Extracorporeal cardiopulmonary resuscitation for refractory in-hospital cardiac arrest: A retrospective cohort study

Abstract

BackgroundExtracorporeal cardiopulmonary resuscitation (ECPR) is a rescue therapy for refractory cardiac arrest, but its high mortality has raised questions about patient selection. No selection criteria have been proposed for patients experiencing in-hospital cardiac arrest. We aimed to identify selection criteria available at the time ECPR was considered for patients with in-hospital cardiac arrest. We analyzed data of in-hospital cardiac arrest patients undergoing ECPR in our extracorporeal membrane oxygenation (ECMO) center (March 2007 to March 2019). Intensive care unit (ICU) and 1-year survival post-hospital discharge were assessed. Factors associated with ICU survival before ECPR were investigated. An external validation cohort from a previous multicenter study was used to validate our results. ResultsAmong the 137 patients (67.9% men; median [IQR] age, 54 [43–62] years; low-flow duration, 45 [30–70] min) requiring ECPR, 32.1% were weaned-off ECMO. Their respective ICU- and 1-year survival rates were 21.9% and 19%. Most 1-year survivors had favorable neurological outcomes (cerebral performance category score 1 or 2). ICU survivors compared to nonsurvivors, respectively, were more likely to have a shockable initial rhythm (53.3% versus 24.3%; P < 0.01), a shorter median (IQR) low-flow time (30 (25–53) versus 50 (35–80) min, P < 0.01) and they more frequently underwent a subsequent intervention (63.3% versus 26.2%, P < 0.01). The algorithm obtained by combining age, initial rhythm and low-flow duration discriminated between patient groups with very different survival probabilities in the derivation and validation cohorts. ConclusionSurvival of ECPR-managed in-hospital cardiac arrest patients in this cohort was poor but hospital survivors' 1-year neurological outcomes were favorable. When deciding whether or not to use ECPR, the combination of age, initial rhythm and low-flow duration can improve patient selection.