Figure 1

Receiver operating characteristic curves of C-reactive protein (CRP) and procalcitonin (PCT) for predicting infection. CRP is shown as solid lines, PCT as broken lines. (a) Receiver operating characteristic (ROC) curves for predicting infection in all patients. CRP: area under the curve (AUC) 0.819 (95% CI 0.782, 0.856), best cutoff value 3.08 mg/dL; sensitivity 81%, specificity 71%; positive predictive value (PPV) 0.75, negative predictive value (NPV) 0.77. PCT: AUC, and best cutoff value 0.831 (95% CI 0.795, 0.866) and 1.1 ng/mL, respectively; sensitivity and specificity 64% and 90%, respectively; PPV and NPV 0.88 and 0.69, respectively. (b) ROC curves for prediction of infection in patients with normal renal function (group I). CRP: AUC, and best cutoff value 0.684 (95% CI 0.587, 0.782) and 2.49 mg/dL, respectively; sensitivity and specificity 82% and 51%, respectively; PPV and NPV 0.61 and 0.75, respectively. PCT: AUC, and best cutoff value 0.766 (95% CI 0.681, 0.851) and 0.08 ng/mL, respectively; sensitivity and specificity 82% and 60%, respectively; PPV and NPV 0.66 and 0.78, respectively. (c) ROC curves for prediction of infection in patients with impaired renal function. CRP: AUC, and best cutoff value 0.876 (95% CI 0.839, 0.912) and 3.08 mg/dL, respectively; sensitivity and specificity 82% and 79%, respectively; PPV and NPV 0.83 and 0.79, respectively. PCT: AUC, and best cutoff value 0.876 (95% CI 0.839, 0.912) and 1.1 ng/mL, respectively; sensitivity and specificity 73% and 89%, respectively; PPV and NPV 0.89 and 0.74, respectively.