Predicted survival in peptic ulcer patients based on computer analysis of preoperative variables

A prospective study designed to emphasize and quantitate the operative risk of patients preparing to undergo surgery for the treatment of complicated peptic ulcer disease is presented. Data were gathered from 347 consecutive patients operated on with benign gastric and/or duodenal ulcers in a Veterans Hospital over an 8-year period. Resident surgeons performed all operations and for the most part decided on the operative procedure used, with advice from attending faculty. Preoperative factors influencing the operative mortality in 34 patients were compared with those in surviving patients and subjected to a multivariant discriminant function analysis by computer. Ten variables were identified as being significantly different (P less than 0.05-P less than 0.01) between the survivor and non-survivor groups. Using the discriminant weights of these variables, a computer program was written to calculate the 30-day operative mortality of any preoperative patient based on this past experience. The accuracy of the program is excellent in good risk patients; i.e., a predicted greater than 90% chance of survival was correct 98.9% of the time with 3 deaths in 279 patients. Patients at the low end of the scale (less than 10%) were predicted with 85% accuracy. In the last 8 months, 29 patients have undergone surgery after prospective computer assessment of their operativ risk. All have survived with a predicted chance of greater than 50%. Four patients died with survival chances predicted at 4, 2, 1, and 1%. The computer may be used as an educational vehicle for sharpening our preoperative assessment of a patient with ulcer disease, particularly regarding operative risk.