Objective: To propose cognitive indices for the PPMI cognitive battery (PPMI-CB).

Background: Cognitive batteries provide meaningful information about Parkinson-related cognitive impairment but are complicated when tests are interpreted in isolation, producing unwieldy analyses with potential for spurious findings. Conversely, individual tests have error and limited sensitivity to change, requiring larger samples or treatment effects to detect reliable differences. These concerns can be addressed with integrative techniques like factor analysis.

Method: Participants were 4,546 PPMI individuals at baseline, mostly men (54.2%), older (M_age=65, SD=8.9), with higher education (M_years=16.3, SD=3.4). The PPMI-CB included: Letter-Number Sequencing, Symbol Digit Modalities, Trail Making, Judgment of Line Orientation, letter fluency, category fluency, and Hopkins Verbal Learning. A random 25% was used for exploratory factor analysis (EFA) and 75% for confirmatory factor analysis (CFA). EFAs used ordinary least squares with oblimin rotation. CFAs used robust Satorra-Bentler estimation with confirmatory fit index, Tucker-Lewis index, standardized root mean square residual, and root mean square error of approximation.

Results: EFAs indicated one- to four-factor models could fit the data. The one-factor model had loadings from all tasks. Memory, executive-processing, and language factors emerged across two-, three-, and four-factor solutions. Spatial and working memory tasks had inconsistent or isolated loadings in three- and four-factor solutions. CFA cross-validation revealed best fit for the three-factor solution (memory, processing-executive, language), followed by both two-factor solutions (memory vs. non-memory; memory vs. processing-executive). The one-factor solution had the worst and insufficient fit.

Conclusion: We tested and cross-validated a parsimonious cognitive approach with the PPMI-CB. Individuals interested in cognitive outcomes can implement fewer, more powerful models instead of unwieldy, inconsistently powered analyses. Future research should explore hierarchical structures to understand poor single-factor fit and consistency of proposed structures across groups.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/proposed-cognitive-indices-for-the-parkinsons-progression-markers-initiative/