Thirty patients with dementia defined by DSM-III-R criteria (Alzheimer's disease (22), vascular dementia (3), Parkinson's disease, frontal lobe dementia, possible diffuse Lewy body dementia, normal pressure hydrocephalus and uncertain diagnosis), with scores below 24 points in the Mini-Mental Status Examination and more than 4 years of education were submitted to a neuropsychological evaluation. The scores in the neuropsychological tests were compared to those obtained by thirty normal volunteers paired for age, sex and education. Sensivity, specificity and accuracy of the tests in the distinction of demented and normal volunteers were determined. The accuracies were calculated using ROC curves. Blessed's information-memory-concentration test showed greatest accuracy, followed by copy of simple figures, delayed memory of 10 figures (after 5 minutes), recognition of 10 figures and verbal fluency test (animals). A linear discriminant function, composed by 6 tests: visual perception, incidental memory, delayed memory (after 5 minutes), drawing of a clock, verbal fluency (animals) and calculation tests, was able to discriminate all controls from patients and only one patient was wrongly classified as normal control. These tests were chosen because they can be applied in less than 10 minutes and are very easy to interpret. This discriminant function must be applied in another group of patients and controls in order to demonstrate its value. When associated to the MMSE it may be useful to discriminate patients with dementia from normal people in epidemiological studies.
dementia; evaluation; neuropsychological tests