Abstract

We developed prediction rules to guide the clinical diagnosis of Alzheimer’s disease (AD) in two community-based cohort studies (the Religious Orders Study and the Rush Memory and Aging Project). The rules were implemented without informant interviews, neuroimaging, blood work or routine case conferencing. Autopsies were performed at death and the pathologic diagnosis of AD made with a modified version of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) criteria. We compared the positive predictive value of the clinical diagnosis in the two community-based studies to the positive predictive value of the clinical diagnosis of AD made by standard clinical practice in a clinic-based cohort study using AD pathology as the gold standard. Of 306 clinic cases with probable AD, 286 (93.5%) met CERAD neuropathologic criteria for AD; the results were comparable for those with possible AD (51 of 54, 94.4%). Of 141 study subjects with probable AD, 130 (92.2%) met CERAD neuropathologic criteria for AD; the results were lower but acceptable for those with possible AD (26 of 37, 70.3%). The results were similar in secondary analyses using alternate neuropathologic criteria for AD. The clinical diagnosis of AD can be made in community-based studies without the use of informant interviews, neuroimaging, blood work or routine case conferencing. This approach holds promise for reducing the operational costs of epidemiologic studies of aging and AD.