What is the volume, diversity and nature of recent, robust evidence for the use of peer support in health and social care? An evidence and gap map

Abstract Background Peer support interventions involve people drawing on shared personal experience to help one another improve their physical or mental health, or reduce social isolation. If effective, they may also lessen the demand on health and social care services, reducing costs. However, the design and delivery of peer support varies greatly, from the targeted problem or need, the setting and mode of delivery, to the number and content of sessions. Robust evidence is essential for policymakers commissioning peer support and practitioners delivering services in health care and community settings. This map draws together evidence on different types of peer support to support the design and delivery of interventions. Objectives The aim of this map was to provide an overview of the volume, diversity and nature of recent, high quality evidence on the effectiveness and cost‐effectiveness of the use of peer support in health and social care. Search Methods We searched MEDLINE, seven further bibliographic databases, and Epistemonikos for systematic reviews (in October 2020), randomised controlled trials (in March 2021) and economic evaluations (in May 2021) on the effectiveness of peer support interventions in health and social care. We also conducted searches of Google Scholar, two trial registers, PROSPERO, and completed citation chasing on included studies. Selection Criteria Systematic reviews, randomised controlled trials and economic evaluations were included in the map. Included studies focused on adult populations with a defined health or social care need, were conducted in high‐income countries, and published since 2015. Any measure of effectiveness was included, as was any form of peer support providing the peer had shared experience with the participant and a formalised role. Data Collection and Analysis Data were extracted on the type of peer support intervention and outcomes assessed in included studies. Standardised tools were used to assess study quality for all studies: assessing the methodological quality of systematic reviews 2 for systematic reviews; Cochrane risk of bias tool for randomised controlled trials; and consensus health economic criteria list for economic evaluations. Main Results We included 91 studies: 32 systematic reviews; 52 randomised controlled trials; and 7 economic evaluations. Whilst most included systematic reviews and economic evaluations were assessed to be of low or medium quality, the majority of randomised controlled trials were of higher quality. There were concentrations of evidence relating to different types of peer support, including education, psychological support, self‐care/self‐management and social support. Populations with long‐term health conditions were most commonly studied. The majority of studies measured health‐related indicators as outcomes; few studies assessed cost‐effectiveness. Studies were unevenly distributed geographically, with most being conducted in the USA. Several gaps were evident regarding the delivery of peer support, particularly the integration of peers and professionals in delivering support and interventions of longer duration. Authors' Conclusions Although there is evidence available to inform the commissioning and delivery of peer support in health and social care, there are also clear gaps that need to be addressed to further support provision, particularly regarding cost‐effectiveness. The effectiveness of peer support in different countries, with varying health and social care systems, is a priority for future research, as is the integration of peers with professionals in delivering peer support.

1 | PLAIN LANGUAGE SUMMARY 1.1 | Evidence and gap map finds unevenly distributed evidence on effectiveness of using peer support in health and social care There is evidence related to educational and emotional peer support, and peer support interventions for people with long-term health conditions, but there are considerable gaps in evidence for peer support in countries and health care systems beyond North America.

| What is this evidence and gap map (EGM) about?
Peer support-people drawing on shared personal experience to help one another-can directly benefit individuals, reducing social isolation and mental ill-health, and potentially improving management of longterm health conditions. If it increases the effectiveness with which people manage their conditions or address their own needs, it could also lessen demand for health and social care services.
However, peer support varies in its design and delivery, the problems or needs that it seeks to address, the setting in which it takes place, the extent to which it is linked to formal care services, and the number and content of sessions. This variation creates a challenge to those seeking to provide peer support services, such as policymakers and practitioners, in finding and understanding evidence relevant to the type of peer support they are planning to deliver to aid decision making.

What is the aim of this EGM?
This EGM is a visual resource presenting recent, high quality evidence on the effectiveness and cost-effectiveness of the use of peer support in health and social care.

| What studies are included?
This EGM includes systematic reviews and impact evaluations (randomised control trials and economic evaluations, if not already included in a systematic review) on the effectiveness and costeffectiveness of peer support.
Included studies are published in English, conducted in highincome countries, and focused on adult populations (aged 18 and over) with a health or social care need.
Any type of peer support was included, as long as peer supporters had the same or a similar health condition as the person they were supporting and had received training, had a contract, or received ongoing support.
The map includes 91 studies: 32 systematic reviews (including 242 impact evaluations) and 59 impact evaluations (52 randomised controlled trials and 7 economic evaluations).

| What are the main findings of this EGM?
Studies included in the map varied in quality; the included systematic reviews and economic evaluations tended to be of low or medium quality, whilst randomised controlled trials were generally of higher quality.
The most investigated peer support interventions included education, emotional and wellbeing support, help with self-care and selfmanagement, and social support. Few studies, particularly systematic reviews, looked at case management by peers. People with long-term health conditions were the most frequently studied population.
Studies looked mostly at the effectiveness of peer support in improving health, both physical and mental, as well as wellbeing and social connection. Few studies examined the cost-effectiveness of peer support. The majority of studies took place in the USA.
On the delivery of peer support, studies tended to focus on inperson peer support. There was a gap regarding the integration of peers and professionals in delivering support.
The evidence on long-term peer support was limited, with studies generally focusing on short interventions of up to 3 months or up to 6 months.

| What do the findings of the map mean?
This EGM provides information for policymakers and practitioners commissioning or delivering peer support. It also indicates a need for more research on the cost-effectiveness of peer support, on different ways of delivering it, and in countries beyond the USA.

| How up-to-date is this EGM?
The authors searched for systematic reviews published from 2015 to 2021 and impact evaluations published up to 2021. The importance of taking a person-centred approach to healthcare is increasingly recognised by health and social care services worldwide, with the World Health Organization publishing a global strategy in (World Health Organization, 2015. Engaging and empowering patients to take a more active role in their own care is particularly important in the context of the increasing rates of non-communicable diseases (NCDs), along with ageing populations, seen in many countries (Vos et al., 2020). These mean that people are living with health conditions for longer periods of time, placing an increasing burden on healthcare systems (World Health Organization, 2015).
Whilst public health has tended to focus on reducing mortality, disability due to NCDs is becoming a greater problem, with a corresponding need to develop interventions to address this issue (Vos et al., 2020). Peer support, which in its simplest form is support exchanged by people who have a shared experience (Mind, 2013), has the potential to be both long-term and low cost, and is an example of a person-centred intervention which could support people to manage their own health (Mind, 2013;World Health Organization, 2015).
In the UK, financial pressure on the NHS and social care services is increasing, with the impact of slowed growth and reductions in funding (Health Foundation Kings Fund & Nuffield Trust, 2018) exacerbated both by increased demand from an aging population and the consequences of Covid-19 NHS Digital, 2016). There was an estimated annual funding deficit of almost £30 billion within the NHS in 2020/2021 (NHS England, 2014), with a minimum annual increase of 4% recommended to maintain and improve services . Social care organisations have sought to reduce the impact of reductions in funding and increasing costs associated with complex care (Cromarty et al., 2019).
Yet despite an injection of ring-fenced funding totalling £10 billion between 2017/2018 and 2019/2020 (Local Government Association, 2017), it is estimated that an annual funding increase of at least £3.9 billion is needed by 2023/24 to meet current funding shortfalls in social care (House of Commons Health and Social Care Committee, 2020). Part of the policy response to these pressures is increased focus on enabling patients and carers to support themselves more effectively .
The shift in policy and practice towards more patient-centred and person-centred care over the last few decades, within the UK and more widely, recognises the capabilities and knowledge of people who manage their conditions themselves and the need to connect with others who have similar problems or care needs (Wilson et al., 2007). The NHS Long Term Plan (for England) outlines an intention to empower patients through increasing their involvement in their own care, focusing on patients' own health and wellbeing goals, improved access to information, and peer support within the community (NHS England, 2019a). This is increasingly important given the prevalence of long-term, often co-occurring, conditions including diabetes and heart disease in the population, and the need to enable and empower people to self-manage associated lifestyle risk factors.
Whilst the use of peer support-based initiatives is currently increasing in the UK, their emergence to complement or substitute for formal, professionally delivered services has a longer history. An early example, launched by the UK Department of Health in 2002, is PRICE ET AL. | 3 of 51 the Expert Patients Programme, a 6-week course for people with longterm health conditions which was adapted from the Stanford Chronic Disease Self-Management Programme (Griffiths et al., 2007;Wilson et al., 2007). In mental healthcare, research suggests the 'recovery movement' and recovery colleges-groups of people with shared experience of living with mental health problems who help each other understand and care for themselves-can aid recovery (Meddings et al., 2015), including reducing service use (Bourne et al., 2018).
Globally, peer support interventions have been used in a range of populations, to address mental health difficulties (Cabassa et al., 2017;, physical health problems including HIV (Kanters et al., 2016) and cancer Lee & Suh, 2018), as well as other health and social care needs such as those of carers (Dam et al., 2016), or new mothers (Leger & Letourneau, 2015). An early example is the Stanford Chronic Disease Self-Management Programme, which was developed in the USA in 1979, originally using lay people to teach self-management for arthritis before being adapted for chronic diseases more widely (Griffiths et al., 2007).

| The intervention
The potential to make use of the 'renewable energy' of the individuals and networks available within the wider community has been recognised within formal services for people with long-term physical conditions such as HIV and diabetes (Health Foundation & Nesta, 2016), with guidelines and research to support the development of peer support initiatives also being developed for other population groups, including people living with dementia (Health Innovation Network, 2015), those experiencing mental health difficulties (NHS England, 2019b;World Health Organization, 2017), and women requiring peri and/or postnatal support (Hetherington et al., 2015;Woodman & NHS Health Scotland, 2013 …people drawing on shared personal experience to provide knowledge, social interaction, emotional assistance or practical help to each other, often in a way that is mutually beneficial (Nesta & National Voices, 2015).
This definition provides a conceptual explanation of peer support but in terms of practical delivery there are a huge variety of peer support interventions (Nesta & National Voices, 2015). Dennis outlines how the different ways people can access peer support lie upon a continuum , with interventions varying in: -Delivery formats, which include face-to-face groups or one-toone contact, online forums, telephone, and email (Mind, 2013;Valenstein et al., 2016). Peer support may also be uni-directional (one peer provides support but does not receive it) or bidirectional (both peers provide and receive support) (Valenstein et al., 2016).
-Content, including the degree of structure (Valenstein et al., 2016), with interventions focusing on providing information and education, emotional support or a combination (Nesta & National Voices, 2015). Differences are a result of the needs of the population, structure of existing services, resources available and intended outcomes.
-Support from, or association with, formal health and social care services. Peer supporters can be paid employees recruited by health care organisations or third sector agencies, or volunteer workers.

| How the intervention might work
Qualitative evidence provides some insight into the underlying mechanisms or processes which may underpin the effectiveness of peer support interventions. In their scoping review synthesising evidence regarding the processes or mechanisms underlying one-toone peer support for adults using mental health services, Watson identified five mechanisms: 'lived experience, love labour, the liminal position of the peer worker, strengths-focussed social and practical support, and the helper role' (Watson, 2019). See Box 1.
The importance of lived experience is also highlighted by Bailie and colleagues, in a study of users of peer support which focused on the core role of the relationship between the professional peer support worker and service user (Bailie et al., 2016). The study emphasised how a shared sense of identity can be developed through disclosure of past experiences, deepening the relationship and helping the service user feel understood. Peer supporters also bring perspectives relating to shared sociodemographic characteristics such as gender and ethnicity (National Collaborating Centre for Mental Health, UCLPartners, Care City Health Education England, & PPL Consulting, 2020); an understanding of the cultural background of the potential participants of proposed peer support may be beneficial in designing interventions that work (Jamison et al., 2017;Mayer et al., 2019).
Peer support-based interventions and initiatives seek to tackle different problems and needs, and are therefore designed and delivered in different ways. Accordingly, we should expect variation between interventions in the outcomes achieved and likely mechanisms underlying these. Outcomes resulting from peer support interventions range from direct benefits to the health of individuals, both physical and mental (Health Foundation & Nesta, 2016;, to intermediate outcomes, such as empowerment, improved self-management, or social inclusion (Mind, 2013), as well as increased knowledge about a health condition . For example, reviews of peer support for Type 2 diabetes have found reductions in blood glucose and increased diabetes knowledge as a result of peer support interventions (Gatlin et al., 2017;Krishnamoorthy et al., 2019). Additionally, some peer support programmes have resulted in reduced use of health services Mind, 2013).
It should be noted, that in some cases, interventions may have no effect, or even negative impacts (Burke et al., 2019;Chien et al., 2019). One critique of peer support programs, as with other health interventions, is that they may fail to recruit those most in need Wilson et al., 2007). Other criticisms can be considered in relation to the mechanisms described above, which can have negative as well as positive impacts (Watson, 2019). Participants may not benefit as they compare themselves negatively to peer supporters, whilst peer supporters might be affected by revisiting past experiences, or the emotional or practical burdens of providing support (Embuldeniya et al., 2013;Watson, 2019 (Kanters et al., 2016), breast cancer , or mental health difficulties ; caring for someone with dementia (Dam et al., 2016), or in the critical care setting (Haines et al., 2018). Broader summaries of peer support, across populations, or considering content or mode of delivery, are limited. Whilst there are no existing EGMs on the effectiveness of using peer support in health and social, there are two systematic reviews which focus on peer support for health promotion and disease prevention and peer support for 'hardly reached' populations, both are restricted to evidence from the USA .
However, reports have been published in the grey literature in recent years. Two key reports from national care/policy organisations in the UK were published in 2015, by National Voices and Nesta, and in 2013, by the mental health charity Mind (Mind, 2013;Nesta & National Voices, 2015). Both looked at the range of peer support interventions available and the different ways they are described, with Mind also collating the experiences of peer support groups with a view to supporting the development of emerging peer support projects. National Voices and Nesta (2015) concluded that it was not possible to identify which method of peer support delivery was most effective, as the majority of studies identified did not compare different modes of delivery to one another.
Furthermore, evidence pertaining to the cost-effectiveness of peer support interventions was inconclusive. Mind highlighted several key issues, including the lack of evidence regarding use of peer support in minority and marginalised communities, the sustainability of funding, the need for adequate training, support and supervision for those delivering peer interventions and whether management of peer support interventions is overseen by services or devolved to service users.
Whilst these reports highlight the variety and broadly perceived value of peer support interventions being implemented across multiple populations and settings, the usefulness of this study to inform the commissioning or delivery of peer support in particular contexts or populations is less certain.

| Why it is important to develop the EGM
Peer support can directly benefit participants, addressing a range of health and social needs, including social isolation (Jones et al., 2014) and mental ill-health (Repper & Carter, 2010). It is an inherently person-centred form of care, which also has the potential to reduce demand on services through improved condition management (NHS England & Local Government Association, 2017) and reduced emergency admissions (Deeny et al., 2018). There has been a desire to professionalise peer supporters and maintain standards through increased accreditation and training. This creates a dilemma, as it may conflict with the wish to maintain the 'authenticity' of peer support services by their separation from formal services (National Collaborating Centre for Mental Health et al., 2020;Q LAB, 2018). Whilst this is beyond the scope of this EGM to address, our focus on the effectiveness of peer support by peer supporters with a formal role is important, as this form of provision is increasing.
By presenting an overview of the evidence in an accessible format, this EGM will allow those designing and providing services, to find studies on the effectiveness and cost-effectiveness of peer support, as well as rapidly discover where evidence of effectiveness is currently BOX 1 Mechanisms underpinning peer support (Watson, 2019) lacking. The decision to produce an EGM, rather than a detailed synthesis of the effectiveness of a current uncertainty, was guided by stakeholders, with the ability to find evidence to inform ongoing decision-making felt to be most helpful in meeting the needs of multiple stakeholders. As peer support interventions vary in terms of delivery, content and underlying actual or intended mechanisms, this presents a challenge to commissioners and service providers as these models may differ in effectiveness, particularly depending on the circumstances in which they are used (Valenstein et al., 2016). Being able to access the spread of evidence relating to specific populations and for specific peer support interventions, or features of these interventions, and determine whether or not they are effective, will aid decision-makers in providing appropriate interventions to achieve their aims.

| OBJECTIVES
Our aim was to identify and appraise the volume, diversity and nature of recent, high quality evidence (systematic reviews, randomised controlled trials and economic evaluations published after 2015) for the use of peer support in health and social care. Nature refers to the information contained within each individual review e.g. the study population, whereas diversity refers to the variation, in terms of content and focus, between reviews.
Specific research objectives were to: -Map the recent, robust evidence for the effectiveness of peer support interventions across health and social care.

| Evidence and gap map: Definition and purpose
EGMs provide an overview of the evidence in a given area. They are produced using the same principles as systematic reviews but give a visual presentation of the types and focus of existing studies available rather than a synthesis of findings (Snilstveit et al., 2013;White et al., 2020). This EGM is presented in two dimensions as a table, with subcategories associated with both dimensions (Saran & White, 2018). The rows typically list types of interventions (i.e. different ways of delivering peer support) and the columns different outcomes; each cell shows the number of studies containing evidence on that particular combination of intervention and outcome. Within cells, studies are separated by type (systematic reviews, and randomised controlled trials and economic evaluations) and their assessed quality.
By displaying the volume, diversity and key characteristics of existing evaluations (including systematic reviews), EGMs allow users to identify and locate the research evidence (or evidence gaps) relevant to their patient or intervention focus. They can therefore support evidence-informed policy, commissioning, and provision, or prioritise the focus of future research (Snilstveit et al., 2013).

| Framework development and scope
The scope of this EGM is to capture recent evidence on the effectiveness and cost-effectiveness of peer support in health and social care. Recent was defined as evidence published after 2015, with this date chosen to focus on evaluations not included in other comprehensive reviews of evidence relating to peer support (Mind, 2013;Nesta & National Voices, 2015).
Framework development was informed by previous research (Nesta & National Voices, 2015), which was used to define key terms and create a list of important categories. This was refined itteratively, through consultation with and input from Fiona Campbell (University of Sheffield), who is experienced in producing Evidence and Gap Maps, and stakeholder engagement as detailed below. Categories were designed to be accessible and usable on the map.
Where categories were identified in the included research that did not fit the Framework, categories were renamed, or their scope adjusted to ensure that all reported population characteristics, and outcomes relating to effectiveness and cost-effectiveness, were included in the EGM. This adjustment was discussed in group meetings, to ensure consistency, and resolved in consultation with stakeholders, where appropriate.
For a list and description of intervention and outcome categories included in the Framework, please see the EGM Glossary, included below as Supporting Information: Appendix 1.

| Stakeholder engagement
User involvement in evidence synthesis is important to ensure useful outputs from the review process (Konnerup & Sowden, 2008). Stakeholders consulted in the production of this EGM included policymakers, health care professionals, academics, third sector organisations providing peer support, and online providers of peer support resources and training, from the following organisations: We also recruited users of peer support to form a public patient involvement (PPI) group. Whilst not the primary intended audience of the EGM, we felt it was important to involve users of peer support for the insight their lived experience could provide on improving and extending use of the map (Gierisch et al., 2019). All stakeholders were identified through word-of-mouth and snowballing techniques and invited to workshops and individual meetings, to suit project progress and stakeholder availability.
We consulted stakeholders throughout the review, with early conversations focused on defining and developing the scope of our research questions and protocol, and later consultation focusing on production of the EGM (Haddaway et al., 2017;Konnerup & Sowden, 2008). We also consulted stakeholders on aspects of the report such as the plain language summaries describing our review and its findings, to ensure they are accessible to a range of audiences.
Initial consultations helped to identify key populations and outcome categories within our EGM as well as focus the definition of peer support. Once a prototype version of the EGM had been developed and populated with some study data, later trials of the use of the map in workshops with stakeholders resulted in discussion of the language used in the EGM. We incorporated feedback from all stages of consultation to ensure the final EGM provides a level of information useful to the intended audiences.

| Conceptual framework
Peer support is defined by the shared lived experience of peers and peer supporters and this is also the key mechanism in its effectiveness [Box 1; (Watson, 2019)]. However, previous reviews of evidence relating to peer support indicate diversity in the forms of peer support available and that people are therefore experiencing . Effectivness may vary depending on the type of support (Mind, 2013;Nesta & National Voices, 2015).
Key categories through which peer support interventions can be described and which may influence its effectiveness are (Nesta & National Voices, 2015): • Who-the people involved, the health condition(s) they are experiencing, and their sociodemographic characteristics; • What-the focus and type of support offered, whether education, emotional or practical, and who facilitates it; • How-whether meetings are in person or online, occur individually or in groups, and the support offered to peer supporters; • Where-the location of its provision, for example, healthcare or community settings; • When-the duration of the intervention.

| Types of study design
Impact evaluations study change in an outcome which can be attributed to an intervention (3ie, 2012). This EGM includes: • Systematic reviews (SRs) of impact evaluations; and two types of impact evaluation: • Randomised controlled trials (RCTs); • Economic evaluations (EEs e.g. comparative costing studies, costeffectiveness studies).
SRs can be defined as studies which collect and synthesise all the research available on a topic to answer a specific question (Chandler et al., 2022). They typically seek to minimise bias by specifying their methods in advance, and assessing the quality of included studies using standard critical appraisal tools. RCTs are experimental studies We included RCTs that were not in the SRs already included in the map because they provide evidence on the effectiveness of peer support in health and social care. We also included EEs, to map evidence relating to the cost-effectiveness of peer support-based interventions.
Both published and ongoing studies were screened for inclusion.

Definitions and language
Having a narrow medicalised definition of peer support as an intervention could be considered problematic. Peer support is a process and not something that is done to people. The word 'intervention' was also removed from the Health Education England Peer Support Worker Framework (National Collaborating Centre for Mental Health et al., 2020).
Discussed more helpful terms and how these depend on the audience; consulted with additional stakeholders to ensure all views were represented.
As the main audience for the EGM are service providers and the term intervention is familiar to them, we decided to keep this language in the EGM but add context to glossary and report.
Measuring outcomes could also be considered problematic for reasons outlined above.
We are limited by the EGM aims, which are to map evidence of effectiveness, but have reflected and contextualised this in the report to acknowledge what is being missed in this EGM.
Definition of community based/community delivered, these can be interpreted differently. For example, peer support could be delivered in a community venue, delivered by a community practitioner, or a community-based peer support group. All may fall in the same category at present but are very different.
We are limited by the information provided by included studies, currently these distinctions are not clearly or consistently defined in the literature, so we are unable to include them in the EGM.
Clarify the 'People at risk' heading and definition: need to be mindful of language and judgement.
We discussed, and consulted on, alternative language and decided to change this category to 'Vulnerable'.

Map presentation
Difficult to distinguish between SRs and RCTs when first looking at the map. The donut rather than bubble style of presenting studies in the cells might represent the study types more clearly.
RCTs and SRs have been clearly defined in the glossary/ definitions section of the map; we have written out the terms in full on the map rather than using abbreviations for clarity.
It is not possible to pre-programme the map to start with the donut style, but we have provided map instructions so that people know how to change the style when viewing.
Filters for countries would be useful as evidence from different countries may have different contextual relevance.
Added country filter to map, with a focus on distinguishing between UK and non-UK studies, as well as studies with potentially similar health systems.
Include filter for year. Communicated more clearly in the EGM description that only research from 2015 onwards is included and added a filter for year to the EGM.
The indication of quality of evidence in the EGM could be more intuitive. A general description might work better than risk of bias and quality appraisal.
Discussion about what might work, particularly in terms of alternative colours.
Changed wording to refer to 'quality' and used broad categories within the EGM. Kept data on specific quality appraisal tool (AMSTAR2 and ROB) in filters on map with full detail in report.
It may be helpful to filter the peer support provider, e.g., NHS/non NHS.
This data is not generally available in SRs and is only available in a limited number of RCTs. We have coded for 'who facilitates the intervention' to indicate the degree of professional involvement in peer support provision.

Useability
The filtering aspect of the map needs more explanation. Information on how to 'select filtering mode' under filtering added to the map instructions.
Having 'Segment' as the default when clicking on the map would indicate which records are SRs and RCTs more clearly.
This is not currently an option when creating the map. Information on how to choose the 'Segment' option has been added to the map instructions.
It would be useful to be able to find studies on specific health conditions in the map.
There is a 'search' function available in the map, this has been indicated in the map instructions.
Find a way to update the map, this is important in such a fast moving field.
The team has discussed putting together a funding application. This could, for example, be to update the map every 6 months and to work on developing the methodology.
Abbreviations: AMSTAR2, assessing the methodological quality of systematic reviews; EGM, evidence and gap map; RCT, randomised controlled trial; ROB, Cochrane risk of bias tool.

| Search methods and sources
As described in the EGM protocol (Shaw et al., 2021), studies were identified in two stages:

| Description of outcomes
Any effectiveness outcome was included in the EGM; we grouped these into five broad categories: health-related indicators; selfregulation; supporting self-regulation; wellbeing and social connectedness; cost-effectiveness and service use; and experience of peer support. Table 3 lists subcategories within each BOX 2 Ovid MEDLINE search strategy 1 (peer* adj3 (administer* or adviser* or advisor* or advocate* or coach* or co-facilitat* or cofacilitat* or consultant* or counsel* or deliver* or educator* or expert* or facilitator* or group* or helper* or instructor* or leader* or led or listener* or mentor* or navigator* or network* or program* or provider* or specialist* or support* or trainer* or trained or tutor* or worker*)).tw. (17015  The revised inclusion and exclusion criteria (listed in Table 4; full details in Supporting Information: Appendix 2) were then applied to the title and abstract of each identified citation independently by two reviewers, with disagreements being resolved by discussion.

| Stage 2: Full text
The full texts were obtained for papers that appeared to meet the inclusion criteria, and those for which a decision was not possible based on the information contained within the title and abstract alone. For SRs, the full text of each record was then assessed independently for inclusion by two reviewers. For impact evaluations, a second calibration exercise was undertaken by two reviewers for a sample of the full text articles (n = 50), before full texts were screened as for SRs. Disagreements were settled by discussion.
Ongoing SRs and RCTs were screened using the screening criteria listed in Table 4 and Supporting Information: Appendix 2. Where limited information at full text made it difficult to assess eligibility, this was discussed by the team, and a decision mechanism agreed and consistently applied. For example, for studies where it was implied, but not explicitly stated, that peers had the same health condition as participants (thus meeting our definition of peer support) these were included.

| Data extraction and management
We imported records from the EndNote libraries into our data management software, EPPI-Reviewer 4 .
Standardised data extraction forms were developed for SRs, RCTs and EEs in EPPI-Reviewer 4 (see Supporting Information: Appendix 4),

Category Definition/example
Case management, health service liaison Helping participant to make contact with available health and social care support. This might include accompanying peers to appointments or identifying relevant services.
Education, coaching, mentoring Provision of information, education, training, mentoring and/or coaching by the peer supporter. This could be through a manualised intervention covering specific topics or informal discussion.
Practical support for health behaviours Practical help with health behaviours such as exercising, or learning to monitor blood glucose levels.
Psychological, emotional, wellbeing support Mental health support, mindfulness, wellbeing, and quality of life interventions.
Self-care, self-management A focus on self-management, self-care and goal setting.

Social, community
By definition all peer support includes a social element. This category related specifically to interventions which aimed to help peers build relationships, participate in the community, or support social interaction.
Not clearly defined Interventions where information was not provided on the content of contact between peer and peer supporter.
to collect information on different aspects of peer support interventions, including the content, method, and duration, and outcomes.
These were piloted by the review team on a selection of included studies.
To ensure an efficient use of limited resources, data for impact evaluations (RCTs and EEs) that were included in SRs was only extracted from information reported within the relevant SR.
However, as the aims, methodology, and reporting of included SRs may have been slightly different to those of the EGM, extracted data on these impact evaluations is more limited than data extracted directly from RCTs and EEs. Data extraction was performed by one reviewer and checked by a second, with disagreements being settled through discussion.
Data were extracted from all included studies identified by the searches; these are listed under 'Included studies'. An additional 4 SRs and 1 RCT were identified after the searches had been completed. Due to time constraints on the construction of the map, data was not extracted from these studies. They can be found in the

| Tools for assessing risk of bias/study quality of included reviews
Quality assessment of SRs, RCTs, and EEs was conducted to indicate confidence in study findings. All assessments were performed by one reviewer and checked by a second, with disagreements settled through discussion. We did not exclude any study based on study quality.

| AMSTAR 2
All SRs identified as eligible following full-text screening were appraised using the AMSTAR2 quality appraisal tool (Shea et al., 2017) for SRs of primary studies of randomised and non-randomised study designs. AMSTAR-2 is a 16-item checklist, covering all aspects of the conduct of a SR, from pre-specifying a protocol to appropriately analysing and discussing risk of bias. Full details of the checklist can be found in Supporting Information: Appendix 5.
Items 2, 4, 7, 9, 11, 13 and 15 of the checklist are considered "critical" in assessing overall study quality, with studies rated from high to critically low depending on the number of weaknesses (Shea et al., 2017). A rating of "high" means the study has no more than one noncritical weakness, "moderate" that there is no critical weakness but more than one noncritical weakness, "low" that a study has one critical weakness, and "critically low" that a study has more than one critical weakness. Low and critically low studies may also have multiple noncritical weaknesses.

| ROB tool
We used the Cochrane ROB tool Higgins & Green, 2011) to assess risk of bias in included RCTs. As our EGM focuses on multiple outcomes, we were assessing all outcomes that met our inclusion criteria in the included studies for risk of bias. We therefore decided to use ROB rather than ROB2, as this tool is more suitable for considering multiple outcomes.
ROB considers trial design, conduct and reporting, focusing on seven domains related to the internal validity of the study: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and other biases. Review authors discussed the application of the domains to the studies throughout the process of critical appraisal.
For the selective reporting domain, as we were not performing a metaanalysis, we considered risk of bias to be low if all measured outcomes were reported in the results; these could be in any format. 'Other biases' are normally prespecified: we did not consider there to be any specific issues relating to peer support so did not use this domain.
As well as considering risk of bias due to each domain/study feature, review authors designate key domains in relation to the review topic which inform a summary of overall risk of bias for a study (Higgins & Green, 2011). Due to the difficulties of blinding in studies of peer support, the study team decided that using the sequence generation, incomplete outcome data, and selective reporting domains would be most indicative of overall risk of bias.

| CHEC list
We used the CHEC list (Evers et al., 2005) for assessing the quality of EEs. The CHEC-list was developed using expert consensus; it is a series of yes/no questions focusing on the methodological quality of EEs, and includes items on generalisability and the distribution of impacts across different groups or users (i.e., equity considerations). It can also be used to assess the quality of comparative costing studies, although in these cases the items relating to outcome identification, measurement and valuation are not applicable. To derive an overall assessment of quality so that EEs could be grouped with RCTs on the map, we calculated the percentage of applicable questions which were answered 'yes' (Wijnen et al., 2017). Studies scoring over 75% were classified as high quality, studies between 50% and 74% as medium quality, and studies below 50% as low quality (  in 2020. EEs varied between 0 and 2 published per year, as can be seen in Table 5.
Included studies can be viewed by visiting the online EGM in Supporting Information: Appendix 8. The map is accompanied by detailed instructions for use and a structured abstract with a summary of key charateristics is available for each study. Figure 3 presents an example of the interactive EGM, with the intervention and outcome categories displayed at the sides of the map; circle size indicates the number of studies found in the cell and colour represents study type and quality.

| Distribution by intervention
Included studies described a wide range of peer support interventions. All SRs and the majority of impact evaluations (30 RCTs and Table 7 shows an aggregate map of outcome categories and described peer support interventions for SRs. Health-related indicators are measured in the majority of described peer support interventions, there is also a concentration of evidence among social and community interventions measuring wellbeing and social connectedness. Table 8

| Logistics of peer support intervention
There were some concentrations of evidence regarding who facilitated peer support interventions. Twenty-six SRs reported on studies in which peers facilitated but it was not clear how professionals were involved although this was not the case among RCTs (12 studies). The next most populated category was interventions led by peers working with professionals (13 SRs, 29 RCTs, and 5 EEs). As can be seen in Figure 8, few studies reported on interventions co-facilitated by peers and professionals, with only 6 SRs, 8 RCTs, and no EEs including evidence on this type of facilitation.
Our inclusion criteria set out three types of support, of which peers needed to have at least one to be included. Of these, training T A B L E 7 Aggregate map showing the number of SRs (n = 32) by description of peer support intervention category and outcome category, with colours indicating greater (purple) and lesser (blue) concentrations of evidence was the most populated category of support (25 SRs, 49 RCTs, and 7 EEs). Fewer studies included peers who were paid or had a contract (9 SRs, 18 RCTs, and 2 EEs).

| Methods used in included EEs
Only 2 of the included EEs were model-based, 5 used data from existing sources such as RCTs. Three of the EEs were linked to RCTs also included in the map. Cost-utility analyses were the most commonly used method for assessing the cost-effectiveness of peer support interventions (3 studies); two EEs were comparative costing studies, with the methods used by the remaining EEs being cost-effectiveness analysis and other (cost-offset) (Figure 9).

| Time period for follow-up
As with the duration of the intervention, there was a concentration of evidence regarding follow-up of interventions in the 'up to 6 months' category (24 SRs and 42 RCTs), with a large number of studies also reporting follow-up at 'up to 12 months' (26 SRs and 23 RCTs). Longer follow-up periods were less common except in EEs.

| Study populations
Included studies tended to focus on populations with chronic health difficulties, as shown in Figure 10. There was a concentration of evidence for chronic physical health difficulties for SRs (22 studies)

| Distribution by outcome category
Any outcome relating to effectiveness of peer support was included in the map. Figure 11 shows that distributions of SRs and RCTs were similar within our five broad outcome categories (Table 3) North America, the most common location for RCTs to be conducted was Europe (6 studies) ( Figure 12). ROB Higgins & Green, 2011) was used to assess RCTs (52 studies); key domains used to calculate overall risk of bias for RCTs can be seen in Figure 13. Overall, the majority of RCTs (25%/13 studies for allocation concealment and 27%/14 studies for incomplete outcome data) ( Figure 13).

| Risk of bias in included studies
For EEs, all except one were judged to be of low or medium quality using CHEC (Evers et al., 2005). No studies satisfied the criteria for whether the generalisability of the results was adequately discussed (CHEC item 17) or whether the article notes ethical aspects, characteristics of the population experiencing the disease/intervention and potential distributional implications (CHEC item 19). Only two studies were judged to have comprehensively explored uncertainty through sensitivity analysis, which is usually seen as an essential feature of a high quality economic evaluation (where, in general, few variables tend to be known with absolute certainty). Of the four economic evaluations, three tended not to clearly report or adequately justify: the time horizon of the analysis, the perspective of the analysis, and the separate presentation and justification of which resources were included, that they were counted in relevant physical units, and that they were valued appropriately. Two of the three costing analyses met these criteria more convincingly, and both were judged to be of 'medium' study quality. Looking at the overall pattern of strengths and weaknesses, Campbell (2014), Ye et al. (2021) and Patel et al. (2017) provide the most valid and reliable evidence about the cost-effectiveness of those peer support interventions [respectively: for people withType 2 diabetes in the USA (Campbell, 2014;Ye et al., 2021) and for people with latent TB in Canada (Patel et al., 2017)]. • education, coaching, and mentoring; • psychological, emotional and wellbeing support; • self-care and self-management; • social and community.
Education, coaching and mentoring contained the most studies of any category describing the type of peer support intervention, as a result they also had the most evidence relating to all types of outcome.
Studies tended to focus on peer support for populations with chronic health difficulties, as do the majority of studies awaiting classification (Berg et al., 2021;Ghahramani et al., 2021;Wan et al., 2021). As might be expected, we found a concentration of evidence focusing on health-related indicators, both physical and mental, of the effectiveness of peer support.
Outcomes relating to wellbeing and social connectedness, and among RCTs self-regulation, were also well represented. There was also a lack of clarity in the reporting of time frames for follow-up. Some studies reported outcomes from baseline, others from the end of the intervention. This creates limitations for users looking for evidence regarding the effectiveness of peer support over different timeframes, as peer support interventions included in the EGM ranged in length from one-off sessions to those lasting several years.
Finally, it remains unclear whether the gaps in the EGM are due to a lack of peer support interventions occurring in these contexts, and among groups with these health and social care needs. It may be that these gaps reflect an absence of recent effectiveness research in these areas.

| Stakeholder engagement throughout the EGM process
The only change to our stakeholder engagement during the EGM process occurred after initial conversations with stakeholders involved in policy and practice highlighted the value of consulting users of peer support, due to the insight their experience could bring to, and as potential users of, the map. As a result, we

| Implications for research
RCTs shown individually in this EGM are those not included in the published SRs already on the map. Areas of the map with high numbers of RCTs therefore indicate the potential for systematic reviews on these topics: • The effectiveness of peer-led case management and health service liaison as a type of peer support.
• The effectiveness of peer support for populations with chronic mental health difficulties and vulnerable populations.
However, it is also evident from this EGM that there is a lack of research relating to some aspects of peer support which are essential for those commissioning and delivering these services. Impact evaluations, particularly RCTs, are needed on: • The cost-effectiveness of peer support compared to other forms of support and/or usual care. Also, research on outcomes beyond physical and mental health, to aid understanding of the mechanisms by which peer support leads to health outcomes.
• The effectiveness of peer support in different geographic locations with different health care systems and contexts.
• The effectiveness of online peer support. Delivering health and social care during the pandemic has led to many adaptations and innovations, such as the provision of services online. This has the potential to increase accessibility, for example, for those with long-term health conditions who may benefit most from peer support.
Research included in the EGM suggests a need for clarity in the reporting of research. Clear reporting of time frames for follow-up is particularly important to allow assessment of the sustainability of programmes and outcomes. Studies of peer support should also provide more information on how and by whom the intervention is organised, as well as aspects such as the location of the intervention.
Similarly, clearer reporting within SRs would have given greater confidence in the results of these studies.

| Implications for policy and practice
• Concentrations of evidence on health-related outcomes of peer support and populations with chronic health conditions provide a source of information for those looking to commission or provide peer support.
• Integrating into multidisciplinary teams and having an unclear role have been identified as challenges for peer supporters (National Collaborating Centre for Mental Health et al., 2020). Evidence was found in impact evaluations (RCTs and EEs) on the delivery of peer support led by peers who were working with care professionals.
These studies could inform policy and practice, particularly as some related to particular types of peer support such as case management and practical support. More evidence is needed, however, on co-facilitated peer support.

CONTRIBUTIONS OF AUTHORS
AP led the development of the interactive EGM and strategic planning for drafting the report and refining map categories. SDB, NS, AB, RA, and JTC contributed to the development of the interactive EGM.
AP and SDB carried out screening, data extraction and quality appraisal. RA carried out screening, data extraction quality appraisal of economic evaluations. NS and AB supported screening, data extraction and quality appraisal. • EGM methods: JTC has previously worked on evidence gaps maps. All authors have prior experience in systematic reviews and are proficient in carrying out the various stages required to produce an EGM, including eligibility screening, quality assessment and data extraction.
• Information retrieval: AB and NS both have training and extensive experience in designing and implementing search strategies.

PLANS FOR UPDATING THE EGM
There are no current plans to update this EGM. However, the authors will consider updating the EGM in the future if relevant funding is available.

DIFFERENCES BETWEEN PROTOCOL AND REVIEW
The only difference in the methods we ultimately used, from those described in the protocol, was to recruit a PPI group with experience of using peer support rather than consult the NIHR PenARC Public Engagement Group (PenPEG) regarding the map.
We did not make any post-hoc decisions regarding eligibility but as planned in the protocol, we clarified the application of inclusion and exclusion criteria. Similarly, as described in the protocol, for the stage 2 search for impact evaluations, we specified a 2015 cut-off date for inclusion in the EGM after full screening and as a team decided to list pre-2015 studies in Supporting Information: Appendix 6.

Internal sources
• No sources of support provided