Methodology

The trial protocol was developed with input from the research team, collaborating partner organisations and the Priment Clinical Trials Unit (University College London, London, UK). A summary of the protocol has been published.43

Design

The study design was a two-arm, parallel-group, researcher-blind pilot RCT with 1 : 1 allocation. Participants were randomly allocated to either the intervention group, where they received one-to-one befriending by a volunteer, or an active control group. Both groups had access to usual care and a booklet of local resources. A process evaluation, based on mixed methods, was conducted to examine the delivery and adherence to the intervention, and to explore stakeholder views on the acceptability of the intervention.

Eligibility criteria

To be enrolled to the study, participants with ID had to:

• be aged ≥ 18 years
• have mild or moderate ID (i.e. an IQ of 35–69) assessed using the Wechsler Abbreviated Scale of Intelligence™ (Second Edition) (WASI-II™)44
• not be attending college/education or a day service for ≥ 3 days per week
• have obtained a score of ≥ 5 on the GDS-LD,42 indicating the presence of depressive symptoms, but did not need to have a clinical diagnosis of depression
• be able to speak English
• be able to provide informed consent.

To take part in the study, volunteers had to be:

• aged ≥ 18 years
• available once per week for at least 1 hour, over a period of 6 months.

Volunteers did not need to have any prior experience of supporting people with ID.

Exclusion criteria

Individuals with ID were excluded if they:

• had limited communication and comprehension that would prevent completion of the questionnaires.

Volunteers were excluded if they:

• had a criminal record (any documented offence, owing to the vulnerability of the befriendee population) recorded on their Disclosure and Barring Service (DBS) check
• were unable to provide two references or had unsuitable references.

Sample size

We did not have any estimates of the number of people with ID who were likely to be eligible and consent to taking part in the trial. If we approached 50 participants who were eligible to take part, this would allow us to estimate an expected recruitment rate of 80% (40 people), with a 95% CI of 68.9% to 91.1%. A sample size of 40 recruited people with ID would allow us to estimate a 30% dropout rate in the trial, with a 95% CI of 25.7% to 54.3%. The recruitment period was 6 months. There were two participating befriending services and, therefore, we needed to recruit 3.3 participants with ID per month at each site.

Recruitment

The recruitment period for the trial was from mid-April 2019 to mid-September 2019 (5 months). Recruitment issues are discussed in Chapter 3.

Two befriending services agreed to take part in the study. One was based in east London (Outward, based in Hackney) and the other was based in Suffolk (The Befriending Scheme). They were responsible for recruiting and training volunteers, matching volunteers to participants with ID, the provision of supervision to volunteers and monitoring of the befriending relationship. Both services had previous experience of supporting befriending relationships with people with ID.

Participants with intellectual disability

Participants with ID were recruited from existing and new referrals to the participating befriending services or were recruited from the caseload of clinicians working in four community ID services within the North East London NHS Foundation Trust (NELFT). Volunteer co-ordinators and clinicians screened referrals and caseloads for participants who were potentially eligible for the study and approached individuals to discuss the study. If the individual was interested in the study, they were given an information sheet and permission was sought for their details to be passed on to the research team. A referral form was then completed by the clinician or staff at the befriending service and e-mailed to the research team. The trial research assistant reviewed the information, contacted the potential participant to discuss the study and carried out the eligibility assessment, which comprised an assessment of IQ using the WASI-II (the two subset version was used).44 The WASI-II was selected because it is a brief and reliable measure of cognitive ability and can be used to estimate the full-scale IQ across all population groups, including people with ID, which is useful in the context of research. The two-subset version consists of vocabulary and matrix reasoning and takes approximately 15 minutes to complete. The eligibility assessment also comprised an assessment of the presence of symptoms of depression using the GDS-LD.42 If the individual was eligible, they were required to sign the consent form before completing the baseline assessment. If the research assistant was uncertain about whether or not a participant had met the eligibility criteria, the case was discussed with the chief investigator before the baseline assessment was carried out.

Volunteers

Volunteers were recruited by the befriending services through newspaper advertisements, befriending and job websites, social media, and recruitment events at colleges and universities. Information about the study was also available on the study website hosted by University College London, Division of Psychiatry (https://www.ucl.ac.uk/psychiatry/research/epidemiology-and-applied-clinical-research-department/bid-project), and study posters were circulated to psychology undergraduate students. Interested volunteers completed an application form and were invited to an informal interview to assess their suitability and motivation for becoming a volunteer. A DBS check was completed to ensure that they had no prior criminal record, and references were obtained. After completing training (see Training and supervision of volunteers), successful candidates were invited to take part in the study. They received an information sheet and were asked to sign a consent form before completing the baseline assessment.

Randomisation

Following the baseline assessment, details of the participant with ID were entered into a web-based randomisation system hosted by sealed envelope™ (Sealed Envelope Ltd, London, UK) by a non-blinded member of the research team. The randomisation procedure used randomly varying block sizes, stratified by befriending service. The randomisation protocol was created by the trial statistician and the set-up of the service was overseen by the Priment Clinical Trials Unit.

Participants were randomly allocated by the system to either the intervention group or the control group. The unblinded researcher informed the befriending service and participants with ID of the group to which they had been allocated. Owing to the nature of the intervention, it was not possible to blind participants to their assigned allocation group. However, outcome assessments were carried out by a research assistant who was blind to the participants’ allocation group.

Intervention group

Participants who were allocated to the intervention group were contacted by the volunteer co-ordinator, who arranged to meet them to discuss their hobbies, interests and activities that they would like support to participate in. Based on this information, participants were matched to a volunteer who could accommodate the person’s interests and availability.

The befriending intervention

Frequency and types of activities

The befriending intervention was adapted from the existing models used by the two participating befriending services and from other studies of befriending.37,40 The purpose of the befriending relationship was to provide friendship and emotional support, to facilitate community access, and to help the individual gain skills and confidence that would allow them to access activities in the community on their own (i.e. to promote sustained social activities beyond the befriending relationship). Figure 1 is a logic model of the intervention and describes the proposed mechanisms of action.

FIGURE 1 Logic model of befriending in people with IDs. (Figure not reproduced here; see the original article.)

The volunteer (befriender) and person with ID were expected to meet once per week for at least 1 hour, over a period of 6 months. However, it was anticipated that there may be breaks due to holidays or illness. The pair received a booklet with information about local activities and amenities, which was to be used as an aid to plan activities. The emphasis was on assisting the individual to make choices about the activities that they wished to pursue. The volunteer was not expected to carry out personal care, administer medication or accompany the individual to medical appointments. Contacts by telephone/social media were permitted in addition to face-to-face contact, and the pair were expected to spend at least 50% of the total number of meetings in the community. Meetings could take place during evenings/weekends, depending on the pair’s availability. Volunteers were asked to keep a record of their activities in a structured log, detailing whether or not they attended, reasons for cancellation, what they did at each visit, the duration of the activity and types of contact other than face to face. Volunteers’ travel expenses were reimbursed, but other expenses had to be agreed with the befriending service. Participants with ID were expected to pay for their travel and the other costs they incurred during outings.

Introduction and monitoring of the befriending relationship

A face-to-face meeting was arranged by the volunteer co-ordinator at the participant’s home to introduce the volunteer to the participant. If both parties were satisfied with the matching process, the pair then arranged to meet on their own. If the volunteer or individual with ID dropped out of the relationship once it had become established, they were offered the option of rematching. Volunteers were permitted to be matched to more than one participant with ID, although this did not occur in the trial.

Six weeks after matching, the volunteer co-ordinator contacted each person by telephone/face to face and then maintained contact every 4 weeks to monitor the progress of the relationship. A further meeting was held with the pair towards the end of the 6-month period to obtain general feedback about the befriending intervention, to discuss ending the relationship and to support the individual with ID in coming to terms with the ending. The pair could continue their relationship if they wished after the 6-month period was over, but arrangements for monitoring the relationship were outside the remit of the trial and varied according to the befriending service. Information was collected on any relationships that had continued beyond 6 months.

Training and supervision of volunteers

Volunteers completed face-to-face training and electronic learning prior to being matched. The training covered the following topics: benefits of befriending and issues related to confidentiality and lone working; advice on how to plan meetings effectively; health and safety; safeguarding; intellectual (learning) disability awareness, and professional boundaries, such as dealing with sensitive issues, ending relationships and expectations of the role of the volunteer. Volunteers received PowerPoint (Microsoft Corporation, Redmond, WA, USA) slides and a manual developed by the research team, with information about the study and additional information relevant to working with people with ID, such as maximising communication, understanding the reasons for challenging behaviour, common mental health problems and what volunteers should do in an emergency.

Volunteers had access to individual supervision with the volunteer co-ordinator, which was delivered face to face or over the telephone once per month. The supervision sessions addressed issues related to the training needs of the volunteer and how to overcome issues that had arisen from the befriending relationship.

Training and support for befriending services

The two befriending services completed good clinical practice training and received training on the trial research processes and procedures as part of the site initiation visit. Volunteer co-ordinators received support from the research team through regular e-mail and telephone contact, and were members of the Trial Management Group.

Control group

To control for the effects of participants in the treatment group receiving more information about local activities than those in the control group, participants in the control group were also given a copy of the activities booklet, and met with a member of the research team who discussed the booklet with them (and their carer, if present) and encouraged them to engage with activities.

Both the control group and the intervention group had access to ‘usual care’. This included access to multidisciplinary input from community ID services, including appointments with their psychiatrist, psychologist and other members of the team. Participants continued to take their usual medication, which included antidepressants and antipsychotics. They had access to primary care and other community and hospital-based health services, as well as day services (e.g. day centres and colleges). Where it was possible, participants in the control group were asked if they wished to be matched with a volunteer after they completed their follow-up assessment.

Outcomes

Feasibility outcomes

Recruitment and retention of participants

We assessed the number of participants with ID recruited from those who were eligible; the number of volunteers who were recruited over a 6-month period; the number of participants who were successfully matched with a volunteer; the number of participants and volunteers who dropped out of the intervention; and the number of participants and volunteers who completed the follow-up assessment.

Adverse events

Adverse events were collected at each follow-up assessment using open-ended questions and were also reported directly to the chief investigator by the befriending services, including concerns about safeguarding. Adverse events were recorded in the medical records and case report forms (CRFs).

Adherence to the intervention

Adherence was assessed as part of the process evaluation (see Treatment adherence).

Acceptability of the intervention

This was informed by data on retention/dropout of volunteers and participants; the data from the process evaluation (see Process evaluation) that examined the extent of engagement with the intervention by participants and volunteers (based on number of sessions attended); and qualitative data obtained from volunteers, participants with ID, carers of people with ID and staff from the befriending service.

Outcome measures in participants with intellectual disability

Health and social outcome measures were assessed with the participant at baseline and at 6 months post randomisation. All the measures that were selected for use in the study have been validated and used in people with ID. A follow-up assessment at 12 months was planned as part of the original protocol but could not be conducted because of delays in setting up the study and poor recruitment (see Chapter 3). Face-to-face assessments were conducted by the research assistant, who was blind to group allocation, and these were carried out at the participants’ homes; where this was not possible, a telephone assessment was carried out. Participants were supported by the research assistant to complete the measures, which included reading out the questions and responses, and explaining and rephrasing items that were difficult to understand. The following health and social outcomes were assessed:

• Depressive symptoms (main outcome of interest) were measured using the GDS-LD.42 This 20-item scale has been developed and validated in people with ID and has been used in recent trials.45 Participants were asked to rate their symptoms over the last week using a three-point response format scored 0 (‘no’), 1 (‘sometimes’) and 2 (‘a lot’). Scores on the scale range from 0 to 40, with higher scores indicating more symptoms. The scale is reported to have good test–retest reliability (r = 0.97) and internal consistency (with an alpha of 0.90).
• Self-esteem was measured using the adapted Rosenberg self-esteem scale for people with intellectual disabilities.46 This is a self-report scale with six items measured on a five-point response scale ranging from ‘never true’ to ‘always true’. Scores range from 6 to 30, with higher scores indicting greater self-esteem. The scale is reported to have an adequate internal consistency (with an alpha of 0.62).
• Quality of life was measured using the 19-item Maslow Assessment of Needs Scale – Learning Disability (MANS-LD),47 and nine items from the adapted World Health Organization Quality of Life measure (WHOQOL-8).48 Both measures are rated using a five-point response format (1–5), where higher scores indicate better quality of life.
• Loneliness and social satisfaction were measured using the Modified Worker Loneliness Questionnaire (MWLQ),49 which has 12 items and two subscales (aloneness and social dissatisfaction). Items are rated on a three-point scale (0–2), with a maximum score of 24. Higher scores indicate more loneliness. The test–retest reliability of the scale is reported to be between 0.76 and 0.89, and inter-rater reliability was 0.87 for aloneness and 0.85 for social dissatisfaction. The internal consistency (Cronbach’s alpha) was 0.65 for aloneness and 0.80 for social dissatisfaction.
• Social support was measured using the Social Support Self Report for intellectually disabled adults (SSSR).50 Participants are asked to rate support from family, staff, friends and their partner based on five items rated on a three-point scale (0–2) for each group. The maximum score is 40, with higher scores indicating receipt of more support. The scale has good internal consistency (with an alpha of 0.85).
• Social participation was measured using the Guernsey Community Participation and Leisure Assessment (GCPLA).51 Participants are asked to rate the frequency of their participation in five types of activities (i.e. services, vocational activities, leisure, social, and facilities and amenities) on a five-point scale ranging from never (0) to daily or more frequently (5). Higher scores indicate more social participation. It has good test–retest reliability (r = 0.87) and internal consistency (with an alpha of 0.93).

Outcome measures in volunteers

The following outcomes were assessed in volunteers at baseline (prior to matching) and 6 months after the baseline assessment:

• Self-esteem was measured using the 10-item Rosenberg self-esteem scale.52 Items are scored on a four-point Likert scale ranging from strongly disagree (1) to strongly agree (4). The maximum score is 40. Higher scores indicate better self-esteem. Reported test–retest reliability ranges from 0.82 to 0.88 and internal consistency (Cronbach’s alpha) ranges from 0.77 to 0.88.
• Psychological well-being and quality of life was measured using the widely used Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS).53 This is a 14-item scale that uses a five-point response format (1 = none of the time; 5 = all of the time). The maximum score is 70 , with higher scores indicating better well-being. It has a high internal consistency (with an alpha of 0.89) and a test–retest reliability value of 0.83.
• Loneliness was measured using the revised University of California, Los Angeles (UCLA) Loneliness Scale.54 This is a 20-item scale that measures feelings of loneliness and isolation. Items are rated on a four-point scale (1 = never; 4 = often). The maximum score is 80 and higher scores indicate more loneliness. The measure has a high internal consistency (with an alpha of 0.96) and a test–retest reliability value of 0.73.
• Attitudes of volunteers was assessed using the 67-item Attitudes Towards Intellectual Disability Questionnaire (ATTID).55 This measure has five subscales (knowledge of causes of ID, knowledge of capacity and rights, discomfort, interaction, and sensibility or tenderness). Higher scores indicate more negative attitudes. Test–retest reliability ranges from 0.44 to 0.88, with an overall internal consistency of 0.92 and a subscale reliability ranging from 0.59 to 0.89.

Feasibility of carrying out a cost-effectiveness analysis

A preliminary health economic analysis was conducted to inform planning of future economic analyses, sources of data required and how best to collect these data. The following were recorded:

• Health-related quality of life was measured using the EuroQol-5 Dimensions – Youth (EQ-5D-Y).56,57 There are currently no EuroQol measures that have been developed specifically for adults with ID. The EQ-5D-Y was selected because it uses simple language and has been used in other studies of people with ID.45 The EQ-5D-Y comprises the descriptive system (five dimensions relating to mobility; looking after myself; doing usual activities; having pain or discomfort; and feeling sad, worried or unhappy) that is rated by selecting the most appropriate statement that reflects the individual’s health state (‘no’, ‘some’ or ‘a lot’) for each dimension, and the EuroQol visual analogue scale (EQ VAS), where self-rated health is recorded on a vertical visual analogue scale between two end points labelled ‘the best health you can imagine’ and the ‘the worst health you can imagine’. The EQ VAS represents the participant’s own subjective view about their health, which may not be captured by the five dimensions, and is therefore considered to provide additional information in its own right. We assessed whether or not it would be feasible to collect self-reported data from individuals with ID on both the descriptive system and the EQ VAS. This was recorded at baseline and 6 months post randomisation.
• The Client Services Receipt Inventory (CSRI)58 was used to assess the feasibility of collecting data on participant-reported health-care service use. Based on a review of the literature and input from the wider study team, we adapted the CSRI to people with ID. The adapted CSRI was completed with participants’ carers, where possible, or health professionals who knew the person well. If they did not have a carer, then information was collected from the participant. They were asked to recall health and social care service use during the last 6 months. This included contacts with primary care, community services, community ID services and hospitals, including outpatient and inpatient care. This was recorded at baseline and at 6 months post randomisation.
• Concomitant medication was recorded on a CRF developed for the study. This information was obtained from carers and, where this was not possible, participants were asked to show medication boxes or blister packs to the research assistant. Information was recorded at baseline and at 6 months post randomisation for the previous 6 months.
• For the befriending intervention, we recorded the resources used by befriending services. Volunteer co-ordinators at the befriending services kept a record of all their contacts with volunteers and participants and the duration of contacts (e.g. monitoring visits and supervision) and other expenses incurred during the intervention.

Process evaluation

A process evaluation using mixed methods59 was conducted to examine:

• Adherence (fidelity) to the intervention: whether or not the different components of the intervention were consistently followed and the extent to which volunteers delivered the intervention as intended
• Acceptability of the intervention and trial procedures. The aim was to understand the perceived value, benefits and unintended consequences of the intervention; to understand the contextual factors influencing how the intervention was delivered; and to develop an understanding of the likely mechanisms of action of the intervention and what modifications that would need to be made to the trial procedures and processes in a future randomised trial.

Treatment adherence

Data were collected on the uptake of volunteer training, supervision and the frequency of monitoring checks carried out by volunteer co-ordinators from routine records at each site to assess fidelity to the intervention by the befriending services. To understand how the befriending intervention was delivered, the location and content of the meetings and the range of activities undertaken were described based on an analysis of structured logbooks completed by volunteers. A framework was developed to categorise different types of activity to enable different types of befriending support to be distinguished and quantified.

We examined (1) how many sessions each volunteer attended and the average number of sessions attended by the volunteers, (2) how many pairs attended at least 10 meetings during the 6-month intervention period, (3) the average duration of sessions and (4) for how many participants the minimum threshold of at least 50% of meetings being outside the participant’s home was achieved.

Acceptability of the intervention

We invited all the participants with ID in both groups of the study, the volunteers who had been matched with a participant with ID, the volunteer co-ordinators at the two befriending services and carers of participants who had received the befriending (after seeking permission from the participant) to take part in semistructured interviews. A topic guide comprising questions and prompts was developed for each of the four groups (see Appendix 1, Boxes 1–4, for a list of questions). Interviews were conducted face to face, but during the COVID-19 pandemic modifications needed to be made to accommodate the lockdown restrictions. Interviews were therefore held by telephone or videoconferencing.

All the participants were asked about their views of the intervention (e.g. training, matching, supervision, support), what they perceived to be helpful and what aspects could be improved or required modification, as well as views about trial procedures (e.g. randomisation and completion of questionnaires). The interviews were audio-taped and transcribed.

Study expenses

Following completion of the baseline assessment and, again, after the 6-month follow-up assessment, all the participants and volunteers received a £15 gift voucher for their time and effort. Participants taking part in the qualitative interviews also received a £15 gift voucher.

Data management

Quantitative data were collected using the study CRFs, which were stored at the study sites. Data were entered into the secure and General Data Protection Regulation-compliant60 Red Pill (web-based) database hosted by Sealed Envelope Ltd and managed by Priment Clinical Trials Unit. A subsample of 5% of the CRFs was checked to ensure consistency in data entry. After completion of relevant data management processes, the data were exported from the database and analysed by the trial statistician.

Statistical analysis

Demographic and clinical characteristics of participants with ID and their mental health, quality of life and social outcomes were summarised by randomised group using mean [standard deviation (SD)] or count (percentage) for continuous and categorical data, respectively, at baseline and at the 6-month follow-up.

Demographic characteristics of befriending volunteers and their mental health and attitudinal outcomes were similarly summarised at baseline and at the 6-month follow-up.

The primary mental health outcome was depressive symptoms, measured using the GDS-LD. All participants were analysed in the groups to which they had been randomised. The effect of the befriending intervention was estimated using a linear regression model, with depressive symptoms at 6 months’ follow-up as the outcome and study group as the main explanatory variable, adjusting for depressive symptoms at baseline.

Owing to the small number of participants recruited to the study, we decided that it was inappropriate to explore the effect of the intervention on the secondary outcomes and further regression analyses were not conducted

Analysis of cost-effectiveness

The main aim of the economic evaluation was to determine the costs of delivering the intervention, and test the feasibility of collecting health-care resource use and health-related quality-of-life data to calculate the incremental mean cost per quality-adjusted life-year (QALY) gained from the intervention, compared with the control (i.e. the cost-effectiveness), in people with ID for a full RCT. The detailed health economic analysis is described in Appendix 2.

We assessed the response rates from the CSRI, EQ-5D-Y and EQ VAS; and collected information on whether the CSRI was completed by carers, professionals or participants with ID. We noted any difficulties reported by participants in completing these measures.

The costs of the intervention were provided by the volunteer co-ordinators at the befriending services. They provided a breakdown of costs related to the intervention, such as costs of recruitment, DBS checks, and staff time in relation to training and supervision of volunteers and monitoring visits. Staffing costs were calculated using an hourly rate.

To calculate the costs of using health and social care services, we adopted the perspectives of the NHS and Personal Social Services. The horizon of the study was 6 months; therefore, discounting was not performed.

The mean (SD) health and social care resource use and costs of medication were calculated and costed using nationally recognised, published unit costs61–63 to estimate the mean (95% CI) costs per participant. Costs were in Great British pounds at 2019 values or adjusted for inflation if costs for 2019 were not available (see Appendix 2, Tables 10 and 11, for unit costs and prescription costs).

Responses from the EQ-5D-Y were converted into a single value (index number). The collection of index values for possible EQ-5D states is referred to as the value set. Research is currently ongoing to produce EQ-5D-Y value sets and, therefore, although not recommended, the regular EuroQol-5 Dimensions, three-level version (EQ-5D-3L), value sets were used to obtain utility values. Utility values range from 0 to 1, with scores close to 1 indicating ‘full health’ (with a score of 0 indicating ‘death’). The EQ VAS was used as a quantitative measure of health outcome that reflects the participant’s own judgement about their health status. We examined whether or not scores on the EQ VAS were consistent with the EQ-5D-Y utility values, as this would help us to determine if the EQ-5D-Y is a suitable measure of health status in people with ID.

Incremental cost per change in the primary health outcome (depression scores on the GDS-LD) between the intervention group and the control group was estimated. The incremental cost-effectiveness ratio (ICER) was calculated by dividing the difference in total costs (incremental cost) by the difference in the health outcome (incremental effect) to provide a ratio of ‘extra cost per extra unit of health effect’ for the more expensive intervention compared with the alternative. ICERs reported by economic evaluations are compared with a predetermined threshold to decide whether or not choosing the new intervention is an efficient use of resources.

Analyses followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) Good Reporting Practices.64

Qualitative data analysis

Transcripts from the interviews with participants with ID, staff at befriending services, volunteers and carers were transcribed and analysed using thematic analysis65 supported by computer software (NVivo 9; QSR International, Warrington, UK). Thematic analysis aims to identify themes or patterns in the data set through a rigorous process of data familiarisation, coding of data, and the development and refinement of themes. It was selected because of its flexibility in incorporating a range of different approaches and its use in a wide range of contexts. Data were coded by two members of the research team, using both deductive (i.e. coding guided by existing concepts and ideas) and deductive (i.e. coding guided by the content of the data) approaches. The codes and initial themes were discussed with the research team, followed by further revision of themes. The analytic strategy aimed to identify themes related to the barriers and facilitators affecting the implementation of the intervention, potential benefits to participants and volunteers, and possible mechanisms of action. Analysis also allowed for the consideration of themes that arose more inductively from the data. The reliability of the coding frame was not assessed.

Progression criteria

The following criteria were used to evaluate whether or not the pilot trial had been successful:

• At least 70% (35) of participants are recruited from the 50 potentially eligible individuals who are approached.
• At least 70% of participants with ID in the intervention group are successfully matched to a volunteer (befriender) and they meet on a minimum of 10 occasions.
• The dropout rate of participants with ID in both groups, post randomisation, is < 30% at 6 months’ follow-up.
• The intervention and trial procedures are considered to be acceptable by volunteers and individuals with ID.
• If the target recruitment rate and number of matched pairs is between 50% and 69%, or the dropout rate is between 31% and 50%, we would consider if these could be rectified by modifying the protocol. If this was the case, a full study could still be considered. If the recruitment rates were < 50% and the dropout rates were > 50%, then a full trial would not be considered feasible.

Management of the study

The study was sponsored by University College London. The study research assistant was supervised by the chief investigator and they met weekly. The Trial Management Group, comprising the research team, co-applicants, and staff from the two befriending schemes and Priment Clinical Trials Unit, met every 3 months. An independent Trial Steering Committee, comprising five members [an academic clinician, clinician in ID, statistician, health economist and patient and public involvement (PPI) representative] was appointed and approved by the funders. Meetings were held annually and minutes were made available to the funders.

Patient and public involvement

Proposal development

In developing the proposal, three befriending services for people with ID were consulted. They agreed that it was important to carry out a trial on befriending that captured a range of outcomes for individuals with ID and volunteers. Two of these services subsequently agreed to take part in the study and contributed to the design of the study. In addition, a consultation group meeting was held in July 2017 with volunteers, befriendees and volunteer co-ordinators at the Suffolk service. We discussed the nature of the befriending intervention, its impact on both befriendees and volunteers and the best choice of outcomes for the study. Feedback was obtained from an advisory group of carers and service users linked One Westminster (London, UK) befriending service who thought that the study was very important. They highlighted areas that they felt should be addressed carefully, such as ensuring participants with ID were safeguarded, that they had opportunities to voice concerns if they were dissatisfied, and that adequate training and supervision were available for volunteers. These suggestions were incorporated into the proposal.

Patient and public involvement during the trial

We advertised for people with ID who had experience of befriending to become members of an advisory group. Unfortunately, despite some initial interest, we were unable to recruit enough members during the initial stages of the trial. Following the closure of recruitment and to understand the reasons for the poor recruitment, we consulted with three people with ID and a volunteer co-ordinator who were members of Gig Buddies, a befriending service provided by My Life My Choice (Oxford, UK), a voluntary organisation. These individuals had experience of befriending as a peer befriender, which meant that they were befrienders to other people with ID. They had a unique perspective on the befriending experience. They suggested some reasons for the poor recruitment, but were largely positive about the study. These suggestions are discussed in Chapter 5.

Ethics issues

The study received ethics approval from the London – City and East Research Ethics Committee (reference 18/LO/2188) and from the Health Research Authority (HRA) on 8 January 2019. All substantial amendments to the study were approved by the sponsor and were submitted for ethics approval. Participants with ID were provided with an accessible information sheet and were required to provide written consent prior to taking part in the study. Volunteers, volunteer co-ordinators and carers were also provided with an information sheet and were required to give written consent. Data confidentiality was maintained by assigning participants study identification numbers. One of the main concerns about the study was the potentially negative impact on the participant with ID of relationships ending, particularly if they had limited access to other forms of support. Unfortunately, it was not possible to explore these issues in detail as we were unable to complete the 12-month follow-up (reasons for which are explained in Chapter 3).

Several ethics issues arose during the study. First, the Suffolk befriending service raised concerns about the randomisation of participants and expressed the view that it was unacceptable to assign participants to the control group. To manage this issue, it was agreed that individuals referred to the service would have the option of taking part in the study or remaining on the waiting list. This had significant consequences for recruitment (discussed further in Chapter 3). Second, carers were occasionally found to obstruct relationships and, on two occasions, appeared to be responsible for the breakdown of the befriending relationships. This situation was distressing for the volunteers. In both cases, the participant with ID did not express dissatisfaction with the relationship. This raises questions about the lack of autonomy that some people with ID may have in making decisions. Finally, one safeguarding alert was raised by a volunteer when their befriendee reported being bullied by a paid carer (discussed further in Chapter 3).