Objectives To determine whether there is a cut off value of serum C-reactive protein (CRP) associated with a higher risk of meningitis in suspected early onset sepsis (EOS) (onset birth to 7 days of life). Methods A retrospective cohort study on neonates admitted in neonatal intensive care unit at McMaster Children's Hospital from January 2010 to 2017 and had lumbar puncture (LP) and CRP for workup of EOS. Included subjects had either (a) non-traumatic LP or (b) traumatic LP with cerebral spinal fluid (CSF) polymerase chain reaction or gram stain or culture-positive or had received antimicrobials for 21 days. Excluded were CSF done for metabolic errors, before cytomegalovirus (CMV) treatment; from ventriculo-peritoneal (VP) shunts; missing data and contamination. Neonates were classified into definite and probable meningitis and on the range of CRP. We calculated sensitivity, specificity, and likelihood ratios for CRP values; and area under the receiver operating characteristic (AUROC) curve. Results Out of 609 CSF samples, 184 were eligible (28 cases of definite or probable meningitis and 156 controls). Sensitivity, specificity, predictive values, likelihood ratios, and AUROC were too low to be of clinical significance to predict meningitis in EOS. Conclusions Serum CRP values have poor discriminatory power to distinguish between subjects with and without meningitis, in symptomatic EOS.
Keywords: C-reactive protein; antibiotics; meningitis; newborn; sepsis.