Abstract
IMPORTANCE: The association between stroke severity and dementia is well established. However, reports on trajectories of cognitive decline comparing stroke survivors with individuals without stroke in large cohorts are insufficient. OBJECTIVES: To examine associations of ischemic stroke incidence and severity with cognitive decline and dementia risk and to explore whether vascular risk factors modify these associations. DESIGN, SETTING, AND PARTICIPANTS: This cohort study pooled longitudinal data on cognitive function of participants aged 45 years or older and without stroke and dementia at baseline from 3 US prospective cohorts: the Atherosclerosis Risk in Communities study (1987-2019), Framingham Offspring Study (1971-2019), and Reasons for Geographic and Racial Differences in Stroke study (2003-2019). First definite ischemic strokes were reported in each cohort using consistent protocols, with severity defined using the National Institutes of Health Stroke Scale (NIHSS). The data analysis was completed February 27, 2026. EXPOSURE: Incident ischemic stroke categorized as minor (NIHSS 0-5), mild to moderate (NIHSS 6-10), or moderate to severe (NIHSS ≥11). MAIN OUTCOMES AND MEASURES: The primary outcomes were decline in global cognition and incident dementia. Secondary outcomes were changes in memory and executive function. Multivariable linear mixed-effects models were used to test the association of stroke incidence and severity with cognitive decline. RESULTS: A total of 42 342 participants from the pooled cohorts were included (mean [SD] age, 61.3 [9.8] years; 55.0% female). Longitudinal cognitive testing data were available for a median of 11.1 years (range, 0-29.7 years) with 397 344 person-years of observation for dementia incidence. Stroke severity data were available for 1055 of 1505 first-ever ischemic stroke survivors (70.1%). Compared with participants with no stroke, adjusted hazard ratios for incident dementia were 1.93 (95% CI, 1.52-2.45) for NIHSS 0 to 5, 3.26 (95% CI, 1.93-5.53) for NIHSS 6 to 10, and 5.06 (95% CI, 2.71-9.45) for NIHSS 11 or higher. Over the follow-up, higher stroke severity was associated with progressively steeper cognitive declines across all domains, with more prevalent dose-response associations for global cognition (ranging from a mean -0.18 [95% CI, -0.19 to -0.18] points per year for no stroke to -0.58 [95% CI, -0.73 to -0.42] points per year for moderate to severe stroke) and memory (ranging from a mean -0.15 [95% CI, -0.16 to -0.14] points per year for no stroke to -0.36 [95% CI, -0.51 to -0.21] points per year for moderate to severe stroke) than for executive function (ranging from a mean -0.33 [95% CI, -0.34 to -0.32] points per year for no stroke to -0.52 [95% CI, -0.66 to -0.39] points per year for moderate to severe stroke). CONCLUSIONS AND RELEVANCE: This large cohort study of participants from 3 prospective cohorts found that greater stroke severity was associated with substantially elevated dementia risk and accelerated decline in global cognition, memory, and executive function. These findings underscore the critical importance of stroke prevention, particularly severe stroke, and identifying mechanisms that may link stroke to cognitive decline.