Clinical prediction rule for differentiating tuberculous from viral meningitis

SETTING: The Professor Dr Matei Bals National Institute of Infectious Diseases, Bucharest, Romania. OBJECTIVE: To create a prediction rule to enable clinicians to differentiate patients with tuberculous meningitis (TBM) from those with viral meningitis. DESIGN: We retrospectively analysed patients admitted to a tertiary care facility between 2001 and 2011 with viral meningitis and TBM. Patients were defined as having TBM according to a recently published consensus definition, and as viral meningitis if a viral aetiology was confirmed, or after ruling out bacterial, fungal and non-infectious causes of meningitis. RESULTS: We identified 433 patients with viral meningitis and 101 TBM patients and compared their clinical and laboratory features. Multivariable analysis showed a statistically significant association between TBM and the following variables: duration of symptoms before admission of ≥5 days, presence of neurological impairment (altered consciousness, seizures, mild focal signs, multiple cranial nerve palsies, dense hemiplegia or paraparesis), cerebrospinal fluid/blood glucose ratio < 0.5 and cerebrospinal fluid protein level > 100 mg/dl. We propose a diagnostic score based on the coefficients derived from the logistic regression model with a sensitivity and specificity for TBM of respectively 92% and 94%. CONCLUSIONS: Our study suggests that easily available clinical and laboratory data are very useful for differentiating TBM from other causes of meningitis.