An investigation of multimodal predictors of adolescent alcohol initiation

Background

Early alcohol initiation is associated with negative, alcohol-related outcomes. While previous work identifies numerous risk factors for early use, the relative contributions of known predictors remains understudied. The current project addresses this gap by 1) prospectively predicting early alcohol initiation using measures of inhibition control, reward sensitivity, and contextual risk factors and 2) interrogating the relative importance of each domain.

Method

This study leverages multimodal data from substance-naïve youth enrolled in the Adolescent Brain Cognitive Development (ABCD) Study® (n=11,694). Early initiation was defined as consuming a full standard drink containing alcohol prior to age 16. Propensity scores were used to match alcohol initiators (n=348) with demographically similar non-initiators at a 1:2 ratio (n=696). Independent logistic regressions were conducted for each domain followed by additive, hierarchical models.

Results

The model of contextual factors (pseudo-R²=0.086, AUC=0.67) outperformed inhibition control (pseudo-R²=0.021, AUC=0.58) and reward sensitivity measures (pseudo-R²=0.020, AUC=0.59). The hierarchical model containing all measures (pseudo-R²=0.106, AUC=0.69) did not significantly improve the model of contextual factors alone (p>0.05). Examples of significant predictors (p<0.05) include externalizing behaviors, number of substances known, and non-religious alcohol sipping.

Conclusion

Contextual risk factors were the strongest predictors of early alcohol use; however, more work is needed to understand the causal nature of this relationship. Measures of inhibition control and reward sensitivity were not adequate in distinguishing initiators from non-initiators. These findings add to a body of evidence that contextual factors play a major role in alcohol initiation while highlighting specific predictor variables that could inform youth alcohol prevention.