[Medical decision sciences in the context of clinical pathology]

Medical decision sciences are the scientific approaches to rational decision making under uncertainty. They are arbitrarily subdivided into prescriptive quantitative approach and descriptive psychological approach. The former approach, i.e., decision analysis, is a method of problem solving according to the following procedures: (1) structure the problem (creating a decision tree), (2) apply probabilities, (3) apply values, (4) calculate expected values, (5) perform sensitivity analysis. This methodology makes it clear that once the sensitivity and specificity of a test are determined, estimation of prior probability becomes critically important in interpreting the test results. Usually, prior probability in a given clinical situation is predicted based on patients' demographics, symptoms and previous medical history, physical examination and prior test results. There are, however, many psychological pitfalls leading to wrong estimation of prior probability as follows: (1) premature closure of information processing in the mind, (2) negligence of prior probability from the beginning, (3) regression toward the mean, (4) effect of the number of specimen on the probabilistic fluctuation, (5) gambler's fallacy, (6) effect of vivid memory, (7) selection bias, (8) negligence of negative data, (9) effect of group norm. In order to estimate prior probability as accurately as possible, several strategies are advocated: (1) cognitive psychological approaches, including always thinking about the results of other possible choices, paying attention to the negative data, and trying not to depend on memory as possible, (2) objective estimation using, e.g., the following formula: prior probability = [proportion of patient with abnormal test result-(1-specificity)]/[sensitivity-(1-specificity)].(ABSTRACT TRUNCATED AT 250 WORDS)