ABSTRACT
AIM: Meningitis may complicate neonatal sepsis, but there is scant evidence to inform the decision to perform a lumbar puncture (LP) and considerable variation in practice. We investigated whether inflammatory markers - C-reactive protein (CRP) and immature-to-total neutrophil ratio (ITR) - were predictive of meningitis or significant cerebrospinal fluid (CSF) pleocytosis and useful in guiding the decision to perform a LP. METHODS: We studied all inpatients in a single tertiary neonatal unit who were <6 months of age who had a LP performed between March 2011 and October 2014. We categorised CSF results as follows: (i) culture-positive meningitis; (ii) probable culture-negative meningitis but meeting a priori criteria for significant CSF leucocytosis; or (iii) no evidence of meningitis. CRP and ITR obtained within 48 h of LP were analysed. We assessed the test performance of CRP and ITR by area under receiver operating characteristic curves. RESULTS: A total of 757 (male 471, 62.2%) infants were included. The median (interquartile range) gestational age was 38.4 weeks (30-40.3), and birthweight was 2940 g (1330-3560). Ten (1.3%) infants had culture-positive meningitis; 71 (9.4%) were classified as probable culture-negative meningitis and 676 (89.3%) as non-meningitis. The area under receiver operating characteristic curve for culture-positive and probable culture-negative meningitis was 0.43 for CRP (95% confidence interval 0.36-0.51) and 0.58 for ITR (0.51-0.65). At a CRP threshold of 30 mg/L, there was a positive likelihood ratio (LR) of 0.77 and a negative LR of 1.44. CONCLUSIONS: CRP and ITR perform poorly in identifying infants with confirmed or probable meningitis. The decision to perform an LP should be more focused on clinical grounds and/or a positive blood culture and less on inflammatory or haematological markers in isolation.