A Mobility Measure for Inpatient Rehabilitation Using Multigroup, Multidimensional Methods

Inpatient rehabilitation facilities (IRFs) report patient functional status to Medicare and other payers using Quality Indicators (QI). While the QI is useful for payment purposes, its measurement properties are limited for monitoring patient progress. A mobility measure based on QI items and additional standardized assessments may enhance clinicians' ability to track patient improvement. Thus, we developed the Mobility Ability Quotient (Mobility AQ) to assess mobility during inpatient rehabilitation.

For 10 036 IRF inpatients, we extracted assessments from electronic health records, used confirmatory factor analysis to define subdimensions of mobility, and then applied multidimensional item response theory (MIRT) methods to develop a unidimensional construct. Assessments included the QI items and standardized measures of mobility, motor performance, and wheelchair and transfer skills.

Confirmatory factor analysis resulted in good-fitting models (root-mean-square errors of approximation ≤0.08, comparative fit indices, and nonnormed fit indices ≥0.95) for 3 groups defined by anticipated primary mode of locomotion at discharge-walking, wheelchair propulsion, or both. Reestimation as a multigroup, MIRT model yielded scores more sensitive to change compared with QI mobility items (dlast-first = 1.08 vs 0.60 for the QI; dmax-min = 1.16 vs 1.05 for the QI). True score equating analysis demonstrated a higher ceiling and lower floor for the Mobility AQ than the QI.

The Mobility AQ demonstrates improved sensitivity over the QI mobility items. This MIRT-based mobility measure describes patient function and progress for patients served by IRFs and has the potential to reduce assessment burden and improve communication regarding patient functional status.Video Abstract available for more insights from authors (see the Video, Supplemental Digital Content 1, available at: http://links.lww.com/JNPT/A341).