Cardiovascular diseases (CVDs) are the leading cause of death globally. Ischemic heart disease (IHD) had the highest global age-standardized disability-adjusted life years (DALYs) among all diseases, at 2,275.9 per 100,000 in 2022. Stroke was the second leading cause of CVD-related age-standardized DALYs.1 Low- and middle-income countries (LMICs) contribute significantly to the global burden of CVD, accounting for 80% of all CVD-related deaths worldwide. Nearly half of all deaths in developing countries are premature.2
The burden of CVD is particularly high in certain LMICs. In 2019, Afghanistan had the highest age-standardized CVD DALY rate in the Middle East and North Africa, at 12,531 per 100,000.3 In Jordan, the age-standardized DALY rate for CVD was 4,647 per 100,000 population in the same year.4 In Mongolia, the age-standardized DALY rates in 2019 were 7,780 for IHD and 5,971 for stroke per 100,000 population.5 In Nepal, age-standardized DALYs due to CVD reached 4,962 per 100,000 population in 2019, reflecting a consistent increase since 1990, with no decline.6 In São Tomé and Príncipe, the leading causes of death among females in 2019 were IHD, with a rate of 57 per 100,000 population, and stroke, with a rate of 53 per 100,000 population.7
Assessing social determinants is essential in primary CVD prevention, as regional variations in cardiovascular risk are influenced by socioeconomic, behavioural, and environmental factors. Prevention efforts should focus on reducing the risk by targeting higher-risk populations and addressing the determinants.8
Previous studies have identified several factors contributing to primary CVD risk, including socioeconomic factors such as marital status, income, and education; behavioural factors like an unhealthy diet, physical inactivity, and tobacco use; and environmental factors such as air pollution. Age, sex, high blood pressure, high cholesterol, diabetes, kidney dysfunction, and autoimmune disorders are also among the predictors of CVD risk.9,10 While traditional risk factors such as hypertension, dyslipidaemia, diabetes, and smoking are well-established, the role of social determinants in primary CVD risk variation is not well understood and is less explored in LMICs.11
The findings of this study can guide targeted primary CVD prevention in LMICs by identifying contextual risk factors, addressing disparities, and reducing the CVD burden in resource-limited settings. The study can support global efforts to reduce health inequalities and improve cardiovascular outcomes in vulnerable populations, aligning with the Sustainable Development Goals.12 This study aimed to quantify the mean ten-year CVD risk scores in Afghanistan, Jordan, Mongolia, Nepal, and São Tomé and Príncipe, and identify individual-level and community-level factors associated with ten-year CVD risk.
METHODS
Design, setting, and participants
The WHO STEPwise approach to non-communicable disease (NCD) risk factor surveillance (STEPS) is a population-based cross-sectional survey that assesses NCD risk factors through standardized questionnaires, physical measurements, and biochemical tests. This study used data from the 2018/2019 WHO STEPS surveys (February 2018 to December 2019) in Afghanistan, Jordan, Mongolia, Nepal, and São Tomé and Príncipe. We analysed using separate models for each country. The study identified that some CVD risk factors were significant in multiple countries, while others were country specific. Comparing these datasets identified shared and unique risk factors across countries with varying socio-economic and demographic characteristics. This comparison enhances the generalizability of our findings and strengthens our understanding of regional variations in CVD risk and its determinants.
The surveys followed consistent protocols across all participating countries. The collected data included sociodemographic factors (age, sex, marital status), behavioural factors (tobacco and alcohol use, physical activity), and biological factors (blood pressure, blood glucose, and cholesterol). The sample design ensured representativeness through stratification and proportional allocations. Sampling techniques followed the WHO STEPS methodology: random cluster selection, systematic or random household selection, and random selection of one individual per household.
In Afghanistan, adults aged 18-69 years were sampled from 34 provinces, stratified into six regions with proportional cluster allocation. A multistage approach was employed, starting with the random selection of districts, followed by systematic sampling of villages and households, with one adult randomly selected per household. In Jordan, a two-stage stratified cluster design ensured the representation of both Jordanians and Syrian refugees aged 18-69 years, with eight households randomly selected per block and one adult per household. In Mongolia, a multistage stratified sampling design was used to select 377 clusters of adults aged 15-69 years, stratified by urban and rural areas. In Nepal, adults aged 15-69 years were sampled using multistage cluster sampling across four strata: metropolitan, sub-metropolitan, municipality, and rural municipality. In São Tomé and Príncipe, a multistage cluster design was used to sample adults aged 18-69 years, with proportional allocation across districts and one adult per household.
Sample sizes for the STEP surveys were determined based on confidence intervals, design effect, and non-response rates, stratified by age. In our analysis, we included participants aged 40-69 years to align with the 2019 WHO CVD risk equations, which are based on individuals aged 40-80 years.13 The upper age limit of 69 was set due to WHO STEPS data restrictions, which exclude individuals older than 69. In Afghanistan, 3,972 adults were sampled, with 1,370 aged 40-69 years included in the analysis. In Jordan, 6,000 participants (3,000 Jordanians and 3,000 Syrians) were sampled, with 1,522 included in the analysis. Mongolia’s sample included 6,497 adults, with 2,689 included in the analysis. Nepal sampled 6,475 adults, with 2,431 included, while São Tomé and Príncipe sampled 2,650 adults, with 702 included in the analysis.
Outcome variable
The outcome variable was the ten-year CVD risk (the probability of a CVD event within ten years), calculated using the 2019 WHO CVD risk equation. This equation is sex-stratified, laboratory-based, and accounts for both fatal and non-fatal myocardial infarction and stroke.13 This equation was validated for the populations of Afghanistan, Jordan, Mongolia, Nepal, and São Tomé and Príncipe through recalibration, accounting for CVD incidence and risk factors in each country.
Explanatory variables
Individual-level factors
The study included sociodemographic factors reported in the survey, such as marital status (never married, currently married/cohabiting, separated/divorced/widowed); education levels (no formal education, less than primary school, primary school completed, secondary school completed, college/university completed); occupation (employed [private skilled workers, farmers, traders, students, homemakers/housewives], retired, unemployed but able to work, unemployed and unable to work); annual income level (categorised into household income quintiles); and area of residence (urban vs. rural). Physical activity levels were assessed using the WHO Global Physical Activity Questionnaire and detailed in the Online Supplementary Document on page 1. The median percentage of missing values for explanatory variables across the five countries was 0.2%, with the highest being 1.6% for educational status in Afghanistan. These data are presented in the Online Supplementary Document on page 1.
Community-level factors
Community-level poverty was categorised into quintiles of annual per capita income, with higher poverty defined as the first two quintiles and lower poverty as the third to fifth quintiles. Community-level education was measured as the proportion of participants within the community who had attained secondary education or higher.
Data analysis
The ten-year CVD risk for WHO STEPS participants was calculated using the 2019 WHO CVD risk equation.13 The equation was recalibrated using age- and sex-specific CVD incidence data from the 2017 Global Burden of Disease Study and risk factor values from the NCD Risk Factor Collaboration. We calculated the ten-year CVD risk using age, smoking status (current vs. non-current), systolic blood pressure (mmHg), history of diabetes (yes/no), and total cholesterol (mmol/L) from the WHO STEPS survey. The ten-year WHO CVD risk prediction model includes interaction terms between age and each predictor variable: smoking status, systolic blood pressure, history of diabetes, and total cholesterol. Sample weighting and adjustments accounted for variations in age, sex, and residential area distribution between the sample and target population and selection probabilities. Survey datasets provided three-step weights (wstep1, wstep2, wstep3) for each participant.
Missing values in explanatory variables were addressed using multiple imputations by chained equations. The imputation model included all explanatory variables used in the multivariable regression analysis. Five imputed datasets were generated and used for the final analysis. Multilevel linear regression models were used to identify individual- and community-level factors associated with overall ten-year CVD risk. We included individual-level factors (marital status, education status, occupation, income levels, area of residence, and physical activity levels) and community-level factors (community-level poverty and community-level education), and adjusted for age and sex. Statistical analyses were conducted using Stata version 17 (StataCorp, College Station, TX, USA, 2023).
RESULTS
Characteristics of the study participants and ten-year CVD risk
Overall, this study included participants nested within clusters in five countries: Afghanistan (1,370 individuals, 262 clusters), Jordan (1,522 individuals, 661 clusters), Mongolia (2,689 individuals, 377 clusters), Nepal (2,431 individuals, 258 clusters), and São Tomé and Príncipe (702 individuals, 108 clusters). The mean (standard deviation) age of participants ranged from 50.9 (8.1) years in São Tomé and Príncipe to 52.4 (8.5) years in Nepal (Table 1). The proportion of women exceeded men in all five countries except Afghanistan, where men constituted 58.5% of the population. Jordan had the highest proportion of women at 63.3%. The prevalence of low physical activity was highest in Jordan (35.9%), followed by Mongolia (32.2%), São Tomé and Príncipe (27.3%), Afghanistan (15.0%), and Nepal (11.0%) (Table 1).
The weighted mean ten-year CVD risk was 18.3% in Jordan, 8.4% in Afghanistan, 6.8% in Mongolia, 5.3% in Nepal, and 5.0% in São Tomé and Príncipe. Widowed, divorced, or separated individuals consistently had a higher mean ten-year CVD risk than those who had never married in Jordan (24.3% vs. 18.5%), Afghanistan (12.2% vs. 7.9%), Mongolia (8.0% vs. 5.7%), Nepal (7.7% vs. 4.1%), and São Tomé and Príncipe (5.7% vs. 5.6%) (Table 2).
Association between individual-level factors and ten-year CVD risk
In the final model, after adjusting for individual- and community-level factors, statistically significant associations with overall ten-year CVD risk were observed for marital status in Nepal; area of residence in Afghanistan and Nepal; occupation in Afghanistan, Mongolia, and Nepal; and physical activity levels in Afghanistan, Jordan, and Mongolia.
In Nepal, widowed, divorced, or separated participants had a 1.36% higher ten-year CVD risk (β [beta coefficient percent] = 1.36, 95% CI=0.19-2.53; p < 0.05) compared to those who had never married. In Afghanistan and Nepal, participants in urban areas had a higher ten-year CVD risk compared to those in rural areas (Afghanistan: β = 0.59, 95% CI=0.01-1.17; p < 0.05; Nepal: β = 0.46, 95% CI =0.003-0.93; p < 0.05). Participants who were retired or unable to work in Afghanistan, Mongolia, and Nepal had a higher ten-year CVD risk compared to those who were employed or unemployed but capable of working (Afghanistan: β = 1.49, 95% CI=0.58-2.41; p < 0.001; Nepal: β = 1.20, 95% CI=0.56-1.85; p < 0.001; Mongolia: β = 0.83, 95% CI=0.11-1.56; p < 0.05). In Jordan, Mongolia, and Afghanistan, participants with low physical activity levels had a higher ten-year CVD risk compared to those with high physical activity levels (Jordan: β = 1.73, 95% CI=0.77-2.69; p < 0.001; Mongolia: β = 1.04, 95% CI=0.46-1.61; p < 0.001; Afghanistan: β = 0.81, 95% CI=0.07-1.56; p < 0.05) (Table 3).
DISCUSSION
The main findings of this study were variations in primary CVD risk and factors associated with the primary CVD risk. The study found variations in the mean ten-year CVD risk across the five countries: 18.3% in Jordan, 8.4% in Afghanistan, 6.8% in Mongolia, 5.3% in Nepal, and 5.0% in São Tomé and Príncipe. Ten-year CVD risk was significantly associated with individual-level factors across countries. In Nepal, widowed, divorced, or separated individuals had a higher ten-year CVD risk than those who had never married. In Afghanistan and Nepal, urban residents had a higher ten-year CVD risk than rural residents. Retired or unable-to-work individuals in Afghanistan, Mongolia, and Nepal had a higher ten-year CVD risk than those who were employed or unemployed but able to work. Low physical activity levels were also associated with a higher ten-year CVD risk in Afghanistan, Jordan, and Mongolia.
In a previous study across 45 low- and middle-income countries, participants aged 40 and above had median ten-year CVD risks of 4.0% for males and 2.8% for females. The highest primary CVD risk was reported in Eastern Europe and the Eastern Mediterranean, while lower estimates were observed in sub-Saharan Africa.14 Similarly, our study observed the highest mean ten-year CVD risk in Jordan, while the lowest was in São Tomé and Príncipe. Variations in mean primary CVD risk between countries may reflect differences in traditional CVD risk factors, such as demographic characteristics (e.g., age), high blood pressure, diabetes, smoking prevalence, and cholesterol levels, as well as non-traditional factors, including healthcare access, socioeconomic factors, diets, and other lifestyle patterns.15,16
In Jordan, the high prevalence of smoking, obesity, diabetes, hypertension, and high cholesterol likely accounts for the highest estimate of primary CVD risk.17,18 Afghanistan had a relatively lower mean CVD risk than Jordan, although factors such as tobacco use and stress from prolonged conflict may contribute to higher CVD risk in Afghanistan.19,20 In Mongolia, in addition to traditional risk factors such as hypertension, smoking, diabetes, and obesity, low physical activity levels and unhealthy dietary patterns may also contribute to higher CVD risk.21,22 Mongolia’s cold climate limits outdoor physical activity, while air pollution in urban areas like Ulaanbaatar, especially during winter’s peak coal burning, may further increase cardiovascular risk.23,24 The relatively lower mean CVD risk in Nepal and São Tomé and Príncipe may reflect differences in CVD risk factors, demographics, health risk behaviours, healthcare access, socioeconomic conditions, and environmental influences across countries.25,26
Widowed, divorced, or separated participants in Nepal had a higher ten-year CVD risk compared to those who were never married. Previous studies indicate that marital status is associated with CVD risk, with widowed and divorced individuals having a higher risk.27–29 Marital disruption, particularly spousal loss, increases psychological stress, loneliness, and emotional distress, exacerbating cardiovascular risk factors. The lack of support reduces adherence to medical advice, increases unhealthy eating, and decreases physical activity levels.30,31
In our study, urban residents in Afghanistan and Nepal had a higher ten-year CVD risk than rural residents. In LMICs, urban areas often experience higher CVD risk due to lifestyle changes, air pollution, physical inactivity, unhealthy diets, obesity, diabetes, and smoking, all strongly associated with a higher risk of CVD.32–34
The evidence on the association between retirement and CVD risk varies. Some studies find no significant association, while others suggest a detrimental effect.35 However, a study identified a beneficial effect of retirement on cardiovascular health.36 Variations in state pensions and the healthy-worker survivor effect may explain differences in the association between CVD risk and retirement. In our study, retirees, or individuals unable to work in Afghanistan, Mongolia, and Nepal, had a higher ten-year CVD risk compared to those employed or able to work. Retirement can disrupt routine, social interaction, and income, negatively affecting physical and mental health. Reduced social networks further contribute to physical health declines and an increased risk of depression. However, retirement can also improve psychological well-being by reducing work stress, offering more free time, and supporting healthier lifestyles. Additional research across various settings is needed.37,38
Previous studies identified that regular physical activity lowers CVD risk, while sedentary behaviours increase CVD and all-cause mortality. Low physical activity level is associated with CVD risk factors such as hypercholesterolemia, hypertension, inflammation, and insulin resistance. Higher physical activity levels lower CVD risk by improving lipid profiles, insulin sensitivity, and inflammation.39–41 In our study, low physical activity levels in Afghanistan, Jordan, and Mongolia were associated with a higher risk of CVD, consistent with previous findings.42 Higher physical activity enhances HDL cholesterol, lowers LDL cholesterol, and triglycerides.43 Regular physical activity improves insulin sensitivity, lowers plasma C-reactive protein levels, and reduces inflammation.44–46
The primary outcome of this study was the ten-year predicted CVD risk score, calculated using the WHO 2019 laboratory-based equation, which includes age, sex, systolic blood pressure, total cholesterol, smoking status, and diabetes. Multilevel regression models were used to assess associations between this risk score and various individual- and community-level factors, including marital status, occupation, physical activity, area of residence, and community-level education or income. Although behavioural and social variables such as marital status showed statistically significant associations with predicted ten-year CVD risk, these should not be interpreted as having the same clinical or biological implications as traditional risk factors. The traditional factors included in the WHO risk model were selected because of their well-established links with CVD. Social and contextual explanatory variables associated with primary CVD risk may contribute indirectly through behavioural, psychosocial, or environmental pathways.47 The regression coefficients reflect statistical associations with the predicted CVD risk score rather than causal effects.48,49
Urban residents, widowed individuals, those retired or unable to work, and people with low physical activity levels had higher predicted cardiovascular risk. Addressing these disparities in LMICs requires low-cost, practical interventions integrated into existing health or community systems. For example, urban health campaigns, community-based physical activity groups, social engagement programs, and routine CVD screening through outreach may support higher-risk groups. Although we did not conduct feasibility studies, we emphasise the need for future research to assess the feasibility of these interventions in resource-limited settings.
While some risk differences, such as the beta coefficient of 1.36% higher 10-year CVD risk among widowed individuals in Nepal, may appear small, they can still have clinical relevance, particularly in LMICs where the burden of CVD is high. Even small increases in population-level risk may justify prioritizing early identification and intervention in vulnerable subgroups.50 Although this study did not conduct a cost-effectiveness analysis, this study emphasises the need for future research to assess both the clinical impacts and cost-effectiveness of primary CVD prevention strategies targeting the higher-risk group in LMICs.
While socioeconomic and behavioural factors like urban residence and low physical activity levels are known CVD risk determinants, this study quantifies primary CVD risk using recalibrated WHO equations across five under-represented LMICs. The study also identifies contextual variations, such as higher risk among widowed/divorced individuals in Nepal and retirees or those unable to work in Afghanistan and Mongolia, identifying vulnerable groups and informing prevention strategies addressing social and structural determinants. However, this study did not directly examine biological, psychosocial, and environmental pathways underlying the association between primary CVD risk and identified explanatory variables. We did not perform mediation analysis due to the absence of prospective, specific social, behavioural, and biological data that could serve as potential mediators in the WHO STEPS surveys. However, prior research suggests potential pathways linking social and behavioural factors to the primary CVD risk. For instance, psychosocial stressors such as grief, loneliness, and emotional distress may mediate the association between widowhood and higher primary CVD risk by influencing health behaviours such as diet and physiological responses, including hormonal dysregulation and systemic inflammation.51–53 Similarly, low physical activity may increase the CVD risk through metabolic changes and inflammatory processes, with biomarkers such as C-reactive protein playing a role.54–56 Future research should include the potential mediators and use mediation analysis to explain the causal mechanisms.57
Strengths and Limitations
This study uses nationally representative data from five countries, enhancing its applicability. The multilevel analysis captures individual and community variations, addressing the hierarchical nature of the WHO STEPS surveys. A recalibrated WHO-2019 CVD risk equation, incorporating local data, predicts ten-year CVD risk more accurately than the non-recalibrated version. A key strength of this study is the use of recalibrated WHO CVD risk equations, which enhances estimation accuracy within countries. However, recalibration may limit the comparability of risk estimates across countries due to differences in the input data used for recalibration, such as baseline risk factor levels and incidence rates. Future studies should consider sensitivity analyses better to understand the effects of recalibration on cross-country comparability. The cross-sectional design limits the ability to infer causal relationships between ten-year CVD risk and associated factors. Residual confounding from unmeasured social or environmental factors could also be a study limitation. The WHO STEPS data used in this study involved limited community engagement during study design and dissemination, despite adherence to WHO protocols and national survey regulations.
This study used nationally representative data from five LMICs across four WHO regions: Afghanistan and Jordan (Eastern Mediterranean), Mongolia (Western Pacific), Nepal (South-East Asia), and São Tomé and Príncipe (Africa). Although these countries provide some diversity, they are not representative of all LMICs, as they exclude much of Sub-Saharan Africa, Latin America, and South Asia, which limits generalizability. Region-specific exposures, such as prolonged conflict and displacement in Afghanistan and severe winter air pollution in urban Mongolia, were not fully accounted for, although these factors may influence cardiovascular risk.58,59 Future research on primary CVD predictions should include a broader range of LMICs and incorporate region-specific factors.
CONCLUSION
This study reaffirms differences in primary CVD risk profiles across countries and indicates the need for focused interventions, such as promoting urban-based CVD prevention in Afghanistan and Nepal, increasing physical activity in Jordan and Mongolia, strengthening social support for widowed individuals in Nepal, and improving access to primary CVD prevention for retirees and those unable to work in Afghanistan, Mongolia and Nepal. Future research should assess cost-effectiveness, feasibility, and implementation challenges of interventions. Studies should also investigate the underlying mechanisms that drive primary CVD risk in higher-risk populations.
Acknowledgements
We would like to acknowledge that YMA is supported by the Australian National University International Research Scholarship (738/2018) and the Free Remission Merit Scholarship (271/2014). This study used WHO NCD STEPS Microdata accessed through the NCD Microdata Repository (1514). We thank the WHO NCD Microdata Repository team in Geneva, Switzerland.
Ethics statement
The study obtained ethics approval from the Australian National University Human Ethics Office (H/2023/1420). The NCD Microdata Repository (1514) approved access to the WHO NCD STEPS microdata across multiple studies. The WHO conducted each country survey in collaboration with the respective Ministry of Health, obtaining ethical clearance through national or regional committees to ensure adherence to country-specific ethical standards. Written informed consent was obtained from all participants before the study.
Data availability
All data used in the analysis are available from the WHO NCD microdata upon reasonable request and approval.
Funding
International Research Scholarship (738/2018) and the Free Remission Merit Scholarship (271/2014).
Authorship contributions
Conceptualization: YMA, NB, AR, DC. Formal analysis: YMA. Methodology: YMA, NB, AR, DC. Supervision: NB, KW, DC, AR. Writing original draft: YMA. Writing with review & editing: YMA, NB, KW, AR, DC. All authors critically reviewed and approved the manuscript.
Disclosure of interest
The authors have completed the ICMJE Disclosure of Interest Form (available upon request from the corresponding author) and report no relevant interests.
Additional material
Online Supplementary Document.
Correspondence to:
Yihun Mulugeta Alemu
National Centre for Epidemiology and Population Health, College of Health and Medicine, The Australian National University, Australia
62 Mills Rd, Acton ACT 2601
Australia
yihun.alemu@anu.edu.au