Previously published community- or population-based studies of brain aging and dementia with autopsy were restricted to a single sex, a single ethnic group, Roman Catholic clergy, or focused pathological assessments. Our goal was to determine the independent pathological correlates associated with dementia in a typical US population.


We evaluated autopsy data from the Adult Changes in Thought study, an ongoing longitudinal, population-based study of brain aging and dementia. Analyses were based on data collected from about 3,400 people 65 years or older who were cognitively intact at the time of enrollment in the Group Health Cooperative in King County, Washington. All consecutive autopsies (n = 221; 20% of deaths) from this cohort were evaluated and analyzed by weighted multivariate analysis to account for potential participation bias.


After adjusting for age, sex, education, and APOE, independent correlates of dementia (relative risk, 95% confidence interval; overall p value) included Braak stage (V/VI vs 0/I/II: 5.89, 1.62–17.60; p < 0.05), number of cerebral microinfarcts in standardized sections (>2 vs none: 4.80, 1.91–10.26; p < 0.001), and neocortical Lewy bodies (any vs none: 5.08, 1.37–18.96; p < 0.05). Estimates of adjusted population attributable risk for these three processes were 45% for Braak stage, 33% for microinfarcts, and 10% for neocortical Lewy bodies.


Our results underscore the therapeutic imperative for Alzheimer's and Lewy body diseases, and provide evidence to support the immediate use of strategies that target cerebral microinfarcts as a means to partially prevent or delay the onset of dementia. Ann Neurol 2007