Table 3. Discriminative performances of conventional risk factors and addition of newer biomarkers for prediction of renal function recoverya after acute tubular necrosis.
| AUROC |
cfNRI (%) |
IDI |
|||||
|---|---|---|---|---|---|---|---|
| AUROC (95% CI) | ΔAUROC (95% CI) | P valueb | cfNRI (95% CI) | P valuec | IDI (95% CI) | P valued | |
| Conventional risk factors only | 0.82 (0.70–0.94) | — | Referent | — | Referent | — | Referent |
| +hemoglobin | 0.86 (0.77–0.96) | 0.05 (−0.02–0.11) | 0.18 | 63.2 (11.2–115.3) | 0.02 | 0.07 (−0.003–0.14) | 0.06 |
| +tubular PRX3 QISV | 0.89 (0.81 –0.98) | 0.08 (0.001–0.15) | 0.048 | 87.4 (38.3–136.6) | <0.001 | 0.13 (0.04–0.22) | 0.01 |
| +hemoglobin + tubular PRX3 QISV | 0.91 (0.84–0.99) | 0.10 (0.01–0.18) | 0.03 | 88.3 (39.1–137.5) | <0.001 | 0.19 (0.09–0.29) | 0.001 |
aRisk prediction was assessed by the AUROC, cfNRI and IDI. Each newer marker was stepwise added to the model of conventional risk factors (base model) to assess the AUROC, cfNRI and IDI for predicting recovery of renal function within 6 months. Conventional clinical risk factors included age, sex, hypertension, diabetes, severity of AKI, urinary protein-to-creatinine ratio and baseline estimated glomerular filtration rate.
b–dThe P value for an increase in AUROC, cfNRI and IDI in a model with conventional risk factors and new biomarkers, compared with conventional risk factors alone. AKI, acute kidney injury; AUROC, area under the ROC curve; cfNRI, category-free net reclassification improvement; CI, confidence interval; IDI, integrated discrimination improvement; PRX3, peroxiredoxin 3; QISV, quantitative immunohistochemical staining value; SE, standard error.