Distinguishing bacterial from viral meningitis: the critical importance of the CSF lactic acid: author's reply

Abstract

Sir: We thank Dr. Cunha for his attention to our multivariate approach of community-acquired bacterial meningitis [1]. He suggests that combining clinical presentation with the clinical severity is helpful in distinguishing bacterial from viral meningitis according to the study by Durand et al. [2], that meningoencephalitis is rare during bacterial meningitis (BM), that CSF lactate values are of interest for the diagnosis, and thus that there was no need to use a multivariate approach. In contrast with Durand et al. who analyzed only community-acquired and nosocomial BM, we compared the clinical presentation and the CSF findings in confirmed, acute community-acquired BM with those observed during viral meningitis. Whereas most of BM series have included nearly 20% unconfirmed BM [2, 3], we paid special attention to enroll only microbiologically documented BM. Therefore we believe that the clinical findings of our population are representative of the clinical presentation of BM. In the past it was thought that BM is only a meningitis, and thus many physicians believe that its diagnosis depends only on the CSF findings. In fact, signs of cerebral dysfunction such as declining consciousness, seizures, and focal neurological deficits are found in 85% of patients presenting with acute BM [4]. Furthermore, in accordance with our results, Behrman et al. [5] demonstrated that all patients with BM have a change in mental status, while none of the aseptic meningitis patients do. We agree with the fact that definitive diagnosis of meningitis depends on CSF analysis, but one of the two predictors of BM in our model was an absolute neutrophil count of PMN greater than 1,000/mm3 and not the absolute CSF white blood cell. Dr. Cunha highlights that CSF glucose was not a predictor of BM in our series. This finding is explained mainly by the important weight of the clinical presentation and conforms with the results of the three studies that have used a multivariate approach to evaluate the performance of initial clinical presentation to distinguish between bacterial and viral meningitis. However, when using only CSF findings, CSF glucose level was a predictor of BM (results not shown). CSF lactate level could have been of interest, but according to the retrospective design of our study, this marker was not determined. Furthermore, its predictive value have never been evaluated in a multivariate model, but according to the weight of the clinical presentation, this parameter may be not predictive. In any case, our main purpose was to highlight the fact that a patient with a high suspicion of meningitis without any sign of severity, i.e., abnormal neurological findings and/or shock at admission is at very low probability of BM, even if the CSF percentage of PMN is high, a common presenting emergency department problem evaluated in the United States at 1 case per 1,000 visits per year [6]. Our model may be useful to reduce unnecessary hospitalization, duration of antibiotics (if used) and imaging. In contrast, all patients with meningoencephalitis must be immediately evaluated and must receive antibiotics or acyclovir according to the CSF findings without delay.