Major haemorrhage and blood product utilisation in patients receiving VA ECMO for cardiogenic shock: a multicentre observational study (OBLEX)

Haemorrhage and blood product usage are common in venoarterial extracorporeal membrane oxygenation (VA ECMO) and associated with increased mortality. A prospective, investigator-initiated, longitudinal observational cohort study on major haemorrhagic events in 12 ECMO centres from 3 continents for three predefined subgroups (VA ECMO initiated during cardio-pulmonary resuscitation (ECPR), after cardiothoracic surgery (CTS), for cardiogenic shock (CS)). The aim was to describe haemorrhagic complications as well as transfusion practice and anticoagulation for the whole population as well as the subgroups. In addition, independent baseline predictors for red blood cell (RBC) transfusions were evaluated. 545 prospective patients were included between 2019 and 2022 (ECPR 149, CTS 169, CS 227). Hospital mortality was 46%. Over 2796 days 406 major haemorrhage events in 286 (52%) patients were recorded. CTS and ECPR patients had more frequent events occurring earlier in their course. 88% received RBC transfusions (1.27 (95%CI 1.22 – 1.31) units/day) with significantly more transfusions for CTS and ECPR patients. Platelet transfusion rates were highest in the CTS group (0.58 (95%CI 0.53–0.64) units/day). Haemoglobin and platelet count prior to transfusion were independent of subgroups and averaged (78 g/L (IQR 73, 84), 58 × 10^9/L (IQR 37, 85), respectively). However, platelet count prior transfusion was only marginally higher on days with major haemorrhage (74 × 10^9/L (IQR 50, 104). Systemic anticoagulation was started within the first 24 h in 83% (95% CI 80–87%) of patients, most frequently in CS patients (90%, CI 85–95%). Independent baseline predictors for RBC transfusion were ECPR (IRR 1.50, 95%CI 1.19–1.89) and prior use of antiplatelets (IRR: 1.43, 95%CI 1.13–1.80). Myocarditis and pulmonary embolism were associated with a lower rate of transfusion when compared to myocardial infarct (IRR: 0.57 (0.37–0.89), IRR: 0.67 (0.45–1.00, respectively). Haemorrhagic complications differ in clinical subgroups and RBC transfusion exposure is by far higher than in other critically unwell populations. We identified ECPR and antiplatelet therapy as additional predictors. Transfusion practice for RBC and platelets is variable and does not always follow international guidelines.