Selection of patients for surgery following peptic ulcer haemorrhage.

Surgery remains the only widely available and well-proven means of stopping haemorrhage from peptic ulcers and preventing its recurrence but carries an unavoidable morbidity. If surgery is to be used to maximum effect with minimum morbidity, an accurate means of predicting which patients will suffer further haemorrhage is needed. Although over 80 per cent of patients who rebleed have the endoscopic stigmata of haemorrhage, a policy of operation in all patients with stigmata would lead to a very high operation rate and a high proportion of unnecessary operations, as one-half of the patients with stigmata do not rebleed. Clinical data were collected prospectively from 278 cases of peptic ulcer haemorrhage. The data from a randomly selected 75 per cent of the cases were analysed by stepwise logistical regression. Patients who had the endoscopic stigmata of haemorrhage and who had a probability of further haemorrhage, calculated from the regression equation, of more than 0.2 were identified as a high risk group. This definition was validated using the 25 per cent of cases not used in the initial analysis. Eighty-four per cent of patients in the high risk group suffered further haemorrhage and all such patients therefore require early surgery: such a policy would have resulted in an operation rate of 28 per cent. Thirty per cent of the patients who had further haemorrhage were not identified as being at high risk but none of them had a severe rebleed. The regression equation greatly enhanced the value of stigmata in guiding surgical decision making and merits further evaluation.