UC San Diego

The current project is a secondary analysis of a treatment sample of adolescents (age 13 to 18; 54% female; N=221) possessing both a DSM Axis I substance use disorder and a comorbid Axis I internalizing (e.g., major depression, generalized anxiety disorder) and/or externalizing (e.g., conduct disorder, oppositional defiant disorder) disorder. Social cognitive theory applied to addictive behaviors assumes that characteristics of situations common to an individual's substance use prior to treatment for a substance use disorder represent situational characteristics that will pose high risk for relapse following treatment. The long-term objective of the current project is to inform the development of prevention and treatment programs for addictive behaviors. Specifically, patterns of substance use situations for "comorbid adolescents" were identified so that psychosocial interventions can be specifically tailored to focus on the needs of the heterogeneous population of adolescent substance users. The current project specifically investigated the structure of comorbid adolescent substance use situations. First, the fit of two adult classification systems of pretreatment substance use situations were compared to an alternative, developmentally informed classification system. Next, pretreatment substance use situational profiles for comorbid adolescents were identified and differences in substance use situations across profile groups were described. Secondarily, differences were explored across substance use situation profile groups in relation to several covariates. Specifically, aims of secondary analyses were to validate the situational profiles obtained in earlier analyses, and to determine associations of specific profile groups with comorbid psychopathology, gender, ethnicity, age, and substance use history. Results from a series of confirmatory factor analyses (CFAs) indicated that the 5-factor model corresponding to the developmentally informed classification system best represented the structure of substance use situations for comorbid adolescents. Separate factors represented substance use situations involving conflict with others/unpleasant emotions (CO/ UE); positive social interactions/urges (PS/U); pleasant emotions (PE); physical/sexual (P/S); and testing personal control (TPC). Substance use situational profiles were determined through the conduct of cluster analyses on the substance use situation factors. A four-cluster solution was identified representing participants who used substances frequently in all situations (all high group); frequently in situations involving pleasant emotions (high PE group); and those who did not use frequently in any situations (all low group); or in pleasant emotions and physical/sexual situations (low PE/P/S group). Exploratory analyses revealed no differences among cluster groups in terms of age and gender, and limited differences in terms of ethnicity and comorbid psychopathology. Most notably, consistent differences were found among situation clusters when examining associations among cluster groups and substance use history. For example, adolescents were 1.1 times more likely to be classified in the high PE cluster, compared to the low PE and P/S cluster, for every 50 times they used marijuana in their lifetime (Wald = 7.68; p < .01; OR = 1.002). Further, adolescents were 1.23 times more likely to be classified in the high PE cluster, compared to the all low cluster, for every lifetime drug- related problem experienced (Wald = 9.13; p < .01; OR=1.23). Findings indicate that the clusters likely represent valid profiles of substance use situations for comorbid adolescents and that severity of substance use is likely associated with situational patterns of use. Evaluation of situational patterns of substance use as part of substance abuse treatment may help tailor coping skills interventions to subgroups of comorbid adolescents with differential situational substance use patterns. Replication of the current study and extension to longitudinal outcomes should be conducted to determine if cluster groups are reliable across samples, and if prospective relationships among clusters and substance use outcomes exist