2 - (IOP 13) Predictors of Anxiety Severity in U.S. School-aged Children Using the 2021 National Survey of Child Health

Anxiety disorder rates in U.S. children doubled during the COVID-19 pandemic (Racine et al., 2021) and remained high afterwards (Wang et al., 2022). Public health data include racially and ethnically diverse study participants; however most public health studies use anxiety disorder diagnosis as an outcome variable, rather than anxiety severity. This project aimed to identify predictive factors of anxiety severity in school-aged children to inform targeted prevention and treatment interventions. Predictor variable selection was guided by the literature and will be discussed during the presentation (Melton et al., 2016; McLean et al., 2011; Fitzallen et al., 2021; Farrell et al., 2011; Swan & Kendall, 2016; De Lijster et al., 2017). Research questions: 1) How do age, gender, need for special healthcare, current depression, friendship difficulties, and premature birth influence anxiety severity in children and adolescents (Model I)? 2) How is predictive anxiety severity influenced by the combined effect of friendship difficulties and age (Model II)?

This study utilized the 2021 National Survey for Child Health publicly available population-level data set (CAHMI, 2023). The study sample (n=29,882) included children aged 6-17 years. Descriptive statistics, preliminary bivariate chi-square analyses, ordered logistic regression analyses, and average marginal effect analyses were conducted. Model I resulted in a model likelihood ratio chi-square of 9494.63 (df=6), p<.001, pseudo R2= 0.3099. All probability differences for each level of each predictor were statistically significant for both Model I and Model II at the .001 level except that for premature birth (p <.05). For Model II, an interaction term of difficulty making and keeping friends and child age was added to Model I. The Model II resulted in a model likelihood ratio chi-square of 9595.64 (df=12), p<.001, pseudo R2= 0.3132. The interaction of child age and friendship difficulties is statistically significant (p <.05) in Model II, showing a difference of 3.008 in the model log-likelihood from Model II with the interaction term and the nested Model I, with an increase in pseudo R2 by .0002. There was a statistically significant chi-square change between models 6.016, p = 0.014 (p <.05).

Child age, being female, having special healthcare needs, current depression symptoms, friendship difficulties, and premature birth were predictive influencers of child anxiety severity in this sample. Friendship difficulties and anxiety symptom severity differed in children by age. Study limitations will be discussed during the presentation. Results may inform the development, adaptation, and implementation of inclusive and affirming interventions targeting moderate to severe anxiety disorder severity in conjunction with the above predictors.