‘We are the eyes and ears of researchers and community’: Understanding the role of community advisory groups in representing researchers and communities in Malawi

Abstract Community engagement to protect and empower participating individuals and communities is an ethical requirement in research. There is however limited evidence on effectiveness or relevance of some of the approaches used to improve ethical practice. We conducted a study to understand the rationale, relevance and benefits of community engagement in health research. This paper draws from this wider study and focuses on factors that shaped Community Advisory Group (CAG) members’ selection processes and functions in Malawi. A qualitative research design was used; two participatory workshops were conducted with CAG members to understand their roles in research. Workshop findings were triangulated with insights from ethnographic field notes, key informant interviews with stakeholders, focus group discussions with community members and document reviews. Data were coded manually and thematic content analysis was used to identify main issues. Results have shown that democratic selection of CAG members presented challenges in both urban and rural settings. We also noted that CAG members perceived their role as a form of employment which potentially led to ineffective representation of community interests. We conclude that democratic voting is not enough to ensure effective representation of community's interests of ethical relevance. CAG members’ abilities to understand research ethics, identify potential harms to community and communicate feedback to researchers is critical to optimise engagement of lay community and avoid tokenistic engagement.

success and sustainability of interventions. 3 There is however no widely agreed definition of community engagement and we employed the Council for International Organisations of Medical Sciences definition of community engagement because it offers ethical guidance on how to strengthen representation of communities in study design. We therefore define community engagement as: a process of engaging potential participants and communities in a meaningful participatory process that involves them in an early and sustained manner in the design, development, implementation, design of the informed consent process, monitoring of research and in the dissemination of its results. 4 While we concur that community members have capabilities to identify their needs and they should be actively involved in finding solutions to these needs, 5 the degree of participation deserves critical attention. According to Sherry Arnstein, there are degrees of participation ranging from nonparticipation to citizen control. 6 Nonparticipation occurs when communities are involved to be educated; tokenism occurs when they are informed or consulted but they lack power to influence decisions; while citizen control is attained when they are actively involved in planning, designing and have power to influence decisions. 7 This implies that meaningful engagement occurs when all parties effectively participate in discussions to identify solutions. The feasibility of integrating both lay and scientific perspectives in research design however remains a challenge particularly in low literacy settings.
One of the approaches used to involve communities in health research is the use of a Community Advisory Board (CAB) or Community Advisory Group (CAG). Community Advisory Groups were initially introduced in HIV/AIDS research to strengthen the representation of people affected by or living with HIV/AIDS in research planning and implementation. 8 Some donors now require establishment of a CAG, particularly in low resource settings, to provide community oversight on ethical conduct of health research. 9 Roles of CAG include reviewing study protocols and informed consent forms, representing community concerns, advocating for the rights of research participants, consulting with potential research participants to provide advice, identifying research priorities, assisting in development of study materials, study design and implementation. 10 While engaging a CAG is designed to strengthen community engagement, existing literature demonstrates challenges. Some of the challenges include limited understanding of health research, monetary expectations, dependence on researchers for finances, and lack of authority to influence decisions concerning research. 11 These challenges have led to scepticism about the advisory roles of CAG members and concerns that their involvement is sometimes tokenistic or 'window dressing' to fulfil donor requirements. 12 In this paper, we report findings from an ethnographic study in Malawi that seeks to understand the purpose, relevance and benefits of community engagement as seen by different stakeholders in research. We start by discussing approaches used to select the CAG members and how these affected their roles. We also discuss contextual factors in urban and rural settings that affected selection and functions of CAG members and community perceptions of the CAG.
Models of CAG vary in terms of both composition and selection processes. In relation to composition, CAG vary in terms of whether they represent the broad community or specific populations. 13 Similarly to the term community engagement, there is no widely agreed definition of a community. 14 As such, the term community can be externally defined to refer to: a group of people residing within a particular geographical location, a group of people with a common characteristic, identity or illness, or simply, a group of people residing within the immediate surroundings of a health facility. The ambiguity of the term 'community' therefore presents challenges on who should legitimately represent community's interests in health research.
Recommendations on composition of CAGs include having a group with equal numbers of representatives of the traditional authority, democratically elected residents and participant representatives; 15 a group of community leaders or a group with equal numbers of representative residents and participant representatives. 16 Selection approaches for CAG members also vary between contexts. A mix of purposive selection, elections and mixed methods approaches have been reported in the literature. 17 One of the recommended approaches to choosing representatives is through democratic elections. 18 Buchanan suggests that CAG members must be selected through democratic elections if they are to have authority to speak on behalf of the community. 19 While recommendations on both CAG composition and selection relate to ideas of representation, the concept of representation is complex, to researchers and vice versa, as well as to balance their conflicting interests have also not been adequately covered in the literature. This paper therefore seeks to contribute to these knowledge gaps.

| Setting
This study was conducted in an urban and a rural district in southern

| Data collection
The findings in this paper were part of doctoral research on community engagement in health research. Data collection took place be-

| Participatory workshops with CAG members and document reviews
We conducted two participatory workshops with CAG members from each district with an aim of understanding their roles in research. A total of 21 CAG members attended the workshop in the rural setting while a total of 16 CAG members attended the workshop in the urban setting (see Table 1). Workshop participants were purposively selected from a list of CAG members based on gender and geographical location. Both workshops were co-facilitated by the first and third authors. Workshop participants were asked to fill a registration form and a summary of socio demographic details has been provided in Table 1. At the workshop, participants were asked to discuss how they were selected, their roles in research and more specifically who they represent and how they represent them. Responses were noted to understand how participants were selected as CAG members while group discussions were used to understand their roles in health research. Workshop participants also shared experiences of how they executed their responsibilities.
Consent was sought from workshop participants to record workshop proceedings and each workshop lasted for about six hours.
Following this, we carried out document reviews of past CAG meeting reports in order to understand concerns raised by the CAG members. Preliminary findings were presented to science communication staff for feedback.

| Focus group discussions, interviews and participant observation
Main themes from the workshops, document reviews and discussions with science communication staff were further explored in subsequent data collection to broaden our understanding of the issues. We conducted eight focus group discussions with men and women from the two sites in order to explore diverse perspectives of how the CAG members functioned in the community. An additional 15 interviews were conducted with key informants to understand certain themes such as selection processes and roles of CAG members in more detail.
These key informants were selected based on their involvement in selected research projects and included community leaders, CAG members, research participants and research staff. Topic guides developed from the workshop themes covering issues of selection processes, roles, communication, community concerns and community representation in research were used to facilitate interviews and focus group discussions.
Data from participatory workshops, interviews and FGDs were triangulated with field notes from observations. Participant observation was used to understand both explicit and unarticulated aspects of how the CAG members functioned in the communities. We participated in activities involving researchers, community engagement staff, field workers, CAG members and community members to observe their interactions and to become familiar with the context where they lived. Interviews and focus group discussions were also recorded using a digital recorder and transcribed. A coding framework was developed by the lead author and transcripts were coded in QSR Nvivo 10. Main themes in relation to selection, roles of CAG and representation were used to support findings from the workshops. Findings were triangulated by using multiple data collection methods and crosschecking responses against various informants and the field notes. Results from this study were presented to CAG members for feedback in a separate workshop. We also sought their views on how to select CAG members in the future and empower them to effectively represent community's interests. Implementation of voting system for CAG members was easier in rural than in urban areas. This was because the villages or geographical locations in rural areas were small and communities were more homogeneous. Communities in rural areas were close-knit, shared the same tribe and language, and demonstrated similar socio-economic characteristics. In addition, people in rural areas were often long-term residents in a particular village, familiar with one another and usually available during community meetings because they were mostly farmers. Since a majority of people from rural areas were available during community meetings and they were familiar with one another; this made it possible to nominate and vote for people they trusted to serve as CAG members. However, while community meetings were easy to organize in rural areas, the election process was not always free and fair. For instance, some community members reported that at times some community leaders influenced their followers to vote for people from their clans which clearly raise questions about democratic selection.

| Challenges with selection of CAG members in urban and rural settings
In contrast, the settings where research was being implemented in urban areas were larger and more densely populated. A majority of people in urban areas were originally from other parts of the country and had migrated to urban areas in search of employment. In addition, there was also high in and out migration. Communities in urban areas were therefore diverse and comprised of people with different professions and tribes. Community leaders in urban areas reported challenges in inviting people for meetings, and general unwillingness of community members to attend community meetings: Only few people come to attend community meetings;

| CAG members' ability to perform their functions in urban and rural settings
We found two major constraints limiting CAG members' roles in performing both the intended role of representing community concerns and the perceived role of communicating study information: community awareness of the CAG, and CAG members' knowledge of the research that they were asked to report on.
We noted that community awareness of a CAG was essential to elicit concerns from community members. Most of the workshop participants in rural settings stated that community members were aware of them and approached them to report problems, seek advice and clarity on issues regarding research. This was seen to help demystify the research when information was obtained from fellow community members and enhanced trust and acceptability of research.
CAG members believed that they were able to clear misconceptions which improved acceptability of research as illustrated in the following quote: We are able to clarify misconceptions in the villages because we have relationships with community members.
For instance, there was a study in our community and people used to say that when they draw blood, they pay you back in exchange for the blood and we would say no, they are reimbursing you for transport to go to the clinic (Workshop participant, rural setting) In urban settings, however, we found that community members who participated in focus group discussions were not aware of the existence of a CAG. This obviously makes it difficult to relay community concerns to the researchers and could have been due to contextual factors discussed earlier, such as large geographical locations, dense population and lack of transparent selection approaches.
The second aspect that affected the CAG's ability to perform their role effectively had to do with communication of scientific procedures. While we noted that in many cases, most CAG members were not able to explain the concept of research accurately through the workshop and reports from previous meetings; we also noted that most CAG members had sometimes difficulties recall- These findings suggest that even though community members selected CAG members who were perceived as health literate, those selected as CAG members were sometimes unable to explain detailed research procedures to others. Overall, the discrepancies in expectations and CAG members' difficulties to perform their role of informing communities about research lead us to question whether CAGs are an appropriate mechanism to represent community interests and ethical concerns.

| DISCUSSION
This paper highlights practical experiences of selecting and engaging CAG in a low resource setting. Whilst democratic selection of community representatives is seen to strengthen the roles of CAG to represent community concerns in research, 25 these results have shown that this is problematic across different settings. Despite using democratic selection, we noted that the CAGs did not reflect all relevant socio-economic characteristics of the communities as intended. Since the CAG members were selected based on geographical location; the CAG did not include people who represented interests of other communities affected by the diseases being researched. In addition, rather than being a diverse group representing the demographics of the community, there were no members younger than 20 and a third of the members were community health workers in the urban setting. We also noted that most of the CAG members across both rural and urban areas (21 out of 37) had additional leadership roles in religious and other social groups; they were selected based on their perceived knowledge of health issues; for being known to others; or were preferentially put forward by powerful community mem- Our results suggest that neither democratic nor purposive selection approaches for CAG members led to effective representation of community concerns to reduce harm in research. Despite efforts to engage communities in research design, decision making was mostly done by researchers because of their expertise. 33 We appreciate that researchers are trained and paid to conduct quality research and are accountable to funders but researchers' obligation to conform to scientific procedures and international research ethics may render community representation ineffective if community feedback deviates from internationally acceptable research procedures. Given the findings from this study that CAG members did not provide a mechanism for collaborative partnerships between researchers and community, the question still remains if trying to establish genuine partnerships with communities using CAGs and share equal decision making power is desirable. 25 Since existing literature has shown that outcomes of similar models of CAG/CAB may vary across different contexts 34 ; several questions remain unanswered on how to optimise the engagement of lay communities and avoid tokenistic engagement across different contexts. In our case, we presented these findings to CAG members and sought their feedback on how to address the challenges. In order to improve representation of community members, CAG members advised that researchers must map social groups in a given context and purposively identify individuals to serve as CAG members. Alternatively, the CAG members suggested using existing selforganised community meetings in urban settings to elect CAG members. We however, acknowledge the limitations of these approaches in that they may not result in a representative group because youths, elderly people and other discriminated groups may be excluded.
Since we have shown that CAG members' role in representing community concerns may be compromised due to their expectations of employment and financial incentives from researchers, we propose that CAGs must be independent. As such, funding for operations of CAG must be unrelated to the projects that they advise on, solicited from independent sources or channelled through independent community based organisations. In order to improve their effectiveness to represent community interests, terms of reference for CAGs should be jointly developed with elected CAG members specifying roles of CAGs and the basis for sharing equal decision-making powers. Such terms of reference should also be made clear to field workers, research staff and community members to improve transparency and accountability. We also concur with other authors that capacity building for CAG members in research ethics, critical thinking and communication is essential for them to function effectively. 35 Most importantly, CAG members should be engaged in determining potential harms to their community and their feedback should be incorporated in research design.

| CONCLUSION
We used a qualitative study design to understand functions and effec-