A cross-sectional survey of P. falciparum prevalence and risk factors for infection was conducted in October and November 2017 shortly after the peak of malaria transmission.
The study was conducted in ten villages in the Banfora Health District, an area of Sudanian savannah in the Cascades region, south-west Burkina Faso (lying between 10°40’ to 10°04’13’’ north latitude and 5°01’21’’ to 4°46’18’’ west longitude) with a population of 713,059 in 2018 (Figure 1) 15. The study site experiences intense seasonal malaria transmission with peaks during the rainy season, from May to November 16, with most cases occurring in September 17. Plasmodium falciparum accounts for 90% of cases 16 and the main malaria vectors are Anopheles gambiae, An. coluzzii and, to a lesser extent, An. arabiensis 14,18. The site has an entomological inoculation rate (EIR) of 80 infective bites per child during the transmission season 8, with another study suggesting up to 52 infective bites per person per night during the peak transmission season in October 19. The NMCP undertook an ITN universal coverage campaign (distribution of one ITN for two persons) in 2010, 2013 and 2016 using nets either treated with permethrin or deltamethrin. No additional nets were distributed by the study team. Indoor residual spraying was not conducted during the study or the preceding 12 months. In Burkina Faso since 2014, children aged 3 to 59 months receive SMC on four occasions during the transmission season using SP-AQ as per WHO recommendations 7.
The map was generated using QGIS 3.16 20. Background layers were downloaded for OpenStreetMap 21, villages were digitised by the authors using GPS coordinates collected in the field using a GARMIN eTREX 10 GPS.
A random sample of 10 villages were selected from a list of villages in the study area using a two-stage process. Five health centres in the study area were chosen with each health centre having a catchment radius of approximately 10 km. Two villages were randomly selected from each catchment area, giving a total of 10 villages, at least 3 km apart. Permission to enter the communities was sought from village leaders.
An age-stratified cross-sectional survey of both children and adults in three age groups (2 to <10 years, 10 to <30 years, ≥30 years) was conducted to determine the prevalence of P. falciparum. The survey aimed to sample 1,200 individuals, 400 from each of the three age strata. 150 study subjects (50 in each age strata) were randomly selected from the Health and Demographic Surveillance System (HDSS) lists of the 10 study villages and entered the screening process. Each study subject was selected from a different household. The first 40 individuals per age strata and per village who provided informed consent were enrolled. Participants were excluded if they were currently participating in a trial of a malaria vaccine or drug, under chemoprophylaxis (except for SMC) or currently participating in a related cohort study 8.
P. falciparum infection survey
A finger-prick blood sample was taken from each participant. Two blood slides were prepared and a malaria rapid diagnostic test (RDT, SD BIOLINE Malaria Ag P.f/Pan screening test Abbott, Geonggi-do, Republic of Korea) performed for point-of-care diagnosis of those with fever (axillary temperature ≥37.5°c) or history of fever in the past 48 h. Individuals with positive RDTs were offered treatment with artemether-lumefantrine (AL) according to national guidelines 22. Thick blood films were stained with Giemsa and examined under 100-fold magnification by experienced microscopists centrally at Centre National de Recherche et de Formation sur le Paludisme (CNRFP) in Banfora. Parasite counts were recorded per high power field and 100 fields counted before a slide was declared negative. Each slide was read separately by two independent microscopists. Discrepancies in positive and negative reads and parasite counts differing by more than 10-fold between the two reads were resolved by a third reader. The final result was average of the two-closer readings. Blood spots were also taken for polymerase chain reaction analysis but unfortunately due to a lack of amplification we did not obtain results.
Risk factor survey
At the same time as the infection survey, a questionnaire was administered to the study participant (or caregiver for children, as indicated) to gather demographic information (age, gender, ethnicity, religion, education, occupation), use of malaria control measures (bednet use the previous night, use of topical or household insecticides e.g. insecticide aerosols, mosquito coils in the past week, receipt of SMC in previous month), house construction (roof material, whether the space between the wall and the roof, i.e. the eaves, were closed), presence of electricity or functioning fan, presence of animals within 5 m of the household, and travel history outside of the village in the previous two weeks. Information was also collected from the study participant or the head of household (as indicated) on asset ownership and household characteristics, following standard procedures used in the Burkina Faso Demographic and Health Survey (DHS) 23.
Human landing catches were carried out inside houses twice a month during the 2017 transmission season in each of the 10 villages 14. Four households were sampled on one night every month, with a different group of households selected the following month to maximise spatial coverage. Volunteers collected mosquitoes landing on their legs between 19.00 h to 06.00 h. Mosquitoes were typed to species using established morphological keys. Phenotypic insecticide resistance was measured using WHO tube tests as per standard procedures 24. Assays were performed with An. gambiae s.l. mosquitoes reared from larvae collected in seven study villages during the 2017 transmission season (because of limited availability of larval habitats in the other three villages during the period of the survey) 14.
Sample size considerations
A random sample of 400 individuals from each of the three age groups (2 to <10 years, 10 to <30 years, ≥30 years) were selected from 10 villages giving a total sample size of 1,200 individuals. Assuming a true parasite prevalence ranging between 40 to 60% across the three age groups 25, the study was able to measure the point prevalence of P. falciparum infection by microscopy with a 5% precision at the 95% confidence level 26.
Data management and statistical analysis
Data were collected on personal digital assistants (PDAs) programmed with an electronic data capture system, Kobo Collect (Version 1.4.8). Forms were piloted in the field prior to use and had drop-down boxes and consistency checks to avoid data entry errors. PDAs were uploaded by fieldworkers weekly to a central computer.
The primary outcome was P. falciparum infection confirmed by microscopy (any level of parasite density). Secondary outcomes were: i) prevalence of symptomatic malaria defined as axillary temperature ≥ 37.5°C (or history of fever within the previous 48 h) with microscopically confirmed P. falciparum infection and ii) prevalence of high-density P. falciparum infection (>5,000 parasites/µL) detected by microscopy.
Principal component analysis was used to calculate the socio-economic status (SES) factor score (based on asset ownership and household characteristics). SES factor scores were ranked and households divided into five equal wealth quintiles (1 being the poorest, through to 5, least poor). The EIR or estimated number of infectious bites per study participant during the transmission season was calculated using the formula EIR = Ma×S×d where Ma is the human biting rate, estimated from the arithmetic mean number of female An. gambiae s.l. caught per human landing catches across the six-month transmission season, where S is the proportion of female An. gambiae s.l. found to be sporozoite positive by village and d is the number of days in the transmission season.
Mean values were compared using a t-test and proportions compared using chi-squared tests. Parasite prevalence was estimated as the proportion of subjects infected divided by the number of subjects tested. Logistic regression was used to investigate the association between malaria infection and risk factors, adjusting for clustering by village. Univariate analysis was conducted followed by construction of a simple multivariate model in which every risk factor was included, irrespective of whether the variable was significant in the univariate model. All analyses were carried out using Stata 15 (Statacorp, Texas, USA).
Malaria transmission dynamic model
A widely used transmission dynamics mathematical model of malaria was used to investigate the impact of modifying SMC in the Cascades region of Burkina Faso. The individual-based stochastic model mechanistically captures transmission of P. falciparum malaria in humans and Anopheles mosquito vectors. All differential equations describing the dynamics of the infection in populations with malaria control interventions have been comprehensively reported in Griffin et al. 27 and Winskill et al. 28 whilst the model code is available from https://github.com/jamiegriffin/Malaria_simulation. This model captures the age distribution of infection in areas with different levels of endemicity 29 and has been used to investigate the impact of SMC 30. Here we use a version of the model calibrated for the region (Lambert et al, unpublished) which is parameterised with local epidemiological 8 and entomological data 14,19 and fitted to estimates of malaria prevalence (assessed by microscopy) across the Cascades region collated by the Malaria Atlas Project (https://malariaatlas.org/). It is used to predict the number of clinical cases per person in the whole population between 2019 and 2022 following the mass ITN campaign in 2019. Results are averaged over the three-year period (the time between ITN campaigns) as cases will depend on ITN use which drops following mass distribution. The type of nets distributed varied at the district level within the Cascades region. Here we assume that the population received either pyrethroid-only or pyrethroid-piperonyl butoxide (PBO) ITNs 31. PBO is an insecticide synergist which inhibits the action of resistance-associated metabolic enzymes of the cytochrome P450 family and improves control of pyrethroid-resistant anopheline mosquitoes. The added advantage of pyrethroid-PBO ITNs over pyrethroid-only ITNs is estimated from a meta-analyses of experimental hut trial data 32 and the level of resistance for the region estimated using discriminating dose bioassay 14. The impact of changing SMC is assessed using the same method as outlined previously 33, either halting SMC, maintaining the existing age range of 0-5 year olds or extending the upper age limit to 10 or 15 years. In all simulations with SMC, coverage is assumed to match previous years (81% of children receiving treatment each round) 2.
Study participants or caregivers of children aged under 12 years of age provided informed consent (assent of child if aged 12-15 years) to participate in the cross-sectional survey. Ethical approval for this study was provided by the Burkina Faso Health Research Ethics Committee (Deliberation No 2016-12-137), CNRFP Institutional Bioethics Committee (No2016/000007/MS/SG/CNRFP/CIB), Durham University’s Department of Biosciences Ethics Committee (SBBS/EC/MIRA) and Liverpool School of Tropical Medicine Ethical Committee (Protocol number: 16/047). The study was conducted in compliance with principles set out by the International Conference on Harmonization Good Clinical Practice, the Declaration of Helsinki and the regulatory requirements of Burkina Faso.