Comparative diagnostic test evaluation of serum procalcitonin and C-reactive protein in suspected bloodstream infections in children with cancer

Purpose. To compute diagnostic test properties of C-reactive protein (CRP) and serum procalcitonin (PCT) levels in bloodstream infections in children with cancer and suspected sepsis, in comparison with blood culture as the gold standard. Methodology. Consecutive paediatric cancer patients, aged <= 14 years, with clinically suspected bloodstream infections were evaluated with blood culture and assay of PCT and CRP levels. Blood culture was taken as the gold standard for comparison. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratio (LR) and receiver operating characteristic (ROC) with area under ROC curve (AUC) were calculated to assess the diagnostic test performance for PCT and CRP. Results/Key findings. The ROC curve for PCT was better than that for CRP, with an AUC of 0.751 for PCT at a cut-off of 2.25 ng ml(-1). The AUC for CRP was 0.638 at a cut-off of 8.0 mg dl(-1). Among the three cut-off values of PCT selected from the ROC curve applicable to the patients under study, the cut-off value of >= 0.49 ng ml(-1) had the maximum sensitivity of 81.4% and an NPV of 94.67 %; >= 2.25 ng ml(-1) had a sensitivity and specificity of 65.12 and 71.6 %, respectively, and >= 6.47 ng ml(-1) had a maximum specificity of 82.10 %. For CRP, the cut-off value of >= 5.3 mg dl(-1) had the maximum sensitivity of 72.09 %; >= 8.0 mg dl(-1) had a sensitivity and specificity of 58.14 and 68.09 %, respectively, and >= 8.4mg dl(-1) had the maximum specificity of 70.04 %. Conclusion. PCT is a better serological marker for excluding bloodstream infections than CRP. The cut-off value of 0.49 ng ml(-1) with a negative predictive value of 94.67% will be ideal in a clinical setting of immune-compromised children with suspected sepsis.