Objective: To develop a diagnostic decision rule based on clinical features to predict the risk of bacterial meningitis in children with signs of meningeal irritation.
Design: Retrospective and prospective.
Method: Predictors for bacterial meningitis were identified by collecting clinical data from the records of 360 patients (aged 1 month to 15 years) who consulted the Casualty Department, Sophia Children's Hospital, Rotterdam, the Netherlands, with signs of meningeal irritation during the period 1988-1998. The diagnostic decision rule derived was prospectively validated on 226 similar children who consulted the casualty departments of four hospitals in the Netherlands during the period 1999-2001.
Results: Predictors for bacterial meningitis were the main complaint and vomiting (in the history) persisting for a long time, the presence of meningeal irritation, cyanosis, petechiae, disturbed consciousness (during physical examination), and a high serum concentration of C-reactive protein. Liquor analysis parameters with an added diagnostic value were the total polymorphonuclear cell count in the liquor and the liquor/blood glucose ratio. The use of these patient characteristics in a decision rule accurately predicted the chance of bacterial meningitis. The rule can also be used to refine the indication for lumbar puncture and empirical antibiotic treatment.