Mental status testing in the elderly nursing home population

The clinical utility of selected brief cognitive screening instruments in detecting dementia in an elderly nursing home population was examined. One hundred twenty nursing home residents (mean age 87.9) were administered the Mini-Mental State Exam (MMSE) and the Modified Mini-Mental State Exam (3MS). The majority of the subjects (75%) were also administered the Dementia Rating Scale (DRS). Both clinically diagnosed demented (n = 57) and non-demented (n = 63) subjects participated in the study. Dementia was diagnosed in accordance with DSM-III-R criteria by physicians specializing in geriatric medicine. Using standard cutoffs for impairment, the 3MS, MMSE, and DRS achieved high sensitivity (82% to 100%) but low specificity (33% to 52%) in the detection of dementia among nursing home residents. Positive predictive values ranged from 52% to 61%, and negative predictive values from 77% to 100%. Higher age, lower education, and history of depression were significantly associated with misclassification of non-demented elderly subjects. Receiver Operating Characteristic (ROC) curve analyses revealed optimal classification of dementia with cutoff values of 74 for the 3MS, 22 for the MMSE, and 110 for the DRS. The results suggest that the 3MS, MMSE, and DRS do not differ significantly with respect to classification accuracy of dementia in a nursing home population. Elderly individuals of advanced age (i.e., the oldest-old) with lower education and a history of depression appear at particular risk for dementia misclassification with these instruments. Revised cutoff values for impairment should be employed when these instruments are applied to elderly residents of nursing homes and the oldest-old.