Impact of a cardiology data bank on physicians' prognostic estimates. Evidence that cardiology fellows change their estimates to become as accurate as the faculty.

To determine whether physicians would be influenced by the prognostic information in a large coronary artery disease data bank, cardiology faculty and fellows made initial estimates of the prognoses of their patients and then made revised final estimates after seeing the outcome of matched patients (OMP) from the data bank. The faculty cardiologists' original estimates proved to be as accurate as those of the data bank's OMP, and the faculty revised their estimates minimally in response to the data bank's OMP. Conversely, the cardiology fellows' original estimates were less accurate than the data bank's OMP, and under all observed circumstances the fellows responded more to the data bank's OMP than did the faculty. As a result, the accuracy of the fellows' final estimates was similar to the accuracies of both the faculty cardiologists and the data bank's OMP. Computerized data banks seem more likely to have impact when their information is provided to physicians who are relatively inexperienced with the disease in question.