The Optimal Short Version of Montreal Cognitive Assessment in Diagnosing Mild Cognitive Impairment and Dementia.

objectivesWhile various short variants of the Montreal Cognitive Assessment (MoCA) have been developed, they have not been compared among each other to determine the most optimal variant for routine use. This study evaluated the comparative performance of the short variants in identifying mild cognitive impairment or dementia (MCI/dementia).

designBaseline data of a cohort study.

settingAlzheimer's Disease Centers across the United States.

participantsParticipants aged ≥50 years (n = 4606), with median age 70 (interquartile range 65-76). MEASURES: Participants completed MoCA and were evaluated for MCI/dementia. The various short variants of MoCA were compared in their performance in discriminating MCI/dementia, using areas under the receiver operating characteristic curve (AUCs).

resultsAll 7 short variants of MoCA had acceptable performance in discriminating MCI/dementia from normal cognition (AUC 87.7%-91.0%). However, only 2 variants by Roalf et al (2016) and Wong et al (2015) demonstrated comparable performance (AUC 88.4-88.9%) to the original MoCA (AUC 89.3%). Among the participants with higher education, only the variant by Roalf et al had similar AUC to the original MoCA. At the optimal cut-off score of <25, the original MoCA demonstrated 84.4% sensitivity and 76.4% specificity. In contrast, the short variant by Roalf et al had 87.2% sensitivity and 72.1% specificity at its optimal cut-off score of <13. CONCLUSIONS/IMPLICATIONS: The various short variants may not share similar diagnostic performance, with many limited by ceiling effects among participants with higher education. Only the short variant by Roalf et al was comparable to the original MoCA in identifying MCI or dementia even across education subgroups. This variant is one-third the length of the original MoCA and can be completed in <5 minutes. It provides a viable alternative when it is not feasible to administer the original MoCA in clinical practice and can be especially useful in nonspecialty clinics with large volumes of patients at high risk of cognitive impairment (such as those in primary care, geriatric, and stroke prevention clinics).