Background

Despite an impressive gain of knowledge throughout the COVID-19 pandemic, the management of COVID-19 patients with acute respiratory distress syndrome (ARDS) remains challenging [1]. For these patients, venovenous extracorporeal membrane oxygenation (VV ECMO) can be a potential life-saving intervention facilitating lung protective ventilation [2, 3]. However, the use of ECMO is associated with serious complications such as an increased bleeding risk and hemotrauma [4]. In addition, patients with critical COVID-19 exhibit unique abnormalities in coagulation which can increase the risk of both bleeding and thrombotic events [31].

Aside from anticoagulation targets, management strategies of critically ill COVID-19 patients have been adapted during the course of the pandemic. In order to counteract profound inflammation, dexamethasone and tocilizumab have been increasingly used, with varying effects on patient outcomes [32,33,34]. From a pathophysiological viewpoint, it would be plausible that attenuated inflammation could decrease the risk of ICH, independent of anti-Xa levels. Indeed, we found decreased CRP and ferritin levels in the lower anti-Xa group, probably reflecting a more frequent use of anti-inflammatory drugs later in the pandemic. Interleukin 6 (IL-6) levels were slightly higher in the low anti-Xa group, in line with observations that IL-6 serum levels even can increase after administration of tocilizumab [35]. Taking dexamethasone and tocilizumab use into account in our analysis, the occurrence of ICH was still the main risk factor for mortality, whereas use of dexamethasone and tocilizumab was not associated with better patient survival. However, these results should be interpreted with caution, as we cannot rule out residual confounders affecting treatment decisions and inflammation. Of note, ECMO runtime was slightly longer and organ support tended to decrease in the low anti-Xa group, both probably affecting ICH risk. Thus, further large-scale studies specifically including patients requiring ECMO support should clarify effects of different anticoagulation strategies and the use of anti-inflammatory medication on global patient outcomes such as ICU length of stay, the need for mechanical ventilation, long-term neurologic sequelae of patients affected by bleeding, and mortality [31].

Our study has to account for limitations. We had a relatively small population of COVID-19 patients of VV ECMO, but overall mortality was similar to previous VV ECMO studies [10, 12, 36] and the consistency of the results across multiple centers adds generalizability to our findings. The retrospective nature of this study prevented us from inferring causality, and we cannot exclude the possibility of unmeasured confounders. Furthermore, our study did not analyze different viral mutants nor did it account for specific ICU patient management strategies such as ventilation, proning or sedation strategies. These all might have contributed to the lower observed overall mortality in the low anti-Xa cohort. The lower anti-Xa target was deployed later in the COVID-19 pandemic in response to the higher rates of ICH observed in earlier phases of the pandemic. Eligibility criteria for VV ECMO and practices for ICU and ventilator management during VV ECMO support evolved to some degree at the participating centers over the course of the pandemic, potentially contributing to higher survival rates in the lower anti-Xa group.

Conclusions

A less intense anticoagulation target (anti-Xa activity 0.15–0.3 U/mL) was associated with a decreased incidence of ICH and lower mortality in COVID-19 patients on VV ECMO support. Our results highlight the need for prospective studies to evaluate anticoagulation regimens in ARDS patients on ECMO support.