Background

Methods

Data Extraction and Coding

All studies meeting the inclusion criteria were read in full by one researcher (NC), and relevant data, relating to stage and level of involvement as well as impact, were extracted using a structured data extraction sheet and a coding framework was developed using an inductive approach to examine the impact of PPI on research. Extracted data were then independently coded by a second lay researcher (RK) to validate the data being collected. Discrepancies in how the coding framework was applied to extracted data were referred to a third independent reviewer (MT) for review and consensus reached.

In this review, categories for stages of the research cycle were based on those described by the NIHR (see figure 1)11:

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

National Institute for Health Research research cycle.

For coding purposes, stages were categorised into four groups: pre-research (identifying, prioritising and commissioning); planning (designing and managing); undertaking research (undertaking); postresearch (disseminating, implementing and evaluating impact). Levels of involvement were coded using NIHR definitions12:

1. Consultation, which is asking the public for their views and using them to inform decision making.

2. Collaboration, which consists of an ‘ongoing partnership’ between research teams and members of the public, ‘where decisions about the research are shared’.

3. User-controlled research, which is ‘actively controlled, directed and managed by’ the public.

No quality assessment was performed on the studies, since the purpose of the review was to identify strategies and the impact of PPI rather than focus on the type or the quality of the study undertaken. However, in order to assist with the interpretation of the results, we extracted key information relating to research design (see online appendix 1).

Finally, as part of the review, we designed and led a workshop attended by LMIC partners from the Improving Mental and Physical Health Multimorbidity and Developing Research Capacity (IMPACT) Group on 7 December 2018 in York. In this, we presented our findings and led a group discussion covering PPI terminology, recruitment strategies and ways of reporting impact, all of which will feed into the IMPACT study design.

Patient and Public Involvement

This narrative review was supported by RK, who is a lay researcher who assisted as an independent coder and author. Additionally, LMIC partners from IMPACT reviewed the findings and contributed to the discussion, as described above.

Results

A total of 1969 studies were identified in the literature search (see figure 2 for Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram). After duplicate removal, 1314 abstracts (and full papers, if required) were screened based on the inclusion criteria by one researcher (NC), resulting in 1184 studies being excluded (see figure 3 for details of exclusion criteria). A total of 61 studies from 34 different countries were included in the narrative synthesis (see online appendix 1 and 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Preferred Reporting Items for Systematic Reviews and Meta-Analyses  flow diagram.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Exclusion criteria. LMIC, low/middle-income country; PPI, patient and public involvement.

Many of the included studies identified their method as CBPR (n=26) or PAR (n=11). The majority of the papers were primary research; either developing or piloting interventions (n=20), qualitative research (n=17), research methods/design (n=6) or research priorities (n=4). In these papers, discussion of PPI was generally secondary to reporting the study results and often comprised only a few sentences scattered throughout the paper. The remaining papers were reflections on research (n=14), particularly focused on community advisory boards and researcher experiences. None of the studies identified would meet the criteria of the only agreed reporting framework for PPI, the Guidance for Reporting Involvement of Patients and the Public (GRIPP)2 checklist.13

The review found that the most common stage to have PPI was the planning stage (n=51) followed by undertaking (n=30), postresearch (n=27) and, finally, pre-research (n=18). Regarding level of involvement, 37 studies were classed as collaboration, with only 4 being classed as user controlled and 20 as consultation (see online appendix 1). Most studies took place in the African subcontinent, followed by India. Using World Bank criteria, the countries where the research took place could be classified as low-income countries (n=8), lower middle-income countries (n=15) and upper middle-income countries (n=11)14 (see online appendix 2).

Studies with consultation-level involvement often had some form of community advisory committee that was used to inform the community about different aspects of the research and sometimes seek their opinion on the research objectives, design and implementation. Mushi et al describe presenting preliminary findings and intervention packages in village meetings as part of the consultation process.15 Similarly, Owolabi et al reports using a task force that included representatives of the Nigerian Stroke Society to ‘review the progress of the community participatory research process and make recommendations about any local adaptations to facilitate its effectiveness’.16

Collaboration-level involvement had varying strategies; some studies described a partnership between researcher and advocacy groups that ensured ongoing, two-way engagement between the community and the research team, with the community identifying the healthcare problem and driving forward solutions. Bradley and Puane research started after the community voiced concerns to community health workers (CHWs) about the increasing prevalence of hypertension and diabetes in their community.17 Following this, a meeting was held with the local community health committee and community leaders. Aims were formulated with the intention of engaging CHWs in many aspects of the research process, including data collection, analysis and dissemination. Zola et al reported that the ‘community based organisation members (CBO), people living with HIV (PLHIV) and researchers were involved, in an equitable partnership’.18 Community members were trained in research methods and ethics, and then involved in developing the questionnaire and conducting the interviews.18

Finally, looking at user-controlled research, common strategies included community-initiated research, involvement in the entire research project, from pre-research through to evaluation and ongoing involvement from peer researchers.19–22 Hayashi et al describe CBPR, which was led by a group of active and former drug users (Thai Drug Users’ Network) who were involved in the whole study from design to analysis and dissemination.22 Similarly, Jongudomkarn’s study was initiated by community members via a forum, and the women were involved in the entire research process right through to the action plan and evaluation.21

The most commonly cited impacts of PPI included increased relevance to the community, empowerment of participants and alterations in trial design. Foster et al  reported that the research team, consisting of US nurses, Dominican nurses and community leaders, continued to meet after the research had ended to drive improvement, hence empowering the team.23 After the conclusion of Jongudomkarn’s research, the women involved became the ‘resource persons responsible for alcohol consumption campaigns’.21 Liu et al explained that the community steering group modified the wording of some of the translations to ensure cultural relevance for a Mandarin-speaking population.24 Following feedback from the community reference team, Mosavel et al decided to ‘refocus the research from cervical cancer to "cervical health"’.25

Another benefit reported was improved quality of results; Bowling et al commented that the ‘partnership with local researchers and community partners strengthened the quality of the findings through their involvement in design, recruitment and interpretation phases’.26 Reflecting on higher levels of engagement in the community, Grinker et al described how PPI ‘facilitates the crucial recruitment phase as well as participant retention by limiting or managing negative views or misunderstandings of the researchers, procedures or goals of the study’.27 Another impact to consider is increased community trust and improved community-researcher relations. In Simwinga et al ’s South African study, community advisory board members provided a ‘protective role for community members’ and also helped ‘resolve tensions between researchers and community’.28 Furthermore, PPI can help challenge common community misconceptions and stigma; Adhikari et al noted that having local villagers involved in their malaria study helped tackle rumours in the community.29 Finally, some studies reported on the difficulties of PPI, largely focusing on the extra time and money required to have PPI in their study.22 30–32 It is important to note that most of the impact reported is from the researcher’s perspective and was often reported as an aim rather than an evidence-based outcome of PPI. In some cases, impact was not reported at all.

We thought it was important to consider whether the study authors were from the study country. The vast majority of studies (93%, n=57) had at least one author from the study country; 43 of these were in partnership with researchers from HICs. Only five studies were conducted without the input of local researchers. Although it is difficult to be certain from study reports, we estimate 10 of the 61 studies included non-governmental organisation or advocacy group members as authors.

Finally, we struggled to find funding guidance to identify whether PPI was a required component of the research. Some of the larger international funders mention public engagement as a concept in their material.33–36 However, many of the studies were funded by smaller local funders, for whom we were unable to find guidance.

Discussion

This review is the first to systematically review PPI in health research in LMICs. None of the studies made explicit reference to PPI as a term nor did they refer to the use of any tools or funding requirements. This could reflect an actual lack of PPI but may also reflect that research teams are simply not reporting PPI in research publications, or that researchers are using a different terminology for involvement activities. It is important to state that the poor reporting of PPI is not unique to research conducted in LMICs. Both Mockford et al and Brett have identified issues with the evidence base behind PPI, as well as the poor quality of reporting, which is essentially limiting our understanding of PPI impact on HICs.37 38

Nevertheless, the findings reveal that researchers are using PPI at different stages and levels in health research taking place in LMICs, highlighting that regardless of the subject, type or location of research, PPI can be integrated into the research process and may consequently have an impact on both the research and the individuals involved. There is a lack of reporting of PPI strategies and impact; those studies that do are still largely reporting impact from the researcher’s perspective.

Earlier research looking at the impact of PPI in the USA, UK and Europe aligns closely with the findings of our review of LMICs, suggesting that although the context may be quite different, benefits can be realised from PPI in LMIC. Brett et al discuss the impact of PPI throughout the research cycle, suggesting that PPI in the planning stage can help identify and prioritise topics according to relevance; in the implementation stage, it can help participant recruitment and researcher–participant rapport, and during analysis and write-up, it allows findings to be interpreted from a user perspective and can also assist with research dissemination.39

Our review identified that most PPIs in LMICs take place during the planning stage of research, which is in contrast to a similar review, not focused on LMIC, which found more examples of PPI in the execution phase.40 It is difficult to say definitively why this might be the case, but it may be related to non-LMIC researchers recognising the need to gain local knowledge in planning research in relatively unfamiliar settings.

PPI for many studies involved the setting up of community advisory groups/boards, but it is not clear if this is the most appropriate source of PPI or whether the formation of these groups acts as a barrier to meaningful engagement with the end user of the research, as the members of these groups were local community leaders, rather than those living with the particular health problem.

Furthermore, the involvement of local researchers was also apparent in the studies identified. This provided research teams with a mechanism to liaise at a grass roots level with local leaders in their local dialects and to remain alert to local sensitivities. In addition, LMIC researchers may gain an advantage by identifying topics relevant to the community and by gaining access and acceptance into the community, ultimately enabling the progression of the research project; this could also be used as a useful opportunity for research capacity building, which was mentioned in a couple of included studies.22 41 The overall relatively low level of patient involvement across the research cycle may be due to the importance, or lack of it, that funders place on PPI. Many of the studies reported in our review were supported by universities or smaller funders, for whom we were unable to find funding guidance. Larger, national and international funders do provide information about engaging communities but tend not to use the term ‘patient involvement’; rather they focus more on public engagement, which, as a concept, has some overlap with community engagement and PPI. For example, National Institutes of Health, USA, encourages community engagement as a ‘process of working collaboratively’.34 Similarly, guidance issued by the Australian Government Research Award Scheme emphasises the importance of ‘research engagement and communication’,35 and the UNC Centre for AIDS Research has a ‘Strategic Community Engagement Education Dissemination Office’,36 with similar aims.

It is therefore unsurprising that researchers focus on the engagement of community leaders as this aligns with those goals, rather than engaging with research participants. What is unclear in the literature from LMICs is whether these were the goals of PPI in these studies, or incidental benefits. The rationale given by many of these researchers for involving people at the design stage is often related to ensuring that gatekeepers support the research and facilitate access to the population of interest, rather than to improve the quality of the research per se.42–44 Many of the studies included in the review describe research in populations that would be defined as hard-to-reach groups, such as people with or at risk of HIV/AIDS, people with mental health problems (eg, schizophrenia) and people with drug/alcohol problems. In the UK, there is now a long-established literature setting out both consumerist and democratic reasons for involving patients and the public in research.45 It could be argued that in at least some of the studies, we found that a third reason dominated, and that was pragmatism.

Considering ways of increasing potential for impact, Brett et al reflected that good training and having clearly defined roles, in a positive supportive environment with mutual trust and respect would be beneficial.39Similarly, the EPIC study encourages researchers to set clear goals and well-developed plans and to advocate for individuals to be involved early in the research in a responsive, managerial way rather than solely to have a general oversight, more often seen in membership of a steering committee (the latter approach was widely used in the studies we identified in our review).46 Since many of the studies included in our review were CBPR, there was often a more hands-on type of involvement from participants, for example, trained community members taking on the role of researcher47–49 and participants shaping/piloting interventions50 51; this type of involvement will also fulfil the secondary aim of engaging participants in the research and perhaps develop local research capacity.

Finally, lack of standardisation in designing and evaluating PPI frameworks and strategies means it is difficult for researchers to develop a comprehensive PPI strategy. Over the last decade, researchers have been developing PPI toolkits but as yet, none have been adopted as a standardised tool.52–54 One of the more recent is the GRIPP2 checklist, which was developed following a systematic review and Delphi study to assist with the reporting of PPI in research, with the aim of improving quality and transparency of the PPI evidence base.13 The authors recommend that the checklist should be used prospectively in research design and retrospectively in evaluation. The GRIPP2 short form includes sections on aims, method, results (both positive and negative), discussion (impact) and reflection, each of these areas requiring information specific to PPI.13 However, it is important to note that this tool, though developed for international use, was developed from an HIC perspective, and though there are similarities, there are also complexities specific to LMIC that need to be considered, particularly the variations in research infrastructure, cultural differences, the power differential between researcher and researched in these contexts, and often, lower research budget. Other than cultural differences and research budget, none of the other areas were explicitly considered in the included studies.

The review suggests that there are positive gains to be had from involving communities from LMICs in research, and the complexities faced by LMIC research are things that PPI can help with, through facilitating communication with communities28 55 and adapting interventions for different cultures.50 56 PPI is still relatively new, even in countries with a well-established research tradition; it may take time for it to gain traction in countries without this tradition.

Conclusion

From this study, we can conclude that PPI does happen in LMIC health research but is generally described using a different terminology and rarely are detailed PPI strategies published. Similarly, at present the impact of PPI on both the participant and research is poorly documented. There is significant work needed to encourage closer engagement with end users of research, not just with community ‘gate-keepers’.

To improve this, LMIC research funders and journal publishers should make PPI an explicit requirement. Work also needs to be done on adapting pre-existing PPI tools for use in LMIC and encouraging their use, to clearly evidence the level, stage and impact of PPI. This will give researchers a generic format and space for reflection and will also capture the voice of the patient and the public to show how it has affected them as individuals, as well as the wider community.