Investigating the Development of Co-occurring Internalising and Externalising Problems using Structural Equation Modelling

• In Event: Continuing the Quest Towards Understanding Co-occurring Internalizing and Externalizing Problems: Methodologies, Risks, and Clinical Implications

Abstract

Children suffering from one mental health problem are much more likely than others to develop additional mental health problems concurrently or later in life (Rhee et al., 2015; Wagner et al., 2017). Trajectories of mental health problems have further been found to be highly correlated such that children following a particular trajectory of internalizing problems likely follow a similar trajectory for externalizing difficulties (e.g., Nivard et al., 2017). A variety of theories have been developed aiming to explain the high co-occurrence of mental health problems. While some theories focus primarily on common causes, that is factors underlying a propensity to developing multiple different mental health problem (Caspi et al., 2014), others focus on causal developmental cascades between different disorders whereby for instance externalising problems are hypothesised to lead to problems with peers and consequently to the development of internalising problems (Capaldi, 1992; Masten & Cicchetti, 2010). Importantly, these different theories focus on different underlying mechanisms. Common cause theories focus primarily on between-person effects, that is factors differing between people making it more likely for some individuals to develop multiple mental health problems, for instance because of a genetic vulnerability. Developmental cascade models focus on within-person effects, that is processes unfolding within an individual such that acting out in school leads to problems with peers and subsequently to emotional problems within the same individual. When investigating the development of co-occurring mental health problems, it is of vital importance to appropriately distinguish between within- and between-person processes. Over the past decade, statistical advancements primarily within a structural equation modelling framework have allowed for much more nuanced investigations of these processes. In particular, group-based trajectory models or parallel process models can be used to define developmental subtypes characterised by multiple co-occurring symptoms or to examine the parallel trajectories of multiple disorders over time. These models can further be extended to investigate how between-person factors such as genetic predispositions, family characteristics or obstetric factors may relate to the development of such joint trajectories (e.g., Speyer, Neaves, et al., 2021). If interested in testing theories focusing on within-person processes, random-intercept cross-lagged panel models can be used to disaggregate repeatedly measured variables into within- and between-person components and to investigate how within-person relations between different mental health problems develop over time (Hamaker et al., 2015). Such models can further be extended to test for mediation (e.g., Speyer, Obsuth, et al., 2021) as well as moderation effects (Speyer et al., 2022), thus offering more detailed insights into the development of co-occurring mental health problems than previously possible. Here, I will give a brief introduction to some of the statistical methods suitable to investigating the development of co-occurring mental health problems and how these methods can be used to test different developmental theories focusing on the development of co-occurring internalising and externalising problems.

Author

• Lydia Gabriela Speyer, University of Edinburgh & University of Cambridge
  Presenting Author