Abstract MP060: Contributors to the Geographic Disparity in Stroke Incidence in the United States: Results From the Reasons for Geographic and Racial Disparities in Stroke Cohort

Abstract

Introduction: Geographic disparities in US stroke risk have recently been confirmed by REGARDS, but contributors to this are unknown. Higher prevalence of risk factors (RF) and lower socioeconomic status (SES) may contribute to this geographic disparity. For a RF/SES to contribute to this disparity it must both : 1) have a large geographic difference in prevalence, and 2) be powerfully associated with stroke risk. Methods: The 1,623 counties of residence of 24,863 REGARDS participants were placed in quartiles of Vital Statistics stroke mortality. Logistic regression assessed the geographic difference in prevalence of each RF/SES by quartile of stroke mortality. Proportional hazards was used to calculate HR stroke for each RF/SES. Mediation analysis then estimated the proportion of increased stroke incidence in counties with high stroke mortality explained by each RF/SES. Results: Higher county-level stroke mortality was significantly associated with low neighborhood SES (nSES), and more weakly associated with low education and presence of RFs (left column of table). Over 8-years follow-up there were 1,194 stroke events. Hypertension, diabetes and heart disease were all more strongly associated with higher stroke risk (HR ≥1.59) than nSES (HR = 1.21) (center column of table). The large differences in nSES between regions overcame the somewhat weaker association of nSES with stroke risk, since nSES was the largest single contributor to the geographic disparity in stroke incidence, accounting for 20.5% of the disparity (95% CI: 7.1 - 34.0) (right column in table). In multivariable analysis nSES, hypertension and diabetes collectively mediated27% of the geographic disparity (95% CI: 20.0% - 34.0%), but the association with county-level mortality remained significant (p = 0.006). Conclusion: Lower nSES played the largest role in explaining the county-level geographic disparity in stroke incidence in REGARDS, however, 73% of the excess risk of stroke incidence was not explained by any studied factors.