Reducing the medical cost of deliveries in Burkina Faso is good for everyone, including the poor

Since 2007, Burkina Faso has subsidized 80% of the costs of child birth. Women are required to pay 20% (900 F CFA = 1.4 Euros), except for the indigent, who are supposed to be exempted. The objective of the policy is to increase service utilization and reduce costs for households. We analyze the efficacy of the policy and the distribution of its benefits. The study was carried out in Ouargaye district. The analysis was based on two distinct cross-sectional household surveys, conducted before (2006; n = 1170) and after (2010; n = 905) the policy, of all women who had had a vaginal delivery in a public health centre. Medical expenses for delivery decreased from a median of 4,060 F CFA in 2006 to 900 F CFA in 2010 (p&0.001). There was pronounced contraction in the distribution of expenses and a reduction in interquartile range. Total expenses for delivery went from a median of 7,366 F CFA in 2006 to 4,750 F CFA in 2010 (p = 0.001). There was no exacerbation of the initial inequalities of the share in consumption after the policy. The distribution of benefits for medical expenses showed a progressive evolution. The greatest reduction in risk of excessive expenses was seen in women in the bottom quintile living less than 5 km from the health centres. Only 10% of those in the poorest quintile were exempted. The subsidy policy was more effective in Burkina Faso than in other African countries. All categories of the population benefited from this policy, including the poorest. Yet despite the subsidy, women still carry a significant cost burden; half of them pay more than they should, and few indigents are fully exempted. Efforts must still be made to reach the indigent and to reduce geographic barriers for all women. © 2012 Ridde et al.

The research was accepted by the ethics committees in Burkina Faso (Health research ethics committee) and Canada (Ethics committee of the CRCHUM). The study was conducted in a rural health district with 260,000 inhabitants (Ouargaye). Vaginal deliveries were carried out in 26 maternity units of health and social promotion centres (CSPS, first line health centers). They were attended by auxiliary midwives (recruited at primary school completion level and trained in two years) or by nurses or male midwives (recruited at high school completion level and trained in two years). Caesareans were done at the district hospital. The analysis was based on two independent cross-sectional household surveys conducted before and after the implementation of government subsidies. The first survey was conducted by another group of researchers in January 2006 as part of a broad study on maternal health (IMMPACT). First, a census of the whole population was conducted. Then, all women of reproductive age (15–49) who had delivered within the previous six weeks were interviewed (n = 1170). The 2006 survey data was never used and published. The post-intervention survey was conducted in February 2010 by the authors of this article. No survey database for households was available and we could not afford to initiate a population census. Therefore, we decided to select a sample of users, namely, women who had delivered in CSPSs. From the centres’ registers, the surveyors sought the names of the women of reproductive ages (15–49) who had delivered in the previous six weeks. Initially, 1,019 women were identified. A field survey, conducted with the help of key informants in the villages, was able to locate 90% of these women (n = 905). The study population included all women who had delivered vaginally in a CSPS. Given the complexity of factors related to caesareans [26], we focused on assisted deliveries in first-line facilities (CSPS), where the policy’s impacts were more readily observable. The same questionnaires were used in 2006 and 2010. Information gathered covered households’ demographic and socio-economic characteristics and expenses incurred for deliveries. All spending was recorded in order to distinguish between medical and non-medical expenses. Dual data entry was done using EpiData©. The analyses were carried out using SPSS©, STATA© and Excel© software. An index reflecting households’ socio-economic status was established by factorial analysis (CPA) based on assets surveyed. To ensure the measures’ comparability, the same assets were considered in both surveys. Household were then classified into quintiles on the basis to their assets scores. For 2010, scores were able to be calculated for only 883 of the 905 respondent women (97%). The medical expenses used in this article are those related to point-of-service fees for medical services in normal deliveries. In calculating the overall mean of expenditure, only women who had reported at least one item of health expenditure were considered. The 2006 healthcare expenditures were adjusted for inflation to be comparable to those of 2010. The effects of the subsidy measures on healthcare spending were studied using non-parametric tests (testing the medians). The impact on the protection of households was based on measuring the change in exposure to risk of excessive expenses. The risk was measured using a methodology based on spending distribution and the breakdown of extreme values pre- and post-subsidy [27]. The process was based on Tukey’s outlier method [28]. Let one person “i”, member of a group “j”, express a given level of access or need. The health cost is considered excessive if it exceeds a threshold value Sij calculated from the interquartile distance, in the group under consideration, i.e., Sj = Q3j+k* (Q3j−Q1j). Q1j = the boundary defining the first quartile, Q3j = the boundary defining the third quartile. K is a constant whose standard value is 1, and whose value is modified in sensitivity analyses. Let P1 j be the prevalence of extreme values of group j in 2006. For a given value of k, this prevalence describes the size of the population at risk for excessive costs. P1 j is sensitive to values of k. If the evaluator selects a lower value for k (e.g. k = 0.5), the value of the threshold Sj is reduced and consequently P1 j increases. Comparing the prevalence of the extreme values before and after the subsidy provides the density of people having moved outside the risk zone defined by the 2006 threshold. Upon comparing P1 j, the prevalence of the extreme values of group j in 2006, to P2 j, that observed in 2010, the distance between them shows the increase in financial protection. To be comparable, these prevalences must be determined on the basis of a common reference standard: the excessive expenses threshold of 2006. The calculation of the increase is carried out for different k values. The analyses are considered to be robust if the relative gains in the prevalence of excessive expenses are comparable, regardless of the k value selected. The financial constraints that households encounter in accessing care vary according to their ability to pay and costs incurred for transportation related to the delivery. Consequently, if the intervention can be expected to reduce the risk of excessive expenses in general, our hypothesis is that this reduction should primarily benefit the poorest populations and those living closest to health facilities. For this reason, our analysis evaluated the intervention’s impacts after stratification based on level of poverty and distance between the residence and the nearest health facility (groups identified by subscript j). As the gain for a given group becomes higher, the subsidy policy’s capacity to protect households in this group from the risk of excessive expenses also increases.

N/A

• Dima Health