Validation of EPIC's Readmission Risk Model, the LACE+ Index and SQLape as Predictors of Unplanned Hospital Readmissions

Overview

The primary objective of this study is to externally validate the EPIC's Readmission Risk model and to compare it with the LACE+ index and the SQLape Readmission model. As secondary objective, the EPIC's Readmission Risk model will be adjusted based on the validation sample, and finally, it´s performance will be compared with machine learning algorithms.

Introduction: Readmissions after an acute care hospitalization are relatively common, costly to the health care system and are associated with significant burden for patients. As one way to reduce costs and simultaneously improve quality of care, hospital readmissions receive increasing interest from policy makers. It is only relatively recently that strategies were developed with the specific aim of reducing unplanned readmissions by applying prediction models. EPIC's Readmission Risk model, developed in 2015 for the U.S. acute care hospital setting, promises superior calibration and discriminatory abilities. However, its routine application in the Swiss hospital setting requires external validation first. Therefore, the primary objective of this study is to externally validate the EPIC's Readmission Risk model and to compare it with the LACE+ index (Length of stay, Acuity, Comorbidities, Emergency Room visits index) and the SQLape (Striving for Quality Level and analysing of patient expenditures) Readmission model. Methods: For this reason, a monocentric, retrospective, diagnostic cohort study will be conducted. The study will include all inpatients, who were hospitalized between the 1st January 2018 and the 31st of January 2019 in the Lucerne Cantonal hospital in Switzerland. Cases will be inpatients that experienced an unplanned (all-cause) readmission within 18 or 30 days after the index discharge. The control group will consist of individuals who had no unscheduled readmission. For external validation, discrimination of the scores under investigation will be assessed by calculating the area under the receiver operating characteristics curves (AUC). For calibration, the Hosmer-Lemeshow goodness-of-fit test will be graphically illustrated by plotting the predicted outcomes by decile against the observations. Other performance measures to be estimated will include the Brier Score, Net Reclassification Improvement (NRI) and the Net Benefit (NB). All patient data will be retrieved from clinical data warehouses.

Conditions

Interventions

Eligibility

Locations (1)

Outcomes