Non-inferiority of a red-blood-cell–only transfusion strategy compared with balanced resuscitation in adults with massive gastrointestinal haemorrhage: a propensity-score–weighted cohort study
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Balanced (1:1:1) transfusion of red blood cells (RBCs), plasma, and platelets is the standard of care in trauma-induced massive haemorrhage, where early coagulopathy is a defining feature. In gastrointestinal (GI) haemorrhage this physiology is non-prominent, and whether plasma and platelets provide benefit when ≥ 10 RBC units are required within 24 hours is unknown.
Objective
To test whether a red-blood-cell-only (RBC-only) transfusion strategy is non-inferior to a balanced (Balanced) strategy for in-hospital mortality in adults meeting massive-transfusion criteria for GI haemorrhage.
Design
Single-centre retrospective cohort of 559 adult massive-transfusion encounters (536 patients; 2021–2025) with a primary admitting diagnosis of upper, lower, or unspecified GI haemorrhage. Exposures were RBC-only versus Balanced (RBCs with any plasma and/or platelets). The primary outcome was in-hospital mortality, with a pre-specified 5-percentage-point (pp) non-inferiority margin on the absolute risk difference and a 3-pp sensitivity margin. Analysis used augmented inverse-probability-of-treatment weighting (AIPTW) with bootstrap inference (2,000 resamples by patient). Five pre-specified sensitivity analyses were performed.
Results
505 encounters (90.3%) received RBC-only and 54 (9.7%) received Balanced transfusion. The AIPTW risk difference for in-hospital mortality (RBC-only − Balanced) was -19.8 pp (95% CI -68.1 − -2.2 pp). Non-inferiority was demonstrated at both the primary 5-pp and the more stringent 3-pp margins. Five pre-specified sensitivity analyses, (1) a propensity-score matched cohort, (2) a complete-case model incorporating INR, (3) a broader GI diagnosis set (n = 749), (4) a first encounter per patient restriction, and (5) E-value bound analysis were concordant with the primary estimate.
Conclusion
In this propensity-score–weighted cohort of adults with massive GI haemorrhage, an RBC-only transfusion strategy was non-inferior to a balanced strategy for in-hospital mortality at both 5-pp and 3-pp margins. The findings support individualized use of plasma and platelets in GI haemorrhage rather than reflexive application of the 1:1:1 trauma protocol; prospective confirmation is warranted.
Significance of this study
What is already known on this topic
Balanced 1:1:1 transfusion is the standard of care for trauma-induced massive haemorrhage, where early coagulopathy, hypothermia, and acidosis frequently coexist. Whether this model should be extrapolated to non-traumatic gastrointestinal haemorrhage, a population in which the trauma triad is generally absent, is uncertain. To date, randomized trials in acute GI bleeding tested restrictive RBC thresholds and antifibrinolytic therapy but have not directly compared red-cell-only with balanced component strategies in patients meeting massive-transfusion criteria. 1–3
What this study adds
In a propensity-score–weighted cohort of 559 adults meeting massive-transfusion criteria for GI haemorrhage, an RBC-only strategy was non-inferior to a balanced strategy for in-hospital mortality at both the pre-specified 5-pp margin and a more stringent 3-pp sensitivity margin. The finding was consistent across all pre-specified sensitivity analyses, including 1:1 propensity-score matching, complete-case adjustment for admission INR, a broader GI-diagnosis cohort, a first encounter per patient restriction, and an E-value bound for unmeasured confounding.
How this study might affect research, practice or policy
These results support individualized use of plasma and platelets rather than universal and reflexive 1:1:1 transfusion in GI haemorrhage. The results support a randomized trial comparing goal-directed component therapy with a fixed-ratio resuscitation in non-traumatic massive GI bleeding.
