High death rate for HIV-associated meningitis reported in African hospitals
Evidence review finds approximately half of hospital patients with cryptococcal, tuberculosis or pneumococcal meningitis died within two weeks or during hospitalisation.
The analysis of 79 studies compared deaths from HIV‐associated meningitis in two different settings: hospitals and clinical trials.
Despite improved access to effective antiretroviral treatment (ART) in sub-Saharan Africa, a high proportion of people living with HIV remain virally unsuppressed, leaving them at increased susceptibility to potentially fatal infections such as meningitis. A 2017 analysis found cryptococcal meningitis alone causes around 15% of HIV‐associated deaths worldwide – almost three‐quarters of which occur in sub-Saharan Africa.
Treatment for HIV‐associated meningitis is challenging, even in high-income countries, and outcomes are even worse in countries with limited resources. The recommended approach for treating cryptococcal meningitis is initial treatment with amphotericin, an effective but highly toxic drug that is given intravenously, followed by treatment with a drug called fluconazole. But in resource‐constrained settings, fluconazole alone is often used as it costs less, can be administered orally, and is less toxic.
The research found approximately half of people diagnosed with cryptococcal, tuberculosis or pneumococcal meningitis – the most common types of HIV-related meningitis – died within two weeks or during hospitalisation in routine care settings. Although outcomes for pneumococcal meningitis were similar in hospitals and clinical trials, a significant difference was found between the two settings for cryptococcal meningitis.
Overall, 44% of people with cryptococcal meningitis died in hospital settings, regardless of whether they were given amphotericin or fluconazole, compared to 21% in clinical trials when only amphotericin was given. This suggests the death rate found in routine care is likely to be linked to multiple factors that arise within hospitals and is not purely the result of the drugs on offer.
When examining the data regionally, more than 50% of patients with cryptococcal meningitis died within two weeks in West and Central Africa; a significantly higher rate than in Southern Africa where 37% of people died.
The proportion of people with pneumococcal meningitis who died within two weeks was similar in clinical trials and hospital settings, at around 54%. This may be reflective of the less intensive treatment currently recommended for pneumococcal disease.
The proportion of people dying within two weeks from TB-related meningitis was found to be 46% in hospital settings. Two studies looked at medium and longer-term outcomes for TB-related meningitis. These found more than two‐thirds of people receiving treatment for TB meningitis in hospital settings died. However, a lack of data meant the review was unable to compare these results with those from clinical trials of a similar nature. One trial from South Africa reported that 12% (3 out of 34) of participants died within nine‐months, but this study excluded people with severe TB and those who did not initiate ART or were judged to have poor drug adherence.
Few studies evaluated by the review looked at longer‐term outcomes among people with HIV-related meningitis who survived in the short or medium term. In addition, most studies reported outcomes for people treated at hospital referral clinics, where staffing, pharmaceutical and laboratory resources are likely to be better than in lower‐level health facilities.
The findings of this study indicate that better strategies and preventative measures, such as improved screening for cryptococcal meningitis and childhood pneumococcal vaccinations, are needed to reduce deaths from HIV‐associated meningitis in the region. It also suggests that increasing access to amphotericin alone will not be enough to significantly reduce the number of people with HIV who die as a result of meningitis.
Community guidelines for comments can be found in our website T&Cs