To evaluate whether head-region injury, as defined by Abbreviated Injury Scale (AIS) coding, is associated with transfusion requirements in patients with severe trauma. This study is a post-hoc analysis of the RESTRIC trial, a multicentre, cluster-randomised, crossover, non-inferiority trial comparing restrictive (7–9 g/dL) and liberal (10–12 g/dL) blood transfusion strategies for patients with severe haemorrhagic trauma. Adult patients with severe trauma anticipated to require transfusion were included. Head injury was defined as an AIS score ≥ 2 in the head region. The primary outcome was cumulative red blood cell (RBC) transfusion volume within 28 days. Key secondary outcomes included RBC transfusion volume within 24 and 48 h. Propensity score matching (PSM) was used to balance baseline injury severity using Injury Severity Score (ISS) and, separately, AIS scores for body regions excluding the head. Sensitivity analyses were conducted using generalised linear mixed models (GLMM) with random intercepts for study centre. Among the 411 patients, 157 (38.2%) had head AIS ≥ 2. In unadjusted analyses, 28-day RBC transfusion volumes were similar between patients with and without head injury (median 10 vs. 10 units). After adjustment using PSM and GLMM, no evidence of an independent association between head-region injury and RBC transfusion volume was observed at 24 h, 48 h, or cumulatively at 28 days (IRR 0.97, 95% CI 0.78–1.21). In contrast, transfusion strategy, abdominal or limb injury, and early interventions were consistently associated with increased transfusion requirements. In this post-hoc analysis of a randomised transfusion trial, we found no evidence of an independent association between head-region injury and transfusion requirements within the limits of statistical power and exposure definition. These findings should be interpreted as an absence of evidence rather than evidence of no effect and should not be extrapolated to isolated intracranial traumatic brain injury. Further studies incorporating precise identification of intracranial injury and coagulation phenotypes are warranted.
Nagasawa et al. (Tue,) studied this question.