E-cigarette Marketing Regulations and Youth Vaping: Cross-Sectional Surveys, 2017–2019

Data are from waves 1 to 3 of the International Tobacco Control Policy Evaluation Project (ITC) Youth Tobacco and Vaping Survey, conducted in Canada, England, and the United States. Online surveys were conducted in July and August 2017 (wave 1), August and September 2018 (wave 2), and August and September 2019 (wave 3). The same methods and recruitment protocols were used in all 3 countries. Respondents were recruited from the Nielsen Consumer Insights Global Panel, which maintains panels in Canada, England, and the United States, as well as their partners’ panels. The Nielsen panel is recruited by using both probability and nonprobability sampling methods in each country. We identified target sample sizes of 4500 for each country at each wave, based on sample size calculations to detect an absolute change of ∼2% in vaping and smoking prevalence across waves. Eligible respondents included youth aged 16 through 19 at the time of recruitment. Respondents were recruited either directly or through their parents. E-mail invitations (with a unique link) were sent to panelists after targeting for age criteria. Panelists known to be parents were also contacted; those who confirmed they had 1 or more children aged 16 to 19 living in their household were asked for permission for their child to complete the survey (if more than 1 child, specifically the one whose birthday was coming up next). After eligibility screening, all potential respondents were provided with information about the study and were asked to provide consent.

The current analyses are based on the cross-sectional samples of Canadian respondents to the ITC Youth Tobacco and Vaping Survey. In Canada, the total number of respondents completing the survey was 4491 in 2017, 4298 in 2018, and 4501 in 2019. Respondents were excluded from the sample if they failed a data integrity check (n = 144 in 2017, n = 134 in 2018, and n = 191 in 2019), had missing or invalid data on key variables (age, sex, province, ever or last smoked and vaped; n = 323 in 2017; n = 86 in 2018; n = 107 in 2019), or had participated in a previous wave of the study (n = 233 in 2018; n = 68 in 2019). The analytic sample included 4024 participants recruited at wave 1, 3845 recruited at wave 2, and 4135 recruited at wave 3. A full description of the study methods, including survey cooperation rates, is available in the Technical Reports.²⁸ 

This study was reviewed and received ethics clearance through a University of Waterloo Research Ethics Committee (ORE#21847) and the King’s College London Psychiatry, Nursing, and Midwifery Research Ethics Subcommittee.

Self-reported exposure to marketing was measured by response to the item, “In the last 30 days, how often have you noticed things that promote e-cigarettes/vaping?” (never, rarely, sometimes, often, very often, don’t know, refused). Overall exposure to marketing was analyzed as a binary outcome (often or very often versus never, rarely, or sometimes). To determine whether the categorization of this outcome impacted the results, parallel models to those reported in the paper were fitted using an alternative categorization (often, very often, or sometimes versus never or rarely), as well as linear models in which the outcome was treated as a continuous variable. The pattern of differences between survey waves and between provinces was similar regardless of which variable or categorization was used. Therefore, results using the binary variable are presented. Respondents reporting “don’t know” or “refused” (n = 296 in wave 1, n = 292 in wave 2, n = 244 in wave 3) were excluded from analysis of this measure.

Respondents who indicated noticing marketing at least rarely were asked whether they noticed e-cigarettes or vaping devices or e-liquid being advertised in each of 15 specific settings.

Variables were created to categorize the strength of marketing restrictions in each province on the basis of information sourced from summaries of legislation.²⁹  Separate policy strength indicator variables were created to indicate whether e-cigarette advertisements were permitted in each of the following channels (0 = prohibited, 1 = permitted): television, radio, billboards (restricted in Quebec, Manitoba, Prince Edward Island, Nova Scotia), and advertisements or product displays inside and outside stores (restricted in British Columbia, Manitoba, Quebec, New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador). A policy strength index was also created to reflect the overall level of marketing restrictions across provinces: low restrictions (4) were in Alberta and Saskatchewan; low-moderate restrictions (3) were in Ontario; moderate restrictions (2) were in New Brunswick and Newfoundland and Labrador; high-moderate restrictions (1) were in British Columbia; and high restrictions (0) were in Quebec, Manitoba, Nova Scotia and Prince Edward Island.

Respondents were asked to report their vaping and smoking behavior, including the last time they used an e-cigarette or vaped (analyzed as past 30-day and past-week vaping) and the number of days they vaped in the past 30 days (analyzed as ≥20 days versus less).

Poststratification sample weights were constructed for each country, based on age, sex, and geographic region, and rescaled to the sample size. In addition, the US National Youth Tobacco Survey and the Canadian Student Tobacco, Alcohol and Drugs Survey were used to calibrate to the trend over time for past 30-day smoking (see Technical Reports²⁸ ).

Regression models were fitted for 3 primary outcomes: (1) overall marketing exposure, (2) exposure to marketing in specific channels, and (3) vaping prevalence. All models included indicator variables for survey wave (2017, 2018, or 2019), province nested within policy strength (low restrictions, low-moderate, moderate, high-moderate, high restrictions), and a 2-way interaction term between survey wave and policy strength. All regression models were adjusted for age (16–17 vs 18–19), sex, and race or ethnicity (white versus other). First, overall exposure to marketing was examined in a logistic regression model using the binary outcome (often or very often versus never, rarely, or sometimes). Second, separate logistic regression models were estimated for exposure to each of the 15 specific marketing channels (inside stores, billboards, etc). For these models, the overall “policy strength” variable was replaced by a binary variable indicating whether the specific marketing channel was permitted or prohibited within each province. Finally, 3 logistic regression models were estimated to examine differences over time in the prevalence of vaping in the past 30 days, past week, and ≥20 days in the past month. The “overall marketing exposure” variable was included as a predictor variable in these models.

Adjusted odds ratios (AORs), 95% confidence intervals (CIs), and weighted estimates are reported, unless otherwise noted. Respondents with missing data on outcome measures or covariates were excluded from models on a case-wise basis. Analyses were conducted by using SAS 9.4.

The weighted sample characteristics across the 3 waves are shown in Table 1. Across all years, participants had an average age of 17.5 years, 51% were male, and 53% self-identified as white. In addition, 58% reported consuming alcohol in the past year, and 10% reported smoking a cigarette in the past 30 days.

The proportion of respondents who reported noticing e-cigarette promotions in the last 30 days increased from 2017 to 2019: in 2017, 13.6% reported noticing promotions often or very often, which increased to 15.5% in 2018 (AOR = 1.18, 95% CI = 1.02–1.36, P = .03) and 26.0% in 2019 (versus 2017: AOR = 2.24, 95% CI = 1.97–2.56, P < .001; versus 2018: AOR = 1.91, 95% CI = 1.68–2.16, P < .001). Often noticing e-cigarette promotions was more prevalent among youth who were 18 or 19 years old (AOR = 1.15, 95% CI = 1.04–1.28, P = .01) and females (AOR = 1.19, 95% CI = 1.07–1.32, P = .002).

As illustrated in Fig 1, noticing e-cigarette promotions often or very often was generally more prevalent in 2019 among provinces with fewer regulatory restrictions on e-cigarette promotions. The interaction between year and policy strength in the logistic regression model was significant (P < .001). The likelihood of often noticing promotions increased from 2017 to 2019 for all provinces except those with high restrictions (high: AOR = 1.27, 95% CI = 0.95–1.71, P = .10; moderate-high: AOR = 2.55, 95% CI = 1.76–3.71, P < .001; moderate: AOR = 3.40, 95% CI = 1.65–7.00, P < .001; low-moderate: AOR = 2.40, 95% CI = 2.00–2.88, P < .001; low: AOR = 3.74, 95% CI = 2.70–5.19, P < .001). In addition, the increase in often noticing promotions between 2017 and 2019 was substantially greater among provinces with fewer restrictions compared with provinces with the most restrictions (moderate-high: AOR = 2.00, 95% CI = 1.25–3.22, P = .004; moderate: AOR = 2.66, 95% CI = 1.22–5.80, P = .01; low-moderate: AOR = 1.88, 95% CI = 1.33–2.65, P < .001; low restrictions: AOR = 2.93, 95% CI = 1.89–4.54, P < .001). For example, the increase seen from 2017 to 2019 was 2.93 times greater in provinces with low restrictions (AOR = 3.74) than in those with high restrictions (AOR = 1.27).

Self-reported noticing of e-cigarette marketing in specific channels is shown in Table 2. Respondents were most likely to notice marketing inside and outside stores that sold cigarettes, followed by digital media, and billboards. As shown in Table 2, exposure to marketing increased between 2017 and 2019 for all channels except kiosks.

Provincial regulations on e-cigarette marketing differed across 5 specific channels: inside stores, outside stores, billboards, television, and print. As illustrated in Fig 2, respondents in provinces with no restrictions on that specific channel were more likely to notice promotions in the following places, and this difference was larger in 2019 than 2017. This was the case for promotions inside stores (AOR = 1.43, 95% CI = 1.15–1.78, P = .001), outside stores (AOR = 1.49, 95% CI = 1.19–1.87, P < .001), and on billboards (AOR = 1.70, 95% CI = 1.24–2.32, P < .001), and to a lesser extent for television (AOR = 1.32, 95% CI = 0.96–1.80, P = .09); there were no differences for print advertising (AOR = 1.16, 95% CI = 0.84–1.62, P = .37).

The overall prevalence of vaping in Canada increased between 2017 and 2019 for all outcomes: vaping in the past 30 days (2017 = 8.4%, 2018 = 12.1%, 2019 = 17.8%; P < .001), in the past week (2017 = 5.1%, 2018 = 7.5%, 2019 = 12.3%; P < .001), and on ≥20 days in the past month (2017 = 1.8%, 2018 = 2.4%, 2019 = 5.7%; P < .001). Respondents who reported noticing marketing often or very often were more likely to report vaping in the past 30 days (AOR = 1.41, 95% CI = 1.23–1.62, P < .001), in the past week (AOR = 1.44, 95% CI = 1.22–1.70, P < .001), and on ≥20 days in the past month (AOR = 1.42, 95% CI = 1.11–1.81, P = .005).

As indicated in Fig 3, the prevalence of vaping in the past 30 days increased from 2017 to 2019 for all provinces (high restrictions: AOR = 2.09, 95% CI = 1.52–2.87, P < .001; moderate-high: AOR = 2.25, 95% CI = 1.45–3.50, P < .001; moderate: AOR = 2.70, 95% CI = 1.24–5.86, P = .01; low-moderate: AOR = 2.84, 95% CI = 2.26–3.57, P < .001; low: AOR = 2.28, 95% CI = 1.59–3.25, P < .001). Similarly, the prevalence of vaping in the past week increased from 2017 to 2019 for all provinces (high restrictions: AOR = 1.98, 95% CI = 1.35–2.91, P < .001; moderate-high: AOR = 3.67, 95% CI = 2.01–6.72, P < .001; moderate: AOR = 3.09, 95% CI = 1.19–8.01, P = .02; low-moderate: AOR = 3.18, 95% CI = 2.37–4.27, P < .001; low: AOR = 2.41, 95% CI = 1.59–3.64, P < .001). The prevalence of vaping on ≥20 days in the past month increased from 2017 to 2019 in all provinces except those with moderate restrictions (high restrictions: AOR = 2.46, 95% CI = 1.41–4.28, P = .001; moderate-high: AOR = 5.00, 95% CI = 1.76–14.2, P = .003; moderate: AOR = 2.38, 95% CI = 0.60–9.45, P = .22; low-moderate: AOR = 4.01, 95% CI = 2.41–6.67, P < .001; low: AOR = 3.59, 95% CI = 1.74–7.41, P < .001).

Compared with provinces with the strongest marketing restrictions, across all waves, the prevalence of vaping in the past 30 days was higher in provinces with moderate-high (AOR = 1.30, 95% CI = 1.06–1.59, P = .01) and low restrictions (AOR = 1.50, 95% CI = 1.25–1.80, P < .001), whereas the prevalence of vaping in the past week was also higher among provinces with moderate-high restrictions (AOR = 1.27, 95% CI = 1.00–1.62, P = .05) and low restrictions (AOR = 1.65, 95% CI = 1.33–2.05, P < .001). The interaction between year and policy was not significant for past 30-day (P = .14), past-week (P = .12), or vaping on ≥20 days in the past month (P = .61).

Young Canadians reported substantial increases in exposure to e-cigarette marketing after the introduction of new federal legislation permitting the sale and marketing of nicotine-containing e-cigarettes in May 2018. The youth surveyed were most likely to report exposure to marketing at retail outlets, which is consistent with the implementation in mid-2018 and 2019 of prominent in-store advertising and product displays for international brands such as JUUL and Vype. In most cases, these promotions were positioned at the point-of-sale counter, immediately adjacent to candy and other products popular among children and youth.³⁰  Along with increased exposure from more traditional marketing channels, including billboards, radio, and print media, youth also reported substantial increases in exposure to digital marketing, including social media and internet ads. Digital marketing may have originated in Canada, as well as through exposure to cross-border digital marketing from the United States to social media campaigns such as those conducted by JUUL and other companies in the United States.^5,6,23  Findings are consistent with research in the United States that documents increased exposure to e-cigarette marketing over the same period is indicated, including the importance of retail stores as a primary source of exposure.^15,21,31 

E-cigarette marketing was strongly associated with increased vaping behavior at the individual level. In addition, vaping increased in all provinces between 2017 and 2019; although this increase occurred at a similar rate across provinces, those with greater marketing restrictions had lower overall vaping prevalence. The findings are consistent with those from both experimental^18,19  and prospective cohort studies,^20,21  the results of which collectively suggest a causal effect of e-cigarette advertisements on use among young people.

Exposure to e-cigarette marketing was less prevalent in provinces with more comprehensive regulations. Indeed, in provinces such as Quebec and Manitoba, the frequency of youth who noticed marketing often or very often did not increase between 2017 and 2019. Also demonstrated in the findings is specificity in terms of the impact of regulations in particular marketing channels: youth living in provinces with restrictions on billboards, television, and in-store and exterior ads were less likely to report noticing marketing in each of these channels. With these findings, we suggest that proposals in Canada and the United States to restrict e-cigarette advertising in these channels may be effective in reducing youth exposure. Several Canadian provinces and US states have also proposed that e-cigarette sales be restricted to adult-only stores, which is an alternative regulatory approach to reduce promotion to youth in retail settings, the most common source of marketing exposure. The findings are also consistent with previous evidence in which national-level differences in marketing regulations were examined: youth and adult exposure to e-cigarette advertising is lower in countries with greater restrictions, such as Australia and Canada before the “opening” of the e-cigarette market in 2018.^12,27 

This study is subject to limitations common to survey research, including response bias. For example, self-reported measures of marketing exposure may be subject to recall biases. In particular, the “overall” measure of marketing may have been interpreted differently across respondents and should, therefore, be interpreted within the context of the data for specific marketing channels. In previous studies, self-reported recall of tobacco marketing has been associated with objective data on the presence of marketing and audience viewership numbers measured using gross rating points.^32,33  Self-reported exposure to media campaigns is higher in markets in which a given campaign aired more frequently or in countries that permit greater marketing.^26,34–36  In the current study, we did not examine other factors that may have contributed to the increase in youth vaping. For example, the effects of marketing on vaping behavior may be mediated by the emergence of nicotine salt–based products with very high concentrations of nicotine, such as JUUL.³⁷  Whereas marketing may increase the likelihood that a young person tries an e-cigarette, nicotine delivery may promote the transition from experimental to regular use, as is the case for conventional cigarettes.^22,23  To our knowledge, there were no other notable differences in the e-cigarette market across provinces, including e-cigarette products, or policies such as taxation. Finally, participants were not recruited using probability-based sampling; therefore, the findings do not provide representative estimates. However, the same methodology was used across survey years, and the prevalence estimates are consistent with those from other national benchmark studies.³ 

Increased exposure to e-cigarette marketing in Canada is strongly associated with increases in youth vaping. With our findings, we provide empirical support for the effectiveness of more comprehensive regulatory restrictions that are being considered at the federal and provincial levels in Canada. Health Canada and each of the 10 provinces have proposed restrictions on e-cigarette marketing, including federal regulations that would align restrictions on e-cigarettes with those on tobacco products.³⁸  With the current findings, we highlight the importance of restrictions on newer forms of marketing, such as digital media, as well as traditional advertising channels, including the point-of-sale. The impact of such restrictions on e-cigarette use among adult smokers is unclear; however, it is notable that the prevalence of e-cigarette use among adult smokers has changed very little over the same period in which advertising exposure and prevalence of use among young people have increased.⁶  Therefore, marketing restrictions may have a relatively greater impact on reducing youth use than discouraging adult smokers from using e-cigarettes.

We acknowledge the coinvestigators of the ITC Youth Tobacco and Vaping Survey, all of whom contributed to the study design and measures: Maansi Bansal-Travers, PhD, Ron Borland, PhD, K. Michael Cummings, PhD, Geoffrey T. Fong, Maciej L. Goniewicz PhD, Bryan Heckman, PhD, Sara C. Hitchman, PhD, David Levy, PhD, Ann McNeill, PhD, Richard O’Connor, PhD, and James F. Thrasher, PhD.

AOR

adjusted odds ratio

CI

confidence interval

e-cigarette

electronic cigarette

ITC

International Tobacco Control Policy Evaluation Project

TVPA

Tobacco and Vaping Products Act