Optimal Transfusion Thresholds in Patients with Acute Brain Injury: A Bayesian Network Meta-analysis of Randomized Controlled Trials

Why clinicians should know about it

• Picked for Critical Care and Intensive Care Medicine (top studies of the week, 18 October 2026): Bayesian network meta‑analysis of transfusion thresholds in acute brain injury

Abstract

BACKGROUND: The optimal transfusion threshold for acute brain injury (ABI) remains uncertain. Previous studies comparing liberal and restrictive transfusion strategies have yielded inconsistent findings. Therefore, we performed a Bayesian network meta-analysis to compare the effectiveness and safety of different transfusion thresholds in patients with ABI. METHODS: A systematic search of PubMed, Embase, Medical Literature Analysis and Retrieval System Online (MEDLINE), and the Cochrane Library was performed to identify randomized controlled trials (RCTs) comparing red blood cell transfusion thresholds in patients with ABI. The primary outcomes were mortality and unfavorable functional outcome. Risk ratios (RRs) with 95% credible intervals (CrIs) were estimated from the Bayesian network meta-analysis, and treatment rankings were assessed using the surface under the cumulative ranking curve (SUCRA). RESULTS: Six RCTs involving 2599 patients were included. Four transfusion thresholds were evaluated (hemoglobin levels of 7, 8, 9, and 10 g/dL). Compared with a transfusion threshold of 7 g/dL, a threshold of 9 g/dL was associated with a lower risk of unfavorable functional outcome (RR 0.86, 95% CrI 0.78-0.94; direct evidence from two trials). Compared with 8 g/dL, the estimated effect favored a 9 g/dL threshold, but the evidence was indirect and did not reach statistical significance (RR 0.79, 95% CrI 0.63-1.00). No clear differences in adverse events were observed across the evaluated thresholds. The 9 g/dL threshold ranked highest according to SUCRA values for both mortality and neurological outcomes. Meta-regression analysis suggested that baseline hemoglobin level may represent a potential source of heterogeneity in the effect of liberal transfusion strategies on functional outcomes. CONCLUSIONS: Among the evaluated transfusion thresholds, a threshold of 9 g/dL ranked most favorably for functional outcome and was associated with a lower risk of unfavorable functional outcome than 7 g/dL, without clear differences in mortality, infectious complications, or thromboembolic events.

Abstract as published, via PubMed.