E-cigarettes remain a contested public health issue, with countries adopting divergent regulatory approaches. Singapore has maintained a comprehensive prohibition on the importation, sale, distribution, and possession of e-cigarettes since 2018, with recently strengthened enforcement measures and penalties [1-4]. While understanding public response to these measures is important, survey- and interview-based approaches are subject to bias from disclosure fears around legal repercussions [1], making social media a valuable complementary data source. This study aims to leverage text-mining tools to provide a fine-grained characterization of vaping-related discourse on social media, with the goal of informing potential strategies for vaping policies.

Descriptive statistics are detailed in Table S1 in Multimedia Appendix 2. Figure 1 illustrates the sentiment and topic composition across time, showing that discourse is predominantly neutral, followed by negative and then positive. Nonetheless, there were occasional periods where positive sentiment exceeded negative, notably around mid-2018, mid-2023, and mid-2024.

Table 1 reveals 18 topics identified through BERTopic, which can be grouped into 4 dimensions: (1) legislation and policy-related, including vape ban legislation (topics 0, 10) and nanny state critiques akin to gum-related laws (topic 6); (2) illicit consumption, covering enforcement-related incidents involving students, train and cable car commuters, and government officials (topics 1, 7, 13, 15); (3) illicit sales and smuggling, including smuggling via borders (topics 2, 4, 16), black market sales (topics 5, 17), and distribution via messaging platforms (topic 11); and (4) health risks and substances.

A breakdown of the type of account, or source of information, across topics is detailed in Multimedia Appendix 2, in addition to external evaluation of BERTopic’s identified topics and assignment of each topic to higher-level themes [7].

“Legislation and policy-related” sentiments fluctuated, with increased discourse during legislative and policy changes. Neutral-dominated discourse encompassed news- or information-sharing by individuals or news outlets on topics such as banning vapes, amendments, and etomidate’s drug classification (topics 0, 10, 12, respectively), per Multimedia Appendix 2 and the quoted examples of topics 10 and 12 in Table 1. Negative posts primarily surrounded nanny state criticisms (topic 6) and individuals opposed to blanket bans on vapes (topic 0) as demonstrated by the topic-specific quotes in Table 1. Positive posts primarily spanned support for such legislation (topic 0), although they were less pronounced than other sentiments.

“Illicit consumption” discourse increased during periods coinciding with vaping incidents involving students, train and cable car commuters, and government officials (topics 1, 7, 13, 15, respectively), including spikes in 2018 and late 2025 during heightened crackdowns. Posts were primarily neutral, encompassing information-sharing from news outlets or individuals, followed by negative sentiments, which encompassed frustration directed at the culprits and the reported incidents themselves, as best illustrated in topics 1 and 7 of the quotes in Table 1.

Similar trends were observed in “illicit sales and smuggling” discourse, where posts were predominantly neutral, followed by negative discourse, with few positive discussions and increased discourse during crackdowns. Discussions surrounded reports of smuggling via checkpoints, neighboring country Malaysia, and airports (topics 2, 4, 16, respectively), illicit sales networks via black markets (topics 5, 17) and Telegram (topic 11), and punishments for smuggling offenders (topics 3, 14). Per Multimedia Appendix 2, a portion of these posts stemmed from news outlets and nongovernmental groups. Most neutral discussions surrounded news sharing, commentary, and individual observations, per the quotes in topics 3, 4, and 5 of Table 1, respectively, while negative discussions leaned toward anger directed at smuggling incidents themselves, per the quote in topic 2 of Table 1.

Finally, discussions on “health risks and substances” were predominantly neutral, which included discussions on dangers and effects of vaping as well as Kpod use (topics 8, 9, respectively). Most discussions stemmed from news outlets disseminating findings and individuals commenting on vaping-related health issues, per Figure 1C and the topic-specific quotes in Table 1.

This study aimed to characterize vaping-related discourse on social media to inform potential implementation strategies, particularly as survey-based approaches are susceptible to disclosure bias in ban jurisdictions [1]. It found that discourse was predominantly neutral and event driven, with discussions on legislation, illicit consumption, sales and smuggling incidents, and health risks.

Echoing qualitative reports of regulation-over-ban preferences and skepticism toward anti-vaping messaging [2,3], such views (topics 0, 6) surfaced in negative posts within predominantly neutral discourse, while positive posts endorsed legislation. The illicit market and youth vaping culture documented qualitatively [2-4] proved publicly visible: neutral reporting dominated (topics 1, 5, 11), with negativity targeting offenders.

These findings also point to implementation frameworks that could guide vaping control efforts. The Consolidated Framework for Implementation Research [8] could help address public acceptability concerns such as nanny state criticisms; the dynamic sustainability framework [9] could inform adaptive enforcement as illicit market channels shift; and policy feedback theory [10] could explain the apparent mutual reinforcement between enforcement actions and public support.

Nonetheless, this study has limitations: data were drawn exclusively from X, which may not represent broader public opinion, with user demographics and location often masked.

As vaping policies evolve worldwide, infodemiology can complement conventional methods to support responsive and adaptive implementation strategies.