Preventing Adolescent Alcohol Use and Delinquency: A Dynamical Systems Analysis of Genetic Moderation of Intervention Effects

Introduction: Dynamic systems can be applied to model adolescent substance use and delinquency (Boker & Graham, 1998), the levels of which fluctuate while the magnitudes amplify or decrease across adolescence. This study intends to investigate intra-individual variation in adolescent alcohol use and delinquency and their coupling. We also examine intervention effects and their interaction with the candidate gene DRD4.

Methods: Data were drawn from 544 adolescent participants (320 in intervention group) from the PROSPER Project (Spoth et al., 2004). Latent differential equation models (Boker, 2007) were fit to four years of self-reported past year alcohol use and delinquent acts. Respondents (45.6% males, 89.7% White/Caucasian, M age = 11.92, SD = .46 at first wave in 6^th grade) were part of a large-scale community-based randomized substance use prevention trial. 35% of intervention and 36% of control participants possessed the DRD4 7+ genotype. Multi-group SEM were performed to examine group differences in the system parameters ζ (dampening) and η (oscillating) that capture linear relationships between the second (acceleration), first derivative (slope), and the value (displacement) of a variable, in addition to group means.

Results:  Alcohol use and delinquency demonstrated oscillation (η = -.43 and -.42 respectively) across time, and alcohol use showed an amplifying magnitude (ζ = 1.08). The processes were coupled in that acceleration of delinquency can be predicted by the displacement of alcohol use (b = .13). The intervention group showed significantly lower means of alcohol use and delinquency than control, particularly in the latter two waves (e.g., .12 versus .16 for alcohol use and .04 versus .06 for delinquency in wave 4). The intervention group demonstrated a dampening magnitude for delinquency (ζ = -2.18), whereas the control group showed amplification (ζ = 1.51) for delinquency. Interactions by DRD4 genotype were not significant, but analyses by DRD4 7+ versus 7- did reveal that the significant main effect for intervention on delinquency only existed among participants with 7+. 

Discussion: Dynamical systems provide a unique approach to model adolescent substance use and delinquency as opposed to conventional latent growth curve (Boker, 2001). Both alcohol use and delinquency demonstrate oscillation, suggesting adolescents experimenting and switching between different states (e.g., non-drinker and drinker). Intervention can not only prevent the onset of alcohol use and delinquency, but also affect system parameters. The null finding regarding genetic effects may be due to its small effect size, sample size, and variations in both behaviors.