Different methods were used in the intervention development. Some of these activities took place concurrently (eg, the focus group study and survey) and were followed by several iterations of consultations with the advisory groups. The review was conducted between October 2015 and December 2015, the focus groups were conducted between January 2016 and September 2017, the survey was conducted between August 2016 and August 2017, and the advisory groups were consulted from February 2017 to February 2020.

A rapid narrative literature review was conducted to enable efficient mapping of the main results related to the specified field. It constitutes a useful technique for intervention development where a broad perspective of the literature is required within a limited timetable [30,31]. Key papers and theories for developing interventions and digital interventions were subsequently searched [32].

Stakeholders potentially linked to the provision of volunteering (ie, mental health professionals and volunteers) were interviewed in a focus group format to explore their views regarding the relationship formats between patients and volunteers [32]. A total of 24 focus groups were conducted with 119 participants. The process and findings of the focus groups have been described elsewhere [33,34].

Patients’ preferences with respect to contact with a volunteer were assessed through a survey that evaluated potential relationship formats between patients and volunteers and the patients’ preferred volunteer characteristics [22]. A total of 151 patients with psychosis followed in outpatient services in London were interviewed. The findings of this study have been published elsewhere [22].

In this research, both patients and volunteers were also included as patient and public involvement advisors [35]. The lead author (MPdC) worked with organizations of volunteers and patients consulting them regarding intervention development [36,37].

The lead author collaborated with a national volunteering association, Befriending Networks [38], which is the umbrella charity for befriending services operating across the United Kingdom and beyond. In September 2017, the lead author organized an event in London that brought several volunteer organizations and volunteers together. The event raised awareness of research on volunteering in mental health and was attended by volunteers involved in charities of mental health. Many of these volunteers contributed to the focus groups conducted during this event. Following the meeting, 2 volunteers also worked with the lead author through the advisory group offering their input for the intervention development process.

To establish patient views, the lead author worked with the Service User and Carer Group Advising on Research at City University of London [39]. This group comprised 14 mental health service users who had personal experience with a variety of mental health conditions, including psychosis.

These groups were consulted throughout the intervention development phase; the meetings involved discussions based on written materials or presentations. The opinions of potential users (ie, both patients and volunteers) were sought concerning which components they thought should be included in an intervention that was aimed toward the provision of opportunities for volunteering in mental health.

The Phone Pal logic model was then developed in consultation with the advisory groups, and it explored a set of person-based intervention components and the potential outcomes that the intervention could affect aligned with the participants’ opinions on what to achieve. The groups advised on the suitability of different components and provided recommendations on the role and characteristics of patients and volunteers in this intervention.

These potential users offered important insights to establish the guiding principles of the intervention. They also discussed and specified the key components and objectives that would improve acceptability and engagement with a digital intervention.

Regular meetings with a multidisciplinary team of 30 people comprising principal investigators, postdoctoral researchers, PhD students, researchers, public health experts, anthropologists, psychologists, and psychiatrists at the Unit for Social and Community Psychiatry were used to discuss the intervention development process. The lead author presented the guiding principles, and the team advised on the Phone Pal logic model development.

The findings emerging from the aforementioned methods were synthesized and presented to the advisory groups. Although there were divergent views on occasions, these group discussions frequently allowed one view to emerge as preferred or, on some matters, there was even consensus from the advisory group members. Further to the consultation with the advisory groups, actions were agreed upon to guide the design of key components of the intervention, the development of the logic model, and the intervention operationalization. The decisions that addressed the identified challenges were acceptable to patients and volunteers, were informed by theory, could be achieved pragmatically within this research, and were endorsed by the expert group.

The process for selecting the intervention core components and objectives was guided by the intervention guiding principles. The guiding principles of the intervention communicate how the objectives and components of the intervention address the particular contextual challenges identified during the process. After an extensive process of gathering the existing evidence, three core intervention components were outlined: (1) a match (ie, patient-volunteer), (2) remote communication, and (3) a smartphone, as depicted in Figure 2.

The objectives established were (1) to provide one-to-one access to a match, (2) to provide this access to a person remotely, and (3) to allow the communication to occur digitally. Table 1 presents the core components, intervention objectives, and contextual factors.

The following section describes the process of logic model development, portraying the decision-making for the selection of suitable intervention components—namely, the volunteers’ characteristics and role, the patients’ characteristics and role, and being connected remotely and through a smartphone—using the methods that informed them (Table 2).

The guiding principles aim to address the contextual challenges that are likely to affect intervention delivery. They consist of 2 aspects (ie, the intervention objectives and the core components used to operationalize them), which were developed to ensure that the intervention met the volunteers’ and patients’ expectations [68].

This logic model was assembled with the contextual factors and anticipated outcomes following a process of collating existing evidence, varying theories, and conceptualizations. Table 3 illustrates the guiding principles of the intervention, hypothesized mechanisms of change, and outcomes that may be affected by intervention use. The intervention components were selected after feedback to theoretically promote engagement via role modeling and self-efficacy, as further explained.

Different guiding principles were then grouped as drivers of the 3 distinct mechanisms of change [68].

A total of 3 guiding principles were linked with engagement as a mechanism of change (ie, personalization, tailoring, and real-world feel). Personalization entails the offer of personalized content; tailoring encompasses the adaptation of the content to potential needs, interests, personality, context, or other individual factors; and the real-world feel emphasizes to the researcher or responsible organization the need to increase the intervention credibility.

In total, 5 guiding principles under the auspice of role model form the second grouping (ie, social facilitation, social learning, social role, social comparison, and normative influence). Social facilitation postulates that participants are more likely to perform a target behavior if they are aware that another person is performing the behavior along with them. Social learning describes that a participant will be more motivated to perform a target behavior if they are able to observe others performing the behavior. Social role suggests that participants will be more likely to perform a target behavior if an intervention adopts a social role. Social comparison describes that participants will have a greater motivation to perform a target behavior if they can compare their performance with that of others. Normative influence describes that an intervention can provide peer pressure to increase the likelihood that a participant will adopt a target behavior.

The 3 remaining guiding principles were linked with the mechanism of change of self-efficacy: self-monitoring, co-operation, and recognition. Self-monitoring postulates that an intervention that allows a participant to keep track of their own performance or status supports the participant to achieve their goals. Co-operation describes that an intervention can motivate participants to adopt a target behavior by leveraging humans’ natural drive to co-operate. Recognition indicates that, by offering public recognition for a participant, an intervention can increase the likelihood that the participant will adopt the target behavior.

The 3 mechanisms of change hypothesized in this model were engagement, role model, and self-efficacy; these are described in the following sections.

There are different types of engagement (eg, active or passive), which are accompanied by a heterogeneity of overarching definitions and measurements [69]. The challenge of engagement has been described as one of the main issues with digital interventions [70]. Typically, initial levels of enthusiasm are only maintained for a period after which diminution of use ensues, especially when interventions are used in the real world [71]. This may be a problem given that most interventions are designed to become effective over a specific duration. In health care, motivations for compliance can be driven by perceived usefulness and achieving goals; these can facilitate the creation of habits [72].

The behavioral activation theory states that, without specific training, activation consists primarily of the scheduling of pleasant activities [73]. This was key for this intervention. An important objective was that participants were motivated by the intervention and considered it relevant to improving their condition and social context. This study aimed to have the patients engaged with the intervention, interacting with a real person—a community volunteer—who would be available to communicate with them and support them.

A key underlying theory for this intervention is the weak ties theory, which postulates the importance of less strong ties such as acquaintances to have access to new information outside the core network [74]. A series of principles related to social support have been proposed in the Oinas-Kukkonen and Harjumaa [68] framework (ie, social learning, social comparison, and social facilitation, which involve some form of connection with others through the technology). The principles of social learning and social comparison require that the person be aware of what the other person is doing and their progress, potentially achieved by mutual sharing of achievements or witnessing them.

The intervention aimed to provide patients with access to another person (ie, a volunteer) who, through communication, could act as a role model. Theories on role model use have advocated that it can be a way of motivating people to perform novel behaviors and inspire them to set ambitious goals [75].

Although the role of volunteers has been described as to be there or do things, this view may overlap with the conceptualizations of taking a more passive or active role [14]. In this intervention, it was envisioned that a volunteer would take a somewhat active role realized through being available to communicate with the patient; participating in mutual encouragement to engage in social and physical activities; and aiming to improve their self-esteem, social contacts, and physical activity. The rationale was to motivate participants to choose and communicate through methods that suited them while remaining in contact with their usual friends, thus linking the intervention to their social world.

Typically, users of digital interventions must feel motivated and confident to independently use them. Therefore, self-determination theory [76,77] is particularly relevant to understanding how users respond to this intervention and how it differs for patients and volunteers.

Therefore, the intervention components were organized under 3 objectives relevant to the constructs of self-determination theory. This postulates that intrinsic motivation to engage in health behavior change will be enhanced by (1) supporting users’ need for autonomy and feeling self-directed; (2) increasing users’ sense of competence, control, and confidence; and (3) enhancing users’ perceived relatedness or support from the intervention.

Having a chronic mental illness (eg, psychosis) presents several challenges that create a barrier to self-efficacy and affect life quality. Self-esteem has been linked to an intrapersonal influence on self-efficacy, suggesting that increasing self-esteem may subsequently increase self-efficacy [78].

In this study, there were several decisions made to design the intervention with the objective of improving self-efficacy. This entailed the provision of 1 volunteer for each patient with whom they were matched and with whom they could communicate remotely. This allowed them to help their match and to be helped, thus improving their self-esteem and quality of life. Having an attachment to their match could enable a normative influence, co-operation, and recognition. Furthermore, to raise awareness of their physical activity, participants were also encouraged by the researchers to use an app, Accupedo, to check their step count, a further means to enhance their self-efficacy [79,80].

Figure 3 depicts the Phone Pal logic model together with the contextual factors and anticipated outcomes. The model embraces the different components that were selected, the guiding principles promoted, the hypothesized mechanisms of change, and how these concepts are linked to outcomes. This logic model describes the main theoretical pathways connecting the components, principles, mechanisms, and outcomes of this intervention. In reality, the various intervention components may operate on multiple principles and, therefore, multiple mechanisms. An explanation follows for the rationale of why particular components and objectives were included in this intervention.

The Phone Pal logic model encompasses the contextual factors on which the intervention has been based: (1) patients’ social isolation, low self-esteem, low quality of life, sedentary lifestyle, and nonsecure attachments; (2) smartphone features (ie, wide availability, portability, and ability to offer digital and remote modes of communication); and (3) community volunteers who provide human contact with others, establishing informal communication about daily life, and offer support, being a role model to patients.

In total, 2 types of outcomes were hypothesized for both patients and volunteers. These were classified according to the time frame (ie, short-term and long-term outcomes).

For patients, potential short-term outcomes were to improve their self-esteem, social contacts, and physical activity. In the longer term, it was felt that additional outcomes could be attained (ie, improving their quality of life, attachment, social comparison, and symptom alleviation).

For volunteers, the short-term outcome goals were to witness a change in their attitudes, with a decrease in their social stigma toward people with mental illness, and to increase their physical activity. Further outcomes could be achieved over a longer timescale (eg, improvement in quality of life and social comparison).

For both patients and volunteers, it was hypothesized that, through a higher level of engagement, they would become more activated and possibly increase their physical activity.

The logic model of the planned intervention conceptualized the matching of 1 patient with 1 volunteer to communicate with each other over a smartphone. The core elements were (1) the remote medium for delivery of the intervention, (2) being in contact with another person, and (3) the use of a smartphone.

Core components are intrinsically linked with the intervention basis; on a higher level are components that, although less fundamental than the core ones, are still part of the operationalized and envisioned intervention.

The main components describing how the intervention should be and the methods that informed them are summarized in Table 4. The methods and decisions made to operationalize the intervention are summarized in Table 5.

For pragmatic reasons, a period of 3 months was chosen in line with several publications reporting interventions with a duration of 12 weeks [54,81-86]. A review identified that the level of commitment should be for a minimum of 4 hours per month [44].

The idea of the communication being tailored by each participant is deemed important to create a persuasive system [68]. Therefore, in this intervention, it was decided to leave it up to each pair to decide which communication modalities they would like to use.

In young adults with a first psychotic episode, it was noted that they preferred a combination of technologies through which to receive mental health care (eg, SMS text message, video, and audio). Among these options, SMS text messages were preferred [87].

Surveyed patients [22] proposed that contacts should be open-ended and weekly and favored SMS text messages followed by email, Skype, Facebook, and audio calls. The modality audio calls, which was not given as a separate option in the survey, was named by different patients when selecting the choice other. The methods of communication initially selected for this intervention (ie, audio calls, video calls, SMS text messages, Facebook messages, WhatsApp messages, and emails) were determined according to the results of this survey and the literature. However, Facebook messages were removed following the instructions of the REC.

It was decided to encourage participants to communicate with each other at least once per week. The participants were informed that one of the study aims was to assess how much or how little each pairing decided to communicate. The focus group study findings indicated that people varied in their attitudes toward specific communication methods. In line with the person-based approach, it was decided not to be prescriptive in this regard.

A pragmatic decision was made to give the participants full access to and choice of all possible intervention components (ie, they could choose from sending SMS text messages, WhatsApp messages, or emails or making audio and video calls). The participants could opt to interact with their match through a particular tool, and their paired match could then choose to reciprocate through the same tool or select another modality. Relinquishing control of communication modality use has been supported by previous research [88]. The lead author encouraged volunteers to take the initiative to call patients in the first instance.

The expert advisory group supported the choice of communication method as all the tools were commonly accessible on any smartphone.

While developing the intervention, the lead author was prescriptive in the duration and frequency of the communication but not in the communication modalities. In the process of operationalizing the intervention, the lead author embraced the challenge of how much or how little guidance to provide when researching a psychosocial intervention. In terms of guidance for intervention adherence and use, there can be tension between supporting autonomy while still providing clear guidance on how participants can best change their behavior. Some studies reported concerns that offering too many options can be overwhelming [89]. Bestowing complete control can result in lower intervention use than tunnelling core intervention content [90]. Some researchers suggest that tunneling is less overwhelming than allowing patients free choice [68], and it is often used in mental health interventions [91]. However, a tunnelled approach [68] in which participants are led sequentially through intervention content, usually in a predefined order or based on a needs assessment, might not always be suitable and was deemed as too imposing. In fact, the person-based approach recommends that, in general, digital interventions should aim to promote user autonomy and offer choices where possible [29].

In this study, communication was between 2 humans and mediated through technology—the smartphone.

The focus groups provided insights into possible suggested conversation content for this intervention. The participants were instructed to establish informal communication about daily life.

The literature indicated that this study had the potential to generate negative feelings in the participants. The sources could be the wait, delay or expectation that their match would make contact, or difficulties encountered in dealing with the end of the study or the end of the relationship. Some organizations provide support throughout to volunteers face to face or by using technology while covering users from a wide age range [99]. Offered volunteer supervision has been described in the form of monthly multidisciplinary meetings, one-to-one supervision sessions, or telephone support [46,95-97,100]. In a previous study, when an intervention was remote, web-based support from a staff member and occasional telephone calls were found to be essential for participants to remain in the study and continue to use the digital intervention [101].

Therefore, it was decided that a study coordinator would be available for the participants, providing one-to-one access to support and supervision whenever the participants felt it was required in addition to proactively contacting all participants once per month. This would enable the provision of constant support to volunteers, who are community laypeople who could end up facing potentially unfamiliar situations that they may not know how to deal with (eg, managing patients’ behavior or difficulties with the relationship ending). Patients equally require support, although negative symptoms could make them less proactive in contacting the study coordinator. Therefore, the lead author carried an additional work phone. The number was given to all participants, and she could be reached at any time (ie, 24 hours a day and 7 days a week). The monthly contact would routinely be through an audio call; if the participants did not respond, a message would be sent asking for their availability. Although evidence suggests that reminders can improve engagement with digital interventions, there is as yet insufficient data to indicate what types of messages are most likely to promote adherence [102-104]. It also remains unclear how people will respond to these motivational messages. What one person sees as encouraging might demotivate another [105,106].

To improve the understanding of the interactions between patients and volunteers, a practical decision was made to monitor the communication between each pair.

Although various members of the expert advisory group initially raised ethical questions about privacy, most subsequently concurred with this choice given that it offered an additional way to ensure that the content was appropriate and did not raise any safeguarding concerns.

The option of creating a technical app for communication monitoring was initially explored. After seeking quotations from different companies, it was established that a purpose-built app would be too costly. An existing app was identified that monitored the content of written communication (SMS text messages, WhatsApp messages, and emails) and the frequency and duration of audio and video calls. Through discussions with the expert advisory group, this app was purchased and used.

In line with the overall aim of the intervention to facilitate improvements in the participants’ mental and physical health by establishing informal communication about daily life and encouraging healthy lifestyles [107], physical activity changes were assessed in 2 ways. First, an existing app to monitor step count was installed on the smartphones. Using this, the participants could check their step count and gain awareness of their physical activity. In addition, to address potential issues of participants not always carrying the smartphone, the International Physical Activity Questionnaire [108] was administered both at the beginning and end of the study.