Parent-other and parent-child ratings agreement on ADHD and co-occuring emotional and behavior symptoms among children with ADHD in the MADDY study

Abstract

Background: Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental condition commonly diagnosed in childhood. Multi-informant (parents, teachers, and children) reporting of symptoms is a standard practice in pediatric behavior research. However, informant discrepancies are frequently observed and may influence screening accuracy. The objective of this study was to examine the level of parent-other adult (teacher, grandparent, aunt/uncle, or neighbor who knew the child’s behavior well) and parent-child agreement for ADHD core symptoms and co-occurring emotional and behavior symptoms in children with ADHD and whether the agreement differs by child sex. Methods: Secondary analysis of data from the MADDY study was conducted to assess the level of parent-other adult and parent-child ratings agreement. Parent-other adult agreement was assessed across CASI-5 domains, including inattention, hyperactivity, social and academic impairment, oppositional defiant disorder (ODD), disruptive mood dysregulation disorder (DMDD), and anger related behavior. Parent-child agreement was assessed across PROMIS domains, including anger, anxiety, depression, peer relations, and sleep. Agreement at all domains were analyzed using data of two timepoints: baseline and week 8. Descriptive statistics summarized demographic and clinical characteristics of study participants. Informant agreement was assessed using concordance correlation coefficients (CCCs), with confidence intervals estimated via bootstrapping. Bland-Altman plots were used to visually assess systematic bias, identify patterns of disagreement, and evaluate the limits of agreement across raters. Stratified analyses by child sex were conducted to explore sex-based differences in informant agreement. Results: A total of 126 children were enrolled in the RCT (mean age = 9.8 years; 73% boys), with analytic sample sizes for this study varying by domain due to data availability. Most families (57.6%) reported annual household incomes > $80,000, and 61.1% parents held a university degree or higher. Parent-other adult agreement on CASI 5 domains was poor at baseline (CCC = 0.12 - 0.43) and week 8 (CCC = 0.34 - 0.53), with agreement for inattention significantly improved at week 8 (diff CCC = -0.27 (95% CI: -0.44 to -0.09)). Parent-child agreement across PROMIS domains was also poor at baseline (CCC = 0.07 - 0.39) and week 8 (CCC = 0.09 - 0.26). Parents tended to report higher symptom severity than other adults across most CASI-5 domains at both time points, and greater anger severity than their children at both time points as well as lower sleep disturbance at week 8. Sex-specific differences in agreement were observed, with higher parent-other adult agreement for inattention among girls at week 8 (diff CCC = -0.33 (95% CI: -0.58 to -0.04)) and higher parent-child agreement for peer relations among boys at baseline (diff CCC = 0.46 (95% CI: 0.08 to 0.81)). Conclusions: Overall low level of parent-other adult and parent-child ratings agreement underscores the differences in informant perspectives. It also reflects variations in contextual demands, informants’ knowledge and expectations, the nature of their relationships with the child, and subjective interpretations of behavioral and emotional symptoms. Children’s sex may play a role in symptom recognition and reporting patterns, and further studies stratified by sex is warranted. Overall, these findings highlight the importance of incorporating sex-sensitive framework and multi-informant assessment approaches to improve the precision of screening in research context.