Laura Coots Daras, Melvin J. Ingber, Anne Deutsch, Jennifer Gaudet Hefele, Jennifer Perloff

To examine whether there are differences in inpatient rehabilitation facilities (IRFs') all-cause, 30-day post-discharge hospital readmission rates by organizational characteristics and geographic regions. Observational study. We analyzed Medicare claims and administrative data sources for Medicare fee-for-service beneficiaries discharged from all IRFs nationally (N=1,166) in 2013 and 2014. We applied specifications for an existing quality measure adopted by CMS for public reporting that assesses all-cause unplanned hospital readmissions for 30 days post-discharge from inpatient rehabilitation. We estimated facility-level observed and risk-standardized readmission rates and then examined variation by several organizational characteristics (facility type, profit status, teaching status, proportion of low-income patients, size) and geographic factors (rural/urban, census division, and state). The mean IRF risk-standardized hospital readmission rate was 13.00 percent (SD 0.77). After controlling for organizational characteristics and practice patterns, we found substantial variation in IRFs' readmission rates: for-profit IRFs had significantly higher readmission rates compared to not-for-profit IRFs (p<0.001). We also found geographic variation: IRFs in the South Atlantic and South Central census regions had the highest hospital readmission rates compared to IRFs in New England that had the lowest rates. Our findings point to variation in the quality of care, as measured by risk-standardized hospital readmission rates following IRF discharge. Thus, monitoring of readmission outcomes is important to encourage quality improvement in discharge care planning, care transitions and follow-up.