Understanding diagnostic tests 3: receiver operating characteristic curves

Abstract

The results of many clinical tests are quantitative and are provided on a continuous scale. To help decide the presence or absence of disease, a cut-off point for 'normal' or 'abnormal' is chosen. The sensitivity and specificity of a test vary according to the level that is chosen as the cut-off point. The receiver operating characteristic (ROC) curve, a graphical technique for describing and comparing the accuracy of diagnostic tests, is obtained by plotting the sensitivity of a test on the y axis against 1-specificity on the x axis. Two methods commonly used to establish the optimal cut-off point include the point on the ROC curve closest to (0, 1) and the Youden index. The area under the ROC curve provides a measure of the overall performance of a diagnostic test. In this paper, the author explains how the ROC curve can be used to select optimal cut-off points for a test result, to assess the diagnostic accuracy of a test, and to compare the usefulness of tests.