Background: Many children who suffer from one mental health issue also suffer from at least one co-occurring disorder and a range of developmental psychopathology theories, including developmental cascade and network models, have been proposed to explain this widespread comorbidity. Autoregressive latent trajectory models with structured residuals (ALT-SR) and multilevel graphical vector autoregression (GVAR) are recently proposed complementary approaches that can help operationalise and test these theories and provide new insights into the reciprocal relationships between multiple mental health domains to advance the understanding of comorbidity development.
Methods: This study uses ALT-SR and multilevel GVAR models to analyse the temporal, contemporaneous, and between-person relationships between key dimensions of child mental health: emotional problems, peer problems, conduct problems, hyperactivity/inattention and prosociality as measured by the parent-reported Strengths and Difficulties Questionnaire (SDQ) in 17,478 children from the UK Millennium Cohort Study at ages 3, 5, 7, 11, 14 and 17 years.
Results: Children’s strengths and difficulties in different domains of psychosocial functioning were dynamically associated with each other over- and within-time. The ALT-SR highlighted that hyperactivity/inattention plays a central role in affecting other domains over developmental time, while the GVAR model highlighted comparably strong bidirectional relationships between conduct problems and prosociality as well as between emotional problems and peer problems.
Conclusion: This study confirms that mental health difficulties influence one another dynamically over time. The complementary techniques of ALT-SR and GVAR models offer different insights into comorbidity and hold promise for supporting the building of more comprehensive developmental psychopathological theories that acknowledge the inter-connectedness of different domains of mental health.
Keywords: developmental psychopathology; socio-emotional strengths and difficulties; ALT-SR; Graphical Vector Autoregression; Millennium Cohort Study