The effects of maternity waiting homes on the health workforce and maternal health service delivery in rural Zambia: A qualitative analysis

Background: Maternity waiting homes (MWHs) are a potential strategy to address low facility delivery rates resulting from access-associated barriers in resource-limited settings. Within a cluster-randomized controlled trial testing a community-generated MWH model in rural Zambia, we qualitatively assessed how MWHs affect the health workforce and maternal health service delivery at their associated rural health centers. Methods: Four rounds of in-depth interviews with district health staff (n = 21) and health center staff (n = 73) were conducted at intervention and control sites over 24 months. We conducted a content analysis using a mixed inductive-deductive approach. Data were interpreted through the lens of the World Health Organzation Health Systems Framework. Results: Nearly all respondents expressed challenges with understaffing and overwork and reported that increasing numbers of facility-based deliveries driven by MWHs contributed substantively to their workload. Women waiting at MWHs allow staff to monitor a woman’s final stage of pregnancy and labor onset, detect complications earlier, and either more confidently manage those complications at the health center or refer to higher level care. District, intervention, and control site respondents passionately discussed this benefit over all time points, describing it as outweighing challenges of additional work associated with MWHs. Intervention site staff repeatedly discussed the benefit of MWHs in providing a space for postpartum women to wait after the first few hours of clinical observation through the first 48 h after delivery. Additionally, intervention site staff perceived the ability to observe women for longer before and after delivery allowed them to better anticipate and plan their own work, adjust their workloads and mindset accordingly, and provide better and more timely care. When understaffing and overwork were frequently discussed, this satisfaction in providing better care was a meaningful departure. Conclusions: MWHs may benefit staff at rural health centers and the health system more broadly, allowing for the provision of more timely and comprehensive obstetric care. We recommend future studies consider how MWHs impact the workforce, operations, and service delivery at their associated health facilities. Considering the limited numbers of skilled birth attendants available in rural Zambia, it is important to strategically select locations for new MWHs. Trial registration: Clinicaltrials.gov, NCT02620436. Registered December 3, 2015, https://clinicaltrials.gov/ct2/show/NCT02620436. The MWH evaluation was conducted in rural health centers in Choma, Kalomo, and Pemba Districts in Southern Province, and Nyimba District in Eastern Province. The population of these districts is primarily rural, ranging from 69% in Choma/Pemba Districts (administratively combined during the 2010 census) to 91% in Kalomo and Nyimba Districts [30]. As of publication of The 2012 Health Facilities List in Zambia, Choma/Pemba Districts had 33 rural health centers serving an average 6650 people per facility, while Kalomo District had 31 rural health centers serving on average 8700 people each; Nyimba District had 17 rural health centers serving 6200 people each [19]. Each district has one or more hospitals (level 1 or level 2), which serve as the obstetric referral center for the district [19]. Between 2012 and 2016, these four districts, in addition to others in Zambia, received the Saving Mothers, Giving Life (SMGL) project, a multi-partner collaboration that took a holistic approach to address challenges around maternal and child health through a series of both supply- and demand-side interventions [18]. SMGL interventions targeted some aspects of the quality of care provided, including training and mentorship for health center staff in BEmONC, and improving electricity, water, and referral systems [22, 31]. Twenty rural health centers were selected from all eligible facilities in the study districts based on the following criteria: (1) distance to a referral facility (< 2 h), (2) capacity to perform five out of seven BEmONC signal functions [23, 24], and (3) volume of deliveries (> 150 per year) [29]. These selection criteria were employed to ensure the facilities were capable of providing basic obstetric care before any new MWHs were constructed, which hypothetically would increase delivery volume at the facilities. The rural health centers were pair-matched on delivery volume and distance to referral facilities, then randomly assigned to the intervention or control study arms. Full details of selection criteria and randomization procedures are available in the published study protocol [29]. Ten new MWHs were constructed according to community standards identified through a formative evaluation that showed community members sought MWHs that are comfortable, safe, culturally appropriate, and sustainable [27, 28, 32]. From this formative evaluation, a core MWH model was designed that included key domains for (1) infrastructure, equipment, and supplies to make the new homes comfortable, safe, and culturally appropriate; (2) policies, management structures, and financial systems to ensure the new homes are operationally and financially sustainable; and (3) health systems linkages and services to ensure women waiting at the MWH receive clinical services at the health center as well as health education [27, 29]. The MWHs are cement buildings with one large dormitory with beds, mattresses, and bedding for women awaiting delivery and one small dormitory for postpartum women, as recommended during formative research, for a total of 14 beds per facility. The homes also have latrines, private bathing and clothes washing areas, lockable cabinets for personal items, a cooking space with available pots and utensils, and a communal verandah for relaxation or health education classes. Nine of the ten new MWHs were opened in September/October 2016; one opened in March 2017. The ten rural health centers randomized to the control group continued to operate under the “standard of care” for waiting women in the districts, which ranged significantly in quality [12]. Six sites had a community-constructed, one-room, mud-brick MWH where women slept on floors. In two sites, women slept on the health center ward floors (or beds if available) at night and waited outside during the day. One control site did not allow women to wait at the health center in preparation for or after delivery. One control site had women waiting in the wards until a new, quality MWH, similar in design to the infrastructure, equipment, and supplies domain of the core MWH model, was constructed during the course of the study. A longitudinal qualitative evaluation was conducted with rural health center staff and district health officers to assess the effect of new and existing MWHs on the health system, which study staff hypothesized would change over time as utilization and staff responsibilities changed. We sought to capture nuanced changes by conducting four rounds of in-depth interviews (IDIs) at intervention and control sites over 24 months (November 2016 through October 2018), starting a few weeks after the first intervention sites opened. Approximately one staff member from each rural health center and one to two staff members from each district health office were interviewed during each round of data collection. Respondents were purposively sampled based on convenience—while we preferred to sample the health center in-charges, district health officer or staff directly involved in the MWHs at the health center or district levels, we were not always able to do this due to time constraints and availability of respondents. While some individuals may have been interviewed at more than one time point, most IDIs were conducted with different individuals due to changes in staffing, individuals being on leave or away for programs, and availability of individuals on the days of data collection. Interview guides elicited information on the MWH strengths and challenges, their perceived impact on the health center workforce and service delivery, perceptions of MWH-associated costs, and sustainability of the MWH. Qualitative data collectors were trained in research ethics and interviewing techniques before each round of interviews. Informed consent and interviews were conducted in English, the common language of staff within the Zambian health system. Interviews lasted between 20 and 60 min and were conducted in a private space at either the health center or the district health office. Interviews were audio-recorded and transcribed verbatim into Microsoft® Word. Demographic data were collected and entered into SurveyCTO Collect Software (Dobility Inc, Cambridge, MA) on tablets, then uploaded to a secure server only accessible by relevant project staff. We used the World Health Organization’s (WHO) Health System Framework, which defines six health systems building blocks and four overall goals, to organize and contextualize how health center staff and district health officers discussed the effect of the MWHs on the operations of rural health centers and the health workers’ perceived ability to provide maternity care to their patients (Fig. ​(Fig.1)1) [33]. Based on the characteristics of the intervention, we focused on the health workforce and service delivery building blocks, through the constructs of improved quality of care and safety, with the goal of improved efficiency of the health system on a microscale within each rural health center. Theoretical framework used to guide the organization and interpretation of qualitative data. Adapted from the WHO [33] IDIs were coded and analyzed in NVivo v11 (QSR International, Doncaster, Australia). The main codes were identified a priori based on the instrument guide, and new codes were added as themes emerged. Project staff conducted a content analysis using a mixed inductive-deductive approach [34, 35]. IDIs from each study arm were analyzed at each round and then compared over time and between study arms for each respondent type to identify patterns and key themes that were related to topics covered by the interview guide. Responses that were deviant from the observed patterns and themes were investigated further by study site to provide explanations, and the research team discussed their importance to the overarching findings. No deviant responses were considered sufficiently important for inclusion in the results below. Demographic data were analyzed in SAS v9.4 (SAS Institute, Cary, NC). The clinical positions of the rural health center staff were collapsed into the following categories based on their international [1] or Zambia-specific [36] classification as a skilled birth attendant: (1) clinical officer, (2) nurse, (3) midwife, and (4) non-SBA staff (environmental health technologist, etc.). The district health staff positions were categorized based on oversight of the MWHs in the district: (1) district health officer, (2) maternal child health officer, (3) nursing officer, and (4) other officer. Some district-level respondents were “acting,” meaning they were standing in for another individual or had yet to be confirmed to their post. For the health center staff, 6.8% (n = 5) were missing employment information; length of employment was missing for one district health officer.