Skip to main content

Table 3 Detective characteristics of the three biomarkers and their combinations for total acute kidney injury and severe acute kidney injury

From: Evaluation of clinically available renal biomarkers in critically ill adults: a prospective multicenter observational study

Logistic regression model

AUC-ROCa

Cutoffb

Sensitivity

Specificity

(+) LR

(−) LR

PPV

NPV

Total AKI (n = 326)

 Univariate models

  sCysC

0.738 (0.703–0.772)

1.26 mg/L

0.44

0.95

7.92

0.59

0.77

0.80

  uNAG

0.650 (0.614–0.686)

27.14 U/g Cre

0.64

0.60

1.59

0.60

0.41

0.80

  uACR

0.683 (0.648–0.718)

61.14 mg/g Cre

0.54

0.76

2.20

0.61

0.49

0.79

 Multivariate models

  sCysC + uNAG

0.756 (0.723–0.789)c

0.43d

0.49

0.91

5.69

0.56

0.71

0.81

  uNAG + uACR

0.661 (0.626–0.697)e

0.27d

0.64

0.62

1.68

0.58

0.42

0.80

  sCysC + uACR

0.740 (0.706–0.774)f

0.45d

0.45

0.94

7.66

0.59

0.77

0.80

Severe AKI (n = 102)

 Univariate models

  sCysC

0.839 (0.798–0.880)

1.25 mg/L

0.67

0.87

5.28

0.38

0.35

0.96

  uNAG

0.706 (0.651–0.761)

32.80 U/g Cre

0.72

0.65

2.03

0.44

0.17

0.96

  uACR

0.771 (0.726–0.817)

71.97 mg/g Cre

0.72

0.74

2.77

0.38

0.22

0.96

 Multivariate models

  sCysC + uNAG

0.863 (0.827–0.900)c

0.09d

0.76

0.83

4.39

0.28

0.31

0.97

  uNAG + uACR

0.715 (0.661–0.768)g

0.08d

0.74

0.64

2.06

0.41

0.18

0.96

  sCysC + uACR

0.838 (0.797–0.879)f

0.08d

0.78

0.75

3.16

0.29

0.25

0.97

  1. Abbreviations: (+) LR Positive likelihood ratio, (−) LR negative likelihood ratio, PPV Positive predictive value, NPV Negative predictive value, sCysC Serum cystatin C, uNAG Urinary N-acetyl-β-d-glucosaminidase, Cre Creatinine concentration, uACR Urinary albumin/creatinine ratio
  2. a Values are presented as AUC-ROC (95% CI)
  3. b Ideal cutoff value according to Youden’s index
  4. c P < 0.05 vs. sCysC, uNAG, uACR, uNAG + uACR, and sCysC + uACR
  5. d Cutoff points of the biomarker panels were the predicted probabilities generated from the multiple logistic regression model
  6. e P < 0.05 vs. sCysC, uNAG, sCysC + uACR, and sCysC + uNAG
  7. f P < 0.05 vs. uNAG, uACR, uNAG + uACR, and sCysC + uNAG
  8. g P < 0.05 vs. sCysC, uNAG, uACR, sCysC + uACR, and sCysC + uNAG