Variation in 30-Day Readmission Rates Among Medically Complex Patients at Inpatient Rehabilitation Facilities and Contributing Factors

Objective Few studies have focused on 30-day readmission rates in inpatient rehabilitation facilities (IRFs) and factors that contribute to this. The purpose of this study was to examine the variation in 30-day readmission rates among medically complex patients at IRFs nationally and explore how patient and facility characteristics are associated with this variation. Design Retrospective review of an administrative database. Setting IRFs throughout the United States. Participants Patients from the medically complex impairment group treated at IRFs nationwide represented in the Uniform Data System for Medical Rehabilitation (UDS_MR) from 2002 to 2011. Measurements Using multilevel logistic regression analysis, 16 patient and facility characteristics were examined to identify and adjust for variables with a significant effect on readmission rates. Unadjusted and adjusted readmission rates were plotted by facility rank. Facilities were grouped by decile, based on readmission rates, and the mean unadjusted and adjusted rates for the lowest and highest deciles were compared to quantify the variation due to adjustment. Patient and facility characteristics for the highest and lowest deciles were compared. Results A total of 117,156 medically complex patients from 682 IRFs nationwide were included. Patients were 46% male and 84% white with an average age of 71.5 years. Nine of 16 characteristics were identified that significantly increased the odds of readmission, including older age, male gender, poorer admission motor function, longer duration of impairment, higher Elixhauser comorbidity index, unemployed or retired status, larger facility size, lower mean facility admission motor function, and eastern or western geographic area. The average unadjusted and adjusted readmission rates for all facilities were 15.63% and 15.86%, respectively. The unadjusted readmission rates for the lowest and highest deciles were 6.71% and 26.48%, respectively. After adjustment, this narrowed to 10.33% and 21.91%, respectively. Patient and facility characteristics accounted for 41% of the variation seen in the readmission rates for these groups. Facilities with the highest readmission rates (highest decile) more commonly cared for patients at highest risk for readmission: unemployed, male patients with higher comorbidity index and poorer motor function on admission. Conclusion This study shows significant variation in readmission rates for medically complex patients across IRFs. However, nearly half (41.4%) of this variation was attributed to 9 patient and facility characteristics, suggesting the need for risk adjustment if readmission rates are to be used as a quality indicator for IRFs.