Beyond accuracy

Diagnostic tests are commonly evaluated using sensitivity, specificity, predictive values, likelihood ratios, and receiver operating characteristic (ROC) curve analysis. Although essential, these measures do not fully answer two practical questions frequently raised in clinical medicine: how often a test makes mistakes, and whether those mistakes are predominantly false positive or false negative. This article introduces two complementary measures that address these questions: the number needed to misdiagnose (NNM), which expresses the frequency of diagnostic error in intuitive clinical terms, and the misdiagnosis direction index (MDI), which indicates whether errors are mainly false positive or false negative. A weighted MDI is also introduced for unequal clinical harms of errors. Together with conventional indices such as Youden's index, these measures provide a broader framework for interpreting diagnostic performance.