Life support by extracorporeal membrane oxygenation plays a major role in treating severe cases of acute respiratory distress syndrome (ARDS) associated with COVID-19 (1–4). It is generally accepted that anticoagulation of the extracorporeal circuit is mandatory to prevent systemic clotting and thromboembolic complications (5–7). At the same time, this requirement increases the risk of bleeding complications, which sometimes may be fatal (8, 9). To strike a correct balance between these potential complications, adequate dosing and efficient monitoring of the anticoagulant drug needs to be ensured (5, 6, 10).

There is ongoing discussion as to which anticoagulant medication has the best safety profile (11). Current guidelines recommend the use of unfractionated heparin for anticoagulation during extracorporeal membrane oxygenation (ECMO) (12). Standardized protocols using low-molecular-weight heparin have yet to find their way into routine clinical practice, even though promising data have been presented for enoxaparin and nadroparin (13–15). There are also reports on major advantages of low-molecular-weight over unfractionated heparin like reduced bleeding complications or heparin-induced thrombocytopenia (16, 17).

A number of intensive care units (ICUs) in our tertiary care center have come to use enoxaparin routinely for anticoagulation during ECMO therapy. The effectiveness of this approach was recently demonstrated in lung transplant patients (13) and we hypothesized, that this might also hold true for patients with COVID-19 associated ARDS. Throughout the COVID-19 pandemic, some of our ICUs continued to administer enoxaparin whereas others have used unfractionated heparin. This situation has enabled us to compare major (i) thromboembolic and (ii) bleeding complications in patients with severe COVID-19 who had received either enoxaparin subcutaneously or unfractionated heparin intravenously for anticoagulation during extracorporeal membrane oxygenation.

Ninety-eight patients could be evaluated, 62 of whom had received enoxaparin and 36 unfractionated heparin. A flow chart of the study is presented in Figure 1. As shown in Table 1, the baseline patient demographics were comparable in both groups, with the exception of sepsis-related organ failure assessments (p = 0.04). Baseline laboratory parameters (conventional coagulation parameters, anti-factor Xa levels, platelet counts, hemoglobin levels) did not differ significantly, except for thrombin time (p = 0.02) that was documented for only 46 patients at values within reference ranges. Additional File 2 in the Supplementary Material presents further details with regard to general treatment, transfusion triggers, transfusion requirements and administration of coagulation factor concentrates.

Except for one patient in the LMWH group, all patients underwent prone positioning during the observational period. Median paO2/FiO2 ratio prior to the start of ECMO therapy was 72.2 mmHg (IQR: 60.5–96.6) in the LMWH group and 74.2 mmHg (IQR: 66.3–155.7) in the UFH group (p = 0.04). Extracorporeal membrane oxygenation was provided over 1,741 patient days (until the occurrence of an event). Venovenous cannulation accounted for the majority of patients in both groups, and venoarterial or venovenoarterial, as used in situations of combined cardiac and respiratory failure, for the remainder (Table 2). The group-specific median durations of these therapies were comparable (p = 0.09). Endpoint events are listed in Table 2. A total of 35 events would per se have terminated the observation span, but given two coinciding events in two cases, only 33 patients were affected: one diagnosed with intracerebral bleeding and major bleeding requiring > two units of packed red blood cells within 24 h, and one with heparin-induced thrombocytopenia in addition to requiring an oxygenator exchange, each on the same day in the ICU.

All endpoint events, with the exception of pulmonary embolism, were found to have occurred less frequently in the enoxaparin group (Table 2 and Figure 2). No clinically relevant cases of deep vein thrombosis were noted. Figure 2 illustrates the timelines of event-free days per endpoint per group. Adjustment in calculating the hazard ratios was required by the baseline findings, listed in Table 1, of a significant difference in sepsis-related organ failure assessments (p = 0.04) and a certain difference in gender distribution (p = 0.07).

All adjusted hazard ratios favored enoxaparin over unfractionated heparin: 3.43 for the thromboembolic primary endpoint (95% CI: 1.08–10.87; p = 0.04); 2.58 for the hemorrhagic secondary endpoint (95% CI: 1.03–6.48; p = 0.04); and 2.86 for the composite endpoint of both event types (95% CI: 1.41–5.92; p = 0.007). As shown in Table 2 and Figure 3, efficient dosing of enoxaparin was reflected by peak levels of anti-factor Xa (median: 0.45 IU ml⁻¹;IQR: 0.38; 0.56). Activated partial thromboplastin time and thrombin time differed significantly (p < 0.001) between the groups mirroring the effect of unfractionated heparin.

This is the first study to report on the successful use of the low-molecular-weight heparin enoxaparin for anticoagulation in COVID-19 patients with acute respiratory distress syndrome during extracorporeal membrane oxygenation. Compared to unfractionated heparin, we found that enoxaparin was associated with superior results in terms of both clinically relevant thromboembolic events and major bleeding complications.

Current guidelines recommend continuous intravenous administration of unfractionated heparin for systemic anticoagulation to prevent intra- and extracorporeal clotting during extracorporeal membrane oxygenation (12). Even with this recommendation implemented, 19 to 50% rates of thromboembolic events have recently been reported in scenarios of respiratory distress syndrome or lung transplantation, compounded by 19 to 66% rates of transfusion due to hemorrhagic side effects of high-dose anticoagulation (10, 13, 30, 31). The present study follows suit in reporting a 25% rate of thromboembolic and a 34% rate of bleeding complications despite the use of unfractionated heparin as recommended.

Given these high complication rates associated with unfractionated heparin, there has been increasing interest lately in low-molecular-weight heparin, with reported rates of 6.5 or 20% for thromboembolic and 12.5 or 30% of bleeding complications (13, 14). Consistent with these figures, the use of enoxaparin in the present study involved 8% thromboembolic and 12.9% bleeding complications. Krueger et al. (14) were the first to report on enoxaparin, administered subcutaneously in standard prophylactic doses, for venovenous extracorporeal oxygenation in acute respiratory distress syndrome, with thrombosis of the centrifugal pump occurring in 5% of patients during the first week. This rate is very similar to the 4.8% rate of oxygenator exchanges in our enoxaparin group, compared to 22.3% in the unfractionated-heparin group (see Table 2).

Only two direct comparisons are currently available on the safety and efficacy of low-molecular-weight vs. unfractionated heparin during extracorporeal oxygenation. While Piwowarczyk et al. (15) noted similar rates of thromboembolic events and bleeding complications for nadroparin and unfractionated heparin, these findings might not be directly relatable to ours, since the comparability of enoxaparin and nadroparin has been questioned (32). Our own study group observed, in perioperative settings of lung transplantation, bleeding events with enoxaparin in 12.5% and with unfractionated heparin in 22.7% of patients (p = 0.31) (13). Our finding herein reported of similar rates (11.3 or 22.2%, respectively; see Table 2) in spite of median daily cumulative dosages of enoxaparin nearly twice as high (4,800 vs 8,000 IU) might point to an inherently increased risk of bleeding in our previous cohort of perioperative transplant patients.

Increasing the risk of heparin-induced thrombocytopenia is a major drawback of unfractionated compared to low-molecular-weight heparin. In the present study, one such diagnosis was made in the unfractionated-heparin group, which, given a 0.5% overall incidence during critical illness (33, 34), corresponds to a notable 2.8% rate in 36 patients. An increased prevalence of heparin-induced thrombocytopenia during extracorporeal membrane oxygenation has been reported previously (35, 36). Also, Daviet et al. recently reported an almost 10-fold increase in positivity for heparin-induced thrombocytopenia, up to a prevalence of 8%, with unfractionated heparin in COVID-19 patients, due to reasons possibly consisting in higher drug concentrations needed for therapeutic anticoagulation and COVID-19-related exacerbation of immune reactions (37).

The Extracorporeal Life Support Organization recommends a multimodal approach to anticoagulation monitoring by adjusting unfractionated heparin based on appropriate standard curves for activated partial thromboplastin time, activated clotting time, thrombin time, or anti-factor Xa levels (12). The anticoagulant regimens herein reported had been guided accordingly, with a 0.45 IU ml⁻¹ median peak level of anti-factor Xa (IQR: 0.38; 0.56) proving the activity of enoxaparin, and the one of unfractionated heparin mirrored by significant differences from enoxaparin for the activated partial thromboplastin time (p < 0.001).

A few limitations of our study should be noted. First of all, retrospective findings will always carry some risk of bias, although we processed all existing data, missing data, and potentially incomplete records with meticulous care. As a case in point, the threshold for oxygenator exchanges was progressively lowered throughout the pandemic while our automated system of patient documentation did not disclose in each specific instance whether clotting of the oxygenator had actually occurred or been imminent. Second, most patients had been referred from hospitals using various anticoagulant regimens. Hence, to avoid potential confounders, we only included patients with complete documentation of anticoagulant medication after in-house cannulation. Third, no routine screening for thromboembolic events had taken place, although this should not formally be regarded as a limitation given that current guidelines recommend against such screening for venous thromboembolism in critically ill patients (26). Still, this lack of screening might account for the small total number of these complications in our study despite a median duration of extracorporeal membrane oxygenation of more than 2 weeks. Lastly, it is important to note that complication rates in terms of thromboembolism and bleeding might vary depending on the ECMO cannulation mode.

In summary, subcutaneously administered enoxaparin is a feasible anticoagulation strategy for extracorporeal membrane oxygenation in COVID-19 patients. Both the thromboembolic primary and the hemorrhagic secondary endpoint of our study yielded results of this approach superior to unfractionated heparin. Jumping to definite conclusions based on this retrospective analysis would be ill-advised, but the current data at the very least support the effective use of enoxaparin in this vulnerable patient cohort and highlight the urgent need for prospective, randomized trials.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Medical University of Vienna; ID: 2024/2020. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

MW, ES, and JG conceived and designed the study, interpreted data, and drafted the manuscript. DL and MM acquired the data. AB performed statistical analysis and interpreted data. PK interpreted data. Critical revisions of the manuscript, in addition to all of the aforementioned authors, were contributed by AH, MH, and CZ. All authors commented on successive versions and have read and approved the final manuscript.

This work was supported by departmental funds of the Department of Anesthesia, Critical Care and Pain Medicine Division of General Anesthesia and Intensive Care Medicine, Medical University of Vienna and the Ludwig Boltzmann Institute Digital Health and Patient Safety, Medical University of Vienna, Vienna, Austria.

MW received honoraria, research funding and travel reimbursement from Boehringer Ingelheim, CSL Behring and Mitsubishi Tanabe Pharma. JG received honoraria, research funding, and travel reimbursement from Alexion, Boehringer Ingelheim, CSL Behring, Johnson & Johnson, Instrumentation Laboratory, Mitsubishi Tanabe Pharma, Octapharma, and Portola. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.