Study staff from the Department of Child and Adolescent Psychiatry, Psychosomatics, and Psychotherapy at the LMU Hospital in Munich recruited adolescents between January 2021 and May 2022. Recruitment strategies included flyers and two contact lists: one provided by the local authority and one comprising adolescents who had previously participated in studies at the department.

The study was approved by the ethics committee of the Medical Faculty of the LMU Munich (No. 20115). All participants were informed in detail about the study procedures and aims. Participants aged 18 years provided written informed consent; those under 18 years provided written assent, with additional consent from a parent or legal guardian.

Inclusion criteria were: age between 12 and 18 years, an intelligence quotient (IQ) ≥ 80, no current psychiatric diagnosis, no past diagnosis of depressive disorder according to ICD-10 (27), and sufficient German language skills to complete study materials. IQ was assessed using the Culture Fair Intelligence Test – Revision [CFT-20-R; (28)]. The absence of ICD-10 diagnoses of mental disorders was confirmed using a well-established diagnostic interview [Kinder-DIPS; Diagnostic Interview for Mental Disorders in Children and Adolescents; (29)].

A total of 77 participants met inclusion criteria and were randomized to either the depression information group (n = 39) or mental health information group (n = 38). According to an a priori power calculation, this sample size provided adequate statistical power to detect the expected effects (see Supplementary Material). Demographic characteristics of both groups were comparable and are shown in Table 1.

This single-blind randomized trial (ClinicalTrials.gov Identifier: NCT05300217) included two parallel experimental groups. Participants were blinded to group allocation, while study personnel involved in data collection were aware of assignments. All sessions were conducted by trained psychology students.

Participants completed one test session (T1) and two follow-up (FU) appointments scheduled two weeks (FU1) and four weeks (FU2) after T1, respectively. Data collected outside these timeframes (FU1: after 2–3 weeks; FU2: after 4–5 weeks) were treated as missing. The flow of participants throughout the study, including dropouts and exclusions, is shown in Figure 1.

The T1 session lasted between two and three hours and comprised three parts: a pre-test (pre-T1), random allocation to the respective information, and a post-test (post-T1). The pre-test included a diagnostic assessment and an IQ test. Once eligibility had been confirmed, participants were randomized into two study groups using a computer-generated random number sequence. Randomization was stratified by age (<15 and ≥15 years) and sex, with an allocation ratio of 1:1. Participants were assigned to the intervention group by investigators (RP and LI), who were not involved in data collection. After the diagnostic assessment, participants completed the pre-test questionnaires and provided their demographic information.

The web-based information was subsequently presented (see below for details). Finally, participants completed the post-test questionnaires. The FU sessions consisted of participants completing the post-test questionnaires again and lasted approximately 30–45 min.

Sessions were primarily conducted at the facilities of the research department. Some experimental sessions had to be conducted online due to restrictions caused by the COVID-19 pandemic during the evaluation period. Online sessions were conducted via screen sharing using RED connect (RED Medical Systems GmbH). Each participant visited the research department at least once for a face-to-face test session.

An offline version of the website “ich bin alles” was used for the study. It contained only the materials relevant to the two intervention conditions: information on depression and information on strategies to promote mental health. Both were similar in structure and design but covered distinct, non-overlapping content areas. The public version of the website was launched in September 2021, while data collection was still ongoing. No participant reported prior exposure to the website “ich bin alles” at any measurement point.

Before the presentation of the web-based information, participants in both groups were told that they would examine material from a website and were asked to review the content attentively. Study personnel notified them when the predefined time window had elapsed. Depending on the length of the text, podcast, or video, participants were given between two and eight minutes. During the presentation, study personnel remained with each participant to ensure compliance.

Participants in the depression information group viewed content explaining depression, including its frequency, symptoms, and course. This material also addressed contributing factors and sources of professional help. Participants in the mental health information group viewed content focusing on strategies to promote mental health. Topics included self-care, exercise and sleep management, positive thinking, stress management, and problem solving.

An overview of all outcome measures applied at each time point is provided in Figure 2, and detailed descriptions are summarized in Supplementary Table S1.

For both the mental health knowledge and depression knowledge questionnaires, raw scores were converted into index scores by dividing the total score obtained by the maximum possible score, and multiplying the result by 100. The index score thus represents the percentage of correct responses on each questionnaire. For the calculation of these indices, mean imputation was used for single missing items; this was the case for 0.29% of items on the depression knowledge questionnaire, and for none of the items on the mental health knowledge questionnaire.

All analyses were performed using IBM SPSS Statistics 29. To examine changes over time, mixed-model analyses of variance (ANOVAs) were employed due to their suitability for longitudinal data. Unlike standard repeated-measures ANOVAs, mixed models (also known as multilevel models) use all available data and do not exclude participants with missing values due to attrition (31). Thus, participants who provided data for at least one measurement point were included in the analyses, and missing data at later measurement occasions were handled using maximum likelihood estimation.

Change in knowledge of depression and in knowledge of strategies to promote mental health over time were analyzed using mixed-model ANOVAs with the factors group (depression information, mental health information) and time (pre-T1, post-T1, FU1, FU2).

Following the ANOVA omnibus tests, and to minimize multiple testing, we examined only specific contrasts of interest: (1) differences between the two groups at each measurement point and (2) differences between the first measurement point (pre-T1) and all subsequent measurement points within each group, to quantify short- and long-term effects.

To facilitate interpretation of the mixed model parameters for these contrasts as effect sizes, the dependent variables were standardized to z-scores (mean = 0, SD = 1), and group and time were represented by binary indicator variables. Standardized regression coefficients were preferred over traditional effect size indices such as η² or Cohen's d, as these indices are not straightforward to compute in mixed models where variance components and sample sizes can differ across levels and time points. By standardizing the dependent variables, the resulting regression coefficients express changes in standard deviations units, allowing them to be interpreted using Cohen's conventions (0.2 = small effect, 0.5 = medium effect, 0.8 = large effect (32, 33).

Descriptive statistics were used to summarize perceived visual aesthetics and the overall reception of the website. For the eight items assessing ease of use, utility, and enjoyment, results were reported as the percentage of participants selecting each agreement category. The VisAWI-S and the global school-grade rating were analyzed as mean values. Differences between the two groups were tested using independent-samples t-tests and chi-square tests.

For changes in knowledge of depression, the mixed-model ANOVA revealed a significant main effect of group, F (1, 77.21) = 11.01, p = .001, a significant main effect of time, F (3, 76.67) = 38.82, p < .001, and a significant group × time interaction F (3, 76.67) = 13.29, p < .001. Means and standard deviations for this outcome are provided in Supplementary Supplementary Table S2. Figure 3 displays changes in knowledge of depression over time for both groups.

Pairwise contrasts and effect sizes are shown in Table 2. The groups did not differ significantly in knowledge of depression at pre-T1 (baseline) but differed significantly at post-T1, FU1, and FU2, with the depression information group showing higher knowledge scores. In both groups, knowledge of depression increased significantly from pre- to post-T1 and remained elevated at FU1 and FU2. These increases corresponded to medium effect sizes for the mental health information group and large effect sizes for the depression information group.

For changes in knowledge of strategies to promote mental health, the mixed-model ANOVA revealed a significant main effect of time, F (3, 76.41) = 34.45, p < .001, and no significant main effect of group, F (1, 77.07) = 0.85, p = .36. However, there was a significant group × time interaction, F (3, 76.41) = 7.89, p < .001. Means and standard deviations for this outcome are presented in Supplementary Supplementary Table S2. Figure 3 displays changes in knowledge of strategies to promote mental health over time for both groups.

Pairwise contrasts and effect sizes are shown in Table 3. The groups did not differ significantly in knowledge at any measurement point (pre-T1, post-T1, FU1, and FU2). Both groups showed a significant increase in knowledge of strategies to promote mental health from pre- to post-T1, which remained elevated at FU1 and FU2. These increases corresponded to large effect sizes in the mental health information group and, by contrast, to small to medium effect sizes in the depression information group.

One of the main strengths of this study relates to the inclusion of two experimental study groups which served as active control conditions for the respective primary knowledge outcomes. This design enabled a direct comparison of the effectiveness of different web-based information materials. In addition, the controlled experimental approach with a standardized test session represents a further strength, as it minimized variability between participants and thereby increased the likelihood of observing true effects attributable to the specific informational content. This enhances the internal validity of the findings.

Another advantage of this study is the scientific evaluation of a publicly available website. Users of the website “ich bin alles” have access to high-quality, evidence-based mental health information, with transparency regarding how the content is developed and by whom. Free access ensures that a broad range of adolescents can benefit from scientifically evaluated mental health information in a format tailored to their specific preferences and needs.

While the controlled experimental setting strengthened internal validity, it may have reduced ecological validity, as it differs from naturalistic website use, where adolescents access the material independently. In everyday contexts, engagement and learning may be influenced by factors such as attention, motivation, or competing online activities. To address this, users' experiences in the context of “ich bin alles” are currently being examined in a naturalistic study (ClinicalTrials.gov: NCT06668701), which will provide complementary insights into the website's benefits across different usage contexts. Future research could also integrate digital usage analytics, such as time on page, navigation patterns, or interaction metrics, to better understand how adolescents engage with web-based mental health information.

A further limitation concerns the study sample, which consisted of adolescents with slightly above-average IQ and predominantly higher socioeconomic status. These characteristics may limit generalizability, as lower parental socioeconomic status and below-average IQ among adolescents have been linked to limitations in knowledge acquisition (36, 37). Adolescents from lower socioeconomic backgrounds might also rate the overall reception of the website in a distinct way. For example, perceived utility (i.e., relevance of information) may depend on how well the topics align with the needs and experiences of less privileged groups. Future studies should therefore aim to recruit more diverse samples in terms of both socioeconomic status and IQ to explicitly test potential influences of these variables. Furthermore, gender differences in the use and perception of the web-based information were not examined and should be addressed in future research.

Another limitation relates to the use of self-developed outcome measures, for which external validity has not been evaluated. While this is a methodological limitation, the use of tailored instruments was necessary given the specific content of the website “ich bin alles”. Nevertheless, their content-based development, expert consultation, and acceptable internal consistency provide preliminary support for their suitability in assessing knowledge within the context of this intervention.

An important point to note is that the present study focused exclusively on the knowledge component of mental health literacy, while other components such as stigma and help-seeking efficacy were not assessed. This focus was chosen because knowledge represents a foundational element of mental health literacy and a prerequisite for reducing stigma and enhancing help-seeking behavior (11, 38). Given the extensive nature of the website's content, the questionnaires needed to be sufficiently comprehensive to capture the informational material. Including additional measures would have substantially increased the length of the assessment and may have increased the risk of response fatigue and measurement bias. Notwithstanding these considerations, future studies are encouraged to extend this research by incorporating measures of stigma and help-seeking efficacy to examine whether increases in knowledge translate into attitudinal and behavioral changes among participants.

The results of our study demonstrate that age-appropriate web-based information presented in an appealing visual design can effectively increase knowledge about mental health and depression in adolescents. As most adolescents seek mental health information online (14, 16, 17) these findings highlight the importance of publicly available, evidence-based, and high-quality web-based resources on these topics.

A web-based dissemination approach has considerable potential to provide free, trustworthy information to adolescents, thereby helping to address the general lack of mental health knowledge in this age group. In the long term, high-quality, freely accessible web-based resources could enhance mental health literacy at the community level, which could facilitate earlier recognition of problems and encourage help-seeking, thereby supporting the prevention of mental health disorders. Integrating validated web-based tools into school mental health curricula or national youth portals could broaden their reach and support early intervention in youth.