Extracorporeal Membrane Oxygenation Support in COVID-19

Quick Takes

  • Critically ill patients with COVID-19 who received care at centers with extensive experience performing extracorporeal membrane oxygenation (ECMO) had an estimated in-hospital mortality rate of 37.4% at 90 days after ECMO initiation.
  • Among the COVID-19 registry subpopulation with acute respiratory distress syndrome (ARDS), the 38% mortality rate is similar to that cited previously in patients with ARDS from other causes.
  • Factors significantly associated with greater risk of mortality were advanced age, underlying chronic lung disease, immunocompromised conditions, severe hypoxia (low PaO2:FiO2 ratio), acute kidney injury, cardiac arrest, and initial use of circulatory support.

Study Questions:

What mortality incidence has been observed among patients diagnosed with coronavirus disease 2019 (COVID-19) within 90 days of receiving extracorporeal membrane oxygenation (ECMO) support during hospitalization at Extracorporeal Life Support Organization (ELSO) member centers, and what factors are independently associated with outcome?


Data from adult patients with a diagnosis of COVID-19 who received ECMO support between January 16 and May 1, 2020 in the ELSO registry were evaluated. The primary outcome was time-to-event in-hospital mortality within 90 days of ECMO initiation; multivariate Cox models were used to determine associations between outcome and both patient- and institution-level factors. Covariates included baseline (age, race, sex, presence of one or more comorbid conditions), and acute patient-level conditions prior to cannulation (PaO2:FiO2 ratio, acute respiratory distress syndrome [ARDS], duration of tracheal intubation, acute kidney injury, cardiac arrest, and initial mode of ECMO support), as well as hospital-level characteristics (2019 ECMO case volume and geographic region).


  • A total of 1,035 patients ≥16 years of age from 213 participating hospitals in 37 countries met analysis criteria, with outcomes reported as median (interquartile range) unless otherwise stated.
  • Patients were 49 (41-57) years of age, with a PaO2:FiO2 ratio of 72 (range, 59-94) mm Hg prior to ECMO initiation.
  • Estimated cumulative 90-day in-hospital mortality was 37.4 (34.4-40.4)%. Among patients meeting criteria for ARDS, the mortality rate was 38.0 (34.6-41.5)%.
  • Within 90 days, 37% of patients died, 6% remained hospitalized, and the rest were discharged to long-term care facilities (10%), other hospitals (17%), or home (30%).
  • Time from intubation to ECMO initiation was 4.0 (1.8-6.4) days, and median ECMO duration was 13.9 (7.8-23.3) days. Hospital length-of-stay was 31.1 (21.0-46.0) days versus 16.0 (6.8-27.6) days in patients who did or did not survive.
  • Age was an independent predictor of mortality; compared to ages 16-39 years, the hazard ratio (HR) (95% confidence interval [CI]) was 1.76 (1.23-2.52), 2.28 (1.42-3.67), and 3.07 (1.58-5.95) in ages 50-59, 60-69, and ≥70 years. Mortality was also independently associated with immunocompromised conditions and previous chronic respiratory disease (whereas not specifically with asthma).
  • Acute factors independently associated with mortality (HR [95% CI]) were cardiac arrest (1.92 [1.32-2.78]), acute kidney injury (1.38 [1.08-1.76]), and initial use of circulatory support, i.e., venoarterial (VA) ECMO (1.89 [1.20-1.97]). Overall, 96% of patients received venovenous (VV) ECMO support only. Lesser severity of hypoxemia was associated with lower mortality risk (HR, 0.68 [0.57-0.81] per doubling of PaO2:FiO2).
  • Mortality within 90 days of ECMO initiation did not differ statistically by region or by the volume of ECMO cases performed in 2019 at each hospital.


In critically ill patients with COVID-19 infection given ECMO at ELSO member hospitals, 30% survived to be discharged home, 37% died, and the remainder continued to require hospitalization or long-term care within 90 days after initiation of ECMO support. Greater 90-day mortality was associated with advanced age, cardiac arrest, and use of circulatory support (VA ECMO).


This 90-day mortality rate of 38% in the critically ill COVID-19 subpopulation with respiratory failure and ARDS receiving ECMO is similar to previously reported mortality rates in patients with ARDS from causes other than COVID-19 receiving ECMO. The magnitude of ECMO benefit for patients with COVID-19 overall is greater than many had anticipated at the start of the pandemic. Although no statistical association between ECMO case volume and outcome could be demonstrated among these ELSO registry hospitals, the authors emphasize that all participating centers were highly experienced, and the apparent lack of case volume versus outcome association should not be extrapolated to centers without extensive, established ECMO experience. Another challenging limitation of these findings is lack of a control arm to determine degree of outcome differences in similar patients without ECMO support.

Clinical Topics: Arrhythmias and Clinical EP, Cardiac Surgery, Heart Failure and Cardiomyopathies, Invasive Cardiovascular Angiography and Intervention, Prevention, Implantable Devices, SCD/Ventricular Arrhythmias, Cardiac Surgery and Arrhythmias, Cardiac Surgery and Heart Failure, Acute Heart Failure

Keywords: Acute Kidney Injury, Cardiac Surgical Procedures, Coronavirus, COVID-19, Critical Illness, Extracorporeal Membrane Oxygenation, Heart Arrest, Heart Failure, Hospital Mortality, Length of Stay, Lung Diseases, Primary Prevention, Respiratory Distress Syndrome, severe acute respiratory syndrome coronavirus 2

< Back to Listings