Background: 

The use of extracorporeal membrane oxygenation (ECMO) in coronavirus disease 2019 (COVID-19) for refractory respiratory failure, severe cardiac dysfunction, and bridge to lung transplantation has been steadily increasing during the ongoing global pandemic.

 Objective: 

Our meta-analysis aims to compare the clinical characteristics between COVID-19 survivors and nonsurvivors requiring ECMO support. 

Methods: 

A systematic search of Pubmed, Cochrane, Embase, Scopus, and Web of Science databases was performed between December first, 2019, to June first, 2021. Studies with comparative data of COVID-19 ECMO patients were selected, in which clinical characteristics and complications were assessed. 

Results: 

Sixteen cohort studies involving 706 COVID-19 patients requiring ECMO support with pooled mortality rate of 40% were included. Younger age (mean 51 years vs 55 years; P < .001), fewer comorbidities (23% vs 31%; odds ratio [OR] 0.55; P  =  .02), and less renal replacement therapy (RRT) (21% vs 39%; OR 0.41; P  =  .007) and vasopressor (76% vs 92%; OR 0.35; P  =  .008) requirement were demonstrated in COVID-19 survivors requiring ECMO support than nonsurvivors. Survivors also had higher pre-ECMO pH (mean 7.33 vs 7.26; P < .001) than nonsurvivors. No difference was observed in gender, body mass index, duration of mechanical ventilation (MV) before ECMO support initiation, total ECMO support duration, and pre-ECMO parameters of PaO2/FiO2 ratio, tidal volume (mL/kg), positive end-expiratory pressure, and plateau pressure. The rate of bleeding complications was lower in survivors (32% vs 59%; OR 0.36; P  =  .001) than nonsurvivors, but no difference was observed in thromboembolism and secondary infections. 

Conclusions: 

We found advanced age, multiple comorbidities, lower pre-ECMO pH, greater RRT, and vasopressor requirements, and bleeding are predictors of death in COVID-19 patients requiring ECMO support. The duration of MV before ECMO support initiation and total ECMO support duration was similar among survivors and nonsurvivors. Our study results have important clinical implications when considering ECMO support in critically ill COVID-19 patients.