Neurological consequences of ischemic and hemorrhagic stroke in the UK Biobank

Introduction: Pathophysiological distinctions among stroke subtypes-acute ischemic stroke (AIS), subarachnoid hemorrhage (SAH), intracerebral hemorrhage (ICH), and other nontraumatic intracranial hemorrhage (ONIH)-are well described, but their associations with long-term mortality, cognitive outcomes, and dementia risk remain incompletely characterized. We examined whether stroke subtypes differ in post-stroke survival, cognitive performance, dementia risk, and neuroimaging phenotypes. Methods: Using data from the UK Biobank, we evaluated subtype-specific associations with all-cause mortality, incident all-cause dementia, post-stroke cognitive performance, and white matter hyperintensity (WMH) burden. We assessed time to mortality and all-cause dementia using multivariable Cox proportional hazards models. Cognitive outcomes were compared cross-sectionally between stroke subtypes and stroke-free participants using covariate-adjusted linear models with ANOVA and Tukey post hoc tests. Neuroimaging analyses assessed associations between stroke subtype and WMH volume. Results: Mortality risk varied substantially by stroke subtype. Of UKB participants who had a stroke during surveillance (n=14,806), we found that 69.24% of strokes were ischemic, 9.51% subarachnoid, 9.78% ONIH, and 11.78% ICH. Compared with stroke-free person-time, intracerebral hemorrhage was associated with the highest post-stroke hazard of death (HR 7.62, 95% CI 7.10-8.18), followed by other nontraumatic intracranial hemorrhage (HR 5.41, 95% CI 4.89-5.98), subarachnoid hemorrhage (HR 3.89, 95% CI 3.52-4.31), and ischemic stroke (HR 3.99, 95% CI 3.82-4.16). At 1 year after stroke, absolute mortality risk was highest following ICH (36.4%), followed by ONIH (31.9%) and SAH (26.0%), while ischemic stroke was associated with substantially lower risk (11.7%); by 5 years, corresponding risks increased to 43.5%, 39.8%, 30.0%, and 20.0%, respectively. Dementia risk also differed by subtype, with the highest risk observed following ICH. Post-stroke cognitive performance varied across domains, with slower reaction times observed across multiple stroke subtypes and lower fluid intelligence scores among AIS and ICH patients. Reaction time and fluid intelligence assessments were completed a median of approximately 6-9 years after stroke. WMH burden was higher in ICH and AIS compared with stroke-free participants. Conclusions: Stroke subtype is associated with distinct patterns of mortality, dementia risk, cognitive performance, and brain imaging phenotypes. These findings highlight the heterogeneity of long-term outcomes following stroke and support the importance of subtype-aware post-stroke monitoring and prognostication.