Journal Logo

Clinical Investigations

Extracorporeal Membrane Oxygenation for Coronavirus Disease 2019-Induced Acute Respiratory Distress Syndrome: A Multicenter Descriptive Study*

Yang, Xiao MD1; Cai, Shuhan MD1; Luo, Yun MD1; Zhu, Fangfang MD1; Hu, Ming MD2; Zhao, Yan MD3; Zheng, Ruiqiang MD4; Li, Xuyan MD5; Hu, Bo MD1; Peng, Zhiyong MD1

Author Information
doi: 10.1097/CCM.0000000000004447
  • Free
  • COVID-19


The coronavirus virus disease 2019 (COVID-19), occurred in Wuhan, China, in December 2019 and subsequently caused an ongoing outbreak from Wuhan, now spread globally. According to the National Health Commission of the People’s Republic of China, the COVID-19 patients in Wuhan had more severe disease in comparison with other regions of China. The recent report demonstrated that the rate of critical illness among patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was about 26%, and critically ill patients had mortality about 61.5% (1,2). Most COVID-19 patients usually develop severe pneumonia and have a high risk of acute respiratory distress syndrome (ARDS) (3). Patients with ARDS have a mortality rate of near 50% (4). In a recently published study from Wuhan, the 28-day mortality of mechanical ventilation (MV) COVID-19 patients was 81% (2). Extracorporeal membrane oxygenation (ECMO) can support gas exchange for patients with ARDS. ECMO was found to be an effective management choice during influenza A (H1N1) outbreaks in 2009 (5–8). However, it remains unclear whether ECMO is effective in treating SARS-CoV-2 pneumonia associated ARDS. Thus, in this study, we aimed to present the clinical characteristics, ECMO-related variables, and outcomes from patients who received ECMO support for COVID-19-related ARDS.


Design, Setting, and Patients

The institutional ethics boards of Zhongnan Hospital of Wuhan University (number 2020020) and Wuhan Pulmonary Hospital (number 2020020) approved this case-series. Patients or families provided informed consent for data analysis with anonymized individual data.

We screened all consecutive adult ICU admissions with SARS-CoV-2 related moderate to severe ARDS in Zhongnan Hospital of Wuhan University and Wuhan Pulmonary Hospital from January 8, 2020, to March 31, 2020. The diagnose of SARS-CoV-2 pneumonia was confirmed by both chest scan and real-time reverse transcription-polymerase chain reaction assays, according to the World Health Organization Interim Guidelines (9). The diagnosis of ARDS was defined based on the Berlin definition (3). We included patients who were intubated and ventilated including those receiving ECMO support in the final analysis.

Data Collection

The medical records of patients were analyzed by the research team of the Department of Critical Care Medicine, Zhongnan Hospital of Wuhan University. We collected demographics, medical history, underlying comorbidities, laboratory findings, imaging studies, vital signs, medications, need for continuous renal replacement therapy (CRRT), ventilator settings (e.g., mode, positive end-expiratory pressure [PEEP], plateau pressure [Pplat], Fio2, static compliance), ECMO-related information (i.e., duration time, flow, gas flow, fraction of oxygen), and outcomes from the medical records. The research coordinators or clinicians obtained relevant data using a priori designed case report forms.

Oxygenation Therapy, Respiratory Support Strategy, and ECMO Protocol

All enrolled patients had a diagnosis of ARDS. For patients with Pao2/Fio2 ratio was about 200–300, nasal cannula and mask oxygen therapy are recommended. If patients were not resolved, high-flow nasal cannula (HFNC) or noninvasive ventilation (NIV) should be initiated. Within the first 2 hours of the HFNC or NIV, if the patient’s Pao2/Fio2 was still less than 150 mm Hg, or respiratory rate (RR) was still more than 30 times per minute with the tidal volume over 9 mL/kg, invasive MV should be considered. MV should be performed with lung-protective strategy. Prone positioning should be continued for at least 12 hours daily. Lung recruitment is not applied routinely unless indicated and PEEP would be usually set at 5–10 mm Hg.

When the optimal lung-protective strategy and prone position were both proved to be ineffective, patients would be considered to initiate ECMO if any one of these criteria was met: 1) Pao2/Fio2 less than 50 mm Hg over 3 hours; 2) Pao2/Fio2 less than 80 mm Hg over 6 hours; 3) arterial blood gas pH less than 7.25 and Paco2 greater than 60 mm Hg over 6 hours, as well as RR greater than 35 breaths per minute; 4) RR greater than 35 breaths per minute, arterial blood gas pH less than 7.2, and Pplat greater than 30 cm H2O; and 5) complicated with cardiogenic shock or cardiac arrest.

During ECMO, the blood flow and oxygen flow were set according to the pulse oxygen saturation and blood gas test, to maintain the Pao2 60–80 mm Hg and the Paco2 35–45 mm Hg. When the ECMO started, the ventilation strategy would be supper lung-protective strategy. Pressure controlled ventilation would be preferred, while the settings would be no more than pressure controlled 15 cm H2O, PEEP 5–10 cm H2O, respiratory rate 8–10 breaths per minute, and Fio2 less than 40%, even if the tide volume less than 4 mL/kg.

All the patients accepted heparin continuous IV infusion according to activated clotting time (ACT) of the whole blood and activated partial thromboplastin time (APTT) test. The goal ACT was 160–200 seconds while keeping APTT no more than two times of the upper limit of normal. If the patient had high risk of bleeding, the goal ACT turned to be 130–160 seconds, and we would give the transfusion of blood products if necessary.

Statistical Analysis

We summarized the data to frequency (percentages) for categorical variables and medians and interquartile ranges for continuous variables. We used chi-square test for categorical and Wilcoxon-Mann-Whitney U test for continuous variables. Statistical analyses were performed using IBM SPSS 26 Version software (IBM, Armonk, NY). p value of less than 0.05 was regarded statistical significantly.


General Characteristics of the Whole Ventilated Patients

From January 8, 2020, to March 31, 2020, 129 critically ill patients with SARS-CoV-2 pneumonia were admitted to the ICUs of the two ECMO referral hospitals and 59 patients received MV. Table 1 summarizes the general characteristics of patients who received MV. In the whole cohort, 21 (35.6%) individuals received ECMO therapy, seven from Wuhan Pulmonary hospital and 14 from Zhongnan Hospital of Wuhan University. The whole patients had a median age of 65.50 years (56.75–76.00 yr), and median body mass index of 23.15 (21.13–24.27), with 40 (67.8%) of them being male. The lymphocytes count was 0.55 (0.28–0.80) and lactic dehydrogenase was 506.00 U/L (421.00–755.00 U/L).

General Characteristics of Patients With Invasive Mechanical Ventilation

Compared to ventilated patients without ECMO therapy (MV only group), patients with ECMO (MV with ECMO group) had a significant younger age (58.50 yr [42.75–67.25 yr] vs 70.50 yr [61.75–79.25 yr]; p = 0.066). Two groups had the similar Acute Physiology and Chronic Health Evaluation II score (17.00 [11.25–24.75] vs 16.50 [12.00–23.00]; p = 0.413), Sequential Organ Failure Assessment score (6.50 [4.00–8.00] vs 6.50 [5.00–9.00]; p = 0.897), and the duration between symptoms and initiation of MV (14.00 d [11.00–19.00 d] vs 15.00 d [11.00–21.50 d]; p = 0.534). Compared to MV only group, although the MV with ECMO group had an elevated tendency in Murray Lung Injury Score (LIS) (3.59 [3.31–4.00] vs 3.00 [2.67–3.33]; p = 0.124), and had a shorten tendency in the duration between hospitalization and initiation of MV (2.00 d [0–7.50 d] vs 6.00 d [2.50–9.00 d]; p = 0.138), there were no significant difference between two groups. By April 7, 2020, nine of 21 patients survived in MV with ECMO group, and 14 of 38 patients survived in MV only group (57.1% vs 63.2%; p = 0.782).

Laboratory Tests of Patients With ECMO Therapy

From laboratory tests (Table 2), the majority of patients with ECMO therapy had leukocytosis with the WBC count of 12.29 × 109/L (7.94–18.04 × 109/L). Although the lymphocytes counts were relatively low at 0.70 × 109/L (0.51–1.10 × 109/L). Included patients had a normal coagulation status, with a median of partial thromboplastin time (14.85 s [13.73–17.40 s]), APTT (32.05 s [29.73–36.98 s]) but with an elevated d-dimer (359.00 mg/L [105.75.00–2,952.50 mg/L]). Liver injury was also common, characterized by elevated alanine aminotransferase (49.50 U/L [18.50–75.00 U/L]), aspartate aminotransferase (51.00 U/L [31.75–115.500 U/L]), and bilirubin (13.92 μmol/L [9.63–19.83 μmol/L]). The creatinine level was normal, with a median of 70.70 μmol/L (58.65–107.35 μmol/L). The level of hypersensitive troponin I was 30.00 ng/mL (5.03–126.00 ng/mL). However, survivors had a significant lower creatinine than nonsurvivors prior to ECMO (61.10 μmol/L [54.70–95.08 μmol/L] vs 82.95 μmol/L [70.55–157.13 μmol/L]; p = 0.027).

Laboratory Tests of Patients With Extracorporeal Membrane Oxygenation Therapy

Lung Functions Before ECMO

Nineteen patients (90.4%) received neuromuscular blockers, and 12 (57.1%) underwent prone positioning before administration of ECMO. Table 3 presents the lung functions of patients who underwent ECMO prior to initiation. The median of LIS was 3.59 (3.31–4.00) (3.33 [3.25–3.67] in survivors, and 4.00 [3.33–4.00] in nonsurvivors; p = 0.211) before initiation of ECMO. Survivors tended to have a shorter duration of onset of illness before ECMO than nonsurvivors (15.00 d [12.50–20.50 d] vs 18.00 d [14.75–21.75 d]; p = 0.382). Meanwhile, the duration of MV before ECMO seemed to be shorter in survivors than in nonsurvivors, but with no significant difference (20.00 hr [2.00–63.00 hr] vs 43.00 hr [12.25–109.00 hr]; p = 0.277). The Pao2/Fio2 was low (60.0 [55.60–72.00]), and Co2 retention occurred in all patients (Paco2, 56.00 mm Hg [54.00–64.00 mm Hg]). Paco2 was higher in nonsurvivors when compared with survivors (63.20 cm H2O [55.40–72.12 cm H2O] vs 54.40 cm H2O [29.20–57.50 cm H2O]; p = 0.006). Accordingly, nonsurvivors were more acidosis than survivors with pH (nonsurvivors 7.23 [7.16–7.33] vs survivors 7.38 [7.28–7.48]; p = 0.023). The static compliance was similar in survivors (20.0 mL/cm H2O [15.50–28.00 mL/cm H2O]) and nonsurvivors (18.00 mL/cm H2O [16.25–23.00 mL/cm H2O]). Serum lactate was mildly elevated with a median of 1.80 mmol/L (1.50–3.10 mmol/L) (1.60 mmol/L [1.35–2.55 mmol/L] in survivors vs 2.25 mmol/L [1.70–3.60 mmol/L] in nonsurvivors; p = 0.211).

ECMO Initiation and Related Variables

The average running time of ECMO was 218.0 hours (142.5–594.0 hr) (193.0 hr [106.0–576.5 hr] in survivors vs 419.3 hr [202.5–629.0 hr] in nonsurvivors; p = 0.382). All patients started an ECMO flow rate at 1,500 revolutions per minute (rpm) and maintained at 4 L/min. All nonsurvivors had a continued need of high-flow rates. After ECMO initiation, the Paco2 was reduced to normal range in both survivors and nonsurvivors. This was regardless of the Paco2 level prior to ECMO initiation. The sweep flow in the survivors decreased gradually, while it remained high among nonsurvivors. During ECMO support, nonsurvivors remained with no significant higher lactate concentrations than survivors (Fig. 1).

Figure 1.
Figure 1.:
Dynamic changes tendency in extracorporeal membrane oxygenation (ECMO) variables and arterial blood gas variables between survivors and nonsurvivors. ECMO variables including blood flow (L/min) and creep flow (L/min), arterial blood gas variables including lactate levels, and Paco 2 level. Data expressed as median (range).

Outcomes and Complications With ECMO

At the end of April 7, 2020, among the 21 ECMO patients, there were 12 patients died. Nine patients weaned from ECMO successfully, six of which had been discharged. The crude mortality rate from ECMO therapy was 57.1%. Among the patients who died, one patient experienced bradycardia, which led to cardiac arrest. This patient did not have any evidence of bleeding or volume depletion based on ultrasonography examination. Six patients died of persistently worsening lung consolidation that was difficult to reverse and got secondary lung infections with multiple drug-resistant bacteria. Three patients died of septic shock from uncontrolled bloodstream infection by multiple drug-resistant Acinetobacter baumannii. One patient had heart arrest before initiation of ECMO and complicated with brain death. One patient died of cerebral hemorrhage who had a stroke history before.

Bradycardia occurred in five patients (23.8%). Except for one patient, the other four patients responded to medical therapy promptly. Three of the patients presented with enlarged right or left ventricular accompanied by low contractility. Hemorrhagic complications occurred in three patients, which manifested as insertion catheter site bleeding (two patients) and intracerebral hemorrhage (one patient). In the cohort of patients requiring ECMO, nine patients received vasopressors, and eight developed acute kidney injury and required CRRT.


To our knowledge, this is the first study to summarize the characteristics of patients with SARS-CoV-2 pneumonia who developed ARDS and treated with ECMO. In this study, the mortality rate was 57.1%. All included patients had severe ARDS with a median of Pao2/Fio2 ratio of 60.00 (55.60–72.00). Our study provides a more optimistic view as we showed using ECMO as an option for salvage therapy of COVID-19 could be associated with lower mortality.

By April 7, the COVID-19 had already involved 1,263,219 patients and resulted in 71,235 deaths globally. Thus, it is critical to have a protocolized and effective strategy to treat these patients, particularly among those who become critically ill. A recent report with a focus on critically ill patients with confirmed SARS-CoV-2 infection from Jinyintan Hospital (Wuhan) demonstrated considerable mortality (81%) within mechanically ventilated patients (2). However, the death rate among patients who received ECMO was not reported. Similar to SARS-CoV-2, patients with severe ARDS associated with influenza A (H1N1) pneumonia when treated with ECMO have lower mortality of 30–40% (5–8). Indeed, the Conventional ventilation or Extracorporeal membrane oxygenation for Severe Adult Respiratory failure (CESAR) study (10) suggested that ECMO could be regarded as rescue therapy for patients with severe ARDS. Also, in the Extracorporeal membrane Oxygenation for severe acute respiratory distress (EOLIA) trial (11), the patient has received ECMO had lower mortality when compared with conventional study by 11%.

The timing of ECMO initiation is still disputed. It has been suggested to initiate ECMO when Pao2/Fio2 ratio less than 50 mm Hg for greater than 3 hours, or Pao2/Fio2 ratio less than 80 mm Hg for greater than 6 hours, or pH less than 7.25 with Paco2 greater than or equal to 60 mm Hg in arterial blood gas for greater than 6 hours, with the RR greater than 35 breaths/min and MV settings adjusted to keep a Pplat of less than or equal to 32 cm H2O (12). During the H1N1 outbreak, the Pao2/Fio2 ratio of less than 70 mm Hg for at least 2 hours and Pplat greater than 30 cm H2O, or Pao2/Fio2 ratio of less than 100 mm Hg associated with Pplat greater than 35 cm H2O, or respiratory acidosis with pH less than or equal to 7.15 treatment with ECMO was considered (5). However, our observations indicated that earlier initiation of ECMO (evaluated by the length of MV before ECMO initiation) maybe associated with improved outcomes. The majority of patients who developed severe ARDS with SARS-CoV-2 pneumonia had delayed treatment and deteriorated rapidly. We considered ECMO should be implemented as soon as possible when the Pao2/Fio2 ratio was less than 80 mm Hg despite being on lung-protective MV strategy and prone positioning.

In our study, ventilated patients had a bad compliance with a median 18.00 mL/cm H2O (16.50–24.00 mL/cm H2O) prior to ECMO. Compliance less than 20 mL/cm H2O was reported to be a prognostic factor of death on venovenous ECMO (12, 13). We noted that patients with more severe Co2 retention tended to have a worse prognosis. In a previous study, higher Pco2 was associated with worse prognosis (14). Also in the study by Nuckton et al (15), the elevated Co2 likely reflects ARDS severity and increased dead space fraction. Thus, we speculated that patients with higher Pco2 for a period might need ECMO support. Besides, acidosis and elevated creatinine prior to ECMO was also associated with poor outcomes.

When the ECMO pump started, many patients developed bradycardia with or without hypotension. We speculated that multiple factors led to such observation. Most patients were severely hypoxemic before ECMO initiation. Following ECMO initiation, receiving highly oxygenated blood could have resulted in ischemia-reperfusion injury and excessive inflammatory response. Patients with hemodynamic instability before ECMO initiation had a higher risk of bradycardia. So we suggested that the ECMO pump rotation should increase slowly from 1,500 rpm to the target rotation with a rate of 500 rpm elevation in every 10 minutes.

Bleeding and thromboembolic events are reported as the most frequent causes of death (16). In our observation, bleeding occurred in the early phase of ECMO therapy in three patients. Bleeding resolved by compression, suturing, and required transfusion. More contemporary ECMO circuitry with higher biocompatibility (17) have markedly contributed to an overall reduction in the need for anticoagulation. Our low prevalence of bleeding may be explained by careful management of the anticoagulation regimen, along with the recruitment of highly experienced teams.

Our study has several limitations inherent to all retrospective studies. First, the power of the study is limited by the small size of our cohort. Therefore, type 2 errors could have led to missing statistical significant differences. Second, we did not adjust for the multiple confounding factors simultaneously; as such, identified risk factors like time to ECMO needed, are likely accounting for ARDS severity and Co2 retention. Third, several data related to the ECMO and mechanical ventilator settings were not available. Last, our observation period was short and further outcomes about these patients may be required for complete analysis.

In conclusion, ECMO might be an effective salvage treatment for patients with SARS-CoV-2 pneumonia associated with severe ARDS. Severe Co2 retention and acidosis prior to ECMO indicated a poor prognosis.

Variables of Lung Function Prior to Extracorporeal Membrane Oxygenation


We would like to thank the staff of the Department of Critical Care Medicine of Wuhan Pulmonary Hospital, who contributed to this study by collecting the required data in the hospital data system.


1. Wang DW, Hu C, Hu B, et al. Cli nical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China. JAMA. 2020; 323:1061–1069
2. Yang X, Yu Y, Xu J, et al. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: A single-centered, retrospective, observational study. Lancet Respir Med. 2020; 8:475–481
3. Ranieri VM, Rubenfeld GD, Thompson BT; ARDS Definition Task ForceAcute respiratory distress syndrome: The Berlin definition. JAMA. 2012; 307:2526–2533
4. Bellani G, Laffey JG, Pham T, et al.; LUNG SAFE Investigators; ESICM Trials GroupEpidemiology, patterns of care, and mortality for patients with acute respiratory distress syndrome in intensive care units in 50 countries. JAMA. 2016; 315:788–800
5. Roch A, Lepaul-Ercole R, Grisoli D, et al. Extracorporeal membrane oxygenation for severe influenza A (H1N1) acute respiratory distress syndrome: A prospective observational comparative study. Intensive Care Med. 2010; 36:1899–1905
6. Noah MA, Peek GJ, Finney SJ, et al. Referral to an extracorporeal membrane oxygenation center and mortality among patients with severe 2009 influenza A(H1N1). JAMA. 2011; 306:1659–1668
7. Patroniti N, Zangrillo A, Pappalardo F, et al. The Italian ECMO network experience during the 2009 influenza A(H1N1) pandemic: Preparation for severe respiratory emergency outbreaks. Intensive Care Med. 2011; 37:1447–1457
8. Pham T, Combes A, Rozé H, et al.; REVA Research NetworkExtracorporeal membrane oxygenation for pandemic influenza A(H1N1)-induced acute respiratory distress syndrome: A cohort study and propensity-matched analysis. Am J Respir Crit Care Med. 2013; 187:276–285
9. World Health OrganizationClinical Management of Severe Acute Respiratory Infection When Novel Coronavirus (2019-nCoV) Infection Is Suspected: Interim Guidance.2020. Available at: Accessed February 8, 2020
10. Peek GJ, Mugford M, Tiruvoipati R, et al.; CESAR trial collaborationEfficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): A multicentre randomised controlled trial. Lancet. 2009; 374:1351–1363
11. Combes A, Hajage D, Capellier G, et al.; EOLIA Trial Group, REVA, and ECMONetExtracorporeal membrane oxygenation for severe acute respiratory distress syndrome. N Engl J Med. 2018; 378:1965–1975
12. Wu MY, Huang CC, Wu TI, et al. Venovenous extracorporeal membrane oxygenation for acute respiratory distress syndrome in adults: Prognostic factors for outcomes. Medicine (Baltimore). 2016; 95:e2870
13. Wu MY, Chang YS, Huang CC, et al. The impacts of baseline ventilator parameters on hospital mortality in acute respiratory distress syndrome treated with venovenous extracorporeal membrane oxygenation: A retrospective cohort study. BMC Pulm Med. 2017; 17:181
14. Buchner J, Mazzeffi M, Kon Z, et al. Single-center experience with veno-venous ECMO for influenza-related ARDS. J Cardiothorac Vasc Anesth. 2018; 32:1154–1159
15. Nuckton TJ, Alonso JA, Kallet RH, et al. Pulmonary dead-space fraction as a risk factor for death in the acute respiratory distress syndrome. N Engl J Med. 2002; 346:1281–1286
16. Beiras-Fernandez A, Deutsch MA, Kainzinger S, et al. Extracorporeal membrane oxygenation in 108 patients with low cardiac output – a single-center experience. Int J Artif Organs. 2011; 34:365–373
17. MacLaren G, Combes A, Bartlett RH. Contemporary extracorporeal membrane oxygenation for adult respiratory failure: Life support in the new era. Intensive Care Med. 2012; 38:210–220

acute respiratory distress syndrome; coronavirus disease 2019; extracorporeal membrane oxygenation

Copyright © 2020 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.