The Comparative Effectiveness of Inpatient Versus Skilled Nursing Facility Rehabilitation Using Linked Acute Stroke Registry Data.

Abstract

Objectives To examine the comparative effectiveness of rehabilitation at inpatient rehabilitation facilities (IRFs) versus skilled nursing facilities (SNFs) on home time-a proxy of functional recovery among patients with acute stroke. Design A retrospective comparative effectiveness cohort of hospitalized acute stroke (International Classification of Disease-10 I61-I63) cases identified by the Michigan Stroke Registry and probabilistically linked to Medicare fee-for-service (FFS) data. We quantified the comparative effectiveness of IRF versus SNF using inverse probability treatment weighted multivariable linear models. Setting IRF and SNF admissions from 31 registry hospitals between 2016 and 2020. Participants FFS Medicare patients with acute stroke. Interventions Not applicable. Main outcome measures Ninety-day and 1-year home time (number of days alive and outside of inpatient care) calculated at 2 different starting points: from the time of discharge from acute hospital care (traditional home time) or the time of discharge from IRF or SNF care (referred to as amended home time). Results We identified 5943 hospitalized Medicare FFS beneficiaries discharged to IRF (n=2995) or SNF (n=2948). After inverse probability treatment weighted adjustment, the mean traditional home time for patients in IRF was 11.1 days (95% confidence interval [CI], 9.5-12.57) longer at 90 days, and 46.3 days (95% CI, 39.8-52.9) longer at 1 year, compared with patients in SNF. However, when amended home time was used, the mean difference in 90-day home time disappeared (0.5d; 95% CI, -1.1 to 2.1) but remained significant at 1 year (35.7d; 95% CI, 29.1-42.2). Conclusions IRF care was associated with longer traditional home time both at 90 days and 1 year. However, amended home time showed that the difference at 90 days was explained by the longer length of stay at SNF facilities. Future studies should avoid using traditional home time as a means of comparing patient outcomes between IRF and SNF settings.