Table 3 Sensitivity analysis of NEWS (M0, M1, M2) and NEWS2 models (M0’, M1’, M2’) for predicting the risk of COVID at threshold NEWS/NEWS2 ≥ 5 (predicted probability of model M0 = 0.130 and M0’ = 0.116 using development dataset) for validation dataset after correcting the calibration-in-the-large
Model | Number of positive cases identified by model | Sensitivity% | Specificity% | PPV | NPV | LR+ | LR- |
|---|---|---|---|---|---|---|---|
M0 | 474 | 32.9 (27.4 to 38.7) | 82.9 (81.3 to 84.5) | 19.2 (15.7 to 23) | 90.9 (89.6 to 92.1) | 1.9 (1.6 to 2.3) | 0.8 (0.7 to 0.9) |
M1 | 600 | 44 (38.1 to 50.1) | 78.7 (76.9 to 80.4) | 20.3 (17.2 to 23.8) | 91.9 (90.6 to 93.1) | 2.1 (1.8 to 2.4) | 0.7 (0.6 to 0.8) |
M2 | 607 | 51.6 (45.6 to 57.6) | 79.3 (77.6 to 81) | 23.6 (20.2 to 27.1) | 93 (91.8 to 94.1) | 2.5 (2.2 to 2.9) | 0.6 (0.5 to 0.7) |
M0` | 681 | 44.4 (38.5 to 50.5) | 75.1 (73.3 to 76.9) | 18.1 (15.2 to 21.2) | 91.6 (90.3 to 92.9) | 1.8 (1.5 to 2.1) | 0.7 (0.7 to 0.8) |
M1` | 781 | 52.7 (46.6 to 58.7) | 71.7 (69.8 to 73.5) | 18.7 (16 to 21.6) | 92.5 (91.1 to 93.7) | 1.9 (1.6 to 2.1) | 0.7 (0.6 to 0.7) |
M2` | 692 | 57 (51 to 62.9) | 76.2 (74.4 to 77.9) | 22.8 (19.8 to 26.1) | 93.5 (92.3 to 94.6) | 2.4 (2.1 to 2.7) | 0.6 (0.5 to 0.6) |