Consistency of clinical diagnosis in a community-based longitudinal study of dementia and Alzheimer's disease.

We evaluated the consistency of the diagnosis of dementia in a multicultural, longitudinal community-based study of cognitive impairment and dementia. We diagnosed dementia using a fixed neuropsychological paradigm; the diagnosis also required historical evidence of functional impairment. In a sample of 656 subjects with at least one annual follow-up examination, dementia was confirmed at 1 year in 89% of the 304 subjects initially demented, and in 90% of the 136 subjects with the initial diagnosis of probable Alzheimer's disease (AD). The 34 initially demented subjects who failed to meet criteria for dementia at follow-up included 13 with an initial diagnosis of probable AD. All 34 still had evidence of cognitive impairment; this group was more likely to have a history of pulmonary disease, multiple medication use, or chronic alcohol use than other demented patients. Consistency of dementia diagnosis did not vary according to educational attainment or ethnic background. The use of a neuropsychological paradigm such as ours in large longitudinal studies of dementia may minimize interobserver diagnostic variability or diagnostic drift over time while contributing the benefits of a comprehensive cognitive evaluation to the diagnostic process.