Effect of education on the mini-mental state examination as a screening test for dementia.

We studied whether Mini-Mental State Examination (MMSE) norms for detecting dementia in elderly outpatients vary according to educational attainment. Subjects were 109 elderly outpatients with Alzheimer's dementia and 100 non-demented outpatient controls. Receiver operating characteristics (ROC) of the MMSE were examined among three strata of educational attainment: middle school, high school, and college/graduate school. MMSE ROC curve areas were .95-.96 in the three educational strata. Assuming a dementia prevalence of 10%-30%, the most accurate lower limits of normal for MMSE scores and their attendant sensitivities and specificities were 21 for middle school (.82/.94), 23 for high school (.79/.97), and 24 for college/graduate school (.83/1.00) attainment. These norms accurately classified over 90% of subjects in all three educational strata. We conclude that education-specific norms optimize performance of the MMSE as a screening test for Alzheimer's dementia in elderly outpatients.