Transfusion Requirements in Gastrointestinal (GI) Bleeding

HYPOTHESIS At the present time there does not exist established criteria to decide when it is necessary a blood transfusion in a patient with digestive hemorrhage, neither which should be the red cells concentrates (RBC) amount that the most of the patients will need. In clinical studies made in critical patients undergoing a by-pass coronary surgery, the strategy of restrictive transfusion showed results similar to the one obtained with more liberal strategies, even with an improvement of the survival and a smaller rate of complications related to transfusion. In animal models of GI bleeding (and also in human studies in the traumatic hemorrhage), precocious or vigorous transfusion made hemostasia more difficult. Also an increase in the rate of rebleeding has been observed, suggesting that arterial hypotension combined with hypovolemia aid hemostasia, stabilize the clot. It leads to diminish by itself the rebleeding rate. In the same way, in patients with portal hypertension associated hemorrhage, aggressive replacement of volemia causes increases on the portal pressure, and that could lead in a condition of bigger difficulty for the control of the hemorrhage and greater rate of recidiva. On the other hand, potential complications associated with the transfusion would be seen potentially reduced. Our randomized prospective study tries to demonstrate that the use of a restrictive strategy in the sanguineous transfusion in patients with acute GI upper bleeding can be at least as beneficial than the habitually used. Moreover, restricted transfusion in these patients could improve short term survival, as well as a smaller rate transfusion-related or rebleeding. In portal hypertension related hemorrhage, restricted transfusion could avoid fluctuations of portal pressure caused by transfusion during the acute phase of hemorrhage, which could favor hemostasia in these patients. OBJECTIVES The main objective is to evaluate if restrictive transfusion criterion in patients with acute upper GI hemorrhage can maintain the rates of survival obtained using habitual transfusion criteria, or to even improve them. The more important secondary targets consist in evaluating if these restrictive transfusional parameters are also accompanied by a better control of the hemorrhage, and also to evaluate if this is accompanied by a smaller rate of complications. Other additional objectives would be: * Effect on changes in portal pressure and its correlation with the clinical evolution. * Hospital stay and estimation of economic cost. STUDY DESIGN Ours is a prospective, randomized and controlled study, in which patients with acute upper GI bleeding will be randomized into two groups of transfusional RBC treatment with: Group 1 (of restricted transfusion), that constitutes the training group: they will receive UCH transfusion when the hemoglobin descends below 70 G/L, to maintain values of hemoglobin of 70 to 90 G/L. Group 2 (of habitual transfusion), that constitutes the group control: they will receive transfusion according to habitual practice, when the hemoglobin descends below 90 G/L, to maintain values of hemoglobin of 90 to 110 G/L. Randomization will be made by means of a closed opaque envelope that will contain the treatment option that will have been obtained by means of a listing of random numbers generated by computer. The patients will be randomized as soon as the inclusion criteria/exclusion has been verified. Randomization will be stratified according to the origin of the hemorrhage (related to portal hypertension or not). NUMBER OF PREDICTED SUBJECTS AND JUSTIFICATION: 430 patients in every group will be required (860 altogether), to objective a mortality reduction of 5%, with a global expected mortality secondary to GI bleeding at the control group of 10%, with a type I error of 5% and a type II error of 20%. 280 patients in each group will be required (560 altogether) to objective a difference of 6%, with the detailed parameters. An expected period of 3 years to include this