Huge fall in deaths after improving meningitis care for people with HIV in African hospitals

An implementation project in three African countries that used an algorithm to quickly identify, diagnose, and treat people with HIV-related central nervous system infections during routine care resulted in a significant decrease in mortality, according to a recent publication in The Lancet HIV. Before the intervention, 49% of patients died within two weeks, reducing to 24% during the implementation phase.

Led by Dr Sayoki Mfinanga and colleagues, the Driving Reduced AIDS-associated Meningo-encephalitis Mortality (DREAMM) project was conducted in five public hospitals in Cameroon, Malawi, and Tanzania. It empowered local health leaders, strengthened existing health systems, administered rapid diagnostic tests at the bedside and provided other tests, medication, and equipment to enable the implementation of the World Health Organization (WHO) cryptococcal meningitis guidelines.

Opportunistic infections affecting the central nervous system (CNS), such as meningitis, are a significant cause of illness and death in people with HIV, particularly in areas with limited resources. Meningitis, which is the inflammation of the spinal cord or brain's membrane, is prevalent in people with advanced HIV and is characterised by symptoms like fever, altered mental status, and neck stiffness. In addition, individuals may experience severe headaches, nausea, and vision difficulties. The symptoms may vary depending on the cause, which is usually an infection, whether that is viral, fungal (one of which causes cryptococcal meningitis) or bacterial (including TB). Cerebral toxoplasmosis is caused by a parasitic infection.

The exact burden of HIV-related CNS infections is unknown, because many cases in routine care are undiagnosed, despite recent advances and updated WHO guidelines, but they might contribute up to a third of HIV-related deaths. Lack of access to tests and medicines is a barrier to saving lives from HIV-related CNS infections.

The study

The researchers implemented a three-phase intervention project: observation, training, and implementation. Its impact was assessed by comparing two-week all-cause mortality between the observation and implementation phases, with ten-week all-cause mortality also measured.

During the observation phase, conducted for several months at each site, the focus was on documenting existing practices and procedures, and identifying the barriers and facilitators involved in delivering care for HIV-related CNS infections.

Common barriers to care recorded included poor retention of patients in care due to ART non-adherence or failure, late presentation to care, limited or no access to rapid diagnostic tests, and little or no access to life-saving medicines for cryptococcal meningitis such as flucytosine and amphotericin B.

During the training phase, local leaders in healthcare settings were empowered to design and execute the DREAMM interventions. A health system strengthening approach was adopted for hospitals to ensure the delivery of quality care through an open-access, co-designed education programme for front-line healthcare workers and laboratory technicians. Clinical and laboratory pathways were mapped and optimised to diagnose patients and rapidly administer targeted treatment. Staff were also trained on the leading causes of HIV-related CNS infection as well as the DREAMM algorithm.

The DREAMM algorithm had two key features. First, clinical staff performed rapid diagnostic testing at the bedside, alongside laboratory testing using standard microbiological techniques such as cerebrospinal fluid culture. All participants underwent testing for cryptococcal antigen (CrAg) using a lateral flow assay in their blood and cerebrospinal fluid, as well as lipoarabinomannan (LAM) testing of their urine.

Second, the algorithm implemented the 2018 WHO cryptococcal meningitis guidelines. In Malawi and Tanzania, the first-line therapy for cryptococcal meningitis was one week of amphotericin B deoxycholate plus oral flucytosine, followed by one week of high-dose fluconazole. In Cameroon, the alternative oral regimen of two weeks of high-dose fluconazole plus flucytosine was chosen due to its ease of administration and monitoring.

In the implementation phase, essential medicines, diagnostic tests and medical equipment for HIV-related CNS infection were provided free of charge.

This study recruited adults aged 18 years or older who were living with HIV and suspected to have a first episode of CNS infection. Most of the participants had advanced HIV disease (over 75%), were ART experienced (69%), had altered mental status (over 75%), and were critically unwell (89%).

During the observation phase, 139 participants, with a median age of 38 and equal gender distribution, were enrolled. However, ten were lost to follow-up. Few investigations were performed: diagnostic lumbar punctures and CT brain scans were done in 30% and 11% of participants, respectively. Cerebrospinal fluid results following lumbar puncture were available in 24% of participants to inform targeted therapy. Treatment was largely empirical, including broad-spectrum antibiotics and fluconazole.

In the implementation phase (which lasted an average of ten months per site), 356 participants were enrolled. The majority (54%) were female, and the median age was 39 years. Lumbar punctures were done for 89% of participants within 90 minutes of enrollment for those with cryptococcal and tuberculous meningitis (lumbar punctures may be contraindicated for those with suspected toxoplasmosis). The investigations confirmed CNS infections in 269 people (148 had cryptococcal meningitis, 52 tuberculous meningitis, 26 bacterial meningitis and 44 cerebral toxoplasmosis), while CNS infections were ruled out in 87 people.

Results

• Out of 129 participants in the observation phase, 49% died within two weeks, while only 24% of 266 participants died in the implementation phase within two weeks. When adjusted for site, sex, age, and ART exposure, all-cause mortality was significantly lower in the implementation phase, with an adjusted risk difference of -23% (95% confidence interval -33 to -13; p<0.001).
• At ten weeks, 55% of participants died in the observation phase, whereas 39% died in the implementation phase (adjusted RD –13%, 95% CI –24 to –3; p=0.01).
• In the observation phase, the median time to death was three days (IQR 2–7), and in the implementation phase, it was nine days (IQR 4–22).

The researchers suggest that scale-up of the DREAMM interventions could save lives from HIV-related CNS infections in routine care settings. They add that costs are likely to be low as they are mainly driven by the time of local leadership to train and mentor healthcare workers and laboratory technicians as well as the cost of existing technologies.

“Our work has shown that African-led health-system strengthening in resource-limited settings is practicable and feasible,” they say.