Serum procalcitonin as a diagnostic marker of neonatal sepsis

Abstract

Purpose: We evaluated serum procalcitonin (PCT) as a diagnostic marker of neonatal sepsis, and compared PCT levels with C-reactive protein (CRP) levels.

Methods: We retrospectively reviewed the medical records of 269 neonates with a suspected infection, admitted to Wonkwang University School of Medicine & Hospital between January 2011 and December 2012, for whom PCT and CRP values had been obtained. Neonates were categorized into 4 groups according to infection severity. CRP and PCT values were analyzed and compared, and their effectiveness as diagnostic markers was determined by using receiver operating characteristic (ROC) curve analysis. We also calculated the sensitivity, specificity, and positive, and negative predictive values.

Results: The mean PCT and CRP concentrations were respectively 56.27±81.89 and 71.14±37.17 mg/L in the "confirmed sepsis" group; 15.64±32.64 and 39.23±41.41 mg/L in the "suspected sepsis" group; 9.49±4.30 and 0.97±1.16 mg/L in the "mild infection" group; and 0.21±0.12 and 0.72±0.7 mg/L in the control group. High concentrations indicated greater severity of infection (P<0.001). Five of 18 patients with confirmed sepsis had low PCT levels (<1.0 mg/L) despite high CRP levels. In the ROC analysis, the area under the curve was 0.951 for CRP and 0.803 for PCT. The cutoff concentrations of 0.5 mg/L for PCT and 1.0 mg/L for CRP were optimal for diagnosing neonatal sepsis (sensitivity, 88.29% vs. 100%; specificity, 58.17% vs. 85.66%; positive predictive value, 13.2% vs. 33.3%; negative predictive value, 98.6% vs. 100%, respectively).

Conclusion: PCT is a highly effective early diagnostic marker of neonatal infection. However, it may not be as reliable as CRP.

Keywords: C-reactive protein; Procalcitonin; Sepsis.

Fig. 1

Comparison of the receiver operating characteristic (ROC) curves of procalcitonin (PCT) and C-reactive protein (CRP). The area under the curve was 0.803 for PCT and 0.951 for CRP. The difference between areas was significant (0.148; 95% confidence interval, 0.056-0.239; P=0.0015).