Screening Substance-Naive Children for Risk of Frequent/Heavy Substance Use in Young Adulthood: A 17-Year Prospective, National Study

Objective: There is no externally validated instrument for predicting which children will show frequent substance use (SU) in young adulthood. This study evaluated whether an instrument previously shown to predict SU outcomes in mid-adolescence could also predict SU outcomes in young adulthood, comparing performance across ages and demographics. Method: The Loeber Risk Score (LRS) is a 5-item, parent-completed screener indexing risk of future SU in children. We evaluated the predictive performance of the LRS in a nationally representative longitudinal birth cohort (N = 4,898, 48% female). Parents completed the LRS when the child was ~5 and ~9 years old. Subsequently, youth reported their SU at age ~22 years old. Results: The LRS at age 9 predicted several outcomes at age 22 better than chance (ps<0.05): daily cigarette use (AUROCs=0.71-0.75), cannabis use ≥3x week (AUROC=0.59), vaping ≥3x week (AUROC=0.55), and receipt of treatment for an alcohol/drug problem (AUROC=0.60). Performance was no better than chance for alcohol outcomes (AUROCs=0.47-0.50). Children with LRS scores ≥2 were 1.3-2.0 times more likely to display frequent substance use outcomes. There were no consistent performance differences when the LRS was measured at age 5 vs. age 9 or by child’s sex, race, or ethnicity. Conclusion: The LRS had modest accuracy when predicting some substance use outcomes in young adulthood, but it does not predict alcohol outcomes any better than chance. Better screeners are needed.