Predicting Performance on the MMSE

To the Editor:—We read with interest the article entitled “Predicting Performance on the Mini-Mental State Examination” by Magaziner et al.1 The conclusion, suggesting use of this abbreviated form of the Mini-Mental Status Exam (MMSE) only for predicting continuous scores rather than for categorizing patients as impaired or not impaired, limits its primary use as a screening tool for dementia. Furthermore, as suspected by the authors, when we applied the equations used for predicting total MMSE scores to a primarily black population as reported below, the equations were not found to be reliable predictors of actual scores. The MMSEs administered in the past 18 months at the Kings County Hospital Center (KCHC) Outpatient Geriatrics Clinic were reviewed and the four age by education-specific predictive equations applied. The age of the 100 patients involved ranged from 62-99, with 89% of the patients being black, 6% hispanic, and 5% white. Results of the actual MMSE score minus predicted MMSE score were colated for the 96 patients at least age 65, with the results between 1 and −1 considered correct predictions. The results were analyzed by age, education, and race. The results for race are shown in Table 1. The equations gave slightly better predictions for younger and more highly educated black patients, but in the best group 50% of the scores were still overestimated, at least 40% by 4 or more points. As can be seen, these results give far from the expected bell-shaped curve in this predominantly black population. The MMSE is specifically designed as a screening tool to assess a patient's cognitive state as well as to separate patients with cognitive impairment from those without impairment.2 If the abbreviated form described by Magaziner et al cannot be used to categorize patients as impaired versus not impaired, it has little utility. Thus, there is no additional benefit from the time saving but merely further limitations on its usefullness as a screening tool. In addition, our data suggest that when used on a predominantly black population, the abbreviated form is a poor predictor of actual MMSE scores. In summary, we feel that the seven item version of the MMSE is neither a useful nor an accurate short-cut for assessing mental status.