Background
The threefold increase in malaria control and elimination funding over the last decade has resulted in widespread increased coverage of malaria control interventions, including vector control with long-lasting insecticidal nets (LLINs) and indoor residual spraying (IRS), improved case management with rapid diagnostic tests (RDTs) and artemisinin-based combination therapy (ACT), and intermittent preventive treatment for high-risk groups [
1]. It has been postulated that the scale-up, specifically of LLINs, ACT, and IRS, is responsible for the estimated 40% decline in malaria case burden in Africa from the year 2000 to 2015 [
2]. Yet, gains have been heterogeneous, and in some endemic areas, despite heavy investments in these interventions, malaria prevalence, morbidity, and mortality remain high. As more countries approach elimination or aim for rapid malaria transmission reduction towards pre-elimination, new tools and strategies are necessary to complement existing ones [
3].
The search for new tools to reduce malaria transmission towards elimination has led to a renewed focus on targeting the infectious human parasite reservoir, particularly the asymptomatic population [
4]. In some low transmission settings, >60% of the infected population is asymptomatic; this proportion typically increases with increasing transmission intensity [
5]. Mosquito infection studies have demonstrated that blood meals from persons with asymptomatic infections can result in transmission [
6‐
8], and it has been postulated that in areas of seasonal transmission, the asymptomatic population may sustain transmission through the dry seasons [
9‐
11].
Additionally, individuals with asymptomatic infections are likely to be infected for longer durations of time, some for many years [
12], as they are less likely to seek care and be directly impacted by strategies such as improved case management. Strategies that specifically target the asymptomatic population, such as mass test and treat (MTaT) campaigns where all community members are tested for malaria, and all positives are treated, and mass drug administration (MDA) where all community members are treated for malaria without testing, are potential complementary tools for rapid malaria transmission reduction towards elimination [
13,
14].
Research trials and public health programmes implementing MDA to control or eliminate malaria have been conducted for at least a century [
15,
16]. The majority of the MDA studies, including the influential Garki Project [
17], have shown a significant decrease in parasite prevalence shortly after MDA rounds, but when the trials were over, and in the absence of ongoing delivery of preventive services, a rebound in infection prevalence was observed, frequently as early as 6 months after trial completion [
18‐
22]. A recently published literature review found that only 12 of 182 (6.5%) published reports of MDA resulted in success towards elimination with a definition of zero indigenous cases of malaria in the target population for a minimum of 6 months after the end of all MDA rounds [
16]. Of note, in many of these MDA trials other key malaria control interventions (high coverage with effective vector control, case management, and surveillance and response including active case detection) were not in place during or after the trials.
In the era of ACT with long chemoprophylactic half-lives (e.g., dihydroartemisinin–piperaquine) and the potential adjunct of drugs with gametocytocidal (e.g., primaquine) or insecticidal (e.g., ivermectin) activity, there is renewed interest in evaluating time-limited mass treatment strategies to accelerate transmission reduction in the context of high coverage with other malaria control interventions. Robust mathematical models have been created to explore the impact of MDAs and MTaTs on malaria transmission. These models suggest that, in conjunction with high and sustained LLIN coverage and treatment with artemisinin-based combinations, interventions targeting the asymptomatic reservoir, such as MTaT and MDA, help to drive transmission in an area towards elimination; however, results differ based on transmission setting [
14,
23‐
28].
Additionally, recently published field studies assessing either modified focal screen and treat strategies, where small discrete geographic areas are identified for screening and treatment [
29,
30], or mass test and treat [
31] in different transmission settings suggest that MTaT may effectively identify the majority of the asymptomatic infected population and reduce malaria transmission. However, a study of MTaT in a low transmission setting in Zanzibar did not reduce malaria incidence [
32], and more recently, and subsequent to the study described in this manuscript, a World Health Organization expert review group stated that with current diagnostic tests, MTaT could not be recommended for malaria transmission reduction [
33]. Additionally, there were concerns that MDA may exert excessive drug pressure that may lead to adverse events in uninfected persons or enhance the development and spread of anti-malarial resistance; there was less concern for these negative outcomes for MTaT strategies [
33,
34]. However, with MTaT, there was concern that the sensitivity of available rapid diagnostic tests would be too low to identify a large enough proportion of infected individuals to make a significant impact on malaria transition.
This manuscript describes and provides a discussion of a study design and methodology to evaluate the acceptability, effectiveness, and costing of a community-based MTaT strategy for malaria in an area of high transmission.
Discussion
Previous results of trials or programmes incorporating mass campaigns, such as MTaT, for transmission reduction on a large-scale have been contradictory or inconclusive [
31,
32]. As the population of individuals harbouring asymptomatic infections has been increasingly implicated in sustaining malaria transmission there has been renewed interest in evaluating time-limited MTaT in different transmission settings with high coverage of malaria control interventions for rapid malaria transmission reduction. The study site in western Kenya provided distinct advantages for evaluating MTaT including; (1) high malaria transmission despite high coverage with LLINs, (2) the existence of a HDSS through which sampling frames and population level data including births, deaths, and migration have been continuously monitored, (3) over 30 years of historical data on epidemiological and entomological malaria indices, (4) advanced laboratories and research infrastructure, and (5) strong relations with the communities and public and private health sectors in the region. In this manuscript, the strengths and limitations of the study design choices are discussed.
Interventions such as MTaT are designed to be effective at high coverage levels and, therefore, must be evaluated at the community level, but the cost and complexity of cluster-randomized trials combined with the heterogeneity of malaria transmission and differential access to health care complicate the design of such trials. To minimize differences in the spatial distribution of malaria transmission, differences in treatment availability and provider proficiency, the effect of distance on health-care seeking, and LLIN coverage, each cluster around the study health facilities was randomly assigned to the intervention or control arm, and only villages with a centroid within 3 kms from the health facilities were included. The result of this decision was that clusters abutted one another and were relatively small for campaign-style interventions, which are intended to be implemented over larger geographic spans for transmission reduction. This smaller size and geographic proximity may artificially increase the impact of migration as individuals may be more likely to migrate out of a smaller geographic cluster than a larger one, and through daily movement, may be more likely to enter a cluster that is closer to where they reside than one that is geographically distant.
Migration is known to play a major role in malaria transmission as parasites are transported from one area to another [
59], or one cluster to another. In this trial, migration and contamination of parasites into other clusters, was likely non-differential, and thus may bias the outcome to the null. It was recognized that human migration was unavoidable and would serve as a limitation to evaluation, and measures were taken to quantify it during MTaT rounds, at cohort visits, and through the HDSS census, however daily movement between clusters was not quantified. To mitigate the impact of daily movement on the outcome measure while maintaining statistical tenets for cluster numbers, there was a consideration for creating much larger clusters that were geographically distant from one another, while maintaining the same number of clusters for statistical validity. However, this would have been cost-prohibitive and the necessary geographic separation would likely have led to significant confounding from transmission differences and health facility associated factors such as stockouts and care-seeking behaviour. When choosing a study design, researchers must carefully weigh the benefits of bias mitigating decisions versus the introduction of contamination. Smaller clusters were used in this study. Other study designs for evaluating population-based interventions such as stepped-wedge approaches, through which larger clusters can be more easily created, were considered, however at the time, the statistical rigor and the practicality of these designs were in question [
60,
61].
The frequency and timing of test and treat rounds in this study are based on robust mathematical models. These models account for multiple variables including LLIN coverage, usage, and baseline transmission levels. All models are based on assumptions from years of research in malaria in general, and specifically to the study site. The models suggest that clearing the parasite reservoir during the dry season, when the number of sub-clinical infections and vectors are lowest, would be the most effective way of reducing transmission, and that spacing the rounds 1–2 months from each other and ensuring at least one of the rounds is delivered immediately before the peak transmission season would be the most cost-effective way of performing this. However, model assumptions are constantly being updated as new field data are collected. Data from this study will be incorporated into the models to refine them.
Authors’ contributions
AS had primary responsibility for writing the manuscript. MD, KL, SPK, SK, LS, MH, JW, AS contributed to study design. All authors suggested revisions. All authors read and approved the final manuscript.