Prognostic indicators of survival and survival prediction model following extracorporeal cardiopulmonary resuscitation in patients with sudden refractory cardiac arrest

Sung Woo Lee, Kap Su Han, Jong Su Park, Ji Sung Lee, Su Jin Kim

Extracorporeal cardiopulmonary resuscitation (ECPR) has been considered in selected candidates with potentially reversible causes during a limited period. Candidate selection and the identification of predictable conditions are important factors in determining outcomes during CPR in the emergency department (ED). The objective of this study was to determine the key indicators and develop a prediction model for survival to hospital discharge in patients with sudden cardiac arrest who received ECPR. This retrospective analysis was based on a prospective cohort, which included data on CPR with ECPR-related variables. Patients with sudden cardiac arrest who received ECPR at the ED from May 2006 to June 2016 were included. The primary outcome was survival to discharge. Prognostic indicators and the prediction model were analyzed using logistic regression. Out of 111 ECPR patients, there were 18.9% survivors. Survivors showed younger age, shorter CPR duration (p < 0.05) and had tendencies of higher rate of initial shockable rhythm (p = 0.055) and higher rate of any ROSC event before ECPR (p = 0.066) than non-survivors. Eighty-one percent of survivors showed favorable neurologic outcome at discharge. In univariate analysis, the following factors were associated with survival: no preexisting comorbidities, initial serum hemoglobin level ≥14 g/dL, and mean arterial pressure ≥60 mmHg after ECPR. Based on multivariate logistic regression, predictors for survival in ECPR were as follows: age ≤56 years, no asystole as the initial arrest rhythm, CPR duration of ≤55 min, and any return of spontaneous circulation (ROSC) event before ECPR. The prediction scoring model for survival had a c-statistic of 0.875. With careful consideration of differences in the inclusion criteria, the prognostic indicators and prediction scoring model for survival in our study may be helpful in the rapid decision-making process for ECPR implementation during CPR in the ED.