Acute upper gastrointestinal bleeding is a potentially life-threatening medical emergency that frequently requires red blood cell (RBC) transfusions. However, the optimal hemoglobin thresholds for transfusion is controversial. The objective of this study was to establish the most efficacious transfusion threshold. Methods: A systematic review of the published literature was completed. MEDLINE, Health technology assessment database, Cochrane central register, Cochrane database of systematic

more »

... ws, and EMBASE were searched from inception to May 2015 using search terms including "blood transfusions", "hemoglobin", and "red blood cell". Studies were included if they: reported original data, were peer-reviewed, studied adult populations, were randomized controlled clinical trials and primarily focused on clinical efficacy or effectiveness of liberal and restrictive pre-transfusion hemoglobin level thresholds. Quality was assessed using the Cochrane Risk of Bias tool. Data were extracted and meta-analysis was conducted using a random effects model to determine the risk ratio for: all-cause mortality, further bleeding and any adverse events. All steps were completed independently by two reviewers. Results: The literature search identified 4037 unique abstracts. Of these, 156 abstracts proceeded to full text review. 154 articles were excluded during full-text review resulting in 2 articles for final analysis. The total number of participants included was 701. The hemoglobin threshold to transfuse RBC varied between 70-80g/L versus 90-100g/L in restrictive and liberal policies, respectively. Both studies were at low risk of bias. Meta-analysis resulted in a pooled decreased risk of all-cause mortality (RR 0.65, 95% CI 0.44-0.96), re-bleeding (RR 0.63, 95% CI 0.46-0.85) and adverse events (RR 0.83, 95% CI 0.73-0.95) in the restrictive blood transfusion group versus the liberal blood transfusion group. Conclusion: While the evidence is limited, the risk of death is lower and there is no significant harm for a restrictive strategy. In this context, there is a decreased risk of transfusion associated adverse events among those receiving a restrictive strategy and should be considered for its impact on patient safety and health system resources.