Abstract

Diagnostic confusion often exists between tuberculous meningitis and other meningoencephalitides. Newer diagnostic tests are unlikely to be available in many countries for some time. This study examines which clinical features and simple laboratory tests can differentiate tuberculous meningitis from other infections. Two hundred and thirty two children (110 tuberculous meningitis, 94 non-tuberculous meningitis, 28 indeterminate) with suspected meningitis and cerebrospinal fluid (CSF) pleocytosis were enrolled. Tuberculous meningitis was defined as positive CSF mycobacterial culture or acid fast bacilli stain, or basal enhancement or tuberculoma on computed tomography (CT) scan with clinical response to antituberculous treatment. Non-tuberculous meningitis was defined as positive CSF bacterial culture or Gram stain, or clinical response without antituberculous treatment. Thirty clinical/laboratory features of patients with tuberculous meningitis and non-tuberculous meningitis were compared by univariate and multiple logistic regression analysis. Five features were independently predictive of the diagnosis of tuberculous meningitis (p < 0.007): prodromal stage ⩾ 7 days, optic atrophy on fundal examination, focal deficit, abnormal movements, and CSF leucocytes < 50% polymorphs. When validated on another set of 128 patients, if at least one feature was present, sensitivity was 98.4% and, if three or more were present, specificity was 98.3%. This simple rule would be useful to physicians working in regions where tuberculosis is prevalent.