We hosted two meetings with 25 stakeholders from relevant fields ((preventive) mental healthcare, youth work, and education). The stakeholders brought the necessary developmental, contextual, and pedagogical expertise to develop guidance on adapting the concept of first aid for mental health problems (Stroobants et al., 2023) for children and adolescents. The panel addressed four key questions: (1) What objectives should the materials achieve?; (2) What topics should be covered?; (3) Which age groups would benefit most from this type of material?; (4) Which settings and didactic methods should be used?

To prepare for collecting evidence to support the recommendations for training material developers, we focused on identifying relevant topics. Unlike the materials developed for adults, the panel advised against focusing on specific symptoms or conditions (e.g., depression, anxiety, or suicidal ideation). Instead, they emphasized that an age-appropriate approach to psychosocial support should address general signs of distress, coping with difficult emotions, and peer-interaction challenges as core themes. These priorities effectively served as the inclusion criteria for determining which topics should guide our evidence gathering, while symptom- or diagnosis-specific approaches were excluded.

Guided by these stakeholder-defined priorities, we formulated two research questions (RQ): (1) “In children and adolescents aged 5–18, what risk or protective factors related to peer interaction and communication are associated with mental health outcomes?”; (2) “In children and adolescents aged 5–18, what are the effects of educational programs promoting social skills, recognition of mental health problems, responses to peers’ mental health issues, or guidance to professional help on mental health or related knowledge, skills, attitudes and behaviors?”.

We systematically collected evidence on mental health risk or protective factors related to peer interactions among children/adolescents (RQ1) and the effectiveness of educational programs targeting their social skills and mental health (RQ2). Both reviews adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Moher et al., 2009).

We translated the evidence conclusions into draft recommendations based on a content analysis of the identified factors (RQ1; from the questionnaires used to measure these factors) and educational initiatives (RQ2). These were supplemented with “Good Practice Points” (GPPs) extracted from scientific articles, (inter)national training materials, and content discussions among authors and their colleagues. GPPs reflect expert-agreed advice without a systemically identified evidence base.

We formulated two types of recommendations: (1) content recommendations describing the desired characteristics of interactions between children and young people, to guide the content of initiatives; and (2) didactic recommendations on intervention characteristics to stimulate supportive peer relationships, guiding structural features, didactic approaches, and overarching content. The draft recommendations were thematically organized into a preliminary guidance document. For each recommendation, we listed supporting references (identified during the systematic searches and/or from additional sources) and the target group (children (5–11 years), adolescents (12–18 years), or both (5–18 years)).

Next, we systematically collected expert input from a large stakeholder panel (partly overlapping with the initial group, see Defining research questions for systematic literature searches), including professionals, peer workers, and volunteers from academia, mental health care, prevention, youth work, and education (see Acknowledgements for a complete list). These stakeholders were selected based on their complementary forms of expertise: academic knowledge on child and adolescent development and mental health, practical experience from daily work with young people in clinical, educational, and community settings, and lived experience through peer workers who understand firsthand the challenges young people face. Together, this spectrum of expertise ensured that the guidance was developmentally appropriate, contextually grounded, and sensitive to real-world experience. We formalized this step using a modified Delphi consensus method (Fitch et al., 2001; Hasson et al., 2000; Murphy et al., 1998). Panel members received the preliminary guidelines and completed an online survey on the draft recommendations. The Delphi process consisted of two rounds with a live panel discussion in between. Recommendations were presented in two cycles (two-week interval) for practical reasons (see Figure 1). In the first round, panel members rated each recommendation on a 5-point scale (“essential”, “important”, “I don’t know/doubtful”, “not important”, or “do not retain”) and could justify their rating or suggest changes. They could also propose new recommendations if they identified gaps. Before the live discussion, members received anonymized feedback comparing their responses with panel-wide results. Consensus was defined as ≥80% rating a recommendation as “essential” or “important”.

Recommendations not reaching consensus in the first round (<80%) were discussed in a face-to-face meeting moderated by the panel chair (EC). After presenting the first-round results, each recommendation was discussed individually. The panel considered the effectiveness and certainty of the underlying evidence, pros and cons, risks, preferences of the target group (children and adolescents), and suitability for the Flemish context. Based on this input, a (possibly revised) recommendation was included in the second online survey round, where participants could reconsider their ratings. Recommendations still not reaching consensus (<80%) after the second round were excluded from the final guideline.

We developed didactic materials for children based on the final guideline and identified needs. The developmental process was informed by principles of human-centered design (Altman et al., 2018), incorporating stakeholder consultations, a structured environmental scan of existing materials, and ethnographic approaches through immersive fieldwork. During the preparatory phase (see Defining research questions for systematic literature searches), we also gathered stakeholder input on the priority target age group, preferred setting, and format of materials for tailored psychosocial support. We conducted preliminary research on existing Dutch and English materials, games, and programs targeting mental well-being in children and adolescents. This was done by targeted online searches and inquiries with Flemish organizations in mental health care, prevention, and promotion. Key characteristics of the identified initiatives were systematically extracted and recorded using a standardized data collection template: name, organization, target age, theme, goal, setting, and method. These findings, together with the stakeholder input, enabled us to determine the initial target group and approach. We then conducted immersion visits to gain an in-depth understanding of the target audience and contextual factors relevant to implementation. These visits included structured observations of primary school children during class and break times, as well as interviews with primary school teachers. Insights from these activities informed the refinement of our methodological approach and guided design considerations for the didactic materials. The final design of the didactic materials was subsequently aligned with the recommendations of the guideline.

As indicated, we formulated two research questions (RQ): (1) “In children and adolescents aged 5–18 years, which risk or protective factors related to peer interaction or communication are associated with their mental health?”; (2) “In children and adolescents aged 5–18 years, what is the effect of educational programs aimed at acting on mental health (problems), social skills, recognizing signs of mental health problems or guiding peers to professional help on their mental health or their mental health-related knowledge, skills, attitudes, and/or behavior (outcome)?”.

For RQ1, we retrieved 39407 references from the searched databases, of which 579 studies met the selection criteria. For these studies, each eligible risk or protective factor was tabulated in an Excel file, together with the corresponding study characteristics, resulting in 1436 data lines. Ten categories were created post hoc to group the factors based on thematic analysis: (pro)social behavior, communication content, perpetration (e.g., bullying, aggression), victimization (e.g., being bullied or exposed to aggression), relationship quality and intensity, being loved and being part of the group, social rejection, social influence, social support, and social skills. We then applied the “best available evidence” principle to determine which studies should undergo full data extraction. We excluded cross-sectional studies, individual victimization studies (as systematic reviews sufficiently covered this topic), and studies lacking a clear description of how the factor was measured. We also excluded studies published before 2000, due to substantial changes in peer interactions following the rise of the internet and social media, which accelerated online social networking (Shah et al., 2019). After applying these criteria, 157 studies remained, corresponding to 341 data lines (see Figure 2 for the study selection flowchart and Supplementary file 3 for the tabulated study characteristics).

The final set included 150 cohort studies, five case-control studies, and two systematic reviews. Most studies (n = 96) were conducted in North America, followed by Europe (n = 49) and Oceania (n = 11). Only 24 studies focused specifically on younger children (5–11 years). The remainder included adolescents aged 12–18 years, or mixed samples of primary and secondary school children (5–18 years).

From the 157 included studies, we extracted 30 risk factors, 12 protective factors, 18 factors with no statistically significant association with mental health outcomes, and four factors for which the evidence was inconsistent. All evidence was of very low certainty. An overview of the identified risk and protective factors is presented in Table 1, and the detailed data extraction for each factor, organized in the ten thematic categories, can be found in Supplementary file 4.

For RQ2, we screened 2404 references and included nine systematic reviews (see Figure 3 for the study selection flowchart). Six reviews contained only experimental studies ((non-) randomized controlled trials), and three included both experimental and observational research. Five reviews focused solely on studies from Western countries, while four also included studies from low-and middle-income countries.

The reviews covered a range of educational programs (see Table 2): interventions targeting help-seeking for mental health problems (Aguirre Velasco et al., 2020), social skills training programs (de Mooij et al., 2020), school-based programs promoting intra- and interpersonal domains (Mertens et al., 2020), a universal social-emotional learning program (“The Second Step”) (Moy et al., 2018), an anti-(cyber)-bullying program (Ng et al., 2022), teen Mental Health First Aid (Ng et al., 2021), adolescent dating violence prevention programs (Russell et al., 2021), mental health literacy programs (Seedaket et al., 2020), and multicomponent positive psychology interventions (Tejada-Gallardo et al., 2020). Three reviews (Aguirre-Velasco et al., 2020; Ng et al., 2021; Seedaket et al., 2020) were used only as study sources because not all included studies met our eligibility criteria.

We found low to moderate certainty evidence supporting all program types, although results varied by program type, specific training components, structural characteristics, and data type (see detailed findings and evidence conclusions in Supplementary file 5).

The evidence identified for RQ1 and RQ2, together with the “Good Practice Points” (see Methods), was used to draft a guideline containing content and didactic recommendations, which was then submitted to the expert panel (see Methods). Table 3 presents the results of the two survey rounds. In total, 146 recommendations were presented in the first round, and the panel reached consensus on 75. During the panel discussions, the remaining recommendations were revised, split, or left unchanged. In the second survey (after the panel discussions), 78 recommendations were resubmitted, and the panel reached consensus on 37 additional recommendations. In total, 112 recommendations reached consensus, while 41 recommendations were not retained.

Content recommendations were grouped into two categories: “Positive, strengthening contacts between children and adolescents”, and “Handling each other’s perceptions and feelings (in case of psychological distress)”. Each category contains 6 subgroups. Didactic recommendations were organized into 4 categories (target group, setting, content, and form). The final guideline structure and example recommendations are shown in Table 4.

We developed an initial set of didactic materials based on the guidelines, stakeholder input (see Defining research questions for systematic literature searches), an analysis of existing initiatives, and insights from immersion visits.

The panel recommended creating materials for various settings that complement existing initiatives, emphasizing interactive, easy-to-share, and free content to increase accessibility. Our preliminary research identified about 80 Dutch and English initiatives, materials, games, and programs. All collected initiatives were thematically analyzed to identify gaps in age groups, settings, and pedagogical approaches. Packages were identified for children and mostly adolescents, mainly focusing on specific mental health themes. Peer support programs existed for adolescents but not for younger children.

Consequently, we developed materials for children aged 8–10, an age just before the first signs of puberty, to fill this gap while ensuring compatibility with existing resources and expert advice. Based on immersion visits, play was chosen as the guiding formative method. This approach is supported by pedagogical literature demonstrating that play-based and game-based learning can enhance engagement, academic performance, and social–emotional outcomes in childhood (e.g., Plass et al., 2015; Skene et al., 2022). These frameworks emphasize learning embedded in children’s natural environments, aligning with our objective of enabling children to practice supportive behaviors in contexts familiar to them. The developed game-based materials follow the guideline recommendations and take the form of ready-to-use, interactive, free digital packages for adults working with elementary school children. Extensive instruction manuals were created to help adult supervisors provide appropriate guidance. The primary aim of these materials is to encourage children to support peers experiencing distress, offering age-appropriate psychosocial support.

Materials were tailored to educational, recreational, and family settings, reflecting children’s main environments. Manuals and methods were adapted accordingly: a board game for schools, an audio story for families, and a story-based interactive game for leisure (in Dutch: Eerste hulp bij moeilijke gevoelens [First aid for difficult feelings]; see Table 5). Practice tests with the materials were conducted in all three environments before being finalized. All materials are freely available for download on the Belgian Red Cross-Flanders website (https://jeugd.rodekruis.be/voor-jou/eerste-hulp-bij-moeilijke-gevoelens).