ROC curves: general characteristics and their usefulness in clinical practice

Abstract

The use of diagnostic tests to determine the presence or absence of a disease is essential in clinical practice. The results of a diagnostic test may correspond to numerical estimates that require quantitative reference parameters to be transferred to a dichotomous interpretation as normal or abnormal and thus implement actions for the care of a condition or disease. For example, in the diagnosis of anemia it is necessary to define a cut-off point for the hemoglobin variable and create two categories that distinguish the presence or absence of anemia. The method used for this process is the preparation of diagnostic performance curves, better known by their acronym in English as ROC (Receiver Operating Characteristic). The ROC curve is also useful as a prognostic marker, since it allows defining the cut-off point of a quantitative variable that is associated with greater mortality or risk of complications. They have been used in different prognostic markers in COVID-19, such as the neutrophil/lymphocyte ratio and D-dimer, in which cut-off points associated with mortality and/or risk of mechanical ventilation were identified. The ROC curve is used to evaluate the diagnostic performance of a test in isolation, but it can also be used to compare the performance of two or more diagnostic tests and define which one is more accurate. This article describes the basic concepts for the use and interpretation of the ROC curve, the interpretation of an area under the curve (AUC) and the comparison of two or more diagnostic tests.