Outcomes in pediatric ECPR for in-hospital cardiac arrest: an ELSO registry analysis

Background: Utilization of in-hospital extracorporeal cardiopulmonary resuscitation (ECPR) in pediatrics has increased significantly, with concurrent improvements in survival. Despite these advances, there remains considerable variability in the criteria for pediatric ECPR candidacy. This study aims to identify the patient demographics as well as pre-, peri-, and post-arrest characteristics associated with ECPR outcome. Methods: This is a retrospective study of patients up to 18 years old with witnessed in-hospital cardiac arrest and ECPR from January 2020 until October 2024 reported to the Extracorporeal Life Support Organization (ELSO) Registry. The primary outcome was a composite measure including survival to hospital discharge, heart transplantation, or placement of permanent ventricular assist device. Univariate and multivariate logistic regression, as well as Kaplan Meier and Joint-Model analysis were performed. Results: A total of 1,903 patients were analyzed in the study, with 1,410 (74.1 %) presenting with cardiac precipitating events, 358 (18.8 %) with non-cardiac events, and 135 (7.1 %) with unknown causes. Overall, 788 patients (41.4 %) achieved a favorable composite outcome, including 641 (45.5 %) in the cardiac group, 100 (27.9 %) in the non-cardiac group and 47 (34.8 %) in the unknown group. On univariate analysis, non-cardiac event (OR 0.46 [0.36–0.60]), longer cardiopulmonary resuscitation (CPR) duration (OR 0.98 [0.98–0.99]), non-shockable rhythm (OR 0.62 [0.49–0.79]), and higher lactate (OR 0.95 [0.93–0.97]) were associated with decreased odds of favorable outcome, whereas signs of life (OR 1.52 [1.22–1.89]) and higher pH (OR 3.26 [2.03–5.27]) were associated with increased odds of favorable outcome. Independent predictors of increased odds of favorable outcome on multivariate analysis included higher pH at 24 h (OR 10.69 [1.46–78.36]), whereas variables associated with decreased odds of favorable outcome included history of prior ECMO run (OR 0.37 [0.22–0.62]), lung disease (OR 0.37 [0.16–0.85]), renal replacement therapy (OR 0.39 [0.16–0.94]), higher PaCO₂ prior to ECMO (OR 0.99 [0.99–0.99]), higher lactate at 24 h (OR 0.80 [0.75–0.85]), and longer CPR time (OR (0.99 [0.99–0.99]). Elevated lactate tertiles at all points (pre-ECPR, 6 h, and 24 h) were associated with lower rates of favorable outcome by Kaplan-Meier (p log-rank < 0.0001) and Joint model analyses (p = 0.018). Conclusion: Pediatric ECPR is a complex, resource-intensive intervention impacted by institutional expertise, patient selection, arrest characteristics, and post-ECPR management. We highlight several prognostic variables that may be useful in determining ECPR candidacy.