Latent class analysis of substance use among adolescents presenting to urban primary care clinics

Abstract

Background: Polysubstance use during adolescence is a significant public health concern. However, few studies have investigated patterns of substance use during this developmental window within the primary care setting. Objectives: This study used an empirical method to classify adolescents into substance use groups, and examines correlates of the empirically defined groups. Methods: Data came from patients, ages 12–18 years, presenting to an urban, primary care community health clinics (Federally Qualified Health Centers) in two cities in the Midwestern United States (n = 1664). Latent class analysis (LCA) was used to identify classes of substance users. Multinomial logistic regression was used to examine variables associated with class membership. Results: LCA identified three classes: class 1 (64.5%) exhibited low probabilities of all types of substance use; class 2 (24.6%) was characterized by high probabilities of cannabis use and consequences; and class 3 (10.9%) had the highest probabilities of substance use, including heavy episodic drinking and misuse of prescription drugs. Those in class 2 and class 3 were more likely to be older and have poorer grades, poorer health, higher levels of psychological distress and more sexual partners than those in class 1. Individuals in class 3 were also less likely to be African-American than those in class 1. Conclusion: Findings provide novel insight into the patterns of substance use among adolescents presenting to low-income urban primary care clinics. Future research should examine the efficacy of interventions that address the complex patterns of substance use and concomitant health concerns among adolescents.

• Adolescent health
• latent class analysis
• primary care
• polysubstance use

Acknowledgements

This project was supported by a grant (DA020075) from the National Institute on Drug Abuse (NIDA). We would like to thank project staff: Cassandra Brooks, Amber Hart, Angela Turkelson, Carolyn Gibson, NaTosha Murry, Janelle Kapa, Karen Ross, Kay Doerr, Bianca Burch and Jessica Lantz. We would also like to thank Linping Duan for statistical support. Finally, special thanks are owed to the patients and medical staff at Hamilton Community Health Centers, Health Delivery Inc., and Mott Children’s Health Center for their support of this project.