The Non-Analytical Basis of Clinical Reasoning.

This paper explores the assertion that much of clinical diagnostic thinking is based on the rapid and unconscious matching of the presenting problem to a similar, previously encountered, problem. This 'non-analytic' form of concept formation has been described in the psychology literature for over a decade. From this theory, we deduce and test several hypotheses: 1) Diagnosis is based in part on similarity to a particular previous example. In studies in dermatology, specific similarity accounts for about 30% of diagnosis. 2) When experts err, these errors are as likely as novices to occur on typical presentations. For residents, general practitioners and dermatologists, about 40% of errors were on typical slides. 3) Features are re-interpreted in light of diagnostic hypotheses. In radiology, attaching a standard positive history to the film bag increased the number of features seen on both normal and abnormal films by about 50%. 4) Experts cannot predict errors of other experts. In dermatology, experts predicted only 11-60% of errors committed by their colleagues. We conclude that amassing prior instances is an important component of expertise, and education should recognize this element.