Skip to main content
. 2020 Jun 1;21:207. doi: 10.1186/s12882-020-01872-z

Table 3.

Calibration and discrimination of the AKI biomarkers in predicting AKD and 5-year mortality

Calibration Discrimination
Goodness –of-fit (χ2) df p-value AUROC±SE 95%CI p-value
For AKD prediction
 Serum IL-18 10.518 8 0.231 0.698 ± 0.033 0.634–0.762 < 0.001
 Serum NGAL 5.756 8 0.675 0.630 ± 0.034 0.563–0.697 < 0.001
 Urine NGAL 12.980 8 0.113 0.660 ± 0.036 0.591–0.730 < 0.001
 Serum CysC 9.448 8 0.306 0.663 ± 0.033 0.598–0.728 < 0.001
For 5-year mortality prediction
 Serum IL-18 0.712 8 0.999 0.522 ± 0.048 0.429–0.616 0.635
 Serum NGAL 9.717 8 0.285 0.573 ± 0.047 0.482–0.665 0.113
 Urine NGAL 19.266 8 0.014 0.712 ± 0.047 0.620–0.803 < 0.001
 Serum CysC 7.780 8 0.455 0.554 ± 0.048 0.460–0.648 0.241

No statistically significant AUROC differences between serum IL-18, serum NGAL, urine NGAL and serum CysC for AKD prediction*.

Statistically significant AUROC differences between urine NGAL vs. serum IL-18, serum NGAL and serum CysC for 5-year mortality prediction, all p-values < 0.01*.

AKI Acute kidney injury, AKD Acute kidney disease, IL-18 Interleukin 18, NGAL Neutrophil gelatinase-associated lipocalin, CysC Cystatin C, AUROC Area under the receiver operating characteristic curve

*Pairwise comparisons of ROC curves, by Delong method, 1988