Study site
The study was conducted in Banfora Health District, in the Cascades Region, south-west Burkina Faso (Figure 2). This is an area of Sudanian savannah covering 6,295 km2 with an estimated population of 407,073 inhabitants [13]. Malaria transmission is intense and seasonal, occurring mainly during the rainy season, from May to November [19]. Plasmodium falciparum accounts for 90% of cases [19]. The main malaria vectors are An. gambiae s.s. and An. coluzzii [20]. In 2016, approximately 1 year before this study took place, a universal coverage campaign distributed ITNs with permethrin or deltamethrin (Sumitomo Chemical, Vestergaard and BASF) at a rate of one net for every two people at risk. No additional ITNs were distributed by the study. No indoor residual spraying was conducted. Typically, single room houses each housing a family are organised into compounds, consisting of on average 4 houses, led by a compound head.
Study design
The study was nested in a cohort study of risk factors for P. falciparum infection in children aged five to 15 years [21]. This study reports on the household and environmental risk factors associated with the density of An. gambiaes.l. in the children’s sleeping room over six months of high malaria transmission from July to December 2017.
Recruitment of study cohort
Sampling and recruitment of the study cohort is described elsewhere [22]. In brief, a random sample of 10 villages were selected from a list of villages in the study area using a two-stage process. Firstly, five health centres in the study area were selected, each with a catchment radius of 10 km. Secondly, two villages, at least 3 km apart, were selected from each catchment area. An enumerated list of children in the study villages was obtained from the Banfora Demographic and Health Surveillance System. From each village, a random sample of 30 children aged five to 15 years were chosen. Each child was selected from a separate house, and, where possible, a separate compound. Children were included in the study if they were of the appropriate age, were likely to remain resident in the village over the duration of the transmission season and the caregiver provided informed consent to participate in the study. 252 children who were successfully cleared of P. falciparum infection were included in the cohort study and this current study reports on the entomological surveillance from the children’s sleeping rooms.
Entomological surveillance
CDC light traps (John Hock, Gainsville, USA) were used to estimate indoor mosquito densities in the study child’s sleeping room. These traps were placed with the bulb 1500 mm above the floor, approximately 500 mm from the foot end of a bed with an ITN occupied by the study child. Houses were sampled from 19.00 h to 06.00 h every four weeks. Typically, houses were sampled once a month, but in some cases two collections were performed. Two villages (Nofesso and Ouangolodougou) were inaccessible for two weeks at the start of the study period due to flooding. Mosquitoes were taken to the laboratory in cool boxes, killed by freezing at −20°C, and identified morphologically using established keys [22]. The presence of circumsporozoites protein (CSP) in An. gambiae s.l. were identified using an enzyme-linked immunosorbent assay [23] and An. gambiae s.l. females typed to species by PCR [24, 25].
Risk factor assessment
In June, a questionnaire was administered to the caregiver of the study child to collect information on ethnicity, education level and occupation of caregivers, ITN use during the previous night, use of other protective measures (e.g. insecticide knockdown spray, mosquito coils, traditional spatial repellent), number of people sleeping in the room with the child, roof, wall and floor construction of the child’s sleeping room, whether the eaves (the gap between the top of the wall) were open or closed, presence of mosquito screening and electricity supply. Information was also collected from the head of the child’s household on asset ownership and household characteristics, following standard procedures used in the Burkina Faso Malaria Indicator Survey [26]. The number and type of large domestic animals (cattle, goats, sheep, pig, dog, donkeys or horses) tethered within 5 m of the house was recorded by a fieldworker. The house was geo-located using a handheld global positioning system (GARMIN eTrex 20). Larval surveys were carried out in each village in September, during the peak of the transmission season. All water bodies within 1 km from a village were mapped, including irrigated fields, streams and ponds, puddles, and foot or hoof prints. The presence of anopheline larvae was recorded with a dipper.
Data management and statistical analysis
Data were collected on Android personal digital assistants programmed using the KoboCollect system and included drop down boxes and consistency checks to reduce data entry errors. Following cleaning, the dataset was locked and saved in Microsoft Access. The primary outcome was the number of An. gambiaes.l. collected in each child’s sleeping room per night. QGIS Geographic Information System (QGIS Development Team (2019), Open Source Geospatial Foundation Project) was used to determine distances between the child’s home and aquatic habitats. Principal component analysis (PCA) was used to calculate the socio-economic status (SES) factor score of the head of the child’s household. SES factor scores were ranked, and households divided into five equal wealth quintiles, from 1, the poorest, to 5, the least poor. The entomological inoculation rate (EIR) or estimated number of infectious bites per study child during the transmission season was calculated using the formula EIR=MaSd where Ma is the human biting rate, estimated from the arithmetic mean number of female An. gambiae s.l. caught per light trap night across the six-month transmission season, where S is the proportion of female An. gambiae s.l. found to be CSP positive by village and d is the number of days in the transmission season (n). Mean values were compared using a t-test and proportions compared using chi-squared tests. A generalised linear mixed-effect model with a negative binomial distribution, to account for overdispersion, and log link function was used to identify risk factors associated with the mean number of An. gambiaes.l. per catch night per house each month. Risk factors were selected a priori based on importance for malaria vector house entry. These were SES quintile, ITN use, use of other protective measures, number of people sleeping in the room with the child, roof, floor and wall material in the sleeping room, eaves (open or closed), electricity supply, presence of large domesticated animals within 5 m of the house and proximity of habitats positive for anopheline larvae. A random effect for study child ID number was used to account for repeated measures on the same house and village was included as a fixed effect. Following univariate analysis, each risk factor with P<0.1 was incorporated into a multivariate model which was refined through a process of backwards stepwise elimination using a likelihood ratio test. Interactions were tested between a subset of variables that were thought to be biologically relevant to explore. Means and 95% confidence intervals were calculated. Statistical analysis was carried out in Stata 15 (Statacorp, Texas, USA). The study is reported following STROBE guidelines [27].