Abstract Background As Turkiye ranks second in cigarette consumption in Europe, the increasing shift to Electronic Nicotine Delivery Systems (ENDS) presents emerging public health concerns. This study aims to introduce a hybrid approach by combining AI-driven large language models (LLM) with traditional thematic analysis to explore 20 years of ENDS discourse on Eksisozluk, a popular Turkish social media platform, using the Health Belief Model (HBM). Methods This mixed-methods study, analyzed 5,907 comments about ENDS from 2003 to 2023, sourced from a renowned online user-generated forum in Turkiye. Utilizing an LLM, comments were first categorized according to the HBM. Subsequently, the LLM’s generative capabilities were employed to identify sub-themes through thematic analysis. A temporal analysis was then conducted to examine the evolving trends over the 20-year span. Results ENDS popularity peaked in 2017, dipped during the COVID-19 pandemic, and later rebounded. Using an LLM with 70% accuracy, validated by a manually labeled sample, we discerned two main themes: perceived benefits (40.5%) - primarily using ENDS as for harm reduction (32.8%) and smoking cessation tools (21.6%)- and perceived barriers (27.7%), with cost and accessibility (21.9%) being the most significant throughout the period. Initially, the focus was on cessation, but shifted to harm reduction post-pandemic. Conclusions Our study indicates that consumers in Turkiye initially used ENDS for cessation but recognized their ineffectiveness over time. During the COVID-19 pandemic, usage shifted towards harm reduction, likely influenced by the rhetoric of the Tobacco Industry. Despite a national ban, ENDS remain widely used in Turkiye. Key messages • Turkiye should develop a social marketing strategy to raise awareness that ENDS is not a valid tool for harm reduction. • Although the sale of ENDS is prohibited in Turkiye, enforcement of bans should be strengthened as part of an effective control strategy.
Read full abstract