Table 3.
Performance for early prediction of severe Covid-19.
| Development cohort (n = 421) | Validation cohort (n = 140) | |||
|---|---|---|---|---|
| AUC (95% CI) | ||||
| Fitted value | 0.886 (0.836–0.937) | 0.826 (0.730–0.922) | ||
| KDDH Score | 0.884 (0.833–0.934) | 0.828 (0.733–0.923) | ||
| Cut-off Points 9 (stable/progression) | ||||
| High-risk (Points ≥ 9) | 31 TP (27.7%) | 81 FP (72.3%) | 17 TP (41.5%) | 24 FP (58.5%) |
| Low-risk (Points < 9) | 6 FN (1.9%) | 303 TN (98.1%) | 7 FN (7.1%) | 92 TN (92.9%) |
| Sensitivity (%) | 83.8% | 70.8% | ||
| Specificity (%) | 78.9% | 79.3% | ||
| Positive predictive value (%) | 27.7% | 41.5% | ||
| Negative predictive value (%) | 98.1% | 92.9% | ||
| Positive likelihood ratio | 3.972 | 3.424 | ||
| Negative likelihood ratio | 0.206 | 0.368 | ||
FN, false negative; FP, false positive; TN, true negative; TP, true positive.