Cerebrospinal Fluid and Plasma Procalcitonin for the Diagnosis of Neonatal Bacterial Meningitis

Objective To determine accuracy of cerebro-spinal fluid (CSF) procalcitonin (PCT) to diagnose neonatal bacterial meningitis (NBM) among septic neonates and compare with other index tests. Design Prospective, cross-sectional, single-gate study Setting Level-3 neonatal unit Patients Neonates with suspected sepsis undergoing lumbar puncture Index tests CSF PCT, leukocyte count and biochemistry; plasma PCT and CSF:plasma PCT ratio Reference standards “Definite meningitis” defined by positive CSF culture and/or gram stain and/or broad-based primer 16S rDNA polymerase chain reaction. “Definite or probable meningitis” defined by definite meningitis or probable meningitis (based on cytochemistry cut-offs). Results Of 216 eligible neonates, 18 had “definite meningitis” and 37 “definite or probable meningitis”. Median (Q 1 , Q 3 ) CSF PCT level was higher in “definite meningitis” compared to “no definite meningitis” [0.429 (0.123, 1.300) vs. 0.181 (0.119, 0.286) ng/ml respectively, p=0.028]. Likewise, it was higher in “definite or probable meningitis” compared to no meningitis [0.245 (0.136, 0.675) vs. 0.170 (0.116, 0.28), p=0.01]. The area under ROC curve (AUC) of CSF PCT level for definite meningitis was 0.656 and for “definite or probable meningitis” 0.635. Paired comparisons of AUC of CSF PCT with other index tests were not significant. Based on a priori cut-off of 0.2 ng/ml, CSF PCT level had a sensitivity (95% CI) of 67% (50, 80), specificity 58% (54, 61), LR + 1.6 (1.1, 2.0) and LR - 0.6 (0.3, 0.9). Conclusions Higher values of CSF PCT are associated with NBM. CSF PCT cannot reliably discriminate between meningitis and no meningitis and is not superior to other CSF tests.