Refining MoCA Interpretation in An Outpatient Clinic Via Logistic Regression and Base Rate Adjustment: A Pilot Study

Objectives: The Montreal Cognitive Assessment (MoCA) is a screening measure for mild cognitive impairment (MCI) and major neurocognitive disorder (MNCD).Typical cutoff scores for the MoCA are 26 for MCI and 18 for MNCD; however; overreliance on cutoff scores can result in misdiagnosis.Our study used logistic regression analyses and predictive probabilities at a range of base rates to view MoCA scores along a continuum rather than a dichotomous variable.Methods: Data from 49 memory clinic outpatients were analyzed.Logistic regressions and base rate adjustments were used to provide predicted probabilities of no diagnosis, MCI, and MNCD for possible scores on the MoCA at different base rates.Results: Using cumulative logistic regression, 11 cases were predicted to have no diagnosis (22.4%), 25 to have MCI (51%), and 13 to have MNCD (26.5%).Using standardized beta coefficients, scores of 24, 18-23, and 17 resulted in predictive probabilities of .50 or greater when diagnosing normal cognition (NC), MCI, and MNCD, respectively.Finally, predictive probabilities were calculated for MoCA scores at diagnosis base rates ranging from 10% to 90%.Conclusions: Cutoff scores commonly used to assist with interpreting MoCA results may limit its diagnostic utility due to treating cognitive impairment as a dichotomous variable.Our study highlights how logistic regression analyses can be used to refine MoCA interpretation by viewing scores along a continuum and accounting for base rates of a condition of