Background and purpose: Stroke increases the risk of dementia; however, bidirectional association of incident stroke and cognitive decline below dementia threshold is not well established. Also, both cognitive decline and stroke increase mortality risk.
Methods: A longitudinal population-based cohort of 7217 older adults without a history of stroke from a biracial community was interviewed at 3-year intervals. Cognitive function was assessed using a standardized global cognitive score. Stroke was determined by linkage with Medicare claims, and mortality was ascertained via the National Death Index. We used a Cox model to assess the risk of incident stroke, a joint model with a piecewise linear mixed model with incident stroke as a change point for cognitive decline process, and a time-dependent relative risk regression model for mortality risk.
Results: During follow-up, 1187 (16%) subjects had incident stroke. After adjusting for known confounders, lower baseline cognitive function was associated with a higher risk of incident stroke (hazard ratio, 1.61; 95% confidence interval, 1.46-1.77). Cognitive function declined by 0.064 U per year before incident stroke occurrence and 0.122 U per year after stroke, a nearly 1.9-fold increase in cognitive decline (95% confidence interval, 1.78-2.03). Both stroke (hazard ratio, 1.17; 95% confidence interval, 1.08-1.26) and cognitive decline (hazard ratio, 1.90; 95% confidence interval, 1.81-1.98) increased mortality risk.
Conclusions: Baseline cognitive function was associated with incident stroke. Cognitive decline increased significantly after stroke relative to before stroke. Cognitive decline increased mortality risk independent of the risk attributable to stroke and should be followed as a marker for both stroke and mortality.
Keywords: cognition; cohort studies; epidemiology; mortality; stroke.
© 2014 American Heart Association, Inc.