Patterns and predictors of rehabilitation therapy among older patients with advanced cancer admitted to nursing homes: A SEER-Medicare analysis.

Abstract

6585 Background: Functional impairments affect > 40% of hospitalized patients (pts) with advanced cancer. After hospital discharge, about 20% of pts received rehabilitation (rehab) in nursing homes (NHs) to maintain functional independence. There is evidence from broad pt cohorts that Medicare Prospective Payment (PP) financially incentivizes NHs to provide extra rehab. This study examines rehab utilization among pts with advanced cancer admitted to NHs. Methods: The 2011-2016 SEER-Medicare data were linked with NH Minimum Data Set 3.0 data, which includes sociodemographic and clinical characteristics at admission. Study cohort included traditional Medicare pts with stage IV breast, lung, and colorectal cancer who were admitted to NHs after hospital discharge. Outcomes: total weekly rehab minutes of physical therapy, occupational therapy, and speech-language pathology; ultra-high rehab (≥720 min/wk); and rehab within 10 minutes of threshold (720-730 min/wk). Function and cognition were assessed by Activities of Daily Living (ADL) [7 domains; total score ranges 0 to 28 (higher = dependent)] and Cognitive Function Scale (intact, mild, moderate, severe impairment). Charlson Comorbidity Index (CCI) and survival from NH admission were computed. Generalized linear mixed models examined predictors of rehab outcomes adjusting for NH random effects. Results: A total of 7,453 pts were included (mean age 78.0, 85.8% White, 74.1% lung/ 16.1% colorectal/ 9.7% breast cancer; 76.1% had surgery, 8.9% had chemotherapy; mean CCI 1.9). The mean ADL score was 18.0, with on average 4.7 impairments; 40.2% reported ≥ mild cognitive impairment. Pts received on average 498 (SD = 245) min/wk rehab, but the distribution was trimodal. The number of pts who received 720-730 min/wk rehab was 2.7 times of the secondary peak at 500-510. From 2011-2016, the proportion of pts receiving ultra-high therapy (19.5%-48.4%) and within-threshold rehab (11.0%-32.0%) more than doubled. Only 5.9% of pts were documented on admission as having a life expectancy < 6 months, yet 32.1% and 74.3% died in 30 days and 6 months, respectively. Multivariable regressions indicate that compared to pts with ≥6 months’ expectancy, those with < 6 months’ expectancy received less rehab (β = -117.6), especially ultra-high rehab (odds ratio = 0.31). Pts with cognitive impairments received less rehab. Conclusions: Rehab utilization in older NH pts with advanced cancer mirrors patterns found in broader cohorts. Under PP, rehab minutes provided strongly followed payment thresholds. Over 5 years, more pts were provided 720-730 min/wk rehab, and 1/3 of these pts were at the end of life. Poor prognostication might contribute to the use of ultra-high rehab. Future work should evaluate whether the new Patient Driven Payment Model avoids excessive rehab use in patients with limited life expectancies.