From the Duke Clinical Research Institute, Durham, NC (Y.X., L.T., L.L., J.J.F., L.E.W., E.D.P., J.P.B.); Department of Neurology, Duke University Medical Center, Durham, NC (Y.X.); Department of Gynecology and Obstetrics, Johns Hopkins Hospital, Baltimore, MD (J.J.F.); Department of Neurology, Wake Forest Baptist Health, Winston Salem, NC (C.D.B., P.W.D.); Stroke Service and Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston (L.H.S.); Department of Rehabilitation and Regenerative Medicine, Columbia University College of Physicians and Surgeons, New York, NY (J.S.); Department of Rehabilitation Medicine, Weill Cornell Medical College, New York, NY (J.S.); Department of Rehabilitation Medicine, NewYork-Presbyterian Hospital (J.S.); Division of Cardiology, David Geffen School of Medicine at University of California, Los Angeles (G.C.F.); Physical Medicine and Rehabilitation, Durham Veterans Administration Medical Center, NC (H.H.); Division of Geriatrics, Department of Medicine, Duke University Medical Center, Durham, NC (H.H.); Department of Rehabilitation, Duke Regional Hospital, Durham, NC (C.M.); Division for Heart Disease and Stroke Prevention, Centers for Disease Control and Prevention, Atlanta, GA (M.G.G.); and School of Nursing, University of North Carolina, Wilmington, NC (B.J.L.).
Checking for direct PDF access through Ovid
Background and Purpose—Rehabilitation is recommended after a stroke to enhance recovery and improve outcomes, but hospital’s use of inpatient rehabilitation facilities (IRFs) or skilled nursing facility (SNF) and the factors associated with referral are unknown.Methods—We analyzed clinical registry and claims data for 31 775 Medicare beneficiaries presenting with acute ischemic stroke from 918 Get With The Guidelines-Stroke hospitals who were discharged to either IRF or SNF between 2006 and 2008. Using a multilevel logistic regression model, we evaluated patient and hospital characteristics, as well as geographic availability, in relation to discharge to either IRF or SNF. After accounting for observed factors, the median odds ratio was reported to quantify hospital-level variation in the use of IRF versus SNF.Results—Of 31 775 patients, 17 662 (55.6%) were discharged to IRF and 14 113 (44.4%) were discharged to SNF. Compared with SNF patients, IRF patients were younger, more were men, had less health-service use 6 months prestroke, and had fewer comorbid conditions and in-hospital complications. Use of IRF or SNF varied significantly across hospitals (median IRF use, 55.8%; interquartile range, 34.8%–75.0%; unadjusted median odds ratio, 2.59; 95% confidence interval, 2.44–2.77). Hospital-level variation in discharge rates to IRF or SNF persisted after adjustment for patient, clinical, and geographic variables (adjusted median odds ratio, 2.87; 95% confidence interval, 2.68–3.11).Conclusions—There is marked unexplained variation among hospitals in their use of IRF versus SNF poststroke even after accounting for clinical characteristics and geographic availability.Clinical Trial Registration—URL: https://clinicaltrials.gov. Unique identifier: NCT02284165.