Veno-venous extracorporeal membrane oxygenation (VV-ECMO) is an established supportive therapy for severe and potentially reversible acute respiratory distress syndrome (ARDS) [1, 2]. Despite improving patient outcomes, including reducing mortality, VV-ECMO is associated with increased need for packed red blood cell (PRBC) transfusion. This includes following the removal of VV-ECMO cannulae and circuit (VV-ECMO decannulation), where large bore cannulae are removed from major veins and discarded along with the membrane lung and the circuit tubing containing the patient’s blood [3, 4]. The administration of PRBC transfusion has a range of potential adverse effects and represents a scarce and expensive resource. Therefore, employing strategies that can reduce transfusion rates is a clinical priority.

One strategy to reduce PRBC transfusion for VV-ECMO patients involves salvaging blood loss at the time of decannulation. There is approximately 500–700 ml of blood within the VV-ECMO circuit which is commonly discarded. Some but not all VV-ECMO centres utilise blood cell salvage (BCS) systems during decannulation to conserve blood and minimise the risk of transfusion-related complications. The BCS systems collect blood from the VV-ECMO circuit and aspirates it into a reservoir. The red cells are isolated from other constituents of the salvaged blood by high-speed centrifugation. The separated erythrocytes are subsequently washed in normal saline fluid, resulting in a concentrated suspension of the patient’s own red cells (approximately 60% haematocrit) ready for reinfusion [5].

It has been hypothesised that BCS could increase the haemoglobin (Hb) level at the time of decannulation, thereby reducing the need for PRBC transfusion. However, the evidence supporting BCS following ECMO decannulation is limited. Furthermore, conducting a prospective clinical trial is expensive and would take several years until data were available given the relatively low rate of patient enrolment. Therefore, we used routinely collected clinical data combined with causal inference methodologies [6], to emulate a target trial [7, 8]. We compared BCS during VV-ECMO decannulation versus decannulation without blood cell salvage (n-BCS) in adults with severe respiratory failure. Our primary objective was to estimate the effect of BCS on subsequent PRBC transfusions. Additionally, we examined changes in Hb levels, other routinely performed laboratory measurements, and other transfusion requirements.