Evolving outcomes of extracorporeal membrane oxygenation during the first 2 years of the COVID-19 pandemic: a systematic review and meta-analysis

Abstract

Background: Extracorporeal membrane oxygenation (ECMO) has been used extensively for coronavirus disease 2019 (COVID-19)-related acute respiratory distress syndrome (ARDS). Reports early in the pandemic suggested that mortality in patients with COVID-19 receiving ECMO was comparable to non-COVID-19-related ARDS. However, subsequent reports suggested that mortality appeared to be increasing over time. Therefore, we conducted an updated systematic review and meta-analysis, to characterise changes in mortality over time and elucidate risk factors for poor outcomes.

Methods: We conducted a meta-analysis (CRD42021271202), searching MEDLINE, Embase, Cochrane, and Scopus databases, from 1 December 2019 to 26 January 2022, for studies reporting on mortality among adults with COVID-19 receiving ECMO. We also captured hospital and intensive care unit lengths of stay, duration of mechanical ventilation and ECMO, as well as complications of ECMO. We conducted random-effects meta-analyses, assessed risk of bias of included studies using the Joanna Briggs Institute checklist and evaluated certainty of pooled estimates using GRADE methodology.

Results: Of 4522 citations, we included 52 studies comprising 18,211 patients in the meta-analysis. The pooled mortality rate among patients with COVID-19 requiring ECMO was 48.8% (95% confidence interval 44.8-52.9%, high certainty). Mortality was higher among studies which enrolled patients later in the pandemic as opposed to earlier (1st half 2020: 41.2%, 2nd half 2020: 46.4%, 1st half 2021: 62.0%, 2nd half 2021: 46.5%, interaction p value = 0.0014). Predictors of increased mortality included age, the time of final patient enrolment from 1 January 2020, and the proportion of patients receiving corticosteroids, and reduced duration of ECMO run.

Conclusions: The mortality rate for patients receiving ECMO for COVID-19-related ARDS has increased as the pandemic has progressed. The reasons for this are likely multifactorial; however, as outcomes for these patients evolve, the decision to initiate ECMO should include the best contextual estimate of mortality at the time of ECMO initiation.

Keywords: Coronavirus disease 2019; Extracorporeal membrane oxygenation; Meta-analysis; Mortality; Severe acute respiratory syndrome coronavirus 2.

Fig. 1

Preferred Reporting Items for Systematic Reviews and Meta-analyses flow diagram

Fig. 2

Forest plot demonstrating the pooled mortality among patients receiving extracorporeal membrane oxygenation for coronavirus disease 2019

Fig. 3

Funnel plot post random-effects trim-and-fill (R₀ estimator) analysis. Bubbles that are black in fill represent the studies included in the meta-analysis, bubbles that are hollow represent filled-in studies based on the trim-and-fill estimator

Fig. 4

Forest plot demonstrating the pooled mortality among patients receiving extracorporeal membrane oxygenation for coronavirus disease 2019 stratified by time period based on the date of final patient enrolment

Fig. 5

World map demonstrating number of studies, patients, and pooled outcomes for each ELSO region reporting on mortality of extracorporeal membrane oxygenation for coronavirus disease 2019

Fig. 6

Meta-regression plot demonstrating the association of age, the date of final patient enrolment, the proportion of patients receiving corticosteroids, and the duration of extracorporeal membrane oxygenation, with mortality. Bubble sizes are inverse-variance weighted, and correspond to the variances of each study, i.e. as the variance decreases, bubble size increases. In the meta-regression of mortality and time (top right), the virus icons refer to new SARS-CoV-2 variants - Black: Beta (May 2020); Blue: Alpha (Sep 2020); Orange: Delta (Oct 2020); Yellow: Omicron (Nov 2020)

Fig. 7.

Three-dimensional linear plot demonstrating the association between age, date of final patient enrolment, and mortality. Bubble sizes are inverse-variance weighted and correspond to the variances of each study, i.e. as the variance decreases, bubble size increases. The 3-dimensional sheet follows a rainbow palette: dark red represents a higher mortality rate, while dark blue represents a lower mortality rate