We report the development of a repository following a convergent design, in which quantitative and qualitative data are concurrently collected aiming at combining their results to obtain a more complete understanding of the topic [17]. Research participants were key informants from the general and professional community involved in the welfare and health care systems, namely children and adolescents, their families, mental health care professionals, pediatricians, teachers, and NGO members. The research was approved by the Research Ethics Committee of the Democritus University of Thrace [approval number: ∆ΠΘ/ΕΗ∆Ε/42772/307]. All participants signed written informed consent; underage participants filled assent forms, and written consent was granted by their legal guardians. Survey data was collected and preserved according to the General Data Protection Regulation (GDPR) National Policy [18], being handled unidentified and kept under a password-protected system with access restricted to research members for a one-year period. Relevant material and records concerning the development of this research can be accessed on our webpage located on the Open Science Framework [https://osf.io/crz6h/], a platform for managing and storing shared knowledge projects [19].

In the quantitative section, nationwide samples of the informant groups were assessed between September 2022 and January 2023 through a cross-sectional survey composed of validated instruments and questions covering areas of interest (see below for detailed recruitment strategies). The survey with children, adolescents, and caregivers assessed: (a) frequency of mental health problems; (b) mental health needs; (c) mental health literacy and stigma; (d) mental health services and access to care. The survey with teachers and healthcare professionals assessed: (1) mental health literacy and stigma; (2) professional practices and experience; (3) training background; (4) training needs and perspectives. For this arm of the research, we followed the study design items of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) (checklist available at https://osf.io/crz6h/) [20].

As for the qualitative section, focus groups were conducted to reach an in-depth comprehension on the target topics and generate insight on solutions with the active participation of community members. To ensure the representation of vulnerable populations, some groups were composed of members of LGBTQIA + groups, refugees, or ethnic minorities. Groups were focused on assessing: (a) views on wellness, mental health, and mental health problems; (b) mental health stigma and discrimination; (c) barriers, facilitators, and opportunities for mental health care; (d) mental health services and access. For teachers and health professionals, discussions also focused on (e) training needs and perspectives. For the items pertaining to this arm of the project, we followed the items on data collection and design from the Consolidated Criteria for Reporting Qualitative Research (COREQ) (checklist available at https://osf.io/crz6h/) [21].

Table 1 shows the selected instruments and Supplementary Table 1 displays the scope of developed questions that were used in the survey, whilst Fig. 1 and Fig. 2 depict how they were applied to each sample. The initial part of the survey consisted of general socio-demographic questions collecting relevant data from all respondents, including preliminary questions on mental health for children, adolescents, and caregivers (whether they face mental health problems, have a diagnosis, medication use, and professional assistance). Then, for each domain of inquiry, we consulted the International Consortium for Health Outcomes Measurement (ICHOM) on patient-relevant outcomes [22, 23]. We consulted the literature on locally validated instruments assessing such outcomes considering their briefness, availability, and reliability, as per the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) (see Supplementary Table 2 for their characteristics and psychometrics) [24]. If no validated Greek version was available, we performed a cross-cultural adaptation of a selected instrument following a five-stage validated procedure (detailed registers of the adaptation are available at https://osf.io/crz6h/) [25]. For a few topics (training background, training needs and perspectives, professional practice and experience, service use and access, and mental health needs), we could not locate adequate instruments covering sufficient aspects of interest, so we developed a set of questions (see Supplementary Table 1 for their scope). In this procedure, a team of local and international experts within our research group formulated survey items considering the context of participants, which were then selected and refined after extensive discussion until reaching a final set of questions. The full version of the questionnaire is available at [https://osf.io/crz6h/].

Survey design: participants and instruments (caregivers)

Survey design: participants and instruments (children, adolescents, health professionals and teachers)

A nationwide sample of caregivers was recruited to a self-applied online questionnaire according to regional and offspring gender and age quotas following the census distribution (see Supplementary Table 3 for quotas) [26]. The questionnaire consisted of questions and instruments on service access and use, literacy and stigma, parenting practices, and mental health symptoms of general and specific conditions (see Fig. 1 for details of which instruments were applied to the sample and its subsets). Participants were reached through an online respondent panel (a list of individuals who agreed to participate in surveys) provided by the research company IQVIA OneKey [27]. This panel was developed based on a census frame and its members were rigorously profiled using over 500 data points. Recruitment occurred online via social media and website campaigns, search engine optimization (SEO), panelists’ friends referrals, and affiliate networks. To avoid self-selection, surveys were automatically routed to respondents based on an algorithm that combines representativeness (via random allocation), appropriate frequency of sampling for each panelist, and management of sample quotas if required. For each survey, a pre-specified maximum number of 1.400 participants applies. Panelists who have been profiled as parents or legal guardians of children under 18 years old were invited to this survey following the algorithm set to manage regional location quotas. A screening question was applied to ensure that the children/adolescents they cared for were between 1.5 and 17 years old at intake.

Surveys were conducted with two groups of children and adolescents answering different sets of instruments (see Fig. 2 for details). The first group consisted of the offspring of the surveyed caregivers, who were invited to a self-applied online questionnaire with instruments measuring mental health symptoms to match equivalent instruments rated by their caregivers. They provided measures on service use and mental health symptoms (including general psychopathology, screen use, and anxiety and depression). A second group was selected by random landline phone calls that recruited children and adolescents aged 8 to 17 years old across the Greek territory (maximum of one per household) to a telephone survey, following census quotas on region, gender, and two age groups (see Supplementary Table 4 for quotas) [26]. An automatic dialer system generated landline and mobile phone numbers to make automatic calls after removing non-operational numbers. This group answered general psychopathology measures, as well as mental health stigma and discrimination questionnaires based on the mental health vignettes (further described in the qualitative arm). Adolescents aged 12 to 17 years old in this second group also filled instruments on self-harm behavior and substance use disorder, as well as questions on gender and sexuality.

A list of 274 regular and special education schools were randomly sampled from the five key areas where the CAMHI hubs are located (Athens, Thessaloniki, Ioannina, Alexandroupoli, and Heraklion). Schoolteachers were recruited in person at the place of work, and then phone call interviews were held at the respondents convenience with a scheduled appointment. To ensure socio-demographic representativeness, the samples were proportionally distributed across and within districts according to the population size following census distribution (see Supplementary Table 3 for quotas) [26]. Teachers answered questionnaires on demographics, mental health literacy and stigma, professional experience, training skills, and perspective on training needs (see Fig. 2).

Healthcare professionals from different fields of practice (child psychiatrists, psychologists, occupational therapists, family physicians, pediatricians, psychiatrists, social workers, and nurses) were randomly sampled from a proprietary database (the IQVIA OneKey) containing contact and profiling information on the specialties intended for the survey [28]. Sampling was done randomly within each specialty area, respecting geographical distribution quotas following the database profiling across the five hub areas designated (see Supplementary Table 3 for quotas). Participants were called and invited to respond to questionnaires that were adapted according to speciality (Fig. 2). Interviews were held online with a computer-assisted interview program (CAWI), being guided by an IQVIA executive by phone to maximize completion rate and quality.

Descriptive analysis was performed to provide measures of central tendency, frequency, and dispersion for each instrument or subset of questions. As relevant, analysis was stratified by subgroups. Statistical analysis was performed in the software R version 3.6.2 [29].

We conducted 14 focus groups (10 online and 4 in person) of 60 to 115 min duration, each composed by 8 to 10 members of an informant group (see Table 2 for a detailed description). Participants were recruited from the five cities where the CAMHI hubs are located after an initial online search for schools, parents’ associations, health centers, professional associations, LGBTQIA + organizations, and legal channels. We contacted these institutions to obtain a list of potential participants who were then invited to participate in the groups via a telephone call. Pomak and Roma adolescents were recruited through a cultural mediator involved in the community. All participants received a 100-euro reward as compensation for their time. Details of the recruitment process according to each group can be seen in Supplementary Table 4.

A vignette-based discussion was carried out to illustrate children and adolescents facing mental health disorders and prompt impressions from participants (see Supplementary Table 5 for complete vignettes). As illustrative instances of internalizing and externalizing prevalent conditions, the children's vignettes featured cases of social anxiety and attention-deficit/hyperactivity disorder (ADHD), while adolescents' vignettes featured depression and conduct disorder. The vignettes on social anxiety and depression were adapted to Greek based on the Australian National Survey on Youth Mental Health Literacy, a work assessing beliefs towards mental health [30]. The vignettes on conduct disorder and ADHD were constructed by our research team following a similar process. All vignettes were adapted according to the cultural background of each focus group by using a common name in that language.

Each focus group was guided by a local moderator with experience in the field and the assistance of manuals containing anchor questions (for the structure of the manuals, refer to the Open Science Framework webpage at https://osf.io/crz6h/; see Supplementary Table 5 for moderators’ credentials). Researchers had no previous relationship with the participants, and disclosed they were professional scholars conducting a work for deepening the understanding on the research topics. The manuals were constructed by an international panel of specialists within our research team in a brainstorming process that generated candidate questions that were then selected, discussed, and refined to reach a final version of anchor questions for the manual. To guarantee that the manuals were context-sensitive, a multicultural panel of specialists with local and external expertise participated in the process, including professionals of diverse fields of activity and experience with refugees and other vulnerable groups. Manuals were adapted according to necessities and specificities of each focus group.

The focus groups’ sessions were video recorded, resulting in a written transcript in the spoken language of the group (Greek or Farsi) and a translated version to English. Written notes were taken in real-time as the group occurred. Immediately following each focus group, the research team conducted a debrief session with the moderator to gather top-of-mind perceptions, using the heightened awareness of the moment to generate initial ideas for analysis and insights to guide future focus group sessions. Considering data was translated to English to be made available for analysis, transcripts were not returned to participants for comment or correction. As a first processing of the transcripts, we summarized the discussion contained within each focus group, providing a narrative description of the major points elicited in the sessions and its supporting quotations.

To illustrate the content of the focus groups, the research team (LEM, EV) looked for pertinent discussions about mental health concepts and stigma, which were purposefully sampled for a brief report. This was examined by attending general principles of inductive thematic analysis, and coding was performed by one researcher (LEM) and discussed in regular meetings with other members of the team (JLF, GAS, EV) [31]. For each focus group, we looked at the selected data to detect meaningful patterns in responses. Initial codes were generated and then clustered into organizing central themes to delineate broad findings. We then compared those findings across the different groups, usefully arranging under four overarching populations that most accurately nested results: professionals, caregivers, children/adolescents of general population, and children/adolescents of underrepresented minorities (namely, Pomak, Roma, LGBTQIA + , and refugee adolescents). The findings were then outlined as bullet points in the form of sentences that more succinctly described their attributes.