It has been suggested that mortality due to upper gastrointestinal haemorrhage may be reduced by restricting blood transfusion [1], We have assessed whether this is due to an anticoagulant effect in a prospective randomised trial.One hundred patients with severe, acute gastrointestinal haemorrhage were randomised to receive either at least 2 units of blood during the first 24 hours of admission, or no blood unless their haemaglobin was lessthan 8g/dl or they were shocked. Minor bleeds and varices were excluded As hypercoagulation cannot be measured using conventional coagulation tests, fresh whole blood coagulation was measured by the Biobridge Impedance Clotting Time (ICT). Coagulation was assessed at 24 hour intervals and compared to age matched controls with the results expressed as mean ± sem.The ICT on admission for the transfusion group (n=50) was 3.2±0.2 mins compared to 10±0.2 mins in controls. This hyper-coagulable state was partially reversed to 6.4±0.3 mins at 24 hours (p<0.001). The 50 allocated to receive no blood had a similar ICT on admission of 4.4±0.4 mins but the hypercoagulable state was maintained with ICT at 24 hours of 4.320.4 mins. Only 2 patients not transfused rebled compared to 15 in the early transfusion group (p<0.001). Five patients died, and they were all in the early transfusion group.These findings show there is a hypercoagulable response to haemorrhage which is partially reversed by blood transfusion leading to rebleeding
