In patients receiving extracorporeal membrane oxygenation (ECMO), standard practice is full-dose intravenous unfractionated heparin (UFH) targeting an activated partial thromboplastin time of 2·0–2·5 times baseline to reduce thrombotic risk. This approach can increase bleeding without further reducing thrombosis compared with low-dose UFH. Robust evidence to guide anticoagulation targets is absent because anticoagulation targets in ECMO have never been assessed in a sufficiently powered randomised trial. We aimed to determine whether low-dose UFH or therapeutic low-molecular-weight heparin (LMWH) is non-inferior to standard-dose UFH in patients receiving ECMO.

In this open-label, three-arm, randomised, non-inferiority trial at seven Dutch intensive care units (ICUs), adults aged 18 years or older supported with veno-venous or veno-arterial ECMO at the ICU without vital indication for full-dose anticoagulation (eg, mechanical mitral valve) were randomly assigned to intravenous standard-dose UFH (activated partial thromboplastin time 2·0–2·5 times baseline), intravenous low-dose UFH (1·5–2·0 times baseline), or therapeutic subcutaneous LMWH. The primary outcome was a composite of severe bleeding during ECMO support, severe thromboembolic complications during ECMO support, or all-cause mortality at 6 months, and was assessed in all randomly assigned patients with deferred informed consent and 6-month follow-up data according to the intention-to-treat principle. Non-inferiority was met if the upper 95% CI limit for the absolute risk difference was less than 7·5 percentage points. This trial is registered at ClinicalTrials.gov (NCT04536272) and the Dutch trial register (NL7976).

Between Oct 22, 2020, and Sept 12, 2024, 330 patients were enrolled: 110 were randomly assigned to standard-dose UFH, 110 to low-dose UFH, and 110 to LMWH. Of 330 enrolled patients, 320 (225 [70%] males, 95 [30%] females; median age 56 years [IQR 45–65]; 255 [80%] White ethnicity) were analysed at 6 months. The composite primary outcome occurred in 87 (81%) of 107 patients with standard-dose UFH, 78 (72%) of 108 with low-dose UFH (absolute risk difference −9·1 percentage points [95% CI −20·3 to 2·1]), and 79 (75%) of 105 with LMWH (−6·1 percentage points [−17·2 to 5·0]), meeting non-inferiority for both interventions. The frequency of severe bleeding was lower with low-dose UFH and LMWH than with standard-dose UFH (63 [58%] patients and 62 [59%] vs 70 [65%]), without excess severe thromboembolic complications (11 [10%] and nine [9%] vs 12 [11%]), although these differences did not reach statistical significance. At 6 months, 54 (50%) patients in the standard-dose UFH group, 45 (42%) in the low-dose UFH group, and 46 (44%) in the LMWH group had died.

Low-dose UFH and therapeutic LMWH were non-inferior to standard-dose UFH. This finding supports reconsideration of anticoagulation targets in ECMO. Given the global expansion of ECMO, these findings suggest that reduced anticoagulation targets could substantially reduce bleeding-related harm.