Clinical predictors of in-hospital mortality in adult patients with refractory cardiogenic shock or cardiac arrest supported with veno-arterial extracorporeal membrane oxygenation

Introduction

Venoarterial (VA) extracorporeal membrane oxygenation (ECMO) is a life-support modality used as a rescue intervention in patients with refractory cardiogenic shock (CS) or cardiac arrest (CA).

Although several predictors of survival in VA-ECMO have been described, clinical variables have received less attention. Therefore, we analysed a single-centre cohort from a high-complexity cardiovascular centre in a low-income country.

Purpose

To identify pre-cannulation clinical variables, ECMO run characteristics, and post-cannulation predictors of in-hospital mortality in VA-ECMO patients with refractory CS or CA.

Methods

A retrospective analysis was conducted including consecutive adult patients with refractory CS or CA requiring VA-ECMO between February 2014 and August 2025. Data were collected from a prospectively maintained ventricular assistance database. The primary outcome was in-hospital mortality and the secondary endpoint was ECMO mortality. Relevant clinical variables were analysed, including pre-cannulation data, ECMO characteristics, and run-related variables. Independent risk factors for in-hospital and ECMO mortality were identified through multivariate analysis using stepwise logistic regression. A p < 0.05 was considered as statistically significant.

Results

During the study period, 130 patients required VA-ECMO due to refractory CS or CA. Sixty one percent were male, median age was 58 (IR 25-75: 47 - 64), 38.6% had hypertension and 27% a history of myocardial infarction. The median left ventricular ejection fraction was 38% (IR 25-75: 23 - 60). Peripheral cannulation was used in 86.7%, and 18.6 % of VA ECMO runs were for extracorporeal cardiopulmonary resuscitation. In-hospital mortality was 56.9%, and ECMO mortality was 41.5%. Median ECMO run was 6 days (IR 25-75: 3-9). Male sex was an independent protective factor (OR 0.4, 95% CI 0.18–0.92, p=0.03) of in-hospital mortality. After multivariable analysis, venting strategy for left ventricle unloading (OR 5.8, 95% CI 1.8-18.3, p=0.003), major bleeding (OR 2.9, 95% CI 1.2-6.8, p=0.01) and requirement of haemodialysis (OR 5.8, 95% CI 1.7-19, p=0.004) were ECMO run predictors of in-hospital mortality. Placement of a left ventricular vent (OR 0.4, 95% CI 0.18–0.92, p=0.03), heart transplant (OR 0.26, 95% IC 0.08-0.8, p=0.018), haemodialysis (OR 5.1, 95% IC 1.4-18.2, p=0.01), and run infections (OR 0.2, 95% CI 0.07-0.6, p=0.004) were associated with ECMO mortality. Prolonged mechanical ventilation was a protective factor for both in-hospital and ECMO mortality.

Conclusions

In our cohort, venting strategy for left ventricular unloading, major bleeding and haemodialysis requirement were independent ECMO-related predictors of in-hospital mortality; whereas male sex and prolonged mechanical ventilation were protective factors. Venting strategy, hemodialysis requirement, infections and heart transplant were independently associated with ECMO mortality.