Screening and case finding tools for the detection of dementia. Part I: evidence-based meta-analysis of multidomain tests

Abstract

Aim: To evaluate the diagnostic accuracy of all brief multidomain alternatives to the Mini-Mental State Examination (MMSE) in the detection of dementia.

Methods: A literature search, critical appraisal, and meta-analysis were conducted of robust diagnostic validity studies involving cognitive batteries. Twenty-nine distinct brief batteries were tested in 44 large-scale analyses. Twenty studies took place in specialist settings (11 in memory clinics and 9 in secondary care), ten studies were conducted in primary care, and 14 in the community.

Results: In community settings with a low prevalence of dementia, short screening methods of no more than 10 minutes had an overall sensitivity of 72.0% (95% confidence interval [CI] = 60.4%-82.3%) and a specificity of 88.2% (95% CI = 83.0%-92.5%). The optimal individual tests were the Telephonic interview based on MSQ, Category fluency/Memory impairment screen-Telephonic interview and 6 item Cognitive Impairment Test (6-CIT), but data were limited by the absence of multiple independent confirmation for any individual test. In primary care where the prevalence of dementia is usually modest, the optimal individual tools were the Abbreviated mental test score/Mental status questionnaire (MSQ), and Prueba cognitive de leganes (PCL). Furthermore, the Abbreviated mental test score (AMTS) was superior to the MMSE for case finding, but for screening the MMSE was optimal. If length is not a major consideration, the MMSE may remain the best tool for primary care clinicians who want to rule in and rule out a diagnosis. In specialist settings where the prevalence of dementia is often high, the optimal individual tools were the DEMTECT, Montreal cognitive assessment (MOCA), Memory Alteration test, and MINI-COG. Two tools were potentially superior to the MMSE for rule in and rule out, namely the 6-CIT and MINI-COG. Only four analyses looked specifically at accuracy in early-stage dementia, and each showed at least equivalent diagnostic accuracy, suggesting these methods might be applicable to early identification.

Conclusion: A large number of alternatives to the MMSE have now been validated in large samples with favorable rule-in and rule-out accuracy. Evidence to date suggests for those wishing to use brief battery tests then the original MMSE or the AMTS should be considered in primary care and either the 6-CIT or the MINI-COG should be considered in specialist settings.