Mental health (MH) problems in youth are prevalent and pose severe health-related, social, and financial burdens on individuals [1-5] and societies [6,7]. Approximately half of all mental disorders first manifest before the age of 18 years [8], and MH problems in youth often persist and aggravate over the life span [9-14]. Therefore, the need for effective prevention and intervention programs targeting young people is an important public health goal. However, while effective MH services exist, most youth with MH problems do not seek professional help. Low uptake has been reported for various conventional [15-18] as well as digital MH services [19-22]. The burden of mental illness can only be alleviated at the population level if a substantial proportion of the population uses the available services [23,24]. Otherwise, the public health impact of MH services remains limited. Thus, increasing the reach of MH services (ie, fostering the uptake and use of professional help) is vital for the improvement of youth MH at the population level.

To facilitate service use, specific barriers to help seeking need to be addressed. Previous research has indicated that attitudinal factors pose larger impediments to help seeking than structural factors (eg, treatment costs and inconvenient scheduling) [25,26]. Specifically, self-reliance, a low perceived need for help [25-27], low treatment expectations [28], stigma [26,27,29,30], and poor MH literacy [27,31,32] have been identified as major contributors to the lack of professional help seeking.

Different approaches to facilitate help seeking and promote positive attitudes toward MH issues and help seeking in youth have been evaluated in previous research, including face-to-face and digital interventions. In a systematic mapping review, 84% (106/126) of the studies focused on school-based interventions, whereas only 10 (8%) articles covered internet-based approaches to improve MH literacy, MH-related attitudes, stigma, and help-seeking behavior in adolescents [33]. The internet-based interventions included both minimal, single-session interventions [34,35] and multisession approaches intended to be used over several weeks [36,37], with different outcome measures. A total of 4 studies focused on MH more broadly, whereas 6 studies investigated interventions for specific MH problems (depression: n=5; eating disorders: n=1). Keeping the limited number of studies in this area of research in mind, the results nevertheless point to the potential of internet-based interventions with respect to reduced stigma (2 studies), enhanced help-seeking intentions (2 studies), and improved help-seeking behaviors (1 study).

Clearly, there is a need for more research in this area, particularly with respect to digital brief and microinterventions (ie, highly focused in-the-moment interventions with a narrower scope and time frame than standard interventions [38]), which allow for a flexible, easily accessible, scalable, and efficient delivery of MH content. Initial research on such brief and microinterventions with psychoeducational and destigmatizing components has shown promising results. For instance, a brief acceptance-facilitating intervention that included a text-based personalized psychoeducation component had a small but significant effect on the intention to use MH services in German university students [39]. More recently, randomized controlled trials (RCTs) in young adults, university students, and adolescents with short video interventions demonstrated effects with regard to public stigma toward schizophrenia [40,41] and depression [42,43], as well as help-seeking intentions [42] and attitudes [43]. Furthermore, an Australian pilot study with international students found that a brief, web-based MH literacy intervention alleviated MH stigma. However, it had no significant effect on help-seeking intentions or MH literacy [44].

Another component of previous help seeking–facilitating strategies has been storytelling. A pilot study on a video-based intervention indicated that storytelling was well accepted and perceived as engaging [45]. In addition, an RCT evaluated internet-based storytelling programs with varying interactivity and stigma-related content. Significant reductions in MH stigma and microaggression toward individuals with MH problems were observed [46].

Concerning the theoretical foundation of interventions, few studies have investigated help seeking–promoting strategies that were explicitly based on the premises of health behavior models. Logsdon et al [47] evaluated an internet-based depression intervention for adolescent mothers, which was conceptualized according to the theory of planned behavior. The intervention led to significant improvements in help-seeking attitudes, intentions, and behavior. Another well-established and yet more recent health behavior model, which incorporates elements of previously developed approaches, is the Health Action Process Approach (HAPA) [48]. It encompasses a stage theoretical perspective on health behavior and includes a motivational, intention-forming phase as well as a volitional phase, where planning and behavior maintenance occur. In both the HAPA model and the updated version of the theory of planned behavior, namely, the reasoned action approach, outcome expectancies (or instrumental attitudes) play a crucial role in the formation of intentions, and intentions significantly predict actual behavior [49,50]. The results of previous research on a trauma recovery internet intervention support the use of the HAPA model for the prediction of e-MH engagement. Specifically, outcome expectations significantly predicted the intention to use the intervention (β=.36) [51]. Skepticism about treatment effectiveness has further been identified as a predictor for not using MH services in another study with university students [28].

Building on the findings of previous research, this study investigates the short-term effectiveness of 2 brief animated video interventions to promote potential professional help seeking in a general sample of adolescents and young adults aged 14 to 29 years using a web-based RCT approach. Both interventions aimed to improve participants’ willingness to seek professional help (ie, psychotherapists, psychiatrists, and counseling services) for 5 MH problems (generalized anxiety disorder [GAD], depression, bulimia, nonsuicidal self-injury [NSSI], and problematic alcohol use). The inclusion of various MH problems allowed for the investigation of potential differential effects. While one intervention followed a destigmatizing and psychoeducational approach, the other intervention aimed to induce positive outcome expectancies in accordance with the HAPA model through storytelling. The interventions were both compared to each other and to a nonintervention control group (CG) where participants were presented with a stand-alone video vignette without an additional intervention video. This approach was chosen due to both contextual (ie, vignette characters were described as experiencing difficulties in several life domains, and thus, additional control videos referring to the vignettes were unfeasible) and practical (ie, the creation of 10 additional videos was not necessary) considerations.

This study had the following objectives:

Within the framework of this study, the videos were evaluated as stand-alone interventions. They were not developed to replace existing interventions. However, in case of favorable outcomes, they have the potential to complement existing health care services. Results and procedures are reported in accordance with the Checklist for Reporting Results of Internet E-Surveys [52] and the CONSORT-EHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth) [53]. The study was preregistered at the German Clinical Trials Register on September 23, 2020 (DRKS00023110).

This anonymous, fully automated, web-based, parallel-group exploratory RCT compared the effects of intervention 1 (psychoeducational intervention) and intervention 2 (positive consequences of help seeking) against those of the CG (no further videos after the case vignette) with regard to potential help seeking, attitudes toward help seeking, and stigma. The design comprised 15 conditions in total (5 MH problems × 3 interventional conditions). Randomization was stratified by gender and implemented using a permuted block design (block sizes: 15 and 30). Due to anonymous participation and automated randomization, researchers were unable to assign specific conditions to individuals. However, 2 of the authors were able to view the randomization list. The video material was aligned with the participants’ gender to increase identification with the character (ie, participants who identified as woman, female, or nonbinary viewed videos with a female protagonist [Paula], and participants identifying as man or male viewed videos with a male protagonist [Paul]). The study components and conditions as well as the study procedure are shown in Figure 1.

Recruitment started in October 2020 and ended in May 2022. Youth aged between 14 and 29 years with sufficient German language skills were eligible for participation. The age of 14 years is widely accepted as appropriate to provide informed consent for medical decisions and participation in studies [54,55]. The upper age limit of 29 years aligns with the definition of emerging adulthood, a separate life stage between adolescence and adulthood [56,57]. Participants were primarily recruited through the web on social media platforms and via mailing lists, web-based marketplaces, and forums for adolescents and young adults (eg, accounts and emails of youth clubs and student associations). As an incentive to complete the study, participants were offered to take part in an optional gift card lottery at the end of the study (100 gift cards of €20 [US $21.58]). We asked participants for a valid email address if they were interested in the lottery and stored email addresses separately from other study data and user IDs to ensure anonymous participation.

We recorded page change time stamps. Participants whose time stamp data indicated that the video or the videos they were assigned to had not been fully viewed (ie, duration of stay<length of the respective videos) were excluded from statistical analyses. Furthermore, only data from participants who completed all questionnaires were included in the final analysis (n=1394; completion rate: 1394/2435, 57.25%). We also excluded 5 cases with duplicate user IDs, which occurred due to a technical error and indicated repeated participation (Figure 1). HTTP cookies were used to assign individual user IDs to participants. For each session, new cookies were generated and used. Therefore, duplicate participation was possible after the completion of each study session and was not registered by the system. In the 5 aforementioned cases, duplicate IDs were mistakenly generated when participants tried to use the “back” button of their web browser and restarted their participation.

Sociodemographic, screening, and outcome data were first analyzed descriptively. Intervention effects on potential professional help seeking (primary outcome) and secondary outcomes in the total sample (ie, across all MH problems and across participants with and without actual help seeking as reported in the screening) were analyzed via analyses of covariance (ANCOVAs) at an α level of P<.05. In addition to the intervention group, the models included age as a covariate, the participants’ actual help seeking (fixed effects), and the 5 MH problems (random effects) as control variables. The results of the main ANCOVA in the total sample are presented in the Results section.

Subgroup ANCOVAs were conducted for each of the 5 MH problems separately. In this case, the respective screening scores (GAD-7, PHQ-9, WCS, number of NSSI events during the last year, and AUDIT-C) were included as additional covariates. Subgroup analyses were further conducted for cases with and without actual help seeking in the total sample and within each of the 5 MH issue groups.

In case of significant (P<.05) and trend ANCOVA effects, pairwise group comparisons were conducted using 2-tailed t tests. All tests were 2-sided with an α level of 5%. Mean differences (MDs) adjusted for covariates are reported in the Results section.

An a priori power analysis was conducted using G*Power (Heinrich-Heine-Universität Düsseldorf) [80]. Under the assumption of a medium effect size (f=0.25), a minimum of 240 participants (80 per condition) were needed to test the expected effect within each of the 5 MH problems via ANCOVAs with a significance criterion of α=.05 and a power of 90%. Statistical analyses were performed using R (version 4.3; R Foundation for Statistical Computing) [81] and SPSS (version 28; IBM Corp) [82]. R was also used to generate the random allocation sequence. Authors involved in data analysis and interpretation were not blinded with respect to the assigned experimental conditions.

In the beginning of recruitment, the upper age limit was raised from 25 years originally to 29 years due to the aforementioned findings of previous research.

Ethics approval was obtained from Ethics Committee I of the Heidelberg Medical Faculty on July 27, 2020 (protocol S378/2020). The procedures were in accordance with the Helsinki Declaration of 1975, as revised in 2000. All participants received information about the study’s aims, scope, procedures, data processing, and data storage on the study website in written form. Furthermore, all participants received contact information if they wished to clarify questions via telephone or email. Only participants who provided their informed consent through a web-based checkbox were eligible for participation. Participants were able to opt out of the study at any time by closing the study website, which they were informed of before their participation. Participants were offered to take part in an optional gift card lottery at the end of the study (100 gift cards of €20 [US $21.58] each). If they were interested in the lottery, they could enter their email address. Email addresses were stored separately from other study data and user IDs to ensure anonymity. All other data were collected and are reported anonymously. Thus, this study does not contain any individual data of identifiable participants.

Figure 1 shows the flow of participants. Of the 2208 participants who were randomized to 1 of the 15 conditions, 472 (21.38%) were excluded because their time spent on the video pages fell below the durations of the videos they were assigned to, indicating that they did not view the entire videos. Of the remaining 1736 participants, 342 (19.7%) were excluded due to incomplete data (ie, they did not complete all the relevant scales that the study entailed [beginning with informed consent up to and including the last acceptability item]). Our final sample consisted of 1394 youths aged 14 to 29 years (mean 20.97, SD 3.67 years). Sociodemographic and screening results are shown in Table 1.

A total of 79.56% (1109/1394) of the sample identified as woman or girl, and 44.76% (624/1394) were help seekers (ie, they used professional MH services at the time of or before data collection). On average, the youth were moderately anxious (mean GAD-7 score 8.38, SD 5.00) and depressed (mean PHQ-9 score 9.56, SD 6.07). While 22.88% (319/1394) reported abstinence in the AUDIT-C, 30.85% (430/1394) indicated hazardous alcohol consumption. A total of 19.58% (273/1394) were at high risk of developing an eating disorder according to the WCS. One-third (479/1394, 34.36%) of the sample reported a lifetime history of NSSI according to the Self-Injurious Thoughts and Behaviors Interview, with a 12-month prevalence rate of 19.01% (265/1394).

The main results are presented in Table 2.

Figure 2 summarizes the results of the overall efficacy and the MH issue–specific subgroup analyses graphically. Specific results of the subgroup analyses can be found in Multimedia Appendices 1, 2, and 3.

On the 7-point scale of the GHSQ, most participants (1046/1394, 75.04%) selected a score of ≥4 (CG: 409/554, 73.8%; intervention 1: 316/410, 77.1%; intervention 2: 321/430, 74.7%). In total, 19.23% (268/1394; CG: 105/554, 19%; intervention 1: 74/410, 18%; intervention 2: 89/430, 20.7%) of participants reported a score of 7 (“extremely likely”), whereas 6.74% (94/1394; CG: 47/554, 8.5%; intervention 1: 24/410, 5.9%; intervention 2: 23/430, 5.3%) responded with a score of 1 (“extremely unlikely”). Across all MH problems, no statistically significant group main effect was found on potential professional help seeking (F_2,1385=0.99; P=.37; Table 2).

This study developed and tested the short-term effectiveness of 2 brief video-based strategies targeted at adolescents and young adults (aged 14 to 29 years) aiming to foster potential professional help seeking (main outcome) and related attitudes for 5 MH problems. In the total sample, we did not find effects of either intervention 1 (psychoeducation) or intervention 2 (positive outcome expectancies) on our primary outcome. However, significant group effects were found with respect to potential informal help seeking, stigma toward others, and indifference to stigma in the total sample. While both intervention groups showed more favorable attitudes than the CG with regard to public stigma, this did not translate to participants’ own indifference to stigma. In this case, the CG showed significantly more positive attitudes in comparison to intervention 1. However, this finding was not apparent in the MH issue–specific subgroup analysis with the exception of GAD. Unintended adverse effects of MH interventions have been reported in previous research [83-85], which underlines the need for thorough evaluations of such interventions before their public dissemination. Accordingly, we would advise against the implementation of our GAD interventions at the current stage and would recommend the development and evaluation of other tailored strategies for this MH problem.

With regard to informal help, participants in intervention 2 showed a greater willingness to approach friends, family members, or romantic partners for help than participants in intervention 1. This might have been due to the interventions’ design as intervention 2 explicitly depicted improvements in social relationships after the main characters in the videos had sought professional support. Overall, the videos were well accepted and rated as quite interesting, with some room for improvement and with the videos of intervention 1 outperforming those of the other 2 conditions. All videos were, on average, rated as very comprehensible, and no significant group differences were observed in this regard. Interestingly, participants felt more transported into the narratives in the CG and intervention 1 as compared to those in intervention 2. As intervention 2 followed a narrative approach, continuing Paul and Paula’s vignette stories, this was surprising. However, as previously stated, intervention 1 was generally more liked and viewed as more interesting in comparison to intervention 2. The animated and fictional third-person approach of intervention 2 seemed to not have sparked as much interest in participants as the facts presented in intervention 1. While we aimed to increase identification with our main characters through the alignment of their genders with those of the participants, intervention 2 might have been insufficient with regard to the perceived “realness” of the story and the characters, which has been identified as crucial for the formation of narrative transportation and, in turn, attitudes and intentions [86]. The rather optimistic portrayal of the help-seeking process might have contributed to a lack of perceived authenticity in this sample as well. Furthermore, implicit MH statements in the videos’ scenarios could have been more fruitful. For instance, the viewer could have watched directly how Paul and Paula discussed their issues with a psychotherapist rather than having a narrator describe the situation to them. These types of videos have been associated with improved health literacy and more beneficial attitudes toward cervical cancer [87], and their application to the field of MH would be interesting.

We further observed differential outcomes with respect to the assigned MH problems and participants’ actual help-seeking status. While the CG outperformed either intervention 1 (total) or both interventions (subgroup of non–help seekers) for GAD with regard to stigma indifference and no further outcomes were found for GAD, different patterns emerged in the other MH issue groups. Results were mixed, where both intervention 1 and intervention 2 outperformed the other conditions in some of the outcomes but not in others (eg, in depression), or a clearer tendency toward the superiority of one of the interventions emerged (eg, in NSSI). In summary, our results point toward the usefulness of tailored interventions with regard to MH issue type and previous help-seeking experiences of potential target groups. Our finding that different strategies might work differently for each of the 5 MH problems included in our study is in accordance with those of previous research. For example, Ebneter and Latner [78] found varying stigmatizing attitudes among different MH problems. The participants in their study blamed a vignette character with an eating disorder more for their condition than a character with depression, whereas the latter was regarded as more impaired. Our finding that a destigmatizing and psychoeducational intervention such as intervention 1 might work better for bulimia fits their recommendation to target stigmatizing attitudes toward specific MH problems [78]. Similarly, alcohol dependency and self-endangering behaviors were perceived as particularly dangerous in a Swiss vignette study [88]. The humanizing depiction of our NSSI and problematic alcohol use characters within a framework of close supportive relationships, which improved in quality through psychotherapy, might have been a relevant factor for reduced distrust scores in the intervention 2 condition as compared to the CG and in the case of non–help seekers in the NSSI condition as compared to intervention 1 as well. This approach might be advantageous to reduce public stigma regarding MH problems that are viewed as particularly dangerous. Decisions for one or another interventional strategy may also depend on the specific goal and targeted outcome. While more research is needed, our study provides preliminary evidence for the tailored strategies suggested in Table 3.

One limitation of this study lies within the sole investigation of effects on hypothetical intentions and attitudes instead of actual help-seeking behavior. While intentions are substantially associated with behavior and provide valuable insights, they do not translate directly to behavior change [49]. Moreover, only short-term effects were investigated. Previous studies on short video-based interventions have demonstrated sustaining destigmatizing effects for 1 [40] and 5 months after their delivery [43]. While potential long-term effects of our interventions in particular remain unknown and might be investigated in the future, the current state of research points toward potentially impactful long-term effects of low-threshold microinterventions. Related to this, we did not investigate dose-response effects. Research on optimal doses (ie, durations, frequencies, and amounts of intervention components) needed for sustainable change through microinterventions and interventions in general is crucial for well-founded recommendations for or against specific interventions [89].

Just as in other microintervention studies among youth [39], group differences in our study were small. However, the potential high reach of easily accessible, low-threshold interventions such as the ones evaluated in this study is apparent in the final sample size of 1394. As it depends on both effectiveness and reach, this allows for a comparably high public health impact [23,24]. We also included trend effects in our overview (Figure 2), which should be interpreted with caution. However, these findings might be useful to inform the planning of subsequent research in this field.

Future research should focus on improvements in the effectiveness of microinterventions. One approach could be the investigation of interventional framing. In a Japanese study on depression, loss-framed messages (ie, emphasizing negative consequences of refraining from help seeking) had a greater impact on help-seeking intentions than gain-framed (positive consequences of help seeking) or neutral (eg, prevalence rates) messages as well as unformatted, plain-text messages in middle-aged adults [90]. Thus, it might be interesting to conduct future studies on the effects of video-based microinterventions with differently framed messaging as the videos in this study emphasized potential gains of help seeking rather than potential losses of help seeking restraint.

Furthermore, our sample showed, on average, a high level of education, very little public stigma, a pronounced willingness to seek help, and a high rate of actual professional help seeking (624/1394, 44.76%), which limits the generalizability of our findings. More than 90% (1285/1394, 92.18%) of our sample knew someone with MH problems. While we were careful not to recruit MH experts, such as university students of medicine and psychology, youth with a personal interest in MH-related topics seemed to have been more inclined to participate. Related to this, we aimed for a community youth sample rather than a clinical sample. The scenarios that our items referred to were hypothetical and did not necessarily reflect participants’ own experiences due to the random assignment to 1 of the 5 MH problems. A similar approach with targeted interventions according to youth’s actual MH status and more individualized elements with regard to gender-related [91] and cultural [92,93] aspects could be promising in future research. Thus, upcoming studies should strive to align their research objectives more closely with the characteristics and needs of the selected target groups.

Finally, we did not include a comprehension check to assess participants’ understanding of and engagement with the content presented in the videos. While we accounted for the time participants spent on the video pages and only included participants with sufficient durations of stay in the final analyses, they may not have fully comprehended or attended to the video material. The substantial number of excluded participants who completed the study without meeting the time threshold (472/2208, 21.38%) underlines this potential issue. Thus, future studies should address this limitation by including comprehension checks to improve the robustness of the findings.

The low uptake of professional MH services in youth reflects the need for appropriate strategies to facilitate professional help seeking. This study investigated the effectiveness of 2 short video-based strategies targeted at youth (aged 14 to 29 years) on potential professional help seeking and related attitudes for 5 MH problems. While we did not find intervention effects on potential professional help seeking (with the exception of previous help seekers in the depression conditions), differential intervention effects depending on each MH problem and participants’ actual help-seeking status were found in our secondary outcomes, such as public stigma. Our study results can be used to inform the development of new antistigma interventions, which, based on our results, we would recommend tailoring to specific MH problems, target groups, and outcomes. While group differences were small, such low-threshold interventions can be easily disseminated and, therefore, hold potential for a high reach and, thus, a meaningful impact at the population level. More research is needed for more robust and generalizable recommendations.