Towards a more standardized assessment of diagnostic tumor-markers

The estimation of sensitivity and specificity are mainly used as summary measures of diagnostic power of new diagnostic tests. Unfortunately, the positive assessment outcome is often recanted, frequently because only a preliminary assessment is done in a highly selected population of already diagnosed patients and healthy individuals. Furthermore, diagnostic marker outcomes are dichotomised for the calculation of sensitivity and specificity, with loss of information as a consequence. We plead for the presentation of all relevant diagnostic information, which includes extensive description of the composition of the study group, distributions of test results for different patient groups and the use of the ROC curve and the area under it as the statistical summary measure.