Relationship between the Montreal Cognitive Assessment and Mini-mental State Examination for assessment of mild cognitive impairment in older adults

Abstract

Background: The Montreal Cognitive Assessment (MoCA) was developed to enable earlier detection of mild cognitive impairment (MCI) relative to familiar multi-domain tests like the Mini-Mental State Exam (MMSE). Clinicians need to better understand the relationship between MoCA and MMSE scores.

Methods: For this cross-sectional study, we analyzed 219 healthy control (HC), 299 MCI, and 100 Alzheimer's disease (AD) dementia cases from the Alzheimer's Disease Neuroimaging Initiative (ADNI)-GO/2 database to evaluate MMSE and MoCA score distributions and select MoCA values to capture early and late MCI cases. Stepwise variable selection in logistic regression evaluated relative value of four test domains for separating MCI from HC. Functional Activities Questionnaire (FAQ) was evaluated as a strategy to separate dementia from MCI. Equi-percentile equating produced a translation grid for MoCA against MMSE scores. Receiver Operating Characteristic (ROC) analyses evaluated lower cutoff scores for capturing the most MCI cases.

Results: Most dementia cases scored abnormally, while MCI and HC score distributions overlapped on each test. Most MCI cases scored ≥ 17 on MoCA (96.3%) and ≥ 24 on MMSE (98.3%). The ceiling effect (28-30 points) for MCI and HC was less using MoCA (18.1%) versus MMSE (71.4%). MoCA and MMSE scores correlated most for dementia (r = 0.86; versus MCI r = 0.60; HC r = 0.43). Equi-percentile equating showed a MoCA score of 18 was equivalent to MMSE of 24. ROC analysis found MoCA ≥ 17 as the cutoff between MCI and dementia that emphasized high sensitivity (92.3%) to capture MCI cases. The core and orientation domains in both tests best distinguished HC from MCI groups, whereas comprehension/executive function and attention/calculation were not helpful. Mean FAQ scores were significantly higher and a greater proportion had abnormal FAQ scores in dementia than MCI and HC.

Conclusions: MoCA and MMSE were more similar for dementia cases, but MoCA distributes MCI cases across a broader score range with less ceiling effect. A cutoff of ≥ 17 on the MoCA may help capture early and late MCI cases; depending on the level of sensitivity desired, ≥ 18 or 19 could be used. Functional assessment can help exclude dementia cases. MoCA scores are translatable to the MMSE to facilitate comparison.

Fig. 1

Scatterplots for MMSE and MoCA scores shown by diagnostic group. Graphs are for all subjects (a), dementia only (b), MCI only (c) and HC only (d). Pearson correlation coefficients between MMSE and MoCA scores are shown for each graph. Vertical lines denote MMSE standard cutoff of 24 points and horizontal lines denote different proposed MoCA cutoffs for MCI (17, 19 and 23). Note that symbols may represent more than one case at that score

Fig. 2

MoCA mapped to MMSE scores using equi-percentile equating method with log-linear smoothing in 618 subjects. Lines in graphs indicate MMSE cutoff of 24 and MoCA cutoffs of 17 and 19. The MoCA value equivalent to a MMSE cutoff of 24 is shaded

Fig. 3

ROC analysis graph of MoCA scores for distinguishing MCI subjects (n = 299) from the AD dementia group (n = 100) and table for MoCA values to consider as lower cutoff values for MCI (17–20) depending on sensitivity and specificity levels preferred in a given situation