Psychometric Refinement of the Cognitive Assessment in Music Therapy for Institutionalized Older Adults (ACPIM-5)

Introduction: This study psychometrically refined the Cognitive Assessment of Older Adults in Music Therapy (ACPIM) and evaluated its convergent validity and longitudinal stability relative to the Montreal Cognitive Assessment (MoCA). Method: Participants included 180 institutionalized older adults (mean age = 79.0 ± 8.2 years; 93.9% cognitively impaired). Psychometric refinement integrated Network Psychometrics, Rasch Partial Credit Modeling, and Confirmatory Factor Analysis. Results: MoCA items required category collapsing to resolve disordered thresholds. Refinement yielded a unidimensional 5-item scale (ACPIM-5) demonstrating adequate fit (χ²(4) = 4.039, p = 0.401; CFI = 1.000; TLI = 1.000; RMSEA = 0.007, 90% CI [0.000, 0.113]; ω = 0.896) and high fidelity with the 18-item version (r = 0.930). A bifactor model confirmed to be a dominant General Factor (H = 0.896; factor determinacy = 0.928) convergent with the MoCA (r = 0.380). ROC analysis demonstrated acceptable preliminary discrimination (AUC = 0.768, 95% CI [0.593, 0.943]; cutoff = 12.5; sensitivity = 72.8%; specificity = 81.8%).Discussion: The ACPIM-5 satisfied 7-day metric invariance (Δχ²(5) = 3.418, p = 0.636), confirming factor loading equivalence. In contrast, the MoCA violated scalar invariance (Δχ²(20) = 43.896, p = 0.0015), indicating threshold drift. The ACPIM-5 provides a context grounded, rapid screening metric for music therapy with institutionalized older adults.