The development and implementation of a community engagement strategy to improve maternal health in southern Mozambique

Delays to seek medical help can contribute to maternal deaths particularly in community settings at home or on the road to a health facility. Community engagement (CE) can improve care-seeking behaviours and complements community-based interventions strengthening maternal health. The purpose of this paper is to describe the process undertaken to develop and implement a large-scale community engagement strategy in rural southern Mozambique. The CE strategy was developed within the context of the “Community-Level Interventions for Pre-eclampsia” (NCT01911494) conducted between 2015–2017 in southern Mozambique. Key CE messages included pregnancy complications and their warning signs, including pre-eclampsia and eclampsia, as well as emergency readiness, birth preparedness, decision-making mechanisms, transport options and information about the trial. CE meeting logs were used to record quantitative and qualitative information on demographic data and feedback. Quantitative data was analyzed using RStudio (RStudio Inc, Boston, United States) and community feedback was qualitatively analyzed on NVivo12 (QSR International, Melbourne, Australia). CE activities reached 19,169 participants during 4,239 meetings. CE activities were reported to be well received by community members though there was a relatively lower participation of men (3565 /18.6%). The use of recognized local leaders and personnel, such as community leaders, nurses and community health workers, allowed for greater acceptance of CE activities and maximized coverage of health messages in the community setting. Our CE strategy was effective in integrating maternal health promoting activities in routine care of community health workers and nurses in the area. Understanding district differences, engaging husbands, partners, mothers-in-law and community-level decision-makers to build local support for maternal health and flexibility to tailor messages to local needs were important in developing sustainable forms of CE. Better strategies are needed to effectively engage men in maternal health promotion who were less available due to working outside of the home or neighbourhoods

Introduction Similar to many countries in sub-Saharan Africa, most maternal and neonatal deaths in Mozambique occur during childbirth and the first few days of life. Lack of access to appropriate medical care contributes significantly to these deaths [1]. A study on the prevalence of the use of institutional delivery services in Mozambique showed that 29% of deliveries occurred outside of health facilities and the proportion of home births was four times higher in rural areas (40%) than in urban areas (10.8%) [2], suggesting sub-optimal health care seeking behaviours [3] and poor distribution of services [4]. The delay to decide to seek medical help is a major contributor to many deaths that occur at home or on the road to a health facility [5]. Appropriate care-seeking behaviour involves first recognizing the danger signs for women and newborns. Knowledge of danger signs in pregnancy, pre-eclampsia in particular, are limited and represent a challenge in the reduction of maternal and infant mortality. A study in southern Mozambique showed that rural communities had little knowledge about pregnancy complications, especially about the origin of pre-eclampsia [6]. Faced with challenges related to access and knowledge, community members may adopt use of traditional beliefs and practices, some of which may carry higher risks to maternal and child health [6,7].
Community engagement (CE) can improve care-seeking behaviours and is an essential part of implementing sustainable health promotion programs [8]. CE in health interventions strengthens communities through increasing their self-confidence, self-awareness and selfreflection, which supports finding local solutions to health problems [9]. Consequently, the World Health Organization (WHO) recognizes the importance of engaging communities in promoting health and describes CE as vital to linking health problems to appropriate health promotion actions [10]. The WHO highlights that CE can improve maternal and newborn health behaviours, increase use of skilled care and increase household and community support for maternal and newborn health [11]. However, there is a gap in identifying sustainable forms of CE [9]. In order to contribute to a better understanding of locally-tailored forms of engaging communities, including the sustainability of such approaches, this paper describes the process of development and implementation of a large-scale CE strategy in rural Mozambique.

Study design and areas
The CE strategy was developed and implemented as an integral part of the "Community-Level interventions for Pre-eclampsia (CLIP) in southern Mozambique: a cluster randomized controlled trial" (NCT01911494) conducted between 2015-2017. CLIP was conducted as a collaboration between the University of British Columbia (UBC) in Canada and Centro de Investigação em Saúde da Manhiça (CISM) and Universidade Eduardo Mondlane (UEM) in Mozambique. The objective of the CLIP Trial was to reduce maternal, perinatal and neonatal mortality and morbidity through a community-based intervention building capacity of community health workers (CHWs), referred to as Agentes Polivalentes Elementares (APE) in Mozambique. These CHWs worked in close collaboration with primary health care nurses [12,13].
The CE strategy was implemented in six regions: two from Manhiça district in Maputo province and four within Bilene Macia and Chibuto districts in Gaza province (Fig 1). The six regions were selected because they constituted the intervention areas of the trial. All three districts are rural, they ranged in size from 1,289.8 to 1,829.1 km 2 with a population of 7,197 to 16,129 ( Table 1). Residents belong to the Xichangana ethnic group. Agriculture represents the economic basis for families' subsistence and the main cash crops are sugarcane, cotton and tobacco.

Community engagement strategy development
CE is a process that encourages community participation in activities to improve population health and/or reduce health inequalities [14]. The WHO defines CE as, "a process of  developing relationships that enable stakeholders to work together to address health-related issues and promote well-being to achieve positive health impact and outcomes" [14]. CE can support building trust, enlisting new resources and allies, and improving communication to support environmental and behavioral changes that will improve the health of the community and its members [15]. Our CE strategy was developed based on the results obtained in a feasibility study to understand women's health care seeking practices during pregnancy and identify underlying social, cultural and structural barriers to accessing timely appropriate care in CLIP Trial communities [16]. Through in-depth interview and focus group discussions with women of reproductive age (WRA), pregnant women, household decision makers, formal and informal health care providers, local health authorities, community leaders and other influencers, the feasibility study identified a number of maternal health barriers that formed topics to be addressed by our CE strategy. These included: 1) Poor awareness of pre-eclampsia, eclampsia and pregnancy complication warning signs; 2) Late disclosure of pregnancy, leading to delayed first antenatal care visit; 3) Women´s low decision making power; and 4) Community transport limitations ( Table 2). Therefore, the CE strategy implementation first aimed to encourage community dialogue on obstetric complications and healthcare-seeking practices. Secondly, as the feasibility study also identified local contextual factors that influence capacity for care-seeking, these were considered by different accompanying components of the PRE-EMPT project. For instance, because lack of transport was identified as a major contributor to poor referrals and represented a barrier to adequate maternal healthcare access [16], a community transport strategy was also developed to leverage existing community resources within a group savings scheme. The community transport strategy is reported further in Amosse et al 2021 [17].
Throughout the development process, we consulted community members, APEs, health care providers at health facilities and the Mozambican Ministry of health regarding our proposed CE strategy to ensure it was locally feasible and appropriate.

Community engagement personnel and activities
CE activities took place in three spaces: primary health centers (PHCs), in the homes of pregnant and puerperal women and in community meetings. CE activities were conducted by three groups of facilitators: 16 nurses conducted activities at PHCs, 48 CHWs during home visits, and five activists and three mobilizers at community meetings. Maternal and child health nurses provide daily health talks as outlined by the Ministry of Health at PHCs and messages regarding pre-eclampsia, eclampsia and other topics were added to these daily talks (Box 1). Daily health talks largely targeted pregnant women and WRA, but also included husbands and other relatives. CHWs are community members selected and trained by the Ministry of Health to provide basic medical care at the community level. For CLIP Trial, they were also trained to recruit pregnant women, conduct follow-up visits, provide messages regarding pre-eclampsia and eclampsia, assess their risk for pre-eclampsia and eclampsia and other complications and refer them to the nearest health facility when necessary (Box 1).
Activists are community members that engaged with the CLIP team and were trained to facilitate community meetings in collaboration with community leaders, who acted as gatekeepers by scheduling and inviting people to participate in community meetings. Because they were based in the community and had a flexible availability, activists frequently conducted the CE meetings. Meetings were also sometimes conducted by mobilizers. Mobilizers are staff based at the central CISM office who made field visits three times a week to conduct community meetings and monitoring activities conducted by nurses, CHWs and activists.
Activities led by nurses and CHWs focused more on health oriented topics, while CISM field staff focused more on research and socio-cultural elements. Overall, facilitators were well trained, equipped with previous research experience, familiar with the community context, and fluent in the local language. Relationships with the communities were established prior to data collection by approaching the administrative post chiefs, traditional leaders, and the neighbourhood secretary for prior permission.
The entire team was coordinated by a Community Liaison Officer (CLO), who oversaw the activities, monitored progress, trouble-shot on operational problems, managed the data, and was the linkage with the investigators' team and the research centre.

Materials
A number of materials were developed to support CE facilitators (nurses, CHW, activists and mobilizers). An informational booklet about pre-eclampsia was developed from questions 6. Feedback on adverse outcomes and 'great saves' 7. Study protocol overview � 8. Details on specific CLIP study interventions (pre-eclampsia diagnosis, treatment and referral) 9. Solutions to the previously identified barriers to maternal and neonatal health � Included discussion of the PRE-EMPT study objectives, target groups to be recruited and study procedures. raised by community members during the feasibility study. Two thousand copies of this booklet were distributed to CE facilitators, community leaders, religious leaders, pregnant woman, WRA, partners and matrons. Pictogram flip chart, posters, and key message scripts which distilled clinical explanations into illustrations and lay language were also developed.

Community engagement messages
The CE activities focused on the key messages regarding pre-eclampsia and eclampsia, pregnancy complications and care-seeking in pregnancy (Box 1).

Data collection and management
CE logs were completed by facilitators after all CE sessions at the health facility, individual homes and community meetings. Information recorded included details on the region, target group, number of participants and messages discussed. CE logs also included community feedback notes where facilitators recorded brief reflections and comments from participants in free text format, which were used for the qualitative analyses. Mobilizers collected CE logs from the community and health facilities and centralized the logs at the CISM office for data entry. CE logs were monitored by mobilizers once a week and quality checked by community liaison officer once a month. While collecting the logs there were opportunities to verify the completeness of the data and clarify or resolve missing data or discrepancies. Missing or questionable data was queried by the study team for follow up. Logs were entered into a REDCap database (Vanderbilt University, Nashville, United States). Data entry from paper records to the electronic database was quality checked for accuracy.

Data analysis
Descriptive statistics were conducted with the quantitative data using RStudio (RStudio Inc, Boston, United States) to compile frequency of demographic characteristics, number of meetings and the topics discussed. Data was then presented in the format of tables and bar charts, and frequencies were compared by region and by target group. These outputs were used to monitor CE activities and adjust when necessary. The community feedback notes from CE logs were imported to NVivo12 (QSR International, Melbourne, Australia) for content analysis. Conventional content analysis is an inductive method of qualitative analysis to interpret meaning from the content of data as expressed by study participants without imposing preconceived categories [18]. Conventional content analysis involves first familiarizing with the data and reading data word by word. Exact words from the data that capture key issues raised by participants are highlighted as codes. Codes are subsequently sorted into categories, which are used to organize codes into meaningful clusters of topics.

Ethical considerations
Approval for this study was obtained from the Institutional Bioethics Research Boards of Centro de Investigação em Saúde da Manhiça (CISM, CIBS-CISM/038/14), the Mozambique National Bioethics for Health Committee (219/CNBS/14) and the University of British Columbia (UBC, H12-03497). Written informed consent was obtained from all households participating in the CLIP Trial, of which community engagement was a nested component. By giving consent to participate in the CLIP trail, participants also agreed for community engagement activities, which were only conducted in CLIP intervention clusters. Following the CLIP protocol [12], adolescents in the Trial gave assent combined with a parental or guardian consent. All CE records were anonymized.

Participants' characteristics
During the two-year study, CE activities reached 19,169 participants between the ages of 12 and 96 years old ( Table 3). Fifty-one percent of the participants were married or cohabiting with their partner. The majority of participants were women (81.4%), and most participants worked in subsistence farming (55.5%), while 18.7% were unemployed.

Community engagement coverage
During CE activities, 4,239 meetings were conducted in Bilene Macia, Chibuto and Manhiça districts ( The primary target group comprised of pregnant women and WRA, who represented the proportion meetings. Pregnant women and WRA were engaged in more than half (69%) of all meetings held throughout the entire CE process ( Table 5). Next were meetings with combined target groups, which represented 13% of meetings. Though the numbers of meetings with community leaders represented a small proportion of overall CE meetings (0.1%), they were engaged in an impactful role in community meetings as gatekeepers and opinion leaders. Bilene Macia, was the district that reached the highest number of meetings while Manhiça registered the lowest number of meetings. Fig 2 illustrates the frequency of messages disseminated. The most discussed topic was "danger signs and symptoms of pregnancy complications", particularly those related to preeclampsia and eclampsia. The second most frequent topic discussed was "identification of skilled birth attendants".
Message frequency varied by region (Fig 3). For example, signs and symptoms of pregnancy complications was more widespread in Manhiça district, while permissions for care seeking was more widely addressed in Bilene-Macia and Chibuto districts. As the study progressed and facilitators and participants became more familiar with the topics, facilitators adapted the focus of discussions to existing gaps in the local communities.

Stakeholders' reflections on engagement activities in communities
Reflections on community engagement format. Community engagement meetings were designed to be an "exchange of ideas" and participants were encouraged to ask questions and share stories from their own communities. The CE activities included dedicated time for open questions and an opportunity to assess comprehension. The quote below from the community feedback illustrates the focus on dialogue and experience sharing to raise awareness of preeclampsia and eclampsia in the communities.  "I am very happy with this care (during the CLIP Trial) and lecture. I liked the information about the pre-eclampsia /eclampsia disease and danger signs in pregnant women." Pregnant woman from Malehice, Chibuto district "The women listened attentively, they were curious and asked questions. . . They promised to go to the Health Facility whenever necessary or if they felt discomfort." Pregnant women from três de Fevereiro, Manhiça district The use of recognized local leaders and personnel, such as community leaders, nurses and community health workers, allowed for greater acceptance of CE messaging. However, women were sometimes shy and reluctant to discuss openly, especially if a community leader was present. A few remarked that women were too tired to participate actively in CE activities.
"The pregnant woman's mother-in-law said she has no time to listen to what the CHW had to say because she was tired and hungry." A mother-in-law during a discussion with women with reproductive age in Chissano, Bilene Macia district "In this neighborhood, the women were shy and didn't make many comments." Pregnant women from Ilha Josina, Manhiça district While most feedback described positive community reception of activities and messages, CE facilitators reported some pushback from men who questioned the focus on maternal health and obstetric emergencies. They wondered why the CLIP Trial and CE messaging did not help them and were sometimes concerned that the messaging would increase fear among their wives and partners. While some men were critical of the project, many men were also positive about the project.
"We would like to understand why only the pregnant women are treated. Men die and the women in the community live." Husbands and partners from Messano, Bilene Macia district "The partners were worried because they heard that this disease has no cure and becomes serious when not controlled. This will make women afraid to give birth." Husbands and partners from Chissano, Bilene Macia district

Discussion
Community engagement activities during the CLIP Trial reached about 19,000 participants with over 4,000 meetings during the course of two years in rural communities in Southern Mozambique. Understanding local issues and barriers during the feasibility study allowed our team to develop a CE strategy that had a wide reach and was well received by community members. The focus on dialogue to deliver CE messaging helped to promote participant discussion on pre-eclampsia and eclampsia.
Lessons learned in developing sustainable forms of CE include understanding differences across districts, the importance of a continued dialogue and opportunities for feedback from communities, and the importance of flexibility to tailor messages to local needs and as participants became more familiar with the topics over duration of the study. Responsiveness to adapt CE strategies according to community needs and their responses to various study components has been identified as a key element of effective community engagement [19]. In our study, this was supported by encouraging CE facilitators' reflection on the process and key issues that emerged in each CE session, which provided a space to record feedback from communities, helped to strengthen facilitators' engagement with activities and supported monitoring and evaluation of the program. Facilitator reflections has been used in quality improvement initiatives to promote buy-in and motivation as well as provide the time and structure to reflect on potential adaptations to improve engagement [20,21].
Formative research has also been identified as a key element of effective community engagement to gain insight into the local socio-economic context and health needs [19]. Formative research can also support understanding differences between regions. Looking at the CE activities, we can see different dynamics of activities by district. Bilene Macia is a larger district in terms of population size (see Table 1) and a larger team of mobilizers recorded more meetings and participants than the other districts. Manhiça district registered fewer activities, which may be related to the low population density in Calanga and Ilha Josina. Activities were also influenced by access. Calanga and Ilha Josina were often affected by severe weather and flooding making the area difficult to access during rainy season [4], therefore during lengthy periods activities relied on a few local facilitators who were able to reach a limited number of target groups. In addition to the challenges of access for community engagement facilitators, the barriers of challenging terrain and poor road infrastructure may also impact access to maternal health services for women [4].
While CE and community partnerships are conceptualized as essential precursors to sustainability of global health projects [22], less is known about the sustainability of CE efforts. Our research shows that it is possible to integrate expanded maternal health messaging into existing health system personnel and routine care, which would support sustainability. However, this may also present some limitations on the delivery of some topics. Because nurses and CHW mostly focused on clinical-oriented topics, the danger signs of pregnancy, especially pre-eclampsia and eclampsia, identification of skilled birth attendants and permission for women to seek medical care were most frequently discussed messages during CE meetings. The reliance of activists' on the capacity of community leaders to mobilize people for meetings and the intermittent presence of the CISM team in the field may have led to lower coverage of the other topics with socio-cultural elements.
CE activities also highlighted the importance and challenges of reaching husbands, partners, mothers-in-law and community leaders that are influential in pregnant women's careseeking decision making. Pregnant women in previous research in these communities shared that they are expected to obtain permission for seeking maternal health care and delays in seeking care occurred when complications arose and husbands were unaware of the warning signs of pregnancies [16]. A study from Ethiopia found that men's knowledge of obstetric danger signs was poor but there was significantly higher levels of birth preparedness among men who knew at least one danger sign [23]. This supports results of two systematic reviews that found male involvement was associated improved utilization of maternal health services, such as improved antenatal attendance, higher likelihood of facility birth, skilled birth attendant and postpartum care, as well as better birth preparedness as maternal nutrition [24,25]. A locally contextualized CE strategy in Pakistan that engaged both men and women supported increased pre-eclampsia knowledge that pregnant women in these communities had around seizures and high blood pressure [26]. This research helped frame the CE strategy to engage men but results suggest that there may be contextual challenges for men to participate. Considering the quantitative and qualitative findings together, both datasets highlight the challenge of engaging husbands and partners. The descriptive statistics show the lower numbers of men who participated in engagement activities, while qualitative feedback found that some men questioned why study activities focused on pregnant women and excluded male health concerns. Lower participation of men may be influenced by competing demands. Men in these communities often work long hours outside their homes, and may be required to relocate to Maputo or South Africa for employment, therefore are usually overlooked in activities related to health information, communication and engagement. Future community engagement initiatives should not only invite men to maternal health discussions, but also identify barriers to their participation and reflect on best practices to improve their participation.

Limitations and strengths
As with any research, there are limitations to our findings. The data were collected in three districts of Maputo and Gaza Provinces; although these results show a good representation of the region, results are not generalizable to other settings. Another limitation was relatively low participation of some groups (decision-makers) and some of the participants were reluctant to share their opinions during the meetings. Geographical challenges such as poor terrain and remoteness of some communities also influenced the reach and pace of CE activities. Additionally, qualitative data was sourced from CE facilitator feedback thus there may be some bias to reporting positive results. While reoccurring participant engagement was encouraged to reinforce key messages and repeated attendance could suggest appreciation of continued community engagement efforts, the anonymization process of CE logs in which names were not collected limits assessment of repeated attendance. Additionally, we designed the community engagement strategy and its assessment to investigate the contribution of the combined efforts of multiple CE approaches within a complex intervention. This is a limitation because it did not allow differentiations explained across the different types of meetings and methods used to engage community members. Lastly, because the purpose of this manuscript was to describe the development and implementation of the community engagement strategy, we focused on descriptive statistics to describe the process. Follow-up is needed to explore the impact of CE on clinical outcomes.
Strengths of these findings include the use of both qualitative and quantitative data to understand the large scale CE effort. The research team had support of leaders at all levels, including involvement of the Mozambique Ministry of Health.

Conclusions
Our CE strategy in southern Mozambique integrated maternal health promoting activities in routine care of community health workers and nurses and the involvement of community leaders maximized coverage of messages among the target groups. The use of a recognized local structure (community leaders, nurses, CHW and local activists) allowed greater acceptance of CE activities and understanding district differences, engaging household decisionmakers and community leaders to build local support for maternal health and flexibility to tailor messages to local needs were important in developing sustainable forms of community engagement. While the current paper demonstrates the effective implementation of a locally developed CE strategy based on maternal health issues identified in the communities, further research is needed to understand potential impacts of CE on maternal health outcomes and strategies to effectively engage men in maternal health promotion.