Broad-range bacterial PCR and meningitis with a negative Gram’s stain

The recent article in CMI by Welinder-Olsson et al. adds evidence to support the use of broad-range PCR assays for the diagnosis of community-acquired bacterial meningitis. However, some additional information is needed to put these results into clinical perspective.

First, in the report of a diagnostic study, readers expect to know how the reference status—diseased or non-diseased—was defined. The authors describe 57 patients with bacterial meningitis: (i) 25 patients with bacteria detected by PCR and culture of the cerebrospinal fluid (CSF); (ii) 19 among 26 patients with bacteria detected by PCR, but not by culture of the CSF; (iii) six among 14 patients with bacteria detected by culture, but not by PCR of the CSF; and (iv) seven patients with no bacteria detected by PCR or culture of the CSF, but with clinical and laboratory findings suggestive of bacterial meningitis and a relevant pathogen identified from blood culture. However, the total number of bacterial meningitis cases is stated to be 74. Some details on how the diagnosis was reached in the 17 missing patients would be welcome.

Second, the concordance between CSF culture and CSF bacterial PCR is only moderate, with kappa = 0.34 (95% CI 0.14–0.55). As both tests obviously provide useful diagnostic information, the authors conclude (appropriately) that they complement each other. Nonetheless, the outcome of bacterial meningitis is highly dependent on prompt initiation of antibiotic therapy, i.e., before the results of CSF culture are available. From a clinical point of view, the difficult problem is thus to make a quick diagnosis. In this respect, a Gram’s stain of CSF is more useful than the CSF culture, since the results are available without delay. A positive result essentially confirms the diagnosis [2] and guides the choice of an appropriate antibiotic regimen. Since the clinician already has the information needed to act efficiently, bacterial PCR is not very useful in such cases. In contrast, PCR might be very helpful in the case of a CSF with a negative result following Gram’s stain. It would therefore be valuable to know the sensitivity and specificity of broad-range bacterial PCR for the subgroup of patients with a negative CSF Gram’s stain.