Setting
Burkina Faso is a West African country with an estimated population of approximately 20 million inhabitants and a surface area of 272,967 km2 [21]. Demographic projections show that the population is growing at a rate of 3.1% per year and would reach 28.2 million by 2030(20). This low-income country has an estimated gross domestic product of 786.895 US dollars, and approximately 40.1% of the entire population lives below the poverty threshold [22]. Children under 5 years of age represent 18% of the entire population, and more than 1 out of every 9 children die before the age of 5 years from preventable febrile illnesses such as malaria, pneumonia and diarrhoeal infections [19, 23]. Moreover only 50% of children have access to formal healthcare in Burkina Faso [19].
Data sources
A secondary data analysis was conducted on three nationally representative surveys from Burkina Faso. A before-after study design was used to compare socioeconomic inequalities among under-5 children with febrile illness seeking care before and after the implementation of the FHCP in Burkina Faso. Data were collected from the Burkina Faso 2010 Demographic and Health Survey (DHS) and the 2014 Malaria Indicator Surveys (MIS) before the implementation of the FHCP, and the 2017–18 MIS after the FHCP implementation. The DHS was only available for 2010; hence, the use of the MIS for 2014 and 2017–18 as the required indicators were available. All surveys used a stratified cluster sampling method for data collection. Overall, 547, 252, and 245 enumeration areas or clusters were selected for the 2010, 2014 and 2017–2018 surveys, respectively. Detailed information regarding the surveys and data collection process has been described previously [21, 24, 25]. This study focused on under-5 children with febrile illness during the two weeks prior to the interview.
Variables
Outcome measure
The outcome variable was care-seeking, which was the proportion of children under five years with fever in the two weeks before the survey, whose caregivers sought treatment in public or private health centres (dichotomous variable, 0 = no and 1 = yes). The healthcare-seeking behaviour was used as a proxy measure of health service utilization (Fig. 1).
Independent measures
The independent variables were selected based on the theoretical framework and similar studies [26, 27].
Socioeconomic variable
The household wealth index was used as the main socioeconomic variable in this study. The wealth index is a composite indicator of inequalities in household characteristics and was categorized (richest, richer, middle, poor, and poorest) based on the household’s ownership of consumer goods; dwelling characteristics; type of drinking water source; toilet facilities; and other characteristics that relate to a household’s socioeconomic status. The detailed construction and explanation of this variable have been previously described [21, 24, 25].
Other variables
The other independent variables constituted mainly sociodemographic variables and determinants of healthcare-seeking for children with fever (biological factors and social structure). Mothers’ ages were re-categorized (15–24 years for younger mothers, 25–34 for middle-aged mothers, and 35–49 for older mothers); likewise, mothers’ education levels into three categories (no education = 0, primary = 1, secondary or more = 2). The age of the child (< 12; 12–35; 36–59 months), head of the household (15–24; 25–34; ≥ 35 years), gender of the household head (male = 0 and female = 1), and religion (Muslim = 1, Christian = 2) were also selected. Other contextual factors were also included, namely, the total number of children (1–2 children = 1, 3–4 children = 2, ≥ 5 children = 3), place of residence (1 = urban, 2 = rural), region (Boucle de Mouhoun, Cascades, Centre, Centre-east, Centre-north, Centre-west, Centre-south, East, Hauts basins, North, Central Plateau, Sahel, South-west), and place of treatment (public, private medical, or other sectors).
Statistical methods
Statistical analyses were performed using household data with children under five who had experienced an episode of fever two weeks prior to the interview. To summarise the socioeconomic and demographic characteristics of the caregivers and children, descriptive analyses were performed and presented as percentages. To compare the characteristics between the three datasets, the Pearson chi-squared test was used. The analyses were weighted for probability sampling and adjusted for stratification and clustering.
Measuring socioeconomic inequalities
Socioeconomic inequalities were measured using concentration curves and the concentration index (CI) at the country level (before and after the implementation of the FHCP), which are well-illustrated tools to quantify the socioeconomic inequalities among health-state indicators. At the regional level, CI was used to assess socioeconomic inequalities in the use of healthcare for children under 5.
Concentration curves
The curves (Lorenz curve) represent the cumulative proportion of ‘treatment-seeking in case of fever for children under 5 years of age’ in relation to the cumulative percentage of ‘wealth’. When computing the cumulative percentage, the wealth quintile variable was ranked from the lowest to the highest. If the dependent variable is evenly distributed, the curve runs diagonally from the bottom left corner to the top right corner (45° line), referred to as the ‘equality line’. In contrast, if the treatment-seeking for fever among children under 5 years is concentrated in the poor, the concentration curve will lie above the line of equality [28].
Concentration index
This was used to describe the magnitude of inequalities. The index is defined as twice the area between the concentration curve and the equality line (the 45° line) and is represented by the formula below:
$$C=\frac{2}{\mu }cov\left(h,r\right),$$
where \(h\) represents healthcare seeking, μ represents its mean, \(r\) is the fractional rank of an individual in the wealth index distribution, and \(cov\) is the covariance between care seeking and the fractional rank of the wealth index [28]. The index varies between -1 and 1, and if the CI is equal to 0, there is no socioeconomic inequality. If the result is positive, the dependent variable is more concentrated among the rich, and the concentration curve will be below the line of equality. In contrast, a negative value means that the dependent variable is more concentrated among the poor, and the concentration curve is above the line of equality [28]. The bounds of the C of a binary health indicator depend on its mean.
As the health indicator varies according to the study period, Erreygers’ normalisation option was selected in STATA. Maps were constructed using QGIS 3.12 software to show the concentration indices for healthcare seeking for under-5 children at the regional level. All analyses were performed using STATA® 14 with a significance level of 0.05.