Associations of attitudes towards electronic cigarettes with advertisement exposure and social determinants: a cross sectional study

Demographic variables included age, gender, race, family income, respondent and parental tobacco smoking status and respondent lifetime e-cigarette use. Respondent smoking status was determined using the Brief Risk Factor Surveillance System (BRFSS) questionnaire criteria from the CDC (2009) [13]. Maternal smoking status was assessed with the question, “How would you classify your mother in terms of smoking cigarettes?”. Possible responses included “My mother is a regular smoker”, “My mother used to be a regular smoker, but has since quit”, and “My mother never smoked cigarettes regularly or at all”. A separate question was asked regarding father’s smoking status. Lifetime e-cigarette use was categorized into “ever used” and “never used” using the question “Have you ever used an electronic cigarette (“e-cigarette”)?”

Perceptions of harm of various substances were assessed by asking participants to rate on an eight point Likert scale (1 = very safe and 8 = very dangerous) “How dangerous or safe do you think each of the following drugs or products are?” Respondents were asked to rate perceived harm of flavored e-cigarettes, non-flavored e-cigarettes, tobacco cigarettes, alcohol, and marijuana. Similarly, perceptions of addictiveness were assessed for the same substances using a Likert scale ranging from 1 (not addictive) to 8 (very addictive).

Attitudes regarding acceptability of e-cigarette use in locations where tobacco smoking is banned was assessed by asking, “Do you believe it is OK for someone to use an e-cigarette where regular tobacco smoking is banned?” This question was asked for the following locations: in a bar, at work, in a dormitory room, in a restaurant or store, and in a classroom. Possible responses included “Yes”, “No”, and “Don’t know/not sure”.

Exposure to e-cigarette advertising was assessed by asking respondents to indicate where they had seen e-cigarette advertisements. Respondents were asked to check all applicable responses, which included “On the side of an automobile, bus or other vehicle”, “Billboard”, “Event (e.g. fair, concert, sporting event)”, “Inside a retail store”, “the Internet”, “Magazine or newspaper”, and “Television or movie”.

Data were analyzed using SAS 9.3 [13]. Respondents who indicated their mother was a former or current smoker were reclassified as having a positive smoking status for their mother. Prior evidence demonstrates maternal smoking status is a stronger predictor of tobacco behavior than paternal smoking status [14]. Similarly, in our analysis maternal smoking status was a much stronger predictor of outcomes than paternal smoking status, therefore the remainder of the analyses focus on maternal smoking status. Race was dichotomized into “white” and “non-white”. Due to the low response rate and potential for skewed data due to non-response bias, the data was weighted using population-based weighting [15]. Because limited information was known about both the population and survey respondents, weighting classes were created using the available data which included gender, race, and graduate/undergraduate status. Weights for each class were created using the inverse probability of response for each group. Bivariate analyses were performed to examine associations between participant and outcome variables. Analyses included two-way tests for perceptions of harm and addictiveness and Chi-squared test of association for attitudes of acceptability of e-cigarette use. Because little difference in results existed between flavored and unflavored e-cigarettes all analyses were conducted using responses for flavored e-cigarettes. To correct for multiple comparisons among bivariate analyses, Bonferonni corrections were applied.

Multinomial logistic analysis was conducted using a cumulative logit model for the ordinal variables harm and addictiveness. For acceptability models a dichotomous logistic regression was performed with “no” as the reference, and “I don’t know” excluded. Dichotomous predictors included: gender, white or non-white, e-cigarette use, ever been a smoker, mother ever been a smoker, advertisement seen on the Internet, advertisement seen on TV or movie, and advertisement seen in a magazine or newspaper.

The most common place students had seen advertisements for e-cigarettes was “Television or movie” (53.2%), followed by the Internet (42.0%), at a retail store (35.7%), and in a magazine or newspaper (34.9%) (Table 1).

More respondents thought it was acceptable (47.1%) versus not acceptable (43.7%) to use an e-cigarette in a bar where tobacco cigarettes are banned, while for all other locations more participants responded it was not acceptable rather than acceptable to use e-cigarettes (Table 2). Nearly 70% of those surveyed indicated that it was not acceptable to use an e-cigarette in class, making it the most unacceptable location to respondents.

Tobacco Cigarettes were perceived as much more harmful (t = 71.5 p < 0.0001) and addictive than e-cigarettes (t = 54.9 p < 0.0001) (Table 3). E-cigarettes were perceived as more harmful (t = 9.5 p < 0.0001) and addictive (t = 24 p < 0.0001) than marijuana, and less harmful (t = 9.6 p < 0.0001) but more addictive (t = 4.3 p < 0.0001) than alcohol.

Being male, an e-cigarette user, a current tobacco smoker, and having a mother who smoked tobacco cigarettes were all associated with having lower perceived harm and addictiveness of e-cigarettes. Race (white) was associated only with perceived harm (Table 4). All background measures were significantly related with perception of acceptability of use in bars, stores, at work, in class, in dorm (data displayed only for in bars).

Each advertising source, except advertisement seen on vehicles, was significantly associated with lower perceived harm. Having seen an advertisement on the Internet, on television, in a retail store, or at an event (e.g., fair, concert, sporting event) were significantly associated with lower perceived addictiveness of e-cigarettes. All forms of advertising exposure were significantly related with perception of acceptability of use in bars, stores, at work, in class, in dorm (data displayed only for in bars).

Being male (aOR 1.7), white (aOR 1.3), ever having used an e-cigarette (aOR 3.2), ever having been a tobacco smoker (aOR 1.8), having a mother who smoked (aOR 1.2), and having seen an advertisement on the Internet (aOR 1.2) all remained significantly associated with lower perceived harm of e-cigarette use (Table 5). Only being male (aOR 1.6) and lifetime e-cigarette use (aOR 2.8) were associated with significantly lower perceived addictiveness of e-cigarettes (Table 5).

All factors, except being white and having seen an advertisement in a magazine, were significantly associated with the likelihood of believing e-cigarette use in a bar was acceptable (Table 6). The factors that were significant for believing it was acceptable to use an e-cigarette in a bar were also significant predictors of acceptability of use in other locations with only one exception. Race was a significant predictor of perceiving e-cigarette use in class acceptable, with Whites 1.4 times more likely to perceive it as acceptable than Non-whites.

Our findings on advertising exposure suggest that the e-cigarette industry’s targeted advertising towards young people has been successful; more than half our respondents, who are all young adults, have seen an advertisement for e-cigarettes on TV or at the movies. Our findings also demonstrate that having seen ads on the Internet, but not on TV or in magazines, is a predictor of lower perceived harm of e-cigarettes. Current restrictions prevent e-cigarette companies from making health claims in TV or magazine ads. While e-cigarette ads on the Internet are also restricted from making health claims, e-cigarette sellers avoid legal repercussions by using third parties to market e-cigarettes online as having benefits, including aiding in smoking cessation [10]. Consequently, the current research has implications for policy regarding regulation and oversight of Internet-based e-cigarette ads.

Our study also found that advertising exposure through both TV and the Internet was indicative of believing e-cigarette use was acceptable multiple locations. While our cross-sectional design cannot determine causality, previous research has shown that among adolescents, exposure to ads, even those without health claims, increases positive perceptions of e-cigarettes [12]. Further research should explore whether general advertising exposure is a stronger predictor of perceived acceptability than the content of the ad. Currently e-cigarettes are exempt from FDA pre-market reviews, are legal to sell over the internet, require no warning labels, and if e-cigarette advertisements do not make health claims, they are subject to little regulation. However, Congress and the FDA are still determining their policy toward e-cigarettes and results of the growing body of literature on the effects of e-cigarette advertisements may inform policy.

The current research also reveals that approximately half of the young adults surveyed believed it was permissible to use an e-cigarette in a bar where traditional tobacco cigarettes are banned and more than 30% thought using an e-cigarette in a restaurant was acceptable. In contrast, a 2008 survey of college students found a large majority of respondents favored traditional tobacco cigarettes bans in restaurants (87%) and bars (78%) [19]. Given the rising rates of e-cigarette use, further research should explore the effect of e-cigarette use on perceptions toward tobacco and e-cigarette bans.

Our finding that college students perceived e-cigarettes as being slightly more harmful than marijuana but much less harmful than traditional tobacco cigarettes is consistent with previous studies [20]. Our results demonstrated college students perceived e-cigarettes as much more addictive than marijuana, whereas Berg and colleagues found marijuana was perceived as slightly more addictive [20]. This could reflect regional attitudes as their study was performed in the Southeast United States, while the current study was performed in the Midwest.

That college students perceive e-cigarettes as less addictive than traditional tobacco cigarettes is concerning because, among adolescents, intention to use cigarettes has been found to be highly correlated with perceived addictiveness [21]. Given this, adolescents who never would have tried traditional cigarettes may use e-cigarettes.

The strongest predictor of perceived addictiveness, perceived harm, and opinions about acceptability of e-cigarettes use in a different location was participant reported e-cigarette use. Unfortunately, because this is a cross-sectional study, we cannot determine whether beliefs about e-cigarettes as less risky and more acceptable influences use, if use impacts these beliefs, if there is a confounding factor we are not measuring, or some combination of the three.

The current research is not without limitations. First, this study was cross-sectional and therefore it can only reveal association and not causation. For example, it is possible that people who are more likely to believe e-cigarettes are less addictive are also more likely to view e-cigarette ads. Furthermore, the study was not able to control for exposure to e-cigarette advertising. For instance, respondents were asked to indicate which, if any, types of e-cigarette advertising they had ever seen. There was no attempt to capture frequency or duration of exposure for the various types of e-cigarette advertising possible. Despite the utilization of strategies to increase response rate through reminders and incentives, the current study achieved a relatively low response rate of 27%, and thus could have resulted in biased results. However, research exists that challenges the belief that low response rate is directly associated with biased study results [22‐24]. Support for the effectiveness of online surveys as a means to collect valid data is mixed [25, 26]. The current study utilized several strategies to increase response rate and thus reduce non-response bias, including the use of a distribution list comprised of currently enrolled students, incentivizing participation through the use of a lottery that included several relatively valuable (e.g., 15-$50) prizes, numerous (e.g., 20 gift cards) chances to win [27], and providing follow-up email reminders to those who had not completed the survey. Lastly, our population consisted of students at one Midwestern university and consequently our findings may not generalize to other college populations or young adults as a whole.