In our study, we were able to confirm a high ICU mortality rate among patients requiring ECMO for COVID-19 in Poland. In-hospital death was confirmed to be independently associated with increased pre-ECMO lactate levels and increased BMI.
Our observation period covered the first 15 months of the pandemic in Poland, i.e., the entire first wave, second wave, and a substantial part of the third wave. The severity of the pandemic waves was different in comparison with that observed in other countries. The first wave was relatively benign, due to the prompt adoption of social distancing and a complete lockdown throughout the country . However, Poland was seriously affected by the second and particularly by the third wave of the pandemic [16, 17]
Nearly, a third of the subjects in our study were transported while on ECMO. According to the ELSO Registry—the largest cohort of COVID-19 patients requiring ECMO to date—nearly half of all patients (47%) were transported while on ECMO support . An even greater proportion of COVID-19 patients—approximately 55%—were transported on ECMO support in the Greater Paris region . The difference in utilization of ECMO transport between the international cohort and our study is most likely an indicator of the early stage of development of the ECMO system in Poland.
Current guidelines recommend the application of ECMO in acute respiratory distress syndrome (ARDS) caused by SARS-CoV-2 infection predominantly in patients without significant comorbidities . Interestingly, only 57% of our patients were without comorbidities. Moreover, approximately 20% of patients in our population had a BMI > 35 kg/m2, however—according to the current guidelines—obesity is not considered a contraindication for ECMO in COVID-19 patients . Therefore, it may be hypothesized that VV ECMO—as a relatively infrequent procedure in Poland prior to the current pandemic—was considered a last resort procedure applied in some cases on compassionate rather than medical grounds.
Available data implicates a direct relationship between mortality and the level of hypoxemia at the initiation of ECMO in the COVID-19 population . The very severe pre-ECMO hypoxemia found in our study (with a mean paO2/FiO2 ratio of only 70.5 mmHg) may explain the high mortality rate. Late referral for ECMO and long periods of respiratory support with positive pressure and high oxygen concentration may expose patients to increased risks of ventilation-induced lung injury and patient self-inflicted lung injury.
Unsurprisingly, lactate levels prior to the initiation of ECMO were one of the two independent risk factors for death in our multivariate analysis. This is not surprising, as levels of lactate are related not merely to its production but also its clearance. Therefore, hypoxemia is not the only reason for elevated lactate levels, which may also be a result of impaired oxygen utilization on the cellular level, while acidemia may be observed only in subjects with concomitant renal failure .
Veno-venous (VV) ECMO was initially used in a vast majority of cases, in line with the medical literature [2, 3, 6]. However, in selected cases the conversion of VV ECMO to more complex ECMO modes may be needed due to various indications and such a conversion may be required in up to 18% of patients .
Approximately, half of our patients received a tracheostomy while on ECMO. There is no consensus on the timing of a tracheostomy in patients on VV ECMO for COVID-19. Hitherto, there is no convincing comparative data concerning this issue, and most of the reports on this subject are case series. The frequency of tracheotomy in this group of patients is also ambiguous and falls within the wide range of 30–61% [8, 23, 24].
There are some negative aspects to performing an early tracheostomy during VV-ECMO use in patients. Bleeding complications are more common, and the hypercoagulable state following discontinuation of anticoagulants raises concerns about the possibility of ECMO circuit malfunction. Due to an increased risk of contaminating personnel, various national organizations recommend caution in this regard . In addition, the technical skills of personnel are decreased by the use of gowns, multiple gloves, and face masks .
In our study, only seven patients were extubated on ECMO support. Kunavarapu et al.  investigated whether this is a viable treatment option. The authors found that the survival rate was higher in patients placed on ECMO prior to mechanical ventilation, but their sample size for such calculations was relatively small .
Complications of ECMO were relatively frequent in the analyzed population. The most common complication was bacterial infections and bleeding (in approximately 80% and 55% of our patients, respectively), occurring with a similar frequency in survivors and non-survivors. Marked immunosuppression (related to the systematic use of corticosteroids), might have contributed to bacterial superinfections, with a predominance of ventilatory-associated pneumonia (VAP). The prevalence of VAP in our study was similar to findings of a recent European study  but higher when compared to the data from Latin America , and much higher in comparison with the EOLIA trial performed during the pre-COVID era .
Another common complication of ECMO in our COVID-19 patients was bleeding. Hemorrhagic complications were reported in 55% of our patients. Similar results were obtained by French investigators, where massive hemorrhages were reported in 42% of their patients .
Another relatively common complication was acute renal failure requiring renal replacement therapy. This complication occurred in 37.3% of patients, and was significantly more frequent in non-survivors (43.6% vs 19.5%, p = 0.011). A similar observation was reported in the majority of publications describing complications of ECMO in COVID-19 patients. In two European multicenter cohorts of patients with COVID-19, it was found that 22% and 46% of patients on ECMO required renal replacement therapy [24, 29], while in the largest ELSO registry, renal complications were reported in 42.9% of patients . A systematic review and meta-analysis of twenty-two observational studies including 1,896 patients done by Ramanathan et al.  reported that 35% of patients developed renal complications while on ECMO support .
The in-hospital mortality in our population was 74.1%. The first experience with ECMO in COVID-19 patients from China revealed very high death rate [1, 30]. Following reports showed significantly lower mortality rates in comparison with our results. In 302 patients from the Greater Paris area, 90-day mortality rate was 54% . Analysis of data from 1,035 US patients from ELSO Registry indicated the mortality of 38% , while the European data from the EuroELSO Survey revealed 44% mortality in 1,602 patients . A recent systematic review including 22 studies and 1,896 COVID-19 patients treated with ECMO, found the in-hospital mortality of 37.1% .
This striking difference in comparison with our results is difficult to explain. However, a recently published analysis of data from German hospitals (based on data from the German insurance institutions) indicated a mortality rate as high as 71% among 119 patients treated with ECMO during the first wave of COVID-19 pandemic . These results are in line with our findings.
Furthermore, our patients had very late qualification for ECMO treatment based on the extremely low values of the paO2/FiO2 ratio and the long duration of mechanical ventilation before ECMO. Finally, despite available guidelines for ECMO therapy in Poland, a rather low compliance to these guidelines was observed in our study. For example, 14 patients (nearly 10%) were above 60 years old, and 31 patients (approx. 20%) were morbidly obese (BMI > 35 kg/m2).
In our study, we were able to identify two independent risk factors for mortality in ECMO patients—pre-ECMO lactate levels and BMI. The identification of these factors contributing to mortality comes as no surprise.
Prolonged tissue hypoxia seen in patients with COVID-19 plays an important role in the development of subsequent organ dysfunction and mortality. A number of studies have reported the correlation between tissue hypoxia, lactate levels, and mortality among critically ill patients with COVID-19 [31, 32]. We should therefore aim for the early implementation of ECMO before a patient’s lactate level becomes elevated. In some ethically difficult situations, pre-ECMO lactate levels may also help in the decision-making process.
Obesity is a widely discussed risk factor for ICU mortality among COVID-19 patients [33,34,35]. However, the morbidity and mortality of obese patients managed in ECMO centers remain ambiguous. A recent systematic review including 6 studies and a total of 1,285 patients did not show significant differences between obese and non-obese patients . Ramanathan et al.  reported a negative correlation between obesity and mortality in COVID-19 patients treated with ECMO . We should be careful in drawing conclusions from our database. Patients deteriorating on mechanical ventilation (assuming no contraindications for ECMO were identified) should be referred to ECMO centers in a timely manner. According to the most recent ELSO Guidelines, obesity per se is not a contraindication to ECMO , although patient selection performed by an experienced clinician must be judicious. In a meta-analysis performed by Zaidi et al., the cutoff value for excessive BMI was set at 30 kg/m2, whereas in our study, it was 35 kg/m2 . The management of obese patients, especially those with BMIs above 35 kg/m2 may be challenging. More data is needed to assess the appropriate utilization of ECMO in patients with high BMIs as this group of patients is under-represented in the literature.
Our study has some significant limitations. This is retrospective research, which is always prone to bias. There were a few high-volume Polish ECMO centers that did not take part in this project. We also have to assume that a significant (but unknown) number of smaller centers also carried out a few ECMO procedures. Therefore, we do not have the entire picture of ECMO utilization for the whole country during the pandemic, as it was possible in a German study that has just been published . Our outcome measures were also limited only to ICU mortality, as we did not have access to the follow-up data of patients successfully discharged from the ICU. We also encountered enormous difficulties in obtaining accurate (or any) data on the circumstances of ECMO implementation—this was particularly true for patients who required ECMO implantation in difficult conditions and remote locations. However, all these deficiencies are balanced by the relatively large sample size which represents a significant part of all ECMO cases performed in Poland during the COVID-19 pandemic.