Online Tobacco Marketing Among US Adolescent Sexual, Gender, Racial, and Ethnic Minorities

Abstract

Introduction:

The tobacco industry has previously targeted sexual/gender and racial/ethnic minorities with focused campaigns in traditional, offline marketing. We assess whether these populations report more engagement with online tobacco marketing compared with heterosexual and non-Hispanic white youth.

Methods:

Data were from 8015 adolescents sampled between 2014–2015 in the nationally-representative Population Assessment for Tobacco and Health (PATH) Study. Engagement with online tobacco marketing within the past year was assessed through eight forms of engagement. A weighted logistic regression model was fit with engagement as outcome and socio-demographic and psychosocial characteristics, internet-related and substance use behavior, tobacco-related risk factors, tobacco use status, and prior engagement with online tobacco marketing as covariates.

Results:

Accounting for other covariates including tobacco use status and prior engagement with online tobacco marketing, the odds of past-year engagement were higher for sexual minority males (aOR=1.57; 95% CI: 1.05–2.35) compared to straight males and higher for sexual minority females (aOR= 1.45; 95% CI: 1.13–1.87) compared to straight females. The odds of past-year engagement were also higher for Hispanics (aOR=1.31; 95% CI: 1.11–1.56) and non-Hispanic Blacks (aOR=1.42; 95% CI: 1.14–1.77) compared to non-Hispanic Whites.

Conclusions:

Sexual/gender and and racial/ethnic minority youth reported higher engagement with online tobacco marketing than their heterosexual and non-Hispanic white peers, respectively.

1. INTRODUCTION.

The tobacco industry has long targeted sexual, gender, racial, and ethnic minority adolescents and adults with focused marketing, advertising campaigns, and economic support of minority advocacy organizations [1–4]. This targeting has likely contributed to higher rates of specific tobacco product use among racial and ethnic minorities (e.g., mentholated cigarettes among non-Hispanic Black adolescents) and overall tobacco and e-cigarette use among sexual and gender minorities compared to their peers [5–7]. For example, 29.8% of bisexual youth and 25.6% of gay and lesbian youth reported tobacco use within the past 30 days in 2013–2014 compared to 11.8% of straight youth.[8] Since the 1998 Master Settlement Agreement, the tobacco industry has shifted away from traditional marketing to internet-based marketing, in part, because the latter is less regulated [9]. Internet-based marketing allows tailored direct-to-consumer marketing that moves beyond passive exposure and instead facilitates engagement in promotional activities and interaction among potential customers [10,11].

As the level of engagement has increased substantially among all adolescents between 2013 and 2015 [12], this rise may be especially problematic in minority populations if they are disproportionately affected as they have been with traditional, offline marketing. In addition to higher risk of exposure to traditional forms of tobacco marketing, sexual and gender minority (SGM) adults are more likely to be exposed to and interact with tobacco marketing on social media [13]. Yet, it is not known if SGM youth, like SGM adults, and racial and ethnic minority youth also face higher risks of engaging with online tobacco marketing, such as signing up to receive e-mails or watching videos about tobacco products. Such engagement is problematic because it may increase the risk of tobacco use initiation, increase the frequency of tobacco use, and decrease the likelihood of tobacco use cessation [14].

Our study addresses this knowledge gap by estimating the prevalence of engagement with online tobacco marketing within the past year among racial, ethnic, and sexual minority adolescents and transgender adolescents. If these minority populations—especially non-tobacco users—are more likely to engage in online tobacco marketing than their majority-group peers, existing disparities in tobacco use could widen. Our study also assesses other socio-demographic, psychosocial, and environmental risk factors for engagement with online tobacco marketing within the past year to identify additional vulnerable subpopulations. Knowledge of the risk factors for engagement with online tobacco marketing could help focus public health campaigns that counter its influence.

2. METHODS.

2.1. Data.

The Population Assessment for Tobacco and Health (PATH) Study is a nationally representative, longitudinal cohort study conducted by the National Institute on Drug Abuse and the Food and Drug Administration’s Center for Tobacco Products. Our analysis primarily utilized data from Wave 2 (2014–2015) of the PATH Study, Youth Interview, Restricted-Use File. Our analysis retrospectively linked engagement with online tobacco marketing measured at Wave 1 (2013–2014).

At Wave 2, the PATH study sampled 12,172 adolescents: 4157 younger adolescents (12–13 year olds) and 8015 older adolescents (14–17 year olds). The younger adolescents were excluded from the analysis because the PATH Study did not assess their sexual orientation or being transgender. Of the 8015 older adolescents, 8012 respondents were previously sampled in Wave 1 and 3 respondents were newly sampled in Wave 2. Those three respondents were excluded from the analysis because they did not have baseline Wave 1 data on engagement with online tobacco marketing. The final sample consisted of the 8012 14–17 year olds sampled in both Wave 1 and Wave 2. Among households that were screened at Wave 2, the overall weighted response rate was 87.3%. For further details about the PATH Study, see Hyland et al [15]. The PATH Study created population and replicate weights that adjusted for complex study design characteristics (e.g., oversampling at Wave 1) and nonresponse at Waves 1 and 2. Combined with the use of a probability sample, the weights enable analyses of the PATH Study data to produce robust estimates that are representative of the non-institutionalized, civilian U.S. population ≥12 years [15].

2.2. Outcome.

Engagement with online tobacco marketing within the past year at Wave 2 was based on eight forms of engagement assessed in PATH: [1] signing up for any email alerts about tobacco products, including e-cigarettes, in past year; [2] reading any articles online about tobacco products, including e-cigarettes, in past year; [3] watching a video online about tobacco products, including e-cigarettes, in past year; [4] liking or following Camel, Marlboro, Newport, Swisher Sweets, Blu, Fin, Vuse, NJOY on Facebook, Twitter or other social media sites; [5] sending a link or information about Camel, Marlboro, Newport, Swisher Sweets, Blu, Fin, Vuse, NJOY to others on Facebook, Twitter or other social media sites in past year; [6] playing an online game related to Camel, Marlboro, Newport, Swisher Sweets, Blu, Fin, Vuse, NJOY in past year [7] receiving any discount coupons or promotions for tobacco products or e-cigarettes by email, the internet, social networking sites, or a text message in past 30 days; and [8] receiving any information from a tobacco company, other than discount coupons or promotions, by email, the internet, social networking sites, or a text message in past 30 days. A respondent was considered to have engaged in online tobacco marketing within the past year if she or he responded affirmatively to at least one form of engagement.

2.3. Primary Variables of Interest.

The primary variables of interest were race/ethnicity (Hispanic, non-Hispanic Black, non-Hispanic Other, and non-Hispanic White), sexual orientation, and being transgender. The PATH Study ascertained sex through the question “What is your sex”, to which respondents could answer “female” or “male” or not answer. The PATH Study ascertained sexual orientation through the question “Do you consider yourself to be straight, lesbian or gay, bisexual, or something else?”. The PATH Study ascertained being transgendered through the question “Do you consider yourself to be transgender?”. Respondents who answered “yes” to this question were considered transgender regardless of their response or non-response to the survey questions about sex and sexual orientation. Respondents who answered “lesbian or gay”, “bisexual”, or “something else” to the question about sexual orientation and “female” to the question about sex were categorized as sexual minority females. Similarly, respondents who answered “lesbian or gay”, “bisexual”, or “something else” to the question about sexual orientation and “male” to the question about sex were categorized as sexual minority males. Respondents who answered “straight” to the question about sexual orientation were categorized as straight females and males (based on their answer to the question about sex).

2.4. Covariates.

Other socio-demographic characteristics of respondents included age and parental education. Psychosocial characteristics included mental health status, which was assessed by the level of internalizing (e.g., feeling very trapped, lonely, sad, blue, depressed, or hopless about the future in the past year) and externalizing (e.g., had a hard time playing attention at school, work, or home two or more times in the past year) problems based on the sum of four and five items of the Global Appraisal of Individual Needs-Short Screener, respectively. Internet-related behavior included the frequency of use of social networking sites and of regular use of smart phones. Substance use behavior included past 30-day binge alcohol drinking, past-year marijuana use, and past-year illicit and non-prescription drug use. Other potential risk factors for tobacco use included close contact with a smoker within the past week; living with anyone who currently used tobacco; receipt of tobacco discount coupons through the mail; receipt of tobacco-related information through the mail; weekly income from a job, family, or allowance; and school performance. Respondents were categorized into the following tobacco use status: non-susceptible never tobacco user, susceptible never tobacco user, ever tobacco user but not within the past year, and ever tobacco user and within the past year. Susceptibility to tobacco use among respondents who had never used tobacco was based on respondent intentionto use a tobacco product soon, willingness to try a product if offered by a friend, and curiosity about using a product. Finally, prior engagement with online tobacco marketing was based onsix forms of engagement measured at Wave 1 including: [1] ever signing up for email alerts, reading articles, or watching videos about tobacco products; [2] ever liking or following tobacco brands on social media sites; [3] ever sending links or information about tobacco brands on social media sites; [4] ever playing online games about tobacco products; [5] ever receiving discount coupons or promotions for tobacco products electronically; and [6] ever receiving tobacco-related information about tobacco products electronically (see Appendix Table A.1 for details on all covariates).

2.5. Analyses.

First, the weighted prevalence of characteristics was estimated among all respondents included in the study. Then, the weighted prevalence of engagement with online tobacco marketing within the past year was estimated by sexual orientation and being transgender and race/ethnicity. The weighted prevalence of individual forms of engagement was also estimated by sexual orientation, being transgender, and race/ethnicity. Spearman’s rank correlation coefficient was calculated for the rank of individual forms of engagement between each pair of sub-populations (e.g., sexual minority females and sexual minority males). Next, weighted logistic regression models were fit to engagement with online tobacco marketing within the past year as the outcome and sexual orientation, being transgender, and race/ethnicity as the primary variables of interest. Covariates included other socio-demographic and psychosocial characteristics, internet-based and substance use behaviors, other potential risk factors for tobacco use, tobacco use status, and baseline engagement with online tobacco marketing at Wave 1. Sets of covariates were added sequentially to assess how they affected the association between the primary variables of interest and past-year engagement. Model 1 included sexual orientation, being transgender, and race/ethnicity. Model 2 added other socio-demographic characteristics. Model 3 additionally added psychosocial characteristics and internet-related behavior. Model 4 additionally added substance use behavior, other potential risk factors for tobacco use, and tobacco use status. Finally, model 5 additionally added ever engagement with online tobacco marketing at Wave 1. The sets of covariates were added sequentially because previous studies identified them as risk factors for exposure to tobacco marketing [16]. The models were fit with straight males as the reference category to allow comparison between sexual minority males and straight males. The models were then fit with straight females as the reference category to allow comparison between sexual minority females and straight females.

All analyses were conducted in R (version 3.4.3; The Comprehensive R Archive Network) and utilized balanced repeated replication weights with Fay’s correction (shrinkage factor set at 0.3) to account for sampling and non-response in the PATH Study. The Dartmouth College Committee for the Protection of Human Subjects determined that the regulatory definition of human subjects research (45 CFR 46.102[f]) did not apply to this study and, therefore, the study was exempted from institutional review board review.

3. RESULTS.

3.1. Study Population.

The population was distributed approximately equally by age (Table 1). The population consisted of approximately 1.0% transgender adolescents, 6.6% sexual minority females, 2.4% sexual minority males, 41.2% straight females, and 48.8% straight males. By race/ethnicity, the population consisted of appropoximately 22.5% Hispanics, 13.5% non-Hispanic Blacks, 9.0% non-Hispanic other/multi-racial adolescents, and 55.0% non-Hispanic Whites. An estimated 81.8% used social networking sites at least once a day. An estimated 38.3% were in close contact with tobacco users with the past week, 30.0% lived with tobacco users, 1.6% received tobacco discount coupons through the mail, and 0.6% received other tobacco-related information through the mail. An estimated 37.9% were non-susceptible never tobacco users, 31.9% were susceptible never tobacco users, 8.0% were ever tobacco users but not within the past year, and 22.1% were tobacco users within the past year. Finally, an estimated 9.4% had engaged with online tobacco marketing at Wave 1 approximately one year earlier.

Table 1.

Characteristics of PATH Wave 2 Study Adolescent Respondents, Aged 14–17 Years (N=8012)¹

	Weighted Prevalence (%)
Age
14	25.5
15	25.5
16	25.3
17	23.7
Sexual Orientation and Being Transgender
Transgender	1.0
Sexual Minority Female	6.6
Sexual Minority Male	2.4
Straight Female	41.2
Straight Male	48.8
Race/Ethnicity
Non-Hispanic White	55.0
Hispanic	22.5
Non-Hispanic Black	13.5
Non-Hispanic Other/Multi-Racial	9.0
School Performance2
Mostly As or School Ungraded	26.1
As and Bs	31.9
Mostly Bs	10.0
Bs and Cs	18.3
Mostly Cs to Mostly Fs	13.6
Weekly Income
None or <$1	30.1
$1-$20	36.3
$21-$50	13.6
$51 or more	19.9
Other Substance Use
0	83.4
1	13.5
2 or More	3.0
Parental Education
Less than High School Graduate	16.5
High School Graduate or Equivalent	18.3
At Least Some College	65.2
Internalizing Problems
Low	38.1
Moderate	26.6
High	35.2
Externalizing Problems
Low	31.6
Moderate	24.8
High	43.6
Social Networking Account Use
No social networking account	5.6
Monthly or less often	6.2
Weekly	6.5
Daily	15.0
Several times a day	66.8
Regularly Used a Smart Phone	75.9
Tobacco Use Status
Never, Non-Susceptible	37.9
Never, Susceptible	31.9
Ever Tobacco Use, Not Past Year	8.0
Past Year Tobacco Use	22.1
Close Contact with a Smoker Within the Past Week	38.3
Lived with Tobacco User	30.0
Received Tobacco Discount Coupon or Promotion by Mail	1.6
Received Tobacco-Related Information by Mail	0.6
Engagement with Online Tobacco Marketing at Wave 1	9.4

¹Unweighted sample size counts cannot be reported per requirements of the PATH Restricted Use File

²The number of adolescent respondents indicating school ungraded fell below reporting requirements for PATH Restricted Use File. Thus, this category was combined with Mostly As.

3.2. Prevalence of Engagement with Online Tobacco Marketing.

An estimated 35.1% of transgender adolescents, 37.2% of sexual minority females, and 30.5% of sexual minority males engaged compared to 22.9% of straight females and 21.3% of straight males (Table 2). An estimated 23.4% of Hispanics and 26.2% of non-Hispanic Blacks engaged compared to 22.3% of non-Hispanic Whites. Across race/ethnicity, sexual orientiation, and being transgender, the leading form of engagement was watching videos online about tobacco products (Table 2). The relative ranking of other forms of engagement across these populations remained approximately consistent (Spearman’s rank correlation between pairs of sub-populations ranged from 0.78 to 1.0).

Table 2.

Prevalence of Engagement with Online Tobacco Marketing Within the Past Year. Overall and by Specific Forms, by Sexual Orientation, Being Transgender, and Race/Ethnicity (%)¹

		Specific Forms2
	≥1 Form of Online Engagement Within Past Year Pt. Est. (95% CI)	Signed Up for Email Alerts Abount Tobacco Products Pt. Est. (95% CI)	Read Articles Online About Tobacco Products Pt. Est. (95% Ci)	W atched Videos Online About Tobacco Products Pt. Est. (95% CI)	Liked or Followed Tobacco Brand on Social Media Pt. Est (95% CI)	Sent Link About Tobacco Brand on Social Media Pt. Est. (95% CI)	Played Online Game Related to a Tobacco Brand Pt. Est. (95% CI)	Received Tobacco Discount Coupon Electronically Pt. Est. (95% CI)	Received Tobacco- Related Information Electronically Pt. Est. (95% CI)
Race/Ethnicity
Hispanic	23.4 (21.6, 25.2)	7.6 (6.4, 8.7)	10.3 (9.0, 11.7)	14.3 (12.8, 15.8)	7.0 (5.9, 8.0)	3.3 (2.5, 4.0)	3.2 (2.4, 4.0)	3.0 (4.0, 3.0)	1.6 (1.0, 2.1)
Non-Hispanic Black	26.2 (23.3, 29.2)	7.0 (5.1, 8.8)	10.6 (8.5, 12.7)	13.8 (11.4, 16.2)	10.5 (8.5, 12.6)	4.7 (3.2, 6.1)	5.6 (4.1, 7.1)	3.9 (7.1, 3.9)	2.3 (1.2, 3.4)
Non-Hispanic Other/Multi	25.4 (21.7, 29.1)	12.0 (9.1, 15.0)	13.0 (10.1, 15.9)	18.2 (14.8, 21.5)	6.4 (4.6, 8.2)	3.2 (1.7, 4.6)	2.4 (1.2, 3.7)	3.0 (3.7, 3.0)	1.2 (0.3, 2.0)
Non-Hispanic White	22.3 (21.0, 23.7)	9.1 (8.2, 10.1)	10.4 (9.4, 11.5)	15.0 (13.8, 16.2)	5.4 (4.7, 6.2)	2.6 (2.1, 3.1)	2.3 (1.8, 2.8)	3.1 (2.8, 3.1)	1.7 (1.3, 2.1)
Sexual Orientation and Being Transgender
Transgender	35.1 (23.8, 46.5)	10.4 (2.3, 18.6)	10.9 (2.8, 19.0)	21.4 (10.8, 31.9)	11.3 (3.8, 18.8)	9.3 (2.3, 16.4)	9.9 (2.8, 17.0)	8.0 (17.0, 8.0)	7.7 (0.9, 14.6)
Sexual Minority Female	37.2 (32.8, 41.6)	10.2 (7.5, 13.0)	17.5 (14.0, 21.0)	22.3 (18.5, 26.1)	14.9 (11.7, 18.1)	7.7 (5.2, 10.1)	4.6 (2.7, 6.4)	4.4 (6.4, 4.4)	2.1 (0.8, 3.4)
Sexual Minority Male	30.5 (23.4, 37.6)	14.7 (9.2, 20.3)	15.5 (9.6, 21.4)	22.8 (16.1, 29.4)	5.1 (1.5, 8.6)	2.4 (0.0, 4.7)	4.7 (1.4, 8.0)	2.5 (8.0, 2.5)	1.8 (0.0, 3.6)
Straight Female	22.9 (21.3, 24.4)	9.2 (8.1, 10.3)	10.0 (8.9, 11.1)	14.9 (13.6, 16.2)	6.6 (5.7, 7.5)	3.3 (2.6, 3.9)	2.1 (1.6, 2.6)	3.3 (2.6, 3.3)	2.2 (1.6, 2.8)
Straight Male	21.3 (19.9, 22.7)	7.9 (6.9, 8.8)	9.9 (8.9, 11.0)	13.5 (12.3, 14.7)	5.3 (4.5, 6.0)	2.2 (1.7, 2.7)	3.2 (2.7, 3.8)	2.8 (3.8, 2.8)	1.1 (0.8, 1.5)

Note: Pt. Est.=Point Estimate; CI=Confidence Interval.

¹Unweighted sample size counts cannot be reported per requirements of the PATH Restricted Use File

²Sum of prevalence of specific forms of engagement not equal to overall prevalence of engagement because individuals could engage with multiple forms within the past year

Most adolescents who engaged did so with a single form of engagement, followed by two forms of engagement (Figure 1). For example, among the estimated 22.9% of straight females who engaged, 72.4% engaged with a single form, 18.5% engaged with two forms, and 9.2% engaged with three or more forms. Among the 37.2% of sexual minority females who engaged, 65.6% engaged with a single form, 22.6% engaged with two forms, and 11.8% engaged with three or more forms. Among the 35.1% of transgender adolescents who engaged, 62.7% engaged with a single form, 13.0% engaged with two forms, and 24.3% engaged with three or more forms.

Figure 1.

Prevalence of Engagement with Online Tobacco Marketing Within the Past Year and by Number of Forms of Engagement Among Adolescents Who Engaged (2014–2015)

Note: See Appendix Table 3 for 95% confidence interval of prevalence values.

3.3. Multivariable Analyses.

In Model 1, which included only race/ethnicity, sexual orientation, and being transgender, the odds of engagement were higher for transgender adolescents (adjusted odds ratio [aOR]=2.13; 95% confidence interval [CI]: 1.29–3.54) and sexual minority males (aOR=1.64; 95% CI: 1.16–2.32) compared to straight males (Table 3) and higher for sexual minority females (aOR=2.01; 95% CI: 1.63–2.48) compared to straight females (Appendix Table 2). The odds were also higher for non-Hispanic Blacks (aOR=1.24; 95% CI: 1.05–1.48) compared to non-Hispanic Whites.

Table 3.

Results from Weighted Multivariable Logistic Regression Models of Engagement with Online Tobacco Marketing Within the Past Year

	Model 1	Model 2	Model 3	Model 4	Model 5 (Full Model)
	aOR (95% CI)	aOR (95% CI)	aOR (95% CI)	aOR (95% CI)	aOR (95% CI)
Race/Ethnicity (Ref: Non-Hispanic White)
Hispanic	1.08 (0.95, 1.23)	1.13 (0.98, 1.31)	1.21 (1.04, 1.41)*	1.33 (1.13, 1.57)*	1.31 (1.11, 1.56)*
Non-Hispanic Black	1.24 (1.05, 1.48)*	1.23 (1.02, 1.48)*	1.33 (1.09, 1.62)*	1.42 (1.15, 1.76)*	1.42 (1.14, 1.77)*
Non-Hispanic Other/Multi-Racial	1.18 (0.96, 1.46)	1.20 (0.97, 1.49)	1.18 (0.94, 1.47)	1.19 (0.93, 1.52)	1.18 (0.92, 1.50)
Sexual Orientation and Being Transgender (Ref: Straight Male)
Transgender	2.13 (1.29, 3.54)	2.21 (1.30, 3.75)*	1.81 (1.01, 3.25)*	1.42 (0.74, 2.71)	1.37 (0.72, 2.60)
Sexual Minority Female	2.18 (1.78, 2.69)*	2.19 (1.76, 2.71)*	1.54 (1.22, 1.94)*	1.29 (1.00, 1.65)*	1.25 (0.97, 1.61)
Sexual Minority Male	1.64 (1.16, 2.32)*	1.66 (1.16, 2.37)*	1.49 (1.02, 2.19)*	1.57 (1.05, 2.33)*	1.57 (1.05, 2.35)*
Straight Female	1.09 (0.96, 1.23)	1.09 (0.96, 1.24)	0.92 (0.80, 1.06)	0.88 (0.75, 1.02)	0.86 (0.74, 1.00)
Age (Ref: 14 Years)
15	-	1.08 (0.91, 1.27)	1.04 (0.88, 1.23)	1.05 (0.87, 1.26)	1.06 (0.88, 1.27)
16	-	1.02 (0.86, 1.21)	0.94 (0.79, 1.12)	0.91 (0.75, 1.10)	0.89 (0.74, 1.08)
17	-	0.91 (0.77, 1.09)	0.87 (0.72, 1.05)	0.79 (0.65, 0.97)	0.79 (0.64, 0.97)
School Performance (Ref: Mostly As or School Ungraded)1
As and Bs	-	0.95 (0.80, 1.11)	0.94 (0.80, 1.11)	0.83 (0.70, 1.00)	0.85 (0.71, 1.02)
Mostly Bs	-	0.98 (0.79, 1.23)	0.96 (0.76, 1.21)	0.81 (0.63, 1.03)	0.81 (0.63, 1.04)
Bs and Cs	-	0.97 (0.80, 1.17)	0.96 (0.79, 1.17)	0.73 (0.59, 0.91)	0.77 (0.62, 0.95)
Mostly Cs to Mostly Fs	-	1.22 (1.00, 1.49)*	1.23 (1.00, 1.52)*	0.89 (0.71, 1.12)	0.90 (0.71, 1.14)
Weekly Income (Ref: None or <$1)
$1-$20	-	1.46 (1.26, 1.70)*	1.30 (1.11, 1.52)*	1.23 (1.04, 1.46)*	1.25 (1.06, 1.48)*
$21-$50	-	1.58 (1.31, 1.92)*	1.45 (1.19, 1.78)	1.31 (1.05, 1.62)*	1.33 (1.07, 1.66)*
$51 or more	-	1.60 (1.33, 1.91)*	1.48 (1.22, 1.79)*	1.21 (0.99, 1.49)	1.25 (1.01, 1.54)*
Parental Education (Ref: At Least Some College)
Less than High School Graduate	-	1.15 (0.95,1.40)	1.12 (0.91, 1.37)	1.11 (0.89, 1.38)	1.08 (0.87, 1.35)
High School Graduate or Equivalent	-	1.15 (0.97,1.36)	1.01 (0.84, 1.20)	1.08 (0.89, 1.30)	1.05 (0.86, 1.27)
Internalizing Problems (Ref: Low)
Moderate	-	-	1.23 (1.03, 1.48)*	1.22 (1.01, 1.48)*	1.22 (1.01, 1.49*)
High	-	-	1.53 (1.27, 1.85)*	1.44 (1.17, 1.76)*	1.41 (1.14, 1.73)*
Externalizing Problems (Ref: Low)
Moderate	-	-	1.44 (1.18, 1.75)*	1.33 (1.08, 1.64)*	1.37 (1.11, 1.70)*
High	-	-	1.91 (1.57, 2.31)*	1.52 (1.23, 1.88)*	1.53 (1.23, 1.89)*
Social Networking Account Use (Ref: No social networking account)
Monthly or less often	-	-	1.26 (0.85, 1.88)	1.42 (0.92, 2.20)	1.43 (0.92, 2.24)
Weekly	-	-	1.26 (0.85, 1.87)	1.25 (0.82, 1.91)	1.23 (0.80, 1.90)
Daily	-	-	1.24 (0.87, 1.75)	1.25 (0.85, 1.84)	1.24 (0.84, 1.84)
Several times a day	-	-	1.32 (0.96, 1.83)	1.22 (0.86, 1.75)	1.26 (0.88, 1.81)
Regularly Used Smart Phone (Ref: No)	-	-	1.12 (0.95, 1.31)	1.09 (0.91, 1.29)	1.10 (0.92, 1.31)
Other Substance Use (Ref: No)
1	-	-	-	1.33 (1.08, 1.64)*	1.35 (1.09, 1.67)*
≥2	-	-	-	1.69 (1.17, 2.44)*	1.65 (1.14, 2.38)*
Close Contact with a Smoker Within the Past Week (Ref: No)	-	-	-	1.65 (1.43, 1.92)*	1.60 (1.37, 1.85)*
Lives with Tobacco User (Ref: No)	-	-	-	1.03 (0.88, 1.19)	1.03 (0.88, 1.20)
Received Tobacco Discount Coupon or Promotion by Mail (Ref: No)	-	-	-	3.00 (1.80, 4.98)*	2.73 (1.58, 4.70)*
Received Tobacco-Related Information by Mail (Ref: No)	-	-	-	4.92 (1.92, 12.63)*	4.55 (1.79, 11.59)*
Tobacco Use Status (Ref: Never, Non-Susceptible)
Never, Susceptible	-	-	-	1.83 (1.54, 2.17)*	1.74 (1.47, 2.07)*
Ever Tobacco Use, Not Past Year	-	-	-	1.45 (1.12, 1.89))*	1.28 (0.98, 1.67)
Past Year Tobacco Use	-	-	-	2.21 (1.77, 2.76)*	2.02 (1.61, 2.53)*
Engagement with Online Tobacco Marketing at Wave 1 (Ref: No)	-	-	-	-	3.35 (2.72, 4.11)*

Note: aOR=adjusted odds ratio; CI=confidence interval; Ref=reference.

1 The number of adolescent respondents indicating school ungraded fell below reporting requirements for PATH Restricted Use File. Thus, this category was combined with Mostly As.

^*p<0.05.

These associations decreased in magnitude when other socio-demographic, psychosocial, environmental characteristics; internet-related and substance use behaviors; other potential risk factors for tobacco use; tobacco use status; and ever engagement with online tobacco marketing were included. In the full model (Model 5), the adjusted odds of engagement were 1.57 (95% CI: 1.05–2.35) times higher for sexual minority males compared to straight males and 1.45 (95% CI: 1.13–1.87) times higher for sexual minority females compared to straight females (Table 3 and Appendix Table 2). In the full model, the adjusted odds of engagement were higher for Hispanics (aOR=1.31; 95% CI: 1.11–1.56) and non-Hispanic Blacks (aOR=1.42; 95% CI: 1.14–1.77) compared to non-Hispanic Whites.

The adjusted odds of engagement within the past year were also higher for adolescents who used other substances compared to those who did not (e.g., aOR=1.35; 95% CI: 1.09–1.67 for substance use score of 1), demonstrated high levels of internalizing (aOR=1.41; 95% CI: 1.14–1.73) or externalizing (aOR=1.53; 95% CI: 1.23–1.89) disorders compared to those with low levels, and were in close contact with a smoker within the past week (aOR= 1.60; 95% CI: 1.37–1.85) compared to those who were not (Table 3). Frequency of using social networking sites was not associated with significantly higher odds of engagement. Compared to non-susceptible never tobacco users, the adjusted odds of engagement were higher for susceptible never tobacco users (aOR=1.74; 95% CI: 1.47–2.07) and past-year tobacco users (aOR=2.02; 95% CI: 1.61–2.53). Finally, the adjusted odds of past-year engagement at Wave 2 were higher for respondents who had ever engaged at Wave 1 (aOR=3.35; 95% CI: 2.72–4.11) compared to those who had never engaged at Wave 1.

4. DISCUSSION.

Our cross-sectional study found SGM adolescents and racial and ethnic minority adolescents reported higher engagement with online tobacco marketing compared to their heterosexual and non-Hispanic white peers, respectively. These minority populations are already exposed to greater levels of tobacco marketing in traditional, offline marketing channels (e.g., ads in magazines) and such exposure has been shown to lead to adolescent tobacco use [17]. The increased risk of online engagement among these minority populations may also pose a public health concern. Engagement with online tobacco marketing, as with exposure to traditional tobacco marketing, is positively associated with tobacco use initiation and increased frequency of tobacco use and negatively associated with tobacco use cessation [14].

The higher level of engagement with online tobacco marketing within the past year among sexual minority and transgender adolescents was not due to higher levels of tobacco use; our study found tobacco use status only partially accounted for this relationship. Instead, the higher level of engagement could be the result of targeting by the tobacco and e-cigarette industries [4]. A substantial body of evidence found the tobacco industry previously marketed to these populations through focused promotional campaigns, magazine advertisements, event sponsorship, and advocacy organizations funding [2,18–21]. For example, a review of internal tobacco industry documents concluded the industry concentrated its marketing and promotion in geographic areas where sexual and gender minority youth gather [5]. The tobacco and e-cigarette industries may now be employing similar tactics through targeted magazine advertisements that encourage consumers to engage with products online [22]. For example, a 2016 ad in The Advocate (a prominent LGBTQ magazine) portrayed a man in drag using a Blu e-cigarette with the caption “JUST YOU & blu” and featured the Blu’s website address (Appendix Figure A.1) [23]. These new marketing techniques may confer harm because adolescents who recall or like advertisements for tobacco products—including e-cigarettes—are at higher risk to begin tobacco use [24]. Sexual and gender minorities may not necessarily perceive targeting as negative; rather targeting may confer external legitimacy and acknowledgement of economic power for some individuals in these populations [25].

In addition to possible targeting by the tobacco and e-cigarette industries, the level of engagement with online tobacco marketing may be higher for sexual minority and transgender adolescents for several other reasons. First, a higher proportion of the peer groups of SGM youth are already tobacco users compared to straight youth [26]. Compared to straight adolescents, lesbian, gay, and bisexual (LGB) adolescents were more than twice as likely to have smoked a whole cigarette before age 13 (12.8% for LGB versus 5.8% for straight) and currently smoke (19.2% for LGB versus 9.8% for straight) [27]. Second, SGM youth may also be more at risk for engagement with online tobacco marketing than straight youth because of their greater overalluse of social media [28]. Third, the in-person discrimination and stigma that many SGM adolescents experience may contribute to higher levels and more frequent use of social networking sites, which enable youth to build supportive communities online [29,30]. However, greater social connectedness through the internet may also increase vulnerability to exposure to online tobacco marketing if their virtual peers share tobacco-related content or if tobacco marketing promotes engagement with their virtual peers [13,31].

Racial and ethnic minority youth may be more likely to engage with online tobacco marketing because they are more familiar with tobacco products through greater exposure to traditional forms of marketing than non-Hispanic White youth [32]. Minority youth experience greater exposure to tobacco advertising at the point of sale [33]. Tobacco retail outlet densityhas been shown to increase as the proportion Black residents increases within a census tract, and Black and Hispanic adolescents have been shown to be more likely to live within a half mile of a tobacco outlet [34,35]. Moreover, these youth are more likely to report that tobacco advertising played a role in their decision to use a tobacco product [36]. Targeted marketing on the part of the tobacco industry has exacerbated these effects. After the 1998 Master Settlement Agreement, tobacco companies increasingly concentrated their magazine advertisement spending on mentholated brands, such as Lorillard’s Newport brand (now a brand of RJ Reynolds, a subsidiary of British American Tobacco), which are popular among Black youth [37]. Additionally, the tobacco industry has long sponsored minority-focused musical and cultural events that prominently featured brand placement (e.g., hip-hop focused Swisher Sweets Artist Project, which includes live music concerts held in convience stores) [3]. Finally, discriminationn-based stress may lead racial and ethnic minorities to seek supportive virtual communities online, which could unintentionally lead to greater engagement with online tobacco marketing through the same processes as described for sexual and gender minorities.

Racial, ethnic, sexual, and gender minority adolescents experience higher levels of mental health distress than their white and straight peers [38,39]. Our results suggest adolescents experiencing moderate and high levels of mental health distress may be more likely to engage with online tobacco marketing than those experiencing low levels of distress. This increased risk of engagement is problematic for several reasons. First, adolescents with behavioral and emotional disorders are already more likely to initiate tobacco use than adolescents without these disorders [40]. Higher levels of engagement with online tobacco marketing—which is associated with an increased risk of tobacco use initiation and decreased likelihood of tobacco use cessation—could widen disparities in tobacco use among adolescents with behavioral and emotional disorders [41]. Second, the tobacco industry previously targeted individuals with serious mental illness to develop brand affinity and bolster sales (e.g., distributing tobacco-brand blankets to homeless shelters and soup kitchens) [42,43]. Future research could assess why adolescents experiencing mental health distress are more likely to engage with online tobacco marketing, which would help develop effective counter-marketing strategies.

In 2015 and 2016, the FDA launched two public education campaigns that focused on youth who identified with hip-hop culture—a large proportion of whom are racial and ethnic minorities—and SGM youth [44,45]. The campaigns promote tobacco-free lifestyles and communicating the negative health consequences and addictive nature of cigarette smoking. If the campaigns prove effective among these youth populations, they could expand and increase awareness about targeting by the tobacco and e-cigarette industries, both through offline and online marketing channels.

Political, civil rights, and advocacy organizations could potentially reduce the level of engagement with online tobacco marketing among youth by publicly disclosing previous relationships with the tobacco industry and divesting from such relationships in the future. Such disclosures and divestures could eliminate conflicts of interest, both perceived and actual, and lead organizations to act in more socially responsible ways [46]. Civic rights and advocacy organizations have been instrumental in countering tobacco marketing focused on minorities (e.g., Philadephia-based tobacco control group the Uptown Coalition efforts against RJ Reynold’s African-American targeted Uptown brand cigarettes) [47]. Future tobacco regulation that actively engages with advocacy organizations—and does so early in the regulation-crafting process—may aid in the production of effective policies [48]. A combination of federal regulation and grass-root public health campaigns could reduce youth engagement with online tobacco marketing; our study suggests such a reduction could reduce the level of and disparities in tobacco use initation, especially among SGM youth.

Several limitations of this study are noted. First, our study may have underestimated the prevalence of gender minorities because the PATH study only asked respondents if they considered themselves transgender and did not ask about other gender minority categories (e.g., genderqueer). Also, some respondents may have interpreted the question “What is your sex?” as biological sex, sex assigned at birth, or gender. Second, our study may conservatively estimate the level of engagement within the past year because three of the PATH Study items used to measure engagement in Wave 2 asked about engagement with only seven specific tobacco brands: three cigarette brands (Camel, Marlboro, and Newport), one cigar brand (Swisher), and three e-cigarette brand (Blu, Fin, and NJOY). While these brands represent a large share of the cigarette, cigar, and e-cigarette markets nationally, the prevalence of use for other brands not assessed in the PATH Study (e.g., JUUL) may be high among youth. Third, the PATH Study did not assesss participants’ level of perceived discrimination, which could interact with exposure to tobacco marketing and increase the risk of tobacco use [49]. Fourth, the PATH Study may not have been sufficiently powered to ascertain statistically significant differences in the odds of engagement among SGM adolescents because of their relatively small number in the study. However, the point estimates across models in the stepwise regression all suggest increased odds of engagement among SGM adolescents. Finally, several factors may have affected youth tobacco use patterns since the time of data collection (2014–2015) including the types of available tobacco products and new brands of e-cigarettes. Similarly, new social networkingsites have grown in prominence and online marketing continues to evolve, both of which could have contributed to the temporal increase in the level of engagement with online tobacco marketing among adolescents [50,51].

5. CONCLUSIONS

In conclusion, sexual, gender, racial, and ethnic minority adolescents are at higher risk of engaging with online tobacco marketing than their majority-group peers. This risk may worsen existing disparities in tobacco use among some of these populations. More rigorous and comprehensive federal regulation of tobacco marketing could help reduce adolescents’ opportunity to engage in online tobacco marketing. Advocacy organizations could also reduce engagement among minority youth by no longer accepting economic support from the tobacco industry. Finally, current and future public health education campaigns, especially focused on high-risk adolescents, could help counter online tobacco marketing.

HIGHLIGHTS.

• Leading form of engagement with online tobacco marketing was watching videos about tobacco products.

• Sexual/gender and racial/ethnic minority adolescents face higher risk of engaging with online tobacco marketing compared to their straight and non-Hispanic white counterparts.