Abstract
OBJECTIVES: The aim of this study was to develop and validate a score to accurately predict the probability of death for adult extracorporeal cardiopulmonary resuscitation (ECPR). BACKGROUND: ECPR is being increasingly used to treat refractory in-hospital cardiac arrest (IHCA), but survival varies from 20% to 40%. METHODS: Adult patients with extracorporeal membrane oxygenation for IHCA (ECPR) were identified from the American Heart Association GWTG-R (Get With the Guidelines-Resuscitation) registry. A multivariate survival prediction model and score were developed to predict hospital death. Findings were externally validated in a separate cohort of patients from the Extracorporeal Life Support Organization registry who underwent ECPR for IHCA. RESULTS: A total of 1,075 patients treated with ECPR were included. Twenty-eight percent survived to discharge in both the derivation and validation cohorts. A total of 6 variables were associated with in-hospital death: age, time of day, initial rhythm, history of renal insufficiency, patient type (cardiac vs noncardiac and medical vs surgical), and duration of the cardiac arrest event, which were combined into the RESCUE-IHCA (Resuscitation Using ECPR During IHCA) score. The model had good discrimination (area under the curve: 0.719; 95% CI: 0.680-0.757) and acceptable calibration (Hosmer and Lemeshow goodness of fit P = 0.079). Discrimination was fair in the external validation cohort (area under the curve: 0.676; 95% CI: 0.606-0.746) with good calibration (P = 0.66), demonstrating the model's ability to predict in-hospital death across a wide range of probabilities. CONCLUSIONS: The RESCUE-IHCA score can be used by clinicians in real time to predict in-hospital death among patients with IHCA who are treated with ECPR.