• Sarah Hearnshaw, Susan Brunskill, Carolyn Doree, Chris Hyde, Simon Travis, and Michael F Murphy.
• Systematic Review Initiative, NHS Blood and Transplant, John Radcliffe Hospital, Oxford, UK, OX3 9BQ. sarah.hearnshaw@nbs.nhs.uk
• Cochrane Db Syst Rev. 2009 Jan 1(2):CD006613.

BackgroundUpper gastrointestinal haemorrhage affects 50 to 150 per 100,000 adults per year and has a high mortality. Red blood cell transfusions are frequently given, but their impact on rebleeding rates and mortality is not known.ObjectivesTo assess the effects of red blood cell transfusion in adults with upper gastrointestinal haemorrhage.Search StrategyWe searched the Cochrane Upper Gastrointestinal and Pancreatic Diseases Group Trials Register to February 2008, the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2008, issue 1), MEDLINE (1950 to February 2008), EMBASE (1974 to February 2008), the Systematic Review Initiative database of randomised controlled trials, haematology and gastroenterology conference proceedings, and reference lists of articles. We also searched databases of ongoing clinical trials.Selection CriteriaRandomised and quasi-randomised studies comparing red blood cell transfusion and standard care with other intravenous fluid and standard care regimens in haemodynamically stable and haemodynamically unstable adults with upper gastrointestinal haemorrhage.Data Collection And AnalysisTwo authors independently assessed trial quality and extracted data. We contacted study authors for additional information.Main ResultsThree trials involving 126 patients were included in the review, with complete data available for 93 patients. The participants were heterogeneous and none of the three studies examined exactly the same interventions or measured the same outcomes. Only two trials reported mortality data and the summary relative risk for mortality of the intervention was 5.4 (95% CI 0.27 to 107.09). One trial reported increased coagulation times in the transfused group, and reported these patients to have increased rates of rebleeding. None of the studies reported adverse events directly related to red blood cell transfusion. Methodological deficiencies, including allocation concealment, generation of random sequences and blinding, simply compound the uncertainty surrounding analysis. None of the studies were appropriately powered and in the largest study less than half the participants were included in the final analysis.One randomised controlled trial of restrictive versus liberal red blood cell transfusion, which aims to recruit 860 patients, has yet to be completed.Authors' ConclusionsThere were more deaths and more rebleeding in the transfusion arms of the combined studies, but the small numbers of participants and large volume of missing data limit the significance of the findings. The studies in this review do not provide useful data regarding outcomes following red blood cell transfusion for acute upper gastrointestinal haemorrhage. They appear to exclude large survival benefit. Large, well-concealed randomised controlled trials of sufficient power are urgently needed.

7 keywords...

hide…