E-cigarette-related Nicotine Misinformation on Social Media

Abstract

Background.

Twitter provides an opportunity to examine misperceptions about nicotine and addiction as they pertain to electronic nicotine delivery systems (ENDS). The purpose of this study was to systematically examine a sample of ENDS-related tweets that presented information about nicotine or addiction for the presence of potential misinformation.

Methods.

A total of 10.1 million ENDS-related tweets were obtained from April 2018 through March 2019 and were filtered for unique tweets containing keywords for nicotine and addiction. A subsample (n=3,116) were human coded for type of account (individual, group, commercial, or news) and presence of potential misinformation.

Results.

Of tweets that presented ENDS-related nicotine or addiction information (n=904), 41.7% (n=377) contained potential misinformation coded as anti-vaping exaggeration, pro-vaping exaggeration, nicotine is not addictive or is never harmful, or unproven health benefits.

Conclusions.

Anti-vaping exaggeration tweets distorted or embellished claims about ENDS nicotine and addiction; pro-vaping exaggeration tweets misinterpreted results from scientific studies. Misinformation that nicotine is not addictive or is never harmful or has unproven health benefits appeared less but are potentially problematic. ENDS-related messaging should be designed to be easily understood by the public and monitored to detect the spread of misinterpretation or misinformation on social media.

Introduction

Use of electronic nicotine delivery systems (ENDS) has been increasing in the United States, particularly among adolescents. The 2019 National Youth Tobacco Survey found that 27.5% of high school students and 10.5% of middle school students use ENDS, representing an increase of 135% and 218%, respectively, since 2017 (Cullen et al., 2019). Although the 2020 National Youth Tobacco Survey reported high school current e-cigarette use declined slightly to 19.6%, the use of disposable e-cigarettes increased by 1000% among high school students from 2019–2020 (T. W. Wang et al., 2020). Additionally, although many youth and young adult e-cigarette users decreased use or quit during the COVID-19 pandemic, those who used e-cigarettes more often or were nicotine dependent were less likely to do so (Gaiha et al., 2020). These trends are concerning considering the increased risk for transition to combustible tobacco products and potential pulmonary and cardiovascular risks associated with ENDS use (Alzahrani et al., 2018; Berry et al., 2019; Kleinman et al., 2020; V P Krishnasamy et al., 2020; Moheimani et al., 2017).

Although ENDS and e-liquids without nicotine are available, most products contain nicotine, often in high concentrations (Jackler & Ramamurthi, 2019). Systematic surveys have identified common misperceptions about ENDS and nicotine. For example, 63% of youth who use the popular JUUL device do not know that it always contains nicotine (Willett et al., 2019), and some youth believe that the nicotine in ENDS is artificial, and therefore harmless (Pepper et al., 2018). Additionally, some youth do not believe that ENDS are addictive; one survey of youth found that 68% of youth who were committed never ENDS users (i.e., never ENDS users who were non-susceptible to use) believed ENDS to be addictive, but this was lower for current ends users (59%) (Bernat et al., 2018). Another survey of youth found that those who were current or former JUUL users were 2.6 and 2.2 more likely, respectively, to believe it is “very unlikely” to become addicted to JUUL compared to never users (Russell et al., 2020). This is despite evidence that ENDS products, particularly JUUL, can produce symptoms of dependence and/or addiction (Amato et al., 2021; Dobbs et al., 2020; Sidani et al., 2019). Finally, the majority of young adults believe that nicotine is the root cause for diseases associated with tobacco product use (Mumford et al., 2017). For example, the majority of young adults believe that a relatively large or very large part cancer risk from cigarette use is due to nicotine, despite nicotine not being carcinogenic (Hecht, 2003; Villanti et al., 2019).

Social media platforms present the opportunity to conduct real-time surveillance of ENDS-related content to investigate misperceptions that may not be captured by traditional surveys (Brett et al., 2019; Kavuluru et al., 2019), and prior research suggests ENDS-related content on social media often contains misperceptions or misinformation (van der Tempel et al., 2016). This is particularly concerning, because false information tends to spread more rapidly on social media compared with factual information (Chou et al., 2018). Recent research also suggests that “bots”—or automated Twitter accounts that produce content and artificially engage with other accounts—disseminate a variety of health-related information on Twitter (Allem, Majmundar, et al., 2019) and are more likely than human users to reference ENDS as a tool for smoking cessation (Allem et al., 2017; Martinez et al., 2018). However, there is little research on the presence and spread of potential misinformation about ENDS on Twitter, especially focused on themes related to nicotine and addiction.

Therefore, the purpose of this study was to systematically examine tweets about ENDS to determine whether such statements contain potential misinformation. These results will inform public health strategies around surveillance, the context in which ENDS-related misinformation is being shared, and the sources of potential misinformation in order to provide intervention targets.

Methods and materials

Data collection and sampling

Using open source “RITHM” software(Colditz et al., 2018), approximately 10.1 million tweets matching ENDS-related keywords (JUUL*, vape*, ecig*, e-cig*) were collected from April 2018 through March 2019. Of these, 4.2 million were original tweets and 5.9 million were “retweets” (i.e., rebroadcasts of others’ content). Tweets containing keywords related to nicotine or addiction (nicotine* and/or addict*) were then identified, which included 155,804 unique tweets and 578,561 retweets. To balance feasibility human coding while maintaining thematic representativeness of the content, a simple random subsample was drawn from the unique tweets (i.e., excluding redundant retweet content) using standard RITHM software functionality. A 2% subsample resulted in 3,116 unique tweets, which was feasible for thematic coding. This data reduction approach is consistent with established practices in Twitter research and the subsample was slightly larger than those used for content analysis of ENDS contexts in prior work (Colditz et al., 2018; Martinez et al., 2018). Codebook development involved a separate pool of random tweets (i.e., not from the 3,116 primary tweets). This study was approved by the University of Pittsburgh Human Research Protection Office Human Research Protection Office.

Codebook development and coding procedures

Initial codebook development was guided by conceptual work in misinformation and science denialism (Diethelm & McKee, 2008). As outlined by Diethelm and McKee (2008), the five characteristics of science denialism are (1) conspiracies—suggesting scientific consensus is the result of a complex and secretive conspiracy; (2) fake experts—citing non-authorities, combined with denigration of established experts; (3) selectivity—referring to isolated papers that challenge scientific consensus; (4) impossible expectations—expecting 100% certain results or health treatments with no potential side-effects; and (5) misrepresentation and false logic—jumping to conclusions and using false analogies. When developing the codebook, coders and the lead author examined ways in which these characteristics manifested in the presentation of ENDS or nicotine information. For example, coders observed that tweets promoting health benefits of nicotine demonstrated selectivity, leading to the creation of a misinformation code for health benefits of nicotine. While this framework was the basis for initial codebook development, the resulting codes did not strictly adhere to science denialism concepts and instead were derived from the available data. Individual codes were developed through a 3-step, iterative coding process in which (1) two independent coders reviewed and annotated a subset of 100 tweets, (2) disagreements were adjudicated with lead author oversight, and (3) codes were added, merged, or clarified as needed. After 3 iterations, a thematic framework was finalized (Table 1).

Table 1.

Definitions for categorical codes, inter-rater reliability, and example tweets.

Code	Cohen’s κ	Definition	Examples
Nicotine or Addiction Information	0.80	The tweet mentions a statement about ENDS-related nicotine or addiction presented in a factual way.	• Vaping isnt healthier one vape has enough nicotine for 20 cigs • Vaping is pushing teenagers’ nicotine use to record highs
Nicotine is Not Addictive or is Never Harmful	1.0	The tweet states that nicotine is entirely safe or it is never harmful.	• A juul contains 4 ingredients which all are harmless to the body & proven so; Nicotine is ONLY addicting. Not harmful.. • The definition of addiction is must cause harm to self or others. Vapers and consumers of smokeless tobacco are simply NOT addicts
Nicotine has Unproven Health Benefits	1.0	The tweet states that there are benefits to nicotine use that are unproven.	• Studies on nicotine have found that it prevents you from developing Parkinson’s • Nicotine is an anti-depressant that you don’t need a prescription for
Pro-Vaping Exaggeration	0.72	The tweet exaggerates research findings or public health statements in a pro-vaping manner.	• Nicotine is only mildly habit forming on its own, it’s when combined with MAOIs and other chemicals in cigarettes that it becomes highly addictive • Vaping nicotine is not any more harmful to kids than coffee
Anti-Vaping Exaggeration	0.79	The tweet exaggerates research findings or public health statements in an anti-vaping manner.	• My doc told me that quitting nicotine is harder than quitting black tar heroin. • A juul pod has the same amount of nicotine as 2 packs of cigarettes
Without Clear Misinformation	0.77	The tweet states a fact that the coder cannot say for certain is misinformation OR the coder believes it to be true.	• All you kids juuling thinking you are cool are just getting nicotine addictions & its dumb asf. 1 pod = nicotine in a pack of cigs • Clinical trial finds e-cigarettes more effective than nicotine-replacement therapy with respect to the 1-year abstinence rate (18% vs. 10%)

Two experienced graduate-level coders were provided with the tweet text and a link to each tweet online. Tweet text was initially coded for relevance to ENDS, defined as mentioning ENDS products as its main topic (“the kids who called me stupid are now addicted to juuls…”). Tweets in which ENDS were mentioned but not as the main topic (“companies in 2018: how do we make this generation of phone addicted juuling teens go outside…”) or that were not related to ENDS (i.e., a tweet mentions “juul” but it is the name of a pet) were excluded. All relevant tweets that remained publicly available at the time of coding were viewed on Twitter.com so that links to external content could be assessed when possible. However, the text from unavailable tweets was still included in thematic analysis to preserve comprehensiveness of the original data.

Relevant tweets were coded for presentation of a statement about ENDS and nicotine or addiction information. A statement was coded as presenting information if the coder believed it could either be proven or disproven based on scientific evidence, consistent with previous research (Mitchell et al., 2018). Tweets coded as presenting information were then coded for the presence of misinformation in four categories: nicotine is not addictive or is never harmful, unproven health benefits, anti-vaping exaggeration, and pro-vaping exaggeration (Table 1). If the information in the tweet could not be disproven based on scientific literature, the tweet was coded as without clear misinformation. Coders were instructed to determine the presence of potential misinformation based on the current literature from peer-reviewed scientific journals or reputable governmental or health agencies, such as the Centers for Disease Control & Prevention, at the time of the tweet. For example, data were collected prior to the widespread reports of “e-cigarette, or vaping, product-use associated lung injury” (EVALI) (Vikram P. Krishnasamy et al., 2020), therefore, tweets claiming that ENDS use was not associated with acute lung injury were coded as without clear misinformation. Codes were not mutually exclusive. For example, a tweet that stated, “vaping is a one-way street to smoking because it is so highly addictive” would be coded as both anti-vaping exaggeration (not all individuals who use ENDS go on to smoke) and without clear misinformation (studies suggest that ENDS can be highly addictive). Additional codes assessed the type of account that authored or rebroadcasted the tweet. Consistent with prior research (Martinez et al., 2018), included categories were news (user bio identifies as a news outlet), commercial (user bio identifies as supplier, distributer, or marketer of e-cigarettes), group (user bio identified as affiliated with an organization or institution), and individual (user bio is not news, commercial, or group) accounts. Accounts that were suspended, deleted, or unclear were coded as unknown.

The iterative coding process involved double-coding 100 tweets by two independent, experienced coders, adjudication of disagreements with the lead author, followed by modifications of the codebook. Inter-rater reliability was assessed using Cohen’s κ (Cohen, 1960). After 6 rounds of this process, Cohen’s κ reached acceptable levels of reliability (Landis & Koch, 1977). For type of account, Cohen’s κ ranged 0.85–1.00; Cohen’s κ for tweet content are presented in Table 1. The two coders then independently coded the remaining tweets in the dataset.

Data Analysis

Frequencies and percentages were calculated for each code. A thematic qualitative content analysis approach was used to inductively assess the tweets coded as presenting information about ENDS-related nicotine or addiction for the presence of potential misinformation (Braun & Clarke, 2006). Coders then met with the lead author to develop and refine thematic units within codes. Qualitative themes and quotes around quantitative findings were organized to contextualize potential misinformation. Thematic analysis is a rigorous and systematic qualitative analysis approach that is meant to result in meaningful, useful, and trustworthy results (Nowell et al., 2017). Quotes were de-identified and unique quotes were slightly rephrased, while preserving the original meaning of the statement, to prevent identification of individual Twitter users (Colditz et al., 2018).

Bot detection

Among the 3,116 tweets included in the subsample for this study, there were 2695 unique users. These users were assessed for likelihood of being a bot using Botometer (Davis et al., 2016). Botometer uses over 1,000 features from the Twitter user profile to determine the probability that an account is completely automated (Allem, Escobedo, et al., 2019; Wojcik et al., 2018). Consistent with prior research, all user accounts with a bot score higher than 0.43 were classified as bots (Marlow et al., 2021; Wojcik et al., 2018). As we were particularly interested in identifying human user activity, this benchmark was seen as favorable to a standard 0.5 cutoff that has greater potential to misclassify bots as humans.

Results

Presentation of information as fact

Of the 3116 human-coded tweets, 2,943 (94.5%) were coded as relevant to ENDS. Of relevant tweets, 30.7% (n=904) presented ENDS-related nicotine or addiction information. Of tweets coded as presenting information, 41.7% (n=377) contained at least one piece of potential misinformation, meaning that they were coded as containing anti-vaping exaggeration, pro-vaping exaggeration, statements indicating that nicotine is not addictive or is never harmful, and/or statements indicating that nicotine has unproven health benefits.

Tweets without clear misinformation

Most tweets that presented at least one piece of information were coded as without clear misinformation (n=719, 79.6%, Table 2) because human coders were not able to find conclusive evidence to refute them, or because scientific evidence at the time was conflicting. Of these tweets, the majority originated from individual accounts (n=493, 68.6%) and group accounts (n=111, 15.4%). Overarching themes in this category included: (1) potential for nicotine to harm adolescent brain development, (2) nicotine can be as addictive as caffeine, (3) conflicting evidence on using ENDS as a cigarette smoking cessation aid, and (4) ENDS regulations domestically and abroad.

TABLE 2—

Presence of misinformation in nicotine and addiction-related tweets presented as fact.

Misinformation Category and Type of Account	Frequency n (%)
Without clear misinformation	719 (79.6)
Individual	493 (68.6)
Group	111 (15.4)
News	42 (5.8)
Commercial	20 (2.8)
Unknown	53 (7.4)
Anti-vaping exaggeration	195 (21.6)
Individual	153 (78.5)
Group	16 (8.2)
News	6 (3.1)
Commercial	1 (0.5)
Unknown	17 (8.7)
Pro-vaping exaggeration	150 (16.6)
Individual	112 (74.2)
Group	12 (8.0)
News	4 (2.6)
Commercial	7 (4.6)
News	4 (2.6)
Unknown	11 (7.9)
Nicotine is not addictive or is never harmful	23 (2.5)
Individual	19 (82.6)
Group	1 (4.3)
News	1 (4.3)
Commercial	0 (0.00)
Unknown	2 (8.7)
Nicotine has unproven health benefits	17 (1.9)
Individual	13 (76.5)
Group	1 (5.9)
News	1 (5.9)
Commercial	1 (5.9)
Unknown	1 (5.9)

^aPercentages total more than 100% due to non- exclusive categories

Misinformation theme: Anti-vaping exaggeration

Anti-vaping exaggeration was found in 21.6% (n=195) of tweets in which information was presented (Table 2). The majority of these tweets originated from individual accounts (n=153, 78.5%). Almost one in ten of these tweets (n=16, 8.2%) originated from group accounts, which were often schools (university and high schools), community education groups, and tobacco-free advocacy groups. Overarching themes in this category were: (1) statements that all ENDS contain nicotine (“your vape [has] more nicotine than my cig”), (2) exaggeration of nicotine contents of certain products (“1 juul pod = 2 packs of cigs”), (3) claims that everyone who uses nicotine will become addicted (“to all…little kids who started juuling…now know ur addicted to nicotine”), and (4) inappropriate translation of results from non-human studies to humans, such as one study conducted with pregnant rats that was reported as if conducted in humans (“Pregnant women who use e-cigarettes increase the risk of their newborns becoming cot-death victims”).

Misinformation theme: Pro-vaping exaggeration

Pro-vaping exaggeration was found in 16.6% (n=150) of tweets in which information was presented (Table 2). The majority of these tweets originated from individual accounts (n=112, 74.2%). Almost one in ten of these tweets (n=12, 8.0%) originated from group accounts, which were generally pro-vaping advocacy organizations. Overarching themes in this category were: (1) statements that the concern about youth ENDS use is exaggerated (“Kids that are using Juul are technically “tobacco-free kids” and thankfully aren’t smoking”), (2) statements that suggest that ENDS are an extremely effective tool for cigarette cessation, which is still unproven (“vaping to stop tobacco smoking is as effective as methadone or Suboxone is to a heroin addiction”), (3) claims that ENDS use is not addictive or harmful (“we’ve been ... engineered to believe nicotine is [an] addictive dangerous drug by Pharma & Public Health”), and (4) false information on the role of monoamine oxidase inhibitors on the addictive properties of nicotine (“Nicotine is only mildly habit forming on its own, it’s when combined with MAOIs and other chemicals in cigarettes that it becomes highly addictive”).

Misinformation themes: Nicotine is not addictive or is never harmful OR Unproven health benefits

There were two other major misinformation themes noted, although less often than the others. Nicotine is not addictive or is never harmful was found in 2.5% (n=23) and statements about unproven health benefits were found in 1.9% (n=17) of tweets that presented information (Table 2). The majority of these tweets originated from individual accounts for both nicotine is not addictive or is never harmful (n=19, 82.6%) and unproven health benefits (n=13, 76.5%). Overarching themes in these categories were: (1) nicotine prevents certain illnesses, such as Parkinson’s disease, attention deficit hyperactivity disorder, and inflammatory disease (“... I dont smoke cigarettes but I started smoking out of a vape cus my ADD gets bad…”), (2) nicotine delivers cognitive benefits (“Nicotine ...increases brain functionality so I’m gonna do it more”), (3) vaping produces a vapor and therefore is not harmful (“Smoke produce harmful chemicals and tar from the burning process. But vapor is just vapor, like water vapor, but with nicotine instead…”), and (4) nicotine is only addictive when paired with burning tobacco (“…CDC and NIH have both found nicotine to be nonaddictive when unpaired from burning cigs...”).

Presence of bots

Of the 2695 unique users, 107 (3.97%) were classified as bots. The Botometer API could not determine the probabilities for 504 (18.7%) accounts due to profiles being private, suspended, deleted, and/or lacking sufficient user content for analysis. The 2084 (77.33%) remaining accounts were classified as actual users. Of the users classified as bots, only 10 (9.3%) were coded as sharing misinformation.

Discussion

In this study of ENDS-related messages on Twitter, approximately 40% of tweets that presented informational statements about nicotine or addiction contained potential misinformation in one of four categories. This misinformation was posted by both individuals and groups. Accounts posting anti-vaping exaggeration were most often individuals, while accounts posting pro-vaping exaggeration were almost exclusively pro-vaping groups.

The two categories of potential misinformation most often identified were anti-vaping exaggeration and pro-vaping exaggeration. In the anti-vaping exaggeration category, most tweets presented true public health- related statements about ENDS nicotine and addiction that were embellished or distorted. For example, a common theme centered around the addictive potential of nicotine, with many tweets containing a variation of “everyone who uses ENDS products gets addicted.” Although nicotine is a highly addictive substance (Benowitz, 2010), not all individuals who use nicotine become addicted. Groups posting anti-vaping exaggeration tended to be schools, community education groups, and tobacco-free advocacy groups. It may be that anti-vaping exaggeration posted by educational and public health-related groups may be unintentional. However, messages such as these play into pro-vaping messaging that ENDS products are intentionally distorted by public health entities and then used to regulate ENDS (Allem et al., 2016; Annechino & Antin, 2016; Harris et al., 2014). Thus, it is imperative that any statements about the nicotine content and addictive potential of ENDS be fact-checked carefully.

In the pro-vaping exaggeration category, many tweets misinterpreted results from scientific studies. For example, many tweets claimed that reports of youth ENDS rates are intentionally exaggerated. In some instances, the tweets contained accusations that ENDS education and regulation campaigns are actually pro-“big tobacco” and will encourage youth to initiate cigarette smoking. Other tweets in this category asserted that, because tobacco smoke contains monoamine oxidase inhibitors and monoamines are known to be involved in neurological reward pathways, nicotine is only addictive when combined with tobacco smoke. These specific messages can be appropriately countered by public health professionals in informational and attitude-based campaigns. Increasing Twitter users’ media literacy may help “inoculate” casual observers against taking this type of misinformation at face value. For example, a core principle of media literacy involves carefully considering the source of the message, which may help youth to carefully scrutinize the financial, political, and other motivations of message creators (Vahedi et al., 2018). Prior research suggests media literacy education for youth centered around cigarette advertising is feasible, acceptable, and may reduce intent to use (Shensa et al., 2015). Future work could expand these programs to include ENDS as well as the application of media literacy to information on social media.

In the anti-vaping exaggeration category, most individuals did not identify themselves as interested in ENDS prevention or public health. However, their statements indicate that they may be misinterpreting the public health ENDS-related messaging to which they are exposed. Conversely, in the pro-vaping category, most individuals identified themselves as pro-vaping and/or former cigarette smokers who had used ENDS to quit in their Twitter biography. Many of these individuals presented information that was misinterpreted from scientific research. Therefore, the non-scientific community may be struggling with understanding both the public health messaging and scientific literature regarding ENDS, nicotine, and addiction. It may be valuable for clinicians and researchers to focus both on strategies to increase scientific literacy in lay populations as well as disseminating scientific information in ways that are more easily understood by the general public. With regard to the latter, training clinicians and researchers in best practices for risk communication may be a useful tool to improve the ability of scientists to communicate about their findings (Covello, 2003).

There were some instances of misinformation in its claim that nicotine is not addictive or is never harmful, which can be problematic. For example, some misinformation indicated that nicotine is only addictive when used in a combustible tobacco product such as cigarettes and not in ENDS products. This is misleading and may convince an otherwise nicotine-naïve individual to initiate use of ENDS products. Although this message was not commonly noted in this study, even scant misinformation such as this can rapidly propagate through social networks and grow into a substantial source of information (Y. Wang et al., 2019).

Limitations

This study had four important limitations. First, Twitter is not representative of the general population. However, approximately one-third of teens and young adults use Twitter. Because this is the same population that heavily uses ENDS (Cullen et al., 2019), analysis of Twitter data can offer insights into rapidly changing health topics that may be difficult to detect in more intensive population studies. Second, interpretation of posts using qualitative analysis can be inherently subjective. However, we employed systematic, rigorous qualitative methods to improve reproducibility of results (Nowell et al., 2017). Third, we coded tweets as without clear misinformation if the information contained in the tweet could not be disproven based on scientific literature at the time the study was conducted. It is possible that the information contained in these tweets have been either proven or disproven since then. Finally, the prevalence of bots in the study dataset was smaller than in other studies (Wojcik et al., 2018), though consistent with other topic-specific datasets (Allem et al., 2017). This may be due to filtering of the dataset to include only nicotine and addiction-related tweets and is consistent with the prevalence of social bots in such topic-specific datasets. This could also be because bot accounts were deactivated or removed from the Twitter platform between data retrieval and the application of Botometer.

Conclusions

The scientific literature around ENDS is rapidly changing. At the time of data collection, a prominent message from the public health community asserted that ENDS were 95% less harmful than combustible cigarettes (McNeill et al., 2015). However, in light of increasing evidence of risks to safety and health, scholars have emphasized the need for health professionals to more clearly articulate the lack of scientific evidence supporting this claim (Eissenberg et al., 2020). The rapidly changing evidence make it difficult for individuals to interpret ENDS-related information. A centralized readily accessible source of up-to-date ENDS information that can be understood by the lay public would be beneficial. Likewise, in a digital age in which information is readily accessible and propagated, ENDS-related public health campaigns are at risk of being delivered out of context or used to develop counter campaigns (Allem et al., 2016). Public health agencies and tobacco control entities should strive to deliver ENDS-related prevention and education campaigns that are easily understood by the lay public. When delivered on social media, these messages should be monitored to detect the spread of misinterpretation or potential misinformation.