Most people can recall an embarrassing moment in secondary school, perhaps making a mistake in front of their crush or feeling excluded during a group project. These experiences are common, as adolescence is a developmental period characterized by heightened emotional sensitivity and increased stress reactivity [1]. This rise in stress is associated with intensified negative emotional responses and a reduced ability to regulate emotions effectively [2]. As a result, adolescence is considered a critical period for the onset of mental health disorders, particularly generalized anxiety and social anxiety disorder [2]. Social anxiety is one of the most prevalent anxiety disorders among adolescents [3,4] and is characterized by an intense fear of social situations, anticipation of negative evaluation, and fear of humiliation, often leading to avoidance and withdrawal [5,6]. These vulnerabilities are further amplified during periods of societal disruptions, such as economic crises and public health emergencies, which can disrupt social routines, increase uncertainty, and limit access to support [7]. The COVID-19 pandemic is a salient recent example, having significantly increased rates of anxiety and social difficulties among adolescents [8]. As large-scale disruptions have become increasingly prevalent, there is an urgent need for effective and accessible interventions to help adolescents manage stress and anxiety. Importantly, such interventions may be especially valuable early in life, aiming to prevent the progression of subclinical symptoms into more severe anxiety disorders.

Traditional therapeutic approaches, such as cognitive behavioral therapy and exposure-based interventions, have been shown to effectively reduce anxiety [9]. However, these in-person treatments present several barriers to accessibility, including limited awareness of symptoms among adolescents, parents, and teachers, as well as stigma, long waiting lists, high costs, and transportation difficulties [10-18]. As a result, many adolescents who could benefit from support do not access it. Digital mental health interventions (DMHIs) offer a promising solution by providing scalable and easily accessible mental health support. However, most DMHIs currently available on the market lack evidence-based treatments, making them potentially harmful [19].

The modes that fall under DMHI are, for example, internet programs, mobile apps, and virtual reality. DMHIs provide the opportunity to autonomously, independently, and with high accessibility get in touch with interventions that would usually be out of reach [20]. In this study, we focus on 2 specific interventions, that is, slow breathing (SB) and detached mindfulness (DM). They were selected for their strong theoretical and empirical foundation in regulating stress and anxiety [21,22]. Both interventions address key mechanisms implicated in social anxiety, which include maladaptive metacognition and dysregulated autonomic function [23,24], while being well-suited for digital implementation. DM, derived from metacognitive therapy, encourages individuals to observe their thoughts without engaging with them, reducing excessive self-focus and negative metacognitive beliefs [24]. SB directly targets autonomic regulation by increasing heart rate variability (HRV), a key physiological marker of emotional resilience [25,26]. By promoting a breathing pattern with prolonged exhalation, SB enhances parasympathetic activity, fostering relaxation and reducing physiological arousal in stressful situations [27].

Research has demonstrated their effectiveness in both immediate and long-term stress reduction. For example, a recent review of diaphragmatic breathing studies in children and adolescents (6-18 y) found that breathing interventions alone or combined with other therapeutic methods significantly lowered perceived stress, and in many cases also reduced anxiety and depression symptoms over the course of multiweek programs [28]. Moreover, a video-delivered slow diaphragmatic breathing curriculum, designed for high-school students and delivered 3 times per week over 5 weeks, also proved feasible and well-tolerated; students rated the program as useful and easy to follow [29]. Similarly, a systematic review of mindfulness-based interventions for adolescents found consistent improvements in cognitive performance, resilience to stress, and overall well-being, with intervention durations typically ranging from 4 to 16 weeks. Effect sizes for stress reduction ranged from moderate to substantial [30].

Recently, both interventions that we test in this study were evaluated in a healthy sample of 120 young adults using a randomized experimental design, in which participants completed a digitally delivered intervention immediately before performing a public speaking task, an established social stressor [31]. Both DM and SB significantly reduced stress and state social anxiety compared with a control condition. The DM intervention also significantly reduced negative metacognitive beliefs compared with the control condition. Moreover, both interventions were rated as highly acceptable in their digital format, supporting the feasibility of brief, app-based delivery [31]. These characteristics, along with their structured, self-guided nature and standardized delivery protocols, make both interventions particularly suitable for digital implementation. While the previous study established the immediate efficacy of both interventions under controlled laboratory conditions, this study extends their effectiveness and acceptability in adolescents during daily life, using a longitudinal design. This allows us to explore how adolescents engage with the interventions in naturalistic settings, and how they perceive their usefulness, usability, and relevance. These insights are critical for tailoring future versions of the interventions to better fit the specific needs and preferences of this age group.

A major challenge in DMHIs is maintaining user engagement and adherence. High dropout rates and low adherence can significantly undermine the effectiveness of these interventions, limiting their real-world impact [32]. While DMHIs offer accessible and evidence-based mental health support, many struggle to fully engage adolescents, leading to high attrition rates [33]. This disengagement suggests that even if an intervention is effective, its benefits may not be fully realized if users disengage prematurely. Therefore, it is crucial to understand the factors that contribute to attrition and identify strategies to enhance user engagement. Adolescents may disengage due to a lack of motivation, difficulty integrating the intervention into their daily routines, or a perceived lack of immediate benefits. Therefore, in this study, we did not only examine the effectiveness of SB and DM but also explore how adolescents experience and interact with these interventions.

This mixed methods study aimed to assess the feasibility and preliminary effects of a 1-week digital intervention using SB or DM in reducing stress, social anxiety levels, and maladaptive metacognitive beliefs in late adolescents compared with a passive control. We hypothesized that both interventions led to a significant reduction in social anxiety, stress, and maladaptive metacognitive beliefs from baseline to the 1-week posttreatment measurement compared with the control group. Additionally, we aimed to examine how these intervention effects were moderated by perceived usability and higher motivation invested in the interventions. Furthermore, we explored day-to-day changes in stress, anxiety, and negative thoughts following each intervention session.

Beyond the quantitative assessment, we aimed to gather late adolescents’ perspectives on key aspects of the interventions. Specifically, we explored (1) whether the interventions are feasible and fit into their daily routines, (2) their motivation to engage with the interventions consistently, and (3) their perceptions of the interventions’ enjoyment and clarity. Additionally, we identified which intervention features are most appealing, which aspects are less engaging, and what improvements could enhance their usability and its effectiveness. Furthermore, we explored how to optimize their implementation in a digital format.

The study used a mixed method approach [34]. For the quantitative part, a randomized controlled trial was used, with a 2×3 within or between-subject design. Participants went through 2 phases in the experiment (within-subject variable)—the baseline and postintervention phases. Additionally, participants were randomly allocated to one of the following 3 different conditions (between-subjects factor): DM, SB, or the passive control condition. For the qualitative part, participants took part in a focus group in which they could share their thoughts and cocreate new perspectives that might otherwise be hidden [35]. There were 4 focus groups per intervention condition (8 total). Only the 2 intervention conditions took part in the qualitative phase; all 3 were included in the quantitative phase. As our study was also used as input for a study on the experiences with online and live focus groups, both formats were organized (in Microsoft Teams and in the university building), and participants were randomly allocated to one of them.

University students were recruited from Leiden University via the online recruitment platform and flyers. Inclusion criteria were: (1) aged 16‐19 years, (2) enrolled as first-year Bachelor students at Leiden University, and (3) proficient Dutch comprehension (C1 level). Exclusion criteria were (1) current or past psychological or neurological disorders, and (2) uncorrected vision problems. A total of 232 participants met the eligibility criteria and provided informed consent (mean age 18.34, SD 1.03 y; n=19, 8.05% male).

Analysis in G*power showed that we needed a sample of 114 participants in total (38 participants per condition; assuming repeated measures, within-between interaction, a small to moderate effect size (f=0.18), and a power of 0.80). We intentionally overrecruited to account for expected attrition [36-39].

Participants were randomly allocated to 1 of the 3 conditions (49 per group). Random sequence generation was performed using a computerized procedure before data collection, with participants assigned to a predetermined random order upon sign-up. Allocation was concealed until after baseline questionnaire completion. If they were assigned to one of the intervention conditions, the same email also contained the day 1 intervention link. Due to the nature of the interventions, participant blinding was not possible; however, data analysts were blinded to condition assignment. Those who did not complete the baseline questionnaire were excluded. Both intervention groups practiced their assigned exercise for 7 days. Those who completed at least 6 days were invited to complete the posttreatment questionnaire. Analyses were conducted on a per-protocol basis, resulting in a final analytical sample of 114 participants (retention rate 77.6%). Participants in the intervention groups who completed the postquestionnaire were subsequently invited to a focus group. A total of 37 participants participated across 8 focus groups with group sizes ranging from 3 to 8 participants.

The study was approved by the Psychology Research Ethics Committee at Leiden University (2024-10-08-R. Akdag-V2-5666). The research procedure complies with the ethical standards of the national and institutional committees and is in accordance with the Declaration of Helsinki. Participants received information about the study’s purpose, procedures, risks, and benefits, and could withdraw at any time. Data were anonymized before analyses, and identifiable information was securely stored, accessible only to authorized personnel. Deidentified datasets were used for analyses to protect privacy. Participants were gifted 6 credits in total if they completed the pre- and postmeasurements, the intervention, and the focus group. The hypotheses, preprocessing of the data, and statistical analyses of the study were preregistered on OSF.

After providing informed consent, participants received a link to the premeasurement questionnaire via email, which they were asked to complete within 1 week using Qualtrics. Upon completion of the premeasurement, participants were informed about the condition to which they had been assigned. Participants in either the DM or SB condition received a daily email for 7 consecutive days containing a link to the online intervention and a brief daily questionnaire. After completing the intervention for at least 6 days, participants received a link to the postmeasurement Qualtrics questionnaire. The pre- and postmeasurements took participants approximately 20 minutes to complete. A suitable date for the focus groups was selected using a date picker tool.

A total of 232 potential participants were assessed for eligibility, with 147 meeting the criteria and being enrolled in the study. A detailed overview of participant enrollment is illustrated in Figure 2. To assess whether randomization was successful, we compared demographic variables and trait measures across conditions. There were no significant differences between conditions in gender, age, and education level (all P values>.338). No differences were observed in baseline levels of anxiety and negative metacognitive beliefs across all conditions (all P values>.134). However, we did find a significant difference between the conditions on stress scores (P<.001). Specifically, the participants in the DM condition scored higher on stress than both the control and SB conditions. On average, participants scored 59.98 on overall stress, which is below the midpoint of the scale (81) and suggests relatively low to moderate levels of self-reported stress in this nonclinical sample. Additionally, participants scored average (mean 22.38, SD 4.81) on anxiety levels and negative metacognitive beliefs (mean 14.09, SD 4.09).

Participants in both intervention groups rated the intervention as highly usable (mean_SB=36.63, SD 4.70 and mean_DM=33.2, SD 6.12 on the SUS) and moderately acceptable (mean_SB=17.3, SD 4.59 and mean_DM=16.67, SD 4.04 on the TAM). On average, participants in the SB group gave the intervention a grade of 7.1 out of 10 and the participants in the DM group a 6.5. The majority (n=67, 89.2%) provided a positive evaluation, describing the intervention as calming, helpful in reducing stress, and easy to follow. A smaller subset (n=9, 10.8%) expressed negative views, noting that the intervention felt repetitive, was difficult to concentrate on, or was not personally effective. A full overview of descriptive statistics is provided in Multimedia Appendix 4.

Regarding recruitment, of the 232 individuals assessed for eligibility, 147 completed the pretest, a recruitment rate of 63.4%. Although this falls below the benchmark of 70% [49], it is consistent with a systematic review and network meta-analysis indicating that attrition is a common challenge in internet-based interventions [50]. Of the 147 who completed the pretest, 114 also completed the postmeasurement, resulting in a retention rate of 77.6% that exceeds the 70%-76% benchmark [49]. Adherence to the daily exercises was high in both intervention conditions. In the SB group, participants completed on average 6.3 of the 7 days, with 88% (n=33) completing at least 6 days. In the DM group, participants completed on average 6.0 days, with 78% (n=29) completing at least 6 days. Finally, no technical problems were reported during any intervention, a technical issue rate of 0%.

The aim of this study was to investigate feasibility and preliminary effects of brief digital SB and DM interventions on stress, anxiety, and negative metacognitive beliefs in youth. In addition, we explored participants’ subjective evaluations of the interventions to inform future design decisions tailored to this target population. Our quantitative results showed that a week-long DM intervention significantly reduced stress compared with the control condition. No other between-group differences emerged for the main outcomes on pre- and postmeasurements. However, in the SB group, higher perceived usability was associated with greater reductions in stress postintervention. Although there were no clear improvements across the 7-day intervention period overall, day-level analyses revealed that participants who exerted more effort experienced greater benefits from the intervention.

As hypothesized, participants in the DM condition experienced a significant reduction in self-reported stress compared with the control group after 1 week. The stress-reducing effect of the DM condition aligns with previous work showing that brief metacognitive strategies can reduce emotional reactivity in stressful situations [41]. In the earlier laboratory-based study with young adults [31], the same intervention reduced stress and social anxiety immediately following a social-evaluative task. This study extends those findings to a real-world adolescent sample, suggesting that even brief, self-guided digital training in DM can lead to noticeable reductions in perceived stress across a week.

The qualitative findings help contextualize and strengthen this result. Late adolescents in both intervention conditions frequently described experiencing a sense of calm or temporary relief from stress following the exercises. These brief, guided moments of stillness and sensory focus appeared to offer a pause in participants’ busy daily lives, supporting the idea that even low-intensity digital interventions may offer a subjective sense of relief when actively engaged with. This convergence between quantitative stress reduction and qualitative reports of calm suggests that the interventions were experienced as emotionally beneficial, even when broader symptom change was limited.

It should be noted that the baseline stress levels were relatively low to moderate, raising questions about the practical significance of the observed stress reduction. The findings are therefore best interpreted as proof-of-concept evidence that the intervention can produce measurable stress reduction under naturalistic conditions, rather than as evidence of clinically meaningful change. Future research should examine the intervention in samples with elevated baseline stress levels to better assess the potential practical impact of the mindfulness condition.

Despite the effects on stress, no intervention effects were observed for anxiety or negative metacognitive beliefs compared with the passive control group. One explanation may lie in the short intervention duration. Meta-analyses suggest that longer programs (eg, 4 weeks or longer) yield stronger and more consistent effects on anxiety and cognitive symptoms [51]. In our case, the 1-week period may have been insufficient to shift deeper cognitive patterns, such as maladaptive metacognitive beliefs. Alternatively, the rather moderate baseline levels of stress of our sample may have reduced the effectiveness of the intervention. A marginal trend toward reduced negative metacognitive beliefs in the DM group provides a promising direction for future studies using longer timelines or higher-intensity delivery formats.

Additionally, the qualitative data showed that while participants often described the exercises as calming, many also reported difficulties with concentration, repetitiveness, or a mismatch between the intervention and their immediate emotional needs. For instance, some participants in the DM group reported that attempting to observe stress-related thoughts without reacting could initially increase distress, particularly when the technique was not fully understood or when stressors felt highly salient. These experiences suggest that late adolescents may require more structure, guidance, or scaffolding to benefit optimally from cognitive or metacognitive techniques, especially in short interventions.

In contrast to the DM, the SB intervention did not lead to significant changes in stress, anxiety, or metacognitive beliefs compared with the passive control group. Although previous studies have shown that slow diaphragmatic breathing enhances parasympathetic regulation and reduces psychological stress, these benefits are typically observed in studies using longer or more structured training programs [26,52-54]. This suggests that the brief, 1-week practice period in our study may have been insufficient for late adolescents to experience meaningful psychological improvements. Moreover, similar to the DM intervention, the lack of significant findings may reflect the relatively low baseline stress and anxiety levels in our sample, limiting room for improvement (ie, floor effects).

Adherence and engagement emerged as critical factors across both data sources. Quantitatively, greater effort and more favorable perceptions of usability, technology acceptance, and outcome expectancy were associated with better outcomes, particularly in the SB group. Late adolescents who found the intervention easier to use and who expected it to be helpful reported greater reductions in anxiety. In the DM group, higher technology acceptance predicted greater stress reduction, suggesting that perceiving the intervention as relevant or useful may be especially important for benefiting from metacognitive strategies.

These patterns were strongly echoed in the qualitative data. Participants frequently described forgetting to complete the exercises, doing them late at night as an afterthought, or rushing through them without full attention. Several participants explicitly noted that they benefited more on days when they were able to focus and less when distracted or unmotivated. This convergence highlights that engagement and effort were associated with better outcomes, potentially outweighing previous familiarity with digital interventions, which did not predict outcomes quantitatively. Rather than general digital experience, participants’ motivation, expectations, and perceived relevance of the specific intervention were associated with better outcomes.

Taken together, the integrated findings suggest that brief digital interventions show preliminary promise for reducing perceived stress in late adolescents, particularly when they are experienced as engaging, credible, and personally relevant. However, achieving broader changes in anxiety and metacognitive beliefs may require longer intervention periods, greater personalization, and enhanced support for motivation and adherence. Future implementations may benefit from adaptive features that tailor content to the user’s emotional state, provide timely reminders, offer brief check-ins to support, and create a sense of togetherness for sustained engagement. In sum, while the interventions showed promising stress-related effects, the mixed methods evidence underscores that real-world effectiveness in adolescent populations depends as much on engagement and context as on intervention content itself.

While overall symptom trajectories across the 1-week period were modest and sometimes flat, our daily analyses revealed that participant engagement, specifically the self-reported effort invested in each session, played a meaningful role in intervention effectiveness. In both, the SB and DM groups, greater total effort was associated with improvements in specific outcomes. For SB, higher effort was linked to greater reductions in daily stress and negative metacognitive beliefs, suggesting that consistent engagement with this relaxation-based intervention can yield emotional benefits, even when average group-level change is limited. Similarly, for DM, effort was associated with lower daily anxiety, indicating that participants who applied themselves more fully to the metacognitive exercise experienced stronger short-term gains. These findings underscore the importance of designing interventions that not only appeal to youth but also sustain motivation and involvement over time.

Several limitations must be acknowledged. First, the 7-day intervention period was relatively short, which may have been insufficient for SB to produce measurable psychological effects. In addition, the absence of follow-up assessments leaves it unclear whether the benefits of DM would persist over time. Second, the passive control condition did not include any matched activity or engagement, making it impossible to rule out nonspecific effects such as expectation, structured daily routine, or simply taking a brief pause. Third, attrition was substantial, which may limit the generalizability of our findings and reflects the ongoing challenge of maintaining engagement in digital interventions. Also, our participants all being university students with 92% (n=104) female limits the generalizability of the results. Fourth, while focus groups provided valuable qualitative insight, participants may have been influenced by group dynamics, potentially limiting the generalizability of their responses. Fifth, baseline stress levels differed in the DM group compared with the SB and control group. Although the sensitivity analysis confirmed that the stress reduction effect remained significant after controlling for baseline stress, the small effect size and regression to the mean cannot be entirely excluded as a contributing factor. Sixth, all outcomes relied on self-report, and daily measures used 3-point scales which may have lacked sensitivity to detect nuanced change. Seventh, participants who completed fewer than 6 intervention days were excluded from the analyses. If these noncompleters differed systematically from completers, this may have introduced selection bias toward a more motivated or engaged sample. Finally, many of the participants experienced a relative delay between completing the intervention and filling out the postintervention questionnaire, which could potentially have influenced results, as circumstances could have changed between these periods.

Future studies should build on these findings by testing enhanced and adaptive DMHI designs in larger, more diverse samples and samples with elevated stress levels. Additionally, future studies should include an active control condition matched for time, digital format, and level of participant contact to allow for stronger causal inference regarding intervention-specific effects. Integrating personalization features and extending the intervention period could help sustain engagement and maximize effectiveness. Moreover, combining self-report with objective measures of stress and intervention use would strengthen the evidence base. Finally, given the documented rise in mental health concerns among adolescents [55,56], if replicated in larger and more diverse samples, DMHIs may offer a scalable approach to promoting youth well-being in educational settings.

In conclusion, this study provides early formative evidence supporting the feasibility and usability of brief digital SB and DM interventions in a nonclinical sample of late adolescents from a single center. A preliminary signal for short-term stress reduction was observed for DM, although the absence of effects on anxiety and metacognitive beliefs, combined with the short 1-week intervention duration, predominantly female sample, and lack of follow-up, underscores the need for replication in larger, more diverse samples before any conclusions about effectiveness can be drawn. Beyond outcome effects, qualitative findings suggest that late adolescents value accessibility, autonomy, and noncompetitive features, offering concrete guidance for future intervention development. Together, these findings underscore the need to move beyond designing solely effective exercises to also designing for motivation. Strategies, such as gamification, social accountability, adaptive feedback, or ecological momentary prompts, may enhance engagement and foster more consistent use. Although effects were modest and the study faced limitations such as a relatively short intervention period and an incomplete sample, the integration of quantitative and qualitative evidence provides a nuanced understanding of both mechanisms of change and user experience in a youth sample. By prioritizing usability and motivation, and by aligning intervention design with late adolescents’ unique preferences and needs, DMHIs hold substantial promise as scalable solutions for addressing the growing mental health challenges facing young people today.