Abstract
BACKGROUND: Predicting substance use in adolescence is a difficult yet important task in developing effective prevention. We aim to extend previous findings on the linear associations between familiarity with (knowledge of) substances in childhood and subsequent substance use in adolescence through a latent class analysis (LCA) to create risk profiles based on substance familiarity. METHOD: Using the ABCD Study(®) sample, we conducted an LCA using 18 binary substance familiarity variables (n = 11,694 substance-naïve youth). Complementary analyses investigated the relationship between LCA groups and (1) longitudinal use, (2) use initiation, and (3) early use. RESULTS: The optimal LCA resulted in a four-class solution: Naïve, Common, Uncommon, and Rare, with each group increasing in both the number and rarity of known substances. Analysis 1 revealed an increased risk in use over time among both the Uncommon and Rare groups (ORs = 2.08 and 5.55, respectively, p's < 0.001) compared to the Common group. Analysis 2 observed a decreased risk for initiation between the Naïve and Common groups (OR = 0.61, p = 0.009); however, the Uncommon and Rare groups were at an increased risk compared to the Common group (ORs = 2.08 and 3.42, respectively, p's < 0.001). Analysis 3 found an increased risk of early use between the Common and Uncommon groups (OR = 1.92, p < 0.001) with a similar trend between the Common and Rare groups (OR = 1.90, p = 0.06). CONCLUSION: These results highlight distinct risk profiles for adolescent substance use based on substance familiarity in middle childhood. Current work could be applied as an early screening tool for clinicians to identify those at risk for adolescent substance use.