Liberal Versus Restrictive Transfusion Strategy in Patients With Subarachnoid Hemorrhage: A Meta-Analysis.

Background: The effect of different thresholds for packed red blood cell transfusion on neurologic outcomes in patients with subarachnoid hemorrhage and anemia is uncertain. This meta-analysis aimed to evaluate the effect of liberal versus restrictive transfusion strategies on functional outcomes in subarachnoid hemorrhage. Methods: We systematically searched MEDLINE (inception-March 26, 2025) for randomized controlled trials comparing functional neurological outcomes in adults with aneurysmal subarachnoid hemorrhage assigned to receive packed red blood cell transfusion using high versus low hemoglobin thresholds. Risk of bias was assessed using Cochrane Risk of Bias tool 2.0. Unfavorable neurologic outcomes were defined based on trial-specific primary neurologic outcome definitions (Glasgow Outcome Scale Extended score ≤5 or modified Rankin Scale score ≥3 or modified Rankin Scale score ≥4). An exploratory analysis using Glasgow Outcome Scale Extended score ≤5 or modified Rankin Scale score ≥3 as unfavorable neurologic outcome was also performed. We calculated the pooled risk ratio (RR) with 95% CI for unfavorable neurologic outcome using random effect models and assessed heterogeneity using Cochran's Q test. Results: Of 1628 studies, 3 trials (n = 953 patients) met the inclusion criteria. Over a follow-up of 3-12 months, 181/470 patients (38.5%) randomized to liberal transfusion strategies (hemoglobin<9-11.5g/dL) compared with 214/483 patients (44.3%) in the restrictive strategy group (hemoglobin<7-10g/dL) experienced unfavorable neurologic outcomes. Liberal (versus restrictive) packed red blood cell transfusion strategies resulted in a nonsignificant reduction (RR, 0.88; 95% CI, 0.77-1.01; P = 0.06; I2 = 0%) in unfavorable neurologic outcomes per trial-specific definitions and RR = 0.89 (95% CI, 0.79-0.99; P = 0.04; I2 = 0%) when unfavorable outcome was defined as Glasgow Outcome Scale Extended score ≤5 or modified Rankin Scale score ≥3. Sensitivity analysis, excluding the small pilot trial (n = 44) with a higher hemoglobin threshold, did not change the summary effect. Conclusions: Our study demonstrated a trend toward better neurologic outcomes with liberal transfusion strategies in patients with subarachnoid hemorrhage. Further studies are required to determine the optimal transfusion strategy for this high-risk population and identify subgroups that are most likely to benefit from the liberal strategy.