Does Implementation of a Previously Validated Prediction Tool Reduce Readmission Rates Into a Medical Intensive Care Unit?

Abstract

PURPOSE: Readmission to the Intensive Care Unit (ICU) is associated with worse outcomes and increased cost. Discharge processes are often fraught with errors and providers are poor judges of the risk of patient readmission. The Stability and Workload Index for Transfer (SWIFT) score is a previously validated tool which predicts unplanned ICU readmission. A score >15 is associated with a 6-15% risk of return to the ICU within 24 hours of discharge. We investigated the impact on discharge plan and patient readmission rates following incorporation of SWIFT score into the discharge planning workflow. METHODS: The SWIFT score and associated percentage risk for readmission were incorporated into daily rounds for the purpose of discharge planning. Providers were encouraged, through education and regular information sessions, to discuss the SWIFT score when it was >15. A record of whether the score was discussed and subsequent decision-making regarding ICU dismissal was made by the bedside nurse. Predefined changes to the discharge plan were recorded by the bedside nurse. RESULTS: Over the implementation period, 1257 patients were discharged alive from the medical ICU. Of these, 473 (38%) had the SWIFT score discussed. Of that group, 44 (9%) had some change in discharge plan; enhanced communication n=12; changed discharge service or location n=23; remained in the ICU n=10. In the group who had SWIFT discussed, 13/473 (2.7%) were readmitted within 24 hours compared to 18/784 (2.3%) in the group with no score discussion. There was no association between SWIFT discussion and readmission within 24 hours (p=0.85). Enhanced communication was significantly associated (p=0.04) with outcome of readmission within 24 hours, in a logistic regression model that also contained SWIFT discussion, change in discharge location/service, as predictors. CONCLUSIONS: Incorporation of the SWIFT score into the discharge workflow led to changes in provider behaviors, without a statistically significant change in readmission rates. Of the provider behaviors which the SWIFT discussion sought to modify, enhanced communication was significantly associated with reduced ICU readmission rates.