We used the Real-time Infoveillance of Twitter Health Messages software framework to collect Twitter posts (ie, tweets) containing the terms “puffbar” or “puff bar” for 1 month from July 13, 2020, to August 13, 2020 [28]. The Real-time Infoveillance of Twitter Health Messages allows for real time collection of all publicly available tweets matching a specified set of keywords through Twitter’s filtered data stream application programming interface. This start date aligned with the date Puff Bar announced ceasing online sales in the United States [9]. This resulted in 13,304 tweets, of which 4785 (36%) were retweets and 8519 (64%) were original tweets.

We obtained a 20% random subsample of original tweets (n=2661) for human coding; this process has previously demonstrated to be both feasible for human qualitative coding and generalizable to the full data set [28].

The University of Pittsburgh Institutional Review Board (IRB) determined that the proposed activity is not research involving human subjects as defined by DHHS and FDA regulations (STUDY19080214).

We developed a codebook based on a hybrid process that included consideration of our research question, prior analysis of e-cigarette discourse on Twitter [14,29], tobacco policy, and an inductive analysis of 100 relevant tweets that were not included in the final sample. First, we included a code for relevance to the research topic of Puff Bar-related discussions (relevant). Tweets that did not contain the disposable e-cigarette Puff Bar as the main topic (eg, “I go to my car to get my puff bar and somebody left me a rose on my car. I can’t stop smiling!”) were deemed not relevant.

Relevant tweets were coded for topics informed by 2 areas of research, which were nicotine dependence and tobacco policy. Topics for nicotine dependence included language that suggests dependence on Puff Bar and signs of nicotine addiction or withdrawal related to Puff Bar use. These categories were informed by prior research that examined similar content posted about JUUL [14]. Topics for tobacco policy included whether posted content were commercially marketing Puff Bar or a business selling Puff Bar, references to purchasing a Puff Bar, references to underage use of Puff Bar, regulations of Puff Bar, price of Puff Bar, references to Puff Bar flavor, and references to other e-cigarette products. The codebook was validated through analysis of 100 relevant tweets that were not included in the final sample by 2 experienced Twitter coders. Following this, the final codebook was codified, presenting clear definitions and examples for each code (Table 1).

The tweets were coded by 2 independent individuals, with adjudication of disagreements by a supervising researcher. The coders were provided with the tweet text and a URL link to each tweet. During the coding process, all relevant tweets that were publicly available at the time of coding were viewed on Twitter so that visuals in the tweet (eg, images, videos, and emoji) could also be assessed. The text from unavailable tweets was still included in the coding and thematic analysis to preserve comprehensiveness of the original data. The codes were not mutually exclusive.

We calculated interrater reliability using the Cohen κ, and disagreements were adjudicated between the 2 coders. In instances where the coders could not reach consensus, the lead author had final determination. After 4 rounds of independent double coding (100 tweets each round), interrater reliability was considered good to excellent (Cohen κ=0.71-1.00) for all categories [30]. The remaining tweets were split between the 2 coders for independent coding.

We calculated descriptive statistics for each coding category and used a thematic content analysis approach to analyze qualitative data [31]. To conduct the content analysis, the coders wrote annotations and memos throughout the coding process and highlighted specific words or phrases within tweets that exemplified the themes. The coders then met with supervising researchers to synthesize themes with salient examples from the observed data.

This study was approved by the University of Pittsburgh Institutional Review Board. To protect tweeters from identification, all examples provided in the text and tables are paraphrased versions of original tweets.

This study examined Puff Bar-related discussions on Twitter to identify themes related to tobacco policy and dependence. Despite federal regulations and FDA warnings against Puff Bar, the results of this study suggest that the purchasing and use of Puff Bar products are still being discussed on social media. Discussions of Puff Bar flavors were prevalent, accounting for a plurality (45.9%) of tweets classified as relevant to tobacco policy; these products are circumventing a federal ban designed to protect youth from the appeal of nicotine products. Similarly, the second most common theme was focused on regulations (n=124, 17.8%). As a result, the outcome of the FDA ban of e-cigarette flavors did not reduce demand, but rather shifted more attention to different e-cigarette brands that utilize a loophole for disposable devices.

We found similar dependence and acute nicotine effect themes as prior research on JUUL products and cigarettes, such as self-report of dependence, frustration over losing access to the device, compulsive use, and self-reports of physical effects of nicotine exposure including suppression of appetite [14,32]. Again, this suggests that despite legislative and policy activities to reduce youth access to and use of flavored, high nicotine-containing and dependence-forming products, they continued to be available due to exploitation of loopholes. The fact that high school students’ use of disposable e-cigarettes such as Puff Bar increased by 1000% from 2019 to 2020 is evidence that these loopholes can contribute to continued use of e-cigarettes by youth.

Due to new federal tobacco policies, discussions about Puff Bar offer a view into topics relevant to tobacco policy including e-cigarette regulations and how they might affect access to Puff Bar (17.9%) and buying Puff Bar (16.8%). The results suggest that while the legality of Puff Bar sales might be recognized and discussed by Twitter users, people continue to purchase them even though the products are banned in the United States. This is especially alarming considering that commercial content (ie, posts by retailers) were very low (2.9% of tobacco policy tweets, 0.9% of all relevant tweets) when compared with other e-cigarette research [33,34].

The results of this study align with data from the NYTS that demonstrate a shift to disposable e-cigarettes following FDA tobacco policies. While the federal flavor ban was enacted to curb youth e-cigarette use, it is unfortunate that exemptions created loopholes to allow continued access to flavored products. While outside the scope of this study, future research should consider identifying what types of accounts are posting this content. Specifically, understanding whether commercial tweets are being posted by Puff Bar users, commercial vendors, news organizations, or others could help to better inform counter messaging or preventive measures. Additionally, it would be valuable for researchers to be guided by a framework that can address both legal and health concerns. For example, legal epidemiology, the scientific study and deployment of law as a factor in the cause, distribution, and prevention of disease and injury in a population, might inform stronger and more comprehensive policies [35]. Legal epidemiology involves the three following components: (1) legal prevention and control—the study and application of laws and legal practices as interventions to prevent disease and injury and as enablers of effective public health administration; (2) legal etiology—the study of laws and legal practices as causes of disease and injury; and (3) policy surveillance—the ongoing, systematic collection, analysis, and dissemination of information about laws and other policies of importance to health [36]. In combination with continued real time infoveillance of e-cigarette discussions, this framework could help better understand how tobacco policy (eg, age or flavor restrictions) and implementation can affect health outcomes.

Puff Bar acknowledged that their products were excluded from the FDA regulation banning most flavors. On February 21, 2020, Puff Bar posted on their official blog a lament about federal regulations “determined to eliminate vaping as a whole” stating that “disposable devices like Puff Bars and e-liquid used in refillable tank systems can still carry flavors like fruits and desserts” [37]. These, and other similar arguments, have been used by JUUL in order to continue promoting their products, particularly through advertisements targeted to adolescents [38]. Puff Bars—which have now resumed sales—have become the latest device to replace previously banned tobacco products. Until we are able to develop comprehensive tobacco policy, new products that capitalize on policy loopholes will continue to supply the demand for nicotine, particularly by adolescents through targeted advertisements or flavored products. We suggest that a framework such as legal epidemiology could offer a unique lens to understand the confluence of tobacco policy and health outcomes, helping inform policy makers to better understand and strengthen the practical implications of tobacco policies.

The results of this study should be considered in the context of the following limitations. Twitter users are not representative of the general population, although it is frequently used by adolescents and young adults [27], a population that also frequently uses e-cigarettes. While interpretation of tweets using qualitative analysis can be subjective, we minimized subjectivity by using a systematic coding procedure and the use of experienced Twitter coders; nonetheless, these tweets are discussion of purchasing, rather than the act itself. We also did not use location data; thus, there is the possibility that some tweets are by non-US users. Finally, our results are constrained by the keywords and time period used in our search parameters. Future research could expand these parameters to widen the scope of the investigation.

Here, we summarize the key findings of our research: (1) laws and regulations around e-cigarettes are rapidly changing in response to increased concern about the use of the products by youth, and prior research suggests the focus of flavor bans on devices such as JUUL may have created a policy loophole that was filled by disposable devices such as Puff Bar; (2) however, it is not yet known how Puff Bar is being used as an alternative to traditional e-cigarettes that fall under federal and state restrictions; (3) our analysis of tweets related to Puff Bar suggests that the FDA ban of e-cigarette flavors did not reduce interest, but rather shifted the discussion to brands utilizing a loophole for disposable devices and suggests the importance of using a framework such as legal epidemiology when researching and evaluating tobacco policy.

This study found similar dependence and acute nicotine effect themes in Puff Bar-related discussions on Twitter compared to prior research on JUUL and cigarettes [14]. We also found that discussions about Puff Bar on Twitter provided insight into topics relevant to tobacco policy, including flavors, e-cigarette regulations, and purchasing Puff Bar. Our results, in conjunction with evidence by the NYTS and other data sources, suggest that the FDA ban of e-cigarette flavors did not reduce demand, but rather shifted the supply to brands utilizing a loophole for disposable devices. Until comprehensive tobacco policy legislation is developed, new products or loopholes will continue to supply nicotine demand.