Objective This work was done to assess the clinical usefulness of blood procalcitonin levels for the diagnosis of nosocomial infections in the neonate.
Methods The enrolment criterion was clinically manifested nosocomial infection. Procalcitonin levels were measured in venous blood from 52 infected neonates (A; mean gestational age 29.7 ± 3.7 weeks; mean birth weight 1264 ± 574 g) and 88 uninfected neonates (B; mean gestational age 30.0 ± 5.1 weeks; mean birth weight 1502 ± 950 g). The results were interpreted against two tests routinely used for the diagnosis of infection, namely C-reactive protein (CRP) and white blood cell count (WBC).
Results Differences between both groups for procalcitonin (median 4.3 vs 0.94 ng/ml) and CRP levels (median 20.3 vs 2.7 mg/l) were highly significant (p = 0.000001 for procalcitonin and p = 0.000005 for CRP, respectively). No significant differences between the groups were noted for WBC. The threshold value on the receiver operator characteristic curve was 2.06 ng/ml for procalcitonin (sensitivity 75% SE 0.07; specificity 80.68% SE 0.06; positive predictive value (PPV) 62.22% SE 0.07; negative predictive value (NPV) 88.75% SE 0.04, area under the curve (AUC) 0.805), 5.0 mg/l for CRP (sensitivity 67.44% SE 0.07; specificity 73.68% SE 0.07; PPV 42.02% SE 0.06; NPV 88.89% SE 0.03, AUC 0.801) and 11.9 g/l for WBC (sensitivity 51.16% SE 0.08; specificity 50.68% SE 0.08; PPV 23.16% SE 0.04; NPV 78.13% SE 0.04, AUC 0.484).
Conclusion Procalcitonin levels in neonatal blood appear to be of use for the diagnosis of nosocomial infections as this parameter demonstrates greater sensitivity and specificity than CRP. WBC appears of little diagnostic value in the early phase of the illness.