DECISION-ANALYSIS IN UPPER GASTROINTESTINAL-BLEEDING DUE TO ULCEROGENIC DRUGS

All patients presenting with acute upper gastrointestinal bleeding due to aspirin or NSAIDs between January 1988 and November 1992 were admi tted to an intercompartmental medical/surgical unit. For each patient an early clinical and endoscopic assessment was carried out and a rapi d surgical intervention was performed in those at high risk. Of the 10 47 patients included in our study 692 (66%) were bleeding from peptic ulcers, 17 (1.6%) had anastomotic ulcers, and the remainder had erosio ns. A major stigma was present in 394 patients (38%), with active blee ding present in 121 patients (31%). Sixty-two patients (5.9%) underwen t surgery. The overall mortality was 2.9%, with the operative mortalit y eight times greater than that for patients treated medically. With o ur data we used an independent Bayes' model using rapid surgical inter vention as the outcome criterion to be predicted. Our analysis has sho wn that the following factors are of value as prediction criteria: ulc er evidence, ulcer operation, prior GI bleeding, shock on admission, h igh location of gastric ulcers, active bleeding, or a visible vessel. The presence of shock on admission and the endoscopic findings have th e greater intrinsic prognostic value and add much information to other clinical data obtained on admission.