In-hospital extracorporeal cardiopulmonary resuscitation for patients with out-of-hospital cardiac arrest: an analysis by time-dependent propensity score matching using a nationwide database in Japan

Abstract

Background: Extracorporeal cardiopulmonary resuscitation (ECPR) has been proposed as a rescue therapy for patients with refractory cardiac arrest. This study aimed to evaluate the association between ECPR and clinical outcomes among patients with out-of-hospital cardiac arrest (OHCA) using risk-set matching with a time-dependent propensity score. Methods: This was a secondary analysis of the JAAM-OHCA registry data, a nationwide multicenter prospective study of patients with OHCA, from June 2014 and December 2019, that included adults (≥ 18 years) with OHCA. Initial cardiac rhythm was classified as shockable and non-shockable. Patients who received ECPR were sequentially matched with the control, within the same time (minutes) based on time-dependent propensity scores calculated from potential confounders. The odds ratios with 95% confidence intervals (CI) for 30-day survival and 30-day favorable neurological outcomes were estimated for ECPR cases using a conditional logistic model. Results: Of 57,754 patients in the JAAM-OHCA registry, we selected 1826 patients with an initial shockable rhythm (treated with ECPR, n = 913 and control, n = 913) and a cohort of 740 patients with an initial non-shockable rhythm (treated with ECPR, n = 370 and control, n = 370). In these matched cohorts, the odds ratio for 30-day survival in the ECPR group was 1.76 [95%CI 1.38–2.25] for shockable rhythm and 5.37 [95%CI 2.53–11.43] for non-shockable rhythm, compared to controls. For favorable neurological outcomes, the odds ratio in the ECPR group was 1.11 [95%CI 0.82–1.49] for shockable rhythm and 4.25 [95%CI 1.43–12.63] for non-shockable rhythm, compared to controls. Conclusion: ECPR was associated with increased 30-day survival in patients with OHCA with initial shockable and even non-shockable rhythms. Further research is warranted to investigate the reproducibility of the results and who is the best candidate for ECPR. Graphical Abstract: [Figure not available: see fulltext.]