Family, Friends Identified as Major Influences of e-Cigarette Use Among Adolescents

Image Credit: yta - stock.adobe.com

A recent study found that friend and family influences are a key risk factor for use of electronic nicotine delivery systems (ENDS) among US adolescents ages 12 to 17 years.

The study, published in JAMA Network Open, was fueled by the rapidly growing nicotine consumer population among middle and high school students over the past 5 years, leading the Surgeon General to declare a youth e-cigarette epidemic in 2018. Data was collected through a prognostic study examining the Population Assessment of Tobacco and Health (PATH) survey responses using machine learning (ML) techniques. ML methods allow high-dimensional survey data to be handled with multicollinearity and complex interactions at a highly accurate rate while identifying any patterns within the data.

Comparisons were analyzed in participants’ who reported at baseline in wave 4.5 (December 2017-December 2018) to their outcomes in wave 5 (December 2018-November 2019). The study’s main purpose was to identify the top risk factors of vaping initiation, including unanticipated risk factors such as body mass index (BMI) and dental or oral hygiene status.

The integration of ML techniques was applied through random forest-recursive feature elimination algorithm (RF-RFE), the eXtreme Gradient Boosting method (XGBoost), and Shapley Additive Explanation (SHAP). The study was ongoing, focusing on individual-specific information, including demographics, physical and mental health, tobacco usage behaviors, and tobacco risk perception.

Participants considered tobacco-naïve were reported to have no history of smoking cigarettes, cigars, cigarillos, filtered cigars, pipe, hookah, bidis, or kretek and have never used e-products, smokeless tobacco, snus pouches, or dissolvable tobacco of any kind. Predictions between current ENDS use within the past 30 days were compared to individuals who were tobacco-naïve at baseline.

Calculations for the most informative risk factors were associated with the ENDS use in wave 5 from wave 4.5 variables. Predictions on the wave 5 ENDS use are made based on training data with RF-RFE features to train XGBoost and optimize its hyperparameters.

Wave 4.5 included a total of 7943 tobacco-naïve adolescents, who then led to 332 participants (4.2%) reporting their current ENDS use in wave 5. Among those who were tobacco-naïve in wave 4.5, 5047 (63.5%) were between 12 and 14 years old, 4066 (51.2%) were male, and 2455 (30.9%) identified as Hispanic.

The most relevant variables found to impact the likelihood of using ENDS were if offered by a best friend (mean SHAP value, 0.184), the number of best friends using e-cigarettes (mean SHAP value, 0.167), household tobacco usage (mean SHAP value, 0.161), curiosity about ENDS use (mean SHAP value, 0.088), future intention to use ENDS (mean SHAP value, 0.068), grade level (mean SHAP value, 0.068), youth’s total average weekly earnings (mean SHAP value, 0.060), BMI (mean SHAP value, 0.055), perceptions of tobacco product safety (mean SHAP value, 0.026), and English writing proficiency (mean SHAP value, 0.024).

The top variables associated with the likelihood of ENDS use in adolescents were if they were offered an ENDS from a best friend, the number of best friends they have who use e-cigarettes, and overall household tobacco usage. Adolescents’ total average weekly earnings were significantly associated with future ENDS use, with higher weekly incomes found to correlate with a higher risk of ENDS use.

Overall, the study found that tobacco-naïve participants with ENDS-using best friends who also reported that they would try the product if offered one or that they lived with a tobacco-using individual were the most at risk for using ENDS in the future.

Limitations of the study included the sole reliability on available variables from the survey data. It is highly possible the study failed to acknowledge other important factors. The ML model used in this study consisted of slightly alternating rankings based on the algorithm’s randomized nature. The results did not create casual relationships within the data but suggested the risk factors in conjunction with outcome variables.

The author concluded, “These findings showed significant associations of family and peer influence with current ENDS use after 1 year of follow-up among adolescents who were tobacco-naive at baseline. Given the popularity of ENDS use in the past years, the prominence of these risk factors raises concerns about their potential contribution to the widespread use of ENDS. Furthermore, these findings also underscored the pivotal role of education in shaping adolescents’ awareness of tobacco-related matters, serving as protective measures against these detrimental products.”

Reference

Le TTT. Key risk factors associated with electronic nicotine delivery systems use among adolescents. JAMA Netw Open. Published online October 20, 2023. doi:10.1001/jamanetworkopen.2023.37101