Introduction The early and prompt diagnosis of TBM is essential to improve outcomes. Delayed treatment leads to increased morbidity and mortality. Cerebrospinal fluid (CSF) findings may assist in early diagnosis as well as potentially identifying patients at high-risk of death.

Methods A retrospective analysis of patients treated for TBM from 2000–2010 was undertaken using the London TB Register. Data was collected from medical records and pathology results.

Results Mortality was 20% (8/41) and 12% (5/41) were left with permanent neurological deficits. Mycobacterium tuberculosis (MTB) was cultured from CSF in 37% (14/38). Mortality in patients in whom CSF was culture positive was higher than those whose CSF culture was negative (33% vs 16%). The mean value of CSF protein was higher in patients with poorer outcomes than those who recovered fully (2.72 vs 1.92 g/L).

Corticosteroids were given to 92% of patients (35/38). Of the three patients who did not receive steroids, one was left with a permanent disability.

Conclusions In our population, we observed high mortality and morbidity rates for patients diagnosed with TBM. A large proportion of diagnoses were not confirmed microbiologically. Higher mortality rates were observed in those whose CSF cultured MTB A possible explanation for this finding is that those patients for whom microbiological confirmation was not obtained, the presumptive diagnosis of TBM was not correct. There is a trend towards higher mortality and morbidity in those with higher CSF protein values. This may be an indicator of increased bacterial load with an associated increase in inflammatory cytokines.