A LONGITUDINAL NETWORK ANALYSIS OF ADHD SYMPTOMS

Jonathan Robert Preszler

Attention deficit/hyperactivity disorder (ADHD) has been extensively studied using the latent variable model (LVM). The LVM models the common variance shared by ADHD symptoms. In contrast, network analysis has been used to model all of the direct relationships between the symptoms of various disorders. Network analysis is particularly suitable for disorders whose symptoms might cause each other, or disorders for which there might be multiple underlying causes. Additionally, recent developments in network models allow for modeling longitudinal and multiple-informant data. Modeling longitudinal data with networks allows for the decomposition of between-subject and within-subject variance. Modeling multiple informant data allows for the examination of consistency of relationships across raters. The current study estimates networks of mother and father ratings of ADHD in first- grade children across multiple occasions, thus separating between-subject and within-subject variance for both mothers and fathers. This procedure allowed for the investigation of the relationships between symptoms across time and within specific time points (i.e., contemporaneous relationships), as well as agreement between parents at both of these levels. Finally, community analyses, which identify subcommunities or clusters of specific networks, were applied to each network at each level. The results of the present study revealed that some ADHD symptoms were more central to the whole network than others, indicating some symptoms had more highly elevated partial correlations with the whole network than other symptoms. Additionally, this study revealed high consistency between mothers and fathers at both levels. However, there were some distinct discrepancies between the between-subject and within-subject levels, suggesting that in-the- moment versus temporally stable ADHD symptom presentations may be distinct from each other. Lastly, community analysis were congruent with LVM studies for inattention symptoms, but were distinct from LVM studies for impulsive/hyperactive symptoms. Overall, this study provides support for additional longitudinal investigations of ADHD as well as some initial support for individualized assessment of ADHD by clinicians.

A LONGITUDINAL NETWORK ANALYSIS OF ADHD SYMPTOMS

Files and links (1)

Abstract

Metrics

Details