Validation and Diagnostic Utility of the Mini-Mental State Examination and Montreal Cognitive Assessment in Screening for Dementia within a Mixed Clinical Sample
MetadataShow full item record
The Mini-Mental Status Examination (MMSE) and Montreal Cognitive Assessment (MoCA) are frequently utilized cognitive screening measures. The goal of the present study was to evaluate: (1) diagnostic utility values (e.g., sensitivity, specificity) of each measure, (2) cutoffs that maximize diagnostic accuracy within a mixed clinical sample, (3) the effect of base rates and severity of cognitive impairment on the efficacy of the screening measures, and (4) the relationship of the screening measure subscores to similar neuropsychological measures. The study included 218 veterans who completed the MMSE, MoCA, and neuropsychological testing. Empirically derived cutoffs across criterion variables – performance at least 1SD or 2SD below average on at least one neuropsychological domain, or dementia versus non-dementia diagnosis -- showed less than 24 and 25 as optimal for the MMSE with sensitivities ranging from 0.32 to 0.44 and specificities ranging from 0.78 to 0.87. Optimal cutoffs for the MoCA were 20, 21, and 25 with sensitivities ranging from 0.44 to 0.73 and specificities ranging from 0.57 to 0.83. Across criterion variables, the area under the receiver operating characteristic (ROC) curve (AUC) with the MMSE total score ranged between 0.59 and 0.70. The AUC of the MoCA ranged between 0.69 and 0.72, which was significantly greater than the MMSE when classifying patients based on the criterion of at least 1SD neuropsychological impairment. The MMSE and MoCA subtest scores showed poor convergent and discriminant validity relative to performance on neuropsychological domains, which indicates poor subscore interpretability. The study provides evidence that use of either the MMSE or MoCA increases classification accuracy beyond the base rate of dementia, although, of the two screening instruments, the MoCA has a relative advantage for classification accuracy at mild levels of neuropsychological impairment.