Abstract
BACKGROUND: Veno-venous extracorporeal membrane oxygenation (VV-ECMO) is a supportive therapy for acute respiratory failure with increased risk of packed red blood cells (PRBC) transfusion. Blood cell salvage (BCS) aims to reduce blood transfusion, but its efficacy is unclear. This study aimed to estimate the effect of BCS at the time of removal of the ECMO circuit (ECMO decannulation) on PRBC transfused. METHODS: To compare BCS to non-blood cell salvage (n-BCS), we conducted an emulated trial of patients at two ECMO centres in the United Kingdom. We used inverse propensity of treatment weighting to control for confounding and estimated the average treatment effect of BCS on PRBC transfused within two days of decannulation, and on changes in haemoglobin (Hb). RESULTS: We included 841 patients who underwent VV-ECMO decannulation. The estimated marginal mean number of PRBC transfused when using BCS was 0·2 (95%CI: 0·16, 0·25) units compared to 0·51 (95%CI: 0·44, 0·59) units with n-BCS; an average treatment effect of -0·31 (95%CI: -0·40, -0·22) units. BCS reduced the risk of receiving any PRBC transfusion by 17·1% (95%CI: 11·1%, 22·9%) equating to a number needed to treat for any PRBC transfusion of 6 (95%CI: 5, 9). The difference in expected Hb levels after decannulation between BCS and n-BCS was 5·0 (95%CI: 4·2, 5·8) g/L. CONCLUSIONS: The use of BCS during VV-ECMO decannulation may be an effective strategy to augment haemoglobin levels and reduce PRBC transfusions.