[Diagnostic results and their clinical interpretation--or: what you never wanted to know about diagnosis]

Diagnostic test results become only relevant if clinically interpreted. A key issue in clinical interpretation is the predictive value which, however, in daily practice and even in medical literature, is frequently misunderstood, as if the predictive value would be about the same as sensitivity. Sensitivity and specificity of a test only together with the prevalence of the target disease allow estimation of the predictive values. This correlation will be exemplified with the diagnosis of chorioamnionitis measuring C-reactive protein: The predictive value is not constant, but depends on prevalence. Prevalence, however, changes with clinical situation, therefore similar test results must be clinically interpreted very differently. Finally, considering these correlations, decision analysis using the "threshold model" shows how to develop a rational, quantitative approach to the selection and interpretation of diagnostic tests. Although more and more studies on decision analysis in diagnostics are published, up to now, there is little response in day-to-day practice. This situation should be changed for the sake of our patients.