Emotion regulation and functioning have well established links to substance use in adolescents. Yet limited research has investigated emotion regulation in very early substance initiators either on self-report or on behavioral measures (i.e., Emotional Stroop). Similarly, there are few prospective investigations of emotional functioning as a predictor of initiation. Given concerns of emotion difficulties preceding and predicting substance use onset, we aim to investigate emotional functioning difficulties in very early (ages 9–13) substance use initiators relative to sociodemographically matched controls, both after initiation and as a predictor of initiation. We hypothesize that initiators would demonstrate greater emotion dysregulation and decreased emotional functioning relative to controls.

ABCD Study Annual Release 4.0 was used. Participants included those who had data available at Y3 follow-up visit and youth-reported use of any full dose of a substance (n=148). Sociodemographic controls were then matched (n=148). General linear mixed effects models were run to assess emotional functioning at Y3 (Emotional Stroop response time and accuracy performance, youth-reported Emotion Regulation Questionnaire, and parent-reported Difficulties in Emotion Regulation Scale and Child Behavior Checklist externalizing and internalizing symptoms) by substance use group status controlling for random effects of family. Further, hierarchical linear models assessed CBCL emotional functioning from Y0 to Y3 predicting SU initiation at Y3, controlling for within-subject change.

At Y3, early substance use initiation predicted higher parent-reported externalizing symptoms significantly (estimate=5.88, p<.001). Substance use initiation also marginally predicted high parent-reported internalizing symptoms (estimate=2.29, p=.08) and DERS (estimate=0.02, p=.07). ERQ and Stroop performance were not significantly associated with group status (p's>.10). For externalizing symptoms predicting SU initiation, regardless of year (baseline through Y3) was significantly predictive of initiation (p's<.001). HLM demonstrated that externalizing symptoms at all time points resulted in the best predictive model (AIC=392.85, BIC=422.80, relative to models including all data through Y2, AIC=433.63, BIC=458.59).

Here we found externalizing symptoms and, to a lesser extent, internalizing symptoms and emotion dysregulation are associated with early substance use initiation. However, results are limited to parent report, despite the consideration of youth-report and a behavioral measure of emotion regulation, the Emotional Stroop task. Further, while marginal effects were found, downstream externalizing symptoms were a better predictor of later substance use initiation. While other metrics of emotion regulation have been linked to substance use in adolescence, emotion regulation abilities may change as a result of substance use, rather than a predictor of use, and thus needs monitoring over time.