Mapping the E-Cigarette Discussion: A Computational Analysis of Textual Stance, Image Categories, and Social Media Engagement on Twitter/X

Background The rise of electronic cigarettes (e-cigarettes) has generated widespread controversy. As a key platform for public discussion, Twitter/X provides a valuable context for examining how textual stance, images, actor types, and user engagement intersect in e-cigarette discourse.Methods This study analyzed 19,983 image-containing tweets, including 24,676 images and 21,976 replies. Image and text classifications were conducted using an AI-assisted coding approach. User engagement was measured by likes, retweets, and replies. BERTopic was used to identify major reply topics.Results Promotions (34.6%), Vaping Advocacy and Rights (21.2%), and Health Warnings and Infographics (16.5%) emerged as the dominant image categories. 16.78% of images categorized as Health Warnings and Infographics showed a text–image mismatch. Retailers and vaping communities were more active in image production, whereas health organizations contributed fewer images. In pro-e-cigarette textual contexts, images depicting vaping acts predicted higher levels of likes (B = 0.258, p < 0.001), replies (B = 0.093, p < 0.001), and retweets (B = 0.099, p < 0.01). By contrast, most image types in anti-e-cigarette textual contexts did not significantly predict engagement. User replies included discussions of everyday e-cigarette use, policy debates, and skepticism toward authoritative institutions.Conclusion E-cigarette images on Twitter/X are not only promotional tools but also part of public debates over health risks, regulation, and vaping rights. Their meanings are shaped by textual stance, and their distribution differs across actor types. These findings suggest that health communication should strengthen its sustained visibility and narrative appeal to respond to pro-vaping narratives and related controversies.