We conducted a double-blind, parallel-group randomized controlled trial (RCT) with a superiority hypothesis and equal allocation (1:1) to a digital mental health self-care service (DMHSS) or a waiting-list control. Assessments were performed at baseline and 1 month after the intervention. The trial protocol was preregistered on the Open Science Framework (6AQSB), and the reporting follows the CONSORT-EHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth) extension [22]. An overview of the trial procedures is shown in Figure 1.

Members of the university community were invited from May to June 2024 to complete a baseline mental health screening via the self-care platform: professors (n=444), administrative staff (n=266), and students (n=7710) at a public Peruvian Andean university. Those aged ≥18 years who responded to the pretest and exhibited mild to moderate symptoms of depression, anxiety, or perceived stress were eligible for inclusion. Exclusion criteria were refusal of virtual informed consent; lack of prior experience with any virtual health service; being on leave or having reserved enrollment during the study period; current psychotherapy or psychopharmacological treatment for depression, anxiety, or stress; a prior diagnosis of severe mental health pathology; or scoring in the severe range on the baseline assessment.

We refer to our self-administered digital intervention throughout as the DMHSS, branded as “Amyga” for the student interface [23]. The DMHSS was developed and overseen by a clinical psychologist specializing in cognitive-behavioral psychotherapy (CBT) to ensure fidelity to evidence-based principles. It comprises 6 self-administered, 5-day modules grounded in a contextual-behavioral model (Figure 2). Each module combined 3‐4 brief (5‐8 min) videos (with matching audio tracks), 20‐30 daily text prompts, and a 3‐ to 8-page PDF workbook, all adaptively tailored to users’ baseline Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder-7 (GAD-7) scores. The modules were organized into 3 pairs by psychotherapeutic technique: modules 1 and 4 applied acceptance and commitment therapy (ACT) to foster emotional acceptance and clarify personal values [24]; modules 2 and 5 used mindfulness practices to cultivate nonjudgmental present-moment awareness and stress tolerance [25]; and modules 3 and 6 targeted experiential avoidance through gradual exposure and behavioral activation to build resilience and engagement in meaningful activities [26]. Detailed content and resource quantities for each module are provided in Table S1 in Multimedia Appendix 1.

In each module, the video icon denotes short tutorials that blend theoretical background with practical examples; the daily-message icon marks brief prompts sent each day to reinforce learning and encourage reflection; the document icon indicates downloadable documents with step-by-step instructions and written exercises; and the audio icon represents narrated versions of the tutorials and exercises for offline or hands-free use.

We reinforced engagement by sending SMS or WhatsApp reminders to participants at the beginning, middle, and end of each module. In addition, reminders were sent to participants who did not complete the activities in the estimated time, in order to encourage compliance with the established deadlines. Control-group participants were placed on a 30-day waiting list; after this period, they received full access to the DMHSS. All participants also received information on standard university mental health services.

In collaboration with the university welfare department, we ran an advertising campaign to recruit students, teachers, and administrative staff via campus announcements and online registration. Eligible individuals were then randomized to the intervention or waiting-list control. The intervention group engaged with the 6-module DMHSS over the next 30 days, while control participants remained on the waiting list and received access afterward. Both groups completed the postintervention assessment immediately on day 30 using the same online instruments.

Continuous variables are presented as mean (SD) or median (IQR); categorical variables as frequencies and percentages. Normality was assessed with Shapiro–Wilk, and independent-samples t tests were applied when appropriate. To isolate the effect of the intervention on posttest scores while accounting for baseline differences, we performed ANCOVA, which adjusts each follow-up outcome for its corresponding pretest value. Intervention effects were quantified with Cohen d (0.20=small; 0.50=moderate; 0.80=large) and partial epsilon-squared (ε²_p) (0.02=small; 0.15=moderate; 0.30=large) for intuitive purposes [32]. All analyses were performed in RStudio (Posit PBC) and Stata 17 (StataCorp LLC).

The study protocol was approved by the Universidad Nacional Santiago Antúnez de Mayolo Research Ethics Committee (report numbers 011‐2022 and 07‐2023). All participants provided electronic informed consent before taking part and were free to withdraw at any time without penalty. No financial or other compensation was offered for participation. To protect privacy and confidentiality, we assigned alphanumeric codes to each record, stored data on secure sockets layer–encrypted servers, and removed all personal identifiers after analysis in compliance with Peru’s data protection law (Number 29733). The study was conducted in accordance with the World Medical Association’s ethical principles [33].

This pilot study assessed the effectiveness of a digital mental health self-care intervention grounded in a contextual-behavioral framework that included acceptance and commitment techniques, experiential avoidance, and mindfulness among students, teachers, and administrative staff at a public university in Peru. Our results suggested that participants in the intervention group experienced a significant reduction in symptoms of depression, anxiety, and stress compared to the control group. Although most participants exhibited high commitment to the platform, satisfaction and usability ratings were low.

Several studies have reported heterogeneous success rates of virtual platforms for mental health and varying user perceptions [34], which in some cases contrasts with our findings. These differences may stem from the therapeutic techniques and approaches used, the duration and format of the intervention, the digital skills of the users, the cultural appropriateness of the intervention, and support personnel.

First, our intervention was grounded in a contextual-behavioral framework, using ACT and mindfulness techniques. Prior studies have demonstrated the effectiveness of these techniques in digital environments. One study assessed an online intervention combining mindfulness and CBT techniques, which showed reductions in depression and anxiety among university students but no significant impact on perceived stress [35]. Another online ACT-based intervention among university students found significant improvements in depression, anxiety, and stress [36]. Together, this evidence suggests that, although mindfulness and CBT techniques effectively address depression and anxiety, complementing them with ACT may be necessary for effective stress management in the university population.

Intervention duration and intensity also influence outcomes. For example, an evaluation of a 6-module digital mental health program found no significant effects on depression and anxiety symptoms [37]. The lack of effectiveness was attributed to low participation and adherence rates, as well as potential issues with intervention delivery. Conversely, another study has indicated that longer interventions (>8 weeks) tend to be more effective in reducing depression and anxiety symptoms [38]. This suggests a need to balance the length of DMHI, tailoring them to specific contexts to achieve effectiveness without hindering user participation. In our study, the reduced module count and diverse content likely supported engagement and contributed to observed mental health improvements, despite overall low completion rates.

Various DMHI formats exist for university students; among these, internet-based interventions are the most commonly used, and our study confirmed their effectiveness in reducing depression and anxiety. However, interactivity and usability are crucial for ensuring this effectiveness [39]. Although our virtual platform featured a responsive design for mobile access, reported satisfaction and usability levels were low, suggesting it failed to provide an optimal user experience to foster participation and maximize benefits. Limited digital skills likely further hindered engagement, as some struggled with navigation, reducing the effective “dose.” In the Peruvian university context, teachers typically possess basic digital skills, and students generally have intermediate skills [40]. These findings partially explain the challenges faced by participants in using the DMHSS, contributing to their low levels of interest, participation, and overall satisfaction.

Several mechanisms may explain why participants maintained high engagement yet rated usability and satisfaction poorly. DMHI may unintentionally create cognitive overload when many modules are displayed simultaneously, a phenomenon linked to “digital fatigue” and reduced app acceptability among university students [41]. Reviews further show that static, nonpersonalized content lowers perceived relevance, whereas adaptive pathways that tailor material to baseline symptom severity improve adherence and satisfaction [42]. Finally, the absence of real-time feedback loops, such as progress indicators or light gamification, diminishes users’ sense of accomplishment; interventions that embed these features consistently achieve high usability scores [43].

In light of this evidence, future platform iterations should streamline navigation via progressive disclosure, incorporate rule-based personalization, and add succinct progress feedback or gamified cues. These refinements, grounded in contemporary usability research, warrant formal A/B testing to quantify their impact on satisfaction, usability, and mental health outcomes, particularly in low-resource university settings. In addition, exploring the incorporation of asynchronous peer-support chats could further assist users with low digital literacy without compromising the autonomy of the intervention.

It has also been emphasized that cultural and contextual aspects should be included in designing DMHI for university students to improve acceptance and effectiveness [44]. In our study, formative studies that assessed the expectations and preferences of students, teachers, and administrative staff regarding DMHI facilitated the intervention design. Including contextually relevant examples and content adapted to the reality of participants likely contributed to the high level of commitment observed despite technological usability or satisfaction challenges.

Moreover, some studies show that support personnel can enhance DMHI effectiveness compared to purely self-administered formats [45], though others find no significant effect [46]. Although our intervention was self-administered, administrative staff provided reminders and technical support, underscoring the value of human support. This highlights the importance of support even in self-administered interventions, particularly among users with procrastination tendencies, which have been shown to reduce the adherence to digital interventions [47], and in contexts of low digital literacy and persistent mental health stigma, such as observed in the Andean region of Peru [7].

Considering the broader applicability of the DMHSS, future integration into existing university mental health services could enhance its impact. Positioning this platform as a first-line self-guided option within a stepped-care model or as an adjunct to traditional counseling services may address unmet mental health needs on campuses. Effective implementation, however, will require adequate IT infrastructure, minimal staff training, clear data-privacy protocols, and consideration of potential barriers such as competing academic demands and limited technical support. Evaluating these factors through future pilot programs is critical to ensuring scalable adoption and sustainability.

Several limitations warrant careful consideration. First, attrition was high: only about 20% of intervention participants completed all modules, which narrowed our analytic sample and may limit the representativeness of the findings (external validity). However, the subset who completed the intervention remained balanced at baseline (Table 2) and yielded sufficient power to detect statistically and clinically meaningful between-group differences, suggesting that internal validity was preserved despite the high dropout. Second, completion rates differed between arms (15% intervention vs 25% control), raising the possibility of differential bias. Nevertheless, baseline characteristics (including age, sex, and initial symptom severity) were equivalent across groups (Table 2), suggesting that this imbalance is unlikely to compromise internal validity. Third, satisfaction and usability scores were low, particularly among administrative and teaching staff, underscoring the need for interface refinements. Fourth, all outcomes were self-reported, introducing potential response and social-desirability bias, though validated instruments and blinded allocation mitigate this risk. Fifth, outcomes were measured only immediately postintervention; thus, durability is unknown and longer follow-up is needed. Sixth, we did not conduct qualitative interviews or focus groups; specific cultural and technological barriers were not explored in depth. Future research will incorporate semistructured interviews and user feedback sessions to address these gaps. Seventh, although we report Cohen d and ε²_p to convey effect magnitude, we did not prespecify or validate minimal clinically important difference thresholds for these instruments, which would aid interpretation of clinical relevance; future work should incorporate such benchmarks. Finally, the study occurred at a single public university in the Andean region, limiting generalizability, although the setting represents typical low-resource universities in the region.

Balanced against these constraints are several noteworthy strengths. The RCT design minimizes selection bias and supports robust causal inference. The intervention is multicomponent and theory-driven, combining ACT, mindfulness, and experiential-avoidance principles that have strong empirical support. In addition, this study extends the scant literature of DMHI in Latin American low-resource contexts [48], and is to our knowledge, one of the first to include teaching and administrative staff alongside students in the effectiveness of digital interventions in mental health in teaching staff, enlarging the scope of evidence on campus mental health of this population [49].

Future studies should evaluate long-term intervention effects with larger samples and replicate research across diverse universities and cultural contexts, including multicountry RCT [50] to assess feasibility, sustainability, and scalability. They should also examine factors influencing participation and engagement, the impact of personalized digital interventions tailored to individual needs, and ways to optimize usability via user-centered, minimalist design, iterative interface evaluation, and enhanced prototypes. It is also recommended to investigate the determinants of low satisfaction, including discrepancies between user expectations and offered functionalities, and the potential role for complementary human support. Future research should use rigorous methodologies such as A/B testing to quantify the effects of these design enhancements on usability, satisfaction, and mental health outcomes in low-resource university settings. Finally, because limited digital literacy may have attenuated our observed effects, future iterations should include targeted digital skills training or onboarding modules to ensure all users can effectively navigate and benefit from the platform.

In conclusion, this pilot study provides preliminary evidence that a multicomponent intervention of a digital self-care intervention based on acceptance and commitment techniques, experiential avoidance, and mindfulness effectively reduces depression, anxiety, and stress among university populations. This highlights the potential of digital interventions to address mental health problems in populations where access to traditional health services is limited. However, intervention format, content, duration, and level of digital skills of users must be carefully considered to ensure adequate usability and satisfaction, promoting future scalability and sustainability.