Construction of receiver operating characteristic curves when disease verification is subject to selection bias

Abstract

The estimates of ROC curves, which are frequently used in the assessment of diagnostic tests, may be biased if the sample is restricted to subjects whose disease status has been definitively verified. A method to provide an unbiased estimate of the ROC curve under these sample conditions is proposed. The new method requires information on the probability distribution of test results in the population from which the verified sample is drawn. It is illustrated using data from a study of computed tomography for fever of uncertain origin.