Mortality in adult patients with culture-positive and culture-negative meningitis in the Botswana national meningitis survey: a prevalent cohort study

Abstract

Background CNS infections are a leading cause of HIV-related deaths in sub-Saharan Africa, but causes and outcomes are poorly defined. We aimed to determine mortality and predictors of mortality in adults evaluated for meningitis in Botswana, which has an estimated 23% HIV prevalence among adults. Methods In this prevalent cohort study, patient records from 2004-15 were sampled from the Botswana national meningitis survey, a nationwide audit of all cerebrospinal fluid (CSF) laboratory records from patients receiving a lumbar puncture for evaluation of meningitis. Data from all patients with culture-confirmed pneumococcal and tuberculous meningitis, and all patients with culture-negative meningitis with CSF white cell count (WCC) above 20 cells per µL were included in our analyses, in addition to a random selection of patients with culture-negative CSF and CSF WCC of up to 20 cells per µL. We used patient national identification numbers to link CSF laboratory records from the national meningitis survey to patient vital registry and HIV databases. Univariable and multivariable Cox proportional hazards models were used to evaluate clinical and laboratory predictors of mortality. Findings We included data from 238 patients with culture-confirmed pneumococcal meningitis, 48 with cultureconfirmed tuberculous meningitis, and 2900 with culture-negative CSF (including 1691 with CSF WCC of up to 20 cells per µL and 1209 with CSF WCC above 20 cells per µL). Median age was 37 years (IQR 31-46), 1605 (50%) of 3184 patients were male, 2188 (72%) of 3023 patients with registry linkage had documentation of HIV infection, and median CD4 count was 139 cells per µL (IQR 63-271). 10-week and 1-year mortality was 47% (112 of 238) and 49% (117 of 238) for pneumococcal meningitis, 46% (22 of 48) and 56% (27 of 48) for tuberculous meningitis, and 41% (1181 of 2900) and 49% (1408 of 2900) for culture-negative patients. When the analysis of patients with culture-negative CSF was restricted to those with known HIV infection, WCC (0-20 cells per µL vs >20 cells per µL) was not predictive of mortality (average hazard ratio 0•93, 95% CI 0•80-1•09). Interpretation Mortality from pneumococcal, tuberculous, and culture-negative meningitis was high in this setting of high HIV prevalence. There is an urgent need for improved access to diagnostics, to better define aetiologies and develop novel diagnostic tools and treatment algorithms.

Mortality from HIV‐associated meningitis in sub‐Saharan Africa: a systematic review and meta‐analysis

“…Restricting to studies after 2000, pooled short‐term mortality was 42% (95% CI: 25% to 59%, I 2 = 93%, seven studies). Ten‐week mortality from a national meningitis audit in Botswana of microbiologically confirmed cases was 46% (95% CI: 31% to 61%) . Long‐term mortality was reported in two studies, within one year in the study from Botswana and after completion of six months of antituberculous therapy in a Nigerian study , both studies with exclusively definite cases.…”

“…Short‐term mortality for 1008 cases was reported in 11 studies (two from a single hospital with combined patient outcomes in Uganda and two from a single hospital with combined outcomes in Zambia ; two‐week mortality specified in one study , in‐hospital mortality in ten studies with length of hospitalization not specified). Pooled mortality was 46% (95% CI: 33% to 59%, I 2 = 92%, 11 studies) (Figure ), and was higher in studies with only definite cases (51% (95% CI: 37% to 65%), I 2 = 78%, six studies) and mostly possible/probable cases (40% (95% CI: 23% to 59%), five studies).…”

“…Study quality was relatively low, with most studies not including details of ART or antituberculous treatment or average duration of hospitalization. Five included cases before 2000 [34,[109][110][111][112], whereas seven had a full period of observation after 2000 [53,60,67,107,108,113,114]. Eight countries were represented, with greatest representation from South Africa (three studies), Uganda (two studies) and Zambia (two studies).…”

“…Several potential studies were excluded on fulltext review (Appendix S1): Reasons included TB case series, for example, exclusively multi-drug resistant cases [115][116][117], high risk of selection bias [42,[118][119][120], significant missingness of outcomes [121][122][123] and unknown or inadequate description of diagnostic criteria [50,74,[124][125][126]. Short-term mortality for 1008 cases was reported in 11 studies (two from a single hospital with combined patient outcomes in Uganda [53,113] and two from a single hospital with combined outcomes in Zambia [67,108]; two-week mortality specified in one study [114], in-hospital mortality in ten studies with length of hospitalization not specified). Pooled mortality was 46% (95% CI: 33% to 59%, I 2 = 92%, 11 studies) ( Figure 5), and was higher in studies with only definite cases (51% (95% CI: 37% to 65%), I 2 = 78%, six studies) and mostly possible/probable cases (40% (95% CI: 23% to 59%), five studies).…”

Epidemiology of adult meningitis during antiretroviral therapy scale-up in southern Africa: Results from the Botswana national meningitis survey

“…Standard CSF evaluation, following national laboratory protocols, remained unchanged during the study period and consisting of white cell count (WCC) and differential (lym phocytes, neutrophils) when WCC ≥10 cells/μL, protein and glucose, microscopy (gram stain, India ink), and bacterial and fungal culture. 13 The full work-up was incomplete for some patients (e.g. missing WCC, protein, or glucose), due to reagent stockout or incomplete laboratory recording.…”