Table 4.
Sensitivity, specificity and predictive values for the CARM model at various cut-offs in the development dataset and validation dataset
| Dataset | Risk value cut-off | No. of patients >cut-off | % Sensitivity (95% CI) | % Specificity (95% CI) | % PPV (95% CI) | % NPV (95% CI) | LR+ (95% CI) | LR− (95% CI) |
| Development dataset | 0.01 | 19 876 | 98.5 (97.9 to 99) | 38 (37.4 to 38.5) | 8.8 (8.4 to 9.2) | 99.8 (99.7 to 99.8) | 1.6 (1.6 to 1.6) | 0 (0 to 0.1) |
| 0.02 | 15 297 | 95.7 (94.6 to 96.6) | 53.4 (52.9 to 54) | 11 (10.6 to 11.6) | 99.5 (99.4 to 99.6) | 2.1 (2 to 2.1) | 0.1 (0.1 to 0.1) | |
| 0.04 | 10 382 | 87.3 (85.6 to 88.8) | 69.8 (69.2 to 70.3) | 14.8 (14.2 to 15.5) | 98.9 (98.8 to 99) | 2.9 (2.8 to 3) | 0.2 (0.2 to 0.2) | |
| 0.08 | 6070 | 72.2 (70 to 74.3) | 83.6 (83.2 to 84) | 21 (20 to 22.1) | 98 (97.8 to 98.2) | 4.4 (4.2 to 4.6) | 0.3 (0.3 to 0.4) | |
| 0.20 | 2190 | 42 (39.6 to 44.3) | 95 (94.8 to 95.3) | 33.8 (31.9 to 35.9) | 96.4 (96.2 to 96.7) | 8.5 (7.9 to 9.1) | 0.6 (0.6 to 0.6) | |
| Validation dataset | 0.01 | 18 338 | 98.4 (97.7 to 99) | 32.1 (31.5 to 32.7) | 9.1 (8.7 to 9.6) | 99.7 (99.5 to 99.8) | 1.4 (1.4 to 1.5) | 0 (0 to 0.1) |
| 0.02 | 14 537 | 95.9 (94.9 to 96.8) | 47.4 (46.8 to 48.1) | 11.2 (10.7 to 11.8) | 99.4 (99.3 to 99.5) | 1.8 (1.8 to 1.9) | 0.1 (0.1 to 0.1) | |
| 0.04 | 10 047 | 89 (87.4 to 90.4) | 65.2 (64.6 to 65.8) | 15.1 (14.4 to 15.8) | 98.8 (98.7 to 99) | 2.6 (2.5 to 2.6) | 0.2 (0.1 to 0.2) | |
| 0.08 | 5871 | 73.2 (71 to 75.3) | 81.2 (80.7 to 81.6) | 21.2 (20.2 to 22.3) | 97.8 (97.5 to 98) | 3.9 (3.7 to 4) | 0.3 (0.3 to 0.4) | |
| 0.20 | 2158 | 43.1 (40.7 to 45.5) | 94.2 (93.9 to 94.5) | 34 (32 to 36.1) | 96 (95.7 to 96.2) | 7.4 (6.9 to 8) | 0.6 (0.6 to 0.6) |
CARM, computer-aided risk of mortality; LR+, positive likelihood ratio; LR−, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value.