Validation of a Diagnosis Model for Differentiating Bacterial from Viral Meningitis in Infants and Children under 3.5 Years of Age

Abstract

 The aim of this study was to validate, in a population of infants and children under 3.5 years of age, a diagnosis model that provides a figure for the probability of bacterial meningitis (pABM), based on four parameters collected at the time of the first lumbar tap: the cerebrospinal fluid (CSF) protein level, CSF polymorphonuclear cell count, blood glucose level, and leucocyte count. The best cut-off value for distinguishing between bacterial and viral meningitis was previously found to be 0.1, since 99% of meningitides associated with pABM<0.1 were viral. The charts of 103 consecutive children aged 0.1–3.5 years who had been hospitalised for acute meningitis were reviewed. Each case was sorted into the following three categories for aetiology: bacterial (positive CSF culture, n=48); viral (negative CSF culture and no other aetiology, and no antibiotic treatment after diagnosis, n=36); and undetermined (fitting neither of the first two definitions, n=19). After computation of pABM values in each case, the predictive values of the model were calculated for different pABM cut-off values. The results confirmed that the best cut-off pABM value was 0.1, for which the positive and negative predictive values in this model were 96% and 97%, respectively. Only one case of bacterial meningitis (lumbar tap performed early in an infant with meningococcal purpura fulminans with negative CSF culture) was associated with a pABM value of <0.1. This model is quite reliable for differentiating between bacterial and viral meningitis in children under 3.5 years of age, and it may enable physicians to withhold antibiotics in cases of meningitis of uncertain aetiology.