Latent Classes of Polysubstance Use Among Middle and High School Adolescents in a Diverse Rural Setting

Background/Significance:Rural youth are an understudied group at risk for a variety of negative outcomes including substance abuse. Given the unique risk factors (e.g., geographic isolation, high unemployment, and community disadvantage), little research has explored patterns of substance use among these youth. In particular, research has neglected lifetime polysubstance use among rural middle and high school students. It is important to consider that youth not only demonstrate different numbers of lifetime substances used, but also different combinations of substances at disparate frequencies. In this study, we fill these gaps in the literature and answer the following question: Do youth in middle and high school sort into discernible patterns of lifetime polysubstance use?

Methods: The total analytic sample was comprised of 6,085 youth and data were collected using the School Success Profile Plus. Five lifetime substance use items (i.e., alcohol, cigarettes, marijuana, inhalants, and prescription drugs) were entered into a latent class analysis (LCA) to identify clusters of youth according to their response patterns. Both imputed and non-imputed data were used when possible. The item responses were treated as categorical variables (4-level) to allow for gradations in use and separate models were estimated for middle- and high-school students. Akaike and Bayesian information criteria, the Vuong-Lo-Mendell-Rubin test, and Bootstrap likelihood ratio test were used to identify the appropriate number of classes at each level of schooling. The resulting classes were examined for their concordance with known trends.

Results:High school students sorted into four non-linear classes showing increasing severity: Abstainers (67%) primarily reported using none of the substances; Initiators (22%) reported moderate levels of alcohol, cigarettes, and marijuana; Users (8%) and Severe Users (4%) exhibited a non-linear pattern with the Users having higher marijuana and cigarette use than in the Severe Users group, who exhibited the highest rates on alcohol, inhalants, and prescription drugs. Middle school students had a simpler linear pattern, with Abstainers (83%) having nearly no use on all substances; Low-Intensity Users (14%) having some use across all substances; and Intensive Users (3%) who used all five substances at the highest rates.  These patterns were the same across imputed and non-imputed data.

Conclusions/Implications: Latent classes provide a way to categorize different levels of child and adolescent substance use and can be used to guide prevention and intervention targeting.  Youth with the same number of substances used demonstrated different combinations of substances that may be associated with different levels of risk. Findings indicate the utility of using LCA to examine groupings of youth on polysubstance use to see how certain patterns of substance use emerge. Prevention services, while universal for all youth, should pay special attention to Initiators and Low-Intensity Users. Meanwhile, Intensive Users in middle school, and Users and Severe Users in high school should be targeted with tertiary substance use treatment before problems become overwhelming. These findings inform further research on predictive modeling of risk factors associated with these classes and change over time in substance use.