Objective: The purpose of this investigation was to replicate the statistical approach used in a previous investigation (Toronto study) within a French population to determine the best predictive model for Alzheimer's disease (AD).
Method: Data from neuropsychological tests from two prospective studies were entered into a regression model.
Results: Replication of the statistical approach in the Montpellier sample produced a three-test model with a specificity of 99% and sensitivity of 73%. This model consisted of a delayed auditory verbal recall test, a construction test, a category fluency test and provides probability estimates for the transition to dementia in individual cases.
Conclusion: The models derived from these two longitudinal studies provide an empirical basis for the selection of tests for the definition of mild cognitive impairment of the Alzheimer type (MCI-A). The small set of tests derived are suitable for use in general practice.