Interpreting diagnostic tests with continuous results and no gold standard

Abstract

Practitioners of evidence-based medicine commonly encounter diagnostic tests with continuous results and no gold standard. In contrast, the traditional critical appraisal teachings assume a binary test (2×2 table) with a gold standard. In this guide, we use the example of the tuberculin skin test to illustrate a simple approach facilitated by using stratum-specific likelihood ratios and odds of developing future patient-important events. This approach can aid practitioners in the interpretation and application of diagnostic tests to patient care.