Answers are grounded in this paper and cite their sources.
Introduction: Accountable care and bundled payment initiatives have increased attention on inpatient post-acute care effectiveness. Empirical evidence for whether care at inpatient rehabilitation (IRF) or skilled nursing facilities (SNF) is associated with differential outcomes is limited. Methods: Data from the American Heart Association Get With The Guidelines-Stroke registry were linked with Medicare claims. We compared post-discharge mortality and a composite of all-cause rehospitalization or death for beneficiaries who received IRF or SNF care immediately after hospital discharge. Proportional hazards regression with inverse propensity weighting (IPW) was used to adjust for measured differences in demographic, clinical and hospital characteristics. Acknowledging the possibility of unmeasured differences between groups, an instrumental variable approach (IV=hospital-specific proportion of patients discharged to IRFs vs SNFs) was used to adjust for both measured and unmeasured confounding. Results: From 1,142 hospitals there were 34,574 AIS patients discharged to IRFs and 34,430 to SNFs. Patients receiving SNF vs IRF care were older (83 vs 79 years), had more comorbid illness, and higher unadjusted 7, 90, and 365 day mortality and rehospitalizations (Table). After IPW adjustment mortality and rehospitalization/mortality rates remained significantly better for patients with IRF vs SNF care (Table). Differences attenuated with IV analysis but IRF care continued to be associated with better outcomes (Table). Findings were similar in sensitivity analyses among patients with a NIHSS score. Conclusions: Receiving care at an IRF after AIS was associated with lower mortality and all-cause rehospitalization/death compared with SNF care even after rigorous adjustment for both measured and unmeasured differences in treatment selection. Further research is needed to discern the mechanisms for these potential differences between settings.
No takes yet. Share an insight, caveat, or question.
Bettger et al. (2015) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: