Purpose: The APPROACH cardiovascular surgical intensive care unit (CVICU) readmission score has excellent discrimination and calibration for CVICU readmission after discharge to a surgical ward; however, it has not been prospectively validated.
Material and methods: In a prospective consecutive cohort of 805 patients ≥18 years admitted to the CVICU after coronary artery bypass and/or valvular surgery, the APPROACH CVICU readmission score was calculated at the time of discharge to a surgical ward. The study compared observed versus predicted CVICU readmission and the model discrimination was evaluated using AUC c-index. The incremental prognostic utility of 6 pre-specified prospectively collected respiratory (re-intubation, tracheostomy, oxygen at discharge) and hemodynamic variables (heart rate, systolic blood pressure, inotropes at discharge) were tested using net reclassification index (NRI) and integrated discrimination improvement (IDI).
Results: A total of 37 (4.6%) patients were readmitted to the CVICU. The median CVICU length of stay (9.0 vs 2.0 days, p < .001) and all-cause in-hospital mortality (8.1% vs 0.4%, p < .001) was higher among readmitted patients. The model had good discrimination (c-index = 0.748). Systolic blood pressure at discharge yielded the largest improvement in model discrimination (c-index = 0.782; Hosmer-Lemshow p = .749).
Conclusions: In a prospective validation cohort, the APPROACH CVICU readmission risk score had good discrimination and could be operationalized in future research and clinical practice.
Keywords: Cardiac surgery; Intensive care unit; Prediction model; Readmission; Validation.
Copyright © 2019 Elsevier Inc. All rights reserved.