morbidity Pattern of Patients in A Rural Field Practice Area of Government Medical College, Miraj, Maharashtra**. *National journal of community medicine* (2019.0) **10** 308-11
21. Kumar R, Kaur N, Pilania M. **Morbidity Pattern of Patients Attending a Primary Healthcare Facility in an Urban Slum of Chandigarh, India**. *Journal of Clinical and Diagnostic Research* (2018.0) **12** 10-3. DOI: 10.7860/JCDR/2018/31331.11297
22. Mash B, Fairall L, Adejayan O, Ikpefan O, Kumari J, Mathee S. **A Morbidity Survey of South African Primary Care**. *PLOS ONE* (2012.0) **7** e32358. DOI: 10.1371/journal.pone.0032358
23. Merali HS, Morgan JF, Uk S, Phlan S, Wang LT, Korng S. **The Lake Clinic—providing primary care to isolated floating villages on the Tonle Sap Lake, Cambodia**. *Rural Remote Health* (2014.0) **14** 2612. PMID: 24811837
24. Mohan M, Dharmendra M, kumar C, Singh S. **Morbidity Pattern In Rural Area Of Jaipur District (Rajasthan)**. *Journal of Research in Medical and Dental Science* (2014.0) **2** 79. DOI: 10.5455/jrmds.20142416
25. Olagundoye OA, van Boven K, van Weel C. **International Classification of Primary Care-2 coding of primary care data at the general out-patients’ clinic of General Hospital, Lagos, Nigeria**. *J Family Med Prim Care* (2016.0) **5** 291-7. DOI: 10.4103/2249-4863.192341
26. Prabhune A, Manoharan A. **Assessment of Healthcare Utilization in a Community-Centric Model of Primary Healthcare for Rural Populations**. *Indian Journal of Public Health Research & Development* (2017.0) **8**
27. Silva V, Pereira I, Rocha M, Caldeira A. **Morbidity in users of Family Health teams in the northeast of Minas Gerais based on the International Classification of Primary Care**. *Revista Brasileira de Epidemiologia* (2014.0) **17** 954-67. DOI: 10.1590/1809-4503201400040013
28. Swain S, Pati S, Pati S. **A chart review of morbidity patterns among adult patients attending primary care setting in urban Odisha, India: An International Classification of Primary Care experience**. *J Family Med Prim Care* (2017.0) **6** 316-22. DOI: 10.4103/2249-4863.220029
29. Torres Rde C, Marques KS, Leal Kde N, Rocha-Filho PA. **Main reasons for medical consultations in family healthcare units in the city of Recife, Brazil: a cross-sectional study**. *Sao Paulo Med J* (2015.0) **133** 367-70. DOI: 10.1590/1516-3180.2014.9490902
30. Ansah EK, Narh-Bana S, Epokor M, Akanpigbiam S, Quartey AA, Gyapong J. **Rapid testing for malaria in settings where microscopy is available and peripheral clinics where only presumptive treatment is available: a randomised controlled trial in Ghana**. *Bmj* (2010.0) **340** c930. DOI: 10.1136/bmj.c930
31. Manguin S, Foumane V, Besnard P, Fortes F, Carnevale P. **Malaria overdiagnosis and subsequent overconsumption of antimalarial drugs in Angola: Consequences and effects on human health**. *Acta Trop* (2017.0) **171** 58-63. DOI: 10.1016/j.actatropica.2017.03.022
32. Leslie T, Mikhail A, Mayan I, Anwar M, Bakhtash S, Nader M. **Overdiagnosis and mistreatment of malaria among febrile patients at primary healthcare level in Afghanistan: observational study**. *BMJ: British Medical Journal* (2012.0) **345** e4389. DOI: 10.1136/bmj.e4389
33. Oladosu OO, Oyibo WA. **Overdiagnosis and Overtreatment of Malaria in Children That Presented with Fever in Lagos, Nigeria**. *ISRN Infectious Diseases* (2013.0) **2013** 914675. DOI: 10.5402/2013/914675
34. 34Unitaid. Fever diagnostic technology landscape, 1st edition. 2018.
35. Gibbs N, Kwon J, Balen J, Dodd PJ. **Operational research to support equitable non-communicable disease policy in low-income and middle-income countries in the sustainable development era: a scoping review**. *BMJ Global Health* (2020.0) **5** e002259. DOI: 10.1136/bmjgh-2019-002259
36. Geldsetzer P, Manne-Goehler J, Marcus ME, Ebert C, Zhumadilov Z, Wesseh CS. **The state of hypertension care in 44 low-income and middle-income countries: a cross-sectional study of nationally representative individual-level data from 1·1 million adults**. *Lancet* (2019.0) **394** 652-62. DOI: 10.1016/S0140-6736(19)30955-9
37. Manne-Goehler J, Geldsetzer P, Agoudavi K, Andall-Brereton G, Aryal KK, Bicaba BW. **Health system performance for people with diabetes in 28 low- and middle-income countries: A cross-sectional study of nationally representative surveys**. *PLOS Medicine* (2019.0) **16** e1002751. DOI: 10.1371/journal.pmed.1002751
38. Rehm J, Shield KD. **Global Burden of Disease and the Impact of Mental and Addictive Disorders**. *Curr Psychiatry Rep* (2019.0) **21** 10. DOI: 10.1007/s11920-019-0997-0
39. Marinho F, de Azeredo Passos VM, Carvalho Malta D, Barboza França E, Abreu DMX, Araújo VEM. **Burden of disease in Brazil, 1990–2016: a systematic subnational analysis for the Global Burden of Disease Study 2016**. *The Lancet* (2018.0) **392** 760-75. DOI: 10.1016/S0140-6736(18)31221-2
40. 40ICPC-3 Consortium. ICPC-3 Browser 2021 [08/07/21]. Available from: https://browser.icpc-3.info/.
|
---
title: Effects of social determinants of health on obesity among urban women of reproductive
age
authors:
- Dickson A. Amugsi
- Zacharie T. Dimbuene
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022252
doi: 10.1371/journal.pgph.0001442
license: CC BY 4.0
---
# Effects of social determinants of health on obesity among urban women of reproductive age
## Abstract
Obesity is a major global public health problem. It is spreading very fast in low- and middle-income countries and has reached world record levels in some of them. In Ghana, it has increased by over $65\%$ among urban women in the past three decades. This study investigated the effects of social determinants of health on obesity among women in urban Ghana. The study analyzed the Ghana demographic and health survey data. These are nationally representative data collective every five years across low- and middle-income countries. A total of 1,204 urban women were included in the analysis. Body mass index was the outcome variable of interest. We used logistic regression to model the effects of the various social determinants of health on obesity. The results showed that $40\%$ ($95\%$ confidence interval (CI) = 25.4, 57.0) and $36.7\%$ ($95\%$ CI = 25.6, 49.3) of women who had higher education and those whose partners had higher education suffered from obesity, respectively. Women living in rich households had a five times higher prevalence of obesity than those in poor households ($28.8\%$ vs $5.7\%$). Further, $33.4\%$ ($95\%$ CI = 18.5, 19.3) of women who occupied managerial positions were obese. The results from the multivariable logistic regression analysis suggested that compared to women in poor households, those in rich households were 3.4 times ($95\%$ CI = 1.31, 8.97) more likely to suffer from obesity. Women whose main occupation was agriculture were $81\%$ (aOR = 0.19; $95\%$ CI = 0.034, 0.99) less likely to suffer from obesity compared to those with no occupation. The results suggest that the various social determinants of health (SDH) included in the analysis significantly influence obesity outcomes. Women and partner education levels, occupying a managerial position, and living in rich households increase the risk of obesity. Interventions to address the rising obesity in urban Ghana should have specific packages targeted at these sub-groups.
## Introduction
Obesity (body mass index greater than or equal to 30kg/m2), an indicator of excessive fat accumulation in the body tissues, is a growing global problem of public health concern [1, 2]. Obesity is an illness that requires deliberate policies and interventions to address it to prevent premature deaths among the world’s population [2]. The available data suggest that since 1975, the global burden of obesity has more than tripled [3]. The data also suggest that no country has experienced a decline in obesity prevalence over this period. Therefore, it is unsurprising that in 2000, the World Health Organisation (WHO) declared obesity a pandemic, which informed the development of a global action plan to combat it [4]. Further evidence suggests that it is more severe among women of reproductive age than men and in urban settings than rural [3].
Obesity is spreading very fast in developing countries and has reached world record levels in some of them [5]. In Africa, the obesity prevalence among women of reproductive age ranged from 6.5 to $50.7\%$ [6]. The problem is most severe among urban women relative to rural women [7]. Similarly, a recent analysis [2] using urban data from 24 African countries, spanning almost three decades, showed that all the countries included in the study experienced a substantial increase in obesity among women of reproductive age. Indeed, a country such as Ghana recorded $65\%$ increase in obesity during the study period [2]. Other studies also observed alarming trends in obesity among women of reproductive age in the African region [8, 9]. There is a need for deliberate policies by governments to address the problem. However, it appears there is lack of policy prioritisation regarding obesity in Africa. This may be the case because many policymakers are yet to appreciate the seriousness of the problem. Therefore, it is critical to provide policymakers and other stakeholders with evidence of the seriousness of the problem and its potential drivers.
The consequences of obesity on population health are enormous. It is associated with an increased risk of cardiovascular diseases (mainly heart disease and stroke), diabetes, musculoskeletal disorders, and some cancers, including but not limited to endometrial, breast, ovarian, prostate, liver, gallbladder, kidney, and colon [10]. An analysis of recent data suggests that obesity is linked to 13 different cancers [11]. Also, the effect of obesity on the incidence of gestational diabetes, pre-eclampsia, high risk of miscarriage/stillbirth, and congenital anomalies is well documented [12–14]. Pregnant women who are obese have higher chances of heart disease, hypertension, and type 2 diabetes [15]. The link between women obesity and future obesity in their children has also been observed [16].
Similarly, the extant literature suggests a strong link between obesity and high mortality rate. An analysis of global data shows that 2.8 million people die annually due to obesity [17], while 35.8 million disability-adjusted life years (DALYs) are attributed to elevated body mass index (BMI) [15]. The adverse effects of obesity on population health, as outlined above, calls for a more detailed analysis to understand the determinants of the problem among women of reproductive age.
Previous research suggests that social determinants of health are key drivers of the rising obesity in urban Africa [18, 19]. According to the WHO commission, "Social determinants of health (SDH) are the conditions in which people are born, grow, live, work and age, and the wider set of forces and systems shaping the conditions of daily life" [3]. The SDH includes socioeconomic status, education, literacy, neighborhood and physical environment, employment, food consumption, place of residence and social support networks, and access to health care. The literature on the effects of socioeconomic status, education, and employment status, among others, abound [20]. However, the impact varies between low- and high-income countries. While obesity is a significant issue among the rich in low-income countries, it is high among the poor in high-income countries [21].
Furthermore, obesity was more prevalent among highly educated women than those without formal education in low-income countries [22]. Women who are formally employed have increased odds of obesity compared to those who are not employed, which vary by educational attainment [23]. This could be attributed to increased income and behavioral changes related to the time allocated to physical activity. In addition, the existing literature suggests a link between urbanization, nutritional transition, and the high prevalence of obesity among women [24]. There have been increasing changes in food consumption, whereby urban dwellers now consume more processed and fast foods and sugary beverages, with consequential adverse effects on obesity outcomes [24, 25].
Although our previous analysis suggests that obesity among women of reproductive age has been rising rapidly in urban Ghana in the past decades [2], there is limited knowledge on the key SDH driving the surge. The present study intends to fill this gap by undertaking a comprehensive analysis of obesity prevalence by various SDH, and the effects of the SDH on the rising obesity in urban Ghana, drawing on existing data sources. This type of analysis is needed to provide detailed and nuanced findings for better decision-making to inform policy and program interventions to address the obesity challenges in urban Ghana. The present study examined the effects of SDH on obesity among women of reproductive age in urban Ghana using existing data.
## Data sources and study design
The study is a secondary analysis of the 2014 Ghana Demographic and Health Surveys (DHS) data [26]. These are nationally representative data collected every five years in Ghana and other low- and middle-income countries (LMICs) [26]. The analysis in the current study is restricted to the urban sub-sample. We downloaded the data from the DHS program website and assessed their completeness regarding the women’s anthropometric data. We excluded cases that were missing anthropometric data.
The DHS utilizes a complex sampling design, whereby the sample selection involves multiple stages. In the first stage, clusters are selected from a master sampling frame constructed from the Ghana National Population and Housing Census data. The clusters are selected using systematic sampling with probability proportional to size. After selecting the clusters, a household listing operation is undertaken in the selected clusters to get a sampling frame to select households to take part in the study. A systematic sampling of households is then undertaken. This stage of selection is aimed to ensure an adequate sample size to estimate the indicators of interest with acceptable precision. Our analyses are restricted to adult non-pregnant women aged 15–49 years. We focus on non-pregnant women because pregnancy is usually associated with weight gain. Therefore, including them in the analysis may present a misleading picture of the obesity situation. The eligible sample size used in the present analysis was 1,204 urban women of reproductive age.
## Ethics statement
The DHS study was undertaken based on high ethical standards. The data collectors were trained to respect the rights of study participants. The participants were made to understand that they have the right to decide whether they wanted to participate in the study or not, as well as to abstain or withdraw their participation at any time without reprisal. The potential risks and benefits associated with the study and steps taken to mitigate potential risks were adequately explained to study participants. A written informed consent was obtained from the parent/guardian of each participant under 18 years of age. A government-recognized institutional review committee granted ethical approval for the conduct of the study. Further ethical clearance was granted by the Institutional Review Board of ICF International, USA before the survey was conducted. The authors obtained permission from DHS Program for the use of the data. The authors did not seek further ethical clearance as the DHS data are highly anonymised.
## Outcome variable
The women’s body mass index (BMI) was the outcome variable of interest in this analysis. Well-trained field technicians collected participants anthropometric data (Height and weight) using recommended techniques [27, 28]. The weight of study participants was measured using electronic Seca scales, while their height was measured using boards produced by Shorr Productions. These anthropometric data were used to estimate the BMI of the study participants. The BMI was calculated by dividing weight in kilograms by the square of height in meters. Based on the WHO guidelines [27], obesity was classified as BMI≥30.0 kg/m2. The prevalence of obesity and its associated factors were estimated. In the DHS dataset, place of residence is classified into rural and urban. However, this study focuses only on urban settings. Our previous analysis [2] informed the choice of the study setting. Also, a place of resident is a critical SDH, therefore, restricting the analysis to the urban settings will help to contextualize the study findings.
## Explanatory variables
The SDH factors used in this analysis included women’s education level, employment status, fruit and vegetable consumption, partner occupation, partner education level, household wealth index, number of trips in the last 12 months, frequency of listening to radio, and sex of household head. Potential control variables included women’s age, height, and breastfeeding status. The household wealth index, a composite measure of the household cumulative living standard, was constructed using principal component analysis (PCA) [29]. The variables used in the PCA included but not limited to household ownership of televisions and bicycles, materials used for housing construction, and types of water access and sanitation facilities. The constructed wealth index is then divided into quintiles (poorest, poor, middle, rich & richest). To preserve the sample size for the analysis, we recoded the wealth index into poor (poorest+poor), middle, and rich (rich+richest). The potential SDH outlined above were identified based on the SDH literature. These variables were subjected to bivariate analysis to establish those associated significantly with women’s obesity. Statistical significant variables were included in the multivariable analysis. In addition, we included variables that were not statistically significant but considered critical as far as obesity was concerned.
## Analytical strategy
The data analysis was performed using STATA V.14. The analyses involved three stages. In the first analysis, we assessed the characteristics of the sample using frequencies and means. In the second stage, we estimated the prevalence of obesity and their associated $95\%$ confidence intervals (CIs) by various SDH. In the third analysis, we examined the effects of SDH on maternal obesity using multivariable logistic regression. We entered SDH variables in the first model (model 1), and adjusted for women’s age, height, and breastfeeding status in the second model (model 2). Adjusted odds ratio (aOR) and CIs are reported for this analysis. We accounted for the complex survey design (CSD) effect in all analyses using the svyset and svy procedures in STATA.
## Background characteristics of the study sample
Table 1 presents the characteristics of the study sample. The average age of study participants was 31 years (30.85±6.29), while the average height was 1.59m (1.59±0.06). Majority ($59\%$) of the women had secondary school education, while only $6\%$ had tertiary/higher education. This was similar to the partner education level where $59\%$ had secondary education. Similarly, $52\%$ of the women had clerical/sales/services as their main occupation. However, the dominant occupation of their partners was manual work ($45\%$). The results also suggested that over $40\%$ indicated they consumed fruits ($42\%$) and vegetables ($45\%$) for 4 or more days in the past 7 days.
**Table 1**
| Variables | Mean±SD or % |
| --- | --- |
| Maternal education | |
| No education | 16.78 |
| Primary | 17.43 |
| Secondary | 59.48 |
| Higher | 6.31 |
| Maternal occupation | |
| No occupation | 21.63 |
| Professional/tech/managerial | 5.99 |
| Clerical/sales/services | 51.85 |
| Agricultural worker | 6.2 |
| Manual worker | 14.32 |
| Days ate fruits in the last 7 days | |
| | 18.25 |
| 1–3 days | 39.8 |
| 4+ days | 41.95 |
| Days ate vegetables in the last 7 days | |
| | 16.76 |
| 1–3 days | 38.58 |
| 4+ days | 44.67 |
| Partner occupation | |
| Manual worker | 45.41 |
| Professional/tech/managerial | 17.25 |
| Clerical/sales/services | 17.14 |
| Agricultural worker | 11.33 |
| Sex of household head | |
| Male | 68.23 |
| Female | 31.77 |
| Partner education level | |
| No education | 12 |
| Primary | 6.12 |
| Secondary | 58.92 |
| Higher | 14.65 |
| Maternal mean age | 30.85±6.29 |
| Maternal mean height | 1.59±0.06 |
Table 2 shows the prevalence of obesity by various SDH. The results showed that $40\%$ ($95\%$ confidence interval (CI) = 25.4, 57.0) of women who had higher education were obese, compared to only $14.4\%$ ($95\%$ CI = 8.6, 23.0) of those with no formal education. Also, $36.7\%$ ($95\%$ CI = 25.6, 49.3) of women whose partners had higher education were obese. The prevalence of obesity was $28.8\%$ ($95\%$ CI = 24.1, 34.0) among women living in rich households compared to $5.7\%$ ($95\%$ CI = 2.4, 13.0) of those in poor households. Further, $33.4\%$ ($95\%$ CI = 18.5, 19.3) of women who occupied managerial positions were obese. Conversely, only $1.8\%$ ($95\%$ CI = 0.4, 7.8) of those whose occupation was agriculture suffered from obesity. Women whose partners held managerial positions suffered more from obesity than those whose partners were agricultural workers (33.6 vs. $7.2\%$). All the above results were statistically significant.
**Table 2**
| Unnamed: 0 | Obese | Obese.1 | Unnamed: 3 |
| --- | --- | --- | --- |
| Variables | % | 95% CI | P-value |
| Women’s education | | | |
| No education | 14.4 | 8.6, 23.0 | 0.026 |
| Primary | 25.4 | 17.2,35.7 | |
| Secondary | 23.5 | 18.6, 29.3 | |
| Higher | 40.2 | 25.4, 57.0 | |
| Household wealth index (HWI) | | | |
| Poor | 5.7 | 2.4, 13.0 | 0.001 |
| Middle | 13.0 | 8.5, 19.3 | |
| Rich | 28.8 | 24.1, 34.0 | |
| Women’s occupation | | | |
| No occupation | 14.5 | 9.1, 22.2 | 0.001 |
| Managerial/technical | 33.4 | 18.5, 52.5 | |
| Clerical/sales/services | 27.9 | 23.2, 33.3 | |
| Agricultural worker | 1.8 | 0.4, 7.8 | |
| Manual worker | 25.2 | 16.2, 37.1 | |
| Number of days ate fruits | | | |
| None (zero days) | 13.6 | 7.7, 22.7 | 0.028 |
| ate fruits 1–3 days | 23.2 | 17.9, 29.6 | |
| Ate fruits 4+ days | 27.7 | 21.9, 34.4 | |
| Number of days ate vegetables | | | |
| None (zero days) | 21.2 | 13.2, 32.3 | 0.76 |
| Ate vegetables 1–3 days | 22.4 | 16.0, 30.4 | |
| Ate vegetables 4+ days | 25.0 | 19.7, 31.1 | |
| Partner’s occupation | | | |
| Manual worker | 25.6 | 20.4, 31.6 | 0.001 |
| Managerial/technical | 33.6 | 24.9, 43.6 | |
| Clerical/sales/services | 24.7 | 16.6, 34.9 | |
| Agricultural worker | 7.2 | 3.3, 15.0 | |
| Sex of household head | | | |
| Male | 24.2 | 19.7, 29.4 | 0.57 |
| Female | 21.4 | 14.6, 30.3 | |
| Partner education level | | | |
| No education | 16.7 | 10.3, 25.8 | 0.001 |
| Primary | 8.2 | 3.6, 17.6 | |
| Secondary | 24.9 | 20.3, 30.1 | |
| Higher | 36.7 | 25.6, 49.3 | |
| Women’s literacy | | | |
| Cannot read at all/no information/virtually impaired | 18.6 | 14.3, 24.0 | 0.072 |
| Able to read only part of the sentence | 32.0 | 17.6, 50.9 | |
| Able to write a whole sentence | 27.1 | 21.3, 33.9 | |
| Frequency listen radio | | | |
| Not at all | 11.6 | 6.5, 19.9 | 0.027 |
| Less than once a week | 26.1 | 19.1, 34.5 | |
| At least once a week | 25.5 | 20.2, 31.7 | |
| Is breastfeeding | | | |
| No | 27.6 | 22.3, 33.6 | 0.031 |
| Yes | 19.7 | 15.2, 25.1 | |
Table 3 presents the effects of SDH on obesity among women. The results showed that compared to women in the poor households, those in the rich households were 3.4 times ($95\%$ CI = 1.31, 8.97) more likely to suffer from obesity. Women whose main occupation was agriculture were $81\%$ (aOR = 0.19; $95\%$ CI = 0.034, 0.99) less likely to suffer from obesity than those with no occupation. Similarly, women whose partners’ occupation was agriculture worker had $71\%$ (aOR = 0.39; $95\%$ CI = 0.15, 0.99) reduced odds of becoming obese compared to those whose partners were manual workers. Partners attaining primary education had a moderate protective effect on obesity among women (aOR = 0.34; $95\%$CI = 0.11, 1.08). Compared to women who did not listen to radio at all, those who listened to radio less than once a week or at least once a week had increased odds of obesity (aOR = 2.64, $95\%$ CI = 1.26, 5.54). A biological variable such as women’s age was associated with increased odds of obesity (aOR = 1.09; $95\%$ CI = 1.06, 1.13).
**Table 3**
| VARIABLES | Model 1 | Model 2 |
| --- | --- | --- |
| Women’s education | | |
| No Education (ref) | | |
| Primary | 1.624 | 1.776 |
| | (0.727–3.627) | (0.767–4.113) |
| Secondary | 0.937 | 0.914 |
| | (0.423–2.076) | (0.395–2.113) |
| Higher | 1.481 | 1.565 |
| | (0.408–5.382) | (0.418–5.861) |
| Household wealth index | | |
| Poor (ref) | | |
| Middle | 1.656 | 1.853 |
| | (0.637–4.306) | (0.698–4.920) |
| Rich | 3.262** | 3.421** |
| | (1.219–8.729) | (1.305–8.966) |
| Women’s occupation | | |
| No occupation (ref) | | |
| Managerial/technical | 1.371 | 1.096 |
| | (0.429–4.380) | (0.334–3.601) |
| Clerical/sales/services | 2.088** | 1.626 |
| | (1.119–3.895) | (0.841–3.145) |
| Agricultural worker | 0.267 | 0.185** |
| | (0.055–1.298) | (0.034–0.989) |
| Manual worker | 1.870 | 1.354 |
| | (0.833–4.200) | (0.595–3.082) |
| Partner’s occupation | | |
| Manual (ref) | | |
| Managerial/technical | 0.998 | 0.985 |
| | (0.474–2.102) | (0.449–2.160) |
| Clerical/sales/services | 0.821 | 0.772 |
| | (0.449–1.501) | (0.403–1.479) |
| Agricultural worker | 0.497 | 0.386** |
| | (0.197–1.250) | (0.150–0.992) |
| Days ate fruits in the last 7 days | | |
| None (ref) | | |
| 1–3 days | 1.585 | 1.667 |
| | (0.777–3.233) | (0.782–3.554) |
| 4+ days | 2.047** | 1.846* |
| | (1.097–3.820) | (0.935–3.647) |
| Days ate vegetables in the last 7 days | | |
| None (ref) | | |
| 1–3 days | 1.030 | 0.985 |
| | (0.489–2.168) | (0.447–2.170) |
| 4+ days | 1.015 | 0.958 |
| | (0.534–1.928) | (0.491–1.871) |
| Partner’s education level | | |
| No education (ref) | | |
| Primary | 0.296** | 0.339* |
| | (0.093–0.941) | (0.107–1.075) |
| Secondary | 0.806 | 0.936 |
| | (0.378–1.716) | (0.425–2.065) |
| Higher | 1.019 | 1.234 |
| | (0.327–3.170) | (0.381–3.997) |
| Number of trips in last 12 months (continuous) | 1.015 | 1.015 |
| | (0.987–1.044) | (0.990–1.040) |
| Frequency listening radio | | |
| Not at all (ref) | | |
| Less than once a week | 2.747*** | 2.640** |
| | (1.345–5.609) | (1.258–5.540) |
| At least once a week | 2.268** | 2.295** |
| | (1.170–4.397) | (1.190–4.426) |
| Women’s age (in single years) (continuous) | | 1.090*** |
| | | (1.055–1.126) |
| Women’s height (in meters) | | 0.092 |
| | | (0.005–1.806) |
| Sex of household head | | |
| Male (ref) | | |
| Female | | 1.059 |
| | | (0.605–1.852) |
| Breastfeeding status (Ref.: No) | | |
| No (ref) | | |
| Yes | | 0.758 |
| | | (0.486–1.183) |
## Discussion
The extant literature shows that social determinants of health (SDH), as captured in the WHO commission on SDH framework, play a critical role in determining obesity and health outcomes [3, 20–23]. In this study, we presented a comprehensive analysis of obesity prevalence by various SDH and their effects on the rising obesity in urban Ghana, using existing data. The results suggest that obesity was three times higher among women with high education relative to those without education ($40\%$ vs $14.4\%$). This is also the case with partner education, where women whose partners were highly educated suffer more from obesity. The findings reflect the results of previous research. Several studies have documented the relationship between the level of education and the high prevalence of obesity [22, 23, 30, 31]. High education appears to place women at an elevated risk of obesity. The negative effect of education on obesity may be because this subgroup is likely to have a higher income, with a consequential impact on lifestyle changes. For example, the extant literature reveals that high-income earners in LMICs practice sedentary lifestyles and consume unhealthy diets [24, 30]. It is essential to point out that while education is a risk factor for obesity in LMICs, it is indeed a protective factor in high-come settings [21]. Thus, the highly educated in high-income countries are less prone to obesity than the less educated.
The analysis of the women’s occupation data reveals that women holding managerial positions had a substantially higher prevalence of obesity than other occupations. The high prevalence may be because of prolonged sitting hours, reduced walkability, increased intake of convenient, highly processed foods, and greater reliance on cars as the primary means of transport [30, 32]. For example, long hours at the office can make it challenging to squeeze in time for exercise [33]. The finding in the present study is consistent with the extant literature. A study undertaken among Korean women showed that obesity prevalence was significantly higher among managers than non-managers [32].
Further, our results suggest that women who engage in agriculture suffer less from obesity compared to the other occupations. This could be attributed to the high physical activity associated with this occupation. This finding is consistent with the multivariable logistic regression results, which revealed that women who engaged in agriculture had an $81\%$ reduced odds of becoming obese than those in other occupations. The low risk associated with agriculture may be linked to agricultural workers consuming their produce rather than relying on high-energy food such as fast food and sugar-sweetened beverages [32].
Similarly, the findings show that the prevalence of obesity among women living in rich households is over five times higher than those in poor households ($28.8\%$ vs $5.7\%$). It implies that wealth is a risk factor for women’s obesity, while poverty is a protective factor. This finding has been corroborated by the multivariable analysis, where women in rich households were over three times more likely to suffer from obesity than those in poor households. The endemicity of obesity in rich households may be due to access to high-energy food and a lack of physical activity. This finding is consistent with the existing literature, whereby several studies have shown that women in rich households suffer more from obesity than those in poor households [34–37]. However, a previous comparative analysis showed some differences between low-and high incomes countries. Indeed, while obesity is a major public health issue among the rich in low-income countries, it is the case among the poor in high-income countries [21]. It may mean that the rich in the high-income countries are more concerned about their weight and therefore live healthy lifestyles.
## Strength and limitations of the study
The study used robust nationally representative data, making it possible to generalise the findings to all women of reproductive age living in urban settings in Ghana. The large sample size enhances the precision of the estimates and the robustness of the associations observed in the analysis. Further, the outcome variable was objectively measured, reducing the chances of misclassification. A noteworthy limitation is the cross-sectional nature of the data, which does not make it possible to establish any causal relationship between the various SDH and obesity. Therefore, the results are interpreted in the context of associations between the independent and dependent variables. The data did not allow the inclusion of all the broader social determinants of health as defined by the WHO commission on SDH framework. However, the SDH framework did not envision that a single study could address all the factors captured in the framework.
## Conclusions
The study examined the effects of the various social determinants of health on obesity among women of reproductive age in urban Ghana. The results suggest that the different SDH substantially influence women’s obesity. Higher education increases the risk of obesity among women. Women occupying managerial positions are also likely to suffer from obesity. Similarly, living in rich households increases the risk of obesity among women. Interventions to address the rising obesity in urban Ghana should have specific packages targeted at these sub-groups. These may include promoting physical activity, and healthy eating behaviours, which the extant evidence suggests are hardly practised by these sub-groups.
## References
1. Yaya S, Ghose B. **Trend in overweight and obesity among women of reproductive age in Uganda: 1995–2016.**. *Obes. Sci. Pract.* (2019.0) **5** 312-323. DOI: 10.1002/osp4.351
2. Amugsi DA, Dimbuene ZT, Mberu B, Muthuri S, Ezeh A. **Prevalence and time trends in overweight and obesity among urban women: an analysis of demographic and health surveys data from 24 African countries, 1991–2014.**. *BMJ Open.* (2017.0) **7** e017344. DOI: 10.1136/bmjopen-2017-017344
3. 3WHO [internet]. Social Determinants of Health; [cited Jun 20]. Available from: http://www.who.int/social_determinants/sdh_definition/en/
4. Adeboye B, Bermano G, Rolland C. **Obesity and its health impact in Africa: a systematic review.**. *Cardiovasc. J. Afr.* (2012.0) **23** 512-521. DOI: 10.5830/CVJA-2012-040
5. Goryakin Y, Lobstein T, James WP, Suhrcke M. **The impact of economic, political and social globalization on overweight and obesity in the 56 low and middle income countries.**. *Soc. Sci. Med.* (2015.0) **133** 67-76. DOI: 10.1016/j.socscimed.2015.03.030
6. Onubi OJ, Marais D, Aucott L, Okonofua F, Poobalan AS. **Maternal obesity in Africa: a systematic review and meta-analysis.**. (2016.0) **38** e218-e231. DOI: 10.1093/pubmed/fdv138
7. Chigbu CO, Aja LO. **Obesity in pregnancy in southeast Nigeria.**. (2011.0) **1** 135-40. PMID: 23209967
8. Asoseg KA, Adebanji AO, Abdul IW. **Spatial analysis of the prevalence of obesity and overweight among women in Ghana.**. *BMJ Open* (2021.0) **11** e041659. DOI: 10.1136/bmjopen-2020-041659
9. **Trends in obesity and diabetes across Africa from 1980 to 2014: an analysis of pooled population-based studies.**. *Int J Epidemiol.* (2017.0) **46** 1421-1432. DOI: 10.1093/ije/dyx078
10. 10WHO [internet]. Obesity and overweight; c2021 [cited May 2]. available from: https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight
11. 11PubMed Health[internet]. Excess body fat now linked to 13 different types of cancers: c2017[cited 30 May]. Available from: https://www.ncbi.nlm.nih.gov/pubmedhealth/behindtheheadlines/news/2016-08-25-excess-body-fat-now-linked-to-13-different-types-of-cancer/
12. Leddy MA, Power ML, Schulkin J. **The impact of maternal obesity on maternal and fetal health.**. *Rev Obstet Gynecol* (2008.0) **1** 170-8. PMID: 19173021
13. Lynch CM, Sexton DJ, Hession M, Morrison JJ. **Obesity and mode of delivery in primigravid and multigravid women.**. *Am J Perinatol* (2008.0) **25** 163-7. DOI: 10.1055/s-2008-1061496
14. Arianas EA, Rankin KM, Norr KF, White-Traut RC. **Maternal weight status and responsiveness to preterm infant behavioral cues during feeding.**. *BMC Pregnancy Childbirth.* (2017.0) **17**. DOI: 10.1186/s12884-017-1298-4
15. Mokdad AH, Ford ES, Bowman BA, Dietz WH, Vinicor F, Bales VS. **Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001.**. *JAMA* (2003.0) **289** 76-9. DOI: 10.1001/jama.289.1.76
16. Godfrey KM, Reynolds RM, Prescott SL, Nyirenda M, Jaddoe VWV, Erickson JG. **Influence of maternal obesity on the long-term health of offspring**. (2017.0) **5** 53-64. DOI: 10.1016/S2213-8587(16)30107-3
17. 17WHO [internet]. Obesity. Situation and trends: Global Health Observatory (GHO) data. c2016[cited May 13]: http://www.who.int/gho/ncd/risk_factors/obesity_text/en/
18. Lee A, Cardel M, Donahoo WT, Feingold KR, Anawalt B, Boyce A. *Social and Environmental Factors Influencing Obesity.* (2000.0)
19. Jaacks LM, Slining MM, Popkin BM. **Recent underweight and overweight trends by rural-urban residence among women in low- and middle-income countries**. *J Nutr* (2015.0) **145** 352-7. DOI: 10.3945/jn.114.203562
20. Goryakin Yevgeniy, Suhrcke Marc. **Economic development, urbanization, technological change and overweight: What do we learn from 244 Demographic and Health Surveys?**. *Economics & Human Biology.* (2014.0) **14** 109-127. DOI: 10.1016/j.ehb.2013.11.003
21. Templin T, Cravo Oliveira Hashiguchi T, Thomson B, Dieleman J, Bendavid E. **The overweight and obesity transition from the wealthy to the poor in low- and middle-income countries: A survey of household data from 103 countries.**. *PLoS Med.* (2019.0) **16**
22. Barlow P.. **The Effect of Schooling on Women’s Overweight and Obesity: A Natural Experiment in Nigeria.**. (2021.0) **58** 685-710. DOI: 10.1215/00703370-8990202
23. Oddo VM, Bleich SN, Pollack KM, Surkan PJ, Mueller NT, Jones-Smith JC. **The weight of work: the association between maternal employment and overweight in low- and middle-income countries.**. (2017.0) **66**. DOI: 10.1186/s12966-017-0522-y
24. Dinsa GD, Goryakin Y, Fumagalli E, Suhrcke M. **Obesity and socioeconomic status in developing countries: a systematic review.**. *Obes. Rev.* (2012.0) **13** 1067-79. DOI: 10.1111/j.1467-789X.2012.01017.x
25. Drewnowski A, Specter SE. **Poverty and obesity: the role of energy density and energy costs**. *Am J Clin Nutr* (2004.0) **79** 6-16. DOI: 10.1093/ajcn/79.1.6
26. 26The DHS Program[internet]. Demographic and Health Surveys. c2017 [cited Jun 8]. Available from: https://dhsprogram.com/data/dataset_admin/download-datasets.cfm
27. 27WHO [internet]. BMI classification. c2006[cited Aug 4]. Available from: http://apps.who.int/bmi/index.jsp?introPage=intro_3.html
28. Yadav K, Krishnan A. **Changing patterns of diet, physical activity and obesity among urban, rural and slum populations in north India.**. *Obes. Rev* (2008.0) **9** 400-8. DOI: 10.1111/j.1467-789X.2008.00505.x
29. 29The DHS Program[internet]. DHS Methodology. c2017[cited Jul 10]. Available from: http://dhsprogram.com/What-We-Do/Survey-Types/DHS-Methodology.cfm
30. Gissing SC, Pradeilles R, Osei-Kwasi HA, Cohen E, Holdsworth M. **Drivers of dietary behaviours in women living in urban Africa: a systematic mapping review.**. *Public Health Nutr* (2017.0) **20** 2104-2113. DOI: 10.1017/S1368980017000970
31. Wardle Jane, Waller Jo, Martin J. **Jarvis, Sex Differences in the Association of Socioeconomic Status With Obesity.**. (2002.0) **92**
32. Eum MJ, Jung HS. **Association between Occupational Characteristics and Overweight and Obesity among Working Korean Women: The 2010–2015 Korea National Health and Nutrition Examination Survey.**. *Int J Environ Res Public Health* (2020.0) **17** 1585. DOI: 10.3390/ijerph17051585
33. Neupane S, Prakash KC, Doku DT. **Overweight and obesity among women: analysis of demographic and health survey data from 32 Sub-Saharan African Countries.**. *BMC Public Health* (2016.0) **16** 30. DOI: 10.1186/s12889-016-2698-5
34. Bishwajit G.. **Household wealth status and overweight and obesity among adult women in Bangladesh and Nepal.**. *Obes Sci Pract* (2017.0) **3** 185-192. DOI: 10.1002/osp4.103
35. Zubery D, Kimiywe J, Martin HD. **Prevalence of Overweight and Obesity, and Its Associated Factors Among Health-care Workers, Teachers, and Bankers in Arusha City, Tanzania.**. *Diabetes Metab. Syndr. Obes* (2021.0) **14** 455-465. DOI: 10.2147/DMSO.S283595
36. Jahnke SA, Kaipust C, Jitnarin N, Hollerback BS, Koeppel MDH, Haddock CK. **Prevalence and predictors of obesity among women in the fire service.**. (2022.0) **79** 289-294. DOI: 10.1136/oemed-2021-107590
37. Sharma AM, Padwal R. **Obesity is a sign—over-eating is a symptom: an aetiological framework for the assessment and management of obesity.**. *Obes Rev.* (2010.0) **11** 362-70. DOI: 10.1111/j.1467-789X.2009.00689.x
|
---
title: Influence of joint volume on range of motion after arthroscopic rotator cuff
repair
authors:
- Jung-Han Kim
- Young-Kyoung Min
- Dae-Yoo Kim
- Jun-Ho Park
- Young-Chae Seo
- Won-Seok Seo
journal: BMC Musculoskeletal Disorders
year: 2023
pmcid: PMC10022253
doi: 10.1186/s12891-023-06306-z
license: CC BY 4.0
---
# Influence of joint volume on range of motion after arthroscopic rotator cuff repair
## Abstract
### Background
Capsular contracture is a well-known etiology in the primary stiff shoulder; thus capsular contracture and resultant decreased joint volume could lead to postoperative stiffness, which is a commonly reported morbidity after arthroscopic rotator cuff repair (ARCR). The purpose of this study was [1] to quantify the joint volume (total joint volume and each quadrant compartmental volume) using computed tomography arthrography (CTA) and [2] to demonstrate the relationship between joint volume and postoperative range of motion (ROM) after ARCR.
### Materials and methods
Eighty-three patients (60 ± 5.11 years, men = 26, women = 57) who had undergone ARCR between January 2015 to December 2020 due to small to medium full-thickness tear and followed by CTA 6 months postoperatively were retrospectively reviewed. An image reconstruction program (3D Slicer, version 4.11.2 software) was used to calculate the joint volume (total joint volume and quadrant compartment joint volumes; anteroinferior, anterosuperior, posterosuperior and posteroinferior). For shoulder ROM, data including scaption (Sc), external rotation on side (ERs), external rotation at 90° (ER90), and internal rotation on back (IRb) were collected 6 months postoperatively. An evaluation of the correlation between joint volume and each shoulder motion was performed.
### Results
There were moderate correlations between the total joint volume and each motion (Sc: Pearson coefficient, 0.32, $$p \leq 0.0047$$; ERs: Pearson coefficient, 0.24, $$p \leq 0.0296$$; ER90: Pearson coefficient, 0.33, $$p \leq 0.0023$$; IRb: Pearson coefficient, 0.23, $$p \leq 0.0336$$). Among the quadrant compartments, the anteroinferior (Sc: Pearson coefficient, 0.26, $$p \leq 0.0199$$; ERs: Pearson coefficient, 0.23, $$p \leq 0.0336$$; ER90: Pearson coefficient, 0.25, $$p \leq 0.0246$$; IRb: Pearson coefficient, 0.26, $$p \leq 0.0168$$) and posterosuperior (Sc: Pearson coefficient, 0.24, $$p \leq 0.029$$; ER90: Pearson coefficient, 0.29, $$p \leq 0.008$$; IRb: Pearson coefficient, 0.22, $$p \leq 0.0491$$) and posteroinferior (Sc: Pearson coefficient, 0.30, $$p \leq 0.0064$$; ER90: Pearson coefficient, 0.29, $$p \leq 0.0072$$) showed moderate correlations with each shoulder motion.
### Conclusion
Total joint volume, anteroinferior compartment joint volume, posterosuperior compartment joint volume and posteroinferior compartment joint volume were related to postoperative ROM after ARCR. Perioperative methods to increase the joint volume, especially the anteroinferior, posterosuperior and posteroinferior parts of the capsule may prevent postoperative stiffness after ARCR.
### Level of Evidence
Level III; Retrospective Case-Control Study.
## Introduction
A rotator cuff tear is a common cause of shoulder pain [1, 2]. For the treatment of such tears, arthroscopic rotator cuff repair (ARCR) generally yields successful outcomes due to the development of surgical devices and techniques [3–5]. As the number of ARCRs performed has increased, postoperative complications have also increased [5]. Among these complications, postoperative stiffness, which is a significant factor affecting outcomes, may occur, and its incidence varies from 2.3 to $40\%$ [1, 2, 6].
Postoperative shoulder stiffness is most likely caused by the surgical violation of tissue planes by an arthroscopic instrument or cannula, resulting in contractures of the soft tissue surrounding the articulations, pathologic connections between motion interfaces [5–13]. Among these pathologies, capsular contracture is a well-known etiology (pathologic conditions) that are related to stiff shoulders [5, 8, 10–12]; change in tissue composition stemming from altered connective tissue fibroblast activity leads to capsular contracture and this prevents enough stretching for humeral head movement, therefore causing decreased range of motion (ROM).[11, 14]. Itoi et al., in their evidence base review (Level V study), suggested that shoulder stiffness after rotator cuff surgery is typically global, but posterior capsular contracture is often accentuated [7]. However, because it is difficult to quantify capsular contracture, there is no study evaluating capsular contracture and the relation between capsular contracture and ROM after ARCR. Moreover, There are also not many studies on which portion of capsular contracture is related to specific motion. Volumetric study of glenohumeral joints has been introduced in shoulder instability studies and measurement of joint volume has been used for evaluation of the degree of redundancy in shoulder joints [15–17]. Considering the joint volume has been used to evaluate the degree of capsular redundancy, we think that the joint volume could also be used to evaluate the capsular contracture. Based on the idea, our study design was [1] to quantify the joint volume (total joint volume and quadrant compartment joint volumes), [2] to demonstrate the relationship between joint volume and postoperative ROM, and [3] to demonstrate the relationship between quadrant compartment joint volume (anteroinferior quadrant; AIQ, anterosuperior quadrant; ASQ, posterosuperior quadrant; PSQ and posteroinferior quadrant; PIQ) and postoperative ROM. The study hypothesis was that the joint volume (total joint volume and quadrant compartment joint volumes) is related to postoperative ROM after ARCR.
## Patient selection
After obtaining approval from the institutional review board, we retrospectively reviewed 1,318 patients who had undergone ARCR for a rotator cuff tear between January 2015 and December 2020. Among them, patients with small to medium (less than 25 × 25 mm anterior-posterior and medial-lateral diameter) full-thickness tears of the posterior-superior (PS) cuff were included. Moreover, patients who had undergone ROM measurement and postoperative shoulder computed tomography arthrography (CTA) 6 months postoperatively were included. The exclusion criteria were as follows: age younger than 50 years (to analyze within similar age groups); traumatic rotator cuff tear, subscapularis tendon tear, irreparable PS cuff tear, and re-tear, calcific tendinitis, revision surgery, biceps pathology, preoperative stiffness, diabetes mellitus (DM) [1, 5, 7, 12] thyroid dysfunction [5, 7] (to exclude medical conditions affecting postoperative stiffness); receipt of worker’s compensation [1, 5, 7] (which may also affect postoperative stiffness); and those who received all 20 mL of contrast medium [18] into the shoulder joint during CTA (to simulate identical filling pressure). Among the 274 patients who satisfied the inclusion criteria, 83 patients were finally selected after applying the exclusion criteria (Fig. 1).
Fig. 1Patient flow chart showing the inclusion and exclusion criteria used in this study
## Surgical method
All surgeries were performed by the one surgeon using a standardized technique. After receiving general anesthesia, patients were positioned in a beach chair at a 70° angle. The arthroscopic evaluation of all associated intra-articular lesions was performed via a standard posterior portal. The anterior portal was created just above the superior margin of the subscapularis tendon and used as a working portal. For PS cuff tears, bone preparation was performed in the footprint area with a shaver and electrocautery device (Edge®, Bipolar Arthroscopic RF System; ConMed Linvatec, Largo, FL) followed by a single-row repair using a 4.5-mm bio- Corkscrew® suture anchor (Arthrex, Naples, Florida, USA). Capsular release was not performed in all patients.
## CTA imaging protocols
All patients were informed about the procedure, timing, and possible complications of CTA. After local anesthesia on the lower one-third level of the glenohumeral joint, the joint was punctured with a spinal needle to deliver a small amount of iodine contrast medium (Ultravist), verifying that the needle tip was positioned inside the joint space. Under fluoroscopic guidance, contrast medium was used to confirm intra-articular location of needle tip. With gentle and progressive injection, flow of contrast medium away from the needle tip and opacification of the joint space confirm adequate position. Then, a solution containing 12.5 mL of saline, 6 mL of iopromide (Ultravist; Bayer Healthcare Pharmaceuticals), and 1.5 mL of mepivacaine (Mevan, 20 mg/mL per vial; Hanlim) was slowly injected by one radiologist until the patient felt pain and pressure. The patients were placed in the supine position in a CT scanner with the affected arm adducted and the head turned to the unaffected side. The scan proceeded from the superior aspect of the acromioclavicular joint to several centimeters inferior to the inferior angle of the scapula. CTA was performed using three different machines: Somatom Drive (180 mA; 120 kV; 1-mm thickness), Somatom Definition AS (117 mA; 140 kV; 1-mm thickness), and Somatom Perspective (110 mA; 130 kV) (all by Siemens) with a slice thickness of 1 mm.
## Image reconstruction
CTA images of the patients were saved as DICOM files, and reconstruction to 3D images was performed using an image reconstruction program (3D Slicer, version 4.11.2 software) [15, 19]. The current study aimed to use not only the total joint volume but also four different quadrant compartments of the joint, which is divided by the scapular plane; therefore, modification of the coordinate plane was warranted prior to joint volume measurement. Three scapular landmarks were first registered as fiducials: the angular inferior, trigonum scapulae, and glenoid center of the scapula (Fig. 2). Then, a plane passing through the three landmarks was set as the scapular plane [13, 15]. With the scapular plane as a reference, the coordinate space of the image was modified, with the scapular plane itself as a coronal plane, the perpendicular plane as an axial plane, and a plane orthogonal to both the scapular plane and axial plane as a sagittal plane.
Fig. 2Registration of three scapular landmarks (angular inferior, trigonum scapulae, and glenoid center) as fiducials before reconstruction. Scapular plane (green plane in the lower left image, green line in the upper left and lower right images): plane that passes through the three scapular landmarks. Axial plane (red plane in the upper left image, red line in the lower right and lower left images): plane that is perpendicular to the scapular plane. Sagittal plane (yellow plane in the lower right image, yellow line in the upper left and lower left images): plane that is orthogonal to the scapular and axial planes Thereafter, the contrast media of the joint was separated from bony structures using the “Grow from seeds” effect, which semi-automatically reconstructs the targeted anatomical structure based on the registered “seeds”. The semi-automated segmentation was followed by a manual erase feature to remove any excessively segmented portions or by manual segmentation for parts that required additional segmentation (Fig. 3). After separation was appropriately performed, a 3D image of the segmented joint was generated (Fig. 3). The total joint volume was calculated by applying the “Segment statistics” module using this image (Fig. 3).
Fig. 3Separation of joint portion of image from others in semi-automated manner. If additional modification was needed, further manual erasing and segmentation were performed (upper right, lower left, and lower right images). When separation was performed appropriately, a 3D image of the segmented joint was generated (upper right image). The total joint volume was then calculated via the “Segment statistics” module In the scapular sagittal plane, the segmented joint image could be further divided into anterior and posterior halves using the scapular coronal plane and into superior and inferior halves using the scapular axial plane. The center of the glenoid was set as the common origin of the coordinate plane. With center of the glenoid as the center of the clock, we defined 6 to 9 h as AIQ, 9 to 12 h as ASQ, 12 h to 3 h as PSQ, 3 to 6 h as PIQ. ( In the remainder of this article, we note clockface positions as ‘‘h’’; e.g., 3 h refers to the 3-o’clock position.) The specific quadrant volume was reconstructed by discarding the other parts using the “Scissors effect” located in the “Segment editor” module (Fig. 4). Then, the same procedure using the “Segment statistics” module was performed to calculate the volume of each compartment.
Fig. 4Further separation of the total joint volume into each quadrant compartment joint volume performed in the scapular sagittal plane. The figure shows that the joint portion was separated into inferior compartment and then posteroinferior joint portions. Thereafter, calculation of the separated joint volume was performed via the “Segment statistics” module To assess the interobserver reliability, two separate orthopedic surgeons measured the total joint volume and each quadrant compartment volume for all patients. For intraobserver reliability, a single surgeon (with 5 years of experience) performed the measurements twice with the same images at a 2-month interval. To evaluate the consistency among the raters, intraclass correlation coefficients (ICCs) were calculated. Values less than 0.5 indicate poor reliability, those between 0.5 and 0.75 indicate fair reliability, those between 0.75 and 0.9 indicate good reliability, and those greater than 0.9 indicate excellent reliability.
## ROM assessment
Measurement of patients’ passive ROM 6 months postoperatively was performed by one independent medical examiner using a goniometer. Four different motions were evaluated: scaption (Sc), external rotation on side (ERs), external rotation at 90° (ER90), and internal rotation on back (IRb).
The ROM patient data were reviewed and evaluated 6 months after ARCR via electric medical records. To facilitate the statistical analysis, IRb values were converted into contiguously numbered groups [20]: buttock for 1, L5 to L1 for 2 to 6, and T12 to T1 for 7 to 18.
## Rehabilitation method
All patients underwent identical postoperative physical therapy regimens. A customized abduction pillow brace was placed on the patient’s shoulder immediately after the surgery. Passive immobilization was performed for 6 weeks. Thereafter, the brace was removed, and active mobilization with coordination training was performed for a further 4 weeks. Finally, specific progressive resistance exercises were prescribed. Patients begin progressive resistance exercise with the Thera-Band® (HCMHygenic Corp, Batu Gajah, Malaysia). With the arm tucked close to the body, use rubber tubing to provide resistance to internal rotation of the arm. Turn around to use the tubing to provide resistance to external rotation of the arm.
## Statistical analysis
A specialized statistician from the author’s institution performed the statistical evaluation using IBM SPSS Version 25.0 (IBM Corp., Armonk, NY, USA). Pearson’s correlation analysis was used to evaluate the relationship between joint volume (total joint volume and quadrant compartment joint volumes) and each shoulder motion. A p-value less than 0.05 was considered statistically significant.
## Results
The epidemiological information, total joint volume, quadrant compartment joint volume, and ROM data of the study participants are summarized in Table 1. The reliability of the measurements for the total joint volume and each quadrant compartment joint volume was excellent, ranging from 0.767 to 0.917 for interobserver reliability and from 0.777 to 0.937 for intraobserver reliability.
Table 1Patients’ demographic dataDemographic DataMean ± Standard Deviation (Range)Age, year59.59 ± 5.11 (50–69)Height, cm159.11 ± 7.99 (133–177)BMI, kg/m223.95 ± 3.09 (15–32.1)Sex, male: female26: 57Tear size (AP), mm11.80 ± 4.95 (2.5–25)Tear size (ML), mm11.56 ± 5.48 (3–26)Scaption, degree159.71 ± 25.23 (80–180)External rotation on side, degree47.21 ± 17.35 (10–90)External rotation on 90°, degree76.58 ± 16.14 (20–90)Internal rotation on back, point7.07 ± 3.67 (1–18)Total joint volume, mL10.38 ± 3.14 (3.14–10.38)Vol. AIQ, mL2.17 ± 1.1 (0.53– 5.94)Vol. ASQ, mL2.28 ± 0.85 (0.84–4.40)Vol. PSQ, mL2.60 ± 1.23 (0.86–7.12)Vol. PIQ, mL3.14 ± 1.34 (0.60– 8.47)aBMI, body mass index; AP, anterior-posterior; ML, medial-lateral; Vol. AIQ, anteroinferior quadrant joint volume; Vol. ASQ, anterosuperior quadrant joint volume; Vol. PSQ, Posterosuperior quadrant joint volume; Vol. PIQ, posteroinferior quadrant joint volume
## Correlations between total joint volume and shoulder motions
The correlation analysis between total joint volume and each shoulder motion is shown in Fig. 5. Total joint volume showed a moderately positive correlation with each shoulder motion (Sc: Pearson coefficient, 0.32, $$p \leq 0.0047$$; ERs: Pearson coefficient, 0.24, $$p \leq 0.0296$$; ER90: Pearson coefficient, 0.33, $$p \leq 0.0023$$; IRb: Pearson coefficient, 0.23, $$p \leq 0.0336$$).
Fig. 5Scatter matrix of the relationship between the total joint volume (Vol. TJ) and each shoulder motion (Sc: scaption, ERs: external rotation on side, ER90: external rotation at 90°, IRb: internal rotation on back). The Vol. TJ showed a moderately positive correlation with each of the four shoulder motions
## Correlations between quadrant compartment joint volumes and shoulder motions
The correlation analysis between quadrant compartment joint volumes and each shoulder motion is shown in Table 2. The AIQ compartment joint volume showed a moderately positive correlation with each shoulder motion (Sc: Pearson coefficient, 0.26, $$p \leq 0.0199$$; ERs: Pearson coefficient, 0.23, $$p \leq 0.0336$$; ER90: Pearson coefficient, 0.25, $$p \leq 0.0246$$; IRb: Pearson coefficient, 0.26, $$p \leq 0.0168$$). Moreover, the PSQ compartment joint volume showed a moderately positive correlation with specific shoulder motion (Sc: Pearson coefficient, 0.24, $$p \leq 0.029$$; ER90: Pearson coefficient, 0.29, $$p \leq 0.008$$; IRb: Pearson coefficient, 0.22, $$p \leq 0.0491$$). The PIQ compartment joint volume showed a positive correlation with specific shoulder motion (Sc: Pearson coefficient, 0.30, $$p \leq 0.0064$$; ER90: Pearson coefficient, 0.29, $$p \leq 0.0072$$).
Table 2Correlation Between Each quadrant Compartment Joint Volume and Each Shoulder MotionMotionVol. AIQVol. ASQVol. PSQVol. PIQScPearson Coefficient0.260 0.000.240.30p-value0.0199*0.99740.029*0.0064*ERsPearson Coefficient0.230.090.180.16p-value0.0336*0.43190.10610.1611ER90Pearson Coefficient0.250.030.290.29p-value0.0246*0.79610.008*0.0072*IRbPearson Coefficient0.26-0.070.220.18p-value0.0168*0.5480.0491*0.0976aSc, Scaption; ERs, External Rotation on Side; ER90, External Rotation on 90°; IRb, Internal Rotation on Back; Vol. AIQ, anteroinferior quadrant joint volume; Vol. ASQ, anterosuperior quadrant joint volume; Vol. PSQ, posterosuperior quadrant joint volume; Vol. PIQ, posteroinferior quadrant joint volume*P value < 0.05 This results expressed as scatterplots in Fig. 6. In addition, tightness in specific directions according to specific capsular quadrant was schematically illustrated in Fig. 7.
Fig. 6Scatter matrix of the relationship between each quadrant compartment joint volumes (Vol. anteroinferior, Vol.anterosuperior, Vol. Posterosuperior, Vol. Posteroinferior) and each shoulder motion (Sc: scaption, ERs: external rotation on side, ER90: external rotation at 90°, IRb: internal rotation on back). The Vol. anteroinferior, Vol. Posterosuperior, Vol. Posteroinferior showed a moderately positive correlation with each of the four shoulder motions Fig. 7A schematic diagram showing the shoulder joint volume affected by postoperative tightening in a specific movement. Among the quadrant (AIQ: Anteroinferior quadrant, ASQ: Anterosuperior quadrant, PSQ: Posterosuperior quadrant, PIQ: Posterosuperior quadrant), the affected area in specific joint motion are filled in blue
## Discussion
The main findings of this study were that [1] there was positive correlation between the total joint volume and each shoulder motion and [2] there were positive correlations between AIQ, PSQ and PIQ compartment joint volume and spectific shoulder motion.
Postoperative stiffness after ARCR is commonly reported morbidity after ARCR and concerns to shoulder surgeons due to inferior functional outcome when developed.[1, 2, 6]. Although the pathophysiology of stiffness after ARCR is not well understood, postoperative ROM is affected by several factors, such as capsular contracture, contracture or atrophy of the rotator cuff itself, and adhesions within the extra-articular glenohumeral motion interface.[7, 12]. However, there is a lack of logical development in the correlation whether the decrease in range of motion was caused by surgery. For proving a cause-effect relationship between range of motion and joint volume after rotator cuff repair, we should have evaluated the relation between change of range of motion and change of joint volume. But, the test to obtain the volume of the shoulder joint (In our study; shoulder CT arthrography) are not performed usually and routinely in the pre-op patients. So we designed the study for evaluating postoperative shoulder joint volume, which could reflect capsule contraction after rotator cuff repair. Our study goal was not to find out the cause of the decrease in joint volume after surgery by comparing the range of joint motion before and after surgery. Several investigators tried to quantitatively measure pathologic regions in MRI images for evaluating capsular contracture; the width, depth, and height of the axillary recess, dimension of rotator interval and the glenohumeral distance [21–23]. However, these studies evaluated pathologic regions in 2D images, which may not accurately reflect the status of the capsular contracture. Other authors measured the capsular volume of the glenohumeral joint according to the volume of fluid injected into the capsule with or without pressure measurement.[1, 12, 24]. Even though these methods are excellent for evaluating capsular contracture, procedures such as fluid injection, volume and/or pressure measurement are not easily applied to patients during follow-up after operation. In the current study, we calculated the volume of the shoulder joint using CTA. CTA has been routinely performed to patients who underwent ARCR at postoperative 6 months in our hospital for evaluating cuff continuity, and selection bias can be reduced. Using 3D Slicer software which has been used in a variety of medical studies, [15, 17, 19] and its accuracy and efficiency in 3-dimensional segmentation and analysis have been well described, we could measure capsular volume through CTA DICOM files.
Decreased total joint volume in a primary stiff shoulder is well documented in previous studies [23, 25, 26]. However, previous studies used comparative analyses and reported decreased joint volume in the stiff group. Therefore, the relation between joint volume and ROM could not be evaluated. Moreover, the subjects in the previous study were primary stiff shoulder patients. Primary shoulder stiffness and postoperative stiffness after ARCR are considered to be different disease entities due to their different natural courses, [6, 9, 19, 27, 28] therefore, the results of the previous studies could not be applied to postoperative stiffness after ARCR. In the current study, we performed a correlation analysis between total joint volume and postoperative ROM after ARCR. The correlation analysis between total joint volume and each shoulder motion showed a moderately positive correlation with each shoulder motion (Sc: Pearson coefficient, 0.32, $$p \leq 0.0047$$; ERs: Pearson coefficient, 0.24, $$p \leq 0.0296$$; ER90: Pearson coefficient, 0.33, $$p \leq 0.0023$$; IRb: Pearson coefficient, 0.23, $$p \leq 0.0336$$). This result implies that total joint volume is related to the postoperative ROM and that a decrease in total joint volume could lead to postoperative stiffness. Therefore, procedures to increase the total joint volume such as selective capsulectomy during operation, posterior capsular stretching exercise after operation etc., are crucial to increase the postoperative ROM, and this can further lead to the prevention of postoperative stiffness. Although the usefulness of capsulectomy and early rehabilitation remain controversial, [1, 6, 9, 14] based on our results, procedures that increase total joint volume may help prevent postoperative stiffness from a clinical perspective.
In the current study, we could easily separate specific areas of joint volume (anteroinferior, anterosuperior, posterosuperior and posteroinferior quadrant) using 3D Slicer software and evaluate the relation between specific area joint volume and ROM after ARCR. Our study showed that anterior compartment volume (especially AIQ) was related to ERs, posterior compartment joint volume (PSQ & PIQ) related to IRb, and inferior compartment joint volume (AIQ & PIQ) related to Sc. These findings were similar to the results of previous studies [11, 29–31]. In our study, Interestingly, AIQ joint volume were related to all shoulder motion and inferior compartment volume (AIQ & PIQ) were also related to ER at 90 and IRb. In addition, posterior compartment volume (PSQ & PIQ) was related to not only IRb but also Sc and ER at 90. Even though we cannot delineate the cause of these findings from current study, this result may be due to methodological differences. Previous studies reproduced joint contracture by selective capsular plications or thermal impacts for ligament shortening to change the specific joint volume and then measured passive range of motion. Whereas, we used shoulder CT arthrography to obtain specific joint volume postoperatively and measured the ROM of patients. In our study, positive correlations between posterior half and inferior half compartment joint volume (AIQ & PSQ & PIQ) and specific shoulder motion are clinically meaningful. Given that the posterior and inferior capsular contracture is often accentuated after ARCR and posterior half and inferior half compartment joint volume are related to specific shoulder ROM, [7] efforts to increase the posterior half and inferior half compartment joint volume such as selective capsulectomy during operation and posterior capsular stretching exercise after operation may be useful in the prevention and treatment of postoperative stiffness.
There are several limitations in the current study. [ 1] Among 1318 consecutive patients during the study period, 274 patients were included after applying the inclusion criteria. During applying inclusion criteria, only $20\%$ patients were selected. In addition, among 274 patients, 83 patients were selected as final subjects after applying the exclusion criteria. Thus, this study may have a selection bias. However, we wanted to evaluate the relation between joint volume and ROM after arthroscopic repair of small to medium sized rotator cuff tears. In addition, we also excluded patients who had factors that are known to affect postoperative stiffness. Considering that the final subjects are selected after applying inclusion and exclusion criteria among consecutive patients, selection bias may be minimized. [ 2] Evaluation of the patients’ ROM was done at 6 months postoperatively, which can be a short observation period for defining postoperative stiffness. Although, goal of our study was not to evaluate joint stiffness after surgery, but to assess the relationship between joint volume and specific range of motion after surgery. Therefore, it is not necessary to evaluate after the stiffness is completely resolved after surgery. In addition, as a retrospective study, we have been conducting CT arthrography for six months after surgery routinely because “naturally resolving” postoperative stiffness patients mostly shows improvement of ROM within 6 months and further observation period makes no difference because tendon stops healing and starts to remodel at this time [1, 20, 28, 29]. [ 3] Although we obtained excellent interobserver reliability (Ranging from 0.767 to 0.917) and intraobserver reliability (0.777 to 0.937), measurement error could not completely be excluded because the joint volume was measured in semi-automated manner. [ 4] For the evaluation of the degree of capsular contracture, measuring the intra-articular pressure is more accurate method than measuring joint volume. However, it is very difficult to measure intra-articular pressure for every follow-up patients and to measure pressure of specific joint compartment. Therefore, we used joint volume to evaluate the degree of capsular contracture. However, joint volume depends on the expansion behavior of the capsule which is not linear but logarithmic. To overcome this problem, we stopped injection of contrast media at the time patient felt pain and pressure during CTA. Furthermore, we excluded the patient who did not felt pain or pressure and received all 20 mL of contrast media in this study. [ 5] Clinical scores are not assessed in this study. Study about the influence of joint volume on clinical scores may have induced another clinically meaningful results and further study regarding this issue seems to be useful.
In conclusion, the total joint volume showed positive correlation with postoperative ROM after ARCR. Specifically, among quadrant compartment joint volumes, posterior half and inferior half compartment joint volume (AIQ, PSQ and PIQ joint volumes) were related to ROM after ARCR. Considering its frequent incidence and effect on poor clinical outcome, preventing the postoperative stiffness after ARCR is an important issue and perioperative methods to increase the total joint volume, especially the posterior or inferior part of the capsule seems to be useful.
## References
1. Park JY, Chung SW, Hassan Z, Bang JY, Oh KS. **Effect of Capsular Release in the treatment of shoulder stiffness concomitant with Rotator Cuff Repair Diabetes as a predisposing factor Associated with Treatment Outcome**. *Am J Sports Med* (2014) **42** 840-50. DOI: 10.1177/0363546513519326
2. Seo SS, Choi JS, An KC, Kim JH, Kim SB. **The factors affecting stiffness occurring with rotator cuff tear**. *J Shoulder Elbow Surg* (2012) **21** 304-9. DOI: 10.1016/j.jse.2011.04.011
3. DeFranco MJ, Bershadsky B, Ciccone J, Yum JK, Iannotti JP. **Functional outcome of arthroscopic rotator cuff repairs: a correlation of anatomic and clinical results**. *J Shoulder Elbow Surg* (2007) **16** 759-65. DOI: 10.1016/j.jse.2007.03.020
4. Galatz LM, Ball CM, Teefey SA, Middleton WD, Yamaguchi K. **The outcome and repair integrity of completely arthroscopically repaired large and massive rotator cuff tears**. *J Bone Joint Surg Am* (2004) **86** 219-24. DOI: 10.2106/00004623-200402000-00002
5. Schneider WR, Trasolini RG, Riker JJ, Gerber N, Ruotolo CJ. **Stiffness after arthroscopic rotator cuff repair: a rehabilitation problem or a surgical indication?**. *JSES Int* (2021) **5** 88-92. DOI: 10.1016/j.jseint.2020.09.001
6. Chung SW, Huong CB, Kim SH, Oh JH. **Shoulder stiffness after rotator cuff repair: risk factors and influence on outcome**. *Arthroscopy* (2013) **29** 290-300. DOI: 10.1016/j.arthro.2012.08.023
7. Itoi E, Arce G, Bain GI, Diercks RL, Guttmann D, Imhoff AB. *Shoulder Stiffness: Current Concepts and Concerns Arthroscopy* (2016) **32** 1402-14. PMID: 27180923
8. Franceschi F, Papalia R, Palumbo A, Vasta S, Maffulli N, Denaro V. **Management of postoperative shoulder stiffness**. *Sports Med Arthrosc Rev* (2011) **19** 420-7. DOI: 10.1097/JSA.0b013e3182393e06
9. Kim I-B, Jung DW. **An intra-articular steroid injection at 6 weeks postoperatively for shoulder stiffness after arthroscopic rotator cuff repair does not affect repair integrity**. *Am J Sports Med* (2018) **46** 2192-202. DOI: 10.1177/0363546518777739
10. Vastamaki H, Vastamaki M. **Postoperative stiff shoulder after open rotator cuff repair: a 3- to 20-year follow-up study**. *Scand J Surg* (2014) **103** 263-70. DOI: 10.1177/1457496913514383
11. Dashottar A, Borstad J. **Posterior glenohumeral joint capsule contracture**. *Shoulder Elb* (2012) **4** 230-6. DOI: 10.1111/j.1758-5740.2012.00180.x
12. Tauro JC. **Stiffness and rotator cuff tears: incidence, arthroscopic findings, and treatment results**. *Arthroscopy* (2006) **22** 581-6. DOI: 10.1016/j.arthro.2006.03.004
13. Casier SJ, Van den Broecke R, Van Houcke J, Audenaert E, De Wilde LF, Van Tongel A. **Morphologic variations of the scapula in 3-dimensions: a statistical shape model approach**. *J Shoulder Elbow Surg* (2018) **27** 2224-31. DOI: 10.1016/j.jse.2018.06.001
14. Kim J-H, Ha D-H, Kim S-M, Kim K-W, Han S-Y, Kim Y-S. **Does arthroscopic preemptive extensive rotator interval release reduce postoperative stiffness after arthroscopic rotator cuff repair?: a prospective randomized clinical trial**. *J Shoulder Elbow Surg* (2019) **28** 1639-46. DOI: 10.1016/j.jse.2019.05.021
15. Lee S-J, Kim J-H, Gwak H-C, Kim C-W, Lee C-R, Jung S-H. **Influence of glenoid labral bumper height and capsular volume on clinical outcomes after arthroscopic Bankart repair as assessed with serial CT arthrogram: can anterior-inferior volume fraction be a prognostic factor?**. *Am J Sports Med* (2020) **48** 1846-56. DOI: 10.1177/0363546520924809
16. Wiater JM, Vibert BT. **Glenohumeral joint volume reduction with progressive release and shifting of the inferior shoulder capsule**. *J Shoulder Elbow Surg* (2007) **16** 810-4. DOI: 10.1016/j.jse.2007.02.117
17. Park JY, Chung SW, Kumar G, Oh KS, Choi JH, Lee D. **Factors affecting capsular volume changes and association with outcomes after Bankart repair and capsular shift**. *Am J Sports Med* (2015) **43** 428-38. DOI: 10.1177/0363546514559825
18. Ahn SJ, Hong SH, Chai JW, Choi J-Y, Yoo HJ, Kim SH. **Comparison of image quality of shoulder CT arthrography conducted using 120 kVp and 140 kVp protocols**. *Korean J Radiol* (2014) **15** 739-45. DOI: 10.3348/kjr.2014.15.6.739
19. Fedorov A, Beichel R, Kalpathy-Cramer J, Finet J, Fillion-Robin JC, Pujol S. **3D slicer as an image computing platform for the quantitative Imaging Network**. *Magn Reson Imaging* (2012) **30** 1323-41. DOI: 10.1016/j.mri.2012.05.001
20. Oh JH, Kim SH, Lee HK, Jo KH, Bin SW, Gong HS. **Moderate preoperative shoulder stiffness does not alter the clinical outcome of rotator cuff repair with arthroscopic release and manipulation**. *Arthroscopy: The Journal of Arthroscopic & Related Surgery* (2008) **24** 983-91. DOI: 10.1016/j.arthro.2008.06.007
21. Lee KH, Park HJ, Lee SY, Youn IY, Kim E, Park JH. **Adhesive Capsulitis of the Shoulder Joint: value of Glenohumeral Distance on magnetic resonance arthrography**. *J Comput Assist Tomogr* (2017) **41** 116-20. DOI: 10.1097/RCT.0000000000000475
22. Mengiardi B, Pfirrmann CW, Gerber C, Hodler J, Zanetti M. **Frozen shoulder: MR arthrographic findings**. *Radiology* (2004) **233** 486-92. DOI: 10.1148/radiol.2332031219
23. Ogul H, Tas N, Tuncer K, Polat G, Ogul Y, Pirimoglu B. **3D volumetric MR arthrographic assessment of shoulder joint capacity in patients with primary adhesive capsulitis**. *Br J Radiol* (2019) **92** 20180496. DOI: 10.1259/bjr.20180496
24. Ko JY, Wang FS. **Rotator cuff lesions with shoulder stiffness: updated pathomechanisms and management**. *Chang Gung Med J* (2011) **34** 331-40. PMID: 21880187
25. Binder AI, Bulgen DY, Hazleman BL, Tudor J, Wraight P. **Frozen shoulder: an arthrographic and radionuclear scan assessment**. *Ann Rheum Dis* (1984) **43** 365-9. DOI: 10.1136/ard.43.3.365
26. Bunker T. **Time for a new name for frozen shoulder—contracture of the shoulder**. *Shoulder & Elbow* (2009) **1** 4-9. DOI: 10.1111/j.1758-5740.2009.00007.x
27. Reeves B. **The natural history of the frozen shoulder syndrome**. *Scand J Rheumatol* (1975) **4** 193-6. DOI: 10.3109/03009747509165255
28. Shaffer B, Tibone JE, Kerlan RK. **Frozen shoulder. A long-term follow-up**. *J Bone Joint Surg Am* (1992) **74** 738-46. DOI: 10.2106/00004623-199274050-00013
29. Audige L, Aghlmandi S, Grobet C, Stojanov T, Muller AM, Felsch Q. **Prediction of shoulder stiffness after arthroscopic rotator cuff repair**. *Am J Sports Med* (2021) **49** 3030-9. DOI: 10.1177/03635465211028980
30. Gagey OJ, Boisrenoult P. **Shoulder capsule shrinkage and consequences on shoulder movements**. *Clin Orthop Relat Res* (2004) **419** 218-22. DOI: 10.1097/00003086-200402000-00036
31. Gerber C, Werner CM, Macy JC, Jacob HA, Nyffeler RW. **Effect of selective capsulorrhaphy on the passive range of motion of the glenohumeral joint**. *J Bone Joint Surg Am* (2003) **85** 48-55. DOI: 10.2106/00004623-200301000-00008
|
---
title: Assessment of independent comorbidities and comorbidity measures in predicting
healthcare facility-onset Clostridioides difficile infection in Kenya
authors:
- Winnie C. Mutai
- Marianne Mureithi
- Omu Anzala
- Brian Kullin
- Robert Ofwete
- Cecilia Kyany’ a
- Erick Odoyo
- Lillian Musila
- Gunturu Revathi
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022263
doi: 10.1371/journal.pgph.0000090
license: CC BY 4.0
---
# Assessment of independent comorbidities and comorbidity measures in predicting healthcare facility-onset Clostridioides difficile infection in Kenya
## Abstract
### Introduction
Clostridioides difficile is primarily associated with hospital-acquired diarrhoea. The disease burden is aggravated in patients with comorbidities due to increased likelihood of polypharmacy, extended hospital stays and compromised immunity. The study aimed to investigate comorbidity predictors of healthcare facility-onset C. difficile infection (HO-CDI) in hospitalized patients.
### Methodology
We performed a cross sectional study of 333 patients who developed diarrhoea during hospitalization. The patients were tested for CDI. Data on demographics, admission information, medication exposure and comorbidities were collected. The comorbidities were also categorised according to Charlson Comorbidity Index (CCI) and Elixhauser Comorbidity Index (ECI). Comorbidity predictors of HO-CDI were identified using multiple logistic regression analysis.
### Results
Overall, $\frac{230}{333}$ ($69\%$) patients had comorbidities, with the highest proportion being in patients aged over 60 years. Among the patients diagnosed with HO-CDI, $\frac{63}{71}$($88.7\%$) reported comorbidities. Pairwise comparison between HO-CDI patients and comparison group revealed significant differences in hypertension, anemia, tuberculosis, diabetes, chronic kidney disease and chronic obstructive pulmonary disease. In the multiple logistic regression model significant predictors were chronic obstructive pulmonary disease (odds ratio [OR], 9.51; $95\%$ confidence interval [CI], 1.8–50.1), diabetes (OR, 3.56; $95\%$ CI, 1.11–11.38), chronic kidney disease (OR, 3.88; $95\%$ CI, 1.57–9.62), anemia (OR, 3.67; $95\%$ CI, 1.61–8.34) and hypertension (OR, 2.47; $95\%$ CI, 1.–6.07). Among the comorbidity scores, CCI score of 2 (OR 6.67; $95\%$ CI, 2.07–21.48), and ECI scores of 1 (OR, 4.07; $95\%$ CI, 1.72–9.65), 2 (OR 2.86; $95\%$ CI, 1.03–7.89), and ≥ 3 (OR, 4.87; $95\%$ CI, 1.40–16.92) were significantly associated with higher odds of developing HO-CDI.
### Conclusion
Chronic obstructive pulmonary disease, chronic kidney disease, anemia, diabetes, and hypertension were associated with an increased risk of developing HO-CDI. Besides, ECI proved to be a better predictor for HO-CDI. Therefore, it is imperative that hospitals should capitalize on targeted preventive approaches in patients with these underlying conditions to reduce the risk of developing HO-CDI and limit potential exposure to other patients.
## Introduction
Clostridioides difficile is a significant nosocomial pathogen contributing to approximately $12\%$ of health care facility-associated diarrhoea in the USA [1]. It is well known that C. difficile forms part of the diverse microbiota in the gut. Nevertheless, gut changes that cause a reduction in the gut microbial diversity potentiate the overgrowth and establishment of pathogenic C. difficile. While antibiotic exposure is typically a prerequisite for C. difficile infection (CDI), epidemiological evidence especially from developed countries has established that advanced age, extended hospital length of stay, comorbidities and the use of acid-suppressive agents are additional predictors implicated in the development of healthcare facility-onset C. difficile infection (HO- CDI) [2].
Despite the investigation of these key risk factors, cases of CDI continue to rise. Furthermore, while the significance of comorbidities on CDI has been well established in developed countries, limited research from resource-limited countries is available. Previous studies have attempted to describe comorbidities commonly associated with HO-CDI and they have shown that chronic kidney disease, HIV/AIDS and other immunodeficiency disorders, chronic obstructive pulmonary disease, inflammatory bowel disease, hematological malignancy, and diabetes mellitus increase the risk of both initial and recurring CDI [3–6]. Other comorbidities that have been implicated include cardiovascular disease, digestive disorders, dementia, cerebrovascular disease, congestive heart failure, peripheral vascular disease, and myocardial infarction [7, 8]. However, the actual pathophysiological mechanisms and specific relationship of how these comorbidities influence the development of CDI is not well understood [6]. Certainly, comorbidities are known to down-regulate the immune system and cause organ dysfunction. However, patients with these comorbidities are more likely to be hospitalized and receive antibiotics, which places them at an increased risk of HO-CDI.
It is important to evaluate the effect of individual comorbidities on HO-CDI, as most of these comorbid conditions are interrelated and therefore the predictions may be overestimated. As such, the concept of stratification of comorbidities using validated comorbidity indices or aggregated scores minimizes the effect of correlation while still controlling for potential confounding variables. The Charlson Comorbidity Index (CCI) and Elixhauser Comorbidity Index (ECI) are widely used measures in health research to access comorbidities and have previously been applied in studies to predict the risk of CDI. While CCI measures 19 comorbid conditions weighted 1 to 6, ECI has a more extensive list of 31 conditions considering additional conditions such as hypertension, weight loss, obesity and psychiatric disorders that are excluded from other indices. Higher Charlson and Elixhauser comorbidity scores have previously been correlated with CDI and represent an increased likelihood of developing HO-CDI [5, 7, 8].
Until recently, studies in Kenya have demonstrated the existence of CDI in hospitalized patients; however, none of them have evaluated the comorbidities defined by the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) as potential risk factors for HO-CDI [9–11]. To build on existing knowledge and generate data from developing countries, we assessed comorbidities in a population previously tested for HO-CDI. The information may help in stratifying patients with significant comorbidities facilitating the design of prevention approaches and targeted treatment at an early stage of HO-CDI diagnosis.
## Study population
Using a cross sectional study approach, we enrolled 333 hospitalized patients between 2016–2018 [9]. The inclusion criteria comprised of all age groups who developed diarrhoea > 3days after admission. Data were obtained by conducting interviews with adult patients or guardians of minors and reviewing their files to check for consistency and additional information. A data collection form was used to collect information on age, gender, admission ward, duration of hospitalization, diagnosis on admission, previous history of admission, medication used, and existing comorbidities. Comorbidities reported by the patients or indicated in the patient files were used to calculate CCI and ECI scores.
## Outcome variable
The study outcome variable was healthcare facility-onset C. difficile infection (HO-CDI). HO- CDI was defined as the onset of diarrhoea >3 days after admission to a healthcare facility and a positive result for amplification of the C. difficile tpi gene, combined with one or more toxin genes (tcdA, tcdB, cdtA/cdtB) based on a previously described nucleic acid amplification test [9].
## Predictor variables and confounders
The study investigated individual comorbid conditions defined by the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). Comorbidity was defined as the pre-existence of one or more medical conditions coexisting with the primary condition [12]. A total of 22 specific comorbidities were considered: congestive heart failure, cardiac arrhythmias, chronic obstructive pulmonary disease (COPD), hypertension, peptic ulcer disease, diabetes, hemiplegia, hypothyroidism, pulmonary circulation disorders, chronic kidney disease (CKD), liver disease, solid tumor without metastasis, metastatic solid tumor, HIV/AIDS, lymphoma, weight loss (malnutrition), anemia, and depression. Each of these individual comorbidities was investigated separately. To construct comorbidity scores, each comorbid condition was assigned a weight based on the relative risk of mortality for risk adjustment (S1 Table). Then the indexes were summed-up to provide the total scores and categories before exploring their association with HO-CDI.
The potential confounders included age, hospitalization duration, medication administered (antibiotics, laxatives, analgesics, antiretrovirals, chemotherapy agents), previous admission, primary disease (a condition present at admission), gastrointestinal procedures (colonoscopy/endoscopy and surgery).
## Ethical approval
This study was approved by the Kenyatta National Hospital-University of Nairobi Ethics and Research Committee (P$\frac{8}{01}$/2014). Written informed consent was obtained from adult participants and legal guardians of the minors.
## Statistical analysis
Descriptive statistics for demographic and clinical information of the study participants were computed and the outcomes were expressed as frequencies and percentages and summarized in Table 1. Individual comorbidities and comorbidity indices were profiled in Table 2 where z test for proportion was used to test for significant differences in individual comorbidities and Pearson Chi-Square applied to test for association between CCI and ECI groups and HO-CDI outcome. A binary logistic regression analysis was conducted in a sequential approach to identify significant comorbidity predictors of HO-CDI and presented in a forest plot.
First, variables with p-values of ≤ 0.2 from Table 2 and those known to be clinically relevant based on literature were selected for the model development stage. The variables were individually fitted into a bivariate binary logistic regression model to obtain crude odds ratios of comorbidities/comorbidity scores associated with the likelihood of developing HO-CDI (Fig 1A. Secondly, variables whose p-values were ≤0.05 in the bivariate analyses were considered in a final multiple binary logistic regression model where potential confounders were controlled (Fig 1B). Adjusted odds ratio (AOR), corresponding p-value and the $95\%$ confidence interval (CI) were used to identify significant independent comorbidities and comorbidity scores associated with the risk of developing HO-CDI. Variables with p ≤ 0.05 were considered statistically significant. Likelihood ratio test was used to assess for goodness-of-fit that is whether adding more parameters to ECI and CCI models had significant impact in predicting the outcome of CDI. Here, the likelihood ratio test static (assumed to follow chi-squared distribution) was generated by getting the difference between log-likelihoods of the simple and complex models, and degrees of freedom represented by additional parameters in the complex model. Finally, to determine the CCI and ECI performance in predicting HO-CDI outcome, the model fit was assessed using a pseudo-R squared where the model with the higher value was considered a better predictor of HO-CDI.
**Fig 1:** *Forest plot depicting odds ratios (OR) with 95% confidence interval (95% CI) of predicting HO-CDI in patients with different comorbidity profiles.(A) Univariate logistic regression model showing unadjusted crude odds ratios. (B) Multiple binary logistic regression model showing adjusted odds ratios. The horizontal lines indicate the width of the confidence interval while the vertical marks on each horizontal line show the odds ratios. An odds ratio of more than 1.0 indicates increased risk. Abbreviations: OR- Odds ratio; CI- Confidence Interval; CCI, Charlson Comorbidity Index; ECI, Elixhauser Comorbidity Index; p-value indicating the level of statistical significance (p < 0.05).*
For each variable, patients without the individual comorbid conditions were the reference group while for comorbidity scores, a score of 0 was the reference value in the analysis. The statistical analysis and visualization were performed using STATA version 13.1.
## Demographic and participants characteristics
The baseline characteristics of the 333 patients with and without comorbidities are summarized in Table 1. A total of $\frac{230}{333}$ ($69\%$) had comorbidities, while $\frac{102}{333}$ ($31\%$) did not report any history of comorbidities. The proportion of comorbidities was greater in patients aged over 60 years. Additionally, higher proportions of medication exposure were recorded in the group with comorbidities: antiretrovirals $\frac{49}{51}$ ($96.1\%$), acid-suppressive agents $\frac{85}{103}$ ($82.5\%$) and antibiotics $\frac{221}{297}$ ($74.4\%$) than in the non-comorbid group. Among the total number of patients who reported prior hospital admission in the preceding three months, $\frac{44}{48}$ ($91.7\%$) had comorbid conditions while those diagnosed with HO-CDI were $\frac{63}{71}$ ($88.7\%$). However, there were no proportional differences in the duration of hospitalization (≤30 or >30 days) in the comorbid and non-comorbid groups.
Independent comorbidities that differed significantly between the HO-CDI patients and the comparison group included hypertension ($23.5\%$ vs. $9.5\%$), anemia ($22.5\%$ vs. $9.5\%$), tuberculosis ($21.1\%$ vs. $9.5\%$), diabetes ($16.9\%$ vs. $6.9\%$), chronic kidney disease ($15.5\%$ vs. $4.6\%$) and chronic obstructive pulmonary disease ($9.9\%$ vs. $1.5\%$). Additionally, the results revealed that although there were more patients with HIV/AIDS ($25.4\%$ vs. $15.7\%$) and peptic ulcer disease ($11.3\%$ vs. $5.3\%$) in the HO-CDI group, the difference in proportions were not significant. A majority of the participants had a score of zero in both the CCI ($$n = 230$$) and ECI ($$n = 158$$). However, more than a third of the participants who had a CCI score of 1 ($33.3\%$), 2 ($44\%$) and ≥ 3 ($34.8\%$) were positive for HO-CDI compared to only $14.4\%$ who had a score of 0. Also, $28\%$, $30\%$ and $48\%$ of the participants positive for HO-CDI had an ECI score of 1, 2, and ≥ 3, respectively. The differences between the categories of both the CCI and ECI were statistically significant at $p \leq 0.001.$ Summary statistics comparing the proportions of HO-CDI outcome by individual comorbidities and comorbidity scores are illustrated in Table 2.
Independent comorbidities including congestive heart failure, peripheral vascular disease, hemiplegia, leukemia, metastatic solid tumor, cardiac arrhythmias, hypothyroidism, lymphoma and depression were not explored further because of their relatively low frequencies.
## Logistic regression analysis of potential comorbidity predictors of HO-CDI
The unadjusted and adjusted logistic regression models were analyzed to explore the association between comorbidities and HO-CDI (Fig 1A and 1B). Eight individual comorbidities with p-values of ≤ 0.2 and all the comorbidity scores in each category were individually fitted into a crude logistic regression model. Following univariate analysis, patients with chronic pulmonary disease (Odds ratio, OR, 7.06; $95\%$ Confidence interval, CI, 2–24.84; $p \leq 0.05$), chronic kidney disease (OR, 3.82; $95\%$ CI, 1.61–9.04; $p \leq 0.001$), hypertension (OR, 2.98; $95\%$ CI, 1.51–5.91; $p \leq 0.05$), diabetes (OR, 2.76; $95\%$ CI, 1.26–6.04; $p \leq 0.05$), anemia (OR, 2.76; $95\%$ CI, 1.38–5.51; $p \leq 0.05$) and tuberculosis (OR, 2.54; $95\%$ CI, 1.26–5.13; $p \leq 0.05$) were more likely to have HO-CDI conditions compared to non-HO-CDI patients. Additionally, patients with CCI scores of 2 (OR, 4.69; $95\%$ CI, 1.96–11.21; $p \leq 0.001$) and ≥ 3 (OR, 3.19; $95\%$ CI, 1.71–5.97; $p \leq 0.001$) were more likely to develop HO-CDI as were those with ECI scores of 1, 2 and ≥ 3. Peptic ulcer disease and HIV/AIDS were not significant in the crude logistic regression model and were therefore not fitted into the adjusted logistic regression model.
After adjusting for all the potential confounders, five independent comorbidities were identified as potential predictors of HO-CDI: chronic obstructive pulmonary disease (OR, 9.51; $95\%$ CI, 1.80–50.1), diabetes (OR, 3.56; $95\%$ CI, 1.11–11.384), chronic kidney disease (OR, 3.88; $95\%$ CI, 1.57–9.62), anemia (OR, 3.67; $95\%$ CI, 1.61–8.34) and hypertension (OR, 2.47; $95\%$ CI, 1–6.07). In comparison to patients who did not have tuberculosis, patients with tuberculosis were $48\%$ more likely to develop HO-CDI although this was not statistically significant. In reference to comorbidity scores, while adjusting for all confounding variables and other comorbidities, the patients who had a CCI score of 2 were 6.67-times ($95\%$ CI: 2.07–21.48; $p \leq 0.001$) more likely to have HO-CDI compared to patients who did not have CCI comorbidities (i.e., CCI = 0), while patients who had an ECI score of 1, 2 and ≥ 3 were associated with a 4.07-times ($95\%$ CI: 1.72–9.65; $p \leq 0.001$), 2.86-times ($95\%$ CI: 1.03–7.89; $p \leq 0.05$) and 4.87-times ($95\%$ CI: 1.40–16.92; $p \leq 0.05$) increased odds of HO-CDI, respectively.
## Discussion
This study is the first to assess independent comorbidities and comorbidity scores that increase the risk of developing HO-CDI in hospitalized patients in a Kenyan cohort. We observed that the majority of the hospitalized patients had underlying conditions with significantly higher proportions in the older population. In addition, an overall significantly higher rate of HO-CDI was observed in the patient population with comorbidities.
Comorbidities with higher prevalence in patients with HO-CDI included HIV/AIDS, followed by hypertension, anemia, tuberculosis, diabetes, chronic kidney disease, peptic ulcer disease and chronic obstructive pulmonary disease. Most of these comorbidities would necessitate polypharmacy and prolonged hospital admission directly influencing the shift from C. difficile colonization to subsequent CDI [13]. Consistent with previous studies, independent comorbidities including hypertension, chronic kidney disease, anemia, diabetes, chronic obstructive pulmonary disease as well as aggregate Charlson Comorbidity scores and Elixhauser Comorbidity scores were significantly associated with increased risk of HO-CDI [5, 8, 14].
In the present study we noted that chronic diseases including diabetes, hypertension, chronic kidney disease and chronic obstructive pulmonary disease were significantly highly ranked predictors of HO-CDI. It is known that a chronic disease naturally compromises the immune system. Consequently, decreased immunological tolerance among our study participants would have increased their susceptibility to infections, likely leading to more antibiotic intake and prolonged hospitalization. Moreover, the descriptive statistics from this study showed that antibiotics and previous history of admission, which are both potential risk factors of developing HO-CDI were significantly higher among the patients with comorbidities. Interestingly, even after adjusting for these factors in the multiple logistic regression, strong correlations were still observed. Thus, our results suggest that chronic diseases are significant predictors of HO-CDI as noted earlier [15–17].
Findings from this study suggest that COPD was among the significant predictors resulting in 9.5 times increased risk of HO-CDI. A possible explanation for this could be that the increased susceptibility to bacterial respiratory tract infections in COPD patients contributes to greater consumption of antibiotics, which in turn predisposes patients to HO-CDI [18]. According to Jasiak et al, COPD resulted in a 3.5-fold increased risk of recurrent HO-CDI [19].
Previous studies comparing patients with and without underlying chronic kidney disease (CKD) noted that the former had a higher risk of initial and recurrent episodes of CDI [20, 21]. Similarly, these findings were supported by a recent study that observed an almost four-fold increased risk (OR:3.68, CI: 1.63–8.31, $$p \leq 0.002$$) of developing CDI in patients with underlying CKD [22]. The reduced function of the kidney not only interferes with the elimination of the toxins from the body but also alters the functions of the intestinal microbiota and activates systemic inflammation [23–25]. These observations could therefore explain the increased susceptibility of HO-CDI in patients with CKD.
Hypertension was the highest comorbidity observed among the study participants and was significantly associated with an increased risk of developing HO-CDI. Previous investigations have reported similar findings [26, 27]. Currently, the reason behind the increased risk is not apparent, however, accumulating evidence using both animal and human models suggests that hypertension influences gut microbiota dysbiosis [28, 29]. On the other hand, antihypertensive drugs have been shown to improve or compromise intestinal microbiota [30, 31]. Verapamil, for example, protected the cells from C. difficile intoxication [32]. We however did not collect any information on antihypertensive medication in this study. Therefore, based on the data we could not ascertain whether hypertension itself or the hypertensive medication was responsible for increased odds of developing HO-CDI.
Another important chronic disease predictor observed in this study was diabetes. Patients with diabetes were three times more likely to develop CDI compared to non-diabetic patients. The relationship between CDI and diabetes has been studied extensively. Diabetes has been established as a possible independent risk factor for primary and recurrent CDI [33–35]. Diabetes causes structural remodeling of the colon that affects various functions of the gastrointestinal tract leading to, amongst other things, impaired motility and an altered composition of the intestinal microbiota, which may contribute to C. difficile driven diarrhoea [36, 37]. On the other hand, in their case-control study Eliakim-Raz et al. observed that diabetic patients treated with metformin an anti-diabetic drug had reduced odds (OR 0.58; $95\%$ CI, 0.37–0.93; $$p \leq 0.023$$) of developing CDI compared to their counterparts [38]. Similarly, an interventional study observed that metformin-treated diabetic patients had a reduced abundance of Clostridium spp., which could significantly impact C. difficile colonization [39]. Although the exact mechanism behind this is not clear, a potential mechanism that has been investigated is that metformin alters the reabsorption of secondary bile acids and as a result inhibits spore germination, vegetative growth and toxin activity of C. difficile strains [40–42]. Therefore, even though a causal relationship has not been established, it is evident that structural and functional changes in the colon induced by diabetes itself or diabetes medication are likely to alter the composition of the gut microbiota, which consequently increases or reduces the risk of CDI [43].
Univariate analysis showed an association between tuberculosis and HO-CDI, however after adjusting for potential confounders including anti-tuberculosis treatment, no statistical difference was observed in patients with HO-CDI in comparison with patients without HO-CDI. Thus, the relationship between tuberculosis and HO-CDI may have occurred because of the confounding effect of anti-tuberculosis drug exposure. Rifampicin was previously shown to induce CDI in patients receiving anti-tuberculosis treatment [44]. Additionally, prolonged use of rifampicin has resulted in high resistance rates in some settings, consequently promoting the persistence of resistant C. difficile strains in patients undergoing tuberculosis treatment [45, 46]. In support of this, we previously reported that a large proportion of C. difficile strains isolated from the same study population showed resistance to rifampicin [9].
Although HIV/AIDS was a frequent comorbidity, this group of patients had $42\%$ lower odds of developing HO-CDI. However, we noted that a majority of these patients were receiving concomitant antiretroviral therapy and consequently this would have an effect in reducing the risk of CDI as previously described [47]. In addition, future studies should provide more insights on the risk of developing HO-CDI in patients with HIV/AIDS as some studies have suggested a possible association between pre-existing HIV/AIDS and CDI in both adults and children [48, 49].
The present study failed to establish a correlation between HO-CDI and underlying peptic ulcer disease, liver disease, inflammatory bowel disease, low levels of vitamin D (rickets), solid tumor without metastasis, and weight loss (malnutrition) as previously described [17, 50–54]. Future clinical studies should explore these associations considering the possibility of increased antibiotic use and hospital admission.
In both CCI and ECI classification, there was sufficient evidence ($p \leq 0.001$) to reject the null hypothesis and conclude that there is an association between the CCI and ECI comorbidity scores and the primary outcome of HO-CDI. Despite their differences in weighting and number of comorbidities, both models performed well with minor differences in their validation values. It is interesting that patients with CCI ≥3 were no more likely to have HO-CDI than those with scores of 0. A possible reason for this might be that patients in this group are regarded as having moderate and severe comorbidity levels raising the likelihood that their diarrhoea is due to causes other than CDI. In testing for goodness-of-fit, the adjusted/complex model was shown to fit the dataset significantly better (p value < 0.0001) for both the CCI and ECI groupings. However, most remarkable observation from the analysis was that the Elixhauser classification emerged as a better predictor than the Charlson classification in both the unadjusted (Pseudo R-squared 7.89 vs 6.09) and adjusted models (Pseudo R-squared 27.55 vs 27.04). These findings are consistent with previous studies where the Elixhauser grouping was reported to be a better predictor of an outcome while compared to the Charlson grouping, albeit by a small margin [55].
Although solid evidence linking comorbidities with HO-CDI was observed, this study, however, had some limitations. First, the study participants were enrolled from a single centre and hence the findings may not be generalized to other healthcare facilities within the country and therefore future studies should consider a multicentre approach. Secondly, data collection relied mostly on what was indicated in each patient’s file, which could have contributed to underreporting of some conditions. Finally, for some conditions like diabetes, it was not classified as complicated or uncomplicated as required by ICD-10-CM.
In summary, chronic obstructive pulmonary disease, chronic kidney disease, anemia, diabetes, and hypertension were significant predictors of HO-CDI in our setting. Therefore, it is recommended that patients with these co-morbidities be identified early and, where possible, procedures implemented that serve to limit potential exposure to other patients with CDI and/or environments likely to be contaminated by spores. Interventions aimed at restoring and maintaining the resident gut microbiota may also be beneficial in this patient population. The study also suggests that stratification of comorbidities according to ECI rather than CCI would further help to identify at-risk patients since it was a better predictor for HO-CDI.
## References
1. Magill SS, Edwards JR, Bamberg W, Beldavs ZG, Dumyati G, Kainer MA. **Multistate point-prevalence survey of health care-associated infections**. *New England Journal of Medicine.* (2014.0) **370** 1198-208. DOI: 10.1056/NEJMoa1306801
2. Eze P, Balsells E, Kyaw MH, Nair H. *Journal of Global Health* (2017.0) **7**
3. Furuya-Kanamori L, Stone JC, Clark J, McKenzie SJ, Yakob L, Paterson DL. **Comorbidities, Exposure to Medications, and the Risk of Community-Acquired Clostridium difficile Infection: A Systematic Review and Meta-analysis.**. *Infection Control & Hospital Epidemiology* **36** 132-41. PMID: 25632995
4. Ticinesi A, Nouvenne A, Folesani G, Prati B, Morelli I, Guida L. **Multimorbidity in elderly hospitalised patients and risk of Clostridium difficile infection: A retrospective study with the Cumulative Illness Rating Scale (CIRS).**. *BMJ Open* (2015.0) **5**. DOI: 10.1136/bmjopen-2015-009316
5. Harris AD, Sbarra AN, Leekha S, Jackson SS, Johnson JK, Pineles L. **Electronically available comorbid conditions for risk prediction of healthcare-associated clostridium difficile infection.**. *Infection Control and Hospital Epidemiology* (2018.0) **39** 297-301. DOI: 10.1017/ice.2018.10
6. Abou Chakra CN, Pepin J, Sirard S, Valiquette L. **Risk factors for recurrence, complications and mortality in Clostridium difficile infection: A systematic review.**. *PLoS ONE.* (2014.0). DOI: 10.1371/journal.pone.0098400
7. Negrut N, Bungau S, Behl T, Khan SA, Vesa CM, Bustea C. **Risk Factors Associated with Recurrent Clostridioides difficile Infection. Healthcare**. (2020.0) **8** 352
8. Davies K, Lawrence J, Berry C, Davis G, Yu H, Cai B. **Risk Factors for Primary Clostridium difficile Infection; Results From the Observational Study of Risk Factors for Clostridium difficile Infection in Hospitalized Patients With Infective Diarrhea (ORCHID).**. *Frontiers in Public Health* (2020.0) **8** 293. DOI: 10.3389/fpubh.2020.00293
9. Mutai WC, Mureithi MW, Anzala O, Revathi G, Kullin B, Burugu M. **High Prevalence of Multidrug-Resistant Clostridioides difficile Following Extensive Use of Antimicrobials in Hospitalized Patients in Kenya.**. *Frontiers in Cellular and Infection Microbiology* (2021.0) **10** 1. DOI: 10.3389/fcimb.2020.604986
10. Oyaro MO, Plants-Paris K, Bishoff D, Malonza P, Gontier CS, DuPont HL. **High rate of Clostridium difficile among young adults presenting with diarrhea at two hospitals in Kenya.**. *International Journal of Infectious Diseases.* (2018.0) **74** 24-8. DOI: 10.1016/j.ijid.2018.06.014
11. Plants-Paris K, Bishoff D, Oyaro MO, Mwinyi B, Chappell C, Kituyi A. **Prevalence of Clostridium difficile infections among Kenyan children with diarrhea**. *International Journal of Infectious Diseases.* (2019.0) **81** 66-72. DOI: 10.1016/j.ijid.2019.01.024
12. Valderas JM, Starfield B, Sibbald B, Salisbury C, Roland M. **Defining comorbidity: Implications for understanding health and health services.**. *Annals of Family Medicine* (2009.0) **7** 357-63. DOI: 10.1370/afm.983
13. Vincent C, Miller MA, Edens TJ, Mehrotra S, Dewar K, Manges AR. **Bloom and bust: Intestinal microbiota dynamics in response to hospital exposures and Clostridium difficile colonization or infection.**. *Microbiome* (2016.0) **4**
14. Khanafer N, Toure A, Chambrier C, Cour M, Reverdy ME, Argaud L. **Predictors of Clostridium difficile infection severity in patients hospitalised in medical intensive care**. *World J Gastroenterol* (2013.0) **19** 8034-41. DOI: 10.3748/wjg.v19.i44.8034
15. Stevens V, Concannon C, van Wijngaarden E, McGregor J. **Validation of the chronic disease score-infectious disease (CDS-ID) for the prediction of hospital-associated clostridium difficile infection (CDI) within a retrospective cohort.**. *BMC Infectious Diseases* (2013.0) **13** 1-8. DOI: 10.1186/1471-2334-13-1
16. Depestel DD, Aronoff DM. **Epidemiology of Clostridium difficile infection.**. *Journal of Pharmacy Practice* (2013.0) **26** 464-75. DOI: 10.1177/0897190013499521
17. Lee HY, Hsiao HL, Chia CY, Cheng CW, Tsai TC, Deng ST. **Risk factors and outcomes of Clostridium difficile infection in hospitalized patients**. *Biomedical Journal.* (2019.0) **42** 99-106. DOI: 10.1016/j.bj.2018.12.002
18. Chmielewska M, Zycinska K, Lenartowicz B, Hadzik-Błaszczyk M, Cieplak M, Kur Z. **Clostridium difficile infection due to pneumonia treatment: Mortality risk models**. *Advances in Experimental Medicine and Biology* (2017.0) **955** 59-63. DOI: 10.1007/5584_2016_160
19. Jasiak NM, Alaniz C, Rao K, Veltman K, Nagel JL. **Recurrent Clostridium difficile infection in intensive care unit patients**. *American Journal of Infection Control* (2016.0) **44** 36-40. DOI: 10.1016/j.ajic.2015.08.013
20. Ramesh MS, Yee J. **Clostridioides difficile Infection in Chronic Kidney Disease/End-Stage Renal Disease**. *Advances in Chronic Kidney Disease* (2019.0) 30-4. DOI: 10.1053/j.ackd.2019.01.001
21. Kim SC, Seo MY, Lee JY, Kim KT, Cho E, Kim MG. **Advanced chronic kidney disease: A strong risk factor for Clostridium difficile infection.**. *Korean Journal of Internal Medicine* (2016.0) **31** 125-33. PMID: 26767866
22. Yang Z, Huang Q, Qin J, Zhang X, Jian Y, Lv H. **Molecular Epidemiology and Risk Factors of Clostridium difficile ST81 Infection in a Teaching Hospital in Eastern China.**. *Frontiers in Cellular and Infection Microbiology* (2020.0) **10** 578098. DOI: 10.3389/fcimb.2020.578098
23. Anders HJ, Andersen K, Stecher B. **The intestinal microbiota, a leaky gut, and abnormal immunity in kidney disease**. *Kidney International.* (2013.0) 1010-6. DOI: 10.1038/ki.2012.440
24. Vaziri ND, Wong J, Pahl M, Piceno YM, Yuan J, Desantis TZ. **Chronic kidney disease alters intestinal microbial flora**. *Kidney International* (2013.0) **83** 308-15. DOI: 10.1038/ki.2012.345
25. Ramezani A, Massy ZA, Meijers B, Evenepoel P, Vanholder R, Raj DS. **Role of the gut microbiome in Uremia: A potential therapeutic target.**. *American Journal of Kidney Diseases* (2016.0) **67** 483-98. DOI: 10.1053/j.ajkd.2015.09.027
26. Hensgens MPM, Dekkers OM, Goorhuis A, Lecessie S, Kuijper EJ. **Predicting a complicated course of Clostridium difficile infection at the bedside**. *Clinical Microbiology and Infection* (2014.0) **20** O301-8. DOI: 10.1111/1469-0691.12391
27. Malik AT, Quatman CE, Phieffer LS, Ly T v., Khan SN. **Clostridium difficile colitis in patients undergoing surgery for hip fractures: an analysis of 17,474 patients.**. *HIP International* (2020.0) **30** 22-32. PMID: 30799640
28. Yang T, Santisteban MM, Rodriguez V, Li E, Ahmari N, Carvajal JM. **Gut Dysbiosis is Linked to Hypertension**. *Hypertension* (2015.0) **65** 1331-40. PMID: 25870193
29. Silveira-Nunes G, Durso DF, LRA de O, Cunha EHM, Maioli TU, Vieira AT. **Hypertension Is Associated With Intestinal Microbiota Dysbiosis and Inflammation in a Brazilian Population.**. *Frontiers in Pharmacology* (2020.0) **11** 258. DOI: 10.3389/fphar.2020.00258
30. Robles‐Vera I, Toral M, Visitación N, Sánchez M, Gómez‐Guzmán M, Muñoz R. **Changes to the gut microbiota induced by losartan contributes to its antihypertensive effects**. *British Journal of Pharmacology* (2020.0) **177** 2006-23. PMID: 31883108
31. Jama H, Kaye DM, Marques FZ. **Population-based gut microbiome associations with hypertension the need for more detailed phenotypes**. *Circulation Research* (2018.0) **123** 1185-7. DOI: 10.1161/CIRCRESAHA.118.313792
32. Caspar M, Florin I, Thelestam M. **Calcium and calmodulin in cellular intoxication withClostridium difficile toxin B**. *Journal of Cellular Physiology* (1987.0) **132** 168-72. DOI: 10.1002/jcp.1041320124
33. Shoaei P, Shojaei H, Shirani K. **Phenotypic and genotypic characteristics of clostridium difficile isolates in patients with type 2 diabetes in Iran.**. *Infection and Drug Resistance* (2020.0) **13** 683-90. DOI: 10.2147/IDR.S225829
34. Shakov R, Salazar RS, Kagunye SK, Baddoura WJ, DeBari VA. **Diabetes mellitus as a risk factor for recurrence of Clostridium difficile infection in the acute care hospital setting.**. *American Journal of Infection Control* (2011.0) **39** 194-8. DOI: 10.1016/j.ajic.2010.08.017
35. Hung YP, Lee JC, Tsai BY, Wu JL, Liu HC, Liu HC. **Risk factors of Clostridium difficile-associated diarrhea in hospitalized adults: Vary by hospitalized duration.**. *Journal of Microbiology, Immunology and Infection* (2019.0) **54** 276-83. DOI: 10.1016/j.jmii.2019.07.004
36. Tottey W, Feria-Gervasio D, Gaci N, Laillet B, Pujos E, Martin JF. **Colonic transit time is a driven force of the gut microbiota composition and metabolism: In vitro evidence.**. *Journal of Neurogastroenterology and Motility* (2017.0) **23** 124-34. DOI: 10.5056/jnm16042
37. Piper MS, Saad RJ. **Diabetes Mellitus and the Colon. Current Treatment Options in Gastroenterology**. *Springer Science and Business Media LLC* (2017.0) **15** 460-74. DOI: 10.1007/s11938-017-0151-1
38. Eliakim-Raz N, Fishman G, Yahav D, Goldberg E, Stein GY, Zvi HB. **Predicting Clostridium difficile infection in diabetic patients and the effect of metformin therapy: a retrospective, case–control study**. *European Journal of Clinical Microbiology and Infectious Diseases* (2015.0) **34** 1201-5. DOI: 10.1007/s10096-015-2348-3
39. Bryrup T, Thomsen CW, Kern T, Allin KH, Brandslund I, Jørgensen NR. **Metformin-induced changes of the gut microbiota in healthy young men: results of a non-blinded, one-armed intervention study**. *Diabetologia* (2019.0) **62** 1024-35. DOI: 10.1007/s00125-019-4848-7
40. Thanissery R, Winston JA, Theriot CM. **Inhibition of spore germination, growth, and toxin activity of clinically relevant C. difficile strains by gut microbiota derived secondary bile acids**. *Anaerobe* (2017.0) **45** 86-100. DOI: 10.1016/j.anaerobe.2017.03.004
41. Tam J, Icho S, Utama E, Orrell KE, Gómez-Biagi RF, Theriot CM. **Intestinal bile acids directly modulate the structure and function of C. Difficile TcdB toxin**. *Proceedings of the National Academy of Sciences of the United States of America* (2020.0) **117** 6792-800. PMID: 32152097
42. Winston JA, Theriot CM. **Impact of microbial derived secondary bile acids on colonization resistance against Clostridium difficile in the gastrointestinal tract**. *Anaerobe* (2016.0) **41** 44-50. DOI: 10.1016/j.anaerobe.2016.05.003
43. Zhang Q, Hu N. **Effects of metformin on the gut microbiota in obesity and type 2 diabetes mellitus**. *Diabetes* (2020.0) 5003-14. DOI: 10.2147/DMSO.S286430
44. Lee YM, Huh KC, Yoon SM, Jang BI, Shin JE, Koo HS. **Incidence and clinical outcomes of clostridium difficile infection after treatment with tuberculosis medication**. *Gut and Liver* (2016.0) **10** 250-4. DOI: 10.5009/gnl14435
45. Choi JM, Kim HH, Park SJ, Park MI, Moon W. **Development of pseudomembranous colitis four months after initiation of rifampicin**. *Case Reports in Gastroenterology* (2011.0) **5** 45-51. DOI: 10.1159/000323753
46. Obuch-Woszczatyński P, Dubiel G, Harmanus C, Kuijper E, Duda U, Wultańska D. **Emergence of Clostridium difficile infection in tuberculosis patients due to a highly rifampicin-resistant PCR ribotype 046 clone in Poland.**. *European Journal of Clinical Microbiology and Infectious Diseases* (2013.0) **32** 1027-30. DOI: 10.1007/s10096-013-1845-5
47. Collini PJ, Kuijper E, Dockrell DH. **Clostridium difficile infection in patients with HIV/AIDS.**. *Current HIV/AIDS Reports* (2013.0) **10** 273-82. DOI: 10.1007/s11904-013-0162-z
48. Haines CF, Moore RD, Bartlett JG, Sears CL, Cosgrove SE, Carroll K. **Clostridium difficile in a HIV-infected cohort: Incidence, risk factors, and clinical outcomes**. *AIDS.* (2013.0) 2799-807. DOI: 10.1097/01.aids.0000432450.37863.e9
49. Johnston PF, Gerding DN, Knight KL. **Protection from clostridium difficile infection in CD4 T cell- and polymeric immunoglobulin receptor-deficient mice**. *Infection and Immunity* (2014.0) **82** 522-31. DOI: 10.1128/IAI.01273-13
50. Wang WJ, Hussain SA, Kim SH, Mehta P, Rubin M, Gray S. **Low vitamin D level is an independent predictor of poor outcomes in Clostridium difficile-associated diarrhea**. *Therapeutic Advances in Gastroenterology* (2014.0) **7** 14-9. DOI: 10.1177/1756283X13502838
51. Abdalla AO, Pisipati S, Elnaggar M, Rishi M, Doshi R, Gullapalli N. **Outcomes of Clostridioides difficile Infection in Patients With Liver Cirrhosis: A Nationwide Study.**. *Gastroenterology Research* (2020.0) **13** 53-7. DOI: 10.14740/gr1240
52. Nitzan O, Elias M, Chazan B, Raz R, Saliba W. **Clostridium difficile and inflammatory bowel disease: Role in pathogenesis and implications in treatment**. *World Journal of Gastroenterology* (2013.0) **19** 7577-85. DOI: 10.3748/wjg.v19.i43.7577
53. Perez-Cruz E, Sandoval-Flores MG, Luna-Camacho Y, Ortiz-Gutierrez S. **Malnutrition as a risk factor for mortality in clostridium difficile infection.**. *Clinical Nutrition* (2018.0) **37** S66
54. Furuya-Kanamori L, Wangdi K, Yakob L, McKenzie SJ, Doi SAR, Clark J. **25-Hydroxyvitamin D Concentrations and Clostridium difficile Infection: A Meta-Analysis**. *Journal of Parenteral and Enteral Nutrition* (2017.0) **41** 890-5. DOI: 10.1177/0148607115623457
55. Sharma N, Schwendimann R, Endrich O, Ausserhofer D, Simon M. **Comparing Charlson and Elixhauser comorbidity indices with different weightings to predict in-hospital mortality: an analysis of national inpatient data.**. *BMC Health Services Research* (2021.0) **21** 1. PMID: 33388053
|
---
title: 'Healthcare-seeking experiences of older citizens in Bangladesh: A qualitative
study'
authors:
- Abdur Razzaque Sarker
- Irfat Zabeen
- Moriam Khanam
- Ruckshana Akter
- Nausad Ali
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022267
doi: 10.1371/journal.pgph.0001185
license: CC BY 4.0
---
# Healthcare-seeking experiences of older citizens in Bangladesh: A qualitative study
## Abstract
Despite improvements in many health indicators in the last few decades, providing access to affordable and quality healthcare for older citizen remains a considerable challenge in Bangladesh. This study aimed to understand individuals ‘experiences regarding their healthcare-seeking, treatment cost, accessibility and coping mechanisms for the promotion of appropriate strategies to enhance the quality of life of the older citizens of Bangladesh. A qualitative descriptive approach was used in this study. A total of 27 In-Depth Interviews (IDIs) were conducted in a district in Bangladesh with older people between January and February 2020, where gender distribution was equal. Face-to-face interviews were conducted by trained and experienced interviewers regarding healthcare-seeking and accessibility, affordability, and healthcare coping strategy. Thematic analysis was conducted to analyse the data. It was found that the health condition of the older population is not satisfactory. Most of them had been suffering from several diseases such as benign tumor, chronic kidney disease, body aches, gastric ulcers for a longer period of time. The majority of the participants were suffering from multiple non-communicable diseases while diabetes and hypertension were the foremost of all diseases. This study provides insight into the challenges of managing healthcare services for older citizens in Bangladesh. Healthcare facilities were available, but high out-of-pocket payments, lack of caregivers, and time distance created a barrier to the service provision. The findings indicated that geriatric care policymakers and service providers should prioritize the older-friendly health infrastructures with affordable cost of treatment for the betterment of the health status of older citizens in Bangladesh.
## Introduction
Despite improvements in many health indicators in the last few decades, providing access to affordable healthcare remains a considerable challenge worldwide, especially for older people living in low- and middle-income countries (LMICs) like Bangladesh [1–3]. In recent years, many countries have experienced significant changes in the age composition of their population, particularly, the population aged 60 years and more has steadily increased in numbers [4]. Changing dynamics in birth and death rates, declining fertility rates, and increasing life expectancy lead to these changes in the global age structure [5]. The demographic structure is changing rapidly in Bangladesh. According to the national estimation conducted by the Bangladesh Bureau of Statistics, approximately $8\%$ of the total population is 60 years or above, which is projected to be increased to $11.5\%$ (21.526 million) by 2030 in Bangladesh [6,7]. The overall life expectancy of the population is also increasing (67.7 years in 2010 and 72.32 years in 2020) which creates both opportunities and public health challenges for Bangladesh [8]. It is expected that everyone should hold better health to live a longer life which will enable them to provide valuable economic, social as well as cultural contributions. Due to shifting disease patterns from communicable diseases (CDs) to non-communicable diseases (NCDs), various non-communicable health issues such as, sensory and cognitive impairments, dementia, physical inactivities, and even the risk of falling during daily activities have become more prevalent among the older population which requires regular health services from various healthcare facilities [9,10]. Further, the gradual rising of nuclear families makes older people more isolated and pushes them to be vulnerable with respect to physical and mental health in Bangladesh [11]. Thus, older citizens with a high risk of diseases and disabilities would put urgent demands on the health systems while *Bangladesh is* not well prepared to meet such healthcare demands. Documenting the current experiences of older people in seeking health care can inform policy decisions on improving the health system’s performance in tackling the growing health care needs of the elderly.
This study tried to document the health-related problem and care-seeking patterns of the older citizen. To understand the care-seeking pattern we followed the theoretical framework proposed by Kasl and Cobb using three important behavior such as- health behavior, illness behavior, and sick role behavior [12,13]. Kasl and Cobb discussed that the likelihood of such behavior is a function of the perceived amount of threat and the value of the behavior. The amount of threat depends on health matters, susceptibility to the disease, and the consequences of diseases while the value of the behavior often depends on the perceived probability of the action that will prevent the disease and the cost of taking the action [12,14]. Indeed, the pattern of diseases and care-seeking behavior of older citizens is heterogeneous and often more complex than that of other ages and thus requires more attention. Many older citizens are often unable to pay for these services and even out of reach of healthcare services [15–17]. Further, aging causes functional deterioration and vulnerability which triggers to increase in the household’s healthcare expenditure [18]. Albeit, the government, and related authorities are often concerned that older citizens will inevitably increase public health care expenditures in near future, public investment is still focused on maternal and child health issues in Bangladesh [19].
The life expectancy of people is increasing worldwide, and the pace of population aging is much faster than in the past. Bangladesh is also facing such challenges and would put urgent demands on the health system. Although a body of literature in Bangladesh focuses on healthcare utilization and associated costs among various population groups, there is little evidence of a study targeting older citizens regarding their views and experiences about healthcare services in Bangladeshi context [8,19–21]. To address this public health concern, this research aimed to understand individuals’ experiences of older people in Bangladesh regarding healthcare-seeking and accessibility, affordability, and coping strategy. The current study also considers the social and cultural aspects of healthcare utilization during their sickness. As such, the present study attempted to bridge the knowledge gap and shed light on emphasizing the promotion of the healthcare condition of senior citizens. The older population is now considered an important contributor to age structure; therefore, without securing their healthcare, universal health coverage (UHC) cannot be achieved.
## Study design
A qualitative descriptive approach was used to understand the experiences of older citizens in terms of their health condition and health care needs, accessibilities, treatment costs, and related factors [22–25]. The present study employed in-depth interviews (IDIs) that provide a comprehensive summary of events reported directly by older adult individuals. This approach delivers more in-depth and circumstantial evidence while providing a comprehensive summary of events in the usual language of the participants [26]. A total of 27 IDIs were conducted up to data saturation with no refusal.
## Study settings & population
This study was conducted in the Tangail district of Bangladesh, where the first ever government-initiated health protection scheme was piloted. Tangail is also nearby to the capital city of Bangladesh where various types of healthcare facilities including tertiarly and specialized healthcare are available. Further, due to minimum-distance and better communications, people can avail the healthcare services from the capital city of Bangladesh. Older citizens or the elderly population are defined as people of sixty years or more, according to the National Policy on Elderly People in Bangladesh [27]. The older citizens residing in this district for a minimum of 1 year were eligible as respondents. For each in-depth interview, one after every 3rd household was targeted and then if any older citizens were found on that very household, were selected for the interview. Additional probing that provided insights relevant to our study was also used. Repetitions within the data were noted, suggesting that the data reached saturation [28]. After interviewing these participants, the research team were satisfied that thematic saturation was reached and recruitment ceased. Therefore, a total of 27 In-Depth Interviews (IDIs) were conducted where gender distribution was equal. Face-to-face interviews were conducted by trained and experienced interviewers using specific interview guidelines.
## Data collection
Considering the convenience and availability of the participants, face-to-face individual in-depth interviews were conducted between January and February 2020. The interviews lasted between 40–60 minutes and were all conducted in Bangla- the country’s local language. The study participants were contacted via in-person visits in their households, and the objectives were described clearly. The household list was collected from the local government of Bangladesh. An interview date was fixed for them, who agreed to participate in the in-depth interview. The interviewers were well trained on the guidelines and had significant experience using qualitative data collection tools. This in-depth guideline was adopted from published qualitative literature based on healthcare utilization and barriers and then finalised considering both cultural and socio-demographic aspects of the Bangladeshi context [29]. Participants were first asked to provide written informed consent to participate, and then verbal consent was recorded at the beginning of the interviews. Each discussion between the participant and interviewer was recorded using an audio recorder with the respondents’ permission. Initially, the questions were broad, according to the guidelines, and later probing was applied to achieve more detailed responses.
## Data analysis
A content analysis was conducted to analyse the data. To ensure reliability in analyzing the interviews, researchers independently checked all the interviews. Transcriptions were made from each of the recorded interviews by the interviewers in their native language (Bangla), and the transcripts were read with field notes for an overall understanding. All transcripts were translated into English by a bilingual translator directly involved in the study. Later, the transcriptions and translations were checked twice by other investigators. Next, the transcriptions were structured using the framework method proposed by Gale et al. and categorized the answers into five themes [30]. Thematic analysis was conducted to ensure methodological accuracy and transparency of qualitative data analysis. To carry out a thematic analysis, several phases were followed according to the methodology adopted by Braun and Clarke, which included familiarizing the data, generating initial codes, defining and reviewing themes, and drawing up general interpretations [31]. Data were manually coded according to meaningful statements (issues, highlights, concerns, and accomplishments) about older health-related experiences and then categorized by the team investigators. Contents were compared across codes, and key concepts were recognized where core themes were identified. A qualitative analysis expert cross-checked the themes for common agreement and refined the identified themes. The investigators finally listed specific themes based on the guidelines (S1 Text) and code categories that included participants’ perceptions, personal beliefs, and understanding. Finally, an evaluation of the themes with a re-reading of the interviews was performed to ensure that the participants’ insistence, meaning, and perception were precisely captured. All explications provided by the respondents were categorized into five distinct themes that described the experiences and expectations of older citizens about healthcare accessibilities and the future scope of improvement.
## Ethics statement
This study was performed in line with the principles of the Declaration of Helsinki. The research was approved by the institutional review board at Bangladesh Institute of Development Studies, IRB Protocol#: PSD/$\frac{01}{2019.20}$/03/REF/HSBAECB. Participants were informed in advance that their participation was voluntary and that all information provided would remain confidential. All study participants provided written consent prior to data collection.
## Results
A total of 27 interviews were conducted up to data saturation with no refusal. Table 1 presents the basic demographic characteristics of the participants. The majority of the respondents were aged between 60–65 years ($59\%$), and most of them ($56\%$) had no formal education. $18\%$ of the total respondents had a monthly household income of 10,000−15,000 BDT, while $48\%$ were earning less than 10,000 BDT.
**Table 1**
| Characteristics | Frequency | Percentage |
| --- | --- | --- |
| Gender | | |
| Female | 13.0 | 48.4 |
| Male | 14.0 | 51.85 |
| Age in Years | | |
| 60–65 | 16.0 | 59.26 |
| 66–69 | 5.0 | 18.52 |
| ≥70 | 6.0 | 22.22 |
| Education | | |
| No formal education | 15.0 | 55.56 |
| Up to primary | 7.0 | 25.93 |
| Up to secondary | 4.0 | 14.81 |
| Higher secondary | 1.0 | 3.7 |
| Monthly Household Income | | |
| Less than 10,000 | 13.0 | 48.15 |
| 10,000–15,000 | 5.0 | 18.52 |
| 15,001–20,000 | 2.0 | 7.41 |
| More than 20,000 | 5.0 | 18.52 |
| Don’t Know | 2.0 | 7.41 |
| Total | 27.0 | 100.0 |
All elucidations provided by the participants were identified under five themes: health condition, care-seeking pattern, expectations, treatment cost and out-of-pocket cost issues and coping strategiesThese themes are presented using the participants’ direct voice to explore the in-depth context and detailed meaning of specific themes that optimally described their experiences. Theme description is quoted in italics (Table 2).
**Table 2**
| Theme | Sub-Theme |
| --- | --- |
| 1. Assessment of health status | 1.1 Current Health Status |
| 1. Assessment of health status | • “Almost all the respondents’ health condition were poor” |
| 1. Assessment of health status | 1.2 Communicable vs. Non-Communicable Diseases |
| 1. Assessment of health status | • “Communicable diseases are mostly common for each respondent” |
| 1. Assessment of health status | 1.3 Long-term illness |
| 1. Assessment of health status | • “Carrying illness till death” |
| 2. Healthcare seeking pattern | 2.1 Decision of choosing health facility |
| 2. Healthcare seeking pattern | • “Prone to seek health care from private clinic/hospital” |
| 2. Healthcare seeking pattern | 2.2 Preferable option for healthcare |
| 2. Healthcare seeking pattern | • “District public hospitals are preferable than sub-district level health facilities” |
| 3. Expectation from service provider | 3.1 Quality improvement |
| 3. Expectation from service provider | • “Quality health care is first priority” |
| 3. Expectation from service provider | 3.2 Affordability and accessibility |
| 3. Expectation from service provider | • “Affordability and Transportation convenience play crucial role for seeking care” |
| 3. Expectation from service provider | 3.3 Other issues |
| 3. Expectation from service provider | “Quality of service provider and environment is also important in deciding care seeking” |
| 4. Out of pocket cost is one of the barrier in accessing healthcare | 4.1 Treatment cost |
| 4. Out of pocket cost is one of the barrier in accessing healthcare | • “Consultation fees should be reduced for the poor older people” |
| 4. Out of pocket cost is one of the barrier in accessing healthcare | • “Surgery and operation costs are not affordable” |
| 5. Coping strategies for the health care expenditure | 5.1 Consequences |
| 5. Coping strategies for the health care expenditure | • “Managing health expenses is a unbearable burden for the older” |
| 5. Coping strategies for the health care expenditure | 5.2 Suggestions from the respondents |
| 5. Coping strategies for the health care expenditure | • “Multiple suggestions are carried out to mitigate healthcare expenditure” |
## Theme 1: Health condition of older people
Participants recounted their experiences during sickness, describing diverse individual illnesses and varied patterns of disease symptoms presenting throughout the body. Almost all of the respondents mentioned that their health condition was abysmal. The majority of the respondents reported that they had been suffering from several diseases such as a benign tumor, chronic kidney disease, body aches, and gastric ulcer for a long time. Some of them stated that they could not even walk or participate in regular work. Most participants suffered from multiple non-communicable diseases (NCDs), while diabetes and hypertension were the foremost of all diseases. The health condition of older people was so fragile that some of them were not in ambulatory condition. Eye infections, benign tumors, chronic kidney disease, body aches, and gastric ulcers were some major causes of illness. One-third of them had been suffering from those diseases for over 10 years. One male respondent aged [64] stated, “My health condition is declining as my heart is weak, the bulb is damaged and there is a hole in my rectum. Though I had an operation 35 years ago, there is no sign of improvement yet.” Most of the respondents explained their miserable health condition by stating that they had been suffering from such a devastating illness for a long time. Indeed, some came to know about their diseases after a major clinical operation (e.g. tumor, eye diseases, and diabetes) and even after experiencing a stroke. Consequently, they are now living with continuous medication which is often unaffordable and are worried about future medication in this regard. One older respondent [68] expressed, “I have high blood pressure, gastric ulcers, and pains in a different organ. A long time ago, maybe 2 years or more, I had a stroke but I had no sense at that time. Since then, I have been on medication and I have to carry on till death.” The other respondents [69] stated, “Many of my friends and relatives are suffering from hypertension and diabetes and some of them advised me to get examined by a physician for diabetes and hypertension, and then the doctors confirmed me about these diseases.” It was almost eight [8] years ago and I am still suffering.” With ageing, people are more likely to experience multiple health conditions which leads them to seek healthcare services which also differ in several aspects. Thus, our second theme features the pattern of care-seeking behaviour of this group.
## Theme 2: Healthcare-seeking pattern
Healthcare-seeking pattern refers to the sequence of remedial actions that individuals undertake to rectify perceived ill health which is essential to provide need-based healthcare delivery, particularly, for the older population, and to make the healthcare system more pro-poor. Almost every respondent was affirmative about the private clinic or hospital when they had been asked about the source of healthcare. Indeed, pharmacies were their first choice due to the convenient and easily approachable nature. A woman [65] replied, “I take my medicines regularly from pharmacies because my doctor advised me to take my medicines regularly, and most of the pharmacies are near to my door.” The other respondent [70] said that “the drug seller is very well known to me and I can trust easily.” Older people prioritized private clinics/hospitals as they believed quality services are available there than any other institutions. Some of them were used to going to the private facilities because their relatives or close persons worked there by which they could avail quality services at a low-cost (at least without doctor fees). One of the older citizens stated [65], “a doctor from Mymensingh Sadar consults with patients in Life Care Hospital (a private hospital). He was one of my relatives and I used to go to him for care. Three months ago, I have consulted with him and he gave me certain medications including antibiotics. He also suggested about going to another doctor of the same private institution for further treatment.” In most of the cases, the private healthcare facilities were either adjacent to their houses or at the center of their city where they can easily go with existing transportation. One of the respondents [67] replied, “Lack of older-friendly transportation is a big problem for me, as I cannot sit properly with the available transportation and also I have no extra family member who could help me as caregiver.” Though the majority of the study participants preferred private clinics, they often visited district public hospitals for regular checkups for chronic diseases like diabetes, asthma, heart diseases, eye operation, and kidney operation as public hospitals are highly subsidized in Bangladesh. Indeed, we observed that the participants choose district public hospitals when it comes to an emergency situation. Very few numbers of them went to Dhaka, the capital of Bangladesh for consulting specialized doctors and for better care. On the other hand, the participants were often unwilling to visit sub-district level public hospitals due to the lack of medical personnel and medical technologies. Nevertheless, very few of them visited these places and had to buy medicines from the pharmacy or other sources due to the scarcity of medicines in those facilities. A statement from a respondent [70], “at first I went to Modhupur Hospital (sub-district level hospital) for treating asthma-related problems, but that wasn’t solved. That’s why, I visited Mymensingh Sadar Hospital (a district-level public hospital), about 8–9 years ago. They prescribed me several medicines after several lab tests and diagnoses. Now my disease is under control to some extent. But, when I stop taking medicine, it further comes out.” Indeed, the distance and the availability of healthcare facilities plays a significant role behind the care-seeking pattern. However, people visits the healthcare centres with expectations to get quality services at an affordable cost. So, we tried to seek for the answers regarding this issue which has been depicted in the next theme.
## Theme 3: Expectations from the health facilities
People usually choose a healthcare facility for their convenience or better quality service which is also cost-effective from a household perspective. The easy and accessible transportation system, amenities, quality service, availability, and accessibility of healthcare are the prerequisites of healthcare services. The majority of our respondents sought healthcare from the private clinics/hospitals because they believed that they could get at least quality healthcare services although there was a very higher treatment cost in the private clinics/hospitals. The quality of services and associated amenities were not satisfactory in sub-district level hospitals according to their observation, therefore, they often sought care at district level hospitals and even shift to private hospitals. A respondent [65] replied, “I normally sought care from a private clinic because in nearby Modhupur Hospital (a sub-district level hospital) there were no services and scope of my treatment (e.g. diabetes). Therefore, the authority suggested me visit a private clinic. However, that particular private clinic suggested me to admit to the Tangail Diabetic Hospital (a specialized diabetes hospital).” The older citizens often sought healthcare from the nearest facilities which they could reach easily with the least cost and time. Moreover, sometimes it cost more than expected if the distance of the healthcare facilities were relatively longer. One-third of the respondents preached that financial issues were the main obstacles to their treatment care. A respondent [67] stated, “I prefer health facilities (private clinic) that are adjacent to my house. However, when I visited a nearby sub-district level hospital, I found nobody was there; not any nurse or even no doctor was there. My time and money both were wasted. As I have to spend my money whether I sought care from either private or public hospital, therefore, nearby private facilities are a better call for me.” Another respondent [64] told us, “I have a relative doctor in a private clinic who doesn’t accept any consultation fees from me. Moreover, if I have to diagnose anything, he manages discounts for me. Most of all, waiting time is very little in there.” Overall, the environment and the facilities of the health centers were not satisfactory even for the private clinics. However, some private clinics had the practice of maintaining cleanliness and better quality. Public hospitals like Upazila Health Complex (sub-district level) lagged behind in these contexts. Most of the respondents replied that the medicine ward was not clean at all and the toilets and floors were not washed daily. Moreover, there were no sufficient seats for the patients, and doctors hardly visit patients. Sometimes the caregivers and the attendants had to take shelter on the floor, and special care for older patients is unavailable.
## Theme 4: Out-of-pocket (OOP) cost—a barrier to healthcare
In Bangladesh, the high cost of medical expenses creates a significant barrier while seeking care. As a result, we wanted to highlight this burning issue and incorporated it as one of our core themes. OOP expenditure includes any payment related to medical fees, purchase of medicines (prescribed or not), user fees for care, and payments for equipment and diagnostic tests [32]. Out-of-pocket cost is generally higher for older people because morbidity complications are higher for older people compared to others. Sometimes the price of drugs becomes excessively high for them. One of them [68] replied, “From the pharmacy, I came to know the price of my drugs is too high, therefore I have purchased only half of the listed medicine and I need to wait for further arrangement of money.” *In a* public hospital, the treatment cost is often shared with households as public hospitals are highly subsidized in Bangladesh. However, all treatment costs are incurred by households if someone seeks care from private facilities. Some of the respondents replied that they could not seek care from a specialized doctor because of the high consultation fee which was unaffordable for them. Every time they visited the private facilities, they had to spend at least Bangladeshi Taka (BDT) 500–700 as consultation fees. Indeed, they also had to spend on other categories like medicines, diagnostic tests, hospital charges, bed fees, etc., which appeared to be a high financial burden for them. Most of the time, the older people did not go to the hospital or take any services if the cost were high. Whether it is a public or private hospital, the in-patient cost is pretty similar for both. Indeed, various surgery-related cost was higher in private facilities, which often becomes a catastrophic health burden for them. One of the respondents stated [68], “The treatment cost is relatively higher where I usually visit for my health-related problems. I had to spend forty thousand BDT for only five days when I had a medical operation.” Another respondent [66] stated, “Sometimes I face financial catastrophes. Sometimes I just restrain myself from going to the doctor as my family is very needy. Once I was suffering from a cataract but couldn’t receive the treatment due to financial constraints.” OOP healthcare expense often absorbs a large proportion of the total household budget which can lead to catastrophic health expenditure (CHE). In many cases, people are forced to cut their total household consumption or, to take out loans, or to take financial help from others just to meet these health expenses which are considered as the major coping mechanisms.
## Theme 5. Coping strategies for the healthcare expenditure
Various coping strategies were observed for mitigating healthcare expenditure. During the in-person interviews, we found that the older citizens mainly met their healthcare expenditure through family funds, financial support from other family members or friends, borrowing, support from others, and even selling assets. One of them [67] replied, “Most of the time, my younger son transferred money using the mobile app and then I got to purchase the medicines from pharmacies.” The other respondent [69] stated, “I received money from my elder daughter whenever I became sick.” Two-thirds of the older people reported having financial difficulties during healthcare payments. Indeed, some people had to take loans and bear the debt burden for longer. Some of them worked in other people’s homes to pay the debt. Meanwhile, a few of them could not seek medical care due to financial constraints. Even if they had managed medical care, they could not buy medicines and administer them properly. Further, some older citizens solely rely on their children’s wealth. For this reason, they sometimes had to wait longer for their children’s compliance. A respondent, aged 71, stated that “I have huge financial problems. I have to go to Upazila Health Complexes by rickshaw. After buying a ticket by *Bangladeshi taka* 5, I have to take medicine from outside the hospital. For mitigating treatment costs, I had to borrow three lac *Bangladeshi taka* from the bank. Now I am repaying the money by working in other’s houses which is difficult for me.” This study tried to capture important suggestions from older citizens regarding healthcare management and treatment care in public hospitals. However, two-thirds of the respondents mentioned various issues related to the hospital management and other support systems; such as the need for separate and furnished waiting rooms where they could easily sit down, ensuring availability of health personnel including physicians, nurses for older care, $\frac{24}{7}$ hospital services, robust referral systems, quality healthcare services, older-friendly accessibilities. They also strongly advised on improving the cleanliness of the facilities and ensuring the availability of medical equipment for investigations in public hospitals as they can’t afford many diagnostic tests in private sectors. One of the respondents, aged 68, stated, “In hospitals, it would be better if we get healthcare services in the separate chamber/ segments where only older patient would be treated; such as establishing an older friendly health unit where an older citizen would go and get treatment faster than other age groups. Currently, many hospitals have insufficient doctors, and nurses than needed, so the number of doctors and nurses has to be increased and those who are already recruited have to do their duty properly. Doctor visit fees should be further reduced, and the government should be strict about it. The price of the drug should be written on the packet. Poor people should be given a card with which they can get a $50\%$ discount on drugs and free medicines from the government hospital. People are not getting exactly what the government is providing; they need to be monitored.”
## Discussion
Improving health and well-being is a global priority in the latest sustainable development goals (SDGs); SDG-3 focuses exclusively on ensuring healthy lives and promoting well-being for all, regardless of age. As the share of the aged population has been growing rapidly and they usually suffer from various chronic diseases, it is crucial to identify their current experiences in seeking health care so that appropriate strategies could be designed to improve their health and well-being. This study tried to capture the individuals’ experiences regarding healthcare-seeking and accessibility, affordability, and coping strategy for older citizens in Bangladesh.
Health status is often measured through the presence of any disabilities that limit full participation in activities. The overall health status of older citizens is found to be poor in Bangladesh. It might be seen that all of the older citizens suffer from various illnesses and disabilities, such as hypertension, gastric ulcer, body aches, diabetes, flu/cough etc. Like other settings, Bangladesh has been experiencing a growing burden of NCDs and $66\%$ of the older people suffers from various NCDs [33]. Several studies have investigated regarding this issues in near-by countries. For example, NCDs contributed to $70\%$ of mortality and morbidity among the population in Bhutan, $50\%$ of the total health burden in terms of mortality in Nepal, $75\%$ of total deaths, and $62\%$ of mortality in India [34]. Due to rapid urbanization, unhealthy diet, increased life expectancy, and lifestyle changes, the rate of cardiovascular diseases, including hypertension, has increased over the years [35]. The prevalence of hypertension in Bangladesh has increased from $17.9\%$ in 2010 to $21\%$ in 2018 [36,37]. Diabetes, the other overwhelming disease, is significantly higher in older than in young people that we have found in our study [38]. People over 65 years are more likely to get affected by gastric ulcers or gastrointestinal disorders, and more than $80\%$ of the death caused by gastrointestinal disorders are happened to be the people of this age group [39]. Moreover, eye infections and body ache called musculoskeletal problems are also widespread for older people in Bangladesh. Besides this, we observed the prevalence of comorbidities is high among older people as found in an earlier study and they require more specialized care [40]. Indeed, the public investment is still focused on maternal and child health issues [41,42]. Therefore, improvements to older-friendly health infrastructure are necessary to improve the health of older citizen.
We observed that the distance of healthcare facilities and lack of an older-friendly transportation system was the vital issues whicle seeking care. Due to physical limitations, older people often faced difficulties for choosing to their desired health facilities; as the communication barriers still exist in Bangladesh [20]. Our study revealed that, overall, hospitals’ cleanliness and surrounding environment is not satisfactory for them. Although few private facilities now maintain cleanliness and relatively better quality, the primary and secondary public hospitals like Upazila Health Complex (sub-district level hospital) lagged in this context, and thus older people are often unwilling to go to such public hospitals that we found during a discussion with respondents [43]. We observed that most of the participants preferred private healthcare facilities, although such care is unaffordable for poor and marginalized people in Bangladesh [43]. The current study also observed that some of the respondents did not seek health care due to high perceived out of pocket costs. There is evidence from developing countries including Bangladesh that due to financial barriers older people usually do not seek care [29]. In this regard, the government should take the initiative to devise special free out-door service units for providing quality care in public hospitals and special free transport services for the older citizen. Health care facilities should focus on improving the quality of health services and strengthening the health care delivery system with referral networks as the existing services are quietly insufficient.
There is some evidence that households often spend more on older care due to various disease complexities and extra care. A study conducted in neighborhood of India found that households’ older member’s monthly per capita healthcare expenditure is 3.8 times higher than younger ones, which is catastrophic for many households [44]. From this study, it seems that older citizens often face financial crises for seeking care which hurts the mental health of older adults [45]. This is also true for Bangladeshi older people as OOP cost bears one of the lion-share for healthcare financing in Bangladesh which is increasing alarmingly. The current study also observed that the older people use different coping strategies like family funds, support from relatives, borrowing and selling assets for their health care expenditure. A study in rural Bangladesh also identified similar coping strategies [29]. Universal social security programs such as universal old-age pension, and old age allowance could be initiated so that they could be free from anxieties and depressions in later life. A similar program was already initiated in many neighborhood countries such as old-age allowance in Thailand, social pensions in Vietnam, senior citizens’ allowance in Nepal etc [46]. The Government of Bangladesh has several Social Safety Net (SSN) Programmes from which older people get some direct and indirect benefits. The government took few initiatives for the older people, such as the pension system, retirement benefits and other initiatives under various programmes. The ‘Old Age Allowance’ policy was introduced for the poor older population of the nation. Still, the coverage is not yet entirely satisfactory as a large number of people from the older population remains outside the ambit of these programmes. Ministry of Health and Family Welfare of Bangladesh still has room for developing new strategies and revising existing policies to address this older age group. Indeed, social health insurance can therefore be viewed as an essential component of financial protection as it aims to make healthcare affordable and accessible to all older citizens.
This research has several limitations. Since the participation was voluntary, a certain bias could not be completely avoided. Further, the findings cannot be generalized to other populations in Bangladesh as a nationwide survey should be required for such interpretations. Nevertheless, given the focused nature of guidelines and the consistency of the data, this study did provide vital information about older healthcare. Again, recall bias might be associated, considering the care-seeking pattern, cost of treatment and even the reporting of specific diseases. Recall bias and incorrect reporting may be underestimating or overestimating the actual situation. All of the above represents possible avenues for further research on this topic.
## Conclusion
With the increase in older people, the health burden is also on the rise, but still there is inadequate data on health care seeking experiences of older population in Bangladesh. There must be strategic implementation of policies focusing on the barriers and shortfalls, which prevent them from seeking health care. This study has documented the experiences of older people regarding healthcare-seeking and accessibility, affordability, and healthcare coping strategy. The study findings suggest that the improvement to older-friendly health infrastructure is necessary and universal pension schemes and social health protection schemes should be introduced to improve the health of older citizens in Bangladesh.
## References
1. 1World Health Organization. World Health Statistics: Monitoring Health for SDGs. 2016.. *World Health Statistics: Monitoring Health for SDGs* (2016.0)
2. Goeppel C, Frenz P, Grabenhenrich L, Keil T, Tinnemann P. **Assessment of universal health coverage for adults aged 50 years or older with chronic illness in six middle-income countries**. *Bull World Health Organ* (2016.0) **94** 276-285C. DOI: 10.2471/BLT.15.163832
3. Goeppel C, Frenz P, Tinnemann P, Grabenhenrich L. **Universal health coverage for elderly people with non-communicable diseases in low-income and middle-income countries: a cross-sectional analysis**. *The Lancet* (2014.0) **384** S6. DOI: 10.1016/S0140-6736(14)61869-9
4. Nagarajan NR, Teixeira AAC, Silva ST. **The impact of an ageing population on economic growth: an exploratory review of the main mechanisms.**. *Anal Soc* (2016.0) **51** 4-35
5. Bloom DE, Canning D, Fink G. **Implications of population ageing for economic growth.**. *Oxf Rev Econ Policy* (2010.0) **26** 583-612. DOI: 10.1093/oxrep/grq038
6. 6BBS. Report on the Households Income and Expenditure Survey 2016. Dhaka, Bangladesh; 2019.
7. 7BBS. Population projection of Bangladesh- Dynamics and Trends 2011–2061. Dhaka, Bangladesh: Bangladesh Bureau of Statistics, Ministry of Planning, Government of The People’s Republic Of Bangladesh; 2015.. *Population projection of Bangladesh- Dynamics and Trends 2011–2061* (2015.0)
8. Rahman MS. **Aging and Negligence in Bangladesh.**. *J Gerontol Geriatr Res* (2017.0) **06** 10-11. DOI: 10.4172/2167-7182.1000423
9. Sarker AR, Khanam M. **Socio-economic inequalities in diabetes and prediabetes among Bangladeshi adults**. *Diabetol Int* (2021.0). DOI: 10.1007/s13340-021-00556-9
10. Mathers CD, Loncar D. *Updated projections of global mortality and burden of disease, 2002–2030: data sources, methods and results.* (2006.0)
11. Sarker AR. **Health-related quality of life among older citizens in Bangladesh**. *SSM—Mental Health* (2021.0) **1** 100031. DOI: 10.1016/j.ssmmh.2021.100031
12. Kasl S v, Cobb S. **Health Behavior, Illness Behavior, and Sick-Role Behavior.**. *Arch Environ Health* (1966.0) **12**. DOI: 10.1080/00039896.1966.10664421
13. Kasl S v., Cobb S. **Health behavior, illness behavior, and sick role behavior. I. Health and illness behavior**. *Archives of environmental health* (1966.0). DOI: 10.1080/00039896.1966.10664365
14. Rosenstock IM. **The Health Belief Model and Preventive Health Behavior**. *Health Education & Behavior* (1977.0) **2**. DOI: 10.1177/109019817400200405
15. Acharya S, Ghimire S, Jeffers EM, Shrestha N. **Health Care Utilization and Health Care Expenditure of Nepali Older Adults**. *Front Public Health.* (2019.0) **7** 1-10. DOI: 10.3389/fpubh.2019.00024
16. Yunus NM, Abd Manaf NH, Omar A, Juhdi N, Omar MA, Salleh M. **Determinants of healthcare utilisation among the elderly in Malaysia.**. *Institutions and Economies* (2017.0) **9** 117-142
17. Richard C, Carol P, Miqdad A, Rosalind R. **Socio‐Economic Inequalities in Health Care in England.**. *Fisc Stud.* (2016.0) **37** 371-403. DOI: 10.1111/j.1475-5890.2016.12109
18. Sarker AR, Mahumud RA, Sultana M, Ahmed S, Ahmed W, Khan JA. **The impact of age and sex on healthcare expenditure of households in Bangladesh.**. *Springerplus* (2014.0) **3** 435. DOI: 10.1186/2193-1801-3-435
19. Hossain T, Akhtar MdT. **The Elderly Care Services and their Current Situation in Bangladesh: An Understanding from Theoretical Perspective.**. *Journal of Medical Sciences* (2006.0) **6** 131-138
20. Abdullah JM, Ahmad MM, Saqib SE. **Understanding accessibility to healthcare for elderly people in Bangladesh.**. *Dev Pract* (2018.0) **28** 552-561. DOI: 10.1080/09614524.2018.1450845
21. Uddin MA, Soivong P, Lasuka D, Juntasopeepun P. **Factors related to quality of life among older adults in Bangladesh: A cross sectional survey.**. *Nurs Health Sci* (2017.0) **19** 518-524. DOI: 10.1111/nhs.12385
22. Fry C, Keith J. *New methods for old-age research: strategies for studying diversity.* (1986.0) **302**
23. Biswas P, Nahar Kabir Z, Nilsson J, Zaman S. **Dynamics of Health Care Seeking Behaviour of Elderly People in Rural Bangladesh**. *International Journal of Ageing and Later Life* (2006.0) **1** 69-89. DOI: 10.3384/ijal.1652-8670.061169
24. Hutchinson SA. *Taking Stock* (2001.0) **11** 505-521. DOI: 10.1177/104973201129119280
25. Pope C, Mays N. **Qualitative Research: Reaching The parts other methods cannot reach: An introduction to qualitative methods in health and health services research**. *Bmj* (1995.0) **311** 42. DOI: 10.1136/bmj.311.6996.42
26. Colorafi KJ, Evans B. *Qualitative Descriptive Methods in Health Science Research. Health Environments Research and Design Journal* (2016.0) **9** 16-25. DOI: 10.1177/1937586715614171
27. Eva SS. *Rights of Elderly in Bangladesh. The Daily Observer. 13 Oct* (2014.0)
28. Cresswell JW, Plano Clark VL. *Designing and conducting mixed methods research* (2011.0) 6
29. Hossain SJ, Ferdousi MtJ, Bakkar SMdA, Tipu SMMU, Qayyum MA, Laskar MS. **Self-reported health problems, health care seeking behaviour and cost coping mechanism of older people: Implication for primary health care delivery in rural Bangladesh.**. *J Family Med Prim Care.* (2019.0) **8** 1209-15. DOI: 10.4103/jfmpc.jfmpc_162_18
30. Gale NK, Heath G, Cameron E, Rashid S, Redwood S. **Using the framework method for the analysis of qualitative data in multidisciplinary health research.**. *BMC Med Res Methodol.* (2013.0) 1-8. DOI: 10.1186/1471-2288-13-117
31. Braun V, Clarke V. **Using thematic analysis in psychology.**. *Qual Res Psychol* (2006.0) **3** 77-101. DOI: 10.1191/1478088706qp063oa
32. Van Doorslaer E, O’Donnell O, Rannan-Eliya RP, Somanathan A, Adhikari SR, Garg CC. **Effect of payments for health care on poverty estimates in 11 countries in Asia: an analysis of household survey data**. *Lancet* (2006.0) **368** 1357-1364. DOI: 10.1016/S0140-6736(06)69560-3
33. Alam N, Chowdhury HR, Bhuiyan MA, Streatfield PK. **Causes of death of adults and elderly and health care seeking before death in rural Bangladesh.**. *J Health Popul Nutr* (2010.0) **28** 520-528. DOI: 10.3329/jhpn.v28i5.6161
34. 34WHO. National NCDs action plan. In: Noncommunicable Disease- SOUTH EAST ASIA. 24
Jul
2022.. *National NCDs action plan. In: Noncommunicable Disease- SOUTH EAST ASIA* (2022.0)
35. Chowdhury MAB, Uddin MJ, Haque MR, Ibrahimou B. **Hypertension among adults in Bangladesh: Evidence from a national cross-sectional survey**. *BMC Cardiovasc Disord* (2016.0) **16** 1-10. DOI: 10.1186/s12872-016-0197-3
36. Riaz BK, Islam MZ, Islam ANMS, Zaman MM, Hossain MA, Rahman MM. **Risk factors for non-communicable diseases in Bangladesh: findings of the population-based cross-sectional national survey 2018**. *BMJ Open* (2020.0) **10** e041334. DOI: 10.1136/bmjopen-2020-041334
37. 37WHO. Non-Communicable Disease Risk Factor Survey Bangladesh 2010. Regional Office for South-East Asia, New Delhi: World Health Organization; 2011.. *Non-Communicable Disease Risk Factor Survey Bangladesh 2010. Regional Office for South-East Asia* (2011.0)
38. Rubin RM, Koelln K. **Out-of-Pocket health expenditure differentials between elderly and non-elderly households.**. *Gerontologist.* (1993.0) **33** 595-602. DOI: 10.1093/geront/33.5.595
39. Jabeen S, Ma B, Wazib A, Umma S, Akm S, Pp DAS. **Morbidity Pattern and Health-Seeking Behaviour Among the Senior Citizens in Selected Rural Areas of Bangladesh**. *Journal Dhaka Med Coll* (2003.0) **22** 129-135
40. Khanam MA, Streatfield PK, Kabir ZN, Qiu C, Cornelius C, Wahlin Å. **Prevalence and patterns of multimorbidity among elderly people in rural Bangladesh: A cross-sectional study.**. *J Health Popul Nutr* (2011.0) **29** 406-414. DOI: 10.3329/jhpn.v29i4.8458
41. Islam A.. **Health System in Bangladesh: Challenges and Opportunities.**. *American Journal of Health Research* (2014.0) **2** 366. DOI: 10.11648/j.ajhr.20140206.18
42. Huq ZK. **Provisions and rights of senior citizens.**. *The New Age* (2022.0)
43. Sharma D, Mazta S, Parashar A. **Morbidity pattern and health-seeking behavior of aged population residing in Shimla hills of north India: A cross-sectional study.**. *J Family Med Prim Care* (2013.0) **2** 188. DOI: 10.4103/2249-4863.117421
44. Mohanty SK, Chauhan RK, Mazumdar S, Srivastava A. **Out-of-pocket Expenditure on Health Care Among Elderly and Non-elderly Households in India.**. *Soc Indic Res.* (2014.0) **115** 1137-1157. DOI: 10.1007/s11205-013-0261-7
45. Wilkinson LR. **Financial Strain and Mental Health Among Older Adults During the Great Recession**. *J Gerontol B Psychol Sci Soc Sci* (2016.0) 71. DOI: 10.1093/geronb/gbw001
46. Suy R, Chhay L, Mbugua CW. **Social Protection for Older Persons: Social Pensions in Asia**. *Journal of Social Research & Policy* (2018.0)
|
---
title: 'Positive association of triglyceride-glucose index with new-onset hypertension
among adults: a national cohort study in China'
authors:
- Qi Gao
- Yuxin Lin
- Ruqi Xu
- Fan Luo
- Ruixuan Chen
- Pingping Li
- Yuping Zhang
- Jiao Liu
- Zhenan Deng
- Yanqin Li
- Licong Su
- Sheng Nie
journal: Cardiovascular Diabetology
year: 2023
pmcid: PMC10022268
doi: 10.1186/s12933-023-01795-7
license: CC BY 4.0
---
# Positive association of triglyceride-glucose index with new-onset hypertension among adults: a national cohort study in China
## Abstract
### Background
Previous studies showed that the triglyceride-glucose (TyG) index was a better predictor of adverse cardiovascular events than triglycerides or fasting blood glucose alone. However, few studies have focused on new-onset hypertension. We aimed to explore the association of TyG index with new-onset hypertension in Chinese adults.
### Methods
A total of 4,600 participants who underwent at least 2 rounds of visits from 2009 to 2015 in the China Health and Nutrition Survey were enrolled in this study. Our outcome of interest was new-onset hypertension. Multivariate Cox hazard regression models and restricted cubic spline were performed to explore the relationship between TyG index and new-onset hypertension.
### Results
The mean (standard deviation, SD) age of the study population was 48.1 (13.6) years, and 2058 ($44.7\%$) of the participants were men. The mean (SD) TyG index level was 8.6 (0.7). A total of 1,211 ($26.3\%$) participants developed new-onset hypertension during a median (interquartile range) follow-up duration of 6.0 (2.0–6.1) years. The incidences of new-onset hypertension were $18.1\%$, $25.3\%$, $28.5\%$, and $33.4\%$ by quartiles of TyG index [from quartile 1 (Q1) to Q4], respectively. The Cox model showed that high levels of TyG index were significantly associated with increased risk of new-onset hypertension (adjusted hazard ratio [aHR]: 1.29, $95\%$ confidence interval [CI] 1.07–1.55, Q2; aHR, 1.24, $95\%$ CI 1.03–1.49, Q3; aHR, 1.50, $95\%$ CI 1.22–1.84, Q4) compared with Q1. Consistently, as a continuous variable, for every 1.0 increase in TyG index, there was a $17\%$ increase in the risk of new-onset hypertension (aHR, 1.17; $95\%$ CI 1.04–1.31). The associations were consistent in various subgroups and sensitivity analysis. The dose–response curve indicated a positive, linear association between TyG index and the risk of new-onset hypertension.
### Conclusions
High TyG index was significantly associated with an increased risk of new-onset hypertension among Chinese adults. Our findings suggest that maintaining a relatively low level of TyG index might be effective in the primary prevention of hypertension.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12933-023-01795-7.
## Introduction
Hypertension is the leading cause of cardiovascular events and all-cause mortality worldwide, which has become an emerging challenge for global public health [1]. In China, with an aging population and changing lifestyles, the prevalence of hypertension is also increasing year by year, with approximately one-third of the adult population, or more than 300 million people, suffering from hypertension between 2014 and 2015 [2, 3]. Therefore, early identifying the high-risk individuals and developing effective primary prevention strategies are very urgent to reverse the rapidly rising trend of hypertension.
Disorders of lipoprotein metabolism, in particular elevated plasma triglycerides (TG), and elevated fasting blood glucose (FBG), are all established risk factors for cardiovascular disease, especially in hypertension [4, 5]. This could be explained by insulin resistance (IR) via at least three mechanisms: inflammatory endothelial dysfunction [6], ectopic synthesis of angiotensinogen [7], and hyperinsulinaemia overstimulating the renin-angiotensin–aldosterone system [8]. Recently, the triglyceride-glucose (TyG) index, which was calculated by using TG and FBG [9], has been proposed as a surrogate of IR, and correlated with various indices of IR [10, 11]. Previous research, however, has primarily focused on the association between TyG index and the incidence of prediabetes [12], diabetes [13], cardiovascular events [14, 15], and all-cause mortality [16, 17]. Very limited studies were conducted to explore the association of TyG index with new-onset hypertension [18–28]. More importantly, these studies found inconsistence in their findings, including positive [18–24, 27] and non-significant [25, 26] associations. In addition, these studies were either limited to cross-sectional study designs, or single-center studies, or did not assess TyG index as a continuous variable. Only three prospective studies reported a positive association of TyG index and hypertension [18, 19, 28]. However, these two studies were both single-center studies and lacked regional representation. Although another study was a national cohort, it was limited to middle-aged and older adults aged ≥ 45 years. To validate these findings, a national, large sample, and prospective study cohort are required.
Based on data from the China Health and Nutrition Survey (CHNS) [29], we aimed to evaluate the potential associations between TyG index and new-onset hypertension in general Chinese adults to fill these important knowledge gaps.
## Study design, population, and data source
The study population was drawn from the CHNS cohort, which has been described previously elsewhere [29]. The data, as well as study materials that support the findings of this study, are available at the CHNS website (http://www.cpc.unc.edu/projects/china). Briefly, the CHNS is an ongoing, large-scale, prospective, multistage cohort established in 1989 among the Chinese population. By 2015, the longitudinal study had enrolled 42,829 people from 388 communities in 15 provinces and autonomous cities/districts. The provinces included in the CHNS constituted $47\%$ of China’s population [30]. To date, 10 follow-up waves (in 1989, 1991, 1993, 1997, 2000, 2004, 2006, 2009, 2011, and 2015) have been completed. At each survey round of follow-up, information on demographics, socioeconomics, diet, lifestyle habits (including smoking and alcohol consumption), and medical health was recorded by trained personnel.
Since blood measurements were first available in 2009, we utilized 3 rounds of CHNS data from 2009 to 2015 in this current study. Among 9,549 eligible participants in 2009, we first excluded those without gender-specific information ($$n = 1$$), younger than 18 years ($$n = 849$$), or who were pregnant ($$n = 62$$). Of the 8,637 participants, we further excluded participants without TyG index ($$n = 29$$) or blood pressure data ($$n = 121$$). In addition, participants who were diagnosed with hypertension in 2009 ($$n = 2$$,636) or who had no follow-up visits ($$n = 998$$), were also excluded. Furthermore, participants who had extreme dietary energy intake (< 800 or > 8000 kcal/d for male and < 600 or > 6000 kcal/d for female) ($$n = 77$$) or without other baseline covariates ($$n = 176$$, Additional file 4: Table S1) were also excluded. Finally, a total of 4,600 participants were included in the analysis (Fig. 1).Fig. 1Flowchart of participants selection. Q quartile, TyG triglyceride-glucose, CHNS China Health and Nutrition Survey The study was approved by the institutional review committees of the University of North Carolina at Chapel Hill, the National Institute of Nutrition and Food Safety, and the Chinese Center for Disease Control and Prevention. Each participant provided their written informed consent. The study complies with the Declaration of Helsinki and the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement [31] was followed in the reporting of this study.
## Standard questionnaire and examinations
A standard questionnaire was used to assess the demographics and socioeconomics data including sex, age, urban residence (yes or no), region (region was divided into north, or south based on the Qinling Mountains-Huaihe River Line.), marital status (married, unmarried, windowed, or others), education level (illiteracy, primary school, middle school, or high school or above), occupation (farmer, worker, unemployed, or others), smoking and drinking status (yes or no). Body weight and height, waist and hip circumstance and blood pressure were measured by trained study staff. Body mass index (BMI) was calculated as weight (kg)/height (m) squared. Waist to hip ratio (WHR) was calculated as waist (m)/hip (m) circumstance. Overweight was defined as BMI > 24 kg/m2 in Chinese adults.
The questionnaire on physician-diagnosed hypertension and antihypertensive treatment included the following questions: “[1] *Has a* doctor ever told you that you suffer from high blood pressure? If yes, [2] for how long have you had it? and [3] are you currently taking anti-hypertension drugs?” In China, hypertension was defined as a clinical systolic blood pressure (SBP) of 140 mmHg or greater, and/or diastolic blood pressure (DBP) of 90 mmHg or greater without the use of antihypertensive medications according to the Chinese Guidelines for Prevention and Treatment of Hypertension (1999, 2005, 2010, and 2018 versions). Overall, all the physicians used the same criteria for the clinical diagnosis and treatment of hypertension during the follow-up period.
## Dietary nutrient intakes
Individual dietary values including energy, carbohydrate, fat, and protein intakes were assessed using 3 consecutive days (randomly allocated from Monday to Sunday and equally balanced across the 7 days of the week for each sampling unit) of 24-h dietary recalls by trained nutritionists [30]. In addition, the dietary recall has been validated as an effective approach to qualifying subjects’ daily nutritional intake. Furthermore, dietary intakes in the 2009 survey were calculated using the China Food Composition Tables (FCT) version 2004. More details about the dietary recall were described on the CHNS official website.
## Blood pressure measurements
At each follow-up survey, seated blood pressure measurements were obtained by trained research staff after the participants had rested for 5 min using a mercury manometer, following the standard method with appropriately sized cuffs. Triplicate measurements on the same arm were taken in a quiet and bright room. The means of SBP and DBP of the 3 independent measures were used in the analysis.
## Blood sample collections and measurements
Participants were asked to fast for 8–12 h prior to all blood sample collections. A total of 12 mL of blood from one participant was stored and measured in the Ministry of Health laboratory of the China-Japan Friendship Hospital. TG, serum total cholesterol (TC), high density lipoprotein cholesterol (HDL-C), and low density lipoprotein cholesterol (LDL-C) were measured by the enzymatic method. FBG was measured using glucose oxidase. Lipids and glucose were analyzed with an automatic biochemical analyzer (Hitachi 7600, Kyowa, Japan). Methods for measuring other biomarkers, including glycosylated hemoglobin (HbA1c), blood urea nitrogen (BUN), uric acid, hemoglobin, serum creatinine (SCr), high sensitivity C reactive protein (hsCRP), and fasting blood insulin (FBI), have been described elsewhere [30].
Diabetes mellitus (DM) was defined as self-reported physician-diagnosed diabetes, taking oral hypoglycemic drugs/insulin injection, FBG ≥ 7.0 mmol/l, or HbA1c ≥ $6.5\%$ by the American Diabetes Association (ADA) [32]. Chronic kidney disease (CKD) was defined as estimated glomerular filtration rate (eGFR) < 60 ml/min/1.73m2 using the Chronic Kidney Disease Epidemiology Collaboration equation (CKD-EPI) [33].
## Exposure
The TyG index was calculated as follows: ln (TG (mg/dL) × FBG (mg/dL) / 2) [9].
## Outcome
The study outcome was new-onset hypertension, defined as mean SBP ≥ 140 mmHg and/or mean DBP ≥ 90 mmHg, or a physician hypertension diagnosis, or undergoing treatments for hypertension during follow-up in accordance with criteria of the WHO.
## Handling of missing variables
The distribution of missing variables was shown in Additional file 4: Table S1. To account for the small proportion of missing data, we only included participants with completed data in our primary analysis. However, in order to control the impact of missing values of some biochemical indicators, random forest imputation method was used to impute the missing variables to conducted an additional analysis.
## Statistical analysis
Shapiro–Wilk’s normality test and Bartlett test were used to detect the normal distribution and homogeneity of variance of continuous variables, respectively. Continuous variables were presented as the mean ± standard deviation (SD) for normally distributed data compared using One-Way ANOVA or a median (interquartile range [IQR]) for data that were not normally distributed compared using Kruskal–Wallis test. Categorical data were presented as a number (percentage) and compared using Pearson χ2 test.
The study population was divided into 4 groups based on the quartiles of baseline TyG index. The 2009 survey was considered baseline, and the follow-up period started from the baseline to the date of the first occurrence of an outcome or the latest survey round (in 2015) or lost to follow-up, which came first. The new-onset hypertension incidence rate was expressed as per 1,000 person-years. The univariate and multivariate Cox proportional hazard regression models were conducted to identify the association of TyG index with new-onset hypertension, and hazard ratios (HRs) were expressed with their $95\%$ confidence intervals ($95\%$ CI). Model 1 adjusted for sex, age, BMI, WHR, baseline SBP and DBP, and smoking/drinking status. Model 2 adjusted for, in addition to variables included in Model 1, urban residence, region, marital status, education level, occupation, dietary intakes, presence of DM, and biochemical variables (including BUN, uric acid, eGFR, hemoglobin, hsCRP, TP, HDL-C, LDL-C, HbA1c, and FBI). The proportional hazards assumption was tested by plotting Schoenfeld residuals against time, followed by a visual inspection for uniformity. The variance inflation factor (VIF) [34] for all predictors in our models was less than 5 (Additional file 1: Figure S1), indicating the absence of significant multicollinearity. We also performed restricted cubic spline (RCS) Cox regression, with 4 knots (5 h, 35th, 65th and 95th percentiles of TyG index), to test for linearity and characterize level-response relationships between TyG index and new-onset hypertension.
## Subgroup analyses
The possible modifications of the association between TyG index and new-onset hypertension were performed in several subgroups. Participants were stratified by age (< 50 vs. ≥ 50 years), sex, BMI (< 24 vs. ≥ 24 kg/m2), WHR (< 0.85 [median] vs. ≥ 0.85), smoking status, drinking status, SBP (< 120 vs. 120- < 140 mmHg), DBP (< 80 vs. 80- < 90 mmHg), residence (urban vs. rural), region (north vs. south), fat (< 70 [median] vs. ≥ 70 g/d), protein (< 63 [median] vs. ≥ 63 g/d), carbohydrate (< 288 [median] vs. ≥ 288 g/d), and presence CKD (yes/no). We included an interaction term in the model for each analysis to assess effect measure modification.
Stratified analyses were conducted to further explore the association between TyG index (Q2-Q4 vs. Q1) and the risk of new-onset hypertension in various subgroups (Fig. 3). Notably, there was a stronger positive association in the carbohydrate intake subgroups (288 vs. ≥ 288 g/day, aHR, 1.56, $95\%$ CI 1.23–1.99 vs. aHR, 1.14, $95\%$ CI 0.93–1.40) (P for interaction = 0.042). None of the other variables, including age, sex, BMI, baseline SBP and DBP, smoking and drinking status, residence, regions, fat intake, protein intake, and presence of CKD or DM, significantly modified this relationship (P for interaction > 0.05).Fig. 3Stratified Analyses by Potential Modifiers of the Association Between TyG index and New-Onset Hypertension. TyG triglyceride-glucose, Q quartile, HR hazard ratio, CI confidence interval, Ref reference, BMI body mass index, WHR waist hip ratio, SBP systolic blood pressure, DBP diastolic blood pressure, eGFR estimated glomerular filtration rate, BUN blood urea nitrogen, hsCRP high sensitivity C reactive protein, LDL-C low density lipoprotein cholesterol, TC total cholesterol, HbA1c glycosylated hemoglobin A1c, FBI fasting blood insulin. aIncident rate was presented as per 1000 person-years of follow-up. The model was adjusted for, if not stratified, sex, age, BMI, WHR, SBP, DBP, smoking, drinking, region, urban resistance, marital status, education, occupation, dietary intake of fat, protein and carbohydrate, diabetes mellitus, eGFR, BUN, uric acid, hsCRP, hemoglobin, total protein, LDL-C, TC, HbA1c, and FBI
## Sensitivity analysis
The robustness of the study results was further verified by various sensitivity analyses. Firstly, considering that the exact time-point of outcome occurred is difficult to capture. We fitted a Cox model using interval-censoring methods [35] to assess whether results were affected. Secondly, we excluded information on hypertension obtained through questionnaires including the presence or absence of hypertension and the use of hypertensive medication, since there might be a recall bias. Thirdly, we reanalyzed the data in participants limited with two follow-up visits to test whether the length of follow-up has an effect on the outcome. In addition, we modified the diagnostic thresholds in accordance with of American College of Cardiology (ACC) and American Heart Association (AHA) guidelines [36]. In this part, hypertension was redefined as an average SBP ≥ 130 mmHg and/or an average DBP ≥ 80 mmHg, a physician hypertension diagnosis, or taking anti-hypertension medication. Finally, the propensity score (PS) of TyG index levels (Q1 vs. Q2-Q4) was estimated using a logistic regression model whose covariates were listed in Table 1, and a baseline balanced cohort was constructed using the 1:1 Propensity Score Matching (PSM) [37] method with nearest-neighbor matching without replacement and within a caliper width of 0.001. A standardized mean difference less than 0.10 was considered a satisfactory balance between the 2 groups. Table 1Baseline characteristics of participants stratified by Quartiles of TyG indexCharacteristicsOverallQuartiles of TyG indexQ1 (≤ 8.1)Q2 (> 8.1 to 8.5)Q3 (> 8.5 to 9.0)Q4 (> 9.0)P valuen46001150115011501150TyG index8.6 (0.7)7.8 (0.2)8.3 (0.1)8.7 (0.1)9.5 (0.5) < 0.001Age, years48.1 (13.6)45.2 (14.2)48.1 (14.1)49.2 (12.8)50.0 (12.6) < 0.001Male (%)2058 (44.7)483 (42.0)475 (41.3)494 (43.0)606 (52.7) < 0.001SBP, mmHg116.4 (11.2)113.8 (11.6)115.7 (11.1)117.0 (11.2)119.1 (10.2) < 0.001DBP, mmHg76.0 (7.5)74.5 (7.8)75.2 (7.6)76.3 (7.4)77.8 (6.9) < 0.001BMI, kg/m222.9 (3.2)21.6 (2.7)22.3 (3.0)23.2 (3.1)24.3 (3.2) < 0.001WHR0.87 (0.08)0.85 (0.08)0.86 (0.08)0.87 (0.10)0.89 (0.07) < 0.001Smoking, n (%)1286 (28.0)292 (25.4)313 (27.2)305 (26.5)376 (32.7) < 0.001Drinking, n (%)1480 (32.2)353 (30.7)333 (29.0)365 (31.7)429 (37.3) < 0.001Urban residence, n (%)1335 (29.0)251 (21.8)318 (27.7)351 (30.5)415 (36.1) < 0.001Region*, n (%)0.011 North1894 (41.2)505 (43.9)455 (39.6)438 (38.1)496 (43.1) South2706 (58.8)645 (56.1)695 (60.4)712 (61.9)654 (56.9)Education, n (%)0.062 Illiteracy937 (20.4)218 (19.0)232 (20.2)259 (22.5)228 (19.8) Primary school927 (20.2)238 (20.7)232 (20.2)211 (18.3)246 (21.4) Middle school2169 (47.2)554 (48.2)548 (47.7)557 (48.4)510 (44.3) High school or above567 (12.3)140 (12.2)138 (12.0)123 (10.7)166 (14.4)Occupation, n (%) < 0.001 Farmer1420 (30.9)424 (36.9)367 (31.9)351 (30.5)278 (24.2) Worker1284 (27.9)332 (28.9)337 (29.3)307 (26.7)308 (26.8) Unemployed1674 (36.4)341 (29.7)405 (35.2)426 (37.0)502 (43.7) Others222 (4.8)53 (4.6)41 (3.6)66 (5.7)62 (5.4)Marital status0.008 Married4033 (87.7)1003 (87.2)996 (86.6)1034 (89.9)1000 (87.0) Unmarried236 (5.1)76 (6.6)68 (5.9)44 (3.8)48 (4.2) Widowed239 (5.2)51 (4.4)58 (5.0)53 (4.6)77 (6.7) Others92 (2.0)20 (1.7)28 (2.4)19 (1.7)25 (2.2)Dietary intake, g/d Energy2098.0 [1712.0, 2533.0]2091.0 [1731.5, 2513.0]2116.5 [1687.3, 2587.8]2091.5 [1720.0, 2500.8]2100.5 [1701.3, 2530.8]0.962 Fat70.0 [50.0, 96.0]69.0 [50.0, 95.0]69.0 [50.0, 94.0]70.0 [49.0, 97.0]70.5 [50.0, 97.0]0.809 Carbohydrate288.0 [229.0, 359.0]289.0 [230.3, 359.0]290.0 [229.0, 370.0]287.0 [228.0, 358.8]284.0 [228.0, 352.0]0.385 Protein63.0 [51.0, 78.0]63.0 [50.0, 78.0]63.0 [50.0, 77.0]63.0 [51.0, 78.0]64.0 [51.0, 80.0]0.334Triglycerides, mmol/L1.18 [0.81, 1.80]0.64 [0.53, 0.74]0.99 [0.88, 1.11]1.44 [1.27, 1.64]2.59 [2.12, 3.48] < 0.001TC, mmol/L4.69 [4.10, 5.37]4.24 [3.77, 4.78]4.60 [4.04, 5.21]4.88 [4.30, 5.45]5.19 [4.58, 5.93] < 0.001LDL-C, mmol/L2.85 [2.31, 3.45]2.58 [2.13, 3.02]2.87 [2.35, 3.40]3.11 [2.56, 3.67]2.96 [2.25, 3.68] < 0.001HDL-C, mmol/L1.40 [1.18, 1.64]1.55 [1.36, 1.80]1.46 [1.27, 1.71]1.36 [1.19, 1.59]1.19 [1.01, 1.41] < 0.001hsCRP, mg/dL1.00 [0.00, 2.00]1.00 [0.00, 2.00]1.00 [0.00, 2.00]1.00 [0.00, 2.00]1.00 [1.00, 3.00] < 0.001Hemoglobin, g/L140.4 (20.5)136.5 (21.0)139.4 (20.2)140.7 (20.4)144.9 (19.3) < 0.001HbA1c, %5.53 (0.80)5.35 (0.50)5.41 (0.53)5.51 (0.62)5.86 (1.23) < 0.001FBG, mmol/L5.28 (1.31)4.74 (0.56)5.01 (0.68)5.26 (0.79)6.10 (2.11) < 0.001FBI, mmol/L9.9 [7.1, 14.2]8.0 [5.7, 10.9]9.2 [7.0, 13.0]10.5 [7.6, 14.6]13.0 [9.0, 19.7] < 0.001Total protein, g/L77.08 (5.12)76.44 (4.81)76.89 (4.99)77.83 (5.11)77.16 (5.44) < 0.001BUN, mmol/L5.38 (1.47)5.45 (1.59)5.33 (1.50)5.22 (1.38)5.53 (1.40) < 0.001Uric acid, μmol/L284.0 [231.0, 346.3]252.0 [206.3, 307.0]266.5 [223.0, 320.0]284.0 [236.0, 343.0]341.0 [282.0, 417.0] < 0.001eGFR, ml/min/1.73m281.1 (15.5)84.6 (15.3)80.8 (15.2)79.3 (15.2)80.0 (15.6) < 0.001CKD, n (%)350 (7.6)52 (4.5)94 (8.2)101 (8.8)103 (9.0) < 0.001Diabetes mellitus, n (%)350 (7.6)15 (1.3)33 (2.9)69 (6.0)233 (20.3) < 0.001Q quartile, TyG triglyceride-glucose, SMD standard mean difference, SBP systolic blood pressure, DBP diastolic blood pressure, BMI body mass index, WHR waist hip ratio, TC total cholesterol, LDL-C low density lipoprotein cholesterol, HDL-C high density lipoprotein cholesterol, hsCRP high sensitivity C reactive protein, HbA1c glycosylated hemoglobin A1c, FBG fasting blood glucose, FBI fasting blood insulin, BUN blood urea nitrogen, eGFR estimated glomerular filtration rate, CKD chronic kidney disease*Region was divided into north (Heilongjiang, Liaoning, Shandong, and Henan), and south (Jiangsu, Hubei, Hunan, Guizhou, and Guangxi) based on the Qinling Mountains-Huaihe River Line
## Additional analyses
We conducted an E-value analysis [38] to assess the extent of unmeasured confounding that would be required to negate the observed results. In addition, considering the impact of DM and CKD on hypertension incidence, we further excluded 651 individuals with DM and/or CKD.
All the above statistical analyses were performed using the R software (version 4.1.2; http://www.r-project.org/). A two-sided $P \leq 0.05$ was considered to be statistically significant in all analyses.
The E-values for the hazard ratio and lower confidence bound for the outcome were 1.69 and 1.36 (Additional file 3: Figure S3). After excluding 651 individuals with DM and/or CKD, or imputing the missing variables, the results were consistent with the primary analysis (Additional file 4: Tables S10, S11).
## Study population and baseline characteristics
The flowchart of the study population selection was shown in Fig. 1. The characteristics of these included and excluded participants were summarized in Additional file 4: Table S2. The distribution of BMI, drinking, residence, and TyG index was similar between two groups. Of the 4,600 participants selected for analysis (of whom 2,058 were male [$44.7\%$]), the mean (SD) age was 48.1 (13.6) years, and the mean (SD) baseline SBP and DBP were 116.4 (11.2) and 76.0 (7.5) mmHg, respectively. Participants who were excluded from the dataset ($$n = 4$$,949) tended to be younger, have a high proportion of illiteracy, and unemployed, and were more likely to have CKD or DM.
The demographic and baseline characteristics of the included participants stratified by the quintiles of TyG index were summarized in Table 1. The mean (SD) TyG index was 8.6 (0.7). *In* general, compared to the lowest quantile group, those in the high quantile were older, more likely to be male, and less likely to live in the north of China and urban residence; they had higher values of SBP, DBP, BMI, WHR, hemoglobin, total protein, triglycerides, HDL-C, LDL-C, FBG, and HbA1c, and a higher prevalence of CKD and DM (all $P \leq 0.05$). In addition, there were no significant differences in dietary intakes in these groups (all $P \leq 0.05$). The distribution of TyG index in the study population was shown in Additional file 2: Figure S2. The baseline characteristics of participants stratified by outcome were compared in Additional file 4: Table S3.
## Association between TyG index and New-Onset hypertension
A total of 1,211 ($26.3\%$) participants developed new-onset hypertension during a median (IQR) follow-up duration of 6.0 (2.0–6.1) years. The incidences of new-onset hypertension were $18.1\%$, $25.3\%$, $28.5\%$, and $33.4\%$ by quartiles of TyG index [from quartile 1 (Q1) to Q4], respectively. After adjusting for confounders, the Cox model showed that high levels of TyG index were significantly associated with an increased risk of new-onset hypertension (adjusted hazard ratio [aHR]: 1.29, $95\%$ confidence interval [CI] 1.07–1.55, Q2; aHR, 1.24, $95\%$ CI 1.03–1.49, Q3; aHR, 1.50, $95\%$ CI 1.22–1.84, Q4) compared with Q1 (Table 2). We also found that this risk increased progressively as the TyG index increased (P for trend < 0.001). Similar trends were observed after combining the Q2–Q4 groups. Consistently in the above analysis, as a continuous variable, for per 1.0 increase in TyG index, there was a $17\%$ increase in the risk of new-onset hypertension (aHR, 1.17; $95\%$ CI 1.04–1.31). The dose–response curve indicated a positive, linear association between TyG index and the risk of new-onset hypertension (Fig. 2).Table 2The association of TyG index with new-onset hypertensionTyG indexTotal NNo. of events (incident ratea)Crude modelModel 1Model 2HR ($95\%$ CI)P valueHR ($95\%$ CI)P valueHR ($95\%$ CI)P valueQuartiles Q1 (≤ 8.1)1150208 (40.4)Ref. Ref. Ref. Q2 (> 8.1 to 8.5)1150291 (56.3)1.45 (1.22–1.74) < 0.0011.27 (1.06–1.52)0.0101.29 (1.07–1.55)0.006 Q3 (> 8.5 to 9.0)1150328 (62.5)1.62 (1.37–1.93) < 0.0011.23 (1.03–1.47)0.0231.24 (1.03–1.49)0.024 Q4 (> 9.0)1150384 (79.8)2.21 (1.86–2.62) < 0.0011.47 (1.23–1.76) < 0.0011.50 (1.22–1.84) < 0.001 P for trend < 0.001 < 0.001 < 0.001Categories Q1 (≤ 8.1)1150208 (40.4)Ref. Ref. Ref. Q2-Q4 (> 8.1)34501003 (65.9)1.74 (1.50–2.02) < 0.0011.31 (1.12–1.53) < 0.0011.30 (1.11–1.53)0.001Continuous Per 1.0 increase46001211 (59.4)1.39 (1.29–1.50) < 0.0011.14 (1.05–1.24)0.0021.17 (1.04–1.31)0.007Model1: adjusted for sex, age, BMI, WHR, SBP, DBP, smoking, and drinkingModel2 (Full model): Model1 + further adjusted for region, urban resistance, marital status, education, occupation, dietary intake of fat, protein and carbohydrate, diabetes mellitus, eGFR, BUN, uric acid, hsCRP, hemoglobin, total protein, LDL-C, TC, HbA1c, and FBITyG triglyceride-glucose, Q quartile, HR hazard ratio, CI confidence interval, Ref reference, BMI body mass index, WHR waist hip ratio, SBP systolic blood pressure, DBP diastolic blood pressure, eGFR estimated glomerular filtration rate, BUN blood urea nitrogen, hsCRP high sensitivity C reactive protein, LDL-C low density lipoprotein cholesterol, TC total cholesterol, HbA1c glycosylated hemoglobin A1c, FBI fasting blood insulinaIncident rate was presented as per 1000 person-years of follow-upFig. 2Levels of TyG index and the Risk of New-Onset Hypertension. TyG triglyceride-glucose, Q quartile, HR hazard ratio, CI confidence interval, Ref reference, BMI body mass index, WHR waist hip ratio, SBP systolic blood pressure, DBP diastolic blood pressure, eGFR estimated glomerular filtration rate, BUN blood urea nitrogen, hsCRP high sensitivity C reactive protein, LDL-C low density lipoprotein cholesterol, TC total cholesterol, HbA1c glycosylated hemoglobin A1c, FBI fasting blood insulin. The model was adjusted for sex, age, BMI, WHR, SBP, DBP, smoking, drinking, region, urban resistance, marital status, education, occupation, dietary intake of fat, protein and carbohydrate, diabetes mellitus, eGFR, BUN, uric acid, hsCRP, hemoglobin, total protein, LDL-C, TC, HbA1c, and FBI
## Sensitivity analyses
The results remained consistent when several methods were utilized to verify the robustness of the relationship between TyG index and new-onset hypertension.
The effect of TyG index on new-onset hypertension was consistent when the interval-censoring Cox model was utilized for analysis (Q2-Q4 vs. Q1, aHR, 1.30; $95\%$ CI 1.13–1.47; per 1.0 increase, aHR, 1.14; $95\%$ CI 1.02–1.26) (Additional file 4: Table S4). After excluding the questionnaire data defining hypertension, the results were not substantially changed (Q2-Q4 vs. Q1, aHR, 1.24; $95\%$ CI 1.06–1.45; per 1.0 increase, aHR, 1.05; $95\%$ CI 1.03–1.28) (Additional file 4: Table S5). In addition, when we restricted participants to only two follow-up visits, the result was consistent with the main analysis (Q2-Q4 vs. Q1, aHR, 1.32; $95\%$ CI 1.12–1.57; per 1.0 increase, aHR, 1.15; $95\%$ CI 1.02–1.29) (Additional file 4: Table S6). Similar trends were observed between TyG index and new-onset hypertension after redefining hypertension by the new ACC and AHA guidelines with a BP threshold of $\frac{130}{80}$, although there was no significantly statistical difference (Q2-Q4 vs. Q1, aHR, 1.09; $95\%$ CI 0.94–1.26; per 1.0 increase, aHR, 0.98; $95\%$ CI 0.87–1.09) (Additional file 4: Table S7). Finally, after conducting 1:1 PSM, we obtained 993 pairs of subjects with baseline-balanced in Q1 and Q2-Q4 groups (Additional file 4: Table S8). The Cox model also showed that high levels of TyG index were significantly associated with a higher risk of new-onset hypertension (Q2-Q4 vs. Q1, aHR, 1.25; $95\%$ CI 1.02–1.52; per 1.0 increase, aHR, 1.23; $95\%$ CI 1.03–1.48) (Additional file 4: Table S9).
## Discussion
In this large, national, longitudinal cohort study among general Chinese adults, containing 4,600 participants with up to 6 years of follow-up, we found a positive association between TyG index and new-onset hypertension after adjusting for confounders, and this relationship was consistent across various subgroups and in sensitivity analyses. The dose–response curve indicated a positive, linear association between TyG index and the risk of new-onset hypertension. To our best knowledge, this is the first national prospective study with adults of all ages (18–94 years old) to investigate the relationship between TyG index and new-onset hypertension. The present study may provide new insights into the primary prevention of hypertension.
TyG index, as a product of FBG and triglyceride, has been confirmed to date to be closely related to the traditional risk factors of cardiovascular disease [39]. Sanchez-Inigo L et al. [ 40] reported, during a 10-year follow-up, that a high TyG index was significantly associated with an increased risk of future ASCVD events. Similarly, two Korean studies found TyG index to be an independent predictor of progression of coronary artery calcification [14, 41]. These evidences suggested that TyG index could have an indirect effect on cardiovascular disease.
The effects of TyG index on blood pressure have been evaluated in several previous studies, which have reported inconsistent results [18–27, 42]. Several cross-sectional studies have shown that an elevated TyG index was associated with an increased risk of developing hypertension, either in the general population [20–23], in children or adolescents [42], or in the elderly population [24]. Moreover, a prospective study in Spanish [18] reported a positive association between TyG index and hypertension in the general population during a long-term follow-up. Zheng et al. [ 19] also conducted another single-center, longitudinal study with 4,686 subjects followed up for 9 years, and demonstrated that TyG index could predict incident hypertension among the Chinese population. Of note, another study [28] reached a similar conclusion using a national cohort, but it was limited to middle-aged and older adults over 45 years. In addition, in a recent meta-analysis of 8 studies involving 200,044 general adult participants [43], the relative risk of hypertension was higher for the highest category of TyG index compared with the lowest. These studies were consistent with our study, which showed that high levels of TyG index were significantly associated with increased risk of new-onset hypertension and that a linear association was observed. Contrary to previous studies and our results, two cross-sectional studies [25, 26] did not find a significant association between TyG index and hypertension in obese or normal‐weight individuals. This may be attributed to the heterogeneity of the selected population and the sample size of the study. Therefore, further research on this topic is needed.
In stratified analysis, we found that most of the variables did not significantly modify the association between TyG index and new-onset hypertension, which indicates that the results of this study are applicable to the majority of the general population. However, the association between TyG index and new-onset hypertension was stronger in patients with low carbohydrate intake (< 288 g/day). Previous studies have reported that higher carbohydrate intake was related to a higher risk of hypertension [44, 45]. Therefore, individuals with higher carbohydrate intake will offset some of the risk from TyG index.
The potential mechanisms for the association between TyG index and new-onset hypertension may be explained by IR. IR has been confirmed by many studies to be significantly associated with hypertension by many mechanisms [6–8]. In recent years, many studies have concluded that TyG index is a surrogate of IR because it was correlated with various indices of IR, such as the M rates in the hyperinsulinaemic–euglycaemic clamp test [10] or the homeostasis of minimal assessment of insulin resistance (HOMA-IR) [46], and with the degree of carotid atherosclerosis [11]. More mechanistic studies are needed to further validate the relationship between TyG index, IR and hypertension.
The major strengths of this study were the 6-year national, longitudinal population-based study and the large number of subjects used to explore the relationship between TyG index and new-onset hypertension. However, certain limitations also existed in this study. First, although we adjusted for potential confounders such as demographics, dietary intake, and blood biochemical markers as much as possible, the E-values for the hazard ratio and lower confidence bound for the primary outcome were also small, which implies that little unmeasured confounding would be needed to reduce the observed association or its $95\%$ CI to the null. However, residual confounding could not be completely eliminated. Second, limited by observational studies, we could not determine the causal relationship between TyG index and new-onset hypertension. Third, since the definition of hypertension was based on physician on-site blood pressure measurement data and questionnaires, there may be recall bias, but the findings remained consistent when we excluded questionnaire-based information on hypertension obtained in the sensitivity analysis. Forth, we could not explore the relationship between TyG index and different hypertension subtypes since information related to 24-h dynamic changes in blood pressure was not available. Fifth, considering that death was an inevitable competitive risk, we may underestimate the relationship between TyG index and new-onset hypertension. However, in our study population, only 74 ($1.6\%$) participants died during the follow-up, which is unlikely to change the trend of results. Last, this study was limited to the *Chinese* general population, and more studies were needed to confirm the consistency of findings in other ethnic and national studies in the future.
## Conclusions
In conclusion, we confirmed that high levels of TyG index were associated with higher risk of new-onset hypertension and showed a linear relationship through a longitudinal national cohort. Our findings suggest that maintaining a relatively low level of TyG index will help with the primary prevention of hypertension. In clinical practice, the TyG index is easily available, and clinicians could use this indicator to risk-stratify the general population in order to provide more personalized prevention or treatment.
## Supplementary Information
Additional file 1: Figure S1. The variance inflation factor (VIF) values for all variables in our model. Additional file 2: Figure S2. Distribution of TyG index in the study population. Additional file 3: Figure S3. E-value analysis to assess the extent of unmeasured confounding that would be required to negate the observed results. Additional file 4: Table S1. Distribution of missing variables. Table S2. Baseline characteristics of excluded and included participants. Table S3. Baseline characteristics of participants stratified by outcome. Table S4. The association of TyG index with new-onset hypertension using interval censored Cox regression model. Table S5. The association of TyG index with new-onset hypertension after excluding the questionnaire data defining hypertension. Table S6. The association of TyG index with new-onset hypertension in participants limited with two follow-up visits. Table S7. The association of TyG index with new-onset hypertension defined by the novel diagnostic criteria (SBP/DBP: $\frac{130}{80}$). Table S8. Baseline characteristics of participants stratified by TyG index quartiles after 1:1 propensity score matching. Table S9. The association of TyG index with new-onset hypertension after 1:1 propensity score matching. Table S10. The association of TyG index with new-onset hypertension after excluding individuals with DM and/or CKD. Table S11. The association of TyG index with new-onset hypertension after imputing the baseline missing values.
## References
1. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM. **Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study**. *J Am Coll Cardiol* (2020.0) **76** 2982-3021. DOI: 10.1016/j.jacc.2020.11.010
2. Lewington S, Lacey B, Clarke R, Guo Y, Kong XL, Yang L. **The burden of hypertension and associated risk for cardiovascular mortality in China**. *JAMA Intern Med* (2016.0) **176** 524-532. DOI: 10.1001/jamainternmed.2016.0190
3. Li Y, Yang L, Wang L, Zhang M, Huang Z, Deng Q. **Burden of hypertension in China: a nationally representative survey of 174,621 adults**. *Int J Cardiol* (2017.0) **227** 516-523. DOI: 10.1016/j.ijcard.2016.10.110
4. Choi JW, Park JS, Lee CH. **Interactive effect of high sodium intake with increased serum triglycerides on hypertension**. *PLoS One* (2020.0) **15** e0231707. DOI: 10.1371/journal.pone.0231707
5. Kuwabara M, Hisatome I. **The relationship between fasting blood glucose and hypertension**. *Am J Hypertens* (2019.0) **32** 1143-1145. DOI: 10.1093/ajh/hpz147
6. Morales-Villegas E. **Dyslipidemia, hypertension and diabetes metaflammation. A Unique mechanism for 3 risk factors**. *Curr Hypertens Rev* (2014.0). DOI: 10.2174/1573402110666140702091315
7. Egan BM. **Insulin resistance and the sympathetic nervous system**. *Curr Hypertens Rep* (2003.0) **5** 247-254. DOI: 10.1007/s11906-003-0028-7
8. Franco OH, Massaro JM, Civil J, Cobain MR, O'Malley B, D'Agostino RB. **Trajectories of entering the metabolic syndrome: the framingham heart study**. *Circulation* (2009.0) **120** 1943-1950. DOI: 10.1161/CIRCULATIONAHA.109.855817
9. Simental-Mendía LE, Rodríguez-Morán M, Guerrero-Romero F. **The product of fasting glucose and triglycerides as surrogate for identifying insulin resistance in apparently healthy subjects**. *Metab Syndr Relat Disord* (2008.0) **6** 299-304. DOI: 10.1089/met.2008.0034
10. Guerrero-Romero F, Simental-Mendia LE, Gonzalez-Ortiz M, Martinez-Abundis E, Ramos-Zavala MG, Hernandez-Gonzalez SO. **The product of triglycerides and glucose, a simple measure of insulin sensitivity. Comparison with the euglycemic-hyperinsulinemic clamp**. *J Clin Endocrinol Metab* (2010.0) **95** 3347-3351. DOI: 10.1210/jc.2010-0288
11. Du T, Yuan G, Zhang M, Zhou X, Sun X, Yu X. **Clinical usefulness of lipid ratios, visceral adiposity indicators, and the triglycerides and glucose index as risk markers of insulin resistance**. *Cardiovasc Diabetol* (2014.0) **13** 146. DOI: 10.1186/s12933-014-0146-3
12. Wen J, Wang A, Liu G, Wang M, Zuo Y, Li W. **Elevated triglyceride-glucose (TyG) index predicts incidence of prediabetes: a prospective cohort study in China**. *Lipids Health Dis* (2020.0) **19** 226. DOI: 10.1186/s12944-020-01401-9
13. Xuan X, Hamaguchi M, Cao Q, Okamura T, Hashimoto Y, Obora A. **U-shaped association between the triglyceride-glucose index and the risk of incident diabetes in people with normal glycemic level: a population-base longitudinal cohort study**. *Clin Nutr* (2021.0) **40** 1555-1561. DOI: 10.1016/j.clnu.2021.02.037
14. Park B, Lee YJ, Lee HS, Jung DH. **The triglyceride-glucose index predicts ischemic heart disease risk in Koreans: a prospective study using National Health Insurance Service data**. *Cardiovasc Diabetol* (2020.0) **19** 210. DOI: 10.1186/s12933-020-01186-2
15. Barzegar N, Tohidi M, Hasheminia M, Azizi F, Hadaegh F. **The impact of triglyceride-glucose index on incident cardiovascular events during 16 years of follow-up: Tehran lipid and glucose study**. *Cardiovasc Diabetol* (2020.0) **19** 155. DOI: 10.1186/s12933-020-01121-5
16. Liao Y, Zhang R, Shi S, Zhao Y, He Y, Liao L. **Triglyceride-glucose index linked to all-cause mortality in critically ill patients: a cohort of 3026 patients**. *Cardiovasc Diabetol* (2022.0) **21** 128. DOI: 10.1186/s12933-022-01563-z
17. Liu X, Tan Z, Huang Y, Zhao H, Liu M, Yu P. **Relationship between the triglyceride-glucose index and risk of cardiovascular diseases and mortality in the general population: a systematic review and meta-analysis**. *Cardiovasc Diabetol* (2022.0) **21** 124. DOI: 10.1186/s12933-022-01546-0
18. Sanchez-Inigo L, Navarro-Gonzalez D, Pastrana-Delgado J, Fernandez-Montero A, Martinez JA. **Association of triglycerides and new lipid markers with the incidence of hypertension in a Spanish cohort**. *J Hypertens* (2016.0) **34** 1257-1265. DOI: 10.1097/HJH.0000000000000941
19. Zheng R, Mao Y. **Triglyceride and glucose (TyG) index as a predictor of incident hypertension: a 9-year longitudinal population-based study**. *Lipids Health Dis* (2017.0) **16** 175. DOI: 10.1186/s12944-017-0562-y
20. Bala C, Gheorghe-Fronea O, Pop D, Pop C, Caloian B, Comsa H. **The association between six surrogate insulin resistance indexes and hypertension: a population-based study**. *Metab Syndr Relat Disord* (2019.0) **17** 328-333. DOI: 10.1089/met.2018.0122
21. Wang K, He G, Zhang Y, Yin J, Yan Y, Zhang Y. **Association of triglyceride-glucose index and its interaction with obesity on hypertension risk in Chinese: a population-based study**. *J Hum Hypertens* (2021.0) **35** 232-239. DOI: 10.1038/s41371-020-0326-4
22. Zhu B, Wang J, Chen K, Yan W, Wang A, Wang W. **A high triglyceride glucose index is more closely associated with hypertension than lipid or glycemic parameters in elderly individuals: a cross-sectional survey from the reaction study**. *Cardiovasc Diabetol* (2020.0) **19** 112. DOI: 10.1186/s12933-020-01077-6
23. Lee DH, Park JE, Kim SY, Jeon HJ, Park JH. **Association between the triglyceride-glucose (TyG) index and increased blood pressure in normotensive subjects: a population-based study**. *Diabetol Metab Syndr* (2022.0) **14** 161. DOI: 10.1186/s13098-022-00927-5
24. Jian S, Su-Mei N, Xue C, Jie Z, Xue-Sen W. **Association and interaction between triglyceride-glucose index and obesity on risk of hypertension in middle-aged and elderly adults**. *Clin Exp Hypertens* (2017.0) **39** 732-739. DOI: 10.1080/10641963.2017.1324477
25. Liu XZ, Fan J, Pan SJ. **METS-IR, a novel simple insulin resistance indexes, is associated with hypertension in normal-weight Chinese adults**. *J Clin Hypertens (Greenwich)* (2019.0) **21** 1075-1081. DOI: 10.1111/jch.13591
26. Morales-Gurrola G, Simental-Mendía LE, Castellanos-Juárez FX, Salas-Pacheco JM, Guerrero-Romero F. **The triglycerides and glucose index is associated with cardiovascular risk factors in metabolically obese normal-weight subjects**. *J Endocrinol Invest* (2020.0) **43** 995-1000. DOI: 10.1007/s40618-020-01184-x
27. Zhang F, Zhang Y, Guo Z, Yang H, Ren M, Xing X. **The association of triglyceride and glucose index, and triglyceride to high-density lipoprotein cholesterol ratio with prehypertension and hypertension in normoglycemic subjects: a large cross-sectional population study**. *J Clin Hypertens (Greenwich)* (2021.0) **23** 1405-1412. DOI: 10.1111/jch.14305
28. Liu T, Xuan H, Yin J, Wang L, Wang C, Xu X. **Triglyceride glucose index increases significantly risk of hypertension development in Chinese individuals aged >/=45 years old: analysis from the China health and retirement longitudinal study**. *J Multidiscip Healthc* (2023.0) **16** 63-73. DOI: 10.2147/JMDH.S391905
29. Popkin BM, Du S, Zhai F, Zhang B. **Cohort profile: the china health and nutrition survey–monitoring and understanding socio-economic and health change in China, 1989–2011**. *Int J Epidemiol* (2010.0) **39** 1435-1440. DOI: 10.1093/ije/dyp322
30. Zhang B, Zhai FY, Du SF, Popkin BM. **The China health and nutrition survey, 1989–2011**. *Obes Rev* (2014.0) **15** 2-7. DOI: 10.1111/obr.12119
31. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. **The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies**. *Ann Intern Med* (2007.0) **147** 573-577. DOI: 10.7326/0003-4819-147-8-200710160-00010
32. Assoc AD. **Diagnosis and classification of diabetes mellitus**. *Diabetes Care* (2005.0) **28** S37-S42. DOI: 10.2337/diacare.28.suppl_1.S37
33. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, Feldman HI. **A new equation to estimate glomerular filtration rate**. *Ann Intern Med* (2009.0) **150** 604-612. DOI: 10.7326/0003-4819-150-9-200905050-00006
34. Kim JH. **Multicollinearity and misleading statistical results**. *Korean J Anesthesiol* (2019.0) **72** 558-569. DOI: 10.4097/kja.19087
35. Kim DK. **Regression analysis of interval-censored survival data with covariates using log-linear models**. *Biometrics* (1997.0) **53** 1274-1283. DOI: 10.2307/2533496
36. 36.Whelton PK, Carey RM, Aronow WS, Casey DE, Jr., Collins KJ, Dennison Himmelfarb C, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Hypertension, 71: 1269–1324, 2018 10.1161/HYP.0000000000000066
37. Zhao QY, Luo JC, Su Y, Zhang YJ, Tu GW, Luo Z. **Propensity score matching with R: conventional methods and new features**. *Ann Transl Med* (2021.0) **9** 812. DOI: 10.21037/atm-20-3998
38. Haneuse S, VanderWeele TJ, Arterburn D. **Using the E-Value to assess the potential effect of unmeasured confounding in observational studies**. *JAMA* (2019.0) **321** 602-603. DOI: 10.1001/jama.2018.21554
39. Tao LC, Xu JN, Wang TT, Hua F, Li JJ. **Triglyceride-glucose index as a marker in cardiovascular diseases: landscape and limitations**. *Cardiovasc Diabetol* (2022.0). DOI: 10.1186/s12933-022-01511-x
40. Sanchez-Inigo L, Navarro-Gonzalez D, Fernandez-Montero A, Pastrana-Delgado J, Martinez JA. **The TyG index may predict the development of cardiovascular events**. *Eur J Clin Invest* (2016.0) **46** 189-197. DOI: 10.1111/eci.12583
41. Won KB, Park EJ, Han D, Lee JH, Choi SY, Chun EJ. **Triglyceride glucose index is an independent predictor for the progression of coronary artery calcification in the absence of heavy coronary artery calcification at baseline**. *Cardiovasc Diabetol* (2020.0) **19** 34. DOI: 10.1186/s12933-020-01008-5
42. Simental-Mendia LE, Hernandez-Ronquillo G, Gamboa-Gomez CI, Gomez-Diaz R, Rodriguez-Moran M, Guerrero-Romero F. **The triglycerides and glucose index is associated with elevated blood pressure in apparently healthy children and adolescents**. *Eur J Pediatr* (2019.0) **178** 1069-1074. DOI: 10.1007/s00431-019-03392-x
43. Wang Y, Yang W, Jiang X. **Association between triglyceride-glucose index and hypertension: a meta-analysis**. *Front Cardiovasc Med* (2021.0) **8** 644035. DOI: 10.3389/fcvm.2021.644035
44. He DL, Sun N, Xiong ST, Qiao YA, Ke CF, Shen YP. **Association between the proportions of carbohydrate and fat intake and hypertension risk: findings from the China Health and Nutrition Survey**. *J Hypertens* (2021.0) **39** 1386-1392. DOI: 10.1097/Hjh.0000000000002803
45. Qi SF, Cao YJ, Wang HJ, Zhang B, Yan J, Tian QB. **Associations of carbohydrate intake with new-onset hypertension subtypes: results from the China Health and Nutrition Survey (2000–2011)**. *Front Nutr* (2021.0) **8** 728774. DOI: 10.3389/fnut.2021.728774
46. Abbasi F, Reaven GM. **Comparison of two methods using plasma triglyceride concentration as a surrogate estimate of insulin action in nondiabetic subjects: triglycerides x glucose versus triglyceride/high-density lipoprotein cholesterol**. *Metabolism* (2011.0) **60** 1673-1676. DOI: 10.1016/j.metabol.2011.04.006
|
---
title: 'Paternal preconception modifiable risk factors for adverse pregnancy and offspring
outcomes: a review of contemporary evidence from observational studies'
authors:
- Tristan Carter
- Danielle Schoenaker
- Jon Adams
- Amie Steel
journal: BMC Public Health
year: 2023
pmcid: PMC10022288
doi: 10.1186/s12889-023-15335-1
license: CC BY 4.0
---
# Paternal preconception modifiable risk factors for adverse pregnancy and offspring outcomes: a review of contemporary evidence from observational studies
## Abstract
### Background
The preconception period represents transgenerational opportunities to optimize modifiable risk factors associated with both short and long-term adverse health outcomes for women, men, and children. As such, preconception care is recommended to couples during this time to enable them to optimise their health in preparation for pregnancy. Historically, preconception research predominately focuses on maternal modifiable risks and health behaviours associated with pregnancy and offspring outcomes; limited attention has been given to inform paternal preconception health risks and outcomes. This systematic review aims to advance paternal preconception research by synthesising the current evidence on modifiable paternal preconception health behaviours and risk factors to identify associations with pregnancy and/or offspring outcomes.
### Methods
Medline, Embase, Maternity and Infant care, CINAHL, PsycINFO, Scopus, and ISI Proceedings were searched on the 5th of January 2023, a date limit was set [2012–2023] in each database. A Google Scholar search was also conducted identifying all other relevant papers. Studies were included if they were observational, reporting associations of modifiable risk factors in the preconception period among males (e.g., identified as reproductive partners of pregnant women and/or fathers of offspring for which outcomes were reported) with adverse pregnancy and offspring outcomes. Study quality was assessed using the Newcastle–Ottawa Scale. Exposure and outcome heterogeneity precluded meta-analysis, and results were summarised in tables.
### Results
This review identified 56 cohort and nine case control studies. Studies reported on a range of risk factors and/or health behaviours including paternal body composition ($$n = 25$$), alcohol intake ($$n = 6$$), cannabis use ($$n = 5$$), physical activity ($$n = 2$$), smoking ($$n = 20$$), stress ($$n = 3$$) and nutrition ($$n = 13$$). Outcomes included fecundability, IVF/ISCI live birth, offspring weight, body composition/BMI, asthma, lung function, leukemia, preterm birth, and behavioural issues. Despite the limited number of studies and substantial heterogeneity in reporting, results of studies assessed as good quality showed that paternal smoking may increase the risk of birth defects and higher paternal BMI was associated with higher offspring birthweight.
### Conclusion
The current evidence demonstrates a role of paternal preconception health in influencing outcomes related to pregnancy success and offspring health. The evidence is however limited and heterogenous, and further high-quality research is needed to inform clinical preconception care guidelines to support men and couples to prepare for a health pregnancy and child.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-023-15335-1.
## Plain English Summary
The time prior to conception, preconception, is widely acknowledged as an integral period whereby a woman’s health, lifestyle, and diet influence the outcomes of future pregnancy and the health of future offspring. Similarly, the influence of a man’s health, lifestyle, and diet during preconception on pregnancy and offspring outcomes must be considered. However, the male reproductive partner’s role during preconception has attracted much less researcher attention when compared to maternal exposures and outcomes and may be undervalued.
Therefore, this review explores the modifiable risk factors of males in the preconception period and how these risks influence adverse pregnancy and/or offspring outcomes. A total of 65 papers are included for review which examined risks associated with factors such as alcohol use, physical activity, stress, and nutrition. Overall, the papers identified some consistent results: paternal smoking increased risk of adverse offspring outcomes, while increased paternal body mass index was associated with higher offspring birthweight. Nevertheless, this review concludes that paternal preconception modifiable risk factors remain largely underexplored. Evidently, more high-quality research must be conducted to better understand the health, lifestyle, and diets of males in the preconception period and how various paternal modifiable risks can influence their partner’s pregnancy and the health and developmental outcomes of their offspring.
## Introduction
Preconception care is defined as the provision of health interventions (behavioural, social, and/or biomedical) to women and couples prior to conception [1]. It addresses the transgenerational opportunity of enabling and optimizing health while limiting risk factors associated with both short- and long-term adverse health outcomes for women, men, and their children. There is global consensus on the key aspects of preconception care [2], yet a consistent definition and clear attributes of the preconception population remain elusive [3]. Preconception research predominately focuses on maternal modifiable risks or health behaviours associated with offspring outcomes [4] as demonstrated by a scoping review of preconception health behaviours which found only $11\%$ of all studies included paternal modifiable risks or health behaviours [5]. Nonetheless, the research community recognizes the father or male partner’s contribution to child health and development before birth [6, 7] and the need to balance our gaze on men in preconception care [8]. This is further supported by the increasing number and diversity of publications about paternal preconception health [9] and formulation of the Paternal Origins of Health and Disease (POHaD) model [10]. As such, the preconception population may include all reproductively aged individuals in a period from their birth to the conception of their (or their partner’s) pregnancy. The care provided during this period must respond to a clear set of identified risk factors and exposures as relevant to each individual.
Indeed, when planning parenthood, males find themselves within a contentious grey zone; concurrently involved while also considered an outsider [11]. A recent survey in the UK found that men are interested in engaging in positive preconception health behaviours [7]. Of the over 500 men surveyed, $19\%$ had visited a primary health provider for preconception health advice, and those who had received advice were more likely to adopt positive health behaviours before their partner’s pregnancy. On the other hand, general practitioners (GPs) report low confidence in their knowledge about paternal preconception health care and modifiable factors affecting male fertility [12, 13]. They describe feeling apprehensive or even sensitive to the subject matter and/or challenged by navigating the stereotypical masculine predispositions toward fertility and preconception care [14]. *In* general, preconception risks are not raised by GPs with male patients unless subfertility is involved and preconception discussions are often encumbered by numerous impediments including the limited time, financial constraints, and knowledge of GPs, plus in some cases, a lack of GP motivation and perceived need for health care [12]. A systematic review of preconception care guidelines found that six of the 11 guidelines included provided preconception care guidance for men [15]. Only one guideline, a position paper from the American Academy of Family Physicians, contained a dedicated section outlining recommendations on preconception interventions for men [16]. Evidently, there is an unmet need for health professionals, and men, to readily access current relevant information regarding paternal preconception health exposures and outcomes, informing clinical practice and directing health decisions.
Evidence supporting paternal preconception care considers males contribution to child health and development before conception via direct (genetic and epigenetic contributions – health and lifestyle behaviours, exposure to environmental toxins, life stressors, and neuroendocrinology) and indirect pathways (the couple’s relationship, and the influence of men on their partner’s health and health behaviours) [17]. Yet, there is a stark contrast between the magnitude of research investigating maternal preconception health risks—including body composition, lifestyle behaviours, and diet/nutrition – and the relative scarcity of research attention directed towards understanding paternal health exposures and outcomes. In direct response, this systematic review aims to advance paternal preconception research by synthesising the current evidence on associations of modifiable paternal preconception health behaviours and risk factors with pregnancy and/or offspring outcomes.
## Methods
This review was prospectively registered in PROSPERO (Registration Number: CRD42021209994), and reported in line with PRISMA 2020 guidance [18] and the AMSTAR 2 critical appraisal tool [19].
## Search strategy
A search was conducted on January 5th 2023, (See Supplementary File 1 Search strategy), through the following databases: 1) Medline (OVID) 2) Embase (OVID), 3) Maternity and Infant care [MIDIRS] (OVID) 4) CINAHL (EBSCO), 5) PsycINFO (EBSCO), 6) Scopus, & 7) ISI Proceedings. For each database, a date limit of 2012–2023 was set. When available, subject headings identified from the controlled vocabulary of each database were also included in the search. On January 11th 2023, a Google Scholar search was conducted for the search term ‘Paternal preconception’, applying the filter to limit articles published since 2022 and searching through to page seven, identifying any other recently published relevant papers. Google Scholar was also used to identify relevant studies citing each included paper. Reference lists of each included paper were then checked for additional relevant studies.
## Selection criteria
Papers were included if they were original contemporary observational research (cross-sectional, cohort or case–control study designs) involving males in the preconception period, examining an association or correlation of a modifiable risk factor or health behaviour to pregnancy and/or offspring health and developmental outcomes. The male participants must identify as being the partner of the pregnant women and/or the biological father of the child for which pregnancy and offspring outcomes were reported (Table 1 – PICO).Table 1PICO (Population, Intervention, Comparison, Outcome) inclusion criteriaPopulationIntervention/ExposureComparisonOutcomeMales who identified as being the partner of the pregnant women and/or the biological father of the child for which outcomes were reportedExposure to modifiable risk factor(s) in the preconception periodNo exposure to modifiable risk factor(s) in the preconception period (or comparison group as defined by individual studies)Adverse pregnancy and offspring outcomes Observational study designs are generally utilized to identify correlations and establish findings at the population level hence are solely considered in this review.
Papers were excluded if they were: reviews, did not report new empirical findings from original studies (i.e. commentaries, opinion-pieces and editorials), not studying humans, not examining male parent exposures, did not differentiate between maternal and paternal preconception exposures, or if the exposure examined specific illness populations. Papers were also excluded when the exposure was not assessed or retrospectively recalled during the preconception period, the outcome was not related to pregnancy or offspring health or development, or the risk factor or health behaviour was not modifiable. Google Translate was used to decipher any studies located in languages other than English.
## Data extraction
Papers were imported into Covidence systematic review software [20], and duplicates removed by automation. Titles and abstracts were screened by TC, AS and DS. Full-text articles were obtained for relevant studies and reviewed based on inclusion criteria by TC who then extracted data from each included paper. AS or DS randomly reviewed the extracted data of ten included studies for accuracy and completeness. Any conflicts were resolved by consensus.
Data extracted from each paper included: the authors and year, study design and duration, location, the preconception population, total number of participants, the paternal exposures (and exposure measures), paternal outcomes (and outcome measures), any covariates considered and the main results from each association reported.
## Quality assessment
The quality of each paper was assessed by TC using the Newcastle–Ottawa Scale (NOS). The NOS comprises three domains 1) selection of participants, 2) comparability of study groups, and 3) outcome of interest (cohort studies) or ascertainment of exposure (case–control studies), assigning stars in each domain to a maximum of nine stars [21]. Papers were then categorized as good quality (7–9 stars), fair quality (4–6 stars) or poor quality (0–3 stars) using groupings employed in previous research [22].
A meta-analysis was considered, but not possible due to exposure and outcome heterogeneity.
## Results
A total of 65 papers were included in this review (Fig. 1 – PRISMA Flowchart) [18], comprising cohort ($$n = 56$$) and case control studies ($$n = 9$$) (Table 2 – Summary Table) & (Table 3 – Summary Table Findings). The majority of papers were conducted in the USA ($$n = 18$$), Europe and the UK ($$n = 19$$), and China ($$n = 17$$), several papers were from Australia [23–29] or included an Australian health centre [30–34]. Approximately half of all papers ($$n = 29$$) included a sample size between 370 and 2,900, while others included > 20,000 ($$n = 11$$) or ≤ 200 participants ($$n = 13$$).Fig. 1PRISMA FlowchartTable 2Summary tableFirst Author & YearLocationDesign & DurationSampleExposure MeasurePaternal ExposureConfoundersOutcome MeasureOutcomeQuality±Body Composition Bowatte et al. 2022 [25]AustraliaCohort [Prospective] Tasmanian Longitudinal Health Study (TAHS) 1968—2021Mothers & Fathers ($$n = 836$$) of offspring ($$n = 1$$,938)Paternal height and weight obtained from school medical recordsBMI – BMI trajectory from early childhood (4–6 years) to late childhood (9–10 years) and adolescence (14–15 years)1)Maternal report of asthma at 14 years 2) Paternal report of asthma at 14 years 3) Grandfather or Grandmother ever asthma 4) *Smoking status* of Grandfather or Grandmother during paternal childhood 5) Grandfather’s occupation1) ‘Ever’ *Allergic asthma* 2) Asthma onset before 10 years old 3) Asthma onset after 10 years oldOffspring asthma5 Broadney et al. 2017 [35]USACohort [Retrospective] Upstate KIDS Study (Population-based) 2008–2010Mothers & Fathers ($$n = 2$$,974) of infants ($$n = 3$$,555)Maternal report of paternal weight & height on baseline questionnaire at 4 months postpartumBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Maternal age, 2) Race/ethnicity, 3) Education, 4) Private insurance, 5) Maternal smoking during pregnancy, 6) Alcohol use during pregnancy, 7) Parity, 8) Infant plurality, 9) Maternal pre-pregnancy BMIInflammatory biomarker [CRP] and Ig levelsInflammation & immune response of neonates6 Casas et al. 2017 [36]SpainCohort [Prospective] INfancia y Medio Ambiente- Environment and Childhood [INMA] (Population-based) 2003–2008Pregnant couples & their expectant children ($$n = 1$$,827)Maternal report of paternal weight & height at first prenatal visit approximately 14 weeks of gestationBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Parental age, 2) Sex of the child, 3) Parental education, 4) Parental social class, 5) Parity, 6) Maternal IQ, 7) Maternal employment status during pregnancy and at 5 years, 8) Breastfeeding duration, 9) Daycare attendance, 10) Child physical activity, 11) Maternal BMI1) McCarthy Scales of Children's Abilities (MSCA) [contexualized to Spanish], & 2) The attention deficit hyoperactivity disorder [ADHD] Criteria of Diagnostic and Statistical Manual of Mental Health Disorders—4th Edition (ADHD-DSM-IV)Neuropsychological development of preschool children around 5 years old9 Chen et al. 2021 [37]ChinaCohort [Retrospective] Women’s Hospital, School of Medicine, Zhejiang University (Hospital-based) 2013—2016Subfertile couples (Males [$$n = 2$$,318]) undergoing IVF/ICSI fresh embryo transfer cycles resulting in singletons ($$n = 1$$,366) and twins ($$n = 952$$)Third Party—Measurement of paternal weight and height by trained nurseBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1)Parental age, 2) type of infertility, 3) duration of infertility, 4) ovulatory dysfunction, 5) endometriosis, 6) maternal prepregnancy BMIInternational classification of Diseases, 10th Revision (ICD-10) into 9 subcategoriesBirth defect5 Fang et al. 2020 [38]ChinaCohort [Retrospective] National Free Preconception Health Examination Project (NFPHEP) (Population-based) 2012–2016Couples planning to conceive [Males [$$n = 50$$,927])Third Party—Measurement of paternal weight and height by physicianBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1)Age, 2) type of household, 3) education, 4) smoking, 5) alcohol consumption, 6) psychosocial pressure and ready for pregnancy 7) cycle regularity, 8) age of menarche, 9) gravidity, 10) spontaneous abortion, 11) induced abortionTime to pregnancy (TTP) = interval between the date of enrolment and last menstrual period (LMP)Fecundability5 Fleten et al. 2012 [39]NorwayCohort [Prospective] Norwegian Mother and Child cohort study (MoBa) (Population-based) 1999–2009Pregnant couples & their expectant children ($$n = 29$$,216)Paternal self-report of weight and height ($20\%$) OR maternal report of paternal weight and height ($80\%$) at approximately 17 weeks of gestationBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Parental educational level (years), 2) Paternal and maternal prenatal smoking, 3) Maternal coffee consumption during pregnancy, 4) Parental BMIBody mass index (BMI) at 3 years oldOffspring adiposity6 Guo et al. 2022 [40]ChinaCohort [Retrospective] National Free Pre-conception Check-up Projects (NFPCP) 2013–2017Nulliparous couples attempting pregnancy (Males [$$n = 4$$,719,813])Third Party—Physician measurement of paternal weight and heightBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared) during participation in the NFPCP1)Maternal and paternal age at last menstrual period, 2) Maternal and paternal height 3) Education level, 4) Parity, 5) Ethnicity, 6) Area of residence 7) Maternal Diabetes, 8) Maternal Hypertension, 9) Smoking 10) Alcohol use 11) Passive smoking 12) History of adverse pregnancy incl preterm birth, stillbirth, or spontaneous abortion in previous pregnancies1) Large-for-gestational- age (LGA) = birthweight above 90th percentile according to birthweight centiles for a Chinese population, & 2) Small-for-gestational-age (SGA) = birthweight below the tenth percentile on birthweight centiles for a Chinese populationOffspring birthweight6 Hoek et al. 2022 [41]The NetherlandsCohort [Prospective] Rotterdam Periconception Cohort (PREDICT Study) (Hospital-based) 2017–2019Subfertile couples (Males [$$n = 221$$]) undergoing IVF/ICSI with cultured embryos ($$n = 757$$)Third party—Anthropometric assessment completed by a trained nurse at baselineBMI—"Preconceptional" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Total motile sperm count [TMSC], 2) Age, 3) Ethnicity, 4) Smoking, 5) Alcohol use, 6) Education1) Fertilization rate, 2) TMSC 3) Embryo developmental morphokinetics, 4) Embryo quality assessed by a time-lapse prediction algorithm (KIDScore), & 5) Live birth rateIVF/ICSI induced live birth8 Johannessen et al. 2020 [33]Northern Europe Denmark Norway Sweden Iceland Estonia & Spain AustraliaCohort [Prospective] The Respiratory Health in Northern Europe, Spain and Australia multigeneration study (RHINESSA) (Population-based) 2013–2016Mothers & Fathers ($$n = 2044$$), of adult offspring ($$n = 2$$,822)Paternal self-report based upon validated figural drawing scale of 9 sex-specific silhouettesBMI—“Overweight status” To identify subjects at risk for overweight body size (BMI, 25–30 kg/m2) at 8 years old, at puberty, and at age 30 years before offspring conception1)*Paternal asthma* status, 2) Education level 3) Maternal overweight status 4) *Maternal asthma* status 5) Offspring sex 6) Offspring ageParent report in the RHINESSA questionnaireAdult offspring asthma with or without nasal allergies6 Lonnebotn et al. 2022 [34]Northern Europe Denmark Norway Sweden Iceland Estonia & Spain AustraliaCohort [Prospective] The Respiratory Health in Northern Europe, Spain and Australia multigeneration study (RHINESSA) (Population-based) 2013–2016Mothers & Fathers ($$n = 308$$) of adult offspring ($$n = 420$$)Paternal self-report based upon validated figural drawing scale of 9 sex-specific silhouettesBMI—“Overweight status” To identify subjects at risk for overweight body size (BMI, 25–30 kg/m2) at 8 years old and at puberty1)Maternal education 2) Paternal education 3) Offspring age 4) Smoking historyPre/post bronchodilator forced expiratory volume in one second (FEV1) & forced vital capacity (FVC)Adult offspring lung function7 Moss et al. 2015a [42]USALongitudinal cohort [Prospective] National Longitudinal Study of Adolescent Health (Add Health) 1994–2008Adolescents (grades 7 -12) followed into adulthood becoming Mothers & Fathers of infants ($$n = 372$$)Third party—Anthropometric assessment completed by a trained professional at baselineBMI—"Preconception" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Parents age at birth, 2) Race/ethnicity, 3) Immigrant status, 4) Education level, 5) Socioeconomic status, 6) Infant sex, 7) Initiation of prenatal care, 8) Parity, 9) Time between wave III interview and conception, 10) Relationship type at wave IIIRespondent self-report on Wave IV questionnaireGestational age & offspring birthweight7 Mutsaerts et al. 2014a [43]The NetherlandsCohort [Prospective] Groningen Expert Center for Kids with Obesity [GECKO] Drenthe cohort (Population-based) 2006–2007Pregnant couples & their expectant children ($$n = 2$$,264)Paternal self-report of weight and height on baseline questionnaire during third trimester or within 6 months postpartumBMI—"Prepregnancy" Body mass index [BMI] at conceptionNilQuestionnaire, shortly after birth, completed by midwife or gynaecologistSpontaneous preterm birth & Small for gestational age (SGA)3 Noor et al. 2019 [44]USALongitudinal cohort [Prospective] Project Viva birth cohort study of mothers and children 1999–2019Pregnant couples & their expectant children ($$n = 429$$)Maternal report of paternal weight & height at first prenatal visit approximately 10 weeks gestationBMI—"Periconception" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Maternal prepregnancy BMI, 2) Maternal Age, 3) Gestational weight gain, 4) Household income, 5) Maternal education, 6) Maternal smoking, 7) Maternal alcohol use, 8) Marital status, 9) Infant's sex, 10) Race/ethnicity, 11) Gestational age at delivery, 12) Mode of delivery, 13) Birth weight, 14) Batch effects, 15) Estimated nucleated cell types from cord blood 16) WBC'sBlood samples collected at birth, age 3 years & 7 yearsGenome-wide DNA methylation patterns and birthweight in offspring7 Pomeroy et al. 2015 [23]AustraliaCohort [Prospective] Mater-University of Queensland Study of Pregnancy (MUSP) 1982–1983Mothers and Fathers of infants ($$n = 1$$,041)Maternal report of paternal weight and height at first prenatal visit approximately 18 weeks of gestationBMI—"Pre-pregnancy" height & body mass index [BMI] (Weight in kilograms over height in meters squared)1) Parity, 2) Maternal education, 3) Maternal smoking in the last trimester, 4) Maternal age at birth1) Birthweight, 2) Neck-rump length 3) Head circumference, 3) Absolute and proportional limb segment and trunk lengths & 4) Subcutaneous fatNeonatal body measurements6 Retnakaran et al. 2021 [45]ChinaCohort [Prospective] Liuyang Preconception cohort 2009 -Newly married couples attempting pregnancy and their expectant children ($$n = 1$$,292)Third party—Anthropometric assessment completed by trained staff at baselineBMI—"Pregravid" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Age, 2) Years of education, 3) Smoking status, 4) BMI, 5) Household income 6) Length of gestation, 7) Total gestational weight gain, 8) Gestational diabetes, 9) Preeclampsia, & 10) Infant sex1) Large-for-gestational- age (LGA) = birthweight above 90th percentile according to birthweight centiles for a Chinese population, & 2) Small-for-gestational-age (SGA) = birthweightbelow the tenth percentile on birthweight centiles for a Chinese populationOffspring birthweight8 Robinson et al. 2020 [46]USACohort [Prospective] Upstate KIDS study (Population-based) 2008–2010Mothers and Fathers of children ($$n = 1$$,915)Maternal report of paternal weight & height on baseline questionnaire at 4 months postpartumBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Maternal & paternal age, 2) Insurance status, 3) Child sex, 4) Maternal race/ethnicity, 5) Education, 6) Marital status, 7) History of polycystic ovary syndrome (PCOS) and/or diagnosis, 8) Smoking, 9) Alcohol intake, 10) Maternal & paternal history of affective disorders, 11) BMI, 12) Maternal prepregnancy BMI1) Positive history of attention deficit hyperactivity disorder (ADHD) or anxiety disorder 2) Positive screening for ADHD and the inattentive or hyperactive/impulse sub scales OR report of clinical ADHD diagnosis 3) Parental report of child borderline behavioural problems at 7 or 8 years of ageOffspring behavioural problems and psychiatric symptoms at 7–8 years7 Sun et al. 2022 [47]ChinaCohort [Prospective] Hunan Maternal and Child Health Hospital (Hospital-based) 2013–2019Couples receiving antenatal care (Males [$$n = 34$$,104)Third party – Paternal height and weight measured at 14–16 weeks gestationBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1)Paternal age, 2) maternal age, 3) maternal BMI, 4) residence location, 5) education level, 6) nationality, 7) history of smoking, 8) history of drinking, 9) history of betel nut consumption, 10) history of drug use, 11) history of preterm birth, 12) per capita monthly household incomeDelivery before 37 weeks gestation & Birth weight < 2,500 gPreterm birth & Low birth weight7 Sundaram et al. 2017 [48]USACohort [Prospective] Longitudinal Investigation of Fertility and the Environment [LIFE]) 2005–2009Couples attempting pregnancy (Males [$$n = 501$$])Third party—Anthropometric assessment completed by a trained nurse at baselineBMI—Body mass index [BMI] (Weight in kilograms over height in meters squared) and waist/hip measurements1) Female partner's age, 2) Difference between the male and female age, 3) Both partner's smoking status, 4) Both partner's number of days of vigorous physical activity per week, 5) Both partner's free cholesterol level 6) Both partner's race 7) Both partner's education 8) Average acts of intercourse per menstrual cycle 9) Menstrual cycle regularityTime to pregnancy (TTP) in menstrual cyclesPregnancy8 Umul et al. 2015 [49]TurkeyCohort [Retrospective]Couples (Males [$$n = 155$$]) undergoing intracytoplasmic sperm injection (ICSI) cycles ($$n = 177$$)Third party—Anthropometric measurementsBMI—Body mass index [BMI] (Weight in kilograms over height in meters squared) during fertility treatmentNil1) Fertilization rate, 2) Implantation rate, 3) Clinical pregnancy rate, & 4) Live birth rateICSI induced live birth2 Wei et al. 2022 [50]ChinaCohort [Prospective] Hunan Provincial Maternal and Children Health Care Hospital (Hospital-based) 2013–2019Pregnant couples (Males [$$n = 40$$,650])Paternal self-report of weight and height on baseline antenatal questionnaire between 8- and 14-weeks’ gestationBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1)Maternal and paternal age 2) ethnicity, 3) educational level, 4) parity, 5) family income per month, 6) active smoking before pregnancy, 7) passive smoking before pregnancy, 8) alcohol consumption before pregnancy 9) folic acid consumption before or during pregnancy, 10) history of adverse pregnancy outcomes, 11) history of pregnancy complications, 12) gestational weight gain recommendation range, 13) pregnancy complications in this pregnancy, 14) smoking status before pregnancy, 15) alcohol consumption before pregnancyLow birth weight = < 2,500 g Very low birthweight = < 1,500 g Extremely low birthweight < 1,000 gOffspring birthweight6 Wei et al. 2021 [51]ChinaCohort [Retrospective] Guangxi Zhuang Birth Cohort (GZBC) (Hospital-based) 2015–2018Parents with singleton birth (Males [$$n = 1$$,082])Paternal self-report of weight and height at first antenatal interviewBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1)Parental age at delivery, 2) offspring sex, 3) gestational age, 4) offspring birth weight), 5) maternal residential place, 6) gravidity, 7) parity, 8) drinking before pregnancy, 9) maternal passive smoking during pregnancy, 10) pregnancy comorbidities or complications, 11) caesarean sectionReal-time polymerase chain reaction (qPCR)Newborn telomere length (TL)6 Xu et al. 2021 [52]ChinaCohort [Prospective] Shanghai Jiao Tong University 2015Pregnant couples and their expectant children ($$n = 1$$,810)Paternal self-report of weight and height at first prenatal visit approximately 16 weeks of gestationBMI—"Preconception" body mass index [BMI] (Weight in kilograms over height in meters squared) during fertility treatment1) Delivery gestational week, 2) Maternal age, 3) Gestational weight gain (GWG), 4) Education, 5) Parity, 6) Family history of metabolic diseases, 7) Haemoglobin, 8) Systolic blood pressure, 9) Diastolic blood pressure, 10) Dyslipidemia, 11) *Fasting plasma* glucose at the first prenatal check-up 12) Offspring sex 13) Preconception BMIAssessed within 1 h of birth using digital scalesOffspring birthweight7 Yang et al. 2015 [53]ChinaCase–control [Retrospective] (Population-based) 2011–2013Mothers & Fathers of cases ($$n = 870$$) and controls ($$n = 5$$,471)Paternal self-report of weight and height at postpartum baseline interviewBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared)1) Infant's gender, 2) Gestational age, 3) Parental age, 4) Family income, 5) Parental education level, 6) Gravidity, 7) Parity, 8) Paternal smoking status during pregnancy, 9) Parental prepregnancy weight,10) Parental height, 11) Parental BMI, 12) Maternal alcohol consumption during pregnancy, 13) Maternal weight gain during pregnancy, 14) Maternal BMI gain during pregnancyLive macrosomic birth (> 4,000 g)Macrosomia6 Zalbahar et al. 2017 [24]AustraliaCohort [Prospective] Mater-University of Queensland Study of Pregnancy (MUSP) 1981–1983Mothers and Fathers of infants ($$n = 1$$,494)Maternal report of paternal weight and height at first prenatal visit at approximately 18 weeks of gestationBMI—"Pre-pregnancy" weight and body mass index [BMI] (Weight in kilograms over height in meters squared)1) Parental education, 2) Family annual income, 3) Maternal gestational weight gained, 4) Maternal smoking habit, 5) Offspring birth weight, 6) Offspring gender, 7) Gestational age, 8) Breastfeeding duration, 9) Offspring's lifestyle at 14 years, 10) Maternal or paternal BMI, 11) Maternal age at birth, 12) Offspring birth weight, 13) Offspring genderPhysical assessment using measuring tape and digital scales at 5, 14 and 21 year follow-upsOffspring weight & BMI changes from childhood (5 years) into adulthood (21 years)5 Zhang et al. 2020 [54]ChinaCohort [Retrospective] National Free Pre-conception Check-up Projects (NFPCP) 2015–2017Nulliparous couples attempting pregnancy (Males [$$n = 2$$,301,782])Third Party—Physician measurement of paternal weight and heightBMI—"Pre-pregnancy" body mass index [BMI] (Weight in kilograms over height in meters squared) during participation in the NFPCP1) Age, 2) Ethnic background, 3) Educational level, 4) Occupation, 5) Household registration and region, 6) Alcohol intake, 7) Tobacco exposure, 8) Hypertension, 9) HBsAg positive status based on male individual model ATime to pregnancy (TTP) = [Date of the last menstruation (pregnant couples) or Date of the most recent follow-up (nonpregnant couples) -Date of baseline questionnaire completion)]/Average menstrualcycle length] + 1Pregnancy9Alcohol Luan et al. 2022 [55]ChinaCohort [Prospective] Shanghai-Minhang Birth Cohort Study 2012 -Mothers and Fathers of infants ($$n = 796$$)Maternal report of paternal preconception alcohol consumption at 12–16 weeks gestationAlcohol – 3 months before conception1)Paternal age 2) Paternal BMI 3) Paternal education 4) Paternal smoking 5) Maternal age 6) Parity 7) Maternal depressive symptoms during pregnancy 8) Maternal preconception folic acid supplements, 9) Multivitamin supplements during pregnancy 10) Gestational weeks 11) SexChild Behaviour Checklist (CBCL) at offspring ages 2, 4, & 6 years oldOffspring behavioural problems7 Milne et al. 2013 [26]AustraliaCase–control [Retrospective] Aus-ALL 2003–2006 Aus-CBT 2005–2010Mothers and Fathers of children with ALL (Cases [$$n = 281$$] Controls [$$n = 672$$) & CBTs (Cases [$$n = 221$$]) and Controls [$$n = 717$$]Paternal self-report on baseline questionnaireAlcohol –Any alcohol 12 months before pregnancy1)Year of birth group 2) Maternal age, 3) Ethnicity 4) Household income 5) Birth order 6) Maternal smoking 7) Child’s age 8) Child’s sex 9) State of residence 10) Paternal smoking 11) Paternal age group 12) Household incomeDiagnosis from one of 10 paediatric oncology centres in AustraliaChildhood acute lymphoblastic leukemia (ALL) & Childhood brain tumours (CBTs)6 Moss et al. 2015a [42]USALongitudinal cohort [Prospective] National Longitudinal Study of Adolescent Health (Add Health) 1994–2008Adolescents (grades 7 -12) followed into adulthood becoming Mothers & Fathers of infants ($$n = 372$$)Paternal self-report of health behaviours at wave III interviewAlcohol—preconception intake greater than once a month1) Parents age at birth, 2) Race/ethnicity, 3) Immigrant status, 4) Education level, 5) Socioeconomic status, 6) Infant sex, 7) Initiation of prenatal care, 8) Parity, 9) Time between wave III interview and conception, 10) Relationship type at wave IIIRespondent self-report on Wave IV questionnaireGestational age & offspring birthweight7 Mutsaerts et al. 2014a [43]The NetherlandsCohort [Prospective] Groningen Expert Center for Kids with Obesity [GECKO] Drenthe cohort (Population-based) 2006–2007Pregnant couples & their expectant children ($$n = 2$$,264)Paternal self-report on baseline questionnaire during third trimester or within 6 months following deliveryAlcohol intake (units/week) 6 months prior to conception and up to deliveryNilQuestionnaire, shortly after birth, completed by midwife or gynaecologistSpontaneous preterm birth & Small for gestational age (SGA)3 Xia et al. 2018 [56]ChinaCohort [Prospective] Shanghai-Minhang Birth Cohort Study 2012Mothers and Fathers of infants ($$n = 980$$)Paternal self-report at baseline interview between 12 to 16 weeks of gestationAlcohol—intake at least once a week 3 months before conception1) Parental age, 2) Parental BMI before conception, 3) Gestational age, 4) Gravidity, 5) Birth weight of offspring, 6) Paternal education, 7) Maternal passive smoking before conception (yes/no), 8) Paternal smoking (yes/no), 9) Days between birth and 12-month measurementMales—AGD-AP (centre of anus to penis) AGD-AS (centre of anus to scrotum) Females—AGD-AC (centre of anus to clitrous) AGD-AF (centre of anus to fourchette)Offspring anogenital distance (AGD)8 Zuccolo et al. 2016 [57]NorwayCohort [Prospective] The Norwegian Mother and Child Cohort Study (MoBa) (Population based) 1999–2009Mothers & Fathers of children ($$n = 68$$,244)Paternal self-report on baseline questionnaire at approximately 17 weeks of gestationAlcohol—intake in the 6 months prior to pregnancy and up to week 18 of gestation1) Year of birth, 2) Folic acid use around conception, 3) Whether the pregnancy was planned, 4) Maternal diabetes, 5) Parity, 6) Ethnicity, 7) Financial strain, 8) Parental age, 9) Height, 10) BMI, 11) Gross income, 12) Education, 13) Smoking/drug use in pregnancy, 14) Other parent's exposureSex-standardised head circumference (expressed as standard deviation [SD] scores), based on the distribution of all MoBa newborns by sexOffspring head circumference4Cannabis Har-Gil et al. 2021 [58]CanadaCohort [Retrospective] (Clinic-based) 2016–2019Female ($$n = 15$$) & male ($$n = 53$$) cannabis users & non-users ($$n = 654$$) undergoing IVFPaternal self-report on baseline questionnaireCannabis—use prior to fertility treatmentNil1) Sperm volume 2) Sperm quality, 3) Fertilization rate 4) Implantation rate (IR) 5) Ongoing pregnancy rate (OPR)IVF/ICSI induced live birth2 Kasman et al. 2018 [59]USACross sectional cohort [Retrospective] National Survey of Family Growth (NSFG) (Population-based) 2002–2015Female ($$n = 1$$,076) & male ($$n = 758$$) respondents of the National Survery of Family Growth (NSFG)Paternal self-report at baseline interviewCannabis—use over the previous 12 months1) Age, 2) Marital status, 3) Previous children, 4) Partner age (for men), 5) Previous fertility evaluation/treatment, 6) Year of survey, 7) Income, 8) Race, 9) EducationEstimated time to pregnancy (TTP) using the current-duration appaorachPregnancy6 Moss et al. 2015a [42]USALongitudinal cohort [Prospective] National Longitudinal Study of Adolescent Health (Add Health) 1994–2008Adolescents (grades 7 -12) followed into adulthood becoming Mothers & Fathers of infants ($$n = 372$$)Paternal self-report of health behaviours at wave III interviewCannabis—use in the last 12 months1) Parents age at birth, 2) Race/ethnicity, 3) Immigrant status, 4) Education level, 5) Socioeconomic status, 6) Infant sex, 7) Initiation of prenatal care, 8) Parity, 9) Time between wave III interview and conception, 10) Relationship type at wave IIIRespondent self-report on Wave IV questionnaireGestational age & offspring birthweight7 Nassan et al. 2019 [60]USACohort [Prospective] Environment and Reproductive Health Study [EARTH] 2005–2017Subfertile couples (Males [$$n = 200$$]) undergoing IVF cycles ($$n = 368$$)Paternal self-report on baseline questionnaireCannabis—use ever1) Age, 2) Race, 3) BMI, 4) Tobacco smoking, 5) Coffee and alcohol consumption, 6) Cocaine use1) Implantation, 2) Clinical pregnancy, 3) Live birth per assisted reproductive technology (ART) cycle, & 4) Pregnancy lossIVF/ICSI induced live birth7 Wise et al. 2018 [61]USACohort [Prospective] Preconception pregnancy planner cohort study online (PRESTO) 2013–2017Couples attempting pregnancy (Males $$n = 1$$,125)Paternal self-report on baseline questionnaireCannabis—use in the previous 2 months1) Age, 2) Race/ethnicity, 3) Education, 4) Annual household income, 5) Cigarette smoking history, 6) Alcohol intake, 7) Caffeine intake, 8) Intercourse frequency, 9) Doing something to improve chances of conception, 10) PSS-10 score, 11) MDI score, 12) Sugar-sweetened soda intake, 13) Average sleep duration 14) Employment statusTime to pregnancy (TTP) = (Menstrual cycles of attempt at study entry) + [(Last menstrual period [LMP] date from the most recent followup questionnaire − date of baseline questionnaire completion)/usualmenstrual cycle length] + 1Fecundability6Physical activity Moss et al. 2015a [42]USALongitudinal cohort [Prospective] National Longitudinal Study of Adolescent Health (Add Health) 1994–2008Adolescents (grades 7 -12) followed into adulthood becoming Mothers & Fathers of infants ($$n = 372$$)Paternal self-report of health behaviours at wave III interviewPhysical activity—sessions in the last week1) Parents age at birth, 2) Race/ethnicity, 3) Immigrant status, 4) Education level, 5) Socioeconomic status, 6) Infant sex, 7) Initiation of prenatal care, 8) Parity, 9) Time between wave III interview and conception, 10) Relationship type at wave IIIRespondent self-report on Wave IV questionnaireGestational age & offspring birthweight7 Mutsaerts et al. 2014a [43]The NetherlandsCohort [Prospective] Groningen Expert Center for Kids with Obesity [GECKO] Drenthe cohort (Population-based) 2006–2007Pregnant couples & their expectant children ($$n = 2$$,264)Paternal self-report on baseline questionnaire during third trimester or within 6 months following deliveryPhysical activity—moderate intensity for 30 min per day ≥ once a week 6 months prior to conception and up to deliveryNilQuestionnaire, shortly after birth, completed by midwife or gynaecologistSpontaneous preterm birth & Small for gestational age (SGA)3Smoking Accordini et al. 2021 [32]Northern Europe Denmark Norway Sweden Iceland Estonia & Spain AustraliaCohort [Prospective] The Respiratory Health in Northern Europe, Spain and Australia multigeneration study (RHINESSA) (Population-based) 2013–2016Mothers & Fathers ($$n = 274$$), investigated in the European Community Respiratory Health Survey (ECRHS), of adult offspring ($$n = 383$$)Paternal self-report at baseline interview and ECRHS examinationsSmoking – Prepubertal smoking [smoking < 15 years old] & smoking ≥ 15 years old1)Grand parents education level 2) Paternal age 3) Paternal education level 4) Paternal occupational class 5) Maternal smoking before or after offspring birth 6) Offspring age 7) Offspring sex 8) Offspring education level 9) Offspring smokingPre/post bronchodilator forced expiratory volume in one second (FEV1) & forced vital capacity (FVC)Adult offspring lung function8 Accordini et al. 2018 [31]Northern Europe Denmark Norway Sweden Iceland Estonia & Spain AustraliaCohort [Prospective] European Community Respiratory Health Survey (ECRHS) (Population-based) 1998–2013Mothers and Fathers ($$n = 1$$,964) of adult offspring ($$n = 4$$,192)Paternal self-report at baseline interview and ECRHS examinationsSmoking –Prepubertal smoking [smoking < 15 years old] & smoking ≥ 15 years old1)Grandmother smoking 2) Father’s ever asthma 3) Education level 4) Smoking initiation 5) Offspring gender 6) AgeParent report in the ECRHS questionnaireAdult offspring asthma with or without nasal allergies7 Carslake et al. 2016 [62]NorwayCombined cohort [Prospective] HUNT Study [Adult ≥ 20 years] (1984 – 2008)/ YoungHUNT Study [Child 13–19 years] (1995 – 2007)Mothers and Fathers ([HUNT] of offspring [YoungHUNT] ($$n = 221$$)Paternal self-report at baseline interviewSmoking –Prepubertal smoking [smoking < 11 years old]1)Offspring birth order 2) Maternal education 3) Paternal employment 4) Maternal and Paternal smoking status at time of offspring conception 5) Offspring sexBody Mass Index (BMI)Offspring adiposity6 Deng et al. 2013 [63]ChinaCase–control [Retrospective] Gene-environmental interaction study on CHD occurrence (Hospital-based) 2010–2011Pregnant couples & their expectant children as CHD cases ($$n = 267$$) & controls ($$n = 386$$)Maternal report at baseline interview during pregnancy but after prenatal diagnosis of CHDSmoking—"Periconceptional" being 3 months before conception through to the first trimester of pregnancy1) Maternal residence, 2) Age, 3) Education, 4) Prepregnancy BMI, 5) Parental alcohol use during the 3 months before and 3 months after conception, 6) Folic acid intake during the 3 months before and 3 months after conception, 7) Family history of CHD, 8) ParityDiagnosed via prenatal echocardiographyCongenital heart defects (CHD) in offspring8 Frederiksen et al. 2020 [64]Costa RicaCase–control [Retrospective] Costa Rican Childhood Leukemia Study (CRCLS) (Population-based) 2001–2003Mothers and Fathers ($$n = 198$$) of offspring suffering leukemia ($$n = 292$$) [Cases] & cancer free age matched offspring ($$n = 578$$) [controls]Paternal self-report at baseline interviewSmoking –Tobacco smoking 12 months before conception1)Child sex 2) Birth year 3) Parental education 4) Paternal age 5) Maternal smokingDiagnosis, between 1995–2000 in Costa Rica while aged < 15 years, of Acute Lymphoblastic Leukemia (ALL) ($$n = 252$$) or Acute Myeloid Leukemia (AML) ($$n = 40$$)Childhood leukemia7 Knudsen et al. 2020 [30]Northern Europe Denmark Norway Sweden Iceland Estonia & Spain AustraliaCohort [Prospective] The Respiratory Health in Northern Europe, Spain and Australia multigeneration study (RHINESSA) (Population-based) 2013–2016Mothers & Fathers ($$n = 2$$,111) of adult offspring ($$n = 2$$,939)Paternal self-report at baseline interview and examinationsSmoking –Prepubertal smoking [smoking before 15 years old] & smoking ≥ 15 years old. Preconception smoking [≥ 2 years before offspring birth year]1)Parental education 2) offspring sex1) BMI [weight (kg)/height (m)2] 2) Bioelectrical impedance analysis 3) Fat mass index (FMI) [fat mass (kg)/height (m)2Adult offspring BMI index and FMI index5 Ko et al. 2014 [65]TaiwanLongitudinal cohort [Prospective] Taiwan Birth Cohort Study (National) 2005–2006Mothers & Fathers of infants ($$n = 21$$,248)Maternal report at baseline interview 6 months postpartumSmoking—Preconception tobacco being before pregnancy and up to four months postpartum1) Maternal age, 2) Nationality, 3) Education, 4) Parity, 5) Total weight gain during pregnancy, 6) Infant gender, 7) Multifetus, 8) Paternal smoking in the same period1) Low Birth weight (LBW) < 2,500 g, 2) Small for gestational age (SGA)—Birth below the 10th percentile of gender-specific birth weight for gestational age based on the 1998–2002 nationwide percentiles & 3) Preterm birth < 37 weeksOffspring birthweight & incidence of preterm delivery5 Milne et al. 2013 [27]AustraliaCase–control [Retrospective] The Australian Study of Childhood Brain Tumors (Aus-CBT) (Population-based) 2005–2010Mothers and Fathers ($$n = 1048$$) of children with childhood malignancy and brain tumors (CBT) ($$n = 247$$) & controls ($$n = 801$$)Paternal self-report on mailed questionnaireSmoking—Average number of cigarettes smoked per day in each calendar year from age 15 until year after child’s birth1)Child’s ethnicity, 2) year of birth group, 3) Mother’s age group, 4) Father’s age group, 5) alcohol consumption during pregnancy, 6) household incomeDiagnosis at one of 10 Australian paediatric oncology centresChildhood brain tumors (CBT)5 Moss et al. 2015a [42]USALongitudinal cohort [Prospective] National Longitudinal Study of Adolescent Health (Add Health) 1994–2008Adolescents (grades 7 -12) followed into adulthood becoming Mothers & Fathers of infants ($$n = 372$$)Paternal self-report of health behaviours at wave III interviewSmoking—At least one cigarette per day over the last 30 days1) Parents age at birth, 2) Race/ethnicity, 3) Immigrant status, 4) Education level, 5) Socioeconomic status, 6) Infant sex, 7) Initiation of prenatal care, 8) Parity, 9) Time between wave III interview and conception, 10) Relationship type at wave IIIRespondent self-report on Wave IV questionnaireGestational age & offspring birthweight7 Mutsaerts et al. 2014a [43]The NetherlandsCohort [Prospective] Groningen Expert Center for Kids with Obesity [GECKO] Drenthe cohort (Population-based) 2006–2007Pregnant couples & their expectant children ($$n = 2$$,264)Paternal self-report on baseline questionnaire during third trimester or within 6 months following deliverySmoking- cigarettes per day in the 6 months prior to conception and up to deliveryNilQuestionnaire, shortly after birth, completed by midwife or gynaecologistSpontaneous preterm birth & Small for gestational age (SGA)3 Northstone et al. 2014 [66]UKCohort [Prospective] The Avon Longitudinal Study of Parents and Children (ALSPAC) 1991–1992Pregnant couples where fathers identified as smoking regularly ($$n = 5$$,376) including before 11 years old ($$n = 166$$)Paternal self-report on baseline questionnaire completed during pregnancySmoking—Prepubertal tobacco before 11 years of age1) Parity of the mother at the time of birth of the offspring (primiparae vs multiparae), 2) Highest maternal education level 3) Housing tenure 4) Maternal smoking during pregnancy 5) Paternal smoking at conception1) BMI, 2) Waist circumference, 3) Total-body fat mass, & 4) Lean massOffspring adiposity7 Orsi et al. 2015 [67]FranceCase–Control [Retrospective] ESTELLE study (Population-based)2010—2011Mothers and fathers ($$n = 247$$) of offspring suffering childhood acute leukemia (CL) ($$n = 69$$) [Cases] & cancer free age matched offspring ($$n = 178$$) [Controls]Paternal self-report on baseline questionnaireSmoking –Tobacco smoking during the 3-month period preceding conception; the “pre-conception period”1)Offspring Age 2) Offspring Sex 3) Mother’s age at child’s birth 4) Mother’s education 5) Birth orderDiagnosed with CL < 15 years old as per the National Registry of Childhood Hematopoietic Malignancies (NRCH) criteriaChildhood acute leukemia (CL)7 Sapra et al. 2016 [68]USACohort [Prospective] Longitudinal Investigation of Fertility and the Environment [LIFE]) 2005–2009Couples attempting pregnancy (Males [$$n = 501$$])Paternal self-report at baseline interviewSmoking—Lifetime exposure to tobacco products (including cigarettes, electronic cigarettes, cigars, pipes, waterpipes, chewing tobacco, snuff and dip)1) Race/ethnicity, 2) Education, 3) Income, 4) Age, 5) Alcohol use, 6) Caffeine use, 7) BMI, 8) *Blood cadmium* in each partner, 9) Couple's mean age, 10) Difference in partners' agesTime to pregnancy (TTP) in menstrual cyclesPregnancy7 Svanes et al. 2017 [69]Northern Europe Norway, Sweden, Iceland, Denmark, EstoniaCombined Cohort [Prospective] European Community Respiratory Health Survey (ECRHS) (1989–1992) & Respiratory Heath in Northern Europe (RHINE) (Population-based) 1991—2012Mothers and Fathers ($$n = 3$$,777) of offspring aged 2–51 years ($$n = 24$$,168)Paternal self-report on RHINE III questionnaireSmoking – Tobacco smoking prior to conception including period around birth1)Age 2) Study centre 3) Parental age 4) *Parental asthma* before age 10, 5) Parental educationDiagnosis via parental reportOffspring asthma before/after 10 years6 Wang et al. 2022 [70]ChinaCohort [Retrospective] National Free Pre-Pregnancy Checkups Project (NFPCP) (Population-based) 2010–2016Non-smoking women and their smoking husbands ($$n = 190$$,529)Paternal self-report at preconception health examinationSmoking—Tobacco while attempting conception in the following 6 months1) Maternal and paternal age at last menstrual period, 2) Higher education, 3) Han ethnicity, 4) Preconception body mass index (BMI), 5) Alcohol drinking, 6) Parental passive smoking, 7) History of adverse pregnancy outcomes, 8) Region of the service stationDelivery before 37 completed gestational weeksPreterm birth (PTB)5 Wang et al. 2018 [71]ChinaCohort [Retrospective] National Free Pre-Pregnancy Checkups Project (NFPCP) (Population-based) 2010–2016Non-smoking women and their husbands ($$n = 5$$,770, 691)Paternal self-report at preconception health examinationSmoking—Tobacco while attempting conception in the following 6 months1) Maternal and paternal age at last menstrual period, 2) Higher education, 3) Han ethnicity, 4) Preconception body mass index (BMI), 5) Alcohol drinking, 6) Parental passive smoking, 7) History of adverse pregnancy outcomes, 8) Region of the service stationFetal death before week 28 of gestationSpontaneous abortion (SA)6 Wesselink et al. 2019 [72]USACohort [Prospective] Preconception pregnancy planner cohort study online (PRESTO) 2013–2018Couples attempting pregnancy (Males $$n = 1$$,411)Paternal self-report on baseline questionnaireSmoking—Tobacco while attempting conception for ≤ 6 menstrual cycles1) Age, 2) Race/ethnicity, 3) Education, 4) Annual household income, 5) BMI, 6) Sugar sweetened beverage intake, 7) Healthy eating index score, 8) Multivitamin or folic acid supplement use, 9) Sleep duration, 10) PSS-10 score, 11) MDI score, 12) Parity, 13) Intercourse frequency, 14) Doing something to improve chances of conceptionPregnancy attempt time = (Menstrualcycles of attempt time at baseline) + [(Last menstrual period [LMP] date from most recent followup questionnaire—date of baseline questionnaire)/Cycle length] + 1Fecundability5 You et al. 2022 [73]ChinaCohort [Prospective] Children lifeway Cohort 2018 -Mothers and Fathers ($$n = 1$$,037) of first grade students (7–8 years old)Paternal self-report at baseline interviewSmoking—Tobacco smoking before conception1)Sex 2) Actual age 3) Father overweight, 4) Mother overweight 5) Percentage of food expenditure 6) Educational level of parents 7) Caesarean Sect. 8) Birthweight 9) Breastfeeding 10) Other household smoking 11) Mother exposed to SHS during pregnancy 12) Picky eaters 13) TV watching time 14) physical exercise 15) Frequency of eating fried/baked food 16) Late-night dinners 17) Vegetable and fruit 18) Snack consumptionAge and sex specific BMI cut-off points according to the growth standard of China “Screening for overweight and obesity among school-age children and adolescents”Offspring overweight/obesity7 Zhou et al. 2020 [74]ChinaCohort [Prospective] National Preconception Health Care Project (NPHCP) (Population-based) 2010–2012 * with matched case controlCouples attempting pregnancy (Males [$$n = 566$$,439])Paternal self-report at baseline interviewSmoking—Tobacco smoking before conception1) Maternal age, 2) Education, 3) Occupation, 4) Residence status, 5) Self-reported medical history, 6) Smoking, 7) Second hand smoking, 8) Alcohol consumption, 9) Folic acid supplement, 10) Paternal alcohol consumption[*Primary] Birth defects = diagnosis on hospital records of first 42 days after delivery [*Secondary] Birth defect types = congenital heart disease, limb anomalies, clefts, digestive tract anomalies, gastroschisis and neural tube defectsOffspring birth defects7 Zwink et al. 2016 [75]GermanyCase–control [Retrospective] (Population based) 2009-OngoingMothers & Fathers of cases ($$n = 158$$) and controls ($$n = 474$$)Maternal report on baseline interview at approximately 8 years postpartumSmoking -"Periconceptional" tobacco being 3 months before conception until the fourth month of pregnancy1) Gender, 2) Birth year of the child, 3) Maternal age, 4) BMI, 5) Maternal body weightDiagnosis of 1) *Esophageal atresia* with or without tracheoesophageal fistula (EA/TEF) or 2) Anorectal malformations (ARM) ARM'sOffspring malformations4Stress Bae et al. 2017 [76]USACohort [Prospective] Longitudinal Investigation of Fertility and the Environment [LIFE]) (Population-based) 2005–2009Couples attempting pregnancy and their expectant children ($$n = 235$$)Paternal self-report at baseline interview assessed by Cohen's Perceived Stress Scale [PSS-4]Stress—& lifetime history of physician-diagnosed anxiety and/or mood disorders1) Age, 2) Serum cotinine, 3) Annual income, 4) Maternal paritySecondary sex ratio (SSR) [Males:Females at birth]Offspring sex6 Mutsaerts et al. 2014a [43]The NetherlandsCohort [Prospective] Groningen Expert Center for Kids with Obesity [GECKO] Drenthe cohort (Population-based) 2006–2007Pregnant couples & their expectant children ($$n = 2$$,264)Paternal self-report on baseline questionnaire during third trimester or within 6 months following deliveryStress—Paid working hours < 16 h per weekNilQuestionnaire, shortly after birth, completed by midwife or gynaecologistSpontaneous preterm birth & Small for gestational age (SGA)3 Wesselink et al. 2018 [77]USACohort [Prospective] Preconception pregnancy planner cohort study online (PRESTO) 2013–2018Couples attempting pregnancy (Males $$n = 1$$,272)Paternal self-report on baseline questionnaire assessed by the Perceived stress scale [PSS]Stress—Perceived stress in the last month1) Age, 2) BMI, 3) Race/ethnicity, 4) Education, 5) Household income, 6) Employment status, 7) Work duration, 8) Physical activityPregnancy attempt time = (Menstrualcycles of attempt time at baseline) + [(Last menstrual period [LMP] date from most recent followup questionnaire—date of baseline questionnaire)/Cycle length] + 1Fecundability7Nutrition Bailey et al. 2014 [29]AustraliaCase–control [Prospective] The Australian Study of Causes of acute lymphoblastic leukemia (ALL) in children (Aus-ALL). ( Population-based) 2003–2007Mothers and Fathers of children with ALL ($$n = 285$$) and controls ($$n = 595$$)Paternal self-report on food frequency questionnaire (FFQ)Folate & Vitamins B6/B12—during the 6 months before conception1)Birth order 2) best parental education, 3) paternal age, 4) paternal smoking in the conception year, 5) year of agreement and FFQ version, 6) supplement use (folate, B6, or B12), 7) control state, 8) control sex, 9) control ageDiagnosis at one of 10 Australian paediatric oncology centresChildhood acute lymphoblastic leukemia (ALL)5 Greenop et al. 2015 [28]AustraliaCase–control [Retrospective] The Australian Study of Childhood Brain Tumors (Aus-CBT) (Population-based) 2005–2010Mothers and Fathers ($$n = 866$$) of children with childhood malignancy and brain tumors (CBT) ($$n = 237$$) & controls ($$n = 629$$)Paternal self-report on food frequency questionnaire (FFQ)Folate & Vitamins B6/B12—during the 6 months before conception1)Control age, 2) control sex, 3) control state of residence, 4) child’s year of diagnosis/recruitment, 5) paternal age, 6) best parental education, 7) child’s ethnicity, 8) paternal preconceptional high alcohol consumptionDiagnosis at one of 10 Australian paediatric oncology centresChildhood brain tumors (CBT)5 Hatch et al. 2018 [78]USACohort [Prospective] Preconception pregnancy planner cohort study online (PRESTO) 2013–2017Couples attempting pregnancy (Males $$n = 1$$,045)Paternal self-report on food frequency questionnaire (FFQ) at baselineSugar sweetened beverage intake –Servings per week in the past month1)Male and female age, 2) male and female BMI, 3) age, 4) race/ethnicity, 5) education, 6) annual household income, 7) smoking history, 8) BMI, 9) physical activity, 10) caffeine intake, 11) alcohol intake, 12) sleep duration, 13) perceived stress scale score, 14) intercourse frequencyTime to pregnancy (TTP) [(menstrual cycles of attempt time at baseline) + [(LMP date from most recent follow-up questionnaire—date of baseline questionnaire)/cycle length] + 1]Fecundability6 Hoek et al. 2019 [79]The NetherlandsCohort [Prospective] Rotterdam Periconception Cohort (PREDICT Study) (Hospital-based) 2010–2015Pregnant couples ($$n = 511$$) producing spontaneous pregnancy ($$n = 303$$) or IVF/ICSI pregnancy ($$n = 208$$)Paternal self-report on baseline questionnaireFolate—"Periconceptional" status being 14 weeks before pregnancy and up to 10 weeks of gestation1) Gestational age at the time of ultrasound, 2) Paternal age, 3) Paternal smoking and alcohol, 4) Geographic origin, 5) Maternal age, 6) Maternal BMI, 7) Maternal smoking and alcohol, 8) Parity, 9) RBC folate levels, 10) Education level, 11) Geographic origin, 12) Fetal gender1) Crown-rump length (CRL) & 2) Embryonic volume (EV) at 7, 9 and 11 weeks of gestationEmbryonic growth trajectories7 Lippevelde et al. 2020 [80]NorwayCombined cohort [Prospective] Young-Health Study in Nord-Trondelag (Young-HUNT 1 1995–1997 & Young-HUNT 3 2006–2008)Adolescents (13–19 years old) followed into adulthood becoming Mothers & Fathers of infants. Young-HUNT 1 Father-offspring dyads ($$n = 2$$,140). Young-HUNT 3 Father-offspring dyads ($$n = 391$$)Adolescent self-report on baseline questionnaireDiet—Dietary exposures during adolescence1) Adolescents age, 2) BMI z-score 3) Education plans 4) Chewing tobacco use 5) Smoking 6) Alcohol use1) Birthweight (g) 2) Length (cm) 3) Head circumference (cm) 4) Placenta weight (g), 5) Gestational length (weeks) & 6) Ponderal index—Adiposity ([Birthweight (g) /Birth length3 (cm)]*100)Neonatal health of offspring8 Martin-Calvo et al. 2019 [81]USACohort [Prospective] Environment and Reproductive Health Study [EARTH] 2007–2017Subfertile couples undergoing fertility treatment (Males $$n = 108$$) producing singletons [$$n = 85$$), twins ($$n = 54$$) & triplets [$$n = 3$$])Paternal self-report on baseline food frequency questionnaire (FFQ)Folate—Preconception intake prior to or up to 12 weeks after the day of peak oestradiol concentration during a fertility treatment cycle (IVF/ICSI/IUI)1) Age, 2) Choline, betaine, methionine, vitamin B6, vitamin B12, 3) Total energy intake, 4) Diet quality, 5) Maternal BMI, 6) Maternal smoking status, 7) Infertility diagnosis, 8) Type of fertility treatment1) Gestational age at delivery (days), 2) Live birth of a neonate ≥ 24 weeks of gestation, & 3) Gestational age-adjusted birthweightIVF/ICSI/IUI induced live birth7 Mitsunami et al. 2021 [82]USACohort [Prospective] Environment and Reproductive Health Study [EARTH] 2007–2018Subfertile couples (Males [$$n = 231$$]) undergoing IVF cycles ($$n = 407$$)Paternal self-report on baseline food frequency questionnaire (FFQ)Diet—patterns 1 (processed foods) & 2 (whole/unprocessed foods) over the previous 12 months1) Men's age, 2) Total caloric intake, 3) BMI, 4) Race, 5) Smoking status, 6) Education level, 7) Physical activity, 8) Women's age + BMI, 9) Couple's primary infertility diagnosis, 10) Treatment protocol, 11) Women's adherence to the two dietary patterns, 12) Women's race, 13) Women's smoking status1) Fertilization rate, 2) Probability of implantation, 3) Clinical pregnancy, & 4) Probability of live birth per initiated treatment cycleIVF/ICSI induced live birth7 Moss et al. 2015a [42]USALongitudinal cohort [Prospective] National Longitudinal Study of Adolescent Health (Add Health) 1994–2008Adolescents (grades 7 -12) followed into adulthood becoming Mothers & Fathers of infants ($$n = 372$$)Paternal self-report of health behaviours at wave III interviewDiet—Fast food consumption1) Parents age at birth, 2) Race/ethnicity, 3) Immigrant status, 4) Education level, 5) Socioeconomic status, 6) Infant sex, 7) Initiation of prenatal care, 8) Parity, 9) Time between wave III interview and conception, 10) Relationship type at wave IIIRespondent self-report on Wave IV questionnaireGestational age & offspring birthweight7 Oostingh et al. 2019 [83]The NetherlandsCohort [Prospective] Rotterdam Periconception Cohort (PREDICT Study) (Hospital-based) 2010–2016Pregnant couples (Males [$$n = 638$$])Paternal self-report on baseline food frequency questionnaire (FFQ) before 8 weeks of gestationDiet—Habitual food intake and dietary patterns in a four week period during periconception being 14 weeks before and up to 10 weeks following conception1) Gestational age, 2) Maternal and paternal total energy intake, 3) Maternal and paternal BMI, 4) Maternal age, 5) Maternal and paternal smoking, 6) Nulliparous, 7) Fetal gender1) Longitudinal crown-rump length (CRL), & 2) Embryonic volume (EV), via transvaginal ultrasound, at 7, 9 and 11 weeks of gestationFirst trimester embryonic growth6 Twigt et al. 2012 [84]The NetherlandsCohort [Prospective] ‘*Achieving a* Healthy Pregnancy’ (AHP) (Hospital-based) 2007–2010Subfertile couples (Males [$$n = 199$$]) with IVF treatment and embryo transfer within 6 months after AHPPaternal self-report on baseline questionnaireDiet –Main food groups 1)Whole wheat 2) Unsaturated oils 3) Vegetables 4) Fruits 5) Meat 6) Fish1)Maternal age 2) Maternal smoking 3)Preconception Dietary Risk Score [PDR] of the partner 4) Maternal and Paternal BMIA pregnancy with positive fetal heart action at around 10 weeks after embryo transfer confirmed by ultrasonographyIVF/ICSI induced ongoing pregnancy5 Wesselink et al. 2016 [85]USACohort [Prospective] Preconception pregnancy planner cohort study online (PRESTO) 2013–2017Couples attempting pregnancy (Males $$n = 662$$)Paternal self-report on food frequency questionnaire (FFQ) at baselineDiet – Caffeinated beverages; approximate servings per week 1) Age, 2) race/ethnicity, 3) education, 4) BMI, 5) smoking history, 6) alcohol intake, 7) intercourse frequency, 8) sleep duration, 9) work timeTime to pregnancy (TTP) [(menstrual cycles of attempt time at baseline) + [(LMP date from most recent follow-up questionnaire—date of baseline questionnaire)/cycle length] + 1]Fecundability6 Xia et al. 2016 [86]USACohort [Prospective] Environment and Reproductive Health Study [EARTH] 2007–2014Subfertile couples (Males [$$n = 142$$]) undergoing IVF/ICSI cycles ($$n = 248$$)Paternal self-report on baseline food frequency questionnaire (FFQ)Diet—Dairy intake in the previous 12 months1) Age, 2) BMI, 3) Smoking status, 4) Total exercise time, 5) Dietary patterns, 6) Alcohol, 7) Caffeine, 8) Total energy intake, 9) Female dairy intake, 10) Female age, 11) Prudent dietary pattern, 12) Western dietary pattern1) Fertilization rate, 2) Implantation rate, 3) Clinical pregnancy rate & 4) Live birth rate per initiated cycleIVF/ICSI induced live birth7 Xia et al. 2015 [87]USACohort [Prospective] Environment and Reproductive Health Study [EARTH] 2007–2014Subfertile couples (Males [$$n = 141$$]) undergoing IVF/ICSI cycles ($$n = 246$$)Paternal self-report on baseline food frequency questionnaire (FFQ)Diet—Meat intake in the previous 12 months1) Age, 2) Total energy intake, 3) BMI, 4) Alcohol, 5) Caffeine, 6) Prudent dietary pattern, 7) Western dietary pattern, 8) Infertility diagnoses, 9) Mode of insemination, 10) Female meat intake1) Fertilization rate, 2) Implantation rate, 3) Clinical pregnancy rate & 4) Live birth rate per initiated cycleIVF/ICSI induced live birth7a Studies covered in multiple exposure sectionsbTotal scores from quality assessment using the Newcastle–Ottawa ScaleTable 3Summary table of findings from included studiesFirst Author & YearResults from paternal exposureQuality score ± Body composition Bowatte et al. 2022 [25]Both ever asthma risk in offspring and asthma before age 10 years old were associated with father’s high BMI trajectory (relative risk ratio [RRR] = 1.72 [$95\%$ CI: 1.00, 2.97] and RRR = 1.70 [$95\%$ CI: 0.98, 2.93], respectively). In the sex-stratified analysis, only the high BMI trajectory of fathers was associated with offspring ever allergic asthma (RRR = 2.04 [$95\%$ CI: 1.12, 3.72]; $$P \leq 0.02$$)5 Broadney et al. 2017 [35]Paternal pre-pregnancy body mass index [BMI] categories overweight [25.0—29.9 kg/m2], obese class I [30.0—34.9 kg/m2], and obese class II/III [> 35 kg/m2] are associated with reduced neonatal IgM levels (β = -0.08, [$95\%$ CI: -0.13, -0.03], $$P \leq 0.001$$); (β = -0.07, [$95\%$ CI: -0.13, -0.01], $$P \leq 0.029$$]); (β = -0.11, [$95\%$ CI: -0.19, -0.04], $$P \leq 0.003$$). Paternal overweight or obesity (class I or II/III) is not associated with the neonatal inflammation score (β = 0.003, [$95\%$ CI: -0.10, 0.11]); (β = 0.05, [$95\%$ CI: -0.07, 0.17]); (β = 0.07, [$95\%$ CI: -0.09, 0.23]) or CRP level (β = 0.02, [$95\%$ CI: -0.04, 0.09]); (β = 0.01, [$95\%$ CI: -0.07, 0.09]); (β = 0.004, [$95\%$ CI: -0.10, 0.10])6 Casas et al. 2017 [36]Zero association identified between paternal pre-pregnancy underweight [< 18.5 kg/m2] or obese fathers [≥ 30 kg/m2] and cognitive and psychomotor scores; Global cognitive index (β = 2.78, [$95\%$ CI: -8.40, 13.97]), (β = 0.51, [$95\%$ CI: -1.68, 2.69]); Memory (β = 4.63, [$95\%$ CI: -7.04, 16.31]), (β = 1.67, [$95\%$ CI: -0.62, 3.95]); Motor (β = -5.42, [$95\%$ CI: -17.51, 6.67]), (β = -0.96, [$95\%$ CI: -3.35, 1.42]). There is also no association between behavioural outcomes at pre-school age and underweight or obese fathers; ADHD Inattention (IRR = 3.46, [$95\%$ CI: 0.77, 15.49]), (IRR = 2.12, ($95\%$ CI: 0.73, 6.17); Hyperactivity (IRR = 1.38, [$95\%$ CI: 0.39, 4.76]), (IRR = 1.38, [$95\%$ CI: 0.96, 1.99]); Childhood Asperger Syndrome Test [CAST] (IRR = 0.85, [$95\%$ CI: 0.50, 1.46]), (IRR = 1.01, [$95\%$ CI: 0.91, 1.13])9 Chen et al. 2021 [37]The birth defect rate was significantly higher when paternal prepregnancy BMI ≥ 25 kg/m2 in IVF cycles (aOR 1.82, $95\%$ CI: 1.06,3.10). Couples with paternal prepregnancy BMI ≥ 25 kg/m2 had a four-fold increased risk of congenital malformations of the musculoskeletal system (aOR 4.38, $95\%$ CI: 1.31,14.65) $$P \leq 0.017$$ compared to couples with paternal prepregnancy BMI < 25 kg/m2. This association still remained after adjustment for confounding factors (aOR 4.55, $95\%$ CI 1.32–15.71). No association was seen between paternal prepregnancy BMI and risk of other subcategories of birth defects5 Fang et al. 2020 [38]Pre-pregnancy BMI was roughly associated with TTP among men with BMI ≥ 24 (FOR 0.97 $95\%$CI: 0.95,0.99); however, this association for men disappeared after adjusting for demographic characteristics (aFOR 1.01 $95\%$CI: 0.98,1.02). Following logistic regression, no association was observed between male pre-pregnancy BMI ≥ 24 and subfecundity (aOR 0.97 $95\%$CI: 0.92 – 1.03)5 Fleten et al. 2012 [39]Using absolute BMI values, paternal pre-pregnancy BMI and offspring BMI at age 3 years are associated (β = 0.038, [$95\%$ CI: 0.033, 0.044], $$P \leq 0.018$$). Using BMI as z-score [standard deviation] (β = 0.125, [$95\%$ CI: 0.107, 0.143], $$P \leq 0.805$$), there is no longer an association6 Guo et al. 2022 [40]Following multivariate adjustment, husbands who were underweight had significantly higher risk (OR = 1·17 [$95\%$ CI: (1·15, 1·19)] of SGA compared with the husband with normal BMI. In addition, a significant and increased risk of LGA was observed for overweight and obese men (OR = 1·08 [$95\%$ CI: 1·06,1·09]); (OR = 1·19 ($95\%$ CI: 1·17, 1·20)] respectively. Reduced paternal BMI was associated with an increased risk of SGA when paternal BMI was less than 22·64 (P non-linear < 0·001). Meanwhile, increasing paternal BMI were associated with an increased risk of LGA when paternal BMI was more than 22·92 (P non-linear < 0·001)6 Hoek et al. 2022 [41]Paternal periconceptional BMI is negatively associated with the fertilization rate (β = − 0.01, [SE = 0.004], $$P \leq 0.002$$]); for every increase in paternal BMI point the fertilization rate decreased $1\%$. Paternal BMI is not associated with the TMSC (β = − 2.48, [SE = 1.53], $$P \leq 0.11$$]), the KIDScore (β = − 0.01, [SE = 0.02], $$P \leq 0.62$$]), the embryo usage rate (β = − 0.001, [SE = 0.004], $$P \leq 0.84$$]), a positive pregnancy (β = 0.03, OR = 1.03, $$P \leq 0.49$$), fetal heartbeat (β = 0.03, OR = 1.03, $$P \leq 0.51$$) or live birth (β = 0.01, OR = 1.01, $$P \leq 0.82$$)8 Johannessen et al. 2020 [33]Among offspring with ECRHS/RHINE fathers who had become overweight during puberty, there was an increased risk of adult offspring’s asthma without nasal allergies (RRR = 2.36 [$95\%$ CI: 1.27, 4.38]), compared with fathers who had never been overweight. Offspring’s overweight status at age 8 years was positively associated with adult offspring’s asthma without nasal allergies (RRR = 1.50 [$95\%$ CI: 1.05, 2.16]. The risk of offspring’s overweight status at age 8 years was greater if the father was overweight at the same period [OR = 2.23 [$95\%$ CI: 1.45, 3.42] compared with the offspring having fathers who had never been overweight6 Lonnebotn et al. 2022 [34]Fathers’ overweight before puberty had a negative indirect effect, mediated through sons’ height, on sons’ forced expiratory volume in one second (FEV1) (beta ($95\%$ CI): − 144 (− 272, − 23) mL) and forced vital capacity (FVC) (beta ($95\%$ CI): − 210 (− 380, − 34) mL), and a negative direct effect on sons’ FVC (beta ($95\%$ CI): − 262 (− 501, − 9) mL); statistically significant effects on FEV1/FVC were not observed7 Moss et al. 2015a [42]Paternal preconception overweight [25.0—29.9 kg/m2] and obesity [> 30 kg/m2] is not associated with gestational age (-0.19, [$95\%$ CI: -1.30, 0.91], $$P \leq 0.37$$); (-0.39, [$95\%$ CI: -1.71, 0.94], $$P \leq 0.28$$), or offspring birthweight (35.6, [$95\%$ CI: -1.40, 211.3], $$P \leq 0.34$$); (76.8, [$95\%$ CI: -74.6, 228.1], $$P \leq 0.16$$)7 Mutsaerts et al. 2014a [43]No association identified between paternal pre-pregnancy BMI and spontaneous preterm birth (OR = 0.99, [$95\%$ CI: 0.93, 1.06]) or SGA (0.96, [$95\%$ CI: 0.91, 1.01])3 Noor et al. 2019 [44]Cord blood DNA methylation at 9 CpG sites is associated with paternal BMI independent of maternal BMI (P = < 0.05). Methylation at cg04763273, between TFAP2C and BMP7, decreased by $5\%$ in cord blood with every 1-unit increase in paternal BMI ($$P \leq 3.13$$ × 10 -҆ꝰ), decreases persist at ages 3 ($$P \leq 0.002$$) and 7 ($$P \leq 0.004$$). Paternal BMI is associated with methylation at cg01029450 in the promoter region of the ARFGAP3 gene; methylation at this site is also associated with lower infant birthweight (β = − 0.0003; SD = 0.0001; $$P \leq 0.03$$)7 Pomeroy et al. 2015 [23]Paternal pre-pregnancy BMI is positively associated with neonatal neck-rump length (β = 0.12, $$P \leq 0.008$$) and the distal limb segments [lower arm/lower leg length] (β = 0.09, $$P \leq 0.006$$);(β = 0.09, $$P \leq 0.003$$). Neonatal birthweight (β = 0.08, $$P \leq 0.003$$), proximal limb segments [upper arm/thigh length] (β = 0.10, $$P \leq 0.001$$);(β = 0.08, $$P \leq 0.008$$), relative upper limb length (β = 0.10, $$P \leq 0.002$$) and relative lower limb length (β = 0.09, $$P \leq 0.004$$) are associated with paternal height only. Neonatal head circumference and adiposity are only associated with maternal pre-pregnancy height and BMI6 Retnakaran et al. 2021 [45]Offspring birthweight increases by 10.7 g per unit increase in paternal pregravid BMI ([$95\%$ CI: 0.5, 20.9], $$P \leq 0.04$$), yet paternal pregravid BMI is not an independent predictor for LGA (aOR = 1.15, [$95\%$ CI: 0.92, 1.44]) or SGA (aOR = 0.88, [$95\%$ CI: 0.67, 1.17]). When modelled separately, paternal pregravid weight ($$P \leq 0.04$$), not height ($$P \leq 0.43$$), is associated with offspring birthweight8 Robinson et al. 2020 [46]No association identified between paternal BMI overweight [≥ 25 kg/m2- < 30 kg/m2], obese class I [≥ 30 kg/m2- < 35 kg/m2] and obese class II [≥ 35 kg/m2] and offspring behavioural issues or psychiatric symptoms at 7–8 years; P trend for behavioural outcomes range from 0.13 [Maternal reported ADHD diagnosis] to 0.79 [Prosocial behaviours]7 Sun et al. 2022 [47]Compared with normal weight men, paternal pre-pregnancy overweight was associated with a significantly increased risk of preterm birth (aOR 1.34 $95\%$ CI: 1.25,1.45) and low birth weight (aOR 1.60 $95\%$ CI: 1.46–1.74) in offspring. There was also an increased risk of preterm birth (aOR 1.26 $95\%$ CI: 1.14,1.40) and low birth weight (aOR 1.40 $95\%$ CI: 1.25,1.58) in offspring of paternal pre-pregnancy obesity7 Sundaram et al. 2017 [48]Male BMI [25—< 35 kg m2] and [≥ 35 kg m2] is not associated with TTP, when modelled individually; (aFOR = 0.92, [$95\%$ CI: 0.70, 1.22]), (aFOR = 0.83, [$95\%$ CI: 0.53,1.28]). Obese class II couples (BMI. > 35.0 kg/m2) associate with fecundability (aFOR = 0.41, [$95\%$ CI: 0.17, 0.98]) having a longer TTP in comparison to couples with normal BMI (< 25 kg/m2) (aFOR = 0.91, [$95\%$ CI: 0.25, 3.37])8 Umul et al. 2015 [49]Increasing paternal BMI is inversely associated with sperm concentration ($$P \leq 0.02$$), sperm motility ($$P \leq 0.04$$), the clinical pregnancy rate ($$P \leq 0.04$$), and the live birth rate ($$P \leq 0.03$$). Zero association identified between paternal BMI and the fertilization rate ($$P \leq 0.89$$) or the implantation rate ($$P \leq 0.62$$)2 Wei et al. 2022 [50]Paternal pre-pregnancy overweight and obesity are associated with a higher risk of low birth weight (LBW) (overweight: OR = 1.637, $95\%$ CI: 1.501,1.784); (obesity: OR = 1.454, $95\%$ CI: 1.289, 1.641) and very low birth weight (VLBW) (overweight: OR = 1.310, $95\%$ CI: 1.097,1.564); (obesity: OR = 1.320, $95\%$ CI: 1.037, 1.681). Paternal pre-pregnancy underweight is associated with a lower risk of LBW (OR = 0.660, $95\%$ CI: 0.519, 0.839). Parents who were both excessive-weights in pre-pregnancy BMI, as well as overweight mothers and normal-weight fathers before pre- pregnancy, were more likely to have offspring with LBW, VLBW, and extremely low birth weight (ELBW)6 Wei et al. 2021 [51]Paternal pre-pregnancy BMI overweight (OW) did not present associations with newborn relative telomere length (TL) in cord blood, even following adjustments (percentage change 0.93 ($95\%$ CI: -5.59,8.14));$$P \leq 0.772$$ or stratification by newborn sex (percentage change 2.09 ($95\%$ CI: -7.53,12.72));$$P \leq 0.686.$$ Analysis of the combined effects of parental weight status on newborn TL showed that TL was significantly shortened among newborns whose mothers were overweight and fathers were of healthy weight when compared with those whose mothers and fathers were both of normal weight (percentage change − 8.38 ($95\%$ CI: − 15.47, − 0.92)); $$P \leq 0.0286$$ Xu et al. 2021 [52]Each standard deviation (SD) increment of paternal BMI (approx 3.27 kg/m2) is associated with an additional 29.6 g increase of birth weight ([$95\%$ CI: 5.7, 53.5], $$P \leq 0.02$$). As a continuous variable, one-unit increase in paternal BMI (1.0 kg/m2) is associated with a 9.6 g increase of offspring birth weight ([$95\%$ CI: 2.3, 17.0], $$P \leq 0.01$$). The association between paternal preconception body weight and offspring’s birth weight is pronounced in male neonates and neonates with overweight mothers or mothers with excessive gestational weight gain [GWG] (P = < 0.05)7 Yang et al. 2015 [53]Fathers overweight [BMI 24.0—27.9 kg/m2] or obese [BMI ≥ 28.0 kg/m2] before pregnancy have an elevated risk of giving birth to a macrosomic infant, compared with their normal weight counterparts (aOR = 1.33, [$95\%$ CI: 1.11, 1.59]);(aOR = 1.99 [$95\%$ CI: 1.49,2.65]). Paternal pre-pregnancy weight only [≥ 75.0 kgs], not height, is associated with increased risk of macrosomia (aOR = 1.49, [$95\%$ CI: 1.16, 1.92])6 Zalbahar et al. 2017 [24]Overweight or obese [OW/OB] fathers [> 25 kg/m2] and normal weight mothers [< 25 kg/m2] have an increased risk of offspring OW/OB at both the 5 to 14 year plus the 14 to 21 year follow-up (aOR = 2.34, [$95\%$ CI: 1.50, 3.65]);(aOR = 2.27, [$95\%$ CI: 1.60, 3.24]). This risk increases further when both parents are OW/OB (aOR = 9.95, [$95\%$ CI: 5.60, 17.69]); (aOR = 12.47, [$95\%$ CI: 7.40, 21.03]); for every unit increase in paternal and maternal BMI z-score, offspring BMI z-score increased, on average, by between $0.15\%$ (kg m2) and $0.24\%$ (kg m2) throughout the 5, 14 and 21 year follow-up5 Zhang et al. 2020 [54]Underweight [< 18.5 kg/ m2] male partners prolong a couples' TTP (aFOR = 0.95, [$95\%$ CI: 0.94, 0.96]) compared to male partners with normal BMI [18.5—23.9 kg/m2]. A combination of normal BMI women and overweight men [24.0—28.9 kg/m2] have the greatest opportunity for pregnancy (aFOR = 1.03, [$95\%$ CI: 1.02, 1.03]), a combination of obese women and underweight men have the least opportunity for pregnancy (aFOR = 0.70, [$95\%$ CI: 0.65, 0.76])9Alcohol Luan et al. 2022 [55]The risks of rating scores on anxious/depressed were increased by $33\%$ (RR = 1.33 [$95\%$ CI: 1.09, 1.61]) and $37\%$ (RR = 1.37 [$95\%$ CI: 1.02,1.84]) among girls in the exposed group at ages 4 and 6, respectively. Risks of somatic complaints were increased by $18\%$ (RR = 1.18 [$95\%$ CI: 1.00, 1.40]) and $65\%$ (RR 1.65,[$95\%$ CI: 1.14, 2.38]) among boys in the exposed group at ages 4 and 6. Also, there was the increased risks of sleep problems (RR = 1.25[$95\%$ C:I 1.00,1.55]) in girls at age 4, thought problems (RR = 1.32 [$95\%$ CI: 1.01, 1.73]) in girls at age 6, and rule-breaking behaviours (RR = 1.35 [$95\%$ CI: 1.09, 1.67]) in boys at age 67 Milne et al. 2013 [27]For both ALL and CBT case/control, there was some evidence of a U-shaped relationship between the amount of alcohol fathers consumed in the 12 months before the pregnancy and risk of both cancers. The odds ratios (ORs) fell with increasing consumption, to a minimum at 14–21 standard drinks a week, ALL (OR = 0.51 [$95\%$ CI: 0.32, 0.81]);CBT (OR = 0.58 [$95\%$ CI: 0.35,0.96]), and rose to a maximum at 28 drinks a week; ALL (OR = 1.20 [$95\%$ CI: 0.79,1.83); CBT (OR = 1.53 [$95\%$ CI:0.95, 2.44]). The p values for the quadratic terms in the ALL and CBT models were 0.005 and 0.02, respectively6 Moss et al. 2015a [42]Paternal preconception alcohol intake > once a month is not associated with offspring birthweight (− 85.9, [$95\%$ CI: -336.2, 164.3], $$P \leq 0.50$$) or offspring gestational age (− 0.10, [$95\%$ CI: -0.96, 0.77], $$P \leq 0.83$$)7 Mutsaerts et al. 2014a [43]Paternal preconception alcohol intake > 7 units/week is not associated with spontaneous preterm birth (OR = 1.08, [$95\%$ CI: 0.64, 1.83]) or SGA (OR = 1.07, [$95\%$ CI: 0.73, 1.56])3 Xia et al. 2018 [56]In the paternal alcohol-exposed group [> 81 g/wk], male offspring have shorter mean AGDs; for AGD-AP at birth (β =—1.73, $$P \leq 0.04$$) and 12 months (β = -7.29, $$P \leq 0.05$$), and shorter mean AGD-AS at 6 months (β =—4.91, $$P \leq 0.02$$). Female offspring have shorter mean AGD-AF (β = -0.72, $$P \leq 0.02$$) at birth yet longer mean AGD AC (β = 2.81, $$P \leq 0.04$$) and AGD-AF ($B = 1.91$, $$P \leq 0.04$$) at 12 months8 Zuccolo et al. 2016 [57]Increased odds of microcephaly at birth with alcohol dose per occasion at 5 + units/sitting; [1—2 units] (OR = 1.48, [$95\%$ CI: 0.77, 2.84], $$P \leq 0.238$$), [3–4 units] (OR = 1.64, [$95\%$ CI: 0.85, 3.16], $$P \leq 0.140$$), [5 + units] (OR = 1.93, [$95\%$ CI: 1.01, 3.70], $$P \leq 0.048$$). The average paternal preconception alcohol dose per occasion and general head circumference at birth is not associated [1—2 units] (β = -0.00, [$95\%$ CI: -0.05, 0.04], $$P \leq 0.831$$), [3–4 units] (β = -0.00, [$95\%$ CI: -0.05, 0.04], $$P \leq 0.915$$), [5 + units] (β = -0.02, [$95\%$ CI: -0.07, 0.02], $$P \leq 0.293$$)4Cannabis Har-Gil et al. 2021 [58]Sperm quality is associated with cannabis use (6 [1.4], $$P \leq 0.022$$), compared with non-use (6[2.2], $$P \leq 0.50$$). Sperm volume ($\frac{2.69}{2.5}$ [1.6]), IVF fertilization ($\frac{53}{53}$ [59]), the IR ($$P \leq 0.46$$) and OPR ($$P \leq 0.508$$) are not associated with male cannabis use2 Kasman et al. 2018 [59]Zero association identified between male cannabis use and TTP, regardless of frequency; [< 1/month] (aTR = 0.9, [$95\%$ CI: 0.7, 1.2], $$P \leq 0.43$$), [Monthly] (aTR = 0.9, [$95\%$ CI: 0.5, 1.8], $$P \leq 0.73$$), [Weekly] (aTR = 1.0, [$95\%$ CI: 0.3, 2.9], $$P \leq 1.00$$), [Daily] (aTR = 1.1, [$95\%$ CI: 0.79, 1.5], $$P \leq 0.65$$)6 Moss et al. 2015a [42]Paternal preconception cannabis use is not associated with gestational age (0.41, [$95\%$ CI: -0.43, 1.25], $$P \leq 0.34$$) or offspring birthweight (201.9, [$95\%$ CI: -97.6, 501.3], $$P \leq 0.19$$)7 Nassan et al. 2019 [60]Compared to males who are past or never cannabis users, couples where the male partner is a cannabis user at enrolment ($$n = 23$$) have increased probability of implantation (77.9, [$95\%$ CI: 53.5, 91.5], P = < 0.05) and live birth (47.6, [$95\%$ CI: 32.4, 63.3], P = < 0.05), independent of women's cannabis use. Clinical pregnancy is not associated with male cannabis use; (60.1, [$95\%$ CI: 42.6, 75.4])7 Wise et al. 2018 [61]Male current cannabis users ($$n = 100$$) present no association between cannabis use and fecundability (aFR = 1.01, [$95\%$ CI: 0.81, 1.27]) even following stratification by intercourse frequency (aFR = 1.35, [$95\%$ CI: 0.72, 2.53]) and timing of sexual intercourse (aFR = 1.05, [$95\%$ CI: 0.76, 1.45]). Paternal cannabis use [< 1 time/week] has slightly decreased fecundability (FR = 0.87, [$95\%$ CI: 0.66, 1.15]), compared with non-current users6Physical activity Moss et al. 2015a [42]Zero association identified between paternal preconception bouts of physical activity per week and gestational age (0.02, [$95\%$ CI: -0.04, 0.07], $$P \leq 0.53$$) or offspring birthweight (1.7, [$95\%$ CI: -13.0, 16.4], $$P \leq 0.82$$)7 Mutsaerts et al. 2014a [43]Paternal preconception physical activity of moderate intensity < 1 time/week is not associated with spontaneous preterm birth (OR = 0.76, [$95\%$ CI: 0.45, 1.27]) or SGA (OR = 1.33, [$95\%$ CI: 0.95, 1.87])3Smoking Accordini et al. 2021 [32]Fathers’ smoking initiation in prepuberty (generation G1) had a negative direct effect on their own FEV1/FVC (Δz-score − 0.36, $95\%$ CI: − 0.68, -0.04) compared with fathers’ never smoking. This exposure had a negative direct effect on both offspring’s FEV1 (− 0.36, $95\%$ CI: − 0.63, − 0.10) and FVC(− 0.50, $95\%$ CI: − 0.80, − 0.20) (generation G2). Fathers’ smoking initiation at later ages also had a negative direct effect on their own FEV1 (− 0.27, $95\%$ CI: − 0.51, − 0.02) and FEV1/FVC (− 0.20, $95\%$ CI: − 0.37, − 0.04), but no effect found on offspring’s lung function8 Accordini et al. 2018 [31]Fathers’ smoking before they were 15 years old were associated with asthma without nasal allergies in their offspring [relative risk ratio ((RRR) = 1.43 $95\%$ CI: 1.01, 2.01]. The risk of fathers’ asthma (generation F1) was higher if their parents (generation F0) had ever had asthma (grandmothers’asthma: (OR = 3.08 [$95\%$ CI: 1.96,4.85]); grandfathers’ asthma: (OR = 2.38 [$95\%$ CI: 1.51, 3.75]). The risk of asthma with or without nasal allergies in offspring (generation F2) was higher if the offspring’s father had ever had asthma (RRR = 2.37 and 1.70), respectively7 Carslake et al. 2016 [62]Paternal smoking during pre-adolescence (< age 11) is not reliably or strongly associated with BMI among sons, with an estimated association close to zero (mean difference in kg m-2 ($95\%$ CI) was -0.18 (-1.75, 1.39) for sons aged 12 ± 19 and 0.22 (-0.53, 0.97) for all ages). Among daughters, early-onset paternal smoking was imprecisely associated with an elevated BMI (mean difference was 1.50 (0.00, 3.00) for daughters aged 12 ± 19 and 0.97 (0.06, 1.87) for all ages)6 Deng et al. 2013 [63]During the periconceptional period, light paternal smoking [1–9 cigarettes/day] increases the risk of isolated conotruncal heart defects (aOR = 2.23, [$95\%$ CI: 1.05, 4.73]). Medium paternal smoking [10–19 cigarettes/day] increases the risk of septal defects (aOR = 2.04, [$95\%$ CI: 1.05, 3.98]) and left ventricular outflow tract obstructions (aOR = 2.48, [$95\%$ CI: 1.04, 5.95]). Heavy paternal smoking (≥ 20 cigarettes/day) provides even greater risk of isolated conotruncal heart defects (aOR = 8.16, [$95\%$ CI: 1.13, 58.84]) and left ventricular outflow tract obstructions (aOR = 13.12, [$95\%$ CI: 2.55, 67.39]). No association identified between paternal smoking and right ventricular outflow tract obstructions; light smoking (AOR = 1.84, [$95\%$ CI 0.88, 3.85]); medium smoking (aOR = 2.04, [$95\%$ CI: 0.71, 5.89]); heavy smoking (aOR = 6.02, [$95\%$ CI: 0.98, 36.77])8 Frederiksen et al. 2020 [64]Nil associations identified between paternal smoking before conception and childhood ALL (OR = 1.00, $95\%$ CI: 0.73, 1.38). Paternal smoking before conception was associated with an increased risk of childhood AML in both the crude (OR = 2.55, $95\%$ CI: 1.25, 5.21) and adjusted models (OR = 2.51, $95\%$ CI: 1.21, 5.17)7 Knudsen et al. 2020 [30]In the unadjusted analysis, father’s preconception smoking, both starting before or from age 15 years, was associated with increased offspring BMI. Following adjustments, father’s smoking onset ≥ 15 years was significantly associated with increased BMI in their adult offspring (0.551, [$95\%$ CI: 0.174, 0.929]) $$P \leq 0.004.$$ Father’s preconception smoking onset ≥ 15 years was also associated with increased offspring FMI (2.590 [$95\%$ CI: 0.544, 4.63]) $$P \leq 0.014.$$ Further, sons of fathers’ who started to smoke ≥ 15 years of age (interaction $$p \leq 0.014$$) had significantly higher FMI compared to sons of never smoking fathers5 Milne et al. 2013 [27]Paternal preconception smoking showed no association with childhood brain tumor (CBT) risk (OR = 0.99 ($95\%$ CI: 0.71, 1.38); $$P \leq 0.54.$$ There was also no association evident when paternal smoking was stratified by child’s age5 Ko et al. 2014 [65]Paternal preconception smoking [11–20 cigarettes/day] has a negative effect on overall infant birthweight (β = -19.17 [7.74], $$P \leq 0.013$$) but is not associated with gestational age (β = -0.05 [0.028], $$P \leq 0.108$$). Paternal preconception smoking [> 20 cigarettes/day] is not associated with preterm delivery (1.07, [$95\%$ CI: 0.84, 1.35]), low birth weight (1.14, [$95\%$ CI: 0.87, 1.27]), or small for gestational age [SGA] (1.12, [$95\%$ CI: 0.90, 1.40])5 Moss et al. 2015a [42]Paternal preconception smoking at least one cigarette/day for one month is not associated with gestational age (− 0.31, [$95\%$ CI: − 1.20, 0.59], $$P \leq 0.50$$) or offspring birthweight (− 219.6, [$95\%$ CI: − 537.0, 97.8], $$P \leq 0.18$$)7 Mutsaerts et al. 2014a [43]Paternal smoking [1–10 cigarettes/day] or [< 10 cigarettes/day] 6 months prior to conception, is associated with an increased risk of SGA (OR = 1.69; [$95\%$ CI: 1.10, 2.59]); (OR = 2.25, [$95\%$ CI: 1.51, 3.37]) but not spontaneous preterm birth (OR = 1.34, [$95\%$ CI: 0.74, 2.41]); (OR = 1.13, $95\%$ CI: 0.59, 2.14)3 Northstone et al. 2014 [66]In sons whose fathers started smoking < 11 years, mean differences in BMI, waist circumference, and fat mass all show increases in measures at ages 13, 15 and 17; at 13 years BMI (2.83, [$95\%$ CI: 1.20, 4.25]), waist circumference and fat mass (4.83, [$95\%$ CI: 0.98, 8.68], $$P \leq 0.014$$);(5.79, [$95\%$ CI: 2.67, 8.91] P = < 0.0001), and at 15 years BMI (2.03 [$95\%$ CI: 0.45, 3.6]), waist circumference and fat mass (4.84, [$95\%$ CI: 0.99, 8.66], $$P \leq 0.006$$); (5.50, [$95\%$ CI: 1.88, 9.30], $$P \leq 0.004$$). At 17 years there is an association with BMI (3.25 [$95\%$ CI: 1.15, 5.35]) and fat mass (10.6 [$95\%$ CI: 5.40, 15.9], P = < 0.0001); waist not recorded. Daughters' measurements vary with associations at ages 9 (all measurements), 11 (lean mass $$P \leq 0.023$$), 13 (waist circumference $$P \leq 0.004$$ & lean mass $$P \leq 0.028$$) and 17 (fat mass $$P \leq 0.012$$)7 Orsi et al. 2015 [67]Pre-conception paternal smoking was significantly associated with ALL (OR = 1.2 [$95\%$ CI: 1.1,1.5)] and AML (OR = 1.5 [$95\%$ CI: 1.0–2.3]). For ALL, the ORs were higher for smoking\10 cigarettes daily than for the highest consumption; no significant trend was evidenced. For AML, significant trends were evidenced for both periods (p trend = 0.03 and 0.02, respectively), with ORs of close to 2.0 for smoking more than 15 cigarettes daily. No joint effect of paternal and maternal smoking was detected7 Sapra et al. 2016 [68]Paternal cigarette smoking is associated with a longer TTP compared with never users (aFOR = 0.41, [$95\%$ CI: 0.24, 0.68]); attenuated slightly after adjusting for cadmium (aFOR = 0.44, $95\%$ CI: 0.24, 0.79). When modelling partners together, paternal cigarette smoking remains associated with a longer TTP (aFOR = 0.46, [$95\%$ CI: 0.27, 0.79]), also attenuated after adjustment for cadmium (aFOR = 0.50, [$95\%$ CI 0.27—0.91]). Zero association identified between TTP and exposure to any other tobacco products including cigars (FOR = 0.70, [$95\%$ CI: 0.45, 1.08]) or snuff and chew tobacco (FOR = 1.17, [$95\%$ CI: 0.70, 1.95]7 Svanes et al. 2017 [69]Non-allergic early-onset asthma (asthma without hay fever) was more common in the offspring with fathers who smoked before conception (OR = 1.68 [$95\%$ CI: 1.18,2.41]). The risk was highest if father started smoking before age 15 years (OR = 3.24 [$95\%$ CI: 1.67,6.27]), even if he stopped more than 5 years before conception (OR = 2.68 [$95\%$ CI: 1.17, 6.13]).*Both a* father’s early smoking debut ($$P \leq 0.001$$) and a father’s longer smoking duration ($$P \leq 0.01$$) before conception increased non-allergic early-onset asthma in offspring, even with mutual adjustment and adjusting for number of cigarettes and years since quitting smoking. A father’s smoking debut before age 11 years (102 fathers) showed the greatest increased risk (OR = 3.95, [$95\%$ CI: 1.07,14.60]), followed by smoking debut ages 11–14 (OR = 1.75, [$95\%$ CI: 1.07,1.86]) and smoking debut after age 15 (OR = 1.37, [$95\%$ CI: 1.00,1.86]). Longer duration of smoking was also associated with an increased risk, up to 1.8-fold for those smoking for more than 10 years (OR = 1.76, [$95\%$ CI: 0.96,3.25])6 Wang et al. 2022 [70]Hazard ratio (HR) of preterm birth (PTB) was 1.07 ($95\%$ CI, 1.06–1.09), compared with women without preconception paternal smoking. Compared with participants without preconception paternal smoking, the fully adjusted HRs of PTB were (1.04 [$95\%$ CI: 0.99,1.08]), (1.05 [$95\%$ CI: 1.01, 1.08]), (1.06 [$95\%$ CI: 1.03, 1.09]), (1.14 [$95\%$ CI: 1.07, 1.21]) and (1.15 [$95\%$ CI: 1.11, 1.19]) for participants whose husband smoked 1–4, 5–9, 10–14, 15–19, and ≥ 20 cigarettes/day respectively (P linear < 0.05)5 Wang et al. 2018 [71]Women with exposure to paternal preconception smoking have increased odds of SA (aOR = 1.11, [$95\%$ CI: 1.08, 1.14], P = < 0.01). This association is evident when smoking > 10 cigarettes/day, P = < 0.01; [10–14 cigarettes/day] (aOR = 1.11, [$95\%$ CI: 1.06, 1.16]), [15–19 cigarettes/day] (aOR = 1.21, [$95\%$ CI: 1.09, 1.33]) and ≥ 20 cigarettes/day (aOR = 1.23, [$95\%$ CI: 1.17, 1.30)6 Wesselink et al. 2019 [72]Male current regular smoking, current occasional smoking, and former smoking is not associated with fecundability (FR = 0.96, [$95\%$ CI: 0.70, 1.34]), (FR = 0.83, [$95\%$ CI: 0.61, 1.13]), (FR = 1.14, [$95\%$ CI: 0.97, 1.35])5 You et al. 2022 [73]For those with only preconception exposure, compared with children without paternal smoking, the risk of childhood overweight and obesity was increased (OR = 1.41 [$95\%$ CI: 1.17, 1.85]). Following further adjustments, for lifestyle and dietary factors, this effect remained statistically significant (OR = 1.54 [$95\%$ CI: 1.14, 2.08]). When stratified by sex, the effects of only preconception exposure on childhood overweight and obesity was statistically significant for only boys ($p \leq 0.05$)7 Zhou et al. 2020 [74]*There is* an increased risk of birth defects in the continued-smoking (OR = 1.87, [$95\%$ CI: 1.36, 2.56], $P \leq 0.001$) and decreased-smoking groups (OR = 1.41, [$95\%$ CI: 1.10, 1.82], $$P \leq 0.007$$). Continued paternal smoking is associated with an elevated risk of congenital heart diseases (OR = 2.51, [$95\%$ CI: 1.04, 6.05], $$P \leq 0.040$$), limb abnormalities (OR = 20.64, [$95\%$ CI: 6.26, 68.02], $P \leq 0.001$), digestive tract anomalies (OR = 3.67, [$95\%$ CI: 1.44, 9.37], $$P \leq 0.007$$) and neural tube defects (OR = 4.87, [$95\%$ CI: 1.66, 14.28], $$P \leq 0.004$$). There is no association between continued paternal smoking and clefts (OR 1.44, [$95\%$ CI: 0.34, 5.90], $$P \leq 0.625$$) or gastroschisis (OR = 2.63, [$95\%$ CI: 0.82, 8.40] $$P \leq 0.103$$)7 Zwink et al. 2016 [75]Paternal periconceptional tobacco consumption is lower in the fathers of EA/TEF patients [Any smoking] $$n = 20$$ ($20\%$) $$P \leq 0.003$$, compared with fathers of isolated ARM patients [Any smoking] $$n = 49$$ ($40\%$) $$P \leq 0.0034$$Stress Bae et al. 2017 [76]*There is* a $76\%$ increase in risk of fathering a male infant (RR = 1.76, [$95\%$ CI: 1.17, 2.65]) in men diagnosed with anxiety disorders compared with those not diagnosed. This association is strengthened (RR = 2.03, [$95\%$ CI: 1.46, 2.84]) when modelled jointly for the couple6 Mutsaerts et al. 2014a [43]Paternal paid working hours < 16 h/week is not associated with spontaneous preterm birth (OR = 2.21, [$95\%$ CI: 0.78, 6.26]) or SGA (OR = 0.76, [$95\%$ CI: 0.23, 2.45])3 Wesselink et al. 2018 [77]Men's baseline PSS scores are not associated with fecundability; [PSS score 10—14] (FR = 0.95 [$95\%$ CI: 0.79, 1.15]), [PSS Score 15–19] (FR = 1.07 [$95\%$ CI: 0.86, 1.33]), [PSS Score 20–24] (FR = 1.02 [0.76, 1.36]), [PSS Score ≥ 25] (FR = 1.03 [0.69, 1.54])7Nutrition Bailey et al. 2014 [29]No significant associations identified with paternal dietary intake of folate or vitamin B6 or vitamin B12 and risk of ALL; (OR = 1.37 $95\%$ CI: 0.78, 2.40)5 Greenop et al. 2015 [28]No significant associations identified between risk of childhood brain tumors (CBT) and energy adjusted dietary folate > 509.5 (mcg) (OR = 0.85 $95\%$ CI: 0.56,1.28) or energy adjusted B6 > 1.71 (mg) (OR = 0.98 $95\%$ CI: 0.66,1.47). A high B12 intake (> 5.91(mcg)) was not significantly associated with an increased risk of CBT (OR = 1.74 $95\%$ CI: 1.14, 2.66)5 Hatch et al. 2018 [78]Male intake of sugar-sweetened beverages is associated with reduced fecundability (aFR = 0.78 $95\%$ CI: 0.63, 0.95) for ≥ 7 sugar-sweetened beverages per week compared with none. Fecundability was further reduced among those who drank ≥ 7 servings per week of sugar-sweetened sodas (aFR = 0.67 $95\%$ CI: 0.51, 0.89). The largest reduction in fecundability was seen in men who consumed seven or more energy drinks per week (FR = 0.42; $95\%$ CI: 0.20, 0.90). Diet sodas did not have significant association with fecundability at ≥ 7 servings per week (aFR = 0.93 $95\%$ CI: 0.71, 1.2)6 Hoek et al. 2019 [79]In spontaneously conceived pregnancies, there is a negative association between paternal RBC folate status and CRL trajectories, in Q2 [875–1,018 nmol/L;] (β = -0.14; [$95\%$ CI:—0.28, -0.006], $$P \leq 0.04$$) and Q4 [1,196–4,343 nmol/L] (β =—0.19, [$95\%$ CI:—0.33, -0.04], $$P \leq 0.012$$). A negative association also exists for EV trajectories in Q4 (β =—0.12, [$95\%$ CI: -0.20, -0.05], $$P \leq 0.001$$). No association identified between paternal RBC folate status and CRL or EV trajectories in IVF-ICSI pregnancies [Q4] (β = 0.03, [$95\%$ CI: -0.07, 0.13], $$P \leq 0.55$$), (β = 0.03, [$95\%$ CI: -0.03, 0.08], $$P \leq 0.32$$)7 Lippevelde et al. 2020 [80]In Young-HUNT1, an extra serving of fruit per week in the paternal diet, during adolescence, is associated with a 2.35 g increase in offspring placenta weight [$95\%$ CI: 0.284, 4.42], $$P \leq 0.03.$$ A slightly shorter birth length is associated with increased paternal vegetable intake during adolescence (β = -0.048, [$95\%$ CI: -0.080, -0.016], $$P \leq 0.003$$) and a lower ponderal index is associated with paternal whole grain bread consumption (β = -0.003, [$95\%$ CI: -0.005, -0.001], $$P \leq 0.01$$). Paternal lunching regularly in adolescence is associated with an increase in offspring head circumference (β = 0.160, [$95\%$ CI: 0.001, 0.320], $$P \leq 0.05$$). Birthweight is not associated with any paternal dietary exposures; [Fruit] (β = 5.84 [$95\%$ CI: -0.983, 12.7], $$P \leq 0.1$$). These associations are not observed in Young-HUNT38 Martin-Calvo et al. 2019 [81]A 400 μg/day increase in preconception paternal folate intake is associated with a 2.6-day longer gestation [$95\%$ CI: 0.8, 4.3], $$P \leq 0.004.$$ This association is strongest in multifetal pregnancies (β = 10.7, [$95\%$ CI: 4.6, 16.8]). Zero association identified between paternal folate intake and gestational age-specific birthweight (β = -11.4, [$95\%$ CI: -28.2, 5.4])7 Mitsunami et al. 2021 [82]Paternal adherence to either dietary patterns 1 or 2 is not associated with the fertilization rate during IVF or ICSI ([Pattern 1] $$P \leq 0.59$$, [Pattern 2] $$P \leq 0.06$$), ([Pattern 1] $$P \leq 0.72$$, [Pattern 2] $$P \leq 0.94$$). Zero association identified between male dietary patterns and probabilities of implantation, clinical pregnancy, or live birth; ([Pattern 1] $$P \leq 0.68$$, [Pattern 2] $$P \leq 0.43$$), ([Pattern 1] $$P \leq 0.35$$, [Pattern 2] $$P \leq 0.68$$), ([Pattern 1] $$P \leq 0.53$$, [Pattern 2] $$P \leq 0.10$$)7 Moss et al. 2015a [42]Males eating fast food more frequently have infants born earlier than men who eat fast-food less frequently (-0.16, [$95\%$ CI: -0.32, 0.00], $$P \leq 0.04$$). There is no association between paternal fast-food consumption and birthweight (-36.0, [$95\%$ CI: -89.8, 17.8], $$P \leq 0.19$$)7 Oostingh et al. 2019 [83]Zero association identified between paternal dietary patterns and CRL or EV in spontaneous pregnancies; [Whole wheat grains and vegetables] (β = -0.006 [$95\%$ CI: -0.069, 0.058]), (β = 0.001 [$95\%$ CI: -0.022, 0.021]), and in IVF/ICSI pregnancies, (β = -0.015 [$95\%$ CI: -0.061, 0.031]), (β = -0.006 [$95\%$ CI: -0.025, 0.013]), independent of maternal dietary patterns6 Twigt et al. 2012 [84]Paternal Preconception Dietary Risk Score [PDR] did not affect the chance of pregnancy after IVF/ICSI treatment (OR = 0.95 [$95\%$ CI: 0.48,1.86]) $$P \leq 0.885$$ Wesselink et al. 2016 [85]Total caffeine intake among males was associated with fecundability for ≥ 300 mg vs. < 100 mg/day (OR = 0.72, $95\%$ CI: 0.54, 0.96)6 Xia et al. 2016 [86]Men's total dairy intake is not associated with the fertilization rate [Conventional IVF] (0.75, [$95\%$ CI: 0.60, 0.86], $$P \leq 0.29$$), [ICSI] (0.72, [$95\%$ CI: 0.58, 0.82], $$P \leq 0.18$$], the implantation rate (0.58, [$95\%$ CI: 0.40, 0.74], $$P \leq 0.87$$), the clinical pregnancy rate (0.51, [$95\%$ CI: 0.34, 0.68], $$P \leq 0.54$$), or the live birth rate (0.46, [$95\%$ CI: 0.28, 0.65], $$P \leq 0.65$$)7 Xia et al. 2015 [87]A positive association identified between paternal poultry intake and the fertilization rate, [Model 1] $$P \leq 0.05$$, [Model 2] $$P \leq 0.03$$, [Model 3] $$P \leq 0.03$$, [Model 4] $$P \leq 0.04$$, with a $13\%$ higher fertilization rate among men in the highest quartile of poultry intake compared with those in the lowest quartile ($78\%$ vs. $65\%$) [Model 4]. Men's total meat intake is not associated with the implantation rate (0.52, [$95\%$ CI: 0.37, 0.67], $$P \leq 0.67$$), clinical pregnancy rate (0.45, [$95\%$ CI: 0.32, 0.59], $$P \leq 0.56$$), or live-birth rate (0.35, [$95\%$ CI: 0.22, 0.50], $$P \leq 0.82$$)7a Studies covered in multiple exposure sectionsbQuality score based on assessment using Newcastle–Ottawa Scale Study participants were diverse consisting of couples either intending pregnancy or pregnant ($$n = 25$$), sub-fertile and seeking fertility treatment undergoing IVF/ICSI ($$n = 11$$), or mothers and fathers of infants ($$n = 26$$). Two studies included adolescents followed into parenthood as adults [42, 80], and one study included individual respondents of a national family growth survey, actively attempting pregnancy [59].
Modifiable preconception risk factors and/or health behaviour exposures examined include paternal body composition ($$n = 25$$), alcohol intake ($$n = 6$$), cannabis use ($$n = 5$$), physical activity ($$n = 2$$), smoking ($$n = 20$$), stress ($$n = 3$$), and nutrition ($$n = 13$$) (including dietary folate intake and consumption of foods and dietary patterns). Two papers investigated multiple exposures [42, 43].
Outcomes examined include fecundability ($$n = 6$$) [38, 61, 72, 77, 78, 85], (time to) pregnancy ($$n = 4$$) [48, 54, 59, 68], IVF/ICSI ongoing pregnancy ($$n = 1$$) [84] or live birth ($$n = 7$$) [41, 49, 58, 60, 81, 82, 86, 87], offspring birthweight or adiposity ($$n = 10$$) [40, 42, 45, 47, 50, 52, 53, 62, 65, 66], including small for gestational age [SGA] [43], neonatal ($$n = 1$$) [23] and offspring body composition ($$n = 4$$) [24, 30, 39, 73]. Other outcomes examined include offspring asthma ($$n = 4$$) [25, 31, 33, 69] and lung function ($$n = 2$$) [32, 34], childhood leukemia ($$n = 4$$) [26, 29, 64, 67], childhood brain tumours ($$n = 2$$) [27, 28], and offspring behavioural issues ($$n = 2$$) [46, 55].
There was an increasing number of papers identified for inclusion in this review with the least number of papers published in 2012 and the greatest number of papers published in 2022 (see Fig. 2).Fig. 2Papers included in this review Results below are described for papers assessed as good quality with approximately half ($$n = 30$$) rated good quality and two receiving a maximum nine-star rating [36, 54] (Table 4—Newcastle Ottawa Scale [NOS] quality assessment (Cohorts)) & (Table 5—Newcastle Ottawa Scale [NOS] quality assessment (Case controls)). Results for the fair and poor-quality papers are not further described. Table 4Newcastle Ottawa Scale [NOS] quality assessment (Cohorts)First author & YearNewcastle Ottawa Scale—CriteriaSELECTIONCOMPARABILITYOUTCOMETOTALExposed cohort (representativeness)Non-exposed cohortAscertainment of exposureOutcome of interest not present at start of studyBased on design or analysis (AGE)Based on design or analysis (OTHER FACTORS)Assessment of outcomeAppropriate length of follow-upAdequacy of follow-up of cohorts Accordini et al. 2021 [32]✶✶✶✶✶✶✶✶8 Accordini et al. 2018 [31]✶✶✶✶✶✶✶7 Bae et al. 2017 [80]✶✶✶✶✶✶6 Bowatte et al. 2022 [25]✶✶✶✶✶5 Broadney et al. 2017 [76]✶✶✶✶✶✶6 Carslake et al. 2016 [62]✶✶✶✶✶✶6 Casas et al. 2017 [74]✶✶✶✶✶✶✶✶✶9 Chen et al. 2021 [75]✶✶✶✶✶5 Fang et al. 2020 [42]✶✶✶✶✶5 Fleten et al. 2012 [67]✶✶✶✶✶✶6 Guo et al. 2022 [61]✶✶✶✶✶✶6 Har-Gil 2021 [50]✶✶2 Hatch et al. 2018 [44]✶✶✶✶✶✶6 Hoek et al. 2022 [49]✶✶✶✶✶✶✶✶8 Hoek et al. 2019 [88]✶✶✶✶✶✶✶7 Johannessen et al. 2020 [33]✶✶✶✶✶✶6 Kasman et al. 2018 [37]✶✶✶✶✶✶6 Knudsen et al. 2020 [30]✶✶✶✶✶5 Ko et al. 2014 [60]✶✶✶✶✶5 Lippevelde et al. 2020 [80]✶✶✶✶✶✶✶✶8 Lonnebotn et al. 2022 [34]✶✶✶✶✶✶✶7 Luan et al. 2022 [73]✶✶✶✶✶✶✶7 Martin-Calvo et al. 2019✶✶✶✶✶✶✶7 Mitsunami et al. 2021 [82]✶✶✶✶✶✶✶7 Moss et al. 2015 [35]✶✶✶✶✶✶✶7 Mutsaerts et al. 2014 [38]✶✶✶3 Nassan et al. 2019 [52]✶✶✶✶✶✶✶7 Noor et al. 2019 [77]✶✶✶✶✶✶✶7 Northstone et al. 2014 [63]✶✶✶✶✶✶✶7 Oostingh et al. 2019 [81]✶✶✶✶✶✶6 Pomeroy et al. 2015 [23]✶✶✶✶✶✶6 Retnakaran et al. 2021 [58]✶✶✶✶✶✶✶✶8 Robinson et al. 2020 [72]✶✶✶✶✶✶✶7 Sapra et al. 2016 [47]✶✶✶✶✶✶✶7 Sun et al. 2022 [66]✶✶✶✶✶✶✶7 Sundaram et al. 2017 [45]✶✶✶✶✶✶✶✶8 Svanes et al. 2017 [69]✶✶✶✶✶6 Twigt et al. 2012 [48]✶✶✶✶✶5 Umul et al. 2015 [57]✶✶2 Wang et al. 2022 [65]✶✶✶✶✶5 Wang et al. 2018 [85]✶✶✶✶✶✶6 Wei et al. 2022 [50]✶✶✶✶✶✶6 Wei et al. 2021 [51]✶✶✶✶✶✶6 Wesselink et al. 2019 [40]✶✶✶✶✶5 Wesselink et al. 2018 [41]✶✶✶✶✶✶✶7 Wesselink et al. 2016 [43]✶✶✶✶✶✶6 Wise et al. 2018 [39]✶✶✶✶✶✶6 Xia et al. 2018 [78]✶✶✶✶✶✶✶✶8 Xia et al. 2016 [55]✶✶✶✶✶✶✶7 Xia et al. 2015 [56]✶✶✶✶✶✶✶7 Xu et al. 2021 [59]✶✶✶✶✶✶✶7 You et al. 2022 [68]✶✶✶✶✶✶✶7 Zalbahar et al. 2017 [24]✶✶✶✶✶5 Zhang et al. 2020 [46]✶✶✶✶✶✶✶✶✶9 Zhou et al. 2020 [86]✶✶✶✶✶✶✶7 Zuccolo et al. 2016 [79]✶✶✶✶4Table 5Newcastle Ottawa Scale [NOS] quality assessment (Case controls)Newcastle–Ottawa Critical Analysis (Case controls)—CriteriaFirst author & YearSELECTIONCOMPARABILITYEXPOSURETOTALAdequacy of case definitionRepresentativeness of the casesSelection of controlsDefinition of controlsComparability of cases and controls (AGE)Comparability of cases and controls (OTHER FACTORS)Ascertainment of exposureSame method of ascertainment for cases and controlsNon-Response rate Bailey et al. 2014 [29]✶✶✶✶✶5 Deng et al. 2013 [83]✶✶✶✶✶✶✶✶8 Frederiksen et al. 2020 [70]✶✶✶✶✶✶✶7 Greenop et al. 2015 [28]✶✶✶✶✶5 Milne et al. 2013 [27]✶✶✶✶✶✶6 Milne et al. 2013 [27]✶✶✶✶✶5 Orsi et al. 2015 [71]✶✶✶✶✶✶✶7 Yang et al. 2015 [64]✶✶✶✶✶✶6 Zwink et al. 2016 [87]✶✶✶✶4
## Body Composition
Twenty-five papers investigated associations between paternal BMI and various offspring outcomes. These papers were derived from studies ($$n = 21$$) conducted in the US, Europe, China, Australia, and Turkey and less than half ($$n = 11$$) rated as good quality.
Less than half of the papers ($$n = 10$$) used anthropometric assessment by the research team to determine BMI [37, 38, 40–42, 45, 47–49, 54]; heights and weights utilized to formulate BMI were determined in the preconception period, generally from males in couples undergoing IVF/ICSI [41, 49] or males in couples currently attempting pregnancy/pregnant [40, 45, 48, 54]. Of the papers validating the BMI utilizing anthropometric assessments, most were good quality and generally affirmed significant results. The remaining papers utilize retrospective reports of preconception paternal weight and height or collect paternal height and weight from medical records. Maternal reporting ($$n = 7$$) occurred at approximately 10 to 18 weeks gestation; or up to four months postpartum [35, 46]. Paternal self-reporting of their own weight and height ($$n = 8$$) occurred at approximately week 17 of gestation [39, 43, 50, 52, 53] or up to 6 months postpartum [43]. In two papers, overweight paternal status, when a child of 8 years, was reported years later through a validated drawing of silhouettes [33, 34].
The outcomes and outcome measures varied, however, ten studies assessed the association of paternal BMI with offspring BMI/bodyweight [23, 39, 40, 42, 45, 47, 50, 52, 53], and one paper assesses offspring weight and BMI changes from childhood (5 years) into adulthood (21 years) [24].
Results of associations between body composition and offspring outcomes were inconsistent. In some studies paternal preconception overweight (25.0—29.9 kg/m2) and obesity (> 30 kg/m2) were not associated with offspring birthweight [42] and paternal pregravid BMI was not an independent predictor for large for gestational age (LGA) or small for gestational age (SGA) [45]. However, other studies found that offspring birthweight increased by 10.7 g per unit increase in paternal pregravid BMI ($95\%$ CI: 0.5, 20.9, $$P \leq 0.04$$) [45], and each standard deviation (SD) increment of paternal BMI (approximately 3.27 kg/m2) was associated with an additional 29.6 g increase of birth weight ($95\%$ CI: 5.7, 53.5, $$P \leq 0.02$$) [52]. Further, compared with normal weight men, paternal pre-pregnancy overweight was associated with a significantly increased risk of preterm birth (aOR 1.34 $95\%$ CI: 1.25,1.45) and low birth weight (aOR 1.60 $95\%$ CI: 1.46–1.74) in offspring [47].
Paternal pregravid weight ($$P \leq 0.04$$), not height ($$P \leq 0.43$$), was associated with infant birth weight [45] and with increased risk of macrosomia (aOR = 1.49, [$95\%$ CI: 1.16, 1.92]) [53], while neonatal birth weight was associated with paternal height only (β = 0.08, $$P \leq 0.003$$) [23]. In another study, paternal pre-pregnancy BMI was only associated with offspring BMI when using absolute BMI values not BMI as a z-score [39].
Fathers’ overweight before puberty had a negative indirect effect, mediated through sons’ height, on sons’ forced expiratory volume in one second (FEV1) (beta ($95\%$ CI): − 144 (− 272, − 23) mL) and forced vital capacity (FVC) (beta ($95\%$ CI): − 210 (− 380, − 34) mL), and a negative direct effect on sons’ FVC (beta ($95\%$ CI): − 262 (− 501, − 9) mL) [34].
Male BMI ≥ 25 kg m2 was not associated with time to pregnancy (TTP) [48], yet underweight (< 18.5 kg/ m2) was associated with a longer TTP (adjusted fecundability odds ratio [aFOR] = 0.95, [$95\%$ CI: 0.94, 0.96]) compared to normal BMI (18.5—23.9 kg/m2) [54].
In couples undergoing IVF/ICSI, paternal periconceptional BMI was negatively associated with fertilization rate (β = − 0.01 [SE = 0.004], $$P \leq 0.002$$]), while paternal BMI was not associated with the total motile sperm count (TMSC), the KIDScore, the embryo usage rate, a positive pregnancy, fetal heartbeat, or live birth [41].
Offspring methylation was associated with paternal BMI independent of maternal BMI (P = < 0.05) [44]. Methylation decreased by $5\%$ in cord blood with every 1-unit increase in paternal BMI ($$P \leq 3.13$$ × 10 -҆ꝰ), decreases persist at 3 years old ($$P \leq 0.002$$) and 7 years old ($$P \leq 0.004$$) [44]. Paternal BMI was associated with methylation at cg01029450 in the promoter region of the ARFGAP3 gene; methylation at this site was also associated with lower infant birthweight (β = − 0.0003; SD = 0.0001; $$P \leq 0.03$$) [44].
No association was found between behavioural outcomes at pre-school age and underweight (< 18.5 kg/m2) or obesity (≥ 30 kg/m2) in fathers [36]. Equally, no associations were found between paternal BMI overweight (≥ 25 kg/m2- < 30 kg/m2), obese class I (≥ 30 kg/m2- < 35 kg/m2) and obese class II (≥ 35 kg/m2) and offspring behavioural issues or psychiatric symptoms at 7–8 years [46].
## Alcohol
Six papers examined alcohol as an exposure [26, 42, 43, 55–57]; three rated as good quality [42, 55, 56]. Excluding one, each paper used paternal self-reports of alcohol consumption with varying definitions; one article specified units/per week [43], the others assessed consumption more broadly either as intake ≥ 1/week [56], ≥ 1/month [42] or general intake [57]. A single study presented a maternal report of paternal preconception alcohol consumption, 3 months before conception, at 12–16 weeks gestation [55].
When examining an outcome of offspring anogenital distance (AGD), in the paternal alcohol-exposed group (> 81 g/wk), male offspring had shorter mean AGDs [56]; for AGD-AP [the centre of the anus to the cephalad insertion of the penis] at birth (β =—1.73, $$P \leq 0.04$$) and 12 months (β = -7.29, $$P \leq 0.05$$), and shorter mean AGD-AS [the centre of the anus to the posterior base of the scrotum] at 6 months (β =—4.91, $$P \leq 0.02$$) [56]. Female offspring had shorter mean AGD-AF [the centre of the anus to the posterior convergence of the fourchette) (β = -0.72, $$P \leq 0.02$$) at birth yet longer mean AGD AC [the centre of the anus to the clitoris] (β = 2.81, $$P \leq 0.04$$) and AGD-AF ($B = 1.91$, $$P \leq 0.04$$) at 12 months [56]. Further, the relative risks of anxiety or depression were increased by $33\%$ (RR = 1.33 [$95\%$ CI: 1.09, 1.61]) and $37\%$ (RR = 1.37 [$95\%$ CI: 1.02,1.84]) among girls in the exposed group at ages 4 and 6, respectively [55]. Paternal alcohol consumption greater than once per month was not associated with offspring birthweight or gestational age [42].
## Cannabis
Paternal cannabis exposure was assessed in five papers [42, 58–61], two rate as good quality [42, 60]. Each paper has a sample size < 1,200 and each utilized paternal self-reporting of cannabis use broadly assessing general use, rather than specific amounts, over a pre-determined period (i.e., last 2 months or 12 months).
In sub-fertile couples undergoing IVF/ICSI, compared to males who were past or never cannabis users, couples where the male partner used cannabis at enrolment had increased probability of implantation (77.9, [$95\%$ CI: 53.5, 91.5], P = < 0.05) and live birth (47.6, [$95\%$ CI: 32.4, 63.3], P = < 0.05), independent of women's cannabis use [60]. Clinical pregnancy was not associated with male cannabis use [60], nor was gestational age or offspring birthweight [42].
## Physical activity
The associations of paternal physical activity with offspring outcomes were assessed in two papers [42, 43], one rated as good quality [42]. This study found no association between paternal preconception bouts of physical activity per week and gestational age or offspring birthweight [42].
## Smoking
The association of tobacco smoking with offspring outcomes was examined in 20 papers [27, 30–32, 42, 62–75, 88]; half ($$n = 10$$) were rated as good quality [31, 32, 42, 63, 64, 66–68, 73, 74] and nine papers adjusted for maternal smoking and/or paternal passive smoking in their analysis [32, 62, 64–66, 71, 73, 74].
Paternal cigarette smoking was associated with a longer TTP compared with never users (aFOR = 0.41, [$95\%$ CI: 0.24, 0.68]), while no associations were found for other tobacco products including cigars or snuff and chew tobacco [68].
Outcomes involving smoking and birth defects report that during the periconceptional period, light paternal smoking [1–9 cigarettes/day] increased the risk of isolated conotruncal heart defects (aOR = 2.23, [$95\%$ CI: 1.05, 4.73]) [63]. Medium paternal smoking [10–19 cigarettes/day] increased the risk of septal defects (aOR = 2.04, [$95\%$ CI: 1.05, 3.98]) and left ventricular outflow tract obstructions (aOR = 2.48, [$95\%$ CI: 1.04, 5.95]) [63]. Heavy paternal smoking (≥ 20 cigarettes/day) increased the risk of isolated conotruncal heart defects (aOR = 8.16, [$95\%$ CI: 1.13, 58.84]) and left ventricular outflow tract obstructions (aOR = 13.12, [$95\%$ CI: 2.55, 67.39]) [63]. Likewise, an increased risk of birth defects was found for continued-smoking (OR = 1.87, [$95\%$ CI: 1.36, 2.56], $P \leq 0.001$) and decreased-smoking groups (OR = 1.41, [$95\%$ CI: 1.10, 1.82], $$P \leq 0.007$$) compared with those fathers that quit smoking during early pregnancy and those who did not smoke at all during preconception [74].
Paternal preconception smoking at least one cigarette/day for one month was not associated with gestational age or offspring birthweight [42]. In contrast, a second study found sons whose fathers started smoking < 11 years, the adjusted mean differences in BMI, waist circumference, and fat mass all showed higher values at ages 13, 15, and 17 [66]. Further, the risk of childhood overweight and obesity was increased among children exposed to paternal preconception smoking compared to children without paternal smoking exposure (OR = 1.41 [$95\%$ CI: 1.17, 1.85]) [73].
Paternal preconception smoking 12 months prior to conception was associated with an increased risk of childhood acute myeloid leukemia (AML) (OR = 2.51, $95\%$ CI: 1.21, 5.17) [64] and paternal smoking just 3 months prior to conception provided significant associations with acute lymphoblastic leukemia (ALL) (OR = 1.2 [$95\%$ CI: 1.1,1.5)] and acute myeloblastic leukemia (AML) (OR = 1.5 [$95\%$ CI: 1.0–2.3]) [67].
Paternal preconception smoking also provided significant associations with offspring lung function and asthma; fathers’ smoking initiation in prepuberty (generation G1) had a negative direct effect on their own FEV1/FVC (difference in offspring’s expected score − 0.36, $95\%$ CI: − 0.68, -0.04) compared with fathers’ never smoking. This exposure had a negative direct effect on both offspring’s FEV1 (− 0.36, $95\%$ CI: − 0.63, − 0.10) and FVC (− 0.50, $95\%$ CI: − 0.80, − 0.20) (generation G2) [32]. Fathers’ smoking before age 15 years was associated with higher risk of asthma without nasal allergies in their offspring [relative risk ratio ((RRR) = 1.43 $95\%$ CI: 1.01, 2.01] [31].
## Stress
Paternal stress exposure was examined in three papers [43, 76, 77]; including one rated as good quality [77]. This study found men's baseline perceived stress scale [PSS] scores were not associated with fecundability [77].
## Nutrition
Papers examining paternal nutrition ($$n = 13$$) evaluated the associations of a range of nutritional exposures including paternal preconception folate, vitamins B6 and B12, and general dietary patterns with numerous offspring outcomes. These papers utilized data from several studies ($$n = 8$$) originating in the US, Norway, The Netherlands, and Australia. Approximately half of these papers ($$n = 7$$) rated as good quality.
Paternal nutritional factors explored included dietary patterns [82, 83] or specific foods groups including dairy [86], and meat [87]. IVF/ICSI-induced live birth was an outcome examined in three papers [82, 86, 87]. A positive association was found between paternal poultry intake and fertilization rate, with a higher fertilization rate among men in the highest quartile of poultry intake [$78\%$] compared with those in the lowest quartile [$65\%$] [87]. Men's total dairy intake was not associated with fertilization rate, implantation rate, clinical pregnancy rate, or live birth rate [86]. Also, paternal adherence to specific dietary patterns [pattern 1 = greater intake of processed foods/meats/high fat/dairy/sugar; pattern 2 = greater intake of fruit/vegetables/legumes/whole grains/nuts/fish] was not associated with fertilization rate [82] when undergoing IVF cycles.
One paper investigated dietary exposures during adolescence and subsequent neonatal health [80]. In a sample of adolescents followed into adulthood becoming fathers ($$n = 2$$,140), an extra serving of fruit per week was associated with a 2.35 g increase in offspring placenta weight [$95\%$ CI: 0.284, 4.42], $$P \leq 0.03$$ [80]. Further, paternal lunching regularly in adolescence was associated with an increase in offspring head circumference (β = 0.160, [$95\%$ CI: 0.001, 0.320], $$P \leq 0.05$$) and whole grain bread consumption was associated with a lower ponderal index (β = -0.003, [$95\%$ CI: -0.005, -0.001], $$P \leq 0.01$$) [80]. Birthweight was not associated with any paternal dietary exposures [80].
Generally, paternal preconception dietary patterns were collected through paternal self-reports on standardised food frequency questionnaires (FFQ) at baseline and include fast foods [42]; males eating fast food more frequently had infants born earlier than men who eat fast food less frequently (-0.16, [$95\%$ CI: − 0.32, 0.00], $$P \leq 0.04$$) [42].
Two papers specifically investigated paternal folate [79, 81]. In males undergoing fertility treatment, a 400 μg/day higher preconception folate intake was associated with a 2.6-day longer gestation [$95\%$ CI: 0.8, 4.3], $$P \leq 0.004$$ [81]. In spontaneously conceived pregnancies, a significant negative association was found between paternal red blood cell [RBC] folate status and crown-rump length (CRL) trajectories, in Quartile 2 [875–1,018 nmol/L;] (β = -0.14; [$95\%$ CI:—0.28, -0.006], $$P \leq 0.04$$) and Quartile 4 [1,196–4,343 nmol/L] (β =—0.19, [$95\%$ CI:—0.33, -0.04], $$P \leq 0.012$$) compared with the reference values in Quartile 3 [79]. A negative association was also found for embryonic volume (EV) trajectories in Quartile 4 (β =—0.12, [$95\%$ CI: -0.20, -0.05], $$P \leq 0.001$$) [79].
## Discussion
This paper reports the first review collating literature assessing modifiable paternal health behaviours and risk factors in the preconception period and highlights clear disparity between the preconception research for women as compared to that for men. While single papers identified in our review do demonstrate adverse pregnancy and offspring outcomes associated with paternal risk factors in the preconception period, current research of paternal health behaviours and risk factors provides an emerging rather than mature evidence-base. Nevertheless, our review did identify a number of important findings.
One consistent finding of this review was the association between paternal preconception smoking and increased risk of adverse infant outcomes, including birth defects and childhood leukemia especially acute myeloid leukemia/acute myeloblastic leukemia (AML). Adverse outcomes such as birth defects are mirrored in maternal preconception smoking literature [89–91], yet the impact of maternal smoking on the risk of AML remains contentious [92, 93]. Smoking in the preconception period may be as perilous for males as for females, as smoking can potentially affect semen quality [94]. Many male smokers (and even more so in smoking couples) consider smoking an indispensable characteristic of their domestic, social and working lives [95] and many report a lack of motivation, willpower, and/or strength to successfully quit [96], in turn influencing female smoking patterns and family environments [95]. Paternal preconception smoking may well be contributing to the estimated 240,000 newborns dying worldwide annually due to birth defects [97]. The finding of paternal preconception smoking and the increased risk of adverse infant outcomes is altogether disconcerting considering the widespread use of tobacco, and that males are more likely than females to engage in risk-taking behaviours, including smoking [98]; the estimated global prevalence of male adolescent smokers in 133 countries is $23.29\%$.
The papers in this review which focus upon body composition with birthweight outcomes generally affirm positive associations between increasing paternal BMI and offspring birthweight. Indeed, this finding aligns with the literature outside this review which acknowledges that mothers and fathers with overweight or obesity are more likely to have children with overweight or obesity [99–102], compared with those with a normal weight. The positive associations between increasing paternal BMI and offspring bodyweight may, in part, be due to paternal contributions of sperm quality and potential changes to the epigenetic profiles of spermatoza [10, 103] resulting from unhealthy preconception environments and relationships with food. Food-based parenting strategies [100] and spending too much time sedentary [104] may also contribute to influencing offspring weight status. One paper in this review did chart offspring weight and BMI changes from childhood into adulthood [24], however, this reported research did not control for the offspring’s diet and physical exercise.
Nonetheless, an individual’s birthweight can influence both their body weight in childhood [105] and their body weight as they transition into adulthood [106]; external literature positively associates both a higher birthweight and childhood obesity with overweight/obesity at 15–20 years of age [107]. Frameworks to maintain healthy bodyweight, in turn promoting healthy birthweights, endure in the Global action plan on physical activity 2018–2030 [108] and in national overweight/obesity guidelines in countries such as Australia [109] and the Unites States [110].
It is important to note that most papers included in this review utilize retrospective reports (paternal self-reports or maternal reports) of anthropometric data collected at baseline. Such retrospective self-reporting is also evident in the maternal preconception literature [111, 112] and is often considered unreliable and subject to inaccuracies due to self-reporting bias or recall bias [113]. Inaccuracies and reporting bias may be present in particular in papers that utilize maternal reports of paternal preconception height and body weight at minimum 10 weeks of gestation in some papers up to 4 months postpartum. Consequently, retrospective reports of data at baseline may undermine the validity, accuracy, and therefore the reliability of BMI data used in these papers.
The majority of papers in our review report research undertaken in distinct geographical regions with the USA, Europe and the UK, and China heavily represented. As such, the implications for reduced geographical spread of the available research examining paternal preconception health exposures and outcomes must also be considered. It may be that existing region-specific idiosyncrasies of paternal health behaviours, and associated adverse health outcomes for their children, are yet to be described due to the absence of research conducted in other countries and cultures. These gaps limit the opportunities for tailored preconception care policies and interventions and constrain the broader understanding of the potential importance of paternal preconception care. Notwithstanding, such issues foster opportunities for other countries and cultures to identify, learn from and support paternal health.
While almost all papers in this review adjust for some confounders, less than half ($$n = 23$$) adjusted for the same maternal exposure (i.e., paternal BMI studies adjusting for maternal BMI). Many papers in this review did not adjust for maternal exposures and thus may present biased results and conclusions. Further, many maternal studies do not control for paternal exposures which is a limitation in the field that requires urgent research attention and refocus.
The date parameters set during the search may also represent a limitation as it may have resulted in manuscripts published before the 2012 being overlooked. However, up until recently the preconception research field has primarily focused on the effects of maternal exposures and as such it is unlikely that significant research was overlooked by this date restriction.
Further limitations of the review include the potential for missed citations due to issues with article indexing. Our search protocol did not employ search term truncations or singular synonyms in the final search string which may have resulted in some citations being missed. However, the search protocol was informed by an experienced health librarian, and additional methods – such as reference list and citation checking—were used to identify relevant manuscripts not identified through the primary search. Furthermore, previous search strings trialed for this review that used different synonyms, truncations and search term categories did not result in any additional relevant manuscripts being identified beyond those included in the final search. As such, the literature review is the most comprehensive review of the topic conducted to date.
This review is innovative in that it provides the first examination of paternal preconception risk factors and their association with adverse pregnancy and offspring outcomes. The rigour of the review is also bolstered through adhering to established systematic review reporting guidelines (PRISMA and AMSTAR).
## Conclusion
Overall, this review shows that paternal preconception modifiable risk factors are largely underexplored; smoking and body composition appear to be important areas for consideration in paternal preconception care. While the current literature identifies an emerging evidence-base around paternal preconception modifiable risk factors, there is a need for further investigation to help better inform paternal preconception care and national and international preconception care guidelines. In particular, further research is necessary to identify and better understand the modifiable risk factors affecting males in the preconception period, and how these risk factors influence offspring outcomes, to inform clinical recommendations and health decisions. The future of paternal preconception care and the integration of such care into frontline health practice and policy rests with informed collaboration between clinicians, researchers and policymakers [8].
## Supplementary Information
Additional file 1.
## References
1. 1.World Health Organization [WHO]. Preconception care: maximising the gains for maternal and child health - Policy brief World Health Organization - All rights reserved; 2013 15 February 2013. Contract No.: WHO REFERENCE NUMBER: WHO-FWC-MCA-13.02.
2. 2.World Health Organization [WHO], editor Meeting to develop a global consensus on preconception care to reduce maternal and childhood mortality and morbidity Meeting to develop a global consensus on preconception care to reduce maternal and childhood mortality and morbidity 2012; Geneva: World Health Organization [WHO].
3. 3.Hill B, Hall J, Currie S. Defining preconception: exploring the concept of a preconception population. BMC Pregnancy and Childbirth. 2020.
4. 4.Caut C, Schoenaker, D., McIntyre, E., Vilcins, D., Gavine, A., Steel, A., . Relationships between women’s and men’s modifiable preconception risks and health behaviors and maternal and offspring health outcomes: an umbrella review. . Seminars in Reproductive Medicine. 2022.
5. Toivonen KI, Oinonen KA, Duchene KM. **Preconception health behaviours: A scoping review**. *Prev Med* (2017.0) **96** 1-15. DOI: 10.1016/j.ypmed.2016.11.022
6. 6.Cardenas S, Morris A, Marshall N, Aviv E, García M, Sellery P, et al. Fathers matter from the start: The role of expectant fathers in child development. Child Develop Perspect. 2021.
7. Shawe J, Patel D, Joy M, Howden B, Barrett G, Stephenson J. **Preparation for fatherhood: a survey of men’s preconception health knowledge and behaviour in England**. *PLoS ONE* (2019.0) **14** e0213897. DOI: 10.1371/journal.pone.0213897
8. Steel A, Carter T. **Balancing our gaze on preconception health and care to include men**. *Advances in Integrative Medicine* (2021.0) **8** 79-80. DOI: 10.1016/j.aimed.2021.04.001
9. Rutkowska J, Lagisz M, Bonduriansky R, Nakagawa S. **Mapping the past, present and future research landscape of paternal effects**. *BMC Biol* (2020.0) **18** 1-24. DOI: 10.1186/s12915-020-00892-3
10. Soubry A. **POHaD: why we should study future fathers**. *Environ Epigenet* (2018.0) **4** dvy007-dvy. DOI: 10.1093/eep/dvy007
11. Kotelchuck M, Lu M. **Father’s role in preconception health**. *Matern Child Health J* (2017.0) **21** 2025-2039. DOI: 10.1007/s10995-017-2370-4
12. Ojukwu O, Patel D, Stephenson J, Howden B, Shawe J. **General practitioners' knowledge, attitudes and views of providing preconception care: a qualitative investigation**. *Upsala J Med Sci* (2016.0) **121** 256-263. DOI: 10.1080/03009734.2016.1215853
13. Mazza D, Chapman A, Michie S. **Barriers to the implementation of preconception care guidelines as perceived by general practitioners: a qualitative study**. *BMC Health Serv Res* (2013.0) **13** 36. DOI: 10.1186/1472-6963-13-36
14. Hogg K, Rizio T, Manocha R, McLachlan RI, Hammarberg K. **Men's preconception health care in Australian general practice: GPs' knowledge, attitudes and behaviours**. *Aust J Prim Health* (2019.0) **25** 353-358. DOI: 10.1071/PY19069
15. 15.Dorney E, Boyle JA, Walker R, Hammarberg K, Musgrave L, Schoenaker D, et al. A Systematic Review of Clinical Guidelines for Preconception Care. Semin Reprod Med. 2022.
16. 16.American Academy of Family Physicians (AAFP). Preconception Care (Position Paper): AAFP; 2015 [Available from: https://www.aafp.org/about/policies/all/preconception-care.html.
17. Cardenas SI, Morris AR, Marshall N, Aviv EC, Martínez García M, Sellery P. **Fathers matter from the start: The role of expectant fathers in child development**. *Child Development Perspectives* (2022.0) **16** 54-59. DOI: 10.1111/cdep.12436
18. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD. **The PRISMA 2020 statement: An updated guideline for reporting systematic reviews**. *PLoS Med* (2021.0) **18** e1003583. DOI: 10.1371/journal.pmed.1003583
19. Shea BJ, Reeves BC, Wells G, Thuku M, Hamel C, Moran J. **AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both**. *BMJ* (2017.0) **358** j4008. DOI: 10.1136/bmj.j4008
20. 20.Covidence systematic review software (COVIDENCE). Melbourne, Australia Veritas Health Innovation; [Available from: www.covidence.org.
21. Luchini C, Stubbs B, Solmi M, Veronese N. **Assessing the quality of studies in meta-analyses: Advantages and limitations of the Newcastle Ottawa Scale**. *World Journal of Meta-Analysis* (2017.0) **5** 80. DOI: 10.13105/wjma.v5.i4.80
22. Sharmin S, Kypri K, Khanam M, Wadolowski M, Bruno R, Mattick RP. **Parental Supply of Alcohol in Childhood and Risky Drinking in Adolescence: Systematic Review and Meta-Analysis**. *Int J Environ Res Public Health* (2017.0) **14** 287. DOI: 10.3390/ijerph14030287
23. Pomeroy E, Wells JCK, Cole TJ, O'Callaghan M, Stock JT. **Relationships of maternal and paternal anthropometry with neonatal body size, proportions and adiposity in an Australian cohort**. *Am J Phys Anthropol* (2015.0) **156** 625-636. DOI: 10.1002/ajpa.22680
24. Zalbahar N, Najman J, McIntyre HD, Mamun A. **Parental pre-pregnancy obesity and the risk of offspring weight and body mass index change from childhood to adulthood**. *Clinical Obesity* (2017.0) **7** 206-215. DOI: 10.1111/cob.12200
25. Bowatte G, Bui DS, Priyankara S, Lowe AJ, Perret JL, Lodge CJ. **Parental preconception BMI trajectories from childhood to adolescence and asthma in the future offspring**. *J Allergy Clin Immunol* (2022.0) **150** 67-74.e30. DOI: 10.1016/j.jaci.2021.11.028
26. Milne E, Greenop KR, Scott RJ, de Klerk NH, Bower C, Ashton LJ. **Parental alcohol consumption and risk of childhood acute lymphoblastic leukemia and brain tumors**. *Cancer Causes Control* (2013.0) **24** 391-402. DOI: 10.1007/s10552-012-0125-5
27. Milne E, Greenop KR, Scott RJ, Ashton LJ, Cohn RJ, De Klerk NH. **Parental smoking and risk of childhood brain tumors**. *Int J Cancer* (2013.0) **133** 253-259. DOI: 10.1002/ijc.28004
28. Greenop KR, Miller M, Bailey HD, Scott RJ, Attia J, Bower C. **Paternal dietary folate, B6 and B12 intake, and the risk of childhood brain tumors**. *Nutr Cancer* (2015.0) **67** 224-230. DOI: 10.1080/01635581.2015.990571
29. Bailey HD, Miller M, Greenop KR, Bower C, Attia J, Marshall GM. **Paternal intake of folate and vitamins B6 and B12 before conception and risk of childhood acute lymphoblastic leukemia**. *Cancer Causes Control* (2014.0) **25** 1615-1625. DOI: 10.1007/s10552-014-0466-3
30. Knudsen GTM, Dharmage S, Janson C, Abramson MJ, Benediktsdóttir B, Malinovschi A. **Parents’ smoking onset before conception as related to body mass index and fat mass in adult offspring: findings from the RHINESSA generation study**. *PLoS ONE* (2020.0) **15** e0235632. DOI: 10.1371/journal.pone.0235632
31. Accordini S, Calciano L, Johannessen A, Portas L, Benediktsdóttir B, Bertelsen RJ. **A three-generation study on the association of tobacco smoking with asthma**. *Int J Epidemiol* (2018.0) **47** 1106-1117. DOI: 10.1093/ije/dyy031
32. Accordini S, Calciano L, Johannessen A, Benediktsdóttir B, Bertelsen RJ, Bråbäck L. **Prenatal and prepubertal exposures to tobacco smoke in men may cause lower lung function in future offspring: a three-generation study using a causal modelling approach**. *Eur Respir J* (2021.0) **58** 2002791. DOI: 10.1183/13993003.02791-2020
33. Johannessen A, Lønnebotn M, Calciano L, Benediktsdóttir B, Bertelsen RJ, Bråbäck L. **Being overweight in childhood, puberty, or early adulthood: Changing asthma risk in the next generation?**. *J Allergy Clin Immunol* (2020.0) **145** 791-9.e4. DOI: 10.1016/j.jaci.2019.08.030
34. Lønnebotn M, Calciano L, Johannessen A, Jarvis DL, Abramson MJ, Benediktsdóttir B. **Parental Prepuberty Overweight and Offspring Lung Function**. *Nutrients* (2022.0) **14** 1506. DOI: 10.3390/nu14071506
35. Broadney MM, Chahal N, Michels KA, McLain AC, Ghassabian A, Lawrence DA. **Impact of parental obesity on neonatal markers of inflammation and immune response**. *Int J Obes* (2017.0) **41** 30-37. DOI: 10.1038/ijo.2016.187
36. Casas M, Forns J, Martinez D, Guxens M, Fernandez-Somoano A, Ibarluzea J. **Maternal pre-pregnancy obesity and neuropsychological development in pre-school children: a prospective cohort study**. *Pediatr Res* (2017.0) **82** 596-606. DOI: 10.1038/pr.2017.118
37. Chen R, Chen L, Liu Y, Wang F, Wang S, Huang Y. **Association of parental prepregnancy BMI with neonatal outcomes and birth defect in fresh embryo transfer cycles: a retrospective cohort study**. *BMC Pregnancy Childbirth* (2021.0) **21** 1-11. DOI: 10.1186/s12884-021-04261-y
38. Fang Y, Liu J, Mao Y, He Y, Li M, Yang L. **Pre-pregnancy body mass index and time to pregnancy among couples pregnant within a year: A China cohort study**. *PLoS ONE [Electronic Resource]* (2020.0) **15** e0231751. DOI: 10.1371/journal.pone.0231751
39. Fleten C, Nystad W, Stigum H, Skjaerven R, Lawlor DA, Davey Smith G. **Parent-offspring body mass index associations in the Norwegian Mother and Child Cohort Study: a family-based approach to studying the role of the intrauterine environment in childhood adiposity**. *Am J Epidemiol* (2012.0) **176** 83-92. DOI: 10.1093/aje/kws134
40. 40.Guo T, Yang Y, Jia J, Deng Y, Wang Y, Zhang Y, et al. Preconception Paternal/Maternal Body Mass Index and Risk of Small/Large for Gestational Age Infant in over 4·7 Million Chinese Women Aged 20–49 Years: A Population-based Cohort Study in China. Br J Nutr. 2022:1–28.
41. Hoek J, Schoenmakers S, van Duijn L, Willemsen SP, van Marion ES, Laven JSE. **A higher preconceptional paternal body mass index influences fertilization rate and preimplantation embryo development**. *Andrology* (2022.0) **10** 486-494. DOI: 10.1111/andr.13128
42. Moss JL, Harris KM. **Impact of maternal and paternal preconception health on birth outcomes using prospective couples' data in Add Health**. *Arch Gynecol Obstet* (2015.0) **291** 287-298. DOI: 10.1007/s00404-014-3521-0
43. Mutsaerts MA, Groen H, Buiter-Van der Meer A, Sijtsma A, Sauer PJ, Land JA. **Effects of paternal and maternal lifestyle factors on pregnancy complications and perinatal outcome. A population-based birth-cohort study: the GECKO Drenthe cohort**. *Human Reproduction* (2014.0) **29** 824-34. DOI: 10.1093/humrep/deu006
44. Noor N, Cardenas A, Rifas-Shiman SL, Pan H, Dreyfuss JM, Oken E. **Association of Periconception Paternal Body Mass Index With Persistent Changes in DNA Methylation of Offspring in Childhood**. *JAMA Netw Open* (2019.0) **2** e1916777. DOI: 10.1001/jamanetworkopen.2019.16777
45. Retnakaran R, Wen SW, Tan H, Zhou S, Ye C, Shen M. **Paternal weight prior to conception and infant birthweight: a prospective cohort study**. *Nutr Diabetes* (2021.0) **11** 28. DOI: 10.1038/s41387-021-00172-1
46. Robinson SL, Ghassabian A, Sundaram R, Trinh M-H, Lin T-C, Bell EM. **Parental Weight Status and Offspring Behavioral Problems and Psychiatric Symptoms**. *J Pediatr* (2020.0) **220** 227-36.e1. DOI: 10.1016/j.jpeds.2020.01.016
47. Sun M, Zhang S, Chen L, Li Y, Diao J, Li J. **Association between paternal pre-pregnancy body mass index with preterm birth and low birth weight**. *Front Pediatr* (2022.0) **10** 955544. DOI: 10.3389/fped.2022.955544
48. Sundaram R, Mumford SL, Buck Louis GM. **Couples' body composition and time-to-pregnancy**. *Hum Reprod* (2017.0) **32** 662-668. DOI: 10.1093/humrep/dex001
49. Umul M, Köse SA, Bilen E, Altuncu AG, Oksay T, Güney M. **Effect of increasing paternal body mass index on pregnancy and live birth rates in couples undergoing intracytoplasmic sperm injection**. *Andrologia* (2015.0) **47** 360-364. DOI: 10.1111/and.12272
50. Wei J, Wang T, Shu J, Liu Y, Song X, Sun M. **Parental pre-pregnancy body mass index and risk of low birth weight in offspring: A prospective cohort study in central China**. *Front Public Health* (2022.0) **10** 1036689. DOI: 10.3389/fpubh.2022.1036689
51. 51.Wei B, Shao Y, Liang J, Tang P, Mo M, Liu B, et al. Maternal overweight but not paternal
overweight before pregnancy is associated with shorter newborn telomere length: evidence from Guangxi Zhuang birth cohort in China. BMC Pregnancy Childbirth. 2021;21(1):283.
52. 52.Xu R, Zhao W, Tan T, Li H, Wan Y. Paternal body mass index before conception associated with offspring’s birth weight in Chinese population: a prospective study. medRxiv. 2021:2021.06.17.21258438.
53. Yang S, Zhou A, Xiong C, Yang R, Bassig BA, Hu R. **Parental Body Mass Index, Gestational Weight Gain, and Risk of Macrosomia: A Population-Based Case-Control Study in China**. *Paediatr Perinat Epidemiol* (2015.0) **29** 462-471. DOI: 10.1111/ppe.12213
54. Zhang Y, Zhang J, Zhao J, Hong X, Zhang H, Dai Q. **Couples’ prepregnancy body mass index and time to pregnancy among those attempting to conceive their first pregnancy**. *Fertil Steril* (2020.0) **114** 1067-1075. DOI: 10.1016/j.fertnstert.2020.05.041
55. Luan M, Zhang X, Fang G, Liang H, Yang F, Song X. **Preconceptional paternal alcohol consumption and the risk of child behavioral problems: a prospective cohort study**. *Sci Rep* (2022.0) **12** 1-11. DOI: 10.1038/s41598-022-05611-2
56. Xia R, Jin L, Li D, Liang H, Yang F, Chen J. **Association Between Paternal Alcohol Consumption Before Conception and Anogenital Distance of Offspring. Alcoholism**. *Clin Exp Res* (2018.0) **42** 735-42. DOI: 10.1111/acer.13595
57. Zuccolo L, DeRoo LA, Wills AK, Davey Smith G, Suren P, Roth C. **Pre-conception and prenatal alcohol exposure from mothers and fathers drinking and head circumference: results from the Norwegian Mother-Child Study (MoBa)**. *Sci Rep* (2016.0) **7** 39535. DOI: 10.1038/srep39535
58. Har-Gil E, Heled A, Dixon M, Ahamed AMS, Bentov Y. **The relationship between cannabis use and IVF outcome—a cohort study**. *J Cannabis Res* (2021.0) **3** 42. DOI: 10.1186/s42238-021-00099-5
59. Kasman AM, Thoma ME, McLain AC, Eisenberg ML. **Association between use of marijuana and time to pregnancy in men and women: findings from the National Survey of Family Growth**. *Fertil Steril* (2018.0) **109** 866-871. DOI: 10.1016/j.fertnstert.2018.01.015
60. Nassan FL, Arvizu M, Mínguez-Alarcón L, Gaskins AJ, Williams PL, Petrozza JC. **Marijuana smoking and outcomes of infertility treatment with assisted reproductive technologies**. *Hum Reprod* (2019.0) **34** 1818-1829. DOI: 10.1093/humrep/dez098
61. Wise LA, Wesselink AK, Hatch EE, Rothman KJ, Mikkelsen EM, Sørensen HT. **Marijuana use and fecundability in a North American preconception cohort study**. *J Epidemiol Community Health* (2018.0) **72** 208-215. DOI: 10.1136/jech-2017-209755
62. Carslake D, Pinger PR, Romundstad P, Davey SG. **Early-onset paternal smoking and offspring adiposity: further investigation of a potential intergenerational effect using the HUNT study**. *PLoS ONE* (2016.0) **11** e0166952. DOI: 10.1371/journal.pone.0166952
63. Deng K, Liu Z, Lin Y, Mu D, Chen X, Li J. **Periconceptional paternal smoking and the risk of congenital heart defects: A case-control study**. *Birth Defects Research Part A - Clinical and Molecular Teratology* (2013.0) **97** 210-216. DOI: 10.1002/bdra.23128
64. Frederiksen LE, Erdmann F, Wesseling C, Winther JF, Mora AM. **Parental tobacco smoking and risk of childhood leukemia in Costa Rica: a population-based case-control study**. *Environ Res* (2020.0) **180** 108827. DOI: 10.1016/j.envres.2019.108827
65. Ko TJ, Tsai LY, Chu LC, Yeh SJ, Leung C, Chen CY. **Parental smoking during pregnancy and its association with low birth weight, small for gestational age, and preterm birth offspring: A birth cohort study**. *Pediatr Neonatol* (2014.0) **55** 20-27. DOI: 10.1016/j.pedneo.2013.05.005
66. Northstone K, Golding J, Davey Smith G, Miller LL, Pembrey M. **Prepubertal start of father's smoking and increased body fat in his sons: further characterisation of paternal transgenerational responses**. *Eur J Hum Genet* (2014.0) **22** 1382-1386. DOI: 10.1038/ejhg.2014.31
67. Orsi L, Rudant J, Ajrouche R, Leverger G, Baruchel A, Nelken B. **Parental smoking, maternal alcohol, coffee and tea consumption during pregnancy, and childhood acute leukemia: the ESTELLE study**. *Cancer Causes Control* (2015.0) **26** 1003-1017. DOI: 10.1007/s10552-015-0593-5
68. Sapra KJ, Barr DB, Maisog JM, Sundaram R, Buck Louis GM. **Time-to-Pregnancy Associated With Couples' Use of Tobacco Products**. *Nicotine Tob Res* (2016.0) **18** 2154-2161. DOI: 10.1093/ntr/ntw132
69. Svanes C, Koplin J, Skulstad SM, Johannessen A, Bertelsen RJ, Benediktsdottir B. **Father's environment before conception and asthma risk in his children: A multi-generation analysis of the Respiratory Health In Northern Europe study**. *Int J Epidemiol* (2017.0) **46** 235-245. PMID: 27565179
70. Wang L, Deng Y, Yang Y, Liu F, Xu Q, Peng Z. **Paternal smoking and preterm birth: a population-based retrospective cohort study among non-smoking women aged 20–49 years in rural China**. *Reprod Health* (2022.0) **19** 72. DOI: 10.1186/s12978-022-01378-x
71. Wang L, Yang Y, Liu F, Yang A, Xu Q, Wang Q. **Paternal smoking and spontaneous abortion: A population-based retrospective cohort study among non-smoking women aged 20–49 years in rural China**. *J Epidemiol Community Health* (2018.0) **72** 783-789. DOI: 10.1136/jech-2017-210311
72. Wesselink AK, Hatch EE, Rothman KJ, Mikkelsen EM, Aschengrau A, Wise LA. **Prospective study of cigarette smoking and fecundability**. *Hum Reprod* (2019.0) **34** 558-567. DOI: 10.1093/humrep/dey372
73. You Y, Liu R, Zhou H, Wu R, Lin R, Li B. **Effect of Exposure to Paternal Smoking on Overweight and Obesity in Children: Findings from the Children Lifeway Cohort in Shenzhen**. *Southern China Obes Facts* (2022.0) **15** 609-620. DOI: 10.1159/000525544
74. Zhou Q, Zhang S, Wang Q, Shen H, Zhang Y, Tian W. **Association between preconception paternal smoking and birth defects in offspring: evidence from the database of the National Free Preconception Health Examination Project in China**. *BJOG* (2020.0) **127** 1358-64. DOI: 10.1111/1471-0528.16277
75. Zwink N, Choinitzki V, Baudisch F, Holscher A, Boemers TM, Turial S. **Comparison of environmental risk factors for esophageal atresia, anorectal malformations, and the combined phenotype in 263 German families**. *Dis Esophagus* (2016.0) **29** 1032-1042. DOI: 10.1111/dote.12431
76. Bae J, Lynch CD, Kim S, Sundaram R, Sapra KJ, Buck Louis GM. **Preconception stress and the secondary sex ratio in a population-based preconception cohort**. *Fertil Steril* (2017.0) **107** 714-722. DOI: 10.1016/j.fertnstert.2016.12.011
77. Wesselink AK, Hatch EE, Rothman KJ, Weuve JL, Aschengrau A, Song RJ. **Perceived Stress and Fecundability: A Preconception Cohort Study of North American Couples**. *Am J Epidemiol* (2018.0) **187** 2662-2671. DOI: 10.1093/aje/kwy186
78. Hatch EE, Wesselink AK, Hahn KA, Michiel JJ, Mikkelsen EM, Sorensen HT. **Intake of sugar-sweetened beverages and fecundability in a North American preconception cohort**. *Epidemiology* (2018.0) **29** 369. DOI: 10.1097/EDE.0000000000000812
79. Hoek J, Koster MPH, Schoenmakers S, Willemsen SP, Koning AHJ, Steegers EAP. **Does the father matter? the association between the periconceptional paternal folate status and embryonic growth**. *J Urol* (2019.0) **202** 446. DOI: 10.1097/01.JU.0000574404.49016.fd
80. Lippevelde W, Vik F, Wills A, Strömmer S, Barker M, Skreden M. **The impact of diet during adolescence on the neonatal health of offspring: evidence on the importance of preconception diet. The HUNT study**. *J Dev Orig Health Dis* (2020.0) **12** 1-13
81. Martín-Calvo N, Mínguez-Alarcón L, Gaskins AJ, Nassan FL, Williams PL, Souter I. **Paternal preconception folate intake in relation to gestational age at delivery and birthweight of newborns conceived through assisted reproduction**. *Reprod Biomed Online* (2019.0) **39** 835-843. DOI: 10.1016/j.rbmo.2019.07.005
82. Mitsunami M, Salas-Huetos A, Mínguez-Alarcón L, Attaman JA, Ford JB, Kathrins M. **Men's dietary patterns in relation to infertility treatment outcomes among couples undergoing in vitro fertilization**. *J Assist Reprod Genet* (2021.0) **38** 2307-2318. DOI: 10.1007/s10815-021-02251-9
83. Oostingh EC, de Vos I, Ham AC, Brouwer-Brolsma EM, Willemsen SP, Eggink AJ. **No independent associations between preconception paternal dietary patterns and embryonic growth; the Predict Study**. *Clin Nutr* (2019.0) **38** 2333-2341. DOI: 10.1016/j.clnu.2018.10.011
84. Twigt JM, Bolhuis ME, Steegers EA, Hammiche F, van Inzen WG, Laven JS. **The preconception diet is associated with the chance of ongoing pregnancy in women undergoing IVF/ICSI treatment**. *Hum Reprod* (2012.0) **27** 2526-2531. DOI: 10.1093/humrep/des157
85. Wesselink AK, Wise LA, Rothman KJ, Hahn KA, Mikkelsen EM, Mahalingaiah S. **Caffeine and caffeinated beverage consumption and fecundability in a preconception cohort**. *Reprod Toxicol* (2016.0) **62** 39-45. DOI: 10.1016/j.reprotox.2016.04.022
86. Xia W, Chiu YH, Afeiche MC, Williams PL, Ford JB, Tanrikut C. **Impact of men's dairy intake on assisted reproductive technology outcomes among couples attending a fertility clinic**. *Andrology* (2016.0) **4** 277-283. DOI: 10.1111/andr.12151
87. Xia W, Chiu YH, Williams PL, Gaskins AJ, Toth TL, Tanrikut C. **Men's meat intake and treatment outcomes among couples undergoing assisted reproduction**. *Fertil Steril* (2015.0) **104** 972-979. DOI: 10.1016/j.fertnstert.2015.06.037
88. Mutsaerts MAQ, Groen H, Buiter-Van Der Meer A, Sijtsma A, Sauer PJJ, Land JA. **Effects of paternal and maternal lifestyle factors on pregnancy complications and perinatal outcome. A population-based birth-cohort study: The GECKO Drenthe cohort**. *Human Reprod* (2014.0) **29** 824-34. DOI: 10.1093/humrep/deu006
89. 89.Lassi ZS, Imam AM, Dean SV, Bhutta ZA. Preconception care: Caffeine, smoking, alcohol, drugs and other environmental chemical/radiation exposure. Reproduct Health. 2014;11(Supplement 3).
90. Perry MF, Mulcahy H, DeFranco EA. **Influence of periconception smoking behavior on birth defect risk**. *Am J Obstet Gynecol* (2019.0) **220** 588.e1-e7. DOI: 10.1016/j.ajog.2019.02.029
91. Bolin EH, Gokun Y, Romitti PA, Tinker SC, Summers AD, Roberson PK. **Maternal Smoking and Congenital Heart Defects, National Birth Defects Prevention Study, 1997–2011**. *J Pediatr* (2022.0) **240** 79-86.e1. DOI: 10.1016/j.jpeds.2021.09.005
92. Metayer C, Dahl G, Wiemels J, Miller M. **Childhood Leukemia: A Preventable Disease**. *Pediatrics* (2016.0) **138** S45-s55. DOI: 10.1542/peds.2015-4268H
93. Chunxia D, Meifang W, Jianhua Z, Ruijuan Z, Xiue L, Zhuanzhen Z. **Tobacco smoke exposure and the risk of childhood acute lymphoblastic leukemia and acute myeloid leukemia: A meta-analysis**. *Medicine (Baltimore)* (2019.0) **98** e16454. DOI: 10.1097/MD.0000000000016454
94. Sharma R, Harlev A, Agarwal A, Esteves SC. **Cigarette Smoking and Semen Quality: A New Meta-analysis Examining the Effect of the 2010 World Health Organization Laboratory Methods for the Examination of Human Semen**. *Eur Urol* (2016.0) **70** 635-645. DOI: 10.1016/j.eururo.2016.04.010
95. Flemming K, Graham H, McCaughan D, Angus K, Bauld L. **The barriers and facilitators to smoking cessation experienced by women’s partners during pregnancy and the post-partum period: a systematic review of qualitative research**. *BMC Public Health* (2015.0) **15** 849. DOI: 10.1186/s12889-015-2163-x
96. Morphett K, Partridge B, Gartner C, Carter A, Hall W. **Why Don't Smokers Want Help to Quit? A Qualitative Study of Smokers' Attitudes towards Assisted vs. Unassisted Quitting**. *Int J Environ Res Public Health* (2015.0) **12** 6591-607. DOI: 10.3390/ijerph120606591
97. 97.World Health Organization [WHO]. Birth defects 2022 [Available from: https://www.who.int/news-room/fact-sheets/detail/birth-defects.
98. Higgins ST, Kurti AN, Redner R, White TJ, Gaalema DE, Roberts ME. **A literature review on prevalence of gender differences and intersections with other vulnerabilities to tobacco use in the United States, 2004–2014**. *Prev Med* (2015.0) **80** 89-100. DOI: 10.1016/j.ypmed.2015.06.009
99. Bahreynian M, Qorbani M, Khaniabadi BM, Motlagh ME, Safari O, Asayesh H. **Association between Obesity and Parental Weight Status in Children and Adolescents**. *J Clin Res Pediatr Endocrinol* (2017.0) **9** 111-117. DOI: 10.4274/jcrpe.3790
100. Patel C, Karasouli E, Shuttlewood E, Meyer C. **Food Parenting Practices among Parents with Overweight and Obesity: A Systematic Review**. *Nutrients* (2018.0) **10** 1966. DOI: 10.3390/nu10121966
101. Toschke AM, Beyerlein A, von Kries R. **Children at high risk for overweight: a classification and regression trees analysis approach**. *Obes Res* (2005.0) **13** 1270-1274. DOI: 10.1038/oby.2005.151
102. Schnurr TM, Morgen CS, Borisevich D, Beaumont RN, Engelbrechtsen L, Ängquist L. **The influence of transmitted and non-transmitted parental BMI-associated alleles on the risk of overweight in childhood**. *Sci Rep* (2020.0) **10** 4806. DOI: 10.1038/s41598-020-61719-3
103. Fleming TP, Watkins AJ, Velazquez MA, Mathers JC, Prentice AM, Stephenson J. **Origins of lifetime health around the time of conception: causes and consequences**. *The Lancet* (2018.0) **391** 1842-1852. DOI: 10.1016/S0140-6736(18)30312-X
104. Hidding LM, Altenburg TM, van Ekris E, Chinapaw MJM. **Why Do Children Engage in Sedentary Behavior? Child- and Parent-Perceived Determinants**. *Int J Environ Res Public Health* (2017.0) **14** 671. DOI: 10.3390/ijerph14070671
105. Qiao Y, Ma J, Wang Y, Li W, Katzmarzyk PT, Chaput JP. **Birth weight and childhood obesity: a 12-country study**. *Int J Obes Suppl* (2015.0) **5** S74-S79. DOI: 10.1038/ijosup.2015.23
106. Zhao Y, Wang S-F, Mu M, Sheng J. **Birth weight and overweight/obesity in adults: a meta-analysis**. *Eur J Pediatr* (2012.0) **171** 1737-1746. DOI: 10.1007/s00431-012-1701-0
107. Evensen E, Emaus N, Kokkvoll A, Wilsgaard T, Furberg A-S, Skeie G. **The relation between birthweight, childhood body mass index, and overweight and obesity in late adolescence: a longitudinal cohort study from Norway, The Tromsø Study, Fit Futures**. *BMJ Open* (2017.0) **7** e015576. DOI: 10.1136/bmjopen-2016-015576
108. 108.World Health Organization [WHO]. Global action plan on physical activity 2018–2030: more active people for a healthier world 2018 [Available from: https://www.who.int/publications/i/item/9789241514187.
109. 109.Australian Government: Department of Health. National Obesity Strategy 2022–2032. Government 2022.
110. 110.Centers for Disease Control and Prevention [CDC]. Prevention Strategies & Guidelines: CDC; 2018 [Available from: https://www.cdc.gov/obesity/resources/strategies-guidelines.html.
111. Wahab RJ, Jaddoe VWV, van Klaveren D, Vermeulen MJ, Reiss IKM, Steegers EAP. **Preconception and early-pregnancy risk prediction for birth complications: development of prediction models within a population-based prospective cohort**. *BMC Pregnancy Childbirth* (2022.0) **22** 165. DOI: 10.1186/s12884-022-04497-2
112. Mastroiacovo P, Nilsen RM, Leoncini E, Gastaldi P, Allegri V, Boiani A. **Prevalence of maternal preconception risk factors: an Italian multicenter survey**. *Ital J Pediatr* (2014.0) **40** 91. DOI: 10.1186/s13052-014-0091-5
113. Althubaiti A. **Information bias in health research: definition, pitfalls, and adjustment methods**. *J Multidiscip Healthc* (2016.0) **9** 211-217. DOI: 10.2147/JMDH.S104807
|
---
title: 'Nutritional status, symptom burden, and predictive validity of the Pt-Global
web tool/PG-SGA in CKD patients: A hospital based cross sectional study'
authors:
- Ishfaq Rashid
- Pramil Tiwari
- Sanjay D’Cruz
- Shivani Jaswal
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022301
doi: 10.1371/journal.pgph.0001301
license: CC BY 4.0
---
# Nutritional status, symptom burden, and predictive validity of the Pt-Global web tool/PG-SGA in CKD patients: A hospital based cross sectional study
## Abstract
### Background
Despite not being frequently recognized, malnutrition, a consequence of chronic kidney disease, negatively affects morbidity, mortality, functional activity, and patient’s quality of life. Management of this condition is made more difficult by the dearth of knowledge regarding the symptom burden brought on by inadequate nutritional status. Additionally, there are multiple tools to evaluate nutritional status in CKD; but, Pt-Global web tool/PG-SGA used in oncology, has not been investigated in chronic kidney disease patients. This study aimed to explore the nutritional status, symptom burden and also investigate the predictive validity of Pt-Global web tool/PG-SGA among pre-dialysis diabetic and non-diabetic chronic kidney disease patients.
### Methodology
This cross-sectional study was carried out at a renal clinic of a tertiary care public teaching hospital. Nutritional status and symptom burden was evaluated by employing a ‘Pt-Global web tool/PG-SGA’ which is considered as a preeminent interdisciplinary tool in oncology and other chronic catabolic conditions. The predictive validity of the Pt-Global web tool/PG-SGA, referred as overall score for malnutrition was ascertained using Receiver Operating Curves (ROC). The conclusions were drawn using descriptive statistics, correlation, and regression analysis.
### Results
In a sample of 450 pre-dialysis CKD patients, the malnutrition was present in 292($64.9\%$) patients. Diabetic CKD patients exhibit higher proportion of malnutrition 159($35.3\%$). The prevalence of malnutrition was exacerbated by eGFR reduction. The overall Pt-Global web tool/PGA-SGA score was significantly influenced by the symptoms of fatigue ($81.5\%$), appetite loss ($54.8\%$), physical pain ($45.3\%$), constipation ($31.78\%$), dry mouth ($26.2\%$), and feeling full quickly ($25.8\%$). The ROC analysis showed that the AUC for the total PG-SGA score was 0.988 ($95\%$ CI: 0.976–1.000), indicating that it is a reliable indicator of malnutrition. The sensitivity ($84.2\%$) for identifying malnutrition was low when using the conventional tool cut off score of ≥9. Instead, it was discovered that a score of ≥3 had a greater sensitivity ($99.3\%$) and specificity ($44.3\%$) and was therefore recommended.
### Conclusions
This study not only presents empirical evidence of poor nutritional status in CKD patients but also reveals that it is worse in patients with diabetes, hypoalbuminemia, and poorer kidney function (well recognized risk factors for cardiovascular disease). Early diagnosis and management of symptoms contributing malnutrition will reduce mortality and CKD progression. The Pt-Global web tool/PG-SGA total score of 3 or more appears to be the ideal cut off score for identifying malnutrition, which can be utilized by dietician for improving malnutrition.
## Introduction
Malnutrition describes a state resulting from lack/inadequate intake or uptake of nutrition that leads to altered body composition (decreased fat free mass) and body cell mass [1]. Even though it is not frequently recognized, this complication of chronic kidney disease has been shown to have an impact on prognosis, physical and mental function, and quality of life [2,3]. Additionally, mortality and the fast advancement of CKD to end-stage renal disease are linked to it [3–5].
According to data from the Global Burden of Disease, chronic kidney disease (CKD) ranks among the leading causes of death globally [6]. Earlier, it was predicted that 1.4 million deaths will be attributable to CKD in 2019, a $20\%$ increase from data from 2010 [7]. Low- and middle-income countries have been severely impacted by its disproportionate rise in prevalence and mortality [8,9]. Poor nutritional status or malnutrition is an under-recognized and undertreated condition that has a high clinical impact on patients across different healthcare settings [10–12]. Patients with chronic renal disease frequently have this complication [13,14]. According to the available research, there is a clear connection between poor nutritional status and morbidity, the geriatric population, and gender [15,16]. The effects are catastrophic, especially in terms of substantial clinical and financial costs, lowered immunity, frequent infections, death, general physical and mental deterioration, and poor quality of life [2,3,17,18].
Nutrition shares a strong bond with the renal outcomes. India has reported a huge prevalence of malnutrition ($56.7\%$) among chronic kidney disease patients [13]. The prevalence of malnutrition, however, varies depending on the stages of CKD, gender, and dialysis status [19]. In developing nations, it has been reported that quantification of malnutrition with respect to dialysis status, CKD stages and gender is poorly characterized before the initiation of dialysis [13]. As a result, routine screening with a validated method to determine the patient’s current nutritional status is a crucial step in identifying individuals at risk of malnutrition and are consequently more vulnerable to illnesses.
Various nutritional assessment tools in CKD patients have been utilized in clinical practice [20–22]. However, the subjective global assessment (SGA) instrument has been recommended by the National Kidney Foundation’s Kidney Disease/Dialysis Outcomes and Quality Initiative (KDOQI) as the diagnostic benchmark for the nutritional assessment [23]. This frequently used tool (SGA) in clinical practice has undergone significant advances in recent years [24]. This tool has been made now available as a web-based programme (Pt-Global web tool/PG-SGA) [25] which is practical, affordable, simple to use, and is non-intrusive in character. This tool also offers automatic solutions to the problems after nutritional assessment.
The predictive validity of manual SGA score system regarding the mortality has been already established in conservatively treated chronic kidney disease patients and/or hemodialysis patients [26–28]. However, the predictive validity of Pt-Global web tool/PG-SGA in non-dialysis chronic kidney disease patients has not been investigated so far. Further, because the patient’s symptom burden has not yet been clarified, it is questionable what underlying factors exist and contribute to malnutrition [29]. Therefore, it is relevant for a healthcare professional to strive for the early detection of malnutrition among CKD patients through the necessary assessment of nutritional status; and, plan for management of the condition through dietary plans, nutritional goals, and intervention measures.
Given these gaps, this prospective observational study aimed to evaluate the nutritional status, symptom burden, and also investigate the predictive validity of Pt-Global web tool/PG-SGA among pre-dialysis diabetic and non-diabetic chronic kidney disease patients at a tertiary care public teaching hospital.
## Study design, setting and subjects
This prospective cross-sectional study was carried out in the renal clinic, Department of General Medicine at a tertiary care public teaching hospital which receive referrals of chronic kidney disease patients from within and outside the state of location.
Patients with non-dialysis chronic kidney disease of either gender were recruited by convenience sampling from the outpatient department (OPD) between August 2019 and November 2021. The inclusion criteria covered patients ≥18 years old, conscious and alert, non-dialyzed, willing to participate and if their medical file contained all the demographic, clinical and biochemical data required for the study. Patients who were on dialysis, seriously ill, or unable to communicate verbally or mentally were not included in the study.
## Study procedure
The study procedure involved the data collection and patient interview for nutritional status using Pt-Global web too/PG-SGA. Data were gathered using a self-administered case record form that asked for socio-demographic details like age, gender, residency, height (measured with an inch tape), weight (recorded using a standard calibrated weighing scale), body mass index (BMI), social history (smoking/drinking), and socio-economic status (occupation, education, and income) by modified kuppuswamy scale. This case record form was also used to collect the clinical status of the patients (i.e., the diagnosis, biochemical parameter estimation, current medications and the number of comorbidities by charlson comorbidity index).
The ideal overall metric of kidney function is the glomerular filtration rate (GFR). Normal GFR varies according to age, sex, and body size, and declines with age. GFR was estimated by using the National Kidney Foundation recommended CKD-EPI equation web application. Estimated GFR was used to stage CKD as stage 1 (GFR ≥ 90 ml/min), stage 2 (GFR 60–89ml/min), stage 3 (GFR = 59–30 ml/min), stage 4 (GFR = 15–29 ml/ min) and stage 5 (GFR < 15ml/min) [30].
## Pt-Global web tool/PG-SGA
There are two sections in this tool [25]. The first section is based on patient grading and contains information about the patient’s medical history (weight change, changes in food intake, gastrointestinal symptoms, activities, and functional capacity), while the second section is based on physician grading and includes information about disease status (comorbidities and advanced age), metabolic demand (fever, fever duration, and use of corticosteroids), as well as a physical examination (subcutaneous fat loss and signs of muscle wasting).
The evaluation method of the Pt-Global web tool/PG-SGA is automatic and quick. It provides "nutritional triage recommendations" for inadequate nutritional status after completion based on its scoring, and it also gives the user the option to save the nutritional status report on email for later use. The additive score of the Pt-Global web tool PG-SGA is taken into consideration by nutritional triage recommendations in order to determine the precise nutritional interventions. Optimal symptom management is a component of the first line nutritional intervention. The nutritional interventions include patient & family education, symptom management including pharmacologic intervention, and appropriate nutrient intervention (food, nutritional supplements, enteral, or parenteral triage). Nutritional triage is determined by the Pt-Global web tool/PG-SGA score. Scores of 0 or 1 indicate that no immediate action is required, but that treatment-related decisions will be periodically and routinely reviewed; Score 2–3: Pharmacologic intervention as indicated by the symptom survey parameter and pertinent lab values, together with patient and family education provided by a dietitian, nurse, or other clinician; Scores of 4 to 8 indicate the need for dietician intervention in conjunction with medical or nursing care as indicated by symptoms, while a score of ≥9 indicates an urgent need for improved symptom management and/or nutrient intervention options. The license to use this tool was purchased and granted by the copyright owner. This Pt-Global web tool is now available at https://pt-global.org/pt-global-app/. On evaluation the patients were either classified as well nourished (category A), moderately malnourished or suspected of being malnourished (category B), or severely malnourished (category C). The patient-generated component and the professional component’s scores were added to create the Pt-Global web tool/PG-overall SGA’s score, with a higher score indicating more severe malnutrition. Age, stages of chronic illness, and diagnosis all had an impact on the final score. These factors were retrieved from medical records. This instrument has been validated by Faith Ottery and the Hanze University of Applied Sciences in populations with cancer and (fragile) elderly individuals [25].
## Ethics approval
The concerned Human Ethics Committee has approved the sample and data collection and granted permission to access and use the patient’s clinical data. The samples collected were assessed by the Department of Biochemistry at the study site. All eligible participants were informed about the study protocol and written, signed consent was obtained from all the participants before their inclusion in the study. This study was conducted in adherence to the Declaration of Helsinki.
## Statistical analysis
Selected demographic and clinical characteristics of the study participants were evaluated using descriptive statistics. The continuous variables were expressed as mean standard deviation, while the categorical variables were expressed as percentages and compared using the chi-square test. After ensuring that the distribution of continuous variables was normally distributed using the Kolmogorov-Smirnov test or by using analysis of variance (ANOVA) when comparing more groups, the student’s t-test for independent samples was used to compare the continuous variables. A univariate spearman’s correlation analysis was also conducted using a simple linear correlation, with the exception of categorical variables, for which the chi-square test was applied. A multivariate linear regression analysis and a multinomial logistic regression analysis were carried out to investigate the variables associated to nutritional status and/or SGA score. To determine the sensitivity, specificity, and ideal PG-SGA total cut off score for identifying malnourished non-dialysis patients, the receiver operating characteristics (ROC) curve was used. An area under the curve (AUC) of 0.9–1 indicates the PG-SGA score was an excellent measure for detecting malnutrition; 0.8–0.89 a good test; and 0.70–0.79 a fair test [31]. Significance was fixed at $P \leq 0.05.$ The statistical analyses were performed by using SPSS software (version 25.0; SPSS Inc, Chicago, IL).
## Patient characteristics
During the course of study, a total of 510 patients were enrolled. The patients with incomplete datasets ($$n = 30$$), those who refused to give their consent ($$n = 15$$), and those who had Ig nephropathy/AKI ($$n = 15$$) were not included in the analysis.
A total of 450 individuals with chronic kidney disease, aged 53.9±14.2 (265 men and 185 women), were eligible for final analysis. [ Table 1] Almost two third ($61.6\%$, $\frac{277}{450}$) of the sample were aged between 30 and 59 years old; and more than one third ($38.4\%$, $\frac{175}{450}$) were aged ≥60 years.
**Table 1**
| Variable | No of participantsn = 450 (%) | Well-nourishedSGA-AN = 158 (35.1%) | Mild-ModerateSGA-BN = 140 (31.1%) | MalnourishedSGA-CN = 152 (33.8%) | p-value |
| --- | --- | --- | --- | --- | --- |
| Age groups (Years) | | | | | ≤0.005 |
| Adult (<60) | 277 (61.6) | 117 (26.0) | 81 (18.0) | 79 (17.6) | |
| Elderly (≥60) | 173 (38.4) | 41 (9.1) | 59 (13.1) | 73 (16.2) | |
| Gender | | | | | 0.866 |
| Male | 265 (58.9) | 96 (21.3) | 78 (17.3) | 91 (20.2) | |
| Female | 185 (41.1) | 62 (13.8) | 62 (13.8) | 61 (13.6) | |
| Body Mass Index [24] | | | | | 0.763 |
| Underweight | 40 (8.9) | 14 (3.1) | 9 (2.0) | 17 (3.8) | |
| Normal | 215 (47.8) | 72 (16.0) | 71 (15.8) | 72 (16.0) | |
| Overweight | 159 (65.3) | 60 (13.3) | 49 (10.9) | 50 (11.1) | |
| Obese | 36 (8.0) | 12 (2.7) | 11 (2.4) | 13 (2.9) | |
| Socio-economic status | | | | | 0.301 |
| Lower | 4 (0.9) | 69 (15.3) | 61 (13.6) | 76 (16.9) | |
| Middle | 240 (53.3) | 87 (19.3) | 77 (17.1) | 76 (16.9) | |
| Higher | 206 (45.8) | 2 (0.4) | 2 (0.4) | 0 (0.0) | |
| Residence | | | | | 0.802 |
| Rural | 241 (53.6) | 83 (18.4) | 76 (16.9) | 82 (18.2) | |
| Urban | 209 (46.4) | 75 (16.7) | 64 (14.2) | 70 (15.6) | |
| Smoking | | | | | 0.661 |
| Yes | 371 (82.4) | 27 (6.0) | 29 (6.4) | 23 (5.1) | |
| No | 79 (17.6) | 131 (29.1) | 111 (24.7) | 129 (28.7) | |
| Alcohol | | | | | 0.680 |
| Yes | 196 (43.6) | 71 (15.8) | 53 (11.8) | 72 (16.0) | |
| No | 254 (56.4) | 87 (19.3) | 87 (19.3) | 80 (17.8) | |
| Diabetes status | | | | | ≤0.005 |
| Yes | 210 (46.7) | 51 (11.3) | 68 (15.1) | 91 (20.2) | |
| No | 240 (53.3) | 107 (23.8) | 72 (16.0) | 61 (13.6) | |
| CKD Stages | | | | | ≤0.005 |
| CKD Stage 1 | 12 (2.7) | 5 (1.1) | 7 (1.6) | 0 (0.0) | |
| CKD Stage 2 | 18 (4.0) | 10 (2.2) | 7 (1.6) | 1 (0.2) | |
| CKD Stage 3a | 34 (7.6) | 19 (4.2) | 11 (2.4) | 4 (0.9) | |
| CKD Stage 3b | 112 (24.9) | 54 (12.0) | 30 (6.7) | 28 (6.2) | |
| CKD Stage 4 | 194 (43.1) | 62 (13.8) | 60 (13.3) | 72 (16.0) | |
| CKD Stage 5 | 80 (17.8) | 8 (1.8) | 25 (5.6) | 47 (10.4) | |
## Prevalence of malnutrition
According to this web based subjective global assessment tool (SGA), severe malnutrition was present in 152($33.8\%$) patients, 140($31.1\%$) patients were mildly or moderately malnourished while 158($35.1\%$) patients were well-nourished. Male patients were more adversely affected with malnutrition (severe 91 ($20.2\%$) & mild/moderate 78($17.3\%$)) than female patients [Table 1].
CKD with diabetes were more severely affected with malnutrition (mild/moderate 68($15.1\%$) & severe 91($20.2\%$)) as compared to chronic kidney disease patients without diabetes (mild/moderate 72($16.0\%$) & severe 61 ($13.6\%$)). The nutritional status was also evaluated for different stages of chronic kidney disease. The prevalence of malnutrition increased with decline of renal residual function; from $3.4\%$ (mild/moderate & severe) in CKD stage 1–2 to $16.2\%$ in CKD stage 3a-3b and $45.3\%$ in CKD stages 4–5 [Table 1].
For sociodemographic details, there was a statistically significant difference in the mean of nutritional status categories for the age groups (adult & elderly), diabetic status (CKD with or without diabetes), CKD stages (stage 1–5) as determined by one-way ANOVA (F [1,448] = 16.56, $p \leq 0.005$), (F [1,448] = 24.96, $p \leq 0.005$) and (F [5,444] = 12.66, $p \leq 0.005$) [Table 1].
For biochemical parameters, a statistically significant difference was observed in the mean of nutritional status categories for urea $p \leq 0.005$, creatinine $$p \leq 0.007$$, phosphorous $$p \leq 0.001$$, eGFR $p \leq 0.005$, serum albumin $$p \leq 0.004$$, and alkaline phosphatase $$p \leq 0.017$$ [Table 2].
**Table 2**
| Variables | OverallMean ± SDN = 450 | Well-nourishedSGA-AN = 158 (35.1%) | Mild-ModerateSGA-BN = 140 (31.1%) | MalnourishedSGA-CN = 152 (33.8%) | P-value |
| --- | --- | --- | --- | --- | --- |
| Age (years) | 53.9 ± 14.2 | 48.2 ± 15.0 | 55.8 ± 13.2 | 58.0 ± 12.0 | 0.0 |
| Biochemical Parameters | | | | | |
| Sodium (mg/dL) | 139.3 ± 8.1 | 139.3 ± 11.3 | 139.6 ± 5.5 | 138.8 ± 5.9 | 0.717 |
| Potassium (mEq/L) | 5.0 ± 4.8 | 4.6 ± 0.8 | 4.8 ± 0.6 | 5.5 ± 8.3 | 0.291 |
| Chloride | 102.8 ± 8.2 | 103.4 ± 8.8 | 102.9 ± 8.0 | 102.0 ± 8.2 | 0.318 |
| Urea (mg/dL) | 89.2 ± 68.5 | 71.8 ± 50.0 | 89.1 ± 86.2 | 107.5 ± 62.2 | 0.0 |
| Creatinine (mg/dL) | 3.1 ± 3.0 | 2.6 ± 4.2 | 2.9 ± 1.8 | 3.7 ± 3.0 | 0.007 |
| Calcium (mg/dL) | 9.0 ± 1.3 | 9.1 ± 1.1 | 9.0 ± 1.4 | 8.8 ± 1.3 | 0.158 |
| Phosphorus (mg/dL) | 4.6 ± 1.7 | 4.2 ± 1.5 | 4.7 ± 1.6 | 5.0 ± 1.8 | 0.001 |
| Uric Acid (mg/dL) | 7.6 ± 2.1 | 7.3 ± 1.9 | 7.5 ± 2.3 | 7.9 ± 2.1 | 0.083 |
| eGFR (ml/min/1.73m2) | 29.7 ± 19.0 | 36.3 ± 19.2 | 31.3 ± 21.5 | 21.2 ± 12.2 | 0.0 |
| Total protein (g/dL) | 7.1 ± 0.86 | 7.2 ± 0.87 | 7.1 ± 0.84 | 7.0 ± 0.86 | 0.102 |
| S Albumin (g/dL) | 4.0 ± 0.69 | 4.1 ± 0.67 | 4.0 ± 0.73 | 3.8 ± 0.65 | 0.004 |
| ALP (IU/L) | 114.0 ± 49.9 | 105.0 ± 40.9 | 118.3 ± 49.9 | 119.6 ± 56.8 | 0.017 |
| Hemoglobin (g/L) | 10.6 ± 8.8 | 11.3 ± 2.3 | 10.1 ± 2.3 | 10.6 ± 5.2 | 0.079 |
| MCV (fL) | 85.9 ± 8.8 | 85.9 ±6.9 | 86.0 ± 11.1 | 85.9 ± 8.8 | 0.925 |
| MCH (pg) | 27.5 ± 3.4 | 27.6 ± 2.18 | 27.2 ±2.23 | 27.5 ± 3.43 | 0.196 |
| TLC /L | 13.4 ± 42.6 | 15.2 ± 64.0 | 14.2 ± 34.4 | 10.9 ± 6.7 | 0.656 |
| TC (g/L) | 188.9 ± 48.7 | 192.8 ± 48.7 | 190.3 ± 58.1 | 183.5 ± 38.0 | 0.223 |
| TG (g/L) | 169.0 ± 102.6 | 169.3 ± 63.1 | 164.0 ±66.4 | 173.4 ± 151.8 | 0.738 |
| HDL-c (g/L) | 48.2 ± 12.7 | 48.0 ± 13.5 | 49.2 ± 13.2 | 47.5 ± 11.4 | 0.523 |
| LDL-c (g/L) | 103.0 ± 38.3 | 103.9 ± 34.9 | 106.2 ± 46.9 | 99.2 ± 32.3 | 0.28 |
| VLDL (g/L) | 34.1 ± 14.5 | 34.3 ± 11.5 | 33.4 ± 14.5 | 34.4 ± 17.2 | 0.818 |
| Anthropometry | | | | | |
| Height (cm) | 163.7 ± 8.8 | 164.3 ± 8.92 | 163.4 ± 8.5 | 163.3 ± 8.97 | 0.527 |
| Weight (kg) | 64.4 ± 13.3 | 65.5 ± 13.2 | 64.4 ± 11.8 | 63.3 ± 14.5 | 0.35 |
| BMI (kg/m2) | 23.9 ± 4.2 | 24.1 ± 4.05 | 24.0 ± 4.01 | 23.6 ± 4.6 | 0.524 |
| SBP (mmHg) | 149.8 ± 24.5 | 148.0 ± 24.2 | 150.3 ± 24.7 | 149.8 ±24.5 | 0.48 |
| DBP (mmHg) | 88.0 ± 13.7 | 89.0 ± 14.5 | 87.4 ± 12.8 | 87.7 ± 13.6 | 0.556 |
## Univariate spearman’s Rho correlations for diabetes and non-diabetes CKD groups
The univariate spearman correlation provided a significant input on correlation of biochemical parameters based on the diabetic status of these patients. The results showed that with decrease in eGFR, serum albumin, and hemoglobin, the SGA score tends to increase continuously, while with increase in urea, serum creatinine and serum phosphorus the SGA score also increases in both the groups [Table 3].
**Table 3**
| Variables | CKD with diabetes(N = 210) | CKD without diabetes(N = 240) |
| --- | --- | --- |
| Variables | Rho correlations with SGA>1 | Rho correlations with SGA>1 |
| Age (years) | 0.44 | 0.41** |
| Gender (male/female) | 0.07 | -0.05 |
| BMI (kg/m2) | -0.07 | -0.10 |
| Sodium (mg/dL) | -0.14* | -0.15* |
| Potassium (mEq/L) | -0.03 | 0.09 |
| Chloride | -0.20** | 0.03 |
| Urea (mg/dL) | 0.21** | 0.34** |
| Creatinine (mg/dL) | 0.28** | 0.28** |
| Calcium (mg/dL) | -0.16* | -0.10 |
| Phosphorus (mg/dL) | 0.24** | 0.15* |
| Uric Acid (mg/dL) | 0.0 | 0.13 |
| eGFR (ml/min/1.73m2) | -0.27** | -0.39** |
| Albumin (g/dL) | -0.18** | -0.14* |
| Hemoglobin (g/L) | -0.21** | -0.30** |
| MCV (fL) | -0.04 | 0.05 |
| MCH (pg) | -0.20** | 0.0 |
| TC (g/L) | 0.05 | -0.08 |
| TG (g/L) | -0.03 | -0.31 |
| HDL-c (g/L) | 0.02 | 0.02 |
| LDL-c (g/L) | 0.04 | -0.08 |
| VLDL (g/L) | -0.01 | -0.07 |
Additionally, in CKD patients with diabetes, SGA score was correlated with chloride, calcium, and mean corpuscular hemoglobin, but in CKD patients without diabetes, SGA score was correlated with age. The strongest correlations were found for age (years) in the non-diabetic CKD group (rho = 0.41; p0.01) and for creatinine in the CKD patients with diabetes (rho = 0.28; p0.01).
## Association of different covariates with Pt-Global web tool/PG-SGA score
A multivariable linear regression was conducted to predict Pt-Global web tool/PG-SGA score from biochemical & demographic parameters like eGFR, phosphorous, glycated hemoglobin (Hb1Ac), serum albumin and age. These variables have statistically significantly predicted the SGA score (F [17, 432] = 8.98, R2 = 0.261, $p \leq 0.05$). However, the three variables phosphorous (β = 1.728; $95\%$ CI, 0.700 to 0.678; $$P \leq 0.039$$), glycated hemoglobin (HB1Ac) (β = 0.186; $95\%$ CI, 0.050 to 0.322; $$P \leq 0.007$$); and age (β =.115; $95\%$ CI, 0.076 to 0.154; $$P \leq 0.000$$) added statistically significantly to the prediction, ($p \leq .05$) positively. A positive coefficient designates that the mean of the dependent variable is directly proportional to the value of an independent variable. Furthermore, the variables like eGFR (β = − 0.060; $95\%$ CI, − 0.093 to − 0.027; $$P \leq 0.000$$) and serum albumin (β = − 0.060; $95\%$ CI, − 2.068 to − 0.457; $$P \leq 0.002$$) were negatively associated with the SGA score. Multicollinearity between the independent variables was not observed.
The multiple correlation coefficient ($R = 0.551$) showed that the dependent variable (SGA score) may be predicted with reasonable accuracy. According to the coefficient of determination (R2 value = 0.261), $26.1\%$ of variance in the dependent variable (SGA score) is explained by these independent variables (biochemical parameters).
A multinomial logistic regression analysis reported that age group (β = 0.959; $95\%$ CI, 1.584 to 4.293; $$P \leq 0.000$$), and diabetes status (β = 1.177; $95\%$ CI, 1.897 to 5.549; $$P \leq 0.000$$) were statistically significantly predicting the nutritional status. The nagelkerke pseudo R2 = 0.110 represents that $11.0\%$ of variance in the nutritional status is explained by the model. Adult age group was found to be more affected with malnutrition, however the box-plot showed that with increase in age, Pt-Global web too/PG-SGA score also tends to increase. The Fig 1 represents the relationship between age and Pt-Global web tool/PG-SGA score.
**Fig 1:** *Relationship between age and nutritional status.*
The regression analysis was also extended to CKD patients based on diabetic status to predict SGA score from biochemical parameters. In non-diabetes CKD patients, the variables (urea $$p \leq 0.029$$, hemoglobin $$p \leq 0.002$$ and age p≤0.005) are significantly predicting the dependent variable (SGA score) while in case of CKD patients with diabetes the variables (phosphorous $$p \leq 0.02$$, glycated hemoglobin (Hb1Ac) $$p \leq 0.02$$, and albumin $$p \leq 0.05$$) are statistically significantly predicting the dependent variable (SGA score).
## Determinants of malnutrition
The Pt-Global web tool/PG-SGA parameters (weight, food intake, symptom score, activities & function) also known as PG-SGA short form were evaluated to explain their contribution towards the poor nutritional status. The symptom score parameter was affected in $81.8\%$ patients. The activities and function parameters ($79.6\%$) have also largely contributed to malnutrition followed by changes in food intake ($69.6\%$) and weight change ($51.6\%$). The Fig 2 represents the frequency of individual Pt-Global web tool/PG-SGA parameters towards malnutrition and/or SGA score.
**Fig 2:** *Frequency of individual Pt-Global web tool/PG-SGA parameters.*
## Symptom burden (symptom score)
This parameter was further evaluated for the problems that have kept the patients away from eating enough prior two weeks of nutritional status evaluation. The results demonstrated that the problems like fatigue ($81.5\%$), loss of appetite ($54.8\%$), body pain ($45.3\%$), constipation ($31.78\%$), dry mouth ($26.2\%$), and feel full quickly ($25.8\%$) were the major contributing factors for higher symptom score in these patients [Table 4].
**Table 4**
| Variables | No of participantsn = 450(%) | Well-nourishedSGA-AN = 158 | MalnourishedSGA-CN = 292 | p-value |
| --- | --- | --- | --- | --- |
| PT-Global web tool Score (median) | 11 | 4 | 16 | ≤0.001 |
| Nutrition impact symptoms | | | | |
| No appetite, just did not feel like eating | 244(54.8) | 4(0.9) | 240(53.3) | ≤0.001 |
| Nausea | 26(5.8) | 0.0 | 26(5.8) | ≤0.001 |
| Constipation | 143(31.7) | 2(0.4) | 141(31.3) | ≤0.001 |
| Mouth sores | 1(0.2) | 1(0.2) | 0.0 | 0.174 |
| Things taste funny or have no taste | 92(20.4) | 1(0.2) | 91(20.2) | ≤0.001 |
| Problems swallowing | 11(2.4) | 0.0 | 11(2.4) | 0.047 |
| Pain | 204(45.3) | 4(0.9) | 200(44.4) | ≤0.001 |
| Vomiting | 34(7.6) | 0.0 | 34(7.6) | ≤0.001 |
| Diarrhea | 17(3.8) | 0.0 | 17(3.8) | 0.002 |
| Dry Mouth | 118(26.2) | 1(0.2) | 117(26.0) | ≤0.001 |
| Smells bother me | 2(0.4) | 0.0 | 2(0.4) | 0.297 |
| Feel full quickly | 116(25.8) | 4(0.9) | 112(24.9) | ≤0.001 |
| Fatigue | 367(81.5) | 90(20.0) | 277(61.5) | ≤0.001 |
| Other Problems (Depression, Money, dental Problems) | 19(4.2) | 0.0 | 19(4.2) | 0.001 |
A significant mean difference was observed between the nutritional status and symptom burden parameters [Table 4]. This involved the higher proportion of problems like fatigue ($61.5\%$, P≤0.001), loss of appetite ($53.3\%$, P≤0.001), body pain ($44.4\%$, P≤0.001), constipation ($31.3\%$, P≤0.001), dry mouth ($26.0\%$, P≤0.001), and feel full quickly ($24.9\%$, P≤0.001) in patients who are malnourished.
## Mobility functions
The “activities and function parameter” was further examined, and the findings revealed that ‘patients with not my normal self but able to be up and about with fairly normal activities’ ($33.11\%$), and ‘patients not feeling up to most things, but in bed or chair less than half a day’ ($28.44\%$) were the major risk factors contributing higher grades/score of decrease in activities and function parameter, which led to malnutrition.
## Predictive validity of Pt-Global web tool/PG-SGA
The findings of the ROC analysis showed that the AUC was 0.988 ($95\%$ CI: 0.976–1.000), demonstrating that the overall PG-SGA score was a reliable indicator of malnutrition. The Receiver Operating Characteristics (ROC) plot of the sensitivity and specificity of the Pt-Global web tool/PG-SGA score for predicting malnutrition is shown in Fig 3.
**Fig 3:** *Receiver Operating Characteristics (ROC) plot of the sensitivity and specificity of the Pt-Global web tool/PG-SGA score for predicting malnutrition.*
In this patient pool, a PG-SGA total cut off score of ≥9 as per the author of the tool [18] only yielded a sensitivity of $84.2\%$ and specificity of $99.4\%$. The sensitivity was increased to $99.3\%$ and specificity comes down to $44.3\%$ when the total PG-SGA score was lowered to ≥3. A sensitivity of $99.7\%$ and specificity of $19.6\%$ was observed when Pt-Global web tool/PG-SGA total score was lowered to ≥2.
As per the ROC analysis, the total score cut off ≥3 was considered the most appropriate score to indicate malnutrition and also invites critical need for intervention among non-dialysis CKD patients. The coordinates of the ROC analysis for the total Pt-Global web tool/PG-SGA score ($$n = 450$$) are summarized in Table 5.
**Table 5**
| Score | Sensitivity | 1—Specificity |
| --- | --- | --- |
| 0.0 | 1.0 | 1.0 |
| 1.5 | 1.0 | 0.987 |
| 2.5 | 0.997 | 0.804 |
| 3.5 | 0.993 | 0.557 |
| 4.5 | 0.99 | 0.266 |
| 5.5 | 0.99 | 0.095 |
| 6.5 | 0.986 | 0.038 |
| 7.5 | 0.962 | 0.013 |
| 8.5 | 0.918 | 0.006 |
| 9.5 | 0.842 | 0.006 |
| 10.5 | 0.777 | 0.006 |
| 11.5 | 0.705 | 0.006 |
| 12.5 | 0.651 | 0.006 |
| 13.5 | 0.596 | 0.006 |
| 14.5 | 0.551 | 0.006 |
| 15.5 | 0.517 | 0.006 |
| 16.5 | 0.473 | 0.006 |
| 17.5 | 0.332 | 0.006 |
| 18.5 | 0.182 | 0.0 |
| 19.5 | 0.099 | 0.0 |
| 20.5 | 0.027 | 0.0 |
| 21.5 | 0.007 | 0.0 |
| 22.5 | 0.003 | 0.0 |
| 24.0 | 0.0 | 0.0 |
## Discussion
This prospective study provides a comprehensive overview of the nutritional status (stratified by the diabetic status, CKD stages, and gender) and symptom burden among chronic kidney disease patients before they underwent on dialysis. To the best of the knowledge of researcher’s, this is the first time a digital tool “Pt-Global web tool/PG-SGA” has been employed into clinical practice for the evaluation of nutritional status among CKD patients. This study has also investigated the predictive validity of “Pt-Global web tool/PG-SGA” in this particular population. Given the benefits over a manual SGA instrument, this web tool has proven itself as the most preeminent interdisciplinary approach of determining nutritional status.
According to Pt-Global web tool/PG-SGA, our results indicated that 140($31.1\%$) patients were mild/moderately malnourished; this number was found to be smaller when compared with the results reported by Tan et al. in Chinese patients 2016, 131($44.8\%$) [33]. The study conducted by Espinosa et al. in Mexican patients reported 1996, 40($44.4\%$) of prevalence in the same population [34]. The findings of this investigation differed slightly from those of Australian patients 85($40.5\%$) reported by Chan et al. in 2014 [35].
Further, the results of the current study also showed that 152($33.8\%$) patients were severely malnourished, which was low as compared to the neighboring states Haryana, India reported by Aggarwal et al. 2018, 58($58.0\%$) [36], and Uttar Pradesh, India reported by Prakash et al. 2007, 131($64.5\%$) [37] despite having the same food habits, but the results were twice as high as the results reported by Jagadeswaran et al. 2019, 19($14.7\%$) Andhra Pradesh, India [38].
In this study, the overall prevalence of malnutrition (mild/moderate and severe) ($64.9\%$) was found to be higher in comparison to the pooled prevalence $44.2\%$ (range 20.7–$70.6\%$) of malnutrition among non-dialysis chronic kidney disease patients reported in the meta-analysis performed by Rashid et al. [ 13]. Inflammation, oxidative stress, carbonyl stress, hormonal imbalances, decreased nutrient absorption from an oedematose gut, increased protein loss during dialysis, particularly peritoneal dialysis, and metabolic acidosis are some of the multiple causes of malnutrition in CKD [39–42].
Our findings also provided insights on the nutritional status based on diabetes, CKD stages and gender. Diabetes mellitus is considered as the most common cause of end stage renal disease. The disproportionate increase in the prevalence of diabetic chronic kidney disease or patients with end stage renal disease is narrated as a real epidemic [43] with an appalling prognosis [44]. In addition, diabetic CKD patients with poor nutritional status are also associated with adverse renal outcomes. The current study also evaluated the nutritional status in chronic kidney disease patients based on the diabetic status of participants. The results demonstrate that the patients with diabetes were substantially affected with malnutrition {severely malnourished 91($20.2\%$)} than non-diabetes CKD group {severely malnourished 61 ($13.6\%$)}.
The study of mechanism behind this larger prevalence was outside the scope of the study. However, the existing literature points to the facts that it may be due to poor glycemic control [45], insulin deprivation (the anabolic effects of insulin on protein homeostasis appear to be impaired in patients with type 1 diabetes mellitus) [46], higher resting energy expenditure [47], and restrictive dietary advice [48].
The results of this investigation showed that patients with chronic renal disease exhibit early signs of malnutrition. The prevalence of malnutrition was found to be increased {$3.4\%$ (mild/moderate & severe) in CKD stage 1–2 to $16.2\%$ in CKD stage 3a-3b and $45.3\%$ in CKD stages 4–5} with the decline of renal residual function, supplementing the data from previous reports both in pediatric and adult chronic kidney disease patients [49–51] [Table 1]. This pattern was also similar in results reported by Anupama et al. [ 52] in chronic kidney disease patients. This decrease in renal function potentiates loss of appetite, reduced food intake due to dietary restriction placed on these patients coupled with inadequate nutritional status monitoring, vomiting, diarrhea, hormonal imbalance etc. [ 53], which contribute significantly to the poor nutritional status in the latter stages of chronic kidney disease. This implies that routine monitoring of nutritional status is imperative at the early stages of chronic kidney disease, as it is more difficult to treat malnutrition when it progresses to higher stages of chronic kidney disease.
Male patients 91 ($20.2\%$) were found to have a greater prevalence of malnutrition than female patients 61 ($13.6\%$). Although the specific cause is not well understood, the observation of an increase in muscle mass loss and protein depletion in male patients suffering from chronic kidney disease supports this. This might also be due to the fact that men are more likely to seek medical attention than women. Hormonal influences, however, could not be completely ruled out. Tayyem et al., Oluseyi et al., and Stenvinkel et al. also corroborated these findings [54–56].
Adult age group 79($17.6\%$) was found to be more severely malnourished as compared to elderly age group 73($16.2\%$). This pattern is surprising because aging is significantly associated with malnutrition in the elderly even without chronic kidney disease [15]. However, the box plot between age and nutritional status showed with increase in age, SGA score continues to increase thus poor nutritional status [Fig 2].
Among non-diabetes CKD patients, the multiple regression analysis showed that the biochemical parameters (urea $$p \leq 0.029$$, hemoglobin $$p \leq 0.002$$ and age p≤0.005) are significantly predicting the SGA score, while in diabetic CKD group the biochemical parameters (phosphorous $$p \leq 0.02$$, glycated hemoglobin (Hb1Ac) $$p \leq 0.02$$, and albumin $$p \leq 0.05$$) were found to be statistically significantly predicting the SGA score. Diabetic chronic kidney disease, end stage renal disease and hypoalbuminemia are well recognized risk factors for cardiovascular disease [57,58].
The evaluation of symptom burden and the predictive validity of the Pt-Global web tool/PG-SGA comprised the study’s main finding. This web application provided a clear grasp of the factors that make up the symptom score. The results of the ROC analysis revealed that the overall PG-SGA score was a credible indication of malnutrition, with an AUC of 0.988 ($95\%$ CI: 0.976–1.000).
The importance of understanding the symptom burden is amply illustrated in the current study, where malnourished patients account for a staggering $80.3\%$ of the symptom score while well-nourished patients make up $19.7\%$. Almutary et al [59] have also highlighted the significant symptom burden among dialysis patients. Notably the symptoms like fatigue, loss of appetite, pain anywhere in the body, constipation, dry mouth, feel full quickly had a strong impact on Pt-Global web tool/PG-SGA total score. This suggests that addressing malnutrition requires multifaceted approach that address both the social determinants of health to improve access to affordable food, as well as treatment of symptoms and underlying potentially medical issues that limit intake. A detailed understanding of symptom clusters may contribute to the quality of life and treatment priorities [60].
## Conclusions
According to this study’s findings, around six out of ten non-dialysis CKD patients experienced malnutrition. Furthermore, it has been demonstrated that individuals with diabetes, hypoalbuminemia, and decreased renal function experience greater malnutrition. The results also revealed that the optimal cutoff threshold for diagnosing malnutrition on the Pt-Global web tool/PG-SGA appears to be a total score of ≥3. The symptoms of fatigue, loss of appetite, body pain, constipation, dry mouth, and feeling full quickly substantially exacerbated the malnutrition. These findings will encourage the usage of the P Pt-Global web tool/PG-SGA for nutritional status evaluation. Early emphasis on nutrition in chronic kidney disease patients holds the key for better health outcomes which sequentially can retard the CKD progression and increase the longevity of these patients.
## References
1. Cederholm T, Barazzoni R, Austin P, Ballmer P, Biolo G, Bischoff SC. **ESPEN guidelines on definitions and terminology of clinical nutrition**. *Clin Nutr.* (2017.0) **36** 49-64. DOI: 10.1016/j.clnu.2016.09.004
2. Feroze U, Noori N, Kovesdy CP, Molnar MZ, Martin DJ, Reina-Patton A. **Quality-of-life and mortality in hemodialysis patients: roles of race and nutritional status**. *Clin J Am Soc Nephrol* (2011.0) **6** 1100-11. DOI: 10.2215/CJN.07690910
3. Macedo C, Amaral TF, Rodrigues J, Santin F, Avesani CM. **Malnutrition and Sarcopenia Combined Increases the Risk for Mortality in Older Adults on Hemodialysis.**. *Front Nutr.* (2021.0) **8** 721941. DOI: 10.3389/fnut.2021.721941
4. Kang SS, Chang JW, Park Y. **Nutritional Status Predicts 10-Year Mortality in Patients with End-Stage Renal Disease on Hemodialysis.**. *Nutrients* (2017.0) **9** 399. DOI: 10.3390/nu9040399
5. de Roij van Zuijdewijn CL, ter Wee PM, Chapdelaine I, Bots ML, Blankestijn PJ, van den Dorpel MA. **A Comparison of 8 Nutrition-Related Tests to Predict Mortality in Hemodialysis Patients**. *J Ren Nutr.* (2015.0) **25** 412-9. DOI: 10.1053/j.jrn.2015.02.005
6. **Global, regional, and national burden of chronic kidney disease, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017.**. *Lancet* (2020.0) **395** 709-733. DOI: 10.1016/S0140-6736(20)30045-3
7. 7The Institute for Health Metrics and Evaluation.
Global Burden of Disease (GBD2019)
2019 [20 Nov 2022]. Available online: https://www.healthdata.org/gbd/2019.. *Global Burden of Disease (GBD2019)* (2019.0)
8. Nugent RA, Fathima SF, Feigl AB, Chyung D. **The burden of chronic kidney disease on developing nations: a 21**. *Nephron Clin Pract* (2011.0) **118** c269-77. PMID: 21212690
9. Fraser SDS, Roderick PJ. **Kidney disease in the Global Burden of Disease Study 2017.**. *Nat Rev Nephrol.* (2019.0) **15** 193-194. DOI: 10.1038/s41581-019-0120-0
10. Lim SL, Ong KC, Chan YH, Loke WC, Ferguson M, Daniels L. **Malnutrition and its impact on cost of hospitalization, length of stay, readmission and 3-year mortality.**. *Clin Nutr.* (2012.0) **31** 345-350. DOI: 10.1016/j.clnu.2011.11.001
11. Sorensen J, Kondrup J, Prokopowicz J, Schiesser M, Krahenbuhl L, Meier R. **EuroOOPS: an international, multicentre study to implement nutritional risk screening and evaluate clinical outcome.**. *Clin Nutr* (2008.0) **27** 340-349. DOI: 10.1016/j.clnu.2008.03.012
12. Guest JF, Panca M, Baeyens JP, de Man F, Ljungqvist O, Pichard C. **Health economic impact of managing patients following a community-based diagnosis of malnutrition in the UK.**. *Clin Nutr* (2011.0) **30** 422-429. DOI: 10.1016/j.clnu.2011.02.002
13. Rashid I, Bashir A, Tiwari P, D’Cruz S, Jaswal S. **Estimates of malnutrition associated with chronic kidney disease patients globally and its contrast with India: An evidence based systematic review and meta-analysis.**. *Clin Epidemiol Global Health* (2021.0) **12** 100855
14. Leij-Halfwerk S, Verwijs MH, van Houdt S, Borkent JW, Guaitoli PR, Pelgrim T. **Prevalence of protein-energy malnutrition risk in European older adults in community, residential and hospital settings, according to 22 malnutrition screening tools validated for use in adults ≥65 years: A systematic review and meta-analysis.**. *Maturitas* (2019.0) **126** 80-89. PMID: 31239123
15. Rashid I, Tiwari P, Lehl SS. **Malnutrition among elderly a multifactorial condition to flourish: Evidence from a cross-sectional study.**. *Clin Epidemiol Global Health* (2020.0) **8** P91-95
16. Kruizenga HM, Wierdsma NJ, van Bokhorst MA, de van der Schueren HJ, Haollander CF. **Screening of nutritional status in The Netherlands.**. *Clin Nutr.* (2003.0) **22** 147-52. DOI: 10.1054/clnu.2002.0611
17. Buys DR, Roth DL, Ritchie CS, Sawyer P, Allman RM, Funkhouser EM. **Nutritional risk and body mass index predict hospitalization, nursing home admissions, and mortality in community-dwelling older adults: results from the UAB Study of Aging with 8.5 years of follow-up**. *J Gerontol A Biol Sci Med Sci* (2014.0) **69** 1146-53. DOI: 10.1093/gerona/glu024
18. Lodebo BT, Shah A, Kopple JD. **Is it important to prevent and treat protein-energy wasting in chronic kidney disease and chronic dialysis patients?**. *J Ren Nutr* (2018.0) **28** 369-79. DOI: 10.1053/j.jrn.2018.04.002
19. Sahay M, Sahay R, Baruah MP. **Nutrition in chronic kidney disease.**. *J Med Nutr Nutraceut* (2014.0) **3** 11-8
20. Fouque D, Kalantar-Zadeh K, Kopple J, Cano N, Chauveau P, Cuppari L. **A proposed nomenclature and diagnostic criteria for protein-energy wasting in acute and chronic kidney disease**. *Kidney Int* (2007.0) **73** 391-398. DOI: 10.1038/sj.ki.5002585
21. Steiber A, Leon JB, Secker D, McCarthy M, McCann L, Serra M. **Multicenter study of the validity and reliability of subjective global assessment in the hemodialysis population.**. *J Ren Nutr* (2007.0) **17** 336-342. DOI: 10.1053/j.jrn.2007.05.004
22. Campbell KL, Ash S, Bauer J, Davies PS. **Critical review of nutrition assessment tools to measure malnutrition in chronic kidney disease.**. *Nutr Diet* (2007.0) **64** 23-30
23. **K/DOQI Clinical practice guidelines for nutrition in chronic renal failure (I. Adult guidelines).**. *Am J Kidney Dis* (2000.0) **35** 17-104
24. Detsky AS, McLaughlin JR, Baker JP, Johnston N, Whittaker S, Mendelson RA. **What is subjective global assessment of nutritional status?**. *JPEN J Parenter Enteral Nutr* (1987.0) **11** 8-13. DOI: 10.1177/014860718701100108
25. Jager-Wittenaar H, Ottery FD. **Assessing nutritional status in cancer: role of the Patient-Generated Subjective Global Assessment.**. *Curr Opin Clin Nutr Metab Care* (2017.0) **20** 322-329. DOI: 10.1097/MCO.0000000000000389
26. Stenvinkel P, Heimburger O, Paultre F, Diczfalusy U, Wang T, Berglund L. **Strong association between malnutrition, inflammation, and atherosclerosis in chronic renal failure**. *Kidney Int* (1999.0) **55** 1899-911. DOI: 10.1046/j.1523-1755.1999.00422.x
27. Steiber A, Leon JB, Secker D, McCarthy M, McCann L, Serra M. **Multicenter study of the validity and reliability of subjective global assessment in the hemodialysis population.**. *J Ren Nutr* (2007.0) **17** 336-42. DOI: 10.1053/j.jrn.2007.05.004
28. **Adequacy of dialysis and nutrition in continuous peritoneal dialysis: association with clinical outcomes.**. *J Am Soc Nephrol* (1996.0) **7** 198-207. PMID: 8785388
29. Li H, Xie L, Yang J, Pang X. **Symptom burden amongst patients suffering from end-stage renal disease and receiving dialysis: A literature review.**. *Int J Nurs Sci.* (2018.0) **5** 427-431. DOI: 10.1016/j.ijnss.2018.09.010
30. Lamb EJ, Levey AS, Stevens PE. **The Kidney Disease Improving Global Outcomes (KDIGO) guideline update for chronic kidney disease: evolution not revolution.**. *Clin Chem* (2013.0) **59** 462-5. DOI: 10.1373/clinchem.2012.184259
31. Šimundić AM. **Measures of Diagnostic Accuracy: Basic Definitions.**. *EJIFCC* (2009.0) **19** 203-11. PMID: 27683318
32. Ikizler TA, Burrowes JD, Byham-Gray LD, Campbell KL, Carrero JJ, Chan W. **KDOQI Clinical Practice Guideline for Nutrition in CKD: 2020 Update.**. *Am J Kidney Dis* (2020.0) **76** S1-S107. DOI: 10.1053/j.ajkd.2020.05.006
33. Tan R, Long J, Fang S, Mai H, Lu W, Liu Y. **Nutritional Risk Screening in patients with chronic kidney disease.**. *Asia Pac J Clin Nutr* (2016.0) **25** 249-56. DOI: 10.6133/apjcn.2016.25.2.24
34. Espinosa A, Cueto-Manzano AM, Velazquez-Alva C, Hernandez A, Cruz N, Zamora B. **Prevalence of malnutrition in Mexican CAPD diabetic and nondiabetic patients.**. *Adv Perit Dial* (1996.0) **12** 302-6. PMID: 8865924
35. Chan M, Kelly J, Batterham M, Tapsell L. **A high prevalence of abnormal nutrition parameters found in predialysis end-stage kidney disease: is it a result of uremia or poor eating habits?**. *J Ren Nutr.* (2014.0) **24** 292-302. DOI: 10.1053/j.jrn.2014.03.008
36. Aggarwal HK, Jain D, Chauda R, Bhatia S, Sehgal R. **Assessment of Malnutrition Inflammation Score in Different Stages of Chronic Kidney Disease.**. *Pril (Makedon Akad Nauk Umet Odd Med Nauki).* (2018.0) **39** 51-61. DOI: 10.2478/prilozi-2018-0042
37. Prakash J, Raja R, Mishra RN, Vohra R, Sharma N, Wani IA. **High prevalence of malnutrition and inflammation in undialyzed patients with chronic renal failure in developing countries: a single center experience from eastern India**. *Ren Fail* (2007.0) **29** 811-6. DOI: 10.1080/08860220701573491
38. Jagadeswaran D, Indhumathi E, Hemamalini AJ, Sivakumar V, Soundararajan P, Jayakumar M. **Inflammation and nutritional status assessment by malnutrition inflammation score and its outcome in pre-dialysis chronic kidney disease patients**. *Clin Nutr* (2019.0) **38** 341-347. DOI: 10.1016/j.clnu.2018.01.001
39. Kalantar-Zadeh K, Ikizler TA, Block G, Avram MM, Kopple JD. **Malnutrition inflammation complex syndrome in dialysis patients: causes and consequences**. *Am J Kidney Dis* (2003.0) **42** 864-881. DOI: 10.1016/j.ajkd.2003.07.016
40. Ballmer PE, McNurlan MA, Hulter HN, Anderson SE, Garlick PJ, Krapf R. **Chronic metabolic acidosis decreases albumin synthesis and induces negative nitrogen balance in humans**. *J Clin Invest* (1995.0) **95** 39-45. DOI: 10.1172/JCI117668
41. Lecker SH, Goldberg AL, Mitch WE. **Protein degradation by the ubiquitinproteasome pathway in normal and disease states**. *J Am Soc Nephrol* (2006.0) **17** 1807-1819. DOI: 10.1681/ASN.2006010083
42. Cueto Manzano AM. **Hypoalbuminemia in dialysis. Is it a marker for malnutrition or inflammation?**. *Rev Invest Clin.* **53** 152-158. PMID: 11421111
43. Jones CA, Krolewski AS, Rogus J, Xue JL, Collins A, Warram JH. **Epidemic of end-stage renal disease in people with diabetes in the United States population: do we know the cause?**. *Kidney Int* (2005.0) **67** 1684-91. DOI: 10.1111/j.1523-1755.2005.00265.x
44. Chantrel F, Enache I, Bouiller M, Kolb I, Kunz K, Petitjean P. **Abysmal prognosis of patients with type 2 diabetes entering dialysis**. *Nephrol Dial Transplant* (1999.0) **14** 129-36. DOI: 10.1093/ndt/14.1.129
45. Morioka T, Emoto M, Tabata T, Shoji T, Tahara H, Kishimoto H. **Glycemic control is a predictor of survival for diabetic patients on hemodialysis**. *Diabetes Care* (2001.0) **24** 909-13. DOI: 10.2337/diacare.24.5.909
46. Charlton M, Nair KS. **Protein metabolism in insulin-dependent diabetes mellitus**. *J Nutr* (1998.0) **128** 323S-327S. DOI: 10.1093/jn/128.2.323S
47. Avesani CM, Cuppari L, Silva AC, Sigulem DM, Cendoroglo M, Sesso R. **Resting energy expenditure in pre-dialysis diabetic patients.**. *Nephrol Dial Transplant.* (2001.0) **16** 556-65. DOI: 10.1093/ndt/16.3.556
48. Gretz N, Lasserre J, Strauch M. **Caloric supplements for patients on low-protein diets?**. *Nephron* (1988.0) **50** 129-32. DOI: 10.1159/000185142
49. Oluseyi A, Enajite O. **Malnutrition in pre-dialysis chronic kidney disease patients in a teaching hospital in Southern Nigeria.**. *Afr Health Sci.* (2016.0) **16** 234-41. DOI: 10.4314/ahs.v16i1.31
50. Olowu WA, Adefehintin O, Aladekomo TA. **Epidemiology and clinicopathologic outcome of paediatric chronic kidney disease in Nigeria; a single centre experience.**. *AJNT* (2013.0) **6** 105-113. PMID: 23656404
51. Kopple JD, Greene T, Chumlea WC, Hollinger D, Maroni BJ, Merrill D. **Relationship between nutritional status and the glomerular filtration rate: result from the MDRD study**. *Kidney Int* (2000.0) **57** 1688-1703. PMID: 10760105
52. Anupama SH, Abraham G, Alex M, Vijayan M, Subramanian KK, Fernando E. **A multicenter study of malnutrition status in chronic kidney disease stages I–V-D from different socioeconomic groups.**. *Saudi J Kidney Dis Transpl* (2020.0) **31** 614-23. DOI: 10.4103/1319-2442.289448
53. Yashpal PJ, Vijah K. **Protein energy wasting in chronic kidney disease: An update with focus on nutritional interventions to improve outcomes**. *Indian J Endocrinol Metab* (2012.0) **16** 246-251. DOI: 10.4103/2230-8210.93743
54. Tayyem RF, Mrayyan MT. **Malnutrition and anthropometric and biochemical abnormalities in end stage renal disease patients.**. *Saudi Med J* (2007.0) **28** 1575-1581. PMID: 17914522
55. Oluseyi A, Enajite O. **Malnutrition in pre-dialysis chronic kidney disease patients in a teaching hospital in Southern Nigeria.**. *Afr Health Sci.* (2016.0) **16** 234-41. DOI: 10.4314/ahs.v16i1.31
56. Stenvinkel P, Heimbürger O, Lindholm B, Kaysen GA, Bergström J. **Are there two types of malnutrition in chronic renal failure? Evidence for relationships between malnutrition, inflammation and atherosclerosis (MIA syndrome).**. *Nephrol Dial Transplant.* (2000.0) **15** 953-60. DOI: 10.1093/ndt/15.7.953
57. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. **Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization**. *N Engl J Med* (2004.0) **351** 1296-1305. DOI: 10.1056/NEJMoa041031
58. Perkovic V, Verdon C, Ninomiya T, Barzi F, Cass A, Patel A. **The relationship between proteinuria and coronary risk: A systematic review and meta-analysis.**. *PLoS Med* (2008.0) **5** e207. DOI: 10.1371/journal.pmed.0050207
59. Almutary H, Bonner A, Douglas C. **Symptom burden in chronic kidney disease: a review of recent literature.**. *J Ren Care* (2013.0) **39** 140-150. DOI: 10.1111/j.1755-6686.2013.12022.x
60. Kalantar-Zadeh K, Lockwood MB, Rhee CM, Tantisattamo E, Andreoli S, Balducci A. **Patient-centred approaches for the management of unpleasant symptoms in kidney disease.**. *Nat Rev Nephrol.* (2022.0) **18** 185-198. DOI: 10.1038/s41581-021-00518-z
|
---
title: Enhancing the knowledge of parents on child health using eLearning in a government
school in the semi-rural community of Karachi, Pakistan
authors:
- Saleema Gulzar
- Sana Saeed
- Salimah Taufiq Kirmani
- Rozina Karmaliani
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022312
doi: 10.1371/journal.pgph.0000500
license: CC BY 4.0
---
# Enhancing the knowledge of parents on child health using eLearning in a government school in the semi-rural community of Karachi, Pakistan
## Abstract
Education is one of the vital social determinants of health. Health and education share a symbiotic relationship for all cadre including children and adolescents to ensure that they are well equipped to combat the health risk in the environment. The current literature globally found some initiatives to create health awareness among school children. However, there is a dearth of studies available addressing parental health awareness through school platforms. Therefore, the current study aims to fill this gap, and the Aga Khan University School of Nursing and Midwifery initiated the School Health Program (SHP) in one of the remote communities in Sindh, Pakistan. The overall goal of the study was to improve children’s health by enhancing the health awareness of the parents through school platforms utilizing online modalities. Another objective of this study was to identify the effect of using eLearning on parental knowledge and perceptions. The study utilized a sequential explanatory mixed-method design. Twelve health awareness sessions relevant to children’s health using eLearning were conducted over one year. Parents’ knowledge was assessed through a pre-posttest, which was administered after each teaching session. Subsequently, focused group discussions were carried out with parents, community leaders, and schoolteachers to gain insights regarding the effectiveness of the health education program. The pre-and post-test results showed again in knowledge in nine out of twelve sessions. The findings from qualitative content analysis yielded three key themes: Perceived usefulness of eLearning, Barriers affecting usability, and Way forward for eLearning through school platforms. The study showed parental satisfaction with the online health education awareness program. They exhibited enthusiasm and desire for further similar sessions in the future. The results demonstrated an enhancement in parental awareness about common health conditions among school children. This study may be replicated on a larger scale in the schools of Pakistan.
## Introduction
Health education is one of the most critical components of universal health coverage [1]. Studies have shown that parents who are better educated about health care are better equipped to cater to the developmental needs of their children [2]. Students whose parents are actively involved with their health are reported to achieve better academic outcomes [3]. Further, the children are likely to be more participative, attentive, and motivated to perform well in their studies. In addition, a better connection between home and school means that more children are likely to complete their education on time, resulting in lower drop-out rates [4]. On the other hand, parents who are not well informed about the health and developmental needs of their children, are also less likely to be involved in their child’s growth and development [5]. Therefore, creating health awareness among parents regarding disease prevention and health promotion of school-aged children would lead to better health outcomes [6, 7]. The available literature has depicted the impact of health education among children and adolescents globally, however, creating health awareness among parents of school children has not been paid much attention in Pakistan. Therefore, the current study was carried out to create awareness among parents of school children because parents influence more on a child’s behaviors [8].
Pakistan being a low-resourced country faces challenges in imparting health education among parents due to a lack of educational reforms which include parents at the core. In several peri-urban and rural areas of Pakistan, the child mortality rate is higher than in many Low and Middle-Income Countries (LMICs). According to Masquelier et al. [ 2018], the mortality rate in Asia and Africa for the age group 5 to 14 is the highest in the region [9]. Similar results are reported in which multiple reasons for child morbidity and mortality were reported, with lack of awareness amongst parents being amongst the most important factors highlighted [10, 11].
School provides a powerful platform to make health accessible for school children. It deals with the curative, preventive, and promotive aspects of health through school settings with the purpose to maintain the health and well-being of children by detecting and treating common illnesses before it gets too complicated. Gulzar et al. [ 2017], is a success story of a school health program in the local context [12]. The school health curriculum was developed and implemented for students in a higher secondary school in urban Karachi. The study identified a vital gap that has occurred due to the lack of including parents in such programs. There is a paucity of literature regarding initiatives that are directed to improve healthcare awareness among parents and technology has rarely been considered as a possible solution to address this gap, which is the focus of the current study.
One such project was started in a semi-rural community in Sindh. This initiative aimed to facilitate the provision of basic health services to the school children and increase the awareness of the caregivers or parents of school children with the basic health knowledge related to childhood illnesses and care to address the gap in the literature in our local context.
In most cases, parents are responsible for children’s primary years of education [13]. Several studies have proven internationally that effective primary care and healthy communication with parents result in improved attendance and engagement at school [14]. Therefore, this project launched a health education program for parents using an eLearning approach in 2017. The infrastructure for the online health education program was developed in the selected school setting. The health education sessions were carried out by content experts relevant to child’s health. The online session was conducted once every month to educate parents on common health conditions of children.
This study aims to evaluate the effect of eLearning-based school health education programs provided in the school in a semi-rural community setting to enhance parental awareness to promote the health of school children.
## Materials and methods
The study was conducted in one of the government schools (grade Early Child Development (ECD)-grade VIII) of the semi-rural community in Sindh, Pakistan, in the year 2016–2019. This study utilized the mixed-method sequential design. To sensitize and mobilize the community, the team of healthcare providers worked alongside the community leaders and school administration in the community. To get the baseline assessment, initially, a formal meeting was carried out with parents and teachers of grades ECD-, i.e., primary, and secondary grades, to identify the common health problems of children that parents found challenging to deal with. The consensus of the parents regarding the time and venue of the health education sessions was sought through an in-person meeting with parents during the need assessment meeting. The total number of students in the school was 253 (ECD-grade VIII). For the current study, parents were invited and asked to voluntary participation in a health awareness program, generally, each parent has three to six children studying at the same school. All the parents who showed up on the day of the session were included through the universal sampling method. Based on this baseline data, the healthcare professionals (nurses, physicians, and nutritionists.) who had the content expertise were approached at one of the private university hospitals in Karachi to facilitate the live session for the parents. The sessions were scheduled once a month on the parents’ preferred day and time. The parents were informed and sent reminders prior to the sessions, and they gathered in the school to attend virtual sessions using MDConsults software. The software helped us cater to the issue of accessibility by parents and travel issues of the session facilitators. The technological solution has assisted the target community in two ways. Firstly, it has helped to mitigate the issue of availability and accessibility of internet and electronic devices individual homes to join in the virtual learning program. Secondly, it has provided the feasibility to the community to attend the sessions virtually in their community that has saved their travel cost and time. Each session was interactive and had pictorial representations for better understanding. All parents who consented to the study were included in the study.
## Ethics approval and consent to participate
This study is approved by The Aga Khan University Ethical Review Committee (4410-SON-ERC-16). All individuals were informed of the ethical issues and were given the opportunity to withdraw from the study at any time. Written informed consent was obtained from all participating individuals.
## Quantitative data collection and analysis
In the quantitative arm, the knowledge of the parents was assessed through pre and post-tests. A specific questionnaire was developed for each health education session, in the local language (Urdu) such as immunization, and the importance of a balanced diet for children. The multiple-choice questions were developed with the help of a content expert who facilitated the sessions. Each questionnaire comprised of 8–10 items with four options. The caregivers/parents had to make the appropriate choice on each question S1 Text. The selected school has one onboard female school health nurse, responsible for administering the pre-posttest questionnaire. In cases of queries, she addressed the participants’ concerns. The data was entered and analyzed using SPSS 20.0 software. Paired t-test was applied to measure the effect of the intervention on the pre-post scores. P-value of <0.05 was taken as significant.
## Qualitative data collection and analysis
In the qualitative arm of this study, the data were gathered through our four focus group discussions (FGDs). These comprise two FGDs with parents and one each with community leaders and teachers separately, were conducted by the research team members, who possessed the rich experience and relevant qualifications for conducting FGDs. The FGDs were conducted and audio-recorded after completing 12 online sessions to explore in-depth perspectives about these sessions. The parents, teachers, and community leaders were invited to share their experiences related to the online health education program. The data was transcribed and translated from Urdu to English. The data analysis was carried out by the research team members who were master’s and PhD prepared nurses with a research background in both quantitative and qualitative research approaches. The qualitative data analysis was carried out independently by two expert researchers, adding trustworthiness to the findings. Similar codes were combined to generate themes, within which subthemes were extracted.
## Quantitative pre-post assessment results
The quantitative arm depicted that the mean participation of parents in the eLearning sessions was found to be 21.8 (max: 34 & min: 13: SD 7.2). There was a significant difference in parents’ knowledge after eight sessions ($$n = 12$$, $66\%$) as measured through a paired t-test. However, no statistical difference was found in pre-post knowledge assessment on the topics of child safety, nutrition, and substance abuse as shown in Table 1. The set of questions from each session was developed by the content experts and reviewed by the team for contextual reference. The translation of the tool into the local language by a team member who possessed bilingual language skills. The questionnaire was multiple-choice, and each participant was given one mark for the correct answer. In the sessions, the number of participants ranged from 15–34 per session. In 8 out of 12 sessions including vaccination, diarrhea, puberty, pneumonia, seizures, personal hygiene, measles, and cognitive development depicted a significant increase in the knowledge of participants.
**Table 1**
| E-learning Modules | Number of Participants | Mean Difference (SD) | P-value |
| --- | --- | --- | --- |
| Child Safety | 25 | -0.04(1.20) | 0.870 |
| Vaccination | 22 | 1.59(1.18) | <0.001* |
| Diarrhea | 18 | 0.38(0.77) | 0.049* |
| Puberty | 34 | 0.58(1.30) | 0.013* |
| Nutrition | 30 | 0.13(1.65) | 0.662 |
| Pneumonia | 15 | 0.40(0.63) | 0.028* |
| Seizures | 28 | 1.17(1.27) | <0.001* |
| Personal Hygiene | 13 | 1.38(1.38) | 0.004* |
| Substance Abuse | 15 | 0.60(1.91) | 0.246 |
| Measles | 15 | 2.60(1.35) | <0.001* |
| Parenting | 17 | -0.76(1.20) | 0.018* |
| Cognitive Development | 29 | 0.89(1.51) | 0.004* |
## Qualitative data results
Content analysis of the responses revealed three key themes that are depicted in Table 2.
**Table 2**
| Unnamed: 0 | Themes | Subthemes |
| --- | --- | --- |
| 1- | Perceived usefulness of eLearning in the community | a) Awareness of common illnesses and home-based management |
| 1- | Perceived usefulness of eLearning in the community | b) Dual service under one roof |
| 2- | Barriers affecting the usability of eLearning in the rural community | a) Cultural issues and community mindset technology-based issues |
| 3- | Way forward for e-learning for distant communities | a) Personal growth |
| 3- | Way forward for e-learning for distant communities | b) Societal awareness |
## Perceived usefulness of eLearning sessions/program
The FGD with the students’ mothers revealed the usefulness of these sessions in improving their knowledge about common childhood and adolescent health conditions. They perceived that this was by far the only opportunity available for them where they could interact with a healthcare expert facilitating the session remotely and addressing their queries. They shared their satisfaction and gratitude towards the initiative as it not only made them aware of their children’s health and well-being but also helped them develop the confidence to manage minor and common health conditions at home.
## Awareness of common illnesses and home-based management
Caregivers were informed about the scarcity of healthcare services in their community and shared that they travel two to three hours to seek a health facility. It becomes close to impossible for them to consult those health services, especially for common illnesses such as diarrhea. They perceived that these sessions empowered them not only with the knowledge but also equipped them with certain skills that can be taken at home during illness.
One of the participants shared, “The doctor/nurse taught us what causes diarrhea, vomiting and how to make ORS (oral rehydration salt), and I used this information at home whenever needed." Another mother shared, "We didn’t know what to teach our daughters about the menstrual cycle. We learned that we don’t necessarily require medication during the menstrual cycle, but maintaining hygiene is the most important action. Secondly, if someone is bleeding less, this could be due to blood deficiency." Overall, the participants appreciated this initiative and shared that, generally, these sessions helped them understand the primary causes of the most common health conditions and how to prevent and manage them at early stages.
The mothers also shared that the eLearning sessions helped them and their children to take care of their health and well-being. The transformative experience of their learning about the health outcome of their family has reinforced their trust in this system of learning.
Another mother appreciated by sharing the experience of attending one of the sessions on a balanced diet by saying, "*Before this* health awareness program, we could not distinguish between right and wrong concerning our children’s upbringing, but now our minds have opened up to various ways to maintain our health, what we should eat now, and why combinations of diet are important in our diet, and this information will always stay in our mind." The participants also shared their experiences about safety measures to be taken for their children. They found these sessions very helpful in taking some preventive measures for the child’s safety.
## Dual services under one roof
School is one such place whereby the children not only learn new ideas and gain new knowledge, but also these learning spaces helped them to understand the principles of healthy living, their rights, and duties as responsible citizens. The project utilized the school space as a platform to gather the caregivers/ parents to help them understand an essential aspect of their children’s health. This added an emotional aspect to this intervention and broaden their minds to reflect upon the importance of the connection between health and education. The parents shared their satisfaction with this.
One of the parents shared, “it gives us immense satisfaction by seeing our children learning in an environment which cares about them and their needs” Another parent reported, “it is very difficult for us to commute due to the non-availability of transport services in our area. Had this intervention taken place at some other space, it would have been close to impossible for us to reach there.”
## Barriers affecting the usability
Despite poor literacy rates, the caregivers understood and valued this intervention. They expressed their desire of having a longing experience whereby they could learn and improve the health and hygiene status of their families. In the next phase of FGD, participants shared their opinions regarding areas which are needed to be improved for the future in such initiatives.
## Cultural issues and community mindset
The participants drew attention to a vital point about maintaining confidentiality related to certain sensitive topics. They informed that discussing those aspects is considered Taboo in their culture. For example, puberty in children is one of the areas which is considered private in their community. They suggested having discussions individually rather than openly in public on those aspects. They shared that discussing those issues in public might have some cultural implications as discussing those matters openly is not considered decent in their community.
One of the schoolteachers informed, "A session in which the doctor discussed personal hygiene during menstruation, though it was very informative for the mothers, the parents shared their concerns of privacy."
## Technology-based challenges
Technology plays a vital role in connecting remote areas. The experience of electronic learning is directly proportional to the unhampered connectivity. To make sure that there is no break in the process, the project team arranged all the possible devices and provided the optimum setup. However, sometimes, during the session, there were challenges related to network connectivity during the sessions.
One of the parents shared, “the whole discussion was going on so well, suddenly the connection dropped, and the doctor disappeared from our screen. Upon rejoining it took a bit of time for me to recall the previous point”.
## Way forward for learning to spread health awareness in rural communities
The participants were found to be enthusiastic and proposed ideas for subsequent sessions. Based on their experience, they identified the need of including various other areas in future activities.
## Societal awareness
Being a close community, the individuals utilized some of their time to discuss their challenges, the politics, school performances of their children, etc. The community leaders informed that they had often seen parents discussing those aspects, which are taught to them, among themselves, and hence they felt that this raised the community’s awareness and brought about a positive change in knowledge, attitude, and practices.
A participant shared: "There needs to be a session on family planning as we do not have enough knowledge regarding birth spacing."
## Personal growth
During the project, it was observed that all stakeholders were very excited and contributed their bit to implement and execute it. One of the most fascinating observations was that they were not only curious to know about physical health, but they also showed interest in getting awareness about mental well-being. During the discussion, it was evident that the community stakeholders wished to learn about basic lifesaving skills, first aid practices, etc.
Other participants shared: "There should be training about the parent-child relationship and first aid in case of injury."
## Discussion
This study aimed to evaluate the effectiveness of interactive health education programs using eLearning modalities for creating awareness among parents and promoting health and well-being among school-aged children in a semi-rural community of Pakistan.
Our study showed that most of the health education sessions were beneficial and informative for parents, and they were very appreciative of the health awareness program initiated through the school platform. eLearning and health education have been widely studied in various countries and cultures, and their effects have been proven to have a significant impact on parental education, awareness, and health-seeking behaviors [15–19]. Akhter et al. [ 2015], reported the odds of under-five mortality $8\%$ lower for the children with mothers having secondary education, compared to the children with uneducated mothers [8]. Another study conducted in India reported that only $39\%$ of parents had knowledge and awareness of dental care which was addressed through health education. Similar studies carried out in western countries such as Australia and Spain on health education among parents for their child’s health showed significant results in terms of change in parental health practices. This sort of initiative has enhanced self-efficacy among parents to provide better health care to their children [20, 21].
Providing access to health through awareness using eLearning modalities has proved to be a paradigm shift and an important step towards addressing this gap among the parents and community leaders. The resource substitution theory states that health education benefits are greater for people in low-resourced settings [22]. In support of this theory, another study conducted on households in the USA determined that low health literacy affects parent acquisition of knowledge, attitudes, and behaviors, affecting child health outcomes across all the disease prevention [23]. Parents having access to online learning in rural areas would be a vital step forward in reducing health disparities. Literature doesn’t report any strong evidence of the use of eLearning for enhancing parental awareness in the local context, however, mobile health has been utilized to improve the attitude and healthcare practices among parents and caregivers in Pakistan [24].
Despite the success of the eLearning sessions, the parents still showed a preference for in-person interactions with the nurses and doctors. This finding was supported by another study conducted in the USA comparing an online and in-person educational program for parents demonstrated that the online session was as effective as the face-to-face program in achieving the required health goals for their children [25]. But the main reason for the preference for in-person interaction was the ability of the health care providers to empathize or understand the parent’s feelings better, rather than the gain in knowledge. A similar study carried out in Australia intending to understand the role of technology in a person-centered learning environment found that "eLearning teaching modalities is the key to revolutionizing education in the future," and utilizing technology will open ways to promote literacy [26]. Another study conducted on adopted children in China concluded that maternal education has a positive and causal effect on the child’s health, regardless after adding control for income, and other socioeconomic variables [27]. A systematic review and meta-analysis further reinforced the importance and impact of parental education on child development and health. In this study, the results showed that parental education was directly associated with reducing under-five mortality, with the mothers’ education being a more robust predictor [11]. Furthermore, a study in China proved that along with health, children with parents with a college education also objectively perform better in academics, owing to the positive interaction between children and their parents [28].
Involving the community leaders and stakeholders was one of the strengths of our study, which helped create a more comprehensive and lasting impact. A similar intervention was implemented in another study where local leaders were involved in encouraging and educating the community, which yielded positive results in the knowledge and decision-making skills of the participants [15].
Along with the overall benefits, this online education program through the school platform has proven that a program like this would help address health care needs by creating awareness among parents where accessibility of health is a significant issue. Using a technology-based health education initiative is a viable solution for providing health education because it is recognized as convenient, efficient, and comparatively economical for the community at large [29]. However, a combination of online and face-to-face may be more effective than exclusively online learning modality alone to improve learning outcomes [30]. Social Reforms are required at the grass-root level (townships, villages, etc.) to educate the community and ensure that the needs of young children in Pakistan are met in this new era of modern technology.
## Limitations
This study is among the few reported evidence from low-middle-income countries such as Pakistan. Though this study has reported the immediate effect of the intervention on parental awareness, our study did not show the subsequent impact on future practices of parents, which is one of the key limitations and an area of future research. Future studies on this subject matter should consider having schoolteachers on the research team utilizing an implementation science approach.
## Conclusion
Overall, the results from the quantitative arm showed significance in terms of increasing the parents’ knowledge level. The qualitative arm showed parents’ satisfaction with the online awareness program and revealed that it enhanced their awareness level about preventable health conditions of children. The participants showed enthusiasm and desire for more similar sessions in the future. The findings from this study have the potential of scaling it up to include various education setups as well as other communities and school teachers, which will be the focus of our future research.
## References
1. Sen A.. **Universal Health Care**. *Harvard Public Health Review* (2015.0) **5** 1-8
2. Van Den Muijsenbergh M, Van Weel C. **The essential role of primary care professionals in achieving health for all**. *The Annals of Family Medicine* (2019.0) **17** 293-5. DOI: 10.1370/afm.2436
3. Semke CA, Sheridan SM. **Family-school connections in rural educational settings: A systematic review of the empirical literature**. *School Community Journal* (2012.0) **22** 21-47
4. Huisman AH, Smits JP. *Keeping children in school: Household and district-level determinants of school dropout in 363 districts of 30 developing countries*
5. Wilder S.. **Effects of parental involvement on academic achievement: a meta-synthesis.**. *Educational Review* (2014.0) **66** 377-97
6. Lee HY, Zhou AQ, Lee RM, Dillon AL. **Parents’ functional health literacy is associated with children’s health outcomes: implications for health practice, policy, and research**. *Children and Youth Services Review.* (2020.0) **110** 104801
7. DeWalt DA, Hink A. **Health literacy and child health outcomes: a systematic review of the literature**. *Pediatrics* (2009.0) **124** S265-74. DOI: 10.1542/peds.2009-1162B
8. Akter T, Hoque DM, Chowdhury EK, Rahman M, Russell M, Arifeen SE. **Is there any association between parental education and child mortality? A study in a rural area of Bangladesh.**. *Public health* (2015.0) **129** 1602-9. DOI: 10.1016/j.puhe.2015.08.004
9. Masquelier B, Hug L, Sharrow D, You D, Hogan D, Hill K. **Global, regional, and national mortality trends in older children and young adolescents (5–14 years) from 1990 to 2016: an analysis of empirical data.**. *The Lancet Global Health* (2018.0) **6** e1087-99. DOI: 10.1016/S2214-109X(18)30353-X
10. Latif MM, Ashraf MF, Zeeshan M. **Factors Determining Poverty and Child Mortality in Pakistan.**. *iRASD Journal of Economics* (2019.0) **1** 32-41
11. Balaj M, York HW, Sripada K, Besnier E, Vonen HD, Aravkin A. **Parental education and inequalities in child mortality: a global systematic review and meta-analysis**. *The Lancet* (2021.0) **398** 608-20. DOI: 10.1016/S0140-6736(21)00534-1
12. Gulzar SA, Nayani P, Husain SS, Khan A. **Development of the School Health Education Program from the perspective of Comprehensive School Health Model: An example at a higher secondary school of Karachi, Pakistan**. *Taiwan Gong Gong Wei Sheng Za Zhi* (2017.0) **36** 545
13. Ishimaru AM. **From family engagement to equitable collaboration.**. *Educational Policy* (2019.0) **33** 350-85
14. Gubbels J, van der Put CE, Assink M. **Risk factors for school absenteeism and dropout: a meta-analytic review**. *Journal of youth and adolescence* (2019.0) **48** 1637-67. DOI: 10.1007/s10964-019-01072-5
15. Grant RH. **Family and self-help education in isolated rural communities.**. *Health Education Monographs* (1977.0) **5** 145-60. DOI: 10.1177/109019817700500205
16. Guilamo-Ramos V, Lee JJ, Kantor LM, Levine DS, Baum S, Johnsen J. **Potential for using online and mobile education with parents and adolescents to impact sexual and reproductive health**. *Prevention Science* (2015.0) **16** 53-60. DOI: 10.1007/s11121-014-0469-z
17. Cowan S, Pease A, Bennett S. **Usage and impact of an online education tool for preventing sudden unexpected death in infancy**. *Journal of Paediatrics and Child Health* (2013.0) **49** 228-32. DOI: 10.1111/jpc.12128
18. de Graaf M, Knol MJ, Totté JE, van Os-Medendorp H, Breugem CC, Pasmans SG. **E-learning enables parents to assess an infantile hemangioma**. *Journal of the American Academy of Dermatology* (2014.0) **70** 893-8. DOI: 10.1016/j.jaad.2013.10.040
19. DeHoff BA, Staten LK, Rodgers RC, Denne SC. **The role of online social support in supporting and educating parents of young children with special health care needs in the United States: a scoping review**. *Journal of medical Internet research* (2016.0) **18** e6722. DOI: 10.2196/jmir.6722
20. McDevitt SE. **While quarantined: An online parent education and training model for families of children with autism in China**. *Research in Developmental Disabilities.* (2021.0) **109** 103851. DOI: 10.1016/j.ridd.2020.103851
21. Hoeft KS, Barker JC, Shiboski S, Pantoja‐Guzman E, Hiatt RA. **Effectiveness evaluation of Contra Caries Oral Health Education Program for improving Spanish‐speaking parents’ preventive oral health knowledge and behaviors for their young children**. *Community dentistry and oral epidemiology* (2016.0) **44** 564-76. DOI: 10.1111/cdoe.12250
22. Ross CE, Mirowsky J. **Sex differences in the effect of education on depression: Resource multiplication or resource substitution**. *Social Science & Medicine* (2006.0) **63** 1400-13. DOI: 10.1016/j.socscimed.2006.03.013
23. Ross CE, Mirowsky J. **The interaction of personal and parental education on health**. *Social Science & Medicine* (2011.0) **72** 591-9. DOI: 10.1016/j.socscimed.2010.11.028
24. Siddiqui M, ul Islam MY, Mufti BAI, Khan N, Farooq MS, Muhammad MG. **Assessing acceptability of hypertensive/diabetic patients towards mobile health based behavioral interventions in Pakistan: A pilot study.**. *International Journal of Medical Informatics* (2015.0) **84** 950-5. DOI: 10.1016/j.ijmedinf.2015.08.009
25. Roberts CA, Smith KC, Sherman AK. **Comparison of online and face-to-face parent education for children with autism and sleep problems**. *Journal of Autism and Developmental Disorders* (2019.0) **49** 1410-22. DOI: 10.1007/s10803-018-3832-2
26. Motschnig-Pitrik R.. **Person-centered e-learning in action: Can technology help to manifest person-centered values in academic environments**. *Journal of Humanistic Psychology* (2005.0) **45** 503-30
27. Chen Y, Li H. **Mother’s education and child health: Is there a nurturing effect**. *Journal of health economics* (2009.0) **28** 413-26. DOI: 10.1016/j.jhealeco.2008.10.005
28. Liu Y, Mao M, Zhang H. **The external effect of better-educated parents on student academic outcome**. *Applied Economics Letters* (2021.0) **24** 1-5
29. van Cranenburgh OD, Ter Stege JA, de Rie MA, Smets E, Sprangers MA, de Korte J. **Implementation of web-based education for patients with chronic skin disease**. *Acta Dermato-Venereologica.* (2016.0) **96** 812-3. DOI: 10.2340/00015555-2365
30. Means B, Toyama Y, Murphy R, Baki M. **The effectiveness of online and blended learning: A meta-analysis of the empirical literature**. *Teachers college record* (2013.0) **115** 1-47
|
---
title: The relationship of insulin resistance and diabetes to tau PET SUVR in middle-aged
to older adults
authors:
- Gilda E. Ennis
- Tobey J. Betthauser
- Rebecca Langhough Koscik
- Nathaniel A. Chin
- Bradley T. Christian
- Sanjay Asthana
- Sterling C. Johnson
- Barbara B. Bendlin
journal: Alzheimer's Research & Therapy
year: 2023
pmcid: PMC10022314
doi: 10.1186/s13195-023-01180-2
license: CC BY 4.0
---
# The relationship of insulin resistance and diabetes to tau PET SUVR in middle-aged to older adults
## Abstract
### Background
Insulin resistance (IR) and type 2 diabetes have been found to increase the risk for Alzheimer’s clinical syndrome in epidemiologic studies but have not been associated with tau tangles in neuropathological research and have been inconsistently associated with cerebrospinal fluid P-tau181. IR and type 2 diabetes are well-recognized vascular risk factors. Some studies suggest that cardiovascular risk may act synergistically with cortical amyloid to increase tau measured using tau PET. Utilizing data from largely nondemented middle-aged and older adult cohorts enriched for AD risk, we investigated the association of IR and diabetes to tau PET and whether amyloid moderated those relationships.
### Methods
Participants were enrolled in either the Wisconsin Registry for Alzheimer’s Prevention (WRAP) or Wisconsin Alzheimer’s Disease Research Center (WI-ADRC) Clinical Core. Two partially overlapping samples were studied: a sample characterized using HOMA-IR ($$n = 280$$ WRAP participants) and a sample characterized on diabetic status ($$n = 285$$ WRAP and $$n = 109$$ WI-ADRC). IR was measured using the homeostasis model assessment of insulin resistance (HOMA-IR). Tau PET employing the radioligand 18F-MK-6240 was used to detect AD-specific aggregated tau. Linear regression tested the relationship of IR and diabetic status to tau PET standardized uptake value ratio (SUVR) within the entorhinal cortex and whether relationships were moderated by amyloid assessed by amyloid PET distribution volume ratio (DVR) and amyloid PET positivity status.
### Results
Neither HOMA-IR nor diabetic status was significantly associated with tau PET SUVR. The relationship between IR and tau PET SUVR was not moderated by amyloid PET DVR or positivity status. The association between diabetic status and tau PET SUVR was not significantly moderated by amyloid PET DVR but was significantly moderated by amyloid PET positivity status. Among the amyloid PET-positive participants, the estimated marginal tau PET SUVR mean was higher in the diabetic ($$n = 6$$) relative to the nondiabetic group ($$n = 88$$).
### Conclusion
Findings indicate that IR may not be related to tau in generally healthy middle-aged and older adults who are in the early stages of the AD clinicopathologic continuum but suggest the need for additional research to investigate whether a synergistic relationship between type 2 diabetes and amyloid is associated with increased tau levels.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s13195-023-01180-2.
## Background
Insulin resistance (IR), a condition that disrupts metabolic homeostasis and increases the risk for type 2 diabetes [1, 2], has been associated with an increased risk for Alzheimer’s clinical syndrome [3, 4]. IR within the peripheral tissues has been hypothesized by some to alter neuronal insulin signaling and contribute to the development of neurofibrillary tangles (NFT) [5], one of the pathologic hallmarks of Alzheimer’s disease (AD). Some studies demonstrate that animals fed high-fat diets to model IR have higher levels of soluble hyperphosphorylated tau [6–8]. Tau hyperphosphorylation is thought to precede aggregation of tau in humans into the insoluble paired helical filaments (PHF) [9–11] that comprise the NFTs of AD [12]. In some human studies [13, 14], IR has been related to higher levels of a soluble form of tau phosphorylated at threonine 181 (P-tau181), an epitope-specific for AD [15]. Higher P-tau181 in the cerebrospinal fluid (CSF) has been associated with higher homeostatic model assessment of insulin resistance (HOMA-IR) values in cognitively unimpaired older adults [13] and in cognitively unimpaired middle-aged and older adults who are carriers of the APOE ε4 allele [14]. However, significant associations between IR and CSF P-tau181 have not been consistently found [16], and increased tau phosphorylation has not been consistently isolated in animal models of IR [17]. Similarly, type 2 diabetes, which is characterized by IR as well as insulin deficiency [2, 18], has been shown to associate with higher CSF P-tau181 in some studies [19] but not in others [20].
Although elevations in CSF P-tau181 have been found to predict the later development of NFTs [21], P-tau may not be a direct marker of NFTs or PHFs like tau positron emission tomography (PET) [22, 23]. Instead, higher concentrations of CSF P-tau181, at least in early AD, may represent a neuronal response to amyloid exposure [21, 22, 24]. In the few neuropathological studies that have investigated the relationship between antemortem IR and NFTs, IR has not been significantly related to regional NFT spread assessed by the Braak score [25, 26]. Similarly, type 2 diabetes has not been related to either the presence or quantity of NFTs in post-mortem studies [27–30]. Approximately one-half of the cases in these studies had an antemortem diagnosis of dementia [26, 30], suggesting that results may be more relevant to a later stage of the AD continuum. Whether IR and diabetes are related to NFTs in adults without dementia who are in an earlier stage of AD, when treatment is more likely to be effective, has not been well-studied.
In addition to their effects upon metabolism, IR and type 2 diabetes also impact the vasculature [31, 32] and increase the risk for cardiovascular disease (CVD) [33, 34]. Some research suggests that cardiovascular disease risk and vascular dysfunction are related to greater tau burden in individuals with higher cortical amyloid [35, 36]. Rabin et al. [ 2019] found that in a sample of cognitively unimpaired older adults, participants with both higher Framingham Heart Study (FHS) CVD risk and higher amyloid PET distribution volume ratio (DVR) had greater tau PET standardized uptake value ratio (SUVR). However, another similar study using the same PET tracers in cognitively unimpaired adults did not find that an interaction between FHS CVD risk and amyloid was related to tau [37]. That study had fewer participants who were APOE4 allele carriers and a smaller range of CVD risk, which could have influenced results. Neither study specifically examined the association of IR and amyloid on tau PET SUVR.
We tested the association of IR and diabetic status to aggregated tau, using MK-6240 PET, in a sample enriched for AD risk due to an increased proportion of APOE ε4 allele carriers and comprised largely of middle-aged and older adults who were cognitively unimpaired. We also explored whether the relationship of IR and diabetic status to aggregated tau was moderated by amyloid burden, assessed using PiB PET.
## Participants
Participants were enrolled in either the Wisconsin Registry for Alzheimer’s Prevention (WRAP), a longitudinal study of middle-aged and older adults enriched for AD risk [38], or the Wisconsin Alzheimer’s Disease Research Center (WI-ADRC) Clinical Core. Diagnosis of mild cognitive impairment and dementia was determined by a multidisciplinary consensus review team [38, 39]. Relevant medical and cognitive data were evaluated to determine cognitive status based upon NIA-AA criteria [40, 41]. All participants provided written informed consent prior to study participation. Study procedures were approved by the University of Wisconsin – Madison Institutional Review Board.
Participants were selected depending upon the availability of tau and amyloid PET, diabetic status data, and fasting glucose and fasting insulin, which are required to calculate HOMA-IR [42] Because WRAP and not WI-ADRC participants have insulin collected at regular study visits, only WRAP participants were included in the sample (i.e., the “HOMA-IR sample”) utilized to test the relationship between IR and tau PET SUVR. $$n = 281$$ WRAP participants had PET as well as fasting glucose and insulin for inclusion in the HOMA-IR sample. Because HOMA-IR is not recommended as a measure for IR in people on insulin therapy [42], one participant with type 2 diabetes who was receiving insulin therapy was excluded, resulting in $$n = 280$$ participants in the HOMA-IR sample. $$n = 394$$ participants ($$n = 285$$ WRAP [which included $$n = 280$$ from the HOMA-IR sample] and $$n = 109$$ WI-ADRC) had PET and diabetic status data; therefore, they were included in the sample used to test the relationship between diabetic status and tau PET SUVR (i.e., the “Diabetic Status sample”). Descriptive statistics of demographic characteristics and study variables for both samples can be found in Tables 1 and 2.Table 1Descriptive statistics of demographic and health characteristics and study variables. Data presented are means (standard deviations) or counts (%). Between-group differences tested using independent samples t-test for continuous variables and Pearson’s chi-square, Fisher’s Exact test, or Fisher-Freeman-Hamilton Exact test for categorical variablesaHOMA-IR sample ($$n = 280$$)Amyloid PET positiveb($$n = 67$$)Amyloid PET negative($$n = 213$$)pDiabetic Status sample ($$n = 394$$)Diabeticc ($$n = 37$$)Non-diabetic ($$n = 357$$)pAge (years)68.1 (6.6)70.8 (4.9)67.3 (6.8)<.00168.0 (7.1)69.8 (7.4)67.8 (7.1).11Sex (female)186 (66.4)41 (61.2)145 (68.1).30261 (66.2)24 (64.9)237 (66.4).85Race/ethnic group:1.0<.001White263 (93.9)63 (94.0)200 (93.9)358 (90.9)24 (64.9)334 (93.6)<.001Black12 (4.3)3 (4.5)9 (4.2)22 (5.6)9 (24.3)13 (3.6)<.001American Indian, Asian, Hispanic/Spanishd5 (1.8)1 (1.5)4 (1.9)14 (3.6)4 (10.8)10 (2.8).01Education (years)16.2 (2.2)16.1 (2.2)16.2 (2.2).7516.1 (2.4)15.6 (2.3)16.2 (2.4).19n=272HOMA2-IRn=22n=250HOMA2-IRe1.1 (0.7)1.0 (0.4)1.1 (0.7).181.1 (0.7)1.6 (0.8)1.0 (0.7)<.001n=381Glucosen=32n=349Glucose (mg/dL)98.6 (14.7)96.1 (9.8)99.4 (15.8).1298.5 (14.1)128.9 (23.5)95.7 (8.7)<.001Prediabetesf74 (26.4)21 (31.3)53 (24.9).30100 (25.9)--100 (28.0)Diabetes22 (7.9)1 (1.5)21 (9.9).0337 (9.4)37 (100.0)--Diabetic medications: Diabetics15 (5.4)0 [0]15 (7.0)--27 (6.9)27 (73.0) Non-diabeticsg2 (0.8)0 [0]2 (0.9)--4 (1.0)--4 (1.1)$$n = 276$$APOE4 allele statusn=66n=210n=379APOE4 allele statusn=33n=346APOE4 allele status:<.001.09 Non-carrier170 (61.6)23 (34.8)147 (70.0)<.001228 (60.2)26 (78.8)202 (58.4) ε2 ε47 (2.5)1 (1.5)6 (2.9)>.059 (2.4)1 (3.0)8 (2.3) ε3 ε486 (31.2)33 (50.0)53 (25.2)<.001120 (31.7)5 (15.2)115 (33.2) ε4 ε413 (4.7)9 (13.6)4 (1.9)<.00122 (5.8)1 (3.0)21 (6.1)Amyloid PET DVRh1.16 (0.22)1.50 (0.23)1.06 (0.05)<.0011.17 (0.23)1.15 (0.24)1.17 (0.23).58Amyloid PET positiveb67 (23.9)67 (100.0)0 [0]--94 (23.9)6 (16.2)88 (24.6).25Tau PET SUVR, ECi1.09 (0.27)1.30 (0.41)1.02 (0.15)<.0011.12 (0.32)1.17 (0.42)1.12 (.31).31Tau PET positivej38 (13.6)25 (37.3)13 (6.1)<.00161 (15.5)9 (24.3)52 (14.6).12Tau PET SUVR, MTLk0.98 (.22)1.15 (.34)0.93 (.12)<.0011.00 (.27)1.03 (.34)1.00 (.26).51Tau PET SUVR, temporal meta-ROIl1.13 (.25)1.31 (.43)1.07 (.11)<.0011.15 (.31)1.17 (.31)1.15 (.31).77Time to tau PETm (years)1.06 (1.05)1.12 (1.10)1.05 (1.03).60.92 (.94).89 (.70).93 (.97).78MCIn9 (3.2)7 (10.4)2 (0.9).00123 (5.8)3 (8.1)20 (5.6).47Dementian0 [0]0 [0]0 [0]--6 (1.5)2 (5.4)4 (1.1).10Abbreviations: DVR Distribution volume ratio, EC Entorhinal cortex, HOMA2-IR Homeostasis model assessment of insulin resistance, MCI Mild cognitive impairment, MTL Medial temporal lobe, PET Positron emission tomography, ROI, Region of interest, SUVR, Standardized uptake value ratioaFor categorical variables with >2 cells, p-value for main effect is noted first followed by p-values for statistically significant post hoc pairwise comparisonsbAmyloid PET positive: average Pittsburgh Compound B (PiB) DVR > 1.19 from 8 bilateral regions at PiB PET closest in time to tau PETcDiabetes identified by clinician or self-report of diabetes or fasting glucose ≥ 126 mg/dLdThe 3 race/ethnic groups were combined to maintain anonymity for groups with < 3 individualseHOMA2-IR has no reference range; a value of 1.0 approximates normal (Wallace, Levy, & Matthews, 2004). HOMA2-IR values in the Diabetic Status sample are from WRAP participants onlyfPrediabetes identified by fasting glucose ≥ 100 mg/dL (American Diabetes Association, 2010)gOff-label use of metformin in non-diabeticshValue represents average PiB PET DVR across 8 bilateral regionsiValue represents average tau PET SUVR from bilateral entorhinal cortexjTau PET positive: average tau PET SUVR > 1.27 from bilateral entorhinal cortexkValue represents average tau PET SUVR from bilateral entorhinal cortex, hippocampus, and amygdalalValue represents average tau PET SUVR from bilateral parahippocampal gyrus, amygdala, fusiform cortex, and inferior and middle temporal gyrusmTime between predictor (HOMA2-IR or diabetic status) and tau PETnDiagnosed using NIA-AA criteria and consensus conferenceTable 2Descriptive statistics of demographic and health characteristics and study variables in the Diabetic Status sample characterized according to amyloid PET positivity status. Data presented are means (standard deviations) or counts (%). Between-group differences tested using t test for continuous variables and Pearson’s chi-square, Fisher’s exact test, or Fisher-Freeman-Hamilton exact test for categorical variablesa($$n = 394$$)Diabeticb amyloid PET positivec($$n = 6$$)Non-diabetic amyloid PET positivec($$n = 88$$)pDiabeticb amyloid PET negative($$n = 31$$)Non-diabetic amyloid PET negative($$n = 269$$)pAge (years)68.0 (7.1)75.12 (6.4)71.3 (5.2).0968.7 (7.2)66.7 (7.2).14Sex (female)261 (66.2)4 (66.7)55 (62.5)1.020 (64.5)182 (67.7).84Race/ethnic group:.002<.001White358 (90.9)3 (50.0)84 (95.5)<.00121 (67.7)250 (92.9)<.001Black22 (5.6)1 (16.7)3 (3.4)>.058 (25.8)10 (3.7)<.001American Indian, Asian, Hispanic/Spanishd14 (3.6)2 (33.3)1 (1.1)<.0012 (6.5)9 (3.3)>.05Education (years)16.1 (2.4)17.3 (2.1)16.1 (2.3).1815.3 (2.2)16.2 (2.4).05n=272HOMA2-IRn=1n=66n=21n=184HOMA2-IRe1.1 (0.7)0.51.0 (0.4)--1.7 (0.9)1.1 (0.7)<.001n=381Glucosen=5n=85n=27n=264Glucose (mg/dL)98.5 (14.1)117.6 (31.7)96.5 (9.1).21131.0 (21.8)95.5 (8.6)<.001Prediabetesf100 (25.9)--28 (32.9)--72 (27.3)Diabetic medications: Diabetics27 (6.9)3 (50.0)--24 (77.4)-- Non-diabeticsg4 (1.0)--0 (0.0)--4 (1.5)$$n = 379$$APOE4 allele statusn=5n=83n=28n=263APOE4 allele status:.14 Non-carrier228 (60.2)2 (40.0)27 (32.5)1.024 (85.7)175 (66.5) ε2 ε49 (2.4)0 (0.0)1 (1.2)1 (3.6)7 (2.7) ε3 ε4120 (31.7)2 (40.0)40 (48.2)3 (10.7)75 (28.5) ε4 ε422 (5.8)1 (20.0)15 (8.1)0 (0.0)6 (2.3)Amyloid PET DVRh1.17 (0.23)1.62 (.31)1.50 (.23).251.06 (.05)1.06 (.05).72Tau PET SUVR, ECi1.12 (0.32)1.94 (.53)1.37 (.48).0061.02 (.15)1.03 (.17).83Tau PET positivej61 (15.5)6 (100.0)37 (42.0).0073 (9.7)15 (5.6).41Tau PET SUVR, MTLk1.00 (.27)1.62 (.49)1.21 (.40).02.92 (.11).93 (.13).52Tau PET SUVR, temporal meta-ROIl1.15 (.31)1.70 (.44)1.37 (.53).151.06 (.11)1.08 (.11).48Time to tau PETm (years).92 (.94).96 (.92)1.02 (1.05).90.88 (.67).90 (.94).89MCIn23 (5.8)1 (16.7)15 (17.0)1.02 (6.5)5 (1.9).16Dementian6 (1.5)1 (16.7)3 (3.4).241 (3.2)1 (0.4).20Abbreviations: DVR Distribution volume ratio, EC Entorhinal cortex, HOMA2-IR Homeostasis model assessment of insulin resistance, MCI Mild cognitive impairment, MTL Medial temporal lobe, PET Positron emission tomography, ROI, Region of interest, SUVR Standardized uptake value ratioaFor categorical variables with > 2 cells, p-value for main effect is noted first followed by p-values for statistically significant post hoc pairwise comparisonsbDiabetes identified by clinician or self-report of diabetes or fasting glucose ≥ 126 mg/dLcAmyloid PET positive: average Pittsburgh Compound B (PiB) DVR > 1.19 from 8 bilateral regions at PiB PET closest in time to tau PETdThe 3 race/ethnic groups were combined to maintain anonymity for groups with < 3 individualseHOMA2-IR has no reference range; a value of 1.0 approximates normal (Wallace, Levy, & Matthews, 2004). HOMA2-IR values in the Diabetic Status sample are from WRAP participants onlyfPrediabetes identified by fasting glucose ≥ 100 mg/dL (American Diabetes Association, 2010)gOff-label use of metformin in non-diabeticshValue represents average PiB PET DVR across 8 bilateral regionsiValue represents average tau PET SUVR from bilateral entorhinal cortexjTau PET positive: average tau PET SUVR > 1.27 from bilateral entorhinal cortexkValue represents average tau PET SUVR from bilateral entorhinal cortex, hippocampus, and amygdalalValue represents average tau PET SUVR from bilateral parahippocampal gyrus, amygdala, fusiform cortex, and inferior and middle temporal gyrusmTime between predictor (HOMA2-IR or diabetic status) and tau PETnDiagnosed using NIA-AA criteria and consensus conference
## General
Participants fasted for a minimum of 8 h prior to having their blood collected during a regular biennial or annual visit. Medical diagnoses (e.g., diabetes) and medication history (e.g., antidiabetic medications) were self-reported through medical history questionnaires and/or clinician interviews as part of the source study evaluations. PET imaging was collected during a regular biennial or annual visit or as part of a PET sub-study using a common acquisition protocol.
## Insulin resistance (IR)
IR was measured using the “updated” homeostasis model assessment of insulin resistance (HOMA2-IR [42];). HOMA2-IR was calculated by entering fasting glucose and insulin into the HOMA2 calculator version 2.2.3 (University of Oxford). HOMA2 modeling has been shown to correlate strongly with the euglycemic clamp and minimal model methods of whole-body insulin sensitivity [42]. A value of 1.0 is considered normal, but there is no reference range. In the HOMA-IR sample, the average HOMA2-IR in non-diabetic participants was 1.0 ($$n = 258$$, SD =.66, interquartile range =.6 to 1.3, full range = 0.1 to 5.6) and the average value in diabetic participants was 1.6 ($$n = 22$$, SD = 0.8, interquartile range = 1.1 to 2.1, full range =.5 to 3.4). Although we are unaware of published cut-points for HOMA2-IR for the US population, a HOMA1-IR cut-point of 2.7 has been used as a threshold for identifying insulin resistance in US samples of nondiabetic adults [43, 44]. This value was equivalent to a HOMA2-IR value of 1.3 in the HOMA2-IR sample. There where $$n = 85$$ ($30.4\%$) participants with HOMA2-IR ≥ 1.3 in that sample. HOMA2-IR and HOMA1-IR, the original HOMA method for calculating IR, were strongly correlated ($r = .98$, $p \leq .001$). Fasting glucose and insulin collected closest in time and within the same month as or prior to tau PET were used to calculate HOMA2-IR. On average glucose and insulin were collected 1.06 years prior to tau PET (year of tau PET minus year of blood collection: SD = 1.05, interquartile range =.17 to 1.75, full range = −.07 to 5.68).
## Diabetic status
Diabetic status was determined by clinician or self-report of diabetes or fasting glucose ≥ 126 mg/dL, a value considered diagnostic for diabetes [45]. $$n = 14$$ WI-ADRC participants had a report of type 2 diabetes determined through clinician interviews. There were no WI-ADRC participants without a clinician report of type 2 diabetes who had a fasting glucose ≥ 126 mg/dL. $$n = 20$$ WRAP participants had a self-report of diabetes, and $$n = 3$$ had a high fasting glucose. WRAP participants did not indicate whether diabetes was type 1 or 2, but no participant self-reported being on insulin therapy in the absence of oral antidiabetic medications. Information collected closest in time and within the same month or prior to tau PET was used to determine diabetic status. On average diabetic status information was collected.92 years prior to tau PET (year of tau PET minus year of diabetic status: SD =.94, interquartile range =.16 to 1.45, full range = −.08 to 5.68).
## Tau and amyloid PET acquisition and processing
18F-MK-6240 and 11C-Pittsburgh Compound B (PiB) were used to quantify aggregated tau and cortical β-amyloid, respectively, using previously published methods [46, 47]. PET data were collected using either a Siemens Biograph Horizon PET/CT or Siemens EXACT HR+ tomograph. Dynamic data were acquired for 20 min (5 min × 4 frames) following a 70-min uptake period for MK-6240 and 0–70 min (2 min × 5 frames, 5 min × 12 frames) beginning with tracer injection for PiB. T1-weighted magnetic resonance imaging (MRI) was performed to delineate anatomical regions. MRI and PET image processing and quality control were performed using a pipeline that uses MATLAB (The Mathworks, Inc., Natick, MA) and SPM12 (University College London). Details regarding radioligand synthesis, image acquisition, processing, and analysis of MRI, MK-6240 PET images [46], and PiB PET images [47] have been previously described.
## Amyloid PET DVR
Cortical PiB DVR (Logan graphical analysis, cerebellum gray matter reference region) was averaged across 8 bilateral regions as previously described [48]. A PiB DVR of 1.19 (equivalent to 20.6 Centiloids) was the cut-point for amyloid PET positivity [49]. We used the amyloid PET scan closest in time to tau PET. On average amyloid PET was performed.02 years (i.e., 7.3 days) prior to tau PET (calculated as year of tau PET minus year of amyloid PET) in both the HOMA-IR (SD =.24; interquartile range = 0 to 0; full range = −.58 to 2.64) and Diabetic Status samples (SD =.25; interquartile range = 0 to 0, full range = -.69 to 2.64.).
## Tau PET SUVR
Tau burden was ascertained using the average MK-6240 standardized uptake value ratio (SUVR; 70–90 min, inferior cerebellum reference region [46]) from the left and right entorhinal cortex (EC). The EC was selected as the primary region of interest (ROI) because it is one of the first to develop NFTs according to Braak pathological staging [50] and tau tracer uptake in this region has been shown to be associated with cognitive decline in preclinical samples [51, 52]. The cut-point for tau PET positivity was an EC tau PET SUVR of 1.27 [51]. To provide context for results obtained using the EC ROI, we examined secondarily a medial temporal lobe (MTL) composite (bilateral entorhinal cortex, hippocampus, and amygdala) and a temporal lobe meta-ROI (bilateral parahippocampal gyrus, fusiform gyrus, inferior and middle temporal gyrus, and amygdala) [53].
## Statistical analyses
Between-group comparisons were used to describe both samples. We examined amyloid PET positivity group differences in the HOMA-IR sample and diabetic status group differences in the Diabetic Status sample. In the latter sample, we also determined diabetic status group differences within amyloid PET positive and negative groups. Between-group differences were tested using independent samples t-test for continuous variables and Pearson chi-square for categorical variables with > 5 cases per cell, Fisher’s Exact test for variables with 2 categories and ≤ 5 cases in a cell, and Fisher-Freeman-Hamilton Exact test for variables with > 2 categories and ≤ 5 cases in a cell. For variables with > 2 categories, post hoc pairwise comparisons for significant main effects were conducted using a z-test for independent proportions and Benjamini-Hochberg correction to adjust for multiple comparisons.
Two separate multiple linear regression models tested the relationship of HOMA2-IR and diabetic status to EC tau PET SUVR. Two separate moderated regression models tested if the relationship of IR and diabetic status to EC tau PET SUVR was moderated by amyloid burden. Prior to calculating the 2 interaction terms (i.e., HOMA2-IR × amyloid PET DVR and diabetic status × amyloid PET DVR), amyloid PET DVR was centered at 1.19, the cut-off for amyloid PET positivity, and HOMA2-IR was centered at the sample mean. Age, sex, cognitive status, and amyloid PET DVR were controlled in all analyses. Cohort was additionally controlled in analyses performed on the Diabetic Status sample. Residual plots from each model were examined to evaluate assumptions of regression. If heteroscedasticity was detected, the Breusch-Pagan test was performed to confirm the presence of non-constant variance. For models demonstrating heteroscedasticity, we performed robust regression employing a heteroscedasticity-consistent standard error estimator (HC3) [54]. The same analysis plan was followed when testing secondary tau PET SUVR ROIs as dependent variables.
IBM® SPSS® version 27 and SAS® version 9.4 were utilized for statistical analyses.
## Results
Between-group differences can be found in Tables 1 and 2. Antidiabetic medication usage by participants in the Diabetic Status sample can be found in Supplementary Table 1. In the HOMA-IR sample, HOMA2-IR was not significantly different between the amyloid PET positive ($$n = 67$$) and negative ($$n = 213$$) groups. In the Diabetic Status sample, amyloid PET DVR and the proportion of participants who were amyloid PET positive were not significantly different between the diabetic and nondiabetic groups. Among amyloid PET-positive participants, the diabetic group ($$n = 6$$) had significantly higher EC and MTL tau PET SUVR than the nondiabetic group ($$n = 88$$). Among amyloid PET-negative participants, EC and MTL tau PET SUVR was not significantly different between the diabetic ($$n = 31$$) and nondiabetic ($$n = 269$$) groups. A scatterplot demonstrating the relationship between amyloid PET DVR and EC tau PET SUVR in diabetic and nondiabetic participants in the Diabetic Status sample is presented in Fig. 1. A second scatterplot demonstrating the relationship between HOMA2-IR and EC tau PET SUVR is presented in Fig. 2.Fig. 1Scatterplot demonstrating relationship between amyloid PET DVR and entorhinal cortex tau PET SUVR in diabetic and nondiabetic participants in the Diabetic Status sample. Horizontal line set at tau PET positivity threshold (SUVR = 1.27); vertical line set at amyloid PET positivity threshold (DVR = 1.19). A, amyloid; T, tauFig. 2Scatterplot demonstrating relationship between HOMA2-IR and entorhinal cortex tau PET SUVR in amyloid PET positive and negative participants (threshold = 1.19 DVR). Horizontal line set at tau PET positivity threshold (SUVR = 1.27) Review of the residual plots revealed a likely violation of the homoscedasticity assumption in all 4 linear regression models. The Breusch-Pagan test was significant for each model (Supplementary Table 2), indicating the presence of non-constant variance. Thus, we ran robust regression employing a heteroscedasticity-consistent standard error estimator (HC3). Results from robust regression demonstrated that HOMA2-IR and diabetic status were not significantly related to EC tau PET SUVR (see Table 3). The HOMA2-IR × amyloid PET DVR and diabetic status × amyloid PET DVR interactions were also not significantly related to EC tau PET SUVR (see Table 3). When examining as dependent variables the secondary tau PET SUVR ROIs, we found that HOMA2-IR and diabetic status, as well as their interaction with amyloid PET DVR, were not significantly related to tau PET SUVR in the MTL and temporal lobe meta-ROIs (see Table 4). As in primary analyses, robust regression was used in these secondary analyses due to the presence of heteroscedasticity. Table 3Results from robust regression testing the relationships of a) HOMA2-IR and diabetic status, as well as b) their interaction with amyloid PET DVR, to tau PET SUVR in the entorhinal cortexaHOMA-IR sampleDiabetic Status sampleb (SE)bp$95\%$ CIb (SE)bp$95\%$ CI(a)Age (years).001 (.002).42−.002 to.005.004 (.002).02.001 to.007Sex (0=female)−.06 (.03).04−.11 to −.003−.02 (.03).56−.07 to.04Cognitive status (0=unimpaired)c.20 (.13).13−.06 to.46.22 (.09).02.04 to.41Amyloid PET DVR.68 (.12)<.001.45 to.91.79 (.10)<.001.60 to.99HOMA2-IR.02 (.02).23−.01 to.05------Cohort (WRAP=0)------.08 (.03).004.03 to.14Diabetic status (0=nondiabetic)------.04 (.04).32-.04 to.13(b)Age (years).001 (.002).39−.002 to.005.004 (.002).04.0002 to.007Sex (0=female)−.06 (.03).04−.11 to −.003−.03 (.03).30−.08 to.03Cognitive status (0=unimpaired)c.19 (.13).16.07 to.45.23 (.09).01.05 to.41Amyloid PET DVR (centered)d.69 (.13)<.001.44 to.93.75 (.11)<.001.54 to.97HOMA2-IR (centered)e.03 (.03).42−.03 to.09------HOMA2-IR × amyloid PET DVR.10 (.23).68−.36 to.55------Cohort (WRAP=0)------.08 (.03).008.02 to.13Diabetic status (0=nondiabetic)------.06 (.05).26-.04 to.16Diabetic status × amyloid PET DVR------.36 (.26).17−.16 to.88Abbreviations: CI Confidence interval; DVR Distribution volume ratio, HOMA2-IR Homeostasis model assessment of insulin resistance, PET Positron emission tomography, SE Standard error, SUVR standardized uptake value ratioaTau PET SUVR was the average value from bilateral entorhinal cortexbComputed using a heteroscedasticity-consistent standard error estimator (HC3)cn=9 with MCI in HOMA-IR sample; $$n = 23$$ with MCI and $$n = 6$$ with dementia in Diabetic Status sampledAmyloid PET DVR centered at cut-off for amyloid PET positivity (DVR = 1.19)eHOMA2-IR centered at meanTable 4Results from robust regression testing the relationship of HOMA2-IR and diabetic status, as well as their interaction with amyloid PET DVR, to tau PET SUVR in the medial temporal lobe and temporal lobe meta-ROITau PET SUVR, medial temporal lobeaTau PET SUVR, temporal meta-ROIbb (SE)cp$95\%$ CIb (SE)cp$95\%$ CIHOMA2-IRd.01 (.01).36−.01 to.04.004 (.01).68−.02 to.03HOMA2-IR × amyloid DVRd.24 (.23).31−.22 to.69−.22 (.20).27−.60 to.17Diabetic statuse (0=nondiabetic).02 (.04).59−.05 to.09.01 (.03).69−.05 to.07Diabetic status × amyloid PET DVRe.24 (.28).40−.32 to.79.22 (.25).37−.26 to.71Abbreviations: CI Confidence interval, DVR Distribution volume ratio, HOMA2-IR Homeostasis model assessment of insulin resistance, PET Positron emission tomography, ROI Region of interest, SE Standard error, SUVR Standardized uptake value ratioaAverage tau PET SUVR from bilateral entorhinal cortex, hippocampus, and amygdalabAverage tau PET SUVR from bilateral parahippocampal gyrus, amygdala, fusiform cortex, and inferior and middle temporal gyruscComputed using a heteroscedasticity-consistent standard error estimator (HC3)dAnalyses controlled for age, sex, and cognitive status. Amyloid PET DVR centered at cut-off for amyloid PET positivity (DVR = 1.19). HOMA2-IR centered at meaneAnalyses controlled for age, sex, cognitive status, and cohort. Amyloid PET DVR centered at cut-off for amyloid PET positivity (DVR = 1.19) Because the diabetic amyloid PET positive group had higher EC and MTL tau PET SUVR than the nondiabetic amyloid PET positive group, we tested whether a diabetic status × amyloid PET positivity interaction would be significantly related to EC and MTL tau PET SUVR when controlling for age, sex, cohort, and cognitive status. Using robust regression with the HC3 estimator, we found that the interaction in both models was significant (EC tau PET SUVR model: $b = .51$, $$p \leq .01$$; MTL tau PET SUVR model: $b = .37$, $$p \leq .03$$; see Supplementary Table 3). To obtain EC and MTL tau PET SUVR estimated marginal means for each of the four diabetic status by amyloid PET positivity status groups, we used robust regression in SAS®. We then tested the difference in marginal means between the diabetic (EC tau PET SUVR mean = 1.79; MTL tau PET SUVR mean = 1.50) and nondiabetic (EC tau PET SUVR mean = 1.32 EC; MTL tau PET SUVR mean = 1.17) amyloid PET positive groups. We found that the diabetic amyloid PET positive group had a significantly higher mean EC tau PET SUVR than the nondiabetic amyloid PET positive group ($t = 2.39$, $$p \leq .02$$; see Fig. 3a). The difference in adjusted MTL tau PET SUVR means between the diabetic and nondiabetic amyloid PET positive groups was close to the significance threshold ($t = 1.94$, $$p \leq .05$$; see Fig. 3b). We then conducted a sensitivity analysis excluding participants with cognitive impairment ($$n = 29$$) and controlling for age, sex, and cohort in a model testing EC tau PET SUVR as the outcome. The pattern of results was similar, but the diabetic status × amyloid PET positivity interaction ($b = .33$, $$p \leq .07$$) and the difference between the EC tau PET SUVR marginal means in the diabetic (mean = 1.56 SUVR, $$n = 4$$) and nondiabetic (mean = 1.26 SUVR, $$n = 70$$) amyloid PET positive groups was not significant ($t = 1.70$, $$p \leq .09$$). In additional robust regressions using the HC3 estimator and controlling for the same variables as in primary analyses, we found that a diabetic status × amyloid PET positivity interaction was not significantly related to tau PET SUVR in the temporal meta-ROI ($b = .29$, $$p \leq .09$$) and that an HOMA2-IR × amyloid PET positivity interaction was not significantly associated with EC tau PET SUVR (see Supplementary Table 3 for latter results).Fig. 3Estimated marginal means of a entorhinal cortex and b medial temporal lobe tau PET SUVR controlling for age, sex, cohort, and cognitive status. Medial temporal lobe tau PET SUVR = average tau PET SUVR from bilateral entorhinal cortex, hippocampus, and amygdala. Amyloid PET positivity threshold = 1.19 DVR. Error bars represent $95\%$ confidence intervals
## Discussion
HOMA2-IR was not significantly related to EC tau PET SUVR in a nondemented middle-aged and older adult sample enriched for AD risk. Additionally, the relationship between HOMA2-IR and EC tau PET SUVR was not moderated by amyloid PET DVR or amyloid PET positivity status. Similarly, diabetic status was not related to EC tau PET SUVR and the interaction between diabetic status and amyloid PET DVR was also not significantly associated with EC tau PET SUVR. HOMA2-IR and diabetic status, as well as their interaction with amyloid PET DVR, were also not related to tau PET SUVR in secondary ROIs, the MTL, and temporal meta-ROI. Despite the small number in the diabetic amyloid PET positive subset ($$n = 6$$), a significant interaction between diabetic status and amyloid PET positivity status was related to EC and MTL tau PET SUVR. Being diabetic and amyloid PET positive was associated with higher EC and MTL tau PET SUVR. We discuss these findings in the context of current research examining relationships between IR, diabetes, and AD-related pathologic tau.
Our HOMA2-IR findings are congruent with the results of two previous post-mortem studies that found no significant relationship between antemortem HOMA1-IR and Braak score [25, 26]. Although other post-mortem studies [55–57] have described IR [57] or a reduction in the levels of insulin signaling kinases (e.g., PI3K) [55, 56] in the brain tissue of AD cases confirmed to have NFTs and amyloid plaques, it is not known from these studies whether peripheral IR contributed to brain IR in the AD cases observed. A measure of antemortem peripheral IR was not reported, and AD cases determined to have brain IR did not have a known antemortem type 2 diabetes diagnosis [57], suggesting that peripheral IR associated with type 2 diabetes was not essential for the development of brain IR. Whether or under what conditions peripheral IR found in prediabetes or type 2 diabetes is related to brain IR warrants investigation. Results from animal studies investigating a linkage between peripheral and central IR have been mixed [58–61].
Despite any linkage between peripheral and brain IR, it remains unclear if brain IR precedes or facilitates tau aggregation. Some have hypothesized that the facilitation of AD-related tau hyperphosphorylation occurs due to overactivity of glycogen synthase kinase-3 (GSK-3) [62], an insulin signaling kinase whose active form phosphorylates tau. However, in contrast to that hypothesis, post-mortem studies have not found higher total levels of GSK-3 in AD cases relative to controls [55, 57]. Furthermore, levels of the suppressed (i.e., inactive) form of GSK-3 were found to be positively (rather than negatively) related to NFT density [57]. Instead of causing tau aggregation, brain IR could be a consequence. In a cellular model, tau hyperphosphorylation specific for AD was shown to drive intracellular insulin aggregation and subsequent IR, demonstrating that abnormal tau in AD could precede brain IR [63]. Whether peripheral IR is related to brain IR in AD and has an upstream influence on the development of tau pathology requires further investigation.
In addition to causing metabolic dysfunction, IR and type 2 diabetes also have adverse effects upon the vascular endothelium [31, 32] and increase the risk for cardiovascular disease [33, 34]. Results from some studies suggest that a synergistic association between vascular disease risk and higher cortical amyloid is related to greater tau pathology [35, 36]. In contrast to these findings, we did not find that individuals with higher HOMA2-IR values and greater cortical amyloid burden, determined by amyloid PET DVR and positivity status, had higher tau PET SUVR in the EC and secondary ROIs. Results may have been limited by the small proportion ($14.6\%$) of participants with HOMA2-IR greater than the mean value (1.6) of the diabetic participants.
Tau PET SUVR in the EC and secondary ROIs was also not related to an interaction between diabetic status and amyloid PET DVR; however, tau PET SUVR in the EC and MTL was significantly associated with an interaction between diabetic status and amyloid PET positivity status. The diabetic amyloid PET positive group had significantly higher EC tau PET SUVR than the non-diabetic amyloid PET positive group. The difference between mean MTL tau PET SUVR in the diabetic and nondiabetic amyloid PET positive groups was similar in direction but weaker statistically ($$p \leq .05$$). The diabetic × amyloid PET positivity interaction was not significantly related to tau PET SUVR in the temporal meta-ROI ($$p \leq .09$$). Since the EC is one of the first regions to develop NFTs according to Braak pathological staging [50], the difference in findings between the EC and secondary ROIs could have been due to greater tau accumulation in the EC and less extensive spread of tau throughout the MTL and temporal lobe. *In* general, the interpretation of the results is limited by the small number of participants who were both diabetic and amyloid PET positive ($$n = 6$$). Participants who were cognitively impaired likely contributed to the significant association between the diabetic status × amyloid PET positivity interaction and EC tau PET SUVR. Following the removal of cognitively unimpaired participants (including 2 of the 6 with diabetes and amyloid PET positivity), that interaction was similar in direction but was no longer significant. It is also likely that the removal of participants from the sample reduced the statistical power to detect a significant effect. Findings suggest the need for additional study to examine if diabetes, through its diverse effects on the vasculature and cellular metabolism, predisposes neurons to tangle formation when amyloid reaches a certain level.
It is noteworthy that neuropathological studies do not support a relationship between type 2 diabetes and NFT or amyloid plaque pathology [27–30], suggesting that either diabetes is not an instigator of AD pathology or possibly that medications for type 2 diabetes help to mitigate AD pathogenesis. In contrast, diabetes has been consistently related to indicators of cerebrovascular disease, such as microinfarcts, lacunes, or white matter hyperintensities [30]. Nevertheless, population-based studies of large sample sizes regularly find that type 2 diabetes is associated with an increased risk for Alzheimer’s clinical syndrome [64–66]. Similar associations to increased Alzheimer’s clinical syndrome risk have been found for elevated HOMA-IR and hyperinsulinemia (an indicator of IR) [3, 4, 67]. Because the diagnosis of dementia in epidemiological studies has not been complemented for practical reasons by CSF or PET biomarkers of AD pathology, it is possible that some individuals with vascular dementia in these studies were misdiagnosed as having dementia due to AD or reflected the increased presentation of the AD clinical syndrome among individuals with mixed AD and vascular pathology [68]. Meta-analyses of epidemiologic studies have indicated that people with type 2 diabetes have a greater risk for vascular than AD dementia; however, when accounting for cerebrovascular and cardiovascular disease, the risk for clinical AD, although reduced, is still present [64].
Mechanisms that may account for the relationship between type 2 diabetes and Alzheimer’s clinical syndrome are unclear [69]. Whether diabetes interacts with amyloid to influence cognitive decline in AD should be examined. Some studies support a synergistic effect between AD biomarkers and vascular risk or vascular endothelial dysfunction upon cognitive function and decline [35, 70]; however, others provide evidence for additive effects, indicating that vascular risk factors and AD pathology promote cognitive decline through independent pathways [71, 72]. Differences in results could be due to differences in sample composition, especially in the proportion of participants effectively treated to reduce vascular risk. Participants untreated for vascular disease risk factors have been shown to have more AD pathology than treated counterparts [73, 74]; thus, identifying a synergistic association between vascular risk and AD pathology upon cognitive decline may be more likely in samples that contain individuals who have not been treated for vascular disease risk factors.
Our study has several strengths and limitations. The use of the tau PET radioligand 18F-MK-6240 to detect aggregated tau was a strength of this study. 18F-MK-6240 is a second-generation tracer that has demonstrated high affinity and selectivity to AD-type NFTs comprised of PHF tau in post-mortem AD brains and little to no binding to tau aggregates in non-AD tauopathy [75]. Our samples were enriched for AD risk due to increased proportion of APOE ε4 allele carriers, and $23.9\%$ of participants in each sample had a positive amyloid PET. There were no participants in the HOMA-IR sample with dementia, and only $1.5\%$ in the Diabetic Status sample had been diagnosed with dementia. This allowed us to examine the relationship of IR and diabetic status to tau PET SUVR in a sample containing individuals who were in the early stages of the AD clinicopathologic continuum, thus extending results from previous studies that investigated the relationship of IR or diabetic status to NFTs in the post-mortem brain tissue of older adults with and without an antemortem dementia diagnosis [26–28].
Because of the correlational nature of our study, causative claims cannot be made. Because of the small number of participants who were diabetic, we did not have adequate power to detect small effects of diabetic status on tau PET SUVR.
Results should be interpreted within the context of sample characteristics. Both samples were comprised predominantly of well-educated and relatively healthy participants. The proportion of participants with diabetes was lower than the population prevalence in Wisconsin for older adults. There were $7.9\%$ in the HOMA-IR and $9.4\%$ in the Diabetic Status sample who were identified as diabetic. These values are slightly less than the $10.9\%$ prevalence rate for diabetes in middle-aged (45–64 years) adults and lower than the $18.3\%$ and $20.6\%$ prevalence rates, respectively for youngest old (65–74 years) and mid to oldest old adults (> 75 years) in Wisconsin in the year 2018 [76]. It is possible that some participants with undiagnosed diabetes were missed due to the lack of 2-hour post-load glucose or hemoglobin A1c measurements.
Because most diabetics ($73.0\%$) in the Diabetic Status sample reported taking antidiabetic medication, we were unable to investigate if untreated diabetes was related to tau. Although a post-mortem study did not find any significant differences in NFT pathology between treated and untreated diabetics [77], another study of living participants found that the amount of CSF P-tau181 in treated diabetic, prediabetic and nondiabetic groups was similar and significantly lower than the concentration in a group of untreated diabetics [74]. Oral antidiabetic medication has been shown in some but not all studies to reduce the risk of dementia [64]. Whether untreated diabetes and the exacerbation of associated conditions, such as inflammation and oxidative stress, play a role in the development of tau pathology deserve investigation.
Relatedly, we acknowledge that effective blood glucose control and diabetes duration could have influenced results. Longer diabetes duration was related in one study to elevated CSF P-tau-181 [74]. Since we did not have hemoglobin A1c to assess glucose control or data on diabetes duration, we were unable to study whether these factors were related to tau PET SUVR. Research is needed to investigate not only untreated diabetes but also the extent that glucose control and diabetes duration are related to AD-associated hyperphosphorylated and aggregated tau.
The majority ($90.9\%$ to $93.9\%$) of participants in our samples self-identified as White. African Americans, American Indians, and some Hispanic and Asian American ethnic groups have higher rates of diabetes than non-Hispanic White Americans [78, 79]. Although there was a greater proportion of African Americans and other minoritized groups in the diabetic relative to the non-diabetic group in our sample, there were insufficient numbers for stratified analyses. Studies with more diverse participant representation are needed.
In conclusion, HOMA2-IR was not associated with EC tau PET SUVR in a sample enriched for AD risk and comprised predominantly of middle-aged and older adults who were cognitively unimpaired. This finding is in congruence with results from previous post-mortem studies [25, 26] and suggests that IR may not be related to the early presence of tau aggregates in AD. Replication is needed in a sample representing a diversity of racial and ethnic groups and with a larger proportion of diabetic participants. Findings for diabetic status were inconclusive but suggest the need for studies investigating whether a synergistic association between diabetes and amyloid is related to tau. Future studies should also examine the role of diabetic treatment effectiveness and diabetes duration upon the development of tau.
## Supplementary Information
Additional file 1: Supplementary Table 1. Antidiabetic medication usage in the Diabetic Status sample. Data presented are counts (%). Supplementary Table 2. Results from linear regression testing (a) the relationship of IR and diabetic status to entorhinal cortex tau PET SUVR1 and (b) amyloid PET DVR as a moderator of the relationship of HOMA2-IR and diabetic status to entorhinal cortex tau PET SUVR. Supplementary Table 3. Results from robust regression testing amyloid PET positivity status as a moderator of the relationship of diabetic status and HOMA2-IR to tau PET SUVR in the a) entorhinal cortex1 and b) medial temporal lobe2.
## References
1. Lebovitz HE. **Insulin resistance: definition and consequences**. *Exp Clin Endocrinol Diabetes* (2001.0) **109** 135-148. DOI: 10.1055/s-2001-18576
2. Martin BC, Warram JH, Krolewski AS, Soeldner JS, Kahn CR, Martin BC. **Role of glucose and insulin resistance in development of type 2 diabetes mellitus: results of a 25-year follow-up study**. *Lancet* (1992.0) **340** 925-929. DOI: 10.1016/0140-6736(92)92814-V
3. Kuusisto J, Koivisto K, Mykkanen L, Helkala EL, Vanhanen M, Hanninen T. **Association between features of the insulin resistance syndrome and Alzheimer’s disease independently of apolipoprotein e4 phenotype: cross sectional population based study**. *BMJ* (1997.0) **315** 1045-1049. DOI: 10.1136/bmj.315.7115.1045
4. Schrijvers EMC, Witteman JCM, Sijbrands EJG, Hofman A, Koudstaal PJ, Breteler MMB. **Insulin metabolism and the risk of Alzheimer disease: the Rotterdam study**. *Neurology* (2010.0) **75** 1982-1987. DOI: 10.1212/WNL.0b013e3181ffe4f6
5. Orr ME, Sullivan AC, Frost B. **A brief overview of tauopathy: causes, consequences, and therapeutic strategies**. *Trends Pharmacol Sci* (2017.0) **38** 637-648. DOI: 10.1016/j.tips.2017.03.011
6. Kim B, Figueroa-Romero C, Pacut C, Backus C, Feldman EL. **Insulin resistance prevents AMPK-induced Tau dephosphorylation through Akt-mediated increase in AMPKSer-485 phosphorylation**. *J Biol Chem* (2015.0) **290** 19146-19157. DOI: 10.1074/jbc.M115.636852
7. Mehla J, Chauhan BC, Chauhan NB. **experimental induction of type 2 diabetes in aging-accelerated mice triggered alzheimer-like pathology and memory deficits**. *j alzheimer’s dis* (2014.0) **39** 145-162. DOI: 10.3233/JAD-131238
8. Yarchoan M, Toledo JB, Lee EB, Arvanitakis Z, Kazi H, Han LY. **Abnormal serine phosphorylation of insulin receptor substrate 1 is associated with tau pathology in Alzheimer’s disease and tauopathies**. *Acta Neuropathol* (2014.0) **128** 679-689. DOI: 10.1007/s00401-014-1328-5
9. Arendt T, Stieler JT, Holzer M. **Tau and tauopathies**. *Brain Res Bull* (2016.0) **126** 238-292. DOI: 10.1016/j.brainresbull.2016.08.018
10. Biernat J, Mandelkow EM, Schröter C, Lichtenberg-Kraag B, Steiner B, Berling B. **The switch of tau protein to an Alzheimer-like state includes the phosphorylation of two serine-proline motifs upstream of the microtubule binding region**. *EMBO J* (1992.0) **11** 1593-1597. DOI: 10.1002/j.1460-2075.1992.tb05204.x
11. Goedert M, Jakes R, Crowther RA, Cohen P, Vanmechelen E, Vandermeeren M. **Epitope mapping of monoclonal antibodies to the paired helical filaments of Alzheimer’s disease: identification of phosphorylation sites in tau protein**. *Biochemical Journal* (1994.0) **301** 871-877. DOI: 10.1042/bj3010871
12. Crowther RA. **Straight and paired helical filaments in Alzheimer disease have a common structural unit**. *Proc Natl Acad Sci* (1991.0) **88** 2288-2292. DOI: 10.1073/pnas.88.6.2288
13. Laws SM, Gaskin S, Woodfield A, Srikanth V, Bruce D, Fraser PE. **Insulin resistance is associated with reductions in specific cognitive domains and increases in CSF tau in cognitively normal adults**. *Sci Rep* (2017.0) **7** 9766. DOI: 10.1038/s41598-017-09577-4
14. Starks EJ, Patrick O’Grady J, Hoscheidt SM, Racine AM, Carlsson CM, Zetterberg H. **Insulin Resistance is Associated with Higher Cerebrospinal Fluid Tau Levels in Asymptomatic APOE ɛ4 Carriers**. *J Alzheimer’s Dis* (2015.0) **46** 525-533. DOI: 10.3233/JAD-150072
15. Hampel H, Buerger K, Zinkowski R, Teipel SJ, Goernitz A, Andreasen N. **Measurement of Phosphorylated Tau Epitopes in the Differential Diagnosis of Alzheimer Disease**. *Arch Gen Psychiatry* (2004.0) **61** 95-102. DOI: 10.1001/archpsyc.61.1.95
16. Westwood S, Liu B, Baird AL, Anand S, Nevado-Holgado AJ, Newby D. **The influence of insulin resistance on cerebrospinal fluid and plasma biomarkers of Alzheimer’s pathology**. *Alzheimers Res Ther* (2017.0) **9** 31. DOI: 10.1186/s13195-017-0258-6
17. Becker K, Freude S, Zemva J, Stöhr O, Krone W, Schubert M. **Chronic peripheral hyperinsulinemia has no substantial influence on tau phosphorylation in vivo**. *Neurosci Lett* (2012.0) **516** 306-310. DOI: 10.1016/j.neulet.2012.04.022
18. Kahn SE. **The Importance of-Cell Failure in the Development and Progression of Type 2 Diabetes**. *J Clin Endocrinol Metab* (2001.0) **86** 4047-4058. PMID: 11549624
19. Moran C, Beare R, Phan TG, Bruce DG, Callisaya ML, Srikanth V. **Type 2 diabetes mellitus and biomarkers of neurodegeneration**. *Neurology* (2015.0) **85** 1123-1130. DOI: 10.1212/WNL.0000000000001982
20. Groeneveld ON, Moneti C, Heinen R, de Bresser J, Kuijf HJ, Exalto LG. **The Clinical Phenotype of Vascular Cognitive Impairment in Patients with Type 2 Diabetes Mellitus**. *J Alzheimer’s Dis* (2019.0) **68** 311-322. DOI: 10.3233/JAD-180914
21. Mattsson-Carlgren N, Andersson E, Janelidze S, Ossenkoppele R, Insel P, Strandberg O. **Aβ deposition is associated with increases in soluble and phosphorylated tau that precede a positive Tau PET in Alzheimer’s disease**. *Sci Adv* (2020.0) **6** eaaz238. DOI: 10.1126/sciadv.aaz2387
22. Kaeser SA, Häsler LM, Lambert M, Bergmann C, Bottelbergs A, Theunis C. **CSF p-tau increase in response to Aβ-type and Danish-type cerebral amyloidosis and in the absence of neurofibrillary tangles**. *Acta Neuropathol* (2022.0) **143** 287-290. DOI: 10.1007/s00401-021-02400-5
23. Zetterberg H, Bendlin BB. **Biomarkers for Alzheimer’s disease—preparing for a new era of disease-modifying therapies**. *Mol Psychiatry* (2021.0) **26** 296-308. DOI: 10.1038/s41380-020-0721-9
24. Sato C, Barthélemy NR, Mawuenyega KG, Patterson BW, Gordon BA, Jockel-Balsarotti J. **Tau kinetics in neurons and the human central nervous system**. *Neuron* (2018.0) **97** 1284-1298.e7. DOI: 10.1016/j.neuron.2018.02.015
25. Matsuzaki T, Sasaki K, Tanizaki Y, Hata J, Fujimi K, Matsui Y. **Insulin resistance is associated with the pathology of Alzheimer disease: the Hisayama study**. *Neurology* (2010.0) **75** 764-770. DOI: 10.1212/WNL.0b013e3181eee25f
26. Thambisetty M, Metter EJ, Yang A, Dolan H, Marano C, Zonderman AB. **Glucose intolerance, insulin resistance, and pathological features of Alzheimer disease in the baltimore longitudinal study of aging**. *JAMA Neurol* (2013.0) **70** 1167. DOI: 10.1001/jamaneurol.2013.284
27. Ahtiluoto S, Polvikoski T, Peltonen M, Solomon A, Tuomilehto J, Winblad B. **Diabetes, Alzheimer disease, and vascular dementia: a population-based neuropathologic study**. *Neurology* (2010.0) **75** 1195-1202. DOI: 10.1212/WNL.0b013e3181f4d7f8
28. Arvanitakis Z, Schneider JA, Wilson RS, Li Y, Arnold SE, Wang Z. **Diabetes is related to cerebral infarction but not to AD pathology in older persons**. *Neurology* (2006.0) **67** 1960-1965. DOI: 10.1212/01.wnl.0000247053.45483.4e
29. Beeri MS, Silverman JM, Davis KL, Marin D, Grossman HZ, Schmeidler J. **Type 2 Diabetes is negatively associated with Alzheimer’s disease neuropathology**. *J Gerontol A Biol Sci Med Sci* (2005.0) **60** 471-475. DOI: 10.1093/gerona/60.4.471
30. Hadley G, Zhang J, Harris-Skillman E, Alexopoulou Z, DeLuca GC, Pendlebury ST. **Cognitive decline and diabetes: a systematic review of the neuropathological correlates accounting for cognition at death**. *J Neurol Neurosurg Psychiatry* (2022.0) **93** 246-253. DOI: 10.1136/jnnp-2021-328158
31. Kim JA, Montagnani M, Kwang KK, Quon MJ. **Reciprocal relationships between insulin resistance and endothelial dysfunction: molecular and pathophysiological mechanisms**. *Circulation* (2006.0) **113** 1888-1904. DOI: 10.1161/CIRCULATIONAHA.105.563213
32. Schulman IH, Zhou MS. **Vascular insulin resistance: a potential link between cardiovascular and metabolic diseases**. *Curr Hypertens Rep* (2009.0) **11** 48-55. DOI: 10.1007/s11906-009-0010-0
33. Kannel WB, McGee DL. **Diabetes and cardiovascular disease**. *JAMA* (1979.0) **241** 2035-2038. DOI: 10.1001/jama.1979.03290450033020
34. 34.Gast KB, Tjeerdema N, Stijnen T, Smit JWA, Dekkers OM. Insulin resistance and risk of incident cardiovascular events in adults without diabetes: meta-analysis. PLoS One. 2012:7(12):e52036. Available from: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0052036.
35. Albrecht D, Lisette Isenberg A, Stradford J, Monreal T, Sagare A, Pachicano M. **Associations between vascular function and Tau PET are associated with global cognition and amyloid**. *Journal of Neuroscience* (2020.0) **40** 8573-8586. DOI: 10.1523/JNEUROSCI.1230-20.2020
36. Rabin JS, Yang HS, Schultz AP, Hanseeuw BJ, Hedden T, Viswanathan A. **Vascular Risk and β-Amyloid Are Synergistically Associated with Cortical Tau**. *Ann Neurol* (2019.0) **85** 272-279. DOI: 10.1002/ana.25399
37. Bilgel M, Bannerjee A, Shafer A, An Y, Resnick SM. **Vascular risk is not associated with PET measures of Alzheimer’s disease neuropathology among cognitively normal older adults**. *Neuroimage* (2021.0) **1** 100068. DOI: 10.1016/j.ynirp.2021.100068
38. Johnson SC, Koscik RL, Jonaitis EM, Clark LR, Mueller KD, Berman SE. **The Wisconsin Registry for Alzheimer’s prevention: a review of findings and current directions**. *Alzheimer’s Dementia* (2018.0) **10** 130-142. PMID: 29322089
39. 39.Langhough Koscik R, Hermann BP, Allison S, Clark LR, Jonaitis EM, Mueller KD, et al. Validity evidence for the research category, “Cognitively unimpaired – declining,” as a risk marker for mild cognitive impairment and Alzheimer’s disease. Front Aging Neurosci. 2021:13. Article 688478. Available from: https://www.frontiersin.org/articles/10.3389/fnagi.2021.688478/full.
40. Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC. **The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease**. *Alzheimer’s Dementia* (2011.0) **7** 270-279. DOI: 10.1016/j.jalz.2011.03.008
41. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Kawas CH. **The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease**. *Alzheimer’s Dementia* (2011.0) **7** 263-269. DOI: 10.1016/j.jalz.2011.03.005
42. Wallace TM, Levy JC, Matthews DR. **Use and abuse of HOMA modeling**. *Diabetes Care* (2004.0) **27** 1487-1495. DOI: 10.2337/diacare.27.6.1487
43. Sumner AE, Cowie CC. **Ethnic differences in the ability of triglyceride levels to identify insulin resistance**. *Atherosclerosis* (2008.0) **196** 696-703. DOI: 10.1016/j.atherosclerosis.2006.12.018
44. Yeni-Komshian H, Carantoni M, Abbasi F, Reaven GM. **Relationship between several surrogate estimates of insulin resistance and quantification of insulin-mediated glucose disposal in 490 healthy nondiabetic volunteers**. *Diabetes Care* (2000.0) **23** 171-175. DOI: 10.2337/diacare.23.2.171
45. **Standards of medical care in diabetes-2010**. *Diabetes Care* (2010.0) **33** S11-S61. DOI: 10.2337/dc10-S011
46. Betthauser TJ, Cody KA, Zammit MD, Murali D, Converse AK, Barnhart TE. **In vivo characterization and quantification of neurofibrillary tau PET radioligand 18F-MK-6240 in humans from Alzheimer disease dementia to young controls**. *J Nuclear Med* (2019.0) **60** 93-99. DOI: 10.2967/jnumed.118.209650
47. Johnson SC, Christian BT, Okonkwo OC, Oh JM, Harding S, Xu G. **Amyloid burden and neural function in people at risk for Alzheimer’s Disease**. *Neurobiol Aging* (2014.0) **35** 576-584. DOI: 10.1016/j.neurobiolaging.2013.09.028
48. Sprecher KE, Bendlin BB, Racine AM, Okonkwo OC, Christian BT, Koscik RL. **Amyloid burden is associated with self-reported sleep in nondemented late middle-aged adults**. *Neurobiol Aging* (2015.0) **36** 2568-2576. DOI: 10.1016/j.neurobiolaging.2015.05.004
49. Racine AM, Clark LR, Berman SE, Koscik RL, Mueller KD, Norton D. **Associations between performance on an abbreviated CogState battery, other measures of cognitive function, and biomarkers in people at risk for Alzheimer’s disease**. *J Alzheimer’s Dis* (2016.0) **54** 1395-1408. DOI: 10.3233/JAD-160528
50. Braak H, Alafuzoff I, Arzberger T, Kretzschmar H, Tredici K. **Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry**. *Acta Neuropathol* (2006.0) **112** 389-404. DOI: 10.1007/s00401-006-0127-z
51. Betthauser TJ, Koscik RL, Jonaitis EM, Allison SL, Cody KA, Erickson CM. **Amyloid and tau imaging biomarkers explain cognitive decline from late middle-age**. *Brain* (2020.0) **143** 320-335. DOI: 10.1093/brain/awz378
52. Sperling RA, Mormino EC, Schultz AP, Betensky RA, Papp K v, Amariglio RE. **The impact of amyloid-beta and tau on prospective cognitive decline in older individuals**. *Ann Neurol* (2019.0) **85** 181-193. PMID: 30549303
53. Jack CR, Wiste HJ, Weigand SD, Therneau TM, Lowe VJ, Knopman DS. **Defining imaging biomarker cut points for brain aging and Alzheimer’s disease**. *Alzheimer’s Dementia* (2017.0) **13** 205-216. DOI: 10.1016/j.jalz.2016.08.005
54. Hayes AF, Cai LI. **Using heteroskedasticity-consistent standard error estimators in OLS regression: An introduction and software implementation**. *Behav Res Methods* (2007.0) **39** 709-722. DOI: 10.3758/BF03192961
55. Liu Y, Liu F, Grundke-Iqbal I, Iqbal K, Gong CX. **Deficient brain insulin signalling pathway in Alzheimer’s disease and diabetes**. *J Pathol* (2011.0) **225** 54-62. DOI: 10.1002/path.2912
56. Moloney AM, Griffin RJ, Timmons S, O’Connor R, Ravid R, O’Neill C. **Defects in IGF-1 receptor, insulin receptor and IRS-1/2 in Alzheimer’s disease indicate possible resistance to IGF-1 and insulin signalling**. *Neurobiol Aging* (2010.0) **31** 224-243. DOI: 10.1016/j.neurobiolaging.2008.04.002
57. Talbot K, Wang HY, Kazi H, Han LY, Bakshi KP, Stucky A. **Demonstrated brain insulin resistance in Alzheimer’s disease patients is associated with IGF-1 resistance, IRS-1 dysregulation, and cognitive decline**. *J Clin Investig* (2012.0) **122** 1316-1338. DOI: 10.1172/JCI59903
58. Arnold SE, Lucki I, Brookshire BR, Carlson GC, Browne CA, Kazi H. **High fat diet produces brain insulin resistance, synaptodendritic abnormalities and altered behavior in mice**. *Neurobiol Dis* (2014.0) **67** 79-87. DOI: 10.1016/j.nbd.2014.03.011
59. Liu Z, Patil IY, Jiang T, Sancheti H, Walsh JP, Stiles BL. **High-Fat diet induces hepatic insulin resistance and impairment of synaptic plasticity**. *PLoS One* (2015.0) **10** e0128274. DOI: 10.1371/journal.pone.0128274
60. Salas IH, Weerasekera A, Ahmed T, Callaerts-Vegh Z, Himmelreich U, D’Hooge R. **High fat diet treatment impairs hippocampal long-term potentiation without alterations of the core neuropathological features of Alzheimer disease**. *Neurobiol Dis* (2018.0) **113** 82-96. DOI: 10.1016/j.nbd.2018.02.001
61. Stanley M, Macauley SL, Caesar EE, Koscal LJ, Moritz W, Robinson GO. **The effects of peripheral and central high insulin on brain insulin signaling and amyloid-β in young and old APP/PS1 mice**. *J Neurosci* (2016.0) **36** 11704-11715. DOI: 10.1523/JNEUROSCI.2119-16.2016
62. Hooper C, Killick R, Lovestone S. **The GSK3 hypothesis of Alzheimer’s disease**. *J Neurochem* (2008.0) **104** 1433-1439. DOI: 10.1111/j.1471-4159.2007.05194.x
63. Rodriguez-Rodriguez P, Sandebring-Matton A, Merino-Serrais P, Parrado-Fernandez C, Rabano A, Winblad B. **Tau hyperphosphorylation induces oligomeric insulin accumulation and insulin resistance in neurons**. *Brain* (2017.0) **140** 3269-3285. DOI: 10.1093/brain/awx256
64. Xue M, Xu W, Ou YN, Cao XP, Tan MS, Tan L. **Diabetes mellitus and risks of cognitive impairment and dementia: a systematic review and meta-analysis of 144 prospective studies**. *Ageing Res Rev* (2019.0) **55** 100944. DOI: 10.1016/j.arr.2019.100944
65. Chatterjee S, Peters SAE, Woodward M, Arango SM, Batty GD, Beckett N. **Type 2 diabetes as a risk factor for dementia in women compared with men: A pooled analysis of 2.3 million people comprising more than 100,000 cases of dementia**. *Diabetes Care* (2016.0) **39** 300-307. DOI: 10.2337/dc15-1588
66. Peila R, Rodriguez BL, Launer LJ. **Type 2 diabetes, APOE gene, and the risk for dementia and related pathologies**. *Diabetes* (2002.0) **51** 1256-1262. DOI: 10.2337/diabetes.51.4.1256
67. Peila R, Rodriguez BL, White LR, Launer LJ. **Fasting insulin and incident dementia in an elderly population of Japanese-American men**. *Neurology* (2004.0) **63** 228-233. DOI: 10.1212/01.WNL.0000129989.28404.9B
68. Beach TG, Monsell SE, Phillips LE, Kukull W. **Accuracy of the Clinical Diagnosis of Alzheimer Disease at National Institute on Aging Alzheimer Disease Centers, 2005-2010**. *J Neuropathol Exp Neurol* (2012.0) **71** 266-273. DOI: 10.1097/NEN.0b013e31824b211b
69. Biessels GJ, Despa F. **Cognitive decline and dementia in diabetes mellitus: mechanisms and clinical implications**. *Nat Rev Endocrinol* (2018.0) **14** 591-604. DOI: 10.1038/s41574-018-0048-7
70. Clark LR, Koscik RL, Allison SL, Berman SE, Norton D, Carlsson CM. **Hypertension and obesity moderate the relationship between β-amyloid and cognitive decline in midlife**. *Alzheimer’s Dementia* (2019.0) **15** 418-428. DOI: 10.1016/j.jalz.2018.09.008
71. Pettigrew C, Soldan A, Wang J, Wang MC, Arthur K, Moghekar A. **Association of midlife vascular risk and AD biomarkers with subsequent cognitive decline**. *Neurology* (2020.0) **95** e3093-e3103. DOI: 10.1212/WNL.0000000000010946
72. Vemuri P, Lesnick TG, Przybelski SA, Knopman DS, Preboske GM, Kantarci K. **Vascular and amyloid pathologies are independent predictors of cognitive decline in normal elderly**. *Brain* (2015.0) **138** 761-771. DOI: 10.1093/brain/awu393
73. 73.Köbe T, Gonneaud J, Pichet Binette A, Meyer PF, McSweeney M, Rosa-Neto P, et al. Association of Vascular Risk Factors with β-Amyloid Peptide and Tau Burdens in Cognitively Unimpaired Individuals and Its Interaction with Vascular Medication Use. JAMA Netw Open. 2020;3(2):e1920780. Available from: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2760441.
74. McIntosh EC, Nation DA. **Importance of treatment status in links between type 2 diabetes and Alzheimer’s disease**. *Diabetes Care* (2019.0) **42** 972-979. DOI: 10.2337/dc18-1399
75. Aguero C, Dhaynaut M, Normandin MD, Amaral AC, Guehl NJ, Neelamegam R. **Autoradiography validation of novel tau PET tracer [F-18]-MK-6240 on human postmortem brain tissue**. *Acta Neuropathol Commun* (2019.0) **7** 37. DOI: 10.1186/s40478-019-0686-6
76. 76.CDC. United States Diabetes Surveillance System. https://gis.cdc.gov/grasp/diabetes/DiabetesAtlas.html2019;
77. Beeri MS, Schmeidler J, Silverman JM, Gandy S, Wysocki M, Hannigan CM. **Insulin in combination with other diabetes medication is associated with less Alzheimer neuropathology**. *Neurology* (2008.0) **71** 750-757. DOI: 10.1212/01.wnl.0000324925.95210.6d
78. Bullock A, Sheff K, Hora I, Burrows NR, Benoit SR, Saydah SH. **Prevalence of diagnosed diabetes in American Indian and Alaska Native adults, 2006–2017**. *BMJ Open Diabetes Res Care* (2020.0) **8** e001218. DOI: 10.1136/bmjdrc-2020-001218
79. Cheng YJ, Kanaya AM, Araneta MRG, Saydah SH, Kahn HS, Gregg EW. **Prevalence of diabetes by race and ethnicity in the United States, 2011-2016**. *JAMA* (2019.0) **322** 2389. DOI: 10.1001/jama.2019.19365
|
---
title: 'Cardiovascular health promotion: A systematic review involving effectiveness
of faith-based institutions in facilitating maintenance of normal blood pressure'
authors:
- Abayomi Sanusi
- Helen Elsey
- Su Golder
- Osayuwamen Sanusi
- Adejoke Oluyase
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022319
doi: 10.1371/journal.pgph.0001496
license: CC BY 4.0
---
# Cardiovascular health promotion: A systematic review involving effectiveness of faith-based institutions in facilitating maintenance of normal blood pressure
## Abstract
Globally, faith institutions have a range of beneficial social utility, but a lack of understanding remains regarding their role in cardiovascular health promotion, particularly for hypertension. Our objective was assessment of modalities, mechanisms and effectiveness of hypertension health promotion and education delivered through faith institutions. A result-based convergent mixed methods review was conducted with 24 databases including MEDLINE, Embase and grey literature sources searched on 30 March 2021, results independently screened by three researchers, and data extracted based on behaviour change theories. Quality assessment tools were selected by study design, from Cochrane risk of bias, ROBINS I and E, and The Joanna Briggs Institute’s Qualitative Assessment and Review Instrument tools. Twenty-four publications contributed data. Faith institution roles include cardiovascular health/disease teaching with direct lifestyle linking, and teaching/ encouragement of personal psychological control. Also included were facilitation of: exercise/physical activity as part of normal lifestyle, nutrition change for cardiovascular health, cardiovascular health measurements, and opportunistic blood pressure checks. These demand relationships of trust with local leadership, contextualisation to local sociocultural realities, volitional participation but prior consent by faith / community leaders. Limited evidence for effectiveness: significant mean SBP reduction of 2.98 mmHg ($95\%$CI -4.39 to -1.57), non-significant mean DBP increase of 0.14 mmHg ($95\%$CI -2.74 to +3.01) three months after interventions; and significant mean SBP reduction of 0.65 mmHg ($95\%$CI -0.91 to -0.39), non-significant mean DBP reduction of 0.53 mmHg ($95\%$CI -1.86 to 0.80) twelve months after interventions. Body weight, waist circumference and multiple outcomes beneficially reduced for cardiovascular health: significant mean weight reduction 0.83kg ($95\%$ CI -1.19 to -0.46), and non-significant mean waist circumference reduction 1.48cm ($95\%$ CI -3.96 to +1.00). In addressing the global hypertension epidemic the cardiovascular health promotion roles of faith institutions probably hold unrealised potential. Deliberate cultural awareness, intervention contextualisation, immersive involvement of faith leaders and alignment with religious practice characterise their deployment as healthcare assets.
## Introduction
Hypertension, the systolic/diastolic blood pressure of $\frac{140}{90}$ mmHg or higher [1–3], is the most important risk factor for cardiovascular morbidity and mortality; and a progressively worsening epidemic in Sub Saharan Africa (SSA) [4–10]. Large hypertension outcome inequalities exist between countries with organised, well-functioning healthcare and public health systems, and populations of SSA with less well organised and funded public health systems [11].
Faith institutions are non-profit entities characterised by expressions of religious creed and implementation of religious worship [12, 13]. They comprise social and cultural networks based on religious traditions or worship practices, and may be involved in providing social services for people within their networks or the wider society. Faith institutions addressed in this review are limited to religious organisations operating worship assembly and social services, not including faith-based higher education institutions and healthcare facilities.
Globally faith institutions are culturally and socio-politically influential [14–17], occupying a peculiar position of influence that potentially frames them as healthcare assets [18–23]. They are part of the structure of societies, and are seen as socially acceptable to contribute to health education and risk reduction campaigns [24–27].
From research and housing to healthcare, there is evidence of involvement and influence of faith-based institutions in a wide range of socially beneficial programmes [19, 28–38]. Knowledge gaps however remain on their role in cardiovascular risk reduction. Specifically, how faith institutions have contributed to hypertension health promotion and facilitation of hypertension screening. Similarly, evidence is needed to understand the features likely to make faith-institution based hypertension interventions acceptable and contextually optimal- particularly for low resource settings and settings without well organised healthcare and public health systems.
The following questions are therefore addressed: *What is* the evidence for the role of faith-institutions in cardiovascular health promotion to reduce hypertension or maintain normal blood pressure? What are the characteristics of those roles?
The objectives are to:
## Methods
The review was registered on the National Institute for Health Research PROSPERO database (CRD42021228938). In accordance with the recommended systematic review process, the protocol was followed and no changes were made. For quantitative synthesis, random effects model meta-analyses were performed using all available primary and secondary outcomes data from the randomised controlled trials. All the randomised controlled studies that had the required data were included in the meta-analyses regardless of risk of bias assessment. Given the relative dearth of studies coupled with clinical and methodological heterogeneity between available ones, attempts such as network meta-analysis or combination of data from randomised studies, nonrandomised studies, before-and-after studies, and cohort studies, were not made. For qualitative synthesis the Joanna Briggs Institute’s integrative meta-aggregation approach was utilised, framing findings on the Diffusion of Innovation theory and the Communication–Behaviour change model. To complete the mixed methods approach using the results based convergent design, a final synthesis was performed of the quantitative and qualitative findings [39].
## Eligibility
All literature addressing hypertension risk reduction targeting adults within faith institutions were eligible. These included literature on hypertension health promotion, hypertension information or education, hypertension risk modification involving weight reduction, increasing physical activity, adoption of healthier diets, hypertension monitoring and engagement with healthcare system. Studies reporting prevention, monitoring, interventions and other activities that are based exclusively on faith, faith-related practices, rituals, activities inexplicable by or incompatible with medical science, or controversial practices were excluded. Studies focussing on yoga were excluded because yoga is also sometimes regarded a physical and spiritual practice, rather a means of providing health information or education for behaviour change [40–43].
Study designs eligible include randomised controlled and non-controlled studies, nonrandomised studies including before and after studies, observational studies, qualitative studies, surveys and reports. The inclusion and exclusion criteria of studies are outlined in Table 1.
**Table 1**
| Unnamed: 0 | Eligible studies for inclusion | Ineligible studies |
| --- | --- | --- |
| P (Population) | • Studies carried out within or in collaboration with a faith-based institution or groups of faith-based institutions • Studies including interventions aimed at adult participants (with adults described as individuals aged at least 16 or 18, depending on the setting) | • Studies exclusively reporting interventions on and data from children or non-adult participants. |
| I (intervention) | • Studies reporting on at least one hypertension prevention, monitoring or treatment activity • Studies including interventions or hypertension related activities that are strictly conventional, universal and based on modern health/medical science | • Studies not reporting on any hypertension intervention or hypertension risk factor intervention • Studies reporting on ritual, spiritual, unscientific or unverifiable activities, or activities with no evidence base in blood pressure reduction. • Studies reporting prevention, monitoring, interventions and other activities that are based exclusively on faith, faith-related practices, rituals, activities inexplicable by or incompatible with health/medical science, or controversial practices (e.g. Yoga, Meditation practices etc.). • Studies incorporating unclassifiable, unstated or unclear interventions. |
| C (Comparison) | There are no comparators. | |
| O (Outcome) | Primary: • Reduction of adult blood pressure measurements from hypertensive to normal blood pressure levels, or maintenance of normal blood pressure.Secondary: • Hypertension risk modifying outcomes including weight reduction, increased physical activity, adoption of healthier diets, increased hypertension awareness, increased hypertension monitoring, increased engagement with healthcare system. • Process indicators of the interventions including intervention acceptability, intervention uptake and participant satisfaction. | |
## Search methods
The following information sources were searched in March 2021, and the results indicating the available literature from their inception to end of March 2021: The Cochrane Library, Epistemonikos, Campbell Library, The International Initiative for Impact Evaluation (3IE), Database of Promoting Health Effectiveness Reviews (DoPHER), Database of Abstracts of Reviews of Effects (DARE), and the Health Technology Assessment (HTA) Database.
Others were MEDLINE, EMBASE, PsycINFO, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Allied and Complementary Medicine Database (AMED), Oxford Bibliographies on Public Health, Scopus, Trials Register of Promoting Health Interventions (TRoPHI), Applied Social Sciences Index & Abstracts (ASSIA), and BiblioMap—The EPPI-Centre database of health promotion research. For grey literature the following databases were searched: The Bielefield Academic Search Engine- BASE, The King’s Fund Library Database, OpenGrey multidisciplinary European database, Open Directory of Open Access Repositories–OpenDOAR, and the National Institute for Health Research—NIHR website and the British Library Catalogue. The British Library E-theses online service database (EThOS) and the ProQuest Dissertations and Theses database were searched. Reference checking was performed on all publications screened eligible for inclusion.
The search strategy combined descriptors for the study population, intervention of interest and the settings are outlined in Tables 2 and 3. The retrieved search results were imported into EndNote and de-duplicated.
## Screening
Title and abstracts were triple screened independently by three reviewers (AS, AO, OS) using Rayyan [44]. Uncertainties and disagreements were resolved by discussion. A fourth reviewer (SG) was available where disagreements could not be resolved. Following screening, two (AS, OS) independently reviewed the full text of the selected publications.
## Data extraction
Using a Microsoft word form (independently made by AS and then checked by OS), data extracted included: author’s name, publication year, country of intervention, type of faith institution, funding (state/private sector/non-governmental/faith institution), physical location of interventions (limited to faith institution/extends outside faith-institution), intervention scope (hypertension focussed/ broad cardiovascular disease/ broad chronic disease), intention (one-off research, cyclical/periodical, permanent programme), basis (voluntary/ subtle coercion), mechanism of action of the intervention, intervention target level (individual, community or population), agents delivering interventions, frequency of interventions, materials used in intervention delivery, religious component of interventions, actual intervention or combination of interventions (publicity/information campaign, patient education, measurement, monitoring, advice, healthcare system interfacing/referral, risk factor management/non pharmaceutical treatment, pharmaceutical treatment), and data on unforeseen implementation challenges.
## Risk of bias assessment
AS independently conducted the risk of bias assessments and OS checked the assessments. The Cochrane Risk of Bias tool was used for randomised studies [45], ROBINS I (Risk Of Bias In Non-randomised Studies—Interventions) tool for non-randomised studies [46] and The ROBINS–E tool (Risk Of Bias In Non-randomised Studies—Exposures) [47] for observational studies. The Joanna Briggs Institute’s Qualitative Assessment and Review Instrument (QARI) was used to assess the risk of bias of qualitative studies [48, 49] and the Mixed Methods Appraisal tool for mixed methods studies [50].
The qualitative studies received a high overall quality rating (Table 5). Of the 6 randomised studies, two were rated low risk of bias, 3 with some concerns, and one high bias risk. Three nonrandomised studies were rated low risk and 12 rated moderate risk of bias. One [59] was adjudged with serious risk of bias due to potential confounding factors, although four sub-domains were low risk and two sub-domains moderate risk. The confounding element of nonrandomised studies was the most frequently adjudged at serious risk—perhaps a reflection of the multiple, complex and interdependent nature of cardiovascular risks (Figs 2A, 2B, 3A and 3B).
**Fig 2:** *Bias assessment of included randomised studies.* **Fig 3:** *Bias assessment of the included non-randomised studies.* TABLE_PLACEHOLDER:Table 5
## Data synthesis
Syntheses were conducted by a single reviewer (AS), and crosschecked by a second (OS). The characteristics of included studies, key characteristics of interventions, and categories of the identified roles of faith institutions are presented. The cardiovascular health promotion and hypertension screening activities described in the included studies as faith institution facilitated roles were categorised.
To synthesise data on characteristics of the roles, the integrative meta-aggregation approach by the Joanna Briggs Institute was undertaken manually using Microsoft word [49]. From the included studies, findings on predefined elements based on the Diffusion of Innovation theory, the Communication–Behaviour change model, and the Template for Intervention Description and Replication (TIDieR) Checklist were collated and combined. From the combined findings, categories were generated on the basis of similarity and meaning of their content. Finally, the categories were combined to generate synthesis statements that integrate the evidence contained in or expressed by the categories.
Evidence of effectiveness of faith institution roles was presented as summary estimates of individual outcome measures where possible. Meta analysis was completed using version 5.4 of Cochrane software Review Manager [51].
Using tables, summary non-aggregated overview of direction of blood pressure change is presented to show indications of the general direction of change of the systolic and diastolic blood pressures. For the multiple distinct secondary outcomes that have an impact on hypertension and cardiovascular health but could not be summarised or aggregated, a tabular representation is presented of their impact.
The final synthesis is presented in tabular form.
## Results
Searches retrieved 10 448 records which after de-duplication reduced to 7 679 (Fig 1). After screening by titles 329 records remained and 23 remained after screening by abstract. Out of these, three were unavailable: a PhD thesis under embargo [52], and two abstract publications [53, 54]. This left 20 publications available for inclusion in the review. Reference checking the included studies within these 20 publications retrieved an additional four records, bringing the total number of included publications to twenty-four.
**Fig 1:** *Preferred reporting items for systematic review and meta-analysis flow diagram showing the process of study selection.*
## Characteristics of the included studies
Each study incorporated data from single countries, except one [55] with data from multiple countries. Twenty were set in the United States of America and one study each was set in China, Norway and South Africa [56–58].
Of the 24 studies, 16 were non-randomised, six were randomised controlled studies and two were qualitative studies. All the randomised controlled studies were conducted in the USA. Of the nonrandomised studies, eight were before and after intervention studies, five were uncontrolled longitudinal studies, two were cohort in design and one was a non-randomised controlled study (Table 4).
**Table 4**
| Included Studies | Details of interventions | Details of interventions.1 | Roles covered by intervention activities | Roles covered by intervention activities.1 | Component mechanisms of the intervention | Evidence of effectiveness | Quality Rating |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Author, Year Country (WHO region) Study design | Name of intervention Setting Funding Scope Intention Number of participants | Summary description of the intervention / programme in the study | Promotion & Education | Screening | Component mechanisms of the intervention | Evidence of effectiveness | Quality Rating |
| Tucker et al., 2019 USA (WHO Americas region) Randomised Controlled Trial | Health-Smart AME interventionChurch/ Church premisesNon Profit OrganisationHypertension risk and chronic diseasesImprove health literacy; Increase health promoting behaviour; Reduce weight; Reduce Blood Pressure172 intervention, 149 control | Church leaders and members, supervised by healthcare professionals, delivered individual and interpersonal level health coaching and physical activity to urban Black / African American adults across all income levels. | ✓ | | Combination of Exercise / Physical Activity with Healthy Lifestyle Coaching, Counselling & Motivation Training | Systolic Blood Pressure (SBP) decrease (2.91mmHg, F value = 2.48, p = 0.117–v–Control: reduction of 1.83 mmHg, F = 1.06, p = 0.303).Diastolic Blood Pressure (DBP) decrease (0.3mmHg, F value = 0.06, p = 0.815 –v–Control: reduction 2.17 mmHg, F = 3.18, p = 0.076).Body Weight decrease (1.69 Lbs., F = 2.95, p = 0.087 –v–Control: reduction of 0.97 Lbs., F value = 1.07, p = 0.303).Nutrition Label Literacy increase (1.2 units, F value = 30.89, p<0.001 –v–Control group non-significant decrease of 0.06 units, F value = 0.09, p = 0.76)Healthy Eating score increase (0.28, F = 26.32, p<0.001 –v–control group score of 0.08, F value = 2.69, p = 0.103).Healthy Drinking score increase (0.88, F = 18.75, p<0.001 –v–control group score of 0.21, F value = 1.40, p = 0.239).Physical Activity score increase (0.30, F = 20.87, p<0.001 –v–control group score of 0.19, F value = 10.95, p<0.01).Overall level of engagement in health- smart behaviors increase (0.76, F = 26.47, p < .001 –v–Control group increase of 0.30, F value = 5.33, p = 0.022). | Overall Low Risk of Bias |
| Sternberg et al., 2007 Multiple Countries, not fully listed Qualitative study | Sternberg et al.,Church/ church facilitated settingsCombination of government and secular bodiesCardiovascular health riskMotivate adoption and maintenance of healthy behaviors. | Faith-based social groups, families, peers from religious community delivered individual, interpersonal and community level education and encouragement linking physical health to spiritual health, to adults from multiple countries. | ✓ | | Indirect intervention consisting the social faith environment, religiosity and religious rituals | Behaviour modification relies on faith related factors including the spiritual and socio-cultural awareness of agents; making members of the faith community potentially effective agents. | Overall high quality rating |
| Schoenthaler et al., 2018 USA (WHO Americas region) Randomised Controlled Trial | TLC-MINT (Therapeutic Lifestyle change- plus Motivational Interviewing)Church/ Church premisesGovernment Agency fundedHypertensionHypertension reduction(373) 172 intervention, 201 control | Trained professional researchers and church members trained as lay health advisors delivered individual and interpersonal level group counseling, therapeutic lifestyle training and motivational interviewing, to urban Black / African American adults across all income levels. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | At 3 months:Mean Arterial Pressure (MAP) decrease (−8.5 mmHg 95%CI (−9.9 to −7.1)–V–control MAP decrease −7.2 95%CI (−7.8 to −6.6)).Systolic Blood Pressure (SBP) decrease (−13.2 in mmHg 95%CI (−14.6 to −11.8)–V–control group decrease of −10.1 95%CI (−10.7 to −9.5).Diastolic Blood Pressure (DBP) decrease −6.1mmHg 95%CI (−6.7 to −5.6)–V–control group decrease of −6.8 95%CI (−7.4 to −6.1).At 6 months:MAP decrease (−9.4 mmHg 95%CI (−11.4 to −7.4)–V–control group decrease of −7.3 95%CI (−8.2 to −6.5).SBP decrease (−14.6 mmHg 95%CI (−19.1 to −10.1)–V–control group decrease of −9.1 95%CI (−10.9 to −7.2).SBP decrease (−5.7 mmHg 95% CI (−6.0 to −5.5)–V–control group decrease of −6.4 95%CI (−6.7 to −6.2). | Overall low risk of Bias |
| Liu et al., 2018 China (WHO Western Pacific region) Cohort study | Routine Buddhist monk religious practiceBuddhist AcademyGovernmental agency fundedAssociation: Tibetan monk religious practice——594 participants | Experimental hypertension screening among Buddhist monks within the context of routine sedentary religious practice. Demonstration of the decreased odds for hypertension. | | ✓ | Indirect intervention consisting the social faith environment, religiosity and religious rituals | Long hours of communal religious rituals and teaching in Tibetan Buddhist settings are associated with a decrease in odds for hypertension. | Overall moderate risk of bias |
| Sørensen et al., 2011 Norway (WHO European Region) Cohort study | The HUNT StudyChurch/churchesGovernment agency fundedHypertensionReligious attendance as hypertension intervention35,964 individuals | Church clergy, social environment within churches and religious attendance as individual level intervention targeting Norwegian adults. | | ✓ | Indirect intervention consisting the social faith environment, religiosity and religious rituals | Mean Systolic Blood Pressure (SBP) & mean Diastolic Blood Pressure (DBP) decreased with increasing Religious Attendance.Bivariate Associations between Religious Attendance and DBP (mmHg) showed mean DBP (and SD) of 71.1 (11.1), 70.9 (10.9), 71.9 (11.5) and 70.9 (11.7) for never attenders, 1–6 time per 6 month attenders, 1–3 times per month attenders and more than 3 times per month attending adults. p = 0.002.Religious Attendance and SBP (mmHg) showed mean SBP (and SD) of 127.6 (19.5), 127.8 (19.7), 134.5 (21.7) and 131.3 (21.2) for never attenders, 1–6 time per 6 month attenders, 1–3 times per month attenders and more than 3 times per month attending adults. p < 0.001. | Overall low risk of bias |
| Kinard, 2016 USA (WHO Americas region) Controlled before- and after study | The My Sister’s Keeper Project ChurchChurch/churchesGovernment agency fundingHypertension and body weightReduction of body weight and blood pressure.(19) 4 intervention, 15 control | Theologically trained ministers and trained health coaches delivered individual and interpersonal level spiritually based nutrition and physical activity education, bible verses and prayers to urban Black / African American adults across all income groups. | ✓ | | Combination of Nutrition & Exercise / Physical Activity | Mean Systolic Blood Pressure (SBP) decrease (0.66 mmHg, Z value = 0.00, p = 0.999,–V–control group mean reduction of 3.0 mm, Z value = -0.734, p = 0.463).No change in mean Diastolic Blood Pressure (DBP) (0mmHg, Z = 0.00, p = 0.999 –V–control group reduction of 4.71 mmHg, Z value = -1.185, p = 0.236).Mean Body Weight increase of 1.0 lbs, Z value = -.816, p = .414, -V- control group mean weight reduction of 1.3 lbs, Z value = -.676, p = 0.499).Z value = Z value in Wilcoxon text | Overall moderate risk of bias |
| Abbot, 2015 USA (WHO Americas region) Controlled before- and after study | With Every Heartbeat is LifeChurch/churchesGovernment agency fundingCardiovascular riskIncrease cardiovascular health knowledge(229) 114 intervention, 115 control | Trained health educators delivered individual and interpersonal level health promotion to rural Black / African American adults. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Greater improvements in cardiovascular health habits (p < .01) and health knowledge (p < .01) compared to the control group on both analyses of repeated measures and gain scores.The intervention positively affected intentions to consume more fruits and vegetables (p = 0.01), reduce dietary fat (p = 0.01), but did not the intentions to increase exercise.The intervention positively affected the norms and attitudes of the participants on consuming more produce (p = 0.01)), reducing dietary fat (p = 0.04), but had no effect on their attitudes and norms on participating in exercising.The intervention had positive effects on participants in increasing their perceived behavioural control/enhancing their self-efficacy (p < 0.01) for increasing their fruit and vegetable intake (p < 0.01), reducing their dietary fat intake (p = 0.03), and increasing their exercise (p < 0.01). | Overall moderate risk of bias |
| Taylor, 2011 USA (WHO Americas region) Case series (uncontrolled longitudinal) study | Way of Faith (WOF) projectChurch/churchesNon Profit Organisation FundingCardiovascular diseaseHealth promotion to reduce obesity, hypertension and high blood sugar15 participants | Qualified wellness counsellors, nurses and clergy delivered individual, interpersonal and community level eating habits and exercise workshops to urban adults of all ethnicities. | ✓ | | Combination of Nutrition & Exercise / Physical Activity | No significant change, with 1.7 mmHg Mean DBP decrease from 86.3 to 84.6 mmHg.Mean pulse rate increase of 3.5 beats per minute from 70.9 to 74.4.Mean weight increase of 1.18 kg from 146.83 to 148.02.Mean blood sugar 5.72 mg/dL increase from 100.57 to 106.29.At 6 months, significant change in attitudes and behavioral understanding: 80% of participants had started exercising, a 60% increase. 90% of participants reported a change of their eating habits. 100% of participants reported greater responsibility in taking care of their bodies. | Overall moderate risk of bias |
| Daye, 2019 USA (WHO Americas region) Controlled before and after study | Faith-based health devotionalChurches/private residencesNon Profit Organisation FundingHypertensionIncrease knowledge on hypertension and hypertension prevention100 participants | Implementation of faith-based health devotional on rural and urban Black / African American adult attenders of faith institutions. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Increase in general knowledge about high blood pressure and its prevention, indicated in statistically significant increase from 8.28 to 9.09 of mean High Blood Pressure Prevention IQ Quiz- 0.81, 95% CI (1.02–0.60), df = 99, p < .000.Non-significant increased understanding of the negative consequences of poor blood pressure control indicated by 3.25 to 3.40 change, a 0.15 increase in Consequences scores 0.15, 95% CI (0.002–0.298), df = 99, p < 0.24.Significant increase in perceived personal control of blood pressure indicated by 4.11 to 4.30 change, a 0.19 increase in Personal Control scores 0.19, 95% CI (0.079–0.311), df = 99, p < 0.0005. | Overall moderate risk of bias |
| Dodani et al., 2014 USA (WHO Americas region) Prospective uncontrolled longitudinal study | Healthy Eating and Living Spiritually (HEALS)Church/churchesFunded by a publicly funded universityHypertensionBlood Pressure reduction34 participants | Pastors and trained church health advisors delivered at the individual and interpersonal level, socio-culturally informed dietary modification, increased physical activity and healthy behavioral change to rural and urban Black / African American adults across all income levels | ✓ | | Combination of Nutrition & Exercise / Physical Activity | Reduction in Mean Systolic Blood Pressure (SBP) of 22 mmHg (p < 0.001).Reduction in Mean Diastolic Blood Pressure (DBP) of 6.5 mmHg (p = 0.0048)A mean weight reduction of 3.11 kg (p < 0.0001). | Overall moderate risk of bias |
| Dodani et al., 2013 USA (WHO Americas region) Prospective uncontrolled longitudinal study | Healthy Eating and Living Spiritually (HEALS)Church/churchesFunded by a publicly funded universityRisk of stroke and hypertensionBlood Pressure reduction31 participants | Pastors and trained church health advisors delivered at the individual and interpersonal level, socio-culturally informed dietary modification, increased physical activity and healthy behavioral change to rural and urban Black / African American adults across all income levels | ✓ | | Combination of Nutrition & Exercise / Physical Activity | Reduction in Mean Systolic Blood Pressure (SBP) of 13.64 mmHg (p = 0.005).Mean Diastolic Blood Pressure (DBP) of 6.12 mmHg (p = 0.01). | Overall moderate risk of bias |
| Dodani et al., 2015 USA (WHO Americas region) Prospective uncontrolled longitudinal study | Healthy Eating and Living Spiritually (HEALS)Church/churchesFunded by a publicly funded universityRisk of stroke and hypertensionBlood Pressure reduction36 participants | Pastors and trained church health advisors delivered at the individual and interpersonal level, socio-culturally informed dietary modification, increased physical activity and healthy behavioral change to rural and urban Black / African American adults across all income levels. | ✓ | | Combination of Nutrition & Exercise / Physical Activity | Reduction of 6.72 mmHg in Mean Systolic Blood Pressure (SBP) (p = 0.0425) and 4.0 mmHg in Diastolic Blood Pressure (DBP) (p = 0.0073).Mean weight reduction of 1.75kg (p = 0.0023). | Overall moderate risk of bias |
| Baig et al., 2015 USA (WHO Americas region) Randomised Controlled Trial | Picture Good HealthChurch/churchesGovernment agency fundedCardiovascular risk factorsSelf management of cardiovascular risk factors(100) 50 intervention, 50 control | Trained lay church leaders delivered individual and interpersonal level cardiovascular self-management classes to urban American Latino adults of all income levels. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | 3 Months:Intervention group Mean Systolic BP decrease of −2.72 95%CI (−7.19 to 1.74). Control group Mean SBP increase of 1.42 mmHg 95% CI (−3.29 to 6.13).6 months:Intervention group Mean SBP decrease of −4.62 95%CI (−9.14 to −0.09). Control group Mean SBP increase of 0.6 mmHg 95% CI (−3.85 to 5.04).3 months:Intervention group Mean DBP decrease, -4.07mmHg 95% CI (-7.19 to -0.94). Control group Mean DBP increase of 1.78 mmHg 95% CI (−1.5 to 5.06).6 months:Intervention group Mean DBP decrease, -3.14mmHg 95% CI (-6.59 to -0.31). Control group Mean DBP increase of 0.91 mmHg 95% CI (−2.5 to 4.33).6 months:Intervention group Mean Glycosylated haemoglobin (%)decrease of -0.27 95%CI (-0.7 to 0.14) in 3 months and -0.27 95%CI (-0.81 to 0.28).6 months:Control group Mean Glycosylated haemoglobin (%) decrease of -0.05 95%CI (-0.59, 0.48) in 6 months and -0.05 95%CI (-0.59, 0.48).6 months:Intervention group Mean Low Density Lipoprotein (mg/dL) decrease of -0.16 95%CI (-6.61, 6.3) in 3 months and -4.94 95%CI (-13.64, 3.77).6 months:Control group Mean Low Density Lipoprotein (mg/dL) increase of 0.97 95%CI (-5.78, 7.72) in 3 months and 2.55 95%CI (-6.04, 11.13).6 months:Intervention group Mean Waist Circumference (cm) decrease of -0.93 95%CI (-2.63, 0.76) in 3 months and -9.8 95%CI (-2.83, 0.88).6 months:Control group Mean Waist Circumference (cm) decrease of -0.22 95%CI (-1.99, 1.55) in 3 months and -0.51 95%CI (-2.34, 1.32). | Overall some concerns risk of Bias |
| Draper et al., 2019 South Africa (WHO Africa region) Prospective uncontrolled longitudinal study | Health through Faith [Impilo neZenkolo]Church/churchesGovernment agency fundingCardiovascular riskHealthy lifestyle to reduce cardiovascular risk84 participants | Church leaders and members delivered individual and interpersonal lifestyle teaching sessions to low income rural and urban Black / African adults. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Mean Systolic BP reduction of 1 mmHg from 123 95%CI (107, 132) to 122 95%CI (116, 134), p = 0.085, z = -1.721.Mean Diastolic BP increase of 3 mmHg from 81, 95%CI (72, 86) to 84, 95%CI (74, 92), p = 0.451, z = -0.753.Mean weight reduction of 2.2kg 80.5 ± 20.1 to 78.3 ± 19.1, p = 0.010Mean BMI reduction of 0.8 kg/m2 from 29.9 ± 7.4 to 29.1 ± 7.1, p = 0.01.Mean waist circumference decrease of 4.1 cm from 92.3 ± 17.4 to 88.2 ± 15.9, p = 0.02.Other improvements in self reported measures of dietary habits, and self-reported health status, psychological distress and self-esteem scores. | Overall moderate risk of bias |
| Lee et al., 2018 USA (WHO Americas region) Qualitative study | Intensive peer group interventionChurches/communityGovernment agency fundingCardiovascular disease riskReduce risks for heart disease | Male church members and peer facilitators delivered individual and interpersonal level intensive heart disease knowledge, awareness and support sessions to urban Black / African American adults. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Intensive peer group intervention was effective to promote cardiovascular health beneficial behaviour change through: Enhancing access to health behavior information and resources; Practicing and applying problem-solving skills with group feedback and support; Discussing health behavior challenges and barriers; Sharing health behavior changes; Sharing perceived health outcome improvements and benefits; The feeling of belonging and being cared for; Addressing health of family and community. | Overall high quality rating |
| Benjamin, 2017 USA (WHO Americas region) Uncontrolled before and after study | Smart Self Management interventionChurches/ homesNon Profit OrganisationHypertensionIncrease hypertension knowledge and self management23 participants | Faith institution participating rural and urban Black Haitian American adults across all income levels delivered individual and interpersonal level hypertension education informed and culturally appropriate self-created SMART management. | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Increases in knowledge gained and ability to create and maintain SMART goals:Significant increase in Basics of hypertension scores from a mean pre-test 55% to mean post-test of 85%. 30%, 95% CI (0.23–0.37). T = 9.376, p < 0.001.Increase in Dietary modifications & BP management scores from a mean pre-test 76% to mean post-test of 89.3%. 11%, 95% CI (0.1–0.32). T = 1.2301, p =0.2584.Significant increase in Physical activity & BP scores from a mean pre-test 64% to mean post-test of 87%. 24%, 95% CI (0.13–0.35). T = 5.0992, p = <0.001.Significant increase in medication management & symptom recognition scores of 31%, 95% CI (0.21–0.42). T = 6.74, p = <0.001. | Overall serious risk of bias |
| Spell-LeSane, 2016 USA (WHO Americas region) Uncontrolled before and after study | I am Working on My Heart: A Cardiovascular Disease Awareness ProgramChurch/churchesGovernment agency fundingCardiovascular riskIncrease cardiovascular disease knowledge, awareness and motivation137 participants | Registered nurses delivered individual and interpersonal level cardiovascular disease knowledge, awareness learning and motivation sessions to rural and urban Black / African American adult women across all income levels | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Participant awareness of cardiovascular disease as leading cause of death significantly increased 27% from 63% to 90% (p = <0.001).Participant knowledge of Cardiovascular disease improved with the proportion of very well informed women increasing 28% from 3% to 31% (p = <0.002).Participant engagement in Physical activity improved.27% of previously not active participants increased the frequency and intensity of physical activity significantly (p <0.0001).18% of participants who pre-intervention engaged sub-optimally in physical activity (PA) increased their PA to meet the guidelines.42% increase in physical activity of women with optimal pre-intervention physical activity levels. | Overall low risk of bias |
| White, 2018 USA (WHO Americas region) Uncontrolled before and after study | Check. Change. Control.Churches/ homesNo Funding identifiedHypertensionReduce Blood Pressure23 participants | Volunteer hypertensive Black / African American adults across all income levels self-delivered an American Heart Association (AHA) self management programme | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Minimal statistically non-significant Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP) reduction:Mean SBP decrease of 3.09mmHg, 95%CI (0–6.558), T = 1.844, p = 0.079.Mean DBP decrease of 2.26mmHg, 95%CI (-0.459–5.59), T = 0.724, p = 0.099. | Overall moderate risk of bias |
| Bittman et al., 2020 USA (WHO Americas region) Randomised Controlled Trial | Gospel Music interventionChurchesNongovernmental Organisation fundingHypertension riskImprove engagement in cardiovascular risk reduction(71) 36 intervention, 35 control | A researcher and a team of local congregation derived cardiologist, musician facilitators, registered nurses and volunteer health workers, delivered Gospel music driven engagement in cardiovascular risk reduction programme to urban Black / African American adults across all income levels | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Statistically non-significant reduction of Systolic Blood Pressure (SBP) (p = 0.54) and Diastolic Blood Pressure (DBP) (p = 0.41).SBP Intervention group Pre-intervention 140.99 95%CI (136.12 to 145.85), post-intervention 133.97 (129.28 to 138.66); Control group pre-intervention 138.48 (312.65 to 144.31), post-intervention 130.10 (124.48 to 135.73).DBP Intervention group Pre-intervention 82.88 95%CI (79.17 to 86.60), post-intervention 82.20 95%CI (78.49 to 85.92),Control group pre-intervention80.33 95%CI (75.88 to 84.78), post-intervention 76.29 95%CI (71.84 to 80.75).Reduction of weight, hip and waist circumference.No significant between group effect for weight, F(1,41) = 0.22, p = 0.64: Intervention group pre-intervention 200.29 95%CI (185.09, 215.48), post-intervention 196.57 95%CI (182.37, 210.77).Control group pre-intervention 174.07 95%CI (153.30, 194.83), post-intervention 178.40 95%CI (159.00, 197.80).No significant between group effect for Hip circumference, F(1,41) = 0.16, p = 0.70:Intervention group pre-intervention 48.07 95%CI (45.99, 501.5), post-intervention 48.21 95%CI (46.07, 50.37). Control group pre intervention 43.73 (40.89, 46.58), post-intervention 44.87 (41.92, 47.82). | Overall high risk of Bias |
| Ray, 2003 USA (WHO Americas region) Non-Randomised Controlled Trial | Church-based nutrition interventionChurchesNo Funding identifiedHypertension and hypertension riskImprove health by weight loss and Blood Pressure reduction(31) 19 intervention, 12 control | Three nurses and a dietician from the local community delivered individual and interpersonal Scripture supported, culturally competent nutrition education workshops to rural and urban Black / African American adults across all income levels | ✓ | | Nutrition | Mean Systolic Blood Pressure (SBP) decrease of 7.2 mmHg–v–control group mean SBP increase of 2.78mmHg (p = 0.935).Mean Diastolic Blood Pressure (DBP) decrease of 6.93 mmHg–v–control group mean DBP increase of 2.33mmHg (p = 0.961).Significant reduction in the weight of participants in the treatment group, F(3,12) = 5.29, p<0.05 compared to non-significant increase in weight of the participants in the control group.Increased awareness about blood pressure, p<0.05, in the intervention group. | Overall low risk of bias |
| Wiist et al., 1990 USA (WHO Americas region) Randomised Controlled Trial | Pilot cholesterol education programmeChurchesGovernment agency fundingCardiovascular riskCardiovascular screening and nutrition education(348) 174 intervention, 174 control | Church leaders and volunteer health professional members delivered individual and interpersonal cardiovascular screening and nutrition education to rural and urban Black / African American adults across all income levels | ✓ | ✓ | Nutrition | Cholesterol levels decrease from 233.9mg/dl to 210.4mg/dl (p<0.0001) compared to control group Cholesterol levels decrease from 241.5mg/dl to 202.9mg/dl (p<0.0001).No statistically significant differences between the intervention and control group baseline or follow-up levels. | Overall some concerns risk of Bias |
| Yanek et al., 2001 USA (WHO Americas region) Randomised Controlled Trial | Project JoyChurchesGovernment agency fundingCardiovascular riskImprove cardiovascular risk profile(529) 455 intervention, 74 control | Lay church leaders and health educators delivered individual, interpersonal and community level cardiovascular risk profile improving nutrition and physical activity education to urban Black / African American adult women across all income levels | ✓ | | Combination of Nutrition & Exercise / Physical Activity | Mean Systolic Blood Pressure (SBP) decrease of 1.6 mmHg–V–control group mean SBP decrease of 0.95, p = 0.47.Mean Diastolic Blood Pressure (DBP) decrease of 0.36 mmHg–V–control group mean SBP increase of 0.22, p = 0.49.In the top 10% cohort by weight loss: Mean SBP decrease of 8.1 mmHg (p = 0.0005)–V–control group mean SBP decrease of 3.3 mmHg (p = 0.5688).Mean DBP decrease of 4.4 mmHg (p = 0.0004)–V–control group mean DBP increase of 0.8 mmHg (p = 0.7805).Mean body weight decrease of 0.5kg–V–control group mean weight increase of 0.38kg (p = 0.0008).Mean Body Mass Index (BMI) decrease of 0.17 kg/m2 –v–control group BMI increase of 0.14 kg/m2 (p = 0.0012).Mean daily Sodium intake decrease of 145mg/day–v–control group decrease of 8 mg/day, p = 0.0167. | Overall some concerns risk of Bias |
| Smith et al., 1997 USA (WHO Americas region) Uncontrolled before and after study | Church-based education and support programmeChurchesGovernment agency fundingHypertensionHypertension reduction97 participants | Nurses trained as church health educators and lay church members trained as organizers and facilitators, delivered individual and interpersonal level hypertension education and practical support to urban Black / African American adults across all income levels | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Mean Arterial Pressure decrease (p ≤ 0.0001, F = 17.80, df = 1,86)Systolic Blood Pressure decrease (p ≤ 0.0001, F = 18.35, df = 1,91)Diastolic Blood Pressure decrease (p ≤ 0.008, F = 17.48, df = 1,91)Hypertension knowledge scores significantly increased and remained higher than baseline at 3 months after end of intervention (p ≤ 0.0001; F = 95.08; df = 1,79). | Overall moderate risk of bias |
| Oexmann et al., 2001 USA (WHO Americas region) Uncontrolled before and after study | Lighten Up lifestyle programmeChurchesNongovernmental Organisation fundingCardiovascular riskReduce cardiovascular risk381 participants | Volunteer lay church members trained as health educators delivered individual and interpersonal level lifestyle modification education to urban White and Black / African American adults across all income levels | ✓ | | Healthy Lifestyle Coaching, Counselling & Motivation Training | Post 0–5 session attendance:Systolic Blood Pressure (SBP) decrease of 1.2 ± 1.2 (mmHg), p>0.5 for Blacks; decrease of 5.6 ± 1.7(mmHg), p≤ 0.01 for Whites.Diastolic Blood Pressure (DBP) decrease of 0.2 ± 0.9(mmHg), p>0.5 for Blacks; decrease of 2.2(mmHg) ± 1.0, p≤ 0.05 for Whites.Post 6–8 sessions attendance:SBP decrease (mmHg) of 6.3 ± 1.2, p>0.5 for Blacks; decrease of 8.1 ± 1.7, p≤ 0.01 for Whites.DBP decrease (mmHg) of 2.5 ± 0.8, p ≤ 0.01 for Blacks; decrease of 3.6 ± 1.2, p≤ 0.01 for Whites.Post 0–5 sessions attendance:Body weight decrease (pounds) of 2.7 ± 0.4, p < 0.001for Blacks; 3.6 ± 0.8, p < 0.001 for Whites.Total cholesterol increase (mg/dl) of 1.0 ± 2.6, p>0.5 for Blacks; and a decrease of 13.2 ± 4.0, p≤ 0.05 for Whites.Post 6–8 sessions attendance:Body weight decrease (pounds) of 3.5 ± 0.4, p < 0.001for Blacks; 6.3 ± 0.6, p < 0.001decrease for Whites.Total cholesterol decrease (mg/dl) of 4.9 ± 2.3, p<0.05 for Blacks; decrease of 14.5 ± 2.9, p≤ 0.001 for White participants. | Overall moderate risk of bias |
## Faith institution roles
Faith institution roles constituted a variety of promotion of, provision for and persuasion of participant engagement in cardiovascular activities. Twenty-two studies incorporated cardiovascular health promotion and education roles. Two cohort studies did not, being mainly blood pressure measurement studies [57, 58]. One study combined cardiovascular health promotion and education roles with blood pressure measurement [60]. Altogether, the 24 studies reported 19 different types of relevant outcomes.
Blood pressures were reported by 17 studies; and mean arterial blood pressure outcomes reported by two [34, 61]. Secondary outcome measures (risk modifying elements) are diverse, mostly quantitative and include body weight, waist circumference and hip circumference measurements, Body Mass Index, nutrition literacy scores, physical activity, healthy behaviour engagement scores, healthy cardiovascular knowledge scores, blood sugar measurements, serum glycosylated haemoglobin measurements, serum Low Density Lipoprotein measurements, serum cholesterol measurements and dietary sodium.
Some secondary outcomes were more complex to measure, including adoption of healthier diets, nutrition label literacy awareness, engagement in health—smart behaviors, increased hypertension monitoring and engagement with healthcare system. Process indicator outcomes were reported for studies and include intervention uptake levels, extent of participation, intervention acceptability and participant satisfaction [56, 62–64].
Four studies incorporated and reported on personal psychological effectiveness [59, 61, 65, 66]. A single study reported on personal control and understanding of consequences as outcome measures [67]. One study reported faith related factors, such as targeted training with religious content, pastoral leadership and sociocultural awareness of educators being influential in personal effectiveness for behaviour modification [55]. A single study addressed nutrition label literacy promoting activities contributing to cardiovascular health promotion [68].
The biological markers targeted in education and health promotion include blood sugar, glycosylated haemoglobin, cholesterol and low-density lipoprotein levels. A single study reported blood sugar levels as part of the intervention outcomes [69]. Lowering serum glycosylated haemoglobin, a marker of the effectiveness of diabetes control, was reported by a single study [62]. Two studies directly reported the outcomes of serum cholesterol levels [60, 70]. A single study reported the closely related outcome of Low Density Lipoproteins, an indirect marker of the risk of coronary heart disease [62].
Alteration of biological markers involves manipulation of energy consumption and expenditure, notably through nutrition and exercise. Two studies directly reported the outcomes of healthy eating or healthy eating habits [68, 69], while a single study captured the same in participant satisfaction outcomes [62]. Closely related is the promotion of healthy drinking habits, which was reported by one study [68].
Motivation or intention for dietary change, usually toward increasing fibre, fruit and vegetable consumption and reducing dietary fat is part of the activities contributing to the health promotion. Four studies reported the outcome of dietary change behaviour or intention [56, 59, 62, 65].
Activities targeting attitudes toward exercise or physical activity contribute to health promotion roles. Five studies incorporate attitudes to or uptake of exercise or general physical activity, and reported them [59, 65, 66, 68, 69]. A single study used measures that influence self reported psychological wellness or self-esteem [56].
## Primary outcomes
The primary outcomes, Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP) indicate the estimate and direction of effect of the interventions.
## Meta-analyses of primary outcomes in randomised studies
Only four RCTs with 865 participants could contribute to the meta-analysis of blood pressure outcomes at 3 months with one study intervention outcomes unavailable before 12 months follow-up [71], and the other having no blood pressure outcomes [60]. Using a random effects model, the overall pooled estimate for mean reduction of systolic BP three months after the interventions was significant at 2.98 mmHg ($95\%$CI -4.39 to -1.57), compared to non-participation in hypertension intervention. Heterogeneity was measured as I2 $0\%$ (Fig 4A). The overall pooled estimate shows a non-significant mean increase of 0.14 mmHg ($95\%$CI -2.74 to +3.01) in the diastolic BP three months after the interventions. Heterogeneity, I2 = $70\%$ (Fig 4B).
**Fig 4:** *Forest plot of meta-analysis of change in systolic and diastolic blood pressures 3 and 12 months post hypertension intervention.*
Meta analysis was conducted on the two RCTs (including 600 participants) that recorded systolic and diastolic blood pressure outcome measures a year post intervention. Using a random effects model, showed an overall pooled estimate of significant mean reduction of 0.65 mmHg ($95\%$CI -0.91 to -0.39) in systolic BP twelve months after the interventions. Heterogeneity, I2 = $0\%$ (Fig 4C). Using a random effects model, showed an overall pooled estimate of non-significant mean reduction of 0.53 mmHg ($95\%$CI -1.86 to 0.80) in diastolic BP twelve months after the interventions. Heterogeneity, I2 = $0\%$ (Fig 4D).
## Meta-analyses of secondary outcomes in randomised studies
Meta-analyses of the RCTs that measured body weight and waist circumferences changes indicates an overall small but beneficial intervention effect at the end of the interventions (3 months for Tucker et al. and Baig et al., 12 months for Yanek et al. and Bittman et al.): significant mean weight reduction of 0.83kg ($95\%$ CI -1.19 to -0.46), I2 = $0\%$ and non-significant mean waist circumference reduction of 1.48cm ($95\%$ CI -3.96 to +1.00), I2 = $41\%$ (Fig 5A and 5B).
**Fig 5:** *Forest plot of meta-analysis of body weight change and waist circumference change at end of intervention.*
## Primary outcomes in Non-randomised studies
Rather than single or summary estimate of effects, an indication is presented of the effect of faith institution facilitated activities on the direction of effect of the available outcomes.
With the exception of a single study where post-intervention diastolic blood pressures increased [56], systolic and diastolic blood pressures decreased. Nearly half of the reductions were statistically significant (Table 6). One study reported a mixed picture of diastolic blood pressure reduction, with significant diastolic blood pressure reductions for white and non-significant diastolic blood pressure reductions for Black participants [70].
**Table 6**
| Study | Systolic Blood Pressure | Diastolic Blood Pressure |
| --- | --- | --- |
| Study | (Significance) | (Significance) |
| Taylor, 2011 | Reduction† | Reduction† |
| Taylor, 2011 | (p = 0.30) | (p = 0.38) |
| Kinard, 2016 | Reduction† | Reduction† |
| Kinard, 2016 | (p = 0.99) | (p = 0.99) |
| Dodani et al., 2014 | Reduction* | Reduction* |
| Dodani et al., 2014 | (P = 0.001) | (P = 0.0048) |
| Dodani et al., 2013 | Reduction* | Reduction* |
| Dodani et al., 2013 | (P = 0.005) | (P = 0.01) |
| Dodani et al., 2015 | Reduction* | Reduction* |
| Dodani et al., 2015 | (P = 0.0425) | (P = 0.0073) |
| Draper et al., 2019 | Reduction† | Increase∂ |
| Draper et al., 2019 | (p = 0.085) | (p = 0.451) |
| White, 2018 | Reduction† | Reduction† |
| White, 2018 | (p = 0.079) | (p = 0.099) |
| Smith et al., 1997 | Reduction* | Reduction* |
| Smith et al., 1997 | (p = 0.0001) | (P = 0.008) |
| Oexmann et al., 2001 | Reduction | Reduction |
| Oexmann et al., 2001 | (Mixed) Blacks p = 0.5 Whites p = 0.01 | (Mixed) Blacks p = 0.01 Whites p = 0.01 |
| Ray, 2003 | Reduction† | Reduction† |
| Ray, 2003 | p = 0.935 | p = 0.961 |
## Secondary outcomes in nonrandomised studies
Four of the included nonrandomised studies reported no secondary outcome measures, while 12 studies recorded a large number of distinct outcomes. These included mean weight; blood sugar; knowledge of hypertension or cardiovascular disease and health; attitudes to diet and exercise; hypertension or cardiovascular disease awareness; personal control; total cholesterol; Body Mass Index; waist circumference and physical activity, among others. A large diversity of methodologies was employed in the studies, rendering aggregation of secondary outcome measures difficult. However individually the distinct secondary outcomes generally indicate beneficial post-intervention effects for hypertension and cardiovascular health (Table 7).
**Table 7**
| Study/ Intervention | Distinct Secondary Outcome | Intervention impact on Hypertension or Cardiovascular Health, based on single distinct secondary outcome as a risk or protective factor | Intervention impact on Hypertension or Cardiovascular Health, based on single distinct secondary outcome as a risk or protective factor.1 | Intervention impact on Hypertension or Cardiovascular Health, based on single distinct secondary outcome as a risk or protective factor.2 |
| --- | --- | --- | --- | --- |
| Study/ Intervention | Distinct Secondary Outcome | Significantly beneficial | Non-significantly beneficial | Not beneficial |
| Kinard (2016)My Sister’s Keeper Project | Weight | | | Intervention group non-significant mean weight increase (p>0.05). Control group non-significant weight reduction (p>0.05). |
| Ray (2003)Church-based nutrition interventionIntervention | Weight | Intervention group significant (p<0.05) mean weight reduction. Control group—non-significant mean weight increase (p>0.05). | | |
| Dodani (2014)Healthy Eating and Living Spiritually (HEALS)Dodani (2015)Healthy Eating and Living Spiritually (HEALS)Draper (2019)Health through Faith [Impilo neZenkolo]Oexmann (2001)Lighten Up lifestyle programme | Weight | Significant mean weight reduction:Dodani (2014) p < 0.05; Dodani (2015) p = 0.0023; Draper (2019) p = 0.010Oexmann (2001) p < 0.001 | | |
| Taylor (2011)Way of Faith (WOF) project | Weight | | | Small non-significant weight increase |
| Daye (2019)Faith-based health devotional intervention | Knowledge | Significant increase in Hypertension Prevention IQ Quiz scores p < .0001 | | |
| Benjamin (2017)Smart Self Management intervention | Knowledge | Significant increase in Basics of Hypertension Scores p < 0.001. | | |
| Spell-LeSane (2016)I am Working on My Heart Cardiovascular Disease Awareness intervention | Knowledge | Significant increase in the proportion of well informed participants in knowledge of cardiovascular disease. p <0.002 | | |
| Smith (1997)Church-based education and support intervention | Knowledge | Significant post intervention hypertension knowledge scores, p ≤ 0.0001 | | |
| Abbot (2015)‘With Every Heartbeat is Life’ intervention, | Knowledge | Significant increase (p < .01) in cardiovascular health knowledge scores of intervention participants compared to the control group | | |
| Ray (2003)Church-based nutrition interventionIntervention | Blood pressure awareness | Significant increase in blood pressure awareness index, p<0.05 | | |
| Spell-LeSane (2016)I am Working on My Heart Cardiovascular Disease Awareness intervention | Awareness as leading cause of death | Significant increase in awareness of hypertension and cardiovascular disease as leading cause of death, p = <0.001 | | |
| Daye (2019)Faith-based health devotional intervention | Perceived personal control | Significant increase in Personal Control scores, p < 0.0005 | | |
| Abbott (2015)‘With Every Heartbeat is Life’ intervention, | Perceived behavioural control to enhance self-efficacy | Significant increase in self-efficacy scores to increase fruit and vegetable intake, p < 0.01.Significant increase in self-efficacy scores to reduce dietary fat intake, p = 0.03.Significant increase in self-efficacy scores to increase exercise, p < 0.01. | | |
| Spell-LeSane (2016)I am Working on My Heart Cardiovascular Disease Awareness intervention | Engagement in physical activity | Significant increase in frequency and intensity (27% of pre-intervention inactive participants, p <0.0001)Increase in physical activity (18% of pre intervention optimally exercising participants)Increase in physical activity (42% of women with adequate pre-intervention physical activity levels) | | |
| Benjamin (2017)‘Smart Self Management’ intervention | Engagement in physical activity | Significant increase in pre-test to post-test physical activity scores, p <0.001. | | |
| Oexmann (2001)‘Lighten Up lifestyle programme’ | Total cholesterol | Significant total cholesterol reduction, p<0.05. | | |
| Abbott (2015)‘With Every Heartbeat is Life’ | Attitudes to fruits/vegetables, dietary fat, and exercise | Significant increased fruits and vegetables consumption, p < 0.01.Significant reduction in dietary fat, p = 0.03. | | No change in attitude toward exercise or physical activity |
| Taylor (2011)‘Way of Faith (WOF) project’ intervention | Attitudes to taking care of self | Post intervention greater self care responsibility:An additional 60% of participants started exercising.90% of participants positively changed their eating habits | | |
| Taylor (2011)‘Way of Faith (WOF) project’ intervention | Blood sugar | | | Post-intervention Blood Sugar increase |
| Draper (2019)Health through Faith [Impilo neZenkolo] | Body Mass Index | Significant post-intervention Body Mass Index reduction, p = 0.01. | | |
| Daye (2019)Faith-based health devotional intervention | Understanding of the negative consequences of poor blood pressure control | | A non-significant increase in understanding of the negative consequences of poor blood pressure control, p < 0.24. | |
| Abbott (2015)‘With Every Heartbeat is Life’ | Intention to change fruits / vegetables, dietary fat, and exercise habits | Significant increase in scores for intention to consume more fruits and vegetables, p < .01.Significant increase in scores for intention to reduce dietary fat, p < .01. | | No change in scores for intention to increase exercise |
| Benjamin (2017)‘Smart Self Management’ intervention | Medication management & symptom recognition | Significant increase in scores for medication management & symptom recognition, p <0.001. | | |
| Benjamin (2017)‘Smart Self Management’ intervention | Dietary modifications & blood pressure management | | Non-significant increase in scores for dietary modifications & blood pressure management, p = 0.2584. | |
Only three interventions reported any non-beneficial findings. The My Sister’s Keeper Project and the Way of Faith (WOF) project interventions did not demonstrate a beneficial or protective effect in terms of the isolated outcome of weight reduction [69, 72]. The ‘With Every *Heartbeat is* Life’ did not show a beneficial change in attitude or intention toward exercise or physical activity [65]. Similarly post-intervention blood sugar increase following intervention in the Way of Faith (WOF) project indicates no beneficial cardiovascular impact. There, however, is a clear demonstration of significantly beneficial impact of the distinct secondary outcomes for hypertension or cardiovascular health, with 14 distinct outcomes showing significantly protective effect and 2 distinct outcomes showing beneficial effect but not to significant levels (Table 7).
## Characteristics of faith institution roles
An outline of the qualitative synthesis based on the Diffusion of Innovation theory, the Communication–Behaviour change model, and the Template for Intervention Description and Replication (TIDieR) *Checklist is* presented (Table 8).
**Table 8**
| The agents delivering the interventionsThe agents delivering hypertension related intervention were, paid or unpaid, trusted based on characteristics as already skilled or trainable, personally influential or health interested, locally-derived and locally-based, personally invested individuals. They were interested adults that qualify as trusted insiders. |
| --- |
| Training received by agents delivering the interventionsThe training received by agents consisted interactive, flexible and creative transfer of skills and competencies from experienced role performers, experts or professionals to adult learners of varied abilities. Training was intervention specific, dependent on required roles, theoretically based and pragmatically delivered to enable empowerment of learners with self-development or skill acquisition without formal assessment, enhancement of existing competencies, and sometimes formal, certifiable achievement. Training was customised to the requirements of the intervention and modified as indicated by needs of the adult learners, and as such was characterised by flexibility to accommodate a variety of styles, techniques and abilities. |
| Materials used in intervention deliveryThe materials used to deliver hypertension interventions comprise communally administered, health content preponderant religion friendly messaging and assets. These include varying combinations of textual and pictorial health and faith content, audio-visual technology assisted health and faith content, and participation-based core religious content delivered with techniques and formats requiring skilled human contact. Locally derived facilitators, influencers and trusted leaders from the faith institutions themselves function as resources and materials for hypertension intervention. |
| Component and combinations of the interventionsA broad range of structured group and individual activities targeting the initiation, achievement and maintenance of behaviour change through contextually applicable tools were used. These include: teaching and learning on and linking nutrition, physical activity, spirituality, lifestyle, health and disease; teaching, learning and practice on personal effectiveness and behaviour change; teaching, learning and practice on monitoring cardiovascular health, disease, lifestyle and behaviour; and contextually applicable creative group and cultural activities suited to reinforcement of learning and an active lifestyle, such as music performance and outdoor recreational pursuits. |
| Frequency of intervention deliveryThe frequency and timing of interventions vary depending on their types, formats and intensities of the intervention activities, resource requirements including training needs, and the local contexts relating to intervention participants. Participant contexts related to urban or rural location, family commitment and employment-determined availability are particularly influential. Intervention durations vary, from six weeks to multiple months, more than a year or several years. Intervention frequencies are as varied as daily or weekly to monthly. |
| Religious components of the interventionsThe religious components of interventions vary, and faith institutions do contribute roles outside faith or spiritually based activities: such as programme endorsement and publicity, participant recruitment, training and facilitation. Many interventions have no inherent religious component, their host faith institutions being relevant solely in utility as physical assets enabling intervention delivery. Proximity of interventions to routine religious activity does not constitute religious component. The religious components however include: teaching, learning and discussion of religious scripture; incorporation of scripture linking lifestyle, physical and spiritual health; use of prayers, meditation and scriptural reasoning; spiritually based singing, chanting and worship; and the incorporation of unique cultural expressions into spiritual or religious songs. |
| Mechanism of action of the interventionsThe processes involved in hypertension interventions securing individual and collective behaviour change appear to be through several interrelated and overlapping mechanisms that all involve empowerment and initiation or sustenance of cardiovascular health beneficial actions. The mechanisms are not always explicitly stated, tend to be combined in interventions and include: education, coaching, counselling, reasoning, goal setting, encouragement, motivation, and habit and ritual development. |
| Socioeconomic and cultural CompatibilityThe interventions had certain features that rendered them trustable and compatible with the prevailing social, economic and cultural values of their target communities. Confidence in some level of pre-existing relationship with key community gatekeepers and the deployment of the Community-based participatory research approach, appear to be constant features.The specific features include the following: ◾ Designs that recognize and respect the spirituality and spiritual values of the target communities, and also incorporate them into the proposed interventions. ◾ Involvement of the target communities in design decisions, promotion of their participation in its delivery and encouragement of a sense of ownership of the intervention. ◾ In pursuing engagement with the target faith communities to secure their participation, according them respect and recognition by working in consultation and collaboration with their established hierarchy. ◾ As much as feasible and achievable in individual contexts, inclusion and appropriate utilisation of local faith institution derived or affiliated experts, decision makers, facilitators and intervention agents. ◾ Deliberate use of culturally appropriate approaches in design and implementation that yield to adaptations informed by the social, economic and environmental realities of the target faith institution users and their communities; and offer participants choice and flexibility. ◾ Use or incorporation of validated tools such as government or professionally validated materials or methods, and trusted secular and religious experts such as healthcare professionals, theologically trained clergy and locally experienced insiders. ◾ Where contextually appropriate, involvement of faith institution and wider community female stakeholders such as pastor’s wives, female clergy, influential gatekeepers and female researchers. |
| Clarity of intended benefitThe intended benefits to participants were varied. Their clarity was promoted in several aspects including: promotion and understanding of cardiovascular health issues; provision of opportunity for peer collaboration and professional support in personal cardiovascular risk reduction; empowerment with awareness, knowledge, acquisition of health-beneficial skills; generating motivation for behaviour change; and achievement of beneficial blood pressure, body weight and healthy dietary change outcomes. |
| Simplicity of the interventionsThe interventions were designed to incorporate simplicity, flexibility and elimination of participant difficulty, whilst focusing on individual participants, encouraging communal implementation and inspiring trust. Included were the sense of communal ownership, empowerment of individual participants, enhancement of active participation, and improvement of accessibility.The feature of sense of communal ownership derive from incorporation of traditional and familiar elements of local culture; use of trusted, influential facilitators and agents for intervention delivery, follow up and encouragement; and customisation to local norms as well as alignment with existing practices including prayers, scripture and faith-based discussion. Individual participant empowerment feature of the interventions derive from the use of transparent and ethical consent processes; the promise and strict maintenance of data confidentiality including by non-inclusion of any data perceivable as sensitive; and where applicable availability of the option for individual participants to extend their participation beyond any basic activities the interventions offer. The features of active participation in the interventions include, in addition to use of familiar or traditional elements from the local culture, generous adaptability of the timing, duration and location of learning or practical sessions; free provision of convenient resources or materials to suit individuals and groups with different abilities and work-life balance; and where applicable, simplification of processes to enable self-administration. The features improving accessibility include the simplification of intervention materials to ensure easy readability and comprehension; and the provision of materials in multiple appropriate languages or dialects. |
| Reversibility, perceived risk or trialability of the interventionsThe interventions had in-built features that encourage participation by signifying negligible risk to participants, and the reversibility of impacts where participants decide to withdraw prematurely. These features rely on participants’ familiarity with intervention elements, availability of alternative intervention elements, provision of individualised intervention elements, confidence of the clergy in the interventions, and availability of volitional exit. The familiarity features are the intervention elements and activities with which participants were already conversant, and which are perceived as admissible or of negligible risk. The alternative activities features are activities that could replace elements of the intervention where required to secure the participation of particular groups or individuals. The individualised activities features are the activities modified from the intervention elements, to secure participation and engagement of particular individuals who would otherwise not be able to participate. The Clergyman’s confidence feature is the credence invested by faith institution leaders in the intervention- the expression and publicity of which encourages participation. The feature of open exit is the availability to individual participants and groups, of the option of unhindered and volitional self-removal from the intervention without adverse consequences. |
| Observability of intervention resultsMost interventions did not have features that enabled potential participants to pre-empt, examine or observe potential effects of intervention on earlier participants. When they did, the observability of results by potential faith institution participants were based on open discussion and peer feedback. This involved early intervention participants sharing experiences and new learning with potential future participants, and demonstrating any new attitudes and resulting behaviour. |
| Perceived sourceInterventions were constructed such that their core messages would be perceived to come from easily identifiable, well-informed, credible sources. The messages of hypertension interventions thus delivered were perceived to derive from four primary sources: healthcare professionals and allied workers; health and healthcare researchers; the teaching from and authority of religious scripture; and, well-informed faith and religious leaders. |
| The message Hypertension interventions were constructed to target the reduction of individual and collective cardiovascular risk, but also include content based on religious traditions and scripture. The core messages of the interventions were clear: cardiovascular risk reduction is achievable by improving health literacy and lifestyle management focussing on nutrition, exercise, stress, sleep and attitudes toward health; and religious scriptures are instructive that divine plans and intentions do exist for sound physical health and healthy living, and to achieve such individuals are responsible and capable. The delivery of the constructed health messages were in sessions that typically lasted 45 to 90 minutes, and were varied in type: from health promotional teaching and scripture infused health topic workshops, to personal individual reflective application guided by learning from teaching and workshops. |
| The channelThe channel of delivering cardiovascular health messages included a diversity of media and techniques. These were determined by the contexts of the individual faith institutions and their participants. Media used consist reading materials like handouts, pamphlets, posters, workbooks, manuals, devotionals; digital media including presentations, animations, videos; and interactive media such as church services, prayers, and existing close personal interactions. The techniques varied from class focussed interactive group teaching, discussion and practical demonstrations to individually focussed coaching, motivation, personal reflection and devotion. |
| ReceiverRural and urban resident adults across all income groups were target audiences and eventual receivers of hypertension interventions delivered within faith institutions. Relative to White American, White European, Latino, Haitian American and other minority ethnic backgrounds however, there was a preponderance of Black African, African American eventual targets and receivers of faith institution delivered hypertension interventions. |
| Intended destination / outcomeWhere explicitly stated the intended outcomes of hypertension interventions included improvement of health literacy and adoption of health promoting behaviour with the outcomes of reduction of body weight and blood pressure. |
| Intervention modificationThere was no evidence that hypertension interventions delivered within faith institutions underwent modification during the course of the intervention. |
Faith institutions roles in hypertension have particular characteristics that hinge on relationships, trust, a sense of ownership, local leadership, a high degree of sociocultural contextualisation and conformity to the practice of faith. Where possible faith institutions used trusted insiders equipped with varied, flexible, faith-friendly materials and direct interpersonal human contact; investing the best available quality context specific training in them. Interventions were flexibly delivered within participants’ contexts, targeted the individual and groups, and comprised or linked lifestyle and spirituality while incorporating or at least respecting their sociocultural and religious realities.
By design the interventions were simplified, transparent, practical and empowering to increase knowledge, give guidance or motivate toward cardiovascular health beneficial habits. The interventions were rationalised as harmless by participants, who then engaged by free choice having perceived the interventions to be respectful, and supported by their faith and community leader.
## Peers and faith related factors are the real drivers of engagement
Sternberg et al. and Lee et al. indicate that the main mechanism of achieving cardiovascular beneficial behaviour change hinges on leveraging the influence of peers and communities within the faith environment to engage in interventions [55, 73].
Sternberg et al. showed that faith-based interventions appear to be influential in reducing cardiovascular risk factors, especially for minorities and women [55]. They found that in order to achieve the outcomes needed to deliver better blood pressure or cardiovascular status, the real drivers of behavior modification are the faith related factors that faith institutions contribute. Among these are pastoral leadership, targeted education and training delivered by members of the faith community. Although the contribution by secular professionals, academia, and governmental and non-governmental agencies in partnership with faith institutions are vital, the required behaviour changes are directly determined by the faith related factors. Similarly, Lee et al. find such faith related factors [73] and peer group influence pivotal in delivering behaviour change that in turn deliver the cardiovascular health benefits. The power of the faith institution peers is influential in engagement with resources, acquisition and fine-tuning of skills needed for personal effectiveness, and the psychological benefit of doing these in the context of a supportive community.
## Faith institution based cardiovascular interventions may be promising
Four studies reported process indicators [56, 62–64]. Baig et al. reflected the view of participants on the motivation for behaviour change delivered by trained lay church leaders to assist them with cardiovascular self-management. They reported high satisfaction rates, with $95\%$ of participants reporting that it was important for them to start classes with prayers. Eighty per cent reported that they had learnt a lot about eating a healthy diet. Seventy per cent reported they had learnt a lot from other participants and felt supported. At the end of the intervention programme $70\%$ felt a lot more confident in discussing their health with medical professionals [62]. Similarly, Draper et al. reported acceptability of the intervention among church pastors, the leaders accepting research partnership with professional researchers on behalf of their congregations, was $100\%$. Participant satisfaction also received good ratings. Each of the 19 components of the intervention received a “very useful” rating of between $81.4\%$ and $93.0\%$. The lowest rating of $81.4\%$ was for regular measurement of body weight, and the highest rating of $93.0\%$ was for physical exercise with fellow participants [56].
In Bittman et al. intervention group participants had a significantly higher retention rate of $83.3\%$ compared to control group retention rate of $54.3\%$, and cardiovascular programme completion was 4.21 times higher for the intervention group compared to control group [63]. Dodani et al. reported a good intervention uptake with $91\%$ of participants attending 7 to 9 sessions out of a 12 sessions and $68\%$ attending 10 to 12 sessions [64].
## Faith institutions are complex, flexible environments for cardiovascular health coaching, motivation and behaviour change
Final synthesis of the qualitative findings on the characteristics of faith institution roles, and the quantitative evidence of the effectiveness of those roles is outlined in Table 9. Participation in the social environment of faith institutions and religious rituals appear to be a form of indirect intervention. The most frequent mechanism of direct cardiovascular intervention was Healthy Lifestyle Coaching, Counselling & Motivation Training, used twice as frequently as the combination of Nutrition and Exercise / Physical Activity, and six times as frequently as nutrition intervention alone. The combination of Exercise / Physical Activity with Healthy Lifestyle Coaching, Counselling & Motivation Training was the least frequently used. Faith institutions are complex assets capable of delivering flexible social environments suited to cardiovascular health coaching / motivation and healthy behaviour change. They are amenable to an eclectic combination of intervention techniques and beneficial hypertension linked outcomes.
**Table 9**
| Categories of interventions by component activities / mechanisms (Number of studies) | Intervention agents | Training received by interventions agents | Intervention target | Channel or technique of delivery | Funding | Frequency of intervention activities | Religious component | Indication of intervention effect |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Healthy Lifestyle Coaching, Counselling & Motivation Training12 studies: [34, 56, 59, 61–63, 65–67, 70, 73, 74] | (1) Trusted researchers, health or healthcare professionals, who deliver the intervention and train others [34, 61, 63, 65, 66](2) Volunteering faith institution leaders and members who were health or healthcare professionals, and received intervention specific training [61, 63, 73](3) Volunteering faith institution leaders and members who were not health or healthcare professionals, but received intervention specific training [34, 56, 61, 62, 70, 73](4) Self delivered intervention activities. [59, 67, 74] | Fully trained and registered professionals received intervention specific, religion friendly training. Training received was intervention-specific, and based on learning theories including Adult Learning Principles, Behavioural Counseling, Motivational Interviewing, Social Cognitive Theory, the Trans-theoretical Model and Self-Determination Theory. | Six were executed in urban areas [34, 61–63, 70, 73]; four in both urban and rural areas [56, 59, 66, 67]; and one in a rural area [65].Eight programmes targeted all income levels [34, 59, 61–63, 66, 70, 74] and one study targeted low income [56].Most programmes were directed at mono-ethnic, mono-cultural demographics.Eight targeted Black or African Americans, one targeted Haitian Black Americans, one targeted American Latinos, one targeted Black Africans and one targeted a mixed ethnicity demographic consisting White and Black Americans. | Self-administered reflection or motivation sessions, using devotionals, videos, presentations, animations and reading materials [67, 74].Small group interactive, direct teaching using tools including presentations, videos, manuals reinforced with post session reading materials [59, 61, 63, 66, 70].Classroom teaching, practicals and demonstration sessions [34, 56, 62, 73]. | Seven programmes were government funded [34, 56, 61, 62, 65, 66, 73], four were funded by non-governmental or non-profit organisations [59, 63, 67, 70], and no source of funding was identified for one programme [74]. | Frequency of intervention activities varied by programme. Personal daily intervention activities and weekly individual or small group activities were interspersed with repeated teaching and reinforcements. Programme activities lasted 3 to 4 months and occasionally up to a year. | Clergy endorsed the programmes, supported recruitment drive with public announcements, donated faith institution premises and faith-based social environment.Religious components of two programmes were limited to these [61, 74]. Seven programmes incorporated inspiration from religious scripture in healthy lifestyle lessons, individual meditation, corporate and individual reading, and in music [34, 56, 63, 66, 67, 70, 73].Seven used prayers [34, 56, 59, 63, 67, 70, 73]. Three incorporated spiritual, gospel or religious scripture based songs [56, 63, 66]. | Pooled mean SBP 0.20 mmHg (95%CI -0.44 to +0.49) reduction & DBP 0.04 mmHg (95%CI -0.24 to +0.31) increase at 3 months [34, 62, 63, 68].Pooled mean BP 0.65 mmHg (95%CI -0.91 to -0.39) reduction & DBP 0.53 mmHg (95%CI -1.86 to 0.80) reduction at 12 months [63, 71].A trend of significant SBP [56, 61, 74] and DBP [61, 74] reduction; significant DBP increase in one programme [56]; and a mixed picture of SBP and DBP reduction with Whites achieving more significant reductions than Blacks [70].Pooled mean body weight reduction of 0.12kg (95% CI -0.28 to 0.03) [63, 68, 71] and a polled mean waist circumference reduction of 1.69cm (95% CI -1.74 to -1.64) [62, 63, 71].Significant knowledge increase on the basics of Hypertension Scores p < 0.001 [59].Significant increase (p < .01) in cardiovascular health knowledge scores [65].Significant increase in awareness of hypertension and cardiovascular disease as leading cause of death, p = <0.001 [66].Significant increase in Perceived Personal Control scores, p < 0.0005 [67]. |
| Combination of Nutrition & Exercise / Physical Activity6 studies [64, 69, 71, 72, 75, 76] | Skilled or trainable insiders, and trusted professionals:(1) Research professionals delivering intervention specific training. (Yanek) (Kinard) (Taylor, 2011) (Dodani, 2014, Dodani, 2013, Dodani, 2015)(2) Qualified health, healthcare and allied health professionals volunteering to receive intervention specific training. (Taylor, 2011)(3) Theologically trained pastors and ministers, and faith institution members volunteering to train as health coaches and advisors. (Yanek) (Kinard) (Taylor, 2011) (Dodani, 2014, Dodani, 2013, Dodani, 2015) | Qualified, registered and practising professionals required were trained in intervention specific procedures (Taylor, 2011). All agents received intervention specific training from the research team (Dodani, 2014, Dodani, 2013, Dodani, 2015), training to deliver nutrition and fitness health promotion curriculum (Yanek, 2001, Kinard, 2016), and blood pressure measurement training delivered by qualified healthcare professionals (Kinard, 2016). | All 6 programmes targeted urban adults of all income groups. One programme targeted all ethnic backgrounds (Taylor, 2011) while the rest targeted Blacks or African Americans (Yanek, Kinard, Dodani, 2014, Dodani, 2013, Dodani, 2015). | Interactive teaching sessions enhanced with visual and practical demonstrations. Post teaching workbooks and handouts. | One programme was funded by a non-profit organisation (Taylor, 2011). Two were government funded (Yanek) (Kinard), and three were funded by a publicly funded university. | Intervention activity sessions were delivered on a weekly basis, and lasted from 7 weeks (Taylor, 2011) or 10 weeks (Kinard), to 12 weeks (Dodani, 2014, Dodani, 2013, Dodani, 2015) and 20 weeks (Yanek). One programme offered an option continuation for up to 36 weeks (Yanek). | Group prayers (Yanek, Kinard, Dodani, Dodani, Dodani). Religious scripture based healthy lifestyle discussion. Religious scripture based music (Kinard). | Significant reduction in SBP & DBP [64, 75, 76]; non-significant reduction in SBP & DBP [72] and non-significant DBP increase [69].Intervention group mean daily Sodium intake reduction of 145mg/day in daily compared to Control group mean daily Sodium intake reduction of 8 mg/day, p = 0.0167 [71].Intervention group mean weight decrease of 0.5kg. Control group mean weight increase of 0.38Kg(p = 0.0008) [71].Intervention group mean SBP decrease of 1.6 mmHg compared to control group mean SBP decrease of 0.95, p = 0.47; Intervention group mean DBP decrease of 0.36 mmHg compared to control group mean DBP increase of 0.22, p = 0.49 [71]. |
| [Indirect intervention] The Social faith environment, Religiosity and Religious Rituals3 studies [55, 57, 58] | No direct intervention agents (Sternberg) (Sørensen).Qualified medical professionals (Liu) | Not applicable (Sternberg) (Sørensen).Medical professionals trained to measure blood pressure (Liu) | Receivers of indirect intervention included social and family groups from multiple countries (Sternberg), Tibetan Buddhist Monks undergoing hypertension screening (Liu), and Norwegian church attenders (Sørensen). | No direct intervention delivery | Government agency funding(Sørensen) (Liu) and funding by a combination of government bodies and secular bodies (Sternberg). | Routine and regular exposure to religious environments | Social interactions of groups and families within the faith or religious environment (Sternberg)Passive religious routines and the religious environment (Liu)Religious attendance and the social environment within churches (Sørensen) | Behavior modification relies on faith related factors including the spiritual and socio-cultural awareness of agents; making members of the faith community potentially effective agents [55].Long hours of communal religious rituals and teaching in Tibetan Buddhist settings are associated with a decrease in odds for hypertension [57].Mean SBP & DBP decreased with increasing Religious Attendance [58]. |
| Nutrition alone2 studies [60, 77] | Agents delivering the intervention activities included volunteering faith institution leaders and members who were health or healthcare professionals, and received intervention specific training (Wiist)(Ray) | Fully trained dietician / nutrition professionals and practicing nurses (Ray). Religion compatible nutrition training was given to non-professional volunteers (Wiist). | Both programmes were directed at African Americans, executed in rural and urban areas, and targeted all income levels (Wiist, Ray). | Teaching sessions aided with posters, manuals and post teaching pamphlets and other reading materials (Ray, Wiist) | Wiist was state funded. Funding for ray was not disclosed. | Intervention frequencies were design specific, with intensive contact over 3 weeks of 16-week programme (Ray) and six weekly intervention contacts (Wiist). | Clergy endorsed the programmes. Faith institution assets were used as classrooms (Wiist), and religious scripture was utilised in healthy nutrition lessons (Ray). | Significant reduction in the weight of intervention participants, F(3,12) = 5.29, p<0.05 compared to non-significant weight increase in non-participants. Significant reduction in SBP & DBP [77].Significant cholesterol levels decrease for participants (233.9 to 210.4mg/dl, p<0.0001), similar to cholesterol levels decrease for Controls (241.5 to 202.9mg/dl, p<0.0001) [60]. |
| Combination of Exercise / Physical Activity with Healthy Lifestyle Coaching, Counselling & Motivation Training1 study [68] | Agents delivering the intervention activities included volunteering faith institution leaders and members who were health or healthcare professionals (Tucker) | Training received by intervention agents included project specific coaching and data collection training (Tucker). | The programme was directed at urban African Americans of all income levels (Tucker). | Delivery of learning was via individual coaching and physical activity sessions; and group and panel discussions (Tucker). | Non-profit organisation | Flexible individualised coaching, 6 weekly physical activity sessions, 6 weekly group discussions | Recruitment of intervention agents, endorsement of programme, and recruitment of participants. | Non-significant SBP reduction (2.91mmHg, F = 2.48, p = p = 0.117 and Control group SBP reduction of 1.83 mmHg, F = 1.06, p = 0.303). Non-significant DBP reduction (0.3mmHg -no change-, F = 0.06, p = 0.815 and Control group 2.17 mmHg decrease, F = 3.18, p = 0.076).Non-significant weight decrease (1.69 Lbs., F = 2.95, p = 0.087, and Control group non-significant decrease of 0.97 Lbs., F = 1.07, p = 0.303).Significant increase in Nutrition label literacy (1.2 units, F = 30.89, p<0.001 v Control group non-significant decrease of 0.06 units, F = 0.09, p = 0.76).Significant increase in Healthy Eating score (0.28, F = 26.32, p<0.001 v control group score increase of 0.08, F = 2.69, p = 0.103).Increased Healthy Drinking score (0.88, F = 18.75, p<0.001 v control group score increase of 0.21, F = 1.40, p = 0.239).Increased Physical Activity score (0.30, F = 20.87, p<0.001 v control group Physical Activity score of 0.19, F = 10.95, p<0.01.)Increased overall level of engagement in health- smart behaviours (increase of 0.76, F = 26.47, p < .001 v Control group increase of 0.30, F = 5.33, p = 0.022). |
| Combination of Nutrition & Exercise / Physical Activity & Lifestyle health coaching, counselling, motivation training(No studies) | | | | | | | | |
| Exercise / Physical Activity alone(No studies) | | | | | | | | |
## Quality of the evidence
All the studies contributed to the qualitative synthesis. Two studies were adjudged with serious or high overall risk of bias, however the characteristics of the interventions synthesised are exclusive to the subdomains contributing to those bias ratings. Thus, the qualitative findings, can be held with confidence, as they did not contribute to any adjudged serious / high risk of bias.
## Discussion
To facilitate achievement or maintenance of normal blood pressures, faith institutions purposefully assume roles in cardiovascular health promotion and blood pressure measurement. Deployed in a variety of innovative ways these comprise: cardiovascular health and disease knowledge teaching, with illustrative linking to lifestyle; facilitation of exercise or physical activity as part of normal lifestyle; facilitation of diet and nutrition change beneficial for cardiovascular health; and cardiovascular health linked measurements. In addition faith institution roles include encouragement of personal psychological control, and opportunistic blood pressure checks. Within intervention programmes, these roles tend to be deployed in combination rather than in isolation. Healthy lifestyle coaching, counselling and motivation training was a more frequently employed mechanism compared to physical activity and nutrition or dietary change.
Globally faith institutions have formed partnerships with healthcare systems and contributed to addressing general unmet general healthcare and chronic disease preventative needs of communities [78–83]. Findings of this review are in agreement with the literature on the well-established record of faith institutions contributing to the implementation of public health interventions, including those attempting to influence behaviour change [84–87].
This review sheds light on faith institution roles, identifying the particular characteristics that those roles must posses: characteristics contributory to why and probably how the mechanisms of behaviour change of interventions work. They include relationships of trust, a collective sense of ownership, an informed leadership derived from the faith institution and the local community, sociocultural contextualisation of the roles or their constituent activities, and conformity of such to the practice of faith.
Of particular relevance is the constant pivotal phenomenon of leaders and influencers within these institutions being probably the most determinant resource [88–90]. The literature is largely focused on cultural and spiritual considerations, intervention design, community engagement, implementation strategies for successful outcomes and contributing recommendations [87, 91, 92]. Peterson et al. however, identified the key elements church-based health promotion programmes require to facilitate successful outcomes, albeit without in-depth discussion or investigation of the characteristics of those elements [93]. They include formation of partnerships between faith and health organisations, positive values held by faith organisations and their leaders, availability within faith organisations of adopted or existing health-related services, access of churches to members able to be engaged in health promoting roles, community focus of churches, inherent capacity to promote health behaviour change, and inherent supportive relationships. The identified elements and characteristics of church-based interventions indicate desirable conditions and characteristics for wider faith-based and faith-placed interventions. They overlap in scope with and are in concordance with the findings of this review. There however were until this review, no identified systematic reviews on the characteristics of those roles or elements of interventions generally, or specifically directed to mitigate hypertension within faith institutions.
This review demonstrates agreement with the general trend of transparency and deliberate, purposeful stakeholder engagement through the approach of Community-based participatory research [94]. This approach involves attention to cultural sensitivity, evidenced by a particularly strong representation of faith communities in formulating interventions, training of volunteers, engagement of contractors, and intervention delivery. Using this approach, researchers, academics and clinical practitioners engaged the religious and faith institutions as partners and full participants. For example, in delivering general health education or specific preventative interventions, research practice has evolved to utilise the internal social networks of people of faith and casual attenders at faith institutions [87, 95]. The use of this approach in studies constituting this review is in keeping with evidence from research practice involving interventions for conditions as diverse as hypertension, diabetes, cancers, and HIV-AIDS [88, 96–98]. With respect to the characteristics of roles played by faith institutions, no areas of disagreement were identified between the findings of this review and existing literature. This may be related to the relative limited literature in the area.
The evidence of effectiveness of faith institution roles is evolving. In the randomised controlled studies, after three months of intervention, there was the small benefit of a non-significant systolic blood pressure reduction, albeit with a non-significant increase in diastolic blood pressure. Over the longer duration of 12 months however, there is persistent and significant systolic blood pressure reduction and a non-significant diastolic blood pressure reduction. There were only a few randomised controlled studies contributing to the meta-analyses, indicating that further research is needed. Nonrandomised studies findings are in agreement with the randomised studies, indicating that the interventions have the generally beneficial and significant effect of lowering systolic and diastolic blood pressures. Similarly for body weight and waist circumference, meta-analyses show an overall beneficial albeit non-significant reduction. Non-aggregated evidence from the non-randomised studies show that distinct biological outcomes targeted by the interventions had individually significantly beneficial effects on cardiovascular health and hypertension. That is, they are protective or beneficial for cardiovascular health and hypertension. These findings are in keeping with the well-established record of faith institutions in contributing to implementation of public health interventions. Spirituality and religious participation–both typically facilitated by faith institutions—are known to be capable of both beneficial and harmful influences on health [99]. And there is significant literature support for the favourable effect of religious participation on personal wellbeing [100–102]. For chronic diseases including cardiovascular problems, more favourable outcomes are associated with spirituality [103], faith-based nature of interventions [104] and direct involvement of faith institution in delivery of interventions [25, 105]. For example there is evidence that short-term effectiveness of health behaviour interventions implemented utilising ordinary or typical users of faith institutions, notably church members as volunteer counsellors, advisers and interventionists, lead to increased physical activity and better dietary change outcomes [106–108]. Although there remain knowledge gaps with regard to individual chronic diseases, there is evidence that the environment within faith-organisations is generally favourable to health promotion [109]; and cardiovascular related and obesity interventions implemented within faith-based organisations have been successful in achieving weight reduction, dietary improvement and increased physical activity [110].
The process indicator outcomes in this review may be indicative of potential future reception of interventions facilitated within the faith environment. Where available, participant retention, intervention uptake, acceptability and satisfaction rates were positive. These are probably reflective of the community based participatory approach adopted in constructing the interventions. This approach was a feature common to the studies contributing to this review. Rather than engagement as research specimen, local communities through faith institutions were decision, research, and implementation partners. This aligns with the literature that full community participation is not only beneficial to enhance the quality of research through direct stakeholder facilitation, but also for the relevance and sustainability of any implementation [111–113].
The evidence contributing to this review derived from a plurality of settings where the participants were Black African and African American Christian adults. This is perhaps unsurprising, keeping with longstanding greater frequency and worse outcomes of hypertension in Blacks compared to non-black peoples in the USA [114]; a profile that fits hypertension in African people outside the USA [115]. Especially since African *American religious* settings and leaders have used their influence to address socio-political and health disparities [23, 116], and widely engaged in research and healthcare partnerships to benefit their communities [117–120].
The interventions did not provide for continuation of blood pressure screening or on-going measurements following the end of the programmes of intervention. Similarly, although the benefits of the health promotion were intended to be lasting, there was no evidence of provision for routine or on-going personal measurement of blood pressure beyond the end of the programmes of intervention. Also, despite the blood pressure measurements being used as opportunity to initiate cardiovascular health beneficial behaviour change, there appeared to be lack of emphasis on the importance of routine and on-going blood pressure measurements as part of normal lifestyle. These may be partly due to long term implementation challenges, especially since resource constraints are a key hindrance to the long-term sustainability of community-based cardiovascular interventions including low cost and effective ones [121]. And, sustainability is generally complex and intervention dependent [122].
Access to healthcare systems for research participants was generally not addressed, and there were no attempts to link up or incorporate faith institution health promotion with the wider healthcare systems. This is a probable indication of the relative ease of access to healthcare in the well-resourced countries where most of the studies were conducted. With respect to faith institution facilitated cardiovascular or hypertension intervention, access to health was not addressed as an intrinsic healthcare system issue. This reflects the fact that nongovernmental, faith or philanthropic healthcare is most likely to be part of the structure in low income countries or those with fragile health systems [19, 123]. Similarly there was no emphasis on the integration of faith institution functions with the health system, a reflection of the resource levels and non-reliance on institutions outside the health system, such as faith institutions.
Ultimately, the size or orthodoxy of invested resources may not be the most impactful determinants of the magnitude of cardiovascular health gains achievable within or through the faith environment. But, probably, in the presence of trust, the gains may be ultimately conditional upon processes incorporating deliberate cultural awareness, immersive involvement of the leaders, respect for the local religious context, and compatibility with the continuation of religious worship. These can be tough to achieve without inside knowledge of the explicit characteristics of individual intervention activities that faith institutions, as unique entities, can agree to on behalf of their patrons and the larger society.
## Limitations
There is inherent language bias, as the search was conducted in and all the publications were in English. Similarly, publication bias cannot be ruled out. It is a given that successful programmes and interventions are more likely to achieve publication. All the programmes and interventions reported an overall positive impact, although to varying degrees. Not a single publication reported an overwhelmingly ineffective intervention or programme.
Otherwise, weaknesses indicated in the findings of this review are in five specific areas.
One: *There is* a paucity of cardiovascular research literature specifically dealing with faith institution facilitated hypertension interventions. Two: The absence of studies from Low to middle income countries including the Africa region, Asia and Latin *America is* an important limitation. This is made more important by the fact that religious affiliation and attendance continue to be an important component of social life in those regions. Three: The absence of hypertension intervention studies from Islamic institutions or targeting Muslim congregations constitutes an important limitation. Islam is a major world religion practiced by significant proportion of populations in every WHO region. Islamic inspired established community serving primary and general healthcare programmes do exist, however absence of any that satisfy the inclusion criteria of this review is perhaps indicative of the current dearth of hypertension intervention literature deployed within faith institutions, an underrepresentation of such in the context of the Islamic faith, and perhaps other factors yet to be identified. For example, the authors of this review were unable to successfully obtain hypertension specific programme information from one of such—the Chicago and Atlanta based Inner-City Muslim Action Network (IMAN) [124].
Four: An overwhelming majority of the studies are from the USA, imposing a limitation on sub-group analyses by countries and WHO regions. Five: The predominant context addressed in the included studies is that of Christianity and Christianity affiliated or professing African Americans, who do have some access to various health systems within the USA. This context is not universally applicable.
## Strengths of the review
To the best of our knowledge, this is the first systematic review to focus on the roles played by faith institutions on hypertension and cardiovascular health of adults, and the characteristics of those roles.
A particular strength of the review lies in the use of the diffusion of innovation framework, a naturally occurring social knowledge transfer process nearly indistinguishable from how faith institutions have naturally operated; and hence a most natural fit for the context. The findings of this review indicate generalisability to contexts of religious participation within urban and rural settings involving adults across all income levels.
In addition there are four unique areas of strength demonstrated by the findings of this review. First: Due to the provenance of the included studies, the findings are potentially compatible with: contexts of state funding or charitable organisation funding of hypertension or cardiovascular research; contexts of well organised and funded health research infrastructure; and contexts where healthcare systems prioritise hypertension or cardiovascular research. Second: The findings are potentially particularly helpful in contexts where religious participation is cultural, socially valued or popular. In these contexts faith institutions could potentially be utilised as viable assets or adjuncts for healthcare systems. Third: The findings potentially have strong applicability to settings bearing similarities to the African American socioeconomic and cultural context, including nutritional, cultural and religious legacies. Four: The findings of the review are potentially particularly useful in contexts where barriers to good cardiovascular health are preponderant. These include populations with high rates or religiosity, urban habitation, inadequate healthcare system access, inadequate hypertension awareness, poor nutrition, poor physical activity, cultural insularity or isolation, poverty or low income and remote, rural communities.
## Implications for research
The findings and observations from this review highlight the need for further research in three areas.
One area of research need is exploration of how to systematize, routinize and normalize blood pressure measurement within faith institutions such that users of faith institutions have constant and permanent access to blood pressure measurements outside of and unrelated to any on-going research activity. Another area of research need is exploration of how healthcare systems at different resource levels can utilise the learning from faith institution based cardiovascular health promotion for early diagnosis, management and other relevant intervention. Finally, research is needed on understanding the barriers, facilitators and feasibility of integration of faith institution health promotion into wider healthcare systems. Example of such could be hypertension referral systems, or on-going support for chronic disease or cardiovascular disease preventative activity within faith institutions.
## Implications for cardiovascular intervention within faith communities
This review agrees with current knowledge on hypertension and cardiovascular interventions within faith institutions that: the faith institution environment is potent resource, religiosity and the practice of faith contribute to the achievement of intervention outcomes, and religious leaders are probably the most influential facilitating factor.
This review contributes the following update to the literature. There is a broad range of roles but they probably are most impactful when their characteristics are purposefully adjusted and highly contextualised to the individual faith setting. These done, small effects beneficial to hypertension and cardiovascular health will probably accrue over time. For maximal impact hypertension and cardiovascular interventions have the following requirements: contextualisation to local circumstances, lifestyle linked hypertension and cardiovascular knowledge impartation, designs incorporating psychological empowerment, designs incorporating multiple measurement of cardiovascular health related parameters, opportunistic hypertension checks in addition to other blood pressure measurements, full involvement of locally extracted professionals, faith and community leaders taking responsibility for all aspects of intervention, and implementation without the disruption of religious practice or on-going religiosity.
The findings of the review also suggest that there is the need for involvement in cardiovascular and hypertension research of more diverse faith and religious cultures, traditions and environments. Similarly, there is the need to understand how faith institutions, in seeking involvement in cardiovascular health and hypertension, compare to and can benefit from non-faith-based institutions.
## Conclusion
Pertinent contribution to knowledge and the key messages for communities globally, policy makers, healthcare professionals, researchers and other stakeholders are highlighted in Table 10.
**Table 10**
| Roles of faith institutions |
| --- |
| To assist adults achieve or maintain a normal blood pressure, faith institutions play a variety of roles including: • Cardiovascular health and disease teaching, with direct linking of daily lifestyle to health and disease. • Promotion of, provision for, and persuasion to exercise or increase physical activity as part of normal lifestyle. • Promotion of, provision for, and persuasion toward diet and nutrition change beneficial for cardiovascular health. • Promotion of, provision for, and persuasion to undertake cardiovascular health linked measurements. • Teaching, training and encouragement of personal psychological control. • Promotion of, provision for, persuasion to undergo opportunistic blood pressure checks. |
| Characteristics of the roles |
| Faith institutions activities that assist adults achieve or maintain normal blood pressure: • Are based on relationships of trust with local leadership • Are contributed to by trusted local insiders and leaders • Foster a sense of ownership • Require simplification, transparency, and health-lifestyle-spirituality linking • Are highly contextualised to individual local sociocultural realities • Work alongside and are in conformity with the practice of faith • Involve investment of training in or cooperation with trusted insiders • Are ethical and harmless to participants • Are volitional but consented to by faith and community leaders |
| Evidence of Effectiveness |
| • The evidence for effectiveness is limited. • Faith institution activities cause reductions in systolic and diastolic blood pressures; but these reductions appear to become less significant over time. • With intervention, body weight and waist circumference reduce. Similarly, multiple, health related outcomes are also impacted in a way beneficial for hypertension and cardiovascular health. |
| Other findings |
| • The literature is limited. • Interventions work through multiple components; and the quantification or separation of their relative contributions is not straightforward. • Within intervention programmes, faith institution roles tend to be deployed as combination of mechanisms rather than in isolation. • The most frequent mechanism was Healthy Lifestyle Coaching, Counselling & Motivation Training. • Interventions work in all socioeconomic, faith and resource settings; but more work is required to further understanding. • Current evidence is predominantly based on research on Black African and African Americans in the Christian faith. • Where available evidence shows high and encouraging rates of intervention uptake, retention, satisfaction and acceptability. • Community and peer influences appear to be an important mechanism of achieving beneficial behaviour change. • More research is needed across different faith, resource and health system settings. |
Faith institutions contribute a variety of roles to assist adults achieve or maintain normal blood pressures. Certain characteristics are important for the feasibility and outcome of hypertension and cardiovascular interventions. Most of the evidence derives from settings with predominantly Black African and African American Christian adults. This implies potentially limited generalizability. However, where applicable, for example in contexts of deprivation of cardiovascular preventative health where religious attendance or participation is prevalent, these findings are invaluable for the prospects of developing low cost, effective and sustainable cardiovascular interventions. Such contexts exist throughout low and middle-income countries globally, and also in pockets of deprivation in high-income countries.
However limited, there is evidence of effectiveness of faith institution facilitated interventions. Faith settings may be amenable to tailored cardiovascular coaching and motivation to specifically address hypertension. Contextualised and tailored innovative application of interventions are attainable, with potentially beneficial outcomes.
Although cultural and religious influences on human behaviour vary across communities globally, this review contributes evidence on faith institution roles and the characteristics of those roles for beneficial cardiovascular public health intervention. These are potentially useful for the construction of community based, long-term, meaningful, sustainable, and perhaps permanent interventions.
In addressing the global hypertension epidemic cardiovascular health promotion roles of faith institutions probably hold unrealised potential in research and utility as viable assets or adjuncts to healthcare systems, crucially in low income, religious or underserved communities across different healthcare settings. Utilising available evidence to construct and deliver bespoke, sustainable interventions that remain effective and stand the test of time remains a challenge.
## References
1. Williams B, Mancia G, Spiering W, Agabiti RE, Azizi M, Burnier M. **2018 ESC/ESH Guidelines for the management of arterial hypertension**. *European Heart Journal* (2018.0) **39** 3021. DOI: 10.1093/eurheartj/ehy339
2. Carey RM, Whelton PK. **Committee* AAHGW. Prevention, detection, evaluation, and management of high blood pressure in adults: synopsis of the 2017 American College of Cardiology/American Heart Association Hypertension Guideline**. *Annals of internal medicine* (2018.0) **168** 351-8. PMID: 29357392
3. Volpe M, Gallo G, Modena MG, Ferri C, Desideri G, Tocci G. **Updated recommendations on cardiovascular prevention in 2022: an executive document of the Italian Society of Cardiovascular Prevention.**. *High Blood Pressure & Cardiovascular Prevention* (2022.0) 1-12. PMID: 34855154
4. Seedat YK. **Hypertension in developing nations in sub-Saharan Africa**. *Journal of human hypertension* (2000.0) **14** 739. DOI: 10.1038/sj.jhh.1001059
5. Olatunbosun ST, Kaufman JS, Cooper RS, Bella AF. **Hypertension in a black population: prevalence and biosocial determinants of high blood pressure in a group of urban Nigerians**. *Journal of human hypertension* (2000.0) **14** 249. DOI: 10.1038/sj.jhh.1000975
6. Cooper RS, Amoah AG, Mensah GA. **High blood pressure: the foundation for epidemic cardiovascular disease in African populations.**. *Ethnicity and disease* (2003.0) 13. PMID: 13677414
7. Opie LH, Seedat YK. **Hypertension in sub-Saharan African populations**. *Circulation* (2005.0) **112** 3562-8. DOI: 10.1161/CIRCULATIONAHA.105.539569
8. Addo J, Smeeth L, Leon DA. **Hypertension in sub-saharan Africa: a systematic review**. *Hypertension* (2007.0) **50** 1012-8. DOI: 10.1161/HYPERTENSIONAHA.107.093336
9. Kuller LH. **Epidemic hypertension in sub-Saharan Africa**. *Hypertension* (2007.0) **50** 1004-5. DOI: 10.1161/HYPERTENSIONAHA.107.095620
10. Keates AK, Mocumbi AO, Ntsekhe M, Sliwa K, Stewart S. **Cardiovascular disease in Africa: epidemiological profile and challenges**. *Nature Reviews Cardiology* (2017.0) **14** 273-93. DOI: 10.1038/nrcardio.2017.19
11. Palafox B, McKee M, Balabanova D, AlHabib KF, Bahonar A, Ismail N. **Wealth and cardiovascular health: a cross-sectional study of wealth-related inequalities in the awareness, treatment and control of hypertension in high-, middle-and low-income countries.**. *International journal for equity in health* (2016.0) **15** 199. DOI: 10.1186/s12939-016-0478-6
12. Bielefeld W, Cleveland WS. **Defining faith-based organizations and understanding them through research**. *Nonprofit and Voluntary Sector Quarterly* (2013.0) **42** 442-67
13. Torry M.. *Managing God’s business: Religious and faith-based organizations and their management: Routledge* (2017.0)
14. James R.. **Handle with care: Engaging with faith-based organisations in development**. *Development in Practice* (2011.0) **21** 109-17
15. Haynes J.. *Faith-based organisations at the United Nations* (2013.0)
16. Clarke G.. **Faith matters: faith‐based organisations, civil society and international development. Journal of International Development**. *The Journal of the Development Studies Association* (2006.0) **18** 835-48
17. Ochillo MA, Van Teijlingen E, Hind M. **Influence of faith-based organisations on HIV prevention strategies in Africa: a systematic review**. *African health sciences* (2017.0) **17** 753-61. DOI: 10.4314/ahs.v17i3.18
18. Lipsky A.. **Evaluating the strength of faith: potential comparative advangates of faith-based organizations providing health services in Sub-Saharan Africa**. *Public Admin Dev* (2011.0) **31** 25-36. DOI: 10.1002/pad.586
19. Olivier J, Tsimpo C, Gemignani R, Shojo M, Coulombe H, Dimmock F. **Understanding the roles of faith-based health-care providers in Africa: review of the evidence with a focus on magnitude, reach, cost, and satisfaction**. *The Lancet* (2015.0) **386** 1765-75. DOI: 10.1016/S0140-6736(15)60251-3
20. Brown RK, Adamczyk A. **Racial/ethnic differences in the provision of health-related programs among American religious congregations**. *J Soc* (2009.0) **36** 105
21. Taylor RJ, Ellison CG, Chatters LM, Levin JS, Lincoln KD. **Mental health services in faith communities: The role of clergy in black churches**. *Social work* (2000.0) **45** 73-87. DOI: 10.1093/sw/45.1.73
22. Maluleke TS. **African Christianity as African Religion**. *The quest for liberation and reconciliation: Essays in honor of J. Deotis Roberts* (2005.0) 116
23. Butler-Ajibade P, Booth W, Burwell C. **Partnering with the black church: recipe for promoting heart health in the stroke belt**. *ABNF Journal* (2012.0) **23** 34-7. PMID: 22774357
24. Clarke M, Charnley S, Lumbers J. **Churches, mosques, and condoms: Understanding successful HIV and AIDS interventions by faith-based organisations**. *Development in Practice* (2011.0) **21** 3-17
25. Maynard MJ. **Faith-based institutions as venues for obesity prevention**. *Current obesity reports* (2017.0) **6** 148-54. DOI: 10.1007/s13679-017-0257-8
26. Levin J.. **Faith-based partnerships for population health: challenges, initiatives, and prospects**. *Public Health Reports* (2014.0) **129** 127-31. DOI: 10.1177/003335491412900205
27. Gildea F.. **Are faith leaders a divine intervention when it comes to public health?**. *Perspectives in Public Health* (2021.0) **141** 249-51. DOI: 10.1177/17579139211026542
28. Green A, Shaw J, Dimmock F, Conn C. **A shared mission? Changing relationships between government and church health services in Africa**. *Int J Health Plann Mgmt* (2002.0) **17** 333-53. DOI: 10.1002/hpm.685
29. Oruwari Y.. **New generation churches and the provision of welfare: A gender study from Port Harcourt, Nigeria**. *Associational life in African cities: Popular responses to the urban crisis* (2001.0) 77-89
30. Johnson-Lawrence V, Bailey S, Sanders PE, Sneed R, Angel-Vincent A, Brewer A. **The Church Challenge: A community-based multilevel cluster randomized controlled trial to improve blood pressure and wellness in African American churches in Flint, Michigan**. *Contemporary Clinical Trials Communications* (2019.0) **14** 100329. DOI: 10.1016/j.conctc.2019.100329
31. Su D, Garg A, Wiens J, Meyer E, Cai G. **Assessing Health Needs in African American Churches: A Mixed-Methods Study**. *Journal of Religion & Health* (2019.0) 08. DOI: 10.1007/s10943-019-00924-5
32. Yi SS, Wyatt LC, Patel S, Choy C, Dhar R, Zanowiak JM. **A Faith-Based Intervention to Reduce Blood Pressure in Underserved Metropolitan New York Immigrant Communities**. *Preventing Chronic Disease* (2019.0) **16** E106. DOI: 10.5888/pcd16.180618
33. Skolarus LE, Cowdery J, Dome M, Bailey S, Baek J, Byrd JB. **Reach Out Churches: A Community-Based Participatory Research Pilot Trial to Assess the Feasibility of a Mobile Health Technology Intervention to Reduce Blood Pressure Among African Americans**. *Health Promotion Practice* (2018.0) **19** 495-505. DOI: 10.1177/1524839917710893
34. Schoenthaler AM, Lancaster KJ, Chaplin W, Butler M, Forsyth J, Ogedegbe G. **Cluster Randomized Clinical Trial of FAITH (Faith-Based Approaches in the Treatment of Hypertension) in Blacks**. *Circulation Cardiovascular quality and outcomes* (2018.0) **11** e004691. DOI: 10.1161/CIRCOUTCOMES.118.004691
35. Chukwuma A, Gong E, Latypova M, Fraser-Hurt N. **Challenges and opportunities in the continuity of care for hypertension: a mixed-methods study embedded in a primary health care intervention in Tajikistan.**. *BMC health services research* (2019.0) **19** 925. DOI: 10.1186/s12913-019-4779-5
36. Rifat M, Rusen ID, Mahmud MH, Nayer I, Islam A, Ahmed F. **From mosques to classrooms: mobilizing the community to enhance case detection of tuberculosis**. *American journal of public health* (2008.0) **98** 1550-2. DOI: 10.2105/AJPH.2007.117333
37. Brand DJ. *The healing station: The role of the African American church as a change agent for health [Ph.D.]* (2011.0)
38. Foco RL. *The role of faith leaders in partnerships among health promotion researchers and faith communities [Ph.D.]* (2011.0)
39. Hong QN, Pluye P, Bujold M, Wassef M. **Convergent and sequential synthesis designs: implications for conducting and reporting systematic reviews of qualitative and quantitative evidence**. *Systematic reviews* (2017.0) **6** 1-14. PMID: 28077170
40. Brown CG. *Debating Yoga and Mindfulness in Public Schools: Reforming Secular Education Or Reestablishing Religion?* (2019.0)
41. Bramadat P.. **A bridge too far: yoga, spirituality, and contested space in the Pacific Northwest**. *Religion, State & Society* (2019.0) **47** 491-507
42. Jain AR. **Who is to say modern yoga practitioners have it all wrong? On Hindu origins and yogaphobia**. *Journal of the American Academy of Religion* (2014.0) **82** 427-71
43. Jeter PE, Slutsky J, Singh N, Khalsa SBS. **Yoga as a therapeutic intervention: a bibliometric analysis of published research studies from 1967 to 2013**. *The Journal of Alternative and Complementary Medicine* (2015.0) **21** 586-92. DOI: 10.1089/acm.2015.0057
44. Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. **Rayyan—a web and mobile app for systematic reviews**. *Systematic reviews* (2016.0) **5** 1-10. DOI: 10.1186/s13643-016-0384-4
45. Higgins JP, Savović J, Page MJ, Elbers RG, Sterne JA. **Assessing risk of bias in a randomized trial**. *Cochrane Handbook for Systematic Reviews of Interventions* (2019.0) 205-28
46. Jüni P, Loke Y, Pigott T, Ramsay C, Regidor D, Rothstein H. **Risk of bias in non-randomized studies of interventions (ROBINS-I): detailed guidance**. *Br Med J* (2016.0)
47. 47University of Bristol CfRS, Analysis D. The ROBINS‐E tool (Risk of Bias in Non‐Randomized Studies‐of Exposures). 2018.. *Analysis D. The ROBINS‐E tool (Risk of Bias in Non‐Randomized Studies‐of Exposures)* (2018.0)
48. McInnes E, Wimpenny P. **Using Qualitative Assessment and Review Instrument software to synthesise studies on older people’s views and experiences of falls prevention**. *International Journal of Evidence‐Based Healthcare* (2008.0) **6** 337-44. DOI: 10.1111/j.1744-1609.2008.00104.x
49. Pearson A.. **Balancing the evidence: incorporating the synthesis of qualitative data into systematic reviews**. *JBI reports* (2004.0) **2** 45-64
50. Pluye P, Gagnon M-P, Griffiths F, Johnson-Lafleur J. **A scoring system for appraising mixed methods research, and concomitantly appraising qualitative, quantitative and mixed methods primary studies in mixed studies reviews**. *International journal of nursing studies* (2009.0) **46** 529-46. DOI: 10.1016/j.ijnurstu.2009.01.009
51. 51Cochrane. Review Manager (RevMan) [Computer program]. Version 5.4. The Cochrane Collaboration, 2020.: The Cochrane Collaboration; 2020.. *Review Manager (RevMan) [Computer program]. Version 5.4* (2020.0)
52. Zimmermann K.. *The Role of Context in the Implementation of a Rural, Church-Based, CVD Risk-Reduction Intervention [Ph.D.]* (2019.0)
53. Author U.. **Churches as an avenue for the control of arterial hypertension**. *Servir* (1990.0) **38** 201-3. PMID: 2237503
54. S Harding CE, Gobin R, Read U, Goberdhan S, Perreira O, Sharma M. **Perspectives from the congregations taking action against non-communicable diseases (CONTACT) study: Opportunities and challenges for a paradigm shift in strengthening primary healthcare systems**. *West Indian Med J* (2019.0) **68** 44
55. Sternberg Z, Munschauer FE. **Faith-placed cardiovascular health promotion: a framework for contextual and organizational factors underlying program success**. *Health Education Research* (2007.0) **22** 619-29. DOI: 10.1093/her/cyl124
56. Draper CE, Tomaz SA, Zihindula G, Bunn C, Gray CM, Hunt K. **Development, feasibility, acceptability and potential effectiveness of a healthy lifestyle programme delivered in churches in urban and rural South Africa**. *PloS one* (2019.0) **14** e0219787. DOI: 10.1371/journal.pone.0219787
57. Liu K, Xu Y, Wang S, Shi R, Gong S, Li X. **Buddhist Activities related to Sedentary behavior and Hypertension in Tibetan monks**. *Journal of human hypertension* (2018.0) **33** 756-62. DOI: 10.1038/s41371-018-0136-0
58. Sørensen T, Danbolt LJ, Lien L, Koenig HG, Holmen J. **The relationship between religious attendance and blood pressure: the HUNT Study, Norway**. *International journal of psychiatry in medicine* (2011.0) **42** 13-28. DOI: 10.2190/PM.42.1.b
59. Benjamin L.. *Culturally Appropriate Hypertension Education for Haitian Immigrants: A Community- and Faith-Based Health Promotion Program [D.N.P.]* (2017.0)
60. Wiist WH, Flack JM. **A church-based cholesterol education program**. *Public Health Reports* (1990.0) **105** 381. PMID: 2116640
61. Smith ED, Merritt SL, Patel MK. **Church-based education: an outreach program for African Americans with hypertension**. *Ethnicity & health* (1997.0) **2** 243-53. DOI: 10.1080/13557858.1997.9961832
62. Baig AA, Benitez A, Locklin CA, Gao Y, Lee SM, Quinn MT. **Picture Good Health: A Church-Based Self-Management Intervention Among Latino Adults with Diabetes.**. *Journal of general internal medicine* (2015.0) **30** 1481-90. DOI: 10.1007/s11606-015-3339-x
63. Bittman B, Poornima I, V MA, Heidel RE. **Gospel Music: a Catalyst for Retention, Engagement, and Positive Health Outcomes for African Americans in a Cardiovascular Prevention and Treatment Program**. *Advances in mind-body medicine* (2020.0) **34** 8-16. PMID: 32277749
64. Dodani S, Arora S, Sealey-Potts C, Christie C, Aldridge P, Kraemer D. **Healthy eating and living spiritually (HEALS): An efficacious model for hypertension control in american african churches**. *Hypertension* (2014.0) **64**
65. Abbott LS. *The effect of a culturally relevant cardiovascular health promotion program on rural African Americans [Ph.D.]* (2015.0)
66. Spell-LeSane DD. *A church based cardiovascular disease awareness program for African American women on increasing awareness behaviors and physical activity [D.P.H.].* (2016.0)
67. Daye GC. *Effects of a Faith-based Health Devotional on Illness Representation of High Blood Pressure in African Americans [Ph.D.].* (2019.0)
68. Tucker C M., Kang S, Ukonu NA, Linn GS, DiSangro CS, Arthur TM. **A Culturally Sensitive Church-Based Health-Smart Intervention for Increasing Health Literacy and Health-Promoting Behaviors among Black Adult Churchgoers**. *Journal of Health Care for the Poor and Underserved* (2019.0) **30** 80-101. DOI: 10.1353/hpu.2019.0009
69. Taylor C.. *Implementing prevention and wellness workshops for individual education and community outreach: A pilot project for Way of Faith Assembly of God Church* (2011.0)
70. Oexmann MJ, Ascanio R, Egan BM. **Efficacy of a church-based intervention on cardiovascular risk reduction**. *Ethnicity & disease* (2001.0) **11** 817-22. PMID: 11763306
71. Yanek LR, Becker DM, Moy TF, Gittelsohn J, Koffman DM. **Project Joy: faith based cardiovascular health promotion for African American women**. *Public health reports (Washington, DC: 1974)* (2001.0) **116** 68-81. DOI: 10.1093/phr/116.S1.68
72. Kinard JQ. *The my sister’s keeper’s project: A spiritually-based weight loss intervention program for African American women, ages 25–64 [D.P.H.]* (2016.0)
73. Lee S, Schorr E, Hadidi NN, Kelley R, Treat-Jacobson D, Lindquist R. **Power of Peer Support to Change Health Behavior to Reduce Risks for Heart Disease and Stroke for African American Men in a Faith-Based Community**. *Journal of racial and ethnic health disparities* (2018.0) **5** 1107-16. DOI: 10.1007/s40615-018-0460-7
74. White HL. *The Effectiveness of a Church-Based Self-Management Program in Reducing Blood Pressure among African American Adults [D.N.P.].* (2018.0)
75. Dodani S, Arora S, Kraemer D. **Heals hypertension control program for stroke prevention in african american communities**. *Stroke* (2015.0) **46**
76. Dodani S, Arora S, Mahin C, Butterfield R. **Heals program implementation in african american churches for hypertension control and prevention**. *Hypertension* (2013.0) **62**
77. Ray CR. **Effectiveness of a church-based nutrition intervention among African American women**. *Dissertation Abstracts International: Section B: The Sciences and Engineering* (2003.0) **64** 4291
78. Harmon BE, Schmidt M, Escobar F, Steele A. **Filling the Gaps: The Role of Faith-Based Organizations in Addressing the Health Needs of Today’s Latino Communities**. *Journal of religion and health* (2021.0) **60** 1198-213. DOI: 10.1007/s10943-020-01002-x
79. Kiger AM, Fagan DM, Van Teijlingen ER. *Faith Communities and the Potential for Health Promotion* (2017.0)
80. Tomlinson S.. **Church-based Health Promotion in Jamaica for the Ageing Population**. *West Indian Medical Journal* (2018.0) **67**
81. Banerjee AT, Landry M, Zawi M, Childerhose D, Stephens N, Shafique A. **A pilot examination of a mosque-based physical activity intervention for South Asian Muslim women in Ontario, Canada**. *Journal of immigrant and minority health* (2017.0) **19** 349-57. DOI: 10.1007/s10903-016-0393-3
82. Islam N, Patel S. **Best practices for partnering with ethnic minority-serving religious organizations on health promotion and prevention**. *AMA journal of ethics* (2018.0) **20** E643. DOI: 10.1001/amajethics.2018.643
83. Levin J.. **Faith-based initiatives in health promotion: history, challenges, and current partnerships**. *American Journal of Health Promotion* (2014.0) **28** 139-41. DOI: 10.4278/ajhp.130403-CIT-149
84. Catanzaro AM, Meador KG, Koenig HG, Kuchibhatla M, Clipp EC. **Congregational health ministries: A national study of pastors’ views**. *Public Health Nursing* (2007.0) **24** 6-17. DOI: 10.1111/j.1525-1446.2006.00602.x
85. Westberg GE, McNamara JW. *The parish nurse: Providing a minister of health for your congregation:* (1990.0)
86. Levin J.. **Partnerships between the faith-based and medical sectors: implications for preventive medicine and public health**. *Preventive medicine reports* (2016.0) **4** 344-50. DOI: 10.1016/j.pmedr.2016.07.009
87. Campbell MK, Hudson MA, Resnicow K, Blakeney N, Paxton A, Baskin M. **Church-based health promotion interventions: evidence and lessons learned**. *Annu Rev Public Health* (2007.0) **28** 213-34. DOI: 10.1146/annurev.publhealth.28.021406.144016
88. Dodani S, Sullivan D, Pankey S, Champagne C. **HEALS: a faith-based hypertension control and prevention program for African American churches: training of church leaders as program interventionists**. *International Journal of Hypertension* (2011.0) **2011** 820101. DOI: 10.4061/2011/820101
89. Bopp M, Baruth M, Peterson JA, Webb BL. **Leading their flocks to health? Clergy health and the role of clergy in faith-based health promotion interventions**. *Family & community health* (2013.0) **36** 182-92. DOI: 10.1097/FCH.0b013e31828e671c
90. Baruth M, Bopp M, Webb BL, Peterson JA. **The role and influence of faith leaders on health-related issues and programs in their congregation**. *Journal of Religion and Health* (2015.0) **54** 1747-59. DOI: 10.1007/s10943-014-9924-1
91. Bopp M, Peterson JA, Webb BL. **A comprehensive review of faith-based physical activity interventions**. *American Journal of Lifestyle Medicine* (2012.0) **6** 460-78
92. Demark-Wahnefried W, McClelland JW, Jackson B, Campbell MK, Cowan A, Hoben K. **Partnering with African American churches to achieve better health: lessons learned during the Black Churches United for Better Health 5 a day project**. *Journal of Cancer Education* (2000.0) **15** 164-7. DOI: 10.1080/08858190009528686
93. Peterson J, Atwood JR, Yates B. **Key elements for church‐based health promotion programs: outcome‐based literature review**. *Public Health Nursing* (2002.0) **19** 401-11. DOI: 10.1046/j.1525-1446.2002.19602.x
94. Wallerstein NB, Duran B. **Using community-based participatory research to address health disparities**. *Health promotion practice* (2006.0) **7** 312-23. DOI: 10.1177/1524839906289376
95. DeHaven MJ, Hunter IB, Wilder L, Walton JW, Berry J. **Health programs in faith-based organizations: are they effective?**. *American journal of public health* (2004.0) **94** 1030-6. DOI: 10.2105/ajph.94.6.1030
96. Austin SA, Claiborne N. **Faith wellness collaboration: a community-based approach to address type II diabetes disparities in an African-American community**. *Social work in health care* (2011.0) **50** 360-75. DOI: 10.1080/00981389.2011.567128
97. Bone L, Edington K, Rosenberg J, Wenzel J, Garza MA, Klein C. **Building a navigation system to reduce cancer disparities in urban black older adults**. *Progress in community health partnerships: research, education, and action* (2013.0) **7** 209. PMID: 23793252
98. Griffith DM, Pichon LC, Campbell B, Allen JO. **YOUR Blessed Health: a faith-based CBPR approach to addressing HIV/AIDS among African Americans**. *AIDS Education and Prevention* (2010.0) **22** 203-17. DOI: 10.1521/aeap.2010.22.3.203
99. Alves RRdN Alves HdN, Barboza RRD Souto WdMS. **The influence of religiosity on health**. *Ciência & Saúde Coletiva* (2010.0) **15** 2105-11. PMID: 20694332
100. Chatters LM. **Religion and health: Public health research and practice**. *Annual review of public health* (2000.0) **21** 335-67. DOI: 10.1146/annurev.publhealth.21.1.335
101. Ellison CG. **Religious involvement and subjective well-being**. *Journal of health and social behavior* (1991.0) 80-99. PMID: 2007763
102. Levin JS, Chatters LM. **Religion, health, and psychological well-being in older adults: Findings from three national surveys**. *Journal of aging and health* (1998.0) **10** 504-31. DOI: 10.1177/089826439801000406
103. Abu HO, Ulbricht C, Ding E, Allison JJ, Salmoirago-Blotcher E, Goldberg RJ. **Association of religiosity and spirituality with quality of life in patients with cardiovascular disease: a systematic review**. *Quality of Life Research* (2018.0) **27** 2777-97. DOI: 10.1007/s11136-018-1906-4
104. Sattin RW, Williams LB, Dias J, Garvin JT, Marion L, Joshua TV. **Community trial of a faith-based lifestyle intervention to prevent diabetes among African-Americans**. *Journal of community health* (2016.0) **41** 87-96. DOI: 10.1007/s10900-015-0071-8
105. Tristão Parra M, Porfírio GJ, Arredondo EM, Atallah ÁN. **Physical activity interventions in faith-based organizations: a systematic review**. *American Journal of Health Promotion* (2018.0) **32** 677-90. DOI: 10.1177/0890117116688107
106. Resnicow K, Campbell MK, Carr C, McCarty F, Wang T, Periasamy S. **Body and soul: a dietary intervention conducted through African-American churches**. *American journal of preventive medicine* (2004.0) **27** 97-105. PMID: 15261895
107. Wilcox S, Laken M, Bopp M, Gethers O, Huang P, McClorin L. **Increasing physical activity among church members: community-based participatory research**. *American journal of preventive medicine* (2007.0) **32** 131-8. DOI: 10.1016/j.amepre.2006.10.009
108. Bopp M, Fallon E. **Community-based interventions to promote increased physical activity**. *Applied health economics and health policy* (2008.0) **6** 173-87. PMID: 19382818
109. Wilcox S, Saunders RP, Kaczynski AT, Forthofer M, Sharpe PA, Goodwin C. **Faith, activity, and nutrition randomized dissemination and implementation study: countywide adoption, reach, and effectiveness**. *American journal of preventive medicine* (2018.0) **54** 776-85. DOI: 10.1016/j.amepre.2018.02.018
110. Lancaster K, Carter‐Edwards L, Grilo S, Shen C, Schoenthaler A. **Obesity interventions in A frican A merican faith‐based organizations: a systematic review**. *Obesity Reviews* (2014.0) **15** 159-76
111. O’Fallon LR, Dearry A. **Community-based participatory research as a tool to advance environmental health sciences**. *Environmental health perspectives* (2002.0) **110** 155-9. DOI: 10.1289/ehp.02110s2155
112. Morgan D, Crossley M, Stewart N, Kirk A, Forbes D, D’Arcy C. **Evolution of a community-based participatory approach in a rural and remote dementia care research program**. *Progress in community health partnerships: research, education, and action.* (2014.0) **8** 337. DOI: 10.1353/cpr.2014.0040
113. Leung MW, Yen IH, Minkler M. **Community based participatory research: a promising approach for increasing epidemiology’s relevance in the 21st century**. *International journal of epidemiology* (2004.0) **33** 499-506. DOI: 10.1093/ije/dyh010
114. Cooper RS, Liao Y, Rotimi C. **Is hypertension more severe among US blacks, or is severe hypertension more common?**. *Annals of Epidemiology* (1996.0) **6** 173-80. PMID: 8827151
115. Ashaye MO, Giles WH. **Hypertension in Blacks: a literature review**. *Ethnicity & disease* (2003.0) **13** 456-62. PMID: 14632264
116. Isaac EP, Rowland ML, Blackwell LE. **Fighting health disparities: The educational role of the African American church**. *CrossCurrents* (2007.0) 261-5
117. Chatters LM, Levin JS, Ellison CG. **Public health and health education in faith communities**. *Health Education & Behavior.* (1998.0) **25** 689-99. DOI: 10.1177/109019819802500602
118. Sutherland M, Hale CD, Harris GJ. **Community health promotion: the church as partner**. *Journal of Primary Prevention* (1995.0) **16** 201-16. DOI: 10.1007/BF02407340
119. Davis DS, Goldmon MV, Coker-Appiah DS. **Using a community-based participatory research approach to develop a faith-based obesity intervention for African American children**. *Health promotion practice* (2011.0) **12** 811-22. DOI: 10.1177/1524839910376162
120. Alio AP, Lewis CA, Bunce CA, Wakefield S, Thomas WG, Sanders E. **Capacity building among African American faith leaders to promote HIV prevention and vaccine research**. *Progress in Community Health Partnerships: Research, Education, and Action* (2014.0) **8** 305-16. DOI: 10.1353/cpr.2014.0050
121. O’Loughlin J, Renaud L, Richard L, Gomez LS, Paradis G. **Correlates of the sustainability of community-based heart health promotion interventions**. *Preventive medicine* (1998.0) **27** 702-12. DOI: 10.1006/pmed.1998.0348
122. Pluye P, Potvin L, Denis J-L. **Making public health programs last: conceptualizing sustainability**. *Evaluation and program planning* (2004.0) **27** 121-33
123. Schumann C, Stroppa A, Moreira-Almeida A. **The contribution of faith-based health organisations to public health**. *International Psychiatry* (2011.0) **8** 62-4. PMID: 31508087
124. **Network I-CMA**. *Inner-City Muslim Action Network*
|
---
title: Peripheral bone structure, geometry, and strength and muscle density as derived
from peripheral quantitative computed tomography and mortality among rural south
Indian older adults
authors:
- Guru Rajesh Jammy
- Robert M. Boudreau
- Iva Miljkovic
- Pawan Kumar Sharma
- Sudhakar Pesara Reddy
- Susan L. Greenspan
- Anne B. Newman
- Jane A. Cauley
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022329
doi: 10.1371/journal.pgph.0000333
license: CC BY 4.0
---
# Peripheral bone structure, geometry, and strength and muscle density as derived from peripheral quantitative computed tomography and mortality among rural south Indian older adults
## Abstract
Multiple studies have observed a relationship of bone mineral density (BMD) measured by Dual energy X-ray absorptiometry (DXA) and mortality. However, areal BMD (aBMD) measured by DXA is an integrated measure of trabecular and cortical bone and does not measure the geometry of bone. Peripheral Quantitative Computed Tomography (pQCT) provides greater insights on bone structure, geometry and strength. To examine whether higher bone phenotypes and muscle density as measured by pQCT are associated with a lower all-cause mortality, we studied 245 men and 254 women (all age >60) recruited in the Mobility and Independent Living among Elders Study in rural south India. Cox proportional hazards models estimated hazard ratios (HR [$95\%$ Confidence Intervals]). After an average follow-up of 5.3 years, 73 men and 50 women died. Among men, trabecular volumetric bone mineral density (vBMD) of radius (HR per SD increase in parameter = 0.59 [0.43, 0.81]) and tibia (0.60[0.45, 0.81]), cortical vBMD of radius (0.61, [0.47, 0.79]) and tibia (0.62, [0.49, 0.79]), cortical thickness of radius (0.55, [0.42, 0.7]) and tibia (0.60, [0.47, 0.77]), polar strength strain index (SSIp) of tibia (0.73 [0.54, 0.98]), endosteal circumference of radius (1.63, [1.25, 2.12]) and tibia (1.54, [1.19, 1.98]) were associated with all-cause mortality. Muscle density (0.67, [0.51, 0.87]) was associated with lower mortality in men. Among women cortical vBMD of radius (0.64, [0.47, 0.87]) and tibia (0.60 [0.45, 0.79]), cortical thickness of radius (0.54, [0.37, 0.79]) and tibia (0.43, [0.30, 0.61]), SSIp of radius (0.70 [0.48,1.01]) and tibia (0.58 [0.37, 0.90]) and endosteal circumference of radius (1.33 [0.97, 1.82]) and tibia (1.83, [1.37, 2.45]) were associated with all-cause mortality. Among men, gait speed mediated the association of muscle density and mortality but there was no mediation for any bone parameters. Conclusion: pQCT bone measures and muscle density were independently associated with mortality among rural south Indian elders.
## Introduction
Multiple studies have observed a relationship of bone mineral density (BMD) measured by Dual energy X-ray absorptiometry (DXA) and mortality [1–6]. A meta-analysis of 10 prospective studies with 46182 participants from 5 countries (US, Netherlands, Sweden, Australia, and Brazil) with median follow up of 7 years, showed an increased all-cause mortality of 1.2 fold (Hazard Ratio (HR) 1.20; $95\%$CI 1.09–1.30) per one standard deviation (SD) decrease in BMD [7]. However, areal BMD (aBMD) measured by DXA is an integrated measure of trabecular and cortical bone and does not measure the geometry of bone. Peripheral Quantitative Computed Tomography (pQCT) provides greater insights on bone structure, geometry and strength.
Among older adults, muscle mass alone cannot fully explain the loss of physical function and muscle strength with age suggesting that there are other aspects of muscle quality which may contribute. Myosteatosis, an excess deposit of fat in the skeletal muscles both at intramuscular and intermuscular levels [8, 9], has been linked with decreased muscular function and physical performance [10–12], aging [13], reduced muscle strength [14], increased hip fractures / fragility fractures [15, 16] and increased mortality [17].
Muscle strength measured using grip strength has been extensively studied in relationship to mortality. A meta-analysis of 42 studies with 3,002,203 participants, including one study from India, showed mortality risk of 1.16- fold ($95\%$ CI 1.12–1.20) for every 5 kg decrease in grip strength [18]. Gait speed a measure of lower limb strength and function has also been linked with mortality: a pooled analysis of 34,485 community dwelling older adults followed for 6 to 21 years reported that for every 0.1 m/s increase of gait speed survival was improved by $12\%$ ($95\%$ CI 10–$13\%$) [19].
Little is known about pQCT derived bone and muscle measures among non-European older adults. The current demographic trends in India predict an increase in the older population, despite a lower life expectancy compared to developed nations and with subsequent higher disability rates. To our knowledge, there have been no studies on the relationship of pQCT bone measures and muscle density with mortality among this high risk population of rural Indian elders.
In this analysis, we tested the hypothesis that higher pQCT bone phenotypes and muscle density are associated with decreased mortality independent of confounding factors and that these associations may differ by sex and be at least partially mediated by grip strength and gait speed.
## Study population
The Mobility and Independent Living Among Elders Study (MILES) is a prospective study which enrolled 562 community dwelling men and women aged 60 years and over, from Medchal region of Telangana state of southern India. MILES was designed to determine the prevalence of age-related chronic diseases, disability and to examine the extent of clinical and subclinical disease [20]. The study was approved by institutional review board at the participating institutions and all subjects provided written informed consent.
At baseline, two visits for data collection were conducted. During the first visit (February 2012-November 2012), participants completed questionnaires including information on health status, smoking, alcohol consumption etc., physical performance tests, fasting glucose tests and blood pressure measurements. At the second baseline visit (June 2012-June 2013), pQCT scans were conducted. Of the total 562 participants recruited, 17 died (11 men, 6 women) between the baseline visit 1 and visit 2; 27 participants (15 men and 12 women) were physically not able to come for pQCT measurement; 3 women moved out of the area and 4 participants (3 men and 1 women) refused to continue in the study. The pQCT scans were conducted on 511 participants; of which, 499 had valid scan data (245 men and 254 women).
## pQCT and calibration
pQCT scans on the radius and tibia were performed using the Stratec XCT-2000 (Stratec Medizintechnik, Pforzheim, Germany). Technicians followed a standardized protocol for positioning and scanning of each subject. Scans were taken at $4\%$ and $33\%$ of the length of radius and at $4\%$, $33\%$ and $66\%$ of the length of tibia. Subject scans were repeated if artefacts due to motion or beam hardening were present. To monitor the stability of the pQCT scanners, a manufacturer supplied cylindrical Quality Assurance (QA) phantom was scanned daily before subject scans were acquired. All pQCT scans were analysed by a single investigator using the manufacturer software package version 6.00 for the XCT scanners. This software provides a suite of segmentation options to quantify total, trabecular and cortical bone properties from each pQCT image. Before each image was analysed, it was checked for artifacts due to motion or beam hardening; scans with artefacts were not analysed. All $4\%$ radius and tibia scans were analysed using the CALCBD option with an automatic gradient search (contour mode 2) applied to segment bone from the soft tissue background and concentric peeling (peelmode 1, $45\%$) to segment trabecular and cortical bone. Proximal scans acquired at the $33\%$ and $66\%$ limb locations were segments using a fixed threshold of 710 mg/cm3(Cortmode1). Coefficients of variation (CVs) were determined for pQCT scans by replicating measurements on 15 subjects (CV ≤ $2.1\%$).
## pQCT parameters
For this analysis, we focused on the following pQCT parameters: at the $4\%$ site of radius and tibia—trabecular vBMD; at the $33\%$ sites of radius and tibia—cortical vBMD, cortical thickness, endosteal circumference and polar strength strain index (SSIp); at $66\%$ tibia–muscle density. These parameters were chosen because vBMD is an indicator of bone matrix mineralisation and represents the mechanical quality of the solid bone tissue both at the trabecular and cortical sites. Endosteal circumference and cortical thickness represent bone geometry and strength. SSIp predicts the failure load [21, 22] and has been shown to be a good predictor of long bone bending [22]. All these parameters also have age-related changes due to adaption of stress, strain, and load on the bone and fractures [23, 24]. Muscle density serves as a surrogate marker of fat infiltration within the muscle [25] and reflects the compactness of muscle fibers, protein within the muscle and other soft tissues and can be viewed as a proxy measure of muscle quality.
## Mortality assessment
As a part of the MILES cohort data collection, each participant was followed for death through community health volunteers at the village level. Based on the information on death from the community health volunteers, MILES research staff visited the participants’ home to ascertain the event of death and administers a verbal autopsy tool.
All deaths until March 31, 2019 were considered in this analysis. There were total of 123 deaths (men $\frac{73}{245}$; women $\frac{50}{254}$); the maximum time to death was 76.7 months since the baseline pQCT measurement. The average follow-up was 64.2 months (5.3 years).
## Covariates
Information on covariates was collected through interviewer-administered questionnaires. Self-reported health status was categorised as good/excellent and fair/poor/very poor. Smoking status was categorised as current smoker and not a current smoker. Alcohol consumption was categorised as consumes alcohol and does not consume alcohol. Self-report of the history of stroke was recorded. Direct measures of weight using SECCA® scale was recorded. Height was measured using a SECCA® stadiometer. Body Mass Index (BMI) was calculated as body weight in kilograms divided by height in meters squared. Diabetes was categorised as present if glucose levels were ≥126mg/dL (after a minimum of an 8-hour fast), self-report of diabetes or used insulin or hypoglycemic medications. Hypertension was considered present if participant self-reported hypertension, reported use of an anti-hypertensive medication or blood pressure assessment (≥ $\frac{140}{90}$ mm of Hg). Activities of daily living (ADL) were assessed using the standard tasks of eating, dressing, bathing, transferring from bed to chair and using the toilet. ADL was categorised as ADL disability if the participant reported difficulty in any one of the tasks. Information on history of fractures (hip, arm, wrist shoulder, spine and any other bones) over the past 5 years was obtained. Serum 25-hydroxy vitamin D levels (ng/ml) were measured using the high-performance liquid chromatography (HPLC) method. Grip strength was measured twice in each hand using a hand-held dynamometer. In this analysis, we used the average of the two readings of the participants’ dominant hand. The Short Physical Performance Battery (SPPB) consists of a group of physical measures used to predict disability among older populations [26] and combines gait speed, balance tests and chair stands with a total score ranging from 0 (worst performance) to 12 (best performance). The 4-meter timed walk was performed twice for each participant as a part of the SPPB. Gait speed was calculated as the average in meters per second. All covariate information was collected during the first baseline visit.
## Statistical analysis
All data underlying the findings described in the manuscript are attached as a supplement.
There were significant differences in the pQCT parameters between men and women; hence, sex specific analyses were done. Serum 25-hydroxy vitamin D values were not normally distributed and were log transformed. We described the participant’s characteristics at the baseline visit using means ± SD or prevalence. Two-sample t-tests or Wilcoxon rank sum tests (continuous variables) or chi-square tests (categorical variables) were used to compare characteristics between men and women and between survivors and deceased participants. The pQCT parameters were categorized as quartiles and Kaplan-Meier survival curves with log rank tests were conducted. Cox proportional hazards models were used to assess the association between the pQCT measures and all-cause mortality; hazards ratio and $95\%$ confidence intervals were calculated. The covariates of interest were age, height, weight, current smoker, consumes alcohol, self-reported health status, hypertension, diabetes, stroke, 25-hydroxy vitamin D level and ADL disability. The minimally adjusted model (model 1) included age, height and weight and fully adjusted model (model 2), model 1 + smoking, alcohol, health status opinion, hypertension, diabetes, stroke, log 25-hydroxy vitamin D and ADL disability. All models with muscle density were also adjusted for muscle cross sectional area (CSA). The proportional hazards assumption was confirmed by Supremum test.
As grip strength and gait speed are associated with mortality and these are in the causal pathway of bone / muscle and mortality, a causal mediation analysis was conducted. The mediator models were model 3, model 2 + grip strength and, model 4 = model 2 + gait speed. Mediation was considered present if there was attenuation of hazards ratio of more than $10\%$ in the models with and without the mediator variables (models 3 and 4 compared to model 2). The pQCT parameters in the mediator models which had significant association with all-cause mortality were considered for the causal mediation analysis. The overall proportion of mediation was considered in the causal mediation analysis. Results were considered statistically significant when a p-value was less than 0.05. All statistical analyses were carried out using Stata/IC 13.1 and SAS 9.4 software.
## Baseline characteristics
The baseline characteristics comparing 245 men and 254 women, and comparing men and women who survived versus deceased are presented in the Table 1. The mean age of the men (68.2 ± 6.62) was similar to women (67.2 ± 6.21). Men compared to women were significantly taller, heavier, had lower BMI, higher waist circumference, lower hip circumference, higher grip strength, faster gait speed, higher SPPB score, higher prevalence of completion of 400-meter walk test, currently smoking and alcohol consumption.
**Table 1**
| Unnamed: 0 | Total participants | Total participants.1 | Total participants.2 | Men | Men.1 | Men.2 | Women | Women.1 | Women.2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Baseline characteristics | Men (N = 245) | Women (N = 254) | p value (Men vs Women) | Alive (N = 172) | Deceased (N = 73) | p value (alive vs deceased) | Alive (N = 204) | Deceased (N = 50) | p value (alive vs deceased) |
| Age (years) | 68.2 ± 6.62 | 67.2 ± 6.21 | 0.0749 | 67.3 ± 6.47 | 70.32 ± 6.54 | 0.0003* | 66.59 ± 5.68 | 69.68 ± 7.59 | 0.01* |
| Height (cm) | 160.62 ± 5.6 | 147.01 ± 5.95 | <0.0001* | 160.06 ± 5.3 | 161.92± 6.09 | 0.0209* | 147.37 ± 5.59 | 145.54 ± 7.12 | 0.18 |
| Weight (kg) | 55.87 ± 11.51 | 49.96 ± 12.21 | <0.0001* | 55.98 ± 11.34 | 55.62± 11.97 | 0.682 | 50.81 ± 12.1 | 46.53 ± 12.12 | 0.02* |
| BMI (kg/m2) | 21.58 ± 3.92 | 22.98 ± 4.8 | 0.0045* | 21.79 ± 3.97 | 21.09 ± 3.78 | 0.204 | 23.28 ± 4.85 | 21.76 ± 4.48 | 0.047* |
| Vitamin D (ng/ml) | 30.62± 16.15 | 29.35 ± 17.65 | 0.1752 | 30.73 ± 16.21 | 30.36± 16.11 | 0.69 | 28.58±18.36 | 32.76±13.68 | 0.026* |
| Average Gait speed (m/s) | 0.69 ± 0.18 | 0.58 ± 0.16 | <0.0001* | 0.72 ± 0.18 | 0.62 ± 0.18 | <0.0001* | 0.61 ± 0.15 | 0.46 ± 0.17 | <0.0001* |
| Average Grip strength (kg) | 20 ± 6.04 | 12.45 ± 4.74 | <0.0001* | 20.99 ± 5.91 | 17.68 ± 5.76 | 0.0003* | 12.86 ± 4.52 | 10.76 ± 5.29 | 0.01* |
| SPPB score | 8.78 ± 2.75 | 7.22 ± 2.82 | <0.0001* | 9.39 ± 2.44 | 7.33 ± 2.89 | <0.0001* | 7.74 ± 2.59 | 5.14 ± 2.78 | <0.0001* |
| Follow up time (years) | 5.22 ± 1.66 | 5.47 ± 1.5 | 0.0603 | 6.12 ± 0.33 | 3.12 ± 1.64 | <0.0001* | 6.12 ± 0.32 | 2.84 ± 1.56 | <0.0001* |
| Health status (Good) | 113 (46.1) | 105 (41.3) | 0.2814 | 85 (49.4) | 28 (38.3) | 0.112 | 93 (45.5) | 12 (24) | 0.006* |
| ADL difficulty (at least one activity) | 197 (80.4) | 199 (78.4) | 0.5694 | 135 (78.5) | 62 (85) | 0.245 | 155 (76) | 44 (88) | 0.06 |
| Current smokers | 107 (43.7) | 1 (0.4) | <0.0001* | 70 (41) | 37 (51) | 0.149 | 1 (0.5) | 0 (0) | 0.62 |
| Consumes alcohol | 175 (71.4) | 146 (57.5) | 0.0011v | 123 (71) | 52 (71) | 0.964 | 118 (57.8) | 28 (56) | 0.81 |
| Hypertension | 149 (60.8) | 157 (61.8) | 0.8196 | 98 (57) | 51 (70) | 0.059 | 120 (58.8) | 37 (74) | 0.05 |
| Diabetes | 41 (16.7) | 57 (22.4) | 0.1087 | 29 (17) | 12 (16,4) | 0.935 | 42 (21) | 15 (30) | 0.15 |
| Stroke | 17 (6.9) | 9 (3.5) | 0.0880 | 10 (6) | 7 (10) | 0.285 | 4 (2) | 5 (10) | 0.01* |
There were 123 deaths ($25\%$) [73 men ($30\%$) and 50 ($20\%$) women among the 499 participants. Men and women who died were significantly older than survivors. Men who died were taller, had lower grip strength, slower gait speed, and lower SPPB scores. Women who died had lower body weight and BMI, higher 25-hydroxy vitamin D levels, slower gait speed, lower grip strength, lower SPPB score and, reported lower health status and a higher prevalence of stroke.
## pQCT parameters and all-cause mortality
Among men, there was a significant decreasing risk of all-cause mortality with increasing quartiles of trabecular vBMD of radius and tibia, cortical vBMD of radius, cortical thickness of radius and tibia, and muscle density, Fig 1. All-cause mortality significantly increased with increasing quartiles of endosteal circumference of radius and tibia and across quartiles of SSIp at both the radius or tibia. Among women, all-cause mortality risk significantly decreased with increasing quartiles of cortical vBMD of radius and tibia, cortical thickness of radius and tibia, Fig 2. The mortality risk increased with the increasing quartiles of endosteal circumference of radius and tibia. All-cause mortality did not vary across quartiles of trabecular vBMD of radius or tibia, and muscle density. The Unadjusted Log Rank test p-value of the pQCT bone and muscle density quartiles and All-cause Mortality are shown in Table 2.
**Fig 1:** *Kaplan Meir curves of quartiles of pQCT parameters and log rank test in men.* **Fig 2:** *Kaplan Meir curves of quartiles of pQCT parameters and log rank test in women.* TABLE_PLACEHOLDER:Table 2 Among men, in multivariable adjusted continuous models, one standard deviation increase in trabecular and cortical vBMD at the radius and tibia, cortical thickness at both the radius and tibia, SSPi of the tibia (but not radius)and muscle density were all associated with significant lower mortality, Table 3.
**Table 3**
| Unnamed: 0 | Model 1 | Model 1.1 | Model 2 | Model 2.1 | Model 3 | Model 3.1 | Model 4 | Model 4.1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | HR | p value | HR | p value | HR | p value | HR | p value |
| Trabecular V BMD (Radius 4%) | 0.63 (0.47, 0.8) | 0.0018 | 0.59 (0.43, 0.81) | 0.0011 | 0.60 (0.44, 0.81) | 0.0009 | 0.59 (0.42, 0.82) | 0.0016 |
| Cortical vBMD (Radius 33%) | 0.63 (0.49, 0.81) | 0.0003 | 0.61 (0.47, 0.79) | 0.0003 | 0.62 (0.47, 0.80) | 0.0003 | 0.60 (0.46, 0.80) | 0.0004 |
| Cortical Thickness (Radius 33%) | 0.56 (0.43, 0.73) | <0.0001 | 0.55 (0.42, 0.72) | <0.0001 | 0.58 (0.44, 0.77) | 0.0001 | 0.52 (0.39, 0.69) | <0.0001 |
| Endosteal circumference (Radius 33%) | 1.56 (1.21, 2.00) | 0.0005 | 1.63 (1.25, 2.12) | 0.0003 | 1.56(1.20, 2.02) | 0.001 | 1.72 (1.32, 2.25) | <0.0001 |
| SSI p (Radius 33%) | 0.85 (0.65, 1.11) | 0.23 | 0.85 (0.64, 1.12) | 0.25 | 0.88 (0.66, 1.16) | 0.36 | 0.86 (0.65, 1.15) | 0.863 |
| Trabecular V BMD (Tibia 4%) | 0.64 (0.48, 0.853) | 0.002 | 0.60 (0.45, 0.81) | 0.001 | 0.64 (0.48, 0.86) | 0.003 | 0.62 (0.45, 0.86) | 0.0036 |
| Cortical vBMD (Tibia 33%) | 0.64 (0.51, 0.80) | 0.0001 | 0.62 (0.49, 0.79) | 0.0001 | 0.60 (0.47, 0.77) | <0.0001 | 0.63 (0.49, 0.80) | 0.0002 |
| Cortical Thickness (Tibia 33%) | 0.61 (0.48, 0.77) | <0.0001 | 0.60 (0.47, 0.77) | <0.0001 | 0.65 (0.51, 0.84) | 0.001 | 0.63 (0.49, 0.81) | 0.0004 |
| Endosteal circumference (Tibia 33%) | 1.54 (1.20, 1.96) | 0.0006 | 1.54 (1.19, 1.98) | 0.0009 | 1.5 (1.15, 1.95) | 0.003 | 1.45 (1.14, 1.94) | 0.0038 |
| SSI p (Tibia 33%) | 0.74 (0.55, 0.99) | 0.04 | 0.73 (0.54, 0.98) | 0.04 | 0.77 (0.57, 1.03) | 0.085 | 0.75 (0.56, 1.01) | 0.058 |
| Muscle Densitya | 0.63 (0.49, 0.81) | 0.0003 | 0.67 (0.51, 0.87) | 0.003 | 0.73 (0.55, 0.96) | 0.025 | 0.78 (0.58, 1.05) | 0.095$ |
One standard deviation increase in endosteal circumference at both the radius was associated with a 53–$64\%$ increase in mortality risk.
Among women, in multivariable adjusted continuous models, one standard deviation increase in cortical vBMD (but not trabecular vBMD) at both the radius and tibia, cortical thickness and SSPi were all associated with lower mortality, Table 4. Similar to findings in men, greater endosteal circumference was associated with an increased mortality. There was no association with muscle density and mortality in women.
**Table 4**
| Unnamed: 0 | Model 1 | Model 1.1 | Model 2 | Model 2.1 | Model 3 | Model 3.1 | Model 4 | Model 4.1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | HR | P value | HR | P value | HR | P value | HR | P value |
| Trabecular V BMD (Radius 4%) | 0.97(0.70,1.34) | 0.84 | 0.86(0.61,1.23) | 0.40 | 1.07(0.72,1.58) | 0.73 | 0.97(0.68,1.37) | 0.86 |
| Cortical vBMD (Radius 33%) | 0.58(0.42,0.7) | 0.0006 | 0.64(0.47,0.87) | 0.003 | 0.68(0.49,0.95) | 0.03 | 0.73(0.55,1.0) | 0.05 |
| Cortical Thickness (Radius 33%) | 0.49(0.32,0.73) | 0.0004 | 0.54(0.37,0.79) | 0.002 | 0.50(0.31,0.81) | 0.005 | 0.60(0.40,0.88) | 0.009 |
| Endosteal circumference (Radius 33%) | 1.40(1.01,1.93) | 0.04 | 1.33(0.97,1.82) | 0.07 | 1.44(0.99,2.09) | 0.056 | 1.29(0.91,1.82) | 0.15 |
| SSI p (Radius 33%) | 0.71(0.50,1.02) | 0.06 | 0.70(0.48,1.01) | 0.05 | 0.75(0.49,1.14) | 0.18 | 0.68(0.47,0.97) | 0.04 |
| Trabecular V BMD (Tibia 4%) | 0.95(0.69,1.30) | 0.74 | 0.91(0.66,1.28) | 0.60 | 1.12(0.77,1.63) | 0.55 | 1.00(0.73,1.39) | 0.96 |
| Cortical vBMD (Tibia 33%) | 0.64(0.48,0.87) | 0.004 | 0.60(0.45,0.79) | 0.0004 | 0.67(0.50,0.91) | 0.011 | 0.66(0.50,0.88) | 0.004 |
| Cortical Thickness (Tibia 33%) | 0.43(0.30,0.61) | <0.0001 | 0.43(0.30,0.61) | <0.0001 | 0.47(0.33,0.68) | <0.0001 | 0.50(0.35,0.71) | 0.0001 |
| Endosteal circumference (Tibia 33%) | 1.79(1.34,2.39) | <0.0001 | 1.83(1.37,2.45) | <0.0001 | 1.83(1.34,2.48) | 0.0001 | 1.77(1.32,2.38) | 0.0001 |
| SSI p (Tibia 33%) | 0.60(0.38,0.94) | 0.02 | 0.581(0.374,0.90) | 0.02 | 0.78(0.49,1.24) | 0.30 | 0.68 (0.44,1.03) | 0.07 |
| Muscle Density | 0.81(0.60,1.10) | 0.18 | 0.84(0.61,1.15) | 0.27 | 0.78(0.56,1.09) | 0.15 | 0.94(0.67,1.34) | 0.74 |
## Mediation analysis
The pQCT parameters which were significant in model 2 were compared with the mediator models (models 3 and 4) for identifying attenuation of hazards ratio of more than $10\%$. Among men, muscle density showed greater than $10\%$ attenuation with the gait speed. Based on the mediation analysis (not shown) gait speed mediates the association between muscle density and all-cause mortality significantly to the extent of $10\%$ ($95\%$ CI: $0.2\%$, $19.7\%$).
Among women, attenuation of more than $10\%$ in the grip strength model was observed for cortical vBMD at radius and tibia, cortical thickness at tibia, and SSIp at radius and tibia. In the gait speed model, attenuation of more than $10\%$ was observed for cortical vBMD at radius and tibia, cortical thickness at radius and tibia and SSIP at tibia. However, the mediation analysis of these pQCT parameters observed no significant mediation by gait speed or grip strength on all-cause mortality among women.
## Discussion
We showed that trabecular vBMD (radius and tibia), cortical vBMD (radius and tibia), cortical thickness (radius and tibia), endosteal circumference (radius and tibia), SSIp (tibia) and muscle density were independent predictors of all-cause mortality among rural south Indian older men. Cortical vBMD (radius and tibia), cortical thickness (radius and tibia), SSIp (radius and tibia) and endosteal circumference (radius and tibia) were independent predictors of all-cause mortality among women. Gait speed significantly mediated the association of mortality and muscle density among men. However, among women no significant mediation by gait speed or grip strength was observed with any of the pQCT parameters.
To our knowledge there have been no studies of pQCT derived bone measures and mortality. In the Age, Gene/Environment Susceptibility (AGES)—Reykjavik Study among men and women, QCT trabecular vBMD at proximal femur was inversely associated with all-cause mortality independent of covariates including gender but cortical vBMD was not associated with mortality [27]. In the African American-Diabetes Heart Study (AA-DHS), QCT was measured at chest and abdomen for thoracic and lumbar spine vBMD. Among men, thoracic and lumbar spine vBMD was inversely associated with all-cause mortality at lumbar vBMD [HR per SD increase = 0.70 ($95\%$ CI 0.52–0.95, $$p \leq 0.02$$)] and thoracic vBMD [HR per SD increase = 0.71 ($95\%$ CI 0.54–0.92, $$p \leq 0.01$$)], but no association was observed among women [28]. These findings were similar to our study wherein trabecular vBMD was associated with mortality among men but not among women.
It has been well established that fractures are associated with increased mortality [29]. Results from the (AGES)—Reykjavik Study showed that history of fracture before the bone assessment did not alter the vBMD and mortality association, whereas incident fracture after the bone assessment attenuated the mortality association with trabecular vBMD [27]. This suggests that fractures may explain the association between lower BMD and mortality. We did not have information on incident fractures but adjusting for history of fracture in the past five years had little effect on the association of trabecular vBMD and mortality among men and women; however, adjustment for fracture history attenuated the SSIp of radius association with mortality among women. Further studies are needed to understand the association and the role of incident fractures among older Indian population and mortality.
Previously, we showed that trabecular vBMD among Indian men was 1.3–1.5 SD units lower compared to Caucasian US men; cortical thickness was 0.8 to 1.2 SD units lower and endosteal circumferences 0.5–0.8 SD units higher among the Indian men. This may suggest that the Indian older population have lower bone density and strength measures that could influence earlier mortality either through fractures or through other mechanisms.
Trabecular bone loss is more pronounced at earlier ages before 50 years (women $37\%$, men $42\%$) compared to cortical bone loss [30]. Throughout adulthood, periosteal apposition counterbalances endosteal bone loss by reconfiguring the available bone mass to maintain biomechanical properties. However, with increasing age, bone loss shifts more to the cortical compartment leading to cortical thinning and increased cortical porosity that in turn leads to loss of biomechanical strength and increased risk for fracture [31]. Bone loss in aging is the net result of periosteal bone formation and endosteal bone resorption [32], however with increasing age, the bone resorption exceeds bone formation leading to bone loss in the endosteal region and an increase in the endosteal circumference. The lower levels of cortical vBMD, lower cortical thickness and higher endosteal circumference in our study among men and women, suggest increased cortical thinning and porosity that may impact mortality among the Indian older population. The lack of association of trabecular vBMD with mortality at least among women needs to be explored further.
The association between lower BMD and mortality may reflect common pathways. An association between cardiovascular disease (CVD) and BMD has been reported but the results of these studies have been inconclusive. A large analysis of NHANES III observed no association of low BMD and CVD mortality among men. However, among women soon after menopause, low BMD was associated with mortality from cardiovascular disease [33]. BMD was associated with mortality independent of coronary artery calcium score and chronic lung disease [27]. In a meta-analysis of 28 studies, low BMD was associated with an increased risk of developing coronary artery disease, cerebrovascular conditions, and CVD-associated death [34]. This may suggest that the CVD may share similar pathways in the link between BMD and mortality.
The association between low BMD and increased mortality may also reflect shared risk factors. Increased low grade inflammation has been linked to higher mortality [35, 36] and lower BMD [37, 38] and fractures [39, 40]. Endogenous sex hormones are associated with mortality [41, 42], lower BMD [43, 44] and fractures [45, 46]. Age at menopause is negatively associated with BMD [47] increased fractures [48, 49] and mortality [48] and Indian women have an earlier menopause [50, 51]. Nevertheless, we have no information on these factors and their influence on these bone-mortality relationships cannot be ruled out.
The mechanostat hypothesis and the concept of bone and muscle crosstalk suggest there may be association of bone and physical performance measures. pQCT bone measures have been associated with grip strength [52–60] and gait speed [57, 59, 60]. Grip strength [18] and gait speed [19] also predict mortality. Considering this, grip strength and gait speed may mediate the association between the pQCT bone measures and mortality. However, in our analysis we observed no mediation between the pQCT bone parameters and mortality.
Over a six year follow up of the inCHIANTI study in Italy, pQCT derived muscle density was associated with all-cause mortality (per SD increase, 0.78 [0.69–0.88]) in models adjusted for height and weight. This association was attenuated and was not significant in fully adjusted models [61]. These results are similar to our findings among women. Among older men in the MrOS with a mean follow up of 7.2 years, muscle density (pQCT derived) was significantly inversely associated with all-cause mortality independent of important covariates [62]. Among Afro-Caribbean men aged 40 years in the Tobago health study, muscle density was also significantly inversely associated with all-cause mortality when adjusted for age and other covariates [17]. In the AGES study, both higher thigh intramuscular and intermuscular fat was associated with an increased mortality in men; intramuscular fat but not intermuscular fat was associated with mortality in women. Overall, findings may differ in men than women perhaps reflecting higher testosterone levels. *In* general, these associations were much smaller in women compared to me [63]. Opportunistic analyses of abdominal pelvic CT scans revealed associations with a total muscle index and psoas muscle index and mortality in a small convenient sample of men and women [64]. The latter study did not stratify by sex. Our study also observed similar inverse relationship of muscle density and mortality among men but not among women. In our study, muscle density did not differ between men and women suggesting a potential gender influence on the adverse effects of fat infiltration into muscle.
Muscle density is a proxy measure of myosteatosis. Insulin resistance [65] and oxidative stress [66] are considered to be factors that influence myosteatosis and mortality. In our study, among men the association of muscle density and mortality persisted even after adjusting for diabetes status. This was similarly observed in other studies [17, 62] suggesting that there may be other pathways which influence the muscle density and mortality association. Gait speed is associated with muscle density [61, 67, 68] and is associated with mortality [19]. The mediation of gait speed between the muscle density and mortality was significant among men in our study, suggesting that gait speed may be in the pathway of myosteatosis and mortality.
The current analysis has several potential limitations. This was an observational study and thus causality cannot be determined. The study sample was small and was limited to a rural south Indian region but the magnitude of our associations were similar to previous reports from larger cohorts of European descent. As some of the key indicators of shared pathways were not measured, there could be residual confounding in these relationships. We were unable to look at cause specific deaths and future studies are needed to determine whether observed associations are driven by mortality from specific causes. As this was an exploratory and hypothesis generating study, we did not adjust for multiple comparisons. Finally, newer techniques to measure microarchitecture and cortical porosity, such as, high resolution pQCT are currently available but we had no access to this technology.
However, our study has some important strengths. To our knowledge this was the first study describing pQCT measures and mortality association among a unique population based random sample of older Indian men and women with excellent longitudinal follow. We adjusted for many important potential confounding variables and conducted mediation analysis.
In conclusion, this study presents unreported independent association of all-cause mortality with trabecular vBMD, cortical vBMD, cortical thickness, endosteal circumference, SSIp and muscle density among rural south Indian older men; and cortical vBMD, cortical thickness, endosteal circumference and SSIp among rural south Indian older women. Grip strength and gait speed did not mediate the association of bone and muscle among women; however, significant mediation was observed by gait speed on muscle density and mortality among men. Further research is needed to confirm our findings in larger Indian older populations and to study the role of mediation of some of the key factors that may underlie these associations.
## References
1. Trivedi DP, Khaw KT. **Bone Mineral Density at the Hip Predicts Mortality in Elderly Men**. *Osteoporosis International* (2001) **12** 259-65. DOI: 10.1007/s001980170114
2. Mussolino ME, Gillum RF. **Low Bone Mineral Density and Mortality in Men and Women: The Third National Health and Nutrition Examination Survey Linked Mortality File**. *Annals of epidemiology* (2008) **18** 847-50. DOI: 10.1016/j.annepidem.2008.07.003
3. Van der Klift M, Pols HAP, Geleijnse JM, Van der Kuip DAM, Hofman A. **De Laet CEDH. Bone mineral density and mortality in elderly men and women: The Rotterdam study**. *Bone* (2002) **30** 643-8. DOI: 10.1016/S8756-3282(02)00670-1
4. Suzuki T, Yoshida H. **Low bone mineral density at femoral neck is a predictor of increased mortality in elderly Japanese women**. *Osteoporosis International* (2010) **21** 71-9. DOI: 10.1007/s00198-009-0970-6
5. Nguyen ND, Center JR, Eisman JA, Nguyen TV. **Bone Loss, Weight Loss, and Weight Fluctuation Predict Mortality Risk in Elderly Men and Women**. *Journal of Bone and Mineral Research* (2007) **22** 1147-54. DOI: 10.1359/jbmr.070412
6. Johansson H, Odén A, Kanis J, McCloskey E, Lorentzon M, Ljunggren Ö. **Low bone mineral density is associated with increased mortality in elderly men**. *MrOS Sweden. Osteoporosis International* (2011) **22** 1411-8. DOI: 10.1007/s00198-010-1331-1
7. Qu X, Huang X, Jin F, Wang H, Hao Y, Tang T. **Bone mineral density and all-cause, cardiovascular and stroke mortality: a meta-analysis of prospective cohort studies**. *Int J Cardiol* (2013) **166** 385-93. DOI: 10.1016/j.ijcard.2011.10.114
8. Kuk JL, Saunders TJ, Davidson LE, Ross R. **Age-related changes in total and regional fat distribution**. *Ageing Res Rev* (2009) **8** 339-48. DOI: 10.1016/j.arr.2009.06.001
9. Miljkovic I, Zmuda JM. **Epidemiology of myosteatosis**. *Current opinion in clinical nutrition and metabolic care* (2010) **13** 260-4. DOI: 10.1097/MCO.0b013e328337d826
10. Visser M, Kritchevsky SB, Goodpaster BH, Newman AB, Nevitt M, Stamm E. **Leg muscle mass and composition in relation to lower extremity performance in men and women aged 70 to 79: the health, aging and body composition study**. *Journal of the American Geriatrics Society* (2002) **50** 897-904. DOI: 10.1046/j.1532-5415.2002.50217.x
11. Marcus RL, Addison O, Dibble LE, Foreman KB, Morrell G, Lastayo P. **Intramuscular adipose tissue, sarcopenia, and mobility function in older individuals**. *Journal of aging research* (2012) **2012** 629637. DOI: 10.1155/2012/629637
12. Delmonico MJ, Harris TB, Visser M, Park SW, Conroy MB, Velasquez-Mieyer P. **Longitudinal study of muscle strength, quality, and adipose tissue infiltration**. *Am J Clin Nutr* (2009) **90** 1579-85. DOI: 10.3945/ajcn.2009.28047
13. Cree MG, Newcomer BR, Katsanos CS, Sheffield-Moore M, Chinkes D, Aarsland A. **Intramuscular and liver triglycerides are increased in the elderly**. *The Journal of clinical endocrinology and metabolism* (2004) **89** 3864-71. DOI: 10.1210/jc.2003-031986
14. Goodpaster BH, Carlson CL, Visser M, Kelley DE, Scherzinger A, Harris TB. **Attenuation of skeletal muscle and strength in the elderly: The Health ABC Study**. *J Appl Physiol (1985)* (2001) **90** 2157-65. DOI: 10.1152/jappl.2001.90.6.2157
15. Lang T, Koyama A, Li C, Li J, Lu Y, Saeed I. **Pelvic body composition measurements by quantitative computed tomography: association with recent hip fracture**. *Bone* (2008) **42** 798-805. DOI: 10.1016/j.bone.2007.12.002
16. Lang T, Cauley JA, Tylavsky F, Bauer D, Cummings S, Harris TB. **Computed tomographic measurements of thigh muscle cross-sectional area and attenuation coefficient predict hip fracture: the health, aging, and body composition study**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2010) **25** 513-9. DOI: 10.1359/jbmr.090807
17. Zhao Q, Zmuda JM, Kuipers AL, Jonnalagadda P, Bunker CH, Patrick AL. **Greater skeletal muscle fat infiltration is associated with higher all-cause mortality among men of African ancestry**. *Age Ageing* (2016) **45** 529-34. DOI: 10.1093/ageing/afw062
18. Wu Y, Wang W, Liu T, Zhang D. **Association of Grip Strength With Risk of All-Cause Mortality, Cardiovascular Diseases, and Cancer in Community-Dwelling Populations: A Meta-analysis of Prospective Cohort Studies**. *J Am Med Dir Assoc* (2017) **18** 551.e17-.e35. DOI: 10.1016/j.jamda.2017.03.011
19. Studenski S, Perera S, Patel K, Rosano C, Faulkner K, Inzitari M. **Gait speed and survival in older adults**. *JAMA* (2011) **305** 50-8. DOI: 10.1001/jama.2010.1923
20. Singh T, Sharma PK, Jammy GR, Cauley JA, Bunker CH, Reddy PS. **Design of the Mobility and Independent Living in Elders Study: An older adult cohort in rural India**. *Geriatr Gerontol Int* (2017) **17** 31-40. DOI: 10.1111/ggi.12668
21. Muller ME, Webber CE, Bouxsein ML. **Predicting the failure load of the distal radius**. *Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA* (2003) **14** 345-52. DOI: 10.1007/s00198-003-1380-9
22. Ferretti JL, Capozza RF, Zanchetta JR. **Mechanical validation of a tomographic (pQCT) index for noninvasive estimation of rat femur bending strength**. *Bone* (1996) **18** 97-102. DOI: 10.1016/8756-3282(95)00438-6
23. Bouxsein ML, Karasik D. **Bone geometry and skeletal fragility**. *Curr Osteoporos Rep* (2006) **4** 49-56. DOI: 10.1007/s11914-006-0002-9
24. Seeman E.. **Bone quality: the material and structural basis of bone strength**. *J Bone Miner Metab* (2008) **26** 1-8. DOI: 10.1007/s00774-007-0793-5
25. Goodpaster BH, Kelley DE, Thaete FL, He J, Ross R. **Skeletal muscle attenuation determined by computed tomography is associated with skeletal muscle lipid content**. *J Appl Physiol (1985)* (2000) **89** 104-10. DOI: 10.1152/jappl.2000.89.1.104
26. Guralnik JM, Ferrucci L, Pieper CF, Leveille SG, Markides KS, Ostir GV. **Lower extremity function and subsequent disability: consistency across studies, predictive models, and value of gait speed alone compared with the short physical performance battery**. *The journals of gerontology Series A, Biological sciences and medical sciences* (2000) **55** M221-31. DOI: 10.1093/gerona/55.4.m221
27. Marques EA, Elbejjani M, Gudnason V, Sigurdsson G, Lang T, Sigurdsson S. **Proximal Femur Volumetric Bone Mineral Density and Mortality: 13 Years of Follow-Up of the AGES-Reykjavik Study**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2017) **32** 1237-42. DOI: 10.1002/jbmr.3104
28. Lenchik L, Register TC, Russell GB, Xu J, Smith SC, Bowden DW. **Volumetric bone mineral density of the spine predicts mortality in African-American men with type 2 diabetes**. *Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA* (2018) **29** 2049-57. DOI: 10.1007/s00198-018-4578-6
29. Haentjens P, Magaziner J, Colon-Emeric CS, Vanderschueren D, Milisen K, Velkeniers B. **Meta-analysis: excess mortality after hip fracture among older women and men**. *Ann Intern Med* (2010) **152** 380-90. DOI: 10.7326/0003-4819-152-6-201003160-00008
30. Riggs BL, Melton LJ, Robb RA, Camp JJ, Atkinson EJ, McDaniel L. **A population-based assessment of rates of bone loss at multiple skeletal sites: evidence for substantial trabecular bone loss in young adult women and men**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2008) **23** 205-14. DOI: 10.1359/jbmr.071020
31. Seeman E.. **Structural basis of growth-related gain and age-related loss of bone strength**. *Rheumatology (Oxford, England)* (2008) **47** iv2-8. DOI: 10.1093/rheumatology/ken177
32. Seeman E.. **Pathogenesis of bone fragility in women and men**. *Lancet (London, England)* (2002) **359** 1841-50. DOI: 10.1016/S0140-6736(02)08706-8
33. Mussolino ME, Armenian HK. **Low bone mineral density, coronary heart disease, and stroke mortality in men and women: the Third National Health and Nutrition Examination Survey**. *Annals of epidemiology* (2007) **17** 841-6. DOI: 10.1016/j.annepidem.2007.06.005
34. Veronese N, Stubbs B, Crepaldi G, Solmi M, Cooper C, Harvey NC. **Relationship Between Low Bone Mineral Density and Fractures With Incident Cardiovascular Disease: A Systematic Review and Meta-Analysis**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2017) **32** 1126-35. DOI: 10.1002/jbmr.3089
35. De Martinis M, Franceschi C, Monti D, Ginaldi L. **Inflammation markers predicting frailty and mortality in the elderly**. *Experimental and molecular pathology* (2006) **80** 219-27. DOI: 10.1016/j.yexmp.2005.11.004
36. Giovannini S, Onder G, Liperoti R, Russo A, Carter C, Capoluongo E. **Interleukin-6, C-reactive protein, and tumor necrosis factor-alpha as predictors of mortality in frail, community-living elderly individuals**. *Journal of the American Geriatrics Society* (2011) **59** 1679-85. DOI: 10.1111/j.1532-5415.2011.03570.x
37. de Pablo P, Cooper MS, Buckley CD. **Association between bone mineral density and C-reactive protein in a large population-based sample**. *Arthritis and rheumatism* (2012) **64** 2624-31. DOI: 10.1002/art.34474
38. Koh JM, Khang YH, Jung CH, Bae S, Kim DJ, Chung YE. **Higher circulating hsCRP levels are associated with lower bone mineral density in healthy pre- and postmenopausal women: evidence for a link between systemic inflammation and osteoporosis**. *Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA* (2005) **16** 1263-71. DOI: 10.1007/s00198-005-1840-5
39. Cauley JA, Danielson ME, Boudreau RM, Forrest KY, Zmuda JM, Pahor M. **Inflammatory markers and incident fracture risk in older men and women: the Health Aging and Body Composition Study**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2007) **22** 1088-95. DOI: 10.1359/jbmr.070409
40. Ahmadi-Abhari S, Luben RN, Wareham NJ, Khaw KT. **C-reactive protein and fracture risk: European prospective investigation into Cancer Norfolk Study**. *Bone* (2013) **56** 67-72. DOI: 10.1016/j.bone.2013.05.009
41. Benn M, Voss Sidsel S, Holmegard Haya N, Jensen Gorm B, Tybjærg-Hansen A, Nordestgaard Børge G. **Extreme Concentrations of Endogenous Sex Hormones, Ischemic Heart Disease, and Death in Women**. *Arteriosclerosis, Thrombosis, and Vascular Biology* (2015) **35** 471-7. DOI: 10.1161/ATVBAHA.114.304821
42. Araujo AB, Kupelian V, Page ST, Handelsman DJ, Bremner WJ, McKinlay JB. **Sex steroids and all-cause and cause-specific mortality in men**. *Archives of internal medicine* (2007) **167** 1252-60. DOI: 10.1001/archinte.167.12.1252
43. Greendale GA, Edelstein S, Barrett-Connor E. **Endogenous Sex Steroids and Bone Mineral Density in Older Women and Men: The Rancho Bernardo Study**. *Journal of Bone and Mineral Research* (1997) **12** 1833-43. DOI: 10.1359/jbmr.1997.12.11.1833
44. Weber TJ. **Battle of the sex steroids in the male skeleton: and the winner is…**. *The Journal of Clinical Investigation* (2016) **126** 829-32. DOI: 10.1172/JCI85006
45. Meier C, Nguyen TV, Handelsman DJ, Schindler C, Kushnir MM, Rockwood AL. **Endogenous sex hormones and incident fracture risk in older men: the Dubbo Osteoporosis Epidemiology Study**. *Archives of internal medicine* (2008) **168** 47-54. DOI: 10.1001/archinternmed.2007.2
46. Cauley JA, Danielson ME, Jammy GR, Bauer DC, Jackson R, Wactawski-Wende J. **Sex Steroid Hormones and Fracture in a Multiethnic Cohort of Women: The Women’s Health Initiative Study (WHI)**. *The Journal of clinical endocrinology and metabolism* (2017) **102** 1538-47. DOI: 10.1210/jc.2016-3589
47. Hedlund LR, Gallagher JC. **The effect of age and menopause on bone mineral density of the proximal femur**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (1989) **4** 639-42. DOI: 10.1002/jbmr.5650040423
48. Svejme O, Ahlborg H, Nilsson J-Å, Karlsson M. **Early menopause and risk of osteoporosis, fracture and mortality: a 34-year prospective observational study in 390 women**. *BJOG: An International Journal of Obstetrics & Gynaecology* (2012) **119** 810-6. DOI: 10.1111/j.1471-0528.2012.03324.x
49. Sullivan SD, Lehman A, Nathan NK, Thomson CA, Howard BV. **Age of menopause and fracture risk in postmenopausal women randomized to calcium + vitamin D, hormone therapy, or the combination: results from the Women’s Health Initiative Clinical Trials**. *Menopause* (2017) **24** 371-8. DOI: 10.1097/GME.0000000000000775
50. Ahuja M.. **Age of menopause and determinants of menopause age: A PAN India survey by IMS**. *Journal of mid-life health* (2016) **7** 126-31. DOI: 10.4103/0976-7800.191012
51. Jungari SB, Chauhan BG. **Prevalence and Determinants of Premature Menopause among Indian Women: Issues and Challenges Ahead**. *Health & social work* (2017) **42** 79-86. DOI: 10.1093/hsw/hlx010
52. Boonen S, Cheng XG, Nijs J, Nicholson PH, Verbeke G, Lesaffre E. **Factors associated with cortical and trabecular bone loss as quantified by peripheral computed tomography (pQCT) at the ultradistal radius in aging women**. *Calcif Tissue Int* (1997) **60** 164-70. DOI: 10.1007/s002239900208
53. Hasegawa Y, Schneider P, Reiners C. **Age, sex, and grip strength determine architectural bone parameters assessed by peripheral quantitative computed tomography (pQCT) at the human radius**. *J Biomech* (2001) **34** 497-503. DOI: 10.1016/s0021-9290(00)00211-6
54. Specker B, Binkley T, Fahrenwald N. **Rural versus nonrural differences in BMC, volumetric BMD, and bone size: a population-based cross-sectional study**. *Bone* (2004) **35** 1389-98. DOI: 10.1016/j.bone.2004.09.005
55. Kaji H, Kosaka R, Yamauchi M, Kuno K, Chihara K, Sugimoto T. **Effects of age, grip strength and smoking on forearm volumetric bone mineral density and bone geometry by peripheral quantitative computed tomography: comparisons between female and male**. *Endocr J* (2005) **52** 659-66. DOI: 10.1507/endocrj.52.659
56. Cousins JM, Petit MA, Paudel ML, Taylor BC, Hughes JM, Cauley JA. **Muscle power and physical activity are associated with bone strength in older men: The osteoporotic fractures in men study**. *Bone* (2010) **47** 205-11. DOI: 10.1016/j.bone.2010.05.003
57. Barbour KE, Zmuda JM, Strotmeyer ES, Horwitz MJ, Boudreau R, Evans RW. **Correlates of trabecular and cortical volumetric bone mineral density of the radius and tibia in older men: the Osteoporotic Fractures in Men Study**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2010) **25** 1017-28. DOI: 10.1002/jbmr.6
58. Chalhoub D, Boudreau R, Greenspan S, Newman AB, Zmuda J, Frank-Wilson AW. **Associations Between Lean Mass, Muscle Strength and Power, and Skeletal Size, Density and Strength in Older Men**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2018) **33** 1612-21. DOI: 10.1002/jbmr.3458
59. Edwards MH, Gregson CL, Patel HP, Jameson KA, Harvey NC, Sayer AA. **Muscle size, strength, and physical performance and their associations with bone structure in the Hertfordshire Cohort Study**. *Journal of bone and mineral research: the official journal of the American Society for Bone and Mineral Research* (2013) **28** 2295-304. DOI: 10.1002/jbmr.1972
60. Patel A, Edwards MH, Jameson KA, Ward KA, Fuggle N, Cooper C. **Longitudinal Change in Peripheral Quantitative Computed Tomography Assessment in Older Adults: The Hertfordshire Cohort Study**. *Calcif Tissue Int* (2018) **103** 476-82. DOI: 10.1007/s00223-018-0442-0
61. Cesari M, Pahor M, Lauretani F, Zamboni V, Bandinelli S, Bernabei R. **Skeletal muscle and mortality results from the InCHIANTI Study**. *The journals of gerontology Series A* (2009) **64** 377-84. DOI: 10.1093/gerona/gln031
62. Miljkovic I, Kuipers AL, Cauley JA, Prasad T, Lee CG, Ensrud KE. **Greater Skeletal Muscle Fat Infiltration Is Associated With Higher All-Cause and Cardiovascular Mortality in Older Men**. *The journals of gerontology Series A* (2015) **70** 1133-40. DOI: 10.1093/gerona/glv027
63. Reinders I, Murphy RA, Brouwer IA, Visser M, Launer L, Siggeirsdottir K. **Muscle Quality and Myosteatosis: Novel Associations With Mortality Risk: The Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study**. *Am J Epidemiol* (2016) **183** 53-60. DOI: 10.1093/aje/kwv153
64. Lenchik L, Lenoir KM, Tan J, Boutin RD, Callahan KE, Kritchevsky SB. **Opportunistic Measurement of Skeletal Muscle Size and Muscle Attenuation on Computed Tomography Predicts 1-Year Mortality in Medicare Patients**. *J Gerontol A Biol Sci Med Sci* (2019) **74** 1063-9. DOI: 10.1093/gerona/gly183
65. Miljkovic I, Vella CA, Allison M. **Computed Tomography-Derived Myosteatosis and Metabolic Disorders**. *Diabetes Metab J* (2021) **45** 482-91. DOI: 10.4093/dmj.2020.0277
66. Garcia-Sanchez A, Gamez-Nava JI, Diaz-de la Cruz EN, Cardona-Munoz EG, Becerra-Alvarado IN, Aceves-Aceves JA. **The Effect of Visceral Abdominal Fat Volume on Oxidative Stress and Proinflammatory Cytokines in Subjects with Normal Weight, Overweight and Obesity**. *Diabetes Metab Syndr Obes* (2020) **13** 1077-87. DOI: 10.2147/DMSO.S245494
67. Cawthon PM, Fox KM, Gandra SR, Delmonico MJ, Chiou CF, Anthony MS. **Clustering of strength, physical function, muscle, and adiposity characteristics and risk of disability in older adults**. *Journal of the American Geriatrics Society* (2011) **59** 781-7. DOI: 10.1111/j.1532-5415.2011.03389.x
68. Cawthon PM, Fox KM, Gandra SR, Delmonico MJ, Chiou CF, Anthony MS. **Do muscle mass, muscle density, strength, and physical function similarly influence risk of hospitalization in older adults?**. *Journal of the American Geriatrics Society* (2009) **57** 1411-9. DOI: 10.1111/j.1532-5415.2009.02366.x
|
---
title: Non-communicable diseases risk factors among the forcefully displaced Rohingya
population in Bangladesh
authors:
- Ayesha Rahman
- Jheelam Biswas
- Palash Chandra Banik
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022334
doi: 10.1371/journal.pgph.0000930
license: CC BY 4.0
---
# Non-communicable diseases risk factors among the forcefully displaced Rohingya population in Bangladesh
## Abstract
Rohingya refugees of Ukhiya, Cox’s bazar are an unaccounted group of people who form the largest cluster of refugees worldwide. Non-communicable disease (NCD) alone causes $70\%$ of worldwide deaths every year therefore, the trend of NCD among Rohingya refugees demands proper evaluation and attention. The objective of this study was to measure the NCD risk factors among a convenient sample of Rohingya refugees. This cross-sectional study was conducted among 249 Rohingya refugees living in Balukhali and Kutupalang Rohinga Camps at Ukhiya, Cox’s bazaar using a survey dataset adapted from the WHO Stepwise approach to NCD Risk Factor Surveillance (STEPS). Data was collected through face-to-face interviews with a structured questionnaire. Anthropometric and biochemical measurements were done by trained medical assistants. Descriptive analysis was applied as appropriate for categorical variables. A Chi-square test and a student t test were performed to compare the categories. *In* general, the findings of NCD risk factors as per STEPS survey was $53.4\%$ for tobacco use including smokeless tobacco, $2.8\%$ for alcohol consumption, $23.7\%$ for inadequate vegetable and fruit intake, $34.5\%$ for taking extra salt, $89.6\%$ for insufficient physical activity, $44.5\%$for confirmed hypertension, $16.9\%$ for overweight, $1.2\%$ for obesity and $0.8\%$ for high blood sugar. Some modifiable non-communicable disease risk factors such as physical inactivity, tobacco smoking, extra salt with food, and hypertension are present among the Rohinga refugees in Bangladesh. These findings were timely and essential to support the formulation and implementation of NCD-related policies among the Rohingya refugees as a priority sub-population.
## Introduction
A vulnerable and socially disadvantaged group of people are considered to be more susceptible to non-communicable disease (NCD) risk factors [1, 2]. Over 15 million of all deaths worldwide are attributed to non-communicable diseases (NCD), which occur between 30 and 69 years of age, and almost a quarter of these untimely deaths are estimated to disproportionally occur in low and middle-income countries [3]. Tobacco use, physical inactivity, unhealthy diet and harmful use of alcohol are considered to be the modifiable risk factors of NCDs that can be prevented by applying prior interventions [3, 4]. In recent times, global health policymakers are more concerned about the importance of the timely prevention, detection and correction of these modifiable risk factors to reduce the overall NCD mortality [5]. The World Health Organization has prioritized adequate monitoring and surveillance of the modifiable risk factors to overcome the NCD epidemics in low resource settings [6]. Refugees from different parts of the world are generally inflicted with poverty and social inequity [5]. Associations between poverty and social inequality with a high risk of morbidity and mortality from NCDs have been established in different studies [6–8]. One study shows that changes in nutrition and lifestyle behaviors contribute to type 2 diabetes mellitus among the migrant population [9]. Another study conducted on refugees from Iraq, Somalia, and Bhutan in the USA, found that limitation in receiving education is partly responsible for the higher prevalence of risk factors of diabetes among them than in the general population [10]. On the other hand, Burmese refugees in Australia have some level of awareness about the negative effects of smoking tobacco and chewing betel quid but little knowledge about the cessation of these habits [11]. Hypertension is widely prevalent among refugees and asylum seekers in Uganda, with a significant number of people unaware of their condition and consequently suffering from uncontrolled hypertension [12].
The Rohingya refugees are the world’s largest group of stateless people seeking asylum in Bangladesh [13]. After August 2017, 6,93,000 adult refugees were forcefully migrated into Cox’s bazar and was resettled forming new camps [14]. As of October 2019, an estimated 905,754 Rohingya refugees resided in Ukhiya, Cox’s bazar in 34 camps [15]. A survey conducted by BRAC identified that food insecurity, inadequate access to health care are two major crises among the Rohingya refugees [16]. A US-based survey on re-settled Rohingya refugees from Myanmar shows a higher trend of chronic diseases like diabetes, hypertension and obesity along with widely prevalent risk factors of NCDs both in urban and camp settings [17]. Although there is a paucity of documentation of the prevalence of NCDs among the Rohingya refugees re-settled in Bangladesh, reviewing their unhealthy dietary pattern, physical inactivity resulting from a shift from rural to sedentary life, mental stress along with depression as a consequence of violently forced migration and some frequently observed NCD risk factors like smoking, using smokeless tobacco indoor air pollution, presence of established chronic diseases among them are highly presumable [18].
However, the Bangladesh government’s health services and various NGOs currently working with the Rohingya refugees focus mainly on infectious diseases and mass vaccination among them. Therefore, the identification and prevention of infectious diseases have received the highest concern from the public health specialists to the government sectors as prime healthcare strategies, resulting in an insubstantial number of studies conducted on chronic diseases [15]. In addition, the treatment approach for NCDs is costly, multiphasic, and time-consuming, and it may pose a significant burden on the economy of a developing country like Bangladesh [15].
The present study aims to explore the non-communicable disease (NCD) risk factors in a convenience sample of the Rohinga refugees. Our study aims to act as baseline data for the larger studies so that the policymakers can put more emphasis on this area.
## Study design and setting
This cross-sectional study was conducted among the Rohinga refugees resettled in two camps Balukhali and Kutupalang, at Ukhiya, Cox’s bazaar using a convenient sampling technique. Data collection was conducted from January 2019 to June 2019 through face-to-face interviews using a structured questionnaire.
## Sample size and criteria
The Rohingya refugees aged over 18 years residing in the above mentioned camps and voluntarily agreed to participate in the study were selected as study participants. Pregnant women, critically ill patients, and people with physical or mental disabilities were excluded from the study. A study conducted among the slum dwellers in Dhaka shows that at least one NCD risk factors was present in $19.5\%$ of participants [19]. Overall prevalence of NCD risk factors were: $36.0\%$ ($95\%$ CI: 31.82–40.41). According to that study $$p \leq 19.5$$%, q = (100–19.5) = $80.5\%$ at $95\%$ CI, $z = 1.96$ & $d = 5$%. So, sample size n = pqz2/d2 = (19.5x80.5)x(1.96)$\frac{2}{52}$ = 241.21. Using this prevalence value as reference our calculated sample size was 241. Due to the deliberate participation of an adequate number of adult refugees in the study, our final sample size was 249.
## Data collection procedure
An adapted (mostly for socio-demographic background) questionnaire for this study was developed using step-I, step-II and, steps III of WHO STEPS protocol based on the 2010 STEPS survey in Bangladesh [20]. Bangla version of the STEPS questionnaire was orally translated into the local Rohingya language by an interpreter during data collection. Although the Rohingya refugees originally live in Mayanmar, the language they speak is similar to Cox’s bazar’s local dialect which assisted the local medical assistants to conduct the interviews. The questionnaire was pre-tested in the field before actual survey. It was administered by the interviewers and no proxy interview was taken. Three medical assistants were recruited from Balukhali and Kutupalong health camps and trained for three days for collecting physical and biochemical measurements as well as conducting interviews. It took twenty minutes on average to conduct each interview along with the physical and biochemical measurements.
## Socio-demographic and behavioral variables (Step I)
Only the core questions about demographic information (step I) with few omissions such as date of birth (due to lack of accuracy on the part of the refugees), marital status etc were used to simplify the questionnaire for the limited time setting. The behavioral components (step I) included the core questions on tobacco, alcohol, physical activity, fruit and vegetable intake and extra salt intake. Information on previous history and treatment of hypertension were also obtained. The average time spent on moderate and vigorous physical activity was transformed into minutes per week. Physical activity less than 150 minutes per week was considered low. A standard measuring cup was used to obtain information on serving sizes of fruits and vegetables in a week.
## Physical and biochemical variables (Step II and III)
In step II, the physical measurements i.e height, weight and blood pressure were measured. The biochemical measurement in step III only included random blood sugar measurement. Systolic and diastolic blood pressure was measured using a manual aneroid sphygmomanometer with an average-sized cuff at sitting and lying positions. The average of the two measurements was used for analysis. Height was recorded in centimeters, and weight was recorded in kilograms using a portable digital weighing scale. Random blood sugar was measured using a standard glucometer.
According to the American Diabetes Association (ADA) guideline, the reference values of random blood sugar between 4.4–7.8 mmol/dl was considered normal, 7.8–11.1 mmol/dl was considered pre-diabetes, and ≥ 11.1 mmol/dl was considered diabetes.
According to the Seventh Joint National Committee (JNC 7), systolic blood pressure <120 mmHg was considered as normal, 120–139 mmHg as pre-hypertension, 140–159 mmHg stage I hypertension, and ≥160 mmHg considered stage II hypertension. Diastolic blood pressure ≤80 mmHg was considered normal, 80–89 mmHg as pre-hypertension, 90–99 mmHg as stage I hypertension and ≥100 considered as stage II hypertension.
According to the WHO guideline 2020, Body Mass Index (BMI) was classified as ≤18.5 as underweight, 18.5–24.99 as normal, 25–29.9 as pre-obesity, and ≥30 as Obesity (class I).
## Statistical analysis
The data was at first entered in Microsoft Excel 2010, and after editing and logical checking; it was analyzed in SPSS version 22.0. Descriptive analysis like frequency, percentage, mean and standard deviation were done as appropriate for the categorical variables. Chi-square test and student t-test were performed to see the association among the categories setting the α level at 0.05. $95\%$ confidence interval was done to see the distribution among the population.
## Ethical considerations
This cross-sectional study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). Ethical clearance was taken from the Research Review Committee and the Ethical Review Committee of the American International University Bangladesh (AIUB). Due permission was taken from the respected authority before entering the restricted camp areas. Written informed consent was taken from each respondent. Blood collection and anthropometric measurement were done by three trained medical assistants with due permission from the respondents.
## Results
Almost an equal number of respondents from both sex (118 men and 131 women) participated in this study with a mean age of 48.6±11.8 years. A majority ($77.8\%$) of the respondents were aged 40 or above. More than half ($67.8\%$) of respondents were unemployed, but the percentage of unemployment was higher ($80.1\%$) among women. The percentage of illiteracy was very high. Eight ($83.5\%$) out of ten respondents were illiterate (Table 1).
**Table 1**
| Variables | Men(n = 118) | 95% CI | Women (n = 131) | 95% CI.1 | Both (n = 249) | 95% CI.2 |
| --- | --- | --- | --- | --- | --- | --- |
| Variables | n(%) | 95% CI | n(%) | 95% CI | n(%) | 95% CI |
| Age, years | | | | | | |
| < 40 | 27 (22.7) | 15.1–30.3 | 30 (23.1) | 15.9–30.3 | 57 (22.4) | 17.2–27.6 |
| 40–50 | 36 (30.2) | 21.9–38.5 | 58 (44.7) | 36.2–53.2 | 94 (37.8) | 31.8–43.8 |
| >50 | 56 (47.1) | 38.1–56.1 | 42 (32.2) | 24.2–40.2 | 98 (40.0) | 33.9–46.1 |
| Mean ± SD | 50.6±12.4 | 48.4–52.9 | 46.8±11.1 | 45.0–49.0 | 48.6±11.8 | 47.2–50.2 |
| Occupation | | | | | | |
| Unemployed | 65 (54.7) | 45.7–63.7 | 104 (80.1) | 73.3–87.0 | 169 (67.8) | 62.0–73.6 |
| Employed | 54 (45.3) | 36.3–54.3 | 26 (19.9) | 13.1–26.7 | 42 (32.2) | 26.4–38.0 |
| Educational status | | | | | | |
| Illiterate | 86 (72.3) | 64.2–80.4 | 122 (93.8) | 89.7–97.9 | 208 (83.5) | 78.9–88.1 |
| Literate | 34 (27.7) | 19.6–35.8 | 8 (6.2) | 2.1–10.3 | 41 (16.4) | 11.8–21.0 |
About one-fourth ($22.9\%$) of the respondents were habitual smokers of whom $31.4\%$ had at least ten years’ habit of smoking. Though smoking was more prevalent among men ($33.9\%$), than women ($13\%$) more than half ($59.5\%$) of the women were habituated with smokeless tobacco and betel leaf consumption. Alcohol consumption was not quite common among the participants. Three ($34.5\%$) out of ten respondents consumed extra salt with their food, and the prevalence was almost the same in both genders. It was also noticeable that the majority ($89.6\%$) of the respondents were sedentary workers. Most ($76.3\%$) of them used to eat ≥5 servings of fruits and vegetables almost 6 days per week (Table 2).
**Table 2**
| Variables | Men(n = 118) | 95% CI | Women (n = 131) | 95% CI.1 | Both (n = 249) | 95% CI.2 | P value* |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Variables | n (%) | 95% CI | n (%) | 95% CI | n (%) | 95% CI | P value* |
| Tobacco smoking | | | | | | | |
| Habitual Smoker | 40 (33.9) | 25.9–43.1 | 17 (13.0) | 6.7–17.9 | 57 (22.9) | 17.7–28.1 | 0.86 |
| Non-smoker | 78 (66.1) | 56.9–64.1 | 114(87.0) | 81.2–92.8 | 192 (77.1) | 71.9–82.3 | 0.86 |
| Years of tobacco smoking | Years of tobacco smoking | | | | | | |
| 10 years of smoking habit | 17 (31.4) | 27.9–37.3 | 16 (12.2) | 4.2–11.8 | 53 (21.3) | 15.4.-28.7 | 0.13 |
| 20 years of smoking habit | 46 (38.9) | 26.6–30.1 | 17 (13.0) | 6.7.2–17.9 | 63(25.3) | 11.2–20.2 | 0.13 |
| Smokeless tobacco and betel leaf consumption | | | | | | | |
| Betel leaf chewing | 69 (58.5) | 50.8–68.6 | 48 (36.6) | 28.0–44.4 | 117 (47.0) | 41.2–53.6 | 0.42 |
| Tobacco leaf chewing | 46 (39.0) | 23.5–30.1 | 30 (22.9) | 15.2–29.4 | 76 (30.5) | 24.8–36.2 | 0.33 |
| Alcohol consumption | 5 (4.2) | 0.6–7.8 | 2 (1.5) | 0.0–3.6 | 7 (2.8) | 0.8–4.8 | 0.78 |
| Extra salt intake | 39 (32.8) | 24.3–41.3 | 47 (36.2) | 28.0–44.4 | 86 (34.5) | 28.6–40.4 | 0.48 |
| Physical activity | | | | | | | |
| Insufficient physical activity | 111(94.9) | 90.9–98.9 | 114(87.0) | 81.2–92.8 | 226 (89.6) | 85.8–93.4 | 0.76 |
| Moderate physical activity** | 0 (0) | 0.0–0.0 | 17 (13.1) | 7.2–18.8 | 17 (6.8) | 3.7–9.9 | 0.76 |
| Heavy physical activity | 6 (5.0) | 1.1–8.9 | 0 (0) | 0.0–0.0 | 6 (2.4) | 0.5–4.3 | 0.76 |
| Fruit and Vegetable Intake | | | | | | | |
| Intake of food and vegetables per week | | | | | | | |
| ≥6 days | 92 (77.3) | 69.7–84.9 | 98 (75.4) | 68.0–82.8 | 190 (76.3) | 71.0–81.6 | 0.07 |
| <6 days | 27 (22.7) | 15.1–30.3 | 32 (24.6) | 17.2–31.9 | 59 (23.7) | 18.4–29.0 | |
| Servings of fruit and vegetable intake | Servings of fruit and vegetable intake | Servings of fruit and vegetable intake | | | | | |
| ≥5 servings per day | 92 (77.3) | 69.7–84.9 | 98 (75.4) | 68.0–82.8 | 190 (76.3) | 71.0–81.6 | 0.07 |
| <5 servings per day | 27 (22.7) | 15.1–30.3 | 32 (24.6) | 17.2–32.0 | 59 (23.7) | 18.4–29.0 | |
About one-third ($33.7\%$) of the participants claimed to have high blood pressure without having any obvious diagnosis, and only one-third ($12.3\%$) take anti-hypertensive medication. The recorded average systolic and diastolic blood pressure was 115.4 ±15.4 mmHg and 74.88± 9.49 mmHg respectively. More than half of the respondents were found to have pre-hypertension according to their systolic ($55.8\%$) and diastolic ($68.3\%$) blood pressure. Three out of ten ($32.2\%$) respondents were newly diagnosed as hypertensive.
The majority ($74.7\%$) of study participants fell under the normal range of BMI (18.5–24.99). In addition, about one-sixth ($16.9\%$) of the participants were documented as overweight and $1.2\%$ as obesity class I. (Table 3).
**Table 3**
| Variables | Men(n = 118) | 95% CI | Women (n = 131) | 95% CI.1 | Both (n = 249) | 95% CI.2 | P value |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Variables | n (%) | 95% CI | n (%) | 95% CI | n (%) | 95% CI | P value |
| History of hypertension | | | | | | | |
| Reported hypertension | 42 (35.3) | 26.7–43.9 | 42 (32.3) | 16.1–30.5 | 84 (33.7) | 27.8–39.6 | 0.05*** |
| H/O taking antihypertensive medications | 42 (35.3) | 26.7–43.9 | 14 (10.8) | 5.5–16.1 | 31 (12.3) | 8.2–16.4 | 0.53*** |
| Systolic blood pressure (mmHg) | | | | | | | |
| <120 | 28 (23.5) | 15.8–31.2 | 35 (26.9) | 19.3–34.5 | 63 (25.3) | 19.9–30.7 | 0.24*** |
| 120–139 | 60 (50.4) | 41.4–59.4 | 79 (60.8) | 52.4–69.2 | 139 (55.8) | 49.6–62.0 | 0.24*** |
| ≥140* | 31 (26.0) | 18.1–33.9 | 16 (12.3) | 6.7–17.9 | 47 (18.9) | 14.0–23.8 | 0.23** |
| Mean ± SD | 116.6±16.8 | | 114.3±13.9 | | 115.4±15.4 | | 0.23** |
| Diastolic blood pressure (mmHg) | | | | | | | |
| <80 | 22 (18.4) | 11.4–25.4 | 24 (18.4) | 11.8–25.0 | 46 (18.4) | 13.6–23.2 | |
| 80–89 | 77 (64.7) | 56.1–73.3 | 93 (71.5) | 63.8–79.2 | 170 (68.3) | 62.5–74.1 | 0.06*** |
| ≥90* | 20 (16.8) | 10.1–23.5 | 13 (10.0) | 4.9–15.1 | 33 (13.3) | 9.1–17.5 | |
| Mean ± SD | 75.5±10.1 | | 74.2±8.9 | | 74.8±9.4 | | 0.64** |
| Newly diagnosed# Hypertension | 51 (42.9) | 27.3–33.8 | 29 (22.3) | 17.7–28.5 | 80 (32.2) | 23.9–44.6 | 0.43*** |
| Confirmed Hypertension## | 93 (78.2) | 64.4–93.2 | 43 (33.1) | 28.9–44.3 | 111 (44.5) | 45.2–79.5 | 0.57*** |
| Body Mass Index (BMI) | | | | | | | |
| Underweight (<18.5 kg/m2) | 4 (3.4) | 0.1–6.7 | 14 (10.8) | 5.5–16.1 | 18 (7.3) | 4.1–10.5 | |
| Normal weight (18.5–24.99 kg/m2) | 93 (78.2) | 70.8–85.6 | 93 (71.5) | 63.8–79.2 | 186 (74.7) | 69.3–80.1 | 0.08*** |
| Overweight (25–29.9 kg/m2) | 20 (16.8) | 10.1–23.5 | 22 (16.9) | 10.5–23.2 | 42 (16.9) | 12.2–21.6 | |
| Obese (≥30 kg/m2) | 2 (1.7) | 0.0–4.0 | 1 (0.7) | 0.0–2.1 | 3 (1.2) | 0.0–2.7 | |
| Mean ± SD | 22.8±2.5 | 22.3–23.2 | 22.0±3.0 | 21.4–22.5 | 22.4±2.8 | 22.0–22.7 | 0.79** |
Although the majority ($84.7\%$) of the study participants showed an average blood glucose level, yet a significant percentage ($14.4\%$) were also discovered to be pre-diabetic (7.8–11.1 mmol/l) from the primary measurement of random blood glucose level (Table 4).
**Table 4**
| Blood sugar | Men | 95% CI | Women (n = 131) | 95% CI.1 | Both (n = 249) | 95% CI.2 | P value |
| --- | --- | --- | --- | --- | --- | --- | --- |
| (mmol/dL) | (n = 118) | 95% CI | n (%) | 95% CI | n (%) | 95% CI | P value |
| (mmol/dL) | n (%) | 95% CI | n (%) | 95% CI | n (%) | 95% CI | P value |
| 4.4–7.8 | 101 (84.8) | 78.3–91.3 | 110 (84.6) | 78.4–90.8 | 211 (84.7) | 80.2–89.2 | |
| 7.8–11.1 | 18 (15.1) | 8.6–21.6 | 18 (13.8) | 7.9–19.7 | 36 (14.4) | 10.0–18.8 | 0.42* |
| >11.1 | 0 (0) | 0.0–0.0 | 2 (1.5) | 0.0–3.6 | 2 (0.8) | 0.0–1.9 | |
| Mean ± SD | 6.3±1.1 | 6.2–6.6 | 6.5±1.5 | 6.3–6.8 | 6.4±1.3 | 6.3–6.6 | |
Among all the risk factors (behavioral, measured and biomedical),insufficient physical activity ($89.6\%$), tobacco use including smokeless tobacco ($53.4\%$) and hypertension ($44.5\%$) were more prevalent among both sexes (Fig 1).
**Fig 1:** *Risk factors of NCD (n = 249).*
## Discussion
According to UNHCR 2017 report, the refugee population living in Ukhiya, Cox’s bazar after the exodus of 2017, have an extremely vulnerable mental and physical state due to past experiences of trauma and persecution along with a bleak living situation in the overcrowded camps [18]. The present study indicates that the course of chronic diseases may rise considering the pattern of living and dietary condition of the camp dwellers at Ukhiya. The study has reported the behavioral risk factors of NCDs among the Rohingya refugees taking shelter in camp settings where most of the participants were unemployed ($67.8\%$), and a significant percentage ($83.5\%$) had no history of schooling. Non communicable diseases have been determined as a significant health challenge among many humanitarian set-ups around the world [21]. According to our study, the trend of NCDs among the Rohingya refugees is consistent with other refugee communities.
Screening of hypertension at refugee set-up has previously shown interesting outcomes in many places. One in five adults from Syrian refugees in Jordan was found to be hypertensive from self-documentation [22]. A need assessment conducted by BRAC on Rohingya refugees in 2018, reported that $51.5\%$ of the refugees had hypertension and $14.2\%$ had diabetes [15]. As per our study, one-third ($32.2\%$) of the adult refugee participants were newly diagnosed as hypertensive and nearly half of them (44.5)%) were confirmed to have stage I and II hypertension. The percentage of confirmed hypertension was higher than the national survey on NCD risk factors among Bangladeshi citizens ($7.9\%$) and the Syrian refugees in Northern Jordan ($39.5\%$) [22, 23]. Only three out of ten hypertensive patients confirmed taking prescribed anti-hypertensive medicines regularly as opposed to their native Bangladeshi counterparts about half of whom take their anti-hypertensive medications regularly [23]. This percentage was also significantly lower than that in adult Syrian refugees ($94.1\%$) who used to take medications on a regular basis [22]. Intake of extra salt with food is considered as the precipitating factor for hypertension, and more than one third of the respondents reported of consuming extra salt with their meals as a daily habit which was much higher than the Bangladeshi nationals ($16.5\%$) [23]. The above scenario indicates that there is importance of screening for hypertension among the refugees in a community setup for early detection and treatment.
Although a negligible ($0.8\%$) number of the respondents were diabetic according to the ADA reference values, $14.4\%$ of the respondents were found to have a pre-diabetic level of random blood sugar. This finding is comparatively better than the Syrian refugees in Jordan, with $9.8\%$ of respondents being reported as diabetic [22]. Since we could only measure random blood glucose, it could not appropriately measure the prevalence of diabetes among the participants. However, our finding reflects that the percentage of pre-diabetics is on the rise, which can create a burden of full-blown diabetes among the refugees in the near future.
Research has found that dependence on tobacco in any form has destructive health consequences [11]. Smoking was found among $22.9\%$ of refugees in our current study which is close to the finding in native Bangladeshi people ($23.5\%$) and nearly double in Palestinian refugees ($36.6\%$) who are habitual smokers [23, 24]. In contrast, the picture is quite the opposite among female smokers. While only $0.8\%$ of Bangladeshi women are smokers, around six out of ten Rohingya women have adopted smoking as a regular habit [23]. On the other hand, the percentage of consumption of smokeless tobacco is almost similar between the Rohingya refugees ($30.5\%$) and the Bangladeshi nationals ($32.0\%$) and higher than the Palestinian refugees ($22.7\%$) in Syria [23, 24]. The refugee populations around the world are commonly seen to be a high-risk group for all form of tobacco addiction [24]. For instance, Palestinian refugees in Syria consume more cigarette and water pipe than the non-refugee residents in Lebanon [24]. In addition, the trend of alcohol and other substance use is high among displaced refugee communities around the globe [25]. For example, the prevalence of harmful alcohol consumption was around $36\%$ among refugee men in Thailand and $23\%$ among male Bhutanese refugees in Nepal. Similar findings have been found among internally displaced men in Uganda and Georgia with a prevalence of hazardous or harmful alcohol consumption of $32\%$ [26]. However, we predicted that religious and cultural preferences have significantly lowered the alcohol consumption rate among the Rohingya refugees ($2.8\%$), which is also compatible with the trend among the Bangladeshi nationals ($1.3\%$). We could not extract information regarding other substance use due to communication barriers.
In the study, $16.9\%$ of the participants were overweight having BMI between 25–29 kg/m2, slightly lower than the Bangladeshi nationals ($20.3\%$), and $1.2\%$ of participants were obese with BMI above 30 kg/m2 [23]. A US-based study on Somalian refugees has shown a thirty times more obese population than the present study on the Rohinga refugees [27]. As we have discussed earlier, the majority of Rohingya refugees are currently workless and living stagnantly, the number of inadequate physical activity ($89.6\%$) in the study reflects the situation precisely. Nearly nine out of ten Rohingya refugees are physically inactive as opposed to three in ten Bangladeshi nationals ($29.1\%$) who have insufficient physical activity [22]. It is presumed that the number of overweight/obesity may rise in the future considering their current condition of living. Moreover, large number of Rohingya refugees ($76.3\%$) reported insufficient amount of fruit and vegetable intake which was a little less than the Bangladeshi nationals ($89.6\%$) [23]. Physical inactivity, overweight and insufficient fruit and vegetable intake are risk factors for cardiovascular diseases, stroke, and cancer [23].
The research has several limitations which are worth mentioning. Primarily, we conducted a cross-sectional study to see the proportion of NCD risk factors among the refugees. It was conducted among the adult refugees of selected camp areas in Ukhiya, Cox’s Bazar, which may not represent all the Rohingya refugees who have currently taken shelter in Bangladesh. The sampling method we used for the study is convenient sampling. We conducted the study among refugees from two pre-selected camps who voluntarily agreed to participate in the study. Since the law enforcement authority of government of Bangladesh maintains a strict surveillance system over the refugees we faced difficulty in conducting a broader study over them including every refugee settlement. In addition, we found many of the refugees who experienced various trauma and persecution from the forceful migration not intent enough to participate in the interviews. This is why we preferred convenient sampling as a method which we understand not to be completely unbiased. Similarly, it was challenging to obtain confirmatory data for diabetes mellitus without measuring the fasting blood glucose level due to the absence of the participants’ adequate cooperation. Hence, we had to rely only on random blood glucose measurement. In addition, single episode of blood pressure measurement may increase the chance of overestimation.
## Conclusion
This study depicts the status of NCD risk factors among a group of adult Rohingya refugees in Ukhiya, Cox’s bazar. Some risk factors in particular, high blood pressure, smoking, consumption of extra salt with food, inadequate physical activity and insufficient fruit and vegetable intake are moderately high among the refugees. Presently, the Bangladesh government and international NGOs working with the refugees prioritize managing infectious diseases, which could be prevented by proper vaccination measures alone. But the course of NCDs among Rohingya refugees cannot be cured by one shot of injection. Even though the patient is cured, there is a chance of developing a disability consequently, which may become a potential threat of a sizeable economic burden for the government of Bangladesh [28] Therefore, the present study may shed some light on this aspect to encourage further research that will guide for implementation of the policies to curb NCD-related behavioral risk factors among the refugees.
## References
1. Montesi L, Caletti MT, Marchesini G. **Diabetes in migrants and ethnic minorities in a changing world**. *World J Diabetes* (2016.0) **7** 34. DOI: 10.4239/wjd.v7.i3.34
2. 2World Health Organization Fact sheets. Obesity and Overweight. 2020. Available from: https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight
3. 3World Health Organization Fact sheets. Non Communicable Diseases. 2018. Available from: https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases
4. Buse K, Aftab W, Akhter S, Phuong LB, Chemli H. **The state of diet-related NCD policies in Afghanistan, Bangladesh, Nepal, Pakistan, Tunisia and Vietnam: a comparative assessment that introduces a ‘policy cube’approach.**. *Health Policy Plan* (2020.0) **35** 503-21. DOI: 10.1093/heapol/czz175
5. Kim HC, Oh SM. **Noncommunicable diseases: current status of major modifiable risk factors in Korea.**. *J Prev Med Public Health* (2013.0) **46** 165. DOI: 10.3961/jpmph.2013.46.4.165
6. Hosseinpoor A.R., Bergen N., Kunst A, Harper S, Guthold R, Rekve D. **Socioeconomic inequalities in risk factors for non communicable diseases in low-income and middle-income countries: results from the World Health Survey.**. (2012.0) **12**. DOI: 10.1186/1471-2458-12-912
7. Allen L, Cobiac L, Townsend N. **Quantifying the global distribution of premature mortality from non-communicable diseases.**. *J Public Health* (2017.0) **39** 698-703. DOI: 10.1093/pubmed/fdx008
8. 8WHO Commission on Social Determinants of Health, World Health Organization.
Closing the gap in a generation: health equity through action on the social determinants of health: Commission on Social Determinants of Health final report. World Health Organization; 2008. Available from: https://www.who.int/publications/i/item/WHO-IER-CSDH-08.1. *Closing the gap in a generation: health equity through action on the social determinants of health: Commission on Social Determinants of Health final report* (2008.0)
9. Misra A, Ganda OP. **Migration and its impact on adiposity and type 2 diabetes**. *Nutrition* (2007.0) **23** 696-708. DOI: 10.1016/j.nut.2007.06.008
10. Berkowitz SA, Fabreau GE, Raghavan S, Kentoffio K, Chang Y, He W. **Risk of developing diabetes among refugees and immigrants: a longitudinal analysis**. *J Community Health* (2016.0) **41** 1274-81. DOI: 10.1007/s10900-016-0216-4
11. Furber S, Jackson J, Johnson K, Sukara R, Franco L. **A qualitative study on tobacco smoking and betel quid use among Burmese refugees in Australia.**. *J Immigr Minor Health* (2013.0) **15** 1133-6. DOI: 10.1007/s10903-013-9881-x
12. Kubiak RW, Sveum EM, Faustin Z, Muwonge T, Zaidi HA. **Prevalence and risk factors for hypertension and diabetes among those screened in a refugee settlement in Uganda.**. *Confl Health* (2021.0) **15** 1-8. PMID: 33390172
13. Marchesini G, Bernardi D, Miccoli R, Rossi E, Vaccaro O. **Under-treatment of migrants with diabetes in a universalistic health care system: the ARNO Observatory**. *Nutr Metabol Cardiovasc Dis* (2014.0) **24** 393-9
14. Joarder T, Sutradhar I, Hasan MI, Bulbul MM. **A record review on the health status of Rohingya refugees in Bangladesh.**. *Cureus* (2020.0) **12**. DOI: 10.7759/cureus.9753
15. Jeffries R, Abdi H, Ali M, Bhuiyan AT, El Shazly M. **The health response to the Rohingya refugee crisis post August 2017: reflections from two years of health sector coordination in Cox’s Bazar, Bangladesh.**. *PLoS One.* (2021.0) **16**
16. 16Rapid Needs Assessment of the Rohingya in Cox’s Bazar. FXB Center for Health & Human Rights | Harvard University. 2018. Available from: https://fxb.harvard.edu/2018/10/09/a-rapid-assessment-of-rohingya-in-coxs-bazaar/
17. Bardenheier B.H, Phares C.R, Simpson D, Gregg E, Cho P., Benoi S. **Trends in chronic diseases reported by refugees originating from Burma resettling to the United States from camps versus urban areas during 2009–2016.**. *J Immigr Minority health* (2019.0) **21** 246-256
18. 18UNHCR, the 2017 year-end report. 2017. Available from: https://reporting.unhcr.org/sites/default/files/pdfsummaries/GR2017-Bangladesh-eng.pdf
19. Rawal LB, Biswas T, Khandker NN, Saha SR, Bidat Chowdhury MM, Khan AN. **Non-communicable disease (NCD) risk factors and diabetes among adults living in slum areas of Dhaka, Bangladesh.**. (2017.0) **12** e0184967. DOI: 10.1371/journal.pone.0184967
20. 20Rohingya refugee crisis in depth | MSF [Internet]. Médecins Sans Frontières (MSF) International. 2022. Available from: https://www.msf.org/rohingya-refugee-crisis-depth
21. Harris P, Kirkland R, Masanja S, Le Feuvre P, Montgomery S, Ansbro É. **Strengthening the primary care workforce to deliver high-quality care for non-communicable diseases in refugee settings: lessons learnt from a UNHCR partnership**. *BMJ Global Health* (2022.0) **7** e007334. DOI: 10.1136/bmjgh-2021-007334
22. Ratnayake R, Rawashdeh F, AbuAlRub R, Al-Ali N, Fawad M, Hani MB. **Access to care and prevalence of hypertension and diabetes among Syrian refugees in northern Jordan**. *JAMA Netw Open* (2020.0) **3** e2021678. DOI: 10.1001/jamanetworkopen.2020.21678
23. 23World Health Organization. Country Office for Bangladesh. National STEPS survey for non-communicable diseases risk factors in Bangladesh 2018. Available from: https://apps.who.int/iris/handle/10665/332886
24. Jawad M, Khader A, Millett C. **Differences in tobacco smoking prevalence and frequency between adolescent Palestine refugee and non-refugee populations in Jordan, Lebanon, Syria, and the West Bank: cross-sectional analysis of the Global Youth Tobacco Survey.**. *Confl Health* (2016.0) **10** 1-8. PMID: 26865857
25. Caetano R, Vaeth PA, Chartier KG, Mills BA. **Epidemiology of drinking, alcohol use disorders, and related problems in US ethnic minority groups.**. *Handb Clin Neurol.* (2014.0) **125** 629-48. DOI: 10.1016/B978-0-444-62619-6.00037-9
26. Fuhr DC, Bogdanov S, Tol WA, Nadkarni A, Roberts B. **Problem Management Plus and Alcohol (PM+ A): A new intervention to address alcohol misuse and psychological distress among conflict-affected populations.**. *Intervention* (2021.0) **19** 141
27. Njeru JW, Tan EM, St Sauver J, Jacobson DJ, Agunwamba AA, Wilson PM. **High rates of diabetes mellitus, pre-diabetes and obesity among Somali immigrants and refugees in Minnesota: a retrospective chart review.**. *J Immigr Minority Health.* (2016.0) **18** 1343-9. DOI: 10.1007/s10903-015-0280-3
28. Bloom DE, Cafiero E, Jané-Llopis E, Abrahams-Gessel S, Bloom LR, Fathima S. *The global economic burden of non-communicable diseases* (2012.0)
|
---
title: A qualitative investigation of facilitators and barriers to DREAMS uptake among
adolescents with grandparent caregivers in rural KwaZulu-Natal, South Africa
authors:
- Dumile Gumede
- Anna Meyer-Weitz
- Thembelihle Zuma
- Maryam Shahmanesh
- Janet Seeley
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022343
doi: 10.1371/journal.pgph.0000369
license: CC BY 4.0
---
# A qualitative investigation of facilitators and barriers to DREAMS uptake among adolescents with grandparent caregivers in rural KwaZulu-Natal, South Africa
## Abstract
Adolescents with grandparent caregivers have experienced challenges including the death of one or both parents due to HIV in sub-Saharan Africa. They may be left out of existing HIV prevention interventions targeting parents and children. We investigated the facilitators and barriers to DREAMS (Determined, Resilient, Empowered, AIDS-free, Mentored and Safe) programme uptake among adolescents with grandparent caregivers across different levels of the socio-ecological model in rural South Africa. Data were collected in three phases (October 2017 to September 2018). Adolescents (13–19 years old) and their grandparent caregivers (≥50 years old) ($$n = 12$$) contributed to repeat in-depth interviews to share their perceptions and experiences regarding adolescents’ participation in DREAMS. Data were triangulated using key informant interviews with DREAMS intervention facilitators ($$n = 2$$) to give insights into their experiences of delivering DREAMS interventions. Written informed consent or child assent was obtained from all individuals before participation. All data were collected in isiZulu and audio-recorded, transcribed verbatim and translated into English. Thematic and dyadic analysis approaches were conducted guided by the socio-ecological model. Participation in DREAMS was most effective when DREAMS messaging reinforced existing norms around sex and sexuality and when the interventions improved care relationships between the adolescents and their older caregivers. DREAMS was less acceptable when it deviated from the norms, raised SRH information that conflicts with abstinence and virginity, and when youth empowerment was perceived as a potential threat to intergenerational power dynamics. While DREAMS was able to engage these complex families, there were failures, about factors uniquely critical to these families, such as in engaging children and carers with disabilities and failure to include adolescent boys in some interventions. There is a need to adapt HIV prevention interventions to tackle care relationships specific to adolescent-grandparent caregiver communication.
## Introduction
The HIV epidemic and the high mortality of the biological parent generation have left many adolescents in the care of grandparents in sub-Saharan Africa (SSA) [1]. In 2019, the General Household Survey in South Africa found that $3.1\%$ of children (0–17 years old) were maternal orphans, $9\%$ of children were paternal orphans, and $2.4\%$ of children were double orphans [2]. The percentage of orphaned children in KwaZulu-Natal was $18.7\%$ and one of the highest in the country [2]. Adolescents (10–19 years old) with grandparent caregivers are a vulnerable and critical population in efforts to prevent HIV acquisition.
The death of one or both parents due to HIV can impact the adolescents’ well-being throughout their lifetime [3] and increase their vulnerability to HIV and risky behaviours [4]. It also has a significant influence on their care arrangements due to household dissolution and migration [5]. The burden of care for these adolescents has often been undertaken by grandparents [6]. Grandparent caregiving can occur suddenly or after a prolonged illness of biological parents [7]. While HIV care and antiretroviral therapy (ART) services have significantly reduced AIDS-related deaths [8], large numbers of orphaned and vulnerable adolescents are still cared for by grandparents in South Africa and elsewhere in SSA [6, 9].
Being raised by grandparent caregivers can be difficult for adolescents, considering that many developmental, psychological, social, and structural transitions converge in this period of their life [10]. It is a development phase characterized by increased risk-taking and greater peer influence in decision-making [11]. During this development phase, new boundaries are explored, and caregiver rules and societal norms that were unchallenged during childhood are re-examined, questioned, and may be challenged in preparation for adulthood [12]. Adolescents with grandparent caregivers have particular challenges which other adolescents may not face. These adolescents face more socio-economic risks including poverty that can affect their development in very unique ways than other adolescents [13]. Researchers have reported that these adolescents are usually at increased risk of poorer health and educational outcomes compared to other adolescents [14]. These challenges place adolescents at increased risk for emotional and behavioural problems [7] as well as loss of social opportunities [6, 15].
Given that young people aged 15 to 24 years represent the largest population living with HIV in South Africa, several studies have explored sex communication between adolescents and their caregivers [16–18]. These studies are based on the premise that effective adolescent-caregiver communication on sex is a protective factor in adolescent risky sexual behaviours [17, 19]. For example, a study conducted in KwaZulu-Natal of adolescents and their caregivers participating in a family-centered HIV prevention intervention found that adolescents’ HIV and condom use knowledge significantly improved [17]. However, South African studies also state that adolescent-parent communication about sex is often negative or punitive [20, 21], which might therefore hinder open communication about minimizing unsafe sexual behaviour [22], with the generation gap between older and younger generations blocking sexuality communication [23]. This gap may be further exacerbated when the carers are grandparents. While many of the previous studies have focused on adolescent-parent communication, adolescent-grandparent communication especially around sexuality and HIV can be difficult yet not much is known about their circumstances. Thus, the adolescents with grandparent caregivers are particularly vulnerable to acquiring HIV [24] and very little is known about how they communicate about sexuality and HIV with their grandparent caregivers.
While HIV prevention interventions are increasingly targeting adolescents in general [25], very few interventions are sensitive to the many adolescents with grandparent caregivers in South Africa. This group of adolescents are a uniquely challenged generation and their invisibility in policy framing could imply that they are left out of existing HIV prevention interventions. Understanding the dynamics and contextual idiosyncrasies shaping the experiences of adolescents with grandparent caregivers is important for identifying factors that impact decision-making relating to initiating sexual activity and preventing pregnancy. Being aware of the issues that influence adolescents’ lives in grandparent families and their participation in HIV prevention interventions can help implementers provide improved, family-oriented and result-driven services.
In recent years, there has been increased public health interest in offering HIV prevention interventions for adolescents [26–29]. However, little is known about the contextual factors that shape the participation of adolescents living with grandparent caregivers in HIV prevention interventions. The socio-ecological model (SEM) acknowledges interrelated individual and contextual factors that influence behaviours [30]. It is described by using multilevel circles that show the individual in the centre and surrounded by the family, community, organisational, and public policy [30]. The behavioural influence of the interactions across the different levels has been discussed by a great number of authors in literature [26, 31, 32]. For example, a recent study in South Africa suggests that socio-ecological factors influenced young people’s intention to access and utilize health services [26]. Although there are many studies, the research on the socio-ecological factors that shape the participation of adolescents with older caregivers in HIV prevention interventions remains limited. To fill this literature gap, in this paper, we investigated the facilitators and barriers to participation in HIV prevention interventions among adolescents with grandparent caregivers across different levels of the SEM covering individual, interpersonal, community and organisational components in the context of DREAMS programme in rural South Africa.
## Study setting
This study was conducted in uMkhanyakude district of northern KwaZulu-Natal, South Africa. The district is predominantly rural with high rates of poverty and unemployment [33]. In this setting, the HIV prevalence and incidence are high [34]. About $19\%$ of adolescent girls and young women (AGYW) and $5.6\%$ of adolescent boys and young men (ABYM) are living with HIV [35].
Within this context, a multi-sectoral HIV combination prevention programme–the DREAMS programme was implemented between April 2016 − September 2018 to reduce HIV infection in AGYW (and their male sexual partners) through evidence-based health, educational and social interventions [34]. The DREAMS programme is an investment by the U.S. Government President’s Emergency Plan for AIDS Relief (PEPFAR) office, Bill and Melinda Gates Foundation, Girl Effect (formerly the Nike Foundation), and other private sector partners, announced in 2014 [36]. The overall aim of DREAMS was to reduce HIV incidence in AGYW through a combination of interventions that target community, family, male partners, and AGYW to promote safer sexual relations, social protection, biological protection, and empower AGYW in 10 countries in sub-Saharan Africa including South Africa [11]. The DREAMS programme had many components, as described elsewhere [37], that were delivered through different DREAMS implementing partners. The DREAMS implementing partners, sometimes, delivered DREAMS through local community-based organisations (CBOs). UMkhanyakude district is one of the sites in which the DREAMS programme was implemented in South Africa [34].
## Study design, sampling and study participants
Adopting a qualitative method study design, we selected a purposive sample of adolescents aged 13–19 years, grandparent caregivers aged 50 years and over, and DREAMS intervention facilitators to painstakingly collate data for this study. Key inclusion criteria were (a) adolescents (boy or girl) aged 13–19 years who were the recipient of at least one DREAMS intervention and their primary caregivers were grandparents (male or female) aged 50 years and above, and (b) DREAMS intervention facilitators (male or female) working for a community-based organisation that was subcontracted by a DREAMS implementing partner for the delivery of DREAMS interventions in the study area.
## Data collection
Repeat in-depth interviews were conducted in the rural community of Mtubatuba sub-district with adolescent-grandparent caregiver dyads and key informant interviews with DREAMS intervention facilitators. Interview guides were used, which consisted of a series of open-ended semi-structured questions tailored for each category of the study participants, to gain insight into the DREAMS interventions and factors shaping adolescents’ participation in DREAMS. Separate interviews were conducted with adolescents and their caregivers to allow each dyad partner to express beliefs and perceptions more freely. Data collection was spread through three phases, at every four months, over 12 months from October 2017 to September 2018 in order to elicit rich qualitative data and for prolonged engagement. Each one-on-one interview was carried out in an undisturbed and private setting of choice, for example, in the participants’ homes or their gardens. The first author (DG) is a social scientist trained in qualitative data collection and speaks both isiZulu (the local language of the participants) and English fluently. DG conducted all the interviews, audio-recorded, transcribed verbatim and translated all data into English. TZ, a local senior social scientist, checked for the accuracy of translations and cultural appropriateness. Throughout the interviews, field notes were kept to detail pre- and post-interview reflective thoughts, observations and impressions.
## Data analysis
Translated data were coded and managed using Atlas.ti 8. Both thematic [38] and dyadic [39] approaches were used in the analysis. The first author (DG) conducted the main analysis of the interviews to develop the initial coding framework based on the levels of the socio-ecological model and salient issues that arose in the data. Over a series of analytical meetings with the researchers (AMW, TZ, MS and JS), the final coding framework was developed. The debriefing also facilitated reflexivity and exploration of alternative explanations. When coding was complete, DG developed data matrices which highlighted prominent themes relevant to adolescents’ facilitators and barriers to DREAMS. Data from the field notes were also used to further inform the development of themes. To increase the rigour of the analysis, data source triangulation, prolonged engagement and debriefing [40] were adopted in this study. Quotes were extracted from the transcripts to illustrate common responses. COREQ guidelines for qualitative papers were followed.
## Ethical considerations
The University of KwaZulu-Natal Humanities and Social Sciences Research Ethics Committee (ref. HSS/$\frac{1109}{017}$D) reviewed and approved the study, along with approval from the Community Advisory Board (CAB). Individuals provided written informed consent before they participated in the study; however, assent and caregiver written consent was obtained for all adolescents aged below 18. No names of participants were recorded; instead, pseudonyms were used.
## Characteristics of study participants
A total of 36 repeat in-depth interviews were conducted with adolescent-grandparent caregiver dyads ($$n = 12$$). Participants’ characteristics are summarized in Table 1. Of the six adolescents, five were adolescent girls and one was an adolescent boy aged between 13 and 19 years. Each adolescent received one or more DREAMS participatory group training interventions through a local CBO. These interventions were Stepping Stones, Let’s Talk, and Vhutshilo.
**Table 1**
| Adolescents (n = 6) | Adolescents (n = 6).1 | Adolescents (n = 6).2 | Adolescents (n = 6).3 | Adolescents (n = 6).4 | Adolescents (n = 6).5 | Grandparent caregivers (n = 6) | Grandparent caregivers (n = 6).1 | Grandparent caregivers (n = 6).2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Namea and age (years) in 2017 | Sex | DREAMS Intervention | In school | Biological mother alive | Biological father alive | Name and age (years) in 2017 | Sex | Relationship |
| Thabani (15) | Mb | Stepping Stones | Yd | Y | Y | MaNdawo (76) | F | Paternal grandmother |
| Neli (14) | Fc | Let’s Talk | Y | Y | N | MaZulu (64) | F | Maternal grandmother |
| Zama (15) | F | Let’s Talk | Ne | Y | Y | MaNgubo (80) | F | Maternal grandmother |
| Sane (13) | F | Stepping Stones, Let’s Talk | Y | Y | N | MaDube (58) | F | Maternal grandmother |
| Thandi (13) | F | Let’s Talk, Vutshilo | Y | Y | N | MaJali (56) | F | Maternal grandmother |
| Mpume (19) | F | Let’s Talk | N | Y | N | MaKhoza (64) | F | Paternal grandmother |
The Stepping Stones intervention focuses on improving gender equity and communication and relationship skills [41]. The Let’s *Talk is* a structured, manualized, small-group HIV prevention intervention with separate and joint sessions for adolescents age 13 or older and their caregivers and is designed to assist in effective communication between caregivers and adolescents, strengthening family relationships, and mitigating adolescent sexual risk [42]. In the same way, the *Vhutshilo is* a curriculum designed for orphaned and vulnerable adolescents, delivered by peer educators from the same background and community, to provide HIV prevention skills and psychosocial support [43].
With regard to their grandparent caregivers, all of them were older women aged 56 to 80 years. Four were maternal grandmothers and two paternal grandmothers to the focal adolescents.
Additionally, one male aged 30 (Stepping Stones intervention facilitator) and one female aged 41 (Let’s Talk intervention facilitator) were key informants, and they were both employed by the local CBO to deliver DREAMS interventions in uMkhanyakude district.
## Facilitators and barriers to DREAMS participation using the socio-ecological framework
The facilitators and the barriers to participation in DREAMS HIV prevention interventions among adolescents in older carer families are organised into four levels of the socio-ecological model (SEM) namely individual, interpersonal, organisational, and community levels, as indicated in Fig 1. For each level, themes are supported by illustrative quotes from the interviews.
**Fig 1:** *Facilitators and barriers to DREAMS participation among adolescents with grandparent caregivers.*
## Obtaining health information and acquiring new skills
Some adolescents mentioned that obtaining relevant information regarding HIV and health issues motivated them to participate in DREAMS as these were contextual issues affecting young people. They listed topics which they discussed with intervention facilitators during DREAMS sessions. Broadly, these topics included HIV, sexually transmitted infections (STIs), Tuberculosis (TB), condoms, contraceptives, communication, and career planning.
It appears from Thabani’s narrative that the factors motivating participation in DREAMS went beyond acquiring relevant health-related information to include gaining knowledge participants described as relevant for ensuring harmonious family relations and stability. In the same way, the DREAMS intervention facilitators also affirmed that the adolescents enjoyed discussing topics such as condoms, sexual relationships, and career planning during the DREAMS participatory sessions.
In addition, the DREAMS intervention facilitators mentioned that the topic of anger management revealed that the adolescents were dealing with anger issues regarding gender inequalities at home and resented their parents’ sexual relationships with new partners other than their biological parents.
Further, the DREAMS intervention sessions provided the opportunity for the adolescents to talk about negative caring behaviours of their caregivers such as unequal gender treatment at home. The opportunity to talk about gender inequalities at home raised emotions for the adolescents. However, learning about anger management in DREAMS interventions equipped the adolescents with skills to control negative emotions in communication, and in adjusting to parents’ sexual relationships, considering that the biological parents of the adolescents that participated in the study were no longer in the relationship.
## Becoming a new person
Some adolescents mentioned that they reflected on themselves and saw a changed version of who they are in terms of their abilities in choosing friends and setting boundaries. One adolescent explained that participating in DREAMS assisted her to redefine the types of friends she wanted and to setting the limits with her friends in terms of what was acceptable and unacceptable towards her: It seems that DREAMS interventions also reinforced some behaviours and values such as showing respect to adults that the grandparents found acceptable. There was a sense that where the messages taught by DREAMS initiatives align with normative cultural ideas upheld by grandparent caregivers, then what was considered positive change was more likely to happen.
Others mentioned they believed they had changed from behaviours that were considered risky and those that were regarded as unacceptable due to participation in DREAMS. They constructed reflections of themselves pre-and during DREAMS participation. From their reflections, they believed that pre-DREAMS participation they had the tendency of teaming up with bad friends to engage in risky behaviours such as flirting to get money from older men and unacceptable behaviours such as not doing household chores. It seems that hanging out with bad friends was associated with leading to engagement in risky and unacceptable behaviours. However, during DREAMS participation, they believed they overcame the tendency of being negatively influenced by bad friends to engage in risky behaviours. They were, thus, motivated to participate in DREAMS due to seeing themselves becoming a new version of themselves.
However, the dyadic analysis revealed some similarities and differences in perspectives between some adolescents and their grandparent caregivers regarding whether the adolescents had adopted acceptable behaviours or not. For example, one adolescent stated that she used to flirt to get money from older men at the taverns and slept away from home, although she did not admit to having sex with the men she seduced for money. However, she went on to mention that she transformed into a new person through her participation in DREAMS: Similarly, in a separate interview with her grandmother, she confirmed that there was a big difference in her granddaughter’s behaviour at the time she participated in DREAMS. She abstained from being absent a lot from home and spent more time at home with the family. The grandmother was happy about the changes she saw in her granddaughter, and this improved their relationship: It was clear that both the adolescent and her grandparent caregiver agreed that the adolescent had changed her behaviour of sneaking out of the house and sleeping away from home. Pre-DREAMS, they had a difficult relationship due to the adolescent’s unacceptable and risky behaviours. Moreover, the transformation that occurred in the adolescent’s behaviours improved the care relationship as they spent time bonding at home.
While some dyads agreed that participation in DREAMS had transformed the adolescents to adopt acceptable behaviours, the other dyads did not have a similar position on this. An example of a disagreement is seen between Thabani and his grandmother. According to Thabani, he perceived himself as having developed the ability to control his negative emotions and attitudes: However, in a separate interview, his grandmother had a contrary view of her grandson and said that she did not notice a difference in his ability to manage anger: It appeared that the members of the dyad had different perceptions of how DREAMS shaped the adolescent’s behaviours. The boy thought that he had become a better person as a result of participating in the intervention while his grandmother felt he had not changed the behaviours that were considered unacceptable.
## Caregiver permission to participate in DREAMS
Caregiver permission was one of the factors that facilitated the participation of adolescents with older caregivers in DREAMS.
Culturally, adolescents could not consent to participate in HIV prevention interventions and thus the caregiver permission necessitated their participation in DREAMS.
## Collaborative relationships between intervention facilitators and adolescents
A sense of collaborative relationships that developed between the adolescent DREAMS recipients and the intervention facilitators over the course of DREAMS positively contributed to the retention of adolescents in the interventions. One intervention facilitator shared his experience that being able to build rapport with the adolescents made them more willing to attend and actively participate in DREAMS: It was evident that the intervention facilitators endeavoured to nurture good relationships with the adolescents in promoting their uptake in DREAMS interventions and emphasizing safe social spaces for the adolescents to tackle health and social issues affecting their lives.
## DREAMS as an opportunity for young people to socialize
In addition, adolescent participation in DREAMS provided an opportunity for the adolescents to socialize with other young people as they connected with peers and friends in a safe space. Many adolescents came along with their friends to participate in DREAMS and, thus, created a sense of cohesion and belonging among the adolescents when they were amongst their friends: Thus, participation in DREAMS created a friendly space for the adolescents to enjoy themselves away from home. It is important to note that the adolescents in older carer families mentioned that the older carers restricted the adolescents’ movements and choice of friends. It was clear that these adolescents had limited opportunities to socialize with their peers as they had responsibilities at home helping older carers and, sometimes, caregivers of their older carers as they had chronic illnesses. As a result, participating in DREAMS brought an opportunity to get away from the home environment and be with friends to have fun, to play games, and to enjoy refreshments.
The DREAMS intervention facilitators also shared similar sentiments that adolescents were motivated to participate in the interventions in order to meet with friends: It seemed DREAMS provided the opportunity for the adolescents to disengage from school work, and housework and they were free from caregiver control. This implies that adolescents desired a space to socialize with other young people.
## Improved care relationships and adolescent-older carer communication
One of the main factors facilitating participation in DREAMS was that it improved care relationships and communication between the adolescents and their grandparent caregivers. One of the DREAMS interventions, Let’s Talk, seems to be playing an important role in strengthening communication between some adolescents and their older carers. This was illustrated by one adolescent who participated in the Let’s Talk intervention together with her grandmother: It seemed the generational gap between the adolescents and their grandparent caregivers shaped the nature of communication between them. The adolescents mentioned fear of expressing their feelings to their grandparent caregivers in order to avoid being labelled as ‘badly-behaved’ or disrespectful to adults. Participating in DREAMS appeared to enable the dyads to discuss issues openly and share their feelings on different aspects.
According to the Let’s Talk intervention facilitator, the intervention offered separate sessions for the adolescents and for the caregivers/parents as well as joint sessions for both the pairs. Only two adolescents in this study participated in the Let’s Talk intervention with their older carers.
Further, the adolescents shared that participation in DREAMS resulted in them being more obedient and respectful towards adult authority. Across the dyads, the older carers emphasised that they expected respect from their adolescent grandchildren, which the adolescents understood as implying ‘obeying adults and doing as instructed’. One adolescent illustrated that her participation in DREAMS was motivated by respect for her grandmother in order to avoid conflict in the relationship with her: It was clear that Zama continued participating in DREAMS to avoid conflict between her and her grandmother and prioritized maintaining a positive relationship with her grandmother.
One dyad shared how DREAMS improved their communication and the care provided by the older carer to the adolescent. The older carer said that participating in the Let’s Talk intervention taught her to see things from the adolescent’s perspective, leading to improved communication and a better relationship: Likewise, in a separate interview with Sane, she concurred with her grandmother: *As a* result of participation in the Let’s Talk intervention, it seems that engaging in open conversations about sex and HIV between the adolescents and their older carers was enhanced and resulted in more positive caring relationships. Unfortunately, a platform for effective communication between the adolescents and their older carers was not sustained since DREAMS ended.
## Attributes of DREAMS intervention facilitator
Some older caregivers described the positive qualities exhibited by the DREAMS intervention facilitator that enabled the grandparents to support the participation of their adolescent grandchildren in DREAMS. She was described as a respected community member, a local pastor’s wife, and a co-founding member of a community crèche. Many older carers regarded her as trustworthy, a reliable source of information, having a special ability to work with community members, and being responsive to their needs. This trust was displayed in interviews with the older carers: The perceptions that the grandmothers had about the intervention facilitator played a significant role in the grandmothers to associate DREAMS as a solution to adolescent risk behaviours and community development.
## Access to community-based DREAMS intervention sites
Community-based DREAMS intervention sites were convenient and easily accessible, thus facilitating the participation of adolescents in the interventions.
Older carers were also comfortable knowing that their adolescent grandchildren were in the neighbourhood as they were concerned with the safety of the adolescents, as one stated:
## Access to service layers
Access to HIV testing and other services which were offered by the CBO implementing DREAMS interventions also facilitated the uptake of DREAMS by the adolescents. One DREAMS intervention facilitator explained that sometimes some adolescents were transferred from one intervention to another within the organisation if the adolescents appeared not ready to exit DREAMS. The local CBO also collaborated with other service organisations to refer the DREAMS recipients for additional services. Some of the services mentioned were HIV Testing Services (HTS), TB screening and testing, and contraceptives.
The local CBO that delivered DREAMS interventions also provided wider social and health services in the community. While delivering DREAMS interventions, the organisation also linked the adolescents with government departments to obtain birth certificates and child support grants. One adolescent stated: Access to HIV testing and the additional social services generated interest and motivated the adolescents to take part in DREAMS.
## Involvement and support of community leadership
The community leaders played an important role in supporting the local CBO in delivering DREAMS interventions by encouraging community members including grandparents, through community meetings, to allow the young people to participate in DREAMS. Additionally, the community leaders reached an agreement with the local CBO to hire people from the community in order to increase employment opportunities for local people and to facilitate better linkages with the organisation on any matter regarding DREAMS interventions.
The involvement and support of community leaders were critical enablers of the continued participation of adolescents in DREAMS interventions, and the leaders articulated their demands to the CBO to maintain social cohesion and community economic development through the employment of local people.
## Diverse learning styles
Diverse learning styles among adolescents hindered learning in the behavioural interventions for HIV prevention. Two adolescents expressed challenges they faced due to the learning methods which were used to deliver DREAMS interventions. One adolescent narrated his challenge of participating in Stepping Stones intervention: It was clear that Thabani struggled with role-play scenarios depicting gender-based violence (GBV) and was not comfortable with sharing his personal experiences about sexual organs within a group setting. However, during the interview, he shared that he never experienced any form of GBV in real life.
On the other side, the other adolescent who participated in Let’s Talk and Vhutshilo interventions shared her frustrations with participating in written tasks as she preferred oral activities: It appeared that the written tasks in DREAMS hindered participation and thus affected her willingness to actively engage in a meaningful way.
## Internalised stigma of being labelled as ‘badly-behaved’
Internalised stigma of being labelled as ‘badly-behaved’ hindered adolescents’ participation in DREAMS. It was mentioned that being a sexually active adolescent was prohibited and thus labelled as ‘badly-behaved’. Mocking of pregnant girls, both at school and in the community, generally hindered their participation in services as they opted to isolate themselves.
It was also mentioned that some pregnant adolescents dropped out of school due to stigma and teenage mothers did not participate in DREAMS interventions to avoid being labelled as ‘badly-behaved’. This account exemplifies the internal challenges and battles adolescents faced once they became pregnant and labelled as ‘badly-behaved’. The DREAMS intervention facilitator explained that there was a misconception that DREAMS was targeting the ‘well-behaved’ young people. However, the facilitator mentioned they consistently corrected the misconception by emphasising that: Tension arose when the DREAMS intervention engaged those who were labelled as ‘badly behaved’ and, thus, facilitated hesitancy and internalised stigma from others to participate in DREAMS.
## Obtaining fragmented Sexual and Reproductive Health (SRH) information
Some adolescents mentioned that they received fragmented SRH information between their churches and DREAMS. They reported that they were taught about contraceptives and different types of contraceptives to prevent unplanned pregnancies and HIV in DREAMS. In contrast, their churches prohibited premarital sex and thus the use of contraceptives by youth. One adolescent narrated teaching from his church: Some adolescents also indicated that their churches were silent about HIV and instead they were taught about abstinence: Further, the church messages were also advocated at home by the older carers. It was clear that adolescents were conflicted with messages they received from their church, home, and DREAMS. Messages that reinforced what the adolescents knew, based on church and home teachings were easier to absorb than those that conflicted with normative thinking related to abstinence and virginity.
## Negative peer pressure
Negative peer pressure emerged as a barrier to participation in DREAMS for some adolescents. This was illustrated by one adolescent as she discussed that she experienced negative peer pressure as her friends demotivated her from participation: Despite the adolescents stating they enjoyed attending DREAMS with their friends, the experience was different for this adolescent whose friends withdrew their participation and thus expected her to pull out from DREAMS.
## Refusal by grandparent caregivers or parents
DREAMS intervention facilitators reported that some adolescents in the community were not allowed by their grandparent caregivers or parents to participate in DREAMS. Two reasons were mentioned for refusals. Firstly, some adolescents were not allowed to participate in DREAMS as a punishment for not carrying out household chores. Some caregivers made restrictions that adolescents were only allowed to attend DREAMS sessions once they had completed the household chores: Secondly, another DREAMS intervention facilitator mentioned that some adolescents left their homes as if they were going to attend DREAMS, but instead went to hang out elsewhere. When caregivers learnt about these lies, they immediately refused to allow the adolescents to continue participating in order to punish them for lying.
## Caregivers’ lack of information
A lack of information by older caregivers about DREAMS hindered young people’s participation in the interventions. This was illustrated by two older caregivers: It appeared that while the grandparents provided permission to the adolescents to participate in DREAMS, the intervention facilitators may not have provided sufficient information for the grandparents to comprehend the content of information to be received by adolescents during the HIV prevention interventions. Therefore, the grandparents needed detailed explanations from their grandchildren who seemed not keen to share openly with their grandparents, unless the grandparents were persistent in finding out from the adolescents. The lack of information limited the support that the older carers provided to the adolescents who were participating in DREAMS and compromised the care relationships.
## Limited financial resources
We found that limited financial resources to reach the many adolescents who needed HIV prevention interventions hindered adolescents’ participation in DREAMS. This was reflected in the explanation by the DREAMS intervention facilitator: Further, limited funding constrained the organisation from hiring sufficient programme facilitators to facilitate planned programme sessions in the various sites. It was indicated the intervention facilitators were overwhelmed by the demand to facilitate the number of groups in the different sites. The facilitators felt that the workload and pressure to meet targets, sometimes made it difficult for them to be punctual for the planned group sessions with the adolescents. In turn, disruption of the group sessions frustrated the adolescents, especially when there was no communication about delays or cancellations of the programme sessions.
Lastly, at the beginning of the interventions, some adolescents were provided with transport fare as an incentive to participate in DREAMS interventions. Many adolescents were motivated to participate as they were informed during recruitment about the incentive money. However, the organisation discontinued the transport fare due to the limited financial resources and upon realising that adolescents did not have to incur transport costs for participation in DREAMS as community-based sites were used. The facilitator explained: Some adolescents who did not receive the transport fare were not happy about it. They expressed their feelings: It was clear that the adolescents were unhappy about not receiving the transport fare as promised and regarded it as dishonesty on the part of the organisation implementing the DREAMS programme.
## Characteristics of DREAMS interventions
Some aspects of the DREAMS interventions were identified as hindering successful adolescent participation in the HIV prevention interventions. Six intervention characteristics were reported.
The first characteristic regarded DREAMS interventions that were school-based and required the participation of adolescents within selected schools. One adolescent was upset that not all schools in the community were selected for DREAMS. As a result, she could not spend time with friends whose schools were not selected in the intervention and was forced to walk home from school alone: The second characteristic that was a major barrier to adolescent participation was the exclusion of adolescent boys in the Let’s Talk intervention. The intervention was only for adolescent girls and their parents/caregivers. One facilitator stated that the exclusion of boys in the intervention influenced HIV risk for adolescent girls: The third barrier was the exclusion of adolescents with disabilities in the HIV prevention interventions. One facilitator mentioned that adolescents with disabilities were overlooked for participation in DREAMS. None of the adolescents with disabilities were reached by the CBO delivering DREAMS interventions to receive the HIV interventions. Reasons cited for the exclusion were that these adolescents lived in boarding facilities and returned home during school holidays when it was also a recess for DREAMS intervention activities: The issue of exclusion of adolescents with disabilities from the DREAMS intervention was also raised by some older carers.
The fourth barrier associated with characteristics of DREAMS interventions was the exclusion of older carers in the Let’s Talk intervention. Some older caregivers were concerned that they were excluded from participating in the intervention with their adolescent grandchildren due to age and physical abilities.
The intervention facilitator explained that they excluded older caregivers to save them from the strain of walking to the venues: Fifth, the timing of programme sessions was not convenient for some adolescents to attend the DREAMS interventions. Participants reported that DREAMS sessions that were scheduled for late afternoons conflicted with adolescents’ housework commitments.
Lastly, recruitment strategies employed by the CBO delivering DREAMS interventions hindered the participation of some adolescents in the HIV interventions. The intervention facilitator explained that the Let’s Talk intervention was community-based and used a door-to-door recruitment strategy whereby caregivers with adolescents aged 13–19 were recruited first. Often, these adolescents were in school while their caregivers were recruited. On the other hand, the Vhutshilo intervention recruited adolescents aged 10–14 directly in schools. It was explained that the Let’s Talk intervention was unable to meet its target as some adolescents were already participating in the Vhutshilo intervention.
## Youth empowerment a potential threat to intergenerational power dynamics
Participants stated that DREAMS interventions were also empowering youth and, thus, viewed as a potential threat to intergenerational power dynamics. At first, it was mentioned that some community members perceived the interventions as ‘a thing for women’ and others labelled the interventions as a campaign to teach young people to control adults: The perception that DREAMS was disrupting generational power dynamics between adults and young people hindered adolescents’ participation in DREAMS.
## Norms of sexual communication
Norms of sexual communication in the community were identified by the DREAMS intervention facilitators as affecting their ability to facilitate some sexual health topics with the adolescents as these topics clashed with community norms around sex and sexuality: It was clear that norms of sexual communication challenged the intervention facilitators when discussing sex and sexual development with young people. However, the young people did not express any problems with the intervention facilitators discussing sex, except, as stated before, for one adolescent boy.
## Discussion
In this poor and rural community hard hit by HIV, DREAMS interventions strengthened the life skills of adolescents in older carer families, particularly interpersonal relationship skills, and coping with emotions. The impact was evident both within the complex family dynamics and through redefining and recognising healthy peer relationships. This was most effective when DREAMS messaging reinforced what adolescents and older caregivers believed to be existing norms around sex and sexuality and when it was delivered by trusted community members in a familiar space. DREAMS was less acceptable when it deviated from these norms, raised SRH information that conflicts with abstinence and virginity and when youth empowerment was perceived as a potential threat to intergenerational power dynamics. While it is reassuring that DREAMS was able to engage these complex families, there were failures, in relation to factors uniquely critical to these families, such as in engaging children and carers with disabilities and failure to engage those busy with chores, particularly in the context where adolescents may themselves be caregivers of their older carers.
Family, peers, and HIV programme facilitators were instrumental in supporting the participation in HIV intervention by creating an enabling environment for adolescents to participate. These aspects are at the interpersonal level of the SEM. This is consistent with other studies showing that support from significant others has a positive impact on adolescents’ uptake of health promotion interventions [44]. The support that adolescents received from their peers and older carers motivated the adolescents to participate in DREAMS interventions. However, refusal by caregivers and the lack of information given to caregivers negatively influenced adolescents’ participation in HIV interventions. The results show that relationships with caregivers played an important role in the participation of adolescents in HIV interventions. Disapproval and lack of support from significant others such as caregivers and peers may limit access to HIV interventions by adolescents. Families are well-positioned to reinforce motivation, decision-making, and adolescent protective behaviours [45]. HIV interventions need to consider the influence of caregivers and therefore promote communication between adolescents and their caregivers. In addition, it is imperative to involve and empower older carers with knowledge about SRH, HIV, and AIDS.
As stated before, domestic work seems to influence care relationships between adolescents and their older carers. In addition, domestic work responsibilities also influenced the participation of young people in DREAMS interventions. These findings mirror those of other researchers that domestic responsibilities were barriers for adolescents in accessing HIV interventions [46]. In this study, the older carers expected young people to prioritise domestic chores over participation in HIV interventions. Refusing to let them attend DREAMS sessions was used as a punishment for not performing domestic chores. While it may be important for young people to perform their domestic responsibilities, this may give the impression that HIV interventions are less important than domestic work. A study in Uganda reported that children from older carer families were often coming late to school as a result of domestic chores [1]. These findings show that domestic tasks are a burden for the adolescents in older carer families and this needs attention as it may compromise their health and well-being.
Moreover, in this study, improved care relationships and communication between adolescents and their older caregivers as a result of the exposure to DREAMS programme components generated interest in the interventions and motivated adolescents to participate in the HIV interventions. According to study participants, DREAMS interventions provided a specific and critically important experience of improved care relationships between adolescents and their older carers. As participants described these, they reflected on connections and bonds that are critical to adolescents’ well-being. The DREAMS Let’s Talk intervention mitigated communication barriers between adolescents and their older carers. Other studies in South Africa have similarly found that a caregiver/parenting programme improved interaction between adolescents and their primary caregivers and facilitated important conversations about sensitive topics [42, 47, 48]. It seems the DREAMS Let’s Talk intervention promoted sexual health communication between adolescents and their older carers and reduced adolescents’ behavioural risks. For example, one adolescent reported delaying engaging in sexual intercourse as a result of communication with her older carer that the programme facilitated. With a clear interest in participating in DREAMS and a stated benefit by some participants for promoting sex communication between adolescents and grandparents, these findings also suggest that DREAMS acted as a platform for young people to discuss sex with adults. Findings from another study in South Africa have described this as a positive outcome whereby the intervention created space for open and meaningful conversations about sex between parents and young people [49]. However, the study raises important issues about the lack of sustainable communication between adolescents and older carers after the completion of the DREAMS programme. Previous research from a family-based HIV prevention intervention in South Africa has also indicated a decline in parent-child relationship quality post-intervention [50], which is an important protective factor for adolescent risky sexual behaviours [17, 19]. Our findings highlight that increased efforts to enhance older carer-child communication about sex post-intervention are still needed. The question of how the communication between adolescents and older carers can be sustained after interventions have come to an end is a matter of great concern.
This study found that the implementing organisation was important in creating greater access to other services in addition to HIV prevention interventions. For example, the programme facilitators assisted and linked the older carer families with other social services to obtain birth certificates and to apply for child support grants. These findings showed that offering a combination of services which adolescents need, motivated their participation in HIV prevention interventions. Also, the findings of this study highlight the utilisation of lay facilitators to deliver HIV prevention interventions. The utilisation of lay facilitators is common in low-income settings [51–53]. While the utilisation of lay facilitators promotes employment opportunities for local community members, they may lack competent skills and self-efficacy to challenge community norms of sexual communication.
Moreover, conflicting sexual health messages and information between different structures confused the adolescents. The research reveals that adolescents obtained conflicting SRH information from their churches, homes, and DREAMS organisation. The contradictory information undermined the HIV prevention initiatives, promoted by a fragmented and uncoordinated approach between different social structures in fighting HIV. This study shows that confusion about conflicting SRH information may be influenced by aspects that occur at the interpersonal, community, and organisational levels of the SEM. Also, their beliefs about whether to use contraceptives and condoms or not were shaped by sources of information they obtained from their families, churches, and HIV implementers. This had important implications on the young people’s desire to participate in HIV interventions as attitudes and beliefs play an important role in the intention to carry out a behaviour [26, 28, 29]. These findings have implications for the design of HIV prevention interventions and highlight the lessons learned in this study that reinforcement of health messages between DREAMS, churches, and families facilitated positive behaviour change in adolescents [52]. Furthermore, the quality of collaboration between information providers and adolescents is also important to bring about positive change. We recommend that integration between community structures and HIV implementers is necessary for HIV interventions to be culturally responsive and acceptable.
Further, adolescents’ attitudes towards participation in the HIV interventions were influenced by how they were being labelled by peers, family, and in the community. Being labelled as ‘badly-behaved’ and internalisation of this label created significant barriers to adolescents’ participation in HIV interventions. It also led to internalised stigma. This is consistent with previous research in South Africa which has shown self-stigmatisation to be a barrier to engaging in HIV and health services by adolescents [54, 55]. Moreover, internalised stigma compromised adolescents’ self-esteem as they avoided engaging in HIV interventions with their peers. In this setting, a study investigating awareness and uptake of DREAMS interventions [56] also found that DREAMS was not effective at reaching AGYW who had ever had sex or ever been pregnant, which may reflect internalised stigma that created barriers for AGYW to participate in DREAMS. The authors suggest that interventions that focus on the individual level of SEM can be addressed by changing the current negative beliefs and attitudes that young people have about themselves as recipients of HIV prevention interventions. A study in Ibadan Nigeria reported that the best predictors for risky sexual behaviour are low self-esteem and low monitoring practices of parents or carers while lower levels of authoritative parenting were found to be associated with risky sexual behaviours [45]. These findings add to the body of qualitative research, which has suggested that internalized stigma contributes to anti-social behaviours and mental health problems [57, 58].
At an organisational level, the exclusion of boys in the DREAMS Let’s Talk intervention was a major barrier to the participation of adolescent boys in the programme. While excluding adolescent boys in the programme may affect family communication with their caregivers, it also placed adolescent girls at increased risk of HIV infection and unwanted pregnancies. Public health researchers have long noted that adolescent girls are disproportionately affected by HIV in SSA [8]. While more interventions are now in place to reduce HIV infections among adolescent girls, it seems adolescent boys are not prioritised. These findings corroborate with those reported in South Africa that exclusion of adolescent boys and young men in HIV prevention interventions is counter-productive, inequitable, and did not play a major role in reducing HIV incidence among AGYW [29]. Being left out of certain HIV prevention interventions may have detrimental effects on the sexual behaviours of adolescent boys. Consequently, adolescent boys may have the belief that they are not at risk, therefore, are less affected by HIV and AIDS. This may further reinforce gender stereotypes that women are responsible for SRH. Earlier work has pointed out the feminization of SRH services [59] and that men view women as responsible to seek reproductive services such as contraceptives while they have little negotiating power regarding their own reproductive health [60]. Moreover, adolescent boys may be ignorant of reproductive health matters and engage in risky sexual behaviours, making them susceptible to HIV infection and other STIs. Therefore, the findings of this study highlight the need for the inclusion of boys in HIV interventions. We support the importance of engaging men as clients in SRH interventions than single-focus interventions [61].
The study findings raise the important issue of exclusion of adolescents with disabilities in DREAMS interventions, education, and social welfare services. This is important because disability is often associated with stigma [62]. The DREAMS implementing organisation in this study had difficulty recruiting and enrolling adolescents with disabilities as recipients of HIV prevention interventions. Consistent with previous reports, adolescents with disabilities in South Africa still lack access to HIV interventions and SRH services [63–66]. The exclusion of adolescents with disabilities might also imply, among other things, that the DREAMS implementing organisation did not appreciate and acknowledge young people with disabilities as beings. A systematic review of studies in SSA reported that barriers to accessing healthcare services for young people with disabilities aged 15 years and above include attitudinal biases of health and social service providers and a lack of adaptation of health information to suit young people with disabilities [67]. Accessing HIV interventions and interacting with other young people may provide comfort to young people with disabilities. Being marginalised from participation in HIV prevention interventions may have detrimental effects on the sense of self-worth of adolescents with disabilities. Consequently, adolescents with disabilities may have the belief that they are asexual beings, therefore, unaffected by HIV and AIDS. Moreover, they may engage in risky sexual behaviours as a mechanism of self-esteem validation to overcome the belief of being asexual [63], making them prone to HIV infection and other STIs as well as pregnancy. A study in South Africa reported that $11.8\%$ of adolescent girls aged 15–18 years with a reported disability were HIV positive compared with $3.3\%$ of girls with no reported disabilities [68]. These results are worrisome considering that DREAMS interventions were implemented in the same district in which this study found the marginalisation of adolescents with disabilities. The findings have implications for future research to bring to light the socio-ecological factors that shape the participation in HIV interventions for adolescents in older carer families and in the general public.
Staying in the house all the time or not attending school may reinforce the social exclusion of these young people as they may have limited opportunities for education and for interactions with their non-disabled peers. The uMkhanyakude District, where the study was conducted, is the poorest rural district of the KwaZulu-Natal province. In 2016, the Section 27 study reported that there were 14 registered schools for children and adolescents with disabilities in the district, and of these, only one was a high school [69]. Furthermore, another study in the district reported that disabled children and adolescents aged 7–18 years showed higher proportions of not attending school ($8.7\%$) compared with children and adolescents without reported disabilities ($4.1\%$) [68]. The findings of this study add depth, in that some of these adolescents with disabilities are also in older carer families which raises a need to focus on these adolescents.
In addition, economic factors may also make adolescents with disabilities vulnerable to poverty. While the monthly disability social grant provided by the South African government makes young people living with disabilities financially better off, the findings of this study showed that older carers caring for young people with disabilities had challenges in accessing social grants. Findings presented here complement those reported in South Africa that only $5.6\%$ of children and adolescents with a reported disability were recipients of disability or care dependency grants, while $21.9\%$ were not receiving any support at all, despite having a reported disability [68].
## Strengths and limitations
The major strength of this study is its ability to identify the socio-ecological factors which are most critical to participation in HIV interventions among adolescents with grandparent caregivers by obtaining the perspectives of the adolescent-grandparent dyads and the programme facilitators. Our design takes advantage of the repeat interviews that enabled discussing and following up on issues that the study participants raised in the previous interviews. Interviewing dyads separately also enhanced understanding of care relationships and dynamics between adolescents and their older carers, from the perspective of each individual member of the pair (dyad).
One important limitation of this study is that the study sample comprised of one adolescent boy. This was because adolescents were recruited through one DREAMS implementing organisation and the researchers found only one adolescent boy who was a recipient of DREAMS interventions and being cared for by a grandparent caregiver. More studies are still needed to investigate contextual issues of adolescent boys with grandparent caregivers.
## Conclusions
In this study, we investigated the adolescents’ facilitators and barriers to participation in DREAMS interventions from the perspectives of adolescents, their grandparent caregivers and DREAMS intervention facilitators in rural South Africa. We found that care relationships could be an overarching barrier and facilitator across the four levels specified in the socio-ecological framework. The importance of this study highlights that continued efforts are needed to strengthen care relationships between adolescents and their grandparent caregivers, involve grandparent caregivers, and meet the needs of adolescents. This may lead to improvement in adolescents’ participation in HIV interventions and subsequently, mitigate adolescent risky behaviours. The findings presented here demonstrate areas of particular need across the SSA and provide a call to action for service providers, donors and policymakers, to think critically about the importance of providing services to adolescents with grandparent caregivers. Adapting HIV interventions to meet the needs of adolescents with complex family backgrounds is foundational to the success of achieving the UNAIDS 95-95-95 targets.
## References
1. Rutakumwa R, Zalwango F, Richards E, Seeley J. **Exploring the care relationship between grandparents/older carers and children infected with HIV in South-Western Uganda: Implications for care for both the children and their older carers**. *Int J Environ Res Public Health* (2015.0) **12** 2120-34. DOI: 10.3390/ijerph120202120
2. 2Statistics South Africa. General Household Survey:2019 [Internet]. Pretoria: Statistics South Africa; 2020 [cited 2021 Feb 18]. Available from: http://www.statssa.gov.za/publications/P0318/P03182019.pdf. *General Household Survey:2019* (2020.0)
3. Sharer M, Cluver L, Shields J. **Mental health of youth orphaned due to AIDS in South Africa: biological and supportive links to caregivers**. *Vulnerable Child Youth Stud* (2015.0) **10** 141-52
4. Bukenya JN, Nakafeero M, Ssekamatte T, Isabirye N, Guwatudde D, Fawzi WW. **Sexual behaviours among adolescents in a rural setting in eastern Uganda: a cross‐sectional study**. *Tropical Medicine & International Health* (2020.0) **25** 81-8. PMID: 31692197
5. Hosegood V, McGrath N, Herbst K, Timæus IM. **The impact of adult mortality on household dissolution and migration in rural South Africa**. *AIDS* (2004.0) **18** 1585-90. DOI: 10.1097/01.aids.0000131360.37210.92
6. Kasedde S, Doyle AM, Seeley JA, Ross DA. **They are not always a burden: Older people and child fostering in Uganda during the HIV epidemic.**. *Soc Sci Med.* (2014.0) **113** 161-8. DOI: 10.1016/j.socscimed.2014.05.002
7. Kelley SJ, Whitley DM, Campos PE. **Behavior problems in children raised by grandmothers: The role of caregiver distress, family resources, and the home environment.**. *Child Youth Serv Rev* (2011.0) **33** 2138-45
8. 8UNAIDS. Prevention Gap Report [Internet]. Geneva: UNAIDS; 2016 [cited 2016 July 01]. Available from: https://unaids.org.br/wp-content/uploads/2016/07/2016-prevention-gap-report_en.pdf
9. Tamasane T, Head J. **The quality of material care provided by grandparents for their orphaned grandchildren in the context of HIV/AIDS and poverty: a study of Kopanong municipality, Free State Résumé.**. *Journal of Social Aspects of HIV/AIDS.* (2010.0) **7** 76-84
10. Bekker LG, Hosek S. **HIV and adolescents: focus on young key populations**. *J Int AIDS Soc* (2015.0) **18** 1-7
11. Karim QA, Baxter C, Birx D. **Prevention of HIV in adolescent girls and young women: Key to an AIDS-free generation.**. *J Acquir Immune Defic Syndr* (1988.0) **75** S17-26
12. Arnett JJ. *Adolescence and emerging adulthood: a cultural approach* (2014.0)
13. Cox C.. **Empowerment as an intervention with grandparent caregivers.**. *J Intergener Relatsh* (2008.0) **6** 465-77
14. Cluver LD, Orkin M, Gardner F, Boyes ME. **Persisting mental health problems among AIDS-orphaned children in South Africa.**. *Journal of Child Psychology and Psychiatry* (2012.0) **53** 363-70. DOI: 10.1111/j.1469-7610.2011.02459.x
15. Skovdal M, Ogutu VO, Aoro C, Campbell C. **Young carers as social actors: Coping strategies of children caring for ailing or ageing guardians in Western Kenya.**. *Soc Sci Med.* (2009.0) **69** 587-95. DOI: 10.1016/j.socscimed.2009.06.016
16. Shenderovich Y, Boyes M, Esposti MD, Casale M, Toska E, Roberts KJ. **Relationships with caregivers and mental health outcomes among adolescents living with HIV: a prospective cohort study in South Africa.**. *BMC Public Health.* (2021.0) **21**. DOI: 10.1186/s12889-020-10147-z
17. Thurman TR, Nice J, Luckett B, Visser M. **Can family-centered programing mitigate HIV risk factors among orphaned and vulnerable adolescents? Results from a pilot study in South Africa**. *AIDS Care—Psychological and Socio-Medical Aspects of AIDS/HIV* (2018.0) **30** 1135-43. DOI: 10.1080/09540121.2018.1455957
18. Arthur K, Christofides N, Nelson G. **Effectiveness of a pre-adolescent inter-generational intervention to address HIV and obesity in South Africa, using a pretest-posttest design.**. *BMC Public Health.* (2021.0) **21**. DOI: 10.1186/s12889-021-12228-z
19. Mullis MD, Kastrinos A, Wollney E, Taylor G, Bylund CL. **International barriers to parent-child communication about sexual and reproductive health topics: a qualitative systematic review.**. *Sex Educ* (2021.0) **21** 387-403
20. Soon CN, Kaida A, Nkala B, Dietrich J, Cescon A, Gray G. **Adolescent experiences of HIV and sexual health communication with parents and caregivers in Soweto, South Africa.**. *Journal of Social Aspects of HIV/AIDS* (2013.0) **10** 163-9. DOI: 10.1080/17290376.2014.902769
21. Vilanculos E, Nduna M. **“The child can remember your voice”: parent–child communication about sexuality in the South African context.**. *African Journal of AIDS Research* (2017.0) **16** 81-9. DOI: 10.2989/16085906.2017.1302486
22. Nambambi NM, Mufune P. **What is talked about when parents discuss sex with children: family based sex education in Windhoek, Namibia.**. *Afr J Reprod Health* (2011.0) **15** 120-9. PMID: 22571114
23. Nilsson B, Edin K, Kinsman J, Kahn K, Norris SA. **Obstacles to intergenerational communication in caregivers’ narratives regarding young people’s sexual and reproductive health and lifestyle in rural South Africa.**. *BMC Public Health* (2020.0) **20** 1-11. PMID: 31898494
24. Rosen JG, Kayeyi N, Chibuye M, Phiri L, Namukonda ES, Mbizvo MT. **Sexual debut and risk behaviors among orphaned and vulnerable children in Zambia: which protective deficits shape HIV risk?**. *Vulnerable Child Youth Stud.* (2021.0) 1-17
25. Doyle AM, Floyd S, Baisley K, Orindi B, Kwaro D, Mthiyane TN. **Who are the male sexual partners of adolescent girls and young women? Comparative analysis of population data in three settings prior to DREAMS roll-out.**. *PLoS One* (2018.0) **13** e0198783. DOI: 10.1371/journal.pone.0198783
26. Ngwenya N, Nkosi B, Mchunu LS, Ferguson J, Seeley J, Doyle AM. **Behavioural and socio-ecological factors that influence access and utilisation of health services by young people living in rural KwaZulu-Natal, South Africa: Implications for intervention**. *PLoS One* (2020.0) **15** 1-15. DOI: 10.1371/journal.pone.0231080
27. Jonas K, Duby Z, Maruping K, Dietrich J, Slingers N, Harries J. **Perceptions of contraception services among recipients of a combination HIV-prevention interventions for adolescent girls and young women in South Africa: A qualitative study.**. *Reprod Health.* (2020.0) **17** 1-14. PMID: 31915022
28. Zuma T, Seeley J, Mdluli S, Chimbindi N, McGrath N, Floyd S. **Young people’s experiences of sexual and reproductive health interventions in rural KwaZulu-Natal, South Africa.**. *Int J Adolesc Youth* (2020.0) **25** 1058-75. DOI: 10.1080/02673843.2020.1831558
29. Thembelihle Zuma, Seeley J, Sibiya LO, Chimbindi N. **The Changing Landscape of Diverse HIV Treatment and Prevention Interventions: Experiences and Perceptions of Adolescents and Young Adults in Rural.**. *Front Public Health.* (2019.0) **7** 1-12. PMID: 30809516
30. 30CDC. The Social-Ecological Model: A Framework for Prevention [Internet]. 2015 [cited 2020 Nov 16]. Available from: https://www.cdc.gov/violenceprevention/ overview/social-ecologicalmodel.html. *The Social-Ecological Model: A Framework for Prevention* (2015.0)
31. Yakob B, Ncama BP. **A socio-ecological perspective of access to and acceptability of HIV/AIDS treatment and care services: A qualitative case study research.**. *BMC Public Health* (2016.0) **16** 1-16. DOI: 10.1186/s12889-016-2830-6
32. Iwuji C, Chimukuche RS, Zuma T, Plazy M, Larmarange J, Orne-Gliemann J. **Test but not treat: Community members’ experiences with barriers and facilitators to universal antiretroviral therapy uptake in rural KwaZulu-Natal, South Africa**. *PLoS One* (2020.0) **15** 1-14. DOI: 10.1371/journal.pone.0239513
33. Herbst K, Law M, Geldsetzer P, Tanser F, Harling G, Bärnighausen T. **Innovations in health and demographic surveillance systems to establish the causal impacts of HIV policies.**. *Curr Opin HIV AIDS* (2015.0) **10** 483-94. DOI: 10.1097/COH.0000000000000203
34. Chimbindi N, Mthiyane N, Birdthistle I, Floyd S, McGrath N, Pillay D. **Persistently high incidence of HIV and poor service uptake in adolescent girls and young women in rural KwaZulu-Natal, South Africa prior to DREAMS.**. *PLoS One.* (2018.0) **13** 1-17. DOI: 10.1371/journal.pone.0203193
35. Francis SC, Mthiyane TN, Baisley K, Mchunu SL, Ferguson JB, Smit T. **Prevalence of sexually transmitted infections among young people in South Africa: A nested survey in a health and demographic surveillance site.**. *PLoS Med* (2018.0) **15** 1-25. DOI: 10.1371/journal.pmed.1002512
36. Saul J, Bachman G, Allen S, Toiv NF, Cooney C, Beamon T. **The DREAMS core package of interventions: A comprehensive approach to preventing HIV among adolescent girls and young women.**. *PLoS One.* (2018.0) **13** e0208167. DOI: 10.1371/journal.pone.0208167
37. Chimbindi N, Birdthistle I, Floyd S, Harling G, Mthiyane N, Zuma T. **Directed and target focused multi‐sectoral adolescent HIV prevention: Insights from implementation of the ‘DREAMS Partnership’ in rural South Africa.**. *J Int AIDS Soc.* (2020.0) **23** e25575. DOI: 10.1002/jia2.25575
38. Babchuk WA. **Fundamentals of qualitative analysis in family medicine.**. *Fam Med Community Health* (2019.0) **7** 1-10. DOI: 10.1136/fmch-2018-000040
39. Eisikovits Z, Koren C. **Approaches to and outcomes of dyadic interview analysis.**. *Qual Health Res.* (2010.0) **20** 1642-55. DOI: 10.1177/1049732310376520
40. Lincoln YS, Guba EG. *Naturalistic inquiry* (1985.0)
41. Gibbs A, Washington L, Abdelatif N, Chirwa E, Willan S, Shai N. **Stepping Stones and Creating Futures Intervention to Prevent Intimate Partner Violence Among Young People: Cluster Randomized Controlled Trial**. *Journal of Adolescent Health* (2020.0) **66** 323-35. DOI: 10.1016/j.jadohealth.2019.10.004
42. Thurman TR, Nice J, Visser M, Luckett BG. **Pathways to sexual health communication between adolescent girls and their female caregivers participating in a structured HIV prevention intervention in South Africa.**. *Soc Sci Med.* (2020.0) **260** 1-7
43. van der Heijden I, Swartz S. **‘Something for something’: The importance of talking about transactional sex with youth in South Africa using a resilience-based approach.**. *African Journal of AIDS Research* (2014.0) **13** 53-63. DOI: 10.2989/16085906.2014.886602
44. Dunbar MS, Maternowska MC, Kang MSJ, Laver SM, Mudekunye-Mahaka I, Padian NS. **Findings from SHAZ!: A Feasibility Study of a Microcredit and Life-Skills HIV Prevention Intervention to Reduce Risk Among Adolescent Female Orphans in Zimbabwe.**. *J Prev Interv Community.* (2010.0) **38** 147-61. DOI: 10.1080/10852351003640849
45. Kuo C, Atujuna M, Mathews C, Stein DJ, Hoare J, Beardslee W. **Developing family interventions for adolescent HIV prevention in South Africa**. *AIDS Care* (2016.0) **28** 106-10. DOI: 10.1080/09540121.2016.1146396
46. Mathews C, Eggers SM, De Vries PJ, Mason-Jones AJ, Townsend L, Aarø LE. **Reaching the hard to reach: longitudinal investigation of adolescents’ attendance at an after-school sexual and reproductive health programme in Western Cape, South Africa.**. *BMC Public Health* (2015.0) **15** 1-12. DOI: 10.1186/s12889-015-1963-3
47. Bhana A, Mellins CA, Petersen I, Alicea S, Myeza N, Holst H. **The VUKA family program: Piloting a family-based psychosocial intervention to promote health and mental health among HIV infected early adolescents in South Africa.**. *AIDS Care* (2014.0) **26** 1-11. DOI: 10.1080/09540121.2013.806770
48. Bogart LM, Skinner D, Thurston IB, Toefy Y, Klein DJ, Hu CH. **Let’s Talk!, A South African Worksite-Based HIV Prevention Parenting Program**. *J Adolesc Health.* (2013.0) **53** 602-8. DOI: 10.1016/j.jadohealth.2013.01.014
49. Myers L, Hajiyiannis H, Clarfelt A, Motuba-Matekane T. *Evaluation of season two of the television drama series, Intersexions* (2014.0)
50. Armistead L, Cook S, Skinner D, Toefy Y, Anthony ER, Zimmerman L. **Preliminary results from a family-based HIV prevention intervention for South African youth.**. *Health Psychology* (2014.0) **33** 668-76. DOI: 10.1037/hea0000067
51. Bhana A, McKay MM, Mellins C, Petersen I, Bell C. **Family-based HIV prevention and intervention services for youth living in poverty-affected contexts: The CHAMP model of collaborative, evidence-informed programme development.**. *J Int AIDS Soc* (2010.0) **13** 1-8. DOI: 10.1186/1758-2652-13-S2-S8
52. Rochat TJ, Mitchell J, Stein A, Mkwanazi NB, Bland RM. **The Amagugu Intervention: A Conceptual Framework for Increasing HIV Disclosure and Parent-Led Communication about Health among HIV-Infected Parents with HIV-Uninfected Primary School-Aged Children.**. *Front Public Health.* (2016.0) **4** 1-12. DOI: 10.3389/fpubh.2016.00183
53. van Rooyen H, Essack Z, Rochat T, Wight D, Knight L, Bland R. **Taking HIV Testing to Families: Designing a Family-Based Intervention to Facilitate HIV Testing, Disclosure, and Intergenerational Communication.**. *Front Public Health.* (2016.0) **4** 1-14. PMID: 26870721
54. Pantelic M, Casale M, Cluver L, Toska E, Moshabela M. **Multiple forms of discrimination and internalized stigma compromise retention in HIV care among adolescents: findings from a South African cohort**. *J Int AIDS Soc* (2020.0) **23** 1-8. DOI: 10.1002/jia2.25488
55. Nkosi B, Seeley J, Ngwenya N, Mchunu SL, Gumede D, Ferguson J. **Exploring adolescents and young people’s candidacy for utilising health services in a rural district, South Africa.**. *BMC Health Serv Res* (2019.0) **19** 1-12. PMID: 30606168
56. Gourlay A, Birdthistle I, Mthiyane NT, Orindi BO, Muuo S, Kwaro D. **Awareness and uptake of layered HIV prevention programming for young women: analysis of population-based surveys in three DREAMS settings in Kenya and South Africa.**. *BMC Public Health.* (2019.0) **19** 1417-35. DOI: 10.1186/s12889-019-7766-1
57. Mason S, Sultzman VO. **Stigma as experienced by children of HIV-positive parents: a narrative review.**. *AIDS Care—Psychological and Socio-Medical Aspects of AIDS/HIV* (2019.0) **31** 1049-60. DOI: 10.1080/09540121.2019.1573968
58. Huggett C, Birtel MD, Awenat YF, Fleming P, Wilkes S, Williams S. **A qualitative study: experiences of stigma by people with mental health problems. Psychology and Psychotherapy: Theory**. *Research and Practice* (2018.0) **91** 380-97
59. Kriel Y, Milford C, Cordero J, Suleman F, Beksinska M, Steyn P. **Male partner influence on family planning and contraceptive use: Perspectives from community members and healthcare providers in KwaZulu-Natal, South Africa.**. *Reprod Health.* (2019.0) **16** 1-15. DOI: 10.1186/s12978-019-0749-y
60. Maharaj P.. **Promoting male involvement in reproductive health.**. *Agenda* (2000.0) **16** 37-47
61. Stern E, Pascoe L, Shand T, Richmond S. **Lessons learned from engaging men in sexual and reproductive health as clients, partners and advocates of change in the Hoima district of Uganda.**. *Cult Health Sex.* (2015.0) **17** S190-205. DOI: 10.1080/13691058.2015.1027878
62. 62WHO. World report on disability [Internet]. Geneva: WHO; 2011 [cited 2011 Dec 14]. Available from: https://www.who.int/teams/noncommunicable-diseases/sensory-functions-disability-and-rehabilitation/world-report-on-disability
63. 63UNAIDS. Disability and HIV [Internet]. Geneva: UNAIDS; 2017 [cited 2017 Aug 15]. Available from: https://www.unaids.org/en/resources/documents/2017/jc2905_disability-and-hiv. *Disability and HIV* (2017.0)
64. 64UNICEF. Towards an AIDS-free generation: Promoting community-based strategies for and with children and adolescents with disabilities [Internet]. New York: UNICEF; 2012 [cited 2013 Jan]. Available from: https://data.unicef.org/resources/towards-an-aids-free-generation-children-and-aids-stocktaking-report-2013/. *Towards an AIDS-free generation: Promoting community-based strategies for and with children and adolescents with disabilities* (2012.0)
65. Chappell P.. **Secret languages of sex: disabled youth’s experiences of sexual and HIV communication with their parents/caregivers in KwaZulu-Natal, South Africa.**. *Sex Educ.* (2016.0) **16** 405-17
66. Hanass-Hancock J, Nene S, Johns R, Chappell P. **The Impact of Contextual Factors on Comprehensive Sexuality Education for Learners with Intellectual Disabilities in South Africa.**. *Sex Disabil.* (2018.0) **36** 123-40
67. Ganle JK, Baatiema L, Quansah R, Danso-Appiah A. **Barriers facing persons with disability in accessing sexual and reproductive health services in sub-Saharan Africa: A systematic review.**. *PLoS ONE. Public Library of Science* (2020.0) **15** 1-19. DOI: 10.1371/journal.pone.0238585
68. Mejia-Pailles G, Berrington A, Seeley J, McGrath N, Hosegood V. **Children and adolescents living with disability in a high HIV prevalence area in rural South Africa.**. *IUSSP.* (2017.0) 1-7
69. Hodgson TF, Khumalo S. **Too many children left behind: Exclusion in the South African Inclusive Education System With a focus on the Umkhanyakude District, KwaZulu-Natal Internet.**. *Durban: Section27* (2016.0)
|
---
title: Stakeholder perspectives around post-TB wellbeing and care in Kenya and Malawi
authors:
- Sarah Karanja
- Tumaini Malenga
- Jessie Mphande
- Stephen Bertel Squire
- Jeremiah Chakaya Muhwa
- Ewan M. Tomeny
- Laura Rosu
- Stephen Mulupi
- Tom Wingfield
- Eliya Zulu
- Jamilah Meghji
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022351
doi: 10.1371/journal.pgph.0000510
license: CC BY 4.0
---
# Stakeholder perspectives around post-TB wellbeing and care in Kenya and Malawi
## Abstract
### Background
There is growing awareness of the burden of post-TB morbidity, and its impact on the lives and livelihoods of TB affected households. However little work has been done to determine how post-TB care might be delivered in a feasible and sustainable way, within existing National TB Programmes (NTPs) and health systems, in low-resource, high TB-burden settings. In this programme of stakeholder engagement around post-TB care, we identified actors with influence and interest in TB care in Kenya and Malawi, including TB-survivors, healthcare providers, policy-makers, researchers and funders, and explored their perspectives on post-TB morbidity and care.
### Methods
Stakeholder mapping was completed to identify actors with interest and influence in TB care services in each country, informed by the study team’s local, regional and international networks. Key international TB organisations were included to provide a global perspective. In person or online one-to-one interviews were completed with purposively selected stakeholders. Snowballing was used to expand the network. Data were recorded, transcribed and translated, and a coding frame was derived. Data were coded using NVivo 12 software and were analysed using thematic content analysis. Online workshops were held with stakeholders from Kenya and Malawi to explore areas of uncertainty and validate findings.
### Results
The importance of holistic care for TB patients, which addresses both TB comorbidities and sequelae, was widely recognised by stakeholders. Key challenges to implementation include uncertainty around the burden of post-TB morbidity, leadership of post-TB services, funding constraints, staff and equipment limitations, and the need for improved integration between national TB and non-communicable disease (NCD) programmes for care provision and oversight. There is a need for local data on the burden and distribution of morbidity, evidence-informed clinical guidelines, and pilot data on models of care. Opportunities to learn from existing HIV-NCD services were emphasised.
### Discussion
This work addresses important questions about the practical implementation of post-TB services in two African countries, exploring if, how, where, and for whom these services should be provided, according to a broad range of stakeholders. We have identified strong interest in the provision of holistic care for TB patients in Kenya and Malawi, and key evidence gaps which must be addressed to inform decision making by policy makers, TB programmes, and funders around investment in post-TB services. There is a need for pilot studies of models of integrated TB care, and for cross-learning between countries and from HIV-NCD services.
## Burden of post-TB morbidity
The global burden of TB disease remains unacceptably high with an estimated 9.9-million incident cases of TB disease in 2020 [1]. However, the treatment success rate for those receiving first-line regimens was $86\%$ in 2019 [1], and there are an estimated 155-million TB-survivors alive today [2].
It is increasingly clear many of these TB-survivors experience long-term morbidity even after treatment completion, including post-TB lung disease (PTLD), with abnormal spirometry and structural lung pathology seen in over a third of those successfully treated for pulmonary tuberculosis (PTB) [3–6], socioeconomic morbidity with difficulty recovering income and employment [7], and long-term psychological morbidity related to stigma and social isolation [8]. TB-survivors have mortality rates which are over three-times higher than TB naïve adults, even after successful treatment completion, with cardiovascular disease identified as a common cause of death [9, 10]. Individuals who have had a first episode of TB disease are at increased risk of recurrent TB disease, compared to the TB naïve population, due to relapse and reinfection [11, 12]. Modelling data suggest that the disability-adjusted life years (DALYs) incurred due to TB sequalae may match or even exceed those incurred during TB disease itself [13, 14].
## Existing context of care
In response to the growing evidence for a high burden of morbidity and mortality amongst TB-survivors, there have been calls from TB-affected groups, healthcare providers, and researchers for the development of clinical guidelines and programmatic standards for post-TB patient care [15–18]. However, there remain many barriers to implementation.
Firstly, the existing paradigm of TB care remains focused on improving diagnosis, treatment and survival during TB disease itself. National and international TB guidelines and targets do not include the long-term wellbeing of TB survivors, and post-TB morbidity is not routinely recorded and not prioritised within treatment programmes or clinical trials [15]. Secondly, post-TB morbidity includes multiple dimensions such as physical, psychological, social and economic wellbeing [8]. The relationships between these parameters are unclear, and there is a lack of robust evidence for interventions which improve patient outcomes across these parameters. Existing guidelines for post-TB care are mostly clinical, and are largely rooted in expert opinion only [18]. Thirdly, even as evidence for the types of support required by TB-survivors improves, there is a lack of pilot studies or implementation based work describing how these services might be delivered in a feasible and sustainable way, within existing National TB Programmes (NTPs) and health systems. This may be particularly challenging in low-resource, high TB-burden settings where resources are stretched, and it remains unclear how, when, and to whom post-TB services might be provided.
## Aims and objectives
In response to this lack of implementation data, we completed an 18-month programme of stakeholder engagement around post-TB care in Kenya and Malawi. The aim of this work was to inform the development of strategies for post-TB care within the region. The objectives were to identify and connect stakeholders in TB service delivery in Kenya and Malawi, to raise awareness of post-TB wellbeing amongst these stakeholders, to understand the existing context of care, and to explore beliefs and perspectives around post-TB morbidity and care. Findings of this stakeholder engagement work are presented here.
## Methods
This study was run as a partnership between The Liverpool School of Tropical Medicine (LSTM) and The African Institute for Development Policy (AFIDEP), in collaboration with the National Tuberculosis, Leprosy and Lung Disease Programme (NTLP) in Kenya, and the National Tuberculosis Control Programme (NTP) in Malawi (Feb 2020 to July 2021).
## Stakeholder mapping
An initial process of stakeholder mapping was completed at the start of the programme, to identify individuals with interest and influence in post-TB wellbeing in Kenya and Malawi. This was informed by the local, regional and international networks of the study team from LSTM and AFIDEP. In-country stakeholders included policy makers, parliamentarians, funders, researchers, health care workers, TB advocates and TB patients and survivors. Key international TB organisations were included to provide a global perspective. A snowballing approach was then used to identify further stakeholders, over the course of the study. The number of interviews completed was determined by the number of key stakeholders identified, and we did not specify a requirement for data saturation.
## Development of data collection tools
Three data collection tools were used to guide semi-structured interviews with different stakeholder groups (Appendix A in S1 File: TB survivors and patient advocates; Local & regional stakeholders; International stakeholders). These tools explored existing practices and perceived need for post-TB care, key evidence gaps, perceived barriers to implementation, and potential structure, content, and leadership of post-TB services. These topics were identified a priori by the study team as relevant to the post-TB agenda, based on their own experience and published literature, with iterative review over the course of data collection. Post-TB sequelae were broadly conceptualised as the multifactorial challenges faced by TB survivors, including physical, psychological, social and economic morbidity, and risk of recurrent TB disease.
## Stakeholder interviews
Key stakeholders identified in the mapping exercise were invited to participate in individual interviews. Interviews were conducted on a one-to-one basis, through phone calls, online or in-person meetings, in keeping with national COVID-19 guidelines and were conducted by researchers from AFIDEP (SK, TM) with previous experience of qualitative research and community engagement. Interviews were completed between October 2020 and March 2021. Participants were provided with a paper or electronic information leaflet about the study, and informed consent was taken either verbally with recording, or in writing. Interviews were conducted in English, Swahili or Chichewa at participant preference, and were audio-recorded and transcribed verbatim.
## Data coding and analysis
Interview transcripts were read by study authors (SK, TM, JMp, JMe). A coding framework was derived (Appendix B in S1 File), and the data were coded using Nvivo 12 software (SK, TM). Data were analysed using thematic content analysis. Coded data were reviewed and discussed regularly in order to identify key concepts, emerging themes, and determine data findings. Notes kept by team members, and records of team meetings were used to inform this analysis‥ Findings were discussed with the broader team at regular intervals to clarify these themes and findings, identify areas of uncertainty, and to inform ongoing stakeholder engagements.
## Data validation and exploration
Once key themes had been identified, virtual workshops were held in Kenya and Malawi, with a broad range of stakeholders invited, whether interviewed or not. International stakeholders were invited to attend the Kenyan workshop. Workshops were video recorded with verbal consent from all participants, and included break-out sessions to explore key themes emerging from interview data (Appendix B in S1 File). The recordings were shared with all participants with their consent. Workshops were not transcribed, but notes were taken by the study team during each session, and used to inform and advance our understanding of the data generated from individual interviews.
## Ethical approval
Formal ethics applications were submitted in Kenya, Malawi and to LSTM. Ethical approval was obtained from LSTM (20–064) for work in both countries and KEMRI for work in Kenya (KEMRI/RES/$\frac{7}{3}$/1). Ethical approval was waived by the National Health Sciences Research Committee in Malawi as the study focused on service design and development.
For confidentiality reasons, the term ‘Ministry of Health’ or MoH is used in this manuscript to describe quotes obtained from NTP or NCD Programme members from Kenya and Malawi.
## Results
In this section we describe the stakeholders who participated in this study, and data collected on the existing context and perceived need for post-TB care. We then describe key themes which emerged on the structure and delivery of post-TB services, and potential barriers to implementation.
## Stakeholder engagements
Forty-seven key multisectoral stakeholders were identified through stakeholder mapping, with 38 interviews completed with TB-survivors, healthcare workers, funders, policy-makers and researchers from Kenya, Malawi and relevant international organisations (Table 1). Nine of the stakeholders we invited to participate in the study did not respond to our invitation. The majority of patient advocates interviewed were themselves TB-survivors.
**Table 1**
| Stakeholder group | Malawi (n = 12) | Kenya (n = 20) | International (n = 6) |
| --- | --- | --- | --- |
| National TB Programme | 2 | 3 | |
| National NCD Programme | 1 | 1 | |
| Other government ministries | 0 | 3 | |
| Healthcare provider | 4 | 6 | |
| TB-survivor | 1 | 1 | |
| TB patient advocacy group | 2 | 4 | |
| In-country researcher | 1 | 0 | |
| In-country non-governmental organisation (NGO) | 1 | 2 | |
| Multi-lateral Organisation (Funding) | 0 | 0 | 2* |
| Multi-lateral Organisation (TB policy) | 0 | 0 | 4† |
A total of 77 stakeholders were invited to the workshops: $\frac{34}{43}$ invited stakeholders attended the Malawi workshop and $\frac{25}{34}$ invited stakeholders attended the Kenya workshop, giving a total of 59 participants. Workshops lasted three hours long for each country.
## Reported context of care
Neither Kenya nor Malawi have active programmes for post-TB care. However, a technical working group for PTLD was established within the Kenyan NTLP in 2019, to “coordinate, operationalize, and entrench” PTLD activities. The next version of the integrated NTLP TB treatment guidelines is expected to include a chapter focused on post-TB lung health. Potential reasons given for the focus on this agenda include that two Kenyan NTLP delegates attended an international post-TB symposium in South Africa in 2019 [8], and that a recent review of the chest X-rays captured in the Kenyan TB Prevalence survey in 2015–16 demonstrated a high prevalence of post-TB lung pathology [19]. However, it is not clear whether these efforts were the cause of, or a response to, the growing interest in this field. Prior to this study, there was no clear working group for post-TB health within either the national TB or NCD programmes in Malawi, and no mention of post-TB care within national guidelines. However, interest in developing a comparable Malawian post-TB working group was expressed in the Malawian workshop held at the end of this programme of work.
## Perceived need for post-TB services (Table 2)
**Table 2**
| Theme | Sub-theme | Quotes |
| --- | --- | --- |
| Not a priority at present | Need to focus on public health | “There are more pressing issues. It is not that this is not important and I am sure it is very important to an individual that goes through this but in the greater scheme of things and in a resource-constrained environment you are going to choose the things that are most common that affect more people. That is the public health perspective. So unfortunately that is the harsh reality.” (MoH staff, Malawi) |
| Not a priority at present | An external agenda | “This is not something that is discussed all the time and it has become important because someone somewhere has decided it is important.” (MoH staff, Malawi) |
| An important agenda | Need to inform patients about their health | “I feel it is very necessary because most people after completing their medication they actually do not know what next. I feel it would be very important so that they have some guidance on what else after the medication, what else after you get cured” (TB patient advocate, Kenya)“Let me speak it from my own point of view. Like for me, I would have found it more helpful if there were already systems in place because I would not have to feel like I am disturbing my surgeon with all those questions. I would have been more comfortable.” (TB-survivor, Kenya)“We do have a lot of patients. TB is quite a burden in Malawi so yes, it means every year we are discharging people from treatment so a lot of post-TB care candidates are coming out of our system each year and those need support and some level of attention to maintain their health in good check. So we do need to have that agenda.” (Researcher, Malawi)“So because on the cut-off date when they do the last tests and declare you free of TB, that is the end of the visitation, the end of going to the clinic, the end of everything. So they have left you with all the problems that you have been talking to them without getting any responses. So for me, I feel that it would have been important to still continue for some time so that they continue to monitor you and they come to a conclusion where you are also satisfied to say I have finished my treatment.” (TB Survivor, Malawi) |
| An important agenda | Need for standardised patient pathways | “I think it is something that we should be advancing because patients are there and we need to have a unified approach to their care because now I believe they are being seen by different doctors everywhere who may make different diagnoses and manage them differently.” (Healthcare provider, researcher and lecturer, Kenya)“…we discussed with my colleagues within the department and at one point or another we refer these patients to the physicians who follow them up. From there we lose track of the patient. We do not have that mechanism of knowing how they have picked up from the consultant level.” (Healthcare Provider, Kenya)“That is why what we have done, we are liaising with our colleagues to say look, instead of saying NCD clinic, ART clinic, TB clinic, why don’t we just say a chronic disease clinic so that when a patient comes in, it is either we are giving them the TB drugs, they are getting ARTs, they are also getting the NCD drugs. In so doing, we will be able to manage these patients under one roof. So that is the integration that we want to have.” (MoH staff, Malawi)“As I said, if you look at TB, HIV and NCDs, I think this just needs to be one family because HIV patients they have TB, they develop NCDs so it is something that yes the three of us we have to be related. So yes, I would say it is good. It is something that needs to be done.” (MoH staff, Malawi) |
| An important agenda | Opportunity to maximise the investment made during TB treatment | “Oh yes. It does not make sense that you invest millions in a person, you cure them and then after that they do not lead normal lives. I mean then what you have invested becomes a waste” (TB Patient advocate, Kenya)“I have observed that within one or two years the close contact of the same patients are coming to the hospital with the TB. So if we have a very well structured way of following up the post-TB cases, we can be able to pick the contacts early enough even without much of interruption or without them spreading to other people. So a timely agenda that if possible it needs to be in place”. (Non-governmental Organisation, Kenya) |
| An important agenda | Means of reducing ongoing health seeking | “…I think [post-TB morbidity] is really rendering these people unproductive because they will spend a lot of time maybe seeking care, they will get all sorts of diagnosis and maybe put on all sorts of antibiotics or other medication and it is going to be expensive for them. It is going to drain their resources in time. From the healthcare system, I think it is also similar. It is a waste of resources because we are not giving these patients the care that they need maybe right from the time that they completed their TB treatment” (Multilateral Organisation–TB policy) |
Post-TB morbidity was acknowledged by all participants as a real phenomenon experienced by TB-survivors. However, opinions about whether post-TB care should be prioritised within existing health care services were mixed (Table 2).
Management of post-TB morbidity was often framed as a form of individual care, and placed in opposition to public health interventions which operate at the population level and prevent transmission, and which must be prioritised within TB programmes. This perspective was seen amongst policy-makers and healthcare providers, and was related to beliefs about the role of national TB services. One researcher and two policy-makers felt that the focus on post-TB wellbeing was an external agenda, imposed by global researchers and policy-makers on countries, rather than being generated in country by National programmes.
In contrast, there were others who considered this a timely agenda, important to the health and wellbeing of TB-survivors. This was particularly the case amongst patient advocates who felt that post-TB morbidity should be addressed in order to allow patients to understand their own health, maximise their long-term wellbeing, and improve access to and ease of care.
Healthcare providers spoke of the need to develop standardised pathways for post-TB care, highlighting challenges around unclear patient pathways, heterogeneity of current practice, and the pressure on individual providers to make clinical decisions in the absence of formal guidelines.
Several participants felt that investing in recovery after TB disease would build on the investment already made in TB treatment, minimise ongoing health seeking by TB-survivors, reduce misdiagnoses of TB sequelae, and provide opportunities for contact screening and further counselling.
## Structure and delivery of post-TB services (Table 3)
**Table 3**
| Theme | Sub-theme | Quotes |
| --- | --- | --- |
| Content of post-TB services | Need for holistic care | “TB has very catastrophic effects and these patients end up being affected not just physically but also mentally. It has also a way of disabling their whole quality of life in totality…so I believe it is something we cannot ignore and it is something that we need to ensure that if we are talking about improving the quality of life of this particular TB patient, then we need to look at them holistically and ensure that they are being followed up beyond their period of treatment” (MoH staff, Kenya)“The consequences of post-TB complications is quite strenuous to the patient…because financially, emotionally, the patient is affected and even the normal health being of the patient is really affected. Nutrition, all those are affected. So if the donors are to be involved or the policy-makers, they should include all these people, nutritional, psychological counsellors such people that can help the patient holistically”. (Healthcare Provider, Kenya) |
| Content of post-TB services | Importance of broader management of TB-comorbidities | “And again we know that not only does TB cause obviously lung damage and longer term problems, but we also know that there are many other health conditions that contribute to a greater risk of someone developing TB, and those are often not treated and continue to pose ongoing risks for people. Whether it be things like smoking…or COPD….you know drinking alcohol” (Multilateral Organisation–TB policy) |
| Content of post-TB services | Importance of psychosocial support, and re-integration | “We forget that psychosocial support is supposed to be like a long term. We focus on it when somebody is on TB medication but medication is not only tablets…it is ensuring that we have proper counsellors or proper professionals that can provide psychosocial support even after treatment to ensure that somebody is well transitioned to the community”. (TB Patient advocate, Kenya)“…then the counselling for me is very very important and even training community counsellors as we have done for HIV. It would go a long way in supporting the TB-survivors.” (TB Survivor, Malawi)“It is a community issue, it is a social issue whether someone has lost their jobs, it is a social issue. Then you see it is also affecting many other aspects of development in terms of housing, in terms of education, in terms of you know livelihoods….[there is a need] to support those people who have had TB in terms of reintegration into the communities because it is not always physical. The impact is not always physical.” (TB Patient advocate, Malawi) |
| Content of post-TB services | Need for economic support | “… when you are transitioning from treatment, not everyone might need money but just to ensure that you have something like IGA–income generating activity” (TB patient advocate, Kenya) “…in terms of work, because some people cannot really do the work they used to do after TB. Maybe making sure people are getting vocational skills or maybe starting smaller businesses.” (TB Community Advocate, Malawi)“So we need to empower them. Economically, we have heard of cash transfer. I think those should be for the beneficiaries so that they can restart their life afresh.” (Healthcare provider, Malawi) |
| Location of post-TB services | Need for decentralisation, with a hierarchy of healthcare services | “The care should also be then provided in an affordable and accessible way closest to the patient. So perhaps the person with a mild presentation, we should be able to attend to them at the primary and even secondary level of care. So that only the persons who have very very severe (inaudible) with this condition would end up in the national referral or tertiary care or hospitals. That is how I would imagine it.” (Multilateral Organisation–TB Policy)“The trend in Malawi, the flow of patients, they will start from the primary healthcare which is the health centre, then secondary health facility like the district hospitals. If they fail there, they will go to the tertiary which is for the TB… Let them follow the same route because for most of the people, going to the tertiary it is expensive.” (Healthcare provider, Malawi) |
| Location of post-TB services | Need for community based care | “They need support and that is where we need civil society, community based organisations or patient support groups who can be linked to those people who are exiting the system that they are able to be integrated and supported to be reintegrated into the community.” (TB Patient advocate, Malawi) |
| Location of post-TB services | Importance of learning from HIV-NCD services | “….so initially you would start with high level cadres and then you need to learn as much as possible and develop detailed training, manuals and guidelines that are easy to follow. The HIV programme should be a good resource for whoever is working on this to look at how these rather complicated tasks were successfully handed over to lower cadres and how the services got as close to the patient as possible…and I would say start conservatively but do not take too long to decentralise.” (In country Researcher, Malawi)“In HIV first of all they all used to come in the tertiary hospitals because nobody knew how to handle it [HIV/AIDS]. But over time we developed our HIV systems, we developed systems protocols even. Now a patient does not need to walk very far. They can go to their nearest health centre and the nurse there has got very clear algorithms and she can manage to some level… So I think we have experience, we can learn lessons from the HIV programme and do it for lung health” (Healthcare Provider, Kenya) |
| Timing of post-TB services | Proactive care, from TB treatment completion | “It should be integrated from the beginning. Let us not wait for them to finish and then come back. In fact, I think if we would start treating this patient with that in mind that there is likely to be complications, there is even a way that we can be able to mitigate those complications. …” (TB Patient advocate, Kenya)“I think we need to be proactive but we should be proactive when we know what we will offer to these individuals. So screening or looking for individuals with the condition is not ethical if you are not able to provide care for them.” (Multilateral Organisation—TB Policy) |
| Timing of post-TB services | Reactive care, responding to patient need | “The onus is on the patient that you feel a certain type of symptoms or whatever, and you will, the expectation is that you will have a health seeking behaviour to go to the hospital and say look I had TB and now I have noticed.” (Policy-maker, Malawi) |
| Delivery of post-TB services | Integration of services, for delivery | “We need to integrate with other services. The HIV services, the TB services, the Covid-19 services that are there…I would love if those activities can be integrated with other services that are already existing in that treatment level and prevention level and at community level. That could not make the work difficult to implement.” (MoH staff, Malawi)“TB programmes in most countries are not a vertical programme. Maybe [at the] national level, [or for] provincial level teams supervising the programmes and so on. But the service delivery level is integrated. I think that is what we need to promote…. I do not think we need to go for post-TB as another vertical approach. It should be integrated. It would make sense because we are talking about TB affecting other different organs and it is beyond TB now—it is post-TB. And that is where it makes sense to think about integration because the same person will have other problems. It could be HIV as well, and now we have COVID and post-COVID.” (Multilateral Organisation–TB Policy)“But I think there has to be a lot more emphasis about …the rest of the system to play its role. I think you will find that Ministries of Health loathe to verticalise things so much. So the thinking as I understand it is to allow the general health system to be able to deal with a wide swathe of problems rather than have specialised units or entities.” (Multilateral Organisation–TB policy) |
| Ownership / leadership of post-TB services | Leadership by the NTP, with support from others | “Who is responsible for it in Malawi, it would obviously be the Malawian health authorities, the NTP programme as far as I can see.” (Multilateral Organisation–TB policy)“I think more than ever, TB is slowly emerging to be a cross-cutting issue…so it is not something or an area that you would want to leave to TB program to solely address but they would obviously take lead…we need other players to come and support this…”(Non-governmental Organisation, Kenya) |
| Ownership / leadership of post-TB services | Challenges of NTP ownership | “I personally do not think TB programmes should be responsible for what happens after. We are a stakeholder but we are not the ones best equipped to deal with you if you are having difficulties breathing. That is not us. The people best equipped to do that would be clinical services and everything under clinical services.” (Policy-maker, Malawi)“Remember it is a National TB control programme so their priority is to control TB but not necessarily to rehabilitate. So it has to be carefully crafted to make sure that it does not look like an add-on to the TB programme but should be part of the TB programme.” (Multilateral Organisation–Funder)“The highlight on post-TB health I think is not new per se, but the emphasis is something that is new to a TB programme. So from a TB programme perspective it is simply designed to deal with the episode of the disease not what happens after” (Policy-maker, Malawi)“….they (NTLP) remain very lean and it will only be restricted to TB and leprosy as it is. So how do they even start to discuss that agenda of post-TB? Are they the right people to discuss that agenda or should they now move this discussion to the non-communicable disease side? So for me, to avoid a lot of back and forth, the easiest way is a lung health programme” (Healthcare Provider, Kenya)“It would be great if [ownership] was the TB programme. I cannot see that it would be responsible…there are so many different post-TB health issues that you could have… how does that work from an implementation standpoint?” (Multilateral Organisation–TB Policy) |
| Ownership / leadership of post-TB services | Challenges of situating within the health system | “For me it is outside the medical approach. So if we bring in another aspect that is outside the medical approach into the medical system then it will not be sustainable, it will not run. ….we need a structure that can be sustainable, that can make sure that they are carrying oversight and making sure that things are being done in the right order.” (TB Patient advocate, Malawi) |
| Framing of post-TB care, in discussion with patients | Need for careful communication | “…we are saying that TB is curable…So now if you say that after your medication there are complications that you can face, people now will be put off, to say what is going on? So I think it is a matter of in a way having clever programmatic and strategic programming because otherwise you might defeat the purpose …People still get cured of TB, but also we want to make sure that we are being realistic enough to say yes the medication might have side effects after but there is help…I think in a way it is a bit overwhelming because it feels like you are going to be a patient for the rest of your life.” (TB Patient advocate, Malawi) |
| Framing of post-TB care, in discussion with patients | Presentation as part of the TB care cascade | “I think if it were to be like a continuation, people would take it more serious rather than when you are told now you have finished this treatment there is another phase because most people would be like why do I need it yet I feel I am okay. I think if it were to be one joint thing, it would work better” (TB Patient advocate, Kenya) |
| Framing of post-TB care, in discussion with patients | Presentation as an optional service | “It should not be mandatory—‘Well you have finished TB we have got this package for you’…some others do not want to feel victimised and feel like they are victims of a certain thing.”(TB Patient advocate, Malawi) |
## Content of care
Although this study was initially framed around post-TB lung health, diverse stakeholders in both countries emphasized the need for multi-disciplinary post-TB services. This was a strong emerging theme. Dimensions of support identified included: health education, counselling and psychological support, nutritional support, social support, vocational training or financial support, and respiratory and physiotherapy services. Of note, discussions with international policy-makers were framed around the need to address broad TB comorbidities rather than just post-TB morbidity, including diabetes, HIV co-infection, smoking, alcohol use disorders, and chronic respiratory diseases such as chronic obstructive pulmonary disease (COPD).
Patient advocates expressed a strong feeling that post-TB morbidity should not be considered to be a purely medical problem, and that support should be provided in the community. Counselling and psychological support were perceived as a critical component, required both during and after TB treatment completion, given the long-term nature of TB related stigma, loss of livelihood, and residual post-TB physical morbidity. It was suggested that psychological support could be provided through individual or group counselling and support groups where people meet and share their experiences. The concept of TB patients being supported to ‘transition’ back into the community or ‘reintegrate’ after treatment completion was widely raised.
It was noted that some TB-survivors may no longer be able to perform activities that they previously performed with ease, and that this might call for a change of profession. There was therefore a suggestion to provide such people with vocational skills training, and the potential for social protection in the form of cash transfers was also raised. Several participants suggested that government departments involved in the provision of social protection should be involved in discussions around holistic TB care.
## Location of services
TB-survivors and advocates noted that post-TB services should be decentralised and provided to participants at the nearest health facility to ensure ease of access, and to allow for continuity of care after TB treatment completion. Several participants suggested that a hierarchy of support may be needed, with basic care provided at the primary care level, but more complex cases referred for specialist care. Several participants suggested that services should initially be developed and refined at the tertiary care level, and then decentralised. Opportunities to learn from the approach taken by HIV programmes for service development were often cited:
## Timing of services
The majority of participants felt that post-TB services should be proactive in providing care–that is, aimed at identifying patients with morbidity at TB treatment completion, with ongoing support provided as part of the TB care cascade. This is the approach which has been outlined in the Kenyan NTLP guidelines. Early intervention was felt by community advocates to be important for preventing complications, optimising delivery of social support, and improving the uptake of services.
However, this was not a universal perspective. Concerns were raised about the merits of adopting this proactive approach in the absence of evidence-informed clinical interventions which have been shown to improve patient outcomes. Some felt that the system should remain reactive, with responsibility for health-seeking if and when complications develop, remaining with TB-survivors:
## Delivery of services
There was consensus that although post-TB services may be vertically led at the national and provincial level, the broad range of support required by TB-survivors will require an integrated approach at the service delivery level, with inclusion of TB and broader health and community services. Given the substantial proportion of TB patients with HIV co-infection in Kenya and Malawi, the need to integrate with HIV services both during and after TB treatment completion was highlighted, and opportunities to access long-term integrated care through HIV services were raised.
Conversations around who should deliver services were closely linked to discussions around responsibility and ownership of post-TB care services.
## Ownership and leadership of services
There was some disagreement about the ownership and management of post-TB services at the national level. In Kenya, the post-TB agenda already sits within the NTLP. In Malawi, this had not been decided.
The majority of national and international participants felt that post-TB care should fall under the remit of NTPs to ensure continuity and consistency in service provision throughout the TB care cascade, and to build on existing resources and monitoring mechanisms. However, the need for broader integration with other programmes at the national level such as the division of NCDs, Ministry of Labour, Ministry of Gender, Social Security services, and implementing partners for service delivery was recognised. The need to position post-TB care as a core element of overall TB service delivery with this approach was highlighted.
However, some caution about this approach was expressed by TB Programmes, healthcare providers, and patient advocacy groups. From the NTP perspective, a focus on post-TB wellbeing was felt to be outside of their remit, which historically has focused on the diagnosis and treatment of TB disease, with input ending at TB treatment completion. There was also a sense that the TB programme is not equipped to deal with the implementation of care for broader comorbidities on the ground, and that responsibility for post-TB service provision should therefore lie elsewhere. TB-survivors and patient advocates expressed some concern that positioning this agenda under the NTP would lead to a medical rather than a holistic approach to care.
## Framing of post-TB care
Patient advocates were keen that messaging on post-TB complications should be delivered carefully, emphasizing that TB is a curable disease, with a minority of TB patients developing residual morbidity, with services available for ongoing support. The need to ensure that TB-survivors are not unnecessarily defined as ‘patients’ or ‘victims’ of TB in the long-term was stressed. It was also felt that post-TB care should be presented to TB-survivors as optional, rather than a compulsory part of the TB care cascade: Barriers to implementation (Table 4). Key barriers identified included the lack of local data on the burden of morbidity, limited funding, limited staff time, lack of equipment, and absence of clinical guidelines and models of care.
**Table 4**
| Theme | Sub-theme | Quotes |
| --- | --- | --- |
| Lack of data | Local data on burden of disease | “That is the first thing we need to know—what is the magnitude of this, before we make a big deal of this. And I do not want to sound careless. I want you to understand this from priorities upon priorities for the system. Someone always gets left behind because you cannot fund everything. So you need to know first, how big a problem is this?” (Policy-maker, Malawi) |
| Lack of data | Data on risk factors for morbidity | “But again, we need to understand the disease better and how it manifests itself. Is it the same here as in India, as in China or is it different depending whether you are a man or a woman, HIV positive or negative, on ART or not on ART, viral load suppressed or viral load not suppressed. Does it depend on your age? What does it depend on? Does it depend on the type of TB you had? If you had Xpert positive TB compared to perhaps clinically diagnosed TB or pleural effusion. I do not know if we have the answers to those questions so we cannot design the interventions until we know what we need to improve.” (Multilateral Organisation–TB Policy) |
| Funding constraints | Need for funding sources from outside the TB programme | “If you are saying well now countries should include longer term post-TB care in their national TB response … it will just probably receive a lot of pushback to say we already stretched enough…The win would need to be having this idea, this concept, bringing in other areas of funding to support it. I think that would make it much more palatable” (Multilateral Organisation–TB Policy) |
| Funding constraints | Need for donor funding | ’Even drugs, drugs are being procured by the Global Fund, PEPFAR and World Bank and GDF of course which means if there is any support which is post-TB care… in my perspective it will definitely be donor driven and donor funded.” (TB Patient advocate, Malawi) |
| Funding constraints | Domestic funding options | “They need probably long term or lifelong support and you cannot expect that from a project or a grant with a shorter period of time. That is why a sustainable financing system should be in place and that is why this group should be prioritised as part of the overall health system and part of the domestic funding.” (Multilateral Organisation–Funder)“One is that if we make it (post-TB care) nationwide and we have an NHIF cover, then we can say that with NHIF you can go to the nearest facility that is near you. So people have the freedom… you can access that service in Nairobi so long as you have a code which is computerised. So with that code you are known that you are a regular recipient of this service so you can go to any facility near you to get that service.” (TB Patient advocate, Kenya) |
| Funding constraints | Need for mandate for care at the National/Global level, in order to secure funding | “So there should be a guidance from you know from WHO and international organizations and countries should also include this [post-TB] group in their national strategic pan and then they can still include some of the support to come from Global Fund.” (Multilateral Organisation—Funder)’I think if it is in line with the WHO guidance I think it is something that we can always look positively but above all I think it has to be a national priority. If the NTP does not look at it as a priority I think it is unlikely that we are going to fund it. For example, the Global Fund this year allowed countries to develop their funding requests based on the National Strategic Plan (NSP). So if it is not in the NSP it is hard to fund it.” (Multilateral Organisation–TB policy) |
| Staffing constraints | Lack of trained staff | “The TB programme at the district level or facility level is not run by medically trained people. It is run by environmental health staff …while post-TB care would require doctors, nurses, clinical officers to do the work and there is not much interest especially from clinical officers and doctors to do TB work at facility level…” (Multilateral Organisation–Funder)“In so many of the countries that we work in there is huge lack of clinicians of trained [staff], whether it be very specialised services like radiologists or other clinical staff, even coming down to nurses. There is a huge gap in human resource capacity especially as we move out of the cities.” (Multilateral Organisation–Funder) |
| Staffing constraints | Need for staff training | “What needs to be done is to train staff and let them know that these patients can have lung complications. I used to see for example some years back and this happens a lot at lower level health facilities when patients with TB who have completed treatment come back with symptoms. Often they have chest pains, they have fever, and they are put back on treatment again. Then they are treated fully again for another 6 months, 8 months, some even up to one year because their symptoms are not resolving. And this is simply because people have not done tests to find out what else could it be if this is not TB…” (Healthcare provider, Kenya) |
| Clinical guidelines | Need to define key clinical interventions which should be used | “Are there other health services, are there medicines, are there procedures that can be offered to them that would make their life better and respiratory function better? So would it be physiotherapy, some sort of pulmonary rehabilitation, education about these airway clearance exercises, regular immunizations against respiratory pathogens, definitely it will include smoking…” (Multilateral Organisation–Funder) |
| Clinical guidelines | Need to define key patient outcomes for monitoring & evaluation | “But having something around longer term health outcomes for people, there would have to be something that is achievable, measurable, something that could be impacted to I think to get funding and donors move around that cause.” (Multilateral Organisation–Funder) |
| Models of care | Lack of existing models of care | “Probably the second reason is that we have not seen really much happening in other countries either so if somebody can bring up a model that has worked elsewhere, I guess we should have an open mind to have a look at it and see how we can adapt that to our setup.” (Multilateral Organisation—Funder) |
| Potential impact of COVID-19 | Greater need for integrated care | “It makes sense to think about integration because the same person will have other problems. It could be HIV as well. Now we have COVID and post-COVID and all those.” (Multilateral Organisation–Funder) |
| Potential impact of COVID-19 | Limited capacity for change | “I think our world is now going through these pandemic times and how exactly we will come out of this also financially and economically will determine at least in short and medium term how the world and how all of us can take up new issues or newer issues. Because you know every organisation, every human being has some sort of saturation point I think.”(Multilateral Organisation–TB Policy) |
## Limited data
Athough many study participants felt that post-TB morbidity was a common problem, there was a broad sense that more in-country data are needed to describe the burden of this morbidity, and risk factors for disease, in order to inform decision making around investment by NTPs and international funders.
## Funding constraints
The lack of secure funding for post-TB care was highlighted as a key barrier to care. Given the heavy reliance of Kenya and Malawi on donor funding for TB service delivery, many participants felt that donor funding would be needed to support post-TB services. There was some discussion of whether this should be requested within the TB envelope, or accessed through non-TB funding streams. The main funder for TB services in Kenya and *Malawi is* The Global Fund, but neither country has included post-TB care in their funding requests submitted to The Global Fund to date. There were a range of perspectives on whether The Global Fund would be responsive to such requests from individual countries, or whether guidelines from external bodies such as the World Health Organization would first be required.
Other participants suggested that post-TB care should be funded by the government to ensure sustainability, with a broad range of options for domestic funding raised, including health insurance programmes such as the National Health Insurance Framework (NHIF) in Kenya, disaster relief schemes, social protection programmes, and pension schemes.
## Staff and equipment constraints
Concern was raised about the human resources needed to deliver post-TB support, particularly using a decentralised approach and in rural areas. Although use of spirometry and chest X-ray are suggested at TB treatment completion within the Kenyan NTLP guidelines, it is noted that access to this equipment in the majority of healthcare facilities is limited. The need for capacity building for post-TB care was expressed.
## Need for clinical guidelines and models of care
The lack of robust clinical guidelines for the diagnosis and management of post-TB complications, lack of existing models of care, and the need for clear patient outcomes which could be used for monitoring and evaluation of interventions were identified as key barriers to implementation.
Although the impact of the COVID-19 pandemic was not explicitly explored in this study, this was raised as a factor which might either expedite the introduction of more integrated models of respiratory care, or act as a barrier to further change.
## Discussion
Our understanding of the burden and nature of post-TB morbidity is growing, and clinical guidelines for post-TB care are under development [18, 20]. However, little work has been done to explore how integrated post-TB services might be delivered by health systems on the ground, particularly in resource-constrained settings. In this study we explored stakeholder perspectives on post-TB care in Kenya and Malawi, including discussion with patient advocates, healthcare providers, policy-makers, national TB and NCD programmes, and funders. Our work has highlighted key challenges including the need for broad multidisciplinary services, uncertainty about governance and leadership, lack of local data on the burden of morbidity, lack of funding, and concerns about staff and equipment capacity. However, the study has also demonstrated widespread interest in this agenda, identified key research opportunities, and highlighted opportunities to learn from HIV-services.
There was broad consensus amongst stakeholders that TB-patients and TB-survivors require holistic care, which addresses the physical, social, psychological, and economic impacts of TB disease. Several international stakeholders emphasised the need to address TB-comorbidities such as smoking, diabetes, and alcohol use disorders alongside post-TB morbidity, to support long-term patient wellbeing, in keeping with existing international guidelines [21, 22]. The importance of addressing TB multimorbidity has been highlighted elsewhere [23–25]. While our initial focus in this study was the management of post-TB lung disease, it is clear that a larger conversation around the delivery of integrated patient-centred care across the TB cascade–including both comorbidities and post-TB sequalae–is needed.
However, there is considerable uncertainty over how this broad holistic care should be managed and delivered. Many stakeholders felt that post-TB care should fall under the remit of NTPs. This would allow continuity of care across the TB care cascade, and would leverage existing systems for the delivery and monitoring of services. Indeed, several pilot programmes of integrated patient care for TB comorbidities such as diabetes and smoking have employed this approach, building off NTP frameworks [26]. However, there was some hesitancy from NTPs to take on this responsibility, alongside their role in the diagnosis and treatment of active TB disease. This may relate to concern about funding, staff time, and expertise, but is also rooted in the belief that the management of comorbidities falls outside of the traditional remit of TB services. If we are to use an NTP-centred approach to post-TB care, the management of comorbidities and sequelae will need to be framed as an integral part of the TB care cascade, and stronger links will be needed between TB and NCD programmes, both at the national-level for governance and leadership, and within health facilities for service delivery [27]. Links with community services and peer support networks may be required, particularly for the delivery of psychosocial support [28].
Operational research will also be needed to understand how TB patients might move between TB, NCD and community-based services. The development of models of care could be informed by existing approaches to the delivery of integrated HIV services in resource-constrained settings [29–31]. Given that approximately $25\%$ of TB patients in Kenya and $45\%$ in Malawi are HIV co-infected [32], with many TB-survivors continuing to engage with HIV services even after TB treatment completion, direct integration with HIV services could be explored. Whatever the approach used, careful evaluation will be needed to determine the cost, impact, barriers and facilitators of implementation, and generalisability of these models of care, and existing guidelines for service development and evaluation should be used in pilot programmes where possible [33–35].
Our study has highlighted several data gaps which must be addressed to inform decision making by funders and policy-makers around investment in post-TB care. These include the need for in-country, local level data on the burden and distribution of post-TB morbidity. Low cost approaches to generating these data should be explored, including routine surveillance of post-TB morbidity amongst TB survivors by National TB programmes, and the use of modelling work. Funders have highlighted the need for clear metrics through which the impact of post-TB services could be evaluated, and international groups currently developing clinical guidelines for the management of post-TB morbidity are well positioned to develop these metrics.
However, our data also suggest some circularity in decision making for the support of post-TB care: NTPs may feel unable to advocate for post-TB care due to the lack of funding and lack of WHO guidelines, but funders and the WHO may require member countries to declare this as a priority area in order to move forward. Further work with policy-makers and funders may help to break this impasse. Advocacy for TB-patients and survivors will be needed to support this work, and our data and that of others suggest that the patient voice may be particularly powerful here [16].
## Strengths and limitations
This study has several limitations. It was designed as a programme of stakeholder engagement rather than primary qualitative research. As such, we did not reach full saturation in our data, and there may have been some bias in both the questions asked and the responses received during data collection. The snowballing approach will likely have identified individuals who are active or influential in this space, but may have missed less powerful individuals. COVID-19 related constraints on time and travel meant that patients and providers from rural areas, and front-line health workers were under-represented. We were largely constrained to one-to-one engagements, with limited opportunity to explore differences in perspectives in group discussions in detail. Lastly, we note the perspective raised by some stakeholders that the post-TB agenda is an external agenda, being imposed from outside. Whilst this was a minority perspective, we are grateful that participants felt comfortable to raise this concern, and acknowledge that this may have shaped some of the discussions held during data collection. This is a reminder that any efforts to take the post TB care agenda forward should be driven by local priorities, buy-in and leadership.
This work also has several strengths. This study is novel in its focus on how post-TB services might be delivered on the ground in LMIC settings. Despite calls for post-TB care, questions about practical implementation within existing health and community services remain largely unanswered. Our findings about the leadership, content and models of care will be critical to the development of pilot studies of post-TB services in Kenya and Malawi, and possibly other African countries with similar disease profiles and health systems. More broadly, our approach highlights the value of formative stakeholder engagement work in defining local priorities and challenges, prior to health system interventions. The study has strengthened the network of local stakeholders with influence and interest in this area, who are well positioned to take this agenda forward.
## Conclusion
Although international TB guidelines and policies remain focused on the management of TB disease, our findings highlight a need for holistic care, which addresses the broad physical, social, psychological, and economic wellbeing of TB patients, across the TB care cascade, including the post-TB period. Delivering patient-centred care in this way will require stronger collaboration between TB and non-communicable disease programmes at the national level, and health facility and community services at the local level. This study highlights key evidence gaps which must be addressed to support decision making by funders and policy-makers, identifies challenges which must be considered prior to implementation, and suggests opportunities to learn from integrated HIV services. Although this work was focused on Kenya and Malawi, inclusion of international partners including the Stop TB Partnership, The Global Fund and The Union has provided some global perspectives that go beyond the Eastern Africa region. We believe these data and the strong networks built through this work provide strong foundations for the development and evaluation of pilot studies of integrated TB care.
## References
1. 1World Health Organisation. Global tuberculosis report 2021. Geneva, Switzerland: World Health Organisation, 2021.
2. Dodd PJ, Yuen CM, Jayasooriya SM, van der Zalm MM, Seddon JA. **Quantifying the global number of tuberculosis survivors: a modelling study**. (2021.0). DOI: 10.1016/S1473-3099(20)30919-1
3. Meghji J, Lesosky M, Joekes E, Banda P, Rylance J, Gordon S. **Patient outcomes associated with post- tuberculosis lung damage in Malawi: a prospective cohort study**. (2020.0) **0** 1-10. DOI: 10.1136/thoraxjnl-2019-213808
4. Meghji J, Simpson H, Squire SB, Mortimer K. **A Systematic Review of the Prevalence and Pattern of Imaging Defined Post-TB Lung Disease**. (2016.0) **11** e0161176. DOI: 10.1371/journal.pone.0161176
5. Amaral AF, Coton S, Kato B, Tan WC, Studnicka M, Janson C. **Tuberculosis associates with both airflow obstruction and low lung function: BOLD results**. (2015.0) **46** 1104-12. DOI: 10.1183/13993003.02325-2014
6. Ravimohan S, Kornfeld H, Weissman D, Bisson GP. **Tuberculosis and lung damage: from epidemiology to pathophysiology**. (2018.0) **27**. DOI: 10.1183/16000617.0077-2017
7. Meghji J, Gregorius S, Madan J, Chitimbe F, Thomson R, Rylance J. **The long term effect of pulmonary tuberculosis on income and employment in a low income, urban setting**. (2020.0) **76** 387-95. DOI: 10.1136/thoraxjnl-2020-215338
8. Allwood BW, van der Zalm MM, Amaral AFS, Byrne A, Datta S, Egere U. **Post-tuberculosis lung health: perspectives from the First International Symposium**. (2020.0) **24** 820-8. DOI: 10.5588/ijtld.20.0067
9. Romanowski K, Baumann B, Basham CA, Ahmad Khan F, Fox GJ, Johnston JC. **Long-term all-cause mortality in people treated for tuberculosis: a systematic review and meta-analysis**. (2019.0) **19** 1129-37. DOI: 10.1016/S1473-3099(19)30309-3
10. Ranzani OT, Rodrigues LC, Bombarda S, Minto CM, Waldman EA, Carvalho CRR. **Long-term survival and cause-specific mortality of patients newly diagnosed with tuberculosis in São Paulo state, Brazil, 2010–15: a population-based, longitudinal study**. (2020.0) **20** 123-32. DOI: 10.1016/S1473-3099(19)30518-3
11. Marx FM, Floyd S, Ayles H, Godfrey-Faussett P, Beyers N, Cohen C. **High burden of prevalent tuberculosis among previously treated people in Southern Africa suggests potential for targeted control interventions**. (2016.0) **48** 1224-7. PMID: 27587562
12. Marx FM, Dunbar R, Enarson DA, Williams BG, Warren RM, van der Spuy GD. **The temporal dynamics of relapse and reinfection tuberculosis after successful treatment: a retrospective cohort study**. (2014.0) **58** 1676-83. DOI: 10.1093/cid/ciu186
13. Menzies NA, Quaife M, Allwood BW, Byrne A, Coussens AK, Harries AD. **The lifetime burden of disease due to incident tuberculosis: a global re-appraisal including post-TB sequelae**. (2021.0)
14. Tomeny EM, Nightingale R, Chinoko B, Nikolaidis GF, Madan J, Worrall E. *TB DALYs overlook the contribution of post-TB disability: evidence from urban Malawi* (2021.0)
15. Harries AD, Dlodlo RA, Brigden G, Mortimer K, Jensen P, Fujiwara PI. **Should we consider a ’fourth 90’ for tuberculosis?**. (2019.0) **23** 1253-6. DOI: 10.5588/ijtld.19.0471
16. Schoeman I, Sifumba Z. **Tuberculosis care does not end at treatment completion—a perspective from tuberculosis survivors**. (2021.0)
17. Allwood B, van der Zalm M, Makanda G, Mortimer K, Amaral AFS, Egere U. **The long shadow post-tuberculosis**. (2019.0) **19** 1170-1. DOI: 10.1016/S1473-3099(19)30564-X
18. Migliori GB, Marx FM, Ambrosino N, Zampongna E, Schaaf HS, van der Zalm MM. **Clinical standards for the assessment, management and rehabilitation of post-TB lung disease**. (2021.0) **25** 797-813. DOI: 10.5588/ijtld.21.0425
19. Mungai BN, Joekes E, Masini E, Obasi A, Manduku V, Mugi B. **’If not TB, what could it be?’ Chest X-ray findings from the 2016 Kenya Tuberculosis Prevalence Survey**. (2021.0)
20. Allwood B, Meghji J. *Management of Tuberculosis: A guide to essential practice* (2019.0)
21. World Health Organisation Disease TIuaTaL. *Collaborative framework for care and control of tuberculosis and diabetes* (2011.0)
22. 22World Health Organisation. A WHO / The Union monograph on TB and tobacco control: joining efforts to control two related global epidemics. Geneva, Switzerland: World Health Organisation, 2007.
23. Siddiki K, Stubbs B, Lin Y, Elsey H, Siddiki N. **Tuberculosis Multimorbidity: A Global Health Challenge Demanding Urgent Attention**. (2021.0) **25** 87-90. PMID: 33656417
24. Marais BJ, Lönnroth K, Lawn SD, Migliori GB, Mwaba P, Glaziou P. **Tuberculosis comorbidity with communicable and non-communicable diseases: integrating health services and control efforts**. (2013.0) **13** 436-48. DOI: 10.1016/S1473-3099(13)70015-X
25. Meghji J, Mortimer K, Mpangama S. **TB and COPD in low-income settings: a collision of old foes**. (2021.0). DOI: 10.1136/thoraxjnl-2021-218220
26. Li L, Lin Y, Mi F, Tan S, Liang B, Guo C. **Screening of patients with tuberculosis for diabetes mellitus in China**. **17** 1294-301. DOI: 10.1111/j.1365-3156.2012.03068.x
27. Workneh MH, Bjune GA, Yimer SA. **Assessment of health system challenges and opportunities for possible integration of diabetes mellitus and tuberculosis services in South-Eastern Amhara Region, Ethiopia: a qualitative study**. (2016.0) **16** 135. DOI: 10.1186/s12913-016-1378-6
28. Khanal S, Elsey H, King R, Baral SC, Bhatta BR, Newell JN. **Development of a Patient-Centred, Psychosocial Support Intervention for Multi-Drug-Resistant Tuberculosis (MDR-TB) Care in Nepal**. (2017.0) **12** e0167559. DOI: 10.1371/journal.pone.0167559
29. Kemp CG, Weiner BJ, Sherr KH, Kupfer LE, Cherutich PK, Wilson D. **Implementation science for integration of HIV and non-communicable disease services in sub-Saharan Africa: a systematic review**. (2018.0) **32** S93-s105. DOI: 10.1097/QAD.0000000000001897
30. Njuguna B, Vorkoper S, Patel P, Reid MJA, Vedanthan R, Pfaff C. **Models of integration of HIV and noncommunicable disease care in sub-Saharan Africa: lessons learned and evidence gaps**. (2018.0) **32 Suppl 1** S33-s42. DOI: 10.1097/QAD.0000000000001887
31. Vorkoper S, Kupfer LE, Anand N, Patel P, Beecroft B, Tierney WM. **Building on the HIV chronic care platform to address noncommunicable diseases in sub-Saharan Africa: a research agenda**. (2018.0) **32 Suppl 1** S107-s13. PMID: 29952796
32. 32World Health Organisation. Global Tuberculosis Programme: Tuberculosis country, regional and global profiles. 2021. https://www.who.int/teams/global-tuberculosis-programme/data (accessed 13th December 2021.
33. O’Cathain A, Croot L, Duncan E, Rousseau N, Sworn K, Tuner KM. **Guidance on how to develop complex interventions to improve health and healthcare**. (2019.0) **9** e029954. DOI: 10.1136/bmjopen-2019-029954
34. Hoffmann TC, Glasziou PP, Boutron I, Milne R, Perera R, Moher D. **Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide**. (2014.0) **348** g1687. DOI: 10.1136/bmj.g1687
35. Kirk MA, Kelley C, Yankey N, Birken SA, Abadie B, Damschroder L. **A systematic review of the use of the Consolidated Framework for Implementation Research**. (2016.0) **11** 72. DOI: 10.1186/s13012-016-0437-z
|
---
title: 'Relationship between cigarette smoking and blood pressure in adults in Nepal:
A population-based cross-sectional study'
authors:
- Renqiao Lan
- Max K. Bulsara
- Prakash Dev Pant
- Hilary Jane Wallace
journal: PLOS Global Public Health
year: 2021
pmcid: PMC10022357
doi: 10.1371/journal.pgph.0000045
license: CC BY 4.0
---
# Relationship between cigarette smoking and blood pressure in adults in Nepal: A population-based cross-sectional study
## Abstract
Smoking and hypertension are two major risk factors for cardiovascular disease, the leading cause of death in Nepal. The relationship between cigarette smoking and blood pressure (BP) in *Nepal is* unclear. This study analysed the data from the 2016 Nepal Demographic and Health Survey to explore the differences in systolic BP (SBP) and diastolic BP (DBP) between current daily cigarette smokers and non-smokers in Nepali adults aged 18 to 49 years. A total of 5518 women and 3420 men with valid BP measurements were included. Age, body mass index, wealth quintile (socio-economic status) and agricultural occupation (proxy for physical activity) were included as potential confounders in multivariable linear regression analysis. Women smokers were found to have significantly lower SBP (mean difference 2.8 mm, $95\%$ CI 0.7–4.8 mm) and DBP (mean difference 2.2 mm, $95\%$ CI 0.9–3.6 mm) than non-smokers after adjustment. There were no significant differences in BP between smokers and non-smokers in males, either before or after adjustment. The lower BP in female cigarette smokers in Nepal may be explained by the physiological effect of daily cigarette smoking per se in women, or unmeasured confounders associated with a traditional lifestyle that may lower BP (for example, diet and physical activity). In this nationally representative survey, daily cigarette smoking was not associated with increased BP in males or females in Nepal.
## Introduction
Cardiovascular disease is the leading cause of death globally [1] and in Nepal [2]. Both cigarette smoking and hypertension (high blood pressure) are well-established risk factors for cardiovascular disease and are thought to act synergistically on disease development [3–6]. According to the Nepal Burden of Disease 2017 report, high blood pressure and smoking are the top two risk factors for death and are responsible for $14\%$ and $13\%$ of all deaths in Nepal respectively [2]. The aetiology of primary hypertension is complex and lifestyle risk factors such as obesity, physical inactivity, excessive alcohol consumption and high salt intake are proposed to be strongly and independently associated with its development [7–9].
The understanding of the role of cigarette smoking in hypertension development continues to be refined. The hemodynamic effects of cigarette smoking are mediated primarily by nicotine [10], which can increase blood pressure (BP) acutely and temporarily via stimulation of the sympathetic nervous system [10–12]. However, with long-term exposure nicotine may have different effects [13]. For example, it is hypothesized that the nicotine metabolite, cotinine, may decrease BP via its vasodilatory effect [13]. Nicotine may also decrease BP via lowering body weight secondary to its effects of appetite suppression or increasing metabolism [14].
Epidemiological studies on the relationship between smoking and BP have produced mixed results. Some studies have found a positive association between current smoking and hypertension [15–17], including in an urban Nepali population [18]. By contrast, BP has also been found to be the same or lower in many groups of smokers compared to non-smokers [17, 19–23].
The nationally representative 2016 Nepal Demographic and Health Survey (NDHS) report found that $17\%$ of women and $23\%$ of men (aged 15 years and over) were hypertensive, and the cigarette smoking rate was $5.5\%$ among women and $26.9\%$ in men (aged 15–49 years) [24]. Cigarette smoking is the most common form of tobacco smoking by men and women in Nepal [24]. Although demographic factors and overweight/obesity were found to be associated with hypertension in adults in this survey [25–27], the relationship between cigarette smoking and BP needs further study. Given the high burden of cardiovascular disease, it is important to have a better understanding of the relationship between its two major risk factors in Nepal’s unique sociodemographic context and add to the evidence available from this specific population to inform mechanistic studies. The aim of this study is to explore the relationship between cigarette smoking and BP (systolic and diastolic) in the Nepali adult population aged 18–49 years, using data from the 2016 Nepal Demographic and Health Survey.
## Study design, setting, and participants
Data were obtained from the 2016 Nepal Demographic and Health Survey (NDHS), a nationally representative cross-sectional household survey, funded by the US Agency for International Development (USAID) [24]. The survey was conducted from June 19, 2016, to January 31, 2017, and the sampling frame was a modified version of the Nepal Central Bureau of Statistics 2011 National Population and Housing Census [24].
Households were selected in two stages in rural areas and three stages in urban areas [24]. In urban areas, wards (smallest units of local government in Nepal) were the primary sampling units (PSU), from which one enumeration area (EA) was selected. Households were subsequently selected from EAs [24]. In rural areas, wards were the PSU from which households were selected directly [24]. Only households containing a woman aged 15–49 years the night before survey administration were eligible for interview. All women aged 15–49 years who were permanent residents of the selected household or visitors who stayed the night in the household the night before the survey were eligible to be interviewed. All men aged 15–49 years from every second household who were permanent residents of the selected household or visitors who stayed the night in the households the night before the survey were eligible to be interviewed. Full details of the NDHS sampling design are discussed elsewhere [24]. Blood pressure measurements were recorded in women and men only in the subsample of households selected for the male survey [24]. Daily cigarette smoking was recorded for all participants aged 15–49 years who were interviewed.
Participants included in this study were men and women aged 18–49 years who were interviewed and with a valid BP measurement. Participants taking BP lowering medication were excluded from the study. Of the 12862 total women aged 15–49 years in the survey, 6452 women had BP measured. After inclusion and exclusion criteria were applied (62 women with technically invalid BP readings, 796 women under 18 years, and 92 women on BP lowering medication) there were 5518 women for analysis. Of the 4063 total men aged 15–49 in the survey, 4040 men had BP measured. After the exclusions were applied (22 men with technically invalid BP readings, 543 men under 18 years, and 71 men on BP lowering medication) there were 3420 men for analysis.
## Variables
Systolic BP (SBP) and diastolic BP (DBP) were the primary outcome variables. Blood pressure was measured three times in each participant at a minimum of five-minute intervals using the UA-767F/FAC blood pressure monitor (A&D Medical, Japan). The first measurement was discarded and the average of the second and third measurements was recorded as the final reading and recorded as a continuous variable (mm Hg) according to the standard DHS biomarker collection protocol [24].
Current cigarette smokers were defined as those who smoked cigarettes daily (manufactured or hand-rolled). Cigarette smokers were further categorized according to number of cigarettes smoked per day (up to 9, 10 or more). For men, this number was the average daily number of cigarettes in the past week, and for women, the number of cigarettes in the last 24 hours.
Demographic variables used to describe the participants were age (years), body mass index (BMI) (weight (kg)/height (m)2), education (Y/N), economic status (poorest wealth quintile; Y/N), social group (marginalised group, non-marginalised group), agricultural occupation (Y/N), place of residence (urban, rural), Province (1–7) and ecological zone (mountains, hills, Terai [plains]). For wealth quintile we used the NDHS household wealth index, derived from detailed information on dwelling and household characteristics, access to a variety of consumer goods and services, and assets [24]. Classification of participants as marginalised or non-marginalised was based on an ethnic grouping which is reflective of the social hierarchy in Nepal [28]. The marginalised group comprised Terai Dalit, Hill Dalit, Hill Janajati, Terai Janajati, Muslim and other Terai castes. Participants not in these groups were classified as non-marginalised.
In exploring the association between cigarette smoking and BP, age, body mass index, socioeconomic status and physical activity were considered potential confounders. Age was classified into the following sub-groups: 18–24 years, 25–34 years, 35–44 years, 45–49 years. Body mass index (BMI) was categorised according to the World Health *Organization* general population BMI classification into underweight (<18.5 kg/m2), normal (18.5 to 24.9 kg/m2), overweight (25.0 to 29.9 kg/m2) and obese (≥30.0 kg/m2). Socioeconomic status [29] was assessed through economic status (wealth quintile) and social group (marginalised ethnic group: Y/N). Physical activity was accounted for, in part, through the proxy variable of agricultural occupation (Y/N), with agricultural occupation representing higher physical activity [30, 31].
## Statistical analysis
The data were analysed using IBM SPSS Ver 26.0 software (IBM Corp., Armonk, N.Y., USA). Data were weighted using sampling weights in accordance with DHS guidelines [32]. All analyses used the Complex Sample Analysis method to account for the multi-stage sample design [32]. Data from men and women were analysed separately. There were no missing data.
The relationship between smoking and BP was assessed with linear regression. The dependent variables in linear regression were the continuous variables SBP and DBP. The potential confounders were treated as categorical variables: age, BMI, wealth quintile, social group, and agricultural work. SBP and DBP were adjusted for age (through linear regression) after stratification into the four age groups (18–24 years, 25–34 years, 35–44 years, 45–49 years). All potential confounders which showed a significant association with BP (in either sex) in age-adjusted linear regression were included in the final multivariable linear regression models. The outcomes are presented as mean SBP and DBP with $95\%$ confidence intervals.
Tests for interactions were also carried out, fitting a smoking X BMI interaction term in the models for men and women, with cigarette smoking fitted as a 2-category variable (Y/N) and BMI as a 4-category variable (underweight, normal, overweight, and obese).
## Power analysis
Using the OpenEpi2 Sample Size calculator for power analysis (comparing two means) [33] and data from the 2016 NDHS (sample size, smoking prevalence and standard deviation), there was $80\%$ statistical power to detect a 2.5 mm Hg mean difference in systolic BP between cigarette smokers and non-smokers in women, and a 1.8 mm Hg mean difference in men.
## Ethics approval
The 2016 NDHS survey protocol was approved by the Nepal Health Research Council (NHRC) and the ICF Institutional Review Board prior to administration. Written informed consent was obtained from individual respondents prior to the interviews during the NDHS data collection. Access to the NDHS 2016 dataset for this project was granted by the DHS Program before the study was carried out. The study was also approved by the Human Research Ethics Committee of the University of Notre Dame Australia, Fremantle (Ref. 2020-066F).
## Results
The characteristics of smokers in our sample (Table 1) showed several differences between men and women and to non-smokers. A smaller proportion of women ($4.8\%$) smoked cigarettes daily than men ($18.9\%$), and the same proportion ($27\%$) of women and men smokers were moderate to heavy smokers (10 cigarettes or more per day [17]). While both men and women cigarette smokers had lower BMI than non-smokers, the mean difference in BMI was larger in women (2.1 units vs. 0.8 units in women and men respectively). A higher proportion of women smokers ($41.8\%$) than men smokers ($25.5\%$) were in the poorest wealth quintile, and a much higher proportion of women smokers ($86.2\%$) had no formal education compared to men who smoked ($16.8\%$) or to women who did not smoke ($34.7\%$). Women smokers were, on average, older (mean 39.7 years) than non-smokers (mean 30.5 years) and men who smoked (mean 33.9 years). Women smokers were more often engaged in agricultural work ($64.9\%$) than non-smokers ($46.7\%$) and men who smoked ($30.8\%$). The proportion of women smokers and men smokers in marginalised social groups was the same ($68.8\%$).
**Table 1**
| Characteristic | Characteristic.1 | Characteristic.2 | Current smoker1 | Current smoker1.1 | Current smoker, cigarettes/day2 (manufactured or hand-rolled) | Current smoker, cigarettes/day2 (manufactured or hand-rolled).1 |
| --- | --- | --- | --- | --- | --- | --- |
| | | | No | Yes | Up to 9 | 10+ |
| Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | | | | |
| | n | n | 2772 | 647 | 471 | 176 |
| | % | % | 81.1 | 18.9 | 13.8 | 5.1 |
| | BMI, kg/m2 (mean ± SEM) | BMI, kg/m2 (mean ± SEM) | 22.3 ± 0.09 | 21.6 ± 0.15 | 21.8 ± 0.18 | 21.1 ± 0.26 |
| | BMI, kg/m2 [age-adjusted] (mean ± SEM) | BMI, kg/m2 [age-adjusted] (mean ± SEM) | 22.4 ± 0.10 | 21.6 ± 0.14 | 21.8 ± 0.16 | 21.0 ± 0.26 |
| | Age, years (mean ± SEM) | Age, years (mean ± SEM) | 31.0 ± 0.24 | 33.9 ± 0.43 | 33.3 ± 0.49 | 35.5 ± 1.00 |
| | Poorest quintile (%3) (n = 518) | Poorest quintile (%3) (n = 518) | 12.8 | 25.5 | 21.7 | 35.7 |
| | No education (%3) (n = 379) | No education (%3) (n = 379) | 9.7 | 16.8 | 16.7 | 17.0 |
| | Social group (%3) | Marginalised group [Terai Dalit, Hill Dalit, Hill Janajati, Terai Janajati, Muslim, other Terai Caste] (n = 2246) | 65.0 | 68.8 | 69.9 | 65.8 |
| | Social group (%3) | Non-marginalised group [Hill Brahmin, Hill Chhetri, Terai Brahmin, Terai Chhetri, Newar] (n = 1174) | 35.0 | 31.2 | 30.1 | 34.2 |
| | Agricultural occupation (%3) (proxy for physical activity) | Yes (n = 984) | 28.3 | 30.8 | 27.7 | 38.9 |
| | Agricultural occupation (%3) (proxy for physical activity) | No (n = 2436) | 71.7 | 69.2 | 72.3 | 61.1 |
| | Place of residence (%3) | Urban (n = 2228) | 64.4 | 68.5 | 71.7 | 59.7 |
| | Place of residence (%3) | Rural (n = 1192) | 35.6 | 31.5 | 28.3 | 40.3 |
| | Province (%3) | Province 1 (n = 578) | 17.1 | 16.0 | 17.8 | 11.4 |
| | Province (%3) | Province 2 (n = 675) | 21.7 | 11.1 | 13.1 | 6.0 |
| | Province (%3) | Province 3 (n = 880) | 24.0 | 33.1 | 28.7 | 44.9 |
| | Province (%3) | Province 4 (n = 302) | 8.9 | 8.8 | 9.4 | 7.0 |
| | Province (%3) | Province 5 (n = 551) | 16.9 | 12.8 | 15.5 | 5.6 |
| | Province (%3) | Province 6 (n = 167) | 4.5 | 6.3 | 5.6 | 8.2 |
| | Province (%3) | Province 7 (n = 267) | 6.9 | 11.8 | 9.9 | 16.7 |
| | Ecological zone (%3) | Mountain (n = 207) | 5.3 | 9.1 | 6.9 | 14.9 |
| | Ecological zone (%3) | Hill (n = 1511) | 43.5 | 47.1 | 40.7 | 64.3 |
| | Ecological zone (%3) | Terai (n = 1702) | 51.2 | 43.8 | 52.4 | 20.8 |
| Women (n = 5518) | Women (n = 5518) | Women (n = 5518) | | | | |
| | n | n | 5251 | 267 | 196 | 71 |
| | % | % | 95.2 | 4.8 | 3.6 | 1.3 |
| | BMI, kg/m2 (mean ± SEM) | BMI, kg/m2 (mean ± SEM) | 22.6 ± 0.11 | 21.3 ± 0.23 | 21.2 ± 0.25 | 21.5 ± 0.54 |
| | BMI, kg/m2 [age-adjusted] (mean ± SEM) | BMI, kg/m2 [age-adjusted] (mean ± SEM) | 22.8 ± 0.12 | 20.7 ± 0.24 | 20.6 ± 0.26 | 20.8 ± 0.52 |
| | Age, years (mean ± SEM) | Age, years (mean ± SEM) | 30.5 ± 0.13 | 39.7 ± 0.58 | 39.2 ± 0.62 | 41.2 ± 1.14 |
| | Poorest quintile (%3) (n = 920) | Poorest quintile (%3) (n = 920) | 15.4 | 41.8 | 41.7 | 42.2 |
| | No education (%3) (n = 2053) | No education (%3) (n = 2053) | 34.7 | 86.2 | 83.9 | 92.7 |
| | Social group (%3) | Marginalised group [Terai Dalit, Hill Dalit, Hill Janajati, Terai Janajati, Muslim, other Terai Caste] (n = 3496) | 63.1 | 68.8 | 71.1 | 62.4 |
| | Social group (%3) | Non-marginalised group [Hill Brahmin, Hill Chhetri, Terai Brahmin, Terai Chhetri, Newar] (n = 2022) | 36.9 | 31.2 | 28.9 | 37.6 |
| | Agricultural work (%3) (proxy for physical activity) | Yes (n = 2625) | 46.7 | 64.9 | 66.3 | 61.0 |
| | Agricultural work (%3) (proxy for physical activity) | No (n = 2893) | 53.3 | 35.1 | 33.7 | 39.0 |
| | Place of residence (%3) | Urban (n = 3460) | 62.7 | 62.1 | 61.1 | 65.1 |
| | Place of residence (%3) | Rural (n = 2058) | 37.3 | 37.9 | 38.9 | 34.9 |
| | Province (%3) | Province 1 (n = 931) | 17.3 | 7.9 | 8.9 | 5.2 |
| | Province (%3) | Province 2 (n = 1119) | 21.0 | 6.5 | 6.1 | 7.4 |
| | Province (%3) | Province 3 (n = 1205) | 21.3 | 32.3 | 27.0 | 47.0 |
| | Province (%3) | Province 4 (n = 543) | 9.8 | 9.8 | 11.0 | 6.1 |
| | Province (%3) | Province 5 (n = 932) | 17.0 | 14.0 | 17.0 | 5.8 |
| | Province (%3) | Province 6 (n = 306) | 5.1 | 13.5 | 12.6 | 16.0 |
| | Province (%3) | Province 7 (n = 482) | 8.4 | 16.0 | 17.3 | 12.4 |
| | Ecological zone (%3) | Mountain (n = 330) | 5.8 | 10.3 | 11.5 | 6.9 |
| | Ecological zone (%3) | Hill (n = 2431) | 43.3 | 60.0 | 56.7 | 69.1 |
| | Ecological zone (%3) | Terai (n = 2757) | 51.0 | 29.7 | 31.8 | 24.0 |
After adjustment for age, mean SBP and DBP were strongly associated with BMI category in both men and women, with significantly higher mean BP in overweight (4–6 mm) and obese (7–10 mm), and lower BP in underweight (3–6 mm), compared to the normal BMI group (Table 2). Men and women in the richest wealth quintile had significantly higher mean BP than those in middle wealth quintile (except for SBP in women), but this was a smaller effect (approximately 2–3 mm) than BMI. Men who were not engaged in agricultural work, but not women, had significantly higher BP than those who were, by approximately 2 mm. Mean BP was similar in the marginalised and non-marginalised social groups.
**Table 2**
| Characteristic | Characteristic.1 | Mean SBP, mm Hg | Mean SBP, mm Hg.1 | Mean DBP, mm Hg | Mean DBP, mm Hg.1 |
| --- | --- | --- | --- | --- | --- |
| Characteristic | Characteristic | Men | Women | Men | Women |
| BMI, kg/m2 | <18.5 | 112.0# | 105.4# | 75.2# | 72.6# |
| BMI, kg/m2 | ≥18.5–25* | 118.0 | 109.3 | 78.9 | 75.6 |
| BMI, kg/m2 | ≥25–30 | 124.3# | 113.6# | 84.9# | 80.1# |
| BMI, kg/m2 | ≥30 | 127.4# | 119.0# | 85.8# | 83.6# |
| Wealth quintile | Poorest | 118.9 | 110.3 | 79.8 | 76.2 |
| Wealth quintile | Poorer | 118.6 | 111.2¥ | 79.2 | 77.2 |
| Wealth quintile | Middle* | 117.0 | 109.9 | 78.5 | 76.1 |
| Wealth quintile | Richer | 118.1 | 109.0 | 79.3 | 75.6 |
| Wealth quintile | Richest | 120.1# | 110.8 | 81.1‡ | 77.8¥ |
| Social group | Marginalised | 118.4 | 110.6 | 79.4 | 76.7 |
| Social group | Non-marginalised* | 118.5 | 109.6 | 80.2 | 76.3 |
| Agricultural work (proxy for physical activity) | Yes* | 117.4 | 110.5 | 78.3 | 76.3 |
| Agricultural work (proxy for physical activity) | No | 118.9¥ | 110.0 | 80.3# | 76.8 |
After age-adjustment, women smokers overall had significantly lower mean SBP (mean difference 3.9 mm; $95\%$ CI 1.7–6.0 mm) and lower mean DBP (mean difference 3.4 mm, $95\%$ CI 1.9–4.8 mm) than non-smokers (Table 3). There were no significant differences in BP between smokers and non-smokers in men, either before or after age-adjustment.
**Table 3**
| BP, mmHg (95% CI) | BP, mmHg (95% CI).1 | Current smoker1 | Current smoker1.1 | Current smoker, cigarettes/day2 (manufactured or hand-rolled) | Current smoker, cigarettes/day2 (manufactured or hand-rolled).1 |
| --- | --- | --- | --- | --- | --- |
| BP, mmHg (95% CI) | BP, mmHg (95% CI) | No* | Yes | Up to 9 | 10+ |
| Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | Men (n = 3420) |
| SBP | Unadjusted | 117.3 (116.4–118.2) | 118.8 (117.0–120.5) | 118.8 | 118.7 |
| SBP | Adjusted | 118.5 (117.6–119.5) | 118.9 (117.1–120.6) | 119.1 | 118.3 |
| DBP | Unadjusted | 78.8 (78.0–79.5) | 79.3 (78.0–80.6) | 79.5 | 78.8 |
| DBP | Adjusted | 79.8 (79.0–80.5) | 79.2 (78.0–80.4) | 79.6 | 78.3 |
| Women (n = 5518) | Women (n = 5518) | | | | |
| SBP | Unadjusted | 108.3 (107.6–108.9) | 109.6 (107.4–111.8) | 109.0 | 111.2 |
| SBP | Adjusted | 110.5 (109.7–111.3) | 106.6 (104.6–108.7) # | 106.3 | 107.5 |
| DBP | Unadjusted | 75.6 (75.0–76.1) | 75.4 (73.9–76.8) | 74.8 | 76.9 |
| DBP | Adjusted | 76.8 (76.2–77.4) | 73.4 (72.0–74.9) # | 73.0 | 74.5 |
Mean BP levels after adjustment for age, BMI, wealth quintile (socio-economic status) and agricultural occupation (proxy for physical activity) are shown stratified by age, BMI and smoking status in Table 4. In both men and women, SBP and DBP increased significantly with increasing age and BMI categories. In men, the mean increase in BP in overweight compared to normal weight was 6–7 mm, and 7–10 mm in obese. In women, the mean increase in BP in overweight compared to normal weight was approximately 5 mm, and 8–10 mm in obese. Women smokers had significantly lower SBP (mean difference 2.8 mm, $95\%$ CI 0.7–4.8 mm) and lower DBP (mean difference 2.2 mm, $95\%$ CI 0.9–3.6 mm) than non-smokers after adjustment. There were no significant differences in BP between smokers and non-smokers in males, either before or after adjustment. Tests for interaction between BMI and the smoking-SBP relationship were not significant in men or women.
**Table 4**
| Characteristic | Characteristic.1 | SBP, mm Hg (95% CI) | SBP, mm Hg (95% CI).1 | DBP, mm Hg (95% CI) | DBP, mm Hg (95% CI).1 |
| --- | --- | --- | --- | --- | --- |
| Characteristic | Characteristic | Unadjusted | Adjusted | Unadjusted | Adjusted |
| Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | Men (n = 3420) | Men (n = 3420) |
| Age, years | 18–24* | 112.7 (111.5–113.9) | 116.1 (114.4–117.8) | 73.4 (72.3–74.4) | 75.6 (74.3–76.9) |
| Age, years | 25–34 | 117.3 (116.4–118.2) # | 119.0 (117.6–120.4) # | 79.4 (78.5–80.2) # | 80.3 (79.2–81.4) # |
| Age, years | 35–44 | 120.5 (119.2–121.8) # | 121.7 (119.9–123.5) # | 82.2 (81.2–83.2) # | 82.8 (81.5–84.1) # |
| Age, years | 45 and over | 123.9 (121.4–126.3) # | 125.5 (122.9–128.1) # | 83.7 (82.2–85.2) # | 84.9 (83.2–86.5) # |
| BMI, kg/m2 | <18.5 | 110.2 (108.7–111.6) # | 112.0 (110.2–113.7) # | 73.4 (72.2–74.4) # | 74.8 (73.6–76.0) # |
| BMI, kg/m2 | ≥18.5–25* | 116.9 (116.0–117.8) | 118.0 (116.8–119.2) | 78.0 (77.2–78.8) | 78.6 (77.7–79.5) |
| BMI, kg/m2 | ≥25–30 | 124.3 (123.0–125.6) # | 124.6 (123.1–126.1) # | 85.3 (84.2–86.4) # | 84.6 (83.4–85.8) # |
| BMI, kg/m2 | ≥30 | 128.3 (124.2–132.4) # | 127.7 (123.4–132.0) # | 86.9 (83.9–90.0) # | 85.4 (82.3–88.6) # |
| Smoking | Current non-smoker*,1 | 117.3 (116.4–118.2) | 120.2 (118.9–121.5) | 78.8 (78.0–79.5) | 81.0 (80.0–82.0) |
| Smoking | Current smoker (manufactured or hand-rolled) | 118.8 (117.0–120.5) | 121.0 (119.0–123.0) | 79.3 (78.0–80.6) | 80.7 (79.4–82.1) |
| Women (n = 5518) | Women (n = 5518) | Women (n = 5518) | Women (n = 5518) | Women (n = 5518) | Women (n = 5518) |
| Age, years | 18–24* | 103.3 (102.6–104.0) | 105.1 (103.8–106.4) | 71.7 (71.1–72.3) | 73.2 (72.2–74.2) |
| Age, years | 25–34 | 106.5 (105.8–107.3) # | 107.2 (105.8–108.5) # | 75.1 (74.5–75.8) # | 75.6 (74.6–76.5) # |
| Age, years | 35–44 | 113.2 (112.1–114.3) # | 113.2 (111.8–114.6) # | 78.8 (78.0–79.6) # | 78.8 (77.8–79.8) # |
| Age, years | 45 and over | 117.9 (116.0–119.8) # | 118.2 (116.3–120.1) # | 80.6 (79.5–81.8) # | 80.9 (79.7–82.1) # |
| BMI, kg/m2 | <18.5 | 102.9 (101.9–104.0) # | 104.0 (102.6–105.4) # | 71.2 (70.4–72.0) # | 71.6 (70.6–72.6) # |
| BMI, kg/m2 | ≥18.5–25* | 107.2 (106.6–107.9) | 108.1 (106.9–109.2) | 74.5 (74.0–75.0) | 74.6 (73.8–75.4) |
| BMI, kg/m2 | ≥25–30 | 112.8 (111.4–114.3) # | 112.9 (111.3–114.5) # | 79.9 (78.7–81.0) # | 79.3 (78.1–80.6) # |
| BMI, kg/m2 | ≥30 | 119.3 (117.2–121.4) # | 118.7 (116.6–120.9) # | 84.0 (82.7–85.4) # | 83.0 (81.4–84.5) # |
| Smoking | Current non-smoker*,1 | 108.3 (107.6–108.9) | 112.3 (111.4–113.2) | 75.6 (75.0–76.1) | 78.2 (77.6–78.8) |
| | Current smoker (manufactured or hand-rolled) | 109.6 (107.4–111.8) | 109.6 (107.5–111.6) ¥ | 75.4 (73.9–76.8) | 76.0 (74.6–77.4) ‡ |
## Discussion
This study used data from a nationally representative survey to examine the relationship between daily cigarette smoking and BP in Nepali adults aged 18–49 years. After adjustment for age, BMI and physical activity, no positive association was observed between cigarettes smoking and BP for men or for women. Indeed, women who smoked cigarettes had significantly lower BP than non-smoking women, by, on average, 2.8 mm and 2.2 mm for SBP and DBP respectively.
While smoking is a major risk factor for cardiovascular disease the association with increased BP is still unclear [34]. Our findings for men and women are consistent with many epidemiological studies showing that BP is either lower or the same in smokers as in non-smokers, the average difference being about 2–8 mm Hg for systolic pressure and 1–5 mm Hg for diastolic [35]. Globally, national surveys and longitudinal studies over the last twenty years have found different patterns for the smoking association with BP related to geography, sex and race. The Health Survey for England [17] found no significant difference in BP of male smokers aged 16–44 years compared to non-smokers, but in other national studies of men, notably in China and Japan, lower BP was found [20, 36]. A longitudinal study in the U.S. with a 30-year follow-up did not find a significant increase in SBP or DBP over time in male or female smokers, and white women smokers had lower DBP [34]. Studies in the UK, Sweden, Israel, and China [17, 22, 37, 38] have also found a lower BP in women current smokers compared to non-smokers, consistent with our findings, and a longitudinal study risk of incident hypertension in women conducted in the U.S. found cigarette smoking does not significantly increase the risk of incident hypertension in women smoking up to 15 cigarettes per day [15]. A meta-analysis of over 20 population-based studies conducted in many geographic locations concluded that there was no causal association between smoking heaviness in current smokers of either sex and SBP or DBP [39].
A systematic review of hypertension in low- and middle-income countries found geographic differences in the relationship between smoking and BP with lower proportions of hypertension among smokers compared to non-smokers in Europe and Central Asia, Latin America and Caribbean, and Middle East and North Africa regions, but higher proportions of hypertension amongst smokers compared to non-smokers in East Asia and Pacific, South Asia, and Sub-Saharan Africa regions [40]. Individual studies in Nepal have produced inconsistent findings, and many have not included adjustments for age and BMI [41–47]. One study in periurban Kathmandu found an association between current cigarette smoking and higher BP after adjustment [18]. Other studies conducted in semi-urban and rural settings in Nepal [48–50], and a nationwide survey [51], found current smoking was not significantly associated with BP in multivariable analyses. A systematic review and meta-analysis of 12 studies undertaken in Nepal in the last 20 years [52] estimated smokers have 1.43 times the odds ($95\%$ CI 1.14–1.79) of hypertension based on unadjusted odds ratios which did not control for confounders. Other recent systematic reviews and meta-analyses of the prevalence of hypertension in Nepal [53, 54] did not examine the effect of cigarette smoking. Two studies using data from the same 2016 NDHS survey as the present study to examine risk factors for hypertension [55, 56] used cigarette smoking in the 30 minutes prior to the BP measurement (Y/N) as their smoking variable and hence measured the short-term impact of nicotine on elevating BP [11], rather than the chronic effect of cigarette smoking on BP. The mixed results between studies demonstrate that methodological differences, different populations, and additional unmeasured confounders (e.g., lifestyle, diet, cultural characteristics, physical activity) may influence the observed relationship between smoking and BP.
The strength of our study compared to other studies conducted in *Nepal is* that we have used data from a nationally representative survey, rather than a specific geographical location, to examine the association between current cigarette smoking and BP. In addition, we used smoking variables that reflect daily smoking patterns rather than smoking in the 30 minutes before BP measurement. We adjusted for age, BMI, socioeconomic status and social differences, but we were unable to adjust for physical activity directly, dietary intake of fruit, vegetables and salt, or alcohol consumption, as these potential confounders were not collected in the 2016 NDHS. To account for physical activity, we used a proxy variable which provided a limited adjustment for this variable. Alcohol intake is strongly associated with smoking in Nepal [49] and may affect the smoking-BP relationship. However, in the study of Primatesta et al. [ 17] alcohol consumption did not alter smoking effects on BP for men or women. We also did not exclude individuals who had smoked or consumed alcohol or caffeine within 30 minutes before the BP readings, or who were users of smokeless tobacco, which may be a source of confounding. Other limitations include a lack of statistical power to undertake sub-group analysis by level of smoking. The use of office BP measurements in the NDHS rather than 24-hour ambulatory BP did not enable detection of BP changes throughout the day, including patterns such as the white-coat effect and masked hypertension [57]. In addition, cigarette smoking might act preferentially on central BP, rather than brachial BP, in the development of hypertensive target organ disease [12]. Since this is a cross-sectional study, we could not establish causality due to the lack of temporal relationship between smoking and BP.
Overall, we showed that BP is either lower or the same in daily cigarette smokers as in non-smokers in the age group 18–49 years. Our finding that women cigarette smokers, but not men smokers, had a significantly lower BP than non-smokers may reflect either, [1] the physiological effect of cigarette smoking per se in women, or [2] unmeasured dietary, lifestyle or health factors associated with low education, poverty, and agricultural work that were characteristics of the women smokers in the study and which may independently lower BP. Possible unmeasured factors include the consumption of a traditional diet (i.e., high in fiber and vegetables, low in fat) and less sedentary behaviour as part of a more traditional lifestyle [58]. While smoking was associated with a lower BMI in women, BMI was adjusted for in the final model.
## Conclusions
This study describes the association between cigarette smoking and blood pressure in adults in Nepal aged 18–49 years. Our finding that daily cigarette smoking was not associated with increased BP in men or women in this population contributes to the understanding of the relationship between these two major risk factors for cardiovascular disease in populations that share characteristics with Nepal. However, the results of this study should not be used to influence the public health campaigns on smoking cessation as cigarette smoking is a strong independent risk factor for cardiovascular disease. Future research employing longitudinal studies, the use of ambulatory BP monitoring, adjusting for additional confounders, and studies on arterial stiffness and central blood pressure [12] may provide further insight into of the effect of smoking on BP.
## References
1. 1World Health Organization. Global Health Estimates 2020: Deaths by cause, age, sex, by country and by region, 2000–2019. Geneva; 2020.
2. 2Nepal Health Research Council (NHRC), Ministry of Health and Population (MoHP), Monitoring Evaluation and Operational Research (MEOR). Nepal Burden of Disease 2017: A Country Report based on the Global Burden of Disease 2017 Study. Kathmandu, Nepal; 2019.
3. Njolstad I, Arnesen E, Lund-Larsen PG. **Smoking, serum lipids, blood pressure, and sex differences in myocardial infarction. A 12-year follow-up of the Finnmark Study**. *Circulation* (1996.0) **93** 450-6. DOI: 10.1161/01.cir.93.3.450
4. Prescott E, Hippe M, Schnohr P, Hein HO, Vestbo J. **Smoking and risk of myocardial infarction in women and men: longitudinal population study**. *BMJ* (1998.0) **316** 1043-7. DOI: 10.1136/bmj.316.7137.1043
5. Lewington S, Clarke R, Qizilbash N, Peto R, Collins R, Prospective Studies C. **Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies**. *The Lancet* (2002.0) **360** 1903-13. DOI: 10.1016/s0140-6736(02)11911-8
6. De Cesaris R, Ranieri G, Filitti V, Bonfantino MV, Andriani A. **Cardiovascular effects of cigarette smoking.**. *Cardiology* (1992.0) **81** 233-7. DOI: 10.1159/000175809
7. Forman JP, Stampfer MJ, Curhan GC. **Diet and lifestyle risk factors associated with incident hypertension in women**. *JAMA* (2009.0) **302** 401-11. DOI: 10.1001/jama.2009.1060
8. Sonne-Holm S, Sorensen TI, Jensen G, Schnohr P. **Independent effects of weight change and attained body weight on prevalence of arterial hypertension in obese and non-obese men**. *BMJ* (1989.0) **299** 767-70. DOI: 10.1136/bmj.299.6702.767
9. Carnethon MR, Evans NS, Church TS, Lewis CE, Schreiner PJ, Jacobs DR. **Joint associations of physical activity and aerobic fitness on the development of incident hypertension: coronary artery risk development in young adults**. *Hypertension* (2010.0) **56** 49-55. DOI: 10.1161/HYPERTENSIONAHA.109.147603
10. Benowitz NL. **Cigarette smoking and cardiovascular disease: pathophysiology and implications for treatment**. *Progress in Cardiovascular Diseases* (2003.0) **46** 91-111. DOI: 10.1016/s0033-0620(03)00087-2
11. Benowitz NL. **Clinical pharmacology of nicotine**. *Annu Rev Med* (1986.0) **37** 21-32. DOI: 10.1146/annurev.me.37.020186.000321
12. Virdis A, Giannarelli C, Neves MF, Taddei S, Ghiadoni L. **Cigarette smoking and hypertension**. *Curr Pharm Des* (2010.0) **16** 2518-25. DOI: 10.2174/138161210792062920
13. Benowitz NL, Sharp DS. **Inverse relation between serum cotinine concentration and blood pressure in cigarette smokers**. *Circulation* (1989.0) **80** 1309-12. DOI: 10.1161/01.cir.80.5.1309
14. Chiolero A, Faeh D, Paccaud F, Cornuz J. **Consequences of smoking for body weight, body fat distribution, and insulin resistance**. *Am J Clin Nutr* (2008.0) **87** 801-9. DOI: 10.1093/ajcn/87.4.801
15. Bowman TS, Gaziano JM, Buring JE, Sesso HD. **A prospective study of cigarette smoking and risk of incident hypertension in women**. *J Am Coll Cardiol* (2007.0) **50** 2085-92. DOI: 10.1016/j.jacc.2007.08.017
16. Groppelli A, Giorgi DM, Omboni S, Parati G, Mancia G. **Persistent blood pressure increase induced by heavy smoking**. *J Hypertens* (1992.0) **10** 495-9. DOI: 10.1097/00004872-199205000-00014
17. Primatesta P, Falaschetti E, Gupta S, Marmot MG, Poulter NR. **Association between smoking and blood pressure: evidence from the health survey for England**. *Hypertension* (2001.0) **37** 187-93. DOI: 10.1161/01.hyp.37.2.187
18. Dhungana RR, Pandey AR, Bista B, Joshi S, Devkota S. **Prevalence and associated factors of hypertension: A community-based cross-sectional study in municipalities of Kathmandu, Nepal.**. *Int J Hypertens* (2016.0) **2016** 1656938. DOI: 10.1155/2016/1656938
19. Kim BJ, Han JM, Kang JG, Kim BS, Kang JH. **Association between cotinine-verified smoking status and hypertension in 167,868 Korean adults.**. *Blood Press.* (2017.0) **26** 303-10. DOI: 10.1080/08037051.2017.1344539
20. Li G, Wang H, Wang K, Wang W, Dong F, Qian Y. **The association between smoking and blood pressure in men: a cross-sectional study**. *BMC Public Health* (2017.0) **17** 797. DOI: 10.1186/s12889-017-4802-x
21. Rhee MY, Na SH, Kim YK, Lee MM, Kim HY. **Acute effects of cigarette smoking on arterial stiffness and blood pressure in male smokers with hypertension**. *Am J Hypertens* (2007.0) **20** 637-41. DOI: 10.1016/j.amjhyper.2006.12.017
22. Green MS, Jucha E, Luz Y. **Blood pressure in smokers and nonsmokers: Epidemiologic findings**. *American Heart Journal* (1986.0) **111** 932-40. DOI: 10.1016/0002-8703(86)90645-9
23. Imamura H, Tanaka K, Hirae C, Futagami T, Yoshimura Y, Uchida K. **Relationship of cigarette smoking to blood pressure and serum lipids and lipoproteins in men**. *Clin Exp Pharmacol Physiol* (1996.0) **23** 397-402. DOI: 10.1111/j.1440-1681.1996.tb02748.x
24. 24Ministry of Health Nepal (MOHN), New ERA, ICF. Nepal Demographic and Health Survey 2016. Kathmandu, Nepal; 2017. Available from: https://dhsprogram.com/publications/publication-fr336-dhs-final-reports.cfm.
25. Gupta RD, Talukdar A, Haider SS, Haider MR. **Prevalence and associated factors of hypertension subtypes among the adult population in Nepal: Evidence from Demographic and Health Survey data.**. *Osong Public Health Res Perspect* (2019.0) **10** 327-36. DOI: 10.24171/j.phrp.2019.10.6.02
26. Hasan M, Sutradhar I, Akter T, Das Gupta R, Joshi H, Haider MR. **Prevalence and determinants of hypertension among adult population in Nepal: Data from Nepal Demographic and Health Survey 2016.**. *PLoS One.* (2018.0) **13** e0198028. DOI: 10.1371/journal.pone.0198028
27. Kibria GMA, Swasey K, Sharmeen A, Sakib MN, Burrowes V. **Prevalence and associated factors of pre-hypertension and hypertension in Nepal: Analysis of the Nepal Demographic and Health Survey 2016.**. *Health Sci Rep.* (2018.0) **1** e83. DOI: 10.1002/hsr2.83
28. Devkota B, Maskey J, Pandey AR, Karki D, Godwin P, Gartoulla P. **Determinants of home delivery in Nepal—A disaggregated analysis of marginalised and non-marginalised women from the 2016 Nepal Demographic and Health Survey.**. *PLoS One.* (2020.0) **15** e0228440. DOI: 10.1371/journal.pone.0228440
29. Busingye D, Arabshahi S, Subasinghe AK, Evans RG, Riddell MA, Thrift AG. **Do the socioeconomic and hypertension gradients in rural populations of low- and middle-income countries differ by geographical region? A systematic review and meta-analysis.**. *Int J Epidemiol.* (2014.0) **43** 1563-77. DOI: 10.1093/ije/dyu112
30. Patel S, Ram U, Ram F, Patel SK. **Socioeconomic and demographic predictors of high blood pressure, diabetes, asthma and heart disease among adults engaged in various occupations: evidence from India.**. *J Biosoc Sci.* (2020.0) **52** 629-49. DOI: 10.1017/S0021932019000671
31. Sorensen TB, Matsuzaki M, Gregson J, Kinra S, Kadiyala S, Shankar B. **Is agricultural engagement associated with lower incidence or prevalence of cardiovascular diseases and cardiovascular disease risk factors? A systematic review of observational studies from low- and middle-income countries.**. *PLoS One.* (2020.0) **15** e0230744. DOI: 10.1371/journal.pone.0230744
32. 32Croft TN, Aileen MJM, C. K., Allen EA. Guide to DHS Statistics2018. Available from: https://dhsprogram.com/data/Guide-to-DHS-Statistics/index.cfm.
33. 33Dean AG, Sullivan KM, Soe MM. OpenEpi: Open Source Epidemiologic Statistics for Public Health 2013. Available from: https://www.openepi.com/SampleSize/SSCohort.htm.
34. Luehrs RE, Zhang D, Pierce GL, Jacobs DR, Kalhan R, Whitaker KM. **Cigarette smoking and longitudinal associations with blood pressure: The CARDIA Study.**. *J Am Heart Assoc.* (2021.0) **10** e019566. DOI: 10.1161/JAHA.120.019566
35. Pickering TG. **Effects of stress and behavioral interventions in hypertension—the effects of smoking and nicotine replacement therapy on blood pressure**. *J Clin Hypertens (Greenwich).* (2001.0) **3** 319-21. DOI: 10.1111/j.1524-6175.2001.00483.x
36. Okubo Y, Suwazono Y, Kobayashi E, Nogawa K. **An association between smoking habits and blood pressure in normotensive Japanese men: a 5-year follow-up study.**. *Drug Alcohol Depend.* (2004.0) **73** 167-74. DOI: 10.1016/j.drugalcdep.2003.10.005
37. Janzon E, Hedblad B, Berglund G, Engstrom G. **Changes in blood pressure and body weight following smoking cessation in women**. *J Intern Med* (2004.0) **255** 266-72. DOI: 10.1046/j.1365-2796.2003.01293.x
38. Wang M, Li W, Zhou R, Wang S, Zheng H, Jiang J. **The paradox association between smoking and blood pressure among half million Chinese people**. *Int J Environ Res Public Health* (2020.0) **17**. DOI: 10.3390/ijerph17082824
39. Linneberg A, Jacobsen RK, Skaaby T, Taylor AE, Fluharty ME, Jeppesen JL. **Effect of smoking on blood pressure and resting heart rate: A Mendelian randomization meta-analysis in the CARTA Consortium**. *Circ Cardiovasc Genet* (2015.0) **8** 832-41. DOI: 10.1161/CIRCGENETICS.115.001225
40. Sarki AM, Nduka CU, Stranges S, Kandala NB, Uthman OA. **Prevalence of hypertension in low- and middle-income countries: A systematic review and meta-analysis**. *Medicine (Baltimore).* (2015.0) **94** e1959. DOI: 10.1097/MD.0000000000001959
41. Vaidya A, Pokharel PK, Karki P, Nagesh S. **Exploring the iceberg of hypertension: a community based study in an eastern Nepal town**. *Kathmandu University Medical Journal* (2007.0) **5** 349-59
42. Manandhar K, Koju R, Sinha NP, Humagain S. **Prevalence and associated risk factors of hypertension among people aged 50 years and more in Banepa Municipality, Nepal.**. *Kathmandu Univ Med J (KUMJ).* (2012.0) **10** 35-8. DOI: 10.3126/kumj.v10i3.8015
43. Shrestha DB, Dhungel S. **Prevalence and risk factors of hypertension in Hansposa VDC of Sunsari District, Nepal.**. *Medical Journal of Shree Birendra Hospital.* (2017.0) **15** 48-53
44. Shrestha S, Devkota R. **Prevalence of hypertension and its associated risk factors in a sub-urban area of central Nepal**. *International Journal of Community Medicine and Public Health* (2016.0) 2477-86
45. Maharjan B. **Prevalence and Awareness of Hypertension among Adults and its Related Risk Factors**. *J Nepal Health Res Counc* (2017.0) **15** 242-6
46. Kafle R, Sharma D, Paudel N, Sapkota S, Alurkar V. **Prevalence and associated risk factors of hypertension in a rural community of Western Nepal: A cross sectional study**. *Journal of Advances in Internal Medicine* (2018.0) **7** 11-6
47. Karki DK, Nepal S, Bhandari KR. **Prevalence and associated risk factors of hypertension among adults in Palpa District, Nepal.**. *Journal of Lumbini Medical College.* (2019.0) **7** 107-12
48. Chataut J, Khanal K, Manandhar K. **Prevalence and associated factors of hypertension among adults in rural Nepal: A community based study**. *Kathmandu Univ Med J (KUMJ).* (2015.0) **13** 346-50. DOI: 10.3126/kumj.v13i4.16835
49. Khanal MK, Dhungana RR, Bhandari P, Gurung Y, Paudel KN. **Prevalence, associated factors, awareness, treatment, and control of hypertension: Findings from a cross sectional study conducted as a part of a community based intervention trial in Surkhet, Mid-western region of Nepal.**. *PLoS One.* (2017.0) **12** e0185806. DOI: 10.1371/journal.pone.0185806
50. Gyawali B, Mishra SR, Ghimire S, Hansen MRH, Shah KJ, Subedee KC. **The burden and correlates of multiple cardiometabolic risk factors in a semi-urban population of Nepal: a community-based cross-sectional study.**. *Sci Rep* (2019.0) **9** 15382. DOI: 10.1038/s41598-019-51454-9
51. Koju R, Manandhar K, Risal A, Steiner TJ, Holen A, Linde M. **Undertreated hypertension and its implications for public health in Nepal: Nationwide population-based survey**. *Kathmandu Univ Med J (KUMJ).* (2015.0) **13** 3-7. DOI: 10.3126/kumj.v13i1.13744
52. Shrestha DB, Budhathoki P, Sedhai YR, Baniya A, Lamichhane S, Shahi M. **Prevalence, awareness, risk factors and control of hypertension in Nepal from 2000 to 2020: A systematic review and meta-analysis.**. *Public Health in Practice.* (2021.0) **2** 100119. PMID: 36101638
53. Dhungana RR, Pandey AR, Shrestha N. **Trends in the prevalence, awareness, treatment, and control of hypertension in Nepal between 2000 and 2025: A systematic review and meta-analysis.**. *Int J Hypertens.* (2021.0) **2021** 6610649. DOI: 10.1155/2021/6610649
54. Huang Y, Guo P, Karmacharya BM, Seeruttun SR, Xu DR, Hao Y. **Prevalence of hypertension and prehypertension in Nepal: a systematic review and meta-analysis**. *Glob Health Res Policy* (2019.0) **4** 11. DOI: 10.1186/s41256-019-0102-6
55. Agho KE, Osuagwu UL, Ezeh OK, Ghimire PR, Chitekwe S, Ogbo FA. **Gender differences in factors associated with prehypertension and hypertension in Nepal: A nationwide survey**. *PLoS One* (2018.0) **13** e0203278. DOI: 10.1371/journal.pone.0203278
56. Hasan MM, Tasnim F, Tariqujjaman M, Ahmed S, Cleary A, Mamun A. **Examining the prevalence, correlates and inequalities of undiagnosed hypertension in Nepal: a population-based cross-sectional study**. *BMJ Open.* (2020.0) **10** e037592. DOI: 10.1136/bmjopen-2020-037592
57. Muntner P, Shimbo D, Carey RM, Charleston JB, Gaillard T, Misra S. **Measurement of blood pressure in humans: A scientific statement from the American Heart Association**. *Hypertension* (2019.0) **73** e35-e66. DOI: 10.1161/HYP.0000000000000087
58. 58Aryal KK, Neupane S, Mehata S, Vaidya A, Singh S, Paulin F, et al. Non communicable diseases risk factors: STEPS Survey Nepal 2013. Kathmandu, Nepal; 2015.
|
---
title: Association of the retail food environment, BMI, dietary patterns, and socioeconomic
position in urban areas of Mexico
authors:
- Elisa Pineda
- Diana Barbosa Cunha
- Mansour Taghavi Azar Sharabiani
- Christopher Millett
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022358
doi: 10.1371/journal.pgph.0001069
license: CC BY 4.0
---
# Association of the retail food environment, BMI, dietary patterns, and socioeconomic position in urban areas of Mexico
## Abstract
The retail food environment is a key modifiable driver of food choice and the risk of non-communicable diseases (NCDs). This study aimed to assess the relationship between the density of food retailers, body mass index (BMI), dietary patterns, and socioeconomic position in Mexico. Cross-sectional dietary data, BMI and socioeconomic characteristics of adult participants came from the nationally representative 2012 National Health and Nutrition Survey in Mexico. Geographical and food outlet data were obtained from official statistics. Densities of food outlets per census tract area (CTA) were calculated. Dietary patterns were determined using exploratory factor analysis and principal component analysis. The association of food environment variables, socioeconomic position, BMI, and dietary patterns was assessed using two-level multilevel linear regression models. Three dietary patterns were identified—the healthy, the unhealthy and the carbohydrates-and-drinks dietary pattern. Lower availability of fruit and vegetable stores was associated with an unhealthier dietary pattern whilst a higher restaurant density was associated with a carbohydrates-and-drinks pattern. A graded and inverse association was observed for fruit and vegetable store density and socioeconomic position (SEP)—lower-income populations had a reduced availability of fruit and vegetable stores, compared with higher-income populations. A higher density of convenience stores was associated with a higher BMI when adjusting for unhealthy dietary patterns. Upper-income households were more likely to consume healthy dietary patterns and middle-upper-income households were less likely to consume unhealthy dietary patterns when exposed to high densities of fruit and vegetable stores. When exposed to a high concentration of convenience stores, lower and upper-lower-income households were more likely to consume unhealthy dietary patterns. Food environment and sociodemographic conditions within neighbourhoods may affect dietary behaviours. Food environment interventions and policies which improve access to healthy foods and restrict access to unhealthy foods may facilitate healthier diets and contribute to the prevention of NCDs.
## Introduction
Food choice is influenced by a complex set of determinants that include biological, economic, and social factors as well as attitudes, knowledge and beliefs [1, 2]. Unhealthy food choices are a major modifiable risk factor for diet related non-communicable diseases (NCDs) [3]. The retail food environment has been identified as a key driver of food choice [4, 5]. It encompasses the accessibility of food retailers, the availability of healthy or unhealthy foods and beverages within these establishments, and their affordability and promotion [6]. Unhealthy retail food environments have been associated with a higher risk of diet related NCDs [7].
The implementation of the North American Free Trade Agreement (NAFTA) agreement between Mexico, the US and Canada in 1994 allowed an increased importation of ultra-processed foods and growth in internationally owned fast-food chain retailers along with the export of fruits and vegetables to the US [8, 9]. Mexico has experienced a nutritional transition, including a decline in fruit and vegetable intake, and an increasing trend of overweight and obesity [10, 11], such that the country now has one of the highest prevalence of obesity worldwide at $75\%$ [12]. With the intention to curb the obesity trend, in 2014, Mexico became one of the first countries to implement a sugar-sweetened beverage (SSB) tax [13]. Since then, several policies have targeted nutritional labelling and televised food marketing. However, the burden of obesity continues to grow along with an increasingly obesogenic food environment which offers a wide range of high-calorie and low-nutrient foods and beverages at low prices [12, 14, 15].
Previous studies have shown that healthier dietary intakes can be enabled through supportive food environments which facilitate access to healthy and affordable food choices, such as fruits and vegetables [16]. Inequalities in access to affordable, healthy, and nutritious food can contribute to health disparities. Research has shown that low-income areas have limited access to healthy foods which may limit the ability of individuals to have healthier diets, exposing them to a greater risk of obesity and NCDs [16].
The price of fruits and vegetables has been rising more than most other foods, including energy-dense processed foods in Mexico [17]. This is concerning and reflects a current lack of policies to address the declining consumption of fruit and vegetables in the country. In addition, the consumption of energy dense foods high in saturated fat, salt and sugar (HFSS) and ultra-processed foods, has increased [18]. Mexico has become one of the highest consumers of HFSS and SSBs in Latin America [19]. From 2009 to 2014, sales of ultra-processed foods increased by $5\%$ Mexico and SSBs contributed to $22\%$ of the total energy intake per capita [19]. Main policy recommendations include to reduce consumption of ultra-processed foods. Yet, although previous research has assessed the relationship of the food environment and obesity in Mexico [20, 21], evidence of the impact of the retail food environment on dietary intake and particularly dietary patterns is limited in Mexico. Furthermore, no actions have been undertaken at national level to improve the food environment to enable healthier food choices. Therefore, the aims of this study were 1) to test the association of dietary patterns and the retail food environment, 2) to test the association of the retail food environment and BMI whilst considering dietary patterns and 3) to test the role of socioeconomic position (SEP) in dietary patterns and the retail food environment.
This study builds on previous research in which the association of body mass index (BMI) and food environment in Mexico were studied [20]. In addition to BMI the present study considers the interaction of dietary patterns with the food environment and the confounding effect of diet when testing the association of the food environment and BMI. We hypothesized that a healthy dietary pattern would be associated with a higher supermarket and fruit and vegetable store availability due to the potential higher availability of healthy foods (e.g. fruits and vegetables) and that this association would be more evident in upper-income SEP households due to previous studies showing greater availability of healthy foods and outlets in higher-income neighbourhoods [22–24]. We also hypothesized that unhealthy dietary patterns would be associated with food stores which focused their sales on unhealthy low-nutrient, energy-dense foods.
## Ethics statement
Ethical approval was sought and obtained by the National Institute of Health (INSP) in Mexico from the NIH Research Ethics Committee to carry out the National Health and Nutrition Survey (ENSANUT).
## Study design
This study involved a secondary analysis of cross-sectional and population-based survey data. Dietary and sociodemographic data were obtained from the 2012 Mexican National Survey of Health and Nutrition (ENSANUT, acronym in Spanish) [25]. This was a national probabilistic survey with state level representation by urban and rural strata and an oversampling of households with the greatest deprivation levels in the country. The sample for ENSANUT included the overrepresentation of households in the country in conditions of greater vulnerability, on the assumption that the support of health and social programs is focused on these households. All used survey data was previously anonymised by the National Institute of Health and Nutrition (INSP, acronym in Spanish) in Mexico.
## Anthropometric and sociodemographic data
Body weight and height were measured by trained personnel for the 2012 ENSANUT survey [26]. Sociodemographic data, including sex, age, car ownership, type of health service user, participation in food programmes (type of programme is described in the results section) and region were extracted for this study from the 2012 ENSANUT survey [26] and captured in a database in STATA 14 [27].
Physical activity was self-reported and was assessed by the ENSANUT survey through the International physical activity questionnaire (IPAQ short), which was previously validated [26]. Physical activity levels were defined as active, moderately active, and inactive/sedentary according to the criteria stablished by the World Health Organization [26, 28].
Household socioeconomic position (SEP) was obtained from the 2012 ENSANUT [12] which considered demographic and socioeconomic characteristics, including characteristics of the head of the family, sociodemographic structure, characteristics of the home, household goods, family consumption patterns and characteristics of the geographical area of residence based on the 2010 National Income and Expenditure Survey [29]. Distribution of SEP household characteristics was described by predicted decile. Deciles were then compared with a measure of poverty to create quintiles which were then equally assigned to each household member. A lower quintile indicates lower-income whilst a higher quintile indicates a higher-income [20, 29]. Study population data focused on the general adult population and excluded data from women who were pregnant, survey participants <18 years of age, and participants without a valid, measured weight and height. Participants with BMI values of >3 standard deviations from the mean were excluded (<15 kg/m2 and >58 kg/m2) in case of possible underlying illnesses, eating disorders or implausible values.
## Dietary intake
Dietary information was obtained through a validated, semi quantitative food frequency questionnaire (FFQ) [26, 30], which was applied to $11\%$ of the ENSANUT participant sample population [26]. The analytical sample size was $$n = 1$$,572. The FFQ included data regarding the consumption of 140-food items. For each food item, portion size and frequency per day, week and year was registered [26].
## Geocoding of individuals and food outlets
The geographical areas of study were urban neighbourhoods in the country of Mexico. Census tract areas (CTA) were used as a proxy for neighbourhoods. A CTA in *Mexico is* defined as a geographic area formed by of a set of blocks delimited by streets or identifiable pathways with land used for residential industrial or commercial services [31]. Urban CTAs contain a population of ≥2,500 inhabitants. There were 55,427 urban CTAs in this study, with mean area of 0.59 km2. The smallest CTA was 0.009 km2 and the largest 5.20 km2 [31].
Anonymised participants from the 2012 ENSANUT were geocoded to the centroid of their urban CTA with ArcGIS 10.2.2 (ESRI, Redlands, CA). Exact ENSANUT participant’s geolocation (home address) was unavailable to protect participant’s privacy. Geographic coordinates of food outlets were obtained from INEGI, 2014 [32]. On-site verification of nine geographic area samples was undertaken to verify the geolocation, existence, and type of food outlet. Urban CTAs were grouped into a single shapefile, which was spatially merged with geolocation and sociodemographic characteristics of participants and food outlets.
## Food retail data
Retail data were obtained from the 2014 Economic Census from the National Institute of Statistics and Geography (INEGI, acronym in Spanish) [31]. Food outlets encompassed convenience stores fast-food outlets, restaurants, supermarkets and fruit and vegetable stores. Classification of food outlets was according to INEGI and revised on our previous study [20]. To summarise, for this study, food outlets, including informal and mobile food carts, which specialised in pizzas, hamburgers, hotdogs, and fried chicken were classified as fast-food outlets. Outlets that mainly sold SSBs and unhealthy snacks were classified as convenience stores. We assumed that all convenience stores and fast-food outlets sold mainly SSBs, snacks and ultra-processed foods. Food outlets with an á la carte menu, that included healthy food alternatives with sitting options available, were classified as restaurants. Mega-supermarkets and grocery stores, which offered greater food options than convenience stores, including fruits and vegetables, were classified as supermarkets. Fruit and vegetable stores outlets included informal fruit stands, small shops, and farmer market style locations. These establishments were characterised by mainly selling fruit and/or vegetables.
Store data, which included type of store and location, was obtained from the 2014 economic census from the National Institute of Statistics and Geography (INEGI) [31]. Density of food outlets was calculated considering the total number of food outlet (e.g. restaurants, fast-food outlets, convenience stores and fruit and vegetable stores) per CTA. Density units are expressed as food outlet per CTA km2. The density of food outlets by state was mapped to visualize the distribution in the country. Due to a lower availability of food outlets, a higher availability of informal commerce not in record and a higher density of food crops) [20, 21] rural areas were excluded from this study.
## Dietary patterns
The 140-food items from the FFQs were aggregated grouping together food items according to their nutritional and common habitual dietary consumption (i.e., how foods are usually paired up for consumption in the study population (e.g., tortillas with beans or milk with cereal), based on previous research [33, 34]. Dietary patterns were computed using exploratory factor analysis (FA), principal component as the extraction method and varimax rotation. Kaiser-Meyer-Olkin test was undertaken to investigate the adequacy of FA to the data. A scree plot was used to select the number of factors to retain (S1 Fig). Food items with factor loading ≥0.30 were retained in the pattern. Factor scores were computed and included in the regression analysis using SAS OnDemand for Academics (SAS Institute Inc., Cary, USA) [35].
Kaiser’s measure of sampling adequacy was 0.79 which indicated that FA was appropriate method due to a correlation of variables [38]. Three dietary patterns were retained, according to the scree plot (S1 Fig), which together explained a variance of $26.3\%$ for factor 1; $26.1\%$ for factor 2; and $16.4\%$ for factor 3 of the total variance of the data. We retained the factors above the inflection point of the curve of the scree plot (patterns that showed and eigen value >$\frac{1}{0}$).
For each of the food groups, a factor loading value ≥0.28 indicated that the food group was included in the factor. Therefore, considering all food groups and corresponding factor loadings, factor 1 was classified as a healthy pattern, characterized by a higher consumption, in comparison of other factors, of fruits and vegetables, cooked meals, pulses, fish and seafood, meat, fermented dairy, soups, bread, and natural drinks. Factor 2 was classified as an unhealthy dietary pattern, mainly composed of fats, high meat and fatty meals, sugar and desserts, sausages, dressings, soda, fast-food, ready to eat soups, potato chips and candy. Factor 3 was denominated as carbohydrates-and-drinks pattern, which encompassed juice and natural drinks, whole wheat products, milk, and refined cereal (S1 Table).
## Association of dietary patterns and the retail food environment
The main outcome of this study was dietary intake represented as dietary patterns and the exposure was food outlet store density (food outlet count per census tract area). Statistical models were constructed after drawing the postulated relationship of variables through directed acyclic graphs (DAG) [36], which captured the dependence structure of multiple variables and allow more robust conclusions about the direction of association. Three multilevel linear regression models were used to test the association between food outlet density and dietary patterns. State, or CTA were used as a second random effect in the three models (Table 1).
**Table 1**
| Model | Covariates | Second random effect level |
| --- | --- | --- |
| Model A | Age, sex, and SEP. | State |
| Model B | Age, sex, SEP, physical activity level, car ownership, neighbourhood deprivation level. | CTA |
| Model C | Age, sex, SEP, CTA deprivation level, and urbanity level. | State |
## Association of the retail food environment and BMI whilst considering dietary patterns
Models A, B and C were replicated to test first the association of the food environment and BMI considering the interaction of the food environment with the dietary patterns and second, to test the association of the food environment and BMI whilst considering the dietary patterns as confounders. To assess sex differences, results were also stratified by sex.
When testing the association of the food environment and dietary patterns, a higher density of supermarkets and fruit and vegetable stores was inversely associated with the carbohydrates-and-drinks dietary pattern (factor 3) for model A (βsupermarkets = -0.36, $95\%$CI: -0.65, -0.06, $$P \leq 0.02$$); (βfruit and vegetable stores = -0.40, $95\%$CI: -0.69, -0.11, $$P \leq 0.01$$); model B (βsupermarkets = -0.33, $95\%$CI: -0.63, 0.03, $$P \leq 0.03$$); (βfruit and vegetable stores -0.37, $95\%$CI: -0.67, -0.08, $$P \leq 0.01$$); and model C (βsupermarkets = -0.35, $95\%$CI: -0.65, -0.06, $$P \leq 0.02$$); (βfruit and vegetable stores = -0.40, $95\%$CI: -0.69, -0.10, $$P \leq 0.01$$) (S2 Table).
When testing the association of the food environment and BMI whilst considering dietary patterns as a confounder, a higher density of convenience stores showed a statistically significant association with a higher risk of obesity (βconvenience stores = 0.06, $95\%$CI: 0.01, 0.12, $$P \leq 0.03$$) (S3 Table). No statistically significant findings were identified when stratifying results by sex (S4 Table).
## Role of SEP in dietary patterns and the retail food environment
To assess the influence of household SEP, Model A, B and C were considered and stratified by household SEP to understand if there were variations within the population. Also, the interaction between socioeconomic position and food retail density was tested. In addition, two-level multinomial logistic regressions with random effects were undertaken to assess the role of socioeconomic aspects of the environment and their influence on dietary patterns in urban CTAs of Mexico.
Multicollinearity was measured for each model by considering variance inflation factors. Variance inflation factors did not exceed the value of 4.0 for any of the included variables and were therefore all included in the models. Survey design and weights were accounted in all models and statistical analyses were undertaken in STATA 14 [27].
When assessing the relationship of fruit and vegetable store density and household SEP, a graded and inverse association was observed in which lower-income households had a reduced availability of fruit and vegetable stores, compared with higher-income households (β: -0.007, $95\%$ CI -0.0011, -0.004), $P \leq 0.001$) (Table 4). Furthermore, when stratifying the association of dietary patterns with fruit and vegetable store density by household SEP, upper-income households were more likely to consume healthy dietary patterns (β: 0.004, $95\%$ CI: 0.0004, 0.007, $$P \leq 0.027$$). Additionally, middle-upper-income households were less likely to consume unhealthy dietary patterns when exposed to high densities of fruit and vegetable stores (β: -0.003, $95\%$ CI: -0.006, -0.0002, $$P \leq 0.036$$) (Table 5).
A low fast-food outlet density was associated with lower-income (β: -0.061, $95\%$ CI: -0.072, -0.050, $P \leq 0.001$) and upper-lower-income households (β: -0.026, $95\%$ CI: -0.034, -0.018, $P \leq 0.001$). Similarly, a low density of restaurants and supermarkets was associated with low-income neighbourhoods, but a graded association showed that as income increased, a higher availability of restaurants and supermarkets became available to higher-income sectors of the population. Convenience stores were widely available in all socioeconomic strata (Table 4). No statistically significant associations were identified when testing the interaction between socioeconomic position and the density of each of the food outlets tested in this study.
Regarding the association of dietary patterns and food outlet density, stratified by household SEP, when exposed to a high density of fast-food outlets, middle-upper-income populations were associated with the consumption of a carbohydrates-and-drinks type of dietary (β: 0.013, $95\%$ CI 0.0008, 0.025, $$P \leq 0.036$$). Whereas middle-upper-income (β: -0.009, $95\%$ CI -0.019, -0.0003, $$P \leq 0.043$$) and lower-income populations (β: -0.026, $95\%$ CI 0.0002, 0.053, $$P \leq 0.036$$) were associated with an unhealthy dietary pattern (Table 5). Regarding restaurants, lower-income (β: 0.017, $95\%$ CI: 0.007, 0.027, $$P \leq 0.001$$) and upper-lower-income households (β: 0.005, $95\%$ CI: 0.002, 0.009, $$P \leq 0.005$$) were associated with a dietary pattern rich in carbohydrates and drinks whilst middle-upper-income (β: -0.002, $95\%$ CI: -0.005, 0.0003, $$P \leq 0.027$$) and upper-income households (β: -0.002, $95\%$ CI: -0.005, -0.0001, $$P \leq 0.038$$) were inversely associated with unhealthy dietary patterns when exposed to a high density of restaurants. When exposed to a high density of supermarkets, lower-income neighbourhoods (β: -0.184, $95\%$ CI: -0.29, -0.076, $$P \leq 0.001$$) were inversely associated with unhealthy dietary patterns whilst for upper-lower-income households there was an increased association with consuming an unhealthy dietary patterns (β: 0.108, $95\%$ CI: 0.014, 0.203, $$P \leq 0.025$$) and middle-income households were associated with a carbohydrates-and-drinks type of pattern (β: 0.095, $95\%$ CI: 0.013, 0.177, $$P \leq 0.023$$) (Table 5). In geographical areas with a high concentration of convenience stores, lower-income (β: 0.069, $95\%$ CI: 0.029, 0.011, $$P \leq 0.001$$) and upper-lower-income households (β: 0.023, $95\%$ CI: 0.003, 0.043, $$P \leq 0.025$$) were associated with unhealthy dietary patterns (Table 5).
## General characteristics of the population and the retail food environment
From the study sample, ($$n = 5$$,080), $56\%$ were female, $35\%$ ($$n = 2$$,824) had a sedentary lifestyle, $12\%$ ($$n = 585$$) owned a car and $14\%$ ($$n = 690$$) participated in a form of food programme (Table 2). Food programmes encompassed the Oportunidades programme ($$n = 395$$, $68\%$), a national conditional cash transfer program targeting poor and extremely poor households; Oportunidades school scholarships ($$n = 22$$, $4\%$); funding support for the elderly ($$n = 8$$, $1\%$); Progresa medical support ($$n = 21$$, $4\%$); monetary support from the Ayuda programme; milk supply programmes Liconsa or Conasupo ($$n = 46$$, $8\%$); food pantries from DIF ($$n = 24$$, $4\%$); food pantries from other organisations, social kitchens and canteens ($$n = 11$$, $2\%$); school breakfasts ($$n = 1$$, $0.2\%$); other educational scholarships ($$n = 13$$, $2\%$); non-governmental or civil organization ($$n = 1$$, $0.2\%$); other financial support for the elderly ($$n = 1$$, $0.2\%$); other ($$n = 38$$, $6\%$).
**Table 2**
| Variable | Total N (%) |
| --- | --- |
| Gender | |
| Men | 2,256 (44) |
| Women | 2,824 (56) |
| Age | |
| 18–24 | 640 (13) |
| 25–34 | 506 (10) |
| 35–44 | 628 (12) |
| 45–54 | 451 (9) |
| 55–64 | 336 (7) |
| 65+ | 323 (6) |
| Missing | 2,196 (43) |
| Physical activity | |
| Active | 401 (8) |
| Moderately active | 266 (5) |
| Inactive | 1,793 (35) |
| Missing | 2,620 (52) |
| Household SEPb | |
| Highest | 1,027 (22) |
| Second highest | 960 (21) |
| Middle | 938 (20) |
| Second lowest | 955 (20) |
| Lowest | 803 (17) |
| Missing | 397 (8) |
| Car ownership | |
| Owns a car | 585 (12) |
| Does not own a car | 2,295 (45) |
| Missing | 2,200 (43) |
| Region | |
| South | 1,729 (34) |
| North | 1,242 (24) |
| Centre | 1,769 (35) |
| Metropolitan area | 340 (7) |
| Area deprivation level | |
| Low | 3,041 (60) |
| High | 2,039 (40) |
| Urbanicity | |
| Rural (excluded) | 1,830 (36) |
| Urban | 1,025 (20) |
| Metropolitan | 2,225 (44) |
| Food programme | |
| Participated | 581 (11) |
| Did not participate | 2,303 (45) |
| Did not respond/did not know | 2,196 (43) |
| Health service | |
| Covered | 663 (13) |
| Not covered | 2,221 (44) |
| Did not respond/did not know | 2,196 (43) |
The assessment of the food environment in Mexico was undertaken as part of the first phase of this study and has been published and is described elsewhere [20]. To summarise, out of 72,892 CTAs in Mexico, only 10,145 CTAs ($14\%$) had access to a fruit and vegetable store. When looking at the distribution of food outlets in Mexico, a higher fruit and vegetable store concentration was evident in the centre and South of Mexico compared with the North and metropolitan regions of Mexico. The North of Mexico shows a very low availability of fruit and vegetable stores compared to the rest of the country. In addition, convenience stores and fast-food outlets were widely available throughout the country.
## Associations of dietary patterns and the retail food environment
Fruit and vegetable store density was inversely associated with unhealthy dietary patterns. This finding was replicated in the three statistical models that were tested (Models A, B and C) (Table 2). Restaurants were repeatedly and positively associated with the carbohydrates-and-drinks type of dietary pattern for models A (β: 0.003 $95\%$ CI: 0.0016, 0.005, $P \leq 0.001$), B (β: 0.004, $95\%$CI: 0.002, 0.005, $P \leq 0.001$) and C (β: 0.003 $95\%$CI: 0.001, 0.005, $P \leq 0.001$) and inversely associated with the unhealthy dietary pattern for models B and C. Fast-food outlets, supermarkets and convenience stores were not statistically significantly associated with any type of dietary pattern (Table 3).
**Table 3**
| Food outlet density (number of stores/CTA) | Factor 1 Healthy pattern | Factor 1 Healthy pattern.1 | Factor 2 Unhealthy pattern | Factor 2 Unhealthy pattern.1 | Factor 3 Carbohydrate & drinks pattern | Factor 3 Carbohydrate & drinks pattern.1 |
| --- | --- | --- | --- | --- | --- | --- |
| | β (95% CI) | P-value | β (95% CI) | P-value | β (95% CI) | P-value |
| Fruit and vegetable stores Model A | 0.002 (-0.0001, 0.005) | 0.064 | -0.002 (-0.004, -0.0002) | 0.029 | 0.001 (-0.001, 0.003) | 0.435 |
| Fruit and vegetable stores Model B | 0.002 (-0.0006, 0.004) | 0.131 | -0.003 (-0.005, -0.001) | <0.001 | 0.002 (-0.0006, 0.004) | 0.129 |
| Fruit and vegetable stores Model C | 0.002 (-0.0001, 0.005) | 0.060 | -0.002 (-0.004, -0.0003) | 0.025 | 0.001(-0.001, 0.003) | 0.397 |
| Fast-food outlets Model A | 0.001 (-0.005, 0.008) | 0.692 | -0.004 (-0.01, 0.001) | 0.127 | 0.006 (-0.007, 0.014) | 0.077 |
| Fast-food outlets Model B | 0.001 (-0.006, 0.009) | 0.734 | -0.004 (-0.01, -0.36) | 0.094 | 0.007 (-0.0001, 0.014) | 0.053 |
| Fast-food outlets Model C | 0.001 (-0.0062, 0.008) | 0.781 | -0.005 (-0.011, 0.0003) | 0.066 | 0.005 (-0.002, 0.012) | 0.141 |
| Restaurants- Model A | 0.0008 (-0.0008, 0.002) | 0.344 | -0.001 (-0.002, 0.00005) | 0.060 | 0.003 (0.0016, 0.005) | <0.001 |
| Restaurant Model B | 0.0006 (-0.001, 0.002) | 0.482 | -0.002 (-0.004, -0.001) | <0.001 | 0.004 (0.002, 0.005) | <0.001 |
| Restaurant Model C | 0.0007 (-0.0009, 0.002) | 0.404 | -0.001 (-0.003, -0.0002 | 0.027 | 0.003 (0.001, 0.005) | <0.001 |
| Supermarkets Model A | 0.014 (0.024, 0.053) | 0.478 | -0.03 (-0.057, 0.004) | 0.094 | 0.012 (-0.027, 0.052) | 0.541 |
| Supermarkets Model B | 0.01 (-0.031, 0.05) | 0.635 | -0.028 (-0.060, -0.003) | 0.081 | 0.020 (-0.020, 0.060) | 0.323 |
| Supermarkets Model C | 0.014 (-0.025, 0.053) | 0.479 | -0.028 (-0.059, 0.003) | 0.077 | 0.012 (-0.27, 0.522) | 0.533 |
| Convenience stores Model A | 0.0002 (-0.001, 0.001) | 0.797 | -0.0007 (-0.002, 0.0003) | 0.163 | -0.0007 (-0.002, 0.0007) | 0.327 |
| Convenience stores Model B | 0.0002 (-0.001, 0.001) | 0.792 | 0.005 (-0.003, 0.12) | 0.239 | 0.000007 (-0.001, 0.001) | 0.991 |
| Convenience stores Model C | 0.0002 (-0.001, 0.001) | 0.708 | -0.0009 (-0.002, 0.0002) | 0.112 | -0.0005 (-0.002, 0.0009) | 0.480 |
## Discussion
This study assessed 1) the dietary patterns from the Mexican population according to the ENSANUT survey; 2) the association between food outlet density and dietary patterns; 3) the association of the food environment and BMI a) considering diet as a confounder and b) considering the interaction of the food environment and diet, 4) the association and interaction of food outlet density and SEP and 5) the association of dietary patterns and SEP.
## Food outlet density and dietary pattern association
Fruit and vegetable store density in Mexico was inversely associated with an unhealthy dietary pattern in the Mexican population, which may be indicating that having a low availability of fruit and vegetable stores could influence unhealthy dietary patterns. Other studies that assessed fruit and vegetable store availability and dietary intake have observed similar findings [39, 40]. Ollberding et al. [ 41], identified that living in areas with a greater healthy food outlet access was associated with a higher mean intake of fruits and vegetables [41]. Another study by Menezes et al. [ 42], observed that the average consumption of fruit and vegetables was higher in neighbourhoods with higher-income and concentration of food stores, and better access to healthy foods. Additionally, fruit and vegetable consumption is low, particularly in low- and middle-income countries (LMIC) where affordability poses an important barrier [43]. Worldwide policies that enhance the availability and affordability of fruits and vegetables may be key to improve dietary intake [43].
Intervention studies that have focused on increasing fruit and vegetable intake have identified that affordability, palatability, and accessibility are some of the key factors that can influence healthy food selection such as fruits and vegetables. However, many interventions focused on increasing fruit and vegetable availability have mostly identified that fruit, but not vegetable consumption, has increased after interventions [44, 45]. Other studies suggest that greater spatial accessibility to food outlets comprising the local food environment may not guarantee fruit and vegetable consumption. This could be explained by a number of factors including high price, low quality and the availability of unhealthy alternatives including ultra-processed foods and beverages which tend to be more affordable, widely accessible and heavily marketed [46]. Therefore, in addition to the availability of fruit and vegetable stores which offer affordable and good quality foods, the restriction of unhealthy and ultra-processed foods to the population are important elements to consider when aiming to improve dietary intake.
Between 2012 and 2014, Mexico exported $3.8 billion a year worth of fruits and nuts; by 2015–17, Mexican fruit and nut exports increased over $55\%$ to $6 billion a year [47] whilst ultra-processed goods were imported to Mexico [48–50]. As the availability of food outlets selling mostly ultra-processed foods has increased, fruit and vegetable stores availability has not [20, 21]. Therefore, increasing accessibility to fruit and vegetable stores that offer a wide variety and high quality of fruits and vegetables along the regulation of unhealthy food offers to the local population could be an important factor to promote healthy food choices.
Our study also identified that restaurants were associated with a dietary pattern rich in carbohydrates and drinks. Complementary food and a high availability of sugar-sweetened beverages (SSBs) and alcoholic drinks, which are all high in simple carbohydrates are common in restaurant environments. Previous studies have identified that restaurant consumers may consume a higher level of carbohydrates [51], sugar-sweetened beverages [52] and alcohol [51] which relates to a higher caloric intake when eating out of home [51].
Previous interventions have focused on including nutritional information on restaurant menus [53]. However, replacing unhealthy appetizers and complementary foods with healthy alternatives (e.g., fruits, vegetables, pulses, nuts) and increasing the ratio of healthy vs unhealthy options may have a more effective impact on improving and facilitating healthier dietary patterns [53].
## BMI, diet, and food environment
When considering dietary patterns as confounders, we identified that a high density of convenience stores was associated with unhealthy dietary patterns and a higher BMI risk. This result aligns with the results identified on our previous study which showed a significant association between a higher density of convenience stores and a higher risk of obesity [20]. Although a higher availability of healthy food options (e.g., fruit and vegetables) and consumption of healthier dietary patterns may be important for the prevention of obesity and NCDs this does not seem to override the abundant availability of food outlets which offer unhealthy food options such as HFSS and ultra-processed foods. This coincides with findings from other studies which suggest that unhealthy food cues have a larger effect than healthier food. Thus, reducing the availability of unhealthy food outlets and removing or restricting less healthy food choices as opposed to only increasing the availability of healthier food options may have a greater impact on dietary intake and the prevention of obesity and diet related NCDs [54–57].
## Socioeconomic position, dietary patterns, and food outlet density
An inverse association was observed between SEP and fruit and vegetable stores and supermarket availability, indicating low-income households were more likely to inhabit areas with a lower access to fruit and vegetable stores and supermarkets. In addition, when exposed to areas with a high availability of fruit and vegetable stores, upper-income households were more likely to consume healthier dietary patterns whilst upper middle-income households were less likely to consume unhealthy dietary patterns.
Fruit and vegetable stores and supermarkets offer a greater availability and variety of healthy food choices (e.g., fruits and vegetables, whole grain alternatives and low-fat options) compared with fast-food outlets, restaurants, and convenience stores. Therefore, having a lower availability of food stores that offer healthy foods such as fruit and vegetable stores and supermarkets in deprived areas may be key barrier to access healthy and affordable foods. Similar findings have been reported in other middle-income countries such as Brazil, where it was observed that socially disadvantaged neighbourhoods had lower access to fruits and vegetables or were of a lower quality [58].
In contrast, convenience stores, which may offer more unhealthy food and beverage options, were widely available for lower-income populations as well as all other household strata. These type of food stores tend to offer a wide range of foods high in saturated fat, salt and sugar which may increase the risk of obesity and NCDs as shown by previous studies [20, 59].
Regarding fast-food outlet exposure and SEP, lower-income households were less exposed to high concentrations of fast-food outlets (β: -0.061 $95\%$CI: -0.072, -0.050; $P \leq 0.001$) and were less likely to consume unhealthy dietary patterns when exposed to high concentrations of fast-food outlets (β: -0.026, $95\%$ CI: 0.0002, 0.053; $$P \leq 0.04$$). This could be explained by the low level of development and commercial infrastructure to which lower-income populations may be exposed to [20, 60, 61]. Our previous study assessed the distribution of food outlets according to the level of urbanicity and it was observed that in areas of higher urbanicity, there was a higher availability of fast-food outlets [20]. Similarly, previous studies have shown that fast-food outlets tend to aggregate in the same geographic areas and locate in areas of greater visibility and connectivity, such as city centres, commercial areas and highstreets which may have a higher street intersection density, a higher availability of public transport and infrastructure that facilitates walkability and safe mobility (e.g., sidewalks, bike lanes, parks) [62]. This strategy helps fast food outlets dominate the market and maximise profits [62] which explains why fast-food outlets in lower-income neighbourhoods were less available for low-income households in this study. Previous studies have also observed that a high availability of fast‐food outlets increases the risk of obesity by discouraging healthy dietary behaviours along an high exposure to unhealthy food outlets which enable unhealthy food options [63].
Similarly, for restaurants, these were less likely to be found in lower-income areas compared with upper-income areas. However, upper-income populations were less likely to acquire unhealthy dietary patterns when highly exposed to unhealthy food outlets whereas lower-income populations who were exposed to fast-food outlets were more likely to consume a diet rich in carbohydrates. This could be explained by higher-income populations having more purchase power, options, and mobility (e.g., car ownership) to acquire food from a diversity of food outlets and may typically be able to travel further to access a greater diversity of food shops and healthier food choices [64, 65]. Supporting this finding, as indicated in the 2019 Pan American Health Organization report [19], ultra-processed foods purchases increased as available money increased. In Mexico, gross domestic product (GDP) per capita increased by $23\%$ between 2009 and 2014.
Supermarkets were less available for lower-income populations; however, when lower-income populations did have access to supermarkets, a lower likelihood of an unhealthy diet was observed. This could be explained by an increased access to staple foods at more affordable prices and an increased availability of fruits and vegetables. Previous studies have observed that lower-income areas are less likely to have access to supermarkets and grocery stores that carry healthy foods compared with predominantly middle- and higher-income areas [16, 66, 67]. However, when exposed to a high density of supermarkets, upper-lower-income populations were more likely to have an unhealthy dietary pattern. This may be explained by a higher purchase power and an increased exposure to purchase snacks and additional discretionary foods [68]. Previous studies have found a correlation between an increase in purchase power unhealthy food purchases in Latin American countries [68].
Middle- and low-income populations may be more vulnerable to unhealthy foods and beverages commonly displayed at cash points which previous research has observed to incentive unhealthy food choices. Point of purchase policy regulations may target smaller portion sizes for unhealthy foods and beverages; unhealthy combo meal-like promotions; and food choice architecture (e.g. placement and marketing) restrictions on HFSS and ultra-processed foods and increased saliency for healthier alternatives [69]. Adoption of these type of policies could contribute significantly to the prevention of obesity and diet-related NCDs [69].
Our findings indicate that the food environment might explain some of the socioeconomic disparities related with dietary intake. Supporting these finding, a study by Pérez-Ferrer et al. indicated that household wealth can be an effect modifier in the association between education and obesity, mainly in women [70]. The study also indicated that as countries like Mexico develop economically, there tends to be a cross-over to higher rates of obesity among socially disadvantaged groups [70]. As our study indicates, a lack of healthy food store availability for the most deprived may increase the susceptibility to unhealthy diets which may increase the risk of obesity and NCDs.
In Mexico, the nutrition transition and an increase in obesity coincided with the NAFTA agreement [50, 71], which may be related to growth in unhealthy food retail outlets [50]. Our findings indicate that environmental and sociodemographic conditions within neighbourhoods may affect dietary behaviours [58]. Additionally, household income can influence access to food products and food stores, thereby making it difficult for low-income families to prioritize the purchase healthy foods, especially when these are more expensive or not as appetising [58].
## Strengths and limitations
The findings presented here should be interpreted with caution. Among the limitations of our study is the use of cross-sectional data, which does not permit examination of how changes in fruit and vegetable store density influenced consumption. Retail food environments are dynamic, and a longitudinal design could help understand the effect of the retail food environment on dietary intake. Similarly, a natural experimental evaluation could have minimized potential bias and provide critical information about the impacts of food retail interventions on dietary intake [72] or obesity [73]. There was a two-year difference between the health and geographic data that were used in this study. However, data-verification two years later found the prevalence, position and type of food store was still accurate, suggesting little change over time. Additionally, due to data confidentiality, dietary intake data were recorded at the centroid of the residents’ CTA as precise individual level area data were not available. CTA was used to calculate food outlet density as a proxy for individual’s food environment. CTA has been considered a gold standard for measuring food environment and has been used by various studies as the unit of analysis to study food environments [74, 75]. However, individuals often cross the boundaries of their residential area to access food which may underestimate food availability [74]. Additionally, residents in impoverished areas may have limited capital resources, such as car ownership, making it feasible to assume that there may be greater reliance on proximal food sources [76] or low-income populations may travel long distances for work during the day and thus shift across food environments. Additionally, even if healthier food options are available, food affordability may be a barrier if working or commuting via high-income areas. Missing data could have impacted the association of the food environment, diet, obesity, and SEP. ENSANUT carried out a dietary assessment to $11\%$ of the population [26]; therefore, the study’s sample size was restricted to participants who had a measure of dietary intake and the retail food environment. Individuals who did not have a measure of diet, lived in a rural area, or did not have a measure of the retail food environment were excluded from this study.
Although informal food vendors were identified in the food outlet data verification stage, it is possible that some mobile food units, which represent an important influence on dietary intake in Mexico [77], may not have been included in the food outlet database used in this study, which could have been due to lack of registration compliance to sell food or because of the provision of a home address rather than the place of sale. A lack of data on the affordability and quality of fruit and vegetables sold and the availability of unhealthy alternatives is a limitation of this study. To account for this limitation, household SEP and dietary intake in the form of dietary patterns were included in the statistical models. Lastly, due to the self-reporting nature of FFQ questionnaires to assess dietary intake, bias may occur due to underreporting of food consumption or low accuracy (recall bias). In addition, because an FFQ is composed of a specific list of food items, a single FFQ may not reflect the consumption patterns of a given population [78].
In terms of strengths, this study included a comprehensive dietary assessment which permitted examination of the diet through dietary patterns. Dietary patterns may be a better predictor of dietary intake than individual nutrient analysis because they consider the overall diet consumption and account for the interaction of nutrients within foods and their potential effect on health [79]. Additionally, we used principal component and factor analysis to determine dietary patterns, which is the most common approach to determine dietary patterns and therefore allows comparison with other studies. Instead of classifying individuals into a single pattern, individuals receive a score of each pattern instead of being subjectively classified into a single cluster or group [79]. Other strengths include the use of measured data; the geographical location verification of food outlets; accounting for selection bias; and the use of a statistical method that was able to detect discrepancies between different geographical levels and account for clustering. To our knowledge, this is the only study to has assessed the relationship between food retailer availability, dietary patterns, BMI and its interaction with diet, and socioeconomic position at a national level in a middle-income country setting. This study advances the existing literature of the retail food environment and its relationship with food choice and the application of geographical information systems.
## Conclusions
A lower density of fruit and vegetable stores was associated with unhealthier dietary patterns whilst a high exposure to convenience stores was linked with unhealthy dietary patterns and a higher risk of BMI particularly in low-income populations. Actions to regulate the food environment by decreasing the availability of unhealthy foods and increasing the availability of healthy food and beverage options, particularly in low-income neighbourhoods, may help improve dietary intake in the Mexican population and contribute to the prevention of obesity and diet related NCDs in the country.
## References
1. Butland B, Jebb S, Kopelman P, McPherson K, Thomas S, Mardell J. *FORESIGHT—Tackling Obesities: future Choices—Project Report* (2007.0)
2. 2EUFIC. The determinants of food choice Belgium: European Food Information Council; 2005 [cited 2015 22/08/2015]. Available from: http://www.eufic.org/article/en/expid/review-food-choice/.. *The determinants of food choice Belgium* (2005.0)
3. Branca F, Lartey A, Oenema S, Aguayo V, Stordalen GA, Richardson R. **Transforming the food system to fight non-communicable diseases**. *BMJ* (2019.0) **364** l296. DOI: 10.1136/bmj.l296
4. Vepsäläinen H, Mikkilä V, Erkkola M, Broyles ST, Chaput JP, Hu G. **Association between home and school food environments and dietary patterns among 9-11-year-old children in 12 countries.**. *Int J Obes Suppl* (2015.0) **5** S66-73. DOI: 10.1038/ijosup.2015.22
5. Stafford M, Cummins S, Ellaway A, Sacker A, Wiggins RD, Macintyre S. **Pathways to obesity: Identifying local, modifiable determinants of physical activity and diet.**. *Soc Sci Med* (2007.0) **65** 1882-97. DOI: 10.1016/j.socscimed.2007.05.042
6. Glanz K, Sallis JF, Saelens BE, Frank LD. **Healthy Nutrition Environments: Concepts and Measures.**. *American Journal of Health Promotion* (2005.0) **19** 330-3. DOI: 10.4278/0890-1171-19.5.330
7. Swinburn B, Kraak V, Rutter H, Vandevijvere S, Lobstein T, Sacks G. **Strengthening of accountability systems to create healthy food environments and reduce global obesity**. *Lancet* (2015.0) **385** 2534-45. DOI: 10.1016/S0140-6736(14)61747-5
8. 8Office of the United States Trade Representative. North American Free Trade Agreement (NAFTA) US2017 [cited 2020 2 March]. Available from: https://ustr.gov/trade-agreements/free-trade-agreements/north-american-free-trade-agreement-nafta.
9. Friel S, Hattersley L, Snowdon W, Thow AM, Lobstein T, Sanders D. **Monitoring the impacts of trade agreements on food environments.**. *Obes Rev.* (2013.0) **14** 120-34. DOI: 10.1111/obr.12081
10. Imamura F, Micha R, Khatibzadeh S, Fahimi S, Shi P, Powles J. **Dietary quality among men and women in 187 countries in 1990 and 2010: a systematic assessment**. *The Lancet Global Health* (2015.0) **3** e132-e42. DOI: 10.1016/S2214-109X(14)70381-X
11. Hawkes C.. **Uneven dietary development: linking the policies and processes of globalization with the nutrition transition, obesity and diet-related chronic diseases.**. *Global Health.* (2006.0) **2** 4. DOI: 10.1186/1744-8603-2-4
12. 12INSP. Encuesta Nacional de Salud y Nutrición 2018 [2018 Health and Nutrition National Survey].
Mexico: Instituto Nacional de Salúd Pública [National Health Insitute], 2018.. *Encuesta Nacional de Salud y Nutrición 2018 [2018 Health and Nutrition National Survey].* (2018.0)
13. Colchero MA, Molina M, Guerrero-López CM. **After Mexico Implemented a Tax, Purchases of Sugar-Sweetened Beverages Decreased and Water Increased: Difference by Place of Residence, Household Composition, and Income Level**. *The Journal of nutrition* (2017.0) **147** 1552-7. DOI: 10.3945/jn.117.251892
14. 14PAHO. Alimentos y bebidas ultraprocesados en América Latina: ventas, fuentes, perfiles de nutrientes e implicaciones normativas [Ultraprocessed foods and beverages in Latin America: sales, sources, nutritional profiles and normative implications].
Washington, DC: PAHO, 2019.. *Alimentos y bebidas ultraprocesados en América Latina: ventas, fuentes, perfiles de nutrientes e implicaciones normativas [Ultraprocessed foods and beverages in Latin America: sales, sources, nutritional profiles and normative implications].* (2019.0)
15. 15INSP. Aspectos económicos relacionados con un impuesto al refresco en México [Economic aspects related to the soda tax in Mexico]. Mexico: 2013.
16. 16Institute of Medicine (US) and National Research Council (US) Committee on Childhood Obesity. Local Government Actions to Prevent Childhood Obesity Washington D.C.: National Academies Press; 2009 [cited 2022 February]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK219682/.
17. Institute OD. *The rising cost of a healthy diet—Changing relative prices of foods in high-income and emerging economies* (2015.0)
18. Marrón-Ponce JA, Tolentino-Mayo L, Hernández-F M, Batis C. **Trends in Ultra-Processed Food Purchases from 1984 to 2016 in Mexican Households.**. *Nutrients* (2018.0) **11** 45. DOI: 10.3390/nu11010045
19. 19Pan American Health Organization. Ultra-processed food and drink products in Latin America: Sales, sources, nutrient profiles and policy implications.
Washington, D. C.: PAHO, 2019.. *Ultra-processed food and drink products in Latin America: Sales, sources, nutrient profiles and policy implications.* (2019.0)
20. Pineda E, Brunner EJ, Llewellyn CH, Mindell JS. **The retail food environment and its association with body mass index in Mexico.**. *Int J Obes (Lond).* (2021.0) **45** 1215-28. DOI: 10.1038/s41366-021-00760-2
21. Pérez-Ferrer C, Auchincloss AH, Barrientos-Gutierrez T, Colchero MA, de Oliveira Cardoso L, Carvalho de Menezes M. **Longitudinal changes in the retail food environment in Mexico and their association with diabetes**. *Health Place* (2020.0) **66** 102461. DOI: 10.1016/j.healthplace.2020.102461
22. Shaw SC, Ntani G, Baird J, Vogel CA. **A systematic review of the influences of food store product placement on dietary-related outcomes.**. *Nutr Rev* (2020.0) **78** 1030-45. DOI: 10.1093/nutrit/nuaa024
23. Krukowski RA, West DS, Harvey-Berino J, Elaine Prewitt T. **Neighborhood impact on healthy food availability and pricing in food stores.**. *J Community Health* (2010.0) **35** 315-20. DOI: 10.1007/s10900-010-9224-y
24. Drewnowski A, Moudon AV, Jiao J, Aggarwal A, Charreire H, Chaix B. **Food environment and socioeconomic status influence obesity rates in Seattle and in Paris.**. *Int J Obes (Lond).* (2014.0) **38** 306-14. DOI: 10.1038/ijo.2013.97
25. 25INSP. National Health and Nutrition Survey [Encuesta Nacional de Salud y Nutrición].
Mexico: INSP, 2012.. *National Health and Nutrition Survey [Encuesta Nacional de Salud y Nutrición].* (2012.0)
26. 26INSP. ENSANUT 2012 Encuesta Nacional de Salud y Nutrición 2012. Mexico: INSP, 2012.. *ENSANUT 2012 Encuesta Nacional de Salud y Nutrición 2012* (2012.0)
27. 27STATA. Your data tell a story 2022 [cited 2022 June]. Available from: https://www.stata.com/.
28. 28World Health O. Recomendaciones mundiales sobre actividad física para la salud.
Ginebra: Organización Mundial de la Salud; 2010.. *Recomendaciones mundiales sobre actividad física para la salud.* (2010.0)
29. 29INEGI. Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH) [National Income and Expenditure Survey 2010]
Mexico: Instituto Nacional de Estadística y Geografía [National Institute of Statistics and Geography]; 2010 [cited 2022]. Available from: https://www.inegi.org.mx/rnm/index.php/catalog/36.. *Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH) [National Income and Expenditure Survey 2010]* (2010.0)
30. Denova-Gutiérrez E., Ramírez-Silva I., Rodríguez-Ramírez S., Jiménez-Aguilar A., Shamah-Levy T., R-D J.A.. *Validity of a food frequency questionnaire to assess food intake in Mexican adolescent and adult population* (2016.0)
31. 31INEGI. Censos económicos 2014
Mexico: Instituto Nacional de Estadística y Geografía; 2017 [cited 2017 25/Septebmer]. Available from: http://www.inegi.org.mx/.. *Censos económicos 2014* (2017.0)
32. Caspi CE, Friebur R. **Modified ground-truthing: an accurate and cost-effective food environment validation method for town and rural areas.**. *The international journal of behavioral nutrition and physical activity* (2016.0) **13** 37. DOI: 10.1186/s12966-016-0360-3
33. Pérez-Tepayo S, Rodríguez-Ramírez S, Unar-Munguía M, Shamah-Levy T. **Trends in the dietary patterns of Mexican adults by sociodemographic characteristics.**. *Nutr J.* (2020.0) **19** 51. DOI: 10.1186/s12937-020-00568-2
34. Valerino-Perea S, Lara-Castor L, Armstrong MEG, Papadaki A. **Definition of the Traditional Mexican Diet and Its Role in Health: A Systematic Review.**. *Nutrients* (2019.0) **11**. DOI: 10.3390/nu11112803
35. Cunha DB, Almeida RM, Pereira RA. **A comparison of three statistical methods applied in the identification of eating patterns.**. *Cadernos de saude publica.* (2010.0) **26** 2138-48. DOI: 10.1590/s0102-311x2010001100015
36. Consonni G, Leucari V. **Model Determination for Directed Acyclic Graphs**. *Journal of the Royal Statistical Society Series D (The Statistician).* (2001.0) **50** 243-56
37. CONAPO S, SEGO B. *Catálogo Sistema Urbano Nacional 2012* (2012.0)
38. Hair JF. *Multivariate data analysis* (1998.0)
39. Franco M, Diez-Roux AV, Nettleton JA, Lazo M, Brancati F, Caballero B. **Availability of healthy foods and dietary patterns: the Multi-Ethnic Study of Atherosclerosis**. *The American Journal of Clinical Nutrition* (2009.0) **89** 897-904. DOI: 10.3945/ajcn.2008.26434
40. Pitts SBJ, Gustafson A, Wu Q, Mayo ML, Ward RK, McGuirt JT. **Farmers’ market use is associated with fruit and vegetable consumption in diverse southern rural communities.**. *Nutr J* (2014.0) **13** 1. DOI: 10.1186/1475-2891-13-1
41. Ollberding NJ, Nigg CR, Geller KS, Horwath CC, Motl RW, Dishman RK. **Food outlet accessibility and fruit and vegetable consumption.**. *Am J Health Promot* (2012.0) **26** 366-70. DOI: 10.4278/ajhp.101215-ARB-401
42. Menezes MC, Costa BVL, Oliveira CDL, Lopes ACS. **Local food environment and fruit and vegetable consumption: An ecological study.**. *Preventive Medicine Reports* (2017.0) **5** 13-20. DOI: 10.1016/j.pmedr.2016.10.015
43. Miller V, Yusuf S, Chow CK, Dehghan M, Corsi DJ, Lock K. **Availability, affordability, and consumption of fruits and vegetables in 18 countries across income levels: findings from the Prospective Urban Rural Epidemiology (PURE) study.**. *The Lancet Global Health* (2016.0) **4** e695-e703. DOI: 10.1016/S2214-109X(16)30186-3
44. Appleton KM, Hemingway A, Saulais L, Dinnella C, Monteleone E, Depezay L. **Increasing vegetable intakes: rationale and systematic review of published interventions**. *Eur J Nutr* (2016.0) **55** 869-96. DOI: 10.1007/s00394-015-1130-8
45. Perry CL, Bishop DB, Taylor G, Murray DM, Mays RW, Dudovitz BS. **Changing fruit and vegetable consumption among children: the 5-a-Day Power Plus program in St. Paul, Minnesota.**. *Am J Public Health* (1998.0) **88** 603-9. DOI: 10.2105/ajph.88.4.603
46. Popkin BM, Adair LS, Ng SW. **Global nutrition transition and the pandemic of obesity in developing countries.**. *Nutr Rev* (2012.0) **70** 3-21. DOI: 10.1111/j.1753-4887.2011.00456.x
47. 47Dialogue M. Rising Fruit and Vegetables Imports from Mexico: UC Davis; 2018 [cited 2020 September]. Available from: https://migration.ucdavis.edu/.
48. 48USDA. Fruit and vegetables USA: USDA; 2020. Available from: https://www.ers.usda.gov/data-products/food-availability-per-capita-data-system/.
49. Mendoza-Cano O, Sánchez-Piña RA, González-Ibarra Á J, Murillo-Zamora E, Nava-Garibaldi CM. **Health Impacts from Corn Production Pre-and Post-NAFTA Trade Agreement (1986–2013).**. *Int J Environ Res Public Health* (2016.0) **13**. DOI: 10.3390/ijerph13070709
50. Clark SE, Hawkes C, Murphy SM, Hansen-Kuhn KA, Wallinga D. **Exporting obesity: US farm and trade policy and the transformation of the Mexican consumer food environment.**. *International journal of occupational and environmental health* (2012.0) **18** 53-65. DOI: 10.1179/1077352512Z.0000000007
51. Orfanos P, Naska A, Rodrigues S, Lopes C, Freisling H, Rohrmann S. **Eating at restaurants, at work or at home. Is there a difference? A study among adults of 11 European countries in the context of the HECTOR* project.**. *Eur J Clin Nutr* (2017.0) **71** 407-19. DOI: 10.1038/ejcn.2016.219
52. Ayala GX, Rogers M, Arredondo EM, Campbell NR, Baquero B, Duerksen SC. **Away-from-home food intake and risk for obesity: examining the influence of context.**. *Obesity (Silver Spring, Md).* (2008.0) **16** 1002-8. DOI: 10.1038/oby.2008.34
53. Mandracchia F, Tarro L, Llauradó E, Valls RM, Solà R. **Interventions to Promote Healthy Meals in Full-Service Restaurants and Canteens: A Systematic Review and Meta-Analysis.**. *Nutrients* (2021.0) **13** 1350. DOI: 10.3390/nu13041350
54. Pechey R, Marteau TM. **Availability of healthier vs. less healthy food and food choice: an online experiment.**. *BMC Public Health* (2018.0) **18** 1296. DOI: 10.1186/s12889-018-6112-3
55. Rose D, Hutchinson PL, Bodor JN, Swalm CM, Farley TA, Cohen DA. **Neighborhood Food Environments and Body Mass Index: The Importance of In-Store Contents.**. *Am J Prev Med.* (2009.0) **37** 214-9. DOI: 10.1016/j.amepre.2009.04.024
56. Nakamura R, Suhrcke M, Jebb SA, Pechey R, Almiron-Roig E, Marteau TM. **Price promotions on healthier compared with less healthy foods: a hierarchical regression analysis of the impact on sales and social patterning of responses to promotions in Great Britain**. *The American Journal of Clinical Nutrition* (2015.0) **101** 808-16. DOI: 10.3945/ajcn.114.094227
57. Talukdar D, Lindsey C. **To Buy or Not to Buy: Consumers’ Demand Response Patterns for Healthy versus Unhealthy Food**. *Journal of Marketing* (2013.0) **77** 124-38. DOI: 10.1509/jm.11.0222
58. Pessoa MC, Mendes LL, Gomes CS, Martins PA, Velasquez-Melendez G. **Food environment and fruit and vegetable intake in a urban population: a multilevel analysis.**. *BMC Public Health* (2015.0) **15** 1012. DOI: 10.1186/s12889-015-2277-1
59. Xin J, Zhao L, Wu T, Zhang L, Li Y, Xue H. **Association between access to convenience stores and childhood obesity: A systematic review.**. *Obes Rev* (2021.0) **22** e12908. DOI: 10.1111/obr.12908
60. Ziccardi A. **Poverty and urban inequality: the case of Mexico City metropolitan region*.**. *International Social Science Journal* (2014.0) **65** 205-19. DOI: 10.1111/issj.12070
61. Capron G, Gonzalez Arellano S, Wigle J, Diez AL, Monterrubio A, Castro J. *The urban food system of Mexico City* (2017.0)
62. Thomadsen R.. **Product Positioning and Competition: The Role of Location in the Fast Food Industry.**. *Marketing Science* (2007.0) **26** 792-804
63. Jia P, Luo M, Li Y, Zheng JS, Xiao Q, Luo J. **Fast‐food restaurant, unhealthy eating, and childhood obesity: A systematic review and meta‐analysis.**. *Obes Rev.* (2019.0) **22**. DOI: 10.1111/obr.12944
64. Pechey R, Monsivais P. **Socioeconomic inequalities in the healthiness of food choices: Exploring the contributions of food expenditures.**. *Prev Med.* (2016.0) **88** 203-9. DOI: 10.1016/j.ypmed.2016.04.012
65. Gustat J, O’Malley K, Luckett BG, Johnson CC. **Fresh produce consumption and the association between frequency of food shopping, car access, and distance to supermarkets.**. *Preventive Medicine Reports.* (2015.0) **2** 47-52. DOI: 10.1016/j.pmedr.2014.12.009
66. Baker EA, Schootman M, Barnidge E, Kelly C. **The role of race and poverty in access to foods that enable individuals to adhere to dietary guidelines.**. *Prev Chronic Dis.* (2006.0) **3** A76. PMID: 16776877
67. Franco M, Diez Roux AV, Glass TA, Caballero B, Brancati FL. **Neighborhood characteristics and availability of healthy foods in Baltimore.**. *Am J Prev Med* (2008.0) **35** 561-7. DOI: 10.1016/j.amepre.2008.07.003
68. Popkin BM, Reardon T. **Obesity and the food system transformation in Latin America.**. *Obes Rev.* (2018.0) **19** 1028-64. DOI: 10.1111/obr.12694
69. Cohen DA, Lesser LI. **Obesity prevention at the point of purchase**. *Obesity reviews: an official journal of the International Association for the Study of Obesity* (2016.0) **17** 389-96. DOI: 10.1111/obr.12387
70. Perez-Ferrer C, McMunn A, Zaninotto P, Brunner EJ. **The nutrition transition in Mexico 1988–2016: the role of wealth in the social patterning of obesity by education.**. *Public Health Nutr* (2018.0) **21** 2394-401. DOI: 10.1017/S1368980018001167
71. Siegel AD. **NAFTA Largely Responsible for the Obesity Epidemic in Mexico.**. *Washington University Journal of Law and Policy* (2016.0) **50** 195-226
72. Sadler RC, Gilliland JA, Arku G. **A food retail-based intervention on food security and consumption**. *International journal of environmental research and public health* (2013.0) **10** 3325-46. DOI: 10.3390/ijerph10083325
73. Taillie LS, Grummon AH, Fleischhacker S, Grigsby-Toussaint DS, Leone L, Caspi CE. **Best practices for using natural experiments to evaluate retail food and beverage policies and interventions.**. *Nutr Rev* (2017.0) **75** 971-89. DOI: 10.1093/nutrit/nux051
74. Charreire H, Casey R, Salze P, Simon C, Chaix B, Banos A. **Measuring the food environment using geographical information systems: a methodological review.**. *Public Health Nutr.* (2010.0) **13** 1773-85. DOI: 10.1017/S1368980010000753
75. Ma X, Battersby SE, Bell BA, Hibbert JD, Barnes TL, Liese AD. **Variation in low food access areas due to data source inaccuracies.**. *Appl Geogr.* (2013.0) **45** 10. DOI: 10.1016/j.apgeog.2013.08.014
76. Lee RE, Heinrich KM, Medina AV, Regan GR, Reese-Smith JY, Jokura Y. **A picture of the healthful food environment in two diverse urban cities.**. *Environ Health Insights* (2010.0) **4** 49-60. DOI: 10.4137/ehi.s3594
77. Long-Solís J.. **A Survey of Street Foods in Mexico City**. *Food and Foodways* (2007.0) **15** 213-36. DOI: 10.1080/07409710701620136
78. Shim JS, Oh K, Kim HC. **Dietary assessment methods in epidemiologic studies.**. *Epidemiol Health.* (2014.0) **36** e2014009. DOI: 10.4178/epih/e2014009
79. Schulze MB, Martínez-González MA, Fung TT, Lichtenstein AH, Forouhi NG. **Food based dietary patterns and chronic disease prevention**. *BMJ* (2018.0) **361** k2396. DOI: 10.1136/bmj.k2396
|
---
title: Exploring health-seeking behavior for non-communicable chronic conditions in
northern Bangladesh
authors:
- Fatema Binte Rasul
- Malabika Sarker
- Farzana Yasmin
- Manuela De Allegri
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022368
doi: 10.1371/journal.pgph.0000497
license: CC BY 4.0
---
# Exploring health-seeking behavior for non-communicable chronic conditions in northern Bangladesh
## Abstract
Non-communicable Diseases (NCDs) account for $67\%$ of total deaths in Bangladesh. However, the Bangladeshi health system is inadequately prepared to tackle NCDs. Evidence on NCD-specific health-seeking behavior can help appropriately address the needs of people affected by NCDs in Bangladesh. Our study aims to explore health-seeking behavior for people affected by NCDs in northern Bangladesh. We conducted a qualitative study in Mithapukur, Rangpur, during 2015–2016. We purposely selected respondents and carried out 25 in-depth interviews with individuals affected by non-communicable diseases and 21 healthcare providers. Additionally, we held six focus group discussions in the wider community. We verbatim transcribed all interviews and analyzed the content using thematic analysis, according to the following thematic areas: individual, household, and contextual factors that influence health-seeking behavior for NCDs within the context of the broader socio-economic environment. Study findings indicate that people seek care only when symptoms disrupt their daily lifestyle. Henceforth, people’s health beliefs, religious beliefs, and relations with local providers direct their actions, keeping provider accessibility, cost anticipation, and satisfying provider-encounters in mind. Health-seeking is predominantly delayed and fragmented. Semi-qualified providers represent a popular first choice. Gender roles dominate health-seeking behavior as women need their guardian’s permission to avail care. Our findings indicate the need to sensitize people about the importance of early health-seeking for NCDs, and continuing life-long NCD treatment. Our findings also highlight the need for people-centered care, making preventive and curative NCD services accessible at grassroots level, along with relevant provider training. Furthermore, special provisions, such as financial support and outreach programs are needed to enable access to NCD care for women and the poor.
## Introduction
Globally, Non-Communicable Diseases (NCDs) are on the rise. In 2019, NCDs accounted for 1.62 billion Disability Adjusted Life Years (DALYs), amounting to $63.8\%$ of total DALYs [1]. The Global Burden of Disease Study 2015 stated that worldwide, NCDs led to roughly 40 out of the total 56 million deaths that year, equivalent to $70\%$ of total deaths [2]. In 2012, the World Health Organization (WHO) warned that $68\%$ of the world’s mortality was attributable to NCDs, and about $75\%$ of total NCD deaths occurred in low- and middle-income countries (LMICs), amounting to 28 million deaths. Worse still, $82\%$ of the total premature deaths attributable to NCDs (deaths before turning 70 years) took place in LMICs [3]. The DALYs for NCDs in LMICs have increased from $34\%$ to $52\%$ over the last decade [4].
NCDs affect health, and economically debilitate countries by affecting the young population’s productivity [5]. By 2030, NCDs are predicted to cause a collective loss of $47 trillion [6]. If NCDs are not controlled, LMICs are headed towards an economic loss surpassing 7 trillion US dollars (USD) spanning 2011 to 2025, around 500 billion USD per year [6]. In Southeast Asia, NCDs are the prime killer, accounting for $62\%$ of all deaths, causing around 8.5 million deaths annually [7]. Almost every other NCD death ($48\%$ of total NCD deaths) in this region is premature [7], an amount higher than any other WHO regions [8]. As NCDs are causing both a health burden and an economic burden worldwide, they are explicitly addressed in the Sustainable Development Goal (SDG) 3, which aims at decreasing by at least one-third NCD-induced premature mortality via prevention and treatment [9].
In Bangladesh, $67\%$ of total deaths are attributable to NCDs [10]. The Bangladeshi health system is inadequately prepared to handle the challenges of the NCD epidemic, while still struggling with maternal and communicable conditions [11]. A key reason for such inadequacy is the persistence of a policy implementation gap. Latest NCD action plans and strategies include the “Strategic Plan for Surveillance and Prevention of NCDs in Bangladesh” (2011–2015) [11], and the “Multi-Sectoral Action Plan for Prevention and Control of NCDs 2018–2025” [12]. In the current Health, Population and Nutrition Sector Program 2017–2022 (HPNSP), Bangladesh has increased budgets for non-communicable disease control operation plans [12]. However, concrete strategies to translate the ambitions outlined in the aforementioned policies into everyday practice are absent [11]. For example, in Bangladesh, only 30 out of 209 listed essential drugs are for NCDs, many of which are rarely prescribed in practice [13]. In the public sector, only tertiary facilities are equipped to offer NCD services. Alternatively, NCD services are available in the private and in the NGO sector. NCD service provision in *Bangladesh is* inadequate due to limited health insurance coverage, inadequate resources and insufficient trained personnel in government health facilities [11,13].
While evidence on NCD epidemiology is steadily increasing across LMICs, little is known about people’s health-seeking practices [14]. In the Asian context, health-seeking is often a household decision, influenced by factors such as socio-demographics, convenience, service price, position of women in the family, type of illness, perceived quality of care and accessibility [15]. In 2010, a literature review on Asian health-seeking behavior inferred that most studies have been done quantitatively, failing to investigate the actual decision-making process [16]. This review explicitly called for the application of qualitative and mixed methods to explore health-seeking pathways and explain relevant decision-making processes, considering the broader socio-economic and cultural context [16].
In Bangladesh, only a handful of studies have explored factors affecting demand for NCD care [17,18] or adherence to NCD treatment [19]. One study found hypertension to be the most common self-reported condition in central and southern Bangladesh, detecting an increased likelihood of people from higher socio-economic statuses to be diagnosed, and men to be more likely to seek care from qualified professionals [17]. A second study approached health-seeking behaviors related to NCDs qualitatively, looking at fatal NCDs in women of reproductive age [20]. It found that most women first sought care from semi-qualified providers. Qualified care was sought at a later stage because of the high cost of services, distance, familiarity with semi-qualified providers, and not acknowledging illness severity [20]. A third study adopted a mixed-method approach to explore hypertension and COPD (Chronic Obstructive Pulmonary Disease). Findings indicate that these two conditions impose a financial burden on households and therefore, self-treatment is common in the initial stages of illness. While maintaining a focus on the financial implications of NCDs, this study also revealed a higher propensity to seek formal care among urban residents and suggested a lack of knowledge of care options among rural residents. Providers interviewed explained that people delay seeking care until their NCD deteriorates, largely due to financial barriers and a lack of understanding of the long-term health consequences of NCDs [18].
Our qualitative study aims to fill this gap in knowledge by looking in detail at the health-seeking behavior of people affected by NCDs.
## Study aim, design and setting
This qualitative study represents a follow-up to a prior quantitative study that assessed determinants of health-seeking and its related expenditure [21]. Albeit being partially informed by the prior survey [21], this study is exploratory in nature, hence it aims at openly exploring all factors shaping decisions regarding NCD health-seeking behavior in northern Bangladesh, rather than explaining results from the quantitative assessment. Health seeking behavior refers to the series of actions that people undertake with the aim of curing ill-health as per their perception [22]. For this study, we have focused on pathways to care in response to perceived illness [23].
Our study was conducted in the *Mithapukur upazila* (including one urban and one rural union—Durgapur and Mirzapur) in Rangpur district, northwest Bangladesh. Rangpur is one of the poorest regions in country, with a low literacy rate [24]. We purposely selected the *Mithapukur upazila* as the study site for our qualitative study, because our prior quantitative assessment [21] had revealed a higher reporting of NCDs in Mithapukur compared to other upazillas, hence facilitating respondent identification and sampling.
Healthcare provision in Rangpur reflects the medically pluralistic context of Bangladesh, and as such, relies on a mixture of public, private and Non-Government Organization (NGO) providers. The public sector has six tiers: ward, union, upazila (subdistrict), district, divisional and national. Primary care facilities are at the ward, union and upazila level, where NCD services are in initial stages of development [25]. At the ward and union level, primary care is provided by outlets known as community clinics and union sub-centers, providing basic screening and referral for NCDs [26]. At sub-district level, NCD corners have been established recently (after this study took place). Medical doctors are available, but can do little in absence of resources for NCD diagnosis and management, and systematic guidelines to offer NCD services [27]. The district level tier provides secondary care through the district hospital. It offers limited diagnostic and curative NCD services [11]. Tertiary care is provided at divisional and national level, through medical college hospitals, post-graduate medical hospitals and specialized health institutes [28]. Overall, there is no streamlined NCD care and no standard preventive and referral protocols in place for NCDs in the public facilities [11]. The 2014 Bangladesh Health Facility Survey reported that only $20\%$ of the country’s public health facility could provide adequate comprehensive NCD services (assessment, prescription, medications and diagnostics), with tertiary facilities being the prime providers [29].
The study area Mithapukur, primarily reliant on subsistence and cash-crop farming, has a population of 527,457 [30]. Nearly half of its population ($45\%$) falls under the national poverty line [31]. The community clinics and union sub-centers of Mithapukur are supposed to refer patients to their upazila health complex (50-bed primary facility) in Durgapur, the urban union. If needed, patients are referred to Rangpur Medical College Hospital (1020-bed tertiary facility) [29,32].
## Data collection strategies and sampling
Between December 2015 and January 2016, we collected data from people affected by NCDs, other members of the community, and healthcare providers using a combination of in-depth interviews (IDIs) and Focus Group Discussions (FGDs). We targeted both community members and healthcare providers to enhance completeness of information and triangulate findings across data sources.
We purposely sampled 25 individuals from the pool who had reported at least one non-communicable chronic condition during the prior quantitative survey [21]. We have included participants with at least one chronic condition, not a healthy individual because our aim was to find out how a person navigates the health system for their health needs when they are affected by a chronic non-communicable disease. This has been a purposive sampling decision, as choosing a group who are already dealing with chronic NCDs increased our chances to elicit more information related to health-seeking for these conditions. To better unravel heterogeneity around health-seeking decisions, we sampled individuals in relation to their previously revealed treatment choice (qualified, semi-qualified, or no treatment/self-care), residency (urban or rural), and sex. Likewise, we purposely selected 21 health providers based on their qualifications (qualified or semi-qualified), affiliation (private or public), and location of practice (urban or rural). Both these individuals and the healthcare providers served as respondents for in-depth interviews. Semi-qualified professionals are any allopathic or traditional providers with some degree of training and experience in primary care, but no specific expertise in CNCDs. For example, medical assistants, village doctors, community health workers, drugstore keepers, and traditional healers. Qualified professionals are health providers who are registered and trained physicians (i.e., MBBS doctors) [33,34]. In addition, we conducted six FGDs, reaching a total of 35 community members. We conveniently sampled FGD participants with support from local grassroots health providers, based on location of residence (urban or rural), sex, and age (middle age and elderly).
The first author (Bangla-speaker, trained medical doctor with public health research experience) and two research assistants (trained Bangla-speaker anthropologists) collected the data. They conducted all interviews and FGDs using pre-tested semi-structured interview guides (S1, S2 and S3 Guidelines), developed beforehand by the research team. The IDI guides for community members affected by NCDs and health providers explored the role of individual factors (such as age and sex), household factors (such as socio-economic status), and contextual factors (such as broader cultural values and social norms) on health-seeking for NCDs and related expenses, perceptions and experiences regarding NCDs, and provider preferences. The IDIs with the people affected by NCDs were mostly conducted at their home, except for one respondent who chose to talk at another place of his choice. The interviews with the health-care providers were conducted mostly at their respective workplaces, a couple were done in their home as per their choice. The FGDs were conducted inside the community, in an open common place where 6–8 people could sit and discuss comfortably. The FGD guides focused more specifically on contextual factors shaping NCD health responses at community level. All interviews and FGDs were conducted in Bangla, audio-recorded, and transcribed verbatim in Bangla. Only the quotes used in this article were translated into English by the lead author.
## Ethical approval and consent to participate
Ethical approval received- from Ethical Review Committee of BRAC JPG School of Public Health, BRAC University, on December 21, 2015 (Ethics reference no: 71). Interviewers obtained informed written consent from all respondents before the interview was conducted. Furthermore, written permission was obtained from the Upazila Health and Family Planning Officer (UHFPO) of Mithapukur to interview public sector providers from various tiers of health facilities in Mithapukur. The investigators offered a token gift to the NCD-affected respondents and FGD participants in appreciation of their time.
## Consent for publication
All participants were asked for consent to use non-attributable quotes in publications during the consenting process. Other than this, no individual data, images or videos of participants have been published.
## Analytic approach
This study’s conceptual framework (Fig 1) is rooted in the Andersen conceptual model and considers “Health-seeking behavior for NCDs” as a focal point, which is influenced by individual and household characteristics (including cultural beliefs and values) and contextual factors (including socio-cultural factors, the health system and broader environmental factors) [35].
**Fig 1:** *Conceptual framework: Health-seeking behavior for non-communicable diseases (NCDs).*
We applied thematic analysis to distill findings from the transcripts. We proceeded in steps. First, the first and third author (both native Bangla speakers with clinical and public health experience) independently developed codes which reflected the themes explored in the interview guide. Second, they independently coded the transcribed material using the abovementioned codes, but allowed for additional codes to emerge as they proceeded through the analysis. Third, coded material was displayed and the relationships between codes and emerging themes were explored with support from all authors. Discordant interpretations of the findings were resolved via discussion among all authors and/or by returning to the data when necessary. Last, all authors agreed on an emerging interpretation. We conducted the analysis using Nvivo, version 9. To increase the validity of our findings, we applied data (different data sources), methods (different data collection methods), and investigator (different researchers coding) triangulation [36].
## Results
To give voice to community concerns, we purposely report findings in a very descriptive manner and illustrate the major emerging themes (first, individual and household factors, and second, contextual factors and health-seeking experience) using a wealth of direct quotation from respondents’ narratives. The respondents’ characteristics are given in Tables 1 and 2. Table 3 lists the abbreviations used to describe the respondent attributes.
## Individual and household factors
Respondents reported family, friends, and semi-qualified professionals to be the ones shaping their NCD knowledge and perceptions most. None of the participants received any NCD awareness messages from government. In the context of poor literacy rates, the role of education in health-seeking seemed trivial, as it was not brought up by the respondents. After probing, a few acknowledged that education helped in following prescriptions. Most respondents did not have an understanding that NCDs represent life-long conditions, requiring continuous care. Many found it difficult to acknowledge that NCDs are only controllable, not curable. More than half of all NCD respondents believed that their conditions would be cured if God forgave them, and framed health providers only as a means towards a cure. “ The doctor is a medium, *Allah is* the one who cures.” ( Ur, R-5, F, A- 50y). A handful of respondents expressed that different NCDs require different forms of treatment, as allopathic and other forms of treatment (i.e., homeopathy, traditional) serve different purposes. “ People say doctors cannot treat Baat [rheumatoid arthritis], it [cure] is in the hands of a traditional healer.” ( Ur, R-4, F) A condition is perceived as existing when symptoms are present, and considered severe when symptoms aggravate to an extent that these symptoms interfere with normal daily activities. At this point, the health-seeking process is initiated. People tend to wait until long after the symptoms appear. No one sought care as soon as the symptoms presented. Many reported discontinuing NCD treatments upon disappearance of their symptoms, as they perceived themselves cured. “ I did not do anything else as I was cured.” ( Ru, R-2, F, 24y). Providers expressed their frustration over this behaviour: This fact, in addition to people’s belief that irrespective of providers, only God can ultimately cure the disease, led respondents to shop from provider to provider in search of a cure. “ *It is* Allah’s will… some people get cured… such places are there, I am telling you.” ( Ur, R-8, M, A- 50y) Most respondents relied on personal and social contacts to initiate health-seeking. As such, interpersonal relationships with providers and their accessibility appeared pivotal in shaping decisions on health-seeking. This explains why almost all of our sample respondents reported to have first sought care from semi-qualified professionals: they are known in the community, are available round the clock, and provide in-home consultations upon request. Moreover, as members of the community themselves, semi-qualified providers are considered trustworthy. People feel comfortable approaching them, as they feel relatable. The absence of qualified providers in rural areas further reinforces this preference towards semi-qualified providers.
Providers are also aware of the popularity of semi-qualified professionals: Interviews also revealed that health-seeking experience differed substantially between men and women. A person’s gender appeared to determine whether they enjoy the privilege of autonomous decision-making. All women unanimously mentioned that they needed permission from their “guardian” (male member of the household) to be able to decide on health-seeking, attributing their financial dependency on men. None of our female respondents held a formal job. All female community health workers interviewed confirmed needing permission for their own health-seeking. Several women reported having discontinued treatment due to their husbands’ disapproval, and no woman had sought care in spite of an unsupportive family. “ If I were a man, I would not have needed anyone’s permission.” ( Ur, R-3, A-26y).
Responses from FGDs and interviews with providers confirmed the lack of empowerment expressed by women in individual interviews. “ *If a* woman wants to override a man’s word and do something, she obviously cannot do that. Can she? Never!” ( FGD-Ur, F). Additionally, most women reported needing a companion to go health-seeking, and expressed concerns about household chores when leaving to seek care. Inversely, men could decide for themselves, and did not require permission of any sort to seek care: “I can go [for treatment] as soon as I say. All the money is mine, my own.” ( Ur, R-11, M, A->60y) All financially dependent elderly respondents were found to forgo treatment sometimes, as they wished not to trouble their family for their treatment expenses. They believed that they should endure this discomfort as a natural process.
A few respondents reported fear of facing stigma due to NCDs. Respondents revealed that among their communities, the belief persist that NCDs affect sexual performance, hence reducing one’s chances to find a spouse.
## Contextual factors and health-seeking experience
Geographical accessibility, defined both in terms of actual distance and means of transport, represents a major barrier to NCD care-seeking. The government health workers confirmed that NCD diagnostic tools and medications are only available at the Upazilla Health Complex, and therefore not available at grassroots level (union sub-centers and community clinics).
Respondents indicated the cost of treatment being fundamental in shaping health-seeking behavior, leading one to forgo or delay treatment, discontinue or change providers/the course of treatment. This sentiment was stronger among respondents of poorer households (16 respondents) in our sample. “ I did not go for any treatment for about a year. I did not go due to lack of money.” ( Ur, R-3, A-26y, F). Respondents most often regarded qualified care as an expensive option, motivating their decision not to seek care unless in dire need. When explaining the costs incurred by treatment, narratives show that people incurred substantial expenses while availing semi-qualified care, but did not seem to recognize that.
In contrast, respondents from least poor families (9 respondents) did not single out the importance of cost in availing healthcare, but rather insisted on perceived treatment effectiveness and on provider-patient interactions as aspects determining their satisfaction with provider. People affected by NCDs measure treatment effectiveness with the speed with which their symptoms are alleviated. “ It does not matter if money is spent, I have to stay well.” ( Ur, R-1, F, A-50y, Diabetes) People were found to lean towards patient-centred care. Many expressed that they would like to participate in decision-making concerning their treatment, rather than just receiving a prescription. People seemed to appreciate the opportunity to negotiate or at least express their opinions with a semi-qualified provider, an opportunity that rarely arose in qualified care encounters. Many would discontinue treatment if the provider failed to understand their needs. Both respondents affected by NCDs and providers confirmed that the pluralistic nature of the local healthcare system enables discontinuation and frequent shifts across providers.
## Discussion
This study is one of the very first qualitative investigations conducted in Bangladesh aiming at exploring health-seeking behavior for NCDs. As such, this study unravels how individual, household, and systemic elements come together to shape decision-making on health-seeking for NCDs. In particular, our findings revealed how cultural and social elements interact with poor geographical and financial accessibility elements to deter individuals affected by NCDs from seeking care. This evidence is very much needed to identify relevant barriers to access care and design appropriate policies to overcome them.
Our findings reveal that people often do not perceive chronic conditions to be sufficiently severe to motivate care-seeking, unless they significantly affect their daily life. Similarly, our findings also revealed that people often discontinue treatment as soon as the most disturbing symptoms disappear. These findings are aligned with prior evidence from Bangladesh [37], as well as from other countries [38,39], indicating that the severity of illness shapes one’s care-seeking decisions. Likewise, findings from India [40] and Kenya [41] also revealed that people seek care only once a condition affects their daily life. The practices captured by our analysis do not appear surprising, considering a context of widespread poverty and lack of financial protection [42]. This forces people to ration limited resources cautiously, given sparce access to reliable sources of information on the health impacts of chronic conditions [11]. Nonetheless, such practices can be especially harmful for NCDs, as chronic conditions usually produce symptoms only at an advanced stage, when the treatment becomes more expensive and the health outcomes worse. Considering that major NCDs (cardiovascular diseases, respiratory disease, cancer, diabetes) are responsible for $80\%$ of all premature deaths [43], measures to alleviate the financial burden should be coupled with information and education campaigns to enlighten people on the health and economic penalties of letting a NCD go untreated.
Moreover, we found that in the absence of access to reliable sources of information on NCDs and treatment options available, religious beliefs interact with false notions on the severity of a condition and direct decisions on care-seeking. More specifically, we noted the common belief that NCDs can be cured only with God’s forgiveness, so that health providers only serve as a means to alleviate the presence of symptoms. An earlier Bangladeshi study found similar religious beliefs on sexual health-related health-seeking [44]. Similarly, a study in Kenya reported the belief that people affected by chronic conditions would be healed by prayer [41]. While recognizing the need to respect people’s religious conviction and trust in God, we note the dangerous implications of such beliefs on care-seeking. Our findings highlight the fact that due to a combination of religious beliefs, poor understanding of the etiology of disease and its long-term consequences, and an occasional fear of facing stigma, individuals affected by NCDs tend to delay seeking care. Once they have reached the decision to seek care, they often shop across providers, looking for a satisfying care experience and cure. Such shopping results in discontinuity of care, bearing harmful health and financial consequences in the long run [45–47]. Unlike acute illnesses, it is not possible to fully cure chronic conditions. People require life-long treatment, aiming towards extending their lifespan whilst improving its quality, physically and mentally [48]. For such conditions, shopping across providers poses a challenge to both quality and continuity of care, since it implies that people are receiving uncoordinated and fragmented care from providers of various qualifications. At the same time, this pattern indicates the need for a streamlined NCD care provision plan and points to the need to develop more people-centered healthcare systems [49,50].
In line with the above, our findings also revealed that considerations related to geographical accessibility, cost, and familiarity with a provider interact to determine where a person affected by a chronic condition chooses to seek care. On the one hand, the lack of service provision for NCDs in remote locations and the fear of high costs associated with qualified health-seeking discouraged people from referring first to qualified providers. On the other hand, familiarity with semi-qualified providers, their lower perceived cost, and their accessibility at village level encouraged people to seek them first when in need of care. In this regard, our findings are clearly aligned with prior evidence from Bangladesh, suggesting that semi-qualified providers are often the providers of choice due to their interpersonal skills and closeness to the community [18,20,44]. Interestingly, a recent systematic review has revealed that providers’ interpersonal skills are key in shaping patients’ satisfaction [51]. Similarly, existing evidence suggests that people skills are crucial in ensuring continuity of care for chronic conditions [41]. The policy implications of our findings are straightforward, calling for a radical restructuring of the health system to ensure availability of patient-centered and low-cost basic NCD screening and treatment at lower levels of care. It will be challenging since Bangladeshi public health outlets have not prioritized primary and secondary preventive care for NCDs at the most basic level [11]. The primary and secondary care facilities need skilled professionals, NCD medications, related equipment and coordinated referrals [11,27]. The latest health bulletin from Bangladesh states that basic supplies for the management of hypertension and diabetes are supposed to be available in public primary healthcare outlets [28]. The government has recently activated NCD corners in selected upazila health complexes. The health providers there acknowledge the role of this initiative in increasing NCD-related literacy, services, and referrals, and say the upazila health complex requires standardized procedures, additional training, diagnostic facilities, record-keeping, and medical supplies [52]. A recent study reported that among primary and secondary health facilities, only $24\%$ and $58\%$ were able to provide basic diagnostic and treatment facilities for cardiovascular diseases and diabetes respectively [53].
The fact that respondents in our sample consistently identified the high cost of care as a major barrier to accessing care reinforces findings from our prior quantitative study [21] and from other studies [18,54], suggesting that OOPE for NCDs in *Bangladesh is* significant. Such findings are not surprising, since the literature has highlighted in depth how the expected cost of care represents a decisive factor to treatment adherence [55], especially in cases of NCDs, where treatment continues long-term [56,57]. Likewise, the literature has shown that people are more likely to discontinue treatment in the absence of health insurance and sufficient income [55]. A review article on South Asia found that NCD-households are more likely to experience OOPE, catastrophic expenses and impoverishment compared to non-NCD households [54]. A recent 18-country study that included Bangladesh found that among NCD respondents in low income countries, around $39\%$ of women forewent medicines due to expenses, whereas $13\%$ of men did so [58].
Finally, we would like to note how our findings indicate that cultural values and social norms pertaining to women and their position in society affect their ability to make autonomous decisions about their health-seeking. This lack of empowerment and dependence on male household members has been noted in prior Bangladeshi studies [59–61]. Although the preceding study found no significant gender differences in health-seeking [21], this in-depth exploration unearthed an existing blockade imposed by the patriarchal societal context. This indicates that there might be many more women in need of NCD care than studies have previously shown.
## Study considerations
In spite of its strengths as one of the first rigorously implemented qualitative studies on NCDs in the setting, we need to acknowledge a number of limitations. First, we recognize that data were collected in 2016 and that since then, NCD policies and programs have partially changed. Although NCD corners have been established at subdistrict health facilities, these remain in a very early stage of development [52]. Nonetheless, given the paucity of available evidence [11] and given that the socio-cultural and economic setting has remained largely unchanged, our findings and their implications continue to be relevant.
Second, we recognize that the sample included few respondents with school education, and few elderly respondents, limiting our capacity to make meaningful inferences on the role played by education and health in mediating healthcare seeking, and calling for further research. Similarly, we could not reach any woman with formal employment, hampering us from assessing if and how formal employment can act as an enabling element in NCD-related care decisions.
## Conclusion
This study has looked into the intricacies of health-seeking for NCDs, an aspect not extensively researched in Bangladesh. Our findings highlight delayed and fragmented care seeking for NCDs and the absence of continuity of care, rooting from health and religious beliefs and anticipated expenses. The medically pluralistic context and limited access to trained providers and NCD services negatively impact health-seeking behavior. A popular first choice is semi-qualified providers, due to their easy access and interpersonal skills. Possible solutions emerging from our study are introducing people-centered care, raising NCD awareness and meeting the provider and facility’s NCD resource needs. Women and the very poor need special access to NCD care, given a patriarchal context and no social health protection coverage. Further studies could look into the effect of gender, old age, and stigma on NCD health-seeking.
## References
1. 1Global Health Metrics-Non-communicable diseases-Level 1 cause. 17 Oct 2020 [cited 25 Jan 2021]. Available: https://www.thelancet.com/pb-assets/Lancet/gbd/summaries/diseases/non-communicable-diseases.pdf.
2. Wang H, Naghavi M, Allen C, Barber RM, Bhutta ZA, Carter A. **Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for 249 causes of death, 1980–2015: a systematic analysis for the Global Burden of Disease Study 2015**. *The Lancet* (2016.0) **388** 1459-1544. DOI: 10.1016/S0140-6736(16)31012-1
3. 3World Health Organization. Global status report on noncommunicable diseases 2014: “Attaining the nine global noncommunicable diseases targets; a shared responsibility.”
World Health Organization; 2014. Available: http://apps.who.int/iris/bitstream/10665/148114/1/9789241564854_eng.pdf?ua=1.. *Global status report on noncommunicable diseases 2014: “Attaining the nine global noncommunicable diseases targets; a shared responsibility.”* (2014.0)
4. 4World Health Organization. Global Health Estimates 2019 Summary Tables: DALYs By Cause, Age And Sex, By World Bank Income Group, 2000–2019.
2020. Available: https://www.who.int/data/gho/data/themes/mortality-and-global-health-estimates.. *Global Health Estimates 2019 Summary Tables: DALYs By Cause, Age And Sex, By World Bank Income Group, 2000–2019.* (2020.0)
5. 5World Health Organization. GLOBAL NONCOMMUNICABLE DISEASES: Saving Lives by Reducing Risk. 2017; 1.. *GLOBAL NONCOMMUNICABLE DISEASES: Saving Lives by Reducing Risk* (2017.0) **1**
6. Bloom DE, Cafiero ET, Jané-Llopis E, Abrahams-Gessel S, Bloom LR, Fathima S. *The Global Economic Burden of Non-communicable Diseases* (2011.0)
7. 7World Health Organization. From Vision to Results: Advancing health for billions in the South-East Asia Region. World Health Organization, Regional Office for South-East Asia; 2017. Available: https://apps.who.int/iris/bitstream/handle/10665/258762/9789290225928-eng.pdf?sequence=1&isAllowed=y.. *From Vision to Results: Advancing health for billions in the South-East Asia Region* (2017.0)
8. 8World Health Organization. Action plan for the prevention and control of non-communicable diseases in South-East Asia, 2013–2020.
World Health Organization (WHO); 2013. Available: http://www.searo.who.int/entity/noncommunicable_diseases/documents/sea-ncd-89(reduced).pdf?ua=1.. *Action plan for the prevention and control of non-communicable diseases in South-East Asia, 2013–2020.* (2013.0)
9. 9World Health Organization. World Health Statistics 2016: Monitoring health for the SDGs, sustainable development goals.
Geneva: World Health Organization; 2016. Available: http://apps.who.int/iris/bitstream/10665/206498/1/9789241565264_eng.pdf?ua=1.. *World Health Statistics 2016: Monitoring health for the SDGs, sustainable development goals.* (2016.0)
10. 10World Health Organization. Noncommunicable Diseases Progress Monitor 2020.
2020. Available: https://www.who.int/publications-detail/ncd-progress-monitor-2020.. *Noncommunicable Diseases Progress Monitor 2020.* (2020.0)
11. 11Bangladesh Health Watch. Bangladesh Health Watch Report 2016: Non-communicable diseases in Bangladesh, current scenario and future directions.
Dhaka, Bangladesh: Bangladesh Health Watch Secretariat, James P Grant School of Public Health, BRAC University; 2017. Available: http://www.brac.net/images/news/2017/bhw_report2016_final.pdf.. *Bangladesh Health Watch Report 2016: Non-communicable diseases in Bangladesh, current scenario and future directions.* (2017.0)
12. 12Management Information System, Directorate General of Health Services. Health Bulletin 2019. Jun 2020 [cited 3 Feb 2021]. Available: https://dghs.gov.bd/images/docs/Publicaations/Health%20Bulletin%202019%20Print%20Version%20(2)-Final.pdf.
13. Islam SMS, Islam MT, Islam A, Rodgers A, Chow CK, Naheed A. **National drug policy reform for noncommunicable diseases in low-resource countries: an example from Bangladesh**. *Bull World Health Organ* (2017.0) **95** 382-384. DOI: 10.2471/BLT.15.161117
14. 14World Health Organization. A prioritized research agenda for prevention and control of noncommunicable diseases. Geneva: World Health Organization; 2011.. *A prioritized research agenda for prevention and control of noncommunicable diseases* (2011.0)
15. Tipping G.. *Health care seeking behaviour in developing countries: an annotated bibliography and literature review.* (1995.0)
16. 16Grundy J, Annear P. Health-seeking behaviour studies: a literature review of study design and methods with a focus on Cambodia. Health policy and health finance knowledge hub working paper series no. 2010;7. Available: http://ni.unimelb.edu.au/__data/assets/pdf_file/0011/385967/HPHF_Hub_WP7_Health_Seeking_Behaviour.pdf.
17. Parr JD, Lindeboom W, Khanam MA, Koehlmoos TLP. **Diagnosis of chronic conditions with modifiable lifestyle risk factors in selected urban and rural areas of Bangladesh and sociodemographic variability therein.**. *BMC Health Services Research* (2011.0) **11** 309. DOI: 10.1186/1472-6963-11-309
18. Uddin MJ, Alam N, Sarma H, Chowdhury MA, Alam DS, Niessen L. **Consequences of hypertension and chronic obstructive pulmonary disease, healthcare-seeking behaviors of patients, and responses of the health system: a population-based cross-sectional study in Bangladesh.**. *BMC Public Health.* (2014.0) **14** 547. DOI: 10.1186/1471-2458-14-547
19. Khanam MA, Lindeboom W, Koehlmoos TLP, Alam DS, Niessen L, Milton AH. **Hypertension: adherence to treatment in rural Bangladesh–findings from a population-based study.**. *Glob Health Action* (2014.0) **7**. DOI: 10.3402/gha.v7.25028
20. Sikder SS, Labrique AB, Ullah B, Mehra S, Rashid M, Ali H. **Care-seeking patterns for fatal non-communicable diseases among women of reproductive age in rural northwest Bangladesh.**. *BMC Women’s Health* (2012.0) **12** 23. DOI: 10.1186/1472-6874-12-23
21. Rasul FB, Kalmus O, Sarker M, Adib HI, Hossain MS, Hasan MZ. **Determinants of health seeking behavior for chronic non-communicable diseases and related out-of-pocket expenditure: results from a cross-sectional survey in northern Bangladesh.**. *J Health Popul Nutr* (2019.0) **38** 48. DOI: 10.1186/s41043-019-0195-z
22. Ward H, Mertens TE, Thomas C. **Health Seeking Behaviour and the Control of Sexually Transmitted Disease.**. *Health Policy Plan* (1997.0) **12** 19-28. DOI: 10.1093/heapol/12.1.19
23. MacKian S.. *A review of health seeking behaviour: problems and prospects.* (2003.0). DOI: 10.1016/s1353-8292(02)00054-0
24. 24The World Bank, Bangladesh Bureau of Statistics (BBS), World Food Programme (WFP). Poverty Maps of Bangladesh 2010: Technical Report. 2014. Report No.: 90487 v2. Available: http://www-wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2014/11/11/000442464_20141111221335/Rendered/PDF/904870v20Bangl0LIC000Sept004020140.pdf.
25. Bleich SN, Koehlmoos TLP, Rashid M, Peters DH, Anderson G. **Noncommunicable chronic disease in Bangladesh: Overview of existing programs and priorities going forward.**. *Health Policy.* (2011.0) **100** 282-289. DOI: 10.1016/j.healthpol.2010.09.004
26. 26World Bank, South Asia Human Development, Health Nutrition and Population.
NCDs Policy Brief-Bangladesh.
Washington D.C.: Word Bank; 2011
Feb. Available: http://siteresources.worldbank.org/SOUTHASIAEXT/Resources/223546-1296680097256/7707437-1296680114157/NCD_BD_Policy_Feb_2011.pdf.. *NCDs Policy Brief-Bangladesh.* (2011.0)
27. Zaman MM, Ullah AKMJ, Bhuiyan MR, Rahman SMA. **Noncommunicable Disease Prevention and Control Situation in a Primary Health Care Setting of Bangladesh: Design and Baseline Findings of an Intervention.**. *Chronic Dis Int* (2016.0) **3**
28. 28Management Information System, Directorate General of Health Services. Health Bulletin 2018. 2018. Available: http://www.dghs.gov.bd/images/docs/Health%20Bulletin/Health%20Bulletin%202018%20Pre%20print%20version.pdf.
29. 29National Institute of Population Research and Training (NIPORT), Associates for Community and Population Research (ACPR)for, ICF International. Bangladesh Health Facility Survey 2014. NIPORT, ACPR and ICF International; 2016. Available: https://dhsprogram.com/pubs/pdf/SPA23/SPA23.pdf.
30. 30Bangladesh National Portal. Mithapukur Upazila. 30 Sep 2019 [cited 6 Oct 2019]. Available: http://mithapukur.rangpur.gov.bd/site/page/5cf518d8-18fd-11e7-9461-286ed488c766/%E0%A6%8F%E0%A6%95%20%E0%A6%A8%E0%A6%9C%E0%A6%B0%E0%A7%87%20%E0%A6%AE%E0%A6%BF%E0%A6%A0%E0%A6%BE%E0%A6%AA%E0%A7%81%E0%A6%95%E0%A7%81%E0%A6%B0.
31. 31World Bank. Bangladesh Interactive Poverty Maps. In: World Bank [Internet]. 10
Nov
2016 [cited 6 Oct 2019]. https://www.worldbank.org/en/data/interactive/2016/11/10/bangladesh-poverty-maps.. *Bangladesh Interactive Poverty Maps.* (2016.0)
32. 32Bangladesh Bureau of Statistics (BBS). District Statistics 2011: Rangpur. Reproduction, Documentation and Publication (RDP), FA & MIS, BBS: Bangladesh Bureau of Statistics (BBS); 2013 Jun. Available: http://203.112.218.65:8008/WebTestApplication/userfiles/Image/District%20Statistics/Rangpur.pdf.
33. Ahmed SM, Adams AM, Chowdhury M, Bhuiya A. **Gender, socioeconomic development and health-seeking behaviour in Bangladesh.**. *Social Science & Medicine.* (2000.0) **51** 361-371. DOI: 10.1016/S0277-9536(99)00461-X
34. Ahmed SM, Hossain MdA. **Knowledge and practice of unqualified and semi-qualified allopathic providers in rural Bangladesh: Implications for the HRH problem.**. *Health Policy.* (2007.0) **84** 332-343. DOI: 10.1016/j.healthpol.2007.05.011
35. Andersen RM. **National Health Surveys and the Behavioral Model of Health Services Use.**. *Med Care.* (2008.0) **46** 647-653. DOI: 10.1097/MLR.0b013e31817a835d
36. Padgett DK. *Qualitative and mixed methods in public health* (2012.0)
37. Zaman S, Rahman H, Edgeworth R. *Health Domain of the Ultra Poor: An Exploration* (2007.0)
38. Kanungo S, Bhowmik K, Mahapatra T, Mahapatra S, Bhadra UK, Sarkar K. **Perceived Morbidity, Healthcare-Seeking Behavior and Their Determinants in a Poor-Resource Setting: Observation from India.**. *PLoS One.* (2015.0) **10**. DOI: 10.1371/journal.pone.0125865
39. Wang Q, Brenner S, Leppert G, Banda TH, Kalmus O, Allegri MD. **Health seeking behaviour and the related household out-of-pocket expenditure for chronic non-communicable diseases in rural Malawi**. *Health Policy Plan* (2015.0) **30** 242-252. DOI: 10.1093/heapol/czu004
40. Dodd W, King N, Humphries S, Little M, Dewey C. **Self-reported morbidity and health service utilization in rural Tamil Nadu, India.**. *Social Science & Medicine.* (2016.0) **161** 118-125. DOI: 10.1016/j.socscimed.2016.05.035
41. Rachlis B, Naanyu V, Wachira J, Genberg B, Koech B, Kamene R. **Identifying common barriers and facilitators to linkage and retention in chronic disease care in western Kenya.**. *BMC Public Health.* (2016.0) **16**. DOI: 10.1186/s12889-016-3462-6
42. 42Ministry of Health and Family Welfare (MOHFW). Bangladesh National Health Accounts: 1997–2012. 2015. Available: https://heu.portal.gov.bd/sites/default/files/files/heu.portal.gov.bd/files/102a60c8_1a29_414a_8ae6_1d30406ff7c1/2020-02-18-12-58-494a491415f422543efeb5f19ae85c34.pdf.
43. 43World Health Organization. Non communicable diseases. 1 Jun 2018 [cited 9 May 2019]. Available: https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases.
44. Mahmud I, Chowdhury S, Siddiqi BA, Theobald S, Ormel H, Biswas S. **Exploring the context in which different close-to-community sexual and reproductive health service providers operate in Bangladesh: a qualitative study.**. *Human Resources for Health* (2015.0) **13** 51. DOI: 10.1186/s12960-015-0045-z
45. **Continuity of Care to Optimize Chronic Disease Management in the Community Setting: an evidence-based analysis.**. *Ont Health Technol Assess Ser.* (2013.0) **13** 1-41. PMID: 24167540
46. Chen C-C, Chen S-H. **Better continuity of care reduces costs for diabetic patients.**. *Am J Manag Care* (2011.0) **17** 420-427. PMID: 21756012
47. Hussey PS, Schneider EC, Rudin RS, Fox D, Lai J, Pollack C. **Continuity and the costs of care for chronic disease**. *JAMA Intern Med* (2014.0) **174** 742-748. DOI: 10.1001/jamainternmed.2014.245
48. Grumbach K.. **Chronic Illness, Comorbidities, and the Need for Medical Generalism.**. *Ann Fam Med* (2003.0) **1** 4-7. DOI: 10.1370/afm.47
49. Nolte E, McKee M. *Caring for People with Chronic Conditions: A Health System Perspective.* (2008.0)
50. De Maeseneer J, Roberts RG, Demarzo M, Heath I, Sewankambo N, Kidd MR. **Tackling NCDs: a different approach is needed**. *Lancet* (2012.0) **379** 1860-1861. DOI: 10.1016/S0140-6736(11)61135-5
51. Batbaatar E, Dorjdagva J, Luvsannyam A, Savino MM, Amenta P. **Determinants of patient satisfaction: a systematic review.**. *Perspect Public Health* (2017.0) **137** 89-101. DOI: 10.1177/1757913916634136
52. Rawal LB, Kanda K, Biswas T, Tanim MI, Poudel P, Renzaho AMN. **Non-communicable disease (NCD) corners in public sector health facilities in Bangladesh: a qualitative study assessing challenges and opportunities for improving NCD services at the primary healthcare level.**. *BMJ Open* (2019.0) **9**. DOI: 10.1136/bmjopen-2019-029562
53. Biswas T, Haider MM, Gupta RD, Uddin J. **Assessing the readiness of health facilities for diabetes and cardiovascular services in Bangladesh: a cross-sectional survey**. *BMJ Open* (2018.0) **8** e022817. DOI: 10.1136/bmjopen-2018-022817
54. Rijal A, Adhikari TB, Khan JAM, Berg-Beckhoff G. **The economic impact of non-communicable diseases among households in South Asia and their coping strategy: A systematic review.**. *PLOS ONE.* (2018.0) **13** e0205745. DOI: 10.1371/journal.pone.0205745
55. Jin J, Sklar GE, Min Sen Oh V, Chuen Li S. **Factors affecting therapeutic compliance: A review from the patient’s perspective.**. *Ther Clin Risk Manag* (2008.0) **4** 269-286. DOI: 10.2147/tcrm.s1458
56. Ponnusankar S, Surulivelrajan M, Anandamoorthy N, Suresh B. **Assessment of impact of medication counseling on patients’ medication knowledge and compliance in an outpatient clinic in South India.**. *Patient Education and Counseling* (2004.0) **54** 55-60. DOI: 10.1016/S0738-3991(03)00193-9
57. Ellis JJ, Erickson SR, Stevenson JG, Bemstein SJ, Stiles RA, Fendrick MA. **Suboptimal Statin Adherence and Discontinuation in Primary and Secondary Prevention Populations.**. *J Gen Intern Med* (2004.0) **19** 638-645. DOI: 10.1111/j.1525-1497.2004.30516.x
58. Murphy A, Palafox B, Walli-Attaei M, Powell-Jackson T, Rangarajan S, Alhabib KF. **The household economic burden of non-communicable diseases in 18 countries**. *BMJ Glob Health* (2020.0) **5**. DOI: 10.1136/bmjgh-2019-002040
59. Ahmed SM. *Exploring health-seeking behaviour of disadvantaged populations in rural Bangladesh.* (2005.0)
60. Hossen MA, Westhues A. **Rural Women’s Access to Health Care in Bangladesh: Swimming Against the Tide?**. *Social Work in Public Health.* (2011.0) **26** 278-293. DOI: 10.1080/19371910903126747
61. Hamiduzzaman M, de Bellis A, Abigail W, Kalaitzidis E. **Elderly Women in Rural Bangladesh: Healthcare Access and Ageing Trends.**. *South Asia Research.* (2018.0). DOI: 10.1177/0262728018767018
|
---
title: 'Assessment of health system readiness for routine maternal and newborn health
services in Nepal: Analysis of a nationally representative health facility survey,
2015'
authors:
- Resham B. Khatri
- Yibeltal Assefa
- Jo Durham
journal: PLOS Global Public Health
year: 2022
pmcid: PMC10022376
doi: 10.1371/journal.pgph.0001298
license: CC BY 4.0
---
# Assessment of health system readiness for routine maternal and newborn health services in Nepal: Analysis of a nationally representative health facility survey, 2015
## Abstract
Access to and utilisation of routine maternal and newborn health (MNH) services, such as antenatal care (ANC), and perinatal services, has increased over the last two decades in Nepal. The availability, delivery, and utilisation of quality health services during routine MNH visits can significantly impact the survival of mothers and newborns. Capacity of health facility is critical for the delivery of quality health services. However, little is known about health system readiness (structural quality) of health facilities for routine MNH services and associated determinants in Nepal. Data were derived from the Nepal Health Facility Survey (NHFS) 2015. Total of 901 health facilities were assessed for structural quality of ANC services, and 454 health facilities were assessed for perinatal services. Adapting the World Health Organization’s Service Availability and Readiness Assessment manual, we estimated structural quality scores of health facilities for MNH services based on the availability and readiness of related subdomain-specific items. Several health facility-level characteristics were considered as independent variables. Logistic regression analyses were conducted, and the odds ratio (OR) was reported with $95\%$ confidence intervals (CIs). The significance level was set at p-value of <0.05. The mean score of the structural quality of health facilities for ANC, and perinatal services was 0.62, and 0.67, respectively. The average score for the availability of staff (e.g., training) and guidelines-related items in health facilities was the lowest (0.37) compared to other four subdomains. The odds of optimal structural quality of health facilities for ANC services were higher in private health facilities (adjusted odds ratio (aOR) = 2.65, $95\%$ CI: 1.48, 4.74), and health facilities supervised by higher authority (aOR = 1.96; CI: 1.22, 3.13) while peripheral health facilities had lower odds (aOR = 0.13; CI: 0.09, 0.18) compared to their reference groups. Private facilities were more likely (aOR = 1.69; CI:1.25, 3.40) to have optimal structural quality for perinatal services. Health facilities of Karnali (aOR = 0.29; CI: 0.09, 0.99) and peripheral areas had less likelihood (aOR = 0.16; CI: 0.10, 0.27) to have optimal structural quality for perinatal services. Provincial and local governments should focus on improving the health system readiness in peripheral and public facilities to deliver quality MNH services. Provision of trained staff and guidelines, and supply of laboratory equipment in health facilities could potentially equip facilities for optimal quality health services delivery. In addition, supervision of health staff and facilities and onsite coaching at peripheral areas from higher-level authorities could improve the health management functions and technical capacity for delivering quality MNH services. Local governments can prioritise inputs, including providing a trained workforce, supplying equipment for laboratory services, and essential medicine to improve the quality of MNH services in their catchment.
## Introduction
The health of mother and newborns from conception to postnatal is commonly referred to as maternal and newborn health (MNH). Uptake of health services during antenatal period (conception to before childbirth), and perinatal (28 weeks after conception to the first week after childbirth) is vital for improved health status of mothers and newborns [1]. Inadequate access to quality ANC and perinatal care contribute significantly to preventing several maternal and newborn deaths. Furthermore, high-quality ANC presents a unique and lifesaving opportunity for health promotion, disease prevention, early diagnosis and treatment of illnesses in pregnancy using evidence-based practices. To ensure optimum care, the World Health Organisation (WHO) recommended that every pregnant woman have a minimum of four ANC visits throughout the pregnancy, with the first visit in the first trimester [2]. Furthermore, routine ANC visits and childbirth in health facilities assisted by skilled birth attendants ensure antenatal, intrapartum care and immediate maternal and newborn care, and reduce the risk of adverse pregnancy outcomes, including perinatal morbidities and mortalities [3–5].
In the last two decades, Nepal has made significant progress in access to routine MNH visits such as at least four antenatal care visits (4ANC), institutional delivery, and at least one postnatal care (PNC) visit within 48 hours of childbirth. For instance, the uptake of institutional delivery increased from $3\%$ in 1996 to $57\%$ in 2016, and similar patterns of increment were observed in the 4ANC, and PNC visits [6, 7]. However, this increased access to routine services has not been reflected in MNH outcomes. For example, from 1996 to 2006, Maternal Mortality Ratio (MMR) reduced from 539 (reported as per 100000 live births) to 281, and Neonatal Mortality Rate (NMR) reduced from 50 (reported as per 1000 live births) to 33. But in the subsequent decade (2007–2016), MMR reduced from 281 to 259, and NMR reduced 33 to 22 only [6].
Evidence suggests that the reasons for slow progress in MNH outcomes are partly contributed by equity gaps in access to services, and utilisation of poor-quality health services. For instance, in 2016, access to institutional delivery among women of the lowest and highest wealth quintile was $34\%$ and $90\%$, respectively [6]. Women with multiple forms of disadvantage had the lowest coverage of all MNH visits compared to their privileged counterparts [8]. Socioeconomically disadvantaged women from remote areas of Karnali province face access barriers to reach health facilities, which are further compounded by poor transportation systems [9–12].
Furthermore, women who attended health facilities received poor quality MNH services, especially facilities in peripheral areas [13]. Good quality health services need better health system inputs such as the provision of trained and technically competent health workforces, regular supply of essential medicine, and enabling environment (e.g., infrastructure, equipment). Such health system inputs determine the health facility readiness for quality health services and are the precondition for delivering quality health services [14].
The measurement of health care quality is complex, multifaceted, and depends on context, it also requires multiple data on health system inputs and processes of health services delivery. According to the Donabedian model, health care quality comprises three components: structural (inputs), process, and outcomes [15]. Structure denotes the attributes of the settings in which care occurs. Structural quality is health facility capacity to deliver good quality health services [16]. Process quality is the delivery of good quality technical services [17]. Process denotes what is done in giving and receiving care. Good structural quality usually depends on inputs in the health system and leads to the process of care or delivery of good quality health services. Outcome denotes the effects of care on the health status of patients and populations. The outcome component of quality refers to client satisfaction or improved health status of people [18].
Global health policies, plans and strategies, evidence on quality of care of maternal and child health services [4, 19–23], and focus on universal access to quality health services to achieve health-related Sustainable Development Goals (SDG3). Recent health policies of Nepal such as the Nepal Health Sector Strategy (2016–2021) [24], Strategy for Skilled Health Personnel and Skilled Birth Attendants 2020–2025 [25], Nepal Safe Motherhood and Newborn Health Road Map 2030 [26], and Nepal Newborn Action Plan (2015–2035) prioritise the quality of care for improved health outcomes. These policies have envisioned optimal health facility readiness, delivery, and utilisation of quality MNH services. Identifying the provision of health system inputs in terms of health workforce, equipment, medicine, and services is essential to track the implementation of policies and ensure the progress towards universal health coverage (UHC) and SDG3 [27]. Further analysis of nationally representative surveys (e.g., Nepal Health Facility Survey) can generate evidence on the status of health system readiness (structural quality). However, despite high policy priority on quality health care, there is limited evidence available on the status of health system readiness for MNH services and their determining factors in Nepal. Therefore, this study aimed to examine health facilities’ structural quality (inputs) and their associated factors for MNH services. Findings of this study can be instrumental in planning and monitoring health facilities and provide insights to policymakers to set priorities. Furthermore, findings will help to allocate scarce resources for effective implementation of MNH policies for improved health status of mothers and newborns in Nepal.
## Study design
This was a cross-sectional study based on further analysis of secondary data. Data for this study were derived from the nationally representative Nepal Health Facility Survey (NHFS) 2015 [28]. The detailed methodology for the NHFS 2015 has been described in its full report [28]. In NHFS 2015, health facility level information was collected using facility inventory and conducted interviews with the health facility in-charge. In addition, health workers’ training and competency-related information were collected by interviews with specific health workers who provide specific health services (e.g., ANC service). For this study, data from the health facility inventory and health workers’ interview files were merged using a unique health facility identifier available in each file. Health facilities and workers’ information were compiled to calculate the structural quality of health facilities. The structural quality of health facilities was assessed for 901 health facilities providing ANC services, and 454 health facilities providing perinatal care services.
## Nepal’s health system context for maternal and newborn health services
Nepal has three levels of government: local, provincial, and federal. Health system governance is in line with the government system. For instance, the local health system covers community-level health facilities (e.g., community health clinics, outreach primary health and immunisation clinics) and health facilities at the ward level (e.g., health posts). In addition, primary health care centers (PHCCs) and district hospitals are also included in the local health system [24, 29]. At the community level, the network of Female Community Health Volunteers (FCHVs) supports community-based health programs, especially in providing preventive, promotive health services to women and children in their catchment areas. Community health workers (e.g., auxiliary health workers and nurse midwives) provide primary health care services in community outreach immunisation and community health clinics and health posts. Health posts offer routine MNH interventions during antenatal, facility birth, and postnatal care visits. In addition, some health posts are accredited birthing centres that provide institutional delivery services for normal pregnancies. While PHCCs and district hospitals provide basic emergency obstetric and neonatal care are the first referral health institutions. The provincial health system includes hospitals that offer tertiary services such as comprehensive emergency obstetric and neonatal care and specialist health services. Health facilities of the federal level include central level hospitals that provide tertiary and super-specialised services related to maternal and newborn health.
## Independent variables
Based on the information available in the dataset and previous studies [30–32], seven health facility level independent variables were selected, such as managing authority (Private, Public), facility types, provinces (province 1 -not named yet), Madhesh, Bagmati, Gandaki, Lumbini, Karnali, Sudurpaschim), mechanism of quality assurance (Yes, No), frequency of health facilities’ management meeting (No, Sometimes, and Monthly), the existence of feedback collection system in health facilities (Yes, No), availability of external supervision of staff (Yes, No). In addition, Routine quality assurance activity was coded as “yes” for facilities reporting that it routinely carries out quality assurance activities (documentation of report or minutes of a quality assurance meeting, a supervisory checklist, a mortality review, or an audit of records or registers) and “No” for those without such quality assurance activities [28].
## Outcome variables and measurement
The antenatal period covers the time from conception to before labour pain, and perinatal services cover services provided during labor and within the first week of childbirth [33]. This study has two outcome variables: Structural quality of health facilities for i) ANC visits (poor, optimal), ii) perinatal services (poor, optimal).
In the NFHS 2015, data were collected using the World Health Organization’s (WHO) Service Availability and Readiness Assessment (SARA) manual [34]. The WHO’s SARA manual provides a list of items to be included in assessing the structural quality of health facilities under two domains: a) service availability and b) facility readiness [34, 35]. The service availability domain covers a list of recommended service interventions that should be available and when service users attend those health facilities (Tables A and B in S1 File). Under the domain of service availability for perinatal services, there were two subdomains: newborn care, and delivery care (Table B in S1 File). The facility readiness domain covers four sub-domains for both services: general readiness (e.g., water, electricity), equipment (e.g., delivery beds for childbirth services), medicine/commodities (e.g., misoprostol, magnesium sulphate, iron tablets), and staff and guidelines (e.g., availability of protocols, guidelines for training). Based on national guidelines for maternal and newborn care [36], and availability of information in the dataset [28], we contextualised and extracted information for the domain and subdomain-specific items for structural quality of health facility for MNH services taking reference of previous studies [13, 28, 34, 36, 37]. Based on the information available in dataset, a number of domain and sub-domain-specific items were identified to calculate the structural quality scores of health facilities for ANC, and perinatal services (Tables A and B in S1 File). We calculated health system inputs or structural quality of health facilities considering previous studies [31, 38]. First, sub-domain-specific structural quality scores of health facilities for each service were calculated. Averaging subdomain scores, domains scores were calculated for each outcome variable (e.g., ANC services). The average scores of two domains (service availability and facility readiness) were the structural quality of health facility for MNH service. Structure and distribution (e.g., normality) of structural quality scores of health facilities were checked for regression analysis. Distribution of structural quality of health facilities for each service was skewed. Thus, we considered the mean as the cut-off point for dichotomization of score [39, 40], which allows to estimate the odds ratios (ORs) of determinants associated with structural quality of health facilities. Thus, considering the mean score as the cut-off point, the health facilities score was dichotomised into poor (if health facilities score< mean) or optimal (if health facilities score ≥ mean) structural quality of health facilities for each MNH service.
## Statistical analyses
Binomial logistic regression analysis was conducted to identify the health facility level determinants of the structural quality of health facilities for MNH services. Bivaraible and multivariable regression models were conducted for each outcome variable. In the descriptive analysis, frequency, mean score of structural quality of Health facilities for both services, proportion, p values obtained from the chi-square association of each independent variable and outcome variable were reported. The statistical significance level was $p \leq 0.05$ (two-tailed). Before running the multivariable regression (back ward elimination) model, multicollinearity was checked and excluded independent variables having variation inflation factor ≥3 in the multivariable regression analyses [41]. The model fitness test was conducted using the Hosmer Lemeshow test (non-significant results ($p \leq 0.05$) indicated an adequate fit) [42]. All estimates were weighted otherwise indicated. In addition, we adjusted the clustering effects of sampling design in the data analysis stage using the clients’ weight and accounting for survey strata: region and types of health facilities. All analyses were conducted using the survey (svy) command function and considering the clustering effect in Stata 14.0 (Stata Corp, 2015).
## Ethics approval
We used secondary data from the 2015 NHFS. This survey was approved by an ethical review board of Nepal Health Research Council, Nepal, and ICF Marco International, Maryland, USA. The Ministry of Health and Population (MOHP) (Nepal) oversaw the overall research process of the NHFS 2015. The NHFS data are publicly available for further analysis, and data were deidentified of the research participants. This study did not require ethical approval from respective institutions. However, the first author took approval for the download and use of the dataset for his doctoral thesis and this publication.
## Descriptive analysis of health facilities providing routine MNH services
Table 1 shows the descriptive characteristics of health facilities providing ANC, and perinatal services. Of 901 health facilities providing ANC services, more than nine in ten ($93\%$) were managed by the public sector. Nearly nine in ten ($86\%$) were peripheral level health facilities (health posts and clinics). More than half ($52\%$) of the health facilities were in the Hill region. Nearly four in five ($79\%$) health facilities did not have quality assurance activities or feedback collection systems within the past year. However, two-thirds ($67\%$) had monthly facility management meetings and external supervision visits in the past four months.
**Table 1**
| Determinants | Categories | Facilities providing ANC services (N = 901) | Facilities providing ANC services (N = 901).1 | Facilities providing perinatal services (N = 454) | Facilities providing perinatal services (N = 454).1 |
| --- | --- | --- | --- | --- | --- |
| Determinants | Categories | Frequency | % | Frequency | % |
| Managed by | Private | 64 | 7.1 | 45 | 9.9 |
| Managed by | Public | 837 | 92.9 | 409 | 90.1 |
| Facility types | PHCCs and hospitals | 122 | 13.5 | 105 | 23.1 |
| Facility types | Health posts and clinics | 779 | 86.5 | 349 | 76.9 |
| Region | Mountain | 112 | 12.4 | 66 | 14.6 |
| Region | Hill | 473 | 52.5 | 275 | 60.6 |
| Region | Terai | 316 | 35.1 | 112 | 24.7 |
| Province | One | 160 | 17.8 | 78 | 17.1 |
| Province | Madhesh | 154 | 17.1 | 38 | 8.5 |
| Province | Bagmati | 179 | 19.8 | 80 | 17.7 |
| Province | Gandaki | 116 | 12.9 | 66 | 14.5 |
| Province | Lumbini | 135 | 15.0 | 64 | 14.0 |
| Province | Karnali | 68 | 7.5 | 60 | 13.3 |
| Province | Sudurpaschim | 88 | 9.8 | 67 | 14.9 |
| Health facility meeting | No | 165 | 18.3 | 75 | 16.6 |
| Health facility meeting | Sometimes | 129 | 14.4 | 71 | 15.6 |
| Health facility meeting | Monthly | 607 | 67.4 | 308 | 67.8 |
| Quality assurance activities | No | 714 | 79.3 | 360 | 79.4 |
| Quality assurance activities | Yes | 187 | 20.7 | 93 | 20.6 |
| Feedback collection | No | 489 | 54.2 | 222 | 48.9 |
| Feedback collection | Yes | 412 | 45.8 | 232 | 51.1 |
| Supervision of staff | No | 330 | 36.6 | 128 | 28.2 |
| Supervision of staff | Yes | 571 | 63.4 | 326 | 71.8 |
Of the 454 health facilities assessed for perinatal services, nine in ten ($90\%$) health facilities were peripheral health facilities (health posts and health clinics). Public authorities managed more than three in four ($90.1\%$) health facilities. However, nearly eight in ten ($79\%$) did not have a quality assurance system in the past year. In contrast, two-thirds of health facilities had external supervision and had a facility management meeting (Table 1).
## Services availability and facility readiness items for MNH services
Table 2 shows the service availability and readiness items of health facilities ($$n = 901$$) for ANC service in Nepal. Of items included in the availability of the services, there were low items available in laboratory-related items such as tests for urine test ($14.2\%$), blood for haemoglobin ($8.1\%$), and anaemia ($17.9\%$). In the subdomains of facility readiness domain, health facilities were poorly equipped with staff and guidelines, including ANC screening training ($13.8\%$). In addition, there was low availability of medicine such as misoprostol tablets ($17.1\%$), and equipment such as digital blood pressure tool ($2.2\%$). Only $12.3\%$ of health facilities had 24-hour staff availability for ANC services.
**Table 2**
| Service availability domain: | Unnamed: 1 | Unnamed: 2 |
| --- | --- | --- |
| Services availability items | Frequency | Yes (%) |
| ANC counselling | 898 | 99.7 |
| Birth preparedness package counselling | 890 | 98.8 |
| Albendazole tablets distribution | 883 | 98.0 |
| Newborn care counselling | 869 | 96.4 |
| Family Planning counselling | 866 | 96.1 |
| Breastfeeding counselling | 866 | 96.1 |
| PNC counselling | 862 | 95.7 |
| Tetanus toxoid service | 835 | 92.7 |
| Blood pressure measure service | 809 | 89.8 |
| Weighting clients | 799 | 88.7 |
| HIV prevention counselling | 797 | 88.5 |
| Iron tablet distribution | 628 | 69.8 |
| Folic acid distribution | 541 | 60.0 |
| HIV test and counselling | 319 | 35.5 |
| Measure height | 207 | 22.9 |
| Health education service | 164 | 18.2 |
| Misoprostol distribution | 154 | 17.1 |
| Anaemia test service | 161 | 17.9 |
| Urine protein test | 128 | 14.2 |
| Urine test service | 91 | 10.1 |
| Haemoglobin test services | 73 | 8.1 |
| Facility readiness domain | | |
| General readiness | | |
| Client latrine | 736 | 81.7 |
| Client waiting area | 719 | 79.8 |
| Water supply | 730 | 81.0 |
| Electricity service | 573 | 63.6 |
| Emergency transport | 536 | 59.5 |
| Landline phone | 140 | 15.6 |
| 24-hour staff availability | 111 | 12.3 |
| Medicine | | |
| Albendazole tablets | 883 | 98.0 |
| Tetanus toxoid vaccine | 835 | 92.7 |
| Iron-folic tabs | 628 | 69.8 |
| Folic acid | 541 | 60.0 |
| Misoprostol tablets | 154 | 17.1 |
| Equipment | | |
| Examination table | 838 | 93.1 |
| Autoclave service | 836 | 92.8 |
| Fetoscope available | 835 | 92.7 |
| Weighing scale | 817 | 90.7 |
| Stethoscope | 813 | 90.3 |
| Blood pressure set manual | 797 | 88.5 |
| Thermometer | 739 | 82.0 |
| Disinfectant for Infection Prevention | 598 | 66.4 |
| Soap for Infection Prevention | 496 | 55.0 |
| Water for infection prevention | 445 | 49.4 |
| Examination light | 424 | 47.1 |
| Tape fundal height | 272 | 30.2 |
| Digital blood pressure tool | 20 | 2.2 |
| Staff training and guidelines | | |
| Supervision of staff | 751 | 83.3 |
| IEC materials for ANC service | 621 | 68.9 |
| ANC guideline | 217 | 24.0 |
| Complication and management | 134 | 14.9 |
| ANC counselling training | 131 | 14.5 |
| ANC screening training | 125 | 13.8 |
| Nutritional assessment | 82 | 9.1 |
| Other training (e.g., refresher training on ANC) | 21 | 2.3 |
Similarly, Table 3 shows the subdomain-specific items available in health facilities ($$n = 454$$) for perinatal services. Under the service availability domain, two in five health facilities had availability of injectable medicine for mothers ($41\%$) and parental convalescent such as magnesium sulphate ($9.6\%$). Nearly one in four ($23.7\%$) health facilities had 24-hour staff for perinatal services and poor availability of communication services including landline services ($23.2\%$) and mobile services ($11.5\%$). In addition, health facilities had poorly equipped with trained staff and guidelines such as neonatal sepsis management ($19.4\%$). Furthermore, health facilities had a low stock of equipment and commodities such as alcohol for hand rubs ($26.1\%$) and availability of essential medicine such as Nifedipine capsule ($19.1\%$) and calcium gluconate ($26.4\%$).
**Table 3**
| Service availability domain | Unnamed: 1 | Unnamed: 2 |
| --- | --- | --- |
| Newborn care services | Frequency | Yes (%) |
| Immediate breastfeeding | 450 | 99.1 |
| Wrapping baby | 443 | 97.6 |
| Weighing newborn | 434 | 95.8 |
| Head to toe examination | 429 | 94.6 |
| Kangaroo mother care | 415 | 91.5 |
| Skin to skin contact | 413 | 91.1 |
| Delayed bathing | 308 | 67.9 |
| Use of chlorhexidine | 289 | 63.6 |
| Newborn resuscitation | 167 | 36.9 |
| Injectable antibiotic available | 186 | 41.0 |
| Maternity care services | | |
| Oxytocin parental | 390 | 86.0 |
| Use of paratograph | 384 | 84.7 |
| Injectable antibiotic available | 186 | 41.0 |
| Antibiotics parental | 184 | 40.7 |
| Anticonvulsant parental | 44 | 9.6 |
| Facility readiness domain | | |
| General readiness | | |
| Client latrine | 411 | 90.6 |
| Protected client waiting area | 393 | 86.6 |
| Water supply | 386 | 85.0 |
| Electricity service | 295 | 65.1 |
| Emergency transport | 282 | 62.2 |
| 24-hour duty call | 107 | 23.7 |
| Landline phone | 105 | 23.2 |
| Mobile phone | 52 | 11.5 |
| Medicines | | |
| Betadine solution | 415 | 91.6 |
| Intravenous fluid | 410 | 90.5 |
| Tablet oxytocin | 402 | 88.5 |
| Tablet magnesium sulphate | 329 | 72.6 |
| Chlorhexidine tube | 263 | 58.0 |
| Injectable antibiotics | 186 | 41.0 |
| Calcium gluconate | 120 | 26.4 |
| Nifedipine capsule | 87 | 19.1 |
| Equipment | | |
| Autoclave services | 450 | 99.2 |
| Delivery bed | 440 | 96.9 |
| Fetescope | 419 | 92.4 |
| Latex gloves | 421 | 92.8 |
| Infant scale | 414 | 91.3 |
| Sponge holder | 416 | 91.8 |
| Stethoscope | 405 | 89.2 |
| Delivery set | 415 | 91.6 |
| Cord cutting blade | 411 | 90.7 |
| Needle holder | 400 | 88.1 |
| Suturing blade | 386 | 85.0 |
| Bag and Mask | 380 | 83.8 |
| Epitomy set | 376 | 82.8 |
| Blood pressure set | 380 | 83.9 |
| Forceps | 378 | 83.4 |
| Disinfectant | 364 | 80.3 |
| Blank paratograph | 363 | 80.1 |
| Baby wrappers four sets | 306 | 67.6 |
| Thermometer | 361 | 79.6 |
| Vaginal speculum | 354 | 78.1 |
| Soap available in the maternity room | 326 | 72.0 |
| Cord clamper | 323 | 71.2 |
| Water available in the delivery room | 313 | 69.1 |
| Nayano Jhola set | 308 | 68.0 |
| Examination light | 293 | 64.7 |
| Dee-lee suction | 234 | 51.7 |
| Alcohol for hand rub | 118 | 26.1 |
| Staff training and guidelines | | |
| Supervision of health workers | 409 | 90.1 |
| External supervision in the last four months | 326 | 71.8 |
| Exclusive breastfeeding training | 142 | 31.3 |
| Neonatal resuscitation training | 136 | 29.9 |
| Kangaroo Mother Care training | 128 | 28.2 |
| Reproductive health guideline | 128 | 28.2 |
| Cord cutting training | 127 | 27.9 |
| Integrated management of pregnancy and childbirth | 123 | 27.2 |
| Acute management of the third stage of labour | 124 | 27.4 |
| Thermal care | 120 | 26.4 |
| Routine labour and delivery | 118 | 25.9 |
| Maternal and newborn care update emergency obstetric care | 97 | 21.4 |
| Neonatal sepsis management | 88 | 19.4 |
| Other training (e.g., refresher training) | 6 | 1.4 |
## Structural quality of health facilities for MNH services
Fig 1 shows the structural quality scores of health facilities for the first ANC visit. The mean score of the structural quality of health facilities for ANC services was 0.62 (maximum: 1.00), with service availability scoring 0.67 and facility readiness scoring 0.56. Out of four subdomains of facility readiness, the staff and guidelines subdomain had the lowest score (0.31), whereas the highest score (0.72) was for equipment.
**Fig 1:** *Structural quality score of health facilities for the ANC visit.*
Similarly, Fig 2 shows the structural quality score of health facilities for perinatal services. The average structural quality score of health facilities with perinatal services was 0.67 (maximum: 1.00), with higher scores for service availability (0.72) than facility readiness (0.62). The staff and guidelines’ mean score for the facility readiness subdomain was lower (0.31) than equipment (0.82) for perinatal services.
**Fig 2:** *Structural quality score of health facilities for the perinatal services.*
## Distribution of health facilities with optimal structural quality for MNH services
Table 4 shows the distribution of the structural quality of health facilities for MNH services in Nepal. Higher-level health facilities (PHCCs and hospitals) had the highest percentage of the optimal structural quality of health facilities for ANC services ($64\%$) compared to peripheral health facilities (e.g., health posts and clinics) ($18\%$). Staff supervised in the past four months ($29\%$) demonstrated optimal structural quality of health facilities for ANC services than staff without such supervision ($16\%$). Private health facilities ($57\%$) had optimal structural quality of health facilities for ANC services compared to public facilities ($22\%$) (Table 4).
**Table 4**
| Determinants | Categories | Health facilities providing ANC services (N = 901) | Health facilities providing ANC services (N = 901).1 | Health facilities providing ANC services (N = 901).2 | Health facilities providing perinatal services (N = 454) | Health facilities providing perinatal services (N = 454).1 | Health facilities providing perinatal services (N = 454).2 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Determinants | Categories | Frequency | Optimal quality (%) | p | Frequency | Optimal quality (%) | P |
| National average | | 901 | 24.4 | | 454 | 29.3 | |
| Managed by | Private | 64 | 56.7 | <0.001 | 45 | 49.5 | <0.001 |
| Managed by | Public | 837 | 22.0 | | 409 | 27.0 | |
| Facility type | PHCCs and hospitals | 122 | 64.1 | <0.001 | 105 | 58.6 | <0.001 |
| Facility type | Health posts and clinics | 779 | 18.2 | | 349 | 20.4 | |
| Provinces | One | 160 | 19.8 | 0.192 | 78 | 38.3 | <0.001 |
| Provinces | Madhesh | 154 | 18.2 | | 38 | 48.7 | |
| Provinces | Bagmati | 179 | 31.4 | | 80 | 26.2 | |
| Provinces | Gandaki | 116 | 22.7 | | 66 | 18.0 | |
| Provinces | Lumbini | 135 | 28.9 | | 64 | 42.6 | |
| Provinces | Karnali | 68 | 20.1 | | 60 | 12.6 | |
| Provinces | Sudurpaschim | 88 | 28.3 | | 67 | 24.7 | |
| Quality assurance activities | No | 714 | 23.1 | 0.138 | 360 | 27.6 | 0.161 |
| Quality assurance activities | Yes | 187 | 29.5 | | 93 | 35.6 | |
| health facility meeting | Never | 165 | 16.3 | 0.076 | 75 | 22.8 | 0.479 |
| health facility meeting | Sometimes | 129 | 24.7 | | 71 | 29.1 | |
| health facility meeting | Monthly | 607 | 26.6 | | 308 | 30.9 | |
| Feedback collection | No | 489 | 21.2 | | 222 | 21.1 | |
| Feedback collection | Yes | 412 | 28.3 | 0.052 | 232 | 37.0 | 0.002 |
| Supervision of staff | No | 330 | 16.4 | <0.001 | 128 | 24.1 | 0.199 |
| Supervision of staff | Yes | 571 | 29.1 | | 326 | 31.3 | |
Private health facilities ($49\%$) had optimal structural quality for perinatal services than public facilities ($22\%$). On the other hand, the health facilities of the Karnali province ($12\%$) had poor structural quality for perinatal services compared to province one ($38\%$).
## Determinants of the optimal structural quality of health facilities for MNH services
In the bivariable analysis, out of the seven independent determinants examined, five determinants, including structural (management authority), intermediary (types of health facilities, province), and health system (health facility meeting, supervision of staff in the past four months), were significantly associated with optimal structural quality of health facilities for ANC services. However, in the multivariable analysis, private health facilities had higher odds of (aOR = 2.65, $95\%$ CI: 1.48, 4.74) optimal structural quality than public facilities. Health facilities with external supervision in the past four months were more likely (aOR = 1.96, $95\%$ CI: 1.22, 3.13) to have optimal structural quality for ANC services than health facilities without supervision. The peripheral health facilities (e.g., health posts) had poor structural quality for ANC services; for instance, the odds of optimal structural quality were $87\%$ lower in peripheral health facilities compared to higher-level health facilities.
In the bivariable analysis, out of the seven independent determinants, three intermediary (types of health facilities, province, and region) and one health system (feedback collection) determinants were significantly associated with optimal structural quality of health facilities for perinatal services. However, in the multivariable analysis, two determinants (types of health facilities, and province) were significantly associated with optimal structural quality of health facilities for perinatal services. The odds of optimal structural quality of health facilities for perinatal services were $84\%$ lower (aOR = 0.16; $95\%$ CI: 0.10, 0.27) in peripheral health facilities, compared with higher-level health facilities. Similarly, health facilities of Karnali province had $71\%$ lower odds of having optimal structural quality of health facilities for perinatal services compared to province one (Table 5).
**Table 5**
| Determinants | Categories | Health facilities providing ANC services (N = 901) | Health facilities providing ANC services (N = 901).1 | Health facilities providing perinatal services (N = 454) | Health facilities providing perinatal services (N = 454).1 |
| --- | --- | --- | --- | --- | --- |
| Determinants | Categories | cOR (95% CI) | aOR (95% CI) | cOR (95% CI) | aOR (95% CI) |
| Managing authority | Public | 1.00 | 1.00 | 1.00 | 1.00 |
| Managing authority | Private | 4.64(2.88, 7.50) *** | 2.65(1.48, 4.74) ** | 2.60 (1.35, 4.53) *** | 1.69 (1.25, 3.40) ** |
| Facility type | PHCCs and hospitals | 1.00 | 1.00 | 1.00 | 1.00 |
| Facility type | Health posts and clinics | 0.12(0.09, 0.18) *** | 0.13(0.09, 0.18) *** | 0.18(0.12, 0.28) *** | 0.16 (0.10, 0.27) *** |
| Province | One | | | 1.00 | 1.00 |
| Province | Madhesh | 0.90 (0.43, 1.89) | | 1.53 (0.60, 3.88) | 0.86(0.24, 3.11) |
| Province | Bagmati | 1.86(1.02, 3.36) * | | 0.57(0.28, 1.17) | 0.47(0.20, 1.07) |
| Province | Gandaki | 1.19(0.60, 2.35) | | 0.35(0.13, 0.96) * | 0.39(0.12, 1.24) |
| Province | Lumbini | 1.65(0.88, 3.09) | | 1.19(0.55, 2.60) | 1.20(0.47, 3.06) |
| Province | Karnali | 1.02(0.45, 2.29) | | 0.23(0.08, 0.66) ** | 0.29(0.09, 0.99) * |
| Province | Sudurpaschim | 1.60(0.80, 3.17) | | 0.53(0.24, 1.16) | 0.53(0.22, 1.29) |
| Quality assurance activities | No | 1.00 | | 1.00 | |
| Quality assurance activities | Yes | 1.39(0.90, 2.17) | | 1.45(0.86, 2.43) | |
| health facility meeting | No | 1.00 | | 1.00 | |
| health facility meeting | Sometimes | 1.68(0.85, 3.35) | | 1.39(0.59, 3.25) | |
| health facility meeting | Regularly | 1.86(1.07, 3.24) * | | 1.52(0.77, 2.99) | |
| Feedback collection | No | 0.68(0.46, 1.00) | | 0.46(0.28, 0.75) ** | |
| Supervision | No | 1.00 | 1.00 | 1.00 | |
| Supervision | Yes | 2.10(1.37, 3.22) *** | 1.96(1.22, 3.13) ** | 1.44(0.82, 2.50) | |
## Discussion
This study used facility inventory and health workers interview data from the Nepal Health Facility Survey 2015 examined the health facility readiness for MNH services in Nepal. Overall, the availability of trained workforces, and laboratory-related facilities was low than other subdomains of SARA framework. In addition, this study revealed suboptimal structural quality of health facilities for MNH services. While health facilities supervised by a higher authority had optimal quality ANC services, peripheral health facilities had poor quality ANC services. On the other hand, health system readiness in private health facilities had optimal quality for ANC, and perinatal care services. In addition, Karnali province and peripheral areas’ public and private health facilities had poor quality perinatal services.
Health facilities had poor availability of trained workforce, laboratory, and general readiness (e.g., mobile communication, ambulance) for ANC and perinatal services. Findings of poor general readiness were consistent with the previous study conducted in Karnali province, which showed shortage of medicine for women and newborns [43] and many low and low income countries of South Asia and Africa [44]. Competent staff and equipment availability are important components of health facility readiness for quality service delivery [45]. For instance, attending health facilities for ANC is to screen pregnant women with possible complications and timely referral, but without trained workforces and needed equipment to screen difficult to screen complicated pregnancies [45–48]. These are crucial to identifying and managing potential pregnancy and childbirth complications. In addition, availability of trained workforces, equipment, and laboratory services helps build trust with the health system and increase the service users’ engagement with the system [43, 49]. Thus, local health system authorities need to identify competent staff and equipment availability in their catchment health facilities and ensure optimal health system readiness.
Of the domains listed in the SARA manual, the lowest scores were observed for the staff training and guidelines subdomain in all health facilities for ANC, and perinatal services. A study conducted in Southern Nepal also revealed that less than half of the health workers had received the mandated skilled birth attendants training [45]. Reasons behind low scores on this subdomain could be less focused on compliance with standard protocols and continuing education. Health workers focus on training, but low implementation of skills gained after attending training. In addition, there is social desirability of getting more training in some cases if they reported they had not received training in the interview response. The health workforce is vital for optimal health system readiness and quality MNH service delivery. Possible strategies for optimal health facility quality in the staff and guideline subdomain, could be improving the skills of the health workforce through training on essential MNH services and providing materials and guidelines for specific health services [14, 50]. Moreover, optimal readiness for the staff and guideline subdomain can be strengthened by ensuring the supply of essential medicines and equipment at the health facilities. In Nepal, the federal health system provides an opportunity and the resources to strengthen the inputs, such as recruitment of health trained health workforces and supply of training materials at health facilities through effective collaboration with provincial and municipal governments at the local level [51].
In this study, private health facilities had two-fold higher odds of having optimal structural quality for ANC services, and perinatal services compared to public health facilities. Previous studies showed high-quality scores for primary health care services compared to public facilities in Nepal [31] and Bangladesh [52]. Private health facilities are usually urban-centric, and have health infrastructure, equipment and supplies, and availability of health workforce [53]. In private health facilities, compared to public facilities, the client flow is generally low [54], are more responsive, hospitable and client-oriented [55], and have short waiting time [56]. On the other hand, public health facilities are often compromised by inadequate inputs, including human workforce, equipment, and medicine, with adverse effects on health facilities’ readiness [57]. While private health facilities offer optimal quality ANC services, users also incur high out-of-pocket expenses, including routine MNH services, free at the point of services covered by government funding. In urban or remote areas, women with lower socioeconomic status have limited access to private health services in Nepal, increasing client flow to public facilities, which results in overcrowding and receiving poor quality of care [58]. According to private hospitals’ operational guidelines, private health facilities should allocate $10\%$ of beds to disadvantaged populations [24]. Although there are increasing trends in the utilisation of maternal and child health services in Nepal over the last two decades [55], there are still no functional monitoring mechanisms to evaluate if this is implemented [58, 59]. Proper monitoring and facilitation of the implementation of this policy provision could increase access to private health facilities, especially for women of marginalised groups in urban areas. Private health facilities are also eligible to participate in the Government’s Safe Delivery Incentive Program. This maternity incentive program provides a monetary incentive to women who complete 4ANC visits or give birth at health facilities [60], reimburses the health service provider for services delivered, and provides health facilities with a financial incentive in cases of cesarean delivery. Very few private health facilities participate in this program; however, the amount reimbursed is lower than the private health facility charge. Private health facilities also have high rates of caesarian section delivery [61]. High care costs of routine health services in private health facilities partly contribute to Nepal’s high OOP expenditure. Thus, access to private maternity services could be improved through the linkage of the national health insurance program with the private health providers [62], where women can get maternity services, and health insurance program can reimburse the cost of health care in private health facilities. Nonetheless, more than two thirds of women received maternity services from public faculties in Nepal, mostly by women with lower socioeconomic status, and ethic disadvantaged women [60]. Therefore, improving quality in public facilities is vital to reduce the maternal and neonatal deaths.
The study showed peripheral health facilities had poor structural quality for ANC, and perinatal care services. In contrast, health facilities of Karnali province had poor structural quality perinatal services, which are likely to result in poor quality MNH services. These findings resonate with available evidence; for instance, past studies suggest that peripheral health facilities were poorly prepared for quality primary health services in Nepal [30–32], India [63], and Burkina Faso [64]. This is of concern as peripheral health facilities provide most routine ANC, childbirth PNC services [60]. During pandemic, the flow of health services users decreased in referral health facilities such as tertiary referral hospitals, and services in rural/peripheral facilities increased. The findings of this study highlight the need for improvement in the quality of care in peripheral facilities for better MNH outcomes. For decades, the Karnali province has had difficult geographical settings, poor roadworks, and neglected mainstream development. Evidence shows that rural areas and Karnali province had a high burden of maternal and neonatal morbidity and mortality [6] attributed in part to poor quality of MNH care [65], due to a lack of skilled birth attendants in health facilities [47, 66], and inadequate skills to handle MNH complications at peripheral level [49]. Previous studies have also revealed poor MNH outcomes in peripheral and remote areas in Karnali province, in part due to suboptimal health system readiness [30], poor access to quality care [66, 67], and lack of transportation and health facility accessibility [12]. Improving health system readiness needs tailored and customised approaches according to the context. Contextual approaches can be implemented, including recruitment of local health workers and support from local governments in infrastructure development, supplies of medicines and equipment. Thus, policy and program initiatives focus on improving the health facility capacity of remote regions and peripheral health facilities to deliver optimal quality MNH services.
Health facilities supervised by higher authorities had better health system readiness for ANC services. Supervision and monitoring of health facilities in peripheral areas could improve the quality of MNH services in two ways [58]: improving the technical quality of health services and health management functions. Technical mentors could transfer technical skills, observe procedures during monitoring and supervisory visits and provide inputs to improve health workers’ skills. Available evidence from Nepal shows onsite coaching and mentoring from a higher-level authority can improve the quality of MNH services [68]. Management mentors could identify enabling environment of health facilities, including community support, internal inputs, management function such as monthly meetings, and quality assurance mechanisms in the health facilities. However, Nepal’s difficult geographical terrain might be unfeasible for in-person training at the district level. In this context, periodic supervision visits, onsite coaching, and mentoring support to peripheral health facilities and health workers could be important strategies for better health system readiness for quality MNH service delivery.
## Implications for policy and programmes
This study has implications for programs and policies. First, this study identified the poor structural quality of health facilities in Karnali province and peripheral health facilities and public health facilities. Provincial and local municipal governments should prioritise health system inputs, including local health workforce recruitment, and supplies of necessary health commodities and medicines. In Nepal’s federal health system, local governments (municipalities) have autonomy and budgets to address contextual problems, recruit health workforces, and improve essential medicines and supplies. Second, local rural municipalities should supervise and monitor peripheral health facilities. The health section of municipalities could monitor and supervise ward-level health facilities such as health posts, community health clinics in their catchment. Monitoring and supervision visits from the local health officer of municipality could improve the local health workers’ health management functions and technical skills. Local health offices can use service availability and facility readiness framework and identify the availability of subdomain-specific items during the monitoring and supervision of the health facilities. A study from Pakistan revealed supportive supervision, recognition, training, logistics, and salaries were community and health system motivating factors for lady health supervisors, and motivated by their role in providing supportive supervision and supervisory support from their coordinators and managers [69].
Third, private health facilities have optimal structural quality for ANC services, but disadvantaged women have poor access to private health services. Implementing maternity incentive programs in private health facilities, cost-sharing and ensuring allocation of $10\%$ beds in private health facilities for disadvantaged populations could increase the access to private maternity services, especially for disadvantaged women in urban areas. Fourth, this study also highlighted using the SARA manual to collect input information during routine supervision and monitoring visits from higher-level health facilities. Later such information can be used to calculate the subdomain-specific health system response for quality MNH services. Finally, this study used multiple data sources and calculated the quality score at the health facilities level covering multiple dimensions. There are data available at the local level, and local health facility managers can also use data from multiple sources to identify the quality index of health facilities in their catchment.
## Strengths and limitations
This study has some strengths. First, this study analysed the nationally representative survey data and assessed the availability and structural quality of health facilities for MNH services in Nepal. So, the findings of this study are generalisable for all regions of Nepal. Second, this important study considered a wide range of health items needed to deliver routine MNH services. The SARA manual, other guidelines on maternity care, and national standard recommended several items/interventions needed to deliver quality health service. This study accounted wide range of items to estimate the quality score. Third, we used multiple data sources, and identified the composite quality scores of health facilities based on the data derived from observation, such as medicines and equipment in health facilities by a trained enumerator. Therefore, findings might be more reliable compared to the perceived quality of care assessment. Limitations of this included, first, we analysed data from NHFS 2015 conducted five years earlier; therefore, the data may not reflect recent conditions of Health facilities in the federal health system context of Nepal. However, this study used recent nationwide health facility survey data, thus can give the overall picture of health system readiness for MNH services. Second, NHFS is a cross-sectional survey; the inferences indicate the correlation rather than causality. Third, the outcome variable’s score distribution did not allow us to run the linear regression. Therefore, due to data distribution and structure, we dichotomised scores taking the mean as cut-off point to run logistic regression [39]. Finally, measuring facility readiness and structural quality is difficult. Some researchers have raised concerns about which items are included (vs. excluded) in the creation of scores [70, 71], as well as concerns about the poor correlation of readiness scores with observed service quality [72]. However, adapting the SARA framework [34], national standards [36], and previous studies [13, 38, 73], we created score of structural quality of health facilities for MNH services. Using secondary data for the analysis always has its limitations, and important information might not be available for analysis. Nevertheless, we have included all available information and analysed the recent national-level facility survey data. Findings and methods used in this analysis could be a reference for future research. Authors’ experience with the health system also suggests that Nepalese health facilities are constrained from many health systems inputs, shortage of medicine, equipment, health workforce and general readiness. Finally, this quantitative study could not provide underling factors of suboptimal quality of care, thus, future qualitative studies can explore contextual factors associated with maternal continuum of care.
## Conclusions
Health facilities in Nepal had sub-optimal structural quality of MNH services across the continuum of care, especially health facilities in rural areas and publicly managed. Health facilities were poorly equipped with staff, training, and laboratory-related equipment and services. Private health facilities and health facilities supervised by higher authorities had optimal structural quality for MNH services, while peripheral health facilities, and health facilities of Karnali province had poor structural quality for MNH services. Maternity and newborn incentive programs such as maternity and newborn incentive program can be implemented in private health facilities to use maternity services in those facilities at the subsidised cost reimbursed by the program. There is an urgent need for policy reform to improve the MNH services, particularly in the public and health facilities of Karnali province. Provincial and local governments should focus on improving the health system inputs, including trained health workers, supply of essential medicines, and provision of laboratory-related equipment in those areas. Achieving universal health coverage will not be possible without strengthening service availability and facility readiness in public and peripheral health facilities.
## References
1. Darmstadt GL. **Global perinatal health: accelerating progress through innovations, interactions, and interconnections**. (2010.0) **34** 367-370. DOI: 10.1053/j.semperi.2010.09.003
2. 2WHO: WHO recommendations on antenatal care for a positive pregnancy experience: World Health Organization; 2016.
3. 3WHO: WHO recommendations on postnatal care of the mother and newborn: World Health Organization; 2014.
4. 4WHO: Strategies towards ending preventable maternal mortality (EPMM). In.: WHO Geneva,; 2015.
5. 5WHO & UNICEF: Pregnancy, childbirth, postpartum and newborn care: A guide for essential practice. In. Geneva; 2015.
6. 6Ministry of Health, New ERA and, ICF: Nepal Demographic and Health Survey 2016. In. Kathmandu; 2017.
7. Benova L, Tunçalp Ö, Moran AC, Campbell OMR. **Not just a number: examining coverage and content of antenatal care in low-income and middle-income countries**. (2018.0) **3** e000779. DOI: 10.1136/bmjgh-2018-000779
8. Khatri RB, Alemu Y, Protani MM, Karkee R, Durham J. **Intersectional (in) equities in contact coverage of maternal and newborn health services in Nepal: insights from a nationwide cross-sectional household survey**. (2021.0) **21** 1-12. PMID: 33388037
9. Malqvist M, Pun A, Raaijmakers H, Kc A. **Persistent inequity in maternal health care utilization in Nepal despite impressive overall gains**. (2017.0) **10** 1356083. DOI: 10.1080/16549716.2017.1356083
10. Thapa J, Budhathoki SS, Gurung R, Paudel P, Jha B, Ghimire A, Kc A. **Equity and coverage in the continuum of reproductive, maternal, newborn and child health services in Nepal-projecting the estimates on death averted using the LiST tool**. (2020.0) **24** 22-30. DOI: 10.1007/s10995-019-02828-y
11. Mehata S, Paudel YR, Dariang M, Aryal KK, Lal BK, Khanal MN. **Trends and inequalities in use of maternal health care services in Nepal: Strategy in the search for improvements**. (2017.0) **2017** 5079234. DOI: 10.1155/2017/5079234
12. Byrne A, Hodge A, Jimenez-Soto E, Morgan A. **Looking beyond supply: a systematic literature review of demand-side barriers to health service utilization in the mountains of Nepal**. (2013.0) **25** 438-451. DOI: 10.1177/1010539513496139
13. Khatri RB, Durham J, Assefa Y. **Utilisation of quality antenatal, delivery and postnatal care services in Nepal: An analysis of Service Provision Assessment**. (2021.0) **17** 1-16. PMID: 33397377
14. Kruk ME, Gage AD, Arsenault C, Jordan K, Leslie HH, Roder-DeWan S. **High-quality health systems in the Sustainable Development Goals era: time for a revolution**. (2018.0) **6** e1196-e1252. DOI: 10.1016/S2214-109X(18)30386-3
15. Donabedian A. **The quality of care. How can it be assessed?**. (1988.0) **260** 1743-1748. DOI: 10.1001/jama.260.12.1743
16. 16WHO: Delivering quality health services: a global imperative for universal health coverage. In.; 2018.
17. Gardner G, Gardner A, O’connell J. **Using the Donabedian framework to examine the quality and safety of nursing service innovation**. (2014.0) **23** 145-155. DOI: 10.1111/jocn.12146
18. Ameh S, Gomez-Olive FX, Kahn K, Tollman SM, Klipstein-Grobusch K. **Relationships between structure, process and outcome to assess quality of integrated chronic disease management in a rural South African setting: applying a structural equation model**. (2017.0) **17** 229. DOI: 10.1186/s12913-017-2177-4
19. Hogan DR, Stevens GA, Hosseinpoor AR, Boerma T. **Monitoring universal health coverage within the Sustainable Development Goals: development and baseline data for an index of essential health services**. (2018.0) **6** e152-e168. DOI: 10.1016/S2214-109X(17)30472-2
20. 20WHO: Global Strategy for Women’s, Children’s and Adolescents’ Health 2016–2030. In: Geneva: WHO. 2015.
21. 21Organization WH: Every Newborn: an action plan to end preventable deaths. 2014.
22. 22WHO: World health statistics 2016: Monitoring health for the sustainable development goals: World Health Organization; 2016.
23. Chou VB, Walker N, Kanyangarara M. **Estimating the global impact of poor quality of care on maternal and neonatal outcomes in 81 low- and middle-income countries: A modeling study**. (2019.0) **16** e1002990. DOI: 10.1371/journal.pmed.1002990
24. 24MOHP [Nepal]: Nepal health sector strategy 2016–2021. In. Kathmandu,; 2015.
25. 25Family Health Division: Strategy for Skilled Health Personnel and Skilled Birth Attendants 2020–2025. In. Kathmandu Family Welfare Divison, MOHP, Nepal; 2021.
26. 26Family Welfare Division: Nepal Safe Motherhood and Newborn Health Road Map 2030. In. Ministry of Health, Kathmandu, Nepal; 2019
27. 27WHO: 2018 Health SDG Profile: Nepal. In.; 2018.
28. 28MOHP [Nepal]: Nepal health facility survey 2015. In. Kathmandu, Nepal 2017.
29. 29Government of Nepal MoH: Nepal Health Infrastructure Development Standards 2017 (Unofficial English Translation). In. Ramshahpath, Kathmandu; 2019.
30. Ghimire U, Shrestha N, Adhikari B, Mehata S, Pokharel Y, Mishra SR. **Health system’s readiness to provide cardiovascular, diabetes and chronic respiratory disease related services in Nepal: analysis using 2015 health facility survey**. (2020.0) **20** 1163. DOI: 10.1186/s12889-020-09279-z
31. Acharya K, Paudel YR. **General health service readiness and its association with the facility level indicators among primary health care centers and hospitals in Nepal**. (2019.0) **3** 1-12
32. Acharya K, Subedi RK, Dahal S, Karkee R. **Basic emergency obstetric and newborn care service availability and readiness in Nepal: Analysis of the 2015 Nepal Health Facility Survey**. (2021.0) **16** e0254561. DOI: 10.1371/journal.pone.0254561
33. 33WHO: Compilation of WHO recommendations on maternal, newborn, child and adolescent health. In. Geneva; 2013.
34. 34WHO: Service availability and readiness assessment (SARA): an annual monitoring system for service delivery: reference manual. In. Geneva; 2013.
35. O’Neill K, Takane M, Sheffel A, Abou-Zahr C, Boerma T. **Monitoring service delivery for universal health coverage: the Service Availability and Readiness Assessment**. (2013.0) **91** 923-931. DOI: 10.2471/BLT.12.116798
36. 36MoHP[Nepal]: National Medical Standard for Maternal and Newborn Care. In. Edited by Populaiton MoHa, vol. III. Kathmandu; 2020.
37. Khatri RB, Durham J, Assefa Y. **Investigation of technical quality of antenatal and perinatal services in a nationally representative sample of health facilities in Nepal**. (2022.0) **80** 1-14. PMID: 34983643
38. Sharma J, Leslie HH, Kundu F, Kruk ME. **Poor quality for poor women? Inequities in the quality of antenatal and delivery care in Kenya**. (2017.0) **12** e0171236. DOI: 10.1371/journal.pone.0171236
39. Wilhelm D, Lohmann J, De Allegri M, Chinkhumba J, Muula AS, Brenner S. **Quality of maternal obstetric and neonatal care in low-income countries: development of a composite index**. (2019.0) **19** 154. DOI: 10.1186/s12874-019-0790-0
40. Sharew NT, Bizuneh HT, Assefa HK, Habtewold TD. **Investigating admitted patients’ satisfaction with nursing care at Debre Berhan Referral Hospital in Ethiopia: a cross-sectional study**. (2018.0) **8** e021107. DOI: 10.1136/bmjopen-2017-021107
41. Akinwande MO, Dikko HG, Samson A. **Variance Inflation Factor: As a Condition for the Inclusion of Suppressor Variable(s) in Regression Analysis**. (2015.0) **05** 754-767
42. Lemes IR, Sui X, Fernandes RA, Blair SN, Turi-Lynch BC, Codogno JS. **Association of sedentary behavior and metabolic syndrome**. (2019.0) **167** 96-102. DOI: 10.1016/j.puhe.2018.11.007
43. Tamang P, Simkhada P, Bissell P, van Teijlingen E, Khatri R, Stephenson J. **Health facility preparedness of maternal and neonatal health services: a survey in Jumla, Nepal**. (2021.0) **21** 1-10. PMID: 33388053
44. Leslie HH, Spiegelman D, Zhou X, Kruk ME. **Service readiness of health facilities in Bangladesh, Haiti, Kenya, Malawi, Namibia, Nepal, Rwanda, Senegal, Uganda and the United Republic of Tanzania**. (2017.0) **95** 738-748. DOI: 10.2471/BLT.17.191916
45. Lama TP, Munos MK, Katz J, Khatry SK, LeClerq SC, Mullany LC. **Assessment of facility and health worker readiness to provide quality antenatal, intrapartum and postpartum care in rural Southern Nepal**. (2020.0) **20** 16. DOI: 10.1186/s12913-019-4871-x
46. Khatri RB, Karkee R. **Social determinants of health affecting utilisation of routine maternity services in Nepal: a narrative review of the evidence**. (2018.0) **26** 32-46. DOI: 10.1080/09688080.2018.1535686
47. Rajbhandari R, Rai S, Hathi S, Thapa R, Rai I, Shrestha A. **The quality of skilled birth attendants in Nepal: High aspirations and ground realities**. (2019.0) **14** e0214577-e0214577. DOI: 10.1371/journal.pone.0214577
48. Rajbanshi S, Norhayati MN, Nik Hazlina NH. **High-risk pregnancies and their association with severe maternal morbidity in Nepal: A prospective cohort study**. (2020.0) **15** e0244072. DOI: 10.1371/journal.pone.0244072
49. Wahlström S, Björklund M, Munck B. **The professional role of skilled birth attendants’ in Nepal—A phenomenographic study**. (2019.0) **21** 60-66. DOI: 10.1016/j.srhc.2019.05.003
50. Scott K, Beckham SW, Gross M, Pariyo G, Rao KD, Cometto G. **What do we know about community-based health worker programs? A systematic review of existing reviews on community health workers**. (2018.0) **16** 39. DOI: 10.1186/s12960-018-0304-x
51. Thapa R, Bam K, Tiwari P, Sinha TK, Dahal S. **Implementing federalism in the health system of Nepal: Opportunities and challenges**. (2019.0) **8** 195-198. DOI: 10.15171/ijhpm.2018.121
52. Yusuf SS, Acharya K, Ahmed R, Ahmed A. **Understanding general health service readiness and its correlates in the health facilities of Bangladesh: evidence from the Bangladesh Health Facility Survey 2017**. (2021.0)
53. Gyawali B, Sharma S, Shilpakar R, Dulal S, Pariyar J, Booth CM. **Overview of Delivery of Cancer Care in Nepal: Current Status and Future Priorities**. (2020.0) 1211-1217. DOI: 10.1200/GO.20.00287
54. KC A, Peterson SS, Gurung R, Skalkidou A, Gautam J, Malla H. **The perfect storm: Disruptions to institutional delivery care arising from the COVID-19 pandemic in Nepal**. (2021.0) **11**
55. Karkee R, Lee AH, Pokharel PK. **Womens’ perception of quality of maternity services: a longitudinal survey in Nepal**. (2014.0) **14** 45-45. DOI: 10.1186/1471-2393-14-45
56. Ashworth HC, Roux TL, Buggy CJ. **Healthcare accessibility in the rural plains (terai) of Nepal: physical factors and associated attitudes of the local population**. (2019.0) **11** 528-535. DOI: 10.1093/inthealth/ihz008
57. Kc A, Singh DR, Upadhyaya MK, Budhathoki SS, Gurung A, Malqvist M. **Quality of care for maternal and newborn health in health facilities in Nepal**. (2020.0) **24** 31-38. DOI: 10.1007/s10995-019-02846-w
58. 58Sapkota T, Houkes I, Bosma H: Vicious cycle of chronic disease and poverty: a qualitative study in present day Nepal. Int Health 2020.
59. 59USAID/Nepal: USAID/Nepal Health Private Sector Engagement Assessment In.; 2017.
60. 60Ministry of Health: Annual Report, Department of Health Services 2075/2076 (2018/2019). In.: Ministry of Health, Department of Health Services. Kathmandu, Nepal.; 2020.
61. Tamang L, Goma D, Kumariniraula H, Roshnitui T, Binod A. **Rising cesarean section rates in Nepal: Question of safety and integrity on obstetric emergency practice**. (2017.0) **7** 555716
62. Ranabhat CL, Subedi R, Karn S. **Status and determinants of enrollment and dropout of health insurance in Nepal: an explorative study**. (2020.0) **18** 40. PMID: 33013204
63. Pakhare A, Kumar S, Goyal S, Joshi R. **Assessment of primary care facilities for cardiovascular disease preparedness in Madhya Pradesh, India**. (2015.0) **15** 408. DOI: 10.1186/s12913-015-1075-x
64. Millogo O, Doamba JEO, Sié A, Utzinger J, Vounatsou P. **Constructing a malaria-related health service readiness index and assessing its association with child malaria mortality: an analysis of the Burkina Faso 2014 SARA data**. (2021.0) **21** 20. DOI: 10.1186/s12889-020-09994-7
65. Joshi D, Baral S, Giri S, Kumar A. **Universal institutional delivery among mothers in a remote mountain district of Nepal: what are the challenges?**. (2016.0) **6** 267-272. DOI: 10.5588/pha.16.0025
66. Karkee R, Tumbahangphe KM, Maharjan N, Budhathoki B, Manandhar D. **Who are dying and why? A case series study of maternal deaths in Nepal**. (2021.0) **11** e042840. DOI: 10.1136/bmjopen-2020-042840
67. Sanjel K, Onta SR, Amatya A, Basel P. **Patterns and determinants of essential neonatal care utilization among underprivileged ethnic groups in Midwest Nepal: a mixed method study**. (2019.0) **19** 310. DOI: 10.1186/s12884-019-2465-6
68. Goyet S, Rajbhandari S, Alvarez VB, Bayou A, Khanal S, Pokhrel TN. **On-site clinical mentoring as a maternal and new-born care quality improvement method: evidence from a nurse cohort study in Nepal**. (2020.0) **19** 3. DOI: 10.1186/s12912-019-0396-1
69. Rabbani F, Shipton L, Aftab W, Sangrasi K, Perveen S, Zahidie A. **Inspiring health worker motivation with supportive supervision: a survey of lady health supervisor motivating factors in rural Pakistan**. (2016.0) **16** 397. DOI: 10.1186/s12913-016-1641-x
70. Sheffel A, Zeger S, Heidkamp R, Munos MK. **Development of summary indices of antenatal care service quality in Haiti, Malawi and Tanzania**. (2019.0) **9** e032558. DOI: 10.1136/bmjopen-2019-032558
71. Stierman EK, Ahmed S, Shiferaw S, Zimmerman LA, Creanga AA. **Measuring facility readiness to provide childbirth care: a comparison of indices using data from a health facility survey in Ethiopia**. (2021.0) **6(** e006698. DOI: 10.1136/bmjgh-2021-006698
72. Leslie HH, Sun Z, Kruk ME. **Association between infrastructure and observed quality of care in 4 healthcare services: A cross-sectional study of 4,300 facilities in 8 countries**. (2017.0) **14** e1002464. DOI: 10.1371/journal.pmed.1002464
73. Galvin G, Hirschhorn LR, Shaikh M, Maji P, Delaney MM, Tuller DE. **Availability of safe childbirth supplies in 284 facilities in Uttar Pradesh, India**. (2019.0) **23** 240-249. DOI: 10.1007/s10995-018-2642-7
|
---
title: 'Identity management in the face of HIV and intersecting stigmas: A metasynthesis
of qualitative reports from sub-Saharan Africa'
authors:
- Alanna J. Bergman
- Katherine C. McNabb
- Khaya Mlandu
- Alvine Akumbom
- Dalmacio Dennis Flores
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022386
doi: 10.1371/journal.pgph.0000706
license: CC BY 4.0
---
# Identity management in the face of HIV and intersecting stigmas: A metasynthesis of qualitative reports from sub-Saharan Africa
## Abstract
While stigma experienced by people living with HIV (PLWH) is well documented, intersectional stigma and additional stigmatized identities have not received similar attention. The purpose of this metasynthesis is to identify salient stigmatized intersections and their impact on health outcomes in PLWH in sub-Saharan Africa. Using Sandelowski and Barroso’s metasynthesis method, we searched four databases for peer-reviewed qualitative literature. Included studies [1] explored personal experiences with intersecting stigmas, [2] included ≥1 element of infectious disease stigma, and [3] were conducted in sub-Saharan Africa. Our multinational team extracted, aggregated, interpreted, and synthesized the findings. From 454 screened abstracts, the 34 studies included in this metasynthesis reported perspectives of at least 1258 participants (282 men, 557 women, and 109 unspecified gender) and key informants. From these studies, gender and HIV was the most salient stigmatized intersection, with HIV testing avoidance and HIV-status denial seemingly more common among men to preserve traditional masculine identity. HIV did not threaten female identity in the same way with women more willing to test for HIV, but at the risk of abandonment and withdrawal of financial support. To guard against status loss, men and women used performative behaviors to highlight positive qualities or minimize perceived negative attributes. These identity management practices ultimately shaped health behaviors and outcomes. From this metasynthesis, the Stigma Identity Framework was devised for framing identity and stigma management, focusing on role expectation and fulfillment. This framework illustrates how PLWH create, minimize, or emphasize other identity traits to safeguard against status loss and discrimination. Providers must acknowledge how stigmatization disrupts PLWH’s ability to fit into social schemas and tailor care to individuals’ unique intersecting identities. Economic security and safety should be considered in women’s HIV care, while highlighting antiretrovirals’ role in preserving strength and virility may improve care engagement among men.
## Introduction
Stigma associated with HIV is well described in the literature as a barrier to patient engagement throughout the care continuum. Stigma can be defined as a negatively perceived attribute that precludes an individual from full social acceptance and contributes to social, financial and/or health related inequity [1, 2]. Researchers have identified associations between HIV stigma and healthcare avoidance [3], poorer ART adherence [3–6], anxiety [7] and depression [4, 6–8]. However, HIV stigma is not the only type of stigma that individuals encounter in their daily lives [9, 10]. Each community has its own norms and values. Violations of these norms can, and often do, lead to stigma. In addition to the stigma of living with HIV, individuals are subject to a variety of other stigmas depending on the cultural context, including but not limited to gender stigma, sexual identity stigma, mental health and substance use stigma, and stigma related to other communicable illnesses. People may encounter stigma in different degrees or in variable qualities due to particular combinations of stigmatized identities. Intersectional stigma is the convergence of more than one of these stigmatized identities. Intersectional stigma acknowledges that living with multiple stigmatized identities does not simply create an additive effect; the negative ramifications of intersectional stigma can be multiplicative or exponential [11]. Intersectional stigma also accounts for factors that ease or mitigate stigma [11].
Infectious disease stigma stemming in part from fear of contagion [12, 13] is well documented across history, and is codified in biblical references; but the HIV epidemic introduced a new facet of infectious disease stigma. Because of its association with sexual behavior and injection drug use, HIV stigma became entwined with questions of morality, invoking stereotypes and debates about blame [14]. Financial loss, religious beliefs, and other institutional authoritarianism can also precede community level stigma impacting internalized, perceived, and experienced stigma at the individual level [1]. Internalized stigma refers to how individuals see themselves as a result of stigma. Perceived stigma references how an individual believes the wider community views members of the stigmatized group. Finally, negative attitudes, insults, and discrimination experienced by the stigmatized person is commonly called enacted stigma [1, 15].
Stigmatization is a societal mechanism for the differentiation and negative stereotyping of people living with stigmatized identities which then leads to their poorer quality of life [16]. Explanations of social categorization and intergroup behavior contribute additional context for our interpretations of stigma. Humans take pride in, and extract meaning from social inclusion; each person situates themselves and others into groups that help to explain belonging and identity [1, 2, 16, 17]. These context dependent social categories often lead to conflict between groups in an effort to ascribe positive attributes to one’s own social schema resulting in negative reflection of other social groups [15–17]. As a result, salient intersections contributing to stigma vary widely across geography, culture, and community.
Reducing infectious disease stigma is a public health priority acknowledged by both the World Health Organization and the Joint United Nations Programme on HIV/AIDS [18, 19]. Because intersectional stigma may include a complex combination of identities and attributes that contribute both intensifying and mitigating factors, intersectional stigma is difficult to quantify. To date, there is no validated measure of intersectional stigma. Qualitative research methodologies allow rigorous exploration of intersectional stigma consistent with the theoretical underpinnings of the construct. Qualitative research will help to inform the scope and impact of intersectional infectious disease stigma, and future interventions towards reducing these stigmas.
The purpose of qualitative meta-synthesis is to bridge the gap between individual studies and to identify broad truths between the lived experiences of individuals [20]. Although intersectional infectious disease stigma in the United States and *Canada is* well studied [5, 21–23], it is less studied in sub-Saharan Africa. Twenty-two of the top 30 high burden TB/HIV countries are located in Sub-Saharan Africa [24]. Therefore, our primary purpose in conducting this review was to synthesize the current body of qualitative evidence in order to determine the intersections that impact infectious disease stigma in sub-Saharan Africa and consider future stigma mitigation strategies.
## Methods
Qualitative meta-synthesis is a methodic and rigorous process to identify, abstract and synthesize qualitative data pertinent to a specific research question [20]. We used the process outlined by Sandelowski and Barroso to guide our methods [20]. A thorough search of the literature was conducted in conjunction with a library informationist across four databases, Medline, CINAHL, PsychINFO, and African Index Medicus. Because intersectional stigma is a relatively new phenomenon of interest with broad definitions, we used a combination of search strategies to maximize recall rather than precision seeking the greatest number of relevant documents [20]. Search terms for intersectional stigma included: intersectional stigma; double stigma; layered stigma, and multiple stigmas. Following PRISMA guidelines, two independent reviewers initially performed title and abstract screening on all studies, and then moved to full text screening, discarding studies that were irrelevant to our study purpose. Inclusion criteria were: studies conducted in sub-Saharan Africa, performed an analysis of intersecting stigmas, explored the experiences of people impacted by stigma [through experiential or key informant testimony], included at least one element of infectious disease stigma [HIV, TB, hepatitis B, hepatitis C, human papilloma virus etc.] and had a qualitative approach in the study design. The infectious diseases included in the search strategy were chosen based on geographic disease burden and the stigma literature. We aimed for an adaptable interpretation of intersectional stigma to include studies that contained thematically relevant findings even if the main study purposes were not focused on stigma as a main outcome [20]. We did not exclude studies by publication date or qualitative methodology.
Each included study was imported into F4analyse (Marburg, Germany), a computer-assisted qualitative data analysis software which was used to store and organize data. Together the two reviewers double coded five randomly selected studies to develop an initial codebook. Each reviewer was then assigned additional articles to code independently. Every fifth study was double coded to ensure congruence. Reviewers independently memo-ed reflecting on emerging themes and possible biases. The coders met weekly with a South African researcher to discuss the analytic process and consider varying viewpoints with a cultural insider. New codes were added iteratively. The team used elements of reciprocal translation, imported concepts and use of event timelines to group codes and develop larger themes from the individual studies [20]. After extracting the main themes, the study team developed a conceptual model that captured the key elements of the synthesis.
The team consisted of researchers from diverse racial and national/ethnic backgrounds. The strength of this diversity is that several of the marginalized groups included in the studies were represented on the research team. This allowed us to constantly reflect on our background, positionality, and privilege given our concomitant roles as emic and etic researchers throughout the review. Team members from South Africa and Cameroon allowed us to consider cultural viewpoints, clarify language and interrogate western bias in global research. The methods expert guiding the analysis and synthesis of the project is a non-White male researcher born outside of the United States with expertise in qualitative meta-synthesis. Collectively, our team has a longstanding track-record of working with PLWH, marginalized groups and in global health work.
## Results
The search generated an initial 454 studies for title and abstract review. Forty-nine studies moved on to full text screening. References from included studies were reviewed to identify potentially relevant research that was not found in the initial search. We also reviewed recent publications from the work of global stigma researchers to identify studies not yet indexed. Through this process, we identified two additional papers that met our inclusion criteria. These papers were imported directly into full text review and subsequently included. Ultimately, 34 studies were included for full abstraction and analysis (See Fig 1).
**Fig 1:** *PRISMA diagram.*
Most of the included studies were conducted in either South Africa ($$n = 10$$) or Uganda ($$n = 9$$). Three studies were conducted in Malawi, two in Botswana and individual studies took place in Cameroon, Ethiopia, Ghana, Kenya, Nigeria, Rwanda, Swaziland, Tanzania, Zambia, and Zimbabwe. This meta-synthesis represents the experiences and perspectives of at least of 1,228 individuals from sub-Saharan Africa. Of these, 918 were people testifying to the lived experience of stigma. An additional 310 were key informants that included healthcare workers, program administrators and care givers who reported on internalized, anticipated, and experienced stigma from an observers’ perspective (See Table 1).
**Table 1**
| Type of participant | N (%) |
| --- | --- |
| Experiential participants | 948 (75%) |
| • Female | 557 (59%) |
| • Male | 282 (30%) |
| • Unknown gender | 109 (11%) |
| Key informants | 310 (25%) |
| Total | 1258 |
All of the studies included HIV as a stigmatized identity in their intersectional analysis. The intersections of interest included: tuberculosis and HIV [8], gender and HIV [6], mental health and HIV [5], older age and HIV [3], non-communicable disease such as hypertension or diabetes and HIV [3], lesbian, gay, bisexual, trans, and queer (LGBTQ) communities and HIV [3], substance use and HIV [2], occupational nursing and HIV [2], refugee status and HIV [1] and one study looked at race, socioeconomic status, and HIV. These intersections are shown in S1 Table. Studies that investigated gender as a stigmatized identity did so in the context of gender-based violence, pregnancy, and by comparing the experiences of cis-gender women to that of cis-gender men (intra-categorical stigma) [25]. Studies that evaluated the experiences of HIV within the LBGTQ community included men who have sex with men (MSM) and transgender women. Taken together, the included studies offer an explanatory model for framing identity and stigma in sub-Saharan Africa focused on role expectation and fulfillment. Table 2 shows the characteristics of the included studies.
**Table 2**
| First author & year | Country | Stigma is study objective (Y/N) | Intersection of interest | Conceptual framework | Number and type of participant | Experiential or key informant |
| --- | --- | --- | --- | --- | --- | --- |
| Angwenyi 2018 [38] | Malawi | N–purpose was to examine patient self-management of HIV and other chronic illnesses | HIV + non-communicable disease (hypertension, epilepsy, stroke, asthma etc). | Bandura’s theory of self-efficacy | 14 in-depth interviews33 focus group participants | Experiential participants only |
| Becker 2019 [39] | Botswana | Y | HIV + mental health | Kleinman’s explanatory modelLink & Phelan’s theory of stigma | 42 in-depth interviews | Both |
| Brown 2018 [40] | South Africa | N–study purpose was to develop a curriculum for intersectional HIV education | HIV + race + socioeconomic status | Intersectionality theory | 86 participants using photovoice, narratives, drawing and self-reflective assignments | Key informants only |
| Buregyeya 2012 [41] | Uganda | N–explore healthcare workers utilization of occupational TB & HIV services | HIV + TB | None identified | 8 focus groups (total number of participants not reported) | Experiential participants only |
| Chileshe 2010 [42] | Zambia | N–understand experiences of antiretroviral therapy access for people living with TB and HIV | HIV + TB | None identified | 9 index participants and their households’ using anthropological assessments | Both |
| Crankshaw 2014 [43] | South Africa | N–study objective was to explore disclosure | HIV + unintentional pregnancy | None identified | 62 in-depth interviews | Experiential participants only |
| Daftary 2007 [44] | South Africa | N–explore TB status disclosure and the decision-making process for HIV testing acceptance or refusal | HIV + TB | None identified | 21 in-depth interviews | Experiential participants only |
| Daftary 2012a [45] | South Africa | Y | HIV + TB | Stigma theory from Goffman, Phelan & Link and Farmer | 40 in-depth interviews | Experiential participants only |
| Daftary 2012b [46] | South Africa | N–study looked at care seeking among people living with TB and HIV | HIV + TB | None identified | 40 in-depth interviews | Experiential participants only |
| Ezeanolue 2020 [47] | Nigeria | N–explored barriers to integrating mental health care with HIV care | HIV + mental health | None identified | 80 focus group participants | Key informants only |
| Finnie 2010 [48] | South Africa | N–study explored perceptions of TB and TB care-seeking | HIV + TB | None identified | 12 in-depth interviews | Key informants only |
| Freeman 2017 [49] | Malawi | N–study explored identity among older adults living with HIV | HIV + aging | Burke’s identity control theory and interactionist framework | 43 in depth interview30–45 focus group participants | Experiential participants only |
| Gebremariam 2010 [50] | Ethiopia | Y | HIV + TB | Cumings theoretical framework (1980). | 24 in-depth interviews14 focus group participants | Both |
| Gnauck 2013 [51] | Kenya | Y | HIV + gender | None identified | 60 focus group participants | Experiential participants only |
| Jani 2021 [52] | Tanzania | Y | HIV prevention + gender | Framework for PrEP introduction for adolescent girls and young women | 28 in-depth interviews | Experiential participants only |
| Kellett 2016 [53] | Uganda | Y | HIV + gender | Tsai, Bangsberg & Weiser’s conceptualization of HIV stigma | 54 focus group participants | Experiential participants only |
| Kennedy 2013 [54] | Swaziland | N–understand the health, dignity and prevention needs of MSM living with HIV | HIV + LGBTQ (MSM) | Positive health, dignity and prevention framework | 62 participants in focus groups and in-depth interviews | Both |
| King 2019 [55] | Uganda | N–explore gender identity and expression as they relate to HIV risk, healthcare seeking and STI prevention | HIV + LGBTQ (transgender women) | Syndemic theory on gender identity, HIV, stigma and social determinants of health | 45 in-depth interviews | Experiential participants only |
| Kuteesa 2012 [56] | Uganda | Y | HIV + aging | None identified | 40 in-depth interviews and focus group participants | Experiential participants only |
| Kyakuwa 2009 [57] | Uganda | Y | HIV + nurses | None identified | 6 Nurses living with HIV were interviewed using anthropological methods [life histories, observation, informal conversation, diary analysis etc]. | Experiential participants only |
| Kyakuwa 2012 [58] | Uganda | N–explore the relationship between HIV expert clients and HIV nurses | HIV + nurses/healthcare workers | None identified | 67 in-depth interviews and anthropological assessments | Both |
| Lemasters 2020 [59] | Malawi | N–study explored experiences of post-natal depression among women living with HIV | HIV + mental health (post-natal depression) | None identified | 24 in-depth interviews | Experiential participants only |
| Logie 2021 [60] | Uganda | N–principally studied HIV self-testing among urban refugee youth | HIV testing + refugee status | Health Stigma and Discrimination Framework | 5 focus group discussions | Both |
| Magidson 2019 [61] | South Africa | N–this was a qualitative implementation study used to tailor an intervention for HIV and substance use disorder treatment integration | HIV + substance use | RE-AIM framework | 30 in-depth interviews | Both |
| Matima 2018 [62] | South Africa | N–study looked at patient experiences with HIV/DMII multi-morbidity | HIV + Non communicable disease (diabetes mellitus II) | Cumulative complexity model | 16 in-depth interviews | Both |
| Matlho 2017 [63] | Botswana | N–develop an understanding of barriers and facilitators towards HIV initiatives tailored for older adults | HIV + aging | Shiffman and Smith’s framework on determinants of political priority for global initiatives | 15 in-depth interviews | Key informants only |
| Mburu 2014 [64] | Uganda | N–explore community-based peer support groups and engagement in peer HIV support | HIV + gender | Wyrod’s frameworkIntersectionality theory | 25 in-depth interviews40 focus group participants | Both |
| Mugisha 2020 [65] | Uganda | N–explored barriers to HIV care engagement from the provider perspective | HIV + mental health | None identified | 15 in-depth interviews | Key informants only |
| Njozing 2010 [66] | Cameroon | N–understand barriers and facilitators of HIV testing among people living with TB | HIV + TB | None identified | 12 in-depth interviews | Experiential participants only |
| Owusu 2020 [67] | Ghana | N–study analyzed experiences of HIV along gender lines | HIV + gender | None identified | 38 in-depth interviews | Experiential participants only |
| Regenauer 2020 [69] | South Africa | Y | HIV + substance use | Intersectional stigma framework by Bowleg & Turan | 30 in-depth interviews | Both |
| Russel 2016 [68] | Rwanda | N–describe trauma [gender-based violence and HIV infection] experienced by women due to conflict | HIV + gender(Gender-based violence) | None identified | 22 in-depth interviews | Experiential participants only |
| Tokwe 2020 [70] | South Africa | N–aimed to explore experiences of living with HIV and hypertension | HIV + non-communicable disease(HTN) | None identified | 9 in-depth interviews | Experiential participants only |
| Tsang 2019 [71] | Zimbabwe | Y | HIV + LGBTQ(MSM) | Scambler’s sociological perspective | 15 in-depth interviews | Experiential participants only |
## Conceptual framework
The Stigma Identity Framework captures (Fig 2) the major themes and their inter-relationship. The framework fits together elements of intersectionality, stigma theory and identity theory in order to holistically depict the experiences of the people in these studies. It also illustrates how affected individuals may minimize or emphasize certain traits and identities to protect themselves against status loss and discrimination. Further, the framework lays out the intersectional identities from the included studies and how these identities impact an individual’s capacity to fulfill their expected social roles. Depending on the community and prevailing value systems, individual identity management may focus on the roles of a spouse or parent, level of productivity, extent of community engagement, and other culturally defined identities. The ability to fully engage in these social roles is impacted by disclosure, or community awareness of “undesirable” traits or identities. In order to protect themselves from being outed or discriminated against, many individuals endorsed performative identities to highlight positive qualities or to minimize perceived negative attributes [26]. The extent to which an individual can balance their expected social roles by preventing disclosure, and through identity management, impacts healthcare related behaviors and consequent health outcomes. The thematic results that follow are grouped into [1] identity and visibility disclosure, [2] ability to fulfill expected social roles and [3] performative identities. The ways that individuals cannot, or do not navigate identity, disclosure, and social fulfillment through performative action impact their experience of intersectional stigma and healthcare behaviors which dictate infectious disease outcome.
**Fig 2:** *Stigma identity framework.*
## Adult identity and productivity
Across studies, the role of the adult was closely tied to social and financial productivity and familial/community responsibilities. All participants highlighted a loss of productivity related to HIV and TB, as illness decreased their stamina and strength. This loss of productivity was perceived to undermine their value as productive adults. Many respondents were agricultural workers or walked long distances for work and found that fatigue, weight loss and generalized weakness decreased their strength or endurance. Across studies, participants repeatedly discussed loss of strength as a threat to their identity as productive adults.
Older adults articulated feelings of adult identity loss tied to lost productivity and the physical limitations of age. Despite the respect gained with age, there was a common perception that aging reduced an individual’s ability to contribute to self and society. This was compounded by a diagnosis of chronic disease and appeared particularly acute for those suffering from symptomatic illness. Most participants reported pride in their work, their ability to maintain and provide for themselves, and wished to contribute to their communities. People who were unable to support themselves, required financial support from others, and who were unable to engage in self-care activities qualified their lapses in productivity as temporary setbacks surmountable with hard work and treatment. Framing loss of strength as a temporary condition allowed identity preservation as a return to productivity remained within reach.
## Gendered identity
Across all studies, societal role expectations differed by gender. Participants described males as financial providers. Men were expected to deliver economic support which validated their role as primary decision makers, including family healthcare decisions. Masculinity was tied to personal attributes and characteristics such as physical strength, rationality, and dominance. The chronic illness of HIV, TB, mental illness or noncommunicable disease, was incompatible with masculine cultural expectations as it rendered men weak, reliant on the care and support of others, and sidelined their familial authority. Similarly, the sexual behaviors of MSM challenged the ideals of African masculinity. MSM discussed enormous familial and community pressure to marry and have children which was inconsonant with their actual attractions, behaviors, or identities. Some cited the idea that homosexuality was unnatural to Africans and that same sex behaviors resulted from Eurocentric influence. In this way, same sex behaviors were perceived as a challenge not only to masculinity but to African identity.
In contrast to masculinity, which was tied to specific characteristics or personality traits, femininity and womanhood were tied to childbearing and household maintenance, rather than character or selfhood. During periods of upheaval or stress, women were viewed as emotional supports for all members of the family. Many male and female participants endorsed the idea of the physical subordination of women to men. Male respondents often viewed women as the weaker sex and many women conceded that men were physically stronger and better equipped for manual labor than women. As a result of their social roles within the home, women were generally financially dependent on their male partners increasing risk of volatility and economic precarity.
## Parenting identity
Several of the included groups reported challenges to parenting. Individuals who felt unable to fulfill their expected roles as mothers or fathers deeply internalized and perceived social inadequacy. Both men and women considered childbearing and childrearing as key elements of their social identity. For men, parenthood affirmed their virility and solidified their role as household heads. For women, childbearing was key to their identity as women and as traditional caregivers. Because so much of the female identity was tied to maternal duties, women who experienced attacks to socially constructed ideas of motherhood felt acute identity violations. Women living with HIV who became pregnant worried about the health of their unborn infants and felt torn between their identity as mothers, and their identity as women living with chronic illness. Similarly, women who experienced post-natal depression deeply internalized their mental illness as a role violation unbecoming of a mother. Conversely, for men who endorsed same-sex behaviors, their sexuality was a barrier to traditional marriage and procreation. Many of the men interviewed were Christian and felt that their sexual behaviors and consequent inability to have children also violated religious expectations.
## Identity visibility and involuntary disclosure
Despite best efforts to maintain daily routines and a semblance of normalcy throughout infectious disease care and treatment, participants experienced symptomology or confidentiality betrayals that disclosed their identity status. Unlike people who chose to voluntarily disclose their identities to access support or camaraderie, others experienced involuntary disclosure or outing.
## Symptom visibility
Illness visibility was central to participant fears of disclosure and outing. Individuals who perceived that their physical presentation betrayed their disease status reported an additional burden of stigma and shame. People living with stigmatized illnesses felt that their bodies revealed their illness identities to the community. Many participants viewed weight loss as an indicator of infectious disease that would then cause the community to make assumptions or gossip about their health in general or HIV status in particular. Others reported that the audible and persistent TB cough, or visible lymphadenopathy alerted others to their disease status. PLWH also reported perceived and experienced ostracization–others stared, gossiped, insulted them, and questioned their HIV status making them feel despondent or hopeless. Particularly in South Africa, similar symptomology led to a conflation of HIV with TB, due to overlapping symptoms of the two disease processes. People living with TB reported that community members spread rumors that they were HIV-positive, regardless of their true HIV status.
## Visibility and outing via healthcare
The threat of being seen at HIV and TB clinics raised concerns among participants about privacy, confidentiality, and disclosure. People living with HIV and/or TB feared status disclosure that might invite speculation, physical violence, or abandonment. In most cases, individuals carefully safeguarded their HIV and TB status unless it was absolutely necessary to disclose. Therefore, being seen at infectious disease clinics was strictly avoided to prevent involuntary disclosure. In many cases, people were willing to travel great distances to seek treatment outside of their communities or in places where they felt certain that no one would recognize them. This presented significant barriers to treatment access and adherence.
Study participants who endorsed stigmatized identities or behaviors attempted to withhold pertinent information from healthcare providers. People who used drugs or alcohol frequently hid their substance use behaviors from healthcare providers. This omission was intended to maintain amicable therapeutic relationships with healthcare providers and to secure non-judgmental treatment for other healthcare needs. Similarly, MSM hid their sexual identities or sexual behaviors from healthcare providers believing that they would receive better care without full disclosure. in fact, some MSM felt more comfortable receiving care at foreign funded and staffed non-governmental organizations whom they believed were more accepting of same sex behaviors.
## Performative identities
Although stigma complicated role and identity fulfillment, many people living with stigmatized identities were able to maintain their social roles. Those who maintained their outward facing identities often escaped critique and judgement from others. In order to safeguard themselves from identity violations and stigma, individuals went to great lengths to carefully cultivate and project curated identities that endorsed assimilation within the larger group and thus normalcy.
## Selective disclosure
Throughout the articles, tuberculosis was inextricably intertwined with HIV. However, TB is not as heavily stigmatized as HIV in sub-Saharan Africa. Despite their different pathophysiology, and potential for cure, the TB and HIV share some of the same signs and symptoms; as a result, many PLWH used their TB status to explain physical changes to their health and appearance. When others pointed out noticeable weight changes, people living with both TB and HIV were more willing to disclose their TB status in order to satisfy curiosity. While TB was understood as easily transmissible and non-selective, HIV was perceived as preventable via “good” behavior. Therefore, TB disclosure had few implications for an individual’s character and thus elicited limited or manageable blame. This was an explanatory narrative that could reasonably quell suspicion while maintaining a more desirable social identity.
Selective TB disclosure was more frequently reported by women. Women needed to disclose their TB status in order to secure financial support for treatment and travel but feared abandonment by their partners or families if they also disclosed their HIV status. For many female participants, a full figure indicated health and prosperity. Selective or partial disclosure of their TB diagnosis allowed women to address violations of societal beauty standards without the unrelenting stigma associated with HIV. According to participants, HIV and TB co-infection led to a tightrope walk between acceptable levels of scrutiny and too much disclosure in order to maintain safe housing and financial stability.
When participants were willing to share a stigmatized identity, they chose family or friends who were most likely to respect privacy and offer support. Often this included others who belonged to the same social identity group. MSM disclosed their identity to others within the LGBTQ community rather than family. Similarly, PLWH disclosed to others within treatment support groups. When disclosing to family or friends, individuals generally chose to disclose to female, rather than male, confidantes who were perceived as more empathetic which is also in-line with traditional female identity roles.
## Sexual behaviors
Throughout these articles, we found evidence that men used sexual behavior to assert dominance and reinforce masculine ideals. As noted earlier, illness narratives arising from patriarchal social norms challenged traditional notions of masculinity. In attempts to counter challenges to their masculinity, men wanted to cement their virility through sex by engaging with multiple sexual partners or being in extramarital relationships. Some men asserted sexual dominance by making unilateral decisions about condom use and sexual health, denying female input. By leaning into traditional masculine sexual narratives, men could minimize threats to their identity. Although women generally did not condone this hypersexual behavior, they did concede that it was an expected part of the male role which they tolerated. MSM also used traditional masculine narratives for identity protection. MSM participants endorsed the idea of heterosexual intercourse, marriage, and reproduction as a way of pacifying family and friends. By engaging in relationships with women, MSM could maintain a veneer of masculine heteronormativity.
In contrast to the exaggerated sexual behaviors of young men, older adults minimized their sexual activity in order to reduce the blame and culpability of living with HIV. There was a perception among key stakeholders that as people, and women in particular age, they lose their sexual drive, drastically reducing their risk for HIV infection. Although older adults revealed during confidential interviews that they had active, fulfilling sexual lives, they allowed a narrative of asexuality to continue in public spaces. Older adults were thought to be too well informed and self-disciplined to become infected with HIV through casual or unprotected sex. Older individuals did not challenge this narrative as it portrayed them in a more positive light.
## Healthcare behaviors
Among men, HIV and TB testing avoidance was common. Rather than subject their masculinity to existential threats, many men preferred not to know their disease status. Male participants living with HIV or TB recalled anxiety related to testing and reported that they delayed testing until their health was poor and they imminently feared death. This delay in testing led to longer recovery times, prolonged time away from work and greater physical dependence on others. Women also feared, and sometimes avoided, HIV testing. However, women’s fears were tied to fears of violence and abandonment. Several included men also reported denial following diagnosis, which was not reported by women. In one article, HIV testing behaviors were hindered by a belief that HIV infection was primarily a Ugandan problem. As a result, key populations within the sampled refugee community believed that they were only at risk for HIV if they were associating with, and adopting the behaviors of, their Ugandan hosts. HIV testing was therefore dismissed as unnecessary.
Lastly, across studies, illness visibility consistently threatened status disclosure and caused much distress to participants. However, women reported the physical and aesthetic benefits of treatment adherence to their HIV and TB regimens. Adherence benefits such as weight gain, quelled cough, or reduction in lymphadenopathy was viewed as a central part of stigma prevention. Stigma mediated through illness identity and gender was a key driver of treatment adherence. Women also discussed the support that they received from other women through support groups, and empowerment projects. However, this finding was unique to women as men were reported to infrequently join or utilize support groups despite invitations and encouragement. Table 3 shows the thematic results by included studies.
**Table 3**
| First author and year | Fulfillment of Expected Social Roles | Fulfillment of Expected Social Roles.1 | Fulfillment of Expected Social Roles.2 | Identity Visibility & Disclosure | Identity Visibility & Disclosure.1 | Performative Identity | Performative Identity.1 | Performative Identity.2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| First author and year | Adult Role | Gender Roles | Parental Role | Illness Visibility | Involuntary Disclosure | Selective Disclosure | Minimize Negative | Exaggerate Positive |
| Angwenyi 2018 [38] | X | | | | | | | |
| Becker 2019 [39] | | X | | | | | | |
| Brown 2018 [40] | | | | | | | | |
| Buregyeya 2012 [41] | | | | | X | | | |
| Chileshe 2010 [42] | | X | | X | X | X | | |
| Crankshaw 2014 [43] | | X | X | | | X | | |
| Daftary 2007 [44] | | | | X | | X | X | X |
| Daftary 2012a [45] | X | | | X | | X | X | X |
| Daftary 2012b [46] | | | | X | | X | | |
| Ezeanolue 2020 [47] | | | | | | | | |
| Finnie 2010 [48] | | | | X | | | | |
| Freeman 2017 [49] | X | | | X | | | X | X |
| Gebremariam 2010 [50] | | | | | | X | | |
| Gnauck 2013 [51] | | X | | | | | | |
| Jani 2021 [52] | X | | | | | | | |
| Kellett 2016 [53] | | X | X | | X | | | |
| Kennedy 2013 [54] | | X | | | X | X | X | X |
| King 2019 [55] | | X | | X | | X | | |
| Kuteesa 2012 [56] | X | | | X | | | | |
| Kyakuwa 2009 [57] | | | | | X | | | |
| Kyakuwa 2012 [58] | | | | | X | | | |
| Lemasters 2020 [59] | | | X | | | | | |
| Logie 2021 [60] | | X | | X | X | X | X | X |
| Magidson 2019 [61] | | | | | | | | |
| Matima 2018 [62] | | | | | | X | | |
| Matlho 2017 [63] | X | | | | | | X | X |
| Mburu 2014 [64] | | X | | | | | X | X |
| Mugisha 2020 [65] | | | | | | | | |
| Njozing 2010 [66] | | | | X | | | X | X |
| Owusu 2020 [67] | | X | | X | | X | | |
| Regenauer 2020 [69] | | | | | | | | |
| Russel 2016 [68] | | X | | | | | | |
| Tokwe 2020 [70] | | | | | X | | | |
| Tsang 2019 [71] | | X | X | | | | X | X |
## Discussion
This metasynthesis highlights the variety and scope of identities that contribute to intersectional stigma in Sub-Saharan Africa. Despite a clear focus on HIV stigma, other factors deemed culturally undesirable such as TB status, substance use disorder, and sexual identity contribute to stigma. Additional factors such as gender and age also interact with stigmatized identities to impact degrees of stigma and its manifestations. These themes are less salient in individual studies but become clear after reviewing the totality of intersectional work conducted in Sub-Saharan Africa.
Stigma is a gendered experience. This analysis underscores how both men and women from diverse settings within sub-Saharan Africa experience distinct, yet significant levels of stigma related to their infectious disease identities. Patriarchal institutions and beliefs around gender roles are harmful to women but are also deleterious to men. We do not believe that a universal African experience exists; and one is certainly not evident from the findings of this analysis. However, traditional masculine standards permeate across the included studies and result in the reported narratives of stigma. Because men in sub-Saharan Africa traditionally play a key role in government, policy, community leadership and family structure, hegemonic masculinity also heavily dictates social constructs and definitions of normalcy [27, 28]. Rigid views of manhood in the context of sub-Saharan Africa create a constructed masculinity that, when threatened by a stigmatized illness result in negative coping behaviors and poor health outcomes. This fragile masculinity causes an identity crisis among men when they feel that due to perceived, experienced, or internalized stigma, they are unable to fulfill their role as household and community leaders. Women on the other hand more frequently experience stigma in the form of isolation and abandonment resulting in poverty. Women’s economic reliance on men and the resulting financial precarity feeds back into a syndemic loop of illness and gendered vulnerability. As women experience abandonment and withdrawal of financial support, they experience barriers to health and well-being. Women also experience gender-based violence leading to emotional and physical vulnerability and lack the social and economic supports to improve their circumstances [29–31]. These factors put women at increased risk for poorer disease management thus reinforcing misconceptions about gendered fragility and subordination.
Despite a traditional focus on women from within intersectional analysis, our work shows that men also perceive and internalize stigma related to a variety of identities. Due to their ascribed roles as providers and community leaders, anticipated stigmatization presents a status threat for men that consequently undermines care engagement behaviors. While not intended to undermine or diminish the experiences of women who face disproportionate financial instability, inequity, hostility, and violence based on their gender, contrasting the experiences of men and women helps us to see that neither gender escapes stigma intensified by misogyny. Though men generally have more power to change or maintain the status quo, they also find themselves at the receiving end of stigma created and perpetuated by hegemonic patriarchy. These standards are particularly debilitating for MSM and transgender individuals in sub-Saharan Africa who do not conform to masculine cultural ideals.
## Implications
To make strides in HIV or TB prevention and care efforts, we must address barriers to infectious disease testing and treatment adherence, with renewed consideration for the unique concerns of men in sub-Saharan Africa. Researchers and medical providers have widely accepted undetectable equals untransmissible (U = U) messaging as an essential component of HIV elimination. However, U = U is predicated on earlier successes in the care continuum, including diagnosis, linkage, and retention in care, all of which are potentially compromised by stigma. Heterosexual HIV transmission remains the biggest driver of the epidemic in Sub-Saharan Africa [32]. If we become so focused on the intersection of women and HIV that we allow stigma to prevent male engagement in the care cascade, then we will never halt heterosexual transmission.
The knowledge gained in this review supports innovative and discrete means of testing to engage men in infectious disease care. Men avoid testing due to threats to their masculine identity. Low barrier testing and treatment models should accommodate work schedules and prioritize confidentiality and privacy [33] which were both highlighted by participants as essential for care engagement. Same day initiation strategies are essential to preserving physical strength and earning potential [34]. Men may be more receptive to home-based testing initiatives, mobile clinics, and education about the utility of ART and TB treatments for preserving strength and earning capacity [35]. From a structural standpoint, engaging men will require formal employment protections that prevent employment discrimination and allow reasonable accommodation for healthcare attendance [34].
Due to economic insecurity and potential for violence following HIV status disclosure, healthcare providers should incorporate interpersonal violence and housing instability screening into HIV testing and treatment visits for women. To support motherhood identity, health care providers must reinforce U = U education at all points of healthcare access and ensure that women know this also applies to transmission between mother and child. Similarly, women will benefit from education about the benefits of ART and TB treatment in the restoration of physical health [36]. Messaging centered around ART and healthy weight may be an effective strategy combating internalized and anticipated stigma. Furthermore, bolstering female economic opportunities leads to empowerment in the home and in healthcare decision making [37]. In the few articles that considered female economic empowerment, there was little discussion around how income changed female or motherhood identity and the long-term social consequences of that identity change. The impact of female financial enfranchisement is certainly an intervention area for scale up given its potential for stigma reduction.
The conceptual framework that has emerged from this meta-synthesis has the potential to guide stigma intervention and mitigation strategies that center socially relevant identities. Using the stigma identity framework, community level stigma arises from individuals’ tendency to categorize and differentiate from others based on divergent characteristics. Stigma arises when individuals assess, judge, and differentiate from others. To address stigma at the individual level, healthcare providers must acknowledge the unique identity of their patients and how their patients fit into their nuanced social positions. Without this acknowledgment, healthcare providers will misunderstand the motivations behind seemingly negative healthcare behaviors. An underappreciation of cultural nuance and personal circumstances leads healthcare providers to dismiss patients as “non-compliant” or “defaulters”. When these terms are used, patients report feeling dehumanized and are allowed to slip into a medically constructed narrative rife with its own healthcare-specific biases.
Throughout this analysis it was clear that healthcare providers continue to enact stigma in their daily interactions with patients. Despite an acknowledgment that stigma is a barrier to care and should be addressed, healthcare providers openly endorsed stigma towards substance use disorders, mental illness and LGBTQ identifying individuals. Healthcare providers in HIV and TB care have come far in their treatment of PLWH but must be cognizant of similar biases towards addiction, mental illness, and same sex behavior. Education and empathy remain central to stigma mitigation in healthcare. Without careful self-reflection, healthcare providers will be unable to engage patients throughout the HIV care continuum, especially longstanding key populations who oftentimes are more acutely subjected to intersectional stigma.
## Conclusion
Individual studies of stigma interrogate the role of intersectionality and an infinite combination of traits or characteristics that affect the internalization, experience, and perception of stigma. However, less studied is the impact of multiple identities on stigma and stigma on multiple identities. From our analysis, gender and HIV status does not create a simple intersection. Gender is a role and an identity central to social and self-actualization. This reality elevates gender, adulthood, parenthood, and occupational identity from social determinants of health to core constructs central to human psychosocial fulfillment in the sub-Saharan context. These socially constructed identities actively created and re-shaped by culture and individuals. Throughout the history of HIV and AIDS in the region, cultural identity remains a powerful force that dictates behavior and health outcomes. Future stigma research should consider not only the barriers and difficulties created by these intersections, but also how these intersections may be central to the identity of individuals.
## References
1. Stangl AL, Earnshaw VA, Logie CH, Van Brakel W, Simbayi LC, Barré I. **The Health Stigma and Discrimination Framework: A global, crosscutting framework to inform research, intervention development, and policy on health-related stigmas**. *BMC Med* (2019.0) **17**. DOI: 10.1186/s12916-019-1271-3
2. Goffman E.. *Stigma: Notes on the Management of Spoiled Identity* (1963.0)
3. Onono M, Odwar T, Abuogi L, Owuor K, Helova A, Bukusi E. **Effects of Depression, Stigma and Intimate Partner Violence on Postpartum Women’s Adherence and Engagement in HIV Care in Kenya**. *AIDS Behav* (2020.0) **24** 1807-15. DOI: 10.1007/s10461-019-02750-y
4. Lipira L, Williams EC, Huh D, Kemp CG, Nevin PE, Greene P. **HIV-Related Stigma and Viral Suppression Among African-American Women: Exploring the Mediating Roles of Depression and ART Nonadherence**. *AIDS Behav* (2019.0) **23** 2025-36. DOI: 10.1007/s10461-018-2301-4
5. Turan B, Rice WS, Crockett KB, Johnson M, Neilands TB, Ross SN. **Longitudinal association between internalized HIV stigma and antiretroviral therapy adherence for women living with HIV: The mediating role of depression**. *Aids* (2019.0) **33** 571-6. DOI: 10.1097/QAD.0000000000002071
6. Christopoulos KA, Neilands TB, Dilworth S, Lisha N, Sauceda J, Mugavero MJ. **Internalized HIV stigma predicts subsequent viremia in US HIV patients through depressive symptoms and ART adherence**. *Aids* (2020.0) 1665-71. PMID: 32769764
7. Tesfaw G, Ayano G, Awoke T, Assefa D, Birhanu Z, Miheretie G. **Prevalence and correlates of depression and anxiety among patients with HIV on-follow up at Alert Hospital, Addis Ababa, Ethiopia.**. *BMC Psychiatry* (2016.0) **16** 1-7. DOI: 10.1186/s12888-016-1037-9
8. Relf M V., Pan W, Edmonds A, Ramirez C, Amarasekara S, Adimora AA. **Discrimination, Medical Distrust, Stigma, Depressive Symptoms, Antiretroviral Medication Adherence, Engagement in Care, and Quality of Life among Women Living with HIV in North Carolina: A Mediated Structural Equation Model**. *J Acquir Immune Defic Syndr.* (2019.0) **81** 328-35. DOI: 10.1097/QAI.0000000000002033
9. Rice WS, Logie CH, Napoles TM, Walcott M, Batchelder AW, Kempf MC. **Perceptions of intersectional stigma among diverse women living with HIV in the United States**. *Soc Sci Med* (2018.0) **208** 9-17. DOI: 10.1016/j.socscimed.2018.05.001
10. Sangaramoorthy T, Jamison A, Dyer T. **Intersectional stigma among midlife and older Black women living with HIV**. *Cult Heal Sex* (2017.0) **19** 1329-43. DOI: 10.1080/13691058.2017.1312530
11. Turan JM, Elafros MA, Logie CH, Banik S, Turan B, Crockett KB. **Challenges and opportunities in examining and addressing intersectional stigma and health.**. *BMC Med.* (2019.0) **17** 1-16. PMID: 30651111
12. Beggars Navon L.. **Metaphors, and Stigma: A Missing Link in the Social History of Leprosy**. *Soc Hist Med* (1998.0) **11** 89-105. PMID: 11620156
13. Gussow Z, Tracy G. **Stigma and the Leprosy Phenomenon: The Social History of a Disease in the Nineteenth and Twentieth Centuries.**. *Bull Hist Med* (1970.0) **44** 425-49. PMID: 4919121
14. Leblanc NM, Flores DD, Barroso J. **Facilitators and Barriers to HIV Screening**. *Qual Health Res* (2016.0) **26** 294-306. PMID: 26631679
15. Stangl AL, Lilleston P, Mathema H, Pliakas T, Krishnaratne S, Sievwright K. **Development of parallel measures to assess HIV stigma and discrimination among people living with HIV, community members and health workers in the HPTN 071 (PopART) trial in Zambia and South Africa**. *J Int AIDS Soc* (2019.0) **22** e25421. DOI: 10.1002/jia2.25421
16. Link BG, Phelan JC. **Conceptualizing Stigma.**. *Annu Rev Sociol* (2001.0) **27** 363-85
17. Tajfel H, Billig MG, Bundy RP, Flament C. **Social categorization and intergroup behaviour**. *Eur J Soc Psychol* (1971.0) **1** 149-78
18. 18Joint United Nations Programme on HIV/AIDS. Confronting discrimination: Overcoming HIV-related stigma and discrimination in health-care settings and beyond. 2017.
19. 19Joint United Nations Programme on HIV/AIDS. Evidence for eliminating HIV related stigma and discrimination. Geneva; 2020.
20. Sandelowski M, Barroso J. *Handbook for Synthesizing Qualitative Research* (2007.0)
21. Leddy AM, Turan JM, Johnson MO, Neilands TB, Kempf MC, Konkle-Parker D. **Poverty stigma is associated with suboptimal HIV care and treatment outcomes among women living with HIV in the United States**. *AIDS* (2019.0) **33** 1379-84. DOI: 10.1097/QAD.0000000000002189
22. Turan B, Hatcher AM, Weiser SD, Johnson MO, Rice WS, Turan JM. *American Journal of Public Health.* (2017.0) **107** 863-9. PMID: 28426316
23. Logie CH, Williams CC, Wang Y, Marcus N, Kazemi M, Cioppa L. **Adapting stigma mechanism frameworks to explore complex pathways between intersectional stigma and HIV-related health outcomes among women living with HIV in Canada.**. *Soc Sci Med* (2019.0) **232** 129-38. DOI: 10.1016/j.socscimed.2019.04.044
24. **WHO releases new global lists of high-burden countries for TB, HIV-associated TB and drug-resistant TB**. (2021.0)
25. McCall L.. **The Complexity of Intersectionality**. *Signs (Chic).* (2005.0) **30** 1771-800
26. Goffman E.. *The Presentation of Self in Everyday Life. first.* (1956.0) 1-251
27. Kisigo GA, Ngocho JS, Mwamba RN, Knettel BA, Relf M V, Mmbaga BT. **HIV Stigmatizing Attitudes Among Men Accompanying Their Partners to Antenatal Care in Tanzania: A Mixed-Method Study.**. *AIDS Behav* (2021.0) **25** 3172-82. DOI: 10.1007/s10461-021-03264-2
28. Wyrod R. **Masculinity and the persistence of AIDS stigma.**. *Cult Heal Sex* (2011.0) **13** 443-56. DOI: 10.1080/13691058.2010.542565
29. Beyene AS, Chojenta C, Roba HS, Melka AS, Loxton D. **Gender-based violence among female youths in educational institutions of Sub-Saharan Africa: a systematic review and meta-analysis.**. *Syst Rev 2019 81* (2019.0) **8** 1-14. DOI: 10.1186/s13643-019-0969-9
30. Akamike C, Akamike IC, Uneke CJ, Uro-Chukwu HC, Okedo-Alex IN, Chukwu OE. **Predictors and facilitators of gender-based violence in sub-Saharan Africa: a rapid review**. *JOGHR* (2019.0) **3** 2019076
31. Singer M, Bulled N, Ostrach B, Mendenhall E. **Syndemics and the biosocial conception of health**. *Lancet* (2017.0) **389** 941-50. DOI: 10.1016/S0140-6736(17)30003-X
32. Kharsany ABM, Karim QA. **HIV Infection and AIDS in Sub-Saharan Africa: Current Status, Challenges and Opportunities.**. *Open AIDS J.* (2016.0) **10** 34. DOI: 10.2174/1874613601610010034
33. Kebaabetswe P, Manyake K, Kadima E, Auletta-young C, Sekoto T, Dintwa OM. **Barriers and facilitators to linkage to care and ART initiation in the setting of high ART coverage in Botswana.**. *AIDS Care* (2020.0) **32** 722-8. DOI: 10.1080/09540121.2019.1640843
34. Misra S, Mehta HT, Eschliman EL, Rampa S, Poku OB, Wang WQ. **Identifying “What Matters Most” to Men in Botswana to Promote Resistance to HIV-Related Stigma**. *Qual Health Res.* (2021.0) **31** 1680-96. DOI: 10.1177/10497323211001361
35. Labhardt ND, Ringera I, Lejone TI, Klimkait T, Muhairwe J, Amstutz A. **Effect of Offering Same-Day ART vs Usual Health Facility Referral During Home-Based HIV Testing on Linkage to Care and Viral Suppression Among Adults With HIV in Lesotho: The CASCADE Randomized Clinical Trial**. *JAMA* (2018.0) **319** 1103-12. DOI: 10.1001/jama.2018.1818
36. Dahab M, Charalambous S, Hamilton R, Fielding K, Kielmann K, Churchyard GJ. **That is why I stopped the ART”: Patients’ & providers’ perspectives on barriers to and enablers of HIV treatment adherence in a South African workplace programme**. *BMC Public Health* (2008.0) **8** 1-6. PMID: 18173844
37. Glass N, Perrin NA, Kohli A, Campbell J, Remy MM. **Randomised controlled trial of a livestock productive asset transfer programme to improve economic and health outcomes and reduce intimate partner violence in a postconflict setting**. *BMJ Glob Heal* (2017.0) **2** 1-11
38. Angwenyi V, Aantjes C, Kajumi M, De Man J, Criel B, Bunders-Aelen J. **Patients experiences of self-management and strategies for dealing with chronic conditions in rural Malawi**. *PLoS One* (2018.0) **13** e0199977. DOI: 10.1371/journal.pone.0199977
39. Becker TD, Ho-Foster AR, Poku OB, Marobela S, Mehta H, Cao DTX. **“It’s When the Trees Blossom”: Explanatory Beliefs, Stigma, and Mental Illness in the Context of HIV in Botswana.**. *Qual Health Res* (2019.0) **29** 1566-80. DOI: 10.1177/1049732319827523
40. Brown A, Wood L. **A critical participatory pedagogical approach to enabling life orientation students to develop social literacy through HIV education**. *Afr J AIDS Res* (2018.0) **17** 153-62. DOI: 10.2989/16085906.2018.1478311
41. Buregyeya E, Nuwaha F, Wanyenze RK, Mitchell EMH, Criel B, Verver S. **Utilization of HIV and Tuberculosis Services by Health Care Workers in Uganda: Implications for Occupational Health Policies and Implementation.**. *PLoS One* (2012.0) **7** e46069. DOI: 10.1371/journal.pone.0046069
42. Chileshe M, Bond VA. **Barriers and outcomes: TB patients co-infected with HIV accessing antiretroviral therapy in rural Zambia**. *AIDS Care—Psychol Socio-Medical Asp AIDS/HIV* (2010.0) **22** 51-9. DOI: 10.1080/09540121003617372
43. Crankshaw T, Voce A, King R, Giddy J, Sheon N, Butler L. **Double Disclosure Bind: Complexities of Communicating an HIV Diagnosis in the Context of Unintended Pregnancy in Durban, South Africa. AIDS**. *Behav* (2014.0) **18** 53-9
44. Daftary A, Padayatchi N, Padilla M. **HIV testing and disclosure: a qualitative analysis of TB patients in South Africa**. *AIDS Care* (2007.0) **19** 572-7. DOI: 10.1080/09540120701203931
45. Daftary A.. **HIV and tuberculosis: The construction and management of double stigma**. *Soc Sci Med* (2012.0) **74** 1512-9. DOI: 10.1016/j.socscimed.2012.01.027
46. Daftary A, Padayatchi N. **Social constraints to TB/HIV healthcare: Accounts from coinfected patients in South Africa.**. *AIDS Care* (2012.0) **24** 1480-6. DOI: 10.1080/09540121.2012.672719
47. Ezeanolue EE, Iheanacho T, Adedeji IA, Itanyi IU, Olakunde B, Patel D. **Opportunities and challenges to integrating mental health into HIV programs in a low- and middle-income country: insights from the Nigeria implementation science Alliance.**. *BMC Health Serv Res* (2020.0) 20. PMID: 31910837
48. Finnie RKC, Mabunda T, Khoza LB, van den Borne B, Selwyn B, Mullen PD. **Pilot study to develop a rapid assessment of tuberculosis care-seeking and adherence practices in rural Limpopo Province, South Africa**. *Int Q Community Health Educ.* (2010.0) **31** 3-19. DOI: 10.2190/IQ.31.1.b
49. Freeman E. **Neither “foolish” nor “finished”: identity control among older adults with HIV in rural Malawi**. *Sociol Health Illn* (2017.0) **39** 711-25. DOI: 10.1111/1467-9566.12531
50. Gebremariam MK, Bjune GA, Frich JC. **Barriers and facilitators of adherence to TB treatment in patients on concomitant TB and HIV treatment: a qualitative study**. *BMC Public Health* (2010.0) **10** 651. DOI: 10.1186/1471-2458-10-651
51. Gnauck K, Ruiz J, Kellett N, Sussman A, Sullivan MA, Montoya M. **Economic empowerment and AIDS-related stigma in rural Kenya: a double-edged sword?**. *Cult Health Sex* (2013.0) **15** 851-65. DOI: 10.1080/13691058.2013.789127
52. Jani N, Mathur S, Kahabuka C, Makyao N, Pilgrim N. **Relationship dynamics and anticipated stigma: Key considerations for PrEP use among Tanzanian adolescent girls and young women and male partners**. *PLoS One* (2021.0) **16** e0246717. DOI: 10.1371/journal.pone.0246717
53. Kellett NC, Gnauck K. **The intersection of antiretroviral therapy, peer support programmes, and economic empowerment with HIV stigma among HIV-positive women in West Nile Uganda.**. *African J AIDS Res* (2016.0) **15** 341-8. DOI: 10.2989/16085906.2016.1241288
54. Kennedy CE, Baral SD, Fielding-Miller R, Adams D, Dludlu P, Sithole B. **“They are human beings, they are Swazi”: intersecting stigmas and the positive health, dignity and prevention needs of HIV-positive men who have sex with men in Swaziland**. *J Int AIDS Soc* (2013.0) **16** 18749. DOI: 10.7448/IAS.16.4.18749
55. King R, Nanteza J, Sebyala Z, Bbaale J, Sande E, Poteat T. **HIV and transgender women in Kampala, Uganda—Double Jeopardy.**. *Cult Health Sex* (2019.0) **21** 727-40. DOI: 10.1080/13691058.2018.1506155
56. Kuteesa MO, Seeley J, Cumming RG, Negin J. **Older people living with HIV in Uganda: Understanding their experience and needs**. *African J AIDS Res* (2012.0) **11** 295-305. DOI: 10.2989/16085906.2012.754829
57. Kyakuwa M.. **Ethnographic experiences of HIV-positive nurses in managing stigma at a clinic in rural Uganda**. *African J AIDS Res* (2009.0) **8** 367-78
58. Kyakuwa M, Hardon A, Goldstein Z. **The Adopted Children of ART”: expert clients and role tensions in ART provision in Uganda.**. *Med Anthropol* (2012.0) **31** 149-61. DOI: 10.1080/01459740.2011.603399
59. LeMasters K, Dussault J, Barrington C, Bengtson A, Gaynes B, Go V. **Pain in my heart”: Understanding perinatal depression among women living with HIV in Malawi.**. *PLoS One* (2020.0) **15** e0227935. DOI: 10.1371/journal.pone.0227935
60. Logie CH, Okumu M, Musoke DK, Hakiza R, Mwima S, Kyambadde P. **Intersecting stigma and HIV testing practices among urban refugee adolescents and youth in Kampala, Uganda: qualitative findings**. (2021.0). DOI: 10.1002/jia2.25674
61. Magidson JF, Joska JA, Regenauer KS, Satinsky E, Andersen LS, Seitz-Brown CJ. **“Someone who is in this thing that I am suffering from”: The role of peers and other facilitators for task sharing substance use treatment in South African HIV care**. *Int J Drug Policy* (2019.0) **70** 61-9. DOI: 10.1016/j.drugpo.2018.11.004
62. Matima R, Murphy K, Levitt NS, BeLue R, Oni T. **A qualitative study on the experiences and perspectives of public sector patients in Cape Town in managing the workload of demands of HIV and type 2 diabetes multimorbidity**. *PLoS One* (2018.0) **13** e0194191. DOI: 10.1371/journal.pone.0194191
63. Matlho K, Lebelonyane R, Driscoll T, Negin J. **Policy-maker attitudes to the ageing of the HIV cohort in Botswana**. *SAHARA J J Soc Asp HIV/AIDS Res Alliance* (2017.0) **14** 31-7
64. Mburu G, Ram M, Siu G, Bitira D, Skovdal M, Holland P. **Intersectionality of HIV stigma and masculinity in eastern Uganda: implications for involving men in HIV programmes**. *BMC Public Health* (2014.0) **14** 1061. DOI: 10.1186/1471-2458-14-1061
65. Mugisha J, Kinyanda E, Osafo J, Nalukenge W, Knizek BL. **Health care professionals’ perspectives on barriers to treatment seeking for formal health services among orphan children and adolescents with HIV/AIDS and mental distress in a rural district in central, Uganda**. *Child Adolesc Psychiatry Ment Health* (2020.0) **14** 26. DOI: 10.1186/s13034-020-00332-8
66. Njozing N, Edin K, Hurtig A-K. **‘When I get better I will do the test’: Facilitators and barriers to HIV testing in Northwest Region of Cameroon with implications for TB and HIV/AIDS control programmes | SAHARA-J: Journal of Social Aspects of HIV/AIDS.**. *SAHARAH-J J Soc Asp HIV/AIDS* (2010.0) **7**
67. Owusu AY. **A gendered analysis of living with HIV/AIDS in the Eastern Region of Ghana**. *BMC Public Health* (2020.0) **20** 751. DOI: 10.1186/s12889-020-08702-9
68. Russell SG, Lim S, Kim P, Morse S. **The legacy of gender-based violence and HIV/AIDS in the postgenocide era: Stories from women in Rwanda.**. *Health Care Women Int* (2016.0) **37** 721-43. DOI: 10.1080/07399332.2015.1083026
69. Regenauer KS, Myers B, Batchelder AW, Magidson JF. **“That person stopped being human”: Intersecting HIV and substance use stigma among patients and providers in South Africa**. *Drug Alcohol Depend* (2020.0) 216. DOI: 10.1016/j.drugalcdep.2020.108322
70. Tokwe L, Naidoo JR. **Lived experiences of human immunodeficiency virus and hypertension in the Eastern Cape, South Africa**. *African J Prim Heal care Fam Med* (2020.0) **12** e1-8. DOI: 10.4102/phcfm.v12i1.2472
71. EY ha Tsang, Qiao S, Wilkinson JS, Fung AL chu, Lipeleke F, Li X. **Multilayered Stigma and Vulnerabilities for HIV Infection and Transmission: A Qualitative Study on Male Sex Workers in Zimbabwe.**. *Am J Mens Health* (2019.0) 13
|
---
title: Uric acid is independently associated with interleukin‐1β levels in tear fluid
of hyperuricemia and gout patients
authors:
- Mian Wu
- Xingna Hu
- Ting Lu
- Chenxiao Liu
- Honghong Lu
journal: Immunity, Inflammation and Disease
year: 2023
pmcid: PMC10022423
doi: 10.1002/iid3.805
license: CC BY 4.0
---
# Uric acid is independently associated with interleukin‐1β levels in tear fluid of hyperuricemia and gout patients
## Abstract
Tear urate levels were higher in patients with hyperuricemia and gout than in controls. There was a significant positive correlation between tear uric acid and tear IL‐1β level.
### Objectives
To explore pro‐inflammatory cytokines status in the tear fluid of patients with hyperuricemia and gout and its association with uric acid level.
### Methods
A total of 58 control subjects, 58 hyperuricemia patients including 40 asymptomatic hyperuricemia and 18 gout participants were recruited in this study. For tear analysis, each patient's tears were collected using capillary action microcaps after stimulation. Tear uric acid levels were measured using chemiluminescence. Tear and serum interleukin‐1beta (IL‐1β) and tumor necrosis factor‐alpha (TNF‐α) levels were measured using enzyme‐linked immunosorbent assay. The correlation of serum and tear uric acid levels with IL‐1β and TNF‐α were determined.
### Results
Tear uric acid levels were significantly higher in hyperuricemia group (98.2 ± 51.5 vs. 42.7 ± 24.0 µmol/L, $p \leq .001$) than in controls group. IL‐1β concentrations were significantly higher in hyperuricemia eyes compared to control eyes (210.2 ± 113.9 vs. 142.6 ± 29.8 pg/mL, $p \leq .001$). Multiple linear regression analysis showed that tear uric acid levels were independently positively associated with tear IL‐1β concentrations ($B = 0.192$, $p \leq .001$). However, no significant correlations were found between serum or tear uric acid and TNF‐α level. Moreover, there were no statistically differences of tear IL‐1β and TNF‐α levels between the asymptomatic hyperuricemia and gout groups.
### Conclusions
Tear uric acid levels were higher in patients with hyperuricemia and gout than in controls. There was a significant positive correlation between tear uric acid value and tear IL‐1β level, implying an interaction between hyperuricemia and ocular inflammation responses.
## INTRODUCTION
Uric acid is the end product of purine catabolism. Either overproduction or underexcretion of uric acid can lead to hyperuricemia. It is estimated that up to $21\%$ of the general population and $25\%$ of hospitalized patients have asymptomatic hyperuricemia. 1 *Hyperuricemia is* the most crucial risk factor for gout development; moreover, it is also associated with a variety of comorbidities, including hypertension, diabetes, chronic kidney disease, and coronary artery disease. 2 However, the impact of hyperuricemia on ocular abnormalities remained poorly understood.
The eye is a particularly vulnerable organ, susceptible to vascular abnormalities, metabolic disturbances, and inflammation. For instance, diabetic retinopathy is a leading cause of vision impairment caused by metabolic stress, glucose‐mediated microvascular damage, and inflammation. 3 *In hyperurcemia* and gout patients, urate crystal deposition has been observed in almost all ocular and adnexal locations. 4 Furthermore, in addition to lower temperatures and pH level or pH gradient between plasma and tissue, ocular structures have relatively poor solvent ability, indicating that they are predisposed to tophi deposition. 5 Crystal‐induced inflammation of the affected tissue due to urate deposition is responsible for the clinical presentations. Therefore, it is necessary to study the changes in ocular inflammation in hyperuricemia and gout patients.
Tear fluid is a heterogeneous solution containing mainly proteins, lipids, mucins, and electrolytes, responsible for regulating the physiology of human eye. The complex composition of tears can be altered due to eye inflammations. 6 Research has indicated that pro‐inflammatory factors such as interleukin (IL)‐1β, IL‐17, and IL‐8, NOD‐like receptor protein 3 (NLRP3) inflammasome, tumor necrosis factor‐alpha (TNF‐α), and anti‐inflammatory factors such as IL‐10, IL‐37 have all been implicated monosodium urate (MSU)‐induced gout inflammatory process. 7 However, cytokines in the tears of hyperuricemia and gout patients remained unknown.
Therefore, this study aims to compare the difference of uric acid and pro‐inflammatory cytokine levels in tears between healthy controls and hyperuricemia patients, and explore the correlation between tear uric acid and pro‐inflammatory cytokines levels.
## Study population
Participants were recruited from our inpatient department at the Affiliated Suzhou Hospital of Nanjing Medical University between June 2020 and October 2021. The study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee at the Affiliated Suzhou Hospital of Nanjing Medical University (KL901220). All participants gave written informed consent.
Inclusion criteria for hyperuricemia subjects were (i) age between 18 and 80 years, (ii) serum uric acid level ≥ 420 µmol/L, and (iii) history of hyperuricemia or gout flare. Diagnosis of gout was based on the 2015 American College of Rheumatology/European League Against Rheumatism diagnostic criteria. 8 *Exclusion criteria* for all subjects were as follows: (i) any ophthalmological condition such as infection or allergic conjunctivitis, (ii) severe liver or kidney impairment (alanine aminotransferase [ALT] or aspartate aminotransferase [AST] greater than three times the normal value or estimated glomerular filtration rate (eGFR) lower than 30 mL/min/1.73 m2), (iii) pregnant or breastfeeding women, (iv) any other severe medical, neurological or psychiatric condition, and (v) gout patients during acute flare and patients receiving regular uric acid‐lowering therapy, nonsteroidal anti‐inflammatory drugs (NSAIDs), colchicine, corticosteroids, and diuretics within 1 month.
In total, 68 healthy controls and 65 hyperuricemia patients were assessed for eligibility. Of these, 10 controls and 7 patients were excluded due to insufficient amount of tear fluid sampled, leaving 58 patients and 58 hyperuricemia patients included for final analysis.
## Tear collection
Tear fluid sampling was conducted between 7:00 a.m. and 9:00 p.m. in a fasting condition. Reflex tears were collected as previously described. 9 After sneeze reflex was stimulated by gently inserting a sterile cotton bud into the nasal passages to induce tear production, tear collection took place for approximately 1 min to enable sufficient volume to be obtained. Participants tilted their head toward the side of the collection while looking in the opposite direction. The microcapillary tube rested in the lateral tear meniscus. Participants were allowed to blink during the procedure. The sample was expelled from the capillary tube into a siliconized polypropylene microcentrifuge tube of 0.25 mL capacity and placed on ice until processing. After collection, samples were centrifuged at 4000 rpm for 20 min at 4°C to remove cellular debris. The supernatants were collected and stored at −80°C.
## Tear fluid analyses
Tear uric acid level were determined using chemiluminescence (Nanjing Jiancheng Bioengineering Institute). To detect tear IL‐1β and TNF‐α concentrations, the tear fluids were diluted 20 times with phosphate buffered saline and detected using a commercial enzyme‐linked immunosorbent assay kit, according to the user manual (R&D systems).
## Clinical and laboratory assessments
The clinical data of the subjects were recorded, including height, weight, systolic blood pressure, and diastolic blood pressure. The body mass index (BMI) was calculated as weight (kg)/height (m2). Venous blood samples were collected in the morning after an overnight fast to measure ALT, AST, serum uric acid, urea nitrogen, creatinine, blood glucose, cholesterol, and triglycerides. The estimated glomerular filtration rate (GFR) was calculated from the four‐variable modification of diet in renal disease equation: estimated GFR (mL/min/1.73 m2) = 175 × (serum creatinine [μmol/L]/88.4)–1.154 × (age [years])–0.203 × (0.742 if female) × (1.21 if Black). All subjects underwent standard clinical and laboratory evaluations.
## Statistics
Continuous variables were represented as mean ± standard deviation, whereas categorical variables were displayed as numbers with percentages. Student's t test was used to compare means for normally distributed variables, and the χ 2 test was used to compare frequencies between groups. Spearman's ρ was used to test for correlations. Variables that were significantly associated with tear uric acid were defined as candidate factors for linear regression analysis. A multivariable linear regression model using the stepwise likelihood ratio method was fitted with candidate factors, with entry probability of 0.05 and removal of 0.10. analysis of covariance was used to test differences in tear uric acid between asymptomatic hyperuricemia and gout patients after adjusting for serum uric acid level. All tests were two‐tailed, and $p \leq .05$ were considered statistically significant. Statistical analyses were performed using the SPSS software package, version 21.0.
## General characteristics of subjects in control and hyperuricemia group
The patient demographics are summarized in Table 1. The mean age of control and hyperuricemia subjects was 49.2 and 50.7 years, respectively. $72.4\%$ and $82.8\%$ of participants were males in control and hyperuricemia group. Compared with controls, patients in the hyperuricemia group were more obese, had higher levels of systolic blood pressure and fasting blood glucose, and had worse liver and kidney functions. In all patients, we did not observe significant uric acid deposition in the cornea.
**Table 1**
| Variable | Control | HUA | p Value |
| --- | --- | --- | --- |
| N (male) | 58 (42) | 58 (48) | .219 |
| Age (year) | 48.2 ± 13.0 | 50.7 ± 17.5 | .395 |
| BMI (kg/m2) | 23.5 ± 3.0 | 26.1 ± 4.3 | <.001 |
| SBP (mmHg) | 127.3 ± 16.8 | 136.6 ± 22.1 | .012 |
| DBP (mmHg) | 77.0 ± 11.5 | 79.1 ± 13.0 | .372 |
| ALT (U/L) | 24.6 ± 23.1 | 36.6 ± 36.1 | .012 |
| AST (U/L) | 23.1 ± 11.7 | 27.0 ± 16.1 | .372 |
| BUN (mmol/L) | 4.9 ± 1.4 | 6.5 ± 3.7 | .002 |
| Cr (umol/L) | 70.8 ± 14.1 | 91.7 ± 33.9 | <.001 |
| eGFR (mL/min/1.73 m2) | 106.4 ± 22.6 | 89.5 ± 34.2 | <.001 |
| FPG (mmol/L) | 5.9 ± 1.9 | 6.7 ± 2.1 | .044 |
| TC (mmol/L) | 4.8 ± 0.8 | 4.5 ± 1.1 | .110 |
| TG (mmol/L) | 1.5 ± 1.1 | 1.9 ± 1.3 | .052 |
| sUA (umol/L) | 292.9 ± 67.2 | 479.7 ± 89.1 | <.001 |
| Gout (%) | 0 | 31.0 | ‐ |
## Quantitative analysis of uric acid level, pro‐inflammatory cytokines in tear fluid
As illustrated in Figure 1A, tear uric acid levels in hyperuricemia group were significantly higher than control group (98.2 ± 51.5 vs. 42.7 ± 24.0 µmol/L, $p \leq .001$). Pro‐inflammatory cytokines including IL‐1β and TNF‐α were evaluated in tear fluid. The results illustrated that tear IL‐1β level was significantly higher in hyperuricemia group than in healthy control group (210.2 ± 113.9 vs. 143.6 ± 29.8 pg/mL, $p \leq .001$) (Figure 1B), whereas no significant difference existed in tear TNF‐α levels between control and hyperuricemia groups (130.5 ± 50.5 vs. 134.7 ± 54.8 pg/mL, $$p \leq .665$$) (Figure 1C).
**Figure 1:** *Comparison of tear uric acid and pro‐inflammatory cytokines levels between control and hyperuricemia groups. (A) Comparison of tear uric acid (A), tear IL‐1β (B) and tear TNF‐α (C) between control and HUA groups. HUA, hyperuricemia; tIL‐1β, tear interleukin‐1beta; tTNF‐α, tear tumor necrosis factor‐α; tUA, tear uric acid.*
## Correlation between uric acid and other clinical variables
Tear uric acid level revealed a positively association with BMI, Cr, and TG levels, serum uric acid level was significantly positively correlated with BMI, SBP, ALT, Cr, and TG (Table 2). As shown in Figure 2A–C, tear uric acid level was significantly positively correlated with serum uric acid level in total participants and control group, but not in hyperuricemia group. Moreover, there were positive associations between tear uric acid level and tear IL‐1β level in total participants and the hyperuricemia group, but not in the control group (Figure 2D–F). However, no statistically significant correlations were observed between tear uric acid and tear TNF‐α level in either total participants, healthy controls, or hyperuricemia patients (Figure 2G–I).
## Tear uric acid was independently associated with tear IL‐1β concentration
Variables significantly related with tear uric acid in Spearman correlation analysis including BMI, SBP, ALT, Cr, TG, sUA, and tear IL‐1β were included in multivariable linear regression analysis. As indicated in Table 3, tear uric acid was independently associated with serum uric acid (B, 0.177; $95\%$ confidence interval [CI], 0.118–0.256; $p \leq .001$) and tear IL‐1β (B, 0.192; $95\%$ CI, 0.111–0.273; $p \leq .001$).
**Table 3**
| Unnamed: 0 | B | p Value | 95% CI | 95% CI.1 |
| --- | --- | --- | --- | --- |
| | B | p Value | Upper limit | Lower limit |
| sUA | 0.177 | <.001 | 0.118 | 0.256 |
| tIL‐1β | 0.192 | <.001 | 0.111 | 0.273 |
## Tear uric acid and cytokine levels between asymptomatic hyperuricemia patients and acute gout patients
Eighteen patients in the hyperuricemia group experienced acute gout attacks before and were in intermittent period at the time of enrollment. Therefore, we divided the hyperuricemia subjects into the asymptomatic hyperuricemia group and gout group. As indicated in Table 4, serum uric acid level was slightly higher in the asymptomatic hyperuricemia group than that in gout group; and tear uric acid concentration was significant higher in asymptomatic hyperuricemia group than in gout patients. However, after adjusting for serum uric acid level, the difference of tear uric acid level between two groups was no longer significant ($$p \leq .076$$). Furthermore, there were no statistical differences of tear IL‐1β and TNF‐α levels between asymptomatic hyperuricemia group and gout group, suggesting that ocular inflammation has been overexpressed during the asymptomatic hyperuricemia period.
**Table 4**
| Unnamed: 0 | Asymptomatic HUA | Gout | p Value |
| --- | --- | --- | --- |
| N (male) | 40 (32) | 18 (17) | 0.249 |
| sUA (umol/L) | 497.8 ± 66.1 | 439.3 ± 118.9 | 0.063 |
| tUA (umol/L) | 109.3 ± 54.5 | 73.5 ± 33.7 | 0.013 |
| tIL‐1β (pg/mL) | 222.3 ± 115.2 | 183.1 ± 109.3 | 0.229 |
| tTNF‐α (pg/mL) | 138.6 ± 57.7 | 116.1 ± 47.9 | 0.429 |
## DISCUSSION
The current study demonstrated that hyperuricemia patients had significantly higher tear uric acid levels than healthy controls. Tear IL‐1β concentration was elevated in the hyperuricemia group, and significantly positively linked to tear uric acid level, implying an interaction between hyperuricemia and inflammation responses. However, no significant differences existed in tear IL‐1β or TNF‐α levels between asymptomatic hyperuricemia and acute gout flare groups, suggesting that as early as the asymptomatic hyperuricemia period, eye inflammation may exist.
This study discovered that the average uric acid concentration in tear fluid of control subjects and hyperuricemia patients was 42.7 and 98.2 μmol/L, respectively, significantly lower than those in serum. Consistent with our study, a former study found that mean concentrations of uric acid were lower in tear than in serum (119 vs. 345 μmol/L). 10 At the same time, we found that the level of tear uric acid in this study was lower than that reported in other literatures, which may be related to the different methods of tear collection and uric acid detection techniques used.
Hyperuricemia can cause ocular surface abnormalities, such as tophi deposition, subconjunctival transparent vesicles and hemorrhage, and vascular changes. 11 *Conjunctival hyperemia* and subconjunctival hemorrhage exacerbated by purine intake are two of the more common manifestations. 4 Other ocular and adnexal structures may be affected by urate deposition and associated inflammation, with potentially sight‐threatening consequences. 4 Multiple studies have implicated the relationship between serum uric acid and IL‐1β release. 12 IL‐1β plays a key role in the pathogenesis and in the screen for specific treatments of various inflammatory and degenerative eye diseases. 13 For instance, studies using a uveitis rat model reported a significant upregulation of IL‐1β and TNF‐α gene expression in the stage of active intraocular inflammation. 14, 15 IL‐1 antagonists have been successfully used to treat uveitis in monogenic autoinflammatory diseases such as Blau syndrome and cryopyridine‐associated periodic syndrome or in complex polygenic autoinflammatory diseases such as Behçet's disease. 16, 17 Likewise, treatment with the IL‐1 receptor blocker anakinra has also been shown to be successful in scleritis and episcleritis in different rheumatic conditions. 18, 19 Thus, IL‐1β concentration may be considered a potential indicator and treat target of hyperuricemia‐associated ocular disorders. The current study indicated that hyperuricemia patients had significantly higher tear IL‐1β levels. A positive association was observed between tear IL‐1β and serum/tear uric acid levels, implying that elevated tear uric acid level may be involved in the development of eye inflammation related to hyperuricemia.
The actual hyperuricemia cut‐off is principally based on the saturation point of uric acid. Recent study has shown that the negative impact of cardiovascular disease could occur at lower levels. 20 Another study indicated that the optimal cut‐off values for serum uric acid to identify metabolic syndrome were 6.3 mg/dL in men and 4.9 mg/dL in women. 21 These results emphasizes the importance of treat‐to‐target for approach of uric acid lowering treatment in patients with hyperuricemia. However, there are no data about the uric acid cut‐off associated to the risk for uric acid‐induced ocular damage, which needs further research in the future.
However, compared with patients with asymptomatic hyperuricemia and acute gout attacks, there was no significant difference in the level of IL‐1β. Typically, gout is recognized as painful arthritis due to urate deposition, which can be treated intermittently with anti‐inflammatory drugs and some medical associations recommend initiating uric acid‐lowering therapies only when repeated flares occur. 22 Systemic deposition of urate and resulting chronic inflammation may be the potential link to the frequent comorbidities associated with gout. 23 Our results suggest that as early as asymptomatic hyperuricemia period, eye inflammation may exist.
Several limitations should be considered. First, the sample size in this study was small, especially in some of the subgroup analyses. Second, our cross‐sectional study design precludes commenting on persistence or change in tear uric acid and cytokines concentration, and MSU disposition features in hyperuricemia and gout patients over time. Third, the ophthalmic manifestations of gout are rare but diverse. In this study, we did not find significant corneal urate deposition. However, direct clinical data were not available to assess ocular or extraocular manifestations. Fourth, tear IL‐1β and TNF‐α was explored in this study. Many cytokines such as IL‐1Ra, IL‐33, IL‐37, IL‐38, and IL‐6 levels were not assessed.
In conclusion, we have demonstrated that upregulated uric acid and IL‐1β levels in the tear fluid of hyperuricemia and gout patients. Upregulated tear uric acid and IL‐1β levels identified in the eyes of hyperuricemia patients imply IL‐1β‐mediated inflammation may be a plausible mechanism underlying hyperuricemia and gout‐associated ocular symptoms. Moreover, asymptomatic hyperuricemia population present the same level of IL‐1β and TNF‐α compared with patients with gout flare, highlighting the importance of physicians paying special attention to asymptomatic hyperuricemia. More prospective studies with a larger sample size are required, however, to confirm the current study's findings and explore the association of elevated cytokines and ocular symptoms in hyperuricemia and gout patients.
## AUTHOR CONTRIBUTIONS
Mian Wu: Data curation; funding acquisition; investigation; software; writing—original draft; writing—review and editing. Xingna Hu: Data curation; formal analysis; methodology; software; writing—original draft. Ting Lu: Data curation; formal analysis; methodology. Chenxiao Liu: Data curation; funding acquisition; visualization. Honghong Lu: Conceptualization; data curation; project administration; resources; supervision; validation; writing—review and editing.
## CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
## DATA AVAILABILITY STATEMENT
All the data used to support the findings of this study are available from the corresponding author upon reasonable request.
## References
1. George C, Minter DA. *Hyperuricemia* (2021)
2. Yip K, Cohen RE, Pillinger MH. **Asymptomatic hyperuricemia: is it really asymptomatic?**. *Curr Opin Rheumatol* (2020) **32** 71-79. PMID: 31688126
3. Forrester JV, Kuffova L, Delibegovic M. **The role of inflammation in diabetic retinopathy**. *Front Immunol* (2020) **11**. PMID: 33240272
4. Ao J, Goldblatt F, Casson RJ. **Review of the ophthalmic manifestations of gout and uric acid crystal deposition**. *Clin Experiment Ophthalmol* (2017) **45** 73-80. PMID: 27010973
5. Coassin M, Piovanetti O, Stark WJ, Green WR. **Urate deposition in the iris and anterior chamber**. *Ophthalmology* (2006) **113** 462-465. PMID: 16513461
6. Capaccio A, Sasso A, Rusciano G. **Raman analysis of tear fluid alteration following contact lense use**. *Sensors* (2019) **19** 3392. PMID: 31382386
7. Wu M, Tian Y, Wang Q, Guo C. **Gout: a disease involved with complicated immunoinflammatory responses: a narrative review**. *Clin Rheumatol* (2020) **39** 2849-2859. PMID: 32382830
8. Neogi T, Jansen TLTA, Dalbeth N. **2015 gout classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative**. *Ann Rheum Dis* (2015) **74** 1789-1798. PMID: 26359487
9. Markoulli M, Papas E, Petznick A, Holden B. **Validation of the flush method as an alternative to basal or reflex tear collection**. *Curr Eye Res* (2011) **36** 198-207. PMID: 21275520
10. Horwath‐Winter J, Kirchengast S, Meinitzer A, Wachswender C, Faschinger C, Schmut O. **Determination of uric acid concentrations in human tear fluid, aqueous humour and serum**. *Acta Ophthalmol* (2009) **87** 188-192. PMID: 18547277
11. Lin J, Zhao GQ, Che CY, Yang SS, Wang Q, Li CG. **Characteristics of ocular abnormalities in gout patients**. *Int J Ophthalmol* (2013) **6** 307-311. PMID: 23826523
12. Klück V, Liu R, Joosten LAB. **The role of interleukin‐1 family members in hyperuricemia and gout**. *Joint Bone Spine* (2021) **88**. PMID: 33129923
13. Fabiani C, Sota J, Tosi GM. **The emerging role of interleukin (IL)‐1 in the pathogenesis and treatment of inflammatory and degenerative eye diseases**. *Clin Rheumatol* (2017) **36** 2307-2318. PMID: 28032234
14. Yoshida M, Yoshimura N, Hangai M, Tanihara H, Honda Y. **Interleukin‐1 alpha, interleukin‐1 beta, and tumor necrosis factor gene expression in endotoxin‐induced uveitis**. *Invest Ophthalmol Visual Sci* (1994) **35** 1107-1113. PMID: 8125721
15. Planck SR, Huang XN, Robertson JE, Rosenbaum JT. **Cytokine mRNA levels in rat ocular tissues after systemic endotoxin treatment**. *Invest Ophthalmol Visual Sci* (1994) **35** 924-930. PMID: 8125755
16. Simonini G, Xu Z, Caputo R. **Clinical and transcriptional response to the long‐acting interleukin‐1 blocker canakinumab in Blau syndrome‐related uveitis**. *Arthritis Rheum* (2013) **65** 513-518. PMID: 23124805
17. Gül A, Tugal‐Tutkun I, Dinarello CA. **Interleukin‐1β‐regulating antibody XOMA 052 (gevokizumab) in the treatment of acute exacerbations of resistant uveitis of Behçet's disease: an open‐label pilot study**. *Ann Rheum Dis* (2012) **71** 563-566. PMID: 22084392
18. Knickelbein JE, Tucker WR, Bhatt N. **Gevokizumab in the treatment of autoimmune non‐necrotizing anterior scleritis: results of a phase I/II clinical trial**. *Am J Ophthalmol* (2016) **172** 104-110. PMID: 27663070
19. Botsios C, Sfriso P, Ostuni PA, Todesco S, Punzi L. **Efficacy of the IL‐1 receptor antagonist, anakinra, for the treatment of diffuse anterior scleritis in rheumatoid arthritis. Report of two cases**. *Rheumatology* (2007) **46** 1042-1043. PMID: 17449489
20. Maloberti A, Giannattasio C, Bombelli M. **Hyperuricemia and risk of cardiovascular outcomes: the experience of the URRAH (Uric Acid Right for Heart Health) project**. *High Blood Press Cardiovasc Prev* (2020) **27** 121-128. PMID: 32157643
21. Zhang M, Gao Y, Wang X, Chang H, Huang G. **Serum uric acid and appropriate cutoff value for prediction of metabolic syndrome among Chinese adults**. *J Clin Biochem Nutr* (2013) **52** 38-42. PMID: 23341696
22. Qaseem A, Harris RP, Forciea MA, Denberg TD. **Management of acute and recurrent gout: A clinical practice guideline from the American College of Physicians**. *Ann Intern Med* (2017) **166** 58-68. PMID: 27802508
23. Khanna P, Johnson RJ, Marder B, LaMoreaux B, Kumar A. **Systemic urate deposition: an unrecognized complication of gout?**. *J Clin Med* (2020) **9** 3204. PMID: 33023045
|
---
title: Identification of altered miRNAs and their targets in placenta accreta
authors:
- José M. Murrieta-Coxca
- Emanuel Barth
- Paulina Fuentes-Zacarias
- Ruby N. Gutiérrez-Samudio
- Tanja Groten
- Alexandra Gellhaus
- Angela Köninger
- Manja Marz
- Udo R. Markert
- Diana M. Morales-Prieto
journal: Frontiers in Endocrinology
year: 2023
pmcid: PMC10022468
doi: 10.3389/fendo.2023.1021640
license: CC BY 4.0
---
# Identification of altered miRNAs and their targets in placenta accreta
## Abstract
Placenta accreta spectrum (PAS) is one of the major causes of maternal morbidity and mortality worldwide with increasing incidence. PAS refers to a group of pathological conditions ranging from the abnormal attachment of the placenta to the uterus wall to its perforation and, in extreme cases, invasion into surrounding organs. Among them, placenta accreta is characterized by a direct adhesion of the villi to the myometrium without invasion and remains the most common diagnosis of PAS. Here, we identify the potential regulatory miRNA and target networks contributing to placenta accreta development. Using small RNA-Seq followed by RT-PCR confirmation, altered miRNA expression, including that of members of placenta-specific miRNA clusters (e.g., C19MC and C14MC), was identified in placenta accreta samples compared to normal placental tissues. In situ hybridization (ISH) revealed expression of altered miRNAs mostly in trophoblast but also in endothelial cells and this profile was similar among all evaluated degrees of PAS. Kyoto encyclopedia of genes and genomes (KEGG) analyses showed enriched pathways dysregulated in PAS associated with cell cycle regulation, inflammation, and invasion. mRNAs of genes associated with cell cycle and inflammation were downregulated in PAS. At the protein level, NF-κB was upregulated while PTEN was downregulated in placenta accreta tissue. The identified miRNAs and their targets are associated with signaling pathways relevant to controlling trophoblast function. Therefore, this study provides miRNA:mRNA associations that could be useful for understanding PAS onset and progression.
## Introduction
Placenta Accreta Spectrum (PAS) is the term that integrates the different grades of abnormal placental adherence and invasion [1, 2]. The most current classification of PAS includes three grades: 1. Placenta accreta or adherenta, 2. Placenta increta and 3. Placenta percreta. In adherent placenta accreta, the decidua basalis is partly or completely lost so that the trophoblast layer is directly apposed to the myometrial tissue, but without invading it. Placenta increta is defined by deep invasion of trophoblast cells into the myometrium, and placenta percreta when trophoblast cells invade and penetrate the uterine serosa [2, 3]. An additional classification recently suggested by FIGO included the subclassification of grade 3a: *Placenta percreta* limited to the uterine serosa grade 3b: *Placenta percreta* with urinary bladder invasion; and grade 3c: *Placenta percreta* with the involvement of pelvic organs (2–4). PAS severity can be evaluated during pregnancy by ultrasound examinations [5, 6], but many cases remain undiagnosed antepartum and are later classified based on the intraoperative situation and histological findings related to partial or complete loss of decidua basalis and the depth of myometrial trophoblast cell invasion [2, 3, 5]. The most common diagnosis of PAS cases is placenta accreta (>$70\%$), followed by increta (~$15\%$) and percreta ($10\%$) [5, 7, 8].
The differences in the clinical and histopathological criteria used to define PAS in previous studies make it difficult to compare data and reach a consensus on the etiology of the disease. A failure in normal decidualization caused by previous endometrial damage is currently the most favored hypothesis, but the abnormal invasive capacities of trophoblast cells may also contribute to the disease [9, 10]. PAS is frequently associated with a previous cesarean section, which is increasingly applied, thus, the worldwide prevalence of PAS has risen over the last four decades [11, 12]. However, only a few investigations have focused on the molecular mechanisms associated with PAS that may also explain its development in women during their first pregnancy. As PAS does not naturally occur in animals and due to its uniqueness to human pregnancy [3, 13], its study in animal models is highly constricted despite one published mouse model [14].
A recent study reported a very low correlation between transcriptome and proteome profiling of PAS samples suggesting a significant role of post-transcriptional regulation [15], which may be mediated by non-coding RNAs including miRNAs. The human placenta harbors miRNAs (20-22 nucleotides size), which regulate its development and functionality [16]. This regulation is further demonstrated by the fact that placenta-specific and -associated miRNAs have particular expression patterns during different stages of pregnancy (17–19). As demonstrated in several pioneer studies, dysregulation of specific miRNAs is associated with pregnancy pathologies, including PAS (20–23). Some of the altered miRNAs are linked with intracellular signaling networks implicated in angiogenesis [24], trophoblast apoptosis [20], and epithelial-mesenchymal transition [23].
In this study, next-generation sequencing was used to screen the miRNA signature of PAS placentas and compare it to healthy pregnancies. Altered miRNAs, as well as their targets, were validated and localized in placental tissues. Our results offer molecular elements for understanding the etiology of PAS and promote the identification of markers in PAS.
## Material and methods
The Placenta Lab strictly applies quality management and is certified after DIN EN ISO 9001.
## Patient samples
Patients were recruited for the study from the Department of Obstetrics, University Hospital Jena and the Department of Gynecology and Obstetrics, University Hospital Essen, Germany, between 2014 and 2018. The respective ethics committees approved the study according to the Helsinki Declaration on ethical principles for medical research involving human subjects by obtaining consent forms (Amendment to No: 1509-$\frac{03}{05}$ Jena and 12-5212-BO Essen). Multiple pregnancies, fetal anomalies, and infections were excluded from the study.
Samples obtained in Jena were collected intraoperatively in cases where a postpartum curettage had to be performed due to incomplete placenta or retention of the placenta after delivery, even under administration of uterotonic drugs or controlled cord traction. An additional sample was taken during cesarean section of a patient with placenta praevia, intraoperatively diagnosed to be an abnormal adherent placenta. Placenta tissue samples were taken in all cases from the suspected sight of detachment failure. Histopathological findings such as loss of decidua basalis or the direct trophoblast apposition to the myometrial tissue without invasion were used as confirmers for the diagnosis of placenta accreta. For controls, placental chorionic tissue was collected from normally delivered placentas including (for ISH and IF) or excluding decidua (RNA analysis). Samples were immediately washed with sterile phosphate buffer solution (PBS) and placed in RNA later (cat. No. AM7021; Invitrogen Life Technologies, Darmstadt, Germany) overnight and then stored in cryotubes at -80°C until RNA extraction, or fixed in $4\%$ formalin overnight before paraffin-embedding. A sub-set of 17 samples was used for the initial RNA-*Seq analysis* and a complete set of 26 samples to perform the validation PCR. Samples for immunostaining and in situ hybridization were also selected from the complete set. Clinical characteristics are summarized in Table 1.
**Table 1**
| Unnamed: 0 | Initial RNA-Seq | Initial RNA-Seq.1 | Initial RNA-Seq.2 | PCR Validation cohort | PCR Validation cohort.1 | PCR Validation cohort.2 | Immunostaining | Immunostaining.1 | Immunostaining.2 | ISH | ISH.1 | ISH.2 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | NP (n=9) | PAS (n=8) | p- Value | NP (n=14) | PAS (n=12) | p- Value | NP (n=3) | PAS (n=3) | p- Value | NP (n=4) | PAS (n=6) | p- Value |
| Maternal age (y) | 28.9 ± 3.0 | 30.2 ± 7.5 | 0.62 | 30.4 ± 3.5 | 30.6 ± 6.8 | 0.91 | 31.3 ± 10.0 | 38.7 ± 2.9 | 0.29 | 30.8 ± 7.2 | 37.8 ± 2.3 | 0.05 |
| Gestational age (w) | 39.8 ± 0.7 | 39.4 ± 2.2 | 0.58 | 39.2 ± 1.5 | 39.6 ± 1.8 | 0.51 | 39.5 ± 1.6 | 37.9 ± 3.3 | 0.50 | 40.1 ± 1.4 | 36.6 ± 4.7 | 0.19 |
| Placenta weight (g) | 641.5 ± 93.2 | 509.5 ± 45.5 | *<0.05 | 632.8 ± 116.6 | 536.3 ± 56.9 | 0.07 | 664.3 ± 90.7 | 527.3 ± 112.8 | 0.18 | 654.8 ± 70.2 | 527.3 ± 112.8 # | 0.12 |
| Birth weight (g) | 3512.2 ± 295.2 | 3164.4 ± 434.4 | 0.07 | 3469.3 ± 375.9 | 3429.2 ± 570.0 | 0.83 | 3481.7 ± 123.9 | 3016.7 ± 767.9 | 0.36 | 3598.8 ± 215.6 | 2814.2 ± 918.8.9 | 0.14 |
| Size birth (cm) | 51.7 ± 2.0 | 51.4 ± 2.0 | 0.73 | 51.5 ± 2.4 | 52.2 ± 2.0 | 0.38 | 53.0 ± 1.7 | 49.2 ± 0.8 | 0.02 | 52.5 ± 1.7 | 49.2 ± 0.8 | *<0.05 |
| Neonate gender (% male) | 33.3 | 87.5 | | 37.5 | 83.3 | | 66.7 | 33.3 | | 50 | 50 | |
| Delivery mode (% Cesarean) | 44.4 | 25 | | 64.3 | 16.6 | | 66.7 | 33.3 | | 50 | Accreta: 25Increta:100Percreta: 100 | |
| Gravida | 1.8 ± 1.1 | 1.8 ± 0.9 | 0.95 | 2.4 ± 1.6 | 2.1 ± 1.2 | 0.54 | 4.3 ± 3.0 | 3.7 ± 1.2 | 0.74 | 2.8 ± 2.9 | 4.3 ± 1.6 | 0.29 |
| Parity | 1.7 ± 0.9 | 1.5 ± 0.5 | 0.64 | 2.0 ± 1.4 | 1.7 ± 0.9 | 0.39 | 3.0 ± 2.0 | 1.3 ± 0.6 | 0.24 | 2.2 ± 1.9 | 2.9 ± 1.5 | 0.82 |
| PAS disorder | None (9) | Accreta (8) | | None (14) | Accreta (12) | | None (3) | Accreta (3) | | None (4) | Accreta (4) Increta (1) Percreta (1) | |
Placental tissue from Essen was obtained at the time of vaginal delivery or caesarian section from cases where PAS was diagnosed at the third stage of delivery or based on antepartum ultrasound measurements. Samples were classified into placenta accreta, increta or percreta (each $$n = 1$$) according to the criteria defined by Cali et al. [ 6] and following the International Federation of Gynecology and Obstetrics (FIGO) guidelines based on intraoperative situation and histological findings [2]. Intraoperatively, the area of placental tissue with the highest degree of invasion was chosen for analysis and was collected including surrounding tissue (decidua, myometrium, uterine serosa, broad ligament tissue). In the cases of placenta increta and percreta, the specimens were obtained by focal resection of the placenta or hysterectomy. Tissue samples were fixed in $4\%$ formalin overnight followed by standard processing to obtain paraffin-embedded sections for ISH. The clinical characteristics are summarized in Table 1.
## RNA isolation
Total RNA was isolated using a mirVana™ miRNA Isolation Kit (cat. No. AM1561; Invitrogen), according to the manufacturer’s protocol. Approximately 100 mg placenta tissue per sample was transferred to a Medicon (cat. No. 340591; BD Biosciences, Franklin Lakes NJ, USA) disposable for biological sample disaggregation containing 1 ml of lysis buffer (provided in the kit) and processed in a Medimachine (Dako; BD) for 20 s. Tissue suspension was collected, and Total RNA concentration was determined in a high-speed microfluidic UV/VIS spectrophotometer (QIAxpert System, Qiagen Hilden, Germany). Samples with A260/A280 ratio >1.8 were stored at −80°C until further processing.
## Next-generation sequencing
GATC Biotech AG, Konstanz, Germany, performed the next-generation sequencing. The small RNA libraries were created using Illumina’s small RNA sample preparation protocol (TruSeq Small RNA Sample Prep Kits; Illumina, San Diego, CA, USA) with minor adaptations to the manufacturer’s instructions. Single read sequencing of the libraries was performed on a HiSeq 2500 (Illumina) according to the manufacturer’s protocol. At least 10 million reads per sample were generated.
## Small RNA-Seq library processing, mapping, and differential expression analysis
First, the RA2 adapter sequences (5’-TGGAATTCTCGGGTGCCAAGG) of the TruSeq small RNA preparation kit were clipped from all reads, using cutadapt [25] (version 2.0), and all reads shorter than 15 bp or with a mean quality lower than 20 were removed subsequently. Read quality was monitored using FastQC (v0.11.3; http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Quality reports of the raw and processed RNA-Seq libraries can be found at https://osf.io/8wq9h. Mapping was performed using TopHat2 [26](version 2.1.1) with standard parameters onto the human reference genome (Ensembl release 98), and read counting was done using the respective *Ensembl* gene annotation. For counting, featureCounts [27] (version 1.6.3) with the parameters -M, -O 0.5 was used to count reads for the mature miRNA annotations of each human pre-miRNA separately. Analysis of differentially expressed miRNAs (DEmiRNAs), as well as plotting PCA, was performed by the Bioconductor R package DESeq2 [28] (version 1.10.0). Multiple testing adjustment of the resulting p-values was performed using Benjamini and Hochberg’s FDR approach [29]. Mature miRNA loci with an identified adjusted p-value < 0.05 were considered differentially expressed. RNA-Seq datasets including DEmiRNA results and the count values are available at NCBI’s GEO database https://www.ncbi.nlm.nih.gov/geo/ under the accession IDGSE216742.
## Biological pathway analysis and interaction network of miRNA targets
Target mRNAs of DEmiRNAs were obtained from the miRTarBase [30] (release 8.0) to get a non-redundant list of experimentally verified genes being potentially altered in PAS. Following the strategy used in our earlier study [31], DEmiRNA targets were assigned within the human regulatory pathways of the KEGG database. KEGG pathways were ranked individually according to the number of targeted genes within each pathway. The hypergeometric test was used to assess if specific pathways were significantly targeted by calculating the corresponding p-values for each pathway. Analysis of sub-pathways was performed by manually identifying key altered regions. To estimate which of the genes of the sub-pathway were most likely affected by the altered DEmiRNAs in PAS, an impact score was calculated based on the frequencies of the genes within the enriched pathways and the amount of DEmiRNAs targeting each gene.
## Confirmation of DEmiRNAs
Expression levels of representative miRNAs were analyzed using the TaqMan™ Advanced miRNA cDNA Synthese-Kit (cat. No. A28007, Applied Biosystems, Darmstadt, Germany) and the TaqMan™ Fast Advanced Master Mix, no UNG (cat. No. A44360, Applied Biosystems) according with the manufacturer’s protocol with specific miRNA probes (hsa-miR-193b-3p, Assay ID: 478314_mir; hsa-miR-519d-3p, Assay ID: 478986_mir; hsa-miR-331-3p Assay ID: 478323_mir; hsa-miR-3074-5p Assay ID: 479606_mir; hsa-miR-24-3p Assay ID: 477992_mir; hsa-miR-382-3p Assay ID: 479458_mir; hsa-miR-376c-3p Assay ID: 478459_mir; hsa-miR-495-3p Assay ID: 478136_mir; hsa-miR-370-3p Assay ID: 478326_mir; hsa-miR-423-3p Assay ID: 478327_mir; hsa-miR-222-3p Assay ID: 477982_mir; hsa-miR-106b-3p Assay ID: 477866_mir; hsa-miR-4732-3p Assay ID: 478118_mir; hsa-miR-454-5p Assay ID: 478919_mir; hsa-mir-3615-3p Assay ID: 478837_mir; hsa-miR-16-2-3p Assay ID: 477931_mir; hsa-miR-39-3p Assay ID: 478293_mir). The Caenorhabditis elegans miRNA cel-miR-39 (Assay ID: 000200; 5′-UCACCGGGUGUAAAUCAGCUUG) was added at a concentration of 1.6 x 108 copies/μL and used as spike-in control. PCR reactions were run in duplicates including no-template controls in 96-well plates on a Mx3005P qPCR System (Applied Biosystems) using 40 cycles, at the following conditions: 95°C for 3 sec and anneal/extend at 60°C for 20 sec. Fold changes were calculated by the formula 2–ΔCt using cel-miR-39 as normalizer.
## In situ localization of DEmiRNA
Representative miRNAs were localized in the placenta tissue by using the microRNA in situ hybridization (ISH) Buffer Set for formalin-fixed paraffin-embedded (FFPE) tissue samples (cat. No. 339457; Qiagen). Specific miRCURY LNA™ microRNA detection probes (cat. No. 339501; Qiagen), as well as positive and negative controls, were purchased from Qiagen (cat. No 339451). The one-day microRNA ISH protocol was carried out according to the supplier’s recommendations. In brief, paraffin blocks were cut into 6 μm-thick sections. Slides were dewaxed in a train of different percentages of xylene and ethanol solutions ending in phosphate-buffered saline (PBS) (cat No. 14190-094; Gibco, Schwerte, Germany). Following, slides were incubated with Proteinase-K for 10 min at 37°C in a CytoBrite Duo slide incubation system (SciGene; Sunnyvale, CA, USA) and then washed twice with PBS. Hybridization mix containing 10 nM of double-DIG LNA™ microRNA probe (miR-519d-3p, miR-193b-3p, miR-106b-3p, miR-370-3p or the negative control scramble probe SCR, which represents random sequence) was added to the slides and hybridized for 1 h. Slides were then washed in a slide rack with different concentrations of 5xSSC buffer (cat. No. 15557-044; Invitrogen) and placed in PBS. A hydrophobic barrier was created around the tissue sections using a Dako-Pen (Cat. No. H-4000; Vector Laboratories, Newark, CA, USA), and slides were incubated in a humidifying chamber with a blocking solution for 15 min. The blocking solution was removed, anti-DIG reagent (cat. No. 11093274910; Sigma Aldrich; Taufkirchen, Germany) was applied on the slides for 60 min incubation at RT. Sections were incubated with freshly prepared alkaline phosphatase substrate (cat. No. 11697471001; Merck, Darmstadt, Germany) for 2 h at 30°C, protected from light in the humidifying chamber. The reaction was stopped by incubating slides in KTBT buffer (Potassium-Tris Buffer with Triton). Nuclear Fast Red™ (cat. No. H-3403; Vector Laboratories) was applied for 1 min for nuclear counterstaining. Slides were dehydrated in ethanol solutions and mounted with 1-2 drops of mounting medium (cat. No. 03989; Sigma Aldrich), avoiding air-drying. The precipitate was allowed to settle overnight, and slides were analyzed using an Axio Imager A2 microscope and Zen Blue software (Carl Zeiss Microscopy GmbH, Jena Germany).
## Expression of target mRNAs
Total RNA (300 ng) from NP and placenta accreta samples PAS was used to analyze the expression of selected mRNAs by reverse transcription using High-Capacity RNA-to-cDNA™ Kit (cat. No. 4368814; Applied Biosystems). Quantitative real-time PCR was performed using TaqMan assays (ERK1, Assay ID: Hs00385075_m1; NFKB1, Assay ID: Hs00765730_m1; AKT1, Assay ID: Hs00178289_m1; PTEN, Assay ID: Hs02621230_s1; STAT3, Assay ID: Hs00374280_m1; TGFB1, Assay ID: Hs00171257_m1; and GAPDH, Assay ID: Hs03929097_g1) and TaqMan Universal PCR Master Mix reagents (cat. No. 4440040; Applied Biosystems). qPCR was run on a Mx3005P qPCR System (Applied Biosystems). mRNA expression was normalized using the 2−ΔCt method relative to GAPDH.
## Immunofluorescence staining
Paraffin-embedded tissue sections were deparaffinized, hydrated in a graded ethanol series, and quenched by antigen retrieval with a citrate buffer (10 mM Sodium citrate, $0.05\%$ Tween 20, pH 6.0) at >95° C for 10 min. Tissue sections were blocked with $0.1\%$ BSA for 20 min and incubated with the primary antibodies mouse-anti-cytokeratin-7 (cat. no. MA1-06316; Invitrogen), rabbit-anti-PTEN (cat. No. 9559S; Cell Signaling, Danvers, MA, USA), and rabbit-anti-NF-κB (cat. No. SC-109; Santa Cruz Biotechnology, Heidelberg, Germany) for 2 h at 37°C in a humid atmosphere followed by incubation with the secondary antibodies goat anti-mouseAF488 (cat. No. A11017; Invitrogen) or goat anti-rabbitAF647 (cat. No. A21246; Invitrogen). All antibodies diluted 1:200 were applied and incubated 1 h at 37°C under humidity. DAPI (1 µg/mL) (cat. No. D9542; Sigma Aldrich) was used for nuclei staining. Fluorescence was visualized and recorded using a Zeiss LSM 710 confocal laser scanning microscope (Carl Zeiss Microscopy GmbH).
## Statistical analysis
Unpaired Student t-test with Mann-Whitney test was applied to assess differences between groups using Prism software version 9 (GraphPad, San Diego, CA) as indicated at every figure legend. A p-value < 0.05 was considered significant.
## Identification of DEmiRNA in placenta accreta by high-throughput small RNA-Seq
Normal (NP; $$n = 9$$) and adherent accreta (PAS; $$n = 8$$) placentas were analyzed by sRNA-Seq. At least 10 million reads per sample were obtained and used for library processing and mapping. Principal component analysis (PCA) revealed a separation of samples belonging to NP and PAS groups with some overlaps (Figure 1A). Most small RNA molecules were identified as miRNA species ($39.9\%$), followed by small nucleolar RNAs (snoRNA; $22.2\%$) and long non-coding RNAs (lncRNA; $16.3\%$). A minor proportion included small nuclear RNAs (snRNA; $5.3\%$) and ribosomal RNA (rRNA; $2.0\%$) (Figure 1B). To identify the significant genes in PAS ($p \leq 0.05$), the DESeq2 R package was used. Placental tissues of NP and PAS shared a common miRNA signature consisting of 994 active miRNAs. Exclusively expressed were 95 miRNA species in NP and 37 in PAS placentas (Figure 1C). A total of 147 mature miRNAs were up- and 151 were downregulated in PAS compared to NP (Figure 1D). A selective analysis of miRNAs (17, 32–34) revealed DEmiRNAs in the placenta-associated clusters including the chromosome 19 miRNA cluster (C19MC; 33 out of 46 miRNA species in the cluster), the chromosome 14 miRNA cluster (C14MC; 19 out of 42), the miR-$\frac{17}{92}$ cluster (4 out of 6), the miR-106a cluster (3 out of 6), and the miR-106b cluster (3 out of 3), but not the miR-371 cluster (Figure 1E). A full list of miRNAs included in the clusters is presented in Supplementary Table. 1.
**Figure 1:** *Small RNA-Seq analysis reveals a distinct expression pattern of miRNAs in placenta accreta samples. (A) PCA of the investigated samples based on all detected miRNAs. (B) Mapped sRNA reads were sorted into RNA classes. (C) Overlap of the actively transcribed miRNA genes in NP and PAS samples. (D) MA plot showing mature DEmiRNAs in PAS relative to NP. The x-axis is the log2 average expression over all samples, and the y-axis is the log2 fold change between PAS and NP groups. Red and blue dots represent respectively the significant differentially up- and down-expressed miRNAs. (E) Number of DEmiRNAs in PAS that belong to placental miRNA clusters. Numbers indicate total miRNA species in the cluster/miRNAs upregulated/miRNAs downregulated. NP, Normal pregnancy; PAS, Placenta Accreta Spectrum; C19MC, chromosome 19 miRNA cluster; C14MC, chromosome 14 miRNA cluster.*
## Validation of DEmiRNAs in placenta accreta
DEmiRNAs were sorted according to the adjusted p-value, and a group of 16 DEmiRNAs exhibiting fold-change > 2.0, and good abundance (base mean > 100) were selected for individual validation using RT-qPCR in a larger cohort of samples (NP:14; *Placenta accreta* PAS:12) that includes the ones used for RNA-Seq. In this group, members of the C19MC (miR-519d-3p) and C14MC (miR-370-3p and miR-454-5p), as well as miRNAs not reported in PAS were included. Small RNA-*Seq data* was successfully validated in eight out of eight selected upregulated miRNAs in PAS samples: miR-24-3p, miR-193b-3p, miR-331-3p, miR-376c-3p, miR-382-3p, miR-495-3p, miR-519d-3p and miR-3074-5p (Figure 2A). Among downregulated miRNAs in PAS, five out of eight miRNA species were validated by RT-PCR (miR-106b-3p, miR-222-3p, miR-370-3p, miR-454-5p, and miR-3615-3p (Figure 2B).
**Figure 2:** *Validation of DEmiRNAs in placenta accreta samples. Expression patterns of differentially expressed miRNAs (DEmiRNAs) identified by RNA-Seq were validated by qRT-PCR in a larger sample cohort NP (n = 14) and PAS (n = 12). (A) Upregulated miRNAs and (B) downregulated miRNAs. The relative expression of each unique miRNA was normalized to the value of the exogenous cel-miR-39 using the 2-ΔCt formula. Data are shown as the mean ± SE. Significant differences were determined by unpaired t- and Mann-Whitney test. ***p < 0.001, **p < 0.01, *p < 0.05.*
## DEmiRNAs localize mainly in trophoblast but also in endothelial cells
To determine the possibility of cell-specific expression, localization of DEmiRNAs in PAS was examined by ISH within placental villous tissue (Figure 3). Tissue sections of placenta accreta, placenta increta and percreta were stained with hematoxylin and eosin (H&E) to visualize morphological differences. Abnormally deep anchoring of the placental villi, as well as fibrin and trophoblast cells invaded into decidual tissue, were present in PAS samples (Figure 3). According to CK7 expression by IHC, extravillous trophoblast cells (EVTs) deeply infiltrating the decidual tissue were often observed in invasive PAS but not in NP samples (red arrows in Figure 3). To investigate the location of DEmiRNAs in the tissue, in situ hybridization was performed using digoxigenin-labeled LNA probes, which bind specifically to their target miRNA or that contain a random non-genomic scramble sequence (SCR) as negative control. ISH revealed miR-193b-3p signal in STB of both PAS and NP placentas, elevated miR-193b-3p expression was observed in PAS compared to NP samples, especially in the EVTs and areas of trophoblast invasiveness into the decidual tissue. The expression of miR-519d-3p, a placenta-specific miRNA, was restricted to trophoblast cells and strongly present on invasive trophoblast cells of PAS tissue. In NP tissue, the expression of miR-106b-3p and miR-370-3p was found mainly delimited in trophoblast cells, although endothelial cells were also positive for miR-370-3p. In PAS, miR-370-3p was highly expressed by invasive trophoblast cells. Contrary to PCR results, a downregulation of miR-106b-3p and miR-370-3p was not observable in PAS compared to NP samples.
**Figure 3:** *Differentially expressed miRNAs visualized by in situ hybridization in normal placenta (NP), placenta accreta, and placenta increta and percreta samples (invasive PAS). Areas containing villi and uterine tissue have been selected, in placenta accreta and invasive PAS with implanted villi and extravillous trophoblast cells (EVTs). The hematoxylin and eosin (H&E) staining shows nuclei in blue, cytoplasm in pink; cytokeratin-7 (CK7) staining marks positive cells brown (mainly syncytiotrophoblast and trophoblast cells) red arrows show areas of deep trophoblast infiltration; in situ hybridization of miR-193b-3p, miR-519d-3p, miR-106b-3p, and miR-370-3p shows positive cells in blue. Sections were counterstained with Nuclear Red. IVS, intervillous space; DE, decidua; FV, fetal vessel; S, Syncytiotrophoblast. Scale bar: 100 µm.*
## Biological pathway analysis identifies cell cycle and inflammation pathways as networks of DEmiRNAs in placenta accreta
To explore alterations in gene expression, a hypergeometric test was run to identify KEGG pathways that the identified DEmiRNAs could alter. This analysis is based on the number of genes involved in each pathway which the DEmiRNAs can potentially regulate. As a result, 87 potentially altered pathways were identified. An additional inspection of these pathways allowed the identification of 7 shared sub-pathways, including cell cycle control, actin regulation, TGF-β, MAPK, PI3K-AKT, NF-κB, and the JAK-STAT signaling pathways (Figure 4A). A representative gene from each pathway was selected for validation by PCR. No significant expression difference was found for ERK1 and AKT mRNA (MAPK and PI3K-AKT signaling pathways are targeted by miR-382-3p and miR-495-3p), but a significant reduction of NF-κB mRNA was confirmed in PAS samples. As a representative of the JAK-STAT pathway with high invasion-inducing capacities [35], STAT3 expression was investigated but was not altered in PAS. Among the TGF-β pathway, TGF-β1 was downregulated in PAS samples. PTEN mRNA, which is involved in the cell cycle control pathway and a potential target of miR-106b-3p, miR-222-3p, and miR-519d-3p [36], was also decreased in PAS compared to NP samples (Figure 4B).
**Figure 4:** *Expression of potential DEmiRNA targets in PAS. (A) Enriched pathways found in the KEGG analysis and putative DEmiRNAs targeting components of the pathway. The score assigned to each gene roughly describes the probability that this gene is regulated by the DEmiRNAs. The higher the score, the greater the influence of the DEmiRNAs on the gene within the pathway. (B) Expression of DEmiRNA mRNA targets by qRT-PCR in NP (n = 14) and PAS placentas (n = 12). The relative expression of each unique mRNA was normalized using the formula 2-ΔCt with GAPDH as endogenous control. Data are shown as the mean ± SE. Significant differences were determined by the unpaired t- and Mann-Whitney test. *P < 0.05, **P < 0.01..*
## PTEN is down- while NF-κB is upregulated in placenta accreta
Based on the network of signaling pathways described above, two main transcription factors were selected for further investigation: PTEN, which is involved in cell cycle functions including proliferation, migration, and metabolism [37], and NF-κB, which is involved in the expression of inflammatory factors [38]. To localize these proteins in placental tissues from NP and placenta accreta PAS samples, double immunofluorescence staining was carried out as described in the method section. In NP, the placenta villi appeared well delimitated by CK7 positive STB as observed in IHC staining. In contrast, PAS tissue showed zones with unorganized STB and the presence of large areas of EVT infiltration in the decidual tissue (Figure 5). In NP and PAS placenta, NF-κB was expressed in STB and the stroma of placental villi. Additionally, focal expression in areas of column-like EVTs was also observed in PAS (Figure 5A, white arrow). Contrary to the mRNA analysis, the fluorescence intensity of NF-kB protein was higher in PAS than NP (Figure 5C). In NP, PTEN was localized mainly in STB, endothelial cells surrounding fetal blood vessels, and in minor proportion in the stroma. Contrarily, in PAS samples, PTEN localized mainly in the stroma, partially in Hofbauer cells (white arrowheads), and in minor proportion in STB and EVTs (Figure 5B). In agreement with the mRNA validation, PTEN protein expression was reduced in PAS compared to NP tissues (Figure 5C).
**Figure 5:** *Immunolocalization of NF-κB and PTEN in normal and placenta accreta samples. Double immunolabelling of cytokeratin-7 (CK7, pseudo-green) with (A) NF-κB (pseudo-red) or (B) PTEN (pseudo-violet). Yellow boxes show the zoom-in area of white dotted boxes. White arrows and arrowheads indicate respectively invasive trophoblast columns and potential hofbauer cells. The scale bar represents 100 µm. (C) Extracted fluorescent intensity from representative pictures (n=3-5) in slides of NP (n=3) and PAS (n=3). Significant differences were determined by unpaired t- and Mann-Whitney test ***p < 0.001, *p < 0.05. DE, Decidua; NC, Negative control; NP, normal pregnancy; PAS, placenta accreta spectrum.*
## Regulation of trophoblast invasion and migration are the most common functions of DEmiRNAs in PAS
To clarify the function of DEmiRNAs, literature was screened for investigations in physiological or pathological pregnancies. Among the confirmed DEmiRNAs, only miR-495-3p has not been reported as differentially expressed in any pathological pregnancy. Several DEmiRNAs have been previously identified altered in preeclampsia (PE), pre-term birth (PTB), fetal growth restriction (FGR), or gestational diabetes mellitus (GDM). To the best of our knowledge, no reports on PAS regarding the here described DEmiRNAs exist. *Target* genes confirmed in this study (e.g., TGF-β, MAPK, PETN) have been reported as validated targets of DEmiRNAs in other studies supporting the network associations proposed here. Most of the found studies described miRNA functions in trophoblast cells, and reported their association with cell migration, proliferation and epithelial-mesenchymal transition (Table 2).
**Table 2**
| miRNA | Chr. Location | Expression in healthy and pathological pregnancy | Target gene(s) | Reported function | Reference |
| --- | --- | --- | --- | --- | --- |
| miR-24-3p | chr9: 95,086,064-95,086,085 chr19: 13,836,289-13,836,310 | Upregulated in PE | TGF-β, MAPK, CDK, PI3K, p85, MYC, MM14 | Regulation of actin organization. Cell migration and proliferation. | (39, 40) |
| miR-24-3p | chr9: 95,086,064-95,086,085 chr19: 13,836,289-13,836,310 | Upregulated in PPROM and PTB | TGF-β, MAPK, CDK, PI3K, p85, MYC, MM14 | Regulation of actin organization. Cell migration and proliferation. | (41) |
| miR-193b-3p | chr16: 14,304,017-14,304,038 | Upregulated in PE | TGF-β2 | Promotes trophoblast (HTR-8/SVneo) cell motility, migration and motion | (42–44) |
| miR-193b-3p | chr16: 14,304,017-14,304,038 | Upregulated in FGR | TGF-β2 | Promotes trophoblast (HTR-8/SVneo) cell motility, migration and motion | (45) |
| miR-331-3p | chr12: 95,308,420-95,308,513 | Downregulated in PE | TGF-βR1 | Regulates the invasion of human trophoblastic HTR-8/SVneo cells | (46, 47) |
| miR-376c-3p | chr14: 101,039,732-101,039,752 (miR-379/miR-656 cluster) | Upregulated at term labor | PHLDA2, HBEGF, TGF-β1R (ALK5 and ALK7) | Promotes trophoblast outgrowth and invasion | (48) |
| miR-376c-3p | chr14: 101,039,732-101,039,752 (miR-379/miR-656 cluster) | Present in umbilical cord serum-derived exosomes | PHLDA2, HBEGF, TGF-β1R (ALK5 and ALK7) | Promotes trophoblast outgrowth and invasion | (49) |
| miR-376c-3p | chr14: 101,039,732-101,039,752 (miR-379/miR-656 cluster) | Downregulated in PE | PHLDA2, HBEGF, TGF-β1R (ALK5 and ALK7) | Promotes trophoblast outgrowth and invasion | (50–52) |
| miR-382-3p | chr14: 101,054,352-101,054,372 | Upregulated in PE | STAT1, NEAT1, ROCK1, PTEN | Inhibits cell proliferation and migration and its downregulation promotes invasiveness in cancer models | (53–55) |
| miR-495-3p | chr14: 101,033,804-101,033,825 | Not reported yet | | | |
| miR-519d-3p | chr19: 53,713,400-53,713,421 (C19MC cluster) | Expressed almost exclusively in placenta tissue | CXCL6, NR4A2, FOXL2, PDCD4, PTEN,MMP-2 | Reduces trophoblast cell migration and invasionDownregulates the EVT invasive phenotype | (56, 57) |
| miR-519d-3p | chr19: 53,713,400-53,713,421 (C19MC cluster) | Upregulated in PE | CXCL6, NR4A2, FOXL2, PDCD4, PTEN,MMP-2 | Suppresses invasion and migration of trophoblast cells by targeting MMP-2 | (58) |
| miR-3074-5p | chr9: 95,086,063-95,086,083 | Upregulated in placental villi from recurrent miscarriageLow expressed in placental villi from NP | BCL2, FGF1, P27, BCL-G, DLST, GAP43, CCR3, RUNX2 | Promotes apoptosis but inhibits invasion of HTR-8/SVneo cell line | (59, 60) |
| miR-16-2-3p | chr3: 160,404,797-160,404,818 | Upregulated in placental villi and decidua from recurrent spontaneous abortion | VEGF | Regulates placental angiogenesis and development | (61) |
| miR-16-2-3p | chr3: 160,404,797-160,404,818 | Upregulated in DICE-deficient HTR-8/SVneo trophoblast cell line | COL1A2 | Reduces invasion of HTR-8/SVneo cell line | (62) |
| miR-106b-3p | chr7: 100,094,002- 100,094,023 | Upregulated in PE | MMP-2 | Inhibits the invasion and proliferation of JAR and JEG3 cells | (63) |
| miR-222-3p | chrX: 45,747,036-45,747,056 | Upregulated in PE | BCL2L11 | Promotes apoptosis of mesenchymal stem cells in response to hypoxia | (64) |
| miR-222-3p | chrX: 45,747,036-45,747,056 | Upregulated in PE | HDAC6 | Inhibits trophoblast proliferation and migration | (65) |
| miR-370-3p | chr14: 100,911,186-100,911,207 | Dysregulated in GTDUpregulated in FGRDownregulated in first-trimester healthy placenta | | Involved in regulating proliferation, migration, and invasion of cancer cells | (19, 66, 67) |
| miR-423-3p | chr17: 30,117,131-30,117,153 | Upregulated in early onset PE | MAPK signaling pathway | | (68, 69) |
| miR-454-5p | chr17: 59,137,828-59,137,849 | Downregulated in PE | ALK7 | Promotes proliferation reduces apoptosis and increases invasion of trophoblast cells | (70) |
| miR-3615-3p | chr17: 74,748,663-74,748,683 | Upregulated in plasma exosomes from PTB | TGF-β signaling | Possibly involved in trophoblast proliferation | (71) |
| miR-4732-3p | chr17: 28,861,668-28,861,688 | Dysregulated in serum from PEDownregulated in GDM | | Possibly involved in cellular development and cellular movement | (72, 73) |
## Discussion
The definition of PAS has been constantly revised in the last century because of the heterogeneous histological and clinical characteristics of deliveries complicated by placental retention. The currently recommended terminology includes different degrees of abnormal placentation from abnormally adherent villi towards extended EVT invasion into the uterine wall and beyond into adjacent organs [1]. The incidence of this life-threatening disease is increasing rapidly, affecting at least 1 in 817 pregnancies worldwide and approximately 1 in 500 pregnancies in developed countries. A recent report indicates an incidence as high as 1 in 272 pregnancies [10].
In contrast to placenta accreta, the etiology of severe PAS (placenta increta/percreta) remains largely well defined since in nearly all cases a surgical damage of the decidua preceded. Currently, the most accepted hypothesis is a combination of scarred endometrium caused by damage prior to gestation and the subsequent abnormal invasion of trophoblast cells [74]. Consequently, the most common risk factor for PAS is a history of cesarean deliveries and/or previous uterine surgeries (myomectomy, operative hysteroscopic procedures, dilation, and curettage, etc.), followed by assisted reproductive technology, especially in vitro fertilization and embryo transfer (IVF-ET) and advanced maternal age [75, 76]. In our hands, the number of previous pregnancies was similar among the control and PAS groups, but all women with placenta increta or percreta presented previous uterine surgery. In contrast, several women included in the PAS group that suffered from adherent placenta accreta were primigravida, supporting the role of additional factors other than previous cesarean sections play a role in placenta accreta etiology.
Advanced clinical examinations, including ultrasound and in some cases also magnetic resonance imaging, may allow the diagnosis of severe PAS (placenta increta and percreta) with high sensitivity (88-$97\%$) when used by skilled personnel [6]. However, antenatal identification of adherent placenta accreta is limited and is reported as low as in only $33\%$ of the cases [5]. Making the diagnosis can be more challenging when patients are not considered at risk because they have no placenta previa or no history of previous uterine surgery [77]. Likewise, these measures are insufficient to reliably predicting the exact extent of trophoblast invasion [78, 79]. Consecutively, the final decision on the optimal method to deliver the placenta depends on the knowledge of the degree of placental invasion, that often only can be decided intraoperatively. Therefore, several authors attempted to identify biomarkers, including placental proteins (e.g., PAPP-A, AFP), hormones (e.g., hCG and human placenta lactogen), and, more recently, cell-free fetal DNA and cell-free placental mRNA, that could improve the accuracy of antenatal diagnosis of PAS [revised in [80, 81]]. Although these factors may be altered in PAS, there is a significant overlap with their concentration in unaffected pregnancies, which limits their applicability. Combining these with other markers such as miRNAs may potentially improve the diagnosis and clinical management of PAS. A deeper knowledge about the clinical behavior of PAS trophoblast cells may offer better surgical treatment or preventative procedures.
In the last years, miRNAs have been widely accepted as critical players in placental development. miRNA dysregulation is found in pregnancy complications such as preeclampsia (PE), early pregnancy loss, and fetal growth retardation (FGR) (82–84), but very few studies have sought to identify the miRNA expression profile in PAS. Here, we found that the placental expression of miRNAs differs between the adherent PAS and control groups. Although the statistical tool DESeq2 was initially designed to identify differential expression of mRNA and not miRNA genes, its basic model and normalization assumptions hold true for the investigated RNA-Seq datasets, e.g., most genes are not differentially expressed and there is a balance of over- and under-expression [28] (see Figure 1D). An independent study showed that DESeq2 could maintain a reasonable false-positive rate without a significant loss of power, even when executed on a dataset with a relatively low number of highly expressed genes, which is the case for most sRNA-Seq datasets [85]. Using this strategy, a group of DEmiRNAs was identified which includes some members of the placenta-associated miRNA clusters C19MC, C14MC, miR-106a, miR-106b, and miR-17-92. These miRNA clusters regulate trophoblast functions, cell-cell communication, and are involved in viral infection responses and placental homeostasis (17, 32–34). In our hands, validation PCR for specific miRNAs carried out in a larger cohort of placenta accreta samples confirmed the differential expression of 13 miRNAs (8 upregulated and 5 downregulated), previously not reported concerning a role in PAS development. These miRNAs were localized by in situ hybridization revealing that they are expressed mainly by trophoblast cells and, in some cases, overexpressed by the invasive EVTs, especially observed in PAS placentas strengthening their role as regulators of trophoblast function. Remarkably, for ten of these miRNAs (miR-331-3p, -193b-3p, -376c-3p, -3074-5p, -222-3p, -519d-3p, -106b-3p, -3615, -16-2-3p und -454-5p, see Table 2), in vitro studies have already been carried out in trophoblastic cell lines, and they are reported to control trophoblast invasion and migration. Some of these miRNAs are also altered in PE, FGR, or other pregnancy disorders, suggesting their central role in placental functions and potentially a common alteration of trophoblast regulation in distinct pathologies. However, it cannot be ruled out that these miRNAs regulate different mRNAs to promote PAS development since there are certain redundancies and compensation effects among miRNAs [86]. Likewise, the presence of DEmiRNAs considered of trophoblast origin (e.g., miR-519d-3p) in other cell types such as endothelial cells could indicate intercellular transfer from trophoblast to other placental cell types, which may cause alteration in the function of recipient cells. In the context of PAS, the relevance of this mechanism has not been addressed yet and could contribute to clarifying its etiology.
The majority of here identified miRNAs have been already tested in pregnancy-related pathologies at the placental level (Table 2), but researchers are now seeking to determine whether these miRNAs may serve as serum biomarkers. For instance, plasma miR-139-3p, miR-196a-5p, miR-518a-3p, and miR-671-3p were found downregulated in serum of patients diagnosed with placenta increta or percreta compared to healthy pregnancies [87]. In our hands, miR-139-3p and miR-671-3p were also found downregulated in placenta accreta compared to NP placentas, which may support their use as biomarkers. However, miR-518a-3p appeared upregulated in our study. Likewise, the assessment of the secretory form of clusterin combined with the expression of either miR-21-5p, miR-92a-3p or miR-320a-3p in plasma of pregnant women were reported as potential predictors for the development of different forms of PAS with high specificity and sensibility [88]. While no changes in miR-21-5p or miR-320a-3p were identified in our study, miR-92a-3p was found downregulated and not upregulated as suggested by that publication. Likewise, while our results showed increased miR-382-3p and decreased miR-423-5p expression in placenta tissue from adherent PAS, their serum levels appeared unchanged in the aforementioned studies [87, 88]. Other miRNAs such as miR-24 and miR-519d were found here upregulated in PAS tissue and their plasma levels were found upregulated in other pathologies such as preeclampsia [89]. Considering that placental miRNA expression changes with the gestation age [90] and the samples included in this study were taken after delivery, the low correlation with the reported alterations in plasma may be due to the differences in the gestational age at sampling. Therefore, although having the potential, more comprehensive studies are needed to determine whether the DEmiRNAs reported in this study may in fact serve as early diagnostic markers for PAS.
To further examine the biological relevance of DEmiRNAs in PAS, an in silico analysis was carried out to assign functional meaning for regulation at the mRNA level. To improve the interpretation of biological phenomena related to the extensive list of enriched KEGG pathways, analysis of local regions or sub-pathways has been achieved following a similar strategy as that published by others [91] and in our previous study [31]. In the context of PAS, these bioinformatic strategies allowed the identification of biological pathways involved in angiogenesis, embryonic development, cell migration and adhesion, and tumor-related pathways that are deregulated in serum of PAS patients [87]. Using a similar strategy, a network of lncRNAs, miRNAs and mRNAs implicated in reduced angiogenesis has been reported in PAS placentas [24]. Here, seven sub-pathways, including cell cycle control, actin regulation, TGF-β, MAPK, PI3K-AKT, NF-κB, and the JAK-STAT signaling, were consistently mapped as targets of DEmiRNAs in the enriched KEGG pathways, which highlights them as the major cascades affected in PAS pathophysiology. Some molecules within these pathways have been previously reported as affected in PAS. For instance, an investigation by us reported increased mRNA and protein expression of cell cycle mediators (p21, p16, and CyclinD1) in PAS placentas compared to NP. However, this effect was reported to be significant only when delivery occurred after week 34, suggesting an additional temporary regulation [92]. Other factors previously reported in PAS include TGF-β, which regulates cellular growth, motility, tumorigenesis, and trophoblastic EMT. It suppresses trophoblast invasion by regulating the transcription factors zinc finger protein SNAI (SNAIL) and Twist family basic helix-loop-helix transcription factor (TWIST) (93–95). In addition, silencing TGF-β type 1 receptor (TGFBR1) expression in trophoblastic cells significantly enhanced their trophoblastic invasiveness related to EMT promotion. Congruently, TGF-β negatively regulates trophoblast invasion by upregulating miR-7 in a SMAD2-dependent manner supporting the repression of EMT [96]. Consistent with our findings, a significant decrease in relative TGF-β1 mRNA expression in tissue of PAS versus NP placenta has been reported [97]. TGFB genes have been proposed in literature as targets of three DEmiRNAs identified in this study (miR-24-3p, miR-193b-3p and miR-3615-3p) [42] [71]. Because miR-24-3p and miR-193b-3p were upregulated while miR-3615-3p was downregulated in PAS samples, it is unfeasible to estimate the contribution of each of these miRNAs to the overall decrease in TGF-β1 mRNA expression nor to the development of PAS. Therefore, these observations reinforce the need to consider larger miRNA:mRNA networks as causative of PAS rather than the association of a single miRNA and its targets reported in vitro.
Abnormal expression of other genes identified in sub-pathway analyses has been reported in pregnancy malignancies. For instance, dysregulated PTEN expression in blastocyst implantation, spontaneous abortion, and PE has been reported suggesting its critical role during pregnancy (98–101). Although NF-κB signaling is mainly involved in regulating inflammatory factors, there is evidence that it negatively regulates cell cycle and cell proliferation [102]. Several studies have reported associations between PTEN and NF-κB. Increased PTEN, dependent on the AP-1/NF-κB pathway, impairs human trophoblast cell invasion and is related to PE development [103]. Furthermore, PTEN has been shown to promote NF-κB activation or suppression in other cell systems [104, 105]. In our study presented here, we have found downregulated PTEN and NFKB mRNA in placenta accreta samples, while PTEN protein was downregulated and NF-κB protein was upregulated (in EVTs). Previously, we have reported that overexpression of miR-519d-3p in trophoblast cell lines is related to AKT upregulation but PTEN downregulation. We found miR-519d-3p associated with augmented trophoblast proliferation but reduced migration [56]. Here, PAS samples showed high expression of miR-519d-3p, reinforcing its link with the PTEN/AKT/NF-κB system, which constitutes a vital cell cycle signaling pathway involved in trophoblast proliferation and metabolism. PTEN has been demonstrated as a common target of numerous miRNAs, including miR-21, miR-214, and miR-217, which are involved in regulating several cancer types [106, 107]. In our study, in addition to miR-519d-3p, PTEN appears to be a potential target of miR-222-3p and miR-106b-3p and AKT of miR-382-3p and miR-495-3p.
In summary, this study provides a set of miRNAs as potential biomarkers for the diagnosis of PAS, especially for placenta accreta. Additionally, these miRNAs and their targets are associated with signaling pathways relevant for controlling trophoblast function, providing preliminary evidence for their role in the pathogenesis of PAS.
## Data availability statement
The data presented in the study are deposited in the OSF (https://osf.io/) and NIH (https://www.ncbi.nlm.nih.gov/geo/) repositories, accession numbers: https://osf.io/8wq9h 2022-12-08 and GSE216742 2022-11-01, respectively.
## Ethics statement
The studies involving human participants were reviewed and approved by the ethics committees from the Jena University Hospital and the Department of Gynecology and Obstetrics, University Hospital Essen, Germany. Experiments in the study were approved according to the Helsinki Declaration on ethical principles for medical research involving human subjects by obtaining consent forms (Amendment No: 1509-$\frac{03}{05}$ Jena and 12-5212-BO Essen). The patients/participants provided their written informed consent to participate in this study.
## Author contributions
Conceptualization, DM-P and UM. Formal analysis, JM-C, EB, RG-S, and DM-P. Funding acquisition, UM and DM-P. Investigation, JM-C, EB, PF-Z and RG-S. Project administration, DM-P. Supervision, MM, UM and DM-P. Resources, TG, AG, AK und MM. Visualization, JM-C, EB, P-FZ, RG-S and DM-P. Writing—original draft. JM-C and DM-P. Writing—review and editing, JM-C, EB, TG, AG, AK, MM, UM and DM-P. All authors contributed to the article and approved the submitted version.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fendo.2023.1021640/full#supplementary-material
## References
1. Hecht JL, Baergen R, Ernst LM, Katzman PJ, Jacques SM, Jauniaux E. **Classification and reporting guidelines for the pathology diagnosis of placenta accreta spectrum (PAS) disorders: recommendations from an expert panel**. *Mod Pathol* (2020) **33**. DOI: 10.1038/s41379-020-0569-1
2. Jauniaux E, Ayres-de-Campos D, Langhoff-Roos J, Fox KA, Collins S. **FIGO classification for the clinical diagnosis of placenta accreta spectrum disorders**. *Int J Gynaecol Obstet* (2019) **146**. DOI: 10.1002/ijgo.12761
3. Jauniaux E, Collins S, Burton GJ. **Placenta accreta spectrum: pathophysiology and evidence-based anatomy for prenatal ultrasound imaging**. *Am J Obstet Gynecol* (2018) **218** 75-87. DOI: 10.1016/j.ajog.2017.05.067
4. Jauniaux E, Chantraine F, Silver RM, Langhoff-Roos J. **FIGO consensus guidelines on placenta accreta spectrum disorders: Epidemiology**. *Int J Gynaecol Obstet* (2018) **140**. DOI: 10.1002/ijgo.12407
5. Bluth A, Schindelhauer A, Nitzsche K, Wimberger P, Birdir C. **Placenta accreta spectrum disorders-experience of management in a German tertiary perinatal centre**. *Arch Gynecol Obstet* (2021) **303**. DOI: 10.1007/s00404-020-05875-x
6. Cali G, Giambanco L, Puccio G, Forlani F. **Morbidly adherent placenta: evaluation of ultrasound diagnostic criteria and differentiation of placenta accreta from percreta**. *Ultrasound Obstet Gynecol* (2013) **41**. DOI: 10.1002/uog.12385
7. Bailit JL, Grobman WA, Rice MM, Reddy UM, Wapner RJ, Varner MW. **Morbidly adherent placenta treatments and outcomes**. *Obstet Gynecol* (2015) **125**. DOI: 10.1097/AOG.0000000000000680
8. Matsuzaki S, Mandelbaum RS, Sangara RN, McCarthy LE, Vestal NL, Klar M. **Trends, characteristics, and outcomes of placenta accreta spectrum: a national study in the united states**. *Am J Obstet Gynecol* (2021) **225** 534 e1-534 e38. DOI: 10.1016/j.ajog.2021.04.233
9. Jauniaux E, Burton GJ. **Pathophysiology of placenta accreta spectrum disorders: A review of current findings**. *Clin Obstet Gynecol* (2018) **61**. DOI: 10.1097/GRF.0000000000000392
10. **Obstetric care consensus no. 7: Placenta accreta spectrum**. *Obstet Gynecol* (2018) **132**. DOI: 10.1097/AOG.0000000000002983
11. Jauniaux E, Bunce C, Grønbeck L, Langhoff-Roos J. **Prevalence and main outcomes of placenta accreta spectrum: a systematic review and meta-analysis**. *Am J Obstet Gynecol* (2019) **221**. DOI: 10.1016/j.ajog.2019.01.233
12. Wu S, Kocherginsky M, Hibbard JU. **Abnormal placentation: twenty-year analysis**. *Am J Obstet Gynecol* (2005) **192**. DOI: 10.1016/j.ajog.2004.12.074
13. Chuong EB, Hannibal RL, Green SL, Baker JC. **Evolutionary perspectives into placental biology and disease**. *Appl Transl Genom* (2013) **2**. DOI: 10.1016/j.atg.2013.07.001
14. Zhou J, Chen H, Xu X, Liu Y, Chen S, Yang S. **Uterine damage induces placenta accreta and immune imbalance at the maternal-fetal interface in the mouse**. *Placenta* (2022) **119** 8-16. DOI: 10.1016/j.placenta.2022.01.002
15. Li N, Hou R, Liu C, Yang T, Qiao C, Wei J. **Integration of transcriptome and proteome profiles in placenta accreta reveals trophoblast over-migration as the underlying pathogenesis**. *Clin Proteomics* (2021) **18** 31. DOI: 10.1186/s12014-021-09336-8
16. Liang Y, Ridzon D, Wong L, Chen C. **Characterization of microRNA expression profiles in normal human tissues**. *BMC Genomics* (2007) **8** 166. DOI: 10.1186/1471-2164-8-166
17. Morales-Prieto DM, Ospina-Prieto S, Chaiwangyen W, Schoenleben M, Markert UR. **Pregnancy-associated miRNA-clusters**. *J Reprod Immunol* (2013) **97** 51-61. DOI: 10.1016/j.jri.2012.11.001
18. Mayor-Lynn K, Toloubeydokhti T, Cruz AC, Chegini N. **Expression profile of microRNAs and mRNAs in human placentas from pregnancies complicated by preeclampsia and preterm labor**. *Reprod Sci* (2011) **18** 46-56. DOI: 10.1177/1933719110374115
19. Morales-Prieto DM, Chaiwangyen W, Ospina-Prieto S, Schneider U, Herrmann J, Gruhn B. **MicroRNA expression profiles of trophoblastic cells**. *Placenta* (2012) **33**. DOI: 10.1016/j.placenta.2012.05.009
20. Gu Y, Bian Y, Xu X, Wang X, Zuo C, Meng J. **Downregulation of miR-29a/b/c in placenta accreta inhibits apoptosis of implantation site intermediate trophoblast cells by targeting MCL1**. *Placenta* (2016) **48**. DOI: 10.1016/j.placenta.2016.09.017
21. Umemura K, Ishioka S, Endo T, Ezaka Y, Takahashi M, Saito T. **Roles of microRNA-34a in the pathogenesis of placenta accreta**. *J Obstet Gynaecol Res* (2013) **39** 67-74. DOI: 10.1111/j.1447-0756.2012.01898.x
22. Gu Y, Meng J, Zuo C, Wang S, Li H, Zhao S. **Downregulation of MicroRNA-125a in placenta accreta spectrum disorders contributes antiapoptosis of implantation site intermediate trophoblasts by targeting MCL1**. *Reprod Sci* (2019) **26**. DOI: 10.1177/1933719119828040
23. Li N, Hou R, Yang T, Liu C, Wei J. **miR-193a-3p mediates placenta accreta spectrum development by targeting EFNB2 via epithelial-mesenchymal transition pathway under decidua defect conditions**. *Front Mol Biosci* (2020) **7**. DOI: 10.3389/fmolb.2020.613802
24. Yang T, Li N, Hou R, Qiao C, Liu C. **Development and validation of a four-microRNA signature for placenta accreta spectrum: an integrated competing endogenous RNA network analysis**. *Ann Transl Med* (2020) **8** 919. DOI: 10.21037/atm-20-1150
25. Martin M. **Cutadapt removes adapter sequences from high-throughput sequencing reads**. *EMBnet.journal* (2011) **17** 3. DOI: 10.14806/ej.17.1.200
26. Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. **TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions**. *Genome Biol* (2013) **14** R36. DOI: 10.1186/gb-2013-14-4-r36
27. Liao Y, Smyth GK, Shi W. **featureCounts: an efficient general purpose program for assigning sequence reads to genomic features**. *Bioinformatics* (2014) **30**. DOI: 10.1093/bioinformatics/btt656
28. Love MI, Huber W, Anders S. **Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2**. *Genome Biol* (2014) **15** 550. DOI: 10.1186/s13059-014-0550-8
29. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I. **Controlling the false discovery rate in behavior genetics research**. *Behav Brain Res* (2001) **125**. DOI: 10.1016/S0166-4328(01)00297-2
30. Huang HY, Lin YC, Li J, Huang KY, Shrestha S, Hong HC. **miRTarBase 2020: updates to the experimentally validated microRNA-target interaction database**. *Nucleic Acids Res* (2020) **48**. DOI: 10.1093/nar/gkz896
31. Morales-Prieto DM, Barth E, Murrieta-Coxca JM, Favaro RR, Gutiérrez-Samudio RN, Chaiwangyen W. **Identification of miRNAs and associated pathways regulated by leukemia inhibitory factor in trophoblastic cell lines**. *Placenta* (2019) **88**. DOI: 10.1016/j.placenta.2019.09.005
32. Malnou EC, Umlauf D, Mouysset M, Cavaillé J. **Imprinted MicroRNA gene clusters in the evolution, development, and functions of mammalian placenta**. *Front Genet* (2018) **9**. DOI: 10.3389/fgene.2018.00706
33. Chen DB, Wang W. **Human placental microRNAs and preeclampsia**. *Biol Reprod* (2013) **88** 130. DOI: 10.1095/biolreprod.113.107805
34. Ouyang Y, Bayer A, Chu T, Tyurin VA, Kagan VE, Morelli AE. **Isolation of human trophoblastic extracellular vesicles and characterization of their cargo and antiviral activity**. *Placenta* (2016) **47** 86-95. DOI: 10.1016/j.placenta.2016.09.008
35. Fitzgerald JS, Poehlmann TG, Schleussner E, Markert UR. **Trophoblast invasion: the role of intracellular cytokine signalling via signal transducer and activator of transcription 3 (STAT3)**. *Hum Reprod Update* (2008) **14**. DOI: 10.1093/humupd/dmn010
36. Chaiwangyen W, Ospina-Prieto S, Photini SM, Schleussner E, Markert UR, Morales-Prieto DM. **Dissimilar microRNA-21 functions and targets in trophoblastic cell lines of different origin**. *Int J Biochem Cell Biol* (2015) **68**. DOI: 10.1016/j.biocel.2015.08.018
37. Chen CY, Chen J, He L, Stiles BL. **PTEN: Tumor suppressor and metabolic regulator**. *Front Endocrinol (Lausanne)* (2018) **9**. DOI: 10.3389/fendo.2018.00338
38. Liu T, Zhang L, Joo D, Sun SC. **NF-κB signaling in inflammation**. *Signal Transduct Target Ther* (2017) **2**. DOI: 10.1038/sigtrans.2017.23
39. Luizon MR, Conceição IMCA, Viana-Mattioli S, Caldeira-Dias M, Cavalli RC, Sandrim VC. **Circulating MicroRNAs in the second trimester from pregnant women who subsequently developed preeclampsia: Potential candidates as predictive biomarkers and pathway analysis for target genes of miR-204-5p**. *Front Physiol* (2021) **12**. DOI: 10.3389/fphys.2021.678184
40. Wu L, Zhou H, Lin H, Qi J, Zhu C, Gao Z. **Circulating microRNAs are elevated in plasma from severe preeclamptic pregnancies**. *Reproduction* (2012) **143**. DOI: 10.1530/REP-11-0304
41. Hromadnikova I, Kotlabova K, Krofta L. **A history of preterm delivery is associated with aberrant postpartal MicroRNA expression profiles in mothers with an absence of other pregnancy-related complications**. *Int J Mol Sci* (2021) **22**. DOI: 10.3390/ijms22084033
42. Zhou X, Li Q, Xu J, Zhang X, Zhang H, Xiang Y. **The aberrantly expressed miR-193b-3p contributes to preeclampsia through regulating transforming growth factor-beta signaling**. *Sci Rep* (2016) **6** 19910. DOI: 10.1038/srep19910
43. Xu P, Zhao Y, Liu M, Wang Y, Wang H, Li YX. **Variations of microRNAs in human placentas and plasma from preeclamptic pregnancy**. *Hypertension* (2014) **63**. DOI: 10.1161/HYPERTENSIONAHA.113.02647
44. Betoni JS, Derr K, Pahl MC, Rogers L, Muller CL, Packard RE. **MicroRNA analysis in placentas from patients with preeclampsia: comparison of new and published results**. *Hypertens Pregnancy* (2013) **32**. DOI: 10.3109/10641955.2013.807819
45. Ostling H, Kruse R, Helenius G, Lodefalk M. **Placental expression of microRNAs in infants born small for gestational age**. *Placenta* (2019) **81** 46-53. DOI: 10.1016/j.placenta.2019.05.001
46. Shan L, Hou X. **Circular RNA hsa_circ_0026552 inhibits the proliferation, migration and invasion of trophoblast cells via the miR3313p/TGFbetaR1 axis in preeclampsia**. *Mol Med Rep* (2021) **24** 798. DOI: 10.3892/mmr.2021.12438
47. Yi Y, Cheng JC, Klausen C, Leung PCK. **TGF-beta1 inhibits human trophoblast cell invasion by upregulating cyclooxygenase-2**. *Placenta* (2018) **68** 44-51. DOI: 10.1016/j.placenta.2018.06.313
48. Yadava SM, Feng A, Parobchak N, Wang B, Rosen T. **miR-15b-5p promotes expression of proinflammatory cytokines in human placenta by inhibiting apelin signaling pathway**. *Placenta* (2021) **104** 8-15. DOI: 10.1016/j.placenta.2020.11.002
49. Jia L, Zhou X, Huang X, Xu X, Jia Y, Wu Y. **Maternal and umbilical cord serum-derived exosomes enhance endothelial cell proliferation and migration**. *FASEB J* (2018) **32**. DOI: 10.1096/fj.201701337RR
50. Li J, Du J, Wang Z, Wang C, Bai J, Zhang S. **Expression of miR-376 in blood of pregnant women with preeclampsia and its effect on 25-hydroxyvitamin d**. *Exp Ther Med* (2018) **16**. DOI: 10.3892/etm.2018.6394
51. Yang HL, Zhang HZ, Meng FR, Han SY, Zhang M. **Differential expression of microRNA-411 and 376c is associated with hypertension in pregnancy**. *Braz J Med Biol Res* (2019) **52**. DOI: 10.1590/1414-431x20197546
52. Fu G, Ye G, Nadeem L, Ji L, Manchanda T, Wang Y. **MicroRNA-376c impairs transforming growth factor-beta and nodal signaling to promote trophoblast cell proliferation and invasion**. *Hypertension* (2013) **61**. DOI: 10.1161/HYPERTENSIONAHA.111.203489
53. Zhang H, Zhu C, He Z, Chen S, Li L, Sun C. **LncRNA PSMB8-AS1 contributes to pancreatic cancer progression via modulating miR-382-3p/STAT1/PD-L1 axis**. *J Exp Clin Cancer Res* (2020) **39** 179. DOI: 10.1186/s13046-020-01687-8
54. Liu Y, Wang Y, Fu X, Lu Z. **Long non-coding RNA NEAT1 promoted ovarian cancer cells' metastasis through regulation of miR-382-3p/ROCK1 axial**. *Cancer Sci* (2018) **109**. DOI: 10.1111/cas.13647
55. Wang J, Luo J, Liu G, Li X. **Circular RNA hsa_circ_0008285 inhibits colorectal cancer cell proliferation and migration via the miR-382-5p/PTEN axis**. *Biochem Biophys Res Commun* (2020) **527**. DOI: 10.1016/j.bbrc.2020.03.165
56. Chaiwangyen W, Murrieta-Coxca JM, Favaro RR, Photini SM, Gutiérrez-Samudio RN, Schleussner E. **MiR-519d-3p in trophoblastic cells: Effects, targets and transfer to allogeneic immune cells via extracellular vesicles**. *Int J Mol Sci* (2020) **21**. DOI: 10.3390/ijms21103458
57. Xie L, Mouillet JF, Chu T, Parks WT, Sadovsky E, Knöfler M. **C19MC microRNAs regulate the migration of human trophoblasts**. *Endocrinology* (2014) **155**. DOI: 10.1210/en.2014-1501
58. Ding J, Huang F, Wu G, Han T, Xu F, Weng D. **MiR-519d-3p suppresses invasion and migration of trophoblast cells via targeting MMP-2**. *PloS One* (2015) **10**. DOI: 10.1371/journal.pone.0120321
59. Gu Y, Shi Y, Yang Q, Gu WW, He YP, Zheng HJ. **miR-3074-5p promotes the apoptosis but inhibits the invasiveness of human extravillous trophoblast-derived HTR8/SVneo cells in vitro**. *Reprod Sci* (2018) **25**. DOI: 10.1177/1933719117725823
60. Gu Y, Zhang X, Yang Q, Wang J, He Y, Sun Z. **Aberrant placental villus expression of miR-486-3p and miR-3074-5p in recurrent miscarriage patients and uterine expression of these MicroRNAs during early pregnancy in mice**. *Gynecol Obstet Invest* (2015) **81**. DOI: 10.1159/000435879
61. Zhu Y, Lu H, Huo Z, Ma Z, Dang J, Dang W. **MicroRNA-16 inhibits feto-maternal angiogenesis and causes recurrent spontaneous abortion by targeting vascular endothelial growth factor**. *Sci Rep* (2016) **6** 35536. DOI: 10.1038/srep35536
62. Tang L, Yang M, Qin L, Li X, He G, Liu X. **Deficiency of DICER reduces the invasion ability of trophoblasts and impairs the pro-angiogenic effect of trophoblast-derived microvesicles**. *J Cell Mol Med* (2020) **24**. DOI: 10.1111/jcmm.14917
63. Li J, Wang JM, Liu YH, Zhang Z, Han N, Wang JY. **[Effect of microRNA-106b on the invasion and proliferation of trophoblasts through targeting MMP-2]**. *Zhonghua Fu Chan Ke Za Zhi* (2017) **52**. DOI: 10.3760/cma.j.issn.0529-567X.2017.05.007
64. Qu HM, Qu LP, Pan XZ, Mu LS. **Upregulated miR-222 targets BCL2L11 and promotes apoptosis of mesenchymal stem cells in preeclampsia patients in response to severe hypoxia**. *Int J Clin Exp Pathol* (2018) **11**
65. Liu T, Li W, Zhang J, Zhang Y. **MiR-222-3p inhibits trophoblast cell migration and alleviates preeclampsia in rats through inhibiting HDAC6 and Notch1 signaling**. *Reprod Sci* (2021) **29**. DOI: 10.1007/s43032-021-00793-y
66. Zhao JR, Cheng WW, Wang YX, Cai M, Wu WB, Zhang HJ. **Identification of microRNA signature in the progression of gestational trophoblastic disease**. *Cell Death Dis* (2018) **9** 94. DOI: 10.1038/s41419-017-0108-2
67. Wen H, Chen L, He J, Lin J. **MicroRNA expression profiles and networks in placentas complicated with selective intrauterine growth restriction**. *Mol Med Rep* (2017) **16**. DOI: 10.3892/mmr.2017.7462
68. Lykoudi A, Kolialexi A, Lambrou GI, Braoudaki M, Siristatidis C, Papaioanou GK. **Dysregulated placental microRNAs in early and late onset preeclampsia**. *Placenta* (2018) **61** 24-32. DOI: 10.1016/j.placenta.2017.11.005
69. Vashukova ES, Glotov AS, Baranov VS. **MicroRNAs associated with preeclampsia**. *Russian J Genet* (2020) **56** 1-16. DOI: 10.1134/S1022795419080167
70. Shi Z, She K, Li H, Yuan X, Han X, Wang Y. **MicroRNA-454 contributes to sustaining the proliferation and invasion of trophoblast cells through inhibiting Nodal/ALK7 signaling in pre-eclampsia**. *Chem Biol Interact* (2019) **298** 8-14. DOI: 10.1016/j.cbi.2018.10.012
71. Menon R, Debnath C, Lai A, Guanzon D, Bhatnagar S, Kshetrapal PK. **Circulating exosomal miRNA profile during term and preterm birth pregnancies: A longitudinal study**. *Endocrinology* (2019) **160**. DOI: 10.1210/en.2018-00836
72. Srinivasan S, Treacy R, Herrero T, Olsen R, Leonardo TR, Zhang X. **Discovery and verification of extracellular miRNA biomarkers for non-invasive prediction of pre-eclampsia in asymptomatic women**. *Cell Rep Med* (2020) **1**. DOI: 10.1016/j.xcrm.2020.100013
73. Ding R, Guo F, Zhang Y, Liu XM, Xiang YQ, Zhang C. **Integrated transcriptome sequencing analysis reveals role of miR-138-5p/ TBL1X in placenta from gestational diabetes mellitus**. *Cell Physiol Biochem* (2018) **51**. DOI: 10.1159/000495319
74. Jauniaux E, Jurkovic D, Hussein AM, Burton GJ. **New insights into the etiopathology of placenta accreta spectrum**. *Am J Obstetrics Gynecology* (2022) 227. DOI: 10.1016/j.ajog.2022.02.038
75. Eshkoli T, Weintraub AY, Sergienko R, Sheiner E. **Placenta accreta: risk factors, perinatal outcomes, and consequences for subsequent births**. *Am J Obstet Gynecol* (2013) **208**. DOI: 10.1016/j.ajog.2012.12.037
76. Baldwin HJ, Patterson JA, Nippita TA, Torvaldsen S, Ibiebele I, Simpson JM. **Antecedents of abnormally invasive placenta in primiparous women: Risk associated with gynecologic procedures**. *Obstet Gynecol* (2018) **131**. DOI: 10.1097/AOG.0000000000002434
77. Comstock CH. **Antenatal diagnosis of placenta accreta: a review**. *Ultrasound Obstet Gynecol* (2005) **26** 89-96. DOI: 10.1002/uog.1926
78. D'Antonio F, Palacios-Jaraquemada J, Timor-Trisch I, Cali G. **Placenta accreta spectrum disorders: Prenatal diagnosis still lacks clinical correlation**. *Acta Obstet Gynecol Scand* (2018) **97**. DOI: 10.1111/aogs.13374
79. Morel O, van Beekhuizen HJ, Braun T, Collins S, Pateisky P, Calda P. **Performance of antenatal imaging to predict placenta accreta spectrum degree of severity**. *Acta Obstet Gynecol Scand* (2021) **100 Suppl 1**. DOI: 10.1111/aogs.14112
80. Bartels HC, Postle JD, Downey P, Brennan DJ. **Placenta accreta spectrum: A review of pathology, molecular biology, and biomarkers**. *Dis Markers* (2018) **2018** 1507674. DOI: 10.1155/2018/1507674
81. Li J, Zhang N, Zhang Y, Hu X, Gao G, Ye Y. **Human placental lactogen mRNA in maternal plasma play a role in prenatal diagnosis of abnormally invasive placenta: yes or no**. *Gynecol Endocrinol* (2019) **35**. DOI: 10.1080/09513590.2019.1576607
82. Choi SY, Yun J, Lee OJ, Han HS, Yeo MK, Lee MA. **MicroRNA expression profiles in placenta with severe preeclampsia using a PNA-based microarray**. *Placenta* (2013) **34** 799-804. DOI: 10.1016/j.placenta.2013.06.006
83. Gunel T, Hosseini MK, Gumusoglu E, Kisakesen HI, Benian A, Aydinli K. **Expression profiling of maternal plasma and placenta microRNAs in preeclamptic pregnancies by microarray technology**. *Placenta* (2017) **52** 77-85. DOI: 10.1016/j.placenta.2017.02.019
84. Hosseini MK, Gunel T, Gumusoglu E, Benian A, Aydinli K.. **MicroRNA expression profiling in placenta and maternal plasma in early pregnancy loss**. *Mol Med Rep* (2018) **17**. DOI: 10.3892/mmr.2018.8530
85. Dillies MA, Rau A, Aubert J, Hennequet-Antier C, Jeanmougin M, Servant N. **A comprehensive evaluation of normalization methods for illumina high-throughput RNA sequencing data analysis**. *Brief Bioinform* (2013) **14**. DOI: 10.1093/bib/bbs046
86. Plotnikova O, Baranova A, Skoblov M. **Comprehensive analysis of human microRNA-mRNA interactome**. *Front Genet* (2019) **10**. DOI: 10.3389/fgene.2019.00933
87. Chen S, Pang D, Li Y, Zhou J, Liu Y, Yang S. **Serum miRNA biomarker discovery for placenta accreta spectrum**. *Placenta* (2020) **101**. DOI: 10.1016/j.placenta.2020.09.068
88. Timofeeva AV, Fedorov IS, Pirogova MM, Vasilchenko ON, Chagovets VV, Ezhova LS. **Clusterin and its potential regulatory microRNAs as a part of secretome for the diagnosis of abnormally invasive placenta: Accreta, increta, and percreta cases**. *Life (Basel)* (2021) **11**. DOI: 10.3390/life11040270
89. Li H, Ge Q, Guo L, Lu Z. **Maternal plasma miRNAs expression in preeclamptic pregnancies**. *BioMed Res Int* (2013) **2013** 970265. DOI: 10.1155/2013/970265
90. Morales Prieto DM, Markert UR. **MicroRNAs in pregnancy**. *J Reprod Immunol* (2011) **88**. DOI: 10.1016/j.jri.2011.01.004
91. Li X, Shen L, Shang X, Liu W. **Subpathway analysis based on signaling-pathway impact analysis of signaling pathway**. *PloS One* (2015) **10**. DOI: 10.1371/journal.pone.0132813
92. Duan L, Schimmelmann M, Wu Y, Reisch B, Faas M, Kimmig R. **CCN3 signaling is differently regulated in placental diseases preeclampsia and abnormally invasive placenta**. *Front Endocrinol (Lausanne)* (2020) **11**. DOI: 10.3389/fendo.2020.597549
93. Karmakar S, Das C. **Regulation of trophoblast invasion by IL-1beta and TGF-beta1**. *Am J Reprod Immunol* (2002) **48**. DOI: 10.1034/j.1600-0897.2002.01151.x
94. Haque S, Morris JC. **Transforming growth factor-β: A therapeutic target for cancer**. *Hum Vaccin Immunother* (2017) **13**. DOI: 10.1080/21645515.2017.1327107
95. Cheng JC, Chang HM, Leung PC. **Transforming growth factor-beta1 inhibits trophoblast cell invasion by inducing snail-mediated down-regulation of vascular endothelial-cadherin protein**. *J Biol Chem* (2013) **288**. DOI: 10.1074/jbc.M113.488866
96. Shih JC, Lin HH, Hsiao AC, Su YT, Tsai S, Chien CL. **Unveiling the role of microRNA-7 in linking TGF-beta-Smad-mediated epithelial-mesenchymal transition with negative regulation of trophoblast invasion**. *FASEB J* (2019) **33**. DOI: 10.1096/fj.201801898RR
97. El-Hussieny M, Mohammed EM, Zenhom NM, Refaie MM, Okasha AM, Tawab MAE. **Possible role of TGF-**. *Fetal Pediatr Pathol* (2021) **40**. DOI: 10.1080/15513815.2020.1843574
98. Makker A, Goel MM, Nigam D, Mahdi AA, Das V, Agarwal A. **Aberrant akt activation during implantation window in infertile women with intramural uterine fibroids**. *Reprod Sci* (2018) **25**. DOI: 10.1177/1933719117737844
99. Tokyol C, Aktepe F, Hüsniye Dilek F, Yilmazer M. **Comparison of placental PTEN and beta1 integrin expression in early spontaneous abortion, early and late normal pregnancy**. *Ups J Med Sci* (2008) **113**. DOI: 10.3109/2000-1967-231
100. Xiao J, Tao T, Yin Y, Zhao L, Yang L, Hu L. **miR-144 may regulate the proliferation, migration and invasion of trophoblastic cells through targeting PTEN in preeclampsia**. *BioMed Pharmacother* (2017) **94**. DOI: 10.1016/j.biopha.2017.07.130
101. Lou CX, Zhou XT, Tian QC, Xie HQ, Zhang JY. **Low expression of microRNA-21 inhibits trophoblast cell infiltration through targeting PTEN**. *Eur Rev Med Pharmacol Sci* (2018) **22**. DOI: 10.26355/eurrev_201810_16023
102. Araki K, Kawauchi K, Tanaka N. **IKK/NF-kappaB signaling pathway inhibits cell-cycle progression by a novel Rb-independent suppression system for E2F transcription factors**. *Oncogene* (2008) **27**. DOI: 10.1038/onc.2008.184
103. Xue P, Zheng M, Diao Z, Shen L, Liu M, Gong P. **miR-155* mediates suppressive effect of PTEN 3'-untranslated region on AP-1/NF-kappaB pathway in HTR-8/SVneo cells**. *Placenta* (2013) **34**. DOI: 10.1016/j.placenta.2013.04.015
104. Xie M, Fu Z, Cao J, Liu Y, Wu J, Li Q. **MicroRNA-132 and microRNA-212 mediate doxorubicin resistance by down-regulating the PTEN-AKT/NF-kappaB signaling pathway in breast cancer**. *BioMed Pharmacother* (2018) **102**. DOI: 10.1016/j.biopha.2018.03.088
105. Tian Y, Li H, Qiu T, Dai J, Zhang Y, Chen J. **Loss of PTEN induces lung fibrosis via alveolar epithelial cell senescence depending on NF-kappaB activation**. *Aging Cell* (2019) **18**. DOI: 10.1111/acel.12858
106. Yang H, Kong W, He L, Zhao JJ, O'Donnell JD, Wang J. **MicroRNA expression profiling in human ovarian cancer: miR-214 induces cell survival and cisplatin resistance by targeting PTEN**. *Cancer Res* (2008) **68**. DOI: 10.1158/0008-5472.CAN-07-2488
107. Kato M, Putta S, Wang M, Yuan H, Lanting L, Nair I. **TGF-beta activates akt kinase through a microRNA-dependent amplifying circuit targeting PTEN**. *Nat Cell Biol* (2009) **11**. DOI: 10.1038/ncb1897
|
---
title: Enhanced external counterpulsation improves dysfunction of forearm muscle caused
by radial artery occlusion
authors:
- Zhenyu Wang
- Chun Yao
- Lihan Huang
- Jianwen Liang
- Xiaocong Zhang
- Jian Shi
- Wenbin Wei
- Jing Zhou
- Yahui Zhang
- Guifu Wu
journal: Frontiers in Cardiovascular Medicine
year: 2023
pmcid: PMC10022471
doi: 10.3389/fcvm.2023.1115494
license: CC BY 4.0
---
# Enhanced external counterpulsation improves dysfunction of forearm muscle caused by radial artery occlusion
## Abstract
### Objective
This study aimed to investigate the therapeutic effect of enhanced external counterpulsation (EECP) on radial artery occlusion (RAO) through the oscillatory shear (OS) and pulsatile shear (PS) models of human umbilical vein endothelial cells (HUVECs) and RAO dog models.
### Methods
We used high-throughput sequencing data GSE92506 in GEO database to conduct time-series analysis of functional molecules on OS intervened HUVECs, and then compared the different molecules and their functions between PS and OS. Additionally, we studied the effect of EECP on the radial artery hemodynamics in Labrador dogs through multi-channel physiological monitor. Finally, we studied the therapeutic effect of EECP on RAO at the histological level through Hematoxylin–Eosin staining, Masson staining, ATPase staining and immunofluorescence in nine Labrador dogs.
### Results
With the extension of OS intervention, the cell cycle decreased, blood vessel endothelial cell proliferation and angiogenesis responses of HUVECs were down-regulated. By contrast, the inflammation and oxidative stress responses and the related pathways of anaerobic metabolism of HUVECs were up-regulated. Additionally, we found that compared with OS, PS can significantly up-regulate muscle synthesis, angiogenesis, and NO production related molecules. Meanwhile, PS can significantly down-regulate inflammation and oxidative stress related molecules. The invasive arterial pressure monitoring showed that 30Kpa EECP treatment could significantly increase the radial artery peak pressure ($$p \leq 0.030$$, $95\%$CI, 7.236–82.524). Masson staining showed that RAO significantly increased muscle interstitial fibrosis ($$p \leq 0.002$$, $95\%$CI, 0.748–2.128), and EECP treatment can reduce this change ($$p \leq 0.011$$, $95\%$CI, −1.676 to −0.296). ATPase staining showed that RAO significantly increased the area of type II muscle fibers ($$p \leq 0.004$$, $95\%$CI, 7.181–25.326), and EECP treatment could reduce this change ($$p \leq 0.001$$, $95\%$CI, −29.213 to −11.069). In addition, immunofluorescence showed that EECP increased angiogenesis in muscle tissue ($$p \leq 0.035$$, $95\%$CI, 0.024–0.528).
### Conclusion
EECP improves interstitial fibrosis and hypoxia, and increases angiogenesis of muscle tissue around radial artery induced by RAO.
## Introduction
In 1989, Campeau reported for the first time that it was conducted coronary angiography through transradial artery access (TRA) [1]. The advantages of no braking after TRA operation contribute to improved patient comfort, reduced bleeding and hematoma, shortened hospital stay and reduced medical costs [2]. In addition, TRA also reduced mortality in high-risk patient subgroups, such as those presenting with acute coronary syndromes [3, 4]. In 2018, the European Society of Cardiology and the European Association of Cardiothoracic Surgery recommended TRA as a preferred approach for coronary artery diagnosis and treatment [5].
Radial artery occlusion (RAO) is the most frequent post-procedural complication of TRA. Several studies have demonstrated significant structural changes after TRA catheterization [6, 7]. The primary mechanism of early RAO after TRA consists of acute arterial thrombosis, resulting from the combined effect of catheter-related endothelial and vessel injury, local hypercoagulable state, and decreased blood flow from compressive hemostasis [8]. Most early RAO patients are missed due to radial artery palpation. Chronic RAO will account for a large proportion of these patients. It can be caused by the proliferation of vascular smooth muscle and the progressive thickening of the intima media caused by proliferation [6, 9]. Several studies have shown that the expression of important active factors such as nitric oxide (NO) and vascular endothelial growth factor (VEGF) in vascular endothelial cells is decreased [10, 11], and the expression of inflammatory and procoagulant molecules nuclear factor-kappaB (NF-κB), Von Willebrand factor (vWF) and tissue factor (TF) is increased [8, 12, 13], which may be involved in the occurrence of RAO. Relative studies shows that RAO can be prevented effectively by adequate procedural anticoagulation [14, 15], proper transradial artery duration and magnitude of compression [16, 17], and reduction of sheath and catheter size [6, 18]. However, the increase in the number of percutaneous coronary intervention and non-standard operations still result in a large number of RAO patients. Although asymptomatic from an ischemia standpoint in the vast majority of cases, it precludes ipsilateral TRA for future procedures. If there is symptomatic hand ischemia, it may be necessary to reopen an occluded radial artery for transradial procedure [19, 20]. This will undoubtedly increase the related vascular complications again.
Oscillatory shear stress (OS) will be generated at the stenotic artery, which will lead to vascular dysfunction and atherosclerosis, and may further aggravate the degree of arterial stenosis [21]. Conversely, physiologically high shear stress is protective, which can improve vascular endothelial function and reduce atherosclerosis. Among them, endothelial cells are critical sensors of shear stress [22]. Enhanced external counterpulsation (EECP)is a non-invasive pneumatic technology. It can not only effectively increase diastolic blood pressure, mean coronary artery pressure and coronary artery flow, but also reduce main artery systolic blood pressure by controlling a series of lower limb cuffs to inflate in the diastole and deflate in the systole, generate characteristic double pulse blood flow. Our previous animal studies had demonstrated that EECP intervention could inhibit intimal hyperplasia, restore vascular endothelial function by increasing pulsatile shear (PS) stress and modifying shear stress responsive gene expression, attenuate atherosclerosis progression through modulation of proinflammatory pathway, as well as promote coronary collaterals and angiogenesis (13, 23–25).
As far as we know, there is no noninvasive treatment that can significantly improve RAO. In this study, we analyzed the transcriptome of human umbilical vein endothelial cells (HUVECs) from the longitudinal time series and horizontal comparison through the OS and PS model in vitro. Furthermore, we further studied the evidence that EECP can improve RAO at invasive hemodynamic and histological levels through animal experiments.
## High throughput sequencing data archives
The expression profiles by an array of GSE92506 were retrieved from GEO database. To profile shear stress-regulated endothelial transcriptomes, researchers performed RNA-seq with HUVECs subjected to different shear flow conditions, including atheroprotective PS (12 ± 4 dyn/cm2) and atheroprone OS (0.5 ± 4 dyn/cm2), or kept as static control for four time periods (1, 4, 12 and 24 h). The Platform of high throughput sequencing is GPL15433 (Illumina HiSeq 1000, Homo sapiens). Series matrix files and data table header descriptions of GSE92506 were downloaded from the GEO database to describe the change trend of differential molecules through transcriptome time-series analysis of OS intervention HUVECs. In addition, we have compared the differential molecules between PS and OS at 24 h.
## Data transformations and differential molecular screening
We used the R package “DEseq2” (v3.6.3) to analyze the differentially expressed genes (DEG) between OS and ST at different time points [26], and found that the genes with p.adj < 0.05 and Log2 (Flod Change) > 1 or < −1 had significant differences. R package “Mfuzzy” (v2.20.0) was used to cluster the standardized high-throughput sequencing data of DEG (OS vs. Static) with fuzzy C-means to describe their changing trend over time. The minimum centroid distance of a series of clusters was calculated and classified into different clusters. In addition, we also analyzed the differences between OS and ST at 24 h, DEG was presented by volcano map.
## Pathway enrichment and dynamic analysis
We used Gene Ontology/Kyoto Encyclopedia of Genes and Genomes (GO/KEGG) [27]. The pathway with p.adj < 0.05 is considered to be significantly enriched. The direction of the pathway was calculated using the average value of the difference multiple of the important molecules (p.adj < 0.05) in the significantly enriched pathway. Compared with baseline, all significant mRNA were used as the background for pathway dynamic analysis [28].
## Animals and groups
A total of 9 female Labrador dogs (24 weeks old, average weight 19 ± 0.4 kg) were purchased from Jinan Jinfeng Experimental Animal Co., Ltd. and raised in Shenzhen Leading Medical Service Co., Ltd. The rearing environment: temperature (22–28°C), humidity (50–$70\%$), density (2 dogs/cage), free drinking water, feeding (2 times/day), lighting time (07:30–19:30). The animals were randomly divided into three groups (each group $$n = 3$$): Control, RAO and RAO-EECP. The animal experiment part of this study has been approved by the Medical Research Ethics Committee of the Eighth Affiliated Hospital of Sun Yat-sen University (Futian, Shenzhen) (Research Ethics of the Eighth Affiliated Hospital of Sun Yat-sen University 2021-037-01).
## Radial artery hemodynamic measurements
Hemodynamic parameters of the left radial artery during EECP intervention were measured in 3 dogs to determine the best counterpulsation pressure. Intramuscular injection of 0.5–1 mg/kg (Jilin Fanggong, China) of xylazine hydrochloride injection was used for induction anesthesia, followed by endotracheal intubation for ventilation (Mindray veta5, China). After that, isoflurane (Baxter Healthcare, United States) was used for continuous anesthesia to maintain SO2 at $100\%$. The animal was then placed in supine position on the counter pulsation bed, and a set of modified dog-specific cuff was wrapped to closely fit the lower limbs and buttocks of the dog for EECP (PSK P-ECP/TM Oxygen Saturation Monitor, China). The cuff was inflated with compressed air from the distal end to the proximal end successively in early diastole, and deflated rapidly before systole of the next cardiac cycle. The invasive arterial pressure of the left radial artery was monitored with a multi-channel physiological monitor (BIOPAC, MP150) to evaluate the appropriate counterpulsation pressure for EECP treatment.
Multi-channel physiological monitor shows the arterial flow and pressure in multiple parts of the body, including invasive radial artery pressure, under the intervention of different counterpulsation pressures (Figure 3A). We found that EECP can produce characteristic double pulse blood flow, and the radial artery peak pressure is the highest under the intervention of 30Kpa counterpulsation pressure ($$p \leq 0.030$$, $95\%$CI, 7.236 to 82.524). We took 30Kpa as the subsequent counterpulsation pressure parameter (Figure 3B).
**Figure 3:** *Multi-channel hydrodynamic measurements under EECP intervention. (A) Multi-channel hydrodynamic measurements of different counterpulsation pressures. The eighth channel (red rectangle) is the radial artery pressure. (B) Peak pressure of radial artery under different counterpulsation pressure. Data were expressed as means ± SD (n = 3). Results were evaluated by Student’s t-test (each counterpulsation pressure with baseline).*
## RAO model and EECP treatment
The left radial artery of 6 dogs was ligated, and the RAO-EECP group was performed 30Kpa EECP for 60 min every day within 14 days. In order to eliminate possible circadian influence, EECP were conducted at the same time of every day.
## Tissue preparation
All dogs were anesthetized with xylazine hydrochloride injection (Jilin Fanggong, China), at the same time, all animals were euthanized by intravenous injection of excessive $10\%$ potassium chloride at the same time as well. At the end of the experiment, the lateral muscle samples of left forearm of all dogs were taken immediately and washed in cold normal saline. The sample of each dog was divided into four parts, which were fixed with $4\%$ paraformaldehyde and then embedded in paraffin for HE staining and collagen staining, embedded with OCT embedding agent (SAKURA, Japan) and frozen in liquid nitrogen for ATPase staining [29]. In addition, another $4\%$ paraformaldehyde samples were used for immunofluorescence.
## Morphological evaluation
The detailed experimental protocol of Hematoxylin–Eosin staining, Masson staining, ATPase staining and immunofluorescence are showed in the Supplemental material.
In accordance with the predetermined research protocol (Figure 4A), we completed the establishment of the Labrador dog’s RAO model, the EECP treatment, the collection of skeletal muscle samples and related tests. The pathological staining results showed that (Figure 4B), there was no significant difference in the Hematoxylin eosin staining of the lateral muscle of left forearm among the three groups. However, Masson staining showed that the interstitial fibrosis of lateral muscle of left forearm caused by RAO ($$p \leq 0.002$$, $95\%$CI, 0.748–2.128), while after the treatment of EECP, the interstitial fibrosis was significantly reduced ($$p \leq 0.011$$, $95\%$CI, −1.676 to −0.296, Figure 4C). ATPase staining showed that type II muscle fibers in RAO group were significantly increased compared with control group ($$p \leq 0.004$$, $95\%$CI, 7.181–25.326), while type II muscle fibers were significantly decreased after treatment with EECP ($$p \leq 0.001$$, $95\%$CI, −29.213 to −11.069, Figure 4D). Immunofluorescence showed that the expression of VWF was significantly up-regulated in lateral muscle of left forearm after treatment with EECP ($$p \leq 0.035$$, $95\%$CI, 0.024–0.528, Figure 4E), which proved that EECP increased angiogenesis.
**Figure 4:** *EECP promotes angiogenesis, improves muscle damage and aerobic metabolism. (A) Animal study design. (B) From left to right: H&E staining, Masson staining, ATPase staining and VWF immunofluorescence (IF) of cross section of lateral muscle of left forearm of Labrador dogs in three groups. Scale bars, 50 μ m. There was no significant difference in H&E staining among the three groups. VWF, Von Willebrand Factor. (C) Quantification of fibrotic regions (appearing as blue in Masson’s trichome in C). Data were expressed as means ± SD (n = 3 per group). Results were evaluated by one-way ANOVA followed by Tukey HSD test. (D) Quantification of type II muscle fibers (appearing as dark grey in B). Data were expressed as means ± SD (n = 3 per group). Results were evaluated by one-way ANOVA followed by Tukey HSD test. (E) Quantification of VWF immunofluorescence. Data were expressed as means ± SD (n = 3 per group). Results were evaluated by one-way ANOVA followed by Tukey HSD test.*
## Statistical analysis
We conducted Shapiro Wilk normality test for all quantitative data. For normal distribution samples, Student’s t-tests and ANOVA were used to compare parameter data between two conditions and multiple conditions, respectively. The homogeneity of variance was evaluated by Levene’s test. For ANOVA analysis, multiple hypothesis test (Tukey HSD test) was used and corrected by Bonferroni method. In addition, we used the R package “ggplot2” (v3.3.3) and GraphPad Prism (v9.1.1) to visualize the above statistical analysis data.
## Time-series mRNA expression and pathway dynamic enrichment analysis
The important molecules (p.adj < 0.05) at each time point compared with static under OS intervention were divided into four main longitudinal trajectories by C-means clustering (Figure 1A) according to their expression trend. We can find that the expression of molecules in Cluster 1, 2 and 3 reached the peak at 1 h, and then the expression of molecules in Cluster 1 decreased rapidly at 4 h, and then slowly recovered, while the expression of molecules in Cluster 2 and 3 decreased slowly after 1 h until 24 h. On the contrary, the molecular expression in Cluster 4 was gradually up-regulated until it reached peak at 12 h.
**Figure 1:** *mRNA expression and enrichment of HUVECs under OS intervention. (A) Clustering of longitudinal gene expression trajectories (p.adj < 0.05). Membership represents the association strength between the sample and the cluster. (B) Pathway enrichment analysis using mRNA significantly changing in tissue synthesis, cell cycle, inflammation and oxidative stress and metabolism (p.adj < 0.05, |Log2 Fold change|>1). Pathway direction is the mean log2 fold change relative to baseline of significant transcripts in each pathway (blue, downregulated; red, upregulated). The dot size represents pathway significance.*
At each time point, we use all important mRNA (p.adj < 0.05) to conduct enrichment analysis on dynamic pathways related to tissue synthesis, cell cycle, inflation and oxidative stress and metabolism (Figure 1B). The results showed that with the extension of OS intervention, the cell cycle decreased, blood vessel endothelial cell proliferation and angiogenesis responses were down-regulated. By contrast, the inflammation and oxidative stress responses and the related pathways of anaerobic metabolism of HUVECs were up-regulated. The details of time-series mRNA expression and pathway dynamic enrichment analysis were in *Supplementary data* sheet.
## Horizontal comparative analysis
We performed a differential analysis on mRNA of HUVECs after 24 h PS and OS intervention (Figure 2). We found that compared with OS, PS can significantly up-regulate muscle synthesis related molecules (such as GLI1, TGM2, SULF1, etc.) ( 30–35), angiogenesis related molecules (such as CD34, CYP1B1, AQP1, ECM1, GPER1, HMOX1, RAMP2, etc.) ( 36–48), and NO production related molecules (such as ASS1, GCH1, NOS3, KLF4, KLF2, etc.) ( 49–56). In addition, PS can significantly down-regulate inflammation and oxidative stress related molecules (such as IL7, INHBA, CCL7, CXCL12, TGFBR1, CXCR4, IL1RL1, TNFSF15, ITGA4, etc.) ( 57–72).
**Figure 2:** *Horizontal comparative analysis of PS and OS in HUVECs mRNA. Volcano plots show the significant differences between PS and OS in HUVECs mRNA at 24 h. Each dot represents a mRNA (blue, downregulated; red, upregulated; gray, no significant). In HUVECs mRNA, the molecules with significant difference were defined as p.adj < 0.05 and |Log2 Fold change|>1.*
## Discussion
The real-world reported incidence of RAO remains the most frequent postprocedural complication of transradial access, with limited choice in the uptake of RAO therapeutic strategies. The reduction of sheath and catheter size [18], use of intraprocedural heparin [14], and maintenance of radial artery patency during hemostasis and oral anticoagulation [16] after TRA have been shown to lower the risk of RAO and have been termed best practices. Recently, The PROPHET-II Randomized Trial showed that prophylactic ipsilateral ulnar artery compression during radial artery hemostasis could significantly reduce the risk of RAO at the time of 24 h and 30 days after the procedure [73]. However, there is no dedicated devices capable of dual compression on the market at present, which poses a great challenge for the wide application. In addition, recanalization of the occluded radial artery via the distal radial access (DRA) was reproted to be safe and effective, but the available evidence remains currently limited [74].
OS will be produced at the artery stenosis site, which further will lead to vascular dysfunction and atherosclerosis [21]. In this study, we used the OS and PS models of HUVECs to analyze the longitudinal time-series of HUVECs mRNA. We finally found that with the prolongation of OS intervention, cell cycle, and vascular endothelial cell proliferation and angiogenesis response were down-regulated. At the same time, the inflammatory and oxidative stress responses and the related pathways of anaerobic metabolism of HUVEC were up-regulated. In addition, we found that PS intervention can significantly up-regulate the muscle synthesis, angiogenesis and NO production related molecules of HUVECs relative to OS, and significantly down-regulate the Inflammation and oxidative stress related molecules. This phenomenon may explain why arterial stenosis is aggravated, and the importance of PS for maintaining vascular function.
Our previous studies showed that EECP intervention can produce characteristic double pulse blood flow, and endothelial protective laminar flow to coronary artery, so as to restore vascular endothelial function by increasing pulsating shear stress and changing the expression of shear stress response gene, slow down the progress of atherosclerosis by regulating the proinflammatory pathway, and promote coronary collateral and angiogenesis (13, 23–25). However, there is no relevant research to prove whether EECP can produce similar blood flow pattern and function to coronary artery for radial artery.
In this study, we first evaluated the hemodynamic changes of the radial artery under the intervention of EECP through invasive radial artery pressure monitoring. We also found that the most appropriate counterpulsation pressure was 30Kpa. According to Hagen Poiseuille and Navier Stokes formulas [75], blood viscosity and the diameter of the vascular cavity remain unchanged, and the pressure gradient is proportional to the shear stress. We prove that EECP can generate double pulse blood flow and significantly increase the blood flow shear stress.
After that, we artificially made the local OS of the radial artery through the RAO model of Labrador dogs, and observed the histological changes of the muscle around the radial artery. At the same time, we used the EECP protocol (30Kpa, 1 h/day) to treat other RAO dogs for 14 days. The results showed that H&E staining had no serious muscle damage in the three groups, but Masson staining showed that RAO might cause interstitial fibrosis in local muscles, indicating that RAO may cause certain inflammatory reaction to surrounding muscle tissue, and muscle interstitial fibrosis was significantly reduced after treatment with EECP. In addition, ATPase staining demonstrated that RAO may cause a significant increase of type II muscle fibers in local muscle tissue, while type II muscle fibers are significantly reduced and type I muscle fibers are significantly increased after treatment with EECP. Researches showed that type II muscle fibers are mainly responsible for anaerobic metabolism, while type I muscle fibers are mainly responsible for aerobic metabolism [76]. Therefore, we judged that RAO may cause insufficient local perfusion, leading to ischemia and hypoxia of muscle tissue, and EECP may improve this pathological state. Finally, we evaluated angiogenesis through the vascular marker VWF. The IF results showed that EECP may significantly increase angiogenesis in the local muscle tissue of the radial artery.
## Limitations and future research
Our research also has some limitations. There are some differences between the potential causes of RAO in clinical patients and the RAO animal model in this study, including the atherosclerosis, vascular access, artistic cather, exposure to wires, etc. Therefore, the results of this study cannot fully reflect the clinical efficacy of EECP on RAO. In addition, we have only verified in animal tests that the pulsating shear stress generated by EECP, and have not further performed clinical verification on patients with RAO stenosis. We will further conduct a multicenter randomized controlled trial to further verify the benefits of EECP for RAO patients.
## Conclusion
EECP can produce characteristic double pulse blood flow shear stress, and the radial artery peak pressure reaches the highest under the action of 30Kpa counterpulsation pressure. In addition, EECP may improve RAO induced interstitial fibrosis and hypoxia of muscle tissue around the radial artery, and increase angiogenesis of muscle tissue around the radial artery.
## Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary material.
## Ethics statement
The animal study was reviewed and approved by Medical Research Ethics Committee of the Eighth Affiliated Hospital of Sun Yat-sen University (Futian, Shenzhen) and affiliated to the Eighth Affiliated Hospital of Sun Yat-sen University (Futian, Shenzhen).
## Author contributions
ZW and GW proposed the scientific problems. ZW and CY designed the experiments. ZW and JZ processed bioinformatics analysis. ZW, JL, XZ, JS, and WW processed and collected the animal experimental data. ZW and CY processed and calculated. All authors contributed to the article and approved the submitted version.
## Funding
This work was supported by the National Key R&D Program of China [2020YFC2004400]; National Natural Science Foundation of China [82270477 and 81970367]; Shenzhen Key Medical Discipline Construction Fund (no. SZXK002); Shenzhen Key Clinical Discipline Funds (ZDXKJF-01002); Shenzhen Science and Technology Innovation Committee [JCYJ20160608142215491]. National Natural Science Foundation of China [grant no. 82202292]. Guangdong Medical Science and Technology Research Foundation (no. A2022383) and Guangdong Basic and Applied Basic Research Foundation (no. 2021A1515110738). Shenzhen Futian District Health System Research Foundation (FTWS2022037). Shenzhen Futian District Health System Research Foundation (FTWS2020009).
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcvm.2023.1115494/full#supplementary-material
## References
1. Campeau L. **Percutaneous radial artery approach for coronary angiography**. *Catheter Cardiovasc Diagn* (1989) **16** 3-7. DOI: 10.1002/ccd.1810160103
2. Feldman DN, Swaminathan RV, Kaltenbach LA, Baklanov DV, Kim LK, Wong SC. **Adoption of radial access and comparison of outcomes to femoral access in percutaneous coronary intervention: an updated report from the national cardiovascular data registry (2007-2012)**. *Circulation* (2013) **127** 2295-306. DOI: 10.1161/CIRCULATIONAHA.112.000536
3. Valgimigli M, Frigoli E, Leonardi S, Vranckx P, Rothenbuhler M, Tebaldi M. **Radial versus femoral access and bivalirudin versus unfractionated heparin in invasively managed patients with acute coronary syndrome (MATRIX): final 1-year results of a multicentre, randomised controlled trial**. *Lancet* (2018) **392** 835-48. DOI: 10.1016/S0140-6736(18)31714-8
4. Ferrante G, Rao SV, Juni P, Da Costa BR, Reimers B, Condorelli G. **Radial versus femoral access for coronary interventions across the entire Spectrum of patients with coronary artery disease: a meta-analysis of randomized trials**. *JACC Cardiovasc Interv* (2016) **9** 1419-34. DOI: 10.1016/j.jcin.2016.04.014
5. Neumann FJ, Sousa-Uva M, Ahlsson A, Alfonso F, Banning AP, Benedetto U. **2018 ESC/EACTS Guidelines on myocardial revascularization**. *Eur Heart J* (2019) **40** 87-165. DOI: 10.1093/eurheartj/ehy394
6. Yonetsu T, Kakuta T, Lee T, Takayama K, Kakita K, Iwamoto T. **Assessment of acute injuries and chronic intimal thickening of the radial artery after transradial coronary intervention by optical coherence tomography**. *Eur Heart J* (2010) **31** 1608-15. DOI: 10.1093/eurheartj/ehq102
7. Staniloae CS, Mody KP, Sanghvi K, Mindrescu C, Coppola JT, Antonescu CR. **Histopathologic changes of the radial artery wall secondary to transradial catheterization**. *Vasc Health Risk Manag* (2009) **5** 527-32. DOI: 10.2147/vhrm.s5468
8. Mamas MA, Fraser DG, Ratib K, Fath-Ordoubadi F, El-Omar M, Nolan J. **Minimising radial injury: prevention is better than cure**. *EuroIntervention* (2014) **10** 824-32. DOI: 10.4244/EIJV10I7A142
9. Abdelaal E, Molin P, Plourde G, Machaalany J, Bataille Y, Brousseau-Provencher C. **Successive transradial access for coronary procedures: experience of Quebec heart-lung institute**. *Am Heart J* (2013) **165** 325-31. DOI: 10.1016/j.ahj.2012.10.016
10. Heiss C, Balzer J, Hauffe T, Hamada S, Stegemann E, Koeppel T. **Vascular dysfunction of brachial artery after transradial access for coronary catheterization: impact of smoking and catheter changes**. *JACC Cardiovasc Interv* (2009) **2** 1067-73. DOI: 10.1016/j.jcin.2009.09.010
11. Hutter R, Carrick FE, Valdiviezo C, Wolinsky C, Rudge JS, Wiegand SJ. **Vascular endothelial growth factor regulates reendothelialization and neointima formation in a mouse model of arterial injury**. *Circulation* (2004) **110** 2430-5. DOI: 10.1161/01.CIR.0000145120.37891.8A
12. Mehrhof FB, Schmidt-Ullrich R, Dietz R, Scheidereit C. **Regulation of vascular smooth muscle cell proliferation: role of NF-kappaB revisited**. *Circ Res* (2005) **96** 958-64. DOI: 10.1161/01.RES.0000166924.31219.49
13. Wu G, Du Z, Hu C, Zheng Z, Zhan C, Ma H. **Angiogenic effects of long-term enhanced external counterpulsation in a dog model of myocardial infarction**. *Am J Physiol Heart Circ Physiol* (2006) **290** H248-54. DOI: 10.1152/ajpheart.01225.2004
14. Hahalis GN, Leopoulou M, Tsigkas G, Xanthopoulou I, Patsilinakos S, Patsourakos NG. **Multicenter randomized evaluation of high versus standard heparin dose on incident radial arterial occlusion after Transradial coronary angiography: the SPIRIT OF ARTEMIS study**. *JACC Cardiovasc Interv* (2018) **11** 2241-50. DOI: 10.1016/j.jcin.2018.08.009
15. Pancholy SB. **Comparison of the effect of intra-arterial versus intravenous heparin on radial artery occlusion after transradial catheterization**. *Am J Cardiol* (2009) **104** 1083-5. DOI: 10.1016/j.amjcard.2009.05.057
16. Dangoisse V, Guedes A, Chenu P, Hanet C, Albert C, Robin V. **Usefulness of a gentle and short hemostasis using the Transradial band device after Transradial access for percutaneous coronary angiography and interventions to reduce the radial artery occlusion rate (from the prospective and randomized CRASOC I, II, and III studies)**. *Am J Cardiol* (2017) **120** 374-9. DOI: 10.1016/j.amjcard.2017.04.037
17. Pancholy SB, Heck LA, Patel T. **Forearm arterial anatomy and flow characteristics: a prospective observational study**. *J Invasive Cardiol* (2015) **27** 218-21. PMID: 25840406
18. Rashid M, Kwok CS, Pancholy S, Chugh S, Kedev SA, Bernat I. **Radial artery occlusion after transradial interventions: a systematic review and meta-analysis**. *J Am Heart Assoc* (2016) **5** 1-22. DOI: 10.1161/JAHA.115.002686
19. Pancholy SB. **Transradial access in an occluded radial artery: new technique**. *J Invasive Cardiol* (2007) **19** 541-4. PMID: 18180525
20. Babunashvili A, Dundua D. **Recanalization and reuse of early occluded radial artery within 6 days after previous transradial diagnostic procedure**. *Catheter Cardiovasc Interv* (2011) **77** 530-6. DOI: 10.1002/ccd.22846
21. Barber T. **Wall shear stress and near-wall flows in the stenosed femoral artery**. *Comput Methods Biomech Biomed Engin* (2017) **20** 1048-55. DOI: 10.1080/10255842.2017.1331342
22. Souilhol C, Serbanovic-Canic J, Fragiadaki M, Chico TJ, Ridger V, Roddie H. **Endothelial responses to shear stress in atherosclerosis: a novel role for developmental genes**. *Nat Rev Cardiol* (2020) **17** 52-63. DOI: 10.1038/s41569-019-0239-5
23. Luo JY, Wu GF, Xiong Y, Chen GW, Xie Q, Yang DY. **Enhanced external counterpulsation promotes growth cytokines-mediated myocardial angiogenesis in a porcine model of hypercholesterolemia**. *Chin Med J* (2009) **122** 1188-94. DOI: 10.3760/cma.j.issn.0366-6999.2009.10.014
24. Zhang Y, He X, Chen X, Ma H, Liu D, Luo J. **Enhanced external counterpulsation inhibits intimal hyperplasia by modifying shear stress responsive gene expression in hypercholesterolemic pigs**. *Circulation* (2007) **116** 526-34. DOI: 10.1161/CIRCULATIONAHA.106.647248
25. Zhang Y, He X, Liu D, Wu G, Chen X, Ma H. **Enhanced external counterpulsation attenuates atherosclerosis progression through modulation of proinflammatory signal pathway**. *Arterioscler Thromb Vasc Biol* (2010) **30** 773-80. DOI: 10.1161/ATVBAHA.109.197806
26. Love MI, Huber W, Anders S. **Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2**. *Genome Biol* (2014) **15** 550. DOI: 10.1186/s13059-014-0550-8
27. Yu G, Wang LG, Han Y, He QY. **clusterProfiler: an R package for comparing biological themes among gene clusters**. *OMICS* (2012) **16** 284-7. DOI: 10.1089/omi.2011.0118
28. Contrepois K, Wu S, Moneghetti KJ, Hornburg D, Ahadi S, Tsai MS. **Molecular choreography of acute exercise**. *Cells* (2020) **181** 1112-1130.e16. DOI: 10.1016/j.cell.2020.04.043
29. Hintz CS, Coyle EF, Kaiser KK, Chi MM, Lowry OH. **Comparison of muscle fiber typing by quantitative enzyme assays and by myosin ATPase staining**. *J Histochem Cytochem* (1984) **32** 655-60. DOI: 10.1177/32.6.6202737
30. Yao L, Tichy ED, Zhong L, Mohanty S, Wang L, Ai E. **Gli1 defines a subset of fibro-adipogenic progenitors that promote skeletal muscle regeneration with less fat accumulation**. *J Bone Miner Res* (2021) **36** 1159-73. DOI: 10.1002/jbmr.4265
31. Pucci S, Greggi C, Polidoro C, Piro MC, Celi M, Feola M. **Clusterin silencing restores myoblasts viability and down modulates the inflammatory process in osteoporotic disease**. *J Transl Med* (2019) **17** 118. DOI: 10.1186/s12967-019-1868-5
32. Tran TH, Shi X, Zaia J, Ai X. **Heparan sulfate 6-O-endosulfatases (Sulfs) coordinate the Wnt signaling pathways to regulate myoblast fusion during skeletal muscle regeneration**. *J Biol Chem* (2012) **287** 32651-64. DOI: 10.1074/jbc.M112.353243
33. Piccioni A, Gaetani E, Neri V, Gatto I, Palladino M, Silver M. **Sonic hedgehog therapy in a mouse model of age-associated impairment of skeletal muscle regeneration**. *J Gerontol A Biol Sci Med Sci* (2014) **69** 245-52. DOI: 10.1093/gerona/glt076
34. Ribeiro S, Ribeiro C, Martins VM, Honore B, Neves-Petersen MT, Gomes AC. **Understanding myoblast differentiation pathways when cultured on Electroactive scaffolds through proteomic analysis**. *ACS Appl Mater Interfaces* (2022) **14** 26180-93. DOI: 10.1021/acsami.2c03444
35. Gill R, Hitchins L, Fletcher F, Dhoot GK. **Sulf1A and HGF regulate satellite-cell growth**. *J Cell Sci* (2010) **123** 1873-83. DOI: 10.1242/jcs.061242
36. Asahara T, Murohara T, Sullivan A, Silver M, van der Zee R, Li T. **Isolation of putative progenitor endothelial cells for angiogenesis**. *Science* (1997) **275** 964-6. DOI: 10.1126/science.275.5302.964
37. Verkman AS. **Aquaporins in endothelia**. *Kidney Int* (2006) **69** 1120-3. DOI: 10.1038/sj.ki.5000226
38. Ichikawa-Shindo Y, Sakurai T, Kamiyoshi A, Kawate H, Iinuma N, Yoshizawa T. **The GPCR modulator protein RAMP2 is essential for angiogenesis and vascular integrity**. *J Clin Invest* (2008) **118** 29-39. DOI: 10.1172/JCI33022
39. Sercu S, Lambeir AM, Steenackers E, El Ghalbzouri A, Geentjens K, Sasaki T. **ECM1 interacts with fibulin-3 and the beta 3 chain of laminin 332 through its serum albumin subdomain-like 2 domain**. *Matrix Biol* (2009) **28** 160-9. DOI: 10.1016/j.matbio.2009.02.003
40. Tang Y, Scheef EA, Gurel Z, Sorenson CM, Jefcoate CR, Sheibani N. **CYP1B1 and endothelial nitric oxide synthase combine to sustain proangiogenic functions of endothelial cells under hyperoxic stress**. *Am J Physiol Cell Physiol* (2010) **298** C665-78. DOI: 10.1152/ajpcell.00153.2009
41. Ayer A, Zarjou A, Agarwal A, Stocker R. **Heme oxygenases in cardiovascular health and disease**. *Physiol Rev* (2016) **96** 1449-508. DOI: 10.1152/physrev.00003.2016
42. Trenti A, Tedesco S, Boscaro C, Trevisi L, Bolego C, Cignarella A. **Estrogen, angiogenesis, immunity and cell metabolism: solving the puzzle**. *Int J Mol Sci* (2018) **19** 1-16. DOI: 10.3390/ijms19030859
43. Mongiat M, Fu J, Oldershaw R, Greenhalgh R, Gown AM, Iozzo RV. **Perlecan protein core interacts with extracellular matrix protein 1 (ECM1), a glycoprotein involved in bone formation and angiogenesis**. *J Biol Chem* (2003) **278** 17491-9. DOI: 10.1074/jbc.M210529200
44. Fernandez-Sauze S, Delfino C, Mabrouk K, Dussert C, Chinot O, Martin PM. **Effects of adrenomedullin on endothelial cells in the multistep process of angiogenesis: involvement of CRLR/RAMP2 and CRLR/RAMP3 receptors**. *Int J Cancer* (2004) **108** 797-804. DOI: 10.1002/ijc.11663
45. Palenski TL, Gurel Z, Sorenson CM, Hankenson KD, Sheibani N. **Cyp1B1 expression promotes angiogenesis by suppressing NF-kappaB activity**. *Am J Physiol Cell Physiol* (2013) **305** C1170-84. DOI: 10.1152/ajpcell.00139.2013
46. Trenti A, Tedesco S, Boscaro C, Ferri N, Cignarella A, Trevisi L. **The glycolytic enzyme PFKFB3 is involved in estrogen-mediated angiogenesis via GPER1**. *J Pharmacol Exp Ther* (2017) **361** 398-407. DOI: 10.1124/jpet.116.238212
47. Skora JP, Antkiewicz M, Kupczynska D, Kulikowska K, Strzelec B, Janczak D. **Local intramuscular administration of ANG1 and VEGF genes using plasmid vectors mobilizes CD34+ cells to peripheral tissues and promotes angiogenesis in an animal model**. *Biomed Pharmacother* (2021) **143** 112186. DOI: 10.1016/j.biopha.2021.112186
48. Avolio E, Katare R, Thomas AC, Caporali A, Schwenke D, Carrabba M. **Cardiac pericyte reprogramming by MEK inhibition promotes arteriologenesis and angiogenesis of the ischemic heart**. *J Clin Invest* (2022) **132** 1-17. DOI: 10.1172/JCI152308
49. Mun GI, Kim IS, Lee BH, Boo YC. **Endothelial argininosuccinate synthetase 1 regulates nitric oxide production and monocyte adhesion under static and laminar shear stress conditions**. *J Biol Chem* (2011) **286** 2536-42. DOI: 10.1074/jbc.M110.180489
50. Douglas G, Hale AB, Patel J, Chuaiphichai S, Al Haj Zen A, Rashbrook VS. **Roles for endothelial cell and macrophage Gch1 and tetrahydrobiopterin in atherosclerosis progression**. *Cardiovasc Res* (2018) **114** 1385-99. DOI: 10.1093/cvr/cvy078
51. Lopez-Ramirez MA, Lai CC, Soliman SI, Hale P, Pham A, Estrada EJ. **Astrocytes propel neurovascular dysfunction during cerebral cavernous malformation lesion formation**. *J Clin Invest* (2021) **131** 1-15. DOI: 10.1172/JCI139570
52. Oliveira-Paula GH, Lacchini R, Tanus-Santos JE. **Endothelial nitric oxide synthase: from biochemistry and gene structure to clinical implications of NOS3 polymorphisms**. *Gene* (2016) **575** 584-99. DOI: 10.1016/j.gene.2015.09.061
53. Morris G, Puri BK, Olive L, Carvalho A, Berk M, Walder K. **Endothelial dysfunction in neuroprogressive disorders-causes and suggested treatments**. *BMC Med* (2020) **18** 305. DOI: 10.1186/s12916-020-01749-w
54. Chen CL, Hsu SC, Ann DK, Yen Y, Kung HJ. **Arginine signaling and cancer metabolism**. *Cancers (Basel)* (2021) **13** 1-29. DOI: 10.3390/cancers13143541
55. Parsamanesh N, Asghari A, Sardari S, Tasbandi A, Jamialahmadi T, Xu S. **Resveratrol and endothelial function: a literature review**. *Pharmacol Res* (2021) **170** 105725. DOI: 10.1016/j.phrs.2021.105725
56. Wu Y, Ding Y, Ramprasath T, Zou MH. **Oxidative stress, GTPCH1, and endothelial nitric oxide synthase uncoupling in hypertension**. *Antioxid Redox Signal* (2021) **34** 750-64. DOI: 10.1089/ars.2020.8112
57. Esposito K, Marfella R, Ciotola M, Di Palo C, Giugliano F, Giugliano G. **Effect of a mediterranean-style diet on endothelial dysfunction and markers of vascular inflammation in the metabolic syndrome: a randomized trial**. *JAMA* (2004) **292** 1440-6. DOI: 10.1001/jama.292.12.1440
58. Zhang D, Chen G, Manwani D, Mortha A, Xu C, Faith JJ. **Neutrophil ageing is regulated by the microbiome**. *Nature* (2015) **525** 528-32. DOI: 10.1038/nature15367
59. de Lange KM, Moutsianas L, Lee JC, Lamb CA, Luo Y, Kennedy NA. **Genome-wide association study implicates immune activation of multiple integrin genes in inflammatory bowel disease**. *Nat Genet* (2017) **49** 256-61. DOI: 10.1038/ng.3760
60. He J, Song Y, Li G, Xiao P, Liu Y, Xue Y. **Fbxw7 increases CCL2/7 in CX3CR1hi macrophages to promote intestinal inflammation**. *J Clin Invest* (2019) **129** 3877-93. DOI: 10.1172/JCI123374
61. Valatas V, Kolios G, Bamias G. **TL1A (TNFSF15) and DR3 (TNFRSF25): a co-stimulatory system of cytokines with diverse functions in gut mucosal immunity**. *Front Immunol* (2019) **10** 583. DOI: 10.3389/fimmu.2019.00583
62. Yu X, Wang D, Wang X, Sun S, Zhang Y, Wang S. **CXCL12/CXCR4 promotes inflammation-driven colorectal cancer progression through activation of RhoA signaling by sponging miR-133a-3p**. *J Exp Clin Cancer Res* (2019) **38** 32. DOI: 10.1186/s13046-018-1014-x
63. Li C, Zhang S, Chen X, Ji J, Yang W, Gui T. **Farnesoid X receptor activation inhibits TGFBR1/TAK1-mediated vascular inflammation and calcification via miR-135a-5p**. *Commun Biol* (2020) **3** 327. DOI: 10.1038/s42003-020-1058-2
64. Saikumar Jayalatha AK, Hesse L, Ketelaar ME, Koppelman GH, Nawijn MC. **The central role of IL-33/IL-1RL1 pathway in asthma: from pathogenesis to intervention**. *Pharmacol Ther* (2021) **225** 107847. DOI: 10.1016/j.pharmthera.2021.107847
65. Wang Y, Li B, Zhao Y. **Inflammation in preeclampsia: genetic biomarkers, mechanisms, and therapeutic strategies**. *Front Immunol* (2022) **13** 883404. DOI: 10.3389/fimmu.2022.883404
66. Swan C, Duroudier NP, Campbell E, Zaitoun A, Hastings M, Dukes GE. **Identifying and testing candidate genetic polymorphisms in the irritable bowel syndrome (IBS): association with TNFSF15 and TNFalpha**. *Gut* (2013) **62** 985-94. DOI: 10.1136/gutjnl-2011-301213
67. Janssens R, Struyf S, Proost P. **The unique structural and functional features of CXCL12**. *Cell Mol Immunol* (2018) **15** 299-311. DOI: 10.1038/cmi.2017.107
68. Yang Q, Ren GL, Wei B, Jin J, Huang XR, Shao W. **Conditional knockout of TGF-betaRII /Smad2 signals protects against acute renal injury by alleviating cell necroptosis, apoptosis and inflammation**. *Theranostics* (2019) **9** 8277-93. DOI: 10.7150/thno.35686
69. Bozza MT, Lintomen L, Kitoko JZ, Paiva CN, Olsen PC. **The role of MIF on eosinophil biology and Eosinophilic inflammation**. *Clin Rev Allergy Immunol* (2020) **58** 15-24. DOI: 10.1007/s12016-019-08726-z
70. Markovic I, Savvides SN. **Modulation of signaling mediated by TSLP and IL-7 in inflammation, autoimmune diseases, and cancer**. *Front Immunol* (2020) **11** 1557. DOI: 10.3389/fimmu.2020.01557
71. Yu GI, Song DK, Shin DH. **Associations of IL1RAP and IL1RL1 gene polymorphisms with obesity and inflammation mediators**. *Inflamm Res* (2020) **69** 191-202. DOI: 10.1007/s00011-019-01307-y
72. He H, Del Duca E, Diaz A, Kim HJ, Gay-Mimbrera J, Zhang N. **Mild atopic dermatitis lacks systemic inflammation and shows reduced nonlesional skin abnormalities**. *J Allergy Clin Immunol* (2021) **147** 1369-80. DOI: 10.1016/j.jaci.2020.08.041
73. Pancholy SB, Bernat I, Patel TM. **Prevention of radial artery occlusion after transradial catheterization: the PROPHET-II randomized trial**. *JACC Cardiovasc Interv* (2016) **9** 1992-9. DOI: 10.1016/j.jcin.2016.07.020
74. Shi G, Li F, Zhang L, Gong C, Xue S, Song Y. **Retrograde recanalization of occluded radial artery: a single-centre experience and literature review**. *J Endovasc Ther* (2022) **29** 755-62. DOI: 10.1177/15266028211067732
75. Hirshfeld JW, Nathan AS. **Deriving function from structure: applying Hagen-Poiseuille to coronary arteries**. *JACC Cardiovasc Interv* (2020) **13** 498-501. DOI: 10.1016/j.jcin.2019.11.031
76. Schiaffino S, Reggiani C. **Fiber types in mammalian skeletal muscles**. *Physiol Rev* (2011) **91** 1447-531. DOI: 10.1152/physrev.00031.2010
|
---
title: Reassessment of the Need for Asthmatic Patients for Biologic Treatment in a
Tertiary Care Hospital
journal: Cureus
year: 2023
pmcid: PMC10022474
doi: 10.7759/cureus.36288
license: CC BY 3.0
---
# Reassessment of the Need for Asthmatic Patients for Biologic Treatment in a Tertiary Care Hospital
## Abstract
Background: *Asthma is* a common chronic inflammatory airway condition. In difficult-to-treat asthma, poor control can be linked to multiple factors like the presence of uncontrolled comorbidities (e.g. gastroesophageal reflux and allergic rhinitis), as well as to poor inhaler use techniques and adherence. In this study we wanted to evaluate our severe asthma patients already on a biologic treatment with regard to presence of any of these factors.
Method: A questionnaire-based study, filled by investigators through direct interview with patients. We included all asthma patients on biologic treatment at King Abdul Aziz Medical City, Riyadh, KSA. Started in October 2020 and ended in December 2020. The questionnaire had a demographic section and sections for asthma symptoms, compliance, inhaler techniques, and comorbidities.
Result: Case series of $$n = 38$$ severe asthma patients showed that majority had partially controlled or uncontrolled asthma ($66\%$). Some $42\%$ had intermediate/high risk for obstructive sleep apnea (OSA) based on the common screening tool “STOPBANG” score. Some $47\%$ of our patients had uncontrolled gastro-esophageal reflux disease (GERD), and majority ($80\%$) had uncontrolled allergic rhinitis. Only half of them demonstrated appropriate inhaler technique. And none of them was found exposed to asthma triggers at the time of interview.
Conclusion: Significant number of severe uncontrolled asthmatic patients were shown to be associated with at least one comorbid condition that might be interfering with patients’ improvement in asthmatic symptoms. By taking appropriate measures toward management and controlling of those comorbid conditions and also educating patients about technique to use inhalers might show notable improvement in asthmatic patients’ condition.
## Introduction
Asthma is a common chronic inflammatory condition that intermittently inflames and narrows the airways in the respiratory tract, and it affects more than 300 million individuals globally [1-2]. It develops as a consequence of the inflammatory reactions involving the neutrophils, eosinophils, macrophages, lymphocytes, and mast cells [3-4]. In difficult-to-treat asthma, poor control can be linked to poor adherence to inhaled therapy, incorrect inhaler technique, and coexisting conditions, including exposure to allergens and irritants [2]. Asthma is considered to be severe when control remains poor despite measures that adequately address medication delivery and comorbidities [5].
Adequate treatment of allergic asthma requires an array of different treatments to be administered for an indefinite amount of time; furthermore, before initiating a new drug therapy, practitioners should check the patient’s adherence with existing therapies, such as checking the inhaler techniques and eliminating triggering factors [6]. Causes behind insufficient asthma control are multifactorial which comprise medication access and compliance [7]. A diagnosis of severe asthma is established when all the comorbidities are identified and well specified.
Patients with severe asthma have persistent symptoms or frequent exacerbations that require repetitive glucocorticoid bursts, maintenance oral glucocorticoid therapy, or both, despite adequate treatment with high-dose inhaled glucocorticoids, long-acting β2-agonists, and long-acting muscarinic antagonists [8]. In these patients, add-on treatment, which may include biologic therapies (e.g., omalizumab and mepolizumab) is needed to reduce the disease burden.
There are many contributing factors to poorly controlled asthma. For example, poor inhaler techniques, which can compromise the medication efficacy, were found in $70\%$-$80\%$ of asthmatic patients [8-10]. In addition, $67\%$ of physicians do not know how to illustrate the appropriate steps to use inhalers which can worsen the uncontrolled asthma [11]. Both active/passive electronic cigarette smoking can cause and exacerbate existing asthma [12]. Patient’s compliance can play a major role in guiding the treatment; for instance, youth and female patients were found better in adhering to the treatment plan than elderly and male patients [13]. While obesity is known to cause obstructive sleep apnea (OSA) and sleep-disordered breathing (SDB), obesity also plays a dose-related effect on asthma incidence among males and females, respectively [14-15].
Biologic treatment for severe asthma has got greater popularity among lung specialists, and data looking at justified prescribing habit are lacking. Therefore, the goal of this study was to evaluate our severe asthma patients who already are on a biologic treatment with regard to asthma medication compliance, asthma medication administration technique, existence of comorbidities known to interact with asthma, and how controlled are these comorbidities, as well as for environmental exposures.
## Materials and methods
The study was conducted at King Abdul Aziz Medical City in Riyadh Saudi Arabia (tertiary care hospital). A questionnaire was developed to collect information from the patients who agreed and consented to participate. Inclusion: All severe Asthma patients on biologics at King Abdul Aziz Medical City in Riyadh. Exclusion: any subjects less than 18 years of age, or using biologic treatment for diagnoses other than severe asthma.
Investigators interviewed the patients and filled the questionnaires themselves (started in October 2020 and ended in December 2020). The questionnaire had a demographic section and sections for asthma symptoms, compliance, inhaler techniques, and comorbidities. Then control of asthma symptoms was assessed based on the Global Initiative for Asthma in all of them at the time of interview. Those who had 20-25 points were considered having controlled asthma, those who scored 16-19 were considered partially controlled, and those who scored less than 16 points were considered in an uncontrolled asthma situation. Regarding OSA, patients were divided into three groups based on their STOPBANG score, where patient who scored 0-2 were considered having a mild risk to have OSA, those with scores 3-4 were considered at moderate risk, and those with 5-8 were considered having high risk to have an OSA. For GERD, patients were divided into two groups based on their GERD symptoms; if they have two or more days of symptoms they were given a score of 1, and if less they were given a score zero. Regarding compliance to asthma medications, they were asked 10 specific questions, and then were asked to demonstrate inhalers technique. Finally, they were asked to grade their confidence in using inhalers technique from 1 to 10. Data were entered in an encrypted excel sheet. Statistical Program for Social Sciences (SPSS) was used for data analysis. Frequencies and percentages were calculated for categorical variables, and cut off for significance was p value of less than 0.05.
## Results
An observational cross-sectional study of $$n = 38$$ severe asthma patients (all were on biologic treatment), using Statistical Program for Social Sciences (SPSS) for analysis, showed that majority ($63\%$) were female, and more than two third were overweight or obese, with remaining characters (educational level, age, and smoking status) listed in Table 1. Regarding asthma control, $34\%$ had controlled asthma, $24\%$ had partially controlled asthma, and $42\%$ were with uncontrolled asthma. Regarding OSA risk assessed by STOPBANG scoring, $58\%$ of the patients were with low OSA risk, and $24\%$ with intermediate OSA risk, and $18\%$ with high risk of having an OSA. Some $53\%$ of our patients had controlled or no GERD, and majority ($80\%$) of patients on biologics had uncontrolled allergic rhinitis/rhinosinusitis (Table 2). In 1-10 scoring, most of them ($84\%$) scored themselves 8 and above, with regard to confidence level concerning how well they use their inhalers. However, when asked to demonstrate inhaler technique, only half of them ($50\%$) demonstrated appropriate inhalers technique. Cross-tabulations between asthma control and comorbidities revealed no significant correlation, with the p-values above 0.05 (Tables 3-5). All of the patients were not found exposed to asthma triggers at the time of interview.
## Discussion
To the best of our knowledge, this is the first study assessing retrospectively severe asthma patients on biologics for appropriateness of biologic treatment prescription. This was mainly by looking at the presence of controllable factors associated with poor asthma control. As per asthma guidelines, diagnosing severe asthma needs careful attention to rule out alternative diagnoses and managing morbidities associated with poor asthma control. Also, assessment of patients while applying a step-wise management approach should be regularly done. Our research hypothesized that those measures were not properly followed, so, unnecessary upgrading of asthma treatment was done, or comorbidities remained unaddressed.
A prospective study included 547 participants concluded that asthma patients had a $39\%$ increased risk of OSA after exclusion of other risk factors such as smoking and obesity. Furthermore, asthma symptoms can be improved by controlling the OSA [16-17]. More than half of our severe asthma sample had moderate or high OSA risk which might have led to poor asthma control and upgrading of treatment.
The GERD on the other hand is very closely linked to asthma which goes side by side with our results [18]. Half of our asthmatic patients had uncontrolled GERD which again could have led to unnecessary upgrading of asthma management. Interestingly more than $80\%$ of our patients had an uncontrolled allergic rhinitis/rhinosinusitis. This may have led to the increased number of uncontrolled asthma patients despite the use of biologic therapy, together with aforementioned factors. It was noticed that most of our patients thought they are capable of using their inhalers properly, however, only half of them were found able to use asthma inhalers properly enough (when asked by investigators to demonstrate it). This might have been secondary to either doctor’s factors (like lack of knowledge or lack of time to teach the patients) or patient’s factors (like low socioeconomic or educational level). Based on above findings, we hypothesize that if all above factors were taken in to consideration when treating severe asthma patients, they would achieve better asthma control and probably some of them might not need biologic therapy. Limitation to our study: The small sample size, and the research was confined to a single center in Riyadh city, and being conducted retrospectively. We would suggest a prospective kind of studies to include larger number in a multicenter approach of severe asthma patients who follow them while controlling all of the factors associated with poor asthma control, to see how many of them would achieve better control, and how many would be controlled with less asthma medications.
## Conclusions
We reassessed our severe asthma patients (those on biologic treatment) for appropriateness of biologic therapy prescribing, mainly by looking at the presence of controllable factors known to exacerbate asthma. We found that significant number of them had at least one comorbid condition that might be interfering with their asthma condition. Also half of them were found not using their asthma medication properly enough. By taking appropriate measures to management and controlling those comorbid conditions and also educating patients about technique to use inhalers might show notable improvement and control in asthmatic patient’s condition. Thereby, reducing the need of upgrading to biologic therapy.
## References
1. McGregor MC, Krings JG, Nair P, Castro M. **Role of biologics in asthma**. *Am J Respir Crit Care Med* (2019) **199** 433-445. PMID: 30525902
2. To T, Stanojevic S, Moores G, Gershon AS, Bateman ED, Cruz AA, Boulet LP. **Global asthma prevalence in adults: findings from the cross-sectional world health survey**. *BMC Public Health* (2012) **12** 204. PMID: 22429515
3. Boonpiyathad T, Sözener ZC, Satitsuksanoa P, Akdis CA. **Immunologic mechanisms in asthma**. *Semin Immunol* (2019) **46** 101333. PMID: 31703832
4. Schatz M, Rosenwasser L. **The allergic asthma phenotype**. *J Allergy Clin Immunol Pract* (2014) **2** 645-648. PMID: 25439351
5. Chung KF, Wenzel SE, Brozek JL. **International ERS/ATS guidelines on definition, evaluation and treatment of severe asthma [Internet]**. *Eur Respir J* (2014) **43** 343-373. PMID: 24337046
6. **British Thoracic Society and Scottish Intercollegiate Guidelines Network. British guideline on the management of asthma. A national clinical guideline [Internet]**. (2019)
7. Slejko JF, Ghushchyan VH, Sucher B, Globe DR, Lin SL, Globe G, Sullivan PW. **Asthma control in the United States, 2008-2010: indicators of poor asthma control**. *J Allergy Clin Immunol* (2014) **133** 1579-1587. PMID: 24331376
8. **Global Initiative for Asthma. 2021 GINA report, global strategy for asthma management and prevention (2021 update)**. (2021)
9. Braido F, Chrystyn H, Baiardini I. **"Trying, But Failing" - the role of inhaler technique and mode of delivery in respiratory medication adherence**. *J Allergy Clin Immunol Pract* (2016) **4** 823-832. PMID: 27587316
10. **Global Initiative for Chronic Obstructive Pulmonary Disease (GOLD). Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease [Internet]**. (2018)
11. Rootmensen GN, van Keimpema AR, Jansen HM, de Haan RJ. **Predictors of incorrect inhalation technique in patients with asthma or COPD: a study using a validated videotaped scoring method**. *J Aerosol Med Pulm Drug Deliv* (2010) **23** 323-328. PMID: 20804428
12. Kim SY, Sim S, Choi HG. **Active, passive, and electronic cigarette smoking is associated with asthma in adolescents**. *Sci Rep* (2017) **7** 17789. PMID: 29259221
13. Bracken M, Fleming L, Hall P. **The importance of nurse-led home visits in the assessment of children with problematic asthma**. *Arch Dis Child* (2009) **94** 780-784. PMID: 19546102
14. Kong DL, Qin Z, Shen H, Jin HY, Wang W, Wang ZF. **Association of obstructive sleep apnea with asthma: a meta-analysis**. *Sci Rep* (2017) **7** 4088. PMID: 28642543
15. Beuther DA, Sutherland ER. **Overweight, obesity, and incident asthma: a meta-analysis of prospective epidemiologic studies**. *Am J Respir Crit Care Med* (2007) **175** 661-666. PMID: 17234901
16. Teodorescu M, Barnet JH, Hagen EW, Palta M, Young TB, Peppard PE. **Association between asthma and risk of developing obstructive sleep apnea**. *JAMA* (2015) **313** 156-164. PMID: 25585327
17. Alkhalil M, Schulman E, Getsy J. **Obstructive sleep apnea syndrome and asthma: what are the links?**. *J Clin Sleep Med* (2009) **5** 71-78. PMID: 19317386
18. Mastronarde JG. **Is there a relationship between GERD and asthma?**. *Gastroenterol Hepatol* (2012) **8** 401-403
|
---
title: 'Long-term outcomes of less drug-eluting stents by the use of drug-coated balloons
in de novo coronary chronic total occlusion intervention: A multicenter observational
study'
authors:
- Xi Wang
- Xinyue Yang
- Wenjie Lu
- Liang Pan
- Zhanying Han
- Sancong Pan
- Yingguang Shan
- Xule Wang
- Xiaolin Zheng
- Ran Li
- Yongjian Zhu
- Peng Qin
- Qiangwei Shi
- Shuai Zhou
- Wencai Zhang
- Sen Guo
- Peisheng Zhang
- Xiaofei Qin
- Guoju Sun
- Zhongsheng Qin
- Zhenwen Huang
- Chunguang Qiu
journal: Frontiers in Cardiovascular Medicine
year: 2023
pmcid: PMC10022494
doi: 10.3389/fcvm.2023.1045859
license: CC BY 4.0
---
# Long-term outcomes of less drug-eluting stents by the use of drug-coated balloons in de novo coronary chronic total occlusion intervention: A multicenter observational study
## Abstract
### Background
Data on drug-coated balloons (DCB) for de novo coronary chronic total occlusion (CTO) are limited. We aimed to investigate the long-term outcomes of substitution of drug-eluting stents (DES) by DCB.
### Methods
We compared the outcomes of less DES strategy (DCB alone or combined with DES) and DES-only strategy in treating de novo coronary CTO in this prospective, observational, multicenter study. The primary endpoints were major adverse cardiovascular events (MACE), target vessel revascularization, myocardial infarction, and death during 3-year follow-up. The secondary endpoints were late lumen loss (LLL) and restenosis until 1-year after operation.
### Results
Of the 591 eligible patients consecutively enrolled between January 2015 and December 2019, 281 (290 lesions) were treated with DCB (DCB-only or combined with DES) and 310 (319 lesions) with DES only. In the DCB group, 147 ($50.7\%$) lesions were treated using DCB-only, and the bailout stenting rate was relatively low ($3.1\%$). The average stent length per lesion in the DCB group was significantly shorter compared with the DES-only group (21.5 ± 25.5 mm vs. 54.5 ± 26.0 mm, $p \leq 0.001$). A total of 112 patients in the DCB group and 71 patients in the DES-only group ($38.6\%$ vs. $22.3\%$, $p \leq 0.001$) completed angiographic follow-up until 1-year, and LLL was much less in the DCB group (−0.08 ± 0.65 mm vs. 0.35 ± 0.62 mm, $p \leq 0.001$). There were no significant differences in restenosis occurrence between the two groups ($20.5\%$ vs. $19.7\%$, $p \leq 0.999$). The Kaplan–Meier estimates of MACE at 3-year ($11.8\%$ vs. $12.0\%$, log-rank $$p \leq 0.688$$) was similar between the groups.
### Conclusion
Percutaneous coronary intervention with DCB is a potential “stent-less” therapy for de novo CTO lesions with satisfactory long-term clinical results compared to the DES-only approach.
## 1. Introduction
Coronary chronic total occlusion (CTO) is characterized by thrombolysis in myocardial infarction (TIMI) grade 0, for at least 3 months, determined angiographically or clinically. CTO is found in 15–$25\%$ of individuals undergoing coronary angiography [1, 2]. Successful CTO revascularization can lead to symptomatic angina relief, and improved cardiac function and clinical outcomes [3, 4]. Recently, CTO percutaneous coronary intervention (PCI) has advanced rapidly owing to the growing experience of interventional cardiologists in antithrombotic drug management, expanding application of drug-eluting stents (DES), and procedural strategies [5]. Currently, new-generation DES is the most widely used intervention in recanalized CTO and reduces the incidence of restenosis and major adverse cardiovascular events (MACE) [5, 6]. However, some disadvantages remain, including risks of restenosis and stent thrombosis (ST) due to long stent implantation, unsuitability for stenting immediately after predilation, and distal vessel remodeling after antegrade flow restoration [7, 8].
Drug-coated balloons (DCB) were first reported by Scheller to treat in-stent restenosis (ISR) [9]. Compared with DES, DCB can rapidly deliver anti-proliferative drugs to the coronary vessel wall, and no permanent metal implants are left after PCI. Owing to the potential advantages of reducing adverse events associated with DES and shortening the period of dual antiplatelet therapy (DAPT), DCB angioplasty is becoming a popular alternative treatment for coronary lesions. Previous studies have demonstrated that DCB has sound clinical effects and safety in bare metal stent (BMS)/DES ISR [10].
In terms of small vessel disease, DCB is also demonstrably safe and effective, and increasing evidence supports DCB in treating acute myocardial infarction (MI), bifurcation lesions, and large vessel disease [11]. However, only a few studies with a small sample size and short follow-up duration have evaluated DCB in treating de novo CTO lesions (12–14). We aimed to compare the efficacy and safety of less DES strategy (DCB-only or combined with DES) and DES-only strategy in treating de novo CTO lesions in a prospective, multicenter, observational study with long-term follow-up.
## 2.1. Patient population
This prospective observational study was performed in three high-volume PCI centers in China [15, 16]. Patients with coronary CTO treated with DCB and/or DES between January 2015 and December 2019 were enrolled consecutively. Inclusion criteria were [1] coronary CTO lesions, defined as TIMI grade 0 with a minimum period of 3 months, as demonstrated angiographically or clinically, and [2] recanalized CTO with DCB angioplasty and/or DES implantation. Exclusion criteria were [1] ISR or graft lesions; [2] acute ST-segment elevation MI necessitating primary PCI, and [3] life expectancy of less than 12 months (Figure 1).
**Figure 1:** *Study population. CTO, chronic total occlusion; DCB, drug-coated balloon; DES, drug-eluting stent; ISR, in-stent restenosis; MI, myocardial infarction; PCI, percutaneous coronary intervention.*
This research was performed according to the Declaration of Helsinki. The local ethics committee approved the research and all patients signed written informed consent forms. This research was not sponsored by any external source.
## 2.2. PCI procedure
PCI was performed according to standard procedures. All patients were administered aspirin and clopidogrel or ticagrelor. Heparin was routinely administered during the procedure at a loading dose of 80–100 IU/kg followed by 1,000 IU per hour. Baseline angiography was completed with at least two near-orthogonal images to show the target lesion following an intracoronary injection of nitroglycerin (100–200 μg). Recanalization strategies for CTO lesions were decided by interventionists.
In the DCB group, predilation was performed with either semi-compliant balloons or noncompliant, cutting, scoring, or dual wire balloons. If predilation was satisfactory, DCB was used to cover the entire target or partial lesion in cases of dissection < type C and residual stenosis ≤$50\%$. New-generation DES implantation was performed as part of an initial hybrid strategy combining DCB and DES, or as bailout stents in segments with flow-limiting dissection or significant residual stenosis (> $50\%$) after DCB angioplasty. Paclitaxel-coated balloons (SeQuent ™ Please, B. Braun, Melsungen, Germany) were used in all patients undergoing DCB angioplasty. New-generation DES was used in patients treated with DES only. Four types of DES were implanted: Resolute IntegrityTM (Medtronic, Santa Rosa, CA, USA), SynergyTM (Boston Scientific, Maple Grove, MN, USA), Excrossal (JW Medical System, China), Excel (JW Medical System, China).
Administration of post-procedure glycoprotein IIb/IIIa receptor inhibitors was based on the patient’s condition. Patients without DES implantation received DAPT for at least 1 month, and DAPT was maintained for 6–12 months in patients after stenting.
## 2.3. Angiographic analysis
According to the results of CAG, the calcification lesions were classified as grade 0 (no calcification). Grade I (mild calcification) can only see the faint and fuzzy high-density shadow when the heart is beating, but cannot see the calcification shadow completely when the heart is not beating. Grade II (moderate calcification) sees sharper, easier-to-see dense shadows when the heart beats. Grade III (severe calcification) can be seen with a clear, dense shadow in both beating and non-beating hearts [17]. Edge detection methods (QAngio XA 7.3 version; Medis Medical Imaging, Leiden, Netherlands) were used to provide quantitative coronary angiography (QCA) measurements during intervention and follow-up angiography. Lesion length, reference vessel diameter (RVD), minimum luminal diameter (MLD), and diameter stenosis percentage were measured. The difference in MLD immediately after intervention and during follow-up angiography was used to compute late lumen loss (LLL).
## 2.4. Clinical endpoints and definitions
The primary endpoints were MACE, defined as all-cause death, MI, and target vessel revascularization (TVR). The secondary end points were LLL and restenosis. Periprocedural serious adverse events included ST, MI, and death during hospitalization. According to the QCA results, ISR was defined as a restenosis of $50\%$ within the DCB or DES area or at its adjacent 5-mm segments. Death was considered cardiogenic unless there is an apparent non-cardiogenic cause. The Fourth Universal Definition of Myocardial Infarction defined procedure-related, nonprocedural, or after-discharge MI [18]. Revascularization of the target lesion (lesion within 5 mm from the edge of each end) and target vessel was defined as target lesion revascularization (TLR) and TVR, respectively. The criteria provided by the Academic Research Consortium were used to classify the ST [19].
## 2.5. Follow-up
Follow-up was achieved for a total of 3 years using clinical visits or phone calls every 3 months. Angiograms were performed 6 and 12 months after PCI or when clinically necessary, but this was not mandatory.
## 2.6. Statistical analysis
Data were presented as the mean ± standard deviation or median (interquartile range). The t-test or non-parametric test was used to assess differences between continuous variables, and the chi-square test or Fisher’s exact test was used for categorical variables. The cumulative incidence of outcomes was estimated from the Kaplan–Meier curve and compared using the log-rank test. The multivariate Cox proportional regression model was used to adjust for potential confounders, including baseline patient characteristics, target vessels, calcification, lesion length, MLD after PCI, and RVD. Statistical analysis was conducted using SPSS 23.0 (IBM SPSS, SPSS Inc.). A two-sided p value of <0.05 was considered statistically significant.
## 3.1. Population
We identified 591 consecutive eligible patients with CTO and 281 patients ($47.5\%$) received DCB between January 2015 and December 2019. The mean age was 58.80 ± 10.87 years, and $72.93\%$ of the patients were male. A high percentage of patients had hypertension ($55.16\%$), diabetes ($35.36\%$), and hypercholesterolemia ($57.19\%$). Majority ($77.9\%$) of the patients presented with multivessel diseases. A total of 281 patients treated with DCB (DCB-only or DCB combined with DES for sequential segments of a lesion) and 310 patients with DES-only implantations were enrolled. There were no significant differences in the baseline patient characteristics between the groups (Table 1).
**Table 1**
| Unnamed: 0 | DCB (n = 281) | DES-only (n = 310) | P |
| --- | --- | --- | --- |
| Age, years | 58.4 ± 10.9 | 59.1 ± 10.8 | 0.431 |
| Male | 207 (73.7%) | 224 (72.3%) | 0.712 |
| Comorbidities | Comorbidities | Comorbidities | Comorbidities |
| Hypertension | 154 (54.8%) | 172 (55.5%) | 0.869 |
| Diabetes mellitus | 105 (37.4%) | 104 (33.6%) | 0.344 |
| Hypercholesterolemia | 150 (53.4%) | 188 (60.7) | 0.081 |
| Family history of CAD | 63 (22.4%) | 58 (18.71%) | 0.307 |
| Previous MI | 76 (27.1%) | 90 (29.0%) | 0.647 |
| Previous PCI | 39 (13.9%) | 41 (13.2%) | 0.904 |
| Previous CABG | 4 (1.4%) | 10 (3.2%) | 0.182 |
| Current/ex-smoker | 110 (39.2%) | 122 (39.4%) | >0.999 |
| Unstable angina | 65 (23.1%) | 69 (22.3%) | 0.844 |
| Number of diseased vessels | Number of diseased vessels | Number of diseased vessels | Number of diseased vessels |
| 1 | 62 (22.1%) | 53 (17.1%) | 0.204 |
| 2 | 97 (34.5%) | 103 (33.2%) | |
| 3 | 122 (43.4%) | 154 (49.7%) | |
| Multivessel disease | 202 (71.9%) | 213 (68.7%) | 0.419 |
| Complete revascularization | 210 (74.7%) | 226 (72.9%) | 0.640 |
| Ejection fraction, % | 58.2 ± 7.0 | 57.0 ± 8.0 | 0.060 |
| Medication during hospitalization | Medication during hospitalization | Medication during hospitalization | Medication during hospitalization |
| Aspirin | 281 (100.0%) | 310 (100.0%) | >0.999 |
| Clopidogrel | 99 (35.2%) | 101 (32.6%) | 0.542 |
| ticagrelor | 182 (64.8%) | 209 (67.4%) | 0.542 |
| statin | 281 (100.0%) | 310 (100.0%) | >0.999 |
| Beta-blocker | 194 (69.0%) | 221 (71.3%) | 0.589 |
## 3.2. Procedural characteristics
The most common PCI access was the transradial artery ($93.1\%$ vs. $91.2\%$, $$p \leq 0.452$$) in both groups. The most frequent target vessels were the left anterior descending (LAD) ($39.7\%$) and right coronary artery (RCA) ($40.0\%$) in the DCB group and the RCA ($49.5\%$) in the DES-only group ($$p \leq 0.031$$). The percentage of calcified lesions was relatively higher in the DES-only group ($7.6\%$ vs. $11.9\%$, $$p \leq 0.078$$). Noncompliant balloons, cutting balloons, and nonslip element balloons were more frequently used in the DCB group. The incidence of dissection after post-dilation was higher in the DES-only group than that in the DCB group (No dissection: $30.1\%$ vs. $40.0\%$; A-B dissection: $53.9\%$ vs. $50.7\%$; C-F dissection: $16.0\%$ vs. $9.3\%$, $$p \leq 0.007$$). The incidence of type A-B and type C-F dissection after DCB angioplasty in the DCB group was 49.7 and $12.1\%$, and there was no dissection after stent implantation in the DES-only group ($p \leq 0.001$). Nine lesions ($3.1\%$) in the DCB group required bailout stents. Table 2 summarizes the procedure and lesion characteristics.
**Table 2**
| Unnamed: 0 | DCB (n = 290) | DES-only (n = 319) | P |
| --- | --- | --- | --- |
| Access | Access | Access | Access |
| Trans-radial | 270 (93.1%) | 291 (91.2%) | 0.452 |
| Trans-femoral | 20 (6.9%) | 28 (8.8%) | |
| Target vessel | Target vessel | Target vessel | Target vessel |
| LAD | 115 (39.7%) | 116 (36.4%) | 0.031 |
| LCX | 59 (20.3%) | 45 (14.1%) | |
| RCA | 116 (40.0%) | 158 (49.5%) | |
| Pre-dilation | Pre-dilation | Pre-dilation | Pre-dilation |
| Semi-compliant balloon (%) | 277 (95.5%) | 311 (97.5%) | 0.191 |
| Non-compliant balloon (%) | 45 (15.5%) | 17 (5.3%) | <0.001 |
| Cutting balloon (%) | 56 (19.3%) | 33 (10.3%) | 0.002 |
| NSE balloon (%) | 79 (27.2%) | 24 (7.5%) | <0.001 |
| Dual wire balloon (%) | 13 (4.5%) | 6 (1.9%) | 0.100 |
| Moderate/severe Calcification | 22 (7.6%) | 38 (11.9%) | 0.078 |
| Rotation | 2 (0.7%) | 5 (1.6%) | 0.454 |
| J-CTO score | 1.79 ± 1.07 | 1.94 ± 1.14 | 0.094 |
| Post-dilation | Post-dilation | Post-dilation | Post-dilation |
| Dissection | Dissection | Dissection | Dissection |
| | 116 (40.0%) | 96 (30.1%) | 0.007 |
| A-B | 147 (50.7%) | 172 (53.9%) | |
| C-F | 27 (9.3%) | 51 (16.0%) | |
| Treatment Strategy | Treatment Strategy | Treatment Strategy | Treatment Strategy |
| DCB-only | 147 (50.7%) | / | / |
| DCB combined with DES | 143 (49.3%) | / | / |
| Dissection after DCB angioplasty | Dissection after DCB angioplasty | Dissection after DCB angioplasty | Dissection after DCB angioplasty |
| | 111 (38.3%) | 319 (100%) | <0.001 |
| A-B | 144 (49.7%) | 0 | |
| C-F | 35 (12.1%) | 0 | |
| Bailout stent | 9 (3.1%) | / | / |
The mean number of DCB per lesion and length of DCB in the DCB group were 1.41 ± 0.66 and 35.81 ± 19.92 mm, respectively. DES implanted per lesion in the DCB group was significantly less in number (0.75 ± 0.87 vs. 1.99 ± 0.83, $p \leq 0.001$) and shorter (21.52 ± 25.46 mm vs. 54.45 ± 26.03 mm, $p \leq 0.001$) than that in the DES-only group. Additionally, the average diameter of DES in the DCB group was smaller (2.82 ± 0.28 mm vs. 2.90 ± 0.35 mm, $$p \leq 0.018$$). The total length of DCB + DES in the DCB group was similar with that in the DES-only group (57.33 ± 26.86 mm vs. 54.45 ± 26.03 mm, $$p \leq 0.180$$). ( Table 3).
**Table 3**
| Unnamed: 0 | DCB (n = 290) | DES-only (n = 319) | p |
| --- | --- | --- | --- |
| Characteristics of DCB | Characteristics of DCB | Characteristics of DCB | Characteristics of DCB |
| Number | 1.41 ± 0.66 | / | / |
| Length, mm | 35.8 ± 19.9 | / | / |
| Diameter, mm | 2.63 ± 0.38 | / | / |
| Pressure of inflation, atm | 8.03 ± 1.21 | / | / |
| Time of inflation, s | 60.9 ± 5.1 | / | / |
| Characteristics of DES | Characteristics of DES | Characteristics of DES | Characteristics of DES |
| Number | 0.75 ± 0.87 | 1.99 ± 0.83 | <0.001 |
| Length, mm | 21.5 ± 25.5 | 54.5 ± 26.0 | <0.001 |
| Diameter, mm | 2.82 ± 0.28 | 2.90 ± 0.35 | 0.018 |
| Total length of devices | 57.3 ± 26.9 | 54.5 ± 26.0 | 0.180 |
## 3.3. QCA measurements
QCA measurements showed that the length of lesions was shorter in DCB group than DES-only group (41.82 ± 19.09 mm vs. 45.67 ± 23.70 mm, $$p \leq 0.027$$). Post-procedure MLD (1.83 ± 0.43 mm vs. 2.55 ± 0.40 mm, $p \leq 0.001$), RVD (2.45 ± 0.43 mm vs. 2.87 ± 0.44 mm, $p \leq 0.001$) and acute lumen gain (1.83 ± 0.43 mm vs. 2.55 ± 0.40 mm, $p \leq 0.001$) were smaller in the DCB group. Among the patients with follow-up angiographic results at 1 year, no significant differences in the incidence of restenosis ($20.5\%$ vs. $19.7\%$, $p \leq 0.999$) or reocclusion ($2.7\%$ vs. $2.8\%$, $p \leq 0.999$) were observed, and LLL in the DCB group (−0.08 ± 0.65 mm vs. 0.35 ± 0.62 mm, $p \leq 0.001$) was less than that in the DES-only group (Table 4).
**Table 4**
| Unnamed: 0 | DCB (n = 290) | DES-only (n = 319) | p |
| --- | --- | --- | --- |
| Before PCI | Before PCI | Before PCI | Before PCI |
| Lesion length, mm | 41.8 ± 19.1 | 45.7 ± 23.7 | 0.027 |
| Immediately after PCI | Immediately after PCI | Immediately after PCI | Immediately after PCI |
| MLD, mm | 1.83 ± 0.43 | 2.55 ± 0.40 | <0.001 |
| RVD, mm | 2.45 ± 0.43 | 2.87 ± 0.44 | <0.001 |
| Diameter stenosis, % | 25.4 ± 9.2 | 10.9 ± 3.1 | <0.001 |
| Acute lumen gain, mm | 1.83 ± 0.43 | 2.55 ± 0.40 | <0.001 |
| Angiographic follow-up at 1 year | Angiographic follow-up at 1 year | Angiographic follow-up at 1 year | Angiographic follow-up at 1 year |
| No. of patients | 112 (38.6%) | 71 (22.3%) | <0.001 |
| RVD, mm | 2.65 ± 0.51 | 2.97 ± 0.42 | <0.001 |
| MLD, mm | 1.86 ± 0.65 | 2.28 ± 0.63 | |
| Diameter stenosis, % | 30.5 ± 18.0 | 22.8 ± 18.0 | 0.005 |
| Restenosis | 23 (20.5%) | 14 (19.7%) | >0.999 |
| Total occlusion | 3 (2.7%) | 2 (2.8%) | >0.999 |
| Late lumen loss, mm | −0.08 ± 0.65 | 0.35 ± 0.62 | <0.001 |
| Late lumen enlargement | 68 (60.7%) | / | / |
## 3.4. Clinical outcomes
During hospitalization, no serious adverse events (definite or probable ST, MI, or death) were observed (Table 5). During the 3 years of follow-up, no target lesion thrombosis occurred in any of the groups. Compared with the DES-only group, the Kaplan–Meier estimates of TLR ($6.7\%$ vs. $5.4\%$, log-rank $$p \leq 0.484$$) and MACE ($12.0\%$ vs. $11.8\%$, log-rank $$p \leq 0.688$$) at 3 years were similar in the DCB group (Figure 2). A multivariate Cox proportional regression model was used to adjust for potential confounders (Table 6). No difference in the risk of MACE was observed between groups (Hazard ration [HR] = 0.932, $95\%$ confidence interval [CI] = 0.532–1.633, $$p \leq 0.805$$). Figure 2 shows the Kaplan–Meier survival curves of the endpoints.
## 4. Discussion
To our knowledge, this is the first study to examine the long-term effects of DCB and DES in the treatment of de novo CTO lesions. This prospective, multicenter, observational study demonstrated that the less DES strategy (DCB with or without DES) was effective and safe, with no significant differences in the incidence of MACE during long-term follow-up.
CTO, as a challenging obstacle in PCI accounts for 15–$25\%$ of elective coronary angiographies [1, 2]. With the development of intervention techniques and concepts for recanalization of CTO, the rates of successful recanalization have improved [5]. Compared to optimal medical therapy, successful revascularization through PCI can effectively reduce the clinical symptoms related with myocardial ischemia [3, 4]. CTO-PCI with complete new-generation DES implantation has satisfactory clinical outcomes. However, concerns about undersized stents after balloon angioplasty, ST, and stent malapposition remain [7, 8].
DCB was originally designed to treat coronary ISR after BMS implantation [9]. Clinical trials have demonstrated the efficacy of DCB in BMS/DES-ISR [10]. Meanwhile evidence regarding the use of DCB in treating small vessels and bifurcation lesions is accumulating [11]. Previous studies have demonstrated that the DCB-only strategy for CTO is a feasible and well-tolerated approach [12, 13]. We aimed to investigate the impact of long-term angiographic and clinical outcomes of DCB treatment (DCB alone or combined with DES) in de novo CTO lesions.
This study demonstrated encouraging long-term outcomes of DCB-only or combined with DES in CTO-PCI, with low rates of MACE ($12.0\%$). In the DECISION-CTO trial [20], the primary endpoint rate (death, MI, stroke, or any revascularization) in the CTO-PCI group (receiving DES implantation) was $21.5\%$ at 3 years, which was comparable to our results. Furthermore, no target lesion thrombosis was observed during hospitalization or follow up. The low thrombosis rate associated with the target lesion in our study was similar to that reported in a previous study.
In our study, the restenosis ($20.5\%$ vs. $19.7\%$, $p \leq 0.999$) and reocclusion rates ($2.7\%$ vs. $2.8\%$, $p \leq 0.999$) at 1-year follow-up between the DCB and DES-only groups were similar. In the DCB-CTO study by Köln et al. [ 13], the restenosis rate was $11.8\%$ ($\frac{4}{34}$) and re-occlusion rate was $5.9\%$ with a mean follow-up of 8.6 ± 9.3 months ($\frac{2}{34}$). In another study on DCB treatment in de novo coronary lesions (including bifurcation lesions and CTO lesions) [12], the CTO lesion subgroup had a restenosis rate of $17\%$ ($\frac{2}{12}$) at a follow-up of 8.2 ± 4.0 months after PCI. Yang et al. [ 21] conducted a systematic review and meta-analysis comparing the use of BMS and DES in the treatment of coronary CTO lesions. The restenosis rates until 6 months after operation in the DES and re-occlusion groups were 14.21 and $3.95\%$, respectively. Our study further demonstrated that the patients with CTO lesions receiving less DES strategy have an acceptable restenosis rate in a long-term follow-up period.
One major benefit of DCB therapy for CTO lesions is the possibility of vessel remodeling over time. The LLL until 1-year angiographic follow-up in the DCB group was significantly less than that in the DES-only group (−0.08 ± 0.65 mm vs. 0.35 ± 0.62 mm, $p \leq 0.001$). One possible reason might be attributed to late lumen enlargement (LLE) occurring in the DCB group. In the DCB group, $60.7\%$ ($\frac{68}{112}$) had an enlarged MLD at the target vessel lesions on follow-up angiography. Scheller et al. [ 22] first reported LLE after DCB angiography of de novo coronary lesions and then published a study evaluating LLE after DCB angioplasty using QCA analysis [23]. The study consecutively included 58 de novo lesions that were treated with DCB. The average RVD was 2.58 ± 0.47 mm. After an average follow-up of 4.1 ± 2.1 months, the minimum lumen diameter of the target lesion increased from 1.75 ± 0.55 mm to 1.91 ± 0.55 mm ($p \leq 0.001$), LLE occurred in $69\%$ of patients. Several subsequent studies [24, 25] have observed the LLE phenomenon after DCB angioplasty for the treatment of de novo lesions. LLE associated with DCB angioplasty reduces the incidence of blood-limiting restenosis and has positive implications on patient prognosis as CTO lesions are characterized by negative remodeling. DCB has natural advantages over DES implantation, such as avoid ST risk, allowing the coronary artery to respond to vasomotor reflexes and vessel enlargement without concerns of small-sized DES during PCI and late or very late stent malapposition due to positive remodeling after recanalization.
Meticulous CTO-PCI operations and appropriate lesion preparation are important for reducing stent implantation in CTO revascularization [9]. Lesion preparation has been consistently highlighted in DCB angioplasty to achieve optimal results [26], particularly in CTO lesions. Noncompliant, scoring, and cutting balloons were used more frequently in the DCB group to acquire adequate blood flow and sufficient lumen gain. Non-conventional balloons, such as scoring and cutting balloons, can reduce irregular tears at the target lesions and avoid severe coronary dissection when performing DCB angioplasty. Sufficient lesion preparation can increase the contact area between the DCB surface and the intima, and local dissection without flow limitation contributes to the delivery of anti-proliferative drugs, which is a cause of late lumen enlargement during angiographic follow-up. It should be mentioned that bailout stents were not implanted in the lesions after predilation between 30 and $50\%$ stenosis with TIMI flow grade 3, which was not in accordance with the consensus rules [11]. The situation is mainly due to interventionists avoiding small-sized stent implantation in distal lesions to reduce the incidence of adverse events in the future.
A hybrid method combining DES and DCB is an acceptable option for the treatment of diffuse coronary diseases [27]. DCB as part of a hybrid revascularization technique with DES might be an option for stent length reduction in vessels with CTO lesions (particularly with diffuse disease after CTO restoration). It is well known that a longer stent length is associated with a higher incidence of MACE. Additionally, in the setting of recanalized CTO involving bifurcation or small vessel lesion, DCB provide an intervention option of “leave nothing behind” to avoid jailed ostial lesions or small caged vessel caused by DES. Each CTO is unique and our study provides a flexible and feasible strategy in CTO-PCI based on a combination of DCB and DES to reduce permanent stent length while maintaining the scaffolding properties of stents where they are required. Another potential benefit is the theoretically shorter DAPT duration in CTO lesions treated with DCB-only and a lower risk of bleeding events. Following DCB-only PCI, a 4 week DAPT was considered sufficient, rather than the 6–12 months advised after DES. This is particularly important for those who are at an increased risk of bleeding, as measured by the CRUSADE score, or scheduled for surgery shortly after PCI. Consequently, DCB has the potential to be used as an adjuvant or definitive treatment for CTO.
This study further accumulated clinical experience for DCB in CTO lesions; however, it has the following limitations. First, the current study was a nonrandomized observational study with potential bias, particularly because the intervention strategy was selected by the cardiovascular team. Interventionists may prefer DCB angioplasty for segments of the target lesion with favorable predilation results, and we cannot determine the exact proportion of lesions that cannot be treated with DCB. Second, we did not analyze the changes in ischemic symptoms before and after PCI and compared the improvements between the DCB-and DES-only groups. Furthermore, despite our analysis of restenosis and re-occlusion at 1 year, the population of angiographic follow-up varies, which could influence the results. Larger randomized controlled trials are necessary to evaluate DCB use in de novo CTO.
In conclusion, PCI with DCB is a potential “stent-less” therapy in treating de novo CTO lesions with satisfactory long-term clinical results compared to the DES-only approach if the result after preparation is satisfactory.
## Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.
## Ethics statement
The studies involving human participants were reviewed and approved by First Affiliated Hospital of Zhengzhou University Jincheng People’s Hospital, Jincheng, the Fifth Affiliated Hospital of Zhengzhou University. The patients/participants provided their written informed consent to participate in this study.
## Author contributions
CQ, WL, ZHa, XiW, and XY: conceptualization. XY, SP, XuW, PQ, YS, YZ, PZ, and ZQ: data curation. CQ and XY: formal analysis. CQ, and ZHa: funding and acquisition. CQ, WL, and ZHa: methodology. CQ and ZHu: project administration, resources and supervision. CQ, ZHu, GS, XQ, SP, and PZ: resources. XiW and XY: visualization. XiW and XY: writing original draft. All authors: investigation and writing, review and editing. All authors gave final approval of the manuscript and agreed to be accountable for all aspects of work ensuring integrity and accuracy.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcvm.2023.1045859/full#supplementary-material
## References
1. Azzalini L, Jolicoeur EM, Pighi M, Millan X, Picard F, Tadros VX. **Epidemiology, management strategies, and outcomes of patients with chronic total coronary occlusion**. *Am J Cardiol* (2016) **118** 1128-35. DOI: 10.1016/j.amjcard.2016.07.023
2. Tomasello SD, Boukhris M, Giubilato S, Marza F, Garbo R, Contegiacomo G. **Management strategies in patients affected by chronic total occlusions: results from the Italian registry of chronic total occlusions**. *Eur Heart J* (2015) **36** 3189-98. DOI: 10.1093/eurheartj/ehv450
3. Rossello X, Pujadas S, Serra A, Bajo E, Carreras F, Barros A. **Assessment of inducible myocardial ischemia, quality of life, and functional status after successful percutaneous revascularization in patients with chronic total coronary occlusion**. *Am J Cardiol* (2016) **117** 720-6. DOI: 10.1016/j.amjcard.2015.12.001
4. Jang WJ, Yang JH, Choi SH, Bin Song Y, Hahn JY, Choi JH. **Long-term survival benefit of revascularization compared with medical therapy in patients with coronary chronic total occlusion and well-developed collateral circulation**. *JACC Cardiovasc Interv* (2015) **8** 271-9. DOI: 10.1016/j.jcin.2014.10.010
5. Brilakis ES, Karmpaliotis D, Vo MN, Garcia S, Michalis L, Alaswad K. **Advances in the management of coronary chronic total occlusions**. *J Cardiovasc Transl Res* (2014) **7** 426-36. DOI: 10.1007/s12265-014-9556-6
6. Van den Branden BJL, Rahel BM, Laarman GJ, Slagboom T, Kelder JC, ten Berg JM. **Five-year clinical outcome after primary stenting of totally occluded native coronary arteries: a randomised comparison of bare metal stent implantation with sirolimus-eluting stent implantation for the treatment of total coronary occlusions (PRISON II study)**. *EuroIntervention* (2012) **7** 1189-96. DOI: 10.4244/eijv7i10a190
7. Nakazawa G, Vorpahl M, Finn AV, Narula J, Virmani R. **One step forward and two steps back with drug-eluting-stents from preventing restenosis to causing late thrombosis and nouveau atherosclerosis**. *JACC Cardiovasc Imag* (2009) **2** 625-8. DOI: 10.1016/j.jcmg.2009.01.011
8. Heeger CH, Busjahn A, Hildebrand L, Fenski M, Lesche F, Meincke F. **Delayed coverage of drug-eluting stents after interventional revascularisation of chronic total occlusions assessed by optical coherence tomography: the ALSTER-OCT-CTO registry**. *EuroIntervention* (2016) **11** 1004-12. DOI: 10.4244/eijy14m10_01
9. Alfonso F, Scheller B. **State of the art: balloon catheter technologies-drug-coated balloon**. *EuroIntervention* (2017) **13** 680-95. DOI: 10.4244/eij-d-17-00494
10. Neumann FJ, Chettibi M, Sisakia H, Metzler B, Ibrahimov F, Stelmashok VI. **2018 ESC/EACTS Guidelines on myocardial revascularization**. *Eur Heart J* (2019) **40** 87-165. DOI: 10.1093/eurheartj/ehy394
11. Jeger RV, Eccleshall S, Wan Ahmad WA, Ge J, Poerner TC, Shin ES. **Drug-coated balloons for coronary artery disease: third report of the international DCB consensus group**. *JACC Cardiovasc Interv* (2020) **13** 1391-402. DOI: 10.1016/j.jcin.2020.02.043
12. Onishi T, Onishi Y, Kobayashi I, Umezawa S, Niwa A. **Drug-coated balloon angioplasty for de novo small vessel disease including chronic total occlusion and bifurcation in real-world clinical practice**. *Cardiovasc Interv Ther* (2019) **34** 139-48. DOI: 10.1007/s12928-018-0534-9
13. Koln PJ, Scheller B, Liew HB, Rissanen TT, Ahmad WA, Weser R. **Treatment of chronic total occlusions in native coronary arteries by drug-coated balloons without stenting -a feasibility and safety study**. *Int J Cardiol* (2016) **225** 262-7. DOI: 10.1016/j.ijcard.2016.09.105
14. Wohrle J, Werner GS. **Paclitaxel-coated balloon with bare-metal stenting in patients with chronic total occlusions in native coronary arteries**. *Catheter Cardiovasc Interv* (2013) **81** 793-9. DOI: 10.1002/ccd.24409
15. Pan L, Lu W, Han Z, Pan S, Wang X, Shan Y. **Clinical outcomes of drug-coated balloon in coronary lesions: a real-world, all-comers study**. *Clin Res Cardiol* (2021a) **111** 732-41. DOI: 10.1007/s00392-021-01895-y
16. Pan L, Lu W, Han Z, Pan S, Wang X, Shan Y. **Clinical outcomes of drug-coated balloon in coronary patients with and without diabetes mellitus: a multicenter, propensity score study**. *J Diabetes Res* (2021b) **2021** 5495219-11. DOI: 10.1155/2021/5495219
17. Mintz GS, Popma JJ, Pichard AD, Kent KM, Satler LF, Chuang YC. **Patterns of calcification in coronary artery disease. A statistical analysis of intravascular ultrasound and coronary angiography in 1155 lesions**. *Circulation* (1995) **91** 1959-65. DOI: 10.1161/01.cir.91.7.1959
18. Thygesen K, Alpert JS, Jaffe AS, Chaitman BR, Bax JJ, Morrow DA. **Fourth universal definition of myocardial infarction (2018)**. *J Am Coll Cardiol* (2018) **72** 2231-64. DOI: 10.1016/j.jacc.2018.08.1038
19. Cutlip DE, Windecker S, Mehran R, Boam A, Cohen DJ, van Es GA. **Clinical end points in coronary stent trials: a case for standardized definitions**. *Circulation* (2007) **115** 2344-51. DOI: 10.1161/circulationaha.106.685313
20. Lee SW, Lee PH, Ahn JM, Park DW, Yun SC, Han S. **Randomized trial evaluating percutaneous coronary intervention for the treatment of chronic total occlusion the DECISION-CTO trial**. *Circulation* (2019) **139** 1674-83. DOI: 10.1161/circulationaha.118.031313
21. Yang SS, Tang L, Ge GG, Li RG, Qu XK, Fang WY. **Efficacy of drug-eluting stent for chronic total coronary occlusions at different follow-up duration: a systematic review and meta-analysis**. *Eur Rev Med Pharmacol Sci* (2015) **19** 1101-16. PMID: 25855938
22. Scheller B, Fischer D, Clever YP, Kleber FX, Speck U, Bohm M. **Treatment of a coronary bifurcation lesion with drug-coated balloons: lumen enlargement and plaque modification after 6 months**. *Clin Res Cardiol* (2013) **102** 469-72. DOI: 10.1007/s00392-013-0556-3
23. Kleber FX, Schulz A, Waliszewski M, Hauschild T, Boehm M, Dietz U. **Local paclitaxel induces late lumen enlargement in coronary arteries after balloon angioplasty**. *Clin Res Cardiol* (2015) **104** 217-25. DOI: 10.1007/s00392-014-0775-2
24. Her AY, Shin ES, Chung JH, Kim YH, Garg S, Lee JM. **Plaque modification and stabilization after paclitaxel-coated balloon treatment for de novo coronary lesions**. *Heart Vessel* (2019) **34** 1113-21. DOI: 10.1007/s00380-019-01346-9
25. Funatsu A, Nakamura S, Inoue N, Nanto S, Nakamura M, Iwabuchi M. **A multicenter randomized comparison of paclitaxel-coated balloon with plain balloon angioplasty in patients with small vessel disease**. *Clin Res Cardiol* (2017) **106** 824-32. DOI: 10.1007/s00392-017-1126-x
26. Konishi H, Habara M, Nasu K, Koshida R, Kinoshita Y, Tsuchikane E. **Impact of optimal preparation before drug-coated balloon dilatation for De novo lesion in patients with coronary artery disease**. *Cardiovasc Revasc Med* (2022) **35** 91-5. DOI: 10.1016/j.carrev.2021.03.012
27. Costopoulos C, Latib A, Naganuma T, Sticchi A, Figini F, Basavarajaiah S. **The role of drug-eluting balloons alone or in combination with drug-eluting stents in the treatment of**. *JACC Cardiovasc Interv* (2013) **6** 1153-9. DOI: 10.1016/j.jcin.2013.07.005
|
---
title: Preliminary study on Bioassay of Capparis spinosa L. seed extract and seed
germination
authors:
- Min Wang
- Xiaolu Yuan
- Liping Xu
journal: PeerJ
year: 2023
pmcid: PMC10022504
doi: 10.7717/peerj.15082
license: CC BY 4.0
---
# Preliminary study on Bioassay of Capparis spinosa L. seed extract and seed germination
## Abstract
The present study explored the germination inhibitors present in the seeds of *Capparis spinosa* L., a plant species that is known for its ecological significance in preventing wind erosion and fixing sand in desertified areas. Additionally, its roots, leaves, and fruits possess medicinal properties, and are used to treat a range of ailments such as rheumatism, tumors, and diabetes. However, the plant’s low germination rate under natural conditions is a major limitation. We aimed to improve the germination of C. spinosa seeds by investigating the effects of various infusions of caper seeds on the germination and seedling growth of Chinese cabbage seeds. A range of chemical reagents, hormonal immersions, and sand storage treatments were used to determine the differences in the germination rate of C. spinosa seeds. Our results revealed that among the various water extract concentrations tested, $100\%$ water extract exhibited the strongest inhibitory effect on the germination and growth of the cabbage seeds, with a germination rate of (70.00 ± 0.09)%. Furthermore, the inhibitory effects on the germination and growth of cabbage seeds were found to be strongest when treated with the extract solution 1, yielding a germination rate of (83.33 ± 0.02)%. Notably, the leaves of Chinese cabbage seedlings turned yellow-green and yellow after treatment with the extract solution. These findings highlight the potential inhibitory effects of C. spinosa seed extracts on seed germination and growth and suggest that further research is needed to better understand the underlying mechanisms. The results of the germination experiment with methanol extract showed a sharp decline in the germination rate of Chinese cabbage seeds treated with $50\%$ methanol extract, to (4.67 ± 0.02)%. These findings indicate the presence of germination-inhibiting substances in caper seeds. The highest germination potential was observed when the caper seeds were soaked in $30\%$ PEG, reaching $35.00\%$. The highest germination rate, $19.33\%$, was observed when the seeds were soaked in 250 mg/L GA3 and 25 mmol/L NaCl. These results suggest that the germination inhibitor present in caper seeds affects the germination of cabbage seeds as well. The highest germination rate was achieved when the seeds were soaked with gibberellin. It is hoped that the research on the germination-inhibiting substances in caper seeds will provide a scientific foundation for improving and refining the artificial propagation and cultivation methods of this species.
## Introduction
Capparis spinosa L. is a deciduous subshrub that can grow up to 30–50 cm in height, and it is characterized by its deep, sturdy root system and branches that can reach up to 3 m in length (Foschi et al., 2020; Yang et al., 2008). This plant is widely distributed in Mediterranean coastal countries and the Middle East, and can also be found in regions of China such as Gansu, Xinjiang, and Tibet (Inocencio et al., 2002; Wang, Zhang & Yin, 2016; Fici, 2001). C. spinosa has a range of medicinal and ecological benefits, including anti-rheumatic and liver-protective properties, as well as the ability to reduce soil erosion and resist wind and sand (Zhang & Hai, 2002; Rahnavard & Razavi, 2016). Additionally, the plant’s seeds can be consumed as a pickle or condiment (Mazandarani, Borhani & Fathiazad, 2014). The seeds have a high nutritional value, containing $26\%$ fiber and 19–$22\%$ protein, and they serve as a significant source of oil (Biouki, Khajahhosseini & Rad, 2020). However, in China, caper plants are mainly found in the wild and artificial breeding is relatively rare. The low natural germination rate and scarcity of existing populations has contributed to the plant’s relative scarcity and limited resources (Levizou, Drillas & Kyparissis, 2004; Khaninejad, Arefi & Kaf, 2012; Juan et al., 2020; Baskin & Baskin, 2014).
Seed dormancy constitutes a significant factor contributing to the low germination rate of capers. The causes of plant seed dormancy can be classified into two broad categories: external environmental factors, such as air, moisture, and temperature, and intrinsic seed-related factors. Intrinsic factors encompass the presence of germination-inhibiting substances within the species, physical barriers in the seed coat, and morphological and physiological immaturity of the embryo (Guo, 2016). Their presence obstructs the metabolic connection between seed physiological activities, thereby impacting the germination process (Li et al., 2011). Seed germination inhibitors are substances that can impede or delay the germination of both inter- and intraspecific seeds, which are classified into endogenous (organic acids, abscisic acid, aldehydes, phenols, etc. produced within the seed coat or other parts of the seed) and exogenous (produced by other plants) categories. Among these, endogenous inhibitory substances are more prevalent and exert a greater impact on seed germination (Li et al., 2020; Yan et al., 2014). Their presence obstructs the metabolic connection between seed physiological activities, thereby impacting the germination process (Jia et al., 2021). Studies have demonstrated that GA3 can significantly reduce the concentration of germination inhibitors and enhance seed germination (Guan, 1986). A sand storage treatment has also been shown to improve air permeability of thick seed coats and alleviate seed dormancy to some extent (Sun, Guo & Wei, 2012). Another factor contributing to dormancy is the degradation of the membrane system, leading to changes in cell membrane permeability during storage and inducing a dormant state in seeds (Peng et al., 2016; Zhang, Chen & Zhang, 2010). FeSO4, KNO3, and PEG-6000 have been shown to regulate membrane permeability, thus breaking seed dormancy and elevating germination rate (Meng & Gao, 2008; Lin et al., 2014; Niu et al., 2018; Chen et al., 2017). In this study, we aimed initially to examine the germination inhibitors of capers by exposing seeds to FeSO4, KNO3, PEG-6000, NaCl, and GA3 solutions and through sand storage methods to break dormancy and increase germination rate. Additionally, we investigated the impact of methanolic and aqueous extracts of the caper plant on the germination of Chinese cabbage seeds, contributing to a scientific understanding of this resource for sustainable development and utilization.
## Materials and Methods
The ripe fruits of *Capparis spinosa* L. were harvested from Karayagaqi Township in Yining County (44°10′N, 81°52′E, 1101.37 masl) in late July 2021. The caper seeds were stripped from the fruit and washed and stored at −20 °C in a refrigerator. The Chinese cabbage seeds were procured from Cangzhou Jinke Lifeng Seedling Co., Ltd. and stored at room temperature.
## Seed morphology observation
The caper seeds were dissected by making a longitudinal cut along the umbilical point using a scalpel. The morphological structure of these seeds was then observed under a stereomicroscope.
## Determination of water absorption rate of caper seeds
Two groups were established, an experimental group and a control group. In the experimental group, 50 seeds were subjected to sandpaper abrasion to break the seed coat and this procedure was repeated three times, totaling 150 seeds. Meanwhile, the seeds in the control group were left unaltered with their seed coats intact and replicated three times. Both groups of seeds were weighed and then soaked simultaneously in distilled water. At 2-h intervals, the seeds were drained of surface water using filter paper and weighed. The weight of the seeds was recorded every 2 h, until a point where no further weight gain was observed (Zhao et al., 2016).
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}${\rm Water\; absorption\; rate={\frac{wet\; weight\text{-}dry\; weight}{dry\; weight}}\times 100\%}$\end{document}Waterabsorptionrate=wetweight-dryweightdryweight×$100\%$
## Treatment of cabbage seeds with water and methanol extracts from caper seeds
Caper seeds (2 g) underwent three successive infusions, each performed with 20 mL of sterile distilled water and a 24-h soak time. The resulting extracts were obtained by filtration through filter paper, yielding extracts 1, 2, and 3. Five milliliters of each infusion was used to culture cabbage seeds (50 seeds per infusion treatment), which were incubated in an artificial climate incubator with a 16-h light regime and a temperature of 25 °C. The experiment was conducted in triplicate.
Caper seeds (2 g) were subjected to two extraction methods: one with sterile distilled water (15 mL) and the other with $80\%$ methanol (15 mL). The seeds were ground into a powder and soaked for 24 h at 25 °C prior to filtration, yielding extracts of 20 mL each. These extracts were then diluted to concentrations of $25\%$, $50\%$, $75\%$, and $100\%$ of their original strength, respectively, to form a series of dilutions. The distilled water served as a control for the first extraction method, while the $80\%$ methanol served as a control for the second extraction method.
The experiment involved two distinct extracts, each of which was prepared at four different concentrations. A total of 150 cabbage seeds were employed, with 50 seeds per concentration and three replicates per concentration. The seeds were washed, placed on Petri dishes lined with two layers of gauze (8 × 8 cm), and treated with 5 mL of the various extract concentrations. A control group was also included, resulting in a total of 30 Petri dishes used. The dishes were incubated in an artificial climate chamber with a 16-h photoperiod and a temperature of 25 °C (Luo, 2015). Germination was assessed on the 3rd day, and the root and shoot lengths of the cabbage seedlings were measured on the 5th day, with data collected 30 times.
## Treatment of seeds with chemical agents
Abraded caper seeds were subjected to a light sanding to thin their seed coat and enhance seed germination potential. The seeds underwent sterilization by soaking in $0.5\%$ sodium hypochlorite solution for 15 min, followed by three washes with sterile water. Six concentration gradients of FeSO4 ($0.1\%$, $0.3\%$, $0.5\%$, $0.7\%$, $0.9\%$, and $1.1\%$), KNO3 ($2\%$, $3\%$, $3.5\%$, $4\%$, $4.5\%$, and $5\%$), PEG ($10\%$, $15\%$, $20\%$, $25\%$, $30\%$, and $35\%$), and NaCl (10, 25, 50, 100, 150, and 200 mmol/L) were used to soak the caper seeds for 24 h at each gradient. The treatment was repeated three times, with 50 seeds per repetition.
The soaked seeds were then placed in Petri dishes lined with two layers of gauze (8 × 8 cm) and placed in an artificial climate incubator, with 14 h of light per day and a temperature of 30 °C during the day and 25 °C at night (Fang, Ye & Zhu, 2017; Lu, 2012; Luo et al., 2014; Zhang et al., 2009; Zheng et al., 2020).
## Treatment of seeds with exogenous hormones
The GA3 concentration was varied to six levels: 175, 200, 225, 250, 275 and 300 mg/L. The seed treatment protocol and soaking duration followed the same methodology as described in the previous chemical treatment (Luo et al., 2014).
## Seeds for sand storage
The seeds and sand were disinfected with $0.5\%$ sodium hypochlorite to ensure their sterilization. Subsequently, 50 seeds were placed in pockets of gauze and laid flat on top of the sand. The humidity of the sand was controlled to approximately $40\%$ by carefully manipulating it to form a ball, and loosening it when it became too dense. The seeds were subjected to a temperature of 4 °C or room temperature (25 °C) for 1, 7, 14, and 21 days. At the end of each storage period, the seeds were removed, washed with water, and placed in Petri dishes lined with gauze. The dishes were then incubated (Du, Li & Xue, 2013; Xue et al., 2019; Liu et al., 2022).
## Determination of germination indicators
Caper seed: The germination potential of the caper seeds was evaluated on the 7th day by counting the number of seeds that had sprouted. The germination rate was then determined on the 20th day by counting the number of seeds that had fully developed cotyledons. These were used as the criterion for successful germination.
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$${\rm Germinational\; potential=\frac{number\; of\; normal\; germinated\; seeds\; within\; 7d}{number\; of\; tested\; seeds}\times 100\% }$$\end{document}Germinationalpotential=numberofnormalgerminatedseedswithin7dnumberoftestedseeds×$100\%$ \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$${\rm Germination\; rate=\frac{number\; of\; normal\; germinated\; seeds\; within\; 20d}{number\; of\; tested\; seeds}\times 100\% }$$\end{document}Germinationrate=numberofnormalgerminatedseedswithin20dnumberoftestedseeds×$100\%$ Chinese cabbage seed: The germination rate of Chinese cabbage was counted on the 3rd day.
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$${\rm Germination\; rate=\frac{number\; of\; normal\; germinated\; seeds\; within\; 3d}{number\; of\; tested\; seeds}\times 100\% }$$\end{document}Germinationrate=numberofnormalgerminatedseedswithin3dnumberoftestedseeds×$100\%$
## Observational and statistical methods
In order to make a fair comparison, the control group was set up with seeds that were treated with only sterile distilled water or normal sand storage without any other treatments. The results of the treatments were then compared to the control group to see if there was a significant difference in the germination rate, germination potential, root length, and shoot length. The ANOVA method is a statistical tool that allows for the analysis of multiple variables and the determination of any significant differences between the groups. The results of the ANOVA analysis will help determine if the treatments had a significant effect on the germination rate, germination potential, root length, and shoot length of the seeds.
## Seed morphological characteristics
The stereo microscope was employed to examine the morphology of caper seeds (Figs. 1A and 1B). The seeds that had not developed fully exhibited a dull coloration and the kernels were observed to be desiccated (Fig. 1A). In contrast, mature seeds displayed a creamy white hue and had plump kernels (Fig. 1B).
**Figure 1:** *Abortive seed (A) and mature seed profile (B) (magnification: 10 × 2).*
## Comparison of before and after sanding seeds
The effect of light sandpapering on the seed coat was demonstrated in Fig. 2. Upon sandpapering, the inner layer of the seed coat became visible and appeared to be of a darker shade in comparison to the black surface layer.
**Figure 2:** *Comparison of seeds before and after sandpaper grinding (magnification: 10 × 0.75).*
## Seed water absorption curve
The water uptake of caper seeds with intact seed coats increased progressively from 2 to 16 h, and stabilized at $36.99\%$ after 16 h (Fig. 3). Conversely, the water absorption rate of caper seeds with damaged seed coats escalated gradually between 2 to 10 h and stabilized at $51.11\%$, a difference of $38.17\%$ higher than that of the intact seed coat. This suggests that the seed coat abrasions lead to an increased water absorption in caper seeds.
**Figure 3:** *Water absorption curve of caper seeds.*
## Effects of water extracts soaked with caper seeds at different concentrations and for different time periods on seed germination and seedling growth of Chinese cabbage
Table 1 highlighted the varied effect of two water extracts on the germination and growth of cabbage seeds. The treatment of cabbage seeds by four different concentrations of water extracts exhibited the strongest inhibition of germination and growth of cabbage seeds by $100\%$ water extracts with (70.00 ± 0.09)% germination rate; this was $21.35\%$ lower than the distilled water control; cabbage had the shortest root length and bud length. Three different time periods of the extracts whose effect on the germination and growth inhibition of cabbage seeds were as follows: extract 1 > extract 2 > extract 3.: extract 1 > extract 2 > extract 3. The lowest germination rate of cabbage seeds was (83.33 ± 0.02)% after the treatment with extract 1, which decreased by $6.37\%$. The inhibition of root and hypocotyl of cabbage seedlings by $75\%$, $100\%$ aqueous extract and extract 1 were significantly different ($P \leq 0.01$). The inhibitory effect of $50\%$ aqueous extract on cabbage seedling roots showed significantly difference ($P \leq 0.05$). The inhibition of cabbage seed germination by $100\%$ water infusion extract was highly significant difference ($P \leq 0.01$). After the cabbage seeds were treated with $75\%$ and $100\%$ water extracts, the cotyledons turned yellow.
**Table 1**
| Treatment | Chinese cabbage | Chinese cabbage.1 | Chinese cabbage.2 | Chinese cabbage.3 | Chinese cabbage.4 | Chinese cabbage.5 | Chinese cabbage.6 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Treatment | Germination rate (%) | P-value | Root length (cm) | P-value | Bud length (cm) | P-value | Cotyledon color |
| Controlcheck (distilled water) | 89.00 ± 0.05 | | 2.10 ± 0.71 | | 2.74 ± 0.69 | | Green |
| 25% water extract | 86.00 ± 0.02 | 0.65 | 1.74 ± 0.92 | 0.21 | 2.53 ± 0.49 | 0.99 | Yellow-green |
| 50% water extract | 80.67 ± 0.01 | 0.18 | 1.62 ± 0.91 | 0.09 | 2.29 ± 0.55* | 0.048 | Yellow-green |
| 75% water extract | 73.33 ± 0.11* | 0.02 | 1.35 ± 0.68** | 0.009 | 2.18 ± 0.45 | 0.99 | Yellow |
| 100% water extract | 70.00 ± 0.09** | 0.005 | 0.86 ± 0.48** | 0.00 | 1.82 ± 0.38 | 0.98 | Yellow |
| Extract 1 | 83.33 ± 0.02 | 0.36 | 1.34 ± 1.48** | 0.007 | 1.78 ± 0.38 | 0.97 | Yellow-green |
| Extract 2 | 86.67 ± 0.02 | 0.73 | 1.63 ± 1.20 | 0.10 | 2.60 ± 0.54 | 0.99 | Yellow-green |
| Extract 3 | 90.67 ± 0.04 | 0.73 | 1.82 ± 1.68 | 0.33 | 2.77 ± 0.66 | 0.99 | Green |
## Effects of different concentrations of methanol extracts of caper seeds on the germination of Chinese cabbage seeds
The results revealed a significant reduction in the germination rate of cabbage seeds following treatment with methanolic extracts of caper seeds (Table 2). It was observed that the $50\%$ methanolic extract had the most pronounced inhibitory effect on germination, with a rate of only 4.67 ± $0.02\%$—a decrease of $68.17\%$. The inhibitory effect of both $50\%$ and $100\%$ methanolic extracts on cabbage seed germination was found to be statistically significant ($P \leq 0.05$). Furthermore, treatment with different concentrations of methanolic extracts resulted in yellowing of the cabbage cotyledons.
**Table 2**
| Treatment | Unnamed: 1 | Chinese cabbage | Chinese cabbage.1 |
| --- | --- | --- | --- |
| Treatment | Germination rate (%) | P-value | Cotyledon color |
| Controlcheck (80% methanol solution) | 14.67 ± 0.05 | | Yellow |
| 25% methanol extract | 14.00 ± 0.06 | 0.86 | Yellow |
| 50% methanol extract | 4.67 ± 0.02* | 0.02 | Yellow |
| 75% methanol extract | 8.67 ± 0.01 | 0.14 | Yellow |
| 100% methanol extract | 6.00 ± 0.03* | 0.04 | Yellow |
## Effects of different concentrations of FeSO4 on seed germination
the highest germination potential of caper seeds was achieved at a concentration of $7.3\%$ FeSO4, reaching $32.00\%$ (Fig. 4A). Conversely, the highest germination rate was observed at a concentration of $0.5\%$ FeSO4, with a rate of $7.33\%$. The germination potential of seeds soaked in FeSO4 solutions ranging from $0.3\%$ to $0.9\%$ was found to be significantly different from the control ($P \leq 0.01$), while the germination potential of seeds soaked in $0.1\%$ and $1.1\%$ FeSO4 solutions showed a significant difference from the control, though to a lesser extent ($P \leq 0.05$). No significant differences were found in germination rate.
**Figure 4:** *Effect of different chemical reagents on caper seeds germination (A), FeSO4; (B), KNO3; (C), PEG; (D), NaCl.Asterisks (*, **) indicate significant at the level of 0.01 and 0.05, respectively.*
## Effects of different concentrations of KNO3 on the germination of caper seeds
As demonstrated in Fig. 4B, the highest germination potential and rate were both achieved at a concentration of $3.5\%$ KNO3, with a potential of $29.33\%$ and a rate of $12.00\%$. Seeds treated with KNO3 solutions at concentrations ranging from $3\%$ to $4.5\%$ showed highly significant differences in germination potential compared to the control ($P \leq 0.01$). The germination potential of seeds treated with $2\%$ and $5\%$ KNO3 was also found to be significantly different from the control ($P \leq 0.05$). However, no significant differences were noted in germination rate.
## Effects of different concentrations of PEG on seed germination of caper
The germination potential and rate of caper seeds soaked in different concentrations of PEG were found to vary (Fig. 4C). The highest germination potential was observed at a concentration of $30\%$ PEG, with a germination potential of $34.67\%$. Meanwhile, the highest germination rate was recorded at a concentration of $25\%$ PEG, reaching $18.67\%$. Compared to the control, significant differences were observed in germination potential for treatments with $15\%$, $20\%$, $25\%$, $30\%$, and $35\%$ PEG ($P \leq 0.01$), while a significant difference was observed in germination potential for the treatment with $10\%$ PEG ($P \leq 0.05$). Additionally, a highly significant difference was observed in the germination rate for the $25\%$ PEG treatment ($P \leq 0.01$).
## Effects of different concentrations of NaCl on seed germination
The germination potential of caper seeds was found to reach $30.00\%$ at a NaCl concentration of 50 mmol/L and the germination rate was highest, at $19.33\%$, when the NaCl concentration was 25 mmol/L (Fig. 4D). These results indicate that there were very significant differences in germination potential ($P \leq 0.01$) when compared to the control group at a NaCl concentration of 50 mmol/L, while the germination rate was found to be highly significantly different ($P \leq 0.01$) at a concentration of 25 mmol/L.
## Effects of different concentrations of GA3 on seed germination
Caper seed germination was positively influenced by increasing concentrations of GA3, with a maximum germination potential of $30.67\%$ observed at a concentration of 200 mg/L and a maximum germination rate of $19.33\%$ observed at a concentration of 250 mg/L (Fig. 5). Compared to the control, there was a very significant difference in germination potential ($P \leq 0.01$) when the GA3 concentration was 200 mg/L, while concentrations of 175 and 225 mg/L showed significant differences in germination potential ($P \leq 0.05$). Significant differences in germination rate ($P \leq 0.05$) were also observed for GA3 concentrations of 225 and 250 mg/L.
**Figure 5:** *Effects of different concentrations of GA3 on seed germination of caper.Asterisks (*, **) indicate significant at the level of 0.01 and 0.05, respectively.*
## Effects of different days in sand storage on seed germination
The results of the sand storage treatment on caper seed germination rate and potential showed a gradual increase over time (Fig. 6). Compared with room temperature, the treatment effect of 4 °C was better. The optimal treatment conditions were determined to be a storage time of 21 days at 4 °C, which resulted in the highest germination rate ($17.67\%$) and potential ($26.67\%$) observed. Significant differences in germination rate and potential were observed between the control and treatment conditions of 14 and 21 days at 4 °C ($P \leq 0.01$). The germination potential also showed a significant difference at these conditions ($P \leq 0.05$).
**Figure 6:** *Effects of different sand storage time and temperature on seed germination of caper.Asterisks (*, **) indicate significant at the level of 0.01 and 0.05, respectively.*
## Discussion
Morphological anatomical observations of caper seeds have revealed instances of seed abortion, which negatively impacts seed germination (Willis et al., 2014). There are numerous factors that influence seed germination, including the thickness and structure of the seed coat (Greipsson, 2001; Zhang, 2021). A dense seed coat structure with well-developed palisade tissue can result in poor water and air permeability, causing seed dormancy (Vazquez-Yanes, 1976; Zhu et al., 2022). Results from the caper seed water absorption experiment showed that grinding the seed coat increased the water absorption rate by $38.17\%$ compared to the unground seed coat, highlighting the presence of a water absorption barrier in the seed coat. Lin et al. [ 2016] found that the water uptake rate of caper seeds in the group with the digestive tract of *Teratoscincus scincus* was consistently higher than the control group, indicating that the digestive tract of *Teratoscincus scincus* promotes seed water uptake. One of the primary mechanisms through which fruit-eating animals enhance seed germination is the abrasive and corrosive effects of the digestive tract on the seed coat, which enhances seed permeability to water and gas. Xiao et al. [ 2017] investigated whether the digestive tract of *Hemiechinus auritus* impacted the water uptake and germination of *Capparis spinosa* seeds and found that the digestive tract increased the seed’s water uptake rate and absorption capacity. The germination rate of seeds soaked in GA3 solution and with the digestive tract of H. auritus was significantly higher than the control group, with extended germination days, suggesting that the digestive tract of H. auritus improves seed permeability to water and disrupts mechanical obstructions, promoting seed germination. This result aligns with the present study’s findings, indicating that the caper seed coat does affect its germination rate.
It has been established that the presence of germination inhibitors in seeds can contribute to seed dormancy (Crocker, 1916). The existence of physiological barriers, particularly in fully developed seed embryos, further complicates the germination process (Gou, Shen & Shi, 2019). The study of germination inhibitors in caper seeds revealed that the aqueous extract of seed powder had a more pronounced inhibitory effect on the growth of cabbage seedlings, particularly in regards to root growth, indicating the presence of germination inhibitors in caper seeds. This result was confirmed by the observation that high concentrations of the aqueous extract could lead to yellowing of the cotyledons of cabbage seedlings, potentially due to inhibition of chlorophyll synthesis.
The results of our study indicate that, compared to the aqueous extract, the methanolic extract displayed a more potent inhibitory effect on the germination and growth of cabbage seeds. Our findings suggest that the germination inhibitor present in caper is more soluble in methanolic solution, as evidenced by the lowest germination rate observed in cabbage seeds treated with the methanolic extract. This is consistent with the findings of Sun & Ji [2009], who found that extracts with varying concentrations of $80\%$ methanol had a greater impact on the germination rate of Tilia seeds compared to water extracts. Our findings suggest that the germination inhibitory substances in the methanolic solution are higher compared to other extracts. This is in contrast to the findings of Shao et al. [ 2018], who found that the water extract of Polygonatum seeds exhibited higher inhibitory activity, while the ethyl acetate extract had no significant impact on cabbage seedling germination. These results suggest that the germination inhibitor in *Polygonatum sibiricum* seeds is water-soluble. In this experiment, two controls were established by soaking cabbage seeds in distilled water and $80\%$ methanol solution, with the latter showing a significantly lower germination rate (14.67 ± 0.05) of $83.52\%$ compared to the distilled water control. The inhibitory effects of methanol on seed germination have been well documented, with studies such as those by Zhang [2015] and Meng, Ji & Shao [2017] indicating that aging maize and cotton seeds with $50\%$ methanol respectively, resulted in reduced germination rates and seedling growth. It is speculated that the methanolic extract of caper seeds exhibits a significant inhibitory effect on the germination of cabbage seeds due to the combination of its effects on seed aging and the greater solubility of the germination inhibitor in methanol.
In this study, three methods including the use of chemical reagents, exogenous hormones, and sand storage were employed to treat caper seeds. The highest germination potential ($35\%$) was observed when the caper seeds were soaked in $30\%$ PEG, while the highest germination rate ($19\%$) was achieved when the seeds were treated with 250 mg/L GA3 and 25 mmol/L NaCl. Previous reports have indicated that soaking caper seeds in GA3 can significantly enhance the germination rate over an extended period. When sulfuric acid was combined with GA3, the highest germination rate ($60\%$) was obtained through the sequential application of a 30-min sulfric acid soak followed by GA3 solution immersion (Sottile et al., 2021). Zhang et al. [ 2009] investigated methods to improve the germination rate of prickly mandarin, and the highest germination rate of caper seeds ($16\%$) was recorded when the concentration of PEG reached $25\%$. Sun & Ma [2010] found that the highest germination rate (700 mg/L GA3) was achieved by soaking the seeds in concentrated H2SO4 for 70 min and treating them with H2O2 for 4 h. In the present study, however, the highest germination rate observed was only $19\%$ after GA3 treatment. Such variations could be attributed to the different origins of the seeds and the environmental factors, including climate and genetic variations, at their place of origin. The germination rate and potential of caper seeds exhibited a trend of first increasing and then decreasing with the soaking in varying concentrations of NaCl solution, demonstrating the salt tolerance of these seeds. Sadeghi & Rostami [2016] investigated the effect of salt stress on prickly mountain citrus and found no significant decrease in the fresh weight of the plants or in root tolerance under salt stress. Furthermore, the synthesis of abscisic acid was observed to increase with increasing salinity. This plant has been reported to grow well in poor soils and is known to thrive in sandy soils with low organic matter content, and has a strong resistance to salinity (Yazdani Biouki, Khajahhosseini & Rad, 2021).
During the germination of caper seeds, a high incidence of mold was observed, which may have contributed to the low germination rate recorded in this study. Lian et al. [ 2021] showed that pepper seeds scalded for 15 min at temperatures below 50 °C had a high rate of mildew and poor disinfection. On the other hand, scalding at temperatures above 50 °C resulted in a significant reduction in both germination rate and potential. In a study conducted by Yuan [2012], the impact of different scalding conditions on Mungbean sprout growth was evaluated through measurement of water absorption rate, germination rate, and sprouting bean mass ratio. The results indicated that scalding at 55 °C for 20–30 min produced the optimal results, with a germination rate of $100\%$ and a sprouting bean mass ratio of over 4.81, leading to optimal sprout growth. It is worth noting that caper seeds have a thick seed coat, and scalding can be utilized to reduce the occurrence of mold and improve germination rate.
## Conclusions
This study provided initial evidence of the presence of seed germination inhibitors in caper seeds. The extracts from caper seeds exhibited an inhibitory effect on the germination of cabbage seeds, with methanolic extracts demonstrating the strongest inhibition. The maximum germination rate of $19.33\%$ was observed in seeds treated with a 250 mg/L GA3 and 25 mmol/L NaCl solution. Further investigations into the structural identification and content analysis of these germination inhibitors would shed light on the underlying mechanism of seed dormancy in capers, offering a theoretical foundation for future research.
## References
1. Baskin CC, Baskin JM. **Seeds: ecology, biogeography, and evolution of dormancy and germination**. *Crop Science* (2014) **40** 564. DOI: 10.2135/cropsci2000.0009br
2. Biouki RY, Khajahhosseini S, Rad MH. **The potential of using medicinal herbs in haloculture, a case study of the caper plant (**. *Acta Horticulturae* (2020) **5** 9-14. DOI: 10.17660/ActaHortic.2020.1287.2
3. Chen ZP, Peng MW, Chang YL, Jiang P. **Effects of drought stress intimated by PEG on germination characteristics of**. *Anhui Agriculture Science* (2017) **23** 97-100
4. Crocker W. **Mechanics of dormancy in seeds**. *American Journal of Botany* (1916) **3** 99-120. DOI: 10.2307/2435202
5. Du JH, Li BQ, Xue QX. **Effect of NaCl seed soaking on germination of rice**. *Chinese Agricultural Science Bulletin* (2013) **29** 33-35. DOI: 10.3969/j.issn.1000-6850.2013.03.007
6. Fang HL, Ye JS, Zhu PL. **Characteristics of substances inhibiting the germination of germinating seeds of**. *Jiangsu Agricultural Sciences* (2017) **45** 138-140. DOI: 10.15889/j.issn.1002-1302.2017.01.039
7. Fici S. **Intraspecific variation and evolutionary trends in**. *Plant Systematics and Evolution* (2001) **228** 123-141. DOI: 10.1007/s006060170024
8. Foschi ML, Juan M, Pascual B, Pascual-Seva N. **Water uptake and germination of caper (**. *Agronomy* (2020) **10** 838. DOI: 10.3390/agronomy10060838
9. Gou CC, Shen YB, Shi FH. **Review on seed dormancy of**. *Journal of Nanjing Forestry University (Natural Science Edition)* (2019) **43** 175-183. DOI: 10.3969/j.issn.1000-2006.201710045
10. Greipsson S. **Effects of stratification and GA**. *Seed Science & Technology* (2001) **29** 1-10. DOI: 10.1109/ICALT.2001.943879
11. Guan KL. **A review on seed dormancy characters of trees**. *Journal of Zhejiang Forestry College* (1986) **3** 99-109
12. Guo LP. *Study on dormancy and dormancy breaking of tree peony seeds* (2016)
13. Inocencio C, Alcaraz F, Calderon F, Obon C, Rivera D. **The use of floral characters in Capparis sect. Capparis’ to determine the botanical and geographical origin of capers**. *European Food Research and Technology* (2002) **214** 335-339. DOI: 10.1007/s00217-001-0465-y
14. Jia X, Zhang MQ, Li Y, Xue Y, Zhang Y, Yu J, Wang XQ. **Study on the activity of seed germination inhibitors of**. *Journal of Chinese Medicinal Materials* (2021) **11** 2521-2525. DOI: 10.13863/j.issn1001-4454.2021.11.003
15. Juan M, Pascual SN, Iranzo D, Pascual B. **Improvement of seed germination of caper (**. *Acta Horticulturae* (2020) **26** 433-440. DOI: 10.17660/actahortic.2020.1273.56
16. Khaninejad S, Arefi IH, Kaf M. **Effect of priming on dormancy breaking and seedling establishment of caper (**. *International Conference on Applied Life Sciences* (2012) **10** 365-370
17. Levizou E, Drillas P, Kyparissis A. **Exceptional photosynthetic performance of**. *Photosynthetica* (2004) **42** 229-235. DOI: 10.1023/B:PHOT.0000040594.85407.f4
18. Li X, Dong CM, Xing B, Li X. **Study on the characteristics of seed germination inhibitors of**. *Seed* (2020) **39** 108-110. DOI: 10.16590/j.cnki.1001-4705.2020.06.108
19. Li SX, Liu JJ, Tian SX, Yin TM. **Study on the causes of**. *Journal of Nanjing Forestry University (Natural Science Edition)* (2011) **35** 1-4. DOI: 10.3969/j.issn.1000-2006.2011.05.001
20. Lian YY, Li HY, Li HX, Han B, Jiang W. **Effects of different treatments on seed germination of pepper**. *Jiangsu Agricultural Sciences* (2021) **49** 132-135. DOI: 10.15889/j.issn.1002-1302.2021.10.024
21. Lin JX, Shao S, Sui D, Mu CS, Wang YC, Wang Y. **Comparison of several methods for improving the germination rate of**. *Chinese Journal of Grassland* (2014) **3** 47-50
22. Lin YY, Xiao LR, Ma XY, Shi L. **Effects of Turpan wonder gecko’s digestive tract on water uptake and germination of caper seeds**. *Pratacultural Science* (2016) **33** 1764-1769. DOI: 10.11829/j.issn.1001-0629.2015-0753
23. Liu XZ, Li JF, Xu WM, Ge K. **Effects of different soaking temperature and sand-storage depth on seed germination of**. *Shanxi Forestry Science and Technology* (2022) **51** 40-41. DOI: 10.19989/j.cnki.1007-726X.20220212
24. Lu FG. **Effects of different concentrations of ferrous sulfate on yield and quality of pea seedlings**. *Northern Horticulture* (2012) **8** 46-48
25. Luo J. *The impact of high temperature stress on physiological-biochemical characteristic of Capparis spinosa L. seedlings and screening of the differentially expressed genes* (2015)
26. Luo J, Yang LJ, Xi LQ, Ma CH. **Study on seed germination of the Capparis spinosa**. *Seed* (2014) **33** 5-9. DOI: 10.16590/j.cnki.1001-4705.2014.01.070
27. Mazandarani M, Borhani G, Fathiazad F. **Phytochemical analysis, antioxidant activity and ecological requirements of**. *Journal of Medicinal Plants and By-Products* (2014) **3** 21-26
28. Meng ZJ, Gao Y. *Effects of polyethylene Glycol (PEG)-simulated drought stress on Chamecytisus palmensis seed germination* (2008) 134-136
29. Meng CM, Ji JH, Shao DY. **Effect of aging treatment with methanol on cotton seed germination and seedling growth**. *Seed* (2017) **36** 10-11 15. DOI: 10.16590/j.cnki.1001-4705.2017.03.010
30. Niu XX, Mou M, Li BH, Dong XH. **Physiological mechanism of FeSO**. *Chinese Journal of Eco-Agriculture* (2018) **26** 1828-1835. DOI: 10.13930/j.cnki.cjea.180257
31. Peng YS, Li TH, Wen SZ, He GX, Huang SJ. **Effect of different sand-storage methods on the germination of**. *Journal of Central South University of Forestry & Technology* (2016) **36** 44-48. DOI: 10.14067/j.cnki.1673-923x.2016.08.009
32. Rahnavard R, Razavi N. **A review on the medical effects of**. *Advanced Herbal Medicine* (2016) **2** 44-53
33. Sadeghi H, Rostami L. **Evaluating the physiological and hormonal responses of caper plant (**. *Desert* (2016) **21** 49-55. DOI: 10.22059/JDESERT.2016.58317
34. Shao C, Liu JY, Liu N, Zhang YY. **Preliminary study on inhibitors of**. *Tianjin Agricultural Sciences* (2018) **24** 16-19. DOI: 10.3969/j.issn.1006-6500.2018.03.005
35. Sottile F, Caltagirone C, Peano C, Signore MBD, Barone E. **Can the caper (**. *Horticulturae* (2021) **7** 316. DOI: 10.3390/HORTICULTURAE7090316
36. Sun J, Guo JF, Wei SN. **Overview on inhibitors of plant seed germination**. *Seed* (2012) **31** 57-61. DOI: 10.16590/j.cnki.1001-4705.2012.04.044
37. Sun B, Ji SQ. **Inhibitor of germination for seeds of**. *Protection Forest Science and Technology* (2009) **4** 34-35+88. DOI: 10.13601/j.issn.1005-5215.2009.04.046
38. Sun QH, Ma M. **Effects of sulphuric acid and gibberellins’ on seed germination of**. *Journal of Shihezi University (Natural Science)* (2010) **28** 144-146. DOI: 10.13880/j.cnki.65-1174/n.2010.02.011
39. Vazquez-Yanes C. **Seed dormancy and germination in secondary vegetation tropical plants: the role of light**. *Comparative Physiology and Ecology* (1976) **1** 30-34
40. Wang Q, Zhang ML, Yin LK. **Genetic diversity and population differentiation of**. *Biochemical Systematics and Ecology* (2016) **66** 1-7. DOI: 10.1016/j.bse.2016.02.034
41. Willis CG, Baskin CC, Baskin JM, Auld JR, Venable DL, Cavender-Bares J, Donohue K, Casas RRD. **The evolution of seed dormancy: environmental cues, evolutionary hubs, and diversification of the seed plants**. *New Phytologist* (2014) **203** 300-309. DOI: 10.1111/nph.12782
42. Xiao LR, Lin YY, Liang T, Shi L. **Effects of passage through the digestive tract of long-eared hedgehog on caper seed water uptake and germination**. *Chinese Journal of Ecology* (2017) **36** 995-1001. DOI: 10.13292/j.1000-4890.201704.032
43. Xue JQ, Alami MM, Wang XK, Liu SB, Gong DL, Wang XS. **Research on seed characteristics of**. *Hubei Agriculture Sciences* (2019) **58** 134-137+157. DOI: 10.14088/j.cnki.issn0439-8114.2019.22.031
44. Yan S, Luo GM, Zhu YY, Dong YK, Li L, Wang YY, Zhou Q, Wang XY. **Optimization of seed germination and detection of germination inhibitors in gardenia fruits**. *Journal of Southern Agriculture* (2014) **45** 376-382. DOI: 10.3969/j.issn.2095-1191.2014.3.376
45. Yang T, Liu YQ, Wang CH, Wang ZT. **Interventional effects of traditional Chinese medicine on Alzheimer disease**. *China Journal of Chinese Materia Medica* (2008) **21** 2453-2458
46. Yazdani Biouki R, Khajahhosseini S, Rad MH. **The potential of using medicinal herbs in haloculture, a case study of the caper plant (**. *Acta Horticulturae* (2021) **1287** 9-13. DOI: 10.17660/ActaHortic.2020.1287.2
47. Yuan XY. **Effects of scalding condition on germinating and growth of mungbean sprout**. *Science Technology and Engineering* (2012) **12** 7350-7353. DOI: 10.3969/j.issn.1671-1815.2012.28.042
48. Zhang HY. **Effect of aging treatment with methanol on maize seed germination and seedling growth**. *Journal of Maize Sciences* (2015) **23** 76-79+85. DOI: 10.13597/j.cnki.maize.science.20150313
49. Zhang XY. *Study on release method of seed anatomy and dormancy of Chionanthus retus* (2021)
50. Zhang FY, Chen CY, Zhang P. **Effects of Ethephon and Gbberellin on seed germinating ability and seeding growth of**. *Journal of Anhui Agriculture Science* (2010) **38** 2304-2305. DOI: 10.13989/j.cnki.0517-6611.2010.05.123
51. Zhang XY, Du RB, Fan XM, Zhao BL. **Germination test of**. *Journal of Anhui Agriculture* (2009) **37** 12008-12009. DOI: 10.13989/j.cnki.0517-6611.2009.25.105
52. Zhang LY, Hai Y. **Plant communities excluded in the book of “The vegetation and its utilization in Xinjiang”: I.the desert plant communities**. *Arid Land Geography* (2002) **25** 84-89
53. Zhao D, Jin H, Song PF, Zou JX, Jiang GB, Li QY. **Study on method for breaking dormancy for**. *Northern Gardening* (2016) **12** 16-20
54. Zheng Q, Zhou YT, Zhou RM, Deng LN, Jiang W. **Study on the effect of ferrous sulfate immersion on germination and germination characteristics of**. *Anhui Agricultural Science Bulletin* (2020) **26** 88-90
55. Zhu XH, Wang HH, Lin FX, Yang ZY. **Comparison of seedling rate of**. *Journal of Guangdong Agriculture Industry Business Polytechnic* (2022) **38** 74-76
|
---
title: 'Changes in Cardiovascular Health during Young Adulthood and Subclinical Atherosclerosis
in Middle Age: The CARDIA Study'
authors:
- Xiaomin Ye
- Zhenyu Xiong
- Jiaying Li
- Yifen Lin
- Peihan Xie
- Xiangbin Zhong
- Rihua Huang
- Xiaodong Zhuang
- Xinxue Liao
journal: Global Heart
year: 2023
pmcid: PMC10022532
doi: 10.5334/gh.1179
license: CC BY 4.0
---
# Changes in Cardiovascular Health during Young Adulthood and Subclinical Atherosclerosis in Middle Age: The CARDIA Study
## Abstract
### Background and aims:
The benefits of reaching ideal cardiovascular health (CVH) are well known, but it is unclear whether positive CVH changes from young adulthood to middle age reduce subclinical atherosclerosis risk. This study examined associations of changes in CVH from young adulthood to middle age and CVH in young adulthood with subclinical atherosclerosis.
### Methods:
Data was analyzed from the Coronary Artery Risk Development in Young Adults (CARDIA) study. CVH was examined at years 0 and 20 using Life Simple 7 metrics from AHA guideline. Coronary artery calcium (CAC) was identified at years 20 and 25. Carotid intima-media thickness (IMT) was identified at year 20.
### Results:
Among 2,935 participants ($56.2\%$ women, $46.7\%$ black), the change of CVH score was –1.26 (2.13). For per 1-unit increase in CVH at baseline, the adjusted odds ratios (ORs) of presence of CAC and IMT were 0.81 ($95\%$ CI 0.78, 0.86) and 0.85 ($95\%$ CI 0.76, 0.94), respectively. For per 1-unit increase in CVH changes, the adjusted ORs of CAC and IMT were 0.86 ($95\%$ CI 0.82, 0.90) and 0.81 ($95\%$ CI 0.73, 0.90). Compared with stable moderate CVH, improvement from moderate to high was associated with a lower risk of CAC (0.64 [$95\%$ CI 0.43, 0.96]), while retrogression from moderate to low was associated with a higher risk of CAC (1.45 [$95\%$ CI 1.19, 1.76]).
### Conclusions:
Positive changes of CVH during young adulthood are associated with negative subclinical atherosclerosis risk in middle age, indicating the importance of reaching an ideal cardiovascular health status through young adulthood.
## Introduction
Life’s Simple 7 metrics, a concept of cardiovascular health (CVH) proposed by the American Heart Association (AHA), is assessed with four metabolic risk factors (body mass index, blood pressure, blood glucose, total cholesterol) and three healthy lifestyles (diet, physical activity, smoking) [1]. They have been shown to be important cardiovascular risk factors, participants of which mostly were chosen from the elder population in current studies [234]. However, not enough attention was paid to the younger population or subclinical cardiovascular events, such as coronary artery calcium (CAC) and intima-media thickness (IMT), broadly accepted as marker and maker of cardiovascular events [5678]. Considering components of cardiovascular health are time-varying variables, associations of changes in CVH and subclinical atherosclerosis remain uncertain. To find out association between CVH scores and CAC is critical in preventing atherosclerosis risk in young adulthood.
This study used the Coronary Artery Risk Development in Young Adults (CARDIA) study with the aim of quantifying the association between CVH and its changes through young adulthood with the incidence of CAC and IMT in midlife, and we hypothesized that positive CVH changes from young adulthood to middle age would reduce subclinical atherosclerosis risk.
## Data and material disclosure statement
Data documentation for CARDIA is publicly available online (cardia.dopm.uab.edu.). Data used in this article are available on reasonable request from the CARDIA coordinating Center.
## Study population
Details of the CARDIA study design and examinations have been reported previously [9]. The CARDIA study is a longitudinal cohort study which began in 1985–1986 (Y0). A cohort sample of 5,115 healthy participants were recruited at four US sites: Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California, and were balanced for gender, race, age (18–24 years and 25–30 years), and educational level (high school or less education and greater than high school education). Seven follow-up examinations of the cohort were conducted at years 2, 5, 7, 15, 20 (Y20), and 25. At each study, all participants provided written informed consent and the institutional review boards at each study site and coordinating center have granted approval annually for all examinations.
The flow chart of this study is presented in supplementary figure 3. Of the initial 5,115 recruited participants at baseline, 1,587 did not return at Y20 and 593 had missing CVH data or covariates data, leaving 2,935 participants included in the basic analysis. For analysis of CAC, 123 participants who had not taken CT scan at either Y20 or Y25 were excluded, while for analysis of IMT 244 participants who had not taken carotid ultrasound at Y20 were excluded, leaving 2,812 participants included in the analysis of CAC and 2,691 participants included in the analysis of IMT.
## Cardiovascular health
CVH, consistent with Life’s Simple 7 according to AHA recommendation, consists of four components of metabolic risk: blood pressure, blood glucose, total cholesterol, and body mass index, and three components of lifestyle: smoking, physical activity, and diet. All metrics were measured at baseline and Y20, obtained by trained and certified technicians. Blood pressure was measured three times after five minutes rest at seated and recorded as the average of second and third-time measurement. Participants were asked to fast for 12 hours before venous blood was drawn for blood glucose and total cholesterol analysis. Body mass index was calculated as weight in kilograms divided by height in meters, squared. Cigarette smoking history was self-reported. Physical activity scoring was assessed from a physical activity questionnaire and calculated with the frequency and duration of 13 kinds of activities. Dietary intakes were assessed from interviewer-administrated, quantitative food history [10]. Medication use for controlling blood pressure, cholesterol, or glucose was self-reported. Covariates such as age, gender, race, education level, and history of disease were self-reported by questionnaire.
CVH scores were calculated according to prior studies [111213] (Table 1). According to prior study [11], we chose $40\%$ as a cut-off point for physical activity. Participants with top $40\%$ physical activity were given two points, and participants with lowest $20\%$ were given zero points, the rest were given one point. Dietary intake measure in CARDIA is most useful for ranking individuals based on one-month consumption rather than absolutely quantifying food intake [14]. Therefore, we adjusted the dietary score based on traditional AHA standard and calculated separately for each sex with four food sources: potassium(mg), calcium(mg), fiber(g), and saturated fat(g). A higher score was assigned to participants with higher intake of potassium, calcium, and fiber and a lower intake of saturated fat (High = 5 to low = 1). Four scores summed up to a total score ranging from 4 to 20. The highest $40\%$ get two points. This method was proven to be in concordance with other dietary eating measure patterns and the traditional AHA standard [1112]. Based on CVH scores, participants were assigned to 3 classes: high (12–14 points), moderate (8–11 points), and low (0–7 points). CVH scores at baseline subtracted from those at Y20 are the change of CVH scores from baseline to Y20.
**Table 1**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 |
| --- | --- | --- | --- |
| | CARDIOVASCULAR HEALTH | CARDIOVASCULAR HEALTH | CARDIOVASCULAR HEALTH |
| | IDEAL = 2 POINTS | MODERATE = 1 POINT | POOR = 0 POINT |
| BMI, kg/m2 | <25 | 25–29.9 | >30 |
| Smoking | Never, quit over 1year | Former, quit less than 1 year | Current |
| Diet | Top 40% | Second 40% | Lowest 20% |
| Blood pressure, mmHg | <120/80 without medication | Systolic 120–139Diastolic 80–89Treated <120/80 | >140/90 |
| Total cholesterol, mg/dL | <200 without medication | 200–239Treated <200 | >=240 |
| Fasting glucose, mg/dL | <100 without medication | 100–125Treated <100 | >=126 |
| Physical activity | Top 40% | Second 40% | Lowest 20% |
## Coronary artery calcium
The extent of CAC was quantified at years 20 and 25 using a standardized protocol [15]. At Y20, two sequential CT scans were provided using either a cardiac-gated electron beam CT scanner or a multidetector CT system. At Y25, only a single CT scan was performed, using a 64-channel multidetector CT system. Prior study suggested that two CT scans were comparable [16]. Computed tomography images were read by experienced technicians in a central reading center. Total CAC score was obtained by Agatston method [17], summing up all lesions within a given artery and across all arteries (left main, left anterior descending, left circumflex, and right coronary artery). A CAC score over zero at either Y20 or Y25 was defined as the presence of CAC.
## Intima-media thickness
Images of distal common carotid artery, the carotid bulb, and the proximal internal carotid artery on both sides were obtained by high-resolution B-mode ultrasonography using a standardized protocol at Y20. Intima-media thickness was calculated from the average of the mean intima-media thickness for the internal, bulb, and common carotid near and far walls of the right and left sides [18]. Carotid intima-media thickness over one was defined as a presence of the thickening of carotid intima-media thickness, which is abnormal IMT.
## Statistical analyses
Continuous data are expressed as means ± SD and categorical variables are presented as counts and proportions. Multivariable adjusted logistic regression analyses were conducted to estimate the association of baseline CVH and changes and presence of CAC and abnormal IMT. Multivariable adjusted linear regression analyses were used to estimate the association between changes of CVH and IMT. We adjusted for demographic variables (age at baseline, gender, race, and education), current drinker, history of hypertension, history of diabetes. We also performed subgroup analysis based on gender, race, alcohol use, history of hypertension, history of diabetes. We further adjusted for family income at Y5 to measure the socio-economic status affect.
A 2-sided P value of <0.05 was considered statistically significant. All analyses were performed using SPSS version 26 and Stata SE version 15.
## Characteristic of participants
The characteristics of the participants at Y0 and Y20 were presented at Table 2. Of 2,935 participants at baseline, $43.8\%$ were male, $46.7\%$ were black, and the average (SD) age was 25 (3.59) years. The proportion of participants with high CVH class dropped from $38.3\%$ at the baseline to $21.8\%$ at Y20, while participants with low CVH class climbed from $5.55\%$ to $21.8\%$. The incident rate of CAC was $26.5\%$ ($\frac{744}{2}$,812). The incident rate of abnormal IMT was $3.6\%$ ($\frac{98}{2}$,691). The average change of CVH scores from baseline to Y20 is –1.26 ± 2.13. The distribution of CVH points and CVH classes at both Y0 and Y20, and also the 20-year changes of both were presented at Figure 1 and Supplementary figure 4–6. Most of the CVH scores of participants were in the 10–12 range at Y0, while the majority were in the 9–11 range at Y20, showing the overall downward trend of CVH from Y0 to Y20.
## Association between CVH and CAC
The association between CVH and incidence of CAC was presented in Table 3. After adjustment by age, gender, race, education, current drinker, history of hypertension, and history of diabetes, the odds ratio (OR) is 0.81 ($95\%$ CI 0.78, 0.86) for per 1 point increase in CVH at baseline. For the further-adjusted baseline CVH score, the OR is 0.86 ($95\%$ CI 0.82, 0.90) for per 1 point increase in CVH changes. An inverse association was also found between changes of CVH scores with incidence of CAC (Ptrend < 0.001) (Supplementary Table 1).
**Table 3**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 | Unnamed: 5 | Unnamed: 6 |
| --- | --- | --- | --- | --- | --- | --- |
| | MODEL 1 | MODEL 1 | MODEL 2 | MODEL 2 | MODEL 3* | MODEL 3* |
| | OR (95%CI) | P-VALUE | OR (95%CI) | P-VALUE | OR (95%CI) | P-VALUE |
| Baseline | | | | | | |
| Per 1 point increase | 0.80 (0.77, 0.83) | <0.001 | 0.80 (0.76, 0.84) | <0.001 | 0.81 (0.78, 0.86) | <0.001 |
| Changes of CVH | | | | | | |
| Per 1 point increase | 0.95 (0.91, 0.99) | 0.007 | 0.94 (0.90, 0.98) | 0.005 | 0.86 (0.82, 0.90) | <0.001 |
In order to assess the specific risk of CAC in different situations, we separated the baseline CVH class into low, moderate, and high classes to see whether the change of CVH class altered the risk of CAC (Table 4). For the moderate class at baseline compared to the stable moderate class, participants who climbed to high class at Y20 had a lower risk of CAC (Adjusted OR 0.55, $95\%$ CI 0.35, 0.89), while those who dropped to low class at Y20 had a higher risk of CAC (Adjusted OR 1.71, $95\%$ CI 1.32, 2.23). Compared to stable high class, the high-to-low class had a higher risk of CAC (Adjusted OR 3.48, $95\%$ CI 1.80, 6.73) and the high-to-moderate class had no significant different risk of CAC (Adjusted OR 1.18, $95\%$ CI 0.81, 1.72). For low class at baseline, significant different risk of CAC was not observed between participants with stable low class and low-to-moderate/high class (Adjusted OR 0.77, $95\%$ CI 0.36, 1.65).
**Table 4**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 | Unnamed: 5 | Unnamed: 6 | Unnamed: 7 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| | N/N | MODEL 1 | MODEL 1 | MODEL 2 | MODEL 2 | MODEL 3 | MODEL 3 |
| | N/N | | | | | | |
| | N/N | | | | | | |
| OR (95%CI) | P-VALUE | OR (95%CI) | P-VALUE | OR (95%CI) | P-VALUE | | |
| Change from low class | 65/151 | | | | | | |
| Low to Low | 46/98 | Ref | Ref | Ref | Ref | Ref | Ref |
| Low to Moderate/High | 19/53 | 0.63 (0.32, 1.26) | 0.190 | 0.65 (0.31, 1.36) | 0.252 | 0.77 (0.36, 1.65) | 0.494 |
| Change from moderate class | 483/1584 | | | | | | |
| Moderate to Moderate | 278/973 | Ref | Ref | Ref | Ref | Ref | Ref |
| Moderate to Low | 179/447 | 1.67 (1.32, 2.11) | <0.001 | 1.88 (1.46, 2.43) | <0.001 | 1.71 (1.32, 2.23) | <0.001 |
| Moderate to High | 26/164 | 0.47 (0.30, 0.73) | 0.001 | 0.52 (0.32, 0.82) | 0.005 | 0.55 (0.35, 0.89) | 0.014 |
| Change from High class | 196/1077 | | | | | | |
| High to High | 69/446 | Ref | Ref | Ref | Ref | Ref | Ref |
| High to low | 23/66 | 2.92 (1.66, 5.16) | <0.001 | 3.80 (1.97, 7.32) | <0.001 | 3.48 (1.80, 6.73) | <0.001 |
| High to moderate | 104/565 | 1.23 (0.88, 1.72) | 0.219 | 1.30 (0.90, 1.87) | 0.170 | 1.18 (0.81, 1.72) | 0.397 |
## Association between CVH and IMT
The association between CVH and incidence of abnormal IMT was presented in Table 5. After adjustment by age, gender, race, education, current drinker, history of hypertension, and history of diabetes, the OR is 0.85 ($95\%$ CI 0.76, 0.94) for per 1 point increase in CVH at baseline. For the further adjusted baseline CVH score, the OR is 0.81 ($95\%$ CI 0.73, 0.90) for per 1 point increase in CVH changes. An inverse association was also found between changes of CVH scores with incidence of abnormal IMT (Ptrend < 0.001) (Supplementary Table 2). Linear association was also found between changes of CVH scores with IMT (Adjusted β (se) –0.012 (0.001), p-value < 0.001) (Supplementary Table 3).
**Table 5**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 | Unnamed: 5 | Unnamed: 6 |
| --- | --- | --- | --- | --- | --- | --- |
| | MODEL 1 | MODEL 1 | MODEL 2 | MODEL 2 | MODEL 3* | MODEL 3* |
| | OR (95%CI) | P-VALUE | OR (95%CI) | P-VALUE | OR (95%CI) | P-VALUE |
| Baseline | | | | | | |
| Per 1 point increase | 0.76 (0.70, 0.84) | <0.001 | 0.82 (0.74, 0.91) | <0.001 | 0.85 (0.76, 0.94) | 0.002 |
| Changes of CVH | | | | | | |
| Per 1 point increase | 0.88 (0.80, 0.97) | 0.009 | 0.89 (0.81, 0.98) | 0.020 | 0.81 (0.73, 0.90) | <0.001 |
In order to assess the specific risk of abnormal IMT in different situations, we separated baseline the CVH class into low, moderate, and high classes to see whether the change of CVH class altered the risk of abnormal IMT (Supplement Table 4). For the moderate class at baseline, compared to stable moderate class, moderate-to-low class had a higher risk of abnormal IMT (Adjusted OR 2.46, $95\%$ CI 1.46, 4.12) while those in the moderate-to-high class did not observe a significant lower risk of abnormal IMT (Adjusted OR 0.70, $95\%$ CI 0.20, 2.37). Compared to the stable high class, the high-to-low and high-to-moderate classes had no significant different risk of abnormal IMT (Adjusted OR 0.60, $95\%$ CI 0.06, 5.64 and Adjusted OR 0.76, $95\%$ CI 0.24, 2.41). For the low class at baseline, significant different risk of abnormal IMT was not observed between participants with stable low class and low-to-moderate/high class (Adjusted OR 0.13, $95\%$ CI 0.01, 1.17).
## Subgroup Analysis
Subgroup analyses based on gender, race, history of hypertension, history of diabetes, and alcohol use were done separately for CAC and abnormal IMT (Supplementary Figure 1 and Figure 2). No significant difference of risk was found between subgroups. After adjustment for family income at Y5, the association between CVH and incidence of CAC and abnormal IMT remained positive. ( Supplementary Table 6).
## Discussion
In this study, our central findings were the association between CVH points (at baseline and changes throughout early adulthood) and risk of subclinical atherosclerosis (CAC and IMT). Improvement on CVH points at baseline and changes by one point both could bring down incidence of CAC and IMT. For participants with moderate CVH class at baseline, the majority of the young population, altering CVH class could benefit in the reduction of midlife risk of subclinical atherosclerosis.
Cardiovascular health in our study contains both metabolic factors and healthy lifestyle factors, as AHA recommended [1]. Comparing prior studies [192021], which only focus on one aspect, our study evaluates cardiovascular health better, from more comprehensive assessment criteria. Prior studies have proven CVH scores at baseline were inversely associated with risk of CAC and IMT in middle age to elderly populations [2223]. Our study fulfilled the blank area in the young adulthood to middle age population. Since the influence of cardiovascular health continues throughout life, focusing on the younger population helps prevent cardiovascular disease earlier. Neglecting variation, single measurement was considered in previous studies [2223]. Only one study measured the changes of CVH during youth population, but it did not observe significant association between them, which was opposite to our results [24]. The failure of demonstrating real association in the population may result from its smaller sample and shorter follow-up time. Our study verified the association between CVH changes and subclinical atherosclerosis in a large size, multi-ethnic, long follow-up cohort study, which ensured the high reliability of our results.
The follow-up of the CARDIA study begun in 1985–1986, when young adults in the United States began to change dietary habits because of the high consumption of fast food [25]; therefore, they might have different cardiovascular risk factors from the last generation. Considering cardiovascular risk may be different with the alteration of generations, we use Life Simple 7, a common measurement for cardiovascular risk assessment, to evaluate early subclinical atherosclerosis risk in young adulthood. In this study, we found that lower CVH class or scores in young adulthood was associated with higher risk of CAC in midlife. Moreover, compared to those within the stable CVH class, participants with worsened CVH class in midlife had significantly higher risk of CAC, no matter which CVH class they fell into at baseline. It suggests that we should pay attention to cardiovascular health earlier in young adulthood, and should not relax vigilance even in midlife. In this study, the majority of participants are in the moderate CVH class at baseline. Compared to the stable moderate CVH class, participants might suffer more subclinical atherosclerosis risk at midlife if they neglect healthy lifestyle or metabolic factors in young adulthood and fall down to low class. On the contrary, participants might gain profit in reducing midlife risk with early intervention. For example, a participant can gain two points by quitting smoking or losing weight, and see a reduction in midlife risk of $28\%$ (central figure). Instead of focusing on all aspects, people can choose one aspect to work on based on their situation, which is easier to help achieve the goal of preventing CVD. Young adults can add morning running to their daily routine to get higher scores in both physical activity and BMI aspects if losing weight at the same time. Great social enlightenment has been shown in our study: cardiovascular health in young adulthood is as important as that in midlife. Therefore, more attention should be paid to young adulthood for primary prevention of CVD.
Key strengths of this study include, first, CARDIA study is representative in studying cardiovascular health of young adults for its identity: a multi-center, long follow-up prospective cohort recruiting participants in 18–25 years. Second, our study explored thoroughly the association between baseline and longitudinal change of CVH and subclinical atherosclerosis and presented with reliable results. Third, our study fulfilled the blank area of early cardiovascular risk in the young population and provided a strategy of early intervention for improving cardiovascular health in young adulthood.
Limitations should be considered in our study. First, the measurement of changes in CVH by only considering baseline and Y20 data could cause neglection of variation during the 20-year period. Even though the measurement is crude, positive results were shown in our study, which would be more reliable with increased follow-up measurements. Meanwhile, we are conducting a multiple follow-up cohort about Life Simple 7 to explore the association between its variation and CVD. Second, in real world research, the possibility of residual confounding cannot be ruled out, but we adjusted confounders based on the characteristics of CARDIA study as far as possible. Third, participants lost to follow-up had different baseline characteristics compared to follow-up participants. Potential bias due to different loss to follow-up in younger participants, black male participants, and those with a lower educational attainment may have occurred (Supplementary table 5).
In this study, baseline and change in CVH were associated with the risk of subclinical atherosclerosis in young adulthood-to-middle age population, suggesting the importance of promoting CVH class.
## Additional File
The additional file for this article can be found as follows:
## Ethics and Consent
All authors contributed important intellectual content during manuscript writing or revision, and read and approved the final manuscript.
## Funding Information
This study was partly supported by the National Natural Science Foundation of China (81870195, 82070384 to X.Liao; 82200408 to J.Li), Guangdong Basic and Applied Basic Research Foundation (2019A1515011582, 2021A1515011668 to X.Liao; 2021A1515110266 to Z.Xiong) and China Postdoctoral Science Foundation (2021TQ0386, 2021M703738 to Z.Xiong) and NSFC Incubation Project of Guangdong Provincial People’s Hospital (KY0120220034 to J.Li). The CARDIA study is conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the University of Alabama at Birmingham (HHSN268201300025C and HHSN268 201300026C), Northwestern University (HHSN268201300027C), University of Minnesota (HHSN268201300028C), Kaiser Foundation Research Institute (HHSN268201300029C), and Johns Hopkins University School of Medicine (HHSN268200900041C). CARDIA is also partially supported by the Intramural Research Program of the National Institute on Aging (NIA) and an intra-agency agreement between NIA and NHLBI (AG0005). This article has been reviewed by CARDIA for scientific content.
## Competing Interests
The authors have no competing interests to declare.
## Author Contributions
XMY, ZYX, XDZ and XXL: research idea and study design. XXL and XDZ: data acquisition. XMY, ZYX, and YFL: data analysis/interpretation. XMY, ZYX, and XBZ: statistical analysis. XMY, ZYX, JYL, and RHH: manuscript drafting. XXL and XDZ are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Xiaomin Ye, Zhenyu *Xiong this* two Authors contributed equally.
## References
1. Lloyd-Jones DM, Hong Y, Labarthe D, Mozaffarian D, Appel LJ, Van Horn L. **Defining and setting national goals for cardiovascular health promotion and disease reduction: The american heart association’s strategic impact goal through 2020 and beyond**. *Circulation* (2010) **121** 586-613. DOI: 10.1161/CIRCULATIONAHA.109.192703
2. Ford ES, Greenlund KJ, Hong Y. **Ideal cardiovascular health and mortality from all causes and diseases of the circulatory system among adults in the United States**. *Circulation* (2012) **125** 987-995. DOI: 10.1161/CIRCULATIONAHA.111.049122
3. Lachman S, Peters RJG, Lentjes MAH, Mulligan AA, Luben RN, Wareham NJ. **Ideal cardiovascular health and risk of cardiovascular events in the EPIC-Norfolk prospective population study**. *Eur J Prev Cardiol* (2016) **23** 986-994. DOI: 10.1177/2047487315602015
4. Gaye B, Canonico M, Perier MC, Samieri C, Berr C, Dartigues JF. **Ideal cardiovascular health, mortality, and vascular events in elderly subjects: The Three-City Study**. *J Am Coll Cardiol* (2017) **69** 3015-3026. DOI: 10.1016/j.jacc.2017.05.011
5. Shaw LJ, Giambrone AE, Blaha MJ, Knapper JT, Berman DS, Bellam N. **Long-term prognosis after coronary artery calcification testing in asymptomatic patients: A cohort study**. *Ann Intern Med* (2015) **163** 14-21. DOI: 10.7326/M14-0612
6. Valenti V, Ó Hartaigh B, Heo R, Cho I, Schulman-Marcus J, Gransar H. **A 15-year warranty period for asymptomatic individuals without coronary artery calcium: A prospective follow-up of 9,715 individuals**. *JACC Cardiovasc Imaging* (2015) **8** 900-909. DOI: 10.1016/j.jcmg.2015.01.025
7. Budoff MJ, Shaw LJ, Liu ST, Weinstein SR, Mosler TP, Tseng PH. **Long-term prognosis associated with coronary calcification. Observations from a registry of 25,253 patients**. *J Am Coll Cardiol* (2007) **49** 1860-1870. DOI: 10.1016/j.jacc.2006.10.079
8. Budoff MJ, Young R, Burke G, Carr JJ, Detrano RC, Folsom AR. **Ten-year association of coronary artery calcium with atherosclerotic cardiovascular disease (ASCVD) events: the multi-ethnic study of atherosclerosis (MESA)**. *Eur Heart J* (2018) **39** 2401b-2408b. DOI: 10.1093/eurheartj/ehy217
9. Friedman GD, Cutter GR, Donahue RP, Hughes GH, Hulley SB, Jacobs DR. **Cardia: study design, recruitment, and some characteristics of the examined subjects**. *J Clin Epidemiol* (1988) **41** 1105-1116. DOI: 10.1016/0895-4356(88)90080-7
10. Fulgoni V, Nicholls J, Reed A, Buckley R, Kafer K, Huth P. **Dairy consumption and related nutrient intake in African-American adults and children in the United States: Continuing Survey of Food Intakes by Individuals 1994–1996, 1998, and the National Health and Nutrition Examination Survey 1999–2000**. *J Am Diet Assoc* (2007) **107** 256-264. DOI: 10.1016/j.jada.2006.11.007
11. Liu K, Daviglus ML, Loria CM, Colangelo LA, Spring B, Moller AC. **Healthy lifestyle through young adulthood and the presence of low cardiovascular disease risk profile in middle age: The Coronary Artery Risk Development in (Young) Adults (CARDIA) study**. *Circulation* (2012) **125** 996-1004. DOI: 10.1161/CIRCULATIONAHA.111.060681
12. Perak AM, Ning H, Khan SS, Bundy JD, Allen NB, Lewis CE. **Associations of late adolescent or young adult cardiovascular health with premature cardiovascular disease and mortality**. *J Am Coll Cardiol* (2020) **76** 2695-2707. DOI: 10.1016/j.jacc.2020.10.002
13. Bhupathiraju SN, Tucker KL. **Coronary heart disease prevention: Nutrients, foods, and dietary patterns**. *Clin Chim Acta* (2011) **412** 1493-1514. DOI: 10.1016/j.cca.2011.04.038
14. Thompson FE, Subar AF, Coulston AM, Boushey CJ, Ferruzzi MG, Delahanty LM. *Nutrition in the prevention and treatment of disease* (2017) 5-48
15. Carr JJ, Nelson JC, Wong ND, McNitt-Gray M, Arad Y, Jacobs DR. **Calcified coronary artery plaque measurement with cardiac CT in population-based studies: Standardized protocol of Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in Young Adults (CARDIA) study**. *Radiology* (2005) **234** 35-43. DOI: 10.1148/radiol.2341040439
16. Mao SS, Pal RS, McKay CR, Gao YG, Gopal A, Ahmadi N. **Comparison of coronary artery calcium scores between electron beam computed tomography and 64-multidetector computed tomographic scanner**. *J Comput Assist Tomogr* (2009) **33** 175-178. DOI: 10.1097/RCT.0b013e31817579ee
17. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Detrano R. **Quantification of coronary artery calcium using ultrafast computed tomography**. *J Am Coll Cardiol* (1990) **15** 827-832. DOI: 10.1016/0735-1097(90)90282-T
18. Polak JF, Person SD, Wei GS, Godreau A, Jacobs DR, Harrington A. **Segment-specific associations of carotid intima-media thickness with cardiovascular risk factors: The coronary artery risk development in young adults (CARDIA) study**. *Stroke* (2010) **41** 9-15. DOI: 10.1161/STROKEAHA.109.566596
19. Gao JW, Hao QY, Zhang HF, Li XZ, Yuan ZM, Guo Y. **Low-carbohydrate diet score and coronary artery calcium progression: Results from the CARDIA Study**. *Arterioscler Thromb Vasc Biol* (2020) 491-500. DOI: 10.1161/ATVBAHA.120.314838
20. Laddu DR, Rana JS, Murillo R, Sorel ME, Quesenberry CP, Allen NB. **25-year physical activity trajectories and development of subclinical coronary artery disease as measured by coronary artery calcium: The Coronary Artery Risk Development in Young Adults (CARDIA) Study**. *Mayo Clin Proc* (2017) **92** 1660-1670. DOI: 10.1016/j.mayocp.2017.07.016
21. Cassidy AE, Bielak LF, Zhou Y, Sheedy PF, Turner ST, Breen JF. **Progression of subclinical coronary atherosclerosis: Does obesity make a difference?**. *Circulation* (2005) **111** 1877-1882. DOI: 10.1161/01.CIR.0000161820.40494.5D
22. Hwang SJ, Onuma O, Massaro JM, Zhang X, Fu YP, Hoffmann U. **Maintenance of ideal cardiovascular health and coronary artery calcium progression in low-risk men and women in the Framingham Heart Study**. *Circ Cardiovasc Imaging* (2018) **11**. DOI: 10.1161/CIRCIMAGING.117.006209
23. Kim S, Chang Y, Cho J, Hong YS, Zhao D, Kang J. **Life’s Simple 7 cardiovascular health metrics and progression of coronary artery calcium in a low-risk population**. *Arterioscler Thromb Vasc Biol* (2019) **39** 826-833. DOI: 10.1161/ATVBAHA.118.311821
24. Bensenor IM, Goulart AC, Santos IS, Bittencourt MS, Pereira AC, Santos RD. **Association between a healthy cardiovascular risk factor profile and coronary artery calcium score: Results from the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil)**. *Am Heart J* (2016) **174** 51-59. DOI: 10.1016/j.ahj.2015.12.018
25. Guthrie JF, Lin BH, Frazao E. **Role of food prepared away from home in the American diet, 1977–78 versus 1994–96: Changes and consequences**. *J Nutr Educ Behav* (2002) **34** 140-150. DOI: 10.1016/S1499-4046(06)60083-3
|
---
title: A Meta-Analysis of eHealth Interventions on Ischaemic Heart Disease Health
Outcomes
authors:
- Puteri Sofia Nadira Megat Kamaruddin
- Azmawati Mohammed Nawi
- Mohd Rizal Abdul Manaf
- Mohamad Nurman Yaman
- Abdul Muizz Abd Malek
journal: Global Heart
year: 2023
pmcid: PMC10022534
doi: 10.5334/gh.1173
license: CC BY 4.0
---
# A Meta-Analysis of eHealth Interventions on Ischaemic Heart Disease Health Outcomes
## Abstract
### Background:
Electronic Health (eHealth) interventions as a secondary prevention tool to empower patients’ health in decision-making and behaviour.
### Objective:
With the growing body of evidence supporting the use of eHealth interventions, the intention is to conduct a meta-analysis on various health outcomes of eHealth interventions among ischaemic heart disease (IHD) patients.
### Methods:
Based on PRISMA guidelines, eligible studies were searched through databases of Web of Science, Scopus, PubMed, EBSCOHost, and SAGE (PROSPERO registration CRD42021290091). Inclusion criteria were English language and randomised controlled trials published between 2011 to 2021 exploring health outcomes that empower IHD patients with eHealth interventions. RevMan 5.4 was utilised for meta-analysis, sensitivity analysis, and risk of bias (RoB) assessment while GRADE software for generating findings of physical health outcomes. Non-physical health outcomes were analysed using SWiM (synthesis without meta-analysis) method.
### Results:
This review included 10 studies, whereby, six studies with 895 participants’ data were pooled for physical health outcomes. Overall, the RoB varied significantly across domains, with the majority was low risks, a substantial proportion of high risks and a sizeable proportion of unclear. With GRADE evidence of moderate to high quality, eHealth interventions improved low density lipoprotien (LDL) levels in IHD patients when compared to usual care after 12 months of interventions (SMD –0.26, $95\%$ CI [–0.45, –0.06], I2 = $0\%$, $$p \leq 0.01$$). Significance appraisal in each domain of the non-physical health outcomes found significant findings for medication adherence, physical activity and dietary behaviour, while half of the non-significant findings were found for other behavioural outcomes, psychological and quality of life.
### Conclusions:
Electronic Health interventions are found effective at lowering LDL cholesterol in long-term but benefits remain inconclusive for other physical and non-physical health outcomes for IHD patients. Integrating sustainable patient empowerment strategies with the advancement of eHealth interventions by utilising appropriate frameworks is recommended for future research.
## 1.0 Introduction
Ischaemic heart disease (IHD) persists as a major contributor to premature mortality and death rates worldwide, with economic growth and urbanisation exerting the greatest influence on its onset [1]. Socioeconomic changes, increased life expectancy, and lifestyle-related risk factors have all contributed to the increased IHD mortality in recent decades [2]. Patients with IHD have a greater risk of premature death, myocardial infarction, and readmission. Following diagnosis, international guidelines advocate the implementation of secondary prevention strategies [3]. These strategies include physical activity, lifestyle modification guidance, symptom and medication management, and psychosocial support to improve IHD outcomes [4]. The World Health Organization stated that risk factor modification and self-care can prevent approximately $80\%$ of cardiac events [5].
Secondary prevention through electronic health (eHealth) interventions is a viable substitute for conventional cardiac rehabilitation as they can be implemented immediately. Despite being established in the 1990s, the term ‘eHealth’ did not become widely used until 1999 [6]. Electronic Health, telemedicine, and mobile Health (mHealth) are frequently used interchangeably. Although there are differences between the concepts, it is now increasingly common to use eHealth as a blanket term that includes telemedicine and mHealth [7]. A new discipline at the nexus of medicine, public health, and business, eHealth is defined as the improvement of health-related information and services via electronic means [6]. For the purpose of this research, the umbrella term ‘eHealth’ refers to health-related information and communication technologies, such as smartphone mobile applications (apps), short message service (SMS), websites, emails, telemonitoring, phone calls, and wearables/monitoring devices (pedometer, accelerometer, smartwatch, sleep tracker, heart rate monitors) [8]. Nevertheless, there is limited evidence evaluating the health outcomes of eHealth interventions among patients with IHD.
In particular, eHealth interventions benefit from the vast functionalities of new technologies that enable people to access health information and educational content quickly and easily to continuously monitor their health status and behaviour and to receive individualised feedback about the suitability of their actions and physiological parameters in real-time. Additionally, eHealth interventions enable patients to easily communicate online with their caregivers and other patients. All of these characteristics may be very important for encouraging and assisting people in choosing and maintaining healthy lifestyles, which will subsequently inhibit disease onset or progression. Finally, gamification helped eHealth treatments better encourage people to adhere to long-term preventative and lifestyle change interventions. For all of these reasons, using eHealth interventions is a promising strategy to improve the health of IHD patients.
Previous research in this field focused primarily on telehealth interventions, which are defined as healthcare delivery over the phone, the Internet, or videoconferencing [910]. Telehealth is more hospital-based or clinician-dependent, where the hurdle of resource utilisation in terms of requiring skilled experts for treatment delivery and health budget may also occur with current individualised and patient-centred eHealth interventions. Despite the aforementioned issue, eHealth intervention accessibility has evolved and is available according to individual affordability. Websites, mobile apps, email, mobile phones, SMS, and monitoring sensors are all current common communication modes. These modalities enable eHealth interventions to encourage healthy behaviours in real-time, enabling users to access and interact with data, upload and review records, receive automated feedback, and communicate with peers or healthcare professionals [11]. Individualised eHealth interventions can accommodate individual risk factors, care needs, objectives, and resources to change health behaviour [12]. Approximately $65.6\%$ of the world population uses the Internet [13], and older adults are increasingly using it as an important health information source and a patient empowerment tool in health decision-making and behaviour [14].
Thus, there is growing support for delivering secondary prevention care components via the eHealth platform that best address cardiac patients’ individualised healthcare needs, resulting in additional health benefits and identifying impediments to service access and use [15]. Based on the growing body of evidence supporting the use of eHealth interventions, a meta-analysis was conducted on the health outcomes of eHealth interventions among patients with IHD.
## 2.1 Design
This review followed the procedures outlined in the Cochrane Handbook for Systematic Reviews [16] (The Cochrane Collaboration, Oxford, UK) of Interventions and reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [17]. The PRISMA protocol encourages researchers to obtain accurate information from reliable sources. The systematic literature review was planned using this protocol by developing an appropriate research question. The PRISMA Checklist can be found in Supplementary Document 1 and 2. The systematic search was classified into three stages: identification, screening, and inclusion. PROSPERO had registered the systematic review (CRD42021290091), and the study protocol can be accessed at https://www.crd.york.ac.uk/prospero/displayrecord.php?RecordID=290091.
## 2.2 Search methods
A comprehensive search was conducted across five databases, including Web of Science, Scopus, PubMed, EBSCOHost, and SAGE, covering the 10-year period from 2011 to 2021 in light of recent innovations in the advancement of eHealth interventions. The PICO framework was utilised to identify keywords that aided authors in formulating a fundamental research question. It was characterised by three concepts: Population or Problem, Interest or Intervention, and Outcome. The formulated keywords based on these concepts are ‘Adult’ and ‘Ischaemic Heart Disease’ (Population), ‘eHealth intervention’ (Intervention) and ‘outcome’ (Outcome), which served as the basis for the formulation of the research objective. Comparator element was not defined in this study since there was no comparator to be examined and it was not part of the study’s design to compare the health outcomes against those of eHealth interventions.
All searches were conducted within a week of the 1st–7th December 2021. The identification stage included a search for all possible synonyms, medical subject heading (MeSH) terms, similar terms, and variants of the keywords: ‘Adult’, ‘Ischaemic Heart Disease’, ‘eHealth intervention’ and ‘outcome’ together with the Boolean operators (Supplementary Document 3). This method provided better coverage for locating relevant articles in the selected databases. These databases were distinguished by their extensive literature collections and advanced search features.
## 2.3 Search outcome
The database search revealed 1231 English language articles published between 2011 and 2021. After 87 duplicates were removed, 1144 articles were further screened with the following criteria to determine inclusion: [1] randomised controlled trials (RCTs); [2] IHD patients aged 18 and above were recruited; [3] made use of a website or a mobile apps in addition to other ways of communicating (email, SMS, phone call) and [4] provided information about physical and non-physical health outcomes (behavioural, psychological, quality of life (QoL) and others). This meta-analysis included only RCTs to generate the most robust evidence for eHealth interventions [18]. The type of study that explicitly stated not randomised, quasi-experimental, pre-post, review articles, editorials, proceedings and commentary articles were excluded.
This process resulted in manually sorting 226 articles that concentrated on participants who were directly connected to the Internet and actively used it, with or without the assistance of other mechanisms, rather than relying solely on data transfer via wearable monitors between patients and professionals. Comparative studies included those in which participants received no intervention, standard or usual care from their health care systems. This review omitted articles in which participants were monitored exclusively via devices or received SMS or phone call reminders without the use of Internet. This method resulted in the exclusion of 216 articles based on unsuitable target population, not RCT and irrelevant health outcomes, e.g., medication trial and merely one-way monitoring patients with eHealth interventions. Final eligibility process included 10 articles [19202122232425262728]. However, only six [192022242527] were fit for meta-analysis due to data compatibility. Figure 1 depicts the PRISMA flow for study identification.
**Figure 1:** *PRISMA diagram of identifying studies of eHealth interventions [17].*
Overall, two independent reviewers (PSNMK and AMN) conducted a sequential review and selection of studies, removed duplicates and evaluated the eligibility of the abstract to the full text. Any disagreements about the study’s inclusion were resolved through consultation with a third investigator (MRAM). The data was transferred to Review Manager 5.4 software [29] by one review author (MNY) and a second reviewer (AMAM) verified the accuracy of the data entry.
## 2.4 Assessment of risk-of-bias and certainty of evidence rating
The Cochrane risk of bias (RoB) tool for randomised trials (RoB 2) was used to assess the study quality as it is the most frequently used tool for randomised trials [30]. This tool is an outline for determining the RoB in a single outcome (an estimate of the effect of an experimental intervention compared to a comparator on a specific outcome) from any type of randomised trial. Within each domain, a series of key questions were used to elicit information about trial characteristics that are associated with bias risk. The following seven major criteria were used by two review authors (PSNMK and MNY) to independently assess each included trial for RoB: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting and other bias. Each domain was classified as having a low, moderate, or high RoB. If there was disagreement, a third author (MRAM) led a discussion leading to a consensus. Meanwhile, the ‘Summary of findings’ table contains the certainty of evidence determination from meta-analysis outcome using the well-established and widely used Grading of Recommendations Assessment, Development and Evaluation (GRADE) method (Supplementary Document 4) [31].
## 2.5 Assessment of heterogeneity, meta-analysis
When three or more studies reported the same outcome, the Review Manager 5.4 was used to pool the data; otherwise, synthesis without meta-analysis (SWiM) was performed [32]. SWiM checklist can be found in Supplementary Document 5. The I2 statistic was used to measure the degree of heterogeneity in the results with a $50\%$ threshold deemed significant [33]. Data were pooled and meta-analysis performed where necessary, using the RevMan 5.4 software’s random-effect model [29], as the assumption was there might be a significant effect between the studies in that they differ in terms of population and sample sizes. The findings were presented using the standard mean difference (SMD) for continuous outcomes and their respective $95\%$ confidence intervals (CI) because different methods were utilised to measure the same outcome and a study [24] result needs to be converted to produce the same measures. To explore the differences, subgroup analysis was employed. When the pooled effect displayed a high level of heterogeneity, a sensitivity analysis was conducted [16]. Publication bias was assessed by using funnel plots.
## 3.0 Results
There were 10 RCTs included in this review published between 2015 and 2020. The characteristics of the included studies were summarised in Table 1, and the overall modality of the various eHealth interventions utilised to measure different health outcomes was summarised in Table 2. The number of participants ranged from 48 [21] to 312 [25].
## 3.1 Risk-of-bias in included studies
Derived from the 10 studies included in this review, Figure 2a depicts the proportion of studies with low, high, and unclear RoB in each domain. In contrast, Figure 2b illustrates the RoB judgement for each included study in each domain. Overall, the studies’ RoB varied significantly across domains, with the majority assessed to be low risks in random sequence generation, incomplete outcome data, selective reporting, and a minority were evaluated to be low risks in allocation concealment. A considerable proportion of studies were discovered to have a high RoB in allocation concealment, blinding of participants, personnel and outcome assessment, as well as, incomplete outcome data. Meanwhile, a sizeable proportion of the studies included in this review lacked sufficient information to permit a meaningful RoB assessment.
**Figure 2:** *(a) The risk-of-bias graph: proportions of studies with low, high, and unclear risks of bias in each domain. (b) The risk-of-bias summary: the risk-of-bias judgement of each included study in each domain.*
## 3.2 Physical health outcomes
The effect of eHealth interventions on physical health outcomes assessed were BMI (body mass index), systolic and diastolic resting blood pressure (BP), and lipid profile (low density lipoprotein (LDL), high density lipoprotein (HDL) and total cholesterol). In total, six studies [192022242527] with 895 participants (441 patients who received eHealth interventions and 160 controls) contributed data, while the outcome data of four studies [21232628] were not reported sufficiently for meta-analysis.
Supplementary Document 6 contains a complete list of the physical health outcomes estimates for each comparison. For certainty-of-evidence ratings of the physical health outcomes, and reasons for downgrading, see the Summary of findings (Supplementary Document 4).
## 1. Body Mass Index
BMI refers to the change of participants’ body mass index at six months in kilogram per metre squared (kg/m2). Data from five studies were pooled and indicated that eHealth interventions did not improve BMI, however it was not significant (SMD 0.05, $95\%$ CI [–0.21, 0.30], I2 = $67\%$, $$p \leq 0.72$$; Figure 3a).
**Figure 3:** *The effect of eHealth intervention on physical health outcomes. (a) The effect of eHealth intervention on BMI at 6 months (n = 5). (b) The effect of eHealth intervention on Systolic BP at 6 months (n = 5). (c) The effect of eHealth intervention on Diastolic BP at 6 months (n = 4). (d) The effect of eHealth intervention on LDL at 6 and 12 months (n = 6). (e) The effect of eHealth intervention on HDL at 6 and 12 months (n = 6). (f) The effect of eHealth intervention on Total Cholesterol at 6 and 12 months (n = 6).*
## 2. Blood Pressure
BP refers to the change of resting BP in terms of systolic and diastolic at six months in millimetres of mercury (mmHg). Data from five studies for systolic and four studies for diastolic were pooled and indicated that eHealth interventions did not significantly improve systolic (SMD 0.11, $95\%$ CI [–0.39, 0.60], I2 = $91\%$, $$p \leq 0.67$$; Figure 3b) and diastolic (SMD 0.27, $95\%$ CI [–0.33, 0.86], I2 = $90\%$, $$p \leq 0.38$$; Figure 3c).
## 3. Lipid profile
The effect of eHealth interventions on LDL was evaluated at 6 and 12 months in terms of milimoles per litre (mmol/L). Due to the substantial degree of heterogeneity in the pooled results of the four included studies, they were divided into subgroups according to the timing of outcome measurement (at 6 and 12 months). A difference in LDL levels favouring eHealth interventions was found. eHealth interventions have significantly improved LDL at 12 months (SMD -0.26, $95\%$ CI [–0.45, -0.06], I2 = $0\%$, $$p \leq 0.01$$), but not at 6 months (SMD -0.13, $95\%$ CI [–0.32, 0.06], I2 = $29\%$, $$p \leq 0.17$$; Figure 3d).
Meanwhile, the effect of eHealth interventions on HDL was evaluated via subgroup analysis at 6 and 12 months (mmol/L), but did not show significant differences (SMD 0.04, $95\%$ CI [–0.20, 0.29], I2 = $57\%$, $$p \leq 0.73$$) at 6 months and at 12 months (SMD -0.05, $95\%$ CI [–0.25, 0.14], I2 = $0\%$, $$p \leq 0.34$$; Figure 3e).
Lastly, the effect of eHealth interventions on total cholesterol was evaluated via subgroup analysis at 6 and 12 (mmol/L), but did not show significant differences (SMD -0.04, $95\%$ CI [–0.24, 0.15], I2 = $32\%$, $$p \leq 0.66$$) at 6 months and (SMD -0.12, $95\%$ CI [–0.71, 0.46], I2 = $85\%$, $$p \leq 0.68$$) at 12 months (Figure 3f).
Overall, the funnel plots generated were asymmetrical due to differences among the studies (Figure 4).
**Figure 4:** *Funnel plot comparison of eHealth intervention for physical health outcomes. (a) Funnel plot comparison of eHealth intervention for BMI (n = 5). (b) Funnel plot comparison of eHealth intervention for resting systolic BP at 6 months (n = 5). (c) Funnel plot comparison of eHealth intervention for resting diastolic BP at 6 months (n = 4). (d) Funnel plot comparison of eHealth intervention for LDL at 6 and 12 months (n = 6). (e) Funnel plot comparison of eHealth intervention for HDL at 6 and 12 months (n = 6). (f) Funnel plot comparison of eHealth intervention for Total Cholesterol at 6 and 12 months (n = 6).*
## 3.2.1 Sensitivity Analysis
From the analysis of physical health outcomes, based on the sensitivity analysis of those with significant heterogeneity, the effectiveness of eHealth interventions might be related to the differences in target population, race or ethnicity, as Skobel 2017 [22] is a multi-centre study and Dorje 2019 [25] is a study conducted among the Chinese. By excluding these studies from the meta-analysis, the heterogeneity improved for BP effect estimates for systolic (SMD 0.11, $95\%$ CI [–0.09, 0.32], I2 = $0\%$, $$p \leq 0.29$$; Figure 5a) and diastolic (SMD –0.05, $95\%$ CI [–0.25, 0.16], I2 = $0\%$, $$p \leq 0.65$$; Figure 5b). Nevertheless, the results were still not significant.
**Figure 5:** *Sensitivity analysis on the effect of eHealth intervention on BP at 6 months. (a) Sensitivity analysis on the effect of eHealth intervention on Systolic BP at 6 months (n = 3). (b) Sensitivity analysis on the effect of eHealth intervention on Diastolic BP at 6 months (n = 3).*
## 3.3 Non-physical health outcomes
The utilisation of eHealth interventions can be seen from the domains derived for non-physical health outcomes, which are: [1] Behavioural, (medication adherence, physical activity, dietary behaviour and others); [2] Psychological, (anxiety and depression); and [3] other health outcomes (QoL). For non-physical health outcomes, instead of meta-analysis, synthesis without meta-analysis (SWiM) was performed in view of data variability with various scale measurements utilised and different duration outcomes. The final outcome summary deduced from all 10 studies was categorised accordingly (Table 3). Significance appraisal between comparable papers in each domain found half of the significant findings were medication adherence, physical activity and dietary behaviour, and half of the non-significant findings were other behavioural outcomes, psychological and QoL.
**Table 3**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 | Unnamed: 5 | Unnamed: 6 |
| --- | --- | --- | --- | --- | --- | --- |
| AUTHOR, YEAR, STUDY DURATION OUTCOME | NON-PHYSICAL HEALTH OUTCOME (EHEALTH INTERVENTION VERSUS CONTROL) | NON-PHYSICAL HEALTH OUTCOME (EHEALTH INTERVENTION VERSUS CONTROL) | NON-PHYSICAL HEALTH OUTCOME (EHEALTH INTERVENTION VERSUS CONTROL) | NON-PHYSICAL HEALTH OUTCOME (EHEALTH INTERVENTION VERSUS CONTROL) | NON-PHYSICAL HEALTH OUTCOME (EHEALTH INTERVENTION VERSUS CONTROL) | NON-PHYSICAL HEALTH OUTCOME (EHEALTH INTERVENTION VERSUS CONTROL) |
| AUTHOR, YEAR, STUDY DURATION OUTCOME | | | | | | |
| AUTHOR, YEAR, STUDY DURATION OUTCOME | BEHAVIOURAL | BEHAVIOURAL | BEHAVIOURAL | BEHAVIOURAL | PSYCHOLOGICAL | QUALITY OF LIFE |
| AUTHOR, YEAR, STUDY DURATION OUTCOME | | | | | PSYCHOLOGICAL | QUALITY OF LIFE |
| AUTHOR, YEAR, STUDY DURATION OUTCOME | MEDICATION ADHERENCE | PHYSICAL ACTIVITY | DIETARY BEHAVIOUR | OTHERS | PSYCHOLOGICAL | QUALITY OF LIFE |
| Dale 2015 [19], at 6 months | MMAS-8Mean difference: 0.58, 95% CI 0.19–0.97; ** p = 0.004 | Godin Leisure Time Physical Activity QuestionnaireIG: 28% to 31%CG: 11% to 24% | – | Self-reported composite health behaviourAOR: 1.93, 95% CI 0.83–4.53; p = 0.13IG: 33% to 53%CG: 27% to 39% | HADS (anxiety)Mean difference: 1.18, 95% CI 0.28, 2.08; * p = 0.01 | – |
| Frederix 2015 [20], at 6 months | – | IPAQ in MET-min/week of VMWIG: χ2 (2): 13.7; * p = 0.01CG: χ2 (2): 0.6; p = 0.72 | – | – | – | HRQLIG: χ2 (2): 14.0; ** p < 0.001CG: χ2 (2): 3.1; p = 0.21 |
| Martin 2015 [21] | – | Daily stepsMean difference: 3376, 95% CI 1951–4801; ** p < 0.001 | – | – | – | – |
| Skobel 2017 [22], at 6 months | – | – | – | – | HADS (Anxiety)Mean (SD); p = 0.1IG: 1.6 (3.1) CG: –0.63 (3.8)HADS (Depression)Mean (SD)p = 0.27IG: 1.36 (3.7)CG: 0 (1.6) | EQ–5DMean (SD); p = 0.98IG: 0.64 (13.9)CG: 0.54 (10.7) |
| Kamal 2018 [23], at 3 months | MMAS–8Mean difference (SD): –0.06 (0.13), 95% CI –0.39–0.19; p = 0.69 | – | – | – | – | – |
| Choi 2019 [24], at 6 months | – | – | MDS compliance(** p < 0.001)IG: 27.5% to 64.7%CG: 18.4% to 57.1% | – | – | – |
| Dorje 2019 [25], at 6 months | Adherence to cardioprotective medicationsMean difference: 1.79, 95% CI 1.76–1.87; * p = 0.019 | SF-12 Physical Health scoreMean difference: 1.26, 95% CI –0.74 – 3.26; p = 0.22 | – | Smoking statusMean difference: 2.42, 95% CI 0.42–14.00; p = 0.32 | GAD–7Mean difference: 0.60, 95% CI –0.25 – 1.46; p = 0.17 | – |
| Broers 2020 [26], at 6 months | – | – | – | Health Promotion Lifestyle ProfileF unadjusted (2,271.90) = 8.28; ** p < 0.001 | – | WHOQOL-BREFF(2,135.42) = 9.63; ** p < 0.001 |
| Lunde 2020 [27], at 1 year | – | Exercise habitsMean change (SD)IG: 1.4 (1.5); (** p < 0.001)CG: 0.6 (1.1); (** p < 0.001) | – | – | – | Heart QoLMean change (SD)IG: 0.21 (0.47); (* p < 0.05)CG: 0.09 (0.45) |
| Song 2020 [28], at 6 months | – | Exercise habits based on ACSMχ2 9.826; * p = 0.02 | – | – | - | – |
| Outcome summary | Significant | Significant | Significant | Not significant | Not significant | Not significant |
## 4.0 Discussion
The review determined that eHealth interventions have significant effects on LDL in patients with IHD but had non-significant effects on other health outcomes. The eHealth interventions were used in secondary prevention to monitor patients’ vital signs and lifestyle modification adherence, and hence, improved their overall QoL.
Here, the main finding demonstrated that 12-month eHealth interventions improved LDL levels in IHD patients as compared to usual care. The meta-analysis demonstrated that a long-term and sustainable intervention is essential for reducing LDL. Supported by current IHD treatment guidelines, LDL remains one of the primary treatment target for reducing ischaemic events and in secondary prevention follow-up visits, where lower levels are better [3435]. Notably, LDL particles act as a major cholesterol transporter and are the key contributor to the increased risk of atherosclerotic lesion formation [36]. Similarly, elevated HDL is correlated with decreased risk in atherosclerosis from epidemiological studies but yielded null results in therapeutic clinical trials [37]. This justification was reflected in the non-significant findings for HDL and total cholesterol, as LDL largely replaced total cholesterol as a risk marker and primary treatment target [38]. Long-term interventions require adherence to both pharmacotherapy and non-pharmacotherapy methods by applying modalities such as mobile smartphone apps and SMS [2527]. The results also suggested that eHealth interventions implemented heterogeneously across countries and populations were successful, which was supported by moderate to high GRADE effect estimates (Supplementary Document 4).
Other physical health outcomes (BMI, BP, HDL, total cholesterol) exerted no significant effects with eHealth interventions, even with sensitivity analysis of BP findings to improve heterogeneity due to population differences. This is possibly reflected by the small sample size and high RoB (Figure 2a, 2b) from allocation concealment and participant and personnel blinding. In some studies, the differences in the technologies used in eHealth interventions and the intervention duration, e.g. 6 months or 12 months, might be insufficient to improve certain cardio-metabolic parameters. Furthermore, a limited number of studies were able to explain the non-significant changes. Here, we included a total of six RCTs; therefore, additional research is required to confirm the results. Moreover, the asymmetrical shapes of the overall funnel plots suggested the existence of publication bias, which is induced because statistically significant results are more likely to be published than null or non-significant results. Hence, publication bias may threaten the validity of such analyses, leading to incorrect, typically over-optimistic conclusions [39].
The findings for the non-physical health outcome domains measured in this study revealed that subjective measurements are ambiguous and how patients benefit from eHealth interventions can be interpretd differently. The beneficial effects of eHealth interventions can sustain lifestyle modification (medication adherence, diet, and exercise) via constant professional support and individualised lifestyle behavioural changes [11]. However, high-quality evidence is lacking and most evidence was from high-income countries [40], where most studies [1920212224262740] in the present review were conducted in Europe or North America.
The non-significant finding of psychological health outcomes and QoL revealed that the ambiguous effects of eHealth interventions identified in this review might be related and linked to the underpinning health behaviour theory: the Transactional Model of eHealth Literacy (TMeHL). Transactional Model of eHealth *Literacy is* a continuous communication transaction process that is constantly modified according to eHealth contextual factors and prior eHealth experiences. Based on the seminal component of eHealth literacy levels, the interplay between task-oriented factors (usability of eHealth interventions modality) and user-oriented factors (age, race) would result in health outcomes by empowering patients with eHealth skills [41]. The present review demonstrated the user-oriented factor in certain studies with a mean participant age of >60 years [20222526]. Older people, who typically have deteriorating cognitive ability and memory, lack effective learning methods for advancement via mobile devices and limited Internet applications and exposure [42]. Coupled with the task-oriented factors of various modalities used in the present study, the included studies captured from 2015 to 2020 revealed the evolution of various eHealth interventions used to influence the non-significance of the non-physical health outcome results. Older adults may be less familiar with novel eHealth technologies and thus struggle to accept and adapt to modern technological devices; accordingly, there should be special consideration of the receptivity, memory, and auditory abilities of the elderly [43]. Nonetheless, none of the included studies mentioned the fundamental eHealth literacy level or other literacy levels of its kind. These literacy levels are essential as health experts have concluded that patients’ knowledge or literacy levels for compatible use of eHealth devices with their users should be assessed to produce excellent health outcomes [44].
Despite physical health outcomes being measured in a standardised and universally accepted method, cautious interpretation is vital. Therefore, we conclude that eHealth interventions can improve the long-term LDL for IHD patients. Further randomised trials with adequate blinding and longer-term follow-up may demonstrate that eHealth interventions have better and significant health outcomes. Hence, eHealth interventions are useful for enhancing specific biomarker results. Researchers or physicians should determine whether certain eHealth interventions are appropriate based on their study objective and their patients’ needs. The findings also suggested that future theory-guided trial interventions are needed as is the need to consider other health behaviour moderators and mediators. Issues aligned with the TMeHL theory might explain the non-significant findings that may be incorporated in providing IHD patients with corresponding eHealth modalities that match their eHealth literacy levels and task- and user-oriented factors. This finding is in accordance with the simple mHealth group intervention strategy that renders health information delivered via SMS, WeChat, and email more acceptable to older adults [43]. Empirical studies supported using any kind of eHealth interventions to improve patients’ health outcomes for various non-communicable diseases, for example, cancer [454647], diabetes [484950] and predominantly IHD [51525354].
IHD treatment also includes various target components and combinations. However, there are no specific interventions for patients with various cardio-metabolic components. Therefore, it is necessary to optimise the rapidly developing eHealth interventions to provide precise care to patients with IHD. Following the pathways in developed and developing nations, the implementation could bes advantageous for healthcare providers providing recommendations for patients with distinct characteristics. In addition to governance and regulatory challenges, the remaining hurdles are information management, interoperability, and integration. These hurdles include the capacity to enable communication technologies and the availability of online information for doctors and patients with IHD that can help manage co-morbidities [55]. Moreover, the findings were generated from a small number of RCTs and we believe that studies involving more regions and larger samples are required before eHealth interventions may be suggested in future guidelines [56].
## 4.1 Strengths and Limitations
To our knowledge, this is the first meta-analysis that focused on the health outcomes of IHD patients by utilising eHealth interventions. The findings are relevant regarding IHD outcomes achieved through the suitability of eHealth interventions that address the unique healthcare needs of IHD patients considering age, literacy level, and population.
Only three studies were conducted in Asia, while the remaining studies were conducted in Europe or North America, which limited the generalisability of the results. Further eHealth interventions with rigorous study designs and more diverse populations from different cultural contexts are required to produce credible evidence. Most of the included studies were pilot and feasibility studies with different durations and used a broad range of eHealth interventions. However, patients may be sceptical of new technology for various reasons, including their preconceptions or unfamiliarity with the use of that technology, which can affect user satisfaction and long-term intervention adherence. Thus, the future challenges for researchers and clinicians are to design studies that incorporate patient preferences and to significantly improve intervention study reporting. These aspects are critical so that clinicians and researchers can assess the feasibility of implementing eHealth interventions for patient education and secondary prevention not only in cardiac care but also in other patient groups. Additionally, the participants’ eHealth literacy levels and computer literacy skills are unknown, which may have influenced the study outcomes. As the reviewed RCTs typically featured inadequate concealment and blinding, evaluating their methodological rigour was challenging, which resulted in a high risk of selection and performance bias. Additional studies should be conducted to strengthen concealment and outcome reporting methods to improve the quality of evidence.
## 5.0 Conclusion
Based on moderate to high effect estimates, eHealth interventions in long-term could effectively lower LDL cholesterol. However, given the study limitations, the effects of eHealth interventions on other physical and non-physical health outcomes remain inconclusive. It is recommended that sustainable patient empowerment strategies be integrated with the advancement of eHealth interventions for future research by utilising appropriate frameworks considering other potential health behaviour moderators and mediators.
## Data Accessibility Statements
The dataset used and analysed during this review is available from the corresponding author on reasonable request.
## Additional File
The additional file for this article can be found as follows:
## Ethics and Consent
This study is part of research approved by the Medical Research Ethics Committee of UKM (protocol code FF-2021-117, approved on 7th April 2021).
## Funding Informaton
This research received no specific grant from any funding agency in public, commercial or not-for-profit sectors.
## Competing Interests
The authors have no competing interests to declare.
## Author Contributions
All authors contributed to the design and implementation of the research, analysis of the results and writing of the manuscript.
## References
1. Khan MA, Hashim MJ, Mustafa H, Baniyas MY, AlKatheeri R. **Global Epidemiology of Ischemic Heart Disease: Results from the Global Burden of Disease Study**. *Cureus* (2020.0) **12**. DOI: 10.7759/cureus.9349
2. Wang EY, Dixson J, Schiller NB, Whooley MA. **Causes and Predictors of Death in Patients With Coronary Heart Disease (from the Heart and Soul Study)**. *The American Journal of Cardiology* (2017.0) **119** 27-34. DOI: 10.1016/j.amjcard.2016.09.006
3. Perel P, Avezum A, Huffman M, Pais P, Rodgers A, Vedanthan R. **Reducing Premature Cardiovascular Morbidity and Mortality in People With Atherosclerotic Vascular Disease: The World Heart Federation Roadmap for Secondary Prevention of Cardiovascular Disease**. *Global Heart* (2015.0) **10** 99-110. DOI: 10.1016/j.gheart.2015.04.003
4. Yudi MB, Clark DJ, Tsang D, Jelinek M, Kalten K, Joshi S. **SMARTphone-based, early cardiac REHABilitation in patients with acute coronary syndromes [SMART-REHAB Trial]: a randomized controlled trial protocol**. *BMC Cardiovascular Disorders* (2016.0) **16** 170. DOI: 10.1186/s12872-016-0356-6
5. **Cardiovascular diseases (CVDs)**. (2021.0)
6. Eysenbach G. **What is e-health?**. *Journal Osf Medical Internet Research* (2001.0) **3**. DOI: 10.2196/jmir.3.2.e20
7. Meier CA, Fitzgerald MC, Smith JM. **eHealth: extending, enhancing, and evolving health care**. *Annual Review of Biomedical Engineering* (2013.0) **15** 359-82. DOI: 10.1146/annurev-bioeng-071812-152350
8. Hutchesson MJ, Gough C, Müller AM, Short CE, Whatnall MC, Ahmed M. **eHealth interventions targeting nutrition, physical activity, sedentary behavior, or obesity in adults: A scoping review of systematic reviews**. *Obesity Reviews* (2021.0) **22**. DOI: 10.1111/obr.13295
9. Neubeck L, Redfern J, Fernandez R, Briffa T, Bauman A, Freedman SB. **Telehealth interventions for the secondary prevention of coronary heart disease: a systematic review**. *European Journal of Cardiovascular Prevention & Rehabilitation* (2009.0) **16** 281-9. DOI: 10.1097/HJR.0b013e32832a4e7a
10. Widmer RJ, Allison TG, Lennon R, Lopez-Jimenez F, Lerman LO, Lerman A. **Digital health intervention during cardiac rehabilitation: A randomized controlled trial**. *American Heart Journal* (2017.0) **188** 65-72. DOI: 10.1016/j.ahj.2017.02.016
11. Beatty AL, Fukuoka Y, Whooley MA. **Using mobile technology for cardiac rehabilitation: a review and framework for development and evaluation**. *Journal of the American Heart Association* (2013.0) **2**. DOI: 10.1161/JAHA.113.000568
12. Yardley L, Choudhury T, Patrick K, Michie S. **Current Issues and Future Directions for Research Into Digital Behavior Change Interventions**. *American Journal of Preventive Medicine* (2016.0) **51** 814-5. DOI: 10.1016/j.amepre.2016.07.019
13. **World Internet Users and 2021 Population Stats 2022**
14. Medlock S, Eslami S, Askari M, Arts DL, Sent D, Rooij SEd. **Health Information–Seeking Behavior of Seniors Who Use the Internet: A Survey**. *Journal of Medical Internet Research* (2015.0) **17** 1-11. DOI: 10.2196/jmir.3749
15. Paige SR, Krieger JL, Stellefson M, Alber JM. **eHealth literacy in chronic disease patients: An item response theory analysis of the eHealth literacy scale (eHEALS)**. *Patient Educ Couns* (2017.0) **100** 1-7. DOI: 10.1016/j.pec.2016.09.008
16. Higgins JP, Green S. (2011.0)
17. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD. **The PRISMA 2020 statement: an updated guideline for reporting systematic reviews**. *BMJ Brit Med J* (2021.0) **372** n71. DOI: 10.1136/bmj.n71
18. Burns PB, Rohrich RJ, Chung KC. **The levels of evidence and their role in evidence-based medicine**. *Plastic and Reconstructive Surgery* (2011.0) **128** 305-10. DOI: 10.1097/PRS.0b013e318219c171
19. Dale LP, Whittaker R, Jiang Y, Stewart R, Rolleston A, Maddison R. **Text Message and Internet Support for Coronary Heart Disease Self-Management: Results From the Text4Heart Randomized Controlled Trial**. *Journal of Medical Internet Research* (2015.0) **17**. DOI: 10.2196/jmir.4944
20. Frederix I, Hansen D, Coninx K, Vandervoort P, Vandijck D, Hens N. **Medium-term effectiveness of a comprehensive internet-based and patient-specific telerehabilitation program with text messaging support for cardiac patients: Randomized controlled trial**. *Journal of Medical Internet Research* (2015.0) **17**. DOI: 10.2196/jmir.4799
21. Martin SS, Feldman DI, Blumenthal RS, Jones SR, Post WS, McKibben RA. **mActive: A randomized clinical trial of an automated mHealth intervention for physical activity promotion**. *Journal of the American Heart Association* (2015.0) **4**. DOI: 10.1161/JAHA.115.002239
22. Skobel E, Knackstedt C, Martinez-Romero A, Salvi D, Vera-Munoz C, Napp A. **Internet-based training of coronary artery patients: the Heart Cycle Trial**. *Heart and Vessels* (2017.0) **32** 408-18. DOI: 10.1007/s00380-016-0897-8
23. Kamal AK, Khalid W, Muqeet A, Jamil A, Farhat K, Gillani SRA. **Making prescriptions “talk” to stroke and heart attack survivors to improve adherence: Results of a randomized clinical trial (The Talking Rx Study)**. *PLoS ONE* (2018.0) **13**. DOI: 10.1371/journal.pone.0197671
24. Choi BG, Dhawan T, Metzger K, Marshall L, Akbar A, Jain T. **Image-based mobile system for dietary management in an american cardiology population: Pilot randomized controlled trial to assess the efficacy of dietary coaching delivered via a smartphone app versus traditional counseling**. *JMIR mHealth and uHealth* (2019.0) **7**. DOI: 10.2196/10755
25. Dorje T, Zhao G, Tso K, Wang J, Chen Y, Tsokey L. **Smartphone and social media-based cardiac rehabilitation and secondary prevention in China (SMART-CR/SP): a parallel-group, single-blind, randomised controlled trial**. *Lancet Digital Health* (2019.0) **1** e363-e74. DOI: 10.1016/S2589-7500(19)30151-7
26. Broers ER, Widdershoven J, Denollet J, Lodder P, Kop WJ, Wetzels M. **Personalized eHealth Program for Life-style Change: Results from the “do Cardiac Health Advanced New Generated Ecosystem (Do CHANGE 2)” Randomized Controlled Trial**. *Psychosomatic Medicine* (2020.0) **82** 409-19. DOI: 10.1097/PSY.0000000000000802
27. Lunde P, Bye A, Bergland A, Grimsmo J, Jarstad E, Nilsson BB. **Long-term follow-up with a smartphone application improves exercise capacity post cardiac rehabilitation: A randomized controlled trial**. *European Journal of Preventive Cardiology* (2020.0) **27** 1782-92. DOI: 10.1177/2047487320905717
28. Song Y, Ren C, Liu P, Tao L, Zhao W, Gao W. **Effect of Smartphone-Based Telemonitored Exercise Rehabilitation among Patients with Coronary Heart Disease**. *Journal of Cardiovascular Translational Research* (2020.0) **13** 659-67. DOI: 10.1007/s12265-019-09938-6
29. 29The Cochrane Collaboration. Review Manager 5 (RevMan 5)
Copenhagen: Nordic Cochrane Centre; 2014. https://training.cochrane.org/online-learning/core-software/revman/revman-5-download.. (2014.0)
30. Sterne JAC, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I. **RoB 2: a revised tool for assessing risk of bias in randomised trials**. *British Medical Journal* (2019.0) **28** l4898. DOI: 10.1136/bmj.l4898
31. Schünemann HJ, Oxman AD, Vist GE, Higgins JP, Deeks JJ, Glasziou P, Higgins JP, Green S. *Cochrane Handbook for Systematic Reviews of Interventions Version 510 (updated March 2011)* (2011.0)
32. Campbell M, McKenzie JE, Sowden A, Katikireddi SV, Brennan SE, Ellis S. **Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline**. *BMJ Brit Med J* (2020.0) **368** l6890. DOI: 10.1136/bmj.l6890
33. Higgins JP, Altman DG, Sterne JA. (2011.0)
34. 34Ministry of Health Malaysia. Clinical Paractice Guideline: Management of Acute St Segment Elevation Myocardial Infarction, (STEMI)
4th Edition; 2019.. (2019.0)
35. **Prevention of Cardiovascular Disease: Guidelines for assessment and management of cardiovascular risk**. (2007.0)
36. Huff T, Boyd B, Jialal I. (2021.0)
37. Lorenzatti AJ, Toth PP. **New Perspectives on Atherogenic Dyslipidaemia and Cardiovascular Diseas**. *European Cardiology* (2020.0) **15** 1-9. DOI: 10.15420/ecr.2019.06
38. Wadhera RK, Steen DL, Khan I, Giugliano RP, Foody JM. **A review of low-density lipoprotein cholesterol, treatment strategies, and its impact on cardiovascular disease morbidity and mortality**. *Journal of Clinical Lipidology* (2016.0) **10** 472-89. DOI: 10.1016/j.jacl.2015.11.010
39. Cooper H, Hedges LV, Valentine JC. (2009.0) 632
40. Adler AJ, Martin N, Mariani J, Tajer CD, Owolabi OO, Free C. **Mobile phone text messaging to improve medication adherence in secondary prevention of cardiovascular disease**. *Cochrane Database of Systematic Reviews* (2017.0) **4** CD011851. DOI: 10.1002/14651858.CD011851.pub2
41. Paige SR, Stellefson M, Krieger JL, Anderson-Lewis C, Cheong JW, Stopka C. **Proposing a Transactional Model of eHealth Literacy: Concept Analysis**. *J Med Internet Res* (2018.0) **20** 1-16. DOI: 10.2196/10175
42. Boekel LCv, Peek ST, Luijkx KG. **Diversity in Older Adults’ Use of the Internet: Identifying Subgroups Through Latent Class Analysis**. *Journal of Medical Internet Research* (2017.0) **19**. DOI: 10.2196/jmir.6853
43. Xu Y, Ye H, Zhu Y, Du S, Xu G, Wang Q. **The efficacy of mobile health in alleviating risk factors related to the occurrence and development of coronary heart disease: A systematic review and meta-analysis**. *Clinical Cardiology* (2021.0) **44** 609-19. DOI: 10.1002/clc.23596
44. Yoon J, Lee M, Ahn JS, Oh D, Shin S-Y, Chang YJ. **Development and Validation of Digital Health Technology Literacy Assessment Questionnaire**. *Journal of Medical Systems* (2022.0) **46**. DOI: 10.1007/s10916-022-01800-8
45. Kanera IM, Bolman CAW, Willems RA, Mesters I, Lechner L. **Lifestyle-related effects of the web-based Kanker Nazorg Wijzer (Cancer Aftercare Guide) intervention for cancer survivors: a randomized controlled trial**. *Journal of Cancer Survivorship* (2016.0) **10** 883-97. DOI: 10.1007/s11764-016-0535-6
46. Lee MK, Yun YH, Park H-A, Lee ES, Jung KH, Noh D-Y. **A Web-based self-management exercise and diet intervention for breast cancer survivors: pilot randomized controlled trial**. *International Journal of Nursing Studies* (2014.0) **51** 1557-67. DOI: 10.1016/j.ijnurstu.2014.04.012
47. Larson JL, Rosen AB, Wilson FA. **The Effect of Telehealth Interventions on Quality of Life of Cancer Patients: A Systematic Review and Meta-Analysis**. *Telemedicine and e-Health* (2017.0) **24** 397-405. DOI: 10.1089/tmj.2017.0112
48. Yap JM, Tantono N, Wu VX, Yobas PK. **Effectiveness of technology-based psychosocial interventions on diabetes distress and health-relevant outcomes among type 2 diabetes mellitus: A systematic review and meta-analysis**. *Journal of Telemedicine and Telecare* (2021.0) **25** 1315-22. DOI: 10.1177/1357633X211058329
49. Ramachandran A, Snehalatha C, Ram J, Selvam S, Simon M, Nanditha A. **Effectiveness of mobile phone messaging in prevention of type 2 diabetes by lifestyle modification in men in India: a prospective, parallel-group, randomised controlled trial**. *The Lancet Diabetes & Endocrinology* (2013.0) **1** 191-8. DOI: 10.1016/S2213-8587(13)70067-6
50. Peimani M, Rambod C, Omidvar M, Larijani B, Ghodssi-Ghassemabadi R, Tootee A. **Effectiveness of short message service-based intervention (SMS) on self-care in type 2 diabetes: A feasibility study**. *Primary Care Diabetes* (2016.0) **10** 251-8. DOI: 10.1016/j.pcd.2015.11.001
51. Duan YP, Liang W, Guo L, Wienert J, Si GY, Lippke S. **Evaluation of a Web-Based Intervention for Multiple Health Behavior Changes in Patients With Coronary Heart Disease in Home-Based Rehabilitation: Pilot Randomized Controlled Trial**. *Journal of Medical Internet Research* (2018.0) **20**. DOI: 10.2196/12052
52. Swoboda CM, Miller CK, Wills CE. **Setting Single or Multiple Goals for Diet and Physical Activity Behaviors Improves Cardiovascular Disease Risk Factors in Adults With Type 2 Diabetes: A Pragmatic Pilot Randomized Trial**. *Diabetes educator* (2016.0) **42** 429-43. DOI: 10.1177/0145721716650043
53. Su JJ, Yu DSF, Paguio JT. **Effect of eHealth cardiac rehabilitation on health outcomes of coronary heart disease patients: A systematic review and meta-analysis**. *Journal of Advanced Nursing* (2019.0) **76** 754-72. DOI: 10.1111/jan.14272
54. Hossain MM, Tasnim S, Sharma R, Sultana A, Shaik AF, Faizah F. **Digital interventions for people living with non-communicable diseases in India: A systematic review of intervention studies and recommendations for future research and development**. *Digital Health* (2019.0) **16** 2055207619896153. DOI: 10.1177/2055207619896153
55. Lewis J, Ray P, Liaw S-T. **Recent Worldwide Developments in eHealth and mHealth to more Effectively Manage Cancer and other Chronic Diseases – A Systematic Review**. *Yearbook of Medical Informatics* (2016.0) **25** 93-108. DOI: 10.15265/IY-2016-020
56. Monzani D, Pizzoli SFM, Pravettoni G, Triberti S. *P5 EHealth: An Agenda for the Health Technologies of the Future* (2020.0) 33-51. DOI: 10.1007/978-3-030-27994-3_3
|
---
title: Outcomes of ST Segment Elevation Myocardial Infarction without Standard Modifiable
Cardiovascular Risk Factors – Newer Insights from a Prospective Registry in India
authors:
- Gnanaraj Justin Paul
- Sabarish Sankaran
- Karthikaa Saminathan
- Mohamed Iliyas
- Suryakanth Sethupathy
- Sivasubramaniam Saravanan
- Salai Sudhan Prabhu
- Sijoy Kurian
- Sandeep Srinivas
- Polavarappu Anurag
- Kumaran Srinivasan
- Elavarasi Manimegalai
- Swaminathan Nagarajan
- Rajasekar Ramesh
- P. M. Nageswaran
- Venkatesan Sangareddi
- Ravishankar Govindarajulu
journal: Global Heart
year: 2023
pmcid: PMC10022543
doi: 10.5334/gh.1189
license: CC BY 4.0
---
# Outcomes of ST Segment Elevation Myocardial Infarction without Standard Modifiable Cardiovascular Risk Factors – Newer Insights from a Prospective Registry in India
## Abstract
### Objectives:
Patients with ST elevation myocardial infarction (STEMI) without standard modifiable cardiovascular risk factors (SMuRFs; dyslipidaemia, hypertension, diabetes mellitus and smoking) are reported to have a worse clinical outcome compared to those with SMuRFs. However, robust prospective data and low-and middle-income country perspective are lacking. We aimed to study the patients with first STEMI and assess the influence of SMuRFs on clinical outcomes by comparing the patients with and without SMuRFs.
### Methods:
We included all consecutive STEMI patients without prior coronary artery disease enrolled in the Madras Medical College STEMI Registry from September 2018 to October 2019. We collected baseline clinical characteristics, revascularisation strategies and clinical outcome. We analysed suboptimal self-reported sleep duration as a 5th extended SMuRF (eSMuRF). Primary outcome was in-hospital mortality. Secondary outcomes included in-hospital complications and one-year all-cause mortality.
### Results
Among 2,379 patients, 605 patients ($25.4\%$) were SMuRF-less. More women were SMuRF-less than men ($27.1\%$ vs $22.1\%$; $$P \leq 0.012$$). SMuRF-less patients were older (57.44 ± 13.95 vs 55.68 ± 11.74; $P \leq 0.001$), more often former tobacco users ($10.4\%$ vs $5.0\%$; $P \leq 0.001$), with more anterior wall MI ($62.6\%$ vs $52.1\%$; $$P \leq 0.032$$). The primary outcome [in-hospital mortality ($10.7\%$ vs $11.3\%$; $$P \leq 0.72$$)] and secondary outcomes [in-hospital complications ($29.1\%$ vs $31.7\%$; $$P \leq 0.23$$) and one-year all-cause mortality ($22.3\%$ vs $22.7\%$; $$P \leq 0.85$$)] were similar in both groups. Addition of suboptimal self-reported sleep duration as a 5th eSMuRF yielded similar results.
### Conclusions
$25\%$ of first STEMI patients were SMuRF-less. Clinical outcomes of patients without SMuRFs were similar to those with SMuRFs. Suboptimal sleep duration did not account for the risk associated with the SMuRF-less status.
## Introduction
Coronary artery disease is a major cause of death worldwide. Cardiovascular disease is the most common cause of mortality in India, accounting for a third of the certified deaths [1]. Precursors of coronary artery disease (CAD) have been extensively studied and causal risk factors have been identified. The best established modifiable risk factors like hypertension, diabetes, dyslipidaemia and smoking have been the focus of many risk scoring systems for coronary artery disease [234]. Several risk models that integrate information on conventional cardiovascular risk factors exist [56]. The INTERHEART study suggested that nine potentially modifiable risk factors, including diabetes, hypertension, dyslipidaemia and smoking, could account for >$90\%$ of population-attributable risk of coronary artery disease [7]. Recognition and management of these risk factors have together led to significant improvements in prevention and therapy [8].
However, it is well known that myocardial infarction also occurs among persons without these traditional risk factors [3]. Contemporary data has shown that as much as $15\%$ to $25\%$ of patients with ST segment elevation myocardial infarction (STEMI) do not have the standard modifiable risk factors (SMuRFs) [91011]. It has been also observed that the proportion of SMuRF-less patients with STEMI is on an increasing trend over the last few decades [12]. Further, many studies have brought out the surprising observation of higher mortality in SMuRF-less patients compared with those with SMuRFs [9111213]. However, this information comes predominantly from retrospective analyses of studies conducted in high-income nations. Hence, we planned this prospective study to find the proportion of patients with STEMI who are SMuRF-less, compare their in-hospital and intermediate term mortality with those with SMuRFs, and offer a low-and middle-income country perspective.
## Data source and study population
Madras Medical College STEMI (M-STEMI) *Registry is* a prospective registry enrolling acute STEMI patients above 18 years of age seeking care in a public hospital in a metropolitan city in India. All consecutive patients with first diagnosis of STEMI enrolled from September 2018 to October 2019 were included in this analysis after getting their informed consent. Diagnosis of STEMI was based on classic chest pain and diagnostic ST elevation as per standard guidelines [14]. Patients with ST elevation not related to acute coronary syndrome, like Takotsubo cardiomyopathy and acute pericarditis, were excluded. Patients with prior CAD of any form were excluded. The study protocol was designed in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans. Written informed consent was obtained from all study participants by the authors.
## Data Collection Instrument
The data collection instrument was developed by the authors in hard copy format convertible into corresponding Microsoft excel tables. The data collection instrument and strategy were put to use in 1,500 consecutive patients admitted with STEMI from October 2017 to August 2018. The data collection instrument and strategy underwent multiple revisions and improvements during this period of trial enrolment of patients to ensure no missing variables. Final data collection was done using a hardcopy of the data collection instrument (Supplementary file-1), which was subsequently updated into an online Microsoft excel spreadsheet by one of the authors.
Baseline demographic factors, cardiovascular risk factors, comorbid conditions and present symptoms and their chronology were collected prospectively for all patients. All patients had detailed clinical evaluation, 12-lead electrocardiogram, echocardiogram and risk stratification on admission. Standard echocardiographic techniques were used to measure left ventricular ejection fraction [15]. Tricuspid annular plane systolic excursion (TAPSE) < 17 mm was used to diagnose right ventricular dysfunction [16]. Management, including revascularisation, was at the discretion of the treating cardiologist. Details of revascularisation modalities, like fibrinolysis, primary percutaneous coronary intervention (PCI), pharmaco-invasive approach or delayed PCI (after 24 hr but before discharge), and details of in-house complications were noted.
## Definitions
Patients who have none of the four standard modifiable cardiovascular risk factors (SMuRFs): hypertension, hyperlipidaemia, diabetes mellitus and current tobacco use, were considered as the SMuRF-less group. Patients with one or more of these four risk factors were considered as the SMuRF-plus group. Current tobacco use was defined as regular use of smoking or non-smoking tobacco for at least the previous one year. Patients who had stopped tobacco use at least 12 months before were labelled as former tobacco users. Considering the high prevalence of smokeless tobacco use in India, we sought for smokeless tobacco use and included it under tobacco use [17]. Hypertension was defined as having an earlier diagnosis of hypertension or prior/current antihypertensive drug therapy. Diabetes was defined as having an earlier diagnosis of diabetes or prior/current use of hypoglycaemic therapy. Dyslipidaemia was defined as having earlier diagnosis of dyslipidaemia or prior/ongoing lipid lowering pharmacologic therapy.
In addition to the four main SMuRFs, suboptimal sleep duration, defined as self-reported sleep duration ≤ 6 hours and > 9 hours, was evaluated as a potential fifth modifiable risk factor. The group with any of these five modifiable risk factors was termed as the extended SMuRF (eSMurRF) group, and the group with none of these as the extended SMuRF-less (eSMurRF-less) group.
## Discharge and follow-up
All patients were discharged with aspirin 150 mg, clopidogrel 75 mg and atorvastatin 80 mg as per protocol, unless contraindicated. All patients were prescribed betablockers and angiotensin converting enzyme inhibitors as permitted by the discharge hemodynamics and biochemistry. Follow-up details were collected at 1, 3, 6 and 12 months. Patients who did not turn up to the outpatient department in time were reminded by telephonic calls by our dedicated and trained follow-up team. Postal letters in vernacular language were sent to the patients who were not reachable by telephone. Our follow-up team performed house visit for the final defaulters.
## Outcome
In-hospital mortality is the primary outcome. The secondary outcomes include a composite of in-hospital complications and one-year all-cause mortality. We are continuing active follow-up of patients.
Our study showed that in-hospital mortality was equal in both the groups with and without SMuRFs. Mortality in acute coronary syndromes has been reported to be worse in SMuRF-less patients compared to those with SMuRFs [11121326]. Vernon et al. observed a high in-hospital mortality in the SMuRF-less group, however, with similar rates of major adverse cardiac events, cardiogenic shock and in-hospital reinfarctions, and suggested that the reason for the observed increased mortality needs to be investigated further [12]. The study by Figtree et al. also found higher in-hospital mortality, with similar rates of reinfarctions and heart failure in the SMuRF-less group. In the absence of data on cardiac arrhythmia, they had suggested cardiac arrhythmia as a possible contributor of this increased mortality in SmuRF-less group.
Our study, with a similar proportion of cardiac arrhythmia in both the groups, did not support this postulation. Though we found that SMuRF-less status was more common in women, mortality was similar in women with and without SMuRFs, unlike earlier observations [11]. The information obtained from this prospective study does not contradict the message of the earlier studies, but rather confirms the fact that absence of traditional risk factor does not imply good outcome. The adverse outcome in patients without SMuRFs may be because they harbour unidentified/quantified risk factors and they lack a therapeutic target, where ironically the SMuRFs group has an advantage.
## Statistical Analysis
Differences between the SMuRF-less group and SMuRF-plus groups in baseline demographics, clinical parameters, reperfusion therapy offered, in-hospital course and complications and follow-up outcome were analysed. An additional similar analysis was performed to find the differences between the eSMuRF and eSMuRF-less groups. Categorical variables are presented as frequencies and percentages and compared using Pearson’s Chi-squared test or Fisher’s Exact Test. Continuous variables are presented as mean, standard deviation (SD) and median (interquartile range), and are compared using student’s t-test (normal distribution) or Mann-Whitney test (non-normal). Covariates with $p \leq 0.10$ on univariable analysis were planned to be included in the final multivariable model.
We had a prespecified plan to include any covariates with $p \leq 0.10$ on univariate testing in a final multivariate model. Multivariable analysis was done with logistic regression. The results of regression analyses are expressed as an odds ratio (OR) with respective confidence interval (CI) and p-values. Significance was assumed at a two-sided value of $p \leq 0.05.$ Analyses were performed using SPSS version 28.0 (SPSS for Mac, Version 28.0. Armonk, NY: IBM Corp Released 2021).
## Study population and baseline features
Between September 2018 to October 2019, 2,499 adults with acute STEMI were enrolled in the M-STEMI registry. Of those, 120 patients with a history of prior CAD were excluded. Of the remaining 2,379 patients with first STEMI analysed, 605 patients with no documented SMuRFs constituted the SMuRF-less group. The rest formed the SMuRFs group. Ninety-nine percent of the study population belonged to the lower socioeconomic category. The Baseline differences in the distribution of demographic variables between the groups are given in Table 1.
**Table 1**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 |
| --- | --- | --- | --- | --- |
| VARIABLE | TOTAL (n = 2379) | NO SMURF (n = 605) | ≥1 SMURF (n = 1774) | P VALUE |
| Age | | | | |
| (Mean age ± SD) | 56.13 ± 12.37 | 57.44 ± 13.95 | 55.68 ± 11.74 | <0.001 |
| <60 years | 1392 (58.5%) | 323 (53.4%) | 1069 (60.3%) | 0.003 |
| | | | | 0.003 |
| ≥60 years | 987 (41.5%) | 282 (46.6%) | 705 (39.7%) | 0.003 |
| Sex | | | | |
| Male | 1823 (76.6%) | 441 (72.9%) | 1382 (77.9%) | 0.012 |
| | | | | 0.012 |
| Female | 556 (23.4%) | 164 (27.1%) | 392 (22.1%) | 0.012 |
| Risk Factors | | | | |
| Hypertension | 810 | 0 | 810 | |
| Diabetes | 933 | 0 | 933 | |
| Dyslipidemia | 34 | 0 | 34 | |
| Current tobacco user | 834 | 0 | 834 | |
| Former tobacco user | 152 (6.4%) | 63 (10.4%) | 89 (5.0%) | <0.001 |
| Alcohol | 838 (35.2%) | 113 (18.7%) | 725 (40.9%) | <0.001 |
| F/h/o CAD | 45 | 15 (2.5%) | 30 (1.7%) | 0.219 |
| Sleep duration per day | | | | |
| Duration (mean ± SD) | 7.65 ± 0.89 | 7.67 ± 0.86 | 7.64 ± 0.90 | 0.439 |
| ≤ 6 hours | 300 (12.6%) | 72 (11.9%) | 228 (12.9%) | 0.810 |
| | | | | 0.810 |
| >6 to ≤ 7 hours | 364 (15.3%) | 88 (14.5%) | 276 (15.6%) | 0.810 |
| | | | | 0.810 |
| >7 to ≤ 8 hours | 1570 (66%) | 411 (67.9%) | 1159 (65.3%) | 0.810 |
| | | | | 0.810 |
| >8 to ≤ 9 hours | 127 (5.3%) | 29 (4.8%) | 98 (5.5%) | 0.810 |
| | | | | 0.810 |
| >9 hours | 18 (0.8%) | 5 (0.8%) | 13 (0.7%) | 0.810 |
| CKD | 26 (1.1%) | 5 (0.8%) | 21 (1.2%) | 0.465 |
| CVA | 42 (1.8%) | 4 (0.7%) | 38 (2.1%) | 0.017 |
| COPD | 21 (0.9%) | 6 (1%) | 15 (0.8%) | 0.740 |
The SMuRF-less group had a higher mean age (57.4 vs $55.7\%$; $P \leq 0.001$) with a larger proportion of patients above 60 years of age ($46.6\%$ vs $39.7\%$) compared to the SMuRF-plus group. The study participants were predominantly men ($76.6\%$). However, a larger proportion of women were SMuRF-less compared to men (29.5 % vs $24.2\%$; $$P \leq 0.012$$). The age and sex differences were not significant in the multivariable analysis. Figure 1 shows the number of SMuRFs present in the study population. The SMuRF-less group had a higher proportion of former tobacco users and lower proportion of ethanol users. Sleep duration was not significantly different between the groups. The proportions of patients with chronic kidney disease (CKD) and chronic obstructive pulmonary disease (COPD) were similar, while the proportion with a cerebro-vascular accident (CVA) was significantly lower in the SMuRF-less group.
**Figure 1:** *Number of SMuRFS identified in the enrolled patients.Note: SMuRF: Standard modifiable cardiovascular risk factor.*
## Clinical presentation and reperfusion strategies
The time from symptom onset to presentation in the hospital and presentation Killip class was similar in both the groups. Though a higher proportion of patients in the SMuRF-less group presented with anterior wall MI, the mean left ventricular ejection fraction was similar in both groups (Table 2).
**Table 2**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 |
| --- | --- | --- | --- | --- |
| PARAMETER ANALYZED | TOTAL (n = 2379) | NO SMURF (n = 605) | ≥1 SMURF (n = 1774) | P VALUE |
| Time Window (symptom onset to presentation at hospital) | Time Window (symptom onset to presentation at hospital) | Time Window (symptom onset to presentation at hospital) | Time Window (symptom onset to presentation at hospital) | Time Window (symptom onset to presentation at hospital) |
| Time window (hours) | 13.23 ± 17.47 | 13.08 ± 17.25 | 13.28 ± 17.25 | 0.809 |
| <6 hours | 1234 (51.9%) | 313 (51.7%) | 921 (74.6%) | 0.794 |
| | | | | 0.794 |
| 6–12 hours | 534 (22.4%) | 143 (23.6%) | 391 (22%) | 0.794 |
| | | | | 0.794 |
| 12–24 hours | 315 (13.2%) | 79 (13.1%) | 236 (13.3%) | 0.794 |
| | | | | 0.794 |
| >24 hours | 296 (12.4%) | 70 (11.6%) | 226 (12.7%) | 0.794 |
| Location of infarction | | | | |
| Anterior | 1374 (57.8%) | 379 (62.6%) | 995 (56.1%) | 0.005 |
| | | | | 0.005 |
| Non anterior | 1005 (42.2%) | 226 (37.4%) | 779 (43.9%) | 0.005 |
| Killip Class | | | | |
| Class I | 1773 (74.5%) | 466 (77.0%) | 1307 (73.6%) | 0.105 |
| | | | | 0.105 |
| Class II, III & IV | 606 (25.5%) | 139 (23.0%) | 467 (23.6%) | 0.105 |
| Left ventricular ejection fraction | Left ventricular ejection fraction | | | |
| Mean EFa ± SDb | 46.1 ± 8.78 | 46.13 ± 8.9 | 46.10 ± 8.45 | 0.938 |
| ≤ 40% | 676 (28.4%) | 163 (26.9%) | 513 (28.9%) | 0.643 |
| | | | | 0.643 |
| 41–54% | 1260 (53.0%) | 326 (53.9%) | 934 (52.6%) | 0.643 |
| | | | | 0.643 |
| >54% | 443 (18.6%) | 116 (19.2%) | 327 (18.4%) | 0.643 |
| Right ventricular function | Right ventricular function | | | |
| TAPSEc mean ± SD | 17.81 ± 2.46 | 18.05 ± 2.3 | 17.74 ± 2.51 | 0.03 |
| TAPSE < 17 | 297 (12.5%) | 57 (9.4%) | 240 (13.5%) | 0.008 |
| Primary reperfusion strategyg | Primary reperfusion strategyg | | | |
| Fibrinolysis | 1242 (52.2%) | 320 (52.9%) | 922 (52%) | 0.852 |
| | | | | 0.852 |
| SKd | 1097 (46.1%) | 278 (46%) | 819 (46.2%) | 0.852 |
| | | | | 0.852 |
| TNKe | 121 (5.1%) | 35 (5.8%) | 86 (4.8%) | 0.852 |
| | | | | 0.852 |
| Reteplasef | 24 (1.0%) | 7 (1.2%) | 17 (1.0%) | 0.852 |
| | | | | 0.852 |
| Primary PCI | 238 (10%) | 63 (10.4%) | 175 (9.9%) | 0.852 |
| | | | | 0.852 |
| Neither | 899 (37.8%) | 222 (36.7%) | 677 (38.2%) | 0.852 |
| Overall Reperfusion Strategyh | Overall Reperfusion Strategyh | Overall Reperfusion Strategyh | Overall Reperfusion Strategyh | Overall Reperfusion Strategyh |
| Primary/PI PCI | 354 (14.9%) | 90 (14.9%) | 264 (14.9%) | 0.961 |
| | | | | 0.961 |
| Fibrinolysis only (no PCI) | 1035 (43.5%) | 266 (44.0%) | 769 (43.3%) | 0.961 |
| | | | | 0.961 |
| Neither | 990 (41.6%) | 249 (41.2%) | 741 (41.8%) | 0.961 |
The proportion of patients with inferior infarction and right ventricular dysfunction was lower in the SMuRF-less group. Fibrinolysis ($51\%$) was the predominant mode of reperfusion, used with only $10\%$ receiving primary PCI. However, the proportion of patients receiving the various modes of reperfusion (Primary PCI, pharmaco-invasive PCI, delayed PCI and standalone fibrinolysis) was similar in both groups.
The window period of presentation was similar in both groups, suggesting that being SMuRF-less did not induce delay in seeking medical help. Similar to earlier observations, we observed a higher proportion of patients with anterior wall myocardial infarction and a non-significant higher involvement of LAD as culprit vessel in the SMURF-less group [1112]. The reasons behind LAD disease being more common in SMuRF-less patients are unclear and open to speculation. As a corollary we observed that inferior wall infarction, and consequently RV dysfunction, was more common in the SMuRF group.
## Angiographic analysis
Angiographic details were available for 1,089 of the 2,379 patients. LAD involvement was non-significantly higher in the SMuRF-less group. There was no significant difference in the culprit lesion profile and proportion of patients with multivessel disease between both the groups (Table 3).
**Table 3**
| Unnamed: 0 | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 | Unnamed: 4 |
| --- | --- | --- | --- | --- |
| PARAMETER | TOTAL (n = 1089) | NO SMURF (n = 270) | ≥1 SMURF (n = 819) | P VALUE |
| Culprit Lesions | Culprit Lesions | Culprit Lesions | Culprit Lesions | Culprit Lesions |
| LMCA | 3 (0.2%) | 1 (0.4%) | 2 (0.2%) | 0.605 |
| | | | | 0.605 |
| LAD | 554 (50.9%) | 145 (53.7%) | 409 (49.9%) | 0.605 |
| | | | | 0.605 |
| LCX | 51 (4.7%) | 8 (3.0%) | 43 (5.3%) | 0.605 |
| | | | | 0.605 |
| RCA | 218 (20.0%) | 55 (20.4%) | 163 (19.9%) | 0.605 |
| | | | | 0.605 |
| Unspecified | 263 (24.2%) | 61 (22.6%) | 202 (24.7%) | 0.605 |
| Single Vessel Disease | 508 (46.6%) | 127 (47%) | 381 (46.5%) | 0.883 |
| | | | | 0.883 |
| Multivessel Disease | 581 (53.4%) | 143 (53%) | 438 (53.5%) | 0.883 |
## In-hospital Outcome
Of the 2,379 patients included in the analysis, 265 patients ($11.1\%$) died in the hospital. There was no difference in the in-hospital course or complications between the groups with and without SMuRFs (Table 4). SMuRF-less status did not alter the risk for in-hospital mortality in the stratified analysis done according to age, sex or location of MI (Table 5).
## 12 months outcome
Follow-up information at 12 months was available for $85.5\%$ of the patients included in the study analysis. Of the total, 344 patients were lost to follow-up. The SMuRF status and eSMuRF status of patients who were lost to follow-up and were available for follow-up were similar. ( Supplementary Table-1). Of the 2,114 patients discharged alive, there were 194 additional deaths reported by 12 months. Thus, one-year all-cause mortality was $22.6\%$. Post discharge mortality ($9.8\%$ vs $9.4\%$; P = NS) and all-cause mortality at 12 months ($22.3\%$ vs $22.7\%$, P = NS) were similar in both the groups (Table 4).
## Extended SMuRF analysis
With self-reported suboptimal sleep duration added as a fifth modifiable risk factor, there were 1,851 patients ($87.8\%$) with at least one of the five eSMuRFs and 528 patients ($22.2\%$) without any of these. The eSMURF analysis showed similar results to the main SMuRF analysis, with no significant differences in the in-hospital or 12-month outcome between the groups with and without eSMuRFs (Supplementary Table-2).
## Discussion
Our study has four main findings (Figure 2). First, the incidence of STEMI without SMuRFs is high ($25.4\%$) in patients from low-and middle-income countries. Second, the in-hospital mortality, complications and twelve-month mortality in SMuRF-less STEMI patients was similar to those with SMuRFs. Third, more women with STEMI were SMURF-less than men. Fourthly, suboptimal-sleep duration, a recently identified modifiable risk factor, did not account for the risk associated with SMuRFless STEMI.
**Figure 2:** *Central illustration-methods and outcome of the study.Note: SMuRF—Standard modifiable cardiovascular risk factor; eSMURF—extended standard modifiable cardiovascular risk factor; STEMI—ST segment elevation myocardial infarction.*
## Incidence and baseline demographics
The proportion of SmuRF-less STEMI in our study ($25.4\%$) was higher than observed in earlier studies ($14.9\%$ & $19\%$) [1112]. Since the proportion of SMuRF-lessness in STEMI is likely to vary depending on the vigour at which risk factors are actively looked for in the community, SMuRFs could have been undetected in our study population with poor access to preventive health care, overestimating the SMuRF-less status.
As our study data was collected over 22 months only, we could not comment on the variably reported increasing trend of prevalence of SMuRF-less STEMI over years [91112]. Similar to earlier observations, our patients without SMuRFs were older than those with SMuRFs [1112]. However, age did not influence the neutral effect of SMuRF-less status on the outcome. Our observation of a higher proportion of STEMI in women being SMuRF-less than in men is different from the earlier observation of SMuRF-less status being more common in men than women [91112]. The reason for this observation is unclear, but potentially hypothesis generating. The possibilities include, but are not limited to, women being less likely to undergo preventive health evaluations [18], and having a higher number of ‘yet-to-be-identified atherosclerotic risk factors’ compared to men, particularly from low-and middle-income countries (LMICs).
## Risk Factors and comorbid conditions
Though the proportion of patients with hypertension, diabetes and current tobacco use in our study was similar to the earlier studies, the proportion of patients with dyslipidaemia was very low. This could be reflective of the unmet needs in diagnosis and management of dyslipidaemia in LMICs [19] compounded by the existing risk factor identification and prevention programmes’ focussing more on hypertension and diabetes than dyslipidaemia [2021]. Similar to earlier studies, the proportion of former tobacco users was high in the SMuRF-less group. Though we defined former tobacco users as patients who had stopped tobacco use at least 12 months before, it has been shown that the CVD risk remains significantly high in former smokers compared to never smokers for beyond 5 years after quitting [22]. It is possible that former tobacco use could account for some of the risk attributable to the first STEMI in the SMuRF-less patients. The proportion of former tobacco users in our population is low ($6.3\%$) compared to the earlier studies ($23\%$ & $27\%$) [911]. This could indicate a smaller contribution from former tobacco use in the atherosclerotic risk of the SMuRF-less group from an LMIC population when compared to developed nations. Though earlier data found comorbid conditions less often in SMuRF-less patients [11], we observed a similar proportion of COPD and CKD in both groups, with only CVA being found in a lower proportion. We did not collect data on obesity, cancer or peripheral vascular disease.
## Revascularisation
Only $10\%$ of our patients underwent primary PCI in our study. Though this number is low, it is not different from earlier reported data from low- and middle-income countries [232425]. However, the proportion of patients receiving various modes of reperfusion therapy (primary PCI, fibrinolysis, pharmaco-invasive therapy, delayed PCI and no revascularisation) was similar between both the groups and hence is unlikely to influence the conclusions of the study.
## eSMuRF
Association between reduced self-reported sleep duration and coronary artery disease and adverse outcome [272829] has been reported recently. The distribution of sleeping hours was equal in both the groups in our study. The additional eSMuRF analysis, with suboptimal self-reported sleep duration added as the fifth modifiable risk factor, yielded results similar to the main outcome. Earlier observations have suggested that STEMI in the SMuRF-less group could not be explained by obesity and family history of premature atherosclerotic coronary artery disease [12]. This study adds information that this finding could not be explained by suboptimal sleep duration either. The role of non-conventional and lesser studied risk factors, like lipoprotein(a), high-sensitivity C-reactive protein, psychosocial risk factors, access to preventive health care and education, air particulate matter, etc. in contributing to the risk in SMuRF-less STEMI needs further evaluation.
## Strengths and limitations
Our study has the strength of being a large prospective study, evaluating the role of SMuRF-less status in STEMI. This study also evaluated the role of suboptimal sleep duration as an additional fifth modifiable cardiovascular risk factor in STEMI. Our results are not generalisable to populations with improved and widespread preventive health care availability. Our study has several limitations. Data on potential confounders, such as baseline, in-hospital and discharge pharmacotherapy, history of malignancy and peripheral occlusive vascular disease, were not routinely collected and hence could not be analysed as covariates. Data on risk factors like family history of premature coronary artery disease, body weight, body mass index, waist circumference, HBA1C, lipoprotein (a), high-sensitivity C-reactive protein, socio-cultural factors, or psychosocial risk factors were not available. Information on possible differences in access to or use of preventive healthcare was not available. Though the relationship between the risk factors and MI is loglinear, with no identified threshold above which the likelihood of MI increases, we chose to stick with the conventional definitions of the SMuRFs with thresholds and specific cut offs. This helped us to categorise them to two different groups for comparison purpose. However, we acknowledge the global cardiovascular risk assessment should ideally consider the linear relationship of different risk factors with morbidity and mortality outcomes. Coronary angiogram was not done for all patients, bringing in a possibility of our population having patients with spontaneous coronary artery dissection (SCAD) and myocardial infarction with non-obstructive coronary arteries (MINOCA) as a potential limitation. Twelve-month follow-up data was available for only $85.5\%$ of the study participants. Though we could capture the follow-up event data, the date of event was not available for all patients due to the COVID-19 pandemic. Hence, the follow-up outcome could not be presented in a Cox regression (survival) model.
## Conclusion
We observed that one-fourth of patients with STEMI were SMuRF-less. More women were SMuRF-less than men. The clinical outcomes of patients with STEMI without SMuRFs was similar to those with SMuRFs, highlighting that being SMuRF-less does not necessarily confer a lower risk in STEMI. This underscores the need for evidence based on timely revascularisation therapy and pharmaco-therapy for both patients with and without SMuRFs, and the need for studies to evaluate the role of non-conventional and yet-to-be-identified risk factors in STEMI.
## Data Accessibility Statement
The deidentified data underlying this article can be shared on reasonable request to the corresponding author. However, data shall be shared after approval from the Institutional Ethics Committee of Madras Medical College.
## Additional Files
The additional files for this article can be found as follows:
## Ethics and Consent
The study and its protocol were approved by the “Institutional Ethics committee” of Madras Medical College.
## Competing Interests
The authors have no competing interests to declare.
## Author Contributions
JPG and MI conceived the idea and planned the study. JPG, SS-1, KS, MI, SS-2, SS-3, SSP, SK, SS-4 and AP contributed to data collection. JPG, SS-1 and KS contributed to data management and analysis. JPG wrote the first draft of the manuscript. All other authors contributed to improvement of the original draft, proofreading and approval of the final version of the manuscript. JPG stands guarantee to the overall content of the manuscript. All authors agree to be accountable for the accuracy and integrity of the work.
## References
1. 1Government of India. MCCD 2019 Cause of death. Census of India website: Office of the Registrar General & Census Commissioner, India. https://censusindia.gov.in/2011-Common/mccd.html (accessed 4 Mar 2022).
2. Greenland P, Knoll MD, Stamler J, Neaton JD, Dyer AR, Garside DB. **Major risk factors as antecedents of fatal and nonfatal coronary heart disease events**. *JAMA* (2003.0) **290** 891-897. DOI: 10.1001/jama.290.7.891
3. Ridker PM. **Evaluating novel cardiovascular risk factors: Can we better predict heart attacks?**. *Ann Intern Med* (1999.0) **130** 933-937. DOI: 10.7326/0003-4819-130-11-199906010-00018
4. Conroy RM, Pyörälä K, Fitzgerald AP, Sans S, Menotti A, De Backer G. **Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project**. *Eur Heart J* (2003.0) **24** 987-1003. DOI: 10.1016/S0195-668X(03)00114-3
5. Anderson KM, Odell PM, Wilson PWF, Kannel WB. **Cardiovascular disease risk profiles**. *Am Heart J* (1991.0) **121** 293-298. PMID: 1985385
6. Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, Minhas R, Sheikh A. **Predicting cardiovascular risk in England and Wales: prospective derivation and validation of QRISK2**. *BMJ* (2008.0) **336** 1475-1482. DOI: 10.1136/bmj.39609.449676.25
7. Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F. **Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): Case-control study**. *Lancet Lond Engl* (2004.0) **364** 937-952. DOI: 10.1016/S0140-6736(04)17018-9
8. Piironen M, Ukkola O, Huikuri H, Havulinna AS, Koukkunen H, Mustonen J. **Trends in long-term prognosis after acute coronary syndrome**. *Eur J Prev Cardiol* (2017.0) **24** 274-280. DOI: 10.1177/2047487316679522
9. Vernon ST, Coffey S, Bhindi R, Soo Hoo SY, Nelson GI, Ward MR. **Increasing proportion of ST elevation myocardial infarction patients with coronary atherosclerosis poorly explained by standard modifiable risk factors**. *Eur J Prev Cardiol* (2017.0) **24** 1824-1830. DOI: 10.1177/2047487317720287
10. Mazhar J, Figtree G, Vernon ST, Galougahi KK, Carlo J, Nissen SE. **Progression of coronary atherosclerosis in patients without standard modifiable risk factors**. *Am J Prev Cardiol* (2020.0) **4** 100116. DOI: 10.1016/j.ajpc.2020.100116
11. Figtree GA, Vernon ST, Hadziosmanovic N, Sundström J, Alfredsson J, Arnott C. **Mortality in STEMI patients without standard modifiable risk factors: a sex-disaggregated analysis of SWEDEHEART registry data**. *The Lancet* (2021.0) **397** 1085-1094. DOI: 10.1016/S0140-6736(21)00272-5
12. Vernon ST, Coffey S, D’Souza M, Chow CK, Kilian J, Hyun K. **ST-segment–elevation myocardial infarction (STEMI) patients without standard modifiable cardiovascular risk factors—how common are they, and what are their outcomes?**. *J Am Heart Assoc* (2019.0) **8**. DOI: 10.1161/JAHA.119.013296
13. Canto JG, Kiefe CI, Rogers WJ, Peterson ED, Frederick PD, French WJ. **Number of coronary heart disease risk factors and mortality in patients with first myocardial infarction**. *JAMA* (2011.0) **306** 2120-2127. DOI: 10.1001/jama.2011.1654
14. Ibanez B, James S, Agewall S, Antunes MJ, Bucciarelli-Ducci C, Bueno H. **2017 ESC Guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation: The Task Force for the management of acute myocardial infarction in patients presenting with ST-segment elevation of the European Society of Cardiology (ESC)**. *Eur Heart J* (2018.0) **39** 119-177. PMID: 28886621
15. Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L. **Recommendations for cardiac chamber quantification by echocardiography in adults: An Update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging**. *Eur Heart J – Cardiovasc Imaging* (2015.0) **16** 233-271. DOI: 10.1093/ehjci/jev014
16. Rudski LG, Lai WW, Afilalo J, Hua L, Handschumacher MD, Chandrasekaran K. **Guidelines for the echocardiographic assessment of the right heart in adults: A report from the American Society of Echocardiography: Endorsed by the European Association of Echocardiography, a registered branch of the European Society of Cardiology, and the Canadian Society of Echocardiography**. *J Am Soc Echocardiogr* (2010.0) **23** 685-713. DOI: 10.1016/j.echo.2010.05.010
17. Yadav A, Singh PK, Yadav N, Kaushik R, Chandan K, Chandra A. **Smokeless tobacco control in India: policy review and lessons for high-burden countries**. *BMJ Glob Health* (2020.0) **5**. DOI: 10.1136/bmjgh-2020-002367
18. Healy B. **The Yentl Syndrome**. *N Engl J Med* (1991.0) **325** 274-276. DOI: 10.1056/NEJM199107253250408
19. Marcus ME, Ebert C, Geldsetzer P, Theilmann M, Bicaba BW, Andall-Brereton G. **Unmet need for hypercholesterolemia care in 35 low- and middle-income countries: A cross-sectional study of nationally representative surveys**. *PLoS Med* (2021.0) **18**. DOI: 10.1371/journal.pmed.1003841
20. Prabhakaran D, Jeemon P, Roy A. **Cardiovascular diseases in India: Current epidemiology and future directions**. *Circulation* (2016.0) **133** 1605-1620. DOI: 10.1161/CIRCULATIONAHA.114.008729
21. Ajay. **Development of a smartphone-enabled hypertension and diabetes mellitus management package to facilitate evidence-based care delivery in primary healthcare facilities in India: The mPower Heart Project**. *Journal of the American Heart Association*
22. Duncan MS, Freiberg MS, Greevy RA, Kundu S, Vasan RS, Tindle HA. **Association of smoking cessation with subsequent risk of cardiovascular disease**. *JAMA* (2019.0) **322** 642-650. DOI: 10.1001/jama.2019.10298
23. Xavier D, Pais P, Devereaux PJ, Xie C, Prabhakaran D, Reddy KS. **Treatment and outcomes of acute coronary syndromes in India (CREATE): a prospective analysis of registry data**. *The Lancet* (2008.0) **371** 1435-1442. DOI: 10.1016/S0140-6736(08)60623-6
24. Srimahachota S, Boonyaratavej S, Kanjanavanit R, Sritara P, Krittayaphong R, Kunjara-Na-ayudhya R. **Thai Registry in Acute Coronary Syndrome (TRACS)—an extension of Thai Acute Coronary Syndrome registry (TACS) group: Lower in-hospital but still high mortality at one-year**. *J Med Assoc Thail Chotmaihet Thangphaet* (2012.0) **95** 508-518
25. Mohanan PP, Mathew R, Harikrishnan S, Krishnan MN, Zachariah G, Joseph J. **Presentation, management, and outcomes of 25,748 acute coronary syndrome admissions in Kerala, India: results from the Kerala ACS Registry**. *Eur Heart J* (2013.0) **34** 121-129. DOI: 10.1093/eurheartj/ehs219
26. Roe MT, Halabi AR, Mehta RH, Chen AY, Newby LK, Harrington RA. **Documented traditional cardiovascular risk factors and mortality in non-ST-segment elevation myocardial infarction**. *Am Heart J* (2007.0) **153** 507-514. DOI: 10.1016/j.ahj.2006.12.018
27. Barger. **Short sleep duration, obstructive sleep apnea, shiftwork, and the risk of adverse cardiovascular events in patients after an acute coronary syndrome**. *Journal of the American Heart Association*
28. Kwok CS, Kontopantelis E, Kuligowski G, Gray M, Muhyaldeen A, Gale CP. **Self-Reported Sleep Duration and Quality and Cardiovascular Disease and Mortality: A Dose-Response Meta-Analysis**. *J Am Heart Assoc Cardiovasc Cerebrovasc Dis* (2018.0) **7**. DOI: 10.1161/JAHA.118.008552
29. Kim JH, Hayek SS, Ko YA, Liu C, Samman Tahhan A, Ali S. **Sleep Duration and Mortality in Patients With Coronary Artery Disease**. *Am J Cardiol* (2019.0) **123** 874-81. DOI: 10.1016/j.amjcard.2018.11.057
|
---
title: A realist review of best practices and contextual factors enhancing treatment
of opioid dependence in Indigenous contexts
authors:
- Rita Henderson
- Ashley McInnes
- Ava Danyluk
- Iskotoah’ka Wadsworth
- Bonnie Healy
- Lindsay Crowshoe
journal: Harm Reduction Journal
year: 2023
pmcid: PMC10022548
doi: 10.1186/s12954-023-00740-x
license: CC BY 4.0
---
# A realist review of best practices and contextual factors enhancing treatment of opioid dependence in Indigenous contexts
## Abstract
### Objectives
The objective of this study was to examine international literature to identify best practices for treatment of opioid dependence in Indigenous contexts.
### Methods
We utilized a systematic search to identify relevant literature. The literature was analysed using a realist review methodology supported by a two-step knowledge contextualization process, including a Knowledge Holders Gathering to initiate the literature search and analysis, and five consensus-building meetings to focus and synthesize relevant findings. A realist review methodology incorporates an analysis of the complex contextual factors in treatment by identifying program mechanisms, namely how and why different programs are effective in different contexts.
### Results
A total of 27 sources were identified that met inclusion criteria. Contextual factors contributing to opioid dependence described in the literature often included discussions of a complex interaction of social determinants of health in the sampled community. Twenty-four articles provided evidence of the importance of compassion in treatment. Compassion was evidenced primarily at the individual level, in interpersonal relationships based on nonjudgmental care and respect for the client, as well as in more holistic treatment programs beyond biophysical supports such as medically assisted treatment. Compassion was also shown to be important at the structural level in harm reduction policies. Twenty-five articles provided evidence of the importance of client self-determination in treatment programs. Client self-determination was evidenced primarily at the structural level, in community-based programs and collaborative partnerships based in trust and meaningful engagement but was also shown to be important at the individual level in client-directed care. Identified outcomes moved beyond a reduction in opioid use to include holistic health and wellness goals, such as improved life skills, self-esteem, feelings of safety, and healing at the individual level. Community-level outcomes were also identified, including more families kept intact, reduction in drug-related medical evacuations, criminal charges and child protection cases, and an increase in school attendance, cleanliness, and community spirit.
### Conclusions
The findings from this realist review indicate compassion and self-determination as key program mechanisms that can support outcomes beyond reduced incidence of substance use to include mitigating systemic health inequities and addressing social determinants of health in Indigenous communities, ultimately healing the whole human being.
## Introduction
This realist review examines international literature focused on interventions for opioid dependence in Indigenous communities in countries with similar healthcare systems and colonial histories to Canada to discover activities or strategies that might be relevant in our context. The term “intervention” is commonly used in Canada to refer to programs or policies designed to respond to community- or population-level health concerns [1]. This term has different connotations elsewhere in the world, which is critically recognized and incorporated into the study out of concern that it perpetuates colonial harm [2].
While geographically and culturally diverse, Indigenous peoples internationally are affected by a common experience of colonization (i.e. community disruption, trauma, unequal access to appropriate services, and discrimination) [3–5]. Indigenous communities have variable uptake of harm reduction efforts, and we believe that a review of the international literature can help illuminate cross-cutting challenges in diverse jurisdictions as well as opportunities that may be adaptable to new spaces. In the Canadian context, the term “Indigenous” refers to First Nations, Inuit, or Métis peoples. Within the USA, the term refers to those who are Native American, Hawaiian and Pacific Islander, Alaskan, and any other Nation of reference. Within New Zealand, the term refers to Maori people. In Australia, the term encompasses Aboriginal and Torres Strait Islander peoples. In this article, we capitalize the term as well as the names of any specific nations, similar to the capitalization of other nationalities.
The realist review methodology was selected to examine which programmatic practices are most effective in the context of historical and ongoing colonization, which continues to drive substance dependence in Indigenous communities. This methodology moves beyond synthesizing literature and identifying gaps, to flesh out mid-range theory about the practices, and their contexts, that best facilitate better health outcomes for Indigenous people struggling with opioid use [6]. As such, this research begins with an understanding of the impacts of colonization on health outcomes particular to opioid dependence in Indigenous contexts, given established relationships between substance use and trauma [7], stress [8], and social disconnection [9] caused by colonization. In this context, it is likely that the opioid poisoning crisis experienced for nearly a decade across *Canada is* intensified in Indigenous communities.
Rates and impacts of opioid dependence in Indigenous communities are difficult to quantify. Nonetheless, provincial data from Alberta reveal that opioid dispensation and opioid-related emergency department visits and hospitalizations, as well as rates of apparent accidental opioid drug toxicity deaths, were significantly higher for First Nations (FN) clients than among the non-FN population through 2016 [10]. Rates of dispensation of buprenorphine/naloxone (i.e. Suboxone®) for FN people in the province also increased by over $3000\%$ between 2013 and 2017, indicating interest among FN people to access opioid agonist therapies (OAT) to treat dependence [10]. While this data cannot be generalized to a single FN community, it establishes strong rationale for strengthening FN capacity to lead community-based models of care for responding to the ongoing opioid crisis and emphasizes the importance of cultural safety within environments where healthcare workers deliver care.
Aligned with legislation, principles, and findings from the United Nations Declaration on the Rights of Indigenous Peoples (UNDRIP) [11] and the Truth and Reconciliation Commission of Canada (TRC) [12], we affirm that the health status of Indigenous people is a direct result of historical and ongoing colonial policies and practices in the healthcare system, and that Indigenous people have the right to influence health research, policy and practices that impact them. Accordingly, this realist review was guided by Indigenous knowledge holders and Alberta FN communities affected by the opioid crisis through a knowledge contextualization process that included an initial Knowledge Holders Gathering to inform and focus the literature search and analysis, and five consensus-building working groups that included Indigenous clients and providers to improve understanding of Indigenous contexts in Alberta and appropriate programmes (Fig. 1). Despite synthesizing literature for the purpose of improving the care of FN opioid addiction in Alberta, we believe this realist review could also improve care in any international context where Indigenous populations are similarly bordered by the effects of colonization. Fig. 1Knowledge contextualization process for the realist review
## Methods
This work is located within a broader opioid knowledge synthesis project initiated in response to the release of 10 Recommendations for Immediate Action by the AHS Indigenous Opioid Advisory Sub-Committee in June 2017. The project was based in partnership with the Alberta First Nations Information Governance Centre (AFNIGC), which governs research initiatives in FN communities and advances FN Ownership, Control, Access and Possession (OCAP®) principles. Our team emphasizes that connection between communities and researchers is essential to high quality and relevant research and utilizes a collaborative, consensus-building approach that centres Indigenous principles of respect for diverse perspectives and affirms a decolonizing approach to knowledge that values the community relevance of evidence yet is sufficiently systematic for policy-makers and healthcare providers. As such, a realist review guided by community is appropriate for this research.
The University of Calgary’s Conjoint Health Research Ethics Board approved components involving research with human participants (#REB18-052). The AFNIGC ensured compliance with Indigenous ethical protocols including OCAP® principles.
## Realist methodology
In this review, we use a realist methodology to synthesize evidence on current opioid use interventions to inform policy-makers and healthcare providers on best practices for opioid use programmes in Indigenous communities. A realist review goes beyond the task of descriptively synthesizing the current literature and identifying gaps in the knowledge base by allowing the reviewer to substantively identify mechanisms, these being how and why certain interventions are successful (previous research has provided detailed description of realist methodology [13]). Briefly, a realist review methodology incorporates an analysis of the complex interactions associated with the social determinants of health by identifying program mechanisms [14]. As such, realist reviews support the adaptation of best practices into different contexts and in doing so may better support programme planning and implementation than conventional systematic literature reviews [14].
Realist reviews begin with identification of a preliminary theory in which evidence is sought to determine in which contexts (e.g. programme leadership, community factors) a mechanism (e.g. client motivation) leads to a particular outcome (e.g. reduction in opioid use) [13]. Less explicitly identified than programme components, such as counselling, mechanisms are identified through reporting of the intervention experience by different actors involved in the programme, from managers to clients, or derived by researchers with in-depth understanding of how the programme was implemented in different contexts [13, 14]. Preliminary theory development in this realist review was guided by discussion with stakeholders [13]. The next step was to compile literature to extract evidence for (or against) the preliminary theories and refine the theories through knowledge synthesis [13]. In addition, we hosted five consensus-building meetings with clients and providers. The initial Knowledge Holders Gathering mentioned above and consensus-building meetings made up the knowledge contextualization process that guided the realist review and supported a better understanding of the community contexts.
## Knowledge contextualization process
Emphasizing that appropriate and effective programmes require connection between community members, healthcare providers, researchers, and policy-makers, we enhanced a conventional realist review with a knowledge contextualization process based in two components. First, we convened a two-day Knowledge Holders Gathering in Banff, Alberta in May 2018, which included twenty-eight Elders from FN communities from Alberta, to understand community perspectives on the opioid crisis and situate our research within key principles they identified. Second, in August and September 2018, we facilitated five consensus-building working groups made up of clients and providers to highlight the realities of Indigenous clients and the clinical experiences of healthcare providers to inform policy, funding priorities, and future research. These meetings further identified salient themes and priorities through elevating the voices of the people most closely affected by this crisis, thus adding value to the synthesis by providing contextualization to lived experiences. This contextualization process also ensures that the realist review remains relevant to Indigenous communities and empowers those working in this area to assess the evidence base in terms of appropriateness for their own community members. Following other realist reviews conducted in Indigenous contexts [14], we developed candidate theories based in these knowledge contextualization processes.
## Theory development
Though colonization impacts all Indigenous communities, how it plays out differs in unique contexts and results in different experiences for different communities. In the Knowledge Holders Gathering, Elders emphasized five key contextual factors affecting opioid dependence in their communities, including structural violence, trauma, culture, community, and experiences (full analysis presented elsewhere [15]). Structural violence outlined by knowledge holders through stories of institutional racism and stigma, has driven mistrust of western healthcare programmes and practices. Particular concerns around overprescription of opioids were highlighted, linked with stories of medications for opioid treatment such as opioid agonist therapy (OAT) that are prescribed without adequately informing clients about the goals, procedures, or risks associated with OAT. These experiences were related to processes of ongoing colonization where clients expressed concern that health professionals are wilfully not providing them with adequate information to allow them to be part of the decision-making process regarding treatment options. With these experiences, knowledge holders called for healing initiatives within communities, and that these not be restricted to abstinence-focused programmes alone but emphasize healing from the multigenerational impacts of colonization. Such healing initiatives were seen to support the decolonization of healthcare in Indigenous communities and also present a key opportunity for individuals in recovery journeys to access programmes that are culturally appropriate, support community reconnection, and promote holistic healing from the broad impacts of colonization on health and wellness.
The Knowledge Holders Gathering in Banff contextualized opioid dependence in Indigenous communities as a symptom of wider social breakdown due to historical and ongoing processes of colonization. Trauma and loss of loved ones were common experiences identified as driving dependence, a view consistent with academic research on emotional pain and stress as key risk factors in developing addiction [8]. We recognize that substance use can be a way of self-medicating in the face of colonial conditions in which people may not have the knowledge, ability, or tools to cope with stressors or pain in other, strength-based ways. Social disconnection for individuals and in communities was frequently discussed and linked to past and current government policies including residential schools, the Sixties Scoop, and the large proportion of children removed from families and placed in the child welfare system: policies separating families and extinguishing cultural practices that have ongoing implications for the health of Indigenous communities [16]. Loss of culture and community were outlined not only as drivers of the opioid poisoning crisis, but also as barriers to recovery when individuals remain isolated in recovery programs that do not resonate with them. For many individuals who have experienced substance use challenges, we were told that reconnection with culture and community are vital supports for recovery. Additionally, and also consistent with the literature [6, 17], culture and community were highlighted as important protective factors for substance use. Knowledge holders outlined their experiences with substance use in their communities, highlighting stories of resilience and deep knowledge about healing.
This knowledge and experience directed us to understand that realist review candidate theories guided by these insights must address trauma and structural violence, enhance the supports provided by culture and community, and utilize the experience of community members who have themselves overcome previous addiction. Shifting attention to mechanisms that underlie community strategies, research team members considered which mechanisms might counteract trauma and structural violence, enhance community connections, and integrate Indigenous culture and experience in treatment and harm reduction. While these are terms commonly used within healthcare and social services settings, we take treatment to refer to medical approaches for diagnosing and mitigating addiction and harm reduction to involve a broader set of policies, practices, and/or programmes that work to minimize diverse impacts associated with drug use.
Reviewing the transcripts from our gatherings for underlying themes, compassion and self-determination became evident as mechanisms to support positive outcomes of opioid programmes (e.g. greater supports for substance users, increased integration of community knowledge holders in western healthcare settings, and lower rates of drug-related medical incidences [29]). In other words, compassion and self-determination were identified as key mechanisms capable of enhancing the efficacy of opioid programmes in Indigenous contexts as they both acknowledge the underlying drivers of opioid dependence. For the purpose of this research and guided by the knowledge holders, we came to define compassion as refering to an all-encompassing kindness and consideration for life [10] and self-determination as refering to recognition of an individual or community’s ability and right to sovereignty in the decision-making process. Considering compassion and self-determination as important mechanisms to Indigenous-driven opioids programmes, we proposed two candidate theories:Candidate Theory #1: Treatment and harm reduction models based in compassion (mechanism) for individuals and communities affected by trauma and structural violence (context) counter the stressors driving addiction and lead to saved lives, reconnected families, and people better able to reach their full potential through unhindered achievement of goals and aspirations (outcome).Candidate Theory #2: Treatment and harm reduction models that recognize Indigenous community self-determination (mechanism) through community leadership, and culturally-based models of care integrative of community knowledge and experience with overcoming addiction (context) builds on community resilience in the face of this crisis (outcomes).
## Document selection and appraisal
To locate peer-reviewed literature, research team members searched a total of 13 electronic health and social sciences databases in June 2018 (see Text Box 1). Combinations of search terms used included: Indigenous, First Nations, Aboriginal, American Indian, Metis, Inuit, Pacific Islander, Aborigine, Polynesian, Alaska Native, Oceanic Ancestry Group, American Native Continental Ancestry Group, Native, Samoan, Tribe, opioids, opiates, fentanyl, street drugs, substance abuse, treatment, prevention, wrap-around supports, harm reduction, harm minimization, models of care, naloxone and/or suboxone. After removing duplicates, the articles were screened for inclusion criteria (see Text Box 2) using a three-step process (title, abstract, and full-text review) (see PRISMA diagram in Fig. 2). Of the articles in the initial literature search, $6\%$ of the abstracts were reviewed by a second reviewer for inter-rater reliability. Of full texts assessed for eligibility, $100\%$ were reviewed by 3 reviewers. Text Box 1Databases searchedMedline, EMBASE, Scopus, Cochrane Central Register of Controlled Trials, PubMed, PsycINFO, iPortal (Indigenous Studies Portal), CINAHL Plus, Bibliography of Native North Americans, SocIndex, Web of Science, Healthstar, and Academic Search CompleteText Box 2Search criteria1. Describes an intervention for opioid use, prevention or treatment, or includes information on treating opioid use in the broader context of substance abuse or holistic health programs2. Includes information on elements of intervention success, either through qualitative methods or through quantitative methods with sufficient in-depth description of the intervention combined with quantitative methods3. Focuses exclusively on Indigenous populations4. Maintains a nonoppressive, affirming voice5. Research article6. Focuses on Indigenous communities in countries with similar healthcare systems and settler-colonial histories framing health inequities (Canada, the USA, Australia, New Zealand)7. English8. Full text availableFig. 2PRISMA diagram
## Data extraction, analysis, and synthesis
Articles were managed in a Microsoft Excel chart. Reason for inclusion or exclusion was tracked for all articles. Full-text screens were extracted in a more rigorous extraction framework that included a realist review extraction tool developed by Molnar et al. [ 18]. We extracted study information on contextual factors driving opioid use, as well as program descriptions (context), compassion and self-determination (mechanisms) and conclusions relevant to the study mechanisms including intervention outcomes, participant perspectives, and key lessons or paper conclusions (outcomes).
## Results
The initial literature search yielded 8418 results, of which 129 full-text articles were reviewed for inclusion (Fig. 2). One additional source was retrieved via a scan of article references. A total of 27 sources were identified that met inclusion criteria, including 12 qualitative [19–30], 6 quantitative [31–36], 6 mixed-methods [37–42], and 3 reviews [43–45]. Most of the articles examined perspectives or outcomes of client or community members on substance use programmes wholly or partially aimed at opioid use [19, 24–26, 31–37, 40, 42]. Some articles examined pharmacist programme outcomes [20] or care provider workshop outcomes [21, 27, 38]. Four articles examined perspectives from both clients and care providers [23, 28, 30, 39].
## Context
Contextual factors contributing to opioid dependence described in the articles often included discussions of a complex interaction of social conditions in the sampled community. The social determinants of health outlined included poverty [22, 40, 42], unemployment [22, 42], poor health [22], low education levels [22, 42], low economic development [22], incarceration [39], single-motherhood [40], domestic violence [40], racism [42], residential school experiences [22], legacies of colonialism on territorial and cultural dislocation [27], and emotional responses to colonization including grief [39], and emotional trauma and loneliness through the loss of loved ones and personal identity [43]. The social determinants of health in Indigenous contexts have been outlined in the literature; for the purpose of this realist review, the context extraction focused on program descriptions (Table 1). Table 1Context mechanism outcome tableReferencesLocationStudy designStudy sampleContext: program descriptionMechanism: compassionMechanism: controlOutcomesBeckstead et al. [ 31]USAQuestionnaireAmerican Indian/Alaska Native (AI/AN) youth ($$n = 229$$) from 39 tribesDialectical Behavior Therapy (DBT) evidence-based treatment integrated with traditional models of healing within a AI/AN youth residential treatment centre (average length of stay 120 days)Mindfulness, a core skill taught in DBT which can include traditional models of healing such as ceremony, talking circles and smudging; program considers individual needs for quality of life (e.g. employment) alongside substance misuse treatmentConsultation with tribal leaders; local spiritual leader provided weekly spiritual practices and explained the relationships between traditional practices and the mindfulness skills taught in DBTTreatment outcomes: $96\%$ either recovered or improved using Youth Outcome Questionnaire-Self Report edition pre and post-treatment (clinically significant change criteria)Benoit et al. [ 19]Canada: Downtown Eastside VancouverParticipant observation, semi-structured interviews, and focus groupsAboriginal women living in Vancouver’s Downtown Eastside (DTES) ($$n = 12$$); 25 interviews with staff and health professionals ($$n = 61$$)Vancouver Native Health Society (VNHS), an integrated and holistic (e.g. food bank) service provision and primary care setting based on integrating traditional and western approaches; Sheway, VNHS partner with a model of care based on harm reductionFree healthcare, non-judgmental primary care and establishing trust before initiating health care; providers aim to make a connection first, work on healingThough service providers are primarily non-Indigenous, the clinic has recruited Indigenous volunteers to support agency; participants also noted gaps in traditional healing practicesPatient perspectives: Women primarily indicated a need for a Healing Place, integrated and holistic health care based in respect and influence over decisions and services that impact their healingBlack et al. [ 32]Australia: Winnunga Nimmityjah communitySurveyAboriginal adult opioid dependent patients ($$n = 21$$)Opioid replacement pharmacotherapy for Aboriginal patients in the Winnunga Nimmityjah Aboriginal Health Service, a community controlled primary care settingPrograms include social supports and comprehensive care within a “supportive framework” that ensures stability for patientsCommunity-controlled health service houses the program; community leadership in education to garner support for opioid replacement therapy and peer outreachTreatment outcomes: Comparable to outcomes in mainstream programs ($81\%$ retention; no significant change in self-reported heroin use)Campbell et al. [ 37]USA: Northern Plains Region and Pacific NorthwestSurveys and Interviews conducted 1-week post interventionAmerican Indians and Alaska Natives (AI/AN) ($$n = 40$$)The Therapeutic Education System, a web-based community reinforcement approach for substance misuse treatment completed onsite at two urban outpatient programsTraining included learning to manage negative thinking and improve self confidenceNot evidentParticipant outcomes: 37 completed at least one module; Patient perspectives: Participants indicated less interest in western approaches and a desire for more culturally specific communication styles and traditional practices including spiritualityDooley et al. [ 33]Canada: NW OntarioRetrospective chart reviewMothers and infants ($$n = 2743$$)The Integrated Pregnancy Program at the Sioux Lookout Meno Ya Win Health Centre that integrates prenatal and addiction care, including OAT and tapering in the third trimesterMale partners are involved in the program and offered additional addiction treatment; respect for patient; family-centered care; postpartum care is coordinated with community-based programs to ease transitionsIncludes traditional healing practices along with OAT to encourage community involvement; the authors attribute the decrease in neonatal abstinence syndrome to the local community initiativesTreatment outcomes: Significant decrease in neonatal abstinence syndrome between 2009 and 2015 ($$p \leq 0.001$$); observed positive community-wide changesDuvivier et al. [ 20]USA: Southwest, Midwest, and Great Lake regionsProgram descriptionIndian Health Service pharmacistsThe Prescription Drug Abuse Workgroup, Pharmacy-based interventions including responsible prescribing practices and improved access to medication-assisted treatment, comprehensive services, pharmacist-developed training for first respondersAdvocate respect for the patient including supportive and nonjudgmental relationships; individualized and comprehensive treatment procedures; expansion of more comprehensive services beyond dispensationCollaborations with local governing bodiesCare provider outcomes: pharmacists have pledged to reduce stigma, screen for opioid use disorder, support safe prescribing, and increased access to naloxone and committed to expanding medicated assisted therapiesGray [43]USALiterature Review and case studyAmerican Indian AdolescentsTwo-month, 12-step based, in-patient treatment program with weekly trauma and loss treatment groupsTrauma-informed treatment, psychological and emotional wellness to prevent relapse; safe environment for grief support; holistic healing including spirituality, cultural connection, empowerment and internal strength; family-based care (adult family members invited to attend the final week)Traditional healing practices and individual empowermentPaper conclusions: treatment must include focus on trauma and loss, connection with culture and spirituality, and healthy coping skillsGray et al. [ 44]Australia, New Zealand, Canada, and USEditorialIndigenous populationsN/AHarm reduction policies; solutions that address structural drivers of health inequities; partnerships between Indigenous and non-Indigenous organizations require trustIndigenous people must guide or be involved at all stages of research and interventionsKey lessons: Research and interventions must be community-based, in collaboration with communities; appropriate research, evaluation and policy (including harm reduction) must be community defined; broader structural interventionsJumah et al. [ 21]Canada: NW OntarioProvider workshopService providers in Indigenous communitiesService provider workshop for care of women with opioid dependence while pregnant and postpartum, in rural and remote settingsRecommendations included increased provider education on SDOH, trauma-informed care; improved transitions between services; family-based care including keeping families intact; improved access to medicated assisted therapy to reduce risks in transportingRecommendations included Indigenous-led programs/ partnerships with Indigenous-led programs; integration of Indigenous best practices with ‘gold standard’ medicine; funding models that encourage collaboration rather than competition between communitiesCare provider outcomes: providers committed to Indigenous-led interventions, improved transitions in care, trauma-informed care based in Indigenous worldviews and holistic health and wellbeing, and improved access to treatment (including stable funding for Indigenous programming)Kanate et al. [ 34]Canada: NW Ontario, North Caribou Lake First NationCommunity statistics 1 year before and 1 year after the program initiationCommunity-wide dataCommunity-based, outpatient program that integrates buprenorphine-naloxone opioid substitution and counseling from traditional healers as well as other modes of holistic healingCommunity acceptance and celebration of individuals attending treatment; holistic healing; sense of community purposeProgram is managed by community nurses and healthcare providers; First Nation counselors and healers deliver culturally based and land-based healing programsTreatment outcomes: Community-wide healing: Decrease in drug-related medical evacuations (-$30\%$), criminal charges (-$66.3\%$), child protection cases (-$58.3\%$); increase in school attendance ($33.3\%$); observed increase in community spirit and sense of purposeKatt et al. [ 35]Canada: N Ontario, Nishnawbe Aski Nation communitiesUrine toxicology screeningFN community members aged 16–48 ($$n = 22$$)Community-based 30 day suboxone tapered off or to low dose maintenance program with community-based aftercareAftercare programs included overall health and spiritual supportTreatment in community health centre; collaboration between off-site addiction specialists and on-site care providersTreatment outcomes: $95\%$ completed the program, $88\%$ had no evidence of prescription opioid use in their urine toxology on day 30Katzman et al. [ 38]USAPre/post intervention surveyIndian Health Service clinicians ($$n = 1079$$)5-h virtual education sessions for clinicians on safe prescribing and appropriate pain managementNot evidentNot evidentCare provider outcomes: significant increase in knowledge, self-efficacy, attitudes ($p \leq 0.001$)Kiepek et al. [ 22]Canada: Ontario, Sioux First NationProgram overviewSioux First NationsInpatient medical withdrawal support service for patients seeking abstinence in the Sioux Lookout Meno Ya Win Health Centre (SLMHC)Holistic care (integrates physical, emotional, interpersonal, contextual factors), recognition that individual health stems from community health; trusting and responsive relationships, emphasis on patient goals and additional supports from establishing a daily routine to community leadership; follow-up careCommunity-based program; integrates traditional healing including Elders in residence; individual patient goals are prioritizedTreatment outcomes: All but two patients successfully completed the program between December 2011 and June 2012Landry et al. [ 23]Canada: New Brunswick, Elsipogtog First NationSemi-structured focus groupsThree groups: professional (methadone maintenance treatment program management and delivery) group, patient group, community group ($$n = 22$$)Elsipogtog methadone maintenance treatment program in the Elsipogtog Health & Wellness CentreHolistic healing, based in traditional medicine and spiritual beliefs, including life skills such as parenting practicesCommunity-based program, services offered in Mi'kmaq, Indigenous staff, Elders are availableParticipant perspectives:Program considered effective at the Individual level (improved parenting practices), with some positive community impacts outlined (cleanliness, safety) but patients experienced stigma, marginalization and discrimination within the community (e.g. spiritual centers, employers) and family conflict demonstrating misinformation within the communityLee et al. [ 39]Australia: SydneySemi-structured interviews and surveysAboriginal female clients ($$n = 24$$) and staff ($$n = 21$$)Weekly Aboriginal women's support group at an inner-city outpatient alcohol and other drug treatment service including opioid substitution treatment; format ranges from educational to informal conversation or recreational (e.g. art)Topics vary, including treatment options and broader healing, skill building; group described as non-judgemental, offering skills-based training and broader health education including navigating systems; children welcome to attend the program; program emphasis on coming together and relaxing, providing opportunities for peer support and relationships with staffProgram encouraged client ownership of the groupParticipant perspectives: Group members reported feeling safe, respected, supported and valued, gaining new skills, improved self-esteem and identity, more connected to services; both staff and patients reported a desire to interact more informally with each otherMamakwa et al. [ 36]Canada: NW Ontario, Sioux Lookout regionMedical record review including buprenorphine-naloxone prescribing and urine drug screening6 First Nation communities ($$n = 526$$)Community-based buprenorphine-naloxone treatment combined with traditional healing; 4 weeks of daily treatment and aftercareInductions are in groups of 10–20 within a community-wide celebration; programs are viewed as a "welcoming back" of patients; treatment facilities are used as meeting places for healing circles; healing includes spirituality (Elder-guided, land-based aftercare)Programs were community designed, implemented and administered; healing circles and traditional activities over formal programsTreatment outcomes: Retention rates were high ($72\%$ at 18 months); urine drug screening showed high rates of negative results for illicit opioids (84–$95\%$); Community outcomes: Decline in suicides in six communities; one community had declines in drug-related medical evacuations, criminal charges and child protection cases, and increase in school attendanceMarquina-Marquez et al. [ 24]Canada: N OntarioOpen-ended interviews and oral story tellingOji-Cree reserve residents, recruited through medical facilities ($$n = 35$$)Community-based healing movement, including nature-based therapeutic initiatives, traditional healing methods and physical spaces for healing, often outdoors and informalEmphasis on personal reconnection to land (place attachment integral to wellbeing), spiritual world, and identity, healing from personal traumas, maintaining family ties, love and respect for communityGrassroots movement based in culture; informal and self-pacedParticipant perspectives: Traditional practices support healing and family/community reconnection; the grassroots/community-based nature was important for holistic healing; the informal nature allowed for individuals to engage at their own paceMomper et al. [ 25]USA: Midwestern Indian Reserva- tionEight Talking circlesAI adults and youth ($$n = 49$$)No specific programNot evidentOne tribal council passed a resolution prohibiting OxyContin prescriptions except in terminal casesParticipant perspectives: Reported barriers to treatment: need for group support; worries that returning the same environment would result in relapse; lack of adequate and accessible treatment optionsRadin et al. [ 26]USA: 4 Washington state tribal communitiesSemi-structured interviews and focus groupsCommunity members ($$n = 153$$)General substance use, abuse, and dependence (SUAD) programs and resources across the four communitiesValued program aspects included community support for struggling individuals, family involvement in treatment, sense of identity, attention to “whole person”, life skills, and providers who are supportiveNeed to address prescription drug misuse with better communication between community and healthcare providers; valued aspects of available programs included culturally-based and community driven prevention, treatment and aftercare; individualized treatmentParticipant perspectives: valued community support, 'supportive' care providers; need for family/community wellness, adequate transition housing more supportive of recovery; need for greater community- provider collaboration, community-based and culturally-based care and healing centres; Former patient perspectives highlighted the need for compassion, holistic treatment (including trauma treatment), and the desire to keep families intact an important motivatorRussell et al. [ 45]CanadaScoping ReviewAboriginalMultiple, includes a section on the National Native Drug Abuse Program (NNADAP) and community-based suboxone programsNot evidentScoping review focused on the successes of community-based and culturally-integrative treatmentPaper conclusions: Community-based treatment models have promising outcomes (high completion rates, improved abstinence, community-level improvements); more contextually and culturally appropriate treatment neededSaylors et al. [ 40]USA: San Francisco Bay AreaProgram overview, staff and interviews and clinical dataNative American Women clients ($$n = 742$$)Women's Circle project of the Native American Health Center, focused on integrating western and Indigenous healing, holistic health;Women’s group provides nonthreatening entry into health programs; emphasizes respect for patient/patient comfort, trauma-informed care; a nurse case manager facilitates transitions between services; holistic healing and skills training (e.g. family functioning); meeting the patient 'where they're at'Healers from diverse communities brought in to support traditional healing for culturally diverse clients; incorporation of spirituality into counselling is guided by the patient; emphasis on Native staffTreatment outcomes: Heroin use decreased by $93\%$; women who were using nonprescription methadone stopped after intervention; self-reported improvement in health, living conditions, increase in school attendance, decrease in involvement with criminal justice system; increase in importance of culture to participantsSrivastava et al. [ 27]Canada: Ontario, Sioux Lookout First NationPhysician interviewsFamily physicians in Sioux Lookout ($$n = 18$$)Sioux Lookout Zone Physicians initiated an education program to reduce physician anxiety about prescribing opioids, improve the management of chronic pain, and limit the risk of addictionEducate patients on harm reduction strategiesPhysician to develop treatment agreements with patientPhysician outcomes: Increased awareness of addictive potential of opiates; started titrating lower potency medications for chronic pain; improved physician confidence in opioid prescribing and identification of patients with opioid dependence; increased use of treatment agreements with patientsTeasdale et al. [ 28]Australia: SydneyInterviews and focus groupsIndigenous patients and Indigenous and non-Indigenous staff ($$n = 63$$)The Drug Health Service of Sydney South West Area Health Service, Eastern Zone, provides drug-related services including opioid maintenance pharmacotherapy with medication dispensed daily or every other day at the treatment centreStaff provided after-hours support for family and community; aimed at providing therapeutic, non-authoritarian carePriority assessment for Aboriginal clients; broad and flexible dosing hours; collaboration with Aboriginal Medical ServicesParticipant perspectives: Clients reported misinformation about methadone's health impacts, and culturally-based concerns; non-Aboriginal staff indicated a lack of cultural awareness training and appropriate care for holistic health challenges particularly to help with child protection services, heavy burden of after-hours support; paper conclusions: tighter partnerships with the Aboriginal community and a less formal and more welcoming serviceThomas et al. [ 41]Canada: British ColumbiaSemi-structured interviews and surveys including follow-upFirst Nation community members ($$n = 12$$)“Working with Addiction and Stress” 4-day retreat with ayahuasca-assisted group therapyNOW AN ILLICIT SUBSTANCE – REMOVE?Primary aim was to release pain and heal through participation in ceremony and personal reflectionInvolvement of local First Nations spirit-keeperTreatment outcomes: Opioid use (one participant) had no changeUddin et al. [ 29]Canada: N Ontario, Eabametoong FNPhysician reflectionN/ANorthern Ontario Suboxone Support program, community-based programs; partnership between community and addiction specialistsHolistic healing in aftercare, compassionate staffCommunity design and ownership is key to the program’s successPaper conclusion: patients report the program has made a positive difference in their lives; author argues that Suboxone should be dispensed in the community by community nurses and trained laypeopleVenner et al. [ 30]USAStakeholder MeetingAI/AN Community members and AI/AN and non-AI/AN healthcare providers and agenciesNational Institute on Drug Abuse stakeholder meeting to elicit feedback on medication assisted treatment in the communityProgram needs outlined included holistic healing, traditional healing, patient desires to be medication free, and address systemic barriers (lack of resources in community, discrimination and inadequate care experienced outside of community)Program needs outlined included need for better resources in community, more AI/AN care providersPaper conclusions: must integrate medicated assisted therapy into traditional healing approaches and train providers to honor Indigenous ways of knowingWilliams et al. [ 42]Australia: AdelaideDiscussions and electronic recordsNunga Australians ($$n = 226$$)The 'Way Out' program, a holistic health program including opioid substitutionHolistic health focus on individual needs and trauma, family-based support, patient advocacy; reconciliation displays in the centre; access within a health centre to provide confidentiality; strong links with correctional services to ensure seamless transitions between servicesInitiated by community, strong partnership with Aboriginal health programs in development and implementation; community members are relied upon to raise awareness about the program and ensure its success; community-based activities; Aboriginal staff; flexible appointmentsTreatment outcomes: *Program is* attracting and maintaining more Indigenous clients than any previous program in the region; of those who ever accessed opioid substitution treatment, $40\%$ are in current substitution, $10\%$ have successfully completed, $16\%$ transferred out, and $34\%$ have defaulted ($19\%$ before stabilization, $15\%$ after) The articles included in the realist review provided examples from urban [19, 28, 39, 40], rural and remote [21, 27, 33, 35, 36], in-patient [23, 43], out-patient [34], and web-based programmes [37]. Indigenous substance use programme challenges outlined as common included a lack of stable funding [21], high staff turnover [37], a lack of cultural awareness training for staff [28], and a lack of Indigenous staff members [19]. Some articles did outline programs with Indigenous staff [23, 28].
## Mechanism: compassion
To test Candidate Theory #1, we searched program descriptions for evidence of compassion, as well as study participant (client, provider, and community member) perspectives affirming a need for compassion in treatment programmes. All but three articles [25, 38, 45] provided evidence of the importance of compassion in treatment. Broadly, compassion was evidenced at the individual level, in interpersonal relationships based on nonjudgmental care and respect for the client, as well as in more holistic treatment programmes beyond biophysical supports, such as medically-assisted treatment.
The importance of interpersonal relationships based in compassion was highlighted frequently in terms of respect for patient, therapeutic, responsive and nonjudgmental relationships, and establishing trust [19–22, 26, 28, 29, 33, 39]. In one example, providers aim to establish trust and develop relationships foremost, before initiating treatment [19]. Compassion was also in evidence through outreach workers that seek out clients who fail to attend programmes [19] or who provide after-hours care, indicating a need for systemic supports for frontline workers going beyond their job descriptions to support people [28]. An additional component of the compassion mechanism was working to ensure seamless transitions between services or into community-based care [33], including frontline staff that advocate for the individual as they transition between services [42]. Participants in one study also noted gaps in compassion as barriers to accessing services [19].
Every article that provided evidence supporting compassion as a mechanism for treatment interventions indicated a need for holistic healing programmes. Weaker examples of holistic healing within the mechanism of compassion included a web-based treatment program that included training to manage negative thinking and improve self-confidence [37] and provision of pamphlets on harm reduction strategies [27]. Most articles highlighted the importance of holistic healing. Such programmes support cultural reconnection and spiritual and emotional healing in addition to substance use treatment. Examples of programmes that integrate spirituality included Elders in residence [22], participation in ceremony [31], utilizing treatment facilities as meeting places for healing circles [36] where clients are able to share openly and informally, and land-based aftercare [36], such as fishing, taking walks, and gardening. Emotional care was profiled as including supports to heal from personal traumas [24], and ensuring safe environments for grief support [43].
Care based in compassion that supports holistic healing must focus on individual needs, which one programme defined as attention to the “whole person” [26]. This includes integrating individual client goals for substance treatment that may include a desire to be medication-free [30], which was a popular perspective considering historical and cultural contexts surrounding the relationship between Western and traditional medical practices [30]. Other broader health and wellness goals among clients included the desire to establish healthy routines and build self-esteem [22]. These include social supports and a comprehensive care framework that ensures stability for clients [32] by considering quality of life alongside substance use treatment [31]. Consideration of individual needs includes programmes that address diverse needs, such as interpersonal and contextual factors [22], empowerment and internal strength [43], parenting skills [18], and life skills or skills-based training [26]. Also important here is broader health training including support for navigating health systems [39].
Holistic healing based in compassion includes recognition of the importance of healing for families and communities beyond the individual in treatment. Holistic healing prioritizes keeping families intact [21, 24] and providing support for treatment and healing for the whole family [26, 28, 33, 42, 43]. There is also recognition that individual health stems from community health [22]. As such, several programmes integrated community healing and community support [34], for example, by holding community-wide celebrations of client inductions into treatment [36]. Supporting community health also means that there is a need to address systemic barriers to healing including a lack of resources in community, and addressing discrimination and inadequate care for individuals accessing treatment outside of their community [30].
Though compassion primarily acts as a mechanism at the interpersonal level, there was some evidence for compassion at a structural level. Models of care based on harm reduction provide evidence for compassion at a structural level [19]. Additional evidence for compassion at the structural level included mandates or provider training programmes on social determinants of health and trauma-informed care [21]. The articles indicated broad support for harm reduction policies [19, 20, 22, 23, 28, 42, 44]. Many Indigenous communities may prefer an abstinence-based approach, yet in such contexts it may be possible to increase support for harm reduction measures when it is sensitive and respectful of local needs and preferences, and is based in strong partnerships [42]. Moreover, regardless of perspectives on which models are most appropriate for addressing opioid dependence, harm reduction is an essential part of an opioid treatment conversation in order to save lives even if the ultimate aim for people with substance use disorder is abstinence, healing, and/or recovery.
## Mechanism: self-determination
To test Candidate Theory #2, we searched programme descriptions for evidence of self-determination, as well as study participant (client, provider, and community member) perspectives that there is a need for community self-determination in treatment programmes. All but two articles [37, 38] provided evidence of the importance of self-determination in treatment programs. Broadly, self-determination was evidenced at the structural level, in community-based programmes, but was also shown to be important at the individual level in client-directed care.
Strong examples of community-based care include programmes that are initiated, planned, managed and evaluated by community [21, 34, 36, 44]. Examples include community-based OAT integrated with traditional healing [23, 35, 36], a grassroots healing movement [24], and tribal council resolutions prohibiting OxyContin prescriptions except in terminal cases [25]. Structural supports for community-based programmes include funding models that encourage collaboration rather than competition between communities [21], community treatment and healing centres [35], and increasing resources in communities including more Indigenous care providers [30].
Collaborative partnerships are also important for community self-determination. Such partnerships are based in trust, with meaningful and early engagement of community leadership [21, 26, 32]. This may include collaboration between on-site care providers and off-site addictions specialists in the provision of OAT [30] or—related to the mechanism of compassion—integration of Western treatment approaches with traditional healing practices, led by community spiritual leaders [21, 22, 33, 40, 45]. The latter indicates a need for on-reserve healing spaces and community-developed healing protocols that include holistic goals [22] and community-based aftercare [21]. These programmes may be especially important when effective biophysical treatments such as buprenorphine-naloxone are not available in community [35]. For those who attend out-patient treatment centres outside of their communities, community-based counselling and aftercare programs may be important to support transitions back into community [36]. Optimally, such programmes would be based in collaborative partnerships between on- and off-site care providers and connected to longitudinal treatment plans [35, 36].
Community ownership and self-determination is important for addressing the structural inequities driving substance use, as well as mitigating mistrust [19]. In communities where abstinence-based approaches are preferred, early involvement of community leadership is important in gaining support for OAT [32]. This may include leaders educating their communities about OAT and peer outreach initiatives [32]. One evaluation emphasized the need to be sensitive and respectful of local contexts to support strong partnerships, especially in providing OAT in contexts where abstinence models are valued [42]. Community self-determination over programming initiated outside of the community also requires early inclusion of leadership in decision-making and hiring of local staff [26, 29]. In one OAT programme, the community initiated wrap-around care models, including healing circles, land-based aftercare, traditional activities, and turned inductions into community-wide ceremonies that welcomed clients back into their community and family roles [36]. Urban treatment centres prioritized hiring Indigenous staff and volunteers or relied on peer outreach to raise awareness of the programme [19, 40, 42]. In an urban treatment centre where clients have diverse spiritual backgrounds, one program invited traditional healers from the individual clients’ communities [40].
Some programmes indicated the importance of self-determination at the individual level [22]. Programmes aimed to support individual empowerment [43] and provide the individual with control over their treatment programmes [26, 27]. In one programme, individual client goals were prioritized, and personal goals may include abstaining from substances, but also holistic wellness goals important to the individual, such as exercising or spending time with family 22). Healing programmes that are informal and self-paced also support individual self-determination over their treatment [24]. One programme evaluation indicated that healing circles and traditional activities may be preferred over formal programmes [36]. One programme supported individualized treatment including client-driven incorporation of spirituality into counselling [40]. In one urban programme, the treatment centre maintained priority assessment for Indigenous clients, as well as broad and flexible dosing hours for OAT, providing elements of individual self-determination in accessing treatment [28]. Flexible appointment scheduling supports individuals with competing priorities, supporting holistic healing [42]. Also important here is strong links between services as clients transition from systems, such as correctional services, into community [42].
## Outcomes
Ten of the included articles outlined treatment outcomes, and all of these provided evidence of compassion and self-determination as mechanisms important in treatment programmes [22, 31–36, 40–42]. Of the two articles that provided results from urine toxicology screening, one study showed that $88\%$ of participants had no evidence of prescription opioid use in their urine on day 30 of the programme ($95\%$ completion rate) [35]. The other study provided a review of a community-based OAT program that integrated traditional healing in 6 communities, where negative results for illicit opioids ranged from 84 to $95\%$ [36]. Other evaluation methods included retention rates, which were generally high in the community-based programmes [22, 32, 36], as well as decreased rates of neonatal abstinence syndrome when examining community-wide data in the six years since the programme began [33]. One study noted that the programme, with strong evidence of compassion and self-determination in their family-based, holistic health initiative driven by community, had attracted and retained more Indigenous clients than any previous programme in the region [42], although a similar programme with evidence of compassion and self-determination noted that their programme had comparable outcomes to mainstream (western) programmes [32].
One noteworthy finding in the articles regarding outcomes is the diversity of outcome measures. Compassion as an important programme mechanism means that treatment objectives move beyond a reduction in opioid use. In Ontario, a service provider workshop indicated that keeping families intact is one goal in the management of opioid dependent pregnant and postpartum women, meaning that counselling on parenting and life skills is an important component of treatment programmes [21]. Similarly, participants in another study noted improved parenting practices as one important outcome of the programme [23]. Other holistic outcome measures included improvement in living conditions and decreased involvement in the criminal justice system [40]. When not measured specifically, participants reported outcomes including improved self-esteem, gaining new skills, and feeling safer and more connected within the care system [39]. Papers also reported holistic community-wide outcome measures [23, 33, 34, 36]. These included declines in suicides, drug-related medical evacuations, criminal charges and child protection cases, and increased school attendance [34, 36]. Also reported was an improvement in community spirit and sense of purpose [34] and an increase in community cleanliness and safety [23]. Reported provider outcome measures included decreasing stigma among pharmacists [20], commitments to Indigenous-led, trauma-informed care [21], and improved access to treatment [20, 21], and increased use of treatment agreements with clients [27].
The impacts of individual or spiritual growth beyond addiction are difficult to capture in the current studies; however, even where outcomes were not specifically measured, participants indicated a need for compassion and self-determination in programmes without necessarily naming it as such. For instance, participants in one study stated a need for a healing place with integrated and holistic healthcare based in respect, and where they would have influence over the decisions and services that impact their healing [19]. Common findings across studies included a desire for more traditional, spiritual, and informal healing opportunities [24, 28, 30, 37, 39], and accessible community-based healing centres providing opportunities for both self-determination [2, 25] and community reconnection [24].
## Discussion
Most articles included in the present study provided evidence for compassion and control as important programs mechanisms demonstrated either through measured treatment outcomes or participant perspectives on the studied programs. Our results support two mid-range theories:Candidate Theory #1: Treatment and harm reduction models based in compassion (mechanism) for individuals and communities affected by trauma and structural violence (context) counter the stressors driving addiction and lead to successful outcomes.
The studies outlined that compassion requires trusting, respectful relationships between care providers, holistic treatment that provides emotional and spiritual support beyond biophysical treatment, as well as educational programs that support building life skills. Compassion ensures individuals do not experience gaps in care as they transition between systems, and they remain connected with their families and communities. Broadly, compassion operates at the individual level, but can be operated at a structural level through harm reduction policies and intercultural or trauma-informed training programs for providers. Candidate Theory #2: Treatment and harm reduction models that recognize Indigenous community self-determination (mechanism) through community leadership and culturally based models of care integrative of community knowledge and experience overcoming addiction (context) build on community resilience and lead to successful outcomes.
The studies outlined that self-determination requires community-based care programs that are community led through all stages of initiation, planning, management, and evaluation. Woven with elements of compassion, when self-determination is operated through collaborative partnerships, these partnerships must be based in trust, as well as respectful and meaningful engagement. Self-determination supports community ownership and participation because programs are more meaningful and appropriate for the local context. Broadly, self-determination operates at the structural level with resources and support for community-based programs, but is also important at the individual level, ensuring clients have voice in their treatment programs, ensuring that treatment is appropriate and meaningful to the client.
These mechanisms have been demonstrated across the literature included in the realist review, which includes communities from across Canada, as well as from Indigenous contexts globally. Additionally, the value of compassion has been demonstrated in the international literature, in numerous provider care frameworks for Indigenous clients, such as in Educating for Equity [46], holistic and comprehensive history taking that accounts for social determinants of health including colonization [47], and in a mindfulness and spiritual approach to suicide prevention [48]. The importance of self-determination was the focus of a scoping review of substance use included in this realist review [45] and has also been demonstrated as important in Indigenous health literature including in prenatal health promotion [14], and in Indigenous health promotion more broadly [49].
## Implications
The contexts, mechanisms, and outcomes presented here have a number of implications for care providers, and policy and systems decision-makers.
Context—Public health interventions and structural-level recommendations. Programme context factors such as stable funding and staff can be influenced in the short term with effective government policyContextual factors driving addiction including problem prescribing, poverty, and other social determinants of health are long-term goals, but could also be addressed with adequate program funding Mechanisms—Avenues for care providers and programme planners.3.Provider education on social determinants of health and trauma-informed care with an emphasis on the need for cultural humility may increase compassion4.Effective programming requires wrap-around supports that build skills, support healing, and strengthen family and community ties5.Programmes should be community-based in all stages: initiation, planning, management, and evaluation, or otherwise based in respectful collaborative partnerships6.Collaborative partnerships must include full information and be community-based where possible, for instance integrating OAT into traditional health and healing centres rather than integrating spirituality into OAT Outcomes—Considerations for researchers, care providers, and programme planners.7.Important outcomes include reduced incidence and so-called relapse, but also holistic and individual goals, for example, increased support networks, increased sense of purpose, increased involvement in community and culture, reduced numbers of Indigenous children being removed from their families, reduced incidence with the criminal justice system8.The international literature often focuses on the standard clinical outcomes of reduced incidence and reduced so-called relapse; however, additional outcome measures noted qualitatively in the studies included here are also valuable health outcomes that should be monitored, such as a cultural safety
## Study strengths
The realist review methodology employed here may be more relevant than systematic review methodology to Indigenous communities, leaders, policy-, and decision-makers who critique standard western approaches as reductionist and lacking adequate integration of local contexts. The integration of community perspectives and academic literature represents an important strength of this research. Additionally, the collaborative research partnership between university researchers and community partners ensures adherence to OCAP® principles and represents a key strength of this research.
## Study limitations
The analysis presented here focuses on academic literature and does not include grey literature that may provide thick descriptions and key insights on programme mechanisms. However, it should be noted that a review of the grey literature from the research team is available in a public report.1 This research was limited by the shortage of literature meeting our initial inclusion criteria. We were required to expand our inclusion criteria to include those studies with thin descriptions, which even so resulted in a relatively small number of studies focusing on opioid harm reduction models in Indigenous communities. As well, the literature was pulled in 2018 and, as a result, the research does not cover recent events such as the COVID-19 crisis and its affects on available programming. Because descriptions in some studies were thin, programme mechanisms had to be interpreted. This is not unusual practice for realist reviews, as program mechanisms are not always explicit [49, 50]. Future research that prioritizes thick program descriptions will better inform programmes and policy. If repeated, we would refine search terms to include “Maori”, as well as “methadone” and “buprenorphine”—common treatments for opioid use disorder.
## Conclusion
By identifying the programme mechanisms, realist reviews may support the adaptation of best practices into different contexts and as such may better support programme planning and implementation than conventional systematic literature reviews [14]. The findings from this realist review indicate compassion and self-determination as key programme mechanisms. This includes a need for care provision at the individual level that is based in respect, nonjudgmental care, individualized and holistic supports and goals, and that provides opportunities for clients to have a say in their treatment programmes. The findings indicate that at the structural level, compassion and self-determination can be maintained through policy and programming decisions that promote community-based programmes, trauma-informed care, and harm reduction. Doing so can support outcomes that go beyond reduced incidence of substance use and include mitigating systemic health inequities in Indigenous communities, addressing social determinants of health by strengthening communities, maintaining family connections, building life and job skills for individuals, and supporting a sense of purpose for individuals and communities.
## References
1. Hawe P, Potvim L. **What is population health intervention research?**. *Can J Public Health* (2009.0) **110** I8-I14. DOI: 10.1016/j.puhe.2019.03.027
2. Reid P, Cormack D, Paine S. **Colonial histories, racism and health—the experience of Māori and Indigenous peoples**. *Public Health* (2019.0) **172** 119-124. DOI: 10.1016/j.puhe.2019.03.027
3. King M, Smith A, Gracey M. **Indigenous health part 2: the underlying causes of the health gap**. *Lancet* (2009.0) **374** 76-85. DOI: 10.1016/S0140-6736(09)60827-8
4. Paradies Y. **Colonisation, racism and indigenous health**. *J Popul Res* (2016.0) **33** 83-96. DOI: 10.1007/s12546-016-9159-y
5. Czyzewski K. **Colonialism as a broader social determinant of health**. *Int Indigenous Policy J* (2011.0). DOI: 10.18584/iipj.2011.2.1.5
6. Greenhalgh T, Wong G, Westhorp G, Pawson R. **Protocol–realist and meta-narrative evidence synthesis: evolving standards (RAMESES)**. *BMC Med Res Methodol [Internet]* (2011.0) **11** 115. DOI: 10.1186/1471-2288-11-115
7. Nutton J, Fast E. **Historical trauma, substance use, and indigenous peoples: seven generations of harm from a “Big Event”**. *Subst Use Misuse [Internet]* (2015.0) **50** 839-847. DOI: 10.3109/10826084.2015.1018755
8. Sinha R. **Chronic stress, drug use, and vulnerability to addiction**. *Ann N Y Acad Sci* (2008.0) **1141** 105-130. DOI: 10.1196/annals.1441.030
9. Panksepp J, Herman B, Conner R, Bishop P, Scott JP. **The biology of social attachments: opiates alleviate separation distress**. *Biol Psychiatry* (1978.0) **13** 607-618. PMID: 83167
10. 10.AFNIGC, Alberta Government. Opioids and substances of misuse among First Nations people in Alberta. 2017.
11. 11.United Nations. United Nations declaration on the rights of Indigenous peoples [Internet]. Geneva; 2008 [cited 2018 Jun 15].
12. 12.TRC. Truth and Reconciliation Commission: calls to action. 2015;20. http://www.trc.ca/websites/trcinstitution/File/2015/Findings/Calls_to_Action_English2.pdf.
13. Pawson R, Greenhalgh T, Harvey G, Walshe K. **Realist review : a new method of systematic review designed for complex policy interventions**. *J Health Serv Res Policy* (2005.0) **10** 21-34. DOI: 10.1258/1355819054308530
14. Smylie J, Kirst M, McShane K, Firestone M, Wolfe S, O’Campo P. **Understanding the role of Indigenous community participation in Indigenous prenatal and infant-toddler health promotion programs in Canada: a realist review**. *Soc Sci Med [Internet]* (2016.0) **150** 128-143. DOI: 10.1016/j.socscimed.2015.12.019
15. 15.Henderson RI, Wadsworth I, Healy B, McInnes A, Danyluk A, Bill L, et al. Healing the whole human being: realist review of best practices and contextual factors for preventing & treating opioid misuse in Indigenous contexts in Alberta. Calgary, AB; 2021.
16. Navia D, Henderson RI, First CL. **Uncovering colonial legacies: voices of Indigenous youth on child welfare (dis)placements**. *Anthropol Educ Q* (2018.0) **49** 146-164. DOI: 10.1111/aeq.12245
17. Waller M, Okamoto S, Miles B, Hurdle D. **Resiliency factors related to substance use/resistance: perceptions of Native adolescents of the southwest**. *J Sociol Soc Welf* (2003.0) **30** 79-94
18. Molnar A, Campo PO, Ng E, Mitchell C, Muntaner C, Renahy E. **Protocol: realist synthesis of the impact of unemployment insurance policies on poverty and health**. *Eval Program Plan [Internet]* (2015.0) **48** 1-9. DOI: 10.1016/j.evalprogplan.2014.09.002
19. Benoit C, Carroll D, Chaudhry M. **In search of a healing place: Aboriginal women in Vancouver’s Downtown Eastside**. *Soc Sci Med* (2003.0). DOI: 10.1016/S0277-9536(02)00081-3
20. Duvivier H, Gustafson S, Greutman M, Jangchup T, Harden AK, Reinhard A. **Indian Health Service pharmacists engaged in opioid safety initiatives and expanding access to naloxone**. *J Am Pharm Assoc JAPhA [Internet]* (2017.0) **57** S135-S140. DOI: 10.1016/j.japh.2017.01.005
21. Jumah NA, Bishop L, Franklyn M, Gordon J, Kelly L, Mamakwa S. **Opioid use in pregnancy and parenting: an Indigenous-based, collaborative framework for Northwestern Ontario**. *Can J Public Health Rev Can Sante Publique [Internet]* (2017.0) **108** e616-e620. DOI: 10.17269/CJPH.108.5524
22. Kiepek N, Hancock L, Toppozini D, Cromarty H, Morgan A, Kelly L. **Facilitating medical withdrawal from opiates in rural Ontario**. *Rural Remote Health [Internet]* (2012.0) **12** 2193. PMID: 23094953
23. Landry M, Veilleux N, Arseneault J-E, Abboud S, Barrieau A, Bélanger M. **Impact of a methadone maintenance program on an Aboriginal community: a qualitative study: table 1**. *C Open* (2016.0). DOI: 10.9778/cmajo.20150076
24. Marquina-Márquez A, Virchez J, Ruiz-Callado R. **Postcolonial healing landscapes and mental health in a remote Indigenous community in subarctic Ontario, Canada**. *Polar Geogr* (2016.0). DOI: 10.1080/1088937X.2016.1155673
25. Momper SL, Delva J, Reed BG. **OxyContin misuse on a reservation: qualitative reports by American Indians in talking circles**. *Subst Use Misuse* (2011.0). DOI: 10.3109/10826084.2011.592430
26. Radin SM, Kutz SH, La Marr J, Vendiola D, Vendiola M, Wilbur B. **Community perspectives on drug/alcohol use, concerns, needs, and resources in four washington state tribal communities**. *J Ethn Subst Abuse* (2015.0). DOI: 10.1080/15332640.2014.947459
27. Srivastava A, Kahan M, Jiwa A. **Prescription opioid use and misuse Piloting an educational strategy for rural primary care physicians**. *Can Fam Phys* (2012.0) **58** 210-216
28. Teasdale KE, Conigrave KM, Kiel KA, Freeburn B, Long G, Becker K. **Improving services for prevention and treatment of substance misuse for Aboriginal communities in a Sydney Area Health Service. [Erratum appears in Drug Alcohol Rev. 2008 Jul;27(4):461]**. *Drug Alcohol Rev* (2008.0) **27** 152-159. DOI: 10.1080/09595230701829447
29. Uddin F. **Hope in Fort Hope: First Nations community is winning the battle against prescription drug abuse**. *Can Fam Phys [Internet]* (2013.0) **59** 391-393
30. Venner KL, Donovan DM, Campbell ANC, Wendt DC, Rieckmann T, Radin SM. **Future directions for medication assisted treatment for opioid use disorder with American Indian/Alaska Natives**. *Addict Behav* (2018.0) **86** 111-117. DOI: 10.1016/j.addbeh.2018.05.017
31. Beckstead DJ, Lambert MJ, DuBose AP, Linehan M. **Dialectical behavior therapy with American Indian/Alaska Native adolescents diagnosed with substance use disorders: combining an evidence based treatment with cultural, traditional, and spiritual beliefs**. *Addict Behav [Internet]* (2015.0) **51** 84-87. DOI: 10.1016/j.addbeh.2015.07.018
32. Black A, Khan S, Brown R, Sharp P, Chatfield H, McGuiness C. **An evaluation of opioid replacement pharmacotherapy in an urban Aboriginal Health Service**. *Aust New Zeal J Public Heal [Internet]* (2007.0) **31** 428-432. DOI: 10.1111/j.1753-6405.2007.00113.x
33. Dooley J, Ryan G, Gerber-Finn L, Bollinger M, Matsumoto C, Hopman W. **Maternal opioid use disorder and neonatal abstinence syndrome in northwest Ontario: a 7-year retrospective analysis**. *Can J Rural Med* (2018.0) **23** 39-44. PMID: 29547380
34. Kanate D, Folk D, Cirone S, Gordon J, Kirlew M, Veale T. **Community-wide measures of wellness in a remote First Nations community experiencing opioid dependence: evaluating outpatient buprenorphine-naloxone substitution therapy in the context of a First Nations healing program**. *Can Fam Phys* (2015.0) **61** 160-165
35. Katt M, Chase C, Samokhvalov AV, Argento E, Rehm J, Fischer B. **Feasibility and outcomes of a community-based taper-to-low-dose-maintenance Suboxone treatment program for prescription opioid dependence in a remote First Nations Community in Northern Ontario**. *J Aborig Health [Internet]* (2012.0) **9** 52-59
36. Mamakwa S, Kahan M, Kanate D, Kirlew M, Folk D, Cirone S. **Evaluation of 6 remote First Nations community-based buprenorphine programs in northwestern Ontario: Retrospective study**. *Can Fam Phys* (2017.0) **63** 137-145
37. Campbell AN, Turrigiano E, Moore M, Miele GM, Rieckmann T, Hu MC. **Acceptability of a web-based community reinforcement approach for substance use disorders with treatment-seeking American Indians/Alaska Natives**. *Community Ment Health J [Internet]* (2015.0) **51** 393-403. DOI: 10.1007/s10597-014-9764-1
38. Katzman JG, Fore C, Bhatt S, Greenberg N, Salvador JG, Comerci GC. **Evaluation of American Indian health service training in pain management and opioid substance use disorder**. *Am J Public Health* (2016.0) **106** 1427-1429. DOI: 10.2105/AJPH.2016.303193
39. Lee KS, Dawson A, Conigrave KM. **The role of an Aboriginal women’s group in meeting the high needs of clients attending outpatient alcohol and other drug treatment**. *Drug Alcohol Rev [Internet]* (2013.0) **32** 618-626. DOI: 10.1111/dar.12068
40. Saylors K. **The Women’s Circle comes full circle**. *J Psychoactive Drugs [Internet]* (2003.0) **35** 59-62. DOI: 10.1080/02791072.2003.10399994
41. Thomas G, Lucas P, Capler N, Tupper K, Martin G. **Ayahuasca-assisted therapy for addiction: results from a preliminary observational study in Canada**. *Curr Drug Abuse Rev* (2013.0). DOI: 10.2174/15733998113099990003
42. Williams N, Nasir R, Smither G, Troon S. **Providing opioid substitution treatment to Indigenous heroin users within a community health service setting in Adelaide**. *Drug Alcohol Rev* (2006.0). DOI: 10.1080/09595230600644699
43. Gray N. **Addressing trauma in substance abuse treatment with American Indian adolescents**. *J Subst Abuse Treat [Internet]* (1998.0) **15** 393-399. DOI: 10.1016/S0740-5472(97)00288-2
44. Gray D, Pulver LJ, Saggers S, Waldon J. **Addressing indigenous substance misuse and related harms**. *Drug Alcohol Rev [Internet]* (2006.0) **25** 183-188. DOI: 10.1080/09595230600644616
45. Russell C, Firestone M, Kelly L, Mushquash C, Fischer B. **Prescription opioid prescribing, use/misuse, harms and treatment among Aboriginal people in Canada: a narrative review of available data and indicators**. *Rural Remote Health* (2016.0) **16** 14
46. Crowshoe L, Henderson RI, Jacklin K. **Educating for equity care framework: addressing social contexts of type 2 diabetes with Indigenous patients**. *Can Fam Phys* (2018.0) **65** 25-33
47. Sherwood J. **Colonisation—it’s bad for your health: the context of Aboriginal health**. *Contemp Nurse* (2013.0) **46** 28-40. DOI: 10.5172/conu.2013.46.1.28
48. Le TN, Gobert JM. **Translating and implementing a mindfulness-based youth suicide prevention intervention in a Native American community**. *J Child Fam Stud* (2015.0) **24** 12-23. DOI: 10.1007/s10826-013-9809-z
49. Pyett P, Waples-Crowe P, van der Sterren A. **Challenging our own practices in Indigenous health promotion and research**. *Health Promot J Aust* (2008.0) **19** 179-183. DOI: 10.1071/HE08179
50. Wong G, Greenhalgh T, Westhorp G, Pawson R. **Development of methodological guidance, publication standards and training materials for realist and meta-narrative reviews: the RAMESES (Reality and Meta-Narrative Evidence Syntheses: Evolving Standards) Project**. *Health Serv Deliv Res* (2014.0) **2** 1-252. DOI: 10.3310/hsdr02300
|
---
title: 'Effect of propolis on mood, quality of life, and metabolic profiles in subjects
with metabolic syndrome: a randomized clinical trial'
authors:
- Sana Sadat Sajjadi
- Mohammad Bagherniya
- Davood Soleimani
- Mansour Siavash
- Gholamreza Askari
journal: Scientific Reports
year: 2023
pmcid: PMC10022550
doi: 10.1038/s41598-023-31254-y
license: CC BY 4.0
---
# Effect of propolis on mood, quality of life, and metabolic profiles in subjects with metabolic syndrome: a randomized clinical trial
## Abstract
Metabolic syndrome (MeS) is a common multifaceted disorder. Plants contain antioxidant bioactive compounds, which are beneficial to improve the health condition of patients with MeS. Propolis is a hive natural product that is composed of various constituent. We aimed to assess the effects of *Iranian propolis* as a natural and safe agent on indicators of MeS, quality of life and mood status in individuals with MeS. In total, 66 interested eligible patients recruited to the present study. Participants were randomly assigned to consume a tablet at dose of 250 mg of propolis extract, twice daily for 12 weeks or placebo. Propolis supplementation could lead to a significant reduction in waist circumference (WC), increase in physical functioning, general health and the overall score of SF-36 compared with placebo group (P-value < 0.05). However, no significant differences were observed regarding other anthropometric indices and biochemical parameters between two groups (P-value > 0.05). The current study indicated that propolis can be effective in decreasing WC and improving physical health and quality of life, while had no significant effects on other components of MeS among subjects with this syndrome.
Clinical trials registration Iran Registry of Clinical Trials.ir IRCT20121216011763N49, registration date $\frac{23}{12}$/2020.
## Introduction
Metabolic syndrome (MeS) is a common multifaceted disorder. It is defined by pathological criteria including abdominal obesity, hypertension, impaired fasting blood glucose, and dyslipidemia1,2. People with MeS are at risk for various comorbidities such as cardiovascular diseases, cancers, polycystic ovary syndrome, and type 2 diabetes mellitus3–5. Also, there is a strong association between MeS and the development of oxidative stress and inflammatory condition6. MeS is regarded as one of the health concerns worldwide. Reports have shown that MeS imposes a great financial burden on health care system and its risk factors can lead to reduce mental and physical heaths and quality of life in individuals7. Data have indicated that number of people who suffer from this disorder. In a meta-analysis performed by Tabatabaei-Malazy et al., the prevalence of MeS in Iran based on the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP-III) criteria was reported $38.3\%$8. The etiology of MeS is not completely well understood yet. However, abdominal obesity and insulin resistance play a significant role on MeS pathogenesis. It has been reported that the main treatment approaches of MeS comprise lifestyle modification including dietary changes and increased physical activity9,10. However, it appears that these approaches may be unsuccessful due to poor compliance in a long term11. Recently, several clinical studies have shown potential effects of nutraceuticals and plant products in reduction of chronic disease complications and control MeS components. Plants contain antioxidant bioactive compounds, which are beneficial to improve the health condition of patients with MeS12.
Propolis is a hive natural product with various constituents. It has been recognized that different kinds of propolis contain active components including, phenolic acids, terpenes, amino acids, vitamins, numerous essential metals, and elements. Chemical composition of propolis is varied by bee species, plant origin, region of collection and, climate affect. Propolis is made by various bee species (Apis mellifera, stingless bees Meliponini and others) which play a major role on constituents of propolis13–17. In the past, propolis has been used as a folk medicine in treatment of infections and wounds18. Nowadays, propolis is prescribed as a popular dietary supplement to promote the body's health. Recently, propolis has been suggested to have various biological properties, including anti-tumor, antioxidant, antibacterial, anti-atherogenic, and anti-inflammatory activities19–23. Several animal and human studies have supported that propolis is effective in improving blood pressure, regulating glucose and lipid metabolisms, and enhancing the immune system function24–30. However, some investigations show inconsistent results. We observed that heterogeneity results for glycemic indices and lipid profiles are high. It might due to difference in form, dosage of used propolis, and duration of study. Instant to, result of a meta-analysis included six trials revealed that propolis significantly improved fasting plasma glucose, but did not effect on serum insulin and homeostasis model assessment-insulin resistance (HOMO-IR)31. A clinical study reported that *Iranian propolis* intake (1000 mg/d) in diabetic patients for 90 days significantly reduced HOMO-IR, inflammatory biomarkers and also increase level of high-density lipoprotein cholesterol (HDL-C) compared with controls32. However, a recent meta-analysis found that propolis supplementation alone or along with other components reduced serum fasting blood glucose, (FBG), hemoglobin A1c, and insulin, but had no effect on HOMO-IR and lipid profiles33. Also, a clinical study demonstrated that propolis intake with a dose of 500 mg/day for four months had no significant effect on lipid profiles and glycemic indices among patients with nonalcoholic fatty liver disease (NAFLD)34. Moreover, it is important to mention that most of these studies have evaluated the effect of propolis consumption on patients with type 2 diabetes.
To our knowledge, no randomized clinical trial has not been directly measured the effects of propolis on metabolic factors in patients with MeS. Therefore, we aimed to assess the effects of *Iranian propolis* as a natural and safe agent on indicators of MeS, quality of life and mood status in individuals with MeS.
## Study design
The current study is a prospective, parallel, randomized, double-blind, placebo-controlled clinical trial. The study was conducted on subjects with MeS referred to an endocrine and metabolism research center and an outpatient clinic affiliated to Isfahan University of Medical Sciences, Isfahan, Iran. The ethics committee of Isfahan university of medical science reviewed and approved the study protocol (ID: IR.MUI.RESEARCH.REC1399.595). The clinical trial was registered at the Iranian Registry of Clinical Trial (Code: IRCT20121216011763N49). All methods were conducted based on the approved study plan, as well as with relevant guidelines and regulations.
## Participants
After evaluation the recorded information in the patient files based on inclusion criteria, in total, 66 interested eligible patients recruited to the present study through telephone calls and direct invitation from February 2021. We informed all individuals about study purpose and procedures. Written informed consent was signed by all participants before their participation in the study. The inclusion criteria consisted the following: Individual who had MeS when they met at least three or more of NCEP ATP-III criteria35: fasting plasma glucose ≥ 100 mg/dL, triglyceride (TG) ≥ 150 mg/dL HDL-C < 40 mg/dL in men or < 50 mg/dL in women, waist circumference (WC) ≥ 102 cm in men or ≥ 88 cm in women, blood pressure ≥ $\frac{130}{85}$ mmHg, adults 20–60 years old, having the ability to read and write, a willingness to participate in the study and no change in type and dosage of lipid-lowering, hypotensive or hypoglycemic drugs over the past three months. The exclusion criteria consisted the following: pregnancy, breastfeeding, a sensitivity to bee products, use of smoking, alcohol and drugs, insulin therapy, patients who follow a weight loss diet or exercise program, a history of malignancy and cancer, type 1 diabetes, nephrotic syndrome, kidney and lung diseases, bile diseases and HIV. Patients who are unwilling to continue cooperating, become lactating and pregnant, have the sensitivity to propolis supplement, and suffer a specific disease during the study were withdrawn from follow-up.
## Randomization
All participants were randomly allocated to either propolis ($$n = 33$$) or placebo ($$n = 33$$) groups. Randomization was stratified according to sex (male vs. female), with the use of permuted block size of 4. The assignment sequences were provided by an independent statistician with the use of a random-number table and then were kept in opaque, sealed, numbered envelopes until the end of the eligibility criteria evaluation. Tablet containers were coded as A and B to order to allocation concealment. The study pharmacist coded tablet containers as A and B according to randomized list. Treatment assignments were concealed from researchers and all patients until the completion of data analyses, with the exception of the pharmacist.
## Interventions
Participants in intervention group were asked to consume a 350 mg propolis tablet (containing 250 mg of Iranian green propolis extract and 100 mg of safe and ineffective combination of microcrystalline cellulose) twice a day (a total of 500 mg Iranian green propolis extract/day) and the control group were asked to intake a same placebo tablet (containing 350 mg of microcrystalline cellulose) twice a day, one tablet before lunch and one tablet before dinner, for 12 weeks. Propolis sample was obtained from honey bee (Apis mellifera) colonies located in Rasht, a region in the north of Iran, in the summer season. First, propolis was ground and extracted with $70\%$ ethanol at a ratio of 1:8. Then, the solution was sonicated by ultrasonic bath (Backer vCLEAN1-L20 Ultrasonic, Backer Co., Tehran, Iran) at 20 kHz and 35 degrees centigrade for 45 min. According to Bankova recommendation for chemical standardization of poplar propolis, total polyphenols content and total flavonoids content in poplar propolis tablets were measured using the spectrophotometric assay (JENWAY 7305, Bibby Scientific Ltd. Stone) based on the Folin–Ciocalteu reducing capacity and aluminum complex formation methods, respectively36. Each propolis tablet contains 90 mg gallic acid equivalent and 67 mg flavonoids. In this double-blind study, Pharmaceutical Company of Naghsh Jahan Ryhan (Isfahan, Iran), under the supervision of the study pharmacist produced placebo and propolis tablets with the same size, color, odor, form and packing. All the components of propolis and placebo tablets were manufactured totally the same except the bioactive component of propolis. The propolis and placebo groups conformed the same protocol (twice daily, before lunch and dinner). Investigators, participants, outcome assessors, researchers who measured anthropometric assessments, trained participant on how to fill out the questionnaires and laboratory staff and data analyzers were blinded to treatment assignment until the completion of data analyses.
At the beginning of the study, participants of both groups obtained healthy lifestyle recommendations. Regular use of supplements was reminded to patients through short message service and telephone call weekly and every two weeks, respectively. Also, their compliance to the study was evaluated by counting unused supplements at each visit. A 3-day food record (two weekdays and one weekend day) as the “gold standard” used for dietary intake assessment of each subject. The participants trained by a nutritionist who was unaware to the treatment allocation on how to complete 3-day food records at the beginning, the middle (weeks six), and the end of the intervention. Nutritionist IV software (First Databank Inc., Hearst Corp., San Bruno, CA, USA), which is adapted for Iranian foods was used to calculate the value of nutrition and calorie intake. A 3-day physical activity record (two weekdays and one weekend day) was used to assess physical activity of participants based on metabolic equivalent (MET)-h/day values.
## Measurements
Demographic characteristics, including age, sex, marital status, medical history, level of education, household status, and current drugs use were collected from each participant by completing a general questionnaire. Anthropometric variables of each patient including weight, height, body mass index (BMI), WC, and blood pressure were measured at the beginning and at the end of the study, after an overnight fast. Body weight was measured using a calibrated hand scale to the nearest 0.1 kg while the subjects wearing minimal clothing and no shoes (Seca, Germany). The scale was calibrated daily by a five-kilogram weight. Height without shoes in the standing position, while shoulders were in a normal position was measured using a stadiometer to the nearest 1 cm (Seca, Germany). Then, BMI was calculated as body weight in kilograms/body height in meters squared. WC was measured between the lower rib margin and the iliac crest at the end of normal exhalation, without any pressure to body surface using an inelastic tap, to the nearest 0.1 cm. an Experienced nutritionist measured blood pressure using a standard hand-held sphygmomanometer (ALPK2, Zhejiang, China; Datis Co, Tehran, Iran) over the right arm for each person, twice after the individuals had been sitting for 15 min. The average of two measurements was considered as the final blood pressure.
At the beginning and at the end of the intervention, a 10 cc venous blood sample was taken from participants after overnight 10–12 h fasting in the endocrine and metabolism research center laboratory (Esfahan, Iran). After separation of serum samples from whole blood, all aliquots were stored at −80 °C until biochemical analysis time. Serum insulin levels were measured using the (enzyme-linked immunosorbent assay) ELISA kit (Pars Azmoun kit, Tehran, Iran). FBG, concentration of serum cholesterol total (TC), TG, low-density lipoprotein cholesterol (LDL-C), and HDL-C were determined using the colorimetric technique by available standard kits (Pars Azmoun kit, Tehran, Iran). Also, LDL-C/HDL-C and cholesterol/HDL-C ratios were calculated. HOMO-IR was determined using the following formula: HOMA-IR = fasting glucose (mg/dl)*fasting insulin (μU/ml)/40537. Serum C-reactive protein (CRP) level was measured by the use of an immunoturbidimetric method (Bionik Diagnostic System, Tehran, Iran) and reagent kite (Pars Azmoun kit, Tehran, Iran).
## Quality of life
Quality of life was evaluated using 36-Item Short Form Health Survey (SF-36) by direct interview with participants at the beginning and the end of the intervention. SF-36 consisted of 36 questions that evaluate eight different domains of health. These eight domains can be summarized in to two components including physical component score and mental component score. Physical component score comprised domains of physical functioning, role limitations due to physical health, bodily pain and general health. Mental component score comprised domains of role limitations due to emotional problems, energy and fatigue, emotional well-being, and social functioning. The minimum and maximum scores in this questionnaire are zero and 100, respectively. Better quality of life is indicated by obtained higher scores38.
## Depression, anxiety and stress scale
The participants were asked to complete the Depression, Anxiety and Stress (DASS-21) questionnaire at the beginning and 12 weeks after the intervention. The DASS-21 as a tool to assess mental health, has 21 questions and three subscales (depression, stress, and anxiety). Each subscale of DASS-21 consist of seven questions39,40. Scores of the three subscales are calculated by summing the relevant responses and multiplying by two41. The minimum and maximum scores obtained in each subscale range between zero to 42, and lower score indicates a better situation of anxiety, depression and stress42.
## Dose of propolis
A study performed by Zakerkish et al. showed that the intake of 1000 mg/day of Iranian raw propolis supplement (the equivalent of 500 mg of propolis extract) for 3 months reduced HOMO-IR in patients with type 2 diabetes mellitus, without any side effects43. Also, Soleimani et al. revealed that the consumption of 500 mg/day of propolis extract improved hepatic steatosis and fibrosis among NAFLD patients34. Therefore, based on similar studies, propolis supplement was considered at a daily dose of 500 mg of propolis extract.
## Statistical analysis
All data were analyzed using SPSS 16 software (SPSS, Inc., Chicago, IL). The normal distribution of data was checked using Kolmogorov–Smirnov test and Q–Q plot. Data are presented as frequencies (percentage) for qualitative variables, mean (± SD) for normally distributed continuous data, or median (25th, 75th) for other variables. Within-group changes were assessed with the use of paired t test for normally distributed data, Chi-square test or Fisher's exact test for nominal variables, and Wilcoxon rank-sum test for other data. The between-group differences were assessed with the use of independent t test for normally distributed data, Chi-square test or Fisher's exact test for nominal variables, and Mann–Whitney U test for other data. Analysis of covariance (ACNOVA) (adjusted for baseline values) was used to detect any differences between the two groups at the end of the study. P-value < 0.05 was considered statistically significant. Sample size was calculated based alpha of 0.05, beta of 0.20 (power of $80\%$) and effect size of one unit. HOMA-IR was considered as a primary outcome variable. By considering $20\%$ drop-out rate, in total, the sample size was estimated 30 subjects in each group43.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{N}= \left(\frac{1+\varphi }{\varphi }\right)\left[\frac{{({Z}_{1-\frac{\alpha }{2}}+ {Z}_{1-\beta })}^{2}}{{\Delta }^{2}}+ \frac{{Z}_{1-\alpha /2}^{2}}{\varphi }\right]$$\end{document}$$n = 1$$+φφ(Z1-α2+Z1-β)2Δ2+Z1-α/22φ
## Ethical approval and consent to participate
The ethics committee of Isfahan university of medical science reviewed and approved the study protocol (ID: IR.MUI.RESEARCH.REC1399.595). The clinical trial was registered at the Iranian Registry of Clinical Trial (Code: IRCT20121216011763N49). Written informed consent was signed by all participants before their participation in the study.
## Results
Among the 66 participants with metabolic syndrome who enrolled in the clinical trial, 62 participants completed the trial ($$n = 33$$ in the propolis and $$n = 29$$ in the placebo, groups). During the intervention, four subjects from the placebo group were excluded due to move ($$n = 1$$), gastrointestinal side effects ($$n = 1$$), and not willing to continue ($$n = 2$$) (Fig. 1).Figure 1Flow chart of study participants.
Table 1 shows baseline characteristics of participants who completed the study in both groups. The mean (± SD) of age in propolis group was 54.27 ± 6.58 years and in placebo group was 53.86 ± 5.60 years (P-value = 0.794). Fifty-seven women (30 persons in the intervention group) and five men (three persons in the intervention group) completed the trial study. At baseline, no statistically significant differences were found between groups in terms of demographic characteristics, body weight, BMI, WC, systolic blood pressure, diastolic blood pressure, consumption of vitamin supplements, current medications, and the number of subjects who had current diseases. Based on 3-days food records, there were no statistically significant differences in total energy and nutrients intakes within or between two groups. Moreover, the mean (± SD) of physical activity did not change between two groups (0.13 ± 2.67 MET-h/day in the propolis group vs. −0.27 ± 2.93 MET-h/day in the placebo group, P-value = 0.241) (Table 2).Table 1General characteristic of study participants in propolis and placebo groups at baseline. VariablesPropolis group ($$n = 33$$)Placebo group ($$n = 29$$)P-valueSex, n (%)Female30 (90.9)27 (93.1)0.99#Age, year54.27 ± 6.5853.86 ± 5.600.794Weight, kg79.04 ± 8.6182.01 ± 13.410.298BMI, kg/m232.56 ± 4.1334.03 ± 4.780.201WC, cm107.16 ± 7.88109.82 ± 9.990.246Systolic blood pressure, mmHg12.83 ± 1.6713.01 ± 1.620.663Diastolic blood pressure, mmHg8.37 ± 1.068.55 ± 1.200.551Vitamin supplement, n (%)2(6.1)0[0]0.494Current diseaseThyroid disease, n (%)5(15.2)3(10.3)0.713#Gastrointestinal disease, n (%)10(30.3)12(41.4)0.363Liver disease, n (%)6(18.2)6 (20.7)0.803Cardiovascular diseases, n (%)6(18.2)7(24.1)0.565Current medicationLipid-lowering medication, n (%)29(87.9)26(89.7)0.99#Sugar-lowering medication, n (%)29(87.9)26(89.7)0.99#Pressure-lowering medication, n (%)19(57.6)18(62.1)0.719Levothyroxine medication, n (%)5(15.2)3(10.3)0.713#Change weight, n (%)9(27.3)6(20.7)0.546Marital status, n (%)0.570#Married29 (87.9)25 (86.2)Divorced/widowed4(12.1)4(61.38)Household status, n (%)0.510#Supervisor/Self- supervisor7(21.2)7(24.1)Under supervision26(78.8)22(75.9)Educational level, n (%)0.357#Under-diploma/Diploma30[91]28(96.6)University3(9.1)1(3.4)History of various diseases n (%)33 [100]28 (96.6)0.468#Values are presented as mean ± standard deviation and frequencies (percentages). Abbreviation: BMI, body max index; WC, waist circumference. P-values were derived from the independent-sample t test for quantitative variables and Chi-square test for qualitative variables between the two groups. # Fisher's exact test. Table 2Dietary intake and physical activity of study participants before and after of the intervention. VariablesGroupsBeforeAfterChangesP-value bP-Value #Energy, kcal/dayPropolis1538.67 [1340.16–2107.91]1539.74 [1249.38–1857.92]−39.45 ± 417.110.6620.659Placebo1413.04 [1190.56–2139.44]1393.78 [1122.37–1812.83]16.44 ± 571.090.940Protein, g/dayPropolis67.62 [56.78–99.34]67.04 [51.89–91.16]−1.69 ± 26.810.8090.953Placebo67.32 [58.17–85.48]64.31 [50.96–83.26]−2.07 ± 23.580.738Carbohydrates, g/dayPropolis255.25 [204.02–362.55]237.94 [189.43–293.95]−10.48 ± 64.330.5610.396Placebo199.85 [159.82–330.28]214.82 [157.82–309.91]7.61 ± 100.460.787Fat, g/dayPropolis42.38 [34.47–57.03]42.57 [35.15–49.31]−0.08 ± 16.860.6230.739Placebo43.24 [32.92–58.08]41.5 [28.58–55.64]−1.58 ± 18.360.854SFA, g/dayPropolis13.30 [11.22–16.86]13.48 [11.21–16.87]1.30 ± 90.9640.949*Placebo17.02 [11.22–19.87]14.14 [10.58–20.66]0.27 ± 8.170.787MUFA, g/daypropolis14.15 [12.15–18.61]14.20 [11.96–15.46]0.14 ± 6.950.7410.362Placebo15.17 [11.66–20.82]13.98 [9.98–19.52]−1.51 ± 7.210.496PUFA, g/dayPropolis8.18 [6.65–10.56]7.65 [6.05–10.10]−0.39 ± 3.830.5980.952Placebo8.12 [4.99–11.01]7.35 [5.64–9.76]−0.45 ± 3.840.922Cholesterol, mg/dayPropolis200.19 [115.40–284.92]198.35 [131.17–264.21]14.77 ± 135.800.6880.530*Placebo187 [126.11–308.18]148.85 [119.73–285.53]38.59 ± 345.680.642Fructose, g/dayPropolis19.12 [13.6–24.62]16.58 [8.53–23.35]−2.07 ± 9.110.4370.334*Placebo12.09 [7.05–17.68]13.81 [7.17–19.80]2.01 ± 13.270.596Magnesium, mg/dayPropolis411.42 [270.56–566.36]360.03 [285.64–468.90]−23.21 ± 216.320.4370.857Placebo345.61 [248.72–442.09]295 [219.87–394.40]−32.34 ± 174.760.469Zinc, mg/dayPropolis11.92 [7.63–14.34]10.39 [8.39–13.35]−0.26 ± 5.210.6750.652Placebo10.61 [8.47–12.91]9.65 [6.65–12.87]−0.84 ± 4.990.596Selenium, ug/dayPropolis102.29 [84.51–128.74]95.50 [77.82–122.70]2.92 ± 48.130.6620.334*Placebo84.32 [71.35–128.31]80.91 [65.29–103.38]−12.63 ± 50.220.198Calcium.mg/dayPropolis1164.55 [578.14–1998.19]1062.07 [550.04–1434.69]−221.18 ± 1082.620.2720.668Placebo1064.78 [562.33–1364.04]815.97 [439.13–1284.16]−113.59 ± 846.160.496Vitamin C, mg/dayPropolis117.82 [53.85–228.71]84.45 [66.33–157.08]−24.89 ± 106.550.3130.822Placebo95.69 [67.65–187.27]80.79 [42.93–131.97]−30.50 ± 86.270.122Fiber, g/dayPropolis36.38 [20.90–55.55]31.40 [21.80–45.30]−2.93 ± 22.690.4480.913Placebo29.88 [19.24–37.12]21.34 [16.28–37.65]−3.53 ± 19.420.496Physical activity (MET-h/week)Propolis32.56 ± 2.5832.69 ± 3.070.13 ± 2.670.781a0.241*Placebo34.62 ± 3.7934.35 ± 2.46−0.27 ± 2.930.624aValues are presented as mean ± standard deviation or median [interquartile range]. Abbreviation: SFA, saturated fatty acids; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids. * P-value were derived from Mann–Whitney U test. # P-value were derived from Independent Sample t test. bP-value were derived from Wilcoxon rank-sum test.aP-value were derived from the Paired- sample t-test.
As reported in Table 3, propolis supplementation significantly reduced the mean weight, BMI, and systolic blood pressure in the propolis group (P-value = 0.016, P-value = 0.016, and P-value = 0.020, respectively). However, the differences were not significant between two groups (P-value = 0.550, P-value = 0.217, and P-value = 0.366, respectively). Also, there was a significant reduction in the mean WC compared with baseline value in the propolis group (P-value < 0.001) and this difference was statistically significant compared with the placebo group (P-value = 0.008). A significant reduction in serum concentrations of TC compared with baseline value was observed in the propolis group (P-value = 0.039). In addition, HDL-C level significantly reduced in both groups. However, we did not find any significant differences between two groups for serum levels of TC, HDL-C, LDL-C, CRP, insulin, FBG, TG, LDL/HDL ratio, cholesterol/HDL ratio, and HOMO-IR.Table 3The effects of propolis supplementation on anthropometric indices, blood pressure, inflammation status, glycemic parameters and lipid profiles in two groups. VariablesGroupsBeforeAfterChangesP-Value bP- value#Weight, kgPropolis79.05 ± 8.6177.97 ± 8.52−1.08 ± 2.430.016Placebo82.02 ± 13.4281.64 ± 13.08−0.38 ± 1.980.3100.550*P-Value0.298BMI, kg/m2Propolis32.57 ± 4.1332.12 ± 4.06−0.45 ± 10.016Placebo34.03 ± 4.7933.88 ± 4.73−0.15 ± 0.840.3480.217P-Value0.201Waist circumference, cmPropolis107.17 ± 7.89104.08 ± 6.96−3.09 ± 4 < 0.001Placebo109.83 ± 9.99109.66 ± 10.26−0.17 ± 4.360.8330.008P-Value0.246SBP, mmHgPropolis12.83 ± 1.6712.16 ± 1.23−0.67 ± 1.580.020Placebo13.02 ± 1.6312.67 ± 1.37−0.34 ± 1.210.1370.366P-Value0.663DBP, mmHgPropolis8.38 ± 1.078.08 ± 0.91−0.30 ± 1.410.237Placebo8.55 ± 1.218.26 ± 0.85−0.29 ± 1.310.2370.995P-Value0.5510.438LDL/HDL ratioPropolis2.06 ± 0.472.06 ± 0.32−0.01 ± 0.420.934Placebo2 ± 0.572.11 ± 0.390.11 ± 0.480.2110.296P-Value0.631Cholesterol/HDL ratioPropolis3.66 ± 0.453.70 ± 0.440.05 ± 0.410.525Placebo3.60 ± 0.553.72 ± 0.440.12 ± 0.550.2570.760P-Value0.673TG, mg/dlPropolis239.97 ± 169.76234.33 ± 144.94−5.64 ± 212.180.880Placebo224.34 ± 129.58288.34 ± 19264 ± 225.160.1370.215P-Value0.688Insulin, μU/mlPropolis4.87 [3.73–8.81]7.34 [3.88–8.88]0.75 ± 3.690.357aPlacebo5.38 [3.2–8.89]6.37 [4.12–9.47]0.64 ± 2.510.163a0.890P-Value0.849FBG, mg/dlPropolis126 [112.5–168.5]123 [104–172]−5.06 ± 57.110.381aPlacebo123 [108–174.5]133 [114–160]0.48 ± 53.530.387a0.197*P-Value0.827HOMO-IRPropolis1.93 [1.22–2.88]1.90 [1.29–3.05]0.42 ± 2.330.662aPlacebo1.57 [1.04–2.96]1.84 [1.26–3.43]0.33 ± 1.280.107a0.805*P-Value0.530TC, mg/dlPropolis168 [143.5–198]141 [129–187]−9.42 ± 34.980.039aPlacebo180 [150.5–199.5]162 [140–201.5]−10.55 ± 31.410.106a0.895P-Value0.489HDL-C, mg/dlPropolis45 [39.5–57]41 32,36–50−2.52 ± 14.420.022aPlacebo50 [42.5–54.5]44 [37.5–51.5]−4.14 ± 8.650.003a0.865*P-Value0.259LDL-C, mg/dlPropolis93 [77.5–109.5]79 [69.5–109]−3.70 ± 37.730.114a0.682*Placeo94 [76–124.5]90 [78–123]−2.52 ± 29.030.406aP-Value0.899CRP, mg/lPropolis1.4 [0.2–2.65]1.8 [0.25–4.3]0.56 ± 3.590.084aPlacebo1.8 [0.2–5.15]2.8 [1–5.1]−0.48 ± 7.230.838a0.341*P-Value0.329Values are presented as mean ± standard deviation and median [interquartile range]. Abbreviation: FBS, fasting blood sugar; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; HOMA-IR, homeostasis model of assessment-insulin resistance; CRP, C- reactive protein; TC, total cholesterol; SBD, systolic blood pressure; DBP, diastolic blood pressure; BMI, body mass index. * P-value were derived from Mann-Whiney U test. # P-value were derived from Independent t test. b P-value were derived from paired-sample t test. a P-value were obtained from Wilcoxon rank-sum test.
The effects of propolis supplementation on mood status and quality of life before and after of intervention are reported in Table 4. The results showed that the mean score of anxiety was significantly reduced in the propolis group (P-value = 0.023), however in compression with the placebo group the mean scores of anxiety, stress, and depression were not significant (P-value = 0.921, P-value = 0.071, and P-value = 0.250, respectively). After the intervention, the mean physical functioning (P-value < 0.001), general health (P-value < 0.001), and the total score of SF-36 (P-value < 0.001) was significantly increased in the propolis group compared with the placebo group. Moreover, energy fatigue domain was increased in the propolis group (P-value = 0.005), however this difference was not significant between two groups. There was a significant different between groups regarding bodily pain (P-value = 0.015) (Supplementary file 1).Table 4The effects of propolis supplementation on mood status and quality of life before and after of intervention. VariablesGroupsBeforeAfterChangesP-value aP-value #DASS-21DepressionPropolis10.5 ± 8.949.39 ± 9.09−1.94 ± 7.700.164Placebo11.93 ± 6.7912.43 ± 10.570.44 ± 8.010.7750.250P-value0.494AnxietyPropolis13.13 ± 9.3811.27 ± 7.78−2.69 ± 6.360.023Placebo15.07 ± 9.1612.71 ± 9.68−2.52 ± 6.620.0590.921P-value0.421StressPropolis15.63 ± 10.2414.60 ± 9.49−1.75 ± 9.040.282Placebo19.36 ± 10.2421.21 ± 9.881.85 ± 5.870.1130.071p-value0.169SF-36physical functioningPropolis82.58 ± 15.8291.36 ± 9.948.79 ± 12.75 < 0.001Placebo66.92 ± 18.8570.17 ± 19.343.25 ± 12.970.188 < 0.001*P-value < 0.001Role limitation due to emotion problemPropolis44.44 ± 43.8357.58 ± 48.7913.13 ± 55.860.165Placebo40.23 ± 45.7642.53 ± 47.892.30 ± 53.40.0560.440P-value0.713Limit to physicalpropolis42.42 ± 46.5653.03 ± 46.2510.61 ± 51.930.3520.401*Placebo22.41 ± 39.1627.59 ± 43.995.17 ± 51.050.186P-value0.071Energy fatiguePropolis59.70 ± 22.8167.73 ± 17.058.03 ± 15.150.005Placebo50.69 ± 19.9452.07 ± 19.621.38 ± 17.060.6670.109P-value0.105Emotional well beingPropolis66 ± 19.4968.61 ± 18.792.61 ± 12.930.255Placebo53.14 ± 20.2054.38 ± 22.091.24 ± 17.310.7020.222*P-value0.013Social functioningPropolis72.35 ± 28.9474.95 ± 27.242.61 ± 29.900.6200.732Placebo71.12 ± 29.7176.72 ± 32.695.60 ± 38.610.441P-value0.870Bodily painPropolis69.39 ± 33.6976.74 ± 28.027.35 ± 38.560.282Placebo46.98 ± 33.6850 ± 332.403.02 ± 30.600.60.015*P-value0.011General healthPropolis53.62 ± 19.0668.94 ± 19.4815.32 ± 19.83 < 0.001Placebo49.61 ± 15.1750.13 ± 16.340.52 ± 15.460.858 < 0.001p-value0.367Totalpropolis62.87 ± 15.3673.02 ± 13.4110.15 ± 11.33 < 0.001Placebo51.53 ± 15.9354.16 ± 16.652.64 ± 11.370.222 < 0.001*p-value0.006Values are presented as mean ± standard deviation. Abbreviations: Dass-21, the 21-Item Depression Anxiety and Stress Scale; SF-36; the 36-Item Short Form Health Survey. * P-value were derived from ANCOVA test with baseline values as the covariate. # P-value were obtained from Independent Sample t test. a P-value were derived from paired-sample t test.
## Discussion
The findings of this double-blind, randomized and placebo-controlled study demonstrated that daily intake of 500 mg *Iranian propolis* extract in subjects with MeS for 12 weeks could lead to a significant reduction in WC, increase in physical functioning, general health, and the overall score of SF-36. While, propolis supplementation had no effects on other anthropometric indices and biochemical parameters compared to the placebo group.
In the propolis group, compared with baseline, at the end of the study, a significant reduction was observed in anthropometric indices including WC, BMI, and body weight. Nonetheless, significant differences were observed between two groups for WC, but not regarding BMI and body weight. In three previous interventional studies that assessed the effects of propolis on WC measurement44–46, propolis had no considerable effects on WC. In addition, some clinical trials23,27,34,43–46 indicated that propolis consumption had no favorable effects on weight and BMI. However, a clinical trial showed that daily intake of 900 mg of raw propolis supplementation for 12 weeks in diabetic subjects could reduce weight and BMI24. Nevertheless, a previous study among healthy subjects reported that 1000 mg daily raw propolis consumption for 60 days significantly increased BMI and weight47. It seems that the controversies in outcomes of researches might be related to differences in form and amount of used propolis, the study population, duration of intervention or the effects of confounders such as change in dietary intakes. Although, finding of current meta-analysis showed that propolis had no significant effects on anthropometric indices33. In preclinical studies, several anti-obesity mechanisms for propolis are considered. For example, it has been shown that propolis had a role in expression of factors that involved in lipid metabolism. Propolis inhibits accumulation of visceral adipose tissues and weight gain through controlling factors such as, SREBP- 1, SREBP-2, and down-regulation of PPARγ protein in the adipocytes48,49. PPARα protein regulates genes associated with fatty acid degradation. The increase of PPARα protein in liver by propolis can lead to increase β-oxidation of fatty acids50. Furthermore, it has been suggested that propolis can prevent the absorption of fat from the intestines in animal models29.
We found that propolis supplementation for 12 weeks had no effects on FBG, serum insulin, and HOMO-IR among subjects with MeS. Consist with our study, the result of a trial indicated that intake 500 mg/day of propolis extract for four months did not improve FBG, serum insulin, and HOMO-IR in patients with NAFLD34. Data from two interventional studies reported that daily supplementation of 900 mg raw propolis during 18 weeks had no beneficial effects on FBG, serum insulin and Hemoglobin A1c among patients with type 2 diabetes mellitus28,51. Also, some interventional studies found no significant association between propolis and HOMO-IR and serum insulin24,34,45,47,52. Conversely, a clinical study indicated that propolis extract supplement at a daily dosage 1000 mg for 90 days improved glycemic parameters in diabetic subjects43. Also, other report showed that daily supplementation of 1500 mg raw propolis for eight weeks significantly reduced FBG, serum insulin, and HOMO-IR in diabetic patients25. A currently published meta-analysis conducted by Hallajzadeh et al. showed that propolis supplements lead to reduce FBG, Hemoglobin A1c, and serum insulin, but did not effect on HOMO-IR33. It has been proposed that bioactive components of propolis such as flavonoids can stimulate glucose uptake and up-regulate the expression of insulin‐sensitive glucose transporter (GLUT) 4 in skeletal muscle. Propolis and its derivatives also can down-regulate the expression of genes involved in gluconeogenesis such as, glucose‐6‐phosphatase enzyme, reduce gut glucose absorption, elevate glucose utilization by liver's cells and increase cellular insulin sensitivity45,53,54. It seems that in our study using low dose of propolis resulted in non-significant changes on glycemic indices. The biological activities of propolis are related to its chemical component. Reports have shown that agents such as, bee species, plant origin, region of collection and climate affect the chemical composition of propolis. Generally, the amounts of chemical composition of collected propolis from different parts of the world are various55. According to the Bankova classification, propolis is divided into Brazilian, Canarian, Chinese, poplar Egyptian and pacific types. Studies have shown that identified propolis in the temperate region has more caffeic acid phenethyl ester (CAPE)56. However, the major bioactive component of propolis collected in the tropical region is prenylated phenylpropanoids and diterpenes such as, Brazilian green propolis57. European propolis has more mount of polyphenolic component than Brazilian propolis. Furthermore, it has been proposed that different species of bees can influence on bioactive components of propolis. Several reports have indicated that propolis is produced by stingless bees and *Apis mellifera* from tropical countries has similar composition58. Different species of *Apis mellifera* impact on chemical component of propolis59. Studies have reported that different components of propolis lead to cause various pharmacology activities60. In our study, propolis sample were collected from honey bee (Apis mellifera) colonies located in Rasht, a region in the north of Iran, in the summer season. Each propolis tablet contains 90 mg gallic acid equivalent and 67 mg flavonoids. While, other studies included in the discussion section investigated effect of different kinds of propolis collected from different regions on fasting blood glucose (FBG), insulin levels. Thus, it seems that in addition to difference in doses, difference in bioactive component of propolis may play a role.
In our study, no significant effect was found on lipid profile after taking propolis in subjects with MeS. Our finding is in agreement with the results of two meta-analyses that indicated propolis supplementation had no effects on lipid indices33,61. Also, another study reported that propolis supplementation for four months did not improve components of lipid profile among patients with NAFLD34. In contrast, a previous study observed a significant increase in HDL-C after 1000 mg/day of raw propolis supplementation for 90 days43. Another study, conducted by Samadi et al. found that propolis supplementation could reduce serum levels of LDL-C and TC among diabetic patients24. A meta-analyses with five studies showed that propolis could reduce TG level and increase HDL-C level62. In this regard, some potential pathways have been suggested for the effects of propolis in modulating blood lipid. Propolis can reduce cholesterol accumulation in the macrophage through up-regulation of PPAR gamma, and liver X receptor. Also, propolis can cause reducing in the activity of HMG-COA reductase protein and increase the expression of ATP-binding cassette transporters (ABC) A1 and G1 genes in hepatocytes which related to cholesterol metabolism63,64. In addition, the blood TG lowering effect of propolis can be attributed to insulin-mediated lipoprotein lipase activity65.
In this study, we observed no difference in the concentration of serum CRP between two groups after propolis intake, which is accordance with the data from studies performed by Fukuda et al.52 and Mujica et al.45. However, two meta-analyses33,66 and one systematic review67 indicated that propolis supplementation improves inflammation status. The possible reasons for the effect of propolis on inflammation status might due to the difference in amount of used propolis and duration of intervention. It has been suggested that propolis can reduce the production of pro-inflammatory cytokines by inhibiting the expression of nuclear factor-kappa (NF-κB), Jun N-terminal Kinase (JNK), and cyclooxygenase 2. NF-κB is an essential transcription factor that involved in the expression of inflammatory gene. It has been shown that caffeic acid phenethyl ester (CAPE) derived from propolis reduces the inflammation process through the inhabitation of NF-κB activation and the degradation of NF-κB inhibitor68,69.
Some animal studies have shown a significant reduction in blood pressure following consumption of propolis through decreasing the tyrosine hydroxylase activity which involved in biosynthesis of catecholamine. The presence of antioxidant components in propolis might play a role in vasorelaxation through down-regulation of nicotinamide adenine dinucleotide phosphate oxidase (NOX) and increase nitric oxide synthase (NOS) activity70. Another mechanism has been shown that propolis reduces blood pressure through suppress Na + reabsorption in renal tubules by the reduction of insulin level71. The result of a study indicated that intake 1000 mg/d of propolis for two months could be beneficial on blood pressure among healthy volunteers with normotensive72. Nevertheless, our study reported that there was a significant improvement in systolic blood pressure in the propolis group, but no significant difference was shown in systolic blood pressure and diastolic blood pressure between two groups. In similar to, the data from a study conducted by Silveira et al. reported that propolis extract supplement at a dose of 500 mg/day with antihypertensive drugs for 12 months did not effect on blood pressure among patients with chronic kidney disease73. It seems that difference in dosage of used propolis and the study population may be reasons of our finding. In the study conducted by Silveira et al., participants were with chronic kidney disease and under treatment with antihypertensive drugs.
We observed that the mean score of anxiety reduced significantly in the intervention group. However, there was no significant difference in three subscales of DASS-21 including, stress, anxiety and depression between the two groups. In six weeks randomized trial, Miryan et al. found that daily supplementation of 900 mg propolis improved anxiety and quality of life among irritable bowel syndrome patients44. Some in vivo and in vitro studies have reported that antioxidant agents such as, terpenoids, aromatic and aliphatic components identified in essential oil of propolis have neuroprotective effects and play an important role in the improvement of cognitive functions74. It has been shown that hyperactivity of hypothalamic–pituitary–adrenal (HPA) in brain can lead to increase production of plasma cortisol and adrenocorticotropic hormone concentrations that influence on body physiological processes, as well as depression, stress and anxiety75. Aromatic carboxylic acids and terpenoids contain in propolis essential oil can lead to an improvement in anxiety behavior through increasing the activity of superoxide dismutase (SOD) enzyme, inhibiting the activity of lipid HPA axis in brain tissue76. Also, it has been suggested that propolis has antidepressant activity, which might be done through the modulation of HPA axis and increasing the expression of hippocampal glucocorticoid receptor (GR) in animal models77. According to the above statements, we expected that propolis supplement improve anxiety, stress and depression in the subjects with MeS. However, no significant association between propolis and mood was observed. So, it seems that higher dosages of propolis and longer durations of study may be needed to observe the significant association. It is important to note that our study was conducted during the coronavirus disease (COVID-19) pandemic that lead to cause a stressful condition in the general population. It has been reported that the prevalence of COVID-19 has adverse effects on mental health due to change in life style of most individuals. Most populations in the world experienced psychological problems including, stress, depression and anxiety during COVID-19 pandemic. In the present study, quality of life was evaluated in components of mental and physical health. We observed that propolis could improve physical health and as well as, the overall score of quality of life. This is consistent with the data from studies performed by Miryan et al. and Presicce et al. In the study of Presicce et al., Boswellia resin extract and propolis derived polyphenols consumption ameliorate quality of life in patients with diabetic mellitus after 90 days78. Evidence has been indicated that high reactive oxygen species levels in skeletal muscle reduce muscles force, elevate fatigue and disrupt the cellular functions. This condition can cause reduce the ability of body and eventually has a negative effect on quality of life. It has been shown that polyphenols components and CAPE identified in propolis reduce damage to muscle and improve physical performance through, inhibiting NF-κB signaling pathway and increasing the activity of antioxidant enzymes79.
Some limitations should be noted in our study. The effect of propolis alone could not be investigated, due to ethical issues. The main limitation of the current study is that it was conducted during the Covid-19 pandemic. Covid-19 has caused a stressful condition upon the major of individuals which has negative effects on different aspects of life. Probably, this tissue can distort our results. The strength points of our study included the following: firstly, the present study is a parallel randomized double-blind clinical trial that has been performed for the first time among subjects with MeS. Third, the control and intervention groups were matched in terms of baseline values.
## Conclusion
In total, the current study indicated that propolis as a natural, safe and available supplement can be effective in decreasing WC and improving physical health and quality of life, while had no significant effects on other components of MeS among subjects with this syndrome. So, further studies with higher dosages of propolis will be needed to explain the exact mechanisms theses favorable effects and draw a clear conclusion the association between propolis with components of MeS in subjects with MeS.
## Supplementary Information
Supplementary Information. The online version contains supplementary material available at 10.1038/s41598-023-31254-y.
## References
1. Reaven GM. **Role of insulin resistance in human disease**. *Diabetes* (1988.0) **37** 1595-1607. DOI: 10.2337/diab.37.12.1595
2. Lorenzo C, Williams K, Hunt KJ, Haffner SM. **The National Cholesterol Education Program-Adult Treatment Panel III, International Diabetes Federation, and World Health Organization definitions of the metabolic syndrome as predictors of incident cardiovascular disease and diabetes**. *Diabetes Care* (2007.0) **30** 8-13. DOI: 10.2337/dc06-1414
3. Engin A. **The definition and prevalence of obesity and metabolic syndrome**. *Obes. Lipotoxicity.* (2017.0) **1** 1-17
4. Cowey S, Hardy RW. **The metabolic syndrome: A high-risk state for cancer?**. *Am J Pathol.* (2006.0) **169** 1505-1522. DOI: 10.2353/ajpath.2006.051090
5. Shin J, Lee J, Lim S, Ha H, Kwon H, Park Y. **Metabolic syndrome as a predictor of type 2 diabetes, and its clinical interpretations and usefulness**. *J. Diabetes Investig.* (2013.0) **4** 334-343. DOI: 10.1111/jdi.12075
6. Romeo GR, Lee J, Shoelson SE. **Metabolic syndrome, insulin resistance, and roles of inflammation–mechanisms and therapeutic targets**. *Arterioscler. Thromb. Vasc. Biol.* (2012.0) **32** 1771-1776. DOI: 10.1161/ATVBAHA.111.241869
7. De Silva ST, Niriella MA, Ediriweera DS, Kottahachchi D, Kasturiratne A, de Silva AP. **Incidence and risk factors for metabolic syndrome among urban, adult Sri Lankans: A prospective, 7-year community cohort, follow-up study**. *Diabetol. Metab. Syndr.* (2019.0) **11** 1-7. DOI: 10.1186/s13098-019-0461-7
8. Tabatabaei-Malazy O, Moghaddam SS, Rezaei N, Sheidaei A, Hajipour MJ, Mahmoudi N. **A nationwide study of metabolic syndrome prevalence in Iran; A comparative analysis of six definitions**. *PLoS ONE* (2021.0) **16** 1. DOI: 10.1371/journal.pone.0241926
9. Grundy SM, Hansen B, Smith SC, Cleeman JI, Kahn RA, Participants C. **Clinical management of metabolic syndrome: report of the American Heart Association/National Heart, Lung, and Blood Institute/American Diabetes Association conference on scientific issues related to management**. *Circulation* (2004.0) **109** 551-556. DOI: 10.1161/01.CIR.0000112379.88385.67
10. Ehteshami M, Shakerhosseini R, Sedaghat F, Hedayati M, Eini-Zinab H, Hekmatdoost A. **The effect of gluten free diet on components of metabolic syndrome: A randomized clinical trial**. *Asian Pac. J. Cancer Prev. APJCP.* (2018.0) **19** 2979. PMID: 30362336
11. Yki-Järvinen H. **Non-alcoholic fatty liver disease as a cause and a consequence of metabolic syndrome**. *Lancet Diabetes Endocrinol.* (2014.0) **2** 901-910. DOI: 10.1016/S2213-8587(14)70032-4
12. Casas-Agustench P, Serra M, Pérez-Heras A, Cofán M, Pintó X, Trautwein EA. **Effects of plant sterol esters in skimmed milk and vegetable-fat-enriched milk on serum lipids and non-cholesterol sterols in hypercholesterolaemic subjects: a randomised, placebo-controlled, crossover study**. *Br. J. Nutr.* (2012.0) **107** 1766-1775. DOI: 10.1017/S0007114511005162
13. Zullkiflee N, Taha H, Usman A. **Propolis: Its role and efficacy in human health and diseases**. *Molecules* (2022.0) **27** 6120. DOI: 10.3390/molecules27186120
14. Bankova VS, de Castro SL, Marcucci MC. **Propolis: Recent advances in chemistry and plant origin**. *Apidologie* (2000.0) **31** 3-15. DOI: 10.1051/apido:2000102
15. Abdullah NA, Zullkiflee N, Zaini SNZ, Taha H, Hashim F, Usman A. **Phytochemicals, mineral contents, antioxidants, and antimicrobial activities of propolis produced by Brunei stingless bees Geniotrigona thoracica, Heterotrigona itama, and Tetrigona binghami**. *Saudi J. Biol. Sci.* (2020.0) **27** 2902-2911. DOI: 10.1016/j.sjbs.2020.09.014
16. Salatino A, Salatino MLF, Negri G. **How diverse is the chemistry and plant origin of Brazilian propolis?**. *Apidologie* (2021.0) **1** 1-23
17. Anjum SI, Ullah A, Khan KA, Attaullah M, Khan H, Ali H. **Composition and functional properties of propolis (bee glue): A review**. *Saudi J Biol Sci [Internet].* (2019.0) **26** 1695-1703. DOI: 10.1016/j.sjbs.2018.08.013
18. Sforcin JM. **Biological properties and therapeutic applications of propolis**. *Phyther Res.* (2016.0) **30** 894-905. DOI: 10.1002/ptr.5605
19. Chan GC-F, Cheung K-W, Sze DM-Y. **The immunomodulatory and anticancer properties of propolis**. *Clin. Rev. Allergy Immunol.* (2013.0) **44** 262-273. DOI: 10.1007/s12016-012-8322-2
20. Aga H, Shibuya T, Sugimoto T, Kurimoto M, Nakajima S. **Isolation and identification of antimicrobial compounds in Brazilian propolis**. *Biosci. Biotechnol. Biochem.* (1994.0) **58** 945-946. DOI: 10.1271/bbb.58.945
21. Krol W, Czuba Z, Scheller S, Gabrys J, Grabiec S, Shani J. **Anti-oxidant property of ethanolic extract of propolis (EEP) as evaluated by inhibiting the chemiluminescence oxidation of luminol**. *Biochem. Int.* (1990.0) **21** 593-597. PMID: 2241984
22. Franchin M, Freires IA, Lazarini JG, Nani BD, da Cunha MG, Colón DF. **The use of Brazilian propolis for discovery and development of novel anti-inflammatory drugs**. *Eur. J. Med. Chem.* (2018.0) **153** 49-55. DOI: 10.1016/j.ejmech.2017.06.050
23. Gholaminejad F, Javadi M, Karami AA, Alizadeh F, Kavianpour M, Khadem Haghighian H. **Propolis supplementation effects on semen parameters, oxidative stress, inflammatory biomarkers and reproductive hormones in infertile men with asthenozoospermia; A randomized clinical trial**. *Int. J. Med. Lab. [Internet].* (2019.0) **6** 21-32
24. Samadi N, Mozaffari-Khosravi H, Rahmanian M, Askarishahi M. **Effects of bee propolis supplementation on glycemic control, lipid profile and insulin resistance indices in patients with type 2 diabetes: a randomized, double-blind clinical trial**. *J. Integr. Med.* (2017.0) **15** 124-134. DOI: 10.1016/S2095-4964(17)60315-7
25. Afsharpour F, Javadi M, Hashemipour S, Koushan Y. **Propolis supplementation improves glycemic and antioxidant status in patients with type 2 diabetes: A randomized, double-blind, placebo-controlled study**. *Complem. Therap. Med.* (2019.0) **43** 283-288. DOI: 10.1016/j.ctim.2019.03.001
26. El-Sharkawy HM, Anees MM, Van Dyke TE. **Propolis improves periodontal status and glycemic control in patients with type 2 diabetes mellitus and chronic periodontitis: A randomized clinical trial**. *J. Periodontol.* (2016.0) **87** 1418-1426. DOI: 10.1902/jop.2016.150694
27. Aoi W, Hosogi S, Niisato N, Yokoyama N, Hayata H. **Biochemical and Biophysical Research Communications Improvement of insulin resistance, blood pressure and interstitial pH in early developmental stage of insulin resistance in OLETF rats by intake of propolis extracts**. *Biochem. Biophys. Res. Commun. [Internet].* (2013.0) **432** 650-653. DOI: 10.1016/j.bbrc.2013.02.029
28. Zhao L, Pu L, Wei J, Li J, Wu J, Xin Z. **Brazilian green propolis improves antioxidant function in patients with type 2 diabetes mellitus**. *Int. J. Environ. Res. Public Health.* (2016.0) **13** 1. DOI: 10.3390/ijerph13050498
29. Koya-Miyata S, Arai N, Mizote A, Taniguchi Y, Ushio S, Iwaki K. **Propolis prevents diet-induced hyperlipidemia and mitigates weight gain in diet-induced obesity in mice**. *Biol. Pharm. Bull. [Internet].* (2009.0) **32** 2022-2028. DOI: 10.1248/bpb.32.2022
30. Soleimani D, Miryan M, Hadi V, Gholizadeh Navashenaq J, Moludi J, Sayedi SM. **Effect of propolis supplementation on athletic performance, body composition, inflammation, and oxidative stress following intense exercise: A triple-blind randomized clinical trial**. *Food Sci. Nutr. [Internet].* (2021.0) **9** 3631-3640. DOI: 10.1002/fsn3.2319
31. Karimian J, Hadi A, Pourmasoumi M, Najafgholizadeh A, Ghavami A. **The efficacy of propolis on markers of glycemic control in adults with type 2 diabetes mellitus: A systematic review and meta-analysis**. *Phyther Res.* (2019.0) **33** 1616-1626. DOI: 10.1002/ptr.6356
32. Zakerkish M, Jenabi M, Zaeemzadeh N, Hemmati AA, Neisi N. **The effect of iranian propolis on glucose metabolism, lipid profile, insulin resistance, renal function and inflammatory biomarkers in patients with type 2 diabetes mellitus: A randomized double-blind clinical trial**. *Sci. Rep. [Internet].* (2019.0) **9** 7289. DOI: 10.1038/s41598-019-43838-8
33. Hallajzadeh J, Milajerdi A, Amirani E, Attari VE, Maghsoudi H, Mirhashemi SM. **Effects of propolis supplementation on glycemic status, lipid profiles, inflammation and oxidative stress, liver enzymes, and body weight: A systematic review and meta-analysis of randomized controlled clinical trials**. *J. Diabetes Metab. Disord. [Internet].* (2021.0) **20** 831-843. DOI: 10.1007/s40200-020-00696-w
34. Soleimani D, Rezaie M, Rajabzadeh F, Gholizadeh Navashenaq J, Abbaspour M, Miryan M. **Protective effects of propolis on hepatic steatosis and fibrosis among patients with nonalcoholic fatty liver disease (NAFLD) evaluated by real-time two-dimensional shear wave elastography: A randomized clinical trial**. *Phyther. Res. [Internet].* (2021.0) **35** 1669-1679. DOI: 10.1002/ptr.6937
35. Grundy SM, Brewer HB, Cleeman JI, Smith SC, Lenfant C. **Definition of metabolic syndrome**. *Circulation [Internet]* (2004.0) **109** 433-438. DOI: 10.1161/01.CIR.0000111245.75752.C6
36. Bankova V. **Chemical diversity of propolis and the problem of standardization**. *J. Ethnopharmacol.* (2005.0) **100** 114-117. DOI: 10.1016/j.jep.2005.05.004
37. de Salgado ALF, de Carvalho L, Oliveira AC, dos Santos VN, Vieira JG, Parise ER. **Insulin resistance index (HOMA-IR) in the differentiation of patients with non-alcoholic fatty liver disease and healthy individuals**. *Arq. Gastroenterol.* (2010.0) **47** 165-169. DOI: 10.1590/S0004-28032010000200009
38. Ware JE. **Improvements in short-form measures of health status: introduction to a series**. *J. Clin. Epidemiol.* (2008.0) **61** 1-5. DOI: 10.1016/j.jclinepi.2007.08.008
39. Wise FM, Harris DW, Olver JH. **The DASS-14: Improving the construct validity and reliability of the depression, anxiety, and stress scale in a cohort of health professionals**. *J. All. Health.* (2017.0) **46** e85-90
40. Parkitny L, McAuley J. **The depression anxiety stress scale (DASS)**. *J. Physiother.* (2010.0) **56** 204. DOI: 10.1016/S1836-9553(10)70030-8
41. Lovibond PF, Lovibond SH. **The structure of negative emotional states: Comparison of the Depression Anxiety Stress Scales (DASS) with the Beck Depression and Anxiety Inventories**. *Behav. Res. Ther.* (1995.0) **33** 335-343. DOI: 10.1016/0005-7967(94)00075-U
42. Brown TA, Chorpita BF, Korotitsch W, Barlow DH. **Psychometric properties of the Depression Anxiety Stress Scales (DASS) in clinical samples**. *Behav. Res. Ther.* (1997.0) **35** 79-89. DOI: 10.1016/S0005-7967(96)00068-X
43. Miryan M, Soleimani D, Alavinejad P, Abbaspour M, Ostadrahimi A. **Effects of propolis supplementation on irritable bowel syndrome with constipation (IBS-C) and mixed (IBS-M) stool pattern: A randomized, double-blind clinical trial**. *Food Sci. Nutr. [Internet].* (2022.0) **10** 1899-1907. DOI: 10.1002/fsn3.2806
44. Mujica V, Orrego R, Pérez J, Romero P, Ovalle P, Zúñiga-Hernández J. **The role of propolis in oxidative stress and lipid metabolism: A randomized controlled trial**. *Evid.-Based Complement Altern. Med.* (2017.0) **2017** 1. DOI: 10.1155/2017/4272940
45. Hesami S, Hashemipour S, Shiri-Shahsavar MR, Koushan Y, Khadem HH. **Administration of Iranian Propolis attenuates oxidative stress and blood glucose in type II diabetic patients: A randomized, double-blind, placebo-controlled, clinical trial**. *Casp J. Intern Med.* (2019.0) **10** 48-54
46. Alassaf FA, Jasim MHM, Alfahad M, Qazzaz ME, Abed MN, Thanoon IA-J. **Effects of bee propolis on FBG, HbA1c, and insulin resistance in healthy volunteers**. *Turk. J. Pharm. Sci. [Internet].* (2021.0) **18** 405-409. DOI: 10.4274/tjps.galenos.2020.50024
47. Horton JD, Goldstein JL, Brown MS. **SREBPs: Activators of the complete program of cholesterol and fatty acid synthesis in the liver**. *J. Clin. Invest.* (2002.0) **109** 1125-1131. DOI: 10.1172/JCI0215593
48. Shimano H. **Sterol regulatory element-binding proteins (SREBPs): Transcriptional regulators of lipid synthetic genes**. *Prog Lipid Res.* (2001.0) **40** 439-452. DOI: 10.1016/S0163-7827(01)00010-8
49. Lindén D, Alsterholm M, Wennbo H, Oscarsson J. **PPARα deficiency increases secretion and serum levels of apolipoprotein B-containing lipoproteins**. *J. Lipid. Res.* (2001.0) **42** 1831-1840. DOI: 10.1016/S0022-2275(20)31509-1
50. Gao W, Pu L, Wei J, Yao Z, Wang Y, Shi T. **Serum antioxidant parameters are significantly increased in patients with type 2 diabetes mellitus after consumption of chinese propolis: A randomized controlled trial based on fasting serum glucose level**. *Diabetes Therapy* (2018.0) **9** 101-111. DOI: 10.1007/s13300-017-0341-9
51. Fukuda T, Fukui M, Tanaka M, Senmaru T, Iwase H, Yamazaki M. **Effect of Brazilian green propolis in patients with type 2 diabetes: A double-blind randomized placebo-controlled study**. *Biomed. Rep. [Internet].* (2015.0) **3** 355-360. DOI: 10.3892/br.2015.436
52. Ueda M, Hayashibara K, Ashida H. **Propolis extract promotes translocation of glucose transporter 4 and glucose uptake through both PI3K-and AMPK-dependent pathways in skeletal muscle**. *BioFactors* (2013.0) **39** 457-466. DOI: 10.1002/biof.1085
53. Al-Hariri M, Eldin TG, Abu-Hozaifa B, Elnour A. **Glycemic control and anti-osteopathic effect of propolis in diabetic rats**. *Diabetes Metab. Syndr. Obes. Targets Ther.* (2011.0) **4** 377. DOI: 10.2147/DMSO.S24159
54. Isla MI, Paredes-Guzman JF, Nieva-Moreno MI, Koo H, Park YK. **Some chemical composition and biological activity of northern Argentine propolis**. *J. Agric. Food Chem.* (2005.0) **53** 1166-1172. DOI: 10.1021/jf040130h
55. Samadi N, Mozaffari-Khosravi H, Rahmanian M, Askarishahi M. **Effects of bee propolis supplementation on glycemic control, lipid profile and insulin resistance indices in patients with type 2 diabetes: A randomized, double-blind clinical trial**. *J. Integr. Med. [Internet].* (2017.0) **15** 124-134. DOI: 10.1016/S2095-4964(17)60315-7
56. Fernandes-Silva CC, Freitas JC, Salatino A, Salatino MLF. **Cytotoxic activity of six samples of Brazilian propolis on Sea Urchin (Lytechinus variegatus) Eggs**. *Evid.-Based Complement Altern. Med.* (2013.0) **2013** 1. DOI: 10.1155/2013/619361
57. Salatino A, de Pereira LR, Salatino MLF. **The emerging market of propolis of stingless bees in tropical countries**. *MOJ Food Process Technol.* (2019.0) **7** 27-29
58. Arias MC, Sheppard WS. **Phylogenetic relationships of honey bees (Hymenoptera: Apinae: Apini) inferred from nuclear and mitochondrial DNA sequence data**. *Mol. Phylogenet. Evol.* (2005.0) **37** 25-35. DOI: 10.1016/j.ympev.2005.02.017
59. 59.Fabris, S., Bertelle, M., Astafyeva, O., Gregoris, E., Zangrando, R., Gambaro, A., et al. Antioxidant properties and chemical composition relationship of Europeans and Brazilians propolis (2013).
60. Gheflati A, Dehnavi Z, Yazdi AG, Khorasanchi Z, Raeisi-Dehkordi H, Ranjbar G. **The effects of propolis supplementation on metabolic parameters: A systematic review and meta-analysis of randomized controlled clinical trials**. *Avicenna J. Phytomed.* (2021.0) **11** 551-565. PMID: 34804893
61. Salehi-Sahlabadi A, Chhabra M, Rahmani J, Momeni A, Karam G, Nattagh-Eshtivani E. **The effect of propolis on anthropometric indices and lipid profile: A systematic review and meta-analysis of randomized controlled trials**. *J. Diabetes Metab Disord.* (2020.0) **19** 1835-1843. DOI: 10.1007/s40200-020-00604-2
62. Iio A, Ohguchi K, Maruyama H, Tazawa S, Araki Y, Ichihara K. **Ethanolic extracts of Brazilian red propolis increase ABCA1 expression and promote cholesterol efflux from THP-1 macrophages**. *Phytomedicine* (2012.0) **19** 383-388. DOI: 10.1016/j.phymed.2011.10.007
63. Yu Y, Si Y, Song G, Luo T, Wang J, Qin S. **Ethanolic extract of propolis promotes reverse cholesterol transport and the expression of ATP-binding cassette transporter A1 and G1 in mice**. *Lipids* (2011.0) **46** 805-811. DOI: 10.1007/s11745-011-3568-7
64. Li Y, Chen M, Xuan H, Hu F. **Effects of encapsulated propolis on blood glycemic control, lipid metabolism, and insulin resistance in type 2 diabetes mellitus rats**. *Evid.-Based Complement Altern. Med [Internet].* (2012.0) **2012** 1-8. DOI: 10.1155/2012/793841
65. Jalali M, Ranjbar T, Mosallanezhad Z, Mahmoodi M, Moosavian SP, Ferns GA. **Effect of propolis intake on serum C-reactive protein (CRP) and tumor necrosis factor-alpha (TNF-α) levels in adults: A systematic review and meta-analysis of clinical trials**. *Complement. Therapies Med.* (2020.0) **50** 1
66. Soleimani D, Miryan M, Tutunchi H, Navashenaq JG, Sadeghi E, Ghayour-Mobarhan M. **A systematic review of preclinical studies on the efficacy of propolis for the treatment of inflammatory bowel disease**. *Phytother. Res.* (2021.0) **35** 701-710. DOI: 10.1002/ptr.6856
67. Silva-Carvalho R, Baltazar F, Almeida-Aguiar C. **Propolis: A complex natural product with a plethora of biological activities that can be explored for drug development**. *Evid.-Based Complement Altern. Med.* (2015.0) **2015** 1. DOI: 10.1155/2015/206439
68. Mirzoeva OK, Calder PC. **The effect of propolis and its components on eicosanoid production during the inflammatory response**. *Prostaglandins Leukot. Essent. Fat Acids.* (1996.0) **55** 441-449. DOI: 10.1016/S0952-3278(96)90129-5
69. Silva V, Genta G, Möller MN, Masner M, Thomson L, Romero N. **Antioxidant activity of Uruguayan propolis. In vitro and cellular assays**. *J. Agric. Food Chem.* (2011.0) **59** 6430-6437. DOI: 10.1021/jf201032y
70. Blazer-Yost BL, Esterman MA, Vlahos CJ. **Insulin-stimulated trafficking of ENaC in renal cells requires PI 3-kinase activity**. *Am J Physiol Physiol.* (2003.0) **284** C1645-C1653. DOI: 10.1152/ajpcell.00372.2002
71. Khalaf DA, Thanoon IA. **Effects of bee propolis on blood pressure record and certain biochemical parameter in healthy volunteers**. *Ann. Coll. Med. Mosul.* (2018.0) **40** 20-26. DOI: 10.33899/mmed.2018.159191
72. Silveira MAD, Teles F, Berretta AA, Sanches TR, Rodrigues CE, Seguro AC. **Effects of Brazilian green propolis on proteinuria and renal function in patients with chronic kidney disease: A randomized, double-blind, placebo-controlled trial**. *BMC Nephrol. [Internet].* (2019.0) **20** 140. DOI: 10.1186/s12882-019-1337-7
73. Shimazawa M, Chikamatsu S, Morimoto N, Mishima S, Nagai H, Hara H. **Neuroprotection by Brazilian green propolis against in vitro and in vivo ischemic neuronal damage**. *Evid.-Based Complement Altern. Med.* (2005.0) **2** 201-207. DOI: 10.1093/ecam/neh078
74. Saiyudthong S, Marsden CA. **Acute effects of bergamot oil on anxiety-related behaviour and corticosterone level in rats**. *Phyther. Res.* (2011.0) **25** 858-862. DOI: 10.1002/ptr.3325
75. Li YJ, Xuan HZ, Shou QY, Zhan ZG, Lu X, Hu F-L. **Therapeutic effects of propolis essential oil on anxiety of restraint-stressed mice**. *Hum. Exp. Toxicol.* (2012.0) **31** 157-165. DOI: 10.1177/0960327111412805
76. Lee M-S, Kim YH, Park W-S, Ahn WG, Park OK, Kwon S-H. **Novel antidepressant-like activity of propolis extract mediated by enhanced glucocorticoid receptor function in the hippocampus**. *Evid.-Based Complement Altern. Med.* (2013.0) **2013** 1
77. Sibona M, Destefanis P, Agnello M, Lillaz B, Giuliano M, Cai T. **The association of Boswellia resin extract and propolis derived polyphenols can improve quality of life in patients affected by prostatitis-like symptoms**. *Arch. Ital di Urol e Androl. [Internet].* (2020.0) **91** 251-255. DOI: 10.4081/aiua.2019.4.251
78. Shen Y-C, Yen J-C, Liou K-T. **Ameliorative effects of caffeic acid phenethyl ester on an eccentric exercise-induced skeletal muscle injury by down-regulating NF-κb mediated inflammation**. *Pharmacology* (2013.0) **91** 219-228. DOI: 10.1159/000348412
79. Powers SK, Jackson MJ. **Exercise-induced oxidative stress: Cellular mechanisms and impact on muscle force production**. *Physiol Rev.* (2008.0) **88** 1243-1276. DOI: 10.1152/physrev.00031.2007
|
---
title: 'Severity and geographical disparities of post-COVID-19 symptoms among the
Vietnamese general population: a national evaluation'
authors:
- Bach Tran
- Minh Ngoc Le Vu
- Huong Thi Le
- Tu Huu Nguyen
- Laurent Boyer
- Guillaume Fond
- Pascal Auquier
- Carl A. Latkin
- Roger C. M. Ho
- Cyrus S. H. Ho
- Melvyn W. B. Zhang
journal: Scientific Reports
year: 2023
pmcid: PMC10022561
doi: 10.1038/s41598-023-30790-x
license: CC BY 4.0
---
# Severity and geographical disparities of post-COVID-19 symptoms among the Vietnamese general population: a national evaluation
## Abstract
Post-COVID-19 symptoms have become a significant global health concern. This study focused on assessing the prevalence, severity, and care preference of post-COVID-19 symptoms, as well as identifying determinants to inform evidence-based policy on post-COVID-19 in Vietnam. A national cross-sectional study was conducted in May 2022 among 12,361 recovered COVID-19 patients, providing the largest dataset on health status after COVID-19 in Vietnam. The study utilized ordered logistic, Poisson regression, Multilevel linear random-effects models, and Multilevel random effects ordered logistic model to identify factors associated with various aspects of post-COVID-19 conditions. Results showed that the average number of post-COVID-19 symptoms was approximately 3, with fatigue and headache being the most common symptoms. The number of post-COVID-19 symptoms varied by province, decreased with age, and was significantly correlated with the duration of infection. Age, infection period, underlying conditions, telehealth utilization, and geographical location were identified as significant determinants of post-COVID-19 symptoms. The study concluded that improving resource allocation and health-seeking behavior in underserved areas could help address differences in health outcomes and improve post-COVID-19 control in Vietnam.
## Introduction
Following nearly three years of unprecedented global spread, the COVID-19 pandemic has now entered its post-peak period. At the start of June, global data reported 225,120 new cases daily and a $99\%$ recovery rate, compared to the peak in November 2021, which saw 514,914 new cases daily and a $90.14\%$ recovery rate1,2. In Vietnam, due to rapid response strategies and a comprehensive national vaccination program, the country has been able to considerably control the course of the pandemic3,4. As of June 2022, Vietnam recorded an average of only 910 new cases daily, and there were four deaths in six consecutive weeks5. However, the slowing of the spread does not signify the end of the pandemic but rather the beginning of a new era in the fight against COVID-19. The aftermath of the virus includes economic losses, environmental degradation, increased social inequality, disruption to lifestyles, and, most significantly, long-term or even permanent health consequences6–8.
The World Health Organization (WHO) has defined post-COVID-19 condition as a collection of long-term symptoms that can persist after being infected by COVID-19, either from the initial illness or after recovery. These symptoms may be less severe than those experienced during the initial infection, but they can affect multiple organs simultaneously, including the respiratory, cardiovascular, gastrointestinal, and neurological systems9–11. Research suggests that approximately $68\%$ of COVID-19 survivors experience post-COVID-19 symptoms, and the duration of these symptoms can range from days to several months or even years12. However, investigating post-COVID-19 symptoms is challenging because of the lack of a standardized protocol, confirmed risk factors, and endpoints, which limits the body of evidence on this condition14,15. Despite several proposed guidelines for the diagnosis and management of post-COVID-19 symptoms, interventions remain in the early stages, with emergency recommendations being more common than evidence-based approaches. Moreover, high operational costs and exhausted workforces after a prolonged period of overwork make these interventions hardly feasible in the long term16–19. Existing frameworks for post-COVID-19 intervention also have limitations, with a lack of contextual domains for resource-scarce regions in low- and middle-income countries being a common barrier.
The Ministry of Health in Vietnam has identified more than 200 post-COVID-19 symptoms that have an extended impact on national health, safety, and socioeconomic growth20. Nevertheless, there is still a dearth of comprehensive data on the prevalence and impact of post-COVID-19 symptoms on the Vietnamese population, which impedes the creation of a national strategy and exacerbates the health disparity between regions in Vietnam. Our study represents one of the largest and earliest efforts to investigate the residual effects of COVID-19 on the Vietnamese population. Our objective was to determine the frequency and severity of post-COVID-19 symptoms in Vietnam, and to analyze the patterns and factors that contribute to a post-COVID-19 symptom timeline. Our findings will not only add to the growing corpus of research on post-COVID-19 symptoms but also facilitate the development of evidence-based interventions in Vietnam in the near future.
## Results
Table 1 presented the individual characteristics of the participants. The majority of respondents were male ($65.1\%$), and the median age was 18 (IQR = 17–27). More than one-third of respondents resided in Northern Midlands and Mountains ($34.7\%$), and $22.1\%$ resided in the Mekong Delta regions. Only $4.9\%$ of respondents were smokers, while $19.7\%$ consumed alcohol. A large proportion of respondents ($70.8\%$) exercised after recovery, and $63.1\%$ had a normal BMI Index. Only a minor proportion ($3.1\%$) had comorbidities. Table 1Individual characteristics of participants. Characteristicsn%Gender Male803565.1 Female430534.9Provinces Southeast region5494.8 Northern midlands and mountains398534.7 Red river delta11319.9 North central region4934.3 South central coast183916.0 Central highlands9358.2 *Mekong delta* regions254022.1 Smoking5944.9 Alcohol use241519.7 Exercising after recovering from COVID-19869370.8BMI index Underweight338228.2 Normal757863.1 Overweight/Obese10478.7 Comorbidities3783.1 Diabetes280.2 Cardiovascular disease2081.6 Chronic lung disease880.6 Neurological disease590.4 Cancer390.3MedianIQRAge (range 16–35)1817–27 Table 2 showed that more than two-thirds of respondents were infected with COVID-19 in the recent 1 to 4 months. Participants infected with COVID-19 for less than 7 days had the highest proportion ($49.8\%$), followed by 7 to 14 days ($48.1\%$). In terms of the severity of COVID-19 at the onset, respondents who had mild symptoms had the highest percentage of $79.8\%$, and $10.8\%$ of respondents were asymptomatic. The prevalence of cases in the region (per 100,000 population) and COVID-19 case fatality rate were calculated as 13,695.5 (IQR = 4437.2–17,936.5) and $0.08\%$ (IQR = $0.01\%$–$0.46\%$) in our sampled region. There were statistically significant differences between regions ($p \leq 0.001$).Table 2Characteristics of participants when infected with COVID-19 by province. CharacteristicsSoutheast regionNorthern midlands and mountainsRed river deltaNorth central regionSouth central coastCentral highlandsMekong delta regionsTotalp-valuen (%)n (%)n (%)n (%)n (%)n (%)n (%)n (%)Time since COVID-19 onset 1 month78 (14.3)704 (17.9)129 (11.4)98 [20]210 (11.5)189 (20.5)314 (12.5)1722 (15.1) < 0.001 1–4 months374 (68.4)2773 (70.4)870 (77.2)351 (71.5)1386 (75.6)674 (72.9)1590 (63.2)8018 (70.5) 4–6 months59 (10.8)159 [4]75 (6.7)20 (4.1)124 (6.8)34 (3.7)394 (15.7)865 (7.6) Above 6 months36 (6.6)305 (7.7)53 (4.7)22 (4.5)113 (6.2)27 (2.9)218 (8.7)774 (6.8)COVID-19 infection period Less than 7 days279 (50.9)2147 (54.3)541 (48.1)215 (43.8)897 (48.9)393 (42.6)1209 (47.8)5681 (49.8) < 0.001 7–14 days259 (47.3)1736 (43.9)552 (49.1)269 (54.8)897 (48.9)511 (55.4)1264 [50]5488 (48.1) More than 14 days10 (1.8)70 (1.8)32 (2.8)7 (1.4)41 (2.2)19 (2.1)54 (2.1)233 [2]Severity of COVID-19 at the onset Asymptomatic57 (10.4)408 (10.2)148 (13.1)38 (7.7)209 (11.4)90 (9.6)288 (11.3)1238 (10.8) < 0.001 Mild456 (83.1)3145 (78.9)890 (78.7)396 (80.3)1491 (81.1)741 (79.3)2030 (79.9)9149 (79.8) Moderate30 (5.5)403 (10.1)85 (7.5)52 (10.5)129 [7]99 (10.6)207 (8.1)1005 (8.8) Severe6 (1.1)29 (0.7)8 (0.7)7 (1.4)10 (0.5)5 (0.5)15 (0.6)80 (0.7)Median (IQR)Median (IQR)Median (IQR)Median (IQR)Median (IQR)Median (IQR)Median (IQR)Median (IQR)p-valuePrevalence of case in region (per 100,000 population)11,565.3(11,501.5–11,565.3)17,936.5(17,936.5–17,936.5)19,055.7(16,708.6–19,055.7)13,695.5(3862.9–13,695.5)4877.6(3851.8–8315.2)11,179.7(4437.2–11,179.7)2808.2(2808.2–7527.9)13,695.5(4437.2–17,936.5) < 0.001COVID-19 case fatality rate (%)0.67 (0.19–0.67)0.01 (0.01–0.01)0.08 (0.058–0.08)0.06 (0.06–0.10)0.31 (0.27–0.32)0.06 (0.06–0.17)1.84 (1.24–2.22)0.08 (0.01–0.46) < 0.001 From Fig. 1, fatigue was the most common symptom of participants ($37.7\%$), followed by headache ($33.1\%$). Moreover, symptoms such as difficulty thinking or concentrating, cough, dyspnea, and somnipathy had a high percentage of $30.8\%$, $30.5\%$, $29.5\%$, and $26.4\%$, respectively. Figure 1Prevalence of post-COVID-19 symptoms.
Table 3 revealed that neurological symptoms were the most frequent symptoms of post-COVID-19 across all regions, followed by respiratory and heart symptoms. *In* general, approximately one-third ($31.1\%$) of participants had four or more four symptoms, while in the red river delta region, the highest proportion ($33.4\%$) of participants were asymptomatic. The mean number of neurological symptoms, digestive symptoms, respiratory and heart symptoms, and other symptoms were 1.43 (SD = 1.45), 0.21 (SD = 0.54), 0.84 (SD = 1.04), and 0.35 (SD = 0.62), respectively. The average of post-COVID-19 symptoms was 2.82 (SD = 2.80). Residents in Northern Midlands and Mountains had the highest number of both four types of symptoms (neurological, digestive, respiratory and heat, and others), there were statistically significant differences between regions ($p \leq 0.001$).Table 3Characteristics of post-COVID-19 symptoms by province. CharacteristicsSoutheast regionNorthern midlands and mountainsRed river deltaNorth central regionSouth central coastCentral highlandsMekong delta regionsTotalp-valuen (%)n (%)n (%)n (%)n (%)n (%)n (%)n (%)Neurological symptoms373 (67.9)2673 (67.1)568 (50.2)356 (72.2)1225 (66.6)660 (70.6)1579 (62.2)7434 (64.8) < 0.001Digestive symptoms64 (11.7)800 (20.1)125 (11.1)74 (15.0)265 (14.4)162 (17.3)316 (12.4)1806 (15.7) < 0.001Respiratory and heart symptoms311 (56.6)2107 (52.9)483 (42.7)246 (49.9)899 (48.9)482 (51.6)1309 (51.5)5837 (50.9) < 0.001Other symptoms151 (27.5)1250 (31.4)248 (21.9)136 (27.6)486 (26.4)280 (29.9)632 (24.9)3183 (27.7) < 0.001Number of post-COVID-19 symptoms Asymptomatic89 (16.2)800 (20.1)378 (33.4)69 (14.0)337 (18.3)157 (16.8)537 (21.1)2367 (20.6) < 0.001 One symptom125 (22.8)748 (18.8)230 (20.3)116 (23.5)430 (23.4)185 (19.8)549 (21.6)2383 (20.8) Two symptoms86 (15.7)529 (13.3)169 (14.9)79 (16.0)303 (16.5)158 (16.9)441 (17.4)1765 (15.4) Three symptoms84 (15.3)471 (11.8)121 (10.7)74 (15.0)215 (11.7)135 (14.4)284 (11.2)1384 (12.1) Four or more than four symptoms165 (30.1)1437 (36.1)233 (20.6)155 (31.4)554 (30.1)300 (32.1)729 (28.7)3573 (31.1)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)p-valueNumber of neurological symptoms (range 0–5)1.44 (1.40)1.54 (1.49)1.04 (1.36)1.61 (1.43)1.49 (1.48)1.54 (1.41)1.30 (1.39)1.43 (1.45) < 0.001Number of digestive symptoms (0–3)0.15 (0.48)0.27 (0.60)0.15 (0.46)0.20 (0.52)0.20 (0.53)0.23 (0.56)0.16 (0.48)0.21 (0.54) < 0.001Number of respiratory and heart symptoms (range 0–4)0.88 (0.98)0.92 (1.10)0.65 (0.92)0.81 (1.05)0.81 (1.05)0.8 (0.97)0.84 (1.02)0.84 (1.04) < 0.001Number of other symptoms (range 0–3)0.33 (0.58)0.40 (0.66)0.25 (0.52)0.34 (0.62)0.33 (0.60)0.38 (0.64)0.31 (0.58)0.35 (0.62) < 0.001Number of post COVID-19 symptoms (range 0–15)2.8 (2.58)3.13 (3.01)2.09 (2.52)2.96 (2.65)2.82 (2.80)2.95 (2.74)2.61 (2.60)2.82 (2.80) < 0.001 Table 4 revealed the brief model to identify factors associated with the number of four main types of post-COVID-19 symptoms (the full model was presented in Appendix 4). Compared to male participants, the female was likely to have a lower number of neurological symptoms, a lower number of other symptoms, but they had a higher number of digestive symptoms. Compared to people living in the Southeast region, people living in the Red River Delta were likely to have a lower number of neurological and respiratory, and heart symptoms, people living in North Central Region and South Central Coast also had a lower number of respiratory and heart symptoms, people who lived in South Central Coast and Mekong Delta Region was likely to have a higher number of digestive symptoms. Smoking and Drinking alcohol were the positive factors that increased the number of other symptoms and neurological symptoms, respectively. Overweight/obese people had a lower number of neurological symptoms but had a higher number of respiratory and heart symptoms and other symptoms. Table 4Brief Multilevel linear random-effects models to identify factors associated with neurological, digestive, respiratory, heart, and other symptoms. FactorsNeurological symptomsDigestive symptomsRespiratory and heart symptomsOther symptomsCoefp-valueCoefp-valueCoefp-valueCoefp-valueSocio-economic Gender (Female vs Male -ref)−0.35 < 0.0010.06 < 0.001−0.19 < 0.001 Age (unit: age)−0.0020.005 Provinces (Southeast region -ref) Red River Delta−0.280.023−0.160.019 North Central Region−0.150.029 South Central Coast0.060.005−0.110.023 Mekong Delta Region0.070.001 Smoking (Yes vs No -ref)0.060.002 Alcohol (Yes vs No -ref)0.13 < 0.001BMI Index (vs Underweight -ref) Normal0.060.007 Overweight/Obese−0.13 < 0.0010.080.0180.030.041 Comorbidities (Yes vs No -ref)0.24 < 0.001 COVID-19 infection characteristicsCOVID-19 infection period (vs Less than 7 days -ref) 7–14 days0.12 < 0.001−0.010.0480.060.0020.030.011 More than 14 days0.200.0110.060.030.150.0030.090.001Time since COVID-19 onset (vs 1 month -ref) 1–4 months0.120.001−0.040.001−0.060.007 4–6 months0.130.023−0.040.033Severity of COVID-19 at the onset (vs asymptomatic -ref) Mild0.26 < 0.0010.17 < 0.001 Moderate0.56 < 0.0010.070.0040.40 < 0.0010.10 < 0.001 Severe0.390.001Prevalence of case in region (vs Low -ref) Medium0.030.038Post-COVID-19 symptoms Neurological symptoms––0.07 < 0.0010.25 < 0.0010.12 < 0.001 Digestive symptoms0.40 < 0.001––0.38 < 0.0010.20 < 0.001 Respiratory and heart symptoms0.43 < 0.0010.12 < 0.001––0.07 < 0.001 Other symptoms0.55 < 0.0010.16 < 0.0010.18 < 0.001–– People who were infected with COVID-19 for 7–14 days or more than 14 days were likely to have a higher number of neurological symptoms, respiratory and heart symptoms, and other symptoms, but those who were infected with COVID-19 from 7 to 14 days had a lower number of digestive symptoms than those who infected with COVID-19 less than 7 days. People who recovered from COVID-19 for 1–4 months were likely to have a higher number of neurological symptoms, have a lower number of digestive and respiratory, and heart symptoms, but those who recovered from COVID-19 from 4 to 6 months had a higher number of neurological symptoms and a lower number of other symptoms. Compare to asymptomatic COVID-19 at the onset, people had mild severity had a higher number of neurological and respiratory, and heart symptoms. Moreover, people had moderate severity had a higher number of both of the four main symptoms. An increase in any one type of symptom was likely to increase the number of other symptoms.
Table 5 presented that female participants tended to experience lower negative impacts of COVID-19 such as lower severity of COVID-19 at the onset, lower level of post-COVID-19 symptoms, and less number of post-COVID-19. These characteristics also correlated with age, as younger people had a lower level of severity of COVID-19 at the onset, and fewer post-COVID-19 symptoms. Smoking, drinking alcohol, and having comorbidities were risk factors that were highly associated with the increase in the level of post-COVID-19 symptoms and the number of post-COVID-19 symptoms. Table 5Factors associated with severity of COVID-19 at the onset and post-COVID-19 symptoms. FactorsSeverity of COVID-19 at the onset (From 1″Asymptomatic” to 4″Severe”Post COVID-19 symptom (From 0 “Asymptomatic” to 4 “4 or more than 4 symptoms”)Number of Post COVID-19 symptomsOR$95\%$CIp-valueOR$95\%$CIp-valueCoef$95\%$CIp-valueSocio-economicGender (Female vs Male -ref)0.530.47; 0.61 < 0.0010.410.39; 0.44 < 0.001−1.14−1.25; −1.03 < 0.001Age (unit: age)0.990.98; 1.000.0250.990.98; 1.000.169−0.03−0.04; −0.01 < 0.001Provinces (Southeast region -ref) Northern Midlands and Mountains1.210.68; 2.160.5190.730.51; 1.050.092−0.20−0.63; 0.230.370 Red River Delta0.550.30; 0.980.0430.340.16; 0.720.005−1.15−1.64; −0.65 < 0.001 North Central Region1.460.89; 2.400.1340.810.60; 1.080.155−0.10−0.46; 0.270.613 South Central Coast1.080.70; 1.660.7440.840.68; 1.040.1030.04−0.24; 0.330.763 Central Highlands1.280.80; 2.050.3080.780.47; 1.310.3550.02−0.35; 0.390.903 Mekong Delta Region1.410.97; 2.040.0690.840.70; 1.020.074−0.03−0.27; 0.210.814 Smoking (Yes vs No -ref)0.730.57; 0.920.0071.201.02; 1.420.0290.230.00; 0.450.046 Alcohol (Yes vs No -ref)1.211.05; 1.390.0071.431.32; 1.55 < 0.0010.440.34; 0.54 < 0.001 Exercising after recovering from COVID-19 (Yes vs No -ref)0.910.79; 1.060.2170.830.73; 0.950.008−0.15−0.34; 0.040.115BMI Index (vs Underweight -ref) Normal1.060.95; 1.180.3201.010.93; 1.090.8790.03−0.09; 0.150.671 Overweight/Obese0.820.66; 1.030.0881.050.89; 1.240.5720.03−0.13; 0.180.716 Comorbidities (Yes vs No -ref)1.951.32; 2.890.0011.921.54; 2.40 < 0.0011.040.68; 1.40 < 0.001 Infected with COVID-19 characteristicCOVID-19 infection period (vs Less than 7 days -ref) 7–14 days3.152.87; 3.45 < 0.0011.421.31; 1.53 < 0.0010.500.40; 0.60 < 0.001 More than 14 days14.0410.27; 19.19 < 0.0012.041.45 2.87 < 0.0011.370.91; 1.83 < 0.001Time since COVID-19 onset (vs 1 month -ref) 1–4 months0.900.82; 0.980.0190.990.91; 1.080.837−0.04−0.2;1 0.120.619 4–6 months0.580.50; 0.68 < 0.0011.100.89; 1.370.3590.15−0.10; 0.390.240 Above 6 months0.280.22; 0.38 < 0.0010.840.73; 0.960.013−0.19−0.42; 0.040.105Severity of COVID-19 at the onset (vs asymptomatic -ref) Mild–––2.512.18; 2.89 < 0.0011.010.76; 1.26 < 0.001 Moderate–––7.305.91; 9.02 < 0.0012.912.62; 3.19 < 0.001 Severe–––3.892.36; 6.43 < 0.0012.391.54; 3.24 < 0.001Prevalence of case in region (vs Low -ref) Medium1.230.95; 1.600.1130.940.77; 1.150.5440.10−0.05; 0.250.187 High2.291.43; 3.660.0011.480.87; 2.510.1440.600.19; 1.020.004COVID-19 case fatality rate in region (vs Low -ref) Medium0.890.58; 1.370.5960.790.49; 1.250.311−0.28−0.64 0.090.135 High0.810.48; 1.360.4200.680.41; 1.130.137−0.48−0.91; −0.050.028 *In this* study, the Southeast region was chosen as the reference factor. Southeast Region included Ho Chi Minh, Ba Ria—Vung Tau, Binh Duong, Binh Phuoc, Dong Nai, and Tay Ninh, which were heavily affected by COVID-19. Especially, Ho Chi Minh City and Binh Duong are the two epicenters of the country's outbreak of the COVID-19 pandemic with the highest number of cases and deaths21. The Vietnamese government has many policies to help reduce the damage of COVID-19 such as establishing a mobile medical station, and mobilizing a large force of health workers, the army, and the police to participate in epidemic prevention and control22. Therefore, choosing the Southeast region as a reference helps to objectively compare the severity of COVID-19 at the onset and post-COVID-19 symptoms and provides more information for readers and policymakers. Compared to participants living in the Southeast Region, people who resided in the Red River Delta had a lower level of severity of COVID-19 at the onset, lower level of post-COVID-19 symptoms, and less number of post-COVID-19 symptoms.
Longer infection time of COVID-19 and higher severity of COVID-19 at the onset were associated with an increase in both levels of post-COVID-19 symptoms and the number of post-COVID-19 symptoms. People who lived in the region with a high prevalence of case was likely to have a higher severity of COVID-19 at the onset and more number post-COVID-19 symptoms.
Appendix 1, Appendix 2, and Appendix 3 revealed post-COVID-19 symptoms and the number of post-COVID-19 symptoms by COVID-19 characteristics.
Figure 2 presents the preference of post-COVID-19 care among respondents. The most favorable means of care delivery among respondents was directly at hospitals ($36.3\%$) or through mobile apps ($33\%$). Medical consultation through telephone received the lowest preference level $12.1\%$).Figure 2Preference of post-COVID-19 care among respondents.
## Discussion
Our results suggested a high frequency and severity level of post-COVID-19 symptoms among Vietnamese and identified age, infection duration, daily health behavior, and comorbidities as factors associated with the prevalence and severity of post-COVID-19 symptoms. Multilevel implications were proposed, including clinical development as well as adjustments to health policies and government management strategy.
Correlations were found between demographic, behavioral factors, and post-infection symptoms. Our data indicated a strong association between gender and post-COVID-19 symptoms, in which females tended to experience less severe chronic COVID-19 syndromes than males. This may result from sex differences in immune response23. While some existing studies recorded similar results, others found an opposite pattern of gender severity, such as the longitudinal cohort study in Germany or a cross-sectional study of healthcare workers in Northwest England, where females experienced more severe post-COVID-19 manifestations24–26. Therefore, although findings suggest that gender may not be a significant determinant of post-COVID-19 conditions, more empirical evidence should be collected to determine this correlation. Interestingly, the number of post-COVID-19 conditions decreased with age, meaning older participants recorded fewer symptoms than young participants. Although no previous study has focused on age as a determinant, this finding may underline the need to encourage post-COVID-19 care among the younger population. Overweight and comorbidities were found to correlate with a heightened likelihood of developing post-COVID-19 symptoms in neurological and respiratory systems. Indeed, problems with nutrition levels are determinants of cardiovascular disease and cancer, the two most common comorbidities among patients with severe infection and longer post-infection syndromes in our results27–29. As these dynamics have been observed in the majority of existing literature, this study reaffirms the importance of practicing regular healthy activities as well as highlights a novel risk of alcohol and smoking30–32.
Longer infection duration and severity were correlated with substantially higher severity and several post-COVID-19 symptoms. Our findings demonstrated that infection duration of more than 14 days and moderate severity were most correlated with a higher number of symptoms as well as higher severity. While this trend is neither refuted nor contrasted in any existing studies, the strength of such association varies: from substantial in our findings, briefly discussed in Hall’s et al. to insignificant or not mentioned at all in many31,33,34. The gap in data may be attributed to disparities between the healthcare system and the socioeconomic background of each sample. A more reasonable explanation lies in the difference between assessment scales and definitions of post-infection symptoms between studies. Baig et al. proposed that symptoms prolonging beyond 3 weeks be classified as chronic COVID-19 symptoms, while another study proposed the categorization for different post-infection durations: post-acute COVID-19 for symptoms beyond 3 weeks and long post-COVID-19 for symptoms beyond 12 weeks35,36.
Our findings and previous meta-analysis worldwide pointed out that the most frequent post-infection symptoms were related to the neurological and respiratory systems, such as dizziness, headache, cough, or shortness of breath. The multivariate random effect models demonstrated an important pattern that post-COVID-19 symptoms tended to co-exist across systems. The presence of either one of three symptom groups: neurological, digestive, and respiratory, was significantly associated with the presence of the other two. To date, the approach of pharmaceutical treatments for post-COVID-19 symptoms tends to focus on each symptom at a time instead of finding a solution for the multisystem manifestation of post-COVID-19. If clinically possible, interventions for post-COVID-19 symptoms should be developed to resolve multisystem problems instead of focusing on one single symptom. Furthermore, although these manifestations can cause great inconveniences to daily functioning and result in lower quality of life for recovered patients, they are generally mild and rarely develop into severe complications37–39. However, such symptoms are also indicators of many serious, even fatal, health conditions. For example, chronic migraines suggest the possible onset of a cerebrovascular accident or a brain tumor, while constant coughs can indicate early stages of gastroesophageal reflux disease or pulmonary tuberculosis40–43. The misperception of these mild conditions as post-COVID-19 symptoms prevents individuals from having an early diagnosis and ultimately leads to poor treatment outcomes. Therefore, besides developing pharmaceutical treatments for chronic COVID-19, recovered patients should be reminded of potential misunderstandings and be informed about how to distinguish post-COVID-19 symptoms from early manifestations of other serious health conditions.
Most notably, the proportion of post-COVID-19 symptoms varies across regions and is attributable to differences in socioeconomic characteristics, health delivery, and level of adaptation to response strategy. The severity of post-COVID-19 symptoms nearly doubled in regions with a high prevalence of cases compared to those with low prevalence. In Vietnam, the most notable differences can be observed in the Red River Delta (RRD) and the Northern Midlands and Mountains (NMM). Residents in RRD recorded lower numbers and severity of post-COVID-19 conditions than all other regions. Moreover, while the largest proportion of participants in other regions had 4 or more post-COVID-19 symptoms, most participants from RRD were asymptomatic ($33.4\%$) or had one symptom ($20.3\%$). On the other hand, residents from NMM had both the highest number and highest severity of post-COVID-19 symptoms than the remaining provinces. The pattern observed in RRD and NMM can be attributed to gaps in healthcare quality and the health literacy of residents. The RRD is the region with the highest population density in Vietnam (1,450 people/km2, the population is 21,848,913 people), is the driver of Vietnam’s industrial growth as one of the most developed, populated regions of Vietnam and consists of the capital Hanoi44,45. RRD accounts for $29.4\%$ of Vietnam's GDP and the monthly average income per capita in 2020 in RRD was 5.005 million VND (~ $210)46,47. Understandably, most resources are directed toward this region, resulting in a highly trained healthcare workforce, better health access, and consequently better awareness of post-COVID-19 conditions among residents48. By contrast, NMM is among the most sparsely populated and underinvested regions of Vietnam. NNM is a region with a GDP equal to $\frac{1}{3}$ of RRD ($8.54\%$ of Vienam's GDP)49, monthly average income per capita in 2020 was 2745 million VND (~ $115)46, and received the least healthcare attention during COVID-1950,51. Vietnam has a serious shortage of healthcare workers, which becomes more serious when the COVID-19 pandemic occurs. Specifically, to influence the impact of the COVID-19 epidemic, there were 9,680 health workers who resigned or quit52, leading to NNM being heavily affected. To now, only $19.9\%$ of residents in Son La received the 3rd dose of the COVID-19 vaccine53.
The case of RRD and NMM suggested that COVID-19 awareness and care-seeking intention play a crucial role in reducing post-COVID-19 conditions. The adaptability to national response also contributes to the inconsistency of COVID-19 outcomes across regions. During the peak of the pandemic, NMM recorded a relatively lower prevalence of cases compared to RDD and other metropolitan areas51. However, low infection rates in NMM were primarily due to secluded locations, rather than successful pandemic control and compliance of residents. On the other hand, while being the center of coronavirus spread and suffering the most devastating damages in terms of human, financial capacity and health workforce, RDD is among regions with the highest levels of adaptability to national guidelines such as social distancing, masking, and m-health utilization. As post-COVID-19 symptoms are long-term and rarely fatal, the distinguishing factor is adaptability to the national response. Differences in levels of awareness and compliance to guidelines between RDD and NMM demonstrate a clear correlation to post-COVID-19 outcomes. In this post-COVID-19 era, it is important that we consider levels of adaptation to previous strategies as the key indicator instead of the number of cases or deaths in a population.
Regarding post-COVID-19 care, $33\%$ of patients chose mobile applications as their preferred healthcare platform compared to $36.3\%$ who preferred traditional hospital appointments. This number showed substantial progress in the transformation of healthcare delivery in Vietnam from traditional to online platforms. Before COVID-19, online healthcare models were not popular nor encouraged in Vietnam due to inconsistencies in care quality between national, public facilities, and online services at the current time. In 2020, social restrictions due to COVID-19 have not only served as an incentive for citizens to utilize telehealth services, but also promoted the improvement of care among providers to cater to the increasing needs of the population, such as the national hotline to resolve inquiries about COVID-19 at home, Zalo COVID-19 consultation chatbot or televised and online art therapy sessions by non-profit organizations to name a few54–57. After 2 years of COVID-19, it is clear that Vietnamese are adapting to the telehealth model, and telehealth promotion has also identified as one of Vietnam’s digital transformation goals from 2020 to 2025 under project 2628/QDBYT. Therefore, it is important that this favorable trend continues even after COVID-19 restrictions are lifted.
Several implications can be drawn from the above findings. First, a global framework should be provided, including non-COVID variables such as age, daily health behavior, and comorbidities. In this study, these terms were used as one under the term “post-COVID-19 symptoms” and defined as symptoms presenting for more than 4 weeks. Various terms for post-COVID-19 symptoms such as “long-COVID”, “chronic COVID” or “post-acute COVID-19” should be distinguished with specific time periods. More data on post-COVID-19 conditions should be collected from resource-scarce settings and in LMICs to inform evidence-based interventions and identify multidimensional determinants such as social, financial, and cultural characteristics of each population. Secondly, validated information should be provided more extensively to raise awareness of post-COVID-19 symptoms and avoid potential confusion between post-COVID-19 conditions and symptoms of other diseases. When developing and applying a national strategy for post-COVID-19 conditions, contextual and behavioral characteristics of different settings such as health system capacity, resident behavior, and level of adaptation to guidelines must be considered. At the managerial level, healthcare resources should be distributed more evenly to ensure health access and adequate information in underserved regions, such as the Northern Midlands and Mountains. As these regions are scarce in resources, instead of large-scale events, province leaders can utilize television and social media as effective means of information delivery to improve health literacy, and address hesitancy toward care-seeking practices among residents. In terms of rehabilitation, it is important that post-COVID-19 care be provided regularly for a length of time due to the long-term nature of this condition. We suggest that existing resources be mobilized and managed by regions or communities instead of the government to ensure a rapid response. As discussed, adaptability to national strategy varies across regions, and thus it may be more effective for authorities to decide on best practices for their populations. Finally, policymakers should strengthen digital safety and privacy of healthcare consulting through technology platforms as well as empower youth-led digital healthcare startups in delivering health services to relieve the burden of the national frontline.
Our study's sample size is a strength, as it was drawn from one of the largest post-COVID-19 care initiatives in Vietnam, providing a national reference dataset. However, there are several limitations to our findings. Firstly, because this was a cross-sectional study, we cannot infer causation. Secondly, our data may be under-reported due to the recruitment of participants only from health facilities, which excludes those who did not seek medical attention despite experiencing post-COVID-19 symptoms. This exclusion disproportionately affects older adults. Additionally, our study focused on the young workforce, recognizing their importance in national development, and as a result, the majority of responses were obtained from this age group. Thirdly, it is essential to note that COVID-19 grading criteria were self-reported, which could be subject to recall bias. Therefore, we combined medical records with participants' responses to minimize this potential bias. Fourthly, there may be selection biases resulting from the online survey platform, which relied on volunteers to guide eligible participants through the questionnaire. Lastly, our study only examined the 15 most common post-COVID-19 symptoms, and other less common symptoms were not included. Nonetheless, our findings identified concerning trends in post-COVID-19 symptoms and have significant implications for policymakers, clinicians, and the general public.
Besides, it is important to acknowledge that our sample might be affected by a potential selection bias and decreased representativeness given that two-thirds of respondents were male and the median age was 18. Previous studies have shown sex differences in sex outcomes, thus, our findings might not fully representative of the whole Vietnamese population from an epidemiological view. However, from an evidence-informed policy making perspective, we would emphasize that this is the largest and most optimal sample we could recruit in the Vietnamese population, in a timely manner. First, because of the confidentiality of the information regarding COVID-19 patients, it is impossible to construct a nationally representative sample frame and sample. Secondly, for resource-scare settings, the implementation of free-of-charge health check-ups on this large scale is unique, especially during the COVID-19 pandemic. Last but not least, the timing of evidence ready to inform policy development in Vietnam that this study has provided is a good example of knowledge translation, and could serve as a reference for global health policy practice.
## Conclusion
Our study investigates the prevalence of post-COVID-19 symptoms among recovered patients in Vietnam. A gap in definitions should be addressed before developing a timeframe and response strategy for chronic COVID-19. We were able to identify several determinants, including age, infection period, and underlying conditions as well as an important trend in telehealth utilization. Disproportionate distribution of resources and inconsistent post-COVID-19 patterns were found across regions, which suggested a shift in strategy implementation from national to province-based. In the upcoming phase in the fight against COVID-19, it is important that post-COVID-19 symptoms are monitored closely and tackled by a combination of pharmaceutical, psychological, and public health interventions.
## Study design and sampling methods
On May 2022, a cross-sectional study was conducted on participants who registered for post-COVID-19 medical examination during the Post-COVID-19 Care Program in accordance with Plan No. 52-KH/TWH of the Central Committee of the Vietnam Youth Union. This government program focused on post-COVID-19 health care for the Vietnamese population. Those who come to examination include people who had COVID-19 and were experiencing post-COVID-19 symptoms as listed by the Ministry of Health.
The research team developed a survey questionnaire consisting of 4 parts and 20 questions online. When people participate in Post-COVID-19 Care Program, volunteers have guided participants by scanning the QR code which allowed people access directly to the questionnaire. Volunteers were issued around hospitals and health examination points to guide participants in completing the questionnaire. The severity of COVID-19 at the onset of participants was collected by volunteers from medical records for patients with COVID-19.
## Participants
All people infected with COVID-19 and who had medical records for patients with COVID-19 were invited to participate in the study. There were 49,496 patients are examined and consulted in the Post-COVID-19 Care Program, 17,093 individuals returned to this survey, and 12,361 completed the survey. The completion rate was $72.3\%$.
The criteria for selecting the subjects were as follows:Aged from 16 to 35 years oldAgree to participate in the studyTook part in medical examination during the Post COVID-19 Care ProgramCurrently living in Vietnam Exclusion criteria:Suffered from serious cognitive impairment or were unable to answer questionsWere unable internet access, survey link, and reach out and complete surveys
## Measurements
A structured questionnaire was developed consisting of three main components: [1] Individual characteristics; [2] Characteristics of participants when infected with COVID-19, and [3] Post-COVID-19 symptoms characteristics. The questionnaire was first uploaded to the online survey platform SurveyMonkey (surveymonkey.com), a convenient and cost-effective survey design platform that enables quick, efficient data collection, especially in times of social distancing. This approach may exclude certain population groups in Vietnam who do not have access to smart devices, but smartphone usage in Vietnam ranked top 10 globally58. By March 2022, the total number of smartphone subscribers was 93.5 million, reaching $73.5\%$ of the country’s adults59. Hence, it is reasonable to use this approach in the COVID-19 context in Vietnam.
The structured questionnaire was then finalized with the following main sections:
## Post-COVID-19 symptoms and the number of post-COVID-19 symptoms
We designed 15 items corresponding to 15 common symptoms during COVID-19 recovery, including rash, diarrhea, change in smell or taste, abdominal pain, chest pain, fast beating, anxiety, changes in menstrual cycles, myalgia, somnipathy, dyspnea, cough, difficulty thinking or concentrating, headache, and fatigue. Participants were then divided into 5 groups, including 0 “Asymptomatic”; 1 “1 symptom”; 2 “2 symptoms”; 3 “3 symptoms”; and 4 “4 or more than 4 symptoms”. We also calculated the number of post-COVID-19 symptoms of participants and categorized them as neurological symptoms, digestive symptoms, respiratory and heart symptoms, or other symptoms through 15 common symptoms of patients. It is vital to note that not only is symptom presence required but so are symptom duration and intensity60. Hence, we used the number of post-COVID-19 symptoms as a variable to measure the severity of post-COVID-19 symptoms.
Neurological symptoms: fatigue, difficulty thinking or concentrating, headache, somnipathy, and anxiety. The number of neurological symptoms ranged from 0 to 5.
Digestive symptoms: change in smell or taste, diarrhea, and abdominal pain. The number of digestive symptoms ranged from 0 to 3.
Respiratory and heart symptoms: dyspnea, cough, chest pain, fast beating. The number of respiratory and heart symptoms ranged from 0 to 4.
Other symptoms: rash, myalgia, and changes in menstrual cycles. The number of other symptoms ranged from 0 to 3.
## Individual characteristics
Respondents reported their socio-demographic information, including age, gender (male/female), BMI index, and the city they lived in, which were divided into seven economic zones of Vietnam: Southeast Region, Northern Midlands and Mountains, Red River Delta, North Central Region, South Central Coast, Central Highlands, and Mekong Delta Region. Map of 63 provinces in Vietnam was presented in Appendix 6.
## Behaviors
Participants self-reported their current status of smoking, alcohol usage, and exercise after recovering from COVID-19.
## Co-morbidities
Participants reported their comorbidities which were diagnosed by doctors. There were some common comorbidities such as diabetes, cardiovascular disease, chronic lung disease, neurological disease, and cancer.
## Characteristics of COVID-19 infection
We used three questions to measure the characteristics of COVID-19 infection among participants, including:
## Time since COVID-19 onset
1 month, 1–4 months, 4–6 months, and above 6 months. Time since COVID-19 onset as a period between the date when COVID-19 patients were confirmed negative for COVID-19 using PCR tests to the time they entered this study.
## COVID-19 infection period
Less than 7 days, 7–14 days, and more than 14 days. COVID-19 infection period was defined as a period between when people were confirmed positive for COVID-19 and when they were confirmed negative for COVID-19 using the PCR test.
## The Severity of COVID-19 at the onset
Asymptomatic, mild, moderate, and severe61:Asymptomatic testing positive for COVID-19 but having no symptoms that are consistent with COVID-19 infection. Mild having any of the following signs and symptoms of COVID-19. fever, cough, sore throat, malaise, headache, muscle pain, nausea, vomiting, diarrhea, loss of taste and smell; but not having shortness of breath, dyspnea, or abnormal chest imaging. Moderate experiencing lower respiratory disease during clinical assessment or imaging and having SpO2 above $94\%$ on room air at sea level. Severe having SpO2 below $94\%$ on room air at sea level, PaO2/FiO2 < 300 mm Hg, respiratory rate above 30 breaths/min, or lung infiltrates above $50\%$.
The severity of COVID-19 at the onset of the participant was diagnosed by the doctor and divided into four levels: asymptomatic, mild, moderate, and severe.
## Prevalence of COVID-19 cases by region
The total number of COVID-19 was taken from official data of the Ministry of Health as of June 14, 202262. We measured the prevalence of COVID-19 cases per 100,000 population and the case fatality rate due to COVID-19 using the following formula:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} Prevalence \, of \, COVID - 19 \, cases \, per \, 100,000 \, population\;in\;a\;region\, \hfill \\ = \,(Number \, of \, COVID - 19 \, cases\;in\;a\;region)/(Number \, of \, population\;in\;a\;region)\, \times \,100,000. \hfill \\ \end{gathered}$$\end{document}PrevalenceofCOVID-19casesper100,000populationinaregion=(NumberofCOVID-19casesinaregion)/(Numberofpopulationinaregion)×100,000.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\text{Case fatality rate}}\, = \,(Number \, of \, deaths \, due \, to \, COVID - 19\;in\;a\; \hfill \\ region)/(Number \, of \, COVID - 19 \, cases\;in\;a\;region)\, \times \,100. \hfill \\ \end{gathered}$$\end{document}Case fatality rate=(NumberofdeathsduetoCOVID-19inaregion)/(NumberofCOVID-19casesinaregion)×100.
After that, we divided the prevalence of COVID-19 cases per 100,000 population and the case fatality rate due to COVID-19 into three groups low, medium, and high prevalence.
Data on the number of COVID-19 cases, the number of deaths due to COVID-19, the prevalence of COVID-19 cases, and the case fatality rate due to COVID-19 was presented in Appendix 1.
## Preference of post-COVID-19 care
Participants self-reported their preference for health care and consult after recovering from COVID-19, including medical examination at the hospital, medical examination, and consultation via phone App, getting health care materials at home, direct medical examination at free examination programs, calling the doctor, and consulting by phone.
## Data analysis
We analyzed data using STATA version 16 (Stata Corp. LP, College Station, United States of America). With missing data, we used the Listwise Deletion method to clean data before analyzing. Continuous variables were presented as mean and standard deviation (SD), while categorical variables were presented as frequencies with percentages. We used the Chi-squared and Kruskal–Wallis tests to test the difference between each group. We used the “xtile” function to separate the prevalence of COVID-19 cases and the case fatality rate due to COVID-19 into 3 groups: low, medium, and high prevalence.
A multi-level modeling approach with mixed effect model was used for determining associated factors of post-COVID-19 severity at the regional level while adjusting for differences in provincial policies and regulations on COVID-19 control and preparedness which were treated as random effects.
Potential covariates for full models of four main types of post-COVID-19 symptoms, COVID-19 at the onset, level of post-COVID-19 symptoms, and the number of post-COVID-19 symptoms included individual characteristics, clinical manifestations when the got infected with COVID-19, the prevalence of COVID-19 cases, and case-fatality rate due to COVID-19. We used multilevel linear random-effects models to identify the factor associated with the number of post-COVID-19 symptoms. Multi-level random effects ordered logistic model was used to identify the factors related to the severity of COVID-19 at the onset and the level of post-COVID-19 symptoms. In this study, we used location which includes 63 provinces in Vietnam as a cluster variable. A p-value (P) of < 0.05 was considered statistically significant.
## Ethical approval
Ethical approval is granted by Hanoi Medical University and Youth Research Institute; the research protocol was reviewed by the Vietnam Young Physician Association and the Institute of Health Economics and Technology’s scientific committee. All participants had provided informed consent, and the survey did not have any impact on their medical examination and consultation process. Study data were de-identified for analysis. All procedures performed in studies involving human participants were in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Patients invited to participate in the study have fully explained the content, purpose, and benefits when participating in the study. The collected information is kept confidential and the information is only for research purposes, not for other purposes, the data is encrypted to ensure confidentiality of the information. Participants can refuse to participate in the study at any time. During the participation process, they can still withdraw from the study if the research they feel uncomfortable.
## Supplementary Information
Supplementary Information 1.Supplementary Information 2.Supplementary Information 3.Supplementary Information 4.Supplementary Information 5.Supplementary Information 6. The online version contains supplementary material available at 10.1038/s41598-023-30790-x.
## References
1. 1.Worldometers. COVID Live: Weekly trends.) (2022).
2. 2.Organization WH. Weekly epidemiological update on COVID-19 - 23 November 2021.) (2021).
3. Van Tan L. **COVID-19 control in Vietnam**. *Nat. Immunol.* (2021.0) **22** 261-261. DOI: 10.1038/s41590-021-00882-9
4. Nguyen TV. **In the interest of public safety: Rapid response to the COVID-19 epidemic in Vietnam**. *BMJ Glob. Health* (2021.0) **6** e004100. DOI: 10.1136/bmjgh-2020-004100
5. 5.VN TT. COVID-19 Report 8th June Vietnam.) (2022).
6. Heitzman J. **Impact of COVID-19 pandemic on mental health**. *Psychiatr. Pol.* (2020.0) **54** 187-198. DOI: 10.12740/PP/120373
7. Pizam A. **The aftermath of the corona virus pandemic**. *Int. J. Hosp. Manag.* (2021.0) **95** 102909-102909. DOI: 10.1016/j.ijhm.2021.102909
8. Verma AK, Prakash S. **Impact of covid-19 on environment and society**. *J. Glob. Biosci.* (2020.0) **9** 7352-7363
9. 9.Mahase E. Covid-19: What do we know about “long covid”? BMJ370, (2020).
10. Michelen M. **Characterising long COVID: A living systematic review**. *BMJ Glob. Health* (2021.0) **6** e005427. DOI: 10.1136/bmjgh-2021-005427
11. 11.Excellence TNIfHaC. COVID-19 rapid guideline: managing the long-term effects of COVID-19.) (2022).
12. 12.Soriano, J.B., Murthy, S., Marshall, J.C., Relan, P., Diaz, J.V., Group WCCDW. A clinical case definition of post-COVID-19 condition by a Delphi consensus. Lancet Infect. Dis. (2021).
13. 13.Association AM. Long COVID: Over 200 symptoms, and a search for guidance.). AMA (2022).
14. Sisó-Almirall A. **Long Covid-19: proposed primary care clinical guidelines for diagnosis and disease management**. *Int. J. Environ. Res. Public Health* (2021.0) **18** 4350. DOI: 10.3390/ijerph18084350
15. 15.Sivan, M., Taylor, S. NICE guideline on long covid). British Medical Journal Publishing Group (2020).
16. Afulani PA. **Job satisfaction among healthcare workers in Ghana and Kenya during the COVID-19 pandemic: Role of perceived preparedness, stress, and burnout**. *PLOS Glob. Public Health* (2021.0) **1** e0000022. DOI: 10.1371/journal.pgph.0000022
17. Uphoff EP. **Mental health among healthcare workers and other vulnerable groups during the COVID-19 pandemic and other coronavirus outbreaks: A rapid systematic review**. *PLoS ONE* (2021.0) **16** e0254821. DOI: 10.1371/journal.pone.0254821
18. 18.Galvin, G. Nearly 1 in 5 health care workers have quit their jobs during the pandemic. Morning Consult4, (2021).
19. 19.Le, C. Vietnam lacks 24,000 preventive healthcare workers). VnExpress International (2022).
20. 20.Vietnam MoH. Up to 203 post-COVID-19 symptoms, Ministry of Health guides medical examination and treatment for recovered patients) (2022).
21. 21.September 8: An additional 12,680 cases of COVID-19, Ho Chi Minh City and Binh Duong nearly 10,500 cases). Ministry of Health (2021).
22. 22.COVID-19: Strategic decisions for people's lives and health). Ministry of Health (2021).
23. Takahashi T. **Sex differences in immune responses that underlie COVID-19 disease outcomes**. *Nature* (2020.0) **588** 315-320. DOI: 10.1038/s41586-020-2700-3
24. 24.Gaber, T.A.-Z.K., Ashish, A., Unsworth, A. Persistent post-covid symptoms in healthcare workers. Occup. Med.71, 144–146 (2021).
25. Augustin M. **Post-COVID syndrome in non-hospitalised patients with COVID-19: A longitudinal prospective cohort study**. *Lancet Reg. Health – Eur.* (2021.0) **6** 100122. DOI: 10.1016/j.lanepe.2021.100122
26. Huang L. **1-year outcomes in hospital survivors with COVID-19: A longitudinal cohort study**. *Lancet* (2021.0) **398** 747-758. DOI: 10.1016/S0140-6736(21)01755-4
27. La Vecchia C, Giordano SH, Hortobagyi GN, Chabner B. **Overweight, obesity, diabetes, and risk of breast cancer: Interlocking pieces of the puzzle**. *Oncologist* (2011.0) **16** 726-729. DOI: 10.1634/theoncologist.2011-0050
28. Garfinkel L. **Overweight and cancer**. *Ann. Intern. Med.* (1985.0) **103** 1034-1036. DOI: 10.7326/0003-4819-103-6-1034
29. Powell-Wiley TM. **Obesity and cardiovascular disease: A scientific statement from the American Heart Association**. *Circulation* (2021.0) **143** e984-e1010. DOI: 10.1161/CIR.0000000000000973
30. Sagarra-Romero L, Viñas-Barros A. **COVID-19: Short and long-term effects of hospitalization on muscular weakness in the elderly**. *Int. J. Environ. Res. Public Health* (2020.0) **17** 8715. DOI: 10.3390/ijerph17238715
31. Han Q, Zheng B, Daines L, Sheikh A. **Long-Term sequelae of COVID-19: A systematic review and meta-analysis of one-year follow-up studies on post-COVID symptoms**. *Pathogens* (2022.0) **11** 269. DOI: 10.3390/pathogens11020269
32. 32.Hussein, A.A.M. et al. Post-COVID-19 functional status: Relation to age, smoking, hospitalization and comorbidities. MedRxiv, (2020).
33. Hall J. **Identifying patients at risk of post-discharge complications related to COVID-19 infection**. *Thorax* (2021.0) **76** 408-411. DOI: 10.1136/thoraxjnl-2020-215861
34. Fernández-de-Las-Peñas C. **Prevalence of post-COVID-19 symptoms in hospitalized and non-hospitalized COVID-19 survivors: A systematic review and meta-analysis**. *Eur. J. Intern. Med.* (2021.0) **92** 55-70. DOI: 10.1016/j.ejim.2021.06.009
35. 35.Baig, A.M. Chronic COVID syndrome: Need for an appropriate medical terminology for long-COVID and COVID long-haulers. J. Med. Virol. (2020).
36. 36.Halpin, S., O'Connor, R., Sivan, M. Long COVID and chronic COVID syndromes. J. Med. Virol. (2021).
37. 37.Cavallini, A., Micieli, G., Bussone, G., Rossi, F., Nappi, G. Headache and quality of life. Headache: J. Head Face Pain35, 29–35 (1995).
38. Hsu L-c, Hu H-h, Wong W-j, Wang S-j, Luk Y-o, Chern C-m. **Quality of life in elderly patients with dizziness: Analysis of the Short-Form Health Survey in 197 patients**. *Acta Otolaryngol.* (2005.0) **125** 55-59. DOI: 10.1080/00016480410017512
39. Bryson WJ. **Long-term health-related quality of life concerns related to the COVID-19 pandemic: A call to action**. *Qual. Life Res.* (2021.0) **30** 643-645. DOI: 10.1007/s11136-020-02677-1
40. 40.Prevention CfDCa. Stroke Signs and Symptoms).
41. 41.Service UKNH. Brain tumours).
42. 42.Katz, P.O., Gerson, L.B., Vela, M.F. Guidelines for the diagnosis and management of gastroesophageal reflux disease. Off. J. Am. College of Gastroenterol. | ACG108, 308–328 (2013).
43. Leung AN. **Pulmonary tuberculosis: The essentials**. *Radiology* (1999.0) **210** 307-322. DOI: 10.1148/radiology.210.2.r99ja34307
44. 44.Associates DS. Regional Spotlight: Red River Delta (Hanoi)). Vietnam Briefing (2013).
45. 45.Wikipedia. Red River Delta) (2022).
46. 46.Preliminary results: Popular life survey in 2020). General statistics (2021).
47. 47.Van, H. The Red River Delta must lead the process of restructuring the economy and transforming the growth model of the whole country.). Socialist Republic of Vietnam (12/02/2023).
48. 48.Vietnam MoH. Local curative care & consultation (including CHS) indicators by province, 2019) (2021).
49. 49.Truong, D. Green, sustainable and comprehensive development of the Northern Midlands and Mountains). Vietnam+ (22/02/2023).
50. 50.Wikiwand. List of Vietnamese subdivisions by GDP) (2019).
51. 51.Vietnam MoH. Vietnam COVID Report 6th November) (2021).
52. 52.In 18 months, 9,680 health workers resigned and quit: The Ministry of Health pointed out 4 main reasons). Ministry of Health (19/09/2022).
53. 53.International V. Vaccination data) (2022).
54. 54.Health Mo. Hanoi announces the 2022 call center for health consultation related to the COVID-19 epidemic) (2021).
55. 55.Online S. Chatbot updating citizens on Covid-19 information daily) (2020).
56. 56.Agency VN. Online exhibition – Safe way to enjoy art amid pandemic) (2020).
57. 57.Today DN. Befriending emotions with art therapy) (2021).
58. 58.Vietnam smartphone use in top 10 globally). Ministry of Information and Communications of the Socialist Republic of Vietnam (2023).
59. 59.Vietnam targets 85% smartphone usage by end of 2022). Ministry of Information and Communications of the Socialist Republic of Vietnam (2022).
60. Munblit D. **Studying the post-COVID-19 condition: Research challenges, strategies, and importance of Core Outcome Set development**. *BMC Med.* (2022.0) **20** 50. DOI: 10.1186/s12916-021-02222-y
61. 61.National Institute of Health. Clinical Spectrum of SARS-CoV-2 Infection) (2022). https://www.covid19treatmentguidelines.nih.gov/.
62. 62.Ministry of Health. COVID-19 epidemic bulletin at noon on June 14: 100 more cases of COVID-19 in 4 provinces and cities; Vietnam has 10,730 patients) (2022).
|
---
title: 'Laryngeal sensory neuropathy caused by COVID-19: findings using laryngeal
electromyography'
authors:
- Paulina Krasnodębska
- Agata Szkiełkowska
- Beata Miaśkiewicz
journal: European Archives of Oto-Rhino-Laryngology
year: 2023
pmcid: PMC10022564
doi: 10.1007/s00405-023-07895-0
license: CC BY 4.0
---
# Laryngeal sensory neuropathy caused by COVID-19: findings using laryngeal electromyography
## Abstract
### Purpose
Laryngeal sensory neuropathy (LSN) is caused by a disorder of the superior laryngeal nerve or the recurrent laryngeal nerve. A diagnosis of LSN should include laryngeal electromyography (LEMG) and laryngovideostroboscopy (LVS). The aim of this study was to characterize the physical and subjective symptoms of neuropathy in patients diagnosed with LSN following COVID-19.
### Material and methods
Since the beginning of the COVID-19 pandemic, 6 patients who had recovered from the disease presented to us with LSN symptoms. All patients underwent laryngological and phoniatric examination, objective and subjective voice assessment, and LEMG.
### Results
The most common LSN symptom reported by patients was periodic hoarseness of varying severity. Other common symptoms were the sensation of a foreign body in the throat and voice fatigue. Endoscopy often showed functional abnormalities. The LSN patients could be characterized by LEMG recordings, and all showed abnormal activity of the cricothyroid (CT) muscle. The degree of EMG changes in the CT correlated moderately with the severity of dysphonia.
### Conclusions
Sensory neuropathy of the larynx may be a long-lasting complication of SARS-COV-2 infection. The severity of EMG neuropathic changes in the CT muscle broadly corresponds to the severity of dysphonia.
## Introduction
Sensory neuropathies appear under a diverse array of conditions. The best known examples are diabetic peripheral neuropathy and trigeminal neuralgia. The pathogenesis results from nerve degeneration caused by metabolic damage, mechanical trauma, or viral infection [1]. Laryngeal sensory neuropathy (LSN) is caused by disorders of the superior laryngeal nerve or the recurrent laryngeal nerve. Symptoms of the disease may be throat discomfort (ranging from paresthesia to numbness, often including pain), permanent dysphonia, laryngospasm, or chronic cough [2–6]. Diagnosis of LSN should include laryngeal electromyography (LEMG) and laryngovideostroboscopy (LVS) [7].
In differential diagnosis several well-defined clinical syndromes have overlapping symptomatology: laryngo-pharyngeal reflux, postnasal drip, chronic refractory cough, globus pharyngeus, paradoxical vocal fold movement, and muscle tension dysphonia [8]. It is hypothesized that in all these states laryngeal hypersensitivity is a common underlying factor [9]. In cases of COVID-19 infection, neuropathy may be caused by viral mechanisms and inflammatory responses (host cytokines and release of chemokines) [10–12]. Nervous system disorders are present in $36\%$ of individuals infected with SARS-CoV-2 [13, 14].
In the literature, laryngeal paralysis arising from innervation damage is the most common complication due to COVID-19 [15]. Sensory neuropathy is seldom mentioned as a complication, mentioned only as co-existing with movement dysfunction [16, 17]. When a search was done of medical databases, using the keywords "COVID" or "SARS" and "laryngeal sensory neuropathy" without time limit, we found only 1 article in PubMed and 54 in Google Scholar.
## Aim
The aim of this study was to examine patients who came to the Audiology and Phoniatrics Clinic, after suffering SARS-CoV-2 infection and having LSN as a complication. We wanted to characterize the physical and subjective manifestations of the neuropathy, including any electromyographic features.
## Material
The material of the study consisted of patients presenting to our hospital with symptoms suggestive of post COVID-19 sensorineural dysfunction of the larynx. The patients were diagnosed with LSN and referred for LEMG testing. Since the beginning of the pandemic, 6 patients presenting with LSN symptoms came under our care. All of them had had COVID-19 infection confirmed by a SARS-COV-2 RNA test. There was 1 male and 5 females, with a mean age of 50.6 years (SD 10.9). None of the 6 patients had been in a critical state or required admission to an intensive care unit due to COVID-19. The inclusion criteria for the study were the duration of symptoms, the acute onset of laryngeal-related sensory symptoms and ruling out any other cause of the disorder. Patients were included in the study only when other previously undiagnosed common etiologies for throat discomfort had been excluded, based mainly on previous neurological, pulmonological, or cardiological diagnosis. Other etiologies considered in the differential diagnosis were: cricopharyngeal dysfunction, recent upper respiratory tract infection, current smoking, untreated asthma, untreated rhinitis, untreated gastroesophageal reflux, significant psychological factors (such as psychosis, schizophrenia, or mood disorders that prevented participation in the assessment), or neurological impairment [9]. During diagnostic hospitalisation at our clinic, in addition to the LEMG examination, patients underwent phoniatric and speech therapy assessment to exclude other laryngeal diseases such as muscle tension dysphonias. We have excluded one patient from the statistical analysis (Patient 6). Due to a history of previous right vocal fold dysfunction, electromyographic recordings from the muscles on the right side were not included in further analysis.
In all patients the symptoms that were the reason for referral for otolaryngologic and phoniatry diagnosis appeared up to 1 month after COVID-19 and lasted a minimum of 3 months, which meets the criteria for post-viral vagal neuropathy (PVVN) as a result of COVID-19 [17–19]. *The* general characteristics of the patients are listed in Table 1. Detailed descriptions of the patients' cases are presented below. Table 1Details of 6 patients suffering laryngeal sensory neuropathy (LSN) and their symptomsPatient noAgeSexDate of COVID-19 infectionDate of phoniatric examination and EMGSymptoms159F11.20203.2022Vocal fatigueVoice fadingPeriodic hoarseness243F4.20212.2022Vocal fatiguePeriodic hoarsenessDry throat344F5.20211.2022Periodic hoarsenessPainless sensation of a foreign body in the throat440F11.20216.2022Painful sensation of a foreign body in the throatSensation of tightening of the larynx547F11.20216.2022Vocal fatiguePainful sensation of a foreign body in the throat (right side)Periodic hoarsenessDry throat671M3.202111.2021Vocal fatiguePeriodic hoarsenessPainless sensation of a foreign body in the throatSensation of tightening in the larynx
## Patient 1
A 59 year old female suffering from asthma, hypothyroidism, and arterial hypertension was referred for a phoniatric diagnosis due to a 1-year history of hoarseness, vocal fatigue, and voice weakness. The complaints started about a month after COVID-19. In addition to vocal complaints, the patient developed an exacerbation of asthma requiring intensification of pulmonary treatment. At the time of admission to the phoniatry department, the patient’s asthma had been stabilized.
## Patient 2
A 43 year old female voice worker, without comorbidities, was referred for a phoniatric evaluation due to laryngeal dysfunction after COVID-19 infection. The complaints had lasted 10 months. Initially, intercostal nerve paralysis, paradoxical vocal folds movement, right-sided trigeminal neuralgia, and muscle weakness on the right side of the body were diagnosed. In addition to periodic hoarseness, the patient complained of voice fatigue while speaking, dry throat, and a problem with breath coordination during vocalization. The patient had had a neurological diagnosis prior to admission to the phoniatry department, which excluded other neurological causes of the symptoms.
## Patient 3
A 44 year old female teacher, with a history of seasonal allergy, was referred for a phoniatric evaluation due to laryngeal dysfunction after COVID-19 infection. The complaints had lasted for 8 months. Initially, she complained of hoarseness, increased dryness in the throat, impaired concentration and attention, and general irritability. The patient reported a sensory disturbance on the scalp and in the nasal area for a couple of months after the infection. The patient had a gastrological, otolaryngological, neurological, and rheumatological evaluation prior to admission to the phoniatry department, ruling out other causes of the laryngeal dysfunction.
## Patient 4
A 40 year old female voice worker, without comorbidities, was referred for a phoniatric evaluation due to laryngeal dysfunction after COVID-19 infection. The complaints had lasted for 7 months. The patient initially reported choking while eating. In addition, she had persistent sore throat (aggravated after speaking) and feelings of tightness and burning in the throat. She did not complain of hoarseness. Gastrointestinal dysfunction had been ruled out before admission to hospital.
## Patient 5
A 47 year old female with a 2-year history of trigeminal neuralgia was referred for phoniatric evaluation due to right-sided pain in the throat, dry throat, and voice fading following COVID-19 infection 8 months earlier. A neurological examination showed no abnormalities. Due to a previous history of neuralgia, the neurologist decided to administer pregabalin.
## Patient 6
A 71 year old male with chronic heart failure and a 6-year history of right-sided laryngeal paralysis after thyroid surgery (laryngeal laryngoplasty with hyaluronic acid) was referred for a phoniatric evaluation due to laryngeal dysfunction exacerbation after COVID-19 infection. After the infection, the patient noticed a feeling of discomfort in the larynx, increased fatigue of the voice, and memory deterioration. The patient was reviewed by a pulmonologist, cardiologist, and neurologist. He did not require modification of treatment due to the absence of changes in the medical examinations. Due to a history of previous right vocal fold dysfunction, electromyographic recordings from the muscles on the right side were not included for further analysis.
## Methods
Procedures consisted of a laryngological and phoniatric examination, objective and subjective voice assessment, and LEMG. Endoscopy of the larynx was performed through the nose in two ways: using a flexible 3.2 mm diameter Xion fiberoptic scope and a rigid endoscope. Evaluation with a flexible endoscope avoids misclassification of hyperfunctional features due to a forced tongue position during examination with rigid optics. Laryngovideostroboscopy (LVS) was done and classified according to standards in the literature [20–22]. The following stroboscopic parameters were evaluated: amplitude, mucosal waveform, closed phase, phase differences, and symmetry of vibration. Acoustic voice analysis was assessed according to the Yanagihara scale, assigning a grade of hoarseness on the basis of the spectrogram. Perceptual voice evaluation was based on the 4-point GRBAS scale (G, grade; R, roughness; B, breathiness; A, asthenicity; S, strain) [23]. All patients also completed a detailed self-assessment with the Voice Handicap Index (VHI) questionnaire.
For LEMG evaluation we used the standard percutaneous approach with a concentric needle electrode used clinically for diagnosis of the thyroarytenoid (TA) and cricothyroid (CT) muscles [24]. The procedure records electrical responses of branches of the vagus nerve supplying motor and sensory innervation to the larynx via the recurrent laryngeal nerve and the superior laryngeal nerve. Accurate electrode placement was confirmed by asking the patient to phonate a sustained /a/ and /e/ and seeing appropriate neural activity. The ground electrode was placed at the midline over the forehead [25]. Recordings were made using the Neurosoft EMG apparatus.
All procedures were approved by our bioethics committee (KB.IFPS $\frac{1}{2021}$). Relationships between laryngeal parameters in patients with LSN were assessed using Spearman correlations.
## Results
Since the beginning of the COVID-19 outbreak, six patients with symptoms fulfilling the criteria for the diagnosis of laryngeal sensory neuropathy due to SARS-COV-2 infection have been referred to our clinic. As shown in Table 1, these patients presented mainly within the last year. This was due to epidemiological strictures and limited clinical activity during the outbreak. Table 1 shows the epidemiological and interview data. Half the subjects were professional voice users. The most common LSN symptom reported by patients was periodic hoarseness of varying severity (in $83\%$ of them). Other frequently reported symptoms were the sensation of a foreign body in the throat (two-thirds of them), voice fatigue (two-thirds), dry throat (one-third), and a sensation of tightness in the throat (one-third).
Table 2 shows the results of LVS examination and voice assessment. On endoscopic examination, functional abnormalities in the form of hyperfunctional dysphonia were mainly observed. Distinctive features of hyperfunctional dysphonia observed by LVS were: supraglottic hypertension, reduced open phase, reduced maximum amplitude, and elongated closed phase [21]. We observed these abnormalities in two-thirds of our patients. In 2 of the first 5 patients we observed periodic abnormal movements of the vocal folds. Acoustic voice analysis confirmed varying degrees of hoarseness (from grade 0 to III according to Yanagihara, with a median of 2). Median grades of all GRBAS features were: G-2, R-1, B-0, A-0.5, S-2. We found no correlation between the grade of hoarseness (derived from the acoustic and perceptual analysis) and VHI scores, but there was a correlation between VHI score and the severity of sensory complaints ($r = 0.84$). Patients reporting throat pain scored higher on the VHI. In the whole study group we found positive correlations between grade of hoarseness according to Yanagihara and G score ($r = 0.95$) as well as S score ($r = 0.86$).Table 2Results of LVS examination and voice assessmentPatient noLVSGrade of hoarseness according to YanagiharaGRBASVHI1Glottal hypofunction, hyperfunction of the supraglottis and lower pharynx, periodical paradoxical movements of the vocal foldsII/IIIG2R1B2A1S2812Hyperfunction of the glottis, periodically observed exacerbated laryngeal closure reflexII/IIIG2R0B1A1S2443Full phonatory and respiratory mobility, periodic slower movement of the right vocal fold, mucosal dryness0/IG1R1B0A0S1304Hyperfunction of the glottis0G0R0B0A0S0645Hyperfunction of the glottisIIIG3R2B0A0S2506Right-sided laryngeal paralysisIIG2R1B0A1S29 Table 3 shows the patients' electromyographic findings. In nearly all the patients their LSN showed up as a neuropathic EMG recording. Only in patient 1 did the recording meet the criteria for myopathy [26]: her low EMG amplitude was regarded as indicating muscle weakness, and under endoscopic examination the patient showed features of glottal insufficiency. In all other patients, recordings showed neuropathic features. Neuropathic abnormalities were observed in both the muscles innervated by the superior laryngeal nerve (CT) and the recurrent laryngeal nerve (TA). Abnormal high-amplitude neuropathic units were more frequently recorded from the CT muscle than from the TA muscle. It was noted that when the mean amplitudes in CT was above 500 μV, numerous or very numerous high-amplitude units were observed. One of the subjects (patient 4), who complained only of sensory impairment, was characterized by abnormal EMG recording only in the CT muscle (TA recording was normal). In another patient (patient 5), a correspondence was observed between the location of the reported symptoms and the location of the deviations in the EMG recording. Patient 5 localized sensory complaints on the right side, matching abnormalities in the EMG recording, which were only visible in the right CT muscle. At the same time, it was noted that in all patients who reported sensory dysfunction, an abnormal neuropathic EMG recording was present in the CT.Table 3Electromyographic findings in two muscles from 6 patients with laryngeal sensory neuropathyPatientSymptom duration (months)SideCricothyroid muscleThyroarythenoid musclePhonationRestPhonationRestMean amplitude (μV)Maximum amplitude (μV)Large MUPsMean amplitude (μV)Maximum amplitude (μV)Large MUPs116Right205252–Fibryl118128––Left210246–Fibryl140174––210Right109124––181656––Left3351957 + + + Residual act167331 + + Residual act38Right131273––2732446 + + + Residual actLeft169585 + + Residual act232687––47Right158760 + Fibryl179974–Residual actLeft184824 + Fibryl170900––58Right172679 + + Residual act187 + Residual actLeft183456–Fasciculations149––68Right148300––137336––Left193426 + Residual act185898 + + Residual actWhen the maximum amplitude in the cricothyroid was > 500 μV, multiple high-amplitude motor unit potentials (MUPs) were always observed. [ act = activity] Figure 1 shows examples of electromyographic recordings from two patients with LSN.Fig. 1Electromyographic recordings of two patients with LSN. At left are recordings from patient 3 who exhibited multiple neuropathic lesions from the CT and TA muscles. At right are recordings from patient 4 who had neuropathic lesions detectable only from the CT muscles 200 microvolt gain, 40ms per division sweep speed Patients with LSN symptoms were statistically analyzed by comparing the severity of EMG changes with the severity of hoarseness (from perceptual voice analysis) as well as VHI score. The intensity of EMG changes in the CT (expressed as the number of neuropathic units) correlated moderately with the severity of dysphonia ($r = 0.46$). By way of contrast, for the TA muscle we found no correlation between the severity of hoarseness and the amount of neuropathic deviation on EMG examination. Moreover, we did not observe a correlation with VHI score.
Patients showed abnormal activity of the CT muscle not only during phonation but also during respiratory phases. From examination of the TA muscle, neuropathic abnormalities were registered in 4 patients, suggesting concomitant motor nerve fiber damage in addition to sensory nerve damage. Of the first 5 patients, 3 displayed abnormal neurogenic units during phonation in the TA muscle (2 unilateral, 1 bilateral). Only patient 3 showed endoscopic features of nerve paresis (this patient also had the most intense EMG changes in the TA).
The results of patient 6 (with long-standing right-side laryngeal paralysis), suggest there may be additional damage to the laryngeal innervation on the left side due to complaints that occurred after COVID-19 infection but which were not present before. The analysis of the results did not show a correlation between the severity of hoarseness and the severity of neuropathic changes in the TA muscle. An interesting observation is the presence of increased tension of the examined muscles at rest in patients in whom neuropathic activity was observed during volitional activity.
## Discussion
This paper has presented observations of patients with sensory neuropathy of the larynx following COVID-19 infection. Due to epidemiological strictures, ENT diagnostics was curtailed during the infectious phase so the LSN problem seems to have been underestimated. Of the 6 patients seeking our help for complaints due to LSN, about half were professional voice users. Helding has pointed out that LSN seems to strongly affect the singing population [19]. Despite the lack of a comprehensive literature, perhaps due to under-reporting, LSN is mentioned in a review of the COVID-19 literature as a potential complication [27]. In addition to post-viral LSN, Orsucci lists three other chronic medical conditions affecting the larynx which can be related to COVID-19 infection [14]. Those medical complications, affecting vocal production, are: intubation and cough-related injury, post-viral vocal fold paralysis or paresis, and chronic vocal fatigue [27]. The true incidence and prevalence of post-COVID-19 LSN is unknown [19].
The most commonly reported complaints in people with post-COVID LSN were periodic hoarseness of varying severity, discomfort or even pain in the pharynx and larynx, and vocal fatigue. A possible mechanism behind the symptoms is damage to the sensory nerves, leading to axonal degeneration and synkinesis. During regeneration and healing, axons from one sensory receptor may connect to fibers that previously carried signals from a different sensory receptor [28]. As Latremoliere points out, this damage commonly starts in unmyelinated fibers of afferent nerves [28]. Other authors also suggest a role for central sensitisation of the afferent reflex [29]. In the material of our work we found paradoxical vocal fold movements in 2 patients, and in one other the laryngeal closure reflex was also disturbed.
A comparison of the VHI questionnaire results with exacerbation of hoarseness as rated by acoustic voice analysis showed no correlation. The main complaint of LSN patients was laryngeal discomfort. Its severity (pain in the laryngeal region) correlated strongly with the handicap index. Sensory signals arising from the laryngeal mucosa are transmitted mainly by the internal branch of the superior laryngeal nerve in addition to branches of the recurrent laryngeal nerve. In our research, we recorded responses coming from the motoneurons of both these nerves. As reported in previous studies of sensory neuropathy, nerve fiber damage can affect both sensory and motor fibers.
We found that an increase in the severity of LEMG abnormalities in the CT correlated with the severity of dysphonia (as assessed from the spectrogram) and with associated motor dysfunction. In a study by Norris on 12 patients with symptoms of LSN, $75\%$ exhibited evidence of motor neuropathy on laryngoscopy [6]. In our study 5 of 6 patients with post COVID-19 LSN had neuropathic changes in the superior laryngeal nerve (which is the main sensory nerve of the larynx) and of the recurrent laryngeal nerve (the main motor nerve of the larynx), but only 2 of patients had features of periodical motor dysfunction visible in LVS.
According to Daia, EMG studies can reveal demyelinating polyneuropathy changes in patients over the course of COVID-19 infection and recovery [30]. According to the author, elements of myopathy could be a new pathological entity in COVID-19 [30]. Myopathic changes in the EMG are observed in severe demyelinating neuropathy and suggest a direct action of COVID-19 on muscular fibers [30]. In this study, we observed myopathic changes in only one patient; this subject had the most severe form of the disease, and her complaints lasted the longest.
Previous publications report treatment of LSN with amitriptyline, gabapentin, and pregabalin [1]. As stated by Norris, patients with evidence of motor neuropathy appear to have better outcomes with neuromodulator therapy [6]. The addition of reflux precautions and acid suppression therapy to neuromodulator therapy is helpful in cases of chronic and recurrent laryngospasm [6]. For the patients in our study, we additionally recommended vitamin B preparations and physiotherapy. So far, we have observed resolution of symptoms with normalization of the EMG recording in just one of the patients (patient 3). Laryngeal sensory neuropathy is a diagnosis of exclusion. There is no defined way to test internal branch of the superior laryngeal nerve or sensory branches of the recurrent laryngeal nerve. The study indirectly searched for abnormalities in the motor innervation of the larynx on the basis of the LEMG results. Neuropathy of sensory fibers does not mean that there is an obligatory motor fibre neuropathy and vice versa, but literature data show a very high rate of co-occurrence of the disfunctions in post-viral lesions. The abnormal LEMG findings presented in this study point to damage of the laryngeal innervation originating from the superior laryngeal nerve. Abnormal activity from the CT muscle may indicate the co-occurrence of sensory neuropathy. In the authors opinion this diagnosis permits differentiation from laryngeal hypersensitivity which is symptomatic of other diseases [4, 7, 31]. The detection of SLN injury on LEMG examination permits the diagnosis to be documented and the severity of the condition to be assessed. On this basis, treatment can be modified with elements of neurological therapy. In addition, the awareness of long-term nerve regeneration prepares the patient for the treatment process.
## Conclusions
Sensory neuropathy of the larynx may be one of the long-lasting complications of SARS-COV-2 infection.
Laryngeal sensory neuropathy is characterized by a variety of symptoms, of which the most common is dysphonia of various degrees.
Voice professionals are more likely to be those seeking medical attention for symptoms of post-COVID-19 LSN.
To diagnose LSN, it is helpful to use electromyography of the muscles innervated by the superior laryngeal nerve and the inferior laryngeal nerve (CT and TA respectively).
The degree of neuropathic changes seen by EMG of the CT muscle broadly corresponds to the severity of dysphonia.
## References
1. Halum S, Sycamore D, McRae B. **A new treatment option for laryngeal sensory neuropathy**. *Laryngoscope* (2009.0) **119** 1844-1847. DOI: 10.1002/lary.20553
2. Amin MR, Koufman JA. **Vagal neuropathy after upper respiratory infection: a viral etiology?**. *Am J Otolaryngol* (2001.0) **22** 251-256. DOI: 10.1053/ajot.2001.24823
3. Bastian RW, Vaidya AM, Delsupehe KG. **Sensory neuro-pathic cough: a common and treatable cause of chronic cough**. *Otolaryngol Head Neck Surg* (2006.0) **135** 17-21. DOI: 10.1016/j.otohns.2006.02.003
4. 4.Lee JK, Mintz S (2006) Letter to the editor. Ann Otol Rhinol Lar-yngol 115:871
5. Arcimowicz M, Niemczyk K. **Chronic cough: new concepts and therapeutic possibilities**. *Pol Otorhinolaryngol Rev* (2012.0) **1** 117-123. DOI: 10.1016/S2084-5308(12)70020-5
6. Norris B, Schweinfurth J. **Management of recurrent laryngeal sensory neuropathic symptoms**. *Ann Otol Rhinol Laryngol* (2010.0) **119** 188-191. DOI: 10.1177/000348941011900307
7. Lee B, Woo P. **Chronic cough as a sign of laryngeal sensory neuropathy: diagnosis and treatment**. *Ann Otol Rhinol Laryngol* (2005.0) **114** 253-257. DOI: 10.1177/000348940511400401
8. Vertigan A, Theodoros D, Gibson P. **Voice and upper airway symptoms in people with chronic cough and paradoxical vocal fold movement**. *J Voice* (2007.0) **21** 361-383. DOI: 10.1016/j.jvoice.2005.12.008
9. Vertigan AE, Bone SL, Gibson PG. **Laryngeal sensory dysfunction in laryngeal hypersensitivity syndrome**. *Respirology* (2013.0) **18** 948-956. DOI: 10.1111/resp.12103
10. 10.Aoyagi Y, Ohashi M, Funahashi R, Otaka Y, Saitoh E (2020) Oropharyngeal dysphagia and aspiration pneumonia following coronavirus disease 2019: a case report. Dysphagia 35:545–548.
11. 11.Ellul M, Benjamin L, Singh B, Lant S, Michael B, Easton A (2020) Neurological associations of COVID-19. Lancet Neurol. 19:767–783
12. Yavarpour-Bali H, Ghasemi-Kasman M. **Update on neurological manifestations of COVID-19**. *Life Sci* (2020.0) **257** 118063. DOI: 10.1016/j.lfs.2020.118063
13. 13.Xiong W, Kwan P, Zhou D, Del Felice A, Duncan J, Sander J (2020) Acute and late neurological complications of COVID19: the quest for evidence Brain. [published online ahead of print]
14. 14.Orsucci D, Ienco E, Nocita G, Napolitano A, Vista M (2020) Neurological features of COVID-19 and their treatment: a review. Drugs Context. 9
15. 15.Lechien J, Hans S (2022) Postacute laryngeal injuries and dysfunctions in COVID-19 patients: a scoping review
16. Torres S, Espinosa J. **Vagus nerve neuropathy related to SARS COV-2 infection**. *IDCases* (2021.0) **26** e01242. DOI: 10.1016/j.idcr.2021.e01242
17. Watson N, Karagama Y, Burnay V, Boztepe S, Warner S, Chevretton E. **Effects of coronavirus disease-2019 on voice: our experience of laryngeal complications following mechanical ventilation in severe coronavirus disease-2019 pneumonitis and review of current literature**. *Curr Opin Otolaryngol Head Neck Surg* (2021.0) **29** 437. DOI: 10.1097/MOO.0000000000000768
18. 18.Rapoport S, Alnouri G, Sataloff R, Woo P (2021) Acute vocal fold paresis and paralysis after COVID-19 infection: a case series. Ann Otol Rhinol Laryngol 10.1177/00034894211047829
19. Helding L, Carroll TL, Nix J, Johns MM, LeBorgne WD, Meyer D. **COVID-19 after effects: concerns for singers**. *J Voice* (2020.0) **S0892–1997** 30281-30282
20. 20.Dejonckere PH, Bradley P, Clemente P (2001) A basic protocol for functional assessment of voice pathology, especially for investigating the efficacy of (phonosurgical) treatments and evaluating new assessment techniques. Guideline elaborated by the Committee on Phoniatrics of the European Laryngological Society (ELS). Eur Arch Otorhinolaryngol 258(2):77–82
21. Hegazi M, Neumann K, Rosenfeld J, Zehnhoff-Dinnesen A, Wiskirska-Woznica B, Neumann K, Nawka T. **Prevention of developmental disorders of speech and language**. *Phoniatrics I* (2020.0) 713-724
22. Szkiełkowska A, Krasnodębska P, Miaśkiewicz B, Włodarczyk E, Domeracka-Kołodziej A, Skarżyński H. **Mucosal wave measurements in the diagnosis of functional dysphonia**. *Otolaryngol Pol* (2019.0) **73** 1-7. DOI: 10.5604/01.3001.0013.3215
23. Krasnodebska P, Domeracka-Kołodziej A, Szkiełkowska A, Mularzuk M, Sokołowska-Łazar D, Wilhelmsen K, Panasiewicz A. **Assessment of short-term comprehensive voice therapy in patients with unilateral paralysis of the larynx**. *Otolaryngol Pol* (2018.0) **72** 35-43. DOI: 10.5604/01.3001.0011.7250
24. Krasnodębska P, Szkiełkowska A, Czarkwiani-Woźniakowska L, Miaśkiewicz B, Sinkiewicz A, Skarżyński H. **Laryngeal electromyography in the therapeutic process of patients with vocal fold immobility or dysmobility**. *Life* (2022.0) **12** 390. DOI: 10.3390/life12030390
25. Carey B, Sulica L, Wu A, Branski R. **A novel electrodiagnostic assessment of the laryngeal closure reflex**. *Muscle Nerve* (2013.0) **47** 432-436. DOI: 10.1002/mus.23544
26. Goen A. **Classification of EMG signals for assessment of neuromuscular disorders**. *Int J Electron Electric Eng* (2014.0) **2** 242-248. DOI: 10.12720/ijeee.2.3.242-248
27. 27.Orsucci D. Ienco E.C. Nocita G et al (2020) Neurological features of COVID-19 and their treatment: a review. Drugs Context 9:1–12
28. Latremoliere A, Woolf C. **Central sensitization: a generator of pain hypersensitivity by central neural plasticity**. *J Pain* (2009.0) **10** 895-926. DOI: 10.1016/j.jpain.2009.06.012
29. Groneberg DA. **Increased expression of transient receptor potential vanilloid-1 in airway nerves of chronic cough**. *Am J Respir Crit Care Med* (2004.0) **170** 1276-1280. DOI: 10.1164/rccm.200402-174OC
30. Daia C, Scheau C, Neagu G, Andone I, Spanu A, Popescu C, Onose G. **Nerve conduction study and electromyography findings in patients recovering from Covid-19–Case report**. *Int J Infect Dis* (2021.0) **103** 420-422. DOI: 10.1016/j.ijid.2020.11.146
31. Rosenstock J, Tuchman M, LaMoreaux L, Sharma U. **Pregabalin for the treatment of painful diabetic peripheral neuropathy: a double-blind, placebo-controlled trial**. *Pain* (2004.0) **110** 628-638. DOI: 10.1016/j.pain.2004.05.001
|
---
title: 'On the potential of drug repurposing in dysphagia treatment: New insights
from a real-world pharmacovigilance study and a systematic review'
authors:
- Vera Battini
- Sara Rocca
- Greta Guarnieri
- Anna Bombelli
- Michele Gringeri
- Giulia Mosini
- Marco Pozzi
- Maria Nobile
- Sonia Radice
- Emilio Clementi
- Antonio Schindler
- Carla Carnovale
- Nicole Pizzorni
journal: Frontiers in Pharmacology
year: 2023
pmcid: PMC10022593
doi: 10.3389/fphar.2023.1057301
license: CC BY 4.0
---
# On the potential of drug repurposing in dysphagia treatment: New insights from a real-world pharmacovigilance study and a systematic review
## Abstract
Background: *Polypharmacy is* common in patients with dysphagia. Routinely used drugs may influence swallowing function either improving or worsening it. We aimed to explore the potential effects of three commonly used drug classes on dysphagia and aspiration pneumonia through a systematic review and a real-world data analysis to probe the possibility of drug repurposing for dysphagia treatment.
Material and Methods: Five electronic databases were searched. Studies on adults at risk for dysphagia, treated with Dipeptidyl-Peptidase IV Inhibitors (DPP-4i), Adrenergic Beta-Antagonists (beta-blockers), or Angiotensin-Converting Enzyme Inhibitors (ACEi), and reporting outcomes on dysphagia or aspiration pneumonia were included. A nested case/non-case study was performed on adverse events recorded in the FDA Adverse Event Reporting System (FAERS) on patients >64 years. Cases (dysphagia or aspiration pneumonia) were compared between patients only treated with Levodopa and patients who were concomitantly treated with the drugs of interest.
Results: Twenty studies were included in the review (17 on ACEi, 2 on beta-blockers, and 1 on DPP-4i). Contrasting findings on the effects of ACEi were found, with a protective effect mainly reported in Asian studies on neurological patients. Beta-blockers were associated with a reduced dysphagia rate. The study on DPP-4i suggested no effect on dysphagia and an increased risk of aspiration pneumonia. The FAERS analysis showed a reduction of the risk for dysphagia/aspiration pneumonia with ACEi, beta-blockers, and DPP-4i.
Conclusion: Our study explores the potential drug repurposing of ACEi, beta-blockers and DPP-4i in neurological patients with dysphagia to improve swallowing function and reduce aspiration pneumonia risk. Future randomized controlled studies should confirm these results and clarify the underlying mechanisms of action.
## 1 Introduction
Dysphagia is an impairment in the bolus transit from the mouth to the stomach (Merlo and Cohen, 1988). It may result from a variety of conditions such as neurological diseases, head and neck cancer, chronic respiratory disease, and aging. Its prevalence in the general population is $12.1\%$ (Kertscher et al., 2015), but it dramatically increases in high-risk populations such as patients with stroke (up to $80\%$), Parkinson’s disease (up to $81\%$), and community-acquired pneumonia ($91.7\%$) (Takizawa et al., 2016). Polypharmacy is therefore common in patients with dysphagia due to the symptoms that are associated with the underlying disease and the increasing number of comorbidities while aging, such as hypertension and diabetes mellitus (Miarons et al., 2016; Wolf et al., 2021). Unfortunately, detrimental effects on swallowing function have been reported for several drug classes commonly prescribed to the elderly patients, such as antidepressants, antipsychotics, benzodiazepines, antiepileptics, and drugs for dementia (Miarons et al., 2016; Dzahini et al., 2018; Wolf et al., 2021), thus exposing patients with dysphagia to an additional risk for pulmonary and nutritional complications.
Indeed, dysphagia is associated with severe complications, such as aspiration pneumonia—the leading cause of death in many neurodegenerative diseases (Lanska et al., 1988; Auyeung et al., 2012; Heemskerk and Roos, 2012), malnutrition, and dehydration: these conditions impact survival, clinical management, and health costs (Attrill et al., 2018; Marin et al., 2021). Thus, preventing dysphagia-related complications by early identification and treatment of dysphagia is of crucial importance. Mechanisms associated with the worsening of dysphagia are various and include xerostomia, drug-induced extrapyramidal symptoms, interactions with neural pathways involved in swallowing, and medicinal injury to the mucous membranes of the structures involved in swallowing. Conversely, potential beneficial effects on dysphagia have been suggested for some pharmacological agents. ACEi and Dipeptidyl-Peptidase IV Inhibitors (DPP-4i) have been reported to improve the swallowing reflex (Cunningham and O'Connor, 1997; Nakayama et al., 1998). Beta-blockers were found to be associated with lower dysphagia prevalence in the elderly patients (Miarons et al., 2018). Nevertheless, none of the studies clarified the mechanisms underlying the potential positive effects of these drugs on dysphagia, although a weak role of substance P (SP), a neuropeptide that enhances swallowing and cough reflexes, was hypothesized (Jin et al., 1994; Imoto et al., 2011; Canning et al., 2014). As many drugs have pleiotropic effects because of their interaction with multiple biological targets, known as primary and secondary effects (Jourdan et al., 2020; Hua et al., 2022), the process of finding new uses outside the original approved medical indication for existing drugs—i.e., redirecting, repurposing, repositioning and reprofiling (Kerber, 2003; Longman, 2004; Stuart, 2004), is increasing attention.
At present, dysphagia treatment relies on a variety of approaches including behavioral treatment, alternative feeding methods, neurostimulation techniques, surgical approaches, and pharmacological treatment. However, the pharmacological approach to dysphagia is relatively recent and still poorly investigated: a recent systematic review on the topic concluded that the number of randomized controlled trials (RCTs) for most of the pharmacological agents is very limited and the evidence of their efficacy is still scant (Cheng et al., 2022).
As a consequence of the lack of clinical trials, despite the intrinsic limitations, the use of alternative source (including pharmacovigilance databases) for retrieving potential effective additional uses of drugs has increased exponentially (Gatti et al., 2021; Ganesh and Randall, 2022).
The aim of this study is then to explore the potential effects of some drug classes on dysphagia and aspiration pneumonia through a systematic review and a real-world data analysis from the US Food and Drug Administration Adverse Event Reporting System (FAERS) database. In particular, we focused on three drug classes routinely used in clinical practice: DPP-4i, beta-blockers, and ACEi. These drug classes were selected because they are frequently prescribed in patients with dysphagia (Miarons et al., 2016; Wolf et al., 2021) and share a common secondary target, the aforementioned cleavage of SP. On the other hand, to target a population at high risk of dysphagia (Baijens et al., 2016; Takizawa et al., 2016), the real-data analysis was focused on the reports of patients aged >64 years and treated with Levodopa (i.e., subjects with Parkinson’s disease). The results of this exploratory study may help generate new hypotheses on the potential of drug repurposing for dysphagia treatment, to be verified in future randomized controlled trials.
## 2.1.1 Search strategy
We performed a systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021). We searched PubMed, Embase, CINHAL, Scopus, and the Cochrane electronic databases from inception up to 14 August 2021 with no language restriction. Our search strategy was adapted as necessary for each database and complete details of each search are described in Supplementary Table S1. Text words and database subject headings were used that were synonymous with the interventions and the outcomes of interest. Essentially, we used the following search terms.- Intervention: DPP-4i, Beta-blockers, ACEi, Neprilysin inhibitors (ACNi);- Outcome: dysphagia, aspiration pneumonia.
The terms related to the intervention and the outcome were combined with the Boolean operator “AND”. Additionally, the reference lists of the included studies and relevant reviews were checked for other potentially relevant studies. As no results for ACNi were found, this drug class will not be mentioned in the results and discussion sections.
## 2.1.2 Eligibility criteria
Study eligibility was based on inclusion and exclusion criteria regarding population, intervention, outcome, study design, and publication type. More specifically, inclusion criteria were: (i) studies on adult patients with any condition commonly associated with the onset of dysphagia (e.g., neurodegenerative diseases, stroke, head and neck diseases, and geriatric patients); (ii) studies including patients treated with any DPP-4i, Beta-blockers, ACEi or ACNi; (iii) studies reporting outcomes on swallowing function, aspiration pneumonia or SP concentrations; (iv) randomized (RCT), non-randomized clinical trials (nRCT), single-arm clinical trials (CTs), and observational studies. Literature reviews, case reports, and unpublished thesis were excluded.
## 2.1.3 Study selection
The records identified from the electronic search were imported into the software Rayyan (Ouzzani et al., 2016). After duplicate removal, our search results were screened by title and abstract for potentially eligible studies by two independent researchers (NP and VB). Potentially relevant studies were retrieved in full text and assessed for eligibility based on our prespecified inclusion criteria by two independent researchers (AB and SR). Reasons for the exclusion of full texts were recorded. Disagreements about eligibility were resolved by consensus.
## 2.1.4 Data extraction
Data from all included studies were extracted by two independent researchers (AB and SR) using pre-specified forms. Disagreements were resolved by consensus and consultation with the expert group (NP and VB). For each included study, the following information was extracted: first author, year of publication, study design, study duration, number of subjects, patient diagnoses, age (mean, median, range, standard deviation, interquartile range), sex distribution, generic name of the drug of interest, drug dose, concomitant therapies (concomitant drugs or swallowing rehabilitation), outcomes of interest reported in the study, assessment method for dysphagia, percentage of patients with dysphagia at the baseline and post-treatment, definition of aspiration pneumonia, percentage of patients with first event of recurrence of aspiration pneumonia, serum SP concentrations at the baseline and post-treatment, and main results. We did not contact authors for missing data.
## 2.1.5 Quality and risk of bias
Two authors (SR and GG) assessed the risk of bias of RCTs by using the Cochrane risk-of-bias tool for randomized trials (RoB2) (Sterne et al., 2019) and of nRCTs, single-arm CTs, and observational studies by using the risk of bias tool to assess non-randomized studies of interventions (ROBINS-I tool) (Sterne et al., 2016). For single-arm studies, a modified version of the ROBINS-I tool was used. Disagreements were resolved by consensus and consultation with the expert group (NP and VB).
## 2.2.1 Data source and extraction
Data were obtained from the FAERS, one of the largest and most comprehensive spontaneous reporting system databases. It contains information related to post-marketing safety surveillance reports in the form of adverse events (AEs) submitted by healthcare professionals, consumers, and other sources. AEs are recorded in the FAERS using the Medical Dictionary for Regulatory Activities (MedDRA®) preferred terms (PTs) (Fescharek et al., 2004), as Individual Case Safety Reports (ICSRs). Each ICSR provides administrative information (country, type of report, qualification of the reporter), patient demographics (sex, age, weight), AE characteristics (seriousness, date of onset, outcome), details about suspect drug therapy (drug name, exposure start and stop dates, time to onset, dose, route, indication, de-challenge and re-challenge) and information concerning any drug administered at the time of AE but not held responsible for its occurrence by the reporter, referred to as concomitant medication. However, the level of completeness of information varies from case to case (Sakaeda et al., 2013).
As the number of safety reports sent to the FDA annually is continuously growing, the database is largely used to detect novel drug-related safety events, to identify possible mechanisms of adverse events, to explore potential drug-drug interactions related to adverse events, and to discover promising new concomitant uses of drugs (Carnovale et al., 2019a; Mazhar et al., 2019; Mazhar et al., 2021).
Adverse events recorded in the FAERS were downloaded from the Food and Drug Administration (FDA) website (US FDA FAERS). The database consists of seven datasets, namely, patient demographic and administrative information (file descriptor DEMO), drug and biologic information (DRUG), adverse events (REAC), patient outcomes (OUTC), report sources (RPSR), start and end dates of drug therapy (THER), and indications for use/diagnosis (INDI). These seven datasets were joined by unique identification numbers for each FAERS report and a relational database was built. Data extraction was restricted to reports without missing values for age and gender; when more versions of the same ICSR were available, the last one was retained. Duplicate records were automatically detected and deleted by comparing the following information among the ICSRs: age, sex, event date, primary suspect, and country. Names of pharmaceutical drugs were harmonized by using the American RxTerms terminology (National Library of Medicine, 2020). The final cleaning process removed the list of “deleted cases” provided by the FDA and cases reported from the literature (US FDA FAERS).
This study was designed as a nested case/non-case study. The cohort was retrieved from the FAERS database in the period covering the first quarter of 2010 to the third quarter of 2021 and consisted of reports involving patients with more than 64 years (Baijens et al., 2016); in general, this population is known to be at risk for swallowing difficulties. Since the use and approval of drugs varies significantly between countries, we limited data extraction to Individual Case Safety Reports (ICSRs) from North America and Europe (except eastern countries). After a review of all LLTs in MedDRA (Fescharek et al., 2004), two terms were selected as relevant descriptors of the ADR of interest: Dysphagia and “Pneumonia Aspiration”. ICSRs reporting at least one of the LLT above mentioned were considered “cases” whilst “non-cases” were all the other ICSRs reporting other AEs.
## 2.2.2 Statistical analysis
Descriptive analysis was performed in terms of age, female sex, reporter type, country and the use of concomitant medications known to increase the risk of dysphagia (the list was retrieved from ClinicalKey (Brown, 2013; Clinical Pharmacology, 2020) and from Miarons et al. [ 2016]; see Supplementary Table S2). Between-group differences for the continuous variables were analyzed by the Mann-Whitney U test while categorical variables (sex, country, and the presence of concomitant medications) analyzed by Pearson’s Chi-square test. Tests were two-tailed, with significance set at a p-value of 0.05.
The crude (cROR) and adjusted reporting odds ratio (aROR) were calculated using univariate and multivariate logistic regression analysis respectively, and adjusted for potential confounding factors such as age class, gender, and concomitant drugs that are known to increase the risk of dysphagia. Since it is known that, among people aged over 64, patients with Parkinson’s are at a high risk of developing dysphagia (Chaudhuri et al., 2006; Takizawa et al., 2016; Wolf et al., 2021; Wang et al., 2022), we compared cases between those who were treated with Levodopa only and those who were concomitantly treated at least with a drug inhibiting the degradation of the SP: ACEi [WHO Anatomical Therapeutic Chemical (ATC) code: C09A]; beta-blockers [ATC code: C07A], Gliptins [ATC code: A10BH], ACNi (sacubitril) [ATC code: C09DX]. The reference group consisted of ICSRs where none of the above-mentioned drugs were reported. We assumed that reports involving only Levodopa would have an increased reporting risk compared to the general population aged over 64 and that the concomitant use of beta-blockers/ACEi/DPP-4i would reduce that risk. Signals of disproportionate reporting were detected when the number of reports was ≥3 and ROR—$95\%$ CI was greater than one.
Finally, since dysphagia is a condition that requires a specific clinical diagnosis, we planned a sensitivity analysis by using only ICSRs reported by physicians, in order to control for the potential confounding of this covariate. All analyses were performed using counts of unique cases.
Data reading, filtering, processing, and statistical analysis were performed through RStudio.
## 3.1.1 Search process
The study selection and screening process is presented in the PRISMA flowchart (Figure 1). The electronic search identified 4,728 records. After duplicate removal, 3,334 records were screened. One record was retrieved by manual search in the reference lists of relevant reviews and included studies for full-text analysis. In total, 183 full-text articles were assessed for eligibility. Ultimately, 20 studies met the eligibility criteria and were included in the review (Table 1). The effects of ACEi (Arai et al., 1998a; Arai et al., 1998b; Nakayama et al., 1998; Arai et al., 2001; Arai et al., 2003; Arai et al., 2005; Shimizu et al., 2008; Marciniak et al., 2009; Nakashima et al., 2011; Bosch et al., 2012; Liu et al., 2012; Matsumoto et al., 2012; Lee et al., 2015; Kano et al., 2016; Kumazawa et al., 2019; Fernandes et al., 2021) were analyzed in 17 studies whereas the effects of beta-blockers (Miarons et al., 2016; Miarons et al., 2018) and DPP-4i (Noguchi et al., 2020) were investigated in two and one studies, respectively.
**FIGURE 1:** *PRISMA 2020 flow diagram depicting the flow of information through the different phases of the Systematic review.* TABLE_PLACEHOLDER:TABLE 1
## 3.1.2 Characteristics of the included studies
Studies were published between 1998 and 2021. Most studies (Arai et al., 1998a; Arai et al., 1998b; Nakayama et al., 1998; Arai et al., 2001; Arai et al., 2003; Arai et al., 2005; Shimizu et al., 2008; Nakashima et al., 2011; Liu et al., 2012; Matsumoto et al., 2012; Sato et al., 2013; Lee et al., 2015; Kano et al., 2016; Kumazawa et al., 2019; Noguchi et al., 2020) were from Asian countries; three publications were from European countries (Bosch et al., 2012; Miarons et al., 2016; Miarons et al., 2018); and two publications were from American countries (Marciniak et al., 2009; Fernandes et al., 2021). Seventeen publications were articles in peer-reviewed journals and three were congress abstracts. Concerning study design, seven were retrospective observational studies, six were prospective observational studies, five were RCTs, and two were single-arm CTs. The sample size was <50 in four studies (Arai et al., 1998a; Nakayama et al., 1998; Shimizu et al., 2008; Kano et al., 2016), between 50 and 99 in five studies (Arai et al., 2003; Marciniak et al., 2009; Nakashima et al., 2011; Lee et al., 2015; Miarons et al., 2018) between 100 and 500 in five studies (Arai et al., 2001; Bosch et al., 2012; Matsumoto et al., 2012; Sato et al., 2013; Fernandes et al., 2021) and >500 in five studies (Arai et al., 1998b; Arai et al., 2005; Liu et al., 2012; Miarons et al., 2016; Kumazawa et al., 2019); in one study the overall sample size was not specified (Noguchi et al., 2020). The majority of the studies recruited neurological patients (10 studies on stroke (Arai et al., 1998a; Arai et al., 2001; Arai et al., 2003; Arai et al., 2005; Shimizu et al., 2008; Marciniak et al., 2009; Liu et al., 2012; Matsumoto et al., 2012; Lee et al., 2015; Kumazawa et al., 2019), one study on dementia (Bosch et al., 2012), and one on amyotrophic lateral sclerosis); in five studies participants had multiple etiologies (Nakayama et al., 1998; Nakashima et al., 2011; Miarons et al., 2016; Miarons et al., 2018; Fernandes et al., 2021). One study focused on diabetic (Noguchi et al., 2020) patients, one study included patients with head and neck cancer, and one study recruited patients with hypertension (Arai et al., 1998b). Mean age of recruited patients was >65 years for all the studies, with eight studies including only elderly patients (Arai et al., 1998a; Nakayama et al., 1998; Nakashima et al., 2011; Bosch et al., 2012; Miarons et al., 2016; Fernandes et al., 2021). With regard to the pharmacological intervention, the effects of ACEi (Arai et al., 1998a; Arai et al., 1998b; Nakayama et al., 1998; Arai et al., 2001; Arai et al., 2003; Arai et al., 2005; Shimizu et al., 2008; Marciniak et al., 2009; Nakashima et al., 2011; Bosch et al., 2012; Liu et al., 2012; Matsumoto et al., 2012; Sato et al., 2013; Lee et al., 2015; Kano et al., 2016; Kumazawa et al., 2019; Fernandes et al., 2021) were analyzed in 17 studies, whereas the effects of beta-blockers (Miarons et al., 2016; Miarons et al., 2018) and DPP4i (Noguchi et al., 2020) were investigated in two and one studies, respectively.
## 3.1.3 Methodological quality of the included studies
Supplementary Figure S3; Supplementary Table S4 show the results of the methodological assessment of included studies using the RoB2 and the ROBINS-I assessment tools. Risk of bias was on average high, with only one study classified as having low risk of bias (Nakashima et al., 2011). Concerning RCTs analyzed with ROB2, risk of bias was classified as low in one study (Nakashima et al., 2011) for two of the three outcomes of interest and high for the remaining outcome, moderate in one study (Lee et al., 2015), and high in three studies (Arai et al., 1998a; Nakayama et al., 1998; Kano et al., 2016). Most critical domains were related to the randomization process and the deviations from intended interventions. At the ROBINS-I, studies were classified as moderate risk of bias in 2 cases (Liu et al., 2012; Kumazawa et al., 2019), serious risk of bias in 2 (Shimizu et al., 2008; Miarons et al., 2016), and critical risk of bias in 5 cases (Marciniak et al., 2009; Bosch et al., 2012; Miarons et al., 2018; Noguchi et al., 2020; Fernandes et al., 2021); in six studies (Arai et al., 1998a; Arai et al., 1998b; Arai et al., 2001; Arai et al., 2005; Matsumoto et al., 2012; Sato et al., 2013) (congress abstracts or letters to the editor) there was not enough information to assess risk of bias. The most critical areas were biases due to deviations from the intended intervention, due to lack of information on, and biases in the classification of interventions.
## 3.1.4 Effect of pharmacological treatments on dysphagia
The effect of the drugs of interest on swallowing function was assessed in 11 studies (4RCTs, 2 CTs, one prospective observational study, and four retrospective observational studies) (Table 2). In particular, nine studies analyzed the effects of ACEi, two studied the effects of beta-blockers, and one studied the effects of DPP-4i. Dysphagia detection was based on instrumental assessment in five studies (Arai et al., 1998b; Nakayama et al., 1998; Arai et al., 2003; Shimizu et al., 2008; Matsumoto et al., 2012) clinical assessment in four studies (Lee et al., 2015; Miarons et al., 2016; Miarons et al., 2018; Fernandes et al., 2021), and was patient and/clinician-reported in two studies (Noguchi et al., 2020; Fernandes et al., 2021). Only four studies used validated scales for dysphagia (Lee et al., 2015; Miarons et al., 2016; Miarons et al., 2018; Fernandes et al., 2021).
**TABLE 2**
| Author, year | Study design | Drug of interest | Concomitant therapy | Outcome measures | Assessment methods | % Dysphagia baseline | % Dysphagia post-treatment | Results |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| ACEi | ACEi | ACEi | ACEi | ACEi | ACEi | ACEi | ACEi | ACEi |
| Arai et al. (1998b) | CT | Imidapril 5–10 mg/day (patients wtih symptomless dysphagia) | Simultaneous administration of other medications was allowed except for Levodopa | % silent aspiration | instrumental assessment | 100% | 62.5% | Silent aspiration improved in 10/16 patients after 12 weeks of ACEi |
| Arai et al. (1998b) | CT | Imidapril 5–10 mg/day (patients without dysphagia) | Simultaneous administration of other medications was allowed except for Levodopa | % silent aspiration | instrumental assessment | 0% | 0% | Silent aspiration improved in 10/16 patients after 12 weeks of ACEi |
| Arai et al. (1998b) | CT | Control (healthy) | Simultaneous administration of other medications was allowed except for Levodopa | % silent aspiration | instrumental assessment | 0% | 0% | Silent aspiration improved in 10/16 patients after 12 weeks of ACEi |
| Nakayama et al. (1998) | RCT | Imidapril 5 mg/day | | Duration of swallowing reflex delay | instrumental assessment (sEMG) | 100% | | The latency of response of the swallowing reflex did not differ between placebo and imidapril in the healthy volunteers (baseline 1.4 ± 0.2 s vs. after treatment 1.2 ± 0.1 s). In patients with aspiration pneumonia, ACEi significantly improved the latency of response compared with placebo (baseline 6.3 ± 1.1 s vs. after treatment 2.7 ± 0.3 s) |
| Nakayama et al. (1998) | RCT | Control (placebo) | | Duration of swallowing reflex delay | instrumental assessment (sEMG) | 100% | | The latency of response of the swallowing reflex did not differ between placebo and imidapril in the healthy volunteers (baseline 1.4 ± 0.2 s vs. after treatment 1.2 ± 0.1 s). In patients with aspiration pneumonia, ACEi significantly improved the latency of response compared with placebo (baseline 6.3 ± 1.1 s vs. after treatment 2.7 ± 0.3 s) |
| Arai et al. (2003) | RCT | Imidapril 1.25 mg/day | | % silent aspiration | instrumental assessment | 100% | 26.2% | Silent aspiration disappeared in the majority of the patients (31/42) in the treatment group (12 weeks of ACEi), whereas only in 1/12 patient in the control group. Response to treatment (no silent aspiration post-treatment) was proportional to the ACEi dose: 50% for 0.25 mg/day, 73% for 0.5 mg/day, 77% for 0.625 mg/day, 83% for 1.25 mg/day |
| Arai et al. (2003) | RCT | Imidapril 0.625 mg/day | | % silent aspiration | instrumental assessment | 100% | 26.2% | Silent aspiration disappeared in the majority of the patients (31/42) in the treatment group (12 weeks of ACEi), whereas only in 1/12 patient in the control group. Response to treatment (no silent aspiration post-treatment) was proportional to the ACEi dose: 50% for 0.25 mg/day, 73% for 0.5 mg/day, 77% for 0.625 mg/day, 83% for 1.25 mg/day |
| Arai et al. (2003) | RCT | Imidapril 0.5 mg/day | | % silent aspiration | instrumental assessment | 100% | 26.2% | Silent aspiration disappeared in the majority of the patients (31/42) in the treatment group (12 weeks of ACEi), whereas only in 1/12 patient in the control group. Response to treatment (no silent aspiration post-treatment) was proportional to the ACEi dose: 50% for 0.25 mg/day, 73% for 0.5 mg/day, 77% for 0.625 mg/day, 83% for 1.25 mg/day |
| Arai et al. (2003) | RCT | Imidapril 0.25 mg/day | | % silent aspiration | instrumental assessment | 100% | 26.2% | Silent aspiration disappeared in the majority of the patients (31/42) in the treatment group (12 weeks of ACEi), whereas only in 1/12 patient in the control group. Response to treatment (no silent aspiration post-treatment) was proportional to the ACEi dose: 50% for 0.25 mg/day, 73% for 0.5 mg/day, 77% for 0.625 mg/day, 83% for 1.25 mg/day |
| Arai et al. (2003) | RCT | Control (no treatment) | | % silent aspiration | instrumental assessment | 100% | 91.7% | Silent aspiration disappeared in the majority of the patients (31/42) in the treatment group (12 weeks of ACEi), whereas only in 1/12 patient in the control group. Response to treatment (no silent aspiration post-treatment) was proportional to the ACEi dose: 50% for 0.25 mg/day, 73% for 0.5 mg/day, 77% for 0.625 mg/day, 83% for 1.25 mg/day |
| Shimizu et al. (2008) | CT | Imidapril 5 mg/day | | Pharyngeal transit time | instrumental assessment (VFSS) | 100% | | The ACE inhibitor decreased the pharyngeal transit time in all 10 patients, resulting in a significant reduction of the pharyngeal transit time after 6-week treatment with the ACE inhibitor (baseline 2.5 ± 0.3 s, after treatment 1.6 ± 0.2 s, p < 0.01). The abnormalities in the oral and esophageal stages were not altered by treatment with the ACE inhibitor |
| Nakashima et al. (2011) | RCT | Imidapril 5 mg/day | Statins, angiotensin receptor blockers, calcium channel blockers, L-dopa and amantadine | % delayed swallowing reflex | clinical assessment | 100% | 73.3% | Swallowing reflex delay improved in 15/30 (50%) of patients taking imidapril and 19/30 (63.3%) patients taking nicergoline. There was no significant difference in the overall proportion of patients who showed improvements in dysphagia with imidapril compared to nicergoline |
| Nakashima et al. (2011) | RCT | Control (Nicergoline 15 mg/day) | Statins, angiotensin receptor blockers, calcium channel blockers, L-dopa and amantadine | % delayed swallowing reflex | clinical assessment | 100% | 60% | Swallowing reflex delay improved in 15/30 (50%) of patients taking imidapril and 19/30 (63.3%) patients taking nicergoline. There was no significant difference in the overall proportion of patients who showed improvements in dysphagia with imidapril compared to nicergoline |
| Matsumoto et al. (2012) | pOBS | Perindopril 2–4 mg/day | Hospital-based conventional rehabilitation program and home-based exercise program | Dysphagia | instrumental assessment (VFSS) | | | Swallowing function was improved in parallel with the increase of motor outcomes |
| Lee et al. (2015) | RCT | Lisinopril 2.5 mg/day | | RBHOMS score | clinical assessment | 100% | | At baseline, swallowing function did not differ between the two groups (RBHOMS mean score treatment 3.7 ± 0.8 vs. placebo 4.2 ± 1.5, p = 0.462). At week 12, there was a better swallowing function in patients treated with ACEi than placebo (RBHOMS mean score treatment 4.2 ± 1.5 vs. placebo 3.5 ± 1.5, p = 0.053) |
| Lee et al. (2015) | RCT | Control (placebo) | | RBHOMS score | clinical assessment | 100% | | At baseline, swallowing function did not differ between the two groups (RBHOMS mean score treatment 3.7 ± 0.8 vs. placebo 4.2 ± 1.5, p = 0.462). At week 12, there was a better swallowing function in patients treated with ACEi than placebo (RBHOMS mean score treatment 4.2 ± 1.5 vs. placebo 3.5 ± 1.5, p = 0.053) |
| Fernandes et al. (2021) | rOBS | ACEi | Anticholinergics, antimuscarinics, antihistamines, antidepressants, antipsychotics, opioids, L-dopa, adrenergics, thyroid hormones, cholinergic drugs, diuretics, sedatives, NSAIDs, corticosteroids, antiulcerogenics, antihypertensive, antidiabetics, antiadrenergics | EAT-10 | patient-reported | | | ACEi users complained more about symptoms of dysphagia. Scores were significantly higher on the EAT-10 than non-users (ACEi users mean EAT-10 = 1.7, non-users mean EAT-10 = 0.7, p = 0.038) |
| ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers | ACEi and Beta-blockers |
| Miarons et al. (2016) | rOBS | ACEi | Selective calcium channel blockers, anti-inflammatory and antirheumatic products, non-steroids, antipsychotics, antidepressants, drugs against dementia | % dysphagia at the VVST | clinical assessment | | | ACEis were not associated with potential beneficial actions on swallowing (OR 0.68, 95%CI 0.46–1.02, p = 0.060). Beta-blockers were independently associated with a reduced risk of dysphagia (OR 0.60, 95%CI 0.38–0.95, p = 0.030) |
| Miarons et al. (2016) | rOBS | Beta-blockers | Selective calcium channel blockers, anti-inflammatory and antirheumatic products, non-steroids, antipsychotics, antidepressants, drugs against dementia | % dysphagia at the VVST | clinical assessment | | 10.4% | ACEis were not associated with potential beneficial actions on swallowing (OR 0.68, 95%CI 0.46–1.02, p = 0.060). Beta-blockers were independently associated with a reduced risk of dysphagia (OR 0.60, 95%CI 0.38–0.95, p = 0.030) |
| Miarons et al. (2016) | rOBS | Oral antidiabetics | Selective calcium channel blockers, anti-inflammatory and antirheumatic products, non-steroids, antipsychotics, antidepressants, drugs against dementia | % dysphagia at the VVST | clinical assessment | | 15.1% | ACEis were not associated with potential beneficial actions on swallowing (OR 0.68, 95%CI 0.46–1.02, p = 0.060). Beta-blockers were independently associated with a reduced risk of dysphagia (OR 0.60, 95%CI 0.38–0.95, p = 0.030) |
| Beta-blockers | Beta-blockers | Beta-blockers | Beta-blockers | Beta-blockers | Beta-blockers | Beta-blockers | Beta-blockers | Beta-blockers |
| Miarons et al. (2018) | rOBS | Beta-blockers | Drugs for alimentary tract and metabolism, drugs for blood and blood-forming organs, drugs for the cardiovascular system, drugs for genitourinary system and sex hormones, systemic hormonal preparations (excluding sex hormones), anti-infectives for systemic use, drugs for muscoloskeletal system, drugs for nervous system, drugs for respiratory system | % dysphagia at the VVST | clinical assessment | | 32.1% | Patients taking beta-blockers had a significantly lower frequency of dysphagia than patients not taking beta-blockers |
| Miarons et al. (2018) | rOBS | Control (no beta-blockers) | Drugs for alimentary tract and metabolism, drugs for blood and blood-forming organs, drugs for the cardiovascular system, drugs for genitourinary system and sex hormones, systemic hormonal preparations (excluding sex hormones), anti-infectives for systemic use, drugs for muscoloskeletal system, drugs for nervous system, drugs for respiratory system | % dysphagia at the VVST | clinical assessment | | 67.9% | Patients taking beta-blockers had a significantly lower frequency of dysphagia than patients not taking beta-blockers |
| DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i |
| Noguchi et al. (2020) | rOBS | DPP-4i | Sulfonylurea, glinide, biguanide, thiazolidinedione, a-glucosidase inhibitors, glucagon-like peptide-1 receptor agonist, sodium glucose cotransporter-2 inhibition | dysphagia as adverse event | patient or physician reported | | | 7 cases of DPP-4i induced dysphagia were found. No significant association between DPP4i use and dysphagia was detected (ROR: 0.82, 95%CI: 0.39–1.73) |
ACEi was the main drug class studied. Six studies reported an improvement of dysphagia in the majority of the patients treated with ACEi (Arai et al., 1998a; Nakayama et al., 1998; Arai et al., 2003; Shimizu et al., 2008; Nakashima et al., 2011; Matsumoto et al., 2012; Lee et al., 2015). Additionally, Arai and others (Arai et al., 2003) compared the efficacy of ACEi at different doses and detected a dose-response relationship with the improvement of dysphagia. Improvements in swallowing function targeted the pharyngeal phase of swallowing and included a reduction of the rate of silent aspiration (Arai et al., 1998b; Arai et al., 2003), pharyngeal transit time (Shimizu et al., 2008), and swallowing reflex delay (Nakayama et al., 1998; Nakashima et al., 2011). On the contrary, one study on elderly patients with different etiology failed to detect any association between ACEi use and dysphagia (Miarons et al., 2016), and another study, on a similar population, found that ACEi users complained of more dysphagia symptoms than non-users (Fernandes et al., 2021). However, the mean score on the self-reported Eating Assessment Tool (EAT-10) (Belafsky et al., 2008), used to assess dysphagia in the latter study, was lower than the cut-off for dysphagia (EAT-10 ≥ 3) in both groups.
Two retrospective observational studies by Miarons and others (Miarons et al., 2016; Miarons et al., 2018) reported an independent association between beta-blocker use and reduced risk of dysphagia. This association was clinically detected using a validated clinical assessment protocol (Clavé et al., 2008) in elderly patients with different neurological and non-neurological diseases. However, no data on the frequency of dysphagia before the pharmacological treatment is available because of the retrospective nature of the studies.
Only one study investigated the effects of DPP-4i on swallowing function (Noguchi et al., 2020). Based on the adverse events from a spontaneous report system, Noguchi and others found no association between dysphagia and the use of DPP-4i in diabetic patients.
## 3.1.5 Effect of pharmacological treatments on aspiration pneumonia
The effect of the drugs of interest on aspiration pneumonia was assessed in 12 studies (2RCTs, four prospective observational studies, and six retrospective observational studies) (Table 3). In particular, 11 studies analyzed the effects of ACEi and one study analyzed the effects of DPP-4i. Four studies focused only on aspiration pneumonia (Marciniak et al., 2009; Bosch et al., 2012; Lee et al., 2015; Kumazawa et al., 2019), while the remaining studies focused on pneumonia in general, including aspiration pneumonia, in patients with documented dysphagia or who were at a high risk of dysphagia.
**TABLE 3**
| Author, year | Study design | Drug of interest | Concomitant therapy | Definition of aspiration pneumonia | % 1st aspiration pneumonia | % Recurrent aspiration pneumonia | Results |
| --- | --- | --- | --- | --- | --- | --- | --- |
| ACEi | ACEi | ACEi | ACEi | ACEi | ACEi | ACEi | ACEi |
| Arai et al. (1998a) | pOBS | Imidapril hydrochloride | | | 3.3% | | The rate of pneumonia in the ACEi group was significantly lower than in the calcium-channel blocker group (p = 0.025) |
| Arai et al. (1998a) | pOBS | Calcium-channel blocker | | | 8.9% | | The rate of pneumonia in the ACEi group was significantly lower than in the calcium-channel blocker group (p = 0.025) |
| Arai et al. (1998a) | pOBS | Control (non-hypertensive patients) | | | 8.3% | | The rate of pneumonia in the ACEi group was significantly lower than in the calcium-channel blocker group (p = 0.025) |
| Arai et al. (2001) | pOBS | ACEi | | | 4.4% | | The incidence of pneumonia in the ACEi group was significantly lower than the ARB group (p = 0.013) |
| Arai et al. (2001) | pOBS | ARB | | | 11.2% | | The incidence of pneumonia in the ACEi group was significantly lower than the ARB group (p = 0.013) |
| Arai et al. (2005) | pOBS | ACEi | | | 2.8% | | Patients treated with ACEi had a lower risk of pneumonia than controls (hazard ratio 0.30, 95% CI 0.14–0.66, p = 0.001) |
| Arai et al. (2005) | pOBS | Calcium-channel blocker | | | 8.8% | | Patients treated with ACEi had a lower risk of pneumonia than controls (hazard ratio 0.30, 95% CI 0.14–0.66, p = 0.001) |
| Arai et al. (2005) | pOBS | Diuretics | | | 8.3% | | Patients treated with ACEi had a lower risk of pneumonia than controls (hazard ratio 0.30, 95% CI 0.14–0.66, p = 0.001) |
| Arai et al. (2005) | pOBS | Control (no antihypertensive drugs) | | | 8.8% | | Patients treated with ACEi had a lower risk of pneumonia than controls (hazard ratio 0.30, 95% CI 0.14–0.66, p = 0.001) |
| Marciniak et al. (2009) | rOBS | ACEi | PPI, H2 blockers | Clinical setting of fever, chills, muscle stiffness, chest pain, cough, shortness of breath, rapid heart rate, or difficulty breathing, with chest x-ray confirmation either in rehabilitation or after transfer to acute care | | | Use of ACE inhibitors was similar for both stroke patients with pneumonia and matched-controls; ACE inhibitors did not confer any decreased risk of pneumonia (OR, 0.9; 95% CI, 0.2–3.0) |
| Nakashima et al. (2011) | RCT | Imidapril 5 mg/day | Statins, angiotensin receptor blocker, calcium channel blocker, L-dopa and amantadine | Pneumonia was diagnosed based on the Japanese Respiratory Society guidelines | Inclusion criteria | 30% | No significant difference in the pneumonia recurrence rate was found between the imidapril and the nicergoline groups |
| Nakashima et al. (2011) | RCT | Nicergoline 15 mg/day | Statins, angiotensin receptor blocker, calcium channel blocker, L-dopa and amantadine | Pneumonia was diagnosed based on the Japanese Respiratory Society guidelines | Inclusion criteria | 16.7% | No significant difference in the pneumonia recurrence rate was found between the imidapril and the nicergoline groups |
| Bosch et al. (2012) | pOBS | ACEi | Psychotropic drugs, histamine receptor blocker or PPI treatment in the mont previous to admission, neuroleptics, SSRI, antibiotics before hospitalization | Infiltrate on chest radiography consistent with pneumonia and one major | | | Patients with recurrent aspiration pneumonia were less-frequently prescribed ACEi compared with a first episode of aspiration pneumonia (8.8% vs. 27.9%, p < 0.001) |
| Bosch et al. (2012) | pOBS | ACEi | Psychotropic drugs, histamine receptor blocker or PPI treatment in the mont previous to admission, neuroleptics, SSRI, antibiotics before hospitalization | symptom or sign (cough, sputum production or temperature above 37.8°C) or two minor criteria (dyspnea, pleuritic chest pain, delirium, respiratory rate>20 bpm, signs of pulmonary consolidation, or leukocyte count>12 × 109/L). In addition, to meet our definition of AP, all patients had to have risk factors for oropharyngeal aspiration and a history of witnessed or suspected aspiration | | | Patients with recurrent aspiration pneumonia were less-frequently prescribed ACEi compared with a first episode of aspiration pneumonia (8.8% vs. 27.9%, p < 0.001) |
| Liu et al. (2012) | rOBS | ACEi | Statins, PPI, histamine type 2 receptor antagonists | ICD9-CM: 507 pneumonitis due to solids and liquids, 481 Pneumococcal pneumonia | Inclusion criteria | | ACEi use was associated with a lower pneumonia risk (OR 0.77; 95%CI 0.68–0.87). An increased mean defined daily dose (DDD) was associated with significantly reduced pneumonia risk (DDD>1 mg daily) |
| Liu et al. (2012) | | | | 482 Other bacterial pneumonia | | | |
| Liu et al. (2012) | | | | 483 Pneumonia due to other specified organism, 485 Bronchopneumonia, organism unspecified | | | |
| Liu et al. (2012) | | | | 486 Pneumonia, organism unspecified | | | |
| Matsumoto et al. (2012) | pOBS | Perindopril 2–4 mg/day | Hospital-based conventional rehabilitation program and home-based exercise program | | 2.9% | | 6 events of pneumonia were recorded over a 5-year period |
| Sato et al. (2013) | rOBS | ACE-i | Chemoradiotherapy (cisplatin, carboplatin, docetaxel) | | 0% | | Patients with ACEi had a lower rate of aspiration pneumonia compared to patients with other antihypertensive drugs or patients without antihypertensive drugs (0% vs. 17.8% vs. 12.5%; no statistical analysis was performed) |
| Sato et al. (2013) | rOBS | Antihypertensive drugs other than ACE-i | Chemoradiotherapy (cisplatin, carboplatin, docetaxel) | | 17.8% | | Patients with ACEi had a lower rate of aspiration pneumonia compared to patients with other antihypertensive drugs or patients without antihypertensive drugs (0% vs. 17.8% vs. 12.5%; no statistical analysis was performed) |
| Sato et al. (2013) | rOBS | No antihypertensive drugs | Chemoradiotherapy (cisplatin, carboplatin, docetaxel) | | 12.5% | | Patients with ACEi had a lower rate of aspiration pneumonia compared to patients with other antihypertensive drugs or patients without antihypertensive drugs (0% vs. 17.8% vs. 12.5%; no statistical analysis was performed) |
| Lee et al. (2015) | RCT | Lisinopril 2.5 mg/day | | Presence of new pneumonic changes in the chest x-ray (done in index admission used for References) and 1 major clinical sign: increased | 57.6% | | The incidences of pneumonia and fatal pneumonia (pneumonia-related death) were not significantly different between the groups (pneumonia: treatment 57.6% vs. placebo 47.4%, p = 0.390; fatal pneumonia: treatment 42.4% vs. placebo 26.3%, p = 0.152) |
| Lee et al. (2015) | RCT | Control (placebo) | | sputum production or 2 of the following minor clinical signs: raised or depressed white cell count, hypoxia at room air (SpO2 <92%) and tympanic temperature greater than 38 °C | 47.4% | | The incidences of pneumonia and fatal pneumonia (pneumonia-related death) were not significantly different between the groups (pneumonia: treatment 57.6% vs. placebo 47.4%, p = 0.390; fatal pneumonia: treatment 42.4% vs. placebo 26.3%, p = 0.152) |
| Kumazawa et al. (2019) | rOBS | ACEi | Antipsychotic, anti-dementia drugs, antiemetics, antiepileptics, antitussive drugs, muscle relaxants, antidiabetics, steroids, immunosuppressive drugs, gastric secretion inhibitors, antidyslipidemics, antithrombotics, antihypertensive, diuretics, amantadine, nicergoline, severe pneumonia antibiotics | ICD-10 codes: Aspiration Pneumonia (J69) and bacterial pneumonia (J13-J18) | Inclusion criteria | 0.8% at 14 days 1.3% at 30 days 2.6% at 90 days | Non-significant difference was seen in 14-day, 30-day, or 90-day post-stroke readmission for aspiration pneumonia between patients on ACEi and patients on ARB. |
| Kumazawa et al. (2019) | rOBS | ARB | Antipsychotic, anti-dementia drugs, antiemetics, antiepileptics, antitussive drugs, muscle relaxants, antidiabetics, steroids, immunosuppressive drugs, gastric secretion inhibitors, antidyslipidemics, antithrombotics, antihypertensive, diuretics, amantadine, nicergoline, severe pneumonia antibiotics | ICD-10 codes: Aspiration Pneumonia (J69) and bacterial pneumonia (J13-J18) | Inclusion criteria | 0.7% at 14 days 1.3% at 30 days 2.4% at 90 days | Non-significant difference was seen in 14-day, 30-day, or 90-day post-stroke readmission for aspiration pneumonia between patients on ACEi and patients on ARB. |
| DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i | DPP-4i |
| Noguchi et al. (2020) | rOBS | DPP-4i | Sulfonylurea, glinide, biguanide, thiazolidinedione, a-glucosidase inhibitors, glucagon-like peptide-1 receptor agonist, sodium glucose cotransporter-2 inhibition | | | | 35 cases of DPP-4i induced aspiration pneumonia were found. DPP-4i use was significantly associated with an increased risk of aspiration pneumonia (ROR 1.67, 95%CI: 1.20–2.34). When DPP-4is were analyzed individually, a significant association with aspiration pneumonia risk was detected for trelagliptin (ROR 9.99, 95%CI: 4.10–24.36), linagliptin (ROR 2.66, 95% CI: 1.19–5.94) and sitagliptin (ROR 1.84, 95% CI: 1.04–3.25) |
Six studies reported results indicating the beneficial effects of ACEi on aspiration pneumonia risk (Arai et al., 1998a; Arai et al., 2001; Arai et al., 2005; Bosch et al., 2012; Liu et al., 2012; Sato et al., 2013). ACEi use was associated with a lower rate of pneumonia compared to no treatment (Arai et al., 2005; Sato et al., 2013) or to other hypertensive drugs (Angiotensin II receptor blockers, Calcium-channel blockers). One study suggested a dose-response relationship in reducing pneumonia risk (Liu et al., 2012). Additionally, one study found a protective effect of ACEi on the recurrence of aspiration pneumonia (Bosch et al., 2012). Conversely, four studies failed to detect an association between ACEi use and reduced aspiration pneumonia risk (Nakashima et al., 2011; Liu et al., 2012; Lee et al., 2015; Kumazawa et al., 2019). In these studies, the incidence of the first pneumonia event or pneumonia recurrence did not differ between patients treated with ACEi and patients treated with Angiotensin II receptor blockers (Kumazawa et al., 2019), nicergoline (Nakashima et al., 2011), other drugs or placebo (Lee et al., 2015).
The only study investigating the effects of DPP-4i on aspiration pneumonia based on a spontaneous report system of adverse events found an increased risk of aspiration pneumonia in diabetic patients treated with DPP-4i (Noguchi et al., 2020).
## 3.1.6 Effect of pharmacological treatments on substance P concentration
The effects of the drugs of interest on SP concentration was assessed in five of the included studies (3RCTs, and one retrospective observational study) (Table 4). Among them, four studies analyzed the efficacy of ACEi, and one study focused on beta-blockers. All the studies on ACEi reported a significant increase in mean serum SP levels in patients treated with ACEi (Arai et al., 1998b; Arai et al., 2003; Nakashima et al., 2011; Kano et al., 2016). Improvements in serum SP concentrations have been associated with improvements in swallowing safety (Arai et al., 1998a; Arai et al., 2003; Nakashima et al., 2011). However, this correlation may not always be straightforward. A minority of the patients who did not improve their swallowing function were found to have increased SP concentrations and, on the contrary, some of the patients who did improve their swallowing function did not record an increase in SP (Arai et al., 1998b; Arai et al., 2003).
**TABLE 4**
| Author, year | Study design | Drug of interest | Concomitant therapy | Serum SP baseline (pg/mL) | Serum SP post-treatment (pg/mL) | Results |
| --- | --- | --- | --- | --- | --- | --- |
| Author, year | Study design | Drug of interest | Concomitant therapy | Mean ± SD | Mean ± SD | Results |
| Studies on ACEi | Studies on ACEi | Studies on ACEi | Studies on ACEi | Studies on ACEi | Studies on ACEi | Studies on ACEi |
| Arai et al. (1998b) | CT | Imidapril 5–10 mg/day (patients with symptomless dysphagia) | Simultaneous administration of other meds was allowed except for L-dopa | 23.3 | | Serum SP concentrations at the baseline were lower in patients with dysphagia than the other groups. After treatment, serum SP concentrations increased in 8/10 patients who improved swallowing function (post-treatment mean 79.3 pg/mL, baseline mean 23.3 pg/mL), whereas did not change in the remaining 2/10 responders. Among patients who did not improve swallowing function, 3/5 showed an increase in SP concentration (mean 82.1 pg/mL), while 2/5 had no change |
| Arai et al. (1998b) | CT | Imidapril 5–10 mg/day (patients without dysphagia) | Simultaneous administration of other meds was allowed except for L-dopa | 76.5 | | Serum SP did not change from baseline |
| Arai et al. (1998b) | CT | Control (healthy) | Simultaneous administration of other meds was allowed except for L-dopa | 72.7 | | |
| Arai et al. (2003) | RCT | Imidapril | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | 1.25 mg/day | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | Imidapril | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | 0.625 mg/day | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | Imidapril | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | 0.5 mg/day | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | Imidapril | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | 0.25 mg/day | | 26.0 ± 1.7 (patients with swallowing improvement) 26.4 ± 1.1 (patients with no swallowing improvement) | 68.8 ± 6.0 (patients with swallowing improvement) 45.4 ± 8.6 (patients with no swallowing improvement) | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Arai et al. (2003) | RCT | Control (no treatment) | | 26.4 ± 1.0 | 25.3 ± 1.1 | Serum SP levels significantly increased patients treated with ACEi, regardless of the improvement in silent aspiration, whereas did not increase in the control group. Serum SP levels at the end of the study were significantly higher: (i) in patients treated with ACEi who improved swallowing function than in patients treated with ACEi who did not improve swallowing function; (ii) in patients treated with ACEi than controls |
| Nakashima et al. (2011) | RCT | Imidapril | Statins, angiotensin receptor blocker, calcium channel blocker, L-dopa and amantadine | | | Both imidapril and nicergoline significantly increased serum levels of SP, with no significant differences among the two groups. Patients whose dysphagia was improved showed significantly increased serum levels of SP. By contrast, the patients whose dysphagia failed to improve did not show significant increases in serum levels of SP. |
| Nakashima et al. (2011) | RCT | 5 mg/day | Statins, angiotensin receptor blocker, calcium channel blocker, L-dopa and amantadine | | | Both imidapril and nicergoline significantly increased serum levels of SP, with no significant differences among the two groups. Patients whose dysphagia was improved showed significantly increased serum levels of SP. By contrast, the patients whose dysphagia failed to improve did not show significant increases in serum levels of SP. |
| Nakashima et al. (2011) | RCT | Nicergoline | Statins, angiotensin receptor blocker, calcium channel blocker, L-dopa and amantadine | | | Both imidapril and nicergoline significantly increased serum levels of SP, with no significant differences among the two groups. Patients whose dysphagia was improved showed significantly increased serum levels of SP. By contrast, the patients whose dysphagia failed to improve did not show significant increases in serum levels of SP. |
| Nakashima et al. (2011) | RCT | 15 mg/day | Statins, angiotensin receptor blocker, calcium channel blocker, L-dopa and amantadine | | | Both imidapril and nicergoline significantly increased serum levels of SP, with no significant differences among the two groups. Patients whose dysphagia was improved showed significantly increased serum levels of SP. By contrast, the patients whose dysphagia failed to improve did not show significant increases in serum levels of SP. |
| Kano et al. (2016) | RCT | Enalapril | | | | At 3 months, sputum SP concentration increased in ALS patients treated with enalapril |
| Kano et al. (2016) | RCT | 5 mg/day | | | | At 3 months, sputum SP concentration increased in ALS patients treated with enalapril |
| Kano et al. (2016) | RCT | Control (no treatment) | | | | At 3 months, sputum SP concentration increased in ALS patients treated with enalapril |
| Studies on beta-blockers | Studies on beta-blockers | Studies on beta-blockers | Studies on beta-blockers | Studies on beta-blockers | Studies on beta-blockers | Studies on beta-blockers |
| Miarons et al. (2018) | rOBS | Beta-blockers | Drugs for alimentary tract and metabolism, blood and blood-forming organs, cardiovascular system, genitourinary system and sex hormones, systemic hormonal preparations, musculoskeletal system, nervous system, respiratory system | | 260.68 ± 144.27 | SP serum levels were significantly higher in patients taking beta-blockers than in patients not taking beta-blockers |
| Miarons et al. (2018) | rOBS | Control (no beta-blockers) | Drugs for alimentary tract and metabolism, blood and blood-forming organs, cardiovascular system, genitourinary system and sex hormones, systemic hormonal preparations, musculoskeletal system, nervous system, respiratory system | | 175.46 ± 108.36 | SP serum levels were significantly higher in patients taking beta-blockers than in patients not taking beta-blockers |
Miarons and others detected significantly higher serum and saliva SP levels in adult non-neurological patients taking beta-blockers compared to patients not taking beta-blockers, matched for age, sex, and independence level (Miarons et al., 2018). However, due to the cross-sectional design of the study, no association between treatment with beta-blockers and improvement in SP concentrations can be determined.
## 3.1.7 Adverse events
Four studies reported adverse events related to the treatment with ACEi. Three studies reported excessive cough in $5.3\%$–$6.3\%$ of the patients (Arai et al., 1998a; Arai et al., 2003; Kano et al., 2016), in one study pneumonia onset was associated with the treatment with ACEi in $6.3\%$ of the patients (Arai et al., 2003), and one study reported dizziness or hypotension in $5.3\%$ of the patients (Kano et al., 2016). Regarding serious adverse events, one RCT was prematurely terminated due to significantly higher mortality in the intervention group at interim analysis (Lee et al., 2015).
## 3.2 Pharmacovigilance study
From the FAERS we identified 1,742,491 ICSRs involved elderly subjects. 1,453,966 ICSRs came from North America and Europe; of these, 12,302 ICSRs ($0.8\%$) were related to dysphagia/aspiration pneumonia. The descriptive analysis of demographic data and the characteristics of nested-cases and non-cases populations is presented in Table 5. The median age was significantly different between cases and non-cases ($p \leq 0.05$); however, most of the patients were 70–80 years old (25th–75th percentiles, 70–82 vs. 69–80 for cases and non-cases, respectively). In both groups, the percentage of female reports were >$50\%$, $p \leq 0.05.$ Concomitant medications associated with dysphagia were used in $35\%$ of cases and in $22\%$ of non-cases ($p \leq 0.05$). Dysphagia was mostly reported by consumers ($42\%$). ICSRs were mainly from North America in both cases and non-cases ($71\%$–$75\%$).
**TABLE 5**
| Unnamed: 0 | Cases (Dysphagia/Aspiration pneumonia) | Non-cases (other AEs) |
| --- | --- | --- |
| | (n = 12,302) | (n = 1,441,664) |
| Age, yrs | Age, yrs | Age, yrs |
| Mean (SD) | 76 (8) | 75 (7) |
| Median (25th-75th percentiles) a | 75 (70–82) | 74 (69–80) |
| Age classes n (%) b | Age classes n (%) b | Age classes n (%) b |
| 65–75 | 6,415 (52) | 836,537 (58) |
| 76–85 | 4,217 (34) | 465,872 (32) |
| 86–95 | 1,560 (13) | 132,800 (9) |
| >95 | 110 (0.9) | 6,455 (0.4) |
| Gender, n (%) b | Gender, n (%) b | Gender, n (%) b |
| Females | 6,300 (51) | 798,178 (55) |
| Concomitant drugs known for risk of Dysphagia/Aspiration Pneumonia, n (%) b | Concomitant drugs known for risk of Dysphagia/Aspiration Pneumonia, n (%) b | Concomitant drugs known for risk of Dysphagia/Aspiration Pneumonia, n (%) b |
| Yes | 4,295 (35) | 316,038 (22) |
| Reporter, n(%) b | Reporter, n(%) b | Reporter, n(%) b |
| Consumers | 5,152 (42) | 612,272 (43) |
| Healthcare Professionals | 488 (4) | 52,955 (4) |
| Medical Doctors | 2,789 (23) | 329,202 (23) |
| Others | 2,054 (17) | 237,128 (17) |
| Pharmacists | 1,252 (10) | 155,251 (11) |
| Not Available | 567 (5) | 54,856 (4) |
| Country b | Country b | Country b |
| North America | 8,677 (71) | 1,088,217 (75) |
| Europe (Western/Southern/Northern Countries) | 3,625 (29) | 353,447 (25) |
## 3.2.1 Disproportionality analysis
Results are presented in Table 6. Ranked by the absolute number of reports, the highest number of outcomes of interest were reported for Levodopa ($$n = 810$$), followed by Levodopa associated with beta-blockers ($$n = 82$$) and ACEi ($$n = 51$$). Sacubitril could not be included in the analysis because, even if five non-cases were present, no ICSRs reporting dysphagia were retrieved. Since all the other drugs were chosen as the Reference group, results must be read in comparison with them.
**TABLE 6**
| Unnamed: 0 | Dysphagia/Aspiration pneumonia (n = 12,302) | Other AEs (n = 1,441,664) | cROR (95% CI) | aROR (95% CI) a |
| --- | --- | --- | --- | --- |
| Over 65 patients | 11317 | 1418163 | Ref. Group | Ref. Group |
| Levodopa | 810 | 18366 | 5.5 (5.1; 5.9) b | 4.8 (4.5; 5.2) b |
| Levodopa + ACE inhibitors | 51 | 1295 | 4.9 (3.7; 6.5) b | 3.6 (2.7; 4.7) b |
| Levodopa + β-blockers | 82 | 2696 | 3.8 (3.0; 4.7) b | 2.8 (2.2; 3.4) b |
| Levodopa + Gliptins | 3 | 165 | 2.3 (0.6; 6.0) | 1.8 (0.4; 4.7) |
| Levodopa + More than one drug of interest | 39 | 979 | 5.0 (3.6; 6.8) b | 3.6 (2.6; 4.9) b |
As expected, compared to the general population over 65, which is known to be more at risk for dysphagia and aspiration pneumonia, patients treated with only Levodopa had the highest and most significant ROR, even after adjustment [aROR ($95\%$CI) = 4.8 (4.5; 5.2)]. Then, in general, the concomitant use of drugs inhibiting the degradation of SP reduces the reporting risk of Levodopa: ACEi [aROR ($95\%$CI) = 3.6 (2.7; 4.7)] and studies with more than one drug of interest [aROR ($95\%$CI) = 3.6 (2.6; 4.9)], followed by beta-blockers [aROR ($95\%$CI) = 2.8 (2.2; 3.4)], and gliptin s reaching a non-significant difference from the Reference group [aROR ($95\%$CI) = 1.8 (0.4; 4.7)].
The planned sensitivity analysis involving only ICSRs reported by physicians (total sub-cohort: 331, 991 ICSRs) was not performed because, among 2,789 ICSRs reporting dysphagia, no cases reported Levodopa associated with DPP-4i; 26 ICSRs reported Beta-blockers ($9\%$ vs. $6\%$ of cases in the main analysis) and 19 ACEi ($6\%$ vs. $4\%$ of cases in the main analysis).
## 4 Discussion
Drug repositioning has many advantages that make it an attractive drug discovery strategy. First, it simplifies regulatory procedures because clinical data concerning the safety and toxicity of the drug have already been acquired; the development is, therefore, faster and cheaper than de novo, and the drug is more likely to be introduced on the market (Ashburn and Thor, 2004).
Our study depicts the current knowledge on the effects of three drug classes (ACEi, beta-blockers, and DPP-4i) on dysphagia and aspiration pneumonia through a systematic review of 20 studies and a real-word data analysis from the spontaneous reporting system database FAERS. Here we will discuss main findings separately, for each drug class included in our focus.
## 4.1 ACE inhibitors
In our systematic review, seven studies reported improvements in dysphagia (Arai et al., 1998b; Nakayama et al., 1998; Arai et al., 2003; Shimizu et al., 2008; Marciniak et al., 2009; Nakashima et al., 2011; Matsumoto et al., 2012; Lee et al., 2015), whereas two studies found no changes or even a worsening in swallowing outcome (Brown, 2013; Fernandes et al., 2021). Differences may be related to the study population and design. All except two of the studies that showed favourable results were conducted on stroke patients. Dysphagia in patients with stroke is often characterized by delayed swallowing reflex, impaired protection of the lower airways, and absent cough reflex (Warnecke et al., 2021). Therefore, the improvement of these conditions seems to be the mechanism behind the protective effect of ACEi (Van de Garde et al., 2007).
The studies that failed to show any positive effects of ACEi on dysphagia were on elderly subjects. The pathophysiological mechanisms underlying dysphagia may be different and, consequently, may not be targeted by ACEi. Additionally, the studies reporting an improvement of dysphagia were all Asian studies. It has been suggested that the effect of ACEi in preventing aspiration pneumonia may be different in Asian and non-Asian populations (Ohkubo et al., 2004; Liu et al., 2012). It was hypothesized that the differential distribution of the ACE insertion/deletion polymorphisms between Asian and non-Asian populations may influence the efficacy of ACEi in improving cough reflex. However, the reasons behind the different effects are still not clearly understood and may be related to differences in the study design or to confounding variables. Finally, two of the seven studies with positive ACEi effects were not controlled (Arai et al., 1998a; Shimizu et al., 2008), thus, the causal relationship between the pharmacological treatment and the positive evolution of dysphagia could not be determined, as spontaneous recovery of swallowing function can occur in stroke patients.
Concerning the effect of ACEi on aspiration pneumonia, the literature is divided between studies showing a protective effect of ACEi (Arai et al., 1998a; Arai et al., 2001; Arai et al., 2005; Bosch et al., 2012; Liu et al., 2012; Sato et al., 2013) and studies failing to detect one when compared to controls or to other pharmacological treatments (Marciniak et al., 2009; Nakashima et al., 2011; Lee et al., 2015; Kumazawa et al., 2019). The only two RCTs that reported no reduction of aspiration pneumonia rate in patients treated with ACEi, but the studies were underpowered due to a small sample size (Nakashima et al., 2011; Lee et al., 2015). The high heterogeneity of the populations and the observational nature of the remaining studies limit the possibilities of comparing and interpreting the results.
Concerning the underlying mechanisms of action, two main mechanisms have been hypothesized: the inhibition of angiotensin II immunomodulatory effect, which reduces pro-inflammatory cytokine release, and the inhibition of the metabolism of both SP and bradykinin, which enhance the swallowing and cough reflexes (Raiden et al., 2002; Arai et al., 2003; He et al., 2007). As the prevention of the degradation of SP induced by ACEi causes its accumulation in the upper respiratory tract and induces the cough reflex (Noguchi et al., 2020), these drugs may then improve swallowing reflexes in patients with a history of aspiration pneumonia (Sekizawa et al., 1998; Okaishi et al., 1999. Although both ACEi and Angiotensin II Receptor Blockers (ARBs) inhibit angiotensin II activity, only ACEi has been shown to shorten the pharyngeal transit time through the increase of substance P and bradykinin levels, improving symptomless dysphagia (Holas et al., 1994; Arai et al., 2000; Shimizu et al., 2008). This activity may suggest that the enhanced cough reflex is most likely the mechanism responsible for the protective effect of ACEi on pneumonia (Liu et al., 2012; Kumazawa et al., 2019). However, there may be other mechanisms (e.g., immune-modulating effect) impacting aspiration pneumonia risk (Suzuki et al., 2003; He et al., 2006; Arndt et al., 2006).
There were some AEs associated with the use of ACEi. The most frequently reported AE was excessive dry cough and was associated with local increase in SP by the inhibition of ACE (Sekizawa et al., 1996). Although the frequency of excessive dry cough was limited to a small percentage of patients ($5\%$–$6\%$) in the retrieved studies (Arai et al., 1998b; Arai et al., 2003; Kano et al., 2016), its occurrence should be monitored as it may interfere with adherence to treatment. One RCTs by Lee et al. was prematurely interrupted because of the high mortality rate in the ACEi group (Lee et al., 2015). Patients in this study were particularly frail, being elderly, tube-fed, and with severe dysphagia, all well-known risk factors for aspiration pneumonia (Palmer and Padilla, 2022). Therefore, the use of ACEi to prevent aspiration pneumonia does not seem to be beneficial in frail high-risk patients and a cautious use of these drugs is recommended to avoid systemic effects on blood pressure, and cardiovascular and renal systems (Cheng et al., 2022).
In the FAERS, ACEi resulted in the highest aROR after Levodopaalone [aROR ($95\%$CI) = 3.6 (2.7; 4.7)]. The data reflect the discordant effect shown by the studies we retrieved through the systematic review: a clear reduction in the ROR was present but less effective than the other drug classes.
## 4.2 Beta-blockers
Findings supporting a potential positive effect of beta-blockers on dysphagia are relatively novel. Only two observational studies from Miarons and others (Miarons et al., 2016; Miarons et al., 2018) were retrieved through our systematic literature review. The authors reported an independent protective effect of beta-blockers on dysphagia in a cohort of elderly subjects with different neurological and non-neurological diseases. Nevertheless, the high refusal rate to participate in the study represents a significant limitation. Future RCTs with adequate sample size are necessary to confirm the efficacy of beta-blockers on dysphagia.
The mechanism by which beta-blockers could exert a protective effect on dysphagia is unknown. Previous research has hypothesized that SP could play a role, as seen with propranolol in guinea pigs (Belvisi, 1996; Lin and Lai, 1998) and supported by Miarons and colleagues’ research in the elderly (Miarons et al., 2018). It seems that beta-blockers prevent the occurrence of dysphagia through the release of pharyngeal SP. Another potential mechanism that has been hypothesized is an increase in contractile forces in pharyngeal muscles because of the upregulation of fast skeletal muscle beta-adrenergic receptors mediated by chronic beta2-adrenergic blockade (Murphy et al., 1997; Miarons et al., 2016).
Beta-blockers were the most reported drugs in our FAERS analysis, but they were ranked after the ACEi, compared to the reference group, thus suggesting a lower risk of dysphagia for beta-blockers compared to ACEi. Indeed, a recent study on the prevalence of oropharyngeal dysphagia in geriatric patients found a slightly higher prevalence of dysphagia in patients being treated with beta-blockers than among users of ACEi (Wolf et al., 2021). Unfortunately, the analyses of diseases and drugs were separated. It is therefore not possible to directly compare these results with our FAERS analysis since it only covers patients treated with Levodopa and drugs inhibiting the degradation of substance P (Wolf et al., 2021).
## 4.3 DPP-4 inhibitors
SP and GLP-1, a glucose-dependent insulinotropic peptide, are substrates of DPP-4, which is conversely inhibited by the class of antidiabetic drugs known as DPP-4i or gliptins (Noguchi et al., 2020). Since DPP-4i seem to prevent the degradation of SP (Cunningham and O'Connor, 1997; Brown et al., 2009), they were hypothesized to improve swallowing reflex and prevent dysphagia and aspiration pneumonia (Noguchi et al., 2020). Indeed, there have been studies on the potential of diabetic drug repurposing in patients with Parkinson’s disease (Labandeira et al., 2022). A case-control study showed a significant decrease in the incidence of Parkinson’s disease in diabetic patients treated with DPP-4i (OR = 0.23; $95\%$ CI: 0.07–0.74) (Svenningsson et al., 2016) and a similar result was found in a longitudinal cohort study (incidence rate ratio 0.64; $95\%$ CI: 0.43–0.88; $p \leq 0.01$) (Brauer et al., 2020). Jeong et al. observed a beneficial effect of DPP4-i in a small group of diabetic patients with Parkinson’s disease: they found a higher baseline dopamine transporter availability and better motor performance compared to non-diabetic patients (Jeong et al., 2021). However, even in this specific population, literature concerning the effect of DPP-4i on dysphagia and aspiration pneumonia is very limited. In our systematic review, only one study was retrieved, and it was based on the Japan Adverse Drug Event Report, a *Japanese spontaneous* pharmacovigilance database (Noguchi et al., 2020). The authors analyzed the events of dysphagia and aspiration pneumonia reported from eight anti-diabetic drugs classes. They concluded that there was no effect of DPP-4i on dysphagia, whereas their use was associated with an increased risk of aspiration pneumonia (Noguchi et al., 2020). Indeed, DPP-4 is the same substance as cell membrane surface antigen CD26, which is also expressed in Tcells (Alexandraki et al., 2006; Reinhold et al., 2007). For this reason DPP-4i may affect the immune system, increasing the risk of developing infections (Willemen et al., 2011).
The data from our analysis on the FAERS contrasts with the results by Noguchi and others (Noguchi et al., 2020).
Differences in the findings may be ascribed to differences in the populations targeted by the pharmacovigilance database search. Indeed, Noguchi et al. focused on diabetic patients, whereas in our analysis we included neurological patients with Parkinsonism being treated with L-dopa. Whereas in Parkinson’s disease a reduction of SP has been associated with the presence of dysphagia (Schröder et al., 2019), in diabetes it might be considered a secondary effect of autonomic neuropathy, which mainly leads to a hyperactivity of the cricopharyngeal muscle and a consequent relaxation of the upper oesophageal sphincter during swallowing (Restivo et al., 2006). Finally, the number of cases reported showing dysphagia and aspiration pneumonia associated with the use of DPP-4i was small both in our analysis and in the study by Noguchi. Therefore, there is a need for prospective studies to clarify the effects of DPP-4i on dysphagia and aspiration pneumonia.
## 4.4 Strengths and limits
This is the first study aimed at exploring the effects of some routinely used drugs on dysphagia and aspiration pneumonia through a combined approach, i.e., systematic review and real-world data analysis, to provide the most comprehensive overview of current knowledge on the topic.
The FAERS is the largest repository of spontaneously reported AEs; therefore, it allows access to very large samples, enabling to researchers to elucidate associations between drugs and reported adverse events that would be difficult to investigate with clinical trials.
However, the use of a pharmacovigilance database has some intrinsic limitations. Reporting might be influenced by factors such as notoriety bias, selection bias, and under-reporting, which precludes making causal inferences except in unusual circumstances (Faillie, 2019). As the FAERS is designed to report AEs, unintended positive effects of the drugs on swallowing function could not be recorded.
Furthermore, since case and non-case studies are drawn from different populations, this method cannot be a real substitute for the classical case-control study (Carnovale et al., 2019b; Faillie, 2019; Garcia et al., 2021). The actual risk and incidence rates cannot be determined from the analysis of AE reporting since the primary goal of a spontaneous reporting system is to signal the existence of a possible relationship between therapies and adverse events, without proving any causality.
With regard to the systematic review, since we included both peer-reviewed articles and congress abstracts, the methodological quality of the studies was generally low and the access to the information was limited. Additionally, the population and the outcome measures investigated in the included studies were heterogeneous.
Finally, the mechanisms responsible for the effects of the investigated drugs on swallowing function are still unclear. As described above, a hypothesized common mechanism is the increase of SP levels, being a secondary target for all the drug classes of interest in the present study. The most well-known function of this neuropeptide is the modulation of pain perception (Zieglgänsberger, 2019), but it is also involved in inflammation (Maggi, 1997) and gastrointestinal functions (Saito et al., 2003). With regard to swallowing function, SP stimulates the production of saliva and amylase through a vasodilatory effect in salivary glands (Pikula et al., 1992). In the oropharynx, SP is released by the sensory terminals of the receptors in the pharyngeal mucosa in response to mechanical, thermal and chemical stimuli (Alvarez-Berdungo er al, 2016). As a result, it enhances swallowing and cough reflexes (Jin et al., 1994; Imoto et al., 2011; Canning et al., 2014). Low concentration of SP has been reported in patients with dysphagia (Schröder et al., 2019) and aspiration pneumonia (Nakagawa et al., 1995) and has been associated with reduced spontaneous swallowing frequency (Niimi et al., 2018) and pharyngeal sensitivity (Tomsen et al., 2022). However, the retrospective nature of the case and non-case studies and the lack of data on SP concentrations in the majority of the studies in the systematic review limits the possibility of drawing conclusions on the causal relationship between the increase of SP levels and the protective effects on dysphagia reported in studies on ACEi and beta-blockers. Indeed, other mechanisms, including the primary mechanism of action of these drugs, may be responsible for the positive effects on swallowing function. This might also be in line with the heterogeneity of results in the systematic review when comparing different diseases: instead of a common mechanism of action, the same drug could exert its role in alternative ways in every pathophysiology. Thus, high-quality randomized controlled studies are required both to verify the efficacy of the investigated drugs on dysphagia and its pulmonary complications, and to analyze the role of SP.
## 5 Conclusion
Our study explores the potential repurposing of ACEi, beta-blockers, and DPP-4i in neurological patients with dysphagia to improve swallowing function and reduce aspiration pneumonia risk. Although a weak role for SP was hypothesized as one of the potential mechanisms associated with the protective effect on dysphagia, currently available data is insufficient to support this hypothesis.
Due to the nature of the study, no firm conclusion can be drawn on the role of these drugs in effectively ameliorating dysphagia or aspiration pneumonia. Their efficacy and the mechanisms of action should be verified in future high-quality randomized controlled studies. Nevertheless, caution is always required in frail patients at a high risk of pneumonia because of other systemic effects. Further high-quality RCTs, especially from non-Asian countries, are needed to verify the protective effects and identify best responders.
## Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
## Author contributions
VB and SOR conceptualized and designed the study, coordinated and supervised data collection, interpreted the data, drafted the manuscript, and approved the final manuscript as submitted. AB, GG, MG, GM, MP, and MN participated in the data collection and data analysis and approved the final article as submitted. SAR, EC, and AS participated in the conceptualization and design of the study, critically reviewed the manuscript, and approved the final manuscript as submitted. NP and CC conceptualized and designed the study, interpreted the data, coordinated and supervised data collection and the drafting of the manuscript, critically reviewed the manuscript, and approved the final manuscript as submitted.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphar.2023.1057301/full#supplementary-material
## References
1. Alexandraki K., Piperi C., Kalofoutis C., Singh J., Alaveras A., Kalofoutis A.. **Inflammatory process in type 2 diabetes: The role of cytokines**. *Ann. N.Y. Acad. Sci.* (2006) **1084** 89-117. PMID: 17151295
2. Arai T., Sekizawa K., Ohrui T., Fujiwara H., Yoshimi N., Matsuoka H.. **ACE inhibitors and protection against pneumonia in elderly patients with stroke**. *Neurology* (2005) **64** 573-574. DOI: 10.1212/01.WNL.0000150897.14961
3. Arai T., Yasuda Y., Takaya T., Toshima S., Kashiki Y., Yoshimi N.. **ACE inhibitors and symptomless dysphagia**. *Lancet* (1998a) **352**. DOI: 10.1016/S0140-6736(98)85021-6
4. Arai T., Yasuda Y., Takaya T., Toshima S., Kashiki Y., Yoshimi N.. **Angiotensin-converting enzyme inhibitors, angiotensin-II receptor antagonists, and pneumonia in elderly hypertensive patients with stroke**. *Chest* (2001) **119** 660-661. DOI: 10.1378/chest.119.2.660
5. Arai T., Yasuda Y., Takaya T., Toshima S., Kashiki Y., Yoshimii N.. **Angiotensin-converting enzyme inhibitors, angiotensin II receptor antagonists, and symptomless dysphagia**. *Chest* (2000) **117** 1819-1820. DOI: 10.1378/chest.117.6.1819
6. Arai T., Yasuda Y., Toshima S., Yoshimi N., Kashiki Y.. **ACE inhibitors and pneumonia in elderly people**. *Lancet (London, Engl.* (1998b) **352** 1937-1938. DOI: 10.1016/S0140-6736(05)60437-0
7. Arai T., Yoshimi N., Fujiwara H., Sekizawa K.. **Serum substance P concentrations and silent aspiration in elderly patients with stroke**. *Neurology* (2003) **61** 1625-1626. DOI: 10.1212/01.wnl.0000096395.80826.23
8. Arndt P. G., Young S. K., Worthen G. S.. **The angiotensin converting enzyme regulates lipopolysaccharide-induced pulmonary neutrophil influx through bradykin and plasminogen activator inhibitor-1 pathways**. *Proc. Am. Thor, Society* (2006) **3** A337
9. Ashburn T. T., Thor K. B.. **Drug repositioning: Identifying and developing new uses for existing drugs**. *Nat. Rev. Drug. Discov.* (2004) **3** 673-683. DOI: 10.1038/nrd1468
10. Attrill S., White S., Murray J., Hammond S., Doeltgen S.. **Impact of oropharyngeal dysphagia on healthcare cost and length of stay in hospital: A systematic review**. *Bmc. Health. Serv. Res.* (2018) **18** 594. DOI: 10.1186/s12913-018-3376-3
11. Auyeung M., Tsoi T. H., Mok V., Cheung C. M., Lee C. N., Li R.. **Ten-year survival and outcomes in a prospective cohort of new onset Chinese Parkinson's disease patients**. *J. Neurol. Neurosurg. Psychiatry.* (2012) **83** 607-611. DOI: 10.1136/jnnp-2011-301590
12. Baijens L. W., Clavé P., Cras P., Ekberg O., Forster A., Kolb G. F.. **European society for swallowing disorders - European union geriatric medicine society white paper: Oropharyngeal dysphagia as a geriatric syndrome**. *Clin. Interv. Aging* (2016) **11** 1403-1428. DOI: 10.2147/CIA.S107750
13. Belafsky P. C., Mouadeb D. A., Rees C. J., Pryor J. C., Postma G. N., Allen J.. **Validity and reliability of the eating assessment tool (EAT-10)**. *Ann. Otol. Rhinol. Laryngol.* (2008) **117** 919-924. DOI: 10.1177/000348940811701210
14. Belvisi M.. **Beta-blocker induced asthma: A role for sensory nerve hyperresponsiveness?**. *Clin. Exp. Allergy.* (1996) **26** 1343-1346. DOI: 10.1111/j.1365-2222.1996.tb00534.x
15. Bosch X., Formiga F., Cuerpo S., Torres B., Rosón B., López-Soto A.. **Aspiration pneumonia in old patients with dementia. Prognostic factors of mortality**. *Eur. J. Intern. Med.* (2012) **23** 720-726. DOI: 10.1016/j.ejim.2012.08.006
16. Brauer R., Wei L., Ma T., Athauda D., Girges C., Vijiaratnam N.. **Diabetes medications and risk of Parkinson's disease: A cohort study of patients with diabetes**. *Brain* (2020) **143** 3067-3076. DOI: 10.1093/brain/awaa262
17. Brown L.. *Clin. JMLA* (2013) **101** 342-343. DOI: 10.3163/1536-5050.101.4.023
18. Brown N. J., Byiers S., Carr D., Maldonado M., Warner B. A.. **Dipeptidyl peptidase-IV inhibitor use associated with increased risk of ACE inhibitor-associated angioedema**. *Hypertension* (2009) **54** 516-523. DOI: 10.1161/HYPERTENSIONAHA.109.134197
19. Canning B. J., Chang A. B., Bolser D. C., Smith J. A., Mazzone S. B., McGarvey L.. **Anatomy and neurophysiology of cough: CHEST guideline and expert panel report**. *Chest* (2014) **146** 1633-1648. DOI: 10.1378/chest.14-1481
20. Carnovale C., Mazhar F., Arzenton E., Moretti U., Pozzi M., Mosini G.. **Bullous pemphigoid induced by dipeptidyl peptidase-4 (DPP-4) inhibitors: A pharmacovigilance-pharmacodynamic/pharmacokinetic assessment through an analysis of the vigibase**. *Expert. Opin. Drug. Saf.* (2019a) **18** 1099-1108. DOI: 10.1080/14740338.2019.1668373
21. Carnovale C., Mosini G., Gringeri M., Battini V., Mazhar F., Pozzi M.. **Interaction between paracetamol and lamotrigine: New insights from the FDA adverse event reporting system (FAERS) database**. *Eur. J. Clin. Pharmacol.* (2019b) **75** 1323-1325. DOI: 10.1007/s00228-019-02691-4
22. Chaudhuri K. R., Martinez-Martin P., Schapira A. H., Stocchi F., Sethi K., Odin P.. **International multicenter pilot study of the first comprehensive self-completed nonmotor symptoms questionnaire for Parkinson's disease: The NMSQuest study**. *Mov. Disord.* (2006) **21** 916-923. DOI: 10.1002/mds.20844
23. Cheng I., Sasegbon A., Hamdy S.. **Effects of pharmacological agents for neurogenic oropharyngeal dysphagia: A systematic review and meta-analysis**. *Neurogastroenterol. Motil.* (2022) **34** e14220. DOI: 10.1111/nmo.14220
24. Clavé P., Arreola V., Romea M., Medina L., Palomera E., Serra-Prat M.. **Accuracy of the volume-viscosity swallow test for clinical screening of oropharyngeal dysphagia and aspiration**. *Clin. Nutr.* (2008) **27** 806-815. DOI: 10.1016/j.clnu.2008.06.011
25. **Clinical Pharmacology powered by elsevier ClinicalKey**. (2020)
26. Cunningham D. F., O'Connor B.. **Proline specific peptidases**. *Biochim. Biophys. Acta.* (1997) **1343** 160-186. DOI: 10.1016/s0167-4838(97)00134-9
27. Douglas S. D., Leeman S. E.. **Neurokinin-1 receptor: Functional significance in the immune system in reference to selected infections and inflammation**. *Ann. N.Y. Acad. Sci.* (2011) **1217** 83-95. DOI: 10.1111/j.1749-6632.2010.05826.x
28. Dzahini O., Singh N., Taylor D., Haddad P. M.. **Antipsychotic drug use and pneumonia: Systematic review and meta-analysis**. *J. Psychopharmacol.* (2018) **32** 1167-1181. DOI: 10.1177/0269881118795333
29. Faillie J. L.. **Case-non-case studies: Principle, methods, bias and interpretation**. *Therapie* (2019) **74** 225-232. DOI: 10.1016/j.therap.2019.01.006
30. Fernandes M. S., Castelo P. M., Chaves G. N., Fernandes J., Fonseca F., Zanato L. E.. **Relationship between polypharmacy, xerostomia, gustatory sensitivity, and swallowing complaints in the elderly: A multidisciplinary approach**. *J. Texture. Stud.* (2021) **52** 187-196. DOI: 10.1111/jtxs.12573
31. Fescharek R., Kübler J., Elsasser U., Frank M., Güthlein P.. **Medical dictionary for regulatory activities (MedDRA)**. *Pharm. Med.* (2004) **18** 259-269. DOI: 10.2165/00124363-200418050-00001
32. Ganesh A., Randall M. D.. **Does metformin affect outcomes in COVID-19 patients with new or pre-existing diabetes mellitus? A systematic review and meta-analysis**. *Br. J. Clin. Pharmacol.* (2022) **88** 2642-2656. DOI: 10.1111/bcp.15258
33. Garcia P., Montastruc J. L., Rousseau V., Hamard J., Sommet A., Montastruc F.. **β-Adrenoceptor antagonists and nightmares: A pharmacoepidemiological-pharmacodynamic study**. *J. Psychopharmacol.* (2021) **35** 1441-1448. DOI: 10.1177/02698811211034810
34. Gatti M., Fusaroli M., Caraceni P., Poluzzi E., De Ponti F., Raschi E.. **Serious adverse events with tocilizumab: Pharmacovigilance as an aid to prioritize monitoring in COVID-19**. *Br. J. Clin. Pharmacol.* (2021) **87** 1533-1540. DOI: 10.1111/bcp.14459
35. He X., Han B., Mura M., Xia S., Wang S., Ma T.. **Angiotensin-converting enzyme inhibitor captopril prevents oleic acid-induced severe acute lung injury in rats**. *Shock* (2007) **28** 106-111. DOI: 10.1097/SHK.0b013e3180310f3a
36. He X., Han D., Mura M., Liu M., Liu Z.. **Angiotensin-converting enzyme inhibitor captopril prevents oleic acid-induced acute lung injury**. *Proc. Am. Thorac. Soc.* (2006) **3** A839. DOI: 10.1097/SHK.0b013e3180310f3a
37. Heemskerk A., Roos R. A. C.. **Aspiration pneumonia and death in Huntington’s disease**. *PLoS. Curr.* (2012) **4** RRN1293. DOI: 10.1371/currents.RRN1293
38. Holas M. A., DePippo K. L., Reding M. J.. **Aspiration and relative risk of medical complications following stroke**. *Arch. Neurol.* (1994) **51** 1051-1053. DOI: 10.1001/archneur.1994.00540220099020
39. Hua Y., Dai X., Xu Y., Xing G., Liu H., Lu T.. **Drug repositioning: Progress and challenges in drug discovery for various diseases**. *Eur. J. Med. Chem.* (2022) **234** 114239. DOI: 10.1016/j.ejmech.2022.114239
40. Imoto Y., Kojima A., Osawa Y., Sunaga H., Fujieda S.. **Cough reflex induced by capsaicin inhalation in patients with dysphagia**. *Acta. Otolaryngol.* (2011) **131** 96-100. DOI: 10.3109/00016489.2010.516013
41. Jeong S. H., Chung S. J., Yoo H. S., Hong N., Jung J. H., Baik K.. **Beneficial effects of dipeptidyl peptidase-4 inhibitors in diabetic Parkinson's disease**. *Brain* (2021) **144** 1127-1137. DOI: 10.1093/brain/awab015
42. Jin Y., Sekizawa K., Fukushima T., Morikawa M., Nakazawa H., Sasaki H.. **Capsaicin desensitization inhibits swallowing reflex in Guinea pigs**. *Am. J. Respir. Crit. Care. Med.* (1994) **149** 261-263. DOI: 10.1164/ajrccm.149.1.7509247
43. Jourdan J. P., Bureau R., Rochais C., Dallemagne P.. **Drug repositioning: A brief overview**. *J. Pharm. Pharmacol.* (2020) **72** 1145-1151. DOI: 10.1111/jphp.13273
44. Kano O., Hirayama T., Takazawa T., Ishikawa Y., Miura K., Yanagihashi M.. **Sputum substance P concentration and peak cough experimental flow in patients with ALS after administration of enalapril**. *Amyotroph. Lateral. Scler. Front. Degener.* (2016) **17** 286
45. Kerber R.. **Old drugs, new life**. (2003)
46. Kertscher B., Speyer R., Fong E., Georgiou A. M., Smith M.. **Prevalence of oropharyngeal dysphagia in The Netherlands: A telephone survey**. *Dysphagia* (2015) **30** 114-120. DOI: 10.1007/s00455-014-9584-z
47. Kumazawa R., Jo T., Matsui H., Fushimi K., &Yasunaga H.. **Association between angiotensin-converting enzyme inhibitors and post-stroke aspiration pneumonia**. *J. Stroke. Cerebrovasc. Dis.* (2019) **28** 104444. DOI: 10.1016/j.jstrokecerebrovasdis.2019.104444
48. Labandeira C. M., Fraga-Bau A., Arias Ron D., Alvarez-Rodriguez E., Vicente-Alba P., Lago-Garma J.. **Parkinson's disease and diabetes mellitus: Common mechanisms and treatment repurposing**. *Neural. Regen. Res.* (2022) **17** 1652-1658. DOI: 10.4103/1673-5374.332122
49. Lanska D. J., Lanska M. J., Lavine L., Schoenberg B. S.. **Conditions associated with Huntington’s disease at death. A case-control study**. *Arch. Neurol.* (1988) **45** 878-880. DOI: 10.1001/archneur.1988.00520320068017
50. Lee J. S., Chui P. Y., Ma H. M., Auyeung T. W., Kng C., Law T.. **Does low dose angiotensin converting enzyme inhibitor prevent pneumonia in older people with neurologic dysphagia-A randomized placebo-controlled trial**. *J. Am. Med. Dir. Assoc.* (2015) **16** 702-707. DOI: 10.1016/j.jamda.2015.05.009
51. Lin C. W., Lai Y. L.. **Tachykinins in propranolol-augmented, hyperpnoea-induced bronchoconstriction in Taida guinea-pigs: effects of dimethylthiourea**. *J. Auton. Pharmacol.* (1998) **18** 139-47. PMID: 9754634
52. Liu C. L., Shau W. Y., Wu C. S., Lai M. S.. **Angiotensin-converting enzyme inhibitor/angiotensin II receptor blockers and pneumonia risk among stroke patients**. *J. Hypertens.* (2012) **30** 2223-2229. DOI: 10.1097/HJH.0b013e328357a87a
53. Longman R.. **Jumpstart to products**. *Vivo-New York Then Norwalk* (2004) **222** 17-30
54. Maggi C. A.. **The effects of tachykinins on inflammatory and immune cells**. *Regul. Pept.* (1997) **70** 75-90. DOI: 10.1016/s0167-0115(97)00029-3
55. Marciniak C., Korutz A. W., Lin E., Roth E., Welty L., Lovell L.. **Examination of selected clinical factors and medication use as risk factors for pneumonia during stroke rehabilitation: A case-control study**. *Am. J. Phys. Med. Rehabil.* (2009) **88** 30-38. DOI: 10.1097/PHM.0b013e3181909b73
56. Marin S., Serra-Prat M., Ortega O., Audouard Fericgla M., Valls J., Palomera E.. **Healthcare costs of post-stroke oropharyngeal dysphagia and its complications: Malnutrition and respiratory infections**. *Eur. J. Neurol.* (2021) **28** 3670-3681. DOI: 10.1111/ene.14998
57. Matsumoto S., Shimodozono M., Toyama K., Kawahira K.. **Beneficial effect of perindopril in post-stroke patients with dysphagia**. *Dysphagia* (2012) **27** 605
58. Mazhar F., Battini V., Pozzi M., Invernizzi E., Mosini G., Gringeri M.. **Hyponatremia following antipsychotic treatment:**. *Int. J. Neuropsychopharmacol.* (2021) **24** 477-489. DOI: 10.1093/ijnp/pyab005
59. Mazhar F., Pozzi M., Gentili M., Scatigna M., Clementi E., Radice S.. **Association of hyponatraemia and antidepressant drugs: A pharmacovigilance-pharmacodynamic assessment through an analysis of the US Food and drug administration adverse event reporting system (FAERS) database**. *CNS Drugs* (2019) **33** 581-592. DOI: 10.1007/s40263-019-00631-5
60. Merlo A., Cohen S.. **Swallowing disorders**. *Annu. Rev. Med.* (1988) **39** 17-28. PMID: 3285768
61. Miarons M., Campins L., Palomera E., Serra-Prat M., Cabré M., Rofes L.. **Drugs related to oropharyngeal dysphagia in older people**. *Dysphagia* (2016) **31** 697-705. DOI: 10.1007/s00455-016-9735-5
62. Miarons M., Tomsen N., Nascimento W., Espín À., López-Faixó D., Clavé P.. **Increased levels of substance P in patients taking beta-blockers are linked with a protective effect on oropharyngeal dysphagia**. *Neurogastroenterol. Motil.* (2018) **30** e13397. DOI: 10.1111/nmo.13397
63. Murphy R. J., Gardiner P. F., Rousseau G., Bouvier M., Béliveau L.. **Chronic beta-blockade increases skeletal muscle beta-adrenergic-receptor density and enhances contractile force**. *J. Appl. Physiol.* (1997) **83** 459-465. DOI: 10.1152/jappl.1997.83.2.459
64. Nakagawa T., Ohrui T., Sekizawa H. S., Sasaki H.. **Sputum substance P in aspiration pneumonia**. *Lancet* (1995) **345** 1447. DOI: 10.1016/s0140-6736(95)92638-0
65. Nakashima T., Hattori N., Okimoto M., Yanagida J., Kohno N.. **Nicergoline improves dysphagia by upregulating substance P in the elderly**. *Med. Baltim.* (2011) **90** 279-283. DOI: 10.1097/MD.0b013e318224413b
66. Nakayama K., Sekizawa K., Sasaki H.. **ACE inhibitor and swallowing reflex**. *Chest* (1998) **113** 1425. DOI: 10.1378/chest.113.5.1425
67. National Library of Medicine (2020). Available at: https://lhncbc.nlm.nih.gov/MOR/RxTerms/
(Accessed July 30, 2020).. (2020)
68. Niimi M., Hashimoto G., Hara T., Yamada N., Abo M., Fujigasaki H.. **Relationship between frequency of spontaneous swallowing and salivary substance P level in patients with acute stroke**. *Dysphagia* (2018) **33** 414-418. DOI: 10.1007/s00455-017-9867-2
69. Noguchi Y., Esaki H., Murayama A., Sugioka M., Koyama A., Tachi T.. **Association between dipeptidyl peptidase-4 inhibitor and aspiration pneumonia: Disproportionality analysis using the spontaneous reporting system in Japan**. *Eur. J. Clin. Pharmacol.* (2020) **76** 299-304. DOI: 10.1007/s00228-019-02794-y
70. Ohkubo T., Chapman N., Neal B., Woodward M., Omae T., Chalmers J.. **Effects of an angiotensin-converting enzyme inhibitor-based regimen on pneumonia risk**. *Am. J. Respir. Crit. Care Med.* (2004) **169** 1041-1045. DOI: 10.1164/rccm.200309-1219OC
71. Okaishi K., Morimoto S., Fukuo K., Niinobu T., Hata S., Onishi T.. **Reduction of risk of pneumonia associated with use of angiotensin I converting enzyme inhibitors in elderly inpatients**. *Am. J. Hypertens.* (1999) **12** 778-783. DOI: 10.1016/s0895-7061(99)00035-7
72. Ouzzani M., Hammady H., Fedorowicz Z., Elmagarmid A.. **Rayyan-a web and mobile app for systematic reviews**. *Syst. Rev.* (2016) **5** 210. DOI: 10.1186/s13643-016-0384-4
73. Page M. J., McKenzie J. E., Bossuyt P. M., Boutron I., Hoffmann T. C., Mulrow C. D.. **The PRISMA 2020 statement: An updated guideline for reporting systematic reviews**. *BMJ* (2021) **372** n71. DOI: 10.1136/bmj.n71
74. Palmer P. M., Padilla A. H.. **Risk of an adverse event in individuals who aspirate: A review of current literature on host defenses and individual differences**. *Am. J. Speech. Lang. Pathol.* (2022) **31** 148-162. DOI: 10.1044/2021_AJSLP-20-00375
75. Pikula D. L., Harris E. F., Desiderio D. M., Fridland G. H., Lovelace J. L.. **Methionine enkephalin-like, substance P-like, and β-endorphin-like immunoreactivity in human parotid saliva**. *Arch. Oral. Biol.* (1992) **37** 705-709. DOI: 10.1016/0003-9969(92)90076-k
76. Raiden S., Nahmod K., Nahmod V., Semeniuk G., Pereira Y., Alvarez C.. **Nonpeptide antagonists of AT1 receptor for angiotensin II delay the onset of acute respiratory distress syndrome**. *J. Pharmacol. Exp. Ther.* (2002) **303** 45-51. DOI: 10.1124/jpet.102.037382
77. Reinhold D., Biton A., Goihl A., Pieper S., Lendeckel U., Faust J.. **Dual inhibition of dipeptidyl peptidase IV and aminopeptidase N suppresses inflammatory immune responses**. *Ann. N.Y. Acad. Sci.* (2007) **1110** 402-409. DOI: 10.1196/annals.1423.042
78. Restivo D. A., Marchese-Ragona R., Lauria G., Squatrito S., Gullo D., Vigneri R.. **Botulinum toxin treatment for oropharyngeal dysphagia associated with diabetic neuropathy**. *Diabetes Care* (2006) **29** 2650-2653. DOI: 10.2337/dc05-2486
79. Saito R., Takano Y., Kamiya H. O.. **Roles of substance P and NK(1) receptor in the brainstem in the development of emesis**. *J. Pharmacol. Sci.* (2003) **91** 87-94. DOI: 10.1254/jphs.91.87
80. Sakaeda T., Tamon A., Kadoyama K., Okuno Y.. **Data mining of the public version of the FDA adverse event reporting system**. *Int. J. Med. Sci.* (2013) **10** 796-803. DOI: 10.7150/ijms.6048
81. Sato Y., Itoh H., Suzuki Y., Tatsuta R., Takeyama M.. **Effect of pilocarpine on substance P and calcitonin gene-related peptide releases correlate with salivary secretion in human saliva and plasma**. *J. Clin. Pharm. Ther.* (2013) **38** 19-23. DOI: 10.1111/jcpt.12011
82. Schröder J. B., Marian T., Claus I., Muhle P., Pawlowski M., Wiendl H.. **Substance P saliva reduction predicts pharyngeal dysphagia in Parkinson's disease**. *Front. Neurol.* (2019) **10** 386. DOI: 10.3389/fneur.2019.00386
83. Sekizawa K., Jia Y. X., Ebihara T., Hirose Y., Hirayama Y., Sasaki H.. **Role of substance P in cough**. *Pulm. Pharmacol.* (1996) **9** 323-328. DOI: 10.1006/pulp.1996.0042
84. Sekizawa K., Matsui T., Nakagawa T., Nakayama K., Sasaki H.. **ACE inhibitors and pneumonia**. *Lancet* (1998) **352** 1069. DOI: 10.1016/S0140-6736(05)60114-6
85. Shimizu T., Fujioka S., Otonashi H., Kondo M., Sekizawa K.. **ACE inhibitor and swallowing difficulties in stroke. A preliminary study**. *J. Neurol.* (2008) **255** 288-289. DOI: 10.1007/s00415-006-0157-4
86. Sterne J. A., Hernán M. A., Reeves B. C., Savović J., Berkman N. D., Viswanathan M.. **ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions**. *BMJ* (2016) **355** i4919. DOI: 10.1136/bmj.i4919
87. Sterne J., Savović J., Page M. J., Elbers R. G., Blencowe N. S., Boutron I.. **RoB 2: A revised tool for assessing risk of bias in randomised trials**. *BMJ* (2019) **366** l4898. DOI: 10.1136/bmj.l4898
88. Stuart M.. **Rediscovering existing drugs**. *Start-Up* (2004) **9** 23-30
89. Suzuki Y., Ruiz-Ortega M., Lorenzo O., Ruperez M., Esteban V., Egido J.. **Inflammation and angiotensin II**. *Int. J. Biochem. Cell. Biol.* (2003) **35** 881-900. DOI: 10.1016/s1357-2725(02)00271-6
90. Svenningsson P., Wirdefeldt K., Yin L., Fang F., Markaki I., Efendic S.. **Reduced incidence of Parkinson's disease after dipeptidyl peptidase-4 inhibitors-A nationwide case-control study**. *Mov. Disord.* (2016) **31** 1422-1423. DOI: 10.1002/mds.26734
91. Takizawa C., Gemmell E., Kenworthy J., Speyer R.. **A systematic review of the prevalence of oropharyngeal dysphagia in stroke, Parkinson's disease, alzheimer's disease, head injury, and pneumonia**. *Dysphagia* (2016) **31** 434-441. DOI: 10.1007/s00455-016-9695-9
92. **Adverse event reporting system (FAERS)**. (2020)
93. Tomsen N., Ortega O., Nascimento W., Carrión S., Clavé P.. **Oropharyngeal dysphagia in older people is associated with reduced pharyngeal sensitivity and low substance P and CGRP concentration in saliva**. *Dysphagia* (2022) **37** 48-57. DOI: 10.1007/s00455-021-10248-w
94. van de Garde E. M., Souverein P. C., Hak E., Deneer V. H., van den Bosch J. M., Leufkens H. G.. **Angiotensin-converting enzyme inhibitor use and protection against pneumonia in patients with diabetes**. *J. Hypertens.* (2007) **25** 235-239. DOI: 10.1097/HJH.0b013e328010520a
95. Wang P., Wang B., Chen X., Xiong B., Xie F., Wu S.. **Six-year follow-up of dysphagia in patients with Parkinson's disease**. *Dysphagia* (2022) **37** 1271-1278. DOI: 10.1007/s00455-021-10387-0
96. Warnecke T., Labeit B., Schroeder J., Reckels A., Ahring S., Lapa S.. **Neurogenic dysphagia: Systematic review and proposal of a classification system**. *Neurology* (2021) **96** e876-e889. DOI: 10.1212/WNL.0000000000011350
97. Willemen M. J., Mantel-Teeuwisse A. K., Straus S. M., Meyboom R. H., Egberts T. C., Leufkens H. G.. **Use of dipeptidyl peptidase-4 inhibitors and the reporting of infections: A disproportionality analysis in the world health organization VigiBase**. *Diabetes Care* (2011) **34** 369-374. DOI: 10.2337/dc10-1771
98. Wolf U., Eckert S., Walter G., Wienke A., Bartel S., Plontke S. K.. **Prevalence of oropharyngeal dysphagia in geriatric patients and real-life associations with diseases and drugs**. *Sci. Rep.* (2021) **11** 21955. DOI: 10.1038/s41598-021-99858-w
99. Zieglgänsberger W.. **Substance P and pain chronicity**. *Cell. Tissue. Res.* (2019) **375** 227-241. DOI: 10.1007/s00441-018-2922-y
|
---
title: 'Exploring obesity, physical activity, and digital game addiction levels among
adolescents: A study on machine learning-based prediction of digital game addiction'
authors:
- Mehmet Gülü
- Fatma Hilal Yagin
- Ishak Gocer
- Hakan Yapici
- Erdem Ayyildiz
- Filipe Manuel Clemente
- Luca Paolo Ardigò
- Ali Khosravi Zadeh
- Pablo Prieto-González
- Hadi Nobari
journal: Frontiers in Psychology
year: 2023
pmcid: PMC10022696
doi: 10.3389/fpsyg.2023.1097145
license: CC BY 4.0
---
# Exploring obesity, physical activity, and digital game addiction levels among adolescents: A study on machine learning-based prediction of digital game addiction
## Abstract
Primary study aim was defining prevalence of obesity, physical activity levels, digital game addiction level in adolescents, to investigate gender differences, relationships between outcomes. Second aim was predicting game addiction based on anthropometric measurements, physical activity levels. Cross-sectional study design was implemented. Participants aged 9–14 living in Kirikkale were part of the study. The sample of the study consists of 405 adolescents, 231 girls ($57\%$) and 174 boys ($43\%$). Self-reported data were collected by questionnaire method from a random sample of 405 adolescent participants. To determine the physical activity levels of children, the Physical Activity Questionnaire for Older Children (PAQ-C). Digital Game addiction was evaluated with the digital game addiction (DGA) scale. Additionally, body mass index (BMI) status was calculated by measuring the height and body mass of the participants. Data analysis were performed using Python 3.9 software and SPSS 28.0 (IBM Corp., Armonk, NY, United States) package program. According to our findings, it was determined that digital game addiction has a negative relationship with physical activity level. It was determined that physical activity level had a negative relationship with BMI. In addition, increased physical activity level was found to reduce obesity and DGA. Game addiction levels of girl participants were significantly higher than boy participants, and game addiction was higher in those with obesity. With the prediction model obtained, it was determined that age, being girls, BMI and total physical activity (TPA) scores were predictors of game addiction. The results revealed that the increase in age and BMI increased the risk of DGA, and we found that women had a 2.59 times greater risk of DGA compared to men. More importantly, the findings of this study showed that physical activity was an important factor reducing DGA 1.51-fold. Our prediction model Logit (P) = 1/(1 + exp(−(−3.384 + Age*0.124 + Gender-boys*(−0.953) + BMI*0.145 + TPA*(−0.410)))). Regular physical activity should be encouraged, digital gaming hours can be limited to maintain ideal weight. Furthermore, adolescents should be encouraged to engage in physical activity to reduce digital game addiction level. As a contribution to the field, the findings of this study presented important results that may help in the prevention of adolescent game addiction.
## Introduction
In the last decade, digital games have become one of the most spent online activities among adolescents, and simultaneously, this activity has changed from being partly computer or console-based to a multiplatform activity (Association, 2015). With the proliferation of smart mobile phones, mobile games are one of the most popular entertainment sectors of the multimedia application industry in Europe (Consumer, 2015). With the increase in games that can be played with technological devices such as computers, phones, and tablets, research on the positive and negative aspects of games has also increased. Research shows that active video games and some games can be a motivating source for following an active lifestyle and can help improve indicators of health status in young adults (Zurita-Ortega et al., 2018). However, in addition to the positive effects of the developing digital technology, it causes many problems on health in terms of mental, physical, social, environmental aspects that may cause problems such as not being able to limit its use or using it for some unintended activities. Spending excessive time on digital games in the ongoing daily routine leads to adopting a sedentary lifestyle and creating an antisocial structure (Christakis, 2019; Almourad et al., 2020).
Adolescence is a unique development period of human development and the foundations of a healthy life are laid. According to the definition of the World Health Organization, adolescence is the life stage between childhood and adulthood, between the ages of 10 and 19. Adolescents experience rapid physical, cognitive and psychosocial growth. This influences what adolescents think, how they feel, and how they make decisions, and most importantly, how they interact with the world around them (WHO, 2022). The level of physical activity is insufficient in adolescents and children all over the world, especially female adolescents and decreases with age (Aubert et al., 2021).
Promoting a lifelong healthy lifestyle and participation in physical activity is important for many countries (Pühse and Gerber, 2005). In order to prevent these problems and improve quality of life, physical activity is recommended by WHO. For children, physical activities of moderate to vigorous intensity (such as jumping rope, line games, holding and rolling games, ice skating, gymnastics, skiing, athletics, soccer, swimming, dancing, table tennis, and slow-paced cycling) should be preferred for at least 60 min every day, at least three times a week. In its current guidelines, the WHO recommends physical activity for children and adolescents, including moderate-to-high intensity, mostly aerobic exercise, for at least 60 min a day on average throughout the week. High-intensity aerobic exercise should also be practiced at least 3 days a week to improve bone and muscle health (WHO, 2010, 2019). Physical activity, which is defined as all body movements that increase the activity of the circulatory system and cause fatigue, performed in a way that causes energy expenditure with our skeletal muscles above the basal level (Bronikowska et al., 2021). A physically active lifestyle begins to develop in childhood and continues throughout life with moderate or high stability of physical activity (Telama et al., 2014). Studies on the health benefits of physical activity have reported that physical activity has positive results in terms of academic achievement and musculoskeletal health (Strong et al., 2005). However, inadequacy of physical activity not only causes many problems in adolescents, but also negatively affects their physical development (Storz, 2020). However, studies have reported that physical activity level tends to decrease during adolescence (Dumith et al., 2011). To further promote physical activity, the World Health Organization launched in 2018 the project of more active people for a healthier world, new global action on physical activity, to reduce the global prevalence of insufficient physical activity among adolescents and adults by $15\%$ by 2030 (WHO, 2019). The rate of physical activity declines significantly during adolescence, and girls are less physically active than boys ($56\%$ vs. $39\%$; Kumar et al., 2015). This coincides with the increase in obesity rates among 11–15-year-old adolescents, of whom $38\%$ are overweight (van Jaarsveld and Gulliford, 2015).
Increases in overweight and obesity have become an important public health problem worldwide (Wyatt et al., 2006; Abarca-Gómez et al., 2017). Since the 1970s, there has been an almost three-fold increase in the prevalence of obesity worldwide, and studies have reported that approximately one-fifth of American children are obese (Ogden et al., 2018). Although its positive effects are known, it is stated that approximately $80\%$ of children and adolescents aged 13–18 worldwide do not participate in physical activity (Rhodes et al., 2017). Studies have reported an increase in poor eating habits in adolescents (Abarca-Gómez et al., 2017) and a decrease in physical activity, especially during the pandemic period (Kohl et al., 2012). Problems such as the habit of being sedentary, attitudes and behaviors of not doing physical activity, and nutritional disorders that start in childhood and adolescence continue as the age progresses, causing individuals not to include physical activity adequately in their lives (Pombo et al., 2020). Both inactive lifestyle and poor eating habits are thought to be among the leading causes of overweight and obesity prevalence (Stevens et al., 2012). In a similar study, it was found that obese children had higher levels of food addiction and lower physical activity levels than non-obese children (Gülü et al., 2022). Although efforts to prevent obesity have increased in the last two decades worldwide, these efforts have not been enough to prevent the increase in obesity in children (Brown et al., 2019). Traditional interventions aimed at preventing obesity, targeting energy balance and individual health behavior change, such as calorie restriction and increasing the level of physical activity, were limited in reducing the prevalence of obesity (Al-Khudairy et al., 2017; Mead et al., 2017).
The study reported that children aged 12–18 years who spent a lot of time with computers and television had higher BMI values (Alghadir et al., 2021). Increased body composition has a significant association with physical activity (especially increased daily moderate to vigorous physical activity), with just over $30\%$ of children aged 8–12 years playing digital games for 2 h or more daily (O’Brien et al., 2021). Li et al. [ 2014] found that among 1,150 rural and urban middle school students with digital addiction in China, obesity values ($32.92\%$) differed from those of students without digital addiction ($21.06\%$). Therefore, they found a relationship between obesity and internet addiction and reported that internet addiction is an independent risk factor for obesity (Li et al., 2014). According to a study, 14 of 26 studies ($53\%$) published between 2013 and 2018 reported that there was no relationship between video games and obesity, while 12 reported a positive relationship (Kracht et al., 2020). When the studies in the literature are examined, the relationship between obesity and game addiction remains unclear.
To our knowledge, this is the first study to predict gaming addiction using machine learning based on physical activity and BMI. Our hypothesis that as the level of game addiction increases, the prevalence of obesity will increase due to the decrease in physical activity level and increase in BMI. The primary aim of this study was to define the prevalence of obesity, physical activity levels and digital game addiction level in adolescents, and to investigate gender differences and relationships between outcomes. The second aim was to predict game addiction based on anthropometric measurements and physical activity levels. The concept of “digital addiction,” which emerged with the impact of technological developments, has become a major and widespread problem today. Due to the temporary solutions and the ineffectiveness of some preventive approaches, it is predicted that the concept of physical activity will be more effective as an indispensable prevention method in the lives of individuals.
## Study design and setting
This cross-sectional study was implemented in line with the purpose described in the STROBE checklist (von Elm et al., 2007). A simple random sampling method was used in the data collection process. All the procedures described below were conducted in October/November 2021.
## Participants
With G*Power software (University of Dusseldorf, Dusseldorf, Germany, version 3.0.1), the independent sample t-test was used to calculate sample size and actual power (α = 0.05, power = 0.80, effect size = 0.35). The results revealed that with a sample size of 260 participants, the actual power was $80.2\%$ (Faul et al., 2007). The research group of this study consisted of (231 girls %57 and 174 boys %43) 405 adolescent individuals in Turkey. The participants mean age, height, and body mass were 11.37 ± 1.45 years, 149.42 ± 11.17 cm, and 44.22 ± 13.06 kg, respectively. The research was announced in schools through a poster or verbal and information was given to the participants who wanted to take part in the research. Inclusion criteria: it consists of individuals between the ages of 9 and 14, speaking fluent Turkish, currently residing in Kirikkale and without any mental or chronic disease that prevents them from participating in physical activity. After reading the information form about the research by the families and children, the participants filled out the questionnaires after measuring the height and body mass. Participants who wanted to withdraw from the study could withdraw from the study at any time without completing the questionnaire. This study was approved by the Kirikkale University Social and Human Sciences Ethics Committee in line with the Declaration of Helsinki (Protocol no: $\frac{10}{18.10.2021}$).
## Data collection method
A descriptive survey model and a quantitative method were used in the research (Karasar, 2016). Because of the Cronbach Alpha reliability analysis of the study, it was concluded that the game addiction for adolescents and physical activity inventories for children was reliable in the range of 0.70–0.90. PAQ-C Cronbach’s Alpha 0.781, Game addiction scale Cronbach’s Alpha 0.902 was detected. The data collection method consisted of three parts: The first part was the personal information form, which consisted of questions about gender, age, health status. Second part was evaluation anthropometric measurements were made at the school sport center. All anthropometric measurements were taken in the afternoon and the indoor at sport center ambient temperature was recorded as temperature 22°. In the third part PAQ-C and Digital Game addiction scales were applied. Height was measured using a sensitive measuring up to 0.1 cm (Seca 217, Seca, Hamburg, Germany). During height measurement, the participant was standing, without shoes, with heels and his head in the horizontal plane. Body mass was measured using the sensitive Tanita body analysis system (Tanita Corp., Tokyo, Japan) up to 0.1 kg for a participant. To determine the physical activity status of the participants, Erdim et al. [ 2019] Physical Activity Questionnaire for Older Children (PAQ-C) with 10 questions adapted into Turkish was used. Additionally, the Validity and Reliability of the Game Addiction Scale for Adolescents-Short Form game addiction scale developed by Gazanfer and Taş [2018] was used to determine game addiction levels.
## Body mass index
Anthropometric measurements were measured standing height and body weight with light clothing and no shoes. In this direction, BMI values were calculated by measuring the height and weight of the individuals; The 85th and 95th percentiles were considered overweight, and those above the 95th percentile were considered obese (Neyzi et al., 2008; Ogden and Flegal, 2010; Wei et al., 2020; Cdc "Centers for Disease Control and Prevention (CDC), 2020, 2022) determining the BMI values and obesity status of the participants, the child body mass index calculation application on the website of Centers for Disease Control and Prevention (CDC) was used (CDC). The outcomes extracted for further data treatment was the BMI measured in kg/m2.
## Digital game addiction questionnaire
The scale developed by Lemmens et al. [ 2009] was adapted into Turkish by Gazanfer and Taş [2018]. Scale consists of nine items. 5-point Likert-type grading was used to score the scale. The lowest score to be obtained from the scale is 9 and the highest score is 45. The grading is “Never,” “Rarely,” “Sometimes,” “Often,” and “Very often.” The Cronbach Alpha internal consistency coefficient of the scale was found to be 0.92. These results show that the responses to our scale were consistent before further to statistical analysis. The total scale score was used in the evaluation of digital game addiction.
## Physical activity questionnaire for older children
The scale consists of 10 items, of which nine items are used to determine the level of physical activity. The tenth item gives information about whether the participant does physical activity in case of illness. The tenth item is excluded from the calculation of the physical activity score. The first question in PAQ C provides descriptive options for what they do physically in 22 titles. Responses to this question are evaluated as a 5-point rating (1 = no activity, 5 = 7 times or more), from which the average score is calculated; higher scores indicate greater levels of physical activity. Defining the 22 activities clearly and precisely will be a reminder to the children. The other eight questions relate to the assessment of physical activities performed during the day or at specific time intervals throughout the week (e.g., physical education class, recess, noon, after-school, evening, weekend). These items are scored on a 5-point scale, with higher scores indicating higher activity level. The overall PAQ-C score is obtained by adding the scores of items 1–9, and the final physical activity-level score is the average of the scores of these nine items. An average of 1 point indicates a low PA level, and an average of 5 points indicates a high PA level (Kowalski et al., 2004; Erdim et al., 2019). The total scale score was used in the evaluation of physical activity level.
## Statistical analysis
The conformity of the variables to the normal distribution was examined using visual (histogram and probability plots) and analytical (Shapiro–Wilk Test) methods. The assumption of homogeneity of variances was examined by Levene’s test. Descriptive statistics are expressed as median, interquartile range for non-normally distributed variables, and mean ± standard deviation for normally distributed variables. Independent Samples t-test was used in the comparisons of two groups regarding the variables satisfying the parametric test assumptions. Mann–Whitney U and Kruskal–Wallis H tests were used for comparisons of two or more groups regarding the variables that did not meet the parametric test assumptions, respectively. Frequency (n) and percentage (%) values were calculated for qualitative variables. Comparison of BMI levels with physical activity and game addiction scale scores was done using the Kruskal–Wallis H analysis. After the Kruskal–Wallis H test results, the Conover test was used for pairwise comparisons for universally significant variables. Effect size was calculated using the Cohen’s D. The magnitude of effect size was considered following the thresholds: Cohen suggested that $d = 0.2$ be considered a ‘small’ effect size, 0.5 represents a “medium” effect size and 0.8 a “large” effect size (Cohen, 1988). Relationships between outcomes were conducted using the Pearson-r product moment correlation test. The magnitude of correlations were defined as follows: <0.1 = trivial, 0.1–0.3 = small, 0.3–0.5 = moderate, 0.5–0.7 = large, 0.7–0.9 = very large, and >0.9 = nearly perfect (Batterham and Hopkins, 2006). In multivariate analysis, independent predictors were analyzed using binomial logistic regression analysis and possible factors identified in previous analyses. Logistic regression was performed according to the forward feature selection method. Hosmer-Lemeshow and Omnibus tests were used to evaluate the logistic regression model and its coefficients. In all results, p-value of <0.05 was considered statistically significant. American Psychological Association (APA) 6.0 style was used to report statistical differences (Yağin et al., 2021). Statistical analyzes were performed using Python 3.9 software and SPSS 28.0 (IBM Corp., Armonk, NY, United States) package program.
## Data preprocessing and machine learning approach
This section provides a description of the approach used to evaluate the predictive ability of the machine learning (ML) approach for game addiction prediction. In the study, there was a high level of class imbalance problem in the distribution of the groups for digital game addiction [digital game addiction: 38 ($9.4\%$), no-digital game addiction 367 ($90.6\%$)]. Synthetic Minority Over-sampling Technique for Nominal and Continuous (SMOTE-NC) Gök and Olgun [2021] was used to eliminate the class imbalance problem. Class imbalance problem, when working with real-life data, this problem is highly prevalent and can be defined as a state of imbalance when there are significantly more cases belonging to the majority class than those belonging to the minority class (Paksoy and Yağin, 2022). Because ML techniques like logistic regression might be biased toward the majority class, which causes issues with under or overfitting, balanced data is crucial. The standard oversampling method multiplies the available data, but the SMOTE makes synthetic samples from the minority class using the information in the data, and under sampling eliminates the majority class from the data. As a result, SMOTE is a popular technique for imbalance concerns since it may perform better than straightforward sampling techniques by avoiding over or underfitting difficulties (Chen et al., 2022). Afterwards, a logistic regression model was created based on the advanced feature selection method, and accuracy, sensitivity, specificity, FI-score, positive and negative predictive values were calculated using the Python 3.9 program with the help of the confusion matrix to evaluate the prediction performance of the model.
## Results
The mean age of the participants was 11.38 ± 1.45 years. Of the 405 participants, 231 ($57\%$) were girls and 174 ($43\%$) were boys. The mean age was 11.28 ± 1.47 years for girls participants and 11.51 ± 1.43 years for boys participants. The mean BMI of the participants was 19.48 ± 4.13, and 37 ($9.14\%$) severely underweight, 45 ($11.11\%$) underweight, 187 ($46.17\%$) healthy weight, 76 ($18.77\%$) overweight, and 60 ($14.81\%$) was in the obesity category. The mean total physical activity score of the participants was 2.96 ± 0.82, the mean total out of school score was 2.78 ± 0.94, the mean total school-based score was 3.31 ± 1.05, and the mean digital game addiction score was 1.89 ± 0.70 (Table 1).
**Table 1**
| Variable | Statistics | Value |
| --- | --- | --- |
| Gender | Gender | Gender |
| Girls | | 231.00 (57.04) |
| Boys | | 174.00 (42.96) |
| BMI | BMI | BMI |
| Severely underweight | n (%) | 37.00 (9.14) |
| Underweight | n (%) | 45.00 (11.11) |
| Healthy weight | n (%) | 187.00 (46.17) |
| Overweight | n (%) | 76.00 (18.77) |
| Obesity | n (%) | 60.00 (14.81) |
| Age (year) | M ± SD | 11.37 ± 1.45 |
| Height (cm) | M ± SD | 149.42 ± 11.16 |
| Weight (kg) | M ± SD | 44.22 ± 13.06 |
| TPA | M ± SD | 2.96 ± 0.82 |
| TPA-OS | M ± SD | 2.78 ± 0.94 |
| TPA-SB | M ± SD | 3.31 ± 1.05 |
| DGA | M ± SD | 1.89 ± 0.70 |
According to the findings of the study, age, height, and TBA scores of men and women were similar ($p \leq 0.05$). Body mass, TBA-OS, and TBA-SB scores, and DGA scores of women were significantly higher than men ($p \leq 0.05$). In addition, gender was significant for BMI and was higher in boys ($p \leq 0.05$; Table 2).
**Table 2**
| Variables | Boys (n = 174) | Girls (n = 231) | p-Value | ES |
| --- | --- | --- | --- | --- |
| Anthropometric | Boys (n = 174) | Girls (n = 231) | p-Value | ES |
| Age (year)* | 12 (3) | 12 (3) | 0.16 | 0.14 |
| Height (cm)** | 148.83 ± 10.77 | 150.21 ± 11.65 | 0.22 | 0.12 |
| Body mass (kg)* | 41 (18) | 45 (21) | 0.01 | 0.24 |
| BMI (kg/m2)** | 20.21 ± 4.56 | 18.93 ± 3.67 | 0.002 | 0.31 |
| Physical activity | Physical activity | Physical activity | Physical activity | Physical activity |
| TBA** | 3.00 ± 0.85 | 2.90 ± 0.78 | 0.26 | 0.11 |
| TPA-OS* | 2.66 (1.50) | 2.87 (1.16) | 0.03 | 0.21 |
| TPA-SB* | 3.00 (1.66) | 3.33 (1.33) | 0.04 | 0.20 |
| Game | Game | Game | Game | Game |
| DGA* | 1.66 (1.00) | 1.94 (0.78) | 0.003 | 0.29 |
Participants’ age, body mass, and BMI were significantly higher in the DGA group ($p \leq 0.05$). TPA and TPA-OS scores were similar in the groups ($p \leq 0.05$), but the TPA-SB score was higher in the non-DGA group ($p \leq 0.05$; Table 3).
**Table 3**
| Variables* | No-DGA (n = 367) | DGA (n = 38) | p-Value | ES |
| --- | --- | --- | --- | --- |
| Anthropometric | No-DGA (n = 367) | DGA (n = 38) | p-Value | ES |
| Age (year) | 11 (3) | 12 (2) | 0.019 | 0.23 |
| Height (cm) | 150 (17) | 153.50 (15.00) | 0.061 | 0.19 |
| Body mass (kg) | 42 (19) | 52.50 (20.00) | 0.001 | 0.38 |
| BMI (kg/m2) | 18.77 (5.86) | 22.10 (6.01) | 0.001 | 0.34 |
| Physical activity | Physical activity | Physical activity | Physical activity | Physical activity |
| TPA | 3.00 (1.11) | 2.66 (0.77) | 0.077 | 0.18 |
| TPA-OS | 2.66 (1.33) | 2.50 (1.16) | 0.238 | 0.12 |
| TPA-SB | 3.33 (1.33) | 3.00 (1.33) | 0.049 | 0.19 |
Table 4 shows the changes in physical activity and game addiction scale scores in BMI groups. Results showed that TPA, TPA-OS and TPA-SB scores were similar in BMI groups ($p \leq 0.05$). However, there was a statistically significant difference between the BMI groups in terms of DGA score, and further analysis revealed that the DGA score was significantly higher in the obesity group compared to the healthy weight group ($$p \leq 0.025$$; ES: 0.27).
**Table 4**
| Variable** | BMI Group* | BMI Group*.1 | BMI Group*.2 | BMI Group*.3 | BMI Group*.4 | p-Value# | ES |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Variable** | Severely underweight (n = 37) | Underweight (n = 45) | Healthy Weight (n = 187) | Overweight (n = 76) | Obesity (n = 60) | p-Value# | ES |
| Physical activity | Physical activity | Physical activity | Physical activity | Physical activity | Physical activity | Physical activity | Physical activity |
| TPA | 3.11a (1.22) | 2.66a (1.11) | 2.88a (1.16) | 3.00a (1.03) | 3.11a (1.16) | 0.19 | 0.14 |
| TPA-OS | 3.00a (1.33) | 2.50a (1.33) | 2.66a (1.41) | 2.66a (1.50) | 2.83a (1.04) | 0.23 | 0.12 |
| TPA-SB | 3.33a (1.00) | 3a (1) | 3.00a (1.33) | 3.00a (1.66) | 3.66a (1.41) | 0.25 | 0.11 |
| Game addiction | Game addiction | Game addiction | Game addiction | Game addiction | Game addiction | Game addiction | Game addiction |
| DGA | 1.77ab (0.88) | 1.88a (0.66) | 1.66b (0.83) | 1.88ab (0.88) | 1.88a (1.00) | 0.025 | 0.27 |
The results of the logistic regression model established with the forward feature selection method are given in Table 5. Age, gender, BMI and TPA scores, the coefficients of which were statistically significant in the model, were determined to be predictors of game addiction. The resulting logistic regression equation is as follows:
**Table 5**
| Variable | B | SE | Wald | p-Value | OR | 95% CI for OR | 95% CI for OR.1 | Interpretation |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Variable | B | SE | Wald | p-Value | OR | Lower | Upper | Interpretation |
| Age (year) | 0.124 | 0.061 | 4.126 | 0.042 | 1.132 | 1.004 | 1.277 | Increasing effect |
| Gender-boys | −0.953 | 0.174 | 30.062 | <0.001 | 0.386 | 0.274 | 0.542 | Reducing effect |
| BMI (kg/m2) | 0.145 | 0.023 | 39.327 | <0.001 | 1.157 | 1.105 | 1.210 | Increasing effect |
| TPA | −0.410 | 0.116 | 12.402 | <0.001 | 0.664 | 0.528 | 0.834 | Reducing effect |
| Constant | −3.384 | 0.824 | 16.886 | <0.001 | 0.034 | – | – | |
Our results revealed that one unit increase in age increased DGA risk by 1.132 times (OR = 1.132; $95\%$ CI [1.004–1.277]; $$p \leq 0.042$$). Moreover, women had a 2.59-fold greater risk of DGA compared to men (OR = 1.132; $95\%$ CI [1.004–1.277]; $$p \leq 0.042$$). BMI was a predictor of DGA, and a one-unit increase in BMI increased DGA risk 1.105-fold (OR = 1.132; $95\%$ CI [1.004–1.277]; $$p \leq 0.042$$). Furthermore, physical activity was found to be an important factor reducing DGA 1.51-fold (OR = 0.664; $95\%$ CI [0.528–0.834]; $$p \leq 0.035$$; Table 5).
In Table 6, the results of the performance criteria for the prediction of game addiction of the logistic regression model are reported. The accuracy rate calculated in the DGA estimation is $75.3\%$ and it can be said that the prediction performance of the model is quite high.
**Table 6**
| Metrics | Value (95% CI) |
| --- | --- |
| Accuracy | 0.753 (0.722–0.785) |
| Sensitivity | 0.747 (0.700–0.790) |
| Specificity | 0.760 (0.712–0.803) |
| Positive predictive value | 0.766 (0.719–0.808) |
| Negative predictive value | 0.741 (0.693–0.785) |
| F1-score | 0.756 (0.725–0.787) |
Figure 1 shows the results of the correlation analysis. When Figure 2 was examined, a small negative correlation was found between DGA and TPA, TPA-SB scores. As a results, as TPA and TPA-SB scores increase, DGA scores decrease ($p \leq 0.05$). There was a small negative correlation between BMI and TPA-SB, while a small positive correlation was observed between BMI and DGA scores ($p \leq 0.05$).
**Figure 1:** *The correlation graph between TPA, TPA-OS, TPA-SB, DGA scores, and BMI. TPA, total physical activity; TPA-OS, total physical activity out of school; TPA-SB, total physical activity-school based; DGA, digital game addiction; the correlation coefficients shown in the gray box represent statistical significance (p < 0.05); the size of the circle in the gray box indicates the level of correlation; blue color indicates positive and red color negative correlation.* **Figure 2:** *Study design.*
## Discussion
The primary aim of this study was to define the prevalence of obesity, physical activity levels and digital game addiction level in adolescents, and to investigate gender differences and relationships between outcomes. The second aim was to predict game addiction based on anthropometric measurements and physical activity levels. According to our findings, it was determined that digital game addiction has a negative relationship with physical activity level. We found that physical activity level had a negative relationship with BMI. In addition, physical activity reduces obesity and DGA. Game addiction levels of girl participants were significantly higher than boy participants, and we also found that obesity increased DGA. With the data obtained, it was determined that age, gender, BMI and total physical activity (TPA) score predicted DGA. The results showed that as age and BMI increase the risk of DGA, girls were 2.59-fold more likely than boys to develop DGA. More importantly, the results of this study showed that physical activity was an important factor in reducing DGA 1.51-fold. Our prediction model: Logit (P) = 1/(1 + exp(−(−3.384 + Age * 0.124 + Gender-boys * (−0.953) + BMI * 0.145 + TPA * (−0.410)))).
In a study, the game addiction levels of girls were found to be higher than boys. This result supports our study, and in our findings, it was determined that the game addiction levels of girl participants were higher than boy participants. However, in another study, the problematic internet use behavior of boys were found to be higher than girls (Durmus et al., 2021). According to studies, problematic internet behaviors are more common in adolescent boys than in adults and adolescent girls (Hancox et al., 2004; Tsai et al., 2009). This may be due to the negative effects of adolescence period. The reasons for the similarities and differences in the literature and our findings may be different due to factors such as the period in which the research was conducted, the welfare level of the country in which the research was conducted, and the region (rural or urban) in which the research was conducted. In addition, the main reason for the conflicting results in the literature may be due to factors such as the period of the research (e.g., pandemic). In another study on smartphone use, the daily calorie consumption, the number of steps taken per day were evaluated for individuals more screen time, and it was reported that they were more sedentary than those who spent less time in front of the screen (Kim et al., 2015). This result shows results in line with our findings. In this study, it was an expected result that there was a decrease in the level of physical activity with the increase in game addiction. In a study that supports our findings, physical activity is an important factor in solving the problem of digital game addiction (Hazar et al., 2017). Another study shows that there is a positive and significant relationship between middle school students’ physical activity participation and awareness of digital game addiction (Çar and Ahraz, 2022). It is possible that a study may have found that physical activity is an important factor in solving the problem of digital game addiction. In this study and literature may have found a positive and significant relationship between physical activity participation and awareness of digital game addiction. It is possible that engaging in physical activity may help individuals who are struggling with digital game addiction by providing a healthy outlet for stress and anxiety, and by promoting self-regulation and self-control. It is also important to note that, as with any study, it is important to consider the limitations and the sample size, and how generalizable the findings are to other populations before making a definitive conclusion. Additionally, it is important to keep in mind that digital game addiction is a complex issue and different interventions will be needed for different individuals, including psychological and social support, as well as education and awareness. Studies have shown that problematic internet use is associated with a high BMI and a sedentary lifestyle (Kautiainen et al., 2005; Tao, 2013). In parallel, another study reported associations between excessive internet, cell phone use, and long screen time and obesity among Saudi school-aged children (Alturki et al., 2020). These results are similar to our study. However, in another study in contrast, no correlation was found between problematic internet use and BMI and PA values in both girls and boys (Durmus et al., 2021). This could be due to a variety of factors such as differences in study design, sample size, and population characteristics. In the study Durmus et al. [ 2021] the researchers did not find a correlation between problematic internet use and BMI and PA values in both girls and boys. This suggests that other factors, such as diet and genetics, may play a larger role in the development of obesity among individuals who engage in problematic internet use.
Another study reported that a relationship between problematic internet use behavior and age variables in adolescents (Doğan, 2013; Al-Gamal et al., 2016). In another study, problematic internet use behavior was found to be lower in the 10–14 age group than in the 15–19 age group (Durmus et al., 2021). The different results in the literature are probably due to other factors (e.g., obesity can also be caused by nutritional status and some health problems). In addition, physical activity may differ depending on the region of residence. As a matter of fact, a study found that physical activity differs between rural and urban children (Gülü et al., 2022).
In addition to physical activity, diet has a positive effect on preventing obesity (Hills and Byrne, 2006). Norshakirah et al., found that Digital addiction among Malaysian adolescents can cause various impacts on physical health such as obesity, physical inactivity. These results in line with our findings. In this study from the data obtained, it was determined that age, gender, BMI and total physical activity (TPA) score predicted game addiction. The similarity between the findings of this study and those in the literature suggests that a sedentary lifestyle is probably one of the factors that cause obesity. It may also be that game addiction triggers a sedentary life. Another study found that obese children had lower physical activity levels than children with normal BMI (Gülü et al., 2022). Along with regular physical activity during the developmental period of the child, access to healthy food and proper nutrition is probably important determinants in the development of the physical structure of each child, in line with their genetic potential (Hills et al., 2007; Sallis and Glanz, 2009). In a study, it was found that the prevalence of obesity in adolescents was quite high ($37.6\%$ in boys, $32.9\%$ in girls; Durmus et al., 2021). These results are also consistent with our findings.
When we examine it as participation in physical activity, in the literature, adolescents exhibit a sedentary lifestyle, there is a significant relationship between physical activity rates and gender, and boys are more energetic than girls (Feldman et al., 2003; Yılmaz et al., 2014; Durmus et al., 2021). There appears to be a strong association between physical activity and obesity in children and adolescents (Hills et al., 2011). This result directly support our research. In this study, negative relationship was found between BMI and physical activity level. The effect of physical activity on BMI was limited in this study, meaning that it alone may be insufficient to prevent obesity. For this reason, to prevent obesity in children, it is necessary to increase participation in physical activity and to control their nutritional status. Future studies to determine the underlying causes of obesity may conduct more detailed studies on the nutritional status of children and physical activity level.
A limitation of this study is its cross-sectional methodology. Unfortunately, we were unable to collect longitudinal data in this first part of the study. Future research should include a longitudinal method to better represent the developmental course of this proposed disorder. Another our limitation is that the generalizability of the findings may be limited as the participants of this study were limited to Turkish students and the sample size was small. Additionally, since this study was conducted only on adolescent children, not all results can be generalized to the population. Finally, the findings of this study may be reporting bias due to the use of a self-report scale in data collection. More effective results could have been obtained if we had the opportunity to use smart wristbands that measure activity by providing more precise data, especially in the evaluation of physical activity. On the other hand, another limitation of this study is that obesity status was performed using height and body weight according to the WHO classification, and if we had the possibility of screening with a gold standard method in determining obesity in this study, it would have strengthened our results. In future studies, the effect of digital game addiction can be investigated in more detail by methods that can analyze self-efficacy, mental, and physical health conditions in more detail.
## Conclusion
The main finding of this study is that as the level of physical activity increases, game addiction will decrease. Regular physical activity should be encouraged and digital gaming hours can be limited to maintain ideal weight. In addition, adolescents should be encouraged to engage in physical activity in order to reduce digital game addiction level. This research provides important information in the prevention of DGA and obesity by increasing physical activity. According to these outputs, policy makers should carry out projects to increase participation in physical activity in schools and outside of school in order to reduce the time spent by adolescents on the screen in today’s technology age.
## Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
## Ethics statement
The studies involving human participants were reviewed and approved by Kirikkale University Social and Human Sciences Ethics Committee. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.
## Author contributions
MG: conceptualization and methodology. IG: data collection. EA, HY, and PP-G: analysis. MG, FY, and HN: writing—original draft preparation. MG, FC, AZ, LPA, PP-G, and HN: writing—review and editing. All authors have read and agreed to the published version of the manuscript.
## Funding
FC and this work are funded by the Fundação para a Ciência e Tecnologia/Ministério da Ciência, Tecnologia e Ensino Superior through national funds, and when applicable, co-funded by EU funds under the project UIDB/$\frac{50008}{2020.}$
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## References
1. Abarca-Gómez L., Abdeen Z. A., Hamid Z. A., Abu-Rmeileh N. M., Acosta-Cazares B., Acuin C.. **Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults**. *Lancet* (2017) **390** 2627-2642. DOI: 10.1016/S0140-6736(17)32129-3
2. Al-Gamal E., Alzayyat A., Ahmad M. M.. **Prevalence of I nternet addiction and its association with psychological distress and coping strategies among university students in J ordan**. *Perspect. Psychiatr. Care* (2016) **52** 49-61. DOI: 10.1111/ppc.12102
3. Alghadir A. H., Iqbal Z. A., Gabr A.. **The relationships of watching television, computer use, physical activity, and food preferences to body mass index: gender and nativity differences among adolescents in Saudi Arabia**. *Int. J. Environ. Res. Public Health* (2021) **18** 9915. DOI: 10.3390/ijerph18189915
4. Al-Khudairy L., Loveman E., Colquitt J. L., Mead E., Johnson R. E., Fraser H.. **Diet, physical activity and behavioural interventions for the treatment of overweight or obese adolescents aged 12 to 17 years**. *Cochrane Database Syst. Rev.* (2017) **6** CD012691. DOI: 10.1002/14651858.CD012691
5. Almourad M. B., Mcalaney J., Skinner T., Pleya M., Ali R.. **Defining digital addiction: key features from the literature**. *Psychology* (2020) **53** 237-253. DOI: 10.2298/PSI191029017A
6. Alturki H. A., Brookes D. S., Davies P. S.. **Does spending more time on electronic screen devices determine the weight outcomes in obese and normal weight Saudi Arabian children?**. *Saudi Med. J.* (2020) **41** 79-87. DOI: 10.15537/smj.2020.1.24786
7. Association E. S.. (2015)
8. Aubert S., Brazo-Sayavera J., González S. A., Janssen I., Manyanga T., Oyeyemi A. L.. **Global prevalence of physical activity for children and adolescents; inconsistencies, research gaps, and recommendations: a narrative review**. *Int. J. Behav. Nutr. Phys. Act.* (2021) **18** 81. DOI: 10.1186/s12966-021-01155-2
9. Batterham A. M., Hopkins W. G.. **Making meaningful inferences about magnitudes**. *Int. J. Sports Physiol. Perform.* (2006) **1** 50-57. DOI: 10.1123/ijspp.1.1.50
10. Bronikowska M., Krzysztoszek J., Łopatka M., Ludwiczak M., Pluta B.. **Comparison of physical activity levels in youths before and during a pandemic lockdown**. *Int. J. Environ. Res. Public Health* (2021) **18** 5139. DOI: 10.3390/ijerph18105139
11. Brown T., Moore T. H., Hooper L., Gao Y., Zayegh A., Ijaz S.. **Interventions for preventing obesity in children**. *Cochrane Database Syst. Rev.* (2019) **2019** CD001871. DOI: 10.1002/14651858.CD001871.pub4
12. Çar B., Ahraz A. O.. **A study on the relationship between secondary school students' digital game addiction awareness and participation motivation to physical activity**. *Int. J. Progress. Educ.* (2022) **18** 175-190. DOI: 10.29329/ijpe.2022.459.13
13. Cdc "Centers for Disease Control and Prevention (CDC) (2020). Percentile datafiles with LMS values. Available at: https://www.cdc.gov/growthcharts/percentile_data_files.htm (Accessed January 2023).. (2020)
14. Cdc "Centers for Disease Control and Prevention (CDC) (2022). Available at: https://www.cdc.gov/healthyweight/bmi/calculator.html (Accessed December 2022).. (2022)
15. Chen J., Huang H., Cohn A. G., Zhang D., Zhou M.. **Machine learning-based classification of rock discontinuity trace: SMOTE oversampling integrated with GBT ensemble learning**. *Int. J. Min. Sci. Technol.* (2022) **32** 309-322. DOI: 10.1016/j.ijmst.2021.08.004
16. Christakis D. A.. **The challenges of defining and studying “digital addiction” in children**. *JAMA* (2019) **321** 2277-2278. DOI: 10.1001/jama.2019.4690
17. Cohen J.. **The effect size**. *Statistical Power Analysis for the Behavioral Sciences* (1988) 77-83
18. Consumer D. M.. (2015)
19. Doğan A.. (2013)
20. Dumith S. C., Gigante D. P., Domingues M. R., Kohl H. W.. **Physical activity change during adolescence: a systematic review and a pooled analysis**. *Int. J. Epidemiol.* (2011) **40** 685-698. DOI: 10.1093/ije/dyq272
21. Durmus G., Ortabag T., Ozdemir S.. **Determining the relationship between obesity and problematic internet use among adolescents**. *Iran. J. Public Health* (2021) **50** 1796-1804. DOI: 10.18502/ijph.v50i9.7052
22. Erdim L., Ergün A., Kuğuoğlu S.. **Reliability and validity of the Turkish version of the physical activity questionnaire for older children (PAQ-C)**. *Turk. J. Med. Sci.* (2019) **49** 162-169. DOI: 10.3906/sag-1806-212
23. Faul F., Erdfelder E., Lang A.-G., Buchner A.. **G* power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences**. *Behav. Res. Methods* (2007) **39** 175-191. DOI: 10.3758/BF03193146
24. Feldman D. E., Barnett T., Shrier I., Rossignol M., Abenhaim L.. **Is physical activity differentially associated with different types of sedentary pursuits?**. *Arch. Pediatr. Adolesc. Med.* (2003) **157** 797-802. DOI: 10.1001/archpedi.157.8.797
25. Gazanfer A., Taş İ.. **Ergenler için oyun bağimliliği ölçeği kisa formunun geçerlilik ve güvenirlik çalişmasi. Electronic**. *Turk. Stud.* (2018) **13** 189-203
26. Gök E. C., Olgun M. O.. **SMOTE-NC and gradient boosting imputation based random forest classifier for predicting severity level of covid-19 patients with blood samples**. *Neural Comput. & Applic.* (2021) **33** 15693-15707. DOI: 10.1007/s00521-021-06189-y
27. Gülü M., Yapici H., Mainer-Pardos E., Alves A. R., Nobari H.. **Investigation of obesity, eating behaviors and physical activity levels living in rural and urban areas during the covid-19 pandemic era: a study of Turkish adolescent**. *BMC Pediatr.* (2022) **22** 405. DOI: 10.1186/s12887-022-03473-1
28. Hancox R. J., Milne B. J., Poulton R.. **Association between child and adolescent television viewing and adult health: a longitudinal birth cohort study**. *Lancet* (2004) **364** 257-262. DOI: 10.1016/S0140-6736(04)16675-0
29. Hazar Z., Tekkurşun Demİr G., Namli S., Türkeli A.. **Investigation of the relationship between digital game addiction and physical activity levels of secondary school students**. *J. Phys. Educ. Sports Sci.* (2017) **11** 320-332
30. Hills A. P., Andersen L. B., Byrne N. M.. **Physical activity and obesity in children**. *Br. J. Sports Med.* (2011) **45** 866-870. DOI: 10.1136/bjsports-2011-090199
31. Hills A. P., Byrne N. M.. **State of the science: a focus on physical activity**. *Asia Pac. J. Clin. Nutr.* (2006) **15 Suppl** 40-48. PMID: 16928660
32. Hills A. P., King N. A., Armstrong T. P.. **The contribution of physical activity and sedentary behaviours to the growth and development of children and adolescents**. *Sports Med.* (2007) **37** 533-545. DOI: 10.2165/00007256-200737060-00006
33. Karasar N.. *Bilimsel araştırma yöntemi (31 Basım)* (2016)
34. Kautiainen S., Koivusilta L., Lintonen T., Virtanen S. M., Rimpelä A.. **Use of information and communication technology and prevalence of overweight and obesity among adolescents**. *Int. J. Obes.* (2005) **29** 925-933. DOI: 10.1038/sj.ijo.0802994
35. Kim S.-E., Kim J.-W., Jee Y.-S.. **Relationship between smartphone addiction and physical activity in Chinese international students in Korea**. *J. Behav. Addict.* (2015) **4** 200-205. DOI: 10.1556/2006.4.2015.028
36. Kohl H. W., Craig C. L., Lambert E. V., Inoue S., Alkandari J. R., Leetongin G.. **The pandemic of physical inactivity: global action for public health**. *Lancet* (2012) **380** 294-305. DOI: 10.1016/S0140-6736(12)60898-8
37. Kowalski K. C., Crocker P. R., Donen R. M.. *The Physical Activity Questionnaire for Older Children (PAQ-C) and Adolescents (PAQ-A) Manual* (2004) 1-38
38. Kracht C. L., Joseph E. D., Staiano A. E.. **Video games, obesity, and children**. *Curr. Obes. Rep.* (2020) **9** 1-14. DOI: 10.1007/s13679-020-00368-z
39. Kumar B., Robinson R., Till S.. **Physical activity and health in adolescence**. *Clin. Med.* (2015) **15** 267-272. DOI: 10.7861/clinmedicine.15-3-267
40. Lemmens J. S., Valkenburg P. M., Peter J.. **Development and validation of a game addiction scale for adolescents**. *Media Psychol.* (2009) **12** 77-95. DOI: 10.1080/15213260802669458
41. Li M., Deng Y., Ren Y., Guo S., He X.. **Obesity status of middle school students in Xiangtan and its relationship with internet addiction**. *Obesity* (2014) **22** 482-487. DOI: 10.1002/oby.20595
42. Mead E., Brown T., Rees K., Azevedo L. B., Whittaker V., Jones D.. **Diet, physical activity and behavioural interventions for the treatment of overweight or obese children from the age of 6 to 11 years**. *Cochrane Database Syst. Rev.* (2017) **2017** CD012651. DOI: 10.1002/14651858.CD012651
43. Neyzi O., Gunoz H., Furman A., Bundak R., Gokcay G., Darendeliler F.. **Weight, height, head circumference and body mass index references for Turkish children**. *Cocuk Sagligi ve Hastalikari Dergisi* (2008) **51** 1
44. O’Brien W., Belton S., Fitzpatrick B., Shannon S., Brennan D., Chambers F.. **Relationship between gender, physical activity, screen time, body mass index and wellbeing in Irish children from social-disadvantage**. *Child Care Pract.* (2021) 1-15. DOI: 10.1080/13575279.2021.1887815
45. Ogden C. L., Carroll M. D., Fakhouri T. H., Hales C. M., Fryar C. D., Li X.. **Prevalence of obesity among youths by household income and education level of head of household—United States 2011–2014**. *Morb. Mortal. Wkly Rep.* (2018) **67** 186-189. DOI: 10.15585/mmwr.mm6706a3
46. Ogden C. L., Flegal K. M.. (2010)
47. Paksoy N., Yağin F. H.. **Artificial intelligence-based colon cancer prediction by identifying genomic biomarkers**. *Med. Records* (2022) **4** 196-202. DOI: 10.37990/medr.1077024
48. Pombo A., Luz C., Rodrigues L. P., Ferreira C., Cordovil R.. **Correlates of children's physical activity during the COVID-19 confinement in Portugal**. *Public Health* (2020) **189** 14-19. DOI: 10.1016/j.puhe.2020.09.009
49. Pühse U., Gerber M.. *International Comparison of Physical Education: Concepts, Problems, Prospects* (2005)
50. Rhodes R. E., Janssen I., Bredin S. S., Warburton D. E., Bauman A.. **Physical activity: health impact, prevalence, correlates and interventions**. *Psychol. Health* (2017) **32** 942-975. DOI: 10.1080/08870446.2017.1325486
51. Sallis J. F., Glanz K.. **Physical activity and food environments: solutions to the obesity epidemic**. *Milbank Q.* (2009) **87** 123-154. DOI: 10.1111/j.1468-0009.2009.00550.x
52. Stevens G. A., Singh G. M., Lu Y., Danaei G., Lin J. K., Finucane M. M.. **National, regional, and global trends in adult overweight and obesity prevalences**. *Popul. Health Metrics* (2012) **10** 22. DOI: 10.1186/1478-7954-10-22
53. Storz M. A.. **The COVID-19 pandemic: an unprecedented tragedy in the battle against childhood obesity**. *Clin. Exp. Pediatr.* (2020) **63** 477-482. DOI: 10.3345/cep.2020.01081
54. Strong W. B., Malina R. M., Blimkie C. J., Daniels S. R., Dishman R. K., Gutin B.. **Evidence based physical activity for school-age youth**. *J. Pediatr.* (2005) **146** 732-737. DOI: 10.1016/j.jpeds.2005.01.055
55. Tao Z.. **The relationship between internet addiction and bulimia in a sample of Chinese college students: depression as partial mediator between internet addiction and bulimia**. *Eat. Weight Disord.* (2013) **18** 233-243. DOI: 10.1007/s40519-013-0025-z
56. Telama R., Yang X., Leskinen E., Kankaanpää A., Hirvensalo M., Tammelin T.. **Tracking of physical activity from early childhood through youth into adulthood**. *Med. Sci. Sports Exerc.* (2014) **46** 955-962. DOI: 10.1249/MSS.0000000000000181
57. Tsai H. F., Cheng S. H., Yeh T. L., Shih C.-C., Chen K. C., Yang Y. C.. **The risk factors of internet addiction—a survey of university freshmen**. *Psychiatry Res.* (2009) **167** 294-299. DOI: 10.1016/j.psychres.2008.01.015
58. Van Jaarsveld C. H., Gulliford M. C.. **Childhood obesity trends from primary care electronic health records in England between 1994 and 2013: population-based cohort study**. *Arch. Dis. Child.* (2015) **100** 214-219. DOI: 10.1136/archdischild-2014-307151
59. Von Elm E., Altman D. G., Egger M., Pocock S. J., Gøtzsche P. C., Vandenbroucke J. P.. **The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies**. *Lancet* (2007) **370** 1453-1457. DOI: 10.1016/S0140-6736(07)61602-X
60. Wei R., Ogden C. L., Parsons V. L., Freedman D. S., Hales C. M.. **A method for calculating BMI z-scores and percentiles above the 95th percentile of the CDC growth charts**. *Ann. Hum. Biol.* (2020) **47** 514-521. DOI: 10.1080/03014460.2020.1808065
61. WHO (2010). Population-based prevention strategies for childhood obesity: Report of a WHO forum and technical meeting, Geneva, 15–17 December 2009.. (2010)
62. WHO (2019). Global Action Plan on Physical Activity 2018–2030: More Active People for a Healthier World. Geneva, Switzerland: World Health Organization.. *Global Action Plan on Physical Activity 2018–2030: More Active People for a Healthier World* (2019)
63. WHO (2022). Adolescent health. (n.d.). [Online]. WORLD HEALTH ORGANIZATION (WHO). Available at: https://www.who.int/health-topics/adolescent-health#tab=tab_1 (Accessed November 22, 2022).. (2022)
64. Wyatt S. B., Winters K. P., Dubbert P. M.. **Overweight and obesity: prevalence, consequences, and causes of a growing public health problem**. *Am J Med Sci* (2006) **331** 166-174. DOI: 10.1097/00000441-200604000-00002
65. Yağin B., Yağin F. H., Gözükara H., Colak C.. **A web-based software for reporting guidelines of scientific researches**. *J. Cogn. Syst.* (2021) **6** 39-43. DOI: 10.52876/jcs.916182
66. Yılmaz E., Şahin Y. L., Haseski H. İ., Osman E.. **Lise öğrencilerinin internet bağımlılık düzeylerinin çeşitli değişkenlere göre incelenmesi: Balıkesir ili örneği**. *Eğitim Bilimleri Araştırmaları Dergisi* (2014) **4** 133-144. DOI: 10.12973/jesr.2014.41.7
67. Zurita-Ortega F., Chacón-Cuberos R., Castro-Sánchez M., Gutiérrez-Vela F. L., González-Valero G.. **Effect of an intervention program based on active video games and motor games on health indicators in university students: a pilot study**. *Int. J. Environ. Res. Public Health* (2018) **15** 1329. DOI: 10.3390/ijerph15071329
|
---
title: Sex-dependent association analysis between serum uric acid and spontaneous
hemorrhagic transformation in patients with ischemic stroke
authors:
- Ye Tang
- Ming-Su Liu
- Chong Fu
- Guang-Qin Li
journal: Frontiers in Neurology
year: 2023
pmcid: PMC10022728
doi: 10.3389/fneur.2023.1103270
license: CC BY 4.0
---
# Sex-dependent association analysis between serum uric acid and spontaneous hemorrhagic transformation in patients with ischemic stroke
## Abstract
### Objective
The association between serum uric acid (UA) and spontaneous hemorrhagic transformation (HT) has been seldom studied, and the role of UA in spontaneous HT remains unclear. This study aims to investigate the sex-dependent association between UA and spontaneous HT in patients with ischemic stroke.
### Method
We retrospectively included patients with ischemic stroke in a tertiary academic hospital between December 2016 and May 2020. Patients were included if they presented within 24 h after the onset of symptoms and did not receive reperfusion therapy. Spontaneous HT was determined by an independent evaluation of neuroimaging by three trained neurologists who were blinded to clinical data. A univariate analysis was performed to identify factors related to spontaneous HT. Four logistic regression models were established to adjust each factor and assess the association between UA and spontaneous HT.
### Results
A total of 769 patients were enrolled ($64.6\%$ were male patients and $3.9\%$ had HT). After adjusting the confounders with a $P \leq 0.05$ (model A) in the univariate analysis, the ratio of UA and its interquartile range (RUI) was independently associated with spontaneous HT in male patients (OR: 1.85; $95\%$ CI: 1.07–3.19; $$P \leq 0.028$$), but not in female patients (OR: 1.39; $95\%$ CI: 0.28–6.82; $$P \leq 0.685$$). In models B–D, the results remain consistent with model A after the adjustment for other potential confounders.
### Conclusions
Higher serum UA was independently associated with a higher occurrence of spontaneous HT in male patients who were admitted within 24 h after the stroke onset without receiving reperfusion therapy.
## Introduction
Spontaneous hemorrhagic transformation (HT) is defined as the blood stain of an infarcted cerebral area formed by the overflow of red blood cells and other blood components from blood vessels to the infarcted brain tissue, which is a part of the natural course of ischemic stroke and a crucial complication of treatment [1]. Spontaneous HT occurs in ~13–$43\%$ of patients with ischemic stroke, and parenchymal hematoma is a critical factor in poor outcomes [2]. Thus, it is important to identify the factors that determine the occurrence of HT. However, the pathophysiological mechanism of spontaneous HT remains uncertain.
Uric acid (UA) is an endogenous antioxidant produced by purine metabolism [3, 4]. If the antioxidant substances are abundant, UA will show antioxidant properties. If there are more pro-oxidant substances, it will show pro-oxidant properties [5]. In patients with acute ischemic stroke, oxygen free radicals will be produced after tissue ischemia–reperfusion, and UA presents antioxidant or pro-oxidant properties depending on the surrounding substances. It has been demonstrated that the dose–response relationship between UA and HT and higher UA was independently associated with a lower incidence of HT. On the contrary, higher UA levels are reported to be associated with a lower incidence of HT in different settings [6]. An examination of UA is widely available in almost all clinical settings. For these reasons, UA may be a protective factor for spontaneous HT. Therefore, the critical clinical significance of the relationship between UA and spontaneous HT is a topic of research interest.
Nevertheless, reperfusion injury and blood–brain barrier damage after infarction are considered the two major causes of spontaneous HT, and the reactive oxygen species (ROS)-mediated oxidative stress response has an important role in these two mechanisms [7]. Many studies have explored the role of thrombolysis in HT occurrence [1, 7, 8]. Although thrombectomy is not independently associated with spontaneous HT [9], given its mechanism of reperfusion therapy (e.g., thrombolysis or thrombectomy), the restoration of blood flow to the salvageable ischemic brain tissue is consistent with the aforementioned mechanism of spontaneous HT and the high incidence of spontaneous HT found by previous studies (10–12). None of these prior studies assessed spontaneous HT with respect to non-reperfusion strategies.
There is no consensus on the association between UA and spontaneous HT in patients with acute ischemic stroke. Studies of the relationship between UA levels and spontaneous HT are contradictory. Positive and negative in the male population or both positive in men and women have been described [8, 10, 13, 14]. Furthermore, Brouns et al. found that UA changed with time in patients with stroke and exhibited a U-shaped curve in general, which decreased within 7 days after the stroke onset and then gradually increased to the baseline value [15]. Few studies have explored the impact of UA levels in specific stroke subtypes and treatment strategies in the acute stage. UA levels are sex-dependent and are higher in males. Therefore, a sex-dependent explorative analysis was made using patients with acute ischemic stroke within 24 h after the stroke onset and who did not receive reperfusion therapy (thrombolysis or thrombectomy) after the onset to investigate whether UA was associated with spontaneous HT.
## Population
We retrospectively reviewed the medical records of patients with ischemic stroke admitted to the Department of Neurology, the First Affiliated Hospital of Chongqing Medical University, from December 2016 to May 2020. For this analysis, the patients were included if they: [1] met the diagnostic criteria of acute ischemic stroke (AIS) in the Guidelines for Early Management of Patients With Acute Ischemic Stroke [2019] of the American Heart Association (AHA) [14], [2] were admitted within 24 h from the onset, [3] completed a serum UA test within 24 h after admission, and [4] had an initial neuroimaging scan [computed tomography (CT) scanning or magnetic resonance imaging (MRI)] within 24 h after admission and at least one follow-up neuroimaging scan within 7 days after admission. The exclusion criteria were as follows: [1] patients who received reperfusion therapy (thrombolysis or thrombectomy) after the onset, [2] patients with platelet abnormalities or coagulation dysfunction, [3] patients who received UA-lowering treatment within 1 month before admission, and [4] patients with intracranial arteriovenous malformation or tumor or head trauma. The First Affiliated Hospital of Chongqing Medical University Institutional Review Board approved this study. Written informed consent was obtained from participants or their legal representatives.
## Data collection
The clinical data were collected from each patient by two researchers: [1] demographic characteristics, such as age and sex; [2] medical histories, such as the history of smoking, alcohol consumption, hypertension, diabetes, dyslipidemia, and atrial fibrillation (AF); [3] clinical variables, such as National Institute of Health Stroke Scale (NIHSS), the Trial of ORG 10172 in Acute Stroke classification (TOAST), systolic blood pressure (SBP), diastolic blood pressure (DBP), and time from the stroke onset to admission; [4] laboratory tests, such as platelet count, activated partial thromboplastin time, serum UA, estimated glomerular filtration rate (eGFR), serum creatinine, low-density lipoprotein cholesterol (LDL-C), and hemoglobin A1c (HbA1c); [5] radiological characteristics, such as large hemispheric infarction (LHI) and spontaneous HT; and [6] treatment, such as anticoagulants, antiplatelet drugs, antihypertensive drugs, and antidiabetic drugs. Among them, eGFR was calculated by the serum creatinine level according to the formula of the Chronic Kidney Disease Epidemiology Collaboration [16]. The cerebral infarct, of size >$\frac{2}{3}$ of MCA territory, was defined as LHI [17].
Serum UA concentration was tested by the enzymatic method (Roche Cobas C701) or the dry chemistry method (Ortho-Clinical Diagnostics). The diagnosis of spontaneous HT is based on the following criteria: abnormal hyperdensity within the area of low attenuation (CT) or abnormal hypointensity within the identified ischemic area (MRI) [13]. The images were evaluated by two neurologists who were blinded to the patient's information. For inconsistent interpretations, the imaging was independently assessed by another senior neurologist, and the final diagnosis was determined on the principle of subordination of the minority to the majority. Furthermore, we classified spontaneous HT into four subtypes [type 1 and 2 hemorrhagic infarction (HI1 and 2) and type 1 and 2 parenchymal hemorrhage (PH1 and 2)] according to the European Cooperative Acute Stroke Study III (ECASS III) [18].
## Statistical analysis
Since there is no clinical significance of a 1-unit (1 μmol/L) change in UA in clinical practice, in this study, the ratio of UA and its interquartile range [RUI, male: RUI = (UA of individual male patient)/(IQR of UA in the male group), female: RUI = (UA of individual female patient)/(IQR of UA in the female group)] was used to replace UA in the statistical analysis, for increasing the practicability of the conclusions in clinical diagnosis and treatment. Continuous variables were expressed as the mean and standard deviation, and categorical variables were expressed as frequency and percentage. The comparison of continuous variables between groups was made by performing the t-test or Mann–Whitney U-test, whereas the comparison of categorical variables was made by performing the chi-square test or Fisher's exact test. In addition, the factors with a $P \leq 0.05$ in the univariate analysis and other factors that potentially could affect the study results were included in the subsequent logistic regression analysis. UA levels are lower in female patients, and a sex-dependent association between UA and cardiovascular disease was reported. Therefore, sex-dependent analysis was performed to investigate the impact of UA levels on HT occurrence. In total, four logistic regression models were built, and the association between UA and spontaneous HT was determined by dividing patients into two subgroups of male and female. These variables were chosen based on their known associations with the occurrence of HT, and their demonstrated link to HT in the logistical regression: Model A is adjusted for variables with a $P \leq 0.05$ in male (or female) patient subgroup univariate analysis; model B is adjusted for variables with a $P \leq 0.05$ in both subgroup univariate analysis; model C is adjusted for variables in model B, antiplatelet treatment, and anticoagulant treatment; model D is adjusted for variables in model C, smoking, alcohol consumption, systolic blood pressure, and eGFR. A $P \leq 0.05$ was considered statistically significant. Data analysis of the present study was performed by using SPSS Statistics Software (version 26.0; IBM Corporation) and GraphPad Prism (version 7.0; GraphPad Software Corporation).
## Results
A total of 769 patients were finally included in this study (Figure 1), of whom $64.6\%$ were male patients, with a mean age of (66.9 ± 12.5) years and 30 ($3.9\%$) had spontaneous HT. In this study, $70\%$ of spontaneous HT was diagnosed by CT, $30\%$ of spontaneous HT was diagnosed by MRI (T2WI and T1WI), $13.3\%$ of patients with spontaneous HT performed SWI, and the result of SWI supports the diagnosis of CT/MRI. No patient with spontaneous HT was diagnosed by SWI alone. Among patients with spontaneous HT, one ($3.3\%$) patient with PH2, seven ($23.3\%$) patients with PH1, 13 ($43.3\%$) patients with HI2, and nine ($30.0\%$) patients with HI1 were identified. Compared with female patients, the male patients had a higher UA level (362.8 ± 96.0 vs. 304.8 ± 87.5, respectively; $P \leq 0.001$), RUI (3.2 ± 0.8 vs. 2.7 ± 0.8, respectively; $P \leq 0.001$), drinking, smoking, creatinine level, and eGFR and antiplatelet drug use rate. Female patients were older (70.3 ± 12.0 vs. 65.0 ± 12.4, respectively, $P \leq 0.001$) and have higher occurrences of AF (17.3 vs. $11.5\%$, respectively, $$P \leq 0.024$$) and anticoagulant use than those of men (Table 1).
**Figure 1:** *Flowchart of patients' selection.* TABLE_PLACEHOLDER:Table 1 The male patients with spontaneous HT tended to have higher UA levels (428.3 ± 124.5 vs. 360.3 ± 94.0, respectively, $$P \leq 0.003$$), RUI (3.7 ± 1.1 vs. 3.1± 0.8, respectively, $$P \leq 0.003$$), age, admission NIHSS score, higher occurrence of AF and LHI, and shorter time from the onset to admission compared to patients without. However, there was no significant association between UA/RUI and spontaneous HT in the female patients ($$P \leq 0.336$$) (Table 2).
**Table 2**
| Variables | Male (n = 497) | Male (n = 497).1 | Male (n = 497).2 | Female (n = 272) | Female (n = 272).1 | Female (n = 272).2 |
| --- | --- | --- | --- | --- | --- | --- |
| | With HT | Without HT | P value | With HT | Without HT | P value |
| Demographic | Demographic | Demographic | Demographic | Demographic | Demographic | Demographic |
| Mean age, y (SD) | 71.3 (13.8) | 64.8 (12.3) | 0.028 | 73.5 (14.5) | 70.2 (11.9) | 0.451 |
| Medical history | Medical history | Medical history | Medical history | Medical history | Medical history | Medical history |
| Alcohol consumption, n (%) | 10 (55.6%) | 241 (50.3%) | 0.662 | 0 (0.0%) | 9 (3.5%) | 1.000 |
| Smoking, n (%) | 13 (72.2%) | 340 (71.0%) | 0.909 | 0 (0.0%) | 13 (5.0%) | 1.000 |
| Hypertension, n (%) | 14 (77.8%) | 347 (72.4%) | 0.790 | 9 (75.0%) | 190 (73.1%) | 1.000 |
| Diabetes mellitus, n (%) | 9 (50.0%) | 143 (29.9%) | 0.690 | 4 (33.3%) | 79 (30.4%) | 0.760 |
| Dyslipidemia, n (%) | 6 (33.3%) | 91 (19.0%) | 0.136 | 2 (16.7%) | 44 (16.9%) | 1.000 |
| Atrial fibrillation, n (%) | 6 (33.3%) | 51 (10.6%) | 0.011 | 8 (66.7%) | 39 (15.0%) | < 0.001 |
| Clinical features | Clinical features | Clinical features | Clinical features | Clinical features | Clinical features | Clinical features |
| Time from onset to admission, h (SD) | 8.5 (8.0) | 14.3 (8.5) | 0.007 | 9.8 (9.2) | 14.1 (8.4) | 0.082 |
| Systolic blood pressure, mmHg (SD) | 148.8 (22.5) | 151.9 (23.8) | 0.586 | 161.7 (29.9) | 153.6 (24.1) | 0.265 |
| Diastolic blood pressure, mmHg (SD) | 83.9 (13.3) | 88.5 (16.1) | 0.236 | 85.9 (11.1) | 86.4 (14.7) | 0.901 |
| Admission NIHSS score, mean (SD) | 11.8 (8.1) | 4.4 (4.6) | 0.001 | 10.3 (6.8) | 4.7 (4.7) | 0.015 |
| TOAST classification, n (%) | | | < 0.001 | | | 0.023 |
| Large-artery atherosclerosis | 9 (50.0%) | 225 (47.0%) | 0.801 | 8 (66.7%) | 102 (39.2%) | 0.073 |
| Small-artery occlusion | 0 (0.0%) | 191 (39.9%) | 0.001 | 0 (0.0%) | 100 (38.5%) | 0.005 |
| Cardio-embolism | 7 (38.9%) | 46 (9.6%) | 0.001 | 4 (33.3%) | 42 (16.2%) | 0.126 |
| Undetermined etiology | 2 (11.1%) | 13 (2.7%) | 0.098 | 0 (0.0%) | 10 (3.8%) | 1.000 |
| Other etiology | 0 (0.0%) | 4 (0.8%) | 1.000 | 0 (0.0%) | 6 (2.3%) | 1.000 |
| Laboratorial index | Laboratorial index | Laboratorial index | Laboratorial index | Laboratorial index | Laboratorial index | Laboratorial index |
| Platelet count, *109/L (SD) | 179.7 (70.6) | 194.2 (76.0) | 0.438 | 178.5 (72.4) | 199.8 (65.4) | 0.294 |
| APTT, s (SD) | 25.7 (3.5) | 26.02 (4.9) | 0.795 | 27.0 (5.2) | 25.9 (8.0) | 0.662 |
| Serum UA, μmol/L (SD) | 428.3 (124.5) | 360.3 (94.0) | 0.003 | 328.6 (57.5) | 303.7 (88.6) | 0.336 |
| RUI, mean (SD) | 3.7 (1.1) | 3.1 (0.8) | 0.003 | 2.9 (0.5) | 2.7 (0.8) | 0.336 |
| Serum creatinine, μmol/L(SD) | 88.9 (36.6) | 82.3 (32.7) | 0.401 | 68.6 (16.6) | 65.0 (20.9) | 0.550 |
| eGFR, mL/min/1.73 m2, (SD) | 77.7 (25.7) | 86.4 (20.8) | 0.086 | 76.7 (18.6) | 82.4 (19.4) | 0.320 |
| HbA1c, %,(SD) | 7.2 (2.3) | 6.7 (1.6) | 0.158 | 5.6 (2.1) | 6,8 (1.9) | 0.050 |
| LDL-C, μmol/L (SD) | 2.7 (0.8) | 2.9 (1.1) | 0.597 | 3.1 (1.1) | 3.0 (1.8) | 0.860 |
| Radiological characteristics | Radiological characteristics | Radiological characteristics | Radiological characteristics | Radiological characteristics | Radiological characteristics | Radiological characteristics |
| Large hemispheric infarction, n (%) | 10 (55.6%) | 50 (10.4%) | < 0.001 | 9 (75%) | 16 (6.2%) | < 0.001 |
| Treatment | Treatment | Treatment | Treatment | Treatment | Treatment | Treatment |
| Antiplatelet, n (%) | 18 (100%) | 468 (97.7%) | 1.000 | 10 (83.3%) | 243 (93.5%) | 0.201 |
| Anticoagulant, n (%) | 4 (22.2%) | 41 (8.6%) | 0.700 | 1 (8.3%) | 48 (18.5%) | 0.700 |
| Antihypertensive, n (%) | 8 (44.4%) | 264 (55.1%) | 0.372 | 4 (33.3%) | 131 (50.4%) | 0.248 |
| Antidiabetic, n (%) | 8 (44.4%) | 132 (27.6%) | 0.118 | 4 (33.3%) | 68 (26.2%) | 0.524 |
After adjustment for factors with a $P \leq 0.05$ (model A) in the univariate analysis of each group by logistic regression, the ratio of UA/IQR was found to be independently associated with spontaneous HT in male patients (OR: 1.85; $95\%$ CI: 1.07–3.19; $$P \leq 0.028$$), but not in female patients (OR: 1.39; $95\%$ CI: 0.28–6.82; $$P \leq 0.685$$). Furthermore, in the other three multivariate logistic regression models, the statistical results were consistent with model A after being adjusted for the factors with a $P \leq 0.05$ in the univariate analyses of both subgroups (model B), anticoagulant use and antiplatelet drug (model C), and smoking, alcohol consumption, SBP, and eGFR (model D) (Figure 2).
**Figure 2:** *The multivariate analysis to identify the association between RUI and spontaneous HT. Variables adjusted in logistic regression models: Model A. Factors with a P < 0.05 in male (or female) subgroup univariate analysis were included; model B: factors with a P < 0.05 in both subgroup univariate analysis were included; model C: variables in model B plus antiplatelet treatment and anticoagulant treatment; and model D: variables in model C plus smoking, alcohol consumption, systolic blood pressure, and eGFR.*
## Discussion
In this study, the RUI was independently associated with spontaneous HT in male patients admitted within 24 h after the onset, and the incidence of spontaneous HT increased by $85.0\%$ for each IQR increase in the UA level. Interestingly, no similar association between the UA level and spontaneous HT was found in female patients.
Furthermore, we reported that UA levels were associated with spontaneous HT in male patients with acute ischemic stroke. However, this association was not found in female patients. UA levels are commonly available in medical settings, and the results of our study suggested that UA may be a potential target for interventions. Several previous studies have investigated the sex differences of UA in patients with cerebrovascular diseases [19, 20]. Recently, a similar study reported that the incidence of spontaneous HT was higher in patients with low UA levels than in patients with high UA levels [13]. They included 1,230 patients who received reperfusion therapy within 7 days from the onset of the symptoms. However, in the context of our study, this finding was not confirmed. The reason for these contradictory results may partly be due to inclusion criteria. Moreover, it has been proven that the UA level of patients with stroke decreased gradually within 7 days after the onset, but there was no significant difference between the UA level measured 24 h after admission [15]. This also may reflex the dual effect of UA. In a cross-sectional study of 2,686 patients, Jeong et al. reported that a high UA level was a risk factor for cerebral microbleeds only in male patients [19], and we confirmed and extended this finding in our study. Further studies are needed to explore whether these patients are potential candidates for interventions.
In previous studies, reperfusion treatment is one of the mechanisms of spontaneous HT, subsequently affecting the outcome. Previous studies regarding the relationship between UA and spontaneous HT have shown conflicting results in the thrombolysis population and non-thrombolysis group. Thrombectomy, the restoration of blood flow to the salvageable ischemic brain tissue, is consistent with the aforementioned mechanism of spontaneous HT, and a higher incidence of spontaneous HT was reported in previous studies. The reason for these contradictory results may partly be due to the modifying effect of reperfusion strategies on spontaneous HT in these studies. Thus, we excluded those patients from this study.
The exact underlying mechanism of UA levels on spontaneous HT remains unknown. Generally, UA is an abundant antioxidant in humans and is supposed to play a protective role in cardio-cerebral vascular diseases. The possible explanation of sex-dependent differences in UA levels on spontaneous HT was the uricosuric effect of estrogen [21], the inhibition of oxidative stress of blood vessels by estrogen [22], and the redox shuttle mechanism of UA [23]. These three factors result in higher UA and lower antioxidant capacity in male patients than in female patients. In addition, UA is more effective in promoting oxidation in an environment with relatively lower-antioxidative substances. Therefore, the stronger oxidation-promoting property of UA in male patients may be responsible for the sex difference in the occurrence of spontaneous HT. However, the opposite result has been found in many large-scale clinical studies (24–26). A literature review revealed that UA, which carries over half of the antioxidant capacity in plasma, may be involved in spontaneous HT through oxidative stress [4]. This involvement can be partly explained by the following reasons: first, the production of UA by xanthine oxidase itself produces oxygen free radicals [27] and second, more oxygen free radicals will be produced after ischemia–reperfusion [5]. UA has a redox shuttle effect in which the presentation of the antioxidant or pro-oxidant properties of UA depends on the surrounding environment. Specifically, antioxidant activity occurs when antioxidant substances are abundant, and pro-oxidant activity occurs if there are more pro-oxidants [23]. In the environment of more oxygen free radicals in the ischemia–reperfusion tissue, UA tends to be pro-oxidative. Therefore, UA may further aggravate oxidative stress and increase blood–brain barrier damage through the aforementioned mechanisms, which leads to spontaneous HT.
It should be noted that our study had some limitations. First, it was a single-center retrospective study with a relatively small sample size. The impact of UA on spontaneous HT seems to be limited in the sex-specific subgroups, and this clinical relevance may not be generalizable to patients with reperfusion treatment. In addition, a multicenter prospective study with a large sample size is required to further confirm and explore the association between UA and the subtypes of HT. Second, UA levels have been found to change over time in patients with stroke [15], whereas our study enrolled patients admitted to the hospital within 24 h after the stroke onset only. Hence, there will be a limited scope of application in terms of the findings in our study. In addition, our study retrospectively explained the association between the single UA level and spontaneous HT at admission, so it is still necessary to further clarify such a relationship by dynamical examination of the UA level in a prospective study.
## Conclusion
In conclusion, among the non-reperfusion patients with acute ischemic stroke within 24 h after admission, the level of UA was independently and positively associated with the occurrence of spontaneous HT in male patients. More prospective research is needed to confirm these results.
## Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
## Ethics statement
The studies involving human participants were reviewed and approved by the Ethics Board of the First Affiliated Hospital of Chongqing Medical University. The patients/participants provided their written informed consent to participate in this study.
## Author contributions
YT and M-SL: study concept and design. YT, M-SL, and CF: acquisition of data. YT: statistical analysis and drafting of the manuscript. G-QL: critical revision of the manuscript for important intellectual content and study supervision. Analysis and interpretation of data were done by all authors. All authors contributed to the article and approved the submitted version.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## References
1. Álvarez-Sabín J, Maisterra O, Santamarina E, Kase CS. **Factors influencing haemorrhagic transformation in ischaemic stroke**. *Lancet Neurol.* (2013.0) **12** 689-705. DOI: 10.1016/S1474-4422(13)70055-3
2. Zhang J, Yang Y, Sun H, Xing Y. **Hemorrhagic transformation after cerebral infarction: current concepts and challenges**. *Ann Transl Med.* (2014.0) **2** 81. DOI: 10.3978/j.issn.2305-5839.2014.08.08
3. Squadrito GL, Cueto R, Splenser AE, Valavanidis A, Zhang H, Uppu RM. **Reaction of uric acid with peroxynitrite and implications for the mechanism of neuroprotection by uric acid**. *Arch Biochem Biophys.* (2000.0) **376** 333-7. DOI: 10.1006/abbi.2000.1721
4. Ames BN, Cathcart R, Schwiers E, Hochstein P. **Uric acid provides an antioxidant defense in humans against oxidant- and radical-caused aging and cancer: a hypothesis**. *Proc Natl Acad Sci U S A.* (1981.0) **78** 6858-62. DOI: 10.1073/pnas.78.11.6858
5. Vassalle C, Mazzone A, Sabatino L, Carpeggiani C. **Uric acid for cardiovascular risk: Dr. Jekyll or Mr. Hide?**. *Diseases.* (2016.0) **4** 12. DOI: 10.3390/diseases4010012
6. Chen Z, Chen H, Zhang Y, He Y, Su Y. **Lower uric acid level may be associated with hemorrhagic transformation, but not functional outcomes in patients with anterior circulation acute ischemic stroke undergoing endovascular thrombectomy**. *Metab Brain Dis.* (2020.0) **35** 1157-64. DOI: 10.1007/s11011-020-00601-7
7. Jickling GC, Liu D, Stamova B, Ander BP, Zhan X, Lu A. **Hemorrhagic transformation after ischemic stroke in animals and humans**. *J Cereb Blood Flow Metab.* (2014.0) **34** 185-99. DOI: 10.1038/jcbfm.2013.203
8. Hacke W, Donnan G, Fieschi C, Kaste M, von Kummer R, Broderick JP. **Association of outcome with early stroke treatment: pooled analysis of ATLANTIS, ECASS, and NINDS rt-PA stroke trials**. *Lancet.* (2004.0) **363** 768-74. DOI: 10.1016/S0140-6736(04)15692-4
9. Goyal M, Menon BK, van Zwam WH, Dippel DW, Mitchell PJ, Demchuk AM. **Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trials**. *Lancet.* (2016.0) **387** 1723-31. DOI: 10.1016/S0140-6736(16)00163-X
10. Hao Y, Yang D, Wang H, Zi W, Zhang M, Geng Y. **Predictors for symptomatic intracranial hemorrhage after endovascular treatment of acute ischemic stroke**. *Stroke.* (2017.0) **48** 1203-9. DOI: 10.1161/STROKEAHA.116.016368
11. Jovin TG, Li C, Wu L, Wu C, Chen J, Jiang C. **BAOCHE investigators. Trial of thrombectomy 6 to 24 hours after stroke due to basilar-artery occlusion**. *N Engl J Med. (* **387** 1373-84. DOI: 10.1056/NEJMoa2207576
12. Tao C, Nogueira RG, Zhu Y, Sun J, Han H, Yuan G. **ATTENTION investigators. Trial of endovascular treatment of acute basilar-artery occlusion**. *N Engl J Med.* (2022.0) **387** 1361-72. DOI: 10.1056/NEJMoa2206317
13. Chen G, Wang A, Zhao X, Wang C, Liu L, Zheng H. **Frequency and risk factors of spontaneous hemorrhagic transformation following ischemic stroke on the initial brain CT or MRI: data from the China National Stroke Registry (CNSR)**. *Neurol Res.* (2016.0) **38** 538-44. DOI: 10.1080/01616412.2016.1187864
14. Song Q, Wang Y, Cheng Y, Liu J, Wei C, Liu M. **Serum uric acid and risk of hemorrhagic transformation in patients with acute ischemic stroke**. *J Mol Neurosci.* (2020.0) **70** 94-101. DOI: 10.1007/s12031-019-01404-x
15. Brouns R, Wauters A, Van De Vijver G, De Surgeloose D, Sheorajpanday R, De Deyn PP. **Decrease in uric acid in acute ischemic stroke correlates with stroke severity, evolution and outcome**. *Clin Chem Lab Med.* (2010.0) **48** 383-90. DOI: 10.1515/CCLM.2010.065
16. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF., Feldman HI. **A new equation to estimate glomerular filtration rate**. *Ann Intern Med.* (2009.0) **150** 604-12. DOI: 10.7326/0003-4819-150-9-200905050-00006
17. Torbey MT, Bösel J, Rhoney DH, Rincon F, Staykov D, Amar AP. **Evidence-based guidelines for the management of large hemispheric infarction: a statement for health care professionals from the Neurocritical Care Society and the German Society for Neuro-intensive Care and Emergency Medicine**. *Neurocrit Care.* (2015.0) **22** 146-64. DOI: 10.1007/s12028-014-0085-6
18. Kissela BM, Flaherty ML., Alwell K, Moomaw CJ. **Eligibility for intravenous recombinant tissue-type plasminogen activator within a population: the effect of the European Cooperative Acute Stroke Study (ECASS) III**. *Trial Stroke.* (2012.0) **43** 1591-5. DOI: 10.1161/STROKEAHA.111.645986
19. Jeong SM, Yoo TG, Nam YS, Kim SH, Lee JE, Kim S. **Sex-dependent effects of uric acid on cerebral microbleed: a cross-sectional study in the general population**. *Eur J Neurol.* (2017.0) **24** 1300-6. DOI: 10.1111/ene.13378
20. Chen LH, Zhong C, Xu T, Xu T, Peng Y, Wang A. **Sex-specific association between uric acid and outcomes after acute ischemic stroke: a prospective study from CATIS**. *Trial Sci Rep.* (2016.0) **6** 38351. DOI: 10.1038/srep38351
21. Quiñones Galvan A, Natali A, Baldi S, Frascerra S, Sanna G, Ciociaro D. **Effect of insulin on uric acid excretion in humans**. *Am J Physiol.* (1995.0) **268** E1-5. DOI: 10.1152/ajpendo.1995.268.1.E1
22. Miller AA, De Silva TM, Jackman KA, Sobey CG. **Effect of gender and sex hormones on vascular oxidative stress**. *Clin Exp Pharmacol Physiol.* (2007.0) **34** 1037-43. DOI: 10.1111/j.1440-1681.2007.04732.x
23. Hayden MR, Tyagi SC. **Uric acid: a new look at an old risk marker for cardiovascular disease, metabolic syndrome, and type 2 diabetes mellitus: the urate redox shuttle**. *Nutr Metab (Lond).* (2004.0) **1** 10. DOI: 10.1186/1743-7075-1-10
24. Fang J, Alderman MH. **Serum uric acid and cardiovascular mortality the NHANES I epidemiologic follow-up study, 1971-1992. National health and nutrition examination Survey**. *JAMA.* (2000.0) **283** 2404-10. DOI: 10.1001/jama.283.18.2404
25. Gerber Y, Tanne D, Medalie JH, Goldbourt U. **Serum uric acid and long-term mortality from stroke, coronary heart disease and all causes**. *Eur J Cardiovasc Prev Rehabil.* (2006.0) **13** 193-8. DOI: 10.1097/01.hjr.0000192745.26973.00
26. Storhaug HM, Norvik JV, Toft I, Eriksen BO, Løchen ML, Zykova S. **Uric acid is a risk factor for ischemic stroke and all-cause mortality in the general population: a gender specific analysis from The Tromsø Study**. *BMC Cardiovasc Disord.* (2013.0) **13** 115. DOI: 10.1186/1471-2261-13-115
27. Zweier JL, Kuppusamy P, Lutty GA. **Measurement of endothelial cell free radical generation: evidence for a central mechanism of free radical injury in postischemic tissues**. *Proc Natl Acad Sci U S A.* (1988.0) **85** 4046-50. DOI: 10.1073/pnas.85.11.4046
|
---
title: High-risk population and factors of stroke has changed among middle-aged and
elderly Chinese—Evidence from 1989 to 2015
authors:
- Xue Zhang
- Jing Dai
- Wei Li
- Yunjuan Yang
journal: Frontiers in Public Health
year: 2023
pmcid: PMC10022731
doi: 10.3389/fpubh.2023.1090298
license: CC BY 4.0
---
# High-risk population and factors of stroke has changed among middle-aged and elderly Chinese—Evidence from 1989 to 2015
## Abstract
### Background
Stroke is an acute cerebrovascular disease with high mortality and disability. This study aimed to investigate the trend of stroke prevalence from 1989 to 2015 in China, explore the transition of high-risk population and high-risk factors, and provide some evidence to develop more targeted stroke intervention strategies.
### Material and methods
We derived the baseline data from China Health and Nutrition Survey (CHNS). Participants responded to face-to-face interviews and examinations containing demographic information, behavioral health information, disease history, and physical examination. We applied chi-square test, shapley value decomposition model, and decision tree model to evaluate the changes of high-risk population and high-risk factors of stroke.
### Results
Across 42,419 middle-aged and elderly residents, the prevalence of stroke was decreasing from 1989 to 2015. Hypertension was the leading risk factor of stroke, while its contribution rate was weakened with the increasing of medicine taking rate. As the second risk factor of stroke, the contribution of age decreased either. Meanwhile, the contribution rate of historical health factors, lifestyle factors, and regional factors, such as body mass index, diabetes, and living area to the impact of stroke was increasing. In addition, the first high-risk population of stroke changed from hypertension patients aged 75 years and above to without spouse residents living in stroke belt such as Beijing and Liaoning. The second risk population of stroke transformed from male hypertensive patients under 75 years old into male hypertensive patients living in urban. The third high-risk group turned from the elderly aged 75 and above into the female patients with hypertension and diabetes.
### Conclusions
This study demonstrated that the high-risk population and high-risk factors of stroke changed in China and revealed the direction and internal mechanism of transition of stroke. Targeted stroke intervention strategies should be renewed. Health education for the high-risk population of stroke should be carried out, healthy living habits need be advocated, and the use of antihypertensive drugs for the hypertensive patients should be standardized.
## Introduction
Stroke is the second leading cause of death and disability in the world and the global burden of stroke is of continual major importance for global health [1]. The prevalence of stroke in European countries and United States ranged from $1.5\%$ in Italy to $3\%$ in the UK and United States (2–4). In Asian countries, the prevalence of stroke has been reported in the range of 45–$\frac{471}{100}$,000 [5, 6]. According to global burden of disease study (GBD) in 2019, stroke is the first cause of life expectancy loss in China [7]. The incidence of stroke in China decreased from $\frac{222}{100}$,000 in 2005 to $\frac{201}{100}$,000 in 2019, and the incidence of ischemic stroke increased from $\frac{117}{100}$,000 in 2005 to $\frac{145}{100}$,000 in 2019 [8, 9]. China is the largest developing country with about one-fifth of the world's population and the highest number of stroke cases in the world. A recent survey conducted by the Stroke Screening and Prevention Committee demonstrated that, the number of new stroke patients in China increased by $8.79\%$ every year and the rate of increase was accelerating. According to China's National Stroke *Screening data* in 2019, middle-aged and elderly people were more likely to suffer from stroke. In 40 years old and above, the prevalence of first stroke increased from $1.89\%$ in 2012 to $2.58\%$ in 2019 [10].
Overtime, public policies have been established to promote the prevention and treatment of stroke. As of December 2019, the Brain Prevention Committee of the National Health Commission has licensed 30 demonstration senior stroke centers, 466 senior stroke centers, 181 comprehensive stroke prevention centers, and 717 stroke prevention centers [11]. The prevention and treatment of cardiovascular and cerebrovascular diseases was integrated into the Healthy China Action in 2019 [12]. As an emergency resource [13, 14], the construction of green channels for stroke in hospitals has been improved to shorten the time for stroke patients to be treated. Even so, the prognosis of stroke is still not optimistic. Most survivors have sequelae such as motor disorder, cognitive disorder, and speech swallowing disorder to varying degrees [15], which seriously hurt the self-care ability and mental health of patients, and burden families. Therefore, the prevention of stroke is far more essential than the treatment.
While numerous studies on the etiologies, risk factors, diagnosis, and management of stroke have been conducted in recent decades, it is still one of the main contributors to the burden of disease worldwide [16]. In particularly, *China is* a country with rapid economic growth in recent years, the lifestyle and living standard of residents have modified. The following question is whether the risk factors of stroke, such as hypertension, diabetes, sleep duration, and alcohol, etc. have changed with the lifestyle changes and social progress? Does the high-risk population of stroke change? Is it necessary to determine new screening objects? Thus, Studying the high-risk factors of stroke in China and identifying the changes of high-risk groups has important practical significance for the prevention of stroke and reducing the burden of disease in China. The identification of high-risk population and factors of stroke can also contribute to developing a prevention and treatment roadmap for other non-communicable diseases in the world [17]. Considering the importance of this issue this qualitative study described the high-risk factors influencing stroke prevalence and screened the high-risk population of stroke in the middle-aged and elderly in China between the years 1989 and 2015, and aimed to provide national strategies for the prevention and screening of these conditions.
## Materials and methods
The China Health and Nutrition Survey (CHNS) study is an ongoing open cohort international collaborative project between the Carolina Population Center at the University of North Carolina at Chapel Hill and the National Institute for Nutrition and Health at the Chinese Center for Disease Control and Prevention. CHNS offers freely accessible detailed nutrition and health datasets on a sample drawn from China that is representative of the national population. The survey covers nine (twelve) provinces that vary substantially in geography, economic development, public resources, and health indicators. A multistage, random cluster process was used to draw the samples. A detailed sampling process is presented in Figure 1.
**Figure 1:** *Multistage random cluster sampling process.*
## Data collection
Participants were contacted and visited at their own homes initially to recruit them into the study. The face-to-face interviews were conducted between trained field staffs and each participant (or his/ her spouse). A series of physical and cognitive measurements was recorded, and a standard questionnaire such as basic demography, health status, nutrition and diet status were answered in each interview. The inclusion criteria in the study were as follows: [1] Age 45 years and above. [ 2] Should have clear diagnosis of stroke. [ 3] Should have completed (missing value < $5\%$) major variables data. Then, we excluded participants as follow: age below 45 years ($$n = 92$$,790), with unavailable or unclear stroke diagnosis ($$n = 288$$), with missing value of marital status and educational attainment ($$n = 2$$,380), occupation and individual income ($$n = 3$$,077), smoking and alcohol behavior ($$n = 2$$,471), activity participation and sleep duration ($$n = 2$$,542), hypertension, diabetes and atrial fibrillation ($$n = 621$$), as well as height and weight ($$n = 1$$,368). The detailed sample selection process is presented in Figure 2. This study was approved by the Ethics Review Committee of The First People's Hospital of Yunnan Province. Written informed consent was obtained from all participants.
**Figure 2:** *Flow chart.*
## Stroke definition
Participants were asked “Did your doctor give you a diagnosis of stroke or transient ischemic attack?” in the face-to-face interviews. Response options included “Yes,” “No,” and “Not clear.” If participants responded “Yes,” they would be defined as having incident stroke outcomes. If participants answered “No,” they would be classified as having no incident stroke outcomes. Participants with “Not clear” answer were excluded from this study.
## Potential risk factors
This study covered four dimensions of risk factors including sociodemographic factors, lifestyle factors, historical health variables, and regional information. The sociodemographic factors taken into account were gender, age, residents, marital status, educational attainment, logarithm of individual income, and occupation types. Lifestyle factors included smoking, Alcohol, activity participation, and sleep duration. Smoking and drinking were defined as a self-reported history in last year. Of the six self-reported activities (martial art, track, gymnastics, soccer or basketball, badminton or volleyball, and ping pong), participating in one or more was defined as participation activity. Sleep duration was divided three levels according self-reported sleep time. Historical health variables included hypertension, diabetes, atrial fibrillation, and body mass index (BMI). Hypertension was measured using mean values of three systolic and diastolic blood pressures. According to the latest standards of WHO, systolic blood pressure ≥140 mmHg and diastolic blood pressure ≥90 mmHg were defined as hypertension. Diabetes and atrial fibrillation were defined as a self-reported physician diagnosis. BMI was calculated by dividing weight by height squared. Height and weight were measured with participants wearing light clothing and without shoes during the physical examination. Regional information referred to the province where the respondent was located.
## Statistical analysis
Chi-square tests were used to examine the general characteristics differences between stroke and non-stroke subjects according to the distribution of categorized variables. Count and percentages were used to describe categorical variables. The risk factors, which were significant in the chi-square test, were further analyzed by decision tree model, which was established by Chi-squared Automatic Interaction Detection (CHAID). Shapley value decomposition model was adopted to calculate the contribution rate and ranking of high-risk factors. $P \leq 0.05$ was considered statistically significant. The decision tree analysis method is an analysis tool based on the principle of probability theory [18]. Its basic principle is to use decision points to represent decision problems [19]. Compared with the traditional logic regression method, the decision tree analysis method can directly reflect the characteristics of different subgroups and the proportion of different results, and can also reflect the interaction between variables [20]. It is mostly used for prediction analysis and factor analysis. However, the disadvantage of this method is that cannot reflect the quantitative impact of variables. The combination of decision tree analysis and Shapley Value Decomposition can just make up for their shortcomings.
## Characteristics of all participants
Table 1 shows the distribution of the main characteristics of the study population. A total of 42,419 middle-aged and elderly residents aged 45 and above were included in the survey from 1989 to 2015. The distribution of male and female was uniform, accounting for nearly $50\%$, respectively. Among them, the middle-aged residents aged 45–49, 50–54, and 55–59 were the most. Most residents lived in rural areas, while only $37.5\%$ residents lived in cities and towns. Moreover, the $82.1\%$ of residents had spouses. The proportion of illiterate residents was $41.56\%$, and the proportion of residents with high school education or above was < $20\%$. More than half of the residents had an annual income of < 5,000 yuan. In terms of the type of occupation, the number of residents without jobs was the largest, accounting for $42.92\%$, followed by farmers, accounting for $31.57\%$. In terms of lifestyle, most residents maintained good living habits, no smoke, no drink, and kept normal sleep duration. However, the exercise time and amount of most residents were below the standard ($89.53\%$). In terms of historical health information, $18.52\%$ of residents suffered hypertension, $4.34\%$ suffered diabetes, $1.29\%$ suffered heart disease, and more than $30\%$ kept a BMI of 25 kg/m2 or above. The residents participating in the study were spread across 12 provinces in China, among which Guangxi Province and Jiangsu Province had the largest number, accounting for 11.64 and $11.09\%$, respectively.
**Table 1**
| Item | Total | Total.1 | Normal | Normal.1 | Stroke | Stroke.1 | χ2 | P-value |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | No. | % | No. | % | No. | % | | |
| Total | 42419 | 100.00 | 41384 | 97.56 | 1035 | 2.44 | | |
| Gender | | | | | | | 43.478 | < 0.001 |
| Female | 21425 | 50.51 | 21007 | 98.05 | 418 | 1.95 | | |
| Male | 20994 | 49.49 | 20377 | 97.06 | 617 | 2.94 | | |
| Age groups (years) | | | | | | | 368.663 | < 0.001 |
| 45–49 | 8624 | 20.33 | 8553 | 99.18 | 71 | 0.82 | | |
| 50–54 | 8283 | 19.53 | 8168 | 98.61 | 115 | 1.39 | | |
| 55–59 | 7516 | 17.72 | 7378 | 98.16 | 138 | 1.84 | | |
| 60–64 | 6443 | 15.19 | 6244 | 96.91 | 199 | 3.09 | | |
| 65–69 | 4742 | 11.18 | 4568 | 96.33 | 174 | 3.67 | | |
| 70–74 | 3339 | 7.87 | 3177 | 95.15 | 162 | 4.85 | | |
| 75–79 | 2068 | 4.88 | 1962 | 94.87 | 106 | 5.13 | | |
| 80 above | 1404 | 3.31 | 1334 | 95.01 | 70 | 4.99 | | |
| Residents | 42419 | | | | | | 72.379 | < 0.001 |
| Rural | 26512 | 62.50 | 25996 | 98.05 | 516 | 1.95 | | |
| Urban | 15907 | 37.50 | 15388 | 96.74 | 519 | 3.26 | | |
| Marital status | | | | | | | | |
| No spouse | 7592 | 17.90 | 7342 | 96.71 | 250 | 3.29 | 28.265 | 0.001 |
| Having spouse | 34827 | 82.10 | 34042 | 97.75 | 785 | 2.25 | | |
| Educational attainment | | | | | | | | |
| Illiterate | 17631 | 41.56 | 17100 | 96.99 | 531 | 3.01 | 50.858 | < 0.001 |
| Elementary school | 8905 | 20.99 | 8719 | 97.91 | 186 | 2.09 | | |
| Middle school | 8754 | 20.64 | 8587 | 98.09 | 167 | 1.91 | | |
| High school | 5719 | 13.48 | 5613 | 98.15 | 106 | 1.85 | | |
| College or more | 1410 | 3.32 | 1365 | 96.81 | 45 | 3.19 | | |
| Individual income (RMB) | | | | | | | 13.235 | 0.004 |
| Q1 (< 1,244) | 10604 | 25.00 | 10298 | 97.11 | 306 | 2.89 | | |
| Q2 (1,244–4,456.89) | 10602 | 24.99 | 10351 | 97.63 | 251 | 2.37 | | |
| Q3 (4,456.89–13,800) | 10584 | 24.95 | 10357 | 97.86 | 227 | 2.14 | | |
| Q4 (>13,800) | 10629 | 25.06 | 10378 | 97.64 | 251 | 2.36 | | |
| Occupation types | | | | | | | 128.702 | < 0.001 |
| No jobs | 18205 | 42.92 | 17584 | 96.59 | 621 | 3.41 | | |
| Professional and administrator | 4006 | 9.44 | 3949 | 98.58 | 57 | 1.42 | | |
| Worker | 5636 | 13.29 | 5536 | 98.23 | 100 | 1.77 | | |
| Farmer | 13390 | 31.57 | 13158 | 98.27 | 232 | 1.73 | | |
| Rests | 1182 | 2.79 | 1157 | 97.88 | 25 | 2.12 | | |
| Smoking | | | | | | | | |
| No smoking | 28313 | 66.75 | 27637 | 97.61 | 676 | 2.39 | 0.98 | 0.322 |
| Smoking | 14106 | 33.25 | 13747 | 97.45 | 359 | 2.55 | | |
| Alcohol | | | | | | | | |
| No alcohol | 28961 | 68.27 | 28146 | 97.19 | 815 | 2.81 | 53.6926 | < 0.001 |
| Alcohol | 13458 | 31.73 | 13238 | 98.37 | 220 | 1.63 | | |
| Activity participation | | | | | | | | |
| No activity | 37977 | 89.53 | 37043 | 97.54 | 934 | 2.46 | 0.576 | 0.448 |
| Activity | 4442 | 10.47 | 4341 | 97.73 | 101 | 2.27 | | |
| Sleep duration (h/d) | | | | | | | | |
| < 6 | 1935 | 4.56 | 1884 | 97.36 | 51 | 2.64 | 129.603 | < 0.001 |
| 6−10 | 39346 | 92.76 | 38448 | 97.72 | 898 | 2.28 | | |
| >10 | 1138 | 2.68 | 1052 | 92.44 | 86 | 7.56 | | |
| Hypertension | | | | | | | | |
| No hypertension | 34565 | 81.48 | 34216 | 98.99 | 349 | 1.01 | 16.013 | < 0.001 |
| Hypertension | 7854 | 18.52 | 7168 | 91.27 | 686 | 8.73 | | |
| Diabetes | | | | | | | | |
| No diabetes | 40576 | 95.66 | 39684 | 97.80 | 892 | 2.20 | 229.007 | < 0.001 |
| Diabetes | 1843 | 4.34 | 1700 | 92.24 | 143 | 7.76 | | |
| Atrial fibrillation | | | | | | | | |
| No atrial fibrillation | 41873 | 98.71 | 40900 | 97.68 | 973 | 2.32 | 184.691 | < 0.001 |
| Atrial fibrillation | 546 | 1.29 | 484 | 88.64 | 62 | 11.36 | | |
| BMI (kg/m 2 ) | | | | | | | | |
| < 25 | 28062 | 66.15 | 27502 | 98.00 | 560 | 2.00 | 102.381 | < 0.001 |
| 25–30 | 10313 | 24.31 | 10020 | 97.16 | 293 | 2.84 | | |
| 30 above | 4044 | 9.53 | 3862 | 95.50 | 182 | 4.50 | | |
| Provinces | | | | | | | 95.258 | < 0.001 |
| Beijing | 1070 | 2.52 | 1029 | 96.17 | 41 | 3.83 | | |
| Liaoning | 3603 | 8.49 | 3478 | 96.53 | 125 | 3.47 | | |
| Heilongjiang | 3521 | 8.30 | 3447 | 97.90 | 74 | 2.10 | | |
| Shanghai | 1616 | 3.81 | 1562 | 96.66 | 54 | 3.34 | | |
| Jiangsu | 4705 | 11.09 | 4595 | 97.66 | 110 | 2.34 | | |
| Shandong | 4384 | 10.33 | 4269 | 97.38 | 115 | 2.62 | | |
| Henan | 4460 | 10.51 | 4321 | 96.88 | 139 | 3.12 | | |
| Hubei | 4219 | 9.95 | 4110 | 97.42 | 109 | 2.58 | | |
| Hunan | 4128 | 9.73 | 4052 | 98.16 | 76 | 1.84 | | |
| Guangxi | 4939 | 11.64 | 4809 | 97.37 | 130 | 2.63 | | |
| Guizhou | 4631 | 10.92 | 4576 | 98.81 | 55 | 1.19 | | |
| Chongqing | 1143 | 2.69 | 1136 | 99.39 | 7 | 0.61 | | |
## Prevalence of stroke in different populations
As shown in Table 1, gender, age, residents, marital status, personal income, occupation, drinking, hypertension, sleep duration, diabetes, heart disease, BMI, and provinces were the main influencing factors for stroke among middle-aged and elderly residents in China ($P \leq 0.05$). Among them, middle-aged and elderly men were more likely to suffer from stroke compared with female middle-aged and elderly residents. Compared with rural residents, urban residents were more likely to suffer from stroke. The middle-aged and elderly residents with spouse were less likely to suffer from stroke than those without spouse. What's more, the prevalence of stroke increased with age. The residents with the lowest and highest education levels were more likely to suffer from stroke. The lower the individual's annual income, the higher the possibility of stroke. Moreover, the unemployed middle-aged and elderly residents were more likely to suffer from stroke. The middle-aged and elderly residents who did not drink alcohol were more likely to suffer from stroke than those who drank alcohol. Both excessive sleep and insufficient sleep could promote the onset of stroke. Besides, the prevalence of stroke in middle-aged and elderly residents with hypertension, diabetes, and heart disease was higher than that in normal residents. In addition, the larger the BMI, the more likely to suffer from stroke. What's more, the prevalence of stroke in Liaoning, Beijing, Shanghai, and Henan was the highest, all exceeding $3\%$.
## High-risk factors and high-risk population of stroke
According to the results of the decision tree in Figure 3, hypertension was the primary factor affecting the prevalence of stroke, followed by sleep time and age. BMI, province, heart disease, and gender were the third level risk factors affecting the prevalence of stroke in middle-aged and elderly residents. From the left branch of the decision tree, the prevalence of stroke in middle-aged and elderly residents with hypertension alone was $8.7\%$. Moreover, the prevalence of stroke in middle-aged and elderly residents with hypertension and more than 10 h of sleep per day reached to $22.4\%$, becoming the second high-risk group. The middle-aged and elderly residents with hypertension, sleeping more than 10 h per day, and the BMI of more than 30 kg/m2 were the first risk group for stroke, and the risk of stroke reached $40.6\%$. Otherwise, the elderly residents with hypertension, sleeping within 6–10 h a day, and over 70 years old were the third high-risk group, and the prevalence rate of stroke was $11.3\%$. From the right branch of the decision tree, residents who did not suffer from hypertension, under the age of 70–84, and lived in Beijing and Liaoning, becoming the fourth high-risk group for stroke, with a prevalence rate of $6.7\%$. The fifth high-risk group was middle-aged residents aged 45–54 who did not suffer hypertension but suffered heart disease, and the prevalence rate of stroke was $5.7\%$. Additionally, the prevalence of stroke among middle-aged residents aged between 45 and 60 who did not suffer from hypertension or heart disease was below the average level. The prevalence of stroke was also below the average level for residents who did not suffer from hypertension, 60–69 years old, and living in Shanghai, Jiangsu, Shandong, Hubei, Hunan, Guangxi, Guizhou, and Chongqing.
**Figure 3:** *Assessment of risk factors and the susceptible population of stroke, 1989–2015.*
## Changes trends of stroke prevalence in high-risk factors
As shown in Figure 4, the prevalence of stroke in the general population decreased from 1989 to 2015. The prevalence of stroke in people with hypertension was higher than that in people without hypertension. In recent years, the prevalence of stroke among middle-aged and elderly residents who did not suffer from hypertension had an upward trend. However, the prevalence of stroke among middle-aged and elderly residents suffering from hypertension had shown a rapid decline trend in recent years, especially by 2015, the prevalence of stroke among hypertensive people had dropped to the lowest in history. Figure 5 described the change of the prevalence of stroke among middle-aged and elderly residents of all ages after regrouping by classification decision tree. The prevalence of stroke among middle-aged residents aged 45–54 years showed a downward trend. The prevalence of stroke in residents aged 55–59 years was higher than that in residents aged 45–54 years at the same period, and the prevalence of stroke in residents aged 55–59 years also showed a downward trend. Moreover, the prevalence of stroke among residents aged 60–69 years showed a downward trend. The prevalence of stroke among residents aged 70 years and above fluctuated greatly in different periods, while it was in a downward trend. It could be seen from Figure 6 that the prevalence and change trend of stroke among middle-aged and elderly residents with different sleep duration after regrouping by classification decision tree. The prevalence of stroke among middle-aged and elderly residents who slept < 10 h a day was at a low level and in the process of slow decline. By 2015, the prevalence rate was the lowest. Otherwise, the prevalence of stroke among middle-aged and elderly residents who slept more than 10 h a day fluctuated greatly, while the prevalence of stroke was generally in a downward trend.
**Figure 4:** *Prevalence of stroke in hypertension and no-hypertension groups in 1989–2015.* **Figure 5:** *Prevalence of stroke in different age groups in 1989–2015.* **Figure 6:** *Prevalence of stroke in different sleep duration groups in 1989–2015.*
Table 2 showed the contribution of stroke influencing factors to stroke prevalence in middle-aged and elderly residents. Hypertension was the first risk factor affecting stroke among middle-aged and elderly residents in 2009, 2011, and 2015, but its contribution was gradually declining, with the contribution of $50.91\%$ in 2009, $44.51\%$ in 2011 and $36.39\%$ in 2015. Age was always the second major risk factor for stroke in middle-aged and elderly residents. And the contribution rate of age to the prevalence of stroke was also in a downward trend, from 19.78 to $11.01\%$. Hypertension, age, sex, residence, and occupation type contributed the most to the prevalence of stroke among middle-aged and elderly residents in 2009, with a total contribution of $80.43\%$. In addition, hypertension, age, occupational type, province, and diabetes were the main risk factors affecting stroke among middle-aged and elderly residents in 2011, with a total contribution of $88.01\%$. Compared with 2009, the impact of historical health factors on stroke in middle-aged and elderly people was increasingly prominent. Hypertension, age, province, diabetes, and sleep duration were the main risk factors affecting stroke in middle-aged and elderly people in 2015, with a total contribution of $76.92\%$. Besides, compared with 2009, the contribution of historical health factors and behavioral style factors to stroke had gradually increased.
**Table 2**
| Item | 2009 | 2009.1 | 2011 | 2011.1 | 2015 | 2015.1 |
| --- | --- | --- | --- | --- | --- | --- |
| | Contribution rate (%) | Rank | Contribution rate (%) | Rank | Contribution rate (%) | Rank |
| Gender | 5.31 | 3 | 4.12 | 6 | 6.52 | 7 |
| Age | 19.78 | 2 | 23.26 | 2 | 11.01 | 2 |
| Residents | 4.43 | 4 | 0.48 | 11 | 2.51 | 9 |
| Marital status | 0.32 | 14 | 0.50 | 10 | 2.94 | 8 |
| Educational attainment | 1.19 | 12 | 0.37 | 13 | 0.84 | 11 |
| Individual income | 0.73 | 13 | 0.10 | 14 | 0.52 | 12 |
| Occupation types | 3.51 | 5 | 9.92 | 3 | 7.75 | 6 |
| Alcohol | 2.24 | 9 | 0.88 | 9 | 0.19 | 14 |
| Sleep duration | 2.89 | 6 | 3.38 | 7 | 9.55 | 5 |
| Hypertension | 50.91 | 1 | 44.51 | 1 | 36.39 | 1 |
| Diabetes | 2.77 | 7 | 4.95 | 5 | 9.68 | 4 |
| Atrial fibrillation | 1.85 | 10 | 1.72 | 8 | 1.35 | 10 |
| BMI | 2.36 | 8 | 0.44 | 12 | 0.47 | 13 |
| Provinces | 1.71 | 11 | 5.37 | 4 | 10.29 | 3 |
| Total | 100.00 | — | 100.00 | — | 100.00 | — |
## Changes trends of stroke prevalence in high-risk population
The high-risk population of stroke in 2009 could be identified from the classified statistical results in Figure 7. The prevalence rate of stroke among middle-aged and elderly people was $2.2\%$ in 2009. The prevalence rate of stroke among residents with hypertension was $8.0\%$, which was 8.9 times than that of residents without hypertension. Among hypertension patients, the prevalence rate of stroke in patients aged 75 years and above was as high as $18.0\%$, that was, the elderly hypertensive patients aged 75 years and above were at high risk of stroke. The prevalence of stroke in men aged 45 years and above and under 75 years who also suffered from hypertension was $9.8\%$, which was also a high-risk population. The second was the 55 to 74 years old elderly who did not suffer from hypertension but had diabetes. The prevalence rate of stroke was $5.5\%$, 2.5 times higher than that of ordinary middle-aged and elderly people. In conclusion, the elderly with hypertension was the first high-risk group for stroke in 2009. Middle aged men and young elderly people with hypertension were the second high-risk group for stroke. The elderly who did not suffer from hypertension but were only over 75 years old were the third high-risk group for stroke. The young elderly without hypertension but with diabetes were the fourth high-risk group for stroke. It could be concluded that age and historical health problems were the main factors to divide the high-risk population of stroke. The distribution of high-risk population of stroke in 2011 could be obtained from Figure 8. The overall prevalence of stroke among middle-aged and elderly residents was $2.5\%$ in 2011. The prevalence of stroke in patients with hypertension was $8.1\%$, which was 8.1 times higher than that in residents without hypertension. In particular, the prevalence rates of stroke among the elderly over 60 years old and over 75 years old were 9.9 and $16.9\%$. At the same time, the prevalence of stroke was $23.8\%$ in elderly who suffered from hypertension and lived in Beijing, Shanghai, Jiangsu, Liaoning, Heilongjiang, which either in high level of economic development or northeast areas. They were the first high-risk population. Meanwhile, the prevalence of stroke among hypertensive patients with diabetes aged between 45 and 60 years was $11.1\%$. The prevalence rate of stroke in male hypertensive patients aged 60–74 was $12.6\%$, which was five times higher than that in ordinary middle-aged and elderly residents. For residents who did not suffer from hypertension, excessive sleep time had become the first killer of stroke. For residents who slept more than 10 h a day, the prevalence of stroke was $7.2\%$. In a word, the elderly hypertensive patients who lived in Beijing, Shanghai, Jiangsu, and Northeast China in 2011 were the first risk group for stroke, and the male hypertensive patients aged 60–74 years were the second risk group for stroke. Middle aged residents with diabetes and hypertension were the third risk group for stroke. What's more, middle-aged and elderly residents who slept more than 10 h a day were also at high risk of stroke. In conclude, age and historical health problems were no longer the main risk groups of stroke, regional factors, and healthy life factors had also became the key factors to divide the high-risk groups. As shown in Figure 9, the prevalence of stroke among middle-aged and elderly residents in China was $1.8\%$ in 2015.. The prevalence of stroke in patients with hypertension was $5.0\%$, which was five times higher than that in patients without hypertension. In addition, the prevalence of stroke in male hypertensive patients living in cities and towns was higher, which was $9.1\%$. The prevalence of stroke among middle-aged and elderly female residents with diabetes and hypertension was $7.5\%$, 4.2 times higher than that of ordinary middle-aged and elderly residents. Meanwhile, among the people who did not suffer from hypertension, the prevalence of stroke among the elderly who lived in Beijing, and Liaoning and without spouse was $9.5\%$. In a word, unmatched middle-aged and elderly residents living in Beijing and Liaoning were the first high-risk group for stroke in 2015. Male hypertensive patients living in cities and towns were the second risk group for stroke. The middle-aged and elderly female residents with diabetes and hypertension were the third high-risk group for stroke. It could be seen that the high-risk group of stroke had not only been limited to patients with historical health problems such as hypertension and diabetes, but also started to extend to middle-aged and elderly people without spouse.
**Figure 7:** *Assessment of risk factors and the susceptible population of stroke in 2009.* **Figure 8:** *Assessment of risk factors and the susceptible population of stroke in 2011.* **Figure 9:** *Assessment of risk factors and the susceptible population of stroke in 2015.*
## Discussion
This study was the first to evaluated the transition of high-risk factors and high-risk population of stroke in middle-aged and older Chinese from 1989 to 2015. It has provided some useful information for stroke prevention and intervention in middle-aged and older Chinese. Hypertension was the first risk factor of stroke in China, but in recent years, the influence of hypertension was gradually decreasing. Hypertension could lead to remodeling of cardiovascular and cerebral vessels, damage the function of pressure reflex, induce white matter lesions, and then lead to stroke [21]. In addition, pulmonary hypertension will cause the hemoglobin in the blood to be too viscous, which will lead to blockage of blood vessels, thus causing stroke [22]. However, hypertension has been shown to be the single most important modifiable risk factor in adult stroke [23]. Hypertension patients taking drugs to control hypertension could effectively reduce the prevalence of stroke among residents in the Stroke Belt [24]. The drug taking rate and control rate of hypertension were gradually rising in China. It could be seen from Figure 10 of this study that the treatment rate of hypertension in China had increased from 20.18 to $84.26\%$ from 1991 to 2015. Therefore, with the control rate of hypertension increased, the impact of hypertension on stroke also decreased [25, 26]. Age was the second risk factor of stroke in China, but its influence was gradually weakening in recent years. With the increase of age, patients' arterial elasticity decreased, and intima damage was more likely to occur, which promoted lipid deposition to form plaque, leading to stroke [27]. However, with the improvement of medical level and the improvement of life expectancy in China, the impact of age on stroke was on the decline. Reasonable diet, scientific exercise and regular monitoring could largely compensate for the disease risk brought by aging to the elderly (28–30). Short or long sleep time would increase the risk of stroke in middle-aged and elderly people, and the contribution rate of sleep time to stroke in elderly people was increasing year by year. The existing clinical observation results showed that insufficient and excessive sleep duration could directly or indirectly cause stroke by influencing endocrine and metabolic functions, sympathetic nerve excitability, cortisol level and other mechanisms, such as hypertension, diabetes, and dyslipidemia [31, 32]. In addition, people at high risk of stroke were generally accompanied by hypertension, diabetes, hyperlipidemia, obesity and other diseases, and some cerebrovascular functions were damaged. The cerebral blood flow and cerebral metabolism of such people would change significantly with the change of sleeping habits [33]. Hence, people with insufficient and excessive sleep duration suffered a higher stroke risk.
**Figure 10:** *Medicine taking rate of hypertension in 1991–2015.*
With the progress and development of Chinese society, the high-risk factors of stroke among middle-aged and elderly residents in China had also changed accordingly. Changed from unmodified factors such as age, gender, and residence to behavior factors such as occupation type and sleep duration. Previous studies had shown that there were significant occupational differences in stroke. The service industry, agriculture, forestry, animal husbandry and fishery had higher stroke risk than other occupations [34]. Moreover, these occupations had a higher detection rate of cardiovascular risk factors than other occupations, including poor blood glucose control, and obesity [35]. Targeted workplace interventions could help reduce cardiovascular risk factors in these occupational groups, thereby benefiting health. In addition, compared with residents without jobs, engaging occupations can promote and help sustain healthy lifestyle habits for person for cardiovascular diseases, including stroke [36]. In addition, even those had high generic risk but adhering to healthy lifestyle had a lower risk of stroke than those at low-to-intermediate genetic risk of stroke [37]. This further confirmed the importance of a healthy lifestyle in reducing stroke risk.
The high-risk population of stroke had changed from the elderly residents with historical health problems to the residents living in certain areas and adhering to certain lifestyles. Stroke Belts had been formed in Beijing, Shanghai, Liaoning, Heilongjiang, and Henan in China. Middle-aged and elderly people living in these regions had a higher risk of stroke. On the one hand, studies showed that the regional differences were due to the different levels of physical activity of residents in each region. Adult Stroke Belt residents had markedly lower physical activity levels and were less likely to meet physical activity guidelines than their non-Stroke Belt counterparts [38]. The work pressure in Beijing and Shanghai was high, and the time was tight, especially the middle-aged residents in the career development period were seriously lacking in physical activity level. On the other hand, air pollution was a new modifiable neuro-vascular risk factor [39]. One study reported the risk for ischemic stroke was increased after exposure to air pollution and exposure to air pollution increased the risk of intracerebral hemorrhage [40]. Notably Beijing was one of the most polluted capitals in the world and the average air quality index (AQI) in 2016 was slight pollution [41]. In fact, Shanghai was a mega city in China, while its AQI was classified as “good” on only 58 and 78 days in 2017 and 2016 [42]. What's more, it was confirmed that solid fuel use for cooking and heating associated with increased risk of stroke, while persistent clean fuel use for both heating and cooking associated with lower risk of stroke occurrence [43]. However, Liaoning, Heilongjiang, and Henan were major agricultural production provinces in China, straw burning and solid fuel using was common before the revision of the law on prevention and control of air pollution in 2016. Hence, the stroke risk in China may be closely connected with air pollution in living area. Notably, the impact of living area on stroke was obvious in the classification tree statistics in 2011 and 2015. The possible reason was that Chongqing, Beijing, and Shanghai were included in 2011 and 2015. If these three cities were included in the first wave of survey, more obvious results may appear. Because it can be seen in the shapley value analysis that the contribution rate of living area was growing, and the ranking is getting top. Our study has highlighted the changing role of hypertension and the stronger role of behavior, lifestyle, and living area. In addition, health education is also necessary [44]. Combine the stroke screening with the health examination of residents, carry out health education for the high-risk population of stroke, advocate healthy living habits, and standardize the use of antihypertensive drugs for the hypertensive patients. On this basis, health files of stroke should be built at the community level, and combined the stroke health management with the Chinese family doctor contract system, regularly follow up the stroke high-risk population.
Notwithstanding, this study has certain limitations. First, the diagnosis of stroke relied on self-reporting and, despite face-to-face interviews, there were inevitable misreported. Second, this study lacked specific distinction between ischemic and hemorrhagic stroke, as more than $50\%$ of the cases cannot report their stroke type clearly. Third, the participants across the ten waves of survey lacked coherence for many participants exited or new entered. Accuracy of the trends may be undermined because generation gap and period effect.
## Conclusion
High-risk population and factors of stroke had changed among middle-aged and older Chinese from 1989 to 2015. Hypertension and age had less influence on stroke, while diabetes, sleep duration and marital status had greater influence on stroke. High risk population of stroke had changed from residents with historical health problems and old age to residents living in certain areas and adhering to certain lifestyles. Furthermore, if possible, we can subdivide the types of stroke in future study, and explore the changes of high-risk factors and high-risk population of hemorrhagic stroke and ischemic stroke, respectively.
## Data availability statement
Publicly available datasets were analyzed in this study. This data can be found here: https://www.cpc.unc.edu/projects/china China Health and Nutrition Survey.
## Ethics statement
The study was approved by the Ethics Review Committee of the First People's Hospital of Yunnan Province. The patients/participants provided their written informed consent to participate in this study.
## Author contributions
XZ statistical analysis and manuscript preparation. JD study concept and design. WL study supervision and critical revision of manuscript for intellectual content. YY statistical analysis and interpretation of data. All authors contributed to the article and approved the submitted version.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## References
1. Shahbandi A, Shobeiri P, Azadnajafabad S, Moghaddam SS, Tehrani YS, Ebrahimi N. **Burden of stroke in North Africa and Middle East, 1990 to 2019: a systematic analysis for the global burden of disease study 2019**. *BMC Neurol.* (2022.0) **22** 1-11. DOI: 10.1186/s12883-022-02793-0
2. Zhang Y, Chapman AM, Plested M, Jackson D, Purroy F. **The incidence, prevalence, and mortality of stroke in France, Germany, Italy, Spain, the UK, and the US: a literature review**. *Stroke Res Treat.* (2012.0) **2012** 436125. DOI: 10.1155/2012/436125
3. Bodenlos K, Schuster M, Ananth C V. **Trends in stroke-related maternal mortality in the United States. Autonomic neuroscience**. *Basic Clin.* (2019.0) **5** 222. DOI: 10.1016/j.ajog.2019.11.475
4. Bangee M, Martinez-Garduno CM, Brady MC, Cadilhac DA, Dale S, Hurley MA. **Oral care practices in stroke: findings from the UK and Australia**. *BMC Nurs.* (2021.0) **20** 1. DOI: 10.21203/rs.3.rs-126291/v1
5. Kulshreshtha A, Anderson LM, Goyal A, Keenan NL. **Stroke in South Asia: a systematic review of epidemiologic literature from 1980 to 2010**. *Neuroepidemiology.* (2012.0) **38** 123-9. DOI: 10.1159/000336230
6. Turana Y, Tengkawan J, Chia YC, Nathaniel M, Wang JG, Sukonthasarn A. **Hypertension and stroke in Asia: a comprehensive review from HOPE Asia**. *J Clin Hyperten.* (2021.0) **23** 513-21. DOI: 10.1111/jch.14099
7. Feigin V L, Stark B A, Johnson C O. **Global, regional, and national burden of stroke and its risk factors, 1990–2019: a systematic analysis for the global burden of disease study 2019**. *Lancet Neurol.* (2021.0) **20** 795-820. DOI: 10.1016/S1474-4422(21)00252-0
8. Liu Y, Wang W, Huang X, Zhang X, Lin L, Qin JJ. **Global disease burden of stroke attributable to high fasting plasma glucose in 204 countries and territories from 1990 to 2019: an analysis of the global burden of disease study**. *J Diabetes.* (2022.0) **14** 495-513. DOI: 10.1111/1753-0407.13299
9. Zhang K, Zhu Z, Peng H, Zhang M, Li H, Wang A. **Utility of China-PAR stroke equations for predicting 10-year stroke risk in the rural Inner Mongolian population in China**. *Neurol Res.* (2022.0) **3** 1-6. DOI: 10.1080/01616412.2022.2089833
10. Guan T, Ma J, Li M, Xue T, Lan Z, Guo J. **Rapid transitions in the epidemiology of stroke and its risk factors in China from 2002 to 2013**. *Neurology.* (2017.0) **89** 53-61. DOI: 10.1212/WNL.0000000000004056
11. Longde W, Bin P, Hongqi Z. **China stroke prevention report 2020**. *Chin J Cerebrovasc Dis.* (2022.0) **19** 136-44. DOI: 10.3969/j.issn.1672-5921.2022.02.011
12. Council T S. **Outline of Healthy China 2030[EB/OL]**
13. Molavi-Taleghani Y, Ebrahimpour H, Sheikhbardsiri H. **A proactive risk assessment through healthcare failure mode and effect analysis in pediatric surgery department**. *J Compr Ped.* (2020.0) **3** 8. DOI: 10.5812/compreped.56008
14. Sheikhbardsiri H, Abdar Z E, Sheikhasadi H. **Observance of patients' rights in emergency department of educational hospitals in south-east Iran**. *Int J Human Rights Healthcare.* (2020.0) **3** 72. DOI: 10.1108/IJHRH-09-2019-0072
15. Oh CW, Lee SH, Nam TM, Jang JH, Kim YZ, Kim KH. **Comparative analysis of stroke patients with and without sequelae: a cross-sectional analysis using the KOREA national health and nutrition examination survey (2016–2019)**. *J Clin Med.* (2021.0) **10** 4122. DOI: 10.3390/jcm10184122
16. Wafa HA, Wolfe CDA, Emmett E, Roth GA, Johnson CO, Wang Y. **Burden of stroke in Europe 30-year projections of incidence, prevalence, death, and disability-adjusted life years**. *Stroke.* (2020.0) **51** 2418-27. DOI: 10.1161/STROKEAHA.120.029606
17. Hadian M, Jabbari A, Abdollahi M, Hosseini E, Sheikhbardsiri H. **Explore pre-hospital emergency challenges in the face of the COVID-19 pandemic: a quality content analysis in the Iranian context**. *Frontiers in Public Health.* (2022.0) **10** 4109. DOI: 10.3389/fpubh.2022.864019
18. Wang YY, Zhang M, Wang XX, Liu S, Ding H. **Correlates of cognitive impairment in the elderly in China: a cross-sectional study**. *Frontiers in Public Health.* (2022.0) **10** 3661. DOI: 10.3389/fpubh.2022.973661
19. Banker RD, Cooper A. **Some models for estimating technical and scale inefficiencies in data envelopment analysis**. *Manage Sci.* (1984.0) **30** 1078-92. DOI: 10.2307/2631725
20. Choi H, Kim E, Kim J. **Prognostic factors in diabetes: comparison of Chi-square automatic interaction detector (CHAID) decision tree technology and logistic regression**. *Medicine.* (2022.0) **101** e31343. DOI: 10.1097/MD.0000000000031343
21. Chen L, He W, Liu F. **Stroke in pulmonary hypertension: is it predictable or preventable?**. *Int J Cardiol.* (2020.0) **303** 79. DOI: 10.1016/j.ijcard.2019.11.108
22. Hoeper, M. *J Am Coll Cardiol* (2020.0) **75** 27-8. DOI: 10.1016/j.jacc.2019.11.010
23. Ganten D. **Nomenclature of the renin-angiotensin system. Report of the nomenclature committee of the international society of hypertension (ISH) on the occasion of the 5th meeting of the ISH in Paris, June 12–14, 1978**. *Klinische Wochenschrift* (1978.0) **56** 187. DOI: 10.1007/BF01477472
24. Tran P, Tran L, Tran L. **A comparison of post-stroke hypertension medication use between US stroke belt and non-stroke belt residents**. *J Clin Hyperten.* (2021.0) **23** 1260-3. DOI: 10.1111/jch.14213
25. Sun Z, Han X, Zheng L, Zhang X, Li J, Hu D. **Subtypes of hypertension and risk of stroke in rural Chinese adults**. *Am J Hypertens.* (2014.0) **27** 193-8. DOI: 10.1093/ajh/hpt197
26. **Factors affecting development of stroke in hypertension patients**. (2019.0). DOI: 10.22900/kphr.2019.45.1.009
27. Modig K, Talbäck M, Ziegler L, Ahlbom A. **Temporal trends in incidence, recurrence and prevalence of stroke in an era of ageing populations, a longitudinal study of the total Swedish population**. *BMC Geriatrics* (2019.0) **19** 1. DOI: 10.1186/s12877-019-1050-1
28. Safiri S, Ashrafi-Asgarabad A. **Common and specific risk factors for ischemic stroke in elderly: differences based on type of ischemic stroke and aging; methodological issues**. *J Neurol Sci.* (2017.0) **383** 70-1. DOI: 10.1016/j.jns.2017.10.029
29. Retnakumar C, Chacko M, Ramakrishnan D, George LS, Krishnapillai V. **Prevalence of anemia and its association with dietary pattern among elderly population of urban slums in Kochi**. *J Family Med Primary Care.* (2020.0) **9** 1533. DOI: 10.4103/jfmpc.jfmpc_1113_19
30. Kurdi F, Abidin Z, Priyanti RP. **Management of diabetes mellitus type 2 for elderly: Taichi exercise to reduce blood sugar levels**. *KHD Production.* (2021.0) **2** 51. DOI: 10.53713/NHS.V1I2.51
31. Bathgate CJ, Edinger JD, Wyatt JK, Krystal AD. **Objective but not subjective short sleep duration associated with increased risk for hypertension in individuals with insomnia**. *Sleep.* (2016.0) **3** 5748. DOI: 10.5665/sleep.5748
32. Huang L, Long Z, Lyu J, Chen Y, Li R, Wang Y. **The associations of trajectory of sleep duration and inflammation with hypertension: a longitudinal study in China**. *Nat Sci Sleep.* (2021.0) **13** 1797-806. DOI: 10.2147/NSS.S329038
33. Patyar S. **Correlation between sleep duration and risk of stroke**. *J Nat Stroke Assoc.* (2015.0) **24** 905-11. DOI: 10.1016/j.jstrokecerebrovasdis.2014.12.038
34. Wada K, Eguchi H, Prieto-Merino D. **Differences in stroke and ischemic heart disease mortality by occupation and industry among Japanese working-aged men**. *SSM - Population Health.* (2016.0) **2** 745-9. DOI: 10.1016/j.ssmph.2016.10.004
35. Carlsson S, Andersson T, Talback M. **Occupation and incidence of type 2 diabetes: evidence from all Swedish employees 2006–2015**. (2019.0) **3** 62
36. Malstam E, Asaba E, Akesson E. **“Weaving lifestyle habits”: complex pathways to health for persons at risk for stroke**. *Scand J Occup Ther.* (2022.0) **29** 152-64. DOI: 10.1080/11038128.2021.1903991
37. Wang M, Brage S, Sharp SJ, Luo S, Yeung SLA, Kim Y. **Associations of genetic susceptibility and healthy lifestyle with incidence of coronary heart disease and stroke in individuals with hypertension**. *Eur J Prev Cardiol.* (2022.0). DOI: 10.1093/eurjpc/zwac135
38. Tran P, Tran L, Tran L. **A cross-sectional analysis of differences in physical activity levels between stroke belt and non-stroke belt US sdults**. *J Stroke Cerebrovasc Dis.* (2019.0) **28** 104432. DOI: 10.1016/j.jstrokecerebrovasdis.2019.104432
39. Shah AS, Lee KK, McAllister DA, Hunter A, Nair H, Whiteley W. **Air pollution and stroke. A new modifiable risk factor is in the air**. *Revue Neurologique.* (2019.0) **175** 619-24. DOI: 10.1016/j.neurol.2019.03.003
40. Verhoeven JI, Allach Y, Vaartjes ICH, Klijn CJM, de Leeuw FE. **Ambient air pollution and the risk of ischaemic and haemorrhagic stroke**. *The Lancet Planetary Health.* (2021.0) **5** e542-52. DOI: 10.1016/S2542-5196(21)00145-5
41. Jiang L, Bai L. **Spatio-temporal characteristics of urban air pollutions and their causal relationships: Evidence from Beijing and its neighboring cities**. *Sci Rep.* (2018.0) **8** 1. DOI: 10.1038/s41598-017-18107-1
42. Dong D, Xu X, Xu W, Xie J. **The relationship between the actual level of air pollution and residents' concern about air pollution: evidence from Shanghai, China**. *Int J Environ Res Public Health.* (2019.0) **16** 4784. DOI: 10.3390/ijerph16234784
43. Zhang H, Xia Y, Su H, Chang Q, Zhao Y. **Household solid fuel use and stroke incidence: evidence from a national cohort study**. *Front Public Health* (2022.0) **10** 8023. DOI: 10.3389/fpubh.2022.1018023
44. Sheikhbardsiri H, Mousavi SH, Doustmohammadi MM, Karimi M, Salahi S. **The effect of educational workshop on emergency department nurses' self-efficacy in patient training**. *J Edu Health Promot* (2019.0) **8** 19. DOI: 10.4103/jehp.jehp_52_19
|
---
title: Chemical, biological and in silico assessment of date (P. dactylifera L.) fruits
grown in Ha’il region
authors:
- Abdulmohsen Khalaf Dhahi Alsukaibi
- Khalaf M. Alenezi
- Ashanul Haque
- Irfan Ahmad
- Mohd Saeed
- Mahima Verma
- Irfan Ahmad Ansari
- Ming-Fa Hsieh
journal: Frontiers in Chemistry
year: 2023
pmcid: PMC10022733
doi: 10.3389/fchem.2023.1138057
license: CC BY 4.0
---
# Chemical, biological and in silico assessment of date (P. dactylifera L.) fruits grown in Ha’il region
## Abstract
Background: Dates palm (*Phoenix dactylifera* L.) fruits are among the most widely used fruits in the Middle East and African nations. Numerous researchers confirmed the presence of phytochemicals in P. dactylifera L. fruit and its by-products with broad-ranging biological activities.
Objectives: In the present work, phytochemical and biological assessments of two different cultivars of date fruit (Shishi M1 and Majdool M2 grown in the Ha’il region of Saudi Arabia) have been carried out.
Methods: Date fruits were extracted and analyzed by gas chromatography-mass spectrometry (GS-MS),liquid chromatography-mass spectrometry (LC-MS) and Fourier-transform infrared spectroscopy (FT-IR)techniques. The lyophilized methanolic extracts were analyzed for their in-vitro antiproliferative andcytotoxicity against colon cancer (HCT116) cell line. To identify the possible constituents responsible for the bioactivity, in-silico molecular docking and molecular dynamics (MD) simulation studies were carried out.
Results: Both cultivars exhibited in-vitro anticancer activity (IC50 = 591.3 μg/mL and 449.9 μg/mL for M1 and M2, respectively) against colon cancer HCT-116 cells. The computational analysis results indicated procyanidin B2 and luteolin-7-O-rutinoside as the active constituents.
Conclusion: Based on these results, we conclude that these cultivars could be a valuable source for developing health promoter phytochemicals, leading to the development of the Ha’il region, Saudi Arabia.
## Introduction
Natural products (NPs) play an instrumental role in drug design and remain an inspiration for discovering new drug candidates (Newman and Cragg, 2016). Being the largest source of new pharmacophores, $60\%$–$70\%$ of drugs used today are based directly or indirectly on NPs (Atanasov et al., 2021). Indeed, enormous diversity in their chemical structure, broad-ranging bioactivity, low toxicity, and ability to bind with different proteins (targets) gives natural compounds an edge over synthetic ones (Ali et al., 2010). In the quest for new drug candidates, research is being carried out on the extraction, isolation, and identification of bioactive compounds found in plants, animals and microbes (Altemimi et al., 2017). Among a large pool of NPs, the dates palm (*Phoenix dactylifera* L.), a member of the Asteraceae family, has garnered an immense interest (Al-Alawi et al., 2017; Maqsood et al., 2020). Date fruits of the date palm tree (P. dactylifera L.) is one of the most consumed fruits worldwide, especially in the Middle East and Asian countries (Complexity, 2022). Apart from their high nutritional and commercial values (Mia et al., 2020), date fruits and their by-products have also attracted researchers due to their potential health benefits (Maqsood et al., 2020). Their antibacterial, antifungal, antiviral, antidiabetic, anticancer, anti-inflammatory, antioxidant, antiangiogenic and other protective effects along with negligible side effects, are particularly interesting (Vayalil, 2002; Maqsood et al., 2020). It has been demonstrated that date fruits are rich in carbohydrates, protein, fibres, minerals, vitamins, phenolic acids, flavonoids, and other phytochemicals responsible for bioactivities. The chemical composition depends on various factors, including the type of cultivar, geographical location, irrigation method, ripening stage, processing time, extracting solvents, etc. ( Borochov-Neori et al., 2013) Based on this knowledge, various groups investigated the chemical composition of date fruits and seeds native to different regions (Vayalil, 2012; Al-Alawi et al., 2017; Mia et al., 2020; Echegaray et al., 2021; Ibrahim et al., 2021). The group led by Aviram (Borochov-Neori et al., 2013) has conducted studies on the chemical and biological analysis of several varieties of date fruits (Maqsood et al., 2020). They, along with others, confirmed that the phytoconstituents and bioactivity of the date fruits are the function of the parameters mentioned above. For example, a pilot study showed that Medjool or Hallawi varieties of date fruits vary in phenolics, catechins and quercetin derivative content and antioxidant effect (Rock et al., 2009). The same group reported anti-atherogenic properties of acetone extracts of Hallawi in addition to the eight other variants (Borochov-Neori et al., 2013). As per the group, phenolic compounds exerted anti-atherogenic properties via low-density lipoprotein (LDL) oxidation and serum-mediated cholesterol efflux. On the other hand, alcoholic extract of the Tunisian variety was found to inhibit α-glucosidase and α-amylase enzymes with low IC50 values (El Abed et al., 2017). Zhang et al. [ 2013] performed an extensive chemical and biochemical profiling of Ajwa date fruits. They identified several new compounds such as bis (2-ethylhexyl) terephthalate and bis (2-ethylheptyl) phthalate) in addition to glycoside, terpenoids, triglyceride, phthalates, etc. They noted that the aqueous and organic extract exerts dose-dependent antioxidant and anti-inflammatory effects.
In Saudi Arabia, more than 400 varieties of date fruit are cultivated, which vary in appearance, nutrition, and nutritional value (Zhang et al., 2017). Several researchers reported that the varieties such as Barni, Khalas, and Ajwa show unique biological activities (Eid et al., 2013; Assirey, 2015; Hamad et al., 2015). Recently, Amir and co-workers (Alghamdi et al., 2018) studied the nutritional value of several varieties of date fruits found in Ha’il province; however, the biological activity of the date fruits remains unclear. Prompted by this, we carried out extraction, characterization, in-vitro antiproliferative and cytotoxicity assay of two varieties of date fruits (Shishi M1 and Majdool M2) grown in the Ha’il region of Saudi Arabia). Then, to further identify the possible potential constituent present, ligand-based virtual screening was performed. This is followed by molecular dynamics simulation studies to identify the stability of promising compounds with possible receptor.
## General
All solvents used for isolation and purification were of ACS reagent grade (Sigma-Aldrich Chemical Co., St. Louis, MO, United States). Lyophilization was carried out on BenchTop Manifold Freeze Dryer (MILLROCK, United States) for 24 h at a condenser temperature of −45°C equipped with Edward pump. Attenuated-total-reflectance IR spectra were recorded on pure samples on diamond using a Shimadzu IRSpirit-T spectrometer.
## Sample collection, extraction, and sample preparation
Two different varieties of date fruits (Tamr stage, Supplementary Table S1) grown in Ha’il province were collected from the local market. Authors (AKDA and KMA) and local farmers authenticated the samples, and a voucher specimen was deposited. The samples were stored and kept in a −20°C freezer. First, three pieces of date fruits from each variant (Supplementary Table S1) were pitted to remove seeds and cut into small pieces. Then, cold extraction was performed by shaking and mixing fruit materials in methanol (MeOH) overnight at room temperature, followed by filtration. The residue was further extracted twice with MeOH for 1 h. Finally, the extracts were combined and concentrated using rotatory evaporation at room temperature. The resulting viscous honey-like liquid was lyophilized to afford light yellow water-soluble powder and stored at −20°C till further analysis.
## Chromatographic studies
Date fruit extracts were analyzed by LC-MS system using a reverse phase C18 column (Accucore, 150 × 4.6, 2.6 μm). The LC-MS system comprised a Waters Alliance 2695 HPLC pump, an autosampler, a vacuum degasser, and a column compartment attached to a XEVO-TQD detector with electrospray ionization (ESI). The following gradient of solvents were used: acetonitrile (mobile phase A) and 5 mM acetic acid (mobile phase B); ratio of A to B, 0–1 min, 5:95; 1–10 min, 5:95 to 30:70; 10–16 min, 30:70 to 60:40; 16–24 min, 60:40 to 80:20; 24–32 min, 80:20; 32–40 min, 5:95. In all cases, the columns were reequilibrated between injections with the equivalent of 5 mL of the mobile phase. During the full scan by MS/MS, mass acquisition was set from 150 to 2000 Da. This method utilized ESI-LC/MS/MS operating in MRM mode. The ESI settings were the following capillary voltage, 3.5 kV; cone voltage, 40 V; the flow of desolvation gas (Argon gas), 650 L/h; flow of cone gas, 30 L/h. Gas Chromatography-Mass Spectrometry (GC-MS) analysis was carried out using Agilent $\frac{8890}{5977}$B Series (Agilent 5977B EI/CI MSD) spectrometer. The segments were recognized by examination of their delay times and mass spectra with those of the NIST 11 mass spectral database.
## Cell culture and maintenance
The human colon cancer cell HCT116 was acquired from American Type Culture Collection (ATCC). McCoy’s 5 A media supplemented with $10\%$ v/v Fetal Bovine Serum (FBS), and $1\%$ antibiotic-antimycotic solution (1 mL contains 10,000 U Penicillin, 10 mg Streptomycin and 25 µg Amphotericin B) was used to grow and maintain HCT116 cells. A humidified environment constituted the standard conditions for cell culture at 37°C with $5\%$ CO2.
## Cell viability assay
To determine the cytotoxicity of M1 and M2 extracts on colon cancer HCT-116 cell line, MTT assay was used. In a 96-well plate, the cells (5 × 103 cells/well) were cultured for 24 h. The cells were treated with M1 and M2 at varied concentrations (100, 500, 1,000, and 5,000 μg/mL) for 24 h, respectively. Each well received 10 µL of MTT solution (5 mg/mL) and was subjected to further 3 h incubation at 37°C. In order to dissolve the purple formazan crystals, 100 µL of dimethyl sulfoxide (DMSO) was added to each well. A microplate reader measured absorbance at 570 nm (Bio-Rad, United States). The cell viability was expressed as a percentage (%) over the untreated control. For calculating the IC50 value, GraphPad Prism Professional software was used.
## Morphological analysis
The effects of M1 and M2 extract treatment on the morphology of HCT-116 cells were investigated using a phase contrast microscope. Briefly, HCT-116 cells (5 × 103) were cultivated in a 96-well plate before M1 and M2 (100, 500, 1,000, and 5,000 μg/mL) treatment. The alteration in morphology of M1 and M2-treated cells was then examined using a phase contrast microscope (Labomed, United States).
## Trypan blue exclusion assay
Trypan blue dye exclusion assay was performed further to confirm the M1 and M2-mediated cytotoxicity in HCT-116 cells. A hemocytometer and a microscope were used to count the cells (5 × 104) after they had been co-cultured with and without M1 and M2 (100, 500, 1,000, and 5,000 μg/mL), respectively for 24 h. The proportion of dead cells in each treatment set from studies done in triplicates was used to express the results.
## Lactate dehydrogenase release assay
In accordance with the manufacturer’s instructions, the lactate dehydrogenase (LDH) release assay kit was used to measure the level of cellular cytotoxicity. First, M1 and M2 were applied to the HCT-116 cells using a 96-well plate at different doses (100, 500, 1,000, and 5,000 μg/mL) for 24 h. LDH release kit was then used to detect released LDH in both the M1 and M2—treated HCT-116 cells in the incubation medium.
## Computational details
Computational calculations were carried out on a Dell workstation (Galax GeForce GTX 1660 Ti) equipped with 8-core processors, 64 GB Ram, and NVIDIA graphics card.
## Receptor and ligands preparation
The co-crystal structure of Bcl-2 complex (PDB ID: 5JSN) was selected for virtual screening and molecular dynamics studies. In the crystallographic structure of this complex, there is a gap at position 33–86, which was fixed by homology modelling using the Swiss modeller tool. Based on literature reports (Mia et al., 2020), a total of ninety-four ($$n = 94$$) phytoconstituents of different chemical classes (Supplementary Table S2) were selected. Chemical structures (.sdf format) of the compounds were retrieved from the NCBI PubChem database (Wang et al., 2009). The downloaded files were converted to.pdb format using the Open Babel software. The ligand files were prepared using AutoDock Tools 1.5.7 (the Scripps Research Institute, La Jolla, CA, United States) software and finally written as.pdbqt file format for docking studies (Ahamad et al., 2021a).
## Active site prediction
The Bcl-2 protein (PDB ID: 5JSN) was given as input to identify the active site, which gives significant insight into recognizing surface structural pockets, shape and volume of every pocket, internal cavities of protein and surface areas. Next, the active site and the interactive residues were selected using PDBsum and CASTp online tools (Laskowski et al., 2018; Tian et al., 2018). The ligands were prepared using AutoDock Tools (ADT), and saved in pdbqt format (Trott and Olson, 2010).
## Protein preparation and grid generation
The 3D structure of Bcl-2 was prepared using the ADT protein preparation wizard. The polar and missing hydrogen atoms were added, while water molecules and hetero-atoms were deleted (Forli et al., 2016). Energy minimization was performed with a default constraint of 0.3 Å root mean square (RMS) and charges were assigned. After protein preparation, clean structure was saved as pdbqt file. Grid box (84 Å × 82 Å × 84 Å) was generated around the centroid of compounds with assigned X, Y, and Z axis.
## Virtual screening and binding affinity calculation
To identify the potential compounds found in P. dactylifera L., a dataset of ninety-four compounds was utilized for the virtual screening. The pdbqt files were provided as input and screened against Bcl-2 (Duffy and Avery, 2012). Top two compounds (ranked based on the binding energy scores and the docking poses) were selected for further studies (Trott and Olson, 2010; Forli et al., 2016). The compounds with favourable binding poses were identified with the help of the lowest free energy (ΔG), defined using the equation as follows, ∆G=∆Gcomplex – ∆Genzyme+∆Gligand Where (∆Gcomplex), (∆Greceptor), and (∆Gligand) are the average values of Gibbs free energy for the complex, receptor, and ligand, respectively. The stability of the docked complex between the receptor-ligand exhibits more negative scores, revealing the high potency of the inhibitor. All the other docking parameters were kept default, and the docked complexes final visualization was performed using PyMOL tool (DeLano, 2002). The active pocket of Bcl-2 and docked pose of the top-ranked compounds were compared to find interactive orientations.
## Molecular dynamics (MD) simulation
MD simulations were performed for the best-docked complexes with maximum binding affinity scores using GROningen MAchine for Chemical Simulations (GROMACS) version 5.18.3. Package (Abraham et al., 2015). The topology of Bcl-2 was generated using GROMOS9643a1 force field (Van Der Spoel et al., 2005). Due to the lack of suitable force field parameters for a drug-like molecule in the GROMACS software, the PRODRG server was used for the generation of molecular topologies and coordinate files (Schüttelkopf and Van Aalten, 2004). All the systems were solvated using a simple point charge model (SPC/E) in a cubic box. To neutralize the system 0.15 M counter ions (Na+ and Cl−) were added. The energy minimization of all the neutralized systems was performed using the steepest descent and conjugate gradients (50,000 steps for each). The constant number of particles, volume, and temperature (NVT) ensemble and constant number of particles, pressure, and temperature (NPT) ensemble were run for system equilibration (Ahamad et al., 2021b). Steepest descent followed by conjugate gradient algorithms was utilized on enzyme-ligand complexes. The NVT ensemble was employed at a constant temperature of 300 K and a constant pressure of 1bar. The SHAKE algorithm was used to confine the H atoms at their equilibrium distances and periodic boundary conditions. Moreover, the Particle Mesh Ewald (PME) method defines long-range electrostatic forces (Abraham et al., 2015). The cut-offs for van der Waals and columbic interactions were set as 1.0 nm. LINC algorithm was used to constrain the bonds and angles. Using the NPT ensemble, production runs were performed for 500 ns, with time integration. The energy, velocity, and trajectory were updated at the time interval of 10 ps *The analysis* is performed by using Cα-atom deviations of the protein calculated using root mean square deviations (RMSD). The relative fluctuations of each amino acid were defined with root mean square fluctuations (RMSF). To measure the compactness of a given molecule radius of gyrations (Rg) is implemented, and the solvent accessible surface area (SASA) was employed to know the electrostatic contributions of molecular solvation (Ahamad et al., 2018; Ahamad et al., 2019).
## Extraction and characterization of Shishi (M1) and Majdool (M2) fruits
Aqueous, organic or mixture solvents can extract a natural product, depending upon the analyte of interest (Nematallah et al., 2018). In the past, it has been demonstrated that when date fruits are extracted with an organic solvent, it yields bioactive compounds able to inhibit colon, liver and cervical cancerous cell lines in vitro (Mansour et al., 2011; Ravi, 2017). Especially, using a polar solvent such as methanol allows the extraction of various components (MHM et al., 2015). It was found that the alcoholic extract of the date fruits effectively inhibits α-glucosidase and α-amylase enzymes with low IC50 values in both in-vitro and in-vivo (El Abed et al., 2017). In 2016, Khan et al. [ 2016] demonstrated that the methanolic extract of Ajwa Date (Saudi origin) inhibits breast cancer (MCF-7) cell lines via cell cycle arrest and apoptosis. Motivated by this, we selected methanol as the extracting solvent in this study too. Cold extraction of finely cut M1 (36.11 gm) and M2 (46.22 gm) followed by concentration and lyophilization yielded M1 and M2 as a light-yellow water-soluble powder (9.1 and 15 gm of M1 and M2, respectively). To identify the extract components, chromatographic (LC-MS & GC-MS) and spectroscopic (FT-IR) techniques (Supplementary Figures S1, S2) were employed.
Lyophilized products were analyzed by LC-MS (negative mode) using gradient mobile phase as it allows easy detection of phenolic acids & flavonoids as they contain acidic hydroxy group. It has been reported that the methanolic extract contains phenolic acids, flavonoid diglucosides, monoglucosides, acylated monoglucosides, free aglycones, lipids and others when analyzed under similar conditions (Farag et al., 2014). Table 1 collects the identities and molecular/fragment ions of some major components present in M1 and M2 as identified by comparing LC-MS (negative mode, Figure 1) results with the literature. Both varieties exhibit similar chromatograms, with M1 having relatively more fraction than M2.
GC-MS analysis further indicated the presence of several phytochemical belongings of different classes. For example, quinic acid, oleic acid, trans-13-octadecenoic acid, stearic acid, O-caffeoyl shikimic acid, luteolin, trihydroxy-octadecenoic acid, stearic acid linoleic acid, 6-hydroxy 7 methoxy coumarin, 4-hydroxy 6-methylcoumarin and amino acids were tentatively identified (Supplementary Table S3). These components and other metabolites have been well-identified in different varieties of date fruits (Farag et al., 2016; Abdul-Hamid et al., 2019; Perveen and Bokahri, 2020; Souda et al., 2020; Ibrahim et al., 2021).
The FTIR spectrum of the methanolic extracts (Supplementary Figure S1) is also consistent with previous literature (Alam et al., 2022). It has been reported that IR spectrum of dates extracts exhibits multiple peaks responsible for functionalities present in lipid (2,960–2,850 cm−1), amide (3,299–3,399 cm−1 and 1,591–1,529 cm−1 for amine and 1,619–1,691 for carbonyl) and carbohydrates (900–1,200 cm−1). As it is clear, the IR spectra of M1 and M2 are identical. The spectrum shows a stretching vibrations band at 3,280 cm−1 attributed to -OH group, bands at 2,888 and 2,930 cm−1 attributed to Csp3-H stretching vibration, aromatic C=C stretching vibrations at 1,622 cm−1 and C–O deformation vibrations of aliphatic alcohols at 1,009 cm−1 (Alam et al., 2022).
## Antiproliferative and cytotoxic effect
Using the MTT test, the antiproliferative and cytotoxic effects of the M1 and M2 date extract were assessed against colon cancer HCT116 cells for 24 h (Figures 2A, B). The M1 and M2 extracts exhibited strong and dose-dependent cytotoxic potential in HCT116 cells. The % cell viability of M1-and M2-treated HCT116 cells were found to be $88.80\%$ ± $1.33\%$, $63.26\%$ ± $3.47\%$, $45.24\%$ ± $2.80\%$, and 15.28 % ± $1.53\%$; and $83.07\%$ ± $2.37\%$, $59.35\%$ ± $4.72\%$, $28.90\%$ ± $1.49\%$ and 10.63 % ± $1.47\%$ at a dose of 100, 500, 1,000, and 5,000 μg/mL, respectively. IC50 values were determined to be 591.3 μg/mL and 449.9 μg/mL for M1 and M2, respectively, revealing the inhibitory potential (Figures 2C, D). Our findings thus demonstrated that both the M1 and M2 inhibit colon cancer cell proliferation in a dose-dependent manner.
**FIGURE 2:** *Effect of date extracts M1 and M2 on HCT-116 cells. (A, B) Percent (%) cell viability of HCT-116 cells treated with different doses of M1 and M2 (100–5,000 μg/mL) for 24 h. The results shown are the mean ± SEM of three independent experiments performed in triplicate (ns > 0.01, *p < 0.01, **p < 0.001, and ***p < 0.0001 represent significant differences compared with control). (C, D) Graph showing IC50 of M1 and M2 against HCT-116 colon cancer cell at 24 h.*
## Morphological alterations
Under a phase contrast microscope, the images of control and M1 & M2-treated HCT-116 cells revealed discernible morphological alterations. The control cells showed increased cell growth and intact cell shape. However, in a dose-dependent manner (100, 500, 1,000, and 5,000 μg/mL), significant morphological modifications were observed in the M1 and M2-treated HCT-116 cells (Figures 3A, B). Moreover, M1 and M2-treated HCT-116 cells showed increased detachment and cytoplasmic shrinkage, which led to a rise in the number of floating cells. The findings thus support the hypothesis that treatment with M1 and M2 causes cytotoxicity in HCT-116 colon cancer cells.
**FIGURE 3:** *(A): Phase-contrast images of HCT-116 cells treated with either vehicle or different doses of M1 (100–5,000 μg/mL) for 24 h. The photomicrographs shown are the representatives of three independent experiments. (B): Phase-contrast images of HCT-116 cells treated with either vehicle or different doses of M2 (100–5,000 μg/mL) for 24 h. The photomicrographs shown are the representatives of three independent experiments.*
## M1 and M2 causes cell death in HCT-116 cells
Trypan blue dye exclusion assay was used to assess how M1 and M2-treated HCT-116 cells lost viability. Figures 4A, B illustrates the considerable increase in cell mortality in HCT-116 cells after exposure to M1 and M2 at various doses (100, 500, 1,000, and 5,000 μg/mL) for 24 h. This result supported the cytotoxic action of M1 and M2 on colon cancer cells.
**FIGURE 4:** *Trypan blue dye exclusion assay. Percent (%) dead cells in HCT-116 cells treated with different doses of (A) M1 and (B) M2 (100–5,000 μg/mL) for 24 h. The results shown are the mean ± SEM of three independent experiments performed in triplicate (ns > 0.01, *p < 0.01, **p < 0.001, and ***p < 0.0001 represent significant difference compared with control).*
## Release of cellular LDH in HCT-116 cells
LDH release assay displayed that treatment with both the M1 and M2 in HCT-116 cells mediated significant release of LDH, which showed the degree of cellular membrane damage post-treatment. Higher M1 and M2 concentrations were found to be significantly more cytotoxic, as evidenced by increased cytotoxicity in HCT-116 cells (Figures 5A, B). The percentage cytotoxicity in M1-treated HCT-116 cells, after 24 h of treatment, was found to be $112.08\%$ ± $3.42\%$, $145.78\%$ ± $3.88\%$, $174.07\%$ ± $2.25\%$, and $190.03\%$ ± $2.64\%$ at 100, 500, 1,000, and 5,000 μg/mL dose, respectively. Similarly, after 24 h of treatment with M2, HCT-116 cells exhibited percent cytotoxicity of $120.53\%$ ± $3.13\%$, $144.07\%$ ± $3.00\%$, $168.46\%$ ± $4.29\%$, $192.15\%$ ± $1.98\%$ at 100, 500, 1,000, and 5,000 μg/mL dose, respectively. Thus, our results suggest that both the M1 and M2 were able to decrease the viability and proliferation in colon cancer cells.
**FIGURE 5:** *LDH release assay. Percent cytotoxicity in HCT-116 cells treated with different doses of (A) M1 and (B) M2 (100–5,000 μg/mL) for 24 h. The results shown are the mean ± SEM of three independent experiments performed in triplicate (ns > 0.01, *p < 0.01, **p < 0.001, and ***p < 0.0001 represent significant differences compared to control).*
## Virtual screening
It has been demonstrated that the *Phoenix dactylifera* L. extract exhibits anticancer activity by modulating Bcl-2-family proteins (Khan et al., 2021). To identify the critical component(s) responsible for the anticancer activity, we conducted exhaustive in silico studies. To this end, ligand-based virtual screening was performed using ninety-four compounds found in P. dactylifera against the receptor (PDB: 5JSN). It has been reported that the small molecule may interact with the various receptors of Bcl-2 protein via multiple non-covalent interactions (Khosravi et al., 2022; Taghizadeh et al., 2022). Among others, Lys22, Arg26, Asp102, Ser105, Arg106, Arg109, Phe112, Val156, Val159, Asp163, Glu160, and Glu209 which participates in H-bonds and steric interactions (Khosravi et al., 2022). Based on the free binding energies and docking poses, virtual screening of the ligands resulted in procyanidin B2 and luteolin-7-O-rutinoside as the most potent candidates (Table 2). As depicted in Figure 6A, procyanidin B2 interacted with various amino acid residues via H-bonding (Ala100, Arg107, Asn143, Gly145, and Arg146) and other non-covalent interactions (such as hydrophobic and Van der Waal’s) with a total binding energy of −9.3 kcal/mol. On the other hand, luteolin-7-O-rutinoside formed H-bond with Asp111, Asn143 and Arg146 amino acids and yielded a binding energy of −9.1 kcal/mol (Figure 6B). Overall results revealed that the proposed two compounds have an edge over the Bcl-2 complexes attributable to more potent binding abilities.
## Molecular dynamics (MD) simulations
To understand the complex stability and interaction profile of the most promising hit compounds inside the active site of Bcl-2, MD simulations of Bcl-2-native, procyanidin B2 and luteolin-7-O-rutinoside complexes were performed on a 500 nanosecond (ns) scale. In addition, structural parameters, including RMSD, RMSF, SASA, and Rg were evaluated as a function of time and discussed in the following sub-sections.
## RMS-deviation and RMS-fluctuations
The docked complexes were subjected to RMSD analysis to assess the residual flexibility of the Bcl-2 receptor. It was noted that the native protein exhibits higher RMSD fluctuation and reaches equilibrium between 0.8 nm and 1.0 nm. However, in the presence of procyanidin B2, it reached an equilibrium at 0.6 nm and showed steady RMSD (average RMSD value 0.92 nm, Table 3), which remained stable over the 500 ns MD simulation (Figure 7A). Similarly, luteolin-7-O-rutinoside and Bcl-2 complexes showed stable equilibrium at 0.6 nm–0.7 nm. Furthermore, they displayed minimal fluctuation over the 500 ns MD simulation. The average RMSD of Luteolin-7-O-rutinoside and Bcl-2 complexes was 0.71 nm. Overall, both procyanidin B2 and luteolin-7-O-rutinoside complexes exhibit stable RMSD values and have a stable binding with Bcl-2 under the given simulation conditions. This also indicates that the studied compounds reached stable and reliable dynamic equilibriums, which bolstered the docking results.
Furthermore, RMSF analysis was implemented to identify the flexible and rigid regions of the complexes and to measure the average atomic flexibility of the Cα-atoms of native Bcl-2 and docked complexes. In the case of native Bcl-2, amino acids residues such as Pro46 ∼0.53 nm, Gly47 ∼ 0.71 nm, Ile48 ∼ 0.64 nm, Arg63 ∼ 0.58 nm, Asp64 ∼ 0.65 nm, Pro65 ∼ 0.73 nm and Val66 ∼ 0.67 nm showed higher fluctuations (Figure 7B). However, fluctuation at 104–112, 162–163 and 201–207 amino acids residue also was found to be higher while other amino acids remain stable. For example, in a complex with procyanidin B2, RMS fluctuations were found in the region Gly79 ∼ 0.60 nm and Ala80 ∼0.57 nm, which is acceptable as these amino acids did not participate in the binding. Similarly, the complex with luteolin-7-O-rutinoside showed RMS-fluctuations at Gly54 ∼0.57 nm, Ala61 ∼0.55 nm and Arg63 ∼0.66 nm values. Overall, the RMSF displayed the highest degree of flexibility, exhibiting stable active site residues interaction compared to the native protein.
## Hydrogen bond monitoring
To underpin the stability of the ligand-protein complex, the number of H-bond was monitored by analyzing the MD trajectories (Figures 7C, D). As can be seen, both compounds procyanidin B2 and luteolin-7-O-rutinoside formed 17 and 22 hydrogen bonds, respectively, which increased/remained the same during the 500 ns MD simulation.
## Radius of gyration (Rg) and solvent accessible surface area (SASA)
Rg helps determine protein folding and unfold upon ligand binding, thus giving an idea about the stability of the complex during the simulation. A higher Rg indicates a less compact structure, while a lower Rg means more compactness (Sharma et al., 2022). We found that the average Rg values for the native Bcl-2 protein (1.56 nm) and luteolin-7-O-rutinoside complex (1.57 nm) were almost similar, indicating that the protein will likely maintain a relatively steady value and is stably folded (Figure 8A; Table 2). However, in the case of the procyanidin B2 complex, the average Rg value was 1.65 nm, indicating unfolded structure.
**FIGURE 8:** *Rg (A) and SASA plot (B) during 500 ns MD simulations docked complexes of native Bcl-2, and complexes with procyanidin B2 and luteolin-7-O-rutinoside.*
SASA was also conducted to ascertain the interactions between the protein-ligand complex and solvent during the 500 ns MD simulation (Figure 8B; Table 2). It was noted that the average SASA value for the complexes (112.77 and 115.39 nM2) of procyanidin B2 and luteolin-7-O-rutinoside, respectively, was better than the native Bcl-2 protein (111.73 nM2).
## Discussion
It has been long understood that the phytochemicals found in *Phoenix dactylifera* L. target and inhibit several important biochemical pathways contributing to disease development (Farag et al., 2014; Lamia and Mukti, 2021). The ethnopharmacological significance of P. dactylifera L., such as antioxidant, anti-inflammatory anticancer, antimicrobial, etc., is now well established (El Abed et al., 2018). The amount of phytoconstituents, and thus the bioactivity, depends on several factors, including the part of the plant (fruits, seeds, etc.), stage, geographical location, and others. For example, it has been demonstrated that date fruit seeds extract shows anticancer activity against pancreatic (Habib et al., 2014), colorectal (Rezaei et al., 2015), liver (Al-Sheddi, 2019), lung (Al-Sheddi, 2019), and breast (Al-Sheddi, 2019) and other (Al-Zubaidy et al., 2016; Hilary et al., 2021; Habib et al., 2022; Khan et al., 2022) cancer cell lines. On the other hand, Siddiqui et al. [ 2019] reported that the pulp extract of the Ajwa variety exhibit antiproliferative activity against human liver cancer cells (HepG2, IC50 = 20.03 and 16.78 mg/mL at 24 and 48 h periods, respectively). Moreover, Khan et al. [ 2021] demonstrated the apoptosis-inducing potential of Ajwa date pulp extract against human triple-negative breast cancer cells (MDA-MB-231, IC50 = 17.45 and 16.67 mg/mL at 24 and 48 h, respectively). Khattak et al. [ 2020] have reported the antiproliferative property of Emirati date fruits extract on human triple-negative breast cancer cell line MDA-MB-231. Besides, the antioxidant and apoptotic potentials of the whole fruit (flesh and pit extracts) is also known (Shahbaz et al., 2022). In addition to the above-mentioned factors, the polarity of extracting solvents also plays an important role; therefore both aqueous and organic solvent systems have been investigated in the past. In a study, it was found that the aqueous extract of a number of date varieties (Saudi Arabian origin) was less bioactive than the methanolic counterparts (Zhang et al., 2017). In a remarkable study, Khan et al. [ 2016] noted that the methanolic extract of Ajwa date fruits exhibit strong anticancer effect on human breast adenocarcinoma (MCF7) (Khan et al., 2016). Besides, other researchers also noted the antitumor activity of methanolic extracts (Mansour et al., 2011; Thouri et al., 2019). Therefore, in the present study, we selected methanol as the solvent to extract date fruits of Shishi (M1) and Majdool (M2) cultivars grown in Ha’il region of Saudi Arabia.
As mentioned, (vide-infra), the methanolic extract concentrates were subjected to lyophilization and the resulting water-soluble products were used for further studies without any further purification. LC-MS (negative mode) and GC-MS analyses of the extract revealed the presence of various phytochemicals in both varieties. We noted that the chromatograms of M1 have more peaks than M2; therefore, the former has relatively more constituents. Among the main constituents identified were flavonoids, sphingolipids, and fatty acids classes of phytochemicals. Several researchers already report the presence of these constituents in a wide variety of date fruits (see references in the result section). At the same time, we firmly believe the presence of other components escaped the detection. MTT assay of the extracts against colon cancer cells (HCT-116) revealed a dose-dependent inhibitory nature of the compounds (IC50 = 591.3 μg/mL and 449.9 μg/mL for M1 and M2 at 24, respectively).
It has been demonstrated that the P. dactylifera L. extract exhibits anticancer activity by modulating Bcl-2-family proteins which is also expressed in the HCT-116 cell line (El-Far et al., 2021). Considering this, attempts have been made to identify the principal agent(s) present in the extract using computational approaches, including virtual screening and MD studies. Ligand-based virtual screening identified procyanidin B2 and luteolin-7-O-rutinoside as the most probable candidates since they could bind with Bcl-2 protein efficiently through various amino acids. MD simulation study further strengthens this observation. Considering the earlier reported values and inhibition mechanism on other cell lines, we believe that the anticancer potential of both Shishi and Majdool date extracts against colon cancer cells is interesting and requires further biochemical investigation.
## Conclusion
In conclusion, the anticancer activity of methanolic extract of two varieties of dates fruits (Shishi M1 and Majdool M2) grown in the Ha’il region of Saudi Arabia has been compared. The results of GC-MS and Ft-IR studies indicated the presence of various components in the M1 and M2 extracts, which are responsible for dose-dependent cytotoxicity against colon cancer cells (HCT116 cells) through morphological modifications, including cellular membrane damage. The IC50 value was 591.3 μg/mL and 449.9 μg/mL for M1 and M2, respectively. Furthermore, Trypan blue dye exclusion assay further supported the cytotoxic action of M1 and M2 on colon cancer cells. Extensive virtual screening combined with MD simulations studies at 500 ns revealed that procyanidin B2 and luteolin-7-O-rutinoside could be possible agents for the bioactivities. Overall, our data strongly suggest that the consumption of date fruits might prove helpful against colon cancer. Moreover, we also believe that these varieties of date fruits could be utilized as a source of bioactive phytochemicals, leading to the development of Ha’il, Saudi Arabia.
## Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.
## Author contributions
All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem.2023.1138057/full#supplementary-material
## References
1. Abdul-Hamid N. A., Abas F., Ismail I. S., Tham C. L., Maulidiani M., Mediani A.. **Metabolites and biological activities of phoenix dactylifera L. pulp and seeds: A comparative MS and nmr based metabolomics approach**. *Phytochem. Lett.* (2019) **31** 20-32. DOI: 10.1016/j.phytol.2019.03.004
2. Abraham M. J., Murtola T., Schulz R., Páll S., Smith J. C., Hess B.. **Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers**. *SoftwareX* (2015) **1** 19-25. DOI: 10.1016/j.softx.2015.06.001
3. Ahamad S., Hassan M. I., Dwivedi N.. **Designing of phenol-based β− carbonic anhydrase1 inhibitors through QSAR, molecular docking, and MD simulation approach**. *3 Biotech.* (2018) **8** 256. DOI: 10.1007/s13205-018-1278-z
4. Ahamad S., Islam A., Ahmad F., Dwivedi N., Hassan M. I.. **2/3D-QSAR, molecular docking and MD simulation studies of FtsZ protein targeting benzimidazoles derivatives**. *Comput. Biol. Chem.* (2019) **78** 398-413. DOI: 10.1016/j.compbiolchem.2018.12.017
5. Ahamad S., Kanipakam H., Birla S., Ali M. S., Gupta D.. **Screening Malaria-box compounds to identify potential inhibitors against SARS-CoV-2 Mpro, using molecular docking and dynamics simulation studies**. *Eur. J. Pharmacol.* (2021a) **890** 173664. DOI: 10.1016/j.ejphar.2020.173664
6. Ahamad S., Kanipakam H., Kumar V., Gupta D.. **A molecular journey to check the conformational dynamics of tau tubulin kinase 2 mutations associated with Alzheimer's disease**. *RSC Adv.* (2021b) **11** 1320-1331. DOI: 10.1039/d0ra07659g
7. Al-Alawi R. A., Al-Mashiqri J. H., Al-Nadabi J. S., Al-Shihi B. I., Baqi Y.. **Date palm tree (phoenix dactylifera L.): Natural products and therapeutic options**. *Front. plant Sci.* (2017) **8** 845. DOI: 10.3389/fpls.2017.00845
8. Al-Sheddi E.. **Anticancer potential of seed extract and pure compound from Phoenix dactylifera on human cancer cell lines**. *Pharmacogn. Mag.* (2019) **15** 494-499. DOI: 10.4103/pm.pm_623_18
9. Al-Zubaidy N., Al-Zubaidy A., Sahib H.. **The anti-proliferative activity of phoenix dactylifera seed extract on MCF-7 breast cancer cell line**. *Int. J. Pharm. Sci. Rev. Res.* (2016) **41** 358-362
10. Alam M. Z., Ramachandran T., Antony A., Hamed F., Ayyash M., Kamal-Eldin A.. **Melanin is a plenteous bioactive phenolic compound in date fruits (Phoenix dactylifera L.)**. *Sci. Rep.* (2022) **12** 6614-6712. DOI: 10.1038/s41598-022-10546-9
11. Alghamdi A. A., Awadelkarem A. M., Hossain A., Ibrahim N. A., Fawzi M., Ashraf S. A.. **Nutritional assessment of different date fruits (Phoenix dactylifera L.) varieties cultivated in Hail province Saudi Arabia**. *Biosci. Biotechnol. Res. Commun.* (2018) **11** 263. DOI: 10.21786/bbrc/11.2/11
12. Ali I., Rahisuddin S. K., Haque A., El-Azzouny A.. **Natural products: Human friendly anticancer medications**. *Egypt Pharma J.* (2010) **9** 133-179
13. Altemimi A., Lakhssassi N., Baharlouei A., Watson D. G., Lightfoot D. A.. **Phytochemicals: Extraction, isolation, and identification of bioactive compounds from plant extracts**. *Plants* (2017) **6** 42. DOI: 10.3390/plants6040042
14. Assirey E. R.. **Nutritional composition of fruit of 10 date palm (Phoenix dactylifera L.) cultivars grown in Saudi Arabia**. *J. Taibah Univ. Sci.* (2015) **9** 75-79. DOI: 10.1016/j.jtusci.2014.07.002
15. Atanasov A. G., Zotchev S. B., Dirsch V. M., Supuran C. T.. **Natural products in drug discovery: Advances and opportunities**. *Nat. Rev. Drug Discov.* (2021) **20** 200-216. DOI: 10.1038/s41573-020-00114-z
16. Borochov-Neori H., Judeinstein S., Greenberg A., Volkova N., Rosenblat M., Aviram M.. **Date (Phoenix dactylifera L.) fruit soluble phenolics composition and anti-atherogenic properties in nine Israeli varieties**. *J. Agric. food Chem.* (2013) **61** 4278-4286. DOI: 10.1021/jf400782v
17. Complexity O.-T. O. O. E.. **Dates, fresh or dried**. (2022)
18. Delano W. L.. **Pymol: An open-source molecular graphics tool**. *CCP4 Newsl. protein Crystallogr.* (2002) **40** 82-92
19. Duffy S., Avery V. M.. **Development and optimization of a novel 384-well anti-malarial imaging assay validated for high-throughput screening**. *Am. J. Trop. Med. Hyg.* (2012) **86** 84-92. DOI: 10.4269/ajtmh.2012.11-0302
20. Echegaray N., Gullón B., Pateiro M., Amarowicz R., Misihairabgwi J. M., Lorenzo J. M.. **Date fruit and its by-products as promising source of bioactive components: A review**. *Food Rev. Int.* (2021) **2021** 1-22. DOI: 10.1080/87559129.2021.1934003
21. Eid N. M., Al-Awadi B., Vauzour D., Oruna-Concha M. J., Spencer J. P.. **Effect of cultivar type and ripening on the polyphenol content of date palm fruit**. *J. Agric. food Chem.* (2013) **61** 2453-2460. DOI: 10.1021/jf303951e
22. El Abed H., Chakroun M., Abdelkafi-Koubaa Z., Drira N., Marrakchi N., Mejdoub H.. **Antioxidant, anti-inflammatory, and antitumoral effects of aqueous ethanolic extract from Phoenix dactylifera L. parthenocarpic dates**. *BioMed Res. Int.* (2018) **2018** 1542602. DOI: 10.1155/2018/1542602
23. El Abed H., Chakroun M., Fendri I., Makni M., Bouaziz M., Drira N.. **Extraction optimization and**. *Biomed. Pharmacother.* (2017) **88** 835-843. DOI: 10.1016/j.biopha.2017.01.129
24. El-Far A. H., Ragab R. F., Mousa S. A.. *Date palm bioactive compounds: Nutraceuticals, functional nutrients, and pharmaceuticals," in the date palm genome* (2021)
25. Farag M. A., Handoussa H., Fekry M. I., Wessjohann L. A.. **Metabolite profiling in 18 Saudi date palm fruit cultivars and their antioxidant potential via UPLC-qTOF-MS and multivariate data analyses**. *Food & Funct.* (2016) **7** 1077-1086. DOI: 10.1039/c5fo01570g
26. Farag M. A., Mohsen M., Heinke R., Wessjohann L. A.. **Metabolomic fingerprints of 21 date palm fruit varieties from Egypt using UPLC/PDA/ESI–qTOF-MS and GC–MS analyzed by chemometrics**. *Food Res. Int.* (2014) **64** 218-226. DOI: 10.1016/j.foodres.2014.06.021
27. Forli S., Huey R., Pique M. E., Sanner M. F., Goodsell D. S., Olson A. J.. **Computational protein–ligand docking and virtual drug screening with the AutoDock suite**. *Nat. Protoc.* (2016) **11** 905-919. DOI: 10.1038/nprot.2016.051
28. Habib H. M., El-Fakharany E. M., Souka U. D., Elsebaee F. M., El-Ziney M. G., Ibrahim W. H.. **Polyphenol-rich date palm fruit seed (phoenix dactylifera L.) extract inhibits labile iron, enzyme, and cancer cell activities, and DNA and protein damage**. *Nutrients* (2022) **14** 3536. DOI: 10.3390/nu14173536
29. Habib H., Platat C., Almaqbali F., Ibrahim W.. **Date seed (Phoenix dactylifera) extract reduces the proliferation of pancreatic cancer cells, DNA damage and superoxide-dependent iron release from ferritin**. *FASEB J.* (2014) **28** 829. DOI: 10.1096/fasebj.28.1_supplement.829.20
30. Hamad I., Abdelgawad H., Al Jaouni S., Zinta G., Asard H., Hassan S.. **Metabolic analysis of various date palm fruit (Phoenix dactylifera L.) cultivars from Saudi Arabia to assess their nutritional quality**. *Molecules* (2015) **20** 13620-13641. DOI: 10.3390/molecules200813620
31. Hilary S., Kizhakkayil J., Souka U., Al-Meqbaali F., Ibrahim W., Platat C.. *Front. Nutr.* (2021) **8** 667514. DOI: 10.3389/fnut.2021.667514
32. Ibrahim S. A., Ayad A. A., Williams L. L., Ayivi R. D., Gyawali R., Krastanov A.. **Date fruit: A review of the chemical and nutritional compounds, functional effects and food application in nutrition bars for athletes**. *Int. J. Food Sci. Technol.* (2021) **56** 1503-1513. DOI: 10.1111/ijfs.14783
33. Khan F., Ahmed F., Pushparaj P. N., Abuzenadah A., Kumosani T., Barbour E.. **Ajwa date (Phoenix dactylifera L.) extract inhibits human breast adenocarcinoma (MCF7) cells**. *PloS one 11* (2016) **11** e0158963. DOI: 10.1371/journal.pone.0158963
34. Khan M. A., Siddiqui S., Ahmad I., Singh R., Mishra D. P., Srivastava A. N.. **Phytochemicals from Ajwa dates pulp extract induce apoptosis in human triple-negative breast cancer by inhibiting AKT/mTOR pathway and modulating Bcl-2 family proteins**. *Sci. Rep.* (2021) **11** 10322-10414. DOI: 10.1038/s41598-021-89420-z
35. Khan M. A., Singh R., Siddiqui S., Ahmad I., Ahmad R., Upadhyay S.. **Anticancer potential of phoenix dactylifera L. Seed extract in human cancer cells and pro-apoptotic effects mediated through caspase-3 dependent pathway in human breast cancer MDA-MB-231 cells: An**. *BMC complementary Med. Ther.* (2022) **22** 68-19. DOI: 10.1186/s12906-022-03533-0
36. Khattak M. N. K., Shanableh A., Hussain M. I., Khan A. A., Abdulwahab M., Radeef W.. **Anticancer activities of selected Emirati Date (Phoenix dactylifera L.) varieties pits in human triple negative breast cancer MDA-MB-231 cells**. *Saudi J. Biol. Sci.* (2020) **27** 3390-3396. DOI: 10.1016/j.sjbs.2020.09.001
37. Khosravi A., Sharifi I., Tavakkoli H., Molaakbari E., Bahraminegad S., Salarkia E.. **Cytotoxicity of amphotericin B and AmBisome:**. *Front. Pharmacol.* (2022) **13** 1. DOI: 10.3389/fphar.2022.860598
38. Lamia F. S., Mukti R. F.. **Bangladeshi wild date palm fruits (Phoenix sylvestris): Promising source of anti-cancer agents for hepatocellular carcinoma treatment**. *Int. J. Appl. Sci. Biotechnol.* (2021) **9** 32-37. DOI: 10.3126/ijasbt.v9i1.36110
39. Laskowski R. A., Jabłońska J., Pravda L., Vařeková R. S., Thornton J. M.. **PDBsum: Structural summaries of PDB entries**. *Protein Sci.* (2018) **27** 129-134. DOI: 10.1002/pro.3289
40. Mansour R. B., Lassoued S., Dammak I., Elgaied A., Besbes S., Attia H.. **Cytotoxicity evaluation and antioxidant activity of date seed oil from ‘‘Deglet-nour Tunisian cultivar”(Phoenix dactylifera L.)**. *Nat. Prod.* (2011) **7** 16-20
41. Maqsood S., Adiamo O., Ahmad M., Mudgil P.. **Bioactive compounds from date fruit and seed as potential nutraceutical and functional food ingredients**. *Food Chem.* (2020) **308** 125522. DOI: 10.1016/j.foodchem.2019.125522
42. Mhm A. E.-A., El-Mesalamy A., Yassin F., Khalil S.. **Identification phenolic and biological activities of methanolic extract of date palm pollen (Phoenix dactylifera)**. *J. Microb. Biochem. Technol.* (2015) **7** 047-050. DOI: 10.4172/1948-5948.1000180
43. Mia M. a.-T., Mosaib M. G., Khalil M. I., Islam M. A., Gan S. H.. **Potentials and safety of date palm fruit against diabetes: A critical review**. *Foods* (2020) **9** 1557. DOI: 10.3390/foods9111557
44. Najm O. A., Addnan F. H., Mohd-Manzor N. F., Elkadi M. A., Abdullah W. O., Ismail A.. **Identification of phytochemicals of phoenix dactylifera L. Cv Ajwa with UHPLC-ESI-QTOF-MS/MS**. *Int. J. Fruit Sci.* (2021) **21** 848. DOI: 10.1080/15538362.2021.1939227
45. Nematallah K. A., Ayoub N. A., Abdelsattar E., Meselhy M. R., Elmazar M. M., El-Khatib A. H.. **Polyphenols LC-MS2 profile of Ajwa date fruit (Phoenix dactylifera L.) and their microemulsion: Potential impact on hepatic fibrosis**. *J. Funct. foods* (2018) **49** 401. DOI: 10.1016/j.jff.2018.08.032
46. Newman D. J., Cragg G. M.. **Natural products as sources of new drugs from 1981 to 2014**. *J. Nat. Prod.* (2016) **79** 629-661. DOI: 10.1021/acs.jnatprod.5b01055
47. Otify A. M., El-Sayed A. M., Michel C. G., Farag M. A.. **Metabolites profiling of date palm (phoenix dactylifera L.) commercial by-products (pits and pollen) in relation to its antioxidant effect: A multiplex approach of MS and nmr metabolomics**. *Metabolomics* (2019) **15** 119-217. DOI: 10.1007/s11306-019-1581-7
48. Perveen K., Bokahri N. A.. **Comparative analysis of chemical, mineral and**. *Saudi J. Biol. Sci.* (2020) **27** 1886-1891. DOI: 10.1016/j.sjbs.2019.11.029
49. Ravi L.. **Bioactivity of Phoenix dactylifera seed and its phytochemical analysis**. *Int. J. Green Pharm. (IJGP)* (2017) **11**. DOI: 10.22377/ijgp.v11i02.1037
50. Rezaei M., Khodaei F., Hooshmand N.. **Date seed extract diminished apoptosis event in human colorectal carcinoma cell line**. *MOJ Toxicol.* (2015) **1** 00017. DOI: 10.15406/mojt.2015.01.00017
51. Rock W., Rosenblat M., Borochov-Neori H., Volkova N., Judeinstein S., Elias M.. **Effects of date (phoenix dactylifera L., Medjool or Hallawi variety) consumption by healthy subjects on serum glucose and lipid levels and on serum oxidative status: A pilot study**. *J. Agric. food Chem.* (2009) **57** 8010-8017. DOI: 10.1021/jf901559a
52. Schüttelkopf A. W., Van Aalten D. M.. **Prodrg: A tool for high-throughput crystallography of protein–ligand complexes**. *Acta Crystallogr. Sect. D. Biol. Crystallogr.* (2004) **60** 1355-1363. DOI: 10.1107/s0907444904011679
53. Shahbaz K., Asif J. A., Liszen T., Nurul A. A., Alam M. K.. **Cytotoxic and antioxidant effects of phoenix dactylifera L. (Ajwa date extract) on oral squamous cell carcinoma cell line**. *BioMed Res. Int.* (2022) **2022** 1-16. DOI: 10.1155/2022/5792830
54. Sharma P., Joshi T., Mathpal S., Joshi T., Pundir H., Chandra S.. **Identification of natural inhibitors against Mpro of SARS-CoV-2 by molecular docking, molecular dynamics simulation, and MM/PBSA methods**. *J. Biomol. Struct. Dyn.* (2022) **40** 2757-2768. DOI: 10.1080/07391102.2020.1842806
55. Siddiqui S., Ahmad R., Khan M. A., Upadhyay S., Husain I., Srivastava A. N.. **Cytostatic and anti-tumor potential of Ajwa date pulp against human hepatocellular carcinoma HepG2 cells**. *Sci. Rep.* (2019) **9** 245. DOI: 10.1038/s41598-018-36475-0
56. Souda B., Rami R., Jalloul B., Mohamed D.. **Roasted date palm seeds (phoenix dactylifera) as an alternative coffee: Chemical composition and bioactive properties**. *Biomass Convers. Biorefinery* (2020) **12** 3771. DOI: 10.1007/s13399-020-00896-7
57. Taghizadeh M. S., Niazi A., Moghadam A., Afsharifar A.. **Experimental, molecular docking and molecular dynamic studies of natural products targeting overexpressed receptors in breast cancer**. *Plos one 17* (2022) **17** e0267961. DOI: 10.1371/journal.pone.0267961
58. Thouri A., La Barbera L., Canuti L., Vegliante R., Jelled A., Flamini G.. **Antiproliferative and apoptosis-inducing effect of common Tunisian date seed (var. Korkobbi and Arechti) phytochemical-rich methanolic extract**. *Environ. Sci. Pollut. Res.* (2019) **26** 36264-36273. DOI: 10.1007/s11356-019-06606-9
59. Tian W., Chen C., Lei X., Zhao J., Liang J.. **CASTp 3.0: Computed atlas of surface topography of proteins**. *Nucleic acids Res.* (2018) **46** W363-W367. DOI: 10.1093/nar/gky473
60. Trott O., Olson A. J.. **AutoDock vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading**. *J. Comput. Chem.* (2010) **31** 455-461. DOI: 10.1002/jcc.21334
61. Van Der Spoel D., Lindahl E., Hess B., Groenhof G., Mark A. E., Berendsen H. J.. **Gromacs: Fast, flexible, and free**. *J. Comput. Chem.* (2005) **26** 1701-1718. DOI: 10.1002/jcc.20291
62. Vayalil P. K.. **Antioxidant and antimutagenic properties of aqueous extract of date fruit (Phoenix dactylifera L. Arecaceae)**. *J. Agric. Food Chem.* (2002) **50** 610-617. DOI: 10.1021/jf010716t
63. Vayalil P. K.. **Date fruits (phoenix dactylifera linn): An emerging medicinal food**. *Crit. Rev. food Sci. Nutr.* (2012) **52** 249-271. DOI: 10.1080/10408398.2010.499824
64. Wang Y., Xiao J., Suzek T. O., Zhang J., Wang J., Bryant S. H.. **PubChem: A public information system for analyzing bioactivities of small molecules**. *Nucleic acids Res.* (2009) **37** W623-W633. DOI: 10.1093/nar/gkp456
65. Zhang C.-R., Aldosari S. A., Vidyasagar P. S., Nair K. M., Nair M. G.. **Antioxidant and anti-inflammatory assays confirm bioactive compounds in Ajwa date fruit**. *J. Agric. food Chem.* (2013) **61** 5834-5840. DOI: 10.1021/jf401371v
66. Zhang C.-R., Aldosari S. A., Vidyasagar P. S., Shukla P., Nair M. G.. **Health-benefits of date fruits produced in Saudi Arabia based on**. *J. Saudi Soc. Agric. Sci.* (2017) **16** 287-293. DOI: 10.1016/j.jssas.2015.09.004
|
---
title: Geographic clustering of travel-acquired infections in Ontario, Canada, 2008–2020
authors:
- Vinyas Harish
- Emmalin Buajitti
- Holly Burrows
- Joshua Posen
- Isaac I. Bogoch
- Antoine Corbeil
- Jonathan B. Gubbay
- Laura C. Rosella
- Shaun K. Morris
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022755
doi: 10.1371/journal.pgph.0001608
license: CC BY 4.0
---
# Geographic clustering of travel-acquired infections in Ontario, Canada, 2008–2020
## Abstract
As the frequency of international travel increases, more individuals are at risk of travel-acquired infections (TAIs). In this ecological study of over 170,000 unique tests from Public Health Ontario’s laboratory, we reviewed all laboratory-reported cases of malaria, dengue, chikungunya, and enteric fever in Ontario, Canada between 2008–2020 to identify high-resolution geographical clusters for potential targeted pre-travel prevention. Smoothed standardized incidence ratios (SIRs) and $95\%$ posterior credible intervals (CIs) were estimated using a spatial Bayesian hierarchical model. High- and low-incidence areas were described using data from the 2016 Census based on the home forward sortation area of patients testing positive. A second model was used to estimate the association between drivetime to the nearest travel clinic and incidence of TAI within high-incidence areas. There were 6,114 microbiologically confirmed TAIs across Ontario over the study period. There was spatial clustering of TAIs (Moran’s $I = 0.59$, $p \leq 0.0001$). Compared to low-incidence areas, high-incidence areas had higher proportions of immigrants ($p \leq 0.0001$), were lower income ($$p \leq 0.0027$$), had higher levels of university education ($p \leq 0.0001$), and less knowledge of English/French languages ($p \leq 0.0001$). In the high-incidence Greater Toronto Area (GTA), each minute increase in drive time to the closest travel clinic was associated with a $3\%$ reduction in TAI incidence ($95\%$ CI 1–$6\%$). While urban neighbourhoods in the GTA had the highest burden of TAIs, geographic proximity to a travel clinic in the GTA was not associated with an area-level incidence reduction in TAI. This suggests other barriers to seeking and adhering to pre-travel advice.
## Introduction
Prior to the COVID-19 pandemic, international travel was increasingly prevalent. In 2019, there were 1.5 billion international travellers, about half of whom travelled to low and lower middle-income countries [1, 2]. With increasing travel, more travellers are exposed to infectious agents not endemic to their departure region and more travel-acquired infections (TAI) occur[3, 4]. Pre-travel advice (PTA), which includes patient education on topics such as malaria prophylaxis, vaccination, food and water safety, and strategies for prevention of other vector-borne diseases, has been shown to decrease the rate of TAIs [5]. The World Health Organization recommends all travellers receive pre-travel advice, yet only 15–$54\%$ of travellers do so [4, 6–8]. Surveillance data indicates that upwards of $70\%$ of travellers who become ill due to travel did not receive PTA [9, 10].
Up to $76\%$ of travellers acquire a TAI [1, 11]. Most TAIs are mild and self-limited, such as traveller’s diarrhea. However, up to $8\%$ of travellers become ill enough to seek clinical care during or after travel, and up to $1\%$ of all travellers develop a febrile or systemic illness with elevated morbidity or mortality, such as malaria, enteric fever, dengue fever and chikungunya [9]. A quarter of all travellers report symptoms of TAI after completion of travel, signifying a significant burden of imported infections and healthcare needs [1]. During the 10-year period from 2009–2018, 4,947 imported cases of malaria and 1,536 imported cases of enteric fever were reported in Canada, about half of which were reported in Ontario [12, 13]. Among sub-national jurisdictions in North America, only New York State has a higher reported rate of malaria cases than Ontario [13, 14]. In addition to the morbidity caused by these TAIs, attributable medical costs in Ontario have been calculated to be CAD$4,558 per case of malaria and CAD$7,852 per case of enteric fever [15].
Previous studies have found that travellers visiting friends and relatives (VFR), business travellers, and those travelling with children, on short notice, or for extended durations are less likely to receive PTA [4, 16–19]. These same factors have been associated with increased incidence of TAI [9, 10, 20]. Quality of PTA also varies depending on the provider, with VFRs especially being less likely to consult a travel medicine specialist [6, 11, 19]. Thus, many of the highest incidence travellers are likely not receiving appropriate PTA, leading to missed opportunities for prevention of TAIs.
Ontario’s public health system is subdivided into 34 local public health units. Demographic factors and incidence of TAIs vary significantly between health units, with three health units–Toronto, Peel, and Ottawa–reporting almost $80\%$ of provincial malaria cases [12]. Due to the large population sizes covered by these health three units (one to three million each), it would be impractical to distribute TAI prevention efforts at the level of the public health unit. Geographical analysis by postal code for targeted interventions has yielded specific outcomes for non-communicable diseases [21, 22], but has not previously been described for TAIs in a Canadian context.
Our objective was to review all laboratory confirmed cases of four common TAIs (malaria, enteric fever, dengue fever, and chikungunya) in Ontario between 2008–2020 to identify high-resolution geographical clusters. Our secondary objective was to explore the association between geographic proximity to travel clinics and neighbourhood-level burden of TAIs. We hypothesized that there would be clustering of TAIs in urban centres and that proximity to travel clinics (measured by drive time) would be associated with reduced neighbourhood-level TAI burden.
## Ethics statement
Approval from The Hospital for Sick Children Research Ethics Board (REB) was obtained for this study (REB #1000068880).
## Study design and setting
Our study setting is Ontario, Canada’s most populous province with a population over 15 million as of 2022 [23]. Ontario has a universal single-payer health system that provides free access to a wide range of services including laboratory testing to the vast majority of residents. We reviewed all tests for malaria, enteric fever (caused by *Salmonella enterica* serovar Typhi or Paratyphi), dengue, and chikungunya processed at Public Health Ontario’s laboratory between July 15, 2008 to December 31, 2020. Public Health *Ontario is* Ontario’s provincial public health reference laboratory and conducts confirmatory testing for the listed TAIs across the province. Since residents were eligible for testing without additional cost, we can provide population-level estimates of disease burden robust to detection bias.
## Data source
PHO’s laboratory information system was queried for the time period of July 15, 2008 (date of implementation of the laboratory information system) to December 31, 2020. All test results for malaria (microscopy, rapid diagnostic tests [RDT], and polymerase chain reaction [PCR] tests), enteric fever (blood, bone marrow, urine, and stool cultures), dengue fever (IgM enzyme-linked immunosorbent assay [ELISA], IgG ELISA, and PCR tests), and chikungunya (IgM ELISA, IgG ELISA, and PCR tests) were extracted. Chikungunya coverage begins in 2015 while dengue coverage begins in 2008 with ELISA and PCR added in 2016. Each data point was collected at the test level and included the patient’s home forward sortation area (FSA) from their 6-digit postal code in addition to the specimen type, test performed, and test result. If the FSA was unavailable, the submitter’s FSA was provided instead. Each test was also assigned a unique patient identifier based on health card number, first name, last name, and date of birth.
## Outcome definition
The primary outcome of this study was the population-standardized incidence ratio of TAIs at the FSA level. The FSA is defined by the first three alphanumeric characters within a Canadian postal code and is a common level of geographic analysis. Each FSA comprises roughly 20,000 people but can vary in coverage due to heterogeneity in geography and population size (e.g., FSAs in northern Ontario are very large in area as the region is sparsely populated).
To determine an accurate count of unique TAI episodes, we identified a time period for attribution of positive tests to the same infection episode (as opposed to a reinfection episode). Repeat positive results outside this period were considered to represent a reinfection. For malaria, the defined period for persistently positive microscopy was set at 14 days, and at 90 days for RDT and PCR. The period was set at 14 days for enteric fever culture, dengue PCR, and chikungunya PCR. For dengue or chikungunya IgM ELISA, the period was set at 365 days. Due to small counts and similar incidence profiles (i.e., same vector and similar geographic distribution), dengue and chikungunya cases were pooled during analysis as ‘arboviruses’. The main analyses are presented with all diseases pooled together as TAIs, but disaggregated analyses can be found in the S1 Text.
## Exposures and covariate selection
The primary aim of this study was to identify any geographic clustering in TAI burden within Ontario. The secondary aim was to explore the association between geographic proximity to travel clinics and FSA-level burden of TAIs. In Canada, self-described travel clinics provide heterogeneous services from healthcare providers whose expertise in travel medicine varies considerably. To identify and include in our analysis only travel clinics that meet a minimum standard, we defined a travel clinic as a healthcare site designated by the Public Health Agency of Canada as a yellow fever vaccination centre [24]. All listed sites in Ontario were identified and geocoded with latitude and longitude coordinates using the Google Maps Geocoding application programming interface (API) [25]. Drive times were calculated from the centroid of each FSA polygon to each travel clinic using the Open Source Routing Machine API which determines the shortest path between a series of points on road networks [26]. This list of drive times was then filtered to obtain the drive time from each FSA to the closest travel clinic as a continuous variable in minutes.
Sociodemographic covariates of interest were defined a priori based on expert knowledge and obtained from Statistics Canada’s 2016 Census (https://www12.statcan.gc.ca/census-recensement/2016/dp-pd/index-eng.cfm) at the FSA level. Specifically, we included log-transformed median household income after tax as a continuous variable and immigration status, knowledge of official languages (English and French), ethnicity (Caribbean, African, Latin American, Middle Eastern, East and Southeast Asian, and South Asian), and education level (postsecondary certificate/diploma/degree) as proportions.
## Statistical analyses
A detailed and accessible overview of spatial modelling with Bayesian hierarchical models can be found elsewhere [27]. Counts of unique TAIs were used to calculate annual period-wide FSA-level case counts as well as population-standardized incidence ratios (SIRs). The expected number of cases was determined by multiplying the Ontario-wide rate of pooled cases by the population of each FSA. Smoothed SIRs and $95\%$ posterior credible intervals (CIs) were estimated with a Besag spatial Bayesian hierarchical model (BHM), which accounts for statistical instability and uncertainty in small area incidence and are widely used for small-area rate analyses [28, 29]. Due to overdispersion of case counts and large number of FSAs with zero cases, we used a zero-inflated negative binomial base model. The Moran’s I test was used to test for global spatial autocorrelation in BHM-smoothed SIRs (i.e., if the spatial distribution of values can be explained by random chance) [30].
Posterior CIs were used to identify high- and low- incidence areas of TAIs, which were described using sociodemographic data from the 2016 Canadian census. High-incidence areas were defined as those with smoothed SIR $95\%$ CIs greater than 1 (i.e., lower credible limit > 1), and low-incidence areas with smoothed SIR $95\%$ CIs less than 1 (i.e., upper credible limit < 1). Moderate-incidence areas are those with smoothed SIR $95\%$ CIs that cross 1. To test for significant differences in Census characteristics between groups, we used the Kruskall-Wallis and Wilcoxon rank sum tests. All statistical tests were two-sided and a p-value of <0.05 was considered significant. A second BHM was used to estimate the association between drive time to the nearest travel clinic and incidence of TAI within high-incidence areas, adjusted for potential confounders. We estimated the percent variance explained by our covariates by subtracting the variance of the posterior SIRs in the unadjusted model from the variance of the adjusted model, dividing the difference by the unadjusted variance, and then multiplying the quotient by 100 [31]. All analyses were done in the R programming language (R version 4.0.1, RStudio version 1.3.9, Boston, MA, USA).
## Results
Between July 15, 2008 and December 31, 2020, a total of 171,500 tests for malaria, enteric fever, dengue, and chikungunya were performed on 107,106 unique individuals. To avoid inappropriate clustering around health facilities, including those that serve large immigrant populations, we excluded 388 tests because of missing or invalid (i.e., outside of Ontario) patient home FSA. Of the 171,112 tests that were eligible, a total of 11,398 were positive for at least one of the four TAIs under study. Of these positive tests, 5,284 were excluded because they represented persistently positive repeat test results for the same TAI episode in the same patient. Therefore, the final analytic cohort included 6,114 laboratory confirmed unique TAI episodes (Fig A in S1 Text).
The annual absolute number of microbiologically confirmed enteric fever and malaria infections remained relatively stable at around 200 per year between 2010 and 2019 (Fig 1). There was greater variation in arbovirus infection numbers over the same time period, with annual numbers fluctuating between 100 and 280. Pooled annual TAI episodes peaked in 2019 at 715. The incidence of all TAIs dropped sharply in 2020, coinciding with the COVID-19 pandemic.
**Fig 1:** *Year-over-year burden of travel-acquired infections in Ontario, Canada.The arbovirus category consists of both dengue and chikungunya.*
BHM-smoothed TAI SIRs by FSA ranged from 0 to 8 across Ontario, with higher SIRs generally located in the Greater Toronto Area (GTA) (Fig 2A and 2B). There was spatial clustering of TAIs (Moran’s $I = 0.59$, $p \leq 0.0001$). The majority of FSAs deemed high-incidence for TAIs (i.e., those with SIR $95\%$ CIs greater than 1; $$n = 57$$) were located in the GTA ($$n = 53$$), with the majority of the remaining GTA FSAs considered moderate-incidence (Fig 2D). There were five FSAs with SIRs over 5, and three of those 5 corresponded to the locations of hospitals within downtown Toronto and Ottawa. The majority of FSAs outside the GTA had SIRs less than 1, with the exception of certain FSAs in Northern Ontario and the Greater Ottawa Area (Fig 2C and 2D).
**Fig 2:** *Standardized travel-acquired infection incidence across Ontario, Canada.Bayesian hierarchical model (BHM) smoothed standardized incidence ratios (SIRs) for travel-acquired infections (TAIs) and estimated incidence levels (panels A and C) with insets for the Greater Toronto Area (B and D). High-incidence areas are defined as those with smoothed SIR 95% credible intervals (Cis) greater than 1 and low-incidence areas with smoothed SIR 95% CIs less than 1. Adapted from Statistics Canada, 2016 Census–Boundary Files, 2019-11-13. This does not constitute an endorsement by Statistics Canada of this product.*
The absolute case counts in the GTA for all three disease categories—arboviruses, enteric fever, and malaria—generally followed the spatial pattern observed for the TAI SIRs (Fig 3). Enteric fever cases were the most clustered (Fig 3C) while arboviruses were the most dispersed (Fig 3A). Higher incidence FSAs had a significantly higher proportion of: immigrants; recent immigrants (migrated between 2001–2016); lower household after-tax income; university certificate or diploma above a Bachelors; and lower knowledge of Canada’s official languages English and French (Fig 4). In the GTA, each minute increase in drive time to the closest travel clinic was associated with a $3\%$ reduction in TAI incidence ($95\%$ CI 1–$6\%$) (Table 1). When comparing adjusted and unadjusted models, the Census covariates explained roughly $15\%$ of the variation in TAI incidence.
**Fig 3:** *Burden of travel-acquired infections in the Greater Toronto Area.Raw counts of microbiologically confirmed infections for arboviruses (panel A), malaria (panel B), enteric fever (panel C) and for all diseases (panel D) in the Greater Toronto Area over the study period. Adapted from Statistics Canada, 2016 Census–Boundary Files, 2019-11-13. This does not constitute an endorsement by Statistics Canada of this product.* **Fig 4:** *Characterizing high, moderate, and low-incidence clusters of travel-acquired illness burden across Ontario, Canada.Boxplots comparing clusters of FSAs across Ontario using key characteristics from the 2016 Census. Kruskall-Wallis and Wilcoxon rank sum tests were used as appropriate. All statistical tests were two-sided and a p-value of <0.05 was considered significant.* TABLE_PLACEHOLDER:Table 1
## Discussion
We reviewed all laboratory confirmed cases of four common TAIs—malaria, enteric fever, dengue fever, and chikungunya—in Ontario between 2008–2020 to identify high-resolution geographical clusters. There was spatial clustering of TAIs largely within the GTA, the largest urban centre in Canada. Compared to low-incidence areas, high-incidence areas had higher proportions of immigrants, lower income status, higher university education, and lower knowledge of English and French. Contrary to our hypothesis, each minute increase in drivetime to the closest travel clinic in the GTA was associated with a $3\%$ reduction in TAI incidence.
Previous work in Ontario has suggested that area-level measures of socioeconomic status are not fully representative as proxies for individual-level data, but that they can measure important drivers of health outcomes and inequities [32]. This is particularly important in the context of TAIs. Communities of immigrants may live within similar regions and may share comparable incidences of TAI because of similar travel destinations and/or behaviours associated with travel. We found that Census covariates explained only $15\%$ of the TAI incidence variation between adjusted and unadjusted models, thus other community- and individual-level factors are likely at play. Ethnicity was an important lens for understanding TAI clusters and provides important potential pathways for targeted education. In the GTA, enteric fever was highly clustered in predominantly South Asian communities whereas malaria was clustered in predominantly African communities (particularly Nigerian and Somali communities). These findings correlate with the high incidence rates of enteric fever in South Asia and of malaria in sub-Saharan Africa, likely due to community members recently immigrating from or travelling to these regions. This clustering of TAIs based on travel region may explain the significant spatial autocorrelation in TAIs observed here (Moran’s $I = 0.59$, $p \leq 0.0001$), which is higher than some of the other correlations reported in the literature such as heat-related illness (Moran’s $I = 0.21$) and opioid-related deaths (Moran’s $I = 0.46$) [33, 34]. The highest incidence neighbourhoods in this study are marginalized along multiple social determinants of health including low economic status, recent immigration, and lower knowledge of Canada’s official languages. Intersectional vulnerability in communities has been linked to a range of adverse health outcomes, from chronic conditions to COVID-19 [35, 36]. While it may appear counterintuitive that higher levels of university education were associated with TAI clusters, this could be because many immigrants were highly educated in their home countries prior to their arrival in Canada and that education was favoured in the Canadian points-based immigration system until the early 2010s [37]. However, this finding may be limited to our jurisdiction and we caution readers against making a similar argument in their jurisdiction if circumstances and evidence may differ. Geospatial analyses may create meaningful lines of inquiry, including by challenging conventional knowledge, that are well suited for further exploration by mixed-methods and qualitative research in high-incidence areas.
Access to health services has been conceptualized along three dimensions: physical accessibility, financial affordability, and acceptability [38]. Regarding physical accessibility, our drive time analysis demonstrates that clinics appear to be appropriately located near high-incidence communities. Financial affordability has been previously reported to be a barrier to seeking and following PTA among VFR travellers residing in the GTA [39]. Similar to the United States, Ontario’s universal health insurance program does not cover PTA, travel-related immunizations nor chemoprophylaxis [40]. Given this barrier and the potentially high healthcare costs linked to the management of TAIs, our findings highlight the need for cost effectiveness analyses evaluating the financial benefits of travel clinics on TAI prevention as well as pilot programs for PTA cost coverage in high-incidence communities [15]. The last dimension of access to health services, acceptability, may be the most complex barrier to address as it encompasses numerous dimensions of risk perception [41]. Studies have found that travellers may not accurately perceive risks associated with their travel and thus may believe PTA to be unnecessary for them [7, 19]. These beliefs are accentuated in VFR travellers. Uncertainty regarding itineraries and rushed circumstances surrounding travel (e.g., to attend a funeral) provide logistical challenges that hinder the feasibility of seeking PTA [39]. Since the tools to prevent these TAIs are different and vary in effectiveness, geospatial analyses, both aggregated and disaggregated by pathogen of interest, may serve as the first step towards guiding the local co-design of interventions with trusted community organizations to promote the acceptability of PTA in high-incidence communities [16, 39, 42, 43]. There can be significant ethno-cultural heterogeneity within a high-incidence community and thus clinicians and public health professionals may face the need to provide several distinct cultural tailorings of geographically targeted interventions (i.e., malaria prevention tailored to Nigerian residents, typhoid prevention tailored to South Asian residents). Research into cost-effective and culturally sensitive interventions to improve uptake of PTA could represent a fruitful area for future work.
Our study features numerous strengths and contributions. First, it leverages a comprehensive dataset from a universal single-payer health system. Since residents were eligible for testing without additional cost, we can provide population-level estimates of disease burden robust to detection bias. Moreover, Ontario has a combination of highly urban and rural communities and is among the most ethnically diverse populations in the world, both of which improve the generalizability of our work to high income jurisdictions with high levels of immigration and multicultural populations. Spatial analyses may be especially useful to plan interventions in major metropolitan areas, as similar ethno-cultural clustering of TAIs has been reported in VFRs living in New York City and London [44–47]. Large, representative studies of this patient population and timescale are currently rare, as many of the published studies are based on chart reviews from individual travel clinics, surveys of travellers in airports, or the International Society of Travel Medicine GeoSentinel Surveillance Network collaboration efforts [9, 16]. While GeoSentinel provides a wide-ranging survey of the burden of TAIs across countries, the relatively fewer number of participant sites per country prevents comprehensive studies within a given geospatial jurisdiction. To our knowledge, we are the first group to examine geographic access to travel clinics as a form of pre-event access prior to a health outcome of interest (TAI). Finally, our geospatial approach was rigorous as we used Bayesian modelling to smooth small case counts and computed drive times to measure distance to care as opposed to less accurate straight-line distances. Bayesian modelling also reduces bias due to the modifiable areal unit problem (MAUP), whereby aggregating data into geopolitical units to study trends is sensitive to the administrative boundaries creating these units [48].
Our study also has limitations. Although we expect to have excellent case identification, a limitation is that those without coverage in Ontario’s health insurance plan (e.g., newly landed immigrants/refugees) may be less likely to seek care and be identified with a TAI. Moreover, we did not have patient-level information on the location, nature, and duration of travel, as well as if PTA was individually sought or received. Lacking these details limits our ability to fully understand patterns and incidence groups. Our analysis does not consider differences in travel clinics (e.g., if a certified travel medicine specialist provides care, clinics that provide pre-travel advice without yellow fever certification) and changes in travel clinic locations over time. Unfortunately, no historical repository of travel clinic locations is publicly available. Our analysis also does not consider changes in community demographics over time since we only used data from the 2016 Census to gauge risk factors. As with all area-level studies, it is possible that our findings would be different based on different geographic boundaries due to the MAUP [49]. While smaller geographic boundaries may capture more variation, larger ones may be more relevant for planning interventions [50], reduce the risk of spillover effects [51], and are likely more temporally stable. Future work in Ontario should leverage mixed-methods designs to glean more nuanced, individual level information in the identified high-incidence communities. Jurisdictions with year-specific risk factors could also consider more advanced methods that handle time-varying confounding (e.g., G methods) to account for changing trends. Moreover, there may be misclassification bias due to the difficulty in determining if a positive test result corresponds to a new clinical episode or persistent infection. To mitigate this, we set time cut-offs between positive tests based on the expected persistence of a positive test result, acknowledging that this approach may still yield misclassified episodes. Finally, while PHO performs the majority of testing for TAIs in Ontario, it is possible that some individuals may have obtained testing by other laboratory facilities and would not be captured in this study. As PHO only performs bacterial stool cultures to support outbreaks or at the request of public health units, it is possible that our estimates undercount enteric fever as diagnosed by stool. Since arboviral infections are more likely to be mild and self-limited compared to malaria and enteric fever, it is possible our estimates undercount the burden of dengue and chikungunya.
## Conclusion
While urban neighbourhoods in the Greater Toronto Area had the highest burden of travel-acquired infections in Ontario, geographic proximity to a travel clinic was not associated with an area-level reduction in the incidence of infections. This suggests other barriers to seeking and/or adhering to pre-travel advice. Future research, policy measures, and community-based interventions should consider barriers to the affordability and acceptability of pre-travel advice to better understand and ultimately reduce the burden of travel-acquired infections.
## References
1. Vilkman K, Pakkanen SH, Lääveri T, Siikamäki H, Kantele A. **Travelers’ health problems and behavior: prospective study with post-travel follow-up.**. *BMC Infect Dis* (2016.0) **16** 328. DOI: 10.1186/s12879-016-1682-0
2. 2United Nations World Tourism Organization. UNWTO World Tourism Barometer and Statistical Annex, January 2020. UNWTO World Tourism Barometer (English version).
2020;18: 1–48. doi: 10.18111/wtobarometereng.2020.18.1.1. *UNWTO World Tourism Barometer and Statistical Annex, January 2020. UNWTO World Tourism Barometer (English version).* (2020.0) **18** 1-48. DOI: 10.18111/wtobarometereng.2020.18.1.1
3. Hill DR, Ericsson CD, Pearson RD, Keystone JS, Freedman DO, Kozarsky PE. **The Practice of Travel Medicine: Guidelines by the Infectious Diseases Society of America**. *Clinical Infectious Diseases* (2006.0) **43** 1499-1539. DOI: 10.1086/508782
4. Kain D, Findlater A, Lightfoot D, Maxim T, Kraemer MUG, Brady OJ. **Factors Affecting Pre-Travel Health Seeking Behaviour and Adherence to Pre-Travel Health Advice: A Systematic Review.**. *J Travel Med* (2019.0) 26. DOI: 10.1093/jtm/taz059
5. Reed JM, McIntosh IB, Powers K. **Travel Illness and the Family Practitioner: A Retrospective Assessment of Travel-Induced Illness in General Practice and the Effect of a Travel Illness Clinic.**. *J Travel Med.* (1994.0) **1** 192-198. DOI: 10.1111/j.1708-8305.1994.tb00594.x
6. Duval B, De Serre G, Shadmani R, Boulianne N, Pohani G, Naus M. **A population-based comparison between travelers who consulted travel clinics and those who did not.**. *J Travel Med* (2003.0) **10** 4-10. DOI: 10.2310/7060.2003.30659
7. LaRocque RC, Rao SR, Tsibris A, Lawton T, Anita Barry M, Marano N. **Pre‐travel Health Advice‐Seeking Behavior Among US International Travelers Departing From Boston Logan International Airport.**. *J Travel Med.* (2010.0) **17** 387-391. DOI: 10.1111/j.1708-8305.2010.00457.x
8. Hill DR. **The burden of illness in international travelers**. *N Engl J Med* (2006.0) **354** 115-117. DOI: 10.1056/NEJMp058292
9. Harvey K, Esposito DH, Han P, Kozarsky P, Freedman DO, Plier DA. **Surveillance for travel-related disease—GeoSentinel Surveillance System, United States, 1997–2011.**. *MMWR Surveill Summ* (2013.0) **62** 1-23
10. Boggild AK, Geduld J, Libman M, Yansouni CP, McCarthy AE, Hajek J. **Malaria in travellers returning or migrating to Canada: surveillance report from CanTravNet surveillance data, 2004–2014.**. *CMAJ Open* (2016.0) **4** E352-E358. DOI: 10.9778/cmajo.20150115
11. Hill DR. **Health problems in a large cohort of Americans traveling to developing countries.**. *J Travel Med* (2000.0) **7** 259-266. DOI: 10.2310/7060.2000.00075
12. Nelder MP, Russell C, Williams D, Johnson K, Li L, Baker SL. **Spatiotemporal dynamics and demographic profiles of imported Plasmodium falciparum and Plasmodium vivax infections in Ontario, Canada (1990–2009).**. *PLoS One* (2013.0) **8** e76208. DOI: 10.1371/journal.pone.0076208
13. 13Public Health Agency of Canada. Reported cases from 2009 to 2018 in Canada—notifiable diseases on-line. Ottawa; 2021.
14. Mace KE, Lucchi NW, Tan KR. **Malaria Surveillance—United States, 2017.**. *MMWR Surveill Summ* (2021.0) **70** 1-35. DOI: 10.15585/mmwr.ss7002a1
15. Savage RD, Rosella LC, Crowcroft NS, Horn M, Khan K, Holder L. **Direct Medical Costs of 3 Reportable Travel-Related Infections in Ontario, Canada, 2012–2014**. *Emerg Infect Dis* (2019.0) **25** 1501-1510. DOI: 10.3201/eid2508.190222
16. Hagmann S, Benavides V, Neugebauer R, Purswani M. **Travel health care for immigrant children visiting friends and relatives abroad: retrospective analysis of a hospital-based travel health service in a US urban underserved area.**. *J Travel Med* (2009.0) **16** 407-412. DOI: 10.1111/j.1708-8305.2009.00357.x
17. Angell SY, Behrens RH. **Risk assessment and disease prevention in travelers visiting friends and relatives.**. *Infect Dis Clin North Am* (2005.0) **19** 49-65. DOI: 10.1016/j.idc.2004.11.001
18. Hendel-Paterson B, Swanson SJ. **Pediatric travelers visiting friends and relatives (VFR) abroad: Illnesses, barriers and pre-travel recommendations.**. *Travel Med Infect Dis* (2011.0) **9** 192-203. DOI: 10.1016/j.tmaid.2010.09.007
19. Hamer DH, Connor BA. **Travel health knowledge, attitudes and practices among United States travelers.**. *J Travel Med* (2004.0) **11** 23-26. DOI: 10.2310/7060.2004.13577
20. Leder K, Tong S, Weld L, Kain KC, Wilder-Smith A, von Sonnenburg F. **Illness in travelers visiting friends and relatives: a review of the GeoSentinel Surveillance Network**. *Clin Infect Dis* (2006.0) **43** 1185-1193. DOI: 10.1086/507893
21. Creatore MI, Glazier RH, Moineddin R, Fazli GS, Johns A, Gozdyra P. **Association of Neighborhood Walkability With Change in Overweight, Obesity, and Diabetes**. *JAMA* (2016.0) **315** 2211-2220. DOI: 10.1001/jama.2016.5898
22. Kersten EE, Adler NE, Gottlieb L, Jutte DP, Robinson S, Roundfield K. **Neighborhood Child Opportunity and Individual-Level Pediatric Acute Care Use and Diagnoses**. *Pediatrics* (2018.0) **141** e20172309. DOI: 10.1542/peds.2017-2309
23. 23Ontario Ministry of Finance. Ontario Demographic Quarterly: Highlights of first quarter. In: https://www.ontario.ca/page/ontario-demographic-quarterly-highlights-first-quarter#:~:text=Ontario’s%20population%20reached%2015%2C007%2C816%20on,the%20same%20quarter%20of%202021.&text=Emigrants%20plus%20net%20change%20in,Source%3A%20Statistics%20Canada. 2022.
24. 24Public Health Agency of Canada. Yellow Fever Vaccination Centres in Ontario. In: Travel Health [Internet]. 2022 [cited 5 May 2022]. Available: https://www.canada.ca/en/public-health/services/travel-health/yellow-fever/vaccination-centres-canada-ontario.html
25. 25Google Maps Platform. Geocoding API Overview. In: Documentation [Internet]. 2022 [cited 5 May 2022]. Available: https://developers.google.com/maps/documentation/geocoding/overview
26. Luxen D, Vetter C. *Proceedings of the 19th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems.* (2011.0) 513-516. DOI: 10.1145/2093973.2094062
27. Moraga P.. *Geospatial Health Data Modeling and Visualization with R-INLA and Shiny* (2020.0)
28. Lawson AB. *Bayesian Disease Mapping: Hierarchical Modeling in Spatial Epidemiology* (2018.0)
29. Rue H, Martino S, Chopin N. **Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations.**. *J R Stat Soc Series B Stat Methodol* (2009.0) **71** 319-392. DOI: 10.1111/j.1467-9868.2008.00700.x
30. MORAN PAP. **NOTES ON CONTINUOUS STOCHASTIC PHENOMENA.**. *Biometrika* (1950.0) **37** 17-23. DOI: 10.1093/biomet/37.1–2.17
31. Buajitti E, Watson T, Norwood T, Kornas K, Bornbaum C, Henry D. **Regional variation of premature mortality in Ontario, Canada: a spatial analysis**. *Popul Health Metr* (2019.0) **17** 9. DOI: 10.1186/s12963-019-0193-9
32. Buajitti E, Chiodo S, Rosella LC. **Agreement between area- and individual-level income measures in a population-based cohort: Implications for population health research.**. *SSM Popul Health* (2020.0) **10** 100553. DOI: 10.1016/j.ssmph.2020.100553
33. Lehnert EA, Wilt G, Flanagan B, Hallisey E. **Spatial exploration of the CDC’s Social Vulnerability Index and heat-related health outcomes in Georgia**. *International Journal of Disaster Risk Reduction* (2020.0) **46** 101517. DOI: 10.1016/j.ijdrr.2020.101517
34. Jones CW, Christman Z, Smith CM, Safferman MR, Salzman M, Baston K. **Comparison between buprenorphine provider availability and opioid deaths among US counties.**. *J Subst Abuse Treat* (2018.0) **93** 19-25. DOI: 10.1016/j.jsat.2018.07.008
35. Moin JS, Moineddin R, Upshur REG. **Measuring the association between marginalization and multimorbidity in Ontario, Canada: A cross-sectional study.**. *J Comorb.* (2018.0) **8**. DOI: 10.1177/2235042X18814939
36. Sundaram ME, Calzavara A, Mishra S, Kustra R, Chan AK, Hamilton MA. **Individual and social determinants of SARS-CoV-2 testing and positivity in Ontario, Canada: a population-wide study.**. *Can Med Assoc J* (2021.0) **193** E723. DOI: 10.1503/cmaj.202608
37. Hou F, Lu Y. *Over-education Among University-educated Immigrants in Canada and the United States* (2019.0)
38. Evans DB, Hsu J, Boerma T. **Universal health coverage and universal access**. *Bull World Health Organ* (2013.0) **91** 546-546A. DOI: 10.2471/BLT.13.125450
39. Savage RD, Rosella LC, Crowcroft NS, Arneja J, de Villa E, Horn M. **How Can We Keep Immigrant Travelers Healthy? Health Challenges Experienced by Canadian South Asian Travelers Visiting Friends and Relatives.**. *Qual Health Res* (2017.0) **28** 610-623. DOI: 10.1177/1049732317746381
40. Hughes IV R, Klein Z. **Assessing US traveller vaccination access: an evaluation of US requirements for healthcare payer coverage of recommended travel vaccines.**. *J Travel Med.* (2020.0) **27**. DOI: 10.1093/jtm/taaa118
41. Leggat PA, Franklin R. **Risk Perception and Travelers.**. *J Travel Med.* (2013.0) **20** 1-2. DOI: 10.1111/j.1708-8305.2012.00663.x
42. Seale H, Kaur R, Mahimbo A, MacIntyre CR, Zwar N, Smith M. **Improving the uptake of pre-travel health advice amongst migrant Australians: exploring the attitudes of primary care providers and migrant community groups.**. *BMC Infect Dis* (2016.0) **16** 213. DOI: 10.1186/s12879-016-1479-1
43. Heywood AE, Zwar N. **Improving access and provision of pre-travel healthcare for travellers visiting friends and relatives: a review of the evidence†.**. *J Travel Med.* (2018.0) **25** tay010. DOI: 10.1093/jtm/tay010
44. Adamson R, Reddy V, Jones L, Antwi M, Bregman B, Weiss D. **Epidemiology and Burden of Hepatitis A, Malaria, and Typhoid in New York City Associated With Travel: Implications for Public Health Policy.**. *Am J Public Health* (2010.0) **100** 1249-1252. DOI: 10.2105/AJPH.2009.178335
45. Smith AD, Bradley DJ, Smith V, Blaze M, Behrens RH, Chiodini PL. **Imported malaria and high risk groups: observational study using UK surveillance data 1987–2006**. *BMJ* (2008.0) **337** a120. DOI: 10.1136/bmj.a120
46. Reddy S, Rangaiah J, Addiman S, Wareham D, Wilson P, Sefton A. **Epidemiology, antibiotic resistance trends and the cost of enteric fever in East London, 2005–2010.**. *Travel Med Infect Dis* (2011.0) **9** 206-212. DOI: 10.1016/j.tmaid.2011.03.003
47. Morgan M, Figueroa-Muñoz JI. **Barriers to Uptake and Adherence with Malaria Prophylaxis by the African Community in London, England: Focus Group Study.**. *Ethn Health.* (2005.0) **10** 355-372. DOI: 10.1080/13557850500242035
48. Xu P, Huang H, Dong N, Abdel-Aty M. **Sensitivity analysis in the context of regional safety modeling: Identifying and assessing the modifiable areal unit problem.**. *Accid Anal Prev* (2014.0) **70** 110-120. DOI: 10.1016/j.aap.2014.02.012
49. Kwan M-P. **The Limits of the Neighborhood Effect: Contextual Uncertainties in Geographic, Environmental Health, and Social Science Research**. *Ann Am Assoc Geogr* (2018.0) **108** 1482-1490. DOI: 10.1080/24694452.2018.1453777
50. Séguin A-M, Apparicio P, Riva M. **The Impact of Geographical Scale in Identifying Areas as Possible Sites for Area-Based Interventions to Tackle Poverty: The Case of Montréal.**. *Appl Spat Anal Policy.* (2012.0) **5** 231-251. DOI: 10.1007/s12061-011-9068-6
51. Chaix B, Duncan D, Vallée J, Vernez-Moudon A, Benmarhnia T, Kestens Y. **The “Residential” Effect Fallacy in Neighborhood and Health Studies: Formal Definition, Empirical Identification, and Correction.**. *Epidemiology* (2017.0) **28**
|
---
title: Marker for kidney fibrosis is associated with inflammation and deterioration
of kidney function in people with type 2 diabetes and microalbuminuria
authors:
- Christina Gjerlev Poulsen
- Daniel G. K. Rasmussen
- Federica Genovese
- Tine W. Hansen
- Signe Holm Nielsen
- Henrik Reinhard
- Bernt Johan von Scholten
- Peter K. Jacobsen
- Hans-Henrik Parving
- Morten Asser Karsdal
- Peter Rossing
- Marie Frimodt-Møller
journal: PLOS ONE
year: 2023
pmcid: PMC10022760
doi: 10.1371/journal.pone.0283296
license: CC BY 4.0
---
# Marker for kidney fibrosis is associated with inflammation and deterioration of kidney function in people with type 2 diabetes and microalbuminuria
## Abstract
### Background
Diabetic kidney disease is a major cause of morbidity and mortality. Dysregulated turnover of collagen type III is associated with development of kidney fibrosis. We investigated whether a degradation product of collagen type III (C3M) was a risk marker for progression of chronic kidney disease (CKD), occurrence of cardiovascular disease (CVD), and mortality during follow up in people with type 2 diabetes (T2D) and microalbuminuria. Moreover, we investigated whether C3M was correlated with markers of inflammation and endothelial dysfunction at baseline.
### Methods
C3M was measured in serum (sC3M) and urine (uC3M) in 200 participants with T2D and microalbuminuria included in an observational, prospective study at Steno Diabetes Center Copenhagen in Denmark from 2007–2008. Baseline measurements included 12 markers of inflammation and endothelial dysfunction. The endpoints were CVD, mortality, and CKD progression (>$30\%$ decline in eGFR).
### Results
Mean (SD) age was 59 [9] years, eGFR 90 [17] ml/min/1.73m2 and median (IQR) urine albumin excretion rate 102 (39–229) mg/24-h. At baseline all markers for inflammation were positively correlated with sC3M (p≤0.034). Some, but not all, markers for endothelial dysfunction were correlated with C3M. Median follow-up ranged from 4.9 to 6.3 years. Higher sC3M was associated with CKD progression (with mortality as competing risk) with a hazard ratio (per doubling) of 2.98 ($95\%$ CI: 1.41–6.26; $$p \leq 0.004$$) adjusted for traditional risk factors. uC3M was not associated with CKD progression. Neither sC3M or uC3M were associated with risk of CVD or mortality.
### Conclusions
Higher sC3M was a risk factor for chronic kidney disease progression and was correlated with markers of inflammation.
## Introduction
Diabetic kidney disease (DKD) is a major cause of morbidity and mortality in diabetes. It is the primary cause of end stage kidney disease (ESKD) in western countries and causes up to half of incident cases [1]. However, the majority never reach ESKD because they are more likely to die of cardiovascular disease (CVD). The risk of CVD increases almost exponentially as kidney function declines [2–4].
Regardless of the etiology, a main feature of the progression of chronic kidney disease (CKD) is a pathological deposition of extracellular matrix components, which can trigger renal fibrosis and lead to ESKD [5]. The main structural component of the fibrotic core is collagen and one of the most prominent collagens in the fibrotic kidney is collagen type III. C3M is a degradation product of collagen type III, produced by the matrix metalloproteinase (MMP)-9. C3M thereby reflects turnover of collagen type III in the interstitial matrix and can be considered as a marker for fibrotic activity [6]. Studies have shown increased MMP-9 activity in DKD [7], and increased levels of MMP-9 in plasma were a risk factor for development of microalbuminuria in persons with type 2 diabetes (T2D) [8]. Increased levels of C3M measured in urine has been associated with severity of CKD in persons with type 1 diabetes (T1D) [9], and with both severity and progression of the disease in other CKD cohorts [6, 10]. C3M has not yet been investigated in people with type 2 diabetes and diabetic kidney disease.
Endothelial dysfunction and inflammation play an important role in the onset and progression of fibrosis. In previously reported data in this study population, markers for endothelial dysfunction and inflammation were independently associated with CVD and all-cause mortality [11].
A kidney biopsy is the only current method for detecting renal fibrosis. As fibrosis may be present before clinically detectable kidney disease, fibrotic biomarkers could potentially be used as a non-invasive method for much earlier detection of disease. Additionally, fibrotic biomarkers could be used for disease monitoring and assessment of treatment response.
In this study, we investigated whether C3M, measured in serum and urine, was associated with markers of inflammation and endothelial dysfunction at baseline, and whether it was a risk marker for progression of chronic kidney disease, occurrence of CVD events, and mortality during follow-up in people with T2D and microalbuminuria.
## Participants and study procedures
In 2007–2008 we recruited 200 persons with T2D from the outpatient clinic at Steno Diabetes Center Copenhagen, Denmark to a prospective, observational follow-up study. The enrolment criteria, as previously described [12, 13], included: a diagnosis of T2D according to the WHO criteria, persistent urine albumin excretion rate (UAER) >30 mg/24h (in two out of three consecutive measurements), normal kidney function and no coronary heart disease. All participants received intensive multifactorial treatment consisting of glycemic, lipid and blood pressure control, antithrombotic therapy, and lifestyle intervention according to the Steno 2 study [14]. The study was approved by the local ethics committee (De Videnskabsetiske Komitéer Region Hovedstaden) and complied with the Declaration of Helsinki. All participants provided written informed consent.
## Measurement of C3M in serum and urine
The biomarker C3M was measured in serum (sC3M) and urine (uC3M) from samples collected at baseline and were available in 198 and 190 of the 200 participants, respectively. Samples were stored at -80°C until analyses. Both serum and urine C3M were measured using competitive enzyme-linked immunosorbent assays (ELISAs) developed by Nordic Bioscience, Denmark [10, 15]. MMP-9 mediated degradation of collagen type III, produces a 10 amino acid neo-epitope (610’. KNGETGPQGP’619) (Fig 1). ELISAs were performed using two different monoclonal antibodies for detecting C3M in serum and urine, both antibodies specifically detected all fragments entailing the neo-epitope [15]. Intra- and inter-assay variations of the ELISAs were below 10 and $15\%$, respectively. The assays were carried out as previously described [10, 15]. To normalize for urine output, uC3M levels were divided by urinary creatinine levels measured on ADVIA 1800 Chemistry System (Siemens Healthineers, Germany).
**Fig 1:** *Generation and detection of C3M fragments.C3M is generated by degradation of collagen type III by the matrix metalloproteinase (MMP)-9, which produces a 10 amino acid neo-epitope (KNGETGPQGP). The ELISA antibody detects this neo-epitope and can thereby detect fragments of down to 10 amino acids. ELISA: enzyme-linked immunosorbent assay.*
## Baseline clinical and laboratory measures
UAER was measured in three 24h urine collections by an enzyme immunoassay (Vitros, Raritan, NJ, USA). HbA1c, plasma creatinine and serum cholesterol were determined using standard methods. The estimated glomerular filtration rate (eGFR) was calculated applying the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [16]. Current smoking was defined as one or more cigarettes, cigars, or pipes per day. Brachial blood pressure was the average of two consecutive measurements after 10 minutes of rest.
Markers of inflammation (TNF-α, sICAM-1, sICAM-3, hsCRP, SAA, IL-1beta, IL-6 and IL-8) and endothelial dysfunction (thrombomodulin, sVCAM-1, sE-selectin and sP-selectin) were measured at baseline using MSD multipanel measurements or ELISA as previously described [17].
## Follow-up
Approximately annual measurements of plasma creatinine were performed in 175 of the 198 participants ($88.4\%$) as part of their regular diabetes control at Steno Diabetes Center Copenhagen. The exact time from baseline to each creatinine measurement varied highly between each individual and each measurement. Information on participants with available samples was traced through the Danish National Death and Danish National Health Registries on January 1, 2014. No participants were lost to follow-up. Definitions of the three predefined endpoints have previously been described [13, 18]. The combined cardiovascular endpoint included cardiovascular mortality, non-fatal myocardial infarction, stroke, ischemic CVD, and heart failure. In case of multiple events in one participant, only the first was included. Chronic kidney disease progression was defined as a >$30\%$ decline in eGFR based on approximately annual plasma creatinine measurements (five in total) and evaluated as change from baseline to the last available measurement.
## Statistical analyses
Normally distributed variables are expressed as mean values ± standard deviations (SD). Both sC3M and uC3M had skewed distributions (S1 Fig, panel A and B) and were log2-transformed in all analyses and presented as medians with interquartile range (IQR). Categorical variables are presented as total numbers with corresponding percentages. Clinical characteristics were stratified by tertiles of the concentrations of sC3M and compared using one-way ANOVA for normally distributed variables, Kruskal-Wallis test for skewed distributed variables and the χ2-test for categorical variables.
Cox proportional hazards analysis was used to calculate hazard ratios (HR) with $95\%$ confidence intervals (CI) per doubling of sC3M and uC3M for the three endpoints and with mortality as competing risk in the analysis of eGFR decline. Adjustment for all models included traditional cardiovascular risk factors including sex, age, body mass index, LDL cholesterol, smoking, HbA1c, plasma creatinine, systolic blood pressure and UAER.
Cox proportional hazard analysis was applied to compare the risk of eGFR decline according to tertiles of sC3M and uC3M, adjusted for traditional cardiovascular risk factors as described.
Linear mixed effect models were applied to examine the association between sC3M/uC3M and eGFR decline during follow-up (including the five follow up measurements). The model was applied using the “gls function” with compound symmetry covariance structure in the “nlme package” using R version 4.1.0.
To evaluate the added predictive value for the markers of inflammation over the value of sC3M, we employed receiver operating characteristic (ROC) curves, applying C-statistics for area under curve (AUC) analysis. For reasons of statistical efficiency, we calculated a z-score for the markers of inflammation by averaging the individual biomarkers signed z-scores. The base model included the traditional cardiovascular risk factors and sC3M, and in the second model we added the z-score for the markers of inflammation.
A two-sided P-value <0.05 was considered statistically significant. Statistical analyses were performed using the SAS software (version 9.4, SAS Institute, Cary, NC, USA).
## Baseline characteristics
All participants were Caucasian, 149 ($75\%$) were male, mean (SD) age was 58.6 (8.7) years, diabetes duration 12.7 (7.4) years, eGFR was 90 [17] ml/min/1.73m2, and median (IQR) UAER was 102 (39–229) mg/24h. The median (IQR) concentration of sC3M was 8.4 (6.9–10.3) ng/ml and uC3M was 6.09 (4.52–7.68) ng/mol. Most were treated with oral antidiabetic medications ($85\%$) and nearly all received antihypertensive therapy ($99\%$) and statins ($94\%$). Baseline characteristics of both the total population and in tertiles of sC3M are presented in Table 1. There were no significant differences between the tertiles for any of the variables. However, there was a trend towards more women in the highest tertile. Baseline characteristics for tertiles of uC3M are presented in S1 Table. Individuals in the highest tertile of uC3M are more likely to be female, be younger of age, have a shorter diabetes duration, a higher HbA1c and a higher eGFR at baseline.
**Table 1**
| Characteristic | All | Tertile 1 | Tertile 2 | Tertile 3 | P-value |
| --- | --- | --- | --- | --- | --- |
| Characteristic | n = 198 | n = 61 | n = 68 | n = 69 | P-value |
| sC3M (ng/ml) | 8.4 (6.9–10.3) | 6.3 (5.8–6.8) | 8.3 (7.8–8.8) | 11.4 (10.5–13.5) | |
| Male, n (%) | 149 (75) | 49 (80) | 55 (81) | 45 (65) | 0.06 |
| Age (years) | 58.6 ± 8.7 | 59.3 ± 7.7 | 59.0 ± 8.3 | 57.7 ± 10.0 | 0.30 |
| Known duration of diabetes (years) | 12.7 ± 7.4 | 14.2 ± 6.8 | 11.3 ± 6.7 | 12.8 ± 8.3 | 0.30 |
| Body mass index (kg/m2) | 32.5 ± 5.8 | 32.6 ± 5.5 | 31.8 ± 5.5 | 33.1 ± 6.2 | 0.60 |
| HbA1c (%) | 7.9 ± 1.3 | 7.9 ± 1.4 | 7.6 ± 1.2 | 8.0 ± 1.5 | 0.72 |
| HbA1c (mmol/mol) | 62.3 ± 14.8 | 63.1 ± 14.9 | 60.0 ± 12.8 | 63.9 ± 16.3 | 0.72 |
| Urinary albumin excretion rate (mg/24-h) | 102 (39–229) | 99 (37–191) | 122 (47–240) | 81 (40–361) | 0.48 |
| P-creatinine (μmol/L) | 76.4 ± 18.4 | 76.5 ± 20.4 | 75.0 ± 16.7 | 77.6 ± 18.2 | 0.72 |
| eGFR (ml/min/1.73m 2 ) | 90 ± 17 | 90 ± 17 | 91 ± 16 | 88 ± 19 | 0.43 |
| LDL cholesterol (mmol/L) | 1.9 ± 0.8 | 1.8 ± 0.7 | 1.8 ± 0.8 | 2.0 ± 0.9 | 0.29 |
| Systolic blood pressure (mmHg) | 130 ± 17 | 133 ± 18 | 128 ± 16 | 130 ± 17 | 0.53 |
| Diastolic blood pressure (mmHg) | 75 ± 9 | 76 ± 12 | 74 ± 10 | 75 ± 11 | 0.99 |
| Current smoker, n (%) | 59 (30) | 21 (34) | 16 (24) | 22 (32) | 0.36 |
| Treatment with | | | | | |
| Oral antidiabetic, n (%) | 169 (85) | 55 (90) | 59 (87) | 55 (80) | 0.22 |
| Insulin, n (%) | 122 (62) | 38 (62) | 42 (62) | 42 (61) | 0.99 |
| Antihypertensive drugs, n (%) | 196 (99) | 61 (100) | 68 (100) | 67 (97) | 0.15 |
| RAAS blockade, n (%) | 194 (98) | 61 (100) | 67 (99) | 66 (96) | 0.20 |
| Statin, n (%) | 187 (94) | 56 (92) | 67 (99) | 64 (93) | 0.19 |
| Aspirin, n (%) | 182 (92) | 56 (92) | 64 (94) | 62 (90) | 0.66 |
There was no correlation between sC3M and uC3M (R2 = 0.002; $$p \leq 0.56$$; S1 Fig, panel C) or between sC3M and either eGFR or UAER (p≥ 0.37; S1 Fig, panel D and E). uC3M was positively correlated with eGFR (R2 = 0.09; $p \leq 0.001$), but not correlated with UAER (R2 = 0.015; $$p \leq 0.09$$).
## Baseline correlations of C3M and markers of inflammation and endothelial dysfunction
As shown in Table 2, all markers of inflammation were positively correlated with sC3M both in unadjusted and adjusted analyses (p ≤ 0.034). uC3M was positively correlated with IL-6 in the adjusted analyses ($$p \leq 0.040$$), but no other correlations between uC3M and markers of inflammation were demonstrated. Thrombomodulin and sVCAM-1 were positively correlated with sC3M in unadjusted analyses, and the correlation for sVCAM-1 remained after adjustment ($p \leq 0.001$). Thrombomodulin was negatively correlated with uC3M in unadjusted analyses ($P \leq 0.001$), but not after adjustment. sE-selectin was positively correlated with uC3M both in unadjusted and adjusted analyses (p ≤ 0.048). The other markers of endothelial dysfunction were not correlated with neither sC3M nor uC3M.
**Table 2**
| Unnamed: 0 | Serum C3M | Serum C3M.1 | Serum C3M.2 | Serum C3M.3 | Urine C3M | Urine C3M.1 | Urine C3M.2 | Urine C3M.3 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | Unadjusted | Unadjusted | Adjusted | Adjusted | Unadjusted | Unadjusted | Adjusted | Adjusted |
| | β (95%CI) | P-value | β (95%CI) | P-value | β (95%CI) | P-value | β (95%CI) | P-value |
| Inflammation markers | | | | | | | | |
| TNF-alfa | 0.15 (0.09–0.21) | <0.001 | 0.14 (0.07–0.21) | <0.001 | -0.08 (-0.18–0.03) | 0.15 | -0.04 (-0.14–0.07) | 0.46 |
| sICAM-1* | 0.15 (0.09–0.21) | <0.001 | 0.14 (0.08–0.21) | <0.001 | 0.07 (-0.04–0.17) | 0.20 | 0.03 (-0.07–0.13) | 0.53 |
| sICAM-3* | 0.08 (0.02–0.14) | 0.008 | 0.07 (0.01–0.13) | 0.034 | 0.07 (-0.04–0.17) | 0.22 | 0.06 (-0.04–0.16) | 0.23 |
| hsCRP | 0.18 (0.13–0.24) | <0.001 | 0.20 (0.14–0.71) | <0.001 | 0.07 (-0.03–0.18) | 0.15 | 0.06 (-0.04–0.16) | 0.25 |
| SAA | 0.15 (0.09–0.21) | <0.001 | 0.15 (0.09–0.21) | <0.001 | 0.05 (-0.05–0.15) | 0.33 | 0.05 (-0.05–0.14) | 0.36 |
| IL-1beta | 0.09 (0.03–0.15) | 0.005 | 0.10 (0.02–0.15) | 0.007 | 0.02 (-0.09–0.13) | 0.74 | 0.04 (-0.05–0.14) | 0.38 |
| IL-6 | 0.13 (0.07–0.19) | <0.001 | 0.14 (0.07–0.20) | <0.001 | 0.05 (-0.06–0.15) | 0.39 | 0.10 (0.01–0.20) | 0.040 |
| IL-8 | 0.11 (0.04–0.17) | <0.001 | 0.11 (0.04–0.17) | 0.013 | -0.04 (-0.14–0.06) | 0.44 | -0.009 (-0.10–0.09) | 0.84 |
| Markers of endothelial dysfunction | Markers of endothelial dysfunction | | | | | | | |
| Thrombomodulin | 0.07 (0.01–0.14) | 0.019 | 0.07 (-0.01–0.14) | 0.074 | -0.17 (-0.27- -0.07) | <0.001 | -0.10 (-0.21–0.01) | 0.078 |
| sVCAM-1 | 0.15 (0.09–0.21) | <0.001 | 0.15 (0.08–0.21) | <0.001 | 0.04 (-0.06–0.15) | 0.41 | 0.03 (-0.05–0.16) | 0.53 |
| sE-selectin | -0.01 (-0.07–0.05) | 0.71 | -0.02 (-0.08–0.06) | 0.74 | 0.10 (0.01–0.21) | 0.048 | 0.10 (0.01–0.20) | 0.040 |
| sP-selectin | -0.01 (-0.08–0.04) | 0.49 | -0.02 (-0.09–0.06) | 0.64 | 0.01(-0.09–0.11) | 0.84 | 0.014 (-0.09–0.11) | 0.79 |
## Incidence of cardiovascular events, mortality and decline in eGFR
The median (IQR) follow-up time in years was 6.1 (5.9–6.6) for CVD events, 6.3 (6.0–6.7) for mortality, and 4.6 (3.8–5.1) for eGFR decline. During the follow-up period 40 participants reached the combined CVD endpoint, 26 died and 42 declined >$30\%$ in eGFR from baseline. The combined CVD endpoint included 11 fatal CVD events (two events of acute myocardial infarction, one case of ischemic CVD, six sudden and otherwise unexplained events and two events of heart failure) and 29 non-fatal CVD events (three cases of acute myocardial infarction, three strokes, 19 cases of ischemic CVD and four cases of heart failure). Out of the 26 participants who died, 11 were due to CVD, nine related to cancer and six related to other causes. No participants were diagnosed with kidney failure during follow-up.
## C3M as risk marker
Table 3 shows the associations between sC3M, uC3M and the three endpoints. High sC3M was associated with an increased risk of a >$30\%$ decline in eGFR in the unadjusted and adjusted analyses, including mortality as competing risk. Serum C3M was not a risk marker for CVD events or all-cause mortality, while uC3M was not associated with any of the endpoints.
**Table 3**
| Unnamed: 0 | Cardiovascular events | Cardiovascular events.1 | All-cause mortality | All-cause mortality.1 | Decline in eGFR > 30% with death as competing risk | Decline in eGFR > 30% with death as competing risk.1 |
| --- | --- | --- | --- | --- | --- | --- |
| | (n = 40) | (n = 40) | (n = 26) | (n = 26) | (n = 42)* | (n = 42)* |
| | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value |
| Serum C3M | | | | | | |
| Unadjusted | 1.17 (0.57–2.40) | 0.67 | 1.57 (0.67–3.71) | 0.30 | 2.71 (1.38–5.30) | 0.004 |
| Adjusted | 1.52 (0.67–3.42) | 0.31 | 1.80 (0.69–4.73) | 0.23 | 2.98 (1.41–6.26) | 0.004 |
| Urine C3M | | | | | | |
| Unadjusted | 0.76 (0.51–1.12) | 0.16 | 0.88 (0.52–1.49) | 0.21 | 0.94 (0.70–1.28) | 0.70 |
| Adjusted | 0.83 (0.60–1.14) | 0.25 | 0.93 (0.62–1.40) | 0.74 | 0.96 (0.66–1.38) | 0.82 |
Fig 2 shows the Cox proportional hazard plots with tertiles of sC3M/uC3M and a >$30\%$ decline in eGFR. High levels of sC3M were associated with a significantly increased risk of decline in eGFR > $30\%$ ($$p \leq 0.047$$). The risk was significantly higher in tertile 3 compared with tertile 1 ($$p \leq 0.014$$). Tertile 2 was not significantly different from tertile 1 ($$p \leq 0.132$$). There was no association between tertiles of uC3M and development of a >$30\%$ decline in eGFR ($$p \leq 0.96$$).
**Fig 2:** *Cox survival plot of decline in eGFR >3 0% for sC3M and uC3M.Adjustment included sex, age, body mass index, LDL cholesterol, smoking, HbA1c, plasma creatinine, systolic blood pressure and urinary albumin excretion rate. uC3M: urinary C3M, sC3M: serum C3M.*
The mixed model analysis showed a significant association between sC3M and eGFR decline over time in both the unadjusted ($$p \leq 0.037$$) and adjusted model ($$p \leq 0.048$$), including adjustment for sex, age, body mass index, LDL cholesterol, smoking, HbA1c, plasma creatinine, systolic blood pressure and urinary albumin excretion rate. Urine C3M was not associated with eGFR decline over time in neither the unadjusted ($$p \leq 0.24$$) or adjusted model ($$p \leq 0.36$$).
## Additional analyses
Because of the strong correlation between sC3M and markers of inflammation, we performed additional adjustment for the markers of inflammation for all the outcomes. Results were confirmatory and higher sC3M remained associated with an increased risk of a >$30\%$ decline in eGFR ($$p \leq 0.001$$).
Adding the z-score for the inflammation markers to ROC-curves with sC3M and traditional cardiovascular risk factors did not improve the predictive value on any of the endpoints (CVD events: $$p \leq 0.36$$; all-cause mortality: $$p \leq 0.32$$; >$30\%$ decline in eGFR: $$p \leq 0.75$$).
Urinary C3M were divided with urine creatinine to normalize for urine output. Since this is correlated with the serum creatinine level, we performed a sensitivity analysis not including serum creatinine in the adjusted model for eGFR decline. This did not change the estimate or the lack of association with eGFR decline, HR ($95\%$ CI): 0.88 (0.63–1.25), p-value 0.47.
Additional sensitivity analysis showed that adding antihypertensive treatment to the model of eGFR decline did not change the results, HR ($95\%$ CI): 2.94 (1.40–6.19).
## Discussion
In the present study, we demonstrated that sC3M was positively correlated with markers of inflammation and endothelial dysfunction at baseline. Moreover, higher sC3M was a risk marker for progression of kidney disease during follow-up. In contrast, sC3M was not a risk marker for development of CVD or mortality and uC3M was not a risk marker for any of the endpoints.
We found no correlation between serum and urine C3M. This phenomenon has been observed across several other studies [6, 9]. One explanation could be that C3M fragments have different lengths and that a large proportion of the circulating fragments are not freely filtered into the urine. Another explanation could be that C3M produced by collagen type III degradation in the renal stoma were released directly into the urine and thereby not systemically available.
## Baseline correlations between C3M and markers of inflammation and endothelial dysfunction
Endothelial dysfunction and chronic inflammation are considered important factors in the pathogenesis of DKD. Previous studies have found markers of both endothelial dysfunction and inflammation associated with onset and progression of albuminuria [19, 20], with decline in eGFR [21] and with development of CVD and mortality [17].
Our results showed consistent, positive correlations between sC3M and markers of inflammation and some of the markers of endothelial dysfunction. Supporting these findings, another study showed a positive correlation between sC3M and CRP in CKD [6]. These consistent correlations indicate that these pathways are interconnected in the pathogenesis of DKD. Despite the consistent and strong correlations, sC3M was not associated with development of CVD and mortality in this population, even though associations between the markers of inflammation and endothelial dysfunction and risk of both CVD and mortality have previously been shown in the same population [17]. This adds evidence to the conclusion that sC3M is related to kidney disease but not CVD. The findings from the ROC analyses showed that adding the markers of inflammation to a model including traditional cardiovascular risk factors and sC3M did not add predictive value for any of the endpoints.
In contrast to sC3M, uC3M was only correlated with IL-6 in adjusted analyses, and not correlated with any of the other markers of inflammation and was positively correlated with only one of the markers of endothelial dysfunction.
## Serum C3M as a risk marker
In a study in persons with T1D, Pilemann-Lyberg et al. found higher sC3M to be a risk marker for a >$30\%$ decline in eGFR and for development of ESKD in the unadjusted models, but the association was lost after adjustment [9]. Similar findings were demonstrated in a population with CKD, where higher sC3M was associated with higher risk of CKD progression, but the significance was lost in the adjusted analysis [6].
None of the clinical characteristics, including kidney function and albuminuria, differed between the tertiles of sC3M at baseline. Since sC3M was a risk marker for kidney disease progression, this could mean that sC3M, has a value as risk marker, that are not evident from traditional measures of kidney function.
Consistently with previous findings, sC3M was not a risk marker for CVD or mortality in this population. In T1D, Pilemann-Lyberg et al found an association between higher sC3M and risk of CVD and mortality, but the significance was lost after adjustment for risk markers [9]. Similarly, in the study by Genovese et al, sC3M was not a risk marker for mortality in the adjusted model [6].
Serum C3M is highly associated with markers of systemic inflammation. But despite the evident association between systemic inflammation and both CVD, mortality and kidney disease progression, sC3M seems only to be associated with kidney disease progression.
## Urine C3M as a risk marker
In the population with T1D, the authors found associations between lower uC3M and higher risk of a >$30\%$ decline in eGFR and development of ESKD in unadjusted models, but associations were lost after adjustment [9]. In the present study we saw the same inverse association, but statistically insignificant. Genovese et al showed an association between higher levels of uC3M and lower risk of CKD progression in a mixed CKD population. A sub-analysis confirmed the findings in a diabetes subpopulation (diabetes type was not specified), however the participants were not necessarily diagnosed with DKD [6]. All mentioned uC3M measurements have been adjusted for urine output by dividing with urine creatinine concentration. The potential of uC3M as a risk marker in CKD patients seems to depend on the investigated population, and in the population investigated in this work, which includes type 2 diabetes patients with early to moderate kidney disease, the uC3M marker was not informative. The lack of correlation with markers for inflammation and endothelial dysfunction supports that the urinary level of C3M is not a valuable measure in this population. Further investigations are required in patients with more advanced DKD to establish whether this can be a risk marker in a subpopulation of DKD.
## Strengths and limitations
A strength of the study was the prospective design and the adequate length of follow-up. Participants were consecutively recruited which minimized selection bias and there was a limited loss to follow-up and missing data. The cohort was well-characterized, which enabled adjustment for important risk factors.
The relatively small study population and thereby a low number of events was a limitation. The cohort consisted of persons with T2D and microalbuminuria without known CVD, which limits the generalizability of the results. Comorbidities could have influenced the results, as fibrotic activity in other organs might affect levels of C3M. Likewise levels of uC3M can be affected by impaired excretion following declining kidney function and can be altered by non-selective proteinuria in kidney diseases like DKD [22].
Future studies should elaborate on the role of C3M as a biomarker for predicting kidney disease progression and the potential effects of therapeutic modification. In rodents, sodium-glucose transport protein 2 (SGLT2) inhibitors have shown to have anti-inflammatory and antifibrotic effects [23, 24]. Recently the nonsteroidal mineralocorticoid receptor antagonist finerenone was demonstrated to reduce progression of CKD and development of CVD (particularly reduced hospitalization for heart failure) in two large trials FIDELIO-DKD and FIGARO-DKD including patients with type 2 diabetes and a broad range of CKD with UACR >30 mg/g and eGFR>25 ml/min/1.73m2 [25, 26]. Experimental studies have demonstrated that blocking mineralocorticoid receptor overactivation protects the kidney by reduced inflammation and fibrosis without affecting blood pressure [27]. Decreasing C3M levels with finerenone could be a way to suggest the same mode of action in humans. Unpublished results have shown that treatment with the glucagon-like peptide-1 receptor agonist (GLP1RA) dulaglutide in T2D was associated with increasing levels of urinary C3M. This could potentially explain some of the beneficial treatment effects of both SGLT2 inhibitors and GLP1RAs.
In conclusion, we found a positive correlation between sC3M and markers of inflammation and endothelial dysfunction, and that elevated sC3M was a risk marker for progression of DKD in persons with type 2 diabetes.
## References
1. 1System USRD. 2021 USRDS Annual Data Report: Epidemiology of kidney disease in the United States. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD, 2021. Report No.. *2021 USRDS Annual Data Report: Epidemiology of kidney disease in the United States* (2021)
2. Gansevoort RT, Correa-Rotter R, Hemmelgarn BR, Jafar TH, Heerspink HJ, Mann JF. **Chronic kidney disease and cardiovascular risk: epidemiology, mechanisms, and prevention**. *Lancet* (2013) **382** 339-52. DOI: 10.1016/S0140-6736(13)60595-4
3. DuBose TD. **American Society of Nephrology Presidential Address 2006: chronic kidney disease as a public health threat—new strategy for a growing problem**. *J Am Soc Nephrol* (2007) **18** 1038-45. DOI: 10.1681/ASN.2006121347
4. Adelborg K, Nicolaisen SK, Hasvold P, Palaka E, Pedersen L, Thomsen RW. **Predictors for repeated hyperkalemia and potassium trajectories in high-risk patients—A population-based cohort study**. *PLoS One* (2019) **14** e0218739. DOI: 10.1371/journal.pone.0218739
5. Pei G, Zeng R, Han M, Liao P, Zhou X, Li Y. **Renal interstitial infiltration and tertiary lymphoid organ neogenesis in IgA nephropathy**. *Clin J Am Soc Nephrol* (2014) **9** 255-64. DOI: 10.2215/CJN.01150113
6. Genovese F, Rasmussen DGK, Karsdal MA, Jesky M, Ferro C, Fenton A. **Imbalanced turnover of collagen type III is associated with disease progression and mortality in high-risk chronic kidney disease patients**. *Clin Kidney J* (2021) **14** 593-601. DOI: 10.1093/ckj/sfz174
7. van der Zijl NJ, Hanemaaijer R, Tushuizen ME, Schindhelm RK, Boerop J, Rustemeijer C. **Urinary matrix metalloproteinase-8 and -9 activities in type 2 diabetic subjects: A marker of incipient diabetic nephropathy?**. *Clin Biochem* (2010) **43** 635-9. DOI: 10.1016/j.clinbiochem.2010.02.006
8. Ebihara I, Nakamura T, Shimada N, Koide H. **Increased plasma metalloproteinase-9 concentrations precede development of microalbuminuria in non-insulin-dependent diabetes mellitus**. *Am J Kidney Dis* (1998) **32** 544-50. DOI: 10.1016/s0272-6386(98)70015-0
9. Pilemann-Lyberg S, Rasmussen DGK, Hansen TW, Tofte N, Winther SA, Holm Nielsen S. **Markers of Collagen Formation and Degradation Reflect Renal Function and Predict Adverse Outcomes in Patients With Type 1 Diabetes**. *Diabetes Care* (2019) **42** 1760-8. DOI: 10.2337/dc18-2599
10. Genovese F, Boor P, Papasotiriou M, Leeming DJ, Karsdal MA, Floege J. **Turnover of type III collagen reflects disease severity and is associated with progression and microinflammation in patients with IgA nephropathy**. *Nephrol Dial Transplant* (2016) **31** 472-9. DOI: 10.1093/ndt/gfv301
11. Moeller MJ, Kramann R, Lammers T, Hoppe B, Latz E, Ludwig-Portugall I. **New Aspects of Kidney Fibrosis-From Mechanisms of Injury to Modulation of Disease**. *Front Med (Lausanne)* (2021) **8** 814497. DOI: 10.3389/fmed.2021.814497
12. Reinhard H, Hansen PR, Persson F, Tarnow L, Wiinberg N, Kjær A. **Elevated NT-proBNP and coronary calcium score in relation to coronary artery disease in asymptomatic type 2 diabetic patients with elevated urinary albumin excretion rate**. *Nephrol Dial Transplant* (2011) **26** 3242-9. DOI: 10.1093/ndt/gfr009
13. von Scholten BJ, Reinhard H, Hansen TW, Oellgaard J, Parving HH, Jacobsen PK. **Urinary biomarkers are associated with incident cardiovascular disease, all-cause mortality and deterioration of kidney function in type 2 diabetic patients with microalbuminuria**. *Diabetologia* (2016) **59** 1549-57. DOI: 10.1007/s00125-016-3937-0
14. Gaede P, Lund-Andersen H, Parving HH, Pedersen O. **Effect of a multifactorial intervention on mortality in type 2 diabetes**. *N Engl J Med* (2008) **358** 580-91. DOI: 10.1056/NEJMoa0706245
15. Barascuk N, Veidal SS, Larsen L, Larsen DV, Larsen MR, Wang J. **A novel assay for extracellular matrix remodeling associated with liver fibrosis: An enzyme-linked immunosorbent assay (ELISA) for a MMP-9 proteolytically revealed neo-epitope of type III collagen**. *Clin Biochem* (2010) **43** 899-904. DOI: 10.1016/j.clinbiochem.2010.03.012
16. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, Feldman HI. **A new equation to estimate glomerular filtration rate**. *Ann Intern Med* (2009) **150** 604-12. DOI: 10.7326/0003-4819-150-9-200905050-00006
17. von Scholten BJ, Reinhard H, Hansen TW, Schalkwijk CG, Stehouwer C, Parving HH. **Markers of inflammation and endothelial dysfunction are associated with incident cardiovascular disease, all-cause mortality, and progression of coronary calcification in type 2 diabetic patients with microalbuminuria**. *J Diabetes Complications* (2016) **30** 248-55. DOI: 10.1016/j.jdiacomp.2015.11.005
18. von Scholten BJ, Reinhard H, Hansen TW, Lindhardt M, Petersen CL, Wiinberg N. **Additive prognostic value of plasma N-terminal pro-brain natriuretic peptide and coronary artery calcification for cardiovascular events and mortality in asymptomatic patients with type 2 diabetes**. *Cardiovasc Diabetol* (2015) **14** 59. DOI: 10.1186/s12933-015-0225-0
19. Persson F, Rossing P, Hovind P, Stehouwer CD, Schalkwijk CG, Tarnow L. **Endothelial dysfunction and inflammation predict development of diabetic nephropathy in the Irbesartan in Patients with Type 2 Diabetes and Microalbuminuria (IRMA 2) study**. *Scand J Clin Lab Invest* (2008) **68** 731-8. DOI: 10.1080/00365510802187226
20. Stehouwer CD, Gall MA, Twisk JW, Knudsen E, Emeis JJ, Parving HH. **Increased urinary albumin excretion, endothelial dysfunction, and chronic low-grade inflammation in type 2 diabetes: progressive, interrelated, and independently associated with risk of death**. *Diabetes* (2002) **51** 1157-65. DOI: 10.2337/diabetes.51.4.1157
21. Pena MJ, Heinzel A, Heinze G, Alkhalaf A, Bakker SJ, Nguyen TQ. **A panel of novel biomarkers representing different disease pathways improves prediction of renal function decline in type 2 diabetes**. *PLoS One* (2015) **10** e0120995. DOI: 10.1371/journal.pone.0120995
22. Genovese F, Manresa AA, Leeming DJ, Karsdal MA, Boor P. **The extracellular matrix in the kidney: a source of novel non-invasive biomarkers of kidney fibrosis?**. *Fibrogenesis Tissue Repair* (2014) **7** 4. DOI: 10.1186/1755-1536-7-4
23. Wang XX, Levi J, Luo Y, Myakala K, Herman-Edelstein M, Qiu L. **SGLT2 Protein Expression Is Increased in Human Diabetic Nephropathy: SGLT2 PROTEIN INHIBITION DECREASES RENAL LIPID ACCUMULATION, INFLAMMATION, AND THE DEVELOPMENT OF NEPHROPATHY IN DIABETIC MICE**. *J Biol Chem* (2017) **292** 5335-48. DOI: 10.1074/jbc.M117.779520
24. Ojima A, Matsui T, Nishino Y, Nakamura N, Yamagishi S. **Empagliflozin, an Inhibitor of Sodium-Glucose Cotransporter 2 Exerts Anti-Inflammatory and Antifibrotic Effects on Experimental Diabetic Nephropathy Partly by Suppressing AGEs-Receptor Axis**. *Horm Metab Res* (2015) **47** 686-92. DOI: 10.1055/s-0034-1395609
25. Bakris GL, Agarwal R, Anker SD, Pitt B, Ruilope LM, Rossing P. **Effect of Finerenone on Chronic Kidney Disease Outcomes in Type 2 Diabetes**. *N Engl J Med* (2020) **383** 2219-29. DOI: 10.1056/NEJMoa2025845
26. Pitt B, Filippatos G, Agarwal R, Anker SD, Bakris GL, Rossing P. **Cardiovascular Events with Finerenone in Kidney Disease and Type 2 Diabetes**. *N Engl J Med* (2021) **385** 2252-63. DOI: 10.1056/NEJMoa2110956
27. Kolkhof P, Joseph A, Kintscher U. **Nonsteroidal mineralocorticoid receptor antagonism for cardiovascular and renal disorders—New perspectives for combination therapy**. *Pharmacol Res* (2021) **172** 105859. DOI: 10.1016/j.phrs.2021.105859
|
---
title: The importance of oxytocin neurons in the supraoptic nucleus for breastfeeding
in mice
authors:
- Mitsue Hagihara
- Kazunari Miyamichi
- Kengo Inada
journal: PLOS ONE
year: 2023
pmcid: PMC10022762
doi: 10.1371/journal.pone.0283152
license: CC BY 4.0
---
# The importance of oxytocin neurons in the supraoptic nucleus for breastfeeding in mice
## Abstract
The hormone oxytocin, secreted from oxytocin neurons in the paraventricular (PVH) and supraoptic (SO) hypothalamic nuclei, promotes parturition, milk ejection, and maternal caregiving behaviors. Previous experiments with whole-body oxytocin knockout mice showed that milk ejection was the unequivocal function of oxytocin, whereas parturition and maternal behaviors were less dependent on oxytocin. Whole-body knockout, however, could induce the enhancement of expression of related gene(s), a phenomenon called genetic compensation, which may hide the actual functions of oxytocin. In addition, the relative contributions of oxytocin neurons in the PVH and SO have not been well documented. Here, we show that females with conditional knockout of oxytocin gene in both the PVH and SO undergo grossly normal parturition and maternal caregiving behaviors, while dams with a smaller number of remaining oxytocin-expressing neurons exhibit severe impairments in breastfeeding, leading to the death of their pups within 24 hours after birth. We also found that the growth of pups is normal even under oxytocin conditional knockout in PVH and SO as long as pups survive the next day of delivery, suggesting that the reduced oxytocin release affects the onset of lactation most severely. These phenotypes are largely recapitulated by SO-specific oxytocin conditional knockout, indicating the unequivocal role of oxytocin neurons in the SO in successful breastfeeding. Given that oxytocin neurons not only secrete oxytocin but also non-oxytocin neurotransmitters or neuropeptides, we further performed cell ablation of oxytocin neurons in the PVH and SO. We found that cell ablation of oxytocin neurons leads to no additional abnormalities over the oxytocin conditional knockout, suggesting that non-oxytocin ligands expressed by oxytocin neurons have negligible functions on the responses measured in this study. Collectively, our findings confirm the dispensability of oxytocin for parturition or maternal behaviors, as well as the importance of SO-derived oxytocin for breastfeeding.
## Introduction
Oxytocin (OT) is a nonapeptide hormone produced by OT neurons in the paraventricular (PVH) and supraoptic (SO) hypothalamic nuclei. Recent studies have reported that OT plays important roles in sexual, maternal, and social behaviors [1–3], in addition to the functions documented in classical studies such as the induction of labor and milk ejection. However, studies on whole-body OT knockout (KO) mice have shown that milk ejection is a specific and essential function of OT, but dispensable for parturition [4–6]. Similarly, the expression of maternal caregiving behaviors does not require OT, given that the performance of parental behaviors by OT KO dams is largely similar to that by controls [5, 7], except in food-limited stressful environments [8]. Despite the clear consistencies across studies, the significance of OT signaling in the regulation of labor, milk ejection, and parental behaviors remains unclear, given that the phenotypes of a whole-body KO might be genetically compensated by the upregulation of related gene(s) [9, 10]. For example, a recent study analyzing a different function of OT neurons, weight homeostasis [11–13], reported that PVH-specific OT conditional KO (cKO) mice showed a hyperphagic obesity phenotype that was not apparent in the whole-body OT KO [14]. In addition, although OT neurons can release not only OT, but also other neurotransmitters or neuropeptides, such as glutamate [15], to our knowledge, the functional roles of such non-OT ligands in the regulation of labor, milk ejection, and parental behaviors have not been described.
Here, we show the relevance of OT secretion on labor, milk ejection, and parental behaviors using an OT cKO mouse line described previously [16]. Our approach offers a better temporal resolution, which allows us to avoid the influence of possible developmental or genetic compensation [9, 10]. We also test the relative contributions of OT secretion from the PVH and SO nuclei to maternal physiology and behaviors by restricting the manipulation to a single hypothalamic nucleus. This improved spatial resolution may reveal distinct functions of OT neurons in the PVH and SO, which show distinct input–output organizations [17]. Furthermore, we compare the phenotype of OT cKO with that observed in an OT neuron-specific cell ablation experiment, in which we test an additional function of non-OT ligands expressed in the OT neurons. Taken together, we aim to improve the resolution of loss-of-function studies of OT ligand and OT neurons in terms of maternal physiology and behaviors.
## OT cKO mothers in both PVH and SO fail in raising pups
Previous studies with whole-body OT KO female mice showed that fertility and pregnancy were unaffected [4–6], whereas parturition was either normal or delayed depending on the genetic background [5]. To evaluate such phenotypes in OT cKO females, we prepared OTflox/flox mice described previously [16]. In this line, Cre expression deletes floxed exon 1 of the OT gene, resulting in the loss of transcription of OT mRNA. We crossed OT KO (OT–/–) and OTflox/flox mice and prepared OTflox/–mice [16]. Each OTflox/–female mouse was injected with an adeno-associated virus (AAV) driving Cre into both the PVH and SO (Fig 1A and 1B). Five weeks or longer after the injection, which was sufficient to reduce significantly the expression of OT mRNA in the PVH and SO (Fig 1C and 1D) [14], each female was crossed with a wild-type male (Fig 1B; Materials and Methods). We found that the durations of both pregnancy and pup delivery were indistinguishable between dams that had received vehicle (control) and AAV-Cre (+Cre) (Fig 1E–1G). However, as described previously [4–6], we found that in $54\%$ (= $\frac{7}{13}$) of the +Cre mothers’ cages, all pups were dead in the next day of birth (postpartum day 1 [PPD 1]) without any sign of infanticide, even though they were alive at PPD 0 (Materials and Methods). This phenotype was not observed in the cages of vehicle-injected mothers ($\frac{0}{9}$). The fraction of surviving pups after PPD 1 was largely binomial: nearly $100\%$ of pups survived in the cages of “success” dams, whereas $0\%$ survived in the cages of “failure” dams (Fig 1H). By visualizing the OT mRNA and counting OT mRNA-positive (OT+) neurons, we found that the smaller the number of remaining OT+ neurons in PVH and SO, the more likely the failure phenotype appeared (Fig 1I). In particular, in this mouse line, when the remaining OT+ neurons in the SO were fewer than 200 (orange shadow in Fig 1I), mothers mostly failed to raise their pups at PPD 1. Together with the observation that OT KO dams failed to eject milk [4–6], the failure of raising pups associated with OT cKO dams was most likely due to defects in milk ejection. These results support the indispensable functions of OT ligands secreted from the PVH and/or SO for milk ejection as essential for raising pups.
**Fig 1:** *Grossly normal parturition and abnormal pup survival in OT cKO mothers.(A) Schematic of the viral injection. AAV-Cre or vehicle was injected into both the bilateral PVH and SO. (B) Schematic of the timeline of the experiment. (C) Representative coronal sections. OT and Cre in situ staining are shown in magenta and green, respectively. Blue, DAPI. Scale bar, 20 μm. (D) Number of remaining OT+ neurons in the PVH (left) or SO (right). ***p < 0.001, two-sided Mann–Whitney U-test. n = 9 and 13 mothers for vehicle and +Cre, respectively. (E) The duration of pregnancy was not statistically different (p > 0.56, two-sided Mann–Whitney U-test). n = 9 and 7 for vehicle and +Cre, respectively. Magenta dots indicate stillbirth. (F) Left, number of pups born from the vehicle or +Cre mothers was not statistically different (p > 0.90, two-sided Mann–Whitney U-test). Right, number of pups that died within the day of delivery (PPD 0). n = 13 and 14 mothers for vehicle and +Cre, respectively. (G) Left, mean pup delivery interval in each dam. Middle, coefficient of variation showing the interval variability. Right, cumulative probability of interval. n = 59 and 40 pups for vehicle and +Cre, respectively. The p-value is shown in the panel (Kolmogorov–Smirnov test). n = 8 and 5 mothers for vehicle and +Cre, respectively. (H) Survival rate of pups at PPD 1. Note that pups that died at PPD 0 (F) were excluded. n = 9 and 13 mothers for vehicle and +Cre, respectively. (I) Relationship between the success (black dots) or failure (orange dots) of raising pups at PPD 1 and the number of remaining OT+ neurons in the PVH (x-axis) or SO (y-axis). Data from the same mice shown in D. (J) Average weight of pups per dam. Mothers in which all pups were dead at PPD 1 were excluded from +Cre. No statistical difference was found in vehicle and +Cre (two-way ANOVA with repeated measurements). n = 7 and 6 for vehicle and +Cre, respectively. (K) Cumulative probability of weight of pups at PPD 3, PPD 6, and PPD 9. Mothers in which all pups were dead at PPD 1 were not included in +Cre. The p-value is shown in the panel (Kolmogorov–Smirnov test). n = 53–54 and 46 pups from seven and six mothers for vehicle and +Cre, respectively. Error bars, standard error of the mean.*
If pups born to OT cKO mothers survive PPD 1, do they develop normally or show any growth defects? Given that several dams that had received AAV-Cre in both PVH and SO successfully raised pups after PPD 1, we measured the weight of those pups at PPD 3, PPD 6, and PPD 9. We found that the average pup weights of each dam were not statistically different from those of the controls (Fig 1J). Also, the weight distribution of the pups was not statistically different from that of the pups born to vehicle-injected control mothers (Fig 1K). The development of pups was weakly correlated with the number of remaining OT+ neurons (S1 Fig). These results indicate that the growth of pups is normal even under OT cKO in PVH and SO as long as pups survive PPD 1, suggesting that the reduced OT release in cKO mothers affects the onset of lactation most severely.
## Loss of OT gene from PVH and SO minimally affects maternal behaviors
OT has been shown to promote maternal caregiving behaviors in rodents [18, 19]. However, previous studies with OT KO mothers have reported normal pup-directed parental behaviors [4, 5, 7], except under food-limited stressful environments [8]. To examine whether OT cKO mothers exhibit any defects in maternal caregiving behaviors, we analyzed the OT cKO mothers characterized in Fig 1 (see Materials and Methods). After injecting AAV-Cre into the bilateral PVH and SO, a behavioral assay was performed at PPD 3 (Fig 2A and 2B). We found that all mothers showed retrieval, irrespective of the Cre expression or success/failure phenotypes of raising pups (Fig 2C). The latency to investigate the pups, number of retrieved pups, and parental care duration were not significantly different (Fig 2D–2F), suggesting that parental behaviors were unaffected by OT cKO in the PVH and SO. A significant difference was found in the time course of retrieval, as shown in the cumulative probability of retrieval (Fig 2G): mothers that received AAV-Cre required more time to retrieve pups. Among OT cKO mothers, those who exhibited the success phenotype of raising pups at PPD 1 were slightly but significantly slower in pup retrieval compared with control mothers, whereas those who showed the failure phenotype were much slower (Fig 2H). As the OT cKO mothers who succeeded in raising pups had more abundant experiences of caring for pups, the observed difference in retrieval latency may be explained by the amount of maternal learning. Taken together, although slight defects could be observed in the efficiency of pup retrieval, the overall performance of parental behaviors in mothers was unaffected by OT cKO in the PVH and SO.
**Fig 2:** *Grossly normal parental behaviors of mothers with OT cKO in the PVH and SO.(A) Schematic of the viral injection. AAV-Cre or vehicle was injected into the bilateral PVH and SO. (B) Schematic of the timeline of the experiment. A behavioral assay was conducted at PPD 3. (C) Percentage of mothers showing attack, ignore or retrieve. Note that attack or ignore was not observed in our dataset. (D) Latency to the first investigation of pups was not statistically different (two-sided Mann–Whitney U-test). (E) Number of retrieved pups. (F) Parental care duration was not statistically different (two-sided Mann–Whitney U-test). (G) Cumulative probability of pup retrieval. The p-value is shown in the panel (Kolmogorov–Smirnov test). (H) Cumulative probability of pup retrieval of vehicle dams (gray line), and +Cre dams that exhibited success (black line) or failure (orange line) in raising pups at PPD 1 (***p < 0.001, Kolmogorov–Smirnov test with Bonferroni correction). n = 5 each for the success and failure phenotypes, respectively. n = 7 and 10 for vehicle and +Cre from the mice analyzed in Fig 1D, respectively. Orange dots indicate mothers with the failure phenotype. Error bars, standard error of the mean.*
## Virgin females with a loss of OT gene from PVH and SO were more likely to ignore pups
Although OT neurons are not necessary to induce caregiving behaviors in OT KO mothers [5, 7] and mothers of OT cKO in the PVH and SO (Fig 2), significant defects in the expression of maternal caregiving behaviors were observed in OT KO virgin females [20]. Furthermore, a recent study reported that virgin females co-housed with experienced mothers and pups began to show caregiving behaviors, and chemogenetic inhibition of OT neurons in the PVH impaired the expression of such alloparental caregiving behaviors [21]. Therefore, we next tested if OT cKO induces any defects in the expression of caregiving behaviors in virgin females. We performed cKO of OT gene from bilateral PVH and SO in virgin females (Fig 3A). Those females were allowed to co-house with a dam and pups for six consecutive days (Fig 3B; Materials and Methods). We confirmed a drastic reduction of the number of OT+ neurons in the PVH and SO (Fig 3C). Consistent with the previous study [21], virgin females that received AAV-Cre injection showed a significantly higher rate of ignore (Fig 3D; $p \leq 0.001$, two-tailed Fisher’s exact test). While latency to investigate did not differ (Fig 3E), decreases in the number of retrieved pups (Fig 3F) and parental care duration (Fig 3G) were found in +Cre females but did not reach the level of statistical significance ($p \leq 0.27$ and 0.15 for Fig 3F and 3G, respectively, two-sided Mann–Whitney U-test). Similar to the mothers (Fig 2G), virgin females that received AAV-Cre required more time to retrieve pups (Fig 3H). Taken together, these results suggest that loss of OT expression leads to a modest but significant defect in the expression of alloparental caregiving behaviors in virgin females.
**Fig 3:** *OT cKO in the PVH and SO in virgin females increases the probability to ignore pups.(A) Schematic of the viral injection. AAV-Cre or vehicle was injected into the bilateral PVH and SO. (B) Schematic of the timeline of the experiment. A behavioral assay was performed 6 days after co-habiting with a mother and pups (see Materials and Methods). (C) Number of remaining OT+ neurons (**p < 0.01, two-sided Mann–Whitney U-test). n = 7 each. (D) Percentage of females showing attack, ignore or retrieve. +Cre females were more likely to ignore the pups (p < 0.001, two-tailed Fisher’s exact test). (E) Latency to the first investigation of pups was not statistically different (two-sided Mann–Whitney U-test). (F) Number of retrieved pups (p > 0.27, two-sided Mann–Whitney U-test). (G) A decrease in parental care duration was found in +Cre females but did not reach the level of statistical significance (p > 0.15, two-sided Mann–Whitney U-test). (H) Cumulative probability of pup retrieval. The p-value is shown in the panel (Kolmogorov–Smirnov test). Error bars, standard error of the mean.*
## Cell ablation of OT neurons in both the PVH and SO recapitulates the OT cKO phenotypes
OT neurons release not only OT, but also other neurotransmitters or neuropeptides, such as glutamate [15]. If such non-OT ligands released from OT neurons have any additional functions over OT in labor, milk ejection, or the raising of pups, cell-type-specific ablation of OT neurons would exhibit more severe phenotypes than would those observed in OT cKO dams. To examine this possibility, we performed cell ablation of OT neurons by injecting taCasp3-encoding AAV [22] into both the PVH and SO of OTCre/+ mice (Fig 4A and 4B). As a result, the taCasp3-encoding AAV reduced the number of neurons expressing OT mRNA in the PVH and SO (Fig 4C and 4D) most likely because the virus induced cell death in OT neurons [22]. The females became pregnant, and the number of littered pups was not statistically different from that of the controls (Fig 4E). Similar to the OT cKO (Fig 1), we found that $33\%$ (= $\frac{3}{9}$) of the +taCasp3 dams failed to raise their pups at PPD 1 (Fig 4F). The dams with a smaller number of the remaining OT+ neurons in the SO were more likely to show the failure phenotype at PPD 1 (Fig 4G). We also found that neither the average pup weights in each dam nor the weight distribution of the pups was statistically different from the controls (Fig 4H and 4I). These results largely recapitulate the phenotypes observed in OT cKO in both the PVH and SO (Fig 1) without additional defects. These data do not suggest an additional role of non-OT ligands expressed in the OT neurons in pregnancy, parturition, or milk ejection.
**Fig 4:** *Cell ablation of OT neurons in the PVH and SO leads to a failure of raising pups.(A) Schematic of the viral injection. AAV-FLEx-taCasp3-TEVp or vehicle was injected into the bilateral PVH and SO. (B) Schematic of the timeline of the experiment. (C) Representative coronal sections. OT in situ staining is shown in magenta. Blue, DAPI. Scale bar, 20 μm. (D) Number of remaining OT+ neurons in the PVH (left) or SO (right). **p < 0.01, two-sided Mann–Whitney U-test. n = 7 and 9 mothers for vehicle and +taCasp3, respectively. (E) The number of pups born was not statistically different (two-sided Mann–Whitney U-test). No dead pups were found. n = 7 and 9 mothers for vehicle and +taCasp3, respectively. (F) Survival rate of pups at PPD 1. n = 7 and 9 mothers for vehicle and +taCasp3, respectively. (G) Relationship between the success (black dots) or failure (orange dots) of raising pups at PPD 1 and the number of remaining OT+ neurons in PVH (x-axis) or SO (y-axis). Data from the same mice shown in D. (H) Average weight of pups per dam. Mothers in which all pups were dead at PPD 1 were excluded from +taCasp3. No statistical difference was found in vehicle and +taCasp3 (two-way ANOVA with repeated measurements). n = 7 and 6 for vehicle and +taCasp3, respectively. (I) Cumulative probability of weight of pups at PPD 3, PPD 6, and PPD 9. Mothers in which all pups were dead at PPD 1 were excluded in +taCasp3. The p-value is shown in the panel (Kolmogorov–Smirnov test). n = 58 and 41–44 pups from seven and six mothers for vehicle and +taCasp3, respectively. Error bars, standard error of the mean.*
## Loss of OT neurons from PVH and SO minimally affects maternal behaviors
To examine the functional role of OT neurons, including non-OT neurotransmitters or neuropeptides, we further analyzed the parental caregiving behaviors of mothers with cell ablation of OT neurons (Fig 5A and 5B) characterized in Fig 4. Similar to OT cKO in the PVH and SO (Fig 2), we found no major defects in the execution of caregiving behaviors (Fig 5C–5F), except that dams expressing taCasp3 required more time to retrieve pups (Fig 5G and 5H). Taken together, under the conditions of cell-type-specific ablation experiments, OT neurons were not necessary to execute maternal caregiving behaviors.
**Fig 5:** *Grossly normal parental behaviors of mothers with cell ablation of OT neurons.(A) Schematic of the viral injection. AAV-FLEx-taCasp3-TEVp or vehicle was injected into the bilateral PVH and SO. (B) Schematic of the timeline of the experiment. A behavioral assay was conducted at PPD 3. (C) Percentage of mothers showing attack, ignore or retrieve. Note that attack or ignore was not observed in our dataset. (D) Latency to the first investigation of pups was not statistically different (two-sided Mann–Whitney U-test). (E) Number of retrieved pups. (F) Parental care duration was not statistically different (two-sided Mann–Whitney U-test). (G) Cumulative probability of pup retrieval. The p-value is shown in the panel (Kolmogorov–Smirnov test). (H) Cumulative probability of pup retrieval of vehicle dams (gray line), and +taCasp3 dams that exhibited success (black line) or failure (orange line) in raising pups at PPD 1 (***p < 0.001, Kolmogorov–Smirnov test with Bonferroni correction). n = 6 and 3 for the success and failure phenotypes, respectively. n = 7 and 9 for vehicle and +taCasp3 from the mice analyzed in Fig 4D, respectively. Orange dots indicate mothers with the failure phenotype. Error bars, standard error of the mean.*
## OT ligands from the SO are needed for the success of raising pups
The results of the cKO of the OT gene (Fig 1) and cell ablation of OT neurons (Fig 4) together imply that mothers with a smaller number of the remaining OT+ neurons in the SO are more likely to fail in raising pups at PPD 1 compared to that in the PVH. Given that our approach enables the cKO of the OT gene restricted to a single hypothalamic nucleus, first, we performed cKO of the OT gene selectively in the PVH (Fig 6A–6C). The number of littered pups was not significantly different (Fig 6D) and all dams successfully raised pups at PPD 1 (Fig 6E and 6F). We found no defects in the development of pups in the PVH-specific OT cKO mothers (Fig 6G and 6H).
**Fig 6:** *OT expression in the PVH does not affect the survival of pups.(A) Schematic of the viral injection. AAV-Cre or vehicle was injected into the bilateral PVH. (B) Schematic of the timeline of the experiment. (C) Number of remaining OT+ neurons in the PVH (left) or SO (right). **p < 0.01, two-sided Mann–Whitney U-test. n = 6 and 6 mothers for vehicle and +Cre, respectively. (D) The number of pups born was not statistically different (p > 0.26, two-sided Mann–Whitney U-test). No dead pups were found. n = 6 and 6 mothers for vehicle and +Cre, respectively. (E) Survival rate of pups at PPD 1. (F) Relationship between the success (black dots) or failure (orange dots) of raising pups at PPD 1 and the number of remaining OT+ neurons in the PVH (x-axis) or SO (y-axis). Data from the same mice shown in C. (G) Average weight of pups per dam. No statistical difference was found in vehicle and +Cre (two-way ANOVA with repeated measurements). n = 6 and 6 for vehicle and +Cre, respectively. (H) Cumulative probability of weight of pups in PPD 3, PPD 6, and PPD 9. The p-value is shown in the panel (Kolmogorov–Smirnov test). n = 41 and 51 pups from 6 and 6 mothers for vehicle and +Cre, respectively. Error bars, standard error of the mean.*
Next, we performed cKO of the OT gene in the SO (Fig 7A–7C). The number of littered pups was not statistically different (Fig 7D). Similar to the OT cKO in both the PVH and SO (Fig 1), $50\%$ (= $\frac{4}{8}$) of dams showed a failure in raising pups at PPD 1 (Fig 7E). As expected, dams with a smaller number of the remaining OT+ neurons in the SO, but not in the PVH, were more likely to show the failure phenotype: approximately 200 remaining OT+ neurons in the SO were needed for success in raising pups at PPD 1 (Fig 7F). These results suggest that OT+ neurons in the PVH and SO show differential contributions to pup survival. As described above, we measured the weight of pups at PPD 3, PPD 6, and PPD 9 to analyze their growth. Overall, the growth of the pups was not largely different between +Cre and vehicle (Fig 7G and 7H).
**Fig 7:** *OT expression in the SO is necessary for the survival of pups.(A) Schematic of the viral injection. AAV-Cre or vehicle was injected into the bilateral SO. (B) Schematic of the timeline of the experiment. (C) Number of remaining OT+ neurons in the PVH (left) or SO (right). **p < 0.01, two-sided Mann–Whitney U-test. n = 6 and 8 mothers for vehicle and +Cre, respectively. (D) The number of pups born was not statistically different (p > 0.21, two-sided Mann–Whitney U-test). No dead pups were found. n = 6 and 8 mothers for vehicle and +Cre, respectively. (E) Survival rate of pups at PPD 1. n = 6 and 8 mothers for vehicle and +Cre, respectively. (F) Relationship between the success (black dots) or failure (orange dots) of raising pups at PPD 1 and the number of remaining OT+ neurons in the PVH (x-axis) or SO (y-axis). Data from the same mice shown in C. (G) Average weight of pups per dam. Mothers in which all pups were dead at PPD 1 were excluded from +Cre. No statistical difference was found in vehicle and +Cre (two-way ANOVA with repeated measurements). n = 6 and 4 for vehicle and +Cre, respectively. (H) Cumulative probability of weight of pups in PPD 3, PPD 6, and PPD 9. Mothers in which all pups were dead at PPD 1 were excluded in +Cre. The p-value is shown in the panel (Kolmogorov–Smirnov test). n = 40 and 25 pups from six and four mothers for vehicle and +Cre, respectively. Error bars, standard error of the mean.*
## Discussion
Parturition and lactation are complex biological processes and OT has been considered to be critically involved in both [23, 24]. However, previous studies with whole-body KO of OT have reported that only milk ejection was severely impaired, whereas parturition was less dependent on OT [4–6]. Decades of studies have suggested that animals have mechanisms to maintain fitness in the presence of harmful mutations. Although this can be achieved by multiple strategies, one possible explanation for the dispensability of OT on the parturition can be the existence of compensatory mechanisms: whole-body KO of OT may enhance the expression of related gene(s) [9, 10], thereby compensating for the absence of OT. The findings of the present study do not support this explanation, given that cKO of the OT gene at the adult stage mostly phenocopied the whole-body KO. Furthermore, we showed that cell ablation of OT neurons resulted in similar phenotypes obtained from OT cKO. These findings, together with classical KO studies, suggest that OT is facilitatory but dispensable for parturition.
In the present study, we performed cKO of OT gene by injecting AAV-Cre into the PVH and/or SO. Our approach enables the removal of the OT gene restricted to the brain, and even to a single hypothalamic nucleus, providing a resolution that exceeds previous studies. By visualizing the mRNA of OT, we found that the number of remaining OT+ neurons in SO correlates well with the survival rate of pups (Figs 1I and 7F): the success of raising pups at PPD 1 requires more than 200 OT+ neurons in the SO, though this number may differ depending on the genetic background. What are the underlying mechanisms? During the milk ejection reflex, the pulsatile OT secretion necessary for the contraction of the mammary glands is mediated by synchronous bursts of OT neurons in the PVH and SO of both hemispheres [25, 26]. In one scenario, there is a pulse generator of synchronous spiking of OT neurons in the SO, the activity of which is then transmitted to all OT neurons, including those in the PVH. This intra- and inter-nucleus transmission of activity may be mediated by OT-to-OTR signaling [27–29]. In this case, loss-of-function of OT neurons in the SO would impair the burst firing of OT neurons in the PVH. Alternatively, the OT neurons in the PVH may remain active even in the absence of OT release from the SO, but the total amount of OT released into the peripheral circulation is not sufficient for milk ejection. In this scenario, pituitary-projecting magnocellular neurons in the SO make a greater contribution to milk ejection [17, 30]. Recent advances in the optical recording of OT neurons during lactation [31, 32] may help explore these possibilities in future studies.
Our data revealed that if the pups survived at PPD 1, their growth was largely normal. This suggests that there is a bottleneck in the establishment of the milk ejection reflex by PPD 1. Fiber photometry-based imaging studies of OT neurons in lactating mice [32] and rats [31] have revealed that the amplitudes of pulsatile Ca2+ transients are the lowest at PPD 1 and increased afterward. These findings may explain why milk ejection at PPD 1 is the most sensitive to the perturbation of OT neurons: once milk ejection is achieved at PPD 1, the plasticity of OT neurons permits more efficient milk ejection as the pups grow.
In the present study, we showed that OT cKO or cell ablation of OT neurons led to delayed pup retrieval in the execution of parental behaviors in mothers, as well as a reduction in the expression of caregiving behaviors in virgin females. In principle, these effects could be due to the experimental procedures employed, such as the injection of AAVs or potential damage to the neural connections caused by the cutting of some afferent or efferent connections of OT neurons. However, negative control experiments that employed the same surgical procedures but just injected a vehicle (saline) showed no phenotype. The likely explanation is that the loss of OT ligands (Fig 2) or OT neurons (Fig 5) may reduce the reward value of pups, thereby delaying pup retrieval, as the connections from the OT neurons to the dopamine neurons in the ventral tegmental area have been documented in the context of social behaviors [33, 34]. Except for this relatively minor phenotype, all the OT cKO or OT neuron-ablated mothers showed normal parental behaviors. Although we cannot fully exclude the possibility that a small number of remaining OT neurons in our virus-based procedures may support these behaviors, the results are in clear contrast to the fathers, given that OT KO or OT cKO in the PVH lead to defective parental behaviors in fathers [16]. Our observations, together with classical whole-body KO studies, suggest that maternal behaviors are supported by multiple redundant neural systems that can compensate for the loss of OT functions, including the modulation of sensory systems [19, 35]. This likely echoes the evolutional trait of mammalian species that depends more on maternal care for survival during infancy [36]. The mechanisms of such a redundancy remain an open question. Future studies on how OT facilitates parental behaviors, based on the detailed input–output structures of OT neurons [17, 37], could help define how multiple redundant mechanisms for parental behaviors work in the maternal brain.
## Animals
Animals were housed under a 12-hour light/12-hour dark cycle with ad libitum access to food and water. Wild-type C57BL/6J mice were purchased from Japan SLC. OTflox/flox and OT–/–mice lines, described previously [16], were generated in C57BL/6J background. We chose the OTflox/–model to increase the efficiency of cKO (Figs 1–3, 6 and 7). If we had used the OTflox/flox mice for cKO, a small fraction of the flox alleles that do not experience recombination would easily mask the phenotypes due to the high expression of OT gene. OTCre/+ (Jax #024234), purchased from the Jackson Laboratory, was backcrossed more than five generations to C57BL/6J mouse. All the experimental procedures were approved by the Institutional Animal Care and Use Committee of the RIKEN Kobe branch (A2017-15-13).
## Viral preparations
We obtained AAV serotype 9 hSyn-Cre from Addgene (#105555, 2.3 × 1013 genome particles [gp]/ml). AAV serotype 1 EF1a-FLEx-taCasp3-TEVp (5.8 × 1012 gp/ml) [22] was purchased from the University of North Carolina viral core.
## Stereotactic injection
To target AAV into a specific brain region, stereotactic coordinates were defined for each brain region based on the Allen Mouse Brain Atlas [38]. Mice were anesthetized with 65 mg/kg ketamine (Daiichi Sankyo) and 13 mg/kg xylazine (X1251, Sigma-Aldrich) via intraperitoneal injection and head-fixed to stereotactic equipment (Narishige). The following coordinates were used (in mm from the bregma for anteroposterior [AP] and mediolateral [ML], and dorsoventral [DV] from the surface of the brain): PVH, AP –0.8, ML 0.2, DV 4.5; SO, AP –0.7, ML 1.2, DV 5.5. The injected volume of AAV was 200 nl at a speed of 50 nl/min. After the viral injection, the animal was returned to the home cage. Each viral injection and all subsequent experiments except Fig 1E–1G were conducted by two experimenters: experimenter 1 prepared two identical tubes containing either saline (vehicle) or solution containing AAV, and experimenter 2, who was blinded to the contents of the tubes, conducted the injections, behavioral assay, and data analysis.
## Measurement of the duration of pregnancy and pup birth intervals
A virgin female (13 weeks old) individually housed for 5 weeks or longer (Figs 1, 6 and 7) or 2 weeks (Fig 4) after the viral injection was paired with a wild-type male. The next day, the male was removed from the cage and the vaginal plug was checked. Only the females that successfully formed a plug were used for further experiments. On the day of the delivery (PPD 0), typically 12–16 hours after the birth of pups, the number of pups was counted, and dead pups were removed from the cage. In the calculation of the survival rate of pups at PPD 1 (Figs 1H, 4F, 6E and 7E), the number of dead pups at PPD 0 was excluded. We did not normalize the number of pups, given that the number of pups born was not strongly correlated with the development of pups in our datasets (S1 Fig). As shown in Figs 1I, 2, 4G, 5 and 7, a dam with one or more living pups at PPD 1 was classified as a “success” in raising pups and as a “failure” if all pups were dead at PPD 1. Weight of pups were measured at PPD 3, 6, and 9, in units of 0.1 g. Of note, the genotype of pups born from OTflox/–mothers crossed with wild-type males should be either OT+/–or OTflox/+. We found that the ratio of OT+/–pups and OTflox/+ pups largely followed Mendel’s law and we did not find a statistical significance between mothers that received vehicle or AAV-Cre injection to the bilateral PVH (The ratio of OT+/–pups was 62.4 ± $10.4\%$ and 53.1 ± $8.7\%$ in vehicle and +Cre, respectively. $p \leq 0.81$, two-sided Mann–Whitney U-test. $$n = 6$$ mothers each). This observation suggests that the genotype of pups did not significantly influence our results.
To measure the pup birth intervals, 18.5 days after the formation of a vaginal plug, each pregnant female was moved to a transparent cage that contained a minimal amount of wood chips and shredded paper, with which the dam built its nest. Two cameras (Qwatch, I-O Data and DMK33UX273, The Imaging Source) were equipped under and on the side of the cage, respectively, to capture the entire cage, including the nest. Videotaping (15 frames/second for Qwatch and 4 frames/second for DMK33UX273) was started after adaptation (5 hours or longer). We defined the birth of each pup as the complete exposure of its entire body from the maternal vagina. In the calculation of the distribution of pup birth intervals, we excluded data from one pup with an interval of longer than 10 hours. Meanwhile, the dam moved freely in the cage, including foraging.
## Assay for mothers
A behavioral assay with mothers was conducted using a similar procedure as described previously [16]. In brief, a virgin female (13 weeks old) individually housed for 5 weeks or longer (Fig 2) or 2 weeks (Fig 5) after the viral injection was paired with a male. The next day, the male was removed from the cage and the vaginal plug was checked. Only the females that successfully formed a plug were used for further experiments. A behavioral assay for the mothers was conducted at PPD 3. For the mothers that successfully fed their pups, all pups were removed from the home cage 6–8 hours before the assay, leaving only the mothers. Unfamiliar wild-type pups (pups unrelated to the resident mother) were used for the assay. Although we prepared three behavioral categories (“Attack”, “Ignore”, and “Retrieve”) as defined previously [16], in our dataset, all dams showed “Retrieve”. The following behaviors were further scored: latency to investigate (time after the introduction of pups to the first investigation), pup retrieval, grooming, and crouching. Even if a dam exhibited grooming behavior during crouching, it was only measured as crouching. The duration of animals undergoing either grooming, crouching, or retrieving was scored as parental care duration.
## Assay for virgin females
Virgin females (Fig 3) were prepared using a similar procedure as described previously [21]. A virgin female (13 weeks old) individually housed for 5 weeks after the viral injection was moved to the cage that contains a resident wild-type mother rearing the PPD 1 pups. The virgin female was allowed to co-house for 6 consecutive days. One day before the assay, the virgin female was moved to a new cage. The behavioral assay with the virgin females was the same as that with mothers described above.
## In situ hybridization
Mice were anesthetized with isoflurane and perfused with phosphate-buffered saline (PBS) followed by $4\%$ paraformaldehyde (PFA) in PBS. The brain was post-fixed with $4\%$ PFA overnight. Twenty-micron coronal brain sections were obtained from the entire PVH and SO (typically 32 sections per brain) using a cryostat (Leica), and all sections were subjected to the staining and cell counting. Fluorescent in situ hybridization was performed as previously described [16, 39]. In brief, sections were treated with TSA-plus Cyanine 3 (NEL744001KT, Akoya Biosciences) or TSA-plus biotin (NEL749A001KT, Akoya Biosciences) followed by streptavidin-Alexa Fluor 488 (S32354, Invitrogen). The primers (5’– 3’) to produce RNA probes were: OT, forward, AAGGTCGGTCTGGGCCGGAGA, reverse, TAAGCCAAGCAGGCAGCAAGC, Cre, forward, CCAAGAAGAAGAGGAAGGTGTC, reverse, ATCCCCAGAAATGCCAGATTAC [16]. Brain images were acquired using an Olympus BX53 microscope equipped with a 10× (N.A. 0.4) objective lens. Cells were counted manually using the ImageJ Cell Counter plugin.
## Data analysis
All mean values are reported as mean ± standard error of the mean. The statistical details of each experiment, including the statistical tests used, the exact value of n, and what n represents, are shown in each figure legend. The p-values are shown in each figure legend or panel; nonsignificant values are not noted.
## References
1. Froemke RC, Young LJ. **Oxytocin, Neural Plasticity, and Social Behavior**. *Annu Rev Neurosci* (2021) **44** 359-81. DOI: 10.1146/annurev-neuro-102320-102847
2. Caldwell HK, Aulino EA, Freeman AR, Miller TV, Witchey SK. **Oxytocin and behavior: Lessons from knockout mice**. *Dev Neurobiol* (2017) **77** 190-201. DOI: 10.1002/dneu.22431
3. Insel TR. **The challenge of translation in social neuroscience: a review of oxytocin, vasopressin, and affiliative behavior**. *Neuron* (2010) **65** 768-79. DOI: 10.1016/j.neuron.2010.03.005
4. Nishimori K, Young LJ, Guo Q, Wang Z, Insel TR, Matzuk MM. **Oxytocin is required for nursing but is not essential for parturition or reproductive behavior**. *Proc Natl Acad Sci U S A* (1996) **93** 11699-704. DOI: 10.1073/pnas.93.21.11699
5. Tsuneoka Y, Yoshihara C, Ohnishi R, Yoshida S, Miyazawa E, Yamada M. **Oxytocin Facilitates Allomaternal Behavior under Stress in Laboratory Mice**. *eNeuro* (2022) **9**. DOI: 10.1523/ENEURO.0405-21.2022
6. Young WS, Shepard E, Amico J, Hennighausen L, Wagner KU, LaMarca ME. **Deficiency in mouse oxytocin prevents milk ejection, but not fertility or parturition**. *J Neuroendocrinol* (1996) **8** 847-53. DOI: 10.1046/j.1365-2826.1996.05266.x
7. Young LJ, Winslow JT, Wang Z, Gingrich B, Guo Q, Matzuk MM. **Gene targeting approaches to neuroendocrinology: oxytocin, maternal behavior, and affiliation**. *Horm Behav* (1997) **31** 221-31. DOI: 10.1006/hbeh.1997.1377
8. Ragnauth AK, Devidze N, Moy V, Finley K, Goodwillie A, Kow LM. **Female oxytocin gene-knockout mice, in a semi-natural environment, display exaggerated aggressive behavior**. *Genes Brain Behav* (2005) **4** 229-39. DOI: 10.1111/j.1601-183X.2005.00118.x
9. El-Brolosy MA, Kontarakis Z, Rossi A, Kuenne C, Gunther S, Fukuda N. **Genetic compensation triggered by mutant mRNA degradation**. *Nature* (2019) **568** 193-7. DOI: 10.1038/s41586-019-1064-z
10. Ma Z, Zhu P, Shi H, Guo L, Zhang Q, Chen Y. **PTC-bearing mRNA elicits a genetic compensation response via Upf3a and COMPASS components**. *Nature* (2019) **568** 259-63. DOI: 10.1038/s41586-019-1057-y
11. Leng G, Sabatier N. **Oxytocin—The Sweet Hormone?**. *Trends Endocrinol Metab* (2017) **28** 365-76. DOI: 10.1016/j.tem.2017.02.007
12. Onaka T, Takayanagi Y. **Role of oxytocin in the control of stress and food intake**. *J Neuroendocrinol* (2019) **31** e12700. DOI: 10.1111/jne.12700
13. Sutton AK, Myers MG, Olson DP. **The Role of PVH Circuits in Leptin Action and Energy Balance**. *Annu Rev Physiol* (2016) **78** 207-21. PMID: 26863324
14. Inada K, Tsujimoto K, Yoshida M, Nishimori K, Miyamichi K. **Oxytocin signaling in the posterior hypothalamus prevents hyperphagic obesity in mice**. *Elife* (2022) 11. DOI: 10.7554/eLife.75718
15. Xu S, Yang H, Menon V, Lemire AL, Wang L, Henry FE. **Behavioral state coding by molecularly defined paraventricular hypothalamic cell type ensembles**. *Science* (2020) **370**. DOI: 10.1126/science.abb2494
16. Inada K, Hagihara M, Tsujimoto K, Abe T, Konno A, Hirai H. **Plasticity of neural connections underlying oxytocin-mediated parental behaviors of male mice**. *Neuron* (2022) **110** 2009-23. DOI: 10.1016/j.neuron.2022.03.033
17. Son S, Manjila SB, Newmaster KT, Wu YT, Vanselow DJ, Ciarletta M. **Whole-Brain Wiring Diagram of Oxytocin System in Adult Mice**. *J Neurosci* (2022) **42** 5021-33. DOI: 10.1523/JNEUROSCI.0307-22.2022
18. Pedersen CA, Ascher JA, Monroe YL, Prange AJ. **Oxytocin induces maternal behavior in virgin female rats**. *Science* (1982) **216** 648-50. PMID: 7071605
19. Marlin BJ, Mitre M, D’Amour J A, Chao MV, Froemke RC. **Oxytocin enables maternal behaviour by balancing cortical inhibition**. *Nature* (2015) **520** 499-504. DOI: 10.1038/nature14402
20. Pedersen CA, Vadlamudi SV, Boccia ML, Amico JA. **Maternal behavior deficits in nulliparous oxytocin knockout mice**. *Genes Brain Behav* (2006) **5** 274-81. DOI: 10.1111/j.1601-183X.2005.00162.x
21. Carcea I, Caraballo NL, Marlin BJ, Ooyama R, Riceberg JS, Mendoza Navarro JM. **Oxytocin neurons enable social transmission of maternal behaviour**. *Nature* (2021) **596** 553-7. DOI: 10.1038/s41586-021-03814-7
22. Yang CF, Chiang MC, Gray DC, Prabhakaran M, Alvarado M, Juntti SA. **Sexually dimorphic neurons in the ventromedial hypothalamus govern mating in both sexes and aggression in males**. *Cell* (2013) **153** 896-909. DOI: 10.1016/j.cell.2013.04.017
23. Armstrong WE. **Central nervous system control of oxytocin secretion during lactation**. *Physiology of Reproduction* (2015) 527-60
24. Brunton PJ, Russell JA. **Maternal bain adaptations in pegnancy**. *Physiology of Reproduction* (2015) 1957-2026
25. Belin V, Moos F. **Paired recordings from supraoptic and paraventricular oxytocin cells in suckled rats: recruitment and synchronization**. *J Physiol* (1986) **377** 369-90. DOI: 10.1113/jphysiol.1986.sp016192
26. Belin V, Moos F, Richard P. **Synchronization of oxytocin cells in the hypothalamic paraventricular and supraoptic nuclei in suckled rats: direct proof with paired extracellular recordings**. *Exp Brain Res* (1984) **57** 201-3. DOI: 10.1007/BF00231147
27. Neumann I, Koehler E, Landgraf R, Summy-Long J. **An oxytocin receptor antagonist infused into the supraoptic nucleus attenuates intranuclear and peripheral release of oxytocin during suckling in conscious rats**. *Endocrinology* (1994) **134** 141-8. DOI: 10.1210/endo.134.1.8275928
28. Freund-Mercier MJ, Richard P. **Electrophysiological evidence for facilitatory control of oxytocin neurones by oxytocin during suckling in the rat**. *J Physiol* (1984) **352** 447-66. DOI: 10.1113/jphysiol.1984.sp015302
29. Rossoni E, Feng J, Tirozzi B, Brown D, Leng G, Moos F. **Emergent synchronous bursting of oxytocin neuronal network**. *PLoS Comput Biol* (2008) **4** e1000123. PMID: 18636098
30. Zhang B, Qiu L, Xiao W, Ni H, Chen L, Wang F. **Reconstruction of the Hypothalamo-Neurohypophysial System and Functional Dissection of Magnocellular Oxytocin Neurons in the Brain**. *Neuron* (2021) **109** 331-46. DOI: 10.1016/j.neuron.2020.10.032
31. Thirtamara Rajamani K, Leithead AB, Kim M, Barbier M, Peruggia M, Niblo K. **Efficiency of cell-type specific and generic promoters in transducing oxytocin neurons and monitoring their neural activity during lactation**. *Sci Rep* (2021) **11** 22541. DOI: 10.1038/s41598-021-01818-x
32. Yukinaga H, Hagihara M, Tsujimoto K, Chiang HL, Kato S, Kobayashi K. **Recording and manipulation of the maternal oxytocin neural activities in mice**. *Curr Biol* (2022) **32** 3821-9. DOI: 10.1016/j.cub.2022.06.083
33. Hung LW, Neuner S, Polepalli JS, Beier KT, Wright M, Walsh JJ. **Gating of social reward by oxytocin in the ventral tegmental area**. *Science* (2017) **357** 1406-11. DOI: 10.1126/science.aan4994
34. Sgritta M, Dooling SW, Buffington SA, Momin EN, Francis MB, Britton RA. **Mechanisms Underlying Microbial-Mediated Changes in Social Behavior in Mouse Models of Autism Spectrum Disorder**. *Neuron* (2019) **101** 246-59. DOI: 10.1016/j.neuron.2018.11.018
35. Schiavo JK, Valtcheva S, Bair-Marshall CJ, Song SC, Martin KA, Froemke RC. **Innate and plastic mechanisms for maternal behaviour in auditory cortex**. *Nature* (2020) **587** 426-31. DOI: 10.1038/s41586-020-2807-6
36. Dulac C O. **’Connell LA, Wu Z. Neural control of maternal and paternal behaviors**. *Science* (2014) **345** 765-70. PMID: 25124430
37. Freda SN, Priest MF, Badong D, Xiao L, Liu Y, Kozorovitskiy Y. **Brainwide input-output architecture of paraventricular oxytocin and vasopressin neurons**. *bioRxiv* (2022)
38. Lein ES, Hawrylycz MJ, Ao N, Ayres M, Bensinger A, Bernard A. **Genome-wide atlas of gene expression in the adult mouse brain**. *Nature* (2007) **445** 168-76. DOI: 10.1038/nature05453
39. Ishii KK, Osakada T, Mori H, Miyasaka N, Yoshihara Y, Miyamichi K. **A Labeled-Line Neural Circuit for Pheromone-Mediated Sexual Behaviors in Mice**. *Neuron* (2017) **95** 123-37. DOI: 10.1016/j.neuron.2017.05.038
|
---
title: Diagnostic and prognostic biomarkers for progressive fibrosing interstitial
lung disease
authors:
- Mayuko Watase
- Takao Mochimaru
- Honomi Kawase
- Hiroyuki Shinohara
- Shinobu Sagawa
- Toshiki Ikeda
- Shota Yagi
- Hiroyuki Yamamura
- Emiko Matsuyama
- Masanori Kaji
- Momoko Kurihara
- Midori Sato
- Kohei Horiuchi
- Risa Watanabe
- Shigenari Nukaga
- Kaoru Irisa
- Ryosuke Satomi
- Yoshitaka Oyamada
journal: PLOS ONE
year: 2023
pmcid: PMC10022771
doi: 10.1371/journal.pone.0283288
license: CC BY 4.0
---
# Diagnostic and prognostic biomarkers for progressive fibrosing interstitial lung disease
## Abstract
No biomarkers have been identified in bronchoalveolar lavage fluid (BALF) for predicting fibrosis progression or prognosis in progressive fibrosing interstitial lung disease (PF-ILD). We investigated BALF biomarkers for PF-ILD diagnosis and prognosis assessment. Overall, 120 patients with interstitial pneumonia who could be diagnosed with PF-ILD or non PF-ILD were enrolled in this retrospective study. PF-ILD was diagnosed according to Cottin’s definition. All patients underwent bronchoscopy and BALF collection. We evaluated blood and BALF parameters, high-resolution computed tomography (HRCT) patterns, and spirometry data to identify factors influencing PF-ILD diagnosis and prognosis. On univariate logistic analysis, age, sex, the BALF white blood cell fraction (neutrophil, lymphocyte, eosinophil, and neutrophil-to-lymphocyte ratio), BALF flow cytometric analysis (CD8), and an idiopathic pulmonary fibrosis/usual interstitial pneumonia pattern on HRCT were correlated with PF-ILD diagnosis. Multivariate logistic regression analysis revealed that sex (male), age (cut-off 62 years, area under the curve [AUC] 0.67; sensitivity 0.80; specificity 0.47), white blood cell fraction in BALF (NLR, neutrophil, and lymphocyte), and CD8 in BALF (cut-off 34.2; AUC 0.66; sensitivity, 0.74; specificity, 0.62) were independent diagnostic predictors for PF-ILD. In BALF, the NLR (cut-off 8.70, AUC 0.62; sensitivity 0.62; specificity 0.70), neutrophil count (cut-off 3.0, AUC 0.59; sensitivity 0.57; specificity 0.63), and lymphocyte count (cut-off 42.0, AUC 0.63; sensitivity 0.77; specificity 0.53) were independent diagnostic predictors. In PF-ILD patients ($$n = 77$$), lactate dehydrogenase (cut-off 275, AUC 0.69; sensitivity 0.57; specificity 0.78), Krebs von den Lungen-6 (cut-off 1,140, AUC 0.74; sensitivity 0.71; specificity 0.76), baseline forced vital capacity (FVC) (cut-off 1.75 L, AUC 0.71; sensitivity, 0.93; specificity, 0.46), and BALF neutrophil ratio (cut-off 6.0, AUC 0.72; sensitivity 0.79; specificity 0.80) correlated with death within 3 years. The BALF cellular ratio, particularly the neutrophil ratio, correlated with the diagnosis and prognosis of PF-ILD. These findings may be useful in the management of patients with interstitial pneumonia.
## Introduction
Interstitial lung disease (ILD) comprises heterogeneous subtypes of diffuse lung disorders [1, 2]. Idiopathic pulmonary fibrosis (IPF) is a typical progressive and fatal fibrotic ILD [3]. The prognosis of most non-IPF ILD cases is better than that of IPF cases; however, some non-IPF patients show rapid progression similar to those with IPF. Non-IPF ILDs with progressive fibrosing phenotypes include non-specific interstitial pneumonia, hypersensitivity pneumonia (HP), autoimmune ILDs, and sarcoidosis [4]. Recently, a subset of fibrosing ILD with a progressive course, despite conventional treatment, has been named progressive fibrosing ILD (PF-ILD) [5]. PF-ILD is characterized by worsening respiratory symptoms, a decline in lung function, and the extent of fibrosis on high-resolution computed tomography (HRCT). PF-ILD has a well-defined clinical phenotype, regardless of its cause. Mortality rates and lung function decline were found to be similar between patients with PF-ILD and those with IPF [1]. Because of early mortality, patients with PF-ILD require early diagnosis and prognostic markers to allow for implementation of precision medicine.
Although bronchoscopy is not necessary for patients with a usual interstitial pneumonia (UIP) pattern on CT scan, bronchoscopy remains necessary for patients with ILD in clinical settings [6]. Bronchoscopy is often required to rule out other conditions, such as respiratory infection. Moreover, factors in bronchoalveolar lavage fluid (BALF) have been used as markers of lower respiratory tract inflammation in many respiratory diseases.
Here, we sought to determine whether BALF contains biomarkers that may be useful for the diagnosis of PF-ILD and for predicting prognosis in these patients.
## Study population
Between March 2006 and April 2018, 532 patients underwent bronchoscopy and bronchoalveolar lavage (BAL) for disease evaluation at the Tokyo Medical Center, Tokyo, Japan (Fig 1). We collected the data of these 532 patients. Patients with malignant diseases, infectious pneumonia, or eosinophilic pneumonia were excluded. For this study, 120 patients with interstitial pneumonia who could be diagnosed with PF-ILD or non PF-ILD were enrolled in this retrospective study. PF-ILD was diagnosed according to Cottin’s definition [7], which was lung fibrosis along with one of the following criteria within 24 months of diagnosis despite receiving standard-of-care treatment: [1] relative decline in forced vital capacity (FVC) of ≥$10\%$ or relative decline of ≥$15\%$ in the diffusing capacity of the lungs for carbon monoxide (DLCO) or [2] worsening symptoms or radiological appearance, accompanied by a ≥$5\%$ but <$10\%$ relative decrease in FVC. Seventy-seven patients who met the above definition criteria were assigned to the PF-ILD group and the 43 who did not meet the definition criteria were assigned to the non PF-ILD group.
**Fig 1:** *Population flow chart of the study cohort.Among 532 patients who underwent bronchoscopy and BAL, 189 patients with diseases not relevant to the objective of this study and 223 who did not undergo a respiratory function test were excluded. Eventually, 120 patients were included in the analysis. PF-ILD: progressive fibrosing interstitial lung disease; BAL: bronchoalveolar lavage.*
Informed consent was obtained in the form of opt-out. The Ethics Committee of the Tokyo Medical Center approved the study protocol. All aspects of the study conformed to the principles of the Declaration of Helsinki.
## BALF analysis
The patients underwent BAL using the method reported by Inomata et al. [ 8]. The affected segmental bronchus was identified on chest CT scan and was lavaged two or three times using 50-ml aliquots (total volume, 100 or 150 mL) of sterile $0.9\%$ saline at room temperature through a wedged flexible fiberoptic bronchoscope. The BALF obtained was centrifuged at 1500 rpm for 5 min at 4°C to separate the supernatant from the cells. Cell pellets were counted in a hemocytometer, and Diff-Quik™ (International Reagents, Kobe, Japan)-stained smears were used to identify the differential profiles after cytospin preparation. Differential counts were performed by examining 300 cells using a standard light microscope.
The T lymphocyte subpopulations were determined using flow cytometry. The BALF samples were incubated with fluorescent monoclonal antibodies CD4 and CD8 (Beckman Coulter, Tokyo, Japan) and subsequently stained at room temperature in the dark for 15 min.
## Assessment of clinical parameters
Clinical and survival data of all patients were retrospectively collected from their medical records. Blood samples were collected within 1 month before bronchoscopy: peripheral blood fractions, biochemical tests including LDH and CRP, and markers of interstitial lung disease such as KL-6. Data on BMI and smoking status were collected during the physician’s interview within 1 month before bronchoscopy.
Since the purpose of this study was to search for predictors of PF-ILD from blood tests, BAL tests, and respiratory function tests obtained at the time of bronchoscopy, we analyzed the items which were available at the time of diagnose. In particular, the neutrophil-to-lymphocyte ratio (NLR) in blood tests is widely used in several chronic inflammatory diseases [8–17]. However, there are few reports on the usefulness of the NLR in the BALF [18].
Baseline pulmonary function tests, including FVC and DLCO, were performed up to 3 months before bronchoscopy. HRCT scans were reviewed by a specialized pulmonologist, and the overall pattern was categorized as UIP, defined according to the 2018 American Thoracic Society/ European Respiratory Society/Japanese Respiratory Society/Asociación Latinoamericana de Tórax guidelines [6].
We also analyzed the 3-year mortality in PF-ILD and the parameters related to 3-year mortality. The median survival after diagnosis of IPF is reported to be 2 or 3 years [6]. In the case of IIPs, MCTD-ILDs, RA-ILDs, and SSc-ILDs, the median survival of PF-ILD patients is reported to range from 3.1 to 3.7 years [19]. Based on the above previous reports, we selected “3 years” as the survival duration in this study.
Patients eligible for data collection are those who underwent bronchoscopy between March and April 2018. The final follow-up timepoint is April 2021, after bronchoscopy in April 2018 and a 24-month follow-up as per Cottin’s definition. Alternatively, for cases diagnosed with PF-ILD at the 24-month follow-up period, death was evaluated 3 years after bronchoscopy.
## Statistical analysis
Data are presented as median (interquartile range). Data were compared between the two groups using Student’s t-test, the Mann–Whitney U test, and the χ2 test. Univariate and multivariate logistic regression analyses were performed to assess the effects of various factors on the diagnosis and survival of patients with PF-ILD. Items that differed significantly in the univariate analysis were selected and entered into the multivariate analysis. Receiver operating characteristic (ROC) curves were constructed to assess the areas under the curve (AUCs). The optimal cut-off values for predictors were determined by maximizing the Youden index. For all tests, two-sided p-values of <0.05 were considered statistically significant. Data were analyzed using JMP v16 (SAS Institute, Cary, NC, USA) and IBM SPSS® Statistics version 27.0 (IBM SPSS Inc., Armonk, NY, USA).
## Baseline characteristics of the patients
The baseline characteristics of the study groups are shown in Table 1. Compared with the non-PF-ILD group, the PF-ILD group was older (median age 69.4 years) and was predominantly male. The PF-ILD group had a higher NLR, lower lymphocyte percentage (Lym%), lower CD8 levels, and higher CD4/CD8 ratio in the BALF. In addition, these patients showed a higher white blood cell count. No significant differences in blood fractions were observed between the groups.
**Table 1**
| Unnamed: 0 | All | PF-ILD | Non-PF-ILD | p-value |
| --- | --- | --- | --- | --- |
| | (n = 120) | (n = 77) | (n = 43) | p-value |
| Follow-up duration, days (median, IQR) | 1489.0 (1112.0–2394.0) | 1431.5 (860.8–2154.5) | 1593.0 (1224.0–2944.0) | 0.076 |
| Age, years (median, IQR) | 69 (58.5–76) | 71 (65–77) | 65 (53–72) | 0.003a |
| Male (%) | 61 (50.8) | 46 (59.7) | 15 (34.9) | 0.009b |
| BMI, kg/m2 (median, IQR) | 22.5 (20.4–25.3) | 23.4 (20.6–25.3) | 21.1 (20.1–25.3) | 0.203a |
| Smoking history (%) | 71 (59.1) | 49 (63.6) | 22 (51.2) | 0.155b |
| Smoking status (current) (%) | 11 (9.2) | 9 (11.7) | 2 (4.7) | 0.193b |
| Smoking, pack-year (median, IQR) | 30 (15–50) | 30 (15–50) | 21 (10.5–47.5) | 0.216a |
| BALF (median, IQR) | | | | |
| Recovery rate of the BALF, % | 40 (28.8–49.3) | 39 (28.0–50.0) | 42 (30.7–48.7) | 0.414a |
| Cell Count (×105/μL) | 3(2–5) | 3 (2–5) | 4 (2–7) | 0.228a |
| NLR | 9.0 (1.8–41.1) | 14.3 (2.7–50.0) | 5.4 (1.1–20.9) | 0.027a |
| MLR | 205.4 (67.8–499.7) | 285.3 (84.3–520.4) | 116.5 (39.2–313) | 0.058a |
| ELR | 4.4 (0–16) | 2.6 (0.0–21.7) | 5.3 (0.0–12.3) | 0.448a |
| BLR | 0 (0.0–0.0) | 0.0 (0.0–0.0) | 0 (0.0–0.9) | 0.729a |
| Neutrophils (%) | 2.5 (0.5–9.0) | 3.8 (0.6–10.2) | 1.5 (0.5–6.0) | 0.117a |
| Lymphocytes (%) | 25.5 (12.5–49.5) | 23.0 (11.4–41.9) | 44 (19–64) | 0.015a |
| Monocytes (%) | 54.2 (25.9–77.5) | 65.3 (26.6–78.0) | 44.5 (25.5–71.5) | 0.168a |
| Eosinophils (%) | 1.5 (0.0–4.0) | 1.0 (0.0–3.9) | 2 (0.0–4.5) | 0.074a |
| Basophils (%) | 0 (0.0–0.0) | 0.0 (0.0–0.0) | 0.0 (0.0–0.5) | 0.586a |
| CD4 | 54.9 (31.7–72.1) | 56.7 (35.3–74.1) | 42.9 (28.9–66.3) | 0.056a |
| CD8 | 28.7 (17.8–55.3) | 26.3 (16.0–36.7) | 41.7 (24.0–66.9) | 0.004a |
| CD4/CD8 ratio | 1.9 (0.6–3.6) | 2.1 (1.0–3.9) | 1.0 (0.5–2.7) | 0.021a |
| Laboratory data (median, IQR) | | | | |
| White blood cell (/μL) | 6900 (5500–9400) | 7100 (6025–9600) | 5900 (4400–8700) | 0.026a |
| Neutrophils (%) | 67.0 (57.5–74.5) | 65.0 (60–72.8) | 67.1 (56.4–76.1) | 0.667a |
| Lymphocytes (%) | 20.7 (15.3–29.9) | 23.3 (17.0–29.9) | 20.5 (14.6–29.2) | 0.437a |
| Monocytes (%) | 6.5 (5.1–8.3) | 6.3 (5.4–8.6) | 6.7 (5.4–8.6) | 0.382a |
| Eosinophils (%) | 3.4 (1.6–5.2) | 3.4 (1.6–5.5) | 0.5 (0.4–0.8) | 0.488a |
| Basophils (%) | 0.6 (0.4–0.9) | 0.5 (0.4–0.8) | 0.7 (0.4–0.9) | 0.140a |
| NLR | 3.1 (1.8–4.8) | 3.2 (1.7–5.1) | 2.8 (1.9–4.1) | 0.812a |
| MLR | 0.3 (0.2–0.5) | 0.3 (0.2–0.5) | 0.3 (0.2–0.5) | 0.730a |
| ELR | 0.1 (0.0–0.2) | 0.1 (0.1–0.2) | 0.2 (0.1–0.2) | 0.850a |
| BLR | 0.0 (0.02–0.04) | 0.0 (0.0–0.0) | 0.0 (0.0–0.0) | 0.366a |
| Platelets (×104/μL) | 25.0 (20.2–30.1) | 24.5 (19.7–29.2) | 25.5 (22.5–31.6) | 0.197a |
| CRP (mg/dl) | 0.6 (0.2–3.9) | 0.7 (0.2–4.8) | 0.5 (0.1–2.8) | 0.185a |
| LDH (U/L) | 226 (195–286) | 235 (194–291) | 222 (194–274) | 0.770a |
| KL-6 (U/ml) | 837 (531–1580) | 872 (589–1640) | 724 (418–1433) | 0.181a |
| Management, n (%) | | | | |
| Steroid treatment or immunosuppressive treatment prior to BS | 15 (12.5) | 10 (13) | 5 (11.6) | 0.829b |
| Steroid treatment prior to BS | 14 (11.7) | 10 (13.0) | 4 (9.3) | 0.547b |
| Steroid treatment within 24 months of BS | 74 (61.7) | 46 (59.7) | 28 (65.1) | 0.561b |
| Immunosuppressive treatment within 24 months of BS | 22 (18.3) | 14 (18.2) | 8 (18.6) | 0.954b |
| Antifibrotic therapy within 24 months of BS | 6 (5.0) | 6 (7.8) | 0 (0.0) | 0.060b |
Details of the ILD types that are present in the study population are shown in the S1 Table.
In the PF-ILD group, $\frac{10}{77}$ ($12.5\%$) patients were treated with steroids or immunosuppressive agents before bronchoscopy, which was similar to the proportion of patients in the non-PF-ILD group ($\frac{5}{43}$ [$11.6\%$], $$p \leq 0.829$$). Immunosuppressive treatment included mizoribine, cyclophosphamide, cyclosporine, tacrolimus, azathioprine, and methotrexate. Of the patients, $\frac{46}{77}$ ($59.7\%$) started treatment with steroids and $\frac{14}{77}$ ($18.2\%$) started immunosuppressive treatment after bronchoscopy. There were no significant differences between the PF-ILD and non-PF-ILD groups with respect to the treatment used. In contrast, treatment with antifibrotic drugs was only initiated in the PF-ILD group ($\frac{6}{77}$, $7.8\%$).
## Diagnostic factors for PF-ILD
We performed univariate logistic regression analysis to assess the predictive markers of PF-ILD (Table 2). Male sex, older age, higher BALF NLR, higher BALF neutrophil percentage (Neu%), lower BALF Lym%, lower BALF eosinophil percentage (Eos%), and lower CD8 levels were diagnostic predictors of PF-ILD. The presence of an IPF/UIP pattern on HRCT had a higher odds ratio (OR) for PF-ILD (OR 19573899, $$p \leq 0.0007$$; Table 2). Steroid therapy or immunosuppressive treatment before bronchoscopy was not predictive of PF-ILD (OR 1.006, $95\%$CI 0.314–3.219, $$p \leq 0.092$$).
**Table 2**
| Parameter | Odds ratio | 95%CI | p-value |
| --- | --- | --- | --- |
| Univariate | Univariate | Univariate | Univariate |
| Sex (male) | 2.77 | 1.28–6.01 | 0.009 |
| Age | 1.04 | 1.01–1.08 | 0.003 |
| BALF NLR | 1.01 | 1.00–1.02 | 0.009 |
| BALF Neu% | 1.04 | 1.01–1.09 | 0.01 |
| BALF Lym% | 0.98 | 0.97–1.00 | 0.01 |
| BALF Eos% | 0.91 | 0.81–1.00 | 0.04 |
| CD8 | 0.97 | 0.96–0.99 | 0.0014 |
| IPF/UIP pattern on CT scan | 19573899 | - | 0.092 |
For the predictive markers identified in the univariate analysis, we performed ROC curve analysis for continuous variables (Fig 2). Age, with a cut-off value of 62 years (0.80 sensitivity, 0.47 specificity, AUC 0.67); BALF NLR, with a cut-off value of 8.696 (0.62 sensitivity, 0.72 specificity, AUC 0.62); BALF Neu%, with a cut-off value of $3.0\%$ (0.57 sensitivity, 0.63 specificity, AUC 0.59); BALF Lym%, with a cut-off value of $42.0\%$ (0.77 sensitivity, 0.53 specificity, AUC 0.63); BALF Eos%, with a cut-off value of $0.5\%$ (0.49 sensitivity, 0.70 specificity, AUC 0.60); and BALF CD8, with a value of 34.2 (0.74 sensitivity, 0.62 specificity, AUC 0.66), were identified as diagnostic predictors for PF-ILD.
**Fig 2:** *ROC curve analysis to establish cut-off values for PF-ILD diagnostic predictors.Abbreviations: NLR: neutrophil-to-lymphocyte ratio; AUC: area under the ROC curve; ROC: receiver operating characteristic.*
We performed multivariate logistic regression analysis of three models using these predictive markers (Table 3). Cellular analysis of the BALF (NLR ≥ 8.696, Neu% ≥ $3\%$, Lym% ≤ $42\%$) and CD8 ≥ 34.2 at the time of bronchoscopy were found to be independent predictive markers of PF-ILD. Eos% was no longer significant in the multivariate analysis. The results did not differ even when the presence of the UIP pattern on HRCT was included.
**Table 3**
| Multivariate analysis model 1 | Multivariate analysis model 1.1 | Multivariate analysis model 1.2 | Multivariate analysis model 1.3 |
| --- | --- | --- | --- |
| Parameter | Odds ratio | 95%CI | p-value |
| Sex (male) | 2.017 | 0.806–5.051 | 0.134 |
| Age ≥ 62 years | 3.671 | 1.321–10.206 | 0.013 |
| BALF NLR ≥ 8.696 | 6.252 | 2.327–16.795 | 0.0002 |
| CD8 ≤ 34.2 | 4.436 | 1.715–11.474 | 0.002 |
| Multivariate analysis model 2 | Multivariate analysis model 2 | Multivariate analysis model 2 | Multivariate analysis model 2 |
| Parameter | Odds ratio | 95%CI | p-value |
| Sex (male) | 2.240 | 0.927–5.414 | 0.073 |
| Age ≥ 62 years | 2.922 | 1.138–7.502 | 0.026 |
| BALF Neutrophil% ≥ 3.0 | 3.206 | 1.295–7.934 | 0.012 |
| CD8 ≤ 34.2 | 4.286 | 1.738–10.573 | 0.002 |
| Multivariate analysis model 3 | Multivariate analysis model 3 | Multivariate analysis model 3 | Multivariate analysis model 3 |
| Parameter | Odds ratio | 95%CI | p-value |
| Sex (male) | 2.017 | 0.825–4.926 | 0.124 |
| Age ≥ 62 years | 3.197 | 1.211–8.44 | 0.019 |
| BALF Lymphocyte% ≤ 42.0 | 3.942 | 1.561–9.955 | 0.004 |
| CD8 ≤ 34.2 | 3.358 | 1.365–8.258 | 0.008 |
## Prognostic factors of 3-year mortality among patients with PF-ILD
Next, we explored the predictors of 3-year mortality in the PF-ILD group. Univariate logistic regression analysis showed that high lactate dehydrogenase (LDH), high Krebs von den Lungen-6 (KL-6), low baseline FVC on respiratory function tests, and high BALF Neu% were predictors of death within 3 years. The presence of an IPF/UIP pattern on the HRCT scan was also a predictor of 3-year mortality (Table 4).
**Table 4**
| Parameter | Odds ratio | 95%CI | p-value |
| --- | --- | --- | --- |
| Univariate | Univariate | Univariate | Univariate |
| LDH | 1.008 | 1.000–1.016 | 0.047 |
| KL-6 | 1.001 | 1.000–1.002 | 0.024 |
| FVC baseline | 3.286 | 1.161–9.297 | 0.025 |
| BALF Neu% | 1.03 | 1.003–1.059 | 0.032 |
| IPF/UIP pattern on CT scan | 11.00 | 2.272–53.266 | 0.003 |
As mentioned before, we performed ROC curve analysis for the continuous variables: LDH, KL-6, baseline FVC values, and BALF Neu% (Fig 3). Predictors of 3-year mortality in the PF-ILD group were LDH ≥ 274 U/L, KL-6 ≥ 1110 U/ml, baseline FVC ≤ 1.79 L, and BALF Neu ≥ $5.25\%$. For KL-6, FVC, and BALF Neu%, the AUC exceeded 0.7, which could be interpreted as indicating relatively high accuracy.
**Fig 3:** *ROC curve analysis of the predictor cut-off values for 3-year mortality.Abbreviations: AUC: area under the ROC curve; ROC: receiver operating characteristic.*
Multivariate logistic regression analysis using these predictors showed that BALF Neu% ≥ $5.25\%$ and baseline FVC < 1.79 L at bronchoscopy were independent predictors of 3-year mortality (BALF Neu%: OR 11.467, $95\%$CI 1.151–128.023, $$p \leq 0.038$$; FVC baseline: OR 37.052, $95\%$CI 2.193–625.929, $$p \leq 0.012$$) in the PF-ILD group (Table 5). The results did not differ even when the presence of the UIP pattern on HRCT was included in this analysis.
**Table 5**
| Multivariate analysis | Multivariate analysis.1 | Multivariate analysis.2 | Multivariate analysis.3 |
| --- | --- | --- | --- |
| Parameter | Odds ratio | 95%CI | p-value |
| BALF Neu% ≥ 6 (%) | 11.467 | 1.091–120.532 | 0.042 |
| LDH ≥ 275 (U/L) | 3.067 | 0.274–34.286 | 0.363 |
| FVC baseline ≤ 1.75 L | 3.896 | 0.437–34.735 | 0.223 |
| KL-6 ≥ 1140 (U/ml) | 37.052 | 2.193–625.929 | 0.012 |
## Discussion
The diagnosis of PF-ILD has been based on imaging findings, clinical symptoms, and respiratory function test results, and a time course investigation is required to distinguish it from ILDs other than IPFs [7]. Respiratory function tests are now less commonly performed than before due to the coronavirus disease 2019 pandemic. In this study, we sought to identify biomarkers for the diagnosis and prognosis of PF-ILD. Our study revealed predictive markers of PF-ILD using only blood tests and the BALF obtained at the bronchoscopic evaluation. As diagnostic predictors, we identified age, NLR, and CD8 levels in the BALF. As prognostic markers, we identified serum LDH, KL-6, FVC, and BALF Neu%.
NLR based on blood tests is an inexpensive and widely available marker of chronic inflammation. The usefulness of the NLR in the blood has been reported in patients with stroke [9], cancer [10], cardiovascular diseases [11], hypertension [11], sepsis [12], diabetes mellitus [13], hepatic cirrhosis [14], rheumatoid arthritis [15], chronic obstructive pulmonary disorder [17], and asthma [8]. However, there are few reports on its usefulness in the BALF [18]. In our study, the NLR in the BALF, but not that in the blood, was associated with PF-ILD diagnosis. The difference in the usefulness of the NLR between the BALF and blood suggests that the BALF reflects local inflammation, whereas the blood reflects systemic inflammation.
CD8+ T cells in the BALF were also associated with the diagnosis of PF-ILD in this study. Daniil et al. reported that increased CD8+ T lymphocytes in the lung biopsies of patients with IPF correlated with decreased lung function [20]. In a bleomycin-induced pulmonary fibrosis model, CD8+ T cells were shown to differentiate into profibrotic IL-13-producing cells [21]. These data suggested that CD8+ T cells are associated with lung fibrosis.
According to the guidelines related to IPF, lymphocyte counts in the BALF of patients with IPF are lower than those in patients with non-IPF interstitial pneumonia [22]. Takei et al. reported that the differential lymphocyte count in the BALF was a prognostic factor for acute exacerbation in patients with chronic fibrosing IIPs [23].
Kinder et al. reported that the Neu% in the BALF correlated with mortality in patients with IPF [24]. In acute exacerbation of IPF, neutrophils in the BALF are a poor prognostic factor [25]. However, neutrophil depletion in rats and mice did not protect against bleomycin-induced pulmonary fibrosis [26]. Therefore, the role of neutrophils in lung fibrosis remains unclear. Nevertheless, our study supports the significance of neutrophils in patients with PF-ILD.
Currently, for the diagnosis of PF-ILD, appropriate treatment should have been administered within 24 months of diagnosis. In our study population, approximately $60\%$ of the PF-ILD group was treated with steroids and approximately $18\%$ were treated with immunosuppressive drugs within 24 months of diagnosis (Table 1), which was similar to the numbers reported in a previous UK test and validation cohort [27].
Although bronchoscopy is not necessary for patients with a UIP pattern on CT scan, patients with IPF/UIP had a higher neutrophil percentage in their BALF [6]. Furthermore, patients with hypersensitivity pneumonitis (HP) or sarcoidosis had a higher lymphocyte percentage in their BALF [6]. S1 Table shows the analysis of BAL samples and detailed diseases. Sarcoidosis was included in the IIPs/non-IPF group. The lymphocyte count in the BALF was significantly higher in the HP group than in the IPF and IIPs groups (mean [IQR], HP group: 56.5 [32.0–77.5], IPF group: 16.35 [8.0–42.5], IIPs group: 25.5 [9.5–46.75]; p value [one-way ANOVA]: 0.014). The CD4 count in the BALF was higher in the IPF group than in the CTD-ILD group (mean [IQR], IPF group: 72.6 [63.7–75.7], CTD-ILD group: 33.3 [22.8–55.3]; p value [one-way ANOVA]: 0.010). The CD8 count in the BALF was higher in the CTD-ILD group than in the IPF and IIPs groups (mean [IQR], CTD-ILD group: 55.2 [31.5–65.2], IPF group: 21 [15.9–24.8], IIPs group: 27 [15.1–39]; p value [one-way ANOVA]: 0.005). Both CTD-ILD and HP are pathologies that may or may not progress to PF-ILD.
The purpose of our study was to search for factors that could be used to diagnose PF-ILD at the time of bronchoscopy, irrespective of the final name of the disease. Nintedanib is approved for PF-ILD. Flaherty et al. showed that the annual rate of decline in the FVC was significantly lower among patients with PF-ILD who received nintedanib than among those who received placebo [28]. In our study, six patients received antifibrotic drugs after bronchoscopy, and all were in the PF-ILD group. There may be a difference in the 3-year mortality between patients who were treated with antifibrotic drugs, including nintedanib, and those who were not. Diagnosing PF-ILD at an early stage and ensuring initiation of antifibrotic drugs at an appropriate time is important, and we hope that this study will help in this regard.
There are some potential limitations to the present study. First, the number of enrolled patients was small, and similar investigations should be performed using larger sample sizes. Second, this was a retrospective study conducted at a single center, indicating a potential risk of selection and recall biases. Third, the HRCT scans were reviewed by the person involved in the data analysis of this study; however, we have reviewed the HRCT scans prior to data analysis to try to avoid bias as much as possible.
In conclusion, this study revealed diagnostic and prognostic factors for patients with PF-ILD based on BALF, blood, and respiratory function tests. In particular, cellular analysis of BALF may be useful in the diagnosis and prognosis prediction of PF-ILD.
## References
1. Brown KK, Martinez FJ, Walsh SLF, Thannickal VJ, Prasse A, Schlenker-Herceg R. **The natural history of progressive fibrosing interstitial lung diseases**. *Eur Respir J* (2020) **55** 2000085. DOI: 10.1183/13993003.00085-2020
2. Hyldgaard C, Hilberg O, Muller A, Bendstrup E. **A cohort study of interstitial lung diseases in central Denmark**. *Respir Med* (2014) **108** 793-799. DOI: 10.1016/j.rmed.2013.09.002
3. Schmidt SL, Sundaram B, Flaherty KR. **Diagnosing fibrotic lung disease: when is high-resolution computed tomography sufficient to make a diagnosis of idiopathic pulmonary fibrosis?**. *Respirology* (2009) **14** 934-939. DOI: 10.1111/j.1440-1843.2009.01626.x
4. George PM, Spagnolo P, Kreuter M, Altinisik G, Bonifazi M, Martinez FJ. **Progressive fibrosing interstitial lung disease: clinical uncertainties, consensus, recommendations, and research priorities**. *Lancet Respir Med* (2020) **8** 925-934. DOI: 10.1016/S2213-2600(20)30355-6
5. Wells AU, Brown KK, Flaherty KR, Kolb M, Thannickal VJ. **What’s in a name? That which we call IPF, by any other name would act the same**. *Eur Respir J* (2018) **51** 1800692. DOI: 10.1183/13993003.00692-2018
6. Raghu G, Remy-Jardin M, Myers JL, Richeldi L, Ryerson CJ, Lederer DJ. **Diagnosis of idiopathic pulmonary fibrosis. An official ATS/ERS/JRS/ALAT clinical practice guideline**. *Am J Respir Crit Care Med* (2018) **198** e44-e68. DOI: 10.1164/rccm.201807-1255ST
7. Cottin V, Wollin L, Fischer A, Quaresma M, Stowasser S, Harari S. **Fibrosing interstitial lung diseases: knowns and unknowns**. *Eur Respir Rev* (2019) **28** 180100. DOI: 10.1183/16000617.0100-2018
8. Inomata T, Konno S, Nagai K, Suzuki M, Nishimura M. **Neutrophil predominance in bronchoalveolar lavage fluid is associated with disease severity and progression of HRCT findings in pulmonary Mycobacterium avium infection**. *PLoS One* (2018) **13** e0190189. DOI: 10.1371/journal.pone.0190189
9. Daniela F, Margarida M, Joana N, Rafael D, Goreti M, Nils P. **Neutrophil-to lymphocyte ratio predict cerebral edema and clinical worsening early after reperfusion therapy in stroke**. *Stroke* (2021) **52** 859-867. DOI: 10.1161/STROKEAHA.120.032130
10. Cupps MA, Cariolou M, Tzoulaki I, Aune D, Evangelou E, Berlanga-Taylor AJ. **Neutrophil to lymphocyte ratio and cancer prognosis: an umbrella review of systematic reviews and meta-analysis of observational studies**. *BMC Med* (2020) **18** 360. DOI: 10.1186/s12916-020-01817-1
11. Balta S, Celik T, Mikhailidis DP, Ozturk C, Demirkol S, Aparci M. **The relation between atherosclerosis and the neutrophil-lymphocyte ratio**. *Clin Appl Thromb Hemost* (2016) **22** 405-411. DOI: 10.1177/1076029615569568
12. Huang Z, Fu Z, Huang W, Huang K. **Prognostic value of neutrophil-to-lymphocyte ratio in sepsis: A meta-analysis**. *Am J Emerg Med* (2020) **38** 641-647. DOI: 10.1016/j.ajem.2019.10.023
13. Wan H, Wang Y, Fang S, Chen Y, Zhang W, Xia F. **Associations between the neutrophil-to-lymphocyte ratio and diabetic complications in adults with diabetes: a cross-sectional study**. *J Diabetes Res* (2020) **2020** 6219545. DOI: 10.1155/2020/6219545
14. Peng Y, Li Y, He Y, Wei Q, Xie Q, Zhang L. **The role of neutrophil to lymphocyte ratio for the assessment of liver fibrosis and cirrhosis: a systematic review**. *Expert Rev Gastroenterol Hepatol* (2018) **12** 503-513. DOI: 10.1080/17474124.2018.1463158
15. Erre GL, Paliogiannis P, Castagna F, Mangoni AA, Carru C, Passiu G. **Meta-analysis of neutrophil-to-lymphocyte and platelet-to-lymphocyte ratio in rheumatoid arthritis**. *Eur J Clin Invest* (2019) **49** e13037. DOI: 10.1111/eci.13037
16. Sakurai K, Chubachi S, Irie H, Tsutsumi A, Kameyama N, Kamatani T. **Clinical utility of blood neutrophil-lymphocyte ratio in Japanese COPD patients**. *BMC Pulm Med* (2018) **18** 65. DOI: 10.1186/s12890-018-0639-z
17. Mochimaru T, Ueda S, Suzuki Y, Asano K, Fukunaga K. **Neutrophil-to-lymphocyte ratio as a novel independent predictor of severe exacerbation in patients with asthma**. *Ann Allergy Asthma Immunol* (2019) **122** 337-339. DOI: 10.1016/j.anai.2018.11.029
18. D’Alessandro M, Bergantini L, Carleo A, Cameli P, Perrone A, Fossi A. **Neutrophil-to-lymphocyte ratio in bronchoalveolar lavage from IPF patients: a novel prognostic biomarker?**. *Minerva Med* (2022) **113** 526-531. DOI: 10.23736/S0026-4806.20.06614-8
19. Nasser M, Larrieu S, Boussel L, Si-Mohamed S, Bazin Fabienne, Marque S. **Estimates of epidemiology, mortality and disease burden associated with progressive fibrosing interstitial lung disease in France (the PROGRESS study)**. *Respir Res* (2021) **22** 162. DOI: 10.1186/s12931-021-01749-1
20. Daniil Z, Kitsanta P, Kapotsis G, Mathioudaki M, Kollintza A, Karatza M. **CD8+ T lymphocytes in lung tissue from patients with idiopathic pulmonary fibrosis**. *Respir Res* (2005) **6** 81. DOI: 10.1186/1465-9921-6-81
21. Brodeur TY, Robidoux TE, Weinstein JS, Craft J, Swain SL, Marshak-Rothstein A. **IL-21 promotes pulmonary fibrosis through the induction of profibrotic CD8+ T cells**. *J Immunol* (2015) **195** 5251-5260. DOI: 10.4049/jimmunol.1500777
22. Jooae Choe, Kwon BS Do K-H, Hwang HS Song JW, Chae EJ. **Diagnostic and prognostic implications of 2018 guideline for the diagnosis of idiopathic pulmonary fibrosis in clinical practice**. *Sci Rep* (2021) **11** 16481. DOI: 10.1038/s41598-021-95728-7
23. Takei R, Arita M, Kumagai S, Ito Y, Noyama M, Tokioka F. **Impact of lymphocyte differential count >15% in BALF on the mortality of patients with acute exacerbation of chronic fibrosing idiopathic interstitial pneumonia**. *BMC Pulm Med* (2017) **17** 67. DOI: 10.1186/s12890-017-0412-8
24. Kinder BW, Brown KK, Schwarz MI, Ix JH, Kervitsky A, King TE. **Baseline Bal neutrophilia predicts early mortality in idiopathic pulmonary fibrosis**. *Chest* (2008) **133** 226-232. DOI: 10.1378/chest.07-1948
25. Kono M, Miyashita K, Hirama R, Oshima Y, Takeda K, Mochizuka Y. **Prognostic significance of bronchoalveolar lavage cellular analysis in patients with acute exacerbation of interstitial lung disease**. *Respir Med* (2021) **186** 106534. DOI: 10.1016/j.rmed.2021.106534
26. Gregory AD, Kliment CR, Metz HE, Kim KH, Kargl J, Agostini BA. **Neutrophil elastase promotes myofibroblast differentiation in lung fibrosis**. *J Leukoc Biol* (2015) **98** 143-152. DOI: 10.1189/jlb.3HI1014-493R
27. Oldham JM, Lee CT, Wu Z, Bowman WS, Pugashetti JV, Dao N. **Lung function trajectory in progressive fibrosing interstitial lung disease**. *Eur Respir J* (2022) **59** 2101396. DOI: 10.1183/13993003.01396-2021
28. Flaherty KR, Wells AU, Cottin V, Deveraj A, Walsh SLF, Inoue Y. **Nintedanib in progressive fibrosing interstitial lung disease**. *N Eng J Med* (2019) **381** 1718-1727. DOI: 10.1056/NEJMoa1908681
|
---
title: 'Psychometric properties of an innovative smartphone application to investigate
the daily impact of hypoglycemia in people with type 1 or type 2 diabetes: The Hypo-METRICS
app'
authors:
- Uffe Søholm
- Melanie Broadley
- Natalie Zaremba
- Patrick Divilly
- Giesje Nefs
- Jill Carlton
- Julia K. Mader
- Petra Martina Baumann
- Mikel Gomes
- Gilberte Martine-Edith
- Daniel J. Pollard
- Dajana Rath
- Simon Heller
- Ulrik Pedersen-Bjergaard
- Rory J. McCrimmon
- Eric Renard
- Mark Evans
- Bastiaan de Galan
- Thomas Forkmann
- Stephanie A. Amiel
- Christel Hendrieckx
- Jane Speight
- Pratik Choudhary
- Frans Pouwer
journal: PLOS ONE
year: 2023
pmcid: PMC10022775
doi: 10.1371/journal.pone.0283148
license: CC BY 4.0
---
# Psychometric properties of an innovative smartphone application to investigate the daily impact of hypoglycemia in people with type 1 or type 2 diabetes: The Hypo-METRICS app
## Abstract
### Introduction
The aim of this study was to determine the acceptability and psychometric properties of the Hypo-METRICS (Hypoglycemia MEasurement, ThResholds and ImpaCtS) application (app): a novel tool designed to assess the direct impact of symptomatic and asymptomatic hypoglycemia on daily functioning in people with insulin-treated diabetes.
### Materials and methods
100 adults with type 1 diabetes mellitus (T1DM, $$n = 64$$) or insulin-treated type 2 diabetes mellitus (T2DM, $$n = 36$$) completed three daily ‘check-ins’ (morning, afternoon and evening) via the Hypo-METRICs app across 10 weeks, to respond to 29 unique questions about their subjective daily functioning. Questions addressed sleep quality, energy level, mood, affect, cognitive functioning, fear of hypoglycemia and hyperglycemia, social functioning, and work/productivity. Completion rates, structural validity, internal consistency, and test-retest reliability were explored. App responses were correlated with validated person-reported outcome measures to investigate convergent (rs>±0.3) and divergent (rs<±0.3) validity.
### Results
Participants’ mean±SD age was 54±16 years, diabetes duration was 23±13 years, and most recent HbA1c was 56.6±9.8 mmol/mol. Participants submitted mean±SD 191±16 out of 210 possible ‘check-ins’ ($91\%$). Structural validity was confirmed with multi-level confirmatory factor analysis showing good model fit on the adjusted model (Comparative Fit Index >0.95, Root-Mean-Square Error of Approximation <0.06, Standardized Root-Mean-square Residual<0.08). Scales had satisfactory internal consistency (all ω≥0.5), and high test-retest reliability (rs≥0.7). Convergent and divergent validity were demonstrated for most scales.
### Conclusion
High completion rates and satisfactory psychometric properties demonstrated that the Hypo-METRICS app is acceptable to adults with T1DM and T2DM, and a reliable and valid tool to explore the daily impact of hypoglycemia.
## Introduction
Hypoglycemia remains a daily threat for most people with type 1 diabetes (T1DM) or insulin-treated type 2 diabetes (T2DM). Hypoglycemia impacts on many areas of daily life including sleep duration and quality, mood, cognition, and productivity, and can in extreme cases lead to coma or even death [1–5]. Both the experience of hypoglycemia, and living with the risk of hypoglycemia (including prevention and treatment) can present a significant burden for people with diabetes [6].
Previous studies on the impact of hypoglycemia have important limitations. These include recall bias for retrospective measures [7], low ecological validity for hospital-based studies [8], limited insight into the impact of asymptomatic episodes detected by continuous glucose monitoring (CGM) [9], and scarce investigation of impact beyond health status and fear of hypoglycemia [10,11]. The Hypo-METRICS (Hypoglycemia MEasurement, ThResholds and ImpaCtS) study attempts to address these limitations and further our understanding of hypoglycemia in its different forms, i.e. symptomatic and asymptomatic, severe and self-treated, and while awake and during sleep [12]. To capture data temporally closer to the hypoglycemic episodes as they occur in real-life, an ecological momentary assessment smartphone application (Hypo-METRICS app) in which experiences are repeatedly captured in real-time and in a usual environment, has been developed. The app prompts participants (morning, afternoon, and evening) to respond to questions within a few hours of any hypoglycemic episode occurring in their everyday lives.
To investigate whether broad use of the Hypo-METRICS app in research and practice is indicated, this study investigates the app’s acceptability (completion rates) and psychometric characteristics (structural validity, internal consistency reliability, test-retest reliability, convergent and divergent validity). These investigations are essential instrument validation steps, that are required before we can address the Hypo-METRICS study key objectives (outlined in Divilly et al [13]) in future publications.
## Study participants & procedure
Hypo-METRICS is part of the EU IMI2-funded Hypo-RESOLVE (Hypoglycemia–Redefining SOLutions for better liVEs) project [12,13], across five European countries (Austria, Denmark, France, the Netherlands, and the United Kingdom). The Hypo-METRICS clinical study has received ethical approval at the lead site from the South Central Oxford B Research Ethics Committee (20/SC/0112) and in the other European countries (Ethikkommission der Medizinischen Universität Graz (Austria), Videnskabsetisk Komite for Region Hovedstaden (Denmark), Comité De Protection Des Personnes SUD Mediterranne IV (France), and Commissie Mensgebonden Onderzoek Regio Arnhem-Nijmegen (the Netherlands)).
Eligible participants were aged 18–85 years, had experienced at least one self-reported hypoglycemic episode in the past 3 months, and were willing to complete the app three times per day and wear a blinded CGM for 10 weeks. Informed written consent was obtained from all participants. Participants from the following three groups were recruited: a) T1DM with intact awareness of hypoglycemia (Gold Score <4 [14]), b) T1DM with impaired awareness of hypoglycemia (Gold score ≥4 [14]), and c) T2DM managed with ≥1 insulin injection per day. The sample for the present study was determined a priori and consists of the first 100 participants to complete the Hypo-METRICS study. This sample size was based on guidelines suggesting an item-to-participant ratio of 1:10 [15] for evaluation of structural validity, and on the requirement to confirm the conceptual model fit before analyses could proceed for central study objectives. Before and after the 10-week period, participants completed an online survey (via Qualtrics, Provo, UT) with person-reported outcome measures (PROMs) and demographic and clinical characteristics were collected.
## The Hypo-METRICS app
The app consists of seven modules with 29 unique questions (see S1 Table), most of which use interval response scales (range 0–10, e.g., “not at all [0]” to “extremely [10]”). The app questions were designed to examine the impact of hypoglycemia on various domains of daily functioning, across three “check-ins”: morning, afternoon, and evening (S1 Table). There was a “Skip question” option for each question. The app was developed in English (UK) and translated into Danish, Dutch, French, and German following the ISPOR guidance [16]. Further details about the app design and development are published elsewhere [17].
## Additional PROMs
Validated PROMs were included for the purpose of validating the Hypo-METRICS app. Two of these were also completed via the app platform on a weekly basis, while the remaining were completed via Qualtrics (Qualtrics, Provo, UT) at baseline and end of study (see Table 1). PROMs were selected to examine constructs (e.g., ‘mood’ or ‘cognitive functioning’) either similar to those measured by the app for convergent validity, or dissimilar, for divergent validity. Moderate-to-high correlations (rs>±0.3) were expected to evidence convergent validity, and low or no correlations (rs<±0.3) were expected for divergent validity.
**Table 1**
| Construct measured | Patient reported outcome measure. |
| --- | --- |
| Sleep quality / sleep disturbance | 1Patient-Reported Outcomes Measurement Information System (PROMIS)—Sleep Disturbance–Short Form 8b [18] |
| Depression | Patient Health Questionnaire– 9 (PHQ-9) [19] |
| Anxiety | General Anxiety Disorder-7 (GAD-7) [20] |
| Vitality | Vitality subscale SF-36 [21] |
| Cognitive functioning | Perceived Deficit Questionnaire (PDQ-20) [22] |
| Fear of hypoglycaemia | Hypoglycaemic Fear Survey II (HFS-II) [23] |
| Diabetes Distress | Problem Areas In Diabetes (PAID-20) [24] |
| Diabetes specific Quality of life | Dawn Impact of Diabetes Profile (DIDP) [25] |
| Work and productivity | 1Work Productivity and Activity Impairment Questionnaire: Specific Health Problem (WPAI:SHP) [26] |
## Demographic & clinical data
At the start of the study, age, gender, employment status, education, medical history, previous episodes of hypoglycemia, HbA1c and method of glucose monitoring, were recorded by the study site personnel and entered into an electronic database (REDCAP, Vanderbilt, USA).
## Statistical analysis
Statistical analyses were conducted with R Studio [27]. Descriptive statistics were used to determine sample characteristics, completion rates, distribution of the data, and floor and ceiling effects. In case of non-normality of question responses, non-parametric tests (Spearman’s rho, rs) were applied. Between-person variance was examined using the Intraclass Correlation Coefficient (ICC) [28] and day-to-day variability in scores was examined using the Root Mean Squared Successive Difference (RMSSD) [29]. To examine structural validity, a multi-level confirmatory factor analysis (MCFA) was conducted. The following indices and values were used as indication of good global model fit: comparative fit index (CFI) >0.95, Tucker Lewis index (TLI) >0.95, the Standardized Root-Mean-square Residual (SRMR) <0.08 and Root-Mean-Square Error of Approximation (RMSEA) <0.06 [30,31]. Items were included in the MCFA if they: 1) were asked every day irrespective of whether the participant experienced a hypoglycemic episode, and 2) were not part of the work and productivity module of the app, where most items are relevant only to participants engaged in paid work. With 100 participants and a maximum of 10 unique items per check-in, a 1:10 item to participant ratio was considered acceptable for conducting factor analysis [15]. Internal consistency reliability of the scales were calculated with use of McDonald’s ω with scores rs>0.7 considered to indicate satisfactory internal consistency [32]. Question responses were analyzed according to check-in time: morning, afternoon or evening. These were analyzed separately to account for potential variation in latent factors at different timepoints of the day (e.g., fear of hypoglycemia in the daytime could be different to fear of hypoglycemia in the night-time). Finally, convergent and divergent validity were investigated by correlating question and scale scores with validated PROMs and participant characteristics. For more details on statistical analyses see (S1 Text).
## Participant characteristics
The first 100 participants who completed the Hypo-METRICS study included 64 adults with T1DM and 36 adults with T2DM, from the United Kingdom ($57\%$) or the Netherlands ($43\%$). Table 2 presents the participant characteristics. Overall, participants’ [M±SD] age was 54±16 years, diabetes duration was 23±13 years and most recent HbA1c was 56.6±9.8mmol/mol. For T1DM and T2DM respectively, $22\%$ and $17\%$ had impaired awareness of hypoglycemia (Gold score ≥4), $13\%$ and $8\%$ had moderate-severe anxiety, and $13\%$ and $17\%$ had moderate-severe depression as determined by the GAD-7 and PHQ-9.
**Table 2**
| Demographic and clinical characteristics | Type 1 diabetes (n = 64) | Type 2 diabetes (n = 36) |
| --- | --- | --- |
| Age,years (range)*** | 47 ± 15 (21–80) | 66 ± 9 (43–79) |
| Women | 27 (42%) | 17 (47%) |
| Employment*** | | |
| Full time employment | 32 (50%) | 6 (17%) |
| Part time employment | 12 (19%) | 3 (8.3%) |
| Full time education | 2 (3.1%) | 0 (0%) |
| Unemployed but actively looking for work | 5 (7.8%) | 0 (0%) |
| Unemployed but not actively looking for work | 2 (3.1%) | 4 (11%) |
| Retired | 11 (17%) | 23 (64%) |
| Highest level of education achieved*** | | |
| Primary school | 0 (0%) | 1 (2.8%) |
| Secondary school / high school | 14 (22%) | 14 (39%) |
| College / undergraduate degree | 29 (45%) | 11 (31%) |
| Post graduate degree | 18 (28%) | 1 (2.8%) |
| Other | 3 (4.7%) | 9 (25%) |
| Diabetes duration (years) | 22.95 ± 14.82 | 22.06 ± 9.71 |
| HbA1c (baseline) | | |
| % | 7.32 ± 0.84 | 7.35 ± 1.01 |
| Mmol/mol | 56.53 ± 9.18 | 56.85 ± 11.04 |
| Percentage time below 70mg/dL*** | 6.10 (4.58) | 2.06 (2.01) |
| Microvascular complications, any | 18 (28%) | 17 (47%) |
| Microvascular complications, any* | 10 (16%) | 13 (36%) |
| Usual method of glucose monitoring | | |
| Capillary glucose monitoring only (fingerprick)* | 20 (31%) | 19 (53%) |
| Continuous glucose monitoring without alerts** | 48 (75%) | 17 (47%) |
| Continuous glucose monitoring with alerts | 1 (1.6%) | 0 (0%) |
| Country*** | | |
| United Kingdom | 45 (70%) | 12 (33%) |
| The Netherlands | 19 (30%) | 24 (67%) |
| Gold score | | |
| Impaired awareness (≥4) | 14 (22%) | 6 (17%) |
| Intact awareness (<4) | 49 (78%) | 30 (83%) |
| Missing | 1 | 0 |
| Psychosocial characteristics | | |
| Anxiety symptoms (GAD-7)No anxiety (<5) | 46 (73%) | 28 (78%) |
| Mild anxiety (5–10) | 9 (14%) | 5 (14%) |
| Moderate-Severe anxiety (≥10) | 8 (13%) | 3 (8.3%) |
| Missing | 1 | 0 |
| Depression symptoms (PHQ-9) | | |
| No depression (<5) | 44 (70%) | 27 (75%) |
| Mild depression (5–10) | 11 (17%) | 3 (8.3%) |
| Moderate-Severe depression (≥10) | 8 (13%) | 6 (17%) |
| Missing | 1 | 0 |
| Cognitive functioning (PDQ-20)1 | 18.62 ± 13.44 (n = 63) | 16.51 ± 10.72 |
| Diabetes-specific quality of life (DIDP)2 | | |
| Composite score | 4.48 ± 0.81 | 4.39 ± 0.83 |
| Percentage score | 49.72 ± 11.63 | 48.44 ± 11.88 |
| Missing | 1 | 0 |
| Fear of hypoglycemia (HFS-II total)3 ** | 32.98 ± 22.17 (n = 63) | 21.22 ± 14.72 |
| Sleep-quality score (T-score PROMIS week 10)4 | 49.77 ± 8.85 (n = 58) | 50.49 ± 8.68 |
| Vitality (SF-36 vitality subscale mean)5 | 3.35 ± 0.83 (n = 63) | 3.37 ± 0.63 |
| Diabetes distress (PAID total)6 | 21.08 ± 17.27 (n = 60) | 17.89 ± 15.81 (n = 63) |
| Below 40 | 54 (86%) | 33 (92%) |
| Above 40 | 9 (14%) | 3 (8.3%) |
| Missing | 1 | 0 |
## Completion rates and acceptability
Participants completed 191±16 of 210 possible check-ins, a completion rate of $91\%$. Slight differences in completion between the morning, afternoon and evening check-ins were observed with completion rates of $90\%$, $89\%$ and $94\%$, respectively. When a check-in was submitted, all questions (except the work/productivity questions) were completed more than $99\%$ of the time (Table 3). One question (“How well did you get along with other people today?”), was skipped marginally more frequently ($0.9\%$) than other questions. Questions in the work and productivity section were frequently skipped (range: 36–$47\%$). Inspection of histograms revealed floor effects as indicated by more than $15\%$ (range: 15–$28\%$) of responses on the lowest score for the four negatively phrased questions (“*How anxious* do you feel right now?”, “ How irritable do you feel right now?”, “ How worried are you about having a hypo later today/while asleep?” and “How worried are you about having high blood glucose later today/while asleep?” across the three check-ins).
**Table 3**
| Latent factors | Latent factors.1 | Morning questions | Mean | SD | Skipped n (%)1 | ICC2 | RMSSD3 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Sleep quality | Sleep quality | 1. How did you sleep? | 7.03 | 1.79 | 8 (0.13) | 0.43 | 1.76 ± 0.74 (0) |
| Sleep quality | Sleep quality | 2. When you woke up, how did you feel? | 6.45 | 2.02 | 7 (0.11) | 0.63 | 1.62 ± 0.67 (0) |
| Overall mood | Overall mood | 3. How is your mood right now? | 7.06 | 1.65 | 10 (0.16) | 0.56 | 1.41 ± 0.62 (1) |
| Negative affect | Negative affect | 4. How irritable do you feel right now?* | 8.20 | 1.89 | 12 (0.19) | 0.58 | 1.51 ± 0.85 (2) |
| Negative affect | Negative affect | 5. How anxious do you feel right now?* | 8.23 | 2.01 | 13 (0.21) | 0.64 | 1.34 ± 1.00 (13) |
| Energy level | Energy level | 6. How is your energy level right now? | 6.61 | 1.86 | 7 (0.11) | 0.64 | 1.43 ± 0.66 (2) |
| Cognitive functioning | Cognitive functioning | 7. How alert do you feel right now? | 6.87 | 1.79 | 8 (0.13) | 0.64 | 1.33 ± 0.7 (2) |
| Cognitive functioning | Cognitive functioning | 8. How well are you able to concentrate right now? | 7.15 | 1.63 | 9 (0.14) | 0.65 | 1.22 ± 0.58 (1) |
| Fear of hypoglycemia | Fear of hypoglycemia | 9. How worried are you about having a hypo later today?* | 7.59 | 2.30 | 5 (0.08) | 0.83 | 1.06 ± 0.80 (14) |
| Fear of hyperglycemia | Fear of hyperglycemia | 10. How worried are you about having high blood glucose later today?* | 6.64 | 2.84 | 3 (0.05) | 0.85 | 1.29 ± 0.89 (10) |
| | | Afternoon questions | Mean | SD | Skipped n (%) 1 | ICC 2 | RMSSD 3 |
| Overall mood | Overall mood | 1. How is your mood right now? | 7.33 | 1.55 | 4 (0.06) | 0.52 | 1.37 ± 0.66 (1) |
| Negative affect | Negative affect | 2. How irritable do you feel right now?* | 8.21 | 1.88 | 6 (0.10) | 0.54 | 1.59 ± 0.85 (2) |
| Negative affect | Negative affect | 3. How anxious do you feel right now?* | 8.25 | 2.01 | 10 (0.16) | 0.58 | 1.52 ± 1.02 (10) |
| Energy level | Energy level | 4. How is your energy level right now? | 6.72 | 1.76 | 6 (0.10) | 0.57 | 1.47 ± 0.68 (1) |
| Cognitive functioning | Cognitive functioning | 5. How alert do you feel right now? | 7.04 | 1.68 | 5 (0.08) | 0.59 | 1.35 ± 0.68 (1) |
| Cognitive functioning | Cognitive functioning | 6. How well are you able to concentrate right now? | 7.23 | 1.63 | 3 (0.05) | 0.60 | 1.29 ± 0.64 (1) |
| Fear of hypoglycemia | Fear of hypoglycemia | 7. How worried are you about having a hypo later today?* | 7.61 | 2.34 | 2 (0.03) | 0.82 | 1.18 ± 0.80 (9) |
| Fear of hyperglycemia | Fear of hyperglycemia | 8. How worried are you about having high blood glucose later today?* | 6.61 | 2.88 | 1 (0.02) | 0.85 | 1.30 ± 0.74 (7) |
| | | Evening questions | Mean | SD | Skipped n (%) 1 | ICC 2 | RMSSD 3 |
| Overall mood | Overall mood | 1. How is your mood right now? | 7.14 | 1.60 | 9 (0.14) | 0.52 | 1.39 ± 0.69 (1) |
| Negative affect | Negative affect | 2. How irritable do you feel right now?* | 8.20 | 1.90 | 8 (0.12) | 0.55 | 1.56 ± 0.92 (4) |
| Negative affect | Negative affect | 3. How anxious do you feel right now?* | 8.26 | 1.98 | 10 (0.15) | 0.61 | 1.47 ± 0.95 (9) |
| Energy level | Energy level | 4. How is your energy level right now? | 6.23 | 1.83 | 5 (0.08) | 0.61 | 1.47 ± 0.71 (1) |
| Cognitive functioning | Cognitive functioning | 5. How alert do you feel right now? | 6.54 | 1.82 | 4 (0.06) | 0.64 | 1.36 ± 0.76 (1) |
| Cognitive functioning | Cognitive functioning | 6. How well are you able to concentrate right now? | 6.97 | 1.70 | 4 (0.06) | 0.64 | 1.29 ± 0.67 (2) |
| Memory (today) | Memory (today) | 7. How easy was it for you to remember things today? | 7.43 | 1.56 | 6 (0.09) | 0.66 | 1.15 ± 0.62 (1) |
| Fear of hypoglycemia while asleep | Fear of hypoglycemia while asleep | 8. How worried are you about having a hypo while asleep?* | 7.38 | 2.54 | 1 (0.02) | 0.83 | 1.20 ± 0.87 (11) |
| Fear of Hyperglycemia while asleep | Fear of Hyperglycemia while asleep | 9. How worried are you about having high blood glucose while asleep?* | 6.60 | 2.88 | 1 (0.02) | 0.84 | 1.31 ± 0.97 (9) |
| Social functioning | Social functioning | 10. How well did you get along with other people today? | 7.80 | 1.47 | 58 (0.88) | 0.62 | 1.10 ± 0.61 (2) |
| | | Work and productivity questions | Mean | SD | Skipped n (%) 1 | ICC 2 | RMSSD 3 |
| Work and productivity4 | Work and productivity4 | 1. How many hours did you work today? | 4.25 | 3.83 | 2748 (41.72) | 0.25 | 3.01 ± 2.31 (22) |
| Work and productivity4 | Work and productivity4 | 2. How many hours did you miss from work for ANY reason today? [this includes health issues, vacation, holiday, etc.] | 0.51 | 1.77 | 2899 (44.02) | 0.15 | 0.99 ± 1.50 (40) |
| Work and productivity4 | Work and productivity4 | 3. How many hours did you miss from activities other than work today for ANY reason (e.g. study, housework, shopping, family or leisure activities)? | 0.33 | 1.16 | 2348 (35.65) | 0.51 | 0.68 ± 0.87 (33) |
| Work and productivity4 | Work and productivity4 | 4. How productive were you while working today? | 7.07 | 2.10 | 3079 (46.75) | 0.45 | 1.21 ± 1.11 (22) |
## Structural validity and internal consistency reliability
Based on kurtosis values and histograms, data were considered non-normally distributed. The ICC ranged from 0.15 to 0.85 and all but two questions (both relating to sleep) included at least one participant with zero variability across the study period (RMSSD = 0) (Table 3). Exploring the data further revealed that one participant had zero variability across all questions (with the exception of the two sleep questions), but their data were still included in further analyses as removing it did not change conclusions. Of the non-work related questions, the negatively phrased questions had the least day-to-day variability. Inter-question correlations were acceptable (rs = 0.20–0.81) and multicollinearity was absent (determinant range 0.000922–0.0134 across the three check-ins). Kaiser-Meyer-Olkin values ranged from 0.76–0.93 across all questions suggesting good factorability.
Applying the five-step approach by Huang [33] to the morning check-in suggested that a MCFA was appropriate (see S2 Table). The first model, based on the conceptual framework (model A in S3 Table), had good model fit on several model fit indices, however, at the between-person level, the SRMR>0.8 (morning and afternoon) indicated that model fit could be further improved. Inspection of correlation residuals (as well as modification indices) showed a large residual between the irritability and anxiety questions at both levels, suggesting that a relationship between these two was not captured in the original model. Combining the two into one scale improved (i.e., decreased) the between-SRMR model fit parameter across all three check-ins (model B in S3 Table). The new scale was labelled ‘Negative affect’ (and the single-question mood scale was labelled ‘Overall mood’). Inspection of the internal consistencies (ω) of model B showed that ω was low for the fear of hypo-/hyperglycemia factor on the within-person level, with values ranging between 0.19–0.30. Therefore, it was decided not to combine these two questions in the same factor. This led to the final model (model C in S3 Table), which showed good model fit on several fit indices across the check-ins (CFI>0.95, RMSEA<0.06, SRMR<0.08). In model C, standardized between-person factor loadings for all questions were >0.7 (Table 4), indicating that factors explained the grouping of the questions well, while at the within-person level, the majority of loadings for the two-question ‘Negative affect’ scale (across the three check-ins) were <0.7. There was a similar pattern across other scales, with satisfactory internal consistency (ω>0.7) at the between-person level but slightly lower (ω>0.5) for the ‘Negative affect’ and ‘Cognitive functioning’ scales at the within-person level (Table 4).
**Table 4**
| Latent variable | Unnamed: 1 | Std. factor loadings | Std. factor loadings.1 | Internal Consistency (ω) | Internal Consistency (ω).1 |
| --- | --- | --- | --- | --- | --- |
| Latent variable | | Within | Between | Within | Between |
| | Morning questions | | | | |
| Sleep quality | How did you sleep? | 0.74 | 0.87 | 0.77 | 0.91 |
| Sleep quality | When you woke up, how did you feel? | 0.85 | 0.94 | 0.77 | 0.91 |
| Overall mood | How is your mood right now? | 1.00 | 1.00 | | |
| Negative affect | How anxious do you feel right now? | 0.55 | 0.92 | 0.55 | 0.93 |
| Negative affect | How irritable do you feel right now? | 0.68 | 0.95 | 0.55 | 0.93 |
| Energy level | How is your energy level right now? | 1.00 | 1.00 | | |
| Cognitive functioning | How alert do you feel right now? | 0.80 | 0.93 | 0.78 | 0.95 |
| Cognitive functioning | How well are you able to concentrate right now? | 0.80 | 0.98 | 0.78 | 0.95 |
| Fear of hypoglycemia | How worried are you about having a hypo later today? | 1.00 | 1.00 | | |
| Fear of hyperglycemia | How worried are you about having high blood glucose later today? | 1.00 | 1.00 | | |
| | Afternoon questions | | | | |
| Overall mood | How is your mood right now? | 1.00 | 1.00 | | |
| Negative affect | How anxious do you feel right now? | 0.52 | 0.94 | 0.53 | 0.93 |
| Negative affect | How irritable do you feel right now?* | 0.68 | 0.92 | 0.53 | 0.93 |
| Energy level | How is your energy level right now? | 1.00 | 1.00 | | |
| Cognitive functioning | How alert do you feel right now? | 0.80 | 0.89 | 0.75 | 0.95 |
| Cognitive functioning | How well are you able to concentrate right now? | 0.74 | 1.01 | 0.75 | 0.95 |
| Fear of hypoglycemia | How worried are you about having a hypo later today? | 1.00 | 1.00 | | |
| Fear of hyperglycemia | How worried are you about having high blood glucose later today? | 1.00 | 1.00 | | |
| | Evening questions | | | | |
| Overall mood | How is your mood right now? | 1.00 | 1.00 | | |
| Negative affect | How anxious do you feel right now? | 0.52 | 0.91 | 0.54 | 0.93 |
| Negative affect | How irritable do you feel right now? | 0.70 | 0.96 | 0.54 | 0.93 |
| Energy level | How is your energy level right now? | 1.00 | 1.00 | | |
| Cognitive functioning | How alert do you feel right now? | 0.71 | 0.86 | 0.66 | 0.92 |
| Cognitive functioning | How well are you able to concentrate right now? | 0.70 | 0.99 | 0.66 | 0.92 |
| Memory (today) | How easy was it for you to remember things today? | 1.00 | 1.00 | | |
| Fear of hypoglycemia while asleep | How worried are you about having a hypo while asleep? | 1.00 | 1.00 | | |
| Fear of hyperglycemia while asleep | How worried are you about having high blood glucose while asleep? | 1.00 | 1.00 | | |
| Social functioning | How well did you get along with other people today? | 1.00 | 1.00 | | |
## Test-retest reliability, convergent and divergent validity
Test-retest reliability and convergent and divergent validity were explored for each scale from model C. High test-retest correlations ($r = 0.76$–0.94) were found across all two-questions and single-question scales (Table 5). For convergent validity, the hypothesized pattern of correlations with PROMs was largely supported (Table 5), except for ‘Energy level’ (morning), ‘Cognitive functioning’ (morning, afternoon), ‘Memory’ (evening), ‘Fear of hyperglycemia while asleep’ (evening) and ‘Social functioning’ (evening). On the other hand, the PROMs measuring vitality and cognitive functioning did correlate highest with the respective app scales (‘Energy level’ and ‘Cognitive functioning’) compared to all other app scales (i.e., when reading vertically down the Table 5 columns). Divergent validity was evidenced by the lowest correlations between all app scales and the ‘Financial situation (DIDP)’ question, ‘HbA1c’ and ‘Diabetes duration’ (included solely for expected low correlations). However, many of the correlations in the remaining dark grey boxes in Table 5 (indicating other correlations that were expected to be low) were above rs>±0.3 (e.g., in the morning between ‘Overall mood’ and ‘Cognitive functioning’, ‘Negative affect’ and ‘Vitality’). The validated weekly work-productivity questionnaire (see S4 Table), showed a strong correlation (r>0.5) with ‘number of hours worked’, a moderate correlation (r>0.3) with ‘productivity’ questions, and low correlations (r<0.3) with hours missed from work and activities other than work on the app. Fig 1 provides an overview of the overall domains of daily functioning that the Hypo-METRICS app is believed to assess based on the psychometric analyses performed.
**Fig 1:** *The overall domains of daily functioning assessed by the Hypo-METRICS app.* TABLE_PLACEHOLDER:Table 5
## Discussion
This study examined the acceptability and psychometric properties of an innovative smartphone app (Hypo-METRICS): results of the present study support its use as an innovative research tool to determine the impact of hypoglycemia on daily functioning among adults with T1DM or T2DM using insulin. Average completion rates were high and the percentage of skipped questions low. The Hypo-METRICS scales had satisfactory model fit (demonstrated by a MCFA and overall satisfactory ω values), high test-retest reliability and satisfactory convergent and divergent validity. Overall, these findings indicate that the novel Hypo-METRICS app is both valid and reliable for assessing the impact of hypoglycemia on daily functioning in research, with high ecological validity and low recall bias.
The high completion rates suggest that the Hypo-METRICS app is an acceptable instrument for assessments of daily functioning by people with T1DM and insulin-treated T2DM, up to three times per day, seven days per week for up to 10 weeks. All three check-ins were similarly acceptable, which may be attributed to the broad/flexible timeframes and that participants could select a convenient time for app completion. The low percentage of skipped questions (for the non-work-related questions) indicates that questions were generally applicable for most participants. The non-work-related question that was skipped the most was the ‘Social functioning’ question, which could be explained by the context of the COVID-19 pandemic (i.e., data collection occurred during a period of pandemic restrictions on social gatherings). The high percentage of work and productivity questions skipped was expected, as participants were instructed to skip these if they did not have paid employment or if it did not concern a workday. However, the question “How many hours did you miss from activities other than work today for ANY reason” does not require the participant to have a paid job to respond to, and the high skip rate could suggest that participants found the question difficult to respond to, difficult to understand, irrelevant, or poorly explained.
At a question-level, the ICC values show that most questions, in particular those focused on worries about hypoglycemia and hyperglycemia, have greater variability between than within individuals. Further, RMSSD values show that for some participants and some questions (particularly negatively-worded and work-related questions), there was no day-to-day variability in responses across the 70-day study period. This may suggest stability in the construct or in sample characteristics (e.g., low baseline depression/anxiety symptoms) and is supported by the floor effects on negatively-worded questions (e.g., the “How irritable do you feel right now?”). Alternative explanations could be that the questions were not capable of capturing variability in the construct in this group of participants, or that variability only occurs within days and not between days. The floor effects are not considered problematic, since it is not desirable, or possible, to reach lower scores than ‘not at all’, and most participants had variable responses to these questions over time. The low variability for some questions may also indicate ‘automatic’ or ‘habitual’ responding’, wherein participants select the same responses when presented with the same questions in the same order multiple times [35]. Future studies could explore if question randomization at each check-in would produce different results. The full range on 0–10 scales were used for all app scales, suggesting that the 11-point length was appropriate, however additional work needs to explore minimal important changes on the scales [36].
The structural validity of the app scales was examined using a MCFA. Model C showed good model fit except on the TLI and Chi-square parameters. TLI values were >0.9, which has been considered an acceptable level [37]. The Chi-square test has been argued to provide an unrealistic null-hypothesis and the value is heavily influenced by sample size; therefore, it was considered less important in model selection [30]. The two adjustments made to the original model (model A) were supported by theory. The first adjustment was to move the ‘Irritability’ question (originally paired with ‘Overall mood’), to form a two-question ‘Negative affect’ scale with the ‘Anxiety’ question. As irritability is a facet of mood [38], it was originally paired with mood. However, irritability and anxiety are closely related as they are both aspects of negative emotionality [39]. This latter pairing was better supported by the data. The second adjustment was to separate the ‘Fear of hypoglycemia’ and ‘Fear of hyperglycemia’ questions from an original two-question scale into two, single-question scores. Although these two constructs have previously been found to be significantly correlated [40], the low internal consistency suggests that these did not covary in the current dataset. Further, participants were, on average, more worried about ‘highs’ than ‘lows’, which has been observed clinically and elsewhere [24]. An alternative explanation could be that the variance for the two questions generally was too low to allow them to covary and correlate.
Internal consistency of all app scales was satisfactory (ω >0.7) at the between-person level, but not at the within-person level for ‘Negative affect’ (across all three check-ins) and ‘Cognitive functioning’ (evening check-in). Internal consistency is highly dependent on number of questions in the scale [15], and similar within-person ω-values for two-question scales have been reported in other EMA studies and found acceptable [41,42]. Low ω-values could also reflect greater question heterogeneity than in other pairs of questions, so for analysis at the within-person level only (e.g. $$n = 1$$ studies), researchers could consider analyzing single questions rather than scales [43].
EMA methods allow an exploration of the variation in outcomes from timepoint to timepoint. Expecting perfect test-retest reliability (correlations) between assessments contradicts the general assumption of the method [42]. However, if comparing aggregated data (e.g., averaged over a longer time period), representing a person’s traits or general pattern of responding, one could expect more persistent scores across time [42]. This approach has been used in other EMA studies. For example, Csikszentmihalyi et al reported that mean scores on variables measuring affect from the first part of a week correlated highly ($r = 0.74$) with scores from the second half of the week [42]. The aggregated Hypo-METRICS app scores showed high test-retest reliability, with correlations ranging from rs = 0.76 (for the one-question ‘Overall mood’ scale in the evening) to rs>0.9 (for the ‘Fear of hypoglycemia’ and ‘Fear of hyperglycemia’ single questions). These findings suggest reasonable consistency, across a few weeks, in the average scores on the measured constructs.
The correlations between the app scales and validated PROMs overall showed satisfactory convergent validity. The majority of the hypothesized highest correlations (indicative of convergent validity) and lowest correlations (indicative of divergent validity) were confirmed, although some of the hypothesized lowest correlations (e.g. for the ‘Social functioning’ app scale and ‘Anxiety (GAD-7)” PROM) were higher than anticipated. Correlations between app scales (aggregated over periods of 1–4 weeks) and validated PROMs were, in some cases, high (rs up to -0.70). However, it is important to note that no collinearity was present, suggesting that the app is not a redundant measure. Further, EMA offers advantages over retrospective questionnaires: it captures variation in the outcomes over time, and it allows assessment of the direct impact of events (here, episodes of hypoglycemia) on the outcomes. All app scales correlated highly with several PROMs, which was expected as previous studies have shown associations between symptoms of anxiety and depression [44], sleep and mood [45], and depressive symptoms and cognitive functioning [46]. However, some expected correlations were not confirmed in this dataset. A moderate-to-strong correlation was expected, but not confirmed, between the ‘Social functioning’ question in the app and the single questions on the DIDP and PAID scales, which refer to ‘relationships with others’ and ‘uncomfortable social situations’, respectively. However, unlike the PAID and DIDP questions, the ‘Social functioning’ question has no attribution to diabetes/hypoglycemia and is focused on a single day, which may explain the low correlations. Instead, the ‘Social functioning’ question correlated highly with the general anxiety questionnaire. Previous qualitative research has shown that anxieties about unpredictable hypoglycemic episodes limit social activities [47], which supports the strong link between social functioning and anxiety seen here. Further work is required to establish the convergent validity of this question. Future research could also explore the meaning of this question from the perspective of the person with diabetes, using cognitive debriefing.
Evidence for convergent validity of work and productivity questions was mixed. High correlations on ‘number of hours worked’ and moderate correlations on ‘productivity’ questions suggest minimal recall bias on these questions when asked retrospectively for the previous seven days. Hours missed from work and ‘activities other than work’ showed very low correlations between daily and weekly measures, suggesting high recall bias in the PROM or that the questions in the app and the PROM were capturing different information.
A strength of this study is its innovative character, including use of advanced statistical methods suitable to explore the psychometric properties of an app for ecological momentary assessments. Factor analyses are often conducted on cross-sectional data, but when data are clustered with repeated measures per participant, use of standard techniques would violate a general assumption of independency between observations [33,48]. MCFA specifically enables a between- and within-person model to run simultaneously and make it possible for the factor structure to vary across these levels [33]. Another strength of this study was the use of several validated PROMs to examine convergent and divergent validity, and the use of approximately matched time periods for correlations between short-form measures (app scales) and long-form measures (PROMs). Furthermore, psychometric properties were able to be confirmed in the first 100 participants of Hypo-METRICS: a sample including people with T1DM and T2DM, with a balanced gender distribution, and varied methods of glucose monitoring and levels of awareness of hypoglycemia. This was an optimal sample, as it was a balance of being large enough to conduct the planned analyses, while having data collection completed early enough to conduct essential analyses determining if the app was ‘fit for purpose’ prior to analyses of central study objectives. The data were collected in “real” everyday life settings, thereby improving ecological validity with reduced recall burden.
A limitation of this study is that due to the substantial requirements of participants in the Hypo-METRICS study, the sample may reflect a highly motivated and relatively “high functioning” group [49]. Thus, the acceptability of the app needs to be explored in other samples, and/or by use of qualitative methods. Another potential limitation is that, despite participants receiving notifications for each check-in at certain times, there were wide time-intervals (six hours) in which participants could submit the check-ins. Allowing participants to complete check-ins at the most convenient time likely increased engagement with the app but may have biased responses towards more positive daily functioning. Future studies could explore how shorter time-intervals would impact on both completion rates and daily functioning scores.
This study investigated the psychometric properties of the majority of the Hypo-METRICS app questions in the three daily check-ins. However, there are some hypoglycemia-specific questions in the app that were not explored here, as these are not asked at each check-in but only if a hypoglycemic episode was reported (i.e., much higher percentage of missing data must be anticipated for these). Additional work is needed to determine the acceptability and psychometric properties of these questions. Similarly, this study should be replicated in independent samples with diverse characteristics (e.g., ethnic, socio-economic, and health-related). Further research is also needed to fully understand respondents’ completion patterns, including potential predictors of completion. Qualitative research would enable a subjective evaluation of the app completion, including the perceived value and/or burden of using the app across many weeks and ways to improve user experience. Qualitative research would further allow for additional investigation of the content validity (relevance, comprehensiveness and comprehensibility) of the app questions. Future versions of the app could be automated and include conversational agents that, by combining CGM data with daily functioning data from the app, could deliver daily guidance on how to optimize treatment plans and/or improve quality of life.
Overall, these findings show that the Hypo-METRICS app is an acceptable, valid, and reliable tool for research to advance knowledge of how hypoglycemia impacts on daily functioning. In addition to its potential in research, the app may have utility in clinical practice to enhance personalized treatment and care for people with diabetes.
## References
1. **Hypoglycaemia, cardiovascular disease, and mortality in diabetes: epidemiology, pathogenesis, and management.**. *The lancet Diabetes & endocrinology* (2019.0) **7** 385-96. PMID: 30926258
2. Brod M, Christensen T, Bushnell DM. **Impact of nocturnal hypoglycemic events on diabetes management, sleep quality, and next-day function: results from a four-country survey.**. *Journal of medical economics.* (2012.0) **15** 77-86. PMID: 22029460
3. Brod M, Christensen T, Thomsen TL, Bushnell DM. **The impact of non-severe hypoglycemic events on work productivity and diabetes management**. *Value in health: the journal of the International Society for Pharmacoeconomics and Outcomes Research* (2011.0) **14** 665-71. PMID: 21839404
4. Green AJ, Fox KM, Grandy S. **Self-reported hypoglycemia and impact on quality of life and depression among adults with type 2 diabetes mellitus**. *Diabetes research and clinical practice* (2012.0) **96** 313-8. PMID: 22296853
5. Pramming S, Thorsteinsson B, Theilgaard A, Pinner EM, Binder C. **Cognitive function during hypoglycaemia in type I diabetes mellitus**. *British medical journal (Clinical research ed).* (1986.0) **292** 647-50. DOI: 10.1136/bmj.292.6521.647
6. Nefs G, Pouwer F. **The role of hypoglycemia in the burden of living with diabetes among adults with diabetes and family members: results from the DAWN2 study in The Netherlands.**. *BMC public health.* (2018.0) **18** 156. DOI: 10.1186/s12889-018-5064-y
7. Little SA, Leelarathna L, Barendse SM, Walkinshaw E, Tan HK, Lubina Solomon A. **Severe hypoglycaemia in type 1 diabetes mellitus: underlying drivers and potential strategies for successful prevention**. *Diabetes/metabolism research and reviews* (2014.0) **30** 175-90. DOI: 10.1002/dmrr.2492
8. Warren RE, Frier BM. **Hypoglycaemia and cognitive function**. *Diabetes, obesity & metabolism.* (2005.0) **7** 493-503. DOI: 10.1111/j.1463-1326.2004.00421.x
9. Henriksen MM, Andersen HU, Thorsteinsson B, Pedersen-Bjergaard U. **Asymptomatic hypoglycaemia in Type 1 diabetes: incidence and risk factors**. *Diabetic medicine: a journal of the British Diabetic Association* (2019.0) **36** 62-9. DOI: 10.1111/dme.13848
10. Hendrieckx C, Ivory N, Singh H, Frier BM, Speight J. **Impact of severe hypoglycaemia on psychological outcomes in adults with Type 2 diabetes: a systematic review**. *Diabetic medicine: a journal of the British Diabetic Association* (2019.0) **36** 1082-91. DOI: 10.1111/dme.14067
11. Amiel SA. **The consequences of hypoglycaemia**. *Diabetologia* (2021.0). DOI: 10.1007/s00125-020-05366-3
12. de Galan BE, McCrimmon RJ, Ibberson M, Heller SR, Choudhary P, Pouwer F. **Reducing the burden of hypoglycaemia in people with diabetes through increased understanding: design of the Hypoglycaemia REdefining SOLutions for better liVEs (Hypo-RESOLVE) project.**. *Diabetic medicine: a journal of the British Diabetic Association* (2020.0) **37** 1066-73. DOI: 10.1111/dme.14240
13. Divilly P, Zaremba N, Mahmoudi Z, Søholm U, Pollard DJ, Broadley M. **Hypo-METRICS: Hypoglycaemia—MEasurement, ThResholds and ImpaCtS—A multi-country clinical study to define the optimal threshold and duration of sensor-detected hypoglycaemia that impact the experience of hypoglycaemia, quality of life and health economic outcomes: the study protocol**. *Diabetic medicine: a journal of the British Diabetic Association* (2022.0) e14892. DOI: 10.1111/dme.14892
14. Gold AE, MacLeod KM, Frier BM. **Frequency of severe hypoglycemia in patients with type I diabetes with impaired awareness of hypoglycemia**. *Diabetes care* (1994.0) **17** 697-703. DOI: 10.2337/diacare.17.7.697
15. Terwee C, Bot S, Boer M, van der Windt D, Knol D, Dekker J. **Quality criteria were proposed for measurement properties of health status questionnaires**. *Journal of clinical epidemiology* (2007.0) **60** 34-42. DOI: 10.1016/j.jclinepi.2006.03.012
16. Wild D, Grove A, Martin M, Eremenco S, McElroy S, Verjee-Lorenz A. **Principles of Good Practice for the Translation and Cultural Adaptation Process for Patient-Reported Outcomes (PRO) Measures: report of the ISPOR Task Force for Translation and Cultural Adaptation.**. *Value in health: the journal of the International Society for Pharmacoeconomics and Outcomes Research* (2005.0) **8** 94-104. DOI: 10.1111/j.1524-4733.2005.04054.x
17. Søholm U, Broadley M, Zaremba N, Divilly P, Nefs G, Mahmoudi Z. **Investigating the day-to-day impact of hypoglycaemia in adults with type 1 or type 2 diabetes: design and validation protocol of the Hypo-METRICS application**. *BMJ open* (2022.0) **12** e051651. DOI: 10.1136/bmjopen-2021-051651
18. Yu L, Buysse DJ, Germain A, Moul DE, Stover A, Dodds NE. **Development of short forms from the PROMIS™ sleep disturbance and Sleep-Related Impairment item banks**. *Behav Sleep Med* (2011.0) **10** 6-24. PMID: 22250775
19. Kroenke K, Spitzer RL, Williams JB. **The PHQ-9: validity of a brief depression severity measure**. *Journal of general internal medicine* (2001.0) **16** 606-13. DOI: 10.1046/j.1525-1497.2001.016009606.x
20. Spitzer RL, Kroenke K, Williams JB, Löwe B. **A brief measure for assessing generalized anxiety disorder: the GAD-7**. *Archives of internal medicine* (2006.0) **166** 1092-7. DOI: 10.1001/archinte.166.10.1092
21. Ware JE, Sherbourne CD. **The MOS 36-item short-form health survey (SF-36).**. *I. Conceptual framework and item selection. Medical care* (1992.0) **30** 473-83. PMID: 1593914
22. Sullivan MJ, Hall E, Bartolacci R, Sullivan ME, Adams H. **Perceived cognitive deficits, emotional distress and disability following whiplash injury**. *Pain research & management* (2002.0) **7** 120-6. DOI: 10.1155/2002/502984
23. Cox DJ, Irvine A, Gonder-Frederick L, Nowacek G, Butterfield J. **Fear of hypoglycemia: quantification, validation, and utilization**. *Diabetes care* (1987.0) **10** 617-21. DOI: 10.2337/diacare.10.5.617
24. Polonsky WH, Anderson BJ, Lohrer PA, Welch G, Jacobson AM, Aponte JE. **Assessment of diabetes-related distress**. *Diabetes care* (1995.0) **18** 754-60. DOI: 10.2337/diacare.18.6.754
25. Holmes-Truscott E, Skovlund SE, Hendrieckx C, Pouwer F, Peyrot M, Speight J. **Assessing the perceived impact of diabetes on quality of life: Psychometric validation of the DAWN2 Impact of Diabetes Profile in the second Diabetes MILES—Australia (MILES-2) survey.**. *Diabetes research and clinical practice* (2019.0) **150** 253-63. DOI: 10.1016/j.diabres.2019.03.020
26. Reilly MC, Zbrozek AS, Dukes EM. **The validity and reproducibility of a work productivity and activity impairment instrument.**. *PharmacoEconomics.* (1993.0) **4** 353-65. DOI: 10.2165/00019053-199304050-00006
27. 27Team R. RStudio: Integrated Development Environment for R. [cited 2021. Version 1.4.1717:[Available from: http://www.rstudio.com/.
28. Theobald E.. **Students Are Rarely Independent: When, Why, and How to Use Random Effects in Discipline-Based Education Research.**. *CBE life sciences education* (2018.0) **17** rm2. DOI: 10.1187/cbe.17-12-0280
29. Woyshville MJ, Lackamp JM, Eisengart JA, Gilliland JA. **On the meaning and measurement of affective instability: clues from chaos theory**. *Biological psychiatry* (1999.0) **45** 261-9. DOI: 10.1016/s0006-3223(98)00152-8
30. Knekta E, Runyon C, Eddy S. **One Size Doesn’t Fit All: Using Factor Analysis to Gather Validity Evidence When Using Surveys in Your Research.**. *CBE life sciences education* (2019.0) **18** rm1-rm. DOI: 10.1187/cbe.18-04-0064
31. Mokkink LB, Terwee CB, Patrick DL, Alonso J, Stratford PW, Knol DL. **The COSMIN study reached international consensus on taxonomy, terminology, and definitions of measurement properties for health-related patient-reported outcomes**. *Journal of clinical epidemiology* (2010.0) **63** 737-45. DOI: 10.1016/j.jclinepi.2010.02.006
32. McNeish D.. **Thanks coefficient alpha, we’ll take it from here**. *Psychological methods.* (2018.0) **23** 412-33. DOI: 10.1037/met0000144
33. Huang F.. *Conducting Multilevel Confirmatory Factor Analysis Using R* (2017.0)
34. Granholm E, Holden JL, Mikhael T, Link PC, Swendsen J, Depp C. **What Do People With Schizophrenia Do All Day? Ecological Momentary Assessment of Real-World Functioning in Schizophrenia**. *Schizophrenia bulletin.* (2020.0) **46** 242-51. PMID: 31504955
35. Silvia PJ, Kwapil TR, Walsh MA, Myin-Germeys I. **Planned missing-data designs in experience-sampling research: Monte Carlo simulations of efficient designs for assessing within-person constructs.**. *Behavior research methods.* (2014.0) **46** 41-54. PMID: 23709167
36. de Vet HC, Terwee CB, Ostelo RW, Beckerman H, Knol DL, Bouter LM. **Minimal changes in health status questionnaires: distinction between minimally detectable change and minimally important change.**. *Health and quality of life outcomes.* (2006.0) **4** 54-. DOI: 10.1186/1477-7525-4-54
37. Gonder-Frederick LA, Schmidt KM, Vajda KA, Greear ML, Singh H, Shepard JA. **Psychometric properties of the hypoglycemia fear survey-ii for adults with type 1 diabetes**. *Diabetes care* (2011.0) **34** 801-6. DOI: 10.2337/dc10-1343
38. Vidal-Ribas P, Brotman MA, Valdivieso I, Leibenluft E, Stringaris A. **The Status of Irritability in Psychiatry: A Conceptual and Quantitative Review.**. *J Am Acad Child Adolesc Psychiatry.* (2016.0) **55** 556-70. DOI: 10.1016/j.jaac.2016.04.014
39. Denollet J.. **DS14: standard assessment of negative affectivity, social inhibition, and Type D personality**. *Psychosomatic medicine* (2005.0) **67** 89-97. DOI: 10.1097/01.psy.0000149256.81953.49
40. Singh H, Gonder-Frederick L, Schmidt K, Ford D, Vajda K, Hawley J. **Assessing Hyperglycemia Avoidance in People with type 1 Diabetes.**. *Diabetes Management* (2014.0) **4** 263
41. Forkmann T, Spangenberg L, Rath D, Hallensleben N, Hegerl U, Kersting A. **Assessing suicidality in real time: A psychometric evaluation of self-report items for the assessment of suicidal ideation and its proximal risk factors using ecological momentary assessments**. *Journal of abnormal psychology* (2018.0) **127** 758-69. DOI: 10.1037/abn0000381
42. Csikszentmihalyi M, Larson R, Csikszentmihalyi M. *Flow and the Foundations of Positive Psychology: The Collected Works of Mihaly Csikszentmihalyi* (2014.0) 35-54
43. Tavakol M, Dennick R. **Making sense of Cronbach’s alpha.**. *Int J Med Educ* (2011.0) **2** 53-5. DOI: 10.5116/ijme.4dfb.8dfd
44. Gorman JM. **Comorbid depression and anxiety spectrum disorders.**. *Depression and anxiety.* (1996.0) **4** 160-8. DOI: 10.1002/(SICI)1520-6394(1996)4:4<160::AID-DA2>3.0.CO;2-J
45. Triantafillou S, Saeb S, Lattie EG, Mohr DC, Kording KP. **Relationship Between Sleep Quality and Mood: Ecological Momentary Assessment Study.**. *JMIR mental health.* (2019.0) **6** e12613-e. DOI: 10.2196/12613
46. Shimada H, Park H, Makizako H, Doi T, Lee S, Suzuki T. **Depressive symptoms and cognitive performance in older adults**. *Journal of psychiatric research* (2014.0) **57** 149-56. DOI: 10.1016/j.jpsychires.2014.06.004
47. Vanstone M, Rewegan A, Brundisini F, Dejean D, Giacomini M. **Patient Perspectives on Quality of Life With Uncontrolled Type 1 Diabetes Mellitus: A Systematic Review and Qualitative Meta-synthesis.**. *Ontario health technology assessment series* (2015.0) **15** 1-29. PMID: 26649106
48. Huang FL, Cornell DG. **Using Multilevel Factor Analysis With Clustered Data: Investigating the Factor Structure of the Positive Values Scale.**. *Journal of Psychoeducational Assessment* (2015.0) **34** 3-14
49. Cranford JA, Shrout PE, Iida M, Rafaeli E, Yip T, Bolger N. **A procedure for evaluating sensitivity to within-person change: can mood measures in diary studies detect change reliably?**. *Personality & social psychology bulletin.* (2006.0) **32** 917-29. DOI: 10.1177/0146167206287721
|
---
title: 'Assessing suicidality during the SARS-CoV-2 pandemic: Lessons learned from
adaptation and implementation of a telephone-based suicide risk assessment and response
protocol in Malawi'
authors:
- Kelsey R. Landrum
- Christopher F. Akiba
- Brian W. Pence
- Harriet Akello
- Hamis Chikalimba
- Josée M. Dussault
- Mina C. Hosseinipour
- Kingsley Kanzoole
- Kazione Kulisewa
- Jullita Kenala Malava
- Michael Udedi
- Chifundo C. Zimba
- Bradley N. Gaynes
journal: PLOS ONE
year: 2023
pmcid: PMC10022777
doi: 10.1371/journal.pone.0281711
license: CC BY 4.0
---
# Assessing suicidality during the SARS-CoV-2 pandemic: Lessons learned from adaptation and implementation of a telephone-based suicide risk assessment and response protocol in Malawi
## Abstract
The SARS-CoV-2 pandemic led to the rapid transition of many research studies from in-person to telephone follow-up globally. For mental health research in low-income settings, tele-follow-up raises unique safety concerns due to the potential of identifying suicide risk in participants who cannot be immediately referred to in-person care. We developed and iteratively adapted a telephone-delivered protocol designed to follow a positive suicide risk assessment (SRA) screening. We describe the development and implementation of this SRA protocol during follow-up of a cohort of adults with depression in Malawi enrolled in the Sub-Saharan Africa Regional Partnership for Mental Health Capacity Building (SHARP) randomized control trial during the COVID-19 era. We assess protocol feasibility and performance, describe challenges and lessons learned during protocol development, and discuss how this protocol may function as a model for use in other settings. Transition from in-person to telephone SRAs was feasible and identified participants with suicidal ideation (SI). Follow-up protocol monitoring indicated a $100\%$ resolution rate of SI in cases following the SRA during this period, indicating that this was an effective strategy for monitoring SI virtually. Over $2\%$ of participants monitored by phone screened positive for SI in the first six months of protocol implementation. Most were passive risk ($73\%$). There were no suicides or suicide attempts during the study period. Barriers to implementation included use of a contact person for participants without personal phones, intermittent network problems, and pre-paid phone plans delaying follow-up. Delays in follow-up due to challenges with reaching contact persons, intermittent network problems, and pre-paid phone plans should be considered in future adaptations. Future directions include validation studies for use of this protocol in its existing context. This protocol was successful at identifying suicide risk levels and providing research assistants and participants with structured follow-up and referral plans. The protocol can serve as a model for virtual SRA development and is currently being adapted for use in other contexts.
## Introduction
Mental health disorders are among the leading causes of death and disability worldwide, especially in low- and middle-income countries (LMICs). Globally, mental health disorders account for nearly a third of years lived with disability and are the fifth-leading cause of disability-adjusted life years (DALYs), making up $13\%$ of DALYs worldwide [1–3]. Aproximately three quarters of this burden resides in LMICs [4]. In addition to morbidity and disability, untreated mental health disorders are associated with early mortality with an estimated $75\%$ of suicides occurring in LMICs [5]. Yearly prevalence of suicide attempt and suicidal behavior in *Malawi is* approximately $0.8\%$ and $7.9\%$, respectively [6]. While effective and low cost treatments are available for the most common mental disorders, the treatment gap is yawning [7, 8]. The proportion of mental health workers in LMICs is as low as 2 per 100,000 population; accordingly, most affected individuals in LMICs do not receive care [9]. The average disparity between those with mental illness in need of care and those who receive services is over $90\%$ in most LMICs [10–14]. The extremely limited access to and availability of national or local suicide hotlines in this context further indicated need for a telephone-based Suicide Risk Assessment (SRA) during the SARS-CoV-2.
The limited availability of mental health services makes identification and management of suicidal thoughts and behavior, often referred to as suicidality, especially challenging. Such behavior can include passive and active suicidal ideation (SI), suicide attempts, and death due to suicide [15, 16]. Passive SI is defined as having thoughts of suicide without the intention to act on suicidal thoughts, while active SI is defined as having suicidal thoughts and intention to act on such thoughts [15, 17]. Passive and active SI typically require different clinical interventions. Conducting mental health research often requires SI measurement, with careful assessment and triage, and is traditionally dependent on in-person assessment with the ability to immediately engage participants if clinical safety measures are needed [18–27].
The SARS-CoV-2 pandemic disrupted and further limited mental health care by precluding many of these face-to-face assessments [28]. The pandemic forced ongoing studies of mental heath in LMICs to quickly pivot from in-person to virtual SI assessments with consideration of how to safely triage and respond to suicide risk by telephone. Increased suicidal risk has previously been associated with epidemics, with ongoing studies suggesting the possibility of increased risk of suicide during the current SARS-CoV-2 pandemic [29–32]. Further, assessment of the feasibility of adapting in-person mental health assessments to virtual assessments in low-resource settings during the pandemic is essential and missing from scientific literature.
In the midst of an ongoing clinical trial integrating depression treatment in non-communicable disease (NCD) settings in Malawi, researchers and participants in the Sub-Saharan Africa Regional Partnership for Mental Health Capacity Building (SHARP) study underwent a rapid change from in-person suicidality assessments to telephone-based assessments by developing a protocol to provide a robust, feasible, and rapidly adaptable protocol to assess suicidality in a timely, safe, quality, affordable, and virtual manner [33–35].
This paper describes the feasibility of development and implementation of this suicidality assessment protocol. We aim to: 1) assess the feasibility (defined as the successful use of the tool in this study’s context) of the protocol for assessing suicidality over telephone among patients who screen positive during the Suicide Risk Assessment (SRA), 2) describe the information we were able to collect and categorize, 3) describe the primary challenges identified and lessons learned during protocol development, and 4) discuss future developments and protocol use in other settings [36, 37].
## Methods
The parent, SHARP scale-up study is an implementation science trial comparing the success of a basic versus an enhanced implementation package in achieving integration of depression treatment with NCD treatment at 10 NCD clinics in Malawi [33, 35]. Clinics in the parent study were randomized with a 1:1 ratio to a basic or enhanced implementation strategy to compare implementation and effectiveness outcomes and cost-effectiveness of the two strategies. Patients receiving care at each clinic who meet eligibility criteria described below were invited to enroll in the study to provide clinical outcome measures. In this paper, we describe the feasibility, development, and implementation of the SHARP Safety Response Protocol for Phone Interviews in response to the SARS-CoV-2 pandemic. We do not report results of the parent study trial.
## Participants
The cohort of patients ($$n = 946$$) consisted of participants meeting the following parent study inclusion criteria: aged 18–65 years and being a patient in care for diabetes or hypertension management at a participating NCD clinic. Of these patients, $$n = 739$$ had elevated depressive symptoms (a score ≥5 on the Patient Health Questionnaire-9 [PHQ-9]) at baseline and were eligible for inclusion in this analysis. Exclusion criteria are history of bipolar or psychotic disorder and/or an emergent self-harm threat. One participant was excluded from analysis due to being a prevalent SI case already receiving regular SI follow-up prior to the study period. Among the $$n = 738$$ participants eligible for analysis, $$n = 602$$ SI screenings were conducted during the study period.
## Ethical approval
All study materials and research activities have been approved by the National Health Sciences Research Committee of Malawi (NHSRC; Approval # 17–3110) and the Biomedical Institutional Review Board of the University of North Carolina at Chapel Hill (Approval #$\frac{17}{11}$/1925). All methods were performed in accordance with the relevant guidelines and regulations. The current manuscript does not report results of the parent study. All participants gave written, informed consent to participate in the parent study. Written informed consent clearly explained the purpose of the research, what will be required of the individual, and the risks and benefits of participation in the participant’s preferred language. All informed consent procedures for the parent study were approved by both ethics committees.
## Measures: PHQ-9 and SRA
Participants are asked to complete the PHQ-9, a depression screening tool developed and originally validated in high-income countries but subsequently validated in Malawi and many other low-income countries, as part of baseline and follow-up assessments at 3, 6, and 12 months [38–40]. The final PHQ-9 question asks, “During the past two weeks, how many days have you been bothered by thoughts that you would be better off dead or of hurting yourself in some way?” Participants respond with: “0 days”, “1–7 days”, 8–12 days”, or “13 or 14 days”. Any response other than “0 days” was considered a positive screen for suicidal ideation, requiring a follow-up SRA following this protocol’s schedule. The PHQ-9 questionnaire and SRA were translated into Chichewa and Chitumbuka for use in Malawi [41].
## Setting
From May 13, 2019 through March 24, 2020, interview data were collected with in-person interviews at 10 NCD clinics in all three regions of Malawi. However, as a result of the SARS-CoV-2 pandemic and the patient population’s high risk for severe complications of COVID-19, all in-person participant contact was put on hold from March 25, 2020 until October 16, 2020. The research team decided to transition to telephone-based follow-up interview data collection, necessitating the development of virtual study and safety assessment and response protocols.
## Protocol development
The SRA protocol guides research team members through a series of risk assessments and the creation of a follow-up plan. The protocol was originally designed to be implemented in-person immediately after a participant screens positive on PHQ-9 Question 9 (any response other than “0 days”) or indicates concerns of suicidal thoughts. The tablet-based data collection software, Open Data Kit (ODK), is set to warn the RA to conduct the SRA when the participant answers affirmatively to PHQ-9 Question 9 [42]. If the patient screens positive, the RA is instructed to not leave the patient alone, complete the telephone protocol form, transfer the patient to clinician care, enter the event into the study log for review by team members, and follow up with the participant as indicated by the protocol (SF 1, SF 2). If the participant indicates SI via a method other than the PHQ-9, RA’s also completed the SRA.
The research team met in March 2020 to discuss methods of adapting the in-person protocol to assess and manage suicide risk virtually during, and beyond, the SARS-CoV-2 pandemic. Given the immediate need of a suicidality protocol without time for a full-validation study of the protocol before implementation due to the pandemic, subject matter experts helped develop the guide. Anticipating that hospitals would have increased resource strain, decreased patient capacity, and increased SARS-CoV-2 transmission risk, the team revised the SRA for telephone delivery with branching logic to guide trained research assistants (RAs) in creating a follow-up plan with participants, reducing participants’ immediate suicide risk, engaging a nearby support person, creating a follow-up plan with participants, and referring participants to in-person care when clinically indicated. Research assistants, with a minimum high school education level and training in SHARP study procedures, were trained by study staff and subject matter experts in suicidality and related health protocols and interventions, including psychiatry, health behavior, and epidemiology experts.
The in-person SRA is an algorithmic questionnaire that is administered by phone in Chichewa or Chitumbuka, and responses are translated into English for data entry. Research staff provide feedback about the event to the research coordinator after each time the protocol is administered. The protocol was implemented in April 2020 after an iterative review process including all research team members with diverse areas of subject matter expertise.
## Procedure: Protocol training
The RAs received two trainings for telephone SRA implementation from clinical and research study team members. These trainings consisted of two 1-hour formal trainings via Zoom in which team members discussed the protocol and practiced protocol implementation in simulated scenarios. Team members had a follow-up meeting one week later to answer questions that arose while preparing for protocol implementation after the two training sessions.
## Section 1: Suicide risk assessment (SF 1–2)
The aim of Section 1 is to assess level of SI. Part A (SF 1) assesses active versus passive suicidality and Part B assesses severity of active thoughts, if present. The RA assesses the participant’s current self-harm risk in Part C (SF 1), regardless of the participant’s reporting of active versus passive suicidal thoughts in Part A. Suicide risk level is defined based on participant responses to Parts 1A-C. Importantly, “thoughts of hurting yourself right now” (SF 1, C.1) was added to the telephone protocol, as no clinician (who would have asked this question in-person and provided immediate intervention in the pre-COVID period) was available with this virtual assessment. The addition of this question to the telephone protocol allowed immediate, virtual assessment of active high and acute risk status.
## Section 2: Confirm patient location
The purpose of Section 2 (SF 1) is to confirm the patient’s location and identify if other individuals are nearby to help the participant as needed. The RA continues to Section 3 if the participant’s suicide risk is active-high or active-emergent. The RA skips to Section 4 if the participant’s suicide risk is passive, active-low, or active-moderate.
## Section 3: Safety plan and next steps for active-high or active-emergent
Section 3 (SF 1) is critical in confirming contact information, on-site support persons, and a strategy to assist the patient in reaching mental health care. The RA confirms the participant’s emergency contact and Health Surveillance Assistant (HSA) information. An HSA is a trained community-based health provider who is able to link community members to various health services. The RA asks the participant to go to the nearest health facility to speak with a mental health professional with the help of a nearby trusted contact. If no trusted contact is available, the RA speaks with an emergency contact in the participant’s locator form regarding the immediate safety concern. If no emergency contact is available, the RA assists the participant in contacting a health professional at the nearest health facility and notifies the study’s clinical coordinator at the participant’s health facility.
## Section 4: Safety plan and next steps for passive, active-low, or active-moderate
The overall goal of Section 4 (SF 1) is to create a safety and follow-up plan. The RA asks the participant to inform a trusted contact if the participant feels as though they might harm themselves. If the participant does not agree, the RA returns to Section 3 to proceed with the active-high and active-emergent workflow, including asking the participant if they can speak with a nearby individual. If the participant agrees to inform a trusted contact if the participant feels as though they might harm themselves, the RA continues through Section 4 to assess the presence of potentially dangerous items in the home. If the RA cannot confirm that the participant has handed over potentially dangerous items, the RA completes Section 3. Otherwise, the RA provides the participant with referral information to mental health specialists and asks the participant to seek care if they believe they may harm themselves.
The protocol then defines a follow-up schedule based on suicide risk level (SF 1–5). If the RA cannot reach the participant at the follow-up time, the RA calls the participant’s contact person. If the RA is unable to reach the participant’s contact, the RA informs the study coordinator for further guidance.
## Procedure: Protocol implementation of Section 5: Saftey follow-up and Section 6: Summary of follow-up contacts
During the scheduled follow-up contacts, the RA asks how the participant is doing, for updates on the agreed upon action plan from the previous call, and screens for SI with PHQ-9 Question 9 (SF 1). If no SI is present, the RA continues with the prescribed follow-up schedule. If suicidal risk is present, the RA restarts the protocol, returning to Section 1.
## Passive risk follow-up (SF 1, SF 3)
Follow-up ends if the participant begins with a passive risk and improves to no risk at the one week follow-up. If the participant maintains a passive risk at weekly follow-up for three follow-ups, the RA asks the participant to contact the RA for the 4th follow-up to confirm participant ability to proactively manage their own care. Follow-up ends when the participant successfully contacts the RA for follow-up.
## Active-low follow-up (SF 1, SF 3)
If the participant starts with active-low risk, follow-up is conducted weekly until the participant improves to no suicidal risk (in which case follow-up ends) or the participant improves to passive suicidal risk. If the latter, follow-up is completed according to the passive risk schedule. If the participant remains at passive or active-low risk for 3 follow-ups, the RA requests that the participant initiaties the 4th follow-up contact. If the participant successfully contacts the RA for the 4th weekly follow-up, follow-up can end.
## Active-moderate follow-up (SF 1, SF 4)
Follow-up is conducted every 3 days for participants beginning with active-moderate risk. If the participant improves to active-low or passive risk, follow-up transitions to weekly follow-up according to the above guidelines. Follow-up can end if the participant improves to no suicidal risk at two consecutive weekly follow-ups. Follow-up ends without asking the participant to contact the RA for the last follow-up if the participant can confirm their health professional’s contact information and that they will contact their mental health professional if thoughts of suicide return. The RA contacts the study clinical coordinator for further direction if the participant maintains active-moderate risk after 3 contacts.
## Active high and active-emergent follow-up (SF 1, SF 5)
If the participant begins with active-high or active-emergent risk, follow-up is conducted every day until the participant improves to active-moderate, active-low, or passive risk (in which case follow-up transitions to the follow-up for the participant’s new risk category). If the participant has no suicidal risk at two consecutive weekly follow-ups, follow up can end. The RA is instructed to contact the study coordinator if the participant continues to have active-high or active emergent suicidal risk after 3 contacts.
In all risk levels, the RA instructs the participant to confirm that the participant has mental health professional contact information and will contact the health professional if thoughts of self-harm return. In all cases, the RA fills out the protocol form, including Section 6: Summary of Follow-up Contacts.
## Study analysis and statistics
We describe the feasibility of protocol implementation and discuss results of the implementation of our telephone protocol when participants screened positive for suicide risk during baseline, 3-, 6-, and 12-month surveys during the first 6 months of telephone protocol implementation (March 25, 2020-September 25, 2020). We use frequency distributions to highlight screened population characteristics, including demographic characteristics. Feasibility was measured as the proportion of completed follow-ups out of total number of follow-ups due as indicated by the safety response protocol. We report the number of participants screened, the number of follow-ups completed, and the number of suicide risk cases resolved during telephone protocol use. We apply descriptive analyses to outline protocol implementation, including SRA screening and frequency of positive results. All statistical analyses were conducted using STATA IC (Version 16) software [43].
## Results
A total of $$n = 602$$ follow-up interviews among $$n = 738$$ eligible participants were conducted during the study period. The majority of participants were females aged 50 years or older who were married with children (Table 1). The majority ($81\%$) of participants reported mobile phone access at baseline, with $2\%$ having access to a landline at baseline.
**Table 1**
| Unnamed: 0 | Unnamed: 1 | N (%) |
| --- | --- | --- |
| Age Group | | 738 |
| | 18–29 | 22 (2.98) |
| | 30–39 | 84 (11.38) |
| | 40–49 | 196 (26.56) |
| | 50+ | 436 (59.08) |
| | Missing | 0 (0.00) |
| Gender | Female | 581 (78.73) |
| | Male | 157 (21.27) |
| | Missing | 0 (0.00) |
| Employment | Employed | 700 (94.85) |
| | Unemployed | 34 (4.61) |
| | Missing | 4 (0.54) |
| Marital status | Married | 489 (66.26) |
| | Separated | 68 (9.21) |
| | Divorced | 44 (5.96) |
| | Widowed | 120 (16.26) |
| | Cohabitating with partner | 3 (0.41) |
| | Never married | 13 (1.76) |
| | Missing | 1 (0.14) |
| Parity ** | 0–4 | 192 (33.05) |
| | 5–6 | 150 (25.82) |
| | 7–8 | 142 (24.44) |
| | 9–14 | 95 (16.35) |
| | Missing | 2 (0.34) |
| Phone access | Mobile phone | 599 (81.17) |
| | Non-mobile phone | 15 (2.03) |
| SRA screening in 6 months prior to telephone protocol implementation *** | Yes | 5 (33.33) |
| | No | 10 (66.67) |
## Feasiblity of protocol implementation and challenges
Study RAs used the protocol to successfully identify suicide risk level of all participants who screened positive for SI, with $100\%$ of cases receiving the SRA being resolved. A total of 602 SI screenings took place in the first six months of telephone protocol implementation (during which no new participants were enrolled because of COVID-19 restrictions) (Table 2). During this period, 13 ($2\%$) participants received a SRA after a positive PHQ-9 screening, with follow-up plans varying based on participant risk-level. Three (<$1\%$) participants recieved a SRA after other indication of SI to study staff. In total, 15 ($2\%$) participants received a SRA for any reason. All patients who screened positive received a SRA.
**Table 2**
| Unnamed: 0 | Unnamed: 1 | SRA telephone protocol period* |
| --- | --- | --- |
| Identification of suicidal ideation (SI) | | 15 |
| Identification of suicidal ideation (SI) | Positive PHQ-9 Question 9 | 12 (1.99) |
| Identification of suicidal ideation (SI) | Otherwise reported SI | 3 (0.50) |
| Identification of suicidal ideation (SI) | Missing | 0 (0.00) |
| Self-harm or suicidal thoughts in the last 2 weeks among those who screened positive on PHQ-9 Question 9 | | 12 |
| Self-harm or suicidal thoughts in the last 2 weeks among those who screened positive on PHQ-9 Question 9 | 1–7 days | 8 (66.67) |
| Self-harm or suicidal thoughts in the last 2 weeks among those who screened positive on PHQ-9 Question 9 | 8–12 days | 1 (8.33) |
| Self-harm or suicidal thoughts in the last 2 weeks among those who screened positive on PHQ-9 Question 9 | 13–14 days | 3 (25.00) |
| Self-harm or suicidal thoughts in the last 2 weeks among those who screened positive on PHQ-9 Question 9 | Missing | 0 (0.00) |
| Suicide risk level among those who screened positive on PHQ-9 Question 9 or otherwise reported SI | | 15 |
| Suicide risk level among those who screened positive on PHQ-9 Question 9 or otherwise reported SI | Passive** | 11 (73.33) |
| Suicide risk level among those who screened positive on PHQ-9 Question 9 or otherwise reported SI | Active-Low** | 1 (6.67) |
| Suicide risk level among those who screened positive on PHQ-9 Question 9 or otherwise reported SI | Active-Moderate | 1 (6.67) |
| Suicide risk level among those who screened positive on PHQ-9 Question 9 or otherwise reported SI | Active-High/Active-Acute | 2 (13.33) |
| Number of protocol follow-ups | | |
| Number of protocol follow-ups | Mean | 2.40 |
| Number of protocol follow-ups | Median | 2.00 |
| Number of protocol follow-ups | Min | 1 |
| Number of protocol follow-ups | Max | 5 |
| Duration of protocol follow-up (days) | | |
| Duration of protocol follow-up (days) | Mean | 13.86 |
| Duration of protocol follow-up (days) | Median | 13.00 |
| Duration of protocol follow-up (days) | Min | 0 |
| Duration of protocol follow-up (days) | Max | 35 |
Most participants had passive risk ($$n = 11$$, $73\%$), followed by active-high ($$n = 2$$, $13\%$), active-moderate ($$n = 1$$, $7\%$), and active-low ($$n = 1$$, $7\%$) risk. Most participants with positive screenings reported suicidal thoughts 1–7 days in the previous two weeks (Table 3). Nearly one third of patients who screened positive during telephone protocol implementation ($$n = 5$$, $33\%$) screened positive at least once in the 6 months prior to telephone protocol implementation. No participants screened positive more than one time during telephone protocol use. There were no suicide attempts or deaths during the telephone implementation period.
**Table 3**
| Unnamed: 0 | Unnamed: 1 | Telephone PHQ-9 Q9 N (%) | Depressive severity | N (%) |
| --- | --- | --- | --- | --- |
| Baseline | | 0* | | 0* |
| | 0 days | 0 (0.00) | No depression | 0 (0.00) |
| | 1–7 days | 0 (0.00) | Mild | 0 (0.00) |
| | 8–12 days | 0 (0.00) | Moderate | 0 (0.00) |
| | 13–14 days | 0 (0.00) | Moderately severe | 0 (0.00) |
| | Missing | 0 (0.00) | Severe | 0 (0.00) |
| | | | Missing | 0 (0.00) |
| 3 month follow-up | | 83 | | 83 |
| | 0 days | 80 (96.39) | No depression | 46 (55.42) |
| | 1–7 days | 2 (2.41) | Mild | 33 (39.76) |
| | 8–12 days | 1 (1.20) | Moderate | 4 (4.82) |
| | 13–14 days | 0 (0.00) | Moderately severe | 0 (0.00) |
| | Missing | 0 (0.00) | Severe | 0 (0.00) |
| | | | Missing | 0 (0.00) |
| 6 month follow-up | | 234 | | 234 |
| | 0 days | 231 (98.72) | No depression | 133 (56.84) |
| | 1–7 days | 3 (1.28) | Mild | 73 (31.20) |
| | 8–12 days | 0 (0.00) | Moderate | 21 (8.97) |
| | 13–14 days | 0 (0.00) | Moderately severe | 5 (2.14) |
| | Missing | 0 (0.00) | Severe | 0 (0.00) |
| | | | Missing | 2 (0.85) |
| 12 month follow-up | | 285 | | 285 |
| | 0 days | 279 (97.89) | No depression | 168 (58.95) |
| | 1–7 days | 3 (1.05) | Mild | 93 (32.63) |
| | 8–12 days | 0 (0.00) | Moderate | 13 (4.56) |
| | 13–14 days | 3 (1.05) | Moderately severe | 5 (1.75) |
| | Missing | 0 (0.00) | Severe | 4 (1.40) |
| | | | Missing | 2 (0.70) |
The telephone protocol successfully guided the monitoring of all patients with SI until resolution of SI risk, with a $100\%$ resolution rate of SI during this period indicating that this was an effective strategy to monitoring SI virtually. The first case during the telephone protocol period required study staff and the clinical lead reminding the RA to implement the new protocol. Only one patient required multiple contact attempts for assessment by the RA during their protocol follow-up period and all other contact attempts were successful on the first try. One participant conducted follow-up calls using a clinic phone during clinical visits. The mean duration of follow-up during the telephone protocol period was 14 days (range: [0, 35]), with the mean number of follow-ups was 2.4 (range: [0, 5]).
Researchers identified several challenges to telephone protocol implementation. Telephone visits were occasionally limited by cell phone access due to short, fixed-term phone plans and intermittent network problems affecting audio quality. Prepaid phones were able to receive calls from the research team and clinicians, but participants who had exhausted their prepaid minutes were unable to call research team members and clinicians until a new phone plan was purchased. These phone plans also resulted in some participants changing telephone numbers and access daily, risking delay of scheduled follow-up contacts or precluding follow-up contacts entirely. The use of a contact person’s phone for follow-up risked delaying completion of SI assessments when the contact person was unwilling or unable to give the phone to the participant.
RAs reported being unsure if the contact person was hiding information about harmful items in the home. Additionally, participants could not always receive the needed services when referred to the nearest health facility, forcing them to weigh the costs associated with traveling long distances to district health facilities to access mental health care. Transport problems from the home to the referred health facilities were a challenge to the referral process due to costs associated with travel. RAs reported that poor support services received by referred patients negatively affected trust between RAs, patients, and health providers in some cases.
## Case review
The study team generated adverse event reports weekly. One research coordinator and a psychiatrist reviewed all SI cases weekly. The coordinator asked each RA on a weekly basis if the SI protocol had been implemented in the last week. If so, the RA was instructed to follow up with the research coordinator individually. Noting that the protocol form itself is not submitted to the research team, but functions as a guide followed during telephone assessments, the RAs submitted a narrative of the telephone protocol event to the research team to review.
The research team met weekly to discuss fidelity to the protocol in each case, and one team member followed-up with RAs individually via WhatsApp messaging when more information on a case is needed or discussion about improvements to protocol fidelity was indicated [44]. The research team closed each protocol case when the weekly review team agreed that all protocol components had been completed with fidelity and that suicide risk was no longer present. In these meetings, the team also discussed if protocol modification was needed to address challenges being faced by the research team (e.g., how to modify the protocol in the event of chronic suicidal ideation).
## Protocol implementation example
The following case study demonstrates the application of the telephone-based SRA protocol.
Day 0 –Study RA reported that ODK directed them to complete an SRA with Participant A during a routine, phone-based data collection interview. After completing the SRA, the result was Passive SI. The RA followed the ‘Passive SI’ protocol steps and determined that the participant did not have access to items they could use to harm themselves, established contact with the participant’s spouse (should the RA be unable to reach the participant), and provided mental health referral information should the participant’s suicidal thoughts worsen. The participant mentioned to the RA that they did not feel like they were currently a danger to themselves or others and said they would follow-up with the referral.
Day 7 –Study RA contacted the participant by phone one week later with fidelity, consistent with the protocol’s Passive SI guidelines. During this call the participant endorsed that they no longer had thoughts of harming themself by answering “no” to PHQ-9 question 9.
Day 9 –Study clinical lead met with study coordinator to discuss the case. After noting the RA’s fidelity to the protocol and Participant A’s improvement at follow-up, the case was closed resulting in no further follow-up.
One key point illustrated by this case study is that, although the full protocol is very detailed, its implementation typically was quite efficient in practice. In fact, the protocol was not needed for $98\%$ of study contacts, and when it was needed, safety concerns were appropriately handled with a small amount of follow-up contact. Thus, while a detailed protocol that anticipated all possible eventualities and backup processes was essential, the actual added effort for the research team was quite modest.
## Discussion
Adding to the baseline challenges of assessing and managing suicidal behavior, the SARS-CoV-2 pandemic has disrupted and limited mental health care by precluding many face-to-face assessments, which has been especially problematic for the assessment and management of suicidality [25, 28]. Currently, $93\%$ of countries are experiencing disruptions to mental healthcare and $24\%$ of countries reported a disruption in suicide prevention programs due to the pandemic [45]. Increased suicidal risk has previously been associated with epidemics, with possibility of increased risk of suicide during the current SARS-CoV-2 pandemic [29–32]. This paper reports on a feasible strategy to monitor suicide risk by phone when in-person assessments are not possible.
The rapid adaptation of SHARP’s in-person SRAs for telephone use during the SARS-CoV-2 pandemic allowed us to continue monitoring SI in an ongoing RCT in Malawi during the pandemic. The possibility of increasing suicide risk during the pandemic is concerning, and adapting and implementing a feasible, virtual protocol that addresses study participant suicide risk events and appropriately refers participants to care is critical [45]. Similar SRA protocols have been adapted for use in RCTs, but few have been developed in non-hospital, emergency, and time-sensitive contexts in which mental health care is severely disrupted [46–49].
Overall, transition from in-person to telephone-based SRAs was feasible and functioned well for patients and research team members. Providing a structured protocol is critical to providing research team members with the support and guidance needed to assess participant suicide risk [46]. The protocol identified participants at risk of suicide and referred participants to clinical settings when indicated. As well, the protocol provided research team members with feasible, structured, and actionable responses to participant SI.
Very few participants had active-low or active-high suicide risk during the telephone protocol period, contrary to what we anticipated based on current literature, subject matter expertise, and prior in-person SRAs [50, 51]. With the addition of the “Thoughts of hurting yourself right now” response option, the protocol treats acute-high and active-emergent cases as equally serious with the same follow-up intensities. As well, the “Might hurt yourself before seeing your doctor again” response option may be less meaningful during pandemic times if clinical visits are postponed. Future directions include validation studies to address these questions.
Importantly, one third of participants who had a positive SRA screening during the telephone protocol implementation had a positive SRA screening in the 6 months prior to telephone protocol implementation, a reminder of the importance of feasible and well-designed SRA follow-up schedules [52, 53].
Challenges cited by research team members include inability to predict certain challenges faced during the pandemic (e.g., changes in governmental and institutional policies about quarantine, travel, and other factors affecting participants), network and audio difficulties, and risk of participant mobile phone access delaying or precluding follow-up visits [54–56]. As some participants had phone access through close contacts reported in SRA Sections 2 and 3, it was criticial to collect this information during assessments and participant follow-up was occasionally limited by access to the contact person’s phone. Phone minute-related challenges limited participant ability to initiate calls with clinicians and research team members, despite being able to receive calls from the research team. RAs identified challenges to referring participants to health centers, as not all centers could provide the needed services. Despite inability to predict all possible challenges, the research team created a protocol and multi-layered review process robust enough to meet challenges and circumstances faced by participants and research staff through iterative adjustment and adaptation.
This study includes a small sample of adult participants, limiting generalizability. Lack of availability of qualitative fidelity data for all events precluded formal analysis of protocol fidelity. However, all RA protocol-related actions and responses were monitored by the research team through a weekly, iterative review process and real-time correspondence via WhatsApp. Finally, it is possible that participant responses to questions about suicidality differ when asked about suicidality by phone versus in person, and possibly undreport symptoms [19]. But, prior SI scales have performed well under rapid, self-reporting conditions conditions [18, 22, 49]. The protocol performed conservatively, to error on classifying patients as having a higher severity status. As well, in-depth, weekly case-reviews of all suicide risk events and RA responses by the research team ensured that the protocol was implemented accurately and that participants received appropriate and timely follow-up and care when clinically indicated.
Research implications include the need to assess the protocol’s sensitivity, specificity, positive predictive value (PPV), and negative predicative value (NPV) [57, 58]. Implications also include identifying and including variables for implicit suicidal thoughts to more accurately identify active-low compared to active-moderate and active-high participants, as this may result in different clinical action indication [58–60].
This protocol can serve as a model for adaptation of adult SI protocols for telephone or virtual use in times of pandemic or not. Further, this protocol can function as a model for use in other contexts, particularly in adult populations in LMICs.
## Conclusions
Mental health care barriers magnified by the SARS-CoV-2 pandemic highlight the increasing need to address mental healthcare through technology. In order to conduct mental health research or health care remotely, researchers and clinicians must have access to standardized, reliable, and accurate suicide risk assessment protocols. The telephone suicide risk assessment protocol developed to assess SHARP participant suicide risk during the SARS-CoV-2 pandemic can serve as a model for development of virtual suicide risk assessment protocols, in times of pandemic or not.
## References
1. Russ SJ, Herbert J, Cooper P, Gunnar MR, Goodyer I, Croudace T. **Cortisol levels in response to starting school in children at increased risk for social phobia**. *Psychoneuroendocrinology* (2012.0) **37** 462-74. DOI: 10.1016/j.psyneuen.2011.07.014
2. Whiteford HA, Ferrari AJ, Degenhardt L, Feigin V, Vos T. **The Global Burden of Mental, Neurological and Substance Use Disorders: An Analysis from the Global Burden of Disease Study 2010**. *PLoS One* (2015.0) **10** e0116820. DOI: 10.1371/journal.pone.0116820
3. Vigo D, Thornicroft G, Atun R. **Estimating the true global burden of mental illness**. *The Lancet Psychiatry* (2016.0) **3** 171-8. DOI: 10.1016/S2215-0366(15)00505-2
4. Gureje O, Seedat S, Kola L, Appiah-Poku J, Othieno C, Harris B. **Partnership for mental health development in Sub-Saharan Africa (PaM-D): a collaborative initiative for research and capacity building**. *Epidemiology and Psychiatric Sciences* (2019.0) **28** 389-96. DOI: 10.1017/S2045796018000707
5. Bantjes J, Iemmi V, Coast E, Channer K, Leone T, McDaid D. **Poverty and suicide research in low- and middle-income countries: systematic mapping of literature published in English and a proposed research agenda**. *Global Mental Health* (2016.0) **3**. DOI: 10.1017/gmh.2016.27
6. Pengpid S, Peltzer K. **Prevalence and correlates of suicidal behaviour among adults in Malawi: a nationally representative cross-sectional survey in 2017**. *International Journal of Mental Health Systems* (2021.0) **15** 57. DOI: 10.1186/s13033-021-00483-x
7. Patel V, Chisholm D, Parikh R, Charlson FJ, Degenhardt L, Dua T. **Addressing the burden of mental, neurological, and substance use disorders: key messages from Disease Control Priorities, 3rd edition**. *The Lancet* (2016.0) **387** 1672-85. DOI: 10.1016/S0140-6736(15)00390-6
8. Singla DR, Kohrt BA, Murray LK, Anand A, Chorpita BF, Patel V. **Psychological Treatments for the World: Lessons from Low- and Middle-Income Countries**. *Annu Rev Clin Psychol* (2017.0) **13** 149-81. DOI: 10.1146/annurev-clinpsy-032816-045217
9. Wang ML, Dorer DJ, Fleming MP, Catlin EA. **Clinical Outcomes of Near-Term Infants**. *Pediatrics* (2004.0) **114** 372-6. DOI: 10.1542/peds.114.2.372
10. Demyttenaere Koen, Bruffaerts Ronny, Jose Posada-Villa Isabelle Gasquet, Kovess Viviane, Lepine Jean Pierre. **Prevalence, Severity, and Unmet Need for Treatment of Mental Disorders in the World Health Organization World Mental Health Surveys**. *JAMA* (2004.0) **291** 2581. DOI: 10.1001/jama.291.21.2581
11. Degenhardt L, Glantz M, Evans-Lacko S, Sadikova E, Sampson N, Thornicroft G. **Estimating treatment coverage for people with substance use disorders: an analysis of data from the World Mental Health Surveys**. *World Psychiatry* (2017.0) **16** 299-307. DOI: 10.1002/wps.20457
12. Alonso J, Liu Z, Evans-Lacko S, Sadikova E, Sampson N, Chatterji S. **Treatment gap for anxiety disorders is global: Results of the World Mental Health Surveys in 21 countries**. *Depression and Anxiety* (2018.0) **35** 195-208. DOI: 10.1002/da.22711
13. Thornicroft G, Somnath C, Evans-Lacko S, Gruber M, Sampson N, Aguilar-Gaxiola S. **Undertreatment of people with major depressive disorder in 21 countries**. *The British Journal of Psychiatry* (2017.0) **210** 119-24. DOI: 10.1192/bjp.bp.116.188078
14. 14Mental health: massive scale-up of resources needed if global targets are to be met [Internet]. WHO. World Health Organization; 2018 [cited 2021 May 27]. http://www.who.int/mental_health/evidence/atlas/atlas_2017_web_note/en/
15. Booniam S, Wongpakaran T, Lerttrakarnnon P, Jiraniramai S, Kuntawong P, Wongpakaran N. **Predictors of Passive and Active Suicidal Ideation and Suicide Attempt Among Older People: A Study in Tertiary Care Settings in Thailand**. *Neuropsychiatr Dis Treat* (2020.0) **16** 3135-44. DOI: 10.2147/NDT.S283022
16. Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E. **The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10**. *J Clin Psychiatry* (1998.0) **59** 22-33. PMID: 9881538
17. Barry LC, Wakefield DB, Trestman RL, Conwell Y. **Active and passive suicidal ideation in older prisoners**. *Crisis: The Journal of Crisis Intervention and Suicide Prevention* (2016.0) **37** 88-94. DOI: 10.1027/0227-5910/a000350
18. Harbauer G, Ring M, Schuetz C, Andreae A, Haas S. **Suicidality assessment with PRISM-S—Simple, fast, and visual: A brief nonverbal method to assess suicidality in adolescent and adult patients**. *Crisis: The Journal of Crisis Intervention and Suicide Prevention* (2013.0) **34** 131-6. DOI: 10.1027/0227-5910/a000164
19. Viguera AC, Milano N, Laurel R, Thompson NR, Griffith SD, Baldessarini RJ. **Comparison of Electronic Screening for Suicidal Risk With the Patient Health Questionnaire Item 9 and the Columbia Suicide Severity Rating Scale in an Outpatient Psychiatric Clinic**. *Psychosomatics* (2015.0) **56** 460-9. DOI: 10.1016/j.psym.2015.04.005
20. Forkmann T, Spangenberg L, Rath D, Hallensleben N, Hegerl U, Kersting A. **Assessing suicidality in real time: A psychometric evaluation of self-report items for the assessment of suicidal ideation and its proximal risk factors using ecological momentary assessments**. *Journal of Abnormal Psychology* (2018.0) **127** 758-69. DOI: 10.1037/abn0000381
21. Ellis TE, Rufino KA, Allen JG. **A controlled comparison trial of the Collaborative Assessment and Management of Suicidality (CAMS) in an inpatient setting: Outcomes at discharge and six-month follow-up**. *Psychiatry Research* (2017.0) **249** 252-60. DOI: 10.1016/j.psychres.2017.01.032
22. Garza NDL, Rush AJ, Killian MO, Grannemann BD, Carmody TJ, Trivedi MH. **The Concise Health Risk Tracking Self-Report (CHRT-SR) assessment of suicidality in depressed outpatients: A psychometric evaluation**. *Depression and Anxiety* (2019.0) **36** 313-20. DOI: 10.1002/da.22855
23. Stefansson J, Nordström P, Jokinen J. **Suicide Intent Scale in the prediction of suicide**. *Journal of Affective Disorders* (2012.0) **136** 167-71. DOI: 10.1016/j.jad.2010.11.016
24. Freedenthal S.. **Assessing the Wish to Die: A 30-Year Review of the Suicide Intent Scale**. *Arch Suicide Res* (2008.0) **12** 277-98. DOI: 10.1080/13811110802324698
25. Posner K, Oquendo MA, Gould M, Stanley B, Davies M. **Columbia Classification Algorithm of Suicide Assessment (C-CASA): Classification of Suicidal Events in the FDA’s Pediatric Suicidal Risk Analysis of Antidepressants**. *Am J Psychiatry* (2007.0) **164** 1035-43. DOI: 10.1176/ajp.2007.164.7.1035
26. Lindh ÅU, Beckman K, Carlborg A, Waern M, Salander Renberg E, Dahlin M. **Predicting suicide: A comparison between clinical suicide risk assessment and the Suicide Intent Scale**. *Journal of Affective Disorders* (2020.0) **263** 445-9. DOI: 10.1016/j.jad.2019.11.131
27. Beck RW, Morris JB, Beck AT. **Cross-Validation of the Suicidal Intent Scale**. *Psychol Rep* (1974.0) **34** 445-6. DOI: 10.2466/pr0.1974.34.2.445
28. Huang C, Wang Y, Li X, Ren L, Zhao J, Hu Y. **Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China**. *The Lancet* (2020.0) **395** 497-506
29. Reger MA, Stanley IH, Joiner TE. **Suicide Mortality and Coronavirus Disease 2019—A Perfect Storm?**. *JAMA Psychiatry* (2020.0) **77** 1093. DOI: 10.1001/jamapsychiatry.2020.1060
30. Zortea TC, Brenna CTA, Joyce M, McClelland H, Tippett M, Tran MM. **The impact of infectious disease-related public health emergencies on suicide, suicidal behavior, and suicidal thoughts: A systematic review**. *Crisis: The Journal of Crisis Intervention and Suicide Prevention* (2020.0)
31. Cheung YT, Chau PH, Yip PSF. **A revisit on older adults suicides and Severe Acute Respiratory Syndrome (SARS) epidemic in Hong Kong**. *International Journal of Geriatric Psychiatry* (2008.0) **23** 1231-8. DOI: 10.1002/gps.2056
32. Ammerman BA, Burke TA, Jacobucci R, McClure K. **Preliminary investigation of the association between COVID-19 and suicidal thoughts and behaviors in the U.S**. *Journal of Psychiatric Research* (2021.0) **134** 32-8. DOI: 10.1016/j.jpsychires.2020.12.037
33. Akiba CF, Go V, Mwapasa V, Hosseinipour M, Gaynes BN, Amberbir A. **The Sub-Saharan Africa Regional Partnership (SHARP) for Mental Health Capacity Building: a program protocol for building implementation science and mental health research and policymaking capacity in Malawi and Tanzania**. *Int J Ment Health Syst* (2019.0) **13**. DOI: 10.1186/s13033-019-0327-2
34. 34Interim Briefing Note Addressing Mental Health and Psychosocial Aspects of COVID-19 Outbreak (developed by the IASC’s Reference Group on Mental Health and Psychosocial Support) | IASC [Internet]. Geneva: Inter-Agency Standing Committee (IASC); 2020
Feb [cited 2021 Jan 20] p. 15. https://interagencystandingcommittee.org/iasc-reference-group-mental-health-and-psychosocial-support-emergency-settings/interim-briefing. *Interim Briefing Note Addressing Mental Health and Psychosocial Aspects of COVID-19 Outbreak (developed by the IASC’s Reference Group on Mental Health and Psychosocial Support) | IASC* (2020.0) 15
35. Gaynes BN, Akiba CF, Hosseinipour MC, Kulisewa K, Amberbir A, Udedi M. **The Sub-Saharan Africa Regional Partnership (SHARP) for Mental Health Capacity-Building Scale-Up Trial: Study Design and Protocol**. *Psychiatr Serv* (2021.0) **72** 812-21. DOI: 10.1176/appi.ps.202000003
36. Proctor E, Silmere H, Raghavan R, Hovmand P, Aarons G, Bunger A. **Outcomes for Implementation Research: Conceptual Distinctions, Measurement Challenges, and Research Agenda**. *Adm Policy Ment Health* (2011.0) **38** 65-76. DOI: 10.1007/s10488-010-0319-7
37. Karsh B.. **Beyond usability: designing effective technology implementation systems to promote patient safety**. *Qual Saf Health Care* (2004.0) **13** 388-94. DOI: 10.1136/qhc.13.5.388
38. Cholera R, Gaynes BN, Pence BW, Bassett J, Qangule N, Macphail C. **Validity of the patient health questionnaire-9 to screen for depression in a high-HIV burden primary healthcare clinic in Johannesburg, South Africa**. *Journal of Affective Disorders* (2014.0) **167** 160-6. DOI: 10.1016/j.jad.2014.06.003
39. Gilbody S, Richards D, Brealey S, Hewitt C. **Screening for Depression in Medical Settings with the Patient Health Questionnaire (PHQ): A Diagnostic Meta-Analysis**. *J GEN INTERN MED* (2007.0) **22** 1596-602. DOI: 10.1007/s11606-007-0333-y
40. Monahan PO, Shacham E, Reece M, Kroenke K, Ong’or WO, Omollo O. **Validity/Reliability of PHQ-9 and PHQ-2 Depression Scales Among Adults Living with HIV/AIDS in Western Kenya**. *J GEN INTERN MED* (2008.0) **24** 189. DOI: 10.1007/s11606-008-0846-z
41. Udedi M, Muula AS, Stewart RC, Pence BW. **The validity of the patient health Questionnaire-9 to screen for depression in patients with type-2 diabetes mellitus in non-communicable diseases clinics in Malawi**. *BMC Psychiatry* (2019.0) **19**. DOI: 10.1186/s12888-019-2062-2
42. 42Open data kit: Proceedings of the 4th ACM/IEEE International Conference on Information and Communication Technologies and Development [Internet]. [cited 2021 Feb 21]. https://dl.acm.org/doi/10.1145/2369220.2369236
43. 43StataCorp. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC; 2019.. *Stata Statistical Software: Release 16* (2019.0)
44. 44WhatsApp Messaging [Internet]. Mountain View, California: Facebook, Inc.; 2020. https://www.whatsapp.com/?lang=en
45. 45The impact of COVID-19 on mental, neurological and substance use services [Internet]. [cited 2021 Jan 11]. https://www.who.int/publications-detail-redirect/978924012455
46. Dunn SL, DeVon HA, Collins EG, Luong A, Buursma MP, Gutierrez-Kapheim M. **Suicide Risk Management Protocol for a Randomized Controlled Trial of Cardiac Patients Reporting Hopelessness**. *Nursing Research* (2021.0) **70** 72-9. DOI: 10.1097/NNR.0000000000000474
47. Park SW, Lee JH, Lee EK, Song JJ, Park HS, Hwang SY. **Development of the Suicide Risk Scale for Medical Inpatients**. *Journal of Korean Medical Science* (2018.0) **33**. DOI: 10.3346/jkms.2018.33.e18
48. Arias SA, Zhang Z, Hillerns C, Sullivan AF, Boudreaux ED, Miller I. **Using Structured Telephone Follow-up Assessments to Improve Suicide-Related Adverse Event Detection**. *Suicide and Life-Threatening Behavior* (2014.0) **44** 537-47. DOI: 10.1111/sltb.12088
49. Herbeck Belnap B, Schulberg HC, He F, Mazumdar S, Reynolds CF, Rollman BL. **Electronic protocol for suicide risk management in research participants**. *Journal of Psychosomatic Research* (2015.0) **78** 340-5. DOI: 10.1016/j.jpsychores.2014.12.012
50. Large M, Kaneson M, Myles N, Myles H, Gunaratne P, Ryan C. **Meta-Analysis of Longitudinal Cohort Studies of Suicide Risk Assessment among Psychiatric Patients: Heterogeneity in Results and Lack of Improvement over Time**. *PLoS One* (2016.0) **11** e0156322. DOI: 10.1371/journal.pone.0156322
51. Park S, Lee Y, Youn T, Kim BS, Park JI, Kim H. **Association between level of suicide risk, characteristics of suicide attempts, and mental disorders among suicide attempters**. *BMC Public Health* (2018.0) **18**. DOI: 10.1186/s12889-018-5387-8
52. Exbrayat S, Coudrot C, Gourdon X, Gay A, Sevos J, Pellet J. **Effect of telephone follow-up on repeated suicide attempt in patients discharged from an emergency psychiatry department: a controlled study**. *BMC Psychiatry* (2017.0) **17**. DOI: 10.1186/s12888-017-1258-6
53. Miller IW, Camargo CA, Arias SA, Sullivan AF, Allen MH, Goldstein AB. **Suicide Prevention in an Emergency Department Population: The ED-SAFE Study**. *JAMA Psychiatry* (2017.0) **74** 563. DOI: 10.1001/jamapsychiatry.2017.0678
54. Aranda-Jan CB, Mohutsiwa-Dibe N, Loukanova S. **Systematic review on what works, what does not work and why of implementation of mobile health (mHealth) projects in Africa**. *BMC Public Health* (2014.0) **14** 188. PMID: 24555733
55. Haberer JE, Kiwanuka J, Nansera D, Wilson IB, Bangsberg DR. **Challenges in Using Mobile Phones for Collection of Antiretroviral Therapy Adherence Data in a Resource-Limited Setting**. *AIDS Behav* (2010.0) **14** 1294-301. DOI: 10.1007/s10461-010-9720-1
56. Chang LW, Kagaayi J, Arem H, Nakigozi G, Ssempijja V, Serwadda D. **Impact of a mHealth Intervention for Peer Health Workers on AIDS Care in Rural Uganda: A Mixed Methods Evaluation of a Cluster-Randomized Trial**. *AIDS Behav* (2011.0) **15** 1776. DOI: 10.1007/s10461-011-9995-x
57. Runeson B, Odeberg J, Pettersson A, Edbom T, Adamsson IJ, Waern M. **Instruments for the assessment of suicide risk: A systematic review evaluating the certainty of the evidence**. *PLoS One* (2017.0) **12** e0180292. DOI: 10.1371/journal.pone.0180292
58. Bolton JM, Gunnell D, Turecki G. **Suicide risk assessment and intervention in people with mental illness**. *BMJ* (2015.0) **351** h4978. DOI: 10.1136/bmj.h4978
59. Glenn CR, Millner AJ, Esposito EC, Porter AC, Nock MK. **Implicit Identification with Death Predicts Suicidal Thoughts and Behaviors in Adolescents**. *J Clin Child Adolesc Psychol* (2019.0) **48** 263-72. DOI: 10.1080/15374416.2018.1528548
60. Greenwald AG, McGhee DE, Schwartz JLK. **Measuring individual differences in implicit cognition: The implicit association test**. *Journal of Personality and Social Psychology* (1998.0) **74** 1464-80. DOI: 10.1037//0022-3514.74.6.1464
|
---
title: Modification of the structural stability of human serum albumin in rheumatoid
arthritis
authors:
- Hsien-Jung L. Lin
- David H. Parkinson
- J. Connor Holman
- W. Chad Thompson
- Christian N. K. Anderson
- Marcus Hadfield
- Stephen Ames
- Nathan R. Zuniga Pina
- Jared N. Bowden
- Colette Quinn
- Lee D. Hansen
- John C. Price
journal: PLOS ONE
year: 2023
pmcid: PMC10022781
doi: 10.1371/journal.pone.0271008
license: CC BY 4.0
---
# Modification of the structural stability of human serum albumin in rheumatoid arthritis
## Abstract
Differential scanning calorimetry (DSC) can indicate changes in structure and/or concentration of the most abundant proteins in a biological sample via heat denaturation curves (HDCs). In blood serum for example, HDC changes result from either concentration changes or altered thermal stabilities for 7–10 proteins and has previously been shown capable of differentiating between sick and healthy human subjects. Here, we compare HDCs and proteomic profiles of 50 patients experiencing joint-inflammatory symptoms, 27 of which were clinically diagnosed with rheumatoid arthritis (RA). The HDC of all 50 subjects appeared significantly different from expected healthy curves, but comparison of additional differences between the RA and the non-RA subjects allowed more specific understanding of RA samples. We used mass spectrometry (MS) to investigate the reasons behind the additional HDC changes observed in RA patients. The HDC differences do not appear to be directly related to differences in the concentrations of abundant serum proteins. Rather, the differences can be attributed to modified thermal stability of some fraction of the human serum albumin (HSA) proteins in the sample. By quantifying differences in the frequency of artificially induced post translational modifications (PTMs), we found that HSA in RA subjects had a much lower surface accessibility, indicating potential ligand or protein binding partners in certain regions that could explain the shift in HSA melting temperature in the RA HDCs. Several low abundance proteins were found to have significant changes in concentration in RA subjects and could be involved in or related to binding of HSA. Certain amino acid sites clusters were found to be less accessible in RA subjects, suggesting changes in HSA structure that may be related to changes in protein-protein interactions. These results all support a change in behavior of HSA which may give insight into mechanisms of RA pathology.
## Introduction
Rheumatoid arthritis (RA) is a systemic inflammatory autoimmune disease characterized by non-articular changes, symmetrical polyarthritis, and congenital symptoms [1, 2]. Despite the prevalence of RA, the classification for the disease is considered definite only after the confirmed presence of chronic inflammation of the connective tissue in one joint, no reasonable alternative diagnosis, and scoring 6 or greater across the four different characterization domains (number of joints involved, abnormal antibody count, elevated acute-phase response, and duration of symptoms [3]). The late and tentative diagnosis of RA is mainly due to the poorly understood etiology of the disease and a complex interplay between genetic and environmental factors. Effective RA management is correlated with early and aggressive treatment [2]. Therefore, it remains crucial to develop an accurate, quick, and inexpensive way to diagnose RA, preferably without a tissue biopsy. The prognosis of RA patients depends heavily on early diagnosis since current treatments only relieve symptoms and slow progress, but do not cure the disease. Thus, the earlier the diagnosis, the better the prognosis for the patient.
Several low abundance proteins in human serum, such as C-reactive protein [4], rheumatoid factor (RF) [5], anti-citrullinated peptide antibodies (ACPA) [6], and anti-keratin antibody (AKA) [7] have been investigated for detection of pre-RA symptoms [5], but none have been found to serve as a biomarker for RA initiation. These proteins correlate with autoimmunity, but collectively make up an extremely minor percentage of human serum (<<$1\%$), which often results in low sensitivity [8]. Other metabolites, such as glucose [9], high-density lipoprotein (HDL) cholesterol [10] and vitamin D [11], have also been implicated in RA pathogenesis, but have not been used for diagnosis.
Current diagnostic tests for RA consist of measuring serum concentrations of rheumatoid factor (RF) and cyclic citrullinated peptide (CCP). Although RF concentration is widely used and the most accepted test for RA serologic diagnosis, it is not specific for RA [12]. Elevated RF can be found in many other diseases, including other autoimmune diseases (Sjogren’s syndrome, systemic lupus erythematosus [13, 14]), chronic infections, cardiovascular disease, cancer, and normal aging [15]. The sensitivity and specificity of RF for RA diagnosis are $62\%$ and $89\%$ respectively, and CCP’s sensitivity and specificity for RA diagnosis are 53–$58\%$ and 95–$96\%$ respectively [8]. The RF and CCP tests are useful, but diagnosis often cannot occur until the disease has progressed significantly. Thus, more information about the causes of RA is needed to detect and intervene in RA development earlier and more accurately.
In this study, we compare RA-positive (RA) patients to RA-negative (non-RA) patients, all within a group of 50 who all came in for clinical testing because they were experiencing RA-like symptoms (Fig 1). Comparing RA and non-RA subjects within a cohort in which all subjects were symptomatic allowed us to determine which proteomic changes come from RA-specific pathology, rather than generic inflammatory factors. The complexity of successful RA diagnosis (due to many confounding factors) demonstrates the difficulty of determining RA-specific characteristics of the disease. By using this experimental design where all of our subjects are symptomatic (only some of which are RA-positive), we hope to reduce the confounding effect of comorbidities so that we can detect RA-specific differences that promote further investigation. Here, we show our findings outlining RA-specific serum proteome changes and we propose a model for an RA-induced increase in HSA stability through potential binding partners, which we hope will offer valuable guidance in future research looking for better RA treatments and diagnostic models.
**Fig 1:** *Experimental flow.Blood serum samples from subjects that had physician ordered RA panels were used (n = 50). Based on clinical diagnosis from medical professionals, the samples were separated into an RA group (n = 27) and a non-RA group (n = 23). DSC was used to obtain the HDCs, and a characteristic shift was seen between groups. LC/MS-MS experiments were performed to determine the mechanism behind this shift. Quantification and surface amino acid reactivity analyses were performed to determine significant differences in serum proteins between RA and non-RA groups.*
A potential method to understand disease-specific pathology is from obtaining patient serum calorimetric thermograms (herein referred to as heat denaturation curves, HDCs) using Differential Scanning Calorimetry (DSC) [16–21]. HDC averages the melting temperature of proteins in serum samples. HDCs of serum from patients with several different diseases have been shown to exhibit reproducible shifts in the pattern of protein heat denaturation that are unique for those diseases [22–31]. Although no mechanistic information is obtained, these differences must arise from changes in the concentrations and/or structures of the most abundant proteins (~8 proteins) in the serum [32]. In this study, HDCs were used to characterize the altered serum proteins in RA patients (clinically diagnosed according to symptoms and various biomarker levels). A characteristic HDC shift was seen across samples, and there was a significant relationship between RA diagnosis and HDC appearance. HDC from normal, healthy patients typically contain two peaks, one around 62.8 and 69.8°C [22]. We saw a distinct decrease in the intensity of the lower temperature peak in RA samples. Mass spectrometry (MS) was then used to identify proteomic differences in the RA vs. non-RA samples, and we also looked for correlations between the proteome and HDC appearance (grouping samples by peak ratio, independent of RA diagnosis). Together, these results allowed us to understand which changes in the RA proteome could be attributed to the observed HDC differences.
After noticing the distinct differences in HDC pattern across samples, we considered two possible mechanisms to explain this HDC shift (Fig 2). First, changes in concentration of abundant proteins, such as human serum albumin (HSA), would alter the intensity of high abundance protein peaks, changing overall HDC shape [22, 33] (Fig 2A). In this study, we focus primarily on HSA because it is the most abundant protein in plasma. Second, HDC shape would be significantly altered by a change in thermal stability of abundant proteins, shifting their melting temperatures. For example, loading HSA with a fatty acid (octanoic acid) increases the melting temperature by 5 to 10°C [34]. Such changes in stability would most likely correlate with a change in tertiary structures of a smaller fraction of the total HSA [35] (Fig 2B). We used MS to explore both concentration differences in HSA (and other abundant proteins), as well as HSA tertiary structure changes (by looking at surface reactivity [36, 37]) as potential causes of the characteristic shift in the HDCs. As outlined below, our data support a change in the thermal stability of a portion of the HSA (Fig 2B). This change in thermal stability, along with MS-detected differences, provides clues for HSA structural changes [38, 39].
**Fig 2:** *Two possible models explaining HDC shifts in RA subjects.The decreased intensity of the lower temperature peak in RA samples could be explained by A) a relative HSA concentration decrease, reducing HDC signal intensity, or B) a shift in thermal stability for a portion of the HSA population, shifting the HSA peak to the right, decreasing the first peak’s intensity, and increasing that of the second peak. This thermal stability could be the result of altered binding partners. The structural changes can be seen through biased detection of surface modifications on HSA. When cargo is unbound (top right, light green protein), binding sites are surface accessible for modification, when cargo is bound (top right, dark green protein), these sites are occupied, reducing surface accessibility and ability of these sites to be modified. These structural and functional differences at the molecular level could be the explanation for the observed shift in HDC.*
## Results and discussion
We used a sample population of fifty anonymized serum samples from patients who experienced joint inflammatory symptoms. Since the samples we used were anonymized post-analysis, and the data has no connection to the subjects, no IRB approval was needed. Patients ranged from age 12 to 88, with a median age of 50. Samples were not selected based on gender and the resulting sample set contained males ($$n = 13$$) and females ($$n = 37$$), matching statistical prevalence of RA [40]. The rheumatoid arthritis panel [41] was conducted for the serum samples by ARUP Laboratories, including a rheumatoid factor (RF) and cyclic citrullinated peptide (CCP) test. Professional medical analysis of symptoms, paired with the CCP and RF levels, classified the 50 samples as coming from RA ($$n = 27$$) and non-RA ($$n = 23$$) subjects (S1 Data). Note that the serology results show only some of the factors used for RA clinical diagnosis. Other factors (joint involvement, acute phase reactants, and symptoms duration, etc. [ 42]) were used for diagnosis, but the supplementary diagnostic information was not provided for this study.
## Heat denaturation curves
Heat denaturation curves (HDC) were collected with a NanoDSC (TA Instruments, Lindon, UT). Forty-seven HDCs were obtained (HDCs for three subjects were uninterpretable due to errors during the sample injection). As seen in literature, HDCs for healthy subjects have two distinct peaks around 63 degrees and 71 degrees, which correspond to the known melting points for HSA and immunoglobulin proteins, respectively [30]. Previous studies showed the low temperature peak at 63°C is primarily a combination of HSA and haptoglobin (HAPT), in which HSA dominates due to its much higher concentration [30]. The high temperature peak at 71°C is primarily a combination of Immunoglobulin G (IgG) and Immunoglobulin A (IgA) underlain by the tail of the broad HSA peak [30]. As shown in Fig 3A, the non-RA HDCs show a smaller peak ratio (1.00 ± 0.23), and the RA HDCs show an even more substantial decrease in peak ratio (0.83 ± 0.16). Healthy normal serum samples are reported in the literature with a higher HSA (low temp) peak at 63°C and a comparatively lower Ig (high temp) peak at 71°C [30]. A separate study from Garbett et al. shows the 63 to 71 degree peak ratio in a cohort of healthy samples has a much greater intensity for the low temperature peak (~1.5), reaffirming that none of the current samples can be described as healthy [22]. A two-tailed t-test yields a p-value of 0.007 between RA and non-RA subjects, indicating that the HDC peak ratios are statistically different (Fig 3A, S1 and S2 Figs in S1 File, S2 Data). This pattern is consistent with literature and can be seen in other auto-immune disorders such as lupus [23, 24]. Similar to literature [25, 26], when the HDCs were ranked according to peak ratio (regardless of RA diagnosis), they could be separated into two groups that correlated with the RA diagnosis: low peak ratio (LPR, peak ratio < 1.00, $$n = 32$$), and high peak ratio (HPR, peak ratio > 1.00, $$n = 15$$) (Fig 3B). Associating the HPR group with non-RA and the LPR group with RA gives a point-biserial correlation coefficient of 0.3966, meaning that $39.66\%$ of the variability in peak ratio can be attributed to the RA diagnosis. With this association, a threshold ratio of 1.00 splits the samples (for the 47 HDCs obtained) with the smallest misclassification rate ($27.7\%$). Using this threshold, 22 of the 32 samples ($68.8\%$) in the LPR group are classified as RA while 12 of the 15 samples ($80.0\%$) in the HPR groups are classified as non-RA. While 22 of the 27 RA samples ($81.5\%$) are in the LPR group, and 12 of the 22 non-RA samples ($54.5\%$) are in the HPR group (S1 Data). These classification rates are likely impacted by the imperfect specificity and sensitivity of RA diagnosis mentioned earlier, as well as the presence of comorbidities in RA and non-RA subjects.
**Fig 3:** *DSC results.This study focuses on the two peaks observed between 55 and 75˚C of the heat denaturation curve. (A) The average normalized HDC curve for non-RA and RA samples, with the difference between the two shown in black. The first peak from HSA is consistently found around 63°C (low temp peak) and the second Ig peak is always around 71°C (high temp peak). Inset for A shows the distribution of peak ratios from the HDC of RA and non-RA subjects. The difference in peak ratio between the non-RA and RA groups is statistically significant (p = 0.007) (B) The distribution of peak ratio of all samples, with a peak ratio threshold of 1.00 as the cut-off between the HPR and LPR groups.*
Several of the non-RA samples are categorized in the LPR group, and this could be the result of other diseases or physiological differences [30] that alter HSA and other serum proteins, such as Lyme Disease, Lupus, or diabetes [22, 43]. This seems likely given that all 50 subjects originally came in for testing because they were experiencing symptoms of discomfort and sickness. We are interested in mechanisms behind these HDC shifts (Fig 2), so we used MS to evaluate the differences between both the RA/non-RA subjects and the HPR/LPR groups.
Blood serum samples ($$n = 50$$) were obtained from ARUP Laboratories. Samples were prepared in random order for Nano Differential Scanning Calorimetry DSC measurements by first filtering with a 0.45-micron filter. After being degassed, 40 μL of the blood serum was diluted with 960 μL of buffer. The buffer used for dilution was 10 mM phosphate-buffered saline (PBS) (138 mM NaCl, 2.7 mM KCl at pH 7.50). Samples were refrigerated at 4°C until Nano DSC scans were made. Samples were prepared ten at a time and loaded into the Nano DSC autosampler at 5°C. Samples were scanned from 20° to 110°C at 1 ⁰C/min after a 600 second equilibration period after loading and corrected against a reference cell. The remainder of the undiluted serum samples were used for MS analysis to look for changes in protein concentration, as well as PTM frequency and location.
## Protein concentrations
The 50 serum samples were individually digested to tryptic peptides and analyzed using mass spectrometry to further explore the difference in protein content between RA and non-RA serum samples. Relative protein quantification analysis (PEAKS Studio_8.5, Bioinformatics Solutions Inc. [44], S3 Data) shows there are no significant differences in protein concentration between RA and non-RA groups or HPR and LPR groups for any of the top eight most abundant proteins (significant changes are defined as proteins with a fold change less than 0.5 fold or greater than 2 and a p-value less than 0.05) (Fig 4, S1 Table in S1 File). The concentration fold change for each protein in each comparison was calculated by evaluating the ratio of RA abundance to non-RA abundance and LPR abundance to HPR abundance. The RA, non-RA, LPR, and HPR protein abundances were defined as the average MS intensity across the samples of each respective group. It is expected that specific autoantibody concentrations would increase in patients with RA [45–47], but since the RA antigen specific Ig population is a relatively small percentage of the entire Ig population, and significant sequence homology exists between immunoglobulins, it is difficult to distinguish target-specific antibodies using MS only. Also, the comparison was not against "healthy" controls, so that lack of significance in Ig could likely be because an Ig increase, non-specific to RA, may have occurred across many of the samples, elevating Ig levels altogether. These results suggest that a change in concentration of abundant serum proteins does not contribute to the decreased HDC peak ratio observed in RA samples.
**Fig 4:** *Protein concentration differences in RA.Volcano plots indicating the fold change and p-value for all 421 detected proteins, comparing A) RA/Non-RA samples and (B) LPR/HPR samples. The top eight most abundant proteins are indicated in green (all insignificant), and the statistically significant proteins (-1 > log2(fold change) > 1, p-value < 0.05) are indicated in red. C-reactive protein is the only significant protein in both plots. The fold change is calculated in each comparison by dividing RA abundance by non-RA abundance and LPR abundance by HPR abundance for all proteins that were detected in all samples.*
Aside from the top eight proteins, overall proteomic analysis showed that among the 421 proteins compared, a statistically significant change was only seen in five proteins when comparing RA to non-RA samples (Fig 4A) and 14 proteins when comparing the HPR and LPR groups (Fig 4B) The only common significant protein between these two groups was C-reactive protein (CRP), a protein known to be associated with systemic inflammation. The significance of CRP in both the RA vs. Non-RA and HPR vs. LPR comparisons indicate that CRP may be involved in mechanisms accounting for the HDC shift seen in RA samples. It is important to note that CRP concentration is most likely upregulated for all subjects relative to healthy controls as has been described previously, but it is significantly lower in both LPR and RA groups. Vitamin D binding protein (VDBP, known to be related to RA [11]) was significantly downregulated in RA samples, and had no significant changes between HPR and LPR groups. These results suggests that although VDBP and other proteins may be associated with RA, their relatively low concentration means they are not directly affecting the HDC shift in RA samples. However, the change in the concentration of these proteins may affect our measurements because they have interactions with the very abundant proteins measured in the HDC (Fig 2B).
## Structural changes in high abundance proteins
Since protein concentration doesn’t directly account for the difference in HDCs between RA and non-RA samples, and there is also no link between concentration and the HPR and LPR groups. Therefore, we expect, similar to other diseases explored in literature, that the observed HDC shifts among RA patients and the LPR group are caused by changes in thermal stability for one of the most abundant serum proteins. We simulated shifts in the abundance and/or melting temperatures of various percentages of each of the top eight serum proteins (using individually measured HDCs of these abundant proteins from literature [22]) to recapitulate the observed changes. This simulation was simply a hypothesis-generating technique and tested what the resulting HDC would look like after altering the abundance and/or melting temperature of each of the top eight proteins in the non-RA HDCs. We tested an abundance of 25, 50, 75, 150, 200, and $500\%$ and/or a shift in the melting temperature of -15, -10, -5, 5, 10, and 15°C. We also performed each of these simulations on various percentages of the total protein present (5, 10, 20, 50, 95, $100\%$). By comparing the simulated HDC shift to the difference curve shown in Fig 3A, and by visually analyzing the similarity of the shifted curve and the RA curve (S3 Fig in S1 File), we found the most plausible explanation for the shift to be an increase in the melting temperature for a small fraction (~$10\%$) of the HSA pool by about 5–$15\%$. Changes in HSA melting temperature could result from new ligand binding, protein interactors, or changes in tertiary structure [48–50]. Therefore, we tested for structural changes of HSA through analysis of covalently modified amino acid profiles between the RA and non-RA samples. Both biological and artificially induced modifications were considered. Changes in biological modifications could show altered RA biochemistry, and changes in artificially induced modifications would show variations in surface accessibility of certain regions of a protein. If RA-specific protein conformation changes are responsible for changes in the HDCs, we also expect these amino acid modification (AAmod) profiles between RA and non-RA groups, to be correlated with the observed HDC groups (HPR and LPR).
Protein Prospector (UCSF) and PEAKS studio (Bioinformatics Solutions Inc) were both used for contrasting analysis of the PTM data. Multiple peptide modifications were observed as noncanonical m/z shifts with Protein Prospector, including a modification of +183 m/z, which was the most frequently observed modification (41 peptides) on HSA (S4 Data). PEAKs Studio’s analysis of HSA proteins and each amnio acid modification (AAmod) confirmed the +183 m/z modification as an aminoethylbenzenesulfonylflouride modification (AEBSF) which came from the protease inhibitor cocktail added before processing the serum. Thus, AEBSF was an artificially induced, non-biological PTM. HSA had 185 modified sites that were observed in more than 12 of the samples. Of the 185 total AAmod sites on HSA, there were 33 observed modification types, with the top ten most frequent being AEBSF, 42; Dehydration, 28; Hexose, 17; Deamidation, 14; Iodination, 14; Oxidation, 9; Citrulline, 8; Formylation, 5; Amidation, 5; and Di-iodination, 4. $71\%$ of these AAmod sites are specific for only one type of modification (S4 Data). AEBSF was the only AAmod that showed statistically significant differences between RA and non-RA groups (S5 Data). Since the AEBSF modification was synthetically introduced, it is not causing the change in HSA structure but is reporting the fact that the in vivo structure was changed for these reactive sites. Non-RA subjects have, on average, 1.9 times more AEBSF modifications than RA subjects ($$p \leq 0.023$$). Since there were significantly fewer AEBSF modifications in RA subjects, it suggests that AAmod sites are less accessible in RA HSA, suggesting conformational changes or a potential increase in binding partners in RA HSA.
## AEBSF as a probe of surface reactivity
AEBSF is an irreversible serine protease inhibitor which can react with surface accessible nucleophilic amino acids such as Serine (S), Lysine (K), Tyrosine (Y), Histidine (H), and the amino-terminus (Fig 5A) [51, 52]. Like other good surface modifiers (diethyl pyrocarbonate [37] or diazonium salt [36]), we can use its prevalence to identify changes in surface area accessibility of individual amino acids on proteins between samples. The AEBSF modifications observed on HSA were most frequently observed on lysine (28 different lysine residues), tyrosine (9 different residues), serine (2 different residues), and histidine (2 different residues) (S5 Data).
**Fig 5:** *AEBSF modification of HSA.(A) The chemical reaction of the AEBSF modification on Serine. The reaction is similar for other nucleophilic amino acids. (B) The heatmap generated with PNNL Inferno showing the intensity differences of AEBSF modification at different HSA sites between different samples. The AEBSF modification amino acid number for HSA is listed on the y-axis, and the serum sample number on the x-axis. The samples are separated into 4 groups according to the hierarchy branch of serum samples, from left to right (S5 Data). Group L1 and L2’s AEBSF modifications are less intense than group H1 and H2 (L stands for lower intensity and H stands for higher intensity). The three clusters, C1 (green), C2 (purple), and C3 (black), are the most intense AEBSF modification clusters and are examined to characterize the modification further. (C) The bar graph shows the number of RA/non-RA and HPR/LPR samples expected in each AEBSF modification group (L1, H1, L2, H2). The percentage of RA samples in L1, H1, L2, H2, is 73%, 25%, 69%, and 31% respectively. For LPR, it is 67%, 89%, 73%, and 50%, respectively.*
To visualize patterns in AEBSF modification between samples and across AAmod sites, we used PNNL Inferno [53] to generate a hierarchical grouped heatmap from the patient-specific ion intensities for each modification site (Fig 5B). From our MS data, samples were sorted into clusters with serum samples on the horizontal axis and HSA modification sites on the vertical axis. The hierarchical order separated the samples into 4 groups. Of the 4 groups, two groups have higher signal intensity (H1 & H2), and two groups have lower signal intensity (L1 & L2). H1 has higher signal intensity at sites shown in clusters C1 and C2, H2 has higher signal intensity at the specific sites in cluster C3, and groups L1 and L2 have lower signal intensity across all sites. While various regions of the heatmap could raise interest, the clusters C1, C2, and C3 were selected for further analysis due to their especially high signal. Higher signal intensity indicates a greater level of AEBSF modification, which implies a greater degree of surface accessibility. Each of these four groups (L1, H1, L2, and H2) are made up of $32.6\%$, $17.4\%$, $23.9\%$, and $26.0\%$ of the serum samples, respectively. Therefore, in Fig 5C, we can visualize our data against the null hypothesis that the same percentage of RA and non-RA samples should be present in each group. Also, given the proportion of HPR samples within the non-RA group and the proportion of LPR samples within the RA group, we can visualize the null hypothesis of how random assignment would distribute the HPR and LPR samples into each subset of the 4 groups (given the size of the L1, H1, L2, and H2 subset groups) (null hypothesis, Ho, Fig 5C). As shown in Fig 5C, we found that a much greater proportion of the RA samples were found in the L1 and L2 groups ($42\%$ and $35\%$, respectively) compared to the non-RA samples ($17\%$ and $17\%$, respectively). A lower percentage of RA samples were in the H1 and H2 groups ($8\%$ and $15\%$, respectively) compared to the non-RA samples ($26\%$ and $39\%$, respectively). The L1 and L2 groups contained close to the expected proportion of samples from the HPR and LPR groups, but in the H2 group (containing a high proportion of non-RA samples), we saw a higher-than-expected percentage of HPR samples ($40\%$ of HPR samples were in the H2 group, compared to $19\%$ of LPR samples). However, the H1 group (also containing a high proportion of non-RA samples), had a higher-than-expected percentage of LPR samples ($23\%$ of LPR samples compared to $7\%$ of HPR samples). This suggests that the high AEBSF frequency at the AAmod sites in clusters C1 and C2, which are in H1 region, are connected to a decrease in HDC peak ratio but an RA-negative diagnosis. In fact, more intense AEBSF modifications at these C1 and C2 sites may give insight into why certain non-RA samples exhibited a low HDC peak ratio (increased surface accessibility from other factors not specific to RA). On the other hand, the high AEBSF frequency at the modification sites in clusters C3 are connected to both a higher HDC peak ratio and non-RA subjects (Fig 5B and 5C), indicating that decreased accessibility of the C3 amino acid binding sites seen in RA samples may be directly linked to the observed HDC shift seen in RA samples.
Together, this pattern suggests that HSA in the RA/LPR groups may have binding partners or other ligand interactors that block those C3 sites. Additionally, the association between HPR and RA samples in the H2 group suggest that the decreased accessibility of C1/C2 AAmod sites, likely due to binding partners or other conformational changes, are unlikely to be the cause of the increased HDC shifts observed in RA HSA. These binding partners could be related to other diseases that the RA-negative (yet still discomforted) patients were experiencing when they came in to be tested for RA.
It should be noted that the clustering in our heatmap in Fig 5B is data-driven using these 50 subjects as a training set. Therefore, statistical inference, error bars, and p-values are not appropriate as we analyze how the data in Fig 5C deviates from our null hypothesis. To test the hypothesis that these patterns can be applied to a population with statistical confidence additional groups of non-RA and RA would need to be collected and compared to our clustered model. Therefore, we compared these observations against previously published literature to gain valuable insight into the connection between these AEBSF groups and the modification sites in relation to specific changes in HSA tertiary structure.
## Potential binding surfaces on has
The 3-dimensional structure of HSA has three recognized domains (I, II and III), each with two subdomains ((IA, IB, IIA, IIB, IIIA, IIIB) [54]. There are also nine known binding pockets for long chain fatty acids distributed throughout the three domains. Two drug and drug-like molecule binding sites, Sudlow sites I and II are located in domains IIA and IIIA [55, 56], respectively (Fig 6B). The HSA structure and the AAmod sites for each of the three clusters was visualized with UCSF Chimera (version 1.15) [57], with C1 sites in blue, C2 sites in red, and C3 sites in green (Fig 6A). AEBSF modification sites in C1 and C3 are mostly in domain II: $70\%$ and $55\%$, respectively. AEBSF modification sites in C2 and are mostly in domain I ($50\%$) (Fig 6C, S5 Data). Sudlow Site I (IIA) has the most frequently observed ($33\%$) AEBSF modification sites from all three clusters combined (Fig 6B, S5 Data). The modified amino acids in C1 and C3 are mostly lysine, and mostly tyrosine in C2 (Fig 6D, S5 Data). PyRosetta [58] was used to extract the secondary structure and surface accessible surface area (SASA) scores from a representative crystal structure of HSA (PDB ID: 1N5U [59]); $81\%$ of the modification sites are on an α-helix, and $19\%$ are on a loop (Fig 6E, S5 Data). The average SASA scores of C1, C2, and C3 are 95.9 ± 37.2, 37.1 ± 32.5, and 81.8 ± 37.3 (Fig 6F, S5 Data) where a larger value indicates more surface accessibility.
**Fig 6:** *Characterizing AEBSF modification site of the 3 clusters (C1, C2, C3) on HSA.(A) A representative HSA crystal structure (PDB ID: 1N5U) with the 3 AEBSF modification sites clusters colored. The three clusters, C1, C2, and C3 are colored in blue, red, and green, respectively. Individual C1 and C2 sites that are significantly less accessible in RA HSA are labeled. The red oval indicates a C3-rich region in domain I that could be a plausible binding site for RA-specific interactions that most likely to increase HSA stability. (B) The HSA structure is colored by its 3 domains (I, II, III), and subdomains (A, B). The number of AEBSF modification sites regardless of cluster designation in each subdomain is listed in parentheses. The 9 known cargo binding pockets are shown in the gray circles, and the two drug binding sites, Sudlow I & II, are shown by an arrow. The four bottom-right panels show what percentage of the AEBSF sites in each cluster (C) is in each HSA domain, (D) is on each amino acid residue, (E) has each secondary structure (Helix (H) or Loop (L)), and (F) the average SASA score of each cluster. Only SASA scores between C1 and C2 are statistically different (p = 0.019).*
When each AEBSF modification intensity is compared between RA and non-RA subjects in the C1, C2, and C3 clusters, all appear to be less accessible in RA HSA. Multiple two-sample t-tests (p-value adjusted) comparing the modification intensity between RA and non-RA samples reveal that three C1 sites (Y263, K359, and H367, in subdomain IIB) and two C2 sites (Y401 and Y497, in subdomain IIIB) have p-values below 0.05, indicating potential RA-specific binding sites (Table 1, indicated in Fig 6A). These significant sites are labeled in Fig 6A. As explained for Fig 5C, we do not expect potential binding partners at these C1 and C2 sites to increase HSA thermal stability because they are associated with more HPR samples. This hypothesis is strengthened by the fact that most C1 and C2 sites (particularly the significant ones) appear on the more outer surfaces of HSA (Fig 6A) and potentially mobile helices (Fig 6E), making them less likely to have a significant impact on overall HSA stability. On the other hand, C3 sites appear to be more concentrated to inner folds of HSA, where a large number of core interactions would need to be broken during denaturation. This, along with the fact that the C3 cluster is associated with more LPR samples, aligns with the hypothesis that decreased surface accessibility at these sites is a marker of RA-specific HSA stabilization and a decreased peak ratio (Fig 5C). Interestingly, no individual C3 sites show statistically significant differences between RA and non-RA groups (Table 1), but as a whole, we see that there is an enrichment in non-RA subjects with high C3 sites (Fig 5C). This suggests that the HSA structure is modified by dynamic surface interactors like other proteins, rather than covalently cross-linked molecules. Covalently cross-linked proteins or molecules would be more likely to show significantly decreased accessibility at exact sites. On the other hand, our data (less site-specific changes and more area specific changes) indicates larger, more regional surface interactions.
**Table 1**
| Group Name | AA | Uniprot Position | 1N5U position | HSA subdomain | SS | SASA | AEBSF mod Fold Change (Non-RA/RA) | p-value | Unnamed: 9 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| C1 | K | 205.0 | 181.0 | IB | H | 62.1 | 1.7 | 0.22 | |
| C1 | K | 229.0 | 205.0 | IIA | H | 121.2 | 1.25 | 0.44 | |
| C1 | K | 249.0 | 225.0 | IIA | L | 93.0 | 2.06 | 0.09 | |
| C1 | K | 257.0 | 233.0 | IIA | H | 79.6 | 1.3 | 0.51 | |
| C1 | Y | 287.0 | 263.0 | IIA | H | 52.4 | 3.22 | 0.05 | * |
| C1 | K | 383.0 | 359.0 | IIB | H | 156.8 | 1.82 | 0.05 | * |
| C1 | H | 391.0 | 367.0 | IIB | H | 72.1 | 2.19 | 0.01 | * |
| C1 | K | 402.0 | 378.0 | IIB | H | 140.5 | 1.06 | 0.82 | |
| C1 | K | 456.0 | 432.0 | IIIA | H | 57.6 | 1.27 | 0.38 | |
| C1 | K | 499.0 | 475.0 | IIIA | H | 124.1 | 1.56 | 0.14 | |
| C2 | Y | 108.0 | 84.0 | IA | H | 26.2 | 1.84 | 0.25 | |
| C2 | Y | 162.0 | 138.0 | IB | H | 66.9 | 1.72 | 0.24 | |
| C2 | Y | 164.0 | 140.0 | IB | H | 5.3 | 2.19 | 0.26 | |
| C2 | Y | 377.0 | 353.0 | IIB | H | - | 2.15 | 0.2 | |
| C2 | Y | 425.0 | 401.0 | IIIA | H | 79.9 | 3.01 | 0.05 | * |
| C2 | Y | 521.0 | 497.0 | IIIA | L | 44.5 | 3.34 | 0.05 | * |
| C3 | K | 36.0 | 12.0 | IA | H | 139.3 | 1.63 | 0.14 | |
| C3 | K | 117.0 | 93.0 | IA | L | 76.1 | 1.37 | 0.31 | |
| C3 | K | 160.0 | 136.0 | IB | H | 68.4 | 1.48 | 0.17 | |
| C3 | K | 214.0 | 190.0 | IIA | H | 131.0 | 1.41 | 0.29 | |
| C3 | K | 223.0 | 199.0 | IIA | H | 51.6 | 1.02 | 0.95 | |
| C3 | K | 305.0 | 281.0 | IIA | L | 80.6 | 1.29 | 0.54 | |
| C3 | S | 311.0 | 287.0 | IIA | H | 19.7 | 2.54 | 0.1 | |
| C3 | H | 312.0 | 288.0 | IIA | H | 40.0 | 1.54 | 0.46 | |
| C3 | K | 375.0 | 351.0 | IIB | H | 119.5 | 1.42 | 0.18 | |
| C3 | Y | 476.0 | 452.0 | IIIA | H | 88.9 | 1.52 | 0.16 | |
| C3 | K | 588.0 | 564.0 | IIIB | L | 84.2 | 1.09 | 0.81 | |
| C3 | | | | | | | | | |
The most significantly altered C3 amino acid residue (in terms of surface accessibility) is S287. Compared to RA subjects, non-RA subjects had 2.54 times as much AEBSF modification at the S287 site ($$p \leq 0.10$$). In the PDB structure, S287 already appears quite buried in HSA (Fig 6), and its SASA score is only 19.7, the lowest of all C3 sites (Table 1). The next four most altered C3 sites are K12, H288, Y452, and K136 (all near S287)–sites at which non-RA subjects have 1.63, 1.54, 1.52, and 1.48 times as much AEBSF modification (p-values are 0.14, 0.46, 0.16, and 0.17, respectively). Seven of the C3 sites (including these top 5) are in a small, localized area (oval shaped magnification in Fig 6A) in domain I that could be a plausible binding interface with RA-specific interactors. Binding interactions here could increase the thermal stability of HSA and reduce the surface reactivity of these sites.
## Conclusions
In agreement with literature on other diseases [30, 32], we found that the HDC of serum are characteristically shifted (shown by a decreased first/second peak ratio) in all subjects experiencing inflammatory symptoms. Interestingly, RA subjects displayed an even lower peak ratio compared to non-RA subjects, suggesting a more pronounced HDC shift. In comparison, all 15 of the healthy control subjects used by Garbett et al. [ 22] fall into the HPR group (peak ratio > 1.00, Table 2), $54.5\%$ of non-RA subjects, but only $18.5\%$ RA subjects fell into the HPR group. Our data is consistent with the literature showing that concentrations of the top 8 proteins do not change significantly during RA or other cases of inflammation [60]. Our data supports the proposed mechanism in Fig 2B, that an increase in HSA stability (5–15°C increase in melting temperature for ~$10\%$ of HSA) would be a more plausible explanation for difference between non-RA and RA HDCs. CRP, known to defend against infectious agents and play a significant role in the inflammatory response [4, 61], is the only protein with a significantly different concentration among both comparisons. Both groups are expected to have elevated CRP concentrations, but relative concentrations are less elevated in both RA and LPR groups, compared to non-RA and HPR groups. At the same time, we observed that RA subjects are underrepresented in the C1, C2, and C3 clusters indicating that RA-positive HSA is less accessible compared to that of non-RA subjects, but LPR only appears to be underrepresented in within the C3 cluster, indicating that it is at those sites in particular that the RA-specific decrease in HSA accessibility is associated with the characteristic HDC shift.
**Table 2**
| Quantitative metric | Healthy (literature) | Non-RA (experimental) | RA (experimental) |
| --- | --- | --- | --- |
| HDC ratio | | 1.00 ± 0.23 | 0.83 ± 0.16 |
| Peak Ratio Group | 100% HPR [22] | 54.5% HPR45.5% LPR | 18.5% HPR81.5% LPR |
| Concentration of top 8 abundant proteins | Baseline, similar to RA and symptomatic patients [60] | No change compared to RA | No change compared to non-RA |
| CRP concentration | Baseline, low compared to non-RA and RA [4] | High | Elevated, but lower than non-RA |
| C1 | Binding previously observed:K205 [78]K225 [79]K233 [80]K359 [81]H367 [80]K378 [80]K432 [79, 85, 87] | More surface reactivity compared to RA;significant on sites Y263, K359, H367 | Lower surface reactivity compared to non-RA;appears to be associated with a higher thermal stability |
| C1 | Reactivity previously observed:K181 [90], K263 [91], K475 [90] | More surface reactivity compared to RA;significant on sites Y263, K359, H367 | Lower surface reactivity compared to non-RA;appears to be associated with a higher thermal stability |
| C2 | Binding previously observed:Y84 [82]Y138 [83, 85]Y353 [84]Y401 [85]Y497 [86] | More surface reactivity compared to RA;significance on sites Y401, Y497 | Lower surface reactivity compared to non-RA;appears to be associated with a higher thermal stability |
| C2 | Reactivity previously observed:Y140 [92] | More surface reactivity compared to RA;significance on sites Y401, Y497 | Lower surface reactivity compared to non-RA;appears to be associated with a higher thermal stability |
| C3 | Binding previously observed:K12 [79]K136 [88]K190 [76, 79, 85]K199 [76, 79, 85, 87]K281 [77]S287 [76]H288 [89]K351 [78, 79, 81, 85]Y452 [87] | More surface reactivity, suggesting a lost binding interface that exposes K12, K136, K199, K281, S287, H288, and Y452 | Lower surface reactivity compared to non-RA;appears to be associated with a higher thermal stability |
| C3 | Reactivity previously observed:K93 [90], K564 [90] | More surface reactivity, suggesting a lost binding interface that exposes K12, K136, K199, K281, S287, H288, and Y452 | Lower surface reactivity compared to non-RA;appears to be associated with a higher thermal stability |
These findings suggest a model consistent with Fig 2B. CRP, produced predominantly by hepatocytes in response to stimulation by IL-6, is known to be a promiscuous interactor and recruiter of proteins [62, 63]. For example, CRP binding to immunoglobulin *Fc gamma* receptors (FcgR) promotes the production of proinflammatory cytokines leading to inflammation [61]. It is possible that during inflammation in non-RA subjects, high levels of CRP associate with HSA binding proteins, changing the structural stability of this portion of the HSA. Since CRP is decreased in RA subjects, these potential has interactors would be more available to HSA, specifically at the C3 site surface interface shown in Fig 6A, increasing HSA stability, and reducing surface accessibility.
The HSA BioGRID 4.4 interactome database shows 349 known interactors for HSA and 72 interactors for CRP. Three of these proteins are common to both HSA and CRP: 1-acylglycerol-3-phosphate O-acyltransferase 1 (AGPAT1), complement factor H (CFH), and fibronectin 1 (FN1). Some studies have shown details about CRP’s relationship with AGPAT1 [64, 65], but extensive studies have shown the details of CRP’s interactions with CFH [66–69] and FN1 [70–75]. The specific location of these proteins interactions with HSA is less understood. Nevertheless, most of the C3 modification sites such as those in the plausible domain I binding pocket (K12, K199, K281, S287, H288, and Y452) are known to be high affinity binding sites for drugs or other protein interactors [76–89]. Table 2 shows which of the C1, C2, and C3 sites have been associated with ligands, drug, or protein binding sites in previous studies [76–89]. For all other sites, reactivity has still been observed [90–92].
To better understand RA pathology and changes in HSA stability, future RA studies should look for potential binding partners by extracting lipids and small molecules from purified serum HSA. Other directions to explore the mechanisms that increase HSA stability are: [1] Using more specific surface modifications or chemical crosslinking reagents to carry out in-depth surface probing of HSA, collect specific information about HSA binding partners and coordination changes [93, 94], and [2] comparing HSA and CRP protein binding partners in RA and non-RA patients using immunoaffinity purification together with mass spectrometry to understand how a change in CRP concentration could be contributing to HSA interactor changes. Future research into HSA and other related proteins will continue to enhance our understanding of RA-specific pathology and give insights into the development of, and potential treatments for, RA.
## Calculating the peak values
Calorimetry experimental results were first corrected for the instrument baseline by subtracting a buffer injection control. Nonzero baselines were then corrected by applying a linear baseline between minimum at 25°C and 82°C. Scans were finally normalized for the volume of protein injected (supplemental information). We then looked at the raw HDC curve between 25 and 100°C, setting the minimum of each HDC as 0 and the maximum as 100. This allowed us to take the peak ratio from two positive values. The low temperature peak value (HSA peak) was measured at 63°C, and the high temperature peak value (Ig peak) was measured at 71°C. The HSA/Ig peak ratio was then calculated.
## Protein digestion
The serum samples were denatured with 6M guanidinium chloride (GdmCl) in 100mM Tris/HCl (pH 8.5) and protease inhibitor (Sigma-Aldrich, cat #: P8340), then spun at 21,000xg for 20 minutes at 4°C to remove insoluble cell contents. The supernatant, which contains soluble proteins, was then transferred into new tubes. The BCA assay (Thermo Fisher Scientific cat #: 23227) protocol was followed to measure the protein concentration in each sample. 1.5 μL serum, which contained about 50 μg of protein, was diluted to 50 μL in 1X PBS, and combined with 100 μL 6 M GdmCl. Each sample was transferred into a new tube, then 1.2 μL of 200 mM dithiothreitol (DTT, >$99\%$ sigma # D-5545) in water was added (final concentration 5mM) and the mixture was incubated at 55°C in a sand bath for 15 minutes. The mixture was then cooled for 5 minutes to reduce disulfide bonding. We then added 3.8 μL of 200 mM freshly made iodoacetamide (IAM, $97\%$ sigma # I-670-9) in water (final concentration 15 mM) and incubated for 1 hour at room temperature in the dark to alkylate the reduced proteins.
Next, samples were put onto 30 kDa centrifugal filters and spun at 14,000 g for 10 minutes. Then 100 μL 6M GdmCl in 100mM Tris/HCl (pH 8.5) was added, and the samples were spun at 14,000xg. This was repeated twice. Then 100 μL 25 mM ammonium bicarbonate (ABC) was added, and the samples were spun again at 14,000 g, this was repeated twice. Next, we emptied and cleaned the collection tube with ddH2O three times and 100 μL 25 mM ABC was added to the top of the filter.
MS trypsin (Promega gold MS sequencing grade Trypsin #V5111) was added to the solution above the filter in a 1:50 (w/w) trypsin/protein ratio and the samples were incubated at 37°C overnight on a shaker. After that, each sample was quenched with 300 mM phenylmethylsulfonyl fluoride (PMSF, final concentration 1 mM). Samples were then centrifuged at 14,000xg for 30 minutes, 100 μL of 25 mM ABC was added, and the samples centrifuged again at 14,000 g for 30 minutes. The filtrate was collected in mass spec vials, dried with a Speedvac, and resuspended in $3\%$ acetonitrile (ACN), $0.1\%$ formic acid (FA) to 1 μg/μL.
## Mass spectrometry acquisition for proteomics
Data for the 50 samples was acquired in a randomized order. Digested peptides were separated on a Polaris-HR-C18 HPLC chip in a chip cube nano spray source using an Agilent 1260 HPLC followed by positive ESI and mass detection using an Agilent QTOF mass spectrometer (6530B). The mobile phases consisted of MS grade $3\%$ acetonitrile, $0.1\%$ formic acid for Buffer A; and $97\%$ acetonitrile, $0.1\%$ formic acid for Buffer B. A 50-minute gradient was run at 0.3μL/min flow rate: $0\%$-$5\%$ B Buffer (0–0.5 minutes), $5\%$-$30\%$ B Buffer (0.5–27 minutes), $30\%$-$95\%$ B buffer (27–28 minutes), $95\%$ B Buffer (28–31 minutes), $95\%$ - $5\%$ B buffer (31–33 minutes), $5\%$ - $95\%$ B buffer (33–35 minutes), $95\%$ - $0\%$ B buffer (35–46 min), $0\%$ B (36–49 min). Auto MS/MS fragmentation using variable collision energy determined by ion mass from 290–1700 m/z at 4 spectra/s rate and 250ms/spectrum time, and with an isolation width around 4 m/z. The auto MS/MS method selected precursor ions that were above 2500 counts and have charge state 2 and above for fragmentation. MS/MS scan range 100–1700 m/z, and 10 max processors allowed per cycle. The same spectra were excluded from the MS/MS selection for 0.2 min. This prevented continual acquisition of the same m/z and allowed for other, less abundant species to be acquired by the mass spectrometer.
## Protein identification and quantification
Protein identification and quantification were performed with two programs. The first was Protein Prospector developed in the University of California San Francisco Mass Spectrometry Facility, funded by NIH National Institute for General Medical Sciences. The second one was PEAKs Studio 8.5, developed Bioinformatics Solutions Inc. Both programs compared peptide fragmentation against the SwissProt human database downloaded in August 2017 with the following parameters: monoisotopic for precursor mass search type; semispecific for digest mode, 3 missed cleavage allowed; 20 ppm for parent mass error tolerance; 0.5 Da for fragment mass error tolerance; 3 max variable PTM per peptide allowed, with carbamidomethlyation as fixed modification, and oxidation, Pyro-glu from Q and other 9 customized PTM as variable modification (detailed listed in S4 Data) in PEAKS DataBase step; 311 built-in ptm was used in the PEAKS PTM step; 20ppm mass error tolerance and 3 min retention time shift tolerance were used in the label free quantification step. The raw data are available for download at the chorusporject.org (project ID: 1739, experiment ID:3632).
## Protein structure analysis
Some analyses performed with UCSF Chimera, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from NIH P41-GM103311.
## Inferno hierarchical clustering and heatmap analysis
The heatmap for AEBSF on HSA was created using InfernoRDN created by Pacific Northwest National Laboratory (PNNL, [53]). PTM sites were identified and quantified using PEAKS Studio. PTM sites that were at least present in 12 samples were included in *Heatmap* generation. Files were then loaded into InfernoRDN and Log2 transformed to reduce the noise of outliers in later analysis. A dual-clustered Heatmap was generated with the standard Euclidean modeling parameters. The hierarchical order output was then used to determine the most changed PTM sites between samples and subsequent PTM site groupings. Additionally, the dual-clustering setting allowed for groups to be observed across samples which were statistically examined for correlation with RA diagnosis.
## References
1. Pauk J, Wasilewska A, Ihnatouski M. **Infrared Thermography Sensor for Disease Activity Detection in Rheumatoid Arthritis Patients**. (2019.0) **19**
2. Majithia V, Geraci SA. **Rheumatoid arthritis: diagnosis and management**. (2007.0) **120** 936-9. DOI: 10.1016/j.amjmed.2007.04.005
3. Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT, Bingham CO 3rd. **2010 Rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative**. (2010.0) **62** 2569-81. DOI: 10.1002/art.27584
4. He Y, Tang J, Wu B, Yang B, Ou Q, Lin J. **Correlation between albumin to fibrinogen ratio, C-reactive protein to albumin ratio and Th17 cells in patients with rheumatoid arthritis**. (2020.0) **500** 149-54. DOI: 10.1016/j.cca.2019.10.009
5. Deane KD, Holers VM. **The Natural History of Rheumatoid Arthritis**. (2019.0) **41** 1256-69. DOI: 10.1016/j.clinthera.2019.04.028
6. Paul BJ, Kandy HI, Krishnan V. **Pre-rheumatoid arthritis and its prevention**. (2017.0) **4** 161-5. DOI: 10.5152/eurjrheum.2017.16006
7. Wang XP, Cheng QY, Gu MM, Leng RX, Fan YG, Li BZ. **Diagnostic accuracy of anti-keratin antibody for rheumatoid arthritis: a meta-analysis**. (2019.0) **38** 1841-9. DOI: 10.1007/s10067-019-04464-x
8. Taylor P, Gartemann J, Hsieh J, Creeden J. **A systematic review of serum biomarkers anti-cyclic citrullinated Peptide and rheumatoid factor as tests for rheumatoid arthritis**. (2011.0) **2011** 815038. DOI: 10.4061/2011/815038
9. Pi H, Zhou H, Jin H, Ning Y, Wang Y. **Abnormal Glucose Metabolism in Rheumatoid Arthritis**. (2017.0) **2017** 9670434. DOI: 10.1155/2017/9670434
10. Erum U, Ahsan T, Khowaja D. **Lipid abnormalities in patients with Rheumatoid Arthritis**. (2017.0) **33** 227-30. DOI: 10.12669/pjms.331.11699
11. Harrison SR, Li D, Jeffery LE, Raza K, Hewison M. **Vitamin D, Autoimmune Disease and Rheumatoid Arthritis**. (2020.0) **106** 58-75. DOI: 10.1007/s00223-019-00577-2
12. Silveira IG, Burlingame RW, von Muhlen CA, Bender AL, Staub HL. **Anti-CCP antibodies have more diagnostic impact than rheumatoid factor (RF) in a population tested for RF**. (2007.0) **26** 1883-9. DOI: 10.1007/s10067-007-0601-6
13. Ingegnoli F, Castelli R, Gualtierotti R. **Rheumatoid factors: clinical applications**. (2013.0) **35** 727-34. DOI: 10.1155/2013/726598
14. Westwood OM, Nelson PN, Hay FC. **Rheumatoid factors: what’s new?**. (2006.0) **45** 379-85. DOI: 10.1093/rheumatology/kei228
15. Ahn JK, Hwang J, Chang Y, Ryu S. **Rheumatoid factor positivity increases all-cause and cancer mortality: a cohort study**. (2017.0) **37** 1135-43. PMID: 28516237
16. Schneider G., Kaliappan A., Nguyen T. Q., Buscaglia R.. **The Utility of Differential Scanning Calorimetry Curves of Blood Plasma for Diagnosis, Subtype Differentiation and Predicted Survival in Lung Cancer**. (2021.0) **13** 5326. DOI: 10.3390/cancers13215326
17. Kim N. A., Jin J. H., Kim K. H., Lim D. G.. **Investigation of early and advanced stages in ovarian cancer using human plasma by differential scanning calorimetry and mass spectrometry**. (2016.0) **39** 668-676. DOI: 10.1007/s12272-016-0722-z
18. Fekecs T., Zapf I., Ferencz A., Lᔛrinczy D. **Differential scanning calorimetry (DSC) analysis of human plasma in melanoma patients with or without regional lymph node metastases**. (2012.0) **108** 149-152
19. Zapf I., Fekecs T., Moezzi M., Tizedes G.. **Differential scanning calorimetry of blood plasma in breast cancer patients**. (2012.0) **56** 274-279. PMID: 23236598
20. Ferencz A., Lᔛrinczy D.. **DSC measurements of blood plasma on patients with chronic pancreatitis and operable and inoperable pancreatic adenocarcinoma**. (2017.0) **127** 1187-1192
21. Fish D. J., Brewood G. P., Kim J. S., Garbett N. C.. **Statistical analysis of plasma thermograms measured by differential scanning calorimetry**. (2010.0) **152** 184-190. DOI: 10.1016/j.bpc.2010.09.007
22. Garbett NC, Miller JJ, Jenson AB, Chaires JB. **Calorimetry outside the box: a new window into the plasma proteome**. (2008.0) **94** 1377-83. DOI: 10.1529/biophysj.107.119453
23. Garbett N. C., Brock G. N., Chaires J. B., Mekmaysy C. S.. **Characterization and classification of lupus patients based on plasma thermograms**. (2017.0) **12** e0186398. DOI: 10.1371/journal.pone.0186398
24. Kendrick S. K., Zheng Q., Garbett N. C., Brock G. N.. **Application and interpretation of functional data analysis techniques to differential scanning calorimetry data from lupus patients**. (2017.0) **12** e0186232. DOI: 10.1371/journal.pone.0186232
25. Rai S. N., Pan J., Cambon A., Chaires J., Garbett N.. **Group classification based on high-dimensional data: application to differential scanning calorimetry plasma thermogram analysis of cervical cancer and control samples**. (2013.0) **3** 1-9
26. Rai S. N., Srivastava S., Pan J., Wu X.. **Multi-group diagnostic classification of high- dimensional data using differential scanning calorimetry plasma thermograms**. (2019.0) **14** e0220765. DOI: 10.1371/journal.pone.0220765
27. Michnik A, Drzazga Z, Michalik K, Barczyk A, Santura I, Sozańska E. **Differential scanning calorimetry study of blood serum in chronic obstructive pulmonary disease**. (2009.0) **102** 57-60
28. Chagovetz AA, Jensen RL, Recht L, Glantz M, Chagovetz AM. **Preliminary use of differential scanning calorimetry of cerebrospinal fluid for the diagnosis of glioblastoma multiforme**. (2011.0) **105** 499-506. DOI: 10.1007/s11060-011-0630-5
29. Garbett N. C., Brock G. N.. **Differential scanning calorimetry as a complementary diagnostic tool for the evaluation of biological samples**. (2016.0) **1860** 981-989. DOI: 10.1016/j.bbagen.2015.10.004
30. Garbett NC, Mekmaysy CS, DeLeeuw L, Chaires JB. **Clinical application of plasma thermograms. Utility, practical approaches and considerations**. *Methods* (2015.0) **76** 41-50. DOI: 10.1016/j.ymeth.2014.10.030
31. Sasahara K, Goto Y. **Application and use of differential scanning calorimetry in studies of thermal fluctuation associated with amyloid fibril formation**. *Biophys. Rev* (2013.0) **5** 259-69. PMID: 28510108
32. Garbett NC, Mekmaysy CS, Helm CW, Jenson AB, Chaires JB. **Differential scanning calorimetry of blood plasma for clinical diagnosis and monitoring**. (2009.0) **86** 186-91. PMID: 19146849
33. Johnson CM. **Differential scanning calorimetry as a tool for protein folding and stability**. (2013.0) **531** 100-9. PMID: 23022410
34. Lang BE, Cole KD. **Unfolding Properties of Recombinant Human Serum Albumin Products Are Due To Bioprocessing Steps**. (2015.0) **31** 62-9
35. Durowoju IB, Bhandal KS, Hu J, Carpick B, Kirkitadze M. **Differential Scanning Calorimetry—A Method for Assessing the Thermal Stability and Conformation of Protein Antigen**. (2017.0)
36. Nielsen JB, Nielsen AV, Carson RH, Lin HL, Hanson RL, Sonker M. **Analysis of thrombin-antithrombin complex formation using microchip electrophoresis and mass spectrometry**. (2019.0) **40** 2853-9. PMID: 31373007
37. Limpikirati P, Pan X, Vachet RW. **Covalent Labeling with Diethylpyrocarbonate: Sensitive to the Residue Microenvironment, Providing Improved Analysis of Protein Higher Order Structure by Mass Spectrometry**. (2019.0) **91** 8516-23. PMID: 31150223
38. Garbett N. C., Merchant M. L., Helm C. W., Jenson A. B.. **Detection of cervical cancer biomarker patterns in blood plasma and urine by differential scanning calorimetry and mass spectrometry**. (2014.0) **9** e84710. PMID: 24416269
39. Geyer P. E., Holdt L. M., Teupser D., Mann M.. **Revisiting biomarker discovery by plasma proteomics**. (2017.0) **13** 942. PMID: 28951502
40. de Brito Rocha S, Baldo DC, Andrade LEC. **Clinical and pathophysiologic relevance of autoantibodies in rheumatoid arthritis**. (2019.0) **59** 2. DOI: 10.1186/s42358-018-0042-8
41. 41Laboratories A. ARUP Test Directory: Rheumatoid Arthritis Panel 2021 [Available from: https://ltd.aruplab.com/Tests/Pub/2003277.
42. 42Laboratories A. ARUP Consult: Rheumatoid Arthritis—RA 2021 [Available from: https://arupconsult.com/content/rheumatoid-arthritis#toc-serology.
43. Garbett NC, Merchant ML, Chaires JB, Klein JB. **Calorimetric analysis of the plasma proteome: identification of type 1 diabetes patients with early renal function decline**. (2013.0) **1830** 4675-80. DOI: 10.1016/j.bbagen.2013.05.007
44. Tran NH, Zhang X, Xin L, Shan B, Li M. **De novo peptide sequencing by deep learning**. (2017.0) **114** 8247-52. DOI: 10.1073/pnas.1705691114
45. Min J-K. **The Role of Immunoglobulin G4 in Patients With Rheumatoid Arthritis**. (2020.0) **27** 125-7
46. Aho K, Heliovaara M, Knekt P, Reunanen A, Aromaa A, Leino A. **Serum immunoglobulins and the risk of rheumatoid arthritis**. (1997.0) **56** 351-6. DOI: 10.1136/ard.56.6.351
47. Lin G, Li J. **Elevation of serum IgG subclass concentration in patients with rheumatoid arthritis**. (2010.0) **30** 837-40. DOI: 10.1007/s00296-009-1330-8
48. Koslen M. M., Eskew M. W., Pinkert V., Hoang H.. **Capture Reagent and Strategy for Retrieving Albumin-Bound Ligands from Plasma**. (2019.0) **9**
49. Eskew M. W., Koslen M. M., Benight A. S.. **Ligand binding to natural and modified human serum albumin**. (2021.0) **612** 113843. DOI: 10.1016/j.ab.2020.113843
50. Eskew M. W., Benight A. S.. **Ligand binding constants for human serum albumin evaluated by ratiometric analysis of DSC thermograms**. (2021.0) **628** 114293. DOI: 10.1016/j.ab.2021.114293
51. Whetzel PL, Noy NF, Shah NH, Alexander PR, Nyulas C, Tudorache T. **BioPortal: enhanced functionality via new Web services from the National Center for Biomedical Ontology to access and use ontologies in software applications**. (2011.0) **39** W541-W5. DOI: 10.1093/nar/gkr469
52. 52Protein Modification Ontology (PSIMOD)-AEBSF [Internet]. 2021 [cited April 8, 2020]. Available from: https://bioportal.bioontology.org/ontologies/PSIMOD?p=classes&conceptid=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FMOD_00596.
53. Polpitiya AD, Qian WJ, Jaitly N, Petyuk VA, Adkins JN, Camp DG 2nd. **DAnTE: a statistical tool for quantitative analysis of -omics data**. (2008.0) **24** 1556-8. DOI: 10.1093/bioinformatics/btn217
54. Guizado TRC, Louro SRW, Anteneodo C. **Dynamics of heme complexed with human serum albumin: a theoretical approach**. (2012.0) **41** 1033-42. DOI: 10.1007/s00249-012-0860-2
55. Arroyo V, Garcia-Martinez R, Salvatella X. **Human serum albumin, systemic inflammation, and cirrhosis**. (2014.0) **61** 396-407. DOI: 10.1016/j.jhep.2014.04.012
56. Hauenschild T, Reichenwallner J, Enkelmann V, Hinderberger D. **Characterizing Active Pharmaceutical Ingredient Binding to Human Serum Albumin by Spin-Labeling and EPR Spectroscopy**. (2016.0) **22** 12825-38. PMID: 27460503
57. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC. **UCSF Chimera—a visualization system for exploratory research and analysis**. (2004.0) **25** 1605-12. PMID: 15264254
58. Chaudhury S, Lyskov S, Gray JJ. **PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta**. (2010.0) **26** 689-91. PMID: 20061306
59. Wardell M, Wang ZM, Ho JX, Robert J, Ruker F, Ruble J. **The atomic structure of human methemalbumin at 1.9 angstrom**. (2002.0) **291** 813-9
60. Doherty N. S., Littman B. H., Reilly K., Swindell A. C., Buss J. M., Anderson N. L.. **Analysis of changes in acute-phase plasma proteins in an acute inflammatory response and in rheumatoid arthritis using two-dimensional gel electrophoresis**. (1998.0) **19** 355-363. DOI: 10.1002/elps.1150190234
61. Pope J. E., Choy E. H.. **C-reactive protein and implications in rheumatoid arthritis and associated comorbidities**. (2021.0) **51** 219-229. DOI: 10.1016/j.semarthrit.2020.11.005
62. Boncler M., Dudzińska D., Nowak J., Watała C.. **Modified C-Reactive Protein Selectively Binds to Immunoglobulins**. (2012.0) **76** 1-10. DOI: 10.1111/j.1365-3083.2012.02704.x
63. Alberts A., Klingberg A., Wessig A. K., Combes C., Witte T., Brand K.. **C-reactive protein (CRP) recognizes uric acid crystals and recruits proteases C1 and MASP1**. (2020.0) **10** 6391. DOI: 10.1038/s41598-020-63318-8
64. Bruce K. D., Cagampang F. R., Argenton M., Zhang J., Ethirajan P. L., Burdge G. C.. **Maternal high‐fat feeding primes steatohepatitis in adult mice offspring, involving mitochondrial dysfunction and altered lipogenesis gene expression**. (2009.0) **50** 1796-1808. DOI: 10.1002/hep.23205
65. Hegele R. A.. **Monogenic forms of insulin resistance: apertures that expose the common metabolic syndrome**. (2003.0) **14** 371-377. DOI: 10.1016/s1043-2760(03)00142-5
66. Mihlan M., Stippa S., Józsi M., Zipfel P. F.. **Monomeric CRP contributes to complement control in fluid phase and on cellular surfaces and increases phagocytosis by recruiting factor H**. (2009.0) **16** 1630-1640. DOI: 10.1038/cdd.2009.103
67. Okemefuna A. I., Nan R., Miller A., Gor J., Perkins S. J.. **Complement Factor H Binds at Two Independent Sites to C-reactive Protein in Acute Phase Concentrations**. (2010.0) **285** 1053-1065. DOI: 10.1074/jbc.M109.044529
68. Laine M., Jarva H., Seitsonen S., Haapasalo K., Lehtinen M. J., Lindeman N.. **Y402H Polymorphism of Complement Factor H Affects Binding Affinity to C-Reactive Protein**. (2007.0) **178** 3831-3836. DOI: 10.4049/jimmunol.178.6.3831
69. Mold C., Gewurz H., Du Clos T. W.. **Regulation of complement activation by C-reactive protein**. (1999.0) **42** 23. DOI: 10.1016/s0162-3109(99)00007-7
70. Salonen E. M., Vartio T., Hedman K., Vaheri A.. **Binding of fibronectin by the acute phase reactant C-reactive protein**. (1984.0) **259** 1496. PMID: 6693419
71. Suresh M. V., Singh S. K., Agrawal A.. **Interaction of Calcium-bound C-reactive Protein with Fibronectin Is Controlled by pH**. (2004.0) **279** 52552-52557. PMID: 15456743
72. Seyyed H., Hashemi; Mamani M, Hashemi S. H, Hajilooi M., Saedi F. **ORIGINAL REPORT Evaluation of Fibronectin and C-Reactive Protein Levels in Patients with Sepsis: A Case-Control Study.**. (2012.0) **50** 404. PMID: 22837119
73. Ullah N., Ma F., Han J., Liu X., Fu Y., Liu Y.. **Monomeric C-reactive protein regulates fibronectin mediated monocyte adhesion**. (2020.0) **117** 122-130. PMID: 31765841
74. Man S. F. P., Xing L., Connett J. E., Anthonisen N. R., Wise R. A., Tashkin D. P.. **Circulating fibronectin to C-reactive protein ratio and mortality: a biomarker in COPD?**. (2008.0) **32** 1451-1457. PMID: 18799503
75. Tseng J., Mortensen R. F.. **Binding of human C-reactive protein (CRP) to plasma fibronectin occurs via the phosphorylcholine-binding site**. (1988.0) **25** 679-686. PMID: 2460754
76. Yang F., Zhang Y., Liang H.. **Interactive Association of Drugs Binding to Human Serum Albumin**. (2014.0) **15** 3580-3595. DOI: 10.3390/ijms15033580
77. Barnaby O. S., Cerny R. L., Clarke W., Hage D. S.. **Comparison of modification sites formed on human serum albumin at various stages of glycation**. (2011.0) **412** 277-285. DOI: 10.1016/j.cca.2010.10.018
78. Gokara M., Malavath T., Kalangi S. K., Reddana P., Subramanyam R.. **Unraveling the binding mechanism of asiatic acid with human serum albumin and its biological implications**. (2013.0) **32** 1290. DOI: 10.1080/07391102.2013.817953
79. Fu F., Liu H., Gao R., Zhao P., Lu X., Zhang R.. **Protein adduct binding properties of tabun-subtype nerve agents after exposure in vitro and in vivo**. (2020.0) **321** 1-11. DOI: 10.1016/j.toxlet.2019.12.014
80. Szapacs M. E., Riggins J. N., Zimmerman L. J., Liebler D. C.. **Covalent Adduction of Human Serum Albumin by 4-Hydroxy-2-Nonenal: Kinetic Analysis of Competing Alkylation Reactions**. (2006.0) **45** 10521-10528. DOI: 10.1021/bi060535q
81. Adams R., Griffin L., Compson J. E., Jairaj M., Baker T., Ceska T.. **Extending the half-life of a fab fragment through generation of a humanized anti-human serum albumin Fv domain: An investigation into the correlation between affinity and serum half-life**. (2016.0) **8** 1336-1346. DOI: 10.1080/19420862.2016.1185581
82. Ketrat S., Japrung D., Pongprayoon P.. **Exploring how structural and dynamic properties of bovine and canine serum albumins differ from human serum albumin**. (2020.0) **98** 107601. DOI: 10.1016/j.jmgm.2020.107601
83. Zhu L., Yang F., Chen L., Meehan E. J., Huang M.. **A new drug binding subsite on human serum albumin and drug–drug interaction studied by X-ray crystallography**. (2008.0) **162** 40-49. DOI: 10.1016/j.jsb.2007.12.004
84. Li Wei-Ying, Chen Fang-Fang, Wang Shi-Long. **Binding of Reactive Brilliant Red to Human Serum Albumin: Insights into the Molecular Toxicity of Sulfonic Azo Dyes**. (2010.0) **17**. DOI: 10.2174/092986610791112756
85. Bhattacharya A. A., Grüne T., Curry S., Huber R.. **Crystallographic analysis reveals common modes of binding of medium and long-chain fatty acids to human serum albumin**. (2000.0) **303** 721-732. DOI: 10.1006/jmbi.2000.4158
86. Schmidt M., Townson S., Andreucci A., King B., Schirmer E., Murillo A.. **Crystal Structure of an HSA/FcRn Complex Reveals Recycling by Competitive Mimicry of HSA Ligands at a pH-Dependent Hydrophobic Interface**. (2013.0) **21** 1966-1978. DOI: 10.1016/j.str.2013.08.022
87. Awang T., Niramitranon J., Japrung D., Saparpakorn P., Pongprayoon P.. **Investigating the binding affinities of fructose and galactose to human serum albumin: simulation studies**. (2021.0) **47** 738
88. Garg A., Manidhar D. M., Gokara M., Malleda C., Suresh Reddy C., Subramanyam R.. **Elucidation of the Binding Mechanism of Coumarin Derivatives with Human Serum Albumin**. (2013.0) **8** e63805. DOI: 10.1371/journal.pone.0063805
89. Abdullah S. M. S., Fatma S., Rabbani G., Ashraf J. M.. **A spectroscopic and molecular docking approach on the binding of tinzaparin sodium with human serum albumin**. (2017.0) **1127** 283-288
90. Anguizola J., Matsuda R., Barnaby O. S., Hoy K. S., Wa C., DeBolt E.. **Review: Glycation of human serum albumin**. (2013.0) **425** 64-76
91. Huang Bill X., Kim Hee-Yong, Dass Chhabil. **Probing three-dimensional structure of bovine serum albumin by chemical cross-linking and mass spectrometry**. (2004.0) **15** 1237-1247. DOI: 10.1016/j.jasms.2004.05.004
92. Fu F., Liu H., Lu X., Zhang R., Li L., Gao R.. **Identification of S419 on human serum albumin as a novel biomarker for sarin and cyclosarin exposure**. (2020.0) **34** e8721. DOI: 10.1002/rcm.8721
93. Criddle RS, Lin HL, James I, Park JS, Hansen LD, Price JC. **Proposing a minimal set of metrics and methods to predict probabilities of amyloidosis disease and onset age in individuals**. (2020.0) **12** 22356-69. DOI: 10.18632/aging.202208
94. Lin H.-J. L., James I., Hyer C. D., Haderlie C. T., Zackrison M. J., Bateman T. M.. **Quantifying In Situ Structural Stabilities of Human Blood Plasma Proteins Using a Novel Iodination Protein Stability Assay**. (2022.0) **21** 2920-2935. DOI: 10.1021/acs.jproteome.2c00323
|
---
title: 'E-health psychological intervention in pregnant women exposed to intimate
partner violence (eIPV): A protocol for a pilot randomised controlled trial'
authors:
- Antonella Ludmila Zapata-Calvente
- Stella Martín-de-las-Heras
- Aurora Bueno Cavanillas
- Karen Andreasen
- Vibeke Rasch
- Khalid S. Khan
journal: PLOS ONE
year: 2023
pmcid: PMC10022801
doi: 10.1371/journal.pone.0282997
license: CC BY 4.0
---
# E-health psychological intervention in pregnant women exposed to intimate partner violence (eIPV): A protocol for a pilot randomised controlled trial
## Abstract
Intimate partner violence (IPV) during pregnancy, a condition as common as obstetrics conditions like gestational diabetes, is associated with maternal and neonatal complications. Systematic detection of IPV is not well established in antenatal screening probably because the effectiveness of protective interventions has not been evaluated. E-health interventions may be beneficial among mothers exposed to IPV. Prior to performing a full-scale effectiveness trial for such an intervention, a pilot study is required to assess the feasibility of randomising a sufficiently large number of women exposed to IPV during pregnancy. The eIPV trial is a randomised pilot study nested within a cohort of consenting mothers who screen positive for IPV in the first antenatal visit at <12 weeks’ gestation and accept an e-health package (psychological counselling by videoconference) in Spain and Denmark. Twenty eligible mothers from the above cohort will be randomised to either intervention or control. The intervention group will receive the e-health package as part of the cohort. The control group will be invited to accept a delay in the intervention (e-health package eight weeks later). After consenting to delay, the control group will provide comparative data without losing the opportunity of obtaining the intervention. We will determine estimates of rates of informed consent to randomization, and the rates of adherence and dropout following randomization. Qualitative interviews will be conducted to examine the women’s perception about the benefit of the intervention, reasons for acceptability and non-adherence, and obstacles to recruitment, randomisation and consent. The results will inform the trial feasibility and variance of key clinical outcome measures for estimation of sample size of the full-scale effectiveness trial.
## Background and rationale
Intimate partner violence (IPV) is one of the most common forms of violence against women and includes physical, sexual, and emotional abuse and controlling behaviour [1]. Globally, the lifetime prevalence of physical and sexual IPV for women is around $30\%$ [2, 3]. Violence during pregnancy is more common than preeclampsia or gestational diabetes which are routinely screened for antenatally [4, 5]. In Spain, $3.6\%$ of pregnant women have suffered physical IPV during pregnancy [6, 7]. In Denmark, $2.5\%$ of pregnant women have reported physical violence [8]. Although the true prevalence of IPV during pregnancy in EU is unclear, it is evident that a substantial minority of women experience violence during pregnancy [9]. Moreover, pregnancy violence often continues into the postpartum period with an estimated $10\%$ of hospitalizations due to injury in pregnancy are the result of intentional injuries inflicted upon the pregnant woman [10]. Also, IPV during pregnancy has been linked to depression, both during pregnancy and in the postpartum period [11].
To improve the health of pregnant women and their infants, it is important that targeted IPV interventions are developed and implemented as part of routine antenatal care.
E-health tools are potential solutions to deliver effective treatment and support to women exposed to IPV. eHealth interventions have the potential to provide more flexibility and safer spaces than face-to-face interventions, making it particularly useful for screening violence, empowering women and reducing exposure to IPV [12]. IPV survivors considered appropriate and necessary to be screened using eHealth strategies in antenatal care settings and perceived that video counselling is a viable and acceptable tool for pregnant women who experience IPV [13]. Trials have shown that e-health screening tools are effective in getting women to disclose or detect IPV [14–16]. Trials have also found that women who receive telephone counselling combined with other forms of support or a web-based safety decision aid kit are more likely to adopt safety behaviours than women in a control group, which were not offered e-health solutions [17–20]. It has been found that online safety planning is a feasible method among pregnant women [21].
Based on the available evidence, we have developed an e-health intervention package combining an online screening tool and video counselling sessions by trained providers for pregnant women who screen positive for IPV. Screening for IPV combined with an empowerment intervention that includes brief education and video counselling on safety planning may have potential to address repeated IPV and associated adverse health effects among pregnant women. However, prior to performing a large interventional trial, a pilot study is needed to identify barriers to recruitment, assess feasibility and acceptability of the treatment, and fine-tune study procedures.
## Objective
The objective is to assess the need and feasibility of randomising a sufficiently large number of women exposed to IPV during pregnancy in a full-scale future randomised trial. To achieve this, we will:
## Trial design
A pilot randomised controlled trial (RCT), co-designed by patient input using Zelen’s design with additional qualitative evaluation, will be nested within a cohort study. To write the present protocol, the SPIRIT guidelines were followed (S1 and S2 Checklists).
We will use a modified Zelen’ design with a double consent process [22–24] and a delayed intervention for the control group. In the first stage, informed consent will be sought from all pregnant women to enter a cohort study. A predetermined small number of the cohort will then be randomized, without their knowledge, to intervention or control (to see the SPIRIT schedule of enrolment, interventions and assessments in eIPV trial see Fig 1 and the Flowchart of eIPV trial in Fig 2). The intervention group will receive the e-health package as part of the cohort.
**Fig 1:** *SPIRIT schedule of enrolment, interventions and assessments in eIPV trial.* **Fig 2:** *Flowchart of eIPV trial.*
In the second stage, participants who have been assigned to the control, will be reapproached and given information about their participation in the control group. At this stage, they will be invited to accept a delay in the intervention (e-health package eight weeks later) and be asked to give the second informed consent for a delayed intervention. Those who decline will remain in the cohort study. Those in the cohort only group will not be informed about the randomization, as their subsequent follow up in the study will remain part of the cohort study to which they would already have consented in the first stage. We will substitute the women who not consent being part of the control until we reach the sample required (5 women in each country). This information will be also analysed for a future full-scale RCT.
## Ethics and dissemination
The trial was granted Andalusian Research Ethics Committee (S2 File; 202167133116; 202072112495; 2021128101651) and the Regional Committees on Health Research Ethics for Southern Denmark (20212000–80). Any changes to the protocol are subject to a formal amendment and may not be implemented prior to the approval by the Research Ethics Committee of each country.
## Study setting
Participants will be recruited at twenty-nine urban public primary health antenatal care centers within Andalusia (Spain) and at Odense University Hospital in the Region of Southern Denmark. Participants will be randomised to the intervention group or to the control group (Fig 2).
## Eligibility criteria
All women who fulfil the inclusion criteria will be screened and invited to receive an e-health package. Of those who accept the e-health intervention, 20 women will be randomised, 10 women in each country. In Spain and Denmark, respectively, 5 will be allocated in the intervention group and 5 will be asked to consent to be in the control group.
Inclusion criteria: pregnant women at <12 weeks gestation, who screen positive in IPV at the first antenatal visit and accept the e-health package.
Exclusion criteria: [1] women who cannot be informed about the study without their partners or other family members knowing; [2] mentally or physically incapacity to participate in the study; [3] women below 16 years in Spain or below 18 years in Denmark; [4] inability to understand Danish/Spanish, [5] lack of internet and electronic device and [6] women with extreme severity of IPV. Women selected to participate in the trial in this situation will receive an evaluation of danger before randomisation. If the severity of IPV is confirmed to be at high level of danger, these pregnant women will be routinely treated and supported according to the standard protocol in each country. Following this path, any potential risk for the women will be taken into account from the beginning of the process and women will be protected. Women who have same-sex partners will be screened, but their data will not be used for the purpose of this study.
## Recruitment: Consent procedures
In advance, a written, informed consent (in Spain the informed consent will be oral and recorded) for administering the screening of IPV, will be sought from the pregnant women at the first antenatal visit with a midwife. Those who screen positive, will also be asked for consent to receive the e-health package intervention and to complete baseline and outcome measurement questionnaires (one month after the intervention). For the pilot trial, women who consent to receive the e-health package will be randomized (following a modified Zelen’s design) into the intervention group (without their knowledge) or into the control group. Women in the control group will additionally be asked for a second informed consent, to be allocated in this group and to receive the e-health package 8 weeks later. This approach is suggested by the input of IPV survivors. A few women from the intervention and control groups of the pilot trial will be asked for consent to be contacted for a qualitative interview. Women will be informed that they have the right to refuse participation as well as to withdraw their consent at any time, without giving a reason, and that this will not affect their subsequent care. If they withdraw consent, data collected up to the point of withdrawal will be retained in the study. Participant information sheets and consent forms used for this pilot study are available on request from investigators. The informed consent provides study aims, contact details and sources for further information. Given the importance of taking into consideration the views and opinion of the participants [25], all recruitment materials have been developed with significant input from a patient and public involvement (PPI) group consisting at two participant representatives from Spain and Denmark (see the Discussion section).
## Screening procedures
Eligible women are invited by trained midwives at the first antenatal care visit to fill in an app developed for tablets for the screening of IPV during pregnancy (in Spain). In Denmark, screening for IPV is part of routine antenatal care in the Region of Southern Denmark where pregnant women are asked to fill in an electronical questionnaire on health conditions and pregnancy related life-style measures in the first trimester (PRO-data), including a screening for IPV exposure.
Women who screen positive in IPV will be invited to receive an e-health intervention (see the positivity criteria in the data collection section below).
## Randomisation procedures
The research team will consult the data server on a daily basis to identify women who screen positive in IPV and accept to receive the e-health package to randomly allocate them (using a computer-generated random numbers) in the intervention or control group through randomisation at a ratio of 1:1. IPV counsellors will be informed of the assignments to each group. Women in the intervention group will be blinded but not women in the control group.
## Study interventions and assessments
Intervention group: Women positive for IPV who accept the e-Health intervention and who have been randomly allocated in the intervention group will receive the e-health package as the rest of the cohort, as well as the baseline and outcome measurements. The e-health package will include six video counselling sessions by trained providers—a psychologist in Spain; midwives in Denmark. Women in both countries will be provided access to a mobile app for designing security plans. This will be an adapted version of the mobile app “My Plan”. To organize and host the video counselling sessions the platform “Linkello Medical” will be used in Spain and the app “My Hospital” will be used in Denmark. Both allow health providers to interact with the women in the place they prefer as long as they have Internet connection. In “Linkello Medical”, the psychologist will send an Internet link to the women by her preferred means (via mobile message or by email) and she would just have to click on the link to connect. Women will not need to register nor provide any personal data. The link will be volatile, i.e., it will be not reusable after completion of the communication. In “My Hospital”, is a software developed specifically for confidential hospital-patient communication in the Region of Southern Denmark and is already used in antenatal care as a regular communication tool. When the pregnant woman attend her first antenatal care visit, she will be introduced to My Hospital and instructed in how to access the system via the Internet or a mobile app. The midwives book the women for a video call through the system. Fifteen minutes before the video call is starting, the woman receives a notification to accept the call. Once the woman has accepted the call, the midwife can start the video consultation. This easy and confidential way of interaction will facilitate women’s participation. The mobile app for safety planning will be camouflaged to look like a common pregnancy app. The content and setup of the safety planning app was adapted from the interactive safety decision aid developed by Glass, et al. [ 26] which is freely available. A safety plan is a personal and practical plan, designed by a woman exposed to violence to minimize their risk of danger and exposure to violence. By digitalizing the safety plan into an app, the women will always have it available to remind them of their own strategies and resources. The safety planning element will be hidden inside the pregnancy app for security reasons; the women must log on for this element to be opened. Once a woman has logged on, the front page consists of two buttons, “Help” and “Emergency”. Both buttons consist of default contact information for relevant resources which can help a woman in case she needs help or is in an emergency situation. The woman can edit the information to fit her own needs. The app also includes the following features: “Warning signs”, where the woman can add signs that may trigger situations where violence could occur; “Strategies”‘, where the woman can note down her coping strategies; and, “Diary”, where a woman can write down anything of importance. It will also contain the feature “Knowledge about domestic violence” which will provide information and links to local and national resource’s relevant for women exposed to violence. Under another feature called “Quick Messages” a woman can enable a predetermined message to her emergency contacts. In case the woman needs to exit the app quickly, she will be able to press the “Quick-exit” button, where she will immediately return to the camouflaged part of the app. This app has been made freely downloadable to researchers and users for both android and iOS devices.
We will monitor if participants adhere to all the sessions offered for counselling and in case of non-adherence or failure to engage in counselling, the counsellor will follow specific protocol to encourage reengagement. In Spain, the psychologist will contact the woman two to three times through the preferred mean (email or phone). In Denmark, a project midwife will try to contact the woman on the preferred mean (email or phone) at 3 occasions within 6 weeks. The content of the six individually tailored sessions will be based on the Dutton’s Empowerment Model [27] and the Psychosocial Readiness Model [28]. Specifically, the contents will include (session 1) the evaluation of abusive behaviour (which will also served to protect the pregnant women is the level of violence detected is high); (session 2) safety planning, network and resources; (session 3) psycho-education (healthy relationships, cycle of violence…); (session 4) self-esteem and self-care and (sessions 5 & 6) empowerment, choice making and problem solving. The video consultations will be planned for every 2 weeks at a time where the women feel comfortable. However, schedule flexibility will always prevail to promote women participation. Criteria for discontinuing or modifying allocated interventions will be applied to women with an elevated risk for IPV who will be routinely treated according to the standard protocol in each country. In other words, if during the interventions the health providers detect an increased on the level of violence, they will apply the established national protocols to protect the pregnant women.
Control group: IPV positive women who accept the e-Health intervention package will be asked for a second consent to receive a delayed intervention (8 weeks later) and to complete the baseline and outcome measurements. Women can request to leave the control group at any time and receive the intervention immediately (in which case the data will be part of the cohort study).
Assessments and data collection: data on socio-demographic characteristics and partner violence will be collected during the screening process. The validated screening questionnaires to detect IPV will be the short form of the Women Abuse Screening tool (WAST-Short) [29], a 2-items tool that measures conflict and tension and the Abuse Assessment Screen (AAS) [30], a 5-item screening questionnaire including emotional, physical, and sexual violence as well as fear of partner within the past year or ever. Data from previous studies support the reliability and validity of the WAST [31] and the AAS [30]. In Spain, a woman is initially screened positive for exposure to IPV in the AAS if she answers “yes” to one or more of the five questions if the perpetrator is the partner or ex-partner. In Denmark, a woman screen positive if she answers “yes” to having been afraid of their partner or ex-partner or to having been exposed to emotional, physical or sexual violence within the past 12 months by a partner or ex-partner in the AAS. In Spain, a woman screens positive in WAST-Short if she meets one of the two original criteria: [1] having some or a lot of tension and/or some or a lot of difficulties to solve problems with her partner (all positive responses receive 1, the cut-off = 2) or [2] if she indicates having a lot of tension and/or a lot of difficulties (extreme responses receive 1, cut-off = 1). In Denmark, woman screens positive in WAST-Short if she meets the second criteria. If the women initially are screened positive, the screening questionnaires will automatically be followed by the Index of Spouse Abuse (ISA) questionnaire [29]. The ISA questionnaire is a detailed 30-item questionnaire about IPV (emotional, physical and sexual) in order to confirm the IPV and evaluate the severity of the violence. Two different scores are computed: ISA-P (physical abuse) and ISA-NP (non-physical abuse). The scores range from 0 to 100. In Denmark, the cut-off scores of 10 for ISA-P, and 25 for ISA-NP developed by Hudson and McIntosh will be used [32]. In Spain, based on the Spanish validation of the ISA tool [33] that gives different weights to the original items, cut-off scores of 6 for ISA-P, and 14 for ISA-NP will be applied. The reliability (α) of the ISA range from.82 [33] to.96 [32].
Quantitative questionnaire data will be collected before and after the e-health intervention. Exposure to IPV will be assessed by use of the ISA tool, and post-natal depression will be assessed by use of The Edinburgh Postnatal Depression Scale (EPDS, α =.81) which is a 10-item validated questionnaire designed to detect postnatal depression [34]. Further, participants will be asked to assess their ability to carry out safety behaviour actions by use of a revised version of the 22-item safety action checklist (α =.75) [35] and to complete a Measure of Victim Empowerment Related to Safety (MOVERS; α =.74-.88) [36] All the data will be entered into a bespoke secure online study database using a unique study ID for each participant. The health providers will check the items of the ISA related to physical IPV and the psychological IPV items measuring fear of the partner to determine if the women invited to the pilot are at extreme level of IPV risk. In those women who respond positive to the above items, before randomisation the providers will evaluate the level of danger with the Severe Intimate Partner Violence Risk Prediction Scale-Revised (EPV-R, α =.72) [37]. If the level of danger is confirmed to be high, women will not be included in the study and will be routinely treated and supported according to the standard protocol in each country.
Qualitative data will be collected through individual in-depth interviews with a few women at any point of the flow diagram (Fig 2) to explore their opinion and experiences of the study procedures and intervention. Additionally, we will conduct interviews with the IPV counsellors (midwives in Denmark and psychologist and midwives in Spain) to explore their experiences with the delivery of the intervention and their opinion about a future full-randomised controlled trial. A semi-structured interview guide will be developed for the interviews using the Model for Assessment of Telemedicine Applications (MAST) as theoretical framework [38] with focus primarily on the following domains: users’ perspectives, their safety, ethical and organisational aspects; health problems and technology. All participants will give written consent prior to the interviews starting and data collection will stop once data is saturated. The interviews will be conducted in Danish/Spanish and audio recorded, and will be transcribed and analysed by use of a thematic content analysis using a combined inductive-deductive approach [39]. A coding frame will be developed, and themes will deductively be derived from MAST and supplemented with categories that inductively arise from the data. See Fig 1 for an overview of intervention and assessments.
Data will be collected by trained health personal and researchers in the STOP (Stop Intimate Partner Violence in Pregnancy, https://stop-ipv.eu) Project. On reasonable request, protocol, data collection forms and plans for statistical analysis will be available from investigators. In case of failure to provide data in the follow-up assessments, we will follow the above described STOP protocol for follow up. In Spain, the participant will be contacted by the preferred mean: phone or email. If a woman indicates she wants to be contacted by phone, the psychologist will make a maximum of three phone calls at different times and days. If the contact is made by email or phone message, the counsellor will send a maximum of two attempts leaving between both at least three days. In Denmark, women in the project will indicate if they prefer to be contacted by phone or email at the inclusion. If an included woman is unreachable, a project midwife will try to establish contact at 3 different occasions within 6 weeks. Rate of failure to obtain data will be one of the outcome measures.
Relevant concomitant care and interventions that are permitted during the trial: Women in both the intervention and control groups would be able to benefit from the standard care available for women exposed to IPV thorough the usual provision made by social and health services at any time during the study, if necessary. Indeed, one of the video-counselling sessions will be dedicated to inform women about the available resources in their community. IPV counsellors will specify if women in the intervention and the control group are in need of these resources during the time of the pilot as a valuable information for the future full-scale trial.
To accomplish the objectives, we will collect the following information: Primary outcome: Secondary outcomes: Participant timeline
The participant timeline is illustrated in Fig 1.
## Sample size calculation
As this is a feasibility pilot study, power calculation for sample size determination is not an obligatory requirement [40]. The pilot study will not be powered to detect statistical differences in key clinical outcomes. We will review patient and staff feedback before finalising a protocol for a full-scale multicentre RCT.
## Statistical analysis
To determine the feasibility, we will perform the analysis against established stop-go rules according to the CONSORT statement for pilot and feasibility trials [41]. The feasibility analysis will focus on the outcomes described above. For example, we will compute the percentage and $95\%$ confidence intervals (CI) of IPV positive women, who consented to receive e-health package and consented to randomization in the control group, following the same analytical approach than other pilot studies [42].
## Progression criteria
Table 1 outlines the main criteria that will be considered to assess the feasibility of a full-scale RCT. In addition, our qualitative findings will also be used to support the decision-making around progression to a full-scale trial. For example, if a progression criterion outlined in Table 1 does not meet the threshold for progression, but we have developed a qualitative understanding of why this occurred and how it could be improved, then it may still be possible to proceed with the full trial. The research team of the STOP project will have access to these interim results.
**Table 1**
| Feasibility objectives and related data to be collected | Go criteria to proceed to full trial | Criteria to reassess and adjust full trial protocol | Stop criteria |
| --- | --- | --- | --- |
| Study population | Study population | Study population | Study population |
| 1. Consent rate of eligible women | Rate >25% of eligible women agreeing to participate. | Rate between 11% and 24% women agreeing to participate | Rate <10% of eligible women agreeing to participate |
| Study outcomes | Study outcomes | Study outcomes | Study outcomes |
| 2. Proportion of women in either intervention or control group for whom the allocated treatment is adhered to. | Adherence to allocated treatment in >80% of study sample. | Adherence to allocated treatment in between 51% and 79% of study sample. | Adherence to allocated treatment in <50% of study sample. |
| RCT process | RCT process | RCT process | RCT process |
| 3. Collection of data on clinical outcomes | Complete data available of >80% of study sample. | Missing data between 21% and 49% of study sample. | Data missing of >50% of study sample. |
## Data management
All data management will be undertaken by the University of Granada (UGR) (Spain) and Odense University Hospital (Denmark). In Spain, standard operating procedures will be in place for the collection and handling of data. All study data will be entered directly by trained and delegated research staff included in the STOP project into a secure, bespoke electronic trial database with inbuilt range checks set up and hosted by the University of Granada. User accounts will be allocated and managed centrally by the trial coordinator and restricted to appropriate site level access. Data collected on the forms and entered into the electronic database will only identify the participants by a unique trial number. No identifiable data will be stored in the trial database.
In Denmark, study data are collected by trained project midwives from the STOP project following a manual for data collection and will be entered directly to a secure web-based database designed to support data capture for research studies with inbuilt range checks (REDCap) (project-redcap.org). User access to the database is restricted and assigned by the Danish trial coordinator. Data is entered into the database with a unique study trial number, and there will be no identifiable data in the database. Data with invalid study trial numbers, values outside the range or follow-up id that do not match the baseline study trial number, will be excluded.
## Trial management
The trial is managed and run by the University of Granada and Odense University Hospital. Both centers are responsible for safety reporting, coordination of trial committees, statistical analysis and reporting, trial monitoring, database management and case report form design.
## Pilot trial oversight
An advisory board has been established to oversee and monitor the trial conduct and patient safety. It is chaired by an independent professor from Norway (Dr Mirjam Lukasse), with two other independent IPV experts from Spain and Denmark (Prof Tine Gammeltoft and Dr Carmen Vives Cases). The advisory board provides overall supervision of the trial and ensures that it is being conducted according to the protocol, good clinical practice and relevant regulations. The advisory board also monitor trial progress in relation to recruitment, data capture and completeness, protocol adherence and deviations and subject withdrawals. The board will meet at the request of the investigators. It will be responsible for reviewing the trial data and assessing whether there are any safety issues that need to be brought to the attention of the sponsor, or any ethical reasons why the trial should not continue. Given the low risk of the study intervention and that it is non-blinded, no separate data safety monitoring committee will be established. The sponsor retains the right to audit the study, including any study site or central facility.
## Protecting participants and safety reporting
By taking part in the study, participants will automatically enter a project where a greater level of care will become available to them. The study has well-established security protocols. The screening questionnaires will capture the severity of violence permitting us to refer women at high risk of danger to the relevant national authorities responsible their protection. During of the first videocounselling session, even if the questionnaires did not register a high level of risk, the danger to the life of the participants will be specifically evaluated. If a risk of danger is detected, the same protocol as above will be applied. All the videocounselling sessions will be carried out with a high level of vigilance concerning danger. It should be noted that the intervention itself (as explain above) contains features that are protective, for example, a link that expiries after the counseling sessions, a protection plan back by easy-to-use features to activate emergency contacts, camouflage of the app and emergency exit button. The videocounselling providers will receive in-session protection training, for example, to change the topic of the conversation in case the partner shows up unexpectedly during the sessions. Videoconsellors and women will talk about these options at the beginning of the intervention; in this way both will know what to do if something like this happens.
For safety reporting, if severe or life-threatening abuse of a pregnant woman is identified by the IPV counsellors, the principal investigator will be notified and the woman will be treated according to the standard protocol in each country. The study does not add any risk or harm for the women, we will therefore not provide a post-trial care for compensation. On the contrary, it is anticipated that video counselling addressing safety behavior and safety planning has the potential to increase safe behaviors and thus decrease IPV exposure. We also anticipate that the participating midwives will experience a greater competence and confidence in approaching the topic of violence and handling women who are exposed to IPV. Women will benefit from health professionals who enquire in an appropriate way about violence. In addition, information will be provided about IPV community resources which the women may make use of after the e-health intervention is finished.
## Patient and public involvement
Two women previously exposed to IPV (one Spanish and one Danish) form the target group representatives. They will help develop a more pregnant women-centred information sheet in line with the well-known need to involve citizens in science [43]. The qualitative research embedded within this pilot study will prove integral in evaluating how the consent materials and processes were received in practice.
## Discussion
Even though numerous intervention models to address IPV have been developed, current efforts suffer from limitations. First, IPV services are often not integrated within routine health service delivery, and second, models for integrating IPV service in health service delivery where developed tend not to have generalizability to the European context [14–20]. To address the personally and politically sensitive problems associated with IPV among pregnant women, effective, sustainable, and culturally appropriate health system-based interventions need development. If found effective, e-health interventions would be suitable for incorporation in these care pathways.
Screening for IPV combined with an empowerment intervention that includes education and video counselling on safety planning may have potential to address repeated IPV and associated adverse health effect among pregnant women. The option of receiving counselling and support through video consultation during antenatal care presents an opportunity for more accessible and flexible care addressing some of the barriers associated with in-person care, such as travel distance and time, travel costs, and the stigma of seeking help. In addition, patients may be more motivated to seek and continue treatment if they are in a familiar environment of their own choice and can avoid stressful situations, such as navigating a hospital or maternity facility. The development of the video counselling intervention will capture the need for safety and adhere to strict security features preventing the risk of women’s partners prying on their online activity. These provisos need underpinning effectiveness evidence.
Prior to performing a large interventional trial, a pilot study is needed to identify barriers to recruitment, assess feasibility and acceptability of the treatment, and fine-tune study procedures. No RCT has previously assessed an e-health intervention in IPV among pregnant women in comparison with a control group (with a delay intervention). In this pilot trial, we chose to perform randomisation with a modified Zelen’s design rather than simple randomisation for several reasons. Participants who take part in standard RCTs will make a judgment of their preferred treatment and often expect to be allocated to the treatment group [44]. If this does not occur, it can be followed by dissatisfaction and distrust in those who approached them to take part [45]. Consequently, randomization to a control group may lead to dropout after allocation. The original Zelen design involves randomization before consent, with consent only required from those allocated to the intervention, whereas the control group receive their usual care [24]. Baseline data and outcomes are collected from medical records (with ethical approval). However, it is not possible to have interaction with the control group during follow-up, as they are not informed of their presence in a study. Taking all of this into consideration, we hypothesized that women will accept the intervention because they perceive that it is needed for their support and protection. If they perceive it like this, they may not want to be randomized into control group or they may drop out after been randomized to usual care instead of intervention. To overcome this difficulty with acceptance of allocation to the control group, we followed the input of IPV survivors in a focus group we conducted [13]. Amongst other things the survivors informed us that they would prefer delayed intervention instead of usual care in a control group. The opinion of the participant representative in our project affirmed their support for this approach. A systematic review concluded that a delayed intervention could be an effective way of minimizing dropout [46]. In our pilot, thus we chose to offer to women of the control group a delay in the intervention. It is this specific challenge that is addressed in our pilot study. We wish to estimate the rates of acceptance of randomization primarily. We do not aim to estimate effect sizes. The approaches recommended for use of standardized effect sizes for pilot study sample sizes estimation are not applicable [47]. Our analytic approach will compute simple statistic with $95\%$ CI and we will employ qualitative judgment in assessing the need to prepare a full-scale RCT.
Beyond the information described to be collected during this pilot, we will capture other relevant information in the cohort study useful for planning the future full-scale randomised control trial: number of women who were approached and agreed to fill in the IPV screening question, number of women where there was a study protocol violation, number of women exposed to physical IPV captured by the screening questions, number of pregnant women exposed to physical IPV that accept to participate in the intervention and in the control group, number of women willing to participate in follow-up interviews, the acceptability of the e-health package for ongoing IPV prevention and the acceptability of video-counselling for IPV prevention in terms of compliance [48] to the scheduled counselling sessions.
In conclusion, the pilot study nested within the cohort study will allow us to obtain information about the rates of IPV in pregnancy, the acceptability of an e-health intervention and the availability of participants for randomisation into an effectiveness trial. These results will inform us about the feasibility and variance of key clinical outcome measures for estimation sample size of the full-scale effectiveness trial.
## Trial status
Trial registration number: NCT04978064 (S1 File). Protocol version 1.0, 20 April 2021. The proposed start date of randomised participant recruitment: September, 2021. The proposed project recruitment completion date of any participant including those not randomised: September, 2022. The proposed end of follow-up of all participants including those randomised: September, 2022.
## Confidentiality
In order to protect confidentiality before, during, and after the trial, personal information about potential and enrolled participants will be collected and saved during the screening process in a secured data base and it will be only accessed by authorised research investigators of the STOP project. The information of the baseline and outcome measurements will be saved in different databases, and both could be only aggregated by the research members by the birthdate and telephone number of the women.
## Declarations of interests
Financial and other competing interests for principal investigators for the overall trial and each study site.
## Access to data
Research team members from the STOP project will have access to the final pilot trial dataset only for research purposes according to the objectives established in this protocol. We will share the anonymized dataset and output of the statistical analysis openly and transparently on completion of the study providing it in a repository or a supplementary file at the time of submission of the manuscript containing results of the completed study.
## Dissemination policy
We will share the eHealth package with relevant stakeholders in order to upscale it across the EU as well as globally to better combat IPV. Concrete dissemination activities include the production of research papers, guidelines and white papers detailing the eHealth intervention, which will be made available on the STOP project website and presented at relevant professional conferences reaching both health care professionals and other relevant stakeholders and policymakers.
## Consent for publication
All relevant data from this study will be submitted to peer-reviewed journals for publication following the completion of the study in line with sponsor publication policy. Data will be captured for all study participants, and no patient identifiable data will be used in any publications. The sponsor retains the right to review all publications prior to submission or publication. Responsibility for ensuring accuracy of any publication from this study is delegated to the chief investigator. Authorship will be assigned in compliance with International Committee of Medical Journal Editors (ICMJE) guidelines.
The Advisory board is comprised by Dr. Mirjam Lukasse (Norway), Dr. Tine Gammeltoft (Denmark), Dr. Carmen Vives Cases (Spain) and two participant representatives from Denmark and Spain whom prefer to remain anonymous.
## References
1. 1World Health Organization, Pan American Health Organization. Understanding and addressing violence against women: intimate partner violence. 2012.. *Understanding and addressing violence against women: intimate partner violence* (2012.0)
2. García-Moreno C, Pallitto C, Devries K, Stöckl H, Watts C, Abrahams N. (2013.0)
3. Román-Gálvez RM, Martín-Peláez S, Martínez-Galiano JM, Khan KS, Bueno-Cavanillas A. **Prevalence of Intimate Partner Violence in Pregnancy: An Umbrella Review**. (2021.0) **18** 707. DOI: 10.3390/ijerph18020707
4. Parsons L, Goodwin MM, Petersen R. **Violence Against Women and Reproductive Health: Toward Defining a Role for Reproductive Health Care Services**. (2000.0) **4** 6. DOI: 10.1023/a:1009578406219
5. Velasco C, Luna JD, Martin A, Caño A, Martin‐de‐las‐Heras S. **Intimate partner violence against Spanish pregnant women: application of two screening instruments to assess prevalence and associated factors**. (2014.0) **93** 1050-1058. DOI: 10.1111/aogs.12459
6. Martin‐de‐las‐Heras S, Velasco C, Luna‐del‐Castillo JD, Khan KS. **Breastfeeding avoidance following psychological intimate partner violence during pregnancy: a cohort study and multivariate analysis**. (2019.0). DOI: 10.1111/1471-0528.15592
7. Martin-de-las-Heras S, Velasco C, Luna-del-Castillo J de D, Khan KS, Seedat S. **Maternal outcomes associated to psychological and physical intimate partner violence during pregnancy: A cohort study and multivariate analysis**. (2019.0) **14** e0218255. DOI: 10.1371/journal.pone.0218255
8. Lukasse M, Schroll A-M, Karro H, Schei B, Steingrimsdottir T, Van Parys A. **Prevalence of experienced abuse in healthcare and associated obstetric characteristics in six European countries**. (2015.0) **94** 508-517. DOI: 10.1111/aogs.12593
9. Lukasse M, Schroll A-M, Ryding EL, Campbell J, Karro H, Kristjansdottir H. **Prevalence of emotional, physical and sexual abuse among pregnant women in six European countries**. (2014.0) **93** 669-677. DOI: 10.1111/aogs.12392
10. Chambliss LR. **Intimate partner violence and its implication for pregnancy**. (2008.0) **51** 385-397. DOI: 10.1097/GRF.0b013e31816f29ce
11. Martin SL, Li Y, Casanueva C, Harris-Britt A, Kupper LL, Cloutier S. **Intimate Partner Violence and Women’s Depression Before and During Pregnancy**. (2006.0) **12** 221-239. DOI: 10.1177/1077801205285106
12. van Gelder N, Ligthart S, Ten Elzen J, Prins J, van Rosmalen‑Nooijens K, Oertelt‑Prigione S. **“If I’d Had Something Like SAFE at the Time, Maybe I Would’ve Left Him Sooner.”‑essential features of eHealth interventions for women exposed to intimate partner violence: a qualitative study.**. *J Interpers Violence* (2021.0). DOI: 10.1177/08862605211036108
13. López RF, de-León-de-León S, Martin-de-las-Heras S, Cantero JCT, Megías JL, Zapata-Calvente AL. **Women survivors of intimate partner violence talk about using e-health during pregnancy: a focus group study**. (2022.0) **22** 98. DOI: 10.1186/s12905-022-01669-2
14. Ahmad F, Hogg-Johnson S, Stewart DE, Skinner HA, Glazier RH, Levinson W. **Computer-Assisted Screening for Intimate Partner Violence and Control**. (2009.0) **151** 93-102. DOI: 10.7326/0003-4819-151-2-200907210-00124
15. Klevens J, Sadowski L, Kee R, Trick W, Garcia D. **Comparison of Screening and Referral Strategies for Exposure to Partner Violence**. (2012.0) **22** e45-e52. DOI: 10.1016/j.whi.2011.06.008
16. Rhodes KV, Drum M, Anliker E, Frankel RM, Howes DS, Levinson W. **Lowering the threshold for discussions of domestic violence: a randomized controlled trial of computer screening**. (2006.0) **166** 1107-1114. DOI: 10.1001/archinte.166.10.1107
17. McFarlane J, Malecha A, Gist J. **An intervention to increase safety behaviors of abused women: results of a randomized clinical trial**. (2002.0) **51** 347-354. DOI: 10.1097/00006199-200211000-00002
18. Gillum TL, Sun CJ, Woods AB. **Can a Health Clinic-Based Intervention Increase Safety in Abused Women? Results from a Pilot Study**. (2009.0) **18** 1259-1264. DOI: 10.1089/jwh.2008.1099
19. Koziol-McLain J, Vandal AC, Wilson D. **Efficacy of a Web-Based Safety Decision Aid for Women Experiencing Intimate Partner Violence: Randomized Controlled Trial**. (2018.0) **19** e426. DOI: 10.2196/jmir.8617
20. Tiwari A, Yuk H, Pang P. **Telephone intervention to improve the mental health of community-dwelling women abused by their intimate partners: a randomised controlled trial**. (2012.0) **18** 14-17. PMID: 23249846
21. Bloom TL, Glass NE, Case J, Wright C, Nolte K, Parsons L. **Feasibility of an Online Safety Planning Intervention for Rural and Urban Pregnant Abused Women**. (2014.0) **63** 243-251. DOI: 10.1097/NNR.0000000000000036
22. Land J, McCourt O, Heinrich M. **The adapted Zelen was a feasible design to trial exercise in myeloma survivors**. (2020.0) **125** 76-83. DOI: 10.1016/j.jclinepi.2020.04.004
23. Schellings R, Kessels AG, ter Riet G, Knottnerus JA, Sturmans F. **Randomized consent designs in randomized controlled trials: Systematic literature search**. (2006.0) **27** 320-332. DOI: 10.1016/j.cct.2005.11.009
24. Zelen M.. **A New Design for Randomized Clinical Trials**. (1979.0) **300** 1242-1245. DOI: 10.1056/NEJM197905313002203
25. Daru J, Moores R, Dodds J, Rayment J, Allard S, Khan KS. **Non-anaemic iron deficiency in pregnancy: the views of health service users and health care professionals**. (2015.0) **25** 27-32. DOI: 10.1111/tme.12184
26. 26myPlan. myplanapp. myPlan. https://www.myplanapp.org/. Accessed [20220628], 2022.
27. Dutton MA. (1992.0) **xx** 202
28. Cluss PA, Chang JC, Hawker L, Hudson S, Dado D, Buranosky R. **The process of change for victims of intimate partner violence: support for a psychosocial readiness model.**. (2006.0) **16** 262-274. DOI: 10.1016/j.whi.2006.06.006
29. Brown JB, Lent B, Schmidt G, Sas G. **Application of the Woman Abuse Screening Tool (WAST) and WAST-short in the family practice setting**. (2000.0) **49** 896-903. PMID: 11052161
30. Soeken KL, McFarlane J, Parker B, Lominack MC. (1998.0) 195-203
31. Rabin RF, Jennings JM, Campbell JC, Bair-Merritt MH. **Intimate Partner Violence Screening Tools**. (2009.0) **36** 439-445.e4. DOI: 10.1016/j.amepre.2009.01.024
32. Hudson WW, McIntosh SR. **The Assessment of Spouse Abuse: Two Quantifiable Dimensions.**. (1981.0) **43** 873. DOI: 10.2307/351344
33. Ruiz Pérez I, Plazaola Castaño J, Escribá Agüir V, Jiménez Martín JM. *Observatorio de Salud de la Mujer* (2006.0) 56
34. Eberhard-Gran M, Eskild A, Tambs K, Schei B, Opjordsmoen S. **The Edinburgh Postnatal Depression Scale: Validation in a Norwegian community sample**. (2001.0) **55** 113-117. DOI: 10.1080/08039480151108525
35. Ford-Gilboe M, Varcoe C, Scott-Storey K, Perron N, Wuest J, Wathen N. **Longitudinal impacts of an online safety and health intervention for women experiencing intimate partner violence: randomized controlled trial**. (2020.0) **20** 260. DOI: 10.1186/s12889-020-8152-8
36. Goodman LA, Cattaneo LB, Thomas K, Woulfe J, Chong SK, Smyth KF. **Advancing domestic violence program evaluation: Development and validation of the Measure of Victim Empowerment Related to Safety (MOVERS)**. (2015.0) **5** 355-366. DOI: 10.1037/a0038318
37. Echeburúa E, Amor PJ. **Escala de Predicción del Riesgo de Violencia Grave contra la pareja—Revisada—(EPV-R)**. (2010.0) **22** 1054-1060. PMID: 21044552
38. Kidholm K, Ekeland AG, Jensen LK, Rasmussen J, Pedersen CD, Bowes A. **A model for assessment of telemedicine applications: mast**. (2012.0) **28** 44-51. DOI: 10.1017/S0266462311000638
39. Schreier M.. (2013.0)
40. Arain M, Campbell MJ, Cooper CL, Lancaster GA. **What is a pilot or feasibility study? A review of current practice and editorial policy**. *BMC Med Res Methodol* (2010.0) **10** 67. DOI: 10.1186/1471-2288-10-67
41. Eldridge SM, Chan CL, Campbell MJ, Bond CM, Hopewell S, Thabane L. **CONSORT 2010 statement: extension to randomised pilot and feasibility trials**. (2016.0) **2** 64. DOI: 10.1186/s40814-016-0105-8
42. Rondung E, Leiler A, Sarkadi A, Bjärtå A, Lampa E, Löfving SG. **Feasibility of a randomised trial of Teaching Recovery Techniques (TRT) with refugee youth: results from a pilot of the Swedish UnaccomPanied yOuth Refugee Trial (SUPpORT)**. *Pilot and Feasibility Studies* (2022.0) **8** 40. DOI: 10.1186/s40814-022-00998-1
43. Moss N, Daru J, Lanz D, Thangaratinam S, Khan K. **Involving pregnant women, mothers and members of the public to improve the quality of women’s health research**. (2017.0) **124** 362-365. DOI: 10.1111/1471-0528.14419
44. King M, Nazareth I, Lampe F, Bower P, Chandler M, Moroum M. **Impact of Participant and Physician Intervention Preferences on Randomized Trials: A Systematic Review**. (2005.0) **293** 1089. DOI: 10.1001/jama.293.9.1089
45. Featherstone K, Donovan JL. **“Why don’t they just tell me straight, why allocate it?” The struggle to make sense of participating in a randomised controlled trial**. (2002.0) **55** 709-719. DOI: 10.1016/s0277-9536(01)00197-6
46. Bisschop CNS, Courneya KS, Velthuis MJ, Monninkhof EM, Jones LW, Friedenreich C. **Control Group Design, Contamination and Drop-Out in Exercise Oncology Trials: A Systematic Review**. (2015.0) **10** e0120996. DOI: 10.1371/journal.pone.0120996
47. Whitehead AL, Julious SA, Cooper CL, Campbell MJ. **Estimating the sample size for a pilot randomised trial to minimise the overall trial sample size for the external pilot and main trial for a continuous outcome variable**. *Stat Methods Med Res. junio de* (2016.0) **25** 1057-73. DOI: 10.1177/0962280215588241
48. Forbes G, Newton S, Calvete CC, Birch J, Dodds J, Steed C. **MEMPHIS: a smartphone app using psychological approaches for women with chronic pelvic pain presenting to gynaecology clinics: a randomised feasibility trial**. (2020.0) **10** e030164. DOI: 10.1136/bmjopen-2019-030164
|
---
title: 'Effect of antiretroviral therapy on decreasing arterial stiffness, metabolic
profile, vascular and systemic inflammatory cytokines in treatment-naïve HIV: A
one-year prospective study'
authors:
- Pedro Martínez-Ayala
- Guillermo Adrian Alanis-Sánchez
- Monserrat Álvarez-Zavala
- Karina Sánchez-Reyes
- Vida Verónica Ruiz-Herrera
- Rodolfo Ismael Cabrera-Silva
- Luz Alicia González-Hernández
- Carlos Ramos-Becerra
- Ernesto Cardona-Muñoz
- Jaime Federico Andrade-Villanueva
journal: PLOS ONE
year: 2023
pmcid: PMC10022802
doi: 10.1371/journal.pone.0282728
license: CC BY 4.0
---
# Effect of antiretroviral therapy on decreasing arterial stiffness, metabolic profile, vascular and systemic inflammatory cytokines in treatment-naïve HIV: A one-year prospective study
## Abstract
### Introduction
Cardiovascular disease is a major cause of death among people living with HIV (PLH). Non-treated PLH show increased levels of inflammation and biomarkers of vascular activation, and arterial stiffness as a prognostic cardiovascular disease risk factor. We investigated the effect of one year of ART on treatment-naïve HIV(+) individuals on arterial stiffness and inflammatory and vascular cytokines.
### Methods
We cross-sectionally compared aortic stiffness via tonometry, inflammatory, and vascular serum cytokines on treatment-naïve ($$n = 20$$) and HIV [-] ($$n = 9$$) matched by age, sex, metabolic profile, and Framingham score. We subsequently followed young, treatment-naïve individuals after 1-year of ART and compared aortic stiffness, metabolic profile, and inflammatory and vascular serum biomarkers to baseline. Inflammatory biomarkers included: hs-CRP, D-Dimer, SAA, sCD163s, MCP-1, IL-8, IL-18, MRP$\frac{8}{14.}$ Vascular cytokines included: myoglobin, NGAL, MPO, Cystatin C, ICAM-1, VCAM-1, and MMP9.
### Results
Treatment-naïve individuals were 34.8 years old, mostly males ($95\%$), and with high smoking prevalence ($70\%$). Baseline T CD4+ was 512±324 cells/mcL. cfPWV was similar between HIV[-] and treatment-naïve (6.8 vs 7.3 m/s; $$p \leq 0.16$$) but significantly decreased after ART (-0.52 m/s; $95\%$ CI -0.87 to -0.16; p0.006). Almost all the determined cytokines were significantly higher compared to controls, except for MCP-1, myoglobin, NGAL, cystatin C, and MMP-9. At follow-up, only total cholesterol and triglycerides increased and all inflammatory cytokines significantly decreased. Regarding vascular cytokines, MPO, ICAM-1, and VCAM-1 showed a reduction. D-Dimer tended to decrease ($$p \leq 0.06$$) and hs-CRP did not show a significant reduction ($$p \leq 0.17$$).
### Conclusion
One year of ART had a positive effect on reducing inflammatory and vascular cytokines and arterial stiffness.
## Introduction
Antiretroviral treatment (ART) has become effective in controlling infections caused by human immunodeficiency virus (HIV). Worldwide, by 2021, the United Nations has estimated that approximately 37.7 million people are currently living with human HIV. In 2020, $73\%$ of patients with HIV had access to ART [1]. In Mexico, until May 2021, 113,788 HIV patients were registered as active for ART [2]. HIV infection itself is associated with the development of cardiovascular disease (CVD) [3,4], atherosclerosis [5], and arterial stiffness [6]. Cardiovascular complications are among the leading causes of morbidity and mortality in patients with HIV. ART and HIV infection have a complex interaction with inflammation, coagulation, and other factors. Higher levels of envelope glycoprotein 120, Nef protein, interleukin-6, high-sensitivity C-reactive protein (hs-CRP), and D-dimer have been associated with endothelial dysfunction, cardiovascular disease, and all-cause mortality [7,8]. In addition, HIV infection has been linked with immune activation and low-grade chronic inflammation [9]. It has been reported that traditional CVD scores (i.e., Framingham Risk Score and Atherosclerotic Cardiovascular Disease Risk Score) systematically underestimate cardiovascular risk in HIV [10].
Aortic arterial stiffness is a predictor of cardiovascular events independent of traditional risk factors [11]. Carotid-femoral pulse wave velocity (cfPWV) is considered the gold standard for measuring aortic stiffness [12]. A recent meta-analysis including 17 studies investigating HIV and arterial stiffness found an overall increased cfPWV in individuals with HIV (+0.44 m/s) [13]. Furthermore, some studies have reported adverse effects of ART on aortic stiffness [14,15], whereas others did not report any association [16,17]. A possible explanation for these discrepancies may be the different combinations of ART regimens, different populations, and methods used to assess arterial stiffness. Most studies on arterial stiffness and HIV are cross-sectional and cannot establish a causal relationship; unfortunately, there is a lack of longitudinal clinical studies that investigate the effect of ART and cfPWV over time with the evaluation of inflammatory and vascular cytokines, especially in developing countries. We have previously reported increased arterial stiffness in treatment-naïve HIV individuals compared to HIV[-] controls [18]. The underlying mechanisms linking HIV infection with arterial stiffness remain unclear. Therefore, the present study aimed to investigate the effect of one year of ART on arterial stiffness, inflammatory and metabolic serum biomarkers in treatment-naïve HIV individuals.
## Study population
Between January 2015 and August 2019, people living with HIV (PLH) were enrolled at the "Antiguo Hospital Civil de Guadalajara" in Guadalajara, Mexico. The study complied with the Declaration of Helsinki and was approved by the ethics committee of the Hospital Civil Fray Antonio Alcalde (Register number: $\frac{208}{15}$). After approval from the ethics committee, all individuals who attended the HIV Unit to start ART were invited to participate in the study. Informed consent was obtained from all participants. At study entry, the participants’ medical history and demographic information were obtained using a questionnaire. Inclusion criteria for PLH included: a) Patients 18 years of age or older with confirmed HIV infection and no previous ART, b) Absence of current or previous rheumatological or neoplastic disease or CVD, c) Not taking any vasoactive medication (e.g., antihypertensives, vasopressors, etc.), d) Without opportunistic infections at the time of enrolment. As a pilot prospective study, we analyzed the first 20 PLH from our previous cross-sectional study [18] that achieved and sustained virologic suppression for one year and who had complete data. To avoid selection bias, we confirmed that the demographic and clinical characteristics did not differ from those of the remaining 31 individuals from our cohort, which were not included in the analysis (S1 Table). The control group was paired by age and sex and recruited from our local network of researchers and volunteers within the University of Guadalajara. In addition, general laboratory testing and medical interrogation were performed on control individuals to confirm similar lifestyle as possible (except for the smoking habit which is higher in most HIV cohorts worldwide [19]) and metabolic and cardiovascular risk profiles. Inclusion criteria for the control group included: a) Negative HIV serological test, b) *No previous* cardiovascular, metabolic, or rheumatological disease, c) Not on any medication, including vasoactive drugs.
## Arterial stiffness
Arterial stiffness was measured by cfPWV as described previously [12] by applanation tonometry (PulsePen, Diatechne, Milan, Italy). cfPWV was calculated as the time delay between the arrival of the pulse wave at the carotid and the femoral artery, divided by the distance between the carotid and femoral arteries, with previous automated subtraction of the segment between the carotid and the sternal notch by the software. All measurements were performed by a single trained technician in a temperature-controlled room. The participants rested in a supine position for 15 minutes before the assessment and were instructed to abstain from smoking, and alcoholic, or caffeinated beverages 12 hours before the evaluation [20]. Systolic (SBP) and diastolic blood pressure (DBP) were measured using an automated sphygmomanometer (Omron HEM-907XL). Mean arterial pressure (MAP) was calculated as MAP = DBP + peripheral pulse pressure (pPP)ˑ0.33.
## Viremic control and immunological status
A venous blood sample in an EDTA-tube was obtained from the antecubital vein after 8-hour fasting. CD4+ T-cells count was determined by flow cytometry (FACScalibur System, Becton Dickinson) and HIV-1 viral load was determined using with real-time polymerase chain reaction with retro transcription (Cobas AmpliPrep/Cobas Taqman, Roche Diagnostics) in a federal laboratory.
## Metabolic profile
Serum samples were obtained from the antecubital vein after an 8-hour fasting. Serum lipids, including total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and triglycerides (TG), were determined by colorimetric quantification (AU5800 autoanalyzer, Coulter Beckman, USA). Plasma glucose was determined by photometry (AU5800 autoanalyzer, Beckman Coulter, USA) in a central laboratory.
## Cytokine and vascular inflammation proteins quantification by flow cytometry
Three bead-based multiplex assays were employed to quantify cytokines and vascular biomarkers in all participants: LEGENDplexTM Human Inflammation Panel, LEGENDplex™ Human Vascular Inflammation Panel 12P, and finally detection of hs-CRP was performed with the LEGENDplex™ Human Vascular Inflammation Panel 1S/P Plex (BioLegend, Inc., San Diego, CA, USA). All assays were performed in accordance with the manufacturer’s instructions. Data were acquired in an Attune Acoustic Focusing Cytometer (Life Technologies, Carlsbad, CA, USA). More than 2,000 events for each analyte were acquired. The files were analyzed using LEGENDplexTM Data Analysis Software. Values are expressed in pg/mL.
## D-Dimer and sCD163 quantification
Serum samples were immediately stored at -80°C. Quantification of hs-CRP, D-dimer, and sCD163 was performed using D-Dimer Human SimpleStep and sCD163 (M130) Human, both by ELISA (Abcam®) following the manufacturer’s instructions. Values are expressed as ng/mL.
## Statistical analyses
Depending on their distribution, data are presented as mean± standard deviation and median [interquartile range]. Parametric and non-parametric tests for independent and paired variables were used accordingly. Chi-square was used for non-paired and McNemar´s test for paired categorical data. Statistical significance was set at a two-tailed p-value of <0.05. Cliff’s delta is a non-parametric effect size estimate that was calculated to assess the effect of ART on inflammatory and vascular cytokines [21]. Effect size thresholds for Cliff’s delta were <0.147 for negligible, 0.148–0.33 for small, 0.334–0.474 for medium, and 0.475 for large effects. Based on previous data from our lab, we calculated a Pearson correlation coefficient between cfPWV paired data of 0.81 and the sample size necessary to detect a paired difference of 0.6 m/s and pooled SD 1.1, which resulted in a sample size of 12 individuals with $80\%$ power. In addition, the effect size of ART on IL-10 after a 12-month ART was 5.1 requiring 6 patients in total [22], and 0.96 for VCAM, requiring 19 patients in total [23]. Statistical analysis and graphical representations were performed with GraphPad Prism 6. Cliff’s delta was calculated using RStudio v.1.3.1073 (Vienna, Austria) with the package “effsize” (v0.8.0, Torchiano, 2020) [24]. Sample size calculation was performed using G*Power V.3.1.9.6 (Universität Kiel, Germany). Due to the explorative nature of this study and the small sample size, a correlation analysis was not performed. Future studies with sufficient statistical power to establish an association between cytokine levels and arterial stiffness changes are required.
## Demographic data
Patient´s characteristics at enrolment are shown in Table 1. Age and sex distributions were similar between the HIV[-] and treatment-naïve HIV(+) groups, with a higher prevalence of males in both groups. ART was initiated within one month of enrolment, fourteen patients ($73.6\%$) with EFV/TDF/FTC, two patients ($10.5\%$) with ATV/r/TDF/FTC, and three patients ($15.7\%$) with DVR/r/TDF/FTC. Among modifiable CVD risk factors, there was a significantly higher proportion of active smokers in the treatment-naïve HIV(+) group, with no changes in smoking at follow-up. BMI was similar between HIV[-] and treatment-naïve HIV(+) groups and did not change after one year of ART.
**Table 1**
| Unnamed: 0 | Unnamed: 1 | HIV (+) | HIV (+).1 |
| --- | --- | --- | --- |
| | HIV (-)(n = 9) | Naïve(n = 20) | Post-ART(n = 20) |
| Demographics | | | |
| Age, years | 34.4 ± 8.2 | 34.8 ± 10.1 | - |
| Male sex, % | 8 (88.8) | 19 (95) | - |
| Cigarette smoking, % | 1 (11.1) | 14 (70)* | 14 (70) |
| BMI, kg/m2 | 24.3 ± 1.8 | 25.3 ± 4.1 | 25.9 ± 4.2 |
| Metabolic profile | | | |
| Glucose, mg/dL | 87.5 ± 5.5 | 84.4 ± 8.7 | 85.4 ± 7.2 |
| TC, mg/dL | 185.6 ± 41.0 | 167.0 ± 41.5 | 185.6 ± 44.6† |
| c-LDL, mg/dL | 116.7 ± 37.1 | 100.2 ± 32.5 | 112.8 ± 38.3 |
| c-HDL, mg/dL | 50.2 ± 10.3 | 33.7 ± 9.4* | 35.3 ± 8.5 |
| TG, mg/dL | 93.0 ± 40.3 | 167.0 ± 67.5* | 176.3 ± 72.7* |
| Framingham score, % | 2.8 [1–3.3] | 3.3 (1.9–7.9) | 3.3 (2.3–7.0) |
| HIV status | | | |
| T CD4+, cells/mcL | - | 512 ± 324 | 727 ± 306† |
| T CD8+, cells/mcL | - | 977 (733–11374) | 859 (572–1180) |
| CD4+/CD8+ ratio | - | 0.39 (0.27–0.44) | 0.73 (0.39–1.48) † |
| Viral load ≥50 cop/mL, n (%) | - | 20 (100) | 0 (0) †† |
| Hemodynamics | | | |
| SBP, mmHg | 111 ± 11 | 108 ± 13 | 116 ± 12† |
| DBP, mmHg | 66 ± 8 | 67 ± 8 | 67 ± 10 |
| MAP, mmHg | 81 ± 7 | 80 ± 8 | 83 ± 10 |
| pPP, mmHg | 45 ± 14 | 39 ± 11 | 49 ± 11† |
| cfPWV, median, m/s | 6.8 (5.8–7.7) | 7.3 (6.7–8.1) | 6.8 (5.8–7.8) † |
## Metabolic data
Regarding the lipid profile, the treatment-naïve HIV(+) group exhibited higher TG ($$p \leq 0.003$$) and lower c-HDL ($p \leq 0.01$) compared to HIV[-], with no significant changes at one-year follow-up. Although TC and c-LDL levels were similar between groups at baseline, in the treatment-naïve HIV(+) group, TC significantly increased ($$p \leq 0.011$$) and c-LDL showed a trend toward a significant increase post-ART ($$p \leq 0.055$$). Plasma glucose levels were not significantly different between HIV[-] and treatment-naïve HIV(+) and did not change with ART. Both HIV[-] and treatment-naïve HIV(+) exhibited similar Framingham scores ($$p \leq 0.145$$), with no changes post-ART ($$p \leq 0.400$$).
## HIV variables
All treatment-naïve HIV(+) individuals achieved virologic suppression (viral load <50 copies/mL). CD4+ T-cell count increased by $36\%$ from baseline ($$p \leq 0.003$$), CD8+ T-cells did not change ($$p \leq 0.14$$), but the CD4+/CD8+ ratio increased at follow-up ($p \leq 0.001$).
## Blood pressure and arterial stiffness
The treatment-naïve HIV(+) group presented similar SBP, DBP, MAP, and pPP at baseline (all $p \leq 0.05$), compared to HIV[-], with a significant increase only in SBP ($$p \leq 0.04$$) and pPP ($$p \leq 0.005$$) at 1-year follow-up. Baseline cfPWV was similar between treatment-naïve HIV(+) and HIV[-] ($$p \leq 0.16$$); however, the post-ART group showed a significant reduction (-0.52 m/s; $95\%$ CI -0.87 to -0.16; $$p \leq 0.006$$) (Fig 1).
**Fig 1:** *Carotid-femoral pulse wave velocity (cfPWV) in HIV(-), treatment-naïve HIV(+), and after one-year follow-up of antiretroviral therapy (post-ART).*
## Inflammation Biomarkers
Inflammation biomarkers are shown in Table 2. At baseline, SAA, sCD163, IL-8, IL-18, calprotectin (i.e., MRP$\frac{8}{14}$), hs-CRP, and D-dimer levels were higher in treatment-naïve HIV(+) compared to HIV[-] (all $p \leq 0.05$), while MCP-1 levels were similar ($$p \leq 0.247$$). After ART, most biomarkers of inflammation significantly decreased ($p \leq 0.01$), except for D-dimer which showed a trend toward a significant decrease ($$p \leq 0.063$$). Similarly, hs-CRP ($$p \leq 0.187$$) was reduced to almost half its baseline value, but the difference was not statistically significant. Despite ART, hs-CRP levels remained significantly higher in the post-ART group than in the control group. The effect size of ART on inflammation biomarkers is shown in Fig 2A. Cliff’s delta showed a medium effect on SAA (-0.38; $95\%$ CI -0.10 to -0.65) and a large on sCD163 (-0.63, $95\%$ CI -0.27 to -0.83), IL-8 (-0.62; $95\%$ CI 0.27 to 0.82), IL-18 (-0.54; $95\%$ CI -0.16 to -0.77), calprotectin (-0.52; $95\%$ CI -0.16 to -0.76), and MCP-1 (-0.53; $95\%$ CI -0.16 to -0.77).
**Fig 2:** *Non-parametric effect size (Cliff’s delta) and $95\%$ confidence interval (error bars) of the effect of ART on a) inflammatory cytokines and b) markers of vascular inflammation. sCD163, soluble CD163; MCP-1, monocyte chemoattractant protein-1; SAA, serum amyloid A; IL-8, interleukin 8; IL-18; interleukin 18; MRP$\frac{8}{14}$, myeloid-related protein $\frac{8}{14}$ (calprotectin); NGAL, neutrophil gelatinase-associated lipocalin; MPO, myeloperoxidase; Cys C, cystatin C; ICAM, intercellular adhesion molecule VCAM, vascular cell adhesion molecule 1; MMP-9, matrix metalloproteinase-9. *$p \leq 0.05$ significant reduction compared to baseline.* TABLE_PLACEHOLDER:Table 2
## Vascular inflammation biomarkers
The vascular inflammation biomarkers are shown in Table 2. Treatment-naïve HIV(+) had higher D-dimer, ICAM-1, VCAM-1, and MPO levels, compared to HIV[-] (all $p \leq 0.05$), while neutrophil gelatinase-associated lipocalin (NGAL) tended to be higher ($$p \leq 0.08$$). Myoglobin, Cystatin C, and MMP-9 levels were similar between the treatment-naïve HIV(+) and HIV[-] groups. At follow-up, treatment-naïve HIV(+) showed a significant reduction in ICAM-1, VCAM-1, neutrophil gelatinase-associated lipocalin (NGAL), and MPO. On the other hand, MPO, ICAM-1, and VCAM1 were not significantly different between the post-ART and HIV[-], except for NGAL which was lower than HIV[-] ($$p \leq 0.029$$). The effect size of ART on vascular inflammation biomarkers is shown in Fig 2B. Cliff´s delta effect size showed a large effect on NGAL (-0.66; $95\%$ CI -0.30 to -0.85), MPO (-0.65; $95\%$ CI -0.29 to -0.85), and VCAM (-0.78; $95\%$ CI -0.45 to -0.92).
## Discussion
In this study, we investigated the effects of one-year ART on arterial stiffness and inflammatory and vascular cytokines levels in non-elderly, treatment-naïve PLH. First, we observed that the treatment-naïve HIV(+) group presented higher levels of inflammatory, vascular cytokines, and arterial stiffness compared to HIV[-] controls. Second, after a one-year follow-up, cfPWV and some cytokines significantly decreased from baseline levels (before ART) and reached levels similar to those in the control group. However, hs-CRP remained higher post-ART, which may suggest a persistent low-grade inflammatory state despite ART.
As previously reported by Smith et al., [ 25] we observed a higher smoking prevalence among the HIV population ($70\%$). We also found that the treatment-naïve HIV(+) group presented the typical dyslipidemia pattern in PLH, characterized by high TG and low HDL levels [26]. In our population, the most common treatment was EFV/TDF/FTC ($70\%$), which showed an unfavorable lipid profile outcome, as previously reported by Daar et al. [ 27].
The arterial system acts as a conduit for blood and as a buffering system to deliver a steady flow to key organs such as the brain and kidneys. Their buffering capacity depends on how elastic/stiff arteries are; in this sense, arterial elasticity is influenced by the vasomotor tone and structural characteristics of the arterial wall (elastin and collagen content). The vasomotor tone is regulated by the endothelium, sympathetic tone, and vasoactive hormones such as the renin-angiotensin-aldosterone system (RAAS). Nitric oxide (NO) is a key molecule responsible for regulating vasomotor tone. During inflammation, released cytokines can induce endothelial dysfunction by limiting the availability of tetrahydrobiopterin, the precursor of NO [28], and decrease the capacity of endothelial NO synthase (eNOS) to produce NO [29], subsequently limiting the capacity to reduce vasomotor tone. In HIV, several factors can affect the vasculature, including endothelial dysfunction [30], RAAS hyperactivation, infection of vascular smooth muscle cells, increased coagulation, chronic immune activation (by the virus itself and by microbial translocation caused by enteropathy), abnormal cholesterol metabolism, lipoprotein transportation, and platelet activation [31,32]. In our study, we found a significant decrease in several inflammatory and vascular cytokines, a halving of hs-CRP levels, and a decrease in PWV in the post-ART group. Our findings differ from those of Rose et al., [ 33] who did not observe a decrease in PWV after 1-year of ART. Conversely, Maia-Leite et al., [ 34] reported that in their ART-experienced group, $87.9\%$ of individuals virologically suppressed that aortic stiffness was similar to HIV[-] controls, implying the importance of viral suppression in arterial health. Our findings suggest that there is an improvement in arterial function, based on lower inflammation and a better environment for recovering normal arterial function. Nevertheless, we consider that future studies with complementary arterial evaluation techniques (e.g., flow mediation dilation) could provide better insights into the mechanisms responsible for recovering normal arterial function in inflammatory states.
The data obtained by measuring the vascular inflammation biomarkers in our study were consistent with previously reported data. Teasdale et al., [ 35] reported a significant decrease in D-dimer levels after 6 months of ART; [36,37]. In our study, we observed an overall trend toward a significant decrease in D-dimer post-ART. However, patients treated with ART based on transcriptase inhibitors (both non-nucleoside analogue and nucleoside analogue) presented a reduction ($79\%$) in D-dimer levels, while patients with ritonavir-boosted PI-based ART D-dimer increased by 1.35 times. These data are consistent with the association of ritonavir-boosted PI exposure time with CVD risk and suggest that it can appear as soon as a year of ritonavir-boosted PI-based ART [36].
In the present study, MPO concentrations decreased after ART and could be associated with a PWV improvement. El-Bejanni et al. [ 38] reported a negative correlation between MPO and CD4+ T-cell counts. Thus, immune reconstitution in these individuals is accompanied by a decrease in MPO and PWV. MPO and aortic stiffness have been related to plaque instability in atherosclerotic disease [37,39,40]. Therefore, it could be considered that the global effect of one year on ART is reflected by an improvement in arterial health and probably less CVD risk.
Another interesting vascular marker is NGAL, which is associated with inflammation, leukocyte migration, carotid stenosis, endothelial dysfunction, plaque formation, acute myocardial infarction, and chronic heart failure [41,42]. It has been demonstrated that NGAL can be expressed in macrophages, smooth muscle cells, and endothelial cells; moreover, it can activate the NF-κB pathway, promoting the expression of several proinflammatory cytokines such as IL-8, MCP-1, TNF-α, and IL-1β [43]. We found a significant decrease in NGAL levels, which could be proposed as a novel cardiovascular biomarker in PLWH.
ICAM-1 and VCAM-1 have been associated with atherosclerosis [44] and CVD risk even in HIV[-] population [30,35,37,38]. In our study, we observed a substantial decrease in the levels of VCAM-1 after ART, as reported by other groups [45], reaching similar levels to the HIV[-] population, as well as ICAM-1 which showed a reduction although non statistically significant. MPO and aortic stiffness have been related to plaque instability in atherosclerotic disease [37,39,40]. In the present study, MPO concentrations also decreased after ART and could be associated with a PWV improvement. El-Bejanni et al. [ 38] reported a negative correlation between MPO and CD4+ T-cell counts. Thus, the immune reconstitution of these individuals is accompanied by a decrease in both variables; therefore, it could be considered that the global effect of one year on ART is reflected in an improvement in arterial health and probably less CVD risk.
It has been reported that PLH exacerbates the production of proinflammatory cytokines, such as IL-1β, IL-8, and TNF-α, and molecules related to non-canonical activation pathways of coagulation, such as IL-6, hs-CRP, SSA, and D-dimer [38,46–49]. In line with those studies, we observed that the concentrations of SAA, sCD163, MCP-1, IL-8, and IL-18 were increased in PLH, but after a year of treatment, most HIV (+) patients reached concentrations similar to controls, except for SAA and sCD163. These results suggest that ART reduced the degree of systemic inflammation and improved arterial stiffness after one year of treatment.
It is well described that ART does not completely suppress viral load, particularly in the viral reservoirs. Hence, low-grade chronic inflammation persists and contributes to an increased risk of non-AIDS-defining events such as CVD [50,51]. As we previously mentioned, there was a reduction of SAA, sCD163, and calprotectin after a year of successful ART; however, the levels remained higher compared to controls. SAA, like CRP, is an acute-phase protein produced by the liver. It is known that SAA can establish an inflammatory atherosclerotic and thrombotic microenvironment that impacts immune dysfunction and promotes CVD [52,53]. In our study, we observed that SAA was higher in treatment-naïve compared to controls and reduced post-ART to levels similar to those in HIV[-]. High concentrations of calprotectin were detected before ART; however, serum levels normalized after the virologic control, this pattern is consistent with previous data reported since calprotectin is one of the first proteins to increase in plasma, even earlier than other markers of myocardial necrosis. Finally, calprotectin promotes atherosclerosis, and elevated plasma concentrations of calprotectin are considered predictors of future CV events [42,43,46].
As PLH live longer, new widely available biomarkers are necessary to properly evaluate the overall health of this population and to prepare physicians for the more plausible complications their patients may have. This study has some prospectively analyzed biomarkers that could be of use; however, they require further investigation to evaluate their long-term effectiveness in the daily clinical field.
These results are consistent with the idea that PLH develops a low-grade inflammation state and metabolic imbalance from the beginning of HIV infection, which remains despite ART. Furthermore, it seems that besides the current therapy goals (i.e., CD4+ T-cell count and HIV RNA levels) it would be advisable to monitor arterial health with different available techniques to try to lower the CV risk in this population, which increases with age and ART exposure.
## Strengths and limitations
This study has some limitations. Despite our aim to isolate the effect of HIV on arterial stiffness by choosing PLH patients without any other comorbidities and comparing them with healthy controls, there may be a series of genetic, lifestyle, social, and behavioral factors that differ between the populations studied, which could not be controlled for. Our study sample included mostly males, and sex differences could play a role in the response of arterial stiffness and levels of inflammatory and vascular biomarkers to ART. Another limitation of the present study was the absence of cytokine reference values for the Mexican population to determine whether HIV-infected individuals achieved normal cytokine levels after ART. The sample size of the control group was smaller than that of the HIV(+) group, which may have been underpowered for some biomarkers. Lastly, we only measured PWV twice (baseline and at 12 months); thus, we could not determine how far into the treatment the PWV and cytokines started to decrease. Nevertheless, the strength of this study is the prospective evaluation of arterial health and serum biomarkers of inflammation which allowed us to study the effects of ART.
## Conclusion
One year of ART had a positive effect on reducing both inflammatory and vascular cytokines, and arterial stiffness.
## References
1. 1Global HIV & AIDS statistics—Fact sheet | UNAIDS. [cited 31 Jul 2022]. Available: https://www.unaids.org/en/resources/fact-sheet.
2. 2Noticia—Estadísticas a propósito del Día mundial de la lucha contra el VIH/SIDA (1 de diciembre). [cited 31 Jul 2022]. Available: https://www.inegi.org.mx/app/saladeprensa/noticia.html?id=6976.
3. Hsue PY, Waters DD. **HIV infection and coronary heart disease: mechanisms and management**. *Nat Rev Cardiol* (2019.0) **16** 745-759. DOI: 10.1038/s41569-019-0219-9
4. Vachiat A, McCutcheon K, Tsabedze N, Zachariah D, Manga P. **HIV and Ischemic Heart Disease**. *J Am Coll Cardiol* (2017.0) **69** 73-82. DOI: 10.1016/j.jacc.2016.09.979
5. Di Yacovo S, Saumoy M, Sánchez-Quesada JL, Navarro A, Sviridov D, Javaloyas M. **Lipids, biomarkers, and subclinical atherosclerosis in treatment-naive HIV patients starting or not starting antiretroviral therapy: Comparison with a healthy control group in a 2-year prospective study**. *PLoS One* (2020.0) 15. DOI: 10.1371/journal.pone.0237739
6. Seaberg EC, Benning L, Sharrett AR, Lazar JM, Hodis HN, MacK WJ. **Association between human immunodeficiency virus infection and stiffness of the common carotid artery**. *Stroke* (2010.0) 41. DOI: 10.1161/STROKEAHA.110.583856
7. Jiang J, Fu W, Wang X, Lin PH, Yao Q, Chen C. **HIV gp120 induces endothelial dysfunction in tumour necrosis factor-alpha-activated porcine and human endothelial cells**. *Cardiovasc Res* (2010.0) **87** 366-74. DOI: 10.1093/cvr/cvq013
8. Wang T, Green LA, Gupta SK, Kim C, Wang L, Almodovar S. **Transfer of intracellular HIV Nef to endothelium causes endothelial dysfunction**. *PLoS One* (2014.0) **9** e91063. DOI: 10.1371/journal.pone.0091063
9. Lewis DE, Couturier JP. *Translational Inflammation* (2019.0) 111-131. DOI: 10.1016/B978-0-12-813832-8.00006–6
10. Triant VA, Perez J, Regan S, Massaro JM, Meigs JB, Grinspoon SK. **Cardiovascular Risk Prediction Functions Underestimate Risk in HIV Infection**. *Circulation* (2018.0) **137** 2203-2214. DOI: 10.1161/CIRCULATIONAHA.117.028975
11. Willum-Hansen T, Staessen JA, Torp-Pedersen C, Rasmussen S, Thijs L, Ibsen H. **Prognostic value of aortic pulse wave velocity as index of arterial stiffness in the general population**. *Circulation* (2006.0) **113** 664-70. DOI: 10.1161/CIRCULATIONAHA.105.579342
12. Laurent S, Cockcroft J, Van Bortel L, Boutouyrie P, Giannattasio C, Hayoz D. **Expert consensus document on arterial stiffness: methodological issues and clinical applications**. *Eur Heart J* (2006.0) **27** 2588-605. DOI: 10.1093/eurheartj/ehl254
13. Kuate Defo A, Chalati MD, Labos C, Fellows LK, Mayo NE, Daskalopoulou SS. **Association of HIV Infection and Antiretroviral Therapy With Arterial Stiffness: A Systematic Review and Meta-Analysis**. *Hypertension* (2021.0) **78** 320-332. DOI: 10.1161/HYPERTENSIONAHA.121.17093
14. Lopez-Sublet M, Honoré P, Bentata M, Bratis C, Rouges F, Krivitzky A. **Protease inhibitor treatment effect on aortic stiffness in normotensive patients with human immunodeficiency virus infection**. *J Mal Vasc* (2012.0) **37** 179-85. DOI: 10.1016/j.jmv.2012.05.001
15. Eira M, Bensenor IM, Dorea EL, Cunha RS, Mill JG, Lotufo PA. **Potent antiretroviral therapy for human immunodeficiency virus infection increases aortic stiffness**. *Arq Bras Cardiol* (2012.0) **99** 1100-1107. DOI: 10.1590/s0066-782x2012005000110
16. van Vonderen MGA, Hassink EAM, van Agtmael MA, Stehouwer CDA, Danner SA, Reiss P. **Increase in carotid artery intima-media thickness and arterial stiffness but improvement in several markers of endothelial function after initiation of antiretroviral therapy**. *J Infect Dis* (2009.0) **199** 1186-94. DOI: 10.1086/597475
17. Maloberti A, Dozio D, Betelli M, Bandera A, Squillace N, Gori A. **Brachial and central blood pressure in HIV-infected subjects**. *Hypertens Res* (2015.0) **38** 405-412. DOI: 10.1038/hr.2015.25
18. Martínez-Ayala P, Alanis-Sánchez GA, González-Hernández LA, Álvarez-Zavala M, Cabrera-Silva RI, Andrade-Villanueva JF. **Aortic stiffness and central hemodynamics in treatment-naïve HIV infection: a cross-sectional study**. *BMC Cardiovasc Disord* (2020.0) **20** 440. DOI: 10.1186/s12872-020-01722-8
19. Rahmanian S, Wewers ME, Koletar S, Reynolds N, Ferketich A, Diaz P. **Cigarette smoking in the HIV-infected population**. *Proc Am Thorac Soc* (2011.0) **8** 313-9. DOI: 10.1513/pats.201009-058WR
20. Townsend RR, Wilkinson IB, Schiffrin EL, Avolio AP, Chirinos JA, Cockcroft JR. **Recommendations for Improving and Standardizing Vascular Research on Arterial Stiffness: A Scientific Statement from the American Heart Association**. *Hypertension* (2015.0) 66. DOI: 10.1161/HYP.0000000000000033
21. Cliff N.. **Dominance statistics: Ordinal analyses to answer ordinal questions**. *Psychol Bull* (1993.0) 114. DOI: 10.1037/0033-2909.114.3.494
22. Osuji FN, Onyenekwe CC, Ahaneku JE, Ukibe NR. **The effects of highly active antiretroviral therapy on the serum levels of pro-inflammatory and anti-inflammatory cytokines in HIV infected subjects**. *J Biomed Sci* (2018.0) **25** 88. DOI: 10.1186/s12929-018-0490-9
23. O’Halloran JA, Dunne E, Gurwith M, Lambert JS, Sheehan GJ, Feeney ER. **The effect of initiation of antiretroviral therapy on monocyte, endothelial and platelet function in HIV-1 infection**. *HIV Med* (2015.0) **16** 608-19. DOI: 10.1111/hiv.12270
24. Torchiano M.. **effsize: Efficient effect size computation**. *R package version 08* (2020.0) 1
25. Smith J. C, Sabin A. C, Lundgren D. J, Thiebaut R, Weber R, Law M. **Factors associated with specific causes of death amongst HIV-positive individuals in the D:A:D study**. *AIDS* (2010.0) 24. DOI: 10.1097/qad.0b013e32833a0918
26. Grunfeld C, Pang M, Doerrler W, Shigenaga JK, Jensen P, Feingold KR. **Lipids, lipoproteins, triglyceride clearance, and cytokines in human immunodeficiency virus infection and the acquired immunodeficiency syndrome**. *J Clin Endocrinol Metab* (1992.0) **74** 1045-52. DOI: 10.1210/jcem.74.5.1373735
27. Daar ES, Tierney C, Fischl MA, Sax PE, Mollan K, Budhathoki C. **Atazanavir plus ritonavir or efavirenz as part of a 3-drug regimen for initial treatment of HIV-1**. *Ann Intern Med* (2011.0) **154** 445-56. DOI: 10.7326/0003-4819-154-7-201104050-00316
28. Gunnett CA, Lund DD, McDowell AK, Faraci FM, Heistad DD. **Mechanisms of inducible nitric oxide synthase-mediated vascular dysfunction**. *Arterioscler Thromb Vasc Biol* (2005.0) **25** 1617-22. DOI: 10.1161/01.ATV.0000172626.00296.ba
29. Venugopal SK, Devaraj S, Yuhanna I, Shaul P, Jialal I. **Demonstration that C-reactive protein decreases eNOS expression and bioactivity in human aortic endothelial cells**. *Circulation* (2002.0) **106** 1439-41. DOI: 10.1161/01.cir.0000033116.22237.f9
30. Solages A, Vita JA, Thornton DJ, Murray J, Heeren T, Craven DE. **Endothelial function in HIV-infected persons**. *Clin Infect Dis* (2006.0) **42** 1325-32. DOI: 10.1086/503261
31. Mu H, Chai H, Lin PH, Yao Q, Chen C. **Current update on HIV-associated vascular disease and endothelial dysfunction**. *World J Surg* (2007.0) **31** 632-43. DOI: 10.1007/s00268-006-0730-0
32. Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. **Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events**. *N Engl J Med* (2002.0) **347** 1557-65. DOI: 10.1056/NEJMoa021993
33. Rose H, Low H, Dewar E, Bukrinsky M, Hoy J, Dart A. **The effect of HIV infection on atherosclerosis and lipoprotein metabolism: a one year prospective study**. *Atherosclerosis* (2013.0) **229** 206-11. DOI: 10.1016/j.atherosclerosis.2013.04.010
34. Maia-Leite LH, Catez E, Boyd A, Haddour N, Curjol A, Lang S. **Aortic stiffness aging is influenced by past profound immunodeficiency in HIV-infected individuals: results from the EVAS-HIV (EValuation of Aortic Stiffness in HIV-infected individuals).**. *J Hypertens* (2016.0) **34** 1338-46. DOI: 10.1097/HJH.0000000000000957
35. Teasdale CA, Hernandez C, Zerbe A, Chege D, Hawken M, El-Sadr WM. **Changes in D-dimer after initiation of antiretroviral therapy in adults living with HIV in Kenya**. *BMC Infect Dis.* (2020.0) 20. DOI: 10.1186/S12879-020-05213-1
36. Martínez E, D’Albuquerque PM, Llibre JM, Gutierrez F, Podzamczer D, Antela A. **Changes in cardiovascular biomarkers in HIV-infected patients switching from ritonavir-boosted protease inhibitors to raltegravir**. *AIDS* (2012.0) 26. DOI: 10.1097/QAD.0b013e328359f29c
37. Neuhaus J, Jacobs DR, Baker JV, Calmy A, Duprez D, La Rosa A. **Markers of inflammation, coagulation, and renal function are elevated in adults with HIV infection**. *J Infect Dis* (2010.0) **201** 1788-95. DOI: 10.1086/652749
38. El-Bejjani D, Hazen SL, Mackay W, Glass NE, Hulgan T, Tungsiripat M. **Higher plasma myeloperoxidase levels are not associated with an increased risk for cardiovascular events in HIV-infected adults**. *HIV Clin Trials* (2008.0) **9** 207-211. DOI: 10.1310/hct0903-207
39. O’Brien KD, McDonald TO, Chait A, Allen MD, Alpers CE. **Neovascular expression of E-selectin, intercellular adhesion molecule-1, and vascular cell adhesion molecule-1 in human atherosclerosis and their relation to intimal leukocyte content**. *Circulation* (1996.0) **93** 672-682. DOI: 10.1161/01.cir.93.4.672
40. Hwang SJ, Ballantyne CM, Sharrett AR, Smith LC, Davis CE, Gotto AM. **Circulating adhesion molecules VCAM-1, ICAM-1, and E-selectin in carotid atherosclerosis and incident coronary heart disease cases: the Atherosclerosis Risk In Communities (ARIC) study**. *Circulation* (1997.0) **96** 4219-4225. DOI: 10.1161/01.cir.96.12.4219
41. Papasavvas E, Azzoni L, Pistilli M, Hancock A, Reynolds G, Gallo C. **Increased soluble vascular cell adhesion molecule-1 plasma levels and soluble intercellular adhesion molecule-1 during antiretroviral therapy interruption and retention of elevated soluble vascular cellular adhesion molecule-1 levels following resumption of antiretroviral therapy**. *AIDS* (2008.0) **22** 1153-1161. DOI: 10.1097/QAD.0b013e328303be2a
42. Healy AM, Pickard MD, Pradhan AD, Wang Y, Chen Z, Croce K. **Platelet expression profiling and clinical validation of myeloid-related protein-14 as a novel determinant of cardiovascular events**. *Circulation* (2006.0) **113** 2278-84. DOI: 10.1161/CIRCULATIONAHA.105.607333
43. Croce K, Gao H, Wang Y, Mooroka T, Sakuma M, Shi C. **Myeloid-related protein-8/14 is critical for the biological response to vascular injury**. *Circulation* (2009.0) **120** 427-36. DOI: 10.1161/CIRCULATIONAHA.108.814582
44. Murdaca G, Colombo BM, Cagnati P, Gulli R, Spanò F, Puppo F. **Endothelial dysfunction in rheumatic autoimmune diseases**. *Atherosclerosis* (2012.0) **224** 309-17. DOI: 10.1016/j.atherosclerosis.2012.05.013
45. Martínez E, D’Albuquerque PM, Llibre JM, Gutierrez F, Podzamczer D, Antela A. **Changes in cardiovascular biomarkers in HIV-infected patients switching from ritonavir-boosted protease inhibitors to raltegravir**. *AIDS* (2012.0) **26** 2315-2326. DOI: 10.1097/QAD.0b013e328359f29c
46. Pedersen L, Nybo M, Poulsen MK, Henriksen JE, Dahl J, Rasmussen LM. **Plasma calprotectin and its association with cardiovascular disease manifestations, obesity and the metabolic syndrome in type 2 diabetes mellitus patients**. *BMC Cardiovasc Disord* (2014.0) **14** 196. DOI: 10.1186/1471-2261-14-196
47. Soylu K, Nar G, Aksan G, Gedikli Ö, İnci S, Yuksel S. **Serum neutrophil gelatinase-associated lipocalin levels and aortic stiffness in noncritical coronary artery disease**. *Cardiorenal Med* (2014.0) **4** 147-54. DOI: 10.1159/000365200
48. Shrestha K, Borowski AG, Troughton RW, Klein AL, Tang WHW. **Association between systemic neutrophil gelatinase-associated lipocalin and anemia, relative hypochromia, and inflammation in chronic systolic heart failure**. *Congest Heart Fail* (2012.0) **18** 239-244. DOI: 10.1111/j.1751-7133.2012.00287.x
49. Eilenberg W, Stojkovic S, Piechota-Polanczyk A, Kaun C, Rauscher S, Gröger M. **Neutrophil Gelatinase-Associated Lipocalin (NGAL) is Associated with Symptomatic Carotid Atherosclerosis and Drives Pro-inflammatory State In Vitro**. *Eur J Vasc Endovasc Surg* (2016.0) **51** 623-31. DOI: 10.1016/j.ejvs.2016.01.009
50. Rajasuriar R, Wright E, Lewin SR. **Impact of antiretroviral therapy (ART) timing on chronic immune activation/inflammation and end-organ damage**. *Curr Opin HIV AIDS* (2015.0) **10** 35-42. DOI: 10.1097/COH.0000000000000118
51. Višković K, Židovec Lepej S, Gorenec A, Grgić I, Lukas D, Zekan Š. **Cardiovascular markers of inflammation and serum lipid levels in HIV-infected patients with undetectable viremia**. *Sci Rep* (2018.0) **8** 6113. DOI: 10.1038/s41598-018-24446-4
52. Malle E, Steinmetz A, Raynes JG. **Serum amyloid A (SAA): an acute phase protein and apolipoprotein**. *Atherosclerosis* (1993.0) **102** 131-146. DOI: 10.1016/0021-9150(93)90155-n
53. Samikkannu T, Rao KVK, Arias AY, Kalaichezian A, Sagar V, Yoo C. **HIV infection and drugs of abuse: role of acute phase proteins**. *J Neuroinflammation* (2013.0) **10** 113. DOI: 10.1186/1742-2094-10-113
|
---
title: The impact of access to financial services on mitigating COVID-19 mortality
globally
authors:
- Todd A. Watkins
- Khue Nguyen
- Hamza Ali
- Rishikesh Gummakonda
- Jacques Pelman
- Brianna Taracena
journal: PLOS Global Public Health
year: 2023
pmcid: PMC10022804
doi: 10.1371/journal.pgph.0001137
license: CC BY 4.0
---
# The impact of access to financial services on mitigating COVID-19 mortality globally
## Abstract
The COVID-19 pandemic has disproportionately affected different social and demographic groups, deepening the negative health implications of social and economic inequalities and highlighting the importance of social determinants of health. Despite a deep literature on pandemic-related disparities, specifically regarding social determinants and health outcomes, the influence of the accessibility of financial services on health outcomes during COVID-19 remains largely unexplored. Modeling (pre-omicron) COVID-19 mortality across 142 nations, we assess the impact of national-level usage and access to formal financial services. Two financial access indexes constructed through principal component analysis capture [1] usage of and access to formal financial tools and [2] reliance on alternative and informal financial tools. On average, nations with higher pre-pandemic use of and access to formal financial services had substantially lower population mortality risk from COVID-19, controlling for key population health, demographic, and socioeconomic covariates. The scale of effect is similar in magnitude—but opposite in direction—to major risk factors identified in previous literature, such as lung cancer, hypertension, and income inequality. Findings suggest that financial services deserve greater attention both in the public health literature related to COVID-19 and more broadly in policy discussions about fostering better public health overall.
## Introduction
The COVID-19 pandemic has both highlighted the importance of social determinants of health and exacerbated the negative health implications of social and economic inequalities. There is broad literature and evidence that inequalities in income, access to resources, and other measures of household financial security affect household members’ health [1, 2]. Specific to the pandemic, two systematic reviews of the burgeoning literature on the social determinants related to COVID-19 health outcomes, one early in the pandemic [3] and one more recently by the World Health Organization [4]—more than 200 articles and reports are included in the latter alone—revealed “glaring inequities” among population groups; main contributing factors include poverty, lack of household resources, affordability of prevention measures, and limited access to various health and social services. Despite the hundreds of such studies related to the pandemic, and the obvious importance of, e.g., ability to pay, affordability, and payment options for accessing health services, none addressed access to broader arrays of financial services beyond formal insurance as a determinant of health outcomes from COVID-19. Yet access to savings, credit, money transfer services, and the like can also serve to buffer risks [5]. The preliminary evidence presented in the sections that follow suggests this may be a major gap in the literature.
Before COVID-19 hit, many developing economies had seen expansions in financial access such as banking deposit and credit services, digital payment systems, and other inclusive fintech innovations. For decades, advocates for global-scale investments in inclusive financial services have championed the role that improved access to financial services plays in helping families in developing economies, particularly those reliant on income from informal work, buffer the risks to their livelihoods and health they face from highly variable and unstable income flows [6–11]. Job losses and economic turmoil due to COVID-19 have exacerbated those risks to an unprecedented degree globally.
Yet it remains an open question whether investments in inclusive financial services and inclusive fintech have mattered for global health. Because health is such a complex socio-economic phenomenon, and financial access is only one (perhaps minor) element among many confounding factors, teasing out the health effects of expanding financial services has proven remarkably challenging. The unfortunately stark and broad natural experiment of COVID-19 might have strengthened the signal-to-noise ratio enough to present a unique opportunity to deepen understanding of the role of access to financial services in health outcomes.
Most of the literature to date linking COVID-19 and financial services has focused on how the pandemic has challenged the financial security of financial sector institutions (e.g., [12–14]), of businesses [15–17], and of households [18–20]. Similarly, the public health literature beyond COVID-19 has focused on how health problems can cause financial risk, including so-called distress financing, the phenomenon of lower-income households coping with health crises and health-related negative income shocks by relying on selling assets and over-indebtedness, in turn begetting further financial risk for the households [21–23]. Distress financing tends to be higher where insurance is thin and national systems lack universal healthcare.
But causality should run the other direction as well if financial security is beneficial to health. Advocates for inclusive finance might have hoped that financial access would have some impact on risk mitigation during the pandemic. Did it? If advocates have been correct, families, communities, and nations that had better financial access pre-COVID-19 should have been better able to weather the health storm in measurable ways during the pandemic. Somewhat surprisingly, there so far has been no global study exploring how the pandemic’s effects correlate with financial access metrics and whether there is any evidence of risk mitigation.
The analysis that follows models COVID-19 mortality rates across 142 nations and finds that nations where residents had greater (pre-pandemic) access to an array of formal financial services had substantially lower population mortality risk during the pandemic. Indeed, the risk reduction is surprisingly large, similar in magnitude to but opposite in direction from the increased COVID-19 mortality risks associated with higher rates of lung cancer, hypertension, and greater income inequality.
## Variables and data sources
Table 1 lists the dependent and control variables, details on the specific metrics used, and data sources; all are publicly available and measured at the national level. The dependent variable is (log-transformed) COVID-19 deaths per million population through the end of September 2021. This cutoff date was before the discovery and spread of the omicron variant. The independent variables listed in Table 1 aim to capture and control for factors known to contribute to variation across nations in COVID-19 mortality rates. All independent variables use the most recently available data predating the pandemic, i.e., 2019 or earlier. Variables with highly skewed distributions are natural log transformed. Two additional variables of main interest are financial services access indexes, constructed as discussed in the next section.
**Table 1**
| Variables | Measures | Sources |
| --- | --- | --- |
| COVID-19 death rate, through 9/30/21 | ln(COVID-19 deaths per million population) | Accessed from Our World in Data, https://ourworldindata.org/covid-deaths; original source: COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University, https://github.com/CSSEGISandData/COVID-19 [44] |
| Demographic & Socioeconomic Control Variables | Demographic & Socioeconomic Control Variables | Demographic & Socioeconomic Control Variables |
| Population aged 65 & older, 2019 | % of population | United Nations, 2019 World Population Prospects, https://population.un.org/wpp/ |
| Population aged 0–14, 2019 | % of population | United Nations, 2019 World Population Prospects, https://population.un.org/wpp/ |
| Population density, 2019 | ln(population per sq. km) | United Nations, 2019 World Population Prospects, https://population.un.org/wpp/ |
| Population in urban areas, 2018 | % of population | United Nations, World Urbanization Prospects: the 2018 Revision, https://population.un.org/wup/ |
| Per capita income, 2019 | ln(per capita income, PPP adjusted, constant 2017 international $) | World Bank, World Development Indicators, https://data.worldbank.org/indicator/NY.GDP.PCAP.PP.CD |
| Income inequality, 2019 | Gini coefficient | Gapminder estimates, http://www.gapm.io/dgini |
| Population Health Control Variables | Population Health Control Variables | Population Health Control Variables |
| Mortality from indoor air pollution, 2016 | ln(mortality rate attributed to household & ambient air pollution, age-standardized, per 100,000 population) | World Bank, World Development Indicators, https://data.worldbank.org/indicator/SH.STA.AIRP.P5 |
| Diabetes prevalence, 2019 | ln(% of population ages 20–79 with type 1 or 2 diabetes, age-standardized) | World Bank, World Development Indicators, https://data.worldbank.org/indicator/SH.STA.DIAB.ZS |
| Lung cancer prevalence, 2018 | ln(average lung cancer incidence rate, age-standardized per 100,000, male & female, all ages) | Accessed from https://canceratlas.cancer.org; original source: International Agency for Research on Cancer, Global Cancer Observatory: Cancer Today, https://gco.iarc.fr/today [45] |
| Body mass index, 2016 | Mean body mass index, adults, age-standardized (kg/m2) | World Health Organization, Global Health Observatory, https://apps.who.int/gho/data/view.main.CTRY12461 |
| Raised blood pressure prevalence, 2015 | % of population 18+ years, age-standardized, with systolic blood pressure ≥ 140 or diastolic blood pressure ≥ 90 | World Health Organization, Global Health Observatory, https://www.who.int/data/gho/indicator-metadata-registry/imr-details/2386 |
| Tuberculosis vaccine coverage, 1989–2018 | Mean bacille Calmette-Guérin (BGC) immunization rate (%) among 1-year olds, 1989–2018 | World Health Organization, Global Health Observatory, https://www.who.int/data/gho/data/indicators/indicator-details/GHO/bcg-immunization-coverage-among-1-year-olds-(-) |
| Health Infrastructure Control Variables | Health Infrastructure Control Variables | Health Infrastructure Control Variables |
| Nurses & midwives, 2010–2019 | ln(nursing & midwifery personnel per 10,000 population) | World Health Organization, Global Health Observatory, https://www.who.int/data/gho/indicator-metadata-registry/imr-details/5319. Used most recently available from 2010–2019. |
| Health services effective coverage, 2019 | Universal healthcare (UHC) effective coverage index | Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019, UHC Effective Coverage Index 1990–2019, doi: 10.6069/GT4K-3B35 [46] |
Population-level COVID-19 mortality risks are, by now, well understood in the medical literature to be related to a variety of population-level risk factors. Though our main interest here is not these previously studied population health and socioeconomic variables, we include some key national-level public health risk characteristics as control variables. These control variables were selected based on previous evidence in the literature as covariates with COVID-19 mortality, including population health and respiratory issues (levels of hypertension, lung cancer, tuberculosis, diabetes, obesity, and air pollution); healthcare system availability and effectiveness; and socio-economic characteristics, including age distribution, urbanization, population density, and socioeconomic inequalities (e.g., [24–43]). To protect, temporally, against bias from reverse causality, all independent variables included in our main models (Tables 5, 6, and S4 Table) reflect national conditions in 2019 or earlier, prior to the arrival of COVID-19. ( In S5 Table we do explore the addition of June 2021 COVID-19 vaccination rates and availability, three months before the mortality variable; main conclusions are unaffected).
In addition, we also screened for data availability. Much of the statistical modeling in the national-level literature suffers from limited data availability across nations, restricting sample size. Aiming for the broadest and most representative set of countries, we selected publicly available variables at the national level that had no more than five nations missing from the 142 nations for which financial inclusion variables and COVID-19 mortality rates were also available. S1 Table shows descriptive statistics for all variables in Table 1, and S3 *Table is* the correlation matrix.
## Financial access variables methodology
To assess the extent to which access to financial services has contributed to mitigating COVID-19 risks, we develop financial inclusion indices at the national level using metrics from the World Bank’s Global Findex Database [47]. The Global Findex includes multiple variables measuring the degree to which the population of each country is engaged in the use of various financial services and activities (as percentage of population). The data are collected through national surveys of more than 150,000 adults from 144 countries, of which 142 also had COVID-19 mortality data available.
We construct our financial access indexes from principal component analysis (PCA) on 20 metrics drawn from the Global Findex Database 2017 panel, the most recent available pre-pandemic. The 2017 panel includes indicators on the access and use of both formal and informal financial services as well as the use of various financial technologies. As described in Table 2, with descriptive statistics in S2 Table, several metrics relate to access to and usage of traditional formal financial tools, such as having any sort of financial institution account, borrowing from and saving in financial institutions, making transactions through financial institutions, and owning debit or credit cards. Other variables measure the use of digital financial services such as making transactions through mobile phones or payments via the internet. We also incorporate several variables indicative of access to financial security tools such as the availability of emergency funds, loans from financial institutions, borrowing for health purposes, and saving for old age. Additional variables measure the use of informal or distress finance, e.g., relying on friends and family for borrowing or for emergencies, or selling assets for emergency funding.
**Table 2**
| Metrics capturing access to traditional formal financial tools | Metrics capturing access to traditional formal financial tools.1 |
| --- | --- |
| Financial institution account | Respondents who report having an account (by themselves or together with someone else) at a bank or another type of financial institution |
| Borrowed from a financial institution or credit card | Respondents who borrowed any money from a bank or another type of financial institution, or using a credit card, in the past 12 months |
| Saved at a financial institution | Respondents who report saving or setting aside any money at a bank or another type of financial institution in the past 12 months |
| Debit card ownership | Respondents who report having a debit card |
| Credit card ownership | Respondents who report having a credit card |
| Received wages into a financial institution account | Respondents who received any money from an employer in the past 12 months in the form of a salary or wages for doing work directly into a financial institution account or into a card |
| Paid utility bills using a financial institution account | Respondents who personally made regular payments for water, electricity, or trash collection in the past 12 months directly from a financial institution account |
| Outstanding housing loan | Respondents who report having an outstanding loan (by themselves or with someone else) from a bank or another type of financial institution to purchase a home, an apartment, or land |
| Metrics capturing access to financial technologies (fintech) | Metrics capturing access to financial technologies (fintech) |
| Used the internet to pay bills or to buy something online in the past year | Respondents who report using the internet to pay bills or buy something online in the past 12 months |
| Received wages through a mobile phone | Respondents who received any money from an employer in the past 12 months in the form of a salary or wages for doing work through a mobile phone |
| Paid utility bills using a mobile phone | Respondents who personally made regular payments for water, electricity, or trash collection in the past 12 months through a mobile phone |
| Made or received digital payments in the past year | Respondents who used mobile money, a debit or credit card, or a mobile phone to make a payment from an account, or used the internet to pay bills or to buy something online, in the past 12 months |
| Metrics capturing access to financial security tools | Metrics capturing access to financial security tools |
| Saved for old age | Respondents who saved or set aside any money in the past 12 months for old age |
| Borrowed for health or medical purposes | Respondents who borrowed any money for health or medical purposes in the past 12 months |
| Coming up with emergency funds possible | Respondents who, in case of an emergency, it is possible for them to come up with 1/20 of gross national income (GNI) per capita in local currency within the next month |
| Main source of emergency funds: savings | Among respondents reporting that in case of an emergency it is possible for them to come up with 1/20 of GNI per capita in local currency, the percentage who cite savings as their main source of this money |
| Main source of emergency funds: formal loans | Among respondents reporting that in case of an emergency it is possible for them to come up with 1/20 of GNI per capita in local currency, the percentage who cite borrowing from a bank, an employer, or a private lender as their main source of this money |
| Metrics capturing reliance on informal or distress financing | Metrics capturing reliance on informal or distress financing |
| Borrowed from family or friends | Respondents who borrowed any money from family, relatives, or friends in the past 12 months |
| Main source of emergency funds: family or friends | Among respondents reporting that in case of an emergency it is possible for them to come up with 1/20 of GNI per capita in local currency, the percentage who cite family, relatives, or friends as their main source of this money |
| Main source of emergency funds: sale of assets | Among respondents reporting that in case of an emergency it is possible for them to come up with 1/20 of GNI per capita in local currency, the percentage who cited the sale of assets as their main source of this money |
To reduce the dimensionality of our exploration of whether countries with greater access to financial services have better success mitigating COVID-19 risks while retaining most of the variability of the underlying metrics, we apply PCA. Our PCA incorporates all 20 variables in Table 2, from the Global Findex Database, to create indices of financial access at the national level. As shown in Table 3, the two principal components with the highest eigenvalues together account for nearly three-quarters ($72\%$) of the total variance among the 20 variables. We use these top two to construct our two financial access indexes.
**Table 3**
| Component | Eigenvalue | Difference | Proportion of Variance | Cumulative Variance |
| --- | --- | --- | --- | --- |
| Comp1 | 12.400 | 10.466 | 0.6200 | 0.6200 |
| Comp2 | 1.935 | 0.630 | 0.0967 | 0.7168 |
| Comp3 | 1.305 | 0.289 | 0.0653 | 0.7820 |
| Comp4 | 1.016 | 0.196 | 0.0508 | 0.8329 |
| Comp5 | 0.821 | 0.351 | 0.0410 | 0.8739 |
| Comp6 | 0.469 | 0.067 | 0.0235 | 0.8974 |
| Comp7 | 0.403 | 0.089 | 0.0201 | 0.9175 |
| etc. | … | … | … | … |
| Comp19 | 0.024 | 0.001 | 0.0012 | 0.9989 |
| Comp20 | 0.023 | . | 0.0011 | 1.0000 |
Table 4 shows how the top components correlate with each original variable. The first component correlates positively and relatively evenly across most of the formalized financial tools including both traditional tools (e.g., savings, loans, cards, and other services provided by financial institutions) and using the internet, and it correlates negatively with informal and distress finance, such as relying on family and friends and borrowing for health purposes or selling assets for emergency funds. Thus, we interpret the first component as an index of the extent to which a nation exhibits “Broad access to and use of formal financial tools.” The higher the index, the more widespread and diverse are the formal financial tools in regular use in that country.
**Table 4**
| Variable | Comp1 | Comp2 | Comp3 | Comp4 |
| --- | --- | --- | --- | --- |
| Finance institution account | 0.259 | -0.03 | 0.197 | -0.027 |
| Borrowed from a financial institution or credit card | 0.261 | 0.009 | -0.012 | 0.155 |
| Saved at a financial institution | 0.268 | 0.058 | -0.137 | -0.051 |
| Debit card ownership | 0.264 | -0.008 | 0.153 | -0.056 |
| Credit card ownership | 0.263 | -0.017 | -0.069 | 0.098 |
| Received wages into financial institution | 0.267 | -0.012 | 0.13 | -0.011 |
| Paid utility bills using financial institution account | 0.267 | 0.012 | 0.006 | 0.053 |
| Outstanding housing loan | 0.248 | 0.08 | 0.03 | 0.032 |
| Used the internet for online transaction | 0.272 | 0.066 | 0.032 | -0.021 |
| Received wages through mobile phone | 0.061 | 0.535 | 0.232 | -0.259 |
| Paid utility bills through mobile phone | 0.179 | 0.427 | 0.081 | -0.125 |
| Made or received digital payments | 0.264 | 0.086 | 0.17 | -0.035 |
| Saved for old age | 0.255 | 0.002 | -0.19 | -0.03 |
| Borrowed for health or medical purposes | -0.193 | 0.312 | 0.003 | 0.023 |
| Coming up with emergency funds: possible | 0.194 | -0.234 | 0.014 | -0.167 |
| Main source of emergency funds: savings | 0.246 | -0.02 | -0.275 | -0.142 |
| Main source of emergency funds: formal loan | 0.077 | 0.173 | 0.269 | 0.85 |
| Borrowed from family or friends | -0.163 | 0.368 | 0.251 | -0.237 |
| Main source of emergency funds: family/friends | -0.134 | -0.247 | 0.62 | -0.055 |
| Main source of emergency funds: sale of assets | -0.138 | 0.361 | -0.419 | 0.202 |
By contrast, the second index associates most closely with metrics related to alternative financial tools, such as mobile phone transactions and informal finance of borrowing from family and friends, and distress financing tools, such as selling assets for emergency uses and borrowing for health and medical needs. It is also negatively associated with the ability to raise emergency funds and access to family funding for emergencies, suggesting that high values of the second index associate, in part, with financial resilience challenges. We therefore interpret the second component as an index capturing a nation’s tendency toward “Reliance on alternative, informal, and distress financial tools.” Descriptive statistics on two financial tools indexes created from these two principal components are shown in S1 Table. The influence of these two index variables on COVID-19 mortality is our main interest in the econometric models that follow.
The correlation matrix in S3 Table shows these two financial tools access indexes’ near zero correlation (r = -0.01) with each other, but high correlations with several economic, demographic, and healthcare infrastructure control variables. Chronbach’s alpha tests of internal consistency on the covariance of these two indexes with each other (alpha = 0.02), with national income (0.52), with the two healthcare infrastructure variables (0.39), and among all five of these highly correlated variables together confirms (all alphas <0.6) they capture significantly different characteristics of national ecosystems. Moreover, despite the potential for noisy estimators from multicollinearity, our econometric modeling nevertheless finds statistically significant evidence for these financial tools access indexes—particularly robust for the first.
## Ethics statement
All data used in the study is publicly available secondary data, not collected by the authors. No permissions for data use were required. Protocols for data collection, ethics, and participants’ rights protection vary by source organization, as listed in Table 1, all of which are widely known and internationally reputable. The main variables of interest, specifically the financial tools access indexes, were constructed from the World Bank’s Global Findex, which is made up from national surveys done by Gallup as part of the Gallup World Poll [47]. Gallup World Poll uses nationally representative stratified random sampling and either face-to-face or random-digit-dialing telephone interviewing of residents aged 15 and older. Gender-matched interviewing occurs where cultural norms dictate. In select nations, Gallup excludes some regions for safety or unapproachability reasons.
## Models and results
As the slight differences among variable observation counts in S1 Table indicate, ten nations lack values for one or several of the control variables, but the specific missing variables differed among nations. Because those ten nations did have data for most of our other variables, rather than lose them from the analysis, we opted for a linear model using full information maximum likelihood (FIML) estimation (following [48], which shows FIML estimates are similar to those from established missing variable imputation methods, e.g., [49–52]). Because COVID-19 patterns correlate geographically, we include World Bank region dummies, and robust standard errors are adjusted clustered by region. Our main results are shown in Table 5 (repeated as Model 1 in S4 Table), followed by a comparison OLS model in Table 6 without those ten nations that lacked full data (repeated as Model 2 in S4 Table), again with robust standard errors adjusted clustered by region, as a robustness check. Either model explains more than $72\%$ of the variation in COVID-19 mortality rates across the 142 nations.
The first financial access index, capturing access to formal financial services, is statistically significantly related to COVID-19 mortality rates ($p \leq .001$), controlling for demographic, socioeconomic, population health and health infrastructure variables, and region. In Table 5, the -0.299 coefficient suggests that (pre-omicron) COVID-19 mortality rises $29.9\%$ for a one-unit reduction in the (pre-COVID-19) access to formal finance index. The standard deviation of the index is 3.521, implying that, on average, nations with one standard deviation lower access to formal financial services had COVID-19 mortality rates more than double (0.299*3.521 = $105\%$) nations with average financial tools access, conditional on the other COVID-19 mortality covariates. The alternative coefficient estimate (0.316) from Table 6 is also statistically significant ($$p \leq .008$$) and very similar in scale, again suggesting more than doubling the mortality rate ($111\%$).
As additional validity checks on the stability of estimates of our main financial access index variables of interest, S4 Table expands the number of alternative model specifications, showing models with only various subsets of the independent variables. Models 1 and 2 repeat Tables 5 and 6. Model 3 removes the region dummies. Model 4 uses only the two financial tools indexes. Because financial system strength is significantly correlated with national income, Model 5 adds only a single other variable, (ln) per capita income, to the two financial indexes. Model 6 adds only the other demographic and socioeconomic variables. Model 7 includes only the health infrastructure variables with the financial indexes. Model 8 adds population health variables. Model 9 aims for parsimony by removing from Model 2, post OLS, any variable with $p \leq 0.4.$ ( Not shown, but the variables included in Model 9 are the same as those identified by a different technique, LASSO regression, except the LASSO-suggested model includes population density, which when added has essentially no substantive effect on other significant coefficients.) Finally, Model 10 includes the same variables as Model 9 but employs the FIML estimation method (as used in Model 1) to enable including the ten nations missing one or more specific variables.
To test the internal validity of model specification, a Ramsey RESET test and a LINK test are conducted after each model (except the FIML models) to test functional form and omitted variable problems. Wald tests on variables added beyond the two variables in Model 5 (financial indexes and income only) are also shown. The Ramsey, Link, and Wald test results appear in the bottom rows of S4 Table. The only statistically significant evidence of problems arises in the Models (4, 5, and 7) that omit all or most control variables, supporting the addition of variables such as those included our preferred model. The Wald tests suggest each added set of variables, added sequentially to other sets, is jointly significant. A Wald test also confirms the variables removed from Model 2 for the parsimony Model 9 (post-OLS set of insignificant ($p \leq 0.4$)) are not collectively significant.
S4 Table shows that the estimated size of the relationship between COVID-19 mortality rate—i.e., an approximate doubling per standard deviation reduction in the index of access to formal financial services—is robust to alternative sets of control variables, except in Model 4, which only includes the two financial tools indexes. Compared to Model 4, the addition in Model 5 of a single additional covariate that is significantly positively correlated with the depth of financial systems, per capita income, switches the sign of the coefficient on the first financial index. A similar effect can be seen in Model 7, which adds only the two health infrastructure variables, both also strongly positively correlated with national wealth. In other words, access to formal financial services becomes significantly negatively associated with COVID-19 mortality once any variables also related to national wealth are controlled for.
By contrast, the results for the second financial services index, reliance on alternative, informal, and distress financing, are weaker in scale and statistically less robust. In Table 5, a nation’s pre-COVID-19 tendency to rely on alternative, informal, and distress financial tools is positively associated with higher COVID-19 mortality and statistically significant ($p \leq .001$). According to this model, mitigating COVID-19 impacts apparently was more challenging where financial emergency coping mechanisms rely on informal options and distress financing—an intuitively appealing result. However, compared to the formal finance index, the strength of the association is weaker. A one standard deviation increase in the alternative, informal, and distress finance index is associated with a $15.9\%$ (1.391*.114) increase in mortality. In the OLS version, Table 6, which includes data from fewer nations, the second index, though still positively related to COVID-19 mortality, falls to statistical insignificance ($$p \leq 0.186$$). Moreover, as S4 Table shows, coefficient estimates and significance for this second index are sensitive to which control variables are included.
While the control variables—drawn from the medical and public health literatures—are not our central interest, we now turn to those population health, demographic, and socioeconomic variables. Youthful populations are associated with statistically significantly lower mortality, as are more-equal income distributions. Consistent with existing literature (e.g., [53–56]), nations with higher degrees of income inequality have statistically significantly higher COVID-19 mortality rates. A one-unit increase in national Gini coefficient raises expected COVID-19 mortality rate by about $5\%$–$6\%$, depending on the model. Nations with income inequality one standard deviation (7.79) above the average level would have roughly $40\%$–$50\%$ higher mortality rates.
Given that COVID-19 attacks the respiratory-system [35, 36, 43], it is unsurprising that population respiratory health appears important too. Higher national pre-COVID-19 lung cancer prevalence is a statistically significant COVID-19 mortality risk factor in models with regional dummies. As shown in Fig 1B, a one-unit increase (approximately the standard deviation, 1.08) from the mean ln(lung cancer prevalence) relates to a two-thirds greater COVID-19 mortality, conditional on the other independent variables. So too, higher pre-COVID-19 childhood tuberculosis vaccination rates are negatively correlated with COVID-19 mortality (though the latter is not statistically significant in Models 1 and 8).
**Fig 1:** *Comparing conditional partial correlation of COVID-19 mortality with financial access, lung cancer, and income inequality.(A) Conditional partial correlation of national COVID-19 mortality with pre-COVID-19 Index of access to formal finance; 95% confidence interval shaded. (B) Similar, with pre-COVID-19 ln(average lung cancer incidence rate, age-standardized per 100,000, male & female, all ages). (C) Similar, with pre-COVID Gini index of income inequality.*
We also find that nations with higher per capita incomes had statistically significantly higher mortality rates, even after controlling for characteristics associated with wealth, like aging and obesity, that might be problematic for richer nations. Though curious, this finding is consistent with most of the COVID-19 literature. For example, Goldberg and Reed [28] found, after controlling for several demographic, public health, policy response, weather, and mobility variables, that for a $1\%$ increase in per capita GDP, COVID-19 mortality rose by $0.9\%$. Our results suggest a smaller relation, roughly half that, but still positive. Reasons for this counterintuitive yet widely duplicated result are not well understood and have been discussed at length elsewhere (e.g., [28, 37–39, 56]).
Though not tested or of central interest here, numerous speculations—with often conflicting empirical results—exist in that literature about why wealthier nations, which also tend to be those with deeper financial services, have had substantially higher average levels of COVID-19 infection and mortality. Possible explanations include, for example: measurement issues such as less-robust public health data reporting resulting in undercounting by lower-income nations, or political and cultural differences in reporting transparency; political economic differences in policy responses; socio-economic differences like greater population mobility in wealthier nations either before or after lockdowns, very high population densities in the very largest urban centers in wealthier nations, previous experience in lower-income nations with handling and/or accumulated immunities from similar viral (e.g., SARS) outbreaks. or more-concentrated systems of elder care; and even seasonal differences in northern and southern hemispheres; or a host of other potential reasons. Clarifying what drives the counterintuitive wealth-COVID-19 correlation has so far been frustratingly difficult and remains a work in process and beyond the scope of this article.
Similarly counterintuitive, nations with more effective healthcare systems, as measured by the universal healthcare effective coverage index, have significantly higher COVID-19 mortality. The size of the relationship in our models is similar to that of lung cancer relation: a one standard deviation increase in the UHC index goes with an approximate two-thirds increase in mortality. Greater population shares of nurses and midwifery personnel are also positively related to mortality, however not statistically significantly in most models. The coefficients on diabetes hint at another puzzle. In models with regional dummies, diabetes prevalence associates negatively with COVID-19 mortality—not the direction we would expect based on the medical literature [41, 42]. However, the statistical significance is marginal ($p \leq .05$) except in Model 9. Indoor air pollution, body mass index, and raised blood pressure prevalence are not statistically related to mortality in most models.
An attempt to explore the effective healthcare system puzzle a bit, specifically whether the interaction between deeper financial systems and stronger healthcare systems might help untangle it, is shown in S1 and S2 Figs. The figures show the results of an OLS (robust, region clustered) model like Model 2 but including an interaction effect between the two (continuous) indexes of the formal financial tools access and health services effective coverage. Because of the interaction term, the marginal effects cannot be interpreted directly from the coefficients (not shown) of the regression model. Rather, the figures show the average marginal effect on the predicted (ln) COVID-19 mortality rate of each of the two variables at various levels of the second variable, together with $95\%$ confidence intervals. As S1 Fig shows, the marginal effect of the formal financial tools index suggests higher financial access reduces mortality, and increasingly so at higher levels of health care effectiveness; the predicted effects are generally statistically significantly, except at the lowest levels of the heath system index. S2 Fig then shows the average marginal effects of the health services effectiveness index at different levels of the financial access index. Unfortunately, the healthcare system puzzle remains. At low levels of financial access, higher healthcare system effectiveness is still statistically significantly positively associated with higher COVID mortality. The effect remains positive, though not significant, at higher levels of financial access. In short, on average in nations with the most effective healthcare systems, COVID-19 related mortality is higher and the mitigating effect of better access to financial tools is larger.
Finally, although the independent variables included in Models 1–10 reflect national conditions in 2019 or earlier, prior to the arrival of COVID-19, in S5 Table we do explore adding measures from June 2021 on COVID-19 vaccination rates and availability, which are clearly important for mitigating COVID mortality risk. Model 11 adds COVID-19 vaccination rate as a share of the total population by June, 30, 2021 [57]. Model 12 adds COVID-19 vaccine availability, measured as full doses acquired by the nation as of June 25, 2021 divided by total population [58]. Note the vaccine availability data in Model 12 is only available for 77 of the 142 nations. In all models except the limited sample Model 12, COVID-19 mortality statistically significantly declines with higher vaccination rates. Vaccine availability itself is not statistically related to mortality. In any event, the main conclusions on the financial tools access index variables are unaffected in any of the models including COVID-19 vaccine data. Neither does inclusion of vaccine data help clear up the puzzles of the signs of the GDP and health system effectiveness variables.
## Discussion and conclusions
In summary, we find that greater pre-pandemic national levels of use and access to formal financial services are related to substantially lower death rates from (pre-omicron) COVID-19. The result suggests that financial services deserve substantially greater attention both in the public health literature related to COVID-19 and more broadly in policy discussions about fostering better public health overall. Robust financial services like savings, insurance, credit, payment systems, and the like are clearly potentially useful tools for households paying for food, housing, medicines, and health services and dealing with medical emergencies. Yet despite the extensive literature linking household financial security and health and the obvious potential link to COVID-19 outcomes, we are unaware of any other study exploring how COVID-19 mortality relates to financial systems.
To help assess the relative importance of the financial access variable, Fig 1A, based on the model in Table 5, shows the partial correlation of COVID-19 mortality with pre-COVID-19 levels of access to formal financial services conditional on the other independent variables in Table 5, with $95\%$ confidence interval. The strength of the association in Fig 1A of COVID-19 mortality with pre-COVID-19 levels of access to formal financial services is similar in scale but opposite in direction to the estimated relations with pre-COVID-19 lung cancer rates (Fig 1B) and with income inequality (Fig 1C)—both of which are among the most important determinants of COVID-19 mortality in the medical and public health literatures. Many dozens (perhaps hundreds) of pandemic-related studies have now looked at each of those variables; financial services have been essentially ignored. Considering that the scale of the relationship to population-level COVID-19 mortality rivals that of lung cancer, this appears to be a major exploratory blind spot.
Though the GDP-COVID-19 puzzle is not our central interest, our finding that deeper formal financial systems appear associated with better risk mitigation if anything deepens the perplexity. It is puzzling too that not only the income variable but also both health infrastructure variables relate positively to COVID-19 mortality. All three behave oppositely to our formal financial access metric, despite the clear relationship between national wealth and the depths of both financial systems and health systems. Some analysts have speculated the positive association of national income and COVID-19 might be an artifact of nations with deeper healthcare systems better tracking health statistics [39]. Yet controlling for the healthcare infrastructure here does not solve the GDP riddle. Regardless, the jury is still out on why rich countries have suffered more. It is remarkably counterintuitive that that pattern is replicated in both income and healthcare infrastructure.
Untangling the complex socio-economic and socio-structural determinants of health inequalities remains a major challenge. The role of financial tools in health should feature centrally in that inquiry.
## References
1. Weida EB, Phojanakong P, Patel F, Chilton M. **Financial health as a measurable social determinant of health**. *PLOS ONE* (2020.0) **15** e0233359. DOI: 10.1371/journal.pone.0233359
2. Pickett KE, Wilkinson RG. **Income inequality and health: a causal review**. *Social Science & Medicine.* (2015.0) **128** 316-326. DOI: 10.1016/j.socscimed.2014.12.031
3. Wachtler B, Michalski N, Nowossadeck E, Diercke M, Wahrendorf M, Santos-Hövener C. **Socioeconomic inequalities and COVID-19—a review of the current international literature**. *Journal of Health Monitoring* (2020.0) **5** 3-17. DOI: 10.25646/7059
4. **COVID-19 and the social determinants of health and health equity: evidence brief**. *World Health Organization* (2021.0)
5. Rutherford S.. *The poor and their money* (2000.0)
6. Yunus M.. *Banker to the poor: Micro-lending and the battle against world poverty:* (1999.0)
7. Robinson MS. *The microfinance revolution: Sustainable finance for the poor: Washington* (2001.0)
8. Xu K, David B. **Evans, Guido Carrin, Ana Mylena Aguilar-Rivera, Philip Musgrove, Timothy Evans**. *Protecting households from catastrophic health spending: Health Affairs* (2007.0) **26** 972-983. DOI: 10.1377/hlthaff.26.4.972
9. Adda J, Banks J, Von Gaudecker H. **The impact of income shocks on health: evidence from cohort data**. *Journal of the European Economic Association* (2009.0) **7** 1361-1399. DOI: 10.1162/JEEA.2009.7.6.1361
10. Orton L, Pennington A, Nayak S, Sowden A, White M, Whitehead M. **Group-based microfinance for collective empowerment: a systematic review of health impacts**. *Bull World Health Organ* (2016.0) **94** 694-704A. DOI: 10.2471/BLT.15.168252
11. Watkins TA. *Introduction to Microfinance* (2018.0)
12. Zheng C, Zhang J. **The impact of COVID-19 on the efficiency of microfinance institutions**. *International Review of Economics & Finance* (2021.0) **71** 407-423. DOI: 10.1016/j.iref.2020.09.016
13. Elnahass M, Trinh VQ, Li T. **Global banking stability in the shadow of COVID-19 outbreak**. *Journal of International Financial Markets, Institutions and Money* (2021.0) **72** 101322. DOI: 10.1016/j.intfin.2021.101322
14. Katusiime L.. **COVID-19 and bank profitability in low income countries: the case of Uganda**. *Journal of Risk and Financial Management* (2021.0) 14. DOI: 10.3390/jrfm14120588
15. Brown R, Rocha A, Cowling M. **Financing entrepreneurship in times of crisis: exploring the impact of COVID-19 on the market for entrepreneurial finance in the United Kingdom**. *International Small Business Journal* (2020.0) **38** 380-390. DOI: 10.1177/0266242620937464
16. Alekseev G, Amer S, Gopal M, Kuchler T, Schneider JW, Stroebel J. **The effects of COVID-19 on U.S. small businesses: evidence from owners, managers, and employees**. *NBER Working Paper Series* (2020.0). DOI: 10.3386/w27833
17. Cowling M, Brown R, Rocha A. **Did you save some cash for a rainy COVID-19 day?**. *The crisis and SMEs. International Small Business Journal* (2020.0) **38** 593-604. DOI: 10.1177/0266242620945102
18. Stevenson C, Wakefield JRH. **Financial distress and suicidal behaviour during COVID-19: family identification attenuates the negative relationship between COVID-related financial distress and mental Ill-health**. *J Health Psychol.* (2021.0) **26** 2665-2675. DOI: 10.1177/13591053211014597
19. Görtz C, McGowan D, Yeromonahos M. *Furlough and Household Financial Distress During the COVID-19 Pandemic* (2021.0). DOI: 10.2139/ssrn.3914975
20. Roll S, Despard M. *Income Loss and Financial Distress during COVID-19: The Protective Role of Liquid Assets* (2020.0). DOI: 10.2139/ssrn.3733862
21. Smith JP. **Healthy bodies and thick wallets: the dual relation between health and economic status**. *Journal of Economic Perspectives* (1999.0) **13** 145-166. PMID: 15179962
22. Dhankhar A, Kumari R, Bahurupi Y. **Out-of-pocket, catastrophic health expenditure and distress financing on non-communicable diseases in India: A systematic review with meta-analysis**. *Asian Pacific Journal of Cancer Prevention* (2021.0) **22** 671-680. DOI: 10.31557/APJCP.2021.22.3.671
23. Ir P, Jacobs B, Asante AD, Liverani M, Jan S, Chhim S. **Exploring the determinants of distress health financing in Cambodia**. *Health Policy and Planning* (2019.0) **34** i26-i37. DOI: 10.1093/heapol/czz006
24. Pana TA, Bhattacharya S, Gamble DT, Pasdar Z, Szlachetka WA, Perdomo-Lampignano JA. **Country-level determinants of the severity of the first global wave of the COVID-19 pandemic: an ecological study**. *BMJ Open* (2021.0) **11** e042034. DOI: 10.1136/bmjopen-2020-042034
25. Okeahalam C, Williams V, Otwombe K. **Factors associated with COVID-19 infections and mortality in Africa: a cross-sectional study using publicly available data**. *BMJ Open* (2020.0) **10** e042750. DOI: 10.1136/bmjopen-2020-042750
26. James A, Dalal J, Kousi T, Vivacqua D, Portela Câmara DC, dos Reis IC. **An in-depth statistical analysis of the COVID-19 pandemic’s initial spread in the WHO African region**. *BMJ Global Health* (2021.0) **7** e007295. DOI: 10.1101/2021.08.21.21262401
27. Asem N, Ramadan A, Hassany M, Ghazy RM, Abdallah M, Ibrahim M. **Pattern and determinants of COVID-19 infection and mortality across countries: an ecological study**. *Heliyon* (2021.0) **7** e07504. DOI: 10.1016/j.heliyon.2021.e07504
28. Goldberg PK, Reed T. **The effects of the coronavirus pandemic in emerging market and developing economies: an optimistic preliminary account**. *Brookings Papers on Economic Activity* (2020.0) **2020** 161-235
29. Dowd JB, Andriano L, Brazel DM, Rotondi V, Block P, Ding X. **Demographic science aids in understanding the spread and fatality rates of COVID-19**. *Proceedings of the National Academy of Sciences* (2020.0) **117** 9696-9698. DOI: 10.1073/pnas.2004911117
30. Escobar LE, Molina-Cruz A, Barillas-Mury C. **BCG vaccine protection from severe coronavirus disease 2019 (COVID-19)**. *Proceedings of the National Academy of Sciences* (2020.0) **117** 17720-17726. DOI: 10.1073/pnas.2008410117
31. Flint SW, Tahrani AA. **COVID-19 and obesity—lack of clarity, guidance, and implications for care**. *The Lancet Diabetes & Endocrinology* (2020.0) **8** 474-475. DOI: 10.1016/S2213-8587(20)30156-X
32. Guan W, Liang W, Zhao Y, Liang H, Chen Z, Li Y. **Comorbidity and its impact on 1590 patients with COVID-19 in China: a nationwide analysis**. *European Respiratory Journal* (2020.0) 55. DOI: 10.1183/13993003.00547–2020
33. Ji Y, Ma Z, Peppelenbosch MP, Pan Q. **Potential association between COVID-19 mortality and health-care resource availability**. *The Lancet Global Health* (2020.0) **8** e480. DOI: 10.1016/S2214-109X(20)30068-1
34. Vardavas CI, Nikitara K. **COVID-19 and smoking: a systematic review of the evidence**. *Tobacco Induced Diseases.* (2020.0) **18**. DOI: 10.18332/tid/119324
35. Ahmad S, Manzoor S, Siddiqui S, Mariappan N, Zafar I, Ahmad A. **Epigenetic underpinnings of inflammation: connecting the dots between pulmonary diseases, lung cancer and COVID-19**. *Seminars in Cancer Biology* (2021.0) **83** 384-398. DOI: 10.1016/j.semcancer.2021.01.003
36. Calabrò L, Rossi G, Covre A, Morra A, Maio M. **COVID and lung cancer**. *Curr Oncol Rep* (2021.0) **23** 1-10. DOI: 10.1007/s11912-021-01125-8
37. Chaudhry R, Dranitsaris G, Mubashir T, Bartoszko J, Riazi S. **A country level analysis measuring the impact of government actions, country preparedness and socioeconomic factors on COVID-19 mortality and related health outcomes**. *EClinicalMedicine.* (2020.0) **25** 100464. DOI: 10.1016/j.eclinm.2020.100464
38. Ferreira FH, Sterck O, Mahler D, Decerf B. **Death and destitution: the global distribution of welfare losses from the COVID-19 pandemic**. (2021.0) 1-11. DOI: 10.31389/lseppr.34
39. Valero M, Valero-Gil JN. **Determinants of the number of deaths from COVID-19: differences between low-income and high-income countries in the initial stages of the pandemic**. *International Journal of Social Economics* (2021.0). DOI: 10.1108/IJSE-11-2020-0752
40. Heuveline P, Tzen M. **Beyond deaths per capita: comparative COVID-19 mortality indicators**. *BMJ Open* (2021.0) **11** e042934. DOI: 10.1136/bmjopen-2020-042934
41. Abdi A, Jalilian M, Sarbarzeh PA, Vlaisavljevic Z. **Diabetes and COVID-19: a systematic review on the current evidences**. *Diabetes Res Clin Pract* (2020.0) **166** 108347. DOI: 10.1016/j.diabres.2020.108347
42. Singh AK, Gupta R, Ghosh A, Misra A. **Diabetes in COVID-19: Prevalence, pathophysiology, prognosis and practical considerations. Diabetes & Metabolic Syndrome: Clinical Research &**. *Reviews* (2020.0) **14** 303-310. DOI: 10.1016/j.dsx.2020.04.004
43. Miller A, Reandelar MJ, Fasciglione K, Roumenova V, Li Y, Otazu GH. **Correlation between universal BCG vaccination policy and reduced mortality for COVID-19**. *MedRxiv.* (2020.0). DOI: 10.1101/2020.03.24.20042937
44. Dong E, Du H, Gardner L.. **An interactive web-based dashboard to track COVID-19 in real time**. *The Lancet Infectious Diseases* (2021.0) **20** 533-534. DOI: 10.1016/S1473-3099(20)30120-1
45. Ferlay J, Ervik M, Lam F, Colombet M, Mery L, Piñeros M. **Global Cancer Observatory: Cancer today**. *International Agency for Research on Cancer* (2020.0)
46. Lozano R, Fullman N, Mumford JE, Knight M., Barthelemy CM, Abbafati C. **Measuring universal health coverage based on an index of effective coverage of health services in 204 countries and territories, 1990–2019: A systematic analysis for the Global Burden of Disease Study 2019**. *The Lancet* (2020.0) **396** 1250-1284. DOI: 10.1016/S0140-6736(20)30750-9
47. Demirguc-Kunt A, Klapper L, Singer D, Ansar S.. *The Global Findex database 2017: Measuring financial inclusion and the fintech revolution*
48. Medeiros R.. **Handling missing data in Stata: Imputation and likelihood-based approaches**. *Presented at Swiss Stata Users Group Meeting* (2016.0)
49. Lee KJ, Carlin JB. **Multiple imputation for missing data: fully conditional specification versus multivariate normal imputation**. *Am J Epidemiol* (2010.0) **171** 624-632. DOI: 10.1093/aje/kwp425
50. Little RJ, Rubin DB. **Statistical analysis with missing data**. *Hoboken, NJ: Wiley* (2002.0)
51. Rubin DB. **Multiple imputation after 18 years**. *Journal of the American Statistical Association* (1996.0) **91** 473-489. DOI: 10.1080/01621459.1996.10476908
52. Schafer JL. **Analysis of incomplete multivariate data**. *Boca Raton: Chapman & Hall/CRC* (1997.0)
53. Elgar FJ, Stefaniak A, Wohl MJA. **The trouble with trust: time-series analysis of social capital, income inequality, and COVID-19 deaths in 84 countries**. *Soc Sci Med* (2020.0) **263** 113365. DOI: 10.1016/j.socscimed.2020.113365
54. Tan AX, Hinman JA, Magid HSA, Nelson LM, Odden MC. **Association between income inequality and county-level COVID-19 cases and deaths in the US**. *JAMA Network Open* (2021.0) **4** e218799. DOI: 10.1001/jamanetworkopen.2021.8799
55. Wildman J.. **COVID-19 and income inequality in OECD countries**. *The European Journal of Health Economics* (2021.0) **22** 455-462. DOI: 10.1007/s10198-021-01266-4
56. Deaton A.. **COVID-19 and global income inequality**. *LSE Public Policy Review* (2021.0) **1** 1. DOI: 10.31389/lseppr.26
57. Mathieu E., Ritchie H., Ortiz-Ospina E.. **A global database of COVID-19 vaccinations**. *Nat Hum Behav* (2021.0) **5** 947-953. DOI: 10.1038/s41562-021-01122-8
58. **Duke Global Health Innovation Center**. *Launch and scale speedometer. Duke University* (2020.0)
|
---
title: 'The impact of hyperglycemia upon BeWo trophoblast cell metabolic function:
A multi-OMICS and functional metabolic analysis'
authors:
- Zachary J. W. Easton
- Xian Luo
- Liang Li
- Timothy R. H. Regnault
journal: PLOS ONE
year: 2023
pmcid: PMC10022812
doi: 10.1371/journal.pone.0283118
license: CC BY 4.0
---
# The impact of hyperglycemia upon BeWo trophoblast cell metabolic function: A multi-OMICS and functional metabolic analysis
## Abstract
Pre-existing and gestationally-developed diabetes mellitus have been linked with impairments in placental villous trophoblast cell metabolic function, that are thought to underlie the development of metabolic diseases early in the lives of the exposed offspring. Previous research using placental cell lines and ex vivo trophoblast preparations have highlighted hyperglycemia is an important independent regulator of placental function. However, it is poorly understood if hyperglycemia directly influences aspects of placental metabolic function, including nutrient storage and mitochondrial respiration, that are altered in term diabetic placentae. The current study examined metabolic and mitochondrial function as well as nutrient storage in both undifferentiated cytotrophoblast and differentiated syncytiotrophoblast BeWo cells cultured under hyperglycemia conditions (25 mM glucose) for 72 hours to further characterize the direct impacts of placental hyperglycemic exposure. Hyperglycemic-exposed BeWo trophoblasts displayed increased glycogen and triglyceride nutrient stores, but real-time functional readouts of metabolic enzyme activity and mitochondrial respiratory activity were not altered. However, specific investigation into mitochondrial dynamics highlighted increased expression of markers associated with mitochondrial fission that could indicate high glucose-exposed trophoblasts are transitioning towards mitochondrial dysfunction. To further characterize the impacts of independent hyperglycemia, the current study subsequently utilized a multi-omics approach and evaluated the transcriptomic and metabolomic signatures of BeWo cytotrophoblasts. BeWo cytotrophoblasts exposed to hyperglycemia displayed increased mRNA expression of ACSL1, HSD11B2, RPS6KA5, and LAP3 and reduced mRNA expression of CYP2F1, and HK2, concomitant with increased levels of: lactate, malonate, and riboflavin metabolites. These changes highlighted important underlying alterations to glucose, glutathione, fatty acid, and glucocorticoid metabolism in BeWo trophoblasts exposed to hyperglycemia. Overall, these results demonstrate that hyperglycemia is an important independent regulator of key areas of placental metabolism, nutrient storage, and mitochondrial function, and these data continue to expand our knowledge on mechanisms governing the development of placental dysfunction.
## Introduction
The rates of diabetes mellitus (DM) during pregnancy have increased substantially over the past several decades [1]. It is currently estimated that up to 1 in 10 pregnancies worldwide are impacted by maternal DM [2,3], however these rates may be even higher in certain at risk demographics including Indigenous populations [4]. These increases are particularly concerning as maternal DM during pregnancy, regardless of whether it is pre-existing (such as in type 1 DM or type 2 DM) or develops during gestation (gestational DM (GDM)), is associated with poor fetal health outcomes [5–8]. Specifically, children exposed to DM during intrauterine development have been found to be at a greater risk of developing non-communicable diseases such as obesity, metabolic syndrome, and impaired insulin sensitivity early in their lives [5,9–12]. Understanding the underlying mechanisms that link maternal DM during pregnancy to the development of non-communicable diseases in offspring is critical to develop appropriate prenatal clinical management practices that help reduce health risks to the next generations.
As the placenta is responsible for nutrient, gas and waste exchange between mother and fetus, specific functional alterations in this organ may be important in facilitating the intrauterine programming of metabolic disorders in DM-exposed offspring. Unsurprisingly, morphological and functional abnormalities have been found to be highly prevalent in placentae of diabetic pregnancies [13,14]. For example, diabetic placentae are often heavier [15–17], and display increased glycogen and triglyceride content [18–23], that is suggestive of altered nutrient storage and processing by the placenta and subsequently altered nutrient delivery to the developing fetus. This increase in nutrient storage in DM placentae has been thought to modulate trans-placental nutrient transport and fetal growth trajectories [24,25]. Additionally, the progenitor cytotrophoblasts (CT) and differentiated syncytiotrophoblasts (SCT) cells of the placenta villous trophoblast layer (cells that form the materno-fetal exchange barrier and are a primary site for placental energy (ATP) production) have been found to have impaired mitochondrial function in response to maternal DM that may further impact placental nutrient handling [26]. In particular, cultured primary CT and SCT cells from GDM pregnancies have been found to have reduced basal and maximal mitochondrial respiratory (oxidative) activity compared to non-diabetic control trophoblasts [27,28]. Additionally, pre-existing DM has been found to impact the activities of individual placental Electron Transport Chain (ETC) complexes in whole placental lysates, highlighted by reduced complex I, II and III activity in type 1 DM placentae, and reduced complex II and III activity in type 2 DM placentae [29]. Overall, these studies suggest that impaired placental nutrient storage and mitochondrial oxidative function may be implicated in the development of metabolic diseases in DM-exposed offspring.
Previous work with placental cell lines and ex vivo placental explant preparations have demonstrated that hyperglycemia (a hallmark symptom of both pre-existing and gestationally-developed DM) is an important regulator of placental metabolic function. For example, explants from uncomplicated pregnancies were found to have altered lipid processing when cultured under hyperglycemic (HG) conditions (25 mM glucose) for 18 hours [22]. Further reports have highlighted transcriptomic and metabolomic markers indicative of altered lipid metabolism, β-oxidation, and glycolysis functions in undifferentiated BeWo CT cells cultured under HG-conditions (25 mM) for 48 hours [30]. Independent hyperglycemia (30 mM glucose for 72h) has also been linked to increased Reactive Oxygen Species (ROS) generation in undifferentiated BeWo CTs [31], which may directly promote the development mitochondrial oxidative damage [32,33]. Overall, these reports have suggested that hyperglycemia may independently facilitate the development of aberrant placental metabolic function in diabetic pregnancies. However, the direct and independent impacts of hyperglycemia on placental mitochondrial respiratory (oxidative) function are poorly understood. Previously, elevated glucose levels (25 mM for 48h) have also been associated with altered mitochondrial activity in BeWo CT cells when assessed by endpoint tetrazolium salt (MTT) assay [34]. However, an interrogation of mitochondrial respiratory activity of HG-exposed trophoblast cells using recently developed real-time functional readouts (such as the Seahorse XF Analyzer), as has been performed with DM-exposed Primary Human Trophoblasts (PHT), [27–29] is warranted. In addition, the direct impacts of hyperglycemia on glycogen and lipid nutrient stores of placental trophoblasts and the underlying mechanisms governing placental nutrient storage in HG-conditions remains ill defined.
The first objective of the current study was to characterize the impacts of independent hyperglycemia on placental mitochondrial respiratory activity and nutrient storage by evaluating both undifferentiated BeWo CTs and differentiated BeWo SCTs following a relatively prolonged 72-hour HG (25 mM) exposure. Recently, the integration of transcriptomics with metabolomics has been identified as a useful method to elucidate cellular mechanisms that underlie pathological placental development in pre-clinical models [30,35,36]. In turn, the second objective of this study was to utilize a multi-omics research approach to thoroughly characterize potential mechanisms leading to altered metabolic function in high-glucose exposed BeWo progenitor CT cells. It was postulated that HG-culture conditions would be associated with increased nutrient storage and impaired mitochondrial respiratory function in BeWo CT and SCT cells, in association with altered transcriptome and metabolome signatures in BeWo CT cells indicative of altered metabolic function.
## Materials
All materials were purchased from Millipore Sigma (Oakville, Canada) unless otherwise specified.
## Cell culture conditions
BeWo (CCL-98) trophoblast cells were purchased from the American Type Culture Collection (ATCC; Cedarlane Labs, Burlington, Canada). Cells were cultured in F12K media (Gibco, ThermoFisher Scientific, Mississauga, Canada) as recommended by the ATCC, and supplemented with $10\%$ Fetal Bovine Serum (Gibco) and $1\%$ Penicillin-Streptomycin (Invitrogen, ThermoFisher Scientific, Mississauga, Canada). All cells were utilized between passages 5–15 and were maintained at 37°C and $5\%$ CO2/$95\%$ atmospheric air.
The F12K media contained 7 mM of glucose, a relatively physiological glucose level, and was utilized for low-glucose (LG) controls. F12K media was supplemented to 25 mM glucose for hyperglycemic (HG) culture treatments as previously utilized with BeWo trophoblasts [30,34]. BeWo trophoblasts cells were plated in LG F12K media at the specifically stated experimental densities and allowed to adhere to cell culture plates overnight before being treated with HG culture media. Cell media was replenished every 24 hours. At T24h and T48h subsets of BeWo trophoblasts were treated with 250 μM 8-Br-cAMP to induce differentiation from cytotrophoblast-like (CT) cells to SCT cells. Cell cultures were collected after 72 hours of high glucose exposure. A schematic of the HG culture protocol is available in Fig 1.
**Fig 1:** *Schematic of 72-hour HG cell culture protocol.BeWo trophoblasts were plated in F12K media and allowed to adhere to culture plates overnight. At T0H cells were treated with low glucose (LG, 7 mM) or hyperglycemic (HG, 25 mM) supplemented F12K media. At T24H subsets of BeWo trophoblasts were treated with 250 μM 8-Br-cAMP to induce CT-to-SCT differentiation. Cell media was replenished every 24 hours and cells were collected at T72H for analysis of metabolic function.*
## Cell viability of HG-cultured BeWo trophoblasts
BeWo trophoblasts were plated at 7.5x103 cells/well in black walled 96-well cell culture plates and cultured as described. At T72h cell viability of both CT and SCT cultures was assessed via the CellTiter-Fluor cell viability assay (Promega Corporation, Madison WI, USA) as per manufacturer’s instructions.
## Analysis of BeWo cell fusion under HG culture conditions
To determine the potential of HG culture conditions to impact the ability of BeWo trophoblasts to differentiate into SCT cells, cell fusion of 8-Br-cAMP stimulated cells (expressed as percent loss of the tight junction protein zona occludens-1 (ZO-1)) was examined. In brief, BeWo cells were plated at 1.4x105 cells/well in 6-well plates containing coverslips coated with laminin (2 μg/cm2) and grown under HG conditions as described. Cellular expression of ZO-1 was then examined via immunofluorescent microscopy as previously detailed [37].
## RT-qPCR analysis of BeWo syncytialization under HG conditions
The expression of the transcription factor Ovo Like Transcriptional Repressor 1 (OVOL1) as well as human chorionic gonadotropin subunit beta (CGB) was additionally analyzed to ensure cell fusion in 8-Br-cAMP stimulated BeWo cells was associated with increased expression of syncytialization-related genes. In brief, BeWo trophoblasts were plated at 3.5x105 cells/plate in 60mm cell culture plates and cultured as described above. At T72h cells were collected in TRIzol reagent (Invitrogen), and total RNA was extracted as per the manufacturer’s protocol. RNA integrity was assessed via formaldehyde gel electrophoresis, and RNA concentration was quantified by Nanodrop Spectrophotometer 2000 (NanoDrop Technologies, Inc., Wilmington, DE, USA). RNA (2 μg) was then reverse transcribed with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; ThermoFisher Scientific). RT-qPCR was then performed via the CFX384 Real Time system (Bio-Rad, Mississauga, Canada). *Relative* gene expression of OVOL1 [38] and CGB [39] was then determined using the ΔΔCt method with the geometric mean of PSMB6 and ACTB utilized as a reference gene. Primer sequences and their efficiencies are available in S1 Table in S2 File.
## Seahorse XF Mito Stress Test quantification of mitochondrial respiratory activity
BeWo cells were plated at 7.5x103 cells/well in Seahorse XF24 V7PS plates and cultured under LG and HG conditions as described above. At T72h mitochondrial activity was assessed using the Seahorse XF Mito Stress Test, as previously optimized for BeWo trophoblast cells [37]. In brief, oxygen consumption rate (OCR) of cell culture media was quantified as a proxy measure of mitochondrial respiratory activity. Subsequent injections of oligomycin (1.5 μg/mL), dinitrophenol (50 μM) and Rotenone and Antimycin A (0.5 μM each) were used to interrogate the basal respiration, maximal respiration, proton leak, spare respiratory capacity and coupling efficiency parameters of mitochondrial respiratory activity. OCR measures were normalized to cellular DNA content using Hoechst dye fluorescence as previously described [37].
## Seahorse XF Glycolysis Stress Test quantification of BeWo trophoblast glycolytic activity
BeWo cells were plated at 7.5x103 cells/well in Seahorse XF24 V7PS plates and cultured under LG and HG conditions as described above. At T72h glycolytic activity was assessed via the Seahorse XF Glycolysis Stress Test, as previously optimized for BeWo trophoblasts [37]. In brief, Extracellular Acidification Rate (ECAR) of cell culture media was assessed to quantify glycolytic activity in BeWo trophoblasts. Injections of glucose (10 mM), oligomycin (1.5 μg/mL) and 2-deoxyglucose (50 mM) were utilized to interrogate the basal glycolytic rate, maximal glycolytic rate, glycolytic reserve, and non-glycolytic acidification parameters of glycolytic function. ECAR data was normalized to cellular DNA content using Hoechst dye fluorescence as previously described [37].
## Immunoblot analysis of protein abundance
BeWo trophoblasts were plated at 9.5x105 cells/plate in 100 mm cell culture dishes and cultured under LG and HG conditions as described. At T72h cells were washed once in ice-cold PBS and subsequently lysed in radioimmunoprecipitation assay (RIPA) buffer supplemented with protease and phosphatase inhibitors as previously described [37]. Protein concentrations were then adjusted to 2 μg/μL in Laemmli loading buffer (62.5 mM Tris-Cl (pH 6.8); $2\%$ SDS $10\%$ glycerol; $0.002\%$ bromophenol blue; $4\%$ β-mercaptoethanol).
The relative abundance of target proteins was then determined via SDS-PAGE gel electrophoresis. In brief, protein lysates were separated on acrylamide gels (10–$15\%$), and separated proteins were transfer to polyvinylidene fluoride (PVDF) membranes (EMD Millipore, Fisher Scientific). Total lane protein was then determined via Ponceau stain ($0.1\%$ Ponceau-S in $5\%$ acetic acid) and utilized to normalize densitometry values. PVDF membranes were blocked in $5\%$ dry-milk protein or $5\%$ Bovine Serum Albumin (BSA, BioShop Canada Inc., Burlington, Canada) and membranes were incubated overnight at 4°C with respective antibody solutions (S2 Table in S2 File). Membranes were then washed 3 times with Tris-buffered Saline with $0.1\%$ Tween-20 (TBST) and incubated with respective secondary antibodies for 1 hour at room temperature (S2 Table in S2 File). Membranes were washed 3 more times in TBST and imaged on a ChemiDoc Imager (Bio-Rad) using Clarity western ECL substrate (Bio-Rad). Protein band abundance and total lane protein (ponceau) were quantified with Image-Lab Software (Bio-Rad).
## Analysis of ETC complex I and II activity in HG BeWo trophoblasts
BeWo trophoblasts were plated at 3.5x105 cells/plate in 60mm cell culture dishes and cultured under LG and HG conditions as described. At T72h cells were washed once with PBS and detached from cell culture plates by scraping. Cells were then pelleted (400g, 5 mins), snap frozen in liquid nitrogen and stored at -80°C until analyzed. Cell pellets were subsequently lysed and Complex I activity was assessed as the rate of rotenone-sensitive NADH oxidation, and Complex II activity was assessed as the rate of DCPIP oxidation as previously detailed [37]. ETC complex activity assays were normalized to cell lysate protein content via Bicinchoninic Acid (BCA) assay (Pierce, ThermoFisher Scientific) as per manufacturer’s instructions.
## Analysis of metabolic enzyme activities in HG BeWo trophoblasts
BeWo trophoblasts were plated at 3.5x105 cells/plate in 60mm cell culture dishes and cultured under LG and HG conditions as detailed above. To examine the enzyme activities of Lactate Dehydrogenase (LDH) and Citrate Synthase (CS), cells were collected by scraping and lysed in glycerol lysis buffer (20 mM Na2HPO4, 0.5 mM EDTA, $0.1\%$ Triton X-100, $0.2\%$ BSA, $50\%$ glycerol) containing protease and phosphatase inhibitors as previously described [37,40]. LDH activity was assessed as the rate of NADH oxidation, and CS activity was assessed as the rate of Ellman’s reagent consumption as previously detailed [37]. The enzyme activity of LDH and CS were normalized to cell lysate protein content via BCA assay.
Additional BeWo cultures were collected to analyze the activity of the E1 (rate-limiting) subunit of the Pyruvate Dehydrogenase (PDH) complex. PDH-E1 subunit was assessed on freshly collected BeWo cells as the rate of DCPIP oxidation and normalized to protein content via BCA assay as previously detailed [37].
## Nutrient storage in HG BeWo trophoblasts
BeWo trophoblasts were plated at 3.5x105 cells/plate in 60mm cell culture dishes and cultured under LG and HG conditions. At T72h cells were washed with PBS and collected into fresh PBS (1.5 mL) by scrapping. Cells were then pelleted (400g, 5 minutes) and the PBS was aspirated.
To determine cellular glycogen content, the cell pellets were lysed in 200 μL ddH2O and samples were boiled for 10 minutes to inactivate cell enzymes. Samples were stored at -20 until glycogen content was analyzed. Samples were diluted 5-fold and glycogen content was analyzed via the Glycogen Assay Kit (ABCAM, ab65620) as per manufacturer’s protocol. Glycogen content was then normalized to cell lysate protein content via BCA assay.
To determine cellular triglyceride accumulation, the collected cell pellets were snap frozen in liquid nitrogen and stored at -80°C until analyzed. Cells were then lysed, and cellular triglyceride content was analyzed via the Triglyceride Assay Kit (ABCAM, ab178780) as per kit instructions. Triglyceride content was normalized to cell lysate protein content determined via BCA assay.
## Transcriptomic analysis of gene expression changes in HG BeWo CT cells
BeWo CT cells were plated at 3.5x105 cells/plate in 60 mm cell culture dishes and grown under LG and HG culture conditions as described. At T72h cells with washed once with PBS and collected in 900 μL TRIzol Reagent and stored at -80°C. Samples were then shipped to the Genome Québec Innovation Centre for transcriptomic analysis via Clariom S mRNA microarray. Automated RNA extraction was completed via QIAcube Connect (Qiagen, Toronto, Canada). RNA content was then quantified via NanoDrop Spectrophotometer 2000 (Nanodrop Technologies, Inc) and RNA integrity was determined by Bioanalyzer 2100 (Agilent Technologies, Waldbronn, Germany). All extracted RNA samples had an RNA Integrity Number (RIN) greater than 9.5. RNA was then processed via the Affymetrix Whole Transcript 2 workflow and analyzed using a Clariom S human mRNA microarray. In brief, sense-stranded cDNA was synthesized from 100 ng total RNA, and subsequently fragmented, and labelled with the GeneChip WT Terminal Labeling Kit (ThermoFisher Scientific) as per manufacturer’s instructions. Labelled DNA was then hybridized to Clariom S human GeneChips (ThermoFisher Scientific) and incubated at 45°C in the GeneChip Hybridization oven 640 (Affymetrix, ThermoFisher Scientific) for 17 hours at 60 rpm. GeneChips were washed using GeneChip Hybridization Wash and Stain Kit (ThermoFisher Scientific) according to manufacturer’s specifications. Microarray chips were scanned on a GeneChip scanner 3000 (ThermoFisher Scientific).
Microarray data was then analyzed via Transcriptome Analysis Console v4.0 (ThermoFisher Scientific), and raw data was normalized using the Robust Multiple-Array Averaging (RMA) method. HG-treated samples were paired with respective LG-control for each cell collection for analysis. Genes with a ≥ ±1.3 fold-change (FC) vs LG-control and raw-$p \leq 0.05$ were determined to be differentially expressed.
RMA-normalized microarray signals (for all identified genes) were then imported into the Gene Set Enrichment Analysis (GSEA) v4.2.3 software (Broad Institute Inc., Cambridge MA, USA) [41,42], and the genes annotated on the microarray chip were ranked via a signal-to-noise score. The ranked gene list was subsequently uploaded to the Web-Based Gene Set Analysis Toolkit (WebGestalt) to conduct GSEA analysis using the KEGG, Wikipathways and Reactome functional gene sets, as well as the Gene Ontology (GO) biological processes and molecular function gene sets [43]. Identified pathways with a False Discovery Rate-corrected p-value <0.25 were determined to be significantly enriched the HG or LG-cultured BeWo CT cells as has previously been reported [42,44].
## RT-qPCR validation of differentially expressed genes identified by mRNA microarray
RT-qPCR was utilized to validate differentially expressed genes involved in metabolic pathways that were highlighted by the mRNA microarray. The RNA utilized for the microarray was returned by Genome Québec and 2 μg was reverse transcribed as described above. The CT samples previously utilized to examine expression of syncytialization-related genes were additionally utilized to validate the differentially expressed genes identified in the microarray. RT-qPCR was then performed and analyzed using the ΔΔCt with the geometric mean of ACTB and PSMB6 used as a reference. Primer sequences of validated targets and their efficiencies are available in S1 Table in S1 File.
## Untargeted metabolomic profiling of HG-treated BeWo CT cells
BeWo trophoblasts were plated at a density of 2x106 cells/plate in 150mm cell culture plates and cultured under LG and HG-conditions as described. At T72h cell media was aspirated and the cells were washed three times with cold PBS. Pre-cooled methanol (-20°C) was then added to quench cellular metabolic processes. Cells were then scraped and collected into microcentrifuge tubes, and the methanol was evaporated with a gentle flow of nitrogen gas. Samples were then frozen at -80°C, and subsequently were lyophilized to remove any residual moisture. The samples were then sent to The Metabolomics Innovation Centre (Edmonton, Canada) for subsequent metabolomics analysis via a High-Performance Chemical Isotope Labelling (HP CIL) liquid chromatography mass spectrometry (LC-MS) approach [45,46].
Samples were reconstituted in $50\%$ methanol and freeze-thaw cycles were utilized to lyse the cells. The lysed samples were centrifuged at 16000 g at 4°C and the supernatants were transferred to new vials. The supernatants were dried down and re-dissolved in 41 μL of water. The total concentrations of metabolite were then determined via the NovaMT Sample Normalization kit (Edmonton, Alberta). Water was added to adjust all the concentrations of samples to 2 mM. The samples were split into five aliquots for respective labeling methods. Each of the individual samples was labeled by 12C-DnsCl, base activated 12C-DnsCl, 12C-DmPA Br, and 12C-DnsHz, for amine-/phenol-, hydroxyl-, carboxyl, and carbonyl- metabolomic profiling, respectively [47]. A pooled sample was generated by mixing each individual sample and was labeled by 13C-reagent, accordingly. Each 12C-labeled individual sample was then mixed with a 13C- labeled reference sample by equal volume, and the mixtures were injected onto LC-MS for analysis. The LC-MS system was the Agilent 1290 LC (Agilent Technologies) linked to the Bruker Impact II QTOF Mass Spectrometer (Bruker Corporation, Billerica, US). LC-MS data was then exported to.csv files with Bruker DataAnalysis 4.4 (Bruker Corporation), the exported data were then uploaded to IsoMS pro v1.2.7 for data quality check and data processing.
Metabolite peak pairs were then identified using a three-tier approach [47]. In tier 1, peak pairs were identified by searching against a labelled metabolite library (CIL library) based on accurate mass and retention time. In tier 2, the remaining peak pairs were matched by searching against a linked identity library (LI library), containing predicted retention time and accurate mass information. In tier 3, the rest of peak pairs were matched by searching against MyCompoundID (MCID) library, containing accurate mass information of metabolites and their predicted products. Metabolites with ± 1.5 FC and False Discovery Rate raw-$p \leq 0.05$ vs LG CT were determined to be differentially abundant in the HG-cultured BeWo CT cells.
Metabolites from tiers 1 and 2 (high confidence identifications) with an associated KEGG library numbers were subsequently identified for pathway analysis to elucidate the biological impacts of the differentially abundant metabolites. Enrichment in the Homo sapiens KEGG library and Small Molecule Pathway Database (SMPDB) was performed using MetaboAnalyst v5.0 (https://www.metaboanalyst.ca/). KEGG and SMPDB pathways with an FDR $p \leq 0.05$ were determined to be significantly enriched.
## Integration of transcriptome and metabolome profiles
Differentially expressed genes (±1.3 FC, raw-$p \leq 0.05$), and differentially abundant metabolites (±1.5 FC, raw-$p \leq 0.05$) with a KEGG library identification library were then imported into the Joint Pathway Analysis Tool in MetaboAnalyst v5.0 to identify enriched pathways containing both altered genes and metabolites in the HG-cultured BeWo CT cells. Pathways with a raw-$p \leq 0.05$ were determined to be significantly enriched.
## Statistical analysis of non-omics data
Data collected as a percentage (percent loss of ZO-1 staining data, spare respiratory capacity, coupling efficiency, and glycolytic reserve) were log-transformed and analyzed via Two-Way ANOVA (2WA) and Bonferroni’s Multiple Comparisons post-hoc test. A Randomized Block Design 2WA and Sidak’s Multiple Comparisons Test was utilized to analyze relative transcript abundances; relative protein abundances; metabolic activity parameters; and nutrient storage data, using raw data with data from each experimental replicate blocked together, as previously described [48]. These data were then expressed as percent of LG CT control for visualization in figures. Data from phosphorylated western blot targets were normalized to the average LG CT expression on each independent membrane prior to calculation of the phosphorylated protein-to-total protein abundance ratio, and subsequently analyzed via randomized block-design ANOVA, and Sidak’s multiple comparisons test. A paired T-test was utilized to analyze gene FC data between HG and LG BeWo CT cells in the RT-qPCR validation of the microarray. All data was analyzed for statistically significant outliers using the Rout method ($Q = 1$%). Statistical analysis was performed with GraphPad Prism 8 Software (GraphPad Software, San Diego, CA, USA).
## Characterization of BeWo viability and differentiation under high glucose culture conditions
HG culture conditions were associated with a mean $7\%$ and $10\%$ reduction in cell viability in BeWo CT and SCT cells respectively (Fig 2; $p \leq 0.01$; $$n = 6$$/group). It is important to note that these viability data were consistent with previously reported values in BeWo trophoblasts and PHTs cultured under hyperglycemia (25 mM glucose) for 48h [30,49]. This highlights that a 72H HG-culture protocol does not impact the stability of cultured BeWo trophoblasts cells. Additionally, BeWo SCT cells displayed lower viability relative to BeWo CT cells consistent with trophoblast cells becoming less proliferative while undergoing syncytialization (Fig 2; $p \leq 0.001$).
**Fig 2:** *Viability of BeWo trophoblasts cultured under HG conditions for 72h.BeWo trophoblasts were cultured for 72H under HG culture conditions as described, and cell viability was assessed via the CellTiter-Flour cell viability assay. Data is presented as percent of LG CT cell viability (n = 6/group). Raw cell viability fluorescence data was analyzed via Two-Way Randomized block design ANOVA (2WA). Different lower-case letters denote differentiation state-dependent differences in viability between CT and SCT cultures, and different upper-case letter denote hyperglycemia-dependent differences in viability within each differentiation state (p<0.05).*
BeWo SCT cultures displayed a greater loss of ZO-1 protein expression (and thus increased cell fusion) compared to BeWo CT cultures (Fig 3A and 3B; $p \leq 0.0001$, $$n = 4$$/group). However, HG-culture conditions had no impact on ZO-1 protein expression in BeWo trophoblast cells (Fig 3A and 3B). BeWo SCT cultures additionally displayed increased relative transcript abundance of the syncytialization-associated transcription factor OVOL1 (Fig 3C; $p \leq 0.01$, $$n = 5$$/group) as well as the syncytialization-associated hormone CGB (Fig 3D; $p \leq 0.01$, $$n = 5$$/group). HG culture conditions in BeWo SCT cells was additionally associated with an increased transcript abundance of CGB compared to LG BeWo SCT cells (Fig 3D; $p \leq 0.05$; 2WA: Interaction $p \leq 0.05$).
**Fig 3:** *HG culture conditions do not impact BeWo SCT cell fusion but are associated with increased CGB transcript abundance at 72h.(A) Representative DAPI (blue); Zona occludens-1 (ZO1; red) and merged immunofluorescent images (scale bar = 50 μm) of BeWo CT and SCT cells. (B) Percent fusion of HG-cultured BeWo trophoblasts. Percent fusion data was log-transformed and analyzed via Two-Way Anova (2WA); data is expressed as the percentage of cells lacking ZO-1 expression (n = 4/group). BeWo cell syncytialization was additionally analyzed via quantifying mRNA transcript abundance of the syncytialization markers (C) OVOL1 and (D) CGB. Transcript abundance data was analyzed via Randomized Block 2WA and Sidak’s multiple comparisons test; data is present as transcript FC versus LG CT cultures (n = 5/group). Different lower-case letters denote differentiation state-dependent differences between CT and SCT cultures, and different upper-case letter denote hyperglycemia-dependent differences within each differentiation state (p<0.05).*
## BeWo mitochondrial respiratory and glycolytic activity
High glucose culture conditions did not impact any of the parameters of mitochondrial respiratory function as assessed by the Seahorse XF Mito Stress Test (Table 1; $$n = 5$$/group). However, BeWo syncytialization was associated with reduced Spare Respiratory Capacity (Table 1; $p \leq 0.0001$) and reduced Coupling Efficiency (Table 1; $p \leq 0.01$) independent of culture glucose level.
**Table 1**
| Differentiationstate | Treatment | Basal glycolyticrate(ECAR/μg DNA) | Max glycolyticrate(ECAR/μg DNA) | Reserve glycolytic capacity(% basal glycolysis) | Non-glycolyticacidification(ECAR/μg DNA) |
| --- | --- | --- | --- | --- | --- |
| CT | LG | 75.71 ± 15.38 | 101.4 ± 20.92 | 133.8 ± 1.14 | 19.26 ± 4.43 |
| CT | HG | 67.54 ± 7.30 | 88.23 ± 11.82 | 129.5 ± 4.08 | 18.38 ± 3.10 |
| SCT | LG | 69.21 ± 12.67 | 84.45 ± 14.55 | 123.5 ± 3.70 | 20.50 ± 3.39 |
| SCT | HG | 77.38 ± 6.78 | 88.01 ± 8.17 | 113.6 ± 1.45 | 23.30 ± 2.45 |
| *Differentiation state difference | *Differentiation state difference | NS | NS | * | NS |
The parameters of glycolytic function as measured with the Seahorse XF Glycolysis Stress Test were not impacted by BeWo syncytialization, or by HG culture conditions (Table 2). Representative tracings from the Seahorse Mito Stress Test and Glycolysis Stress Test are available in S1 Fig in S1 File.
**Table 2**
| Differentiationstate | Treatment | Basal respiration(OCR/μg DNA) | Maximal respiration(OCR/μg DNA) | Proton leak(OCR/μg DNA) | Spare respiratory capacity(% basal OCR) | Coupling efficiency(% basal OCR) |
| --- | --- | --- | --- | --- | --- | --- |
| CT | LG | 199.70 ± 39.08 | 249.10 ± 43.78 | 55.48 ± 8.45 | 129.00 ± 5.30 | 70.38 ± 3.34 |
| CT | HG | 173.40 ± 24.69 | 221.30 ± 36.78 | 46.72 ± 3.76 | 125.50 ± 6.79 | 70.58 ± 4.59 |
| SCT | LG | 173.00 ± 7.96 | 154.40 ± 11.97 | 78.20 ± 7.38 | 89.02 ± 5.68 | 55.04 ± 2.99 |
| SCT | HG | 177.5 ± 19.07 | 168.30 ± 18.27 | 75.88 ± 12.12 | 95.76 ± 5.83 | 57.30 ± 3.80 |
| *Differentiation state difference | *Differentiation state difference | NS | NS | NS | **** | ** |
## The impact of high glucose and syncytialization upon BeWo ETC complex protein abundance and activity
High glucose levels did not impact protein abundance of ETC complex subunits in BeWo trophoblasts (Fig 4A–4E; $$n = 4$$-5/group). Likewise, HG-culture conditions did not affect ETC complex I or II activity in BeWo trophoblasts (Table 3, $$n = 5$$-6/group). However, BeWo syncytialization was associated with reduced ETC complex IV Cytochrome c oxidase subunit II (COXII) relative protein abundance in both LG and HG cultures (Fig 4D; $p \leq 0.05$). Additionally, BeWo SCT cells displayed reduced activity of ETC complex II compared to CT cells independent of culture glucose level (Table 3; $p \leq 0.001$, $$n = 6$$/group).
**Fig 4:** *HG culture conditions do not impact protein expression of electron transport chain complexes in BeWo trophoblasts.Relative protein abundance of (A) Complex I (NDUFB8 subunit) (B) Complex II (SDHB subunit) (C) Complex III (UQCRC2 subunit) (D) Complex IV (COX II subunit), and (E) Complex V (ATP5A subunit) of the electron transport chain (ETC) in HG-cultured BeWo trophoblasts. Different lower-case letters denote differentiation state-dependent differences in ETC complex protein abundance between CT and SCT cells ($$n = 4$$-5/group; Two-Way Randomized Block ANOVA (2WA)). ETC complex protein band density was normalized to total lane protein (ponceau) for statistical analysis and the data is presented as percent of LG CT protein abundance for visualization. Full length representative images of ETC complex bands and ponceau staining of total lane protein are available in S2 Fig in S1 File.* TABLE_PLACEHOLDER:Table 3
## Metabolic enzyme activity in HG-cultured BeWo trophoblasts
BeWo SCT cells displayed increased activity of the PDH-E1 subunit compared to BeWo CT cells regardless of glucose level (Table 3; $p \leq 0.01$, $$n = 6$$/group). Additionally, high glucose levels did not impact the activities of the PDH-E1 subunit, citrate synthase (CS) or lactate dehydrogenase (LDH) in BeWo CT and SCT cells (Table 3, $$n = 6$$/group). Syncytialization additionally did not affect the enzyme activities of CS or LDH in BeWo trophoblasts (Table 3).
## HG-cultured BeWo trophoblast mitochondrial fission and fusion dynamics
Regardless of glucose level BeWo SCT cells displayed increased relative protein abundance of the mitochondrial fission marker DRP1 in conjunction with decreased relative protein levels of the mitochondrial fusion marker OPA1 compared to undifferentiated BeWo CT cells (Fig 5A and 5B; $p \leq 0.05$, $$n = 5$$/group).
**Fig 5:** *Syncytialization impacts mitochondrial dynamics in BeWo trophoblasts.Relative protein abundance of (A) total DRP1, (B) OPA1, and (C) ratio of pSER616 DRP1 to total DRP1. Different lower-case letters denote differentiation state-dependent differences in relative protein abundance between CT and SCT cultures (n = 5/group; Two-Way Randomized Block ANOVA (2WA), Sidak’s multiple comparisons test). Protein band density was normalized to total lane protein (ponceau) for statistical analysis and the total protein abundance data is presented as percent of LG CT abundance for visualization. Protein band density data from each membrane was normalized to the average LG CT protein band density prior to calculation of the phosphorylated protein abundance-to-total protein abundance ratio. Uncropped representative images of protein bands and ponceau staining of total lane protein are available in S4 and S5 Figs in S1 File.*
HG culture conditions additionally did not impact total protein abundance of OPA1 and DRP1 in BeWo trophoblasts (Fig 5A and 5B). There was a trend towards increased pSER616 phosphorylation of DRP1 in HG-cultured BeWo trophoblasts (mean $19\%$ increase; S3A Fig in S1 File; $$p \leq 0.0632$$, $$n = 5$$/group) as well as reduced pSER616 DRP1 phosphorylation in syncytialized BeWo trophoblasts (S3A Fig in S1 File; $$p \leq 0.0924$$) when protein levels were normalized to total lane protein via ponceau stain. When the pSER616 DRP1 bands were expressed relative to total DRP1 protein levels, there was increased pSER616 DRP1:DRP1 levels in HG cultured trophoblasts (Fig 5C; pglucose level <0.05, $$n = 5$$/group), and decreased pSER616 DRP1:DRP1 levels in syncytialized BeWo trophoblasts (Fig 5C; pdifferentiation state <0.01). Post-hoc analysis identified trends towards significance in pSER616 DRP1-to-total DRP1 ratio levels in both BeWo CT cells ($$p \leq 0.0725$$) and SCT cells ($$p \leq 0.0512$$) (Fig 5C).
## HG culture conditions impact glycogen storage in BeWo trophoblasts
HG culture conditions in both BeWo CT and SCT cells resulted in increased cellular glycogen content compared to respective differentiation state LG cultures (Fig 6A; $p \leq 0.0001$; $$n = 4$$/group). Furthermore, the glycogen content in BeWo CT cultures was found to be greater than that of the SCT cultures (Fig 6A; $p \leq 0.0001$). In addition, BeWo SCT cells displayed increased GLUT1 protein abundance compared to BeWo CT cells (Fig 6B; $p \leq 0.05$), however, no glucose-dependent impacts to GLUT1 protein abundance were observed.
**Fig 6:** *HG culture conditions impact glycogen storage and regulation in BeWo trophoblasts.(A) Glycogen content and relative protein abundance of (B) GLUT1 (C) glycogen synthase; (D) the pSer641 glycogen synthase-to-total glycogen synthase ratio; (E) GSK3β; and (F) the pSer9 GSK3β-to-total GSK3β ratio in HG-treated BeWo trophoblasts at T72H. Different lower-case letters denote differentiation state-dependent differences in protein abundance between CT and SCT cultures, and different upper-case letters denote glucose-level dependent differences within a differentiation state (Randomized-Block Two-Way ANOVA (2WA), and Sidak’s multiple comparisons test, p<0.05). Glycogen content data are presented as protein normalized glycogen abundance (μg glycogen per μg protein; n = 4/group). Protein band density was normalized to total lane protein (ponceau) for statistical analysis and the data is presented as percent of LG CT abundance for visualization (n = 5/group). Protein band density data from each membrane was normalized to the average LG CT protein band density prior to calculation of the phosphorylated protein abundance-to-total protein abundance ratio. Uncropped representative images of protein bands and ponceau staining of total lane protein are available in S6 Fig in S1 File.*
However, HG culture conditions were associated with reduced relative glycogen synthase protein abundance in both BeWo CT and SCT cells (Fig 6C; $p \leq 0.05$, $$n = 5$$/group). HG cultured BeWo CT and SCT cells displayed increased total abundance of phosphorylated (pSer641) glycogen synthase (S3B Fig in S1 File; $p \leq 0.01$, $$n = 5$$/group), and additionally differentiated BeWo SCT cells displayed increased phosphorylated (pSer641) glycogen synthase levels compared to undifferentiated CT cultures (S3B Fig in S1 File; $p \leq 0.01$, $$n = 5$$/group). When phosphorylated Glycogen Synthase levels were expressed relative to total Glycogen Synthase levels, an increase in inhibitory phosphorylation was observed in both HG-cultured BeWo CT cells and SCT cells (Fig 6D; $p \leq 0.05$, $$n = 5$$/group).
Furthermore, differentiated BeWo SCT cells were found to have increased protein levels of both GSK3β (Fig 6E; $p \leq 0.05$, $$n = 5$$/group) and phosphorylated (pSer9) GSK3β (S3C Fig in S1 File; $p \leq 0.05$, $$n = 5$$/group) compared to BeWo CT cultures. However, levels of phosphorylated (pSer9) GSK3β were found to not impacted by glucose culture condition when the protein levels were expressed as a phosphorylated protein abundance-to-total protein abundance ratio (Fig 6F; $$n = 5$$/group). HG culture conditions were not associated with alterations in the relative protein abundance of either GSK3β or phosphorylated (pSer9) GSK3β (Fig 6E and 6F; $$n = 5$$/group).
## HG culture conditions increase TG abundance in BeWo CT cells
HG culture conditions were associated with increased triglyceride accumulation in BeWo CT cells, but not in BeWo SCT cells (Fig 7A; $p \leq 0.01$, $$n = 4$$/group). Furthermore, BeWo syncytialization and HG culture conditions did not impact the relative abundance of ACSL1 protein, although there was a trend towards increased expression in both differentiated BeWo SCT cells and in HG-treated BeWo trophoblasts (Fig 7B; 2WA: glucose level $$p \leq 0.0894$$; 2WA: differentiation state $$p \leq 0.0695$$, $$n = 5$$/group). Finally, syncytialization and HG culture conditions did not impact the relative abundance of fatty acid synthase (FASN) in BeWo trophoblasts (Fig 7C).
**Fig 7:** *HG culture conditions impact triglyceride content in BeWo CT cells.(A) Triglyceride content and relative protein abundance of (B) ACSL1, and (C) FASN in HG-treated BeWo trophoblasts at T72H relative abundance. Different upper-case letters denote hyperglycemia-dependent differences between LG and HG treatments within each respective differentiation state (Two-Way Randomized Block ANOVA; Sidak’s multiple comparisons test, p<0.05). TG content data is presented as protein normalized TG abundance (pmol TG per μg protein; n = 4/group). Protein band density was normalized to total lane protein (ponceau) for statistical analysis and the data is presented as percent of LG CT abundance for visualization (n = 5/group). Uncropped representative images of protein bands and ponceau staining of total lane protein are available in S4 Fig in S1 File.*
## Transcriptomic profiling of HG-cultured BeWo CT cells
HG-cultured BeWo CT cells displayed 197 differentially expressed genes (75 upregulated, 122 down-regulated) compared to LG BeWo CT cells (S3 Table in S2 File, ≥ ±1.3-FC vs LG CT). The volcano plot (Fig 8) and 2D hierarchical clustering heatmap (S7 Fig in S1 File) were constructed to visualize the degree of gene expression differences between LG and HG cultured BeWo CT cells.
**Fig 8:** *Volcano plot visualization of differentially expressed genes between LG and HG cultured BeWo CT cells.The volcano plot was generated to visualize differentially expressed genes in HG-cultured BeWo CT cells (≥ ±1.3-fold change, p<0.05, n = 5/group). Overall, 197 genes (122 up-regulated, and 75 down-regulated) were identified to be differentially expressed in HG-cultured BeWo CT cells. A summary list of the identified differentially expressed genes is available in S3 Table in S2 File. The x-axis indicates fold-changes vs LG BeWo CTs, and the y-axis indicates the p-value (-log10). The green dots represent statistically significant down-regulated genes, and the red dots represent statistically significant upregulated genes in HG-cultured BeWo CT cells.*
Gene-Set Enrichment Analysis (GSEA) was then performed using all transcripts identified in the microarray panel, with genes ranked according to their signal-to-noise ratio. A list of all genes and their associated signal-to-noise rank score is available in S4 Table in S2 File. Enrichment analysis was performed using the KEGG, Reactome, and Wikipathways functional gene sets, as well as with the Gene Ontology (GO) biological processes and molecular function gene sets. Significant enrichment in KEGG pathways was found in the HG-cultured BeWo CT cells that were associated with an upregulation in the “Fatty Acid Biosynthesis” gene set (the genes involved were: ACSL1 (+1.36 FC), ACACB, ACSL5, and ACSBG1), and the “Glutathione Metabolism” gene set (the gene involved were: LAP3 (+1.32 FC), GPX5, CHAC1, GGT7, MGST1, ANPEP, GSTA3, GLCM, GSTO1, and NAT8) (FDR-$p \leq 0.25$, Fig 9A). A significant down-regulation in the KEGG “Nitrogen Metabolism” gene set (the genes involved were: CA5A (-1.53 FC), CA1, CA6, CA7, and CA13) was also observed (FDR-$p \leq 0.25$, Fig 9A).
**Fig 9:** *Gene set enrichment analysis of differentially expressed genes and RT-qPCR validation of microarray gene changes.(A) Top 10 up-regulated and down-regulated KEGG gene sets and (B) top 10 up-regulated and down-regulated Reactome gene sets (by normalized enrichment ratio) in HG-cultured BeWo CT cells. Green bars represent down-regulated gene sets and red-bars represent up-regulated gene sets, with dark green bars representing significantly down-regulated and dark red bars representing significantly up-regulated gene sets (FDR-p<0.25). The transcripts identified in the microarray were ranked by signal-to-noise ratio using the Gene Set Enrichment Analysis Software, and enrichment analysis was performed using WebGestalt. (C) RT-qPCR was utilized to validate the differential expression of ACSL1, CYP2F1, HK2, HSD11B2, LAP3, and RPS6KA5 in HG-cultured BeWo CT cells. Data was analyzed via paired t-test (*p<0.05, **p<0.01, n = 9-10/group), and data expressed as fold-change vs LG BeWo CT (mean ± SEM). Down-regulated genes ratios were inversed and expressed as a negative (e.g 0.5-fold is expressed as a minus-2 fold-change).*
Analysis using the Reactome functional gene sets was associated with significant up-regulation in the “Synthesis of very long-chain fatty acyl-CoAs” gene set (the genes involved were: ELOVL7, ACSL1 (+1.36 FC), ELOVL4, ELOVL3, ACSL5, and ACSBG1); the “Fatty acyl-CoA biosynthesis” gene set (the genes involved were: ELOVL7, ACSL1 (+1.36 FC), ELOVL4, ELOVL3, ACSL5, CBR4, and ACSBG1); the “CD209 (DC-SIGN) signaling” gene set (the genes involved were: CD209, RPS6KA5 (+1.32 FC), FYN, ICAM3, PRKACB, HRAS, and ICAM2); and the “Cristae formation” gene set (the genes involved were: APOO (+1.5 FC), and CHCHD6) (FDR-$p \leq 0.25$, Fig 9B). No significant enrichment was found in the Wikipathways functional gene sets (S8A Fig in S1 File).
GSEA analysis in the Gene Ontology (GO) databases revealed a significant down-regulation in the GO “Oxygen Binding” molecular pathway database (the genes involved were: CYP2A16, CYP1A1, CYP2F1 (-1.32 FC), ADGB, ALB, CYP2A7, NGB, NOX4, CYP17A, and HBM) (FDR-$p \leq 0.25$; S8B Fig in S1 File). There was no significant enrichment in the GO biological processes database (S8C Fig in S1 File).
The expression of DEGs identified to be involved in significantly upregulated and down-regulated functional pathways (ACSL1, APOO, CA5A, CYP2F1, DGKG, GPAT2, LAP3, LIPF, RPS6KA5) were subsequently validated via RT-qPCR. Individual DEGs involved in cellular metabolic processes (HSD11B2, HK2, and SLC27A2) were also identified from the microarray panel and selected validation via RT-qPCR. The transcripts for the DEGs: CA5A, DGKG, LIPF could not be detected via RT-qPCR in BeWo lysates, and thus the differential expression of these targets could not be validated.
However, the differential expression of ACSL1 (+1.36 FC microarray; +1.22 FC RT-qPCR); CYP2F1 (-1.32 FC microarray; -1.31 FC RT-qPCR); HSD11B2 (+1.35 FC microarray; +1.38 FC RT-qPCR); HK2 (-1.39 FC microarray; -1.39 FC RT-qPCR); LAP3 (+1.32 FC microarray; +2.17 FC RT-qPCR); and RPS6KA5 (+1.32 FC microarray; +1.47 FC RT-qPCR) in HG-cultured BeWo CT cells were validated via RT-qPCR. ( Fig 9C) The differential expression of: APOO (+1.50 FC microarray; +1.15 FC, $p \leq 0.05$ RT-qPCR); GPAT2 (-1.37 FC microarray; +1.10 FC, $p \leq 0.05$ RT-qPCR; and SLC27A2 (-1.40 FC microarray; +1.29 FC, $p \leq 0.05$ RT-qPCR) in HG-cultured BeWo CT cells could not be validated via RT-qPCR (Fig 9C).
## Impacts of HG culture conditions on the metabolome of BeWo CT cells
On average 6541 ± 42 (mean ± SD) metabolite peak pairs were measured in each sample. A summary of the metabolites identified in all tiers is available in S5 Table in S2 File. Of these peak pairs, 179 were positively identified in tier 1 and 602 peak pairs were identified with high confidence in tier 2. Of these identified peak pairs, 4 from tier 1, 89 from tier 2, and 627 from tier 3 were found to be differentially abundant (≥ ±1.5 FC, raw-$p \leq 0.05$) between HG and LG cultured BeWo CT cells. A list of all differentially abundant metabolites between LG and HG-cultured BeWo CT cells is available in S6 Table in S2 File.
Differentially abundant metabolites were subsequently visualized via volcano plot (±1.5 FC, $p \leq 0.05$, Fig 10) and heat map (S9 Fig in S1 File). Additionally, the degree of differences in metabolite profiles between HG and LG-cultured BeWo CT cells was visualized by unsupervised principal component analysis (PCA) 2D plot as well as supervised partial least squares discriminant analysis (PLS-DA) scores plot (S10A and S10B Fig in S1 File).
**Fig 10:** *Visualization of differentially abundant metabolites in HG-cultured BeWo CT cells.A volcano plot was constructed to visualize the differentially abundant metabolites in HG-cultured BeWo CT cells ((≥ ±1.5-fold change, p<0.05, n = 5/group). Overall, 720 metabolites were found to be differentially abundant in HG-cultured BeWo CT cells. Of these metabolites, 4 were identified in tier 1 (all increased in abundance) and 89 were identified in tier 2 (22 decreased in abundance, and 67 increased in abundance) with high confidence. The x-axis indicates log2(fold-change) vs LG BeWo CTs, and the y-axis indicates the p-value (-log10). The red dots represent significantly increased metabolites, and the blue dots represent significantly decreased metabolites in HG-cultured BeWo CT cells.*
The KEGG database pathways involving: glycolysis/gluconeogenesis (associated with significantly increased lactate (+2.72 FC) levels, and significantly decreased acetaldehyde (+2.72 FC) levels); pyruvate metabolism (associated with increased lactate (+2.72 FC) levels, and decreased acetaldehyde (-1.20 FC) levels); drug metabolism–cytochrome P450 (associated with increased 2-hydroxyiminostilbene (+2.52 FC) and 3-carbamoyl-2-phenylpropionic acid (+5.42 FC) levels); ascorbate and aldarate metabolism (associated with increased L-gulonolactone (+2.46 FC) levels); riboflavin metabolism (associated with increased riboflavin (+3.68 FC) levels); fatty acid biosynthesis (associated with increased malonate (+3.74 FC) levels); as well as synthesis and degradation of ketone bodies (associated with increased (R)-3-hydroxybutanoate (+1.60 FC) levels) were identified to be significantly enriched in HG-cultured BeWo CT cells (FDR-corrected $p \leq 0.05$, Fig 11A).
**Fig 11:** *Pathway analysis of metabolites identified in tiers 1 and 2.Identified peak pairs from tiers 1 and 2 were imported into MetaboAnalyst v5.0 for analysis of enriched KEGG and Small Molecule Pathway Database (SMPDB) pathways. A scatterplot was created to visualize identified pathways with pathway impact (ratio of identified metabolites to total metabolites in the pathway) on the x-axis and p-value (-log10) on the y-axis. Each individual node represents a unique KEGG pathway, with the node size corresponding to pathway impact and node colour corresponding to significance level. Pathways with a false discovery rate p<0.05 were determined to be significant and were labelled with roman numerals. (A) The KEGG database pathways involving: (i) Glycolysis/Gluconeogenesis, (ii) Pyruvate Metabolism, (iii) Drug Metabolism–Cytochrome P450, (iv) Ascorbate and Aldarate Metabolism, (v) Riboflavin Metabolism, (vi) Fatty Acid Biosynthesis, and (vii) Synthesis and Degradation of Ketone Bodies pathways were significantly enriched in HG-cultured BeWo CT cells. (B) The SMPDB pathways involving: (i) the Glycerol Phosphate Shuttle; (ii) Glycerolipid Metabolism; and (ii) Riboflavin Metabolism were identified to be significantly enriched in the HG-cultured BeWo CT cells.*
The SMPDB pathways involving: the Glycerol Phosphate Shuttle associated with significantly increased p-Benzoquinone (+3.78 FC) levels); Glycerolipid Metabolism (associated with significantly increased p-Benzoquinone (+3.78 FC) levels); and Riboflavin Metabolism (associated with significantly increased riboflavin (+3.68 FC) levels) were identified to be significantly enriched in the HG-cultured BeWo CT cells (FDR-corrected $p \leq 0.05$; Fig 11B).
## Integration of HG-cultured BeWo cytotrophoblast transcriptome and metabolome signatures
Differentially abundant metabolites and DEGs were pooled into queries for KEGG pathway enrichment analysis. The KEGG Glycerolipid metabolism pathway was enriched in the HG-cultured BeWo CT cell datasets, however only differentially expressed genes (LIPF (+1.30 FC); DGKG (+1.31 FC); and GPAT2 (-1.37 FC)), and not differentially abundant metabolites were found to be involved in this pathway (Fig 12; raw-$p \leq 0.05$).
**Fig 12:** *Joint pathway analysis integration of HG-cultured BeWo CT cell transcriptome and metabolome profiles.Differentially expressed gene and differentially abundant metabolite lists were imported into the MetaboAnalyst v5.0 Joint Pathway Analysis tool for transcriptome-metabolome integration analysis. The differentially expressed gene and differentially abundant metabolite lists were pooled into a single query for KEGG pathway over-representation analysis. A scatterplot was created to visualize identified pathways with pathway impact (ratio of identified metabolites to total metabolites in the pathway) on the x-axis and p-value (-log10) on the y-axis. Each individual node represents a unique KEGG pathway, with the node size corresponding to pathway impact and node colour corresponding to significance level. The (i) Glycerolipid metabolism pathway was found to be enriched in the HG-cultured BeWo CT cells, although this enrichment was only associated with differential gene expression, and not with simultaneous altered metabolite levels (raw-p <0.05).*
## Discussion
This study aimed to expand upon current published literature [22,30,31,34] and more thoroughly explore the independent impacts of hyperglycemia on placental villous trophoblasts by characterizing nutrient storage and mitochondrial respiratory activity in BeWo trophoblasts following a relatively prolonged 72-hour HG-exposure (25 mM glucose). While previous studies utilizing BeWo trophoblasts have primarily highlighted the impacts of high glucose exposure on undifferentiated CT cells [30,31,34] the current study is strengthened through the combined examination of both undifferentiated CT cells and differentiated SCT cells. The more chronic 72-hour culture protocol as utilized in this study also allowed for exposure of villous trophoblast cell populations to hyperglycemia prior to and during differentiation, analogous to in vivo villous trophoblast layer development whereby progenitor CT cells are pre-exposed to dietary nutrients prior to and during syncytialization. More importantly, the specific use of functional readouts of metabolic and mitochondrial activity (including the Seahorse XF Mito and Glycolysis Stress Tests) in this study provided a more in-depth insight into the real-time metabolic function of live CT and SCT cells following a prolonged glucose challenge. Subsequently, the current study utilized a multi-omics research approach and described transcriptomic and metabolomic signatures of HG-exposed BeWo CT cells, allowing a thorough characterization of the glucose-mediated alterations to placental metabolism. Overall, the data presented in this study demonstrated that a 72-hour exposure to hyperglycemia, increased glycogen and triglyceride stores as well as modulated the transcriptomic and metabolomic signatures as well of BeWo trophoblast cells. However, impairments in functional readouts of BeWo trophoblast mitochondrial respiration were not observed following 72-hours of HG-exposure.
## Hyperglycemia and nutrient stores in BeWo trophoblasts
As previously highlighted, DM during pregnancy is associated with aberrant nutrient storage in the villous trophoblast layer of the placenta [18–23]. Our results demonstrated that HG-culture conditions directly promoted increased glycogen content in both BeWo CT and SCT cells, and additionally, that BeWo CT cells have a greater glycogen storage potential than SCT cells. It is important to note that these differentiation state-dependent trends in glycogen content are consistent with previous reports from primary human placental tissue which described that glycogen storage predominately occurs within the CT cells of the placenta [24,50,51]. Increased accumulation of glycogen stores in diabetic placentae has been thought to be a mechanism by which the placenta limits materno-fetal glucose transfer in times of nutrient overabundance to limit fetal overgrowth [24,25]. However, high-glucose exposure in both BeWo CT and SCT cells was also associated with reduced relative protein abundance of glycogen synthase as well as with increased inhibitory Ser641 phosphorylation [52] of glycogen synthase. As there were no glucose-mediated differences in the protein abundance of GSK3β or the inhibitory Ser9 phosphorylation [53] of GSK3β, the increased inhibitory phosphorylation of glycogen synthase in HG-cultured BeWo trophoblasts likely occurs via a GSK3β-independent mechanism.
Overall, these data highlighted that following 72-hours of hyperglycemia BeWo trophoblast cells, despite having elevated stores of glycogen, have a diminished capacity to continue to store excess glucose as glycogen. The changes could reflect the presence of a negative feedback mechanism that acts to prevent excessive glycogen accumulation in placental villous trophoblasts, similar to the feedback mechanisms described in skeletal muscle cells [54]. Excessive glycogen levels have previously been associated with liver inflammation in rodents [55] and reduced lifespan in Caenorhabditis elegans [56], which suggests that glycogen accumulation is damaging. Therefore, the diminished glycogen storage capacity in placental trophoblasts in response to prolonged high glucose could act in a protective manner. However, these changes in glycogen storage dynamics in placental trophoblasts following sustained HG-exposure could also lead to increased materno-fetal glucose transport. This may be an important alteration to placental glucose processing that helps facilitate the development of fetal macrosomia in diabetic pregnancies [57].
In addition to increased glycogen stores, our study also demonstrated that BeWo CT cells, but not SCT cells, have increased triglyceride content in response to excess glucose exposure. These data were consistent with previous reports that have highlighted an increased accumulation of triglyceride species in placental explants from healthy pregnancies under high glucose conditions [22]. These glucose-mediated changes in trophoblast triglyceride abundance will likely modulate materno-fetal nutrient transfer and could have downstream impacts on fetal growth and development. The current study further identified that there were no differentiation state-dependent differences in triglyceride abundance in BeWo trophoblast cells, similar to what has been reported in freshly isolated villous trophoblast samples [58]. Previous readouts from primary placenta samples have highlighted that lipid esterification processes occur primarily within villous CT cells, and it has been suggested that any lipid droplets present in primary SCT cells are remnants from esterification process that occurred prior to syncytialization [58]. We speculate that a similar reduction in esterification activity also occurs in BeWo SCT cells and may explain the absence of a HG-mediated difference in triglyceride content in our BeWo SCT cells. Future studies utilizing functional readouts of lipid esterification activity may be needed to better elucidate the mechanisms underlying differences in TG abundance between undifferentiated BeWo CT cells and differentiated BeWo SCT cells.
## Hyperglycemia and metabolic function in BeWo trophoblasts
The current study demonstrated that cellular mitochondrial respiratory activity (measured by the Seahorse XF Mito Stress Test) as well as that the activities of ETC complexes I and II are not impacted in BeWo trophoblasts cultured under hyperglycemia at 72H. Additional analysis of metabolic function via the Seahorse Glycolysis Stress Test and activity assays for LDH and CS enzymes further indicated that isolated hyperglycemia does not impact functional aspects of metabolism in BeWo trophoblasts. Overall, our data suggests that hyperglycemia alone may not directly facilitate the impairments in mitochondrial respiratory activity that have previously been observed in DM-exposed primary placental samples [27–29].
Previous studies in other cell preparations, however, have demonstrated that the impacts of HG-culture conditions on mitochondrial respiratory function are dependent on the length of high-glucose exposure. For example, human kidney tubule (HK-2) cells have displayed reduced basal and maximal mitochondrial respiratory activity only when cultured under HG-conditions (25 mM) for at least 4 days [59]. Further, mitochondrial respiratory activity in kidney glomerular (HMC) cells was not impacted prior to 8 days of high glucose exposure [59]. Likewise, mitochondrial activity of human umbilical cord endothelial (EA.hy926) cells was not impaired after 3 days of high glucose (25 mM) treatment, but was reduced after 6 days of HG-culture conditions, and this impairment in mitochondrial function was sustained through 9 days of high glucose exposure [60]. Overall, these studies suggest that there are time-course dependent factors that influence whether hyperglycemia impacts cellular mitochondrial respiratory activity. Thus, the conclusions of the current study may be limited due to the single timepoint utilized for all analyses of metabolic function. It is possible that prolonging HG-culture conditions in BeWo trophoblasts could ultimately lead to impaired mitochondrial function, as was observed in HK-2, EA.hy926, and HMC cells. The impacts of a longer duration of high glucose treatment, along with sequential sampling over a time course, in BeWo cells may need to be explored in future investigations.
In addition, previous research has highlighted that hyperglycemia negatively regulates mitochondrial function in some cell types by altering mitochondrial fusion (regulated by OPA1) and fission (regulated by DRP1) dynamics leading increased mitochondrial fractionation [61–64]. In the current study, we observed a significant increase in the pSER616 DRP1-to-total DRP1 ratio in HG-cultured BeWo trophoblasts. This post-translational modification of DRP1 is associated with increased mitochondrial translocation of DRP1 that promotes increased mitochondrial fission [65]. These underlying changes in mitochondrial physiology are important, as increased mitochondrial fission has been associated with increased cellular oxidative stress, and impaired insulin sensitivity, and ultimately mitochondrial respiratory dysfunction [66,67]. Further, these data may indicate that underlying aspects of trophoblast mitochondrial function are negatively regulated by hyperglycemia, despite concurrent our observation that global, real-time measures of BeWo mitochondrial respiratory activity were not impacted. Further, we speculate this could indicate that HG-exposed BeWo trophoblasts are transitioning towards mitochondrial dysfunction mediated by increased mitochondrial fission, comparable to the outcomes reported in skeletal muscle following high fat exposure [66].
While functional global readouts of BeWo trophoblast mitochondrial function were not impacted in response to hyperglycemia, we did observe impaired mitochondrial respiratory activity in differentiated BeWo SCT cells. Syncytialization of BeWo trophoblasts was associated with reduced mitochondrial spare respiratory capacity, reduced coupling efficiency, concomitant with reduced activity of ETC complex II and reduced protein expression of ETC complex IV. Furthermore, BeWo SCT cells displayed increased DRP1 protein abundance, an increased pSER616 DRP1-to-total DRP1 ratio, and decreased OPA1 protein abundance suggestive of increased mitochondrial fractionation. These data may indicate that alterations in mitochondrial dynamics underlies the observed functional differences in mitochondrial respiration between BeWo CT and SCT cells. Previous work by our research group has likewise highlighted that BeWo CT cells are overall more metabolically active than BeWo SCT cells [37], and similar trends have been reported in cultured PHT cells [28,68]. As undifferentiated BeWo CT cells display greater metabolic activity and greater alterations to nutrient stores in response to hyperglycemia than in differentiated SCT cells, we speculated that alterations in transcriptome and metabolome profiles in response to high glucose culture conditions would be more prevalent in these progenitor cells. Thus, the current study sought to examine global gene expression as well as global metabolite abundance solely in HG-cultured BeWo CT cells to further elucidate mechanisms underlying altered placental metabolic function in response to hyperglycemia.
## Transcriptomic analysis of HG-cultured BeWo CT cells
Overall, we identified 197 differentially expressed genes (122 up-regulated, and 75 down-regulated; ≥ ±1.3 FC) in BeWo CT cells cultured under hyperglycemia for 72H. However, previous reports in BeWo CT cells demonstrated more substantial variations in gene expression (>5000 DEGs) between BeWo trophoblasts cultured under similar high and low glucose conditions [30]. The differences between the current study and previous reports may be due in part to differences in study design (pooling samples for arrays vs independent arrays for each sample), cell media formulation (DMEM-F12 vs F12K) as well as length of hyperglycemic exposure (48H vs 72H). Despite these differences in the number of differentially expressed genes, our study did align with this previous transcriptomic report, and further demonstrated that elevated glucose levels impact the expression of genes involved in functional pathways involving metabolic processes [30].
Specifically in the current study, using a Gene Set Enrichment Analysis (GSEA) approach, we highlighted an enrichment in metabolic functional and molecular pathways relating to glutathione metabolism (associated with increased LAP3 expression), fatty acid synthesis (associated with increased ACSL1 expression), CD209 signaling (associated with increased RPS6KA5 expression), and oxygen binding (associated with reduced CYP2F1 expression) in HG-cultured BeWo CT cells. Additionally, we validated the differential expression of genes involved in glucose metabolism (reduced HK2 expression) and glucocorticoid metabolism (increased HSD11B2 expression) in our high-glucose exposed cultures.
Of particular interest was the increased mRNA expression of ASCL1 in BeWo CT cells, a gene involved in functional gene sets associated with fatty acid and lipid synthesis. Previous studies have highlighted that ACSL1 is involved in lipid synthesis in various tissues and that knockdown of ACSL1 is associated with reduced triglyceride and lipid droplet abundance [69–72]. More importantly, cells transfected to overexpress ACSL1 have been found to have increased triglyceride accumulation [72–74]. We speculate that ACSL1 is also important in the trafficking of lipid species to lipid droplets in trophoblast cells and the increased expression of this transcript may underlie the glucose-induced accumulation of triglyceride species that was observed in our BeWo CT cultures. Thus, ACSL1 may be a regulator of trans-placental lipid transport, and its altered expression in response to elevated glucose could impact fetal growth by reducing materno-fetal lipid transport. In contrast, the observed reduction in Hexokinase-2 (HK2, the first enzyme in the placental glycolysis metabolic pathway [27]) transcript abundance could indicate that glycolysis is down-regulated in trophoblasts exposed to hyperglycemia. The reduced expression of glycolytic enzymes combined with a reduced capacity for glycogen storage (reduced glycogen synthase expression and increased inhibitory phosphorylation of glycogen synthase) could facilitate increased materno-fetal glucose transport. The overall impacts of hyperglycemic exposure on transplacental nutrient transport will need to be examined in more depth in future experiments (perhaps through the use of trans well culture systems) to elucidate how fetal nutrient delivery is modulated in response to these changes in placental nutrient handling.
The results from the current study additionally demonstrated that hyperglycemia is an important regulator of the expression of placental HSD11B2, an enzyme involved in the metabolism and inactivation of the glucocorticoid cortisol. In normal pregnancies, cortisol is thought to be involved in mediating the physiological increase in maternal insulin resistance that is necessary to support fetal growth [75,76], however, circulating cortisol levels may be pathologically elevated in some GDM pregnancies [77]. Interestingly, placentae from GDM pregnancies have been found to have increased expression of HSD11B2 leading to increased cortisol inactivation [78]. As, elevated cortisol levels in fetal circulation have been associated with impaired brain development processes [79], increased placental HSD11B2 expression in GDM pregnancies may act in a protective manner to limit fetal glucocorticoid exposures. However, as cortisol also activates the glucocorticoid receptor leading to modulation of gene transcription [80,81], alterations in placental cortisol metabolism may have downstream consequences on placental function through altering gene expression. Overall, it remains poorly understood whether these glucose-mediated impacts to placental cortisol metabolism act in a protective or detrimental manner.
These transcriptomic readouts may also highlight underlying alterations to stress responses in BeWo trophoblast cells exposed to hyperglycemia. For example, we observed a significant up-regulation in the CD209 (DC-SIGN) signaling pathway in GSEA analysis. Although CD209 has been reported to be low in placental chorionic villi [82], its activation has previously been associated with increased inflammatory signaling in endothelial cells [83] as well as increased inflammation and insulin resistance in adipocytes [84,85]. Additionally, the validated DEG involved in this pathway, RPS6KA5, encodes the mitogen-and-stress activated kinase 1 (MSK-1) protein that is highly expressed in placental trophoblasts [86,87]. Interestingly, MSK1 had previously been found to have an anti-inflammatory function via increasing the expression of the anti-inflammatory cytokine, IL-10 [88,89]. Overall, future studies are still needed to elucidate the global impacts of hyperglycemia on BeWo trophoblast inflammatory signaling pathways. Further, we observed an increased expression of LAP3 (involved in glutathione metabolism), a target that has previously found to be positively correlated with fasting blood glucose levels and has, by inhibiting autophagy, been implicated in the progression of Non-Alcoholic Fatty Liver Disease [90]. LAP3 has also been found to degrade glutathione (GSH), an important cellular antioxidant enzyme [91,92]. Overall, these data highlight an increased risk of oxidative stress development in hyperglycemia-exposed BeWo trophoblasts. As mitochondria are a predominant generator of Reactive Oxygen Species, a reduction in glutathione (GSH) availability may promote mitochondrial oxidative damage as well as increased mitochondrial fission, ultimately leading to mitochondrial dysfunction in high-glucose exposed trophoblasts, as reported to occur in other model systems [32,66,67].
## Metabolomic analysis of HG-cultured BeWo CT cells
Multivariate analysis of BeWo trophoblast metabolome profiles highlighted a divergence in metabolite signatures between HG and LG-cultured BeWo CT cells suggesting that high glucose levels also impact the levels of small, polar metabolites in the placenta. Subsequent pathway analysis highlighted specific intracellular accumulations of lactate (involved in the glycolysis/gluconeogenesis and pyruvate metabolism pathways), malonate (involved in the fatty acid biosynthesis pathway), as well as riboflavin (involved in the riboflavin metabolism pathway) in the HG-cultured BeWo CT cells.
The observed lactate accumulation likely suggests that glycolytic flux is in fact increased in HG-cultured BeWo CT cells, despite our concomitant observation of reduced HK2 expression [93]. It is interesting to note that this study did not observe increased basal or maximal glycolytic activity when assessed via the Seahorse XF Glycolysis Stress Test. As the Glycolysis Stress Test utilizes extracellular media acidification (resulting from the co-export of lactate and H+ from the cell) as a proxy measurement of glycolytic activity, this functional assay may underestimate glycolytic activity in the event of reduced lactate export as could potentially occur in a “Cytosol-to-Mitochondrial Lactate Shuttle” metabolic pathway [93].
Increased malonate levels may reflect an increase in de novo lipogenesis via FASN in HG-cultured BeWo CT and may be another mechanism underlying the increased TG levels observed in HG-cultured BeWo CT cells [94]. Future studies and the use of radio-labelled metabolites may be required to further characterize glycolytic activity and de novo lipogenesis in high-glucose exposed BeWo trophoblasts.
The accumulation of riboflavin (the essential vitamin B2) in HG-cultured BeWo CT cells either reflects an increased cellular uptake of riboflavin or an inhibition of riboflavin metabolism into the cofactors flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) [95]. Reduced metabolism of riboflavin to its cofactor intermediaries has previously been implicated in the development of mitochondrial dysfunction [96]. However, riboflavin has also previously been suggested to act as an antioxidant [97] and has been demonstrated to be beneficial in reducing oxidative stress in rodent models of DM [98,99]. Future investigations may be needed to assess the impacts of riboflavin accumulation in HG-cultured trophoblasts and elucidate whether accumulation of this vitamin is beneficial or harmful to the placenta.
Interestingly, our integrated over-representation analysis combining DEGs and differentially abundant metabolites was not associated with an enrichment in any KEGG data sets with concomitant changes in both transcripts and metabolites. These data highlight limited overlap between the transcriptome and metabolome in high glucose treated trophoblasts and could indicate that BeWo trophoblast metabolism is primarily regulated by post-translational rather than transcriptional mechanisms. However, as BeWo placental trophoblasts are a male cell line, our data may also align with previous reports in human pregnancies that have found male placentae display relatively few adaptations to nutrient stresses in vivo compared to female placentae [82,83]. The presence and importance of these sex-dependent responses in placental trophoblasts likely highlights a limitation in the current study, and thus, the data reported here may only be predictive and representative of placental responses in pregnancies with a male fetus. Future studies are still needed to elucidate the differences in molecular mechanisms and responses between male and female placental trophoblasts in response to elevated glucose.
Finally, it is important to highlight that the choriocarcinoma origin of BeWo trophoblasts may limit the ability of our model culture system to fully represent in vivo placentation during hyperglycemia. While these data are a useful first outlook into the impacts of an isolated and prolonged in vitro hyperglycemic exposure on both progenitor CT cells, and differentiated SCT cells, future works utilizing primary-based placental material (including explant, as well as continuously improving trophoblast stem cell and organoid culture systems) are still needed to fully characterize villous trophoblast responses to high glucose.
## Conclusion
The results of the current study highlighted that a 72-hour hyperglycemia exposure independently impacts metabolic function and nutrient storage in BeWo trophoblasts, but does not mediate any global changes in functional readouts of mitochondrial respiratory activity. However, HG-cultured BeWo trophoblasts displayed markers suggestive of increased mitochondrial fission, as well as reduced antioxidant capacity, and we speculate this may highlight these HG-exposed cells are ultimately transitioning towards future mitochondrial dysfunction and failure. Notably, while the 72-hour hyperglycemic exposure utilized in the current study is relatively prolonged in the setting of in vitro cell culture experiments, this timeline is acute in comparison to the 40-week duration of human pregnancy in vivo. Thus, the data from the current study may highlight that preventing even short periods of hyperglycemia is of great importance in the clinical management of diabetic pregnancies. Maintaining appropriate glycemic management during gestation will help to ensure that placental mitochondrial function, and is sustained throughout gestation, and in turn limit the risk of the offspring developing later life non-communicable metabolic diseases.
## References
1. Feig DS, Hwee J, Shah BR, Booth GL, Bierman AS, Lipscombe LL. **Trends in Incidence of Diabetes in Pregnancy and Serious Perinatal Outcomes: A Large, Population-Based Study in Ontario, Canada, 1996–2010**. *Diabetes Care* (2014.0) **37** 1590-1596. DOI: 10.2337/dc13-2717
2. Behboudi-Gandevani S, Amiri M, Bidhendi Yarandi R, Ramezani Tehrani F. **The impact of diagnostic criteria for gestational diabetes on its prevalence: a systematic review and meta-analysis**. *Diabetol Metab Syndr* (2019.0) **11** 11. DOI: 10.1186/s13098-019-0406-1
3. Lawrence RL, Wall CR, Bloomfield FH. **Prevalence of gestational diabetes according to commonly used data sources: an observational study**. *BMC Pregnancy Childbirth* (2019.0) **19** 349. DOI: 10.1186/s12884-019-2521-2
4. Voaklander B, Rowe S, Sanni O, Campbell S, Eurich D, Ospina MB. **Prevalence of diabetes in pregnancy among Indigenous women in Australia, Canada, New Zealand, and the USA: a systematic review and meta-analysis.**. *Lancet Glob Health.* (2020.0) **8** e681-e698. DOI: 10.1016/S2214-109X(20)30046-2
5. Clausen TD, Mathiesen ER, Hansen T, Pedersen O, Jensen DM, Lauenborg J. **High Prevalence of Type 2 Diabetes and Pre-Diabetes in Adult Offspring of Women With Gestational Diabetes Mellitus or Type 1 Diabetes: The role of intrauterine hyperglycemia**. *Diabetes Care* (2008.0) **31** 340-346. DOI: 10.2337/dc07-1596
6. Farrell T, Neale L, Cundy T. **Congenital anomalies in the offspring of women with Type 1, Type 2 and gestational diabetes**. *Diabetic Medicine* (2002.0) **19** 322-326. DOI: 10.1046/j.1464-5491.2002.00700.x
7. Aisa MC, Cappuccini B, Barbati A, Clerici G, Torlone E, Gerli S. **Renal Consequences of Gestational Diabetes Mellitus in Term Neonates: A Multidisciplinary Approach to the DOHaD Perspective in the Prevention and Early Recognition of Neonates of GDM Mothers at Risk of Hypertension and Chronic Renal Diseases in Later Life.**. *J Clin Med* (2019.0) **8** 429. DOI: 10.3390/jcm8040429
8. Pereira TJ, Moyce BL, Kereliuk SM, Dolinsky VW. **Influence of maternal overnutrition and gestational diabetes on the programming of metabolic health outcomes in the offspring: experimental evidence**. *Biochemistry and Cell Biology* (2015.0) **93** 438-451. DOI: 10.1139/bcb-2014-0141
9. Lindsay RS, Nelson SM, Walker JD, Greene SA, Milne G, Sattar N. **Programming of Adiposity in Offspring of Mothers With Type 1 Diabetes at Age 7**. *Years. Diabetes Care* (2010.0) **33** 1080-1085. DOI: 10.2337/dc09-1766
10. Damm P, Houshmand-Oeregaard A, Kelstrup L, Lauenborg J, Mathiesen ER, Clausen TD. **Gestational diabetes mellitus and long-term consequences for mother and offspring: a view from Denmark**. *Diabetologia* (2016.0) **59** 1396-1399. DOI: 10.1007/s00125-016-3985-5
11. Dabelea D.. **The Predisposition to Obesity and Diabetes in Offspring of Diabetic Mothers**. *Diabetes Care* (2007.0) **30** S169-S174. DOI: 10.2337/dc07-s211
12. Pathirana MM, Lassi ZS, Ali A, Arstall MA, Roberts CT, Andraweera PH. **Association between metabolic syndrome and gestational diabetes mellitus in women and their children: a systematic review and meta-analysis**. *Endocrine* (2021.0) **71** 310-320. DOI: 10.1007/s12020-020-02492-1
13. Huynh J, Dawson D, Roberts D, Bentley-Lewis R. **A systematic review of placental pathology in maternal diabetes mellitus**. *Placenta* (2015.0) **36** 101-114. DOI: 10.1016/j.placenta.2014.11.021
14. Carrasco-Wong I, Moller A, Giachini FR, Lima VV, Toledo F, Stojanova J. **Placental structure in gestational diabetes mellitus**. *Biochimica et Biophysica Acta (BBA)—Molecular Basis of Disease.* (2020.0) **1866** 165535. DOI: 10.1016/j.bbadis.2019.165535
15. Strøm-Roum EM, Haavaldsen C, Tanbo TG, Eskild A. **Placental weight relative to birthweight in pregnancies with maternal diabetes mellitus**. *Acta Obstet Gynecol Scand* (2013.0) **92** 783-789. DOI: 10.1111/aogs.12104
16. Taricco E, Radaelli T, Nobile de Santis MS, Cetin I. **Foetal and Placental Weights in Relation to Maternal Characteristics in Gestational Diabetes**. *Placenta* (2003.0) **24** 343-347. DOI: 10.1053/plac.2002.0913
17. Lao TT, Lee C-P, Wong W-M. **Placental weight to birthweight ratio is increased in mild gestational glucose intolerance**. *Placenta* (1997.0) **18** 227-230. DOI: 10.1016/s0143-4004(97)90097-7
18. Gheorman V, Gheorman L, Ivănuş C, Pană RC, Gogănău AM, Pătraşcu A. **Comparative study of placenta acute fetal distress and diabetes associated with pregnancy**. *Rom J Morphol Embryol* (2013.0) **54** 505-11. PMID: 24068397
19. Desoye G, Hofmann HH, Weiss PAM. **Insulin binding to trophoblast plasma membranes and placental glycogen content in well-controlled gestational diabetic women treated with diet or insulin, in well-controlled overt diabetic patients and in healthy control subjects**. *Diabetologia* (1992.0) **35** 45-55. DOI: 10.1007/BF00400851
20. Akison LK, Nitert MD, Clifton VL, Moritz KM, Simmons DG. **Review: Alterations in placental glycogen deposition in complicated pregnancies: Current preclinical and clinical evidence**. *Placenta* (2017.0) **54** 52-58. DOI: 10.1016/j.placenta.2017.01.114
21. Shafrir E, Barash V. **Placental glycogen metabolism in diabetic pregnancy**. *Isr J Med Sci* **27** 449-61. PMID: 1835720
22. Hulme CH, Nicolaou A, Murphy SA, Heazell AEP, Myers JE, Westwood M. **The effect of high glucose on lipid metabolism in the human placenta**. *Sci Rep* (2019.0) **9** 14114. DOI: 10.1038/s41598-019-50626-x
23. Lindegaard MLS, Damm P, Mathiesen ER, Nielsen LB. **Placental triglyceride accumulation in maternal type 1 diabetes is associated with increased lipase gene expression**. *J Lipid Res* (2006.0) **47** 2581-2588. DOI: 10.1194/jlr.M600236-JLR200
24. Tunster SJ, Watson ED, Fowden AL, Burton GJ. **Placental glycogen stores and fetal growth: insights from genetic mouse models**. *Reproduction* (2020.0) **159** R213-R235. DOI: 10.1530/REP-20-0007
25. Desoye G, Korgun ET, Ghaffari-Tabrizi N, Hahn T. **Is fetal macrosomia in adequately controlled diabetic women the result of a placental defect?–a hypothesis. The Journal of Maternal-Fetal &**. *Neonatal Medicine* (2002.0) **11** 258-261. DOI: 10.1080/jmf.11.4.258.261
26. Sobrevia L, Valero P, Grismaldo A, Villalobos-Labra R, Pardo F, Subiabre M. **Mitochondrial dysfunction in the fetoplacental unit in gestational diabetes mellitus.**. *Biochimica et Biophysica Acta (BBA)—Molecular Basis of Disease.* (2020.0) **1866** 165948. DOI: 10.1016/j.bbadis.2020.165948
27. Muralimanoharan S, Maloyan A, Myatt L. **Mitochondrial function and glucose metabolism in the placenta with gestational diabetes mellitus: role of miR-143**. *Clin Sci* (2016.0) **130** 931-41. DOI: 10.1042/CS20160076
28. Valent AM, Choi H, Kolahi KS, Thornburg KL. **Hyperglycemia and gestational diabetes suppress placental glycolysis and mitochondrial function and alter lipid processing**. *The FASEB Journal* (2021.0) 35. DOI: 10.1096/fj.202000326RR
29. Hastie R, Lappas M. **The effect of pre-existing maternal obesity and diabetes on placental mitochondrial content and electron transport chain activity**. *Placenta* (2014.0) **35** 673-83. DOI: 10.1016/j.placenta.2014.06.368
30. Hulme CH, Stevens A, Dunn W, Heazell AEP, Hollywood K, Begley P. **Identification of the functional pathways altered by placental cell exposure to high glucose: lessons from the transcript and metabolite interactome**. *Sci Rep* (2018.0) **8** 5270. DOI: 10.1038/s41598-018-22535-y
31. He M, Wang G, Han S, Jin Y, Li H, Wu X. **Nrf2 signalling and autophagy are involved in diabetes mellitus-induced defects in the development of mouse placenta**. *Open Biol* (2016.0) **6** 160064. DOI: 10.1098/rsob.160064
32. Guo C, Sun L, Chen X, Zhang D. **Oxidative stress, mitochondrial damage and neurodegenerative diseases.**. *Neural Regen Res* (2013.0) **8** 2003-14. DOI: 10.3969/j.issn.1673-5374.2013.21.009
33. Kowaltowski AJ, Vercesi AE. **Mitochondrial damage induced by conditions of oxidative stress**. *Free Radic Biol Med* (1999.0) **26** 463-471. DOI: 10.1016/s0891-5849(98)00216-0
34. Weiss U, Cervar M, Puerstner P, Schmut O, Haas J, Mauschitz R. **Hyperglycaemia in vitro alters the proliferation and mitochondrial activity of the choriocarcinoma cell lines BeWo, JAR and JEG-3 as models for human first-trimester trophoblast**. *Diabetologia* (2001.0) **44** 209-219. DOI: 10.1007/s001250051601
35. Joseph S, Walejko JM, Zhang S, Edison AS, Keller-Wood M. **Maternal hypercortisolemia alters placental metabolism: a multiomics view**. *American Journal of Physiology-Endocrinology and Metabolism* (2020.0) **319** E950-E960. DOI: 10.1152/ajpendo.00190.2020
36. Stuart TJ, O’Neill K, Condon D, Sasson I, Sen P, Xia Y. **Diet-induced obesity alters the maternal metabolome and early placenta transcriptome and decreases placenta vascularity in the mouse†**. *Biol Reprod* (2018.0) **98** 795-809. DOI: 10.1093/biolre/ioy010
37. Easton ZJW, Delhaes F, Mathers K, Zhao L, Vanderboor CMG, Regnault TRH. **Syncytialization and prolonged exposure to palmitate impacts BeWo respiration**. *Reproduction* (2021.0) **161** 73-88. DOI: 10.1530/REP-19-0433
38. Kusama K, Bai R, Imakawa K. **Regulation of human trophoblast cell syncytialization by transcription factors STAT5B and NR4A3**. *J Cell Biochem* (2018.0) **119** 4918-4927. DOI: 10.1002/jcb.26721
39. Malhotra SS, Suman P, Kumar Gupta S. **Alpha or beta human chorionic gonadotropin knockdown decrease BeWo cell fusion by down-regulating PKA and CREB activation**. *Sci Rep* (2015.0) **5** 11210. DOI: 10.1038/srep11210
40. Price ER, McFarlan JT, Guglielmo CG. **Preparing for Migration? The Effects of Photoperiod and Exercise on Muscle Oxidative Enzymes, Lipid Transporters, and Phospholipids in White‐Crowned Sparrows**. *Physiological and Biochemical Zoology* (2010.0) **83** 252-262. DOI: 10.1086/605394
41. Mootha VK, Lindgren CM, Eriksson K-F, Subramanian A, Sihag S, Lehar J. **PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes**. *Nat Genet* (2003.0) **34** 267-273. DOI: 10.1038/ng1180
42. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA. **Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles**. *Proceedings of the National Academy of Sciences* (2005.0) **102** 15545-15550. DOI: 10.1073/pnas.0506580102
43. Liao Y, Wang J, Jaehnig EJ, Shi Z, Zhang B. **WebGestalt 2019: gene set analysis toolkit with revamped UIs and APIs**. *Nucleic Acids Res* (2019.0) **47** W199-W205. DOI: 10.1093/nar/gkz401
44. Khan MI, Dębski KJ, Dabrowski M, Czarnecka AM, Szczylik C. **Gene set enrichment analysis and ingenuity pathway analysis of metastatic clear cell renal cell carcinoma cell line**. *American Journal of Physiology-Renal Physiology* (2016.0) **311** F424-F436. DOI: 10.1152/ajprenal.00138.2016
45. Guo K, Li L. **Differential 12 C-/ 13 C-Isotope Dansylation Labeling and Fast Liquid Chromatography/Mass Spectrometry for Absolute and Relative Quantification of the Metabolome**. *Anal Chem* (2009.0) **81** 3919-3932. DOI: 10.1021/ac900166a
46. Zhao S, Li L. **Chemical derivatization in LC-MS-based metabolomics study**. *TrAC Trends in Analytical Chemistry* (2020.0) **131** 115988. DOI: 10.1016/j.trac.2020.115988
47. Zhao S, Li H, Han W, Chan W, Li L. **Metabolomic Coverage of Chemical-Group-Submetabolome Analysis: Group Classification and Four-Channel Chemical Isotope Labeling LC-MS**. *Anal Chem* (2019.0) **91** 12108-12115. DOI: 10.1021/acs.analchem.9b03431
48. Lew M.. **Good statistical practice in pharmacology Problem 2**. *Br J Pharmacol* (2009.0) **152** 299-303. DOI: 10.1038/sj.bjp.0707372
49. Hahn T, Hahn D, Blaschitz A, Korgun ET, Desoye G, Dohr G. **Hyperglycaemia-induced subcellular redistribution of GLUT1 glucose transporters in cultured human term placental trophoblast cells**. *Diabetologia* (2000.0) **43** 173-180. DOI: 10.1007/s001250050026
50. Georgiades P, Ferguson-Smith AC, Burton GJ. **Comparative Developmental Anatomy of the Murine and Human Definitive Placentae**. *Placenta* (2002.0) **23** 3-19. DOI: 10.1053/plac.2001.0738
51. Wislocki GB, Bennett HS. **The histology and cytology of the human and monkey placenta, with special reference to the trophoblast**. *American Journal of Anatomy* (1943.0) **73** 335-449. DOI: 10.1002/aja.1000730303
52. Bouskila M, Hunter RW, Ibrahim AFM, Delattre L, Peggie M, van Diepen JA. **Allosteric Regulation of Glycogen Synthase Controls Glycogen Synthesis in Muscle**. *Cell Metab* (2010.0) **12** 456-466. DOI: 10.1016/j.cmet.2010.10.006
53. Krishnankutty A, Kimura T, Saito T, Aoyagi K, Asada A, Takahashi S-I. **In vivo regulation of glycogen synthase kinase 3β activity in neurons and brains.**. *Sci Rep.* (2017.0) **7** 8602. DOI: 10.1038/s41598-017-09239-5
54. Lai Y-C, Stuenæs JT, Kuo C-H, Jensen J. **Glycogen content and contraction regulate glycogen synthase phosphorylation and affinity for UDP-glucose in rat skeletal muscles**. *American Journal of Physiology-Endocrinology and Metabolism* (2007.0) **293** E1622-E1629. DOI: 10.1152/ajpendo.00113.2007
55. Floettmann E, Gregory L, Teague J, Myatt J, Hammond C, Poucher SM. **Prolonged Inhibition of Glycogen Phosphorylase in Livers of Zucker Diabetic Fatty Rats Models Human Glycogen Storage Diseases.**. *Toxicol Pathol.* (2010.0) **38** 393-401. DOI: 10.1177/0192623310362707
56. Gusarov I, Pani B, Gautier L, Smolentseva O, Eremina S, Shamovsky I. **Glycogen controls Caenorhabditis elegans lifespan and resistance to oxidative stress.**. *Nat Commun* (2017.0) **8** 15868. DOI: 10.1038/ncomms15868
57. KC K, Shakya S, Zhang H. **Gestational Diabetes Mellitus and Macrosomia: A Literature Review**. *Ann Nutr Metab* (2015.0) **66** 14-20. DOI: 10.1159/000371628
58. Kolahi K, Louey S, Varlamov O, Thornburg K. **Real-time tracking of BODIPY-C12 long-chain fatty acid in human term placenta reveals unique lipid dynamics in cytotrophoblast cells**. *PLoS One* (2016.0) **11** 1-23. DOI: 10.1371/journal.pone.0153522
59. Czajka A, Malik AN. **Hyperglycemia induced damage to mitochondrial respiration in renal mesangial and tubular cells: Implications for diabetic nephropathy.**. *Redox Biol.* (2016.0) **10** 100-107. DOI: 10.1016/j.redox.2016.09.007
60. Koziel A, Woyda-Ploszczyca A, Kicinska A, Jarmuszkiewicz W. **The influence of high glucose on the aerobic metabolism of endothelial EA.hy926 cells**. *Pflugers Arch* (2012.0) **464** 657-669. DOI: 10.1007/s00424-012-1156-1
61. Kumari S, Anderson L, Farmer S, Mehta SL, Li PA. **Hyperglycemia Alters Mitochondrial Fission and Fusion Proteins in Mice Subjected to Cerebral Ischemia and Reperfusion**. *Res* (2012.0) **3** 296-304. DOI: 10.1007/s12975-012-0158-9
62. Kobayashi S, Zhao F, Zhang Z, Kobayashi T, Huang Y, Shi B. **Mitochondrial Fission and Mitophagy Coordinately Restrict High Glucose Toxicity in Cardiomyocytes.**. *Front Physiol.* (2020.0) **11**. DOI: 10.3389/fphys.2020.604069
63. Wang W, Wang Y, Long J, Wang J, Haudek SB, Overbeek P. **Mitochondrial Fission Triggered by Hyperglycemia Is Mediated by ROCK1 Activation in Podocytes and Endothelial Cells**. *Cell Metab* (2012.0) **15** 186-200. DOI: 10.1016/j.cmet.2012.01.009
64. Detmer SA, Chan DC. **Functions and dysfunctions of mitochondrial dynamics**. *Nat Rev Mol Cell Biol* (2007.0) **8** 870-879. DOI: 10.1038/nrm2275
65. Kashatus JA, Nascimento A, Myers LJ, Sher A, Byrne FL, Hoehn KL. **Erk2 Phosphorylation of Drp1 Promotes Mitochondrial Fission and MAPK-Driven Tumor Growth**. *Mol Cell* (2015.0) **57** 537-551. DOI: 10.1016/j.molcel.2015.01.002
66. Jheng H-F, Tsai P-J, Guo S-M, Kuo L-H, Chang C-S, Su I-J. **Mitochondrial Fission Contributes to Mitochondrial Dysfunction and Insulin Resistance in Skeletal Muscle**. *Mol Cell Biol* (2012.0) **32** 309-319. DOI: 10.1128/MCB.05603-11
67. Wu S, Zhou F, Zhang Z, Xing D. **Mitochondrial oxidative stress causes mitochondrial fragmentation via differential modulation of mitochondrial fission-fusion proteins**. *FEBS Journal* (2011.0) **278** 941-954. DOI: 10.1111/j.1742-4658.2011.08010.x
68. Kolahi K, Valent A, Thornburg KL. **Cytotrophoblast, Not Syncytiotrophoblast, Dominates Glycolysis and Oxidative Phosphorylation in Human Term Placenta**. *Sci Rep* (2017.0) 1-12. DOI: 10.1038/srep42941
69. Joseph R, Poschmann J, Sukarieh R, Too PG, Julien SG, Xu F. **ACSL1 Is Associated With Fetal Programming of Insulin Sensitivity and Cellular Lipid Content**. *Molecular Endocrinology* (2015.0) **29** 909-920. DOI: 10.1210/me.2015-1020
70. Li LO, Ellis JM, Paich HA, Wang S, Gong N, Altshuller G. **Liver-specific Loss of Long Chain Acyl-CoA Synthetase-1 Decreases Triacylglycerol Synthesis and β-Oxidation and Alters Phospholipid Fatty Acid Composition**. *Journal of Biological Chemistry* (2009.0) **284** 27816-27826. DOI: 10.1074/jbc.M109.022467
71. Soupene E, Kuypers FA. **Mammalian Long-Chain Acyl-CoA Synthetases**. *Exp Biol Med* (2008.0) **233** 507-521. DOI: 10.3181/0710-MR-287
72. Li T, Li X, Meng H, Chen L, Meng F. **ACSL1 affects Triglyceride Levels through the PPARγ Pathway.**. *Int J Med Sci* (2020.0) **17** 720-727. DOI: 10.7150/ijms.42248
73. Parkes HA, Preston E, Wilks D, Ballesteros M, Carpenter L, Wood L. **Overexpression of acyl-CoA synthetase-1 increases lipid deposition in hepatic (HepG2) cells and rodent liver in vivo**. *American Journal of Physiology-Endocrinology and Metabolism* (2006.0) **291** E737-E744. DOI: 10.1152/ajpendo.00112.2006
74. Zhao Z, Abbas Raza SH, Tian H, Shi B, Luo Y, Wang J. **Effects of overexpression of ACSL1 gene on the synthesis of unsaturated fatty acids in adipocytes of bovine**. *Arch Biochem Biophys* (2020.0) **695** 108648. DOI: 10.1016/j.abb.2020.108648
75. Keller-Wood M, Feng X, Wood CE, Richards E, Anthony RV. **Elevated maternal cortisol leads to relative maternal hyperglycemia and increased stillbirth in ovine pregnancy**. *American Journal of Physiology-Regulatory, Integrative and Comparative Physiology.* (2014.0) **307** R405-R413. DOI: 10.1152/ajpregu.00530.2013
76. Hornnes PJ, Kühl C. **Gastrointestinal hormones and cortisol in normal pregnant women and women with gestational diabetes**. *Acta Endocrinol Suppl (Copenh).* (1986.0) **277** 24-6. PMID: 3464150
77. Ahmed SA, Shalayel MH. **Role of cortisol in the deterioration of glucose tolerance in Sudanese pregnant women**. *East Afr Med J* (1999.0) **76** 465-7. PMID: 10520355
78. Ma R, Liu J, Wu L, Sun J, Yang Z, Yu C. **Differential expression of placental 11β-hydroxysteroid dehydrogenases in pregnant women with diet-treated gestational diabetes mellitus**. *Steroids* (2012.0) **77** 798-805. DOI: 10.1016/j.steroids.2012.03.007
79. Nath A, Murthy GVS, Babu GR, Di Renzo GC. **Effect of prenatal exposure to maternal cortisol and psychological distress on infant development in Bengaluru, southern India: a prospective cohort study.**. *BMC Psychiatry* (2017.0) **17** 255. DOI: 10.1186/s12888-017-1424-x
80. Saif Z, Hodyl NA, Hobbs E, Tuck AR, Butler MS, Osei-Kumah A. **The human placenta expresses multiple glucocorticoid receptor isoforms that are altered by fetal sex, growth restriction and maternal asthma**. *Placenta* (2014.0) **35** 260-268. DOI: 10.1016/j.placenta.2014.01.012
81. Clifton VL, Cuffe J, Moritz KM, Cole TJ, Fuller PJ, Lu NZ. **Review: The role of multiple placental glucocorticoid receptor isoforms in adapting to the maternal environment and regulating fetal growth**. *Placenta* (2017.0) **54** 24-29. DOI: 10.1016/j.placenta.2016.12.017
82. Sheridan MA, Yunusov D, Balaraman V, Alexenko AP, Yabe S, Verjovski-Almeida S. **Vulnerability of primitive human placental trophoblast to Zika virus**. *Proceedings of the National Academy of Sciences* (2017.0) 114. DOI: 10.1073/pnas.1616097114
83. Hwang H-J, Jung TW, Hong HC, Seo JA, Kim SG, Kim NH. **LECT2 induces atherosclerotic inflammatory reaction via CD209 receptor-mediated JNK phosphorylation in human endothelial cells**. *Metabolism* (2015.0) **64** 1175-1182. DOI: 10.1016/j.metabol.2015.06.001
84. Kim J, Lee SK, Kim D, Lee E, Park CY, Choe H. **Adipose tissue LECT2 expression is associated with obesity and insulin resistance in K orean women.**. *Obesity* (2022.0) **30** 1430-1441. DOI: 10.1002/oby.23445
85. Jung TW, Chung YH, Kim H-C, Abd El-Aty AM, Jeong JH. **LECT2 promotes inflammation and insulin resistance in adipocytes via P38 pathways**. *J Mol Endocrinol* (2018.0) **61** 37-45. DOI: 10.1530/JME-17-0267
86. Liu Y, Fan X, Wang R, Lu X, Dang Y-L, Wang H. **Single-cell RNA-seq reveals the diversity of trophoblast subtypes and patterns of differentiation in the human placenta**. *Cell Res* (2018.0) **28** 819-832. DOI: 10.1038/s41422-018-0066-y
87. Li J, Liu X, Wang W, Li C, Li X. **MSK1 promotes cell proliferation and metastasis in uveal melanoma by phosphorylating CREB**. *Archives of Medical Science* (2020.0) **16** 1176-1188. DOI: 10.5114/aoms.2019.85810
88. Reyskens KMSE Arthur JSC. **Emerging Roles of the Mitogen and Stress Activated Kinases MSK1 and MSK2**. *Front Cell Dev Biol* (2016.0) 4. DOI: 10.3389/fcell.2016.00056
89. Elcombe SE, Naqvi S, Van Den Bosch MWM, MacKenzie KF, Cianfanelli F, Brown GD, Lang R. **Dectin-1 Regulates IL-10 Production via a MSK1/2 and CREB Dependent Pathway and Promotes the Induction of Regulatory Macrophage Markers.**. *PLoS One* (2013.0) **8** e60086. DOI: 10.1371/journal.pone.0060086
90. Feng L, Chen Y, Xu K, Li Y, Riaz F, Lu K. **Cholesterol-induced leucine aminopeptidase 3 (LAP3) upregulation inhibits cell autophagy in pathogenesis of NAFLD.**. *Aging* (2022.0) **14** 3259-3275. DOI: 10.18632/aging.204011
91. Kaur H, Ganguli D, Bachhawat AK. **Glutathione Degradation by the Alternative Pathway (DUG Pathway) in Saccharomyces cerevisiae Is Initiated by (Dug2p-Dug3p)2 Complex, a Novel Glutamine Amidotransferase (GATase) Enzyme Acting on Glutathione.**. *Journal of Biological Chemistry* (2012.0) **287** 8920-8931. DOI: 10.1074/jbc.M111.327411
92. Panga V, Kallor AA, Nair A, Harshan S, Raghunathan S, Mishmar D. **Mitochondrial dysfunction in rheumatoid arthritis: A comprehensive analysis by integrating gene expression, protein-protein interactions and gene ontology data.**. *PLoS One* (2019.0) **14** e0224632. DOI: 10.1371/journal.pone.0224632
93. Rogatzki MJ, Ferguson BS, Goodwin ML, Gladden LB. **Lactate is always the end product of glycolysis.**. *Front Neurosci.* (2015.0) **9**. DOI: 10.3389/fnins.2015.00022
94. Honda A, Yamashita K, Ikegami T, Hara T, Miyazaki T, Hirayama T. **Highly sensitive quantification of serum malonate, a possible marker for de novo lipogenesis, by LC-ESI-MS/MS**. *J Lipid Res* (2009.0) **50** 2124-2130. DOI: 10.1194/jlr.D800054-JLR200
95. Balasubramaniam S, Christodoulou J, Rahman S. **Disorders of riboflavin metabolism**. *J Inherit Metab Dis* (2019.0) **42** 608-619. DOI: 10.1002/jimd.12058
96. Balasubramaniam S, Yaplito-Lee J. **Riboflavin metabolism: role in mitochondrial function**. *J Transl Genet Genom* (2020.0). DOI: 10.20517/jtgg.2020.34
97. Ashoori M, Saedisomeolia A. **Riboflavin (vitamin B 2) and oxidative stress: a review.**. *British Journal of Nutrition* (2014.0) **111** 1985-1991. DOI: 10.1017/S0007114514000178
98. Wang G, Li W, Lu X, Zhao X. **Riboflavin Alleviates Cardiac Failure in Type I Diabetic Cardiomyopathy**. *Heart Int* (2011.0) **6**. DOI: 10.4081/hi.2011.e21
99. Alam MdM, Iqbal S, Naseem I. **Ameliorative effect of riboflavin on hyperglycemia, oxidative stress and DNA damage in type-2 diabetic mice: Mechanistic and therapeutic strategies**. *Arch Biochem Biophys* (2015.0) **584** 10-19. DOI: 10.1016/j.abb.2015.08.013
|
---
title: Designing HIV Vaccine Efficacy Trials in the Context of Highly Effective Non-vaccine
Prevention Modalities
authors:
- Holly Janes
- Yifan Zhu
- Elizabeth R. Brown
journal: Statistics in biosciences
year: 2020
pmcid: PMC10022814
doi: 10.1007/s12561-020-09292-1
license: CC BY 4.0
---
# Designing HIV Vaccine Efficacy Trials in the Context of Highly Effective Non-vaccine Prevention Modalities
## Abstract
The evolving HIV prevention landscape poses challenges to the statistical design of future trials of candidate HIV vaccines. Study designs must address the anticipated reduction in HIV incidence due to adding new prevention modalities to the standard prevention package provided to trial participants, and must also accommodate individual choices of participants with regard to the use of these modalities. We explore four potential trial designs that address these challenges, with a focus on accommodating the newest addition to the prevention package-antiretroviral-based oral pre-exposure prophylaxis (PrEP). The designs differ with respect to how individuals who take up oral PrEP at screening are handled. An All-Comers Design enrolls and randomizes all eligible individuals, a Decliners Design enrolls and randomizes only those who decline PrEP at screening, and Single and Multi-Stage Run-In Designs enroll all but randomize only those who decline PrEP or show inadequate adherence to PrEP after one or multiple run-in periods. We compare these designs with respect to required sample sizes, study duration, and resource requirements, using a simulation model that incorporates data on HIV risk and PrEP uptake and adherence among men who have sex with men (MSM) in the Americas. We advocate considering Run-In Designs for some future contexts, and identify their advantages and tradeoffs relative to the other designs. The design concepts apply beyond HIV vaccines to other prevention modalities being developed with the aim to achieve further reductions in HIV incidence.
## Introduction
Dramatic advancements have been made in recent years in antiretroviral (ARV)-based prevention of HIV infection. In particular, ARV treatment of HIV-infected individuals, or Treatment-as-Prevention (TasP), has been shown to reduce the risk of HIV transmission by a dramatic $96\%$ [8, 9], and prophylactic ARV use among HIV-uninfected individuals, or pre-exposure prophylaxis (PrEP), has been found to have high efficacy in populations that adhere to current regimens which entail daily pill taking [5, 7, 15, 17, 24, 27, 28, 38, 40]. Most recently, cabotegravir as injectable PrEP has been found to be highly effective in a trial of men who have sex with men (MSM) [32]; a trial of this same intervention in women in *Africa is* ongoing. Yet, HIV remains a significant public health burden with 1.7 million new infections in 2018, and $38\%$ of HIV-infected individuals not accessing treatment [37]. Implementation of ARV-based preventive interventions has been hindered by social, behavioral, ethical, and economic factors, and uptake and sustained adherence has been variable [25, 36, 41]. Vaccines have generally been the tools used to contain and eliminate other infectious diseases and an effective HIV vaccine will ultimately be needed to bring an end to the epidemic.
The design of future vaccine efficacy trials is challenged by the rapidly evolving HIV prevention landscape [22]. Three vaccine trials underway or recently completed utilize prototypical designs that randomize HIV-negative participants to Vaccine or Placebo and follow them for HIV infection endpoints over a fixed duration of follow-up [19]. A state-of-the-art HIV prevention package that includes risk reduction counseling, free condoms, STI testing and treatment, referral for voluntary medical male circumcision, and education around and access to oral PrEP, is provided to all participants throughout the course of the trial. Going forward, this design may no longer be optimal or feasible. A specific challenge will be balancing the ethical mandate to provide participants the best standard of HIV prevention—which in turn will reduce HIV incidence among trial participants—and enabling the assessment of vaccine efficacy through an adequately powered trial. As well, designs must reflect the reality that diverse cultural, lifestyle, and biological circumstances influence individual decision-making around the use of HIV prevention strategies, and that these choices are dynamic even over the course of a trial.
This paper discusses four potential design approaches for future vaccine efficacy trials. We focus on approaches that may apply in the next few years, anticipating that oral PrEP use will increase but remain heterogeneous. In the discussion, we comment on application of the design approaches to an era in which injectable PrEP is added to the HIV prevention package. Importantly, while our focus is vaccines, the concepts and approaches also apply to evaluating other HIV prevention products under development that are alternatives to daily oral PrEP, e.g., to microbicides or on-demand products. Reflecting the consensus that has been achieved in the field in recent years, we presume that all future designs will be conducted among individuals with “unmet need”, i.e., that individuals already using and persisting in using oral PrEP at the time of screening would not be enrolled. For these individuals, a favorable risk-benefit ratio is not achieved as they have minimal HIV risk but would be subject to the potential risks that any participant of an experimental vaccine trial takes on, e.g., due to local reactogenicity and repeated blood draws. The four designs we consider take different approaches with regard to the enrollment and randomization of individuals who take up oral PrEP at screening or during the course of the trial. Our paper is structured as follows. First, we describe the four designs and the objectives they address. Second, we implement a simulation study to investigate the relative sizes and resources required for the designs, as a means of highlighting the key parameters that may influence the choice of design for a given future context. Our simulations reflect features of the HIV epidemic and current status of PrEP use among MSM in the Americas. Last, we highlight other considerations around use of these designs and discuss variations on them that deserve future exploration.
## Study Designs
*In* general, we consider vaccine efficacy trial designs that randomize HIV-negative participants to Vaccine or Placebo and follow them for HIV infection endpoints over a fixed duration of follow-up. We explore four specific variations on this general design, motivated by recent consultations with stakeholders in HIV prevention. The four designs differ in how participants who take up oral PrEP are handled. Under each design, at trial screening otherwise eligible individuals are educated about oral PrEP and queried about usage and interest in it. Those who are already using oral PrEP, and who are satisfied with the product and would like to continue using it, are not eligible for participation under any design. This reflects consensus that these individuals do not have a favorable risk-benefit for inclusion in a vaccine trial. However, individuals not already using PrEP but interested in receiving it, or those who make informed decisions not to use PrEP, may be enrolled as described below. Given the challenges with adhering to oral PrEP and the many factors that affect individual usage of oral PrEP [35], these individuals may benefit from a vaccine that reduces their HIV risk. In practice, the manner in which trial sites provide PrEP to participants is expected to vary; in some instances, participants may be referred to another location where PrEP can be accessed, e.g., to a demonstration project or public health access program, and in others the study site physicians may prescribe PrEP to the participants. In either situation, the cost of PrEP is covered, including all safety monitoring and clinic visits. The provision of PrEP in the context of a vaccine trial does necessitate that all laboratory testing is done at the study site, and this is especially important for the HIV testing that is part of PrEP clinical care, given that vaccines can induce false positive test results with standard HIV diagnostics.
The first design we consider, which we call the “All-Comers Design”, enrolls and randomizes all eligible individuals to Vaccine or Placebo, without regard to their acceptance of the offer of oral PrEP at screening. Participants are followed for incident HIV infection for a fixed-24 month follow-up; this is a duration that is typical of current vaccine efficacy trial designs. Importantly, education about and access to PrEP continues throughout the duration of the trial- as part of the standard prevention package and participants may elect to take up PrEP at any time and are provided access in the same manner as those who take up PrEP at screening. The All-Comers Design serves as a useful reference when evaluating the other designs that in some fashion enrich for participants who choose not to use PrEP. This is essentially the design employed by two recent HIV vaccine efficacy trials, the recently completed HVTN 702 trial and the ongoing HVTN 705 trial.
The second design we consider begins with a 6-month run-in period, during which participants who take up the offer of oral PrEP at screening are enrolled and provided it. At the end of the run-in period, adherence to PrEP is assessed by measuring ARV levels in participants’ blood using standard assays [4, 42]. Only the participants without adequate adherence or who decline further use of PrEP, and who are still at risk of HIV infection, are randomized to Vaccine or Placebo and followed for HIV infection for 24 months; those who have adequate adherence and elect to continue PrEP are terminated from the study at that time point. Individuals who decline the offer of oral PrEP at screening are immediately randomized and followed for 24 months (see Fig. 1). We call this the “1-Stage Run-In Design”. The adherence threshold that is employed is a key parameter of the design, and in practice is chosen to correspond to a level of HIV incidence at which there is a favorable benefit/risk of randomization to Vaccine. Importantly, under this design all participants, including randomized participants, continue to have education around and access to oral PrEP throughout the duration of follow-up. The 1-Stage Run-In *Design is* expected to be more efficient than the All-Comers Design, by virtue of the enrichment for individuals who choose not to use PrEP or who are not adherent to it after the run-in period.
The third design is an extension of the previous design (Fig. 1). Under the “3-Stage Run-In Design”, three different run-in periods are used. At the end of each period, participants’ PrEP adherence is assessed through measuring blood ARV levels, and individuals without adequate adherence or who decline further use of PrEP and who are still at risk of HIV infection are randomized to Vaccine or Placebo and followed for 24 months. As under the 1-Stage Run-In Design, individuals who decline PrEP at screening are enrolled and immediately randomized and followed. Individuals who continue to be interested in using PrEP and who have adequate drug levels at the end of the third run-in period are terminated from the study at that time point. All participants, including those randomized, continue to have ongoing education around and access to oral PrEP throughout the duration of follow-up. While the design we consider employs three run-in periods, the concept is general and the number of run-in periods is a design parameter. The 3-Stage Run-In Design may have improved efficiency relative to the 1-Stage Run-In, if some individuals take more than one run-in period to determine whether oral PrEP works for them.
The fourth and last design we consider is called the “Decliners Design”. At screening, individuals who express an interest in initiating PrEP are provided it but are not enrolled. Only the individuals who decline the offer of PrEP at screening are enrolled and randomized. This design is the most aggressive of the four considered in the manner in which it enriches for participants who choose not to use oral PrEP. Again, however, ongoing education around and access to oral PrEP is provided and participants may choose to take up PrEP at any time. The Decliners Design concept is being employed in the ongoing HVTN 706 HIV vaccine efficacy trial.
An important attribute of all designs is that individuals who are not enrolled due to interest in PrEP (under the Decliners Design) or who are enrolled but not randomized (under the 1- or 3-Stage Run-In Designs) continue to have PrEP provided and covered by the study for 24 months, if they wish to continue using it. This is for the protection of these individuals, and it also removes a potential incentive for individuals to enroll or change their responses or behavior, simply to access PrEP. This strategy is being pursued for the HVTN 706 trial.
Table 1 outlines the primary efficacy and key PrEP-related secondary objectives that can be assessed with each design. These are fundamental for gauging the relative merits of the designs. Notably, the population randomized to Vaccine or Placebo differs across the designs, and therefore the vaccine efficacy (VE) parameter that is assessed differs. While the All-Comers Design evaluates VE for the most expansive population and the Decliners Design the most restrictive population, the Run-In Designs evaluate VE for intermediate-sized populations that are more difficult to characterize as they depend on outcomes of run-in periods.
Current vaccine efficacy trial designs prospectively collect and store blood specimens for all trial participants. At the end of the trial these specimens may be assayed for HIV-infected cases and frequency-matched controls to permit a variety of secondary analyses, including to assess VE among individuals not on PrEP at HIV acquisition (who either declined PrEP or took up PrEP post-enrollment were not adherent). As shown in Table 1, all designs considered here could have such specimen collection and assaying performed to assess this secondary objective.
Under a variation on the Run-In Designs, individuals who remain interested and adherent to PrEP after each run-in period continue to be followed for incident HIV infection. This variation allows two additional secondary objectives to be addressed (Table 1). Specifically, HIV incidence among PrEP users can be evaluated and used to evaluate PrEP effectiveness, by comparing HIV incidence among PrEP users vs. those randomized to Placebo, and also to compare Vaccine and PrEP effectiveness, by comparing HIV incidence among PrEP users vs. those randomized to Vaccine. To address these objectives, however, statistical methods would need to control for the many possible differences in risk between individuals who do and do not choose to use PrEP.
For the purposes of comparing the operating characteristics of the designs in the next section, we focus on their capacity to address the primary vaccine efficacy objectives. In the discussion, we comment on relative power of the designs to address Secondary Objective 1.
## Simulations to Compare Study Designs
We use a model that links PrEP, vaccine, and HIV infection status to simulate data for an MSM population in the Americas. The simulations are used to compare the design attributes for a few specific scenarios of interest, and to identify the parameters that may influence the choice of design more generally.
## Simulation Model
To capture the heterogeneity in HIV risk in efficacy trial populations-attributable to demographic and risk behavior characteristics—we assume that there are three latent HIV risk groups in the absence of PrEP or Vaccine, denoted by W ∈ {1, 2, 3} and called low, average, and high risk. See the full set of model parameters in Table 2. We assume there are equal proportions of individuals in each risk group. The time of HIV infection, Y, is assumed to follow an exponential distribution with a marginal annual incidence of 3 per 100 person-years (λ0), and the hazard ratios for the three risk groups are HRW = (0.5, 1.0, 2.0) which results in incidence rates of 1.5, 3 and 6 per 100 person-years, respectively. The marginal incidence of 3 per 100 person-years is consistent with rates seen in recent efficacy trials in the MSM population in the Americas [21]. Sensitivity analyses described in Online Resources show the results for $4\%$ marginal incidence.
Individuals are assumed to belong to one of three latent PrEP adherence groups, denoted by $A = 1$, 2, 3 (see Fig. 2). The groups are based on data from HPTN 069 [18], a phase II PrEP safety and tolerability study in MSM and at-risk women in the US and Puerto Rico. In this study, Wisepill™ electronic device monitoring was used to measure participants’ daily pill-taking. Linear change-point models fitted to the HPTN 069 data ($$n = 406$$ MSM and $$n = 188$$ women; $$n = 182$$ subjects with dense Wisepill™ data) [43] identified three adherence groups, corresponding to consistently high adherence, slowly declining adherence, and rapidly declining adherence as measured by the fraction of pills taken per week, with group membership probabilities P(A = a) ∈ {.4,.3,.3}. We use these previously published latent PrEP adherence groups, and assume that once an individual in group A = a takes up the offer of PrEP, the latent individual-level adherence trajectory, defined in terms of the fraction of prescribed pills taken and denoted by A(t) = μ(t) + σ(t), is assumed to follow the time-varying linear change-point adherence model with mean process μa(t) = μ0a + β1a(t ∧ t0a) + β2atI(t > t0a) where t0a is the change-point, and additive random noise process σa(t)~N(0,σa2). To account for additional measurement error in the adherence measurements we assume that the observed adherence trajectory is Aobs(t) = μ(t) + σ*(t), where σa*(t)~N(0,4⋅σa2). See Online Resources for the parameter values. This model allows for variability among individuals within an adherence group, but importantly it assumes that expected adherence declines monotonically over time. The HPTN 069 data form a useful basis for the adherence model because they are rich in time and therefore permit simulating daily individual-level adherence measurements. Qualitatively, the HPTN 069-based adherence model is consistent with recent demonstration project data showing modest uptake and persistence of PrEP use among MSM in the Americas [11, 20, 33, 34].
Given that many PrEP efficacy trials have found that factors that associate with higher HIV risk also tend to predict lower adherence to oral PrEP [2, 12, 27], our model allows for potentially correlated latent adherence and risk groups. Specifically, we assume the marginal responses of A and W are derived from probit-transformed latent continuous random variables that are bivariate normal with correlation ρAW; see [39] for details. Qualitatively, positive ρAW is consistent with low-risk individuals tending to be more adherent, while negative ρAW indicates that high-risk individuals tend to be more adherent.
Reflecting the current context in which PrEP uptake is heterogeneous, our model allows for a fraction of participants to decline the offer of PrEP at screening, and for some of these initial decliners to take up PrEP at some time point post-screening. Specifically, let tiuptake be the PrEP uptake time for subject i. Individuals who take up the offer of PrEP at screening have tuptake = 0. Individuals who decline PrEP at screening have tuptake > 0; some will never take up PrEP during the maximum 42-month follow-up across designs, i.e., tuptake > T where T is the maximum follow-up time, and others will take up PrEP at certain follow-up visit after initial enrollment (0 < tuptake < T). We assume a fixed value for P(tuptake = 0) and that post-screening PrEP uptake is uniformly distributed across 3-monthly follow-up visits until truncation at time T. Therefore the distribution of min{0, tuptake} is a zero-and-T-inflated uniform distribution at 3-month follow-up intervals between first (tuptake = 3 months) and last (tuptake = T) visits, where $T = 42$ months. Based on data from the ongoing HVTN 704 monoclonal antibody prevention trial in MSM in which oral PrEP is offered to participants and uptake is approximately $25\%$ (Peter Gilbert, personal communication), but anticipating increasing uptake in coming years, we set P(tuptake = 0) = 0.5, P(0 < tuptake < T) = 0.3, and P(tuptake > T) = 0.2. At tiuptake, subject i’s adherence trajectory is assumed to follow from the linear change-point model for adherence category Ai. We assume that Ai(t) = 0 for t<tiuptake, i.e., that participants do not procure PrEP outside the study. If tiuptake>T we set Ai(t) = 0 for all t. Importantly, this model assumes that PrEP uptake is independent of latent adherence and HIV risk categories. Sensitivity analyses shown in Online Resources consider lower rates of PrEP uptake.
Data from PrEP efficacy trials and studies with directly observed dosing [3, 16] form the basis for our model linking time-varying adherence A(t) with HIV outcomes. We assume the hazard ratio associated with PrEP follows [1] HRPrEP(A(t))=e−θ1A(t)θ2.
An exponential model with θ2 fixed to 1 was used in [3] to estimate the association between adherence, as measured by tenofovir concentration in peripheral blood mononuclear cells (PBMCs), and HIV risk among MSM on PrEP, based on data from the iPrEX trial [15]. Figure 3 shows point estimates of adherence and PrEP hazard ratios from [3, 16], based on calculations that convert drug concentration levels in PBMCs or dried blood spots to the fraction of pills taken using the published relationship between dosing and drug concentration [3, 6], and that assume that risk among placebo recipients does not vary with adherence to daily pill-taking. We estimated the parameters θ1 and θ2 by fitting Model 1 to the points in Fig. 3 using least-squares. This model motivated the choice of PrEP adherence threshold below which the Run-In Designs randomize a participant who takes up PrEP. The threshold A0 = 0.15 corresponds to an estimated PrEP efficacy of $54\%$; this is a level of PrEP efficacy below which it may be deemed ethical to randomize an individual to ascertain whether the vaccine can reduce HIV risk even further.
Finally, we assume that participants are randomized to Vaccine ($Z = 1$) or Placebo ($Z = 0$) with equal probability. Vaccine efficacy is measured by the multiplicative reduction in HIV risk due to assignment to vaccine, VE = 1 − HRV, and is assumed constant over 24 months follow-up. The vaccine is also assumed not to interact with HIV risk or usage of oral PrEP. Given (W, A, Z), λ(t|W, A(t), Z) = λ0 · HRW · HRPrEP(A(t)) · (1 − VE)Z is the instantaneous hazard of HIV infection. The cumulative probability distribution function of Y is FY(t|W,A,Z)=1−e−∫0tλ(s∣W,A(s),Z)ds. We assume an exponentially distributed non-informative censoring time, C ~ Exp(λc), independent of (W, A, Z, Y), and a $10\%$ annual censoring rate. Table 2 describes the full set of model parameters and assumed values.
To capture the resources required for the designs, we describe the number of participants who must be screened and enrolled to achieve adequate power and discuss the implications with respect to accrual time. We also characterize the resources required for PrEP provision and adherence monitoring. For the purposes of this work, we make the simplifying assumption that the time of screening is the same as the time of enrollment. In practice, there may be a small gap of several days between the two time points.
## Simulation Algorithm
We describe the steps in simulating the data for the All-Comers and 1-Stage Run-In Designs, and briefly highlight the major differences in approaches to the 3-Stage Run-In and Decliners Designs below.
For the All-Comers Design, we begin by simulating each individual’s latent HIV risk and adherence group membership, Wi ~ pW(w) and PrEP adherence Ai ~ pA(a) for $i = 1$, …, n, based on the bivariate normal latent variables with correlation ρAW. Next, we simulate each individual’s PrEP uptake time, tiuptake. The individual-level PrEP adherence trajectory is simulated from the time of PrEP uptake, according to the linear change-point model for group Ai. Given a random treatment assignment Zi and a censoring time Ci we next simulate the HIV infection time Yi with the cumulative intensity process Λi(t)=∫0tλ0⋅HRWi⋅HRPrEP(Ai(s))ds. Let Fi(t)=1−e−Λi(t). We use inverse transform sampling to simulate Yi=Fi−1(U) where U ~ Unif[0, 1] and then calculate the observed event time Yiobs=min(Yi,Ci,24months) and censoring indicator δi=1−I(Yi=Yiobs). The Decliners *Design is* simulated similarly but randomization only occurs for those with tiuptake>0.
For the 1-Stage Run-In Design, Wi, Ai, tiuptake, and Ai(t) are simulated as above. For individuals with tiuptake=0, data for the 6-month run-in are generated as follows. The infection time during 1-Stage Run-In, Yi1, is generated using the intensity process λi1(t)=λ0⋅HRWi⋅HRPrEP(Ai(t)). The censoring time, Ci1 follows an exponential distribution with intensity λC. The measured adherence level at the end of the 6-month run-in period is denoted by Aiobs($t = 6$month). Randomization occurs for those with tiuptake=0, Aiobs(6month)<A0 and min(Yi1,Ci1)≥6months. Let δi,1Vtrial=I[min(Yi1,Ci1)≥6months]*I[Aiobs(6month)<A0]*I[tiuptake=0] be an indicator of satisfying these conditions. *We* generate Zi for each individual with δi,1Vtrial=1. We simulate the post-randomization outcome Yi2 with intensity process λi2(t)=λ0⋅HRWi⋅HRPrEP(Ai(t+6month))⋅(1−VE)Zi similar to the generation process of Yi1, and Ci2 as exponential with rate λC. The observed post-randomization event time is Yi2,obs=min(Yi2,Ci2,24months) and the censoring indicator is δi2=1−I(Yi2=Yi2,obs). Individuals with tiuptake>0 are randomized at enrollment and have outcomes generated as under the All-Comers Design.
Under the 3-Stage Run-In Design, subject i is randomized after the kth run-in stage if δi,kVtrial∏$l = 1$k−1(1−δi,lVtrial)=1, where δi,kVtrial=I[min(Yi1,Ci1)≥6*kmonth]*I[Aiobs(6*kmonths)<A0]*I[tiuptake=0]. Post-randomization event times (Yik,s,$k = 2$,3,4) are generated as under the 1-Stage Run-In Design, although the subjects’ time-varying hazard function will begin at $t = 6$ * k months instead of $t = 0$, since the 24 months follow-up period after run-in randomization excludes previous run-in periods.
## Power Comparison
We compare the power of the designs to reject H0: VE ≤ $25\%$ under the alternative Ha: VE = $50\%$ in the randomized and modified intent-to-treat (MITT) population, defined as the set of randomized participants who are retrospectively determined to have been HIV-uninfected at the time of randomization. The choice of null and alternative hypotheses is consistent with recent vaccine efficacy trial designs [13], and is based on consideration of properties of a minimally useful vaccine and informed by the anticipated level of efficacy of current vaccine regimens.
An 0.05-level 2-sided log-rank test is used to evaluate VE in the randomized MITT population. For the 1-Stage and 3-Stage Run-In Designs, the test is stratified on the stage of randomization for improved power.
Note that even though the designs enroll and randomize different populations (see Table 1), under our assumption that the vaccine does not interact with baseline risk, PrEP uptake, or PrEP adherence the true VE parameter is the same across all populations. Therefore all four designs produce unbiased VE estimates.
We calculate empirical power by simulating designs for a grid of possible total sample sizes, calculating the probability of H0 rejection under Ha for each sample size based on 1000 simulations, and plotting the results to find the number needed to randomize to achieve $90\%$ power.
## Required Sample Sizes
Table 3 shows the fraction of participants randomized at enrollment and after each run-in period for each of the designs. While individuals who decline PrEP at screening are immediately randomized under all designs, those who take up PrEP at screening follow different paths depending on the design. For the All-Comers Design, those who take up PrEP at screening are also immediately randomized, while for the Decliners Design those who take up PrEP at screening are not enrolled (or randomized). For the 1- and 3-Stage Run-In Designs, the time point at which participants are randomized depends on what PrEP adherence category they fall in to. Those who fall in to the consistently high adherence category ($A = 1$) never have adherence levels that fall below the adherence threshold of $15\%$ of pills per week, and thus are never randomized under the Run-In Designs. Those in slowly declining adherence category ($A = 2$) tend to be randomized after two or three run-in periods; a small fraction have low enough adherence to be randomized after one run-in period. Those with rapidly declining adherence ($A = 3$) quickly achieve low adherence levels and are randomized after one run-in period for both Run-In Designs.
Table 4 compares the numbers of participants involved in each of the trial designs, where we distinguish between the number of individuals who are offered PrEP at screening, the number who are enrolled, and the number who are randomized to Vaccine or Placebo and used to evaluate vaccine efficacy. Results are shown for three different values of ρAW, the correlation between baseline risk and PrEP adherence.
For the All-Comers Design, the numbers offered PrEP, enrolled, and randomized are the same- and are large by virtue of the lower placebo-group incidence due to PrEP.
The Decliners Design, which enrolls and randomizes only those who decline PrEP at screening, requires considerably fewer participants, 7000 vs. 9000, or a $22\%$ reduction in sample size under ρAW = 0. However, under our assumptions, in order to enroll these 7000 participants, PrEP is provided for 24 months to an additional 7000 individuals who are not enrolled and do not contribute to meeting the scientific objectives of the trial. This is an important consideration in terms of resources.
The 1-Stage Run-In Design requires slightly fewer participants to achieve $90\%$ power. With ρAW = 0, 6762 are required vs. 7000 for the Decliners Design, and thus the 1-Stage Run-In achieves a $25\%$ reduction in sample size relative to the All-Comers Design. Most of those randomized are randomized at enrollment ($83\%$) as opposed to after the single run-in period. The slight gain in power relative to the Decliners *Design is* due to the enrollment and randomization of individuals who take up PrEP at screening but who fall into the low adherence category (see Table 3), and who therefore have an HIV incidence that is minimally reduced by PrEP.
An important cost consideration of the 1-Stage Run-In *Design is* that many participants enrolled are provided PrEP for the run-in period and have adherence assessed at its end, but are not randomized ($43\%$ under ρAW = 0); these participants do not contribute to assessing vaccine efficacy. The design accrues the cost of enrolling and following participants for the run-in period for the purposes of adherence assessment, and this needs to be factored in to cost assessments. As well, the 1-Stage Run-In *Design is* longer in duration than the All-Comer or Decliners Designs, with 30 months maximum follow-up as compared to 24 months follow-up per participant under the other designs.
Interestingly, under our model the 3-Stage Run-In *Design is* less powerful than the 1-Stage Run-In Design. With ρAW = 0, a total of 7265 individuals must be randomized to achieve $90\%$ power vs. 6762 for the 1-Stage Run-In Design. The reason for the lower trial power is that the only difference between the designs is the randomization of a relatively small fraction of individuals who take up PrEP at screening and who fall into the intermediate adherence category (see Table 3), and these individuals are randomized at the expense of those who decline PrEP or who have low PrEP adherence and who have higher HIV incidence. Note that the vast majority of those randomized under the 3-Stage Design are randomized at enrollment ($63\%$) or after the first run-in ($14\%$). Also observe that the 3-Stage Run-In *Design is* considerably longer in duration than the other designs, with a maximum follow-up time of 42 months as compared to 24 months for the All-Comers and Decliners Designs and 30 months for the 1-Stage Run-In Design. However, in contrast to the 1-Stage Run-In Design, most ($76\%$ under ρAW =0) of those enrolled are randomized and contribute to evaluating vaccine efficacy.
Under the likely scenario in which higher risk subgroups are likely to be less adherent (ρAW = 0.5), the results are similar, with the 1-Stage Run-In Design achieving the smallest number of randomized participants ($75\%$ the size of the All-Comers Design). All four designs require fewer participants to have $90\%$ power, since PrEP has a smaller effect on HIV risk. When the opposite is true and lower risk subgroups are likely to be less adherent (ρAW = −0.5), all four designs require more participants, and the designs that enrich for individuals who choose not to use PrEP are less powerful relative to the All-Comers Design because under this scenario enriching for poor adherence is akin to enriching for lower risk.
The impact of the enrichment for individuals who choose not to use PrEP is visually apparent in Fig. 4, which compares the power of the designs as a function of the number of participants randomized to Vaccine or Placebo (NR). The All-Comers Design, which enrolls without regard to PrEP uptake and adherence, has the lowest power for a fixed NR among the designs. The 1-Stage Run-In Design has the highest power, by virtue of its randomization of individuals who decline PrEP or are not adherent after 6 months; with 5000 participants randomized when ρAW = 0.5, the 1-Stage Run-In Design has $80\%$ power as compared to $66\%$ for the All-Comers Design. The 3-Stage Run-In Design has slightly lower power at $77\%$ and the Decliners Design has $75\%$ power.
When ρAW = −0.5, larger sample sizes are required to achieve the same power for all four proposed designs. Interestingly, in this scenario the Run-In and Decliner Designs are quite similar to each other, and the 1-Stage Run-In Design has less advantage over the 3-Stage Run-In Design.
## Study Resource Requirements
Figure 5 summarizes the expected duration of the PrEP screening, enrollment, and follow-up periods for the four designs. To enable their comparison, we assume that 350 individuals can be screened for interest in PrEP per month-with PrEP access provided if interested- and that 250 trial participants can be enrolled per month. However, the relative screening and enrollment periods of the designs are invariant to these assumed rates. Typically, screening and enrollment periods would occur in parallel, but examining them separately shows where the time resources are invested for each of the designs. We see that, by virtue of the Decliners Design’s need to screen large numbers of individuals to enroll only those who decline PrEP at screening, this design has the longest screening period, but also the shortest enrollment and per-participant follow-up periods; the Run-In Designs require shorter screening but longer enrollment and follow-up times.
Figure 6 compares the designs with respect to the person-years of off-study PrEP provision. Off-study PrEP includes 24 months of PrEP for individuals who are screened but not enrolled, and 24 months of PrEP for participants continuing PrEP after one or more run-in periods, who are not randomized and therefore do not contribute to meeting primary study objectives. The designs are also compared with respect to required number of PrEP adherence tests. The figure emphasizes that the Run-In and Decliners Designs involve providing considerable person-years of off-study PrEP. The Run-In Designs also require PrEP adherence testing that the other designs do not. These factors must be weighed against any decrease in sample size.
## Influence of Simulation Model Parameters
Parallel results for additional parameter settings are contained in Online Resources. The relative power of the designs is found to be similar in general for higher marginal HIV incidence. When the rate of PrEP uptake at screening is assumed to be lower ($25\%$ vs. $50\%$) and the fraction who never take up PrEP is higher ($30\%$ vs. $20\%$), the All-Comers *Design is* seen to be more powerful but still not competitive against the Run-In or Decliners Designs. When the PrEP adherence threshold is increased from A0 = 0.15 to A0 = 0.3, corresponding to $82.5\%$ PrEP efficacy (see Fig. 3), the Run-In Designs have slightly improved power and require smaller NR relative to the All-Comers Design.
A parameter with particular influence is the patterns of PrEP adherence. Under our simulation model, motivated by the HPTN 069 data, the majority of individuals have stable adherence or rapidly declining adherence. In populations with sizeable subgroups with slowly declining adherence, the 3-Stage Run-In *Design is* expected to have improved power.
Another parameter that is influential in general is the length of the run-in periods, which we took to be 6 months long. As shown in the Online Resources, when a shorter 3-month run-in period is used, the 3-Stage Run-In Design has higher power (requires fewer randomized participants) compared to all other designs as long as ρ = 0 or 0.5. This is due to the less frequent randomization of individuals in the intermediate ($A = 2$) adherence category.
## Discussion
This paper compared four potential designs for future HIV vaccine efficacy trials that take into account what is anticipated to be increasing but variable use of oral PrEP in at-risk populations- both within and among individuals. While recent HIV vaccine efficacy trials (HVTN 702 and 705) utilize an “All-Comers” approach that enrolls participants without regard to PrEP uptake, and a trial just underway (HVTN 706) is using a “Decliners” approach that only enrolls those who decline oral PrEP at screening, we found that Run-In Designs that provide interested individuals with PrEP for fixed durations of time, assess interest and adherence at the end of run-in periods, and randomize those not interested in continuing PrEP or not adherent to it, can require smaller numbers of randomized participants. However, this sample size advantage must be traded off against longer duration of follow-up, the additional cost of measuring adherence to PrEP, and the cost of providing PrEP to individuals during run-in, before randomization.
An important attribute of the various designs is the relative interpretability of their vaccine efficacy parameter estimates. While the efficacy estimates from the All-Comers and Decliners Designs are simple to interpret, those from the Run-In Designs are not, since the populations they apply to are defined based on outcomes measured after run-in(s). Thus, another limitation of the Run-In *Designs is* parameter interpretability, and this must be weighed against the potentially reduced sample size.
There are additional challenges with implementing the proposed designs that require separate investigation. How will trial feasibility be assessed, in the face of uncertain PrEP uptake and adherence in advance of the trial? Current efficacy trials utilize operational futility monitoring plans that assess HIV incidence in the treatment-arm-pooled trial population, to enable design modification or termination for operational futility if HIV incidence is lower than anticipated. Will such monitoring among randomized participants suffice, or will additional procedures be needed? As well, will there be licensure implications if vaccine efficacy is evaluated using a design that in some way enriches for individuals not using PrEP? In particular, will regulators be concerned with the generalizability of the trial results, and the safety and efficacy of the vaccine in populations with higher rates of PrEP use- and how will these concerns be addressed? Also challenging, how will community and site investigators be engaged meaningfully to permit such designs, which will require careful communication regarding the scientific and ethical rationale for the designs and the procedures in place to protect participant safety? The designs discussed will clearly not be possible without robust support from these key stakeholders. Furthermore, how will participant informed consent be procured? Participants will need to demonstrate not only an understanding of the vaccine and its potential risks and benefits but also of PrEP and the trial’s approach to it, and to authentically choose whether and when to use PrEP [23]. Finally, will retention of trial participants be more challenging with these designs? Given that some of the barriers to sustained adherence to PrEP may also be barriers to attending clinic visits and adhering to other clinical procedures, there is a potential for participants who choose not to continue on PrEP to be harder to retain on-study. How will the designs anticipate and tackle this issue? The challenges summarized here are considerable.
In exploring the proposed designs, our simulation study did not endeavor to exhaustively examine the potential scenarios in which one design may be preferred over another. Instead, our goal was to illustrate and compare the designs in a handful of reasonable scenarios, and to use the simulations to identify parameters that are especially influential in terms of design optimality. Online Table 10 lists the key attributes of the target population, the study design, and the clinical context, and highlights those that will have major impact on the relative power of the designs we considered. These include the patterns of PrEP adherence, uptake, and the efficacy of PrEP in the target population; and the length of the run-in period and the PrEP adherence randomization threshold which are design attributes under control of the investigator. As well, when considering the relative resource requirements of the designs, the costs listed in Online Table 11 must be considered, although formal cost comparison of designs is complex given that typically these costs will vary by site for multi-site trials, and over study time.
Our simulations made simplifying assumptions by necessity. For example, we assumed that the rate and timing of PrEP uptake is independent of baseline HIV risk and latent PrEP adherence groups. How a dependency would affect design performance is difficult to predict. For simplicity, we also assumed a time-constant vaccine efficacy, whereas in reality vaccine trials commonly anticipate and accommodate ramping efficacy (prior to full immunity) and waning efficacy given the limited durability of vaccine-induced immune responses. Since this ramping and waning is likely independent of PrEP use, we expect that the simplification does not influence the relative performance of the designs.
There are variations on the Run-In Designs that merit further investigation. As mentioned above, Run-In Designs that proceed to follow participants who continue on PrEP post-run-in for incident HIV infection have the merit that additional secondary objectives can be assessed around PrEP effectiveness and Vaccine vs. PrEP effectiveness. Alternatively, Run-In Designs may provide the vaccine to all or a random subset of participants who continue on PrEP post-run-in, in order to collect safety and immunogenicity outcomes among PrEP users. This may aid in licensure decisions and facilitate rollout of vaccine programs in populations using PrEP. As well, Run-In Designs may employ PrEP adherence monitoring throughout the duration of the run-in period(s), not just at the end of each period. While additional resources would be required for this more frequent adherence monitoring, there may be an advantage in terms of increased power and shorter average follow-up times as participants are randomized earlier in time. *More* generally, the length and number of the run-in periods require optimization from both a statistical and operational vantage-point, and need to be informed by anticipated PrEP uptake and adherence patterns.
The Run-In Designs we explored are similar to Sequential Multiple Assignment Randomized Trial (SMART) Designs which have been used to evaluate behavioral interventions that require modification under lack of adherence or response to intervention [1, 10, 26, 29–31]. SMART designs are appealing in that they reflect the need for certain interventions to be modified based on response, use, or tolerability. As well, they can provide data to discover optimal individual treatment policies. If in the future there are multiple PrEP variations for individuals to consider, a SMART design could be employed to compare the effectiveness of PrEP-A vs. PrEP-B, to evaluate vaccine efficacy, and also to discover an optimal treatment policy that combines vaccine and PrEP and takes into account subject preferences and adherence to PrEP.
The HIV prevention package continues to expand. Recent results from a phase 3 efficacy trial of injectable PrEP (Cabotegravir) suggest high efficacy relative to oral PrEP (tenofovir disoproxil fumarate/emtricitabine) in MSM and transgender women in North and South America, Asia, and southern Africa [32]. Another trial of the same intervention in women is ongoing. While there remain uncertainties about the long-term safety profile and acceptability of injectable PrEP, a likely scenario is that this intervention will be added to the HIV prevention package at some point in the near future. The designs discussed herein could have application to this context, as long as there remains a subpopulation of individuals who decline or are not able to adhere to either oral or injectable PrEP. Importantly, however, for the designs to be feasible, these individuals would need to be willing to be randomized to receive a vaccine. This may be unlikely; whereas oral PrEP and vaccines are different experiences for the participant (dosing, mode of delivery, side effects), injectable PrEP and vaccines provide more similar participant experiences, e.g., the HPTN 083 regimen entails an injection every 8 weeks and current HIV vaccine regimens involve 3–6-monthly injections. On the other hand, the current injectable PrEP regimen entails combining injections with oral PrEP; when an individual is taken off the injections, oral PrEP is given to “cover the tail”, i.e., to prevent HIV infection during a period when drug resistance could occur. Therefore, individuals who refuse oral PrEP currently are not eligible for injectable PrEP. Population acceptability of injectables, and future research on covering the tail of injectables, may modify these considerations.
While explored for the vaccine context, the designs we discussed may have application to other HIV prevention interventions that are viewed as alternatives to oral PrEP. It is recognized that many factors influence individual decision-making around products and practices to protect against HIV, and much like the field of contraception, multiple products will ultimately be needed to provide all at-risk populations with strategies that are effective, acceptable, and available [14]. Products such as vaginal rings, rectal microbicides, and other on-demand products will face similar challenges for efficacy trial design as for vaccines, and the designs we describe have direct application.
## References
1. Almirall D, Nahum-Shani I, Sherwood NE, Murphy SA. **Introduction to SMART designs for the development of adaptive interventions: with application to weight loss research**. *Transl Behav Med* (2014.0) **4** 260-274. PMID: 25264466
2. Amico KR, Mehrotra M, Avelino-Silva VI, McMahan V, Veloso VG, Anderson P, Guanira J, Grant R. **Self-reported Recent PrEP Dosing and Drug Detection in an Open Label PrEP Study**. *AIDS Behav* (2016.0) **20** 1535-1540. PMID: 26992393
3. Anderson PL, Glidden DV, Liu A, Buchbinder S, Lama JR, Guanira JV, McMahan V, Bushman LR, Casapia M, Montoya-Herrera O, Veloso VG, Mayer KH, Chariyalertsak S, Schechter M, Bekker LG, Kallas EG, Grant RM. **Emtricitabine-tenofovir concentrations and pre-exposure prophylaxis efficacy in men who have sex with men**. *Sci Transl Med* (2012.0) **4** 125-151
4. Anderson PL, Liu AY, Castillo-Mancilla JR, Gardner EM, Seifert SM, McHugh C, Wagner T, Campbell K, Morrow M, Ibrahim M, Buchbinder S, Bushman LR, Kiser JJ, MaWhinney S. **Intracellular tenofovir-diphosphate and emtricitabine-triphosphate in dried blood spots following directly observed therapy**. *Antimicrob Agents Chemother* (2018.0) **62** e01710. PMID: 29038282
5. Baeten JM, Donnell D, Ndase P, Mugo NR, Campbell JD, Wangisi J, Tappero JW, Bukusi EA, Cohen CR, Katabira E, Ronald A, Tumwesigye E, Were E, Fife K, Kiarie J, Farquhar C, John-Stewart G, Kakia A, Odoyo J, Mucunguzi A, Nakku-Joloba E, Twesigye R, Ngure K, Apaka C, Tamooh H, Gabona F, Mujugira A, Panteleeff D, Thomas KK, Kidoguchi L, Krows M, Revall J, Morrison S, Haugen H, Emmanuel-Ogier M, Ondrejcek L, Coombs RW, Frenkel L, Hendrix C, Bumpus NN, Bangsberg D, Haberer JE, Stevens WS, Lingappa JR, Celum C, Team PPS. **Antiretroviral prophylaxis for HIV prevention in heterosexual men and women**. *N Engl J Med* (2012.0) **367** 399-410. PMID: 22784037
6. Castillo-Mancilla JR, Zheng JH, Rower JE, Meditz A, Gardner EM, Predhomme J, Fernandez C, Langness J, Kiser JJ, Bushman LR, Anderson PL. **Tenofovir, emtricitabine, and tenofovir diphosphate in dried blood spots for determining recent and cumulative drug exposure**. *AIDS Res Hum Retroviruses* (2013.0) **29** 384-390. PMID: 22935078
7. Choopanya K, Martin M, Suntharasamai P, Sangkum U, Mock P, Leethochawalit M, Chiamwongpaet S, Kitisin P, Natrujirote P, Kittimunkong S, Chuachoowong R, Gvetadze R, McNicholl J, Paxton L, Curlin M, Hendrix C, Vanichseni S. **the Bangkok Tenofovir Study Group: antiretroviral prophylaxis for HIV infection in injecting drug users in Bangkok, Thailand (The Bangkok Tenofovir Study): a randomised, double-blind, placebo-controlled phase 3 trial**. *Lancet* (2013.0) **381** 2083-2090. PMID: 23769234
8. Cohen M, Chen Y, McCauley M, Gamble T, Hosseinipour M, Kumarasamy N, Hakim J, Kumwenda J, Grinsztejn B, Pilotto J, Godbole S, Chariyalertsak S, Santos B, Mayer K, Hoffman I, Eshleman S, Piwowar-Manning E, Cottle L, Zhang X, Makhema J, Mills L, Panchia R, Faesen S, Eron J, Gallant J, Havlir D, Swindells S, Elharrar V, Burns D, Taha T, Nielsen-Saines K, Celentano D, Essex M, Hudelson S, Redd A, Fleming T, Team HS. **Prevention of HIV-1 infection with early antiretroviral therapy**. *N Engl J Med* (2011.0) **365** 493-505. PMID: 21767103
9. Cohen M, Chen Y, McCauley M, Gamble T, Hosseinipour M, Kumarasamy N, Hakim J, Kumwenda J, Grinsztejn B, Pilotto J, Godbole S, Chariyalertsak S, Santos B, Mayer K, Hoffman I, Eshleman S, Piwowar-Manning E, Cottle L, Zhang X, Makhema J, Mills L, Panchia R, Faesen S, Eron J, Gallant J, Havlir D, Swindells S, Elharrar V, Burns D, Taha T, Nielsen-Saines K, Celentano D, Essex M, Hudelson S, Redd A, Fleming T, Team HS. **Prevention of HIV-1 infection with early antiretroviral therapy**. *N Engl J Med* (2016.0) **375** 830-839. PMID: 27424812
10. Dawson R, Lavori PW. **Efficient design and inference for multistage randomized trials of individualized treatment policies**. *Biostatistics* (2012.0) **13** 142-152. PMID: 21765180
11. Doblecki-Lewis S, Liu AY, Feaster DJ, Cohen SE, Elion R, Bacon O, Coleman M, Cardenas G, Kolber MA. **Patterns and correlates of participant retention in a multi-city pre-exposure prophylaxis demonstration project**. *J Acquir Immune Defic Syndr* (2018.0) **79** 62-69. PMID: 29771790
12. Donnell D, Baeten J, Bumpus N, Brantley J, Bangsberg D, Haberer J, Mujugira A, Mugo N, Ndase P, Hendrix C, Celum C. **HIV protective efficacy and correlates of tenofovir blood concentrations in a clinical trial of prep for HIV prevention**. *J Acquir Immune Defic Syndr* (2014.0) **66** 340-348. PMID: 24784763
13. Gilbert PB, Grove D, Gabriel E, Huang Y, Gray G, Hammer SM, Buchbinder SP, Kublin J, Corey L, Self SG. **A sequential phase 2b trial design for evaluating vaccine efficacy and immune correlates for multiple HIV vaccine regimens**. *Stat Commun Infect Dis* (2011.0) **3** 1037. PMID: 23181167
14. Glidden D. **Advancing novel prep products - alternatives to non-inferiority**. *Stat Commun Infect Dis* (2019.0) **65** 324
15. Grant R, Lama J, Anderson P, McMahan V, Liu A, Vargas L, Goicochea P, Casapia M, Guanira-Carranza J, Ramirez-Cardich M, Montoya-Herrera O, Fernandez T, Veloso V, Buchbinder S, Chariyalertsak S, Schechter M, Bekker L, Mayer K, Kallas E, Amico K, Mulligan K, Bushman L, Hance R, Ganoza C, Defechereux P, Postle B, Wang F, McConnell J, Zheng J, Lee J, Rooney J, Jaffe H, Martinez A, Burns D, Glidden D. **Preexposure chemoprophylaxis for HIV prevention in men who have sex with men**. *N Engl J Med* (2010.0) **363** 2587-2599. PMID: 21091279
16. Grant RM, Anderson PL, McMahan V, Liu A, Amico KR, Mehrotra M, Hosek S, Mosquera C, Casapia M, Montoya O, Buchbinder S, Veloso VG, Mayer K, Chariyalertsak S, Bekker LG, Kallas EG, Schechter M, Guanira J, Bushman L, Burns DN, Rooney JF, Glidden DV. **Uptake of pre-exposure prophylaxis, sexual practices, and HIV incidence in men and transgender women who have sex with men: a cohort study**. *Lancet Infect Dis* (2014.0) **14** 820-829. PMID: 25065857
17. Harel C, Coll J, Ruane P, Molia JM, Mayer KHJ, Grant R, De Wet J, Thompson M, DeJesus E, Ebrahimi RRMG, Das M, Brianard D, McCallister S. *The Phase 3 Discover Study: Daily F/TAF or F/TDF for HIV Preexposure Prophylaxis* (2019.0)
18. 18.HIV Prevention Trials Network (2018) HPTN 069: A Phase II Randomized, Double-Blind, Study of the Safety and Tolerability of Maraviroc (MVC), Maraviroc + Emtricitabine (MVC+FTC), Maraviroc + Tenofovir disoproxil fumarate (MVC+TDF), or Tenofovir disoproxil fumarate + Emtricitabine (TDF+FTC) for Pre-Exposure Prophylaxis (PrEP) to Prevent HIV Transmission in At-Risk Men Who Have Sex with Men and in At-Risk Women (Cited 2018). https://hptn.org/research/studies/hptn069. *HPTN 069: A Phase II Randomized, Double-Blind, Study of the Safety and Tolerability of Maraviroc (MVC), Maraviroc + Emtricitabine (MVC+FTC), Maraviroc + Tenofovir disoproxil fumarate (MVC+TDF), or Tenofovir disoproxil fumarate + Emtricitabine (TDF+FTC) for Pre-Exposure Prophylaxis (PrEP) to Prevent HIV Transmission in At-Risk Men Who Have Sex with Men and in At-Risk Women* (2018.0)
19. 19.HIV Vaccine Trials Network (2017) HVTN Protocols (Cited Mar 2018). https://www.hvtn.org/content/dam/hvtn/Science/studies/IR%20table%20Dec%202017.pdf. *HVTN Protocols* (2017.0)
20. Huang Y, Tao G, Smith D, Hoover K. *Persistence with HIV Preexposure Prophylaxis in the United States, 2012–2016* (2019.0)
21. 21.HVTN Efficacy Trials Working Group (2019) HIV Incidence in Prevention Trial Control Groups and Observational Studies: A Summary Table (Cited 2019). https://www.hvtn.org/en/resources/hivresources-links.html. *HIV Incidence in Prevention Trial Control Groups and Observational Studies: A Summary Table* (2019.0)
22. Janes H, Donnell D, Gilbert PB, Brown ER, Nason M. **Taking stock of the present and looking ahead: envisioning challenges in the design of future HIV prevention efficacy trials**. *Lancet HIV* (2019.0) **6** e475-e482. PMID: 31078451
23. Janes H, Donnell D, Nason M. **Designing the next generation of HIV prevention efficacy trials: synopsis of a 2018 symposium**. *Stat Commun Infect Dis* (2019.0)
24. Kerr T, Walsh J, Lloyd-Smith E, Wood E. **Antiretroviral prophylaxis for HIV prevention in heterosexual men and women**. *Curr HIV/AIDS Rep* (2005.0) **2** 200-205. PMID: 16343379
25. Knight R, Small W, Thomson K, Gilbert M, Shoveller J. **Implementation challenges and opportunities for HIV treatment as prevention (Tasp) among Young Men in Vancouver, Canada: a qualitative study**. *BMC Public Health* (2016.0) **15** 262
26. Lei H, Nahum-Shani I, Lynch K, Oslin D, Murphy SA. **A SMART design for building individualized treatment sequences**. *Annu Rev Clin Psychol* (2012.0) **8** 21-48. PMID: 22224838
27. Marrazzo J, Ramjee G, Richardson B, Gomez K, Mgodi N, Nair G, Palanee T, Nakabiito C, van der Straten A, Noguchi L, Hendrix C, Dai J, Ganesh S, Mkhize B, Taljaard M, Parikh UM, Piper J, Masse B, Grossman C, Rooney J, Schwartz J, Watts H, Marzinke M, Hillier S, McGowan I, Chirenje Z. **Tenofovir-based preexposure prophylaxis for HIV infection among African women**. *N Engl J Med* (2015.0) **372** 509-518. PMID: 25651245
28. Molina J, Capitant C, Spire B, Pialoux G, Cotte L, Charreau I, Tremblay C, Le Gall J, Cua E, Pasquet A, Raffi F, Pintado C, Chidiac C, Chas J, Charbonneau P, Delaugerre C, Suzan-Monti M, Loze B, Fonsart J, Peytavin G, Cheret A, Timsit J, Girard G, Lorente N, Prau M, Rooney J, Wainberg M, Thompson D, Rozenbaum W, Dor V, Marchand L, Simon M, Etien N, Aboulker J, Meyer L, Delfraissy J. **On-demand preexposure prophylaxis in men at high risk for HIV-1 infection**. *N Engl J Med* (2015.0) **373** 2237-2246. PMID: 26624850
29. Murphy SA. **An experimental design for the development of adaptive treatment strategies**. *Stat Med* (2005.0) **24** 1455-1481. PMID: 15586395
30. Murphy SA, Collins LM, Rush AJ. **Customizing treatment to the patient: adaptive treatment strategies**. *Drug Alcohol Depend* (2007.0) **88** 1-3. PMID: 17049753
31. Murphy SA, Lynch KG, Oslin D, McKay JR, TenHave T. **Developing adaptive treatment strategies in substance abuse research**. *Drug Alcohol Depend* (2007.0) **88** 24-30
32. 32.National Institutes of Allergy and Infectious Diseases: Long-Acting Injectable Drug Prevents HIV Among Men Who Have Sex with Men and Transgender Women (2020). https://www.niaid.nih.gov/news-events/long-acting-injectable-drug-prevents-hiv-among-men-who-have-sex-men-and-transgender. *National Institutes of Allergy and Infectious Diseases: Long-Acting Injectable Drug Prevents HIV Among Men Who Have Sex with Men and Transgender Women* (2020.0)
33. Rolle C, Onwubiko U, Jo J, Sheth A, Kelley C, Holland D. *PrEP Implementation and Persistence in a County Health Department in Atlanta, GA* (2018.0)
34. Serota D, Rosenberg E, Lockard A, del Rio C, Luisi N, Cutro S, Rolle C, Siegler A, Sanchez T, Sullivan P, Kelley C. *PrEP Persistence and Discontinuations in a Cohort of Young Black MSM in Atlanta, GA* (2019.0)
35. Sidebottom D, Ekstrom AM, Stromdahl S. **A systematic review of adherence to oral preexposure prophylaxis for HIV - how can we improve uptake and adherence?**. *BMC Infect Dis* (2018.0) **18** 581. PMID: 30445925
36. Sugarman J. **Bioethical challenges with HIV treatment as prevention**. *Clin Infect Dis* (2014.0) **59** S32-S34. PMID: 24926031
37. 37.The Joint United Nations Programme on HIV/AIDS: Global HIV and AIDS statistics – 2018 fact sheet (Cited 2018). http://www.unaids.org/en/resources/fact-sheet. *The Joint United Nations Programme on HIV/AIDS: Global HIV and AIDS statistics – 2018 fact sheet*
38. Thigpen M, Kebaabetswe P, Paxton L, Smith D, Rose C, Segolodi T, Henderson F, Pathak S, Soud F, Chillag K, Mutanhaurwa R, Chirwa L, Kasonde M, Abebe D, Buliva E, Gvetadze R, Johnson S, Sukalac T, Thomas V, Hart C, Johnson J, Malotte C, Hendrix C, Brooks J. **the TDF2 Study Group: Antiretroviral Preexposure Prophylaxis for Heterosexual HIV Transmission in Botswana**. *N Engl J Med* (2012.0) **367** 423-434. PMID: 22784038
39. Touloumis A. **Simulating correlated binary and multinomial responses under marginal model specification: The simcormultres package**. *The R Journal* (2016.0)
40. Van Damme L, Corneli A, Ahmed K, Agot K, Lombaard J, Kapiga S, Malahleha M, Owino F, Manongi R, Onyango J, Temu L, Monedi M, Mak’Oketch P, Makanda M, Reblin I, Makatu S, Saylor L, Kiernan H, Kirkendale S, Wong C, Grant R, Kashuba A, Nanda K, Mandala J, Fransen K, Deese J, Crucitti T, Mastro T, Taylor D. **Preexposure prophylaxis for HIV infection among African women**. *N Engl J Med* (2012.0) **367** 411-422. PMID: 22784040
41. Wilson D, Taaffe J, Fraser-Hurt N, Gorgens M. **The economics, financing and implementation of HIV treatment as prevention: what will it take to get there?**. *Afr J AIDS Res* (2014.0) **13** 109-119. PMID: 25174628
42. Zheng JH, Rower C, McAllister K, Castillo-Mancilla J, Klein B, Meditz A, Guida LA, Kiser JJ, Bushman LR, Anderson PL. **Application of an intracellular assay for determination of tenofovir-diphosphate and emtricitabine-triphosphate from erythrocytes using dried blood spots**. *J Pharm Biomed Anal* (2016.0) **122** 16-20. PMID: 26829517
43. Zhu Y, Di C, Chen YQ. **Clustering functional data with application to electronic medication adherence monitoring in HIV prevention trials**. *Statistics in Biosciences* (2019.0) 1-24
|
---
title: 'Epidemiological status and associated factors of frailty and pre-frailty in
older adults with asthma in China: A national cross-sectional study'
authors:
- Xue-zhai Zeng
- Ling-bing Meng
- Na Jia
- Jing Shi
- Chi Zhang
- Ying-ying Li
- Xing Hu
- Jia-bin Hu
- Jian-yi Li
- Di-shan Wu
- Hui Li
- Xin Qi
- Hua Wang
- Qiu-xia Zhang
- Juan Li
- De-ping Liu
journal: Frontiers in Public Health
year: 2023
pmcid: PMC10022817
doi: 10.3389/fpubh.2023.1136135
license: CC BY 4.0
---
# Epidemiological status and associated factors of frailty and pre-frailty in older adults with asthma in China: A national cross-sectional study
## Abstract
### Objective
There are few studies on the prevalence and factors associated with frailty and pre-frailty in older adults with asthma worldwide. The aim of this study was to examine the epidemiological status and factors associated with frailty and pre-frailty in older adults with asthma in China.
### Research design and methods
Data were obtained from the Sample Survey of Aged Population in Urban and Rural China in 2015, a nationwide cross-sectional survey covering 224,142 older people aged 60 years or older in 31 provinces/autonomous regions/municipalities in mainland China. We performed frailty and pre-frailty assessments using the frailty index, and the diagnosis of asthma in the older adults was self-reported based on the history of the physician's diagnosis.
### Results
Nine thousand four hundred sixteen older adults with asthma were included in the study. The age-sex standardized prevalence of frailty and pre-frailty in Chinese older adults with asthma was $35.8\%$ ($95\%$ CI $34.8\%$−$36.7\%$) and $54.5\%$ ($95\%$ CI $53.5\%$−$55.5\%$). Multinomial logistic regression analysis showed that increased age, female, illiteracy, living alone, poor economic status, ADL disability, comorbid chronic diseases, previous hospitalization in the past year, and residence in northern China were associated with frailty and pre-frailty in older adults with asthma.
### Conclusion
The prevalence of frailty and pre-frailty in Chinese older adults with asthma is very high, and assessment of frailty should become routine in the management of older adults with asthma. Appropriate public health prevention strategies based on identified risk factors for frailty in older adults with asthma should be developed to reduce the burden of frailty in Chinese older adults with asthma.
## Introduction
Asthma is a common non-communicable lung disease, with an aging population, the number of older people suffering from asthma is increasing. Epidemiological data show that the prevalence of asthma in people aged 65 years or over is $4\%$−$15\%$ (1–3). The 2017 global burden of disease study revealed asthma to be the second most prevalent chronic respiratory disease after chronic obstructive respiratory disease and the second leading cause of death from chronic respiratory disease [4]. Asthma poses a serious threat to the quality of life and health of older people, and increases the use and burden of healthcare resources on society [5]. Frailty is a clinical syndrome characterized by reduced physiological reserve and multisystem dysregulation, which limits the body's ability to respond to internal and external stresses and maintain the stability of the internal environment, increasing the body's susceptibility to stressful events (6–8). The weighted prevalence of frailty in older people in the community is $11\%$ (range $4\%$−$59\%$) [9]. Frail older people are at significantly increased risk of falls, delirium, incapacity, hospitalization and death, and increase the burden on society's healthcare resources [8, 10, 11]. There are few large-scale studies in the world on frailty in older adults with asthma. The paucity of research on frailty in older patients with asthma has led to many challenges in the comprehensive management of asthma in older adults. On top of the existing medical care for older adults with asthma, frailty assessment can provide additional valuable information and help clinicians choose better medical care for their patients. This study applied data from the 2015 Sample Survey of Aged Population in Urban and Rural China (SSAPUR) to analyse the prevalence of frailty and pre-frailty and their associated factors in older adults with asthma in China. The aim was to provide information to reduce the decline in cognitive and physical functioning and to prevent frailty and disability in older people with asthma.
## Study design and participants
Data were obtained from the 4th SSAPUR, a cross-sectional study of 224,142 older adults aged 60 years or older in 31 provinces, autonomous regions, and municipalities in mainland China in 2015. The survey used a stratified, multi-stage, proportional probability sampling by size and equal probability sampling design in the final stage. The sampling proportion was ~1 in 1,000 of the national older population in 2015, and the sample obtained was self-weighted to ensure national representativeness. The sample number was assigned based on the proportion of the older population in each province, autonomous regions, and municipalities of the country, after that the number of counties, towns and communities sampled was determined and 462 counties were selected. Based on PPS sampling, four towns were selected from each county and four communities (villages or neighborhood councils) were selected from each town. Finally, 30 older adults were selected from each community using equidistant sampling. Data on the living conditions of older adults were collected through household interviews and questionnaires. More information on the design and sampling methods of the 4th SSAPUR study has been reported in previous studies (12–14). We used the frailty index (FI) for frailty and pre-frailty assessment, and 15,756 ($7.0\%$) older adults were excluded because the number of constructed FI items was <28. Of the 208,386 older adults, 9,416 older adults with asthma, determined on the basis of a self-reported history of diagnosis by a physician, were included in our study (Figure 1). The study protocol was approved by the National Bureau of Statistics (No. [ 2014] 87) and the Ethics Committee of Beijing Hospital (2021BJYYEC-294-01). Written informed consent was provided by all participants.
**Figure 1:** *Flowchart of study participants on frailty and pre-frailty prevalence in older adults with asthma in China.*
## Demographics
Demographic characteristics: age, sex, education, marital status, ethnicity, residence, living status, health checkup within the past year, hospitalization within the past year, financial status, ease of medical reimbursement, activities of daily living (ADL) disability (Inability to do one or more of the following: bathing, dressing, toileting, getting in and out of bed, eating and moving around the room is considered a disability), comorbid chronic diseases, and residence in southern or northern China.
## Identification and assignment of health deficit variables for FI
FI refers to the proportion of deficits that are present when a person undergoes a health assessment. We constructed FI following Searle's standard procedure [15]. The FI items ($$n = 33$$, each subject needs at least $\frac{28}{33}$ variables) were selected from the baseline questionnaires of demographic characteristics, physical health, physical functioning, lifestyle, social activity, and mental health status. The variables included eight items of basic activities of daily living (bathing, dressing, toileting, getting in and out of bed, eating, walking around the room, urinary incontinence, and fecal incontinence); 10 items focusing on chronic diseases included glaucoma/cataract, cardiovascular disease, hypertension, diabetes, gastric disease, bone and joint disease, chronic lung disease, asthma, malignancy, and reproductive system disease; two items focused on feelings of loneliness and happiness; three items focused on geriatric syndrome, including visual impairment, hearing impairment, and history of falls; five items focused on assistive devices (hearing aids, dentures, crutches, wheel-chairs, and adult diapers/nursing pads); three items focused on mobility (needing care from others in daily life, self-rated health status, and exercise); two items focused on social activity (regular leisure activities and regular public service activities). The FI was calculated by summing the number of deficits recorded for a patient and dividing this by the total number of possible deficits. The exact construction method has been described in our previous study [12]. FI scores ≥0.25 are considered frailty, <0.12 are considered robust, and FI 0.12–0.25 are considered pre-frailty.
## Statistical analysis
We used SPSS 24.0 software for statistical analysis. Missing data were interpolated using the Markov chain Monte Carlo (MCMC) multiple fill method [15]. The age-standardized prevalence of frailty and pre-frailty among Chinese older adults with asthma was calculated based on the weights established in our study. For continuous variables, we assessed the significance of differences by ANOVA or Student's t-test, and for categorical variables, we passed the χ2 test. Trends in prevalence of covariates were examined by using the Cochran-Armitage test. Multinomial regression analysis was used to identify factors associated with frailty and pre-frailty, including age group, gender, ethnicity, urban/rural, education level, marital status, living alone, economic status, health insurance, ease of medical reimbursement, comorbid chronic diseases, ADL disability, and residence in southern or northern China, with $p \leq 0.05$ being statistically significant.
## Results
From August 1, 2015 to August 31, 2015, 224,142 older adults aged 60 years or older were invited to participate in the fourth SSAPUR, of which 15,756 participants were excluded because fewer than 28 items were used to construct the FI. Nine thousand four hundred sixteen older adults with self-reported asthma out of 208,386, and the self-reported prevalence of asthma among older adults was $4.5\%$, with $4.8\%$ prevalence in men and $4.3\%$ prevalence in women. Demographics and frailty risk factors by frailty stage in older adults with asthma are shown in Table 1, demographics and frailty risk factors for older asthmatics by North and South are shown in Supplemental Table S1.
**Table 1**
| Unnamed: 0 | Total (n = 9,416) | Men (n = 4,774) | Women (n = 4,642) | p-value | Men | Men.1 | Men.2 | Men.3 | Women | Women.1 | Women.2 | Women.3 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | | | | | Robust (546) | Pre-frailty (2,762) | Frailty (1,466) | p for difference | Robust (383) | Pre-frailty (2,380) | Frailty (1,879) | p for difference |
| Proportional of participants | 100% | 50.7% | 49.3% | | 11.4% | 57.9% | 30.7% | <0.001 | 8.3% | 51.3% | 40.5% | <0.001 |
| Age (years) | 71.9 ± 8.1 | 71.4 ± 7.9 | 72.3 ± 8.3 | <0.001 | 69.8 ± 7.6 | 71.0 ± 7.6 | 72.8 ± 8.1 | <0.001 | 70.6 ± 8.1 | 71.5 ± 7.9 | 73.7 ± 8.6 | <0.001 |
| Age group | Age group | Age group | Age group | Age group | Age group | Age group | Age group | Age group | Age group | Age group | Age group | Age group |
| 60–69 | 44.4% | 45.5% | 43.2% | <0.001 | 56.2% | 47.4% | 37.8% | <0.001 | 51.9% | 47.2% | 36.5% | <0.001 |
| 70–79 | 35.8% | 36.9% | 34.7% | | 30.8% | 36.9% | 39.4% | | 31.1% | 34.3% | 35.9% | |
| ≥80 | 19.8% | 17.6% | 22.1% | | 13.0% | 15.7% | 22.8% | | 17.0% | 18.5% | 27.6% | |
| Urban or rural area (Urban) | 41.4% | 41.0% | 41.9% | 0.361 | 44.9% | 40.2% | 40.9% | 0.130 | 47.3% | 41.1% | 41.7% | 0.078 |
| Education (illiteracy) | 36.9% | 19.0% | 55.3% | <0.001 | 15.0% | 18.8% | 20.9% | 0.010 | 45.7% | 52.4% | 60.9% | <0.001 |
| Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status | Marital status |
| Married | 65.9% | 76.7% | 54.8% | <0.001 | 83.5% | 76.6% | 74.1% | <0.001 | 68.4% | 58.8% | 47.0% | <0.001 |
| Widowed | 31.4% | 18.6% | 44.7% | | 13.6% | 17.8% | 22.0% | | 31.6% | 40.6% | 52.4% | |
| Divorced | 0.8% | 1.1% | 0.4% | | 1.1% | 1.4% | 0.8% | | 0 | 0.5% | 0.5% | |
| Unmarried | 1.9% | 3.6% | 0.1% | | 1.8% | 4.2% | 3.1% | | 0 | 0.1% | 0.1% | |
| Ethnicity (non-Han) | 7.0% | 6.5% | 7.6% | 0.035 | 6.8% | 6.4% | 6.3% | 0.940 | 9.1% | 8.4% | 6.1% | 0.008 |
| Living alone | 16.2% | 13.8% | 18.5% | <0.001 | 5.0% | 15.9% | 24.6% | <0.001 | 6.2% | 14.0% | 16.4% | <0.001 |
| Health checkup within 1 year | 57.8% | 58.2% | 57.4% | 0.469 | 56.8% | 58.4% | 58.3% | 0.782 | 53.0% | 59.9% | 55.2% | 0.002 |
| Hospitalized within 1 year | 44.6% | 44.7% | 44.4% | 0.800 | 70.7% | 58.3% | 54.1% | <0.001 | 76.2% | 59.2% | 46.8% | <0.001 |
| Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status | Economic status |
| Very rich | 0.8% | 0.8% | 0.7% | 0.545 | 1.5% | 0.8% | 0.7% | <0.001 | 1.3% | 0.7% | 0.7% | <0.001 |
| Rich | 12.0% | 12.1% | 12.0% | | 16.9% | 12.5% | 9.7% | | 21.7% | 12.8% | 9.1% | |
| Adequate | 55.1% | 55.8% | 54.4% | | 63.7% | 57.7% | 49.2% | | 57.4% | 58.1% | 49.0% | |
| Poor | 26.6% | 25.9% | 27.2% | | 15.9% | 25.1% | 31.2% | | 18.3% | 25.0% | 31.9% | |
| Very poor | 5.5% | 5.4% | 5.7% | | 2.0% | 3.9% | 9.2% | | 1.3% | 3.4% | 9.3% | |
| Medicare (no) | 1.0% | 1.0% | 1.0% | 0.810 | 0.9% | 0.9% | 1.0% | 0.960 | 0.5% | 0.9% | 1.3% | 0.268 |
| Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement | Convenience of medical cost reimbursement |
| Highly convenient | 30.6% | 30.2% | 31.1% | 0.555 | 36.6% | 29.3% | 29.5% | <0.001 | 28.7% | 31.8% | 30.5% | 0.003 |
| Convenient | 43.4% | 43.1% | 43.6% | | 42.5% | 44.0% | 41.7% | | 50.9% | 42.7% | 43.3% | |
| Less convenient | 18.9% | 19.6% | 18.2% | | 16.1% | 20.2% | 19.7% | | 15.9% | 19.1% | 17.7% | |
| Inconvenient | 5.0% | 5.0% | 4.9% | | 3.1% | 4.8% | 6.2% | | 3.7% | 4.1% | 6.2% | |
| Highly inconvenient | 2.1% | 2.1% | 2.2% | | 1.7% | 1.7% | 2.9% | | 0.8% | 2.3% | 2.3% | |
| Comorbidities (≥1) | 92.0% | 90.9% | 93.2% | <0.001 | 48.5% | 94.7% | 99.6% | <0.001 | 53.3% | 94.6% | 99.5% | <0.001 |
| ADL disability | 7.8% | 6.6% | 9.1% | <0.001 | 0.2% | 2.0% | 17.8% | <0.001 | 0 | 2.0% | 20.0% | <0.001 |
| Living southern or northern China (Northern China) | 25.3% | 22.9% | 27.7% | <0.001 | 15.4% | 21.1% | 29.1% | <0.001 | 21.9% | 23.4% | 34.4% | <0.001 |
The FI of older adults with asthma was gamma distributed with a statistical value of 0.075, $p \leq 0.0001.$ The FI of older adults with asthma was 0.21 (0.11; ranging from 0.04–0.70), with 0.23 (0.12) for female asthmatics, which was higher than that of male asthmatics 0.20 (0.11; z = −11.686, $p \leq 0.001$).
The prevalence of frailty and pre-frailty in older adults with asthma was $35.5\%$ and $54.6\%$, respectively, significantly higher than the prevalence of frailty ($8.2\%$) and pre-frailty ($46.4\%$) in older adults without asthma, both $p \leq 0.0001.$ The prevalence of frailty was higher in female older adults with asthma ($40.5\%$) than in males ($30.7\%$), while the prevalence of pre-frailty was lower in female older adults with asthma ($51.3\%$) than in males ($57.9\%$), both $p \leq 0.001.$ The age-sex standardized prevalence of frailty and pre-frailty among Chinese older adults with asthma was $35.8\%$ ($95\%$ CI $34.8\%$−$36.7\%$) and $54.5\%$ ($95\%$ CI $53.5\%$−$55.5\%$). The age-sex standardized prevalence of frailty and pre-frailty was $40.9\%$ ($95\%$ CI $39.5\%$−$42.3\%$) and $51.4\%$ ($95\%$ CI $49.9\%$−$52.8\%$) for women, and $30.8\%$ ($95\%$ CI $29.5\%$−$32.1\%$) and $57.5\%$ ($95\%$ CI $56.1\%$−$58.9\%$) for men.
The prevalence of frailty in older adults with asthma increased with age, from $29.7\%$ in the 60–69 years age group to $45.8\%$ in the ≥80 years age group. There was no difference in the prevalence of frailty between rural and urban older adults with asthma. The prevalence of frailty was higher in Han Chinese older adults with asthma than in non-Han Chinese, mainly in Han Chinese women than in non-Han Chinese women. Frailty was mostly seen in older adults with asthma who were illiterate, widowed, living alone, had been hospitalized in the past 1 year, had financial difficulties, had difficulties in reimbursing medical expenses, and had comorbid chronic diseases and disabilities. The prevalence of frailty was higher among older adults with asthma in northern China than in southern China (Table 2). The prevalence of frailty among older adults with asthma in the seven administrative regions of mainland China was highest in northwest China, followed by north China, then northeast China, and then southwest, central, southeast, and south China (Figure 2).
Multinomial regression analysis revealed that female, increased age, illiteracy, living alone, hospitalization in the past 1 year, difficult financial situation, comorbid chronic diseases, ADL disability, and north of residence were risk factors for frailty and pre-frailty in older adults with asthma (Table 3).
**Table 3**
| Variables | Variables.1 | Pre-frailty vs. robust | Pre-frailty vs. robust.1 | Pre-frailty vs. robust.2 | Pre-frailty vs. robust.3 | Frailty vs. robust | Frailty vs. robust.1 | Frailty vs. robust.2 | Frailty vs. robust.3 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| | | OR | 95%CI | 95%CI | p -value | OR | 95%CI | 95%CI | p -value |
| | | | Lower | Upper | | | Lower | Upper | |
| Sex | Male | 1 (ref) | | | | | | | |
| Sex | Female | 1.1094 | 0.909 | 1.316 | 0.341 | 1.423 | 1.163 | 1.742 | <0.001 |
| Age (years) | 60–69 | 1 (ref) | | | | | | | |
| Age (years) | 70–79 | 1.364 | 1.126 | 1.652 | 0.001 | 1.618 | 1.313 | 1.993 | <0.001 |
| Age (years) | ≥80 | 1.830 | 1.388 | 2.412 | <0.001 | 2.469 | 1.833 | 3.326 | <0.001 |
| Urban or rural area | Urban | 1 (ref) | | | | | | | |
| Urban or rural area | Rural | 1.250 | 1.052 | 1.482 | 0.011 | 1.211 | 1.004 | 1.460 | 0.046 |
| Marriage | Married | 1 (ref) | | | | | | | |
| Marriage | Widowed | 0.938 | 0.740 | 1.189 | 0.595 | 0.945 | 0.731 | 1.221 | 0.663 |
| Marriage | Divorced | 1.249 | 0.444 | 3.512 | 0.674 | 0.870 | 0.278 | 2.723 | 0.811 |
| Marriage | Unmarried | 1.632 | 0.752 | 3.543 | 0.595 | 1.056 | 0.452 | 2.456 | 0.901 |
| Education | Non-illiterate | 1 (ref) | | | | | | | |
| Education | Illiterate | 1.341 | 1.090 | 1.650 | 0.006 | 1.555 | 1.243 | 1.944 | <0.001 |
| Ethnicity | Han | 1 (ref) | | | | | | | |
| Ethnicity | Others | 0.972 | 0.704 | 1.343 | 0.863 | 0.769 | 0.538 | 1.099 | 0.149 |
| Living alone | No | 1 (ref) | | | | | | | |
| Living alone | Yes | 4.810 | 3.298 | 7.014 | <0.001 | 8.691 | 5.848 | 12.917 | <0.001 |
| Medical checkup within 1 year | Yes | 1 (ref) | | | | | | | |
| Medical checkup within 1 year | No | 0.927 | 0.782 | 1.100 | 0.385 | 1.021 | 0.370 | 2.820 | 0.968 |
| Hospitalized within 1 year | No | 1 (ref) | | | | | | | |
| Hospitalized within 1 year | Yes | 1.609 | 1.342 | 1.929 | <0.001 | 2.670 | 2.198 | 3.244 | <0.001 |
| Economic status | Very rich | 1 (ref) | | | | | | | |
| Economic status | Rich | 1.608 | 0.772 | 3.351 | 0.205 | 1.371 | 0.586 | 3.209 | 0.466 |
| Economic status | Adequate | 2.175 | 1.063 | 4.448 | 0.033 | 2.257 | 0.986 | 5.165 | 0.054 |
| Economic status | Poor | 4.077 | 1.948 | 8.531 | 0.0002 | 6.658 | 2.849 | 15.558 | <0.001 |
| Economic status | Very poor | 9.953 | 3.912 | 25.326 | <0.001 | 33.890 | 12.074 | 95.126 | <0.001 |
| Medical reimbursement | Very convenient | 1 (ref) | | | | | | | |
| Medical reimbursement | Convenient | 1.034 | 0.854 | 1.252 | 0.729 | 1.026 | 0.832 | 1.265 | 0.809 |
| Medical reimbursement | Less convenient | 1.203 | 0.936 | 1.547 | 0.149 | 1.122 | 0.854 | 1.474 | 0.408 |
| Medical reimbursement | Inconvenient | 1.328 | 0.828 | 2.130 | 0.239 | 1.569 | 0.951 | 2.588 | 0.078 |
| Medical reimbursement | Very inconvenient | 1.654 | 0.771 | 3.549 | 0.197 | 1.846 | 0.829 | 4.110 | 0.133 |
| Co-morbidities | <1 | 1 (ref) | | | | | | | |
| Co-morbidities | ≥1 | 25.971 | 21.077 | 32.001 | <0.001 | 556.186 | 309.531 | 999.391 | <0.001 |
| ADL disabilities | No | 1 (ref) | | | | | | | |
| ADL disabilities | Yes | 49.491 | 6.632 | 369.356 | <0.001 | 613.664 | 81.615 | 4614.157 | <0.001 |
| Medicare | Yes | 1 (ref) | | | | | | | |
| Medicare | No | 0.976 | 0.374 | 2.547 | 0.961 | 1.021 | 0.370 | 2.820 | 0.968 |
| Living in southern or northern China | Southern | 1 (ref) | | | | | | | |
| Living in southern or northern China | Northern | 1.453 | 1.171 | 1.803 | <0.001 | 2.229 | 1.772 | 2.804 | <0.001 |
## Discussion
Asthma has long been recognized as a disease that is prevalent in adolescents. Our study showed a self-reported asthma prevalence of $4.5\%$ in older adults, indicating that the prevalence of asthma in older adults is not low. Data from the 2010–2012 China Epidemiological Survey of Asthma Prevalence and Risk Factors study showed that the prevalence of asthma increased with age, with $2.26\%$ of people aged 61–70 years and $3.10\%$ of people aged ≥71 years [2], and data from the China Adult Lung Health Study from 2012 to 2015 showed that the prevalence of asthma was $6.0\%$ in people aged 60–69 years and $7.4\%$ in people aged ≥70 years [3]. The above study showed that asthma prevalence was not low among older adults in China. Foreign epidemiological data showed that the prevalence of asthma in people aged 65 years or over was $4\%$−$15\%$ [1, 16]. Asthma surveillance data released by the Centers for Disease Control and Prevention show that the prevalence of asthma among older adults ≥65 years of age in the United States was $8.1\%$ in 2010 [1]. The 2017 Global Burden of Disease Study shows that the prevalence of asthma has decreased since 1990, from $3.9\%$ in 1990 to $3.6\%$ in 2017, but it remains the second most prevalent chronic respiratory disease after chronic obstructive respiratory disease, and asthma is also the second leading cause of death from chronic respiratory disease [4].
The rate of severe asthma and mortality in older adults with asthma is higher than in other age groups. The risk of severe and frequent acute exacerbations, resulting in frequent hospitalizations or emergency room visits, significantly increases the direct medical costs of older adults with asthma [5]. Prevention and early management of asthma, a common non-communicable disease in older adults, is also one of the key goals for achieving healthy aging, and studies of frailty as an assessment indicator of biological aging have shown a significant increase in negative clinical events in frail older adults [10]; therefore, studying the frailty status and risk factors of older asthmatics can help inform public health policymakers to reduce cognitive and physical decline and prevent frailty and disability in older asthmatics.
Our large cross-sectional national study used a stratified, multi-stage, size-proportional probability sampling and a final-stage equal-probability sampling design with a sample covering 31 provinces, municipalities, and autonomous regions in mainland China with different geographic regions and economic development status, and the age distribution, gender, and urban-rural ratios of older adults in the obtained sample were consistent with the demographic characteristics of older adults in the 2015 China Population Survey, ensuring national representativeness. Our study accurately reports the prevalence of frailty and pre-frailty in older Chinese patients with asthma. Our study found a high age-sex standardized prevalence of frailty in older adults with asthma of $35.8\%$ ($95\%$ CI $34.8\%$−$36.7\%$) and an age-sex standardized prevalence of pre-frailty of $54.5\%$ ($95\%$ CI $53.5\%$−$55.5\%$), with a 3.3-fold increase in the prevalence of frailty in older adults with asthma compared to older adults without asthma. The results of our study are generally consistent with those of a small study who found a $37\%$ prevalence of frailty among 203 older outpatients with asthma aged 60 years or older [17], and we report a higher prevalence of frailty among older adults with asthma than a Brazilian cohort study that reported $13\%$ of older patients with current asthma had frailty in 2015 [18]. Chronic inflammation in asthmatics is not only present in the respiratory tract but is systemic, characterized by increased levels of peripheral blood eosinophils, immunoglobulin E and type 2 cytokines. Recent studies have found that chronic systemic inflammation is associated with the development of frailty and that chronic systemic inflammation in older adults with asthma may be an important cause of frailty (19–21). Our study showed that the proportion of older asthmatics who never exercised was significantly higher than that of older adults without asthma, and that the reduction in exercise also resulted in older asthmatics being more likely to develop sarcopenia, one of the key factors in the development of frailty syndromes. Until the recent advent of biologic agents, oral corticosteroids (OCS) have been the key controller medication for the treatment of severe intractable asthma. Ryu et al. reported a higher prevalence of frailty in older adults with asthma with longer lifetime OCS exposure ($33\%$ of patients with no lifetime OCS use, $59\%$ of low-dose users, and $68\%$ of high-dose users; $p \leq 0.005$ for trend). Suggesting that lifetime cumulative OCS exposure was associated with a high prevalence of weakness and muscle weakness [17]. Although OCS is effective for asthma, it causes side effects including osteoporosis, fractures, diabetes, obesity, cardiovascular disease, and infections that may promote the onset of frailty. These aforementioned factors may promote the development of frailty in older adults with asthma.
Our study found that being female, increasing age, illiteracy, living alone, hospitalization in the past 1 year, economic hardship, comorbid chronic diseases, ADL disability, and living in northern China were risk factors for frailty and pre-frailty in older asthmatic patients. Studies have shown that aging is an independent risk factor for the onset of frailty, as older adults age, several physiological systems throughout the body undergo degenerative changes, the reserve function of various organs decreases, and the risk of frailty increases [7, 22]. Decreased estrogen levels in older women lead to a decrease in muscle strength and a negative impact on neuromuscular function and postural stability, leading to an increased incidence of frailty in older women [23]. Educational attainment is highly correlated with income, and difficult economic status makes it difficult to obtain adequate medical care, which increases the risk of frailty [24, 25]. Older asthmatics who are living alone are more likely to experience loneliness and depression due to decreased family support, leading to an increased risk of frailty [26]. Older adults with asthma have multiple co-morbidities and these chronic diseases contribute to the increased risk of frailty in older adults with asthma [27]. Therefore, in the management of older adults with asthma, attention should be paid to identifying risk factors for frailty and providing early targeted interventions to reduce or delay the risk of frailty in older adults with asthma.
Compared with male older adults with asthma, our study found that female older adults with asthma were older, had higher rates in the advanced age group, higher rates of illiteracy, widowhood, living alone, comorbid chronic diseases, disability, and hospitalization in the past 1 year, which are factors associated with higher prevalence of frailty and pre-frailty in female older adults with asthma than in men, and the findings of our study provide information for policy makers to take appropriate measures to reduce and prevent the occurrence of frailty in female asthmatics.
Our study also found that the prevalence of frailty was higher among older adults with asthma living in the north than in the south. We found higher rates of female, widowed, no physical examination in the past 1 year, difficult financial status, inconvenient medical reimbursement, comorbid chronic diseases and ADL disability among older adults with asthma in northern China than in southern China. Differences in the demographic characteristics of older adults with asthma and risk factors for frailty in northern and southern China may explain the higher prevalence of frailty among older adults with asthma in the north than in the south. These findings provide information for policy makers to take appropriate measures to reduce and prevent the occurrence of frailty in asthmatic patients in northern China.
Limitations: firstly, the self-reported diagnostic information of asthma in this study may be subject to recall bias, secondly, there is a lack of information on the course and extent of disease in older adults with asthma, and at least two phenotypes exist in older adults with asthma; long-term asthmatics have more severe airflow limitation and are less fully reversible than asthmatics with late onset asthma. Third, smoking is an important risk factor for chronic airway disease, and this study did not assess the association between smoking and frailty in older adults with asthma. Fourth, this study was a cross-sectional study and could not determine the causal relationship between associated factors and frailty in older adults with asthma.
In conclusion, the prevalence of frailty and pre-frailty in older adults with asthma is very high, and frailty assessment should become a routine in the management of older adults with asthma, and attention should be paid to the early identification of risk factors for frailty in older adults with asthma and targeted interventions to prevent and delay the onset of frailty in older adults with asthma.
## Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.
## Ethics statement
The studies involving human participants were reviewed and approved by the National Bureau of Statistics (No. [ 2014] 87) and the Ethics Committee of Beijing Hospital (2021BJYYEC-294-01). The patients/participants provided their written informed consent to participate in this study.
## Author contributions
X-zZ and NJ wrote the various drafts of the manuscript. X-zZ, L-bM, and JS conducted the statistical analyses. CZ, Y-yL, J-bH, HL, XQ, HW, XH, D-sW, and J-yL participated in data interpretation. D-pL, Q-xZ, JL, and X-zZ conceived and designed this study. X-zZ, NJ, L-bM, CZ, Y-yL, J-bH, JS, HL, XQ, HW, XH, D-sW, J-yL, Q-xZ, JL, and D-pL were revised the drafts of the manuscript for important scientific content. D-pL was the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors gave final approval of the version to be published.
## Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
## Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
## Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2023.1136135/full#supplementary-material
## References
1. Moorman JE, Akinbami LJ, Bailey CM, Zahran HS, King ME, Johnson CA. **National surveillance of asthma: United States, 2001-2010**. *Vital Health Stat* (2012) **35** 1-58. PMID: 24252609
2. Lin J, Wang W, Chen P, Zhou X, Wan H, Yin K. **Prevalence and risk factors of asthma in mainland China: the CARE study**. *Respir Med.* (2018) **137** 48-54. DOI: 10.1016/j.rmed.2018.02.010
3. Huang K, Yang T, Xu J, Yang L, Zhao J, Zhang X. **Prevalence, risk factors, and management of asthma in China: a national cross-sectional study**. *Lancet.* (2019) **394** 407-18. DOI: 10.1016/S0140-6736(19)31147-X
4. Collaborators GBDCRD. **Prevalence and attributable health burden of chronic respiratory diseases, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017**. *Lancet Respir Med.* (2020) **8** 585-96. DOI: 10.1016/S2213-2600(20)30105-3
5. Skloot GS, Busse PJ, Braman SS, Kovacs EJ, Dixon AE, Vaz Fragoso CA. **An Official American Thoracic Society Workshop Report: evaluation and management of asthma in the elderly**. *Ann Am Thorac Soc.* (2016) **13** 2064-77. DOI: 10.1513/AnnalsATS.201608-658ST
6. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J. **Frailty in older adults: evidence for a phenotype**. *J Gerontol A Biol Sci Med Sci.* (2001) **56** M146-56. DOI: 10.1093/gerona/56.3.M146
7. Rockwood K, Howlett SE. **Age-related deficit accumulation and the diseases of ageing**. *Mech Ageing Dev.* (2019) **180** 107-16. DOI: 10.1016/j.mad.2019.04.005
8. Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. **Frailty in elderly people**. *Lancet.* (2013) **381** 752-62. DOI: 10.1016/S0140-6736(12)62167-9
9. Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. **Prevalence of frailty in community-dwelling older persons: a systematic review**. *J Am Geriatr Soc.* (2012) **60** 1487-92. DOI: 10.1111/j.1532-5415.2012.04054.x
10. Hoogendijk EO, Afilalo J, Ensrud KE, Kowal P, Onder G, Fried LP. **Frailty: implications for clinical practice and public health**. *Lancet.* (2019) **394** 1365-75. DOI: 10.1016/S0140-6736(19)31786-6
11. Jin HY, Liu X, Xue QL, Chen S, Wu C. **The association between frailty and healthcare expenditure among Chinese older adults**. *J Am Med Dir Assoc.* (2020) **21** 780-5. DOI: 10.1016/j.jamda.2020.03.008
12. Zeng X, Jia N, Meng L, Shi J, Li Y, Hu X. **A study on the prevalence and related factors of frailty and pre-frailty in the older population with diabetes in China: a national cross-sectional study**. *Front Public Health* (2022) **10** 996190. DOI: 10.3389/fpubh.2022.996190
13. Dang JW, WY NN. *Survey Report on the Living Conditions of China's Urban and Rural Older Persons.* (2018)
14. Meng L, Xu J, Li J, Hu J, Xu H, Wu D. **Self-reported prevalence and potential factors influencing cardio-cerebral vascular disease among the Chinese elderly: a national cross-sectional study**. *Front Cardiovasc Med.* (2022) **9** 979015. DOI: 10.3389/fcvm.2022.979015
15. Searle SD, Mitnitski A, Gahbauer EA, Gill TM, Rockwood K. **A standard procedure for creating a frailty index**. *BMC Geriatr.* (2008) **8** 24. DOI: 10.1186/1471-2318-8-24
16. McHugh MK, Symanski E, Pompeii LA, Delclos GL. **Prevalence of asthma among adult females and males in the United States: results from the National Health and Nutrition Examination Survey (NHANES), 2001-2004**. *J Asthma.* (2009) **46** 759-66. DOI: 10.3109/02770900903067895
17. Ryu K, Fukutomi Y, Nakatani E, Iwata M, Nagayama K, Yano K. **Frailty and muscle weakness in elderly patients with asthma and their association with cumulative lifetime oral corticosteroid exposure**. *Allergol Int.* (2022). DOI: 10.1016/j.alit.2022.10.005
18. Landre B, Nadif R, Goldberg M, Gourmelen J, Zins M, Ankri J. **Asthma is associated with frailty among community-dwelling adults: the GAZEL cohort**. *BMJ Open Respir Res* (2020) **7** e000526. DOI: 10.1136/bmjresp-2019-000526
19. Wu IC, Lin CC, Hsiung CA. **Emerging roles of frailty and inflammaging in risk assessment of age-related chronic diseases in older adults: the intersection between aging biology and personalized medicine**. *Biomedicine.* (2015) **5** 1. DOI: 10.7603/s40681-015-0001-1
20. **[Chinese guidelines for the diagnosis and treatment of allergic asthma (2019, the first edition)]**. *Zhonghua Nei Ke Za Zhi* (2019) **58** 636-55. DOI: 10.3760/cma.j.issn.0578-1426.2019.09.004
21. Hanania NA, King MJ, Braman SS, Saltoun C, Wise RA, Enright P. **Asthma in the elderly: current understanding and future research needs–a report of a National Institute on Aging (NIA) workshop**. *J Allergy Clin Immunol.* (2011) **128** S4-24. DOI: 10.1016/j.jaci.2011.06.048
22. Kojima G, Iliffe S, Taniguchi Y, Shimada H, Rakugi H, Walters K. **Prevalence of frailty in Japan: a systematic review and meta-analysis**. *J Epidemiol.* (2017) **27** 347-53. DOI: 10.1016/j.je.2016.09.008
23. Gordon EH, Peel NM, Samanta M, Theou O, Howlett SE, Hubbard RE. **Sex differences in frailty: a systematic review and meta-analysis**. *Exp Gerontol.* (2017) **89** 30-40. DOI: 10.1016/j.exger.2016.12.021
24. Li Y, Xue QL, Odden MC, Chen X, Wu C. **Linking early life risk factors to frailty in old age: evidence from the China Health and Retirement Longitudinal Study**. *Age Ageing.* (2020) **49** 208-17. DOI: 10.1093/ageing/afz160
25. Dugravot A, Fayosse A, Dumurgier J, Bouillon K, Rayana TB, Schnitzler A. **Social inequalities in multimorbidity, frailty, disability, and transitions to mortality: a 24-year follow-up of the Whitehall II cohort study**. *Lancet Public Health.* (2020) **5** e42-50. DOI: 10.1016/S2468-2667(19)30226-9
26. Ge L, Yap CW, Heng BH. **Associations of social isolation, social participation, and loneliness with frailty in older adults in Singapore: a panel data analysis**. *BMC Geriatr.* (2022) **22** 26. DOI: 10.1186/s12877-021-02745-2
27. Hanlon P, Nicholl BI, Jani BD, Lee D, McQueenie R, Mair FS. **Frailty and pre-frailty in middle-aged and older adults and its association with multimorbidity and mortality: a prospective analysis of 493 737 UK Biobank participants**. *Lancet Public Health.* (2018) **3** e323-32. DOI: 10.1016/S2468-2667(18)30091-4
|
---
title: Efficacy of Young Cinnamomum zeylanicum Blume Bark on Hyperglycemia and PTPase
Activity in Type 2 Diabetes
journal: Cureus
year: 2023
pmcid: PMC10022837
doi: 10.7759/cureus.35023
license: CC BY 3.0
---
# Efficacy of Young Cinnamomum zeylanicum Blume Bark on Hyperglycemia and PTPase Activity in Type 2 Diabetes
## Abstract
Diabetes is a major public health concern and natural easy-going remedies are being searched. Since *Cinnamomum zeylanicum* Blume has a low coumarin concentration and possible insulin-enhancing properties, it is preferred over all other cinnamon species. Although similar research has been done on humans, there have been very few studies on this particular species, and none among South Asians. Moreover, no human trial that properly described their intervening agent (C. zeylanicum) and checked its efficacy at the molecular level along with clinical variables was conducted. Therefore, the current research aimed to explore the effects of C. zeylanicum on the glycemic index, lipid profile, and expression of the protein tyrosine phosphatase 1 B (PTP1B) enzyme in the peripheral blood mononuclear cells (PBMC) in type 2 diabetes.
We examined the presence of bioactive compounds in young C. zeylanicum bark (Alba grade) from native Sri Lanka using gas chromatography-mass spectrometry, high-performance thin-layer chromatography, and thin-layer chromatography before introducing it in the clinical study where trans-Cinnamaldehyde was found to be a major chemical constituent (>$60\%$). Then, from January 2020 to March 2022, a randomized double-blinded placebo-controlled trial was carried out in the Diabetic Clinic at AIIMS Rishikesh. A total of 154 diabetic patients were enrolled and were taken either cinnamon or placebo capsules (1.5 g/day) for 120 days on an empty stomach with warm water along with their conventional treatment. Reduction in fasting blood glucose levels in the cinnamon group was found -$35.50\%$ ($95\%$ CI, -173 to 58.4), whereas in the placebo group change was $5.00\%$ ($95\%$ CI, -165 to 224). For glycosylated hemoglobin, it differed -$0.85\%$ ($95\%$ CI, -8.2 to 1.6) in the cinnamon group compared to the placebo where it was found $0.15\%$ ($95\%$ CI, -6.1 to 5.5). PTP1B expression in PBMC was determined from pre- and post-trial blood samples using the Western Blot, and significant inhibition was also observed ($$p \leq 0.039$$). The study result depicts, C. zeylanicum is emerging as a beneficial plant for type 2 diabetes in Northern India and could be used as an adjunctive treatment rather than as a standalone managerial remedy.
## Introduction
Type 2 diabetes mellitus (T2DM) is a chronic complex metabolic disease with impaired insulin sensitivity and diminished insulin secretion. It is caused by multiple etiologies and the confluence of genetic, environmental, and behavioral risk factors, including inactivity, sedentary behavior, cigarette smoking, and excessive alcohol consumption [1]. The prevalence of T2DM has been progressively growing globally. Among people over the age of 18 years, it has hiked from $4.7\%$ in 1980 to $8.5\%$ in 2014 and was expected to reach $9.3\%$ in 2019 [2,3]. Limited options for decreasing blood sugar, adverse effects of modern medicine, unpleasant insulin therapy, unexpected hypoglycemia, and the expense of treatment necessitate the quest for a hassle-free cure. Since the 1990s, when nuclear receptor proteins that regulate gene expression were recognized as a potential therapeutic target against diabetes, cinnamon had become the subject of research [4]. The name "cinnamon" is derived from the Latin and medieval French intermediate forms of the word "kínnamon," which means "sweet wood" [5]. Various species belonging to the Lauraceae family are cultivated as sources of cinnamon spice such as "Ceylon cinnamon" or "Cinnamomum zeylanicum," "Chinese cinnamon," or "Cinnamomum cassia," "Vietnamese cinnamon" or "Cinnamomum loureiroi," "Malabar cinnamon" or "Cinnamomum tamala," "Indonesian cinnamon" or "Cinnamomum burmannii," and "Camphor laurel" or "Cinnamomum camphora" [6]. Cinnamon extracts (CE) containing trans-Cinnamaldehyde (majorly found in cinnamon bark) and phenolic compound (majorly found in leaf) may improve insulin sensitivity, and insulin release by inhibiting the enzyme protein tyrosine phosphatase 1B [6]. CE improves type 2 diabetes by prompting GLUT 4 translocation, a major glucose transporter in skeletal muscle and adipose tissue which plays a key role in the uptake of glucose from the bloodstream, storing it as glycogen, and oxidizing it to produce energy [7]. CE affects the genes related to lipid metabolism, e.g., fatty acid synthase, sterol-responsible binding protein-1c, lipoprotein lipase, hormone-sensitive lipase, etc., in a way to control the metabolic biohazards accompanied by diabetes [8]. Pyruvate kinase (PK) and phosphoenol pyruvate carboxykinase (PEPCK) are two more important enzymes that regulate hepatic glucose metabolism by increasing glycolysis and inhibiting gluconeogenesis. In diabetes, the diminished PK activity causes a reduction of glycolysis and elevated PEPCK activity catalyzes gluconeogenesis. CE reverses the increased hepatic PEPCK mRNA expression and restoration to near-normal values of PK [9]. It also decreases plasma oxidative stress markers and delays stomach emptying [10,11].
Though animal studies had provided evidence of the positive effects of cinnamon and there is gaining popularity in the traditional use against diabetes [12], very few human clinical trials on this particular species only in the Iranian population make up the paucity of sufficient scientific evidence as well as there has been no work proving it at the biochemical level in humans [13,14]. Studies on other species of cinnamon (C. cassia or C. burmannii) had also some clashes, e.g., lack of species name, origin, quality, purity, grade, etc. Methodological differences across previous studies (small sample size, dose, duration of trial, form of intervention, e.g., grind bark to water or alcohol extract), disparities in outcomes, and lack of hands-on proof from bimolecular levels in humans have made it imperative to carry out. Using cinnamon powder as a supplement in the diet could be inexpensive and combined with other diabetes care strategies it could lower blood sugar levels more successfully in nations like India where it is widely accessible. So, a well-defined methodology was used to develop and design the current randomized clinical trial (RCT) in North India.
## Materials and methods
For our research certified young C. zeylanicum Blume bark (Alba grade) was imported from the farm “True Ceylon Spice” of native Sri Lanka. After that, phytochemical analysis had been conducted using gas chromatography-mass spectrometry (GCMS), high-performance thin-layer chromatography (HPTLC), and thin-layer chromatography (TLC) in the Department of Natural products, National Institute of Pharmaceutical Education and Research (NIPER), Mohali, India. This was done to assess the purity and also to determine the active compound present in those imported barks prior to performing the trial and then crude bark powder was used for the actual randomized trial in the capsulated form.
Essential oil extraction Dried bark (100 g powder) was subjected to steam distillation in a Clevenger-type apparatus for 8 hours. The distillate containing water along with essential oil was transferred into a 250 mL separating funnel. Dichloromethane (DCM) was added to the separating funnel for performing partitioning. The separating funnel was then capped tightly and shaken vigorously with occasional venting. When the two layers were separated, the lower DCM layer was transferred to a clean conical flask. The extraction process was repeated with two more portions of DCM and all of the DCM fractions were collected. Then, the pooled DCM layers were washed again with distilled water followed by separating into a dry container. The traces of aqueous solvent if present in the pooled fraction were removed by passing it through anhydrous sodium sulfate (Na2SO4). The combined DCM fraction was finally evaporated using a rotary evaporator. The residue was obtained in the form of a yellow oil (0.8 mL).
TLC analysis The isolated compound (cinnamaldehyde) from C. zeylanicum bark oil was dissolved in chloroform (CHCl3) and then applied on a silica gel TLC plate and developed using a mobile phase comprising of toluene:ethyl acetate (EtOAc) (8:2). The TLC plate was visualized under UV light. The standard trans-cinnamaldehyde was used as a reference standard (Figure 1).
**Figure 1:** *TLC profile of cinnamon oil with the trans-cinnamaldehyde standard.TLC profile at (a) 254 nm and at (b) 330 nm, (c) after derivatization (anisaldehyde sulphuric acid reagent) (1: cinnamaldehyde standard, 2: cinnamon oil).*
Soxhlet extraction A 100 g of bark material was used for the soxhlet extraction using methanol (MeOH) solvent. After completion of the soxhlet extraction (24 hours) methanol was concentrated using a rotary evaporator to give 8.66 g of the dried extract. Of the total extract, 6 g was further used for partitioning using hexane, EtOAc, and n-butanol (n-BuOH) in the order of increasing polarities. The amount of hexane extract, EtOAc extract and n-BuOH extract obtained after partitioning are 700 µL, 1.71 g, and 3.2 g, respectively.
HPTLC analysis It was performed using two different mobile phases. The results with each of the mobile phases were recorded in Figures 2, 3.
**Figure 2:** *HPTLC profile at 254 nm (a), HPTLC profile at 330 nm (b), TLC profile after derivatizing with natural product (NP) reagent (c).1: methanol extract, 2: rutin standard, 3: quercetin standard, 4: EtOAc extract, 5: cinnamaldehyde standard, 6: kaempferol standard, 7: hexane extract, 8: catechin standard, 9: caffeic acid standard, and 10: butanol extract.TLC Plate 1: Mobile Phase used: Toluene: EtOAc (8:2)* **Figure 3:** *TLC profile at 254 nm (a), TLC profile at 330 nm (b), TLC profile after derivatizing with NP reagent (c).1: methanol extract, 2: rutin standard, 3: quercetin standard, 4: EtOAc extract, 5: cinnamaldehyde standard, 6: kaempferol standard, 7: hexane extract, 8: catechin standard, 9: caffeic acid standard, and 10: butanol extract.TLC Plate 2: Mobile Phase: Chloroform: Methanol: Formic acid (9.5:0.4:0.1).*
HPTLC result The TLC plates developed in two different mobile phases did not match the spots of standards which included rutin, quercetin, kaempferol, catechin, and caffeic acid. Only the presence of trans-cinnamaldehyde was confirmed with its standard using both HPTLC and TLC. The cinnamon oil percentage was calculated using the formula given as follows: % Oil = essential oil weight/sample weight × 100 × specific gravity = 0.8 mL/100g × 100 × 1.01 = 0.808. The percentage of cinnamon oil pre-reported in the bark of C. zeylanicum is 0.5-$1\%$. The percentage of oil isolated was found to be $0.8\%$ which is in the specified range [15].
GCMS analysis It was carried out to identify the chemical components present in the cinnamon oil obtained from the C. zeylanicum bark. Twenty-four chemical constituents were identified in the oil extracted from bark using the *Clevenger apparatus* viz. benzaldehyde, α-phellandrene, delta-3-carene, limonene, β-phellandrene, 3,7-dimethyl-1,6-octadien-3-ol, α-pinene, naphthalene-1,2,3,4,4a,7-hexahydro, endo-borneol, β-terpineol, d-limonene, cinnamaldehyde, eugenol, α-cubebene, copaene, α-cedrene, ylangene, caryophyllene, humulene, 3-methoxy cinnamaldehyde, caryophyllene alcohol, epiglobulol, α-santalol, and benzyl benzoate. The major components of the oil were found to be trans-cinnamaldehyde ($33\%$) and its isomer ($31\%$), followed by eugenol ($7\%$), 3,7-dimethyl-1,6-octadien-3-ol ($5\%$) caryophyllene ($3\%$), naphthalene,1,2,3,4,4a,7-hexahydro ($3\%$), benzyl benzoate ($1.8\%$), d-limonene ($1.5\%$), benzaldehyde ($1.3\%$), and other constituents in trace amounts (Figure 4).
**Figure 4:** *GC-MS spectrogram of cinnamon oil.GC-MS: gas chromatography-mass spectrometry*
Trans-cinnamaldehyde is present as a major chemical constituent (>$60\%$) in the cinnamon essential oil extracted through the Clevenger apparatus, confirmed by HPTLC, TLC, and GCMS analysis by using the reference standard of cinnamaldehyde. However, HPTLC analysis revealed the absence of rutin, catechin, kaempferol, quercetin, caffeic acid, and other polyphenols in the bark extracts of C. zeylanicum.
Capsule making After the quality checking of cinnamon bark, the authors prepared both cinnamon and placebo capsules with the help of “Green Gold Pharmaceuticals,” Haridwar, India. These pharmaceuticals are a non-profit foundation for making Ayurvedic medicines, herbal extracts, essential oils, crude herbs, etc. A manual method was used and nearly about one-week period was taken to complete the whole process. Here, the cinnamon capsule contained 500 mg C. zeylanicum grind bark powder and the placebo capsule contained 500 mg grind Bengal gram flour. Both capsules were white in color and identical in shape. Both boxes were white, identical in shape, have the same aroma spray before closing, and were labeled as “Ayurvedic medicine for diabetes.” We didn’t make any extract capsules as our aim was to use whole-product in the clinical trial to evaluate effectiveness in their natural form.
Randomized controlled trial The RCT was conducted from January 2020 to March 2022 in the Diabetic Clinic at All India Institute of Medical Sciences (AIIMS), Rishikesh. A total of 697 T2DM patients were assessed for the eligibility criteria and finally, 154 people were recruited as per the estimated sample size. The sample size estimation was done through GPower software 3.1 with moderate effect size 0.5, α 0.05, power 0.8, and $20\%$ attrition rate. Patients with [1] non-insulin-dependent type 2 diabetes, [2] age more than 30 years [3] glycosylated hemoglobin 6.5, [4] fasting blood glucose level 126 mg/dL, [5] cooperative and want to participate, [6] no intake of herbs or other complementary therapy in recent eight weeks, [7] no acute infection (pneumonia, urinary tract infection, otitis), and [8] no insulin therapy were included in the study. Chain smoking, heavy alcohol, pregnancy and lactation, allergy or sensitivity to cinnamon, liver disease, and critically ill patients were excluded from the trial.
The study was randomized where block randomization had been done through a computer-generated random list created by sealedenvelope.com (block size four). Participants were randomized to both groups (77 persons in each group) and made their allocation concealed. The trial was double-blinded as researchers, e.g., recruiting physicians, blood sample collectors, and outcome assessors along with participants were blind. Allocation concealment and blinding were maintained throughout the trial. After taking written informed consent from participants and completing socio-demographic proforma, the pretrial blood sample was drawn for fasting blood glucose (FBG), glycosylated hemoglobin (HbA1c), and lipid profiles. Thereafter, cinnamon 500 mg three capsules daily were taken by the experimental group in the morning on an empty stomach with warm water for 120 days or four months. The control group did the same with placebo capsules (which look identical, white in color) containing finely grounded Bengal gram flour. The outcome was re-investigated at the end of the study as a routine checkup. Conventional management (dietary modification, exercise, and oral hypoglycemic agent, e.g., metformin, glimepiride, teneligliptin, was continued for both groups during the study. The third person who was a medicine box supplier and was responsible to analyze PTP1B was open and took $10\%$ pre- and post-trial blood samples of the experimental group in a randomized way from the sample collection center separately.
Sample collection procedure Whole blood sample in fasting (at least eight hours) condition was drawn from a brachial vein in a sitting position with the help of a needle 22-gauge, needle holder, and tourniquet from study participants at the sample collection center of AIIMS, Rishikesh. Blood samples were collected in one ethylenediaminetetraacetic acid (EDTA) (lavender color, 2 mL), one sodium fluoride (gray color, 2 mL), and one non-anticoagulant (red, 4 mL) vials for estimation of HbA1c, FBG, and lipid profile, respectively. Thereafter study participants were undergone with their concealed allocated intervention or placebo for four months with their previous prescribed treatment. After completion of the trial, fasting blood samples of FBG, HbA1c, and lipid profiles were re-collected in the same way. The collection of blood had been performed by health care professionals only. The aseptic technique was maintained. No extravasation injury was noticed.
Fasting blood glucose analysis Blood glucose analysis was performed with the enzymatic hexokinase method on the Fully Automated Chemistry Analyser of Beckman Coulter Inc. (Brea, CA), model no. Au480, installed at the Clinical Biochemistry Laboratory, AIIMS, Rishikesh.
Estimation of glycosylated hemoglobin HbA1c is the greatest indicator of average glucose levels during the previous one to three months. Estimation of HbA1c was conducted on TOSOH.HLC-723 (Tessenderlo, Belgium: Tosoh Europe N.V.) ion-exchange high-performance liquid chromatography principle at the Clinical Biochemistry laboratory AIIMS Rishikesh.
Lipid profile estimation Fasting levels of serum triacylglycerol, total cholesterol, low-density lipoprotein, and high-density lipoprotein were also estimated on AU680 fully automated clinical chemistry analyzer at the Clinical Biochemistry Laboratory, AIIMS, Rishikesh.
Western blot Expression of protein tyrosine phosphatase 1B was compared against beta-actin expression in the Bioengineering Lab of the Indian Institute of Technology, Roorkee. Total protein lysates were prepared from the collected patient PBMCs using radioimmunoassay (RIPA) buffer (50 mM Tris-HCL, 150 mM NaCl, $1\%$ NP40, $0.1\%$ sodium deoxycholate, $0.1\%$ SDS, pH 7.4) containing protease inhibitor cocktail. Protein concentrations were determined using the Bradford method (BioRad) then equal amounts of proteins were prepared in Laemmli buffer for SDS-PAGE and separated by $12\%$ SDS-polyacrylamide gel. The separated proteins were electro-transferred into a polyvinylidene difluoride (PVDF) membrane. The membranes were blocked with $3\%$ BSA in TBST (20 mM, Tris, pH 7.5, 150 mM NaCl, and $0.1\%$ Tween 20) for an hour at room temperature with constant rocking. The primary antibody (PTP1B rabbit the polyclonal antibody, catalog no. 11334-1-AP, Proteintech) was diluted at 1:2000 with $3\%$ BSA in TBST and incubated for 2 hours at room temperature. Membranes were washed three times (10 minutes each) with TBST and incubated for 1 hour in horseradish peroxidase-conjugated secondary antibodies (Goat Anti-Rabbit IgG, catalog no. SA00001-2, Proteintech) diluted at 1:5000 in $3\%$ BSA in TBST at room temperature [16,17].
The research has followed the guidelines of the Declaration of Helsinki and Tokyo for humans and was approved by the institutional review board (#AIIMS/IEC/$\frac{18}{577}$), AIIMS Rishikesh. The trial was registered in Clinical Trial Registry, registration no. CTI/$\frac{2019}{04}$/018386.
A total of 132 participants out of 154 had completed the trial. The reason for the $14\%$ dropout was explained in Figure 5 (CONSORT flow diagram, as per 2010 guidelines). However, some of the reasons were allergies, nausea, diarrhea, inconvenience, non-compliance, avoiding the hospital, shift to insulin therapy, etc. No adverse effect was found during the trial and per protocol, analysis had been done for 68 participants in the experimental group and 64 participants in the control group.
**Figure 5:** *CONSORT flow diagram as per 2010 guidelines.CONSORT: Consolidated Standards of Reporting Trial*
Statistical analysis *Statistical analysis* had been performed based on the data. Categorical data were reported as frequency and percentages. The difference between categorical data was assessed by chi-square test. Continuous data were reported as mean±standard deviation (SD), or median (minimum to maximum) for non-normally distributed data, which was tested by use of the Kolmogorov-Smirnov and Shapiro-Wilk test. The difference between the intergroup mean±SD was tested by an independent sample t-test and the inter-group median (minimum to maximum) was tested by the Mann-Whitney U test. The differences between clinical data (non-normally distributed) were analyzed by the Mann-Whitney U test (2-tailed non-parametric) where intra-individual delta values were taken for calculating the difference. P-values less than 0.05 were regarded as statistical significance. It was a per-protocol analysis, performed by the use of IBM SPSS Statistics for Windows, version 23 (Armonk, NY: IBM Corp.) [18].
## Results
Among socio-demographic details data regarding age, BMI, gender, marital status, education, occupation, religion, type of family, habitat, and duration of diabetes were obtained using a self-structured questionnaire. The data was tabulated and depicted in Table 1. Table 2 depicted the changes in fasting blood glucose, glycosylated hemoglobin, triglycerides, total cholesterol, low-density lipoprotein, and high-density lipoprotein from the initial to the end result of the trial.
As indicated in Table 2, there was a significant decrease in FBG, and HbA1C levels in the cinnamon group compared to the placebo ($$p \leq 0.000$$). No such reduction was observed in the placebo group. LDL-C level was found to increase in the placebo group compared to cinnamon ($$p \leq 0.047$$). There were no notable changes in the levels of triglycerides, total cholesterol, and HDL cholesterol ($p \leq 0.05$).
To determine the inhibition of protein tyrosine phosphatase, the western blot method was applied and beta-actin was taken as a housekeeping gene. A total of 13 participants’ blood samples from the cinnamon group who had higher glycosylated hemoglobin at the time of the pre-test and maintained good compliance throughout the trial were collected randomly. One obese non-diabetic control was also taken. Therefore, every participant had four bands of protein: pre-PTP1B, post-PTP1B, pre-beta-actin, and post-beta-actin (Figure 6).
**Figure 6:** *Changes in protein band of protein tyrosine phosphatase 1B and beta-actin before and after the trial.*
Figure 7 represents the inhibition of protein tyrosine phosphatase in peripheral blood mononuclear cells of type 2 diabetes patients after consuming C. zeylanicum for four months. Student's t-test had been used for statistical analysis between the mean pre- and post-PTP1B band’s intra-optical density. Such results depict the anti-diabetic properties of C. zeylanicum.
**Figure 7:** *Expression of protein tyrosine phosphatase 1B in pre- and post-C. zeylanicum treatment.*
## Discussion
Besides the multiple etiology, complex treatment regimens, unexpected hypoglycemia, high expenditure on treatment, patients' mistaken beliefs, and the negative impacts of modern medicine have all been cited as reasons for low compliance and poor adherence to current treatment methods for diabetes. While plant resources became a top priority in the quest for new medicines, cinnamon or dalchini, the common aromatic condiment obtained from the inner bark of the tree species genus Cinnamomum was touted as a promising spice for diabetes management [19].
A 62.09-$89.31\%$ of trans-Cinnamaldehyde could be found in cinnamon bark oil as a major compound and in the present research, it was found in nearly $64\%$ which was in the pre-specified range [20-22]. Although the use of Ceylon cinnamon against diabetes was mentioned in the Charaka Samhita and cassia was well-known in Chinese folk medicine. Khan et al. in 2003 were the first to show scientific evidence through a human trial on cassia bark powder (1 g, 3 g, and 6 g) in Pakistan, where type 2 diabetic participants' fasting blood glucose and lipid profile was significantly reduced over the course of 40 days [23]. Following that trial, Crawford in 2009 in Las Vegas, Akilen et al. in 2010 in London, and Sengsuk et al. in 2015 in Thailand conducted research with a similar methodology and approved that 1-2 g of grind cinnamon cassia for 2-3 months could decrease HbA1C in type 2 diabetes [24-26].
But comparable results were not found by Vanschoonbeek et al. in 2006 in the Netherlands, Suppapitiporn et al. in 2006 in Thailand, Blevins et al. in 2007 in Oklahoma City, US, and Hasanzade et al. in 2013 in Iran [27-30]. The study of Vanschoonbeek et al. was primarily confined small sample size, a short duration, and solely female participants (post-menopausal women) who might be experiencing hormonal changes. In the study of Blevins et al., both groups were different by age, multi-ethnicity in a small study population was seen and the initial value of outcome variables was near to normal range. The study by Hasanzade et al. in 2013 majorly performed on a low socio-economic group of participants (housekeepers) which makes a methodological flaw in the research [30]. Besides those factors, the lack of a proper description of the intervening agent is a major discrepancy across all the studies. It's sometimes only described as to where it was acquired, e.g., from a local vendor. Cinnamon's origin, geographical wellness, and limit were unknown, and no validation or authentication was offered. It should be explained in detail, including the type of cinnamon, its origin, grade, whether it was new or old, and the optimal time to utilize it. Khan et al. in 2010 in Pakistan and Sharma et al. in 2012 in India didn't even specify the species' name though these studies conveyed a significant decrease in anthropometric, glycemic, and lipid index of patients with type 2 diabetes after consuming cinnamon bark [31,32].
When talking about extract, Ziegenfuss et al. in 2006 and Anderson et al. in 2016 employed cassia extract 500 mg/day for three and two months, respectively, and found reduced plasma glucose in type 2 diabetes [33,34]. The use of cassia extract for diabetes was positively impacted by intergroup differences in glycosylated hemoglobin along with fasting blood glucose levels by Lu et al. 2012 [35]. However, in those studies, there was no precise explanation of extract synthesis. Depending on the species and even the formulation, the number of active molecules (cinnamaldehyde otherwise polyphenols) may vary. Furthermore, phytochemicals may be harmed by differences in manufacturing techniques, because herbal drugs aren't subjected to similar quality control standards as other pharmacy items. Because cinnamon contains an evaporating agent, the essential oil might evaporate over time if it was old or mishandled, and there would be no active molecule to work with at that moment. The species of cinnamon was predetermined in the current investigation, along with the plant's age, location, and quality of the sticks. According to the plant list at http://www.theplantlist.org, C. zeylanicum *Blume is* a synonym of *Cinnamomum verum* J. Presl which is currently accessible in a variety of forms, including quills (usually 42-inch cinnamon sticks), quilling (broken cinnamon tubes), bark powder, bark oil, and leaf oil. The diameter of the cinnamon quills is divided into the following four classes in Sri Lankan grading: Alba (<0.24 inches), Continental (<0.63 inches), Mexican (<0.75 inches), and Hamburg (<1.3 inches). C. zeylanicum bark from native Sri Lanka was collected as young, fresh bark (Alba grade) for investigation. It was also measured whether the product was pure and legitimate and whether trans-cinnamaldehyde and eugenol, two active compounds, were present in the ground bark, and capsules were prepared just after grinding for best durability and to provide an equal quantity to the study subjects.
Vafa et al. carried out the first study on C. zeylanicum in Tehran, Iran, in 2012 where 44 participants with type 2 diabetes were enrolled (22 persons in each group; cinnamon and placebo), and 3g of grind C. zeylanicum was administered in the treatment group for eight weeks. Despite the study's small sample size and brief duration, the cinnamon group showed significantly lower FBG, HbA1c, and triglyceride levels than others [13]. Similar results regarding C. zeylanicum had also been found in the study by Zare et al. in 2019 in Iran [14]. The study was conducted on a relatively large sample size (138 participants) where cinnamon supplementation for three months significantly decreased levels of fasting and post-prandial blood glucose, HbA1c, triglycerides, total cholesterol, LDL cholesterol, and a significant increase in HDL cholesterol. The current research had been performed on 132 participants, and 1.5 g grind C. zeylanicum bark per day for four months resulted in a significant reduction in fasting blood sugar, and glycosylated hemoglobin among type 2 diabetic patients. Though present research didn’t observe any change in the levels of triglyceride and other lipids which might be related to a low dose of cinnamon or the near-normal value of the initial level. Along with significant inhibition of PTPase expression in peripheral blood mononuclear cells after four months of oral consumption of Ceylon cinnamon bark evidences it as an anti-diabetic herb from a biomolecular level (Figure 8).
**Figure 8:** *Schematic representation of active cinnamon compound in insulin signaling transduction.The image is created by the author (Anindita Mandal) of this study.*
When insulin binds to α unit of the insulin receptor, phosphorylation of tyrosine protein residue of β unit takes place. Kinase enzyme helps in the addition of phosphate from ATP to tyrosine amino acid is termed tyrosine kinase. The opposite enzyme protein tyrosine phosphatase removes a phosphate group from the target molecule and causes de-phosphorylation that inactivates insulin receptors followed by insulin resistance [36-38]. There is evidence that certain diabetic individuals have a variant of PTPase that is abnormally active and for this purpose, PTPase inhibitors, e.g., cinnamon are being explored (Figure 8) [39]. The result was similar to the study by Saifudin et al. in 2013 and Imparl-Radosevich et al. in 1998. Water and methanol extracts of *Cinnamomum burmannii* exhibited ≥$70\%$ inhibition at 25 μg/mL (IC50, 2.47 μg/mL) which was comparable with that of the positive control, RK-682 (IC50, 2.05 μg/mL). The PTP1B inhibitory activity of the constituents of C. Burmannii was then evaluated and trans-cinnamaldehyde (5; IC50, 57.6 μM) was found as an active constituent [40,41].
Though there's the fact that this type of human trial is difficult, as it would be among primary care patients. Trials of this nature could not be tightly controlled. Doctor shopping, medication changes, dose adjustments by patients exclusively, treatment discontinuation, participants' failure to comply with dietary restrictions, workouts, interactions with several pharmacotherapies, and lack of follow-up all had a significant impact on the trial's outcome. In the present study, researchers tried to control the extraneous variables by discussing with participants for both groups. The researchers didn’t explore about contents of the capsules, neither cinnamon nor placebo, and never prioritize anyone in front of the participants.
Major limitations of the present study are as follows: this is a single-centered study in single ethnic group and only one enzyme PTPase was studied through a single technique. Because topographical changes affect body mechanics, the study might be reproduced as a multi-centric trial including a broad population in a different geographic area. Instead of using the western blot, other biochemical tests could be utilized to acquire more specific information at the molecular level. Genes involved in the insulin pathway, rather than the insulin molecule, could also be assessed.
## Conclusions
C. zeylanicum has the potential to improve the glycemic index in persons with poorly managed diabetes. Its anti-diabetic properties have made it an herbal adjuvant treatment modality. This was further substantiated by the fact that ingestion of Ceylon cinnamon inhibited protein tyrosine phosphatase in peripheral blood mononuclear cells of type 2 diabetic individuals. Triglycerides, total cholesterol, and HDL-C levels were all unaltered. The clinical relevance and durability of these effects in multi-centric trials have yet to be determined. It's still worth looking into genes linked to glycemic pathways after consumption of C. zeylanicum as future prospects of research.
## References
1. Olokoba AB, Obateru OA, Olokoba LB. **Type 2 diabetes mellitus: a review of current trends**. *Oman Med J* (2012.0) **27** 269-273. PMID: 23071876
2. Roglic G. **WHO global report on diabetes: a summary**. *Int J Non-Commun Dis* (2016.0) **1** 3-8
3. Saeedi P, Petersohn I, Salpea P. **Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: results from the International Diabetes Federation Diabetes Atlas, 9th edition**. *Diabetes Res Clin Pract* (2019.0) **157**
4. Sheng X, Zhang Y, Gong Z, Huang C, Zang YQ. **Improved insulin resistance and lipid metabolism by cinnamon extract through activation of peroxisome proliferator-activated receptors**. *PPAR Res* (2008.0) **2008**
5. Medagama AB. **The glycaemic outcomes of Cinnamon, a review of the experimental evidence and clinical trials**. *Nutr J* (2015.0) **14**
6. Sharma S, Mandal A, Kant R, Jachak S, Jagzape M. **Is cinnamon efficacious for glycaemic control in type-2 diabetes mellitus**. *J Pak Med Assoc* (2020.0) **70** 2065-2069. PMID: 33341863
7. Absalan A, Mohiti-Ardakani J, Hadinedoushan H, Khalili MA. **Hydro-alcoholic cinnamon extract, enhances glucose transporter isotope- 4 translocation from intracellular compartments into the cytoplasmic membrane of C2C12 myotubes**. *Indian J Clin Biochem* (2012.0) **27** 351-356. PMID: 24082459
8. Soliman MM, Ahmed MM, El-Shazly SA. **Cinnamon extract regulates gene expression of lipids and carbohydrate metabolism in streptozotocin-induced diabetic Wistar rats**. *Am J Biochem Biotechnol* (2013.0) **9** 172-182
9. Anand P, Murali KY, Tandon V, Murthy PS, Chandra R. **Insulinotropic effect of cinnamaldehyde on transcriptional regulation of pyruvate kinase, phosphoenolpyruvate carboxykinase, and GLUT4 translocation in experimental diabetic rats**. *Chem Biol Interact* (2010.0) **186** 72-81. PMID: 20363216
10. Roussel AM, Hininger I, Benaraba R, Ziegenfuss TN, Anderson RA. **Antioxidant effects of a cinnamon extract in people with impaired fasting glucose that are overweight or obese**. *J Am Coll Nutr* (2009.0) **28** 16-21. PMID: 19571155
11. Hlebowicz J, Hlebowicz A, Lindstedt S. **Effects of 1 and 3 g cinnamon on gastric emptying, satiety, and postprandial blood glucose, insulin, glucose-dependent insulinotropic polypeptide, glucagon-like peptide 1, and ghrelin concentrations in healthy subjects**. *Am J Clin Nutr* (2009.0) **89** 815-821. PMID: 19158209
12. Almubayedh H, Ahmad R, Naqvi AA, Ahmad N. **Ethnopharmacological Uses and Public Knowledge Regarding Cinnamomum zeylanicum in Khobar, Saudi Arabia**. *J Pharm Bioallied Sci* (2018.0) **10** 159-165. PMID: 30237686
13. Vafa M, Mohammadi F, Shidfar F, Sormaghi MS, Heidari I, Golestan B, Amiri F. **Effects of cinnamon consumption on glycaemic status, lipid profile and body composition in type 2 diabetic patients**. *Int J Prev Med* (2012.0) **3** 531-536. PMID: 22973482
14. Zare R, Nadjarzadeh A, Zarshenas MM, Shams M, Heydari M. **Efficacy of cinnamon in patients with type II diabetes mellitus: a randomized controlled clinical trial**. *Clin Nutr* (2019.0) **38** 549-556. PMID: 29605574
15. Gursale A, Dighe V, Parekh G. **Simultaneous quantitative determination of cinnamaldehyde and methyl eugenol from stem bark of Cinnamomum zeylanicum Blume using RP-HPLC**. *J Chromatogr Sci* (2010.0) **48** 59-62. PMID: 20056038
16. Zheng L, Zhang W, Li A, Liu Y, Yi B, Nakhoul F, Zhang H. **PTPN2 downregulation is associated with albuminuria and vitamin D receptor deficiency in type 2 diabetes mellitus**. *J Diabetes Res* (2018.0) **2018**
17. Khadir A, Tiss A, Abubaker J. **MAP kinase phosphatase DUSP1 is overexpressed in obese humans and modulated by physical exercise**. *Am J Physiol Endocrinol Metab* (2015.0) **308** 71-83
18. **IBM Corp: released in 2015, IBM SPSS statistics for windows, version 23.0**
19. Vinitha M, Ballal M. **In vitro anticandidal activity of Cinnamomum verum**. *J Med Sci* (2008.0) **8** 425-428
20. Vazirian M, Alehabib S, Jamalifar H, Fazeli MR, Toosi AN, Khanavi M. **Antimicrobial effect of cinnamon (Cinnamomum verum J. Presl) bark essential oil in cream-filled cakes and pastries**. *Res J Pharmacogn* (2015.0) **2** 11-16
21. Kamaliroosta L, Gharachorloo M, Kamaliroosta Z, Alimohammad Zadeh KH. **Extraction of cinnamon essential oil and identification of its chemical compounds**. *J Med Plants Res* (2012.0) **6** 609-614
22. Ainane T, Khammour F, Merghoub N. **Cosmetic bio-product based on cinnamon essential oil “Cinnamomum verum” for the treatment of mycoses: preparation, chemical analysis and antimicrobial activity**. *MOJ Toxicol* (2019.0) **5** 5-8
23. Khan A, Safdar M, Ali Khan MM, Khattak KN, Anderson RA. **Cinnamon improves glucose and lipids of people with type 2 diabetes**. *Diabetes Care* (2003.0) **26** 3215-3218. PMID: 14633804
24. Crawford P. **Effectiveness of cinnamon for lowering hemoglobin A1C in patients with type 2 diabetes: a randomized, controlled trial**. *J Am Board Fam Med* (2009.0) **22** 507-512. PMID: 19734396
25. Akilen R, Tsiami A, Devendra D, Robinson N. **Glycated haemoglobin and blood pressure-lowering effect of cinnamon in multi-ethnic type 2 diabetic patients in the UK: a randomized, placebo-controlled, double-blind clinical trial**. *Diabet Med* (2010.0) **27** 1159-1167. PMID: 20854384
26. Sengsuk C, Sanguanwong S, Tangvarasittichai O, Tangvarasittichai S. **Effect of cinnamon supplementation on glucose, lipids levels, glomerular filtration rate, and blood pressure of subjects with type 2 diabetes mellitus**. *Diabetol Int* (2016.0) **7** 124-132. PMID: 30603255
27. Vanschoonbeek K, Thomassen BJ, Senden JM, Wodzig WK, van Loon LJ. **Cinnamon supplementation does not improve glycemic control in postmenopausal type 2 diabetes patients**. *J Nutr* (2006.0) **136** 977-980. PMID: 16549460
28. Suppapitiporn S, Kanpaksi N, Suppapitiporn S. **The effect of cinnamon cassia powder in type 2 diabetes mellitus**. *J Med Assoc Thai* (2006.0) **89** 200-205
29. Blevins SM, Leyva MJ, Brown J, Wright J, Scofield RH, Aston CE. **Effect of cinnamon on glucose and lipid levels in non insulin-dependent type 2 diabetes**. *Diabetes Care* (2007.0) **30** 2236-2237. PMID: 17563345
30. Hasanzade F, Toliat M, Emami SA, Emamimoghaadam Z. **The effect of cinnamon on glucose of type II diabetes patients**. *J Tradit Complement Med* (2013.0) **3** 171-174. PMID: 24716174
31. Khan R, Khan Z, Shah S. **Cinnamon may reduce glucose, lipid, and, Cholesterol level in type 2 diabetic individuals**. *Pakistan J Nutr* (2010.0) **9** 430-433
32. Sharma P, Sharma S, Agrawal RP, Agrawal V, Singhal S. **A randomized double blind placebo control trial of cinnamon supplementation on glycaemic control and lipid profile in type 2 diabetes mellitus**. *Australian Journal of Herbal Medicine* (2012.0) **24** 4-9
33. Ziegenfuss TN, Hofheins JE, Mendel RW, Landis J, Anderson RA. **Effects of a water-soluble cinnamon extract on body composition and features of the metabolic syndrome in pre-diabetic men and women**. *J Int Soc Sports Nutr* (2006.0) **3** 45-53. PMID: 18500972
34. Anderson RA, Zhan Z, Luo R. **Cinnamon extract lowers glucose, insulin and cholesterol in people with elevated serum glucose**. *J Tradit Complement Med* (2016.0) **6** 332-336. PMID: 27774415
35. Lu T, Sheng H, Wu J, Cheng Y, Zhu J, Chen Y. **Cinnamon extract improves fasting blood glucose and glycosylated hemoglobin level in Chinese patients with type 2 diabetes**. *Nutr Res* (2012.0) **32** 408-412. PMID: 22749176
36. Salway JG. **Medical Biochemistry at a Glance. Third Edition**. *John Wiley & Sons Ltd* (2012.0) **61** 62
37. Boucher J, Kleinridders A, Kahn CR. **Insulin receptor signaling in normal and insulin-resistant states**. *Cold Spring Harb Perspect Biol* (2014.0) **6**
38. Ardito F, Giuliani M, Perrone D, Troiano G, Lo Muzio L. **The crucial role of protein phosphorylation in cell signaling and its use as targeted therapy (Review)**. *Int J Mol Med* (2017.0) **40** 271-280. PMID: 28656226
39. Ahmad F, Goldstein BJ. **Increased abundance of specific skeletal muscle protein-tyrosine phosphatases in a genetic model of insulin-resistant obesity and diabetes mellitus**. *Metabolism* (1995.0) **44** 1175-1184. PMID: 7666792
40. Saifudin A, Kadota S, Tezuka Y. **Protein tyrosine phosphatase 1B inhibitory activity of Indonesian herbal medicines and constituents of Cinnamomum burmannii and Zingiber aromaticum**. *J Nat Med* (2013.0) **67** 264-270. PMID: 22645080
41. Imparl-Radosevich J, Deas S, Polansky MM, Baedke DA, Ingebritsen TS, Anderson RA, Graves DJ. **Regulation of PTP-1 and insulin receptor kinase by fractions from cinnamon: implications for cinnamon regulation of insulin signalling**. *Horm Res* (1998.0) **50** 177-182. PMID: 9762007
|
---
title: High-dose intravenous iron reduces myocardial infarction in patients on haemodialysis
authors:
- Mark C Petrie
- Pardeep S Jhund
- Eugene Connolly
- Patrick B Mark
- Michael R MacDonald
- Michele Robertson
- Stefan D Anker
- Sunil Bhandari
- Kenneth Farrington
- Philip A Kalra
- David C Wheeler
- Charles R V Tomson
- Ian Ford
- John J V McMurray
- Iain C Macdougall
journal: Cardiovascular Research
year: 2021
pmcid: PMC10022850
doi: 10.1093/cvr/cvab317
license: CC BY 4.0
---
# High-dose intravenous iron reduces myocardial infarction in patients on haemodialysis
## Abstract
### Aims
To investigate the effect of high-dose iron vs. low-dose intravenous (IV) iron on myocardial infarction (MI) in patients on maintenance haemodialysis.
### Methods and results
This was a pre-specified analysis of secondary endpoints of the Proactive IV Iron Therapy in Hemodialysis Patients trial (PIVOTAL) randomized, controlled clinical trial. Adults who had started haemodialysis within the previous year, who had a ferritin concentration <400 μg per litre and a transferrin saturation <$30\%$ were randomized to high-dose or low-dose IV iron. The main outcome measure for this analysis was fatal or non-fatal MI. Over a median of 2.1 years of follow-up, $8.4\%$ experienced a MI. Rates of type 1 MIs ($\frac{3.2}{100}$ patient-years) were 2.5 times higher than type 2 MIs ($\frac{1.3}{100}$ patient-years). Non-ST-elevation MIs ($\frac{3.3}{100}$ patient-years) were 6 times more common than ST-elevation MIs ($\frac{0.5}{100}$ patient-years). Mortality was high after non-fatal MI (1- and 2-year mortality of $40\%$ and $60\%$, respectively). In time-to-first event analyses, proactive high-dose IV iron reduced the composite endpoint of non-fatal and fatal MI [hazard ratio (HR) 0.69, $95\%$ confidence interval (CI) 0.52–0.93, $$P \leq 0.01$$] and non-fatal MI (HR 0.69, $95\%$ CI 0.51–0.93; $$P \leq 0.01$$) when compared with reactive low-dose IV iron. There was less effect of high-dose IV iron on recurrent MI events than on the time-to-first event analysis.
### Conclusion
In total, $8.4\%$ of patients on maintenance haemodialysis had an MI over 2 years. High-dose compared to low-dose IV iron reduced MI in patients receiving haemodialysis.
### EudraCT Registration Number
2013-002267-25.
## Graphical Abstract
Graphical Abstract
## Introduction
Contemporary data on the incidence and outcomes of myocardial infarction (MI) in patients on maintenance haemodialysis are sparse. MIs in patients receiving dialysis for end-stage kidney disease (ESKD) can be type 1 MIs (classical plaque rupture with thrombus formation) or type 2 MIs (secondary to ‘supply-demand mismatch’ with or without underlying coronary disease).
Concerns have been expressed that intravenous (IV) iron could increase the prevalence or severity of coronary artery disease, and even increase coronary artery events. These concerns have emerged from observational studies in humans, animal studies, and small mechanistic studies.1–4 Others have suggested that IV iron may not increase coronary events5 but there are no data to suggest that IV iron might reduce MIs. In the Proactive IV Iron Therapy in Hemodialysis Patients trial (PIVOTAL), we compared a regimen of IV iron administered proactively in a high-dose regimen, with a low-dose regimen, administered reactively.6 MI was an adjudicated outcome [type 1, type 2, ST-elevation MI (STEMI), non-ST-elevation MI (NSTEMI)] and was a pre-specified secondary endpoint in the trial. Here, we describe the rates and prognostic significance of MI in patients on maintenance haemodialysis and the effect of the high- vs. low-dose IV iron therapy regimens on MIs.
## 2. Methods
The design, baseline characteristics, and main results of PIVOTAL have been published.6,7 In summary, 2141 adults who had started haemodialysis within the previous year, who had a ferritin concentration <400 μg per litre and a transferrin saturation <$30\%$, and who were receiving an erythropoiesis-stimulating agent were enrolled. Patients were randomized in a 1:1 ratio, to receive open-label high-dose IV iron administered proactively or low-dose IV iron administered reactively. Ferritin concentration and transferrin saturation were measured monthly and the results were used to determine the monthly dose of iron sucrose. In the high-dose group, 400 mg of iron sucrose was prescribed, with safety cut-off limits (ferritin >700 μg per litre or transferrin saturation >$40\%$) above which further iron was withheld until the next blood test 1 month later. Patients in the low-dose group received 0–400 mg of iron sucrose monthly to maintain ferritin ≥200 μg per litre and transferrin saturation ≥$20\%$, in line with current guidelines. The protocol required the use of an erythropoiesis-stimulating agent in a dose sufficient to maintain a haemoglobin of 100–120 g per litre, but otherwise, patients were treated according to usual practice. The trial was reviewed and approved by the ethics committee. Each patient provided informed consent. The trial conformed to the principles outlined in the Declaration of Helsinki.
## 2.1 Baseline information related to MI
Investigators were asked about the presence of prior MI and other cardiovascular comorbidities on an electronic case report form. The use of cardiovascular medications, including statins, renin–angiotensin system blockers, and beta-blockers, were documented.
## 2.2 Clinical outcomes
The primary outcome of the trial was the composite of MI, stroke, hospitalization for heart failure, or death from any cause, analysed as time-to-first event. MI was a pre-specified secondary outcome. For this article, the outcomes of time-to-first MI (type 1 or type 2 MI, STEMI, or NSTEMI) and the composite outcome of MI and death due to MI were reported. In addition to time-to-first MI, we also performed a post hoc analysis of total (first and recurrent) MI events, to account for the cumulative burden of events over time. We also performed a post hoc analysis of mortality after (initially) non-fatal MI.
## 2.3 Adjudication of outcomes
All potential endpoints and all deaths were adjudicated by an independent committee, blinded to treatment allocation. The endpoint charter is included in the main results paper.6 The endpoint charter is based on the Food and Drug Administration (FDA)-endorsed Standardized Data Collection for Cardiovascular Trials Initiative (SCTI).8 For the confirmation of MI, there was a requirement for a rise and/or fall of cardiac biomarkers (preferably cardiac troponin) with at least one value above the 99th percentile upper reference limit and with at least one of the following: symptoms of myocardial ischaemia; new, or presumed new, significant ST-segment T-wave (ST-T) changes, or new left bundle branch block; development of pathological Q waves on the electrocardiogram; imaging evidence of new loss of viable myocardium or new regional wall motion abnormality; and identification of an intracoronary thrombus by angiography or autopsy. Two independent reviewers adjudicated each potential event with discordant results being discussed at a consensus committee meeting involving all adjudicators (comprising cardiologists, a renal physician, and a stroke physician).
## 2.4 Statistical analysis
The time-to-event analyses of the primary, secondary, and post hoc outcomes were performed in the intention-to-treat population using Cox proportional hazards regression, with treatment group and stratification factors [(dialysis catheter vs. arteriovenous fistula or graft), diagnosis of diabetes (yes vs. no), and duration of haemodialysis treatment (<5 months vs. ≥5 months)] as explanatory variables. The Kaplan–Meier method was used to estimate mortality rates and cumulative incidence functions corrected for the competing risk of deaths not included in the outcome of interest. Recurrent events were analysed using the proportional means model of Lin et al. and described in the form of mean frequency functions (method of Ghosh and Lin9). Baseline characteristics were summarized as means and standard deviations, medians and lower and upper quartiles, or counts and percentages as appropriate. P-values for between-group differences based on two-sample t-tests or χ2 tests/Fisher’s exact tests, as appropriate, are provided. Analyses were performed using SAS software, version 9.4 (SAS Institute) and R version 3.6.0.
## 3. Results
A total of 2141 eligible men and women were randomized. In total, 180 ($8.4\%$) patients experienced at least one confirmed fatal or non-fatal MI during the median follow-up of 2.1 years (maximum 4.4 years).
## 3.1 Patients experiencing a MI vs. those not experiencing a MI: baseline characteristics
Patients who had a fatal or non-fatal MI were older (67 vs. 62 years of age, $P \leq 0.001$) and more often Asian than other races (Table 1). Patients who had a fatal or non-fatal MI were more likely to have a history of previous MI, previous heart failure, diabetes, and peripheral artery disease than those who did not experience an MI. Patients with fatal or non-fatal MIs were more likely to have a diabetic or renovascular cause of renal failure than those who did not have an MI. Patients with a type 2 MI had higher systolic blood pressures than those with a type 1 MI (156 mmHg vs. 145 mmHg, $$P \leq 0.01$$, Supplementary material online, eTable S1).
**Table 1**
| Unnamed: 0 | MI (N = 180) | No MI (N = 1961) | P-value |
| --- | --- | --- | --- |
| Age (years) | 67.0 (12.3) | 62.4 (15.2) | <0.001 |
| Male sex (%) | 70 | 64.9 | 0.17 |
| Race (%) | | | 0.03 |
| White/European | 81 | 79 | 0.03 |
| Black/African descent | 5 | 9 | 0.03 |
| Asian | 13 | 8 | 0.03 |
| Other | 2 | 3 | 0.03 |
| BMI (kg/m2) | 28.8 (5.8) | 28.7 (7.0) | 0.92 |
| Systolic BP (mmHg) | 147 (25) | 144 (24) | 0.09 |
| Median duration of dialysis (months) | 4.9 (2.6–7.9) | 4.8 (2.8–8.3) | 0.8 |
| History (%) | | | |
| Hypertension | 81 | 72 | 0.05 |
| Atrial fibrillation | 11 | 7 | 0.03 |
| MI | 21 | 7 | <0.001 |
| PVD | 16 | 8 | 0.002 |
| Heart failure | 10 | 3 | <0.001 |
| Stroke | 12 | 8 | 0.12 |
| Diabetes | 65 | 42 | <0.001 |
| Aetiology of renal failure | | | <0.001 |
| Hypertension | 7 | 11 | <0.001 |
| Diabetic nephropathy | 51 | 32 | <0.001 |
| Glomerular disease | 11 | 19 | <0.001 |
| Tubulointerstitial disease | 7 | 10 | <0.001 |
| Renovascular disease | 15 | 6 | <0.001 |
| Polycystic kidney disease | 1 | 6 | <0.001 |
| Unknown | 5 | 10 | <0.001 |
| Smoking status (%) | | | |
| Never | 54 | 64 | 0.04 |
| Previous | 30 | 25 | |
| Current | 16 | 11 | |
| Laboratory measurements | | | |
| Haemoglobin | 105 (13) | 106 (14) | 0.73 |
| Ferritin | 208 (138–294) | 217 (133–304) | 0.45 |
| Transferrin saturation | 20 (15–24) | 20 (16–24) | 0.45 |
| C-reactive protein | 8 (4–17) | 6 (3–14) | 0.04 |
| Cardiovascular medications (%) | | | |
| β-Blocker | 48 | 44 | 0.25 |
| ACE inhibitor | 12 | 18 | 0.07 |
| ARB | 8 | 12 | 0.16 |
| Any diuretic | 47 | 43 | 0.34 |
| Statin | 75 | 58 | <0.001 |
| Any antiplatelet agent | 68 | 43 | <0.001 |
## 3.2.1 Time-to-first event analysis
Rates of fatal and non-fatal type 1 MIs ($$n = 142$$, $\frac{3.3}{100}$ patient-years) were 2.5 times greater than fatal and non-fatal type 2 MIs ($$n = 57$$, $\frac{1.3}{100}$ patient-years) (Table 2). Fatal and non-fatal NSTEMIs ($$n = 141$$, $\frac{3.3}{100}$ patient-years) were more than 6 times more common than fatal and non-fatal STEMIs ($$n = 22$$, $\frac{0.5}{100}$ patient-years).
**Table 2**
| Unnamed: 0 | n (% of total in patients in trial) | Rate/100 patient-years |
| --- | --- | --- |
| Time-to-first event analyses | Time-to-first event analyses | Time-to-first event analyses |
| Fatal or non-fatal MI | 180 | 4.2 |
| Fatal or non-fatal type 1 MI | 142 | 3.3 |
| Fatal or non-fatal type 2 MI | 57 | 1.3 |
| Fatal or non-fatal STEMI | 22 | 0.5 |
| Fatal or non-fatal NSTEMI | 141 | 3.3 |
| Recurrent (first and subsequent) event analyses | Recurrent (first and subsequent) event analyses | Recurrent (first and subsequent) event analyses |
| Fatal or non-fatal MI | 259 | 6.1 |
| Fatal or non-fatal type 1 MI | 193 | 4.5 |
| Fatal or non-fatal type 2 MI | 65 | 1.5 |
| Fatal or non-fatal STEMI | 24 | 0.6 |
| Fatal or non-fatal NSTEMI | 200 | 4.7 |
## 3.2.2 Recurrent event analysis
Of the total number of MIs ($$n = 259$$), 79 ($30.5\%$) were subsequent (i.e. not first events). Most of the recurrent MIs were type 1 MIs ($$n = 193$$) with few type 2 MIs ($$n = 65$$).
## 3.3 Death due to MI
MI was the cause of death in $5\%$ of patients who died of any cause (Supplementary material online, eTable S3). Fourteen percent of cardiovascular deaths were due to MI.
## 3.4 Mortality after non-fatal MI
Mortality after a non-fatal MI was $40\%$ at 1 year and $60\%$ at 2 years (Supplementary material online, eFigure S1). Mortality was similar for type 1 MIs and NSTEMIs but appeared to be higher for STEMIs (although numbers of these were small) (Supplementary material online, eFigure S1).
## 3.5 Effect of high-dose iron vs. low-dose iron on fatal and non-fatal MIs
In the time-to-first event analysis, fatal or non-fatal MIs occurred in 78 of 1093 patients ($7.1\%$; 3.5 events per 100 person-years) in the high-dose iron group and in 102 of 1048 patients ($9.7\%$; 4.9 events per 100 person-years) in the low-dose group [hazard ratio (HR) 0.69, $95\%$ confidence interval (CI) (0.52–0.93); $$P \leq 0.01$$] (Table 3 and Figure 1). Fatal or non-fatal type 1 MI occurred in $5.7\%$ in the high-dose group and $7.6\%$ in the low-dose group (HR 0.71; $95\%$ CI 0.51–0.99, $$P \leq 0.04$$). No reduction in non-fatal type 2 MIs was seen. In the recurrent event analysis, only fatal and non-fatal type 1 NSTEMIs were reduced by high-dose IV iron (Table 3 and Figure 2).
**Figure 1:** *Effect of high- vs. low-dose IV iron on fatal and non-fatal MI (time-to-first event analysis). (A) Fatal and non-fatal MI; (B) type 1 MI; (C) type 2 MI; (D) NSTEMI; and (E) STEMI.* **Figure 2:** *Effect of high- vs. low-dose IV iron on fatal and non-fatal MI (recurrent event analysis, i.e. first and subsequent events).* TABLE_PLACEHOLDER:Table 3
## 4. Discussion
In patients who had started haemodialysis less than 12 months prior to enrolment, MI was a common event, with $8\%$ having a fatal or non-fatal MI over 2 years of follow-up. Type 1 MIs and NSTEMIs were much more frequent than type 2 MIs and STEMIs. MIs were more common in this randomized trial than other cardiovascular events (e.g. heart failure hospitalization or stroke).6 IV iron administered in a high-dose regimen reduced acute fatal and non-fatal MI compared with low-dose iron. High-dose IV iron is the first therapy to reduce MIs in patients undergoing maintenance haemodialysis. These results suggest that the use of high-dose IV iron may be warranted to reduce MI.
Comparing the rate of acute MI over the last two decades is problematic as more sensitive biomarkers of myocardial damage (troponins) have been introduced. In more recent studies, clinical events, which would not previously have met the criteria for MI, have been classified as such. For example, at first glance, an incidence of MI of $10\%$ over 2 years over 20 years ago in a US Medicare cohort on dialysis appears very similar to that seen in PIVOTAL. It is, however, very likely that if troponins had been used in these earlier studies more clinical episodes would have been classified as MIs and substantially higher rates reported.10 *Observational data* using hospital discharge coding suggested that NSTEMI is more common than STEMI in patients receiving dialysis.11 *Our data* from a randomized clinical trial using formally adjudicated events supports this but also demonstrates that most acute MIs are type 1 rather than type 2.
Until the current trial, no treatment had been shown to reduce acute MI in patients receiving haemodialysis. Notably, statin or other lipid-lowering therapies have not resulted in reductions in acute MI.12 In the PIVOTAL trial, patients who had an acute MI were more likely to have ESKD of diabetic or renovascular aetiology. It is possible that efforts to prevent MIs might be particularly targeted at groups with these aetiologies of renal disease.
In the PIVOTAL trial, the reduction in MI by high-dose compared to low-dose iron might be due to several factors. As this was a clinical trial our ability to determine mechanisms is limited. More IV iron is likely to result in more oxygen delivery; haemoglobin levels increased more rapidly in those receiving high-dose IV iron than those on the low-dose regimen.6 Such an action might be more likely to prevent type 2 than type 1 MI (by improving the ‘supply’ aspect of the oxygen ‘supply/demand mismatch’ in type 2 MIs), yet we saw more of an effect on type 1 MI. An effect secondary to more oxygen delivered by higher haemoglobin levels is supported by the greater HRs seen for time-to-first events than for recurrent events. The difference in haemoglobin is greatest between high- and low-dose arms early in the trial when the first events are happening and there is no difference in haemoglobin when recurrent events are taking place. If an increase in haemoglobin levels was the main mechanism of reduction in MI in PIVOTAL, it is difficult to explain why darbepoetin did not result in a reduction in MI in the TREAT trial.13 This trial was, however, conducted in a different population: patients with chronic kidney disease that were not receiving haemodialysis. It seems likely that the beneficial effects of high-dose iron are contributed to by other effects. An increase in platelets is known to be associated with iron deficiency.14 In the PIVOTAL trial, high-dose iron was associated with lower platelet levels than low-dose iron.6 This may be an additional or alternative mechanism explaining how IV iron reduces acute MIs. Acute MIs could also be reduced due to the direct effects of iron on the endothelium and circulating monocytes but data to support this hypothesis in patients on dialysis are lacking.
In the current trial, rates of death following non-fatal MI were very high (1- and 2-year mortality was $40\%$ and $60\%$, respectively). Although these data are striking, these numbers appear to represent an improvement when compared with data from the 1980s and 1990s when extremely high mortality rates were reported (1-year mortality post-MI of $60\%$).15 Data from the USA reported a 2-year mortality of $71\%$ as recently as 2008. On the other hand, it is possible that the lower mortality rate seen in the contemporary PIVOTAL trial does not reflect a reduction in mortality rates but from the inclusion of clinical events that are now classified as acute MIs because of the introduction of troponin. These acute MIs are likely to have been smaller and associated with better outcomes. In the current trial, the mortality rate in STEMI was very high but this must be verified in studies with larger numbers. Such high death rates after MI highlight that this is an area of major unmet need in cardiovascular medicine.
Whether or not the rate of acute MIs can be reduced in patients on haemodialysis can be improved is finally attracting some attention. ISCHEMIA-CKD 30172098 was an NHLBI funded trial that randomized 777 patients with estimated glomerular filtration rate <30 or on haemodialysis ($53\%$) to a routine invasive strategy (i.e. routine coronary angiography) or a conservative approach with angiography only for ‘failure’ of optimal medical therapy.16 No benefit of a routine invasive strategy on the primary endpoint of death or MI was seen.
Very few trials have investigated the role of medical therapy to reduce MI in patients on haemodialysis. Patients on haemodialysis receiving 10 mg rosuvastatin had similar outcomes to placebo over a mean follow-up of 3.8 years.17 Patients with diabetes on dialysis had no benefit from atorvastatin 20 mg compared to placebo.18 Patients receiving dialysis have been excluded from trials establishing the beneficial effects of angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARBs), and beta-blockers, meaning that it remains unclear as to whether these agents are efficacious or harmful.19 Over half of the patients in the current trial were receiving statins but prescription rates of ACE inhibitors or ARBs were low.
Another potential therapeutic target in patients receiving haemodialysis is hypertension. In the current trial, the mean systolic blood pressure was 145 mmHg. Patients with type 2 MIs had higher systolic blood pressures than those with type 1 MIs. The optimum blood pressure target for patients on haemodialysis is unknown. Perhaps trials of blood pressure lowering could reduce MIs.
## 4.1 Limitations
The PIVOTAL trial was conducted in the UK so the findings may not be generalizable to other countries or regions. The varying international nature of renal disease and cardiovascular disease could plausibly result in different results. Troponins can be chronically elevated in patients on dialysis and clinical presentation of MIs can be atypical.20,21 To overcome this, the Clinical Events Committee combined cardiology as well as nephrology expertise. The PIVOTAL trial design did not require documentation of all troponin values during each possible presentation with an MI. Cardiac magnetic resonance imaging can be useful to identify MIs but late gadolinium cannot be used in patients on haemodialysis. Systematic coronary angiography (including intracoronary imaging) can help to differentiate type 1 from type 2 MIs but is not practical to mandate during a large clinical trial. The capture of unrecognized or ‘silent’ MI was not performed during this trial. Other limitations of this analysis include a shared problem with all studies including type 2 MIs. Before a type 2 MI can be adjudicated a rise or fall in troponin must be seen. To see such a change, at least two measurements must be performed. In other words, if only one troponin (or no troponin at all) is measured, a type 2 MI cannot be diagnosed. Type 2 MIs are therefore always, to some extent, investigator-dependent events.
## 5. Conclusion
MIs occurred in $8\%$ of patients over 2 years of follow-up in patients on maintenance haemodialysis. Most of these MIs are type 1 MIs and NSTEMIs. Mortality remains high after non-fatal MI (1- and 2-year mortality of $40\%$ and $60\%$, respectively). High-dose vs. low-dose IV iron reduces MI in patients in their first year of haemodialysis.
## Supplementary material
Supplementary material is available at Cardiovascular Research online.
## Authors’ contributions
M.C.P.—first draft of the manuscript, study design, and Clinical Events Committee. P.S.J.—critical revision of the manuscript, study design, and Clinical Events Committee. E.C.—critical revision of the manuscript, study design, and Clinical Events Committee. P.B.M.—critical revision of the manuscript, study design, and Clinical Events Committee. M.R.M.D.—critical revision of the manuscript, study design, and Clinical Events Committee. M.R.—statistical analysis, critical revision of the manuscript, study design, and member of Clinical Events Committee. S.D.A.—Steering Committee, trial design, and critical review of the manuscript. S.B.—Steering Committee, trial design, and critical review of the manuscript. K.F.—Steering Committee, trial design, and critical review of the manuscript. P.A.K.—Steering Committee, trial design, and critical review of the manuscript. D.C.W.—Steering Committee, trial design, and critical review of the manuscript. C.R.V.T.—Steering Committee, trial design, and critical review of the manuscript. I.F.—Steering Committee, trial design, and critical review of the manuscript. J.J.V.Mc. M.—Steering Committee, trial design, and critical review of the manuscript. I.C.M.—Chief Investigator, Steering Committee, trial design, and critical review of the manuscript.
## Funding
This work was supported by the Kidney Research UK, which was supported by an unrestricted grant from Vifor Fresenius Medical Care Renal Pharma. M.C.P. and J.J.V.Mc. M. are supported by a British Heart Foundation Centre of Research Excellence (RE/$\frac{18}{6}$/34217).
## Data availability
Although the PIVOTAL Steering *Committee is* not making the data available at present, the committee will consider public release of data in the future.
## References
1. Stadler N, Lindner RA, Davies MJ.. **Direct detection and quantification of transition metal ions in human atherosclerotic plaques: evidence for the presence of elevated levels of iron and copper**. *Arterioscler Thromb Vasc Biol* (2004) **24** 949-954. PMID: 15001454
2. Bailie GR, Schuler C, Leggett RE, Li H, Li H-D, Patadia H, Levin R.. **Oxidative effect of several intravenous iron complexes in the rat**. *Biometals* (2013) **26** 473-478. PMID: 23681275
3. Kuo K-L, Hung S-C, Lee T-S, Tarng D-C.. **Iron sucrose accelerates early atherogenesis by increasing superoxide production and upregulating adhesion molecules in CKD**. *J Am Soc Nephrol* (2014) **25** 2596-2606. PMID: 24722448
4. de Valk B, Marx JJ.. **Iron, atherosclerosis, and ischemic heart disease**. *Arch Intern Med* (1999) **159** 1542-1548. PMID: 10421276
5. Kshirsagar AV, Freburger JK, Ellis AR, Wang L, Winkelmayer WC, Brookhart MA.. **Intravenous iron supplementation practices and short-term risk of cardiovascular events in hemodialysis patients**. *PLoS One* (2013) **8** e78930. PMID: 24223866
6. Macdougall IC, White C, Anker SD, Bhandari S, Farrington K, Kalra PA, McMurray JJV, Murray H, Tomson CRV, Wheeler DC, Winearls CG, Ford I.. **Intravenous iron in patients undergoing maintenance hemodialysis**. *N Engl J Med* (2019) **380** 447-458. PMID: 30365356
7. Macdougall IC, White C, Anker SD, Bhandari S, Farrington K, Kalra PA, McMurray JJV, Murray H, Steenkamp R, Tomson CRV, Wheeler DC, Winearls CG, Ford I. **Randomized trial comparing proactive, high-dose versus reactive, low-dose intravenous iron supplementation in hemodialysis (PIVOTAL): study design and baseline data**. *Am J Nephrol* (2018) **48** 260-268. PMID: 30304714
8. Hicks KA, Mahaffey KW, Mehran R, Nissen SE, Wiviott SD, Dunn B, Solomon SD, Marler JR, Teerlink JR, Farb A, Morrow DA, Targum SL, Sila CA, Thanh Hai MT, Jaff MR, Joffe HV, Cutlip DE, Desai AS, Lewis EF, Gibson CM, Landray MJ, Lincoff AM, White CJ, Brooks SS, Rosenfield K, Domanski MJ, Lansky AJ, McMurray JJV, Tcheng JE, Steinhubl SR, Burton P, Mauri L, O'Connor CM, Pfeffer MA, Hung HMJ, Stockbridge NL, Chaitman BR, Temple R. **2017 cardiovascular and stroke endpoint definitions for clinical trials**. *J Am Coll Cardiol* (2018) **71** 1021-1034. PMID: 29495982
9. Ghosh D, Lin DY.. **Nonparametric analysis of recurrent events and death**. *Biometrics* (2000) **56** 554-562. PMID: 10877316
10. Foley RN, Herzog CA, Collins AJ.. **Smoking and cardiovascular outcomes in dialysis patients: the United States Renal Data System Wave 2 study**. *Kidney Int* (2003) **63** 1462-1467. PMID: 12631362
11. Shroff GR, Li S, Herzog CA.. **Trends in discharge claims for acute myocardial infarction among patients on dialysis**. *J Am Soc Nephrol* (2017) **28** 1379-1383. PMID: 28220031
12. Baigent C.. **Cholesterol metabolism and statin effectiveness in hemodialysis patients**. *J Am Coll Cardiol* (2015) **65** 2299-2301. PMID: 26022818
13. Pfeffer MA, Burdmann EA, Chen C-Y, Cooper ME, de Zeeuw D, Eckardt K-U, Feyzi JM, Ivanovich P, Kewalramani R, Levey AS, Lewis EF, McGill JB, McMurray JJ, Parfrey P, Parving HH, Remuzzi G, Singh AK, Solomon SD, Toto R.. **A trial of darbepoetin alfa in type 2 diabetes and chronic kidney disease**. *N Engl J Med* (2009) **361** 2019-2032. PMID: 19880844
14. Kulnigg-Dabsch S, Schmid W, Howaldt S, Stein J, Mickisch O, Waldhör T, Evstatiev R, Kamali H, Volf I, Gasche C.. **Iron deficiency generates secondary thrombocytosis and platelet activation in IBD: the randomized, controlled thromboVIT trial**. *Inflamm Bowel Dis* (2013) **19** 1609-1616. PMID: 23644823
15. Herzog CA, Ma JZ, Collins AJ.. **Poor long-term survival after acute myocardial infarction among patients on long-term dialysis**. *N Engl J Med* (1998) **339** 799-805. PMID: 9738087
16. Bangalore S, Maron DJ, O'Brien SM, Fleg JL, Kretov EI, Briguori C, Kaul U, Reynolds HR, Mazurek T, Sidhu MS, Berger JS, Mathew RO, Bockeria O, Broderick S, Pracon R, Herzog CA, Huang Z, Stone GW, Boden WE, Newman JD, Ali ZA, Mark DB, Spertus JA, Alexander KP, Chaitman BR, Chertow GM, Hochman JS. **Management of coronary disease in patients with advanced kidney disease**. *N Engl J Med* (2020) **382** 1608-1618. PMID: 32227756
17. Fellström BC, Jardine AG, Schmieder RE, Holdaas H, Bannister K, Beutler J, Chae D-W, Chevaile A, Cobbe SM, Grönhagen-Riska C, De Lima JJ, Lins R, Mayer G, McMahon AW, Parving H-H, Remuzzi G, Samuelsson O, Sonkodi S, Sci D, Süleymanlar G, Tsakiris D, Tesar V, Todorov V, Wiecek A, Wüthrich RP, Gottlow M, Johnsson E, Zannad F. **Rosuvastatin and cardiovascular events in patients undergoing hemodialysis**. *N Engl J Med* (2009) **360** 1395-1407. PMID: 19332456
18. Wanner C, Krane V, März W, Olschewski M, Mann JFE, Ruf G, Ritz E. **Atorvastatin in patients with type 2 diabetes mellitus undergoing hemodialysis**. *N Engl J Med* (2005) **353** 238-248. PMID: 16034009
19. Konstantinidis I, Nadkarni GN, Yacoub R, Saha A, Simoes P, Parikh CR, Coca SG.. **Representation of patients with kidney disease in trials of cardiovascular interventions: an updated systematic review**. *JAMA Intern Med* (2016) **176** 121-124. PMID: 26619332
20. Stacy SR, Suarez-Cuervo C, Berger Z, Wilson LM, Yeh H-C, Bass EB, Michos EB.. **Role of troponin in patients with chronic kidney disease and suspected acute coronary syndrome: a systematic review**. *Ann Intern Med* (2014) **161** 502-512. PMID: 25111593
21. Herzog CA, Littrell K, Arko C, Frederick PD, Blaney M.. **Clinical characteristics of dialysis patients with acute myocardial infarction in the United States: a collaborative project of the United States Renal Data System and the National Registry of Myocardial Infarction**. *Circulation* (2007) **116** 1465-1472. PMID: 17785621
|
---
title: 'Cognitive and Behavioral Development of 9-Year-Old Children After Maternal
Cancer During Pregnancy: A Prospective Multicenter Cohort Study'
authors:
- Indra A. Van Assche
- Evangeline A. Huis in 't Veld
- Kristel Van Calsteren
- Mathilde van Gerwen
- Jeroen Blommaert
- Elyce Cardonick
- Michael J. Halaska
- Robert Fruscio
- Monica Fumagalli
- Jurgen Lemiere
- Elisabeth M. van Dijk-Lokkart
- Camilla Fontana
- Harm van Tinteren
- Jessie De Ridder
- Martine van Grotel
- Marry M. van den Heuvel-Eibrink
- Lieven Lagae
- Frédéric Amant
journal: Journal of Clinical Oncology
year: 2023
pmcid: PMC10022854
doi: 10.1200/JCO.22.02005
license: CC BY 4.0
---
# Cognitive and Behavioral Development of 9-Year-Old Children After Maternal Cancer During Pregnancy: A Prospective Multicenter Cohort Study
## Abstract
Clinical trials frequently include multiple end points that mature at different times. The initial report, typically based on the primary end point, may be published when key planned co-primary or secondary analyses are not yet available. Clinical Trial Updates provide an opportunity to disseminate additional results from studies, published in JCO or elsewhere, for which the primary end point has already been reported.
This multicenter cohort study reports on the long-term effects of prenatal exposure to maternal cancer and its treatment on cognitive and behavioral outcomes in 9-year-old children. In total, 151 children (mean age, 9.3 years; range, 7.8-10.6 years) were assessed using a neurocognitive test battery and parent-report behavioral questionnaires. During pregnancy, 109 children ($72.2\%$) were exposed to chemotherapy (only or in combination with other treatment modalities), 18 ($11.9\%$) to surgery only, 16 ($10.6\%$) to radiotherapy, one to trastuzumab, and 16 ($10.6\%$) were not exposed to oncologic treatment. Mean cognitive and behavioral outcomes were within normal ranges. Gestational age at birth showed a positive association with Full Scale Intelligence Quotient (FSIQ), with the average FSIQ score increasing by 1.6 points for each week increase in gestational age ($95\%$ CI, 0.7 to 2.5; $P \leq .001$). No difference in FSIQ was found between treatment types (F[4,140] = 0.45, $$P \leq .776$$). In children prenatally exposed to chemotherapy, no associations were found between FSIQ and chemotherapeutic agent, exposure level, or timing during pregnancy. These results indicate a reassuring follow-up during the critical maturational period of late childhood, when complex functions develop and rely on the integrity of early brain development. However, associations were observed with preterm birth, maternal death, and maternal education.
## INTRODUCTION
Maternal cancer during pregnancy is an emerging challenge and may affect child development through both direct treatment effects and indirect environmental and psychosocial effects, in both prenatal and postnatal periods.1 As part of the International Network on Cancer, Infertility and Pregnancy, our group previously published reports showing reassuring general outcomes of children prenatally exposed to maternal cancer and its treatment.2-7 Nevertheless, specific results highlight a need for continued follow-up. At a median age of 22 months, prematurity predicted poorer cognitive outcomes.2,3 At age 6 years, prenatally exposed children showed lower verbal intelligence and visuospatial long-term memory scores, and chemotherapy during pregnancy was associated with poorer emotional regulation.6,7 Maternal death also represented an important risk factor for poorer neurocognitive and behavioral outcomes.7-10 The question remains whether children born after a pregnancy complicated by maternal cancer are at risk of growing into deficits when complex cognitive and executive functions are developing and relying on brain structures that developed aberrantly during gestation and early childhood.11,12 This report describes the cognitive and behavioral outcomes of 9-year-old children prenatally exposed to maternal cancer and its treatment.
## Study Design
The INCIP Child Follow-up study (Protocol, online only) evaluates the general health and neurocognitive development after prenatal exposure to maternal cancer (treatment) using age-adapted standardized test batteries.7 Children are included longitudinally at the ages 18 months, 36 months, and subsequently once every 3 years until age 18 years.
We report a cross-sectional analysis of the 9-year-old children (Data Supplement, online only). Ethical approval and written informed consent was obtained for all participating subjects.
## Outcomes
A neuropsychologic examination, assessing intelligence quotient, attention, memory, and behavior, was performed (Data Supplement). The primary outcome was the Full Scale Intelligence Quotient (FSIQ), derived from the Wechsler Intelligence Scale for Children.13-15 Secondary outcomes included all other neurocognitive test scores and behavioral questionnaire results. All children underwent a clinical neurologic and general pediatric examination, and parents completed a questionnaire on their child's general health and educational level. All tests were conducted in the child's native language.
Methodologic details are reported in the Data Supplement.
## Participants
A total of 151 children (including seven pairs of twins) were included: 95 from Belgium, 34 from Netherlands, nine from Italy, six from Czech Republic, and seven from New Jersey (Table 1). The Data Supplement contains further information about maternal cancer types, treatment characteristics, substance use during pregnancy, fertility treatment, bilingualism, congenital malformations, and labor types and delivery modes.
## Cognitive Development and Behavior
Group outcomes for all intelligence outcomes, verbal and visuospatial memory, attentional function, and behavioral measures were within normal ranges (Table 2 and Fig 1).
No difference in FSIQ was found between girls and boys (t[143] = 0.25, $$P \leq .802$$), treatment types (F[4,140] = 0.45, $$P \leq .776$$), or cancer stages (F[3,140] = 1.53, $$P \leq .211$$).
Children who scored below normal ranges were more likely to have been born preterm. The average FSIQ score increased by 1.6 points ($95\%$ CI, 0.7 to 2.5; $P \leq .001$; Data Supplement) for each week increase in gestational age at birth (GA). GA also explained verbal intelligence (β = 1.49 points/wk; CI, 0.6 to 2.4; $$P \leq .002$$), performance intelligence (β = 1.39 points/wk; CI, 0.6 to 2.2; $$P \leq .002$$), and processing speed (β = 1.21 points/wk; CI, 0.3 to 2.1; $$P \leq .009$$).
We found an effect of maternal bereavement (F[2, 144] = 3.94, $$P \leq .022$$). FSIQ was lower in children with a deceased mother before age 2 years (93.13 ± 12.65) than in children with a surviving mother (104.08 ± 14.88). When adjusting for GA, this association disappeared ($P \leq .3$): Children with a deceased mother before age 2 years were on average born earlier (33.4 weeks ± 2.7) than children with a deceased mother after age 2 years (37.1 weeks ± 1.7) and children with a surviving mother (36.4 weeks ± 2.5; $P \leq .001$). No associations were found between maternal death and cancer type, cancer stage, or treatment type.
Multiple linear regression (Data Supplement) with GA, death of mother, and parental (maternal and paternal) education level as explanatory variables shows that GA ($$P \leq .006$$) and maternal education level ($$P \leq .005$$) remained to explain FSIQ. Specifically, this model estimated an increase of 1.27 points (± 0.45) in FSIQ for every week increase in GA.
When looking specifically at the subgroup of children prenatally exposed to chemotherapy, a mixed-effect regression model, with parental education levels as random variables, found no significant associations with FSIQ (all $P \leq .08$). When parental education levels were separately included as fixed effects, maternal education level remained a significant predictor of FSIQ ($$P \leq .011$$).
## General Health
Data from the parent-reported health questionnaire revealed no specific problems across the group. An overview of the reported medical problems is enlisted in the Data Supplement.
Four children were diagnosed with attention-deficit hyperactivity disorder, of which three took supportive medication. One child was diagnosed with autism spectrum disorder. Thirty-one children ($24.4\%$) received remedial care, including speech therapy (17 children), remedial teaching at school [10], a type of physical or exercise therapy (four), and neurofeedback therapy (one).
From these children receiving remedial support, eight children ($25.8\%$) lost their mother (of which seven before age 2 years), with a chi-squared test showing a significant relation between maternal bereavement before age 2 years and remedial care (χ2 [2,127] = 7.33, $$P \leq .026$$, φ = 0.240). An association was also found between remedial support and prematurity: 24 children ($77.4\%$) who received remedial care were born preterm (χ2 [1,127] = 6.67, $$P \leq .010$$, φ = 0.229).
## DISCUSSION
Cognitive and behavioral outcomes of 9-year-old children born to mothers diagnosed with cancer during pregnancy did not differ with norms of the general population. We reported average group scores for all intelligence index scores, verbal and visuospatial memory outcomes, attentional functioning outcomes, and behavioral questionnaire measures. We found no impact of sex, treatment type, and cancer stage.
Children who scored below normal ranges on intelligence index scores were more likely to have been born preterm. We report an increase in Full Scale Intelligence Quotient of almost 6.5 points for each month increase in GA. Hence, preterm birth should be avoided as much as possible in the obstetric management of pregnant women with cancer.
Seventeen percent of the children in this study had lost their mother (of which almost $58\%$ died before their child turned 2 years old). The results suggest a relationship between maternal bereavement in the first 1,000 days of life and Full Scale Intelligence, although a larger sample is necessary to disentangle effects of maternal loss and often coinciding premature birth.
A longitudinal study in the general population also demonstrates multicollinearity between the impact of early-life adversity exposure and prematurity.18 As maternal death was not associated with cancer type, cancer stage, or treatment type, it could be possible that maternal stress or an insecure mother-child attachment may play a role in determining child cognitive development.
In the subgroup of children prenatally exposed to chemotherapy, only maternal education level was associated with Full Scale Intelligence. In this group, no effect of prematurity was found in both the current 9-year-old cohort and the 6-year-old cohort.7 Further research is needed to determine whether the effect of prematurity is hidden by an underpowered analysis in the smaller chemotherapy group, or whether administering chemotherapy during pregnancy perhaps reduces the beneficial effect of delivery at term.
Maternal education level explained part of the outcome on Full Scale Intelligence. This effect was found in the entire sample and chemotherapy subgroup. Although both paternal and maternal education level were associated with Full Scale Intelligence at the univariate level, only maternal education remained a significant predictor in fitted models.
Almost a quarter of parents reported a need for remedial support for their child. As rates of pediatric remedial support in the general population remain under-reported, disallowing a comparison of the observed rates and the rates of the general population, further research is needed to put this percentage in perspective. The need for remedial support may be related to prematurity and maternal bereavement, with $77\%$ of the children who received support born preterm, and $26\%$ of the children having lost their mother to cancer. In combination with a recent study showing increased child trauma–related cognitions and emotion regulation difficulties after parental cancer (bereavement),19 it tentatively supports the need to follow these families in the long term to determine their psychosocial needs.
In conclusion, children prenatally exposed to maternal cancer (treatment) showed on average a normal cognitive and behavioral development at age 9 years. No associations of cognition and behavior were found with treatment type, exposure level, cancer stage, and gestational age at the start of treatment. These results convey a reassuring follow-up of neurocognitive development during late childhood, which represents a time of critical maturation when complex functions are developing. However, these children still require a close follow-up, as maternal cancer during pregnancy is associated with preterm delivery and maternal death, which are risk factors for developmental problems.
## AUTHOR CONTRIBUTIONS
Conception and design: Indra A. Van Assche, Evangeline A. Huis in 't Veld, Kristel Van Calsteren, Michael J. Halaska, Harm van Tinteren, Marry M. van den Heuvel-Eibrink, Lieven Lagae, Frédéric Amant Administrative support: Indra A. Van Assche, Evangeline A. Huis in 't Veld, Frédéric Amant Provision of study materials or patients: Evangeline A. Huis in 't Veld, Kristel Van Calsteren, Mathilde van Gerwen, Robert Fruscio, Frédéric Amant Collection and assembly of data: Indra A. Van Assche, Evangeline A. Huis in 't Veld, Kristel Van Calsteren, Mathilde van Gerwen, Elyce Cardonick, Michael J. Halaska, Robert Fruscio, Monica Fumagalli, Camilla Fontana, Jessie De Ridder, Martine van Grotel, Marry M. van den Heuvel-Eibrink, Frédéric Amant
Data analysis and interpretation: Indra A. Van Assche, Evangeline A. Huis in 't Veld, Kristel Van Calsteren, Mathilde van Gerwen, Jeroen Blommaert, Michael J. Halaska, Robert Fruscio, Monica Fumagalli, Jurgen Lemiere, Elisabeth M. van Dijk-Lokkart, Camilla Fontana, Harm van Tinteren, Marry M. van den Heuvel-Eibrink, Frédéric Amant Manuscript writing: All authors Final approval of manuscript: All authors Accountable for all aspects of the work: All authors
## Cognitive and Behavioral Development of 9-Year-Old Children After Maternal Cancer During Pregnancy: A Prospective Multicenter Cohort Study
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.
Open *Payments is* a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).
## References
1. Van Assche IA, Lemiere J, Amant F. **Direct and indirect effects on child neurocognitive development when maternal cancer is diagnosed during pregnancy: What do we know so far?**. *Crit Rev Oncol Hematol* (2022) **179** 103824. PMID: 36174901
2. Amant F, Van Calsteren K, Halaska MJ. **Long-term cognitive and cardiac outcomes after prenatal exposure to chemotherapy in children aged 18 months or older: An observational study**. *Lancet Oncol* (2012) **13** 256-264. PMID: 22326925
3. Amant F, Vandenbroucke T, Verheecke M. **Pediatric outcome after maternal cancer diagnosed during pregnancy**. *N Engl J Med* (2015) **373** 1824-1834. PMID: 26415085
4. Blommaert J, Zink R, Deprez S. **Long-term impact of prenatal exposure to chemotherapy on executive functioning: An ERP study**. *Clin Neurophysiol* (2019) **130** 1655-1664. PMID: 31330451
5. Blommaert J, Radwan A, Sleurs C. **The impact of cancer and chemotherapy during pregnancy on child neurodevelopment: A multimodal neuroimaging analysis**. *EClinicalMedicine* (2020) **28** 100598. PMID: 33294813
6. van Gerwen M, Vandenbroucke T, Gorissen AS. **Executive functioning in 6 year old children exposed to chemotherapy in utero**. *Early Hum Develop* (2020) **151** 105198
7. Vandenbroucke T, Verheecke M, van Gerwen M. **Child development at 6 years after maternal cancer diagnosis and treatment during pregnancy**. *Eur J Cancer* (2020) **138** 57-67. PMID: 32858478
8. Paksarian D, Eaton WW, Mortensen PB. **A population-based study of the risk of schizophrenia and bipolar disorder associated with parent–child separation during development**. *Psychol Med* (2015) **45** 2825-2837. PMID: 25952483
9. Thomas AW, Caporale N, Wu C. **Early maternal separation impacts cognitive flexibility at the age of first independence in mice**. *Dev Cogn Neurosci* (2016) **18** 49-56. PMID: 26531108
10. van Gerwen M, Huis in 't Veld E, van Grotel M. **Long-term neurodevelopmental outcome after prenatal exposure to maternal hematological malignancies with or without cytotoxic treatment**. *Child Neuropsychol* (2021) **27** 822-833. PMID: 33876721
11. Sesma HW, Georgieff MK. **The effect of adverse intrauterine and newborn environments on cognitive development: The experiences of premature delivery and diabetes during pregnancy**. *Dev Psychopathol* (2003) **15** 991-1015. PMID: 14984135
12. Luciana M. **Cognitive development in children born preterm: Implications for theories of brain plasticity following early injury**. *Dev Psychopathol* (2003) **15** 1017-1047. PMID: 14984136
13. Wechsler D. *WISC-V: Technical and Interpretive Manual* (2014)
14. Wechsler D. *Wechsler Intelligence Scale for Children* (1991)
15. Wechsler D. *Wechsler Intelligence Scale for Children—Fourth Edition (WISC-IV)* (2003)
16. 16.UNESCO Institute for Statistics: International Standard Classification of Education: ISCED 2011. Montreal, Quebec, Canada, UNESCO Institute for Statistics, 2012. *International Standard Classification of Education: ISCED 2011* (2012)
17. Meekes J, van Schooneveld MMJ, Braams OB. **Parental education predicts change in intelligence quotient after childhood epilepsy surgery**. *Epilepsia* (2015) **56** 599-607. PMID: 25705968
18. Turpin H, Urben S, Ansermet F. **The interplay between prematurity, maternal stress and children’s intelligence quotient at age 11: A longitudinal study**. *Sci Rep* (2019) **9** 450. PMID: 30679588
19. Egberts MR, Verkaik D, van Baar AL. **Child posttraumatic stress after parental cancer: Associations with individual and family factors**. *J Pediatr Psychol* (2022) **47** 1031-1043. PMID: 35595308
|
---
title: Rapid neutrophil mobilization by VCAM-1+ endothelial cell-derived extracellular
vesicles
authors:
- Naveed Akbar
- Adam T Braithwaite
- Emma M Corr
- Graeme J Koelwyn
- Coen van Solingen
- Clément Cochain
- Antoine-Emmanuel Saliba
- Alastair Corbin
- Daniela Pezzolla
- Malene Møller Jørgensen
- Rikke Bæk
- Laurienne Edgar
- Carla De Villiers
- Mala Gunadasa-Rohling
- Abhirup Banerjee
- Daan Paget
- Charlotte Lee
- Eleanor Hogg
- Adam Costin
- Raman Dhaliwal
- Errin Johnson
- Thomas Krausgruber
- Joey Riepsaame
- Genevieve E Melling
- Mayooran Shanmuganathan
- Adrian Banning
- Adrian Banning
- Raj Kharbanda
- Neil Ruparelia
- Mohammad Alkhalil
- GianLiugi De Maria
- Lisa Gaughran
- Erica Dall’Armellina
- Vanessa Ferreira
- Alessandra Borlotti
- Yujun Ng
- Christoph Bock
- David R F Carter
- Keith M Channon
- Paul R Riley
- Irina A Udalova
- Kathryn J Moore
- Daniel C Anthony
- Robin P Choudhury
journal: Cardiovascular Research
year: 2022
pmcid: PMC10022859
doi: 10.1093/cvr/cvac012
license: CC BY 4.0
---
# Rapid neutrophil mobilization by VCAM-1+ endothelial cell-derived extracellular vesicles
## Body
See the editorial comment for this article ‘Extracellular vesicles selectively mobilize splenic neutrophils’, by R. Panda and P. Kubes, https://doi.org/10.1093/cvr/cvad015.
## Abstract
### Aims
Acute myocardial infarction rapidly increases blood neutrophils (<2 h). Release from bone marrow, in response to chemokine elevation, has been considered their source, but chemokine levels peak up to 24 h after injury, and after neutrophil elevation. This suggests that additional non-chemokine-dependent processes may be involved. Endothelial cell (EC) activation promotes the rapid (<30 min) release of extracellular vesicles (EVs), which have emerged as an important means of cell–cell signalling and are thus a potential mechanism for communicating with remote tissues.
### Methods and results
Here, we show that injury to the myocardium rapidly mobilizes neutrophils from the spleen to peripheral blood and induces their transcriptional activation prior to arrival at the injured tissue. Time course analysis of plasma-EV composition revealed a rapid and selective increase in EVs bearing VCAM-1. These EVs, which were also enriched for miRNA-126, accumulated preferentially in the spleen where they induced local inflammatory gene and chemokine protein expression, and mobilized splenic-neutrophils to peripheral blood. Using CRISPR/Cas9 genome editing, we generated VCAM-1-deficient EC-EVs and showed that its deletion removed the ability of EC-EVs to provoke the mobilization of neutrophils. Furthermore, inhibition of miRNA-126 in vivo reduced myocardial infarction size in a mouse model.
### Conclusions
Our findings show a novel EV-dependent mechanism for the rapid mobilization of neutrophils to peripheral blood from a splenic reserve and establish a proof of concept for functional manipulation of EV-communications through genetic alteration of parent cells.
## Graphical Abstract
Graphical Abstract
## 1. Introduction
Acute myocardial infarction (AMI) is a substantial sterile injury that leads to a rapid increase in peripheral blood neutrophils.1–5 Elevated peripheral blood neutrophil number post-AMI correlates with the extent of myocardial injury, degree of cardiac dysfunction, and mortality.1–3 Neutrophil depletion enhances susceptibility to cardiac rupture6 and antibody depletion of neutrophils prior to AMI increases infarct size, enhances fibrosis, and lowers the number of M2 macrophages in the healing myocardium.1,7 However, inhibition of neutrophil recruitment in AMI reduces infarct size.1 These competing findings suggest a complex role for neutrophils in the contexts of myocardial ischaemic injury and repair.
The bone marrow is the primary site for granulopoiesis8,9 and has been regarded as the principal source of neutrophils that are mobilized to peripheral blood after injury.4,10 Mature neutrophils are held in large numbers in the haemopoietic cords, separated from the blood by the sinusoidal endothelium.11 In the current paradigm, these cells are retained in the marrow by the interaction of CXCR4 and CXCL12 [stromal cell-derived factor (SDF-1α)]12 and mobilized in response to soluble factors. Intravascular injection of a range of chemotactic factors, including leukotriene B4, C5a, interleukin-8 (IL-8),13 CXCL chemokines,12,14 and granulocyte-colony stimulating factor (G-CSF)15,16 can drive the rapid mobilization of neutrophils across the sinusoidal endothelium. However, numerous strands of evidence question whether chemokines derived from injured tissues are responsible for very early neutrophil mobilization in vivo. Intra-cardiac mRNA levels for cytokines peak 12 h after injury17 and pro-inflammatory proteins are very modestly elevated in coronary sinus following reperfusion therapy in AMI.18,19 Furthermore, in vivo blood chemokine profiles peak 24 h post-AMI and do not precede the rise in blood neutrophil counts in humans or mice, which occurs within 2 h in mice following injury,1,7 whereas a large majority of rodent AMI studies investigating neutrophils dynamics following AMI have focused on neutrophil elevations 6–24 h post-injury.1,4 Moreover, a putative source of chemokine generation in the acutely ischaemic myocardium prior to neutrophil infiltration has not been identified.
These observations suggest that neutrophils may be mobilized from alternative reserves following injury and by mechanisms that are not dependent on chemokines. One possible source is extramedullary haematopoiesis in the spleen20 from where neutrophils are mobilized to peripheral blood following bacterial infection.21 By analogy, it is known that monocytes are deployed from a splenic reserve following sterile injury in mice,22 and that this can be driven by extracellular vesicles (EVs) that are derived from the vascular endothelium.23 EVs are membrane-enclosed envelopes24 that are actively secreted by many cell types.25–27 These vesicles bear bioactive cargo that includes proteins and microRNAs (miRNAs), which are derived from the parent cell. EV can alter the biological function and cellular status of cells locally28 and remotely following liberation into the blood.29 Endothelial cell (EC)-derived EVs (EC-EVs) bearing vascular cell adhesion molecule-1 (VCAM-1) are elevated in the blood following AMI23,28 and have a role in the mobilization and transcriptional programming of splenic monocytes in AMI.23 Here, we sought to establish whether EC-EVs contribute to the very early mobilization and programming of neutrophils and, if so, through which of their component parts. We hypothesized that EC-EVs released during AMI would localize to neutrophils in remote reserves in a process mediated by VCAM-1, which has been shown to bind to neutrophils via surface integrins.30 Furthermore, we reasoned that once localized to neutrophils in reserve pools, EC-EV-miRNA cargo could induce functionally relevant transcriptional programmes in those target tissues and cells prior to recruitment to the injured myocardium. An understanding of these mechanisms would immediately suggest possibilities for cell-selective immuno-modulatory interventions that are relevant in AMI and, potentially, other pathologies with an inflammatory component.
## 2. Methods
Translational studies using whole blood, plasma, plasma neutrophils, plasma EV and CMR imaging in human patients, mouse models of AMI with and without antagmiR treatment, RNA-sequencing, flow cytometry, in vivo EV injections, and in vitro studies using human and mouse ECs were employed here. Full experimental details are provided in the Supplementary material online.
## 2.1 AMI patients
All clinical investigations were conducted in accordance with the Declaration of Helsinki. The Oxfordshire Research Ethics Committee (references 08/H$\frac{0603}{41}$ and 11/SC/0397) approved human clinical cohort protocols and conformed to the principles outlined in the Declaration of Helsinki. All patients provided informed written consent for inclusion in the study.
## 2.2 LAD ligation model
All animal procedures were approved by an ethical review committee at the University of Oxford or NYU Lagone Health. Animal experiments conform to the guidelines from Directive $\frac{2010}{63}$/EU of the European Parliament on the protection of animals used for scientific purposes or the current NIH guidelines. UK experimental interventions were carried out by UK Home Office personal licence holders under the authority of a Home Office project licence. AMI was induced in adult wild-type (WT) female C57B6/J mice as previously described.1 Due to the higher incidence of acute ventricular rupture in male mice.3 Mice were anesthetized with $4\%$ isofluorane and maintained under $2.5\%$ isoflurane under assisted external ventilation through the insertion of an endotracheal tube (∼200 strokes min − 1; stroke volume ∼200 μL min − 1). Buprenorphine (buprenorphine hydrochloride; Vetergesic) was delivered as a 0.015 mg/mL solution via intraperitoneal injection at 20 min before the procedure to provide analgesia. Post-AMI animals were euthanized by cervical dislocation and peripheral blood cells, splenocytes, bone marrow, and cardiac cells were isolated.
## 2.3 Statistical analysis
All values are group mean ± standard deviation (SD). Paired and unpaired two-tailed Student’s t-test was used to compare two groups, a one-way or two-way analysis of variance (ANOVA) or mixed model effects with post-hoc Bonferroni or Tukey correction was used to compare multiple group (>2) means with one, two or more independent variables. P-values <0.05 were considered significant. Hierarchical clustering analysis and generation of heatmap plots was performed using the pheatmap R package v1.0.12.
## 3.1 Plasma neutrophil number correlates with the extent of AMI
In acute ST-segment-elevation AMI (STEMI) peripheral blood neutrophil number at the time of presentation [median time from onset of symptoms to percutaneous coronary intervention (PCI) 3 h] correlated with the extent of ischaemic injury, as determined by oedema estimation on T2-weighted magnetic resonance imaging (MRI) images obtained within 48 h of AMI (R2 = 0.365, $$P \leq 0.017$$) (Figure 1A) and with final infarct size, determined by late gadolinium enhancement (LGE) MRI 6 months post-AMI (R2 = 0.507, $$P \leq 0.003$$) (Figure 1B).
**Figure 1:** *Human peripheral blood neutrophils correlate with the extent of myocardial injury in AMI. (A) Pearson’s correlation of peripheral blood neutrophil number (109/L) in patients experiencing AMI significantly correlated with the extent of myocardial injury (T2-weight MRI) and (B) LGE MRI 6-months post-AMI (n=15). (C) Schematic representation of mouse AMI and tissue harvesting for flow cytometry. (D) Percentage of neutrophils in peripheral blood, spleen, bone marrow, and heart 2 h after AMI in mice relative to the levels of intact controls (controls n = 4, AMI n = 3). (E) Mean fluorescence intensity of CD62L/L-selectin on mouse neutrophils in peripheral blood, spleen, bone marrow, and heart 2 h after AMI relative to the levels of intact controls (controls n = 4, AMI n = 3). (F) Percentage of monocytes in peripheral blood, spleen, bone marrow, and heart 2 h after AMI in mice relative to the levels of intact uninjured controls (controls n = 4, AMI n=3). Pearson’s correlation was used in (A) and (B), dotted lines represent 95% confidence interval and an unpaired t-test was used in (D)–(F) for statistical analysis. Error bars represent mean ± SD **P < 0.01, ***P < 0.001.*
## 3.2 AMI mobilizes neutrophils from the spleen without alterations in systemic chemokines
This rapid increase in peripheral blood neutrophils is consistent with mobilization from an existing reserve. To determine the source of neutrophil mobilization in the very early hours post-AMI, we performed left anterior descending artery ligation in a mouse model of AMI and analysed cell populations from blood, spleen, bone marrow, and the heart 2 h after AMI, by flow cytometry (Figure 1C). AMI induced a 6.3-fold ($P \leq 0.01$) increase in peripheral blood neutrophils (Live, CD45+, CD11b+, Ly6G+) (Supplementary material online, Figure S1) and simultaneously lowered splenic-neutrophil number by 0.7-fold ($P \leq 0.001$) (Figure 1D). As described previously, to obtain an indication of the mobilization between reserves, we calculated a neutrophil mobilization ratio23 [peripheral blood neutrophils/splenic [or bone marrow] neutrophils] and found an increase in splenic-neutrophil mobilization (8.5-fold) ($P \leq 0.01$), but no alteration in bone marrow neutrophil number relative to control animals. There was no significant alteration in CD62L/L-selectin (which is shed during neutrophil activation) in mobilized peripheral blood neutrophils (Figure 1E). At this very early time point (2 h post-AMI), we found no differences in LyC6high monocyte number in the peripheral blood or spleen, indicating that neutrophils mobilize from the spleen prior to splenic-monocyte mobilization22 (Figure 1F).
The prevailing paradigm is that chemokines are rapidly released from ischaemic tissues and mobilize reserves of neutrophils to the peripheral blood following AMI. To determine this in the hyper-acute phase, when splenic-neutrophils are deployed, we undertook a quantitative protein-detection array for 25 different proteins that influence neutrophil function in plasma obtained in 2 h and 24 h post-AMI in our mouse model. We found no alterations in systemic cytokines 2 h post-AMI and only found a significant increase in CCL6 24 h post-AMI ($P \leq 0.05$) (Supplementary material online, Figure S2A).
S100A8 and S100A9 are released following AMI by activated neutrophils.4 We measured S100A8/S100A9 heterodimer in the plasma of mice following AMI and found a significant 6.6-fold induction 2 h post-AMI, which was maintained 24 h post-AMI (6.8-fold) when compared to control mice (both, $P \leq 0.01$) (Supplementary material online, Figure S1B). These data demonstrate a rapid increase in peripheral blood neutrophils from the splenic reserve 2 h post-AMI without significant alterations in systemic plasma cytokines.
## 3.3 Human plasma EVs correlate with the extent of peripheral blood neutrophil counts in AMI and myocardial scar 6-months post-AMI
In agreement with our previous findings, patients with AMI had significantly more plasma EVs at time of presentation (24.3 × 108 ± 25.7 EV/mL) vs. a 6-month follow-up measurement (11.0 × 108 ± 12.5 EV/mL, $P \leq 0.010$) but they exhibited a similar EV size distribution profile, with an elevation in EVs in the size range 100–200 nm diameter (Figure 2A and B). Plasma-EV number at presentation significantly correlated with the extent of myocardial scarring as determined by LGE at 6 months post-AMI (R2 = 0.423, $$P \leq 0.009$$) (Figure 2C). We also found a highly significant relationship between plasma-EV number and peripheral blood neutrophil number (R2 = 0.753, $P \leq 0.001$) at time of presentation (Figure 2D), consistent with a possible role for plasma EVs in neutrophil mobilization post-AMI.
**Figure 2:** *VCAM-1+ plasma EVs are elevated in peripheral blood following AMI. (A) Human plasma-EV number (108/mL) at time of presentation following AMI and 6 months later in the same patients (n =15). (B) Size and concentration profile of human plasma EVs at time of presentation following AMI and 6 months later in the same patients (n =15) determined by Nanoparticle Tracking Analysis. Pearson’s correlation of human plasma EVs at time of presentation vs. and: (C) LGE MRI 6-months post-AMI, (D) number of peripheral blood neutrophils following AMI (109/L) (n =15) in the same patients. (E) Heat map showing human plasma-EV markers CD9, CD63, CD81, ALIX, TSG101, flotillin-1, annexin V and (F) heat map showing human plasma-EV EC markers thrombomodulin, VEGFR2, endoglin, MCAM, ICAM-1, VCAM-1, VE-cadherin, tissue factor and CD16 in the same patients at: presentation, immediately following post-PCI, 6, 24, and 48 h post-PCI and 6 months post-AMI (n = 10 per time point). A paired t-test was used for statistical analysis in (A). Error bars in (B) represent mean ± SD. Heat maps in (E) and (F) are group means per time point. Values were normalized to the 6-month time point per patient. Pearson’s correlation was used in (C) and (D), dotted lines represent 95% confidence interval for statistical analysis. **P < 0.01.*
## 3.4 Human VCAM-1+ plasma EVs are enriched at time of presentation with AMI
In the same patients, we determined the composition of plasma EVs at six different time points: at presentation (prior to PCI), immediately following PCI and at 6, 24, 48 h, and 6 months post-AMI using a validated high throughput immunoaffinity EV-protein array.31 To interpret the time course of the EV response after AMI, we normalized the EV-markers in each patient to those obtained at the 6 months post-AMI time point.
Generic EV-markers CD9, CD63, flotillin-1, CD81, ALIX, and annexin V were highly abundant in plasma EVs at the time of presentation (Figure 2E). At presentation plasma-EV number and EV-CD63 (R2 = 0.863, $$P \leq 0.001$$) and EV-ALIX (R2 = 0.724, $$P \leq 0.018$$) showed significant associations. Following PCI, there was augmentation of plasma-EV CD81, ALIX, TSG101, and flotillin-1, which subsided 6, 24, and 48 h post-presentation/post-PCI, approaching levels that were comparable to those at 6 months post-AMI (Figure 2E). Plasma EVs displayed typical morphology by transmission electron microscopy (TEM) were negative for markers of cellular contamination by histone H3 and washing of plasma EVs with phosphate-buffered saline lowered levels of apoB and albumin in isolated plasma EVs (Supplementary material online, Figure S3A and B).
EVs carry proteins on their surface, which can reflect their cellular origin. We determined the composition of EV in relation to EC markers, measuring thrombomodulin, vascular endothelial growth factor receptor 2 (VEGFR2), endoglin, melanoma cell adhesion molecule (MCAM), intercellular adhesion molecule-1 (ICAM-1), VCAM-1, and VE-cadherin on plasma EVs. We additionally examined tissue factor and CD16.
Plasma EVs were enriched for VCAM-1 at the time of presentation, prior to PCI and showed significant associations with plasma-EV number at presentation (R2 = 0.745, $$P \leq 0.013$$) and with EV-markers CD63 (R2 = 0.714, $$P \leq 0.020$$) and ALIX (R2 = 0.651, $$P \leq 0.042$$). VCAM-1+ plasma EV was highest at the earliest time point and diminished over time (Figure 2F). Plasma-EV thrombomodulin and VE-cadherin were elevated following PCI but showed no associations with plasma-EV number at presentation. Tissue factor and monocyte/neutrophil marker CD16 also showed distinct patterning post-PCI (Figure 2F), but with later peaks than for VCAM-1. These data suggest orchestrated rapid enrichment of plasma EVs-bearing VCAM-1 in the context of AMI.
## 3.5 Human EC-EVs are enriched for miRNA-126
In order to probe the function of EVs derived exclusively from activated ECs, we studied an established model of primary human umbilical cord vein ECs in vitro. Compared with basal conditions, treatment of ECs with pro-inflammatory tumour necrosis factor-α (TNF-α) activated ECs as evidenced by enhanced VCAM-1 protein expression ($P \leq 0.001$) (Figure 3A) and increased EV production ($P \leq 0.001$) (Figure 3B and C), with a significant increase in small EVs, of similar size range (100–200 nm) to those found to increase in patients with AMI. EC-EVs displayed typical EV morphology (Figure 3D and E), the EV-protein marker CD9 (Figure 3F) and were positive for endothelial nitric oxide synthase (eNOS).
**Figure 3:** *Human umbilical cord vein endothelial cells (HUVEC) release more EVs after inflammatory stimulation. (A) HUVECs: express more VCAM-1 following treatment with recombinant human TNF-α (10 ng/mL) (n = 9 per group); (B) release more EVs (n = 8 per group). (C) Size and concentration profile of HUVEC-derived EVs under basal conditions and after inflammatory stimulation with recombinant human TNF-α (n = 8 per group). (D) TEM of HUVEC-derived EVs (scale bar 100 nm) and (E) cryo-TEM HUVEC-derived EVs (scale bar 50 nm). (F) Ponceau stain and western blot of HUVEC-derived EV from basal and after inflammatory stimulation with TNF-α for eNOS, TSG101, CD9, ATP5A, and Histone H3. HUVEC cell pellets, EV-depleted cell culture supernatants (EV-dep), and cell culture media that was not exposed to cells (control media) were used as controls. EC-EV miRNA levels of (G) hsa-miRNA-126-3p and (H) hsa-miRNA-126-5p under basal conditions and after inflammatory stimulation with TNF-α (n = 8 per group). miRNA-126-mRNA targets in human and mouse and their target pathways. (I) Euler plot of miRNA-126-mRNA targets from TargetScanHuman, TargetScanMouse, miRWalk, miRDB for human and the mouse. (J) Euler plot of GO terms for miRNA-126-mRNA targets for the human and mouse. Shape areas are approximately proportional to number of genes. An unpaired t-test was used in (A), (B), (C), (G), and (H) for statistical analysis. Error bars represent mean ± SD **P < 0.01, ***P < 0.001.*
EC-EVs derived from pro-inflammatory stimulations show significant enrichment for miRNA-126-3p ($P \leq 0.010$) (Figure 3G) and miRNA-126-5p ($P \leq 0.010$) (Figure 3H), consistent with previous observations of changes in miRNA, measured in the unselected plasma-EV pool, following AMI23 at a time point consistent with elevated VCAM-1+ plasma EVs and prior to PCI.
## 3.6 miRNA-126-mRNA targets cluster selectively in neutrophil motility pathways
To explore the potential role of EC-EV-miRNA-126, we employed in silico techniques, curating miRNA-126 putative-mRNA target genes from three separate miRNA–mRNA target prediction databases for human and mouse.32–34 We determined whether the mRNAs putatively regulated by miRNA-126 for the human, mouse or the overlap gene set (targeted in both the human and mouse) (Figure 3I and J and Supplementary material online, Tables S1 and S2) were present in Gene Ontology (GO) terms for neutrophil function. miRNA-126-putative-mRNA targets were significantly overrepresented when compared by Fisher’s exact test to neutrophil pathway GO terms for neutrophil migration (GO: GO1990266) and neutrophil chemotaxis (GO: GO0030593) in the human (both $P \leq 0.001$), the mouse (both $P \leq 0.001$), and the overlap gene set (both $P \leq 0.001$) (Supplementary material online, Table S3). Whereas, other neutrophil GO terms, such as neutrophil-mediated killing of a fungus (GO: GO0070947), neutrophil clearance (GO: GO0097350), and regulation of neutrophil-mediated cytotoxicity (GO: 0070948) were not enriched (Supplementary material online, Table S3), suggesting a possible role for EC-EV-miRNA-126 in orchestrating processes related to neutrophil mobilization post-AMI.
## 3.7 AMI alters human and mouse neutrophil transcriptomes
To determine whether neutrophil transcriptomes are altered post-AMI, we obtained peripheral blood neutrophils from newly recruited patients presenting with STEMI ($$n = 3$$) and non-STEMI (NSTEMI) control patients ($$n = 3$$) at time of presentation and matched-control samples 1 month later. STEMI patients had a greater number of differentially expressed genes at time of presentation vs. NSTEMI control patients (STEMI 933 genes vs. NSTEMI 8 genes) (Figure 4A–C).
**Figure 4:** *RNA sequencing of human peripheral blood neutrophils. STEMI and NSTEMI patients at the time of presentation vs. a control sample obtained from the same patients 1 month post-AMI (n = 3 per group). MA plots show differential transcriptome at the time of presentation vs. a control sample obtained from the same patients 1 month post-AMI in (A) NSTEMI and (B) STEMI patients. Significantly altered genes are highlighted in red. (C) Euler plot showing similarity and differences in the number of differentially expressed (DE) genes in NSTEMI and STEMI patients at time of presentation vs. 1 month follow-control samples or between all NSTEMI and all STEMI patients (n = 3 per group). (D) miRNA-126 antagomiR treatment of WT mice prior to induction of AMI. (E) TTC staining of the myocardium 24 h post-AMI in scramble and antagomiR treated mice (scramble n = 7 and antagomiR n = 5 per group). Significant DE genes in (A)–(C) were determined by adjusted P-values below the 5% FDR threshold. Error bars represent mean ± SD **P < 0.01.*
To further understand the potential target pathways for the differentially enriched genes in blood neutrophils following AMI, we used GO term enrichment analysis and Reactome pathway analysis35 ranked by false discovery rate (FDR)-adjusted P-values. GO analysis showed that differentially expressed neutrophil genes at the time of presentation favoured pathways for signal recognition particle (SRP)-dependent co-translational protein targeting to membrane (GO: 0006614 and R-HSA-1799339) (both, $P \leq 0.001$), co-translational protein targeting to membrane (GO: 0006613) ($P \leq 0.001$), and neutrophil degranulation (R-HSA-6798695) ($P \leq 0.001$) (Supplementary material online, Table S4).
Next, we used single cell (sc)-RNA-sequencing data to determine whether neutrophil populations in the peripheral blood of mice subjected to AMI exhibited similar transcriptomic alterations prior to recruitment to the heart. We found differential enrichment in neutrophil populations in the blood following AMI,36 which favoured pathway terms for neutrophil aggregation (GO: 0070488) (Supplementary material online, Tables S5 and S6) ($P \leq 0.05$), platelet activation (GO: 0030168) ($P \leq 0.001$), platelet activation, signalling, and aggregation (R-HSA-76002) ($P \leq 0.001$) (Supplementary material online, Table S7).
There were significant overlaps between the genes that are differentially expressed following AMI in the blood of the human and the mouse ($P \leq 0.001$) (Supplementary material online, Table S8) and significant similarity in target pathways (GO terms: biological process, molecular function and cellular component, and Reactome pathways) between the human and mouse (Supplementary material online, Table S9).
## 3.8 miRNA-126-mRNA targets are overrepresented in neutrophil transcriptomes following AMI
Human miRNA-126-mRNA targets were significantly overrepresented in human neutrophil transcriptomes at the time of injury ($P \leq 0.05$) (Supplementary material online, Table S10). Similarly, in mice, neutrophils within the myocardium (but not peripheral blood) showed differential enrichment for miRNA-126-mRNA targets (Supplementary material online, Table S10).
To test the functional significance of these findings, we treated WT mice with an antagomiR for miRNA-126 ($$n = 5$$) or a scramble control ($$n = 7$$) prior to induction of experimental AMI. miRNA-126 (Figure 4D) reduced infarct size by $12\%$ compared with scramble control ($P \leq 0.01$) (Figure 4E) (representative images—Supplementary material online, Figure S4A and B).
## 3.9 EC-EVs localize to the spleen
These accumulating data suggest that EC-EVs, enriched for miRNA-126 and VCAM-1 provide an ‘ischaemia signal’ to neutrophils in the spleen, resulting in mobilization and transcriptional activation. Accordingly, we tested whether EC-EV localized to the spleen after intravenous injection and whether there were consequent alterations in neutrophil-associated chemokine gene and protein expression. Primary mouse and human ECs release more EVs following inflammatory stimulation23 and hypoxia.37 In agreement with these data, mouse sEND.1 ECs produced EV under basal conditions and released significantly more EVs after pro-inflammatory stimulation with TNF-α ($P \leq 0.001$) (Figure 5A and B).23 sEND.1 TNF-α-activated ECs produced more VCAM-1 protein ($P \leq 0.001$) (Figure 5C). sEND.1-derived EVs displayed typical EV morphology (Figure 5D and E), EV-protein markers (ALIX, TSG101, and CD9) (Figure 5F) and were positive for eNOS and VCAM-1. EC-EVs derived from pro-inflammatory stimulations showed significant enrichment for miRNA-126-3p ($P \leq 0.001$) (Figure 5G) and miRNA-126-5p ($P \leq 0.01$) (Figure 5H), indicating similarities in the EC-EV response between human and mouse ECs. We labelled the mouse EC-EV by transfection with non-mammalian miRNA-39, which belongs to Caenorhabditis elegans and allows quantitative tracing of EVs in vivo (Figure 6A). EC-EVs accumulated preferentially in the spleen compared with bone marrow ($P \leq 0.05$), brain ($P \leq 0.001$), heart ($P \leq 0.05$), kidney ($P \leq 0.05$) (Supplementary material online, Figure S5A) 1 h post-injection, and remained detectable in the spleen for 2, 6 ($P \leq 0.001$), and 24 h ($P \leq 0.001$) (Figure 6A).
**Figure 5:** *Mouse sEND.1 ECs release more EVs after inflammatory stimulation. (A) Mouse sEND.1 ECs express more VCAM-1 following treatment with recombinant mouse TNF-α (10 ng/mL) (n = 9 per group); (B) release more EVs (n = 11 per group). (C) Size and concentration profile of sEND.1-derived EVs under basal conditions (n = 3) and after inflammatory stimulation with recombinant mouse TNF-α (n = 4). (D) TEM of sEND.1-derived EVs (scale bar 1000 nm) and (E) cryo-TEM sEND.1-derived EVs (scale bar 1000 nm). (F) Ponceau stain and western blot of sEND.1-derived EV from basal and after inflammatory stimulation with TNF-α for ALIX, TSG101, CD9, eNOS, VCAM-1, ATP5A, and Histone H3. sEND1 cell pellets, EV-depleted cell culture supernatants (EV-dep), and cell culture media that was not exposed to cells (control) were used as controls. EC-EV miRNA levels of (G) hsa-miRNA-126-3p and (H) hsa-miRNA-126-5p under basal conditions (n= 8) and after inflammatory stimulation with TNF-α (n = 7). An unpaired t-test was used in (A), (B), (C), (G), and (H) for statistical analysis. Error bars represent mean ± SD **P < 0.01, ***P < 0.001.* **Figure 6:** *Mouse EC-EVs localize to the spleen in WT mice and influence gene and protein expression and mobilize splenic-neutrophils. (A) RT–qPCR detection of EC-EV labelled with miRNA-39-3p in the spleen of mice following intravenous injection of 1×109 EVs by tail vein at: 2 (n = 4) and 6 h (n = 4) post-injection and control injections (n = 5); and 24 h (n = 6) post-injection and control injections (n = 5). Control represents a media only preparation with no EC-EVs. (B) Heat map showing gene expression in the spleen of mice following intravenous injection of 1×109 EVs by tail vein at 2 (n = 4) and 6 h (n = 4) post-injection and control injections (n = 5); and 24 h (n = 6) post-injection and control injections (n = 5). Control represents a media only preparation with no EC-EVs. Data shown as ΔΔCt values normalized to row mean ΔΔCt value for each gene. (C) Heat map showing protein expression in the spleen of mice following intravenous injection of 1×109 EVs by tail vein at 2 h (n = 4) post-injection. Control (n = 4) represents a media only preparation with no EC-EVs. Data shown are chemokine array dot blot density values normalized to mean row value for each protein. (D) Schematic of experiment. (E) Percentage of neutrophils as a proportion of the total leukocytes (live, CD45+, CD11b+, and Ly6G+) in peripheral blood, bone marrow, and spleen (n = 5 per group). (F) Splenic-neutrophil mobilization ratio (peripheral blood neutrophils/spleen neutrophils) shows net contributions of neutrophil reserves to mobilized peripheral blood neutrophils following intravenous injections of EC-EV (1×109 EVs/mL) injections (n = 5 per group). (G) Mean fluorescent intensity of CD62L/L-selectin on neutrophils in peripheral blood, spleen, and bone marrow 2 h after (n = 5 per group) intravenous injections of EC-EV (1×109 EVs/mL). A one-way ANOVA with post-hoc Bonferroni correction was used in (A), (B), and an unpaired t-test was used in (C). An unpaired t-test was used in (E)–(G). Error bars represent mean ± SD *P < 0.05, **P < 0.01, ***P < 0.001.*
## 3.10 EC-EVs alter chemokine and protein expression in the spleen
Informed by the earlier in silico studies suggesting regulation of neutrophil activation and motility by miRNA-126-mRNAs, we hypothesized that EC-EV localization in the spleen would alter gene expression within spleen tissue, with a focus on CXC chemokine and cytokine activity.
EC-EVs significantly induced mRNA expression for Cxcr2, Itag4, Gapdh (all, $P \leq 0.05$), Il-1β, Cxcl1 (both, $P \leq 0.01$), Cxcr4, and Il-6 (both, $P \leq 0.001$) post-EC-EV injection (Figure 6B).
We further determined whether delivery of EC-EVs to the spleen altered chemokine protein levels, including for the retention chemokine CXCL12/SDF-1. In the same mice, we undertook the quantitative protein-detection array for 25 different proteins that influence neutrophil function, including CXCL12/SDF-1, CCL2,38 and CCL3,39 which are known miRNA-126-mRNA targets and CCL27/CCL28, which are predicted miRNA-126-mRNA targets. There were significant reductions in CCL21 ($P \leq 0.01$), CXCL13 ($P \leq 0.05$), chemerin/retinoic acid receptor responder protein 2 ($P \leq 0.01$), IL-16 ($P \leq 0.05$), MCP-5/CCL12 ($P \leq 0.05$), and CXCL12/SDF-1 ($P \leq 0.05$) (Figure 6C). These findings are consistent with a role for EC-EV-miRNA-126 in silencing genes involved in cell retention.
## 3.11 EC-EVs mobilize neutrophils from the spleen
Given the effects of the EC-EVs derived from TNF-α activated cells on gene expression and silencing of retention chemokines, we injected EC-EVs, derived from TNF-α activated ECs, intravenously into healthy WT mice vs. control media only injections with no EC-EVs (Figure 6D). Flow cytometry (Live, CD45+, CD11b+, Ly6G+) showed that EC-EVs significantly increased the number of circulating peripheral blood neutrophils (Figure 6E), and simultaneously lowered splenic-neutrophil numbers in the same mice (Figure 6F), confirming splenic-neutrophil mobilization induced by EC-EV. Consistent with our observations in AMI, we found that EC-EVs mediated greater neutrophil mobilization from the spleen ($P \leq 0.001$) than from the bone marrow (Figure 6F). As in the context of AMI, there was no alteration in CD62L/L-selectin expression in blood neutrophils (Figure 6G).
## 3.12 EC-EV-VCAM-1 mediates neutrophil mobilization
VCAM-1 positive EVs increases in the immediate hours after AMI (Figure 2G). Similarly, ECs in culture produce EVs enriched for VCAM-1 following pro-inflammatory stimulation (Figure 5E). Given its well-established role in mediating interactions between activated vascular endothelium and circulating leukocytes, we hypothesized that VCAM-1 on the surface of EC-EVs might perform the converse role by mediating the capture of circulating EC-EVs by static neutrophils in the spleen. We confirmed the presence of VCAM-1 on the surface of plasma EVs using immunoaffinity capture. Magnetic beads of iron oxide were conjugated to IgG control or anti-VCAM-1 antibodies and incubated with isolated human plasma EVs. Subsequent TEM shows specific capture of VCAM-1+ plasma EVs from a heterogenous pool through EV-surface expression of VCAM-1 (Figure 7A and Supplementary material online, Figures S6A–E and S7A–F).
**Figure 7:** *EV VCAM-1 is necessary for EC-EV splenic-neutrophil mobilization in mice. (A) TEM of a VCAM-1+ plasma EV bound to a magnetic bead of iron oxide conjugated with anti-human VCAM-1 antibodies, scale bar is 200 nm. (B/C) Western blot of sEND.1 WT and CRISPR-cas9 base-edited VCAM-1 KO cell pellets under basal conditions (WT n = 4 and VCAM-1 KO n = 4 per group) and after inflammatory stimulation with recombinant mouse tumour necrosis (TNF-α). (D) The number of mouse sEND.1 EC-EVs from WT and CRISPR-cas9 base-edited VCAM-1 KOs under basal conditions (WT n = 12 and VCAM-1 KO n = 11 per group) and after inflammatory stimulation with recombinant mouse TNF-α (n = 8 per group). (E) Size and concentration profile of mouse sEND.1 EC-EVs from WT and CRISPR/Cas9 base-edited VCAM-1 KOs under basal conditions (WT n = 12 and VCAM-1 KO n = 11 per group) and after inflammatory stimulation with recombinant mouse TNF-α (n = 8 per group). (F) Ponceau stain and western blot of WT and CRISPR-case9 base-edited VCAM-1 KO sEND.1-derived EVs from basal and after inflammatory stimulation with recombinant mouse TNF-α for TSG101, CD9, and VCAM-1. Inflammatory stimulated sEND1 cell pellets and EV-depleted cell culture supernatants were used as controls. (G) RT–qPCR detection of WT sEND.1 and CRISPR-cas9 base-edited VCAM-1 KOs EC-EV labelled with miRNA-39-3p in the spleen of mice following intravenous injection of 1×109 EVs by tail vein at 2 h post-injection (n = 5 per group). (H/I) Percentage of neutrophils as a proportion of the total leukocytes (live, CD45+, CD11b+, and Ly6G+) in peripheral blood and spleen (control and VCAM-1 KO EC-EV n = 4 and WT EC-EV n = 5 per group). (I) Splenic-neutrophil mobilization ratio (peripheral blood neutrophils/spleen neutrophils) shows net contributions of splenic reserves to mobilized peripheral blood neutrophils following intravenous injections of WT or CRISPR-cas9 base-edited VCAM-1 KO EC-EVs 1×109 EVs by tail vein at 2 h post-injection. Control represents a media only preparation with no EC-EVs (control and VCAM-1 KO EC-EV n = 3 and WT EC-EV n = 5 per group). (J) Heat map showing mRNA expression in the spleen of mice following intravenous injection of WT or CRISPR-cas9 base-edited VCAM-1 KO EC-EVs 1×109 EVs by tail vein at 2 h post-injection. Control represents a media only preparation with no EC-EVs (n = 5 per group). Data shown as ΔΔCt values normalized to row mean ΔΔCt value for each gene. One-way (H–J) and two-way (B–E) ANOVA with post-hoc Bonferroni correction was used for statistical analysis. An unpaired t-test was used in (F). Error bars represent mean ± SD *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.*
We next used CRISPR-Cas9 base editing of ECs to produce VCAM-1 deficient EC-EVs by introducing a stop codon in the VCAM-1 sequence. To confirm CRISPR-Cas9 editing of VCAM-1 from ECs, we stimulated WT and VCAM-1 knock-out (KO) cells with TNF-α. WT mouse ECs expressed more VCAM-1 following inflammatory stimulation ($P \leq 0.001$) (Figure 7B and C), whereas VCAM-1 KO cells did not express VCAM-1, confirming successful CRISPR-Cas9 base editing in ECs. VCAM-1 KO cells released EC-EV under basal conditions similar to WT cells (Figure 7D and E) but VCAM-1 KO ECs did not release more EVs following inflammatory stimulation with TNF-α (Figure 7D and E). WT and VCAM-1 KO EC-EV were positive for EV-markers TSG101 and CD9 but only WT inflammatory-derived EC-EVs were positive for VCAM-1 (Figure 7F). EC-EVs derived from either TNF-α-activated WT or TNF-α-activated VCAM-1 KO cells were injected into healthy, WT mice at the same concentration (1 × 109/mL EC-EVs). Using the miRNA-39-3p labelling technique (described above), we found that VCAM-1 deficient EC-EVs and WT EC-EV accumulate in the spleen at similar levels (Figure 7G), but VCAM-1 deficient EC-EVs did not induce alteration in gene expression that were comparable to WT EC-EVs responses in the spleen for Il-6 ($P \leq 0.001$), Il-1β ($P \leq 0.05$), and Cxcl1 ($P \leq 0.05$) (Figure 7J). Deletion of VCAM-1 in EC-EVs prevented mobilization of splenic-neutrophils to peripheral blood when compared to WT VCAM-1+ EC-EVs (Figure 7I and J).
## 4. Discussion
Mobilization of neutrophils occurs rapidly after AMI in mice and humans1,3,4 and their number in peripheral blood correlates with the extent of myocardial injury.1–3 The bone marrow has been regarded as the principal source for neutrophils that are mobilized to peripheral blood after injury, because (i) it is the primary site for granulopoiesis;8–10 (ii) it contains ample reserves of mature cells; and (iii) releases neutrophils in response to injection of exogenous chemokines.12–16 However, the divergent timings of neutrophil mobilization (rapid)1,7 and chemokine elevation (delayed) in vivo17,18,40 suggest that additional processes may be involved.
Here, we have identified a previously unknown mechanism by which ischaemic injury to the myocardium signals to mobilization of neutrophils from a splenic reserve. We show that: (i) EC-EV generated under conditions of inflammation are enriched for VCAM-1, miRNA-126-3p, and miRNA-126-5p and are elevated in peripheral blood at presentation; (ii) EC-EVs are delivered to the spleen, where they alter gene and protein expression; and (iii) induce the mobilization of splenic-neutrophils to peripheral blood. Notably, (iv) these EC-EV effects are dependent on VCAM-1. Furthermore, (v) we show that neutrophil transcriptomes are differentially regulated following AMI, prior to entry into the myocardium. ( vi) Targets of miRNA-126 are significantly altered in neutrophil transcriptomes post-AMI and (vii) administration of miRNA-126 antagomir significantly reduces infarct size in vivo.
We utilized a systemic antagomiR strategy to determine the influence of miRNA-126 on infarct size in our rodent model. Treatment of mice by intraperitoneal injection with an antagomiR by repeated injections 5 and 2 days prior to AMI surgery may result in off target effects and does not selectively interfere with neutrophil mobilization from the spleen. However, the data reported here are consistent with a role for miRNA-126 in neutrophil activation and show lower cardiac injury following antagomiR treatment, possibly through abrogated neutrophil activation or recruitment to the injured heart. Future investigations into the role of miRNA-126 in neutrophil activation in the rodent model of AMI may benefit from more selective targeting of neutrophils through genetic approaches or the use of bioengineered EC-EV for specific immunomodulation.
Mature neutrophils are held in large numbers in the haemopoietic cords through interactions with the neutrophil receptors CXCR2 and CXCR4.11,41,42 Loss of CXCL12 induces an increase in peripheral blood neutrophils. Injection of chemotactic factors,13 CXCL chemokines,12,14 and G-CSF15,16 can drive the rapid mobilization of neutrophils across the sinusoidal endothelium through alterations in CXCR4-CXCL12.
Numerous studies have shown that neutrophil elevation in the blood and myocardium within 24 h in the rodent model of AMI but neutrophils are already elevated in patient blood by the time of arrival at the hospital. Scrutiny of the relative timings of cytokine elevation after ischaemic injury in relation to neutrophil mobilization does not support their role in this early mobilization, since both the onset and peaks in neutrophil mobilization occur prior to those for cytokine elevation.17,18,40 IL-8 injection mobilized neutrophils from the bone marrow,13 but after reperfusion in AMI, even in blood from the coronary sinus (undiluted myocardial effluent), the elevation is modest (0.1-fold).18,19 Furthermore, we calculate that the absolute concentration based on these physiological measurements is ∼2–3 orders of magnitude less than the concentration used to elicit neutrophil mobilization in mice.14 Finally, neutrophils are the first cells to arrive in the acutely injured tissue. Neutrophil depletion dampens plasma chemokines levels following AMI7 and in a mouse air pouch model.43 *It is* not clear which other cells in the profoundly ischaemic myocardium could be capable of the rapid synthesis of chemokines, that would be of sufficient magnitude to mediate neutrophil mobilization from a remote site, such as the bone marrow. The sympathetic nervous system is also activated following AMI and mobilizes committed myeloid lineage cells and neutrophil progenitors from the bone marrow.44 However, unlike terminally differentiated neutrophils, blood numbers of myeloid lineage committed cells and neutrophil progenitors do not peak until >6 h post-AMI, subsequent to the increase in peripheral blood neutrophils.
By contrast, numerous studies have shown that hypoxia promotes the rapid (<30 min) release of EVs by ECs.37 We show that activated ECs in culture liberate large amounts of EV that contain VCAM-1 in their membranes. Using CRISPR/Cas9 genome base editing of cultured ECs, we generated VCAM-1-deficient EV and showed that while VCAM-1 was not essential for splenic localization, its absence removed the ability of EV to provoke the rapid mobilization of neutrophils. Importantly, EVs are taken up rapidly and selectively by the spleen where they become locally concentrated,23,45 unlike chemokines, which have a systemic effect.
Following injury to the myocardium there is a marked increase in VCAM-1-bearing EVs. VCAM-1 is a glycoprotein, which is expressed on activated endothelium and has a well-established role in the recruitment of circulating leukocytes by binding integrins,30,46,47 including CD49d.48 Therefore, our findings suggest an efficient signalling system, in which neutrophils are activated and mobilized by engaging VCAM-1-bearing EVs that are taken up in the spleen, having been released remotely from activated endothelium. A subsequent interaction between neutrophils in circulation and static VCAM-1 on activated ECs mediates their recruitment to the original site of injury.
Deficiency of VCAM-1 by CRISPR/Cas9 in ECs impaired EC-EV release following TNF-α stimulation. The underlying mechanism for this remains unknown, but cellular integrins, such as MAC-1 form important signalling pathways for EV biogenesis in neutrophils49 and similarly, VCAM-1 may be necessary for inflammation induced EC-EV biogenesis.
The recruitment of neutrophils to the injured myocardium is an essential step in tissue response to injury and repair1,7 and thus modulating the neutrophil response raises possibilities for immuno-modulatory interventions in selected inflammatory pathologies, including AMI. Peripheral blood neutrophils are elevated at time of presentation with AMI in patients and rapidly increase in peripheral blood following AMI in mice. Here, we show that splenic-neutrophils are rapidly mobilized to peripheral blood by EC-EV-bearing VCAM-1. These findings complement the current paradigm in which neutrophils are liberated from bone marrow reserves through elevations in blood chemokines. We demonstrate a novel and efficient signalling mechanism between the injured heart microvasculature and the spleen. ECs are ideally placed for the rapid release of EVs to peripheral blood during ischaemia. EV clearance is rapid and predominately to the spleen, which contains neutrophils in the sup-capsular red pulp. Precisely how neutrophils are retained in the spleen is not known, but our findings suggest that local chemokine signalling may be important, as delivery of EC-EVs rich in miRNA-126 down-regulates retention chemokines, including the miRNA-126 target CXCL12, and induces expression of neutrophil mobilization signals, namely CXCL1.
Importantly, we show that these processes are dependent on EV-VCAM-1, an integrin ligand with a well-documented role in immune cell recruitment. Furthermore, we make a new observation that peripheral blood neutrophils are transcriptionally activated prior to recruitment to the injured myocardium, with a bias towards miRNA-126-mRNA targets. Our bioinformatics analysis revealed SRP-dependent co-translational protein targeting to membrane as the most significantly enriched pathway that was conserved between the human and the mouse blood neutrophils following AMI. SRP is necessary for transferring newly synthesized nascent proteins from the ribosome, which are destined for cellular excretion. Enrichment of SRP-dependent pathways prior to tissue recruitment may prime neutrophils for subsequent degranulation and protein secretion. At 1 day post-AMI neutrophils in the myocardium display significant enrichment for degranulation and protein secretion.50 A significant enrichment for the universally conserved SRP supports our conclusion that neutrophils are transcriptionally active prior to recruitment to the injured heart. Targeting SRP may open novel opportunities to target neutrophils prior to tissue recruitment to modulate their survival and function, thereby protecting injured tissues from pro-inflammatory neutrophil mediated damage.
In conclusion, we demonstrate that the injured myocardium can rapidly mobilize splenic-neutrophils through generation and release of EC-EVs that bear VCAM-1. These findings provide novel insights into how neutrophils are mobilized to peripheral blood following ischaemic injury, without the need for immediate generation and release of chemokines. EVs are decorated in surface proteins and integrins, which allows them to interact with cells and home to specific sites.51,52 A functionally efficient reciprocity may operate, in which VCAM-1 on EC-EVs is required for the mobilization of splenic-neutrophils, complementing the known role of static VCAM-1 in the recruitment of circulating neutrophils to activated endothelium. Neutrophils are the first cells recruited to the ischaemic myocardium and are a major source of chemokines.7 Thus, the well-established mobilization of neutrophils from bone marrow reserves in response to chemokines12–16 represent a secondary response, which is consistent with the time course of earlier reports and with our observations that there is no rapid bone marrow mobilization in the early phase.
We have shown proof of concept that genetic manipulation can alter EV properties in functionally important ways. Immunomodulation of the neutrophil and monocyte response to AMI using EV vectors may provide therapeutic opportunities in AMI.
## Supplementary material
Supplementary material is available at Cardiovascular Research online.
## Authors’ contributions
N.A. isolated, characterized, and utilized EVs in the described experiments and performed western blots, RT–qPCR, chemokine arrays, analysis of infarcted hearts, and assisted with in silico bioinformatics. A.T.B., C.L., D.P., E.L., L.E., G.E.M., E.M.C., and C.v. S. assisted with experimentation. E.M.C., G.J.K., and C.v. S. undertook antagomiR experiments. R.B. and M.M.J. undertook the EV-array. A.T.B. performed bioinformatics analysis and generated transcriptome analysis graphs and heat maps. A.L.C. and D.P. prepared and analysed flow cytometry preparations. A.C., R.D. and E.J. imaged EVs by TEM/Cryo-TEM. M.G.H. performed some AMI surgeries. T.K. and C.B. performed RNA sequencing. J.R. performed CRISPR-cas9 base editing of endothelial cells. K.J.M., P.R., K.M., and D.A. led animal investigations. R.P.C. and K.M.C. led human AMI investigations. N.A. and R.P.C. conceived the study. All authors participated in study design, coordination, and helped to draft the manuscript. All authors have seen the final version of the manuscript and approve of its submission.
## Funding
This work was supported by research grants from the British Heart Foundation (BHF) Centre of Research Excellence, Oxford (N.A. and R.P.C.: RE/$\frac{13}{1}$/30181 and RE/$\frac{18}{3}$/34214); British Heart Foundation Project Grant (N.A. and R.P.C.: PG/$\frac{18}{53}$/33895); the Tripartite Immunometabolism Consortium, Novo Nordisk Foundation (R.P.C.: NNF15CC0018486); Oxford Biomedical Research Centre (BRC); Nuffield Benefaction for Medicine and the Wellcome Institutional Strategic Support Fund (ISSF) (N.A.); the National Institutes of Health (NIHR) [R35HL135799, P01HL131478 (K.J.M.) and T32HL098129 (C.v. S.)]; the American Heart Association [19CDA346300066 (C.v. S.)] and D.R.F.C. and G.E.M. acknowledge Biotechnology and Biological Sciences Research Council (BB/P$\frac{006205}{1}$).
## Data availability statement
RNA-*Sequencing data* are deposited at Gene Expression Omnibus (GSE187571) and available by contacting the corresponding author. The data underlying this article are available in Gene Expression Omnibus at https://www.ncbi.nlm.nih.gov/geo/, and can be accessed with GSE187571 and upon reasonable request to the corresponding author.
## References
1. Schloss MJ, Horckmans M, Nitz K, Duchene J, Drechsler M, Bidzhekov K, Scheiermann C, Weber C, Soehnlein O, Steffens S.. **The time-of-day of myocardial infarction onset affects healing through oscillations in cardiac neutrophil recruitment**. *EMBO Mol Med* (2016) **8** 937-948. PMID: 27226028
2. Guo T-M, Cheng B, Ke L, Guan S-M, Qi B-L, Li W-Z, Yang B.. **Prognostic value of neutrophil to lymphocyte ratio for in-hospital mortality in elderly patients with acute myocardial infarction**. *Curr Med Sci* (2018) **38** 354-359. PMID: 30074197
3. Kong T, Kim TH, Park YS, Chung SP, Lee HS, Hong JH, Lee JW, You JS, Park I.. **Usefulness of the delta neutrophil index to predict 30-day mortality in patients with ST segment elevation myocardial infarction**. *Sci Rep* (2017) **7** 15718. PMID: 29146994
4. Sreejit G, Abdel-Latif A, Athmanathan B, Annabathula R, Dhyani A, Noothi SK, Quaife-Ryan GA, Al-Sharea A, Pernes G, Dragoljevic D, Lal H, Schroder K, Hanaoka BY, Raman C, Grant MB, Hudson JE, Smyth SS, Porrello ER, Murphy AJ, Nagareddy PR.. **Neutrophil-derived S100A8/A9 amplify granulopoiesis after myocardial infarction**. *Circulation* (2020) **141** 1080-1094. PMID: 31941367
5. Chia S, Nagurney JT, Brown DFM, Raffel OC, Bamberg F, Senatore F, Wackers FJT, Jang I-K.. **Association of leukocyte and neutrophil counts with infarct size, left ventricular function and outcomes after percutaneous coronary intervention for ST-elevation myocardial infarction**. *Am J Cardiol* (2009) **103** 333-337. PMID: 19166685
6. Lörchner H, Pöling J, Gajawada P, Hou Y, Polyakova V, Kostin S, Adrian-Segarra JM, Boettger T, Wietelmann A, Warnecke H, Richter M, Kubin T, Braun T.. **Myocardial healing requires Reg3beta-dependent accumulation of macrophages in the ischemic heart**. *Nat Med* (2015) **21** 353-362. PMID: 25751817
7. Horckmans M, Ring L, Duchene J, Santovito D, Schloss MJ, Drechsler M, Weber C, Soehnlein O, Steffens S.. **Neutrophils orchestrate post-myocardial infarction healing by polarizing macrophages towards a reparative phenotype**. *Eur Heart J* (2017) **38** 187-197. PMID: 28158426
8. Evrard M, Kwok IWH, Chong SZ, Teng KWW, Becht E, Chen J, Sieow JL, Penny HL, Ching GC, Devi S, Adrover JM, Li JLY, Liong KH, Tan L, Poon Z, Foo S, Chua JW, Su IH, Balabanian K, Bachelerie F, Biswas SK, Larbi A, Hwang WYK, Madan V, Koeffler HP, Wong SC, Newell EW, Hidalgo A, Ginhoux F, Ng LG. **Developmental analysis of bone marrow neutrophils reveals populations specialized in expansion, trafficking, and effector functions**. *Immunity* (2018) **48** 364-379.e368. PMID: 29466759
9. Zhu YP, Padgett L, Dinh HQ, Marcovecchio P, Blatchley A, Wu R, Ehinger E, Kim C, Mikulski Z, Seumois G, Madrigal A, Vijayanand P, Hedrick CC. **Identification of an early unipotent neutrophil progenitor with pro-tumoral activity in mouse and human bone marrow**. *Cell Rep* (2018) **24** 2329-2341.e2328. PMID: 30157427
10. Horckmans M, Bianchini M, Santovito D, Megens RTA, Springael J-Y, Negri I, Vacca M, Di Eusanio M, Moschetta A, Weber C, Duchene J, Steffens S.. **Pericardial adipose tissue regulates granulopoiesis, fibrosis, and cardiac function after myocardial infarction**. *Circulation* (2018) **137** 948-960. PMID: 29167227
11. Furze RC, Rankin SM.. **Neutrophil mobilization and clearance in the bone marrow**. *Immunology* (2008) **125** 281-288. PMID: 19128361
12. Martin C, Burdon PCE, Bridger G, Gutierrez-Ramos JC, Williams TJ, Rankin SM.. **Chemokines acting via CXCR2 and CXCR4 control the release of neutrophils from the bone marrow and their return following senescence**. *Immunity* (2003) **19** 583-593. PMID: 14563322
13. Terashima T, English D, Hogg JC, van Eeden SF.. **Release of polymorphonuclear leukocytes from the bone marrow by interleukin-8**. *Blood* (1998) **92** 1062-1069. PMID: 9680376
14. Jagels MA, Hugli TE.. **Neutrophil chemotactic factors promote leukocytosis. A common mechanism for cellular recruitment from bone marrow**. *J Immunol* (1992) **148** 1119-1128. PMID: 1310708
15. Semerad CL, Liu FL, Gregory AD, Stumpf K, Link DC.. **G-CSF is an essential regulator of neutrophil trafficking from the bone marrow to the blood**. *Immunity* (2002) **17** 413-423. PMID: 12387736
16. Köhler A, De Filippo K, Hasenberg M, van den Brandt C, Nye E, Hosking MP, Lane TE, Männ L, Ransohoff RM, Hauser AE, Winter O, Schraven B, Geiger H, Hogg N, Gunzer M.. **G-CSF-mediated thrombopoietin release triggers neutrophil motility and mobilization from bone marrow via induction of Cxcr2 ligands**. *Blood* (2011) **117** 4349-4357. PMID: 21224471
17. Deten A, Volz HC, Briest W, Zimmer HG.. **Cardiac cytokine expression is upregulated in the acute phase after myocardial infarction. Experimental studies in rats**. *Cardiovasc Res* (2002) **55** 329-340. PMID: 12123772
18. Neumann F-J, Ott I, Gawaz M, Richardt G, Holzapfel H, Jochum M, SchöMig A.. **Cardiac release of cytokines and inflammatory responses in acute myocardial infarction**. *Circulation* (1995) **92** 748-755. PMID: 7543831
19. Marx N, Neumann FJ, Ott I, Gawaz M, Koch W, Pinkau T, Schömig A.. **Induction of cytokine expression in leukocytes in acute myocardial infarction**. *J Am Coll Cardiol* (1997) **30** 165-170. PMID: 9207638
20. Puga I, Cols M, Barra CM, He B, Cassis L, Gentile M, Comerma L, Chorny A, Shan M, Xu W, Magri G, Knowles DM, Tam W, Chiu A, Bussel JB, Serrano S, Lorente JA, Bellosillo B, Lloreta J, Juanpere N, Alameda F, Baró T, de Heredia CD, Torán N, Català A, Torrebadell M, Fortuny C, Cusí V, Carreras C, Diaz GA, Blander JM, Farber C-M, Silvestri G, Cunningham-Rundles C, Calvillo M, Dufour C, Notarangelo LD, Lougaris V, Plebani A, Casanova J-L, Ganal SC, Diefenbach A, Aróstegui JI, Juan M, Yagüe J, Mahlaoui N, Donadieu J, Chen K, Cerutti A.. **B cell-helper neutrophils stimulate the diversification and production of immunoglobulin in the marginal zone of the spleen**. *Nat Immunol* (2011) **13** 170-180. PMID: 22197976
21. Deniset JF, Surewaard BG, Lee WY, Kubes P.. **Splenic Ly6G(high) mature and Ly6G(int) immature neutrophils contribute to eradication of S. pneumoniae**. *J Exp Med* (2017) **214** 1333-1350. PMID: 28424248
22. Swirski FK, Nahrendorf M, Etzrodt M, Wildgruber M, Cortez-Retamozo V, Panizzi P, Figueiredo J-L, Kohler RH, Chudnovskiy A, Waterman P, Aikawa E, Mempel TR, Libby P, Weissleder R, Pittet MJ.. **Identification of splenic reservoir monocytes and their deployment to inflammatory sites**. *Science* (2009) **325** 612-616. PMID: 19644120
23. Akbar N, Digby JE, Cahill TJ, Tavare AN, Corbin AL, Saluja S, Dawkins S, Edgar L, Rawlings N, Ziberna K, McNeill E, Johnson E, Aljabali AA, Dragovic RA, Rohling M, Belgard TG, Udalova IA, Greaves DR, Channon KM, Riley PR, Anthony DC, Choudhury RP. **Endothelium-derived extracellular vesicles promote splenic monocyte mobilization in myocardial infarction**. *JCI Insight* (2017) **2** e93344. PMID: 28878126
24. Thery C, Witwer KW, Aikawa E, Alcaraz MJ, Anderson JD, Andriantsitohaina R, Antoniou A, Arab T, Archer F, Atkin-Smith GK, Ayre DC, Bach JM, Bachurski D, Baharvand H, Balaj L, Baldacchino S, Bauer NN, Baxter AA, Bebawy M, Beckham C, Bedina Zavec A, Benmoussa A, Berardi AC, Bergese P, Bielska E, Blenkiron C, Bobis-Wozowicz S, Boilard E, Boireau W, Bongiovanni A, Borràs FE, Bosch S, Boulanger CM, Breakefield X, Breglio AM, Brennan MÁ, Brigstock DR, Brisson A, Broekman ML, Bromberg JF, Bryl-Górecka P, Buch S, Buck AH, Burger D, Busatto S, Buschmann D, Bussolati B, Buzás EI, Byrd JB, Camussi G, Carter DR, Caruso S, Chamley LW, Chang YT, Chen C, Chen S, Cheng L, Chin AR, Clayton A, Clerici SP, Cocks A, Cocucci E, Coffey RJ, Cordeiro-da-Silva A, Couch Y, Coumans FA, Coyle B, Crescitelli R, Criado MF, D'Souza-Schorey C, Das S, Datta Chaudhuri A, de Candia P, De Santana EF, De Wever O, Del Portillo HA, Demaret T, Deville S, Devitt A, Dhondt B, Di Vizio D, Dieterich LC, Dolo V, Dominguez Rubio AP, Dominici M, Dourado MR, Driedonks TA, Duarte FV, Duncan HM, Eichenberger RM, Ekström K, El Andaloussi S, Elie-Caille C, Erdbrügger U, Falcón-Pérez JM, Fatima F, Fish JE, Flores-Bellver M, Försönits A, Frelet-Barrand A, Fricke F, Fuhrmann G, Gabrielsson S, Gámez-Valero A, Gardiner C, Gärtner K, Gaudin R, Gho YS, Giebel B, Gilbert C, Gimona M, Giusti I, Goberdhan DC, Görgens A, Gorski SM, Greening DW, Gross JC, Gualerzi A, Gupta GN, Gustafson D, Handberg A, Haraszti RA, Harrison P, Hegyesi H, Hendrix A, Hill AF, Hochberg FH, Hoffmann KF, Holder B, Holthofer H, Hosseinkhani B, Hu G, Huang Y, Huber V, Hunt S, Ibrahim AG, Ikezu T, Inal JM, Isin M, Ivanova A, Jackson HK, Jacobsen S, Jay SM, Jayachandran M, Jenster G, Jiang L, Johnson SM, Jones JC, Jong A, Jovanovic-Talisman T, Jung S, Kalluri R, Kano SI, Kaur S, Kawamura Y, Keller ET, Khamari D, Khomyakova E, Khvorova A, Kierulf P, Kim KP, Kislinger T, Klingeborn M, Klinke DJ, Kornek M, Kosanović MM, Kovács ÁF, Krämer-Albers EM, Krasemann S, Krause M, Kurochkin IV, Kusuma GD, Kuypers S, Laitinen S, Langevin SM, Languino LR, Lannigan J, Lässer C, Laurent LC, Lavieu G, Lázaro-Ibáñez E, Le Lay S, Lee MS, Lee YXF, Lemos DS, Lenassi M, Leszczynska A, Li IT, Liao K, Libregts SF, Ligeti E, Lim R, Lim SK, Linē A, Linnemannstöns K, Llorente A, Lombard CA, Lorenowicz MJ, Lörincz ÁM, Lötvall J, Lovett J, Lowry MC, Loyer X, Lu Q, Lukomska B, Lunavat TR, Maas SL, Malhi H, Marcilla A, Mariani J, Mariscal J, Martens-Uzunova ES, Martin-Jaular L, Martinez MC, Martins VR, Mathieu M, Mathivanan S, Maugeri M, McGinnis LK, McVey MJ, Meckes DG, Meehan KL, Mertens I, Minciacchi VR, Möller A, Møller Jørgensen M, Morales-Kastresana A, Morhayim J, Mullier F, Muraca M, Musante L, Mussack V, Muth DC, Myburgh KH, Najrana T, Nawaz M, Nazarenko I, Nejsum P, Neri C, Neri T, Nieuwland R, Nimrichter L, Nolan JP, Nolte-'t Hoen EN, Noren Hooten N, O'Driscoll L, O'Grady T, O'Loghlen A, Ochiya T, Olivier M, Ortiz A, Ortiz LA, Osteikoetxea X, Østergaard O, Ostrowski M, Park J, Pegtel DM, Peinado H, Perut F, Pfaffl MW, Phinney DG, Pieters BC, Pink RC, Pisetsky DS, Pogge von Strandmann E, Polakovicova I, Poon IK, Powell BH, Prada I, Pulliam L, Quesenberry P, Radeghieri A, Raffai RL, Raimondo S, Rak J, Ramirez MI, Raposo G, Rayyan MS, Regev-Rudzki N, Ricklefs FL, Robbins PD, Roberts DD, Rodrigues SC, Rohde E, Rome S, Rouschop KM, Rughetti A, Russell AE, Saá P, Sahoo S, Salas-Huenuleo E, Sánchez C, Saugstad JA, Saul MJ, Schiffelers RM, Schneider R, Schøyen TH, Scott A, Shahaj E, Sharma S, Shatnyeva O, Shekari F, Shelke GV, Shetty AK, Shiba K, Siljander PR, Silva AM, Skowronek A, Snyder OL, Soares RP, Sódar BW, Soekmadji C, Sotillo J, Stahl PD, Stoorvogel W, Stott SL, Strasser EF, Swift S, Tahara H, Tewari M, Timms K, Tiwari S, Tixeira R, Tkach M, Toh WS, Tomasini R, Torrecilhas AC, Tosar JP, Toxavidis V, Urbanelli L, Vader P, van Balkom BW, van der Grein SG, Van Deun J, van Herwijnen MJ, Van Keuren-Jensen K, van Niel G, van Royen ME, van Wijnen AJ, Vasconcelos MH, Vechetti IJ, Veit TD, Vella LJ, Velot É, Verweij FJ, Vestad B, Viñas JL, Visnovitz T, Vukman KV, Wahlgren J, Watson DC, Wauben MH, Weaver A, Webber JP, Weber V, Wehman AM, Weiss DJ, Welsh JA, Wendt S, Wheelock AM, Wiener Z, Witte L, Wolfram J, Xagorari A, Xander P, Xu J, Yan X, Yáñez-Mó M, Yin H, Yuana Y, Zappulli V, Zarubova J, Žėkas V, Zhang JY, Zhao Z, Zheng L, Zheutlin AR, Zickler AM, Zimmermann P, Zivkovic AM, Zocco D, Zuba-Surma EK.. **Minimal information for studies of extracellular vesicles 2018 (MISEV2018): a position statement of the International Society for Extracellular Vesicles and update of the MISEV2014 guidelines**. *J Extracell Vesicles* (2018) **7** 1535750. PMID: 30637094
25. Akbar N, Azzimato V, Choudhury RP, Aouadi M.. **Extracellular vesicles in metabolic disease**. *Diabetologia* (2019) **62** 2179-2187. PMID: 31690986
26. Sluijter JPG, Davidson SM, Boulanger CM, Buzás EI, de Kleijn DPV, Engel FB, Giricz Z, Hausenloy DJ, Kishore R, Lecour S, Leor J, Madonna R, Perrino C, Prunier F, Sahoo S, Schiffelers RM, Schulz R, Van Laake LW, Ytrehus K, Ferdinandy P.. **Extracellular vesicles in diagnostics and therapy of the ischaemic heart: Position Paper from the Working Group on Cellular Biology of the Heart of the European Society of Cardiology**. *Cardiovasc Res* (2018) **114** 19-34. PMID: 29106545
27. Russell AE, Sneider A, Witwer KW, Bergese P, Bhattacharyya SN, Cocks A, Cocucci E, Erdbrügger U, Falcon-Perez JM, Freeman DW, Gallagher TM, Hu S, Huang Y, Jay SM, Kano S-I, Lavieu G, Leszczynska A, Llorente AM, Lu Q, Mahairaki V, Muth DC, Noren Hooten N, Ostrowski M, Prada I, Sahoo S, Schøyen TH, Sheng L, Tesch D, Van Niel G, Vandenbroucke RE, Verweij FJ, Villar AV, Wauben M, Wehman AM, Yin H, Carter DRF, Vader P.. **Biological membranes in EV biogenesis, stability, uptake, and cargo transfer: an ISEV position paper arising from the ISEV membranes and EVs workshop**. *J Extracell Vesicles* (2019) **8** 1684862. PMID: 31762963
28. Loyer X, Zlatanova I, Devue C, Yin M, Howangyin K-Y, Klaihmon P, Guerin CL, Kheloufi M, Vilar J, Zannis K, Fleischmann BK, Hwang DW, Park J, Lee H, Menasché P, Silvestre J-S, Boulanger CM.. **Intra-cardiac release of extracellular vesicles shapes inflammation following myocardial infarction**. *Circ Res* (2018) **123** 100-106. PMID: 29592957
29. Cheng M, Yang J, Zhao X, Zhang E, Zeng Q, Yu Y, Yang L, Wu B, Yi G, Mao X, Huang K, Dong N, Xie M, Limdi NA, Prabhu SD, Zhang J, Qin G.. **Circulating myocardial microRNAs from infarcted hearts are carried in exosomes and mobilise bone marrow progenitor cells**. *Nat Commun* (2019) **10** 959. PMID: 30814518
30. Ross EA, Douglas MR, Wong SH, Ross EJ, Curnow SJ, Nash GB, Rainger E, Scheel-Toellner D, Lord JM, Salmon M, Buckley CD.. **Interaction between integrin alpha9beta1 and vascular cell adhesion molecule-1 (VCAM-1) inhibits neutrophil apoptosis**. *Blood* (2006) **107** 1178-1183. PMID: 16223772
31. Jorgensen M, Baek R, Pedersen S, Sondergaard EK, Kristensen SR, Varming K.. **Extracellular Vesicle (EV) Array: microarray capturing of exosomes and other extracellular vesicles for multiplexed phenotyping**. *J Extracell Vesicles* (2013) **2**. DOI: 10.3402/jev.v2i0.20920
32. Agarwal V, Bell GW, Nam JW, Bartel DP.. **Predicting effective microRNA target sites in mammalian mRNAs**. *Elife* (2015) **4** e05005. PMID: 26267216
33. Sticht C, De La Torre C, Parveen A, Gretz N.. **miRWalk: an online resource for prediction of microRNA binding sites**. *PLoS One* (2018) **13** e0206239. PMID: 30335862
34. Liu W, Wang X.. **Prediction of functional microRNA targets by integrative modeling of microRNA binding and target expression data**. *Genome Biol* (2019) **20** 18. PMID: 30670076
35. Jassal B, Matthews L, Viteri G, Gong C, Lorente P, Fabregat A, Sidiropoulos K, Cook J, Gillespie M, Haw R, Loney F, May B, Milacic M, Rothfels K, Sevilla C, Shamovsky V, Shorser S, Varusai T, Weiser J, Wu G, Stein L, Hermjakob H, D'Eustachio P.. **The reactome pathway knowledgebase**. *Nucleic Acids Res* (2020) **48** D498-D503. PMID: 31691815
36. Vafadarnejad E, Rizzo G, Krampert L, Arampatzi P, Arias-Loza A-P, Nazzal Y, Rizakou A, Knochenhauer T, Bandi SR, Nugroho VA, Schulz DJJ, Roesch M, Alayrac P, Vilar J, Silvestre J-S, Zernecke A, Saliba A-E, Cochain C.. **Dynamics of cardiac neutrophil diversity in murine myocardial infarction**. *Circ Res* (2020) **127** e232-e249. PMID: 32811295
37. Davidson SM, Riquelme JA, Zheng Y, Vicencio JM, Lavandero S, Yellon DM.. **Endothelial cells release cardioprotective exosomes that may contribute to ischaemic preconditioning**. *Sci Rep* (2018) **8** 15885. PMID: 30367147
38. Arner E, Mejhert N, Kulyté A, Balwierz PJ, Pachkov M, Cormont M, Lorente-Cebrián S, Ehrlund A, Laurencikiene J, Hedén P, Dahlman-Wright K, Tanti J-F, Hayashizaki Y, Rydén M, Dahlman I, van Nimwegen E, Daub CO, Arner P.. **Adipose tissue microRNAs as regulators of CCL2 production in human obesity**. *Diabetes* (2012) **61** 1986-1993. PMID: 22688341
39. Agudo J, Ruzo A, Tung N, Salmon H, Leboeuf M, Hashimoto D, Becker C, Garrett-Sinha L-A, Baccarini A, Merad M, Brown BD.. **The miR-126-VEGFR2 axis controls the innate response to pathogen-associated nucleic acids**. *Nat Immunol* (2014) **15** 54-62. PMID: 24270517
40. Rusinkevich V, Huang Y, Chen Z-y, Qiang W, Wang Y-G, Shi Y-F, Yang H-T.. **Temporal dynamics of immune response following prolonged myocardial ischemia/reperfusion with and without cyclosporine A**. *Acta Pharmacol Sin* (2019) **40** 1168-1183. PMID: 30858476
41. Eash KJ, Means JM, White DW, Link DC.. **CXCR4 is a key regulator of neutrophil release from the bone marrow under basal and stress granulopoiesis conditions**. *Blood* (2009) **113** 4711-4719. PMID: 19264920
42. Suratt BT, Petty JM, Young SK, Malcolm KC, Lieber JG, Nick JA, Gonzalo J-A, Henson PM, Worthen GS.. **Role of the CXCR4/SDF-1 chemokine axis in circulating neutrophil homeostasis**. *Blood* (2004) **104** 565-571. PMID: 15054039
43. Garcia-Ramallo E, Marques T, Prats N, Beleta J, Kunkel SL, Godessart N.. **Resident cell chemokine expression serves as the major mechanism for leukocyte recruitment during local inflammation**. *J Immunol* (2002) **169** 6467-6473. PMID: 12444156
44. Dutta P, Courties G, Wei Y, Leuschner F, Gorbatov R, Robbins CS, Iwamoto Y, Thompson B, Carlson AL, Heidt T, Majmudar MD, Lasitschka F, Etzrodt M, Waterman P, Waring MT, Chicoine AT, van der Laan AM, Niessen HWM, Piek JJ, Rubin BB, Butany J, Stone JR, Katus HA, Murphy SA, Morrow DA, Sabatine MS, Vinegoni C, Moskowitz MA, Pittet MJ, Libby P, Lin CP, Swirski FK, Weissleder R, Nahrendorf M.. **Myocardial infarction accelerates atherosclerosis**. *Nature* (2012) **487** 325-329. PMID: 22763456
45. Saunderson SC, Dunn AC, Crocker PR, McLellan AD.. **CD169 mediates the capture of exosomes in spleen and lymph node**. *Blood* (2014) **123** 208-216. PMID: 24255917
46. Forrester JV, Lackie JM.. **Adhesion of neutrophil leucocytes under conditions of flow**. *J Cell Sci* (1984) **70** 93-110. PMID: 6501440
47. Taooka Y, Chen J, Yednock T, Sheppard D.. **The integrin alpha9beta1 mediates adhesion to activated endothelial cells and transendothelial neutrophil migration through interaction with vascular cell adhesion molecule-1**. *J Cell Biol* (1999) **145** 413-420. PMID: 10209034
48. Issekutz TB, Miyasaka M, Issekutz AC.. **Rat blood neutrophils express very late antigen 4 and it mediates migration to arthritic joint and dermal inflammation**. *J Exp Med* (1996) **183** 2175-2184. PMID: 8642327
49. Lőrincz ÁM, Bartos B, Szombath D, Szeifert V, Timár CI, Turiák L, Drahos L, Kittel Á, Veres DS, Kolonics F, Mócsai A, Ligeti E.. **Role of Mac-1 integrin in generation of extracellular vesicles with antibacterial capacity from neutrophilic granulocytes**. *J Extracell Vesicles* (2020) **9** 1698889. PMID: 31853340
50. Daseke MJ, Valerio FM, Kalusche WJ, Ma Y, DeLeon-Pennell KY, Lindsey ML.. **Neutrophil proteome shifts over the myocardial infarction time continuum**. *Basic Res Cardiol* (2019) **114** 37. PMID: 31418072
51. Hoshino A, Costa-Silva B, Shen T-L, Rodrigues G, Hashimoto A, Tesic Mark M, Molina H, Kohsaka S, Di Giannatale A, Ceder S, Singh S, Williams C, Soplop N, Uryu K, Pharmer L, King T, Bojmar L, Davies AE, Ararso Y, Zhang T, Zhang H, Hernandez J, Weiss JM, Dumont-Cole VD, Kramer K, Wexler LH, Narendran A, Schwartz GK, Healey JH, Sandstrom P, Labori KJ, Kure EH, Grandgenett PM, Hollingsworth MA, de Sousa M, Kaur S, Jain M, Mallya K, Batra SK, Jarnagin WR, Brady MS, Fodstad O, Muller V, Pantel K, Minn AJ, Bissell MJ, Garcia BA, Kang Y, Rajasekhar VK, Ghajar CM, Matei I, Peinado H, Bromberg J, Lyden D.. **Tumour exosome integrins determine organotropic metastasis**. *Nature* (2015) **527** 329-335. PMID: 26524530
52. Rodrigues G, Hoshino A, Kenific CM, Matei IR, Steiner L, Freitas D, Kim HS, Oxley PR, Scandariato I, Casanova-Salas I, Dai J, Badwe CR, Gril B, Tešić Mark M, Dill BD, Molina H, Zhang H, Benito-Martin A, Bojmar L, Ararso Y, Offer K, LaPlant Q, Buehring W, Wang H, Jiang X, Lu TM, Liu Y, Sabari JK, Shin SJ, Narula N, Ginter PS, Rajasekhar VK, Healey JH, Meylan E, Costa-Silva B, Wang SE, Rafii S, Altorki NK, Rudin CM, Jones DR, Steeg PS, Peinado H, Ghajar CM, Bromberg J, de Sousa M, Pisapia D, Lyden D.. **Tumour exosomal CEMIP protein promotes cancer cell colonization in brain metastasis**. *Nat Cell Biol* (2019) **21** 1403-1412. PMID: 31685984
|
---
title: 'Spotlight on the Mucormycosis Outbreak: A Deadly Fungal Infection That Followed
the COVID-19 Pandemic'
journal: Cureus
year: 2023
pmcid: PMC10022911
doi: 10.7759/cureus.35095
license: CC BY 3.0
---
# Spotlight on the Mucormycosis Outbreak: A Deadly Fungal Infection That Followed the COVID-19 Pandemic
## Abstract
Background: The COVID-19 pandemic along with its treatment has brought myriad potential complications including the heightened risk of secondary fungal infections like mucormycosis. Mucormycosis is a rare angioinvasive fungal infection that has traditionally been highly fatal despite surgical intervention and antifungal medications.
Aim: To re-evaluate the risk factors, epidemiology, and possible COVID-19-associated conditions on a larger sample size than the existing data.
Methodology: We studied the possible risk factors, clinical presentations, treatment, and outcome of 203 patients with mucormycosis in a single-center retrospective-prospective observational study for three months at a tertiary care hospital after obtaining due permission from the institutional ethics committee.
Results: The mean age of patients was 52 ± 11.5 years, and $92.61\%$ had a history of COVID-19 infection. Around $86.7\%$ of patients were suffering from diabetes mellitus with $50\%$ being already known cases whereas the other $50\%$ developed post-COVID-19 infection; $65.02\%$ of patients were administered corticosteroids during their COVID-19 treatment. About $51.72\%$ of patients required hospital admission and among them, $16.25\%$ of patients required ICU support. The mean oxygen saturation (SpO2) levels on admission were 84.61 ± $12.96\%$, and $38.92\%$ of patients required mechanical respiratory support. The mean duration between COVID-19 infection and the onset of mucormycosis was 18.80 ± 16.61 days. The most common clinical presentations were facial pain and swelling ($26.6\%$) and ophthalmic symptoms including eye swelling, pain, and ptosis ($25.12\%$). Antifungal treatment was given to all the patients and $89.36\%$ of the patients underwent surgical debridement of fungal mass. At the end of three months, $60.59\%$ of the 188 patients survived with improvement, $13.30\%$ had no improvement and/or deterioration of health, and $18.72\%$ succumbed to mucormycosis. Intracranial involvement and leukocytosis were positively associated with mortality whereas surgical intervention was significant for positive outcomes at the end of three months in patients with mucormycosis ($p \leq 0.05$).
Conclusion: The sudden rise of mucormycosis during the second wave of COVID-19 can be attributed to uncontrolled blood sugar levels along with high corticosteroid usage as well as various nosocomial factors during the COVID-19 treatment. Early and aggressive treatment with surgical intervention and antifungal drugs can improve disease outcomes.
## Introduction
The COVID-19 or severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was declared a global pandemic by the World Health Organisation (WHO) in March 2020 [1]. The pandemic continued to be an ongoing public health concern with more than 275 million cases recorded and more than 5 million deaths globally. In India, more than 34 million cases are reported to date [2]. For some people, opportunistic fungal infections by mucormycosis, candidiasis, and aspergillosis proved to be more morbid than COVID-19 itself [1].
Mucormycosis is an angio-invasive fungal infection caused by the ubiquitous filamentous fungi of the class of Mucoromycetes resulting in vessel thrombosis and tissue necrosis [3]. Mucorales are thermotolerant organisms and are known to infect via ingestion of contaminated food, inhalation of spores, or inoculation into disrupted skin or wounds leading to localized and disseminated infections, especially in an immunocompromised patient [4]. The spectrum of mucormycosis includes rhino-orbital-cerebral, pulmonary, cutaneous, gastrointestinal, and disseminated forms; the most common being rhino-cerebral mucormycosis (ROCM) [5].
The estimated prevalence of mucormycosis in India before COVID-19 was as low as 0.02 to 9.5 cases per 100,000 population but the country witnessed an alarming surge in the number of mucormycosis cases during the COVID-19 pandemic [6]. Spellberg et al. proposed that the viral infection leads to an overexpression of inflammatory cytokines and impaired cell-mediated immunity (cluster of differentiation (CD)4 and CD8 T-cells) which increases the susceptibility to opportunistic fungal infections. Major risk factors for mucormycosis include uncontrolled diabetes mellitus (DM), diabetic ketoacidosis (DKA), metabolic acidosis, malignant hematologic disorders, and deferoxamine therapy in patients receiving hemodialysis [7]. Additionally, factors such as corticosteroids and humidifier use during oxygen therapy in COVID-19 management contributed to the process [8].
Despite aggressive treatment protocols involving surgical intervention and administration of antifungal drugs (amphotericin B, posaconazole, and isavuconazole) mortality rate is increasing in patients of mucormycosis with COVID-19 infection. So, in this study, we aim to assess the risk factors in patients suffering from invasive fungal infection, especially mucormycosis, at a tertiary care teaching hospital.
## Materials and methods
Study design *This is* a retrospective-prospective observational, single-center study conducted at the mucormycosis ward of the B.J. Medical College & Civil Hospital, Ahmedabad, Gujarat, India. Due ethics approval was obtained from the Institutional Ethics Committee of the B.J. Medical College & Civil Hospital (approval no. EC/Approval/$\frac{67}{2021}$/$\frac{12}{06}$/2021) before starting the study. All patients admitted in this mucormycosis ward in the 15 days of data collection and who met the undermentioned inclusion criteria were included in the study and were followed up for three months.
The inclusion criteria for the study are as follows: patients above the age of 18 years and willing to give written informed consent; patients whose mucormycosis was defined by clinical-radiological diagnosis and confirmed by positive biopsy results before admission to the ward. All the patients who were determined to have COVID-19 are those who tested positive in RT-PCR testing of the throat and nasopharyngeal swabs. Patients who were not willing to give written consent were excluded. The included patients were followed up telephonically at the end of three months for updates on any further surgery, medications, and the clinical outcome. Written informed consent was taken either from the patient or nearest of kin.
The investigator visited the mucormycosis ward at the Civil Hospital and enrolled the patients as per inclusion criteria and recorded the data in case record forms (CRF). Patient details were recorded during the hospital stay and follow-up details were obtained at the end of three months, telephonically. A total of 203 patients were enrolled in the study.
At the first visit, the following baseline data were collected: demographic details, personal history, past history, COVID-19-related treatment, and clinical history. And during hospitalization, details of ICU stay, oxygen requirement, existing co-morbid conditions, hematological investigations, and treatment data were recorded. At the end of three months, the outcome of mucormycosis was recorded.
Data were entered in Microsoft Excel 2019 (Microsoft Corp., Redmond, WA, USA) worksheet and was analyzed in Excel 2019 itself for various risk factors associated with mucormycosis. Clinical outcome in the patients was analyzed using descriptive statistics. Descriptive statistics were presented as mean and standard deviation (SD) for quantitative variables, and as frequencies with percentages for qualitative data. The chi-square test was used to check the association of risk factors and outcomes in patients with mucormycosis using GraphPad Prism version 9.3.0. ( GraphPad Software Inc., San Diego, CA, USA). A p-value <0.05 was considered statistically significant.
## Results
Demographic details Out of the 203 patients enrolled, 146 ($71.92\%$) were males and 57 ($28.07\%$) were females with a mean age of 52 ± 11.50 years. Around $63.5\%$ of patients belonged to the 40 to 59 years age group. The mean BMI of the patients was 24.43 ± 4.21 kg/m2; 116 patients were healthy, and 49 patients were overweight. Around $37.93\%$ of patients were farmers by occupation followed by housewives at $21.67\%$. Ninety ($44.33\%$) patients had no addiction history, and 113 ($55.67\%$) patients were found to be addicted to tobacco/alcohol/smoking.
Risk factors Co-morbidities Out of 203 patients, 185 ($91.13\%$) were COVID-19-positive and 18 ($8.86\%$) were COVID-19-negative with a mean COVID-19 duration of 14.43 ±5.87 days. The most common comorbidity observed was DM in 176 ($86.7\%$) patients; out of which 88 patients had a history of DM pre-COVID-19 and 88 patients developed DM after the onset of COVID-19. Apart from that, 64 ($31.5\%$) patients had hypertension, 26 ($12.8\%$) patients had DKA, 19 ($9.35\%$) patients had cardiovascular disorders, and 12 ($5.91\%$) patients had respiratory disorders. Other co-morbidities observed were hypothyroidism at $3.44\%$, history of (H/O) tuberculosis at $2.95\%$, H/O fungal infection at $2.46\%$, cancer at $1\%$, and H/O transplantation at $0.5\%$.
COVID-19 and Mucormycosis Out of 203 patients, 162 ($73.80\%$) were non-vaccinated and 41 ($20.19\%$) were vaccinated against COVID-19. Mucormycosis was diagnosed in 74 ($40\%$) patients who had active COVID-19 infection, and 111 ($60\%$) patients were diagnosed with mucormycosis post-COVID-19 infection. Based on the high-resolution computed tomography (HRCT) score, 25 ($15.82\%$) patients had mild, 118 ($74.68\%$) patients had moderate, and 15 ($9.49\%$) patients had severe COVID-19 infection. Out of the 185 patients diagnosed with COVID-19, 105 ($56.75\%$) were hospitalized with a mean duration of hospitalization of 11.35 ± 8.22 days. The medications administered during the COVID-19 treatment are in Table 1.
**Table 1**
| Medication (n=203) | Medication (n=203).1 | Medication (n=203).2 | Patients administered with drugs |
| --- | --- | --- | --- |
| Corticosteroids ( IV or Oral ) | Corticosteroids ( IV or Oral ) | Corticosteroids ( IV or Oral ) | 65.02% (132) |
| Duration of steroids | <5 Days | <5 Days | 3.44% (7) |
| Duration of steroids | 5-10 days | 5-10 days | 34.97% (71) |
| Duration of steroids | >10 days | >10 days | 27.09% (55) |
| Remdesivir | Remdesivir | Remdesivir | 33.0% (67) |
| Tocilizumab | Tocilizumab | Tocilizumab | 7.88% (16) |
| Favipiravir | Favipiravir | Favipiravir | 56.15% (114) |
| Zinc supplements | Zinc supplements | Zinc supplements | 81.2% (165) |
| Iron supplements | Iron supplements | Iron supplements | 18.22% (37) |
| Anticoagulants | Anticoagulants | Anticoagulants | 1.08% (124) |
| Anti-microbials (n=203) | Anti-microbials (n=203) | Azithromycin | 83.25% (169) |
| Anti-microbials (n=203) | Anti-microbials (n=203) | Doxycycline | 12.80% (26) |
| Anti-microbials (n=203) | Anti-microbials (n=203) | Ivermectin | 29.55% (60) |
The mean SpO2 levels on admission ($$n = 157$$) were 84.61 ± 12.96 (Spo2: $40.12\%$ of patients had >90; $59.87\%$ of patients had <=90), and 80 ($43.24\%$) patients were given home treatment. Intensive care unit treatment was required in 33 ($31.43\%$) hospitalized patients and the mean ICU stay duration was 6.9 ± 6.17 days. Around 79 ($75.2\%$) patients required respiratory support ($12.6\%$ of patients required mechanical ventilation), and 26 ($24.8\%$) patients did not need respiratory support. The mean duration of mechanical ventilation was 8.7 ± 8 days. Forty-three patients used only a single mask (20 nasal cannulas, 18 simple oxygen masks, and five non-rebreather masks (NRBM)) for the whole duration of treatment whereas 36 patients used multiple masks for ventilation. The mean duration from the onset of COVID-19 to mucor symptom presentation was 18.80 ± 16.61 days. The mean time between mucormycosis symptoms appearance and treatment was 10.86 ± 10.47 days.
Clinical presentation The most common clinical presentations were facial pain and swelling ($26.6\%$) and ophthalmic symptoms including eye swelling, pain, and ptosis ($25.12\%$). Dental pain and moving teeth ($22.6\%$), headache ($13.79\%$), and nasal symptoms ($4.43\%$) were also observed. Other symptoms were paralysis, fever, weakness, and unconsciousness ($7.38\%$). The presentations are demonstrated in Figure 1.
**Figure 1:** *Clinical presentation of mucormycosis*
Medical management of mucormycosis All the patients in our study were administered intravenous amphotericin. Patients were started on amphotericin B or liposomal form in some cases as soon as admission and were shifted to either liposomal or lipid complex preparations in case of hazardous effects on the other systems. The patients after discharge were prescribed oral posaconazole tablets as continuation therapy. Table 2 states the drugs administered to patients and the doses of amphotericin administered. Two patients each opted for ayurvedic and homeopathic therapies.
**Table 2**
| Mucormycosis medication | Value |
| --- | --- |
| Conventional amphotericin B | 82.27% (167) |
| Mean doses administered | 23.12 ± 10.27 |
| Liposomal amphotericin | 38.92% (79) |
| Lipid complex amphotericin | 6.89% (14) |
| Posaconazole therapy | 76.84% (156) |
| Mean duration of posaconazole (days) | 34.19 ± 21.26 |
Surgical intervention Out of 188 patients, 168 ($89.36\%$) patients underwent surgical removal of the fungal mass whereas 20 patients ($10.63\%$) did not. Sixty-three patients had had revision surgeries (more than one) while 105 had a one-time surgery (total number of surgeries=243). The most common surgery was functional endoscopic sinus surgery (FESS)/debridement in $60.49\%$ of patients followed by dental procedures in $11.93\%$ of patients followed by orbital exenteration in $11.11\%$ of patients. Surgical intervention was found to have a positive impact on the outcome at the end of the follow-up period (p-value=0.034). Figure 2 demonstrates the surgical interventions done during the treatment of mucormycosis.
**Figure 2:** *Surgical intervention in patients with mucormycosisFESS: Functional endoscopic sinus surgery*
Outcomes of mucormycosis At the end of three months, 150 patients survived (65 recovered, 58 were recovering, 20 had persistent symptoms, and seven had progressive symptoms), and 38 patients died. Fifteen patients were lost to follow-up. Figure 3 represents the comparison between the clinical outcome at the end of the follow-up period.
**Figure 3:** *Outcomes of mucormycosis at the end of three months*
Table 3 depicts the statistical significance of various factors in the clinical outcome at the end of the follow-up period and is compared with the statistical significance of those factors determined as risk factors of mucormycosis by a contemporary study [9].
**Table 3**
| Risk factors (n=188) | Survivors | Non-Survivors | p-value | p-value for the occurrence of mucormycosis ** |
| --- | --- | --- | --- | --- |
| Diabetes mellitus (DM) | 129 (79.14) | 34 (20.85) | 0.86 | 0.001 |
| Known cases | 64 | 15 | 0.74 | 0.035 |
| Newly diagnosed DM | 65 | 19 | 0.56 | 0.04 |
| Diabetic ketoacidosis | 17 (70.8) | 7 (29.2) | 0.63 | |
| Hypertension | 47 (80) | 12 (20) | 0.85 | 0.74 |
| Cardiovascular disorders | 12 (75) | 4 (25) | 0.45 | |
| Respiratory disorders | 10 (83) | 2 (17) | 0.71 | |
| Tuberculosis | 4 (80) | 1 (20) | 1 | |
| Hypothyroidism | 5 (83.3) | 1 (16.7) | 0.71 | |
| History of fungal infection | 3 (75) | 1 (25) | 0.49 | |
| History of acute conditions in the past 3 months | 44 (85) | 8 (15) | 0.27 | |
| COVID-19 vaccination (non-vaccinated) | 121 (81.2) | 28 (18.8) | 0.30 | |
| COVID-19 Positive | 138 (74.59) | 35 (18.91) | 0.45 | |
| Mucor diagnosis during active COVID-19 (n=185) | 54 (77.1) | 16 (22.9) | 0.48 | |
| Spo2 (n=157) =<90 | 74 (82.2) | 16 (17.8) | 0.56 | |
| HRCT score | 99(79.2) | 26(20.8) | 0.86 | |
| Intracranial mucor involvement | 36 (64.2) | 20 (35.8) | 0.0006* | |
| ICU admission | 27 (84.3) | 5 (15.7) | 0.46 | |
| Neutrophil/lymphocyte ratio High (>3.53) | 65 (73) | 24 (27) | 0.08 | 0.93 |
| Leukocytosis (>10000/cu.mm) [Approximately equal to the average = 10500] | 61 (70.1) | 26 (29.9) | 0.01* | 0.56 |
| Eosinophils % (> average eosinophils = 2.83%) | 18 | 88 | 0.20 | <0.0001 |
| Steroid use | 98 (79.6) | 25 (20.4) | 1 | 0.2 |
| Steroid duration (More than the guideline) | 21 (77.7) | 6 (2) | 0.73 | 0.001 |
| Remdesivir | 51 | 13 | 0.98 | 0.004 |
| Tocilizumab use | 12 (75) | 4 (25) | 0.49 | 0.55 |
| Zinc | 121 (79) | 32 (21) | 0.58 | |
| Iron | 29 (82.8) | 6 (17.2) | 0.58 | |
| Anticoagulants | 90 | 25 | 0.51 | <0.001 |
| Surgical intervention | 141 | 27 | 0.034* | |
## Discussion
Table 4 contains the comparison of similar parameters in our study with other contemporary studies by Mishra et al. and Selarka et al. [ 5,1].
**Table 4**
| Parameter | Parameter.1 | Parameter.2 | Our study (n=203) | Mishra et al. [5] (n=32) | Selarka et al. [1] (n=47) |
| --- | --- | --- | --- | --- | --- |
| Age (mean Age) | Age (mean Age) | Age (mean Age) | 52 ± 11.50 | 58.28 ± 8.57 | 55 ± 12.80 |
| Diabetes mellitus | Diabetes mellitus | Diabetes mellitus | 86.7% | 87.5% | 76.6% |
| Hypertension | Hypertension | Hypertension | 31.5% | 50.0% | 57.4% |
| Covid severity | Covid severity | Severe | 9.49% | 56.2% | 10.6% |
| Covid severity | Covid severity | Non-severe | 91.51% | 43.8% | 89.6% |
| Corticosteroid usage | Corticosteroid usage | Corticosteroid usage | 65.02% | 93.7% | 95.7% |
| Remdicivir usage | Remdicivir usage | Remdicivir usage | 33.0% | - | 74.4% |
| Features | Headache | Headache | 13.79% | 93.8% | 74.5% |
| Features | Orbital symptoms | Orbital symptoms | 25.12% | 59.4% | 40.4% |
| Features | Facial symptoms | Facial symptoms | 26.60% | - | 34.0% |
| Features | Nasal symptoms | Nasal symptoms | 4.43% | 62.5% | - |
| Surgical intervention | Surgical intervention | Surgical intervention | 89.36% | 93.8% | 80.9% |
| Total leukocyte count | Total leukocyte count | Total leukocyte count | 10104 ± 4413 | 11478 ± 3693 | - |
| Timelag (COVID-19 and mucormycosis) | Timelag (COVID-19 and mucormycosis) | Timelag (COVID-19 and mucormycosis) | 18.80 ± 16.61 | 17.28 ± 11.36 | |
| Mechanical ventilation | Mechanical ventilation | Mechanical ventilation | 38.91% | - | 80.85% |
Mucormycosis is a rare and potentially fatal invasive fungal infection with an increased risk of occurrence in immunocompromised patients and thus COVID-19 patients are at a higher risk of getting infected [1]. The fungal elements invade the blood vessels leading to pro-coagulative complications like thrombosis, ischaemic infarction, and ultimately necrosis of the host tissue [10]. India has the highest prevalence of mucormycosis around the world, accounting for about 70 times more than the global data [6,11]. A recent study of coronavirus disease-associated mucormycosis in India reported the incidence of mucormycosis to be $0.27\%$ among hospitalized patients and that the caseload had increased about 2.1-fold as compared to the previous year [12]. Mucorales spores invade the body through either inhalation (most common) or ingestion through the mouth. The inhaled spores reach and damage the paranasal sinuses causing necrosis of the tissue structures around and subsequently spreading to the surrounding areas causing local symptoms of invasion and necrosis in case of non-treatment or high fungal load [7]. The majority of the patients in our study were farmers ($37.93\%$) by occupation creating a chance of association with the high exposure of disseminated Mucorales spores in agricultural areas and thus being more susceptible to acquiring the invasive fungal infection.
Diabetes mellitus is the most common risk factor for the development of mucormycosis and is being reported in more than $50\%$ of mucormycosis cases in India [13]. Existing history of diabetes and hyperglycemia was present in $43.34\%$ of our subjects and another $43.43\%$ reported a history of hyperglycemia post the COVID-19 treatment. Post-COVID-19 hyperglycemia was suspected to be associated with the administration of corticosteroids (oral and intravenous) during the treatment of COVID-19. Out of $65.02\%$ of patients who were administered corticosteroids, $16.67\%$ of patients received steroids more than the recommended dose for COVID-19 treatment [14]. Unmonitored use of corticosteroids in COVID-19 treatment and DM leads to an increased susceptibility to contracting mucormycosis due to an interplay of factors causing immunosuppression and hyperglycemia through (I) defective phagocytic mechanism; (II) upregulation of GRP78 receptors in humans and CotH (Mucorales specific protein) in the fungi (entry into human endothelial cells by fungus is facilitated by the interaction of GRP78 and CotH); (III) hyperglycation of iron sequestering proteins causing increased iron delivery to the Mucorales that is essential for their growth [1,15,16].
Hypertension is not an established risk factor for the occurrence and the outcome of mucormycosis but this needs to be further evaluated as there was a significant proportion of subjects with hypertensive history in both our study as well as contemporary studies. Diabetic ketoacidosis is an established risk factor for the occurrence of mucormycosis. The presence of a special ketone reductase system in fungi leads to increased glucose utilization in an acidic medium which boosts fungal growth along with a compromised phagocytic mechanism. In our study, $12.80\%$ of patients had a DKA episode which was higher than in other similar studies where no such cases were reported [5]. History of acute medical conditions (within the past three months before mucormycosis) was present in $26.10\%$ of our patients and could be further investigated as possible risk factors for the development of mucormycosis. These included dental procedures ($33.96\%$), wounds and mucosal exposures ($30.18\%$), upper respiratory tract infections ($22.64\%$), and ophthalmologic procedures ($16.98\%$). The potential modes of transmission of the spores of Mucorales also coincide with the occurrence of these acute medical histories further strengthening the possibility of an association.
The established incubation period of mucormycosis is around seven to 10 days after exposure to the fungus whereas the mean duration between COVID-19 infection and the appearance of mucormycosis symptoms in our study was 18.80 ± 16.61 days. Thus, increasing the possibility of contracting a mucor infection during COVID-19 treatment and hospitalization [17]. Around $31.43\%$ of patients required ICU admission with a mean duration of 6.9 ± 6.17 days where the ICU is already a hotspot for acquiring a multitude of nosocomial and opportunistic infections. Mechanical ventilation was required by 79 ($75.2\%$) out of 105 patients hospitalized for COVID-19 pneumonia for a mean duration of 9.34 ± 8.80 days. Out of 79 patients, 43 patients used only a single mask (20 nasal cannulas, 18 simple oxygen masks, and five NRBM) for the whole duration of treatment whereas the rest of the 36 patients used multiple masks for ventilation. Non-adherence to regular cleaning and replacement of mask may lead to the growth of multiple microorganisms including fungi predisposing the patients to develop mucormycosis. External oxygen in mechanical ventilation is passed through a humidifier which increases the moisture content in the mask, tubing as well as the pharyngeal area of the patient making the conditions favorable for fungal growth. In cases of normal undistilled water usage or non-replacement of the water regularly in the humidifier, a lot of minerals present in tap water along with the moisture play a perfect breeding ground for fungal growth. The oxygen for 60 out of 79 ($75.95\%$) patients was supplied through cylinders whereas only 19 ($24.05\%$) received oxygen produced from in-house oxygen plants. The second wave of COVID-19 was a time when there was a severe shortage of all medical facilities and equipment including oxygen cylinders, so various industrial oxygen suppliers were brought in to fulfill the medical needs. Industrial cylinders are not sterilized before re-filling, and they may have been a source of nutrients and minerals responsible for fungal growth. However, due to the limitations of our study, we could not test the cylinders used but the relation between oxygen delivery systems and the growth of micro-organisms needs to be further studied.
Tocilizumab, an interleukin (IL)-6 inhibitor, is used to treat cytokine storms in COVID-19 patients and was administered in $7.88\%$ of subjects in our study. Tocilizumab acts as an immunomodulator by interleukin suppression and thus may contribute to the occurrence of mucormycosis by downregulation of the natural immune system. Around $81.28\%$ of patients took supplemental zinc during their COVID-19 treatment, which has been demonstrated to increase the fungal growth in vitro due to more efficient substrate utilization and can be one of the factors in the rapid rise of mucormycosis cases among patients being treated for COVID-19 [18]. Around $18.22\%$ of patients had taken supplemental iron therapy which enhances the expression of GRP78 receptors, facilitating fungal entry in cells [16].
The high leucocyte count in $42.70\%$ of patients was associated with unfavorable outcomes at the end of three months and proved to be statistically significant (p-value 0.01). The neutrophil-lymphocyte ratio in our study was high (>3.53) in $49.18\%$ of patients suggestive of relative lymphocytopenia. Lymphocytopenia has been associated with COVID-19 and its severity but its correlation with the occurrence of mucormycosis and outcome needs to be further evaluated [19]. Rhino-cerebral mucormycosis had shown a significant association with a three-month survival rate ($$p \leq 0.0006$$) in patients with mucormycosis.
A delay in diagnosis is an important determinant of the outcome of treatment and even a delay of six days has been seen to almost double the mortality rate [20]. In our study, the mean delay or lag in receiving treatment was 10.86 ± 10.47 days. The lag period in accessing treatment has not been compared with the outcome in this study as the clinical outcome depends on a multitude of factors rather than just delayed diagnosis. The treatment modality of mucormycosis consists of surgical removal of the fungal mass along with antifungal medications like amphotericin B, and posaconazole. In our study, 168 patients underwent surgical removal whereas 20 patients did not. Surgical intervention was a statistically significant parameter positively affecting the outcome at the end of 12 weeks suggesting that aggressive removal or debridement of the invasive fungal mass is a potent step to acutely reduce the progression and in turn obtain a better prognosis.
At the end of 12 weeks, 150 patients survived ($32.01\%$ recovered, $28.57\%$ were recovering, $9.85\%$ had persistent symptoms, and $3.44\%$ had progressive symptoms) whereas $18.71\%$ of patients succumbed, and $7.38\%$ of patients were lost to follow-up. Traditionally the survival rates for mucormycosis are low ($56\%$ in previous studies), whereas in our study $81\%$ of the patients survived at the end of three months which is attributed to aggressive anti-fungal treatment and early surgical management of patients [21]. The results of our study can be generalized in the aspect of treatment of mucormycosis as we had a good enough sample size from a single center for such a rare disease.
Limitations Despite having a larger sample size for a single-center study than any of those previously conducted, we did not find significant patients for rare risk factors like transplant, malignancy, etc. Also, being an observational study, we did not have controls that were COVID-19-positive but did not develop mucormycosis to analyze the specific factors that led to the development of the fungal infections. However, to the best of our knowledge and considering the rare occurrence of the disease, this is the largest single-center case series studied.
## Conclusions
The sudden rise of mucormycosis during the second wave of COVID-19 can be attributed to uncontrolled blood sugar levels along with irrational corticosteroid usage combined with various nosocomial factors such as mechanical ventilation and the oxygen used, drug therapy, and ICU admissions during COVID-19 treatment. The role of COVID-19 infection by itself in the causation of invasive fungal infection needs to be evaluated further. Early and aggressive treatment with surgical intervention and antifungal drugs can be pivotal in improving disease outcomes in patients with mucormycosis.
## References
1. Selarka L, Sharma S, Saini D. **Mucormycosis and COVID-19: an epidemic within a pandemic in India**. *Mycoses* (2021) **64** 1253-1260. PMID: 34255907
2. **India: WHO Coronavirus Disease (COVID-19) Dashboard With Vaccination Data | WHO Coronavirus (COVID-19) Dashboard With Vaccination Data**. (2022)
3. Bala K, Chander J, Handa U, Punia RS, Attri AK. **A prospective study of mucormycosis in north India: experience from a tertiary care hospital**. *Med Mycol* (2015) **53** 248-257. PMID: 25587084
4. Reid G, Lynch JP 3rd, Fishbein MC, Clark NM. **Mucormycosis**. *Semin Respir Crit Care Med* (2020) **41** 99-114. PMID: 32000287
5. Mishra Y, Prashar M, Sharma D, Akash Akash, Kumar VP, Tilak TV. **Diabetes, COVID 19 and mucormycosis: clinical spectrum and outcome in a tertiary care medical center in Western India**. *Diabetes Metab Syndr* (2021) **15** 102196. PMID: 34246939
6. Prakash H, Chakrabarti A. **Global epidemiology of mucormycosis**. *J Fungi (Basel)* (2019) **5** 26. PMID: 30901907
7. Spellberg B, Edwards J Jr, Ibrahim A. **Novel perspectives on mucormycosis: pathophysiology, presentation, and management**. *Clin Microbiol Rev* (2005) **18** 556-569. PMID: 16020690
8. Yang X, Yu Y, Xu J. **Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study**. *Lancet Respir Med* (2020) **8** 475-481. PMID: 32105632
9. Gupta D, Kulkarni R, Pujari S. **Covid-19 associated mucormycosis: a case-control study**. *MedRxiv* (2021)
10. Afroze SN, Korlepara R, Rao GV, Madala J. **Mucormycosis in a diabetic patient: a case report with an insight into its pathophysiology**. *Contemp Clin Dent* (2017) **8** 662-666. PMID: 29326525
11. Chander J, Kaur M, Singla N. **Mucormycosis: battle with the deadly enemy over a five-year period in India**. *J Fungi (Basel)* (2018) **4**
12. Patel A, Agarwal R, Rudramurthy SM. **Multicenter epidemiologic study of coronavirus disease-associated mucormycosis, India**. *Emerg Infect Dis* (2021) **27** 2349-2359. PMID: 34087089
13. Prakash H, Ghosh AK, Rudramurthy SM. **A prospective multicenter study on mucormycosis in India: Epidemiology, diagnosis, and treatment**. *Med Mycol* (2019) **57** 395-402. PMID: 30085158
14. **Clinical management protocol for covid-19 (In Adults) Ministry of Health and Family Welfare, Government of India**. (2022)
15. Ibrahim AS, Spellberg B, Walsh TJ, Kontoyiannis DP. **Pathogenesis of mucormycosis**. *Clin Infect Dis* (2012) **54 Suppl 1** 0
16. Alqarihi A, Gebremariam T, Gu Y. **GRP78 and integrins play different roles in host cell invasion during mucormycosis**. *mBio* (2020) **11**
17. Smith RM, Lee J, Mody RK. **Determining the incubation time of mucormycosis: a systematic review**. *Open Forum Infect Dis* (20151) **2** 1
18. Wegener WS, Romano AH. **Zinc stimulation of RNA and protein synthesis in rhizopus nigricans**. *Science* (1963) **142** 1669-1670. PMID: 14075699
19. Guan WJ, Ni ZY, Hu Y. **Clinical characteristics of coronavirus disease 2019 in China**. *N Engl J Med* (2020) **382** 1708-1720. PMID: 32109013
20. Chamilos G, Lewis RE, Kontoyiannis DP. **Delaying amphotericin B-based frontline therapy significantly increases mortality among patients with hematologic malignancy who have zygomycosis**. *Clin Infect Dis* (2008) **47** 503-509. PMID: 18611163
21. Jeong W, Keighley C, Wolfe R, Lee WL, Slavin MA, Kong DC, Chen SC. **The epidemiology and clinical manifestations of mucormycosis: a systematic review and meta-analysis of case reports**. *Clin Microbiol Infect* (2019) **25** 26-34. PMID: 30036666
|
---
title: 'Association Between Lipid Profile Measurements and Mortality Outcomes Among
Older Adults in a Primary Care Setting: A Retrospective Cohort Study'
journal: Cureus
year: 2023
pmcid: PMC10022913
doi: 10.7759/cureus.35087
license: CC BY 3.0
---
# Association Between Lipid Profile Measurements and Mortality Outcomes Among Older Adults in a Primary Care Setting: A Retrospective Cohort Study
## Abstract
Background Lipid profile components play a role in predicting the development of cardiovascular disease and hence mortality, but recent studies have shown mixed results in the older population. The aim of our study was to investigate the association between levels of lipid profile components with all-cause mortality and cardiovascular outcomes among older adults in a primary care setting in Riyadh, Saudi Arabia.
Methods A retrospective cohort study was performed among 485 individuals aged 60 years and older who visited the family medicine clinics linked to a tertiary care hospital during the first six months of 2010. The electronic charts of the participants were reviewed up to April 2022 to gather relevant data. Each lipid profile component, including total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TGs), was categorized into four quartiles. LDL was calculated using the Friedewald formula. Cardiovascular outcomes included ischemic heart disease (IHD), heart failure (HF), and stroke.
Results The mean follow-up period was 12 years. The elderly participants with the lowest HDL-C quartile (<1.1 mmol/L) were at higher risk of all-cause mortality (adjusted hazard ratio of 2.023 ($95\%$ CI 1.21-3.38)) and IHD (adjusted hazard ratio 3.2 ($95\%$ CI 1.6-6.2)). High TC (≥5.7 mmol/L) was associated with an increased risk of HF (adjusted hazard ratio 2.1 ($95\%$ CI 1.1-4.0)).
Conclusion In patients aged 60 years and older, low HDL-C (<1.1 mmol/L) was associated with a higher risk for all-cause mortality and IHD, and high TC was associated with an increased risk of having HF. No significant association was found for LDL-C, TC, and TGs with all-cause mortality.
## Introduction
The impact of elevated total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C) on cardiovascular and all-cause mortality in the general population is well established [1,2]. However, this association seems to be more complex in older individuals as multiple studies have shown that elevated TC is not associated or even inversely associated with mortality in the elderly [3-5]. Similarly, with LDL-C, the results are inconsistent, with some studies showing an inverse association with all-cause mortality [6,7].
High-density lipoprotein cholesterol (HDL-C) has a well-known role as the “good” cholesterol in protecting against cardiovascular and all-cause mortality, but in the older population, there are different results, with studies showing that only the lowest levels are associated with the highest mortality [8,9], while other studies show a U-shaped relationship with the lowest and highest quartiles corresponding with the highest mortality [10,11]. For triglycerides (TGs), a meta-analysis study showed that elevated levels were associated with a higher risk for cardiovascular and all-cause mortality [12], although some studies showed that the relationship weakens with older age [13,14].
Cardiovascular disease (CVD) is the most common cause of death globally according to the World Health Organization [15], and one of the most important risk factors for CVD is dyslipidemia [16]. Studies involving the older population found that low HDL-C was significantly associated with an increased risk for ischemic heart disease (IHD) and stroke [9,17]. For LDL-C, some studies show a higher risk for CVD with high levels [18,19], while other studies show no association [20,21]. Moreover, a study reviewing the association between high TC and CVD in the elderly found a strong association with IHD [22]. For TGs, only a few studies have targeted the elderly population and found that higher levels are associated with increased CVD risk [23,24]. No regional or local studies are available in the literature on the association between lipid profile measurements and mortality in the older population. This study investigates the association of plasma lipid profile concentrations with all-cause mortality and cardiovascular outcomes among older adults (60 years of age and older).
## Materials and methods
We performed a retrospective cohort study using electronic health records from King Faisal Specialist Hospital and Research Centre (KFSH&RC). All patients aged 60 years and older who presented to the family medicine clinic in KFSH&RC during the period from January 1, 2010, to June 30, 2010, and underwent lipid profile blood tests were enrolled. We excluded patients who died within one year after the blood test or those who were lost to follow-up. Demographic data collected included age and gender.
Lipid profile levels The lipid profile included TC, LDL-C, HDL-C, and TGs. If a participant had more than one lipid profile reading during the study period, only the first reading was collected. The LDL-C was calculated using the Friedewald formula.
Lipid profile components were categorized into four groups: LDL-C: <2.6, 2.6 to 3.09, 3.1 to 3.59 and ≥3.6 mmol/L.
TC: <4.5, 4.5 to 5.09, 5.1 to 5.69 and ≥5.7 mmol/L.
HDL-C: <1.1, 1.1 to 1.39, 1.4 to 1.59 and ≥1.6 mmol/L.
TG: <0.9, 0.9 to 1.2, 1.21 to 1.69 and ≥1.7 mmol/L.
Assessment of confounders The confounders considered included body mass index (BMI), statin use, diabetes mellitus (DM), and hypertension (HTN). BMI was categorized as underweight for BMI < 18.5 kg/m2, normal weight for BMI of 18.5-24.9 kg/m2, overweight for BMI of 25-29.9 kg/m2, and obese for BMI > 30 kg/m2). The use of statins was determined electronically by reviewing patient prescriptions. The presence of DM was based on an HbA1C level of ≥$6.5\%$. The presence of HTN was based on multiple charted blood pressure readings of >$\frac{140}{90}$ mmHg.
Endpoints Death and cardiovascular outcomes were determined electronically using the patient’s electronic health records in April 2022. The cardiovascular outcomes collected included IHD, heart failure (HF), and stroke.
Statistical analysis *Statistical analysis* of the data was done using IBM SPSS Statistics for Windows, Version 20 (Released 2011; IBM Corp., Armonk, New York, United States). Descriptive statistics were reported as the mean and standard deviation for continuous variables, and the categorical variables were summarized as frequencies and percentages. Continuous variables were compared using an independent t-test, while categorical variables were compared using the chi-squared and Fisher’s exact tests. Cox regression was used to estimate the hazard ratio after adjusting for the other variables.
## Results
A total of 670 participants were reviewed. Of those, 485 participants met our inclusion criteria, the mean age was 69.2 ± 6.3 years, and $55.7\%$ of them were female. During the study period, a total of 115 ($23.7\%$) individuals died. The mean follow-up period was 12 years.
Table 1 summarizes the baseline characteristics of the participants based on mortality status. There was no significant difference between alive and deceased participants with regard to the percentage of females, obesity rate, presence of HTN, and the number of people on statins. Factors associated with shorter survival included older age, low HDL-C (<1.1 mmol/L), DM, stroke, and HF.
**Table 1**
| Variables | Alive (n=370) | All-Cause Death (n=115) | P value |
| --- | --- | --- | --- |
| Age (years) | 68.0±5.4 | 73.3±7.2 | <0.0001 |
| Female gender | 207 (55.9) | 63 (54.8) | 0.826 |
| BMI | | | 0.008 |
| <18.5 kg/m2 | 0 (0) | 2 (1.9) | 0.008 |
| 18.5-24.9 kg/m2 | 38 (10.4) | 18 (16.7) | 0.008 |
| 25-29.9 kg/m2 | 132 (36) | 27 (25) | 0.008 |
| ≳30 kg/m2 | 197 (53.7) | 61 (56.5) | 0.008 |
| Hypertension | 233 (63) | 83 (72.2) | 0.070 |
| Diabetes mellitus | 245 (66.2) | 89 (77.4) | 0.024 |
| TC | | | 0.987 |
| <4.5 mmol/L | 200 (54.1) | 62 (53.9) | 0.987 |
| 4.5-5 mmol/L | 78 (21.1) | 23 (20.0) | 0.987 |
| 5.1-5.69 mmol/L | 60 (16.2) | 19 (16.5) | 0.987 |
| ≥5.7 mmol/L | 32 (8.6) | 11 (9.6) | 0.987 |
| LDL-C | | | 0.520 |
| <2.6 mmol/L | 155 (41.9) | 57 (49.6) | 0.520 |
| 2.6-3.09 mmol/L | 90 (24.3) | 26 (22.6) | 0.520 |
| 3.1-3.59 mmol/L | 56 (15.1) | 14 (12.2) | 0.520 |
| >3.59 mmol/L | 69 (18.6) | 18 (15.7) | 0.520 |
| HDL-C | | | 0.009 |
| <1.1 mmol/L | 65 (17.6) | 36 (31.3) | 0.009 |
| 1.1-1.39 mmol/L | 138 (37.3) | 38 (33.0) | 0.009 |
| 1.4-1.59 mmol/L | 84 (22.7) | 16 (13.9) | 0.009 |
| ≥1.6 mmol/L | 83 (22.4) | 25 (21.7) | 0.009 |
| TG | | | 0.182 |
| <0.9 mmol/L | 46 (12.4) | 15 (13.0) | 0.182 |
| 0.9-1.2 mmol/L | 135 (36.5) | 30 (26.1) | 0.182 |
| 1.21-1.69 mmol/L | 87 (23.5) | 29 (25.2) | 0.182 |
| ≥1.7 mmol/L | 102 (27.6) | 41 (35.7) | 0.182 |
| Ischemic heart disease | 76 (20.5) | 30 (26.1) | 0.209 |
| Stroke | 43 (11.6) | 38 (33.0) | <0.0001 |
| Heart failure | 33 (8.9) | 40 (34.8) | <0.0001 |
| Statins | 321 (86.8) | 95 (82.6) | 0.266 |
Figure 1 shows the relationship between HDL-C and all-cause mortality. The multivariable hazard ratios ($95\%$ CI) for all-cause mortality were 0.585 (0.371-0.924), 0.363 (0.201-0.655), and 0.494 (0.296-0.825) for the 2nd, 3rd, and 4th quartiles of HDL-C, respectively, when compared to the first quartile, after adjusting for age and DM. TG, LDL-C, and TC levels were not associated with all-cause mortality. Regarding cardiovascular outcomes, low HDL-C (<1.1 mmol/L) was associated with an increased risk of having IHD (adjusted hazard ratio 3.2 ($95\%$ CI 1.6-6.2)), and high TC (≥5.7 mmol/L) was associated with an increased risk of having HF (adjusted hazard ratio 2.1 ($95\%$ CI 1.1-4.0)).
**Figure 1:** *The relationship between HDL-C and all-cause mortality. Adjusted hazard ratios in the second, third, and fourth quartiles compared with the first quartile.HDL-C: high-density lipoprotein cholesterol*
## Discussion
Dyslipidemia is strongly associated with CVD and mortality, but this association is complicated in the older population. In our study, we explored the relationship between levels of plasma lipids, all-cause mortality, and cardiovascular outcomes. Our findings showed that low HDL-C was significantly associated with a higher risk for all-cause mortality, which is in line with the literature. A Taiwanese cohort study found that low HDL-C (< 1.1 mmol/l) was significantly associated with a high risk for all-cause mortality in people older than 65 years [5]. Low HDL-C was also linked to higher mortality in elderly subjects from a community-based cohort in the United States [11]. Other studies show a similar link [8-10,19], which is likely related to the loss of the well-known protective antioxidative, antithrombotic, and anti-inflammatory effects of HDL-C [25].
Our study did not show a significant association of low TC, LDL-C, or TG levels with all-cause mortality. A recent meta-analysis reviewed the association between LDL-C and mortality in the elderly population and included 19 studies. Of these, four studies similarly found no association, while the rest of the studies showed an inverse relationship with all-cause mortality. For cardiovascular mortality, seven studies showed no association, and only two studies had an inverse association [6].
Regarding TC, there is strong evidence that low levels are associated with higher mortality, particularly all-cause mortality in the older population [3-5]. However, this association is typically attributed to the fact that older people with low TC are linked to having more frailty, nutritional deficiencies, and comorbidities like cancer and other terminal diseases [26]. In our study, we tried to eliminate these factors by excluding subjects who died within one year of follow-up.
Globally, CVD imposes the most significant disease burden among people aged 60 years and older [27]. We investigated the association between the plasma lipid profile and cardiovascular outcomes (IHD, CHF, and stroke). Our results showed that low HDL-C was significantly associated with an increased risk of IHD, similar to what other studies targeting the elderly have reported [8-10, 19]. Furthermore, our results showed a significant association between high TC and CHF. A previous study on the association between TC and cardiovascular outcomes showed a similar link with a higher risk in subjects with high TC [22].
The role of low HDL-C levels as a predictor of cardiovascular morbimortality and all-cause mortality in the older population has been proven in studies done in Western countries and East Asia [5,8-11,19]. Yet global lipid management guidelines mainly target LDL-C and not HDL-C, including those of the American College of Cardiology/American Heart Association (ACC/AHA) [28] and the National Institute for Health and Care Excellence (NICE) [29]. Additionally, these guidelines do not emphasize the elderly and primarily target the middle age group.
To the best of our knowledge, this study is the first in Saudi Arabia and the region to investigate the association of plasma lipid concentrations with all-cause mortality and cardiovascular outcomes in an older population. This is particularly important as the country is following the global increase in the aging population, with the proportion of those aged 60 years and above projected to increase to $9.5\%$ and $18.4\%$ in 2035 and 2050, respectively [30].
Limitations Due to the nature of this study design, reverse causality may be a concern, as the direction of the relationship between exposure and outcome may be unclear in retrospective studies. Also, we could not identify cause-specific mortality due to the lack of current infrastructure for retrieving data from the death registry to specify the cause of death. Furthermore, a larger prospective cohort study will be necessary to obtain more robust data regarding the older Saudi population.
## Conclusions
Our study showed that in patients aged 60 years and older, low HDL-C (<1.1 mmol/L) was associated with a higher risk of all-cause mortality and IHD. High TC was associated with an increased risk of having CHF. No significant association was found for LDL-C, TC, and TG with increased risk for all-cause mortality. With current dyslipidemia management guidelines targeting LDL-C, it is safe to argue that more focus on targeting HDL-C is needed. Additionally, future research should explore the causes and risk factors of low HDL-C in the elderly.
## References
1. Anderson KM, Castelli WP, Levy D. **Cholesterol and mortality. 30 years of follow-up from the Framingham study**. *JAMA* (1987.0) **257** 2176-2180. PMID: 3560398
2. Okamura T, Tanaka H, Miyamatsu N. **The relationship between serum total cholesterol and all-cause or cause-specific mortality in a 17.3-year study of a Japanese cohort**. *Atherosclerosis* (2007.0) **190** 216-223. PMID: 16529754
3. Kronmal RA, Cain KC, Ye Z, Omenn GS. **Total serum cholesterol levels and mortality risk as a function of age. A report based on the Framingham data**. *Arch Intern Med* (1993.0) **153** 1065-1073. PMID: 8481074
4. Cabrera MA, de Andrade SM, Dip RM. **Lipids and all-cause mortality among older adults: a 12-year follow-up study**. *ScientificWorldJournal* (2012.0) **2012** 930139. PMID: 22666169
5. Wang MC, Hu HY, Lin IF, Chuang JT. **Plasma lipid concentrations and survival in geriatric population: a retrospective cohort study**. *Medicine (Baltimore)* (2019.0) **98** 0
6. Ravnskov U, Diamond DM, Hama R. **Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly: a systematic review**. *BMJ Open* (2016.0) **6** 0
7. Zhou L, Wu Y, Yu S, Shen Y, Ke C. **Low-density lipoprotein cholesterol and all-cause mortality: findings from the China health and retirement longitudinal study**. *BMJ Open* (2020.0) **10** 0
8. Li JZ, Chen ML, Wang S, Dong J, Zeng P, Hou LW. **Apparent protective effect of high density lipoprotein against coronary heart disease in the elderly**. *Chin Med J (Engl)* (2004.0) **117** 511-515. PMID: 15109440
9. Weverling-Rijnsburger AW, Jonkers IJ, van Exel E, Gussekloo J, Westendorp RG. **High-density vs low-density lipoprotein cholesterol as the risk factor for coronary artery disease and stroke in old age**. *Arch Intern Med* (2003.0) **163** 1549-1554. PMID: 12860577
10. Santos RD, Barter PJ. **HDL-cholesterol level and mortality occurrence in the elderly: is the good cholesterol always good?**. *J Clin Endocrinol Metab* (2019.0)
11. Li ZH, Lv YB, Zhong WF. **High-density lipoprotein cholesterol and all-cause and cause-specific mortality among the elderly**. *J Clin Endocrinol Metab* (2019.0) **104** 3370-3378. PMID: 30869791
12. Liu J, Zeng FF, Liu ZM, Zhang CX, Ling WH, Chen YM. **Effects of blood triglycerides on cardiovascular and all-cause mortality: a systematic review and meta-analysis of 61 prospective studies**. *Lipids Health Dis* (2013.0) **12** 159. PMID: 24164719
13. Schupf N, Costa R, Luchsinger J, Tang MX, Lee JH, Mayeux R. **Relationship between plasma lipids and all-cause mortality in nondemented elderly**. *J Am Geriatr Soc* (2005.0) **53** 219-226. PMID: 15673344
14. Criqui MH, Heiss G, Cohn R. **Plasma triglyceride level and mortality from coronary heart disease**. *N Engl J Med* (1993.0) **328** 1220-1225. PMID: 8464432
15. **World Health Organization. Cardiovascular diseases (CVDs). 2017**
16. Nelson RH. **Hyperlipidemia as a risk factor for cardiovascular disease**. *Prim Care* (2013.0) **40** 195-211. PMID: 23402469
17. Hayashi T, Kawashima S, Itoh H. **Low HDL cholesterol is associated with the risk of stroke in elderly diabetic individuals: changes in the risk for atherosclerotic diseases at various ages**. *Diabetes Care* (2009.0) **32** 1221-1223. PMID: 19509009
18. Yu J, Liu X, Chen S. **Effects of low-density lipoprotein cholesterol on cardiovascular disease and all-cause mortality in elderly patients (≥75 years old)**. *Endocrine* (2022.0) **75** 418-426. PMID: 34618323
19. Zimetbaum P, Frishman WH, Ooi WL, Derman MP, Aronson M, Gidez LI, Eder HA. **Plasma lipids and lipoproteins and the incidence of cardiovascular disease in the very elderly. The Bronx Aging Study**. *Arterioscler Thromb* (1992.0) **12** 416-423. PMID: 1558833
20. Nanna MG, Navar AM, Wojdyla D, Peterson ED. **The association between low-density lipoprotein cholesterol and incident atherosclerotic cardiovascular disease in older adults: results from the National Institutes of Health Pooled Cohorts**. *J Am Geriatr Soc* (2019.0) **67** 2560-2567. PMID: 31411740
21. Ravnskov U, de Lorgeril M, Diamond DM. **LDL-C does not cause cardiovascular disease: a comprehensive review of the current literature**. *Expert Rev Clin Pharmacol* (2018.0) **11** 959-970. PMID: 30198808
22. Aronow WS, Ahn C. **Risk factors for new coronary events in a large cohort of very elderly patients with and without coronary artery disease**. *Am J Cardiol* (1996.0) **77** 864-866. PMID: 8623741
23. Aberra T, Peterson ED, Pagidipati NJ. **The association between triglycerides and incident cardiovascular disease: what is "optimal"?**. *J Clin Lipidol* (2020.0) **14** 438-447. PMID: 32571728
24. LaRosa JC. **Triglycerides and coronary risk in women and the elderly**. *Arch Intern Med* (1997.0) **157** 961-968. PMID: 9140266
25. White R, Giordano S, Datta G. **Role of HDL-associated proteins and lipids in the regulation of inflammation**. *Advances in Lipoprotein Research* (2017.0) 118
26. Hwang AC, Liu LK, Lee WJ, Chen LY, Peng LN, Lin MH, Chen LK. **Association of frailty and cardiometabolic risk among community-dwelling middle-aged and older people: results from the I-lan longitudinal aging study**. *Rejuvenation Res* (2015.0) **18** 564-572. PMID: 26556635
27. Prince MJ, Wu F, Guo Y. **The burden of disease in older people and implications for health policy and practice**. *Lancet* (2015.0) **385** 549-562. PMID: 25468153
28. Arnett DK, Blumenthal RS, Albert MA. **2019 ACC/AHA Guideline on the primary prevention of cardiovascular disease: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines**. *J Am Coll Cardiol* (2019.0) **74** 0-232
29. Duerden M, O'Flynn N, Qureshi N. **Cardiovascular risk assessment and lipid modification: NICE guideline**. *Br J Gen Pract* (2015.0) **65** 378-380. PMID: 26120133
30. **United Nations. World Population Prospects: The 2012 Revision 2012**
|
---
title: A Comparison of the Quality of Life of People With Epilepsy Receiving Home-Based
and Clinic-Based Epilepsy Care Using the European Quality of Life Five-Dimension
Three-Level (EQ-5D-3L) Scale
journal: Cureus
year: 2023
pmcid: PMC10023070
doi: 10.7759/cureus.35045
license: CC BY 3.0
---
# A Comparison of the Quality of Life of People With Epilepsy Receiving Home-Based and Clinic-Based Epilepsy Care Using the European Quality of Life Five-Dimension Three-Level (EQ-5D-3L) Scale
## Abstract
Background and objective *Epilepsy is* a chronic neurological condition that, both physically and psychologically, puts a person at risk for poor quality of life (QOL). People with epilepsy (PWE) may experience shame, fear, and rejection and feel discriminated against, hence avoiding social interactions. To avoid being labeled as having epilepsy, patients may conceal their disease and refuse medical attention, which can lead to treatment discontinuation and significantly impact the quality of life. Epilepsy care in India has fallen back on primary care physicians because there are not enough neurologists available to treat the condition. Home-based care (HBC) may overcome many barriers by providing free antiepileptic drugs (AEDs), eliminating the “distance to a health facility,” and providing correct information that may improve QOL. This study is therefore conducted to compare the QOL between people with epilepsy receiving home-based care (HBC) and routine clinic-based care (CBC).
Methodology The people with epilepsy enrolled in this study were already part of a community-based randomized controlled trial conducted to compare the effect of regular home-based epilepsy care with routine clinic-based epilepsy care on antiepileptic adherence among urban and peri-urban areas of Ludhiana, Punjab, India (explained further in the study). The present study is a cohort study where the two cohorts, one receiving home-based epilepsy care ($$n = 97$$) and the other receiving routine clinic-based epilepsy care ($$n = 76$$), were compared for QOL at two points in time, i.e., at baseline (at enrolment) and after 24 months of receiving epilepsy care, using the European Quality of Life Five-Dimension Three-Level (EQ-5D-3L) scale.
Results The mean EQ-5D-3L index scores for the HBC group at baseline were 0.88 ± 0.15, and after 24 months, the scores increased to 0.94 ± 0.17. The baseline mean index scores for the CBC group were 0.89 ± 0.21, and after 24 months, the value increased to 0.90 ± 0.19. The mean difference in QOL in the HBC group showed a higher difference than in the CBC group (0.06 ± 0.1 versus 0.01 ± 0.1), but the difference was found to be statistically not significant ($$p \leq 0.067$$). As per the five dimensions of the EQ-5D-3L scale, i.e., mobility, self-care, usual activities, pain/discomfort, and anxiety/depression, there was a decrease in the number of PWE reporting problems among both groups after 24 months of epilepsy care. Sociodemographic and clinical variables such as level of education, working status, age at the onset of seizures, frequency of seizures, treatment regimen, presence of comorbidities, and adverse drug reactions significantly affect the QOL of people with epilepsy at $p \leq 0.05.$
Conclusion The results of the study emphasize that epilepsy has a negative impact on QOL. The results showed a higher QOL among the people in the HBC group as compared to the CBC group, but the difference was not statistically significant. There was an improvement in QOL from baseline after dedicated care in both groups. The problems related to mobility, self-care, usual activities, pain/discomfort, and anxiety/depression have been significantly reduced in the HBC group. Having low levels of education, not having a job, starting to have seizures at a young age, having seizures more often, receiving more than one type of treatment, and the presence of other health problems and side effects are factors associated with poor QOL among people with epilepsy.
## Introduction
Epilepsy is a chronic neurological disorder characterized by recurrent seizures, which may vary from a brief lapse of attention or muscle jerks to severe and prolonged convulsions. The seizures are caused by sudden, usually brief disruptions of electrical activity in the brain cells [1,2]. The International League Against Epilepsy (ILAE) defined epilepsy as a disease diagnosed by any of the following criteria: (i) at least two unprovoked (or reflex) seizures occurring more than 24 hours apart, (ii) one unprovoked (or reflex) seizure and a probability of further seizures similar to the general recurrence risk (at least $60\%$ recurrence risk) after two unprovoked seizures occurring over the next 10 years, or (iii) having an epilepsy syndrome [3]. According to the World Health Organization (WHO), approximately 50 million people worldwide suffer from epilepsy, with more than $85\%$ living in developing countries. Based on a conservative estimate of $1\%$ as the prevalence of epilepsy, India has more than 12 million PWE, accounting for nearly one-sixth of the global burden [4,5].
Epilepsy affects patients’ physical, psychological, social, cognitive, and behavioral health and puts a person at risk for poor health-related quality of life (HRQOL). The psychological impact is related to the uncertainty of seizures, fear of recurrent seizures, associated physical injuries, treatment-related side effects, lifestyle restrictions, physical difficulties, and perceived stigmatization. As a result of stigma, people with epilepsy (PWE) may experience shame, fear, and rejection and feel discriminated against, hence avoiding social interactions. To avoid being labeled as having epilepsy, patients may conceal their disease and refuse medical attention.
Additionally, the availability of epilepsy care in low- and middle-income countries (LMICs) is problematic due to both supply-side and demand-side constraints (e.g., lack of antiepileptic drugs (AEDs), distances to healthcare facilities, a lack of expertise in treating epilepsy, treatment costs, cultural beliefs regarding epilepsy, and faith in traditional treatment providers). Together, these factors have the potential to result in psychological stress and low self-esteem, which can lead to discontinuation (nonadherence) of treatment. Nonadherence results in seizure relapse, status epilepticus, hospitalizations, increased healthcare costs, and, in rare cases, sudden unexpected death from epilepsy (SUDEP) and could significantly impact the quality of life (QOL). The QOL is also influenced by the type of seizures, frequency of seizure, treatment regimen, duration of epilepsy, presence of comorbidities, and adverse effects, and by some sociodemographic characteristics such as age, marital status, and employment status. Therefore, appropriate care must be provided to prevent complications related to epilepsy [6-8].
Epilepsy care in India has fallen back on primary care physicians because there are not enough neurologists available to treat the condition. The WHO encourages primary healthcare providers to deliver epilepsy care in countries and regions where adequate facilities are not available [9]. Home-based epilepsy care (HBC) may overcome barriers by providing routine, free AEDs, eliminating the “distance to a health facility,” and providing information and advice to reduce stigma and false beliefs, as well as supporting self-management by primary care workers. Therefore, a community-based cluster randomized trial named Community Interventions for Epilepsy (CIFE) was conducted to ascertain whether home-based care with the community and primary healthcare workers’ support improves adherence to AEDs over routine clinic-based care (CBC) among people with epilepsy [10,11]. The present study was an extension of this community-based trial, where the same groups receiving home-based epilepsy care and routine clinic-based epilepsy care were followed to assess and compare the quality of life.
Over the past few decades, many quality of life questionnaires have been developed and used to assess the quality of life of patients with various diseases. Some are disease-specific, and others are generic. While disease-specific instruments are more sensitive to detecting changes in health related to disease, generic instruments were built to assess and compare the lifestyles of people suffering from various pathologies [12-14]. The European Quality of Life Five-Dimension Three-Level (EQ-5D-3L) scale, which is a generic questionnaire, was used in this study with the aim to compare QOL among people with epilepsy receiving home-based epilepsy care and clinic-based epilepsy care.
## Materials and methods
This is a cohort study in which the quality of life of two cohorts receiving home-based and routine clinic-based epilepsy care was compared. The study included a total of 173 PWE, more than one year of age, screened, diagnosed, and recruited by a panel of neurologists under a community-based trial (the Community Interventions for Epilepsy (CIFE) project) [10,11]. PWE were assessed monthly for 24 months by trained field-workers and neurologists. CIFE was a two-step cluster-randomized trial during which 59,509 people were screened for epilepsy in urban and peri-urban areas of Ludhiana, Punjab. The trained field-workers carried out door-to-door screening among 24 selected clusters of around 2,000 people each using a validated questionnaire. Screen-positive people were then invited for evaluation by neurologists specializing in epilepsy at a tertiary care hospital facility. It was finally conducted on 240 people with epilepsy, equally divided among both groups, to compare the effect of regular home-based care with the routine clinic-based care provided by neurologists on AED adherence in people with epilepsy (Figure 1).
**Figure 1:** *Flowchart showing PWE selectionPWE: people with epilepsy*
PWE under the routine clinic-based epilepsy care were asked to attend monthly clinics at the Government District Hospital for review visits and to receive free antiepileptic drugs (AEDs), while those under the home-based epilepsy care received an interventional package comprising (a) free delivery of AEDs, (b) education and counseling about self-management, social functioning, and stigma abrogation, and (c) adherence monitoring, all provided at home on a monthly basis by study personnel with qualifications equivalent to auxiliary nurse midwives (ANMs). The duration of the trial was from December 2017 to August 2020 (24 months for each patient).
In this present study, the same cohort of PWE receiving home-based and routine clinic-based epilepsy care were assessed for QOL using the EQ-5D-3L scale. All PWE who have completed 24 assessments (over 24 months) under the CIFE project, are older than one year, and are willing to participate in the study were selected as part of the study using the total enumerative sampling technique. The study was approved by the Research and Development Centre of Dayanand Medical College and Hospital (approval number DMCH/R&D/$\frac{2018}{825}$). Informed consent was obtained from all subjects over the age of 18, as well as from a parent or guardian in the case of children (in addition to oral consent from children 7-11 years old and written consent from children over 12 years old). The details of clinical condition, frequency of seizures, treatment regimen, presence of comorbidities, and others were taken from medical records. The selected PWE were interviewed for demographic data, clinical profile, and quality of life using the EQ-5D-3L scale.
The EQ-5D-3L scale is a generic instrument that assesses the overall condition of the patient (physical, psychological, and social) regardless of the pathology [15]. It helps compare the lifestyles of groups of subjects to many pathologies. The questionnaire has two parts: the EQ-5D descriptive system and the EQ visual analog scale (EQ-VAS).
EQ-5D descriptive system This includes five dimensions of health: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. Each dimension has three levels: no problems, some problems, and extreme problems. Patients were asked to select one of three health levels for each of the five dimensions. The answers given can be combined into a number of five digits that describes the patient’s health state, where “11111” is a perfect health state. The results of the health state can be transformed into a single index value using the value sets given for different countries. As the value sets are not available for India, the value set for China has been used for the study (as per the EQ-5D-3L instrument guide instructions).
EQ visual analog scale The EQ-VAS records the patient’s self-rated health on a vertical visual analog scale where the endpoints are labeled “best imaginable health state” and “worst imaginable health state.” The VAS can be used as a quantitative measure of health outcomes that reflect the patient’s own judgment. The people were then classified based on their self-reported health states as very good [81-100], good [51-80], normal [31-50], and bad [0-30].
The EuroQOL International Group provided permission to use the instrument via email. The scale was already available in Hindi and Punjabi on the EuroQOL website. The questionnaire was validated by the experts for English, Hindi, and Punjabi, and the reliability of the scale was tested using Cronbach's alpha (EQ-5D descriptive system: $r = 0.94$; EQ-VAS: $r = 0.98$). Data were collected through face-to-face interviews with the researchers. The questionnaire was pretested on a selected group of people with epilepsy to ensure that all the questions were properly understood by the respondents. Changes were made, and the final version of the questionnaire was used to collect the data. The groups were then compared for quality of life at baseline and after 24 months of receiving home-based and routine clinic-based epilepsy care. The Statistical Package for the Social Sciences (SPSS) version 20 (IBM SPSS Statistics, Armonk, NY, USA) was used to analyze the data. The results were expressed as means and standard deviation (SD) for variables that met the criteria of normality. For variables that were not normally distributed, non-parametric tests (Mann-Whitney U and Kruskal-Wallis tests) were used to find out the association of EQ-5D scores with sociodemographic and clinical variables. The statistical significance level was set at 0.05.
## Results
Sociodemographic profile of people with epilepsy As per the sociodemographic profile of people with epilepsy, the majority were in the age group of 13-30 years, i.e., 87 ($50.2\%$), with a mean age of 25.64 ± 15.1. Most of the PWE were male ($65.3\%$) and unmarried ($61.8\%$). Fifty-four ($31.2\%$) PWE were secondary or senior secondary educated. Only about one-third of people with epilepsy ($37.6\%$) were working, and of them, almost half were unskilled workers ($40.1\%$). There was no difference between the groups as checked using the chi-square test of homogeneity (Table 1).
**Table 1**
| Sociodemographic variables | Groups | Groups.1 | Total | ᵡ2 value | p value |
| --- | --- | --- | --- | --- | --- |
| Sociodemographic variables | HBC (n = 97) | CBC (n = 76) | Total | ᵡ2 value | p value |
| Sociodemographic variables | Number (%) | Number (%) | Total | ᵡ2 value | p value |
| Age (in years) | | | | | |
| 13-30 | 46 (47.4) | 41 (54) | 87 (50.2) | 2.25 | 0.521NS |
| 30-50 | 36 (37.1) | 25 (32.9) | 61 (35.3) | 2.25 | 0.521NS |
| 50-70 | 15 (15.5) | 9 (11.8) | 24 (13.9) | 2.25 | 0.521NS |
| 70-90 | - | 1 (1.3) | 1 (0.6) | 2.25 | 0.521NS |
| Gender | | | | | |
| Male | 65 (67.1) | 48 (63.1) | 113 (65.3) | 0.279 | 0.597NS |
| Female | 32 (32.9) | 28 (36.8) | 60 (34.7) | 0.279 | 0.597NS |
| Marital status | | | | | |
| Unmarried | 59 (60.7) | 48 (63.2) | 107 (61.8) | 1.622 | 0.654NS |
| Married | 35 (36.1) | 23 (30.3) | 58 (33.5) | 1.622 | 0.654NS |
| Widow/widower/separated | 2 (2.1) | 3 (3.9) | 5 (2.9) | 1.622 | 0.654NS |
| Divorced | 1 (1.1) | 2 (2.6) | 3 (1.7) | 1.622 | 0.654NS |
| Education | | | | | |
| Illiterate | 19 (19.6) | 27 (35.5) | 46 (26.6) | 10.21 | 0.337NS |
| Primary | 20 (20.6) | 12 (15.8) | 32 (18.5) | 10.21 | 0.337NS |
| Secondary/senior secondary | 38 (39.2) | 16 (21.1) | 54 (31.2) | 10.21 | 0.337NS |
| Graduate and above | 4 (4.1) | 3 (3.9) | 7 (4.1) | 10.21 | 0.337NS |
| Studying | 16 (16.5) | 18 (23.7) | 34 (19.7) | 10.21 | 0.337NS |
| Working status | | | | | |
| Working | 38 (39.2) | 27 (35.5) | 65 (37.6) | 0.242 | 0.623NS |
| Not working | 59 (60.8) | 49 (64.5) | 108 (62.4) | 0.242 | 0.623NS |
| Occupation of those who are working (HBC: n = 38; CBC: n = 27) | | | | | |
| Professional | 2 (5.3) | - | 2 (3.1) | 4.058 | 0.398NS |
| Semiskilled worker | 13 (34.2) | 6 (22.2) | 19 (29.2) | 4.058 | 0.398NS |
| Skilled worker | 11 (28.9) | 7 (25.9) | 18 (27.6) | 4.058 | 0.398NS |
| Unskilled worker | 12 (31.5) | 14 (51.8) | 26 (40.1) | 4.058 | 0.398NS |
Clinical profile of people with epilepsy As per the clinical profile, the majority of people with epilepsy were diagnosed with symptomatic focal epilepsy in both home-based care and clinic-based care, i.e., 40 ($41.2\%$) and 32 ($42.1\%$), respectively. Among the home-based care group, the age of onset of seizures was less than five years among most PWE ($34.1\%$), while the duration of seizures was 10-15 years in the majority of them. The majority of PWE in the clinic-based care group had seizures between the ages of 10 and 15, and the duration was five years. The majority of PWE in both the HBC and CBC groups had monthly seizures, i.e., $28.7\%$ and $35.5\%$, respectively. Most of the PWE in the HBC group were prescribed polytherapy ($54.6\%$) as a treatment regimen, while in the CBC group, an equal number of PWE were prescribed monotherapy and polytherapy, i.e., $50\%$. Adverse drug reactions were present in 26 ($26.8\%$) PWE in the home-based care group as compared to 13 ($17.1\%$) in the clinic-based care group. Comorbidities were present in approximately half of the PWE among both groups, i.e., 47 ($48.5\%$) and 31 ($40.8\%$), respectively. After 24 months, 27 ($27.8\%$) PWE had controlled epilepsy, 24 ($24.7\%$) had drug-resistant epilepsy (DRE), and 46 ($47.4\%$) had indeterminate epilepsy (which could not be classified under the above two categories). As depicted by the p value, there was no significant difference between the groups as per their clinical profile (Table 2).
**Table 2**
| Clinical profile | Groups | Groups.1 | ᵡ2 value | p value |
| --- | --- | --- | --- | --- |
| Clinical profile | HBC (n = 97) | CBC (n = 76) | ᵡ2 value | p value |
| Clinical profile | Number (%) | Number (%) | ᵡ2 value | p value |
| Diagnosis | | | | |
| Idiopathic focal epilepsy | 2 (2.1) | 1 (1.3) | 3.145 | 0.790NS |
| Idiopathic generalized epilepsy | 18 (18.6) | 10 (13.1) | 3.145 | 0.790NS |
| Symptomatic focal epilepsy | 40 (41.2) | 32 (42.1) | 3.145 | 0.790NS |
| Symptomatic generalized epilepsy | 5 (5.2) | 6 (7.9) | 3.145 | 0.790NS |
| Unestablished | 32 (33.1) | 27 (35.5) | 3.145 | 0.790NS |
| Age at onset of seizures (in years) | | | | |
| <5 | 33 (34.1) | 18 (23.7) | 7.865 | 0.164NS |
| 5-10 | 16 (16.5) | 10 (13.1) | 7.865 | 0.164NS |
| 10-15 | 11 (11.3) | 20 (26.3) | 7.865 | 0.164NS |
| 15-20 | 11 (11.3) | 11 (14.5) | 7.865 | 0.164NS |
| 20-25 | 15 (15.5) | 10 (13.1) | 7.865 | 0.164NS |
| >25 | 11 (11.3) | 7 (9.2) | 7.865 | 0.164NS |
| Duration of epilepsy (in years) | | | | |
| <5 | 19 (19.6) | 22 (28.9) | 6.173 | 0.290NS |
| 5-10 | 21 (21.6) | 12 (15.8) | 6.173 | 0.290NS |
| 10-15 | 22 (22.6) | 8 (10.5) | 6.173 | 0.290NS |
| 15-20 | 13 (13.4) | 9 (11.8) | 6.173 | 0.290NS |
| 20-25 | 7 (7.2) | 4 (5.3) | 6.173 | 0.290NS |
| >25 | 15 (15.5) | 21 (27.6) | 6.173 | 0.290NS |
| Frequency of seizures | | | | |
| Daily | 10 (10.3) | 4 (5.3) | 5.747 | 0.332NS |
| Weekly | 9 (9.3) | 8 (10.5) | 5.747 | 0.332NS |
| Monthly | 28 (28.7) | 27 (35.5) | 5.747 | 0.332NS |
| Annually | 6 (6.2) | 4 (5.3) | 5.747 | 0.332NS |
| Sporadic | 32 (33.1) | 17 (22.4) | 5.747 | 0.332NS |
| Biannually | 12 (12.4) | 16 (21.1) | 5.747 | 0.332NS |
| Treatment regimen | | | | |
| Monotherapy | 44 (45.4) | 38 (50) | 2.134 | 0.544NS |
| Polytherapy | 53 (54.6) | 38 (50) | 2.134 | 0.544NS |
| Adverse drug reactions | | | | |
| Present | 26 (26.8) | 13 (17.1) | 2.296 | 0.130NS |
| Not present | 71 (73.2) | 63 (82.9) | 2.296 | 0.130NS |
| Comorbidities | | | | |
| Present | 47 (48.5) | 31 (40.8) | 1.011 | 0.315NS |
| Not present | 50 (51.5) | 45 (59.2) | 1.011 | 0.315NS |
| Seizure control | | | | |
| Controlled epilepsy | 27 (27.8) | 23 (30.3) | 2.838 | 0.242NS |
| Drug-resistant epilepsy | 24 (24.7) | 26 (34.2) | 2.838 | 0.242NS |
| Indeterminate | 46 (47.4) | 27 (35.5) | 2.838 | 0.242NS |
Assessment of self-reported quality of life using the EQ-5D-3L scale The EQ-5D-3L scale was used to assess the health status of people with epilepsy. Under the descriptive system, the following dimensions were assessed: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. These dimensions were assessed at two points in time: baseline (after recruitment) and after 24 months. Figure 2 depicts the percentage of PWE having some problems (level 2 + 3 of EQ-5D-3L scale) at baseline and after 24 months of receiving home-based epilepsy care. The bar in the diagram depicts the presence of some problems (level 2 + 3). The results show that after 24 months, there was a decrease in the percentage of PWE having some problems on all dimensions of the EQ-5D-3L scale (mobility, self-care, usual activities, pain or discomfort, and anxiety or depression).
**Figure 2:** *Percentage of PWE having some problems (level 2 + 3) at baseline and after 24 months of receiving home-based epilepsy care (n = 97)PWE: people with epilepsy, HBC: home-based care, EQ-5D-3L: European Quality of Life Five-Dimension Three-Level*
Figure 3 shows the percentage of PWE having some problems (level 2 + 3) at baseline and after 24 months of receiving clinic-based epilepsy care. There was a decrease in the percentage of PWE having problems as per the dimensions mobility, self-care, and pain/discomfort; however, there was no change in the percentage of PWE having problems related to usual activities. Most importantly, the percentage of PWE having anxiety or depression increased after 24 months of care.
**Figure 3:** *Percentage of PWE having some problems (level 2 + 3) at baseline and after 24 months of receiving clinic-based epilepsy care (n = 76)PWE: people with epilepsy, CBC: clinic-based care, EQ-5D-3L: European Quality of Life Five-Dimension Three-Level*
Figure 4 shows the percentage change in problems after 24 months among PWE receiving home-based and clinic-based epilepsy care. There was a negative change in percentage for all dimensions, which means that there was a decrease in the percentage of PWE having problems, but the anxiety/depression dimension in the clinic-based epilepsy care group showed a positive change, i.e., there was an increase in anxiety or depression after 24 months. The possible reason could be that the PWE in home-based care were receiving education and counseling about self-management of epilepsy, social functioning, and stigma abrogation, while the clinic-based epilepsy care group was receiving routine care (although this is not under the study).
**Figure 4:** *Percentage change in PWE reporting some problems (level 2 + 3) after receiving 24 months of epilepsy care as per various dimensions of the EQ-5D-3L scale (N = 173)HBC: home-based care, CBC: clinic-based care, PWE: people with epilepsy, EQ-5D-3L: European Quality of Life Five-Dimension Three-Level*
Calculation of index scores The obtained health states were noted and converted into index scores as per the given value sets. The value sets for the health states in India were not available; therefore, Chinese value sets were used in the study [16]. The EuroQOL Foundation recommends selecting one of the nearby/similar country value set in the absence of a country-specific value set [15].
A score of “1” indicates perfect health. The higher the value (close to 1), the better the health. Both groups were similar as per the baseline index scores ($$p \leq 0.503$$). The mean EQ-5D-3L index scores for the HBC group at baseline were 0.88 ± 0.15, while after 24 months, they increased to 0.94 ± 0.17. The baseline mean index scores for the CBC group were 0.89 ± 0.21, while after 24 months, the value increased to 0.90 ± 0.19 (Table 3). The mean difference in index scores in the HBC group showed a higher difference than in the CBC group (0.06 ± 0.1 versus 0.01 ± 0.1). However, the difference found between the two groups was statistically not significant at $$p \leq 0.067.$$
**Table 3**
| Groups | EQ-5D-3L index scores | EQ-5D-3L index scores.1 | Mean difference | Z value | p value |
| --- | --- | --- | --- | --- | --- |
| Groups | Baseline | After 24 months | Mean difference | Z value | p value |
| Groups | Mean ± SD | Mean ± SD | Mean difference | Z value | p value |
| HBC (n = 97) | 0.88 ± 0.15 | 0.94 ± 0.17 | 0.06 ± 0.1 | -0.722 | 0.067NS |
| CBC (n = 76) | 0.89 ± 0.21 | 0.90 ± 0.19 | 0.01 ± 0.1 | -0.722 | 0.067NS |
| Z value | -0.669 | | | | |
| p value | 0.503NS | | | | |
Assessment of the EQ visual analog scale (EQ-VAS) Health was self-assessed using the EQ-VAS scale at two points in time: baseline and after 24 months. Both groups were similar as per baseline EQ-VAS scores at $$p \leq 0.439.$$ Table 4 shows that people in both groups tended to improve their quality of life after receiving 24 months of care. However, the difference found between the two groups is statistically not significant as calculated using the Mann-Whitney test (Table 4).
**Table 4**
| Groups | EQ-VAS score | EQ-VAS score.1 | Mean difference | Z value | p value |
| --- | --- | --- | --- | --- | --- |
| Groups | Baseline | After 24 months | Mean difference | Z value | p value |
| Groups | Mean ± SD | Mean ± SD | Mean difference | Z value | p value |
| HBC (n = 97) | 67.1 ± 24.7 | 75.2 ± 27.6 | 8.1 ± 15.0 | -1.605 | 0.10NS |
| CBC (n = 76) | 64.1 ± 25.2 | 73.8 ± 25.4 | 9.7 ± 12.7 | -1.605 | 0.10NS |
| Z value | -0.773 | | | | |
| p value | 0.439NS | | | | |
Association of EQ-5D-3L scores with sociodemographic and clinical profile The EQ-5D-3L index scores were significantly associated with education level ($$p \leq 0.003$$), working status ($$p \leq 0.001$$), age at seizure onset ($$p \leq 0.012$$), frequency of seizures ($$p \leq 0.001$$), presence of comorbidities ($$p \leq 0.002$$), and adverse drug reactions ($$p \leq 0.012$$). The EQ-VAS scores were significantly associated with education level ($$p \leq 0.012$$), working status ($$p \leq 0.001$$), age at seizure onset ($$p \leq 0.004$$), frequency of seizures ($$p \leq 0.001$$), treatment regimen ($$p \leq 0.005$$), presence of comorbidities ($$p \leq 0.005$$), and adverse drug reactions ($$p \leq 0.031$$).
## Discussion
The study aimed to assess the quality of life of people with epilepsy using a generic EQ-5D-3L scale. Most of the studies done previously to assess HRQOL in people with epilepsy measured HRQOL at a single point in time [7,12,14,17]. There is very limited literature on quality of life among PWE using the EQ-5D-3L scale, and this is the first study where two different types of epilepsy care (home-based epilepsy care and clinic-based epilepsy care) were compared for change in the quality of life.
In the present study, the five dimensions of the EQ-5D-3L scale, namely, mobility, self-care, usual activities, pain or discomfort, and anxiety or depression, were also compared for the percentage of problems present at baseline and after 24 months of epilepsy care. There was a decrease in the percentage of problems on all dimensions of the EQ-5D-3L scale among PWE receiving home-based epilepsy care. Among the PWE receiving clinic-based care, there was a decrease in the percentage of PWE having problems with mobility, self-care, and pain/discomfort. However, there was no change in the percentage of PWE having problems related to usual activities, and the percentage of PWE having anxiety or depression increased after 24 months of care. The possible reason could be that the PWE in home-based care were receiving education and counseling about self-management of epilepsy, social functioning, and stigma abrogation, while the clinic-based epilepsy care group was receiving routine care (although this is not under the study). Similar results were found in another study, where an improvement in all dimensions of the EQ-5D-3L scale was observed when compared to baseline values [14].
The mean EQ-5D-3L index scores for the HBC group at baseline were 0.88 ± 0.15, while after 24 months, they increased to 0.94 ± 0.17. The baseline mean index scores for the CBC group were 0.89 ± 0.21, while after 24 months, the value increased to 0.90 ± 0.19. The mean difference in index scores in the HBC group showed a higher difference than in the CBC group (0.06 ± 0.1 versus 0.01 ± 0.1). However, the difference found between the two groups was statistically not significant at $$p \leq 0.067.$$ Similar results were found by de Souza et al. [ 2018], where the sample population tended toward a gain in quality of life after second follow-up visits to the outpatient clinic [13].
The EQ-5D-3L index scores and EQ-VAS scores were significantly associated with education level, working status, age at seizure onset, frequency of seizures, treatment regimen, presence of comorbidities, and adverse drug reactions. The variables found to be associated with poor HRQOL in people with epilepsy are similar to the factors found in other studies [7,17].
The study has some limitations, as the chosen cohorts were people with epilepsy who had previously been enrolled in a community-based randomized controlled trial. *The* generic instrument has been used to compare the burden of epilepsy in comparison to other disorders, but it does not give complete details about epilepsy. The number of subjects was limited in the groups, the survey location could also be expanded, and further studies can be replicated with larger cohorts to establish more evidence. The quality of life between children and adults could be compared separately to learn much about the differences between them.
## Conclusions
The results of the study emphasize that epilepsy has a negative impact on quality of life. There was an improvement in QOL from baseline after dedicated care in both groups. The results showed a higher QOL among the people in the HBC group as compared to the CBC group, but the difference was not statistically significant. The problems related to mobility, self-care, usual activities, pain/discomfort, and anxiety/depression have been significantly reduced in the HBC group. Having low levels of education, not having a job, starting to have seizures at a young age, having seizures more often, receiving more than antiepileptic drugs, and the presence of other health problems and side effects are factors found to be associated with poor QOL among people with epilepsy.
## References
1. **National Health Portal of India: Epilepsy**. (2022)
2. Fisher RS, Acevedo C, Arzimanoglou A. **ILAE official report: a practical clinical definition of epilepsy**. *Epilepsia* (2014) **55** 475-482. PMID: 24730690
3. Amudhan S, Gururaj G, Satishchandra P. **Epilepsy in India I: epidemiology and public health**. *Ann Indian Acad Neurol* (2015) **18** 263-277. PMID: 26425001
4. Feigin V, Nichols E, Allam T. **Global, regional, and national burden of neurological disorders, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016**. *Lancet Neurol* (2019) **18** 459-480. PMID: 30879893
5. Kultu A, Mülayim S. **Psychiatric comorbidities and quality of life in epilepsy**. *Epileptology - the modern state of science* (2016) 226-231
6. Berto P. **Quality of life in patients with epilepsy and impact of treatments**. *Pharmacoeconomics* (2002) **20** 1039-1059. PMID: 12456200
7. Bani-Issa W. **Evaluation of the health-related quality of life of Emirati people with diabetes: integration of sociodemographic and disease-related variables**. *East Mediterr Health J* (2011) **17** 825-830. PMID: 22276489
8. Hixson JD, Kirsch HE. **The effects of epilepsy and its treatments on affect and emotion**. *Neurocase* (2009) **15** 206-216. PMID: 19204849
9. **World Health Organization: Global burden of epilepsy and the need for coordinated action at the country level to address its health, social and public knowledge implications**. (2022)
10. Singh G, Sharma S, Bansal RK. **A home-based, primary-care model for epilepsy care in India: basis and design**. *Epilepsia Open* (2019) **4** 264-274. PMID: 31168493
11. Singh G, Sharma S, Bansal N. **A cluster-randomized trial comparing home-based primary health care and usual clinic care for epilepsy in a resource-limited country**. *Epilepsia Open* (2022) **7** 781-791. PMID: 36213959
12. Kehailou F, Jabari M, Labriji A. **Quality of life assessment with EQ-5D-3L in a Moroccan diabetic population**. *OALibJ* (2021) **8** 1-14
13. Souza IA, Pereira CC, Monteiro AL. **Assessment of quality of life using the EQ-5D-3L instrument for hospitalized patients with femoral fracture in Brazil**. *Health Qual Life Outcomes* (2018) **16** 194. PMID: 30249245
14. Bosić-Zivanović D, Medić-Stojanoska M, Kovacev-Zavisić B. **[The quality of life in patients with diabetes mellitus type 2]**. *Vojnosanit Pregl* (2012) **69** 858-863. PMID: 23155606
15. EuroQol Research Foundation. **EuroQol Research Foundation: EQ-5D user guides**. (2023)
16. Zhuo L, Xu L, Ye J, Sun S, Zhang Y, Burstrom K, Chen J. **Time trade-off value set for EQ-5D-3L based on a nationally representative Chinese population survey**. *Value Health* (2018) **21** 1330-1337. PMID: 30442281
17. Latif FI, Wahid HA, Mohamed AA, Farg HK. **Quality of life of type 2 diabetic patients in relation to gender and socio-economic status in Egypt**. *Int J Pharm Sci & Scient Res* (2016) **2** 152-160
|
---
title: Differentially methylated regions identified in bovine embryos are not observed
in adulthood
authors:
- Luna Nascimento Vargas
- Allice Rodrigues Ferreira Nochi
- Paloma Soares de Castro
- Andrielle Thainar Mendes Cunha
- Thainara Christie Ferreira Silva
- Roberto Coiti Togawa
- Márcia Marques Silveira
- Alexandre Rodrigues Caetano
- Maurício Machaim Franco
journal: Animal Reproduction
year: 2023
pmcid: PMC10023072
doi: 10.1590/1984-3143-AR2022-0076
license: CC BY 4.0
---
# Differentially methylated regions identified in bovine embryos are not observed in adulthood
## Abstract
The establishment of epigenetic marks during the reprogramming window is susceptible to environmental influences, and stimuli during this critical stage can cause altered DNA methylation in offspring. In a previous study, we found that low levels of sulphur and cobalt (low S/Co) in the diet offered to oocyte donors altered the DNA methylome of bovine embryos. However, due to the extensive epigenetic reprogramming that occurs during embryogenesis, we hypothesized that the different methylation regions (DMRs) identified in the blastocysts may not maintain in adulthood. Here, we aimed to characterize DMRs previously identified in embryos, in the blood and sperm of adult progenies of two groups of heifers (low S/Co and control). We used six bulls and characterized the DNA methylation levels of KDM2A, KDM5A, KMT2D, and DOT1L genes. Our results showed that all DMRs analysed in both groups and tissues were hypermethylated unlike that noticed in the embryonic methylome profiles. These results suggest that embryo DMRs were reprogrammed during the final stages of de novo methylation during embryogenesis or later in development. Therefore, due to the highly dynamic epigenetic state during early embryonic development, we suggest that is essential to validate the DMRs found in embryos in adult individuals.
## Introduction
DNA methylation has been studied in embryos of various species ever since techniques were first developed for the analysis of DNA methylation (Stevens et al., 1988; Kafri et al., 1992; Kang et al., 2001). The technological advancements of the past few decades have made it possible to access the embryonic methylome through whole-genome sequencing. Several studies that have analysed DNA methylation during embryogenesis in cattle, sheep, and humans have since been published (Guo et al., 2014; Zhu et al., 2018; Duan et al., 2019; Zhou et al., 2019; Zhang et al., 2021b). Although these analyses of the various embryonic stages have provided valuable information to elucidate some regulatory mechanisms, the epigenetic state during the early embryonic development is highly dynamic and requires further study.
During the initial stages of development in mammals, particularly gametogenesis and embryogenesis, extensive epigenetic reprogramming occurs to support proper embryonic and foetus growth (Reik et al., 2001; Canovas et al., 2017). First, a wide loss of DNA methylation is initiated during primordial germ cell (PGC) formation in the foetal phase (Reik et al., 2001). After demethylation, a subsequent de novo DNA methylation process occurs, establishing a new sex-specific pattern in developing gametes (Lee et al., 2002). This de novo methylation process differs between male and female germ lines. In the male germ line, the process is initiated in the foetus; hence, the paternal allele is hypermethylated at birth in this cell lineage (Davis et al., 2000). In the female germ line by contrast, the process is arrested during meiosis in the foetal period, and de novo methylation begins only after birth during folliculogenesis/oogenesis, which is around puberty (Obata et al., 1998; Fagundes et al., 2011). A follicle is recruited for growth, and de novo DNA methylation is initiated in the oocytes. However, the process is not completed without the aid of appropriate hormonal stimuli.
During embryogenesis, the parental pronucleus undergoes a differential demethylation process, where the paternal genome is significantly demethylated by an active mechanism closely following fertilization (Oswald et al., 2000). The maternal genome, however, loses DNA methylation at a later stage due to cleavage divisions through a passive mechanism (Sasaki and Matsui, 2008). After DNA demethylation, global de novo methylation begins at the 8-16 cell stage in cattle (Dean et al., 2001; Ivanova et al., 2020). At the blastocyst stage, where several methylome analyses takes place, de novo methylation has been started; however, a wide range of reprogramming continues until the establishment of DNA methylation patterns in the embryonic and extra-embryonic tissue (Greenberg and Bourc'his, 2019). Therefore, several mechanisms of epigenetic remodeling still happen from the blastocyst stage until the establishment of the tissue epigenome of the adult animal (Xu et al., 2021). Now, how informative could the methylome of the embryos be if the DNA methylation patterns are analysed before the structures have been completely reprogrammed?
One of the main reasons for the increased interest in embryonic DNA methylation is the Developmental Origins of Health and Disease (DOHaD) study and the long-term consequences for the progeny, which is a crucial concern for humans (Almeida et al., 2019; Lapehn and Paquette, 2022). The DOHaD theory states that the foetus undergoes an intrauterine environmental adaptation process in order to cope with those same conditions following birth (Wadhwa et al., 2009; Moreno-Fernandez et al., 2020). Therefore, adverse environmental stimuli during foetal programming can affect the establishment of epigenetic marks. As the offspring may not face the same conditions after birth, these adaptions can lead to susceptibility to diseases in adulthood (Cropley et al., 2006; Martinez et al., 2014).
As a result of several studies in humans and animal models, the maternal diet during early pregnancy is known to affect the embryo and long-term conceptus (Tobi et al., 2015; Wang et al., 2015; Finer et al., 2016; Knight et al., 2018; Serrano-Perez et al., 2020). We found that low levels of sulphur and cobalt in the diet offered to oocyte donors altered the DNA methylome of bovine embryos (Nochi et al., 2022). The inheritance of differential methylation regions (DMRs) by the next generation is known as an intergenerational epigenetic inheritance (Skvortsova et al., 2018). Analysing embryonic methylomes can be helpful for the analysis of epigenetic inheritance as an initial screening strategy. However, caution must be exercised when considering the information extracted from these data and when projecting the embryonic methylome onto adult tissues (Nochi et al., 2022).
Thus, this study tests the hypothesis that the DMRs identified in the embryos are not maintained in the somatic tissues of the animal in adulthood, considering that the embryos have a high probability of losing these DMR patterns during the second wave of epigenetic reprogramming, which occurs during early development (Reik et al., 2001). Accordingly, we aimed to characterize four DMRs in genes, which were previously identified in embryos and are involved in the epigenetic machinery, in the blood and sperm of adult progenies of two groups of heifers used in a related previous study in our laboratory (Nochi et al., 2022). *These* genes of special interest are related to histone methylation: writer lysine methyltransferase 2D (KMT2D), DOT1-like histone lysine methyltransferase (DOT1L), erasers lysine demethylase 2A (KDM2A), and lysine demethylase 5A (KDM5A).
## Ethical approval
This experimental study has been approved by the Ethics Committee on Animal Use (CEUA-Protocol n° $\frac{98}{2010}$), School of Veterinary Medicine and Animal Science, Universidade Estadual Paulista “Júlio de Mesquita Filho.”
## Animals and experimental diets
In this study, we used animals from a previous study conducted in our laboratory (Nochi et al., 2022). Briefly, the heifers were separated into groups with different diets, the control and the group with low sulfur and cobalt (low S/Co). The respective diets were offered to the animals for six months (pre- and periconceptional periods). At the end of the experiment, the heifers were inseminated with the same bull used for the in vitro embryo production (IVP) performed in Nochi et al. [ 2022]. Among the progeny of those heifers, we collected the blood and sperm of the bulls in adulthood.
## Sample collection
The blood and sperm cells were collected from six bulls (Bull 1, Bull 2, Bull 3, Bull 4, Bull 5, and Bull 6) — the progenies of heifers (two from the control and four from the low S/Co group). Semen from six Nellore bulls (*Bos taurus* indicus) was collected from the ejaculate via electroejaculation. Sperm quality, concentration, motility, plasma membrane integrity, and morphology were evaluated (Table 1). Semen samples were stored in liquid nitrogen (-196 ºC) until DNA isolation was performed.
**Table 1**
| Unnamed: 0 | Animals | Animals.1 | Animals.2 | Animals.3 | Animals.4 | Animals.5 |
| --- | --- | --- | --- | --- | --- | --- |
| | Bull 1 | Bull 2 | Bull 3 | Bull 4 | Bull 5 | Bull 6 |
| Concentration (×106/mL) | 1710 | 1510 | 1250 | 2920 | 650 | 1750 |
| Total motility (%) | 20% | 90% | 70% | 90% | 90% | 90% |
| Plasma membrane integrity (%) | 43% | 79% | 82% | 66,5% | 78% | 61% |
| Sperm normal morphology (%) | 31% | 80,5% | 66,5% | 67,5% | 64,5% | 67% |
## DNA isolation
Genomic DNA was isolated from white blood cells using the DNeasy Blood & Tissue Kit (Qiagen, CA, USA) according to the manufacturer’s instructions. Sperm DNA was isolated using a protocol based on salting out as described by Carvalho et al. [ 2012]. The DNA samples were stored at -20 °C for sodium bisulphite treatment. The quality of the DNA samples was evaluated using agarose gel electrophoresis.
## Sodium bisulphite treatment
Blood and sperm genomic DNA (500 ng) were treated with sodium bisulphite using the EZ DNA Methylation-Lightning kit (Zymo Research, Irvine, CA, USA), according to the manufacturer’s instructions. Sodium bisulphite-treated DNA were stored at -80 °C until PCR amplification was performed.
## Bisulphite PCR
PCR was performed to amplify the DMRs in the genes KDM2A, KDM5A, KMT2D, and DOT1L, which are associated with histone-active methylation marks (Nochi et al., 2022). Primers were designed using the MethPrimer (Li and Dahiya, 2002) and Bisulphite Primer Seeker software (Zymo Research) to flank the DMRs, which were located on CpG islands in all genes except KDM5A (Figure 1).
**Figure 1:** *Representation of the KDM2A, KDM5A, KMT2D, and DOT1L gene structures, GC content, and CpG island prediction. Green bars represent the input sequence; below, blue lines represent introns, blue arrows represent exons, and orange arrows represent primer positions. The GC content and CpG islands are predicted for each gene. The graphs were generated using Geneious v2020.0.5 (Biomatters, Auckland, New Zealand).*
The primer sequences, GenBank accession numbers, number of CpG sites, amplicon sizes, and annealing temperatures are listed in Table 2. The total volume of each reaction prepared was 20 μL and comprised of 1× Taq buffer, 1.5 mM MgCl2, 0.4 mM dNTPs, 1 U Platinum™ Taq polymerase (Invitrogen, CA, USA), 0.5 μM primers (forward and reverse), and 2 μL of bisulphite-treated DNA. PCR was performed with an initial denaturing step at 94 °C for 3 min, followed by 29 cycles of 94 °C for 40 s, annealing (Table 2) for 1 min, and 72 °C for 1 min. The final extension was at 72 °C for 15 min. After PCR, amplicons were purified from agarose gels using the Wizard® SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA), according to the manufacturer’s instructions.
**Table 2**
| Gene | Unnamed: 1 | Primer Sequence (5’-3’) | Genbank acession number | CpG sites | Amplicon length (bp) | Annealing (°C) |
| --- | --- | --- | --- | --- | --- | --- |
| KDM2A | F: | GGTAAGTGTAGAGGGTTTTGAAGAAAGGAGATATTG | 540141 | 28 | 387 | 60 |
| KDM2A | R: | TTAACTTTCTCAACTTCAAACAACTCCTTTTTACC | 540141 | 28 | 387 | 60 |
| KDM5A | F: | AAATTGGTTAAGAAGTTAGTAAAAGAAGAAGAGAG | 507962 | 13 | 334 | 55 |
| KDM5A | R: | ATAATACAAAACCAAATCCTAAAATCAAAACAAACC | 507962 | 13 | 334 | 55 |
| KMT2D | F: | TAGTTAGAGTGGAGTAGATTTTGTGGGGTTT | 506805 | 11 | 333 | 60 |
| KMT2D | R: | CACAACTAAAAACCAAACTACCCCCTTATC | 506805 | 11 | 333 | 60 |
| DOT1L | F: | GTTATGGGTATTTTTTAGGTTGGTGGTTG | 510442 | 19 | 335 | 60 |
| DOT1L | R: | TACAAAATAAAAACCATATTCCAAACCCAC | 510442 | 19 | 335 | 60 |
## Cloning and bisulphite sequencing
The purified amplicons were cloned into the TOPO TA Cloning® vector (Invitrogen, CA, USA) and transferred into DH5α cells using a heat shock procedure. Plasmid DNA was isolated using Pure Yield Plasmid Miniprep (Promega, Madison, WI, USA), and individual clones were sequenced using BigDye® cycle sequencing chemistry and an ABI3100 automated sequencer (Applied Biosystems, Foster City, CA). Electropherogram quality was analysed using Chromas® (Technelysium Pty Ltd, South Brisbane, Australia), and methylation patterns were processed using the QUantification tool for Methylation Analysis (QUMA) (Kumaki et al., 2008). DNA sequences were compared with GenBank reference sequences (Table 2), and only those sequences originating from clones with ≥ $95\%$ identity and ≥ $97\%$ cytosine conversion were used in the analysis ($$n = 684$$). The efficiency of the bisulphite treatment was calculated based on the percentage of CpH (H = A, C, or T) site conversion divided by the total number of CpH sites in the sequence.
## Statistics analysis
Comparison of methylation data between two groups was done using the Mann-Whitney test and more than two groups were performed using the Kruskal-Wallis test followed by Dunn’s multiple comparison test. Comparative methylation analysis of CpG site was performed using Fisher's exact test. Statistical significance was set at $p \leq 0.05.$ Data analyses were performed using QUMA and GraphPad Prism software.
## Results
Overall, we analyzed 688 clones and compared the DNA methylation patterns of the four genes KDM2A, KDM5A, KMT2D, and DOT1L (detected in the blood and sperm of six Nellore bulls) in the control group against that of the low S/Co groups. The DNA methylation levels (Figures2-5) were classified as low (0-$20\%$), moderate (21-$50\%$), and high (51-$100\%$) according to Zhang et al. [ 2016] and Silveira et al. [ 2018].
**Figure 2:** *DNA methylation profile of KDM2A gene in blood and sperm for control and low S/Co groups. (A) Blood samples, (B) Sperm samples, and (C) Comparative analysis of methylation by CpG sites between control and low S/Co in blood and sperm. Each line represents an individual DNA clone, and each circle represents a CpG dinucleotide. Black circles represent methylated cytosines and white circles represent unmethylated cytosines. The DNA methylation percentage for each animal (Bull 1, Bull 2, Bull 3, Bull 4, Bull 5, and Bull 6) is represented as mean ± standard deviation of the mean. Differences in DNA methylation among animals within the same group are shown by letters a and b (p < 0.05). (*) represents significant difference in the mean values for methylation of individual CpGs using Fisher's exact test (p<0.05). (n) represents the number of sequenced alleles of each sample.* **Figure 5:** *DNA methylation profile of DOT1L gene in blood and sperm for control and low S/Co groups. (A) Blood samples, (B) Sperm samples, and (C) Comparative analysis of methylation by CpG sites between control and low S/Co in blood and sperm. Each line represents an individual DNA clone, and each circle represents a CpG dinucleotide. Black circles represent methylated cytosines and white circles represent unmethylated cytosines. The DNA methylation percentage for each animal (Bull 1, Bull 2, Bull 3, Bull 4, Bull 5, and Bull 6) is represented as mean ± standard deviation of the mean. Differences in DNA methylation among animals within the same group are shown by letters a and b (p < 0.05). (*) represents significant difference in the mean values for methylation of individual CpGs using Fisher's exact test (p<0.05). (n) represents the number of sequenced alleles of each sample.*
*In* general, a hypermethylated pattern was observed in DNA isolated from both the blood and sperm for all genes, groups, and animals (Figures 2-5). However, the KDM2A and KMT2D genes of three animals showed a lower methylated pattern [KDM2A/sperm/low S/Co/Bull 4 ($57.7\%$), Figure 2B; KMT2D/blood/low S/Co/Bull 6 ($60.2\%$), Figure 4A; and KMT2D/sperm/control/Bull 2 ($54.5\%$), Figure 4B] as the same animal showed alleles with $100\%$ and $0\%$ methylation. We also found more variation in the methylation profile among the sperm alleles in other animals [KDM5A/sperm/low S/Co/Bull 3 ($75.8\%$), Figure 3B; KMT2D/sperm/low S/Co/Bull 5 ($94.1\%$) and Bull 6 ($79.4\%$), Figure 4B; DOT1L/sperm/control/Bull 2 ($86\%$), Figure 5B; DOT1L/sperm/low S/Co/Bull 3 ($88.4\%$) and Bull 5 ($90.1\%$), Figure 5B], but it did not influence the higher methylation pattern.
**Figure 4:** *DNA methylation profile of KMT2D gene in blood and sperm for control and low S/Co groups. (A) Blood samples, (B) Sperm samples, and (C) Comparative analysis of methylation by CpG sites between control and low S/Co in blood and sperm. Each line represents an individual DNA clone, and each circle represents a CpG dinucleotide. Black circles represent methylated cytosines and white circles represent unmethylated cytosines. The DNA methylation percentage for each animal (Bull 1, Bull 2, Bull 3, Bull 4, Bull 5, and Bull 6) is represented as mean ± standard deviation of the mean. Differences in DNA methylation among animals within the same group are shown by letters a and b (p < 0.05). (n) represents the number of sequenced alleles of each sample.* **Figure 3:** *DNA methylation profile of KDM5A gene in blood and sperm for control and low S/Co groups. (A) Blood samples, (B) Sperm samples, and (C) Comparative analysis of methylation by CpG sites between control and low S/Co in blood and sperm. Each line represents an individual DNA clone, and each circle represents a CpG dinucleotide. Black circles represent methylated cytosines and white circles represent unmethylated cytosines. The DNA methylation percentage for each animal (Bull 1, Bull 2, Bull 3, Bull 4, Bull 5, and Bull 6) is represented as mean ± standard deviation of the mean. Differences in DNA methylation among animals within the same group are shown by letters a and b (p < 0.05). (*) represents significant difference in the mean values for methylation of individual CpGs using Fisher's exact test (p<0.05). (n) represents the number of sequenced alleles of each sample.*
Interestingly, when we compared the methylation status of each CpG site individually, we found specific CpGs differentially methylated between control and low S/Co for KDM2A in the blood (CpG 17) and sperm (CpG 25) (Figure 2C), for KDM5A in sperm (CpG 4) (Figure 3C), and DOT1L in sperm (Figure 5C). However, when we compared all CpG sites among themselves, there were no differences in DNA methylation patterns for any of the genes between the control and low S/Co groups in the blood or sperm samples (Figure 6A). Therefore, treatment with a low S/Co diet in the heifers during the pre-and periconceptional periods did not affect the DNA methylation pattern of the gamete and blood cells of the progeny in adulthood for the DMRs analyzed.
**Figure 6:** *Percentage of methylation in KDM2A, KDM5A, KMT2D, and DOT1L genes (A) Comparison of DNA methylation levels between control and low S/Co groups for blood and sperm samples. (B) Comparison of DNA methylation levels in blood and sperm samples in the control and low S/Co groups, respectively. Numbers represent significant differences in the mean values for methylation using the Mann-Whitney test (p ≤ 0.05)*
We also analyzed DNA methylation in the blood and sperm tissues of the control and low S/Co groups and found a difference in DNA methylation only in the control groups for KDM2A (Figure 6B); however, despite the statistical difference between the tissues analyzed, all samples were considered hypermethylated.
## Discussion
In mammals, extensive epigenomic remodelling occurs during the initial stages of development. During gametogenesis and embryogenesis, epigenetic marks are more vulnerable to environmental influences. In our previous study, we presented DMR candidates for investigation, focusing on the impact of maternal nutrition on foetal epigenetic reprogramming during the pre- and peri-conceptional periods (Nochi et al., 2022). Therefore, to validate whether DMRs in blastocysts are maintained in adulthood, we characterized four DMRs in the sperm and blood from F1 animals.
Our previous study applied the experimental diet during the de novo methylation phase of F0 gametogenesis after the animals reached puberty (Nochi et al., 2022). Although Nochi et al. [ 2022] found an altered DNA methylation pattern in the blastocyst stage between the low S/Co and control groups in their study, we identified a hypermethylated pattern for both groups in all the DMRs analyzed in both the blood and sperm DNA of F1. This result suggests that embryos from both groups reprogrammed their epigenetic profiles correctly in the blood and sperm cells during development. Thus, epigenetic reprogramming during embryogenesis prevents the transmission of F0 gametic epimutations to F1.
Despite a second wave of epigenetic reprogramming, some regions are not reprogrammed during embryogenesis. The DMRs established during gametogenesis are known as germ line DMRs (gDMRs). Those that are reprogrammed are known as transient DMRs (tDMRs) (Proudhon et al., 2012; Smallwood and Kelsey, 2012). In contrast, the imprinted DMRs (iDMRs) are those DMRs that are protected from loss of methylation after fertilization and are not methylated during embryo or tissue differentiation (Proudhon et al., 2012; MacDonald and Mann, 2014; Thakur et al., 2016). In our study model, the diet on final gametogenesis did not exert a permanent effect in the DMRs studied. However, since iDMRs are protected from reprogramming during embryogenesis, if the diet had affected these DMRs in any way, those effects probably would have been retained into adulthood to create metastable epialleles.
In addition to evaluating DNA methylation patterns in white blood cells in F1 adults, we also analysed DNA methylation in the sperm cells of these animals. Despite the hypermethylated state in the blood and sperm samples, we found more variation in DNA methylation patterns among sperm alleles. Interestingly, several studies in humans and cattle have described the potential use of sperm DNA methylation-epimutations as biomarkers of infertility and susceptibility to diseases (Kropp et al., 2017; Nasri et al., 2017; Capra et al., 2019; Lujan et al., 2019; Garrido et al., 2021). Thus, further studies characterizing whether maternal diet can influence the sperm DNA methylation of the offspring will provide valuable information.
A previous study in humans revealed some sensitive environmental hotspots in the embryonic methylome (Silver et al., 2022). In contrast, a low S/Co diet administered during gametogenesis stochastically affected the embryonic methylome (Nochi et al., 2022). Thus, the regions of the embryonic epigenome that are impacted by the dietary effects may be reprogrammed without deleterious changes in the offspring.
It is well known that the diet during pregnancy may affect the offspring. Several studies have confirmed the effect of different maternal diets on the offspring in humans (Roseboom et al., 2006), mice (Guo et al., 2018; Mao et al., 2018; Xie et al., 2018), rats (Carlin et al., 2019; Pedrana et al., 2020), and bovines (Devos et al., 2021; Liu et al., 2021; Noya et al., 2021). Moreover, studies reported that the maternal diet during gestation affects the DNA methylation pattern in the placenta and offspring of mice (Ge et al., 2014; Mahajan et al., 2019; Zhang et al., 2021a), cattle (Liu et al., 2021), and humans (Daniels et al., 2020; Kupers et al., 2022). Interestingly, a study showed that the exposure of IVP embryos to choline in the culture medium alters the DNA methylation profile in the offspring muscle (Estrada-Cortes et al., 2021). However, these studies evaluated the effects during embryogenesis. DMRs, by contrast, are more likely to propagate in the tissue of the offspring when the stimuli affect epigenetic reprogramming beyond the stage of gametogenesis but also embryogenesis. Therefore, determining the time and duration to which the dietary stimuli exert its effect is essential to study and understand the consequences of the maternal diet on the offspring.
Interestingly, studies in livestock have only evaluated the dietary effects during the final stages of de novo methylation (Sinclair et al., 2007; Zglejc-Waszak et al., 2019; Toschi et al., 2020), but Nochi et al. [ 2022] evaluated the impact of nutrition starting from the initial stage of de novo methylation during gametogenesis. A broad experimental design for the study of environmental influence on gametogenesis should contemplate the erasure of DNA methylation during foetal programming and ensure the dietary effect on oocytes during de novo methylation. However, this experimental design is easier to implement in mice models than it is in cattle because of its expensive and time-consuming nature. Moreover, based on our observations, embryonic methylome can be used only as an initial screening tool because DMRs may be reprogrammed in the final stages of de novo methylation during embryogenesis and foetal growth; therefore, it is crucial to validate the DMRs in adulthood.
## Conclusion
In this study, we characterized the DMRs identified in the previous experiment, which showed that the pre-and periconceptional diet affected the DNA methylation profile of embryos. Among the 2,320 DMRs identified in blastocysts by Nochi et al. [ 2022], the six that we analyzed underwent extensive epigenetic reprogramming in both blood and sperm cells. These results confirm our hypothesis that the DMRs found in embryos may not be maintained in adult animals. Thus, we suggest that after the first screening using WGBS, it is crucial to confirm the inheritance before projecting the embryonic methylome onto adult tissues.
## References
1. Almeida DL, Pavanello A, Saavedra LP, Pereira TS, Castro-Prado MAA, Mathias PCF. **Environmental monitoring and the developmental origins of health and disease**. *J Dev Orig Health Dis* (2019) **10** 608-615. DOI: 10.1017/S2040174419000151
2. Canovas S, Ross PJ, Kelsey G, Coy P. **DNA Methylation in Embryo Development: Epigenetic Impact of ART (Assisted Reproductive Technologies)**. *BioEssays* (2017) **39**. DOI: 10.1002/bies.201700106
3. Capra E, Lazzari B, Turri F, Cremonesi P, Portela AMR, Ajmone-Marsan P, Stella A, Pizzi F. **Epigenetic analysis of high and low motile sperm populations reveals methylation variation in satellite regions within the pericentromeric position and in genes functionally related to sperm DNA organization and maintenance in Bos taurus**. *BMC Genomics* (2019) **20** 940. DOI: 10.1186/s12864-019-6317-6
4. Carlin G, Chaumontet C, Blachier F, Barbillon P, Darcel N, Blais A, Delteil C, Guillin FM, Blat S, van der Beek EM, Kodde A, Tome D, Davila AM. **Maternal high-protein diet during pregnancy modifies rat offspring body weight and insulin signalling but not macronutrient preference in adulthood**. *Nutrients* (2019) **11** 96. DOI: 10.3390/nu11010096
5. Carvalho JO, Michalczechen‐Lacerda VA, Sartori R, Rodrigues FC, Bravim O, Franco MM, Dode MA. **The methylation patterns of the IGF2 and IGF2R genes in bovine spermatozoa are not affected by flow‐cytometric sex sorting**. *Mol Reprod Dev* (2012) **79** 77-84. DOI: 10.1002/mrd.21410
6. Cropley JE, Suter CM, Beckman KB, Martin DIK. **Germ-line epigenetic modification of the murine Avy allele by nutritional supplementation**. *Proc Natl Acad Sci USA* (2006) **103** 17308-17312. DOI: 10.1073/pnas.0607090103
7. Daniels TE, Sadovnikoff AI, Ridout KK, Lesseur C, Marsit CJ, Tyrka AR. **Associations of maternal diet and placenta leptin methylation**. *Mol Cell Endocrinol* (2020) **505**. DOI: 10.1016/j.mce.2020.110739
8. Davis TL, Yang GJ, McCarrey JR, Bartolomei MS. **The H19 methylation imprint is erased and re‐established differentially on the parental alleles during male germ cell development**. *Hum Mol Genet* (2000) **9** 2885-2894. DOI: 10.1093/hmg/9.19.2885
9. Dean W, Santos F, Stojkovic M, Zakhartchenko V, Walter J, Wolf J, Reik W. **Conservation of methylation reprogramming in mammalian development: aberrant reprogramming in cloned embryos**. *Proc Natl Acad Sci USA* (2001) **98** 13734-13738. DOI: 10.1073/pnas.241522698
10. Devos J, Behrouzi A, Paradis F, Straathof C, Li C, Colazo M, Block H, Fitzsimmons C. **Genetic potential for residual feed intake and diet fed during early- to mid-gestation influences post-natal DNA methylation of imprinted genes in muscle and liver tissues in beef cattle**. *J Anim Sci* (2021) **99**. DOI: 10.1093/jas/skab140
11. Duan JE, Jiang ZC, Alqahtani F, Mandoiu I, Dong H, Zheng X, Marjani SL, Chen J, Tian XC. **Methylome dynamics of bovine gametes and in vivo early embryos**. *Front Genet* (2019) **10** 512. DOI: 10.3389/fgene.2019.00512
12. Estrada-Cortes E, Ortiz W, Rabaglino MB, Block J, Rae O, Jannaman EA, Xiao Y, Hansen PJ. **Choline acts during preimplantation development of the bovine embryo to program postnatal growth and alter muscle DNA methylation**. *FASEB J* (2021) **35**. DOI: 10.1096/fj.202100991R
13. Fagundes NS, Michalczechen-Lacerda VA, Caixeta ES, Machado GM, Rodrigues FC, Melo EO, Dode MA, Franco MM. **Methylation status in the intragenic differentially methylated region of the IGF2 locus in Bos taurus indicus oocytes with different developmental competencies**. *Mol Hum Reprod* (2011) **17** 85-91. DOI: 10.1093/molehr/gaq075
14. Finer S, Iqbal MS, Lowe R, Ogunkolade BW, Pervin S, Mathews C, Smart M, Alam DS, Hitman GA. **Is famine exposure during developmental life in rural Bangladesh associated with a metabolic and epigenetic signature in young adulthood? A historical cohort study**. *BMJ Open* (2016) **6**. DOI: 10.1136/bmjopen-2016-011768
15. Garrido N, Cruz F, Egea RR, Simon C, Sadler-Riggleman I, Beck D, Nilsson E, Ben Maamar M, Skinner MK. **Sperm DNA methylation epimutation biomarker for paternal offspring autism susceptibility**. *Clin Epigenetics* (2021) **13** 6. DOI: 10.1186/s13148-020-00995-2
16. Ge ZJ, Luo SM, Lin F, Liang QX, Huang L, Wei YC, Hou Y, Han ZM, Schatten H, Sun QY. **DNA methylation in oocytes and liver of female mice and their offspring: effects of high-fat-diet-induced obesity**. *Environ Health Perspect* (2014) **122** 159-164. DOI: 10.1289/ehp.1307047
17. Greenberg MVC, Bourc’his D. **The diverse roles of DNA methylation in mammalian development and disease**. *Nat Rev Mol Cell Biol* (2019) **20** 590-607. DOI: 10.1038/s41580-019-0159-6
18. Guo H, Zhu P, Yan L, Li R, Hu B, Lian Y, Yan J, Ren X, Lin S, Li J, Jin X, Shi X, Liu P, Wang X, Wang W, Wei Y, Li X, Guo F, Wu X, Fan X, Yong J, Wen L, Xie SX, Tang F, Qiao J. **The DNA methylation landscape of human early embryos**. *Nature* (2014) **511** 606-610. DOI: 10.1038/nature13544
19. Guo Y, Wang Z, Chen L, Tang L, Wen S, Liu Y, Yuan J. **Diet induced maternal obesity affects offspring gut microbiota and persists into young adulthood**. *Food Funct* (2018) **9** 4317-4327. DOI: 10.1039/C8FO00444G
20. Ivanova E, Canovas S, Garcia-Martinez S, Romar R, Lopes JS, Rizos D, Sanchez-Calabuig MJ, Krueger F, Andrews S, Perez-Sanz F, Kelsey G, Coy P. **DNA methylation changes during preimplantation development reveal inter-species differences and reprogramming events at imprinted genes**. *Clin Epigenetics* (2020) **12** 64. DOI: 10.1186/s13148-020-00857-x
21. Kafri T, Ariel M, Brandeis M, Shemer R, Urven L, McCarrey J, Cedar H, Razin A. **Developmental pattern of gene-specific DNA methylation in the mouse embryo and germ line**. *Genes Dev* (1992) **6** 705-714. DOI: 10.1101/gad.6.5.705
22. Kang Y-K, Koo D-B, Park J-S, Choi Y-H, Chung A-S, Lee K-K, Han Y-M. **Aberrant methylation of donor genome in cloned bovine embryos**. *Nat Genet* (2001) **28** 173-177. DOI: 10.1038/88903
23. Knight AK, Park HJ, Hausman DB, Fleming JM, Bland VL, Rosa G, Kennedy EM, Caudill MA, Malysheva O, Kauwell GPA, Sokolow A, Fisher S, Smith AK, Bailey LB. **Association between one-carbon metabolism indices and DNA methylation status in maternal and cord blood**. *Sci Rep* (2018) **8** 16873. DOI: 10.1038/s41598-018-35111-1
24. Kropp J, Carrillo JA, Namous H, Daniels A, Salih SM, Song J, Khatib H. **Male fertility status is associated with DNA methylation signatures in sperm and transcriptomic profiles of bovine preimplantation embryos**. *BMC Genomics* (2017) **18** 280. DOI: 10.1186/s12864-017-3673-y
25. Kumaki Y, Oda M, Okano M. **QUMA: quantification tool for methylation analysis**. *Nucleic Acids Res* (2008) **36** W170-5. DOI: 10.1093/nar/gkn294
26. Kupers LK, Fernandez-Barres S, Nounu A, Friedman C, Fore R, Mancano G, Dabelea D, Rifas-Shiman SL, Mulder RH, Oken E, Johnson L, Bustamante M, Jaddoe VWV, Hivert MF, Starling AP, de Vries JHM, Sharp GC, Vrijheid M, Felix JF. **Maternal Mediterranean diet in pregnancy and newborn DNA methylation: a meta-analysis in the PACE Consortium**. *Epigenetics* (2022) **17** 1419-1431. DOI: 10.1080/15592294.2022.2038412
27. Lapehn S, Paquette AG. **The placental epigenome as a molecular link between prenatal exposures and fetal health outcomes through the DOHaD hypothesis**. *Curr Environ Health Rep* (2022) **9** 490-501. DOI: 10.1007/s40572-022-00354-8
28. Lee J, Inoue K, Ono R, Ogonuki N, Kohda T, Kaneko-Ishino T, Ogura A, Ishino F. **Erasing genomic imprinting memory in mouse clone embryos produced from day 11.5 primordial germ cells**. *Development* (2002) **129** 1807-1817. DOI: 10.1242/dev.129.8.1807
29. Li L-C, Dahiya R. **MethPrimer: designing primers for methylation PCRs**. *Bioinformatics* (2002) **18** 1427-1431. DOI: 10.1093/bioinformatics/18.11.1427
30. Liu L, Amorin R, Moriel P, DiLorenzo N, Lancaster PA, Penagaricano F. **Maternal methionine supplementation during gestation alters alternative splicing and DNA methylation in bovine skeletal muscle**. *BMC Genomics* (2021) **22** 780. DOI: 10.1186/s12864-021-08065-4
31. Lujan S, Caroppo E, Niederberger C, Arce JC, Sadler-Riggleman I, Beck D, Nilsson E, Skinner MK. **Sperm DNA Methylation Epimutation Biomarkers for Male Infertility and FSH Therapeutic Responsiveness**. *Sci Rep* (2019) **9** 16786. DOI: 10.1038/s41598-019-52903-1
32. MacDonald WA, Mann MR. **Epigenetic regulation of genomic imprinting from germ line to preimplantation**. *Mol Reprod Dev* (2014) **81** 126-140. DOI: 10.1002/mrd.22220
33. Mahajan A, Sapehia D, Thakur S, Mohanraj PS, Bagga R, Kaur J. **Effect of imbalance in folate and vitamin B12 in maternal/parental diet on global methylation and regulatory miRNAs**. *Sci Rep* (2019) **9** 17602. DOI: 10.1038/s41598-019-54070-9
34. Mao J, Pennington KA, Talton OO, Schulz LC, Sutovsky M, Lin Y, Sutovsky P. **In utero and postnatal exposure to high fat, high sucrose diet suppressed testis apoptosis and reduced sperm count**. *Sci Rep* (2018) **8** 7622. DOI: 10.1038/s41598-018-25950-3
35. Martinez D, Pentinat T, Ribo S, Daviaud C, Bloks VW, Cebria J, Villalmanzo N, Kalko SG, Ramon-Krauel M, Diaz R, Plosch T, Tost J, Jimenez-Chillaron JC. **In utero undernutrition in male mice programs liver lipid metabolism in the second-generation offspring involving altered Lxra DNA methylation**. *Cell Metab* (2014) **19** 941-951. DOI: 10.1016/j.cmet.2014.03.026
36. Moreno-Fernandez J, Ochoa JJ, Lopez-Frias M, Diaz-Castro J. **Impact of early nutrition, physical activity and sleep on the fetal programming of disease in the pregnancy: a narrative review**. *Nutrients* (2020) **12** 3900. DOI: 10.3390/nu12123900
37. Nasri F, Gharesi-Fard B, Namavar Jahromi B, Farazi-Fard MA, Banaei M, Davari M, Ebrahimi S, Anvar Z. **Sperm DNA methylation of H19 imprinted gene and male infertility**. *Andrologia* (2017) **49**. DOI: 10.1111/and.12766
38. Nochi ARF, Vargas LN, Sartori R, Junior RG, Araujo DB, Figueiredo RA, Togawa RC, Costa MMC, Grynberg P, Mendonca AS, Kussano NR, Pivato I, Silva BDM, Spricigo JFW, Leme LO, da Silva JP, Caetano AR, Dode MAN, Franco MM. **Low levels of sulfur and cobalt during the pre- and periconceptional periods affect the oocyte yield of donors and the DNA methylome of preimplantation bovine embryos**. *J Dev Orig Health Dis* (2022) **13** 231-243. DOI: 10.1017/S2040174421000222
39. Noya A, Ripoll G, Casasus I, Sanz A. **Long-term effects of early maternal undernutrition on the growth, physiological profiles, carcass and meat quality of male beef offspring**. *Res Vet Sci* (2021) **142** 1-11. DOI: 10.1016/j.rvsc.2021.10.025
40. Obata Y, Kaneko-Ishino T, Koide T, Takai Y, Ueda T, Domeki I, Shiroishi T, Ishino F, Kono T. **Disruption of primary imprinting during oocyte growth leads to the modified expression of imprinted genes during embryogenesis**. *Development* (1998) **125** 1553-1560. DOI: 10.1242/dev.125.8.1553
41. Oswald J, Engemann S, Lane N, Mayer W, Olek A, Fundele R, Dean W, Reik W, Walter J. **Active demethylation of the paternal genome in the mouse zygote**. *Curr Biol* (2000) **10** 475-478. DOI: 10.1016/S0960-9822(00)00448-6
42. Pedrana G, Viotti H, Lombide P, Cavestany D, Martin GB, Vickers MH, Sloboda DM. **Maternal undernutrition during pregnancy and lactation affects testicular morphology, the stages of spermatogenic cycle, and the testicular IGF-I system in adult offspring**. *J Dev Orig Health Dis* (2020) **11** 473-483. DOI: 10.1017/S2040174420000306
43. Proudhon C, Duffie R, Ajjan S, Cowley M, Iranzo J, Carbajosa G, Saadeh H, Holland ML, Oakey RJ, Rakyan VK, Schulz R, Bourc’his D. **Protection against de novo methylation is instrumental in maintaining parent-of-origin methylation inherited from the gametes**. *Mol Cell* (2012) **47** 909-920. DOI: 10.1016/j.molcel.2012.07.010
44. Reik W, Dean W, Walter J. **Epigenetic reprogramming in mammalian development**. *Science* (2001) **293** 1089-1093. DOI: 10.1126/science.1063443
45. Roseboom T, de Rooij S, Painter R. **The Dutch famine and its long-term consequences for adult health**. *Early Hum Dev* (2006) **82** 485-491. DOI: 10.1016/j.earlhumdev.2006.07.001
46. Sasaki H, Matsui Y. **Epigenetic events in mammalian germ-cell development: reprogramming and beyond**. *Nat Rev Genet* (2008) **9** 129-140. DOI: 10.1038/nrg2295
47. Serrano-Perez B, Molina E, Noya A, Lopez-Helguera I, Casasus I, Sanz A, Villalba D. **Maternal nutrient restriction in early pregnancy increases the risk of late embryo loss despite no effects on peri-implantation interferon-stimulated genes in suckler beef cattle**. *Res Vet Sci* (2020) **128** 69-75. DOI: 10.1016/j.rvsc.2019.10.023
48. Silveira MM, Bayão HXS, dos Santos Mendonça A, Borges NA, Vargas LN, Caetano AR, Rumpf R, Franco MM. **DNA methylation profile at a satellite region is associated with aberrant placentation in cloned calves**. *Placenta* (2018) **70** 25-33. DOI: 10.1016/j.placenta.2018.08.007
49. Silver MJ, Saffari A, Kessler NJ, Chandak GR, Fall CHD, Issarapu P, Dedaniya A, Betts M, Moore SE, Routledge MN, Herceg Z, Cuenin C, Derakhshan M, James PT, Monk D, Prentice AM. **Environmentally sensitive hotspots in the methylome of the early human embryo**. *eLife* (2022) **11**. DOI: 10.7554/eLife.72031
50. Sinclair KD, Allegrucci C, Singh R, Gardner DS, Sebastian S, Bispham J, Thurston A, Huntley JF, Rees WD, Maloney CA, Lea RG, Craigon J, McEvoy TG, Young LE. **DNA methylation, insulin resistance, and blood pressure in offspring determined by maternal periconceptional B vitamin and methionine status**. *Proc Natl Acad Sci USA* (2007) **104** 19351-19356. DOI: 10.1073/pnas.0707258104
51. Skvortsova K, Iovino N, Bogdanovic O. **Functions and mechanisms of epigenetic inheritance in animals**. *Nat Rev Mol Cell Biol* (2018) **19** 774-790. DOI: 10.1038/s41580-018-0074-2
52. Smallwood SA, Kelsey G. **De novo DNA methylation: a germ cell perspective**. *Trends Genet* (2012) **28** 33-42. DOI: 10.1016/j.tig.2011.09.004
53. Stevens ME, Maidens PM, Robinson ES, Vandeberg JL, Pedersen RA, Monk M. **DNA methylation in the developing marsupial embryo**. *Development* (1988) **103** 719-724. DOI: 10.1242/dev.103.4.719
54. Thakur A, Mackin SJ, Irwin RE, O’Neill KM, Pollin G, Walsh C. **Widespread recovery of methylation at gametic imprints in hypomethylated mouse stem cells following rescue with DNMT3A2**. *Epigenetics Chromatin* (2016) **9** 53. DOI: 10.1186/s13072-016-0104-2
55. Tobi EW, Slieker RC, Stein AD, Suchiman HE, Slagboom PE, van Zwet EW, Heijmans BT, Lumey LH. **Early gestation as the critical time-window for changes in the prenatal environment to affect the adult human blood methylome**. *Int J Epidemiol* (2015) **44** 1211-1223. DOI: 10.1093/ije/dyv043
56. Toschi P, Capra E, Anzalone DA, Lazzari B, Turri F, Pizzi F, Scapolo PA, Stella A, Williams JL, Marsan PA, Loi P. **Maternal peri-conceptional undernourishment perturbs offspring sperm methylome**. *Reproduction* (2020) **159** 513-523. DOI: 10.1530/REP-19-0549
57. Wadhwa PD, Buss C, Entringer S, Swanson JM. **Developmental origins of health and disease: brief history of the approach and current focus on epigenetic mechanisms**. *Semin Reprod Med* (2009) **27** 358-368. DOI: 10.1055/s-0029-1237424
58. Wang X, Lan X, Radunz AE, Khatib H. **Maternal nutrition during pregnancy is associated with differential expression of imprinted genes and DNA methyltranfereases in muscle of beef cattle offspring**. *J Anim Sci* (2015) **93** 35-40. DOI: 10.2527/jas.2014-8148
59. Xie R, Sun Y, Wu J, Huang S, Jin G, Guo Z, Zhang Y, Liu T, Liu X, Cao X, Wang B, Cao H. **Maternal high fat diet alters gut microbiota of offspring and exacerbates DSS-induced colitis in adulthood**. *Front Immunol* (2018) **9** 2608. DOI: 10.3389/fimmu.2018.02608
60. Xu R, Li C, Liu X, Gao S. **Insights into epigenetic patterns in mammalian early embryos**. *Protein Cell* (2021) **12** 7-28. DOI: 10.1007/s13238-020-00757-z
61. Zglejc-Waszak K, Waszkiewicz EM, Franczak A. **Periconceptional undernutrition affects the levels of DNA methylation in the peri-implantation pig endometrium and in embryos**. *Theriogenology* (2019) **123** 185-193. DOI: 10.1016/j.theriogenology.2018.10.002
62. Zhang S, Chen X, Wang F, An X, Tang B, Zhang X, Sun L, Li Z. **Aberrant DNA methylation reprogramming in bovine SCNT preimplantation embryos**. *Sci Rep* (2016) **6** 1-11. DOI: 10.1038/srep30345
63. Zhang Q, Xiao X, Zheng J, Li M, Yu M, Ping F, Wang T, Wang X. **Maternal high-fat diet disturbs the DNA methylation profile in the brown adipose tissue of offspring mice**. *Front Endocrinol (Lausanne)* (2021) **12**. DOI: 10.3389/fendo.2021.705827
64. Zhang Z, Xu J, Lyu S, Xin X, Shi Q, Huang Y, Yu X, Zhu X, Li Z, Wang X, Lang L, Xu Z, Wang E. **Whole-genome DNA methylation dynamics of sheep preimplantation embryo investigated by single-cell DNA methylome sequencing**. *Front Genet* (2021) **12**. DOI: 10.3389/fgene.2021.753144
65. Zhou F, Wang R, Yuan P, Ren Y, Mao Y, Li R, Lian Y, Li J, Wen L, Yan L, Qiao J, Tang F. **Reconstituting the transcriptome and DNA methylome landscapes of human implantation**. *Nature* (2019) **572** 660-664. DOI: 10.1038/s41586-019-1500-0
66. Zhu P, Guo H, Ren Y, Hou Y, Dong J, Li R, Lian Y, Fan X, Hu B, Gao Y, Wang X, Wei Y, Liu P, Yan J, Ren X, Yuan P, Yuan Y, Yan Z, Wen L, Yan L, Qiao J, Tang F. **Single-cell DNA methylome sequencing of human preimplantation embryos**. *Nat Genet* (2018) **50** 12-19. DOI: 10.1038/s41588-017-0007-6
|
---
title: 'The Effects of Gestational Diabetes on Fetus: A Surveillance Study'
journal: Cureus
year: 2023
pmcid: PMC10023128
doi: 10.7759/cureus.35103
license: CC BY 3.0
---
# The Effects of Gestational Diabetes on Fetus: A Surveillance Study
## Abstract
Introduction: Gestational diabetes is an intolerance to glucose diagnosed during pregnancy that goes away postpartum. Gestational diabetes may result in outcomes such as birth trauma, increased rates of cesarean sections, and macrosomia. This study aims to determine the outcomes of gestational diabetes mellitus (GDM) on maternal and fetal health in a tertiary care hospital setting.
Materials and methods: *This is* a retrospective study of 52 patients who presented with gestational diabetes mellitus (GDM) and were treated at Tentishev Satkynbai Memorial Asian Medical Institute, Kyrgyzstan, between April 2021 and January 2022. The information was taken from the medical records of the patients. The baby's age, the mother's body mass index (BMI), history of pregnancy, deaths, birth weight, and the number of births were all taken into account.
Results: Out of all the cases during the study period at the Tentishev Satkynbai Memorial Asian Medical Institute, Kyrgyzstan, 52 were found to be complicated with gestational diabetes mellitus, which is $2.7\%$ of the total deliveries. There was a significant difference found among both study groups in gestational age and history of GDM. The neonatal intensive care unit (NICU) admission rate of neonates born to GDM mothers was found to be significant with a difference of $10.9\%$ ($p \leq 0.0003$), which is higher compared to the control group.
Conclusion: Incidences of macrosomia, NICU admissions of preterm babies, and large for gestational age (LGA) and increased rates of hypertensive disorders were found among GDM pregnancies compared to control cases. The study shows higher rates of maternal and fetal/neonatal complications in females with GDM.
## Introduction
Gestational diabetes mellitus (GDM) is an intolerance to glucose that is diagnosed during pregnancy [1]. This disorder commonly fades away postpartum; also, in a few cases, this sustains even after the pregnancy [2]. If their blood glucose levels are well controlled, females with GDM disorder usually have healthy neonatal births. These results can be achieved with either diet or insulin, combined with maintaining an appropriate body weight [3]. Gestational diabetes cases can result in negatively impacting pregnancy outcomes such as birth trauma, increased rates of cesarean sections ($10\%$ absolute risk), and macrosomia ($14\%$ absolute risk), and the outcome measurements may differ depending on variables such as screening methods, population cohorts, and diagnostic criteria [4].
The management of GDM is based on establishing control over the serum glucose levels in GDM patients by measuring glucose levels repeatedly both during home monitoring and via glycosylated hemoglobin [5]. A large population affected by GDM usually responds to a controlled diet via diet therapy alone, but the remaining subjects need insulin in addition to the dietary changes [6]. Controlling the blood sugar levels of patients affected by GDM via lifestyle and diet has shown improvement in perinatal outcomes in $70\%$-$85\%$ of patients in a recent study [7].
This study aims to determine the outcomes of phototherapy, jaundice, macrosomia birth, and other factors related to GDM on both mothers and fetuses/neonates in a Kyrgyz population and to compare them with nondiabetic pregnancy outcomes in a tertiary care hospital setting.
## Materials and methods
This is a retrospective study of 52 patients presenting with GDM and treated at the Tentishev Satkynbai Memorial Asian Medical Institute, Kyrgyzstan, between April 2021 and January 2022. The ethical committee of the Tentishev Satkynbai Memorial Asian Medical Institute, Kyrgyzstan, approved the undertaking of the study, and the data used in the current study were obtained from the medical records of the patients, where the age, body mass index (BMI), pregnancy complications, mortality, and birth weight data of the neonate and parity were collected and processed for this study.
The control group was selected using hospital databases, as they had not presented with GDM at any stage of their pregnancy, and the age, parity, and BMI of the control group were matched with those of the GDM group in the study. All the subjects selected for the study were hospitalized and delivered to the hospital's gynecology department. The measurements taken to record neonatal outcomes were weight at the time of birth, hypoglycemia, respiratory distress syndrome (RDS), hypocalcemia, bilirubin, and intensive care unit (ICU) admissions. The Apgar scores used in the study were collected from the hospital records at one, five, and 10 minutes postpartum.
The screening for GDM was done during the study period using a selective screening method that is based on factors such as prepregnancy GDM, age of the mother, weight, family history of diabetes, and macrosomia birth history. During the gestational period, the screening took place between weeks 20 and 25. If the results obtained via venipuncture using the glucose oxidase method exceeded 140 mg/dL, the patient would then go through another test in which a 100 g glucose oral test was administered, and after three hours, the following cutoff values were used: fasting blood sugar of 100 mg/dL; after one hour, 190 mg/dL; after two hours, 165 mg/dL; and after three hours, 165 mg/dL [3]. Patients who were diagnosed with GDM have been prescribed a diet within the 1800 kcal diabetic diet for one week, followed by a fasting blood sugar test if the results showed a fasting blood sugar value of less than 100 mg/dL and a postprandial value of less than 125 mg/dL; those patients were managed by adhering to a specialized diet only. Patients presenting with results higher than these cutoff values were prescribed insulin to control their blood glucose levels. The number of GDM patients treated only with dietary changes was 40; the remaining 12 patients required insulin on top of diet changes. The patients included in the study were scheduled for follow-up every two weeks after GDM was confirmed. Labor was induced after 40 weeks in patients who had been treated for GDM solely with diet. Some patients required an earlier labor induction due to toxemia, depending on their biophysical profile. The blood sugars of neonates born to diabetic mothers were measured after delivery and continued to be measured until the values were stable. Intravenous glucose was administered to babies with hypoglycemia, and feeding was initiated as soon as possible. Blood sugar measurements on neonates in the control group were taken only when indicated.
Statistical Package for Social Sciences (SPSS) version 23 (IBM SPSS Statistics, Armonk, NY) was used to perform statistical analysis, and a chi-square test and Fisher's exact test were used to assess statistical significance. Confidence intervals of $90\%$ and odds ratios (OR) were calculated. A p-value of 0.05 was considered significant. To estimate the confidence interval and odds ratio, a multivariate logistic regression was employed. The results are shown as mean standard deviation, "n" represents frequency, and nominal data are presented as a percentage (%).
## Results
Out of all the cases during the study period at the Tentishev Satkynbai Memorial Asian Medical Institute, Kyrgyzstan, 52 were found to be complicated with gestational diabetes mellitus, which is $2.7\%$ of the total deliveries. The demography of the females affected with GDM can be viewed in Table 1.
**Table 1**
| Characteristics | GDM, N = 52 (%) | Control, N = 52 (%) | P-value | OR (95% CI) | Adjusted OR (95% CI) |
| --- | --- | --- | --- | --- | --- |
| Age (in years) | 28 ± 7.5 | 29.2 ± 6.8 | 0.3947 | - | - |
| Parity: 0, 1, and ≥2 | 9 (16.9%), 15 (29.3%), and 28 (53.8%) | 8 (15.3%), 18 (34.6%), and 26 (50.1%) | - | - | - |
| Delivery time (weeks) mean age for gestation | 38.5 ± 1.4 | 39.4 ± 1.6 | 0.0001* | (0.082-0.518) | (0.101-0.645) |
| BMI (kg/m2) | 28.4 ± 1.5 | 27.1 ± 1.6 | 0.1490 | - | - |
| DM family history | 22 (41.43%) | 17 (32.34%) | 0.060 | - | - |
| Prior history of (H/O) GDM | 10 (19.5%) | 4 (7.7%) | 0.0004* | 2.901 (1.597-5.268) | 2.072 (1.064-4.745) |
| Previous H/O macrosomia | 4 (7.3%) | 2 (4.5%) | 0.3121 | - | - |
| Previous stillbirth | 1 (1.4%) | 1 (1.8%) | 1.000 | - | - |
| Hypertension | 9 (18.2%) | 3 (5.9%) | <0.0001* | 3.538 (1.834-6.824) | 2.958 (1.251-6.313) |
| Preterm delivery | 6 (11.42%) | 3 (5.11%) | 0.0233 | 2.434 (1.167-5.082) | 2.013 (1.059-4.862) |
| C-section | 12 (24.11%) | 6 (12.33%) | 0.0018 | 2.269 (1.364-3.769) | 2.134 (1.123-2.934) |
| Polyhydramnios | 2 (3.22%) | 1 (1.44%) | 0.337 | - | - |
| Oligohydramnios | 1 (2.71%) | 0 (0.91%) | 0.2845 | - | - |
| Labor induction | 17 (31.82%) | 6 (12.31%) | <0.0001* | 3.334 (2.037-5.458) | 2.702 (1.852-5.224) |
Both study groups showed a significant difference in gestational age and history. On multivariate logistic regression high-risk variable, the values found are as follows: Hypertensive disorders were found to be $p \leq 0.0001$, the p-value was 0.001 for the induction of labor, and the value of preterm delivery was $p \leq 0.236.$ The multivariate analysis showed that females who were previously diagnosed with GDM were at a higher risk of developing GDM in future pregnancies.
Babies born to GDM females had a higher mean birth weight (macrosomia) compared to control cases' mean birth weight, which is elaborated in Table 2.
**Table 2**
| Outcome | GDM, N = 52 (%) | Control, N = 52 (%) | P-value | Odds ratio (95% confidence interval) | Odds ratio (adjusted) (95% confidence interval) |
| --- | --- | --- | --- | --- | --- |
| Mean birth weight (grams) | 3544 ± 466 | 3357 ± 332 | <0.0001 | (113.18-264.82) | (105-231.40) |
| Large for gestational age | 8 (14.54%) | 3 (5.11%) | 0.0012 | 3.233 (1.586-6.395) | 3.342 (1.465-6.376) |
| Macrosomia birth | 7 (12.71%) | 3 (5.1%) | 0.0185 | 2.76 (1.343-5.618) | 2.68 (1.233-5.494) |
| Small for gestational age | 4 (7.4%) | 3 (6.87%) | 1.01 | | |
| Birth weight of <2.500 kg | 2 (3.68%) | 2 (3.21%) | 1.0 | | |
| ICU entry | 9 (16.4%) | 3 (5.5%) | 0.0003* | 3.391 (1.713-6.712) | 2.954 (1.732-6.805) |
| RDS | 1 (1.4%) | 0 (0.9%) | 0.6233 | | |
| Hypoglycemia | 1 (2.74%) | 0 (0.91%) | 0.2789 | | |
| Incidence of jaundice in newborns | 4 (8.2%) | 2 (4.5%) | 0.1707 | | |
| Phototherapy | 3 (5.1%) | 1 (2.7%) | 0.2017 | | |
| Apgar score of <7 at five minutes | 2 (3.2%) | 1 (2.3%) | 0.543 | | |
| Congenital anomalies in neonates | 1 (1.4%) | 1 (1.8%) | 1.000 | | |
| Perinatal deaths | 0 13.6 | 0 9.1 | 1.0000 | | |
The neonatal intensive care unit (NICU) admission rate of neonates born to GDM mothers was $10.9\%$, which is significantly higher than the control group ($p \leq 0.0003$). No perinatal death was recorded in the study duration.
## Discussion
Gestational diabetes mellitus is a clinical complication diagnosed during pregnancy that is associated with a person having up to a $60\%$ chance of developing type 2 diabetes mellitus in the later stages of life [8]. GDM, if left untreated during pregnancy, is strongly related to a higher risk of developing maternal and perinatal risks that are responsible for morbidity and mortality in some cases [9]. Patients are evaluated and screened for GDM to prevent perinatal morbidity, stillbirth, and large for gestational age (LGA) babies, which are the most common complications to avoid [10]. According to one study, excessive fetal growth has an indirect impact, with risk factors such as the patient's parity, ethnicity, maternal age, and maternal obesity being significant factors [11]. The results of this study largely conform with the results of previous studies; however, there are a few differences in outcomes. This study was conducted under the assumption that GDM patients are more likely to experience complications during pregnancy, as well as adverse effects on fetal outcomes. The results show a higher incidence of obstetric complications, including but not limited to preterm labor, preeclampsia, LGA, macrosomia, and a higher need for cesarean section, in GDM-affected females compared to the control group; the results are similar to another GDM risk score study [12].
The rate of the induction of labor was found higher in GDM cases in a recent study, which is similar to the results of our study where, in GDM cases, the rate was $31.8\%$ [13]. The rate of cesarean section was found to almost double in the GDM group as compared to the control group in our study, which is consistent with a review study [14]. Previous cesarean section history, hypertension, macrosomia, and insubstantial fetal heart tracing were the main indicators of cesarean section in GDM-affected subjects. This study confirms the findings of a previous study that changes in diet and GDM treatment through insulin, if required, drastically change the serious perinatal morbidity rate of patients and common birth issues of neonates [15]. The GDM group included in this study showed significant rates of neonatal intensive care unit (NICU) admissions ($16.4\%$, $p \leq 0.0003$) compared to the control group, and the rate of admission was found to be slightly higher in the GDM group that was treated with insulin. Significant differences were not found in small for gestational age (SGA) neonates and neonatal hypoglycemia and phototherapy among both groups. However, due to pregnancy risk factors and fetal complications, neonates of GDM mothers spent more time in the NICU after admission than the control group; this is due to the hospital policy of keeping GDM neonates under observation for 24 hours. In our study, the rate of NICU admissions was comparable to a previous study done in Australia [16].
An Iranian study discovered that even minor changes in blood glucose levels can result in macrosomia (abnormal fetal growth) and other complications, which can be avoided by implementing simple measures such as a controlled diet and insulin use during the gestational period [17]. The standard treatment for GDM management is diet control and insulin therapy. Oral hypoglycemic agents have also shown promise in controlling blood glucose levels during GDM without having a negative impact on fetal outcomes [18].
The study's limitations include the small sample size and the lack of evaluation of numerous biomarkers. There needs be more research with a larger sample size to assess the connections between maternal glucose levels and perinatal outcomes.
## Conclusions
In this study, the incidence of macrosomia, preterm baby NICU admissions, and LGA, and a higher incidence of hypertensive issues were found in GDM pregnancies compared to the non-GDM control group. The results of this study support the previous findings of higher rates of maternal and neonatal complications in females affected by GDM. The study concludes that tight control of GDM during pregnancy can turn out to be the one major variable that can significantly reduce the complications correlated with GDM. The study concludes that GDM is directly related to obesity and recommends that all patients be screened for GDM because even mild diabetes can have a significant impact on fetuses and pregnant females' outcomes. The study found weight as a major variable that can be altered to get positive results for both maternal and fetal outcomes.
## References
1. Tang L, Xu S, Li P, Li L. **Predictors of insulin treatment during pregnancy and abnormal postpartum glucose metabolism in patients with gestational diabetes mellitus**. *Diabetes Metab Syndr Obes* (2019) **12** 2655-2665. PMID: 31853192
2. Shub A, Miranda M, Georgiou HM, McCarthy EA, Lappas M. **The effect of breastfeeding on postpartum glucose tolerance and lipid profiles in women with gestational diabetes mellitus**. *Int Breastfeed J* (2019) **14** 46. PMID: 31708997
3. Domanski G, Lange AE, Ittermann T, Allenberg H, Spoo RA, Zygmunt M, Heckmann M. **Evaluation of neonatal and maternal morbidity in mothers with gestational diabetes: a population-based study**. *BMC Pregnancy Childbirth* (2018) **18** 367. PMID: 30200916
4. Buchanan TA, Xiang AH, Page KA. **Gestational diabetes mellitus: risks and management during and after pregnancy**. *Nat Rev Endocrinol* (2012) **8** 639-649. PMID: 22751341
5. Rani PR, Begum J. **Screening and diagnosis of gestational diabetes mellitus, where do we stand**. *J Clin Diagn Res* (2016) **10** 0-4
6. Kusinski LC, Murphy HR, De Lucia Rolfe E. **Dietary intervention in pregnant women with gestational diabetes; protocol for the digest randomised controlled trial**. *Nutrients* (2020) **12** 1165. PMID: 32331244
7. Lende M, Rijhsinghani A. **Gestational diabetes: overview with emphasis on medical management**. *Int J Environ Res Public Health* (2020) **17** 9573. PMID: 33371325
8. Noctor E, Dunne FP. **Type 2 diabetes after gestational diabetes: the influence of changing diagnostic criteria**. *World J Diabetes* (2015) **6** 234-244. PMID: 25789105
9. Langer O, Yogev Y, Most O, Xenakis EM. **Gestational diabetes: the consequences of not treating**. *Am J Obstet Gynecol* (2005) **192** 989-997. PMID: 15846171
10. Wang M, Athayde N, Padmanabhan S, Cheung NW. **Causes of stillbirths in diabetic and gestational diabetes pregnancies at a NSW tertiary referral hospital**. *Aust N Z J Obstet Gynaecol* (2019) **59** 561-566. PMID: 30663043
11. Gaudet L, Ferraro ZM, Wen SW, Walker M. **Maternal obesity and occurrence of fetal macrosomia: a systematic review and meta-analysis**. *Biomed Res Int* (2014) **2014** 640291. PMID: 25544943
12. Phaloprakarn C, Tangjitgamol S. **Risk score for predicting primary cesarean delivery in women with gestational diabetes mellitus**. *BMC Pregnancy Childbirth* (2020) **20** 607. PMID: 33032545
13. Muche AA, Olayemi OO, Gete YK. **Effects of gestational diabetes mellitus on risk of adverse maternal outcomes: a prospective cohort study in Northwest Ethiopia**. *BMC Pregnancy Childbirth* (2020) **20** 73. PMID: 32013909
14. Berger H, Melamed N. **Timing of delivery in women with diabetes in pregnancy**. *Obstet Med* (2014) **7** 8-16. PMID: 27512412
15. Buhary BM, Almohareb O, Aljohani N. **Glycemic control and pregnancy outcomes in patients with diabetes in pregnancy: a retrospective study**. *Indian J Endocrinol Metab* (2016) **20** 481-490. PMID: 27366714
16. Seimon RV, Natasha N, Schneuer FJ. **Maternal and neonatal outcomes of women with gestational diabetes and without specific medical conditions: an Australian population-based study comparing induction of labor with expectant management**. *Aust N Z J Obstet Gynaecol* (2022) **62** 525-535. PMID: 35347699
17. Tabrizi R, Asemi Z, Lankarani KB. **Gestational diabetes mellitus in association with macrosomia in Iran: a meta-analysis**. *J Diabetes Metab Disord* (2019) **18** 41-50. PMID: 31275873
18. Kintiraki E, Goulis DG. **Gestational diabetes mellitus: multi-disciplinary treatment approaches**. *Metabolism* (2018) **86** 91-101. PMID: 29627447
|
---
title: 'Hertel Exophthalmometry Values in a Greek Adult Outpatient Clinic-Based Population:
Association With Demographic Factors and Systemic Disease'
journal: Cureus
year: 2023
pmcid: PMC10023137
doi: 10.7759/cureus.35027
license: CC BY 3.0
---
# Hertel Exophthalmometry Values in a Greek Adult Outpatient Clinic-Based Population: Association With Demographic Factors and Systemic Disease
## Abstract
Purpose: To investigate correlations of exophthalmometry values (EVs) with age, gender, and the presence of diabetes mellitus, arterial hypertension, and dyslipidemia.
Methods: *In a* cross-sectional, clinic-based study, consecutive adult Greek patients presenting for evaluation at the outpatient general clinic on a scheduled appointment basis at a tertiary care referral center were submitted to Hertel exophthalmometry in both eyes by the same observer. Subjects with signs of history or orbital pathology, including thyroid-associated ophthalmopathy, were excluded. Demographics, as well as a detailed systemic history report, were recorded. Mixed effect linear regression analysis was performed to account for the correlation between the eyes of the same participant.
Results: A total of 800 eyes (400 subjects) were included, 194 males and 206 females, with a mean age of 67.82 ± 12 years (range: 18-92 years). The mean exophthalmometry value was 15.7 ± 2.6 mm (range: 11-21 mm). Every one year of increase in age is associated with a decrease in EVs by 0.03 mm ($95\%$ CI -0.04, -0.02/p-value<0.001). Female gender was associated with lower EVs by 0.33mm ($95\%$ CI-0.56, -0.1/p-value=0.005). Patients with diabetes mellitus had higher EVs by 0.47 mm ($95\%$ CI 0.25, 0.70/p-value<0.001) compared to patients without diabetes, and patients with arterial hypertension had lower EVs by 0.26 mm ($95\%$ CI -0.5, -0.02/p-value=0.034) compared to patients without hypertension. No association was found between dyslipidemia and systemic history of thyroid dysfunction.
Conclusions: A negative correlation of EVs was noted with increasing age, female gender, as well as history of arterial hypertension and a positive correlation with diabetes mellitus.
## Introduction
The orbit is a protective bony socket in the skull that contains and protects the eyeball, the optic nerve, other nerves of the eye, adipose tissue (fat), blood vessels, and six extraocular muscles [1,2]. It is pear-shaped, and its volume is 30cc [1]. Enlargement of one or more elements of the orbital tissue can cause ocular protrusion [1]. Exophthalmometry is a routine examination method for patients with suspected ocular protrusion [3]. The Hertel exophthalmometer is a reliable method for measuring ocular protrusion [4] and the most commonly used exophthalmometer in clinical settings [5]. Normal exophthalmometry values are trying to be established by the scientific community for different populations [4]. Normative exophthalmometry values (EVs) may vary and may be affected by ethnic origin, age, gender, refraction, and axial length of the eye [2,4]. Well-known pathological conditions causing proptosis include thyroid-associated orbitopathy, head and orbital trauma, tumors, and craniofacial abnormalities [2,6]. The purpose of this study was to determine exophthalmometry values in the Greek population and the impact of age, gender, and common systemic diseases, including diabetes mellitus, arterial hypertension, and dyslipidemia.
## Materials and methods
Study design and population *As a* part of a larger ongoing study designed to characterize normative values for exophthalmometry readings in the Greek population, we randomly recruited patients presenting at the outpatient general clinic on a scheduled appointment basis at a tertiary care referral center (Athens General Hospital “G. Gennimatas”).
The study was approved by the Institutional Review Board of “G. Gennimatas” General Hospital of Athens and was conducted in accordance with the Declaration of Helsinki. Recruiting and assessment took place between January 2017 and December 2019. All subjects were informed and provided their consent before enrolling in the study.
Inclusion criteria were age greater than 18 years, Caucasian origin, and absence of medical or surgical retinal intervention for a time interval shorter than four weeks before the exophthalmometry measurement. Exclusion criteria were: history of strabismus or orbital pathology, including thyroid-associated ophthalmopathy, orbital tumors, orbital trauma, and craniofacial abnormalities or asymmetry, as well as EVs difference between the two eyes equal or greater than 2 mm.
Data collection and outcome measures Demographic information (age, gender, race) and a detailed systemic history were obtained for all study participants. Patient history was obtained and recorded by recruiting physicians (KC, P.P) while the patients waited for evaluation. Participants were specifically asked about the presence of thyroid dysfunction and systemic vasculopathy disease, including arterial hypertension, diabetes mellitus, and dyslipidemia; evidence of the above was further supported by documentation of pertinent, concurrent, or past medical and/or past surgical treatment (i.e., thyroidectomy for thyroid dysfunction). The absence of thyroid-associated orbitopathy was based on the patient's reported history of absence of obvious ocular pathology, including exophthalmos and diplopia, as well as the absence of reported concurrent or past medical or surgical treatment suggestive of thyroid-associated orbitopathy, including orbit, extraocular muscles, or eyelid surgery, corticosteroid, immunosuppressive or selenium administration.
All subjects underwent Hertel exophthalmometry in both eyes (Hertel exophthalmometer K-0161, Inami& Co. Ltd. Japan). Exophthalmometry readings to the closest millimeter (mm) were measured at one setting for both eyes; the distance between the lateral canthi (intercanthal distance, ICD), as well as the interpupillary distance (IPD), were measured by the same observer (AN) and recorded as well.
Statistical analysis The normal distribution of demographic and clinical information was assessed by plots (histogram and probability graphs) and corresponding statistical tests (Kolmogorov-Smirnov/Shapiro-Wilk test). Normally distributed continuous values were summarized by the mean and standard deviation (SD), and categorical data by number (N) and presentence (%). The correlation between EVs and all studied variables was investigated by applying univariate and multivariate linear regression analysis for each eye separately and for both eyes together. To assess the sensitivity of our findings for the latter analysis, we further applied mixed effects linear regression to account for the correlation between the eyes from the same subject. All multivariate models were adjusted for age, gender, diabetes mellitus, arterial hypertension, thyroid dysfunction, and dyslipidemia. Statistical significance was set at $P \leq 0.05.$ Analysis was conducted in StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP.
## Results
From 430 subjects who initially consented, 409 were eligible to participate in the study. Patients who demonstrated poor cooperation with exophthalmometry testing, some of them with unstable fixation due to low vision, four subjects with Asian racial backgrounds, and five subjects with an exophthalmometry value difference of equal or larger than 2 mm between the two eyes were excluded. Hence a total of 400 patients (800 eyes) were enrolled in the study. All patients were of Greek (Caucasian) ethnicity.
The majority of patients were under medical treatment for the underlying medical condition: 182 ($45.50\%$) were diagnosed with diabetes mellitus, 173 ($43.25\%$) with arterial hypertension, 121 ($30.25\%$) with dyslipidemia and 57 ($14.25\%$) with thyroid dysfunction. The clinical characteristics of the studied variables under investigation are demonstrated in Table 1.
**Table 1**
| Unnamed: 0 | Unnamed: 1 | N (%) | Exophthalmometry values (Mean±SD) |
| --- | --- | --- | --- |
| Gender | Males | 194 (48.50) | 16.07±1.67 |
| | Females | 206 (51.50) | 15.60±1.66 |
| Diabetes Mellitus | No | 218 (54.50) | 15.59±1.63 |
| | Yes | 182 (45.50) | 16.10± 1.70 |
| Arterial Hypertension | No | 227 (56.75) | 16±1.68 |
| | Yes | 173 (43.25) | 15.61±1.66 |
| Thyroid dysfunction | No | 343 (85.75) | 15.87±1.68 |
| | Yes | 57 (14.25) | 15.54±1.67 |
| Dyslipidemia | No | 279 (69.75) | 15.84±1.69 |
| | Yes | 121 (30.25) | 15.80±1.66 |
| N=Number; SD=standard deviation | N=Number; SD=standard deviation | N=Number; SD=standard deviation | N=Number; SD=standard deviation |
The mean Hertel exophthalmometry value was 15.8±1.7 mm (range; 12 to 22 mm). The mean interpupillary distance was 60.97±2.5 (range; 55-70), and the mean intercanthal distance was 97.81 ± 3.82 mm (range; 90 to 112 mm). The mean difference of the exophthalmometry value (relative exophthalmometric value) between the eyes of each participant was 0.08 ± 0.06 mm in the study population, with a maximum difference of 1.5 mm.
We performed univariate linear regression analysis between EVs and each of the variables studied (only for the left eyes, only for the right eyes, and for both eyes, respectively). There was a statistically significant negative correlation between EVs and age, female gender, and arterial hypertension and a statistically significant positive correlation with diabetes mellitus in each analysis. In the univariate analysis for both eyes, thyroid dysfunction had a bordered statistically significant negative correlation with EVs. Univariate analysis of the left and right eye separately for thyroid dysfunction and each univariate analysis for dyslipidemia were not proven of statistical significance. The results are presented in Table 2. Scatterplots of exophthalmometric values with respect to age and box plots of the descriptive relationship between EVs and thyroid dysfunction, diabetes, hypertension, and dyslipidemia are presented in Figure 1.
We also performed multivariable linear regression analysis among EVs; and all the variables studied (only left eyes, only right eyes, or both eyes, respectively). The above-noted correlations of univariate linear regression analysis of EVs with age, presence of diabetes mellitus, and arterial hypertension persisted. The bordered correlation with thyroid dysfunction no longer existed in the multivariate linear regression analysis.
Finally, we applied mixed-effects linear regression adjusted for each variable to account for the dependence that both eyes correspond to the same individual (Table 3). Analysis revealed that the mixed effects linear regression model was the optimal fit that better accounted for the structure of the observation since the likelihood ratio test versus linear regression showed a statistically significant difference (p-value<0.001) in every comparison, keeping all other confounders not statistically significant.
**Table 3**
| Exophthalmometry values | Unnamed: 1 | Unnamed: 2 | Unnamed: 3 |
| --- | --- | --- | --- |
| | Coef. | p-value | 95% CI |
| Age | -0.03 | <0.001*** | -0.04, -0.02 |
| Gender (females) | -0.33 | 0.005 ** | -0.56, -0.1 |
| Diabetes | 0.47 | <0.001*** | 0.25, 0.70 |
| Hypertension | -0.26 | 0.030* | -0.5, -0.02 |
| Dyslipidemia | 0.11 | 0.395 | -0.14, 0.36 |
| Thyroid Dysfunction | -0.23 | 0.172 | -0.55, 0.1 |
| ***p<0.001, **p<0.01, *p<0.05; Coef= coefficient; CI=confidence intervals | ***p<0.001, **p<0.01, *p<0.05; Coef= coefficient; CI=confidence intervals | ***p<0.001, **p<0.01, *p<0.05; Coef= coefficient; CI=confidence intervals | ***p<0.001, **p<0.01, *p<0.05; Coef= coefficient; CI=confidence intervals |
We conclude that there was a statistically significant negative correlation between EVs and age, female gender, and arterial hypertension and a statistically significant positive correlation with diabetes mellitus. More specifically, every one year of increase in age is associated with a decrease in exophthalmometry measurements by 0.03 mm ($95\%$ CI -0.04, -0.02/ p-value<0.001). Females had decreased EVs by 0.33mm ($95\%$ CI-0.56, -0.1 / p-value=0.005). Moreover, patients with hypertension had lower EVs by 0.26 mm ($95\%$ CI -0.5, -0.02/ p-value=0.030 compared to patients without hypertension. In addition, patients with diabetes had a higher exophthalmometry measurement by 0.47 mm ($95\%$ CI 0.25, 0.70/ p-value<0.001) compared to patients without diabetes. Thyroid dysfunction and dyslipidemia were not proven of statistical significance.
We also performed multivariate and mixed-effect linear regression analysis adding intercanthal and interpupillary distance to the above-noted variables. The above-noted correlations of EVs with age, presence of diabetes mellitus, and arterial hypertension persisted the negative correlation with female gender no longer persisted in the multivariate and mixed-effects regression analysis. This was the result of the preponderance of the correlation of the variable “intercanthal distance” as a statistically significant positive correlation with EVs.
## Discussion
In this clinic-based cross-sectional study, common vasculopathy systemic diseases were found to be correlated with differences in exophthalmometry measurements. More specifically, arterial hypertension was negatively correlated with EVs, while diabetes mellitus was positively correlated. To our knowledge, this is the first study that evaluates exophthalmometry in relation to diabetes mellitus, arterial hypertension, and dyslipidemia.
The relationship of exophthalmometric values with demographic factors has been extensively investigated. Kashkouli et al. [ 7] and Nath et al. [ 8] demonstrated that children and teenagers EVs significantly increased as they grow older into adulthood. Fledelius [9] found EVs to remain stable after the late teenage years. Several studies have demonstrated a negative correlation between EVs and advancing age, reporting a statistically significant reduction in exophthalmolmetry values with every year of age increase starting from 20 years of age [4,10-12]. Our study results are consistent with these findings.
The association between gender and ocular protrusion remains controversial. Some studies [4,13-15] reported significantly higher values in males than in females. Kashkouli et al. [ 3] and Wu et al. [ 12] did not find a statistically significant correlation between gender and EVs in their study. Our study demonstrated statistically significantly lower EVs in female participants, a finding related to facial metric parameters since it was overshadowed by statistically significant positive associations of exophthalmometric values with intercanthal distance measurements.
Ethnic and racial background may affect facial metric parameters [13,16]; we thus opted for excluding the few number subjects of non-Caucasian origin who had initially been recruited in this study.
In a report on 46 participants by Detorakis et al. [ 17], the mean exophthalmometry value was reported to be 16 mm for males and 15mm for females, in a hospital-based sample of the Greek adult population, with differences between gender not being statistically significant. In addition, the EVs detected in the current study were comparable to those described by other studies. In a cross-sectional study on 236 Turkish adult participants by Karti et al. [ 2], mean Hertel exophthalmometric readings were 15.7±2.6 mm (range; 11 to 21 mm). The mean value for males was 16.1±2.6 mm, and for females, 15.5±2.6 mm. Bageri et al. [ 10], in a metanalysis of four studies in the Iranian population and 3,696 subjects, reported a mean exophthalmometry value for males at 16.5 mm and females at 16.2 mm. Barrette et al. [ 17], in a cohort that included 65 white adults, reported mean EVs of 17±2.65 mm for white males and 15.98±2.22 mm for white females.
Other than thyroid eye disease, certain systemic diseases have been reported to affect eye protrusion. Schwarz et al. [ 18] postulated that there was possibly an exophthalmos-producing activity in the serum and pituitary of patients with Cushing's syndrome and acromegaly. Obesity has been shown to enhance ocular protrusion [15,19]. Smolders et al. [ 19] demonstrated that $33\%$ of obese patients in their series had bilateral exophthalmos.
Obesity is a predisposing factor for diabetes mellitus and arterial hypertension [20]. Type 2 diabetes mellitus is an expanding global health problem. The most common ophthalmic complications of longstanding diabetes mellitus, macular edema, and proliferative diabetic retinopathy are associated with the thickening of intraocular structures, including the retina [21]; reports on the impact of diabetes on the choroid are controversial [21-23].
The extensive investigation of diabetic complications on the thickness of intraocular structures has led investigators to propose specific choroidal imaging indices as biomarker candidates. Endo et al. [ 23], using enhanced depth imaging optical coherence tomography and a binarisation method, suggested a potential biomarker role for the ratio of the luminal area (LA) in the total choroidal area (TCA), designated as the L/C ratio: they found that a lower L/C ratio was associated with longer duration of diabetes.
On the other hand, there is no data in access to our literature on the impact of diabetes mellitus on the orbital soft tissues. A majority of individuals suffering from type 2 diabetes are obese, with central visceral adiposity [24]. Several adipose-tissue-centric mechanisms have been proposed in the pathogenesis of diabetes mellitus, with chronic, low-grade adipose tissue inflammation receiving considerable attention [25]. Although the intraorbital fat has not been the subject of related investigations [26], being assigned to a different origin, the neural crest as opposed to mesoderm for the most well-studied subcutaneous or visceral adipose tissue [27], we might hypothesize that orbital adipose perivascular tissue could potentially be involved in low-grade chronic inflammation in diabetic patients accounting for the relative increase in noted EVs series.
It is also noteworthy that the presence of diabetes mellitus has been associated with more severe forms of thyroid-associated ophthalmopathy [28]. Side effects of diabetic medications could also be considered as possible contributors to ocular proptosis. Dorkhan et al. [ 29] described a subgroup of type 2 diabetic patients responding with increased eye protrusion when treated with pioglitazone, one of the thiazolidinediones, a class of glucose-lowering drugs that promote insulin sensitivity.
Arterial hypertension was shown to correlate with lower EVs in our study. There is extensive literature associating elevated blood pressure with decreased thickness in intraocular structures, including decreased choroidal thickness [30], decreased macular thickness [31], inner retinal layer, particularly ganglion cell-inner plexiform layer (GC-IPL) thinning [32], which is significantly correlated with a decrease in retinal blood flow [33]. Antihypertensive medication, specifically angiotensin-converting enzyme inhibitors or diuretics, was reported to be associated with thinning of the retinal nerve fiber layer [32]. The above well-studied and reproduced associations have led to suggesting a possible role of Optical Coherence Tomography angiography in identifying subclinical microvascular damage [34].
On the antipodes of extensive investigation regarding intraocular structures, there is a paucity of data on the impact of arterial hypertension on extraocular orbital tissues in the access to our literature. We could theorize that arteriosclerosis-related reduction in vascular flow in the intraorbital soft tissues associated with systemic arterial hypertension could account for decreased EVs in our series.
It is noteworthy that the presence of isolated systemic thyroid dysfunction (without signs of thyroid-associated orbitopathy) was not associated with a change in exophthalmometry readings either in the multivariate or the mixed-effects regression analysis in this study. We postulate that these data indirectly support the statistical significance of our findings on the associations with arterial hypertension and diabetes mellitus. Dyslipidemia was not found to be correlated with the exophthalmometry readings either.
The knowledge that certain systemic diseases (diabetes mellitus and arterial hypertension) but not others (dyslipidemia or thyroid dysfunction without signs of thyroid-associated orbitopathy) may affect exophthalmometry readings may be of value in the interpretation of exophthalmometry when coexisting orbital or systemic (dysthyroid) pathology is contemplated. The above-noted associations may also shed light on a better understanding of the pathophysiology of diabetes mellitus and the impact of microvascular systemic disease in the microenvironment of the orbit.
This study has to be viewed in light of its limitations. Systemic history was based on a patient-reported history. Several possible confounding factors, including obesity, were not documented in this study. The exact concomitant retinal disease diagnosis, as well as a detailed medication intake history, were not recorded and analyzed. Medication history is important because apart from systemic, topical ophthalmic medications have been associated with a change in ocular protrusion as well: use of prostaglandin intraocular pressure lowering drops, for example, notably bimatoprost, have been shown to decrease orbital fat and induce enophthalmos [35].
The external validity of our results needs to be confirmed, as measurements have been taken on a selected patient sample. Possible bias towards more severe and longstanding cases of diabetes mellitus or hypertension, as evidenced by their complicated course and referral for ophthalmic manifestations, may exist.
Future studies could validate our results by expanding to a population-based sample and addressing additional confounding factors, including specific ophthalmic diagnosis, the axial length of the eye, body mass indices, and a detailed systemic and topical medication intake history.
## Conclusions
In this clinic-based cross-sectional study, a negative correlation of EVs was noted with increasing age, female gender, as well a history of arterial hypertension, and a positive correlation with diabetes mellitus. To our knowledge, this is the first study that evaluates exophthalmometry in relation to common vasculopathy systemic diseases. Future studies could validate our results by expanding to a population-based sample and addressing additional confounding factors.
## References
1. Shumway CL, Motlagh M, Wade M. **Anatomy, Head and Neck, Orbit Bones**. (2022)
2. Karti O, Selver OB, Karahan E, Zengin MO, Uyar M. **The effect of age, gender, refractive status and axial length on the measurements of hertel exophthalmometry**. *Open Ophthalmol J* (2015) **9** 113-115. PMID: 26089994
3. Kashkouli MB, Beigi B, Noorani MM, Nojoomi M. **Hertel exophthalmometry: reliability and interobserver variation**. *Orbit* (2003) **22** 239-245. PMID: 14685897
4. Cheung JJ, Chang DL, Chan JC. **Exophthalmometry values in the Hong Kong Chinese adult population from a population-based study**. *Medicine (Baltimore)* (2019) **98** 0
5. Dunsky IL. **Normative data for hertel exophthalmometry in a normal adult black population**. *Optom Vis Sci* (1992) **69** 562-564. PMID: 1635761
6. Limongi RM, Feijó ED, Rodrigues Lopes E Silva M. **Orbital bone decompression for non-thyroid eye disease proptosis**. *Ophthalmic Plast Reconstr Surg* (2020) **36** 13-16. PMID: 31373985
7. Kashkouli MB, Nojomi M, Parvaresh MM, Sanjari MS, Modarres M, Noorani MM. **Normal values of hertel exophthalmometry in children, teenagers, and adults from Tehran, Iran**. *Optom Vis Sci* (2008) **85** 1012-1017. PMID: 18832980
8. Nath K, Gogi R, Rao GS. **Normal exophthalmometry**. *Indian J Ophthalmol* (1977) **25** 47-52. PMID: 614274
9. Fledelius HC, Stubgaard M. **Changes in eye position during growth and adult life as based on exophthalmometry, interpupillary distance, and orbital distance measurements**. *Acta Ophthalmol (Copenh)* (1986) **64** 481-486. PMID: 3492853
10. Bagheri A, Shahraki K, Arabi A, Kashkouli MB. **Normal Exophthalmometry Values in Iranian Population: A Meta-analysis**. *J Ophthalmic Vis Res* (2021) **16** 470-477. PMID: 34394874
11. Ahmadi H, Shams PN, Davies NP, Joshi N, Kelly MH. **Age-related changes in the normal sagittal relationship between globe and orbit**. *J Plast Reconstr Aesthet Surg* (2007) **60** 246-250. PMID: 17293280
12. Wu D, Liu X, Wu D, Di X, Guan H, Shan Z, Teng W. **Normal values of Hertel exophthalmometry in a Chinese Han population from Shenyang, Northeast China**. *Sci Rep* (2015) **5** 8526. PMID: 25703959
13. Migliori ME, Gladstone GJ. **Determination of the normal range of exophthalmometric values for black and white adults**. *Am J Ophthalmol* (1984) **98** 438-442. PMID: 6486215
14. Mourits MP, Lombardo SH, van der Sluijs FA, Fenton S. **Reliability of exophthalmos measurement and the exophthalmometry value distribution in a healthy Dutch population and in Graves' patients. An exploratory study**. *Orbit* (2004) **23** 161-168. PMID: 15545129
15. Chan W, Madge SN, Senaratne T, Senanayake S, Edussuriya K, Selva D, Casson RJ. **Exophthalmometric values and their biometric correlates: The Kandy Eye Study**. *Clin Exp Ophthalmol* (2009) **37** 496-502. PMID: 19624347
16. Barretto RL, Mathog RH. **Orbital measurement in black and white populations**. *Laryngoscope* (1999) **109** 1051-1054. PMID: 10401839
17. Detorakis ET, Drakonaki E, Papadaki E, Pallikaris IG, Tsilimbaris MK. **Effective orbital volume and eyeball position: an MRI study**. *Orbit* (2010) **29** 244-249. PMID: 20812829
18. SC F, der KI PJ, HO-HO M. **Exophthalmos-producing activity in the serum and in the pituitary of patients with Cushing's syndrome and acromegaly**. *J Clin Endocrinol Metab* (1962) **22** 718-725. PMID: 13909760
19. Smolders MH, Graniewski-Wijnands HS, Meinders AE, Fogteloo AJ, Pijl H, de Keizer RJ. **Exophthalmos in obesity**. *Ophthalmic Res* (2004) **36** 78-81. PMID: 15017102
20. Saxton SN, Clark BJ, Withers SB, Eringa EC, Heagerty AM. **Mechanistic links between obesity, diabetes, and blood pressure: Role of perivascular adipose tissue**. *Physiol Rev* (2019) **99** 1701-1763. PMID: 31339053
21. Wang W, Liu S, Qiu Z, He M, Wang L, Li Y, Huang W. **Choroidal thickness in diabetes and diabetic retinopathy: A swept source OCT study**. *Invest Ophthalmol Vis Sci* (2020) **61** 29
22. Lutty GA. **Diabetic choroidopathy**. *Vision Res* (2017) **139** 161-167. PMID: 28535994
23. Endo H, Kase S, Ito Y. **Relationship between choroidal structure and duration of diabetes**. *Graefes Arch Clin Exp Ophthalmol* (2019) **257** 1133-1140. PMID: 30887100
24. Scheen AJ. **Pathophysiology of type 2 diabetes**. *Acta Clin Belg* (2003) **58** 335-341. PMID: 15068125
25. Kusminski CM, Bickel PE, Scherer PE. **Targeting adipose tissue in the treatment of obesity-associated diabetes**. *Nat Rev Drug Discov* (2016) **15** 639-660. PMID: 27256476
26. Borzenok SA, Afanasyeva DS, Gushchina MB. **[Orbital adipose tissue: Just a fat pad or terra incognita in ophthalmology]**. *Vestn Ross Akad Med Nauk* (2015) 464-467. PMID: 26710530
27. Billon N, Iannarelli P, Monteiro MC. **The generation of adipocytes by the neural crest**. *Development* (2007) **134** 2283-2292. PMID: 17507398
28. Stein JD, Childers D, Gupta S. **Risk factors for developing thyroid-associated ophthalmopathy among individuals with Graves disease**. *JAMA Ophthalmol* (2015) **133** 290-296. PMID: 25502604
29. Dorkhan M, Lantz M, Frid A, Groop L, Hallengren B. **Treatment with a thiazolidinedione increases eye protrusion in a subgroup of patients with type 2 diabetes**. *Clin Endocrinol (Oxf)* (2006) **65** 35-39. PMID: 16817816
30. Waghamare SR, Mittal S, Pathania M, Samanta R, Kumawat D, Gupta N, Mittal SK. **Comparison of choroidal thickness in systemic hypertensive subjects with healthy individuals by spectral domain optical coherence tomography**. *Indian J Ophthalmol* (2021) **69** 1183-1188. PMID: 33913856
31. Kong M, Kwun Y, Sung J, Ham DI, Song YM. **Association between systemic hypertension and macular thickness measured by optical coherence tomography**. *Invest Ophthalmol Vis Sci* (2015) **56** 2144-2150. PMID: 25736785
32. Chong RS, Chee ML, Tham YC. **Association of antihypertensive medication with retinal nerve fiber layer and ganglion cell-inner plexiform layer thickness**. *Ophthalmology* (2021) **128** 393-400. PMID: 32739337
33. Lim HB, Lee MW, Park JH, Kim K, Jo YJ, Kim JY. **Changes in ganglion cell-inner plexiform layer thickness and retinal microvasculature in hypertension: An optical coherence tomography angiography study**. *Am J Ophthalmol* (2019) **199** 167-176. PMID: 30502337
34. Donati S, Maresca AM, Cattaneo J. **Optical coherence tomography angiography and arterial hypertension: A role in identifying subclinical microvascular damage?**. *Eur J Ophthalmol* (2021) **31** 158-165. PMID: 31617414
35. Higashiyama T, Minamikawa T, Kakinoki M, Sawada O, Ohji M. **Decreased orbital fat and enophthalmos due to bimatoprost: Quantitative analysis using magnetic resonance imaging**. *PLoS One* (2019) **14** 0
|
---
title: Identification of a weight loss-associated causal eQTL in MTIF3 and the effects
of MTIF3 deficiency on human adipocyte function
authors:
- Mi Huang
- Daniel Coral
- Hamidreza Ardalani
- Peter Spegel
- Alham Saadat
- Melina Claussnitzer
- Hindrik Mulder
- Paul W Franks
- Sebastian Kalamajski
journal: eLife
year: 2023
pmcid: PMC10023155
doi: 10.7554/eLife.84168
license: CC BY 4.0
---
# Identification of a weight loss-associated causal eQTL in MTIF3 and the effects of MTIF3 deficiency on human adipocyte function
## Abstract
Genetic variation at the MTIF3 (Mitochondrial Translational Initiation Factor 3) locus has been robustly associated with obesity in humans, but the functional basis behind this association is not known. Here, we applied luciferase reporter assay to map potential functional variants in the haplotype block tagged by rs1885988 and used CRISPR-Cas9 to edit the potential functional variants to confirm the regulatory effects on MTIF3 expression. We further conducted functional studies on MTIF3-deficient differentiated human white adipocyte cell line (hWAs-iCas9), generated through inducible expression of CRISPR-Cas9 combined with delivery of synthetic MTIF3-targeting guide RNA. We demonstrate that rs67785913-centered DNA fragment (in LD with rs1885988, r2 > 0.8) enhances transcription in a luciferase reporter assay, and CRISPR-Cas9-edited rs67785913 CTCT cells show significantly higher MTIF3 expression than rs67785913 CT cells. Perturbed MTIF3 expression led to reduced mitochondrial respiration and endogenous fatty acid oxidation, as well as altered expression of mitochondrial DNA-encoded genes and proteins, and disturbed mitochondrial OXPHOS complex assembly. Furthermore, after glucose restriction, the MTIF3 knockout cells retained more triglycerides than control cells. This study demonstrates an adipocyte function-specific role of MTIF3, which originates in the maintenance of mitochondrial function, providing potential explanations for why MTIF3 genetic variation at rs67785913 is associated with body corpulence and response to weight loss interventions.
## Introduction
Over 650 million people are obese and often suffer from metabolic abnormalities, including dyslipidemia, type 2 diabetes, and hypertension (Adams et al., 2006; May et al., 2020). It is widely believed that obesity results from an interplay between genetic and environmental factors (Thomas, 2010), but the biological mechanisms behind these interactions are poorly understood.
Genetic variation (rs12016871) at MTIF3 (encoding the Mitochondrial Translation Initiation Factor 3 protein [Kuzmenko et al., 2014]) has been robustly associated with body mass index (BMI) in humans (Locke et al., 2015). Several subsequent studies have linked MTIF3 genetic variation with the response to weight loss interventions, including diet, exercise, and bariatric surgery (Papandonatos et al., 2015; Rasmussen-Torvik et al., 2015), and with weight-related effects of habitual diet (Nettleton et al., 2015). For example, analyses in two of the world’s largest randomized controlled weight loss trials (Diabetes Prevention Program [DPP] and Look AHEAD) found that homozygous minor allele carriers (rs1885988) were slightly more prone to weight gain in the control arm, yet achieved significantly greater weight loss at 12-month post-randomization and retained lost weight longer (18–36 months) than major allele carriers (Papandonatos et al., 2015). Elsewhere, the same locus has been associated with greater and more sustained weight loss following bariatric surgery (Rasmussen-Torvik et al., 2015).
Mtif3 loss in the mouse results in cardiomyopathy owing to impaired translation initiation from mitochondrial mRNAs and uncoordinated assembly of OXPHOS complexes in heart and skeletal muscle (Rudler et al., 2019). In the human hepatocyte-like HepG2 cell line, MTIF3 loss decreases the translation of the mitochondrial-encoded ATP synthase membrane subunit 6 (ATP6) mRNA without affecting cellular proliferation (Chicherin et al., 2020). No human genomic mutations leading to total MTIF3 deficiency have been reported, but the studies outlined above suggest that MTIF3 may influence obesity predisposition and weight loss potential by modulating mitochondrial function; thus, MTIF3 may play a key role in adipose tissue metabolic homeostasis, as adipocyte mitochondria not only provide ATP, but also impact adipocyte-specific biological processes such as adipogenesis, lipid metabolism, thermogenesis, and regulation of whole-body energy homeostasis (Gregoire et al., 1998; Boudina and Graham, 2014).
In this study, we aimed to experimentally dissect the molecular mechanisms that could underlie the correlation between MTIF3 genetic variation and weight loss intervention outcomes. We hypothesized that among the common genetic variants in MTIF3, one (or more) is causal for altered MTIF3 expression. Secondly, we hypothesized that MTIF3 content in human white adipocytes influences adipocyte-specific, obesity-related traits under basal and perturbed metabolic conditions. For the latter, we used glucose restriction to mimic the effects of in vivo lifestyle interventions focused on energy restriction and expenditure.
## rs67785913 is a regulatory variant for MTIF3 expression
The MTIF3 rs1885988 C allele is associated with enhanced weight loss and weight retention following lifestyle intervention trials in DPP and Look AHEAD cohorts (Papandonatos et al., 2015). In the GTEx database, the rs1885988 associates with an eQTL in subcutaneous fat (Figure 1A), with C allele carriers having significantly higher MTIF3 expression (normalized effect size: 0.15, $$p \leq 0.0000032$$).
**Figure 1.:** *Identification of rs67785913 as a causal cis-eQTL for MTIF3.(A) Violin plot of MTIF3 expression in subcutaneous adipose tissue for rs1885988 from Genotype-Tissue Expression (GTEx) Project eQTL. (B) Same as in (A), but for rs67785913. (C) Representative Sanger sequencing traces of rs67785913 CTCT/CTCT and CT/CT clones obtained after CRISPR/Cas9-mediated allele editing and single-cell cloning. (D) Normalized Z-score plot of luciferase reporter assays using vectors carrying different DNA fragments of the MTIF3 gene cloned into pGL4.23 luciferase reporter vector. Hypothesis testing was performed by comparing the transcriptional enhancer activity of each of the 12 vectors (F1–12) to the empty vector (minP). All data were plotted as mean ± standard deviation (SD), n = 4 independent experiments, p values are presented in each graph; ordinary one-way analysis of variance (ANOVA) was used for statistical analysis. (E) Relative MTIF3 expression (mRNA) in rs67785913 allele-edited cells 2 days before, at, or 2 days post-differentiation induction (day −2, 0, and 2, respectively). n = 3 clonal populations for CTCT/CTCT genotype, n = 5 clonal populations for CT/CT genotype, error bars show SD. (F) as in (E), but for GTF3A (mRNA) expression. Two-tailed Student’s t-test was used; p values are presented in each graph.*
To experimentally validate and fine map the potential causal DNA variation in the haplotype block tagged by rs1885988, we looked up all tightly linked (r2 > 0.8) single-nucleotide polymorphisms (SNPs) in HaploReg database v4.1 (Ward and Kellis, 2012). We then PCR-amplified and cloned 12 DNA fragments from that haploblock, altogether comprising the linked SNP loci, into luciferase reporter plasmids. As shown in Figure 1D, by comparing the luciferase signals with minimal promoter (minP) construct, only one DNA fragment (F11), encompassing the rs67785913 locus, could enhance luciferase transcription. Coincidentally, the rs67785913 also shows an eQTL effect on MTIF3 expression in subcutaneous adipose tissue in GTEx database, with the major CT allele associated with lower expression than the minor CTCT allele (normalized effect size: −0.16, $$p \leq 3.0$$ × 10−8) (Figure 1B). To demonstrate an allele-specific regulatory effect on MTIF3 expression, we then used CRISPR-Cas9 to substitute the major CT for the minor CTCT allele at the rs67785913 locus in the pre-adipocyte hWAs cell line. Due to rather low CRISPR editing efficiency of that locus, we needed to genotype over 700 single-cell clones to obtain five CT/CT and three CTCT/CTCT clones without random indels, as confirmed by Sanger sequencing (Figure 1C). We then examined MTIF3 expression in these clones at pre- and post-adipogenic differentiation induction, and found rs67785913 CTCT/CTCT to confer higher MTIF3 expression at all time points (Figure 1E), although without apparent change on adipogenic differentiation markers (Figure 1—figure supplement 1). As rs67785913 also correlates with an altered GTF3A expression in other tissues (e.g., muscle, lung), we also detected, but found no apparent difference in GTF3A expression in rs67785913-edited cells (Figure 1F).
## Generating inducible Cas9-expressing pre-adipocyte cell line (hWas-iCas9)
Next, we intended to use the rs67785913-edited cells in functional genomics experiments to examine the phenotypic consequences of the eQTL. To conduct meaningful studies of gene × environment interaction, it is desirable to use similarly differentiated cells with comparable baselines (e.g., similar triglyceride or mitochondrial content). Unfortunately, marginally different passage numbers between control and experimental groups can confound adipogenic differentiation. This problem can originate during single-cell cloning to create genetic knockouts/knockins, and became apparent with our rs67785913 allele-edited cells. While the mean values of adipogenic markers were similar in both rs67785913 genotypes (Figure 1—figure supplement 1), the variation between clones of the same genotype precluded the use of these cells in gene × environment studies. To circumvent this, we instead established an inducible Cas9-expressing pre-adipocyte cell line that allowed us to knockout MTIF3 after completed adipogenic differentiation. As illustrated in Figure 2, we co-transfected hWAs with two plasmids: one encoding piggyBac transposase, and the other carrying piggyBac transposon-flanked doxycycline-inducible Cas9 and constitutively expressed puromycin resistance genes. In this setup, piggyBac transposase drives the integration of the piggyBac-transposon-flanked genes, and transgenic cells are then selected and expanded in puromycin-supplemented culture medium. We have thus obtained an hWAs cell line with doxycycline-inducible Cas9 expression and maintained adipogenic differentiation capacity (henceforth called hWAs-iCas9). The Cas9-expressing differentiated cells could then be transfected with relatively low molecular weight synthetic single guide RNAs (sgRNAs) that complex with intracellularly expressed Cas9 and target the gene exon of interest to generate random indels (in essence, gene knockouts). We used this method here to determine the functional role of MTIF3 in adipocyte biology.
**Figure 2.:** *The workflow of establishing hWAs-iCas9 cell line and its application in studying MTIF3 and environment interactions in vitro.*
## Generation of MTIF3 knockout in hWAs-iCas9 mature adipocytes
To investigate the role of MTIF3 in human adipocyte development and energy metabolism we generated stable MTIF3 knockouts in differentiated hWAs-iCas9 adipocytes. We designed Cas9-specific sgRNA to generate random indels in the exon expressed in all three MTIF3 protein-encoding transcripts (Figure 3A) and obtained a >$80\%$ reduction in MTIF3 protein levels in every experiment, as assessed by western blotting (Figure 3B–D, Figure 3—figure supplement 1, and Figure 4A, B). To assess off-target effects of CRISPR-Cas9, we also performed T7EI assays on PCR-amplified top 5 predicted off-target sites and did not observe any detectable off-targeting (data are not shown).
**Figure 3.:** *MTIF3 perturbation in mature adipocytes does not affect adipocyte-specific protein expression or total triglyceride content.(A) An illustration of Cas9-specific single guide RNA (sgRNA)-binding site in the exon expressed in all three MTIF3 protein-encoding transcripts. (B) Representative Sanger sequencing of control and knockout hWAs mature adipocytes. (C) Immunoblots of adipocyte markers in scrambled control and MTIF3 knockout adipocytes, n = 5 independent experiments. (D) Quantitative analysis of MTIF3 band densities in (C). (E) Quantitative analysis of ACC, FABP4, and FAS band densities in (C). (F) Representative Oil-red O staining images of control and MTIF3 knockout in hWAs mature adipocytes. Scale bar is 200 µm. (G) Total triglyceride content in scrambled control (SC) and MTIF3 knockout (KO) cells. n = 3 independent experiments. Error bars show standard deviation in all plots. Statistical analysis was performed using two-tailed Student’s t-test, p values are presented in each graph. Uncropped blot images for (C) and raw.scn data files can be found in Figure 3—source data 1.
Figure 3—source data 1.Raw data files for western blots shown in Figure 3C.* **Figure 4.:** *MTIF3 perturbation in mature adipocytes disrupts mitochondrial gene expression and OXPHOS complex assembly.(A) Immunoblots of mitochondrial genome-encoded proteins in scrambled control and MTIF3 knockout adipocytes. (B) Quantitative analysis of band densities in (A). (C) qPCR for mitochondrial gene expression in scrambled control and MTIF3 knockout adipocytes, n = 5 independent experiments. (D) Relative mitochondrial DNA content in scrambled control and MTIF3 knockout adipocytes, n = 5 independent experiments. (E) Immunoblots of mitochondrial OXPHOS complex assembly after Blue Native-PAGE electrophoresis, n = 4 independent experiments. (F) Quantitative analysis of band densities in (E). Error bars show standard deviation in all plots. Statistical analysis was performed using two-tailed Student’s t-test, p values are presented in each graph. Uncropped blot images for (A) and raw.scn data files can be found in Figure 4—source data 1. Uncropped blot images for (E) and raw.scn data files can be found in Figure 4—source data 2.
Figure 4—source data 1.Raw data files for western blots shown in Figure 4A.
Figure 4—source data 2.Raw data files for western blots shown in Figure 4E.*
## MTIF3 knockout in mature adipocytes does not affect adipogenic marker or lipid content
Although the MTIF3 knockout in hWAs-iCas9 cells was generated after the cells were differentiated, we wanted to ensure the genetic perturbation did not affect adipogenic markers or triglyceride content, as that could confound results from downstream functional studies. Incidentally, we observed that the quantities of the adipogenic markers, including ACC, FABP4, and FAS were comparable in control and MTIF3 knockout cells (Figure 3C, E; see also Figure 3—figure supplement 1). Similarly, there were no apparent differences in Oil-red O or total triglyceride content (Figure 3F, G).
## MTIF3 knockout disrupts mitochondrial DNA-encoding gene and protein expression, mitochondrial content, as well as mitochondrial OXPHOS assembly in hWAs-iCas9 adipocytes
MTIF3 is a mitochondrial translation initiation factor; thus, we examined the effects of MTIF3 ablation on differentiated hWAs adipocyte mitochondrial respiration chain. Assessed by western blotting, the MTIF3 knockout cells had significantly decreased COX II (subunit of OXPHOS complex IV) and ND2 (subunit of OXPHOS complex I), trending decrease of CYTB (subunit of OXPHOS complex III), and unchanged ATP8 (subunit of OXPHOS complex V) content (Figure 4A, B). Moreover, using qPCR, we observed an altered expression of several mitochondrial DNA-encoding genes. Specifically, MTIF3 deficiency led to higher expression of MT-ND1, MT-ND2, a trending increase of MT-ND4, and lower expression of MT-ND3, and MT-CO3 (Figure 4C). In addition, we also found significantly reduced mitochondrial content in MTIF3 knockout adipocytes (Figure 4D). Taken together, the above data suggest MTIF3 knockout disrupts mitochondrial DNA-encoding gene and protein expression.
Next, we hypothesized the above observations could have originated from the insufficient MTIF3 supply during OXPHOS complex assembly (a role previously ascribed to MTIF3 [Rudler et al., 2019]). To test this, we used Blue Native-PAGE to examine OXPHOS complexes in mitochondria isolated from MTIF3 knockout adipocytes. MTIF3 deficiency led to decreased complex III2/IV2 and IV1, and a trending decreased complex V/III2 + IV1 assembly. In contrast, OXPHOS complex II assembly was significantly increased in MTIF3 knockout cells (Figure 4E, F). Interestingly, we also observed faster-migrating undefined bands in MTIF3 knockout adipocytes (Figure 4E), which could be single chain proteins, or mistranslation or degradation products. Lastly, The OXPHOS complex I + III2 + IVn appeared to be less abundant in MTIF3 knockout mitochondria, although the bands appeared more diffuse and could not be quantified (Figure 4E).
## MTIF3 knockout affects mitochondrial respiration in hWAs-iCas9 adipocytes
Having established the role of MTIF3 in adipocyte mitochondria OXPHOS complex assembly, and in mitochondrial gene expression, we then investigated the mitochondrial function in MTIF3-ablated differentiated hWAs adipocytes using Seahorse Mito Stress Test. Additionally, to avoid potential cofounders caused by the high glucose content in the differentiation medium, we adapted the cells to 1 g/l glucose growth medium for 3 days before running the assay. As shown in Figure 3, MTIF3 knockout cells exhibited lower basal oxygen consumption rate (OCR), as well as lower ATP-forming capacity, the latter estimated by calculating OCR decrease after blocking ATP synthase with oligomycin (Figure 5A–C). MTIF3 knockout cells also showed a trending decrease in maximal respiration OCR (Figure 5D). Furthermore, both MTIF3 knockout and control cells, had comparable proton leak OCR, non-mitochondrial respiration OCR and coupling efficiency (Figure 5E–G).
**Figure 5.:** *Cellular mitochondrial respiration in hWAs adipocytes.(A) The average oxygen consumption rate (OCR) traces during basal respiration, and after addition of oligomycin, FCCP, and rotenone/antimycin A. (B) Basal respiration OCR, n = 4 different cell passages. (C) ATP production OCR, n = 4 different cell passages. (D) Maximal respiration OCR, n = 4 different cell passages. (E) Proton leak OCR, n = 4 different cell passages. (F) Non-mitochondrial respiration OCR, n = 4 different cell passages. (G) Coupling efficiency, n = 4 different cell passages. Error bars show standard deviation. Statistical analyses were performed using paired Student’s t-test in each condition, p values are presented in each graph.*
## MTIF3 knockout affects hWAs-iCas9 adipocyte endogenous fatty acid oxidation
Next, we used Seahorse to assess the endogenous fatty acid oxidation in MTIF3 knockout versus control cells treated with etomoxir (an inhibitor of carnitine palmitoyl transferase). We found that MTIF3 ablation mimics the effect of etomoxir on basal endogenous fatty acid oxidation OCR. Furthermore, while etomoxir decreases basal fatty acid oxidation OCR in control cells, it does not markedly decrease it in MTIF3 knockout cells (Figure 6A, B).
**Figure 6.:** *MTIF3 perturbation affects adipocyte fatty acid oxidation.(A) A representative Seahorse oxygen consumption rate (OCR) trace for endogenous fatty acid oxidation assay. MTIF3 knockout and scrambled control adipocytes were treated with or without etomoxir for 15 min before the assay. Following the basal OCR measurement, oligomycin, FCCP (carbonyl cyanide-p-trifluoromethoxyphenylhydrazone), and rotenone + antimycin A were added sequentially to measure the detection of ATP production OCR, maximal respiration OCR and non-mitochondrial respiration OCR. (B) Basal endogenous fatty acid oxidation OCR in scrambled control (SC) and MTIF3 knockout (KO) adipocytes, n = 4 independent experiments. (C) Upper panel: workflow of glucose restriction in differentiated adipocytes; Lower left panel: total triglyceride content in scrambled control (SC) and MTIF3 knockout (KO) adipocytes in 25 mM glucose conditions; Lower right panel: triglyceride content in adipocytes cultured in glucose-restricted conditions (5, 3, and 1 mM) relative to adipocytes cultured in 25 mM glucose, n = 4 independent experiments. (D) Z-score-normalized data for glycerol release in scrambled control and MTIF3 knockout adipocytes under basal, insulin-stimulated, and isoproterenol-stimulated conditions, n = 4 independent experiments. (E) qPCR for mitochondrial and adipocyte-related gene expression in scrambled control and MTIF3 knockout adipocytes. Error bars show standard deviation in all plots. Statistical analysis was performed using two-tailed Student’s t-test, p values are presented in each graph.*
## MTIF3 knockout affects hWAs-iCas9 adipocyte triglyceride content after glucose restriction challenge
To mimic the interactions between MTIF3 content and dietary intervention on weight change, we generated hypertrophic control and MTIF3 knockout hWAs-iCas9 adipocytes and then used glucose-limited medium, not supplemented with free fatty acids (FFAs), to mimic energy restriction in vivo (schematic shown in Figure 6C). Triglyceride content decreased both in control and MTIF3 knockout cells after 3 days of different levels of glucose restriction when compared with 25 mM glucose medium. Interestingly, a more extensive decrease in triglyceride content occurred in control cells cultured in 1 mM glucose medium ($$p \leq 0.053$$), and a similar trending decrease, albeit with higher coefficient of variation, occurred in 3 and 5 mM glucose medium (Figure 6C).
## MTIF3 knockout does not affect lipolysis-mediated glycerol release in hWAs adipocytes
Owing to the effects of MTIF3 ablation on triglyceride content and on fatty acid oxidation, described above, we then examined the effects of MTIF3 knockout on lipolysis. We measured basal, insulin-attenuated, and isoproterenol-stimulated glycerol release in differentiated hWAs cells. As shown in Figure 6D, in all three conditions, glycerol release in control and MTIF3 knockout cells was comparable. In addition, basal glycerol release in glucose-restricted conditions was similar (Figure 6—figure supplement 1), and significantly reduced in low glucose versus high glucose assay medium.
## MTIF3 knockout affects mitochondrial function- and fatty acid oxidation-related gene expression
Next, we examined how MTIF3 ablation affects the gene expression programmes pertinent to mitochondrial function, fatty acid oxidation, lipolysis and lipid catabolism. As shown in Figure 6E, MTIF3 knockout cells had decreased expression of the mitochondria-related MT-CO1, and the fatty acid oxidation-related ACADM and ACAT1, but unchanged expression of other genes involved in mitochondrial function and lipid metabolism (TFAM, TOMM20, PRDM16, CPT1B, ABHD5, PNPLA2, and ACACB).
## MTIF3 knockout results in glucose level-depending alterations in metabolism
Considering the observed effect of MTIF3 ablation on mitochondrial function and fatty acid oxidation, we assessed the metabolite profile in MTIF3 knockout cells. Using combined GC/MS and LC/MS metabolite profiling resulted in relative quantification of 110 metabolites. First, we analyzed metabolite profiles at a global level using PCA. The score plot reveals a clear systematic difference in the metabolite profile between cells in 25 mM glucose versus in glucose restriction (Figure 7A). Interestingly, differences between MTIF3 knockout and control cells at 25 mM glucose are observed along principal component 1 (PC1), whereas differences between genotypes at glucose restriction are observed along PC2, suggesting the effect of MTIF3 ablation to depend on the calorie level. Next, to identify alterations in metabolite levels underlying this differential response, we analyzed data using orthogonal projections to latent structures discriminant analysis (OPLS-DA) separately at 25 mM glucose condition (two components R2 = 0.82, Q2 = 0.66) and at glucose restriction (two components, R2 = 0.95, Q2 = 0.52). These analyses revealed systematic differences between genotypes at both growth conditions (Figure 7 B, C). Next, to examine whether the differences between genotype depended on growth condition, we combined the correlations from the two OPLS-DA models into a shared and unique structures plot (Figure 7D). These analyses revealed levels of intermediates in cytosolic metabolic pathways connected to the glycolysis, such as glycerate 3-phosphate, glycerol 2-phosphate, UDP-N-acetylglucosamine, and ribose 5-phosphate, to be lower in MTIF3 knockout cells at both glucose concentrations. Interestingly, levels of fatty acids, ranging from 9 to 17 carbons and including several odd-chain fatty acids, were lower in the MTIF3 knockout cells only at 25 mM glucose condition. At glucose-restricted conditions, levels of both essential and non-essential amino acids were lower in the knockout. Finally, we analyzed data using two-way analysis of variance (ANOVA), incorporating glucose concentration and genotype, thereby providing information on effects at the individual metabolite level. These analyses revealed 18 and 20 significantly different metabolites between control and MTIF3 knockout cells at 25 mM glucose condition and glucose-restricted conditions, respectively (q <0.05). These included ribose 5-phosphate, glycerate 3-phosphate, glycerol 2-phosphate, and glycerol 3-phosphate (Figure 7E).
**Figure 7.:** *Mass spectrometry-based metabolomics data for control (SC) and MTIF3 knockout (KO) cells in 25 mM glucose (NF, normal feeding) and 5 mM glucose (GR, glucose-restricted) conditions.(A) Principal component analysis (PCA) score plot displaying the discrimination between MTIF3 knockout and control cells in normal and glucose-restricted conditions (PC1: 28%, PC2: 19%). (B) Orthogonal projections to latent structures discriminant analysis (OPLS-DA) score plot showing classification of MTIF3 knockout and control cells in 25 mM glucose condition. (C) OPLS-DA score plots showing classification of MTIF3 knockout and control cells in glucose-restricted condition. (D) Shared and unique structures (SUS) plot, based on OPLS-DA models in (B, C), showing glucose concentration-dependent differences between MTIF3 knockout and control cells. (E) Box plots showing the abundance of some of the significantly altered metabolites in normal and MTIF3 knockout cells in normal and glucose-restricted conditions. Statistical analysis was performed using two-way analysis of variance (ANOVA) test, p values are presented in each graph.*
## Discussion
Excessive weight gain caused by dietary excess, and its effects on adipocyte lipid metabolism, can cause life-threatening disease (Appleton et al., 2013; Denis and Obin, 2013; Chu et al., 2017; Lotta et al., 2018). Findings from clinical trials (Papandonatos et al., 2015), a bariatric surgery case series (Rasmussen-Torvik et al., 2015) and epidemiological cohorts (Nettleton et al., 2015) showed the MTIF3 variation modulates weight loss-promoting exposures on body weight (see also UK *Biobank analysis* in Supplementary file 1b). Here, we validated MTIF3 rs1885988 C allele correlates with higher MTIF3 expression in subcutaneous fat tissue, and our in vitro luciferase reporter assay and CRISPR-Cas9 genome editing results revealed that the tightly linked rs67785913 variant is likely to be the actual eQTL for MTIF3 expression (Figure 1).
Pre-adipocyte cell lines have been used extensively for studying adipogenic differentiation (Ruiz-Ojeda et al., 2016), but their application for lipid metabolism studies has been limited, especially in the context of gene–environment interaction. This is largely due to the variation of differentiation capacity across cell passages (Poulos et al., 2010) and differential genetic effects on adipocyte differentiation (Kamble et al., 2020), which can alter baseline phenotypes in differentiated cells. Therefore, we established an inducible Cas9-expressing human pre-adipocyte cell line (hWAs-iCas9), which enabled us to generate gene knockout of interest in differentiated adipocytes, thus circumventing these limitations (Figure 2). Using the inducible knockout cell model, we then tested interactions between MTIF3 and environmental changes (lifestyle mimetics). Our data reveal that MTIF3 deficiency mediated disrupted mitochondrial respiration, probably as a consequence of decreased OXPHOS complex assembly. This led to perturbed cellular functions, including reduced fatty acid oxidation and a trending increase in intracellular triglyceride content. These data indicate that MTIF3 plays an important role in lipid metabolism in human adipocytes.
In adipose tissue, mitochondria play an essential role not only by ensuring ATP supply but also by triggering cellular signaling pathways that require reactive oxygen species generated by OXPHOS complexes I and III (Tormos et al., 2011). These, and complexes IV and V, are partially encoded by mitochondrial DNA (Maechler and Wollheim, 2001). The post-transcriptional rate-limiting translation step can be promoted by MTIF3, as it facilitates initiation complex formation on mitochondrial 55S ribosomes in the presence of MTIF2, fMet-tRNA, and poly(A,U,G) (Bhargava and Spremulli, 2005). Previous studies have shown that loss of MTIF3 results in an imbalanced assembly of OXPHOS complexes in muscle and heart in mouse (Rudler et al., 2019) and decreased translation of ATP6 mRNA in hepatocyte-like HepG2 cells (Chicherin et al., 2020). Our data show that in adipocytes MTIF3 deficiency results in lower content of mitochondrial DNA-encoding proteins and altered expression of several mitochondrial DNA-encoding genes. Furthermore, it leads to disassembly of mitochondrial respiration OXPHOS complex, along with impaired mitochondrial respiration rate (Figures 4 and 5). Altogether, our results suggest that MTIF3 vastly affects the mitochondrial electron transport chain.
Mitochondrial dysfunction in white adipose tissue has been frequently associated with obesity (Kaaman et al., 2007), with the presumed mechanisms being decreased fatty acid oxidation and ATP production (reviewed in Heinonen et al., 2020). We have found that one causal link connecting these factors may be the MTIF3 content (Figures 5 and 6). Furthermore, a previous study described an inverse relationship between mitochondrial capacity and weight change (Jokinen et al., 2018); thus, conceivably, any genetic variation-modulated change in MTIF3 expression could influence a dietary intervention outcome. We attempted to test this hypothesis in vitro, exposing hypertrophic adipocytes to glucose restriction, thereby mimicking weight loss-promoting exposures in vivo. We found that MTIF3-deficient adipocytes exposed to glucose restriction challenge responded less through changed triglyceride content, indicating limited capacity for lipid metabolism under glucose restriction (Figure 6). Intriguingly, we did not observe altered lipolysis (glycerol release) in MTIF3 knockouts, either in the presence or absence of insulin and isoproterenol, or during glucose restriction. Here, we speculate that MTIF3 knockouts, having reduced fatty acid oxidation, re-esterify the released fatty acids to triglycerides, and thus retain more triglycerides during glucose restriction.
Mitochondrial function is robustly associated with insulin sensitivity (Böhm et al., 2020; Pietiläinen et al., 2008; Heinonen et al., 2015), both of which can be improved through lifestyle intervention aiming at weight loss (Phielix et al., 2010; Toledo et al., 2007; Larson-Meyer et al., 2006; Civitarese et al., 2007). Furthermore, impaired adipocyte differentiation derived from mitochondrial dysfunction in adipose tissue often associates with insulin resistance in humans and animal models (Sakers et al., 2022; Zhu et al., 2022). In our study, we did not observe impaired adipocyte differentiation due to MTIF3 deficiency or rs67785913 eQTL. These results, however, may not be directly translatable to in vivo conditions, in part because in vitro differentiation protocols employ relatively highly concentrated adipogenic compounds. Further in vivo studies are therefore needed to establish any link between MTIF3 content and insulin resistance. Similarly, long-term in vivo studies on the effect of altered MTIF3 expression on body weight are warranted to illuminate the translatability of the short-term or low effect size in vitro experiments (e.g., Figures 5 and 6). We envisage that the MTIF3 effect on adipocyte metabolism, while not dramatic in some of the presented data, could translate into larger effect size over time in vivo. Here, MTIF3 effect on metabolism in other tissues (e.g., muscle) would further contribute to body weight regulation. Lastly, one should bear in mind that our in vitro data were generated in cells highly depleted of MTIF3, and the extent to which a more moderate MTIF3 deficiency in vivo (e.g., conferred by common genetic variants) influences adipogenic differentiation or long-term diet-induced weight loss is currently unknown.
Finally, the metabolomic analysis added another dimension to the role of MTIF3 in regulating adipocyte metabolism (Figure 7). The lower content of glycolysis intermediates and odd-chain fatty acids in MTIF3 knockout adipocytes could indicate blunted lipogenesis, while the decreased essential and non-essential amino acid level in MTIF3 knockout cells under glucose restriction could be an adaption to the lower energy supply owing to impaired mitochondrial function (Hansson et al., 2004). The metabolite data, and the earlier described lower fatty acid oxidation capacity of MTIF3-deficient cells, suggest that MTIF3 plays a vital role in triglyceride metabolism in adipocytes, and provides further insight into the previously reported role of this protein on weight loss induced by dietary intervention (Papandonatos et al., 2015).
In summary, we experimentally demonstrated that the common genetic variant rs67785913 is a functional polymorphism that causes modulated MTIF3 expression. Our functional genomics studies demonstrate that MTIF3 is essential for mitochondrial electron transport chain complex assembly, mitochondrial function, and lipid metabolism in human adipocyte cell lines. MTIF3 content may also influence adipocyte triglyceride content in conditions that mimic dietary restriction, reflecting a gene x environment interaction. Since our findings show that higher MTIF3 content in adipocytes increases mitochondrial function, this helps explain the previously observed interaction between MTIF3 locus and weight loss interventions. Furthermore, this suggests, although it remains to be demonstrated, that people who carry genetic variants that increase MTIF3 expression (e.g., the rs67785913 CTCT allele) may benefit more from lifestyle interventions targeting weight loss.
Lastly, we established a novel, efficient, and simple method to generate gene knockouts in differentiated adipocyte cell lines, and the method should be suitable for generating other gene knockouts.
## Allele-specific MTIF3 expression in subcutaneous adipose tissue
Data and plots presented in this manuscript for allele-specific MTIF3 expression in subcutaneous adipose tissue were obtained from GTEx Portal (https://www.gtexportal.org/home/) on $\frac{10}{5}$/2022.
## Cell culture
A human white pre-adipocyte cell line (hWAs) was previously isolated and immortalized from human subcutaneous white adipose tissue of a female subject, aged 56 with a BMI of 30.8. *The* generation and characterization of hWAs were described previously (Xue et al., 2015), and the cell line was kindly shared by Professor Yu-Hua Tseng (Joslin Diabetes Center, Harvard Medical School, USA). hWAS cells. For expansion, cells were cultured in 25 mM Dulbecco’s Modified Eagle Medium (DMEM) with GlutaMAX (10566016, Thermo Fisher Scientific), $10\%$ fetal bovine serum (FBS; HyClone, GE Healthcare, Uppsala, Sweden), and $1\%$ (100 U/ml) penicillin/streptomycin (15140122, Thermo Fisher Scientific). The cells were passaged at $90\%$ confluence, and tested negative for mycoplasma.
## DNA isolation and luciferase reporter assays
Genomic DNA was isolated from hWAs cells using DNeasy Blood and Tissue kit (69506, QIAGEN) according to the manufacturer’s manual. To fine map the transcriptional regulatory regions in the MTIF3 locus, we first identified the common genetic variants which were in tight linkage disequilibrium (r2 ≥ 0.8) with the lead variant rs1885988 in HaploReg v4.1 (Ward and Kellis, 2012). The thus identified 31 SNPs were tiled down into 12 DNA segments of the MTIF3 gene, as shown in Supplementary file 1a. These segments were then PCR amplified from hWAs DNA, all ranging from 700 to 1600 bp in size (depending on PCR primer design constraints), and with all SNP loci located several hundred bp from the ends of each fragment. The PCR primers were also designed to include flanking KpnI and EcoRV sites to allow cloning into the pGL4.23 minimal promoter luc2 luciferase reporter vector (Promega). For the reporter assays, hWAs were seeded into 96-well plates, and on the following day transfected with 95 ng of the pGL4.23 vectors and 5 ng pGL4.75 CMV-Renilla reporter vectors (for normalization), using Lipofectamine 3000 (Thermo Fisher Scientific), in technical duplicates. Two days after transfection, the luc2 and Renilla signals were detected using Dual-Glo Stop&Glo reagents (E2920, Promega). The averages of technical duplicates were used to calculate luc2:Renilla ratios, which were then Z-score normalized to allow statistical evaluation across four independent experiments.
## gRNAs and ssDNA design for CRISPR/Cas9 mediated editing of rs67785913 in hWAs cells
To edit the rs67785913 CT allele to the minor CTCT allele in hWAs cell genome, CRISPR/Cas9 D10A nickase (Alt-R S.p. Cas9 D10A Nickase V3, IDT) and two sgRNAs, and an ssDNA donor template were used. The sgRNA spacer sequences were: 5′-TTCAATAAGAAATTCCTCAA-3′ and 5′-GAAGAAAAAGGGGGGACACG-3′. The ssDNA sequence was 5′-TGTGGACTCGCAGTCTGCCCTTGAGGAATTTCTTATTGAAGAAGAAAAAGAGGGGGGACACGGGGCCCAGACCCCCAGCACCCGGCTTTCGAGCAGGCTC-3′. All oligonucleotides, sgRNAs, and ssDNA were purchased from Integrated DNA Technologies. The transfection was performed using Nucleofector 2b device (program A-033) (Lonza, Sweden) in nucleofector reagent L (Lonza, Sweden) mixed with 5 × 105 hWAs cells, 120 pmol Cas9 nickase, 104 pmol sgRNA, and 300 pmol ssDNA. To increase the homology directed repair (HDR) editing efficiency, cells were incubated at 32°C for 2 days in growth medium containing 30 μM HDR enhancer (Alt-R HDR Enhancer V2, IDT). Subsequently, cells were transferred to 37°C for 3 days. For single-cell cloning, the hWAs cells were seeded at low density (2 cells/well in a 96-well plate) and allowed to expand for 3 weeks. Then the genomic DNA was extracted using QuickExtract DNA Extraction Solution (Lucigen) from the apparent single-cell clonal populations. To identify the allele-edited homozygous clones, PCR was used to amplify the DNA fragment surrounding rs67785913 using primer pairs as below: Forward 5′–3′ GATTTGCAGGTGAGCAGACA, Reverse 5′–3′ ACTTGGAAATGGCCAAGATG; the amplicon was then subjected to Sanger sequencing to confirm the DNA sequence of each clone.
## Generation of inducible CRISPR/Cas9-expressing hWAs cell line (hWAs-iCas9)
hWAs cells were first seeded at 80,000 cells/well in 6-well plates and transfected with 200 ng Super PiggyBac transposase (PB210PA-1, System Biosciences) and 500 ng pPB-rtTA-hCas9-puro-PB plasmid (kind gift from Dr. William Pu) (Wang et al., 2017) using Lipofectamine 3000 (Thermo Fisher Scientific). The plasmid carries a doxycycline-inducible promoter driving the expression of Cas9, and a puromycin resistance gene, all flanked by piggyBac transposon integration sequences. After 2 days, the transfected cells were selected and expanded for 3 weeks in growth medium with 1 μg/ml puromycin, to obtain cells with genomically integrated inducible Cas9 construct.
## Differentiation of hWAs-iCas9 pre-adipocytes into mature adipocytes
hWAs-iCas9 pre-adipocytes were seeded into 24- or 96-well plates at the density of 40,000 or 8000 cells/well, respectively. After 3 days, the cells reached confluency and were then incubated for 12 days with the differentiation cocktail, with medium changes every 3 days, as described before (Shamsi and Tseng, 2017). To increase the accumulation of lipid droplets, 30 μM FFA (Linoleic Acid-Oleic Acid-Albumin) (L9655, Sigma-Aldrich) was added to the differentiation medium (Aprile et al., 2020).
## CRISPR/Cas9 guide RNA design and off-targeting check
*To* generate MTIF3 knockout adipocytes, guide RNA spacer sequence targeting MTIF3 exon 5, expressed in all known MTIF3 protein-encoding transcripts (as reported at https://www.ensembl.org), was selected. The spacer sequence was 5′-GCAATAGGGGACAACTGTGC-3′, and full-length sgRNA was purchased from IDT. Furthermore, the hWAs genomic sequence surrounding the gRNA-binding site was amplified by PCR using the primers 5′-CCACTTGTCTTGGGGACAGT-3′ and 5′-CTGGGAATGGTGGTTGAATC-3′, then analyzed by Sanger sequencing to ensure sequence match between gRNA spacer and the intended target locus. The potential off-target sites were predicted using CRISPR-Cas9 guide RNA design checker (https://eu.idtdna.com), and the genomic regions surrounding the top 5 off-target sites were PCR amplified from the genomic DNA extracted from MTIF3-knockout and scramble control cells. The amplicons were then analyzed for any heteroduplexes generated by off-targeting using T7EI assay (IDT, Alt-R Genome Editing Detection Kit).
## sgRNA transfection and MTIF3 knockout in mature adipocytes
After 12 days of differentiation, Cas9 expression was induced in mature adipocytes by adding 2 μg/ml doxycycline to the growth medium. On the following day, 30 nM pre-designed sgRNA was delivered into the cells using Lipofectamine RNAiMAX (13778075, Thermo Fisher Scientific) according to the manufacturer’s protocol; in parallel, 30 nM negative control crRNA (1072544, IDT) was used to transfect the scrambled control cells. One day post-transfection, cells were washed with phosphate-buffered saline (PBS) and incubated in normal growth medium for at least 3 days before functional assays carried out.
## Oil-red O staining
After MTIF3 knockout, the differentiated white adipocytes were washed twice with PBS and fixed for 10–20 min with $4\%$ buffered formalin at room temperature. The cells were then stained with Oil-red O solution for 30 min at room temperature, followed by five washes with distilled water. The stained cells were visualized using light microscopy.
## Glucose restriction challenge for hWAs-iCas9 adipocytes
The hWAs-iCas9 adipocytes were firstly differentiated to mature adipocytes as described above (in FFA-supplemented medium), then the MTIF3 knockouts and scrambled controls were generated, also as described above. After 3 days, the mature adipocytes were incubated in DMEM medium (11966025, Thermo Fisher Scientific) without FFA, and supplemented with different glucose concentrations (5, 3, and 1 mM) for the glucose restriction test. The cells were then incubated for 3 days, and the triglyceride content was determined as described above.
## RNA isolation and qPCR gene expression assays
Total RNA was extracted from cells using RNeasy plus Kit (74034, QIAGEN) together with Qiazol reagent (79306, QIAGEN). RNA purity was assessed using Nanodrop (Nanodrop, Wilmington, USA), and cDNA was synthesized using SuperScript IV VILO Master Mix (11756500, Thermo Fisher Scientific). Then, RT-qPCR was performed on ViiA7 qRT-PCR system (PE Applied Biosystems, Foster City, CA, USA), using predesigned Taqman assays following the manufacturer’s instructions. The Taqman assays (Thermo Fisher Scientific, Uppsala, Sweden) were: MTIF3 (Hs00794538_m1), GTF3A (Hs00157851_m1), ADIPOQ (Hs00977214_m1), PPARG (Hs01115513_m1), CEBPA (Hs00269972_s1), SREBF1 (Hs02561944_s1), FASN (Hs01005622_m1), TFAM (Hs01073348_g1), MT-CO1 (Hs02596864_g1), PRDM16 (Hs00223161_m1), TOMM20 (Hs03276810_g1), CPT1B (Hs00189258_m1), ACADM (Hs00936584_m1), ACAT1 (Hs00608002_m1), ABHD5 (Hs01104373_m1), PNP1A2 (Hs00386101_m1), ACACB (Hs01565914_m1), MT-ND1 (Hs02596873_s1), MT-ND2 (Hs02596874_g1), MT-ND3 (Hs02596875_s1), MT-ND4 (Hs02596876_g1), MT-CO2 (Hs02596865_g1), MT-CO3 (Hs02596866_g1), HPRT-1 (Hs99999909_m1), TBP (Hs00427620_m1), and RPL13A (Hs03043885_g1). The relative gene expression was calculated using the delta Ct method, and the target gene expression was normalized to the mean Ct of three reference genes HPRT-1, TBP, and RPL13A.
## Western blotting
Cells were washed twice with ice-cold PBS and lysed in $1\%$ sodium dodecyl sulfate buffer for 10 min, then passed through a QIAshredder (79654, QIAGEN) and centrifuged for 15 min at 14,000 × g. The supernatant was subsequently collected and protein concentration quantified using the BCA assays (23225, Thermo Fisher Scientific). To assess target protein expression, 10 μg lysates were loaded into 4–$20\%$ Mini-PROTEAN TGX Stain-Free Protein Gels (Bio-Rad Laboratories AB, Solna, Sweden) and separated, followed by transfer of polyvinylidene difluoride (PVDF) membranes (1704156, Bio-Rad Laboratories AB). After blocking in $5\%$ bovine serum albumin (BSA) solution for 1 hr, the membranes were incubated with primary antibodies against MTIF3 (14219-1-AP, Proteintech), OXPHOS complex (45-8199, Thermo Fisher Scientific), FABP4, ACC, FAS (12589, Cell Signalling Technology), ATP8 (26723-1-AP, Proteintech), ND2 (19704-1-AP, Proteintech), CYTB (55090-1-AP, Proteintech), and corresponding horseradish peroxidase (HRP)-conjugated secondary antibodies (anti-mouse IgG, Cell Signalling Technology; anti-rabbit IgG, Cell Signaling Technology). TBS with $0.1\%$ (vol/vol) Tween-20 was used for membrane washing, and TBS with $2\%$ BSA was used for antibody incubation. To visualize the blots, Clarity western ECL substrate was added to the membrane and a CCD camera used to acquire images and Image Lab software (Bio-Rad Laboratories AB, Solna, Sweden) were used to develop the images. ImageJ software was used to quantify the protein bands. After detection of the protein targets, the membranes were stripped using Restore Western Blot Stripping Buffer (21059, Thermo Fisher Scientific) and blotted using anti-β-Actin antibody (4967, Cell Signaling Technology) or anti-GAPDH antibody (ab37168, Abcam).
## Blue Native polyacrylamide gel electrophoresis and immunoblotting
Differentiated scrambled control and MTIF3 knockout hWAs-iCas9 cells were adapted to 5.5 mM glucose growth medium for 3 days to mimic the physiological glucose concentration. The Blue Native polyacrylamide gel electrophoresis (BN-PAGE) was performed as described previously (Singh and Duchen, 2022). Briefly, mitochondria were isolated using Mitochondria Isolation Kit (89874, Thermo Fisher Scientific). NativePAGE Sample Prep Kit (BN2008, Invitrogen) was then used for mitochondrial protein extraction and BN-PAGE sample preparation. For native gel electrophoresis, 20 µg mitochondrial protein was loaded to precast 3–$12\%$ gradient Blue Native gels (BN1001, Invitrogen) and separated according to the manufacturer’s instructions. The proteins were then electroblotted onto PVDF membrane and probed with anti-OXPHOS antibody cocktail (45-8199, Thermo Fisher Scientific) and a corresponding secondary antibody. The blots were then imaged and analyzed as described above.
## Relative mitochondrial content measurement
To examine the effects of MTIF3 knockout on mitochondrial biogenesis in white adipocytes, relative amount of mtDNA was quantified using a qPCR-based method described previously (Ajaz et al., 2015). Briefly, total DNA was extracted and quantified using QIAamp DNA Mini Kit (catalogue number: 56304, QIAGEN) from scrambled control and MTIF3 knockout cells. For qPCR, equal amounts of total DNA from each sample were mixed with SYBR Green master mix (catalogue number: A25742, Thermo Fisher Scientific) and with primers targeting mitochondrial and nuclear genes, then the samples were run on ViiA7 qRT-PCR system (PE Applied Biosystems, Foster City, CA, USA). The relative mtDNA content was calculated as ΔCt (Ct of nuclear target − Ct of mitochondrial target).
## Mitochondrial function in MTIF3 knockout adipocytes
To directly assess the effects of MTIF3 on mitochondrial respiration in adipocytes we used the Seahorse XF (Seahorse Bioscience, North Billerica, MA) to measure cellular respiration OCR under different conditions. hWAs-iCas9 cells were seeded at 8000 cells/well in a Seahorse 24-well plate, then differentiated and induced for MTIF3 knockout or with scrambled control, as described above. Then, cells were adapted in 1 g/l growth medium (31885049, Thermo Fisher Scientific) for 3 days. Mitochondrial function was then assessed using the Seahorse XF-24 instrument according to a protocol optimized for the adipocyte cell line. Briefly, to measure OCR independent of oxidative phosphorylation, 2 μM oligomycin (O4876, Sigma-Aldrich) was added to the cells. Subsequently, 2 μM FCCP (carbonyl cyanide-p-trifluoromethoxyphenylhydrazone) (C2920, Sigma-Aldrich) and 5 μM respiratory chain inhibitors: rotenone (R8875, Sigma-Aldrich) and antimycin A (A8674, Sigma-Aldrich) were added to measure maximal respiration and basal rates of non-mitochondrial respiration. Cells were then frozen at −80°C for at least 4 hr, then the plate was dried, and DNA was extracted with CyQUANT Cell Lysis Buffer (C7027, Thermo Fisher Scientific). Total DNA was then quantified by Quant-iT PicoGreen dsDNA Assay Kit (P7589, Thermo Fisher Scientific) against a lambda DNA-generated standard curve.
## Endogenous long chain fatty acid oxidation in adipocytes
The Seahorse mitochondrial analyzer was used to test the effects of MTIF3 loss on endogenous long chain fatty acid oxidation in adipocytes. Prior to the assay, adipocytes were incubated overnight with substrate-limited medium: DMEM (A14430, Thermo Fisher Scientific); 0.5 mM glucose (103577-100, Angilent); 1.0 mM glutamine (103579-100, Angilent); 0.5 mM carnitine (C0283, Sigma-Aldrich); $1\%$ FBS (SV30160.03, HyClone). On day of the assay, the substrate-limited medium was replaced with FAO assay medium: 1× Krebs-Henseleit Buffer (KHB) was supplemented with 2.5 mM glucose, 0.5 mM carnitine, and 5 mM N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid (HEPES), and the pH was adjusted to pH 7.4 with NaOH. The cells were then treated for 15 min with either 40 μM etomoxir (E1905, Sigma-Aldrich) or only with the solvent (dimethyl sulfoxide, DMSO). Etomoxir inhibits carnitine palmitoyltransferase (CPT)-1 and diglyceride acyltransferase (DGAT) activity in mitochondria, and thus inhibits mitochondrial fatty acid oxidation (Xu et al., 2003; Griesel et al., 2010). The OCR was then measured as described above.
## Mass spectrometry-based metabolite profiling
The mature hWAs-iCas9 adipocytes were firstly induced for MTIF3 knockout, followed by glucose restriction challenge as described above. The cells were quenched on dry ice and metabolites were extracted using a previously optimized protocol (Danielsson et al., 2010).
For analysis of low molecular weight metabolites, extracts were reconstituted in 100 µl of MeOH/water ($\frac{8}{2}$, vol/vol) and 60 µl was transferred to new Eppendorf tubes and evaporated to dryness using a miVac concentrator (SP Scientific, NY) for 3 hr at 30°C. Dried samples were methoximated using 20 µl of methoxyamine hydrochloride in pyridine (Thermo Scientific, MA) by shaking at 3000 rpm for 30 min at room temperature (VWR, PA). Afterward, 20 µl of N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA) + $1\%$ trimethylsilyl chloride (Thermo Scientific, MA) was added to each sample and shaken at 3000 rpm at room temperature for 1 hr. Samples were transferred to glass vials and immediately analyzed using an Agilent 6890 gas chromatograph connected to an Agilent 5975CL VL MSD mass spectrometer controlled by MassHunter Workstation software 10.0 (Agilent, Atlanta, GA). One µl sample was injected at 270°C on an HP-5MS column (30 m length, 250 µm ID, 0.25 µm phase thickness). with a helium gas flow rate of 1 ml/min and a temperature gradient starting at 70°C for 2 min, increasing 15°C/min to 320°C and held for 2 min. Data were acquired using electron ionization at 70 eV in either full scan (50–550 m/z) or single ion monitoring mode. The MS-DIAL version 4.7 was used for raw peak extraction, peak alignment, deconvolution, peak annotation, and integration of peaks.
Amino acids and free fatty acids were chemically derivatized and analyzed using a previously described method (Meng et al., 2021). Briefly, 40 µl of the samples were mixed with 20 µl of 3-nitrophenylhydrazine (3-NPH) (Sigma-Aldrich, MO), followed by addition of 20 µl of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) (Thermo Scientific, MA) and shaking at 3000 rpm at room temperature for 1 hr. Samples were analyzed using an Agilent 1260 ultra-performance liquid chromatograph coupled with an Agilent 6495 tandem mass spectrometer and controlled by MassHunter version 8.0 (Agilent Technologies, CA). Three µl sample was injected on an Agilent Eclipse RRHD C18 column (2.1 × 150 mm, 1.8 µm) (Agilent Technologies, CA) with a flow rate of 0.6 ml/min and a column oven temperature of 50°C. The mobile phases A and B were $0.1\%$ formic acid (Fisher Chemical, Prague, Czech Republic) in Milli-Q water (Merck, Millipore, MO) and acetonitrile (VWR, Paris, France), respectively. Gradient elution was performed as follows: held at $5\%$ B from 0 to 1 min, changed linearly to $90\%$ B in 10 min, changed from $90\%$ B to $100\%$ B in 13 min, held at $100\%$ B for 2 min, returned to $5\%$ B (initial condition) in 0.1 min, and held at $5\%$ B for 2 min. Analyses were conducted in negative electrospray ionization mode (ESI) mode with the nebulizer gas pressure set at 20 psi, ion capillary voltage at 2500 V, gas temperature at 150°C, and sheath gas temperature at 250°C. Data were recorded in multiple reaction monitoring (MRM) mode, with two transitions for each analyte.
## Lipolysis quantification in differentiated hWAs cells
Differentiated scrambled control or MTIF3 knockout cells were washed twice with PBS and then incubated with DMEM containing $2\%$ free fatty acid-free BSA for 2 hr. For the insulin or isoproterenol-stimulated lipolysis, 100 nM insulin (I2643, Sigma-Aldrich) or 10 μM isoproterenol (1351005, Sigma-Aldrich) was added in the medium separately. After the incubation, the medium was collected, and the glycerol content was measured using Glycerol-Glo Assay (J3150, Promega).
## Total triglyceride measurement
Triglyceride-Glo Assay kit (J3161, Promega) was used to quantify total triglyceride content in scrambled control or MTIF3 knockout cells cultured either in 25 mM glucose or glucose restriction medium. Briefly, cells were collected in 50 μl kit lysis buffer at room temperature for 1 hr. Then 2 μl lysate was mixed with 8 μl glycerol lysis solution with lipase, and incubated at 37°C for 30 min. Subsequently, 10 μl glycerol solution was mixed with 10 μl glycerol detection solution supplemented with reductase substrate and kinetic enhancer, and transferred into a 384-well plate. After 1-hr incubation at room temperature, the luminescence was detected using CLARIOstar plate reader (BMG Labtech, Germany), and the triglyceride concentration was calculated using a standard curve generated from glycerol standards and normalized to total protein measured using BCA assays (23227, Thermo Fisher Scientific).
## Statistics
For each assay, the number of biological and technical replicates, standard deviation and statistical significance are reported in the figure legends. Hypothesis tests were performed using two-tailed Student’s t-test, one-way ANOVA, or paired t-test. A nominal p value of <0.05 was considered statistically significant. All analyses were undertaken using Prism GraphPad 9.0 software (La Jolla California, USA), SIMCA 17.0 (Sartorius Stedim Data Analytics, Malmö, Sweden), Rstudio 1.4, and Microsoft Excel 365. For the metabolome data, ANOVA (aov) was performed in R with genotype and glucose concentration as independent variables with Tukey’s test post hoc (TukeyHSD). Significance was defined as q < 0.05 using multiple testing adjustment according to the false discovery rate (p.adjust).
## Funding Information
This paper was supported by the following grants:
## Data availability
All data generated or analyzed in this study are included in the figures and the source data files. Source data files are provided for Figures 3 and 4, and for Figure 3—figure supplement 1.
## References
1. Adams KF, Schatzkin A, Harris TB, Kipnis V, Mouw T, Ballard-Barbash R, Hollenbeck A, Leitzmann MF. **Overweight, obesity, and mortality in a large prospective cohort of persons 50 to 71 years old**. *The New England Journal of Medicine* (2006) **355** 763-778. DOI: 10.1056/NEJMoa055643
2. Ajaz S, Czajka A, Malik A. **Accurate measurement of circulating mitochondrial DNA content from human blood samples using real-time quantitative PCR**. *Methods in Molecular Biology* (2015) **1264** 117-131. DOI: 10.1007/978-1-4939-2257-4_12
3. Appleton SL, Seaborn CJ, Visvanathan R, Hill CL, Gill TK, Taylor AW, Adams RJ, North West Adelaide Health Study T. **Diabetes and cardiovascular disease outcomes in the metabolically healthy obese phenotype: a cohort study**. *Diabetes Care* (2013) **36** 2388-2394. DOI: 10.2337/dc12-1971
4. Aprile M, Cataldi S, Perfetto C, Ambrosio MR, Italiani P, Tatè R, Blüher M, Ciccodicola A, Costa V. **In-vitro-generated hypertrophic-like adipocytes displaying PPARG isoforms unbalance recapitulate adipocyte dysfunctions in vivo**. *Cells* (2020) **9**. DOI: 10.3390/cells9051284
5. Bhargava K, Spremulli LL. **Role of the N- and C-terminal extensions on the activity of mammalian mitochondrial translational initiation factor 3**. *Nucleic Acids Research* (2005) **33** 7011-7018. DOI: 10.1093/nar/gki1007
6. Böhm A, Keuper M, Meile T, Zdichavsky M, Fritsche A, Häring HU, de Angelis MH, Staiger H, Franko A. **Increased mitochondrial respiration of adipocytes from metabolically unhealthy obese compared to healthy obese individuals**. *Scientific Reports* (2020) **10**. DOI: 10.1038/s41598-020-69016-9
7. Boudina S, Graham TE. **Mitochondrial function/dysfunction in white adipose tissue**. *Experimental Physiology* (2014) **99** 1168-1178. DOI: 10.1113/expphysiol.2014.081414
8. Chicherin IV, Baleva MV, Levitskii SA, Dashinimaev EB, Krasheninnikov IA, Kamenski P. **Initiation factor 3 is dispensable for mitochondrial translation in cultured human cells**. *Scientific Reports* (2020) **10**. DOI: 10.1038/s41598-020-64139-5
9. Chu AY, Deng X, Fisher VA, Drong A, Zhang Y, Feitosa MF, Liu C-T, Weeks O, Choh AC, Duan Q, Dyer TD, Eicher JD, Guo X, Heard-Costa NL, Kacprowski T, Kent JW, Lange LA, Liu X, Lohman K, Lu L, Mahajan A, O’Connell JR, Parihar A, Peralta JM, Smith AV, Zhang Y, Homuth G, Kissebah AH, Kullberg J, Laqua R, Launer LJ, Nauck M, Olivier M, Peyser PA, Terry JG, Wojczynski MK, Yao J, Bielak LF, Blangero J, Borecki IB, Bowden DW, Carr JJ, Czerwinski SA, Ding J, Friedrich N, Gudnason V, Harris TB, Ingelsson E, Johnson AD, Kardia SLR, Langefeld CD, Lind L, Liu Y, Mitchell BD, Morris AP, Mosley TH, Rotter JI, Shuldiner AR, Towne B, Völzke H, Wallaschofski H, Wilson JG, Allison M, Lindgren CM, Goessling W, Cupples LA, Steinhauser ML, Fox CS. **Multiethnic genome-wide meta-analysis of ectopic fat depots identifies loci associated with adipocyte development and differentiation**. *Nature Genetics* (2017) **49** 125-130. DOI: 10.1038/ng.3738
10. Civitarese AE, Carling S, Heilbronn LK, Hulver MH, Ukropcova B, Deutsch WA, Smith SR, Ravussin E. **Calorie restriction increases muscle mitochondrial biogenesis in healthy humans**. *PLOS Medicine* (2007) **4**. DOI: 10.1371/journal.pmed.0040076
11. Danielsson APH, Moritz T, Mulder H, Spégel P. **Development and optimization of a metabolomic method for analysis of adherent cell cultures**. *Analytical Biochemistry* (2010) **404** 30-39. DOI: 10.1016/j.ab.2010.04.013
12. Denis GV, Obin MS. **“ metabolically healthy obesity ”: origins and implications**. *Molecular Aspects of Medicine* (2013) **34** 59-70. DOI: 10.1016/j.mam.2012.10.004
13. Gregoire FM, Smas CM, Sul HS. **Understanding adipocyte differentiation**. *Physiological Reviews* (1998) **78** 783-809. DOI: 10.1152/physrev.1998.78.3.783
14. Griesel BA, Weems J, Russell RA, Abel ED, Humphries K, Olson AL. **Acute inhibition of fatty acid import inhibits GLUT4 transcription in adipose tissue, but not skeletal or cardiac muscle tissue, partly through liver X receptor (LXR) signaling**. *Diabetes* (2010) **59** 800-807. DOI: 10.2337/db09-1542
15. Hansson A, Hance N, Dufour E, Rantanen A, Hultenby K, Clayton DA, Wibom R, Larsson NG. **A switch in metabolism precedes increased mitochondrial biogenesis in respiratory chain-deficient mouse hearts**. *PNAS* (2004) **101** 3136-3141. DOI: 10.1073/pnas.0308710100
16. Heinonen S, Buzkova J, Muniandy M, Kaksonen R, Ollikainen M, Ismail K, Hakkarainen A, Lundbom J, Lundbom N, Vuolteenaho K, Moilanen E, Kaprio J, Rissanen A, Suomalainen A, Pietiläinen KH. **Impaired mitochondrial biogenesis in adipose tissue in acquired obesity**. *Diabetes* (2015) **64** 3135-3145. DOI: 10.2337/db14-1937
17. Heinonen S, Jokinen R, Rissanen A, Pietiläinen KH. **White adipose tissue mitochondrial metabolism in health and in obesity**. *Obesity Reviews* (2020) **21**. DOI: 10.1111/obr.12958
18. Jokinen R, Rinnankoski-Tuikka R, Kaye S, Saarinen L, Heinonen S, Myöhänen M, Rappou E, Jukarainen S, Rissanen A, Pessia A, Velagapudi V, Virtanen KA, Pirinen E, Pietiläinen KH. **Adipose tissue mitochondrial capacity associates with long-term weight loss success**. *International Journal of Obesity* (2018) **42** 817-825. DOI: 10.1038/ijo.2017.299
19. Kaaman M, Sparks LM, van Harmelen V, Smith SR, Sjölin E, Dahlman I, Arner P. **Strong association between mitochondrial DNA copy number and lipogenesis in human white adipose tissue**. *Diabetologia* (2007) **50** 2526-2533. DOI: 10.1007/s00125-007-0818-6
20. Kamble PG, Hetty S, Vranic M, Almby K, Castillejo-López C, Abalo XM, Pereira MJ, Eriksson JW. **Proof-of-concept for CRISPR/cas9 gene editing in human preadipocytes: deletion of FKBP5 and PPARG and effects on adipocyte differentiation and metabolism**. *Scientific Reports* (2020) **10**. DOI: 10.1038/s41598-020-67293-y
21. Kuzmenko A, Atkinson GC, Levitskii S, Zenkin N, Tenson T, Hauryliuk V, Kamenski P. **Mitochondrial translation initiation machinery: conservation and diversification**. *Biochimie* (2014) **100** 132-140. DOI: 10.1016/j.biochi.2013.07.024
22. Larson-Meyer DE, Heilbronn LK, Redman LM, Newcomer BR, Frisard MI, Anton S, Smith SR, Alfonso A, Ravussin E. **Effect of calorie restriction with or without exercise on insulin sensitivity, beta-cell function, fat cell size, and ectopic lipid in overweight subjects**. *Diabetes Care* (2006) **29** 1337-1344. DOI: 10.2337/dc05-2565
23. Locke AE, Kahali B, Berndt SI, Justice AE, Pers TH, Day FR, Powell C, Vedantam S, Buchkovich ML, Yang J, Croteau-Chonka DC, Esko T, Fall T, Ferreira T, Gustafsson S, Kutalik Z, Luan J, Mägi R, Randall JC, Winkler TW, Wood AR, Workalemahu T, Faul JD, Smith JA, Zhao JH, Zhao W, Chen J, Fehrmann R, Hedman ÅK, Karjalainen J, Schmidt EM, Absher D, Amin N, Anderson D, Beekman M, Bolton JL, Bragg-Gresham JL, Buyske S, Demirkan A, Deng G, Ehret GB, Feenstra B, Feitosa MF, Fischer K, Goel A, Gong J, Jackson AU, Kanoni S, Kleber ME, Kristiansson K, Lim U, Lotay V, Mangino M, Leach IM, Medina-Gomez C, Medland SE, Nalls MA, Palmer CD, Pasko D, Pechlivanis S, Peters MJ, Prokopenko I, Shungin D, Stančáková A, Strawbridge RJ, Sung YJ, Tanaka T, Teumer A, Trompet S, van der Laan SW, van Setten J, Van Vliet-Ostaptchouk JV, Wang Z, Yengo L, Zhang W, Isaacs A, Albrecht E, Ärnlöv J, Arscott GM, Attwood AP, Bandinelli S, Barrett A, Bas IN, Bellis C, Bennett AJ, Berne C, Blagieva R, Blüher M, Böhringer S, Bonnycastle LL, Böttcher Y, Boyd HA, Bruinenberg M, Caspersen IH, Chen Y-DI, Clarke R, Daw EW, de Craen AJM, Delgado G, Dimitriou M, Doney ASF, Eklund N, Estrada K, Eury E, Folkersen L, Fraser RM, Garcia ME, Geller F, Giedraitis V, Gigante B, Go AS, Golay A, Goodall AH, Gordon SD, Gorski M, Grabe H-J, Grallert H, Grammer TB, Gräßler J, Grönberg H, Groves CJ, Gusto G, Haessler J, Hall P, Haller T, Hallmans G, Hartman CA, Hassinen M, Hayward C, Heard-Costa NL, Helmer Q, Hengstenberg C, Holmen O, Hottenga J-J, James AL, Jeff JM, Johansson Å, Jolley J, Juliusdottir T, Kinnunen L, Koenig W, Koskenvuo M, Kratzer W, Laitinen J, Lamina C, Leander K, Lee NR, Lichtner P, Lind L, Lindström J, Lo KS, Lobbens S, Lorbeer R, Lu Y, Mach F, Magnusson PKE, Mahajan A, McArdle WL, McLachlan S, Menni C, Merger S, Mihailov E, Milani L, Moayyeri A, Monda KL, Morken MA, Mulas A, Müller G, Müller-Nurasyid M, Musk AW, Nagaraja R, Nöthen MM, Nolte IM, Pilz S, Rayner NW, Renstrom F, Rettig R, Ried JS, Ripke S, Robertson NR, Rose LM, Sanna S, Scharnagl H, Scholtens S, Schumacher FR, Scott WR, Seufferlein T, Shi J, Smith AV, Smolonska J, Stanton AV, Steinthorsdottir V, Stirrups K, Stringham HM, Sundström J, Swertz MA, Swift AJ, Syvänen A-C, Tan S-T, Tayo BO, Thorand B, Thorleifsson G, Tyrer JP, Uh H-W, Vandenput L, Verhulst FC, Vermeulen SH, Verweij N, Vonk JM, Waite LL, Warren HR, Waterworth D, Weedon MN, Wilkens LR, Willenborg C, Wilsgaard T, Wojczynski MK, Wong A, Wright AF, Zhang Q, Brennan EP, Choi M, Dastani Z, Drong AW, Eriksson P, Franco-Cereceda A, Gådin JR, Gharavi AG, Goddard ME, Handsaker RE, Huang J, Karpe F, Kathiresan S, Keildson S, Kiryluk K, Kubo M, Lee J-Y, Liang L, Lifton RP, Ma B, McCarroll SA, McKnight AJ, Min JL, Moffatt MF, Montgomery GW, Murabito JM, Nicholson G, Nyholt DR, Okada Y, Perry JRB, Dorajoo R, Reinmaa E, Salem RM, Sandholm N, Scott RA, Stolk L, Takahashi A, Tanaka T, van ’t Hooft FM, Vinkhuyzen AAE, Westra H-J, Zheng W, Zondervan KT, Heath AC, Arveiler D, Bakker SJL, Beilby J, Bergman RN, Blangero J, Bovet P, Campbell H, Caulfield MJ, Cesana G, Chakravarti A, Chasman DI, Chines PS, Collins FS, Crawford DC, Cupples LA, Cusi D, Danesh J, de Faire U, den Ruijter HM, Dominiczak AF, Erbel R, Erdmann J, Eriksson JG, Farrall M, Felix SB, Ferrannini E, Ferrières J, Ford I, Forouhi NG, Forrester T, Franco OH, Gansevoort RT, Gejman PV, Gieger C, Gottesman O, Gudnason V, Gyllensten U, Hall AS, Harris TB, Hattersley AT, Hicks AA, Hindorff LA, Hingorani AD, Hofman A, Homuth G, Hovingh GK, Humphries SE, Hunt SC, Hyppönen E, Illig T, Jacobs KB, Jarvelin M-R, Jöckel K-H, Johansen B, Jousilahti P, Jukema JW, Jula AM, Kaprio J, Kastelein JJP, Keinanen-Kiukaanniemi SM, Kiemeney LA, Knekt P, Kooner JS, Kooperberg C, Kovacs P, Kraja AT, Kumari M, Kuusisto J, Lakka TA, Langenberg C, Marchand LL, Lehtimäki T, Lyssenko V, Männistö S, Marette A, Matise TC, McKenzie CA, McKnight B, Moll FL, Morris AD, Morris AP, Murray JC, Nelis M, Ohlsson C, Oldehinkel AJ, Ong KK, Madden PAF, Pasterkamp G, Peden JF, Peters A, Postma DS, Pramstaller PP, Price JF, Qi L, Raitakari OT, Rankinen T, Rao DC, Rice TK, Ridker PM, Rioux JD, Ritchie MD, Rudan I, Salomaa V, Samani NJ, Saramies J, Sarzynski MA, Schunkert H, Schwarz PEH, Sever P, Shuldiner AR, Sinisalo J, Stolk RP, Strauch K, Tönjes A, Trégouët D-A, Tremblay A, Tremoli E, Virtamo J, Vohl M-C, Völker U, Waeber G, Willemsen G, Witteman JC, Zillikens MC, Adair LS, Amouyel P, Asselbergs FW, Assimes TL, Bochud M, Boehm BO, Boerwinkle E, Bornstein SR, Bottinger EP, Bouchard C, Cauchi S, Chambers JC, Chanock SJ, Cooper RS, de Bakker PIW, Dedoussis G, Ferrucci L, Franks PW, Froguel P, Groop LC, Haiman CA, Hamsten A, Hui J, Hunter DJ, Hveem K, Kaplan RC, Kivimaki M, Kuh D, Laakso M, Liu Y, Martin NG, März W, Melbye M, Metspalu A, Moebus S, Munroe PB, Njølstad I, Oostra BA, Palmer CNA, Pedersen NL, Perola M, Pérusse L, Peters U, Power C, Quertermous T, Rauramaa R, Rivadeneira F, Saaristo TE, Saleheen D, Sattar N, Schadt EE, Schlessinger D, Slagboom PE, Snieder H, Spector TD, Thorsteinsdottir U, Stumvoll M, Tuomilehto J, Uitterlinden AG, Uusitupa M, van der Harst P, Walker M, Wallaschofski H, Wareham NJ, Watkins H, Weir DR, Wichmann H-E, Wilson JF, Zanen P, Borecki IB, Deloukas P, Fox CS, Heid IM, O’Connell JR, Strachan DP, Stefansson K, van Duijn CM, Abecasis GR, Franke L, Frayling TM, McCarthy MI, Visscher PM, Scherag A, Willer CJ, Boehnke M, Mohlke KL, Lindgren CM, Beckmann JS, Barroso I, North KE, Ingelsson E, Hirschhorn JN, Loos RJF, Speliotes EK. **Genetic studies of body mass index yield new insights for obesity biology**. *Nature* (2015) **518** 197-206. DOI: 10.1038/nature14177
24. Lotta LA, Wittemans LBL, Zuber V, Stewart ID, Sharp SJ, Luan J, Day FR, Li C, Bowker N, Cai L, De Lucia Rolfe E, Khaw K-T, Perry JRB, O’Rahilly S, Scott RA, Savage DB, Burgess S, Wareham NJ, Langenberg C. **Association of genetic variants related to gluteofemoral vs abdominal fat distribution with type 2 diabetes, coronary disease, and cardiovascular risk factors**. *JAMA* (2018) **320** 2553-2563. DOI: 10.1001/jama.2018.19329
25. Maechler P, Wollheim CB. **Mitochondrial function in normal and diabetic beta-cells**. *Nature* (2001) **414** 807-812. DOI: 10.1038/414807a
26. May CE, Rosander J, Gottfried J, Dennis E, Dus M. **Dietary sugar inhibits satiation by decreasing the central processing of sweet taste**. *eLife* (2020) **9**. DOI: 10.7554/eLife.54530
27. Meng X, Pang H, Sun F, Jin X, Wang B, Yao K, Yao L, Wang L, Hu Z. **Simultaneous 3-nitrophenylhydrazine derivatization strategy of carbonyl, carboxyl and phosphoryl submetabolome for LC-MS/MS-based targeted metabolomics with improved sensitivity and coverage**. *Analytical Chemistry* (2021) **93** 10075-10083. DOI: 10.1021/acs.analchem.1c00767
28. Nettleton JA, Follis JL, Ngwa JS, Smith CE, Ahmad S, Tanaka T, Wojczynski MK, Voortman T, Lemaitre RN, Kristiansson K, Nuotio M-L, Houston DK, Perälä M-M, Qi Q, Sonestedt E, Manichaikul A, Kanoni S, Ganna A, Mikkilä V, North KE, Siscovick DS, Harald K, Mckeown NM, Johansson I, Rissanen H, Liu Y, Lahti J, Hu FB, Bandinelli S, Rukh G, Rich S, Booij L, Dmitriou M, Ax E, Raitakari O, Mukamal K, Männistö S, Hallmans G, Jula A, Ericson U, Jacobs DR, Van Rooij FJA, Deloukas P, Sjögren P, Kähönen M, Djousse L, Perola M, Barroso I, Hofman A, Stirrups K, Viikari J, Uitterlinden AG, Kalafati IP, Franco OH, Mozaffarian D, Salomaa V, Borecki IB, Knekt P, Kritchevsky SB, Eriksson JG, Dedoussis GV, Qi L, Ferrucci L, Orho-Melander M, Zillikens MC, Ingelsson E, Lehtimäki T, Renström F, Cupples LA, Loos RJF, Franks PW. **Gene × dietary pattern interactions in obesity: analysis of up to 68 317 adults of European ancestry**. *Human Molecular Genetics* (2015) **24** 4728-4738. DOI: 10.1093/hmg/ddv186
29. Papandonatos GD, Pan Q, Pajewski NM, Delahanty LM, Peter I, Erar B, Ahmad S, Harden M, Chen L, Fontanillas P, Wagenknecht LE, Kahn SE, Wing RR, Jablonski KA, Huggins GS, Knowler WC, Florez JC, McCaffery JM, Franks PW. **Genetic predisposition to weight loss and regain with lifestyle intervention: analyses from the diabetes prevention program and the look ahead randomized controlled trials**. *Diabetes* (2015) **64** 4312-4321. DOI: 10.2337/db15-0441
30. Phielix E, Meex R, Moonen-Kornips E, Hesselink MKC, Schrauwen P. **Exercise training increases mitochondrial content and ex vivo mitochondrial function similarly in patients with type 2 diabetes and in control individuals**. *Diabetologia* (2010) **53** 1714-1721. DOI: 10.1007/s00125-010-1764-2
31. Pietiläinen KH, Naukkarinen J, Rissanen A, Saharinen J, Ellonen P, Keränen H, Suomalainen A, Götz A, Suortti T, Yki-Järvinen H, Oresic M, Kaprio J, Peltonen L. **Global transcript profiles of fat in monozygotic twins discordant for BMI: pathways behind acquired obesity**. *PLOS Medicine* (2008) **5**. DOI: 10.1371/journal.pmed.0050051
32. Poulos SP, Dodson MV, Hausman GJ. **Cell line models for differentiation: preadipocytes and adipocytes**. *Experimental Biology and Medicine* (2010) **235** 1185-1193. DOI: 10.1258/ebm.2010.010063
33. Rasmussen-Torvik LJ, Baldridge AS, Pacheco JA, Aufox SA, Kim K-YA, Silverstein JC, Denham EW, Hungness E, Smith ME, Greenland P. **Rs4771122 predicts multiple measures of long-term weight loss after bariatric surgery**. *Obesity Surgery* (2015) **25** 2225-2229. DOI: 10.1007/s11695-015-1872-7
34. Rudler DL, Hughes LA, Perks KL, Richman TR, Kuznetsova I, Ermer JA, Abudulai LN, Shearwood AMJ, Viola HM, Hool LC, Siira SJ, Rackham O, Filipovska A. **Fidelity of translation initiation is required for coordinated respiratory complex assembly**. *Science Advances* (2019) **5**. DOI: 10.1126/sciadv.aay2118
35. Ruiz-Ojeda FJ, Rupérez AI, Gomez-Llorente C, Gil A, Aguilera CM. **Cell models and their application for studying adipogenic differentiation in relation to obesity: a review**. *International Journal of Molecular Sciences* (2016) **17**. DOI: 10.3390/ijms17071040
36. Sakers A, De Siqueira MK, Seale P, Villanueva CJ. **Adipose-Tissue plasticity in health and disease**. *Cell* (2022) **185** 419-446. DOI: 10.1016/j.cell.2021.12.016
37. Shamsi F, Tseng YH. **Protocols for generation of immortalized human brown and white preadipocyte cell lines**. *Methods in Molecular Biology* (2017) **1566** 77-85. DOI: 10.1007/978-1-4939-6820-6_8
38. Singh K, Duchen MR. **Analysis of organization and activity of mitochondrial respiratory chain complexes in primary fibroblasts using blue native page**. *Methods in Molecular Biology* (2022) **2497** 339-348. DOI: 10.1007/978-1-0716-2309-1_25
39. Thomas D. **Gene -- environment-wide association studies: emerging approaches**. *Nature Reviews. Genetics* (2010) **11** 259-272. DOI: 10.1038/nrg2764
40. Toledo FGS, Menshikova EV, Ritov VB, Azuma K, Radikova Z, DeLany J, Kelley DE. **Effects of physical activity and weight loss on skeletal muscle mitochondria and relationship with glucose control in type 2 diabetes**. *Diabetes* (2007) **56** 2142-2147. DOI: 10.2337/db07-0141
41. Tormos KV, Anso E, Hamanaka RB, Eisenbart J, Joseph J, Kalyanaraman B, Chandel NS. **Mitochondrial complex III ROS regulate adipocyte differentiation**. *Cell Metabolism* (2011) **14** 537-544. DOI: 10.1016/j.cmet.2011.08.007
42. Wang G, Yang L, Grishin D, Rios X, Ye LY, Hu Y, Li K, Zhang D, Church GM, Pu WT. **Efficient, footprint-free human iPSC genome editing by consolidation of cas9/CRISPR and piggyBac technologies**. *Nature Protocols* (2017) **12** 88-103. DOI: 10.1038/nprot.2016.152
43. Ward LD, Kellis M. **HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants**. *Nucleic Acids Research* (2012) **40** D930-D934. DOI: 10.1093/nar/gkr917
44. Xu FY, Taylor WA, Hurd JA, Hatch GM. **Etomoxir mediates differential metabolic channeling of fatty acid and glycerol precursors into cardiolipin in H9c2 cells**. *Journal of Lipid Research* (2003) **44** 415-423. DOI: 10.1194/jlr.M200335-JLR200
45. Xue R, Lynes MD, Dreyfuss JM, Shamsi F, Schulz TJ, Zhang H, Huang TL, Townsend KL, Li Y, Takahashi H, Weiner LS, White AP, Lynes MS, Rubin LL, Goodyear LJ, Cypess AM, Tseng Y-H. **Clonal analyses and gene profiling identify genetic biomarkers of the thermogenic potential of human brown and white preadipocytes**. *Nature Medicine* (2015) **21** 760-768. DOI: 10.1038/nm.3881
46. Zhu Q, An YA, Scherer PE. **Mitochondrial regulation and white adipose tissue homeostasis**. *Trends in Cell Biology* (2022) **32** 351-364. DOI: 10.1016/j.tcb.2021.10.008
|
---
title: A parabrachial to hypothalamic pathway mediates defensive behavior
authors:
- Fan Wang
- Yuge Chen
- Yuxin Lin
- Xuze Wang
- Kaiyuan Li
- Yong Han
- Jintao Wu
- Xingyi Shi
- Zhenggang Zhu
- Chaoying Long
- Xiaojun Hu
- Shumin Duan
- Zhihua Gao
journal: eLife
year: 2023
pmcid: PMC10023160
doi: 10.7554/eLife.85450
license: CC BY 4.0
---
# A parabrachial to hypothalamic pathway mediates defensive behavior
## Abstract
Defensive behaviors are critical for animal’s survival. Both the paraventricular nucleus of the hypothalamus (PVN) and the parabrachial nucleus (PBN) have been shown to be involved in defensive behaviors. However, whether there are direct connections between them to mediate defensive behaviors remains unclear. Here, by retrograde and anterograde tracing, we uncover that cholecystokinin (CCK)-expressing neurons in the lateral PBN (LPBCCK) directly project to the PVN. By in vivo fiber photometry recording, we find that LPBCCK neurons actively respond to various threat stimuli. Selective photoactivation of LPBCCK neurons promotes aversion and defensive behaviors. Conversely, photoinhibition of LPBCCK neurons attenuates rat or looming stimuli-induced flight responses. Optogenetic activation of LPBCCK axon terminals within the PVN or PVN glutamatergic neurons promotes defensive behaviors. Whereas chemogenetic and pharmacological inhibition of local PVN neurons prevent LPBCCK-PVN pathway activation-driven flight responses. These data suggest that LPBCCK neurons recruit downstream PVN neurons to actively engage in flight responses. Our study identifies a previously unrecognized role for the LPBCCK-PVN pathway in controlling defensive behaviors.
## Introduction
Living in an environment full of diverse threats, animals develop a variety of defensive behaviors for survival during evolution (Anderson and Adolphs, 2014; Blanchard et al., 2001; Crawford and Masterson, 1982; Fanselow and Bolles, 1979; LeDoux, 2012; Yilmaz and Meister, 2013). From the perspective of predator and prey interactions, defensive responses are generally divided into two categories: to avoid being discovered (such as freezing or hiding behaviors) or being caught (such as escaping or attacking behaviors) (Crawford and Masterson, 1982; Fanselow and Bolles, 1979; Yilmaz and Meister, 2013). Which type of defensive behavior is selected depends on many factors, including the intensity and proximity of the threat, as well as the surrounding context. *In* general, escape is triggered when animals are faced with an imminent threat with a shelter nearby (Eilam, 2005; Lin et al., 2023; Perusini and Fanselow, 2015; Sun et al., 2020b; Zhang et al., 2018).
Several brain regions, including the periaqueductal gray (PAG), amygdala, and medial hypothalamic zone (MHZ), have been implicated in mediating escape behaviors (Evans et al., 2018; Shang et al., 2018; Shang et al., 2015; Silva et al., 2013; Tovote et al., 2016; Wang et al., 2015; Wang et al., 2021b; Wei et al., 2015). Recent studies have highlighted a role for the hypothalamus, in particular the MHZ, including the anterior hypothalamic nucleus (AHN), the dorsomedial part of ventromedial hypothalamic nucleus (VMHdm) and the dorsal pre-mammillary nucleus (PMd), in mediating escape behaviors. For example, a subset of VMH neurons collaterally project to the AHN and PAG to promote both escape and avoidance behaviors (Wang et al., 2015). Moreover, PMd neurons project to the dorsolateral periaqueductal gray region (dlPAG) and the anteromedial ventral thalamic region (AMV) also control escape behaviors (Wang et al., 2021b). In addition, PVN, an area enriched with endocrine neurons, has also been associated with escape behaviors (Mangieri et al., 2019). Optogenetic activation of Sim1+ neurons in the PVN triggers escape and projections from the PVN to both the ventral midbrain region (vMB) and the ventral lateral septum (LSv) mediate defensive-like behaviors, including hiding, escape jumping and hyperlocomotion (Mangieri et al., 2019; Xu et al., 2019). Despite a growing number of studies illustrating the importance of PVN in mediating defensive behaviors, upstream inputs that transmit the danger signals to the PVN remain unclear (Isosaka et al., 2015; Penzo et al., 2015).
Located in the dorsolateral pons, PBN, serves as an important relay station for sensory transmission (Fulwiler and Saper, 1984). While PBN is best known for various sensory processes to protect the body from noxious stimuli, emerging evidence suggests that it may act as a key node in mediating defensive behaviors (Campos et al., 2018; Day et al., 2004; Han et al., 2015a). Electrical lesions of the PBN or local microinjections of kainic acid into the PBNinduced defensive behaviors (Mileikovskii and Verevkina, 1991). In addition, exposure of the olfactory predator cue, trimethylthiazoline (TMT), increased c-fos expression in the LPB (Day et al., 2004). More importantly, calcitonin gene-related peptide (CGRP) neurons, located in the external lateral subnuclei of PBN (PBel), have been shown to mediate alarm responses and defensive behaviors under stressful or threatening circumstances via projections to the amygdala and the bed nucleus of stria terminalis (BNST) (Han et al., 2015a; Zhang et al., 2020). While both PBN and PVN neurons appear to be involved in defensive responses, it remains unclear whether they connect with each other to mediate defensive behaviors (Fulwiler and Saper, 1984).
Both CCK and CCK receptor-expressing neurons have been implicated in defensive behaviors (Bertoglio et al., 2007; Chen et al., 2022; Wang et al., 2021a). PVN neurons also express CCK receptors and are involved in defensive behaviors (Mangieri et al., 2019; O’Shea and Gundlach, 1993; Xu et al., 2019), raising a possibility that upstream CCK inputs to the PVN (Meister et al., 1994) may be implicated in defensive behaviors. In an attempt to test this possibility, we injected the Cre-dependent retrograde tracer (AAV2-Retro-DIO-EYFP) into the PVN of Cck-cre mice and found that the LPB is an important upstream CCKergic input to the PVN. We further showed that LPBCCK neurons were recruited upon exposure to various threat stimuli. Optogenetic activation of either LPBCCK neuronal somas or their projections to the PVN promotes defensive responses, whereas their inhibition attenuates defensive-like behaviors, suggesting an essential and sufficient role for LPBCCK-PVN pathway in mediating defensive behaviors.
## LPBCCK neurons provide monosynaptic glutamatergic projections to the PVN
To identify upstream CCKergic inputs to the PVN and visualize the range of viral infection, we injected a mixture of retrograde viral tracer (AAV2-Retro-DIO-EYFP, hereafter referred to as AAV2-Retro) and CTB 555 into the PVN of knock-in mice expressing Cre recombinase at the CCK Locus (Cck-ires-cre, referred to as Cck-cre hereafter; Figure 1A). We observed abundant EYFP+ neurons in the LPB, medial orbital cortex (MO), and cingulate cortex (CC), with scattered EYFP+ cells in the PAG and dysgranular insular cortex (DI) (Figure 1B–H). Since LPB is a region critical for sensory signal processing, we mainly focused on the LPB in the following study (Fulwiler and Saper, 1984; Garfield et al., 2014).
**Figure 1.:** *LPBCCK neurons project to the PVN.(A) Scheme of viral strategy for retrograde tracing from the PVN in Cck-cre mice using AAV2-Retro virus. (B) Representative image showing the injection site as marked by CTB555. (C) Representative histological images of EYFP+ neurons in the LPB. (D–H) A heatmap (D) demonstrating the distribution of EYFP+ neurons in the MO (E), CC (F), DI (G), PAG (H), n = 3 mice. (I–J) Anterograde tracing of LPBCCK neurons using AAV-hSyn-FLEX-mGFP-2A-Synaptophysin-mRuby virus. (K) Projections of LPBCCK neurons to the PVN. The right panel shows a magnified view of boxed area; scale bar, 100 μm; n = 3 mice. (L) Schematic of recording from PVN cells after optogenetic activation of LPBCCK axonal terminals. (M) Representative traces of light-evoked EPSCs recorded from PVN neurons following light stimulation of LPBCCK axonal terminals in the presence of ACSF (Ctrl), TTX (1 μM) and 4-AP (100 μM). (N) Quantification of excitatory postsynaptic currents (EPSCs) from identified PVN neurons receiving inputs from the LPBCCK neurons in the presence of ACSF (Ctrl), TTX (1 μM) and 4-AP (100 μM). (ACSF vs. TTX ****p < 0.0001; TTX vs. 4-AP ****p < 0.0001; ACSF vs. 4-AP p > 0.9999, one-way ANOVA Bonferroni’s multiple comparisons test). (O) Representative traces of light-evoked EPSCs recorded from PVN neurons following light stimulation of LPBCCK axonal terminals in the presence of CNQX (20 μM) and AP5 (50 μM) (n = 9 neurons). (P) Quantification of EPSCs and inhibitory postsynaptic currents (IPSCs) from identified PVN neurons receiving inputs from the LPBCCK neurons. (oEPSC, p = 0.0003 t = 5.378 df = 10; oIPSC, p = 0.8793 t = 0.1558 df = 10; unpaired t test).
Figure 1—source data 1.Quantification of the labeled cells, EPSCs and IPSCs.*
LPB contains several subregions with diverse peptide-expressing neuronal subtypes and CCK+ neurons have been shown to primarily locate within the superior lateral PB (PBsl; Garfield et al., 2014). To dissect the identity of CCK+ neurons in the LPB, we performed in situ hybridization in Cck-cre::Ai14 mice using probes for Slc17a6 and Slc32a1, markers specific to glutamatergic and GABAergic neurons (Figure 1—figure supplement 1A–C). We found that LPBCCK neurons ($84.0\%$ ± $1.7\%$) predominantly expressed Slc17a6, with a tiny subpopulation ($2.0\%$ ± $1.2\%$) expressing Slc32a1, suggesting that LPBCCK neurons were mostly glutamatergic neurons (Figure 1—figure supplement 1D–E). In addition, co-labeling with another neuropeptide, CGRP, revealed minimal co-localization ($2.2\%$ ± $0.37\%$) between CCK and CGRP (Figure 1—figure supplement 1F–G), suggesting primary separation between these two subpopulations of neurons. Collectively, these data demonstrate that LPBCCK neurons comprise predominantly glutamatergic neurons, which barely overlap with CGRP neurons.
To verify the anatomical connections between the LPB and PVN, we injected the anterograde viral tracer (AAV-hSyn-FLEx-mGFP-2A-Synaptophysin-mRuby) into the LPB of Cck-cre mice (Figure 1I–J). We observed prominent GFP+ fibers and mRuby+ bouton-like structures, reminiscent of axonal terminals, within the PVN (Figure 1K) and other brain regions (Figure 1—figure supplement 2A–L). Next, we injected the Cre-dependent adeno-associated virus (AAV) expressing channelrhodopsin-2 virus (AAV$\frac{2}{9}$-DIO-ChR2-EYFP) into the LPB of Cck-cre mice (Figure 1L) and optogenetically stimulated the axonal terminals of virus-labeled LPBCCK neurons within the PVN in brain slices. We successfully recorded light stimulation-induced excitatory postsynaptic currents (EPSCs), rather than inhibitory postsynaptic currents (IPSCs). These EPSCs were blocked in the presence of the sodium channel antagonist tetrodotoxin (TTX), but rescued by the potassium channel antagonist 4-Aminopyridine (4-AP) (Figure 1M–N). Further, NMDA receptor antagonist AP5, and AMPA receptor antagonist CNQX also blocked light stimulation-induced EPSCs (Figure 1O–P), validating that PVN neurons receive monosynaptic excitatory innervations from LPBCCK neurons.
## Photostimulation of LPBCCK neurons induces aversion and defensive behaviors
LPBCCK neurons have been shown to regulate glucose homeostasis and body temperature (Garfield et al., 2014; Yang et al., 2020). However, it remains unclear whether activation of LPBCCK neurons may elicit direct behavioral changes. Before optic stimulation in vivo, we first tested the efficacy of optogenetic stimulation by whole-cell recording in ChR2-expressing LPBCCK neurons. We observed that 5–20 Hz blue laser pulses induced time-locked action potential firing, with 20 Hz inducing maximal firing capacities. We then chose 20 Hz for the following in vivo stimulation (Figure 2A).
**Figure 2.:** *Activation of LPBCCK neurons triggers aversion, defensive-like flight-to-nest behavior and autonomic responses.(A) Left, schematic of light stimulation of and patch-clamp recording from ChR2-EYFP expressing CCK neurons in the LPB. Right, example of action potentials evoked by optogenetic stimulation LPBCCK neurons using whole cell patch-clamp slice recording. (B) Schematic diagram of optogenetic activation of LPBCCK neurons. (C) Representative image showing the ChR2-EYFP expression and optical fiber tip locations in the LPB of a Cck-cre mouse. Scale bar, 100 μm. (D) Schematic of the timing and behavioral paradigm with optical activation of LPBCCK neurons. (E–F) Diagram of the real-time place aversion (RTPA) test and the example traces of the RTPA test from the mice. (G) Quantification of the time of mice spent in the laser-paired chamber (EYFP: n = 5 mice, ChR2: n = 5 mice; df = 16; two-way ANOVA test) after optogenetic activation. (H) Diagram of the flight to nest test. (I–K) Quantification of latency (I), speed (J) and time in the nest (K) (EYFP: n = 7 mice, ChR2: n = 7 mice; for latency, p = 0.0006, U = 0; Mann-Whitney test; for speed, p = 0.0016, t = 4.052, df = 12; unpaired t test; for time in the nest, p < 0.0001, t = 19.82, df = 12; unpaired t test). (L) Analyses of heart rate changes induced by photostimulation of LPBCCK neurons. (EYFP: n = 7 mice, ChR2: n = 7 mice; p = 0.0006, t = 4.603, df = 12; unpaired t test). (M–N) Example images of computer-detected pupils (M) and quantitative analyses of pupil size before and during photo-stimulation of LPBCCK neurons (N) (EYFP: n = 6 mice, ChR2: n = 6 mice; p = 0.0022, U = 0; Mann-Whitney test). (O) Cartoon of the arousal state during the activation of LPBCCK neurons. (P) Plasma corticosterone levels in EYFP and ChR2 groups. (EYFP: n = 5 mice, ChR2: n = 5 mice; p = 0.0121, t = 3.23, df = 8; unpaired t test).
Figure 2—source data 1.Quantification of the flight-to-nest behavior and autonomic responses upon activation of LPB CCK neurons.*
We injected the control or ChR2-expressing viruses (AAV$\frac{2}{9}$-EF1a-DIO-EYFP or AAV$\frac{2}{9}$-EF1a-DIO-ChR2-EYFP) into the LPB of Cck-cre mice, followed by optical fiber implantation and optic stimulation (Figure 2B–D). Since LPB has been shown to mediate aversion (Chiang et al., 2019), we first tested whether optogenetic activation of LPBCCK neurons affects aversion using the real-time place aversion (RTPA) test (Figure 2E). We found that activation of LPBCCK neurons significantly reduced the duration of mice in the laser-paired chamber, accompanied by rapid running or flight to the other laser-unpaired chamber, indicating an obvious aversion-like avoidance and/or fear-related defensive-like behaviors (Figure 2F–G).
To further investigate whether LPBCCK neurons plays a role in defensive behaviors, we used a well-established flight-to-nest behavioral test by putting a nest in the corner of an arena to test the ability of mice to actively search for hiding (Figure 2H). Of note, activating LPBCCK neurons induced robust flight-to-nest behavior in mice, with much shorter latency and faster speed running towards the nest (latency: EYFP, 87.71 ± 22.38 s vs. ChR2, 7.571 ± 0.8123 s; speed: EYFP, $184.4\%$ ± $30.4\%$ vs. ChR2, $361.7\%$ ± $31.5\%$), followed by longer stay in the nest (EYFP, $63.906\%$ ± $3.645\%$ vs. ChR2, $97.26\%$ ± $2.737\%$; Figure 2I–K). These data suggest that activation of LPBCCK neurons induces aversion-like avoidance and defensive-like flight-to-nest behaviors.
Defensive behaviors are usually accompanied by changes in the autonomic nervous system, such as increased heart rates, dilated pupil size and elevated cortisol levels (Dong et al., 2019; Gross and Canteras, 2012; Wang et al., 2015). We found that photostimulation of LPBCCK neurons also significantly increased heart rates and enlarged pupil diameters, along with elevated plasma levels of corticosterone (Figure 2L–P), suggesting that activating LPBCCK neurons induces fast sympathetic changes, leading to an arousal state accompanying the defensive responses in mice (Salay et al., 2018).
Since activation of LPBCCK neurons induced fear-associated flight-to-nest behaviors and long-term fear may trigger anxiety-like states in animals (Tovote et al., 2005; Yang et al., 2016), we also examined whether prolonged activation of LPBCCK neurons induces anxiety-like behaviors (Figure 2—figure supplement 1A). After 10 min of light stimulation, mice injected with ChR2 exhibited significantly reduced center time and center entries in the open field tests without affecting total distance, and decreased open arm entries and time in the elevated plus maze tests (Figure 2—figure supplement 1B–J), suggesting that prolonged activation of LPBCCK neurons also induces anxiety-like behaviors in mice.
## LPBCCK neurons encode threat stimuli-evoked flight behaviors
LPB neurons were shown to be activated when animals are in proximity to dangerous stimuli (Campos et al., 2018; Day et al., 2004; Han et al., 2015b). To track the endogenous activities of LPBCCK neurons in vivo, we injected the Cre-dependent fluorescent calcium indicator AAV-DIO-GCaMP7s into the LPB of Cck-cre mice, and recorded the calcium responses by fiber photometry (Figure 3A–C). Consistent with previous observation (Yang et al., 2020), we observed elevated calcium responses of LPBCCK neurons by heat stimuli (43 °C) (Figure 3—figure supplement 1A–D). We then tested the response of these neurons in a predator-exposure assay, in which an awake but restrained rat was placed at one end of a rectangular arena (Reis et al., 2021; Weisheng et al., 2021), and a mouse was placed at the other end, away from the rat. After perceiving the presence of the rat, mice usually exhibit risk assessment behaviors by curiously approaching and investigating the rat (Olivier et al., 1991). We observed that the calcium signals of LPBCCK neurons gradually rise when mice approached the rat. At the moment of escape initiation, when mice turned back and dashed away from the rat, the calcium signals reached a peak (Figure 3D–G and L). However, the signals ramped down as mice gained distance from the rat. Notably, activities of LPBCCK neurons were unaffected in the presence of a toy rat (Figure 3—figure supplement 1E–H).
**Figure 3.:** *Threat stimuli recruit LPBCCK neurons to elicit defensive behaviors.(A) Schematic of the in vivo recording system for the Ca2+ signal. (B) Schematic showing the injections and recording of LPB neurons in Cck-cre mice. (C) A representative image (left panel) and a magnified view of neurons labeled by AAV-hsyn-DIO-GCaMP7s, scale bar, 100 μm; n = 5 mice. (D) Schematic of the rat exposure assay. (E–F) A heatmap (E) and a peri-event plot (F) of calcium transients of LPBCCK neurons in a mouse evoked by rat exposure (5 trials) (gray dotted line, onset of approach to the live rat; dark dotted line, onset of flight). (G) Average calcium transients of the tested animals during the rat exposure assay (n = 5 mice). Shaded areas around means indicate error bars. (H) Schematic paradigm of looming stimulus in a nest-containing open-field apparatus. (I–J) A heatmap presentation (I) and a peri-event plot (J) of calcium transients of LPBCCK neurons in a mouse upon looming stimulus (5 trials). (K) Average calcium transients of the tested animals during the looming assay (n = 5 mice). (L–M) Long session calcium recordings of LPBCCK neurons during rat exposure (L) or looming tests (M). (N–O) Correlation analyses between the elevation of calcium transients and the onset of flight during rat exposure or looming tests. (p < 0.0001 Linear regression). (P) Comparison of calcium signals LPBCCK neurons evoked by different stimuli. (Q–R) Plot depicting the differences of the amplitude or the area under the curve (AUC) of calcium signal changes in response to different stimuli. ∆AUC, AUC stimulus signal- AUC basal signal. (Q, Rat vs. Looming: p > 0.9999; Rat vs. TMT: p > 0.9999；TMT vs. Looming: p > 0.9999; one-way ANOVA; R, Rat vs. Looming p > 0.0676; Rat vs. TMT p > 0.1654; TMT vs. Looming p >0.9999; one-way ANOVA).
Figure 3—source data 1.Plot depicting the difference of Ca2+ activity.*
In the presence of the visual predatory cue, such as appearing and expanding looming shadows to mimic an approaching predator (Yilmaz and Meister, 2013), LPBCCK neurons also showed increased calcium signals (Figure 3H). Similar to rat exposure assay, we also tested the response of these neurons in the looming assay, the elevated calcium signals reached peak when the animal initiated escape, but reduced once it ran into the nest (Figure 3H–K and M). LPBCCK neurons also showed elevated calcium signals when mice were exposed to the olfactory predatory cue, TMT odor (Figure 3—figure supplement 1–L). Correlative analysis revealed that the elevation of the calcium signal was correlated with the onset of escape responses (Figure 3N–O). Thus, threatening stimuli of different sensory predatory cue stimulated the LPBCCK neurons (Figure 3P–R). Together, these data demonstrate that LPBCCK neurons encode innate threat stimuli-evoked aversion and flight behaviors.
## Inhibition of LPBCCK neurons suppressed predatory cue-evoked flight responses
To examine whether LPBCCK neurons are required for innate threat-evoked defensive behaviors, we then inhibited LPBCCK neurons using optogenetic tools. By injecting the Cre-dependent AAV expressing the guillardia theta anion channel rhodopsins-1 (GtACR1) into the LPB of Cck-cre mice, followed by optic stimulation, we were able to effectively inhibit the firing of LPBCCK neurons (Figure 4A). Next, we bilaterally injected the control or GtACR1 viruses into the LPB and photo-inhibited these neurons in vivo (Figure 4B–C). In the rat exposure test (Figure 4D–E), after a quick exploration of the rat in the corner (the danger zone), control mice usually fled to the other end of the box (the putative safe zone). However, optic inhibition of LPBCCK neurons significantly increased the time (EYFP, $7.247\%$ ± $2.329\%$ vs. GtACR1, $26.57\%$ ± $5.375\%$) and entries (EYFP, 6.167 ± 1.579 vs. GtACR1, 12.67 ± 2.141) of mice to the danger zone, with no effect on total travel distance (EYFP, 8.837 ± 1.934 m vs. GtACR1, 12.69 ± 1.421 m), suggesting a delayed flight response (Figure 4F–H). In the looming test (Figure 4I–J), inhibition of LPBCCK neurons also increased the latency fleeing to the nest (EYFP, 10.18 ± 2.828 s vs. GtACR1, 27.14 ± 6.526 s), followed by reduced hiding time in the nest (EYFP, $85.36\%$ ± $9.846\%$ vs. GtACR1, $39.17\%$ ± $16.36\%$; Figure 4K–L). Our data suggest that LPBCCK neurons are required for proper defensive behaviors to rat exposure and looming stimuli.
**Figure 4.:** *Optogenetic inhibition of LPBCCK neurons suppresses predator- and visual predatory cue-evoked innate flight responses.(A) Left, schematic of light stimulation of and patch-clamp recording of GtACR1-expressing CCK neurons in the LPB. Right, example of action potentials evoked by optogenetic inhibition of LPBCCK neurons from using whole cell patch-clamp recording. (B) Schematic diagram of optogenetic inhibition of LPBCCK neurons. (C) Representative image showing the GtACR1-EYFP expression in the LPB and optical fiber tip locations above the LPB of a Cck-cre mouse. (D–E) Schematic and the timing and behavioral paradigm of rat exposure assay. (F–H) Photoinhibition of LPBCCK neurons increased number of entries toward the rat (danger zone), time spent in the danger zone, with unchanged travel distance (EYFP: n = 6 mice, GtACR1: n = 9 mice; for times of entries, p = 0.0456, t = 2.21, df = 13; unpaired t test; for time in the danger zone, p = 0.0153, t = 2.791, df = 13; unpaired t test; for total distance, p = 0.1246, t = 1.642, df = 13; unpaired t test). (I–J) Schematic of the looming test apparatus, the timing and behavioral paradigm of looming-evoked flight-to-nest behavioral test. (K–L) Photoinhibition of LPBCCK neurons increased the latency towards the nest and reduced the hiding time in the nest. (EYFP: n = 11 mice, GtACR1: n = 7 mice; for latency, p = 0.0153, t = 2.716, df = 16; unpaired t test; Mann-Whitney test; for time in the nest, p = 0.0201, t = 2.582, df = 16; unpaired t test).
Figure 4—source data 1.Quantification of the defensive responses upon inhibition of LPB CCK neurons.*
## Stimulation of the LPBCCK-PVN pathway triggers defensive-like flight-to-nest behaviors
To further investigate whether activation of the LPBCCK-PVN pathway induces defensive-like flight-to-nest behavior, we unilaterally injected the control or ChR2 virus into the LPB of Cck-cre mice and implanted an optical fiber above the PVN (Figure 5A–E). By photostimulating the axonal terminals of LPBCCK neurons within the PVN, we also observed shorter latency (EYFP, 73.43 ± 21.08 s vs. ChR2, 8 ± 1.047 s), faster speed (EYFP, $122.3\%$ ± $17.81\%$ vs. ChR2, $277.9\%$ ± $32.85\%$) running towards the nest, and longer stay in the nest, (EYFP, $3.751\%$ ± $1.23\%$ vs. ChR2, $89.29\%$ ± $10.71\%$; Figure 5F–H), similar to soma activation. Post-hoc c-fos staining further verified the activation of PVN neurons after optic stimulation (Figure 5I–J). As well, terminal activation also increased heart rates (Figure 5K) and plasma corticosterone levels (Figure 5N), but with no effects on pupil size (Figure 5L–M). Together, these data suggest that activation of the LPBCCK-PVN pathway is sufficient to trigger flight-to-nest behavior, along with increased heart rates and corticosterone levels.
**Figure 5.:** *Photostimulation of the LPBCCK-PVN pathway induces defensive-like flight-to-nest behavior.(A) Schematic diagram of optogenetic activation of LPBCCK-PVN pathway. (B–C) A representative image showing the ChR2-EYFP expression in the LPB (B) and the optical fiber tip locations in the PVN (C) of a Cck-cre mouse. (D) Schematic of the timing and behavioral paradigm with optical activation of LPBCCK-PVN pathway. (E) Schematic of the experimental apparatus with a nest in the corner. (F–H) Optogenetic activation of the LPBCCK-PVN pathway shortened the latency but increased the speed of animals towards the nest, with increased hiding time in the nest (EYFP: n = 7 mice, ChR2: n = 7 mice; for latency, p = 0.0012, U = 0; Mann-Whitney test; for speed, p = 0.0013, t = 4.163, df = 12; unpaired t test; for time in the nest, p = 0.0006, U = 0; Mann-Whitney test). (I–J) C-fos staining in the PVN (I) and quantification of c-fos positive cells in the PVN (J). Scale bar: 100 μm. (EYFP: n = 3 mice, ChR2: n = 3 mice; p = 0.0033, t = 6.253, df = 4; unpaired t test). (K) Mean heart rate analyses in EYFP and ChR2 groups (EYFP: n = 7 mice, ChR2: n = 7 mice; p = 0.0002, t = 5.249, df = 12; unpaired t test). (L) Example image of computer-detected pupil size before and during photoactivation of LPBCCK neurons. (M) Relative pupil size of animals (during/before photostimulation of LPBCCK neurons) (EYFP: n = 6 mice, ChR2: n = 6 mice; p = 0.0022, U = 0; Mann-Whitney test). (N) Plasma corticosterone levels in EYFP and ChR2 groups (EYFP: n = 5 mice, ChR2: n = 5 mice; p = 0.0079, U = 0; Mann-Whitney test).
Figure 5—source data 1.Quantification of the flight-to-nest behavior and autonomic responses upon activation of LPB CCK-PVN pathway.*
## PVN is required for defensive responses to threatening situations
Since LPBCCK neurons project to multiple regions (Figure 1—figure supplement 2A–I), the effects observed upon their activation may arise from different downstream targets, other than the PVN. To test this possibility, we investigated whether inhibition of the PVN is required for the LPBCCK neurons-induced defensive-like flight-to-nest behavior. To achieve this aim, we used a dual virus-mediated optogenetic activation and chemogenetic inhibition strategy, by injecting AAV$\frac{2}{9}$-DIO-ChR2-EYFP into the LPB, and AAV$\frac{2}{9}$-hsyn-hM4Di-mCherry into the PVN of Cck-cre mice. By putting an optical fiber above the PVN, we were able to activate the LPBCCK terminals and induce defensive-like flight-to-nest behavior as shown above (Figure 6A–B). Upon CNO administration to inhibit PVN neurons, we observed a much longer latency, lower speed of the mice in returning to the nest, and reduced hiding time in the nest (latency: saline, 8.286 ± 1.475 s vs. CNO, 18.29 ± 2.843 s; speed: saline, $289.3\%$ ± $25.22\%$ vs. CNO, $124.7\%$ ± $13.67\%$; time in the nest: saline, $81.55\%$ ± $11.16\%$ vs. CNO, $25.83\%$ ± $10.32\%$; Figure 6C–E). These data suggest that the PVN is a required downstream target for LPBCCK neurons activation-induced defensive behavior.
**Figure 6.:** *PVN is involved in the defensive-like flight-to-nest behavior evoked by LPBCCK neurons.(A) Schematic of experimental setup. (B) Left, representative image showing the ChR2-EYFP expression in the LPB of a Cck-cre mouse. Scale bar, 100 μm. Right, representative image showing the hM4Di-mCherry expression in the PVN and optic fiber placement in the PVN from a ChR2-EYFP expressing mouse. (C–E) Chemogenetic inhibition of PVN neurons before optical activation of LPBCCK-PVN terminals increased latency and reduced the speed of animals towards the nest, with reduced hiding time in the nest (saline: n = 7 mice, CNO: n = 7 mice; for latency, p = 0.0088, t = 3.122, df = 12; unpaired t test; for speed, p < 0.0001, t = 5.736, df = 12; unpaired t test; for time in the nest, p = 0.0032, t = 3.665, df = 12; unpaired t test). (F) Optogenetic activation of LPBCCK -PVN terminals with a cannula implanted in the PVN for aCSF, CNQX +AP5, or Devazepide +L-365 260 delivery. Scale bar, 100 μm. (G) Representative image showing the ChR2-EYFP expression in the LPB of a Cck-cre mouse. Scale bar, 100 μm. (H) Representative image showing the cannula placement in the PVN from a ChR2-EYFP expressing Cck-cre mouse. Scale bar, 100 μm. (I–K) Microinjection of the glutamate receptor antagonists (CNQX +AP5), rather than CCK receptor antagonists (Devazepide +L-365260), into the PVN increased the latency to the nest and reduced the hiding time in the nest (ACSF: n = 6 mice, CNQX +AP5: n = 6 mice, Devazepide +L-365260: n = 5 mice; for latency, F(2,14) = 7.658, p = 0.057; One-way ANOVA; for speed, F(2,14) = 11.49, p = 0.0011; for time in the nest, One-way ANOVA; F(2,14) = 16.44, p = 0.0002; One-way ANOVA). (L) Schematic diagram of optogenetic activation of PVNVglut2 neurons. (M) Representative image showing the ChR2-EYFP expression and optical fiber tip locations in the PVN of a Vglut2-cre mouse. Scale bar, 100 μm. (N–P) Optogenetic activation of PVNVglut2 neurons reduced the latency, increased the speed of animals towards a nest and the time in the nest. (EYFP: n = 6 mice, ChR2: n = 6 mice; for latency, p = 0.0065, U = 2; Mann-Whitney test; for speed, p = 0.0221, t = 2.705, df = 10; unpaired t test; for time in the nest, p = 0.022, U = 0; Mann-Whitney test). (Q) Graphical summary showing the LPBCCK-PVN pathway in mediating defensive behaviors.
Figure 6—source data 1.Quantification of the flight-to-nest behavior upon manipulation of PVN neurons.
Figure 6—source data 2.Maps for virus expression and optical fiber location.*
LPBCCK neurons may release either glutamate or neuropeptide CCK to induce the defensive behavior. To determine which neurotransmitter or modulator is engaged in the above responses, we combined pharmacologic with optogenetic manipulation. Two types of CCK receptors are present in the central nervous system and *Cckar is* widely distributed throughout the PVN, whereas the expression of Cckbr in PVN is low (Figure 6—figure supplement 1A–B). We infused glutamate receptor antagonists (CNQX +AP5), CCK receptor antagonists (Devazepide +L-365,260), or artificial cerebrospinal fluid (ACSF) through an implanted cannula 30 min before optogenetic activation of LPBCCK neurons (Figure 6F–H). We found that blocking glutamate receptors induced a longer latency, slower speed in returning to the nest, along with reduced hiding time in the nest (Figure 6I–K), when compared with ACSF-treated mice. However, blocking CCK receptors showed no significant effects (Figure 6I–K). These results suggest that glutamatergic transmission from LPBCCK neurons to the PVN mediates defensive-like flight-to-nest behavior.
## Photostimulation of PVNVglut2 neurons promotes defensive-like flight-to-nest behavior
Of note, PVN neurons are mostly glutamatergic (Vong et al., 2011; Xu et al., 2013). We further assessed whether optogenetic activation of PVNVglut2 neurons elicit defensive-like behavior (Figure 6L–M). We found that optogenetic activation of PVNVglut2 neurons also induced flight-to-nest behavior (latency: EYFP, 127.5 ± 22.5 s vs. ChR2, 12.83 ± 2.937 s; speed: EYFP, 142.4 ± $32.07\%$ vs. ChR2, $384.4\%$ ± $83.52\%$), followed by longer hiding time in the nest (EYFP, $0.2783\%$ ± $0.2783\%$ vs. ChR2, $61.11\%$ ± $17.58\%$) (Figure 6N–P), compared to the control. These data suggest that optogenetic activation of PVNVglut2 neurons was sufficient to evoke defensive-like flight-to-nest behavior.
PVNCRH neurons have recently been shown to predict the occurrence of defensive behaviors in mice (Daviu et al., 2020). Interestingly, optogenetic activation of the LPBCCK-PVN pathway primarily activated PVNCRH neurons, rather than vasopressin or oxytocin neurons (Figure 6—figure supplement 1C–F). However, optogenetic activation of PVNCRH neurons was not sufficient to drive defensive-like flight-to-nest behavior (Figure 6—figure supplement 1G–K; latency: EYFP, 77 ± 20.69 s vs. ChR2, 90.17 ± 22.03 s; speed: EYFP, 156.6 ± $22.59\%$ vs. ChR2, $217.2\%$ ± $61.54\%$; time in the nest: EYFP, $20.95\%$ ± $13.89\%$ vs. ChR2, $35.28\%$ ± $20.48\%$).
## Discussion
Defensive behaviors are actions that are naturally selected to avoid or reduce potential harm for the survival of animal species. While emerging evidence has suggested that LPB are associated with sensation of danger signals and aversive stimuli (Campos et al., 2018; Han et al., 2015b), we found that LPBCCK neurons were actively recruited upon exposure to various predatory stimuli. Activating LPBCCK neurons triggers aversive-like avoidance and defensive-like flight responses, whereas inhibition of these neurons suppressed predatory stimuli-induced defensive behaviors. Inhibition of downstream PVN neurons attenuated photoactivation of LPBCCK-PVN terminals-promoted flight responses, suggesting that LPBCCK-PVN pathway is important for the regulation of defensive responses. Our study thus reveals a new connection from the brainstem to the hypothalamus to regulate innate defensive behaviors (Figure 6Q).
Hippocampal and cortical CCK-expressing neurons have been thought of mainly GABAergic (Liu et al., 2020; Sun et al., 2020a; Whissell et al., 2015). We found that LPBCCK neurons are predominantly glutamatergic, similar to those in the amygdala (Shen et al., 2019). While the roles of LPBCGRP neurons have been associated with the passive defensive responses (Han et al., 2015b; Keay and Bandler, 2001), we show that activation of LPBCCK neurons primarily triggers active defensive responses, such as flight-to-nest behavior (Keay and Bandler, 2001). The fact that CGRP and CCK neurons are spatially segregated, responsive to distinct threat signals and involved in different defensive responses, indicating that LPB acts as an important integration center or hub to gate defensive responses when animals encounter different threats (Liu et al., 2022; Tokita et al., 2009). It would be interesting to investigate whether LPBCCK and LPBCGRP neurons coordinately regulate defensive responses under different threatening situations.
In vivo calcium imaging showed that LPBCCK neurons increase firing upon exposure to predators and/or predatory cues, which occurs before the initiation of an escape behavior. These findings indicate that LPBCCK neurons may produce a preparatory signal before the escape initiation, which allows the downstream targets to further assess the threat signals, plan escape routes and take appropriate motor actions. LPBCCK neurons are well positioned to play such an important role in linking the imminent threat to escape initiation, as they receive inputs from both the peripheral and visceral sensory system and project to many brain regions involved in defensive responses including the hypothalamus, PVT and PAG (Cooper and Blumstein, 2015; Ellard and Eller, 2009; Han et al., 2015b; Krout and Loewy, 2000; Saper and Loewy, 1980; Sun et al., 2020b; Tokita et al., 2009). While PVNCRH or PAGCCK neurons have also been shown to be recruited during flight, their peak activity did not match the flight initiation, but occurred during flight (Daviu et al., 2020; La-Vu et al., 2022). The peak activities of LPBCCK neurons at the onset of escape suggest that LPBCCK neurons likely gate and instruct the escape rather than directly control locomotive behaviors.
Despite many early studies mapping downstream projections for LPB neurons (Han et al., 2015a; Saper and Loewy, 1980; Sun et al., 2020b), we are the first to demonstrate that LPBCCK neurons directly connect with PVN neurons. Direct projections from the brainstem to the hypothalamus for the escape responses may be more conducive to escape in emergency situations. Besides the PVN, we also observed that LPBCCK neurons project to the PVT, VMH and PAG, which have also been associated with aversion and/or defensive behaviors (LeDoux, 2012; Vianna et al., 2001; Wang et al., 2015). It would be interesting to further investigate how these different downstream areas coordinately contribute to defensive responses.
PVN is important for neuroendocrine and autonomous regulation (Canteras et al., 2001; Coote, 2005; Ferguson et al., 2008; Sutton et al., 2016). While earlier c-fos staining analyses suggest that PVN neurons are activated in response to different threat stimuli (Canteras et al., 2001; Faturi et al., 2014; Martinez et al., 2008; Staples et al., 2008), it remains unclear whether they are only involved in neuroendocrine responses or also defensive behaviors. Increasing evidence has demonstrated that PVN neurons play important roles in defensive behaviors independent of hormonal actions (Daviu et al., 2020; Mangieri et al., 2019; Xu et al., 2019). Sim1, as a specific marker of PVN, studies have reported that photoactivation of PVNSim1 neurons induced defensive responses (Mangieri et al., 2019). Our findings support a role for PVN neurons in mediating behavioral responses, as photoactivation of PVNVglut2 neurons promoted flight-to-nest behavior. Intriguingly, stimulating the LPBCCK-PVN pathway promoted flight-to-nest behaviors but activating of PVNCRH neurons did not induce apparent flight-to-nest behaviors. Since PVNCRH neurons receive multiple inputs and activation of the PVNCRH neurons have been shown to induce aversion (Kim et al., 2019), future studies using input-specific activation of PVNCRH neuronal subpopulations would further dissect the role of LPBCCK- PVNCRH pathway in defensive responses. Future studies using additional projection-specific approaches and more genetically defined cell tools may help resolve this problem.
Our data show that activation of LPBCCK neurons drives aversion, defensive and anxiety-like behaviors, evoking an arousal and stressed states. Fear, anxiety and stress responses are closely associated with mental illnesses with dysregulated neural circuits (Blanchard et al., 2001). It is worth further investigation to examine whether alterations of the LPBCCK-PVN pathway are implicated in these diseases and whether manipulation of this pathway might be an effective strategy for therapeutic targeting.
## Materials and methods
**Key resources table**
| Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
| --- | --- | --- | --- | --- |
| Strain, strain background (Mus musculus) | Cck-ires-cre(Ccktm1.1(cre)Zjh/J) | The Jackson Laboratory | JAX012706 | |
| Strain, strain background (Mus musculus) | Crh-ires-cre (B6(Cg))-Crhtm1(cre)Zjh/J | The Jackson Laboratory | JAX012704 | |
| Strain, strain background (Mus musculus) | Vglut2-ires-Cre (Slc17a6tm2(cre)Lowl/J) | The Jackson Laboratory | JAX 016963 | |
| Strain, strain background (Mus musculus) | Ai14 (B6;129S6-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J) | The Jackson Laboratory | JAX007908 | |
| Strain,strainbackground(Rattus norvegicus) | SD | ShanghaiSLACLaboratoryAnimal Co.Ltd | http://www.slaccas.com/ | |
| Strain, strain background (Mus musculus) | C57BL/6 J | Shanghai SLAC Laboratory Animal Co. Ltd | http://www.slaccas.com/ | |
| Antibody | Anti-Dsred, rabbit polyclonal | Takara | Cat# 632496RRID: AB_10013483 | (1:800) |
| Antibody | Anti-CGRP, mouse monoclonal | Abcam | Cat# 81887RRID: AB_1658411 | (1:800) |
| Antibody | Anti-GFP, goat polyclonal | Abcam | Cat# 5450RRID: AB_304896 | (1:500) |
| Antibody | Anti-CRH, rabbit polyclonal | Phoenix Biotech | Cat# H-019–06 | (1:500) |
| Antibody | Anti-c-fos,Guinea pig polyclonal | Synaptic Systems | Cat# 226004RRID: AB_2619946 | (1:10,000) |
| Antibody | Alexa Fluor 488 donkey anti-guinea pig IgG (H+L) polyclonal | Jackson | Cat#112-486-068RRID: AB_2617153 | (1:1000) |
| Antibody | Alexa Fluor 555 donkey anti-rabbit IgG (H+L) polyclonal | Invitrogen | Cat#A31572RRID: AB_162543 | (1:1000) |
| Antibody | Alexa Fluor 488 donkey anti-mouse IgG (H+L) polyclonal | Invitrogen | Cat# R37114RRID: AB_2556542 | (1:1000) |
| Antibody | Alexa Fluor 488 donkey anti-goat IgG (H+L) polyclonal | Invitrogen | Cat# A11055RRID: AB_2534102 | (1:1000) |
| Antibody | Alexa Fluor 488 donkey anti-rabbit IgG (H+L) polyclonal | Jackson | Cat# R37118RRID: AB_2556546 | (1:1000) |
| Commercial assay or kit | RNAscope Multiplex Fluorescent Reagent Kit v2 | Advanced Cell Diagnostics | Cat# 323100 | |
| Sequence-based reagent | RNAscope probe Slc17a6 | Advanced Cell Diagnostics | accessionnumber NM_080853.3 | probe region1986–2998 |
| Sequence-based reagent | RNAscope probe Slc32a1 | Advanced Cell Diagnostics | Accessionnumber NM_009508.2 | probe region894–2037 |
| Sequence-based reagent | RNAscope probe Cckar | Advanced Cell Diagnostics | accessionnumber NM_009827.2 | probe region328–1434 |
| Sequence-based reagent | RNAscope probe Cckbr | Advanced Cell Diagnostics | accessionnumber NM_007627.4 | probe region136–1164 |
| Chemical compound, drug | CNQX disodium salt hydrate | Sigma-Aldrich | Cat#1045 | |
| Chemical compound, drug | Devazepide | Sigma-Aldrich | Cat#2304 | |
| Chemical compound, drug | Clozapine N-oxide | Sigma-Aldrich | Cat#C0832 | |
| Chemical compound, drug | L-365,260 | Sigma-Aldrich | Cat#143626 | |
| Chemical compound, drug | DAPI | Sigma-Aldrich | | |
| Chemical compound, drug | Tween-20 | Sigma-Aldrich | | |
| Strain, strain background (AAV2/9) | AAV2/9-hEF1a-DIO-hChR2(H134R)-EYFP-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat# S0199-9 | Viral titers: 2.95x1013 particles/ml |
| Strain, strain background (AAV2/9) | AAV2/9-hEF1a-DIO-EYFP-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat#S0196-9 | Viral titers:1.0x1012 particles/ml |
| Strain, strain background (AAV2/9) | AAV2/9-CAG-DIO-hGtACR1-P2A-EGFP-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat# S0311-9 | Viral titers:5×1013 particles/ml |
| Strain, strain background (AAV2/9) | rAAV2/9-hSyn-DIO-mGFP-2A-Synaptophysin-mRuby | Shanghai Taitool Bioscience Co. | Cat# S0250-9 | Viral titers:1.55x1013particles/ml |
| Strain, strain background (AAV2/2) | rAAV2/2-Retro-hEF1a-DIO-EYFP-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat# S0196-2R | Viral titers: 2.52x1013 particles/ml |
| Strain, strain background (AAV2/9) | AAV2/9-hSyn-mCherry-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat# S0238-9 | Viral titers:≥1.0 × 1013 particles/ml |
| Strain, strain background (AAV2/9) | AAV2/9-hSyn-hM4D(Gi)-mCherry-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat# S0279-9 | Viral titers:≥1.0 × 1013 particles/ml |
| Strain, strain background (AAV2/9) | AAV2/9-hsyn-DIO-jGCaMP7s-WPRE-pA | Shanghai Taitool Bioscience Co. | Cat# S0590-9 | Viral titers:≥1.0 × 1013 particles/ml |
| Software, algorithm | ANY-Maze software 5.3 | Global Biotech Inc | http://www.anymaze.co.uk/ | |
| Software, algorithm | Image J | NIH | https://imagej.nih.gov/ij/index.html;%20RRID:SCR_003070 | |
| Software, algorithm | GraphPad Prism 6 | GraphPad Software | https://www.graphpad.com/scientificsoftware/prism/;RRID: SCR_002798 | |
| Software, algorithm | MatLab R2016a | MathWorks | https://www.mathworks.com/products.html;RRID:SCR_001622 | |
## Animals
Cck-ires-cre (Ccktm1.1(cre)Zjh/J; Stock No. 012706), Crh-ires-cre (B6(Cg)-Crhtm1(cre)Zjh/J; Stock No. 012704), Vglut2-ires-cre (Slc17a6tm2(cre)Lowl/J; Stock No. 016963), Ai14 (B6;129S6-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J; Stock No. 007908), C57BL/6 mice and SD rats were obtained from the Shanghai Laboratory Animal Center. Adult male mice and SD rats were used in our study. Mice and rats housed at 22 ± 1 °C and 55 ± $5\%$ humidity on a 12 hr light/12 hr dark cycle (light on from 07:00 to 19:00) with food and water ad libitum. All experimental procedures were approved by the Animal Advisory Committee at Zhejiang University and were performed in strict accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. All surgeries were performed under sodium pentobarbital anesthesia, and every effort was made to minimize suffering.
## Immunohistochemistry
Animals were transcardially perfused with saline and $4\%$ PFA. Brains were post-fixed overnight in $4\%$ PFA at 4 °C, followed by immersed in $30\%$ sucrose solution. Coronal sections (40 μm) were cut by a CM1950 Microtome (Leica). Immunostaining was performed as previously described (Zhang et al., 2018). The brain slices were permeabilized in $0.5\%$ Triton X-100 in Tris-buffered saline, blocked with 100 mM glycine and $5\%$ bovine serum albumin (BSA) containing $5\%$ normal donkey serum. Tissue sections were subsequently incubated with diluted primary and secondary antibodies as indicated, nuclei stained with 6-diamidino2-phenylindole (DAPI), and slides mounted with antifade reagents. The primary antibodies used were: Guinea pig anti-c-fos (Synaptic Systems, Cat# 226004), Rabbit anti-Dsred (Takara, Cat# 632496), Goat anti-GFP (Abcam, Cat# 5450), Mouse anti-CGRP (Abcam, Cat# 81887). Slides were imaged with a confocal microscope (Olympus FluoView FV1200).
## RNAscope in situ hybridization
We used RNAscope multiplex fluorescent reagent kit and designed probes (ACDBio Inc) to perform fluorescence in situ hybridization. Mouse brain tissue was sectioned into 20 μm sections by cryostat (Leica CM 1950). Then sections were mounted on slides and air-dried at room temperature. Subsequently, the sections were dehydrated in $50\%$ EtOH, $70\%$ EtOH, and $100\%$ EtOH for 5–10 min each time and air-dried at room temperature again. Thereafter, protease digestion was performed in a 40 °C HybEZ oven for 30 min pretreatment, slides were hybridized with pre-warmed probe in a 40 °C HybEZ oven for 2 hr. Probes used in our paper were: Slc17a6 probe (VGLUT2, accession number NM_080853.3, probe region 1986–2998), Slc32a1 probe (VGAT, accession number NM_009508.2, probe region 894–2037), Cckar probe (accession number NM_009827.2, probe region 328–1434), Cckbr probe (accession number NM_007627.4, probe region 136–1164). After hybridization, the brain sections went through four steps of signal amplification fluorescent label. Anti-DsRed or GFP staining was performed after the RNAscope staining. Slides were imaged with a confocal microscope (Olympus FluoView FV1200).
## Stereotaxic injections and optical fiber/cannula implantation
For surgical procedures, mice were anaesthetized with sodium pentobarbital (0.1 g/kg) and placed in a stereotaxic apparatus (RWD). Stereotaxic surgery was performed as described. Briefly, holes were made into the skull over the target areas to inject the virus with glass pipettes (diameter 10–15 mm) or to implant optical fibers (outer diameter [o.d.]: 200 μm; length: 6.0 mm, 0.37 NA; Inper) or to implant of guide cannula (outer diameter [o.d.]: 0.41 mm; RWD). The coordinates relative to bregma were as follows according to the Paxinos and Franklin [2001] atlas. For all experiments, mice with incorrect injection sites were excluded from further analysis.
For optical activation of LPBCCK neurons, AAV$\frac{2}{9}$-hEF1a-DIO-hChR2(H134R)-EYFP-WPRE-pA (viral titers: 2.95×1013 particles/ml; Taitool Bioscience) or AAV$\frac{2}{9}$-hEF1a-DIO-EYFP-WPRE-pA (viral titers: 1.0×1012 particles/ml; Taitool Bioscience) virus was unilaterally microinjected into the LPB (AP: –4.8; ML: –1.35; DV: –3.4; mm relative to bregma) of Cck-cre mice. The virus was diluted into 5.9×1012 genomic copies per ml with phosphate-buffered saline (PBS) before use and injected with 65 nl into the LPB. Virus was delivered at a flow rate of 10 nl/min. The glass capillary was left in place for an additional 10 min after injection to allow diffusion of the virus. The cannulas were held in place with dental cement above the LPB (AP: –4.8; ML: –1.35; DV: –3.2; mm relative to bregma).
For optical activation of PVNVglut2 neurons, AAV$\frac{2}{9}$-EF1a-DIO-hChR2-EYFP (viral titers: 2.95×1013 particles/ml; Taitool Bioscience) or AAV$\frac{2}{9}$-EF1α-DIO-EYFP (viral titers: 1.0×1012 particles/ml; Taitool Bioscience) virus was unilaterally microinjected into the PVN (AP: –0.4; ML: –0.15; DV: –4.95; mm relative to bregma) of Vglut2-cre mice. The virus was diluted into 2.0×1012 genomic copies per ml with PBS before use and injected with 70 nl into the PVN. Virus was delivered at a flow rate of 15 nl/min. The glass capillary was left in place for an additional 10 min after injection to allow diffusion of the virus. The cannulas were held in place with dental cement above the PVN (AP: –0.4; ML: –0.15; DV: –4.7; mm relative to bregma).
For optical activation of PVNCRH neurons, AAV$\frac{2}{9}$-EF1a-DIO-hChR2-EYFP (viral titers: 2.95×1013 particles/ml; Taitool Bioscience) or AAV$\frac{2}{9}$-EF1α-DIO-EYFP (viral titers: 1.0×1012 particles/ml; Taitool Bioscience) was unilaterally microinjected into the PVN (AP: –0.4; ML: –0.15; DV: –4.95 mm relative to bregma) of Crh-cre mice. The virus was diluted into 4.0×1012 genomic copies per ml with PBS before use and injected with 75 nl into the PVN. Virus was delivered at a flow rate of 15 nl/min. The glass capillary was left in place for an additional 10 min after injection to allow diffusion of the virus. The cannulas were held in place with dental cement above the PVN (AP: –0.4; ML: –0.15; DV: –4.7; mm relative to bregma).
For optical inhibition of LPBCCK neurons, AAV$\frac{2}{9}$-EF1α-DIO-hGtACR1-P2A-eYFP-WPRE (viral titers: 5×1013 particles/ml; Taitool Bioscience) or AAV$\frac{2}{9}$-hEF1a-DIO-EYFP-WPRE-pA (viral titers: 1.0×1012 particles/ml; Taitool Bioscience) virus was bilaterally infused to the LPB (AP: –4.8; ML: –1.35; DV: –3.4; mm relative to bregma) of Cck-cre mice. The virus was diluted into 1.67×1012 genomic copies per ml with PBS before use and injected with 70 nl into the LPB. Optical fibers were bilaterally implanted at an angle of 5° above the LPB (AP: –4.8; ML: –1.72; DV: –2.94; mm relative to bregma).
In order to study the downstream of LPBCCK neurons, AAV$\frac{2}{9}$-hSyn-DIO-mGFP-2A-Synaptophysin-mRuby (viral titers: 1.55×1013 particles/ml; Taitool Bioscience) virus was unilaterally injected into the LPB (AP: –4.8 mm; ML: –1.35 mm; DV: –3.4 mm) of Cck-cre mice; in order to study the source of CCKergic upstream of PVN, AAV$\frac{2}{2}$-Retro-hEF1a-DIO-EYFP-WPRE-pA (viral titers: 2.52×1013 particles/ml; Taitool Bioscience) virus was unilaterally injected into the PVN (AP, –0.4 mm; ML, –0.2 mm; DV, –4.85 mm) of Cck-cre mice.
For LPBCCK-PVN axon terminal stimulation, AAV$\frac{2}{9}$-hEF1a-DIO-hChR2(H134R)-EYFP-WPRE-pA (viral titers: 2.95×1013 particles/ml; Taitool Bioscience) or AAV$\frac{2}{9}$-hEF1a-DIO-EYFP-WPRE-pA (viral titers: 1.0×1012 particles/ml; Taitool Bioscience) virus was unilaterally microinjected into the LPB (AP: –4.8; ML: –1.35; DV: –3.4; mm relative to bregma) of Cck-cre mice. The virus was diluted into 5.9×1012 genomic copies per ml with PBS before use and injected with 65 nl into the LPB. After the virus was expressed for 2 weeks, for LPBCCK-PVN axon terminals stimulation, optical fibers were unilaterally implanted above the PVN (AP: –0.4; ML: –0.2; DV: –4.7; mm relative to bregma).
For pharmacological experiments (procedures of optical activation of LPBCCK neurons have been mentioned above), drug cannulas were ipsilaterally implanted into the PVN (AP, –0.4 mm; ML, –0.2 mm; DV, –4.7 mm).
For prolonged inhibition of PVN neurons, AAV$\frac{2}{9}$-hSyn-mCherry-WPRE-pA (viral titers:≥1.0 × 1013 particles/ml; Taitool Bioscience) or AAV$\frac{2}{9}$-hSyn-hM4D(Gi)-mCherry-WPRE-pA (viral titers:≥1.0 × 1013 particles/ml; Taitool Bioscience) virus was microinjected into the PVN (AP, –0.4 mm; ML, –0.2 mm; DV, –4.85 mm) of Cck-cre mice.
For fiber photometry experiments, AAV$\frac{2}{9}$-hSyn-DIO-GCaMP7s-WPRE virus (viral titers: 2.0×1012 particles/ml; Taitool Bioscience) were injected into the LPB of CCK-ires-Cre mice. After two weeks, an optical fiber was implanted into the LPB, then each mouse was allowed to recover for 1 week before recording. Each mouse was handled for 3 days prior to fiber photometry recording.
## Fiber photometry
Mice were allowed to recover from surgery for at least 7 days before the behavioral experiments. The fiber photometry system (RWD Life Science Co., Ltd, China) was used for recording fluorescence signal (GCaMP7s and isosbestic wavelengths) which produced by an exciting laser beam from 470 nm LED light and 410 nm LED light. Calcium fluorescence signals were acquired at 60 Hz with alternating pluses of 470 nm and 410 nm light. The power at the end of the optical fiber (200 μm, 0.37NA, 2 m) was adjusted to 20 μW. Recording parameters were set based on pilot studies that demonstrated the least amount of photobleaching, while allowing for the sufficient detection of the calcium response. We used the camera for behavioral video recordings to synchronize calcium recordings. On the experimental day, mice were allowed to acclimate in the home cage for 30 min. Regarding quantification, the filtered 410 nm signal was aligned with the 470 nm signal by using the least-square linear fit. ΔF/F was calculated according to (470 nm signal-fitted 410 nm signal)/(fitted 410 nm signal). And the standard z-score calculation method is used, that is, Z-score = (x-mean)/std, x = △F/F. During the behavior experiments, the GCaMP7s fluorescence intensity was recorded.
## In vivo optogenetic manipulation
For optogenetic manipulation experiments, an implanted fiber was connected to a 473 nm laser power source (Newdoon Inc, Hangzhou, China). The power of the blue (473 nm; Newdoon Inc, Hangzhou, China) was 0.83–3.33 mW mm2 as measured at the tip of the fiber. 473 nm laser (ChR2: power 5–15 mW, frequency 20 Hz, pulse width 5ms; GtACR1: power 15 mW, direct current) was supplied to activate or inhibit neurons, respectively.
## Pharmacological antagonism
Antagonists were delivered 30 min before optical activation of LPBCCK neurons. For blocking glutamatergic neurotransmission, we firstly connected guiding cannula with a Hamilton syringe via a polyethylene tube. Then we infused 0.25 μl mixed working solution containing CNQX disodium salt hydrate (0.015 μg; Sigma-Aldrich), a glutamate AMPA receptor antagonist, and AP5 (0.03 μg; Sigma-Aldrich), a NMDA receptor antagonist, into the PVN with a manual microinfusion pump (RWD, 68606) over 5 min. For blocking CCKergic neurotransmission, 0.25 μl mixed solution containing Devazepide (0.0625 μg; Sigma-Aldrich), a Cckar receptor antagonist, and L-365,260 (0.0625 μg; Sigma-Aldrich), a Cckbr receptor antagonist, was infused into the PVN over a period of 5 min. To prevent backflow of fluid, the fluid-delivery cannula was left for 10 min after infusion.
## Chemogenetic manipulation
Clozapine Noxide (CNO, Sigma) was dissolved in saline (5 mg in 10 µL DMSO and 190 µL $0.9\%$ NaCl solution). CNO was injected intraperitoneally at 0.3 mg per kg of body weight for chemogenetic manipulation.
## c-Fos staining and analysis
For c-fos quantification, mice were perfused 1.5 hr after 10 min blue photostimulation illumination, and sections were cut. The boundaries of the nuclei were defined according to brain atlases Mouse Brain Atlas (Franklin and Paxinos, 2008). Cell counting was carried out manually.
## Behavioral task
For all behavioral tests, experimenters were blinded to genotypes and treatments. Mice were handled daily at least seven days before performing behavioral tests. All the apparatuses and cages were sequentially wiped with $70\%$ ethanol and ddH2O then air-dried between stages. At the end of behavioral tests, mice were perfused with $4\%$ PFA followed by post hoc analysis to confirm the viral injection sites, optic fiber and cannula locations. Mice with incorrect viral injection sites, incorrect positioning of optical fibers or cannula were excluded.
## Real-time place aversion test
Mice were habituated to a custom-made 20×30 × 40 cm two-chamber apparatus (distinct wall colors and stripe patterns) before the test. First stage: each mouse was placed in the center and allowed to explore both chambers without laser stimulation for 10 min. After exploration, the mouse indicated a small preference for one of the two chambers. Second stage: 473 nm laser stimulation (20 Hz, 8 mW, 5ms) was delivered when the mouse entered or stayed in the preferred chamber, and the light was turned off when the mouse moved to the other chamber for 10 min.
## Flight-to-nest test
Flight-to-nest test was performed using previously described methods (Zhou, Z., et al, 2019). Flight-to-nest test was performed in a 40×40 × 30 cm closed box with a 27-inch LED monitor stationed on top to display the stimulus. A nest in the shape of a 20 cm wide ×12 cm high triangular prism was in the corner of the closed Plexiglas box. There are two cameras, one from the top, one from the side (Logitech) that record the mouse’s activity simultaneously. Briefly, on day 1, the mice were habituated to the box conditions for 15 min. On day 2, the mice were first allowed to explore the box for 5–10 min. When the mice were in the corner furthest from the nest and within in a body-length distance from the wall, they were given optical stimulus. For optogenetic activation of LPBCCK neurons or LPBCCK-PVN experiments, mice received a 20 s 473 nm blue laser (frequency 20 Hz, pulse width 5ms) with 15–20 mW (terminal) or 5–10 mW (soma) light power at the fiber tips. For prolonged inhibition of PVN neurons, Clozapine Noxide (CNO, Sigma-Aldrich) was dissolved in saline to a concentration of 3 mg/ml. The flight-to-nest test was performed 1.5 hr after CNO intraperitoneally injection.
## Flight-to-nest behavioral analysis
The behaviors of the mice were recorded and analyzed automatically with Anymaze software (Global Biotech Inc). Behavioral analysis was performed as previously described (Zhou, Z., et al, 2019). Flight-to-nest behavior was characterized on the basis of the three aspects: latency to return nest, speed (% of baseline speed), time spent in nest (% of 2 min). Latency to return nest refers to the moment from optical stimulus onset to moment when the mouse first went into the nest. Speed (% of baseline speed) refers to the ratio of the post-stimulation speed to the baseline speed. We recorded the speed of the mice in the 50 s before photostimulation presentation as the baseline speed. The post-stimulation speed was record from the time of stimulation to 15 s after. It was averaged over a 1 s time window centered on the maximum speed. Time spent in nest (% of 2 min) refers to the time from mouse’s body was first completely under the shelter after photostimulation to the time when the mouse left the nest.
## Open-field test
The open-field chamber was made of plastic (50 × 50 × 50 cm). At the start of the test, mice were placed in the periphery of the open-field chamber. The open-field test lasted 5 min.
## Elevated plus-maze test
The elevated plus maze was made of plastic with two open arms (30 × 5 cm), two closed arms (30 × 5 × 30 cm) and a central platform (5 × 5 × 5 cm). At the beginning of the experiments, mice were placed in the center platform facing a closed arm. The elevated plus-maze test lasted 5 min.
## Heart rate measurements
Heart rate was measured via a pulse oximeter (MouseOx Plus; Starr Life Sciences). Mice were placed in the home cage with a detector fixed around the neck. After habituation for 10 min, heart rate was simultaneously measured for 10 min while intermittent 2 min period blue light was applied. Each mouse was tested three times, and the mean heart rate was calculated. Heart rate data was analyzed with the MouseOx Plus Conscious Applications Software.
## Pupil size measurements
Mice were adapted to constant room light (100 lx) for 1 hr before testing. Mice were kept unanaesthetized and restrained in a stereotaxic apparatus during the experiment. The pupil size was recorded using Macro module camera under constant light conditions before stimulation, during stimulation, and after stimulation. The test lasted 90 s, consisting of 30 s light off-on-off epochs. The pupil size was later measured by Matlab software.
## Heat exposure assay
We used a translucent plastic box (38×25 cm) divided into two parts, the smaller part is the unheated comfortable zone (5×25 cm), and the larger part with rubber heating pad is the hot uncomfortable zone (33×25 cm). The rubber heating pad was heated to 43℃, and the heat insulated pad was placed in the comfortable zone to avoid the heat. The heating temperature 43℃ was chosen because it would cause heat escape but not heat pain (Wang et al., 2021b; Weisheng et al., 2021).
## Rat exposure assay
We used a rectangular chamber (70×25 × 30 cm). Mice were acclimated to this environment for three days for 10 min each day. During the rat exposure period, a live rat was restrained to one end of the chamber using a harness attached to the chamber wall. As a control, before a live rat exposure, we exposed mice to a toy rat during fiber photometry recording (similar in size and shape to a live rat). For fiber photometry recordings, all mice underwent rat exposure for 20 min. For photoinhibition tests, mice were exposed to live rat and all trials lasted 10 min.
## Looming test
The looming test was performed in a closed Plexiglas box (40×40 × 30 cm) with a shelter in the corner. For looming stimulation, an LCD monitor was placed on the ceiling to present multiple looming stimuli, which was a black disc expanding from a visual angle of 2° to 20° in 0.5 s. The expanding disc stimulus was repeated for 20 times in quick succession and each repeat is followed by a 0.15 s pause. Animals were habituated for 10–15 min in the looming box one day before testing. During the looming test, mice were first allowed to freely explore for 3–5 min. For calcium signal experiment, total five trials of looming stimuli were presented and analyzed. Behavior was recorded for 20 min. For photoinhibition tests, light stimulation was given 1 s before the looming stimulus appears and continue until the looming stimulus ends.
## TMT odor test
During this test, all mice were habituated to the testing environment for three days before any experimental manipulation. For the photometry studies involving odor presentations, mice were placed on a plastic chamber (30×30 × 20 cm). A dish (diameter, 5 cm) with cotton was positioned on the side beside the chamber. When TMT (Ferro Tec, 4 µl of $100\%$) was used as the stimulus, cotton was first wetted with the equivalent volume of 0.1 M PBS. After a habituation period, the dish was replaced with a new one scented with the TMT. The mice were free to explore in the chamber, and behavior was recorded for 20 min.
## Measurements of corticosterone
Blood samples were collected to determine the hormone levels. Taking Blood was performed in the morning by rapidly collecting heart blood after anesthetization. We immediately collected blood after 10 min of light stimulation (20 Hz, 5ms pulse width, 15 s per min, 473 nm). Blood samples were temporarily placed in iced plastic tubes coated with heparin. All serum was prepared after every blood sample was centrifuged at 2000 g for 2.5 min at 4 °C. Supernatant was collected and plasma corticosterone concentration was measured using commercially-available ELISA kits (Enzo ADI-900–097).
## In vitro electrophysiology
Each mouse was anesthetized with pentobarbital sodium (100 mg/kg, i.p.) and decapitated. Then the whole brain was quickly dissected into ice-cold oxygenated ($95\%$ O2 and $5\%$ CO2) artificial cerebrospinal fluid (aCSF) (93 mM N-methyl-D-glucamine, 2.5 mM KCl, 1.2 mM NaH2PO4, 20 mM HEPES, 25 mM D-glucose, 30 mM NaHCO3, 10 mM MgSO4, 0.5 mM CaCl2, 5 mM sodium ascorbate, 3 mM sodium pyruvate, and 1 mM kynurenic acid), followed by cutting coronally into 300 μm slices on a microtome (VTA-1200S; Leica). Slices containing the LPB were transferred to a similar solution (93 mM NaCl, 2.5 mM KCl, 1.2 mM NaH2PO4, 20 mM HEPES, 25 mM D-glucose, 30 mM NaHCO3, 2 mM MgSO4, 2 mM CaCl2，5 mM sodium ascorbate, 3 mM sodium pyruvate, and 1 mM kynurenic acid), and incubated for at lowest 1 h at room temperature (24–26℃). Then the brain slices were transferred to a recording chamber attached to the fixed stage of an BX51WI microscope (Olympus) (solution containing: 125 mM NaCl, 3 mM KCl, 1.25 mM NaH2PO4, HEPES, 10 mM D-glucose, 26 mM NaHCO3, 2 mM MgSO4 and 2 mM CaCl2). Patch glass electrodes were pulled from borosilicate capillaries (BF150-86-10; Sutter Instrument Co, Novato, CA, USA) and filled with artificial intracellular fluid following component: 135 mM CsMeSO3, 10 mM HEPS, 0.5 mM EGTA, 3.3 mM QX-314, 4 mM Mg-ATP, 0.3 mM Na2-GTP, 8 mM Na2-Phosphocreatine. Whole-cell voltage-clamp recordings were made with a MultiClamp 700B amplifier (Molecular Devices). To tested the efficacy of ChR2-mediated activation, LED-generated blue light pulses were applied to recorded neurons using 4 different frequencies (5, 10, 20, and 40 Hz). To test the effects of photoactivation of LPBCCK projection terminals within PVN, blue light pulses were applied to the recorded PVN neurons. To confirm that postsynaptic currents were monosynaptic, the blue light-evoked currents were recorded in the presence of ACSF (Ctrl), TTX (1 μM) and 4-AP (100 μM). To confirm that postsynaptic currents were monosynaptic. CNQX (20 μM) and AP5 (50 μM) were perfused with ACSF to examine the neurotransmitter type used on LPBCCK-PVN projection. Signals were low-pass filtered at 10 kHz and digitized at 10 kHz (MICRO3 1401, Cambridge Electronic Design). Data were acquired and analyzed using Spike2 7.04 software (Cambridge Electronic Design).
## Quantification and statistical analysis
All data analyses were conducted blinded. All statistical analyses were performed with GraphPad Prism (version 7.0) and analyzed by Unpaired Student’s t-tests, one-way ANOVA, two-way ANOVA according to the form of the data. Nonparametric tests were used if the data did not match assumed Gaussian distribution. Animals were randomly assigned to treatment groups. All data were presented as Mean ± SEM, with statistical significance taken as *$p \leq 0.05$, **$p \leq 0.01$ and ***$p \leq 0.001.$
## Funding Information
This paper was supported by the following grants:
## Data availability
All data generated or analysed during this study are included in the manuscript and supporting file. Source data files have been uploaed.
## References
1. Anderson DJ, Adolphs R. **A framework for studying emotions across species**. *Cell* (2014) **157** 187-200. DOI: 10.1016/j.cell.2014.03.003
2. Bertoglio LJ, de Bortoli VC, Zangrossi H. **Cholecystokinin-2 receptors modulate freezing and escape behaviors evoked by the electrical stimulation of the rat dorsolateral periaqueductal gray**. *Brain Research* (2007) **1156** 133-138. DOI: 10.1016/j.brainres.2007.04.038
3. Blanchard DC, Griebel G, Blanchard RJ. **Mouse defensive behaviors: pharmacological and behavioral assays for anxiety and panic**. *Neuroscience and Biobehavioral Reviews* (2001) **25** 205-218. DOI: 10.1016/s0149-7634(01)00009-4
4. Campos CA, Bowen AJ, Roman CW, Palmiter RD. **Encoding of danger by parabrachial CGRP neurons**. *Nature* (2018) **555** 617-622. DOI: 10.1038/nature25511
5. Canteras NS, Ribeiro-Barbosa ER, Comoli E. **Tracing from the dorsal premammillary nucleus prosencephalic systems involved in the organization of innate fear responses**. *Neuroscience and Biobehavioral Reviews* (2001) **25** 661-668. DOI: 10.1016/s0149-7634(01)00048-3
6. Chen L, Liu Y, Su P, Hung W, Li H, Wang Y, Yue Z, Ge MH, Wu ZX, Zhang Y, Fei P, Chen LM, Tao L, Mao H, Zhen M, Gao S. **Escape steering by cholecystokinin peptidergic signaling**. *Cell Reports* (2022) **38**. DOI: 10.1016/j.celrep.2022.110330
7. Chiang MC, Bowen A, Schier LA, Tupone D, Uddin O, Heinricher MM. **Parabrachial complex: a hub for pain and aversion**. *The Journal of Neuroscience* (2019) **39** 8225-8230. DOI: 10.1523/JNEUROSCI.1162-19.2019
8. Cooper WE, Blumstein DT. *Escaping from Predators: Escape Decisions Prior to Pursuit* (2015). DOI: 10.1017/CBO9781107447189
9. Coote JH. **A role for the paraventricular nucleus of the hypothalamus in the autonomic control of heart and kidney**. *Experimental Physiology* (2005) **90** 169-173. DOI: 10.1113/expphysiol.2004.029041
10. Crawford M, Masterson FA. **Species-specific defense reactions and avoidance learning an evaluative review**. *The Pavlovian Journal of Biological Science* (1982) **17** 204-214. DOI: 10.1007/BF03001275
11. Daviu N, Füzesi T, Rosenegger DG, Rasiah NP, Sterley TL, Peringod G, Bains JS. **Paraventricular nucleus CRH neurons encode stress controllability and regulate defensive behavior selection**. *Nature Neuroscience* (2020) **23** 398-410. DOI: 10.1038/s41593-020-0591-0
12. Day HEW, Masini CV, Campeau S. **The pattern of brain c-fos mrna induced by a component of fox odor, 2,5-dihydro-2,4,5-trimethylthiazoline (TMT), in rats, suggests both systemic and processive stress characteristics**. *Brain Research* (2004) **1025** 139-151. DOI: 10.1016/j.brainres.2004.07.079
13. Dong P, Wang H, Shen X-F, Jiang P, Zhu X-T, Li Y, Gao J-H, Lin S, Huang Y, He X-B, Xu F-Q, Duan S, Lian H, Wang H, Chen J, Li X-M. **A novel cortico-intrathalamic circuit for flight behavior**. *Nature Neuroscience* (2019) **22** 941-949. DOI: 10.1038/s41593-019-0391-6
14. Eilam D. **Die hard: a blend of freezing and fleeing as a dynamic defense -- implications for the control of defensive behavior**. *Neuroscience and Biobehavioral Reviews* (2005) **29** 1181-1191. DOI: 10.1016/j.neubiorev.2005.03.027
15. Ellard CG, Eller MC. **Spatial cognition in the gerbil: computing optimal escape routes from visual threats**. *Animal Cognition* (2009) **12** 333-345. DOI: 10.1007/s10071-008-0193-9
16. Evans DA, Stempel AV, Vale R, Ruehle S, Lefler Y, Branco T. **A synaptic threshold mechanism for computing escape decisions**. *Nature* (2018) **558** 590-594. DOI: 10.1038/s41586-018-0244-6
17. Fanselow MS, Bolles RC. **Naloxone and shock-elicited freezing in the rat**. *Journal of Comparative and Physiological Psychology* (1979) **93** 736-744. DOI: 10.1037/h0077609
18. Faturi CB, Rangel MJ, Baldo MVC, Canteras NS. **Functional mapping of the circuits involved in the expression of contextual fear responses in socially defeated animals**. *Brain Structure & Function* (2014) **219** 931-946. DOI: 10.1007/s00429-013-0544-4
19. Ferguson AV, Latchford KJ, Samson WK. **The paraventricular nucleus of the hypothalamus-a potential target for integrative treatment of autonomic dysfunction**. *Expert Opinion on Therapeutic Targets* (2008) **12** 717-727. DOI: 10.1517/14728222.12.6.717
20. Fulwiler CE, Saper CB. **Subnuclear organization of the efferent connections of the parabrachial nucleus in the rat**. *Brain Research Reviews* (1984) **7** 229-259. DOI: 10.1016/0165-0173(84)90012-2
21. Garfield AS, Shah BP, Madara JC, Burke LK, Patterson CM, Flak J, Neve RL, Evans ML, Lowell BB, Myers MG, Heisler LK. **A parabrachial-hypothalamic cholecystokinin neurocircuit controls counterregulatory responses to hypoglycemia**. *Cell Metabolism* (2014) **20** 1030-1037. DOI: 10.1016/j.cmet.2014.11.006
22. Gross CT, Canteras NS. **The many paths to fear**. *Nature Reviews. Neuroscience* (2012) **13** 651-658. DOI: 10.1038/nrn3301
23. Han S, Mao H, Dally WJ. **Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding**. *arXiv* (2015a)
24. Han S, Soleiman MT, Soden ME, Zweifel LS, Palmiter RD. **Elucidating an affective pain circuit that creates a threat memory**. *Cell* (2015b) **162** 363-374. DOI: 10.1016/j.cell.2015.05.057
25. Isosaka T, Matsuo T, Yamaguchi T, Funabiki K, Nakanishi S, Kobayakawa R, Kobayakawa K. **Htr2a-expressing cells in the central amygdala control the hierarchy between innate and learned fear**. *Cell* (2015) **163** 1153-1164. DOI: 10.1016/j.cell.2015.10.047
26. Keay KA, Bandler R. **Parallel circuits mediating distinct emotional coping reactions to different types of stress**. *Neuroscience and Biobehavioral Reviews* (2001) **25** 669-678. DOI: 10.1016/s0149-7634(01)00049-5
27. Kim J, Lee S, Fang Y-Y, Shin A, Park S, Hashikawa K, Bhat S, Kim D, Sohn J-W, Lin D, Suh GSB. **Rapid, biphasic CRF neuronal responses encode positive and negative valence**. *Nature Neuroscience* (2019) **22** 576-585. DOI: 10.1038/s41593-019-0342-2
28. Krout KE, Loewy AD. **Parabrachial nucleus projections to midline and intralaminar thalamic nuclei of the rat**. *The Journal of Comparative Neurology* (2000) **428** 475-494. DOI: 10.1002/1096-9861(20001218)428:3<475::aid-cne6>3.0.co;2-9
29. La-Vu M, Sethi E, Maesta-Pereira S, Schuette PJ, Tobias BC, Reis FM, Wang W, Leonard SJ, Lin L, Adhikari A. **A Genetically-Defined Population in the Lateral and Ventrolateral Periaqueductal Gray Selectively Promotes Flight to Safety**. *bioRxiv* (2022). DOI: 10.1101/2022.01.19.476981
30. LeDoux J. **Rethinking the emotional brain**. *Neuron* (2012) **73** 653-676. DOI: 10.1016/j.neuron.2012.02.004
31. Lin S, Zhu MY, Tang MY, Wang M, Yu XD, Zhu Y, Xie SZ, Yang D, Chen J, Li XM. **Somatostatin-positive neurons in the rostral zona incerta modulate innate fear-induced defensive response in mice**. *Neuroscience Bulletin* (2023) **39** 245-260. DOI: 10.1007/s12264-022-00958-y
32. Liu X, Dimidschstein J, Fishell G, Carter AG. **Hippocampal inputs engage CCK+ interneurons to mediate endocannabinoid-modulated feed-forward inhibition in the prefrontal cortex**. *eLife* (2020) **9**. DOI: 10.7554/eLife.55267
33. Liu S, Ye M, Pao GM, Song SM, Jhang J, Jiang H, Kim J-H, Kang SJ, Kim D-I, Han S. **Divergent brainstem opioidergic pathways that coordinate breathing with pain and emotions**. *Neuron* (2022) **110** 857-873. DOI: 10.1016/j.neuron.2021.11.029
34. Mangieri LR, Jiang Z, Lu Y, Xu Y, Cassidy RM, Justice N, Xu Y, Arenkiel BR, Tong Q. **Defensive behaviors driven by a hypothalamic-ventral midbrain circuit**. *ENeuro* (2019) **6**. DOI: 10.1523/ENEURO.0156-19.2019
35. Martinez RCR, Carvalho-Netto EF, Amaral VCS, Nunes-de-Souza RL, Canteras NS. **Investigation of the hypothalamic defensive system in the mouse**. *Behavioural Brain Research* (2008) **192** 185-190. DOI: 10.1016/j.bbr.2008.03.042
36. Meister B, Broberger C, Villar MJ, Hökfelt T. **Cholecystokinin B receptor gene expression in hypothalamic neurosecretory neurons after experimental manipulations**. *Neuroendocrinology* (1994) **60** 458-469. DOI: 10.1159/000126782
37. Mileikovskii BY, Verevkina SV. **Influence of brain stem structures on the formation of the defensive behavior of animals**. *Neuroscience and Behavioral Physiology* (1991) **21** 126-130. DOI: 10.1007/BF01182887
38. Olivier B, Mos J, Slangen JL. *Animal Models in Psychopharmacology* (1991). DOI: 10.1007/978-3-0348-6419-0
39. O’Shea RD, Gundlach AL. **Regulation of cholecystokinin receptors in the hypothalamus of the rat: reciprocal changes in magnocellular nuclei induced by food deprivation and dehydration**. *Journal of Neuroendocrinology* (1993) **5** 697-704. DOI: 10.1111/j.1365-2826.1993.tb00542.x
40. Penzo MA, Robert V, Tucciarone J, De Bundel D, Wang M, Van Aelst L, Darvas M, Parada LF, Palmiter RD, He M, Huang ZJ, Li B. **The paraventricular thalamus controls a central amygdala fear circuit**. *Nature* (2015) **519** 455-459. DOI: 10.1038/nature13978
41. Perusini JN, Fanselow MS. **Neurobehavioral perspectives on the distinction between fear and anxiety**. *Learning & Memory* (2015) **22** 417-425. DOI: 10.1101/lm.039180.115
42. Reis FM, Lee JY, Maesta-Pereira S, Schuette PJ, Chakerian M, Liu J, La-Vu MQ, Tobias BC, Ikebara JM, Kihara AH, Canteras NS, Kao JC, Adhikari A. **Dorsal periaqueductal gray ensembles represent approach and avoidance states**. *eLife* (2021) **10**. DOI: 10.7554/eLife.64934
43. Salay LD, Ishiko N, Huberman AD. **A midline thalamic circuit determines reactions to visual threat**. *Nature* (2018) **557** 183-189. DOI: 10.1038/s41586-018-0078-2
44. Saper CB, Loewy AD. **Efferent connections of the parabrachial nucleus in the rat**. *Brain Research* (1980) **197** 291-317. DOI: 10.1016/0006-8993(80)91117-8
45. Shang C, Liu Z, Chen Z, Shi Y, Wang Q, Liu S, Li D, Cao P. **A parvalbumin-positive excitatory visual pathway to trigger fear responses in mice**. *Science* (2015) **Jun 26** 1472-1477. DOI: 10.1126/science.aaa8694
46. Shang C, Chen Z, Liu A, Li Y, Zhang J, Qu B, Yan F, Zhang Y, Liu W, Liu Z, Guo X, Li D, Wang Y, Cao P. **Divergent midbrain circuits orchestrate escape and freezing responses to looming stimuli in mice**. *Nat Commun* (2018) **9**. DOI: 10.1038/s41467-018-03580-7
47. Shen C-J, Zheng D, Li K-X, Yang J-M, Pan H-Q, Yu X-D, Fu J-Y, Zhu Y, Sun Q-X, Tang M-Y, Zhang Y, Sun P, Xie Y, Duan S, Hu H, Li X-M. **Cannabinoid CB1 receptors in the amygdalar cholecystokinin glutamatergic afferents to nucleus accumbens modulate depressive-like behavior**. *Nature Medicine* (2019) **25** 337-349. DOI: 10.1038/s41591-018-0299-9
48. Silva BA, Mattucci C, Krzywkowski P, Murana E, Illarionova A, Grinevich V, Canteras NS, Ragozzino D, Gross CT. **Independent hypothalamic circuits for social and predator fear**. *Nat Neurosci* (2013) **16** 1731-1733. DOI: 10.1038/nn.3573
49. Staples LG, McGregor IS, Apfelbach R, Hunt GE. **Cat odor, but not trimethylthiazoline (Fox odor), activates accessory olfactory and defense-related brain regions in rats**. *Neuroscience* (2008) **151** 937-947. DOI: 10.1016/j.neuroscience.2007.11.039
50. Sun X, Bernstein MJ, Meng M, Rao S, Sørensen AT, Yao L, Zhang X, Anikeeva PO, Lin Y. **Functionally distinct neuronal ensembles within the memory engram**. *Cell* (2020a) **181** 410-423. DOI: 10.1016/j.cell.2020.02.055
51. Sun L, Liu R, Guo F, Wen MQ, Ma XL, Li KY, Sun H, Xu CL, Li YY, Wu MY, Zhu ZG, Li XJ, Yu YQ, Chen Z, Li XY, Duan S. **Parabrachial nucleus circuit governs neuropathic pain-like behavior**. *Nature Communications* (2020b) **11**. DOI: 10.1038/s41467-020-19767-w
52. Sutton AK, Myers MG, Olson DP. **The role of PVH circuits in leptin action and energy balance**. *Annual Review of Physiology* (2016) **78** 207-221. DOI: 10.1146/annurev-physiol-021115-105347
53. Tokita K, Inoue T, Boughter JD. **Afferent connections of the parabrachial nucleus in C57BL/6J mice**. *Neuroscience* (2009) **161** 475-488. DOI: 10.1016/j.neuroscience.2009.03.046
54. Tovote P, Meyer M, Ronnenberg A, Ogren SO, Spiess J, Stiedl O. **Heart rate dynamics and behavioral responses during acute emotional challenge in corticotropin-releasing factor receptor 1-deficient and corticotropin-releasing factor-overexpressing mice**. *Neuroscience* (2005) **134** 1113-1122. DOI: 10.1016/j.neuroscience.2005.05.027
55. Tovote P, Esposito MS, Botta P, Chaudun F, Fadok JP, Markovic M, Wolff SB, Ramakrishnan C, Fenno L, Deisseroth K, Herry C, Arber S, Luthi A. **Midbrain circuits for defensive behaviour**. *Nature* (2016) **534** 206-212. DOI: 10.1038/nature17996
56. Vianna DM, Landeira-Fernandez J, Brandão ML. **Dorsolateral and ventral regions of the periaqueductal gray matter are involved in distinct types of fear**. *Neuroscience and Biobehavioral Reviews* (2001) **25** 711-719. DOI: 10.1016/s0149-7634(01)00052-5
57. Vong L, Ye C, Yang Z, Choi B, Chua S, Lowell BB. **Leptin action on GABAergic neurons prevents obesity and reduces inhibitory tone to POMC neurons**. *Neuron* (2011) **71** 142-154. DOI: 10.1016/j.neuron.2011.05.028
58. Wang L, Chen IZ, Lin D. **Collateral pathways from the ventromedial hypothalamus mediate defensive behaviors**. *Neuron* (2015) **85** 1344-1358. DOI: 10.1016/j.neuron.2014.12.025
59. Wang W, Schuette PJ, La-Vu MQ, Torossian A, Tobias BC, Ceko M, Kragel PA, Reis FM, Ji S, Sehgal M, Maesta-Pereira S, Chakerian M, Silva AJ, Canteras NS, Wager T, Kao JC, Adhikari A. **Dorsal premammillary projection to periaqueductal gray controls escape vigor from innate and conditioned threats**. *eLife* (2021a) **10**. DOI: 10.7554/eLife.69178
60. Wang W, Schuette PJ, Nagai J, Tobias BC, Cuccovia V Reis FM, Ji S, de Lima MAX, La-Vu MQ, Maesta-Pereira S, Chakerian M, Leonard SJ, Lin L, Severino AL, Cahill CM, Canteras NS, Khakh BS, Kao JC, Adhikari A. **Coordination of escape and spatial navigation circuits orchestrates versatile flight from threats**. *Neuron* (2021b) **109** 1848-1860. DOI: 10.1016/j.neuron.2021.03.033
61. Wei P, Liu N, Zhang Z, Liu X, Tang Y, He X, Wu B, Zhou Z, Liu Y, Li J, Zhang Y, Zhou X, Xu L, Chen L, Bi G, Hu X, Xu F, Wang L. **Apr 9). processing of visually evoked innate fear by a non-canonical thalamic pathway**. *Nat Commun* (2015) **6**. DOI: 10.1038/ncomms7756
62. Weisheng W, S. P. J, Jun N, Christine TB, Midea C, Shiyu J, Avishek A. *Coordination of escape and spatial navigation circuits orchestrates versatile flight from threats* (2021)
63. Whissell PD, Cajanding JD, Fogel N, Kim JC. **Comparative density of CCK- and PV-GABA cells within the cortex and hippocampus**. *Frontiers in Neuroanatomy* (2015) **9**. DOI: 10.3389/fnana.2015.00124
64. Xu Y, Wu Z, Sun H, Zhu Y, Kim ER, Lowell BB, Arenkiel BR, Xu Y, Tong Q. **Glutamate mediates the function of melanocortin receptor 4 on SIM1 neurons in body weight regulation**. *Cell Metabolism* (2013) **18** 860-870. DOI: 10.1016/j.cmet.2013.11.003
65. Xu Y, Lu Y, Cassidy RM, Mangieri LR, Zhu C, Huang X, Jiang Z, Justice NJ, Xu Y, Arenkiel BR, Tong Q. **Identification of a neurocircuit underlying regulation of feeding by stress-related emotional responses**. *Nature Communications* (2019) **10**. DOI: 10.1038/s41467-019-11399-z
66. Yang H, Yang J, Xi W, Hao S, Luo B, He X, Zhu L, Lou H, Yu Y, Xu F, Duan S, Wang H. **Laterodorsal tegmentum interneuron subtypes oppositely regulate olfactory cue-induced innate fear**. *Nature Neuroscience* (2016) **19** 283-289. DOI: 10.1038/nn.4208
67. Yang WZ, Du X, Zhang W, Gao C, Xie H, Xiao Y, Jia X, Liu J, Xu J, Fu X, Tu H, Fu X, Ni X, He M, Yang J, Wang H, Yang H, Xu XH, Shen WL. **Parabrachial neuron types categorically encode thermoregulation variables during heat defense**. *Science Advances* (2020) **6**. DOI: 10.1126/sciadv.abb9414
68. Yilmaz M, Meister M. **Rapid innate defensive responses of mice to looming visual stimuli**. *Current Biology* (2013) **23** 2011-2015. DOI: 10.1016/j.cub.2013.08.015
69. Zhang B, Wang W, Zhang Z, Hu Y, Meng F, Wang F, Lou H, Zhu L, Godbout R, Duan S, Gao Z. **Alternative splicing of disabled-1 controls multipolar-to-bipolar transition of migrating neurons in the neocortex**. *Cerebral Cortex* (2018) **28** 3457-3467. DOI: 10.1093/cercor/bhx212
70. Zhang L, Hernández VS, Eiden LE. **Hindbrain PBN to CEC and ovbnst HIFI circuits modulate medial and lateral preoptic hypothalamus for defensive locomotion: the discovery of perisomatic PACAP‐glutamatergic calyx‐of‐held‐like synapse in the rodent pallidum nuclei**. *The FASEB Journal* (2020) **34**. DOI: 10.1096/fasebj.2020.34.s1.03234
|
---
title: Bacterial Profile of External Ocular Infections, Its Associated Factors, and
Antimicrobial Susceptibility Pattern among Patients Attending Karamara Hospital,
Jigjiga, Eastern Ethiopia
authors:
- Tigist Abebe
- Zelalem Teklemariam
- Tadesse Shume
- Surafel Mekuria
- Kedir Urgesa
- Fitsum Weldegebreal
journal: International Journal of Microbiology
year: 2023
pmcid: PMC10023229
doi: 10.1155/2023/8961755
license: CC BY 4.0
---
# Bacterial Profile of External Ocular Infections, Its Associated Factors, and Antimicrobial Susceptibility Pattern among Patients Attending Karamara Hospital, Jigjiga, Eastern Ethiopia
## Abstract
### Background
External ocular infection is a global public health problem. Frequently, bacteria cause an ocular infection that ranges from morbidity to loss of vision. The increasing bacterial resistance in ocular infections leads to the risk of treatment failure with possibly serious consequences.
### Objective
The study aimed to assess the bacterial profile of external ocular infections, their associated factors, and antimicrobial susceptibility pattern among patients admitted to Karamara hospital, Jigjiga, Eastern Ethiopia.
### Method
Institutional-basedcross-sectional study was conducted on 288 conveniently selected patients among patients admitted to Karamara hospital from May 1 to June 30, 2020. Data were collected using a structured questionnaire. The ocular sample was collected and cultured in the appropriate culture media and identified using a series of biochemical tests. Antimicrobial susceptibility testing of isolates was performed by using the disk diffusion method. Data were double entered onto EpiData version 3.1 then exported to SPSS version 20 and analyzed to calculate descriptive frequency and odds ratio, and p value ≤0.05 was taken as the significant value.
### Result
The prevalence of bacterial infection in external ocular samples was $62.2\%$ ($95\%$ CI: $56.6\%$, $68.4\%$). Out of the 179 isolates, the majority of the bacterial isolates ($87.7\%$) were Gram-positive. Staphylococcus aureus ($53.1\%$) was the predominant isolate. Using soap for washing the face (AOR = 0.43; $95\%$ CI: 0.29, 0.95), having diabetes mellitus (AOR = 3.11; $95\%$ CI: 1.45, 6.75), and history of hospitalization (AOR = 2.82; $95\%$ CI: 1.44, 5.54) were significantly associated with external ocular infection. Most ($95.5\%$) of the Gram-positive bacteria showed resistance to penicillin, but they were susceptible to vancomycin, clindamycin, and ciprofloxacin.
### Conclusion
The study showed a high prevalence of bacterial infections with the predominant isolate was S. aureus. Penicillin-resistant bacteria were identified among Gram-positive bacterial isolates. Soap usage, hospitalization, and diabetes mellitus were associated with the infection. Antibiotics that were susceptible to the specific bacteria should be used as a drug of choice and using soap for washing the face is advisable to protect against external ocular infection.
## 1. Introduction
Microorganisms are closely associated with external ocular infection. Particularly, infections caused by bacteria are quite common [1]. The most common external ocular infections include conjunctivitis, blepharitis, dacryocystitis, orbital, and periorbital cellulitis. These infections are among the leading causes of ocular morbidity and blindness worldwide, chiefly in developing countries like Ethiopia [2, 3]. Despite considerable resident microbiota, the eye is exposed to an external environment where a range of microorganisms is also inhibited which can cause eye infections opportunistically [4]. Several bacteria play a great role in triggering eye infections and corneal [5, 6]. The common bacterial agents responsible for ocular infections include Gram-positive bacteria such as Staphylococcus aureus, Staphylococcus epidermidis, and several *Streptococcus and* Bacillus spp. as well as Gram-negative bacteria such as Pseudomonas aeruginosa, Moraxella spp., and Haemophilus spp [7]. These organisms may come from the patient's skin, upper respiratory tract, or caught from another person with an external ocular infection [8].
Although antibiotics have been used systemically or topically to control ocular infection, bacterial resistance has been emerging and increasing worldwide for treating ocular infections, more likely due to widespread and inappropriate dosing of broad-spectrum antibiotics for systemic infections, exacerbated by inadequate compliance to full-treatment duration [9, 10].
External ocular infections are affecting and leading to vision loss globally [11]. According to the World Health Organization (WHO), 285 million people were visually impaired worldwide. Out of those, 39 million people were blinded by the year 2010. The report also disclosed that more than $90\%$ of the world's visually impaired people live in developing countries, and surprisingly $82\%$ of the visual impairment, including blindness, was preventable [12]. In Africa, it is estimated that approximately 2.2 million people were blinded due to ocular infection [13]. One report [2015] in Sudan showed that bacterial external ocular infections are significantly prevalent among the pediatrics population and cause more than $65\%$ of morbidity in all cases [14].
A report in 2015 showed that, blindness in Ethiopia reached $1.6\%$. Out of this, $87.4\%$ of blindness was a result of a bacterial pathogen [15]. The morbidity of ocular infections occurs ranging from mild and self-limiting conditions to extremely serious and visually threatening [16]conditions. People who are living in a rural area, children and old aged people are the most affected group compared to others [17]. Several factors such as personal hygiene, living condition, sociodemographic or economic status, ocular trauma, frequency of face washing, the occurrence of systemic disease, and cigarette smoking were considered as associate factors for bacterial-based ocular infections [7, 15].
Most ocular infections in the world have been treated using commonly known antimicrobials. Due to this, microbial resistance to antimicrobial agents has become increasingly prevalent in ocular infections including systemic infections on a global basis [18, 19]. Particularly, in Ethiopia, there is an inordinate habit of using different antibiotics without prescription [9, 10] and there are poor personal hygiene and infection control practices, which lead to increased antimicrobial resistance in the community [20].
In Ethiopia, there are inadequate published resources on this topic. Moreover, in the study area, there is no single study conducted related to this topic. Therefore, this study was designed to determine the bacterial profile of external ocular infections, their associated risk factors, and antimicrobial susceptibility pattern among patients admitted to Karamara hospital Jigjiga, Eastern Ethiopia.
## 2.1. Study Area and Period
The study was conducted in Karamara hospital in Jigjiga from May 1 to June 30, 2020. Jigjiga town is found in the eastern part of Ethiopia, and it is the capital city of the Somali region. It is found 635 km away from Addis Ababa. Karamara hospital renders health services for over seven million people living in all zones and districts of the Somali region. It has high patient flow in the eye clinic.
## 2.2. Study Design and Population
An institutional-basedcross-sectional study was employed. Two hundred eighty-eight [288] patients who visited the eye clinic of Karamara hospital with suspected external ocular infections and fulfilled the inclusion criteria during the study period were included. Patients on antibiotics, anti-inflammatory drugs, and those diagnosed with allergic problems and trachoma were excluded.
## 2.3. Sample Size Determination and Sampling Techniques
The sample size of the study was determined using a single population proportion formula by considering the prevalence of bacterial pathogens among patients with external ocular infection ($21\%$) from the study conducted in Hawassa University Teaching and Referral Hospital, southern Ethiopia [15], with $95\%$ confidence interval (CI), $5\%$ margin of error and $10\%$ nonresponse rate. Then, the final sample size was 288. The study participants were recruited conveniently until we got the required sample size.
## 2.4.1. Physical Examination, Specimen Collection, and Transportation
All patients suspected with external ocular infections were physically examined using a slit-lamp biomicroscope and diagnosed by an ophthalmologist. Specimens from the external part of the eye, such as conjunctiva, eyelid, and lacrimal sac, were taken by an ophthalmologist. Conjunctival specimens were collected using a sterile saline moistened cotton swab, applied by passing the swab gently over the lower and upper conjunctiva 2-times [21]. In cases of dacryocystitis, specimens were taken by puncture and aspiration of the lacrimal sac. An antiseptic was first applied to the area of the puncture, and then the lacrimal sac was punctured in the area below the medial canthal ligament [22]. In the case of blepharitis, discharge from the margin of the eyelid was collected using cotton swabs and placed into a sterile tube. All the swabs were finally immersed in a tube that had 3 ml brain heart infusion (BHI) [23] and transported to the Somali regional microbiology laboratory by using the cold box. After specimen collection, data on sociodemographic and associated factors with external ocular infection were collected by a trained optometrist from each study participant using a pretested structured questionnaire adapted from the previous studies [15, 24].
## 2.5. Bacterial Isolation and Identification
Gram staining was done for differentiating Gram-positive and Gram-negative bacteria and to observe the presence and morphology of cells. Smears were prepared at the collection sites from swabs by gently circularly spreading the specimen on a glass slide [25]. Each specimen was inoculated on a blood agar plate (BAP), chocolate agar plate (CAP), MacConkey agar (MAC), and mannitol salt ager (MSA) culture media with sterile wire loops and incubated at 37°C for 48 hours. Chocolate agar plates were incubated within a candle-jar to facilitate the CO2 atmosphere. After 24 hours of incubation, the plates were observed and examined for bacterial pathogen growth, and plates with no growth were reincubated for further 24 hours [26].
The identification of bacterial pathogens was done initially by Gram stain and colony morphology from culture followed by biochemical tests. Biochemical tests like catalase, coagulase, optochin disk, and bile solubility tests were applied to identify and differentiate Gram-positive cocci, while biochemical tests, such as triple sugar iron agar (TSI), citrate utilization, lysine decarboxylase agar (LDC), oxidase, urease, indole, Methyl Red-Voges-Proskauer (MR-VP), and tributyrin tests were used to identify Gram-negative bacterial isolates [26, 27]. Gram-positive bacteria were identified using hemolytic activity on sheep blood agar, catalase for differentiation of Gram-positive and Gram-negative, coagulase test for S. aureus, bile solubility, and optochin disk test sensitivity for S. pneumonia [26].
## 2.6. Antimicrobial Susceptibility Testing
An antimicrobial susceptibility testing was carried out on each identified bacterium using the disc diffusion method on Mueller Hinton agar (MHA). Besides, MHA medium containing $5\%$ defibrinated sheep blood was used for fastidious bacterial isolate like S. pneumoniae. Primarily, 3–5 bacterial colonies of the test organism were picked and emulsified in 5 ml of nutrient broth and mixed gently. To standardize the density of the inoculum for a susceptibility test, a 0.5 McFarland standard solution was used. The plates were then inoculated by streaking the swab over the entire agar surface then the common antimicrobials were used for patients treated in the *Karamara* general hospital with the following concentrations: ceftriaxone (CRO) (30 μg), ciprofloxacin (CIP) (5 μg), trimethoprim-sulfamethoxazole (SXT) (25 μg), gentamicin (CN) (10 μg), tetracycline (TE) (30 μg), penicillin (P) (10U) and clindamycin (DA) (2 μg), vancomycin (VA) (30 μg), and doxycycline (DO) (30 μg) were placed using sterile forceps on the plate's surface on the MHA plate and incubated at 37°C for 18–24 hours but for S. aureus, it was incubated for only 16–18 hours, and then the zone of inhibition was determined. The zone diameters were measured and recorded. Finally, bacterial susceptibilities were interpreted following the Clinical and Laboratory Standards Institute (CLSI) guidelines as susceptible (S), intermediate (I), or resistant [28].
## 2.7. Quality Control
The reliability of this study was ensured by actualizing quality control (QC) measures all through the entire process of the laboratory procedures. All necessary materials, equipment, and procedures were controlled enough. The questionnaire was prepared in English language and translated to Amharic and the Somali language then retranslated to English to check the consistency. The data were collected by a trained optometrist. Needed specimens were collected following the standard operating procedures (SOPs) that were prepared specifically for external ocular specimen collection. Culture media sterility was ensured by incubating uninoculated media. The prepared culture media performance was checked by inoculating the standard strains, such as *Escherichia coli* (ATCC 25922), *Staphylococcus aureus* (ATCC 25923), and *Pseudomonas aeruginosa* (ATCC 27853) [25, 27] obtained from the Ethiopian Public Health Institute, Addis Ababa, Ethiopia. These strains were also used to check the qualities of biochemical tests. Furthermore, the quality of the data entry was maintained by double data entry.
## 2.8. Data Analysis
The data were cleaned, coded, and double entered using EpiData version 3.1 software and then exported to statistical package for Social Sciences version 20 software for analysis. The descriptive statistics (mean, percentages or frequency) were calculated to summarize the findings. The magnitude of the association between the different variables to the outcome variables was measured by the odds ratio with a $95\%$ confidence interval (CI). Bivariate and multivariable logistic regression analyses were performed to assess the association between dependent and independent variables. Crude odds ratio (COR) and adjusted odds ratio (AOR) at $95\%$ confidence interval was used to measure the strength of association. Those variables with p value <0.2 at bivariate logistic regression were considered for the multivariable logistic regression model to control the confiding variables. Statistical significance was declared at a p value less than 0.05.
## 2.9. Ethical Consideration
Ethical clearance was obtained from the Institutional Health Research Ethics Review Committee (IHRERC) Health and Medical Sciences College, Haramaya University. An official permission letter was written to the Somali regional health office which wrote a permission letter to Karamara hospital. The objective, purpose, risk, and benefits were explained and the signed consent was obtained from the hospital head, the study participants and guardian, or parents of children under 18 years. All the information obtained from the study participants were kept confidential but positive culture result with the possible drug of choice was reported to the ophthalmologist for proper treatment.
## 3.1. Sociodemographic Characteristics
A total of 288 patients were clinically diagnosed with external ocular infections and included in this study with a response rate of $100\%$. About $52.8\%$ of the study participants were males. The mean age was 38.5 (SD ± 16.2) years and $49\%$ of the study participants were of the age between 18 and 39 years. Approximately $71.2\%$ of the participants were from urban and $31.3\%$ were businessmen. One-fourth of the study participants had formal education up to the primary school level (Table 1).
## 3.2. Behavioral and Clinical Related Factors
This result showed that among the participants, 48 ($16.7\%$) had swimming habits, 112 ($38.9\%$) of them frequently washed their faces, 176 ($61\%$) used soap for washing their faces, and 80 ($27.8\%$) used eye cosmetics. Patients who had a previous history of ocular surface disease were 70 ($24.3\%$), hospitalized 78 ($27\%$), ocular trauma 35 ($12\%$), using contact lens 10 ($3.5\%$) making eye surgery 43 ($14.9\%$), and having diabetes were 56 ($19.4\%$) (Table 2).
## 3.3. Bacterial Profile
Among 288 ocular specimens subjected to culture, $62.2\%$ ($\frac{179}{288}$) ($95\%$ CI: $56.6\%$, $68.4\%$) were positive for different bacterial species. Among the 179 isolates, $87.7\%$ were Gram-positive cocci with a predominant isolate of S. aureus ($53.1\%$), followed by Coagulase-negative staphylococcus ($932.4\%$), and *Streptococcus pneumonia* ($2.2\%$). However, $12.3\%$ of isolates were Gram-negative bacteria with predominantly E. coli spp ($6.2\%$). In addition, other species were less frequent such as Klebsiella-pneumonia ($3.3\%$) were isolated only in the case of conjunctivitis, and *Pseudomonas aeruginosa* ($1.7\%$) was isolated in the case of both conjunctivitis and blepharitis (Table 3).
## 3.4. The Magnitude of External Ocular Infection
The prevalence of conjunctivitis was $48.3\%$ ($95\%$ CI: $42.7\%$, $54.5\%$) followed by blepharitis ($29.9\%$) ($95\%$ CI: $24.3\%$, $35.1\%$), blepharoconjunctivitis ($14.6\%$) ($95\%$ CI: $10.4\%$, $18.7\%$), and dacrocystitis $7.3\%$ ($95\%$ CI: $4.5\%$, $10.4\%$). In this study, S. aureus ($55.8\%$ in conjunctivitis, $50\%$ in blepharitis, $40\%$ blepheroconjunctivitis, and $50\%$ dacryocystitis) was the most dominant amongst the other strains.
## 3.5. Associated Factors
In this study, washing with soap, educational status, presence of ocular surface disease, diabetes mellitus, hospitalization, and making surgery showed a significant association in bivariate logistic regression analysis ($p \leq 0.25$) and were considered as a candidate for multivariable logistic regression. In multivariable analysis, washing with soap, hospitalization, and presence of diabetes mellitus was statistically significant in patients who had an external ocular infection at p value less than 0.05.
Patients who used soap for washing their faces were $56.7\%$ (AOR = 0.43; $95\%$ CI: 0.29, 0.95) less likely to be infected with bacterial external ocular infection compared to counterparts. Patients who had a history of hospitalization up to 30 days were 2.8 times (AOR = 2.82; $95\%$ CI: 1.44, 5.54) more likely to develop an external ocular infection compared to their counterparts. Patients who were diabetic were 3.11 times (AOR = 3.11; $95\%$ CI: 1.45, 6.75) more likely to develop the infection than those who did not have diabetes (Table 4).
## 3.6. Antimicrobial Susceptibility Patterns of Gram-Positive Bacterial Isolate
The antimicrobial susceptibility pattern of bacteria was done on nine antimicrobial agents. Out of those, vancomycin, penicillin, and doxycycline were tested for Gram-positive bacteria. Most (150/$95.5\%$) of the Gram-positive showed resistance for penicillin, but they were susceptible to vancomycin (152/$96.8\%$), clindamycin (148/$94.3\%$), ciprofloxacin (145/$92.3\%$), doxycycline (132/$84.1\%$), ceftriaxone (112/$71.3\%$), gentamicin (104/$66.2\%$), tetracycline (88/$56.0\%$), and trimethoprim-sulfamethoxazole (76/$48.4\%$) (Table 5).
## 3.7. Antimicrobial Susceptibility Patterns of Gram-Negative Bacterial Isolate
Over $68\%$ and $63\%$ of Gram-negative bacteria isolates were sensitive to gentamicin and ceftriaxone, respectively. However, $50\%$ and $77.3\%$ were resistant to tetracycline and trimethoprim-sulfamethoxazole, respectively (Table 6).
## 3.8. Multi-Drug Resistance
In this study, the overall multidrug resistance (resistance to two or more antimicrobials) was $87.7\%$. Only $2.2\%$ were sensitive to all tested antimicrobials (Table 7).
## 4. Discussion
External ocular infection is one of the major problems affecting many individuals and is responsible for the increased incidence of morbidity and blindness globally [9, 29]. In the present study, the overall prevalence of bacterial pathogens was $62.2\%$. This is comparable with the reports which were done in Gondar ($58.3\%$) [24], Dessie ($59.4\%$) [15], Gondar ($60.8\%$) [30] in Ethiopia, Sudan ($63.7\%$) [14], and Uganda ($59.5\%$) [31]. On the other hand, it is lower compared to reports from Jimma, Southwest Ethiopia ($74.7\%$) [6], Nigeria ($81.7\%$) [32], and India ($88\%$) [33]. However, it is higher from the study conducted in Hawassa, Ethiopia ($48.8\%$) [34]. This variation might be due to the difference in geographical variation, time variation, study population, and the practice of infection control in the community.
In this study, Gram-positive cocci ($87.7\%$) were the most common isolates. This is supported by previous studies conducted in Gondar ($88\%$) [24], Dessie ($55.6\%$) [15], and Jimma ($52\%$) [6] in Ethiopia, and Nigeria ($50.3\%$) [35]. In the current study, the predominant bacterial isolates were S. aureus ($53.1\%$) similar to other previous studies conducted in Gondar [24], Jimma [6], Nigeria [35], and India [1]. Other studies reported that CoNS was the predominant isolate [15]. The occurrence of different bacteria as an etiological agent for external ocular infection may be due to differences in the environmental conditions [36].
The current study showed a higher prevalence of conjunctivitis ($58\%$) and blepharitis as the next most dominant types of eye infection ($33.5\%$). This is consistent with a study conducted in northwest Ethiopia [15]. Staphylococcus aureus was the most common isolate in conjunctivitis ($55.8\%$), blepharitis ($50\%$), and blepharoconjunctivitis ($40\%$). A similar conclusion was reached by studies conducted in Ethiopia [6] and India [33]. On the other hand, S. aureus was isolated from blepharitis ($47.6\%$) and conjunctivitis ($26.6\%$) as reported from northern Ethiopia [9]. This dominance of S. aureus might be due to contamination of the eye from skin normal flora as a result of touching the eyes with contaminated hands [37].
In the present study, those who used soap were less likely to develop an external ocular infection. It increases personal hygiene, which prevents the growth of bacterial pathogens on the exterior part of the eye, and it is supported by a similar study conducted in France [38]. However, another study reported that there is no significant association between soap usage and external ocular infection [15]. This protective association might be due to the chemical characteristics of soap, which destroys the pathogen from the infection site.
History of hospitalization was significantly associated with external ocular infection. This is consistent with the study conducted in Portugal [39] and in the USA, Central California [40]. The main reason for this significant association is due to the characteristics of the bacteria that cause external ocular infections. These bacteria cause nosocomial infection, that can be acquired during hospitalization [41].
Being diabetes mellitus was significantly associated with ocular infection, this result is supported by several studies in China [42], Denmark [43], England [44], and Iran [3]. This is due to individuals with diabetes mellitus having lower immunity, which may result in loss of control for systemic infections with subsequent spread to ocular tissues [43].
The drug susceptibility patterns of Gram-positive cocci bacterial isolates showed that sensitivity to vancomycin ($96.8\%$) followed by ciprofloxacin ($92.4\%$). This finding agrees with studies conducted in Ethiopia [6] and India [33]. However, most of the isolates were resistant to penicillin and a similar pattern of results was obtained in Jimma [6] and Gondar [24]. This reduction in the effectiveness of penicillin could be due to the frequent usage that results from its low price and accessibility without a prescription.
Most of the Gram-negative isolates were sensitive to gentamicin ($68\%$) followed by ceftriaxone ($63\%$), but they were resistant to trimethoprim-sulphamethoxazole ($78\%$). Several reports also showed similar patterns of drug resistance among Gram-negative bacteria Dessie, [15] Gondar, [20], and India [2]. Moraxella species were $100\%$ sensitive to ciprofloxacin, gentamicin, and tetracycline. This might be due to the few numbers of isolated Moraxella species. Their sensitivity to ciprofloxacin is consistent with the studies conducted in Jimma [6] and Hawassa [45]. However, they showed $100\%$ resistance to trimethoprim-sulphamethoxazole. This might be due to a few isolates of the species.
In this study, most bacterial isolates were resistant to penicillin. This might be due to the usage of those broad-spectrum antimicrobial agents without taking appropriate diagnosis. This result is supported by the study conducted in Jimma [6].
The prevalence of multidrug resistance (MDR) to two or more bacterial isolates to the commonly prescribed antimicrobials was observed in $87.7\%$ of the isolates. This is consistent with what has been found in previous studies conducted in Gondar, northwest Ethiopia [18]. However, a lower prevalence of multidrug resistance was previously reported in Hawassa, south Ethiopia [34]. This may be due to the difference in type and generation of antibiotics that we used for susceptibility testing.
## 5. Conclusion
In this study bacterial external ocular infections are highly prevalent. Conjunctivitis was the dominant external eye infection followed by blepharitis. Gram-positive bacteria constitute more than eighty-five percent of isolates with S. aureus being the most predominant ones. Vancomycin and clindamycin were the drugs of choice for Gram-positive bacterial isolate and gentamicin and ceftriaxone were the drugs for a Gram-negative bacterial isolate. The prevalence of MDR to the commonly prescribed antimicrobials was very high. In this study, soap usage, hospitalization, and diabetes mellitus were statistically significant. Therefore, the community should keep themselves from systemic diseases like DM and practice good personal hygiene to minimize the probability of getting external ocular infections. Antibiotics that have high sensitivity for each bacterial isolate should be used as a drug of choice for patients with external ocular infection. Using soap for washing the face is advisable to protect against external ocular infection. Additionally, Somali regional health offices should give health education to the community to minimize the effect of possible risk factors.
## Data Availability
All the generated data and the analysis developed in this study are included within the article.
## Conflicts of Interest
The authors declare that they have no conflicts of interest.
## Authors' Contributions
All authors made a significant contribution to the work reported such as conception, study design, execution, and acquisition of data, analysis, and interpretation. They took part in drafting and gave final approval to the version to be published. They have agreed on the journal to which the article has been submitted and agreed to be accountable for all aspects of the work.
## References
1. Bharathi M., Ramesh S., Ramakrishnan R., Amuthan M., Viswanathan S.. **Prevalence of bacterial pathogens causing ocular infections in South India**. (2010) **53** p. 281. DOI: 10.4103/0377-4929.64336
2. Sharma S.. **Antibiotic resistance in ocular bacterial pathogens**. (2011) **29** 218-222. DOI: 10.4103/0255-0857.83903
3. Falavarjani K. G., Parvaresh M., Modarres M.. **Isolates and antibiotic resistance of culture-proven endophthalmitis cases presented to a referral center in Tehran**. (2012) **19** p. 361. DOI: 10.4103/0974-9233.102740
4. Linden S. K., Sutton P., Karlsson N. G., Korolik V., McGuckin M. A.. **Mucins in the mucosal barrier to infection**. (2008) **1** 183-197. DOI: 10.1038/mi.2008.5
5. Jung J. S., Lim N. K., Kang D. H.. **Delayed intraorbital infection after craniofacial bone surgery**. (2019) **20** 324-328. DOI: 10.7181/acfs.2019.00360
6. Tesfaye T., Beyene G., Gelaw Y., Bekele S., Saravanan M.. **Bacterial profile and antimicrobial susceptibility pattern of external ocular infections in Jimma University specialized hospital, Southwest Ethiopia**. (2013) **1** 13-20. DOI: 10.12691/ajidm-1-1-3
7. Castañeda-Sánchez J., García-Pérez B., Muñoz-Duarte A.. **Defensin production by human limbo-corneal fibroblasts infected with mycobacteria**. (2013) **2** 13-32. DOI: 10.3390/pathogens2010013
8. Salako A. O., Okesola A. O.. **Microbiological profile of bacterial conjunctivitis in Ibadan, Nigeria**. (2010) **8** 20-24. DOI: 10.4314/aipm.v8i1.63953
9. Teweldemedhin M., Gebreyesus H., Atsbaha A. H., Asgedom S. W., Saravanan M.. **Bacterial profile of ocular infections: a systematic review**. (2017) **17** p. 212. DOI: 10.1186/s12886-017-0612-2
10. Anagaw B., Biadglegne F., Belyhun Y., Anagaw B., Mulu A.. **Bacteriology of ocular infections and antibiotic susceptibility pattern in Gondar University Hospital, northwest Ethiopia**. (2011) **49** 117-123. PMID: 21796911
11. Sthapit P. R., Tuladhar N. R.. **Conjunctival flora of normal human eye**. (2014) **2** p. 1021
12. Bn N., kshmi M., Rao A.. **Aerobic bacteriological profile in cases of ocular infections in a tertiary care hospital (navodaya medical College and research centre, raichur)**. (2014) **13** 14-21. DOI: 10.9790/0853-131141421
13. Lewallen S., Courtright P.. **Blindness in Africa: present situation and future needs**. (2001) **85** 897-903. DOI: 10.1136/bjo.85.8.897
14. Mazin O. M., Lemya A. K., Samah O. M.. **External ocular bacterial infections among Sudanese children at Khartoum State, Sudan**. (2016) **10** 1694-1702. DOI: 10.5897/ajmr2016.8092
15. Shiferaw B., Gelaw B., Assefa A., Assefa Y., Addis Z.. **Bacterial isolates and their antimicrobial susceptibility pattern among patients with external ocular infections at Borumeda hospital, Northeast Ethiopia**. (2015) **15** p. 103. DOI: 10.1186/s12886-015-0078-z
16. Chaudhary M., Bhattarai A., Adhikari S., Bhatta D.. **Bacteriology and antimicrobial susceptibility of adult chronic dacryocystitis**. (1970) **2** 105-113. DOI: 10.3126/nepjoph.v2i2.3716
17. Sthapit P. R., Tuladhar N. R., Marasini S., Khoju U., Thapa G.. **Bacterial conjunctivitis and use of antibiotics in dhulikhel hospital-kathmandu university hospital**. (2012) **9** 69-72. DOI: 10.3126/kumj.v9i2.6292
18. Muluye D., Wondimeneh Y., Moges F., Nega T., Ferede G.. **Types and drug susceptibility patterns of bacterial isolates from eye discharge samples at Gondar University Hospital, Northwest Ethiopia**. (2014) **7** p. 292. DOI: 10.1186/1756-0500-7-292
19. Cavuoto K., Zutshi D., Karp C. L., Miller D., Feuer W.. **Update on bacterial conjunctivitis in South Florida**. (2008) **115** 51-56. DOI: 10.1016/j.ophtha.2007.03.076
20. Muhie O. A.. **Antibiotic use and resistance pattern in Ethiopia: systematic review and meta-analysis**. (2019) **2019** 8. DOI: 10.1155/2019/2489063
21. Choudhury R., Panda S., Sharma S., Singh D. V.. (2012)
22. Mah F. S., Davidson R., Holland E. J.. **Current knowledge about and recommendations for ocular methicillin-resistant**. (2014) **40** 1894-1908. DOI: 10.1016/j.jcrs.2014.09.023
23. Quinn C., Mathews D., Noyes R., Oliver G., Thimons J., Thomas R.. (2010)
24. Getahun E., Gelaw B., Assefa A., Assefa Y., Amsalu A.. **Bacterial pathogens associated with external ocular infections alongside eminent proportion of multidrug resistant isolates at the University of Gondar Hospital, northwest Ethiopia**. (2017) **17** p. 151. DOI: 10.1186/s12886-017-0548-6
25. Vandepitte J., Verhaegen J., Engbaek K., Rohner P., Piot P., Heuck C.. (2003)
26. Sharma S. J. E.. **Diagnosis of infectious diseases of the eye**. (2012) **26** 177-184. DOI: 10.1038/eye.2011.275
27. Pinar-Sueiro S., Sota M., Lerchundi T.-X.. **Dacryocystitis: systematic approach to diagnosis and therapy**. (2012) **14** 137-146. DOI: 10.1007/s11908-012-0238-8
28. 28Clsi
Performance Standards for Antimicrobial Susceptibility Testing
202030edWayne, PA, USAClinical and Laboratory Standards Institute Wayne. (2020)
29. Whitcher J. P., Srinivasan M., Upadhyay M. P.. **Corneal blindness: a global perspective**. (2001) **79** 214-221. PMID: 11285665
30. Assefa Y., Moges F., Endris M.. **Bacteriological profile and drug susceptibility patterns in dacryocystitis patients attending Gondar University Teaching Hospital, Northwest Ethiopia**. (2015) **15** p. 34. DOI: 10.1186/s12886-015-0016-0
31. Mshangila B., Paddy M., Kajumbula H., Ateenyi-Agaba C., Kahwa B., Seni J.. **External ocular surface bacterial isolates and their antimicrobial susceptibility patterns among pre-operative cataract patients at Mulago National Hospital in Kampala, Uganda**. (2013) **13** p. 71. DOI: 10.1186/1471-2415-13-71
32. Onyedikach A. P., Innocent N. E., Emmanuel E. I.. **Prevalence, biofilm formation and anti-microbial susceptibility testing of bacterial and fungal etiological agents associated with external ocular infections in Nigeria**. (2019) **12** 677-681. DOI: 10.3923/ajbs.2019.677.681
33. Bharathi M. J., Ramakrishnan R., Shivakumar C., Meenakshi R., Lionalraj D.. **Etiology and antibacterial susceptibility pattern of community-acquired bacterial ocular infections in a tertiary eye care hospital in south India**. (2010) **58** p. 497. DOI: 10.4103/0301-4738.71678
34. Anteneh A., Tamirat A., Adane M., Demoze D., Endale T.. **Potential bacterial pathogens of external ocular infections and their antibiotic susceptibility pattern at Hawassa University teaching and referral Hospital, Southern Ethiopia**. (2015) **9** 1012-1019. DOI: 10.5897/ajmr2014.7282
35. Iwalokun B. A., Oluwadun A., Akinsinde K. A., Niemogha M. T., Nwaokorie F. O.. **Bacteriologic and plasmid analysis of etiologic agents of conjunctivitis in Lagos, Nigeria**. (2011) **1** 95-103. DOI: 10.1007/s12348-011-0024-z
36. Wolfgang M. C., Kulasekara B. R., Liang X.. **Conservation of genome content and virulence determinants among clinical and environmental isolates of**. (2003) **100** 8484-8489. DOI: 10.1073/pnas.0832438100
37. Sato K.-i. **External ocular infections due to methicillin-resistant**. (2015) **50** e97-e99. DOI: 10.1016/j.jcjo.2015.05.006
38. Sauer A., Meyer N., Bourcier T.. **Risk factors for contact lens–related microbial keratitis: a case–control multicenter study**. (2016) **42** 158-162. DOI: 10.1097/icl.0000000000000180
39. Dias C., Gonçalves M., João A.. **Epidemiological study of hospital-acquired bacterial conjunctivitis in a level III neonatal unit**. (2013) **2013** 5. DOI: 10.1155/2013/163582
40. Walvick M. D., Khan A.. (2013)
41. Klein E., Smith D. L., Laxminarayan R.. **Hospitalizations and deaths caused by methicillin-resistant**. (2007) **13** 1840-1846. DOI: 10.3201/eid1312.070629
42. Wong J.-S., Chan T.-K., Lee H.-M., Chee S.-P.. **Endogenous bacterial endophthalmitis: an east Asian experience and a reappraisal of a severe ocular affliction**. (2000) **107** 1483-1491. DOI: 10.1016/s0161-6420(00)00216-5
43. Kruse A., Thomsen R. W., Hundborg H. H., Knudsen L. L., Sørensen H. T., Schonheyder H. C.. **Diabetes and risk of acute infectious conjunctivitis—a population‐based case‐control study 1**. (2006) **23** 393-397. DOI: 10.1111/j.1464-5491.2006.01812.x
44. Ansari A. S., de Lusignan S., Hinton W., Munro N., McGovern A.. **The association between diabetes, level of glycaemic control and eye infection: cohort database study**. (2017) **11** 421-429. DOI: 10.1016/j.pcd.2017.05.009
45. Aweke T., Dibaba G., Ashenafi K., Kebede M.. **Bacterial pathogens of exterior ocular infections and their antibiotic vulnerability pattern in southern Ethiopia**. (2014) **1** 19-25
|
---
title: Identification of Potentially Functional Circular RNA/Long Noncoding RNA-MicroRNA-mRNA
Regulatory Networks Associated with Vascular Injury in Type 2 Diabetes Mellitus
by Integrated Microarray Analysis
authors:
- Yi Leng
- Ming-zhu Wang
- Kang-ling Xie
- Ying Cai
journal: Journal of Diabetes Research
year: 2023
pmcid: PMC10023230
doi: 10.1155/2023/3720602
license: CC BY 4.0
---
# Identification of Potentially Functional Circular RNA/Long Noncoding RNA-MicroRNA-mRNA Regulatory Networks Associated with Vascular Injury in Type 2 Diabetes Mellitus by Integrated Microarray Analysis
## Abstract
This research is aimed at figuring out the potential circular RNA (circRNA)/long noncoding RNA- (lncRNA-) microRNA- (miRNA-) mRNA regulatory networks associated with a vascular injury in type 2 diabetes mellitus (T2DM). Differentially expressed genes (DEGs) screened in T2DM-related expression datasets were intersected with genes associated with vascular injury in T2DM to obtain candidate DEGs, followed by the construction of an interaction network of DEGs. The upstream miRNAs of candidate genes were predicted by mirDIP, miRWalk, and DIANA TOOLS databases, and the upstream lncRNAs/circRNAs of miRNAs by DIANA-LncBase/circBank database, followed by the construction of circRNA/lncRNA-miRNA-mRNA regulatory networks. Peripheral blood was attained from T2DM patients with macroangiopathy for clinical validation of expression and correlation of key factors. Differential analysis screened 37 candidate DEGs correlated with vascular injury in T2DM. Besides, MAPK3 was a core gene associated with vascular injury in T2DM. Among the predicted upstream miRNAs of MAPK3, miR-4270, miR-92a-2-5p, miR-423-5p, and miR-613 ranked at the top according to binding scores. The upstream lncRNAs and circRNAs of the 4 miRNAs were further predicted, obtaining 11 candidate lncRNAs and 3 candidate circRNAs. Moreover, KCNQ1OT1, circ_0020316, and MAPK3 were upregulated, but miR-92a-2-5p was downregulated in the peripheral blood of T2DM patients with macroangiopathy. Mechanistically, KCNQ1OT1 and circ_0020316 bound to miR-92a-2-5p that inversely targeted MAPK3. Collectively, KCNQ1OT1/circ_0020316-miR-92a-2-5p-MAPK3 coexpression regulatory networks might promote vascular injury in T2DM.
## 1. Introduction
Type 2 diabetes mellitus (T2DM) accounts for more than $90\%$ of patients with diabetes and is featured with relative insulin deficiency resulting from pancreatic β-cell dysfunction and insulin resistance in target organs [1]. T2DM is a multifactorial metabolic disease that is attributable to the interplay between multiple environmental and genetic predispositions [2]. T2DM results in microvascular and macrovascular complications, which leads to severe psychological and physical distress to both patients and careers and imposes a massive burden on the health care system [3]. Besides, T2DM can adversely impact the microvasculature in various organs, and therefore, T2DM may develop microvascular injury/dysfunction as a chronic outcome [4]. Therefore, the vascular injury may assume a role in T2DM. This background calls for the necessary exploration of the molecular mechanism behind vascular injury in T2DM.
Circular RNA (circRNA) is a novel kind of endogenous noncoding RNA with tissue-specific and cell-specific expression patterns. It has been demonstrated to be involved in multiple diseases like neurological disorders, cancer, diabetes mellitus, and cardiovascular diseases [5]. It was previously reported that hsa_circ_0054633 could be utilized as a diagnostic biomarker of T2DM [6]. Another report elucidated that circANKRD36 was linked to inflammation in T2DM patients [7]. As transcripts of more than 200 nucleotides that are not translated into proteins, long noncoding RNAs (lncRNAs) have been documented to participate in the physiology of tissues and organs and disease processes by orchestrating numerous cell processes, including cell division, differentiation, survival, and senescence [8, 9]. Recently, lncRNA MALAT1 was found to reduce insulin resistance in T2DM [10]. Additionally, lncRNA MEG3 could assume a crucial role in T2DM-induced vascular disease [11]. Therefore, these findings illustrated that circRNAs and lncRNAs might function as novel potential biomarkers for T2DM.
Recently, great attention has been paid to the competing endogenous RNA (ceRNA) regulatory network that lncRNAs/circRNAs act as a sponge for microRNAs (miRNAs/miRs) to indirectly upregulate miRNA downstream target genes, thus modulating disease progression [12, 13]. For instance, mmu_circ_0000250 enhanced wound healing in diabetes by upregulating SIRT1 through binding to miR-128-3p [14]. In addition, LINC-P21 could repress insulin secretion and proliferation of pancreatic β-cells in T2DM through binding to miR-766-3p to upregulate NR3C2 [15].
With the rapid development of sequencing and large sample analysis, bioinformatics analysis has been widely utilized in biological research and therapeutic progress [16]. It has been widely used to analyze public databases to construct a more comprehensive lncRNA/circRNA-miRNA-mRNA regulatory network and mine more accurate prognostic markers [17]. Moreover, there exist few researches to predict the prognostic lncRNA-miRNA-mRNA [18] and circRNA/lncRNA-miRNA-mRNA [19] networks in T2DM. Gene Expression Omnibus (GEO) database can provide circRNA, miRNA, and mRNA of numerous diseases, which can be applied in data mining and biological discovery [20].
In our study, we aimed to figure out the potential circRNA/lncRNA-miRNA-mRNA regulatory networks associated with vascular injury in T2DM. Our study provided a theoretical basis for further understanding the pathogenesis of T2DM vascular injury and searching for potential diagnostic and therapeutic targets.
## 2.1. Ethics Statement
Written informed consents were acquired from all participants prior to enrollment. Study protocols were ratified by the Ethics Committees of our hospital and strictly adhered to the Declaration of Helsinki.
## 2.2. Acquisition and Analysis of Data in Expression Datasets from GEO Database
The T2DM-related expression datasets GSE15653 and GSE21340 were attained from GEO database. The GSE15653 dataset encompasses 9 liver biopsy tissue samples from T2DM obese patients and 5 obese without T2DM samples. The GSE21340 dataset consists of 15 plasma samples from T2DM patients and 5 normal control samples.
Differentially expressed genes (DEGs) in these 2 datasets were identified using the R language “limma” package with a threshold of p value < 0.05. Volcano plots were drawn using the R language “ggplot2” package. Meanwhile, the correlation analysis of mRNA expression of candidate genes was performed using the R language “corrplot” package.
## 2.3. Retrieval of Disease-Related Database
T2DM-related target genes were researched through databases of GeneCards (relevance score ≥ 20) and Comparative Toxicogenomics Database (CTD; inference score ≥ 30).
## 2.4. Functional Enrichment Analysis of Candidate Genes
Venn analysis of the result of microarray analysis and databases of GeneCards and CTD was conducted using the draw Venn diagram tool to identify candidate genes. Gene Ontology (GO) and Kyoto Encyclopedia at Genes and Genomes (KEGG) enrichment analyses were performed on candidate genes using the R language “ClusterProfiler” package to analyze the cellular functions and signaling pathways that were mainly impacted by potential targets and key targets. $p \leq 0.05$ was considered statistically significant.
## 2.5. Construction of Protein Interaction Network of Candidate Genes
The interaction network of target genes was obtained through STRING with species condition limited to “Homo sapiens” to construct the regulatory relationship network. Minimum required interaction score was set as 0.4. Active interaction sources, including text mining, experiments, databases, coexpression, neighborhood, gene fusion, and cooccurrence, were used to construct a network of regulatory relationships. Then, the PPI enrichment value less than 1.0e-16 in enrichment network was selected to import into Cytoscape (v3.8.2) software. The network relationship plots were subjected to result analysis and ranking. The degree value and combine score value were indicated by colors, and candidate genes were ranked based on the degree value.
## 2.6. Prediction of Upstream miRNA-lncRNA/circRNA of Candidate Genes
The upstream miRNAs of candidate target genes were predicted using mirDIP, miRWALK, and DIANA TOOLS databases, which were intersected by the draw Venn diagram tool. The upstream lncRNAs of miRNAs were predicted by DIANA-LncBase database, and the upstream circRNAs of miRNAs were predicted by circBank database. Afterwards, the lncRNA/circRNA-miRNA-mRNA coexpression regulatory network was constructed, which was visualized using Cytoscape (v3.8.2).
## 2.7. Clinic Sample Collection
A total of 50 T2DM patients with macroangiopathy (36 males and 14 females; mean age: 54.5 ± 7.1 years) treated in our hospital from January 2020 to January 2021 were enrolled as the case group. In addition, 30 normal healthy individuals (15 males and 15 females; mean age: 55.2 ± 4.9 years) who underwent physical examination during the same period were enrolled as the control group. Fasting peripheral blood (5 mL) was harvested from all participants during admission or the early morning of the physical examination day. All patients were diagnosed for the first time and had no history of long-term medication, cancer, or chronic diseases.
## 2.8. Reverse Transcription Quantitative Polymerase Chain Reaction (RT-qPCR)
Total RNA was extracted from peripheral blood of healthy controls and T2DM patients with macroangiopathy using a TRIzol Kit (Invitrogen, Carlsbad, California, USA) and reversely transcribed to cDNA using as per the manuals of TaqMan MicroRNA Assays Reverse Transcription Primer (4427975, Applied Biosystems, Carlsbad, CA, USA). lncRNA, circRNA, and mRNA expressions were normalized by glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and miRNA expression was normalized by U6. Relative differences in target gene expression were calculated using a 2-ΔΔCT. Primers are manifested in Supplementary Table 1 (primer design using NCBI's primer design function).
## 2.9. Human Vascular Smooth Muscle Cell (HVSMC) Culture
HVSMC (bio-73393, Biobw, Beijing, China) were cultured with Ham's F-12K medium (PM150910, Biobw) encompassing 0.05 mg/mL vit. C+0.01 mg/mL insulin+0.01 mg/mL transferrin+10 ng/mL sodium selenite+0.03 mg/mL endothelial cell growth supplement+$10\%$ fetal bovine serum+10 mM 4-(2-hydroxyethyl)-1-piperazineëthanesulfonic acid+10 mM TES+$1\%$ P/S in a 37°C with $5\%$ CO2 and saturated humidity. The cells were passaged by trypsinization when they reached more than $90\%$ confluence, and the cells in the exponential growth phase were used for follow-up experiments.
## 2.10. Dual Luciferase Gene Reporter Assay
The 3′ untranslated region (UTR) of the mitogen-activated protein kinase 3 (MAPK3) gene was clonally amplified, and the PCR product was cloned into the multicloning site at downstream of the luciferase gene of pmirGLO (E1330, Promega, Madison, WI, USA) luciferase gene, named as pMAPK3 wild type (WT). The pMAPK3 mutant type (MUT) vector was constructed by site-directed mutagenesis of the binding sites for bioinformatically predicted miR-92a-2-5p to target genes. With pRL-TK vector expressing Renilla luciferase (E2241, Promega) as an internal reference, miR-92a-2-5p mimic and negative control were cotransfected with luciferase reporter vector, respectively, into HVSMCs, and luciferase activity was detected according to the method provided by Promega.
## 2.11. Statistical Analysis
Statistical analyses were performed using SPSS21.0 software (IBM Corp. Armonk, NY, USA). Measurement data were described as mean ± standard deviation. Data between the two groups were compared using an unpaired t-test. Correlation analysis was performed by the Pearson method. A value of $p \leq 0.05$ was regarded as statistically significant.
## 3.1. There Were 37 Candidate Genes Associated with Vascular Injury in T2DM
We screened out the key factors associated with vascular injury in T2DM by GEO database. Then, circRNA/lncRNA-miRNA-mRNA coexpression networks were constructed, followed by further exploration of the biological functions of the circRNA/lncRNA-miRNA-mRNA coexpression networks in the occurrence of vascular injury in T2DM. The flow of bioinformatics screening of key genes involved in the development of vascular injury in T2DM is detailed in Figure 1.
First, 227 DEGs were obtained from the T2DM-associated dataset GSE21340, of which 110 DEGs were upregulated and 117 DEGs were downregulated (Figure 2(a)). Then, 598 DEGs were identified in the T2DM-associated GSE15653 dataset, among which 373 DEGs were upregulated and 225 were downregulated (Figure 2(b)). The T2DM-associated genes were retrieved through GeneCards database, and 1470 genes were filtered using relevance score ≥ 20, whereas 2677 genes were filtered from the CTD database using inference score ≥ 30. The DEGs from GSE21340 and GSE15653 were intersected with the results of GeneCards and CTD databases, which filtered out 61 candidate DEGs (Figure 2(c)).
Further vascular injury-related genes were retrieved from GeneCards database, and 1137 genes were filtered by relevance score ≥ 5, which was then intersected with the above 61 candidate DEGs. A total of 37 candidate DEGs related to vascular injury were harvested (Figure 2(d)). Subsequently, the 37 candidate DEGs were imported into the STRING database with the corresponding filtering conditions to obtain protein interaction relationships which were then imported into Cytoscape software to construct the protein-protein interaction (PPI) network. There existed 37 nodes and 116 edges in the PPI network relationship graph (PPI enrichment p value < 1.0e-16): a larger shape indicated a larger degree value, and the greater the degree value and combine score value corresponded to the color from yellow to purple (Figure 2(e)). In the network, the top 5 genes by degree value were albumin (Alb), MAPK3, MAPK1, Hras, and Nras, respectively (Figure 2(f)).
## 3.2. MAPK3 Might Be a Core Gene in the Development of Vascular Injury in T2DM
These 37 candidate DEGs further underwent GO and KEGG enrichment analyses. The GO functional analysis results indicated that in terms of biological processes (BP), DEGs were mainly enriched in positive regulation of gene expression, response to hypoxia, regulation of stress-activated MAPK cascade, and inflammatory response; in terms of cellular component (CC), DEGs were mainly enriched in protein binding, MAP kinase activity, scaffold protein binding, protein-containing complex binding, and MAP kinase activity; in terms of molecular function (MF), DEGs were mainly enriched in protein-containing complex, Golgi apparatus, early endosome, and external side of plasma membrane (Figure 3(a)). As reflected by KEGG pathway analysis results, DEGs were mainly enriched in vascular endothelial growth factor signaling pathway, apoptosis, phosphoinositide 3-kinase-Akt signaling pathway, MAPK signaling pathway, and ErbB signaling pathway (Figure 3(b)). These results suggested that DEGs mainly assumed a role in MAPK cascade, inflammatory response, and stress activation and were enriched in structures such as mitochondria, extracellular space, and cytoplasm. In addition, the MF of DEGs mainly manipulated cell-related enzyme activities and protein binding.
In the progression of T2DM, vascular system disease is one of the critical complications, and the mechanism of vascular injury is associated with the pathogenesis of insulin resistance, diabetic nephropathy, and peripheral arterial disease [21]. When T2DM patients experience hypoglycemia, it promotes the release of inflammatory factors, increases the aggregation of platelets, alters hemodynamics, damages vascular endothelial cells, and enhances the incidence of cardiovascular system disease [22]. It has been documented that inhibition of the MAPK3 signaling pathway decreases VSMC proliferation and migration to alleviate vascular neointimal hyperplasia induced by vascular injury in rats [23]. In addition, GSE15653 analysis manifested that MAPK3 was remarkably highly expressed in T2DM (Figures 3(c) and 3(d)). Therefore, we speculated that MAPK3 might be the core gene of T2DM-induced vascular injury.
## 3.3. Screening Upstream miRNAs and circRNAs/lncRNAs of MAPK3 and Coexpression Network Construction
In the mechanism of vascular injury development in T2DM, miRNAs, lncRNAs, and circRNAs are involved in mediating vascular endothelial cell injury and repair capacity [24, 25]. The upstream miRNAs of MAPK3 were further predicted by 3 common online databases (mirDIP, miRWALK, and DIANA TOOLS) with the species as human. In total, 88 miRNAs were predicted by mirDIP database, 721 miRNAs were predicted by miRWALK database, and 35 miRNAs were predicted by DIANA TOOLS database (score > 0.5), the intersection of which was obtained to screen out 8 miRNAs (Figure 4(a)). The scores of target binding to MAPK3 for the 8 miRNAs screened are depicted in Supplementary Table 2, among which miR-4270, miR-92a-2-5p, miR-423-5p, and miR-613 ranked at the top and all participated in T2DM or vascular injury [26–29].
The upstream lncRNAs of miR-4270, miR-92a-2-5p, miR-423-5p, and miR-613 were further predicted through DIANA-LncBase database, obtaining 1109, 1014, 714, and 505 lncRNAs, respectively. Moreover, 14 intersected lncRNAs were screened by *Venn analysis* (Figure 4(b)). The lncRNA data that could not be utilized for subsequent experiments were excluded by querying the NCBI website, which obtained 11 lncRNAs, and the specific binding relationships are displayed in Supplementary Table 3. The network regulation diagram was drawn by Cytoscape software, and the lncRNA-miRNA-MAPK3 coexpression regulatory network was constructed (Figure 4(c)). Furthermore, TSIX, KCNQ1OT1, and LOC101926935 were upregulated in T2DM of GSE20966 (Figures 4(d) and 4(e)).
Further, the circBank database was adopted to predict the upstream circRNAs of miR-4270, miR-92a-2-5p, miR-423-5p, and miR-613, and the top 300 targeted circRNAs were, respectively, subjected to *Venn analysis* to yield three candidate circRNAs (circ_0020316, circ_0091807, and circ_0091808) in the intersection (Figure 5(a)). The specific binding relationships are manifested in Supplementary Table 4. Then, Cytoscape software was applied to draw a network regulatory relationship diagram and construct a circRNA-miRNA-MAPK3 regulatory network (Figure 5(b)).
## 3.4. KCNQ1OT1, circ_0020316, and MAPK3 Expressions Were High, but miR-92a-2-5p Expression Was Low in Peripheral Blood of T2DM Patients with Macroangiopathy
To identify the mechanism of MAPK3 and its upstream miRNAs, lncRNAs, and circRNAs in vascular injury of T2DM, peripheral blood samples were harvested from 50 T2DM patients with macroangiopathy and 30 normal healthy controls. RT-qPCR revealed that compared with healthy control, MAPK3 mRNA expression was obviously augmented in the peripheral blood of T2DM patients (Figure 6(a)), whereas miR-92a-2-5p expression was potently diminished (Figure 6(b)) and negatively correlated with MAPK3 expression (Figure 6(c)). In addition, KCNQ1OT1 and circ_0020316 expressions was noticeably enhanced in the peripheral blood of T2DM patients compared with healthy controls and exhibited positive correlations with MAPK3 expression (Figures 6(d)–6(g)). Conclusively, KCNQ1OT1/circ_0020316-miR-92a-2-5p-MAPK3 coexpression regulatory networks may be a key molecular pathway involved in the development of vascular injury in T2DM.
## 3.5. KCNQ1OT1 and circ_0020316 Bound to miR-92a-2-5p to Upregulate MAPK3
Dual luciferase gene reporter assay results manifested that miR-92a-2-5p mimic reduced luciferase activity of MAPK3-WT but did not impact that of MAPK3-MUT in HEK293 cells (Figure 7(a)), indicating that MAPK3 was negatively targeted by miR-92a-2-5p.
Moreover, the luciferase activities of KCNQ1OT1-WT and circ_0020316-WT were diminished by miR-92a-2-5p mimic, while there was no obvious difference in luciferase activities of KCNQ1OT1-MUT and circ_0020316-MUT (Figures 7(b) and 7(c)). Therefore, both KCNQ1OT1 and circ_0020316 bound to miR-92a-2-5p.
## 4. Discussion
T2DM is a prevalent disease resulting in major neurologic, renal, and vascular complications, and it is of significance to prevent and treat T2DM and its complications [30]. The clinical management of T2DM is through a healthy diet and lifestyle combined with glucose-lowering agents aimed at preventing or delaying the acute symptoms of hyperglycemia and complications of the disease [31]. In spite of the therapeutic benefits of glucose-lowering agents for the treatment of T2DM, most of the drugs can contribute to some undesirable side effects [32]. Therefore, it is imperative to deepen the understanding of biomarkers for the diagnosis and treatment of T2DM. Considering this, this research was conducted through bioinformatics analysis and experiments on peripheral blood from T2DM patients and revealed that KCNQ1OT1/circ_0020316-miR-92a-2-5p-MAPK3 regulatory networks might promote vascular injury in T2DM.
Initially, we screened out 37 candidate DEGs associated with vascular damage in T2DM through differential analysis of T2DM-associated datasets GSE15653 and GSE21340, among which Alb, MAPK3, MAPK1, Hras, and Nras were the top five genes ranked by the degree value in the PPI network. A prior research elucidated that dynamic change of serum Alb level was correlated with T2DM risk [33]. Also, it was noted in another research that MAPK3 and MAPK1 were involved in heart failure caused by diabetes [34]. Hras has been demonstrated to accelerate glucose-induced apoptosis of retinal capillary cells in diabetes [35]. Besides, diabetes could elevate Nras expression during rat oral oncogenesis [36]. Moreover, combined with the results of PPI network analysis, GO analysis, and KEGG analysis, the present research indicated that MAPK3 might be a core gene associated with vascular injury in T2DM. Additionally, RT-qPCR revealed that MAPK3 was upregulated in peripheral blood from T2DM patients. Consistently, it was predicted by the research of Du and Uversky that MAPK3 was a moderately disordered protein in T2DM [37]. Moreover, integrated microarray analysis in the research of Li et al. manifested predicted MAPK3 as the hub gene in circMYO9B/circGRAMD1B/circTHAP4/circTMC7-miRNA-mRNA regulatory network in T2DM [38]. Of note, another report predicted that MAPK3 might be involved in antidiabetic and antihyperlipidemic effects of Gegen in T2DM [39]. Also, it was previously reported that MAPK3A was differentially expressed in atherosclerosis, a common chronic vascular inflammatory disease, which suggested the potential role of MAPK3 in vascular injury [40].
It has been generally accepted that lncRNAs/circRNAs may bind to miRNAs through their miRNA response elements, thus orchestrating the expression of target genes of miRNAs [41]. Based on this, we further predicted the circRNA/lncRNA-miRNA-MAPK3 network involved in vascular injury in T2DM, which was further screened by clinic sample experiments and dual luciferase gene reporter assay. Our data documented that KCNQ1OT1/circ_0020316-miR-92a-2-5p-MAPK3 regulatory networks were involved in vascular injury in T2DM. Corroborating findings were reported in a prior work that miR-92a was poorly expressed in animal models of diabetes [42]. Moreover, mmu-miR-92a-3p upregulation was involved in the rescue of diabetes-impaired angiogenesis by reconstituted high-density lipoproteins, which suggested its implications for the treatment of diabetes-related vascular complications [43]. Of note, the involvement of KCNQ1OT1 has been displayed in T2DM susceptibility [44]. Yang et al. observed KCNQ1OT1 upregulation in patients with diabetes, high glucose-induced cardiomyocytes, and diabetic mouse cardiac tissue, and silencing KCNQ1OT1 inhibits diabetic cardiomyopathy [45]. In addition, a prior study uncovered that KCNQ1OT1 was overexpressed in diabetic nephropathy and that its silencing depressed proliferation and fibrosis and elevated apoptosis in diabetic nephropathy cells [46]. circRNAs have been implicated in vascular diseases, including vascular dysfunction, diabetes mellitus-related retinal vascular dysfunction, and hepatic vascular invasion in T2DM [47]. For instance, cZNF532 upregulation alleviated human diabetes-induced retinal pericyte degeneration and vascular dysfunction [48]. circHIPK3 silencing contributed to the enhancement of retinal endothelial cell proliferation, migration, and tube formation, thus depressing retinal vascular dysfunction in diabetes [49]. Of note, our research predicted the involvement of a novel circRNA, circ_0020316, in the circRNA-miRNA-MAPK3 network in vascular injury in T2DM. Furthermore, clinic sample detection showed that circ_0020316 and KCNQ1OT1 expressions were high in the peripheral bloods of T2DM patients, and dual luciferase gene reporter assay documented that circ_0020316 and KCNQ1OT1 bound to miR-92a-2-5p.
## 5. Conclusion
Conclusively, our data revealed the involvement of KCNQ1OT1/circ_0020316-miR-92a-2-5p-MAPK3 regulatory networks in T2DM-induced vascular injury (Figure 8). Our results provide new insight for mechanistic investigations and may offer potential therapeutic targets for T2DM. Future studies are needed to better understand the role of these two regulatory networks in vascular injury in T2DM.
## Data Availability
The datasets generated and/or analyzed during the current study are available in the manuscript and supplementary materials.
## Ethical Approval
Study protocols were ratified by the Ethics Committees of our hospital and strictly adhered to the Declaration of Helsinki.
## Consent
Written informed consents were acquired from all participants prior to enrollment.
## Conflicts of Interest
The authors declare no competing interest.
## Authors' Contributions
Ying Cai designed the study. Yi Leng and Ming-zhu Wang were involved in the clinical data collection. Ying Cai, Kang-ling Xie, and Yi Leng performed GEO microarray data acquisition and analysis. Ying Cai, Yi Leng, and Ming-zhu Wang completed the experiments such as HVSMC culture, RT-PCR, and dual luciferase gene reporter assay. Ying Cai and Yi Leng performed the statistical analysis, preparation of figures, and drafting of the paper. All authors read and approved the final manuscript.
## References
1. Zheng Y., Ley S. H., Hu F. B.. **Global aetiology and epidemiology of type 2 diabetes mellitus and its complications**. (2018) **14** 88-98. DOI: 10.1038/nrendo.2017.151
2. Javeed N., Matveyenko A. V.. **Circadian etiology of type 2 diabetes mellitus**. (2018) **33** 138-150. DOI: 10.1152/physiol.00003.2018
3. Chatterjee S., Khunti K., Davies M. J.. **Type 2 diabetes**. (2017) **389** 2239-2251. DOI: 10.1016/S0140-6736(17)30058-2
4. Barrett E. J., Liu Z., Khamaisi M.. **Diabetic microvascular disease: an endocrine society scientific statement**. (2017) **102** 4343-4410. DOI: 10.1210/jc.2017-01922
5. Kristensen L. S., Andersen M. S., Stagsted L. V. W., Ebbesen K. K., Hansen T. B., Kjems J.. **The biogenesis, biology and characterization of circular RNAs**. (2019) **20** 675-691. DOI: 10.1038/s41576-019-0158-7
6. Zhao Z., Li X., Jian D., Hao P., Rao L., Li M.. **Hsa_circ_0054633 in peripheral blood can be used as a diagnostic biomarker of pre-diabetes and type 2 diabetes mellitus**. (2017) **54** 237-245. DOI: 10.1007/s00592-016-0943-0
7. Fang Y., Wang X., Li W.. **Screening of circular RNAs and validation of circANKRD36 associated with inflammation in patients with type 2 diabetes mellitus**. (2018) **42** 1865-1874. DOI: 10.3892/ijmm.2018.3783
8. Noh J. H., Kim K. M., McClusky W. G., Abdelmohsen K., Gorospe M.. **Cytoplasmic functions of long noncoding RNAs**. (2018) **9**
9. Kopp F., Mendell J. T.. **Functional classification and experimental dissection of long noncoding RNAs**. (2018) **172** 393-407. DOI: 10.1016/j.cell.2018.01.011
10. Liu S. X., Zheng F., Xie K. L., Xie M. R., Jiang L. J., Cai Y.. **Exercise reduces insulin resistance in type 2 diabetes mellitus via mediating the lncRNA MALAT1/MicroRNA-382-3p/resistin axis**. (2019) **18** 34-44. DOI: 10.1016/j.omtn.2019.08.002
11. Chang W. W., Zhang L., Yao X. M.. **Upregulation of long non-coding RNA MEG3 in type 2 diabetes mellitus complicated with vascular disease: a case-control study**. (2020) **473** 93-99. DOI: 10.1007/s11010-020-03810-x
12. Ghosh S., Stepicheva N., Yazdankhah M.. **The role of lipocalin-2 in age-related macular degeneration (AMD)**. (2020) **77** 835-851. DOI: 10.1007/s00018-019-03423-8
13. Feng W., Han S.. **lncRNA ADAMTS9-AS1/circFN1 competitively binds to miR-206 to elevate the expression of ACTB, thus inducing hypertrophic cardiomyopathy**. (2022) **2022** 13. DOI: 10.1155/2022/1450610
14. Shi R., Jin Y., Hu W.. **Exosomes derived from mmu_circ_0000250-modified adipose-derived mesenchymal stem cells promote wound healing in diabetic mice by inducing miR-128-3p/SIRT1-mediated autophagy**. (2020) **318** C848-C856. DOI: 10.1152/ajpcell.00041.2020
15. Cao Z., Yao F., Lang Y., Feng X.. **Elevated circulating LINC-P21 serves as a diagnostic biomarker of type 2 diabetes mellitus and regulates pancreatic beta-cell function by sponging miR-766-3p to upregulate NR3C2**. (2022) **130** 156-164. PMID: 33007789
16. Tao Z., Shi A., Li R., Wang Y., Wang X., Zhao J.. **Microarray bioinformatics in cancer- a review**. (2017) **22** 838-843. PMID: 29155508
17. Cheng Y., Su Y., Wang S.. **Identification of circRNA-lncRNA-miRNA-mRNA competitive endogenous RNA network as novel prognostic markers for acute myeloid leukemia**. (2020) **11** p. 868. DOI: 10.3390/genes11080868
18. Lin Z., Li X., Zhan X.. **Construction of competitive endogenous RNA network reveals regulatory role of long non-coding RNAs in type 2 diabetes mellitus**. (2017) **21** 3204-3213. DOI: 10.1111/jcmm.13224
19. Yang F., Chen Y., Xue Z.. **High-throughput sequencing and exploration of the lncRNA-circRNA-miRNA-mRNA network in type 2 diabetes mellitus**. (2020) **2020** 13. DOI: 10.1155/2020/8162524
20. Li K., Zhang A., Li X., Zhang H., Zhao L.. **Advances in clinical immunotherapy for gastric cancer**. (2021) **1876**. DOI: 10.1016/j.bbcan.2021.188615
21. Groeneveld O. N., Moneti C., Heinen R.. **The clinical phenotype of vascular cognitive impairment in patients with type 2 diabetes mellitus**. (2019) **68** 311-322. DOI: 10.3233/JAD-180914
22. Al‐Qaissi A., Papageorgiou M., Deshmukh H.. **Effects of acute insulin-induced hypoglycaemia on endothelial microparticles in adults with and without type 2 diabetes**. (2019) **21** 533-540. DOI: 10.1111/dom.13548
23. Qiu M., Yang Z., Guo X. H.. **Trichosanthin attenuates vascular injury-induced neointimal hyperplasia following balloon catheter injury in rats**. (2017) **80** 1212-1221. DOI: 10.1080/15287394.2017.1367140
24. Kuschnerus K., Straessler E. T., Muller M. F., Luscher T. F., Landmesser U., Krankel N.. **Increased expression of miR-483-3p impairs the vascular response to injury in type 2 diabetes**. (2019) **68** 349-360. DOI: 10.2337/db18-0084
25. Jiang Y., Wang C., Zhou S.. **Targeting tumor microenvironment in ovarian cancer: premise and promise**. (2020) **1873**. DOI: 10.1016/j.bbcan.2020.188361
26. Ma Z., Li Q., Liu P., Dong W., Zuo Y.. **METTL3 regulates m6A in endometrioid epithelial ovarian cancer independently of METTl14 and WTAP**. (2020) **44** 2524-2531. DOI: 10.1002/cbin.11459
27. Kong L. Y., Li Y., Rao D. Y.. **miR-666-3p mediates the protective effects of mesenchymal stem cell-derived exosomes against oxygen-glucose deprivation and reoxygenation- induced cell injury in brain microvascular endothelial cells via mitogen-activated protein kinase pathway**. (2021) **18** 20-77. DOI: 10.2174/1567202618666210319152534
28. Mu X., Wang H., Li H.. **Silencing of long noncoding RNA H19 alleviates pulmonary injury, inflammation, and fibrosis of acute respiratory distress syndrome through regulating the microRNA-423-5p/FOXA1 axis**. (2021) **47** 183-197. DOI: 10.1080/01902148.2021.1887967
29. Gao R., Feng Q., Tan G.. **microRNA-613 exerts anti-angiogenic effect on nasopharyngeal carcinoma cells through inactivating the AKT signaling pathway by down-regulating FN1**. (2019) **39**. DOI: 10.1042/BSR20182196
30. Vijan S.. **Type 2 diabetes**. (2019) **171** ITC65-ITC80. DOI: 10.7326/AITC201911050
31. Tang W., Ma J., Gu R.. **Lipocalin 2 suppresses ocular inflammation by inhibiting the activation of NF-**. (2018) **46** 375-388. DOI: 10.1159/000488472
32. Xu L., Li Y., Dai Y., Peng J.. **Natural products for the treatment of type 2 diabetes mellitus: pharmacology and mechanisms**. (2018) **130** 451-465. DOI: 10.1016/j.phrs.2018.01.015
33. Hu F., Lou Y., Shi J.. **Baseline serum albumin and its dynamic change is associated with type 2 diabetes risk: a large cohort study in China**. (2020) **36**. DOI: 10.1002/dmrr.3296
34. Mai Z., Li H., Chen G.. **A bioinformatics investigation into the pharmacological mechanisms of sodium-glucose co-transporter 2 inhibitors in diabetes mellitus and heart failure based on network pharmacology**. (2022) **36** 713-726. DOI: 10.1007/s10557-021-07186-y
35. Kowluru V., Kowluru R. A.. **Increased oxidative stress in diabetes regulates activation of a small molecular weight G-protein, H-Ras, in the retina**. (2007) **13** 602-610. PMID: 17515880
36. Vairaktaris E., Goutzanis L., Kalokerinos G.. **Diabetes increases both N-ras and ets-1 expression during rat oral oncogenesis resulting in enhanced cell proliferation and metastatic potential**. (2007) **21** 615-621. PMID: 17708355
37. Du Z., Uversky V. N.. **A comprehensive survey of the roles of highly disordered proteins in type 2 diabetes**. (2017) **18** p. 2010. DOI: 10.3390/ijms18102010
38. Li Z., Deng X., Lan Y.. **Identification of a potentially functional circRNA-miRNA-mRNA regulatory network in type 2 diabetes mellitus by integrated microarray analysis**. (2021). DOI: 10.23736/S2724-6507.21.03370-8
39. Yuan G., Shi S., Jia Q.. **Use of network pharmacology to explore the mechanism of Gegen (Puerariae lobatae Radix) in the treatment of type 2 diabetes mellitus associated with hyperlipidemia**. (2021) **2021** 14. DOI: 10.1155/2021/6633402
40. Chen Y., Zeng A., He S.. **Autophagy-related genes in atherosclerosis**. (2021) **2021** 11. DOI: 10.1155/2021/6402206
41. Seok H., Ham J., Jang E. S., Chi S. W.. **MicroRNA target recognition: insights from transcriptome-wide non-canonical interactions**. (2016) **39** 375-381. DOI: 10.14348/molcells.2016.0013
42. Zhu H., Leung S. W.. **Identification of microRNA biomarkers in type 2 diabetes: a meta-analysis of controlled profiling studies**. (2015) **58** 900-911. DOI: 10.1007/s00125-015-3510-2
43. Hourigan S. T., Solly E. L., Nankivell V. A.. **The regulation of miRNAs by reconstituted high-density lipoproteins in diabetes-impaired angiogenesis**. (2018) **8** p. 13596. DOI: 10.1038/s41598-018-32016-x
44. Travers M. E., Mackay D. J., Dekker Nitert M.. **Insights into the molecular mechanism for type 2 diabetes susceptibility at the KCNQ1 locus from temporal changes in imprinting status in human islets**. (2013) **62** 987-992. DOI: 10.2337/db12-0819
45. Yang F., Qin Y., Wang Y.. **LncRNA KCNQ1OT1 mediates pyroptosis in diabetic cardiomyopathy**. (2018) **50** 1230-1244. DOI: 10.1159/000494576
46. Jie R., Zhu P., Zhong J., Zhang Y., Wu H.. **LncRNA KCNQ1OT1 affects cell proliferation, apoptosis and fibrosis through regulating miR-18b-5p/SORBS2 axis and NF-ĸB pathway in diabetic nephropathy**. (2020) **12** p. 77. DOI: 10.1186/s13098-020-00585-5
47. Ding S., Zhu Y., Liang Y., Huang H., Xu Y., Zhong C.. **Circular RNAs in vascular functions and diseases**. (2018) **1087** 287-297. DOI: 10.1007/978-981-13-1426-1_23
48. Jiang Q., Liu C., Li C. P.. **Circular RNA-ZNF532 regulates diabetes-induced retinal pericyte degeneration and vascular dysfunction**. (2020) **130** 3833-3847. DOI: 10.1172/JCI123353
49. Shan K., Liu C., Liu B. H.. **Circular noncoding RNA HIPK3 mediates retinal vascular dysfunction in diabetes mellitus**. (2017) **136** 1629-1642. DOI: 10.1161/CIRCULATIONAHA.117.029004
|
---
title: Mechanism of SET8 Activates the Nrf2-KEAP1-ARE Signaling Pathway to Promote
the Recovery of Motor Function after Spinal Cord Injury
authors:
- Xin Li
- Yan Qian
- Wanling Shen
- Shiying Zhang
- Hui Han
- Yu Zhang
- Shuangmei Liu
- Shaokun Lv
- Xiuying Zhang
journal: Mediators of Inflammation
year: 2023
pmcid: PMC10023234
doi: 10.1155/2023/4420592
license: CC BY 4.0
---
# Mechanism of SET8 Activates the Nrf2-KEAP1-ARE Signaling Pathway to Promote the Recovery of Motor Function after Spinal Cord Injury
## Abstract
### Background
Spinal cord injury (SCI) is a common injury of the central nervous system (CNS), and astrocytes are relatively abundant glial cells in the CNS that impairs the recovery of motor function after SCI. It was confirmed that the oxidative stress of mitochondria leads to the accumulation of reactive oxygen species (ROS) in cells, which plays a key role in the motor function of astrocytes. However, the mechanism by which oxidative stress affects astrocyte motility after SCI is still unexplained. Therefore, this study investigated the influence of SET8-regulated oxidative stress on astrocyte autophagy levels after SCI in rats and the potential mechanisms of action.
### Methods
We used real-time quantitative PCR, western blotting, and immunohistochemical staining to analyze SET8, Keap1, and Nrf2 expression at the cellular level and in SCI tissues. ChIP to detect H4K20me1 enrichment in the Keap1 promoter region under OE-SET8 (overexpression of SET8) conditions. Western blotting was used to assess the expression of signature proteins of astrocytes, proteins associated with autophagy, proteins associated with glial scar formation, reactive oxygen species (ROS) levels in cells using DHE staining, and astrocyte number, morphological alterations, and induction of glial scar formation processes using immunofluorescence. In addition, the survival rate of neurons after SCI in rats was examined by using NiSSl staining.
### Results
OE-SET8 upregulates the enrichment of H4K20me1 in Keap1, inhibits Keap1 expression, activates the Nrf2-ARE signaling pathway to suppress ROS accumulation, inhibits oxidative stress-induced autophagy and glial scar formation in astrocytes, and leads to reduced neuronal loss, which promoted the recovery and improvement of motor function after SCI in rats.
### Conclusion
Overexpression of SET8 alleviated oxidative stress by regulating Keap1/Nrf2/ARE, inhibited astrocyte autophagy levels, and reduced glial scar formation as well as neuronal loss, thereby promoting improved recovery of motor function after SCI. Thus, the SET8/H4K20me1 regulatory function may be a promising cellular therapeutic intervention point after SCI.
## 1. Preface
Spinal cord injury (SCI) is a common central nervous system (CNS) injury that causes movement dysfunction and even paralysis. In recent years, the prevalence of SCI has been increasing and trending younger, with over two million SCI cases worldwide and an annual increase of 12 thousand new cases [1]. The incidence of SCI has been increasing in recent years, and the trend is in younger populations. Pathophysiological studies have found that SCI occur and develop in two main stages, with the first stage being the primary injury caused by physical tension. In the second stage, SCI could cause secondary injury, such as mitochondrial dysfunction, oxidative stress damage, and apoptosis [2]. Secondary injury can produce neuronal apoptosis and generate glial scars [3, 4], which further inhibit spinal cord regeneration after SCI [5]. SCI has a significant influence on the recovery of motor function. Therefore, exploring the mechanisms that inhibit or interrupt secondary injury holds promise for promoting recovery of motor function after SCI.
As a highly conserved mechanism of selective intracellular degradation, autophagy mainly occurs through the lysosomal pathway as the steady response of cells to stress stimulation [6]. While most studies on autophagy have focused on immune regulation of tumors, an increasing number of researchers have confirmed that autophagy has a key role in central nervous system (CNS) diseases, especially after SCI, in recent years [7, 8]. Astrocytes are the most abundant and widely distributed class of glial cells in the CNS and have a key role in synapse formation, the production of nutritional factors, the uptake and release of neurotransmitters, and the control of neuron survival [9]. After SCI, astrocytes show abnormal “responsiveness” in the form of upregulation of GFAP and cellular hypertrophy [10]. In addition, astrocytes migrate to the lesion and proliferate, leading to the formation of glial scars [11]. The underlying mechanism is astrocyte autophagy, and Kanno et al. [ 12] found that astrocyte autophagy was activated and Beclin1 expression was upregulated after SCI [13], showing that astrocyte autophagy is a potential mechanism for astrocyte activation. Activated astrocytes are closely associated with neuronal injury [14, 15]. A study showed that motor neurons cocultured with astrocytes died after 72 hours [4]. Therefore, inhibition of astrocyte autophagy may improve functional prognosis after SCI.
Reactive oxygen species (ROS) are a universal or general name for a class of single-electron reduction products of oxygen with high oxidative activity or ions in the organism, which are produced by mitochondria and act as redox signal messengers in the organism [16]. After SCI, ROS in mitochondria increase [17]. ROS are produced by the mitochondria and as redox signaling messengers in the body, leading to the formation of oxidative stress [18]. Oxidative stress directly affects the function of astrocytes [19], and induces autophagy in astrocytes [20]. Therefore, an effective way to protect the injured spinal cord from mitochondrial oxidative stress is to inhibit the production of ROS. The keap1-Nrf2/ARE pathway is the main defense mechanism of cells against oxidative stress, in which Keap1 can negatively regulate Nrf2 and the activated Nrf2 molecule can activate antioxidant genes by combining with antioxidant response elements (AREs) [21]. It was shown that activated Nrf2 prevents astrocytes from killing cocultured motor neurons in a glutathione-dependent mechanism [22, 23]. Thus, the Keap1-Nrf2/ARE signaling pathway could reduce or inhibit astrocytes autophagy and motor neuron apoptosis under the activated state, promoting a potential therapeutic approach for motor function recovery after SCI.
SET domain-containing protein 8 (SET8) is a lysine monomethyltransferase, also called SETD8, which specifically methylates H4 lysine 20 [24]. SET8/H4K20me1 participates in the regulation of many signaling pathways, including the induction of apoptosis in cancer cells [25], mediating endothelial cell inflammation [26] and cellular oxidation [27]. Chen et al. [ 28] showed that SET8/H4K20me1 regulates Keap1-Nrf2/ARE to inhibit ROS accumulation from oxidative damage in hyperglycemia-induced endothelial injury. This study hypothesized that SET8/H4K20me1 plays a key role in the regulation of the Keap1-Nrf2/ARE signaling pathway after SCI, thereby participating in the regulation of autophagy in astrocytes and promoting the recovery of motor function after SCI.
## 2.1. Laboratory Animals
Healthy female SD rats of the SPF class (Animal Experiment Center of Kunming Medical University), weighing 275 ± 25 g, were utilized. The experimental scheme of this study was approved by the Animal Ethics Committee of Hunan Provincial People's Hospital, and it fully met the requirements of the National Institutes of Health Laboratory Animal Care Guide. This study included the following groups: sham-operated group ($$n = 10$$), SCI+OE-sET8 group ($$n = 10$$), and SCI group ($$n = 20$$), all the animals were distributed randomly.
## 2.2. Post-SCI Model Group in Rats
Before the skin and muscles on the covering layer were cut, rats were anesthetized with $4\%$ pentobarbital sodium, and a laminectomy was performed on T9-11 to keep the dura mater intact. Epidural impingement was performed on the rat spinal cord with a weight-reducing impinger (20 g) placed into the T10 region of the spinal cord after descending 0.25 cm [29]. After the operation, to prevent dehydration, all animals were injected with $0.9\%$ saline (30 mL/kg). In this study, all rats were housed in cages with controlled feeding conditions individually, with enough food and water, and had a light/dark cycle at 12-hour intervals. Assisted urination was performed 3 times daily after SCI during the study period.
## 2.3. Astrocyte Culture
Rat astrocytes were provided by Wuhan Pronosai Life Co. Rat astrocytes were inoculated at 1 × 105 cells/mL and cultured at 37°C with saturated humidity and CO2. Before the cells covered $90\%$ of the bottom of the culture dish, the medium was changed every other day. To purify the astrocytes, the mixed astrocytes were placed at a concentration of 1 × 105 cells/mL into culture dishes coated with polylysine. Ara-C medium (5 μg/mL) was added after 24 h of culture and replaced with DMEM+$10\%$ fetal bovine serum after 48 h of culture. During 10 days of continuous cell culture, the medium was changed every 3 days.
## 2.4. Chromatin Immunoprecipitation Analysis
A ChIP Assay Kit (Beyotime, Shanghai, China) was used to perform the ChIP assays in this study. AsTs (1 × 107 cells) were fixed with $1\%$ formaldehyde for 10 min to crosslink DNA and protein, and the crosslinking reaction was stopped by adding glycine. Subsequently, the cells were ultrasonically treated with a Microson ultrasonic cell disruptor to shear chromatin (every 15 s, every 2 min, power = 15 W, and amplitude = 10). Ten microliters of the ultrasonically treated extract was obtained from each group as an input sample, and the anti-h4k20me1 antibody or IgG negative control (Abcam, Cambridge, UK) was added to the remaining samples and incubated at 4°C for 12 h. DNA–protein crosslinks were reversed with protein G magnetic beads bound to the immunoprecipitates for 2 h at 65°C. Subsequently, the enriched DNA fragments were purified and analyzed by qPCR. The primers for Keap1 were as follows: 5′-TGACAAAACTGAGCCTCCTAGC-3′ and Rev 5′-GCATCAAAGAGTGATGCTGAATG-3′.
## 2.5. ROS Detection
ROS was tested by the ROS Assay Kit (Beyotime, Shanghai, China). In this study, rats were euthanized 3 days after SCI. Samples were lysed using 0.01 mol/L PBS to suspend the brain tissue and centrifuged at 500 × g for 10 min at low temperature (4°C). Then, the samples were mixed with 190 μL supernatant and 10 μL dichlorodihydrofluorescein diacetate (DCFH-DA; 1 M) in microtiter wells for 30 minutes at room temperature. Then, a BCA Protein Assay Kit (Beyotime, Shanghai, China) was used to obtain the protein levels. Ultimately, the ROS levels were displayed as fluorescence/mg protein. Similarly, the mixture of astrocytes and 10 micron DCFH-DA was incubated in a 96-well plate at 25°C for 30 minutes and measured using a fluorometer.
## 2.6. NiSSL Dyeing
After drying, 7 micron transverse frozen slices were directly immersed in the mixture (ethanol/chloroform = 1: 1) for 12 h in the dark at 22 ± 1°C. Then, the sections were dyed with $0.1\%$ cresyl violet (Sango Biotech, Shanghai, China) solution for 5 min, divided, dehydrated, and rinsed. Finally, the slices were fixed using Permount™ and observed using an Olympus optical microscope (Leica DMI4000B, Germany).
## 2.7. Quantitative Real-Time Polymerase Chain Reaction (qRT–PCR)
In this study, we extracted total RNA from tissues and cells using a Total RNA Extractor (Sangon Biotech). A cDNA synthesis kit (Vazyme, Nanjing, China) was used to reverse transcribe 2 μg mRNA into cDNA, which was then diluted 10 times. One microliter of the prepared cDNA was used for qPCR. All primers (Table 1) used in this study were designed with Premier 5.0. The two-step reaction conditions for PCR were as follows: predenaturation (maintained at 95°C for 5 min), maintained at 95°C for 10 s, annealing (30 s), and extension (30 s). Both annealing and extension were cycled 40 times. The confidence of the PCR results was assessed by the dissociation curve and cycle threshold (CT) values. The results were calculated by the 2-ΔΔCt method after repeated at least 3 times.
## 2.8. Western Blotting
In this study, proteins (including nuclear and cytoplasmic proteins) in spinal cord-injured rats were extracted by using RIPA lysis buffer (Sangon Biotech, Shanghai), and the total protein concentration was determined using a BCA assay (Sangon Biotech, Shanghai). A $10\%$ SDS–PAGE gel was used to separate the total proteins, which were then transferred to PVDF membranes by constant current flow at 200 mA. Subsequently, PVDF membranes were cultured with antibodies (Abcam, USA) for 12 h at 4°C. The PVDF membranes were washed with TBS buffer and incubated with secondary antibodies (Abeam) at 25°C for 1 h. After washing the membranes 3 times, chemiluminescent reagents were added, and the bands were analyzed for grayscale values using ImageJ software. Each experiment was repeated 3 times independently.
## 2.9. Immunohistochemistry (IHC)
In this study, IHC experiments were carried out by 3,3′-diaminobenzidine (DAB) analysis. After the glass slide was baked at 65°C for 2 h, it was placed in xylene for 10 minutes. The sections were incubated in the following ethanol gradient (5 min for each solution): $90\%$, $80\%$, $70\%$, and distilled water. In a wet room, citric acid buffer was used to treat the slices, and hydrogen peroxide ($3\%$) was used to remove endogenous peroxidase (25°C, 10 min). Sections were blocked with $5\%$ bovine serum at 37°C for 30 min and then incubated with anti-Set8 and anti-keap1 antibodies (1: 200) for 12 h at 4°C. They were incubated with goat anti-rabbit antibodies (IgG, 1: 100) for 30 min at 37°C after washing the slices with PBS buffer. 3,3′-DAB was used to observe the sections, and a light microscope was used to acquire the images.
## 2.10. Immunofluorescence
On the 3rd, 7th, and 21st days after SCI, all rats were euthanized with deep pentobarbital and then infused with 0.1 mol/L PBS and $4\%$ paraformaldehyde (PFA). AsTs of each group were resuspended and inoculated onto sterile coverslips coated with polylysine and then dissolved in goat serum containing $0.3\%$ Triton X-100. Subsequently, the cells were incubated with $5\%$ BSA for 1 h, and primary antibody (1: 200, Abcam, UK) was added and incubated at 4°C for 12 h. Afterward, the AsTs were incubated with the corresponding fluorescent secondary antibodies. The nuclei were stained with DAPI, and images were obtained by fluorescence microscopy.
## 3.1. Low Expression of Keap1 Alleviates LPS-Induced Oxidative Damage Induction and Inhibits Astrocyte Autophagy
After SCI, mitochondrial dysfunction leads to oxidative stress in rats; astrocytes migrate to the lesion site and become abnormally “reactive,” preventing recovery of motor function and leading to glial scar formation after SCI. KEAP1 is a sensor of the oxidative stress pathway [30]. To detect the potential influence of KEAP1 on oxidative stress behavior after SCI, we used LPS to induce oxidative stress behavior in simulated rats after SCI and explored the effect of KEAP1 knockdown on LPS-induced astrocyte autophagy after oxidative injury. The experimental data showed that LPS enhanced the accumulation of ROS (Figure 1(e)) and glial scar formation (Figure 1(d)), activated astrocytes and enhanced astrocyte autophagy, as evidenced by enhanced expression of GFAP and Vimentin (Figure 1(c)) and autophagy-related proteins LC3II/I and Beclin1, while P62 expression was diminished (Figure 1(b)). In addition, knockdown of Keap1 downregulated Keap1 expression compared to the LPS (Figure 1(a)), which alleviated the oxidative stress response under LPS induction, as evidenced by a significant decrease in ROS accumulation detected by DHE staining (Figure 1(e)). It also impaired astrocyte autophagy as well as glial scar formation, and immunoblotting revealed diminished expression of the astrocyte autophagy-related proteins LC3II/I and Beclin1, while P62 expression was enhanced (Figure 1(b)). These results suggest that si-KEAP1 alleviates LPS-induced oxidative damage and attenuates astrocyte autophagy and glial scar formation after LPS.
## 3.2. Low Expression of Keap1 Reduces the Inhibitory Effect on the Nrf2-ARE Signaling Pathway
To detect the influence of KEAP1 on the Nrf2/ARE signaling pathway after SCI in rats, we used transfection of si-Keap1 to induce LPS-treated astrocytes. The effect of si-Keap1 was verified by qPCR and western blotting analysis. si-Keap1 reversed the LPS-mediated inhibition of the Nrf2/ARE signaling pathway (Figures 2(a) and 2(b)). Keap1 negatively regulates the Nrf2/ARE signaling pathway in oxidative injury in rat SCI.
## 3.3. SET8 Reduces Keap1 Expression by Enhancing H4K20me1 Enrichment in the Keap1 Promoter Region
SET8 is a key regulator of DNA methylation and is the only known modifier enzyme that catalyzes monomethylation of histone H4K20me1 [31]. We investigated whether SET8 regulates the expression of Keap1, a key factor in oxidative stress. First, we used protein blotting and RT–qPCR to detect H4K20me1 levels in OE-SET8 cells. The results showed that H4K20me1 levels were significantly upregulated in OE-SET8 cells (Figure 3(a)). Second, this study tested the distribution of H4K20me1 in astrocyte genomes by the ChIP method, and the results showed that H4K20me1 can be enriched in the Keap1 promoter (Figure 3(b)). In addition, compared with the LPS, OE-SET8 significantly decreased Keap1 mRNA expression, as shown by protein blotting and qPCR (Figures 3(a) and 3(c)). This result indicated that SET8 inhibits Keap1 by promoting the enrichment of H4K20me1 in the Keap1 promoter, which in turn activates the Nrf2-ARE signaling pathway.
## 3.4. SET8 Inhibits Oxidative Stress-Induced Autophagy and Glial Scar Formation in Astrocytes through the KEAP1-Nrf2-ARE Signaling Pathway
We investigated whether SET8 regulates oxidative stress via the KEAP1-Nrf2-ARE signaling pathway and thus affects astrocyte autophagy and glial scar formation. First, we used western blotting to detect SET8, KEAP1, and Nrf2-related proteins in the signaling pathway. Compared with the LPS+OE-SET8+OE-KEAP1 group, the LPS+OE-SET8 group significantly decreased the expression of KEAP1 and promoted SET8 and Nrf2 (Figure 4(a)). Next, we detected ROS levels using DHE staining and found that the LPS+OE-SET8 group had significantly reduced ROS content (Figure 4(e)). Meanwhile, compared with the LPS+OE-SET8+OE-KEAP1 group, LPS+OE-SET8 significantly decreased the expression of astrocyte signature proteins, autophagy-related proteins LC3II/I and Beclin1, and glial scar formation-related proteins according to the western blotting results (Figures 4(b)–4(d)). These results suggest that SET8, through activation of the KEAP1-Nrf2-ARE signaling pathway, inhibits oxidative stress and suppresses astrocyte autophagy as well as glial scar formation, thereby promoting recovery after SCI.
## 3.5. SET8 Promotes Motor Function Recovery after SCI in Rats In Vivo
We further validated that SET8 promotes motor function recovery by regulating the KEAP1-Nrf2-ARE signaling pathway after SCI in rats. We simulated post-SCI rats and established an in vivo model. First, we analyzed the expression of the key genes SET8 and KEAP1 after SCI in rats by immunohistochemistry (Figure 5(a)), and the experimental results were similar between the in vivo and in vitro experiments. Next, we detected astrocyte autophagy proteins using western blotting, and the experimental results showed that OE-SET8 significantly decreased the level of astrocyte autophagy, and the autophagy-related proteins LC3II/I and Beclin1 were diminished in astrocytes in OE-SET8-induced tissues, while P62 expression was enhanced (Figure 5(b)). Meanwhile, western blotting assays showed that the glial scar formation-related proteins Brevican and Nucrocan were downregulated (Figure 5(c)). The immunofluorescence assay also showed more visually that in the Model group, a wide thick glial scar was formed by proliferating, hypertrophic astrocytes, while in the OE-SET8 rats, normal, more elongated astrocytes were observed, and the glial scar was shorter and narrower (Figure 5(d)) [32]. Here, we obtained a good phenotypic characterization of the astrocytes observed in the in vivo animal group.
Neurons are known to be critical for recovery after SCI, and Liu et al. [ 33] showed that neurons innervate changes in neuroplasticity and improve functional recovery after SCI in the spinal cord. However, after SCI, neuronal death occurs in large numbers and is largely driven by reactive astrocytes [34, 35]. The results of this study are presented below. To determine the effect of OE-SET8 on neurons in rat SCI, we used Nissl staining to examine the neurons in different groups. After SCI in rats, the number of surviving neurons decreased in the SCI group, while there was a larger number of surviving neurons in OE-SET8-induced SCI tissue (Figure 5(e)).
In summary, we fully validated in vivo that SET8 can inhibit the autophagy phenomenon in astrocyte and glial scar formation by regulating the KEAP1-Nrf2-ARE signaling pathway to improve neuronal survival after SCI in rats, thereby promoting the recovery of motor function after SCI in rats.
## 4. Discussion
SCI usually results in disability or even paralysis, because the oxidative stress triggers astrocyte activation, glial scar formation, and neuronal cell death in the lesion[36], which severely hinders prominent regeneration and motor function recovery in SCI. Currently, there is no effective clinical treatment for the recovery of motor function after SCI. In this study, we studied the activation of the Keap1-Nrf2-ARE signaling pathway by SET8/H4K20me1 to inhibit oxidative stress-induced autophagy in astrocytes and promote recovery of motor function after SCI.
After traumatic injury to the CNS, including SCI, the surrounding astrocytes are reactive, proliferate, hypertrophy, and migrate to the lesion site, intertwining to form a glial scar [37–39]. This pathological phenomenon leads to the production of axon growth inhibitors and prevents the regeneration of the spinal cord. In the mammalian central nervous system, axon regeneration can be hindered by the formation of glial scars, which is the main reason for the low regeneration ability [40]. In this study, glial scar formation mediated by LPS-induced astrocyte activation after SCI was investigated in rats. LPS-induced astrocyte activation was associated with upregulation of GFAP, LC3II/I and Beclin1, downregulation of P62 and enhanced autophagy. Brevican and versican are produced by reactive astrocytes in glial scars, are the main inhibitory extracellular matrix molecules [41], and play a key role in the regeneration and motor function of the spinal cord in rats. Western blotting results showed that neuroglial scar brevican and versican were significantly elevated after LPS-induced enhancement of astrocyte autophagy. In our vivo experiments, we also foundthat astrocytes sequentially exhibited phenotypically distinct changes, first increasing the number of hypertrophied reactive astrocytes and then developing into glial scars after SCI. It was suggested that the enhanced autophagic capacity of activated astrocytes plays a key role in the recovery of motor function in rats [42]. Therefore, the main therapeutic strategy for glial scarring focuses on the regulation of astrocyte activation as well as autophagy [43].
As a nuclear factor red lineage 2-related factor, Nrf2 regulates cellular defense against oxidative damage through its expression in response to oxidative stress [44]. Nrf2 plays a key role in the recovery of motor function after SCI, as it regulates oxidative stress-related molecules such as ROS, thioredoxin (TXN), and glutathione (GSH) [45]. Nrf2 inhibits oxidative stress and blocks the apoptotic cascade by activating Nrf2 in the damaged spinal cord [46]. The mechanism suggests that Nrf-2 is released from Keap-1 and translocates to the nucleus to bind to the ARE, thereby activating antioxidant defense enzymes and attenuating cellular oxidative stress [47, 48]. Several studies have shown that inhibition of Keap1 activates Nrf2, thereby inhibiting oxidative stress [49, 50]. In this study, we revealed that the Nrf2/ARE signaling pathway can be inhibited by LPS-induced astrocyte activation after SCI in rats. In addition, si-Keap1 ameliorated LPS-mediated oxidative damage, as evidenced by a decrease in ROS content by DHE staining and a decrease in the astrocyte marker proteases GFAP and VimentinD by western blotting. The expression of the autophagy-associated proteins LC3II/I and Beclin1 was downregulated by western blotting, and the expression of P62 was upregulated, while the expression of the glial scar-associated proteins brevican and neurocan was low. Therefore, Keap1 could be used as a therapeutic target for motor function recovery after SCI.
In multicellular organisms, SET8 is the only enzyme that produces histone H4 monomethylation on lysine 20 (H4K20me1) [51]. SET8 plays a key role in the epigenetic regulation of genes in many cellular processes [52]. The mechanism of SET8 regulation in human umbilical vein endothelial cells (HUVECs) with hyperglycemia showed that low expression of SET8 increased the production of cellular ROS, leading to increased oxidative stress, while overexpression of SET8 attenuated oxidative stress damage to endothelial tissue by blocking ROS accumulation [27]. However, the mechanism of SET8 regulation in rat SCI is not clear. In this study, our results showed that overexpression of SET8 reversed the LPS-mediated Keap1/Nrf2/ARE signaling pathway, alleviated ROS accumulation, inhibited LPS-mediated astrocyte autophagy and created glial scars, and thus improved the recovery of motor function after SCI. Furthermore, ChIP assays showed that overexpression of SET8 enriched the downstream target H4K20me1 in the Keap1 promoter region and showed by RT–qPCR and western blotting that overexpression of SET8 downregulated Keap1 and upregulated Nrf2 expression. These data fully demonstrate that overexpression of SET8 can inhibit Keap1 to reverse LPS-induced repression of the Nrf2/ARE signaling pathway.
Nissl staining is a commonly used method for neuronal morphology and cytoarchitecture detection [53]. In this study, to further verify the regulatory mechanism of overexpressed SET8 in rat SCI, we detected the survival of neurons in rat SCI tissues by Nissl staining. In the model rat group, there were significantly fewer neurons involved in regulation by overexpressed SET8 after SCI, which fully indicated in vivo that overexpressed SET8 could increase the recovery of motor function after SCI in rats.
In summary, these results showed that SET8 reduces glial scar generation and inhibits astrocyte autophagy through regulation of Keap1/Nrf2/ARE, thereby improving motor function recovery after SCI. OE-SET8/H4K20me1 inhibits Keap1, and the Nrf2-ARE signaling pathway was activated to inhibit oxidative stress-induced autophagy in astrocytes and promote motor function recovery after SCI. Here, we provide information on how SET8 regulates astrocyte activity by the Keap1/Nrf2/ARE signaling pathway after SCI, thereby generating a permissive environment that promotes motor function recovery after SCI, which can help optimize cellular therapies after SCI and develop therapeutic strategies for secondary SCI.
## Data Availability
The data used to support the findings of this study are included within the article.
## Conflicts of Interest
The authors have no conflicts of interest to declare.
## Authors' Contributions
Xin Li and Yan Qian are responsible for the conceptualization. Shaokun Lv, Xin Li, and Yan Qian are assigned to the methodology. Shiying Zhang and Wanling Shen are assigned to the software. Hui Han and Yu Zhang did the validation. Shaokun Lv, Shuangmei Liu, and Wanling Shen performed the formal analysis. Xin Li, Yan Qian, Shaokun Lv, and Shiying Zhang did the investigation. Shaokun Lv and Xiuying Zhang are responsible for the resources. Hui Han, Yu Zhang, Shuangmei Liu, and Shiying Zhang curated data. Xin Li, Yan Qian, Shaokun Lv, and Xiuying Zhang are responsible for writing—original draft preparation. Shaokun Lv and Xiuying Zhang are responsible for writing—review and editing. Shaokun Lv and Xiuying Zhang did visualization. Xin Li, Shaokun Lv, and Xiuying Zhang are assigned to the supervision. Shaokun Lv and Xiuying Zhang acquired funding. All authors have read and agreed to the published version of the manuscript. Xin Li and Yan Qian contributed equally to this work.
## References
1. Rosner J., Avalos P., Acosta F., Liu J., Drazin D.. **The potential for cellular therapy combined with growth factors in spinal cord injury**. (2012) **2012** 11. DOI: 10.1155/2012/826754
2. Fakhri S., Abbaszadeh F., Moradi S. Z., Cao H., Khan H., Xiao J.. **Effects of polyphenols on oxidative stress, inflammation, and interconnected pathways during spinal cord injury**. (2022) **2022** 34. DOI: 10.1155/2022/8100195
3. Kotipatruni R. R., Dasari V. R., Veeravalli K. K., Dinh D. H., Fassett D., Rao J. S.. **p53- and Bax-mediated apoptosis in injured rat spinal cord**. (2011) **36** 2063-2074. DOI: 10.1007/s11064-011-0530-2
4. Allen S. P., Hall B., Castelli L. M.. **Astrocyte adenosine deaminase loss increases motor neuron toxicity in amyotrophic lateral sclerosis**. (2019) **142** 586-605. DOI: 10.1093/brain/awy353
5. Yang T., Dai Y., Chen G., Cui S. S.. **Dissecting the dual role of the glial scar and scar-forming astrocytes in spinal cord injury**. (2020) **14** p. 78. DOI: 10.3389/fncel.2020.00078
6. Shin H. J., Kim H., Oh S.. **AMPK-SKP2-CARM1 signalling cascade in transcriptional regulation of autophagy**. (2016) **534** 553-557. DOI: 10.1038/nature18014
7. Galluzzi L., Bravo-San Pedro J. M., Blomgren K., Kroemer G.. **Autophagy in acute brain injury**. (2016) **17** 467-484. DOI: 10.1038/nrn.2016.51
8. Lipinski M. M., Wu J., Faden A. I., Sarkar C.. **Function and mechanisms of autophagy in brain and spinal cord trauma**. (2015) **23** 565-577. DOI: 10.1089/ars.2015.6306
9. Matias I., Morgado J., Gomes F. C. A.. **Astrocyte heterogeneity: impact to brain aging and disease**. (2019) **11** p. 59. DOI: 10.3389/fnagi.2019.00059
10. Sofroniew M. V., Vinters H. V.. **Astrocytes: biology and pathology**. (2010) **119** 7-35. DOI: 10.1007/s00401-009-0619-8
11. Lagos-Cabré R., Burgos-Bravo F., Avalos A. M., Leyton L.. **Connexins in astrocyte migration**. (2020) **10, article 1546**. DOI: 10.3389/fphar.2019.01546
12. Kanno H., Ozawa H., Sekiguchi A., Itoi E.. **Spinal cord injury induces upregulation of Beclin 1 and promotes autophagic cell death**. (2009) **33** 143-148. DOI: 10.1016/j.nbd.2008.09.009
13. Huang R., Zhang Y., Han B.. **Circular RNAHIPK2regulates astrocyte activation via cooperation of autophagy and ER stress by targetingMIR124–2HG**. (2017) **13** 1722-1741. DOI: 10.1080/15548627.2017.1356975
14. Lau J. W., Senok S., Stadlin A.. **Methamphetamine-induced oxidative stress in cultured mouse astrocytes**. (2000) **914** 146-156. DOI: 10.1111/j.1749-6632.2000.tb05192.x
15. Clark K. H., Wiley C. A., Bradberry C. W.. **Psychostimulant abuse and neuroinflammation: emerging evidence of their interconnection**. (2013) **23** 174-188. DOI: 10.1007/s12640-012-9334-7
16. Andrabi S. S., Yang J., Gao Y., Kuang Y., Labhasetwar V.. **Nanoparticles with antioxidant enzymes protect injured spinal cord from neuronal cell apoptosis by attenuating mitochondrial dysfunction**. (2020) **317** 300-311. DOI: 10.1016/j.jconrel.2019.12.001
17. Visavadiya N. P., Patel S. P., Vanrooyen J. L., Sullivan P. G., Rabchevsky A. G.. **Cellular and subcellular oxidative stress parameters following severe spinal cord injury**. (2016) **8** 59-67. DOI: 10.1016/j.redox.2015.12.011
18. Mittal M., Siddiqui M. R., Tran K., Reddy S. P., Malik A. B.. **Reactive oxygen species in inflammation and tissue injury**. (2014) **20** 1126-1167. DOI: 10.1089/ars.2012.5149
19. Phatnani H., Maniatis T.. **Astrocytes in Neurodegenerative Disease**. (2015) **7**. DOI: 10.1101/cshperspect.a020628
20. Zhu L., Zang J., Liu B.. **Oxidative stress-induced RAC autophagy can improve the HUVEC functions by releasing exosomes**. (2020) **235** 7392-7409. DOI: 10.1002/jcp.29641
21. Magesh S., Chen Y., Hu L.. **Small molecule modulators of Keap1-Nrf2-ARE pathway as potential preventive and therapeutic agents**. (2012) **32** 687-726. DOI: 10.1002/med.21257
22. Vargas M. R., Pehar M., Cassina P.. **Fibroblast growth factor-1 induces heme oxygenase-1 via nuclear factorerythroid 2-related factor 2 (Nrf2) in spinal cord astrocytes:**. (2005) **280** 25571-25579. DOI: 10.1074/jbc.M501920200
23. Vargas M. R., Pehar M., Cassina P., Beckman J. S., Barbeito L.. **Increased glutathione biosynthesis by Nrf2 activation in astrocytes prevents p75NTR-dependent motor neuron apoptosis**. (2006) **97** 687-696. DOI: 10.1111/j.1471-4159.2006.03742.x
24. Beck D. B., Oda H., Shen S. S., Reinberg D.. **PR-Set7 and H4K20me1: at the crossroads of genome integrity, cell cycle, chromosome condensation, and transcription**. (2012) **26** 325-337. DOI: 10.1101/gad.177444.111
25. Xu L. F., Shi L., Zhang S. S.. **LukS-PV induces apoptosis via the SET8-H4K20me1-PIK3CB axis in human acute myeloid leukemia cells**. (2021) **11, article 718791**. DOI: 10.3389/fonc.2021.718791
26. Shen X., Chen X., Wang J.. **SET8 suppression mediates high glucose-induced vascular endothelial inflammation via the upregulation of PTEN**. (2020) **52** 1715-1729. DOI: 10.1038/s12276-020-00509-3
27. Chen X., Wu Q., Jiang H.. **SET8 is involved in the regulation of hyperglycemic memory in human umbilical endothelial cells**. (2018) **50** 635-642. DOI: 10.1093/abbs/gmy051
28. Chen X., Qi J., Wu Q.. **High glucose inhibits vascular endothelial Keap1/Nrf2/ARE signal pathway via downregulation of monomethyltransferase SET8 expression**. (2020) **52** 506-516. DOI: 10.1093/abbs/gmaa023
29. Kimura-Kuroda J., Teng X., Komuta Y.. **An**. (2010) **43** 177-187. DOI: 10.1016/j.mcn.2009.10.008
30. Baird L., Yamamoto M.. **The molecular mechanisms regulating the KEAP1-NRF2 pathway**. (2020) **40**. DOI: 10.1128/MCB.00099-20
31. Oda H., Okamoto I., Murphy N.. **Monomethylation of histone H4-lysine 20 is involved in chromosome structure and stability and is essential for mouse development**. (2009) **29** 2278-2295. DOI: 10.1128/MCB.01768-08
32. Goncalves M. B., Malmqvist T., Clarke E.. **Neuronal RAR**. (2015) **35** 15731-15745. DOI: 10.1523/JNEUROSCI.1339-15.2015
33. Liu Z. H., Yip P. K., Adams L.. **A single bolus of docosahexaenoic acid promotes neuroplastic changes in the innervation of spinal cord interneurons and motor neurons and improves functional recovery after spinal cord injury**. (2015) **35** 12733-12752. DOI: 10.1523/JNEUROSCI.0605-15.2015
34. Filous A. R., Silver J.. **Targeting astrocytes in CNS injury and disease: a translational research approach**. (2016) **144** 173-187. DOI: 10.1016/j.pneurobio.2016.03.009
35. Lukacova N., Kisucka A., Kiss Bimbova K.. **Glial-neuronal interactions in pathogenesis and treatment of spinal cord injury**. (2021) **22** p. 13577. DOI: 10.3390/ijms222413577
36. Li X. Q., Wang J., Fang B., Tan W. F., Ma H.. **Intrathecal antagonism of microglial TLR4 reduces inflammatory damage to blood-spinal cord barrier following ischemia/reperfusion injury in rats**. (2014) **7** p. 28. DOI: 10.1186/1756-6606-7-28
37. Anderson M. A.. **Targeting central nervous system regeneration with cell type specificity**. (2021) **32** 397-405. DOI: 10.1016/j.nec.2021.03.011
38. Kjell J., Olson L.. **Rat models of spinal cord injury: from pathology to potential therapies**. (2016) **9** 1125-1137. DOI: 10.1242/dmm.025833
39. O'shea T. M., Burda J. E., Sofroniew M. V.. **Cell biology of spinal cord injury and repair**. (2017) **127** 3259-3270. DOI: 10.1172/JCI90608
40. Silver J., Miller J. H.. **Regeneration beyond the glial scar**. (2004) **5** 146-156. DOI: 10.1038/nrn1326
41. Mckeon R. J., Jurynec M. J., Buck C. R.. **The chondroitin sulfate proteoglycans neurocan and phosphacan are expressed by reactive astrocytes in the chronic CNS glial scar**. (1999) **19** 10778-10788. DOI: 10.1523/JNEUROSCI.19-24-10778.1999
42. Okada S., Nakamura M., Katoh H.. **Conditional ablation of Stat3 or Socs3 discloses a dual role for reactive astrocytes after spinal cord injury**. (2006) **12** 829-834. DOI: 10.1038/nm1425
43. Yang T., Xing L., Yu W., Cai Y., Cui S., Chen G.. **Astrocytic reprogramming combined with rehabilitation strategy improves recovery from spinal cord injury**. (2020) **34** 15504-15515. DOI: 10.1096/fj.202001657RR
44. He F., Ru X., Wen T.. **NRF2, a transcription factor for stress response and beyond**. (2020) **21**. DOI: 10.3390/ijms21134777
45. Tonelli C., Chio I. I. C., Tuveson D. A.. **Transcriptional regulation by Nrf2**. (2018) **29** 1727-1745. DOI: 10.1089/ars.2017.7342
46. Lv R., Du L., Zhang L., Zhang Z.. **Polydatin attenuates spinal cord injury in rats by inhibiting oxidative stress and microglia apoptosis via Nrf2/HO-1 pathway**. (2019) **217** 119-127. DOI: 10.1016/j.lfs.2018.11.053
47. Kim H. J., Zheng M., Kim S. K.. **CO/HO-1 induces NQO-1 expression via Nrf2 activation**. (2011) **11** 376-382. DOI: 10.4110/in.2011.11.6.376
48. Zhao Y., Sun Y., Wang G., Ge S., Liu H.. **Dendrobium officinale polysaccharides protect against MNNG-induced PLGC in rats via activating the NRF2 and antioxidant enzymes HO-1 and NQO-1**. (2019) **2019** 11. DOI: 10.1155/2019/9310245
49. Canning P., Sorrell F. J., Bullock A. N.. **Structural basis of Keap1 interactions with Nrf2**. (2015) **88, Part B** 101-107. DOI: 10.1016/j.freeradbiomed.2015.05.034
50. Uruno A., Yagishita Y., Yamamoto M.. **The Keap1-Nrf2 system and diabetes mellitus**. (2015) **566** 76-84. DOI: 10.1016/j.abb.2014.12.012
51. Jørgensen S., Schotta G., Sørensen C. S.. **Histone H4 lysine 20 methylation: key player in epigenetic regulation of genomic integrity**. (2013) **41** 2797-2806. DOI: 10.1093/nar/gkt012
52. Couture J. F., Collazo E., Brunzelle J. S., Trievel R. C.. **Structural and functional analysis of SET8, a histone H4 Lys-20 methyltransferase**. (2005) **19** 1455-1465. DOI: 10.1101/gad.1318405
53. Oertel D., Wu S. H., Garb M. W., Dizack C.. **Morphology and physiology of cells in slice preparations of the posteroventral cochlear nucleus of mice**. (1990) **295** 136-154. DOI: 10.1002/cne.902950112
|
---
title: 'A new continuous glucose monitor for the diagnosis of gestational diabetes
mellitus: a pilot study'
authors:
- Daria Di Filippo
- Amanda Henry
- Chloe Bell
- Sarah Haynes
- Melissa Han Yiin Chang
- Justine Darling
- Alec Welsh
journal: BMC Pregnancy and Childbirth
year: 2023
pmcid: PMC10023314
doi: 10.1186/s12884-023-05496-7
license: CC BY 4.0
---
# A new continuous glucose monitor for the diagnosis of gestational diabetes mellitus: a pilot study
## Abstract
### Background
Gestational Diabetes Mellitus (GDM) incidence and adverse outcomes have increased globally. The validity of the oral glucose tolerance test (OGTT) for GDM diagnosis has long been questioned, with no suitable substitute reported yet. Continuous Glucose Monitoring (CGM) is potentially a more acceptable and comprehensive test. The aim of this study was to assess the Freestyle Libre Pro 2 acceptability as a diagnostic test for GDM, then triangulating its results with OGTT results as well as risk factors and sonographic features of GDM.
### Methods
Women wore the CGM device for 7 days at 24–28 weeks, undergoing the OGTT before CGM removal. CGM/OGTT acceptability as well as GDM risk factors evaluation occurred via three online surveys. CGM distribution/variability/time in range parameters, combined in a CGM Score of Variability (CGMSV), were triangulated with OGTT results and a risk-factor-based Total Risk Score (TRS). In a subgroup, GDM ultrasound features (as modified Ultrasound Gestational Diabetes Score – m-UGDS) were also incorporated.
### Results
Of 107 women recruited, 87 ($81\%$) were included: 74 ($85\%$) with negative OGTT (NGT) and 13 ($15\%$) positive (GDM). No significant difference was found between NGT and GDM in terms of demographics (apart from family history of diabetes mellitus), CGM parameters and perinatal outcomes. Women considered CGM significantly more acceptable than OGTT ($81\%$ versus $27\%$ rating $\frac{5}{5}$, $p \leq 0.001$).
Of the 55 NGT with triangulation data, 28 were considered ‘true negative’ (TRS concordant with OGTT and CGMSV): of these $\frac{4}{5}$ evaluated at ultrasound had m-UGDS below the cut-off. Five women were considered ‘false negative’ (negative OGTT with both TRS and CGMSV above the respective cut-offs). Triangulation identified also six ‘false positive’ women (positive OGTT but TRS and CGM both below the cut-offs). Only one woman for each of the last two categories had m-UGDS evaluated, with discordant results.
### Conclusions
CGM represents a more acceptable alternative for GDM diagnosis to the OGTT. CGM triangulation analysis suggests OGTT screening may result in both false positives and negatives. Further research including larger cohorts of patients, and additional triangulation elements (such as GDM biomarkers/outcomes and expanded m-UGDS) is needed to explore CGM potential for GDM diagnosis.
### Supplementary Information
The online version contains supplementary material available at 10.1186/s12884-023-05496-7.
## Background
Gestational Diabetes Mellitus (GDM) is a major public health issue, with steeply increasing incidence in the last decades due to a combination of maternal and environmental factors as well as changes in diagnostic strategies [1, 2]. Maternal and neonatal outcomes continue to be deeply impacted by this condition in both the short and long term, contributing to the current obesity and type 2 diabetes (T2DM) pandemic [2–4].
Although diagnostic thresholds and exact methods have varied widely over time, among different countries and even among different organizations within the same country, the current ‘gold standard’ for GDM diagnosis is still the oral glucose tolerance test (OGTT). However, an extensive list of pre-analytical (encompassing pre-testing (including preparatory diet and time of fasting, glucose load, collection tubes) and physiological factors (e.g. hydration, stress levels), as well as analytical (e.g. traceability/bias) and post-analytical limitations (results reporting and interpretation) have been reported for the OGTT [1, 5].
A number of potential substitutes for the OGTT have been proposed in the literature although none has been yet reported as a sufficiently robust candidate [6]. Of promise is Continuous Glucose Monitoring (CGM), which allows for evaluation of interstitial glucose levels during up to 14 days of ordinary life, as opposed to a one-off response to an artificial glucose load, offering a completely new perspective on glucose homeostasis [7]. To date, most research on CGM use in pregnancy has been regarding the management of type 1 diabetes (T1DM) and GDM, with little data on its application to GDM diagnosis [1, 8–10]. Our previous work demonstrates good acceptability of CGM as a diagnostic test for GDM, and its potential to unmask OGTT misdiagnosis [11].
The main limitation to developing a CGM-based diagnostic test for GDM is the lack of a gold standard aside from the deeply flawed OGTT. A solution to this issue may be provided by the concept of triangulation, which consists in evaluating an object from different perspectives to identify overlapping areas which represent the base for a new definition of the object [12]. Therefore, the aim of this study was to trial the Freestyle Libre Pro 2 as a diagnostic test for GDM, assessing its acceptability as opposed to the OGTT for a general population of pregnant women, and triangulating its results with risk factors and sonographic features of GDM as well as with the OGTT results.
## Study design
This prospective cohort study was held in two metropolitan hospitals in Sydney between April 2021 and April 2022; delays occurred secondary to the Sydney lock-down in response to the Covid-19 pandemic, resulting in 3 months suspension of recruitment. Women enrolled in the antenatal care clinics of the two hospitals were eligible and were invited to participate in the study via SMS/phone calls. Exclusion criteria were pre-existing T1DM or T2DM, first/early second trimester diagnosis of GDM and mental illness precluding informed consent.
Interested women had the opportunity to clarify further details over the phone or during the first study appointment. After signing the consent form, the Freestyle Libre iPro 2 CGM monitor was applied on the back of the participant’s upper arm [13]. The application side was decided by each participant depending on preferred side for sleeping, writing and carrying heavy loads. Using the PRO version, participants were blinded to their CGM data and were then asked to download a free app to keep track of their diet and exercise sessions [14]. After 7 days of CGM wearing, each participant’s routine OGTT was undertaken by study staff at the recruitment site before having the sensor removed. The OGTT was performed using a 75 g glucose beverage and interpreted against the IADPSG criteria [15].
Prior to completion of their participation, women were requested to share their diet/physical activity diary from their app through an email, and to complete three questionnaires: one on their risk factors for GDM, one on the acceptability of OGTT and one on the acceptability of CGM (Additional files 1 and 2). Participants could find more information on the study and the link to the questionnaire on the study website www.cgm4gdm.net.
Regarding sample size, we aimed to recruit similar numbers of patients to our previous study using the Medtronic iPro 2, in order to compare pilot results between the two devices. Our previous study’s combined dropout/non-usable data rate was $39\%$, with reasons including poor compliance with food intake recording and finger pricks, as well as incomplete OGTT results [11]. Given that the Freestyle Libre PRO does not require finger pick calibration and that the OGTT was offered at the recruitment sites as part of this study, a lower dropout/data exclusion rate was anticipated. To allow data comparison, the recruitment goal was set at 100 women, accounting for a $20\%$ dropout/data exclusion rate. Rates of negative outcomes (macrosomia defined as birthweight > 4 kg, preterm delivery < 37 weeks’ gestation, respiratory distress and elective/emergency caesarean sections) in the results of the NGT group in this study were used together with their OR described in a recent meta-analysis in GDM women to calculate the sample size needed to explore their correlation with CGM data in future using online software package “Select Statistical Service” [16, 17].
In a previous study, our group developed a survey for extensive evaluation of well-established and recently proposed risk factors for GDM, including data from 21 participants of this current pilot study [12]. Questions were on ethnicity, BMI, medical history (obstetric inclusive) but also exercise and dietary patterns, season of conception and ART use. We additionally created surveys on OGTT and CGM consisting of five questions regarding the overall acceptability as well as acceptability of insertion, wearing and removal, and the likelihood of recommending CGM as a diagnostic test for other women in a Likert scale format of 0–5. A final free text box allowed participants to share any recommendation or comment. The survey on CGM acceptability is the same used in our previous Medtronic pilot study to allow comparison [11].
## Data collection and analysis
Data collection and synthesis were based on the protocols of our previous studies on the use of the Medtronic iPro2 for GDM diagnosis and the development of a questionnaire for GDM risk factors [11, 12]. Clinical data was obtained from the hospitals’ obstetric database (Additional file 3). Cases were followed until birth.
Data from the Freestyle Libre PRO was downloaded using the web-based software portal (LibreView, app) and exported for analysis [18]. Glycaemic reports generated for each patient in Microsoft Excel were individually considered to determine validity for analysis. Only CGM output with 96 measurements per day for seven days were considered valid and analysed. Daytime was considered from 06:00 am to 23:59 h and night-time from 00.00 am to 05:59am. The CGM parameters considered in our analysis are outlined in Table 1.Table 1Continuous glucose monitor parameters used for data analysisSigle – NameDefinition/cut-off (reference)MeanMean of blood glucose level registered at CGM [19]SD – Standard deviationDispersion of the dataset relative to its mean [19]CV – Coefficient variationMean corrected for SD (SD/Mean) [20]TIR – Time in range3.5–7.8 mmol/L [16]TBR – Time below range = 3.0 – 3.4 mmol/L, 2 = < 3.0 mmol/L [16]TAR – Time above range > 7.8 mmol/L [16], 2 = > 10 mmol/LMAGE – Mean amplitude of glycaemic excursionMeasure of intra-day glycaemic variability [19]MODD – Mean of daily differencesMeasure of inter-daily glycaemic variability [19] *Statistical analysis* was performed using Microsoft Excel (Microsoft, WA, USA) and SPSS (SPSS Inc, IL, USA). Normally distributed continuous variables are presented as mean ± standard deviation (SD); non-normally distributed continuous variables are presented as median with interquartile range. Continuous variables were compared between groups using t-test (normally distributed) and Mann Whitney U test (non-normally distributed) as appropriate. Categorical variables are presented as percentages and were compared using Chi-Square or Fisher's exact test as appropriate. Values of $p \leq 0.05$ were considered statistically significant. Due to the pilot/exploratory nature of the study no statistical adjustments for multiple comparisons were made.
## Triangulation
Triangulation may help to explore the issue when one is attempting to introduce a new measure that can only be compared against a current flawed gold-standard test. In order to not have to solely rely on OGTT as comparison for CGM data (combined in a score of variability (CGMSV)), we additionally triangulated CGMSV with risk factors, combined in a Total Risk Score (TRS) as already described in our previous publications [11, 12]. In a subgroup of women, triangulation also included the evaluation of ultrasound features of GDM, combined in a modified version of the Ultrasound Gestational Diabetes Score (m-UGDS) [21].
The CGMSV was calculated based both on the first three days and the complete 7 days period of CGM wearing. The parameters considered were of glucose levels’: (a) distribution: mean, SD, coefficient of variation; (b) variability: MAGE(Mean Amplitude of Glycaemic Excursion) for intra-day variability, and MODD (Mean of Daily Differences) for inter-day variability; (c) percentage of time spent in the range recommended for pregnant women (3.5–7.8 mmol/L) [7, 22]. These values were calculated on Excel after downloading raw data from the CGM system. MAGE and MODD were calculated using the Easy GV software [23]. CGMSV was calculated as a sum of the normalised values of mean, SD, CV, MAGE, MODD, TBR, TAR.
To calculate the TRS, each alternative response to the risk factors questionnaire was allocated a value based on the odds ratio (OR) for likelihood of development of GDM, with the baseline risk being 1 for each risk factor in its absence (i.e. 1 = baseline, risk factor not present). A total risk score was then calculated as the sum of the values (normalised against the baseline) recorded for each answer [12]. The cut-off value for CGMSV and TRS was estimated by finding the midpoint of the sum of the highest value in the NGT women and the lowest in the GDM women as already described in our previous publications [11, 12].
In a subgroup of women ($$n = 25$$) the triangulation analysis included a sonographic score of GDM, based on a modified protocol of the UGDS (m-UGDS), which was found to be a promising indicator of GDM in a recent systematic review published by our group, when assessed against the WHO ASSURED criteria (affordable, sensitive, specific, user-friendly, rapid and robust, equipment-free and deliverable to end-users) [6]. The ultrasound was performed by the study sonographer during or after the CGM monitoring period (24–28 weeks), excluding the day of the OGTT. The m-UGDS consisted of six parameters: fetal adipose subcutaneous tissue, asymmetrical macrosomia, cardiac circumference, cardiac width, interventricular septum thickness, immature appearance of placenta. We did not include the sonographic features of the UGDS that were less used in the recent years due to conflicting evidence in the literature, namely: breathing movements, placental thickness and immature placental appearance [24–26]. For Fetal Subcutaneous Adipose Thickness (SCAT), measures were taken from the inner edge of skin to the outer aspect of the echogenic subcutaneous fat surrounding the abdomen at the level of the fetal kidneys (as per Perovic et al.) and at level of the abdominal circumference, to then calculate the mean value and increase reproducibility [21]. All the other variables were measured as described in the original protocol, of which we also adopted the cut-offs values [21, 27]. The cut-off of the m-UGDS was set as > 3.
The maximum TRS in the NGT population was 0.82 and the minimum score in the GDM population was 0.56. Therefore, the cut off value was determined to be 0.69; those above this value was considered to be at high-risk of GDM. Similarly, the maximum CGMSV in the NGT population was 5.52 for 3 days and 5.54 for 7 days and the minimum score in the GDM population was 2.85 for 3 days and 3.41 for 7 days. Therefore, the cut off value for high-risk from CGMSV was determined to be 4.18 for 3 days and 4.47 for 7 days. Triangulation of TRS-OGTT with concordant CGMSV3 and CGMSV7 ($$n = 63$$/65 = $97\%$) is outlined in Fig. 2.Fig. 2OGTT, TRS and CGMSV TriangulationTRS = Total Risk Score, OGTT = Oral Glucose Tolerance Test, CGMSV = Continuous Glucose Monitoring Score of Variability, NGT = Normal Glucose tolerance Test, GDM = Gestational Diabetes Mellitus, TN = True Negative, FN = False Negative, FP = False Positive, TP = True Positive, m-UGDS = modified Ultrasound Gestational Diabetes Score. Perinatal data: macrosomia (> 4.5 kg), hypoglycaemia, respiratory distress. * one additional case suggested by CGMSV3 only Nine potential misdiagnoses of the OGTT were suggested by triangulating results of the CGMSV3 and CGMSV7 with TRS: five ‘false positive’ (positive OGTT but TRS and CGMSV $\frac{3}{7}$ all below the cut-off) and four’false negative’ diagnoses (negative OGTT with TRS and CGMSV $\frac{3}{7}$ all above the cut off). CGMSV3 suggested two additional misdiagnoses: one false positive (being below the cut-off in a GDM woman as opposed to CGMSV7) and one false negative (being above the cut-off in an NGT woman as opposed to CGMSV7).
Three of the twenty-five patients who underwent the ultrasound had an m-UGDS > 3 ($12\%$). Adding UGDS fortified the true negative diagnosis (4 cases confirmed as not having GDM features) but not the potential misdiagnosis suggested by CGMSV3 (m-UGDS discordant in $\frac{1}{6}$ patients considered false positive and $\frac{1}{5}$ considered false negative who had been scanned). The analysis of outcomes in terms of macrosomia, respiratory distress and hypoglycaemia was additionally discordant. None of the five women considered to be false negative and the one considered true positive had any of the considered outcomes, whereas one of the six women considered false positive had a macrosomic newborn.
Table 5 shows the difference in TRS and CGMSV as well as the CGM parameters described above in the ‘NGT by triangulation’ group (including the false positives OGTT as well as the true negatives) with’GDM by triangulation’ group (including the false negative OGTT and the true positive) when considering CGMSV3 for triangulation. Table 5TRS and CGM parameters of 3 and 7 days in NGT and GDM by triangulation NGT by triangulation($$n = 34$$)Median (IR) GDM by triangulation($$n = 6$$)Median (IR) p -value TRS 0.60 (0.07) 0.72 (0.07) < 0.001 CGMSV - 3 days 3.74 (0.63) 4.45 (0.23) < 0.001 - 7 days 3.99 (0.63) 4.66 (0.63) < 0.001 TIR - 3 days$40.71\%$ (16.69)$39.32\%$ (56.71)0.625 - 7 days$70.80\%$ (33.1)$70.1\%$ (40.70)0.571Mean ± SDMean ± SDMean - 3 days3.88 ± 0.404.12 ± 0.430.24 - 7 days 3.86 ± 0.90 4.34 ± 0.35 0.03 SD - 3 days 0.78 ± 0.11 1.04 ± 0.11 0.01 - 7 days 0.81 ± 0.11 1.06 ± 0.21 0.03 CV - 3 days0.20 ± 0.030.26 ± 0.050.06 - 7 days 0.20 ± 0.03 0.25 ± 0.06 0.103 MAGE - 3 days 1.87 ± 0.33 2.52 ± 0.34 0.01 - 7 days 1.93 ± 0.31 2.51 ± 0.45 0.02 MODD - 3 days0.76 ± 0.130.91 ± 0.160.07 - 7 days 0.73 ± 0.11 0.92 ± 0.13 0.02 TRS Total risk factors score, CGMSV Continuous glucose monitoring score of variability, TBR Time below range, TAR Time above range, SD Standard deviation, CV Coefficient variation, MAGE Mean amplitude of glycaemic excursion, MODD Mean of daily differences Women defined as NGT by triangulation had significantly lower TRS, CGMSV, SD and MAGE than those considered GDM, both when 3 and 7 days of CGM data were considered. No significant difference was found for TIR.
## Results
Of 107 women recruited to the study (Fig. 1), 87 were included ($81\%$) in data analysis. The most common reason for exclusion was CGM data recording period < 7 days ($$n = 11$$). Four cases had less than 6 days recorded and seven had less than $100\%$ coverage (96 readings) in the seventh day. Additionally, two cases had missing CGM data due to sensor misplacement. Fig. 1Consort diagramCGM = continuous glucose monitoring, TRS = total risk score, OGTT = oral glucose tolerance test, CGMSV = continuous glucose monitoring score of variability, m-UGDS = modified ultrasound gestational diabetes score Seventy-four participants ($85\%$) had Normal Glucose Tolerance as per the OGTT (NGT group) and 13 ($15\%$) were positive to the OGTT (GDM group). Triangulation was completed for 65 participants who completed the risk factor questionnaire enabling calculation of the TRS. Twenty-two of these patients also underwent an ultrasound for evaluating the m-UGDS. Perinatal outcomes were analysed for all included participants.
Maternal demographic characteristics are summarised in Table 2. Women classified as having GDM were significantly more likely to have a family history of diabetes mellitus ($54\%$ vs $23\%$, $$p \leq 0.03$$). All the OGTT values of time 0, 1 h and 2 h after the glucose load were significantly higher in the GDM group. Table 2Participant Demographic Characteristics NGT ($$n = 74$$) n (%) GDM ($$n = 13$$) n (%) p -value High Risk Backgrounda 18 ($24\%$)5 ($38\%$)0.23 Family History of DM 17 ($23\%$) 7 ($54\%$) 0.03 Previous macrosomia1 ($3\%$)1 ($8\%$)0.39Previous GDM3 ($4\%$)1 ($8\%$)0.48Primiparity40 ($54\%$)9 ($75\%$)0.15 Mean ± SD Mean ± SD Age32.4 (±) 4.832.1 (±) 2.90.82BMI22.9 (±) 4.823.0 (±) 5.00.98 OGTT time 0 (mmol/L) 4.3 (±) 0.3 4.6 (±) 0.6 0.01 OGTT 1 h (mmol/L) 6. 9 (±) 1.4 9.2 (±) 1.2 < 0.001 OGTT 2 h (mmol/L) 5. 7 (±) 1.2 8.1 (±) 1.7 < 0.001 GDM Gestational diabetes mellitus, NGT normal glucose tolerance, DM Diabetes Mellitus, SD Standard deviation, IR interquartile range, BMI Body Mass Index aHigh risk background = Southeast Asian, Chinese, Middle Eastern, Hispanic, South American, Aboriginal, Torres Strait Islander Perinatal outcomes are described in Table 3. No significant difference was found in terms of perinatal outcomes for mothers and newborns in women classified by the OGTT as having NGT versus GDM. Of the 13 women diagnosed with GDM, 8 were managed with diet only and five with medication (one with insulin, one with oral hypoglycaemic agents and three with both insulin and oral hypoglycaemic agents).Table 3Perinatal outcomes in NGT versus GDM NGT ($$n = 74$$) n (%) GDM ($$n = 13$$) n (%) p -value Macrosomia suspected4 ($5\%$)2 ($15\%$)0.22Induction of labour21 ($29\%$)5 ($39\%$)0.34Second degree tear14 ($19\%$)3 ($23\%$)0.49Caesarean Section: Elective21 ($29\%$)4 ($31\%$)0.56 Emergency3 ($4\%$)0 (%)0.61 Post-partum haemorrhage for atonic uterus8 ($11\%$)2 ($15\%$)0.46 Neonatal Respiratory distress5 ($7\%$)1 ($8\%$)0.64 Mean ± SD Mean ± SD Gestational Age at birth, weeks39.1 (± 1.3)39.6 (± 0.8)0.08Birth weight, kilograms3.46 (± 0.49)3.43(± 0.29)0.77Apgar 5 min8.9 (± 0.6)8.8 (± 0.6)0.63
## Acceptability and feasibility of OGTT and CGM
Women reported CGM to be significantly more acceptable than OGTT ($81\%$ vs $27\%$ $\frac{5}{5}$ general acceptability rate, $p \leq 0.001$). One participant had uncontrollable nausea and vomiting, which she had also experienced with OGTT during her previous pregnancy. Her OGTT had to be stopped after the first hour, with only the first two blood glucose values being considered for diagnostic purposes by the treating team.
In the free comments’ section of the questionnaire on CGM acceptability, the most frequently reported issue ($$n = 10$$) was difficulty with keeping track of diet and exercise due to the requested time commitment and malfunctioning of the app.
## TRS and CGM parameters
One outlier for TRS was identified and removed (not included in the analysis as per Fig. 1) for a patient with a score deemed extreme compared to the rest of the cohort. This was due to the patient having selected “6 + servings/ day” for beef consumption, driving up the OR for iron and total red meat serving [12].
In the total cohort, the difference between 7 and 3 days of CGM data was significant (all $p \leq 0.001$) for sensor mean (4.1 ± 0.4 vs 3.9 ± 0.4 mmol/L), max value (7.3 ± 0.9 vs 6.8 ± 0.1 mmol/L), TIR during the day (81.8 ± $14.4\%$ vs 72.5 ± $21.1\%$), and TIR at night (63.9 ± $26.5\%$ vs 54.5 ± $28.8\%$). CV and MODD were significantly but only slightly higher when considering 3 vs 7 days of CGM (0.21 ± 0.1 vs 0.22 ± 0.04 and 0.80 ± 0.16 vs 0.76 ± 0.1 respectively). The difference between 3 and 7 days of CGM monitoring was not significant for SD, Min Value and MAGE.
Table 4 illustrates the differences in terms of TRS and CGM parameters (with both 3 and 7 days of monitoring considered) between women classified as NGT and GDM. No statistically significant differences were found. Women in the GDM group had higher TRS, CGMSV, SD, CV, MAGE and MODD and lower TIR and mean glucose values, both when 3 and 7 days were considered. Table 4TRS and CGM parameters of 3 and 7 days in NGT versus GDM NGT ($$n = 74$$)Median (IQR) GDM ($$n = 13$$)Median (IQR) p -value TRS0.59 (0.69)0.61 (0.41)0.94CGMSV - 3 days3.83 (0.72)3.91 (0.78)0.95 - 7 days4.14 (0.65)4.33 (0.67)0.55TIR - 3 days$42.6\%$ (33.6)$31.2\%$ (10.5)0.13 - 7 days$75.8\%$ (34.6)$74.9\%$ (29.5)0.28 Mean ± SD Mean ± SD Mean - 3 days3.96 ± 0.453.90 ± 0.400.67 - 7 days4.16 ± 0.414.13 ± 0.430.85SD - 3 days0.85 ± 0.170.86 ± 0.140.84 - 7 days0.85 ± 0.160.91 ± 0.190.27CV - 3 days0.21 ± 0.050.22 ± 0.040.69 - 7 days0.20 ± 0.410.22 ± 0.460.24MAGE - 3 days2.00 ± 0.432.13 ± 0.360.27 - 7 days2.01 ± 0.392.20 ± 0.460.18MODD - 3 days0.80 ± 0.160.81 ± 0.140.95 - 7 days0.75 ± 0.130.81 ± 0.140.24 TRS Total risk factors score, CGMSV Continuous glucose monitoring score of variability, TBR Time below range, TAR Time above range, SD Standard deviation, CV Coefficient variation, MAGE Mean amplitude of glycaemic excursion, MODD Mean of daily differences
## Discussion
To the best of our knowledge, this is the first study to assess the Freestyle Libre PRO for GDM diagnosis based on but not exclusive to the OGTT results. As expected, GDM women were more likely to have family history of diabetes mellitus and higher OGTT values [1, 11]. No significant difference was found in terms of demographics and perinatal outcomes. This could be due to the small sample size of the GDM group, but also to the non-reliable classification of glycemic metabolism offered by the OGTT.
Triangulation of OGTT results with CGM data, combined in the CGMSV3, and a comprehensive list of risk factors (TRS), suggested eleven potential misdiagnoses of the OGTT. The results of previous studies demonstrate the potential for CGM to unmask OGTT misdiagnosis [8–10]. In a study by Tartaglione et al., 33 of 53 women classified as NGT with the OGTT were then found to have blood glucose levels above or below the recommended thresholds at CGM and managed with one week of self-blood glucose monitoring and diet [10]. Twelve of these women ended up requiring insulin [10]. As in our study, Tartaglione et al. found no difference in average daily glucose, time spent in the different ranges and maternal and fetal outcomes between GDM and NGT [10]. In 2009 Hijazi found dysglycaemia with CGM in 2 of 9 OGTT negative patients [8]. A study by Milln on 28 women (20 GDM, 8 controls) reported instead potential false positives of the OGTT, with CGM glucose variability of women classified as having GDM being not different from those having a negative OGTT result once at home [9].
Our group has conducted preliminary studies on more than 80 patients using the Medtronic iPro2 CGM device and in an initial cohort of twenty-one women recruited in this pilot study ($$n = 21$$) [11, 12]. In the Medtronic pilot study, CGM was found to be safe and acceptable by the recruited pregnant women, with CGM values correlating well with 1-h ($$p \leq 0.003$$) and 2-h OGTT values ($$p \leq 0.004$$), and uncovering glycaemic variability that OGTT could not detect [11]. However, some women complained of irritation due to the overlying tape on their already sensitive abdomen, whilst others commented that they would prefer not to have daily finger pricking for calibration of the CGM device [11]. Our group proposed Abbott’s Freestyle Libre 2 CGM device to be more tolerable for pregnant patients, being wearable on the arm and not requiring finger pricking for calibration. We therefore sought advice from the Australian TGA who subsequently approved use of the FreeStyle Libre PRO for this study. The Freestyle Libre 2 CGM was reported as highly acceptable for GDM diagnosis by participating women, significantly more than the OGTT and with increased acceptability compared to the Medtronic Ipro2 pilot study [11].
During our recruitment, the woman suspending the OGTT after 1 h due to uncontrollable nausea and vomiting underscored the low acceptability of the OGTT deeply impacting completion rates, as previously reported in an Australian study [28].
The main disadvantage of our protocol was identified by women as having to keep track of diet and physical activity. Women also stated that they would have preferred a shorter CGM wearing period of three days. For this reason, a comparison between the first 3 days and the total period of CGM data was performed, showing significant differences for some CGM parameters only with contrasting results (e.g. higher distribution parameters (mean, max value) but lower variability (CV and MODD) and higher time in range (both daytime and night-time). No difference was found for the remaining parameters of distribution (min value, SD) and variability (MAGE). The variation in CGM3 and CGM7 parameters regarding OGTT diagnoses of GDM or NGT was similar.
The concept of triangulation is based on observing a phenomenon from different perspectives to fully comprehend it, adding a new frame of reference to consolidate the evaluation [29]. Triangulation with both well-established (e.g. family history of diabetes mellitus, age, BMI) and newly identified risk factors (diet composition, physical activity, season of conception, use of assisted reproductive technologies) in our cohort suggested OGTT misdiagnosis. This confirms the findings of our recent study on the development of an online questionnaire to recruit women at high and low risk of developing GDM, where triangulation analysis suggested six ($13\%$) misdiagnoses (one false positive and five false negative cases) when both TRS and CGMSV resulted discordant with OGTT [12].
Considering 3 versus 7 days of CGM data resulted in conflicting differences regarding distribution/variability/time in range parameters, (e.g. better distribution but worse variability) suggesting that neither of the two timeframe performs better than the other in identifying a clear pattern of good/poor glycaemic control. This is reflected by the fact that at the triangulation analysis the results of CGMSV3 and CGMSV7 were concordant in $97\%$ of the cases.
The additional two misdiagnosis cases (one false positive and one false negative) suggested by CGMSV3 compared to CGMSV7 favour its use for an initial screening phase. Evaluation of TRS and CGM data differences between women considered as NGT (true negatives and false positives) versus those considered GDM (true positives and false negatives) with triangulation adopting CGMSV3 highlighted significantly higher TRS as well as distribution (SD) and variability (MAGE) parameters in the GDM group. This result underlines the potential of CGM and triangulation in classifying glucose dysmetabolism of new onset in pregnancy. Adopting 3 days of CGM monitoring as a first step for GDM screening could represent a good compromise to increase acceptability whilst retaining diagnostic ability.
In the subgroup of 25 women who underwent an ultrasound, m-UGDS reinforced the true negative diagnosis but contrasted with the triangulation in one case considered false positive and one case considered false negative.
## Strengths, limitations and future directions
This pilot study reinforces the potential role of CGM in unmasking OGTT misdiagnosis and introduces the role of triangulation in aiding development of a new GDM screening tool when OGTT remains the ‘gold-standard’. Patients found CGM to be acceptable for GDM diagnosis, although suggested that the protocol could improve with a multistage approach, not encompassing diet and physical activity tracking during the screening phase. Only CGM data with complete acquisition (96 readings a day for 7 days of monitoring period) was included in our analysis to maximise its accuracy. Data collected with this pilot study on diet and training sessions are not reported in this manuscript. Our group is currently working on automating lifestyle data analysis independent of and in correlation with CGM data to allow for a more comprehensive and expedited evaluation of glucose metabolism in the everyday setting.
Our modified ultrasound score (m-UGDS) was evaluated in a small subgroup only given the delayed recruitment due to the COVID-19 pandemic. The results of this analysis need to be verified in further studies:we hope to adopt this method in larger future cohorts to verify its usefulness for triangulation. Only neonatal hypoglycaemia, macrosomia and respiratory distress were evaluated in terms of perinatal outcomes, with limited impact on the triangulation. An extended and systematic evaluation of perinatal outcomes as well as biomarkers, potentially combined in a score, could improve the triangulation. All the cut-off scores used for triangulation were based on the maximum and minimum values observed in the NGT and GDM group of this pilot study, limiting the comparison with our previous studies. The expansion of data acquisition at a multicentre level could permit the development of cut-offs based on and applicable to different settings, allowing for more reliable comparison of results. Based on the OR for macrosomia (> 4 kg), respiratory distress, preterm delivery and elective/emergency caesarean section reported in a recent meta-analysis for GDM women not using insulin, considering a relative precision of $50\%$, confidence level of $95\%$, and the rates of these outcomes resulted in this pilot study, a minimum sample size of 243 is required to explore the correlation of CGM data with at least one of these outcomes (caesarean section) (Additional file 4). To examine all these perinatal outcomes, at least 1041 patients is required [16].
The recent adaptation from WHO ASSURED to RE-ASSURED criteria, including now ‘Real-time-connectivity’, and ‘Ease-of-specimen-collection’ underlines the importance of investing in and expanding the promising potential of CGM as a screening test for GDM [30]. CGM fits well with both of these newly adapted criteria, and could allow for a more minimally invasive, remotely visualised and realistic picture of daily glycaemic control than the one represented by the OGTT.
## Conclusions
Freestyle Libre PRO 2 is an acceptable and feasible tool for CGM diagnosis. Future research on larger cohorts of patients considering additional biomarkers and multicentre-based scores is warranted to assess the use of CGM for the diagnosis of CGM on a broader scale and develop a triangulation system applicable to the general population of pregnant women in Australia. The realization of a multistage CGM diagnostic test for GDM could improve its acceptability and patients’ compliance as well as “inform in real-time, strengthen the efficiency of health care systems and improve patient outcomes” [30].
## Supplementary Information
Additional file 1. Additional file 2. Additional file 3. Additional file 4.
## References
1. Sweeting A, Wong J, Murphy HR, Ross GP. **A clinical update on gestational diabetes mellitus**. *Endocr Rev* (2022.0) **43** 763-793. DOI: 10.1210/endrev/bnac003
2. 2.Jonathan Shaw ST. Diabetes the silent pandemic and its impact on Australia. https://www.diabetesaustralia.com.au/wp-content/uploads/Diabetes-the-silent-pandemic-and-its-impact-on-Australia.pdf: Diabetes Australia.
3. 3.Diabetes Australia. Gestational Diabetes. https://www.diabetesaustralia.com.au/about-diabetes/gestational-diabetes/: Diabetes Australia; 2019.
4. 4.Diabetes Australia. Gestational Diabetes in Australia - Position statement. https://www.diabetesaustralia.com.au/wp-content/uploads/Gestational-Diabetes-in-Australia-Position-Statement-2020.pdf; 2020.
5. 5.Bogdanet D, O’Shea P, Lyons C, Shafat A, Dunne F. The oral glucose tolerance test-is it time for a change?-a literature review with an emphasis on pregnancy. J Clin Med. 2020;9(11):3451.
6. Di Filippo D, Wanniarachchi T, Wei D, Yang JJ, Mc Sweeney A, Havard A. **The diagnostic indicators of gestational diabetes mellitus from second trimester to birth: a systematic review**. *Clin Diabetes Endocrinol* (2021.0) **7** 19. DOI: 10.1186/s40842-021-00126-7
7. Battelino T, Danne T, Bergenstal RM, Amiel SA, Beck R, Biester T. **Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range**. *Diabetes Care* (2019.0) **42** 1593-1603. DOI: 10.2337/dci19-0028
8. 8.Hijazi SAB, H. Issa, B.G.,. Continuous glucose monitoring as a diagnostic tool in gestational diabetes. Diabet Med. 2010;27:169.
9. Milln JM, Walugembe E, Ssentayi S, Nkabura H, Jones AG, Nyirenda MJ. **Comparison of oral glucose tolerance test and ambulatory glycaemic profiles in pregnant women in Uganda with gestational diabetes using the FreeStyle Libre flash glucose monitoring system**. *BMC Pregnancy Childbirth* (2020.0) **20** 635. DOI: 10.1186/s12884-020-03325-9
10. Tartaglione L, di Stasio E, Sirico A, Di Leo M, Caputo S, Rizzi A. **Continuous glucose monitoring in women with normal OGTT in pregnancy**. *J Diabetes Res* (2021.0) **2021** 9987646. DOI: 10.1155/2021/9987646
11. Di Filippo D, Ahmadzai M, Chang MHY, Horgan K, Ong RM, Darling J. **Continuous glucose monitoring for the diagnosis of gestational diabetes mellitus: a pilot study**. *J Diabetes Res* (2022.0) **2022** 5142918. DOI: 10.1155/2022/5142918
12. Di Filippo D, Bell C, Chang MHY, Darling J, Henry A, Welsh A. **Development and evaluation of an online questionnaire to identify women at high and low risk of developing gestational diabetes mellitus**. *BMC Pregnancy Childbirth* (2022.0) **22** 321. DOI: 10.1186/s12884-022-04629-8
13. 13.Abbott. FreeStyle Libre Pro System https://provider.myfreestyle.com/freestyle-libre-pro-product.html2020.
14. 14.Easy diet diary. About research food diary https://support.easydietdiary.com/hc/en-us/articles/360036920592-About-Research-Food-Diary-Frequently-Asked-Questions: Easy diet diary.
15. 15.International Association of Diabetes Pregnancy Study Groups CP, Metzger BE, Gabbe SG, Persson B, Buchanan TA, Catalano PA, et al. International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care. 2010;33(3):676–82.
16. Ye W, Luo C, Huang J, Li C, Liu Z, Liu F. **Gestational diabetes mellitus and adverse pregnancy outcomes: systematic review and meta-analysis**. *BMJ* (2022.0) **377** e067946. DOI: 10.1136/bmj-2021-067946
17. 17.Select statistical services. Sample size calculator: odds ratio https://select-statistics.co.uk/calculators/sample-size-calculator-odds-ratio/.
18. 18.Abbott. FreeStyle Libre Pro https://www.freestyle.abbott/in-en/products/freestyle-libre-pro.html2021.
19. Wyckoff JA, Brown FM. **Time in range in pregnancy: is there a role?**. *Diabetes Spectr.* (2021.0) **34** 119-132. DOI: 10.2337/ds20-0103
20. 20.Gáborová M, Doničová V, Bačová I, Pallayová M, Bona M, Peregrim I, et al. Glycaemic variability and risk factors of pregnant women with and without gestational diabetes mellitus measured by continuous glucose monitoring. Int J Environ Res Public Health. 2021;18(7). 10.3390/ijerph18073402.
21. Perovic M, Garalejic E, Gojnic M, Arsic B, Pantic I, Bojovic DJ. **Sensitivity and specificity of ultrasonography as a screening tool for gestational diabetes mellitus**. *J Matern Fetal Neonatal Med* (2012.0) **25** 1348-1353. DOI: 10.3109/14767058.2011.634458
22. Yu W, Wu N, Li L, OuYang H, Qian M, Shen H. **A Review of research progress on glycemic variability and gestational diabetes**. *Diabetes Metab Syndr Obes* (2020.0) **13** 2729-2741. DOI: 10.2147/DMSO.S261486
23. 23.University of Oxford. EasyGV https://www.phc.ox.ac.uk/research/resources/easygv2011.
24. Goda MA, Telek T, Kovacs F. **Novel phonography-based measurement for fetal breathing movement in the third trimester**. *Sensors (Basel)* (2020.0) **21** 211. DOI: 10.3390/s21010211
25. Sun X, Shen J, Wang L. **Insights into the role of placenta thickness as a predictive marker of perinatal outcome**. *J Int Med Res* (2021.0) **49** 300060521990969. DOI: 10.1177/0300060521990969
26. Turowski G, Vogel M. **Re-view and view on maturation disorders in the placenta**. *APMIS* (2018.0) **126** 602-612. DOI: 10.1111/apm.12858
27. Patil V, Srinivas G, Ms S, Kiran Das S, Hiremath R, Shabadi N. **Diagnostic significance of ultrasonographic markers and score in detection of gestational diabetes mellitus in the indian subcontinent**. *Ultrasound Q* (2019.0) **37** 362-369. PMID: 31233438
28. Kirke AB, Atkinson D, Moore S, Sterry K, Singleton S, Roxburgh C. **Diabetes screening in pregnancy failing women in rural Western Australia: an audit of oral glucose tolerance test completion rates**. *Aust J Rural Health* (2019.0) **27** 64-69. DOI: 10.1111/ajr.12465
29. 29.Flick. An Introduction to Qualitative Research: Sage Publications, London. ; 2010.
30. Land KJ, Boeras DI, Chen X-S, Ramsay AR, Peeling RW. **REASSURED diagnostics to inform disease control strategies, strengthen health systems and improve patient outcomes**. *Nat Microbiol* (2019.0) **4** 46-54. DOI: 10.1038/s41564-018-0295-3
|
---
title: Deoxyguanosine kinase mutation F180S is associated with a lean phenotype in
mice
authors:
- Cédric Francis Borreguero
- Stephan Wueest
- Constanze Hantel
- Holger Schneider
- Daniel Konrad
- Felix Beuschlein
- Ariadni Spyroglou
journal: International Journal of Obesity (2005)
year: 2023
pmcid: PMC10023562
doi: 10.1038/s41366-023-01262-z
license: CC BY 4.0
---
# Deoxyguanosine kinase mutation F180S is associated with a lean phenotype in mice
## Abstract
### Background
Deoxyguanosine kinase (DGUOK) deficiency is one of the genetic causes of mitochondrial DNA depletion syndrome (MDDS) in humans, leading to the hepatocerebral or the isolated hepatic form of MDDS. Mouse models are helpful tools for the improvement of understanding of the pathophysiology of diseases and offer the opportunity to examine new therapeutic options.
### Methods
Herein, we describe the generation and metabolic characterization of a mouse line carrying a homozygous DguokF180S/F180S mutation derived from an N-ethyl-N-nitrosourea-mutagenesis screen. Energy expenditure (EE), oxygen consumption (VO2) and carbon dioxide production (VCO2) were assessed in metabolic cages. LC-MS/MS was used to quantify plasma adrenal steroids. Plasma insulin and leptin levels were quantified with commercially available assay kits.
### Results
Mutant animals displayed significantly lower body weights and reduced inguinal fat pad mass, in comparison to unaffected littermates. Biochemically, they were characterized by significantly lower blood glucose levels, accompanied by significantly lower insulin, total cholesterol, high density lipoprotein and triglyceride levels. They also displayed an almost 2-fold increase in transaminases. Moreover, absolute EE was comparable in mutant and control mice, but EE in mutants was uncoupled from their body weights. Histological examination of inguinal white adipose tissue (WAT) revealed adipocytes with multilocular fat droplets reminiscent of WAT browning. In addition, mRNA and protein expression of Ucp1 was increased. Mutant mice also presented differing mitochondrial DNA content in various tissues and altered metabolic activity in mitochondria, but no further phenotypical or behavioral abnormalities. Preliminary data imply normal survival of DguokF180S/F180S mutant animals.
### Conclusion
Taken together, DGUOK mutation F180S leads to a lean phenotype, with lower glucose, insulin, and lipid levels rendering this mouse model not only useful for the study of MDDS forms but also for deciphering mechanisms resulting in a lean phenotype.
## Introduction
Mitochondrial DNA encodes a restricted number of genes including subunits of the triphosphate adenosine (ATP) synthase, cytochrome oxidase and NADH dehydrogenase, all involved in the oxidative phosphorylation and ATP production [1]. Mitochondrial DNA synthesis, itself, relies on nuclear genes. De novo synthesis of deoxyribonucleotides (dNTPs) occurs during nuclear DNA replication, but in resting cells, dNTP supply depends on a rescue pathway, with sequential phosphorylation of deoxyribonucleotides catalyzed by thymidine kinase 1 (TK1) and deoxycytidine kinase (DCK), acting in the cytosol and thymidine kinase 2 (TK2) and deoxyguanosine kinase (DGK or DGUOK), localized in the mitochondria [2]. When the activity of these kinases is reduced or absent, the limited availability of dNTPs leads to mitochondrial DNA depletion syndromes (MDDS). While this group of autosomal recessive disorders has as common root with the reduction of mitochondrial DNA, clinically they present with a wide range of symptoms caused by any combination of hepatopathy, myopathy and encephalopathy, also depending on the specific genetic alteration [3].
Along the same line, DGUOK mutations can present in two different forms: one characterized by neonatal-onset hepatopathy and encephalopathy and one by isolated liver disease. In the early-onset form, symptoms appear within the first weeks of life with lactic acidosis and hypoglycemia and progressive liver failure with hepatomegaly, elevated transaminases, cholestasis, and jaundice along with hypotonia, nystagmus and psychomotor retardation [4]. The prognosis of this form is poor, with the majority of affected individuals dying before the age of four [5]. Patients affected by the later-onset isolated form of hepatic disease present in infancy with progressive liver failure with an overall less severe phenotype. However, at later stages patients can develop hepatocellular carcinoma [6, 7]. Remission of hepatic disease has been reported in a single case [8, 9]. Currently, no satisfactory therapy is available for this MDDS. Liver transplantation remains an option but does not seem to improve survival in patients with liver failure and should be considered only in the absence of neurological features [5, 10]. Several DGUOK mutations have been identified in patients with hepatocerebral or isolated hepatic form of the disease. No clear genotype-phenotype correlation could be established; however, missense mutations seem to result into a later onset of symptoms and slower disease progression, in comparison to frameshift or non-sense mutations [11]. Furthermore, residual DGUOK activity seems to play an important role in disease progression [12]. Several frameshift or non-sense and missense mutations (e.g., M1I, M1V, S52F, E44K, K51Q, R105*, S107P, E165V, Q170R, W178X, Y191C, H226R, L248P, L250S, F256*) have been reported in patients with DGUOK deficiency [7, 11, 13–16], addressing the need for various in vitro and in vivo models to improve understanding of the varying clinical characteristics of this disease.
Murine models serve as a helpful tool to understand pathophysiology of human disease, allow detailed phenotyping, and can provide the basis for the development of new therapeutic strategies. So far, two Dguok animal models have been described: a complete Dguok knock-out (KO) rat model which did not display a pathological phenotype [17] and a complete murine Dguok KO model, indicative of a hepatic phenotype [18]. Herein, we report the generation and metabolic characterization of a Dguok mutant mouse line, carrying a phenylalanine (F) to serine (S) substitution of residue 180 (DguokF180S/F180S), leading to a mild form of MDDS.
## Generation of the DguokF180S/F180S mouse line
Our previous research has focused on the identification of mouse lines with hyperaldosteronism using an N-ethyl-N-nitrosourea (ENU)-mutagenesis screen. This alkylating agent induces point mutations in the spermatogonial stem cells of treated mice. Following this procedure, among others, we established a mouse line of familial hyperaldosteronism, carrying four different point mutations all located on chromosome 6 [19, 20]. By applying an iterative breeding strategy, we were able to segregate the potential candidate genes and established four different mouse lines, each one carrying only one single point mutation. For the DguokF180S/F180S (c.539 A > G) mouse line described herein, LC-MS/MS was used to quantify plasma adrenal steroids and, thereby, no significantly elevated aldosterone values could be documented in mutant animals (Supplementary Fig. 1). While the mouse line carrying this specific Dguok mutation was not the appropriate for pursuing the aldosterone driven phenotype, we observed significant weight alterations in genetically affected animals, deserving further metabolic characterization.
More specifically, for the generation of this mouse line, sperm of the previously reported mouse line, archived at the European Mouse Mutant Archive – EMMA, was used for embryo transfer [19, 20]. Revitalized mice, with C3HeB/FeJ background, were imported from Munich Helmholtz Center, Institute for Experimental Genetics. Heterozygous mice carrying the DguokF180S/F180S mutation were cross-bred to obtain homozygous mice, together with wild types, used herein as controls, and heterozygous animals, from the same breeding.
## Genotyping
Genotyping of the DguokF180S/F180S mouse line was performed with RT-qPCR probes containing Locked Nucleic Acids (LNA), designed to be specific for either the wild-type (WT) or the mutant sequence (TIB Molbiol, Berlin, Germany, Supplementary Table 1). Genotyping was also confirmed with Sanger sequencing of selected samples, performed at Microsynth AG, Balgach, Switzerland).
## Blood and organ sampling
Mice were housed in groups of 2–5 in IVC cages in a controlled environment (20 °C) on a 12 h light/dark photoperiod, in the Laboratory Animal Service Center of the University of Zurich. All animal experiments were approved by the Zurich cantonal authorities (License number $\frac{090}{2019}$). Animals were fed normal chow diet (KLIBA NAFAG 3436: $18.5\%$ protein, $4.5\%$ fat, $4.5\%$ fiber, $6.5\%$ ash, $54\%$ NFE) ad libitum and had free access to water. To minimize the effect of the circadian rhythm on hormones level, all mice were euthanized between 9 h and 11 h AM. Mice were killed by decapitation under isoflurane anesthesia at the age of 16 weeks. Trunk blood was collected in empty tubes for serum measurements or EDTA coated tubes for plasma measurements. The blood was then centrifuged at 10.000 g for 10 min at room temperature and the supernatant was separated and directly frozen at −20 °C.
Brain, liver, and bilateral inguinal fat pads, subscapular brown adipose tissue (BAT) and adrenal glands were prepared and resected. Adrenal glands were cleaned from adjacent fat tissue. Livers and right sided bilateral organs were stored in paraformaldehyde solution (PFA) for further histological analyses. Brains, livers and left sided bilateral organs were snap frozen in liquid nitrogen and stored at −80 °C for further uses.
## Glucose, lipid, electrolytes, and hormonal measurements
Blood glucose levels were determined using the Accu-Chek Aviva glucose meter (Roche, Basel, Switzerland). Serum electrolytes (sodium, potassium and chloride were quantified with the Stat Profile Prime Electrolyte Analyzer (Nova Biomedical, Zürich, Switzerland). Total cholesterol, high density lipoprotein, triglycerides were measured with the AU480 Clinical Chemistry System (Beckman Coulter, Indianapolis, USA). Plasma insulin and leptin levels were quantified with following commercially available assay kits: Mouse insulin ELISA Kit (Mercodia, Uppsala, Sweden), Mouse leptin ELISA kit (Sigma-Aldrich, St. Louis, MO, USA), according to the respective manufacturer’s instructions.
Liquid Chromatography/Mass Spectrometry (LC/MS) for adrenal steroids was performed in EDTA plasma samples in collaboration with the University Hospital Carl Gustav Carus at TU Dresden, Institute of Clinical Chemistry and Laboratory Medicine, Experimental Mass Spectrometry and Trace Elements Lab (Dr. Mirko Peitzsch). The method for analysis of the plasma steroid panel, including validation and assay performance characteristics has been described in detail elsewhere [21].
## Metabolic cage analysis
15-week-old male mice were placed individually in air-tight cages designed for metabolic phenotyping in an open-circuit indirect calorimetric system (PhenoMaster, TSE Systems, Bad Homburg, Germany) for four days as previously described [22]. The average of days 3 and 4 was used for data calculation, since body weight was stable during these two days. A total of 72 data points for food intake, O2 consumption, and CO2 production were recorded over both 24 h periods. Locomotor activity was measured using a 2-dimensional infrared light-beam. Energy expenditure (EE), oxygen consumption (VO2) and carbon dioxide production (VCO2) were calculated using the manufacturer’s software and values were additionally corrected for lean body mass (LBM). LBM was calculated according to manufacturer’s software as body weight raised to the power of 0.75. Body weight was daily updated in the software ensuring that EE data normalized to LBM were always taking into account the current body weight of the mice.
## Fecal bomb calorimetry
Two animals of the same genotype were housed together in a clean IVC cage. After 24 h, 1 g of feces were sampled and frozen at −20 °C until analysis. Samples were then dried under a ventilated hood overnight and placed in the decomposition vessel. The decomposition vessel was then placed in the calorimeter bomb and the samples were processed in adiabatic mode. This analysis was performed by the center of Phenogenomics of the École Polytechnique Fédérale in Lausanne.
## Real-time PCR
Whole organs were homogenized in the RNA lysis buffer (Zymo Research, Irvine, CA, USA) using a shaking homogenizer with ceramic beads (MP Biomedical INC, Illkirch, France). Upon 5 min centrifugation at 10.000 g the clean supernatant was used for RNA extraction. For adipose tissue, the upper phase containing fat was discarded and not used. Upon extraction, the concentration of RNA was measured with NanoDrop One UV Spectrophotometer (ThermoFisher Scientific, Waltham, MA, USA). 2 µg of RNA was converted to cDNA using High-Capacity cDNA Reverse Transcription kit with RNase Inhibitor (Applied Biosystems, Waltham, MA, USA). The cDNA was finally diluted to a concentration of 5 ng/μL. The SsoFast EVAGreen Supermix Mastermix was used to quantify the investigated genes. For the gene expression analysis, 10 ng of cDNA was pipetted to the Mastermix. The Cycle program in the Quant 5 was 5’ @ 95 °C then 40 cycles with following conditions: 15”@95 °C and 30”@60 °C. To ensure the absence of unspecific product, a melting curve analysis between 55 °C and 95 °C was also performed at the end. Quantification of gene expression was adjusted using the Tbp mouse gene expression for the investigated adipose tissues and using Gapdh as housekeeping gene for all other tissues.
## Western blotting
Acrylamide gels were cast one day before running the gel or at the same day using standard procedures. Whole organs were homogenized using a shaking homogenizer with 4 mm ceramic beads in RIPA Buffer with protease inhibitor (cOmplete mini, Roche, Basel, Switzerland) and then centrifuged. The protein concentration was measured with the Pierce BCA Protein Assay kit (ThermoFisher, Waltham, MA, USA) and the absorption measured at 565 nm. Protein was then dissolved to the desired concentration using RIPA buffer with Proteinase inhibitor and 2x loading buffer with DTT (Roche, Basel, Switzerland). For the inguinal fat pad 30 μg of protein were loaded. For the other organs, 7.5 μg of protein were loaded. Proteins were transferred to a Nitrocellulose membrane (GE Healthcare, Chicago, IL, USA) with a Blot transfer machine (Bio-Rad Laboratories, Hercules, CA, USA). The membrane was then blocked with a blocking buffer containing $5\%$ dry milk. Next, the membrane was incubated overnight at 4 °C with the primary antibody (dilution 1:1000). The membrane was washed 3 times and then incubated 1 h at RT with the secondary antibody (dilution 1:10000). Finally, after further washing (3x), the membrane was incubated with the detection reagent (West Pico PLUS, ThermoFisher, Waltham, MA, USA) and measured with the Western Blot Imaging System (Fujifilm, Tokyo, Japan). Primary antibodies (UCP1 polyclonal antibody (PA5-29575, ThermoFisher, Waltham, MA, USA), GAPDH - D16H11 Rabbit mAb (Cell Signaling Technology, Danver, MA, USA)) were diluted at 1:1000, while secondary antibodies were used at 1:10000 concentration (Rabbit HRP linked, NA9340, Merck, Darmstadt, Germany).
## Histological analysis
Tissues remained in $4\%$ paraformaldehyde overnight and then were dehydrated, embedded in paraffin, sectioned, and stained with hematoxylin and eosin following standard protocols. Hematoxylin and eosin (H/E)-stained adrenal sections were examined with a light microscope using magnifications of ×40 and ×400. For immunohistochemical staining for Ki-67 paraffin-embedded sections were rehydrated, heated in EDTA 1 mM, pH 9.0, SDS $0.05\%$ in the microwave for antigen retrieval, blocked with $3\%$ H2O2 in methanol for 10 min, and incubated with blocking buffer for 15 min. Ki-67 was immunolocalized overnight at 4 °C by means of a rabbit monoclonal antibody (RM9106-s Thermofisher, MA, USA) in a dilution of 1:300 in blocking buffer. After rinsing for 15 min in PBS, SignalStain® Boost IHC Detection Reagent (HRP, Rabbit, CellSignal, MA, USA) and Sigma Fast DAB (Sigma, Munich, Germany) were used for visualization. Transmission electron microscopy of BAT of a DguokF180S/F180S mouse and an unaffected littermate was performed in collaboration with the Center for Microscopy and Image Analysis of the University of Zurich (https://www.zmb.uzh.ch/en.html).
## Statistical analysis
Sample size was calculated based on the body weight difference observed between WT and mutant animals of the same sex in a previous pilot experiment (with an alpha error of 0.05 and a power of 0.8) resulting in an n-number of 5 animals per genotype. The n-number of samples included in all experiments/results presented herein was at least five per genotype. Animals/samples were included in the analysis depending on their availability after breeding. No randomization was applied. The investigators were not blinded to the group allocation during the experiments. Statistical analysis was carried out with the Prism 3.02 (GraphPad Software). Statistical significance was determined using the unpaired t-test for normally distributed parameters and Mann–Whitney test for non-normally distributed parameters. To investigate body mass-dependence of EE, a regression-based analysis-of-covariance (ANCOVA) was performed as previously described [23, 24]. Statistical significance was denoted by asterisks in the figures as *$P \leq 0.05$, **$P \leq 0.01$ and ***$P \leq 0.001.$
## Significantly lower 11-deoxycorticosterone and corticosterone levels in DguokF180S/F180S male mice
As part of the characterization of the ENU-derived mouse lines, the DguokF180S/F180S line was examined for its adrenal steroid phenotype by LC-MS/MS. Thereby, no significant differences could be observed in the aldosterone levels of this mouse line (Supplementary Fig. 1, E and J). In contrast, male DguokF180S/F180S mutant mice displayed significantly lower 11-deoxycorticosterone and corticosterone levels than their unaffected littermates (Supplementary Fig. 1B, C).
## Reduced body weight and fat mass of DguokF180S/F180S mice
On gross observation, DguokF180S/F180S mutant animals of both sexes at the age of 16 weeks moved and behaved normally without obvious phenotypic alterations but displayed significantly lower body weights than their unaffected littermates (Fig. 1A, B). For all subsequent analyses we focused on male animals. In line with the previous finding, inguinal fat pads of DguokF180S/F180S male mutant animals were significantly lighter compared to controls (Fig. 1, C). While serum sodium, potassium and chloride levels did not differ between controls and Dguok mutant mice (data not shown), blood glucose levels were significantly reduced in DguokF180S/F180S mutants (Fig. 1D). Of note, the low blood glucose levels of mutant animals were accompanied by low insulin levels (Fig. 1E). Plasma leptin levels were significantly reduced in Dguok mutant animals (Fig. 1F). Finally, DguokF180S/F180S mutant animals presented significantly lower lipid levels, that is, lower total cholesterol, lower HDL cholesterol and lower triglyceride levels (Fig. 1G–I).Fig. 1Reduced body weight and fat mass of DguokF180S/F180S mice. Body weight of male animals (A), body weight of female animals (B), weight of the inguinal fat pad (C), blood glucose levels in male animals (D), insulin levels (E), plasma leptin levels (F), cholesterol (G), HDL (H) and triglycerides (I). WT: Wild type animals, MUT: DguokF180S/F180S, HDL High density lipoproteins. * $p \leq 0.05$, **$p \leq 0.01$, ***$p \leq 0.001$, ****$p \leq 0.0001$ (Student’s t test). Values are expressed as mean ± SEM.
## Mild hepatic impairment in DguokF180S/F180S mice
Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels of DguokF180S/F180S mutant animals were significantly increased in comparison to controls (Fig. 2A, B), however, not exceeding a two-fold increase. Liver weights did not differ between DguokF180S/F180S and control mice (Fig. 2C). H/E staining of the hepatic tissue demonstrated larger hepatocytes with larger nuclei in mutant mice, without signs of increased steatosis and mild to absent cholestasis (Fig. 2D, E). The Ki-67 expression in the liver was increased in Dguok mutant animals (Fig. 2F, G), possibly suggesting increased proliferation due to regenerative stress. Fig. 2Mild hepatic impairment in DguokF180S/F180S mice. Comparison of alanine aminotransferase (ALT) (A), aspartate aminotransferase (AST) levels (B) and of liver weights (C). H/E staining of hepatic tissue (D, E). Ki-67 expression (F, G). WT: Wild type animals, MUT: DguokF180S/F180S. *$p \leq 0.05$, **$p \leq 0.01$, ***$p \leq 0.001$, ****$p \leq 0.0001$ (Student’s t test). Values are expressed as mean ± SEM.
## Energy expenditure is uncoupled from body mass in Dguok F180S/F180S mice
In order to elaborate on the mechanisms contributing to blunted weight gain in mutant mice (Fig. 1A, B), male animals were placed in metabolic cage units. Thereby, mutant animals did not display any significant difference in their locomotor activity (Fig. 3A, F, K), or food consumption (Fig. 3E, J) compared to unaffected littermates during both the dark and the light phase. Similarly, mean respiratory exchange ratio (RER) did not differ between the two genotypes during dark and light phase (Fig. 3B, G). Still, when observing the RER values plotted over 24 h, an RER of almost 1 was observed towards the end of the light phase (when animals were still asleep) in mutant animals. Such RER of mutant mice was rapidly normalized back to the level of control animals at the beginning of the dark phase, when animals started eating again (Fig. 3L). However, when analyzing the different time points, a statistically significant difference could only be documented in the time frame 14–16 h between mutant and controls ($P \leq 0.05$, Fig. 3L). Furthermore, although DguokF180S/F180S mutant mice did not differ from controls in terms of absolute EE during both the light and dark phase (Fig. 3C, H), DguokF180S/F180S mutant mice demonstrated significantly higher EE during the dark phase when EE was normalized to lean body mass (LBM) (Fig. 3D, I, N). When analyzing body mass-dependence of EE using ANCOVA, control but not mutant animals displayed a significant positive correlation, leading to significantly different slopes of the regression lines (Fig. 3M). Finally, to exclude impaired intestinal nutrient absorption capacity as the cause of reduced body weight in mutant mice, nutrient absorption was assessed by fecal bomb calorimetry. As depicted in Fig. 3O, no significant difference was observed in excreted caloric loss between mutant and littermate mice. Fig. 3Energy expenditure is uncoupled from body mass in DguokF180S/F180S mice. Comparison of locomotor activity (A, F, K), RER (B, G, L), absolute energy expenditure (C, H), energy expenditure adjusted to LBM (D, I, N) and food consumption (E, J) in the dark (upper panels) and light phase (middle panels) and plotted over a period of 24 h (lower panels). Correlation of energy expenditure to body weight (M, black dots: WT, red dots: MUT). Energy excreted measured with fecal bomb calorimetry (O). WT: Wild type animals, MUT DguokF180S/F180S, RER Respiratory exchange ratio, EE energy expenditure. * $p \leq 0.05$ (Student’s t test). Values are expressed as mean ± SEM.
## Increased browning of inguinal WAT in DguokF180S/F180S mice
Increased EE may suggest activation of BAT or browning of WAT in mutant mice. We therefore performed histological analysis of inguinal WAT. As depicted in Fig. 4A, B, mutant mice showed smaller adipocytes, with reduced fat content and multilocular fat droplets reminiscent of WAT browning. Moreover, mutant mice exhibited significantly higher Ucp1 (Fig. 4C), but lower leptin (Fig. 4D) mRNA expression in inguinal fat pads. Similarly, UCP1 protein levels were higher in mutant mice (Fig. 4E, F). In contrast, UCP1 protein levels were similar in subscapular BAT of Dguok mutant and littermate mice (Supplementary Fig. 2A). In addition, brown adipocytes of subscapular BAT were smaller in Dguok mutant compared to littermate mice (Supplementary Fig. 2B, C).Fig. 4Increased browning of inguinal WAT.Hematoxylin/eosin (H/E) staining of inguinal fat pad of control littermate (A) and mutant mice (B). mRNA expression of Ucp1 (C) and leptin (D). Semiquantitative UCP1 to GAPDH protein expression (E). Representative western blot of UCP1 protein levels in inguinal WAT (F). WT: Wild type animals, MUT: DguokF180S/F180S, Ucp1: Uncoupling protein 1, GAPDH: Glyceraldehyde-3-Phosphate Dehydrogenase. * $p \leq 0.05$, **$p \leq 0.01$ (Student’s t test). Values are expressed as mean ± SEM.
## Altered mRNA expression of mitochondrial enzymes in DguokF180S/F180S mice
As DGUOK plays an important role in mitochondrial function, mice were characterized for possible mitochondrial alterations. Mitochondrial DNA quantification in various tissues/organs displayed significantly lower mitochondrial DNA levels in brain, liver, and adrenal glands, but significantly higher DNA level in BAT of DguokF180S/F180S mutant mice (Fig. 5A–D). We next determined mRNA expression of enzymes involved in the Krebs cycle and the electron transport chain. As depicted in Fig. 5E, expression of citrate synthase, which catalyzes the first step providing substrate to the Krebs cycle, was significantly suppressed in mutant animals. In contrast, expression of Idh1, Idh2 and Sdha, all three enzymes catabolizing different steps of the Krebs cycle were significantly higher in Dguok mutant animals (Fig. 5F–H). Similarly, all three enzymes involved in the NAD + biosynthetic pathway were significantly upregulated in Dguok mutant animals (Fig. 5I–K). Moreover, mRNA expression of ATP synthase Atp5b was significantly reduced in mutant mice (Fig. 5L).Fig. 5Altered mRNA expression of mitochondrial enzymes in DguokF180S/F180S mutant mice. Mitochondrial DNA quantity in brain (A), liver (B), adrenal gland (C) and BAT (D). mRNA expression of enzymes involved in the Krebs cycle (Citrate synthase (E), Idh1 (F), Idh2 (G), Sdha (H)) and in the electron transport chain (Nmnat1 (I), Nadsyn1 (J), Nampt (K)), mRNA expression of ATP synthase - Atp5b (L) in BAT. WT: Wild type animals, MUT: DguokF180S/F180S, BAT: Brown adipose tissue, Idh1: Isocitrate Dehydrogenase 1, Idh2: Isocitrate Dehydrogenase 2, Sdha: Succinate Dehydrogenase Complex Flavoprotein Subunit A, Nmnat: Nicotinamide mononucleotide adenylyltransferase 1, Nadsyn1: NAD Synthetase 1, Nampt: Nicotinamide Phosphoribosyltransferase, Atp5b: ATP synthase F1 subunit beta. * $p \leq 0.05$, **$p \leq 0.01$, ***$p \leq 0.001$, ****$p \leq 0.0001$ (Student’s t test). Values are expressed as mean ± SEM.
## Increased mRNA expression of deoxycytidine kinase (DCK) in liver but not BAT of Dguok F180S/F180S mice
Given the mild phenotype observed in this mouse line based on the absence of neurological or hepatic features, we hypothesized on possible compensatory mechanisms. As a rescue pathway for deoxyribonucleoside biosynthesis has been described in the literature, the activity of this pathway was investigated [2]. Specifically, it is appreciated that deoxycytidine kinase (DCK) is able to transform deoxyguanosine to deoxyguanosine monophosphate outside the mitochondria and overcome a relative or absolute DGK absence (Fig. 6A, red labeled pathway). However, in DguokF180S/F180S animals, although Dck was expressed at higher levels in the liver (Fig. 6B), this was not the case in BAT, with Dck expression levels being comparable between the genotypes, suggesting that this rescue pathway is inactive in BAT (Fig. 6C). The mRNA expression of the mitochondrial gene Nd1 in the liver, BAT (Fig. 6D, E) and in the adrenal gland (data not shown) did not differ between controls and DguokF180S/F180S mice, either. DguokF180S/F180S mutant animals did not display a significantly different number of mitochondria in their BAT, neither showed any pronounced differences in their size or microscopic structure (Fig. 6F, G).Fig. 6Increased mRNA expression of deoxycytidine kinase (DCK) in liver but not BAT of DguokF180S/F180S.A Simplified schematic representation of the deoxyribonucleotide rescue pathway, with dNDPs generated outside the mitochondria. mRNA expression of the nuclear-encoded gene deoxycytidine kinase (Dck) in the liver (B) and in BAT (C) and of the mitochondrial *Nd1* gene in the liver (D) and BAT (E) of DguokF180S/F180S mutant and control littermate mice. Electron microscopy image of BAT of control (F) and DguokF180S/F180S mutant (G) mouse (scale bar 3 µm). WT Wild type animals, MUT DguokF180S/F180S, BAT Brown adipose tissue, dGuo deoxyguanosine, dAde deoxyadenosine, dCyt deoxycytidine, dThd deoxythymidine, DCK deoxycytidine kinase, TK1 Thymidine kinase 1, dGMP deoxyguanosine monophosphate, dAMP: deoxyadenosine monophosphate, dCMP deoxycytidine monophosphate, dTMP deoxythymidine monophosphate, DGK deoxyguanosine kinase, TK2 Thymidine kinase 2, dNDP deoxyribonucleotide diphosphate, dNTP deoxyribonucleotide triphosphate. ** $p \leq 0.01$ (Student’s t test). Values are expressed as mean ± SEM.
## Discussion
In summary, the mouse model presented herein carries a homozygous Dguok point mutation (F180S) leading to an aberrant metabolic phenotype, characterized by reduced body weight and subcutaneous fat pads, while no other phenotypic abnormalities, such as hepatic or neurological involvement are present that typically characterize MDDS. The mild hepatic impairment documented biochemically and histologically did not seem to phenotypically affect the mice. Thereby, DguokF180S/F180S mice extend the spectrum of MDDS with the possibility to specifically characterize metabolic consequences of the disease.
In variance to currently described DGUOK animal models, the DguokF180S/F180S mutant mouse line provides a metabolic phenotype. Specifically, mutant animals had significantly lower body and inguinal fat pad weights, but no other gross phenotypical abnormalities. Furthermore, DguokF180S/F180S mice presented lower blood glucose values, which is a typical initial sign of MDDS due to Dguok deficiency in humans. However, Pronicka et al. describe an islet cell hyperplasia and a hyperinsulinemia in two patients with Dguok deficiency [13], and, similarly, a further case report presents a hyperinsulinemic hypoglycemia due to a homozygous DGUOK Phe256* mutation [25]. Unlike these cases, the DguokF180S/F180S animals presented with significantly lower insulin levels than their unaffected littermates, suggestive of a negative feedback mechanism to compensate for the lower glucose levels. In parallel, these mutant mice displayed low corticosterone values, originating from the adrenal glands, that also presented a very low mitochondrial DNA content. As adrenal steroidogenesis is partly dependent on mitochondrial function, Dguok deficiency could contribute to lack of adrenal counter-regulation thereby accentuating the reduction in blood glucose levels. The DguokF180S/F180S mouse line also displayed significantly lower cholesterol, HDL and triglyceride levels. The lipid profile of affected individuals with MDDS due to DGUOK deficiency has not been described in detail so far, but the Dguok KO mouse line, previously described, presents an opposed biochemical profile, with significantly increased cholesterol levels in the KO animals [18]. Furthermore, the DguokF180S/F180S mice displayed up to two-fold increased transaminase levels (ALT and AST), suggestive of a mild hepatic impairment, whereas the Dguok KO mouse model, previously described, presents with a 4- to 5-fold increase in transaminases. Histological signs of hepatic damage are present at different extent in both mouse models [18].
DguokF180S/F180S mutant animals had normal or even increased intestinal absorption, excluding a malabsorption of nutrients as causative for their body weight phenotype. Additionally, they did not differ in their absolute EE from control animals. However, in DguokF180S/F180S animals, EE appeared uncoupled from their body mass. The latter may be due to increased browning of white adipose tissue in DguokF180S/F180S animals, as reflected by increased UCP1 protein and mRNA levels in inguinal WAT as well as the appearance of multilocular fat cells on histological examination. These mice did not display any neurological abnormalities, possibly due to the lower but maintained at >$50\%$ mitochondrial DNA content in the brain. The fact that they presented a mild hepatic pathology, despite the almost nonexistent mitochondrial DNA in their liver can potentially be explained by the increased hepatic expression of deoxycytidine kinase, the key enzyme for the rescue pathway for dNTP synthesis. Another compensatory mechanism is suggested since we observed an unaffected mRNA expression of mitochondrial genes such as Nd1 in DguokF180S/F180S animals, despite the low mitochondrial DNA content in this tissue. Surprisingly, these animals present increased mitochondrial DNA in the BAT, with an apparently intact mRNA expression of mitochondrial genes, such as Nd1 in this tissue. It has been previously acknowledged, that mitochondria from different cell types are functionally unique, depending on their respective nuclear background, to address the needs of different cells and BAT is recognized as a tissue with high mitochondrial concentration [26, 27]. Whether the necessity for an increased mitochondrial DNA replication in this tissue serves as a mechanism to escape the mutational effect remains unclear. In line with this concept, electron microscopy revealed no pronounced differences in number, size or structure of mitochondria in the BAT of mutant animals. Still, in spite of the increased mitochondrial DNA in BAT, the expression of various enzymes involved in the Krebs cycle and electron chain transport presented alterations, suggestive of a reduced flow of substrates in the Krebs cycle and a compensatory increased catalyzing of intermediate products. The current mouse model has some phenotypic overlap with the previously described Dguok KO model, that is characterized by low body weight and decreased subcutaneous fat layer [18]. Furthermore, this mouse line also presents an altered expression of the enzymes of the Krebs cycle [28].
DguokF180S/F180S mice have not yet been systematically observed for the assessment of their life span. According to our preliminary observation, these animals survived up to 30 weeks without further apparent phenotypical or behavioral abnormalities, in line with the preliminary survival estimates of the murine Dguok KO mouse line [18, 28]. The differential mitochondrial DNA content in various tissues might play a role in the lean phenotype presented herein, and it seems, that the DguokF180S/F180S mouse line also possesses sufficient compensatory pathways ensuring sufficient mitochondrial DNA levels, that do not further influence their phenotypical appearance and survival.
In MDDS due to DGUOK mutations, the phenotypical abnormalities, the time of onset and the course of the disease present with a large variety among genetically affected patients. Both human and mouse *Dguok* genes contain 277 amino acids and present homology with $75\%$ identities and $85\%$ positive residues [29]. The position F180 is well conserved among species (Supplementary Fig. 3). In the case of F180S substitution, the non-polar, hydrophobic phenylalanine is replaced by a polar and hydrophilic serine, affecting helix propensity, and causing structural changes in the predicted three-dimensional structure of the protein (Supplementary Fig. 4). The previously described W178X mutation in close proximity to the F180S mutation, is associated with a severe and lethal hepato-cerebral form of MDDS in the affected individual [11]. In contrast, as the F180S mutation originates from an ENU mutagenesis screen, this setting might have favored the milder phenotype described herein.
Taken together, we herein describe the generation and metabolic characterization of a DguokF180S/F180S mutant mouse line, that displays a lean phenotype, with reduced subcutaneous fat pads, characteristics of WAT browning, and increased EE. Furthermore, mutant animals are characterized by lower blood glucose, insulin, and lipid levels. This mouse line presents differential mitochondrial DNA quantities in various tissues and altered metabolic function in the mitochondria, but no further phenotypical abnormalities observed in MDDS forms. These data are suggestive of the presence of compensatory mechanisms in the context of this specific mutation, ensuring sufficient mitochondrial DNA levels that do not further influence phenotype and survival. This mouse model could serve in the future not only in the study of MDDS forms but also in the understanding of mechanisms resulting in a lean phenotype.
## Supplementary information
Supplemental Material The online version contains supplementary material available at 10.1038/s41366-023-01262-z.
## References
1. Barchiesi A, Vascotto C. **Transcription, Processing, and Decay of Mitochondrial RNA in Health and Disease**. *Int J Mol Sci* (2019.0) **20** 2221. DOI: 10.3390/ijms20092221
2. Camara Y, Gonzalez-Vioque E, Scarpelli M, Torres-Torronteras J, Marti R. **Feeding the deoxyribonucleoside salvage pathway to rescue mitochondrial DNA**. *Drug Discov Today* (2013.0) **18** 950-7. DOI: 10.1016/j.drudis.2013.06.009
3. Elpeleg O. **Inherited mitochondrial DNA depletion**. *Pediatr Res* (2003.0) **54** 153-9. DOI: 10.1203/01.PDR.0000072796.25097.A5
4. 4.El-Hattab AW, Scaglia F, Wong LJ Deoxyguanosine Kinase Deficiency. In: Adam MP, Everman DB, Mirzaa GM, Pagon RA, Wallace SE, Bean LJH, et al., editors. GeneReviews((R)). Seattle (WA)1993.
5. Dimmock DP, Dunn JK, Feigenbaum A, Rupar A, Horvath R, Freisinger P. **Abnormal neurological features predict poor survival and should preclude liver transplantation in patients with deoxyguanosine kinase deficiency**. *Liver Transpl* (2008.0) **14** 1480-5. DOI: 10.1002/lt.21556
6. Scheers I, Bachy V, Stephenne X, Sokal EM. **Risk of hepatocellular carcinoma in liver mitochondrial respiratory chain disorders**. *J Pediatr* (2005.0) **146** 414-7. DOI: 10.1016/j.jpeds.2004.10.023
7. Freisinger P, Futterer N, Lankes E, Gempel K, Berger TM, Spalinger J. **Hepatocerebral mitochondrial DNA depletion syndrome caused by deoxyguanosine kinase (DGUOK) mutations**. *Arch Neurol* (2006.0) **63** 1129-34. DOI: 10.1001/archneur.63.8.1129
8. Mousson de Camaret B, Taanman JW, Padet S, Chassagne M, Mayencon M, Clerc-Renaud P. **Kinetic properties of mutant deoxyguanosine kinase in a case of reversible hepatic mtDNA depletion**. *Biochem J* (2007.0) **402** 377-85. DOI: 10.1042/BJ20060705
9. Ducluzeau PH, Lachaux A, Bouvier R, Duborjal H, Stepien G, Bozon D. **Progressive reversion of clinical and molecular phenotype in a child with liver mitochondrial DNA depletion**. *J Hepatol* (2002.0) **36** 698-703. DOI: 10.1016/S0168-8278(02)00021-1
10. Jankowska I, Czubkowski P, Rokicki D, Lipinski P, Piekutowska-Abramczuk D, Ciara E. **Acute liver failure due to DGUOK deficiency-is liver transplantation justified?**. *Clin Res Hepatol Gastroenterol* (2021.0) **45** 101408. DOI: 10.1016/j.clinre.2020.02.018
11. Dimmock DP, Zhang Q, Dionisi-Vici C, Carrozzo R, Shieh J, Tang LY. **Clinical and molecular features of mitochondrial DNA depletion due to mutations in deoxyguanosine kinase**. *Hum Mutat* (2008.0) **29** 330-1. DOI: 10.1002/humu.9519
12. Wang L, Eriksson S. **Mitochondrial deoxyguanosine kinase mutations and mitochondrial DNA depletion syndrome**. *FEBS Lett* (2003.0) **554** 319-22. DOI: 10.1016/S0014-5793(03)01181-5
13. Pronicka E, Weglewska-Jurkiewicz A, Taybert J, Pronicki M, Szymanska-Debinska T, Karkucinska-Wieckowska A. **Post mortem identification of deoxyguanosine kinase (DGUOK) gene mutations combined with impaired glucose homeostasis and iron overload features in four infants with severe progressive liver failure**. *J Appl Genet* (2011.0) **52** 61-6. DOI: 10.1007/s13353-010-0008-y
14. Nobre S, Grazina M, Silva F, Pinto C, Goncalves I, Diogo L. **Neonatal liver failure due to deoxyguanosine kinase deficiency**. *BMJ Case Rep* (2012.0) **2012** bcr1220115317. DOI: 10.1136/bcr.12.2011.5317
15. Waich S, Roscher A, Brunner-Krainz M, Cortina G, Kostl G, Feichtinger RG. **Severe Deoxyguanosine Kinase Deficiency in Austria: A 6-Patient Series**. *J Pediatr Gastroenterol Nutr* (2019.0) **68** e1-e6. DOI: 10.1097/MPG.0000000000002149
16. Shimura M, Kuranobu N, Ogawa-Tominaga M, Akiyama N, Sugiyama Y, Ebihara T. **Clinical and molecular basis of hepatocerebral mitochondrial DNA depletion syndrome in Japan: evaluation of outcomes after liver transplantation**. *Orphanet J Rare Dis* (2020.0) **15** 169. DOI: 10.1186/s13023-020-01441-5
17. Bennett B, Helbling D, Meng H, Jarzembowski J, Geurts AM, Friederich MW. **Potentially diagnostic electron paramagnetic resonance spectra elucidate the underlying mechanism of mitochondrial dysfunction in the deoxyguanosine kinase deficient rat model of a genetic mitochondrial DNA depletion syndrome**. *Free Radic Biol Med* (2016.0) **92** 141-51. DOI: 10.1016/j.freeradbiomed.2016.01.001
18. Zhou X, Curbo S, Zhao Q, Krishnan S, Kuiper R, Karlsson A. **Severe mtDNA depletion and dependency on catabolic lipid metabolism in DGUOK knockout mice**. *Hum Mol Genet* (2019.0) **28** 2874-84. DOI: 10.1093/hmg/ddz103
19. Perez-Rivas LG, Rhayem Y, Sabrautzki S, Hantel C, Rathkolb B, Hrabe de Angelis M. **Genetic characterization of a mouse line with primary aldosteronism**. *J Mol Endocrinol* (2017.0) **58** 67-78. DOI: 10.1530/JME-16-0200
20. Spyroglou A, Wagner S, Gomez-Sanchez C, Rathkolb B, Wolf E, Manolopoulou J. **Utilization of a mutagenesis screen to generate mouse models of hyperaldosteronism**. *Endocrinology.* (2011.0) **152** 326-31. DOI: 10.1210/en.2010-1081
21. Peitzsch M, Dekkers T, Haase M, Sweep FC, Quack I, Antoch G. **An LC-MS/MS method for steroid profiling during adrenal venous sampling for investigation of primary aldosteronism**. *J Steroid Biochem Mol Biol* (2015.0) **145** 75-84. DOI: 10.1016/j.jsbmb.2014.10.006
22. Lucchini FC, Wueest S, Challa TD, Item F, Modica S, Borsigova M. **ASK1 inhibits browning of white adipose tissue in obesity**. *Nat Commun* (2020.0) **11** 1642. DOI: 10.1038/s41467-020-15483-7
23. Tschop MH, Speakman JR, Arch JR, Auwerx J, Bruning JC, Chan L. **A guide to analysis of mouse energy metabolism**. *Nat Methods* (2011.0) **9** 57-63. DOI: 10.1038/nmeth.1806
24. Muller TD, Klingenspor M, Tschop MH. **Revisiting energy expenditure: how to correct mouse metabolic rate for body mass**. *Nat Metab* (2021.0) **3** 1134-6. DOI: 10.1038/s42255-021-00451-2
25. Arya VB, Dhawan A, Kapoor RR. **Hyperinsulinaemic hypoglycaemia in deoxyguanosine kinase deficiency**. *Clin Endocrinol (Oxf)* (2019.0) **91** 900-3. DOI: 10.1111/cen.14084
26. Herbers E, Kekalainen NJ, Hangas A, Pohjoismaki JL, Goffart S. **Tissue specific differences in mitochondrial DNA maintenance and expression**. *Mitochondrion.* (2019.0) **44** 85-92. DOI: 10.1016/j.mito.2018.01.004
27. Wilkins HM, Carl SM, Swerdlow RH. **Cytoplasmic hybrid (cybrid) cell lines as a practical model for mitochondriopathies**. *Redox Biol* (2014.0) **2** 619-31. DOI: 10.1016/j.redox.2014.03.006
28. Zhou X, Mikaeloff F, Curbo S, Zhao Q, Kuiper R, Vegvari A. **Coordinated pyruvate kinase activity is crucial for metabolic adaptation and cell survival during mitochondrial dysfunction**. *Hum Mol Genet* (2021.0) **30** 2012-26. DOI: 10.1093/hmg/ddab168
29. 29.HomoloGene:8456 [Internet]. Available from: https://www.ncbi.nlm.nih.gov/homologene?Db=homologene&Cmd=Retrieve&list_uids=8456.
|
---
title: 'Genetic regulation of body size and morphology in children: a twin study of
22 anthropometric traits'
authors:
- Karri Silventoinen
- José Maia
- Weilong Li
- Reijo Sund
- Élvio R. Gouveia
- António Antunes
- Gonçalo Marques
- Martine Thomis
- Aline Jelenkovic
- Jaakko Kaprio
- Duarte Freitas
journal: International Journal of Obesity (2005)
year: 2023
pmcid: PMC10023566
doi: 10.1038/s41366-023-01253-0
license: CC BY 4.0
---
# Genetic regulation of body size and morphology in children: a twin study of 22 anthropometric traits
## Abstract
### Background
Anthropometric measures show high heritability, and genetic correlations have been found between obesity-related traits. However, we lack a comprehensive analysis of the genetic background of human body morphology using detailed anthropometric measures.
### Methods
Height, weight, 7 skinfold thicknesses, 7 body circumferences and 4 body diameters (skeletal breaths) were measured in 214 pairs of twin children aged 3–18 years (87 monozygotic pairs) in the Autonomous Region of Madeira, Portugal. Factor analysis (Varimax rotation) was used to analyze the underlying structure of body physique. Genetic twin modeling was used to estimate genetic and environmental contributions to the variation and co-variation of the anthropometric traits.
### Results
Together, two factors explained $80\%$ of the variation of all 22 anthropometric traits in boys and $73\%$ in girls. Obesity measures (body mass index, skinfold thickness measures, as well as waist and hip circumferences) and limb circumferences loaded most strongly on the first factor, whereas height and body diameters loaded especially on the second factor. These factors as well as all anthropometric measures showed high heritability ($80\%$ or more for most of the traits), whereas the rest of the variation was explained by environmental factors not shared by co-twins. Obesity measures showed high genetic correlations (0.75–0.98). Height showed the highest genetic correlations with body diameter measures (0.58–0.76). Correlations between environmental factors not shared by co-twins were weaker than the genetic correlations but still substantial. The correlation patterns were roughly similar in boys and girls.
### Conclusions
Our results show high genetic correlations underlying the human body physique, suggesting that there are sets of genes widely affecting anthropometric traits. Better knowledge of these genetic variants can help to understand the development of obesity and other features of the human physique.
## Introduction
Anthropometric measures are the key method to assess a child’s nutrition and development [1]. While body mass index (BMI), waist circumference and skinfold thicknesses are important to measure excess energy intake [2], height [3], upper arm circumference [4] and chest circumference [5] provide important information on malnutrition. Genetic studies of anthropometric traits are important for understanding the factors behind physical development and can thus also provide new insight into the role of environmental factors. *The* genetics of height and BMI have been extensively studied in children using the classic twin design [6, 7]. Further, molecular genetic studies using mainly the genome-wide-association (GWA) design [8, 9] have identified thousands of loci affecting adult height and BMI which show strong genetic correlations with these traits over childhood and adolescence [10]. There are also genetic twin studies on other traits, such as waist circumference [11], skinfold thicknesses [12], and chest circumference [13], as well as head circumference and several other craniofacial measures [14]. Still, generally, less is known about the genetics of anthropometric traits other than height and BMI. Collectively, these studies highlight the importance of genetic factors behind the variation of anthropometric traits.
However, an area which is still poorly understood is how much these different anthropometric measures share common genetic variation. Previous twin studies have shown genetic correlations between BMI and waist circumference [11], as well as BMI and several skinfold thicknesses [15, 16]. Genetic correlations were also found in a family-pedigree study including detailed obesity and other anthropometric measures [17]. These results based on twin and family designs have been confirmed by a GWA study finding genetic correlations between childhood BMI and percentage of body fat as well as waist and hip circumferences in adulthood [18]. There can also be shared genetic background even between distinct body traits as demonstrated in a family-pedigree study finding genetic correlations between craniofacial traits and body composition [19]. *These* genetic correlations can reflect genetic pleiotropy going back to fetal development [20]. Further, genetic factors can affect the adipose tissue both directly [21] and indirectly through, for example, eating behavior [22], thus creating correlations between indicators of obesity.
Knowledge on the genetic correlations between anthropometric traits can provide insight into the genetic regulation and development of body morphology. This knowledge may also have practical implications since it can guide which traits are most informative to assess obesity. However, a limitation in the previous studies is that they include only a few traits and thus can only partly capture the complexity of the human body physique. In this study, we use a twin data set of children that includes 22 anthropometric measures providing detailed information on human body size and morphology. *Using* genetic twin modeling, we analyze how these traits are mutually correlated and how much they share common genetic variation.
## Data and methods
The data were derived from the Madeira Twin Study conducted in the Autonomous Region of Madeira, Portugal [23]. First, all public and private schools were contacted and asked if they had twins as students and inquired about their contact information. Together, 434 twin families were identified, and an invitation letter to participate in the study was sent to them. From these families, 216 families having twin children 3 to 18 years of age ($51\%$ girls) participated in a detailed clinical examination in the capital city of Funchal in 2007 and 2008. During the examination, the children gave a blood sample. Zygosity was assessed by the polymerase chain reaction (PCR) amplification of short tandem repeat analyzed with a commercially available panel (AmpFlSTR Identifiler kit) comprising 15 autosomal, codominant, unlinked loci and the sex-determining marker [24]. Among the twin pairs, 87 were monozygotic (MZ), 73 same-sex dizygotic (SSDZ) and 56 opposite-sex dizygotic (OSDZ) pairs. The twins themselves and/or their parents/legal guardians provided written informed consent. The Scientific Board of the University of Madeira approved the study protocol.
A team of six experienced researchers from the Laboratory of Growth and Development of the University of Madeira conducted detailed anthropometric measures based on a standardized protocol [25]. All measures were done in a swimsuit, without shoes and with jewelry removed. All one-sided measurements were taken on the left side of the body. Height was measured using a Harpenden wall-mounted stadiometer accurate to 1 mm (Holtain, UK). Body weight was measured on a balance-beam scale accurate to 0.1 kg (Scena Optima 760, UK). BMI was then calculated by dividing weight in kg by the square of height in meters (kg/m2). Four body diameters (biacromial, bicristal, humerus and femur) were measured with a spreading caliper with an accuracy of 1 mm (Siber-Hegner, GPM, Switzerland). Seven body circumferences (waist, hip, calf, thigh, upper arm, forearm and upper arm flexed) were measured with a flexible steel tape accurate to 1 mm (Holtain, UK). Seven skinfold thicknesses (triceps, biceps, subscapular, suprailiac, calf, front thigh and abdominal) were assessed using a skinfold caliper and recorded to the nearest 2 mm (Siber-Hegner, GPM, Switzerland). We did not have missing cases in our data. However, we removed a few outliers (1 measurement for femur diameter and calf circumference and 2 measurements for humerus diameter, as well as thigh and forearm circumferences) since otherwise they may have disproportionally affected heritability estimates.
We found that the distributions of waist and hip circumferences, BMI, weight and all seven skinfold thicknesses were skewed and thus used logarithmic transformation to normalize them. After this transformation, the distributions of all traits were roughly normal (the skewness parameters varied between 0.06 and 0.98). In our previous study reporting the heritability estimates of 10 of these 22 traits, we did not find systematic differences when comparing children younger and older than 12 years of age [26]. Thus, in this study, we decided to report the results for the whole age range to increase the statistical power. All traits were adjusted by age and age-squared separately in boys and girls using a linear regression model by Stata statistical package, version 17.0 for Windows (StataCorp, College Station, TX, USA). The linear regression model was also used for statistical testing after correcting the standard errors and confidence intervals (CI) by the cluster option for the lack of statistical independence of twins sampled as pairs [27].
We started the statistical modeling with a factor analysis using the Varimax rotation, which creates uncorrelated (orthogonal) factors, separately in boys and girls. The Eigenvalue statistics suggested a two-factor solution in boys and a three-factor solution in girls. However, in girls, the Eigenvalue for the third factor was only slightly over 1 (1.027), and the factor explained only $5\%$ of the total variance. Thus, we used the two-factor solution in both boys and girls to have comparable results. In this analysis, the first factor explained $67\%$ of the variation in boys and $53\%$ in girls whereas the second factor explained 13 and $20\%$ of the variation in boys and girls, respectively. These factor scores were estimated using the maximum likelihood estimator and then stored as additional variables for the genetic modeling. The factor analysis was conducted using the SPSS statistical software version 28.0 for Windows (IBM Corp, Armonk, NY, USA).
We continued the analyses using genetic twin modeling based on the principle that while MZ twins are virtually genetically identical at the gene sequence level, DZ twins share, on average, half of their segregating genes, as with ordinary siblings [28]. Since the underlying correlation structure between co-twins is known, it is possible to decompose trait variance into genetic and environmental components. *Additive* genetic variance (A; correlation 1 within MZ and 0.5 within DZ pairs) includes the effects of all loci affecting the trait. Shared environmental variance (C; correlation 1 within both MZ and DZ pairs) includes the effects of all environmental factors making co-twins similar. Unique environmental variation (E; correlation 0 within both MZ and DZ twins) includes the effects of all environmental factors making co-twins dissimilar including measurement error.
We started the genetic modeling with univariate models to find the best fitting model and calculate heritability estimates. Based on co-twin correlations (Supplementary Table 1), we selected the additive genetic/ shared environment/ unique environment (ACE) model as the baseline model. The model fit statistics are presented in Supplementary Table 2. The assumptions of twin modeling were first tested by comparing the fit of the ACE model to the saturated model, which does not make any assumptions but freely estimates all possible statistics. The fit of the ACE model was good; only 6 traits showed poorer fit as compared to the saturated model if using a conventional p-value of 0.05 and none of them was statistically significant if using the Bonferroni corrected p-value for 24 tests ($p \leq 0.002$). We did not find any evidence for a sex-specific genetic effect, which would be seen as a lower genetic correlation of OSDZ pairs than the 0.5 expected for SSDZ pairs. Additionally, we were able to eliminate the shared environmental component from the model without a statistically significant decrease in the model fit. Thus, we used the additive genetic/ unique environment (AE) model without the sex-specific genetic effect in further analyses; this model showed good fit when compared to the saturated model. However, for some of the traits, we found a decrease in the model fit if using the same estimates for boys and girls. Nevertheless, since this was the case for only a few traits, we presented the genetic modeling results for boys and girls together using the AE model and then compared them to the sex-specific results (see supplementary files).
Using univariate models, we first calculated the proportions of variation explained by additive genetic factors – i.e., (narrow sense) heritability estimates – and unique environmental factors. Then, we utilized bivariate Cholesky decomposition, a model-free method to decompose all variation and covariation in the data into uncorrelated latent factors [29]. This method was used to decompose the covariation between the anthropometric measures into genetic and environmental covariances. Standardizing these covariances provides us the estimates of additive genetic and unique environmental correlations. *The* genetic twin modeling was conducted using the OpenMx package, version 3.0.2, of R statistical software, estimating the parameters based on the linear structural equations methodology and using the maximum likelihood estimator [30].
## Results
Table 1 presents the descriptive statistics of all anthropometric traits by sex. Girls had thicker skinfolds than boys, whereas boys had broader humerus and femur diameters. Forearm circumference was larger in boys and thigh circumference in girls. Otherwise, the anthropometric measures were roughly similar in boys and girls. Table 1Means and standard deviations (SD) of anthropometric measures by sex. Boys ($$n = 210$$)Girls ($$n = 222$$)p-value of sex differenceMeanSDMeanSDWeight measuresWeight (kg)37.615.5037.514.950.987BMI (kg/m2)18.43.5918.83.760.326Skinfolds (mm)Triceps10.34.6712.95.34<0.0001Biceps6.53.608.54.1<0.0001Subscapular8.65.611.77.0<0.0001Suprailiac10.47.9713.48.900.002Calf10.15.1113.56.51<0.0001Front thigh15.57.0522.59.31<0.0001Abdominal12.88.9517.110.64<0.0001Circumferences (cm)Waist62.79.8061.99.500.506Hip73.912.0776.413.660.109Upper arm21.04.0921.14.120.906Upper arm flexed22.24.2621.84.080.367Forearm20.63.0019.72.930.011Thigh42.98.2445.59.160.011Calf28.64.7028.75.050.758Diameters (cm)Biacromial30.34.4230.03.960.585Bicristal22.03.2222.03.280.937Humerus5.70.785.30.59<0.0001Femur8.20.987.70.81<0.0001Height measures (cm)Height139.818.99138.117.160.427Sitting height74.48.9673.58.080.398 Figure 1 presents the correlation matrices between all anthropometric traits in boys (right triangular matrix) and girls (left triangular matrix); the $95\%$ CIs are available in Supplementary Table 3. The correlation structure was roughly similar in boys and girls; only 36 of these 231 correlations showed a p-value of sex-difference <0.05 (Supplementary Table 4), which can be because of multiple testing. BMI showed the highest correlations with body circumferences, but the correlations were also high with skinfold thicknesses and somewhat lower with body diameters. On the other hand, height and sitting height showed the highest correlations with body diameters, whereas weaker correlations were found with body circumferences, and they were lowest with skinfold thicknesses. Fig. 1Trait correlations between anthropometric measures in boys (right triangular matrix) and girls (left triangular matrix).SF skinfold, C circumference, D diameter.
We then conducted the factor analysis to obtain more insight into the correlation structure of anthropometric measures (Table 2). Obesity measures (BMI, skinfold thicknesses and waist and hip circumferences) and limb circumferences loaded strongly on the first factor, whereas height, sitting height and body diameters loaded strongly on the second factor. However, all anthropometric measures loaded positively on both factors, except height and sitting height showing only weak loadings on the first factor. Communalities were generally high ($80\%$ or more for most of the traits) showing that these two factors largely explained the variation of these anthropometric measures. The exceptions were the body diameters in boys and girls and some of the skinfold thicknesses in girls showing only moderate communalities (from 40 to $70\%$).Table 2Factor loadings and communalities of anthropometric measures using a two-factor solution in boys and girlsa. First factorSecond factorCommunalitiesBoysGirlsBoysGirlsBoysGirlsWeight measuresWeight0.6250.8180.7750.5740.9910.999BMI0.8300.9850.4800.1640.9190.996SkinfoldsTriceps0.9280.7960.1530.2250.8840.684Biceps0.8870.7340.1340.1420.8050.559Subscapular0.8760.7980.2880.1850.8510.672Suprailiac0.9060.8060.250.2640.8830.719Calf0.8770.6960.1450.3310.7910.593Front thigh0.8990.7260.1560.2780.8320.604Abdominal0.9150.7780.2190.2340.8850.660CircumferencesWaist0.7700.8760.5470.3090.8920.863Hip0.7020.8090.6630.5110.9320.916Upper arm0.7730.8560.4960.3120.8430.831Upper arm flexed0.7030.8320.5100.3260.7540.799Forearm0.5830.7710.6250.4420.7310.790Thigh0.7190.8280.5820.4140.8560.857Calf0.5700.7890.6470.4330.7430.811DiametersBiacromial0.2030.4760.7600.6030.6190.589Bicristal0.4030.5210.7090.6030.6650.636Humerus0.2490.4310.7790.4690.6690.406Femur0.4450.5870.6730.3620.6500.476Height measuresHeight−0.0060.0970.8920.9950.7960.999Sitting height0.0210.1580.8700.8430.7560.735aVarimax rotation is used.
Next, we conducted the univariate twin modeling for these factor scores and all anthropometric measures in boys and girls (Table 3). *Additive* genetic variation explained a major part of the variation of all the traits, and the heritability estimates were more than $80\%$ for most of them. The remaining variation was explained by unique environmental factors. In the sex-specific results, we found that the heritability estimates for most of the traits were somewhat higher in boys than in girls (Supplementary Table 5). The largest differences were found for body diameters; however, CIs were also wide in these sex-specific analyses. Table 3Additive genetic and unique environmental variance components of anthropometric measures and underlying factors in boys and girls. *Additive* genetic factorsUnique environmental factorsa$295\%$ confidence intervalse$295\%$ confidence intervalsLLULLLULWeight measuresWeight0.890.850.920.110.080.15BMI0.890.850.920.110.080.15SkinfoldsTriceps0.840.780.890.160.110.22Biceps0.760.680.830.240.170.32Subscapular0.870.810.910.130.090.19Suprailiac0.840.770.890.160.110.23Calf0.790.700.840.210.160.30Front thigh0.850.790.890.150.110.21Abdominal0.830.750.880.170.120.25CircumferencesWaist0.830.760.880.170.120.24Hip0.890.850.920.110.080.15Upper arm0.740.640.810.260.190.36Upper arm flexed0.670.560.760.330.240.44Forearm0.630.510.730.370.270.49Thigh0.750.660.820.250.180.34Calf0.800.720.860.200.140.28DiametersBiacromial0.820.750.870.180.130.25Bicristal0.790.710.850.210.150.29Humerus0.820.740.870.180.130.26Femur0.620.490.720.380.280.51Height measuresHeight0.920.880.940.080.060.12Sitting height0.910.880.940.090.060.12FactorsFirst factor0.880.820.910.120.090.18Second factor0.910.880.940.090.060.12 We continued the genetic modeling by analyzing genetic and environmental correlations between the anthropometric traits in boys and girls (Fig. 2; the $95\%$ CIs are available in Supplementary Table 6). *Additive* genetic correlations (right triangular matrix) were generally high and followed the same pattern as found in the trait correlations. Between the obesity-related traits (BMI, skinfold thicknesses and waist and hip circumferences), the genetic correlations varied between 0.72 and 0.98, indicating that 52 to $96\%$ of the genetic variation is shared between the obesity-related traits. Height and sitting height showed the highest genetic correlations with the body diameter measures, but they were lower than among the obesity measures, i.e., from 0.58 to 0.76, indicating that 34 to $58\%$ of the genetic variation is shared between these traits. The unique environmental correlations (left triangular matrix) were also substantial but remarkably lower than the additive genetic correlations. Fig. 2Additive genetic correlations (right triangular matrix) and unique environmental correlations (left triangular matrix) between anthropometric measures in boys and girls. SF skinfold, C circumference, D diameter.
Finally, we replicated the analyses in boys and girls to see whether there were any sex differences in these correlation patterns. Supplementary Fig. 1 presents the additive genetic correlations in boys (right triangular matrix) and girls (left triangular matrix); the $95\%$ CIs are available in Supplementary Table 7. We did not find any systematic differences between the genetic correlations in boys and girls. When analyzing the unique environmental correlations, no systematic sex differences were found either (Supplementary Fig. 2; the $95\%$ CIs are available in Supplementary Table 8).
## Discussion
In this comprehensive twin study of 22 anthropometric traits, we found that the same genetic factors underlined the different anthropometric traits traditionally used to measure obesity, namely, BMI, skinfold thicknesses and waist and hip circumferences. Based on the genetic correlations, we estimated that from half to nearly all the genetic variation was shared between these different obesity-related traits. Further, the heritability estimates for these traits were high: genetic factors explained from 80 to $90\%$ of the variation for most of them. Thus, there is a set of genes explaining a substantial proportion of variation of different obesity-related traits in children. A large number of loci associated with childhood BMI have been identified in a GWA study and are also associated with obesity measures in adulthood [18]. Thus, these loci may also underlie the variation of other obesity-related traits. We also identified genetic correlations of height with body diameters and somewhat lower correlations with body circumferences and skinfold thicknesses. Thousands of genetic variants have been identified for height in a GWA study [8]. Thus, it would be important to study how these genetic variants are associated with other anthropometric traits.
The underlying mechanisms behind the genetic correlations are poorly known and can vary between the traits. *The* genetic variants for height and BMI have largely similar associations within sibling pairs as those found at the population level, suggesting that they affect independently of family environment [31]. *The* genetic variants associated with higher BMI have been found to be enriched in the brain, especially in the hypothalamus, pituitary gland, hippocampus and limbic system [9, 32, 33]. These brain areas are important in appetite regulation, learning, cognition, emotion and memory [34]. Together with the previous direct evidence on the genetic component behind eating behavior [22], these results suggest that the genetic factors underlying covariation between different obesity traits can partly be associated with energy intake. However, it is also noteworthy that even when high, the genetic correlations between most obesity-related traits were much less than 1, suggesting that different genetic factors also affect different obesity measures. There is evidence from a GWA study that the genetic variants associated with body fat distribution are related to lipid metabolism and adipose tissue regulation in particular [35]. On the other hand, the expression of genes associated with height have been found to be enriched in growth plate chondrocytes [36]. It is interesting to note that there are genetic correlations between height and anthropometric traits not related to the ossification of bones, such as skinfold thicknesses, which is also consistent with a previous family study [17]. Thus, it is likely that part of the genes associated with height affect through other mechanisms and may, for example, reflect nutrition choices that promote both weight gain and height growth. These associations can have a basis starting from fetal life when the same genes regulate the development of different body parts [20], but these molecular level mechanisms are complex and still poorly understood [37]. More studies are thus needed to identify these different pathways from genes to various anthropometric traits.
The correlation pattern between different anthropometric traits suggests that it is possible to create summary scales capturing body morphology variation. The best known of these scales is probably the somatotype, based on 10 anthropometric measures and classifying the physique or body form through three specific components that characterize the configuration of the body: endomorphy (relative fatness), mesomorphy (relative musculoskeletal development) and ectomorphy (relative linearity) [38]. In our previous study based on these same data, we found that these somatotype components showed high heritability [26]. In this current study, we found that a large part of the variation and covariation of 22 anthropometric traits can be captured by two orthogonal factors. In particular, obesity-related traits (BMI, skinfold thicknesses and waist and hip circumferences) were loaded on the first factor. However, the loadings on this factor were also high for limb circumferences, which are a combination of bone, muscle, and fat tissues. Height, sitting height and body diameters loaded strongly on the second factor, but substantial loadings were also found for all body circumferences. Thus, we could interpret that the first factor reflects body fatness and the second factor body tallness/robustness. Both factors showed high heritability. It is well known that excess body fat is associated with higher [39], and body tallness with lower [40], risk of cardiovascular diseases, and therefore a better understanding of the biological background of these factors may have important public health implications. On the other hand, the high correlations between different obesity measures suggest that they largely capture the same information. Thus, the detailed anthropometric measures increase the accuracy when measuring body fatness and underlying genetic susceptibility. However, if detailed measures are not possible, such as in large epidemiological studies, only one measure may be enough to offer sufficient information on obesity.
We found that the heritability estimates were higher for most of the anthropometric traits in boys than in girls. This sexual dimorphism parallels the findings of a large pooled twin study in that heritability estimates of BMI were systematically higher in boys than in girls over childhood [6]; for height, the results were somewhat less systematic but also showed higher heritability in boys at most of the ages [7]. These results may suggest that the female body shows more environmental plasticity as compared to the male body. The sexual dimorphism of phenotype environmental plasticity is very common in the animal kingdom, but it is affected by traits such as evolutionary pressure and cross-sex genetic correlations [41]. Thus, more studies are needed to analyze this issue in humans to discover whether this reflects, for example, evolutionary pressure for the female body to better adapt to the changing environment. However, in light of these results, it is interesting to note that all correlations (i.e., trait correlations, additive genetic correlations and unique environmental correlations) between these anthropometric traits were similar in boys and girls. This suggests that despite the somewhat different role of genetic and environmental factors behind the variation of anthropometric traits, the pleotropic effects behind body size and morphology are roughly similar in both sexes.
Our study has both strengths and limitations. Our main strength is the very detailed measures of the human body – 22 anthropometric traits together – in a twin data set allowing us to analyze the genetic regulation of human body morphology in detail. In addition, genetic studies in Southern European populations are rare compared to Northern European and North American populations of European ancestry. Our main limitation is that the sample size was not large enough to study potential differences over the age range studied. For example, in a very large twin study pooling data from several cohorts, environmental factors shared by co-twins affected BMI variation in early childhood but its effect disappeared in adolescence [6]. In our previous study, we analyzed the heritability estimates of 10 of the traits also used in the current study and found no systematic differences between children younger or older than 12 years of age [26]. However, separating shared environmental effects from additive genetic effects requires considerable statistical power [42]. Thus, shared environmental factors may also affect anthropometric traits in early childhood in our data, but because of lack of power, we cannot identify these factors and their effect is thus pooled with additive genetic factors. Furthermore, the cross-sectional data do not allow analyzing developmental trajectories and, for example, studying whether the same genetic factors affect these anthropometric traits at different ages. To study these issues, larger studies, preferably with follow up data over childhood, are needed. Finally, we had only anthropometric measures and not dual-energy X-ray absorptiometry (DEXA), bioimpedance, computer tomography or other measures of body composition allowing us to directly assess fat and fat free mass. This information would have allowed us to calculate genetic correlations between fat mass, fat free mass and different anthropometric measures. However, considering the high genetic correlations between obesity related anthropometric traits in our data, we can speculate that the genetic correlations between fat mass and these anthropometric traits would also be high.
In conclusion, the correlation structure of detailed anthropometric measures suggested that there are two factors – general body fatness and body height/robustness of the skeleton – underlying body morphology. In particular, body fatness measures showed high genetic correlations suggesting that there is a set of genes affecting overall body fatness. *These* genetic variants common for various anthropometric traits probably play an important role in the formation of human body size and morphology. Considering the role of obesity and other human physique features behind metabolic and many other chronic diseases, a better understanding on these pleiotropic effects can also shed more light on individual variation in health risk profiles.
## Supplementary information
Supplemental material The online version contains supplementary material available at 10.1038/s41366-023-01253-0.
## References
1. **Physical status: the use and interpretation of anthropometry**. *World Health Organ Tech Rep Ser* (1995.0) **854** 1-452. PMID: 8594834
2. Himes JH. **Challenges of accurately measuring and using BMI and other indicators of obesity in children**. *Pediatrics.* (2009.0) **124** S3-22. DOI: 10.1542/peds.2008-3586D
3. Millward DJ. **Nutrition, infection and stunting: the roles of deficiencies of individual nutrients and foods, and of inflammation, as determinants of reduced linear growth of children**. *Nutr Res Rev* (2017.0) **30** 50-72. DOI: 10.1017/S0954422416000238
4. Shinsugi C, Gunasekara D, Takimoto H. **Use of mid-upper arm circumference (MUAC) to predict malnutrition among Sri Lankan schoolchildren**. *Nutrients.* (2020.0) **12** E168. DOI: 10.3390/nu12010168
5. Goto E. **Meta-analysis: identification of low birthweight by other anthropometric measurements at birth in developing countries**. *J Epidemiol* (2011.0) **21** 354-62. DOI: 10.2188/jea.JE20100182
6. Silventoinen K, Jelenkovic A, Sund R, Hur YM, Yokoyama Y, Honda C. **Genetic and environmental effects on body mass index from infancy to the onset of adulthood: an individual-based pooled analysis of 45 twin cohorts participating in the COllaborative project of Development of Anthropometrical measures in Twins (CODATwins) study**. *Am J Clin Nutr* (2016.0) **104** 371-9. DOI: 10.3945/ajcn.116.130252
7. Jelenkovic A, Sund R, Hur YM, Yokoyama Y, Hjelmborg JVB, Möller S. **Genetic and environmental influences on height from infancy to early adulthood: An individual-based pooled analysis of 45 twin cohorts**. *Sci Rep* (2016.0) **6** 28496. DOI: 10.1038/srep28496
8. Yengo L, Vedantam S, Marouli E, Sidorenko J, Bartell E, Sakaue S. **A saturated map of common genetic variants associated with human height**. *Nature.* (2022.0) **610** 704-12. DOI: 10.1038/s41586-022-05275-y
9. Turcot V, Lu Y, Highland HM, Schurmann C, Justice AE, Fine RS. **Protein-altering variants associated with body mass index implicate pathways that control energy intake and expenditure in obesity**. *Nat Genet* (2018.0) **50** 26-41. DOI: 10.1038/s41588-017-0011-x
10. Silventoinen K, Li W, Jelenkovic A, Sund R, Yokoyama Y, Aaltonen S. **Changing genetic architecture of body mass index from infancy to early adulthood: an individual based pooled analysis of 25 twin cohorts**. *Int J Obes* (2022.0) **46** 1901-9. DOI: 10.1038/s41366-022-01202-3
11. Wardle J, Carnell S, Haworth CM, Plomin R. **Evidence for a strong genetic influence on childhood adiposity despite the force of the obesogenic environment**. *Am J Clin Nutr* (2008.0) **87** 398-404. DOI: 10.1093/ajcn/87.2.398
12. Peeters MW, Beunen GP, Maes HH, Loos RJF, Claessens AL, Vlietinck R. **Genetic and environmental determination of tracking in subcutaneous fat distribution during adolescence**. *Am J Clin Nutr* (2007.0) **86** 652-60. DOI: 10.1093/ajcn/86.3.652
13. Silventoinen K, Kaprio J, Dunkel L, Yokoyama Y. **Genetic and environmental influences on chest circumference during infancy: a longitudinal study of Japanese twins**. *Paediatr Perinat Epidemiol* (2012.0) **26** 553-60. DOI: 10.1111/ppe.12003
14. Silventoinen K, Iacono WG, Krueger R, McGue M. **Genetic and environmental contributions to the association between anthropometric measures and IQ: a study of Minnesota twins at age 11 and 17**. *Behav Genet* (2012.0) **42** 393-401. DOI: 10.1007/s10519-011-9521-y
15. Beunen G, Maes HH, Vlietinck R, Malina RM, Thomis M, Feys E. **Univariate and multivariate genetic analysis of subcutaneous fatness and fat distribution in early adolescence**. *Behav Genet* (1998.0) **28** 279-88. DOI: 10.1023/A:1021671313974
16. Hasselbalch AL, Benyamin B, Visscher PM, Heitmann BL, Kyvik KO, Sørensen TIA. **Common genetic components of obesity traits and serum leptin**. *Obesity.* (2008.0) **16** 2723-9. DOI: 10.1038/oby.2008.440
17. Jelenkovic A, Rebato E. **Association among obesity-related anthropometric phenotypes: analyzing genetic and environmental contribution**. *Hum Biol* (2012.0) **84** 127-37. DOI: 10.3378/027.084.0202
18. Vogelezang S, Bradfield JP, Ahluwalia TS, Curtin JA, Lakka TA, Grarup N. **Novel loci for childhood body mass index and shared heritability with adult cardiometabolic traits**. *PLoS Genet* (2020.0) **16** e1008718. DOI: 10.1371/journal.pgen.1008718
19. Ghosh S, Kasher M, Malkina I, Livshits G. **Is craniofacial morphology and body composition related by common genes: Comparative analysis of two ethnically diverse populations**. *Am J Phys Anthropol* (2021.0) **176** 249-61. DOI: 10.1002/ajpa.24373
20. Tickle C. **How the embryo makes a limb: determination, polarity and identity**. *J Anat* (2015.0) **227** 418-30. DOI: 10.1111/joa.12361
21. Pan S, Chen Y, Yan J, Li F, Chen X, Xu X. **The emerging roles and mechanisms of exosomal non-coding RNAs in the mutual regulation between adipose tissue and other related tissues in obesity and metabolic diseases**. *Front Endocrinol* (2022.0) **13** 975334. DOI: 10.3389/fendo.2022.975334
22. Silventoinen K, Konttinen H. **Obesity and eating behavior from the perspective of twin and genetic research**. *Neurosci Biobehav Rev* (2020.0) **109** 150-65. DOI: 10.1016/j.neubiorev.2019.12.012
23. Maia JAR, Santos D, de Freitas DL, Thomis M. **Physical activity, physical fitness, gross motor coordination, and metabolic syndrome: focus of twin research in Portugal**. *Twin Res Hum Genet* (2013.0) **16** 296-301. DOI: 10.1017/thg.2012.128
24. von Wurmb-Schwark N, Schwark T, Christiansen L, Lorenz D, Oehmichen M. **The use of different multiplex PCRs for twin zygosity determination and its application in forensic trace analysis**. *Leg Med* (2004.0) **6** 125-30. DOI: 10.1016/j.legalmed.2003.12.002
25. Claessens AL, Eynde BV, Renson R, Gerven DV, Simons J, Beunen G, Renson R, Claessens A, Vanreusel B, Lefevre J. **The description of tests and measurements**. *Growth and Fitness of Flemish girls—The Leuven Growth Study* (1990.0) 21-39
26. Silventoinen K, Maia J, Jelenkovic A, Pereira S, Gouveia É, Antunes A. **Genetics of somatotype and physical fitness in children and adolescents**. *Am J Hum Biol* (2021.0) **33** e23470. DOI: 10.1002/ajhb.23470
27. Williams RL. **A note on robust variance estimation for cluster-correlated data**. *Biometrics.* (2000.0) **56** 645-6. DOI: 10.1111/j.0006-341X.2000.00645.x
28. Posthuma D, Beem AL, de Geus EJC, van Baal GCM, von Hjelmborg JB, Iachine I. **Theory and practice in quantitative genetics**. *Twin Res* (2003.0) **6** 361-76. DOI: 10.1375/136905203770326367
29. Kaprio J, Silventoinen K. **Advanced methods in twin studies**. *Methods Mol Biol* (2011.0) **713** 143-52. DOI: 10.1007/978-1-60327-416-6_11
30. Neale MC, Hunter MD, Pritikin JN, Zahery M, Brick TR, Kirkpatrick RM. **OpenMx 2.0: extended structural equation and statistical modeling**. *Psychometrika.* (2016.0) **81** 535-49. DOI: 10.1007/s11336-014-9435-8
31. 31.Howe LJ, Nivard MG, Morris TT, Hansen AF, Rasheed H, Cho Y, et al. Within-sibship genome-wide association analyses decrease bias in estimates of direct genetic effects. Nat Genet. 2022;54:581–92.
32. Locke AE, Kahali B, Berndt SI, Justice AE, Pers TH, Day FR. **Genetic studies of body mass index yield new insights for obesity biology**. *Nature.* (2015.0) **518** 197-206. DOI: 10.1038/nature14177
33. Akbari P, Gilani A, Sosina O, Kosmicki JA, Khrimian L, Fang YY. **Sequencing of 640,000 exomes identifies GPR75 variants associated with protection from obesity**. *Science.* (2021.0) **373** eabf8683. DOI: 10.1126/science.abf8683
34. Lenard NR, Berthoud HR. **Central and peripheral regulation of food intake and physical activity: pathways and genes**. *Obesity.* (2008.0) **16** S11-22. DOI: 10.1038/oby.2008.511
35. Justice AE, Karaderi T, Highland HM, Young KL, Graff M, Lu Y. **Protein-coding variants implicate novel genes related to lipid homeostasis contributing to body-fat distribution**. *Nat Genet* (2019.0) **51** 452-69. DOI: 10.1038/s41588-018-0334-2
36. Renthal NE, Nakka P, Baronas JM, Kronenberg HM, Hirschhorn JN. **Genes with specificity for expression in the round cell layer of the growth plate are enriched in genomewide association study (GWAS) of human height**. *J Bone Miner Res* (2021.0) **36** 2300-8. DOI: 10.1002/jbmr.4408
37. Alberti C, Cochella L. **A framework for understanding the roles of miRNAs in animal development**. *Development.* (2017.0) **144** 2548-59. DOI: 10.1242/dev.146613
38. Carter JEL, Heath BH. *Somatotyping development and applications* (1990.0)
39. Wormser D, Kaptoge S, Di Angelantonio E, Wood AM, Pennells L. **Separate and combined associations of body-mass index and abdominal adiposity with cardiovascular disease: collaborative analysis of 58 prospective studies**. *Lancet.* (2011.0) **377** 1085-95. DOI: 10.1016/S0140-6736(11)60105-0
40. **Adult height and the risk of cause-specific death and vascular morbidity in 1 million people: individual participant meta-analysis**. *Int J Epidemiol* (2012.0) **41** 1419-33. DOI: 10.1093/ije/dys086
41. Hangartner S, Sgrò CM, Connallon T, Booksmythe I. **Sexual dimorphism in phenotypic plasticity and persistence under environmental change: An extension of theory and meta-analysis of current data**. *Ecol Lett* (2022.0) **25** 1550-65. DOI: 10.1111/ele.14005
42. Visscher PM, Gordon S, Neale MC. **Power of the classical twin design revisited: II detection of common environmental variance**. *Twin Res Hum Genet* (2008.0) **11** 48-54. DOI: 10.1375/twin.11.1.48
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.