The transdiagnostic approach and the network model approach to psychopathology have emerged as alternatives to traditional diagnostic systems. This is the first study to date that has examined the network structure of the transdiagnostic dimensions of emotional disorders in a large sample of adolescents with subclinical emotional symptoms and the relationship with risk factors and protective factors. New approaches such as transdiagnostic and network models may provide new perspectives on the origin, maintenance, clinical management, and recovery from experiences of disabling mental distress (Conway et al., 2019; Dalgleish et al., 2020; Fonseca-Pedrero, 2018; Widiger, 2021).
The first objective of this study was to analyze the network structure of the transdiagnostic dimensions of emotional disorders put forward by Brown and Barlow (2009) and their applicability in a representative sample of adolescents with subclinical emotional problems. The results demonstrate a strong interaction between the transdiagnostic dimensions, with the exception of PT. This is consistent as positive temperament is understood as the tendency to experience a positive affect in response to social tasks and goal-directed tasks (e.g. extraversion, behavioral activation, positive affectivity) (Rosellini & Brown, 2019). Previous studies have shown that PT is significantly and inversely correlated with the other transdiagnostic dimensions (Brown & Barlow, 2009; Rosellini & Brown, 2019; Osma et al., 2022; Pérez-Esteban et al., 2024). In addition, the scientific literature has demonstrated that low positive affect is related to developing greater depressive disorder and anxiety disorders (Barlow & Kennedy, 2016). Our results add to the previous evidence as PT was the dimension with the least expected influence within the network.
In terms of strength and expected influence, the nodes in the network with the greatest weight were AVD, IC, TRM, NT and DM. The strong association and the weight of TRM in the network, along with the IC dimension, is particularly interesting. It is well known in the literature that adverse experiences in childhood and adolescence mean a greater risk of experiencing generalized anxiety and/or symptoms of depression during or after subsequent stressful life events (Ródenas-Perea et al., 2022; Li et al., 2022; Zhou et al., 2022; Reis et al., 2024). One of the explanations for this is that traumatic childhood experiences result in the development of ineffective coping strategies (Barlow, 2004; Snyder et al., 2019; Cludius et al., 2020; Sætren et al., 2023) such as excessive focusing on events themselves, the emotions they provoke, and analyzing their causes and effects (Aldao et al., 2010; Nolen-Hoeksema et al., 2008; Snyder et al., 2019). This is known as rumination and arises as an attempt to understand and resolve the situation, although rumination has been shown to be negatively related to problem solving and is maladaptive for coping with stressful life events (Battista et al., 2023; Michael et al., 2007; Papageorgiou & Wells, 2003; Snyder et al., 2019). Another response to traumatic experiences is the appearance of intrusive thoughts. Research has found a strong relationship between the presence of intrusive thoughts and the experience of traumatic events in early ages (Barzilay et al., 2019; Ródenas‐Perea et al., 2022). These people also present higher levels of avoidance behaviors and symptoms of anxiety or depression than those who have not had such experiences (Jhang, 2020; Pérez et al., 2017). This is consistent with our results. Lastly, it is worth noting that in addition to the strong relationship between these variables, there were also mediating relationships. For example, maladaptive strategies such as rumination and the appearance of phenomena such as intrusive thoughts are mediating variables between traumatic experiences in childhood and the subsequent appearance of symptoms of stress, anxiety, and depression in later life (Kim et al., 2021; Ródenas‐Perea et al., 2022; Sætren et al., 2023).
The second objective of our study was to examine the network structure of the transdiagnostic dimensions in relation to other psychometric indicators of psychopathology and socio-emotional adjustment. In the first place, the results corroborate the importance of the AVD, IC, and TRM dimensions. In addition, it seems logical to have found a negative relationship between the transdiagnostic dimensions (except PT) and the other protective variables (perceived social support, sense of belonging at school, prosocial behavior, quality of life, and self-esteem). These factors were closely interrelated with each other, and seem to act as an important source of protection against different types of psychopathological domains in the adolescent population (Fonseca-Pedrero et al., 2021; Fonseca-Pedrero, Ortuño-Sierra, et al., 2019). Our results also indicate the importance of perceived social support, a sense of belonging at school, and adolescents’ reported quality of life (Kumcağız & Şahin, 2017; Nowicki, 2008; Singstad et al., 2021). Having a positive, cohesive, social support network seems to be a fundamental protective factor for mental health in adolescence. In fact, research has shown that there is a significant positive relationship between social support and psychological wellbeing (Agbaria & Bdier, 2020; Singstad et al., 2021; Tian et al., 2013) and that such support allows adolescents to better cope with stressful life events (Camara et al., 2017; Reife et al., 2020). In addition, satisfaction with teachers and feeling a sense of belonging at school may protect young people from the effects of stressful life events, promoting resilience and reducing the likelihood of developing depression or behavioral problems (Höltge et al., 2021; Wang et al., 2013). In contrast, dissatisfaction with teachers is a potential risk factor for developing anxiety, emotional, and behavioral problems (Arslan, 2021; Mameli et al., 2018). Lastly, in our study, prosocial behavior was positively related to a feeling of belonging at school and perceived social support, and negatively related to externalizing symptomatology and poor behavioral adjustment (Inglés et al., 2015).
Our results have notable clinical implications. On the one hand, studying subclinical states may help us understand the dynamics between different variables that may develop into different mental health problems. Network analysis has confirmed that problems that manifest sub-clinically in specific dimensions in early stages of development are consistent over subsequent years in more serious psychopathological processes (Groen et al., 2019). These aspects are related to the clinical staging model that seeks to facilitate treatment selection based on the progression and severity of the problem (Hartmann et al., 2020). On the other hand, implementing network analysis is useful for uncovering specific treatment pathways, which may encourage the development of more effective interventions for addressing symptoms of emotional problems based on where they are in their progression and the severity. Previous studies have shown that intervention in specific symptoms indirectly triggers a wave of change in other symptoms, altering the connections between elements in the network (Jurado-González et al., 2024). Incorporating the network model together with the transdiagnostic approach applied to adolescents’ emotional problems opens the door to better understanding of the psychopathology and possible focuses of individualized treatment (Borsboom et al., 2011; Manfro et al., 2023). In addition, the results of our study may feed into novel theoretical dimensional and transdiagnostic models applied to psychopathology, such as the internalizing spectrum within the HiTOP model (Snyder et al., 2023; Watson et al., 2022). Similarly, the relationship with school-related risk and protective factors allows us to consider the importance of relationships in adolescents’ developmental surroundings (social support, sense of belonging), encouraging development of structural initiatives or universal prevention programs in schools that promote protective factors and reduce risk factors (Fonseca-Pedrero et al., 2023).
The present study is not without limitations. Firstly, the use of solely self-reported data limits the conclusions that can be drawn. Secondly, the study was transversal, which means it does not allow us to examine the dynamic individual interactions between the variables nor to do so longitudinally. A time-series analysis would avoid this issue, but we must be cautious, as the field of network analysis is in its infancy, which means there still fundamental questions open about the best way to estimate a network over time. Thirdly, only some risk and protective factors were selected, although they were consistent and based on the literature. It would be useful to consider others that are involved in various problems that are particularly prevalent in adolescence (González-Roz et al., 2023). Another limitation was the structure of the networks being restricted by the tool used in the study. Network analysis is still in the early stages and is not without its critics. This means we should be careful when it comes to extrapolating from these results to other populations in which a wider range of scores may lead to networks behaving differently. Nonetheless, it is a promising methodology for obtaining important information in a variety of research fields (Bringmann et al., 2021; Bringmann et al., 2019; Guloksuz et al., 2017).
Future lines of research should include network models with variables from different levels of analysis, especially those providing information about other transdiagnostic variables (such as rumination), clinical variables (such as frequency and type of traumatic experiences, suicidal behavior, etc.), and contextual variables (such as bullying). In addition, it would be extremely interesting to collect longitudinal information via new methodologies such as ambulatory assessment in order to move closer towards dynamic, contextual, tailored models. Finally, it would also be interesting to be able to incorporate new analytical methodologies such as Directed Acyclic Graphs (DAG) (Briganti et al., 2021; Moffa et al., 2017).