Patterns of adverse childhood experiences and hourly social media use among U.S. high school students: a cross-sectional analysis of the 2023 YRBS

Adverse childhood experiences (ACEs) may be associated with adolescent social media engagement, but cumulative counts can obscure differences between co-occurring adversity patterns. We examined these associations in the 2023 national Youth Risk Behavior Survey (YRBS). We analyzed 9,603 students with complete data on eight ACE indicators, social media frequency and analysis covariates. The primary outcome was social media use at least hourly; the original threshold of at least several times a day was retained as a sensitivity outcome. Exploratory latent class analysis identified ACE profiles. Survey logistic regression incorporated weights, strata and primary sampling units. ACE-only XGBoost models and out-of-fold Shapley additive explanations (SHAP) were examined overall and by sex. At least hourly use was reported by 3,844 students (40.0% unweighted; 38.2% survey-weighted). Five latent profiles were identified. Compared with the low-ACE profile, the household mental illness/emotional abuse profile had an aOR of 1.35 (95% CI 1.11–1.63), while the high multi-type profile had an aOR of 1.18 (95% CI 0.99–1.39) and was not statistically significant. Each additional ACE had an aOR of 1.04 (95% CI 1.01–1.07). Overall out-of-fold XGBoost AUC was 0.529, indicating weak discrimination. Cumulative ACE burden showed a modest association with hourly social media use, while profile-specific findings depended on the outcome threshold and covariate adjustment. Exploratory ACE-only prediction had weak discrimination. These cross-sectional findings do not establish problematic use, addiction or causal effects.