Zhang · Addictive behaviors 2021 · Retrospective longitudinal cohort study · n=10,644

Adolescent drug use initiation and transition into other drugs: A retrospective longitudinal examination across race/ethnicity.

Cited 29 times in the scientific literature.

Level 3 - non-randomized controlled study

Retrospective longitudinal analysis of survey cohort data using survival models

PubMed 33032193 · doi:10.1016/j.addbeh.2020.106679 · record verified 2026-08-26

What was done

Analyzed pooled retrospective data from the National Survey on Drug Use and Health (2015–2018) for adolescent illicit drug users aged 14–17 years (excluding tobacco and alcohol; n = 10,644). Drug initiation history and progression to other illicit substances were mapped using reported age of initiation. Transition trajectories and racial/ethnic disparities were evaluated using life table methods and multivariate Cox proportional hazards regression models.

What was found

Two-thirds of adolescent drug users initiated with marijuana, one-quarter with inhalants, and the remainder with hallucinogens, prescription drugs, or hard drugs. By year 8 post-initiation, the cumulative probability of transitioning to a new drug was approximately 40% for inhalant initiators and 70% to 80% for initiators of other drugs. The probability of transition accumulated stably over time for initial marijuana and inhalant users. Black/African American adolescents were significantly less likely than other racial/ethnic groups to transition to an additional drug.

Why it matters

The study maps initial adolescent illicit drug choices and subsequent progression risks over an 8-year span. Identifying substance-specific and demographic transition pathways supports early, targeted screening and intervention to halt drug escalation.

Limits

Relies entirely on retrospective self-reported age of initiation, which is prone to recall bias and underreporting. Excluded alcohol and tobacco from the initiation definitions, omitting two common early substances. The abstract does not provide exact hazard ratios, confidence intervals, or specific covariate adjustments.

Cited by