[Differential diagnosis of bacterial and viral meningitis in childhood acute meningitis: a statistical model].

Acute bacterial meningitis (BM) which is a pediatric emergency with high mortality and morbidity, must be diagnosed and treated promptly. There is no unique method to prove or rule out the diagnosis of BM in a patient with cerebrospinal fluid (CSF) findings consistent with BM but negative Gram stain and culture results. For this purpose the combination of CSF parameters are used for diagnosis. The aims of this study were to compare retrospectively the mean leukocyte count, serum C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), CSF leukocyte and neutrophil, CSF protein and glucose values in 40 bacterial and 29 viral meningitis (VM patients, ages between 1 month and 14 years, and to develop a statistical model for the differentiation of BM and VM cases. Logistic regression analysis was used to investigate the relationship between BM and age, CPR, ESR, leukocyte count, CSF leukocyte, neutrophil, protein and glucose values. Based on CSF protein and neutrophil ratio which were found as independent variables, the regression model could predict the patients having BM with 95% and viral meningitis with 93.2% accuracy.