Background: A prediction model based on clinical and cerebrospinal fluid (CSF) analysis has been proposed for the differentiation of Lyme meningitis (LM) from non-Lyme aseptic meningitis (NLAM) in the United States. No similar model has ever been proposed for European patients. The objective of our study was to develop a prediction model to differentiate LM from NLAM based on clinical and CSF biologic data.
Methods: The medical charts of all children older than 2 years of age admitted to our hospital from 1996 through 2006 with a definite diagnosis of LM were analyzed and compared retrospectively with those having a diagnosis of NLAM. Chart review included the duration of symptoms, the presence of cranial neuropathy, and CSF analysis.
Results: A total of 93 patients were included (LM: 26 patients; NLAM: 67 patients) in the study. Patients with LM had statistically more frequent cranial neuropathy (73% vs. 4%), displayed a longer duration of symptoms before admission (8.8 vs. 1.8 days), had a higher CSF protein (71 vs. 38 mg/d), and had a lower percentage of neutrophil cells in the CSF (3.4% vs. 51%) than patients with NLAM. A predicted probability was derived from these 4 variables. At a cutoff point of >0.432, the model had a negative predictive value of 100% and a positive predictive value of 92.3%, with a sensitivity of 100% and a specificity of 97%.
Conclusions: We report the first European prediction model for LM. Owing to its high negative predictive value, this model may assist physicians in managing aseptic meningitis (AM) while awaiting serologic tests, especially in Lyme endemic regions.
From the *Département de Pédiatrie, †Unité de Biostatistique et Documentation Médicale, and ‡Département de Microbiologie, Université Catholique de Louvain, Cliniques, Universitaires de Mont Godinne, Yvoir, Belgium.
Accepted for publication October 20, 2008.
Address for correspondence: David Tuerlinckx, MD, Département de Pédiatrie, Cliniques, Universitaires de Mont-Godinne, 5530 Yvoir, Belgium. E-mail: firstname.lastname@example.org.