Presentation (CTI)

OBJECTIVES: This study aimed to estimate recent trends in the frequency of adolescents feeling depressed or anxious using nationally representative data 2021-2023.
METHODS: The study utilized the 2021-2023 National Health Interview Survey’s Sample Child Interview, focusing on adolescents aged 12 to 17. The outcome measures were the frequency of adolescents seeming sad or depressed or anxious or worried, categorized as 1 if they experienced it monthly, weekly, or daily, and 0 otherwise. Independent variables included age, sex, race/ethnicity, self-reported general health, community support, family income, parental education, insurance status, urban/rural residence, U.S.-born status, and year indicators. Logistic regression models were applied to estimate the trends. Additionally, logistic regression, k-nearest neighbor (KNN) and decision tree machine learning models were trained to determine which approach best classified adolescents experiencing frequent depression or anxiety.
RESULTS: The final sample included 8,516 adolescents. Adolescents in 2022 reported significantly higher frequencies of feeling depressed and anxious compared to 2021. In 2023, the frequency of feeling depressed was lower, while the frequency of feeling anxious was higher compared to 2021, though these differences were not statistically significant. Among the machine learning models, logistic regression demonstrated the highest accuracy in classifying whether an adolescent frequently felt depressed or anxious.
CONCLUSIONS: Adolescents in 2022 had significantly higher frequencies of feeling depressed and anxious compared to 2021. Logistic regression proved to be the most accurate model for classifying adolescents who frequently experienced depression or anxiety. Further research is needed to explore the factors contributing to these trends and the utility of machine learning models in mental health surveillance.