highly significantp < 0.0001
The ROC analysis (sensitivity and specificity; Table 3 ) revealed that a highly significant ( p < 0.0001) difference between two medians with a large effect size does not necessarily represent a good classifier, even if the effect size is proportional to the AUC.