INTRODUCTION SUMMARY There is high co-occurrence between attention-deficit/hyperactivity disorder (ADHD) and disruptive behavior problems in children, including oppositional defiant disorder (ODD) and conduct disorder (CD).1 Recommended non-pharmacological interventions for managing ADHD, ODD, and CD characteristics in children largely overlap, with behavioral parent training programs showing empirical evidence of effectiveness.2,3 By improving parent-child interaction patterns, these programs may also interrupt cycles linking characteristics of ADHD with ODD and CD in children. For example, after intervention, children may be less likely to argue and fight with their parents when leaving their seat. However, much parenting program research relies on composite scores, which capture overall intervention effects but offer limited insight into how changes in specific characteristics contribute to overall improvement. In the present study, we will pool data from around 44 randomized controlled trials (RCTs) of parenting programs for children with ADHD and/or disruptive behavior problems, yielding a large sample suitable for applying network analysis to better understand how parenting programs change the interplay between child ADHD, ODD, and CD characteristics. In network theory, mental health problems arise from interrelations between characteristics, which may reinforce one another in reciprocal cycles.4 For instance, network studies have shown that running excessively, blurting out answers, and interrupting are strongly interrelated with arguing, defying, and blaming others in children.5,6 However, most network studies have focused on general population samples, limiting the extent to which their findings can inform hypotheses about potential feedback loops among characteristics of ADHD, ODD, and CD in clinical populations, where problems might be more entrenched (see Figure 1 for an example). To address this gap, we will first examine the interplay of ADHD, ODD, and CD characteristics in children with ADHD and/or disruptive behavior problems who receive a parenting program. Second, we will explore if and how parenting programs change these interrelations, and whether changes differ between intervention and control group children. Figure 1. Example of Potential Baseline Network. Note. This figure demonstrates a selection of child attention-deficit/hyperactivity disorder (ADHD), obsessive-compulsive disorder (OCD), and conduct disorder (CD) characteristics for illustrative purposes. Nodes represent individual characteristics. Thicker edges between nodes represent stronger interrelations (ie, partial correlations). The example depicts a densely connected network with several strong associations. Larger nodes reflect relatively more central characteristics that are more strongly connected to other characteristics in the network. Parenting programs may change both the overall pattern of interrelations among characteristics and the strength to which one characteristic is interconnected with others. Network connectivity reflects how strongly all characteristics co-occur and may reinforce each other, with higher connectivity suggesting greater persistence of problems over time.7 Changes in connectivity may thus provide insight into whether parenting programs reduce feedback loops among characteristics. Centrality quantifies how strongly a characteristic is connected to others in the network (see Figure 1), with highly central characteristics assumed to be key intervention targets because improving them may have cascading effects.8 Changes in centrality following intervention may therefore indicate shifts in interrelations and reveal which characteristics drive broader improvements. METHOD SUMMARY Our source of data is the Psychosocial ADHD and disruptive behavior INTerventions Individual Participant Data Meta-Analysis database (PAINT-IPDMA; PROSPERO CRD42022355664). We will include studies of behavioral parenting programs, defined as interventions that promote positive child behaviors and reduce disruptive ones by targeting parenting practices through behavioral and/or cognitive techniques.9 Eligible studies must include children with either (1) a diagnosis of ADHD and/or ODD/CD, confirmed through clinical interviews, or (2) scores in the clinical or subclinical (at-risk) range on ADHD and/or ODD/CD scales as defined by each trial using established questionnaire cutoffs or normative data. Studies should also provide baseline and post-intervention item-level data on ADHD, ODD, and/or CD characteristics. Trials with a mean child age above 12 years will be excluded. Based on current data harmonization, we expect to include ∼44 studies (≈ 5,789 families). Measures Child behaviors will be harmonized using an item-matching approach, selecting items from established parent-rated questionnaires that map onto DSM-5 criteria for ADHD, ODD, and CD. To ensure comparability across studies, all items will be rescaled to a 7-point scale, consistent with the most common response format. Details of the harmonization process are available at: https://osf.io/s3m48/?view_only=d2221defce214f52b0aee4427f4577c6. Analytic Strategy We will estimate regularized Gaussian graphical models (GGMs)10 to examine how children’s ADHD, ODD, and CD characteristics cluster and interrelate at baseline. Network connectivity (sum of all absolute edge weights) and node strength centrality (sum of edge weights per node) will characterize the network’s global structure and hierarchy of characteristics. To examine how parenting programs change the network, we will make two comparisons: 1) networks for the intervention versus the control group, and 2) networks at baseline versus at post-intervention for the intervention and control families. Connectivity will be compared with the network comparison test (NCT), and node strength centrality with Spearman’s rank-order correlations. All analyses will be exploratory. SIGNIFICANCE SUMMARY First, we will examine the interplay among child ADHD, ODD, and CD characteristics, advancing our theoretical understanding of how they co-occur and which characteristics share more connections to others. Findings may inform hypotheses regarding potential feedback loops among characteristics and priority targets for intervention. Second, our study will help clarify whether parenting programs alter patterns of interrelations and if they disrupt the connections of certain characteristics to others, thereby preventing escalation into more entrenched problems. This knowledge may help explain overall improvements and differential effects in child outcomes. Also, we may identify characteristics for which current intervention techniques are insufficient, highlighting priorities for future work on strengthening or adapting intervention strategies to reduce characteristics and their interconnections that may be less amenable to change.

Keywords