Predicting Discharge Destination with Admission Outcomes Following Stroke

Abstract

Objective(s): To use inpatient rehabilitation facility (IRF) admission outcomes for patients with stroke to predict discharge destination. Design: Retrospective survey study. Setting: Inpatient Rehabilitation Hospital. Participants: Patients (NZ407) hospitalized for an acute stroke. Patients were included in the study if they had a stroke within 6 months of admission to rehabilitation. Patients were excluded if the time between onset and admission was greater than 6 months or if they had a previous stroke. Interventions: N/A. Main Outcome Measure(s): Admission data from the Functional Independence Measure (FIM) and Simplified Stroke Rehabilitation Assessment of Movement (S-STREAM) were used to determine predictive factors for a community discharge. Results: Admission variables that predict discharge destination were identified using logistic regression analysis. Logistic regressions and chi-square analysis were then used to determine the cut off score for each predictive variable and to create a predictive tool. Specifically a score of 12 and above on the S-STREAM, 26 and above on Motor FIM, 47 and above on Total FIM, and 3 and above for the following individual FIM items: FIM Bladder, FIM bed transfer, FIM toilet transfer, FIM bathing, and FIM memory were all predictive of a community discharge. The predictive tool with the greatest sensitivity and specificity combined 4 admission variables (S-STREAM, FIM Bladder, FIM Bed transfer, and FIM memory) and was able to predict a community discharge with a sensitivity of 76% and a specificity of 64%. Conclusions: By using admission outcomes, discharge destination can be predicted with significant sensitivity and specificity. A predictive tool for discharge destination has clinical utility in identifying appropriate patients for acute rehabilitation and in determining the best plan of care for patients at risk of an institutional discharge. Prospective evaluation of this predictive tool is warranted.