Table 4.

Accuracy of urine neutrophil gelatinase-associated lipocalin normalized concentration for the prediction of dialysis or in-hospital death

VariableBefore CPBBefore CPBWithin 1 h of CPBWithin 1–3 h of CPBWithin 3–6 h of CPBWithin 6–12 h of CPBWithin 12–18 h of CPBWithin 18–24 h of CPB
Urine creatinine–normalized NGAL concentration cutoff (μg/mmol creatinine)12.521.2186.5126.09.57.67.17.2
Sensitivity (95% CI)a0.64 (0.53 to 1.00)0.81 (0.58 to 1.00)0.55 (0.47 to 1.00)0.61 (0.43 to 0.93)0.81 (0.56 to 1.00)0.87 (0.62 to 1.00)0.92 (0.83 to 1.00)0.57 (0.32 to 1.00)
Specificity (95% CI)a0.76 (0.32 to 0.84)0.68 (0.40 to 0.94)0.85 (0.31 to 0.96)0.85 (0.56 to 0.99)0.68 (0.53 to 0.92)0.61 (0.47 to 0.94)0.58 (0.38 to 0.82)0.51 (0.21 to 0.99)
Positive likelihood ratio (95% CI)a2.73 (1.33 to 4.52)2.56 (1.45 to 8.52)3.61 (1.39 to 14.17)4.26 (1.72 to 44.62)2.54 (1.90 to 8.24)2.25 (1.76 to 11.59)2.18 (1.63 to 5.28)1.16 (1.17 to 23.67)
Negative likelihood ratio (95% CI)a0.47 (0.00 to 0.68)0.27 (0.00 to 0.59)0.52 (0.00 to 0.66)0.45 (0.00 to 0.68)0.27 (0.00 to 0.55)0.20 (0.00 to 0.53)0.14 (0.00 to 0.33)0.84 (0.00 to 0.85)
Diagnostic odds ratio (95% CI)a5.85 (2.17 to ∞)9.57 (3.32 - ∞)6.87 (2.04 - ∞)9.48 (4.01 - ∞)9.24 (4.19 to ∞)11.03 (4.93 to ∞)15.17 (7.94 to ∞)1.37 (0.47 to ∞)
Net reclassification index (95% CI)0.81 (0.26 to 1.36)0.99 (0.38 to 1.61)0.80 (0.13 to 1.47)0.94 (0.36 to 1.52)0.97 (0.27 to 1.29)0.97 (0.46 to 1.49)0.99 (0.38 to 1.61)0.53 (-0.05 to 1.10)
 P value0.0040.0010.020.001<0.0010.0030.0010.07
Integrated discrimination improvement (95% CI)0.07 (0.02 to 0.13)0.07 (0.03 to 0.10)0.03 (-0.006 to 0.07)0.08 (0.02 to 1.51)0.06 (0.03 to 0.08)0.06 (0.02 to 0.09)0.08 (0.03 to 0.12)0.04 (-0.001 to 0.08)
 P value0.01<0.0010.100.006<0.001<0.001<0.0010.05
Number needed to screen (95% CI)a1.97 (1.55 to 2.75)2.63 (1.95 to 4.13)1.51 (1.30 to 1.83)1.53 (1.28 to 1.96)2.60 (1.95 to 4.02)3.65 (2.41 to 8.14)3.86 (2.11 to 20.00)−27.00 (−77 to ∞)
  • NGAL, neutrophil gelatinase-associated lipocalin.

  • a 95% CIs were estimated by boot-strapping with 1000 resamples.