Chen Yu, Wang Zining, Jiang Shaoying, Cai Yujiang, Xu Jing, Wang Ying
School of Art and Communication, Fujian Polytechnic Normal University, Fuqing, China.
School of Journalism and Communication, Peking University, Beijing, China.
Tob Induc Dis. 2025 Jul 4;23. doi: 10.18332/tid/204743. eCollection 2025.
E-cigarette use is rising among young adults globally, and college students are particularly vulnerable due to high social media engagement and targeted promotions. Understanding which factors predispose this population to initiate vaping is critical for designing effective prevention strategies.
We conducted a cross-sectional survey of 303 never-smoking, never-vaping Chinese college students (aged 18-24 years) recruited via online platforms and referrals. The 25-item questionnaire assessed six domains: demographics, parental smoking, peer e-cigarette use, 'quasi-deviant' behaviors (regular alcohol consumption and bar attendance), social media use and trust, and exposure to e-cigarette marketing across five media channels. A three-item susceptibility scale was combined into a single index via principal component analysis. An Extremely Randomized Trees classifier (n_estimators=60, max_depth=6) with grid-search and five-fold cross-validation on a 75:25 train-test split, identified the strongest predictors of high susceptibility. Model performance was evaluated by accuracy and area under the receiver operating characteristic curve (AUC).
The model achieved 81% classification accuracy. Feature importance (FI) indicated that bar attendance (FI=0.21), alcohol consumption frequency (FI=0.12), exposure to e-cigarette marketing messages (FI=0.08), social media use (FI=0.08), peer e-cigarette use (FI=0.05), and parental smoking (FI=0.05) were the most influential predictors. Among the participants, 18.8% were classified as high-susceptibility, indicating elevated risk for future vaping initiation.
'Quasi-deviant' behaviors (regular alcohol use and bar attendance), social media marketing exposure, and social influences (peer and parental smoking) are key predictors of e-cigarette susceptibility in Chinese college students. Multi-level prevention strategies - enforcing digital marketing restrictions, peer-focused education, and integrated substance-use interventions - may effectively reduce susceptibility and avert vaping initiation in this high-risk group.
全球范围内，年轻人使用电子烟的情况呈上升趋势，由于大学生在社交媒体上的活跃度高且受到针对性推广，他们尤其容易受到影响。了解哪些因素使这一人群开始使用电子烟对于制定有效的预防策略至关重要。
我们对通过在线平台和推荐招募的303名从未吸烟、从未使用过电子烟的中国大学生（年龄在18 - 24岁之间）进行了横断面调查。这份包含25个条目的问卷评估了六个领域：人口统计学特征、父母吸烟情况、同伴使用电子烟情况、“准越轨”行为（经常饮酒和去酒吧）、社交媒体使用与信任，以及在五个媒体渠道接触电子烟营销的情况。通过主成分分析将一个包含三个条目的易感性量表合并为一个单一指标。使用极端随机树分类器（n_estimators = 60，max_depth = 6），在75:25的训练-测试分割上进行网格搜索和五折交叉验证，确定了高易感性的最强预测因素。通过准确率和受试者工作特征曲线下面积（AUC）评估模型性能。
该模型的分类准确率达到81%。特征重要性（FI）表明，去酒吧（FI = 0.21）、饮酒频率（FI = 0.12）、接触电子烟营销信息（FI = 0.08）、社交媒体使用（FI = 0.08）、同伴使用电子烟（FI = 0.05）和父母吸烟（FI = 0.05）是最具影响力的预测因素。在参与者中，18.8%被归类为高易感性，表明未来开始使用电子烟的风险升高。
“准越轨”行为（经常饮酒和去酒吧）、社交媒体营销接触以及社会影响（同伴和父母吸烟）是中国大学生对电子烟易感性的关键预测因素。多层次预防策略——加强数字营销限制、以同伴为重点的教育以及综合的物质使用干预——可能有效降低易感性，并避免这一高风险群体开始使用电子烟。
