Receiver operating characteristic (ROC) curves to summarize the performance of detection of EBV DNA in plasma to discriminate cases of BL from controls. (A) All cases of BL and controls, N = 800, and (B) cases of BL and controls, excluding those lacking detectible RPP30 and EBV, N = 689. These ROC curves plot the false positive rate (1 – specificity) against the true positive rate (sensitivity) for the detection of EBV in participants with BL and controls, separately for all 800 participants with BL and controls (A), and for 698 participants with BL and controls that had either EBV or RPP30 detected. The shaded purple region represents the 95% CIs based on the bootstrap method with 1000 replicates. The gray diagonal line is the line of no discrimination (AUC = 50%). Because most controls (85%) had no detectable EBV, there is a vertical incline at a false positive rate of 0 because controls without detectable EBV are correctly classified. There is a plateau at 77.3% (panel A) and 88.3% (panel B) because that is the proportion of cases of BL with detectable EBV.