What puts some kids at risk for using alcohol, cannabis, or nicotine?
What is it about?
We identified factors at ages 9 – 10 that predict substance use by early adolescence. Understanding predictors of early substance use is important, because early substance use significantly increases the likelihood of developing problems with alcohol, cannabis, and other drugs, and is related to a number of health, educational, and occupational issues. In a sample of 6,829, approximately 14% of youth initiated substance use initiation by age 12. Substance use initiation was defined as trying a single sip of alcohol, puff from a vape pen or cigarette, or trying cannabis or other substances. We examined over 400 predictors of substance use initiation and found our top predictors of initiation were in the self-report category (i.e., demographics, mental health, physical health). Top predictors of substance use initiation included self-report factors. Religious preference showed mixed results for different religions (e.g., some religions were associated with less substance use and some with more). Black youth were less likely to initiate than White youth, and youth from lower income backgrounds and with less parental education were also less likely to initiate. Other top predictors of early substance use that may be modifiable included history of more detentions or suspensions in schools, more prenatal substance use exposure, and greater substance use availability. These factors were associated with greater likelihood of substance use initiation. The addition of more time-consuming and expensive measures such as hormones, neurocognitive factors, or neuroimaging did not improve our ability to predict substance use initiation. Modifiable factors associated with substance use initiation may be targets in the context of prevention or early intervention efforts to delay onset of substance use among youth. The longer youth can delay substance use initiation, the better their health outcomes are.
Why is it important?
Previous studies have primarily examined a smaller subset of variables across 1 – 2 domains as predictors of substance use among youth. To provide a comprehensive assessment of predictors of substance use initiation, we included 420 variables from the following domains: demographics, self- and peer- risk factors for substance use, parenting behaviors, mental health, physical health, culture and environment, neurocognition, hormones, and neuroimaging variables. We used a machine learning approach to assess a) whether the addition of more resource-intensive measures such as hormones, neurocognitive functioning, and neuroimaging improved our ability to predict substance use initiation and b) of the 420 variables, which were most strongly associated with substance use initiation. Our findings expand upon previous studies by including an extensive number of predictors across various domains and using advanced quantitative methods that address the complexity of the data in a large sample. The most predictive factors were those that are fairly inexpensive and easy to collect for clinicians.