Air pollution and stroke; effect modification by sociodemographic and environmental factors. A cohort study from Denmark

Objectives: Air pollution increases the risk of stroke, but the literature on identifying susceptible subgroups of populations is scarce and inconsistent. The aim of this study was to investigate if the association between air pollution and risk of stroke differed by sociodemographic factors, financial stress, comorbid conditions, and residential road traffic noise, population density and green space. Methods: We assessed long-term exposure to air pollution with ultrafine particles, PM2.5, elemental carbon and NO2 for a cohort of 1,971,246 Danes aged 50-85 years. During follow-up from 2005 to 2017, we identified 83,211 incident stroke cases. We used Cox proportional hazards model (relative risk) and Aalen additive hazards models (absolute risk) to estimate associations and confidence intervals (CI) between 5-year running means of air pollution at the residence and risk of stroke in population strata.Results: All four pollutants were associated with higher risk of stroke. The association between air pollution and stroke was strongest among individuals with comorbidities, with shorter education, lower income and being retired. The results also indicated stronger associations among individuals living in less populated areas, and with low noise levels and more green space around the residence. Estimates of absolute risk seemed better suited to detect such interactions than estimates of relative risk. For example for PM2.5 the hazard ratio for stroke was 1.28 (95%CI: 1.22-1.34) and 1.26 (95%CI: 1.16-1.37) among those with mandatory and medium/long educa-tion respectively. The corresponding rate difference estimates per 100,000 person years were 568 (95%CI: 543-594) and 423(95%CI: 390-456)Conclusion: The associations between air pollution and risk of stroke was stronger among individuals of lower socioeconomic status or with pre-existing comorbid conditions. Absolute risk estimates were better suited to identify such effect modification.