Table 3: Diagnostic performances of non invasive means for the prediction oesophageal varices and large oesophageal varices in the training sample.

AUROC SE Cut-off TP (n) TN (n) SS (%) SP (%) PPV (%) NVP (%) DA (%) LR+ Oesophageal varices  Platelet count <110500 68 18 80 69 89.5 51 77.5 2.6  Spleen diameter >140 44 18 51.8 72 86.3 30.5 55.9 1.9  PC/SD ratio ≤868 70 20 82.4 76.9 92.1 57.1 81.1 3.6  Regression function 0.3 82 7 96.5 28 82 70 80.9 1.3 0.4 81 7 95.3 28 81.8 63.6 80 1.3 0.5 80 10 94.1 40 84.2 66.7 81.8 1.6 0.6 79 17 92.9 68 90.8 73.9 87.3 2.9 0.7 69 18 81.2 72 90.8 52.9 79.1 2.9 0.8 62 18 72.9 72 89.9 43.9 72.7 2.6 Large oesophageal varices  Platelet count <106500 61 20 78.2 60.6 82.4 54.1 73 2  Spleen diameter >137 49 25 62.8 46.3 86 46 66.7 1.2  PC/SD ratio ≤897 66 21 84.6 63.6 84.6 61.8 78.4 2.3 0.3 75 9 96.2 28.1 76.5 75 76.4 1.3 Regression function 0.4 73 11 93.6 34.4 77.7 68.8 76.4 1.4 0.5 68 15 87.2 46.9 80 60 75.5 1.6 0.6 64 20 82.1 62.5 84.2 58.8 76.4 2.2 0.7 59 25 75.6 78.1 89.4 56.8 76.4 3.5 0.8 51 27 65.4 84.4 91.1 50 71 4.2

AUROC: area under receiver operating characteristic curve, SE: standard error, PC/SD ratio: platelet count-spleen diameter ratio, TP: true positive TN: true negative SS: sensitivity SP: specificity PPV: positive predictive value NVP: negative predictive value DA: diagnostic accuracy, LR+: positive likelihood ratio, Prevalence of OV: 76.6%, logistic regression analysis was computed on 110 patients due to one missing value in those with no OV.