The Effect of E-cigarette Commercials on Youth Smoking: A Prospective Study

Abstract

Objectives:

In this study, we assessed whether commercials for electronic cigarettes (e-cigarettes) influence the use of e-cigarettes, cigarettes, and cigars among high-risk youth in southern California.

Methods:

We recruited students (N = 1060) from 29 alternative high schools into a prospective cohort study. We used multilevel Poisson regression models to examine whether exposure to e-cigarette commercials and perceptions of their appeal predicted increased use of e-cigarettes, cigarettes, and cigars one year later. We also tested the potential moderating effect of gender and ethnicity.

Results:

Models with and without covariates suggest that exposure to e-cigarette commercials is a statistically significant predictor of increased use of e-cigarettes. When gender was added to the models as a moderator, the relationships between commercial exposure and future use of e-cigarettes and cigars were found to be stronger among females. Unadjusted and adjusted models also indicated that students with favorable perceptions of e-cigarette commercials reported greater use of e-cigarettes, cigarettes, and cigars one year later.

Conclusions:

E-cigarette commercials may play an important role in persuading high-risk youth to use nicotine and tobacco products. Extending the Broadcast Advertising Ban of 1971 to include a broader range of products may be critical to preventing future generations from becoming addicted to nicotine.

In the 1960s, television commercials were the preeminent form of tobacco advertising.¹ Following the release of a report by the Surgeon General in 1964 that detailed the dangers of smoking,² the Federal Communications Commission (FCC) took steps in 1967 to protect the public by mandating that the persuasive power of cigarette commercials be counterbalanced by anti-smoking ads. Within a year, the 3 major networks had broadcast over 1300 anti-smoking commercials.³ By 1971, the Broadcast Advertising Ban ended cigarette advertising on television. Tobacco companies embraced this change as a preferable alternative to the relentless stream of counteradvertising.⁴

In the midst of the 2013 Super Bowl, a 30-second regional ad for NJOY e-cigarettes signaled the reemergence of television commercials promoting nicotine products. The ad instantly reached over 10 million viewers.⁵ It concluded with a statement that e-cigarettes contained nicotine and were ‘not for sale to minors.’⁶ No mention was made of the health risks. Research at the time reported adverse health effects in individuals who used e-cigarettes.^7–9 Subsequent research revealed the presence of potential carcinogens including formaldehyde, acetaldehyde, and acrolein.¹⁰ Researchers conducting a systematic review of 76 studies concluded that e-cigarettes could not be considered a safe product.¹¹ Emerging evidence also suggested that while no sidestream vapor was generated from e-cigarettes, vapor exhaled by users involuntarily exposed non-users to harmful compounds.^12–14

The Broadcast Advertising Ban had been implemented to protect the public from a dangerous product. E-cigarettes also posed risks,^11,15,16 but the same restrictions were not applied.¹⁷ Some argued that e-cigarettes deserved special consideration as a potentially safer alternative to traditional cigarettes.^18–21 Others argued that this was a strawman comparison because most products are a safer alternative to traditional cigarettes.^11,22 Moreover, the harm reduction argument presumed that individuals would only use e-cigarettes. In a meta-analysis involving 17,389 adolescents and young adults, youth who tried e-cigarettes were significantly more likely to try cigarettes.²³

Examining the Effect of E-cigarette Commercials on High-risk Youth

Research indicates that e-cigarettes have the potential to save lives.²⁴ However, the advertising strategy implemented by the nicotine and tobacco industry suggests a different priority. Commercials for e-cigarettes have been aired alongside television programs with substantial youth audiences.²⁵ An estimated 24 million youth have been exposed to imagery that glamorizes the use of nicotine products.²⁶ Given the high awareness of e-cigarettes among youth,²⁷ as well as evidence from experimental studies showing that youth exposed to e-cigarette advertising are more likely to use e-cigarettes,^28–30 it is not surprising that by 2019 an estimated 3 million high school students reported the use of e-cigarettes in the past 30 days.³¹

Although prior studies³² have examined the impact of tobacco advertising on adolescent smoking, to our knowledge, no study has examined the effects of e-cigarette commercials on high-risk youth in the United States (US). One such population that warrants investigation is students who are referred to alternative high schools due to poor academic performance, conduct problems, or extenuating life circumstances. Reports of past month use of traditional cigarettes among these students have ranged from 38.7% to 56.3%^33–37 compared to the national rate of 8.1%.³¹ Based on prior research,^38–43 it was hypothesized that the frequency with which these high-risk youth viewed e-cigarette commercials and their reaction to those commercials would predict future use of e-cigarettes, cigarettes, and cigars even after adjusting for multiple covariates commonly cited by the tobacco industry.⁴⁴

METHODS

Sampling

Using data obtained from the California Department of Education, we identified 183 eligible alternative high schools. We classified schools as eligible if they had at least 100 students and were within 100 miles of the program offices in Claremont, California. After receiving approval from the Claremont Graduate University Institutional Review Board, all schools were contacted in a randomly selected order and invited to participate in the study. Schools were accepted on a first-come, first-served basis until 29 sites were enrolled. Research staff visited the schools between October 14, 2014 and May 18, 2015. Interest forms were distributed to 6870 students who were in attendance at the schools. Completed forms were returned by 2726 students. Each student that returned a form was assigned to a specific staff member. The staff member obtained written consent and provided a link to a Web-based survey. Parental consent and youth assent were obtained for students under the age of 18. All students were given until September 1, 2015 to complete a 90-minute survey programmed with Inquisit 4 software (http://www.millisecond.com/). Data were gathered on the variables presented as well as a number of additional variables beyond the scope of this article.^45,46 A total of 1060 students took part in the initial assessment. Each of these students was given a $45 gift card to compensate them for their time.

We tracked students using procedures modeled on longitudinal studies conducted with high-risk populations.^47–50 Contact points included: (1) a reminder flyer mailed to each student 6 months prior to the one-year follow-up assessment; (2) a personalized text message delivered on the student’s birthday; (3) a scripted email exchange performed 3 months prior to the follow-up assessment; and (4) a scripted text message exchange initiated one month before the follow-up assessment. One-year follow-up assessments were conducted between September 21, 2015 and September 1, 2016. The average follow-up assessment took place 330 days (SD = 26.6) after the initial assessment. Most assessments (96.6%) were administered on a Web-enabled device utilizing Inquisit or Qualtrics (http://www.qualtrics.com). Students without access to a Web-enabled device (3.4%) were given the option to take a computer-assisted telephone interview. Each student that completed an assessment received a $50 gift card. The completion rate was 87.1%. Overall, 137 students did not complete a follow-up assessment due to withdrawal from the study (5.8%), incarceration (0.7%), or failure to respond to repeated contact attempts (93.5%).

Measures

Demographics.

We asked students to report their gender and ethnicity. Students also provided their birthdate which was used to calculate their age at the time of the initial assessment.

Exposure to e-cigarette commercials.

A single item utilized in prior research^51,52 was administered to quantify exposure to e-cigarette commercials. We asked students: About how often did you see an electronic cigarette commercial in the last 6 months? Response options included ‘Never’, ‘Less than once a month’, ‘Once a month’, ‘2–3 times a month’, ‘Once a week’, ‘2–6 times a week’, and ‘Every day’.

Likeability of e-cigarette commercials (α = .77).

We used a 5-item scale modeled on prior studies^53–56 to gauge the extent to which students liked the e-cigarette commercials they had seen. The first 3 questions were: When you see electronic cigarette commercials on TV or online…, ‘Do you think they are funny? ‘, ‘Do you think they are sexy? ‘, ‘Do you wish you were like the people in the commercials?’ Response options for all 3 items were: ‘No, never’, ‘No, usually not’, ‘Yes, usually’, and ‘Yes, always’. The next question was: When you see electronic cigarette commercials, how often do you pay attention to them? Response options included: ‘Never’, ‘Some of the time’, ‘Most of the time’, and ‘Always’. The final question was: Of all the commercials you see, how much do you like electronic cigarette commercials? Students responded on a 4-point scale ranging from ‘I like electronic cigarette commercials the least’ to ‘I like electronic cigarette commercials the most.’

Exposure to other forms of advertising for nicotine and tobacco products (α = .81).

To account for the influence of other advertising channels, a 4-item scale was adapted from the National Youth Tobacco Survey (NYTS).⁵⁷ The scale assessed the extent to which students had been exposed to: (1) newspaper and magazine ads; (2) posters and signs; (3) radio spots; and (4) Web banners for nicotine and tobacco products. Response options included: ‘None’, ‘1–3 times in the past 30 days’, ‘1–3 times per week’, ‘Daily or almost daily’, and ‘More than once a day’.

Family use of nicotine and tobacco products.

Numerous studies have documented the impact of family tobacco use on youth.^58–62 To adjust for this effect, 3 items were adapted from the NYTS to determine whether students had at least one family member who currently used: (1) cigarettes; (2) e-cigarettes, vaporizers, or vape pens; or (3) cigars, cigarillos, or little cigars.

Peer use of nicotine and tobacco products.

The influence of peers on youth tobacco use is another well-documented finding.^63–67 To adjust for the effect of peer use of nicotine and tobacco products, 3 items were adapted from the California Student Tobacco Survey⁶⁸ to measure whether students had at least one friend who currently used cigarettes, e-cigarettes, vaporizers, or vape pens, or cigars, cigarillos, or little cigars.

Use of nicotine and tobacco products in the past 30 days.

To facilitate comparisons across products, a previously validated drug use questionnaire^69,70 was modified to inquire about cigarettes, e-cigarettes, vaporizers, or vape pens, and cigars, cigarillos, or little cigars. Students were asked how many times they had used each product in the past 30 days. Response options included: ‘0 times’, ‘1–10 times’, ‘11–20 times’, ‘21–30 times’, ‘31–40 times’, ‘41–50 times’, ‘51–60 times’, ‘61–70 times’, ‘71–80 times’, ‘81–90 times’, and ‘91+ times’. The response provided at the one-year follow-up assessment was used as the dependent variable and the responses for each product at the initial assessment were integrated as covariates.⁷¹

Data Analysis

The sample consisted of 1060 students nested within 29 schools. Calculating the intra-class correlation revealed that 5.5% to 11.7% of the variance in the use of e-cigarettes, cigarettes, and cigars could be attributed to clustering effects at the school level. Two approaches were considered to account for similarities between students attending the same school.⁷² The first approach involved the use of generalized estimating equations to model population-averaged responses. The second approach utilized generalized linear mixed models to estimate subject-specific responses. Based on prior research indicating that marginal models have inflated type 1 error rates when analyzing datasets with a relatively small number of clusters (ie, schools),^73–75 the decision was made to use generalized linear mixed models. Because the dependent variables exhibited a count distribution and consisted of only non-negative integers, a multilevel Poisson regression was used in all analyses. Continuous independent variables included in the Poisson regression models were group-mean centered.

A comparison between participants with complete data versus those with missing responses revealed no statistically significant differences by gender (45.8% male vs 55.7% male, p = .916), ethnicity (78.1% Hispanic vs 72.1% Hispanic, p = .956), age (17.5 vs 17.5, p = .550), exposure to e-cigarette commercials (1.73 vs 1.80, p = .610), or the perceived likeability of e-cigarette commercials (0.54 vs 0.57, p = .483). Use of e-cigarettes (0.53 vs 0.56, p = .779), cigarettes (0.43 vs 0.54, p = .412), and cigars (0.27 vs 0.37, p = .362) in the past 30 days at the initial assessment were also not significantly different. Based on this preliminary analysis, missing responses were assumed to be missing at random and a multiple imputation analysis was performed using SAS PROC MI and MIANALYZE.⁷⁶

Models that assessed whether exposure to e-cigarette commercials predicted product use in the past 30 days at the one-year follow-up assessment were estimated utilizing SAS PROC GLIMMIX. The first set of models analyzed raw data and used LaPlace’s method⁷⁷ to assess the unadjusted effect of exposure. The second set of models analyzed forty imputed datasets⁷⁸ and incorporated multiple covariates including gender (female = 0 vs male = 1), ethnicity (Non-Hispanic = 0 vs Hispanic = 1), age, family use of nicotine and tobacco products (nonuse = 0 vs use = 1), peer use of nicotine and tobacco products (nonuse = 0 vs use = 1), use of nicotine and tobacco products in the past 30 days at the initial assessment, and exposure to promotions for nicotine and tobacco products through other advertising channels. A third set of models tested whether ethnicity or gender moderated the relationships between exposure to e-cigarette commercials at the initial assessment and the use of nicotine and tobacco products at the one-year follow-up assessment.

The next set of analyses was restricted to a sub-sample of 720 students who had seen at least one e-cigarette commercial. The first set of models used raw data to estimate the unadjusted effect of e-cigarette commercial exposure and likeability. The second set of models incorporated the same covariates utilized previously. The third set of models examined whether there was a statistically significant interaction between the likeability of e-cigarette commercials and either ethnicity or gender.

RESULTS

Descriptive Statistics

The sample was 50.7% male and 75.2% Hispanic as Table 1 shows. The mean age was 17.5 years (SD = 0.9). Compared to high school students in California,⁷⁹ students in the current sample reported greater use of e-cigarettes (19.8% vs 8.6%, p < .001), cigarettes (15.9% vs 4.3%, p < .001), and cigars (11.9% vs 4.3%, p < .001) in the past 30 days. Students attending alternative high schools also reported greater use of 2 or more nicotine or tobacco products in the past 30 days (13.1% vs 6.1%, p < .001).

Table 1.

Descriptive Statistics for 1060 Students from 29 Alternative High Schools

Initial Assessment
Gender, N (%)
Male	534 (50.7%)
Female	520 (49.3%)
Ethnicity, N (%)
Hispanic	777 (75.2%)
Non-Hispanic	256 (24.8%)
Age, Mean (SD)	17.5 (0.9)
Exposure to E-cigarette Commercials, Mean (SD)	1.8 (1.9)
Likeability of E-cigarette Commercials, Mean (SD)	0.6 (0.5)
Exposure to other Forms of NTP Advertising, Mean (SD)	1.2 (1.0)
Family Use of NTPs, N (%)
E-cigarettes	150 (15.3%)
Cigarettes	283 (29.7%)
Cigars	102 (10.4%)
Peer Use of NTPs, N (%)
E-cigarettes	435 (52.8%)
Cigarettes	435 (51.1%)
Cigars	230 (24.1%)
Use of NTPs in the Past 30 Days, N (%)
E-cigarettes	198 (19.8%)
Cigarettes	160 (15.9%)
Cigars	119 (11.9%)
Only One Product	152 (15.4%)
2 Products	73 (7.4%)
All 3 Products	56 (5.7%)
One Year Follow-uv Assessment
Use of NTPs in the Past 30 Days, N (%)
E-cigarettes	124 (14.2%)
Cigarettes	134 (15.3%)
Cigars	88 (10.1%)
Only One Product	110 (12.8%)
2 Products	58 (6.7%)
All 3 Products	35 (4.1%)

Note.

NTPs = Nicotine and tobacco products

More than two-thirds of the students (67.9%) indicated they had seen at least one e-cigarette commercial. More than one-third (39.2%) had seen 2 or more commercials in the past month. Among the 720 students who had seen a commercial, 68.1% reported watching the commercial on television while 30.3% reported seeing it online.

Exposure to E-cigarette Commercials and Product Use One Year Later

The first set of models estimated the unadjusted effect of exposure to e-cigarette commercials. Table 2 presents the unstandardized coefficients (b), standard errors (SE), and p-values. A one unit change in exposure to e-cigarette commercials was associated with a 21.8% increase in the number of times students used e-cigarettes one year later (b = 0.20, SE = 0.03, p < .001), a 10.0% increase in the number of times students smoked cigarettes (b = 0.10, SE = 0.02, p < .001), and a 10.1% increase in the number of times students smoked cigars (b = 0.10, SE = 0.03, p = .001).

Table 2.

Multilevel Poisson Regression Models Examining the Relationship between E-cigarette Commercial Exposure and Product Use One Year Later Among 1060 Alternative High School Students

Variables	E-cigarettes			Cigarettes			Cigars
	b	SE	p	b	SE	p	b	SE	p
	Unadjusted Models
Exposure to E-cigarette Commercials	0.20	0.03	<.001	0.10	0.02	<.001	0.10	0.03	.001
	Covariate Adjusted Models
Gender (Male)	0.76	0.13	<.001	0.37	0.10	<.001	0.81	0.13	<.001
Ethnicity (Hispanic)	−0.35	0.13	.010	−0.53	0.10	<.001	−0.59	0.13	<.001
Age	0.01	0.07	.856	0.01	0.06	.925	−0.11	0.07	.123
Family Use of E-cigarettes	0.25	0.17	.144	0.24	0.13	.074	0.21	0.17	.214
Family Use of Cigarettes	−0.15	0.16	.368	0.28	0.12	.022	−0.05	0.17	.777
Family Use of Cigars	0.15	0.22	.491	−0.02	0.15	.893	0.31	0.19	.100
Peer Use of E-cigarettes	0.02	0.19	.915	−0.21	0.14	.155	0.30	0.19	.121
Peer Use of Cigarettes	0.39	0.18	.032	0.55	0.19	.004	−0.36	0.19	.066
Peer Use of Cigars	−0.16	0.14	.265	0.24	0.13	.074	0.61	0.15	<.001
Previous Use of E-cigarettes	0.17	0.02	<.001	−0.02	0.02	.328	−0.03	0.03	.318
Previous Use of Cigarettes	0.03	0.03	.178	0.22	0.02	<.001	0.16	0.03	<.001
Previous Use of Cigars	−0.23	0.06	<.001	−0.02	0.03	.567	0.06	0.03	.052
Exposure to other Tobacco Advertising	0.10	0.06	.125	0.09	0.05	.089	0.23	0.06	<.001
Exposure to E-cigarette Commercials	0.07	0.03	.024	0.02	0.03	.412	−0.01	0.03	.742

The second set of models integrated multiple covariates. Use of nicotine and tobacco products in the past 30 days at the initial assessment was a statistically significant predictor in most of the models. Other variables, such as gender and ethnicity, were also relevant. After accounting for these covariates, exposure to e-cigarette commercials remained a statistically significant predictor of e-cigarette use (b = 0.07, SE = 0.03, p = .024) but not cigarette use (b = 0.02, SE = 0.03, p = .412) or cigar use (b = −0.01, SE = 0.03, p = .742). A third set of analyses that examined the moderating effect of ethnicity revealed that the interaction between ethnicity and exposure to e-cigarette commercials was not statistically significant in models estimating the use of e-cigarettes (b = 0.11, SE = 0.06, p = .056), cigarettes (b = 0.08, SE = 0.05, p = .143), or cigars (b = 0.07, SE = 0.07, p = .294). However, gender had a moderating effect on the relationship between e-cigarette commercial exposure and the use of e-cigarettes (b = −0.17, SE = 0.07, p = .011) and cigars (b = −0.22, SE = 0.07, p = .003). Table 3 presents parameter estimates and Figure 1 provides a visualization of the interaction. The interaction between exposure to e-cigarette commercials and gender was not statistically significant in models estimating the use of cigarettes (b = −0.04, SE = 0.05, p = .446) at the one-year follow-up assessment.

Table 3.

Multilevel Poisson Regression Models Examining the Moderating Effect of Gender on the Relationship between E-cigarette Commercial Exposure and Product Use One Year Later among 1060 Alternative High School Students

Variables	E-cigarettes			Cigarettes			Cigars
	b	SE	p	b	SE	p	b	SE	p
Gender (Male)	0.84	0.14	<.001	0.38	0.10	<.001	0.88	0.14	<.001
Ethnicity (Hispanic)	−0.33	0.13	.015	−0.53	0.10	<.001	−0.59	0.13	<.001
Age	0.01	0.07	.860	0.01	0.06	.932	−0.12	0.07	.108
Family Use of E-cigarettes	0.24	0.17	.162	0.24	0.13	.078	0.18	0.17	.275
Family Use of Cigarettes	−0.14	0.16	.395	0.28	0.12	.022	−0.05	0.17	.783
Family Use of Cigars	0.20	0.22	.353	−0.01	0.15	.960	0.38	0.19	.043
Peer Use of E-cigarettes	0.02	0.19	.904	−0.21	0.14	.142	0.34	0.19	.085
Peer Use of Cigarettes	0.40	0.18	.026	0.55	0.19	.004	−0.37	0.20	.058
Peer Use of Cigars	−0.20	0.14	.167	0.24	0.13	.079	0.58	0.15	<.001
Previous Use of E-cigarettes	0.18	0.02	<.001	−0.02	0.02	.352	−0.03	0.03	.327
Previous Use of Cigarettes	0.03	0.03	.209	0.22	0.02	<.001	0.16	0.03	<.001
Previous Use of Cigars	−0.23	0.06	<.001	−0.02	0.03	.570	0.06	0.03	.046
Exposure to other Tobacco Advertising	0.08	0.06	.204	0.09	0.05	.107	0.22	0.06	<.001
Exposure to E-cigarette Commercials	0.19	0.06	.001	0.05	0.04	.255	0.14	0.06	.012
Gender X Commercial Exposure	−0.17	0.07	.011	−0.04	0.05	.446	−0.22	0.07	.003

Figure 1.

Visual Depiction of How Gender Moderates the Effect of E-cigarette Commercial Exposure on Product Use One Year Later among 1060 Alternative High School Students

Note.

Poisson regression models estimate a multiplicative effect characterized by a logarithmic curve.

Likeability of E-cigarette Commercials and Product Use One Year Later

Within the subsample of students who had seen at least one e-cigarette commercial, exposure to e-cigarette commercials was a statistically significant predictor of future use of e-cigarettes (b = 0.16, SE = 0.04, p <.001), cigarettes (b = 0.13, SE = 0.03, p <.001), and cigars (b = 0.09, SE = 0.04, p = .002) as Table 4 shows. Likeability also was a statistically significant predictor of increased use of e-cigarettes (b = 0.91, SE = 0.10, p < .001), cigarettes (b = 0.71, SE = 0.10, p < .001), and cigars (b = 0.80, SE = 0.10, p < .001). After adding multiple covariates to the models, exposure remained a significant predictor of increased use of e-cigarettes (b = 0.11, SE = 0.04, p = .004) and cigarettes (b = 0.10, SE = 0.04, p = .008), whereas likeability remained a statistically significant predictor of increased use of e-cigarettes (b = 0.44, SE = 0.12, p < .001), cigarettes (b = 0.23, SE = 0.10, p = .031), and cigars (b = 0.47, SE = 0.13, p < .001).

Table 4.

Multilevel Poisson Regression Models Examining the Relationship between E-cigarette Commercial Likeability and Product Use One Year Later among A Subsample of 720 Alternative High School Students Who Had Seen at Least One E-cigarette Commercial

Variables	E-cigarettes			Cigarettes			Cigars
	b	SE	p	b	SE	p	b	SE	p
	Unadjusted Models
Exposure to E-cigarette Commercials	0.16	0.04	<.001	0.13	0.03	<.001	0.09	0.04	.019
Likeability of E-Cigarette Commercials	0.91	0.10	<.001	0.71	0.10	<.001	0.80	0.10	<.001
	Covariate Adjusted Models
Gender (Male)	0.66	0.16	<.001	0.26	0.12	.023	0.48	0.15	.002
Ethnicity (Hispanic)	−0.40	0.16	.013	−0.40	0.13	.002	−0.48	0.16	.003
Age	0.02	0.08	.831	0.01	0.07	.889	−0.12	0.09	.188
Family Use of E-cigarettes	0.13	0.17	.447	0.22	0.14	.113	0.13	0.18	.473
Family Use of Cigarettes	−0.20	0.17	.262	0.31	0.13	.015	−0.03	0.18	.871
Family Use of Cigars	0.27	0.22	.210	0.02	0.17	.920	0.35	0.21	.100
Peer Use of E-cigarettes	0.09	0.26	.722	−0.05	0.18	.793	0.19	0.22	.379
Peer Use of Cigarettes	0.25	0.23	.278	0.64	0.24	.008	−0.32	0.24	.175
Peer Use of Cigars	−0.21	0.17	.203	0.14	0.15	.339	0.44	0.17	.011
Previous Use of E-cigarettes	0.15	0.02	<.001	0.03	0.02	.211	−0.03	0.03	.374
Previous Use of Cigarettes	0.08	0.03	.010	0.21	0.02	<.001	0.16	0.03	<.001
Previous Use of Cigars	−0.16	0.05	.003	−0.01	0.04	.759	0.10	0.04	.003
Exposure to other Tobacco Advertising	0.17	0.07	.013	0.13	0.06	.029	0.22	0.07	.002
Exposure to E-cigarette Commercials	0.11	0.04	.004	0.10	0.04	.008	0.00	0.04	.910
Likeability of E-cigarette Commercials	0.44	0.12	<.001	0.23	0.10	.031	0.47	0.13	<.001

A one unit change in exposure to e-cigarette commercials was associated with an 11.6% increase in the number of times students used e-cigarettes and a 10.0% increase in the number of times students smoked cigarettes in the past 30 days. A one unit change in likeability was associated with a 56.0% increase in the number of times students used e-cigarettes, a 25.2% increase in the use of cigarettes, and a 60.7% increase in the use of cigars in the past 30 days. The interaction between ethnicity and the likeability of e-cigarette commercials was not statistically significant in models predicting the use of e-cigarettes (b = −0.08, SE = 0.23, p = .734), cigarettes (b = 0.08, SE = 0.21, p = .714), or cigars (b = 0.13, SE = 0.25, p = .594). Similarly, gender did not moderate the relationship between e-cigarette commercial likeability and the use of e-cigarettes (b = −0.25, SE = 0.27, p = .352), cigarettes (b = 0.13, SE = 0.21, p = .531), or cigars (b = −0.15, SE = 0.28, p = .592).

DISCUSSION

E-cigarette Commercials and the Use of Multiple Nicotine and Tobacco Products

We explored the relationship between e-cigarette commercials and the use of nicotine and tobacco products in high-risk youth in southern California. In the unadjusted models, exposure to e-cigarette commercials was associated with increased use of e-cigarettes, cigarettes, and cigars one year later. After adding covariates to the model, only the relationship between commercial exposure and the use of e-cigarettes remained statistically significant. However, a moderating effect also was detected, which indicated that the relationship between commercial exposure and the use of e-cigarettes and cigars was stronger among females. Within the subsample of students who had viewed at least one e-cigarette commercial, repeated exposure was associated with increased use of e-cigarettes and cigarettes in both the unadjusted and adjusted models. Moreover, the degree to which students liked the commercials predicted an increase in the use of e-cigarettes, cigarettes, and cigars one year later in models with and without covariates. Taken together, these findings suggest that e-cigarette commercials may play an important role in persuading high-risk youth to use nicotine and tobacco products. This interpretation aligns with prior research indicating that e-cigarette advertising increases susceptibility to smoking in youth^41,42 and that likeable commercials influence youth behavior.^52,53 The moderating effect of gender that we found and supported by other studies^80–82 is particularly troubling given that females metabolize nicotine faster than males,⁸³ which may make subsequent attempts at cessation more challenging.^84–85

Although the use of traditional cigarettes among youth has declined, dual-use and poly-use have been rising since 2011.⁸⁶ The timing is noteworthy given that whereas e-cigarettes were introduced to US consumers in 2006,⁸⁷ advertising expenditures remained relatively low until 2011⁸⁸ after which they increased 10-fold.⁸⁹ It was only after this increase that e-cigarette use among students who had not previously used tobacco products rose from approximately 79,000 in 2011 to more than 263,000 in 2013.⁹⁰ Past research shows a dose-response relationship between exposure to pro-tobacco advertising channels and youth experimentation with alternative nicotine and tobacco products.^91–93 The addition of a new product (e-cigarettes) promoted through a new medium (commercials) may be a key factor contributing to recent estimates indicating that over 1.6 million youth have used 2 or more nicotine and tobacco products in the past 30 days.³¹

Limitations

Conclusions derived from the current investigation must be weighed against threats to validity⁹⁴ such as mono-method bias and maturation. Among smokers, self-report measures have been known to produce imprecise estimates of cigarette use.^95–97 Recall bias may have similarly resulted in inaccurate reports of the use of cigarettes, e-cigarettes, and cigars in the past 30 days. Prior longitudinal investigations also have revealed that youth tobacco use progression is associated with increased receptivity to tobacco advertising⁹⁸ as well as greater amounts of time spent watching television.^99,100 Although randomized experiments^{29,30,101–103} suggest a causal mechanism in which e-cigarette advertising encourages youth to use nicotine and tobacco products, it also may be the case that reported increases in the use of e-cigarettes, cigarettes, and cigars are being driven by a natural progression in the use of addictive products and that the higher levels of e-cigarette commercial exposure and receptivity observed in the current investigation are incidental.

Another noteworthy limitation is that the sample was predominantly Hispanic and restricted to students attending alternative high schools in southern California, which limits the generalizability of the findings. Future research should test whether the reported effects can be replicated in national and international longitudinal datasets such as those provided through the Population Assessment of Tobacco and Health Study and the International Tobacco Control Policy Evaluation Project. It is also worth emphasizing that whereas the current investigation highlighted the importance of commercial likeability it did not delineate the types of content that youth find appealing. Focus groups have revealed the attractive qualities of novel tobacco products.^104,105 Similar qualitative research should be conducted to determine the types of commercials youth classify as engaging.

Policy Implications

Despite the limitations, findings from the study suggest that policies should be developed to protect high-risk youth from the influence of e-cigarette commercials utilizing evidence-based¹⁰⁶ frameworks such as the World Health Organization Framework Convention on Tobacco Control^107–109 (WHO FCTC). A sensible approach described in Article 13 of the WHO FCTC would be to extend the Broadcast Advertising Ban to include a broader range of nicotine and tobacco products.^110,111 Such policies have been shown to be effective at reducing smoking^112,113 and have been implemented in multiple countries across the globe,¹¹⁴ including all nations within the European Union.¹¹⁵ Unfortunately, this form of government intervention is likely to be challenged on the grounds of free speech in the US.^116,117 Consequently, policymakers may wish to support counteradvertising campaigns in accordance with Article 12 of the WHO FCTC. Numerous studies^118–126 have demonstrated the efficacy of mass media interventions including those that specifically target adolescent females.¹²⁷ In light of the moderating effect of gender that we detected, both prevention and cessation programs tailored toward young women may help reduce the use of nicotine and tobacco products among high-risk youth. It may also be prudent to require the Federal Trade Commission to track advertising expenditures for e-cigarettes given the known correlation between the launch of tobacco marketing campaigns and gender-specific increases in youth smoking.^128,129

Article 6 of the WHO FCTC suggests that raising the price of nicotine and tobacco products through taxation may be an effective deterrent to youth smoking. This approach is strongly supported in California,¹³⁰ has been implemented in multiple locations throughout the US,¹³¹ and is buttressed by prior research^132–134 including a study indicating that Hispanics, females, and low-SES populations are especially responsive to increases in the price of tobacco products.¹³⁵ Increasing the cost of cigars may be especially powerful given evidence indicating these are some of the cheapest tobacco products sold near alternative high schools,⁴⁵ and that the sale of cigars may be escalating following the passage of the Family Smoking Prevention and Tobacco Control Act.¹³⁶ An additional course of action recommended by Article 16 of the WHO FCTC and supported by extensive research^137–140 is to increase the minimum age at which youth can purchase nicotine and tobacco products. Public support for these laws is broad¹⁴¹ and nearly half of the states and over 400 cities within the country have already raised the minimum age for tobacco purchases to 21.¹⁴² Applying the same restriction to e-cigarettes may be a politically feasible way to combat the current vaping epidemic.¹⁴³

As policymakers consider these approaches, they also may find it necessary to reframe the conversation surrounding commercials for e-cigarettes. Although the debate about whether e-cigarettes represent a safer alternative to traditional cigarettes is warranted, it should not overshadow the discussion about whether advertising that could persuade youth to use hazardous products should be disseminated without limitations. Although society may benefit by encouraging existing smokers to switch to a less harmful alternative, such benefits must be balanced against the cost of permitting a new generation of youth to become addicted to nicotine and tobacco products.