Toward communities as systems: a sequential mixed methods study to understand factors enabling implementation of a skilled birth attendance intervention in Nampula Province, Mozambique

Mozambique, located in southeastern Africa, is a country in which the majority of its 26.4 million population relies on the public health system for health services [12, 13]. Having moved from a GDP of 5.02 USD Billion in 2000 to 11.01 in 2016, the country has seen improvement in its economic situation in recent years, though in 2009 more than half of the population lived below the poverty line [14]. Maternal health improvement initiatives have been funded by various donors, necessarily engaging all four levels of care in the Mozambican health system [15, 16]. The first point of access to labor and delivery services for the majority of the country’s 70% rural population is through type 2 health centers [11, 17]. These centers are typically staffed with one MCH nurse and up to three maternity beds. Type 2 facilities address complications and need for comprehensive emergency obstetric care through referral to hospitals, which are typically located in district capitals [17]. Type 2 facilities provide Basic Emergency Obstetric Care, with every 2.3 type 2 facilities serving a catchment population of 500,000 on average [17]. This study looks at maternal health related service delivery in two type 2 facilities in the country’s most populous province: Nampula, located in northern Mozambique [11].

We adopted a sequential explanatory mixed methods case study design in which quantitative methods are followed by qualitative methods to explain the results derived from the quantitative phase [18‐20]. To ensure thorough reporting of our approach, we describe it here in line with the Mixed Methods Appraisal Tool (MMAT), created in 2014 to support appropriate planning, execution, and appraisal of mixed methods research [20, 21].

Drawing on the positive case study approach, our design identified primary health facilities (herein referred to as “sites”) where clinic data reflected an increase in institutional deliveries, and then constructed the “story” of this increase [22]. We employed three phases: 1) Quantitative analysis of service delivery data to select two positive cases, 2) Integrated analysis of quantitative service delivery data and qualitative data via expert panels to understand the intervention’s implementation, and 3) In-depth interviews with key informants to understand the contextual factors contributing to the increase in institutional deliveries (Table 2).

In 2009, Damschroder et al. conducted a systematic review of 19 articles, themselves comprising over 500 studies of the factors found to influence the implementation quality of evidence-based interventions [24]. Given our interest in explaining the factors that may have contributed to the observed increase in institutional deliveries in the two sites, we determined the authors’ resulting “Consolidated Framework for Implementation Research” (CFIR) as appropriate to our aim and applied it in the data collection and analysis phases of the study, discussed below.

Adaptability was discussed as a factor facilitating implementation by the majority of respondents, particularly in relation to exchange of responsibilities. When needed—most commonly, to attempt to respond to high client volume—TBAs and MCH nurses in both sites interchanged their tasks during deliveries. But actors’ relationship to adaptability and its meaning to their implementation differed between the two sites. In Upper Province, respondents discussed their exchange of roles as a complement to training and an expression of respect and mutual appreciation between TBAs and nurses. Respondents discussed adaptability in the context of mentorship to build TBAs’ skills, and accountability in their division of responsibilities. Nurses also retained the authority over how services should be delivered. As a TBA described:

“[If] TBAs find the [nurse] with a lot of work, [the nurses] allow…TBAs to assist childbirth and advise…and are there to follow if anything abnormal is happening.”—Upper Province TBA

In Lower Province, respondents’ relationship to the need for adaptability in implementation varied between stakeholder groups and was not uniformly positive. All respondents described the direct involvement of TBAs in delivery attendance, including the facility nurse who reflected,

“But now I can even leave [TBAs] alone [to] do the delivery…without any problem. …[If the delivery] is not normal, so together we…help the [laboring client].” –Lower Province Nurse

However, half of all Lower Province respondents discussed TBAs’ adaptability to attend deliveries alone primarily as a result of facility implementers’ lack of accountability to fulfill their own responsibilities. Both TBAs and beneficiaries recounted births reflecting this:

“When…the nurse came to observe…she came rather drunk and…said…to [wait]. …After a while [I] gave birth but it was the TBA who assisted.” – Lower Province Beneficiary

Nearly all respondents discussed the intervention as responsive to beneficiary needs. First, respondents discussed the intervention’s meeting of clients’ medical needs, particularly regarding unexpected complications:

“…At the facility…if you have a complicated delivery, can first apply injection and if you have problem giving delivery [of] placenta...[women] are supported. While at the community level that does not happen” --Upper Province TBA

Beneficiaries’ needs were also described as met by the intervention through introduction of TBAs as escorts to the facility, and to its added emphasis on TBAs’ and nurses’ help to women to find better laboring positions and address their emotional needs:

“I say some sweet words to her, I moralize her so that she may feel motivated saying that…this child may save her life one day, be a governor, a teacher…I say all this to all women I take to hospital. …in the past I did not…” –Lower Province TBA

Across the two sites, beneficiaries echoed this perception of a caring, respectful birth experience:

“After giving birth the TBA looked after me, she washed me and gave bath to my baby and put her on scale to weigh [her], and she took me by the hand and advised me…to try to sleep. After some time when I was still in bed she brought my child…and told me to breastfeed…and…she talked to me.” –Lower Province beneficiary

In Upper Province, nurses and TBAs shared mutual appreciation of each other’s unique contributions to meeting patients’ needs, both emotionally and medically.

“When the pregnant woman is in the delivery process [TBAs] help [the beneficiaries] morally. They…are the connect[ing] bond between the pregnant woman and the community.…The pregnant woman feels more protected.”–Facility Manager, Upper Province (01UPF)

“…Any emergency…[the nurses] step in”—Upper Province TBA

In Lower Province respondents pointedly distinguished between implementers when discussing contributions to meeting patients’ needs. Facility staff described themselves and were described by others as meeting patients’ emergency medical needs:

“I know what to give, and I call for the district, I can call the chief medical officer, I can call the director, I can call the district head of MCH, any [available] line that I can use to communicate, to say that I have a patient to evacuate.” –Lower Province Nurse

However, across respondent interviews, numerous examples were given in Lower Province in which patients’ needs went poignantly unmet. In all instances, this was discussed as a result of facility staff’s failure to uphold their role as skilled birth attendants, resulting in mistreatment:

“When I go to call her, the nurse doesn’t come…unless the situation is serious and she needs to come and to transfer the [laboring woman] to another health facility...” –Lower Province TBA

“When there is delivery complication and TBA goes to present the case to the nurse, [the nurse] normally answers to go and find out to know if the patient’s family has money. If the family doesn’t have money, the patient won’t be assisted. …the nurse says you have to go to the family to find out if they have 100 meticais” –Lower Province Beneficiary

The majority of respondents demonstrated deep knowledge of the intervention, and a belief that the intervention’s formally established linkages between community- and facility-based actors contributed to the observed increase in institutional deliveries.

The quality of implementers’ interpersonal relationships as they collaborated came up throughout both sites’ interviews, though not in uniform ways. In Upper Province, collaboration between TBAs and nurses was characterized by mutual appreciation and support:

“Both of them, both TBA and nurses always worked together. [At] no time…[was] there…someone who said, ‘Let me go to work alone...’ They have been working together. They work well.” –Upper Province Beneficiary.

By contrast, in Lower Province the quality of relationships between facility- and community-based implementers was less consistent. Though facility-based implementers described a positive working relationship, most community-based implementers portrayed a strained one. Here, a beneficiary echoes this sentiment:

“…The [TBAs] complain because…the nurses do absolutely nothing especially if the expectant mother comes at night. The nurses just remain in their houses and do not set their feet in the health centre. However, the TBA who came with you…stays by your side” –Lower Province Beneficiary.

Despite variation in the quality of interpersonal relationships, the intended division of labor was uniformly understood, as were the resulting locusts of control between implementers, which helped inform how and when hand-offs should take place:

“…The TBAs have the duty of taking the expectant mothers to hospital and hand them to the professional nurses in the maternity. …I mean they work in coordination with health centre personnel…[and, with] the TBAs…the community leaders lead the meetings and they work with us in sensitizing the people”—Upper Province TBA.

This linkage between facility and community actors was also discussed in detail when respondents described the intervention and their perception of its value. TBA and CLC respondents richly recounted the processes set in place by the intervention and the roles and reinforcing relationships between each other. Community-based implementers seemed to draw pride from their system-like way of generating and sustaining institutional deliveries demand.

Community-based respondents’ pride and belief in their role in the intervention’s effectiveness also appeared to bolster their commitment and ownership of the structured processes they relied on for its implementation, creating a virtuous cycle. For example, rather than deferring to hierarchical processes, in which one might expect community stakeholders to wait for directives, community-based implementers demonstrated ownership of their processes, using them for improvement.

There was a self-starter quality to community implementers’ collaboration, enabled by the organized processes they adopted and maintained. In this way, these themes shared a relationship with implementer’s commitment to implementing the intervention, captured in Individual Stage of Change below.

Across sites, respondents demonstrated clarity about the intervention’s formally-appointed leaders. Compatibility between implementers’ roles in the intervention versus their pre-existing social and professional roles facilitated their ability to adopt and implement the intervention, without need to “prove” the appropriateness of their post to their peers. In both sites and across all respondent groups, nurses understood their tasks and were similarly understood and respected as medical authorities by their co-implementers. Similarly, TBAs were regarded and respected for their effective emotional partnership to pregnant and laboring women, an authority that was also uncontested by their co-implementers or beneficiaries. Compatibility brought ease and speed in implementers’ understanding, adoption, and demonstration of mastery in their roles as formally-appointed leaders.

This synergy also facilitated implementers’ ability to overcome emergent challenges to implementation—particularly in the case of CLC members.

“…Last month we had a situation where…the husband of the pregnant lady, he did not believe she…[needed to deliver in the facility]. [They] came to present the case to the CLC and [we] as members along with the TBAs referred the lady to the health facility.”—Upper Province CLC member

TBAs and CLC members in particular stewarded and reinforced each other’s leadership roles, leveraging the compatibility of their roles as implementation leaders with their pre-established roles as respected leaders in the community. CLC members upheld TBAs’ leadership over community mobilization and tracking of pregnant beneficiaries from ANC to delivery. TBAs in turn enabled and reinforced CLC members’ authority to review performance data and set agendas for their collective work, and both groups of community-based implementers seemed to have an explicit understanding of the reinforcing value of the contribution of their colleagues’ roles to their own ability to perform effectively.

The authority to collect and use data featured strongly in respondents’ discussion of leadership. Here, the constructs of Engaging Formally-Appointed Leaders and Compatibility overlapped with structured Networks & Communications, as CLC members and TBAs intertwined discussion of their roles as Formally-Appointed Leaders with descriptions of the procedures they used to organize, make sense of, and validate their work—the structures they used as leaders to lead. Through generation and review of their own data, community-based implementers took ownership of their performance and in doing so seemed to draw confidence in the legitimacy of their roles as leaders of the intervention, thereby supporting continued implementation:

“I know that in this community there is (xxx) number of pregnant women who are about to give birth, or who are still registering their pregnancies, I know everything. I…organize them in this way. …Women who are leaving and women who are pregnant, [so we know] when it is time for [beneficiaries to] give birth.” –Lower Province TBA

“TBAs bring [us] the data … how many women are pregnant and how many deliveries they have seen. …Members of the CLC are first to thank the TBAs, [for] the data they provide us, [so we can] know what is the state of health [in] our community...” –Lower Province CLC member

Though this factor emerged as a facilitator to community-based implementation, it is important to note that in Lower Province, its positive influence did not translate to facility-based implementation. There, where community respondent interviews suggest that nurses and facility managers did not fully uphold their formally-appointed leadership roles, community-based implementers also failed to address this problem as leaders.

Across the two sites, majority of respondents demonstrated ownership of the intervention and their role in its implementation. Facility respondents discussed the importance of sustained collaboration with the community. Community-based implementers similarly described their implementation activities as “permanent” and happening “without fail.” Some of the most powerful statements reflecting implementers’ Stage of Change came as community-based implementers intertwined their reasons for supporting the intervention with stories of sorrow, suffering, and loss due to the previous norm of home births. Respondents recounted examples of family members using pestles to attempt to expedite births by pushing on the fundus of laboring women, protracted and painful labors, lack of hygienic practices, hemorrhage, and death. These experiences seemed to bolster TBA and CLC members’ desire to protect women and families from preventable deaths, in turn fortifying their commitment to delivering change through the intervention.

This commitment to the intervention did not come without cost to implementers. TBAs described long hours and time away from home, the challenges of crossing flooded rivers to escort laboring women to the clinic, pain imposed by the distances traveled, and sleepless nights. Despite these challenges and lack of pay, these stories were interlaced with their expressed intention to sustain implementation.

TBAs also discussed these hardships in terms of stress in their personal lives, as their commitment to ensuring adherence to institutional deliveries challenged their ability to fulfill household and gender roles at home. A TBA poignantly articulated this friction between her commitments:

“[Women] ask me…‘please escort me [to the facility] I want to give birth,’…I cannot clean my farm, I cannot have my meals at home, there is nothing I can do at home I only leave my husband worried and sometimes angry, but I tell him please…let me go to work. …Even if my husband is in bed and very erected, I…get up and go [to work].” –Lower Province TBA

TBAs and CLC members’ commitment to sustaining the intervention bore a relationship with their perception of increased recognition, appreciation, and deference from their community as a result of their role in the intervention. This increased social capital may have acted as an informal reward or remuneration for community-based implementers’ efforts, sustaining their resolve and commitment despite lack of pay.

Across the two sites, beneficiaries’ views about the intervention and their resulting demand for the services of TBAs, CLC members, and facility staff clearly intermingled with implementers’ own motivation to implement.

Across both sites, motivation based on past negative experience centered on beneficiaries’ previous experiences during births based in the community. Nearly all beneficiaries who discussed their previous births described risk, fear, suffering, and loss of children as motivating factors in adherence to institutional delivery.

“[I] had a very bitter experience, in the first delivery that…at home, in the community. …It had complications and at the moment [of] the delivery, [I] was unconscious, [I] did not know what was happening and [I] ended up losing the child. …Then [I] started having advices from the [TBAs]. And so when [I had] the second pregnancy, [I knew that] the deliveries are safe when they are done in the maternities…” – Upper Province Beneficiary

Beneficiaries’ perceived quality of care was a strong motivator as well. Notably, this was also the case in Lower Province, despite respondents’ accounts of misconduct at the hands of facility staff.

“What motivates us [to come for delivery] even being far away…Is that when we reach the hospital we are well attended [by] the TBAs…The nurses do not appear…they only come to charge 20 meticais...” –Lower Province Beneficiary

Social momentum arose as a motivating factor in both sites, acting as a sort of peer pressure among community members to continue demand for institutional deliveries.

“What make [me] decide to give birth in the hospital is the fact that [I] liked to hear that ‘She gave birth in the hospital.’ [I] would not like to hear somebody comment saying that [I] gave birth at home again. [Interviewer:] In your community here, do they talk well about people who give birth in hospital? Yes they do. A person likes it and takes advantage.” —Lower Province Beneficiary

“…The tendency…to deliver in the facility, [I don’t] think twice on that. …Other women are [also] aware, they are informed and they are really into the spirit.”—Upper Province Beneficiary

The role of context, collective action, relational structures, and shared perspectives in empowering women to voice their desires to deliver at the facility and, specific to this study’s focus, facilitating implementation have been found in previous studies [33, 43‐45]. This study builds on this evidence base. The contrasting homogeneity versus heterogeneity in implementers’ perspectives on their implementation helped to construct the picture of how these actors worked together as a system [33, 46]. Olivier de Sardan et al. have articulated the roles of congruence (or lack thereof) between patients’ and health workers’ norms on implementation [47]. We additionally find this relevant to health system and community workers.

Debpuur et al. have found combined health system and community health coalition-based implementation to be effective [48]. These findings deepen understanding of the extent of communities’ influence in implementation of respectful care and EPMM. Community lay health workers, and particularly traditional healers, have been found to be effective custodians of care, supporting beneficiary satisfaction with services [49]. In this study, community implementers demonstrated high confidence and precise technical understanding of the practices they used to enable their sustained implementation, including techniques for respectful, compassionate care, routine data generation and review, and joint data-based decision-making in pursuit of optimal coverage of beneficiaries. Gimbel et al. have found the significant role that integration of quality improvement into management strengthening can play in unifying health system actors from various levels for sustained implementation [46]. In this case, the intervention’s management strengthening targeted community level actors, and the resulting unification extended beyond the health system, to include the communities’ unified functioning as a system.

Global dialogue has long pointed to the importance of communities’ involvement in global health programming, as lay health workers for MCH care [7], as participants in health care planning and quality improvement [50], and as decision-makers to address their own health needs [51]. However, it has also been recognized that health programs often cite “community empowerment”—e.g. community engagement (passive or active) with information, materials, or skills-building—as a primary objective, rather than working to achieve full community participation [52‐55]. In this study, we find community implementers were core contributors to the successful implementation observed in the two sites, enabled by their commitment to the intervention and their capacity to manage their own implementation. Community implementers acted as significant counterparts to the health system in achieving sustained implementation, leveraging their social status to enhance their and their co-implementers’ effectiveness and authority in the intervention, driving demand, influencing community norms, and innovating to ensure service coverage and provision of compassionate, respectful care responsive to women’s emotional and physical needs. The significance of their role in this study echoes the findings of studies in West Africa regarding community leaders’ roles in sustaining performance-based financing schemes [40], and in exercising control over quality and reliability of implementation [56]. The need for this additional influence beyond purely health system-based implementation for quality, respectful care has been well documented [57, 58]. In this case, we find far more than community “involvement” to support the health system’s delivery of quality, respectful care.

Implementers, donors, and decision-makers may use these insights in considering the design of system strengthening programs aimed at sustainable solutions for EPMM and respectful care. Intervention strategies to sustain EPMM interventions may benefit from investing in communities not only to enable their involvement in EPMM, but to enable their full partnership with the health system, as complementary systems for sustained implementation. In this way, when complexity strikes—whether due to logistical or ethical challenges—EPMM interventions will have two capable allies to weather the storm.

We join a discrete but growing body of studies that have applied the CFIR in low- and middle- income countries (LMIC), and have found the CFIR to offer a useful, evidence-based framework to organize and study complex implementation variations, to gain insight into how and why successful implementation could be achieved in these two sites [46, 59]. Though we were pleased to find the CFIR’s utility to our implementation setting, we also found some challenges to its applicability. First, many complex interventions in LMIC include volunteer-based implementation with actors outside of the formal health system, but the CFIR does not explicitly recognize these implementation settings [24, 30, 60]. Because of this, we were originally unsure of its applicability to our implementation scenario. In the future, it may prove useful to expand the CFIR’s guidance to discuss how the constructs apply specifically to such informal volunteer implementation settings. Second, as discussed above, constructs and sub-constructs often overlapped, shaping how they influenced implementation. Though the CFIR has been recognized for attempting to address the integrated nature of implementation and context [60], we found the framework to lack guidance on this, thereby somewhat lessening our findings’ ability to immediately inform implementers’ real-time decision-making because of the additional interpretation and time required. Related to this, we originally aimed for analysis to be a collective process with implementation managers so as to inform their decision-making [61]. However, by following the recommended process for use of the CFIR—in which it is applied both to inform study design as well as to analyze and code data [31]—our process proved too lengthy and exhaustive for these initial purposes. In the end, we found it necessary to divorce the study from implementation in order to minimize disruption to implementation. As a result, though useful to future implementation, findings could not be used to inform decision-making during the project’s lifecycle. Though recent research by Keith et al. suggest the CFIR may be used for rapid evaluation, we note that these authors were themselves a dedicated research team separate from the implementation team [62]. We suggest the impact of the CFIR may be expanded by design of instruments that can support implementers’ direct and rapid diagnosis of factors influencing evidence-based interventions’ implementation. This may prove particularly useful as implementers take interventions to large-scale and, in so doing, must consider how to adapt and tailor their implementation to exponentially more varied contexts.

Though our study did not aim to measure the impact of the intervention and so did not require it, it was nonetheless not possible to create control sites for the quantitative portion of this study. Consistency and reliability of the clinic register data collected on institutional deliveries and beneficiaries’ community of origin were not uniform in the province. As a result, the authors were required to invest considerable resources to ensure quality data collection in the selected Lower Province and Upper Province sites during the observation period and were not able to invest the same for control sites. Additionally, the authors were able to collect only 8 months of data preceding launch of the SCIP project owing to similar challenges with consistency and reliability of data collection in clinic registers. Our study design has delivered data and findings in which the authors have high confidence. Nonetheless, by nature of our study design, we cannot—and have not aimed to—claim causality between the intervention’s implementation and the observed increases in institutional deliveries, nor generalizability to other settings. Additionally, because of resource constraints similar to those mentioned above, we were unable to include a greater number of communities in key informant interviews. In-depth interview respondents, having been initially identified through facility and project staff, may have introduced bias. To address this, we were careful to ensure that snowball sampling introduced additional respondents in each community and ensured that any new themes triggered subsequent identification of additional respondents until saturation was reached. In analysis, we also used beneficiary respondents’ accounts to verify or validate those of implementers, and vice versa. We also recognize that these sites, as in most of the province, had a previous norm of home births and relatively low health-service-seeking behavior prior to the intervention. As such, our findings regarding respondents’ motivation for institutional deliveries due to perceived quality of care may also be influenced by the relative “novelty” of the facility and may change over time as this novelty reduces and potential exposure to poor quality care at the facility increases. Our observation timeline may not have been sufficient to capture this change. Finally, as this was a study conducted by and for implementers, study limitations must also acknowledge that two of the authors were directly responsible for implementation of the SCIP intervention, and six of the authors have been employed at some time or are employed by the organization that managed the project.