The Mini-Mental State Examination score and the clinical diagnosis of dementia.

To set a working cutoff score for the referral to diagnostic examination, we evaluated 150 consecutive patients with complaints potentially related to dementia, using the Mini-Mental State Examination (MMSE). All patients were later given a complete, standardized work-up and diagnosis as part of our Alzheimer's Disease Patient Registry protocol. Dementia diagnosis was made, consistent with accepted criteria, by consensus of the physicians and psychologist. Diagnosis was reaffirmed after 1-year follow-up exam; 133 of the 150 original patients completed follow-up (80 dementia, 53 no dementia). We evaluated the initial MMSE score compared with the follow-up diagnosis. Sensitivity, specificity, and predictive values were calculated for MMSE scores ranging from 22 through 29. The conventional cutoff score of < 24 shows a sensitivity of 0.63 and a specificity of 0.96; sensitivity increased at higher cutoff scores. Multivariate analysis showed that educational level added significant prediction only at scores of > or = 27. We conclude that an MMSE score of 26 or 27 should be used as a cutoff score in symptomatic populations with similar educational and socioeconomic backgrounds when the goal is to miss few true cases. Population surveys where the expected prevalence is low may require a different cutoff score to indicate the need for further diagnostic evaluation.