Discrepant analysis: a biased and an unscientific method for estimating test sensitivity and specificity

Discrepant analysis is a widely used technique for estimating test performance indices (sensitivity, specificity, etc.) of DNA-amplification tests for detecting infectious diseases. It has recently been claimed that the discrepant analysis-based estimates of specificity are typically less biased than those based on culture and that the discrepant analysis-based specificity shows little appreciable bias. In this article, I show that those conclusions are incorrect. Using a typical example from the published literature, I show that the discrepant analysis-based estimates of sensitivity and specificity can generate a significant and clinically important overestimation of the true sensitivity and specificity values. Moreover, I demonstrate that the concept of discrepant analysis is profoundly flawed and unscientific. It violates a fundamental principle of diagnostic testing-the principle that the new test should not be used to determine the true disease status. Thus, the major problem with discrepant analysis is not only that it is biased but that it is unscientific. Therefore, discrepant analysis should not be adopted for the evaluation of any diagnostic or screening test.