Listening to the community: Using formative research to strengthen maternity waiting homes in Zambia

Background The WHO recommends maternity waiting homes (MWH) as one intervention to improve maternal and newborn health. However, persistent structural, cultural and financial barriers in their design and implementation have resulted in mixed success in both their uptake and utilization. Guidance is needed on how to design a MWH intervention that is acceptable and sustainable. Using formative research and guided by a sustainability framework for health programs, we systematically collected data from key stakeholders and potential users in order to design a MWH intervention in Zambia that could overcome multi-dimensional barriers to accessing facility delivery, be acceptable to the community and be financially and operationally sustainable. Methods and findings We used a concurrent triangulation study design and mixed methods. We used free listing to gather input from a total of 167 randomly sampled women who were pregnant or had a child under the age of two (n = 59), men with a child under the age of two (n = 53), and community elders (n = 55) living in the catchment areas of four rural health facilities in Zambia. We conducted 17 focus group discussions (n = 135) among a purposive sample of pregnant women (n = 33), mothers-in-law (n = 32), traditional birth attendants or community maternal health promoters (n = 38), and men with a child under two (n = 32). We administered 38 semi-structured interviews with key informants who were identified by free list respondents as having a stake in the condition and use of MWHs. Lastly, we projected fixed and variable recurrent costs for operating a MWH. Respondents most frequently mentioned distance, roads, transport, and the quality of MWHs and health facilities as the major problems facing pregnant women in their communities. They also cited inadequate advanced planning for delivery and the lack of access to delivery supplies and baby clothes as other problems. Respondents identified the main problems of MWHs specifically as over-crowding, poor infrastructure, lack of amenities, safety concerns, and cultural issues. To support operational sustainability, community members were willing to participate on oversight committees and contribute labor. The annual fixed recurrent cost per 10-bed MWH was estimated as USD543, though providing food and charcoal added another $3,000USD. Respondents identified water pumps, an agriculture shop, a shop for baby clothes and general goods, and grinding mills as needs in their communities that could potentially be linked with an MWH for financial sustainability. Conclusions Findings informed the development of an intervention model for renovating existing MWH or constructing new MWH that meets community standards of safety, comfort and services offered and is aligned with government policies related to facility construction, ownership, and access to health services. The basic strategies of the new MWH model include improving community acceptability, strengthening governance and accountability, and building upon existing efforts to foster financial and operational sustainability. The proposed model addresses the problems cited by our respondents and challenges to MWHs identified by in previous studies and elicits opportunities for social enterprises that could serve the dual purpose of meeting a community need and generating revenue for the MWH. Our study was conducted in the contiguous districts of Choma and Kalomo in Southern Province, Zambia. At the time of data collection, Choma District included what is now the administrative district of Pemba, and Kalomo District included what is now the administrative district of Zimba. Choma and Kalomo Districts are primarily rural with some peri-urban areas. Choma District has a population of 247,860 [28], with approximately 11,830 births per year [29]. Choma District has 33 rural health centers (RHC), three hospitals, and eight health posts [30]. Kalomo District has a population of 258,570 [28], with approximately 13,737 births per year [29]. Kalomo District has 31 RHC, two hospitals, and three health posts (30). Deliveries typically happen at the rural health centers, with referrals to hospitals when necessary. Twenty-eight of the health facilities in Choma and Kalomo Districts had an existing MWH structure, while three had no physical structure but the health facility staff allowed pregnant women to sleep in the wards at night [31]. In general, the existing MWHs had poor infrastructure with few amenities such as beds or mosquito nets [31]. Over a period of 5 months in 2013–2014, we employed mixed methods, using a concurrent triangulation design wherein methods were applied at the same time to confirm and cross-validate the findings [32]. The approach included gathering community input, engaging key stakeholders, and creating cost projections for operating a MWH. The sample size for each qualitative data collection method was estimated to ensure we reached saturation or predictability, the point at which no new themes or issues emerge [33]. In the catchment areas of four randomly selected facilities with existing MWHs, we approached every nth household (dependent on catchment size) to gather community input from a total of 167 randomly sampled women who were pregnant or had a child under the age of two, men with a child under the age of two, and community elders all from three distances from the health facilities: within 5 km, between 5 and 10 km, and greater than 10 km. We initially estimated a sample of 120–200 respondents, and reached saturation after 167. After administering a short survey, we used free listing (FL) [34] (S1 File), in which each respondent generated an exhaustive list of responses to the following broad, open-ended questions: 1) What are the biggest problems for pregnant women through delivery in your community?; 2) What do people in your community know or believe about MWHs at health facilities?; and 3) What businesses or services are needed but not currently available in your community? This third question was used to explore possible revenue generating activities that could help support the MWH financially. FL results were used to inform the development of the focus group discussion (FGD) and key informant interview (KII) guides, and generate a sample of key informants (S2 File and S3 File). In the same catchment areas, we also conducted 17 FGDs among a purposive sample of pregnant women, mothers-in-law, traditional birth attendants (TBAs) or community members who were part of a cadre of trained maternal health promoters called Safe Motherhood Action Groups (SMAGs) [35], and men with children under the age of two. FGDs captured information on perceptions of place of delivery, barriers and facilitators to access, and perceptions of MWHs. We conducted 38 semi-structured interviews with key informants who were recommended most frequently by the FL respondents, and who were considered to have a stake in improvement in the condition and use of MWH in their communities. These key informants included health facility staff, headmen, and TBAs, among others. KIIs were used to more deeply explore the ideas that emerged from the FL. We collected cost information from government and private sector records to estimate the fixed and variable recurrent costs of operating a basic, functional MWH, and to inform a cost recovery plan (S4 File). To inform a MWH financial sustainability strategy, we captured willingness to pay among all respondents; detailed results are presented elsewhere [36]. We anticipated that the communities might be able to operate a small business or income generating activity (IGA), using the profits to contribute to the cost of operating the MWH. The cost-volume-profit analysis included only the recurrent costs per month, assuming that any MWH major renovations or construction would be funded entirely by an outside source. We included the cost of the specific components detailed by respondents as necessary for good quality (e.g., bed linens, locks, etc.). Nine local data collectors fluent in English and Tonga, the local language, attended a 5-day training in research ethics, research methods, and quantitative and qualitative interviewing techniques immediately before data collection. FL questions were designed to elicit perspectives on problems that pregnant women face in the community, community perceptions of MWHs, and ideas for sustainability. FGD and KII guides were initially developed based on a review of the maternal health literature, then refined based on themes that emerged during the analysis of FL results. Socio-demographic characteristics of all respondents were measured. Costs were collected from expenditure records or direct quotations from vendors. Socio-demographic data were captured in Microsoft® Excel and analyzed in SAS v9.1.3 [37]. FL responses were captured on paper and analyzed nightly using pile sorting wherein responses from each participant were written on individual cards, shuffled, and grouped by data collectors into piles of similar constructs to detect emerging themes [34,38] (S1 Dataset). KII and FGD transcripts were coded for themes and the themes were linked into a theoretical model guided by the Three Delay model and the sustainability framework [1,39]. Qualitative data were translated and transcribed into Microsoft® Word and coded and analyzed in NVivo v10 [40]. We performed a cost-volume-profit analysis to determine the fixed and variable costs and revenue needed to function at various levels of activity [41]. We then conducted break-even analyses based on alternative assumptions and potential revenue stream identified by the respondents. Cost data were captured in Microsoft® Excel. Ethical approval was obtained from the Boston University Institutional Review Board (IRB) and the ERES Converge IRB in Zambia. Prior to data collection, we secured letters of support from the Ministry of Health at the national, provincial and district levels. We also had support and approval from the four Chiefs (traditional leaders) overseeing the catchment areas where we had planned research activities. We obtained verbal informed consent for each participant.