A concise model for the management of possible appendicitis.

Abstract

A conventional decision tree model for the management of patients with possible appendicitis requires at least 15 pieces of information. It makes three assumptions. A more concise decision tree model of the problem is presented here. It requires six pieces of information and prevents unintended mathematical shifts between disease states. It makes an additional five assumptions. The concise decision tree model was compared with the larger original model. Expert surgeons gave opinions on the management of a theoretical patient with possible appendicitis and the information necessary to solve the problem by decision analysis. Information from 25 surgeons whose opinions were consistent with analyses of their information using the original large decision tree was used to reanalyze the problem using the concise decision tree. The concise decision tree agreed with the original analysis in 24 of 25 cases (96%), despite violation of the five additional assumptions in 103 of 125 possible instances. The surgeon's decision can be shown by this model to be a function of the probability of appendicitis, the probability of perforation during observation, and the relative differences in outcomes between avoidable perforation and unnecessary operation.