Table 4.
Prediction model | Incremental value of DNI added to predictive model composed of traditional risk factors | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
AUROC (95% CI) | Difference of AUROC (95% CI) | P | NRI (95% CI) | P | IDI (95% CI) | P | ||||
(A) | ||||||||||
Reference (Ref.) model | 0.814 (0.750–0.879) | |||||||||
Ref. model + DNIa | 0.819 (0.754–0.884) | 0.005 (−0.022–0.031) | 0.731 | 0.127 (0–0.254) | 0.049* | 0.027 (0.004–0.05) | 0.023* | |||
(B) | ||||||||||
Prediction model | Comparison between predictive model composed of traditional risk factors and traditional risk factors combined with DNI | |||||||||
Sensitivity (95% CI) | P | Specificity (95% CI) | P | Accuracy (95% CI) | P | NPV (95% CI) | P | PPV (95% CI) | P | |
Ref. model | 72.1 (60.9–83.4) | Ref. | 81.4 (76.5–86.3) | Ref. | 79.5 (75–84.1) | Ref. | 49.4 (39.1–59.8) | Ref. | 92.1 (88.4–95.7) | Ref. |
Ref. model + DNIa | 63.9 (51.9–76) | 0.020 | 90.1 (86.3–93.8) | <0.001 | 84.8 (80.8–88.9) | 0.001 | 61.9 (49.9–73.9) | <0.001 | 90.8 (87.2–94.5) | 0.170 |
*P < 0.05, aVariable was logarithmically transformed before analyses. Abbreviations: AUROC, area under the receiver operating characteristic; CI, confidence interval; DNI, delta neutrophil index; IDI, integrated discrimination improvement; NRI, net reclassification index; PPV, positive predictive value; NPV, negative predictive value; Reclassification tables comparing two prediction models (See Supplement 4). Reference model = qSOFA score + Heart rate + Body temperature + Platelet count + Neutrophil ratio + Prothrombin ratio + Serum albumin level + Serum alanine aminotransferase level + Total bilirubin + C-reactive protein.