Change in Mobility Quality Measure for Inpatient Rehabilitation Facilities: Exclusion Criteria and the Risk Adjustment Model

Objective: To describe the exclusion criteria and updated risk adjustment model developed for the Change in Mobility quality measure in the inpatient rehabilitation facility (IRF) quality reporting program. Facility-level quality measures focused on patient outcomes usually require risk adjustment to account for varied admission characteristics of patients across facilities. Design: This cohort study analyzed admission demographic and clinical factors associated with mobility change scores using the Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) data for Medicare patients discharged from IRFs in calendar year 2017. Setting: A total of 1129 IRFs in the United States. Participants: A total of 493,209 (N=493, 209) Medicare fee-for-service and Medicare Advantage IRF patient stays discharged in calendar year 2017. Interventions: Not applicable. Main Outcome Measures: Mobility change scores using admission and discharge standardized assessment data from the IRF-PAI. Results: Approximately 53% of patients in the study were female, 67% were aged 65-84 years, and nearly 80% were White. In the final risk adjustment model, 105 covariates were included, explaining 20% of variance in mobility change scores. Key risk adjusters included IRF primary diagnosis group, prior indoor ambulation functioning, age older than 90 years, and 14 of the comorbidities. The model showed good calibration across the range of deciles of predicted IRF mobility change scores; the ratio of the average expected to observed change scores ranged from 0.93-1.03, with all but 1 within ±0.03. Conclusions: The updated risk adjustment model uses IRF patients' demographic and clinical characteristics to predict their mobility change scores. The exclusion criteria and resulting risk model are used to calculate the risk adjusted Change in Mobility quality measure scores, enabling comparisons of Change in Mobility scores across IRFs.