research-article

Open access

Between Rhetoric and Reality: Real-world Barriers to Uptake and Early Engagement in Digital Mental Health Interventions

Authors:

Gavin DohertyAuthors Info & Claims

ACM Transactions on Computer-Human Interaction, Volume 31, Issue 2

Article No.: 27, Pages 1 - 59

https://doi.org/10.1145/3635472

Published: 05 February 2024 Publication History

PDF eReader

Abstract

Digital mental health interventions (DMHIs) have potential to provide effective and accessible care to entire populations, but low client uptake and engagement are significant problems. Few prior studies explore the lived experiences of non-engagers, because reaching this population is inherently difficult. We present an observational inquiry into the barriers to sign-up and early use of a DMHI, along with reasons for initial interest in the DMHI. We collected 205 online questionnaire responses and 20 interviews from self-referring participants across four healthcare ecosystems in the UK and US. Questionnaire results revealed that uncertainty about DMHI usefulness and usability were the main barriers to uptake, whereas forgetting about it, not finding time for it and not finding it useful were the main barriers to early engagement. Participants reported multiple reasons for considering the DMHI, reflecting the contextual, subjective nature of mental health. Our thematic analysis generated themes around (1) the need for human connection, (2) the impact of self-stigma on help-seeking, (3) the lack of knowledge around DMHIs and psychological therapy, (4) the desire for personally relevant care, and (5) the fluctuating, perennial nature of mental health. We discuss implications for DMHI design, implementation and future research, as well as transdisciplinary opportunities.

1 Introduction

1.1 Digital Mental Health Interventions

Mental health is a subject of significant global concern [134, 170]. It is also a contested and normative construct that can be understood differently across cultures and generations [25, 83, 119]. The prevailing medical or psychiatric model of mental health (where mental illness is viewed as an “illness like any other” [132]) dominates discourse, service provision and scientific research [25, 133, 170], however, there are other ways to understand mental health and distress. Some perspectives, for example, contextualize distress in terms of its social, political and environmental determinants, and thus highlight the complex value systems at play in both its framing and management [2, 4, 25, 37, 83, 142]. Due to the ubiquity of the medical model, mental distress has traditionally been addressed on an individual rather than a societal level, i.e., with psychological interventions aimed at improving a person's mood or behaviour [25, 37, 142].

Within this contested space, digital mental health interventions (DMHIs) have been developed to address the escalating care gap between the vast number of people in need of support and those receiving it [6, 117, 135]. DMHIs usually take the form of interactive, online programs where clients read or watch psychoeducational content and use tools to develop awareness and self-management skills [46]. Many DMHIs are based on Cognitive Behavioural Therapy (CBT), due to its structured format and the abundance of clinical data demonstrating its effectiveness in addressing common (clinically diagnosable) forms of distress, such as depression and anxiety [29, 174, 176]. CBT has also been critiqued, similarly to the medical model, for promoting a singular, neoliberal view of distress as contained within individuals, which can decontextualize and depoliticize emotional responses and sometimes amplify distress [40, 66, 129, 130, 159]. CBT remains a dominant therapeutic method, however, due to its accessibility and ability to relatively quickly bring people out of clinical ranges of distress, an effect that is easily quantifiable through standardized outcome measures and randomized controlled trials (RCTs) [66, 176]. The RCT evidence for digitally delivered CBT (accompanied by human guidance or support) is substantial [28, 77, 160, 174], and qualitative research has shown that DMHIs are often experienced positively by clients, who can find solace, support and empowerment via these self-guided tools [50, 54, 70, 120]. Combined with their cost-effectiveness and ability to reduce waiting lists and ease pressure on overburdened services [108, 135], DMHIs are becoming a core element of mainstream mental healthcare services in countries such as the UK and Australia [38, 135]. These interventions are being used by millions of people, yet one crucial factor which undermines their potential to alleviate distress is whether clients actually begin and continue to use them.

1.2 Uptake and Early Engagement

Terms like uptake and engagement are both widely used and disputed in the DMHI and broader Human-Computer Interaction (HCI) spheres [48, 167]; for the purposes of this study, we will use uptake to describe signing up or creating an account for an intervention and early engagement to refer to a second use of the intervention after the initial signup process. We focus specifically on early engagement because this is historically low with DMHIs; many clients disengage after their first use, despite having actively sought help [13, 30, 57, 103]. This early engagement issue is not exclusive to the digital realm, as high early dropout rates are also present in FTF (face-to-face) psychotherapy [32, 61], however, research indicates that clients could be twice as likely to drop out of a DMHI compared to other therapies [171]. Uptake rates are also generally low when it comes to DMHIs, [57, 171] and research indicates that uptake can vary widely between different groups of people [38]. Furthermore, uptake and engagement rates are lower in real-world contexts than the structured research settings typically used [19, 30, 57, 91, 167]. Our study presents an observational exploration into the barriers that prevent people from signing up for and using DMHIs in organic settings, which is a key focus area within the field [147, 165, 167]. The quantitative aspect of our research aimed at understanding the most common barriers to and reasons for seeking help across this naturalistic cohort, while our qualitative analysis delved deeper into the various factors underpinning these experiences.

1.3 Reflexive Considerations

This research was undertaken in partnership with a private company (SilverCloud, acquired by Amwell in 2021) that creates DMHIs; one of the authors is a founder of this company and two others are employees. Hence, the study originated and was designed from a medically positioned stance, albeit still within the HCI discourse. During the qualitative analysis process however, the first author's reflexive engagement with the data necessitated a shift to a broader, more socially engaged perspective. An additional author with expertise in cultural psychology was brought in as a result of this repositioning. A detailed position statement for the first author can be found in the Methodology section and short position statements for all other authors can be found in Appendix A.1. As a team, we represent a variety of incongruent perspectives on mental health. We believe there is value in acknowledging and balancing these tensions, as they speak to wider challenges faced by the field. We have therefore taken a reflexive, pluralist approach with this article [17, 25, 33, 56].

2 Background

2.1 Barriers to Uptake and Early Engagement

DMHIs circumvent some of the structural barriers that impede access to traditional forms of therapy, such as prohibitive costs and long waiting lists [108], however, new structural barriers include lack of reliable internet access, a personal technology device, and a basic level of computer literacy [108]. These structural barriers disproportionately affect those with lower education levels, lower socioeconomic status and those living in rural areas, factors that intersect with other personal characteristics such as ethnicity and age [131]. Across all therapy types, it has been widely identified that those in minoritized or marginalized groups face additional barriers to seeking help and receiving psychological care, both in terms of access and attitudes towards therapy [127, 131, 155]. Attitudinal barriers play a crucial role in uptake and engagement with therapy [6, 52, 148]; two of the most common attitudinal barriers reported are low perceived need for support and a preference to deal with the problem on one's own [6, 16, 109, 127, 151]. Being mediated through technology, DMHIs face additional attitudinal barriers, such as technology acceptance [113]. Attitudes are shaped by dominant cultural narratives about health and wellbeing and the social and economic systems we exist within [25, 27, 67], and thus intersect with gender, class and other social factors [79]. For example, the clinical literature consistently demonstrates that women are more likely than men to perceive a need for support, to seek help, and to stay engaged with DMHIs [6, 11, 19, 76, 139], a finding intricately linked with how gender is culturally formulated [5]. We will now explore some of these common barriers in more detail, along with the implementation factors that impact the delivery of DMHIs [108].

2.2 Perceived Need

Clinical research indicates that the primary reason clients seek out mental health support is to reduce psychological distress, i.e., to feel better [114, 139]. This can be underpinned by a biological, hedonic drive to alleviate pain, but equally by a social drive towards normative conceptions of the ideal healthy, happy self [1, 45, 162]. Medically framed research on the link between need and engagement has usually focused on objective need, or symptom severity, as assessed by standardized outcome measures; these studies have shown that the more severe a person's symptoms, the more likely they are to seek support in general [139] and stay engaged in a DMHI [11, 63, 103], despite DMHIs being recommended by the National Institute for Health and Care Excellence (NICE) for those with mild to moderate symptoms [116]. Conversely, other research has indicated that baseline symptoms do not predict adherence to DMHIs [76], or that only certain symptom measures predict dropout [53]. There is clearly a difference between objective need as evaluated by standardized measures and perceived or subjective understandings of need. Perceived need can be multifaceted and is deeply enmeshed with a person's life and circumstances, as research suggests that distress is as much a product of social, contextual factors as it is an individual emotional response [4, 83, 119]. In line with this, common reasons for seeking help cited by clients include reference to precipitating stressors, such as interpersonal or relationship problems, productivity in work or study [104, 114], and bereavement [86]. Additional stressors such as discrimination and inequality are a core aspect of why women, those who are less privileged and minoritized groups are more likely to experience emotional distress [5, 79, 107, 155]. Another area linked to perceived need and context is busyness; lack of free time is a commonly reported barrier to engagement with DMHIs [15, 19, 103, 127, 145, 171], which highlights both external pressures and internal evaluations around the importance and urgency assigned to the task of self-care [72, 162]. In overlooking these patterns of experiences, purely quantitative clinical research often decontextualizes and dehumanizes distress [119]. Understanding the complex interactions between DMHIs and the context of their use and non-use is a key area for consideration in HCI [14], thus, we aimed to qualitatively understand the lived experiences of non-engagers as situated within their distinct, personal settings.

2.3 Culture and Stigma

Lack of mental health literacy, i.e., knowledge about recognizing, managing and preventing mental distress [45], has been suggested as a factor influencing the pervasiveness of low perceived need for mental health support [4, 102]. Mental health literacy implies a singular truth that is devoid of cultural context, but understandings of mental health are culturally and historically constructed; culture influences the development of attitudes towards mental distress, perceptions of need and subsequent decisions around help-seeking [25, 27, 82]. Established ideologies and social norms shape our beliefs about what constitutes normal vs aberrant behaviour; diagnostic definitions are affected by and in turn also affect these judgements [39, 89]. Despite advancements since the deinstitutionalization of asylums, psychiatry's history of systematic oppression, colonialism and social control has left, as its legacy, the mental illness as societal threat paradigm [58, 119]. Thus, seeking support for mental distress is often attributed with negative characteristics such as helplessness and weakness [163]. Asking for help can be seen as an admittance of failure, an act that signifies the individual is now labelled “someone who needs psychological help” [35]; this effect can be more extreme for men, due to masculinity norms which promote strength and independence [55]. Research in western contexts has shown that mental health stigma is directly related to lower help-seeking in the general population [151], and global research shows that, of those who do express the need for mental health support, a preference to deal with the problem on one's own is frequently cited as the main reason for not commencing or engaging with interventions [6, 16, 52, 127, 148]. Even when clients reach the stage of being able to acknowledge the need for external help, there are clearly culturally mediated risks associated with obtaining this help that need to be assessed [148]. These risks can be more pertinent among minoritized groups, who have historically been mistreated by coercive healthcare systems and exposed to structural stigma and racism [75, 119].

Mental health stigma within individuals (as opposed to structural stigma) has been theorized as consisting of two linked constructs: public stigma, (i.e., prejudice or discrimination against others experiencing mental distress), and self-stigma (i.e., the internalization of public stigma) [10, 34]. Research suggests that these two constructs have different relationships with attitudes towards seeking professional help, with self-stigma playing a more adverse role due to its impact on self-esteem and self-efficacy [10, 109]. There have been calls for further research into the effects of self-stigma on help-seeking [69, 151] and little is known, in particular, about the impact of self-stigma on help-seeking via technology. Using a DMHI circumvents some of the barriers associated with public stigma, due to its anonymity and ease of access [126], but the relationship it has with self-stigma is less clear. We sought to understand more about the prevalence of public vs. self-stigma amongst non-engagers, how people make sense of stigma in relation to DMHI use and the association between this method of seeking help and internalized beliefs about mental health. The complexities inherent in the connection between culture, mental health paradigms, awareness of the need for support, access and engagement with care are intricate and under-researched [4, 79, 108] and therefore an area we unpicked further with this study.

2.4 Technology Acceptance

Attitudes towards DMHIs are often negative among those who have not used them; studies show that DMHIs are perceived as less helpful, credible and motivational than FTF therapy and that FTF therapy is often the preferred choice over digital interventions [7, 51, 111]. A common barrier linked to low uptake and engagement with DMHIs is the belief that they will not work or be useful [15, 19, 103]. Perceived usefulness is a concept explained within various theoretical models, such as the Health Information Technology Acceptance Model (HITAM) [80]. The HITAM proposes that the perceived threat of a particular illness, combined with the perceived usefulness and perceived ease of use of a piece of technology, affects attitudes, intentions and resulting health related behaviours [80]. Under the HITAM, the reliability of the DMHI and the self-efficacy experienced by the client in relation to the DMHI collectively impact perceived usefulness and perceived ease of use [80]. In terms of ease of use, technical difficulties have been reported as significant barriers to DMHI use [145, 150], and engagement is facilitated if clients are informed or trained about how to use the technology [19]. Perceived fit, or the relevance of content to the client's needs, its cultural appropriateness and whether the DMHI can be cersonaliz or personalized are also factors effecting perceived usefulness [19]. We explored how clients construe usefulness in relation to DMHIs, and what influences the development of these perceptions, as few studies to date have taken this approach.

2.5 Implementation Factors

The wider healthcare ecosystem of the DMHI and how it is implemented can also have a substantial impact on client uptake and early engagement [41, 62]. Self-referral pathways are often a core component of population-health approaches, despite the fact that clients referred to DMHIs by medical professionals have much greater odds of engaging than those who self-refer online [11, 76]. Little is known about how clients experience the self-referral process and why it leads to lower engagement. Engagement is also higher when interventions are guided by a human therapist, rather than being self-guided [19, 174], but again it is unclear how clients perceive these different support models. There is a dearth of research on systemic or implementation factors that might affect uptake and engagement with DMHIs [139], as well as a lack of holistic, qualitative exploration of personal responses to these interventions [81]. Our study therefore examined the experiences of those self-referring to a DMHI online, within a range of natural implementation settings (e.g., a health service, universities, and a not-for-profit organization). In comparing different implementation variables (e.g., the care pathways and support options available), we explored how a client's journey is situated within this ecosystem and how these often-hidden variables affect the lived experiences of DMHIs.

2.6 Studying Non-engagers

Much of the existing research on barriers to uptake and early engagement in DMHIs focuses on quantitative predictors of dropout, rather than exploring the lived experiences of clients who dropout early or do not engage [3, 11, 15, 53, 76, 150]. A core reason for this could be that non-engagers are, by their nature, difficult to recruit, and even if reached might not be able to fully understand or express their own reasons for not engaging, due to the complex social and emotional constructs (e.g., stigma) that surround help-seeking for mental health [10, 150]. Consequently, little is known about what these non-engagers want or need, what prevents them from engaging [167], or what role dissatisfaction plays in disengagement; research shows that many clients who drop out of therapy early do so because they feel better and no longer perceive the need for it [86, 158]. In HCI, non-use is usually understood through the frame of lagging adoption (i.e., non-users are potential clients who are just not using the technology yet), which implies that non-use is temporary and rooted in individual clients, and therefore a “problem to be solved” [149]. Satchell and Dourish explain that other forms of non-use can include active resistance (those who make a positive, intentional effort not to engage) and disenchantment (those reluctant to engage due to nostalgia for past realities) [149]. Keeping these pluralities of non-use in mind while exploring the perspectives of non-users can help us to broaden our understanding of non-use in the context of DMHIs. Calls have been made for HCI researchers to listen and attend to negative feelings at the sites of technology use, and to consider not only how we might improve a technology to better suit those lagging adoption, but also where, how and why that technology might not be suitable [14, 90].

2.7 The Current Study

The current study presents an exploratory, observational inquiry into the experiences of non-engagers of a DMHI in relation to uptake and early engagement. Our main aims were to understand the barriers that prevented people from signing up for the DMHI, and for those who did sign up, the factors that led to them not returning after this initial use. We were also interested in learning more about the perceived needs of this cohort and the reasons why they decided to investigate the DMHI (i.e., visited the self-referral website or signed up), to contextualize our findings. Due to the difficulties inherent in recruiting non-engagers, we aspired for a comprehensive understanding of this particular group. Hence, we targeted both common perspectives across the wider non-use population as well as more in-depth insights based on lived experiences. Our research questions were:

(1)

What prevents people from signing up for a DMHI or using it after signup?

(2)

Why do non-engagers show initial interest in the DMHI?

3 Methodology

3.1 Study Design

This was a cross-sectional, observational study using mixed methods. Data collection included a short online questionnaire, through which we recruited participants for semi-structured online interviews. Data collection for this study ran between December 2021 and April 2022. We used the consolidated criteria for reporting qualitative research (COREQ) checklist [169] for the qualitative component of the study. This study was approved by the UK's National Health Service (NHS) Health Research Authority (21/NW/0351) for the health service site in the study, and by the Research Ethics Committee of the Trinity College Dublin School of Computer Science and Statistics for the universities and not-for-profit organization sites in the study.

3.2 Intervention

SilverCloud by Amwell is a digital mental health platform, designed through HCI research [47] and subsequently validated through a series of RCTs [46, 135, 137]. SilverCloud offers over 20 programs, each aimed at a specific category of distress (depression, anxiety, stress, insomnia, etc.). It has delivered human-supported interventions to over a million users via more than 300 partner organizations in both the public health and private sectors and across different geographies [50, 135, 136]. The programs offered are most often based on CBT and typically include 6–8 modules of psychoeducational content, as well as tools like a journal, mood monitoring and quizzes. Programs are delivered to clients via their service (e.g., health service, employer, university), resulting in a variety of different ecosystems that can surround the platform (see Figure 1 and Appendix A.2 for screenshots of the SilverCloud platform).

Fig. 1.

3.2.1 Support Models.

SilverCloud can be delivered unsupported in self-guided mode, but it is most often delivered with human support in the form of weekly written or telephone feedback [46]. Supporters are usually early career stage clinicians, but trained volunteers have also demonstrated effectiveness at providing this non-judgmental encouragement and feedback [137]. Supporters and support models vary by service and support usually lasts for 6–8 weeks, after which clients have one year to continue using the platform alone.

3.2.2 Referral and Signup Flow.

Clients access the SilverCloud platform either via clinician referral (where a link to sign up is sent by email) or through the self-referral website. The self-referral website provides an overview of how the intervention works, explains the programs on offer and allows clients to self-select to sign up. Clients choose a program (e.g., Space from Depression, Space from Anxiety, Space from Stress) either on the self-referral website before they click to sign up, or during the signup flow. The signup flow differs between services, but usually involves clients clicking a link to sign up, entering their email address as a username, creating a password and clicking that they understand and agree to the privacy notices. The next screen presents a series of personal goals that the client can select from (e.g., Feel calmer, Be more active), followed by one or more sets of standardized measures (e.g., Patient Health Questionnaire-9 or PHQ-9) associated with the particular program the client selected. These again differ based on service and support type, and are used to track progress, monitor and escalate risk and assess intervention effectiveness across the service. Finally, the client is shown three illustrations that represent a mental model of the platform (program, tools, supporter) and is prompted to start the first module of their program immediately, to help them get the benefits of the program as soon as possible (see Appendix A.3 for screenshots of the SilverCloud self-referral website and signup process).

3.3 Setting

This study was conducted in collaboration with four SilverCloud services, three in the UK and one in the US, within the health service, not-for-profit and higher education sectors.

3.3.1 Service 1: UK Health Service.

The Berkshire Healthcare Foundation Trust is an NHS Improving Access to Psychological Therapies (IAPT) service, covering a population of 900,000. The IAPT program is a stepped care approach to treating people experiencing anxiety and depression, as recommended by NICE [115]. Step 2 interventions are offered to those experiencing mild or moderate symptoms and include low-intensity CBT-based interventions such as guided self-help, SilverCloud and group therapy, whereas step 3 services, for those with more severe symptoms, involve high-intensity FTF CBT and counselling interventions. Clients are monitored throughout their treatment with weekly clinical measures (e.g., PHQ-9 for depression) and stepped up or down accordingly. All clients offered SilverCloud are provided with support from trained psychological wellbeing practitioners.

3.3.2 Service 2: UK Not-for-profit.

The remit of the Ben charity (Motor and Allied Trades Benevolent Fund) is to support the health and wellbeing of those working in the automotive industry in the UK, and their families. Ben provides a number of support avenues including life coaching, financial advice and assistance, mental health assessments, FTF therapy and SilverCloud, as well as industry training (e.g., mental health first aid). The Ben service is most often accessed by clients via their employers, who provide access to Ben resources as part of Employee Assistance Programs. Support on SilverCloud is provided by Ben counsellors in an opt-in manner, whereby clients can self-select support at any stage of their journey.

3.3.3 Service 3: UK University.

Nottingham Trent University is one of the largest higher education institutions in the UK, with over 40,000 students and staff spanning five university sites. The university's wellbeing teams include Student Support Advisers, who offer general advice and signposting, the Counselling Service, which offers phone, video and FTF counselling and email support, and Pastoral support from chaplains. SilverCloud is provided without support (as self-help) to students and employees of the university, along with other recommended resources under a “wellbeing toolkit” offering.

3.3.4 Service 4: US University.

Colorado State University is a public research university with approximately 40,000 students and staff across three campuses. The university's Health Network provides an extensive range of mental health resources including group therapy, workshops, substance use screening and support, FTF counselling, and psychiatry services. SilverCloud is offered to students with support from the counselling team or as self-help, and only as self-help to staff of the university.

3.4 Participants

We focused only on people who accessed the DMHI via the SilverCloud self-referral website, because engagement is generally lower without the motivational structure of clinician referral. As we were interested in both uptake and early engagement, we decided to recruit two different groups of participants, (1) those who visited the SilverCloud self-referral website and were undecided about signing up for the intervention, and (2) those who signed up for the SilverCloud intervention and didn't return for at least 2 weeks after their initial use. To be included in the study, participants had to be over 18 years of age, have provided informed consent and have accessed SilverCloud via the self-referral website.

3.5 Recruitment

Those who were undecided about signing up for the intervention were recruited via static, unintrusive notifications on the self-referral website, which read “Undecided about signing up? Do you have 2 mins to take this short research questionnaire?”. Those who signed up for the intervention and didn't return were recruited via an automated email, sent from the SilverCloud platform at 2 weeks post signup. Both the notification and the email contained a link to the questionnaire, which was preceded by a participant information sheet and consent form. Interview participants were recruited via the questionnaire, through an option to enter their email address to be contacted for the interview; a separate interview information sheet and consent form were then emailed to participants. Interview participants received a $/£50 Perx Reward virtual gift card after the interview was conducted.

Due to the naturalistic nature of this study, we had to adjust our protocol to address arising issues. Firstly, due to a very low response rate from the questionnaire email, we sent a second email which directly invited participants to the interview and presented the questionnaire as an alternative participation option. This approach yielded a much higher recruitment rate for both the questionnaire and the interview. Secondly, many of the clients who were sent the email in fact had multiple accounts under the same email address. Therefore, despite the intentions of the study being clearly stated in the email and participant information sheet, we received responses from users who had already engaged with the platform. After interviewing three of these participants, we decided to send a qualifying email to check whether participants were using the platform, reiterating that we wanted to talk to people who were not using the platform. We decided to keep these engaged participants in the corpus because they still provided insights on the barriers they experienced when signing up, and in fact many of the interview participants recruited via the self-referral website notification had signed up for the intervention at the time of the interview and were actively using it, despite being “unsure” about it initially. To manage this in the data analysis process we sorted the corpus into three groups based on their engagement status at time of interview: not signed up (NSU), signed up and not using (SUNU), and signed up and using (SUU). We did not have this same engagement status data for the questionnaire respondents.

3.6 Questionnaire Design

As this was an observational enquiry, and the cohort in question were non-engagers and thus likely to be difficult to recruit, we designed a short questionnaire (see Appendix A.4) which took only 1–2 minutes to complete, with the aim of increasing the response rate for this difficult to access group [93]. Multiple choice options were provided, as previous research indicates that participants often find it hard to identify or articulate the factors that affect their decisions around engagement with support [148, 150]. The questions and multiple-choice options for the two main research questions were developed based on the HITAM [80] and existing clinical literature on common barriers to uptake and engagement with digital mental health interventions [3, 6, 15, 19, 52, 53, 109, 127, 145]. One of the most common reasons clients give in the literature for not engaging with support is that they would “prefer to deal with it alone”. We felt that this statement could be underpinned by either self-stigma or public stigma [10, 34], so we decided to unpack it into two options: “I should be able to deal with my problems by myself” (self-stigma) and “I wouldn't want anyone to find out I was using it ” (public-stigma). See Appendix A.5 for further details on the multiple-choice options to the primary research question and their associated constructs from the literature. The order of choices was randomized to prevent order bias, participants could select more than one option, and “Other” options were also provided for those clients who wished to articulate their own reasons [148, 150]. The questionnaire began with five demographic questions (age, gender, ethnicity, education level, and employment) to understand participant characteristics and because placing simple, easy to answer questions at the start of a questionnaire has been shown to increase participant confidence and overall questionnaire completion [93].

3.7 Quantitative Data Analysis

Quantitative data were analyzed in IBM SPSS Statistics for PC, Version 28. Descriptive summary statistics were used to answer the main research questions and understand the characteristics of the data and the representativeness of the sample. Due to the non-standardized nature of the questionnaire and the multiple-choice options, no statistical tests were performed.

3.8 Interview Design

The semi-structured interview script was designed based on the same literature as the questionnaire and acted as a more detailed exploration into the same two core research questions. Through the interview, we explored how the client came to access the platform and the barriers they experienced, along with background information on their opinion about mental health and technology in general. Feedback was sought about the interview process from participants in the initial interviews, but there were no major issues or changes made to the protocol. Minor adjustments were made to the script after each interview, but the initial question “Can you tell me about the first time you heard about SilverCloud?” always remained the same, as it allowed participants to reflect back on the start of their journey. Sensitivity and care were taken not to push participants on their personal reasons for seeking help, if these were not forthcoming in answer to the question “Why did you decide to check out SilverCloud?”. The interviewer recognized her limits in terms of therapeutic training and the risk of creating a space where people might feel exposed. The researcher's role and experience were clearly stated at the start of the interview, and a risk protocol was created (including contact details for the relevant emergency services) and kept on hand throughout the interviews. Participants were also reminded of the value of critical feedback and the researcher's impartiality in terms of the design of the intervention, to stem social desirability bias.

3.9 Qualitative Data Analysis

Qualitative interview data were analyzed using reflexive thematic analysis (TA) [22–24], underpinned by a critical realist stance [124, 125]. Critical realism proposes that the world exists independent of human perception, but this objective world is of little importance to us in light of the more compelling (individually and socially) knowledge-constructed realities we live within [125]. This resonates with the inner and outer worlds of mental health – there are outwardly objective truths at play (people die by suicide), yet mental distress is primarily a reality felt individually, and deeply entwined with a world construed socially. Reflexive TA was chosen because it allowed for this critical realist foundation, but also because the complex nature of the topic at hand (decision-making around mental health support seeking) necessitated a method which supported creative oscillation between surface-level felt experience and underlying generative mechanisms. This approach offered movement between semantic and latent coding of the data, and between descriptive and interpretive reporting [22]. At different stages, the analysis moved along the inductive-deductive spectrum, never being fully deductive because there was no attempt to fit the data into existing theories, but never being fully inductive either, as theory was used to inform the study design [22]. Considering recent calls for greater detail in the reporting of reflexive TA practice in healthcare HCI [21], we have endeavoured to provide a comprehensive methodological account of our analytic process. In line with these best practices, and because all interviews and analysis of the data was conducted by the first author, we will hence switch to the first person narrative for an explanation of the first author's positionality and process.

3.9.1 Primary Researcher Position Statement.

I am a white, Irish, middle class, cisgender woman in my mid-thirties. I have an educational background in fine art and psychology, and experience working in many diverse creative roles. I have been working for the digital mental health company SilverCloud since 2017, first as a research assistant, then writing program content, and now doing a PhD while embedded in the UX team. In my role as a research assistant, I conducted diagnostic interviews for a large-scale clinical trial with the NHS, and analyzed qualitative data collected as part of this trial [70]. I have always been drawn to qualitative methods because I recognize and value the complexity of humans, and the fundamental subjectivity of all research undertaken by humans [89, 125, 173].

In terms of my own help-seeking for mental health, I have previously been to and for the duration of the data analysis process attended regular FTF psychotherapy sessions, with positive experiences on both occasions. I recognize the considerable privilege I hold in this regard, as private support of this kind was not an option for many of our participants. I have not used digital programs as a client, but I have an intimate understanding of their practical use from over three years of volunteering with the Irish charity Aware as an online supporter, helping adults experiencing distress to work through SilverCloud. In this sense I am not an inside researcher; I have extensive knowledge of SilverCloud and how it functions. Where I do consider myself, an inside researcher is in my preference for in-person interactions over digital. I would not be interested in using a digital intervention for my mental health, because I am generally reluctant to spend time using technology. In keeping with the reflexive TA method, I used my subjectivity as an analytic resource and kept a reflexive journal throughout the interview and data analysis process, to remain cognizant and questioning of how my beliefs were shaping this work (see Appendix A.6 for selected excerpts from this journal). I also used conversations with colleagues, friends, and my therapist as reflexive opportunities to untangle the fluctuating position I held in relation to this topic and my evolving relationship with the data.

3.9.2 The Analytic Process.

The interviews were transcribed by an external transcription company and I checked the recordings against the transcripts, updating the text where discrepancies were found. I also made field notes during each interview, which I typed up immediately afterwards. Phase 1 of the TA process involves familiarization with the data [22]. I printed the transcripts (with field notes attached), randomised their order and read through each one in detail, making notes in pencil on the transcripts and notes for each case in a TA notebook. I then consolidated these notes into a familiarization mind map (see Appendix A.7).

Coding (Phase 2) began by hand using the printed transcripts. Working with hardcopies allowed for freedom and flexibility in terms of working environment, and for a physiological, tactile experience of coding. Coded text was highlighted in the transcript and tagged with a number, which linked to a code label written down in my TA notebook. I used three different colours to highlight the text, two for the main research questions (barriers and reasons for seeking help) and the third for miscellaneous codes which were of interest. After eight transcripts were coded in this way there were 219 codes and checking back and forth in my TA notebook for existing codes became unwieldy, so I typed up all the codes into a Microsoft Word document and organised them under headings, to make finding codes easier. I was conscious of the danger in this premature grouping of codes as it segregated the ideas into topics, but I was confident that I could amalgamate the codes again in subsequent rounds of coding. This new code list was printed and used for the rest of the first round of coding, with new codes being written in by hand (see Appendix A.8 for image of first round coding process). The handwritten codes were typed up into the Word document at the end of this round; there were 362 codes at this stage. Between the first and second rounds of coding I took a three week break from the analysis.

The second round of coding was performed using QSR International's NVivo 12 software, because I am familiar with it from previous work and find it useful as a means of organizing and working with large numbers of codes. The transcripts were ordered differently to the first round of coding (see Appendix A.9 for ordering based on the coding round). The paper transcripts were used to check the first-round codes, the corresponding code was then searched for using the digital Word document and the transcript was coded in Nvivo. Codes were placed into two sections reflecting the two main research questions. Adjustments were made to the coding throughout this process; codes were combined if similar, split if they covered more than one meaning, left out if they were not relevant enough to the research questions, and new codes were created. The third round of coding was brief and involved double checking over transcripts to make sure nothing was missed. At the end of the third round, there were 173 codes. The final stage in the coding process involved systematically examining each code and its associated quotes, making sure the code label captured the essence of the meaning contained within the group of quotes, and removing or moving quotes that did not fit. The final list of codes came to 142. During this time (and throughout the theming process) I was attending regular TA discussion groups with other researchers, where we talked about methodological issues, we were having with our various TA analyses. These sessions were vital to the development of my understanding of TA and provided a much-needed outlet for what can often be an overwhelming process.

The generation of initial themes (Phase 3) began with a printed list of the codes, again to allow for creativity and a tangible connection to the data. Each code was cut out on a separate small piece of paper (see Appendix A.10 for image of candidate theme creation). Candidate themes were ideated upon in a creative process of playing with the codes and moving them about to create different groupings. In developing and reviewing the themes (Phase 4), I drew thematic maps to work through the layers of meaning within potential themes, to draw connections between them and to process questions I had. For the most part, the more semantic/descriptive themes persisted through the different theming rounds, whereas the more latent/interpretive themes took longer to work through and changed more drastically as the theming developed. It was around this time that I began engaging with socially situated mental health literature [4, 20, 25, 39, 67, 83, 133, 138, 142], because the data and the meaning I was making of it did not fit with a medical framing of distress. Once I felt comfortable with a cohesive list and map of themes, I began Phase 5, which involved refining, defining and naming themes. I wrote up theme definition paragraphs and spent some time adjusting the theme names to make sure they reflected the particular meaning and direction of each theme. It was only upon writing up the results section and choosing quotes for each theme (Phase 6) that I realized the boundaries between my themes were somewhat blurred, I had been trying to cover too much with each theme. To address this, I returned to Phase 5 and wrote up a one sentence tagline for each theme, which succinctly highlighted the main focus and thus brought clarity and defined the edges of the themes. The themes were further refined in consultation with the other authors.

4 Results

4.1 Questionnaire Results

There were 205 responses to the online questionnaire, 162 from participants who visited the self-referral website and accessed the questionnaire via the notification (notification group) and 43 who were SilverCloud users and accessed the questionnaire via the email (email group). A total of 93 participants (45.3%) came from the two university services, 31.2% from the health service and 23.4% from the not-for-profit service. Due largely to the university services, many participants were aged between 18–24 (40%), aside from this there was a balanced spread of age groups represented in the sample. The majority of participants were also female, white, and educated to college or university level, as is typical of service users [31, 139]; see Table 1 for questionnaire participant characteristics. In terms of the employment question, we used a “select all that apply” response option to reflect the varied circumstances of each participant, and 37 participants (18.2%) selected more than one option, including 3.4% who were parenting young children or on maternity or paternity leave, and 2% who were caring for the sick or elderly. Of those who only selected one option, the majority were employed 30–50 hours per week (40.5%) or full-time students (19.5%). See Appendix A.11 for full table of employment characteristics.

Table 1.

		All (N = 205)	Notification (N = 162)	Email (N = 43)
		n (%)	n (%)	n (%)
Service	UK Health service	64 (31.2)	51 (31.5)	13 (30.2)
	UK University	62 (30.2)	40 (24.7)	22 (51.2)
	UK Not-for-profit	48 (23.4)	43 (26.5)	5 (11.6)
	US University	31 (15.1)	28 (17.3)	3 (7.0)
Age	18-24	82 (40.0)	65 (40.1)	17 (39.5)
	25-34	46 (22.4)	34 (21.0)	12 (27.9)
	35-44	32 (15.6)	23 (14.2)	9 (20.9)
	45-54	26 (12.7)	24 (14.8)	2 (4.7)
	55-64	15 (7.3)	12 (7.4)	3 (7.0)
	65+	4 (2.0)	4 (2.5)	-
Gender	Female	155 (75.6)	124 (76.5)	31 (72.1)
	Male	46 (22.4)	36 (22.2)	10 (23.3)
	Non-binary	3 (1.5)	2 (1.3)	1 (2.3)
	Prefer not to say	1 (0.5)	-	1 (2.3)
Ethnicity	White	161 (78.5)	126 (77.8)	35 (81.4)
	Asian	12 (5.9)	10 (6.2)	2 (4.7)
	Black	11 (5.4)	9 (5.6)	2 (4.7)
	Mixed	11 (5.4)	9 (5.6)	2 (4.7)
	Indian	4 (2.0)	4 (2.5)	-
	Latino/Hispanic	4 (2.0)	3 (1.9)	1 (2.3)
	Native American /Alaskan native	1 (0.5)	-	1 (2.3)
	South Pacific Islander	1 (0.5)	1 (0.6)	-
Education	College/University	132 (64.4)	105 (64.8)	27 (62.8)
	High School/Secondary	44 (21.5)	37 (22.8)	7 (16.3)
	Postgraduate masters/Doctorate	28 (13.7)	19 (11.7)	9 (20.9)
	Primary	1 (0.5)	1 (0.6)	-

Table 1. Questionnaire Participant Characteristics

Responses to the two main research questions were “select all that apply”, so the distribution of answers was accumulative, i.e., for the question “Why did you decide to check out SilverCloud?”, more than half of participants (131/205, 63.9%) selected more than one multiple choice option. This was also true of the question “Why are you unsure about signing up for SilverCloud? ”, where 59.6% of participants (118/198) selected more than one option. This indicates that the reasons people have both for seeking help via the DMHI and for being uncertain about it, are manifold.

Regarding the first question around reasons for considering using the DMHI (see Table 2 for full results), the most prevalent response was “to feel better” (54.6%), followed closely by “because my mental health is important to me” (51.2%). Many participants wanted to deal with circumstantial or external stressors, and interestingly 60 participants (29.3%) decided to investigate the DMHI because it was recommended to them. Of the 11 participants (5.4%) who selected the “other” option and entered free text, responses centered around specific stressful life events or situations (e.g., “To cope with an unbearable workload and work environment”), alleviating distress, easy access and self-care (e.g., “to give myself better support”).

Table 2.

Multiple choice options, n (% of cases)	(N = 205)	(N = 162)	(N = 43)
	All	Notification	Email
To feel better	112 (54.6)	93 (57.4)	19 (44.2)
Because my mental health is important to me	105 (51.2)	88 (54.3)	17 (39.5)
To improve my work, study or home life	78 (38.0)	64 (39.5)	14 (32.6)
To cope with a stressful life event	75 (36.6)	61 (37.7)	14 (32.6)
It was recommended to me	60 (29.3)	44 (27.2)	16 (37.2)
To improve my relationships	46 (22.4)	40 (24.7)	6 (14.0)
I was just curious	39 (19.0)	30 (18.5)	9 (20.9)
To help someone else with their mental health	13 (6.3)	11 (6.8)	2 (4.7)
Other	11 (5.4)	8 (4.9)	3 (7.0)
Total	539 (262.9)	439 (271)	100 (232.7)

Table 2. Responses to the Question “Why did you Decide to Check out SilverCoud?”

As the primary research question (around what prevented participants from engaging with the DMHI) was presented slightly differently to the email and notification groups, the results will be displayed separately for each group. For the notification group, the question was “Why are you unsure about signing up for SilverCloud?”. Of this group, 155/162 participants completed this question (see Table 3 for full results). The most prevalent response to this question was “I'm not sure how useful it will be”, selected by 59.4% of participants. The two most widely selected responses following on from this centered around perceived ease of use, “I'm not sure how to use it” (33.5%) and “I can't decide which program to choose” (32.9%). The option based on self-stigma (“I should be able to deal with my problems by myself”) was selected by 27.1% of the non-user group. Free text responses to the “other” option were varied and included reasons related to stigma and apprehension about seeking help (e.g., “I'm worried someone will say I'm making it up”), not having enough time to use the DMHI or information about it (e.g., “I don't know enough about what it entails”), and having more suitable options elsewhere.

Table 3.

Multiple choice options	n	% of cases	% of total responses
I'm not sure how useful it will be	92	59.4%	24.8%
I'm not sure how to use it	52	33.5%	14.0%
I can't decide which program to choose	51	32.9%	13.7%
I should be able to deal with my problems by myself	42	27.1%	11.3%
I'm not sure if I will be able to fit it into my life	29	18.7%	7.8%
I need more support than this	28	18.1%	7.5%
I'm not sure if I need it	27	17.4%	7.3%
I wouldn't want anyone to find out I was using it	27	17.4%	7.3%
Other	10	6.5%	2.7%
I'm worried my data won't be secure	8	5.2%	2.2%
My mental health isn't important to me	4	2.6%	1.1%
My mental health isn't a priority for me right now	1	0.6%	0.3%
Total	371	239.4%	100%

Table 3. Responses from Notification Group (N = 155) to the Question “Why are you unsure about signing up for SilverCloud?”

For the email group, participants were first asked “Do you plan on logging back in to SilverCloud?”. In response to this question 46.5% of participants (20/43) answered “Yes”, 41.9% (18/43) selected “Undecided”, and 11.6% (5/43) said “No”. Different follow-up questions were asked and different multiple-choice options were presented depending on which answer participants chose. For the “Yes” response group, the follow up question was “What has been preventing you so far?” (see Table 4 for “Yes” group results). The most widely selected response among this group was “I forgot about it” (45%, 9/20), followed by “I'm finding it hard to fit it into my life” (35%, 7/20). Responses to the “other” option were related to complications with switching programs, difficulties in prioritizing self-care (e.g., “I always put myself last”) and issues with public stigma (e.g., “There's no accountability because I don't want people at work to know I am using it”).

Table 4.

Multiple choice options	n	% of cases	% of total responses
I forgot about it	9	45.0%	26.5%
I'm finding it hard to fit it into my life	7	35.0%	20.6%
I can't decide which program to choose	4	20.0%	11.8%
I'm not sure how to use it	3	15.0%	8.8%
Other	3	15.0%	8.8%
My mental health isn't a priority for me right now	2	10.0%	5.9%
I should be able to deal with my problems by myself	2	10.0%	5.9%
I haven't found it useful so far	1	5.0%	2.9%
I've been feeling better	1	5.0%	2.9%
I wouldn't want anyone to find out I was using it	1	5.0%	2.9%
I'm worried my data won't be secure	1	5.0%	2.9%
Total	34	170.0%	100.0%

Table 4. Responses from the Group who Answered “Yes” (N = 20) to “Do you plan on Logging Back in to SilverCloud?” to the Follow up Question “What has been Preventing you so far?”

The follow-up questions for the “No” response group was “Why not?” and for the “Undecided” response group the question was “What is affecting your decision?” (see Table 5 for “No” and “Undecided” group results). The multiple-choice options for both of these questions were the same. Of the “No” response group, 80% of participants (4/5) selected “I haven't found it useful so far”, while 60% (3/5) selected “I need more support than this”. In the “Undecided” group 44.4% (8/18) chose the option “I'm not sure if I can fit it into my life”, 27.8% (5/18) selected “I'm not sure how to use it” and 22.2% (4/18) chose “I need more support than this”. Free text responses from the “other” option concerned feeling constrained by program choice (e.g., “Felt restricted by signing up to one particular area, although there would have been more than one relevant for me at that time”), not finding the DMHI useful and not needing it because help was found elsewhere.

Table 5.

Multiple choice options	n	% of cases	% of total responses	n	% of cases	% of total responses
	No response group (N = 5)			Undecided response group (N = 18)
I'm not sure if I can fit it into my life	0	-	-	8	44.4%	24.2%
I'm not sure how to use it	2	40.0%	15.4%	5	27.8%	15.2%
I haven't found it useful so far	4	80.0%	30.8%	3	16.7%	9.1%
I need more support than this	3	60.0%	23.1%	4	22.2%	12.1%
I wouldn't want anyone to find out I was using it	1	20.0%	7.7%	3	16.7%	9.1%
I can't decide which program to choose	0			4	22.2%	12.1%
I'm worried my data won't be secure	1	20.0%	7.7%	2	11.1%	6.1%
Other	2	40.0%	15.4%	1	5.6%	3.0%
I should be able to deal with my problems by myself	0	-	-	2	11.1%	6.1%
I got what I needed from it already	0	-	-	1	5.6%	3.0%
My mental health isn't a priority for me right now	0	-	-	0	-	-
My mental health isn't important to me	0	-	-	0	-	-
I don't need it	0	-	-	0	-	-
Total	13	260.0%	100.0%	33	183.3%	100.0%

Table 5. Responses from the Group who Answered “No” and “Undecided” to “Do you Plan on Logging back in to SilverCloud?” to the Follow up Questions “Why not?” and “What is Affecting your Decision?” Respectively

Across all of the questions, there did not appear to be any trends between demographic groups and selected responses, however, we cannot be certain whether any significant differences between groups existed because statistical tests were not performed.

4.2 Interview Participants

A total of 20 participants were interviewed, 13 recruited via the self-referral website notification and 7 via email. At the time of interview, 5 participants had not signed up for the intervention (NSU), 7 had signed up and were not using the intervention (SUNU), and 8 had signed up and were actively using the intervention (SUU). Despite half of the services involved being universities, participants were quite evenly distributed in terms of age, however, there was again a skew towards more white women in the sample; see Table 6 for interview participant characteristics.

Table 6.

		n (%)
Service	Health Service	8 (40)
	UK University	5 (25)
	Not-for-profit	5 (25)
	US University	2 (10)
Age	18–24	7 (35)
	25–34	5 (25)
	35–44	4 (20)
	45–54	3 (15)
	55–64	1 (5)
Gender	Female	14 (70)
	Male	6 (30)
Ethnicity	White	14 (70)
	Asian	1 (5)
	Black	1 (5)
	Mixed	2 (10)
	Latino	2 (10)
Education	College/University	11 (55)
	High School/Secondary	3 (15)
	Postgraduate masters/Doctorate	6 (30)
Employment	Employed (30 – 50 hrs per week)	5 (25)
	Employed (<30 hrs per week) & Student (full-time)	4 (20)
	Student (full-time)	2 (10)
	Unemployed	2 (10)
	Student (part-time)	1 (5)
	Employed (<30 hrs per week)	1 (5)
	Employed (<30 hrs per week) & student (part-time)	1 (5)
	Employed (30–50 hrs per week) & student (full-time)	1 (5)
	Employed (<30 hrs per week) & parenting young children	1 (5)
	Employed (30–50 hrs per week) & parenting young children	1 (5)
	Employed (30–50 hrs per week) & student (full-time) & parenting young children	1 (5)

Table 6. Interview Participant Characteristics (N = 20)

4.3 Qualitative Analysis

Five themes and two subthemes were identified through the reflexive thematic analysis, which fall at different temporal positions along the timeline of engaging with mental healthcare (see Figure 2 for thematic map). The first theme, Humans need humans, covers the idea that human connection is a large proportion of what people are looking for when they seek psychological help. Needing help means I'm weak or I've failed centers on the deep emotional, self-stigmatized beliefs that can underpin reluctance to seek help, and even stifle awareness of the need for help. The third theme, Getting the right help is confusing contains two subthemes; What is “help”? deals with the lack of understanding, education and guidance around therapy in general and how to choose appropriate care, while What the hell are DMHIs? Centers more specifically on the dearth of knowledge around DMHIs and how this impacts expectations about these relatively novel interventions. The fifth theme, “One size fits all” fits few references the need for choice and personal relevance in terms of care options and content; and finally Accounting for changing needs highlights the diverse, intersecting trajectories of mental health goals and motivation and how they are not adequately catered for by existing care systems. Quotes are presented with the participant number, gender, service type and engagement status group (NSU: not signed up, SUNU: signed up and not using, and SUU: signed up and using), to situate each participant's words within their particular context. See Appendix A.12 for theme and code tables.

Fig. 2.

4.3.1 Theme 1: Humans Need Humans.

A fundamental part of what many participants in this study desired, in terms of therapeutic care, was human connection. This was either because human interaction was perceived as a crucial component of the therapeutic healing process, or simply because participants wanted an antidote to feeling alone. The importance of feeling heard and understood by another human was reported widely across the data, and many participants felt that this was not something that could be translated to the digital realm. As participant 7 here describes, it is the relational quality of human-to-human contact that is key,

“I think if you're talking to a person, you feel more interacted, I guess…maybe you feel more of a connection, you really feel like someone's listening. Obviously, you can tell when someone, you're talking face to face with someone you can tell if they're listening, if they're interested in what you're saying. Through a computer I guess you don't get that.” (P7, male, not-for-profit, SUNU).

The positive, healing relationship between therapist and client is a large component of the effectiveness of FTF therapies [85], and clearly something that participants in this study felt was missing from the DMHI. There were, in fact, a small number of participants who expressed that using a computer to improve their mental health further enhanced their feelings of seclusion and disconnection, even in services where human support was provided as part of the DMHI. One aspect of this preference for human interaction over digital was a desire to offload and externalize pent up worries or thoughts, “I sort of feel like I need to actually talk to someone and maybe just vent about it or to have someone to listen to me” (P8, female, university, SUNU). For this participant, who was struggling with multiple interacting forms of distress, it was being able to express her complex concerns that mattered most. The role of the other person in this interaction, therefore, was simply to facilitate a space for expression. As participant 12 notes,

“I think having someone to hear you, to be heard, to actually listen to your problem, is, is quite important to influence changes not only thinking, reflecting on your behaviour but to influence any changes in mental health.” (P12, female, health service, SUNU).

The reflective power opened up by being listened to was, for this participant, a core mechanism expected to bring about cognitive and behavioural change. Therapy has informally been termed the “talking cure”, and this name underpins common perceptions and expectations of the therapeutic process [118]. The relational function of talking was mentioned by participants as a key therapeutic process that was integral to change. For example, participant 10 here describes the art of conversation as an externalisation whereby thoughts are taken out of the inner realm of the mind and made concrete,

“having an actual conversation with somebody, it felt a bit more real and you know, like I had to maybe think about it for more than two seconds, trying to articulate what I needed to say, rather than having it in front of me you know, choosing which option” (P10, female, health service, SUNU).

The inherent difficulty in verbalising thoughts (“had to think about it”, “trying to articulate”) was recognised by this participant as a positive and integral aspect of processing distressing cognitions. The alternative, working through content digitally and alone, was seen as easier but markedly less effective for this individual. Aside from the beliefs participants held about the therapeutic value of conversing with another human, the need for human connection also resonated through the data as a solution to the common circumstance of social isolation and loneliness. There were different reasons across the data for this isolation, from recent emigration, “I move to the UK very recently like six- five months ago. So, I feel like I'm alien and I don't see people much” (P16, female, health service, SUU), to the working conditions that have become mainstream as a result of the COVID-19 pandemic,

“I actually work at home but I've got a log cabin at the bottom of the garden so I'm completely detached from even the family in the house. So, I don't- unless I choose to, I don't get a lot of interaction with other people” (P1, male, not-for-profit, NSU).

For some participants, however, there was no precipitating event that brought about their loneliness, they simply did not have any close friends or family members. A small number reported that they didn't have a single person in their life who they felt they could talk to about their mental health. Even when participants did have support networks, people were sometimes seen as too busy to be supportive,

“people say you know they'll check-in on someone and whatever else. But people don't, people are wrapped up in whatever they have going on. They're not bad people for it, it's just life and I think everyone gets wrapped up in their day and they just forget” (P13, female, not-for-profit, SUU).

Participant 13 had been through a recent bereavement and noted that the support she received immediately following her loss waned consistently over time. The social cue of loss triggers supportive instincts in those around us [94], but when it comes to distress that is not linked to an overt external event, or continues long after an event, these supportive social cues are less obvious. From a sociopolitical perspective, theorists have noted that modern capitalist economic and social systems tend to isolate us and consume our attention, obscuring our capacity to recognise and care for each other [100]. This participant wanted to be reached out to, rather than proactively asking for help, because of the stigma associated with not being able to cope alone (discussed further in the next theme). A number of participants noted that apprehension about burdening others with their problems prevented them from opening up or seeking help. There is evidently a complicated relationship between interpersonal bonds and mental health, as conversely participants noted that the influence of other people contributed to them seeking help. For some this came in the form of a recommendation from a trusted person or the testimonial recommendations on the self-referral website as the main influences that lead them to use the DMHI. For participant 19, his girlfriend played a key role not only in his decision to reach out for help, but also in his recognition of the need for help in the first place. He notes here the difficulty that people might experience if they don't have this same level of support,

“But a lot of people I suppose wouldn't be in that position where they want to address that issue themselves. So, it might take someone else to then, but if they haven't got anyone then it's a bit of an awkward situation for them I suppose.” (P19, male, health service, SUU).

Overall, we can see that human connection appears to be a crucial component of the need for help, yet the interpersonal weight of mental health makes attaining this support difficult, as we shall see in the following theme.

4.3.2 Theme 2: Needing Help Means I'm Weak or I've Failed.

Many participants in the study revealed that they felt ashamed of their mental distress and feared the judgement of others, feelings which had previously prevented them from seeking help or even talking about their mental health. This emotional response was sometimes rooted in experiences of public stigma, such as that which participant 8 encountered,

“I've been in therapy before and even when people don't mean to be rude or show that there is a stigma around it, I've had people say like, “Oh, yeah, I could never, like I'm not that crazy” or something like that, which is shocking.” (P8, female, university, SUNU).

This type of negative social assessment can lead to shame and the internalisation of public attitudes as self-stigma [34]. Even nowadays, when public stigma and discrimination have become far less socially acceptable (participant 8 was shocked by this overt display of stigma), self-stigma persists. Stigma was mentioned by participants from a variety of diverse backgrounds and across each of the included services, yet it appeared to be slightly more prevalent among participants from the not-for-profit service, and some of the male participants noted that they believed mental health stigma was worse for men than women. In addition, a small number of participants described how their culture impacted their feelings towards seeking help. Participant 5, who was from Zimbabwe, described mental health as not being part of her culture at all, but this is perhaps a reflection of how distress and suffering can be conceptualised and expressed differently across cultures [25],

“when people think mental health, they start thinking, okay there's some sort of witchcraft involved in it, or you're crazy basically… people tend to not want to talk about it because they are afraid of what other people might think like, okay, if I say that I'm feeling depressed right now, people will just be like, “ah you're just trying to get attention, you're not depressed”. So, you tend to just keep it in.” (P5, female, university, NSU).

When participant 5 moved to the UK, she was inundated with mental health and wellbeing information; she explained that it was a relief to admit to herself that she might need help, and to have these resources available to her. Other participants communicated recognition of the general decline of mental health stigma in recent years, the increase in public debate around the subject, and the widespread availability of information and self-care resources. Yet there was a sense that underneath this surface-level awareness, we are still a long way from living in a society that fully grasps the many reasons how and why distress affects us and what can be done to prevent or alleviate it. As participant 18 notes,

“there is still a disconnect between the rhetoric and the reality. So, you know, we talk a good game in society around [mental health] and its importance. I think there's a bit of a lag in terms of the actions or the real understanding of it” (P18, male, not-for-profit, SUU).

The hidden, stigmatised emotional beliefs surrounding mental health could be a key factor in the perpetuation of self-stigma, as many participants in this study described how scared they had been to admit that they had a problem and needed help. To cross behavioural help-seeking barriers, people must first cross the more significant psychological barrier between being what they perceived as “healthy” and what they perceive as “someone in need of help” [35, 36]. Deciding what constitutes “bad enough” to cross this barrier is a delicate balancing act, where comparisons are constantly made against other people,

“I guess, also, this thought of, oh, I'm not doing that bad, to actually go to someone. We think sometimes that oh there are a lot of people that are doing worse than me so, I'm not going to waste their time, like the therapists’ time.” (P8, female, university, SUNU).

Other participants, along with participant 8, held the perception that a person needs to be at a certain, socially mediated, level of severity in terms of distress to seek and receive help. Identifying the boundaries of this level can be problematic because it requires not only social comparisons between the self and others, but introspective evaluations between current experience and previous states [152]. Participant 6 was so unsure about the validity of her own distress, that she believed she might have been imagining it, “sometimes I almost convince myself that I've made up any issue I've got in my head” (P6, female, university, SUNU). Many participants in this study declared that they did not know whether they needed help or not, or what was normal vs. “help-worthy” in terms of emotional distress, a finding that is consistent with previous qualitative research [18].

Looking more closely at the social aspect of crossing this needing help barrier, some participants remarked that they felt pressure to cope with the difficult emotions they were experiencing, which usually involved keeping them in or not expressing them. Participant 13 describes how she remembers her mother coping with bereavement, “I don't know how she coped, but I don't ever remember her crying, I don't ever remember her falling apart and I shouldn't. I should be able to do what she did.” (P13, female, not-for-profit, SUU). Sadness, along with other dysphoric states and emotions such as anxiety and anger, hold the position in many cultures around the world of being unwanted, and therefore broadly classified as unhealthy [94]. Expressing these emotions, for participant 13, was equated with “falling apart” and seen as the opposite of coping. Coping therefore is an ideal state, something that one should be able to do. The term healthism has been used to refer to the common but harmful assumption that individuals have sole responsibility over their own health, and therefore upholding one's status of good health (e.g., through coping) is a moralized duty [37]. Coping as a moral, socially promoted act becomes complicated when we observe the opinions of those towards people who do cope by concealing their distress, as participant 1 describes,

“I know I've got a friend who has always been very sort of happy, very sort of outward and we found out a few months ago that actually, he'd been suffering from depression, which none of our circle of friends actually knew. But I know that some of the other people I heard talking about it seemed to say ‘well you know, why would he suffer from depression, he's always happy’.” (P1, male, not-for-profit, NSU).

A paradox exists here whereby neither expressing your distress (“falling apart”) nor concealing it (people think you are lying or making it up) are socially tolerated. The dominant social norm or assumption, therefore, is that you shouldn't be experiencing these emotions in the first place, to do so means there is something wrong with you (you have an illness), or you're either imagining it or you failed to look after yourself adequately (it is your fault) [138]. This has been termed the “mad or bad, brain or blame” dichotomy, and it is deeply enmeshed with the medical model of distress [73]. Rather than situating distress in the context of external stressors, medical framings can individualize suffering and enhance self-stigma [73, 132, 133, 146]. In order to seek help, people must face their own internal beliefs about what it means for them to need this help, a finding that was seen widely across the data in this study.

4.3.3 Theme 3.1: Getting the Right Help is Confusing: What is “Help”?.

In addition to uncertainty around when to seek help, many participants in the study also communicated that they were unsure what type of help they needed, and in a more practical sense what getting help entailed. In terms of awareness of mental health support, participants stated that their understanding came from either direct personal experience, specialised education (e.g., studying psychology or working in the health sector), or television programs, movies and talking to friends and family. The pervasiveness of the medical model means that many people assume the healing process in relation to distress is akin to physical health, for example, a small number of participants mentioned being “fixed” or “cured” of their distress. Participant 13 was worried about her not-for-profit service telling her they couldn't help when she sought out the DMHI,

“if they say they can't help me, then I'm really going to have to go to the doctor's, it's just someone else telling me that. But actually, if they turn around and say, “I can help” then I'm not “not fixable”. I know that sounds stupid” (P13, female, not-for-profit, SUU).

Participant 13 was conscious that her worry about being “not fixable” was perhaps a social faux pas, as were others in the study. These participants believed there was something inappropriate about suggesting that emotional distress can be cured, but for want of an alternative, they were unsure how else to think about it. In line with this confusion about the goals or outcomes of therapy, disconcertion was experienced around whether the DMHI was the right fit for certain individuals and their specific circumstances. For these participants, it appeared that their satisfaction with any solution presented to them, no matter how simple, depended primarily on its suitability, as participant 4 explains,

“I want to be sure that someone who's qualified can listen and then say, ‘okay, yes, you know that's just everyone goes through this from time to time, it's fine’. Or you know, ‘okay I think you should try this program’ or even just you know, even if they said you know, ‘do some breathing exercises’.” (P4, male, health service, NSU).

The most important aspect of getting help for this participant was having a trusted professional hear his story, assess whether he needed help and if he did, guide him towards the right course of action. Several other participants also mentioned wanting advice from qualified experts, or specifically to be triaged, noting that this aspect of getting help was missing from the DMHI self-referral process they experienced. This could be a factor impacting the higher DMHI engagement rates seen amongst those clients referred by a doctor or other professional [11, 76]. DMHIs accessed via self-referral, perhaps because of this lack of triage that can be discouraging for some, are consequently not seen as “help” in the same way that more traditional forms are. Participant 1 held the opinion that his interest in the DMHI was something different to the more meaningful act of seeking help, “I mean I'm not actually sort of seeking physical help myself but certainly I've used a multitude of different Apps over the years for different things” (P1, male, not-for-profit, NSU). Thus, in using or investigating a DMHI, one can remain comfortably on the healthy side of the needing help barrier. As participant 9 explains, the digital route to help allows for a less overwhelming experience of seeking support,

“I'd rather not go and cry to my doctor and tell them that I need help, I'd rather go on the computer… I think that's what's nice as well actually about this online thing because it's not so much of a big deal, is it? Just turn your computer on and doing it that way, it's not quite as difficult to get yourself to do it.” (P9, female, not-for-profit, SUNU).

This positioning of DMHIs “on the fence”, so to speak, between help and not help, could contribute to the notion that they are a good first step, an easy route into the daunting world of help, as was mentioned by a number of participants. Participant 17 had a traumatic past experience with distress, so the idea of “going back there”, signified by seeking “treatment”, was a frightening prospect for her,

“everything to do with mental health is negative to me because I don't want it, I don't like it. And I needed to turn it into a positive, that's something you don't really need but it's to help you so, that's how I've seen it.” (P17, female, health service, SUU).

This participant was unsure about signing up for the DMHI because of these fears, but the ambiguity of the DMHI meant that she was able to reframe it as something she “didn't really need”, which allowed her to circumvent the barrier of admitting she needed help. Seeing the DMHI as a coping strategy or preventative measure is precisely why she went on to engage with it, but other participants had different reactions to this aspect of the DMHI. Participant 12 also saw the DMHI as preventative, but for her this was a problem because she was seeking a more reactive solution, “It is good for someone with stable mental health, not for people who are looking [for help]” (P12, female, health service, SUNU). Other participants also believed that digital solutions were not as effective or as serious as FTF therapies, precisely because they were not perceived as real help,

“if I'm at the stage where I'm like sitting in the shower like properly sobbing because I don't know what, what the hell is happening with my life, I can't- like sitting on a computer just clicking on boxes is not what I- it's not going to help me at that point.” (P10, female, health service, SUNU).

Participant 10 desired immediate relief from her overwhelming experience of distress, which she did not find with the digital solution presented to her. She mentions getting to a point where she needed something more than the DMHI could provide, an area that we will discuss further in theme 4. Crossing the needing help barrier was not a concern for this participant, but for those who do experience this stigma, DMHIs appear to hold a unique position in terms of being able to help without being considered real help. Underpinning this is a deep uncertainty about what it means to get help for mental distress, a confusion that we found was exacerbated by the lack of knowledge and understanding of digital solutions more broadly.

4.3.4 Theme 3.2: Getting the Right Help is Confusing: What the Hell are DMHIs?.

One of the most prevalent codes across the entire dataset involved participants not being sure what the DMHI was. Most participants had not heard of or used a digital solution previously, and therefore had little frame of reference to go on, “I didn't have any expectations because I never used anything like that before” (P3, female, university, NSU). Even when participants had prior experience with the platform (e.g., one participant was a nurse who had previously recommended it to her patients) there were still many layers of uncertainties and unanswered questions. Participant 1 remarked that “until you sign up you can't really see the sort of increased detail as to what exactly it is” (P1, male, not-for-profit, NSU), which points to a lack of practical explanation of the intervention on the self-referral website and in other materials provided to clients in anticipation of sign up. Many other participants also declared that there was not enough information given in these preliminary materials (e.g., service websites, outreach emails, and wellbeing resource landing pages) or that the information provided oversold the intervention, leading to unrealistically high expectations which were not met, and therefore left participants feeling disappointed. Recent HCI research has called for greater transparency and honesty in client-facing marketing of mental health technologies [162], because implying that psychological change is easy cultivates expectations that are unattainable [157]. In line with this, a major request from participants was for more information to be provided before clients are expected to enter their details and commit to signing up for the DMHI. Participant 14 was enthusiastic about the digital option but had doubts about it due to previous negative experiences with FTF CBT, as she explains,

“I just wanted to see I guess what it was actually offering, go into more depth about it… I like structure so it would have been nice to have like- I don't know, like a course detail thing of like what it all goes through.” (P14, female, health service, SUU).

For those participants who did decide to sign up, as in participant 14’s case, the desire for more guidance and information continued into experiences of the platform itself. A number of participants professed that they didn't know how to use the platform, this confusion often stemming from not understanding the fundamental nature of how interventions like this operate. For example, participant 15 expected someone to contact her to initiate the start of her journey with the DMHI, an expectation possibly based on traditional forms of therapy, which are interactional rather than self-directed,

“I didn't actually know how to start the program, if that makes sense? Because I kind of sort of logged in, did the things and then I was kind of waiting for somebody- Waiting for the, well, when do I start? How do I start it? So, I'd signed up for a few days and it was like, oh okay, so I've got to kind of go back in. So there was a slight gap.” (P15, female, health service, SUU).

The autonomous nature of DMHIs was not something that this participant fully comprehended before she signed up, so she was left waiting for direction. There was also uncertainty around the sign-up process, as some participants believed that the self-referral website was part of the intervention platform itself; without a mental model for the DMHI, it can also be difficult for people to grasp the systems and processes surrounding the DMHI. These unknowns left many participants feeling frustrated and lost. Participant 12’s exasperation was palpable,

“I didn't know did I, did I click on the right page? Where, or did I already sign up? What is it? There was nothing, nothing really to guide me, to explain what that Silver Cloud is, why clouds are silver. I was- I couldn't understand really what it is.” (P12, female, health service, SUNU).

This excerpt introduces another area that perplexed some participants, relating to the ecosystem of the intervention itself and the relationship between the participant's service and this external company “SilverCloud” who were delivering the intervention. Previous research found that endorsement by a trusted service and the presence of familiar logos indicated credibility when people were seeking mental health information online [126], and for many participants in our study this was found to be true, e.g., “I thought oh, I might as well give it a go because I saw that it was… done with the NHS so, I knew it would be reliable” (P19, male, health service, SUU). However, other participants found the presence of branding outside of their “trusted service” to be a point of ambiguity and concern. Participant 4 expressed distrust in private sector companies and queried the usefulness of the intervention considering what he perceived to be the potentially immoral values and goals of the corporation behind it,

“anything that's for-profit healthcare… has that implicit you know incentive for the company, assuming they are not that scrupulous, to you know, effectively try and keep their customers on board rather than providing them with help.” (P4, male, health service, NSU).

When faced with this dearth of knowledge about DMHIs, participants tended to forge expectations based on the closest comparable thing they had knowledge about, which was often wellbeing apps and other similar digital products they had used before, “my expectation was that an online tool would be similar to you know, other stuff out there” (P18, male, not-for-profit, SUU). For example, some participants expected to have to pay for the intervention, and that the payment structure might be like consumer facing apps (e.g., restricted access to content based on subscription level). In addition, because most of these apps are delivered as self-help, many participants did not expect human support from the DMHI; some participants from the supported services in the study didn't even realize there was a human support aspect. For the most part, attitudes towards other widely available mental health and wellbeing apps were negative; participant 9 dismissed the apps she had used previously as frivolous and unhelpful,

“I think the ones I'd seen were more just like, here's a daily mantra for you and write your dream and [laughs] I don't know, like weird things like that. Because I did actually download some of them ages ago and they were rubbish, I never used them” (P9, female, not-for-profit, SUNU).

Pre-determined negative attitudes towards the broad category of “wellbeing technologies”, means that DMHIs can become tarred with the same brush, as was found in this study. Many participants had low expectations for how useful the intervention was going to be, as did participant 8, who put off engagement with the DMHI because of this pessimism, “I feel like I'm delaying it just because I'm like, oh yeah it's not going to work” (P8, female, university, SUNU). Belief that DMHIs will not work is a common barrier to uptake and engagement found in the literature [15, 19, 103], but another explanation for these attitudes could be linked to the inherently standardised nature of digital solutions. For many participants, the usefulness of the DMHI was directly linked to its relevance to their personal situation and circumstances. Participant 6, who it should be noted was provided the DMHI without human support, questioned the effectiveness of a universally applicable digital intervention considering her specific circumstances and needs,

“How are they going to know exactly what I need help for if they're not asking me… I thought how can this really help, like it might be one of those things that they say is going to help and it might make you think it's going to help, like a bit placebo effect but it doesn't really help.” (P6, female, university, SUNU).

Not knowing how DMHIs function, what their underlying mechanisms are, or how they are tailored can lead to huge gaps in understanding, gaps which are often filled with inference. The notion of tailoring and personalisation of digital interventions will be explored in more detail in the following theme.

4.3.5 Theme 4: “One Size Fits all” Fits Few.

When asked what brought participants to the DMHI, they conveyed a range of different reasons for seeking help, involving one or a number of external stressors, such as recent bereavement, job loss, starting university, domestic violence, or moving country, “I'm an international student so I just wanted some aids to help me because it's a new environment” (P5, female, university, NSU). These stressors often precipitated help seeking, so understandably participants wanted to address these issues specifically with the help they sought. From a clinical standpoint, many of these issues would be classified as related to “adjustment disorder”, a diagnostic category that raises taxonomic dilemmas around the boundary between normality and pathology [164]. While the range of program options available through the DMHI was mentioned by several participants as an appealing aspect of this particular intervention, “it covers such a wide aspect, a wide range, a wide variety if you want, of issues that people can seek help with” (P7, male, not-for-profit, SUNU), many participants still felt that their particular problem was not covered or represented. There was an overall perception among most participants that a digital solution implies generic content, and that generic content is undesirable, as has been found in other recent HCI studies [8, 9]. Participant 12 found a disconnect between her subjective experience of her own distress and the nonspecific solution presented to her, “it's just I find it strange that we all have different problems and we all get treated with the same, the same generic videos” (P12, female, health service, SUNU). Frequently coupled with this aversion to a “one size fits all” approach was a preference for human interaction, due to its intrinsically personal nature. Participant 1 articulated the desire for a solution that could be adapted to him and his needs,

“I think where it would fall down is if I felt that it was too generic and therefore didn't suit me and therefore talking to someone in person is probably much more tuneable to personal circumstance.” (P1, male, not-for-profit, NSU).

Not understanding the role of the human supporter in the DMHI, or not knowing that there was a human support element at all, led many participants to underestimate the personalised nature of the intervention. However, even when people did understand the supporter role and were actively using the intervention with support, as in participant 16’s case, there was still a sense that the content of the DMHI was indistinguishable from the plethora of mental health information that is now widely available online,

“I've been reading I know these modules, they maybe they are helpful but it's still these things are, we can, we can find them outside like any other websites.” (P16, female, health service, SUU).

The global therapeutic and self-help industries have grown substantially in recent decades [138, 162], meaning that most people have been exposed to “self-help” style wellbeing advice in one form or another, often including techniques from the widely popularised CBT [115]. Some participants who had this existing knowledge and already used CBT techniques as part of their own self-care routine, found the provision of this advice through the DMHI to be simplistic and even patronising,

“Have you tried, I don't know, sitting down and having a cup of tea? Or just writing out your thoughts? And I'm like, yes, I have, I've thought about all that. I've done all of that. I keep a diary so I'm very mindful about how I'm feeling, when I'm feeling it… I almost felt it was a bit- I don't know if- I don't really know how to say it. Like I don't want to say like it was talking to a child, you know like explaining things.” (P10, female, health service, SUNU).

Being given these nonspecific recommendations was an extremely frustrating experience for participant 10, and it is clear that the DMHI was not an appropriate option for her at that time. For many participants in fact, the DMHI was simply not suitable. This was either because they were not comfortable using technology, “I'm just quite bad with technology, it takes me a while to figure it out…I'm just, I don't really know how to work it properly” (P8, female, university, SUNU), or they were experiencing types of distress for which CBT or the digital format was not suitable, “I guess the stuff I have kind of been struggling with it was- I didn't know that it was really going to provide the help that I felt I needed” (P2, male, university, NSU). That said, we can see from some of the engaged participants in the sample that the DMHI was an ideal solution for some, due to its adaptable, self-directed nature and relative convenience compared to FTF modalities. Even amongst the non-engaging participants, these beneficial attributes were regularly mentioned. For participant 13, a busy working mother with young children, the DMHI was a perfect fit,

“I can actually help myself rather than feeling like I've got to have a childminder, I've got to try and get somewhere, I've got to book an appointment, no one else is around at that time. So I don't think I had realised before actually looking into it that it was flexible enough that it helped us, but it's also not feeling like I've got to add an extra pressure into my day.” (P13, female, not-for-profit, SUU).

Choice and agency over support options surface here as essential mediating factors in client satisfaction. Giving clients transparent information and multiple options from which to choose can help them find a solution that meets their needs, and crucially, one that they are happy with. Some participants felt like they were pushed towards the DMHI because no other options were available to them, when their preference was to talk to a therapist, “I went again to the health centre and they didn't have enough staff, so they tried to direct me to the SilverCloud platform” (P3, female, university, NSU). Providing DMHIs as an alternative to FTF therapy can be problematic if people feel like they are being given something inferior and unsuitable for them, thus the framing of DMHIs is a key area for consideration. Human variability is so wide and complex that each different support option presented will inevitably suit a portion of the population, but no single option will suit everyone. Add to this the fluctuating nature of mental health and the different trajectories of distress that clients can experience, which we will explore in the next theme, and the picture becomes even more complex.

4.3.6 Theme 5: Accounting for Changing Needs.

Client needs or goals for mental health support can be conceptualised as immediate and hedonic (i.e., the reduction of pain and distress) or more eudemonic and related to long term wellbeing, the attainment of happiness and the prevention of future distress [45, 121, 162]. There is a complex relationship between these intersecting, fluctuating goals, which caused confusion for participants in terms of what they expected or wanted from the DMHI at different times. The connection between perceived need or level of distress and motivation was mentioned by several participants, for example, participant 6 noted that she tends to only work on her mental health when she is feeling low,

“I kind of go through like phases so, when I don't, when I'm not going through a phase of feeling low then I kind of stop looking for help. Which isn't good but that is part of the issue. So, I think I do need to just start doing it and force myself to keep doing it.” (P6, female, university, SUNU).

Participant 6 recognized the inherent issue with her mood-dependent help-seeking behaviour, but her instinct was to avoid the unpleasant prospect of working on her mental health when she was feeling well. She references “forcing” herself to work on the DMHI, while she would engage effortlessly when experiencing low mood. Alleviating distress feels good, whereas experiences of growth and meaning may not be as hedonically pleasant, and can in fact be subjectively uncomfortable [94, 153, 157]. A number of participants in this study referenced the challenging nature of working on their mental health and the difficulties they experienced in finding motivation, which has also been found in previous research [44, 70]. Under most health behaviour change models, including the HITAM, intentions are theorised as a core predictor of behaviour change [80]. Many participants in this study, who emphasised the importance of mental health and had signed up with intentions to use the DMHI, still found the motivation to use it varied and was hard to come by. This highlights the emotional, fluid nature of motivation, a factor which could underpin the intention-behaviour gap that has been found in relation to multiple areas including mental health [12, 88, 156].

For other clients, motivation was not an issue because their only goal was immediate distress relief. Participant 11 no longer felt the necessity to spend time and energy on her mental health because she was no longer experiencing distress, “I should give credit to my mental health but I'm just, just not giving any attention to it anymore because… I'm not in need of it.” (P11, female, university, SUNU). We cannot assume that clients seeking support want to build long term skills for future wellbeing or engage in sustained use of the technology. This potential conflict between the (fluctuating) goals of the client and the (usually static) goals of the intervention evokes debate on the power dynamics that exist between clients and the governments, employers, private corporations and institutions creating and delivering DMHIs [45, 96, 130]. The weight of responsibility is often directed towards individual clients to maintain their own wellbeing through self-care [37], yet some participants in this study indicated that there were too many competing demands on their time to find the space to work on the DMHI. Prioritizing use of the DMHI amidst work and childcare responsibilities was not an option for participant 9, “I kept thinking oh, I'll do it when I get to bed or you know, sit and have a bit of time and I just kept falling asleep and just not bothering” (P9, female, not-for-profit, SUNU), which speaks to the contextual nature of need and motivation, as influenced by external factors. In many instances, what is best for the wellbeing of the client is not to use the DMHI, particularly when this “self-care” work is perceived as burdensome. DMHIs do have the potential to provide more immediate support, due to their accessibility, however, for many participants the signup process itself was a barrier to attaining this momentary relief,

“unfortunately when your thoughts are racing and you're not feeling your best, the last thing you want to do is read and sit down and take things in. Because you don't take it in, you just want to go straight to it.” (P17, female, health service, SUU).

Participant 17 was experiencing heightened distress caused by the prospect of seeking help, which impeded her ability to process information and make decisions. Distress experienced while going through the signup process comes at a time when clients are often most in need of support. Several participants pointed out this gap in care provision between crisis services and preventative solutions, and there was a collective call for better “middle ground” support, i.e., immediate but not emergency care. Participant 14 gives a potential solution to this issue:

“maybe it would be good if you had like an option of like if you're feeling you know, hopeless and unmotivated like speak to the Crisis Team first. Like speaking to someone about it, just about your mental health in general would really help and then people who actually want help and are in the right mindset for help, SilverCloud would be really useful for.” (P14, female, health service, SUU).

This participant's idea involves options that depend on the client's attitude towards help, which is shaped by their culture and prevailing mental health narratives, e.g., healthism and the medical model [27, 37, 133]. Thus the “right mindset for help” is potentially one that assumes responsibility over and the ability to control distress. These narratives also determine how interventions are assessed and assigned, with diagnostic definitions and symptom measures being standard practice. For example, under NICE guidance, the stepped care model used by the health service in this study advocates that DMHIs are more suitable for those with mild to moderate “symptoms” [116], despite evidence suggesting those experiencing more severe symptoms still benefit [49, 77, 105]. In fact, as we saw above, fluctuating periods of severity could actually be motivational in terms of client engagement with DMHIs. What's more, clients who struggle to fit their unique experiences into these standardized symptom measures and rigid diagnostic categories could feel enhanced distress, as we found in this study,

“I will absolutely work myself up into a state where I can't get any work done for like one or two weeks at a time and then the problem will magically resolve itself and I'll be fine for like a month. So, it's not constant and if you do any of those questionnaires they're always like, ‘oh in the last two weeks you know, how many days roughly has this been affecting you?’ And it's like if I answer it honestly then there's never any kind of solution presented. It's always like, ‘oh you know, you've just- you're having a bad week, sod off, come back when you've got a problem’.” (P4, male, health service, NSU).

This participant was suffering from intermittent bouts of anxiety, that were extreme when he was experiencing them, but were not constant. The progression of distress experience over time involves multiple interacting trajectories, which are not accounted for in current diagnostic classification systems [20]. Mental health is hugely varied and contextual and thus needs multiple ways to be formulated and understood [73]. There was a sense from this data that the reality of mental health as a complex and subjective experience, with oscillating needs and shifting trajectories, was perceived by participants as not being adequately addressed by current care pathways and models.

5 Discussion

5.1 Overview

The primary aim of this study was to understand what prevents people who are interested in a DMHI from signing up for it or using it after they sign up. Our secondary aim was to explore why these non-engagers considered the DMHI in the first place, to understand the perceived need of our sample for mental health support. Our quantitative results showed that over 50% of participants were interested in the DMHI because they wanted to “feel better”, and over 50% because mental health was important to them. Similarly, when it came to the reasons why participants were unsure about signing up for the DMHI or why they were not engaging, very few stated that mental health was not important to them or was not a priority for them right now. These results indicate that participants in this sample perceived both a threat associated with emotional distress and the need for support, findings that align with prior studies [114, 139]. This need was found to be multifaceted and temporal, as over 60% of questionnaire participants selected more than one reason for checking out the DMHI, and interviewees described multiple stressors influencing their need for help (e.g., emigration, bereavement, job loss, domestic abuse) and how their needs changed over time (Theme 5: Accounting for changing needs), findings that again are consistent with previous literature [38, 104, 114].

Regarding the primary aim, perceived usefulness and perceived ease of use were prevalent issues; almost 60% of questionnaire participants who were unsure about signing up stated that this was because they were not sure how useful the DMHI would be for them. Considering our qualitative findings, we can see that this may be because of a lack of perceived fit, i.e., not believing that the DMHI would be relevant to their specific needs (Theme 4: “One size fits all” fits few), or not knowing whether the DMHI was the right type of support for them (Theme 3.1: Getting the right help is confusing: What is “help”?). It could also be down to not understanding what the DMHI entailed (Theme 3.2: What the hell are DMHIs?). Two other common factors preventing uptake (both also key barriers among the signed up but not engaged groups) were not being sure how to use the DMHI and not knowing which program to choose. This again speaks to the dearth of information and guidance around DMHIs as discussed in Theme 3.2 and could be related to the self-referred nature of the cohort in question, who did not have the scaffolding benefits of clinician referral. It also brings to mind issues inherent in the categorization of mental distress [20, 119], as discussed in Theme 5, and the importance of the suitability and relevance of the intervention to personal needs, as detailed in Theme 3.1 and Theme 4. The need for more information during the uptake process was also evident in the disparity between notification and email response rates; those on the self-referral website were actively seeking out information and perhaps saw the research as an opportunity for their questions to be answered.

Of those questionnaire participants who had signed up and were planning to return to the DMHI, forgetting about it was the most common factor preventing engagement. This implies that there could be a lower perceived need for immediate distress relief among this group, as we saw in Theme 5 how motivation can often be linked with distress rather than long term goals. There could also be lower perceived threat of emotional distress and its impacts, or the DMHI could lack salience due to skepticism about its effectiveness (Theme 3.2). Although, from the next most prevalent selection among this group (not being sure if they would be able to fit the DMHI into their lives), we can see that the participant's environment could play as much of a role as their internal perceptions. This selection was also common across the other participant groups, which suggests that participants’ lives were too busy to allow time for their mental health. This evokes debate on the wider contextual and social determinants of mental health [4, 25], and the issues inherent in the positioning of responsibility on the individual to address their own welfare, when they exist within structures that undermine this welfare (e.g., work cultures that propagate burnout [26]) [83, 107, 138, 162]. Questionnaire participants who did not plan to use the DMHI again mostly reported that this was because they had not found the DMHI useful so far or they needed more support than it could offer (i.e., a lack of perceived fit). This relates back to ideas around pluralities of non-use, and how not all non-engagers are simply lagging adoption; for some, the DMHI was simply not an appropriate form of support for them at that time (Theme 4).

We will discuss the implications of this research for the design and implementation of DMHIs, for further research, and consider some wider, transdisciplinary implications (see Table 7 for a summary of the main implications).

Table 7.

Implications for Design & Implementation	• Be transparent about what exactly is involved in the DMHI, rather than overselling it.
	• Consider the framing of DMHIs and the mental health narratives they might be propagating.
	• Emphasise the presence of human support in marketing or introductory content.
	• Endorse DMHIs with real-world testimonials and recommendations as well as clinical evidence.
	• Create adaptive care models that expand the spectrum of options between purely digital and purely FTF support (e.g., include human touchpoints at critical stages such as triage).
	• Position DMHIs on the ‘help’ side of the ‘help or not help’ fence by distinguishing them from wellbeing apps and including more human contact or leverage the ‘not help’ aspects by providing information about care options and mental health narratives, to help clients experiencing self-stigma to make sense of their own needs and goals.
	• Provide flexible, customisable pathways to address clients’ unique and changing needs.
	• Design DMHIs that normalize emotional distress in response to situational stressors, or specifically to support people through common life transitions or difficult circumstances.
Implications for Research	• When recruiting non-engagers, repeatedly highlight the study aims and the value of non-engagement experiences, as these are likely to be misunderstood.
	• Explore different forms of non-use in the context of broader sociotechnical and political trends.
	• Critically examine the success metrics of DMHIs and whose needs they are serving.
	• Reconsider non-engagement as a ‘problem to be solved’ by exploring more contextual, fluid and personally meaningful ways to assess the benefits of DMHIs.
Transdisciplinary Implications	• Address the loneliness epidemic by facilitating real-life interactions and community support, rather than creating individual technological solutions.
	• Explore alternative understandings of mental health that account for social norms and determinants, because the medical model can enhance self-stigma by individualizing distress and masking situational factors.
	• Consider a transdisciplinary approach to mental healthcare, combining knowledge and focus from both the medical and social sciences.

Table 7. Summary of the Main Implications of the Study

5.2 Implications for Design and Implementation

A central implication of this research is the need for greater transparency and sincerity in the framing and marketing of DMHIs, a need also called for by other interaction design and HCI research [157, 162, 175]. Rather than overselling DMHIs as solving all the client's mental healthcare needs or being experientially comparable to FTF therapy, providers should focus on clearly articulating what is involved in the DMHI in practical terms, and the fundamentally challenging nature of therapeutic change [157]. For example, despite the critique surrounding CBT, it provides practical techniques that can be highly beneficial in helping people deal with emotional distress [54, 66], and research has shown that individuals who viewed CBT as a learning process, rather than an opportunity to offload, were better able to learn skills and use them after the intervention [59]. This provision of information extends beyond the specific features of the intervention in question, to the type of help it provides and the associated model of mental health it promotes. While the majority of DMHIs available on the market might present a narrative in line with the medical model of mental health [73, 142, 174], designers cannot assume that potential clients are familiar with this narrative and how it shapes their perceptions. Disregarding core ontological elements such as this risks creating normative interventions that propagate singular stories about mental health, rather than giving people the knowledge to choose their own story [45, 97, 98, 124, 161]. It can also result in people internalizing responsibility for the environmental factors which might be causing their distress [25, 73]. In line with this, pluralities of knowledge should also be considered in terms of the “evidence” that can validate DMHIs for clients [33]. While RCT evidence is usually presented as the primary scientific endorsement for DMHIs, real-world client testimonials and recommendations from trusted persons could play a key role in guiding expectations and beliefs about intervention usefulness, as was mentioned in Theme 1: Humans need humans. Finally, in terms of transparency, if there is human support provided as part of the DMHI pathway (e.g., interactions with mental health professionals, trained volunteers, or peers), this should be emphasized strongly (but genuinely) in marketing or introductory content, as clients often do not expect human contact from digital care options.

In addition, we found that there is a significant learning curve involved in understanding and getting started with DMHIs, which can be intensified when clients are experiencing heightened distress brought on by the act of seeking help. An adaptive care model could incorporate the option for human touchpoints at critical stages in the client's help-seeking journey, such as during the referral process or in triage, to help these clients. Thinking more flexibly about mental health support could help us to expand the spectrum of care options that can exist between purely digital and purely FTF support, moving from regarding DMHIs as standalone products to thinking instead about technology-enabled services [106]. Furthermore, designers should consider on which side of the “help or not help” fence to situate their DMHI. To be seen as real or legitimate help, DMHIs need to distinguish themselves from wellbeing apps and include human contact, as this is where many of the core benefits of therapy can be found [85, 123, 128]. Alternatively, a positioning on the “not real help” side could help those clients experiencing self-stigma to safely make sense of their needs, learn about different narratives of mental health [73], understand the care options available to them, and consequently make more informed decisions [69]. In this sense, DMHIs that focus more on the provision of information could be a stepping stone to further help, providing a trusted resource on mental health amidst the overwhelming torrents of information available online.

This study also bears significance in movements towards greater personalisation of mental health technologies [8, 70, 143]. Our findings suggest that static, standardised solutions are not perceived by clients as sufficient to address the individual, shifting and contextual nature of emotional distress. Systems that offer multiple branching pathways could forefront client choice and agency by helping clients to identify and address their specific needs at diverse timepoints [102], as digital has the potential to help alleviate momentary distress as well as providing more future-oriented, skills-based learning [126]. In developing personalised interventions, designers should be mindful of the social and environmental factors that influence distress and the idiosyncratic complexity of human emotional responses [4, 67]. DMHIs could play a role in helping to normalize emotional distress in response to life stressors, along with enabling understandings of distress that can be both normal and help-worthy, rather than one or the other [132, 162]. Finally, as many of our participants were seeking support for specific situational stressors, DMHIs that address common developmental or life transitions (e.g., migration, bereavement) or external stressors (e.g., domestic violence, job loss) are an important area for future development [99, 166].

5.3 Implications for Research

The principal learning, we took from trying to recruit a non-engaging cohort was that many non-engagers do not understand or value their position as a non-engager in terms of research. We found that when it came to the interviews, individuals tended to think that they would have nothing useful to say about the intervention if they had not engaged with it. This was especially the case with our email recruitment arm; the first line of the email and the participant information sheet clearly stated that the study was targeting those who had not used the intervention, yet we still had participants ask if they needed to cancel their interview because they had not had time to use the intervention yet. A qualifying question in the consent form could have helped to filter out engaged clients, but due to these preconceptions, most non-engagers would not even get that far. General awareness raising about the value of non-engagement experiences to research could help to shift these perceptions, but when it comes to individual studies, we recommend strong emphasis and repeated highlighting of the study and recruitment aims.

That being said, some of the most important insights in our study came from currently engaged clients who had overcome initial periods of non-engagement. This research has thus highlighted the need for a broadening of our understanding of non-engagers as one homogenous group, to explore different forms of non-use and the evolution of engagement over time [113, 149]. Satchell and Dourish describe non-use as more than simply an absence of use, rather it is “active, meaningful, motivated, considered, structured, specific, nuanced, directed, and productive” [149]. In addition to researching those who are (or were) simply lagging adoption, investigating the people who actively resist or are disenchanted by DMHIs could facilitate a better understanding of where mental health technologies fit within broader sociotechnical trends, while considering the perspectives of people who are simply disinterested in DMHIs might allow for alternative views of “problems to be solved” like uptake and engagement [14, 16, 67, 90, 141, 149]. Addressing low uptake and engagement could go beyond a focus on individual clients and their (lack of) motivation, to explore the social, political, and economic dynamics of DMHIs and the power structures they are bound up in [45]. This shift in mindset could help us reconsider the goals of these interventions and whose interests they are serving [90]. For example, while personal recovery can be a meaningful goal for many of those who are suffering, disability justice and movements against the recovery focus of mental health support advocate that a sole focus on personal recovery at the level of service provision negates political responsibility for widespread distress [101, 159]. When considering the personal level, these movements also remind us that lives lived with disability and illness can, for many people, be lives lived meaningfully and well [42, 74, 101, 122, 172]. Thus, in evaluating the success of DMHIs in practice, instead of adhering to one-dimensional, normative ideals of health and wellbeing, we could look to the many cultural and personal variations of what it means to be and live well [25, 27, 95, 138]. Usage-based success metrics and standardized outcome measures, while important for large-scale service evaluation, do not adequately account for these unique client experiences and personal goals [86]. A pluralist approach to success in terms of DMHIs would be to account for broad sociopolitical agendas, personally meaningful objectives and the use of standardized outcome measures, thus making explicit the many layered goals of these interventions [33, 73, 138].

5.4 Transdisciplinary Implications

With this study, we have endeavoured to present a comprehensive picture of the barriers to DMHI uptake and early engagement, which means that many of our findings relate not only to DMHI use and non-use, but to the field of mental healthcare more broadly. In terms of our first theme (Humans need humans), recent research from the health field has recognised the pressing nature of the emerging loneliness epidemic [71, 92, 112], which has in part been exacerbated by the rapid evolution of technology and its facilitation of now familiar practices such as remote working, as well more direct disruption of traditional social networks [60, 64, 71, 154]. While technology can undoubtedly be used to connect people, it can equally further isolate, e.g., as was found with a recent HCI study around homesickness [78]. Developing digital interventions to address loneliness on an individual level (e.g., [168]), could be compared to the startling suggestion in a recent JAMA Psychiatry paper that the loneliness epidemic might be tackled pharmacologically [71]. Opportunities exist here for more creative and socially conscious thinking around the use of technology to address social disconnection [87], e.g., digital communication channels could be used to facilitate online community forums and real-world community interventions such as urban walk and talk groups [110] and governments could digitally disseminate interventions teaching support skills as public health initiatives, thus making the prevention of social disconnection a societal responsibility, rather than an individual one [138, 162].

Putting the use of technology aside, this research indicates the need for mental healthcare to move away from singular, individualising ontologies of health towards more balanced approaches that account for social norms and determinants [1, 27, 65, 67, 94, 139]. The “mad or bad, brain or blame” dichotomy, which we saw in Theme 2: Needing help means I'm weak or I've failed, can underpin self-stigma and thus inhibit help-seeking, yet it is rooted in the very medical model of distress that shapes most current thought around mental health [25, 67, 73]. Critique of the medical model is widespread, even within the field itself [73, 84] and often extends to the neoliberal or late capitalist economic system within which the model has flourished [43]. The confusion that we saw in Themes 2 and 3 around when to seek help, what is “normal” in terms of distress and what help is or should entail, could originate in the juxtaposition of this reductive, categorical model with the fluid, contextual nature of lived experiences of distress [132, 133]. One of our participants described a disconnect between the rhetoric around mental health and the felt reality of it; perhaps clients are caught in this gap, confused and unsure how to make sense of their experiences [25, 138].

Alternatives to the medical model, such as the Power Threat Meaning Framework (PTMF) [73, 133] offer more contextually and culturally sensitive approaches that focus on patterns of response [119, 140]. In line with our pluralist approach, we advocate not for one model over another, but for multiple perspectives to be considered [33]. A natural extension of this is to consider a more transdisciplinary approach to mental healthcare, which would combine knowledge and focus from both the medical and social sciences [25, 68, 144]. Rogers and Pilgrim, in discussing the discrepant value positions held around mental health, note that “currently there is no single platform from which to launch a campaign for improved “mental health literacy”, or to set out a single blueprint for a public mental health policy” [142]. How can we expect clients to understand mental health and the help-seeking process, when so much uncertainty surrounds it at the policy and service provision level? Thus, rather than debating the merits of either perspective, transdisciplinarity would allow us to harness the benefits of both sides, i.e., addressing the global mental health crisis at the crucial root level of societal and structural change, while simultaneously providing tangible and timely solutions for those who are suffering [37, 39, 83]. The emerging field of deep medicine advocates for a shift in focus to the ways in which systems interact to create health or illness, because we evolved as systems within systems [100]; systems thinking could be a core part of dealing with the complexities inherent in a transdisciplinary approach to distress [102].

5.5 Limitations

Due to the exploratory, real-world nature of this study, there was an inevitable self-selection response bias, which we noted particularly in relation to the interviews. The cohort in question put themselves forward to take part and were therefore, by their nature, more comfortable with and predisposed towards talking; this could affect the transferability of these findings. Across the study, there was a skew towards white, educated women, which is typical of service users within high-income countries [139]. In a global sense, we acknowledge the relative privilege of our participant pool; specific steps should be taken to understand the perspectives of more diverse groups of non-engagers in future studies, as their experiences are likely to be different [107]. Additionally, while we felt that multiple-choice options in the questionnaire were necessary to aid response rates for this hard to engage cohort [16], there was a risk of priming participants, and many of the multiple-choice options were influenced by our own positioning within a culture that prioritises medical understandings of distress. Finally, due to the naturalistic, observational nature of this study, a number of our interview participants were current active users of the intervention. Many of these participants had overcome periods of “non-engagement” however, and thus we felt their insights were relevant for this research.

5.6 Conclusion

In this article, we presented an observational, mixed-methods study on the barriers that prevent uptake and early engagement with DMHIs. We found that needs for mental health support are multifaceted, temporal, and contextually situated. Static, generalised interventions are therefore not always seen as sufficient to address these complex, shifting needs. The difficulties people experience in forming mental models of DMHIs was also reported on, and participants expressed the need for more transparent, honest information and guidance about what DMHIs are and how to use them. Furthermore, we recognise that while this type of digital support is well suited to the vast numbers of people who have and continue to use it with success, it is not suitable for many others. The need for customisable, flexible care models that incorporate human interaction at critical touchpoints, such as triage, was discussed. This study contributes to the limited literature on non-engagers, and we suggest that broadening definitions and metrics around non-use could help facilitate a deeper understanding of how the fluctuating goals of the individual intersect with those of the technology itself, the service or ecosystem surrounding the technology, and wider society. Finally, in order to fully understand the complexity of mental health and address the escalating global mental health crisis, we need to move away from reductionist approaches that focus solely on individual clients or users of technology. Allowing for society or humanity level perspectives and concentrating on the systems and structures that cause and perpetuate distress, could help us to take a more transdisciplinary, reflexive and ultimately effective approach to the use of technology in the mental health space.

A.1 Position Statements

Camille Nadal is a Postdoctoral Research Fellow at Trinity College Dublin where her research focuses on human-centered technology for health and mental health care. Her background is in Human-Computer Interaction and Computer Science. She is interested in research making a difference for vulnerable or less privileged populations and contributing towards better research practices in the field of Human-Computer Interaction. She is currently leading research in partnership with SilverCloud by Amwell, and has previously collaborated with the company during her doctoral research.

Sarah Robinson works as a Senior Postdoctoral Researcher with Lero - Science Foundation Ireland's Research Centre for Software, in University College Cork. She has a background in socio-cultural and community psychology, and Human-Computer Interaction. She is interested in cultural understandings of emotional distress, and how these understandings inform practice and experience.

Marcus Hanratty is a Lecturer In Product, Interaction, and Service Design in the National College of Art and Design, Dublin. His background is in Industrial design and Interaction design, with a strong focus on Human-Centered Design approaches. His research focuses on the role design and technology play in shaping people's behaviours, with a particular interest in design for behaviour change and design for health and wellbeing.

Angel Enrique works as a Senior Manager Digital Health Scientist at SilverCloud by Amwell. He holds a Psychology degree and PhD in Clinical Psychology from Jaume I University (Spain). His background is in the development and research of clinical and positive psychological interventions, face-to-face and internet delivered, among different psychological conditions, such as anxiety and depression disorders. He has been working in the digital health space for over a decade through a combination of academic and industry experience across Europe and US. In his current role he works at a strategic level, leveraging science to generate insights that derive in platform and service improvements. He is passionate about the continuous innovation and optimization of digital mental health products.

Gavin Doherty is a Professor in the School of Computer Science and Statistics at Trinity College Dublin, where he leads the Health Technology Design Group. He has a background in Computer Science and Human-Computer Interaction, and led the development of the SilverCloud digital health platform. The focus of his work in digital health has been on supporting and extending the reach of mental health professionals, with a strong interest in designing innovative and engaging systems which can be implemented in routine clinical environments.

A.2 Screenshots of the SilverCloud Platform

A.3 Screenshots of the SilverCloud Self-referral Website and Signup Process

A.4 Questionnaire Text

Firstly, we'd like to know a little more about you.

•

(1) Age group:

•

18–24

•

25–34

•

35–44

•

45–54

•

55–64

•

65+

•

(2) Gender:

•

Female

•

Male

•

Non-binary

•

Prefer not to say

•

Prefer to self-disclose: free text box

•

(3) Ethnicity:

•

Asian

•

Latino

•

Black

•

White

•

Arab

•

Indian

•

Native American/Alaskan Native

•

Mixed

•

Other: free text box

•

(4) Highest level of education:

•

Elementary/Junior/Primary

•

High School/Secondary

•

College/University

•

Postgraduate masters/doctorate

•

(5) Employment: please select all the options that apply to you

•

Employed (50 + hrs per week)

•

Employed (30–50 hrs per week)

•

Employed (less than 30 hrs per week)

•

Furloughed/temporarily out of work

•

On sick leave

•

On maternity/paternity leave

•

Unemployed

•

Retired

•

Student (full-time)

•

Student (part-time)

•

Caring for someone sick or elderly

•

Parenting young children (under age 12)

Please tell us about your experience with SilverCloud so far.

(6)

Why did you decide to check out SilverCloud?

Please select as many options as you like, or write your own

•

To feel better

•

To cope with a stressful life event

•

To improve my relationships

•

To improve my work, study or home life

•

Because my mental health is important to me

•

I was just curious

•

To help someone else with their mental health

•

It was recommended to me

•

Other (describe below): free text box

Notification group

(7)

Why are you unsure about signing up for SilverCloud?

Please select as many options as you like, or write your own

•

I'm not sure if I will be able to fit it into my life

•

My mental health isn't a priority for me right now

•

My mental health isn't important to me

•

I'm not sure if I need it

•

I need more support than this

•

I wouldn't want anyone to find out I was using it

•

I should be able to deal with my problems by myself

•

I'm not sure how useful it will be

•

I'm worried my data won't be secure

•

I'm not sure how to use it

•

I can't decide which program to choose

•

Other (describe below): free text box

Email group

(7)

Do you plan on logging back into SilverCloud?

•

Yes

•

Undecided

If Yes:

(7a)

What has been preventing you so far?

Please select as many options as you like, or write your own

•

I'm finding it hard to fit it into my life

•

My mental health isn't a priority for me right now

•

I forgot about it

•

I've been feeling better

•

I wouldn't want anyone to find out I was using it

•

I should be able to deal with my problems by myself

•

I haven't found it useful so far

•

I'm worried my data won't be secure

•

I'm not sure how to use it

•

I can't decide which program to choose

•

Other (describe below): free text box

If No:

(7b) Why not?

Please select as many options as you like, or write your own

If Undecided:

(7c) What is affecting your decision?

Please select as many options as you like, or write your own

•

I'm not sure if I can fit it into my life

•

My mental health isn't a priority for me right now

•

My mental health isn't important to me

•

I don't need it

•

I got what I needed from it already

•

I need more support than this

•

I wouldn't want anyone to find out I was using it

•

I should be able to deal with my problems by myself

•

I haven't found it useful so far

•

I'm worried my data won't be secure

•

I'm not sure how to use it

•

I can't decide which program to choose

•

Other (describe below): free text box

(8)

Do you have any further comments or thoughts on this topic?

Free text box

We're interested in hearing more about your experience.

Would you be available for a 20–30 minute online interview on this topic? We will reimburse you £/$50 for your time. Please enter your email address if you would like to be contacted with further details on the interview: ________________

Your participation in this research will help us improve SilverCloud for thousands of other people, thank you.

Submit

Debriefing Page

Thank you for taking part in our research.

If you are experiencing any distress caused by this survey, please visit the “Help” page of the SilverCloud website.

A.5 Multiple Choice Options and Associated Constructs from the Literature

Multiple choice answers to the primary outcome question: Why are you unsure about signing up for SilverCloud/ What has been preventing you from logging in? (Please select as many options as you like, or write your own)

Constructs from literature	Multiple choice options
Perceived Need/ Busyness/	I'm not sure if I will be able to fit it into my life
Perceived Fit	I forgot about it
	I'm not sure if I need it
	I need more support than this
Perceived Threat	My mental health isn't important to me
	My mental health isn't a priority for me right now
Stigma	I wouldn't want anyone to find out I was using it
	I should be able to deal with my problems by myself
Perceived Usefulness/Perceived Fit	I'm not sure how useful it will be
Internet Anxiety	I'm worried my data won't be secure
Perceived Ease of Use	I'm not sure how to use it
	I can't decide which program to choose

A.6 Reflexive Journal Excerpts

Interview period (21 April 2022)

Some of these interviews feel like therapy sessions, I'm not sure if my participants consciously realized it or not but it felt like some people took part because they just wanted someone to talk to. Or they wanted more information – I think one or two people actually signed up at the end of the interview because I explained how the DMHI works! There is so much going on here in terms of barriers, so many layers to unpick, more than I even imagined. There definitely seems to be a lot about human connection coming through (or is that just what is resonating with me?) and about people not really understanding what digital treatments are or having a mental model for what they might be like.

Familiarization (28 July 2022)

Coming to the end of the familiarization period has felt overwhelming, my brain is muddy with bits of information. How can I make sense of so many different opinions and experiences? What does it all mean? Do I have enough time to do the data justice? I feel overcome by the state of the world and all the suffering out there - so many of these people were just lonely. How did we get here, as a species that relies so much on social connection? I think covid catalyzed this disconnection, we all suddenly don't see our colleagues, miss family gatherings, would rather stay home than go out. Is this permanent damage, or can we go back? I can see that this is something I'm particularly interested in and concerned about, I need to be mindful of this and try not to get too bogged down in the emotional side of this research. I know I have a tendency to get side-tracked by the bigger picture, even when the bigger picture isn't the point. I do think there is value in understanding the root of the problem though, even if you can't do anything about it. It's what researchers are supposed to do, right?

TA discussion group (25 August 2022)

We just had out first TA discussion group and its so interesting to hear different opinions on the process and how it can be construed. We talked about philosophical perspectives and its funny how much easier it is to know what your position is on something when you discuss it with others. I definitely think critical realism makes the most sense for my data, it fits so well with the layered realities of mental health and what my participants said about outward “truths” not matching up with their inner experiences.

Coding – first round (5 September 2022)

I'm finding it hard to be reflexive and understand how my particular perspective is shaping the coding process. I almost have to force myself out of my head at intervals, to look at what I'm doing and thinking from the outside. I am skeptical of digital mental health support, even though I know that it can work and has worked for thousands of people. I feel that there is something fundamental missing from it. It's a feeling I also have about society in general and the rise of technology use, it sort of all misses the point, of what's important. I wonder would someone else, someone who is less of a luddite, have coded this very differently? To me it seems like a core part of the interviews, but then again there is so much in them, there are so many barriers. Coding around ideas of the uniqueness of mental distress and the causes of problems has been interesting, it makes me think about how we understand and interpret mental health, often as something that the individual is responsible for and can control, rather than as a reaction to external stressors or circumstances.

Coding – second round (12 October 2022)

I recognize that I have this certain narrative in my head that individualism is bad for our mental health, isolation is the root of much of our distress and technology is making this worse; disconnection in a hyper-connected world. I definitely felt this before I began this research, but sitting in the interviews listening to those confused, lonely people brought a certain sense of emotional resonance to this viewpoint. But now I'm worried that this emotional resonance could overpower my analysis. I mean, I think it is important and it was definitely there in the data, I didn't make it up. And I suppose that is the benefit of this method, I don't have to be objective because, as a human, its impossible to be objective anyway. This element that I picked up in the data is important, it resonates emotionally, and what is wrong with that? I think I'm slowly relinquishing the fear of subjectivity instilled in me from studying psychology and finally starting to grasp the beauty of reflexive TA. It feels more like my fine art training, with all its fluidity and critical awareness. It's exciting to combine both sides now in this research.

Theme development (29 November 2022)

Working more on theming I am coming across two things I'm wary of – 1. That some people close to me have been struggling with getting help and I'm wondering how dealing with this in my personal life is affecting the work, and 2. I'm aware of the impact of my own understanding of mental health on this research and how hard it is to comprehend these things from outside of the emotional/relational paradigm (and wider worldview) that I exist within. This is frustrating because I can see that the paradigm is part of the problem, but its really hard to see from outside of it, it is fundamental to the very core of how I see myself and how I relate to other people. Thinking about all of that changing or being different somehow seems impossible. I suppose even the fact that I'm aware of it as a paradigm and not an objective fact is useful.

A.7 Familiarization Mind Map

A.8 Image of First Round Coding Process

A.9 Ordering of Transcripts Based on Coding Round

Participant No.	Coding 1^st round order	Coding 2^nd round order	Coding 3^rd round order
1	1	10	10
2	2	5	9
3	3	15	8
4	4	7	7
5	5	19	6
6	6	2	5
7	7	12	4
8	8	1	3
9	9	17	2
10	10	9	1
11	11	20	20
12	12	6	19
13	13	13	18
14	14	3	17
15	15	18	16
16	16	8	15
17	17	16	14
18	18	4	13
19	19	11	12
20	20	14	11

A.10 Image of Candidate Theme Creation

A.11 Questionnaire Employment Characteristics

This question was multiple choice with a “select all that apply” response (N = 205).

One option chosen only	n (%)	Multiple options chosen	n (%)
Employed (30–50 hrs per week)	83 (40.5)	Employed (<30 hrs per week) & student (full time)	21 (10.2)
Student (full time)	40 (19.5)	Employed (30–50 hrs per week) & student (full time)	3 (1.5)
Employed (<30 hrs per week)	14 (6.8)	Employed (30–50 hrs per week) & student (part time)	3 (1.5)
Employed (50+ hrs per week)	12 (5.9)	Employed (<30 hrs per week) & student (part time)	2 (1.0)
Unemployed	8 (3.9)	Employed (30–50 hrs per week) & student (full time) & parenting	1(0.5)
Student (part time)	2 (1.0)	Employed (30–50 hrs per week) & parenting & caring	1(0.5)
Retired	3 (1.5)	Employed (30–50 hrs per week) & caring	1(0.5)
On sick leave	3 (1.5)	Employed (<30 hrs per week) & caring	1(0.5)
On maternity/paternity leave	1(0.5)	Student (full time) & parenting	1(0.5)
Parenting a young child (under age 12)	1(0.5)	Student (part time) & on sick leave	1(0.5)
Caring for sick or elderly	1(0.5)	Unemployed & parenting	1(0.5)
		On maternity/paternity leave & parenting	1(0.5)

A.12 Theme and Code Tables

Theme 1: Humans need humans

Codes	No. of Participants	No. of References
Afraid of being a burden	3	4
DMHI is isolating/impersonal	4	7
Expected human interaction	5	8
Humans are more reliable/familiar than tech	6	6
Lonely (no support network)	8	20
People forget to support each other	2	2
People need other people to help with mental health	6	8
Recommendation from others	5	12
Research interview was an opportunity to connect	2	7
Talking helps to process things	4	5
Testimonials normalise the process	4	6
Wants a human connection (to not feel alone)	12	27
Wants support for someone else	1	3
Wants to be accountable to someone	4	5
Wants to talk and feel heard	9	26

Theme 2: Needing help means I'm weak or I've failed

Codes	No. of Participants	No. of References
Admitting you have a problem is scary	7	19
Afraid of being judged by others	11	15
Afraid of wasting doctor/therapist's time (other people are worse than me)	3	4
Ashamed of mental health issues	13	26
Culture affects knowledge and stigma	3	5
Stigma to cope/not express emotions	5	8
Mental health awareness is better but long way to go	7	8
Mental health stigma is worse for men	3	4
Mental health talk is all talk (wellbeing washing)	5	6
Not sure whether they need help or not	9	11
People aren't aware of mental health	14	17
People don't know when to get help	10	13
People don't talk (openly) about mental health	12	14
People think mental health is shameful	7	8
Seeking help takes time	3	4
Uncomfortable talking about own mental health	5	6

Theme 3.1: Getting the right help is confusing: What is “help”?

Codes	No. of Participants	No. of References
Afraid of being put on medication	1	2
DMHI isn't help (don't need to admit you need help to do it)	7	16
DMHI is for general prevention not specific current issues	5	12
DMHI is for less severe issues	6	10
Less stigma for using DMHI compared to fact-to-face	5	9
Little knowledge about/experience of getting support	5	10
Mental health education is a specialised subject	2	3
Mental health info comes from tv/media (not trustworthy)	3	7
Mental health therapy is silver bullet (can be cured of mental health problems)	3	5
Negative past experience with mental health/seeking help	5	11
Not sure what type of help they need	8	10
People don't know how to get help	5	5
People don't understand mental health	13	16
Seeking help is for when you're severe	4	6
Talking is for more severe cases	10	13
Understanding mental health comes from experience with it	9	15
Wants credible/evidence-based help	6	10
Wants help from a qualified professional	5	13
Wants to be triaged/guided by a human	6	17
Wants to know what is suitable for them	7	12

Theme 3.2: Getting the right help is confusing: What the hell are DMHIs?

Codes	No. of Participants	No. of References
Curious to find out more	7	12
Decision/opinion based on quick first impression	5	13
Did questionnaire to find out more	4	7
Distrust in private sector tech industry/social media	3	12
DMHI is provided to cut costs for service	3	4
Expectations based on other apps/DMHIs	13	29
Expected it not to be interactive/tailored (same program for everyone)	5	10
Expected it not to work/be useful	7	12
Expected no human interaction	6	10
Expected signup to be easier	2	4
Familiarity with DMHI name increased interest in it	5	6
Marketing doesn't give enough info	10	15
Marketing led to unrealistically high expectations	4	7
More marketing needed	3	7
Not sure how signup works/how to start it	3	9
Not sure how to use DMHI	7	15
Not sure how useful it will be	6	10
Not sure if its free or not	2	3
Not sure of link between DMHI and service	3	9
Not sure what the DMHI is	16	48
Not sure which program to choose	4	7
Questionnaire/interview increased interest in DMHI	4	4
Signing up is effortful/complicated	7	13
Trusts service so trusts DMHI	8	12
Wants more info about DMHI before signup	9	21

Theme 4: “One size fits all” fits few

Codes	No. of Participants	No. of References
Already done CBT/uses CBT techniques	6	9
Already getting help elsewhere	5	7
CBT content isn't enough to help	5	9
Choice/different treatment options are important	4	6
Content wasn't relevant	5	11
Dislikes using tech/finds it difficult to use	3	6
DMHI is good for accessibility (disabilities)	1	2
DMHI is less effort than fact-to-face (don't have to travel/meet people)	9	19
DMHI is self-directed/flexible	8	15
DMHI wasn't suitable	7	17
DMHI wasn't what they were looking for	7	21
Each to their own	8	11
Generic content isn't helpful (can be found anywhere)	10	21
Individual preferences in what to do first (after signup)	4	8
Low self-efficacy/needs accountable other	5	9
Prefers fact-to-face (its 1st choice)	7	20
Pushed towards DMHI/only option available	7	15
Range of programs is appealing	8	10
Videos after signup are frustrating	2	7
Wants specific issues addressed	8	18
Wants to deal with a stressful life event	7	18
Wants to improve work/study	4	4

Theme 5: Accounting for changing needs

Codes	No. of Participants	No. of References
Comforting to know its there if you need it	4	5
Desperate for help	3	4
Didn't sign up until they felt they needed it	2	5
Distress makes you more confused	6	7
Distress reduces focus/self-motivation	8	11
DMHI can be used in the moment (coping strategy)	6	13
DMHI is immediate access/no wait	9	13
Expected it to be burdensome	3	5
Forgot about it	3	3
Getting the help you need takes time (waitlists)	7	8
Hard to prioritise/find the motivation to work on mental health	9	15
Help seeking motivation comes in waves	3	4
Improved so doesn't need it anymore	4	6
Measures don't capture real experience of mental distress	3	7
Mental health isn't taken seriously (not enough support for it)	6	7
No middle ground support	4	5
Plans on using DMHI in future	7	12
Signing up is doing something to help self (empowered)	7	12
Sought help only when it got really bad	3	4
Too busy to use it/sign up	8	15
Using DMHI doesn't feel good/make them feel better now	4	10
Working on mental health is challenging	3	4

Author Statement – Prior Publication Policy

A brief outline of the protocol for this study was accepted as a short workshop article (4 pages) to the ACM CHI 2022 Conference on Human Factors in Computing Systems workshop “Challenges, Tensions, and Opportunities in Designing Ecosystems to Support the Management of Complex Health Needs”. This short article is not archived in the ACM digital library and only included background literature, study design and methodology; no results were discussed. Aside from this short workshop article, this manuscript has not been submitted elsewhere and bears no relation whatsoever to the authors’ prior publications.

Acknowledgements

We wish to thank our participants for sharing their experiences and our editor and reviewers for their valuable feedback.

References

[1]

Sarah Ahmed. 2010. The Promise of Happiness. Duke University Press. DOI:

Participant No.	Coding 1^st round order	Coding 2^nd round order	Coding 3^rd round order
1	1	10	10
2	2	5	9
3	3	15	8
4	4	7	7
5	5	19	6
6	6	2	5
7	7	12	4
8	8	1	3
9	9	17	2
10	10	9	1
11	11	20	20
12	12	6	19
13	13	13	18
14	14	3	17
15	15	18	16
16	16	8	15
17	17	16	14
18	18	4	13
19	19	11	12
20	20	14	11

Participant No.	Coding 1^st round order	Coding 2^nd round order	Coding 3^rd round order
1	1	10	10
2	2	5	9
3	3	15	8
4	4	7	7
5	5	19	6
6	6	2	5
7	7	12	4
8	8	1	3
9	9	17	2
10	10	9	1
11	11	20	20
12	12	6	19
13	13	13	18
14	14	3	17
15	15	18	16
16	16	8	15
17	17	16	14
18	18	4	13
19	19	11	12
20	20	14	11

Abstract

1 Introduction

1.1 Digital Mental Health Interventions

1.2 Uptake and Early Engagement

1.3 Reflexive Considerations

2 Background

2.1 Barriers to Uptake and Early Engagement

2.2 Perceived Need

2.3 Culture and Stigma

2.4 Technology Acceptance

2.5 Implementation Factors

2.6 Studying Non-engagers

2.7 The Current Study

3 Methodology

3.1 Study Design

3.2 Intervention

3.2.1 Support Models.

3.2.2 Referral and Signup Flow.

3.3 Setting

3.3.1 Service 1: UK Health Service.

3.3.2 Service 2: UK Not-for-profit.

3.3.3 Service 3: UK University.

3.3.4 Service 4: US University.

3.4 Participants

3.5 Recruitment

3.6 Questionnaire Design

3.7 Quantitative Data Analysis

3.8 Interview Design

3.9 Qualitative Data Analysis

3.9.1 Primary Researcher Position Statement.

3.9.2 The Analytic Process.

4 Results

4.1 Questionnaire Results

4.2 Interview Participants

4.3 Qualitative Analysis

4.3.1 Theme 1: Humans Need Humans.

4.3.2 Theme 2: Needing Help Means I'm Weak or I've Failed.

4.3.3 Theme 3.1: Getting the Right Help is Confusing: What is “Help”?.

4.3.4 Theme 3.2: Getting the Right Help is Confusing: What the Hell are DMHIs?.

4.3.5 Theme 4: “One Size Fits all” Fits Few.

4.3.6 Theme 5: Accounting for Changing Needs.

5 Discussion

5.1 Overview

5.2 Implications for Design and Implementation

5.3 Implications for Research

5.4 Transdisciplinary Implications

5.5 Limitations

5.6 Conclusion

A.1 Position Statements

A.2 Screenshots of the SilverCloud Platform

A.3 Screenshots of the SilverCloud Self-referral Website and Signup Process

A.4 Questionnaire Text

A.5 Multiple Choice Options and Associated Constructs from the Literature

A.6 Reflexive Journal Excerpts

A.7 Familiarization Mind Map

A.8 Image of First Round Coding Process

A.9 Ordering of Transcripts Based on Coding Round

A.10 Image of Candidate Theme Creation

A.11 Questionnaire Employment Characteristics

A.12 Theme and Code Tables

Author Statement – Prior Publication Policy

Acknowledgements

References

Cited By

Index Terms

Recommendations

Searching for Mental Health: A Mixed-Methods Study of Young People's Online Help-seeking

Measuring and Fostering Engagement with Mental Health e-Coaches

Adolescent Engagement in Mental Health Services

Comments

Information

Published In

Publisher

Publication History

Check for updates

Author Tags

Qualifiers

Funding Sources

Contributors

Other Metrics

Participant No.	Coding 1^st round order	Coding 2^nd round order	Coding 3^rd round order
1	1	10	10
2	2	5	9
3	3	15	8
4	4	7	7
5	5	19	6
6	6	2	5
7	7	12	4
8	8	1	3
9	9	17	2
10	10	9	1
11	11	20	20
12	12	6	19
13	13	13	18
14	14	3	17
15	15	18	16
16	16	8	15
17	17	16	14
18	18	4	13
19	19	11	12
20	20	14	11