RESUMO
BACKGROUND: Studies of dementia in populations avoid many of the selection biases in clinical samples but require special evaluation and diagnostic methods to obtain high participation rates. To address this issue, we developed a unique in-home dementia assessment. We assessed validity of these assessments using neuropathologic confirmation of the clinical diagnosis in 3 epidemiologic samples. METHODS: Subjects were 175 participants in 3 ongoing studies of dementia. Two were population based and identified dementia by cognitive screening. The third study sought volunteers via advertisements. Dementia evaluations were then conducted at the participants' residences by specially trained nurses and psychometricians. Evaluation results were interpreted, and preliminary diagnoses were assigned by a geropsychiatrist or neurologist and a psychologist. Final diagnoses were assigned by a consensus panel of neurologists, geropsychiatrists, and psychologists. We compared the clinical diagnoses with the gold-standard neuropathologic diagnoses for those participants who subsequently underwent autopsy. RESULTS: Among the demented, the sensitivity of a clinical diagnosis of probable or possible Alzheimer's disease (AD) was 93% across the 3 studies. The rate of overall diagnostic agreement was 81%. Measures of agreement did not differ meaningfully across varying levels of dementia severity. CONCLUSIONS: Rates of neuropathologic confirmation for clinical AD diagnoses in these studies were similar to those reported from clinic-based samples. These results support the validity of clinical diagnoses of AD from a structured in-home assessment of community dwelling and institutionalized individuals using relatively economical methods of dementia screening and assessment.