- Has inpatient rehabilitation facility (IRF) quality improved since 2004?
- How have quality changes and case-mix changes interacted over the period?
Quality metrics play an increasingly important role in the evaluation and reimbursement of post-acute providers. Currently, it is difficult to ascertain whether changes in inpatient rehabilitation facility (IRF) patient outcomes are due to changes in treatment or the case mix of patients seen in IRFs. Risk adjustment, however, has the potential to improve the comparability of quality metrics both across providers and over time. This report (1) develops risk-adjusted quality metrics at the provider level for IRFs, (2) develops methods to address low case volume, and (3) uses these metrics to estimate national trends in IRF quality from 2004 to 2009. It presents the results for five IRF outcomes: (1) functional gain, (2) discharge to the community, (3) 30-day readmission to acute care given discharge to the community, (4) 30-day readmission to skilled nursing facility (SNF), given discharge to the community, and (5) discharge directly to acute care.
Raw Outcome Rates for Most of the Quality Measures Studied Worsened over the Study Period
- The Functional Independence Measure (FIM) gain improved by about two points, but discharge to the community, 30-day readmission to acute care facility, 30-day readmission to a skilled nursing facility, and discharge directly to acute care all worsened.
- The declining raw rates were caused by a worsening case mix.
But Increasing Severity in Case Mix Masks Real Improvements in Quality for Each of the Outcomes Considered
- After adjusting for case mix, the researchers found that quality improved on every metric; in the case of FIM gain, quality improved by more than the improvement in the raw rate.
- Quality at individual inpatient rehabilitation facilities (IRFs) persisted over the study period.
- Risk-adjustment models should be used as the basis for revealing overall quality trends in the market.
- Because risk-adjustment parameter estimates change over time, the models should be recalibrated regularly and, if feasible, year-specific models should be used.
Table of Contents