Frameworks for supporting patient and public involvement in research: Systematic review and co‐design pilot

Abstract Background Numerous frameworks for supporting, evaluating and reporting patient and public involvement in research exist. The literature is diverse and theoretically heterogeneous. Objectives To identify and synthesize published frameworks, consider whether and how these have been used, and apply design principles to improve usability. Search strategy Keyword search of six databases; hand search of eight journals; ancestry and snowball search; requests to experts. Inclusion criteria Published, systematic approaches (frameworks) designed to support, evaluate or report on patient or public involvement in health‐related research. Data extraction and synthesis Data were extracted on provenance; collaborators and sponsors; theoretical basis; lay input; intended user(s) and use(s); topics covered; examples of use; critiques; and updates. We used the Canadian Centre for Excellence on Partnerships with Patients and Public (CEPPP) evaluation tool and hermeneutic methodology to grade and synthesize the frameworks. In five co‐design workshops, we tested evidence‐based resources based on the review findings. Results Our final data set consisted of 65 frameworks, most of which scored highly on the CEPPP tool. They had different provenances, intended purposes, strengths and limitations. We grouped them into five categories: power‐focused; priority‐setting; study‐focused; report‐focused; and partnership‐focused. Frameworks were used mainly by the groups who developed them. The empirical component of our study generated a structured format and evidence‐based facilitator notes for a “build your own framework” co‐design workshop. Conclusion The plethora of frameworks combined with evidence of limited transferability suggests that a single, off‐the‐shelf framework may be less useful than a menu of evidence‐based resources which stakeholders can use to co‐design their own frameworks.


| BACKG ROU N D
It is a truth universally acknowledged by policymakers, researchers and research funding bodies that patients and the public should be "involved" in research, though there are different perspectives on what such involvement should look like and why it should happen.
The first argument, described by some as normative and others as emancipatory, holds that patients have a right to have an input to research on their condition and that reducing the known power imbalances between researchers and patients is a moral duty of researchers, especially with oppressed and seldom-heard groups. [2][3][4]9 The second, which some have described as consequentialist or efficiency-oriented, 3 is that patient and public involvement, by bringing a real-world and lived-experience perspective, improves the efficiency and value of research via a number of mechanisms: increasing its relevance to patients; improving recruitment and retention rates of research participants; extending the range of people represented in research studies; and improving dissemination of findings beyond academic audiences 6,7,10,11 -though the evidence base for all these claims has been questioned. 10,12 The third argument is political and practical: that forming alliances with patients and the public is a defining feature of contemporary Mode 2 science (in which knowledge is co-constructed by scientists and citizens, often beyond the walls of the university 13 ); it increases the accountability and transparency of research and may be an effective way of attracting resources. 5,10 Notwithstanding the different (and to some extent incommensurable) perspectives represented by the above literature, it is clear that improving patient and public involvement in research is a high priority for research policymakers, 14-16 research funders, [17][18][19][20] researchers, 21-23 some academic journals 1 and patient and lay organizations. [24][25][26] Many of these groups have developed, or are in the process of developing, structured frameworks, tools, guidelines and checklists in an attempt to improve their own performance and (in some cases) critique or assess the performance of others.
As a multi-stakeholder research collaboration based in one of the UK's leading medical and biotechnology research regions, we had a strong commitment to strengthening patient and public input to our research. When we began this study, the UK National Institute for Health Research had recently put out for consultation its draft benchmarks for patient involvement in research. 27 The conditions of our funding required us to report annually on our patient and public involvement activity. We sought, therefore, to identify one or more tools or frameworks that would help us support, evaluate, improve and report on the patient and public involvement work of research teams across our collaboration.
An initial browsing search identified numerous potential frameworks in both academic and grey literature, many of which appeared to have been carefully researched and some formally validated and field tested. Different groups had produced different frameworks, drawing on different principles, applying different theories and prioritizing different potential use cases. It was clear that developing a new framework from scratch was almost certainly unnecessary, but that the existing literature could benefit from a taxonomy and improved accessibility.
Accordingly, we set out to achieve three objectives. First, to identify, critically examine, summarize and synthesize existing tools, frameworks, benchmarks, guidelines and critical appraisal checklists for patient and lay involvement in research. Second, to determine which of the frameworks were actually used and why (and explain why others were not used). Third, to work with patient and lay groups and designers to adapt, simplify and annotate existing frameworks and improve their aesthetic appeal and usability. As the study unfolded (and for reasons explained in the results section below), this last aim evolved to incorporate a major focus on optimizing the process of running workshops aimed at generating, adapting and operationalizing frameworks for involving patients and lay people in research.

| Study design
Narrative systematic review, drawing on the principles of hermeneutic review, 28 along with lay consultation and co-design. 29 Hermeneutic review consists of two interlinked cycles (described in more detail below): (a) accessing and interpreting the literature and (b) developing an argument. Searching is systematic but flexible and iterative. As sources accumulate, it becomes necessary to interpret, clarify and understand the emerging ideas and perspectives and to reject less relevant sources through progressive focusing.
We have argued elsewhere that narrative review, which adds successive primary studies to an increasingly rich picture of a complex field of study, is the method of choice for synthesizing and making sense of a large and diverse body of primary literature where different groups of authors have approached the topic in very different ways. 30  We also searched selected grey literature sources (eg, guidance produced by national and international patient organizations and advocacy groups, health services or think tanks), collated sources already known to the authors and put out requests to our professional networks (including social media followers). When we identified papers that met our inclusion criteria, we checked the references of those papers and also put the title into Google Scholar to subsequent citations of it (an "ancestry and snowball" approach 32 ). Where papers cited a specific theoretical underpinning, we obtained the original paper describing that theory. If a framework had been described in both academic and grey literature, we included only the former.

| Inclusion and exclusion criteria
We included any published, systematic approach designed to inform, evaluate or report on patient and public involvement in health-related research. There were no language restrictions. The main exclusion criteria were as follows: not a framework, not about research or lacking provenance (ie, unable to trace its source). We excluded frameworks that were focused only on communication or engagement (defined as explaining research to the public) as opposed to involvement (involving patients and the public in some way in planning, undertaking and disseminating research). Largely for practical purposes, grey literature was limited to publications from national or international organizations (eg, James Lind Alliance, INVOLVE, Canadian Institute for Health Research).

| Data extraction and appraisal of quality
We used an Excel spreadsheet to summarize key aspects of each study (both theoretical and empirical). For each framework, data were extracted on the rationale for its development; provenance (including funding/sponsorship); patient/public input; theoretical basis (if any); orientation (initially using the taxonomy set out in the background above: "emancipatory," "efficiency-focused" and "practical," and evolving as new categories emerged); fields and topics covered; format and accessibility; intended user(s) and purpose(s); examples of use; and critiques. Three reviewers (TG, AM and LH) undertook data extraction; each study was looked at by two reviewers with disagreements resolved by discussion. We attempted to contact lead authors of all papers to ask whether and by whom the framework had been used since its publication.
Using data from these domains, we applied the Canadian Centre for Excellence on Partnerships with Patients and the Public (CEPPP) evaluation tool, which assesses four aspects of a tool or framework 23 : • scientific rigour (graded as 3 = good, 2 = moderate or 1 = weak); • incorporation of patient/public perspective (graded as 3 = extensive, 2 = limited, 1 = absent or not reported); • comprehensiveness (graded as 3 = good, covering all intended dimensions; 2 = limited, covering only some key dimensions; 1 = very limited); and • usability (graded as 3 = good, extensive evidence of use beyond the study in which it was developed; 2 = emerging [for recently published frameworks with some evidence of use]; 1 = limited or unknown).

| Analysis and synthesis of primary literature
Using the iterative hermeneutic methodology developed by Boell and Cecez-Kecmanovic, 28 we built an overall picture of the different kinds of frameworks and their strengths and limitations, adding detail and nuance as successive studies were incorporated.
As an example of our approach, our hand search turned up a paper by Staniszewska et al 33 on the GRIPP1 (Guidance for Reporting Involvement of Patients and Public) framework for structuring how researchers report lay involvement in a clinical trial. 33 Through citation tracking of that paper, we identified a number of additional "report-focused" frameworks, [34][35][36] including GRIPP2. 34 Whilst we initially grouped all these as a subset of frameworks designed for planning and organizing patient involvement in research studies ("study-focused"), detailed analysis revealed that these were separate categories with limited cross-referencing between them.
We synthesized a preliminary set of resources based on the frameworks in our data set. To inform the practical workshops, rather than reproduce all the frameworks (since many covered similar ground), we worked with lay colleagues to select the "best-inclass" from different categories in our data set. In this process, we were guided by three questions: (a) did the framework score well using the CEPPP tool (see above)? (b) does it make sense to patients and lay people as well as researchers-and is it potentially usable by both? and (c) will it allow valid measurement and iterative improvement of patient and public involvement work by research teams?

| Co-design phase
We shared our preliminary set of 12 "best-in-class" resources in two preliminary 2-hour development workshops attended by a total of 16 participants recruited from three local pre-existing academic-lay research partnerships (including researchers, patient involvement leads, patients, carers and advocates). We adapted the interactive and participatory methodology described by previous authors. 29,37 Prior to the first workshop, we made large-scale diagrams of the different "best-in-class" frameworks from our systematic review and invited the groups to talk about them and use sticky notes to annotate them. We systematically captured and incorporated their suggestions for adaptation, and sought input from a professional design service to produce resources in multiple formats.
The workshop materials, suggested format, resources and facilitator notes produced in the two development workshops were refined through three further pilot workshops in contrasting clinical and research settings: a long-established patient participation group for a specialist research group in blood disorders; a recently established lay partner group for a community-based mental health research programme; and an academic-lay-industry partnership seeking to establish working principles and evaluation methods for lay participation in industry-led clinical trials. Full details of these workshops will be presented in a separate paper.

| Description of data set
The study flow chart is shown in Figure 1. Of over 5000 titles, 150 papers were retrieved in full text; this sample was extended to 250 using ancestry and snowball searches. After applying exclusion criteria, our final data set consisted of 64 papers describing 65 frameworks from 10 countries (one paper described The included publications described toolkits, tools, frameworks, checklists, benchmarks or maps for informing, guiding, assessing or reporting on patient and/or public involvement in research. A total of 56 frameworks were written up in 55 academic papers. 5,10,12,31,[33][34][35][36][38][39][40][41][42][43][44][45][47][48][49][50]52,53,[55][56][57][60][61][62][63][64][65][66][67][69][70][71][72][73][74][75][76][77][78][79][80][81][82][83][84][87][88][89][90]92 Of these, 44 were available open access. Nine frameworks were in the grey literature, all of which were publicly available. 20,27,46,51,54,58,68,85,91 The data extraction and scoring spreadsheet for the 65 frameworks is available from the authors. Almost all frameworks in our sample scored moderately or very highly on the CEPPP tool for scientific rigour (our scoring acknowledged a wide range of study designs). Most frameworks had been developed using a systematic approach with substantial input from patients or lay people, though approaches used varied considerably. Some groups had used primary qualitative research 50,55-57,81,84 and/or qualitative, thematic or narrative literature review, 5 71 Other frameworks had been developed in a more pragmatic way by working groups (typically involving lay people, researchers and/or research funders) with extensive consultation but without an in-depth review of the relevant academic literature. 20,27,36,59,69,78,91 Some groups used a combination of literature review, qualitative research and workshops. 31,34,53,58,65,72,73,75,77,[80][81][82]85,88,89 Some covered all lay involvement; others were restricted to specific groups such as older people, 61,80 those with a specific clinical condition, 36,59,72,75,84,87 those with or at risk of a genetic condition 72,85 or underserved or marginalized groups. 65,69,76,77,79 Most grey literature frameworks gave limited details of methodology, though one drew on academic sources 91 and two described and referenced a literature review. 46,54 Four papers proposed a "framework of frameworks" taxonomy of approaches to patient and public involvement in research (see Discussion for details). [62][63][64]83 The remaining 61 frameworks could be grouped into five main categories (though several had features of more than one): In the first and last of these, the presumed unit of analysis was a partnership (actual or desired). In the middle three, the presumed unit of analysis was a research study (usually, a clinical trial). In Table 1, we summarize the features of the five categories of framework, highlighting the ones we selected as "best in class" (high score on CEPPP tool and liked by our patient advisers). Below, we describe the frameworks in each category in more detail, giving one example of each. The other "best in class" frameworks are reproduced in the Appendix S1.
The number of publications per year in our sample is shown in Figure 2. Despite there being no date limit on our database search, no frameworks had been published before 2003. Since then, the number published annually has grown steadily. There has been a recent steep rise in the publication of study-focused and partnershipfocused frameworks and (in 2018) reviews of frameworks.

| Power-focused frameworks
Thirteen frameworks (eight from UK 5,39-45 and five from United States [65][66][67]76,77 ) were developed by academically led teams whose primary interest was studying and challenging power differentials in researcher-lay partnerships. They applied theories from critical sociology such as Foucault (who proposed that knowledge and power are intimately related), Habermas (who explored the concealed power games between those based in the "system" and those outside it) or Bourdieu (who wrote about different kinds of social and cultural capital, of which specialized knowledge is one component), or from critical public health (notably, theories of power-sharing in community-based participatory research).
Whilst power-focused frameworks addressed similar domains to those in other categories in our taxonomy, they asked more TA B L E 1 Taxonomy of frameworks for supporting and evaluating patient and public involvement in research

Category with selected "best in class" examples Main focus of frameworks in this category Comment
Power-focused frameworks Oliver et al 44 Morrow et al 42 Gibson et al 40 Gradinger et al 41  Frameworks in this category tend to link a specific value or principle with a particular set of metrics of involvement and impact radical questions. They were particularly interested, for example, in surfacing power imbalances, values and hidden motives. For example, they did not merely talk about "empowering" patients and lay people; they asked questions such as "who gets to define what empowerment is?" and "whose interests are served by so-called empowerment?" The earliest power-focused framework in our data set was led by Oliver's group at the Institute of Education. It was published in 2004 44 and updated in 2008 39 (see Appendix S1 for diagram).
Drawing on Arnstein's ladder of participation, they rated lay input on a continuum from none to consultation to collaboration to control 93 ; they also used Mullen's distinction between proactive and reactive behaviour by researchers (researchers could invite lay groups, invite individuals, respond to lay action or do little or nothing). 94

| Priority-setting frameworks
Eight frameworks, from Canada, 78 Netherlands, 86  decide whether a comprehensive approach is needed at all), efforts to include the voices of marginalized groups, and included a followup evaluation (see full framework in Appendix S1). 89 More recently, researchers from Zambia published a similar health system-oriented framework based on a systematic analysis of previous frameworks and two user workshops. 88 The UK-based James Lind Alliance developed a framework A similar framework, oriented to priority-setting in comparative effectiveness research, was produced by PCORI in the United States. 68 Pollock et al 36 50 and one on the involvement of patients in setting clinical outcomes. 74 One considered the economic costs and benefits of lay involvement in different phases of a clinical trial 71 ; and one addressed how to maintain recruitment to successive trials over time. 49 A framework for improving patient engagement in Alzheimer's disease trials highlighted specific challenges with this target population and offered solutions based on a literature review. 87 Most studies in this category were funded by bodies that sponsor clinical trials and/or seek to ensure patient input to such trials.

These include the UK National Institute for Health Research (NIHR)
Research Design Service (Figure 4), 51 PCORI in United States, 70 and international development funders. 90 All publications in this category emphasized that, in the view of the authors, patient and lay input throughout a clinical trial would make the trial more relevant, more appealing to potential participants, more likely to reach its target recruitment, more likely to retain participants and more likely to generate and disseminate highquality research knowledge.
Whilst study-focused frameworks differed in detail, common They applied this in a novel way to patient engagement in the research process. They found that more patient involvement substantially lowers the chances (and hence the cost) of protocol amendments and also improves the participant experience, leading to fewer withdrawals from the study (again, with major predicted cost savings).  Figure 5, through which key aspects of patient and lay involvement can be tied to specific and measurable outcomes (see Appendix S1 for a table of specific metrics). For example, asking patients to feed back on the wording of questionnaire items (process) would be expected to increase the completion rate (outcome) and hence the robustness of the findings (impact).

| Report-focused frameworks
Four frameworks, all from UK, offered a checklist for critically appraising a published study for the quality and comprehensiveness of patient and lay involvement. Three covered primary studies [33][34][35] ; and one covered systematic reviews. 53 All addressed (at least in broad outline) the structure of a clinical trial report (eg, rationale, methodology, findings, discussion, evaluation or reflection) or systematic review equivalent. Two were produced by the GRIPP team as part of the EQUATOR network; the latest version is the 34-item long-form GRIPP-2 checklist (shown in short form in Table 2). 34 Report-focused frameworks for primary studies assumed that the research design was a clinical trial; they addressed the same elements in broadly the same way as study-focused frameworks, but did so retrospectively (as a quality checklist) rather than prospectively (to guide activity).

| Partnership-focused frameworks
Seventeen frameworks (from United States, 10 Canada, 20,82 Australia, 31  reproducibly. Most such frameworks placed particular emphasis on governance, public release of data (transparency) and accountability.
Details of some exemplar partnership-focused frameworks are listed in the Appendix S1.
The James Lind Alliance (described above in the "Priority-setting" category above) was one of the first groups to propose some ing researcher-lay partnerships more generally (see Appendix S1 for details). These early initiatives are typical of approaches that seek to deliver what Daniels has termed "accountability for reasonableness" in the public sector-that is, demonstrating a systematic, transparent and auditable process through which citizens and service users can contribute to, and help oversee, the work of a public body. 98 A number of academic-lay partnerships have produced similar frameworks, typically as a result of hybrid funding from academic, service and patient organizations. 31,41,55,[59][60][61]77,79,82,84,85,91 Common themes in this category included governance mecha-  Table 3, and the standards are reproduced in full in the Appendix S1.

| Evidence of framework use
The only dimension of the CEPPP tool on which a high proportion of frameworks scored poorly was usability (which we interpreted to include actual evidence of use). Power-focused frameworks were rarely used directly, but they informed and underpinned subsequent work on more applied categories of framework. 46 Technology Assessment programme as "best practice." 38 The most recent report-focused framework (GRIPP2 34 ) is recommended by several leading journals, though few currently make its use mandatory. Three frameworks that were developed within a particular clinical field (elderly care, 80  In sum, frameworks to guide patient and lay involvement in research developed in one setting do not appear to have transferred readily to other settings, except when they have been oriented to a specific clinical field and actively disseminated within that field.
Our data set also revealed a number of examples of efforts to operationalize a theoretically derived framework using some kind of practical workshop. For example, the Public Involvement Section and topic Item

Aim
Report on the aim of PPI (patient and public involvement) in the study

Methods
Provide a clear description of the methods used for PPI in the study

Study results
Outcomes-report the results of PPI in the study, including both positive and negative outcomes

Discussion and conclusions
Outcomes-comment on the extent to which PPI influenced the study overall. Describe positive and negative effects 5. Reflections/ critical perspective Comment critically on the study, reflecting on the things that went well and those that did not, so others can learn from this experience  ing with well-established patient involvement groups, produced a definitive framework, which suggests that a frameworking process is likely to require a series of facilitated workshops, not a one-off event. Additional findings from the workshop study (which is ongoing) will be reported in a subsequent paper.

| Summary of principal findings
This study, which to our knowledge is the first attempt at a comprehensive synthesis of frameworks for supporting patient and lay involvement in health research, has produced four main findings.
First, well over 60 frameworks already exist, many though not all of which have been robustly developed using both theoretical principles and extensive patient and lay involvement.
Second, we have developed a new taxonomy of these frameworks-power-focused, priority-setting, study-focused, report-focused and partnership-focused-based on their primary focus and intended purpose.
Third, we have ascertained that most published frameworks have been little used beyond the groups that developed them (with the exception of frameworks oriented to a particular clinical field and disseminated via networks within that field).
Finally, we have refined a provisional format and set of resources for an evidence-based "develop your own framework" workshop to be run adaptively by researcher-lay partnerships.
Whilst the frameworks in our data set were developed in dif-

| Comparison with other studies
Four previous "framework of frameworks" publications offered a taxonomy of published approaches to patient and lay involvement in health research, though each took a narrower focus than our own review. In an early non-systematic review (written when only six of the frameworks in our sample had been published), Savory arranged previous literature broadly along two axes: focus of involvement (patient, carer, group, interested layperson, general public) and purpose of participation ("on," "with," "by" and "led by" lay people). 64 Fransman explored various theoretical discourses used to analyse public engagement in research (not limited to health). 63 Hughes and Duffy used concept analysis to consider how power-sharing had been theorized in previous public involvement frameworks. 62 Boivin et al 83 summarized and critiqued evaluation tools for patient and lay involvement in research.
The emerging literature on the use of practical workshops in knowledge creation helps explain why our focus on building one's own framework appeared to be more successful than inviting groups to use off-the-shelf frameworks. This literature includes reviews of approaches to co-creation of knowledge 102 and the sociology of design, 103 and (more specifically relevant to our empirical work) a recent theorization of "collective making." 29 In the last of these, • influence on the process of implementation (the intervention generated through creative play is "owned" by end-users; the intervention incorporates research, experiential and contextual knowledge and comes with the testimony of end-users who were involved in the making; it includes a "boundary object" in physical or visual form that acts to engage others beyond the co-design group; and it typically includes "core" and "adaptable" elements).
The shift in our focus from identifying published frameworks to supporting local co-design of frameworks reflects an emerging philosophical shift in the way knowledge is conceptualized: from a highly objective view of knowledge (positivism, which views knowledge as "facts" that are empirically derived and to a large extent context-independent) and a more subjective view (interpretivism, which views knowledge as socially constructed and perspectival) to a hands-on view of knowledge (known as performative and defined as something that is brought into being in and through human action). 102 In other words, actively building a framework may be more effective and enduring than attempting to apply someone else's framework. Van de Ven and Johnson 104 explain how the principles and philosophy of pragmatism (attending primarily to the practical and context-dependent use to which the outputs of practical work will be put) can aid a performative approach to collaborative knowledge creation: "By exploiting multiple perspectives, the robust features of reality become salient and can be distinguished from those features that are merely a function of one particular view or model" (page 810).
Ours is not the first study to grapple with the tension between an academic ideal and a local, pragmatic solution. Deborah Ghate recently described an attempt to co-produce a parenting programme that was both "evidence-based" (ie, drawing on the research literature, which in this case was characterized by intensive interventions that were difficult to replicate and prohibitively expensive) and "home-grown" (ie, co-produced by local practitioners and the groups they sought to serve, taking account of contextual realities and resource constraints). 105 Published research evidence was used to develop a sophisticated theory of change that was fed into local activities to produce what Ghate called "evidence-supported design."

| Strengths and limitations
To our knowledge, this is the most comprehensive and systematic summary of patient and lay involvement frameworks yet published.
The literature search was extensive and used multiple methods (including database searching, hand searching and citation tracking) to amass an extensive primary data set. Through detailed data extraction and theoretical analysis, we have produced a new taxonomy into which future studies can be classified-and which has the potential to be extended if other groups develop new approaches to exploring the field. This is also the first systematic review on this topic to have gone beyond an academic synthesis: we produced, and empirically tested, a set of resources intended for use in practical workshops, allowing different researcher-lay partnerships to draw on them in different ways through evidence-informed serious play.
One limitation of this review is that few primary studies were based in low-or middle-income settings. A reviewer of an earlier draft of this paper suggested that not all countries or settings have a strong culture of patient involvement in research, so frameworks or framework-building activities that implicitly assume such a culture may have limited success.
Another key limitation of this study is that the empirical component reported here was preliminary. We tested the practical resources on only three researcher-lay partnerships, all of which were linked to the University of Oxford and did not represent the potential range of diversity of such partnerships. Whilst we believe we have demonstrated proof of concept for our "co-design your own framework" approach, we invite other groups to explore their use of our workshop resources and facilitator guides in a wider range of target groups and settings. We have made these resources available free for download from the Health Expectations website to those using them in non-profit initiatives.

| CON CLUS ION
This study has shown that numerous published frameworks for supporting and evaluating patient and public involvement in research already exist. They have different provenances, intended purposes, strengths and limitations. But being evidence-based and theoretically informed is no guarantee that a framework will be used. A single, onesize-fits-all framework may be less useful than a range of resources that can be adapted and combined in a locally generated co-design activity.
We suggest that those who seek to develop or strengthen the patient or lay involvement in their own research use a three-step process. First, explore the published examples described in this paper and the Appendix S1. Depending on context and intended use case, a framework may be found that is fit for purpose-perhaps with some adaptation. In the absence of such a framework, download and study the facilitator guide and evidence-based resources and prompts, which are based on the findings of this review. Finally, work with patient collaborators and (ideally) professional facilitators to plan and deliver a series of co-design workshops to generate a locally relevant and locally owned framework.

ACK N OWLED G EM ENTS
We thank the patients, lay people, researchers and research support staff who attended the co-design workshops, without whom this study would not have been possible. We also thank the authors of primary studies, many of whom helpfully provided additional contextual data on their study (included where relevant in the spreadsheet in the Appendix S3). We thank Robert Macfarlane for help with hand searching.

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C S A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
This study is part of the "Partnerships for Health, Wealth and