Objectives: To examine the accuracy of clinical diagnoses of Alzheimer's disease (AD) in subjects enrolled in the Rochester Alzheimer's Disease Project (RADP) who were examined at autopsy, and to present a list of clinical "red flags."
Design: Autopsy examination of both prospective and retrospective subjects consecutively enrolled in this clinicopathologic study of the RADP.
Setting: University hospital and research center, using a multidisciplinary geriatric neurology clinic, satellite clinics, nursing home visits, and home visits.
Patients: One hundred seventy subjects clinically diagnosed as having AD who were enrolled in the RADP between 1983 and 1993 underwent neuropathologic examination. Of these, 93 had been enrolled prospectively and 77 retrospectively.
Main outcome measures: Agreement between clinical and pathologic diagnoses.
Results: One hundred forty-nine subjects of 170 clinically diagnosed as having AD fulfilled the pathologic criteria for AD, yielding an accuracy rate of 88%. Of 93 subjects enrolled prospectively and diagnosed as having AD, 83 (90%) met the histologic criteria for AD. Of the 77 subjects enrolled retrospectively, neuropathologic examination indicated definite AD in 66 (86%).
Conclusions: There was a high correlation between clinicians' diagnoses and final pathologic diagnoses. The most common clinical errors involved the misdiagnosis of dementias due to Parkinson's disease and cerebrovascular disease. There was no significant difference in the accuracy rates of subjects enrolled prospectively and retrospectively.