As part of the EMPATIA project [36], the BePart study constituted the predesign phase of an iterative co-design process to develop a preventive transdiagnostic DMHI for adolescents. Through semistructured interviews, adolescents’ needs, preferences, and perceived barriers were explored to inform the development of a prototype, which was subsequently evaluated and refined through focus group interviews with adolescents during the design phase. The developed intervention was then evaluated in a randomized controlled trial as part of the broader EMPATIA project. The BePart study was conducted in German-speaking cantons of Switzerland. Between January and June 2022, adolescents from the general population were recruited to participate in one semistructured interview.

This study was approved by the Ethics Commission of the Faculty of Human Sciences of the University of Bern (2021-12-00003) and is in accordance with Articles 3(k) and 23 of the Swiss Human Research Act (HFG), which permit capable adolescents aged ≥14 years to provide their own written informed consent for minimal-risk research. Upon contacting the study team, potential participants received the study information and informed consent documents by mail. Participants aged ≥14 years signed the informed written consent form themselves and returned it via mail. For participants aged 12 or 13 years, written informed consent was obtained from a parent or legal guardian, and written assent was obtained from the participant. Participation was voluntary, and participants were informed that they could withdraw at any time without consequences.

Interview data were pseudonymized. Prior to the interview, each participant generated a personal code. Audio recordings were stored under this code. At no point during the interview was the participant’s name recorded. The code key linking participants’ identities to their pseudonyms was stored separately and was accessible only to the principal investigator. Audio recordings were deleted following transcription. All data were stored in a password-protected folder accessible only to the principal investigator. Only deidentified data were used for analysis and reporting. Participants received CHF 20 (∼US $21.2, as of January to June 2022) in the form of a voucher after the interview. Additionally, they were reimbursed for potential traveling expenses.

Participants were recruited through voluntary response sampling. Study information was distributed online (via social media profiles and the project website), through flyers, media interviews, and cooperation partners. Interested participants were asked to contact the study team via email, direct messages on social media, or phone call to arrange a date for the interview. To be included, participants had to (1) be between 12 and 21 years old, (2) have their main residency in Switzerland, and (3) understand and speak German. Prior personal experience with DMHIs and/or mental health problems did not serve as an inclusion or exclusion criterion. Although participants ranged from 12 to 21 years, we use the term adolescents throughout the paper in line with contemporary developmental definitions that conceptualize adolescence as an extended developmental period [43,44].

After having provided consent, participants and interviewers agreed on the study information via mail. Participants (or their legal guardians when aged <14 years) needed to sign the informed consent form and to return it via mail. Participants and interviewers then agreed on a time and date for the interview, either in person or via online video call. Prior to the interview, all participants were asked to complete an anonymous online survey about help-seeking and psychological distress (refer to Materials section). Only the interviewer and the participant were present at the date of the interview. The interview started with demographic questions about age, gender identity (female, male, and other), and sociodemographic data. Race and ethnicity were not collected because the study was not designed to examine differences in intervention perceptions or engagement across demographic groups. Thereafter, participants were again informed on the study procedures in a standardized manner (Multimedia Appendix 1). If the participants had no further questions, the interviewer started the audio recording and carried out the interview.

Given the sensitive nature of the interview topics, a participant-centered approach emphasizing participant safety and autonomy was applied, consistent with recommendations for qualitative research on such topics [45-47]. All interviewers were advanced master’s students specializing in clinical child and adolescent psychology with prior experience conducting clinical interviews. Before data collection, they completed a 2-day training workshop covering communication and active listening skills, recognizing signs of distress, maintaining appropriate boundaries, and procedures to support participant safety and autonomy. A predefined procedure for managing participant distress was established. If participants showed signs of distress, interviewers paused the interview, assessed the participant’s immediate well-being, and offered the option to continue, take a break, skip questions, or terminate the interview. Participants were informed that participation was voluntary and that they could pause or withdraw at any time without consequences. Participants requiring further support were provided with written information on available support services. If additional clinical support was required, interviewers could consult SJS, a licensed psychotherapist, during or after the interview, and participants could be referred to the outpatient service of the involved clinical division. Established clinical procedures for acute safety concerns, including urgent referral when clinically indicated, were available. Potential challenges and critical situations were discussed during regular team supervision meetings.

The research team regularly reviewed the latest interview audios to monitor data saturation, assessing whether any new key topics emerged. This process continued until the team agreed that no new information leading to possible codes for answering the research questions had emerged, indicating that recruitment could be concluded.

The semistructured interviews were conducted by 4 female master’s students in their last term (Christina Hiltbrunner, Nika Saxer, Taina Thees, and Vera Bächler). The interviewers did not have previous experience in qualitative interviewing but were experienced in conducting clinical diagnostic interviews with adolescents. The interviews were coded by NW, PhD, and AH, MSc. The coders and interviewers were supervised by SJS, a professor in Clinical Psychology and licensed psychotherapist for children and adolescents.

Prior to data collection, interviewers conducted pilot interviews with adolescents that were reviewed within the research team to refine interview questions and ensure comprehensibility. These pilot interviews were complemented by structured role-plays among the interviewers, which were discussed to develop a shared understanding of appropriate open-ended follow-up questioning and to identify potential interviewer biases. Throughout data collection, interviewers participated in weekly team supervision meetings with SJS, during which emerging challenges were discussed. These procedures were implemented to enhance the credibility and trustworthiness of the data, consistent with established quality criteria for qualitative research [48,49].

The online survey included the General Help Seeking Questionnaire (GHSQ) [50], the Actual Help Seeking Questionnaire (AHSQ) [51], and the Kessler Psychological Distress Scale (K10) [52]. The semistructured interview guide was developed specifically for this study (Multimedia Appendix 1). It was informed by factors influencing help-seeking [2], the COM-B model [22], and by the intervention-level factors associated with engagement in DMHIs identified in the systematic review by Liverpool et al [16]. Accordingly, the interview questions covered the entire pathway from help-seeking to the uptake and continued engagement with a proposed preventive DMHI. The interview guide was reviewed by an external qualitative researcher who was not otherwise involved in the study. Based on this feedback, several questions were revised to place greater emphasis on participants’ behaviors and experiences rather than opinions. Additionally, the interview questions were subsequently piloted with 2 adolescents to assess comprehensibility and estimate the interview duration. The pilot interviews were not included in the final dataset.

The interview began by exploring participants’ general use of digital applications and online help-seeking behaviors (eg, “Where would you look online for information about mental health problems and for help if needed?”). Participants were then introduced to transdiagnostic, preventive DMHIs to establish a shared understanding of the intervention type before discussing factors that would influence their willingness to start using and continue engaging with such interventions (eg, “What would you need to tell your friends about the DMHI to get them to start using and engaging with it?”). Participants subsequently evaluated 21 predefined intervention features [16] (eg, video content or text) by rating their importance on a scale ranging from 1 (“not important at all”) to 10 (“very important”). The ratings served solely as structured prompts to facilitate discussion. Participants were explicitly asked to elaborate on their responses, suggest implementation strategies, introduce additional features, or reject items they considered unimportant or irrelevant. The predefined features were drawn from the systematic review by Liverpool et al [16] and reflect their 3 intervention-specific dimensions of acceptability, usability, and suitability. Within the acceptability dimension, predefined features included content format options (information texts, animated videos, videos with real-life adolescents or influencers, and everyday-life tasks), exchange options (with peers, e-coaches, and professionals), and the inclusion of parents. Within the usability dimension, features included layout and design, offline availability, and the self-administered format of the program. Within the suitability dimension, features included options to adapt content (eg, additional examples, own profile, and choice of avatar), data security and anonymity settings, and feedback mechanisms (specific progress feedback or integrated progress bar). This study thus operationalizes intervention-specific dimensions by Liverpool et al [16] at a more granular level, drawing directly on adolescents’ own perspectives in a preventive transdiagnostic context. The interview concluded with questions on personalization (eg, “How could we best tailor the DMHI to each individual participant?”). All interview data including all follow-up calls were used in the qualitative analysis.

All interviews were transcribed verbatim following the transcription rules described by Kuckartz [53]. Data were analyzed using structured qualitative content analysis [54] in MAXQDA [55]. This approach aims to identify and systematize topics and subtopics and examine relationships between categories through an iterative process comprising the following seven phases: (1) initial work with the text, (2) development of main topics, (3) coding data with main topics (first coding cycle), (4) inductive development of subtopics, (5) coding data with subtopics (second coding cycle), (6) simple and complex analyses, and (7) writing up of results. Research questions can continuously be adapted up to phase 6 [54]. In the present analysis, the originally planned research questions distinguished between personal, intervention, and system-level factors. During the analysis, it became apparent that factors relevant to the uptake and engagement with preventive transdiagnostic DMHIs could not always be meaningfully assigned to a single level. Therefore, the third research question was refined to explicitly capture factors operating across levels.

A hybrid deductive-inductive approach was applied, consistent with theory-informed qualitative content analysis [54]. The initial coding framework was developed deductively based on the COM-B model [22] and a systematic review on factors influencing engagement with DMHIs among young people [16]. At the person level, initial topics represented the COM-B constructs capability, opportunity, and motivation. At the intervention level, topics were based on intervention-related factors identified in previous DMHI research [16], namely acceptability, usability, and suitability. In addition, a broad help-seeking topic was included [2] to reflect the structure of the interview guide, which examined adolescents’ help-seeking pathway from awareness of mental health needs to considering, accessing, and engaging with preventive DMHIs. These topics guided the first coding cycle, which was conducted independently by both coders, NW and AH, using a subset of 4 interviews.

During subsequent coding cycles, the framework was refined iteratively based on the interview material. Subtopics were developed inductively, allowing participants’ perspectives and unexpected concepts to shape the final analytical structure. Participants frequently described expectations extending beyond a single intervention toward broader visions of preventive mental health care; therefore, system-level factors were added as an inductively derived main topic. Similarly, discrepancies between adolescents’ expectations and perceived implementation possibilities emerged as a cross-cutting topic. The distinction between anticipated uptake and expected engagement also became increasingly relevant during analysis, as perspectives on DMHI characteristics differed across stages of the help-seeking process. The “access to health care” framework by Levesque et al [56], adapted for young people by Werlen et al [9], was consulted to support interpretation of this process dimension. Additionally, the health belief model [57] was used to refine terminology for inductively developed belief-related subtopics. These frameworks supported interpretation but were not used as predefined coding categories. Table 1 provides an overview of the deductive theoretical starting points to the inductively developed final topics, indicating the theoretical frameworks and interview questions associated with each, and providing commentary on the analytic decisions made. After establishing the coding system, the coders independently coded 2 interviews to assess intercoder agreement. Agreement was evaluated using MAXQDA [55] by comparing all coded segments on a document level, resulting in 73.4% agreement and a Cohen κ of 0.72. NW and AH subsequently discussed all coding discrepancies, including differences in assigned codes and uncoded segments. Most differences occurred in the subtopic coding of Beliefs about DMHIs influencing anticipated uptake and engagement. Therefore, corresponding code definitions were refined to improve distinction between subtopics. Following this discussion, full agreement was reached, and the revised coding system was applied to the remaining interviews, with NW and AH independently coding half of the interviews each. Weekly meetings were held throughout the coding process to ensure consistency. After completion of coding, AH reviewed all coded segments, and remaining uncertainties were resolved through discussion. Representative participant quotations were selected from the original German transcripts and translated into English by the first author (AH) and additionally reviewed by the senior author (SJS), both native German speakers proficient in English. The translations aimed to preserve the conceptual meaning and participants’ intended message rather than providing a literal word-for-word translation.

The feature ratings were analyzed quantitatively and reported separately in Multimedia Appendix 2. They were not incorporated into the qualitative coding or theme development.

The first author (AH) was actively engaged in developing a transdiagnostic preventive DMHI at the time of data collection, and NW brought in expertise from prior research on a similar DMHI for adolescents with social anxiety. Both initially framed topics within existing theoretical models, which they actively challenged through reflexive diary-keeping and regular team discussions to remain open to new insights. Additionally, 4 master’s students conducted the interviews. Their closer age to participants may have facilitated rapport and open discussion, while also increasing the possibility that shared assumptions about digital engagement were left unexplored. Further reflexivity considerations, including the influence of the all-female team composition and the time lag between data collection and analysis, are detailed in Multimedia Appendix 3.

A total of 48 participants were interviewed (6 in person, 42 via online video call). Interviews lasted between 30 and 85 minutes (mean 50.04, SD 11.58 min). Thirty of 48 participants were recontacted via telephone to elaborate on previous responses such as provided feature ratings, with these follow-up conversations lasting between 1 and 28 minutes (mean 12.28, SD 7.95 min). Moreover, 4 participants had a personal connection with one of the interviewers. To minimize potential bias, they were interviewed by an interviewer with whom they had not previously interacted.

The sample consisted of predominantly well-educated, female participants aged 12 to 21 years (refer to Table 2 for the sample description). During the interviews, 9 (18.8%) participants reported prior use of a digital application related to (mental) health (eg, information websites, chats, and meditation apps).

This study is among the first to qualitatively investigate adolescents’ perspectives on factors influencing the anticipated uptake and expected engagement with transdiagnostic preventive DMHIs. Using a hybrid deductive-inductive approach, we identified factors operating within and across person, intervention, and system levels. Beyond level-specific considerations, trust, support, and discrepancies between adolescents’ expectations and implementation possibilities emerged as important cross-level themes shaping how preventive DMHIs are perceived and how they may be used.

The findings of this study suggest that adolescents’ decision to engage with a preventive DMHI begins well before they encounter the intervention itself. Person-level factors were particularly relevant during the awareness phase of uptake, highlighting the importance of mental health literacy and social contexts that facilitate recognition of and access to available support options. These findings align with previous research on help-seeking and DMHI engagement and correspond to the capability and opportunity domains of the COM-B model [2,3,61,62].

A noteworthy finding was that adolescents’ informational needs extended beyond general mental health literacy. Before considering uptake, they required concrete information about the intervention’s purpose, credibility, expected commitment, and access procedures. This suggests that uptake barriers may arise before first contact with the intervention and highlights the importance of targeted preuptake communication. Furthermore, the prominent role of peers and family members as sources of awareness suggests that adolescents’ social networks may represent an underused pathway for disseminating preventive DMHIs. However, because this reflects hypothetical rather than observed dissemination, relying on peer- and family-based promotion could also disadvantage adolescents with weaker social support or in more stigmatizing environments.

Perceived usefulness and anticipated stigma further shaped adolescents’ willingness to engage with preventive DMHIs. These findings are consistent with previous research identifying perceived usefulness as a key determinant of DMHI uptake [63] and align with the health belief model, which proposes that factors such as knowledge and previous experiences influence health-related decision-making by shaping beliefs about benefits, barriers, and the personal relevance of preventive interventions [57,64]. Thus, adolescents’ willingness to use DMHIs may depend not only on awareness of available interventions, but also on whether they perceive them as meaningful, applicable, and worthwhile in relation to their own needs and concerns.

Once adolescents considered using a preventive DMHI, their expectations shifted from whether to use the intervention to whether it met their expectations regarding quality, usability, and accessibility. A key finding was that participants evaluated preventive DMHIs against the standards of commercial digital applications rather than traditional health care services. Consequently, appealing design, intuitive navigation, and affordability appeared to function as prerequisites for uptake rather than features that actively promote engagement. While these findings are consistent with previous research on facilitators of DMHI engagement [3,12,16], they additionally highlight the importance of distinguishing between factors influencing initial uptake and those supporting sustained engagement.

Regarding content of the DMHI, participants further emphasized the importance of reducing cognitive burden during engagement. Consistent with previous research on usability and suitability [16], adolescents valued concise, accessible content and inclusive representation rather than demographic segmentation. Personalization and individualization were also considered important for enhancing engagement as in previous research [16,65], but only when combined with user control over data disclosure. This highlights that personalization cannot be considered independently from privacy and autonomy [66], a tension further reflected in the cross-level discrepancies identified in this study.

Beyond individual intervention characteristics, participants envisioned preventive DMHIs as integrated components of adaptive and low-threshold mental health care systems. A particularly noteworthy finding was that adolescents spontaneously described a stepped-care approach, including different levels of digital and professional support depending on individual needs, despite not having been introduced to this concept. This convergence with established stepped-care models [58-60] suggests that adolescents may intuitively recognize the value of flexible care pathways and provides further support for embedding preventive DMHIs within broader health care structures rather than positioning them as standalone tools. Participants also highlighted the potential value of integrated peer and human support options, suggesting that social connection may represent an important component of continuity within digital mental health care [60]. However, as participants favored passive over active peer exchange, such features would need moderation to safeguard against harmful peer advice and negative social comparison.

Notably, although the study focused on preventive, subthreshold DMHIs, participants frequently described expectations more characteristic of clinical or crisis-oriented services, including immediate professional availability and emergency support. This may reflect difficulties in conceptualizing prevention as an early support strategy and in distinguishing preventive support from early intervention, treatment, and crisis care, extending existing concerns about mental health literacy [67,68] to the domain of digital prevention. Clear communication about the purpose and boundaries of preventive DMHIs may therefore be essential to support appropriate uptake and engagement. In particular, preventive DMHIs may benefit from psychoeducational information that clarifies what the intervention can provide, as well as clear pathways for accessing professional or crisis-oriented support when needed.

Beyond factors operating within individual levels, trust and support highlighted that the willingness to try and engage with preventive DMHIs emerges through interactions between person-, intervention-, and system-level factors. Although initially identified through specific intervention expectations, both concepts extended beyond the boundaries of a single analytical level. Trust was not merely a feature of the intervention but a prerequisite for engagement shaped by adolescents’ willingness to disclose information, perceptions of transparency and privacy, and confidence in the institutions delivering the intervention. Similarly, support was not perceived as an isolated intervention component but as an experience emerging from the interaction between individual needs, intervention characteristics, and available care structures.

Interestingly, while both human- and AI-supported DMHIs have been shown to enhance engagement [12,19,69], participants in this study expressed a clear preference for human support. This preference suggests that adolescents may not perceive support solely as an intervention feature intended to increase engagement, but also as a relational component providing reassurance, validation, and a sense of being taken seriously. For preventive DMHIs, this highlights the importance of clearly defining and communicating the role of human involvement: while access to professional support may strengthen perceived safety, unlimited availability of individual professional contact may conflict with the preventive and self-guided nature of these interventions. Clear boundaries around the availability and role of human support, together with defined escalation pathways for situations requiring more intensive care, may therefore be important to avoid unrealistic expectations of continuous or 24/7 professional support. These findings suggest that trust and support cannot be achieved through intervention design alone but also depend on transparent communication and credible integration within trusted health care structures. Future development should also consider the operational feasibility of human support, including staffing requirements, legal liability, and sustainable funding mechanisms, as key implementation questions for integrating human support into preventive DMHIs.

This study identified three recurring tensions between adolescents’ expectations of preventive DMHIs and the practical realities of implementation: the desire for personalized support while minimizing personal data disclosure, the wish to benefit from peers’ experiences without active self-disclosure, and the expectation of immediate human support despite limited intended engagement with the intervention. Rather than representing contradictions, these findings reflect competing priorities that require careful balancing during intervention development.

These findings have important implications for the development of future DMHIs and their implementation. Clear communication at the point of uptake regarding the intervention’s purpose, expected level of engagement, and handling of personal data may help align users’ expectations with the preventive aims of the intervention. Similarly, the balance between anonymity and personalization should be guided by the primary objective of the intervention, as greater anonymity may reduce barriers to uptake whereas more personalized support may require increased disclosure [70]. Participants’ own implementation suggestions reflected this tension directly, ranging from anonymity-related requests such as optional data provision and the use of nicknames to personalization-related requests such as customizable privacy settings within individualized profiles (Multimedia Appendix 5). The expectation of continuous human support warrants particular attention because it conflicts with the preventive and self-guided nature of DMHIs. Preventive interventions should therefore clearly communicate their scope and limitations, provide transparent pathways to higher-intensity care when needed, and position human support as an escalation pathway rather than a substitute for professional mental health services. Specifically, these findings informed the subsequent design of the EMPATIA intervention. Adolescents’ need for concrete information before uptake informed the inclusion of clear and prominently placed information about the intervention’s content, duration, and provider/source. Their preference for accessible human support was addressed by the inclusion of human e-guidance. Furthermore, participants’ preferences for individualized support informed the use of self-assessment exercises at the beginning of modules to support the selection of relevant modules, strategies, and examples.

Several limitations should be considered when interpreting these findings. First, the sample was predominantly female (73%) and highly educated, reflecting a self-selecting sample of adolescents willing to participate in mental health research. Consequently, the findings may not generalize to male adolescents, those with lower educational attainment, migrant or minority-language backgrounds, or adolescents with little prior experience of mental health support, with no or little interest in mental health topics, or with low trust in health or academic institutions. Furthermore, race, ethnicity, migration background, and socioeconomic status were not assessed, and eligibility was restricted to adolescents residing in Switzerland who understood and spoke German. This may have excluded the perspectives of multilingual, migrant, or otherwise minoritized youth. In addition, although age and gender were recorded, participants may have differed substantially in developmental stage, autonomy, privacy expectations, help-seeking, and parental involvement, which may have shaped their interview responses. However, the sample composition did not allow meaningful exploration of age-related or other subgroup differences, including the potential influence of intersecting social identities. Moreover, prior mental health symptoms and help-seeking experiences were not related to the qualitative findings, limiting interpretation of whether participants’ perspectives varied according to their prior mental health experience. Future research should therefore include more diverse samples and examine developmental differences in participants’ perspectives.

Second, the limited prior experience of the master’s student interviewers in qualitative research methods represents a potential limitation. Although this was mitigated through structured preparation, pilot interviews, reflexive discussions, and ongoing expert supervision, future studies may benefit from involving interviewers with more extensive qualitative research experience.

Third, shared assumptions about digital engagement between the master’s student interviewers and participants may have limited the exploration of certain topics during follow-up questioning. Furthermore, the all-female research team may have influenced data interpretation. These potential influences were addressed through reflexive diary-keeping, regular discussions within the research team, and repeated review of coding and interpretations.

Finally, the semistructured, theory-derived interview guide and initial coding framework may have influenced participants’ responses by directing attention toward elements of the research team’s proposed intervention and established theoretical frameworks [36]. Although the analysis allowed inductive themes to emerge, the influence of these predefined starting points cannot be excluded. Moreover, this study focused on the discovery phase of co-design rather than involving adolescents throughout the full design process. Future co-design efforts should therefore involve adolescents throughout all stages of intervention development and evaluation [27,30,31,39].

This study identified person-, intervention-, system-, and cross-level factors shaping adolescents’ anticipated uptake and expected engagement with preventive transdiagnostic DMHIs during the predesign phase of intervention development. At the person level, targeted preuptake communication addressing adolescents’ informational needs and leveraging social networks may facilitate uptake. At the intervention level, high-quality design and a balanced approach to personalization that preserves user control over data disclosure may support both uptake and sustained engagement. At the system level, adolescents’ descriptions of adaptive, low-threshold, and integrated mental health services highlight the importance of embedding preventive DMHIs within broader care pathways rather than positioning them as standalone tools. Beyond these level-specific findings, participants’ difficulties conceptualizing prevention suggest that implementation should be accompanied by psychoeducational efforts that clarify the purpose of preventive DMHIs and distinguish them from crisis and treatment services. Moreover, the tension between adolescents’ desire for immediate human support and the preventive, self-guided nature of these interventions highlights the need for clear clinical models defining the role and limits of digital preventive support. These findings should be understood as generating implementation hypotheses to be tested in subsequent phases of intervention development. Furthermore, they emphasize that successful implementation of preventive DMHIs requires coordinated attention to adolescents’ needs, intervention design, and the wider health care context.