Education and stroke: evidence from epidemiology and Mendelian randomization study

Stroke is a leading cause of death and disability globally. Identifying modifiable risk factors is crucial for prevention. Previous research suggests an association between lower education levels and increased stroke risk, but this association may be confounded by other factors. Randomized controlled trials are not feasible to study the impact of education on stroke. Mendelian randomization (MR), an epidemiological method using genetic variants as proxies for modifiable risk factors, offers a way to assess causality. While some MR studies have suggested a protective effect of education against ischemic stroke, the prospective association between education and incident stroke in a large, long-term cohort study remains unclear, as does the causal relationship regarding different stroke subtypes (ischemic and hemorrhagic). This study aimed to characterize this prospective association in the ARIC Study and to use MR to assess the causal relationship between education and various stroke types.

Existing literature demonstrates a correlation between lower education levels and a higher risk of stroke, independent of many traditional risk factors. However, observational studies are prone to confounding and cannot definitively establish causality. Mendelian randomization (MR) studies offer a powerful tool to address this limitation by using genetic variants associated with education as instrumental variables to estimate the causal effect on stroke risk. A recent two-sample MR analysis indicated a protective effect of education against ischemic stroke, independent of cognitive function, suggesting a causal relationship. However, the literature lacks a comprehensive, long-term prospective cohort study and investigation into the causal effect on hemorrhagic stroke.

This study utilized data from the Atherosclerosis Risk in Communities (ARIC) Study, a large population-based prospective cohort study. After excluding participants with missing data or pre-existing conditions, the final sample included 11,509 individuals. Education level was self-reported at baseline (visit 1) and categorized into basic, intermediate, and advanced levels. Incident stroke was ascertained through 2014, using data from annual telephone interviews, visits, and hospital surveillance. Cox proportional hazards regression models were used to analyze the association between education level and incident stroke (total, ischemic, and hemorrhagic), adjusting for various cardiovascular risk factors and family income. Two-sample Mendelian randomization (MR) analysis was conducted using publicly available summary-level data from genome-wide association studies (GWAS) to estimate the causal associations between genetically determined education and different stroke types. Several MR methods were employed, including inverse variance weighted (IVW), weighted median, weighted mode, and MR-Egger regression, along with sensitivity analyses such as leave-one-out analysis and funnel plots to assess the robustness of the findings. Baseline characteristics were compared across education levels using appropriate statistical tests (ANOVA, χ², Kruskal-Wallis).

During a median follow-up of 25.3 years, 915 stroke events (8.0%) occurred. Cox proportional hazards regression models showed that individuals with intermediate and advanced education levels had a significantly decreased risk of total stroke (HR 0.83, 95% CI 0.71, 0.98 and HR 0.67, 95% CI 0.56, 0.80, respectively, after adjusting for several covariates). In the final model which adjusted for family income, those with advanced education had a 25% reduction in the risk of total stroke (HR 0.75, 95% CI 0.62, 0.91). Similar trends were observed for ischemic stroke, with a decreased risk associated with intermediate and advanced education levels in the final model. However, no significant association was found between education level and hemorrhagic stroke incidence. Subgroup analysis showed consistent results across different demographic and lifestyle factors, except for interactions with BMI and age, suggesting potential effect modification. Mendelian randomization analysis provided further evidence for a causal relationship. The IVW method showed a significant protective causal effect of education on ischemic stroke (OR 0.764, 95% CI 0.585–0.998, P=0.048), but not on total or hemorrhagic stroke. Sensitivity analyses supported the robustness of this finding for ischemic stroke. The MR analysis revealed no significant association between genetically determined education and total stroke or hemorrhagic stroke risk.

This study provides strong evidence for a negative association between higher education and the incidence of total and ischemic stroke in a large, long-term cohort. The MR analysis further supports a likely causal link between genetically determined education and a reduced risk of ischemic stroke. This supports the hypothesis that education's protective effect is not simply due to confounding factors, but rather reflects a genuine causal relationship, at least concerning ischemic stroke. This effect may be due to education’s association with healthier lifestyles, safer working conditions, and improved access to healthcare. However, the absence of a significant association with hemorrhagic stroke highlights the heterogeneity of stroke subtypes and suggests that the mechanisms linking education to stroke risk may differ across subtypes. The different pathophysiological processes involved in ischemic and hemorrhagic stroke may explain this discrepancy. Future research should explore potential mediating factors and explore the interaction with other factors such as BMI and age.

Higher education levels were associated with a decreased risk of total and ischemic stroke, but not hemorrhagic stroke. Mendelian randomization analysis suggests a likely protective causal effect of education on ischemic stroke. This emphasizes the importance of education in stroke prevention. Further research should investigate the mechanisms underlying this association and explore the role of education across diverse populations.

Several limitations should be considered. The ARIC study may have missed some stroke events due to ascertainment bias. The MR analysis relied on publicly available GWAS data, which may limit the generalizability of findings to non-European populations. Additional potential confounders, not measured in the ARIC study, could influence the results. The study's focus on European populations limits the generalization of results to other ethnic and racial groups.