Community conversations (CCs) are a powerful tool that encourages critical thinking among participants, stimulates finding tailored solutions to the problem detected, and can lead to lasting community empowerment. Opportunities are created for diverse stakeholders to develop critical consciousness; an awareness of the ways in which relationships between structural contexts and our experiences come about and contribute to the difficulties people face in their everyday lives [25]. Dialogue has been identified as key to processes of social change [26] and the CC method enables community-centred dialogue through a light touch role for the facilitator.

The Nelson Mandela Foundation’s approach to CCs involves five broad stages to generate critical dialogue: Relationship building, Concern Identification, Concern Exploration, Decision making, and Reflection and Review [25]. It was popularised within HIV-related work in Southern Africa and has been used to develop a range of community-led interventions that challenge norms, shift action and provide community-defined solutions to change, including HIV adherence [26] or issues related to gender-based violence [27-29]. However, this approach remains under-utilised.

The CC method is not without its challenges. Given its interest in combining actors across a range of social positions within shared spaces for thinking and action, some evidence warns that consensus building can be challenging [27]. Furthermore, structural challenges that communities identify as targets for change, such as poverty and political instability, have created obstacles to problem-solving leading to frustration and barriers to community empowerment [26]. However, these issues can be overcome, as seen in a recent application of CC in Zambia for the improvement of women and child health outcomes, where a range of community-led activities responded to structural drivers of poor outcomes, including the construction of new health centres, alongside typical health information programmes [29].

Given the capacity for CCs to illuminate a wide range of concerns linked to health challenges, while providing a pathway to meaningful dialogue between communities and researchers, we felt the methodology could provide a more systematic approach to community engagement in the design of our intervention. We describe our approach, as well as findings generated from its use in subsequent sections.

The INSPIRING intervention: a whole system approach to tackling infectious diseases in children under five

Formative research was carried out as part of a larger INSPIRING programme in Jigawa State Nigeria, which uses a cluster RCT approach to impact evaluation [30]. The project involves multiple partners and was designed through an iterative co-design process with the funder, implementer, evaluators, government and local community members. The intervention is a locally adapted ‘whole systems strengthening’ package of three evidence-based activities: community Participatory Learning and Action (PLA) groups; Partnership Defined Quality Scorecard (PDQS); training and engagement between caregivers and health facility staff, mentorship and provision of basic essential equipment for child health. The inclusion of community members in the co-design process was achieved through an adapted CC approach, which allowed us to confirm the details and implementation/delivery strategy of PLA and PDQS community link intervention.

Research setting

Research was completed from November 2019 to March 2020. Jigawa is one of Nigeria’s 36 States, with an estimated population of 4.3 million constituting mainly Hausa-Fulani tribes, and a small proportion of Manga and Badawa, tribes [31]. People live in extended family compounds typically comprising of two to five households with children, parents, grandparents and other siblings. 98.9% of the population practices Islam, while 1.1% are indigenous Christians. Agriculture is the occupation for 80% of the population and most live in rural areas. Two-thirds of the population (69%) live in severe poverty, with 50.3% belonging to Nigeria’s lowest wealth quintile [31].

Implementing community conversations for formative co-design research at scale: a new method

To ensure meaningful involvement of the community in shaping the intervention design, RAB developed an adaptation of the Mandela Foundation CC approach, to engage with community members at scale. Though typically applied as an intervention itself, our reformulation was designed to deepen engagement with the communities of men, women and health practitioners who will engage with the proposed interventions. Conversations were structured in a similar manner to the PLA group component of the intervention, providing an additional layer of depth for understanding the acceptability of the approach in this context. Our adaptation included four of the five CC stages: Relationship building, Concern identification, Concern Exploration and Reflection and Review. Our decision to exclude the decision-making stage was linked to the ability for this stage to turn the approach into an intervention itself, as it is oriented to deciding and acting on how to answer the problems that have been prioritised in previous stages. To run this stage would have comprised the ability to establish true baseline prior to the trial intervention being delivered in later months. Interactive CC discussions explored: perceptions of key concepts underpinning the proposed interventions (e.g., challenges related to child health); relationships with health care workers; and how key intervention components would work best in their communities, including location, timing for delivery of groups and incentives.

Sampling

Three categories of Jigawa community members were invited to participate in CCs: men, women and healthcare workers. Participants were recruited using a blend of convenience and purposive sampling. The inclusion criteria included any adults who were parents and caregivers of children under five, aged 18-49 years, or health care staff. To identify eligible candidates, the research team walked around the communities with a village gatekeeper someone with intimate knowledge of the geographical and social community, who is well respected and known by those living in the area. This resulted in a combined population-level availability and targeted sampling. Consent was obtained from participants who were then invited to gather at the meeting point.

During CC meetings, participants were broken down into sub-groups. This ensured that any participants with have limited agency in mixed settings (such as younger women or lower ranking wives, which we observed in this community in earlier work (see [32]) had opportunities to contribute ideas and shape discussions and at each round. Older women (age 31–49), younger women (age 18–30), and men (age 18 above) were placed in groups. Given that men with children in community settings were found to have so much power in our scoping research [32], and based on the perspectives of research team members who are local to this context (FS, IH, AB, ) we did not anticipate the need to separate men into sub-groups by age. Because of the number of participants at each site, we had groups running in parallel on each day, with a maximum number of 10 participants per sub-group working together at a time.

Procedure

CCs were conducted across 2–3 days, per ward. Wards were selected based on consultation with the INSPIRING team, to ensure a balance between rural and peri-urban sites, as well as moving across the whole local government area. Conversations were held at primary school sites, which was agreed upon following earlier consultations with community gatekeepers.

Each CC session started with the researcher welcoming participants and recording their socio-demographic characteristics. Facilitators provided a brief introduction in which they explained the project, its partners, aims of the conversations, and established ground rules for the conversations. Facilitators were responsible for guiding the conversations with open-ended questions and activities, which are detailed in the Supplementary materials Discussions in sub-groups were not recorded, but recordings were taken during the larger discussions where each group reported back to the full group. Note-takers captured additional reflections from sub-group and large-group discussions.

We ensured the process was not dominated by any single voices or groups, by giving all groups the opportunity to participate and by distributing tasks within the smaller groups, i.e. some volunteered to draw, while others generated ideas of which they all deliberated on. Refreshments and transportation reimbursements were given to participants at the end of each session (Table 1).

Table 1 Overview of procedures in CC adaptation processParticipatory activities

Participatory learning activities [33] provided opportunities for people to share and contribute to debates, without direct questioning, and created a level playing field among participants with limited formal education — Figs. 1 and 2 demonstrate some of these activities.

Fig. 1

Men’s group: showing body map of a child with associated risks factors

Fig. 2

Older women’s group — Venn diagram of Community power dynamics

Localising risks to child health: body mapping

Body mapping activities are widely used in research studies that seek to understand how participants make sense of bodied manifestations of pain, emotions, and illness, using illustrations [34]. We used the approach here to make discussions about risks facing children more tangible, allowing participants to anchor discussions to maps of an imagined child. Our approach to facilitating this activity is described in supplementary materials document 1.

Understanding relationships: stakeholder and power mapping (Venn diagram)

A key part of intervention development is the process of stakeholder mapping. We expanded stakeholder mapping, by including a Venn diagram activity (see Fig. 2). Using these methods, allowed us to explore not only stakeholders, but to tangible represent the relationships between different actors. This is particularly important given existing understandings of gendered power dynamics in the region [3].

Understanding space and place in relation to child health — community mapping

Community-based and community-led interventions demand an understanding of community that is as complex as possible, though this is rarely the case. Typically, the emphasis on community stops at the geographical, with concerns about distributions of illness or demographics. In our context, we sought to explore community as a more complex entity, acknowledging three often under-appreciated dynamics related to communities of place: (1) social identity processes (how people identify and relate to each other) linked to the use of space; (2) community of practice and action — created through engagement in shared activities; and (3) symbolic communities — linked to cultural sub-groups. The mapping activity (see [35]) began with a process of identifying landmarks and resources within the space but expanded beyond this to enquire about how different groups within a community linked to the above dimensions make use of this space. Details on how this activity was carried out are included in supplementary data document 1 (Fig. 3).

Fig. 3

Young women’s group community mapping activities

Focus group discussions: reflection and review through concept testing

The final stage in the community conversation process involved dialogue-based concept testing [34]. This allowed us to share findings from the community conversations with additional community members. Broadly speaking, our efforts were in line with member-checking approaches in qualitative research, which increases the trustworthiness of data and enables researchers to ensure that participant meanings are not supplanted by researcher priorities and aims for knowledge production [34].

We presented synthesised data and a proposed intervention structure to participants in two types of focus group settings: (1) discussions with sets of new participants in the six wards we had been working in previously (n=54), and (2) discussions with participants in the 5 governmental wards where we had yet to collect data (n=45). This member-checking approach fits with the wider epistemological purpose of community conversations, where the goal is to remain rooted in community-owned knowledge production. However, by also including new participants, our emphasis extended beyond validation, allowing us to explore whether the perceptions and views of other community members resonated with others, to identify dissonant voices, and create opportunities to add new data or perspectives to our findings before rolling out the intervention [34, 36]. This adaptation is particularly important for trial-related work, where the approach will eventually be rolled out to large portions of a population. Details on activities are described in supplementary data file 1.

Data collection, processing and analysis

Three local research assistants (RAs) were recruited and trained by FS, who is local to our study community. In addition to exploring study objectives and sampling processes, role play was used to introduce RAs to community conversation procedures. This allowed us to confirm the resonance of the activities with people from this community and refine our approach if needed. The standard operating procedures for the method were provided and included sample probes to expand on responses in sessions. All conversations and interviews were recorded, transcribed verbatim and then translated into English by the local RAs.

Data analysis was completed using a codebook approach to thematic analysis [37]. Following an initial reading of a selection of transcripts by the lead author RAB, a preliminary codebook was created, to structure a focused reading of data, to provide answers to specific questions in relation to intervention development. This was then used by additional members of the analysis team (FS, AI, AAB), who expanded on these initial structures individually. Individual updated codebooks created by other analysis team members were consolidated by RAB, looking for convergence/divergence across themes. A final codebook was circulated to the analysis team, and additional members of the study team (TC) for agreement on the summary analysis at the preliminary stage.