Female Anopheles mosquitoes transmit malaria sporozoites to humans in the context of a blood meal.
Female Anopheles mosquitoes transmit malaria sporozoites to humans in the context of a blood meal. In malaria-endemic areas, most of the ensuing infections are asymptomatic. Some, however, progress to an uncomplicated illness (fever, headache, body aches, and pains). Younger individuals, with less clinical immunity to malaria, are at highest risk of developing severe disease (anemia, cerebral malaria, and/or respiratory distress) and of dying (1). Because the relationship between malaria transmission and malaria mortality is so variable, and because both are challenging to measure, it has remained unclear whether decreases in malaria transmission, resulting from control measures, would actually decrease malaria mortality. On page 926 of this issue, Paton et al. (2) find that the higher the prevalence of malaria infection in a given community, the higher the incidence of severe malaria disease. These findings may be useful in tracking the impact of various malaria control measures over time.
Measuring malaria transmission is not straightforward. Two of the traditional metrics of exposure to malaria parasites, the entomological inoculation rate and cohort incidence studies, are expensive and difficult to measure. The entomological inoculation rate, or the number of infectious bites per person per unit time (usually per year), is the product of the human biting rate and the sporozoite rate. (Sporozoites are carried in mosquito salivary glands and are injected into skin when female Anopheles mosquitoes take blood meals from humans.) To estimate the human biting rate, volunteers (protected by prophylactic doses of antimalarial drugs) bare their legs and collect mosquitoes as they land. Sporozoite rates can be determined by analyzing the salivary glands of the mosquitoes. Measuring incidence rates for new malaria infections requires treating a study cohort with an effective antimalarial drug, ensuring that the treatment was successful, and then following the cohort longitudinally at frequent intervals to identify when new infections appear.
Paton et al. took a different approach and used community prevalence of malaria infection in 26 communities in Uganda, Kenya, and Tanzania, measured directly through surveys of households and in school children or extracted from published literature, as a proxy for malaria transmission. Malaria infection can be measured directly, by microscopy or polymerase chain reaction (PCR) to detect parasite DNA, or indirectly, using rapid diagnostic tests that capture parasite antigens. Of these, PCR is the most sensitive, followed by microscopy and then rapid diagnostic tests (3); the sensitivities of the latter two vary according to age, treatment history, and transmission intensity (4). PCR was not used in any of the sampling sites, and for the sites that used rapid diagnostic tests, the results were converted to the standard microscopy metric, parasites per microliter of blood (4).
Measuring malaria-associated mortality is also not straightforward. Many deaths occur outside of hospitals and are not captured systematically. Even within hospitals in malaria-endemic areas, “malaria infection” is not synonymous with “malaria illness.” Paton et al. chose patients who were “sick enough to be admitted to hospital with malaria” as the proxy for malaria-associated mortality and measured three clinical phenotypes (cerebral malaria, severe malarial anemia, and respiratory distress) individually and together. Observations were thus limited to hospitals with the capacity to detect malaria infections, characterize the level of consciousness, measure the degree of anemia, provide blood for transfusion, follow the patient through the entire clinical course, and capture the data reliably.
Broadly, Paton et al. found that for every 25% increase in community parasite prevalence (above a baseline of 17.6% and below an upper limit of 75%), annual rates of admission for severe malaria double, and that as prevalence rates rise, the average age of children admitted to the hospital drops (see the figure). A potential, but unavoidable, bias in these results is that these analyses involved populations with easy geographic access to hospitals (5). Whether a patient even presents to a hospital or not depends on a multiplicity of factors: distance between home and the hospital and availability of transportation and community perceptions of the quality of facility-based care (e.g., availability of drugs or competence of health care workers). In difficult to reach rural areas, severe illnesses unfold at home and never touch the health care system.
It is important to appreciate that the study sites were chosen to represent a range of parasite prevalence rates and are thus a substantial collection of (largely) cross-sectional studies. Longitudinal data are difficult and expensive to collect, but following the effect of changing parasite prevalence rates over time in response to control activities would help to define the kinetics of the change in the incidence of severe malaria. How much time is required for a given community prevalence rate to affect hospitalizations for severe malaria?
Over the course of the observational period included in this analysis (2006 to 2020), there was a sea change in recognizing and managing uncomplicated malaria in sub-Saharan Africa. Countries rapidly shifted to using highly effective artemisinin combination therapy (ACT) to treat patients with positive results on malaria rapid diagnostic tests (6). Community health workers were empowered to test members of their community with fever and to immediately provide ACTs to anyone with a positive rapid diagnostic test. This public health intervention could stop the progression of a malaria infection to severe disease and diminish the association between malaria prevalence and mortality described by Paton et al., but the design of the study precluded addressing this potential confounder.
The findings of Paton et al. are highly relevant to the evaluation and implementation of malaria vaccines. At present, for the most advanced candidate, RTS,S/AS01, a regimen of three monthly doses (with a fourth dose 12 to 18 months later) is envisaged for children between the ages of 5 to 17 months, the age group shown by Paton et al. to have the highest rates of severe malaria. The phase 3 trial results of RTS,S/AS01 showed a 30% reduction in severe malaria in 5 to 17 month olds in settings with good health care (to decrease malaria mortality) and high bed net coverage (to decrease malaria transmission) (7). But would widespread use of an effective malaria vaccine skew the association between community parasite prevalence and severe disease? Community parasite prevalence is unlikely to change in this vaccination scenario because the vaccine targets a small proportion of the population—but an effective vaccine would decrease the rate at which severe disease develops in the vaccinated population. Programmatic evaluation of RTS,S/AS01 is ongoing, with an interim data review and consideration for recommendations from the World Health Organization (WHO) expected late in 2021.
The study of Paton et al. is a useful high-resolution image of the status quo in three East African settings. The utility of the relatively easily acquired, robust data is clear and emphasizes the value of maintaining health systems that have been strengthened over the course of the COVID-19 pandemic. These systems are the source of the programmatic data needed to devise strategies and to evaluate responses to a variety of public health threats, including, but not limited to, malaria
originally from Science
