Article Text

Download PDFPDF

Testing messages about comparative risk of electronic cigarettes and combusted cigarettes
  1. Bo Yang,
  2. Daniel Owusu,
  3. Lucy Popova
  1. Tobacco Center of Regulatory Science, School of Public Health, Georgia State University, Atlanta, Georgia, USA
  1. Correspondence to Dr. Lucy Popova, Tobacco Center of Regulatory Science, School of Public Health, Georgia State University, Atlanta GA30302, USA; lpopova1{at}gsu.edu

Abstract

Introduction Health agencies are grappling with communicating risks of electronic cigarettes (e-cigarettes) compared with combusted cigarettes. This study examined smokers’ responses to two types of comparative risk messages with one type incorporating more negative antismoking elements in the design.

Methods In an online experiment, 1400 US adult (18+ years) current smokers and recent quitters were randomised to view one of three comparative risk messages about e-cigarettes (CR messages), one of three comparative risk messages that included more negative antismoking elements in the design (CR− messages) or a control message. Selection of outcomes was guided by the antismoking message impact framework. Multivariate analyses of covariance and logistic regression models analysed effects of messages on message evaluations, e-cigarette-related and cigarette-related beliefs and behavioural intentions.

Results Both CR and CR− messages decreased smokers’ intentions to smoke cigarettes, increased intentions to switch to e-cigarettes completely and increased perceptions that e-cigarettes are less harmful than combusted cigarettes. Neither message type increased dual use intentions relative to exclusive e-cigarettes use or smoking cessation. CR messages decreased perceived absolute risks of e-cigarettes and self-exempting beliefs about smoking, whereas CR− versus CR messages produced higher self-efficacy to quit smoking.

Conclusion Comparative risk communication might encourage smokers to switch to lower-harm tobacco products. Comparative risk messages with more negative antismoking elements in the design might be particularly effective, because they led to higher self-efficacy to quit smoking. Regulatory agencies may consider using comparative risk messages with more negative antismoking elements to educate the public about lower risk of e-cigarettes.

  • harm reduction
  • electronic nicotine delivery devices
  • advertising and promotion

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Tobacco smoking carries significant economic and health burdens.1 Some argue that alternative tobacco products such as electronic cigarettes (e-cigarettes) that effectively deliver nicotine with lower levels of harmful chemicals than combusted tobacco may reduce the overall harm of tobacco use by attracting smokers not willing to quit.2–4

Some countries began communicating about lower risk of e-cigarettes. In the UK, high nicotine e-cigarettes can be licensed as medicinal products and make health claims.5 ‘Tobacco Control Plan for England’ discusses e-cigarettes as one of the strategies to reach a ‘smoke-free generation’.4 Canada recently passed a bill to regulate e-cigarettes to achieve both smoking prevention and harm reduction.6 The US Food and Drug Administration (FDA) has been urged to communicate to the public about relative risks of different tobacco products.2 3 7 The FDA can also authorise marketing of products with modified risk claims—statements that a certain tobacco product is less harmful than other tobacco products currently on the market.8

The challenge with comparative risk communication (including modified risk claims) is to develop messages that would motivate smokers who are unwilling to quit smoking to switch completely to less harmful products, but would not result in unintended consequences, such as lower cessation and increased dual use among smokers, initiation among non-users or relapse in former smokers. Few studies outside tobacco industry examined communication strategies for comparative risks of novel tobacco products,9–14 particularly e-cigarettes, and more research is needed.

Greater risk perceptions make people less likely to engage in harmful behaviour (eg, smoking).15 Past studies predominantly focused on perceived risk of a single product, such as cigarettes or e-cigarettes.16–18 Communicating comparative risks of two products, such as electronic and combusted cigarettes, raises additional issues. Since public health education resulted in engrained opinions about risks of cigarettes, comparative risk messages may only influence e-cigarette risk perceptions. Alternatively, telling people that one risky behaviour is better than another might reduce perceived risk of both behaviours. To our knowledge, no study examined how claims comparing risks of e-cigarettes and cigarettes influence the perceived risks of each product.

Consistent with theories on risk perception, qualitative and quantitative research,19 20 longitudinal and cross-sectional observations21 22 and research conducted among youth and adults23 24 found that lower perceived risk of e-cigarettes as compared with cigarettes was associated with greater likelihood of e-cigarette initiation. Also, former and current smokers named lower harm of e-cigarettes as one of the most important reasons for e-cigarette use.20 25 Hence, for e-cigarettes to be promoted as a harm reduction tool, smokers should be able to recognise that e-cigarettes present lower risks than cigarettes.

Most people perceive e-cigarettes as less harmful than combusted cigarettes,26 27 particularly the young, highly educated and ever e-cigarette users.28 29 However, beliefs that e-cigarettes are less harmful have been declining.27 28 This might stem from the changing tobacco regulatory environment or negative media coverage of e-cigarettes.30–32

The few studies9 12 33–37 on comparative risk messages about e-cigarettes and cigarettes typically used an explicit statement that e-cigarettes are less harmful than cigarettes, similar to e-cigarette advertisements and tobacco companies’ modified risk statements (eg, “Switching completely from conventional cigarettes to the iQOS system can reduce the risks of tobacco-related diseases”38). However, it is unclear whether this way of communicating comparative risk about e-cigarettes is the most effective. Research on different approaches to communicating lower risk of e-cigarettes on risk perceptions and intentions surrounding use of e-cigarettes and cigarettes is lacking. To our knowledge, only one study examined the effect of different comparative risk messages on smokers’ dual use interest, finding that messages emphasising cigarettes had more harmful chemicals than e-cigarettes (10 times more, 100 times more; and harmful chemicals only contained in cigarettes) evoked higher dual use interest than the message that e-cigarettes and cigarettes had same amount of harmful chemicals.9

To evaluate different approaches to communicating comparative risk of e-cigarettes and cigarettes, we developed and tested two types of comparative risk messages based on two knowledge frameworks. One type was developed from the knowledge about comparative risk messages from prior studies9 12 33–37 where comparative risk messages were similar to e-cigarette advertisements and tobacco companies’ modified risk claims. These messages emphasised the benefits of e-cigarettes compared with cigarettes (labelled ‘CR messages’). The other type was developed from the knowledge about effective antismoking messages. Prior studies have identified several elements that enhance antismoking communication effectiveness,18 39–42 such as a strong smoking risk message and graphic visuals emphasising severe health problems caused by smoking. We incorporated these elements to create messages that emphasised the harms of continued smoking while presenting e-cigarettes as a safer alternative (labelled ‘CR− messages’). Given that comparative risk messages might have unintended effects, such as delayed cessation or increased dual use, we evaluated whether comparative risk messages developed from the effective antismoking message framework that included many antismoking elements in the design would have smaller unintended effects than the messages tested in prior studies and that were similar to e-cigarette advertisement appeals. We assessed message effects on multiple outcomes, guided by the antismoking message impact framework.43

Methods

Materials

We combined a top-down deductive and an empirical inductive approach to develop the messages. In collaboration with a research company, John Snow, Inc., we reviewed existing e-cigarette messages from public health agencies (such as California’s Still Blowing Smoke campaign44) and the extant literature on e-cigarette health effects and comparative risks. This review combined with the team’s expertise in communication, public health and tobacco control informed the development of 12 initial message concepts, executed as full-colour pictures and text. These initial message concepts were tested with 12 focus groups comprising 72 adult (18+ years) current or recently former (quit in the past 2 years) smokers in Atlanta, Georgia. Each group viewed 8 out of 12 messages. We evaluated which message concepts participants liked and perceived as effective and which messages generated the most engaged discussion that were relevant to e-cigarettes and dual use. Of the 12 initial concepts, six were selected and further revised into the final six messages (online supplementary appendix).

In developing the messages, we aimed to evaluate two different ways to deliver comparative risk information about e-cigarettes. One (CR messages) followed the way messages in prior comparative risk communication studies, most e-cigarette advertisements and tobacco companies’ modified risk claims have been constructed. The other (CR− messages) capitalises on knowledge of what makes an effective antismoking appeal. The two ways of delivering comparative risk messages are different and incorporate different visual and textual elements. As a result, the two types of messages were different not only in inclusion of antismoking appeal but also in other features.

All six messages stated that if smokers are not ready to quit, switching to e-cigarettes completely can reduce their health risks (online supplementary appendix). Since comparative risk information likely appeals to positive emotions (eg, hope), based on focus group discussion, more positive images and lighter colours were used in the three CR messages. In contrast, the three CR− messages were designed to arouse negative thoughts about smoking and motivate smoking cessation. CR− messages included a strong smoking risk message and used darker colours and graphic images. As CR messages emphasised that switching to e-cigarettes can reduce health risks and used more positive images and CR− messages highlighted dangers of smoking and used threat-based images, respectively, CR and CR− messages resemble gain-framed (emphasising benefits of staying away from smoking) and loss-framed (emphasising costs of smoking) antismoking messages. Loss-framed antismoking messages were found to be more effective in encouraging smoking cessation than gain-framed ones.39 45 46 For instance, current smokers rated loss-framed graphic warning labels more effective in motivating people to stay away from smoking than gain-framed graphic warning labels.46 Based on prior studies, CR− messages might have greater intended effects of comparative risk communication. The three CR messages and three CR− messages were paired up by three themes: taking control, e-cigarettes having lower amount of toxic chemicals and smoking effects on lifespan.

Participants

Participants were 1400 adult (18+ years) current smokers (smoked at least 100 cigarettes in their life and were currently smoking cigarettes every day or some days) or recent former smokers (quit smoking within the past 2 years). Toluna, a survey market research company, recruited participants through different online recruitment strategies (eg, web banners, website referrals, affiliate marketing and pay-per-click). All participants completed electronic informed consent. All protocols were approved by the Georgia State University Institutional Review Board.

Procedure

We conducted a pilot test with 100 participants recruited by Toluna to examine feasibility and refine the questionnaire. In the main study, participants began by answering questions about demographics, past tobacco use, efficacy to quit smoking, perceived comparative risk of e-cigarettes and combusted cigarettes and smoking identity. Participants were then randomised to view one of the six messages (groups 1–6) or a control message (a bottled water advertisement; group 7). Randomisation was stratified on the current versus former smoker status and on intentions to quit smoking (never planning to quit; planning to quit in the future, but not in the next 6 months and planning to quit in the next 6 months or sooner). Participants viewed the message without time constraint. After seeing the message, participants were asked questions about their emotional responses and perception of the messages. Then, their risk perceptions and use intentions about e-cigarettes and cigarettes were assessed, followed by other belief questions (eg, self-efficacy in quitting smoking).

At the end, all participants saw a debriefing page stating that the messages were used for research only, and that stopping smoking completely is the best thing they can do for health. All participants were given a quit-line telephone number and referred to smoking cessation websites.

Outcome measures

The selection of the outcome measures was based on the antismoking message impact framework43 and included:

  1. message reactions and perceived effectiveness (negative emotions,47 48 positive emotions,49 50 reactance,51 message novelty and perceived message effectiveness in arousing health concerns52 and in motivating switching to e-cigarettes),

  2. e-cigarette-related and cigarette-related beliefs (perceived risks and benefits of e-cigarettes and cigarettes,53 perceived comparative risk of e-cigarettes to cigarettes, response efficacy,54 self-efficacy,54 self-exempting beliefs about smoking55 and support for tobacco control56),

  3. behavioural intentions (intentions to smoke cigarettes, intentions to switch to e-cigarettes,14 dual use intentions and for current smokers only: intentions to quit57 or to engage in various cessation-related behaviours18).

Detailed measures are shown in table 1. Variables measured by multiple items were averaged to create composite scores.

Table 1

Key measures

Covariates

Covariates included demographics (sex, age, race and education level), perceived comparative risk of e-cigarettes and cigarettes, response efficacy and self-efficacy at pretest, daily cigarette use (yes vs no), e-cigarettes use (never vs ever but not current vs current), last year quit attempt (yes vs no) and smoking identity46 (table 1). Current e-cigarettes use was defined as past 30-day use.

Analysis plan

We first examined whether the effects of message type varied across message theme.58 Univariate analyses of variances did not identify significant interaction effects between message type and message theme on the outcome variables. Based on prior practice,59 60 we collapsed three messages within each type of comparative risk messages.

We performed multivariate analyses of covariance (MANCOVAs) on each block of outcome variables with experimental condition as the independent variable. For the MANCOVA on message reactions and perceived effectiveness, we excluded the control group because reactions to the control message (water bottle advertisement) were not relevant. For other MANCOVAs, we included the control condition. Significant multivariate effects of message condition were followed by univariate analyses of the condition effect on each dependent variable.

We performed binomial logistic regression for perceived comparative risk of e-cigarettes and cigarettes (comparing e-cigarettes being ‘less harmful’ to all other answers—‘about the same’, ‘more harmful’ and ‘don’t know’, which represented potentially incorrect beliefs or lack of knowledge). Multinomial logistic regression examined whether exclusive e-cigarette use and cessation (more desirable outcomes) versus dual use (undesirable outcome) differed across conditions. For both binomial and multinomial logistic regressions, we created two message dummy variables using orthogonal coding. One variable compared CR and CR− with control. The other dummy variable compared CR with CR− messages.

All analyses used the same covariates detailed previously. Significance level was p<0.05. Multiple pairwise comparisons involving three groups employed Bonferroni procedure. We used SPSS V.24.

Results

The sample was 53% female, 81.6% White, 63.7% college graduates, 61% daily smokers, 9.4% former smokers, 33.6% current e-cigarette users and 31.3% current dual users averagely aged 47 years (table 2).

Table 2

Participant characteristics overall and by experimental condition

Message reactions and perceived effectiveness

Condition was significant at the multivariate level, F(6, 1183)=52.93, p<0.001, Wilks’ λ=0.79. Follow-up univariate analyses found that CR messages evoked less negative emotions, more positive emotions, were rated more novel and more effective in motivating switching to e-cigarettes, but less effective in increasing health concerns than CR− messages. Reactance was low and not significantly different between conditions (table 3).

Table 3

Means, standard errors and univariate F statistics of each treatment condition on message reactions and perceived effectiveness

E-cigarette-related and cigarette-related beliefs

Condition was significant at the multivariate level, F(16, 2418)=2.23, p<0.01, Wilks’ λ=0.97. Follow-up univariate analyses found that CR messages reduced self-exempting beliefs and perceived risks of e-cigarettes compared with other conditions. CR− messages led to greater self-efficacy compared with CR messages. Differences in perceived benefits of cigarettes and e-cigarettes and perceived cigarette risks were not significant (table 4).

Table 4

Means, standard errors and univariate F statistics of each message condition on e-cigarette-related and cigarette-related beliefs

Perceived comparative risk of e-cigarettes and cigarettes

Compared with control, CR and CR− messages increased perceptions that e-cigarettes are less harmful than cigarettes (OR 1.29, 95% CI 1.12 to 1.48). However, CR and CR− messages did not differ in comparative risk perceptions (table 5).

Table 5

Logistic regressions predicting perceived comparative risk of e-cigarettes to cigarettes at post-test and dual use intentions

Dual use intentions

Compared with control, CR and CR− messages did not significantly increase dual use intentions relative to exclusive e-cigarette use or smoking cessation. In addition, CR and CR− messages did not differ in dual use intentions (table 5).

Smoking and switching intentions

Condition was significant at the multivariate level, F(4, 2768)=5.32, p<0.001, Wilks’ λ=0.99. Compared with control, both CR and CR− messages reduced intentions to smoke cigarettes and increased intentions to switch to e-cigarettes. CR and CR− messages did not differ from each other for both intentions (table 6).

Table 6

Means, standard errors and univariate F statistics of each message condition on behavioural intentions

Current smokers’ intentions to quit

Message condition was not significant at the multivariate level, F(8, 2500)=1.21, p=0.29, Wilks’ λ=0.99 or at the univariate level (table 6).

Discussion

Despite the increasing calls to inform the public of comparative risks of various tobacco products,2 few studies examined comparative risk communication strategies.9–14 We developed and tested two types of comparative risk messages: one emphasised the benefits of switching to e-cigarettes (CR messages) and the other incorporated more negative antismoking elements, such as graphic images highlighting dangers of smoking (CR− messages). Compared with the control, both types of messages decreased intentions to smoke, increased intentions to completely switch to e-cigarettes and made smokers more likely to report that e-cigarettes are less harmful than combusted cigarettes. CR messages reduced e-cigarette risk perceptions and self-exempting beliefs about smoking. CR− messages led to higher self-efficacy to quit smoking than the CR messages. While the difference in self-efficacy was small (Cohen’s d=0.10 or r=0.05, a small effect size), it is comparable with other effect sizes in communication research.61 For example, in media campaigns for smoking prevention and cessation, an average effect size is r=0.03.62 However, these small effects are important considering a large number of people who would view the messages and be influenced by them.63

Compared with CR messages, CR− messages evoked more negative emotions, less positive emotions, were perceived as less novel and less likely to encourage complete switching to e-cigarettes but more likely to arouse health concerns in smokers. Hence, CR and CR− messages might influence behaviour through different mechanisms.

This study begins to answer some important questions about comparative risk communication around tobacco products and identifies areas of future research:

  1. Can comparative risk communication have a positive impact on consumers? We did not find unintended effects of comparative risk messages and found some positive effects on comparative risk perceptions and smoking and switching intentions; thus, the answer is a tentative yes. However, our results differed from Pepper et al’s finding9 where smokers exposed to comparative risk claims about e-cigarettes reported more dual use interest. One explanation for the discrepancy might be the use of different measures to assess dual use tendencies. We assessed dual use intentions in a direct way. Participants picked one most likely future action among smoking cessation, exclusive e-cigarette or cigarette use or dual use. In contrast, Pepper et al assessed dual use interest, with a measure similar to perceived message effectiveness (participants reported whether the message would make them do a particular thing). They assessed dual use interest indirectly by asking smokers’ interest in using e-cigarettes and cigarettes separately. Participants who indicated intentions to increase, decrease or continue smoking and also to initiate or increase e-cigarettes use were classified as having dual use interest. Indirect and direct measures might have led to the observed difference, similar to how indirect and direct comparative risk measures resulted in different findings in other studies.64–66 In addition, different control messages (bottled water advertisement in our study vs message that e-cigarettes and cigarettes had the same amount of harmful chemicals in Pepper et al) might explain the difference. Additional research is needed to evaluate effects of comparative risk communication on dual use intentions.

  2. Which way of communicating about comparative risks would have the most positive effect? CR and CR− messages generally produced similar effects (or lack thereof) with three exceptions: CR messages lowered perceived risks of e-cigarettes and reduced self-exempting beliefs about smoking, whereas CR− messages increased smokers’ self-efficacy to quit smoking. Meta-analyses67–70 of fear appeal studies found that perceived efficacy has a greater impact on positive outcomes (such as intentions to quit smoking) than perceived risk (or threat). Furthermore, higher self-efficacy facilitates smoking cessation.71 Thus, we recommend CR− messages. However, other types of messages need to be explored. For example, few studies directly compared messages about absolute and comparative risk of different tobacco products.72 It is possible that the same intentions to switch to e-cigarettes might be more effectively motivated by traditional messages communicating absolute risk of tobacco products, such as the antismoking messages from the Tips from Former Smokers campaign.73

  3. How do comparative risk communications affect risk perceptions? Both CR and CR− messages produced greater perceptions that e-cigarettes are less harmful than cigarettes and reduced absolute perceptions of e-cigarette risks, although the difference was only significant for CR messages. CR− messages did not significantly reduce perceived risk of e-cigarettes possibly because CR− messages emphasised harms of cigarettes, making tobacco product risk perceptions salient. For instance, seeing messages about smoking risks increased perceived risks of e-cigarettes in young adults.74 Both CR and CR− messages did not change perceived cigarette risks, potentially due to a uniformly high awareness of cigarette risks. Messages on comparative risk of two products with lower perceived harms (eg, different alcohol products) might find different results.

Limitations include participants viewing messages without distraction, possibly amplifying our effect sizes. Our non-probability sample contained more whites and higher educated participants than the US census figures, limiting the generalisability of our findings. Only adult smokers were included; future studies should include non-smokers, adolescent smokers and long-term former smokers. We analysed self-reported data and did not measure actual behaviours. Lacking a validated measure of dual use intentions, we created our own measure, which might explain the lack of differences among conditions on the measure. Using measures from other studies resulted in varying referents (self vs others vs no referent); consistently treating respondents as referents might provide more useful information.75 While measuring people’s efficacy beliefs and comparative risk perceptions before exposure and controlling for them increased statistical power, estimates’ precision and internal validity,76 pretest might have interacted with the messages to produce outcomes not generalisable to contexts without pretest. Finally, we compared two formats of comparative risk messages: one based on the knowledge about effective antismoking communication and one based on the knowledge about e-cigarette advertisements, modified risk claims and prior comparative risk communication studies. Owing to the differences in the two knowledge frameworks, there were many differences across the two types of messages (eg, different health risks highlighted, images used, size of image, portrayal of persons, font sizes and colour), which might be responsible for the differences we found. Future studies should evaluate individual antismoking message features to uncover the mechanisms of effective comparative risk communication.

Strengths of our study include comparing two types of comparative risk messages and assessing multiple messages in each message type, which reduces case-category confounding77 and increases ecological validity. Evaluating message effects on multiple outcomes provided a comprehensive understanding of how comparative risk messages influence smokers.

In conclusion, comparative risk communication might motivate smokers to switch to reduced risk products. Comparative risk messages with many negative antismoking elements in the design may be particularly effective by increasing smokers’ self-efficacy. While further research is needed, our results indicate that public health agencies may consider comparative risk messages with negative antismoking elements to increase population-level benefits.

What this paper adds

  • Messages that frame electronic cigarettes (e-cigarettes) as a less harmful option to combusted cigarettes could be beneficial if smokers who are unwilling to quit switch to e-cigarettes completely; however, comparative risk communication about e-cigarettes might have unintended effects on smokers, such as delayed smoking cessation or increased dual use.

  • This study evaluated two types of comparative risk messages with one type of messages incorporating more negative elements in the design.

  • Regardless of message type, we found some positive effects and no evidence of unintended effects of messages communicating lower risk of e-cigarettes than cigarettes. Furthermore, comparative risk messages incorporating more negative antismoking elements in the design increased smokers’ self-efficacy to quit smoking—a critical factor contributing to the success of smoking cessation.

References

Footnotes

  • Contributors LP and DO planned the study with contributions from BY. BY analysed the data, wrote the first draft and submitted the manuscript. LP conceived the study and supervised the data collection, analysis, interpretation and writing. All authors contributed to the writing, revision and approval of the final version.

  • Funding Research reported in this publication was supported by the National Institute of Drug Abuse of the National Institutes of Health and Food and Drug Administration Center for Tobacco Products (P50DA036128) and the National Cancer Institute of the National Institutes of Health and Food and Drug Administration Center for Tobacco Products (R00CA187460).

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval Georgia State University IRB.

  • Provenance and peer review Not commissioned; externally peer reviewed.