Peer support for people with severe mental illness versus usual care: study protocol for a pragmatic multicentre randomised controlled trial (UPSIDES-RCT)

Background: Peer support is an established intervention involving a person in recovery from mental illness being engaged to offering support to others with mental illness. Peers are an under-used resource in global mental health. Building upon comprehensive formative research, this study will rigorously evaluate the impact of peer support at the levels of service users (psychosocial and clinical outcomes), peer support workers (work role, empowerment), services (cost-effectiveness, return on investment), and implementation (adoption, sustainability, organisational change). Methods: UPSIDES-RCT is a pragmatic parallel-group multi-centre randomised controlled trial assessing the effectiveness of UPSIDES at four measurement points over one year (baseline, 4-, 8-, and 12-month follow-up), and embedded process evaluation and cost-effectiveness analysis. Research will take place in a range of high-, middle- and low-income countries (Germany, United Kingdom, Israel, Uganda, Tanzania, India). The primary outcome is social inclusion of service users with severe mental illness (N= 558; N = 93 per site) at 8-month follow-up, measured with the Social Inclusion Scale. Secondary outcomes include empowerment (Empowerment Scale), hope (HOPE scale), recovery (Stages of Recovery), and health and social functioning (Health of the Nations Outcome Scales). Mixed-methods process evaluation will investigate mediators and moderators of effect, and implementation experiences of four UPSIDES stakeholder groups (service users, peer support workers, mental health workers, and policy makers). A cost-effectiveness analysis examining cost-utility and health budget impact will estimate the value for money of UPSIDES peer support. Discussion: By implementing and evaluating a manualized peer support intervention for people with severe mental illness across low-, middle-, and high-income countries, this study will contribute to harmonising core elements of peer support across different

cultural and organisational dimensions. The UPSIDES-RCT will explore the essential components necessary to create a peer support model in mental health care, while providing the evidence required to sustain and eventually scale-up the intervention.
Performance of mental health services will be maximised by actively involving and empowering service users, generating system changes towards user-centeredness, recovery orientation, community participation, and realising mental health as a human right.

Background
Peer support is part of a broader recovery agenda in mental health that places emphasis on user-centred outcomes such as social inclusion and empowerment [1]. Peer support is an established intervention whereby a person in recovery from mental illness is engaged to offer support to others with mental illnesses [2]. Peer support workers (PSWs) support their own recovery and the recovery of others by drawing on their lived experiences, employing positive self-disclosure, expanding social networks, and promoting hope, empowerment and self-efficacy. Around the world, diverse PSW roles have been developed and formalised: peer companions, peer advocates, consumer case managers, peer specialists, peer counsellors and more [3]. PSWs providers offer a wide range of services, which may include social support, management of symptoms, counselling, outreach, coaching and advocacy [4]. Peer support can also be provided in different settings -as an alternative to, an independent service within, or an integral part of professional care [5].
Findings from meta-analyses synthesizing a large number of controlled studies indicate that PSWs are able to achieve outcomes comparable to professionally trained staff, and might therefore represent cost-effective additions to task-sharing models in low-and middle-income countries (LMICs) as well as in high-income countries (HICs) [6,7].
Qualitative and quantitative studies have also shown that peer support has a positive impact on recovery-related outcomes that may not be possible to achieve using exclusively clinical interventions [8,9]. However, there are significant gaps in the evidence base for peer support in LMICs [10] and in non-Anglophone countries [11], with most reviews about peer support only identifying studies from HICs, primarily Englishspeaking countries [12].

The UPSIDES consortium
The project Using Peer Support In Developing Empowering Mental Health Services UPSIDES takes place over a five-year period (2018-2022) and is divided into two phases, with different objectives and methods. In phase 1, which lasts two years, implementation stage of peer support across study sites will be assessed and a culturally appropriate peer support intervention will be developed [13]. This paper focuses on the three years of phase 2, when the finalised peer support intervention will be implemented and evaluated at the levels of service users (psychosocial and clinical outcomes), peer support workers (empowerment, professional development) services (cost-effectiveness, return on investment), and implementation (adoption, sustainability, organisational change).
Framework of the current study UPSIDES' conceptual background is the model of change processes in mental health peer support developed by Gillard et al [14] (Fig. 1). The model, based on a large multisite qualitative study, provides a measurable set of outcomes expected to change in response to processes of peer support. These address active ingredients of peer support and relate to a host of process outcomes of peer workers including hope, social functioning and increased engagement with services, and downstream impacts refer to recovery, wellbeing, and service use.

UPSIDES research is also guided by the Consolidated Framework For Implementation
Research (CFIR [15], Fig. 2) which was created to promote implementation theory development, helping to standardise investigation into and verification into what works, where, and why across multiple contexts. CFIR covers five major domains that can influence effective implementation: (i) intervention, (ii-iii) outer and inner setting, (iv) individuals involved, and (v) process.
Building on results of formative research described elsewhere [13,16], UPSIDES peer support will be implemented across sites in line with an implementation manual, taking into account differences across study sites. The development of the manuals as well as their implementation and the overall study design will take into consideration the five CFIR domains by, for example: incorporating adaptations to local contexts; planning for organizational readiness; paying attention to the selection and involvement of relevant stakeholders; and allowing for a staged approach to continuously modify the intervention as needed.

Objectives and research question
The aim of UPSIDES-RCT is to explore the implementation and effectiveness of peer support delivered in a range of high-, middle-and low-income country contexts. The main objectives are:

1.
To evaluate the outcomes of delivering peer support, for service users, peer support workers and organisations, through a multi-centre pragmatic parallel-group randomised controlled trial and additional qualitative methods.

2.
To assess the value for money of peer support for persons with severe mental illness, by carrying out a cost-effectiveness analysis.

3.
To evaluate the process of implementing the UPSIDES peer support intervention, with special attention to differences in context across the study sites, using both quantitative and qualitative methods with PSWs, service users, mental health workers and wider stakeholders.

Methods/Design
This protocol covers the UPSIDES pragmatic parallel-group multi-centre randomised controlled trial, cost-effectiveness analysis, and process evaluation, described further below. This study protocol adheres to the SPIRIT statement [17].

Setting
Six of the eight UPSIDES collaborating institutions will host study sites:

4.
Dar es Salaam, Tanzania: The intervention will be implemented at Muhimbili National Hospital at the Department of Psychiatry and Mental Health.

5.
Beer Sheva, Israel: The intervention will be implemented at institutions in Israel which provide UPSIDES peer support.

6.
Pune, India: The intervention will be implemented at Hospital for Mental Health in Ahmedabad, Gujarat, which is a public mental health facility and has been implementing peer support since April 2015.
Randomised controlled trial UPSIDES-RCT is a pragmatic parallel-group randomised controlled multicentre trial including a waiting list, with four measurement points (t 0 = baseline; t 1 = 4 months; t 2 = 8 months; t 3 = 12 months). Exclusion criteria: Main diagnosis of learning disability, dementia, substance disorder or organic brain disorder; cognitive impairment severe enough to make it impossible to give informed consent or complete study measures.

Processes, intervention and comparison
Intervention All participants will receive treatment as usual (TAU). Participants allocated to the intervention group will additionally receive the UPSIDES peer support, which is a direct service that is delivered by a person who has personal experience of mental health problems to a person or a group of persons with a serious mental illness. UPSIDES peer support workers (PSWs) will be persons of adult age, having experienced mental ill health and having progressed along in their recovery as indicated by having been stable or out of hospital for at least three months. Peer support workers will have progressed along in their recovery beyond controlling symptoms, to a focus on self-definition, growth and participation in the community. UPSIDES PSWs will be using these personal experiences, along with UPSIDES training and supervision, to facilitate, guide, and mentor another person's recovery journey [19]. and site-specific topics (e.g. stigma, rights and advocacy, trauma and disasters). An online training platform will facilitate exchange among trainers, PSWs and staff members of different study sites. The platform will provide training and supervision, and help build an international PSW community.
UPSIDES peer support will be provided without hierarchy and judgement, taking into account that mental health states may be variable. UPSIDES PSWs will focus on: providing tangible supports; role modelling that recovery is possible; sharing experiences of mental health and ill-health; and promoting hope, a sense of control and opportunities. Specific tasks include: initial assessment of strengths and resources to build on; practical support with daily life as needed (e.g. accompany to appointments or activities); support during crises; and actively promote recovery planning. Providing an additional recovery group setting is possible and recommended. UPSIDES peer support will be delivered for up to six months, with a minimum of three contacts. Weekly or biweekly meetings are recommended, but frequency may vary, depending on the needs of service users, PSWs, and study sites. The intervention manual provides additional materials to be used by the PSWs during the intervention. To support the implementation process, organisationalreadiness workshops are held across trial sites. Participants allocated to the control group (wait list) will start receiving the intervention after completion of follow-up (month 12).

Control
The control intervention is treatment-as-usual as provided at each of the respective UPSIDES-RCT the study sites: Ulm: Psychiatric routine care in Germany is mainly provided by psychiatric hospitals, psychiatric outpatient clinics and office-based psychiatrists and psychotherapists. In

Outcomes
Primary outcome is social inclusion at t 2 (8 months). This time point has been chosen because it is shortly after the end of the intervention but also gives enough time for changes in social inclusion to take place. Social inclusion is a key outcome in global mental health [29] and will be measured with the Social Inclusion Scale (SIS [30]) which is a service-user-reported measure with 16 items answered on a 4-point Likert scale ("not at all"; "not particularly"; "yes a bit"; "yes definitely"), yielding a total score consisting of the prorated sum over all 16. The total score can range from 16-64 with higher scores representing higher levels of social inclusion. The SIS has previously shown adequate psychometric properties [30] which will be further established for the different language versions to be used in this trial. Secondary outcomes are empowerment (Empowerment Scale, ES [31]), hope (HOPE scale [32]), recovery (Stages of Recovery, STORI-30 [33]), and health and social functioning (Health of the Nations Outcome Scales, HoNOS [34]).
Established state-of-the-art translation guidelines [35] will be followed to translate and locally validate the standardised measures (including administration instructions) used, with most attention paid to the psychometric evaluation of the primary outcome. See Puschner et al. [13] for further details on translation.
Participant timeline Figure 3 gives an overview of this trial's participant timeline for the major stages enrolment, allocation, and assessments in line with SPIRIT recommendations ("SPIRIT figure").

Recruitment
Recruitment of study participants will vary among sites due to their differences in mental health service systems. Potential participants will be approached in various ways, including outpatient/ community mental health services, patient and carer organisations, local newspapers, social media, community leaders, and word-of-mouth. Duration of recruitment period is 12 months, starting January 2020.

Allocation
Participants will be randomly assigned to either control or intervention group with a 1:1 allocation as per a computer-generated randomisation schedule stratified by site using permuted blocks of random sizes. The block sizes will not be disclosed, to ensure concealment. Participants will be randomised by an independent unit (Institute for Medical Biometry and Epidemiology, UUlm, Germany) using the software ROM [36]. All participants who give consent for participation and fulfil the inclusion criteria will be randomised.
Allocation concealment will be ensured, as the service will not release the randomisation code until the participant has been recruited into the trial and the baseline assessments completed.
Randomisation will be requested by the staff member responsible for recruitment and clinical interviews from UUlm's Institute for Medical Biometry and Epidemiology by e-mail.
The randomisation form includes site ID, participant ID, gender of participant, date of informed consent and approval of eligibility. The requesting research worker will get a response by mail within one working day. For participants in the intervention group, the research worker will inform the local PSW team and the participant. Throughout the study, the randomisation will be conducted by UUlm's Institute for Medical Biometry and Epidemiology in order to keep the data management and the statistician blind to the study allocation for as long as the data bank is open. The randomisation list remains with UUlm's Institute for Medical Biometry and Epidemiology for the duration of the study. Thus, randomisation will be conducted without any influence of the principal investigators or raters.

Blinding
Due to the nature of the intervention, neither participants nor peer support workers can be blinded. Efforts will be made to blind staff collecting data during study visits, e.g. by separating recruitment tasks from data collection, as feasible. Researchers analysing study data will be blinded until the entire analysis has been completed, as described above.
Trial management and oversight The Trial Management Group (TMG) consists of leads of study sites and is responsible for identification, recruitment and follow-up of study participants, data collection, and adherence to the study protocol. Oversight bodies Trial Steering Committee (TSC) and Data Safety and Monitoring Board (DSMB) are independent from the sponsor and will be continuously informed of study progress including data quality issues. The three members of the TSC will ensure that the trial is conducted in line with Good Clinical Practice. The TSC will have ultimate responsibility for the trial and will assume primacy over the DSMB and TMG. The TSC can prematurely terminate the trials, e.g. in case of violations of patient safety. The DSMB will safeguard the interests of the study participants and monitor the data collected in the trial.

Statistical analysis and power calculation
Analyses will start once baseline data have been collected and cleaned. Descriptives of all outcome measures will be produced, and outcome trajectories from t 0 to t 3 will be examined via exploratory analyses. The intervention's effect on primary and secondary outcomes will be tested by means of random-effect regression models including a fixed group effect over time, allowing the inclusion of cases with incomplete (unbalanced) data across panels [37]. Post-hoc analyses will examine early (at t 1 ) and late effects (at t 3 ) for primary and secondary outcomes.

Measures
Comprehensive societal costs of mental illness will be estimated for each participating country using an adapted version of the Client Sociodemographic and Service Receipt Inventory [38] at t0-t3. The CSSRI-UPSIDES will be adapted for use at all sites in order to assess mental health service use and productivity losses due to mental illness. If not already available, country-specific unit costs of health and social care services will be determined by investigating publicly accessible sources (price lists, catalogues of fees and charges), by expert interviews or by calculations of staff and capital costs for used services drawing on the principles from the reference case for global health costing, recently developed by the Global Health Cost Consortium. All costs will be converted into international $ using purchasing power parities (PPP) [39]. Quality-adjusted life years (QALYs) will be estimated using the EQ-5D 3L at t0-t3 which is used in most multinational cost-utility studies [40] and available in the languages of all study sites. As country-specific population-based value sets are not available for most study sites, QALYs will be estimated on the basis of the European EQ-5D visual analogue scale (VAS) for all countries [41].

Data analysis
Incremental cost-utility ratios (ICUR) will be estimated for each country as the ratio between the average difference in total costs of illness and the average QALY difference over 12 months. Stochastic uncertainty of incremental cost-effectiveness ratios will be assessed by nonparametric bootstrapping with 2.000 replications. The maximum willingness-to-pay (MWTP) necessary to cover 95% of the ICER variance will be estimated by means of the cost-effectiveness acceptability curves [42]. In order to take into account the economic differences between participating countries, country-specific MWTP thresholds suggested by Woods et al. [43] and Leech et al. [44] will be applied.
Net monetary benefit regression models with random effects for the UPSIDES treatment, fixed country effects and fixed treatment by country interaction effects will be computed to estimate the impact of country specific differences on cost-utility [45]. Individual netmonetary benefit values will be computed for each study participant by multiplying the individual country specific QALYs with MWTP thresholds ranging from 0 to 50,000 $ (steps will be defined on the basis of results from the incremental cost-utility analysis (ICUA) and subtracting country specific individual costs [45]. Marginal effects of the net monetary benefit regression models will be computed to estimate country-specific net-monetary benefits for defined MWTP thresholds. Robust standard errors will be estimated to account for skew distribution of net-benefit values [46,47]. Results from the primary data ICUA will be used to develop a decision tree model [48] to simulate cost-utility and health budget impact of the UPSIDES intervention for each country over a five year time frame [49]. The sensitivity analysis will consider variance of program coverage and equity of service access [50].
Participants in the process evaluation include service users, peer support workers, mental health workers and other key informants involved in the UPSIDES intervention. Peer support workers and service users will be recruited from the participants in the RCT (see above). Purposive sampling strategies will be applied for mental health workers and key informants. Mental health workers will be recruited from multidisciplinary teams who cooperate with peer support workers. Key informants will be local stakeholders with relevant expertise relating to implementation of peer support work, including clinicians and managers who employ peer support workers.

Quantitative part of the process evaluation
Information on important process variables will be collected from different stakeholders (Table 1). Data analysis will apply random-effect regression models and structural equation modelling including process variables as moderators or mediators of effect.
Additional multivariate analyses will be carried out to analyse the effect of fidelity including dose (number, frequency and duration of peer support sessions, recorded by routinely administered M&E forms) and implementation outcomes.  Practitioners' experiences of peer support; and (4) Key informants' experiences of peer support.
A common overall methodology shared by each of these studies is described below and followed by detailed information on each of the individual studies. Table 2 gives an overview of qualitative studies which are part of the process evaluation.  Data analysis will involve reviewing all field notes, reading and re-reading all the transcripts for familiarization, consensual coding and generating themes. The data analysis will be conducted by a core group, to be supported by partners at each study site involved in data collection as needed. Data analysis will adhere to the following steps: (i) Based on 1-2 interviews with participants, the task lead will develop preliminary codes and categories (themes). (ii) Preliminary codes and categories (themes) will be reviewed by the core group and commented upon. (iii) Modifications of the coding tree will be commented upon by the core group. The final coding tree will be discussed and revised as necessary until consensus is reached. (iv) Subsequent inclusion of interviews and modifications of sub-categories (themes). (v) Reviewing, refining and defining themes. (vi) Validation of the coding structure with each local site.
A code book will be developed using the data from all sites, then will be shared across sites for review, discussion and standardization. The pre-determined code book will be used to develop notes, data managed into units of information that cover broad categories with grouping of relevant emerging themes of importance. Each site will comment on the emerged themes, and the core analysis group will finalise the themes.
The study-specific core group will decide on a software to be used for data management and text retrieval (e. g. QSR International's NVivo 12 qualitative software, MAXQDA). The focus will be on the semantic content of the data (as responses to semi-structured questions). Quality will be improved by the use of multiple analysts to ensure that a range of perspectives inform the interpretation of the data; the use of verbatim quotes for each theme, to ensure the interpretation is as close to the data as possible; and local validation to maximise cross-cultural validity of the coding framework.

#1: Service users' experiences of peer support intervention
Objectives: In order to identify factors that contribute to the impact of peer support on service users, this study will explore service users' views on positive/ negative effects of peer support, and factors which moderate the positive/negative effects of peer support on social inclusion, (self)stigmatization, empowerment, hope, recovery, and illness concepts.
By using an open approach, the study will shed light on service users' everyday experiences of the intervention beyond what is already known or expected regarding the mechanisms of impact including active ingredients and conditions for optimal experiences as well as barriers and contextual influences.
Participants: Following a mixed-method approach, 6-8 service users from the intervention group at each site will be purposively selected at the end of intervention using their quantitative outcome data to include those with and without benefit (1:1) of peer support with regard to social inclusion and experiences of peer support. Participants will be selected based on their ratings on the Social Inclusion Scale (SIS [30]) and Brief INSPIRE Participants: Mental health workers working in the same team as an UPSIDES PSW.
Methods: Between 6 and 8 mental health workers in each site will take part in a focus group before and after the intervention. A replacement clinician will be identified for follow-up if a baseline focus group attendee is no longer available (e.g. because they have moved jobs). Participant diversity in age, gender and years since qualifying will be maximised. Data will be collected using an open focus group guide based on an implementation scale developed on a systematic review, pertaining to elements of organisational culture, training, role clarity, resourcing, and access to a peer network and more that came up as crucial for successful peer support implementation [16]. In addition, at baseline and follow-up, focus group participants will be asked to fill-out a 16 item questionnaire (Mental Illness: Clinicians' Attitudes Scale, MICA [53]) in order to assess changes of stigmatising attitudes towards people with mental illness following working with PSWs.

# 4: Key informants' experiences of peer support
Objectives: To explore the practical consequences on dealing with the promises and challenges of peer support from the perspectives of key informants and to assess implementation outcomes in line with the Consolidated Framework for Implementation Research (CFIR [15]). This focus group study will evaluate stakeholders' views on (a) perceived changes/difficulties prior to the intervention vs. implementation outcome; (b) barriers and facilitators for successful implementation of peer support in the given institution, and (c) the need for changes or alterations to make the intervention work effectively in a specific context.
Participants: Key informants will be defined as experts with specific authorities and/or responsibilities in regard to the implementation processes at study sites. Experts are supposed to have specific contextual (organisational) knowledge about implementation of peer support in mental health settings (e.g. policy makers, representatives of service user, provider, and funder organisations). Experts will be selected according to the local structures of mental health services.
Methods: At each site, focus groups with 4-6 key informants will be held at the beginning and end of the intervention. Purposive sampling strategies will be applied based on study site researchers' knowledge about local relevant key informants. Topics covered will include acceptability, appropriateness and feasibility, to be guided by the CFIR Interview Guide Tool [54], including characteristics at the levels of the intervention (e.g.
advantages, risks/costs, cultural adaptability), the outer setting (e.g. barriers and facilitators for meeting service users' needs, policies and incentives), the individual (e.g. knowledge, values), and the intervention processes (e.g. obstacles and facilitators during planning, execution, and evaluation).

Discussion
With a balanced representation from high-, middle-and low-income countries, at different stages of peer support, UPSIDES represents a significant contribution to international implementation research on peer support. The UPSIDES-RCT offers an opportunity to explore the essential components necessary to create a peer support model in mental health care, while providing the evidence required to sustain and eventually scale-up the intervention. UPSIDES involves an effort to internationally implement and evaluate a peer support intervention for people with severe mental illness across low-middle and highincome countries, harmonising core elements across different cultural and organisational dimensions. Given the scarcity of mental health workers in LMICs and recent paradigm shift towards mental health recovery in high-income countries, involvement of peers in mental health care has the potential to improve and to transform mental health services in low-, middle-and high-resource settings.
This study addresses several knowledge gaps and limitations in previous RCT designs and is in line with calls to continue to research and improve the quality of studies on peer support in mental health [6,55]. First, evaluating the development of an effective peer intervention using a longitudinal design combining both quantitative and qualitative methods has been previously recommended [56]. Second, the qualitative component of the study will yield insights into the complex processual aspects of implementing peer support; for example, what is the role of a PSW and what do peer providers do to successfully create change, both for recipients and for organizations? Third, the RCT addresses ethical concerns regarding the use of comparison groups in intervention studies for vulnerable populations, by adopting a wait-list design. Finally, as much as possible this study protocol follows best practices for the minimization of research biases, while taking into account differences in local contexts [57].
Previous reviews of randomised trials of peer support have highlighted the heterogeneity of interventions and outcome measures used in many studies, positing that lack of standardisation may contribute to null or mixed results in this area of research (e.g. [6]).
The current study addresses these concerns by developing and testing a peer support intervention based on core elements common to all participating study sites, and further investigates processes of change that have previously been identified as relevant to mental health peer support interventions. Further, it will focus on social inclusion as the primary outcome, alongside other outcomes directly related to personal recovery, in line with established definitions of mental health peer support [8,56,58]. In addition, the study will include mental health and economic outcomes reported to be missing from many studies of peer support, according to previous reviews [6,55].
Conducting an international implementation research study will allow for the identification of universal versus local/contextual elements of the intervention, contributing to the evidence base for peer support and its theoretical underpinnings. It will characterise what this intervention looks like in different high-, middle-and low-income countries, and specify contextual factors required for its successful implementation. Identifying nonmodifiable and modifiable components of peer support will help to inform the implementation of peer support in other countries and settings.

Limitations
A multicentre RCT can prove challenging, particularly in low resource settings and given differences in mental health care structures. For example, aligning timelines for tasks can be challenging with varied infrastructural and contextual challenges. Ethical review protocols and committee recommendations vary across settings, requiring flexibility within the overall framework without compromising on quality and fidelity to the model. While overall there has been much involvement in the development of the intervention by users and peer support workers, the study planning was more heavily reliant upon researchers in academic institutions. It is also necessary to plan for the high likelihood of attrition among peer staff, who may need to take sick-leave or drop out unexpectedly. Proper supports and accommodations for PSWs will be put in place to maximise retention and sustainability.

Conclusions
UPSIDES is the first multisite RCT to study peer support in a range of low-, middle-and high-income countries, addressing a number of geographical, methodological and other knowledge gaps in international research on mental health peer support. By explicitly studying process as well as outcomes, UPSIDES asks not just whether peer support works in these settings, but how, in order to provide practical guidance on the implementation and scale-up of peer support in different settings. Ultimately, UPSIDES aims to inform mental health policy, implementation and practice, to ensure that the perspectives and potential contributions of people with lived experience are taken into account.

Abbreviations
Not applicable.

Competing interests
The author(s) declare that they have no competing interests. Change model underpinning peer worker interventions [14].

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.