Introduction

Digital technologies to support mental health have become increasingly popular; a trend that looks set to continue. The shift towards utilising distance-based digital solutions for mental health care was accelerated by the COVID-19 pandemic [1] and the increasing demand-versus-supply gap for mental health services [2]. Announcements by the UK government in England of investment in digital mental health, as well as recommendations for how digital solutions can be utilised in the healthcare system [2], suggest digital technologies will become more and more commonplace in our health services. Digital mental health interventions (DMHIs) are appealing because they improve accessibility to support [2], are more easily scalable in comparison to non-digital interventions [3], and have some evidence of effectiveness particularly when compared to a waitlist control [4]. DMHIs are interventions addressing a mental health-related target, delivered by commonly available technologies such as mobile applications, websites, or text messages [5].

A plethora of DMHIs are already available but few have been thoroughly empirically tested. A 2020 review by Marshall et al. [6] found 293 smartphone applications available through the Apple App Store or Google Play Store that offered a therapeutic treatment for anxiety and/or depression. This does not even include applications offering singular aspects of treatment, such as mood monitoring, or applications targeting other mental health conditions. Despite the number of applications available, only 55.3% claimed to have been developed using an evidence-based theoretical framework, and only 10 had their effectiveness investigated in a published research study [6]. Although Marshall et al’s [6] review was limited to only results appearing in the App stores, and the information disclosed in the app’s description, it demonstrates a lack of published evidence to support apps that are widely available to users. The quality of these apps and whether they are delivering evidence-based support is therefore uncertain.

Moreover, it has been reported that 90% of users fail to engage with mental health applications for longer than 10 days, with some studies reporting an average retention of just 5.5 days [4]. If digital technologies are to be recommended and have impact there is a clear need to increase engagement with their intended users. One way to do this is to engage with the DMHI’s key stakeholders from the conception and design phase [7]. Understanding users’ lived experiences can uncover their requirements, challenges, and any potential barriers or facilitators to engagement that may otherwise be overlooked [7–12]. The end result are digital solutions that are more relevant to their intended user and context, integrating more easily into daily life and use, improving their overall acceptability and feasibility [7,13].

Young people who self-harm in particular may benefit from access to effective DMHIs. In the UK, 7.8% of 7–16-year-olds (based on parental report), and 32.8% of 17–24-year-olds (via self-report) reported having ever self-harmed in 2022 [14]. Young people who self-harm often experience difficulties meeting referral criteria for specialist mental health services and long waitlists to receive in-person support once they are referred [15–17]. DMHIs are one possible solution that may be helpful in providing support to these individuals because they eliminate the need for access to in-person services. DMHIs overcome additional help-seeking barriers young people who self-harm may face such as accessibility from any location, preferences to manage their problems on their own, and supporting those who may struggle to identify a trusting adult to approach for help [18].

Young people are also an ideal target group generally for DMHIs. From 11 years old over 90% of young people in the UK own their own mobile phone [19], and most young people spend more time online than older adults [20]. Additionally, young people report liking digital forms of communication when it comes to their healthcare [21], and are satisfied with DMHIs finding them acceptable and helpful [22–24].

Currently most DMHIs designed to address self-harm are based on Cognitive-Behavioural Therapy (CBT) [25]. Whilst some of these DMHIs have found positive effects post-intervention, others had to abandon testing due to problems with feasibility, or have not published findings yet. In addition, although CBT’s focus on changing thinking and behavioural patterns is helpful for many individuals [26,27], different therapeutic approaches may be valuable for others.

Solution-focused therapy (SFT) is one therapeutic approach that differs from CBT and is not currently delivered digitally. SFT focuses on exploring the individual’s ideal future and what ‘better’ looks like for them [28]. SFT is therefore flexible and considerate of individual goals and perspectives. This is important given the complexity of self-harm presentations [29], and the importance of aligning with an individual’s goals to effectively promote healthier ways of coping. The focus on their desired future and what it looks like when things go well can be empowering and motivating [30,31], compared to more traditional approaches which often focus more on altering problematic behaviours or uncovering the cause of a problem [28].

There is lots of evidence for the use of SFT with children and young people in a variety of settings [32] and some evidence for its applicability to treating individuals presenting after self-harm [31]. However, what is yet to be established is how young people who self-harm feel about engaging with SFT delivered digitally. SFT has typical questions and steps that are core to the approach [28], which in theory should be possible to deliver through a digital medium. SFT has been delivered previously through a self-directed web-based programme to college students with positive results [33], but not through a DMHI delivered to a mobile phone.

We decided to develop our DMHI to be delivered by text message since approximately 1 million children and young people in the UK do not have access to a personal smartphone or the internet at home, but usually have access to a simple mobile phone capable of sending and receiving text messages [34]. Text message-based digital interventions to support individuals who self-harm do exist, but mostly consist of simple, unidirectional interventions [25,35–37], such as reminders of supportive messages and links to support. However, two-way interaction via text message is possible that would enable a more engaging and conversational approach, like that provided by a chatbot.

Chatbots are promising since they are accessible and scalable given that they are automated, rather than requiring an individual or service to respond in real-time. Additionally, users have reported overall positive perceptions of therapeutic chatbots for mental health, liking their availability, perceived lack of judgement, ease of use, and usefulness for reflection and skills development [38]. Utilising a chatbot-style approach would eliminate the issues with waitlists and accessibility so often faced by young people who self-harm, and would allow us to develop a DMHI that could explore an individual’s goals and desires for the future, as is done typically within SFT.

We are proposing development of an automated, chatbot-style, DMHI delivered by text message, based on SFT. Studies by Kruzan et al. [39] and Hetrick et al. [40] have previously explored the preferences of young people who self-harm for a digital intervention to support their mental health. These studies focused primarily on understanding how participants managed their mental health and urges to self-injure, and how they perceived technology to help with this management. Neither seemed to explore a specific digital modality or therapeutic framework in their questions. This study therefore will be the first to explicitly explore the development of a DMHI based in SFT and delivered by text message with key stakeholders. Key stakeholders for this project are young people with lived experience of self-harm, as the target user, and mental health professionals who work with young people who self-harm, to ensure the DMHI developed is clinically relevant and has a solid therapeutic foundation.

This study aims to explore stakeholder perceptions of a chatbot-style DMHI based on SFT for young people who self-harm. This will enable us to develop a novel intervention within an evidence-based framework (SFT), and consider user requirements and facilitators in its design. This aim was achieved through focusing on the following objectives:

1. Obtain clinicians’ and young people’s views of a text message intervention based on solution-focused therapy.
2. Determine what factors will promote and hinder engagement with a DMHI delivered by text message.
3. Determine what content is important to be included within the intervention.

Methods

The reporting of the study has been organised in accordance with the Standards for Reporting Qualitative Research (SRQR) checklist [41]. The protocol for this study was preregistered at Open Science Framework: https://osf.io/3mpwf/.

This study uses focus groups and individual interviews with young people with lived experience of self-harm, and with clinicians who have experience of working with young people who self-harm and of SFT. Focus groups were sought in the first instance as they would enable ideas and suggestions relevant to the research aims to develop through collective discussion, ultimately reflecting the perspectives of the participants [42]. Focus groups encourage discussion where opposing views can be presented and challenged, enabling a range of different opinions on a topic to be explored. This was particularly helpful for the aims of this study, where different attitudes towards digital technologies and therapeutic approaches could be explored in the discussions. Individual interviews supported the focus group discussions by providing more in-depth exploration of individual views of the topic. Individual interviews were primarily organised for participants who were unable to attend one of the arranged focus groups.

Recruitment

We aimed to conduct four focus groups in total, two with young people and two with clinicians, to capture a range of perspectives on our proposed DMHI. We aimed to recruit 6–8 participants per focus group, which is recommended as an ideal sample size for focus groups [42]. This number would also ensure there were enough participants in each focus group for a discussion with different and opposing views to ensue, even if we encountered some drop-outs.

We took a convenience sampling approach to recruitment. Participants were recruited through adverts disseminated online through East London NHS Foundation Trust, or were individuals already in contact with the research team who had either previously engaged in, or expressed an interest in, the wider project this study is a part of. All individuals recruited through their existing engagement with the research project had been identified originally through UK-based organisations. Interested individuals who self-identified as meeting the eligibility criteria (see Table 1) contacted the research team. Although we were conscious of the representativeness of our participants, particularly in terms of gender, age, ethnicity, and professional role (for clinician participants) we did not purposively recruit based on these characteristics.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Eligibility criteria.

https://doi.org/10.1371/journal.pdig.0001276.t001

Given that young people with lived experience of self-harm are the target audience for this novel intervention it is essential to include them in the development process to ensure their needs and requirements are sufficiently considered. Additionally, as the planned intervention will be based on SFT, it is necessary to consult clinicians with relevant experience of using this approach with young people who self-harm in order to obtain their perspectives on what works well and what is important to consider when using this approach in practice.

Results

Separate focus groups were held with clinician and young people participants. We conducted two focus groups with young people (n = 4, n = 5), and one focus group with clinicians (n = 12). Additional individual interviews were conducted with three young people participants and four clinician participants. Although we aimed to conduct two focus groups with each participant group, it was difficult to organise a second with the clinician participants due to their availability. The young people participant focus groups were originally planned to be split by age (under v. over 16 years), but due to only recruiting individuals over 16 years of age, focus groups were organised based on participant availability. The focus groups lasted around 60 minutes whereas the individual interviews lasted around 45 minutes.

Participant characteristics

Table 2 provides details of the participants in the study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Participant socio-demographic information.

https://doi.org/10.1371/journal.pdig.0001276.t002

We recruited 14 young people participants in total. Two young people withdrew from the focus groups prior to them taking place so the information is reported for the 12 who participated in a focus group (n = 9) or interview (n = 3).

All young people participants had their own smartphone, but two participants did not have consistent access to the internet at home. Participants used their mobile phones for a range of purposes, but all used it for social media (including WhatsApp). No participants identified any access needs when using their mobile phone. Only two participants reported using existing digital apps to support their health, these included Calm, HeadSpace, and the Health app provided by Apple.

We recruited 16 clinician participants in total. Clinician participants were from a range of clinical roles, predominantly social workers, support workers, nurses, and a range of different therapist roles. Most clinician participants worked predominantly with young people, and reported using a wide range of therapeutic approaches including Solution-Focused approaches, Cognitive Behavioural Therapy (CBT), Dialectical Behavioural Therapy (DBT), and Systemic Therapy, amongst others.

Main findings

Our primary aim was to explore key stakeholder views of our proposed DMHI so their requirements can be considered throughout its design and development. A summary of the overall key recommendations is presented in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Summary of the key recommendations from our findings.

https://doi.org/10.1371/journal.pdig.0001276.g001

Discussion

This study explored what young people with lived experience of self-harm, and mental health professionals, want a DMHI based on SFT to look like. Overall, the perception of a chatbot-style DMHI, delivered by text message, based on SFT, was positive. SFT delivered through a digital medium was seen as a different approach that could prompt the user to think about things in a new way. Considering the requirements and preferences highlighted in our findings when we develop our DMHI will help our final DMHI to be acceptable, engaging, and effective.

Personalisation was one of the most prominent themes in the discussions we had with participants in this study. This included both wanting the content of the messages within the intervention to be personalised, as well as discussions of whether digital interventions could be personalised enough to be engaging and impactful. The Affect-Integration-Motivation and Attention-Context-Translation (AIM-ACT) framework conceptualises engagement with digital stimuli as a multi-stage process, and proposes engagement is most likely to translate to actual behaviour when outcomes are internalised and valued by one’s self [45]. This suggests goals personalised to a user that align with their values will have the best chance of being achieved. Given the complex and unique nature of how young people understand their self-harm thoughts and behaviours [39], personalisation of a DMHI may be particularly important in this context in order for it to be relevant to their individual needs and goals. SFT centres primarily around whatever the client wishes to discuss, helping it to align with their values. It is well established that accommodating client preferences within traditional psychotherapies has a positive impact on treatment adherence and outcomes [46]. It is therefore important that the same accommodation needs to be made in DMHIs.

However, the evidence for the impact of such personalisation within DMHIs is lacking, which may in part be due to discrepancies in how the term ‘personalisation’ is used and understood in DMHI research [47]. Furthermore, most existing digital interventions use rule-based or user choice methods of personalisation, making adaptations quite simple (i.e., different content if a user is above or below a certain symptom score), rather than employing more complex models that could additionally account for a wider variety of user characteristics and preferences (for example including sensor data like GPS location or mobile phone usage like the number of texts or calls made) [47]. Using more complex models requires more complicated programming and data input, but the result would be an intervention with an illusion of better memory and rapport with less effort required by the user [48]. Although more effort is required by the developer, our participants felt tailored and customised content would make a novel DMHI more appealing. This fits with existing research suggesting that generic content does not tell users anything new and is not wanted, instead customised content is key [39,40]. In our proposed DMHI, this could include using both algorithms and scale scores to determine whether users are struggling at that moment, based on their responses to the chatbot, and thus direct them to messages exploring coping rather than moving forwards. Clarity in future research of how exactly DMHIs are personalised will help to evaluate which aspects of personalisation are important.

Privacy and confidentiality was another topic that featured consistently in the discussions we had in this study, particularly with the young people participants. Privacy and security are considered by the NHS Confederation as one of the biggest challenges to successful integration of digital technologies into mental healthcare [2]. Although data protection regulations and policies exist, there needs to be an approach to ensure digital interventions adhere to these standards [2,49]. In recent years there has been an increased focus on developing regulatory guidance to match the development of digital solutions, including project funding such as that granted by the Wellcome Trust to produce guidance regulating digital mental health tools specifically [50]. Including key stakeholders in this work will ensure the resulting regulations reflect the needs and priorities of the end users. What is apparent from our findings is that young people have clear expectations for how their information is kept private when using digital technologies. This highlights a need for DMHIs to prioritise privacy and confidentiality for their users, whilst including appropriate pathways to break confidentiality in situations where concerns for safety and risk take precedence.

Clinician participants in this study highlighted a need for clarity around how and when users would be escalated for additional support, and what happens if users disclose something to the intervention which raises a concern for their safety. In order for digital interventions to be successfully adopted and sustainable, the health/care systems and wider context they will be delivered in are important to consider from the design phase [51], as well as the individual user. Careful consideration of how to balance user preferences (i.e., users may not want to provide personal details) so benefits such as encouraged disclosure to the intervention are retained whilst prioritising safety will be key to engagement and adoption. In the development of our intervention, the safety issues raised (such as what happens if someone says they are going to harm themselves) will inform the approach to risk escalation integrated into the intervention that can then be tested with young people for acceptability in its testing phase. This could include an automatic triage of messages flagged for safety concerns to a clinical team when a user’s identity is known, or responses from the chatbot being triggered which include helplines to contact in a crisis, and exploration of a person the user can contact if they feel unsafe.

There are some commonalities in our findings with the studies conducted by Kruzan et al. [39] and Hetrick et al. [40]. Kruzan et al. [39] found participants used technology to learn about self-injury and alternative ways to cope, which is similar to our finding whereby participants would like the option to receive educational content on coping. Both Kruzan et al. [39] and Hetrick et al. [40] found that participants valued peer support for feeling they are not alone, but recognised this could become negative if not appropriately moderated. We also found participants wanted a space to talk to peers going through the same difficulties. Kruzan et al. [39] also found a range of activities that were specific to the user were deemed most helpful and engaging, which needed to reflect changing needs and goals over time. This is similar to our finding that participants wanted personalised content, and valued the use of solution-focused questions like the ‘best hopes’ question to make their unique goals and needs the theme for the work in the intervention. Both Kruzan et al. [39] and Hetrick et al. [40] findings share similarities with those reported in our study. We add to this existing work by specifically exploring the delivery of SFT in a digital format, and report direct recommendations of the desired content for a DMHI. Specifically one delivered as a chatbot via text message.

To our knowledge this is the first study to provide evidence for the perspective of young people who self-harm on SFT as an approach to support their mental health. Previous work has explored the perceptions of adults [30] and young people with learning difficulties [52] of SFT, after they had engaged in SFT sessions. In those studies participants liked being supported to define their aims, the attitude towards change and the future, and focusing on strengths, resources, and moving forward. This is similar to the findings of our study. Participants liked the focus on the future and strengths, and the use of solution-focused questions felt like a new and different approach to what they were used to, providing an opportunity to change their perspective and challenge their thinking. Some questions, such as exploring exceptions to the problem, were considered challenging; highlighting careful consideration of techniques that might require more support to explore. These findings add to the evidence-base for the use of SFT with young people who self-harm.

However, some participants in our study also highlighted that SFT could feel a little simplistic. Some participants stated that they knew focusing on positives for example was a good step to take, but either struggled to see the benefit of this approach or were unlikely to find these kinds of questions engaging. A challenge in developing our proposed DMHI will be to convey the therapeutic power of the approach whilst being simple and straightforward, and keeping users engaged in a self-directed autonomous delivery of the approach. Referring again to the AIM-ACT framework [45], digital stimuli need to balance complexity and simplicity. A positive affective response to stimuli will increase the likelihood of motivation to carry out the end behaviour. Nahum-Shani et al. [45] propose this is most likely if the intervention is not either frustrating or results in boredom. This suggests we need to think carefully about how questions are framed and concepts explored so that the approach feels novel and not boring. For example, in SFT typically the therapist will ask the client ‘and what else?’ to uncover additional detail of what their ideal future and better looks like for them. In a DMHI we will need to balance including these questions to probe for further detail, with moving on or sufficient rephrasing if the user cannot think of something so the conversation does not feel repetitive and frustrating.

Additionally, other types of questions and activities that are not typically used in SFT were still seen as valuable by our participants (e.g., education on coping, mood tracking, peer support). This raises a question of whether developing a DMHI strictly within one therapeutic approach is most effective, or whether a more integrative approach is beneficial. It will be important when examining effectiveness to try to determine which aspects of the DMHI are beneficial, and whether the SFT content alone is sufficient to result in positive outcomes. Studies like O’Sullivan et al’s [53] have used methods such as causal analyses to begin to try to determine which aspects of DMHI’s influence engagement and outcomes. Similar methods could be employed at the evaluation stage of other DMHIs.

Despite the clear importance of involving end users in DMHI design and development to promote acceptance and engagement, how we measure and test these constructs is important. A review by Bear et al. [13] of mental health apps for young people, focusing on assessing implementation, consistently found that engagement decreased over time. Whilst this may appear concerning, inconsistency in how engagement is conceptualised makes it difficult to determine whether this disengagement is in fact a problem [13]. Choosing which metrics to measure acceptance and engagement by is important, and in fact lack of usage may actually represent a positive outcome [3]. For example, a mobile application containing a safety plan for individuals at risk of suicide may not be used because the individual does not need to use it, not because they dislike the app itself. When evaluating DMHIs it needs to be decided apriori what good engagement looks like, or whether measures of improvement are in fact more appropriate to determine the DMHIs success [54]. Measuring both usage of the DMHI as well as user wellbeing or symptom improvement could help to determine whether users simply do not feel a need to engage with the DMHI, or whether they have not engaged because they do not perceive it as helpful. Furthermore, obtaining feedback on the DMHI from users will be most helpful to establish its acceptability, and in what scenarios users may or may not engage with it.

Conclusions

Young people with lived experience of self-harm, and mental health professionals, were positive about a novel chatbot-style DMHI, delivered by text message, based on SFT, to support the mental health of young people who self-harm. Our participants’ positive perceptions of SFT and suggestions for its delivery adds to the evidence-base for its use with young people who self-harm. Participants in this study provided us with numerous recommendations that can be actioned in the development of DMHIs to improve the likelihood of acceptability and engagement. In particular, careful consideration of how to make our DMHI suitably personalised to each user, how to deliver SFT simply but with enough complexity to remain engaging, and how to determine the effectiveness of our DMHI and its components is needed.

Supporting information

Acknowledgments

First and foremost we would like to thank all the participants who took part in the study and provided their insights and experiences to make this piece of work possible. We would also like to thank our colleagues in the Unit for Social and Community Psychiatry and the Youth Resilience Unit at Queen Mary University of London for their support in developing and refining this manuscript.