Background: Several African countries have recently reduced/removed user fees for maternal care, producing considerable increases in the utilization of delivery services. Still, across settings, a conspicuous number of women continue to deliver at home. This study explores reasons for home delivery in rural Burkina Faso, where a successful user fee reduction policy is in place since 2007. Methods: The study took place in the Nouna Health District and adopted a triangulation mixed methods design, combining quantitative and qualitative data collection and analysis methods. The quantitative component relied on use of data from the 2011 round of a panel household survey conducted on 1130 households. We collected data on utilization of delivery services from all women who had experienced a delivery in the previous twelve months and investigated factors associated with home delivery using multivariate logistic regression. The qualitative component relied on a series of open-ended interviews with 55 purposely selected households and 13 village leaders. We analyzed data using a mixture of inductive and deductive coding. Results: Of the 420 women who reported a delivery, 47 (11 %) had delivered at home. Random effect multivariate logistic regression revealed a clear, albeit not significant trend for women from a lower socio-economic status and living outside an area to deliver at home. Distance to the health facility was found to be positively significantly associated with home delivery. Qualitative findings indicated that women and their households valued facility-based delivery above home delivery, suggesting that cultural factors do not shape the decision where to deliver. Qualitative findings confirmed that geographical access, defined in relation to the condition of the roads and the high transaction costs associated with travel, and the cost-sharing fees still applied at point of use represent two major barriers to access facility-based delivery. Conclusions: Findings suggest that the current policy in Burkina Faso, as similar policies in the region, should be expanded to remove fees at point of use completely and to incorporate benefits/solutions to support the transport of women in labor to the health facility in due time.
The study took place in the Nouna Health District (NHD), north-western Burkina Faso, in 2011–2012. At the time of the study, the district had a population of approximately 311,000 distributed in 300 villages, and counted 34 first-line facilities, Centres de Santé et de Promotion Sociale (CSPS)—33 located in rural areas and one in Nouna town – and one district hospital, also located in Nouna town. The 34 CSPS were equipped and staffed as Basic Emergency Obstetric Care facilities (BEmOC) capable of managing uncomplicated deliveries, while only the district hospital was equipped and staffed as a Comprehensive Emergency Obstetric Care facility (CEmOC) capable of managing complicated deliveries, including C-sections. The government was and continues to be the exclusive provider of formal healthcare services in the area. A sub-portion of the district has been part of a Health and Demographic Surveillance System (HDSS) for over 15 years [35]. This study adopted a triangulation mixed methods design [36]. The triangulation design was chosen as the most appropriate mixed methods design since the research question focused on one single phenomenon, i.e. the decision to deliver at home vs. in a health care facility, but aimed at capturing all its possible dimensions, in such a way that neither qualitative nor quantitative methods alone could do [36]. The conceptual framework guiding the study recognized the decision to deliver at home vs. to deliver in a health care facility as the product of the interplay between access factors (acting at the household and at the community level) and individual and household knowledge, beliefs, and practices concerning labor and delivery [37, 38]. The adoption of a mixed methods design allowed us to integrate both series of elements into one single study and to address them from multiple perspectives. The quantitative dimension of the study relied on household survey data to identify socio-demographic, economic, and health system factors associated with the decision to deliver at home. The qualitative dimension of the study relied on a series of in-depth interviews to explore knowledge, beliefs, and practices concerning labor and delivery. Data for the quantitative and the qualitative study components were obtained through two sequential data collection activities which relied on independent data collection tools. Data analysis occurred separately for the two components. At the end, we pooled together quantitative and qualitative findings to derive a final interpretation of the material collected in relation to the over-arching research question. Recognizing that decisions concerning delivery are made collectively [39, 40], the study targeted women with a recent history of delivery, their households, and the community leaders of the villages where these women resided. We used data from the 2011 round of a panel household survey conducted in the region since 2006. Data used for this study represents a sub-set of a survey originally designed to monitor progress towards coverage with malaria control interventions and access to maternal care services. Data from previous rounds of the survey has already been used to produce a number of evaluations, including three pertaining to maternal care services [30, 32, 41]. The survey sampling procedures have been described in detail elsewhere [42]. In 2011, data was collected between February and March from a total of 1130 households, selected using a three-stage cluster sampling procedure. First, clusters were defined according to the catchment area of each CSPS (clusters were defined according to the number of CSPS present at baseline, i.e. 27). Second, two villages in each cluster were selected. Third, 20 households were randomly selected in each village, using modified EPI sampling procedures [43]. To take into account its larger population, 70 households were selected in Nouna town. The survey relied on four core modules to assess a household socio-demographic and economic profile. In the 1130 sampled households, all women who had completed a pregnancy in the twelve months prior to the interview date were administered one additional survey module, gathering information on health care seeking during pregnancy and at time of delivery. Based on a preliminary analysis of the quantitative findings, we applied maximum variation sampling to purposely identify the respondents (n = 55) for the qualitative study component [44]. We sampled 25 households where women had delivered in a health care facility; 24 households where women had delivered at home; and 6 households which had experienced both home and facility-based deliveries in the prior twelve months. These 55 households were distributed in 13 villages (out of a total of 54 included in the household survey, 24 %), which were purposely selected to display maximum variation in terms of health seeking behavior at delivery. We included villages: i) located at more than 7 km from the health care facility and with less than 80 % of all deliveries taking place in a health care facility; ii) located at more than 7 km from the health care facility and with more than 80 % of all deliveries taking place in a health care facility; iii) located within 7 km of a health care facility and with less than 80 % of all deliveries taking place in a health care facility. We purposely did not include villages located within 7 km of a health care facility and with more than 80 % of all deliveries being facility-based, because we did not see potential to explore remaining barriers to access in such settings. We set the threshold at 7 km because this was the mean distance used to define the CSPS catchment areas on a national level [33]. We set the threshold at 80 % in relation to facility-based delivery because our own data indicated that across the district, 89 % of all women delivered in a health care facility. Within each village, we purposely selected, to the extent possible, households belonging to all socio-economic strata. To do so, we relied on the same quartile classification used for the quantitative analysis. We made the explicit decision to interview both households where women had delivered at home and households where women had delivered in a health facility to be able to explore systematic differences between the two. Our aim in doing so was to understand what made it possible for certain households to overcome barriers which other households were not able to overcome. This decision allowed us to explore possible copying strategies used to overcome remaining barriers to facility-based delivery, in a context of substantial user fee reduction. In each household, we interviewed the woman having delivered, her husband, and/or any other person indicated by the woman as influential in the process of seeking care at delivery, such as the mother, the mother in law, and/or the household head (in cases where the husband was not the household head). In addition, we interviewed all thirteen village leaders in the selected villages. All interviews took place in 2012 and were conducted by trained qualitative interviewers, working under the direct supervision of the authors. The interview guide used at the household level was developed to induce respondents to recall the latest completed pregnancy in the household (most frequently the one reported in the household survey) and the decision making process which had led either to a home or to a facility-based delivery. In addition, respondents at the household level were invited to express their opinion on perceived benefits and problems associated with home vs. facility-based delivery, on remaining barriers to access, and on cultural beliefs and practices surrounding labor and delivery. The interview guide used for the village leader did not explore own experiences, but focused exclusively on this latter set of elements. All interviews were conducted in the local languages, tape-recorded, and later verbatim transcribed and translated into French. Quantitative household survey data were analysed using Stata 12 (Stata Corporation, Texas, USA). Multivariate logistic regression was used to explore the association between a theoretically relevant set of individual, household head, household, and village characteristics and the outcome variable, defined as “home delivery”. To account for the hierarchical structure of the data, i.e. women are clustered in villages, we applied random effects modelling. In our analysis, we defined women as level 1 and village as level 2. Table 1 provides a comprehensive list of all variables included in the analysis, the answer categories, their distribution in the sample, and the hypothesized coefficient sign. Given the small sample size, we estimated a relatively parsimonious model, including only the most important theoretically relevant and objectively measurable variables identified in prior studies and available in our dataset. Variables, their distribution in the study sample, and the expected coefficient sign (Observations (women) = 420; Clusters (villages) = 54) Most variables included in the analysis are self-explanatory. In line with previous research [45–47], socio-economic status was estimated by computing the total monetary value of all animals (cows, sheep, goats, donkeys, horses, pigs, and poultry) and durable assets (cart, plough, telephone, radio, television, bicycle, gas cooker, fridge, and motorbike) owned by the household. The value of each asset was set at the average market price, which was assessed though a parallel small survey carried out at major markets in the district. To compute per capita wealth estimates, we simply divided the total monetary asset value by household size (i.e. number of people living within a household). To align the quantitative and the qualitative analysis, distance to the referral CSPS was computed using 7 km as cut-off point. We included a variable to distinguish households residing in villages under health and demographic surveillance from households residing in villages beyond this area. Analysis of the qualitative material took place on the transcribed material, directly in French, using a mixture of inductive and deductive coding based on major determinants of access [3, 37]. One of the authors worked as the primary analyst, coding all transcribed material. As a source of triangulation [44], to check the consistency of the emerging interpretation, two senior authors checked the coding scheme, the coding process, and independently read two different sub-sets of the transcripts. We translated into English only the citations used in this manuscript. The interpretation of the findings as presented in this paper is based on the joint appraisal of the quantitative and qualitative findings. The process of bringing together into major access dimensions [3, 37] findings from the two study components was managed at the end of the two distinguished and parallel analytical approaches. Institutional ethical review of the study protocol was obtained from University of Heidelberg, Germany and from the Ethical Board of the CRSN, Nouna, Burkina Faso. Oral consent was obtained from all study participants separately for the quantitative survey and the qualitative interviews.
N/A