Prognostic factors and mortality predictive scale in patients treated with in-hospital extracorporeal cardiopulmonary resuscitation

Abstract Background Extracorporeal cardiopulmonary resuscitation (ECPR) is a rescue option in cardiac arrest (CA) refractory to advanced life support. Patient selection is challenging, evidence is limited and randomized studies are lacking regarding in-hospital CA. Purpose Our objective was the identification of factors associated with survival in in-hospital ECPR. Methods Retrospective analysis of consecutive cases of venoarterial extracorporeal membrane oxygenator (VA-ECMO) for ECPR indication in a referral center. The ECPR alert includes in-hospital cardiologist/intensivist, on-call interventional cardiologist/cardiac surgeon for cannulation, and on-call perfusionist. We analyzed baseline variables related to CA and ECPR technique and studied their relationship with in-hospital survival. Results Out of 207 VA-ECMO implants between 2014-feb 2023, we selected the ECPR cases: N=59 (28.5%). Baseline, admission and ECPR characteristics are collected in the table. ECMO-VA support intention was bridge to recovery in 68% of cases, transplant 5%, ventricular assist device 5% and decision 22%. The most frequent access was percutaneous femoro-femoral, and ECPR was unsuccessful (no flow, no rhythm, or impossibility of cannulation in 11.8% of cases). All patients underwent temperature control at 36ºC with ECMO. Overall in-hospital survival was 18.6% (n=11) (cerebral performance cathegory 1-2), and the most frequent causes of death were refractory shock/multiple organ failure (49%) and anoxic encephalopathy (25%). The patients who died were significantly younger, had peripheral vascular disease (PAD), higher creatinine, LDH and lactate levels, longer CA duration, and its ECMO implantation was more frequent during on-call hours (table). Lactate >10.05 predicted mortality (AUC 0.693, Sens 67,3%, Esp 80%). The time under VA-ECMO support and mechanical ventilation, and the use of adrenaline infusion were more frequent in the survivors (table). At multivariate analysis, a longer CA duration (HR 0.947, CI95% 0.900-0.977, p0,040) was the only factor independently associated with in-hospital mortality. We developed a predictive scale including factors related to survival (lactate >10.05, >60 minutes CA, PAD, on-call time) which predicted 100% with 3 or more factors (p