Adolescent Marijuana Use: Developmental Trajectory and Multi-Level Predictors

Background: The increasing trend of adolescent marijuana use is a threat to public health. Untreated adolescent marijuana use is closely associated with various negative health issues and juvenile delinquency. The findings of longitudinal studies of marijuana use in adolescence are inconsistent, with some studies indicating a linearly increased trend and other suggesting a quadratic trend. While previous studies have widely examined the individual- and family-level predictors of adolescent marijuana use, few studies have focused on the influences of neighborhood factors. Informed by the ecological theory that assumes multi-level system influences on human development, this study aims to understand: (1) the trajectory of marijuana use during adolescence; (2) the influences of individual-, family-, and neighborhood-level predictors on adolescent marijuana use.

Methods: Using four waves of public data from the National Longitudinal Study of Adolescent to Adult Health, latent growth curve modeling (LGCM) was conducted. A total of 4,461 adolescents aged from 11 to 19 at baseline were followed over a ten-year period. Adolescents’ self-reported marijuana use frequency in the previous 30 days was measured at four time points. Log-transformed variables were used in analysis. The neighborhood-level predictor was measured by two variables: trash on streets and neighborhood drug use. Adolescents evaluated these two neighborhood problems using a 3-point scale. Adolescent depression was measured using the Center for Epidemiologic Studies–Depression (CES-D) scale. Child maltreatment was a latent variable including three items from Parent-Child Conflict Tactics. Caregivers’ substance use was measured using two observed variables: evidence of smoking and drinking in household. Control variables included gender, race, family income, adolescents’ depression, parents’ education level, and parents’ marital status.

Results: The LGCM model had a good model fit (CFI = .95; RMSEA = .026 [90% CI: .02 - .03]; SRMR = .02). The quadratic trajectory was supported by this study, suggesting that adolescent marijuana use peaked in the mid-20s and declined after that. Child maltreatment experiences and depression were associated with the initial level of marijuana use (unstandardized mean intercept: child maltreatment = 0.078, p = 0.007; depression = 0.17, p < 0.001). Caregivers’ smoking in the home also predicted the initial level of marijuana use (unstandardized mean intercept = 0.13, p < 0.001). Neighborhood trash problems predicted both the linear and quadratic increase of marijuana use over time (unstandardized mean slope = 0.14, p = 0.019; quadratic slope = 0.14, p = 0.011).

Conclusion: This study identified the predictors of adolescent marijuana use from multi-level systems, including individual, family, and neighborhood environments. Prevention programs that aim to prevent marijuana use during adolescence should address the influences of child maltreatment experiences, depressive symptoms, and caregivers’ substance use. In particular, the influences of neighborhood problems, which have received little attention in previous studies, were highlighted as a significant risk factor for youth behavioral health problems in our study. Practitioners should closely screen adolescents’ living environment when recommending treatment for marijuana use. There is a critical unmet need for neighborhood-level preventive programs in order to reduce adolescent marijuana use.