Abstract
Background: The discrimination of tuberculous meningitis and bacterial meningitis remains difficult at present, even with the introduction of advanced diagnostic tools. This study aims to differentiate these two kinds of meningitis by using the rule of clinical and laboratory features. Methods: A prospective observational study was conducted to collect the clinical and laboratory parameters of patients with tuberculous meningitis or bacterial meningitis. Logistic regression was used to define the diagnostic formula for the discrimination of tuberculous meningitis and bacterial meningitis. A receiver operator characteristic curve was established to determine the best cutoff point for the diagnostic formula. Results: Five parameters (duration of illness, coughing for two or more weeks, meningeal signs, blood sodium, and percentage of neutrophils in cerebrospinal fluid) were predictive of tuberculous meningitis. The diagnostic formula developed from these parameters was 98% sensitive and 82% specific, while these were 95% sensitive and 91% specific when prospectively applied to another 70 patients. Conclusion: The diagnostic formula developed in the present study can help physicians to differentiate tuberculous meningitis from bacterial meningitis in high-tuberculosis-incidence-areas, particularly in settings with limited microbiological and radiological resources.