Abstract
Numerous studies linked fine particulate matter (PM(2.5)) to ischemic stroke. However, only a few investigated the differential associations with specific PM(2.5) components and sources. We utilized electronic health records (EHR) from the Mount Sinai Health System in the New York City metropolitan area during 2011-2019 and assessed the associations of PM(2.5) components and sources with ischemic stroke. We used mixed-effect Poisson survival regressions to assess the single-exposure associations with the chemical components. We used multivariable regression to assess the simultaneous associations with source-apportioned PM(2.5) exposures estimated using non-negative matrix factorization. Then, we assessed the sensitivity of our results to different specifications of EHR data continuity: (1) using a less strict definition of censorship year, (2) adjusting the model for EHR data continuity index, a validated algorithm measuring EHR-data continuity based on indicators of primary care service utilization. We observed higher risks for ischemic stroke (Risk ratio [95 % confidence intervals] per interquartile range increase) associated with higher exposure to nickel (1.080 [1.045; 1.116]), vanadium (1.070 [1.033; 1.109]), zinc (1.076 [1.031; 1.122]), and nitrate (1.084 [1.039; 1.132]). In the multivariate models we found higher risk for ischemic stroke associated with exposure to oil combustion sourced PM(2.5) (1.061 [1.012; 1.113]). The results remained consistent under different model specifications accounting for EHR data continuity. In conclusion, we found an increased risk of ischemic stroke associated with specific PM(2.5) components and sources. These findings were robust to different specifications of EHR-data continuity. Our findings can inform policy and interventions aimed at reducing cardiovascular disease burden.