Table 7.
Diagnostic ability of SLPI to predict AKI in the subgroup of patients with serum creatinine at baseline ≤1.25 mg/dL.
| Time of measurement | Optimal Cut-Off (Youden index) | AUC | ||||
|---|---|---|---|---|---|---|
| Cut-Off, ng/ml | Sensitivity [%] | Specificity [%] | LR+ | LR− | ||
| Baseline | ≥ 49.84 | 69.23 [38.6, 90.9] | 55.56 [21.2, 86.3] | 1.56 | 0.55 | 0.496 [0.235, 0.756] |
| Admission to ICU | ≥ 49.67 | 50.00 [21.1, 78.9] | 83.33 [35.9, 99.6] | 3.00 | 0.60 | 0.569 [0.253, 0.886] |
| 12 h after ICU | ≥ 57.02 | 76.92 [46.2, 95.0] | 100 [66.4, 100] | — | 0.23 | 0.932 [0.832, 1] |
| 24 h after ICU | ≥ 43.20 | 91.67 [61.5, 99.8] | 77.78 [40.0, 97.2] | 4.13 | 0.11 | 0.898 [0.763, 1] |
| 48 h after ICU | ≥ 71.24 | 81.82 [48.2, 97.7] | 88.89 [51.8, 99.7] | 7.36 | 0.20 | 0.798 [0.560, 1] |
| 72 h after ICU | ≥ 68.81 | 45.45 [16.7, 76.6] | 100 [63.1, 100] | — | 0.55 | 0.716 [0.477, 0.955] |
ROC analysis was performed to evaluate the diagnostic ability of perioperative SLPI levels during the first 72 h on ICU with regard to AKI. If an elevated SLPI value indicates that the patient is likely to develop an AKI, the ROC curve should be farther from the bisecting line (Sensitivity = 1-Specificity). Sensitivity, specificity and likelihood ratios (LR+/−), are reported for the Youden optimal cut-off. 95%-confidence intervals are shown in parentheses.