Objective: To evaluate physicians' probability estimates of acute appendicitis based on structured collection of clinical data.
Design: Open prospective study.
Setting: District hospital, Norway.
Subjects: 304 patients admitted with suspected acute appendicitis.
Interventions: Initial diagnostic accuracy of physicians was compared with corresponding results from a computer model.
Main outcome measures: The estimated probabilities of appendicitis in different testing groups were analysed using receiver operating characteristic (ROC) curves.
Results: Physicians' estimates had a mean area under ROC-curve of 0.81 (95% CI 0.79 to 0.82), not significantly different from the computer model. Both correlated well with the actual rate of appendicitis, but the physicians tended to overestimate the probability by 10%.
Conclusion: Physicians' probability estimates perform rather well. Further attempts to implement a probabilistic approach in the diagnostic process of acute appendicitis therefore seem justified.