Liberal or restrictive transfusion for veno-arterial Extracorporeal Membrane Oxygenation patients: a target trial emulation using the OBLEX study data

Liberal or restrictive transfusion for veno-arterial Extracorporeal Membrane Oxygenation patients: a target trial emulation using the OBLEX study data | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Liberal or restrictive transfusion for veno-arterial Extracorporeal Membrane Oxygenation patients: a target trial emulation using the OBLEX study data Thao Thi Phuong Le, Hergen Buscher, Tri-Long Nguyen, Gennaro Martucci, and 28 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6241374/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: The optimal transfusion threshold for patients undergoing venoarterial extracorporeal membrane oxygenation (VA-ECMO) remains uncertain. Methods: We used data from OBLEX (ClinicalTrials.gov: NCT03714048), an international, prospective, observational study conducted across 12 centres in Australia, Europe, and North America between 2019 and 2022. The study collected information on patient demographics, bleeding risk factors, transfusion practices during the first seven days of ECMO, and in-hospital mortality. Using these data, we emulated a target trial comparing the effects of liberal transfusion practice (transfusion initiated at Hb ≥ 90 g/L) and restrictive transfusion practice (transfusion initiated at Hb ≤ 70g/L) on hospital mortality within seven days of ECMO initiation. Sequential trials approach was used to estimate the causal contrast. Results: A total of 534 patients were included, with 46% dying during hospitalisation. After accounting for potential confounders, the liberal transfusion practice demonstrated a modest survival benefit within the first two days of ECMO, with differences in survival probabilities of 12% (95% CI: 3% to 21%) at day 2 and 13% (95% CI: 2% to 25%) at day 3, corresponding to the number needed to treat (NNT) of 8 and 7 respectively. These results were consistent across sensitivity and exploratory analyses. Conclusion: This target trial emulation study suggests that a liberal transfusion threshold may provide a modest survival benefit during the early course of VA-ECMO. Prospective studies are needed to confirm these findings, assess clinical adoption, and investigate underlying mechanism. ECMO VA ECMO transfusion threshold target trial emulation Figures Figure 1 Figure 2 Take home message A liberal transfusion strategy (Hb ≥ 90 g/L) may offer a small survival advantage over a restrictive approach (Hb ≤ 70 g/L) during the first few days of VA-ECMO. Further studies are needed to determine whether adoption in clinical practice is warranted. 140-character Tweet Target trial emulation: Liberal transfusion (Hb ≥90 g/L) may offer a small survival benefit over restrictive (Hb ≤70 g/L) in early VA-ECMO Introduction Extracorporeal Membrane Oxygenation (ECMO) is an invasive and resource-intensive treatment used to support critically ill patients with refractory cardiac arrest, or severe cardiac or respiratory failure [1]. Red blood cell (RBC) transfusion is common in patients undergoing venoarterial ECMO (VA ECMO) due to underlying critical illness and high rates of bleeding in the setting of circuit-induced coagulopathy, the need for anticoagulation, as well as invasive procedures[2]. In a recent study of 419 patients from an international cohort, nearly 90% of all patients received at least one RBC transfusion during VA ECMO[3]. RBC transfusion increases the risk of complications such as transfusion-related acute lung injury (TRALI), transfusion-related circulatory overload (TACO) and transfusion-related immunomodulation (TRIM) [4, 5]. It has been found to be associated with higher mortality rates in ECMO patients, although this relationship may be confounded by the severity of the underlying disease [6, 7]. Identifying the optimal transfusion strategy is crucial for optimising patient outcomes and minimising transfusion-related risks. Two primary transfusion strategies have been explored: restrictive transfusion (triggered by a haemoglobin (Hb) level of 70–90 g/L) and liberal transfusion (triggered by Hb level of 100–120 g/L). Evidence from observational studies suggests that a restrictive strategy was associated with an acceptable outcome and no signal for harm compared with a liberal transfusion strategy[8–10]. However, the findings across the included studies were inconsistent and subject to a high level of bias [9]. Clinical trials in critically ill patients have shown lower RBC transfusion triggers to be non-inferior to higher triggers [11]. However, the certainty of evidence is low [12] and there are differences between these patients and those requiring VA ECMO. Therefore, it remains unknown whether these trial results may be extrapolated to patients requiring VA ECMO. Despite the high use of RBC transfusions and their potential impact on mortality, complications, and costs, as well as the resource limitations that exist for blood products, current knowledge is insufficient to provide strong evidence-based recommendations for transfusion triggers in this highly vulnerable patient group. Large scale multicentre observational studies have only been published in patients on venovenous ECMO support [13] and no randomised controlled studies have been conducted [14]. To address this knowledge gap, we launched the International Observational Study on Blood Management for Mechanical Circulatory Support using Extracorporeal Membrane Oxygenation (OBLEX, ClinicalTrials.gov: NCT03714048). Using the OBLEX data, we emulated a target trial designed to compare the effect of different transfusion thresholds on hospital mortality within 7 days following commencement of ECMO. Method Study population design and participants OBLEX was an international, prospective, observational study aiming to capture detailed information on current blood management in patients treated with VA ECMO at 12 participating centres from three continents (Australia, Europe and North America). OBLEX collected information on patient’s demographics, diagnosis classification, risk factors for bleeding 24 hour prior to ECMO, bleeding and blood management information during the first 7 days of ECMO, and in-hospital mortality. Data was collected from 2019 to 2022. Full details of the study, data collection procedures, data definitions and case report form are described elsewhere. Statistical analysis Target trial To compare the effect of different transfusion strategies on mortality, we emulated a hypothetical randomised controlled trial (so-called “target trial”) using the OBLEX data. Key components of the protocol for our target trial are outlined in Table 1 . Patients were included in the trial if they were older than 18 years of age and received ECMO for mechanical circulatory support. In the target trial, we compared two transfusion strategies (Table 1 ): restrictive vs. liberal transfusion. Under the restrictive transfusion strategy, one unit of RBC was transfused if a patient’s Hb concentration was recorded below 70 g/L. Under the liberal transfusion strategy, one unit of RBC was transfused if a patient’s Hb concentration was recorded below 90 g/L. The primary outcome of the trial was in-hospital mortality. The estimand of interest was the difference in survival probability between the two randomisation arms in a per-protocol population where patients adhered to their assigned transfusion strategy throughout the entire duration of ECMO treatment. Table 1 Description of the target trial Protocol component Target trial Emulation with the OBLEX data Eligibility Inclusion criteria: ECMO patients: ECMO for mechanical circulatory support ECMO using a temporary device containing an oxygenator and an active blood pump Age > 18 Exclusion criteria: ECMO for respiratory support only ECMO treatment outside the intensive care unit only (e.g. theatre or angiography) Enrolment in other studies where a randomized intervention is targeting anticoagulation or blood management. Inclusion criteria: adults patients who received ECMO for cardiac support, observed in the OBLEX study between 2019 and 2022 Exclusion criteria: As in the target trial We also exclude patients with missing data on time-varying confounders or diagnostics group at baseline (n = 11), and patients whose ECMO treatment outside of the participating centre for >24 hours Treatment strategies Restrictive transfusion: maintain Hb concentration ³ 70 g/L Liberal transfusion: maintain Hb concentration ³ 90 g/L No transfusion during 7 days of ECMO Restrictive transfusion trigger arm: Transfusion at Hb £ 70g/L (Low) Liberal transfusion trigger arm: Transfusion at Hb ³ 90g/L (High) Intermediate transfusion arm: Transfusion at Hb between 70g/L and 90g/L Assignment Randomisation at the time of eligibility to either of the treatment strategies. In the emulated trial, patients are not randomly assigned to the treatment strategy. We attempt to emulate baseline randomisation through the sequential trial approach in the analysis Follow-up period From ECMO treatment until hospital discharge or death, whatever comes first As in the target trial Outcome In-hospital death As in the target trial Causal contrasts Per-protocol As in the target trial Analysis plan Difference in survival probabilities during 7 days ECMO between treatment strategies. Estimated using Aalen additive hazards model for the outcome, with an indicator for treatment group Confounding by measured baseline and time-varying covariates is addressed by using sequential trials approach, with inverse probability of censoring weights (details are in the statistical analysis) In the emulated trial, we used observational data to mimic the target trial design while accounting for the characteristics of our data set (Table 1 ). Key modifications were required, particularly in defining the treatment strategies. Based on the original definitions, a patient could be eligible for multiple treatments at a single time point. For example, a patient with a Hb concentration of 95 g/L who had not received a transfusion could theoretically be allocated to both the restrictive and liberal transfusion arms. However, as we did not record Hb levels for those who did not receive a transfusion in the OBLEX study, we were unable to map all patients to the target trial’s specified transfusion strategies. To address this, we redefined the transfusion strategies, using daily average Hb threshold values that triggered transfusions. Specifically, the strategies were: no transfusion initiated during the first 7 days of ECMO (the reference level); transfusion initiated at Hb ≤ 70g/L; transfusion initiated at 70g/L < Hb < 90g/L; and transfusion initiated at Hb ≥ 90 g/L. These three transfusion practices mirrored a strategy of restrictive, intermediate and liberal transfusion pattern, respectively. Including the no-transfusion arm as the reference allowed us to compute the contrast of interest. Specifically, by first comparing each transfusion practice to the reference, we could then indirectly estimate the contrast between restrictive and liberal transfusion. This was due to the transitive property of contrasts defined on the additive scale (i.e. survival probability difference). To account for the lack of randomisation to transfusion strategies in the emulated trial, we used directed acyclic graphs (DAG) to show the assumed relationship between the initiation of transfusion and outcome, thereby informing which variables were potential confounders (see Figure S1 ). We considered both time-fixed and time-varying confounders in our analysis: Baseline (i.e. time-fixed) confounders included age, gender, diagnosis group, renal replacement prior to ECMO, extracorporeal cardiopulmonary resuscitation (ECPR) and surgery; time-varying confounders included severe bleeding (Bleeding Academic Research Consortium (BARC)[15] score≥ 3), any mechanical support, and any renal replacement during ECMO. Further details on how these variables were defined can be found in the Supplement. With regards to the primary outcome, as the exact date of death was not recorded, if a patient died during hospitalisation, we assumed they died soon after their last ECMO episode. This assumption is reasonable based on data from the national ECMO registry (EXCEL)[16], that is, the majority of hospital mortality (97/101, 96%) occurred during ICU stay. If a patient was discharged from the hospital alive, we censored them at the last day of their ECMO or day 7, whichever occurred earlier. Further assumptions we made in order to estimate the difference in marginal survival probability from the OBLEX data are: no interference, positivity, consistency, and no unmeasured confounding[17, 18]. Sequential trials We employed the sequential trials approach[19, 20] to estimate the causal effects of different transfusion practices while acknowledging the observational nature of the OBLEX data. This method involves creating a sequence of nested trials (Figure S2 ). Specifically, a new trial was “started” at each time (day) a patient had initiated a transfusion during ECMO. Any individuals who meet the inclusion and exclusion criteria of the target trial and were still transfusion-naive at that time (i.e. had not yet received a transfusion while on ECMO) were included in the trial. The trial starting at time 1 thus included everyone on the first day of ECMO, comparing patients who were transfused on day 1 and those who were not. The trial starting at time 2 compared patients who initiated transfusion on day 2 and those who did not receive any transfusions on day 1 and 2, etc. Trial starting at time t + 1 are hence nested in trial at time t, comparing patients who initiated transfusion at time t + 1 and those who did not receive any transfusions from day 1 to t + 1 (Figure S2 ). As a result, we had seven sequential trials corresponding to the seven days of ECMO. Within each trial, we implemented artificial censoring at the moment patients deviated from their initial transfusion strategy (e.g., when patients initially assigned to the restrictive transfusion received a liberal transfusion). Thus, all patients in each trial sustained their allocated transfusion strategies (Figure S3). Data from all seven trials were combined for the subsequent analysis. To address selection bias due to artificial informative censoring, we applied stabilized inverse probability of artificial-censoring weights (IPACW) [19]. These weights were derived from the probability of remaining uncensored at time t , given patient covariates. Further details on weight calculation based on the assumed DAG are available in the Supplement. We estimated survival difference between transfusion strategies by fitting a marginal structural model with stabilized IPACWs. Assuming a consistent treatment effect across trials, we fitted the MSM to combined data from seven sequential trials using the Aalen additive hazards model [21], which accommodates time-varying hazard effects [22]. We adjusted for transfusion strategies, all time-fixed confounders, and time-varying confounders measured at the beginning of each trial. Survival probabilities for each strategy were derived from the model’s time-varying hazards [19]. To capture overall uncertainty, we constructed percentile-based 95% confidence intervals via bootstrapping with 1000 samples. In each sample, we repeated the construction of sequential trials, the estimation of weights, and MSM model to account for total uncertainty across all steps of the analysis. Missing data were minimal. We excluded 11 patients with missing baseline diagnosis groups and imputed missing values for 17 cases of renal replacement and one case of severe bleeding during ECMO using the last-observation-carried-forward method. In sensitivity analyses, we refitted the MSM using truncated stabilized IPACWs at the 95th percentile and adjusted for massive transfusion before ECMO as a time-fixed confounder. Additionally, we restricted the cohort to patients whose transfusion episodes fully aligned with one of the three transfusion strategies. Specifically, patients were mapped to strategies based on individual transfusion trigger values rather than the daily average value. All analyses were performed using R, version 4.4.0 [23]. Results Among 545 patients eligible for the target trial, we excluded 11 patients due to missing baseline diagnosis groups. In total, 534 unique patients were included in the sequential trials analysis. Of these, 243 (46%) died during hospitalization. Baseline characteristics of patients by transfusion strategies in the first day of ECMO are presented in Table 2 . There were 192/534 (36%) patients that received transfusion at an intermediate Hb threshold value. Only 48/534 (9.0%) patients received transfusions at a restrictive threshold, and 51/534 (9.6%) at a liberal threshold. Compared to the liberal transfusion arm, patients who received restrictive transfusions were slightly younger, and had undergone more surgery and renal replacement prior to ECMO. Table 2 Baseline characteristics overall and by treatment arm on the first day of ECMO Characteristic Overall, N = 534 1 No transfusions, N = 243 1 Restrictive, N = 48 1 Liberal, N = 51 1 Intermediate, N = 192 1 Age 57 (46, 65) 58 (47, 68) 54 (46, 65) 57 (46, 65) 57 (45, 64) Gender (Female) 177 (33%) 88 (36%) 12 (25%) 19 (37%) 58 (30%) BMI 27.2 (24.2, 31.0) 26.8 (24.2, 31.2) 27.2 (24.6, 32.3) 26.8 (23.6, 30.6) 27.4 (24.2, 30.6) Unknown 8 1 1 1 5 Region Australia 260 (49%) 121 (50%) 28 (58%) 26 (51%) 85 (44%) Europe 215 (40%) 92 (38%) 16 (33%) 23 (45%) 84 (44%) US 59 (11%) 30 (12%) 4 (8.3%) 2 (3.9%) 23 (12%) Diagnosis group Acute myocardial infarction (AMI) 152 (28%) 66 (27%) 16 (33%) 15 (29%) 55 (29%) Chronic cardiomyopathy 54 (10%) 22 (9.1%) 4 (8.3%) 10 (20%) 18 (9.4%) Pulmonary embolism 43 (8.1%) 22 (9.1%) 0 (0%) 4 (7.8%) 17 (8.9%) Myocarditis 41 (7.7%) 18 (7.4%) 2 (4.2%) 3 (5.9%) 18 (9.4%) Other acute cardiomyopathy 244 (46%) 115 (47%) 26 (54%) 19 (37%) 84 (44%) Other characteristics Surgical 188 (35%) 90 (37%) 20 (42%) 14 (27%) 64 (33%) Other mechanical circulatory support 86 (16%) 39 (16%) 9 (19%) 7 (14%) 31 (16%) ECPR 145 (27%) 56 (23%) 15 (31%) 15 (29%) 59 (31%) Renal replacement 49 (9.2%) 20 (8.2%) 7 (15%) 3 (5.9%) 19 (9.9%) 1 Median (IQR); n (%) ECPR: Extracorporeal cardiopulmonary resuscitation The number of patients who received transfusions during the seven days of ECMO is presented in Table S1 . 54% of patients received at least one transfusion on day 1. Among them, 60% received transfusions on two or more separate occasions. Over the following days, approximately 40% of patients received at least one transfusion per day, with most patients receiving only one transfusion episode. Patients changed their transfusion strategy over time. Figure 1A describes the flow of patients between transfusion strategies over the seven days of ECMO. Most transfusions occurred at the intermediate level of Hb threshold. The proportion of patients who received transfusion at the restrictive and liberal levels were smaller and very few of them maintained an assigned transfusion strategy throughout the course of ECMO. Figure 1B and Table S2 depict the number of patients enrolled in the emulated sequential trials analysis. In the combined data across all seven sequential trials, 20/70 (29%) hospital deaths were recorded in the restrictive transfusion arm, while this number in the liberal transfusion arm was 14/71 (20%) (Table S2 ). Estimated survival curves for treatment arms are presented in Fig. 2A. Patients in the restrictive arm had the worst survival outcome compared to those in other transfusion thresholds, especially in the first three days of ECMO. The difference in survival probability between the restrictive and liberal arms is presented in Fig. 2B. A positive difference indicates a favourable effect of the liberal transfusion. Receiving a liberal transfusion practice appeared to improve survival in the first 3 days of ECMO, but showed no benefit in the following days. Estimated differences in the survival probabilities at day 2 and 3 were 12% (95% CI: 3%, 21%) and 13% (95% CI: 2%, 25%), respectively. These differences correspond to numbers needed to treat (NNTs) of 8.3 and 7.7 at the two time points. This implies that treating approximately eight patients with the liberal transfusion practice instead of the restrictive transfusion would save one additional life by day 2 and 3. The distribution of the original weights and the truncated weights are presented in Figure S4. Using truncated weights did not change our conclusion about the difference in survival between the liberal and restricted transfusion practice (Figure S5A). We observed the same result in a sensitivity analysis where we included massive transfusion as a time-fixed confounder (Figure S5B). In the additional analysis, we included 363 patients whose all daily transfusion triggers, rather than just the daily average Hb trigger, were fully consistent with the transfusion practices. Of these, 149 (41%) died during hospitalisation. In terms of the primary endpoint, we observed similar conclusions, with a slightly larger benefit of the liberal transfusion in the first two days (Figure S5C). Specifically, estimated differences in the survival probabilities at day 2 and 3 were 16% (95% CI: 5%, 27%) and 21% (95% CI: 7%, 36%), respectively. Discussion Using a large international cohort database, we evaluated the effects of restrictive versus liberal transfusion practices. Our target trial emulation suggested a potential advantage of a liberal transfusion practice (i.e. transfusion initiated at Hb ≥ 90g/L) over a restrictive strategy (i.e. transfusion initiated at Hb ≤ 70g/L) in the initial days of ECMO. Liberal transfusion was associated with a slight survival benefit during the first three days, but this advantage disappeared afterward, suggesting that others factors may become more important for patient outcomes later in the ECMO course. Oxygen delivery is directly related to cardiac output, oxygen saturation and haemoglobin. In VA ECMO patients, cardiac output includes both native cardiac blood flow and ECMO-delivered flow. If left ventricular ejection fraction is severely impaired in the early phase of VA ECMO, ECMO blood flows may not fully compensate for reduced native output, potentially leading to inadequate oxygen delivery. Additionally, severe shock is common in the early phase of VA ECMO, which can impair endothelial function and tissue oxygenation, while the lack of pulsatility may worsen this situation further. Previous studies have limitations, either focusing on VV ACMO instead of VA ECMO [13], or being observational with a high risk of publication bias and moderate to severe heterogeneity [12]. Trials of restrictive vs. liberal red cell transfusion have reported non-inferiority of restrictive transfusion strategy compared to liberal [11]. However, the recently published MINT trial[24] in patients with myocardial infarct and anaemia, found that a liberal transfusion target of 100 g/L may improve a composite outcome of survival and myocardial infarction at 30 days by 15% compared to a restrictive transfusion strategy (risk ratio: 1.15; 95% CI: 0.99–1.34; P = 0.07). About half of these patients were in ICU at the time of randomisation but only 14% were mechanically ventilated. The European Society of Intensive Care Medicine has recently reviewed the evidence for transfusion strategies in critically ill patients and could not find enough evidence to extrapolate evidence from trials in other cohorts to ECMO patients and did not make a recommendation [12]. The results of the target trial should be interpreted with caution for the following reasons. First, our treatment definitions were based on daily average Hb trigger values in patients who received transfusion. This introduces two complications. First, the transfusion strategy definitions in the emulated trial differ from those in the target trial. In the emulated trial, we compared patients who consistently received transfusions at a liberal Hb threshold with those who consistently received transfusions at a restrictive threshold, mimicking the liberal and restrictive transfusion protocol in the target trial. Second, defining transfusion strategies based on the average Hb trigger may not fully capture actual transfusion practices for patients who received multiple transfusions in a single day. For instance, the intermediate transfusion arm could include patients who were transfused at both restrictive and liberal thresholds on the same day. This heterogeneity within treatment arms challenges the assumption of consistency in the sequential trials analysis. To address this, we performed the additional analysis restricted to patients whose transfusion practices were entirely consistent the defined thresholds. The results of this additional analysis agreed with the main analysis, indicating the observed benefit of liberal transfusion was robust to the variation of Hb triggers within each treatment arm in our data set. Second, even though we employed a carefully designed observational study with the sequential trials approach to mitigate confounding factors, unmeasured confounding due to factors not captured in our analysis, such as baseline illness severity, remains a possibility. Randomized controlled trials are still needed to definitively establish causal effects of different transfusion strategies on mortality in ECMO patients. Third, while the OBLEX study is one of the largest databases for this patient population, the number of patients who received transfusions at the restrictive and liberal thresholds is relatively small. Although the sequential trials approach efficiently utilizes data in this scenario compared to other causal inference methods for time-varying confounders, the limited sample size in these groups might restrict the generalizability of the findings, particularly regarding the potential benefit of restrictive transfusions. In conclusion, our findings suggest that a liberal transfusion practice (transfusions initiated at Hb ≥ 90 g/L) might yield a small survival advantage in the initial days of VA-ECMO compared to a restrictive transfusion practice (transfusions initiated at Hb ≤ 70 g/L). Future studies are needed to confirm this finding, ideally in a randomised controlled trial design to assess clinical adoption, and investigate underlying mechanism. Declarations Acknowledgement The authors acknowledge the financial support of the Blood Synergy program, established under the Australian National Health and Medical Research Council’s Synergy Grants (1189490). This work was completed with thanks to the EXCEL Management Committee, EXCEL Investigators, International ECMO Network, Australian and New Zealand Intensive Care Society Clinical Trials Group, Alfred Hospital, Melbourne, Vic: Jasmin Board, Aidan Burrell, Annalie Jones, Emma Martin, Phoebe McCracken, Vincent Pellegrino, Jayne Sheldrake, Shirley Vallance and Meredith Young; Gold Coast, Qld: Dee Figures, Maimoonbe Gough, Maree Houbert, James McCullough, Julie Pitman, Mandy Tallott and James Winearls; Liverpool Hospital, Sydney, NSW: Anders Aneman, Danielle Austin, Peter McCanny and Jennene Miller; Princess Alexandra Hospital, Brisbane, Qld: Meg Harward, Chris Joyce, Josie McKay, Jason Meyer, and James Walsham; Prince Charles Hospital, Brisbane, Qld: Rachel Bushell, John Fraser, Jayshree Lavana, Dawn Lockwood and Raymond Marteene; Royal Prince Alfred Hospital, Sydney, NSW: Heidi Buhr, Ruaidhri Carey, Jennifer Coles, David Gattas and Richard Totaro; St Vincent’s Hospital Sydney, Sydney, NSW: Hergen Buscher, Priya Nair, Sally Newman and Claire Reynolds; References Lindstrom SJ, Pellegrino VA, Butt WW (2009) Extracorporeal membrane oxygenation. Med J Aust 191:178–182. https://doi.org/10.5694/j.1326-5377.2009.tb02735.x Guimbretière G, Anselmi A, Roisne A, et al (2019) Prognostic impact of blood product transfusion in VA and VV ECMO. Perfusion 34:246–253. https://doi.org/10.1177/0267659118814690 Raasveld SJ, Karami M, Schenk J, et al (2023) Transfusion of red blood cells in venoarterial extracorporeal membrane oxygenation: A multicenter retrospective observational cohort study. Transfusion (Paris) 63:1809–1820. https://doi.org/10.1111/trf.17505 Semple JW, Rebetz J, Kapur R (2019) Transfusion-associated circulatory overload and transfusion-related acute lung injury. Blood 133:1840–1853. https://doi.org/10.1182/blood-2018-10-860809 Blet A, McNeil JB, Josse J, et al (2022) Association between in-ICU red blood cells transfusion and 1-year mortality in ICU survivors. Crit Care 26:307. https://doi.org/10.1186/s13054-022-04171-1 Olek E, Pasierski M, Słomka A, et al (2023) Blood product transfusions on extracorporeal membrane oxygenation: a narrative review. Ann Blood 8:. https://doi.org/10.21037/aob-21-30 Omar HR, Mirsaeidi M, Socias S, et al (2015) Plasma Free Hemoglobin Is an Independent Predictor of Mortality among Patients on Extracorporeal Membrane Oxygenation Support. PLoS ONE 10:e0124034. https://doi.org/10.1371/journal.pone.0124034 Doyle AJ, Richardson C, Sanderson B, et al (2020) Restrictive Transfusion Practice in Adults Receiving Venovenous Extracorporeal Membrane Oxygenation: A Single-Center Experience. Crit Care Explor 2:e0077. https://doi.org/10.1097/CCE.0000000000000077 Abbasciano RG, Yusuff H, Vlaar APJ, et al (2021) Blood Transfusion Threshold in Patients Receiving Extracorporeal Membrane Oxygenation Support for Cardiac and Respiratory Failure—A Systematic Review and Meta-Analysis. J Cardiothorac Vasc Anesth 35:1192–1202. https://doi.org/10.1053/j.jvca.2020.08.068 Agerstrand CL, Burkart KM, Abrams DC, et al (2015) Blood conservation in extracorporeal membrane oxygenation for acute respiratory distress syndrome. Ann Thorac Surg 99:590–595. https://doi.org/10.1016/j.athoracsur.2014.08.039 Carson JL, Stanworth SJ, Guyatt G, et al (2023) Red Blood Cell Transfusion: 2023 AABB International Guidelines. JAMA 330:1892–1902. https://doi.org/10.1001/jama.2023.12914 Vlaar AP, Oczkowski S, de Bruin S, et al (2020) Transfusion strategies in non-bleeding critically ill adults: a clinical practice guideline from the European Society of Intensive Care Medicine. Intensive Care Med 46:673–696. https://doi.org/10.1007/s00134-019-05884-8 Martucci G, Schmidt M, Agerstrand C, et al (2023) Transfusion practice in patients receiving VV ECMO (PROTECMO): a prospective, multicentre, observational study. Lancet Respir Med 11:245–255. https://doi.org/10.1016/S2213-2600(22)00353-8 Hughes T, Zhang D, Nair P, Buscher H (2021) A Systematic Literature Review of Packed Red Cell Transfusion Usage in Adult Extracorporeal Membrane Oxygenation. Membranes 11:251. https://doi.org/10.3390/membranes11040251 Mehran R, Rao SV, Bhatt DL, et al (2011) Standardized Bleeding Definitions for Cardiovascular Clinical Trials: A Consensus Report From the Bleeding Academic Research Consortium. Circulation 123:2736–2747. https://doi.org/10.1161/CIRCULATIONAHA.110.009449 Hodgson CL, Higgins AM, Bailey MJ, et al (2022) Incidence of death or disability at 6 months after extracorporeal membrane oxygenation in Australia: a prospective, multicentre, registry-embedded cohort study. Lancet Respir Med 10:1038–1048. https://doi.org/10.1016/S2213-2600(22)00248-X Smit JM, Krijthe JH, Kant WMR, et al (2023) Causal inference using observational intensive care unit data: a scoping review and recommendations for future practice. Npj Digit Med 6:1–11. https://doi.org/10.1038/s41746-023-00961-1 Hernán M, Robins J (2020) Causal Inference: What If. Boca Raton: Chapman & Hall/CRC. Keogh RH, Gran JM, Seaman SR, et al (2023) Causal inference in survival analysis using longitudinal observational data: Sequential trials and marginal structural models. Stat Med 42:2191–2225. https://doi.org/10.1002/sim.9718 Gran JM, Røysland K, Wolbers M, et al (2010) A sequential Cox approach for estimating the causal effect of treatment in the presence of time-dependent confounding applied to data from the Swiss HIV Cohort Study. Stat Med 29:2757–2768. https://doi.org/10.1002/sim.4048 Aalen OO (1989) A linear regression model for the analysis of life times. Stat Med 8:907–925. https://doi.org/10.1002/sim.4780080803 Samuelsen SO (2023) Cox regression can be collapsible and Aalen regression can be non-collapsible. Lifetime Data Anal 29:403–419. https://doi.org/10.1007/s10985-022-09578-0 R Core Team (2024) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria Carson JL, Brooks MM, Hébert PC, et al (2023) Restrictive or Liberal Transfusion Strategy in Myocardial Infarction and Anemia. N Engl J Med 389:2446–2456. https://doi.org/10.1056/NEJMoa2307983 Supplementary Files STROBEchecklistcohortOBLEX2.docx SupplementOBLEX2.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6241374","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":433395689,"identity":"af3ab159-70f9-437c-aa0f-ba26d6c8042b","order_by":0,"name":"Thao Thi Phuong Le","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzklEQVRIiWNgGAWjYDACCRBRwCAH5jA2AIkDBHTwgLUYMBiTriWxgWgt9tLNDx8wGNikr23vPfjx5w4GOb4bCQRskTlmbMBgkJa77cy5ZGneMwzGkgS1SCSYSTAYHM7ddiPHQJqxjSFxA2Et6d+AWv6nm93IMf75s42hnggtOSBbDiQAtZhJ8LYxJBgQ1HIjp9ggwSDZcNuZM2bWvG0ShjPPPMCvhX1G+sYHHyrs5M2O9xjf/NlmI893nIAtYICkRoII5aNgFIyCUTAKCAIA4pVCYJRErCkAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-9661-6835","institution":"Monash University School of Public Health and Preventive Medicine","correspondingAuthor":true,"prefix":"","firstName":"Thao","middleName":"Thi Phuong","lastName":"Le","suffix":""},{"id":433395690,"identity":"25d884f7-843c-4886-9bc7-0b7f2c9d5eb1","order_by":1,"name":"Hergen Buscher","email":"","orcid":"","institution":"St Vincent's Hospital Sydney","correspondingAuthor":false,"prefix":"","firstName":"Hergen","middleName":"","lastName":"Buscher","suffix":""},{"id":433395691,"identity":"9638df56-9b74-4877-ae5b-89cfda19dceb","order_by":2,"name":"Tri-Long Nguyen","email":"","orcid":"","institution":"University of Copenhagen Department of Public Health: Kobenhavns Universitet Institut for Folkesundhedsvidenskab","correspondingAuthor":false,"prefix":"","firstName":"Tri-Long","middleName":"","lastName":"Nguyen","suffix":""},{"id":433395692,"identity":"2a027535-5a61-4770-879d-611347fda8da","order_by":3,"name":"Gennaro Martucci","email":"","orcid":"","institution":"Istituto Mediterraneo per i Trapianti e Terapie ad Alta Specializzazione","correspondingAuthor":false,"prefix":"","firstName":"Gennaro","middleName":"","lastName":"Martucci","suffix":""},{"id":433395693,"identity":"9527d5d7-88cc-4dc2-b0af-6909ef1b5535","order_by":4,"name":"Johannes Gratz","email":"","orcid":"","institution":"Medical University of Vienna: Medizinische Universitat Wien","correspondingAuthor":false,"prefix":"","firstName":"Johannes","middleName":"","lastName":"Gratz","suffix":""},{"id":433395694,"identity":"7928624c-10c2-44a4-a244-6dfaa821b261","order_by":5,"name":"Georg Trummer","email":"","orcid":"","institution":"Medical Center - University of Freiburg Institute of Immunology: Universitatsklinikum Freiburg Institut fur Immundefizienz","correspondingAuthor":false,"prefix":"","firstName":"Georg","middleName":"","lastName":"Trummer","suffix":""},{"id":433395695,"identity":"3bca3ef0-c6a6-41aa-85d5-47f1c37c2bed","order_by":6,"name":"Matthieu Schmidt","email":"","orcid":"","institution":"Assistance Publique - Hopitaux de Paris","correspondingAuthor":false,"prefix":"","firstName":"Matthieu","middleName":"","lastName":"Schmidt","suffix":""},{"id":433395696,"identity":"32416270-776a-4b96-9812-83e11717eaf8","order_by":7,"name":"Melchior Gautier","email":"","orcid":"","institution":"Assistance Publique - Hopitaux de Paris","correspondingAuthor":false,"prefix":"","firstName":"Melchior","middleName":"","lastName":"Gautier","suffix":""},{"id":433395697,"identity":"f745979b-a181-4e61-95da-437e5d26531f","order_by":8,"name":"Alexis Serra","email":"","orcid":"","institution":"New York University","correspondingAuthor":false,"prefix":"","firstName":"Alexis","middleName":"","lastName":"Serra","suffix":""},{"id":433395698,"identity":"88cb6757-1f92-4e9e-b77c-066bdd848087","order_by":9,"name":"Koji Takeda","email":"","orcid":"","institution":"Columbia University Vagelos College of Physicians and Surgeons","correspondingAuthor":false,"prefix":"","firstName":"Koji","middleName":"","lastName":"Takeda","suffix":""},{"id":433395699,"identity":"b464ec33-0723-4161-b94f-f3971bdab917","order_by":10,"name":"Jan-Steffen Pooth","email":"","orcid":"","institution":"University Medical Center Freiburg: Albert-Ludwigs-Universitat Freiburg Universitatsklinikum Freiburg","correspondingAuthor":false,"prefix":"","firstName":"Jan-Steffen","middleName":"","lastName":"Pooth","suffix":""},{"id":433395700,"identity":"e7cf982c-eb04-4590-8667-8170b38d7c27","order_by":11,"name":"Kevin Rahn","email":"","orcid":"","institution":"University Medical Center Freiburg: Albert-Ludwigs-Universitat Freiburg Universitatsklinikum Freiburg","correspondingAuthor":false,"prefix":"","firstName":"Kevin","middleName":"","lastName":"Rahn","suffix":""},{"id":433395701,"identity":"8b6dacb4-394c-41b6-85be-6d539c7fc9da","order_by":12,"name":"Florian Geismann","email":"","orcid":"","institution":"University Hospital Regensburg Department for Internal Medicine II: Universitatsklinikum Regensburg Klinik und Poliklinik fur Innere Medizin II","correspondingAuthor":false,"prefix":"","firstName":"Florian","middleName":"","lastName":"Geismann","suffix":""},{"id":433395702,"identity":"80399c91-ce8a-4c47-b1f3-1e3443995fd7","order_by":13,"name":"Matthias Lubnow","email":"","orcid":"","institution":"University Hospital Regensburg Department for Internal Medicine II: Universitatsklinikum Regensburg Klinik und Poliklinik fur Innere Medizin II","correspondingAuthor":false,"prefix":"","firstName":"Matthias","middleName":"","lastName":"Lubnow","suffix":""},{"id":433395703,"identity":"c87ec53b-bbf7-4276-9e16-a7992d630b67","order_by":14,"name":"Andrew Retter","email":"","orcid":"","institution":"Guy's and Saint Thomas' Hospitals NHS Trust: Guy's and St Thomas' NHS Foundation Trust","correspondingAuthor":false,"prefix":"","firstName":"Andrew","middleName":"","lastName":"Retter","suffix":""},{"id":433395704,"identity":"f594dfd2-3c98-4001-ac40-20b09900af19","order_by":15,"name":"Priya Nair","email":"","orcid":"","institution":"St Vincent's Hospital Sydney","correspondingAuthor":false,"prefix":"","firstName":"Priya","middleName":"","lastName":"Nair","suffix":""},{"id":433395705,"identity":"6af8ea84-7e39-4461-8489-6a1c575981cb","order_by":16,"name":"Ruan Vlok","email":"","orcid":"","institution":"Royal North Shore Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ruan","middleName":"","lastName":"Vlok","suffix":""},{"id":433395706,"identity":"cb3bae45-c9bf-4f92-9023-8a80bee8addb","order_by":17,"name":"Maithri Siriwardena","email":"","orcid":"","institution":"The Prince Charles Hospital","correspondingAuthor":false,"prefix":"","firstName":"Maithri","middleName":"","lastName":"Siriwardena","suffix":""},{"id":433395707,"identity":"d45ff296-30f6-4b34-ba49-0df0bb3dabb1","order_by":18,"name":"James Winearls","email":"","orcid":"","institution":"Gold Coast University Hospital","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"","lastName":"Winearls","suffix":""},{"id":433395708,"identity":"4ac9522a-5cdd-4349-9606-c22508a4cd0d","order_by":19,"name":"James Walsham","email":"","orcid":"","institution":"Princess Alexandra Hospital","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"","lastName":"Walsham","suffix":""},{"id":433395709,"identity":"f56fd180-3a68-4464-a462-b86b1afbe54d","order_by":20,"name":"David Gattas","email":"","orcid":"","institution":"Royal Prince Alfred Hospital","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Gattas","suffix":""},{"id":433395710,"identity":"c822486f-0195-46ec-990e-07313f5455b0","order_by":21,"name":"Anders Aneman","email":"","orcid":"","institution":"Liverpool Hospital ICU: Liverpool Hospital Intensive Care Unit","correspondingAuthor":false,"prefix":"","firstName":"Anders","middleName":"","lastName":"Aneman","suffix":""},{"id":433395711,"identity":"ca300bc3-37ac-489d-811e-9a80772d5fe5","order_by":22,"name":"Bentley Fulcher","email":"","orcid":"","institution":"Monash University School of Public Health and Preventive Medicine","correspondingAuthor":false,"prefix":"","firstName":"Bentley","middleName":"","lastName":"Fulcher","suffix":""},{"id":433395712,"identity":"22932504-eca5-4e24-8d5a-64dd53eaebcb","order_by":23,"name":"Sally Newman","email":"","orcid":"","institution":"St Vincent's Hospital Sydney","correspondingAuthor":false,"prefix":"","firstName":"Sally","middleName":"","lastName":"Newman","suffix":""},{"id":433395713,"identity":"2adb1d18-0588-461c-9da5-aa0ca5451d13","order_by":24,"name":"Claire Reynolds","email":"","orcid":"","institution":"St Vincent's Hospital Sydney","correspondingAuthor":false,"prefix":"","firstName":"Claire","middleName":"","lastName":"Reynolds","suffix":""},{"id":433395714,"identity":"77586e21-7653-4e77-a409-34c260492fe1","order_by":25,"name":"Anthonio Arcadipane","email":"","orcid":"","institution":"Istituto Mediterraneo per i Trapianti e Terapie ad Alta Specializzazione","correspondingAuthor":false,"prefix":"","firstName":"Anthonio","middleName":"","lastName":"Arcadipane","suffix":""},{"id":433395715,"identity":"8346f3fb-bea5-453b-8f9f-4d7e58ab9f21","order_by":26,"name":"Kiran Shekar","email":"","orcid":"","institution":"The Prince Charles Hospital","correspondingAuthor":false,"prefix":"","firstName":"Kiran","middleName":"","lastName":"Shekar","suffix":""},{"id":433395716,"identity":"27676a7e-7846-4987-9561-43b59166fb5c","order_by":27,"name":"Carol Hodgson","email":"","orcid":"","institution":"Monash University School of Public Health and Preventive Medicine","correspondingAuthor":false,"prefix":"","firstName":"Carol","middleName":"","lastName":"Hodgson","suffix":""},{"id":433395717,"identity":"cd8454d7-b23d-4391-a45d-057f7c221ccd","order_by":28,"name":"Vincent Pellegrino","email":"","orcid":"","institution":"Alfred Hospital: The Alfred","correspondingAuthor":false,"prefix":"","firstName":"Vincent","middleName":"","lastName":"Pellegrino","suffix":""},{"id":433395718,"identity":"fe5650ed-2619-4fb6-8051-76d0b692788a","order_by":29,"name":"Thomas Mueller","email":"","orcid":"","institution":"University Hospital Regensburg Department for Internal Medicine II: Universitatsklinikum Regensburg Klinik und Poliklinik fur Innere Medizin II","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Mueller","suffix":""},{"id":433395719,"identity":"f63e4a3f-64aa-4396-b1fe-62958615a7a2","order_by":30,"name":"Daniel Brodie","email":"","orcid":"","institution":"Johns Hopkins University Department of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Brodie","suffix":""},{"id":433395720,"identity":"48eb4459-4c70-4c46-bcc4-51d702c92c92","order_by":31,"name":"Zoe McQuilten","email":"","orcid":"","institution":"Monash University School of Public Health and Preventive Medicine","correspondingAuthor":false,"prefix":"","firstName":"Zoe","middleName":"","lastName":"McQuilten","suffix":""}],"badges":[],"createdAt":"2025-03-17 06:06:27","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6241374/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6241374/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":80046210,"identity":"a82a4e92-5552-4c37-bc84-62b8f89d5114","added_by":"auto","created_at":"2025-04-07 09:49:45","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":140206,"visible":true,"origin":"","legend":"\u003cp\u003eA) Transfusion pattern in the OBLEX study B) Number of patients in the sequential trials\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6241374/v1/cfe048211028ec245c333e87.png"},{"id":80045054,"identity":"ed021ff1-e28f-45ad-934d-3f49dcdbec7e","added_by":"auto","created_at":"2025-04-07 09:41:45","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":152066,"visible":true,"origin":"","legend":"\u003cp\u003eA) Estimated survival curve under all transfusion strategies, B) Estimated survival difference between the liberal and restrictive transfusion strategies. The shaded areas show the 95% confidence intervals.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6241374/v1/9588a23ee5bc994e7ab7592f.png"},{"id":80049786,"identity":"a3ada5ff-98dd-4b09-8284-f9528420559d","added_by":"auto","created_at":"2025-04-07 10:13:46","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1028703,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6241374/v1/7ed7a546-d3e8-418e-a454-c5fe7708b23f.pdf"},{"id":80045049,"identity":"a82d9136-bbc6-4daf-abd3-ec5389009f6a","added_by":"auto","created_at":"2025-04-07 09:41:45","extension":"docx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":41333,"visible":true,"origin":"","legend":"","description":"","filename":"STROBEchecklistcohortOBLEX2.docx","url":"https://assets-eu.researchsquare.com/files/rs-6241374/v1/baffd62a8555fff84199c5b2.docx"},{"id":80046213,"identity":"e258fdb3-bbdd-4edb-991d-f835faeb8265","added_by":"auto","created_at":"2025-04-07 09:49:46","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":331762,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementOBLEX2.docx","url":"https://assets-eu.researchsquare.com/files/rs-6241374/v1/71a76ef5256cd90dce2225ff.docx"}],"financialInterests":"","formattedTitle":"Liberal or restrictive transfusion for veno-arterial Extracorporeal Membrane Oxygenation patients: a target trial emulation using the OBLEX study data","fulltext":[{"header":"Take home message","content":"\u003cp\u003eA liberal transfusion strategy (Hb ≥ 90 g/L) may offer a small survival advantage over a restrictive approach (Hb ≤ 70 g/L) during the first few days of VA-ECMO.\u0026nbsp;Further studies are needed to determine whether adoption in clinical practice is warranted.\u003c/p\u003e\n\u003ch2\u003e140-character Tweet\u003c/h2\u003e\n\u003cp\u003eTarget trial emulation: Liberal transfusion (Hb ≥90 g/L) may offer a small survival benefit over restrictive (Hb ≤70 g/L) in early VA-ECMO\u003c/p\u003e"},{"header":"Introduction","content":"\u003cp\u003eExtracorporeal Membrane Oxygenation (ECMO) is an invasive and resource-intensive treatment used to support critically ill patients with refractory cardiac arrest, or severe cardiac or respiratory failure [1]. Red blood cell (RBC) transfusion is common in patients undergoing venoarterial ECMO (VA ECMO) due to underlying critical illness and high rates of bleeding in the setting of circuit-induced coagulopathy, the need for anticoagulation, as well as invasive procedures[2]. In a recent study of 419 patients from an international cohort, nearly 90% of all patients received at least one RBC transfusion during VA ECMO[3].\u003c/p\u003e \u003cp\u003eRBC transfusion increases the risk of complications such as transfusion-related acute lung injury (TRALI), transfusion-related circulatory overload (TACO) and transfusion-related immunomodulation (TRIM) [4, 5]. It has been found to be associated with higher mortality rates in ECMO patients, although this relationship may be confounded by the severity of the underlying disease [6, 7]. Identifying the optimal transfusion strategy is crucial for optimising patient outcomes and minimising transfusion-related risks. Two primary transfusion strategies have been explored: restrictive transfusion (triggered by a haemoglobin (Hb) level of 70–90 g/L) and liberal transfusion (triggered by Hb level of 100–120 g/L). Evidence from observational studies suggests that a restrictive strategy was associated with an acceptable outcome and no signal for harm compared with a liberal transfusion strategy[8–10]. However, the findings across the included studies were inconsistent and subject to a high level of bias [9].\u003c/p\u003e \u003cp\u003eClinical trials in critically ill patients have shown lower RBC transfusion triggers to be non-inferior to higher triggers [11]. However, the certainty of evidence is low [12] and there are differences between these patients and those requiring VA ECMO. Therefore, it remains unknown whether these trial results may be extrapolated to patients requiring VA ECMO.\u003c/p\u003e \u003cp\u003eDespite the high use of RBC transfusions and their potential impact on mortality, complications, and costs, as well as the resource limitations that exist for blood products, current knowledge is insufficient to provide strong evidence-based recommendations for transfusion triggers in this highly vulnerable patient group. Large scale multicentre observational studies have only been published in patients on venovenous ECMO support [13] and no randomised controlled studies have been conducted [14]. To address this knowledge gap, we launched the International Observational Study on Blood Management for Mechanical Circulatory Support using Extracorporeal Membrane Oxygenation (OBLEX, ClinicalTrials.gov: NCT03714048). Using the OBLEX data, we emulated a target trial designed to compare the effect of different transfusion thresholds on hospital mortality within 7 days following commencement of ECMO.\u003c/p\u003e "},{"header":"Method","content":"\u003cp\u003eStudy population design and participants\u003c/p\u003e\n\u003cp\u003eOBLEX was an international, prospective, observational study aiming to capture detailed information on current blood management in patients treated with VA ECMO at 12 participating centres from three continents (Australia, Europe and North America). OBLEX collected information on patient\u0026rsquo;s demographics, diagnosis classification, risk factors for bleeding 24 hour prior to ECMO, bleeding and blood management information during the first 7 days of ECMO, and in-hospital mortality. Data was collected from 2019 to 2022. Full details of the study, data collection procedures, data definitions and case report form are described elsewhere.\u003c/p\u003e\n\u003ch2\u003eStatistical analysis\u003c/h2\u003e\n\u003cp\u003eTarget trial\u003c/p\u003e\n\u003cp\u003eTo compare the effect of different transfusion strategies on mortality, we emulated a hypothetical randomised controlled trial (so-called \u0026ldquo;target trial\u0026rdquo;) using the OBLEX data. Key components of the protocol for our target trial are outlined in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. Patients were included in the trial if they were older than 18 years of age and received ECMO for mechanical circulatory support. In the target trial, we compared two transfusion strategies (Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e): restrictive vs. liberal transfusion. Under the restrictive transfusion strategy, one unit of RBC was transfused if a patient\u0026rsquo;s Hb concentration was recorded below 70 g/L. Under the liberal transfusion strategy, one unit of RBC was transfused if a patient\u0026rsquo;s Hb concentration was recorded below 90 g/L. The primary outcome of the trial was in-hospital mortality. The estimand of interest was the difference in survival probability between the two randomisation arms in a per-protocol population where patients adhered to their assigned transfusion strategy throughout the entire duration of ECMO treatment.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eDescription of the target trial\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eProtocol component\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTarget trial\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEmulation with the OBLEX data\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eEligibility\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eInclusion criteria:\u003c/p\u003e\n \u003cp\u003eECMO patients:\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eECMO for mechanical circulatory support\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eECMO using a temporary device containing an oxygenator and an active blood pump\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eAge \u0026gt; 18\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eExclusion criteria:\u003c/p\u003e\n \u003cp\u003eECMO for respiratory support only\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eECMO treatment outside the intensive care unit only (e.g. theatre or angiography)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eEnrolment in other studies where a randomized intervention is targeting anticoagulation or blood management.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eInclusion criteria: adults patients who received ECMO for cardiac support, observed in the OBLEX study between 2019 and 2022\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eExclusion criteria:\u003c/p\u003e\n \u003cp\u003eAs in the target trial\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eWe also exclude patients with missing data on time-varying confounders or diagnostics group at baseline (n = 11), and patients whose ECMO treatment outside of the participating centre for \u0026gt;24 hours\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eTreatment strategies\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003col\u003e\n \u003cli\u003eRestrictive transfusion: maintain Hb concentration\u0026nbsp;\u0026sup3;\u0026nbsp;70 g/L\u003c/li\u003e\n \u003cli\u003eLiberal transfusion: maintain Hb concentration\u0026nbsp;\u0026sup3;\u0026nbsp;90 g/L\u003c/li\u003e\n \u003c/ol\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003col\u003e\n \u003cli\u003eNo transfusion during 7 days of ECMO\u003c/li\u003e\n \u003cli\u003eRestrictive transfusion trigger arm: Transfusion at Hb\u0026nbsp;\u0026pound;\u0026nbsp;70g/L (Low)\u003c/li\u003e\n \u003cli\u003eLiberal transfusion trigger arm: Transfusion at Hb\u0026nbsp;\u0026sup3;\u0026nbsp;90g/L (High)\u003c/li\u003e\n \u003cli\u003eIntermediate transfusion arm: Transfusion at Hb between 70g/L and 90g/L\u003c/li\u003e\n \u003c/ol\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eAssignment\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eRandomisation at the time of eligibility to either of the treatment strategies.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eIn the emulated trial, patients are not randomly assigned to the treatment strategy. We attempt to emulate baseline randomisation through the sequential trial approach in the analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eFollow-up period\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eFrom ECMO treatment until hospital discharge or death, whatever comes first\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eAs in the target trial\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eOutcome\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eIn-hospital death\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eAs in the target trial\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eCausal contrasts\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003ePer-protocol\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eAs in the target trial\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eAnalysis plan\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eDifference in survival probabilities during 7 days ECMO between treatment strategies.\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eEstimated using Aalen additive hazards model for the outcome, with an indicator for treatment group\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 33.3333%;\"\u003e\n \u003cp\u003eConfounding by measured baseline and time-varying covariates is addressed by using sequential trials approach, with inverse probability of censoring weights (details are in the statistical analysis)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003eIn the emulated trial, we used observational data to mimic the target trial design while accounting for the characteristics of our data set (Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e). Key modifications were required, particularly in defining the treatment strategies. Based on the original definitions, a patient could be eligible for multiple treatments at a single time point. For example, a patient with a Hb concentration of 95 g/L who had not received a transfusion could theoretically be allocated to both the restrictive and liberal transfusion arms. However, as we did not record Hb levels for those who did not receive a transfusion in the OBLEX study, we were unable to map all patients to the target trial\u0026rsquo;s specified transfusion strategies. To address this, we redefined the transfusion strategies, using daily average Hb threshold values that triggered transfusions. Specifically, the strategies were: no transfusion initiated during the first 7 days of ECMO (the reference level); transfusion initiated at Hb\u0026thinsp;\u0026le;\u0026thinsp;70g/L; transfusion initiated at 70g/L\u0026thinsp;\u0026lt;\u0026thinsp;Hb\u0026thinsp;\u0026lt;\u0026thinsp;90g/L; and transfusion initiated at Hb\u0026thinsp;\u0026ge;\u0026thinsp;90 g/L. These three transfusion practices mirrored a strategy of restrictive, intermediate and liberal transfusion pattern, respectively. Including the no-transfusion arm as the reference allowed us to compute the contrast of interest. Specifically, by first comparing each transfusion practice to the reference, we could then indirectly estimate the contrast between restrictive and liberal transfusion. This was due to the transitive property of contrasts defined on the additive scale (i.e. survival probability difference).\u003c/p\u003e\n \u003cp\u003eTo account for the lack of randomisation to transfusion strategies in the emulated trial, we used directed acyclic graphs (DAG) to show the assumed relationship between the initiation of transfusion and outcome, thereby informing which variables were potential confounders (see Figure \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e). We considered both time-fixed and time-varying confounders in our analysis: Baseline (i.e. time-fixed) confounders included age, gender, diagnosis group, renal replacement prior to ECMO, extracorporeal cardiopulmonary resuscitation (ECPR) and surgery; time-varying confounders included severe bleeding (Bleeding Academic Research Consortium (BARC)[15] score\u0026ge; 3), any mechanical support, and any renal replacement during ECMO. Further details on how these variables were defined can be found in the Supplement.\u003c/p\u003e\n \u003cp\u003eWith regards to the primary outcome, as the exact date of death was not recorded, if a patient died during hospitalisation, we assumed they died soon after their last ECMO episode. This assumption is reasonable based on data from the national ECMO registry (EXCEL)[16], that is, the majority of hospital mortality (97/101, 96%) occurred during ICU stay. If a patient was discharged from the hospital alive, we censored them at the last day of their ECMO or day 7, whichever occurred earlier. Further assumptions we made in order to estimate the difference in marginal survival probability from the OBLEX data are: no interference, positivity, consistency, and no unmeasured confounding[17, 18].\u003c/p\u003e\n \u003ch2\u003eSequential trials\u003c/h2\u003e\n \u003cp\u003eWe employed the sequential trials approach[19, 20] to estimate the causal effects of different transfusion practices while acknowledging the observational nature of the OBLEX data. This method involves creating a sequence of nested trials (Figure \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e). Specifically, a new trial was \u0026ldquo;started\u0026rdquo; at each time (day) a patient had initiated a transfusion during ECMO. Any individuals who meet the inclusion and exclusion criteria of the target trial and were still transfusion-naive at that time (i.e. had not yet received a transfusion while on ECMO) were included in the trial. The trial starting at time 1 thus included everyone on the first day of ECMO, comparing patients who were transfused on day 1 and those who were not. The trial starting at time 2 compared patients who initiated transfusion on day 2 and those who did not receive any transfusions on day 1 and 2, etc. Trial starting at time \u003cem\u003et\u003c/em\u003e\u0026thinsp;+\u0026thinsp;1 are hence nested in trial at time t, comparing patients who initiated transfusion at time \u003cem\u003et\u003c/em\u003e\u0026thinsp;+\u0026thinsp;1 and those who did not receive any transfusions from day 1 to \u003cem\u003et\u003c/em\u003e\u0026thinsp;+\u0026thinsp;1 (Figure \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e). As a result, we had seven sequential trials corresponding to the seven days of ECMO. Within each trial, we implemented artificial censoring at the moment patients deviated from their initial transfusion strategy (e.g., when patients initially assigned to the restrictive transfusion received a liberal transfusion). Thus, all patients in each trial sustained their allocated transfusion strategies (Figure S3). Data from all seven trials were combined for the subsequent analysis.\u003c/p\u003e\n \u003cp\u003eTo address selection bias due to artificial informative censoring, we applied stabilized inverse probability of artificial-censoring weights (IPACW) [19]. These weights were derived from the probability of remaining uncensored at time \u003cem\u003et\u003c/em\u003e, given patient covariates. Further details on weight calculation based on the assumed DAG are available in the Supplement.\u003c/p\u003e\n \u003cp\u003eWe estimated survival difference between transfusion strategies by fitting a marginal structural model with stabilized IPACWs. Assuming a consistent treatment effect across trials, we fitted the MSM to combined data from seven sequential trials using the Aalen additive hazards model [21], which accommodates time-varying hazard effects [22]. We adjusted for transfusion strategies, all time-fixed confounders, and time-varying confounders measured at the beginning of each trial. Survival probabilities for each strategy were derived from the model\u0026rsquo;s time-varying hazards [19]. To capture overall uncertainty, we constructed percentile-based 95% confidence intervals via bootstrapping with 1000 samples. In each sample, we repeated the construction of sequential trials, the estimation of weights, and MSM model to account for total uncertainty across all steps of the analysis.\u003c/p\u003e\n \u003cp\u003eMissing data were minimal. We excluded 11 patients with missing baseline diagnosis groups and imputed missing values for 17 cases of renal replacement and one case of severe bleeding during ECMO using the last-observation-carried-forward method.\u003c/p\u003e\n \u003cp\u003eIn sensitivity analyses, we refitted the MSM using truncated stabilized IPACWs at the 95th percentile and adjusted for massive transfusion before ECMO as a time-fixed confounder. Additionally, we restricted the cohort to patients whose transfusion episodes fully aligned with one of the three transfusion strategies. Specifically, patients were mapped to strategies based on individual transfusion trigger values rather than the daily average value.\u003c/p\u003e\n \u003cp\u003eAll analyses were performed using R, version 4.4.0 [23].\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eAmong 545 patients eligible for the target trial, we excluded 11 patients due to missing baseline diagnosis groups. In total, 534 unique patients were included in the sequential trials analysis. Of these, 243 (46%) died during hospitalization.\u003c/p\u003e\n\u003cp\u003eBaseline characteristics of patients by transfusion strategies in the first day of ECMO are presented in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. There were 192/534 (36%) patients that received transfusion at an intermediate Hb threshold value. Only 48/534 (9.0%) patients received transfusions at a restrictive threshold, and 51/534 (9.6%) at a liberal threshold. Compared to the liberal transfusion arm, patients who received restrictive transfusions were slightly younger, and had undergone more surgery and renal replacement prior to ECMO.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eBaseline characteristics overall and by treatment arm on the first day of ECMO\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCharacteristic\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOverall, N\u0026thinsp;=\u0026thinsp;534\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eNo transfusions, N\u0026thinsp;=\u0026thinsp;243\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRestrictive, N\u0026thinsp;=\u0026thinsp;48\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLiberal, N\u0026thinsp;=\u0026thinsp;51\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eIntermediate, N\u0026thinsp;=\u0026thinsp;192\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e57 (46, 65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e58 (47, 68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e54 (46, 65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e57 (46, 65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e57 (45, 64)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGender (Female)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e177 (33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e88 (36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12 (25%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19 (37%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e58 (30%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.2 (24.2, 31.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26.8 (24.2, 31.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.2 (24.6, 32.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26.8 (23.6, 30.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.4 (24.2, 30.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRegion\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAustralia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e260 (49%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e121 (50%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e28 (58%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (51%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e85 (44%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEurope\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e215 (40%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e92 (38%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16 (33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (45%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e84 (44%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e59 (11%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30 (12%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4 (8.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (3.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (12%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiagnosis group\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAcute myocardial infarction (AMI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e152 (28%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e66 (27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16 (33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (29%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e55 (29%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eChronic cardiomyopathy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e54 (10%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22 (9.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4 (8.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18 (9.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePulmonary embolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (8.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22 (9.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4 (7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (8.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMyocarditis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e41 (7.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18 (7.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (4.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (5.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18 (9.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOther acute cardiomyopathy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e244 (46%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e115 (47%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (54%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19 (37%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e84 (44%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\" align=\"left\"\u003e\n \u003cp\u003eOther characteristics\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSurgical\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e188 (35%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e90 (37%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (42%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14 (27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e64 (33%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOther mechanical circulatory support\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e86 (16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e39 (16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 (19%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 (14%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e31 (16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eECPR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e145 (27%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e56 (23%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (31%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (29%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e59 (31%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRenal replacement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e49 (9.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (8.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 (15%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (5.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19 (9.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\" align=\"left\"\u003e\n \u003cp\u003e\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e Median (IQR); n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\"\u003eECPR: Extracorporeal cardiopulmonary resuscitation\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eThe number of patients who received transfusions during the seven days of ECMO is presented in Table \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e. 54% of patients received at least one transfusion on day 1. Among them, 60% received transfusions on two or more separate occasions. Over the following days, approximately 40% of patients received at least one transfusion per day, with most patients receiving only one transfusion episode. Patients changed their transfusion strategy over time. Figure\u0026nbsp;1A describes the flow of patients between transfusion strategies over the seven days of ECMO. Most transfusions occurred at the intermediate level of Hb threshold. The proportion of patients who received transfusion at the restrictive and liberal levels were smaller and very few of them maintained an assigned transfusion strategy throughout the course of ECMO.\u003c/p\u003e\n\u003cp\u003eFigure 1B and Table \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e depict the number of patients enrolled in the emulated sequential trials analysis. In the combined data across all seven sequential trials, 20/70 (29%) hospital deaths were recorded in the restrictive transfusion arm, while this number in the liberal transfusion arm was 14/71 (20%) (Table \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003e). Estimated survival curves for treatment arms are presented in Fig.\u0026nbsp;2A. Patients in the restrictive arm had the worst survival outcome compared to those in other transfusion thresholds, especially in the first three days of ECMO. The difference in survival probability between the restrictive and liberal arms is presented in Fig.\u0026nbsp;2B. A positive difference indicates a favourable effect of the liberal transfusion. Receiving a liberal transfusion practice appeared to improve survival in the first 3 days of ECMO, but showed no benefit in the following days. Estimated differences in the survival probabilities at day 2 and 3 were 12% (95% CI: 3%, 21%) and 13% (95% CI: 2%, 25%), respectively. These differences correspond to numbers needed to treat (NNTs) of 8.3 and 7.7 at the two time points. This implies that treating approximately eight patients with the liberal transfusion practice instead of the restrictive transfusion would save one additional life by day 2 and 3.\u003c/p\u003e\n\u003cp\u003eThe distribution of the original weights and the truncated weights are presented in Figure S4. Using truncated weights did not change our conclusion about the difference in survival between the liberal and restricted transfusion practice (Figure S5A). We observed the same result in a sensitivity analysis where we included massive transfusion as a time-fixed confounder (Figure S5B). In the additional analysis, we included 363 patients whose all daily transfusion triggers, rather than just the daily average Hb trigger, were fully consistent with the transfusion practices. Of these, 149 (41%) died during hospitalisation. In terms of the primary endpoint, we observed similar conclusions, with a slightly larger benefit of the liberal transfusion in the first two days (Figure S5C). Specifically, estimated differences in the survival probabilities at day 2 and 3 were 16% (95% CI: 5%, 27%) and 21% (95% CI: 7%, 36%), respectively.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eUsing a large international cohort database, we evaluated the effects of restrictive versus liberal transfusion practices. Our target trial emulation suggested a potential advantage of a liberal transfusion practice (i.e. transfusion initiated at Hb\u0026thinsp;\u0026ge;\u0026thinsp;90g/L) over a restrictive strategy (i.e. transfusion initiated at Hb \u0026le; 70g/L) in the initial days of ECMO.\u003c/p\u003e \u003cp\u003eLiberal transfusion was associated with a slight survival benefit during the first three days, but this advantage disappeared afterward, suggesting that others factors may become more important for patient outcomes later in the ECMO course. Oxygen delivery is directly related to cardiac output, oxygen saturation and haemoglobin. In VA ECMO patients, cardiac output includes both native cardiac blood flow and ECMO-delivered flow. If left ventricular ejection fraction is severely impaired in the early phase of VA ECMO, ECMO blood flows may not fully compensate for reduced native output, potentially leading to inadequate oxygen delivery. Additionally, severe shock is common in the early phase of VA ECMO, which can impair endothelial function and tissue oxygenation, while the lack of pulsatility may worsen this situation further.\u003c/p\u003e \u003cp\u003ePrevious studies have limitations, either focusing on VV ACMO instead of VA ECMO [13], or being observational with a high risk of publication bias and moderate to severe heterogeneity [12]. Trials of restrictive vs. liberal red cell transfusion have reported non-inferiority of restrictive transfusion strategy compared to liberal [11]. However, the recently published MINT trial[24] in patients with myocardial infarct and anaemia, found that a liberal transfusion target of 100 g/L may improve a composite outcome of survival and myocardial infarction at 30 days by 15% compared to a restrictive transfusion strategy (risk ratio: 1.15; 95% CI: 0.99\u0026ndash;1.34; P\u0026thinsp;=\u0026thinsp;0.07). About half of these patients were in ICU at the time of randomisation but only 14% were mechanically ventilated. The European Society of Intensive Care Medicine has recently reviewed the evidence for transfusion strategies in critically ill patients and could not find enough evidence to extrapolate evidence from trials in other cohorts to ECMO patients and did not make a recommendation [12].\u003c/p\u003e \u003cp\u003eThe results of the target trial should be interpreted with caution for the following reasons. First, our treatment definitions were based on daily average Hb trigger values in patients who received transfusion. This introduces two complications. First, the transfusion strategy definitions in the emulated trial differ from those in the target trial. In the emulated trial, we compared patients who consistently received transfusions at a liberal Hb threshold with those who consistently received transfusions at a restrictive threshold, mimicking the liberal and restrictive transfusion protocol in the target trial. Second, defining transfusion strategies based on the average Hb trigger may not fully capture actual transfusion practices for patients who received multiple transfusions in a single day. For instance, the intermediate transfusion arm could include patients who were transfused at both restrictive and liberal thresholds on the same day. This heterogeneity within treatment arms challenges the assumption of consistency in the sequential trials analysis. To address this, we performed the additional analysis restricted to patients whose transfusion practices were entirely consistent the defined thresholds. The results of this additional analysis agreed with the main analysis, indicating the observed benefit of liberal transfusion was robust to the variation of Hb triggers within each treatment arm in our data set. Second, even though we employed a carefully designed observational study with the sequential trials approach to mitigate confounding factors, unmeasured confounding due to factors not captured in our analysis, such as baseline illness severity, remains a possibility. Randomized controlled trials are still needed to definitively establish causal effects of different transfusion strategies on mortality in ECMO patients. Third, while the OBLEX study is one of the largest databases for this patient population, the number of patients who received transfusions at the restrictive and liberal thresholds is relatively small. Although the sequential trials approach efficiently utilizes data in this scenario compared to other causal inference methods for time-varying confounders, the limited sample size in these groups might restrict the generalizability of the findings, particularly regarding the potential benefit of restrictive transfusions.\u003c/p\u003e \u003cp\u003eIn conclusion, our findings suggest that a liberal transfusion practice (transfusions initiated at Hb\u0026thinsp;\u0026ge;\u0026thinsp;90 g/L) might yield a small survival advantage in the initial days of VA-ECMO compared to a restrictive transfusion practice (transfusions initiated at Hb \u0026le; 70 g/L). Future studies are needed to confirm this finding, ideally in a randomised controlled trial design to assess clinical adoption, and investigate underlying mechanism.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAcknowledgement\u003c/h2\u003e\n\u003cp\u003eThe authors acknowledge the financial support of the Blood Synergy program, established under the Australian National Health and Medical Research Council’s Synergy Grants (1189490).\u003c/p\u003e\n\u003cp\u003eThis work was completed with thanks to the EXCEL Management Committee, EXCEL Investigators, International ECMO Network, Australian and New Zealand Intensive Care Society Clinical Trials Group, Alfred Hospital, Melbourne, Vic: Jasmin Board, Aidan Burrell, Annalie Jones, Emma Martin, Phoebe McCracken, Vincent Pellegrino, Jayne Sheldrake, Shirley Vallance and Meredith Young; Gold Coast, Qld: Dee Figures, Maimoonbe Gough, Maree Houbert, James McCullough, Julie Pitman, Mandy Tallott and James Winearls; Liverpool Hospital, Sydney, NSW: Anders Aneman, Danielle Austin, Peter McCanny and Jennene Miller; Princess Alexandra Hospital, Brisbane, Qld: Meg Harward, Chris Joyce, Josie McKay, Jason Meyer, and James Walsham; Prince Charles Hospital, Brisbane, Qld: Rachel Bushell, John Fraser, Jayshree Lavana, Dawn Lockwood and Raymond Marteene; Royal Prince Alfred Hospital, Sydney, NSW: Heidi Buhr, Ruaidhri Carey, Jennifer Coles, David Gattas and Richard Totaro; St Vincent’s Hospital Sydney, Sydney, NSW: Hergen Buscher, Priya Nair, Sally Newman and Claire Reynolds;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e Lindstrom SJ, Pellegrino VA, Butt WW (2009) Extracorporeal membrane oxygenation. Med J Aust 191:178\u0026ndash;182. https://doi.org/10.5694/j.1326-5377.2009.tb02735.x\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Guimbreti\u0026egrave;re G, Anselmi A, Roisne A, et al (2019) Prognostic impact of blood product transfusion in VA and VV ECMO. Perfusion 34:246\u0026ndash;253. https://doi.org/10.1177/0267659118814690\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Raasveld SJ, Karami M, Schenk J, et al (2023) Transfusion of red blood cells in venoarterial extracorporeal membrane oxygenation: A multicenter retrospective observational cohort study. Transfusion (Paris) 63:1809\u0026ndash;1820. https://doi.org/10.1111/trf.17505\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Semple JW, Rebetz J, Kapur R (2019) Transfusion-associated circulatory overload and transfusion-related acute lung injury. Blood 133:1840\u0026ndash;1853. https://doi.org/10.1182/blood-2018-10-860809\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Blet A, McNeil JB, Josse J, et al (2022) Association between in-ICU red blood cells transfusion and 1-year mortality in ICU survivors. Crit Care 26:307. https://doi.org/10.1186/s13054-022-04171-1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Olek E, Pasierski M, Słomka A, et al (2023) Blood product transfusions on extracorporeal membrane oxygenation: a narrative review. Ann Blood 8:. https://doi.org/10.21037/aob-21-30\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Omar HR, Mirsaeidi M, Socias S, et al (2015) Plasma Free Hemoglobin Is an Independent Predictor of Mortality among Patients on Extracorporeal Membrane Oxygenation Support. PLoS ONE 10:e0124034. https://doi.org/10.1371/journal.pone.0124034\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Doyle AJ, Richardson C, Sanderson B, et al (2020) Restrictive Transfusion Practice in Adults Receiving Venovenous Extracorporeal Membrane Oxygenation: A Single-Center Experience. Crit Care Explor 2:e0077. https://doi.org/10.1097/CCE.0000000000000077\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Abbasciano RG, Yusuff H, Vlaar APJ, et al (2021) Blood Transfusion Threshold in Patients Receiving Extracorporeal Membrane Oxygenation Support for Cardiac and Respiratory Failure\u0026mdash;A Systematic Review and Meta-Analysis. J Cardiothorac Vasc Anesth 35:1192\u0026ndash;1202. https://doi.org/10.1053/j.jvca.2020.08.068\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Agerstrand CL, Burkart KM, Abrams DC, et al (2015) Blood conservation in extracorporeal membrane oxygenation for acute respiratory distress syndrome. Ann Thorac Surg 99:590\u0026ndash;595. https://doi.org/10.1016/j.athoracsur.2014.08.039\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Carson JL, Stanworth SJ, Guyatt G, et al (2023) Red Blood Cell Transfusion: 2023 AABB International Guidelines. JAMA 330:1892\u0026ndash;1902. https://doi.org/10.1001/jama.2023.12914\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Vlaar AP, Oczkowski S, de Bruin S, et al (2020) Transfusion strategies in non-bleeding critically ill adults: a clinical practice guideline from the European Society of Intensive Care Medicine. Intensive Care Med 46:673\u0026ndash;696. https://doi.org/10.1007/s00134-019-05884-8\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Martucci G, Schmidt M, Agerstrand C, et al (2023) Transfusion practice in patients receiving VV ECMO (PROTECMO): a prospective, multicentre, observational study. Lancet Respir Med 11:245\u0026ndash;255. https://doi.org/10.1016/S2213-2600(22)00353-8\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Hughes T, Zhang D, Nair P, Buscher H (2021) A Systematic Literature Review of Packed Red Cell Transfusion Usage in Adult Extracorporeal Membrane Oxygenation. Membranes 11:251. https://doi.org/10.3390/membranes11040251\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Mehran R, Rao SV, Bhatt DL, et al (2011) Standardized Bleeding Definitions for Cardiovascular Clinical Trials: A Consensus Report From the Bleeding Academic Research Consortium. Circulation 123:2736\u0026ndash;2747. https://doi.org/10.1161/CIRCULATIONAHA.110.009449\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Hodgson CL, Higgins AM, Bailey MJ, et al (2022) Incidence of death or disability at 6 months after extracorporeal membrane oxygenation in Australia: a prospective, multicentre, registry-embedded cohort study. Lancet Respir Med 10:1038\u0026ndash;1048. https://doi.org/10.1016/S2213-2600(22)00248-X\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Smit JM, Krijthe JH, Kant WMR, et al (2023) Causal inference using observational intensive care unit data: a scoping review and recommendations for future practice. Npj Digit Med 6:1\u0026ndash;11. https://doi.org/10.1038/s41746-023-00961-1\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Hern\u0026aacute;n M, Robins J (2020) Causal Inference: What If. Boca Raton: Chapman \u0026amp; Hall/CRC.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Keogh RH, Gran JM, Seaman SR, et al (2023) Causal inference in survival analysis using longitudinal observational data: Sequential trials and marginal structural models. Stat Med 42:2191\u0026ndash;2225. https://doi.org/10.1002/sim.9718\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Gran JM, R\u0026oslash;ysland K, Wolbers M, et al (2010) A sequential Cox approach for estimating the causal effect of treatment in the presence of time-dependent confounding applied to data from the Swiss HIV Cohort Study. Stat Med 29:2757\u0026ndash;2768. https://doi.org/10.1002/sim.4048\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Aalen OO (1989) A linear regression model for the analysis of life times. Stat Med 8:907\u0026ndash;925. https://doi.org/10.1002/sim.4780080803\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Samuelsen SO (2023) Cox regression can be collapsible and Aalen regression can be non-collapsible. Lifetime Data Anal 29:403\u0026ndash;419. https://doi.org/10.1007/s10985-022-09578-0\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e R Core Team (2024) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e Carson JL, Brooks MM, H\u0026eacute;bert PC, et al (2023) Restrictive or Liberal Transfusion Strategy in Myocardial Infarction and Anemia. N Engl J Med 389:2446\u0026ndash;2456. https://doi.org/10.1056/NEJMoa2307983\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"ECMO, VA ECMO, transfusion threshold, target trial emulation","lastPublishedDoi":"10.21203/rs.3.rs-6241374/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6241374/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eThe optimal transfusion threshold for patients undergoing venoarterial extracorporeal membrane oxygenation (VA-ECMO) remains uncertain.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eWe used data from OBLEX (ClinicalTrials.gov: NCT03714048), an international, prospective, observational study conducted across 12 centres in Australia, Europe, and North America between 2019 and 2022. The study collected information on patient demographics, bleeding risk factors, transfusion practices during the first seven days of ECMO, and in-hospital mortality. Using these data, we emulated a target trial comparing the effects of liberal transfusion practice (transfusion initiated at Hb ≥ 90 g/L) and restrictive transfusion practice (transfusion initiated at Hb ≤ 70g/L) on hospital mortality within seven days of ECMO initiation. Sequential trials approach was used to estimate the causal contrast.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eA total of 534 patients were included, with 46% dying during hospitalisation. After accounting for potential confounders, the liberal transfusion practice demonstrated a modest survival benefit within the first two days of ECMO, with differences in survival probabilities of 12% (95% CI: 3% to 21%) at day 2 and 13% (95% CI: 2% to 25%) at day 3, corresponding to the number needed to treat (NNT) of 8 and 7 respectively. \u0026nbsp;These results were consistent across sensitivity and exploratory analyses.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThis target trial emulation study suggests that\u003cstrong\u003e \u003c/strong\u003ea liberal transfusion threshold may provide a modest survival benefit during the early course of VA-ECMO. Prospective studies are needed to confirm these findings, assess clinical adoption, and investigate underlying mechanism.\u003c/p\u003e","manuscriptTitle":"Liberal or restrictive transfusion for veno-arterial Extracorporeal Membrane Oxygenation patients: a target trial emulation using the OBLEX study data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-07 09:41:41","doi":"10.21203/rs.3.rs-6241374/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"2614583c-2e4f-4972-b46e-ba391288752a","owner":[],"postedDate":"April 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-04-07T09:41:43+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-07 09:41:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6241374","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6241374","identity":"rs-6241374","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}