Inpatient Stroke Rehabilitation Lengths of Stay in Canada Derived From the National Rehabilitation Reporting System, 2008 and 2009

OBJECTIVE: To determine rehabilitation length of stay (LOS) for patients with stroke in Canada, and to evaluate which factors contribute to variations in LOS. DESIGN: A retrospective cohort study of Canadians rehabilitating from stroke using medical, functional, and sociodemographic variables extracted from the Canadian Institute for Health Information's National Rehabilitation Reporting System. SETTING: Canadian rehabilitation hospitals providing stroke rehabilitation services. PARTICIPANTS: Patients with stroke (N=11,983) admitted to rehabilitation hospitals from January 2008 through December 2009. INTERVENTIONS: None. MAIN OUTCOME MEASURES: Rehabilitation LOSs were calculated nationally and regionally. Regression models incorporating sociodemographic and clinical measures were constructed to test their effect on LOS. RESULTS: The median stroke rehabilitation LOSs was 35 days (quartiles: 20d, 54d). LOSs varied regionally within Canada. A multivariable regression model including age, FIM motor function scores at admission, and geographic region explained 20% of the variation in LOSs. Modeling these data using a Function-Related Groups case-mix model explained 16% of the variation in LOSs. CONCLUSIONS: FIM motor function scores at admission along with age and geographic region best predicted rehabilitation LOS. These variables explained 20% of the variation in LOSs. Despite regional differences in LOSs, patient characteristics were similar between regions. Other nonpatient factors not captured in these data may contribute to a greater extent in determining stroke rehabilitation LOS.