[Diagnosis and clinical decision making: a conceptional framework for predictive pathology].

Abstract

In the clinical pathway of diagnosis and therapy of diseases two decisions are distinguished: diagnostic and therapeutic decision. The former is analysed by decision tables, the latter by decision trees. In both decisions pathology plays a dominant role, especially as a gold standard that is a test to which most people have developed trust. This definition is remarkably soft. An efficient diagnostic prediction depends on a high prevalence of the disease. This is frequently forgotten when tests have a high sensitivity and specificity. The mathematical concept behind this observation is the Bayesian theorem. This is highly important for predictive pathology because it allows to combine attributes with high likelihood ratio simply by multiplication and has been shown to be remarkably stable, e. g. in the differential diagnosis of acute abdominal pain. Pathology should take the leadership in prediction since it has a considerable power as the gold standard of many tests. However, a network is advisable with other basic disciplines.