Latent class analysis of substance use and predictors of latent class membership among adolescents in the Republic of Korea

Abstract

ABSTRACTPurpose: We aimed to investigate substance use patterns and identify predictors of homogeneous subgroups of adolescent substance users.Methods: We analyzed nationally representative secondary data collected from Korean adolescents (N = 72,435). To investigate substance use patterns, we conducted latent class analysis using seven behaviors linked to alcohol, cigarette, and e-cigarette use. After choosing the best latent class model, we investigated predictors of latent class membership (LCM) for substance use, using demographics and mental health conditions.Results: A four-latent class model best fit the data. Non-users (74%) had low likelihoods of reporting lifetime and current use of alcohol, cigarette, and e-cigarette. Experimenters (10%) had high likelihoods of reporting lifetime alcohol and cigarette use. Current drinkers (10%) had high likelihoods of reporting lifetime and current alcohol use. Multi-substance users (6%) had high likelihoods of reporting lifetime and current use of alcohol and cigarettes, lifetime e-cigarette use, and current binge drinking. Additionally, demographics (gender, grades, socioeconomic status, co-residence with family members, and grade point average) and mental health conditions (depression, suicidal ideation, and subjective unhappiness) successfully predicted LCM.Conclusions: In developing interventions for addressing substance-related issues, health professionals should focus on adolescent substance use patterns and take into account factors predicting LCM.