Brief screening tests for dementia.

OBJECTIVE To compare brief dementia screening tests as candidates for routine use in primary care practice. METHOD We selected screening tests that met 2 criteria: 1) administration time of 10 minutes or less in studies including individuals with, and without, dementia; and 2) performance characteristics evaluated in at least 1 community or clinical sample of older adults. We compared tests for face validity, sensitivity, and specificity in a clearly defined subject sample; for vulnerability to sociodemographic biases unrelated to dementia; for direct comparison with an accepted standard; for acceptability to patients and doctors; and for brevity and ease of administration, scoring, and interpretation by nonspecialists. RESULTS Thirteen instruments met our inclusion criteria. Very short tests (1 minute or less) proved unacceptable by several criteria. Standard instruments requiring more than 5 minutes to complete, including the best-studied Mini-Mental State Examination (MMSE), were found to be too long for routine application. Several failed other performance tests or could not be adequately assessed. Short tests taking between 2 and 5 minutes that can be administered by nonspecialists with little or no training and are relatively unbiased by language and education level appear to be superior to both shorter and longer instruments. CONCLUSIONS Three tests showed the most promise for broad application in primary care settings: the Mini-Cog, the Memory Impairment Screen, and the General Practitioner Assessment of Cognition (GPCOG). Formal practice intervention trials are now needed to validate the utility of short screens with regard to implementation, effect on rates of diagnosis and treatment of dementia patients, and outcomes for patients, families, and health care systems.