The Mex-Cog cognitive assessment battery: discriminant analysis of the cognitive performance profile in older adults

Abstract

Objective. To analyze the cognitive profile of a clinical sample using the Mex-Cog cognitive battery and establish which cognitive measures and domains contribute most to group separation. Materials and methods. A group of 145 older adults previously diagnosed with dementia (n= 47), mild cognitive impairment MCI (n= 47), or as cognitively normal (n= 51) were assessed with the Mex-Cog cognitive battery. Six linear discriminant analyses (LDA) were estimated to compare dementia vs. cognitively normal, MCI vs. cognitively normal, and MCI vs. dementia, using ten individual measures and six cognitive domains. We used a leave-one-out cross-validation procedure to evaluate the predictive capacity of LDA models. Results. Discriminant functions using individual measures and domains distinguished correctly 100% of dementia and cognitively normal groups showing a memory and executive function profile. The predictive group membership for MCI versus cognitively normal varied between 82 and 85%, with a cognitive profile associated with attention-executive function followed by memory. Group separation between MCI and dementia was between 80 and 87%, characterized by orientation, memory, and visuospatial abilities. Conclusions. The Mex-Cog cognitive battery is useful for identifying cognitive impairment in older adults.