IntroductionRelevance of mental health in adolescence

Adolescent mental health is a major global health issue, with more than 30% of all mental disorders beginning by the age of 14 and 60% by the age of 25.1 Notably, the subclinical symptoms of mental disorders (ie, mental health problems that do not fulfil the number and/or duration of symptoms required by the Diagnostic and Statistical Manual of Mental Disorders (DSM) or the International Classification of Diseases (ICD) to assign a diagnosis) start even earlier.2 3 Many mental disorders persist into adulthood,4 with poor long-term outcomes including higher levels of drug abuse,5 self-harm and suicidality,6 and negative effects on educational attainment,7 social relationships8 and physical health.8 9 Furthermore, mental health problems do not only place a significant burden on family and friends,10 11 but they also have important societal implications as they lead to increased healthcare costs and reduced economic productivity in the long term.12 If mental health disorders remain untreated, the risk of developing comorbid mental health problems increases,13–15 particularly in the first year after onset,13 16 17 leading to even higher levels of distress, functional deficits and need for care.17–19 Therefore, adolescence seems to be an optimal window of opportunity to prevent mental health problems and their associated burden. The promotion of mental health in this age group is generally regarded as a key priority and is globally recognised.20–22

The need to improve early detection in adolescence

Despite the significant impact of mental health problems at the individual and societal levels, early detection of mental health problems in adolescence is challenging as only a minority of those in need of care actively seek help.23 24 The main obstacles to help-seeking in adolescence are perceived stigma and embarrassment, a preference for self-reliance, and poor mental health literacy among adolescents, parents and teachers, including difficulties in recognising mental health problems.4 25 26 Consequently, adolescents have the most limited access to mental health services across the lifespan and tend to not seek help until numerous mental health problems and psychosocial difficulties have developed.24 27 This might also contribute to the risk enrichment found in individuals seeking help at specialised early intervention services,28 and highlights the promise of a community-based approach to detecting early signs of emerging psychopathology even among non-help-seeking adolescents and before the development of comorbid disorders and functional difficulties.29–31

From specific risk syndromes to transdiagnostic mechanisms

Despite long-standing efforts to define risk syndromes for specific mental disorders, such as psychosis, depression or bipolar disorder,32 there is increasing support for a transdiagnostic approach that goes beyond traditional diagnostic boundaries and that recognises psychopathology as being dimensional, highly dynamic and changeable in clinical picture with high rates of comorbidity between mental health problems.33–35 This is well in line with concepts such as the Research Domain Criteria,36 37 the Hierarchical Taxonomy of Psychopathology model38 and staging models.39

In this study, ‘transdiagnostic’ refers to the mechanisms causing and maintaining multiple mental disorders.40 These mechanisms can be linked to symptoms of multiple mental disorders across diagnostic categories and could therefore also explain comorbidities.3 40 Several transdiagnostic mechanisms underlying a range of mental disorders with a typical onset in adolescence have already been derived theoretically and empirically, including behavioural avoidance,41–43 repetitive negative thinking,44 45 intolerance of uncertainty,14 emotion regulation,46 47 self-perfectionism48 49 and rejection sensitivity.50–52 Many cross-sectional studies have found significant associations between these transdiagnostic mechanisms and mental health problems in adolescents,53–55 but only few studies have investigated these relationships in adolescent samples longitudinally. They examined the effect of a single47 48 52 56 or a few transdiagnostic mechanisms (ie, sleep–wake disturbances and rumination,57 or experiential avoidance, rumination and emotion dysregulation58) on the development of one or two mental disorders (ie, anxiety and depression). However, this hampers the comparison of the relative strength of these transdiagnostic mechanisms.59 Furthermore, with regard to subclinical symptoms, only two studies have investigated the effect of more than one transdiagnostic mechanism on the course of these subthreshold mental health problems in adolescents.57 60 One of these studies explored the effects of intolerance of uncertainty, negative affect, and emotion regulation strategies on the subclinical levels of anxiety and depression.60 The second study examined the role of rumination and sleep–wake disturbances in depressive symptom progression in at-risk youth.57 These studies found that intolerance of uncertainty, negative affect and difficulties in emotion regulation, as well as sleep–wake disturbances, were significantly associated with the development, progression and maintenance of subclinical levels of depression and anxiety. Given that both studies only looked at anxiety and/or depressive symptoms, this study will prospectively investigate the effect of several transdiagnostic mechanisms on subclinical symptoms of a broad range of mental disorders.

Dynamic modelling of emerging psychopathology

Regarding the temporal course of emerging psychopathology, symptoms seem to vary considerably over time, gradually differentiating in a non-linear way into more distinct syndromes with increasing diagnostic specificity.61–68 Thus, the challenge is to get a detailed understanding of how subclinical symptoms emerge, stabilise or fade over time and eventually develop into full-threshold disorders. There is evidence for both a homotypic course (ie, symptoms belonging to one category evolve into the same category) and a heterotypic course (ie, symptoms belonging to one category evolve into another category) of psychopathology across the lifespan.69–71 However, many previous studies were interested in the course of mental disorders instead of focusing on subclinical symptoms.72–74 Recent studies investigating the course of subclinical symptoms in adolescents have mainly been limited by focusing on the subclinical symptoms of specific mental disorders, namely depression,75 76 anxiety77 78 or psychosis.79 However, this precludes the investigation of the heterotypic aspects and does not take the transdiagnostic nature of emerging psychopathology into account.80–82 Moreover, many of these studies were based on few assessment points over a period of several years.77–79 However, to better capture the dynamics and complexity of emerging psychopathology, models with several assessment points taking interindividual and intraindividual changes in symptoms into account are warranted.34 65 Thus, this study will assess the subclinical symptoms of various mental disorders multiple times to model the dynamic course of the homotypic and heterotypic patterns and the association with transdiagnostic mechanisms, level of functioning and onset of mental disorder.

Network approach of subclinical symptoms

The network approach to the analysis of emerging psychopathology83 84 has rapidly gained substantial influence in the field of mental health and prevention.84–87 The core assumption is that a mental disorder emerges from a complex system of symptoms that interact dynamically with each other.84 This seems especially appealing to better understand emerging psychopathology, as subclinical symptoms have shown to be differentially predictive of the onset of a mental disorder and to improve prediction accuracy of summary scores at the syndrome level.88 In a network, symptoms are represented as nodes and the associations between them as edges. The most important nodes (ie, core symptoms) are at the centre of the network and their effects are supposed to spread quickly by activating other, strongly interrelated symptoms.65 The density of a network can be estimated to investigate if symptoms are highly interrelated and may be best understood as a single entity or if they form distinct clusters of symptoms. This could help investigate the assumption that, with an increased risk of onset of a mental disorder, emerging psychopathology gradually develops into more distinct syndromes,62 or that symptoms become increasingly interconnected, which allows adverse effects to spread more quickly through the network.84 Network models have been applied to different mental disorders in adults, including anxiety,89 mood disorders,90 substance dependence91 and eating disorders.92 Although studies have mainly focused on single disorders using cross-sectional data,87 88 93 an increasing number of studies in adult samples are based on longitudinal data94–96 and across several diagnostic categories.68 97 While promising, few studies have used network analysis in adolescents so far and most of them have focused on a single or a few mental disorders.98–102 For example, one study investigated networks of anxiety and depressive symptoms between the ages of 5 and 14 and found evidence of a highly dense network and overall lack of distinct subgroups of symptoms with slightly increasing interconnectivity over time.99 However, when studying a broader range of psychopathology in adolescents, there is evidence of both distinct symptom cluster (eg, for externalising and internalising symptoms) and also for several between-cluster connections (eg, depression and oppositional behaviour).88 99 101 Given that most studies in adolescents are only cross-sectional in nature and mainly focused on a disorder level instead of a symptom level, this study will advance current knowledge by investigating the network structure of subclinical symptoms across several mental disorders in adolescents cross-sectionally and longitudinally.

Taken together, the EMERGE-study aims to address the overarching research question of detecting temporal patterns and dynamic interactions between subclinical symptoms and their longitudinal associations with transdiagnostic mechanisms, functioning and onset of mental disorders. Specifically, dynamic modelling techniques will be used to investigate the following research questions:

Do subclinical symptoms follow a homotypic or a heterotypic course and is this course differentially predicted by transdiagnostic mechanisms?

Can different risk trajectories of subclinical symptoms be detected and are they predictive of the level of functioning and onset of a mental disorder?

What are the subclinical symptoms that are most central in symptom networks?

Are symptom networks becoming more central, interconnected or distinct over time?

Methods and analysisStudy design

The EMERGE-study is a prospective, 1-year follow-up epidemiological study. Assessments of subclinical symptoms and transdiagnostic mechanisms, conducted through online questionnaires, will take place at baseline (t0) and at 3-month (t1), 6-month (t2), 9-month (t3) and 12-month (t4) follow-up. In order to exclude adolescents with a current or lifetime diagnosis of a mental disorder, clinical diagnostic interviews and level of functioning will be carried out at baseline and at 12-month follow-up (see table 1).

Table 1

Overview of the assessment schedule

Participants

The sample will be recruited from the Swiss general population. To be included, subjects need to (1) be between 11 and 17 years, (2) read and speak German fluently so that they can take part in the interview and fill in the questionnaires, (3) have their main residency in Switzerland and (4) have access to the internet. Subjects will be excluded if they have a current or lifetime diagnosis of a mental disorder according to the ‘Diagnostic Interview for Mental Disorders for Children and Adolescents’ (Kinder-DIPS103 104) or a known developmental disorder (‘Diagnostic and Statistical Manual of Mental Disorders’, Fifth Edition105), as this study focuses on the development of psychopathology. Notably, adolescents with past and/or present diagnosis of specific phobia will not be excluded, as specific phobias are highly prevalent in adolescence but only a minority of them are clinically relevant and require treatment.106 107 Further, specific phobias are currently discussed as a potential marker and predictor of the onset of mental disorder(s), making them a valuable factor in characterising preventive stages.108

Procedures

As shown in figure 1, the contact details of potential participants (ie, those between 11 and 17 years of age and with main residency in a German-speaking canton of Switzerland) will be provided by the Federal Statistical Office, including name, address and, if available, phone number. The sample will be stratified according to age, gender and degree of regionalisation (as defined by the Federal Statistical Office: rural area, intermediate area, urban area). Personalised invitation letters, including study information, will be sent via postal mail to adolescents and their parents/legal guardians. Within the next 2 weeks, potential participants will be contacted via phone and informed about the study (ie, oral study information/verbal briefing). Potential participants (if ≥14 years) or their parent/legal guardian (if <14 years) will then be asked to provide informed oral and written electronic consent. For the written electronic consent, a personalised link will be sent to the email address of the adolescent or to the parent/legal guardian. After receiving written consent, the presence of past and present mental disorders will be assessed via phone using the Kinder-DIPS.103 104 All interviews will be conducted by advanced master’s and PhD students specialising in clinical child and adolescent psychology, who received a 2-day training on conducting and documenting diagnostic interviews. Additionally, all clinical interviews and study inclusions will be supervised in weekly meetings. Eligible individuals will then receive an email with personalised access to the online questionnaires via a web-secured server (Research Electronic Data Capture, REDCap).109 110 At each follow-up assessment, participants will receive a personalised link to the online questionnaires via email, with a total of four reminders sent every 4 days. After the last assessment point, participants will be paid an expense allowance in the form of vouchers for each completed online questionnaire and interview.

Figure 1

Flowchart of the recruitment process.

Assessments

Table 1 provides an overview of the assessments. Self-report instruments to assess transdiagnostic mechanisms and subclinical symptoms were selected based on the criteria that they needed to be short, validated for adolescents and available in German. The subclinical symptoms of mental disorders that typically have their onset in adolescence will be assessed, that is, depression, bipolar disorder, anxiety disorders, obsessive-compulsive disorders, psychosis, substance abuse, eating disorders, conduct disorders and non-suicidal self-injurious behaviours.1 Six transdiagnostic mechanisms that have been theoretically and empirically found to be related to various mental disorders in adolescents will be assessed at each time point: behavioural avoidance,43 111 repetitive negative thinking,44 45 112 intolerance of uncertainty,14 55 113 emotion regulation,114 self-critical perfectionism115 116 and rejection sensitivity.52 117 All self-report assessments will be provided as online questionnaires via REDCap.109 110 Research has shown that adolescents can accurately fill in online questionnaires and take an average of 17 min (SD=6) to answer 116 questions online.118 Therefore, the maximal number of items per assessment point will be restricted to 210 items (ie, 30 min).

In a telephone interview at baseline and at 12-month follow-up, current and lifetime diagnosis of mental disorders will be assessed using the Kinder-DIPS103 104 and social and role functioning using the ‘Global Functioning Social and Role Scale’.119 Furthermore, sociodemographic data, socioeconomic status, adjustment to daily life, self-perceived stress level and adverse life events will be recorded.120–122 The duration of the interviews will be approximately 60 min.

Statistical analyses

To investigate whether subclinical symptoms follow a homotypic or a heterotypic course and whether this course is predicted by transdiagnostic mechanisms (research question 1), dynamic panel models within the structural equation modelling framework will be constructed.123–127 To this aim, autoregressive cross-lagged panel models with summary scores of the subclinical symptom domains (eg, depression and anxiety) at each assessment point will be estimated. Therein, symptoms at one time point (t) will be regressed on these symptoms and predictors (transdiagnostic mechanisms, life events, gender, age, socioeconomic status) at the previous time point (t-1). It is expected that transdiagnostic mechanisms will be interrelated as these processes often overlap.128 129 Residuals of model variables at each time point will be allowed to correlate. Next, time-invariant components will be added as latent variables correlating with predictors and symptom variables at baseline and loading on all symptom variables at later assessment points. These fixed models will then be compared with hybrid models.130 131 Overall model fit will be evaluated using a combination of absolute, incremental and parsimonious fit indices, including χ2 test, comparative fit index, Tucker-Lewis index, root mean square error of approximation and standardised root mean square residual. To identify subclasses of individuals with different trajectories of subclinical symptoms and to use them to predict distal outcomes, such as level of functioning and onset of a mental disorder (research question 2), latent growth mixture modelling and growth mixture survival analysis132 will be carried out. To identify the core subclinical symptoms cross-sectionally (research question 3), network analysis84 133 will be applied, and an undirected, weighted network will be estimated including subclinical symptoms assessed by the self-report questionnaires. The relative importance of a node in a network (ie, its centrality) will be quantified by estimating the following parameters: strength (ie, level of connectivity of a symptom in the network), betweenness (ie, frequency of emergence of a symptom as part of interactions between other symptoms) and closeness (ie, to what extent symptoms tend to cluster together).84 To analyse changes in networks of subclinical symptoms over 12 months and to investigate if they are becoming more central, interconnected or distinct over time (research question 4), longitudinal, dynamic networks and the corresponding centrality parameters (strength, closeness and betweenness) will be estimated including each assessment point. The most central symptoms at each assessment point will then be identified by standardised z-scores. Furthermore, symptoms will be rank-ordered concerning their centrality estimates, and correlations will be calculated to investigate the stability of centrality. To test for distinct networks, a modularity index will be calculated for each network and the Q-index will be estimated ranging from 0 to 1, with values >0.3 indicating significant modularity.134 The distribution of modularity indexes across time will be compared using χ2 test. Changes in the strength of cross-domain/bridging edges will be examined to determine whether these constructs become more interconnected, using non-parametric permutation tests. Comparisons will be performed for specific edges across and in overall connectivity (ie, global strength). Notably, to assess the potential impact of past or present specific phobias (see Participants section) and restrictions associated with the COVID-19 pandemic during recruitment, these variables will be investigated as potential moderators in all statistical analyses.

Power considerations

This study is part of a larger project consisting of two studies, combining basic and intervention research: a prospective, naturalistic, 1-year follow-up study described in this publication (ie, EMERGE-study) and a consecutive transdiagnostic internet intervention study (ie, EMPATIA-study). The studies are linked to each other in that the follow-up assessment of the EMERGE-study corresponds to the baseline assessment of the EMPATIA-study. Therefore, the power calculation of the EMERGE-study is based on the required sample size of the EMPATIA-study. To detect at least a small effect size (F=0.10) in the EMPATIA-study, a power of 0.80, an alpha level of 0.05 and an expected dropout rate of 20%, a sample size of 152 participants is required. Based on an expected participation rate of 60%,135 254 participants who must meet the inclusion criteria of the EMPATIA-study are needed, that is, reporting at least mild subclinical symptoms but not meeting the diagnostic criteria for a mental disorder. It is expected that at least 25% of the participants available at 12-month follow-up of the EMERGE-study will meet these inclusion criteria of the EMPATIA-study.136 137 Thus, 1016 participants are needed at 12-month follow-up and 1196 at baseline assessment of the EMERGE-study (expected dropout rate between baseline and 12-month follow-up: 15%). Because of the expected insufficient German skills in 1% and lifetime/current mental disorders in 13%,138 an expected participation rate in the EMERGE-study of 42%139 140 and an expected contactability rate by phone of 48%,139 3312 persons need to be contacted, of whom 1391 will give consent to study participation. Thus, 6900 addresses need to be drawn from the register of the Federal Statistical Office (see figure 1).

Patient and public involvement

Patients and public were not involved in this study.

Ethics and dissemination

The study will be conducted in accordance with national and international regulations and guidelines, including the Declaration of Helsinki, the principles of Good Clinical Practice, the Human Research Act and the Human Research Ordinance, as well as other locally relevant regulations. Ethical approval was obtained from the Cantonal Ethics Committee (KEK) Bern (ID 2020-02108). All data will be handled with utmost discretion and only accessible to authorised personnel who require the data to fulfil their duties within the scope of the research project. All study personnel need to sign a confidentiality agreement, which ensures the secure handling of all sensitive data. For independent data review and monitoring, quality visits by an external person from the research management of the Faculty of Human Sciences of the University of Bern, who is not otherwise involved in the study, will be conducted every 6 months. In order not to violate the ‘right not to know’ in this non-help-seeking general population sample, the adolescents and their parents/legal guardians will not be informed about individual test results. Yet, if need for counselling or therapy is voiced, information about local mental health services will be provided. Adolescents and their legal guardians are informed in the study information document and reminded before each interview that participation is voluntary and can be terminated at any time without giving reasons.

The results of this study will be disseminated through open-access publications, presentations at conferences, and meetings with researchers, clinicians, school classes and other relevant stakeholders. In addition, all results will be published on our website and distributed to all study participants via a factsheet sent by email.

Discussion

This paper presents the rationale and methodology of a large epidemiological study to better understand emerging psychopathology in the very early stages in the adolescent general population before they may actively seek help. The EMERGE-study represents a shift from rather static, disorder-specific prevention models to dynamic network models of emerging psychopathology. The detection of typical pattern and trajectories of subclinical symptoms and their predictive value in emerging psychopathology and functional difficulties may enhance predictive accuracy under which conditions and through which transdiagnostic mechanisms subclinical symptoms progress into full-threshold mental disorders. This may help identify individuals in need of care earlier and identify the most promising targets for the indicated prevention. Thereby, this study may contribute to overcoming the current gap in care among adolescents by supporting their mental health in the long term.

Potential limitations of this study include the relatively short follow-up duration of 1 year with assessments every 3 months. This time span and frequency of assessments may be too short to fully capture the dynamics of emerging psychopathology and the onset of mental disorders. Moreover, it is important to note that while the study focuses on well-established psychological transdiagnostic mechanisms, other common social and neurobiological transdiagnostic mechanisms contributing to the development of subclinical symptoms (eg, executive functions,54 emotional awareness141 or pubertal timing142) are not fully considered. This highlights the need for future studies that encompass a broader range of transdiagnostic mechanisms and more intensive longitudinal assessments over longer follow-up periods to better model the complexity of emerging psychopathology in adolescence.

Acknowledgments

The authors would like to thank the following persons for their support during the study: Fabian Steiner, Aline Banz, Alina Hunkeler, Irina Lory, Livia Ruckli, Maxine Schmidt, Nadja Hornburg, Nadja Mauerhofer, Nika Saxer, Pelin Koyuncu, Samira Zurbrügg, Sarah Oberli, Stella Ludwig, Stéfanie Mayer, Sarah Wüthrich, Taina Thees, Vera Bächler and Wenja Käch.