RESUMO
BACKGROUND AND PURPOSE: To validate a Short Form of the Mini-Mental State Examination (SMMSE) as a screening test for dementia in older ambulatory individuals followed in a memory clinic for a memory complaint. METHODS: A total of 202 cognitively healthy individuals, 100 individuals with a mild cognitive impairment and 304 demented individuals sent for a memory complaint by their primary care physician to a memory clinic were prospectively included in this cross-sectional study. They were randomized into derivation (n = 303) and validation (n = 303) groups. The SMMSE score was built from six memory items of MMSE, with a score ranging from 0 to 6 (i.e. best performance). RESULTS: The receiver operating characteristic curve showed an area under the curve of 0.98 for the derivation group and 0.97 for the validation group without differences between curves (P = 0.254). The cut-off between the sensitivity and the specificity of the SMMSE score for clinically diagnosed dementia was ≤4. The performance of the SMMSE for the diagnosis of dementia was high in the derivation and validation groups: sensitivity at 93.1% and 93.8%, specificity at 93.8% and 90.5%, positive predictive value at 94.3% and 90.1%, negative predictive value at 92.5% and 94.0%, likelihood ratio of positive test at 14.9 and 9.8 and of negative test at 0.07 and 0.07, respectively. CONCLUSIONS: The Short Form of the Mini-Mental State Examination was a good screening test for dementia in older individuals followed in a memory clinic for a memory complaint. The next step should be the confirmation of its discriminative value in older primary care patients.