Menu
Menu
Menu
Menu
Background: Poor and marginalized segments of society often display the worst health status due to limited access to health enhancing interventions. It follows that in order to enhance the health status of entire populations, inequities in access to health care services need to be addressed as an inherent element of any effort targeting Universal Health Coverage. In line with this observation and the need to generate evidence on the equity status quo in sub-Saharan Africa, we assessed the magnitude of the inequities and their determinants in coverage of maternal health services in Burkina Faso. Methods: We assessed coverage for three basic maternal care services (at least four antenatal care visits, facility-based delivery, and at least one postnatal care visit) using data from a cross-sectional household survey including a total of 6655 mostly rural, poor women who had completed a pregnancy in the 24 months prior to the survey date. We assessed equity along the dimensions of household wealth, distance to the health facility, and literacy using both simple comparative measures and concentration indices. We also ran hierarchical random effects regression to confirm the presence or absence of inequities due to household wealth, distance, and literacy, while controlling for potential confounders. Results: Coverage of facility based delivery was high (89%), but suboptimal for at least four antenatal care visits (44%) and one postnatal care visit (53%). We detected inequities along the dimensions of household wealth, literacy and distance. Service coverage was higher among the least poor, those who were literate, and those living closer to a health facility. We detected a significant positive association between household wealth and all outcome variables, and a positive association between literacy and facility-based delivery. We detected a negative association between living farther away from the catchment facility and all outcome variables. Conclusion: Existing inequities in maternal health services in Burkina Faso are likely going to jeopardize the achievement of Universal Health Coverage. It is important that policy makers continue to strengthen and monitor the implementation of strategies that promote proportionate universalism and forge multi-sectoral approach in dealing with social determinants of inequities in maternal health services coverage.
Burkina Faso is a landlocked, francophone country in West Africa demarcated into 13 regions with 63 districts. In 2016, average life expectancy was estimated at 53.4 years for men and 57.6 years for women. In 2016, about 70% of the population were estimated to live in rural areas and only 36% to be literate [36]. Poverty is widespread, with about 41.1% of the population living below the national poverty line of US$1.90 a day [36]. The public health system in Burkina Faso is organized along three levels: primary level (named Centre de Santé et de Promotion Sociale – CSPS) in rural areas some urban areas, secondary level in district capitals and tertiary referral level in regional capitals and in Ouagadougou, the capital city. Total expenditure on health was at 5% of GDP in 2014 [37]. As noted earlier, Burkina Faso has made access to maternal and child health services one of its key policy objectives. It has done so by introducing a series of reforms, first to reduce (in 2007) and then to remove (in 2016) user fees for maternal care services [38–40]. While it is still early to evaluate the impact of the 2016 policy, evaluations of the 2007 user fee reduction policy indicate equity-neutral increases in health service utilization and decreases in out-of-pocket expenditure [41]. This is to say that the 2007 user fee reduction policy neither increased nor decreased existing gaps in service utilization between socio-economic strata, but kept them constant by improving access to care and financial protection across all socio-economic strata. This study used data from a cross-sectional household survey conducted as part of the baseline assessment of the impact evaluation of a performance-based financing pilot intervention launched in Burkina Faso in 2014. Data was collected between October 2013 and March 2014 in 24 districts (38% of all districts in the country) on a mainly rural population (91.8%) – hence after the 2007 user fee reduction policy, but before the 2016 user fee removal policy. Sampling followed a three-stage clustering procedure. First, clusters were defined according to the catchment areas of 561 primary health facilities in the 24 districts. Second, one village was randomly selected from each cluster. Third, for each sampled village, teams of interviewers drafted a comprehensive list of all households with at least one woman who was either pregnant at the time of the visit or had completed a pregnancy in the prior 24 months. Subsequently, 15 households were to be randomly selected from the list for inclusion in the survey. The final sample included 7844 households, somewhat less than intended as it was not always possible to identify 15 eligible households per sampled village. For this study, we focus on the sub-sample of the 6655 mostly rural and poor women with a completed pregnancy in the prior 24 months residing in the sampled households [42]. However, each of the three outcome variables (i.e.: at least four ANC visits, facility-based delivery and at least one postnatal care) had a different sample size. This was due to the following reasons: i) some women included in the main sample had incomplete pregnancies such as abortions and miscarriages and hence did not attend any of the three maternal health services, and ii) some women attended only one service and not the other services along the continuum of maternal care, and iii) some women might have been missed due to either not completing giving responses to the questionnaire as the survey progressed or due to interviewer mistakes. The survey questionnaire assessed households’ and women’s socio-demographic characteristics, and their use of essential maternal health care services during pregnancy (ANC, facility-based delivery, and postnatal care). Table 1 provides an overview of all variables included in our analysis, their measurement, and their distribution in the sample. Variables considered in the analyses, their measurement, and distribution in the sample aOutcome variables not adding up to 100% due to missing values as the survey proceeded from one section to the other bANC = Antenatal care; cPNC = Postnatal care; dPNC1 = at least one postnatal care visit We defined our primary outcomes to capture coverage along the maternal care continuum, hence we included: a. having attended at least four antenatal care visits (ANC4+); b. having had a facility-based delivery, as a proxy measure for skilled attendance at delivery [43, 44]; and c. having attended at least one postnatal care visit (PNC1) within six weeks after birth. Equity is defined as the absence of systematic disparities in health, its social determinants, and/or in health service utilization between more or less disadvantaged social groups [7, 45]. Inequities exist in the presence of disparities or determinants that are deemed avoidable, unfair and unjust [7]. In this study, equity refers to equal utilization of health services given equal need for such services [7], and when “need” is defined as the capacity to benefit from any service along the continuum of maternal health care. The literature recognizes multiple dimensions to equity in health service utilization, such as gender, wealth, education, place of residence/geographic region, ethnicity, age, migratory status, religion, occupation, indigenous status, or sexual orientation [12, 24, 46]. In this study, we investigated inequities in maternal health service coverage along three equity dimensions: i) household wealth; ii) woman’s education (measured in relation to literacy); and iii) distance to catchment primary health facility. Household wealth was selected, as overcoming socio-economic inequities in maternal health service coverage remains a top priority for the government and also a priority for the achievement of the SDGs [28, 29, 47]. In order to measure the household wealth, a wealth index using assets and living conditions was developed using the multiple correspondence analysis (MCA) [48]. The following variables were used to compute the household wealth index: housing (type of building materials, number of rooms, water and energy supply sources), assets (TV, radio, fridge etc.), house and fields owned, and animals. After calculation of wealth scores, households were split into quintiles from the poorest (Q1) to least poor (Q5). Distance and literacy were chosen because they were identified as important barriers to access by earlier studies [22, 33, 34]. Distance was dichotomized so as to reflect the World Health Organization standard of having a primary health facility within a radius of 5 km as well as those living outside this recommended World Health Organization 5 km radius standard. Literacy was used instead of education: although this is recommended for equity analysis, in our sample, less than 1% of the respondents had formal education. Based on Andersen’s behavioral model [49], we included in our analysis a number of additional explanatory variables that were available in our data set as potential relevant confounders. These included: woman’s marital status, age, parity, and religion, as well as household size. Our analysis proceeded in steps. First, we looked at the differences in coverage for each of the three outcome variables by districts, and explanatory variables through descriptive bivariate statistics. The chi-square test was used to identify significant associations between the outcomes of interest and selected explanatory variables. Second, to measure equity, we used simple comparative rates/measures of coverage for two groups [12, 50]. Simple comparative rates/measures draw on data from two subgroups and include differences and ratios to demonstrate absolute and relative inequalities, respectively [12, 46, 50, 51]. The absolute gap for socio-economic inequity was computed by subtracting the outcome of the first quintile from that of the fifth quintile (Q5-Q1) of the respective outcome variable. The ratio of socio-economic inequity was established by dividing the outcome of the fifth quintile to that of the first quintile (Q5/Q1), respectively. Because distance and literacy were binary variables, we computed inequity gap and ratio in the same way as with the continuous variable. It is important to note that absolute measures provide an idea of the actual gap that exists between groups and thus the required effort to close them while relative measures provide an insight into the degree of unfairness between groups [52]. To correct for the weaknesses of the simple comparative rates, especially for socio-economic position (since they only take into account the two extreme groups, leaving out other groups in the middle [53]), we used concentration indices, which are estimated using concentration curves, to draw on data from more than two subgroups [12, 46, 51, 54]. Concentration curves provide a graphical display of the share of health or health services accounted for by cumulative proportions of individuals in a population ranked from poorest to richest at a given point in time [51, 55]. A concentration curve that lies below the line of equality (45 degrees) signifies presence of inequality favouring the rich, while a curve that lies above the equality line signifies presence of inequality favouring the poor. When it overlaps with the diagonal line (the line of equality), this implies there are no inequalities [51, 55]. Concentration indices quantify the degree of socioeconomic-related inequality in a given health or health service variable [51, 55], defined as twice the area between the concentration curve and the diagonal (line of equality) ranging between − 1 and 1. The index takes a negative value when the curve lies above the line of equality, indicating disproportionate concentration of the health or health service variable among the poor and a positive value when it lies above the line of equality, indicating disproportionate concentration of the health or health service variable among the rich and takes the value of zero when there is equality [56]. Third, we ran three separate regressions (one per outcome variable) to confirm the presence or absence of inequities due to household wealth, distance, and literacy, while controlling for all potential confounders. As such, we performed a regression analysis using a hierarchical model to allow for clustering at the district level, attempting to capture the variance in the outcome variables across districts captured by the descriptive analysis. We operationalized our random effects models using Stata version 14 (Stata Corporation, Texas, USA), defining women as first level and district as second level. Our estimated model is of the form: where of each observation ‘i’, Y is one of the 3 outcome variables ‘j’ [j = 1(ANC4+, 2(facility-based delivery), 3 (PNC1)] and X is the explanatory variables, β0jis the intercept of the respective model for outcome variable ‘j’, uj is the district-specific effects, and εij is the error term. As is the case with hierarchical models, our assumption is that each of the levels (districts) has a different (i.e. district-specific) effect uj on the outcome variables yij, which are independent of the explanatory variables xij.
N/A
