Objectives: To refine and to re-validate the best current tool (the Nigrovic rule: ''outpatient management may be considered for children without seizure, blood neutrophil count>or=10,000/mm(3), positive cerebrospinal fluid -CSF- Gram-staining, CSF protein>or=80 mg/dl, or CSF neutrophil count>or=1,000/mm(3)'') proposed to distinguish between aseptic meningitis (AM) and bacterial meningitis (BM) in the emergency department.
Methods: Children hospitalized for BM between 1995 and 2004, or AM between 2000 and 2004 were included, and randomly divided into derivation (111 children, 14 BM) and internal validation (57 children, 7 BM) sets. The Nigrovic rule was refined on the derivation set, introducing new variables (purpura, toxic appearance and high serum procalcitonin), changing variables thresholds (CSF protein) and withdrawing some variables (blood neutrophil count, CSF neutrophil count), according to previous results, with the aim to obtain 100% sensitivity user friendly tool. The refined rule was then applied on the internal validation set, stayed blinded during the derivation process.
Results: The refined rule was: start antibiotics in case of seizure, purpura, toxic appearance, procalcitonin>or=0.5 ng/ml, positive CSF Gram-staining, or CSF protein>or=50 mg/dl. The refined rule had 100% sensitivity on the derivation and the internal validation sets (95% confidence interval 78-100, and 65-100, respectively) with 62 and 51% specificity, respectively.
Conclusion: The refined rule (called Meningitest) was a highly sensitive, specific and user friendly tool that could allow to safely avoid>50% a posteriori unuseful antibiotic treatments for patients with AM.