Apr 26, 2026

Strengthening resilience in university students with a scalable digital intervention: a randomized controlled trial

Reserved DOI:
10.17504/protocols.io.x54v9qbjml3e/v1
  • David Daniel Ebert1,
  • Sophie Nestler2,
  • Marvin Franke3,
  • Harald Baumeister4,
  • Rocio Herrero5,
  • Esther Beier1,
  • Anna-Carlotta Zarski6,
  • Rosa Banos5,
  • Cristina Botella7,
  • Matthias Berking8,
  • Claudia Buntrock2
  • 1Psychology & Digital Mental Health Care, Department of Sport and Health Sciences, School for Medicine and Health, Technical University Munich, Munich, Germany.;
  • 2Institute of Social Medicine and Health Systems Research, Medical Faculty, Otto-von-Guericke-University, Magdeburg, Germany.;
  • 3Clinical Psychology and Psychotherapy, Friedrich-Alexander University Erlangen Nuremberg, Erlangen, Germany.;
  • 4Clinical Psychology and Psychotherapy, University of Ulm, Ulm, Germany.;
  • 5Departamento Psicología Básica, Clínica y Psicopatología. Universitat Jaume I (Spain).;
  • 6Division of eHealth in Clinical Psychology, Department of Clinical Psychology, Philipps-Universität Marburg, Marburg, Germany.;
  • 7Departamento Psicología Básica, Clínica y Psicopatología. Universitat Jaume I (Spain);
  • 8Department for Clinical Psychology and Psychotherapy, Friedrich Alexander University Erlangen-Nuremberg, Germany.
  • PROTECTLAB
david.daniel.ebert
Icon indicating open access to content
QR code linking to this content
Protocol Citation: David Daniel Ebert, Sophie Nestler, Marvin Franke, Harald Baumeister, Rocio Herrero, Esther Beier, Anna-Carlotta Zarski, Rosa Banos, Cristina Botella, Matthias Berking, Claudia Buntrock 2026. Strengthening resilience in university students with a scalable digital intervention: a randomized controlled trial. protocols.io https://dx.doi.org/10.17504/protocols.io.x54v9qbjml3e/v1
This is an unpublished protocol. If you are citing it in a manuscript, ensure the protocol is fully published upon manuscript publication. We reserve the right to publish the protocol if we identify it in a published research article.
License: This is an open access  protocol  distributed under the terms of the  Creative Commons Attribution License,  which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: March 26, 2026
Last Modified: April 26, 2026
Protocol  Integer ID: 313994
Keywords: Resilience, Mental Health, Mental Health Promotion, Digital Intervention, Prevention, Digital Therapeutics, digital resilience intervention, grounded digital resilience intervention, resilience in university student, role of digital resilience program, resilience across unselected student population, based resilience intervention, resilience intervention, digital resilience program, scalable digital resilience intervention, baseline resilience, strengthening resilience, resilience, davidson resilience scale, scalable digital intervention, controlled trial university student, prevention strategies in higher education setting, universal prevention approaches in routine university context, intervention effect, scalable intervention, perceived stress, trial university student, intervention condition, effects of an internet, psychological well, evidence on scalable intervention, efficacy, controlled trial design, prevention strategy, related indicators of mental health, mental health, high burden of mental health probl
Funders Acknowledgements:
EU Horizon 2020
Grant ID: 634757
BARMER
Grant ID: Studicare
Abstract
University students experience a high burden of mental health problems during a developmental period marked by major academic, social, and personal transitions. Although resilience has been identified as a potentially important protective factor, evidence on scalable interventions that can sustainably strengthen resilience across unselected student populations remains limited. In particular, little is known about the longer-term effects of theoretically grounded digital resilience interventions when delivered as universal prevention approaches in routine university contexts. The present study addresses this gap by evaluating the efficacy of a digital resilience intervention in an unselected sample of university students.
This study uses a randomized controlled trial design (N = 262) to examine the effects of an internet-based resilience intervention compared with care as usual. University students are randomly allocated to either the intervention condition or a control condition. The primary outcome is resilience at post-test, assessed with the Connor–Davidson Resilience Scale. Secondary outcomes include depressive symptoms, anxiety, perceived stress, affect, self-esteem, self-compassion, psychological well-being, and related indicators of mental health and functioning. Outcomes are assessed at post-treatment and 6 & 12 months follow up. Analyses will be conducted according to the intention-to-treat principle, with additional per-protocol analyses. Moderator analyses will examine whether intervention effects vary according to baseline resilience, participant characteristics, and contextual stressors. This trial will provide important evidence on the short- and longer-term effects of a scalable digital resilience intervention for university students. The findings are expected to inform the role of digital resilience programs as mental health promotion and prevention strategies in higher education settings.
Guidelines
This paper follows the guidelines of the CONSORT Statement as well as the Guidelines for Executing and Reporting Research on Internet Interventions (Proudfoot et al., 2011).
Before start
Participants were randomized only after informed consent was obtained, the screening completed, and no exclusion criteria were identified.
Background & Objectives
Rationale
The study aimed to evaluate the long-term effectiveness of a theory-based digital resilience intervention (CORE) in an unselected university student sample. Additionally, it examined whether intervention effects were moderated by baseline resilience, participant characteristics, and contextual stressors.
Hypotheses
Primary hypothesis: The intervention will lead to higher resilience at post-treatment (T2), assessed with the CD-RISC-25, relative to the control condition.
Secondary hypotheses: Compared with the control condition, the intervention will be associated with sustained improvements in resilience (CD-RISC-25 at T3/T4; RS-14 at T2/T3/T4), lower depressive symptoms (PHQ-9), anxiety symptoms (GAD-7), and perceived stress (PSS-4), more favorable affect (higher positive affect, lower negative affect on the PANAS), higher self-esteem (RSES at T2), higher psychological well-being (PWBS-29 subscales), higher self-compassion (SCS-SF), higher enjoyment orientation (EOS) and lower alcohol consumption at T2 (AUDIT-C)
Target population
University students
Study design
The study employed a two-arm randomized controlled trial design. Participants were allocated either to the digital resilience intervention (intervention group, IG; n = 131) or to a no intervention control group (CG; n = 131) with unrestricted access to treatment as usual. Control group participants received access to the intervention after completion of the final 12 months follow-up assessment. The study followed a prospective longitudinal design with assessments at baseline (T1), post-intervention at 8 weeks (T2), and follow-ups at 6 months (T3) and 12 months (T4).
Participants / Eligibility / Setting
Recruitment
Recruitment took place between March 2018 and January 17, 2020, with final follow-up assessments completed on January 20, 2021. Participants were recruited from Germany, Austria, and Switzerland through the German site of the WHO World Mental Health College Student Initiative, a collaborative effort among universities to provide online mental health interventions. Nineteen partner universities invited students to participate in randomized trials targeting student populations. In addition, promotional activities were conducted during freshman events at Friedrich-Alexander University Erlangen-Nürnberg and the University of Ulm (Germany). Interested students were directed to the StudiCare website (studicare.com), which provided detailed information on the interventions and study participation requirements. Students who expressed interest by completing an online registration form received a welcome email from the study team. This email included instructions to sign an informed consent form and complete a screening questionnaire. Participants were randomized only after informed consent was obtained, the screening completed, and no exclusion criteria were identified.
Inclusion criteria
  • German-speaking university students in Germany, Austria and Switzerland
  • Internet access and the ability to use a computer
  • written informed consent
Exclusion criteria
  • a history of a common mental disorder within the past 12 months
  • currently receiving or awaiting psychotherapy, or had received psychotherapy within the past 12 months
  • had a current or past diagnosis of bipolar or psychotic disorder
  • reported suicidal plans or an elevated risk of suicide, indicated by a score greater than 1 on item 9 of the Beck Depression Inventory-II (BDI-II)
Setting
online
Centers
monocentric
Intervention & Comparator
Intervention
CORE was a 6-week internet-based prevention program designed to teach skills and strategies for managing everyday stressors, with the goal of strengthening resilience and coping abilities, fostering self-empowerment, and improving well-being. The intervention comprised seven interactive modules intended for weekly sessions (see Table 1). Each module consisted of psychoeducation and exercises that guided the user in practicing the proposed skills and took an estimated 45 to 60 minutes to complete. The intervention was enriched with images, audio, videos, and vignettes of affected persons, representing different target population characteristics. The program could be accessed via either a webbrowser or a mobile app. To facilitate the transfer of learned strategies into everyday life, participants could fill out a structured mobile diary, focussing on the modules’ topic. The therapeutic components of the program were evidence-based techniques selected following the Riff model of well-being (Ryff, 1989; Ryff, 2014), and organized in 6 dimensions: Autonomy, Self-Acceptance, Environmental Mastery, Purpose in Life, Positive Relations, and Personal Growth. CORE was running on the Minddistrict platform (Minddistrict B.V., Amsterdam, The Netherlands), a web-based eHealth platform.


AB
Module Objective
(0) Welcome Introductory module presenting the program, including an overview of the tools and instructions on how to use CORE.
(1) Psychoeducation Introduction to psychological well-being and resilience: * Understand the concept of psychological well-being, its key components, and its relevance in everyday life. * Understand the concept of resilience and the importance of developing and strengthening it.
(2) Autonomy: building my way Promotion of autonomy and self-direction: * Develop a healthy lifestyle by achieving balance across key areas such as physical activity, nutrition, and sleep, enabling individuals to pursue their personal goals. * Enhance psychological well-being by fostering personal strengths and aligning actions with individual values and life goals.
(3) Mindfulness and self-compassion Training in mindfulness, savoring, and self-compassion: * Understand the concept of mindfulness, how to cultivate it, and its benefits. * Learn to create distance from one’s thoughts and manage them more effectively. * Recognize, appreciate, and savor positive experiences. * Develop kindness toward oneself and strengthen self-care through self-compassion.
(4) Overcoming obstacles Development of coping strategies for everyday challenges: * Recognize the importance of addressing problems constructively. * Learn and apply problem-solving techniques. * Understand how thoughts influence emotions and develop greater cognitive flexibility in interpreting situations.
(5) Connecting to others Recognizing the importance of social relationships for well-being: * Acknowledge the value of social connections. * Learn strategies to maintain and improve relationships. * Foster high-quality relationships that support and strengthen resilience.
(6) Purpose in life and personal growth Encouraging a positive and goal-oriented outlook on the future: * Reflect on personal values and what is meaningful to the individual. * Plan for the future in alignment with personal goals and priorities.
Table 1: CORE modules and their objectives

Comparator
Participants allocated to the control condition (Treatment as usual) were assessed at baseline, 8 weeks, and at 6- and 12-month follow-ups. They did not receive the intervention during the controlled study phase. They received access to the prevention program by the end of the last follow-up assessment.
Outcomes
Primary outcome
Resilience was assessed using the Connor–Davidson Resilience Scale (CD-RISC; Connor & Davidson, 2003), a 25-item self-report measure of stress coping ability. Items are rated on a 5-point Likert scale ranging from 0 (strongly disagree) to 4 (strongly agree), yielding total scores between 0 and 100, with higher scores indicating greater resilience. The CD-RISC has demonstrated good internal consistency, with Cronbach’s alpha coefficients above 0.70 (Yu & Zhang, 2007; Singh & Yu, 2010).
Secondary outcomes
  • Resilience: Resilience was additionally measured using the 14-item Resilience Scale (RS-14; Wagnild, 2009), with scores ranging from 14 to 98 and reported internal consistency between α = 0.76 and 0.94.
  • Well-being: Psychological well-being was assessed with the 29-item version of the Ryff Scales of Psychological Well-Being (PWBS-29; Ryff, 1989), a theory-driven instrument capturing six dimensions: autonomy, environmental mastery, personal growth, positive relations, purpose in life, and self-acceptance. Items are rated on a 6-point scale (1 = completely disagree to 6 = completely agree). The scale has demonstrated good psychometric properties (e.g., Van Dierendonck, 2004; Díaz et al., 2006), with internal consistency ranging from α = 0.70 to 0.90.
  • Depression: Depressive symptoms were measured using the 9-item Patient Health Questionnaire (PHQ-9; Kroenke et al., 2001; Kroenke & Spitzer, 2002). Items are rated from 0 (not at all) to 3 (nearly every day), resulting in total scores from 0 to 27. Cut-off scores of 5, 10, 15, and 20 indicate mild, moderate, moderately severe, and severe depression, respectively. The PHQ-9 has shown good reliability and validity (e.g., Wittkampf et al., 2007), with α ≈ 0.89.
  • Positive and Negative Affect: Emotionality was assessed using the Positive and Negative Affect Schedule (PANAS; Watson et al., 1988), a 20-item measure comprising two subscales: positive affect (PANAS+) and negative affect (PANAS−). Each subscale ranges from 10 to 50. The PANAS is a reliable and valid instrument with strong convergent and discriminant validity (e.g., Sandín et al., 1999).
  • Anxiety: Anxiety symptoms were measured using the 7-item Generalized Anxiety Disorder scale (GAD-7; Spitzer et al., 2006). Items are rated from 0 (not at all) to 3 (nearly every day), yielding total scores from 0 to 21. Cut-offs of 5, 10, and 15 represent mild, moderate, and severe anxiety, respectively. The GAD-7 has demonstrated excellent internal consistency (α ≈ 0.92) and strong validity (e.g., Löwe et al., 2008).
  • Stress: Perceived stress was assessed with the 4-item Perceived Stress Scale (PSS-4; Cohen et al., 1983), with scores ranging from 0 to 16. Items are rated on a 5-point scale (1 = never to 5 = very often). The PSS-4 has shown good internal consistency (α = 0.84–0.86) and has been validated across multiple studies.
  • Self-esteem: Self-esteem was measured using the Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965), a 10-item instrument assessing global self-worth. Items are rated on a 4-point scale (1 = strongly disagree to 4 = strongly agree), with total scores ranging from 0 to 30. The scale has demonstrated high internal consistency and test–retest reliability (e.g., Gray-Little et al., 1997; Robins et al., 2001), with α ≈ 0.91.
  • Self-compassion: Self-compassion was assessed using the short form of the Self-Compassion Scale (SCS-SF; Raes et al., 2011), a 12-item measure rated on a 5-point scale (1 = almost never to 5 = almost always). It captures key facets such as self-kindness, common humanity, and mindfulness. The SCS-SF has demonstrated adequate reliability and validity (α = 0.80–0.92).
  • Enjoyment (Anticipatory Pleasure): Anticipatory pleasure was measured using the 6-item Enjoyment Orientation Scale (EOS; Hervás & Vázquez, 2006). Items are rated on a 7-point Likert scale (1 = strongly disagree to 7 = strongly agree), with total scores ranging from 6 to 42. The scale assesses the tendency to seek and engage in enjoyable experiences.
  • Substance Abuse: Alcohol abuse was assessed using the Alcohol Use Disorders Identification Test – Consumption (AUDIT-C; Bush et al., 1998), a 3-item screening tool with scores ranging from 0 to 12. Scores ≥4 for men and ≥3 for women indicate hazardous drinking. The AUDIT-C has demonstrated good psychometric properties (e.g., Frank et al., 2008).
  • Treatment Credibility and Expectancy: Treatment credibility and outcome expectancy were assessed using the Credibility and Expectancy Questionnaire (CEQ; Devilly & Borkovec, 2000), a 6-item measure with responses on 10-point and percentage scales. The instrument has demonstrated high internal consistency across different populations.
  • Program Satisfaction: Satisfaction with the intervention was measured using the Client Satisfaction Questionnaire (CSQ; Larsen et al., 1979; Attkisson & Greenfield, 1996), an 8-item scale with total scores ranging from 8 to 32. The CSQ has shown excellent internal consistency.
  • Working Alliance: The therapeutic alliance in the digital intervention was assessed using the Working Alliance Inventory for Technology-Based Interventions (WAI-TECH; Kiluk et al., 2014), an 8-item questionnaire covering agreement on goals and tasks. Items are rated on a 5-point scale (1 = never to 5 = always). The instrument has demonstrated adequate reliability and validity.
Measurement points / Definitions
Primary endpoint
The primary endpoint was resilience at post-treatment (T2), measured with the 25-item Connor–Davidson Resilience Scale (CD-RISC-25), analyzed as a continuous total score.

Secondary endpoints
• CD-RISC-25 at T3 and T4
• Resilience Scale, 14-item version (RS-14) at T2, T3, and T4
• Patient Health Questionnaire-9 (PHQ-9) at T2, T3, and T4
• Generalized Anxiety Disorder-7 (GAD-7) at T2, T3, and T4
• Perceived Stress Scale-4 (PSS-4) at T2, T3, and T4
• Positive and Negative Affect Schedule (PANAS), positive affect and negative affect, at T2, T3, and T4
• Rosenberg Self-Esteem Scale (RSES) at T2
• Ryff Psychological Well-Being Scales (PWBS-29) subscales at T2, T3, and T4
• Self-Compassion Scale short form (SCS-SF) at T2, T3, and T4
• Enjoyment Orientation Scale (EOS) at T2, T3, and T4
• Alcohol Use Disorders Identification Test–Consumption (AUDIT-C) at T2


Intervention-related outcomes
The following intervention-related outcomes were analyzed descriptively in the intervention group:
• number of completed modules
• intervention completion rate
• number of diary entries
• client satisfaction (CSQ-8)
• working alliance (WAI-TECH)
Safety-/Harms-Outcomes
Because the intervention was a low-intensity preventive self-help programme, serious harm was not expected to be common. Students screened positive for elevated suicide risk (indicated by a score greater than 1 on item 9 of the Beck Depression Inventory-II, BDI-II) were excluded and referred to appropriate support options. No safety-relevant concerns were brought to the attention of the study team during trial conduct that needed to be documented and handled according to study procedures.
Sample size
To inform the sample size calculation, we considered the evidence available at the time the trial was planned. At that point, no meta-analytic benchmarks were available specifically for digital resilience interventions. The most relevant meta-analysis on resilience reported a small-to-moderate pooled effect of resiliency training on resilience (SMD = 0.37), but also highlighted substantial heterogeneity and limited study quality (Leppin et al., 2014). Similarly, contemporary reviews described the resilience intervention literature as methodologically heterogeneous and not sufficiently standardized to support robust quantitative benchmarking (Robertson et al., 2015). We therefore also drew on meta-analytic evidence from digital stress interventions as the closest digital comparator, which indicated small-to-moderate post-treatment effects on stress- and depression-related outcomes (e.g., d =0.43 for stress and d = 0.34 for depressive symptoms; Heber et al., 2017). On this basis, we chose a conservative expected between-group effect size of d = 0.35. Although the primary outcome was intended to be analyzed using ANCOVA, the sample size calculation was based on a conventional two-sample t-test framework because ANCOVA-based estimation requires a baseline-to-follow-up correlation for the primary outcome (Faul et al., 2009), and no robust estimate for the CD-RISC was available in a comparable sample at the time of planning. To avoid underpowering the study by assuming an overly optimistic correlation, we therefore adopted the more conservative t-test approach. Power calculations in GPower 3.1.5.2 for a two-sample comparison with equal allocation, 80% power, and a two-sided α of 0.05 indicated a required sample size of N = 260. Power calculations in GPower 3.1 (Erdfelder et al., 2009) for a two-sample comparison with equal allocation, 80% power, and a two-sided α of 0.05 likewise yielded a required sample size of N = 260.
Allocation / Randomisation / Concealment / Blinding
Block randomization was employed to allocate participants to either the intervention or control group, ensuring comparable group sizes. Participants were randomly assigned in blocks of four or six using a 1:1 allocation ratio. Group assignment was generated using a computer-based randomization tool (Randlist, Datinf GmbH, Tübingen, Germany). The randomization sequence was precomputed by an independent researcher not otherwise involved in the study and uploaded to a secure randomization platform. Study personnel accessed this platform to assign participants, ensuring allocation concealment until the point of assignment. Due to the nature of the intervention-control comparison, full participant blinding was not feasible.
Statistical analysis plan
1. Introduction
This Statistical Analysis Plan specifies the statistical methods for the randomized controlled trial evaluating the efficacy of a self-guided digital resilience intervention compared with a no-intervention control condition with unrestricted access to treatment as usual in university students.
2. Study objectives
2.1 Primary objective
The primary objective was to determine whether the intervention improves resilience at post-treatment compared with the control condition.
2.2 Secondary objectives
Secondary objectives were to determine whether the intervention improved resilience, mental health, psychological well-being, and related outcomes at post-treatment, 6-month follow-up, and 12-month follow-up.
2.3 Exploratory objectives
Exploratory objectives were to examine whether intervention effects on the primary outcome differ according to baseline resilience, baseline perceived stress, age, or gender.
3. Trial design
This study was a two-arm, parallel-group, randomized controlled superiority trial with a 1:1 allocation ratio. Participants were assigned either to the intervention group or to a no-intervention control group with unrestricted access to treatment as usual. Assessments took place at baseline (T1), post-treatment at 8 weeks (T2), 6-month follow-up (T3), and 12-month follow-up (T4).
4. Planned sample size
The trial planed to enroll 260 participants.
5. Analysis populations
5.1 Randomized population
The randomized population included all participants who were randomized.
5.2 Intention-to-treat population
The intention-to-treat population included all randomized participants and analyzed them according to their allocated treatment group, irrespective of intervention uptake, adherence, or use of concomitant care. Participants who requested deletion of all study data for data protection reasons were excluded from all analyses because no analyzable data remained available.
5.3 Study-completer population
For a given assessment point, the study-completer population included all participants with observed non-missing data for the respective outcome at that time point.
5.4 Intervention-completer population
The intervention-completer population included all participants allocated to the intervention group who completed all intervention modules. This population was used in a sensitivity analysis comparing intervention completers with the control group.
6. Outcome definitions
6.1 Primary endpoint
The primary endpoint was resilience at post-treatment (T2), measured with the 25-item Connor–Davidson Resilience Scale (CD-RISC-25), analyzed as a continuous total score.
6.2 Secondary endpoints
• CD-RISC-25 at T3 and T4
• Resilience Scale, 14-item version (RS-14) at T2, T3, and T4
• Patient Health Questionnaire-9 (PHQ-9) at T2, T3, and T4
• Generalized Anxiety Disorder-7 (GAD-7) at T2, T3, and T4
• Perceived Stress Scale-4 (PSS-4) at T2, T3, and T4
• Positive and Negative Affect Schedule (PANAS), positive affect and negative affect, at T2, T3, and T4
• Rosenberg Self-Esteem Scale (RSES) at T2
• Ryff Psychological Well-Being Scales (PWBS-29) subscales at T2, T3, and T4
• Self-Compassion Scale short form (SCS-SF) at T2, T3, and T4
• Enjoyment Orientation Scale (EOS) at T2, T3, and T4
• Alcohol Use Disorders Identification Test–Consumption (AUDIT-C) at T2
6.3 Intervention-related outcomes
The following intervention-related outcomes were analyzed descriptively in the intervention group:
• number of completed modules
• intervention completion rate
• number of diary entries
• client satisfaction (CSQ-8)
• working alliance (WAI-TECH)

Adherence in the intervention group was assessed based on intervention completion, defined as completion of all modules. User satisfaction (CSQ-8) and working alliance were additionally evaluated in the intervention group using descriptive analyses of the respective measures.]
7. Baseline characteristics
Baseline demographic and clinical characteristics were summarized descriptively by randomized group. Continuous variables were reported using means and standard deviations or, where appropriate, medians and interquartileranges. Categorical variables were reported using frequencies and percentages. No formal significance testing of baseline group differences was planned.
8. General analysis principles
All statistical tests were two-sided. The primary hypothesis test concerns the primary endpoint only and uses a significance level of alpha = 0.05. No formal multiplicity adjustment was applied. All analyses of secondary outcomes, moderator analyses, and sensitivity analyses were considered secondary or exploratory and were interpreted accordingly. For these analyses, effect estimates, 95% confidence intervals, and nominal p-values were reported. All analyses were conducted in R using standard statistical packages. Missing data were handled using multiple imputation, and pooled estimates were combined according to Rubin’s rules.
9. Missing data handling
The primary analysis assumed that missing data were missing at random conditional on the variables included in the imputation model. Missing data were handled using multivariate imputation by chained equations (MICE) with 50 imputations and 50 iterations. Imputation was performed in the full dataset with treatment group included as a predictor. The imputation model included treatment group, all repeated outcome measures, age, gender, baseline covariates used in moderator models, all outcomes assessed, and all assessed socio-demographic and clinical baseline variables. Parameter estimates and standard errors from the analyses fitted in each imputed dataset were pooled according to Rubin’s rules.
10. Primary efficacy analysis
The primary estimand was the between-group difference in mean resilience at T2 under a treatment-policy strategy, comparing participants according to randomized assignment regardless of intervention uptake or use of concomitant care. The primary analysis used analysis of covariance (ANCOVA) with CD-RISC-25 at T2 as the dependent variable, randomized group as the fixed effect, and baseline CD-RISC-25 score as the covariate. Treatment group was coded with the control group as reference and the intervention group as the comparison group. The adjusted group effect was reported as the adjusted mean difference, F statistic, two-sided p-value, and 95% confidence interval. In addition, standardized between-group effect sizes are reported with 95% confidence intervals using the pooled standard deviation of changes in outcome. The number needed to treat (NNT) was derived from the standardized mean difference using the Kraemer and Kupfer approach.
11. Secondary outcome analyses
Secondary outcomes were analyzed separately for each post-baseline assessment point using ANCOVA models. For each outcome and time point, the model included the post-baseline outcome as the dependent variable, randomized group as the fixed effect, and the baseline score of the same outcome as the covariate. No additional covariates were included in the main ANCOVA models. For each model, the adjusted group effect, F statistic, two-sided nominal p-value, 95% confidence interval, SMD with 95% confidence interval, and NNT, where applicable, were reported.
12. Moderator analyses
Moderator analyses were conducted for the primary outcome measure, CD-RISC-25, at T2, T3, and T4.
Each moderator model included randomized group, baseline CD-RISC-25, the moderator of interest, and the group-by-moderator interaction term. The moderators examined were baseline resilience, baseline perceived stress (PSS-4 at baseline), Depressive Symptoms (PHQ-9), anxiety (GAD-7), lifetime psychopathology, age analyzed as a continuous variable, and gender (analyzed in participants reporting female or male gender only) The main parameter of interest was the group-by-moderator interaction term. For significant interactions simple-slope analyses were conducted at low baseline (mean − 1 SD), average baseline (mean), and high baseline (mean + 1 SD). Given prior empirical findings suggesting possible moderation by baseline resilience, these simple-slope analyses were conducted for baseline resilience when the interaction was at least trend-level.
13. Sensitivity analyses
Two sensitivity analyses were conducted. A study-completer analysis was conducted for each assessment point and included participants with observed non-missing data for the respective outcome at that time point. An intervention-completer analysis compared intervention participants who completed all intervention modules with the control group. This sensitivity analysis was conducted using the same multiple-imputation framework as the main analysis.

14. Concomitant care and intervention adherence
Participants in both trial arms were free to use usual healthcare, psychotherapy, counseling, pharmacological treatment, and preventive services during the study period. Such care was not prohibited by the protocol and was not considered a protocol deviation. The intervention was self-guided. No additional trial-specific adherence-enhancing measures were implemented beyond standard platform delivery and routine study procedures.
15. Model diagnostics and statistical assumptions
For the ANCOVA models, standard model assumptions were evaluated, including linearity of the relationship between baseline covariates and outcomes, approximate normality of residuals, homoscedasticity, and the absence of highly influential observations. Diagnostics were based on standard graphical and numerical procedures, including residual plots and inspection of influential cases. Because the primary and secondary analyses were prespecified ANCOVA models for continuous outcomes, minor deviations from assumptions did not automatically trigger alternative modeling. If applicable, any substantial deviations judged to materially affect inference were documented and reported transparently.
16. Interim analyses and stopping rules
No interim analyses were planned. No formal stopping rules were defined.
17. Safety analyses
No formal comparative safety analysis was planned. Safety-relevant events brought to the attention of the study team were documented and described narratively. Reliable deterioration on the CD-RISC-25 was evaluated using the Jacobson–Truax method, applying a test–retest reliability coefficient of 0.87 and reliable deterioration defined as RCI ≤ −1.96
18. Statistical software
All analyses were conducted in R. Multiple imputation was implemented using the mice package. Pooling across multiply imputed datasets and related post-estimation procedures were conducted using standard R packages available at the time of analysis.
19. Reporting of results
Results were reported in accordance with CONSORT recommendations. The main report and supplementary materials included participant flow, baseline characteristics by randomized group, descriptive statistics by group and time point, primary and secondary ANCOVA results, moderator analyses, simple-slope analyses for baseline resilience, sensitivity analyses, and descriptive intervention-related outcomes.
20. Deviations from the Statistical Analysis Plan (SAP)
Any deviations from this SAP occurring after finalization were documented, justified, dated, and reported in the final manuscript or supplement.
21. References - Statistical Analysis Plan (SAP)
Connor, K. M., & Davidson, J. R. T. (2003). Development of a new resilience scale: The Connor-Davidson Resilience Scale (CD-RISC). Depression and Anxiety, 18(2), 76–82.

Kraemer, H. C., & Kupfer, D. J. (2006). Size of treatment effects and their importance to clinical research and practice. Biological Psychiatry, 59(11), 990–996.

Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. John Wiley &Sons.

van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45(3), 1–67.
Harms / Safety / Monitoring
Individuals identified as being at risk for suicide (indicated by a score greater than 1 on item 9 of the Beck Depression Inventory-II, BDI-II) were provided with information about appropriate treatment options and support services.

No independent data monitoring committee has been established for thistrial. This decision reflected the characteristics of the study as a low-risk,single-centre, self-guided psychological intervention trial without investigational medicinal products, invasive procedures, or anticipated major safety concerns. Trial oversight was instead be provided by the principal investigator and the study team, who monitored study conduct, protocol adherence, data quality, and any safety-relevant issues arising during the trial. Should unexpected safety concerns have emerged, these would have been documented, reviewed, and handled in accordance with study procedures and ethics requirements.
Ethics / Consent / Dissemination / Data sharing
The trial was registered in the German Clinical Trials Register (DRKS00013765) and approved by the Ethics Committee of the University of Erlangen–Nürnberg (8_18 B). Anonymized participant-level data will be provided to qualified researchers upon reasonable request to David Daniel Ebert ([email protected]), depending on to be specified data security and data exchange regulation agreements. Requests should be accompanied by a brief research proposal and analysis plan. Access will be granted following review and approval by the study team and completion of a data use agreement
Protocol deviations
Following completion of recruitment in the parallel sister trial targeting students with markedly low baseline resilience (Herrero et al., 2019), the prespecified exclusion criterion for very low baseline resilience was removed. This amendment was made for primarily pragmatic and conceptual reasons. Most importantly, it allowed the trial to evaluate a broader and more ecologically valid implementation scenario, better reflecting how such interventions are likely to be delivered in routine practice, namely without extensive screening procedures. This was aligned with the aim of evaluating the programme as a scalable mental health promotion intervention for university students. Under the original eligibility criteria, participants were excluded if they scored below 63 on the Connor–Davidson Resilience Scale (CD-RISC-25), corresponding to approximately one standard deviation below the mean. Because students with markedly low resilience had already been specifically addressed in the sister trial, maintaining this restriction in the present study was no longer necessary within the wider research programme. A further advantage of removing the criterion was that it preserved a wider range of baseline resilience, which also improved the basis for exploratory analyses of whether intervention effects differed by initial resilience. Overall, the amendment increased the practical relevance and external validity of the trial.

Additional deviations from the protocol was to add a post hoc moderation analysis examining whether the COVID-19 pandemic influenced intervention effects at the 12-month follow-up (T4). This was operationalized by comparing participants who completed the T4 assessment before versus after March 10, 2020, the date on which the university first announced the suspension of in-person teaching in response to the pandemic. Six days later, a nationwide lockdown was announced, which came into effect on March 22, 2020.]

Additionally we examined reliable deterioration in resilience, depressive symptoms, and anxiety symptoms as a safety outcome using the Jacobson–Truax method.

R Analysis Script

Dataset
R Code for all analyses
NAME
https://doi.org/10.17605/OSF.IO/DC2AZ
LINK

Protocol references
Herrero et al., 2019; Proudfoot et al., 2011; Heber et al., 2017; Faul et al., 2009; Erdfelder et al., 1996; Ryff, 1989; Ryff, 2014; Connor 6 Davidson, 2003; Wagnild, 2009; Spitzer, 2002; Watson et al., 1988; Spitzer et al., 2006; Cohen et al., 1983; Rosenberg, 1965; Raes et al., 2011; Hervás 6 Vázquez, 2006; Bush et al., 1998; Rammstedt 6 John, 2007; Devilly 6 Borkovec, 2000; Attkisson 6 Greenfield, 1994; Larsen et al., 1979; Kiluk et al., 2014; Buuren 6 Groothuis-Oudshoorn, 2011; Grund et al., 2021; Kraemer 6 Kupfer, 2006.
Acknowledgements
**Authors contribution**
DDE initiated this study. DDE, MF contributed to the design of this study. RH, MF, RB, CBO developed the intervention content. MF and DDE obtained ethics approval, MF coordinated the trial and conducted participant management. MF, SN and EB carried out data analyses supervised by CBu. MF drafted the method section, which SN expanded and revised, supervised by CBu. DDE developed first draft of the manuscript circulated to the co-authors, all co-authors reviewed and edited the original draft, read and approved the final version of the manuscript.

**Funding**
This work was partly supported by the European Union’s Horizon 2020 research and innovation programme (grant no. 634757), which supported the development of the intervention. Trial management was supported by the StudiCare project funded by BARMER, a German statutory health insurance provider. The funders had no role in the analysis or interpretation of the data, drafting of the manuscript, or the decision to submit the manuscript for publication.

**Declaration of competing interest**
MF, RH, RB, CBO were involved in the development of the CORE intervention. DDE reports having received consultancy fees from, and served on the scientific advisory boards of, several companies such as Sanofi, Minddistrict, Lantern, Schoen Kliniken, and German health insurance companies (Techniker Krankenkasse, BARMER). He is a shareholder of the Institute for health training online GmbH (GET.ON/HelloBetter), a provider of digital therapeutics for mental disorders in routine mental health care (unrelated to the present work). HB and DDE report having received consultancy fees and fees for lectures/workshops from chambers of psychotherapists and training institutes for psychotherapists in the e-mental-health context.

**Data availability statement**
Anonymized participant-level data will be provided to qualified researchers upon reasonable request to David Daniel Ebert ([email protected]), depending on to be specified data security and data exchange regulation agreements. Requests should be accompanied by a brief research proposal and analysis plan. Access will be granted following review and approval by the study team and completion of a data use agreement.