Exposure to tobacco marketing in bars predicts subsequent use of multiple tobacco products among non-tobacco-using college students

Abstract

Background

Due to other marketing restrictions, one venue where tobacco companies concentrate their marketing efforts to reach young adults is bars/nightclubs.

Objective

This study examined the relationship between exposure to tobacco marketing in bars/ nightclubs and number of alternative tobacco/nicotine products used 6 months later among college students.

Methods

Participants were 1,406 students aged 18–29 years old who reported going to bars or nightclubs at least rarely (M age=21.95; 67% female; 46% non-Hispanic white). Students completed an online survey in fall 2014/spring 2015 (wave 1) and again 6 months later (wave 2). Multilevel Poisson regression models were used to assess the relationship between exposure to three types of marketing at bars/nightclubs at wave 1 (tobacco/nicotine product advertisements; free samples; industry representatives) and number of tobacco products used (range=0–5) at wave 2, controlling for school type (2 year vs 4 year), age, sex, race/ethnicity and frequency of bar visits. An interaction between the number of wave 1 products and each marketing variable was tested.

Results

Greater exposure to free samples and tobacco industry representatives at bars/nightclubs predicted a greater number of products used 6 months later, but only among wave 1 non-tobacco users and not among tobacco users. Exposure to advertisements at bars/nightclubs did not predict the number of products used 6 months later.

Conclusion

Tobacco companies claim that marketing is targeted to those who already use the product, not to non-users. However, the current study indicates tobacco marketing in bars and nightclubs may encourage use among non-users and has no influence on current users.

INTRODUCTION

Since the Master Settlement Agreement and Smokeless Tobacco Master Settlement Agreement in 1998, which restricted cigarette and smokeless tobacco marketing to those less than 18 years of age, young adults have become tobacco companies’ youngest legal targets.¹ While these policies helped decrease overall cigarette and smokeless tobacco use in the USA,^1–3 the use of alternative tobacco/nicotine products (ATNP) (e.g., electronic nicotine delivery system (ENDS), smokeless tobacco and cigar products) continues to be a concern among young adults.^{4 5} According to a nationally representative study, approximately 41% of young adults between 18 and 24 years of age report any tobacco use within the past 30 days⁶ and this age group is more likely to use two or more tobacco products than any other age group.^{7 8} College students, in particular, are at an increased risk for using ATNPs such as cigarillos, smokeless tobacco, or ENDS^{5 9–11} and often report concurrent use of these products,^{4 12–14} thereby increasing their risk for nicotine addiction and dependence.¹⁵ One reason for increased use may be that tobacco companies are capitalizing on this vulnerable developmental period with targeted marketing strategies to promote conventional, new, and emerging products.^{16 17}

Internal tobacco industry documents revealed the importance of marketing to young adults during this important developmental period.¹⁸ Young adults entering college, the military or workforce explore their identity, may experience increased levels of stress and pressure, and experience changes in their social networks, all of which may increase their susceptibility for tobacco use.¹⁹ Recent evidence suggests that young adults may be at an increased risk for tobacco use initiation. Perry et al²⁰ examined the onset of ever and current (past 30 days) tobacco use among youth and young adults (ages 18–24 years) using longitudinal data from the Texas Adolescent Tobacco and Marketing Surveillance System (adolescents), the Marketing and Promotions Across Colleges in Texas (M-PACT) project (young adults) and the Population Assessment of Tobacco and Health (adolescents and young adults) nationally representative study, and found that young adults are now initiating at greater rates compared with youth across multiple tobacco products. Therefore, it is important to understand the marketing strategies being used by tobacco companies that may be contributing to this shift to greater young adult onset.

Tobacco companies use a variety of marketing strategies to reach young adults, including the sponsorship of marketing promotions at adult-only entertainment facilities such as bars and nightclubs.^{21 22} These venues represent an opportunity for tobacco companies to link their brands and products to popular social places, thereby establishing the credibility and relevance of their products among young adults.²³ While research on tobacco promotions in bars and nightclubs is limited, studies suggest the tobacco industry has increased its use of these locations as promotional venues since 1990.^{23 24} One well-documented strategy tobacco companies used was to sponsor social events at bars and nightclubs where cigarette samples and promotional items were distributed to encourage the initiation or progression of cigarette use.^25–27 As of August 2016, all tobacco products are subject to the Federal Drug Administration’s (FDA) rule, which prohibits the distribution of free samples, with the exception of smokeless tobacco distributed in a ‘qualified adult-only facility’ (i.e., facilities that verify that all customers are at least 18 years of age).²⁸ Furthermore, two longtime marketing tactics that may be used to attract young adults are still permitted: (1) tobacco brand sponsorship of athletic, musical or cultural events, which may occur at bars and nightclubs, and (2) the distribution of non-tobacco merchandise (eg, t-shirts, hats, and so on) for products other than cigarettes and smokeless.²⁹ Previous research documents the association between exposure to cigarette promotions such as attendance at a tobacco industry-sponsored event at a bar or nightclub,¹ tobacco-branded paraphernalia¹⁷ and a higher risk of smoking. Therefore, it is possible that exposure to promotions at these locations for other tobacco products, such as ENDS and cigars, may also be associated with a higher risk of tobacco use.

Interestingly, while the amount spent on cigarette advertising and promotions increased from $8.03 billion in 2014 to $8.64 billion in 2017,^{30 31} the amount spent in adult-only promotions decreased from $138.9 million in 2010³² to $72.4 million in 2015.³³ Similarly, expenditures for smokeless tobacco in adult-only facilities decreased from $32.4 million in 2010 to $15.6 million in 2017.³⁴ However, as tobacco products other than cigarettes and smokeless tobacco do not report their expenditures to the Federal Trade Commission, it is difficult to obtain total marketing expenditure estimates. Limited evidence suggests that tobacco companies continue to increase their marketing expenditures for ENDS, including e-cigarettes,^{35 36} which may include allocating funds for promotions at adult-only facilities.

Ultimately, relatively little is known about the influence of marketing tactics used to promote tobacco products in bars and nightclubs among college students, and no studies have examined the influence on subsequent tobacco use. Tobacco companies have long claimed that the purpose of their marketing strategies is to promote brand loyalty and brand switching among current users.³⁷ However, research shows that these marketing tactics also influence youth and young adult non-users by encouraging them to experiment and initiate regular tobacco use.¹⁶ With the current surge in availability of a variety of tobacco products, we have also seen an expansion of the marketing for these products that may be contributing to the use of multiple tobacco products, or polytobacco use among young adults.^{19 38} Recent studies indicate that polytobacco use is becoming increasingly popular among young adults^{15 39 40} and that young adults are likely to concurrently use cigarettes with emerging tobacco products (eg, ENDS, cigar products and hookah).^{40 41} Although the polytobacco use phenomenon is understudied,⁴² it is known that youth and young adults who use multiple tobacco products may be more susceptible to tobacco advertising compared with non-users or exclusive cigarette users.¹⁵ However, it is not known how marketing in bar/nightclub venues influences tobacco use behaviours among both tobacco users and non-users.

Based on current evidence linking exposure to tobacco marketing to an increased risk for tobacco use,¹⁶ this study examines if exposure to tobacco marketing and free samples in bars and nightclubs predicts the number of products used 6 months later among college students. We hypothesise that greater exposure to tobacco marketing will be associated with greater tobacco use 6 months later. Second, given that tobacco users may be more susceptible to tobacco marketing,¹⁵ we will examine the interaction between the number of tobacco products used and marketing exposure. We hypothesise that the effect of tobacco marketing on subsequent overall tobacco use will be greater among those who use one or more tobacco products compared with those who do not use tobacco products.

METHODS

Participants

Participants were 1,406 college students participating in the M-PACT surveillance study, which evaluated tobacco use in a sample of 5,482 young adults attending 1 of 12 two-year or 12 four-year Texas colleges. Only students pursuing full or part-time vocational training/certificates were recruited from the 2-year colleges. Additionally, students at 4-year universities were eligible to participate in the study if they were either part-time or full-time undergraduate degree-seeking students. Students were eligible to be included in the study if at wave 1, they responded that they at least ‘rarely’ visited bars or nightclubs on a scale that ranged from ‘never’ to ‘frequently’ in response to the question, ‘In general, how often do you go to bars/nightclubs?’ Of the 1,945 students who went to bars at wave 1, there were 1,406 who completed the wave 2 survey and are included in the present study (response rate=72.3%). Two-thirds of the 1,406 participants were female, 46% were non-Hispanic white, 29% were Hispanic, 11% were Asian, 6% were African-American and 8% were of another race/ethnicity. Most participants attended 4-year universities (see table 1).

Table 1.

Participant demographics, tobacco use and reported exposure to marketing variables at bars or nightclubs at wave 1 of M-PACT project (n=1406)

Characteristic	Mean/percentage
Age mean (SD)	21.95 (2.36)
Gender
Male	33.4%
Race/ethnicity
Non-Hispanic white	46.2%
Hispanic	29.0%
African-American	6.1%
Asian-American	10.7%
Other	8.0%
School level
2-year college	5.0%
4-year college/university	95.0%
Exposure to marketing types at bars/nightclubs
Tobacco advertisements	61.0%
Free samples	23.2%
Tobacco industry representatives	25.8%
Number of tobacco products used at wave 1
None	53.3%
One product	20.2%
Two products	15.2%
Three or more products	11.3%
Most common products used at wave 1 (among users)
Cigarettes	28.2%
ENDS	21.8%
Hookah	22.3%
Cigars	12.1%
Smokeless tobacco	3.6%
Most common products used 6 months later (among users)
Cigarettes	23.6%
ENDS	18.8%
Hookah	18.9%
Cigars	8.9%
Smokeless tobacco	2.9%

ENDS, electronic nicotine delivery system; M-PACT, Marketing and Promotions Across Colleges in Texas.

Attrition analyses were conducted to compare those who were in the study (i.e., responders, n=1,406) and those who did not have data at wave 2 (i.e., non-responders, n=539). None of the demographic variables differed between responders and non-responders: age (X² (1)=2.41, p=0.121), sex (X² (1)=0.97, p=0.324), race (X² (4)=6.54, p=0.162) and attending a 4-year college (X² (1)=0.392, p=0.532). However, the non-responders were significantly more likely to go to bars/clubs frequently (X² (1)=9.74, p=0.002), use more tobacco products at wave 1 (X² (1)=11.85, p<0.001) and report seeing tobacco samples in bars (X² (1)=13.98, p<0.001), but did not differ on whether they saw tobacco advertisements (X² (1)=0.576, p=0.448) or tobacco representatives (X² (1)=0.710, p=0.400).

Procedures

The 24 colleges were selected from the five counties surrounding the four largest cities in Texas. Three colleges of each type were selected from the five counties surrounding the four target cities (Austin, Dallas/Fort Worth, Houston and San Antonio), for a total of 6 per city, 24 total. Eligible students were invited to participate via email invitation and screened for eligibility. If eligible, they provided informed consent and completed the online survey. Of 13 714 students meeting the eligibility criteria (ie, part-time or full-time enrolment at 1 of the 24 colleges, between 18 and 29 years old), 40%, or 5482 provided consent and completed the survey. This rate of participation is similar to, or exceeds, other online studies of college students.^{43 44} The first wave of data collection occurred between November 2014 and February 2015 and a second wave was collected approximately 6 months later between May and June 2015.

χ² tests were conducted to examine if there were differences between 2-and 4-year students on sociodemographic and exposure variables. No significant differences were found for sex, race or exposure variables (tobacco advertisements, free samples and tobacco industry representatives). There was an age difference between 2-and 4-year students, with 4-year students being significantly younger; however, age was included in all models as a control variable.

Measures

Three items assessed exposure to tobacco marketing at wave 1: (A) How often have you been to a bar/nightclub when free samples of tobacco products or ENDS products (ie, e-cigarettes, vape pens or e-hookahs) were available? (B) How often do you see tobacco or ENDS product advertisements in bars or nightclubs on walls or furniture, or on napkins, coasters or giveaway items? (C) How often have you seen or interacted with a tobacco or ENDS representative at a bar or nightclub? Each item had a response scale ranging from ‘never’ (0) to ‘always’ (3). For each item, participants who reported any exposure (i.e., any response other than never) were assigned as exposed and given a ‘1’. However, if participants answered ‘never’, they were assigned as non-exposed and given a ‘0’. These items were adapted from previous work by Gilpin and colleagues⁴⁵ to include ENDS products.

The number of products used was assessed with five questions 6 months later (wave 2) including past 30-day use of cigarettes, ENDS, smokeless tobacco, cigars and hookah. Each question asked about past 30-day use, for example: ‘During the past 30 days, have you used any ENDS product (ie, an e-cigarette, vape pen, or e-hookah), even one or two puffs, as intended (ie, with nicotine cartridges and/or e-liquid/e-juice)?’ Response options were 0=‘No’; 1=‘Yes’. Each product was counted as 1, therefore the number of products used ranged from 0 to 5.

All models contained the following wave 1 covariates: sex (0=female, 1=male), age at baseline z score, race/ethnicity (dummy coded for Hispanic ancestry, African ancestry, Asian ancestry and other race/ethnicity), 4-year college (0=2 years, 1=4 years), frequency of bar visits and wave 1 number of tobacco products used.

Statistical models

To determine if exposure to tobacco marketing and free samples in bars and nightclubs predicted the number of tobacco products used 6 months later, generalised linear mixed models were fit for a Poisson distribution using the log-link function to estimate the number of products used.⁴⁶ The Poisson distribution was selected for two reasons: (1) the outcome is a count response variable, and (2) as many observations of the outcome had a value of 0, this can cause the sample to be skewed. Poisson models were evaluated for overdispersion using the χ² distribution based on an approximate estimate of an overdispersion parameter, the Pearson residuals sum of squares, using the model’s residual df.⁴⁷ The Poisson distribution using the log-link function was used to test the effect of marketing exposure among tobacco users and non-users and also used to test the interaction between the number of products used at wave 1 and the marketing variables. All models were implemented using the R glmer function.⁴⁸ Respondents were nested within their wave 1 school. The intercept was treated as a random coefficient. In addition to age, sex, race/ethnicity and school type, the models also contained the number of products used at wave 1 as a covariate. Each tobacco marketing variable was evaluated in separate models. To test our hypothesis that the effect of marketing exposure would be greater among those who are current tobacco users at wave 1, we tested the interaction between the number of wave 1 products used and the marketing variables in a separate model for each type of marketing. This allowed us to examine the relationship between marketing exposure and the outcome across the range of number of products used. The interaction effects were probed using the simple slopes technique,⁴⁹ in which the significance of the simple slope of wave 2 number of products regressed on a putative marketing variable (eg, tobacco representative) is estimated at low and high values of wave 1 number of tobacco products used. We selected prototypical values of 0 for low levels and 2 for high levels of number of products used at wave 1. The values of 0 and 2 are approximately 1 SD below and above the mean (M=0.88, SD=1.14), which are typical values used to evaluate an interaction effect.⁴⁹

RESULTS

As shown in table 1, over half (61.0%) of participants reported seeing a tobacco advertisement, while about a quarter of participants reported seeing either free samples (23.2%) or a tobacco representative (25.8%) when they visited bars/nightclubs. Although more than half of the participants did not use any products (53.3%), among those who were past 30-day users, most used 1 (20.2%) or 2 (15.2%) products at wave 1. The most common products used at both waves were cigarettes (28% at wave 1 and 24% at wave 2), followed by hookah (22% at wave 1, 19% at wave 2) and ENDS (22% at wave 1 and 19% at wave 2).

Free samples

Evaluation of the overdispersion parameter indicated that the data were not overdispersed (X² (1,394)=1,390.03, p=0.525). In the main effects model, there was no significant main effect of exposure to free samples at wave 1 on number of products used 6 months later (z=1.16, p=0.25, see table 2). However, in the interaction model, we found that there was a significant two-way interaction between wave 1 free samples and number of products used at wave 2 (z=−2.53, p=0.01). When probing the interaction, we found that exposure to free samples was a significant predictor of wave 2 number of products used when estimated for low wave 1 product use (ie, the simple slope estimated at wave 1 products use=0) (z=2.83, p=0.01; see figure 1), but found no effect of exposure to free samples on number of products used at wave 2 when estimated at high wave 1 product use (ie, the simple slope estimated at wave 1 products use=2) (z=1.04, p=0.30). These findings indicate that exposure to free samples at bars or nightclubs did not influence those who were already users but did impact those who did not use any tobacco products at wave 1.

Table 2.

Main and interaction effects of the relationship between wave 1 tobacco marketing exposure variables and count of products used 6 months later among college students who attend bars/clubs (n=1406)

Wave 1 variables	Free samples				Industry representatives				Tobacco ads
	Main effects model		Interaction model		Main effects model		Interaction model		Main effects model		Interaction model
Parameter	Incidence rate ratios† (95% CI)	P value	Incidence rate ratios† (95% CI)	P value	Incidence rate ratios† (95% CI)	P value	Incidence rate ratios† (95% CI)	P value	Incidence rate ratios† (95% CI)	P value	Incidence rate ratios† (95% CI)	P value
Intercept	0.21 (0.15 to 0.30)	<0.001	0.20 (0.14 to 0.27)	<0.001	0.22 (0.16 to 0.30)	<0.001	0.17 (0.12 to 0.24)	<0.001	0.21 (0.15 to 0.29)	<0.001	0.19 (0.13 to 0.27)	<0.001
Age (z score)	1.00 (0.94 to 1.06)	0.981	1.00 (0.94 to 1.07)	0.980	1.00 (0.93 to 1.06)	0.929	1.00 (0.94 to 1.06)	0.929	1.00 (0.94 to 1.07)	0.943	1.00 (0.94 to 1.07)	0.932
Male	1.13 (0.99 to 1.28)	0.075	1.13 (0.99 to 1.28)	0.071	1.12 (0.99 to 1.28)	0.091	1.12 (0.98 to 1.27)	0.091	1.13 (0.99 to 1.29)	0.066	1.13 (0.99 to 1.29)	0.062
Hispanic/Latino	1.05 (0.91 to 1.22)	0.503	1.06 (0.91 to 1.23)	0.465	1.04 (0.90 to 1.21)	0.351	1.07 (0.93 to 1.25)	0.351	1.05 (0.91 to 1.22)	0.505	1.06 (0.91 to 1.23)	0.449
African ancestry	1.05 (0.80 to 1.39)	0.711	1.04 (0.79 to 1.38)	0.764	1.06 (0.80 to 1.40)	0.682	1.06 (0.80 to 1.40)	0.682	1.05 (0.79 to 1.39)	0.739	1.07 (0.81 to 1.41)	0.651
Asian ancestry	1.10 (0.90 to 1.36)	0.353	1.11 (0.90 to 1.37)	0.316	1.10 (0.89 to 1.36)	0.345	1.11 (0.90 to 1.36)	0.345	1.10 (0.90 to 1.36)	0.354	1.12 (0.91 to 1.38)	0.301
Other race/ethnicity	1.13 (0.90 to 1.41)	0.307	1.14 (0.91 to 1.44)	0.247	1.12 (0.89 to 1.41)	0.222	1.15 (0.92 to 1.45)	0.222	1.13 (0.90 to 1.41)	0.309	1.13 (0.90 to 1.42)	0.292
Four-year institution	0.96 (0.72 to 1.29)	0.804	0.96 (0.72 to 1.30)	0.838	0.96 (0.72 to 1.28)	0.965	1.01 (0.75 to 1.35)	0.965	0.97 (0.72 to 1.29)	0.813	0.97 (0.72 to 1.29)	0.825
In general, how often do you go to bars/clubs?	1.12 (1.03 to 1.22)	0.011	1.12 (1.03 to 1.22)	0.10	1.11 (1.02 to 1.21)	0.007	1.13 (1.03 to 1.23)	0.007	1.13 (1.03 to 1.23)	0.006	1.13 (1.04 to 1.23)	0.005
Number of products used	4.06 (3.60 to 4.58)	<0.001	4.42 (3.85 to 5.08)	<0.001	4.04 (3.57 to 4.56)	<0.001	4.83 (4.18 to 5.59)	<0.001	4.08 (3.62 to 4.60)	<0.001	4.53 (3.75 to 5.47)	<0.001
Free samples*	1.10 (0.94 to 1.29)	0.248	1.57 (1.15 to 2.15)	0.005	–	–	–	–	–	–	–	–
Number of products used × free samples*	–	–	0.69 (0.52 to 0.92)	0.11	–	–	–	–	–	–	–	–
Industry representatives*	–	–	–	–	1.13 (0.98 to 1.30)	0.097	2.00 (1.53 to 2.63)	<0.001	–	–	–	–
Number of products used × industry representatives*	–	–	–	–	–	–	0.56 (0.43 to 0.71)	<0.001	–	–	–	–
Advertisements*	–	–			–	–	–	–	1.03 (0.91 to 1.17)	0.617	1.21 (0.94 to 1.56)	0.140
Number of products used × advertisements*	–	–	–	–	–	–	–	–	–	–	0.84 (0.67 to 1.07)	0.158

Bolded values indicate statistical significance at p<.05

^*Separate models were run for each marketing exposure, controlling for covariates.

^†Incidence rate ratios are multiplicative changes in the counts of products used at wave 2.

Figure 1.

Fitted values for interaction between number of tobacco products/ATNPs used at wave 2 and exposure to free tobacco samples at wave 1 for number of products=0 and 2. ATNP, alternative tobacco/nicotine product.

Tobacco industry representatives

Evaluation of the overdispersion parameter indicated that the data were not overdispersed (X² (1,394)=1,376.77, p=0.624). In the main effects model, there was no significant main effect of exposure to tobacco industry representatives at wave 1 on number of products used 6 months later at wave 2 (z=1.66, p=0.09, see table 2). However, in the interaction model, the two-way interaction between exposure to tobacco industry representatives at wave 1 and number of products used at wave 1 was significant (z=−4.62, p<0.001). When probing the interaction, we found that the relationship between exposure to tobacco industry representatives and number of products used at wave 2 was significant when estimated at low wave 1 product use (z=4.99, p<0.001, see figure 2), but not when estimated at high wave 1 product use (z=1.16, p=0.25), indicating that exposure to tobacco industry representatives at bars or nightclubs had an influence on non-users, but not current users.

Figure 2.

Fitted values for interaction between number of tobacco products/ATNPs used at wave 2 and exposure to tobacco industry representatives at wave 1 for number of products=0 and 2. ATNP, alternative tobacco/nicotine product.

Tobacco advertisements

Evaluation of the overdispersion parameter indicated that the data were not overdispersed (X² (1,394)=1,391.93, p=0.511). Finally, no significant main effect was found for exposure to advertisements on product use 6 months later (z=0.5, p=0.62, see table 2). The interaction between exposure to tobacco advertisements and number of products used at wave 1 was also not significant (z=−1.41, p=0.16).

For each of the models, more frequent bar/nightclub attendance and number of products used at wave 1 predicted a greater number of products used 6 months later (see table 2). Sex, race/ethnic group and school type (2 or 4 years) did not predict the number of products used 6 months later in any of the models (see table 2).

DISCUSSION

This study found that exposure to tobacco marketing, specifically free samples and tobacco industry representatives in bars and nightclubs, predicted an increased number of tobacco products used 6 months later, but only for college students with low tobacco product use at wave 1 (e.g., wave 1 non-tobacco users). Tobacco companies have long understood the importance of social environments and making smoking a normal and essential part of the bar and nightclub scene.^{22 50–52} Therefore, it was not surprising that college students who go to bars/nightclubs more often are also more likely to initiate or increase tobacco use 6 months later. Research has already documented the strong association between alcohol and tobacco use,^{53 54} and this long-standing relationship has led to a new kind of tobacco user, the ‘social smoker’, typically defined as ‘smoking primarily in social contexts such as parties, bars, or nightclubs’.⁵⁵ While studies focusing on this phenomenon have identified social cigarette smokers, it is possible that because of widespread smoke-free indoor air laws and with the rising popularity of emerging tobacco products, such as ENDS and smokeless tobacco, young adults may replace cigarettes with other tobacco products in these settings. For example, a study conducted by Adams et al⁵⁶ found that smokers used smokeless tobacco as a cigarette substitute in public spaces, including bars, to continue their nicotine consumption in light of existing state-wide smoke-free laws. Even though the new FDA deeming rule has prohibited free samples of most tobacco products, it exempted smokeless tobacco products in adult-only facilities. Future research can determine whether the free sample ban influences the prevalence of smokeless tobacco in these settings, and overall.

Tobacco industry-sponsored events at bars and nightclubs are among the few promotional activities permitted by the US law. More often than not, tobacco representatives attend these events. Even though they will no longer be able to provide free samples of tobacco products, with the exception of smokeless tobacco in adult-only facilities, the deeming rule does not prohibit the use of coupons, as long as the coupon is not used to obtain a free tobacco product.⁵⁷ While not directly measured in this study, it is not unusual for tobacco representatives to hand out coupons and/or request personal contact information to send coupons via direct mail during promotional events.⁵⁸ The tobacco industry maintains that the purpose of these strategies is only to compete for brand share among current users and not to increase consumption,³⁷ but our findings and other research suggest otherwise.^59–61 Future longitudinal research is needed to determine whether this trend continues among this population.

Finally, while exposure to tobacco advertising was not found to have an overall significant association with use at follow-up, it is likely that these messages are still being processed, but at a subconscious level. Moreover, it is possible that a significant effect between exposure to tobacco advertising and use at follow-up was not found because advertising at bars/nightclubs may not be as prominent as tobacco advertising found in other settings such as the tobacco retail outlet. Additionally, the bar/nightclub environment may be more distraction filled than exposure at a retail outlet or via the internet or print media. As such, even though some young adults did report exposure this may not have been enough to influence use 6 months later. The elaboration likelihood model posits that while only messages with high importance are processed with care, other messages are still processed, but with less effort.⁶² As such, tobacco advertising at bars and nightclubs may serve as an environmental cue that triggers curiosity and use behaviours without directly affecting attitudes and beliefs, similar to how food advertising works on food-related behaviours.^{18 63} Future research should investigate how multiple exposures to tobacco advertising from a variety of marketing channels influence tobacco behaviours among young adults.

This study is subject to several limitations. First, the measure used to determine exposure to free samples did not distinguish between seeing and taking the free sample. Therefore, we are unable to determine how these samples were used to encourage subsequent use. Future research should include measures that distinguish between simple exposure and actual use of free samples. Furthermore, we did not measure brand loyalty. While research on brand loyalty for other tobacco products is limited, cigarette smokers do tend to exhibit brand loyalty, that is, each time a smoker wants to smoke a cigarette, they look for their favourite branded cigarette⁶⁴; this may be how marketing influences tobacco users. Future research should include measures of brand loyalty to determine if there are similar effects found for ATNPs. Also, we only asked about exposure to tobacco and ENDS advertising in general, not specific to each product. Future research should examine how exposure to tobacco marketing (ie, advertising and promotional activities) and tobacco use in bars and nightclubs affect social norms, harm perceptions of these products and use rates. Another limitation is the classification of non-users, as it is possible that even though they did not use a tobacco product in the past 30 days, they may have used a tobacco product in their lifetime. However, we believe that current use (past 30 days) is a better measure to capture sustained users of tobacco products. Finally, we do not know how often these tobacco promotions occur. Therefore, future research should focus on developing a systematic approach for identifying, defining and evaluating tobacco promotions in these spaces on a larger scale. The continued monitoring of the promotional activities that take place in bars and nightclubs can help improve our understanding of how these activities influence tobacco use. Furthermore, information collected can be used to improve and develop tobacco control policies.

There are hundreds, even thousands of bars and clubs across the four Texas cities where study participants reside, and it is safe to assume that young adults, including college students, will continue to attend these venues. Therefore, it is important to focus tobacco control efforts in bars and clubs, as young adults remain a high priority population for tobacco companies. Comprehensive (ie, no exceptions) smoke-free policies in bars have been associated with a significant decline in individual smoking risk and intensity, as well as an increase in the likelihood of quitting among young to middle-aged adults over 25.⁶⁵ Therefore, tobacco control efforts should focus on extending policies to make them more comprehensive by including outdoor spaces (ie, patios), eliminating coupons for all tobacco products and banning all tobacco sponsorship in order to prevent the initiation and/or continued use of tobacco products among young adults.

What this paper adds.

• Adult-only venues such as bars and nightclubs remain increasingly important venues for tobacco companies to reach young adults.

• This study found that greater exposure to free samples and tobacco industry representatives in bars and nightclubs predicted a greater number of tobacco products used at follow-up, among wave 1 non-users, but with no effect among wave 1 users.

• Given that the effects of marketing were found only among non-users, this suggests that these venues may be important areas for future regulations to reduce initiation.