Abstract 2224: The Relationship Between Process and Outcomes for Stroke Post Prospective Payment in Inpatient Rehabilitation Facilities

Abstract

Background and Objective: Previous studies have reported decreasing length of stay (LOS) for inpatient rehabilitation facilities (IRFs), with conflicting effects on discharge Functional Independence Measure (FIM) scores and discharge destination (DD). This study was the first to examine the post prospective payment system (PPS) period using Medicare data drawn from the national Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) database. The purposes were to examine trends over time for process and outcome factors, and to describe the relationships between process, LOS, and outcomes (discharge FIM scores and DD), for Medicare patients with stroke. Methods: The study included 371,211 Medicare Part A beneficiaries aged 65 and older with stroke, admitted to United States (US) IRFs between January 1, 2002 and June 30, 2007. Descriptive statistics and generalized estimating equations (GEE) modeling for clustered data were used for analysis. Continuous GEE evaluated LOS and discharge FIM scores, and binomial GEE evaluated LOS and DD (community verses institution). Covariates in models were: admission FIM scores, age, gender, race/ethnicity, comorbidities, complications, and stroke type. Time interactions with admission FIM scores, LOS, and discharge FIM scores (binomial model only) were examined. Results: During the study period, mean LOS decreased from 17.9 (SD=9.9) to 16.1(SD=8.3) days (p<. 0001), mean discharge FIM scores decreased from 80.1 (SD=24.5) to 76.5 (SD=24.5) points (p<. 0001), and rate of community discharge decreased from 66.6% to 61.2% (p<. 0001). LOS predicted discharge FIM scores (95% CI, .48, .52, p<.0001), but the relationship to community discharge was weak (OR .997, p=.007). Discharge FIM scores predicted discharge destination (OR 1.07, p<.0001). Covariates of admission FIM scores, age, gender, race/ethnicity, comorbidities, complications, and stroke type also predicted outcomes. Time interactions were present for LOS, admission FIM scores, and discharge FIM scores. Conclusions: During the first 5.5 years of PPS, declining trends were found for LOS, discharge FIM scores, and rate of community discharge for Medicare beneficiaries with stroke. LOS was a strong predictor for discharge FIM scores, but weak for DD. Discharge FIM scores were a better predictor of DD than LOS. Effects of covariates on discharge FIM scores and DD have clinical implications for IRF stroke rehabilitation in the US. Because of the reduced time for treatment, dose of IRF rehabilitation for Medicare beneficiaries may not be achieving expected results in the post-PPS period.