Measuring Inpatient Rehabilitation Facility Quality of Care: Discharge Self-Care Functional Status Quality Measure
OBJECTIVE: This paper describes the calculation and psychometric properties of the discharge self-care functional status quality measure implemented in the Centers for Medicare & Medicaid Services' (CMS) Inpatient Rehabilitation Facility (IRF) Quality Reporting Program (QRP) on October 1, 2016.
DESIGN: Medicare fee-for-service (FFS) patients from 38 IRFs that participated in CMS' Post-Acute Care Payment Reform Demonstration were included in this cohort study. Data came from the Continuity Assessment Record and Evaluation (CARE) Item Set, IRF-Patient Assessment Instrument (IRF-PAI), and Medicare claims. For each patient, we calculated an expected discharge self-care score, risk-adjusted for demographic and baseline clinical characteristics. Each IRF's performance score equaled the percentage of patient stays where the observed discharge self-care score met or exceeded the expected score. We assessed the measure's discriminatory ability across IRFs and reliability.
PARTICIPANTS: Medicare FFS patients, 21 years and older.
MAIN OUTCOME MEASURES: Facility-level discharge self-care quality measure performance score.
RESULTS: 4,769 patient stays were included; 57% were female, 12.1% under 65 years. Stroke was the most common diagnosis (21.8%). The performance score mean (SD) was 55.1% (16.6%), range was 25.8% to 100%. About 54% of IRFs had scores significantly different from the percentage of stays that met or exceeded the expected discharge self-care score in the overall demonstration sample. The quality measure showed strong reliability, with intra-class correlation coefficients of 0.91.
CONCLUSIONS: The discharge self-care quality measure showed strong discriminatory ability and reliability, representing an important initial step in evaluation of IRF self-care outcomes. A wide range in performance scores suggested a gap in quality of care across IRFs. Future work should include testing the measure with nationwide data from all IRFs.