Background
Central nervous system infections cause high morbidity and mortality in high HIV-prevalence, low-resource settings where tuberculous meningitis (TBM) may be common but is rarely microbiologically diagnosed and may be under-treated due to a lack of diagnostic testing. We integrated routine TB diagnostics into diagnostic algorithms in Botswana and Zimbabwe to determine the potential role of early empiric TB treatment in individuals presenting with suspected meningitis.
Methods
Enhanced diagnostic packages were introduced at Princess Marina Hospital, Gaborone, Botswana between 2021-2024 & Parirenyatwa Hospital, Harare, Zimbabwe between 2022-2024. Xpert MTB/RIF Ultra was performed on all CSF samples with sufficient volume and results reported to clinicians in real-time. Patients were followed to hospital discharge. Clinical details were captured from medical records.
Results
1334 patients were recruited; HIV status was determined in 1153 individuals, of whom 78.9% were living with HIV (PLHIV, 910/1153). TBM prevalence was 7.1% (80/1131) amongst patients who had a successful Xpert MTB/RIF Ultra performed. The proportion of patients with confirmed TBM increased with CSF white cell counts, from 18.1% (37/204) at >10 cells/mm³ to 19.6% (20/102) at >50 cells/mm³ and 22.2% (18/81) at >100 cells/mm³ overall, and from 17.7% (27/54) to 20.3% (16/54) and 24.2% (15/54), respectively, among PLHIV. A further 17 patients had probable TBM and 300 had possible TBM based on a uniform case definition. Amongst PLHIV with TBM, 54.3% (19/35) were receiving ART for a median of 3.5 months (IQR 2.1-72.4) and 45.7% (16/35) had interrupted treatment. The median time from admission to receipt of TB treatment was 71.5 hours (IQR 37.1-140.1) and positive CSF Xpert MTB/RIF Ultra result triggered initiation of treatment in 61.1% (33/54) in patients with complete data on treatment times. Inpatient mortality for patients with a positive CSF Xpert MTB/RIF Ultra was 55.7% (44/79).
Conclusions
TBM prevalence in high HIV-prevalence African settings reached 24% in patients with HIV and cellular CSF and mortality from TBM was high. Initiation of TBM treatment was slow and empiric treatment was uncommon with only a third of patients with confirmed TBM receiving anti-tuberculous treatment prior to receiving a positive Xpert MTB/RIF result. Implementation of routine TB testing on CSF samples and consideration of early empiric treatment in high HIV-prevalence settings is warranted.
Table or Figure
