Individual and Contextual Factors Associated With Malaria Among Children 6–59 Months in Burkina Faso

Objective: This study aims to understand the individual and contextual factors associated with malaria among children aged 6–59 months in Burkina Faso. Methods: This cross-sectional study used secondary data extracted from the Burkina Faso Malaria Indicator Survey 2017–2018. Descriptive analysis was used to analyse socio-demographic characteristics. We performed a multilevel logistic regression model to highlight individual and contextual factors of children’s exposure to malaria. Results: Our analysis included 5,822 children aged 6–59 months. Of these, 15% had a positive rapid diagnostic test. Factors associated with malaria among children 6–59 months were age, maternal education, household wealth, rural residence, and region. The variability in malaria exposure was 16% attributable to the strata level and 23% to the primary sampling unit level. Some factors, such as the family’s socio-economic status, access to hospital care, and place of living, were positively associated withs malaria cases in children. Conclusion: The study identified some individual and contextual determinants of malaria among children aged 6–59 months in Burkina Faso. Taking them into account for the design and implementation of policies will undeniably help in the fight against malaria in Burkina Faso.

Burkina Faso is located in sub-Sahara Africa with a superficies of 272,200 Km2 and is bordered to the north and west by Mali, to the northeast by Niger, to the southeast by Benin and to the south by Togo, Ghana, and Côte d’Ivoire. It ranks 185th out of 188 countries in the 2016 Human Development Index (HDI), published by the UNDP in 2017. The country’s population is characterised by its youth. The average age of the population was 16.6 years in 2006. Children under the age of 5 and 18 represented 17% and 53% of the population, respectively. It is divided into 13 administrative regions, characterised by cultural, socio-economic, and environmental diversity Figure 1. Map of Burkina Faso with administrative regions (Malaria Indicator Survey, Burkina Faso, 2017–2018). Data from the “Enquête sur les Indicateurs du Paludisme au Burkina Faso (EIPBF 2017–2018)” were used for this study. This is a nationally representative cross-sectional survey in which data were collected by the National Institute of Statistics and Demonography (“Institut National de la Statistique et de la Démographie (INSD)”) between December 2017 and March 2018 [6, 8]. The four databases set up were household (HR), household member (PR), mother (IR) and child (KR). The data from the mother (IR) and the child (KR) databases were aggregated for our study. MIS aim to provide quality data to assess the progress of goals and targets necessary for effective monitoring and evaluation of NMCP meaurement implementation. Specifically, these surveys aim to assess insecticide-treated net (ITNs) ownership and use, coverage of the sporadic preventive care programme for pregnant women, and treatment-seeking behaviours. Additionally, it is an assessment of awareness, behaviours and behavioural indicators related to malaria control. Depending on the country’s needs, MIS can also identify the factors related to malaria and anaemia. In addition to these items, several other questions are asked about basic demographics and education. A two-stage stratified cluster sample was used to determine the study sample. The primary sampling unit is the enumeration area (EA). The sampling was described in detail in the survey report (8). Each area was separated into urban and rural parts to form the sampling strata and the sample was drawn independently in each stratum. In total twenty-six sampling strata were created. In the first stage, 252 EAs (52 urban, 200 rural) were drawn with probability proportional to size, where size is the number of households in the EA during the mapping exercise for the 2006 census. In the second stage, from each of the EAs selected in the first stage, 26 households were selected (a total of 1,352 in urban areas and 5,500 in rural areas) that best represent the cultural and socio-economic diversity of the country as well as regional differences in malaria prevalence with a systematic equal probability draw from newly established lists at the time of enumeration. The sample size was calculated to provide statistically representative results on malaria prevalence in children aged 6–59 months [8, 9]. For this study, we examined 5,822 children aged 6–59 years who had a febrile episode during the two weeks preceding the survey and for whom the result of the malaria test (RDT) was available. The response variable in our study is the result of the rapid malaria test (RDT) performed in children aged 6–59 months. The test result is coded “Positive” for a positive test for Plasmodium falciparum (Pf) and “Negative” if it is not. Laboratory microscopy on blood smears and thickened drops was done for three-quarters of the households where RDTs were performed. Malaria results were also classified as positive or negative. There was a strong positive correlation of 0.581 (95% CI: 0.57–0.60; p < 0.0001) between these two test results. However, the laboratory microscopy test was performed primarily as a confirmation test for the RDT [9]. Thus, a malaria case was determined by a positive RDT with fever or a history of fever in the previous two weeks. The explanatory variables considered in our study were identified from the literature data [3, 7, 10–12]. The databases of children under five (KR) and household members (PR) were merged using a common primary key for both databases. The variables are divided into three groups, individual characteristics, household-level factors, and contextual factors [7, 9–14]. A descriptive analysis using frequencies was used to establish the distribution of malaria status among children aged 6–59 months in Burkina Faso with the explanatory variables considered in our study. The explanatory variables were subjected to bivariate analyses to estimate the significance of their association with malaria (Table 2). The chi-square test with the second-order correction of Rao and Scott was used to compare proportions [15, 16]. Cross-tabulation of malaria status with child, parent, community, and administrative area predictors (Malaria Indicator Survey, Burkina Faso, 2017–2018). Multilevel logistic regression was performed to identify individual and contextual effects. The hierarchical nature of the 2017–2018 EIPBF data easily allows the use of multilevel logistic regression models [17, 18]. Variables significant at the 20% level were retained for multilevel modelling. Despite non-significance at the 20% threshold, considering the data in the literature, some variables were retained for the following modelling [19]. Three multilevel logistic models were considered. The principle of parsimony was followed. A likelihood ratio test was performed to determine the most appropriate model [18, 20]. Measures of association (i.e., fixed effects) were described using an adjusted odds ratio (AOR) with corresponding p-values and 95% confidence intervals (CIs). Measures of variation (i.e., random effects) were captured using the intra-class correlation (ICC). The ICC represents the proportion of the total variation in the dependent variable attributable to the contexts (strata or EAs). This coefficient will be 0 when there is no variance between the groups. In our model (three-level model), we identified two ICCs: one concerning children nested at the strata level and the groups at the strata level nested in the group at the administrative zone level. Therefore: From Eqs. 1, 2, the variance between EAs, is the variance between strata, and ≃ 3.29 is the variance between children/individuals with scale factor one for logistic regression [17, 18, 21]. The values of the CCIs help to establish the need for multilevel analysis versus single-level analysis. The rule of thumb could be: when the ICC is less than 5% in the null model, hierarchical modelling may not be necessary [22]. All analyses were performed using R software version 4.0.5. The “svydesign” command in the survey extension was used to adjust for under- and over-reporting in the survey, using a weighting factor of (v005/1000000), where v005 is the sample weight. This study is based on the analysis of secondary data without the use of information about the identity of the participant. All DHS were approved by ICF International and a national ethics committee in each host country. All participants gave written informed consent before taking part in the survey. Although, additional ethical approval was not required in this study, we obtained written permission from the DHS programme to use the data.

N/A

• Dima Health