Association between free tobacco product sample receipt and tobacco use in youth and adults in the PATH Study, 2014–2016

Abstract

Tobacco companies have distributed free samples of tobacco products in the past. While prior studies have found a relationship between various marketing strategies and tobacco use, no study has assessed the prevalence of free sample receipt or the relationship between receipt and subsequent tobacco use. We analyze three waves of Population Assessment of Tobacco and Health (PATH) Study data to provide the first nationally representative prevalence estimates of free tobacco product receipt among US youth and adults in 2014 to 2016. This analysis also examines the relationship between free tobacco product receipt and ever, past 30-day, and new tobacco use one year later. The prevalence of free sample receipt in the US population in 2014–2016 was 0.8% for any tobacco product but 1.6% for ENDS/e-liquid in adults, and 0.5% for any tobacco product but 1.05% for ENDS/e-liquid in youth. Free sample receipt was higher among vulnerable subgroups. Receipt of free tobacco product samples was associated with tobacco use status using cross-sectional data in youth and adults, as well as one year later in youth. Receipt of a free ENDS or e-liquid sample was associated with any tobacco and ENDS use status using cross-sectional data in youth and adults. After these data were collected, the US Food and Drug Administration broadened the ban on distributing free samples to include all tobacco products except smokeless tobacco in adult-only facilities. Our findings support limiting free samples to further prevent youth access to tobacco products.

1. Introduction

Tobacco company distribution of free product samples had long been a standard practice among brand marketers, and is often used to encourage trial of a new product (Schultz et al., 1998; Peattie, 1998). Tobacco companies historically distributed free samples of tobacco products at retailers, bars, and events, such as concerts, baseball games or festivals, or through mailed or online redeemable vouchers (Lewis et al., 2004; Lewis and Ling, 2016; David and Jason, 1988). More recently in 2014, free samples of e-cigarettes and e-liquids were distributed at vape shops (Cheney et al., 2015).

The relationship between industry marketing and tobacco use among youth and young adults is well established (Cruz et al., 2018; Pierce et al., 1998; Henriksen et al., 2010; Lovato et al., 2011; Pierce et al., 2018; Soneji et al., 2014). Several studies found a relationship between youth (Choi and Forster, 2014a), young adult (Lewis et al., 2015; Choi and Forster, 2014b; Choi et al., 2018a), and adult (Choi et al., 2018b) receipt of direct mail or coupon receipt and tobacco trial and use. However, few studies have examined the prevalence of free samples or their relationship with tobacco use. Only one study examined the effect of free samples, finding that receipt of free samples of cigarette or smokeless tobacco products is positively associated with susceptibility among adolescents (Altman et al., 1996).

Currently in the United States, smokeless tobacco products are the only tobacco product that companies can distribute for free to adults in “qualified adult-only facilities” as defined in the 2009 Family Smoking Prevention and Tobacco Control Act (FDA, 2009; FDA, 2010). Although free tobacco product samples are no longer allowed in the US except for smokeless tobacco, this was not always the case. Before the 2016 “deeming rule” that extended FDA’s authority over all tobacco products, some tobacco products, including electronic nicotine delivery systems (ENDS), could be given as free samples (FDA, 2016). Cigarettes, cigarette tobacco and roll-your-own tobacco could not be given as free samples. Understanding the effects of free samples on tobacco use could be useful for other countries where free samples are still distributed (Henriksen, 2012) and for other consumer products. Furthermore, understanding the role of free tobacco samples in trial and subsequent use among youth and adults contributes to the broader literature about the influence of other tobacco price promotion strategies such as discount coupons (Chaloupka et al., 2002). Free samples differ from coupons or promotions because they don’t require purchase.

In this study, we analyzed data on free samples and tobacco use from the US Population Assessment on Tobacco and Health (PATH) Study. First, we describe the prevalence of receipt of free tobacco product samples, including ENDS and e-liquid in a nationally representative sample of US youth and adults from 2014 to 2016. Second, we examine whether receipt of a free sample is associated with tobacco use status (ever, past 30-day, and recent new use). Third, we examine the longitudinal association between receipt of a free sample and tobacco use (ever use, past 30-day use, and recent new use) one year later. We hypothesized that free sample receipt is associated with tobacco use status and subsequent tobacco use one year later.

2. Methods

The National Institutes of Health, through the National Institute on Drug Abuse, is partnering with the Food and Drug Administration’s Center for Tobacco Products to conduct the PATH Study under a contract with Westat. The PATH Study is an ongoing, nationally-representative, longitudinal cohort study of adults and youths in the US. The PATH Study uses audio computer-assisted self-interviews available in English and Spanish. Data collection dates were as follows: Wave 1 (September 12, 2013 to December 14, 2014), Wave 2 (October 23, 2014 to October 30, 2015), and Wave 3 (October 19, 2015 to October 23, 2016). PATH Study recruitment employed a stratified address-based, area-probability sampling design at Wave 1 that oversampled adult tobacco users, young adults (18 to 24 years), and African-American adults.

Full-sample and replicate weights were created that adjusted for the complex study design and nonresponse. Combined with the use of a probability sample, the weights allow analyses to compute robust estimates that approximate the non-institutionalized, civilian US population ages 12 years and older at Wave 1 who are still in the resident population at Wave 2 and Wave 3. Further details regarding the PATH Study design and methods are published elsewhere (Hyland et al., 2017; https://doi.org/10.3886/Series606). The PATH Study was conducted by Westat and approved by the Westat Institutional Review Board.

2.1. Analytic sample

The current study analyzes data from adults (n = 25,384) and youth (n = 9332) with information at all three waves. For analyses, youth or adult status was based on age at Wave 2. This cut-off permits analyses to track the association between Wave 2 free sample receipt and Wave 3 use while defining age at free sample receipt. Eighteen-year-olds at Wave 3 (n = 1737) were considered youth and were not included in Wave 3 tobacco product free sample receipt estimates by demographics in Table 1b. Twelve-year-olds that entered the sample at Wave 2 or 3 were not included in this analysis.

Table 1b.

Free sample receipt by demographic characteristics and coupon or promotion receipt among youth ages 12–17 years at Wave 2 (2014–2015) and Wave 3 (2015–2016).

	Wave 2 (2014–2015)	Wave 3 (2015–2016)
	Free sample of any tobacco product in past 30 days	Free sample of any tobacco product in past 30 daysa	Free sample of E-cigarette or E-liquid in past 30 days
Count (weighted %) [SE]	71 (0.73) [0.09]	42 (0.47) [0.1]	93 (1.05) [0.11]
Weighted population	149,693	96,481	215, 067
Total N	9302	7568	9299
	Weighted % (95%CI)	Weighted % (95%CI)	Weighted % (95%CI)
Demographics (at wave 2)
Age
12–14 years	0.69 (0.46, 1.02)	0.4 (0.28, 0.93)	0.67 (0.45, 1.01)
15–17 years	0.76 (0.54, 1.07)	0.6 (0.38, 1.0)	1.33 (1.03, 1.72)
Gender
Female	0.56 (0.37, 0.84)	0.39 (0.22, 0.70)	0.91 (0.66, 1.27)
Male	0.88 (0.65, 1.20)	0.76 (0.48, 1.21)	1.19 (0.90, 1.59)
Race/ethnicity
White, Non-Hispanic	0.62 (0.43, 0.88)	0.61 (0.34, 1.08)	1.1 (0.87, 1.51)
Black, Non-Hispanic	1.21 (0.71, 2.06) d	0.98 (0.44, 2.20)d	0.86 (0.46, 1.62)
Hispanic or Latino	0.88 (0.58, 1.32) d	0.38 (0.18, 0.79)d	0.9 (0.63, 1.43)
Otherb	0.33 (0.12, 0.92) d	0.33 (0.10, 1.03)d	1.0 (0.50, 2.38)d
Receipt of coupon or promotionc
Yes	5.64 (3.76, 8.38)	3.38 (2.05, 5.54)	6.59 (4.91, 8.79)
No	0.48 (0.35, 0.65)	0.31 (0.17, 0.55)	0.58 (0.43, 0.80)

Note. The Wave 3 all-wave weights were used for the analyses presented in this table. Therefore, the individuals represented in this table had to be present at all three waves – Wave 1, Wave 2, and Wave 3. Bolded values indicate statistical significance (p < 0.05) based on a chi-square test.

^aReceipt of Wave 3 free tobacco product sample receipt was asked only to those 12–17 years old at Wave 3. Since we used age at Wave 2 as our definition for youth and adults in this analysis, free tobacco product sample receipt at Wave 3 does not include the 1737 youth who turned 18 years old between Wave 2 and Wave 3.

^bOther race/ethnicity includes American Indian/Alaskan Native, Non-Hispanic; Asian, Non-Hispanic; NH or OP, Non-Hispanic; more than one race, Non-Hispanic.

^cAt Wave 2, youth were asked whether they received a coupon or promotion for a tobacco product or e-cigarette in the past 30 days. At Wave 3, youth were asked whether they received coupons or promotions for tobacco products in the past 12 months. Youth at Wave 3 were asked about receipt of coupons or promotions for shisha or hookah tobacco. This item was not asked of adults at Wave 3 or of adults or youth at any prior waves.

^dEstimate has been flagged because it is statistically unreliable. It is based on a sample size of less than 50, or the coefficient of variation of the estimate is larger than 30%.

2.2. Measures

2.2.1. Free sample receipt

Our independent variable was free tobacco product sample receipt assessed as follows: “In the past 30-days, have you gotten a free sample of a tobacco product?” Receipt of a free tobacco product sample was assessed among all youth at Waves 2 and 3 and among all adults at Wave 2. Receipt of a free ENDS sample was assessed with the following item: “In the past 30-days, have you gotten a free sample of an e-cigarette or e-liquid/or other electronic nicotine product (including e-liquid)?” The item was only asked of ever e-cigarette users in Wave 2 and of all respondents in Wave 3. In Wave 2, the item asks about receipt of “e-cigarette or e-liquid,” while in Wave 3, the item is expanded to include “other electronic nicotine products (including e-liquid).” The youth and adult items are provided in Supplemental Table 1.

2.2.2. Tobacco use

Our outcome variables were ever, current, and new tobacco product use. New tobacco use was defined as a change from never use of a given tobacco product to use of one or more products one year later. New use was assessed by asking never users of a given product if they had used the product in the past 30 days or the past 12 months. For example, youth and adults who started using one or more of the following products for the first time between Wave 2 and Wave 3 were considered new tobacco users at Wave 3: cigarettes, traditional cigars, cigarillos, filtered cigars, smokeless tobacco, hookah, ENDS, dissolvable tobacco, snus, or pipes.

Ever use of a tobacco product was assessed at each wave among all adults and among youth that have heard of or seen each product with the following items: “Have you ever tried [cigarette smoking/smoked a traditional cigar, cigarillo, or filtered cigar/tobacco in a hookah/a pipe filled with tobacco], even one or two puffs?” “Have you ever used [dissolvable tobacco products/snus pouches, loose snus, moist snuff, dip, spit, or chewing tobacco, electronic nicotine products (including e-cigarettes, e-cigars, e-pipes, e-hookahs, personal vaporizers, vape pens and hookah pens)] even one or two times?” We defined ever use of a tobacco product as ever use of one or more products.

Current tobacco product use was assessed at each wave among all adults and among youth who had ever heard of or seen each product and had ever used it, even one or two times. In this analysis, we define current tobacco use as past 30-day use of one or more of the following tobacco products: cigarettes, traditional cigars, cigarillos, filtered cigars, smokeless tobacco including snus pouches, hookah, ENDS, dissolvable tobacco, or pipes.

At Wave 3 in youth and adults, we assessed ever use, past 30-day use, and new use in the past 12 months of ENDS, such as e-cigarettes, e-cigars, e-pipes, e-hookahs, personal vaporizers, vape pens, and hookah pens.

2.2.3. Coupon receipt

We assessed coupon receipt among adults at Wave 2 and Wave 3 with the following: “In the past 12 months, have you received [e-mail message/mail] with discount coupons or promotions for [cigarettes, e-cigarettes or other electronic nicotine products (including e-liquid), cigars, shisha or hookah tobacco, snus pouches, other types of smokeless tobacco (such as dip, spit or chew), some other type of tobacco product]?”

We assessed receipt of coupons in youth with the following items: Wave 2: “In the past [30-days], have you received any discount coupons or promotions for tobacco products or e-cigarettes, whether it was online or some other way?” Wave 3: “In the past 12 months, have you received discounts or coupons for [cigarettes, e-cigarettes or other electronic nicotine products (including e-liquid), cigars, shisha or hookah tobacco, snus pouches, other types of smokeless tobacco (such as dip, spit or chew), some other type of tobacco product]?”

2.2.4. Demographic characteristics

We included the following in our analyses: age, race/ethnicity, and gender. In adults, we also examined education, income, and sexual orientation. All demographic characteristics were assessed at Wave 2. Missing data on age, gender, race, Hispanic ethnicity, and adult education were imputed as described in the PATH Study Restricted Use Files User Guide (https://doi.org/10.3886/Series606).

2.3. Analytic approach

We conducted analyses using SVY procedures in Stata/MP version 15.1. We estimated the prevalence of receipt of free samples by demographics and tobacco use status among youth and adults who received free samples of tobacco products and e-cigarettes or e-liquids between October 2014 and October 2016. Weighted percentages and logit-transformed 95% confidence intervals were calculated. p-Values based on weighted chi-square tests assessed differences in free sample receipt by demographic characteristics at Wave 2 and coupon receipt at Wave 2 or Wave 3. We conducted analyses separately for youth and adults because we anticipated relationships of interest to differ.

Logistic regression models tested whether Wave 2 receipt of a free tobacco product sample was associated with Wave 2 ever use, past 30-day use or new tobacco product use, adjusting for Wave 2 demographics and coupon receipt. Models for past 30-day and new use also included past 30-day tobacco use assessed at Wave 1 to account for potential differences in sample receipt by use history. Logistic regression models tested whether Wave 2 receipt of a free tobacco product sample was associated with subsequent Wave 3 ever tobacco use, past 30-day tobacco use and new tobacco product use, adjusting for Wave 2 demographics and coupon receipt. The models for past 30-day and new use also included past 30-day tobacco use assessed at Wave 1 and Wave2. Lastly, logistic regression models tested whether Wave 3 receipt of a free sample of ENDS, including e-liquid, was associated with Wave 3 ever ENDS use, past 30-day ENDS use, or new ENDS use, adjusting for Wave 2 demographics and coupon receipt. Models for past 30-day and new ENDS use also included past 30-day tobacco use at Wave 1 or Wave 2. We could not examine the relationship between ENDS free sample receipt and use at Wave 2 or 3 since the Wave 2 e-cigarette free sample question was only provided to ever ENDS users.

3. Results

3.1. Free sample receipt by demographic and tobacco use characteristics

At Wave 2, 0.8% of adults and 0.7% of youth received a free tobacco product sample in the past 30 days. At Wave 3, 1.6% of adults and 1.0% of youth received a free sample of an e-cigarette or e-liquid in the past 30 days. Among ever users of e-cigarettes at Wave 2, 5.2% of adults and 5.6% of youth received a free sample of an e-cigarette or e-liquid in the past 30 days.

Among adults, a higher proportion of younger adults (ages 18–24) received a free sample of an e-cigarette or e-liquid in the past 30 days as compared to older adults over age 45 at Wave 3 (Table 1a). Among youth, a greater proportion of youth ages 15–17 received a free e-cigarette or e-liquid sample at Wave 3 compared to youth ages 12–14 (Table 1b). For adults, a greater proportion of Black respondents received a free tobacco product sample in the past 30 days at Wave 2 compared to other racial and ethnic groups with the exception of Hispanic respondents. A greater proportion of lesbian/gay/bisexual adults received a free sample of a tobacco product or ENDS/e-liquid in the past 30 days compared to adults identifying as straight at Wave 2. A greater proportion of adults with the lowest household income reported receipt of free samples of any tobacco product at Wave 2 and ENDS/e-liquid at Wave 3 compared to those with the highest household income. Adults with a bachelor’s degree or higher had a lower prevalence of free tobacco or ENDS/e-liquid sample receipt compared to adults with less education. Adults and youth at both waves who received coupons had a higher prevalence of free sample receipt.

Table 1a.

Free sample receipt by demographic characteristics and coupon or promotion receipt among adults 18 years and older at Wave 2 (2014–2015) and Wave 3 (2015–2016).

	Wave 2 (2014–2015)	Wave 3 (2015–2016)
	Free sample of any tobacco product in past 30 days	Free sample of E-cigarette or E-liquid in past 30 days
Count (weighted %) [SE]	323 (0.8) [0.1]	624 (1.65) [0.09]
Weighted population	1,901,252	3,967,522
Total N (non-missing)	25,349	25,328
	Weighted % (95%CI)	Weighted % (95%CI)
Demographics (at wave 2)
Age
18–24	0.8 (0.62, 1.0)	2.7 (2.3, 3.1)
25–44	1.1 (0.87, 1.3)	2.2 (1.9, 2.6)
45–64	0.8 (0.59, 1.0)	1.5 (1.2, 1.9)
65+	0.4 (0.19, 0.85) b	0.3 (0.09, 0.87) b
Gender
Female	0.7 (0.58, 0.87)	1.4 (1.2, 1.7)
Male	0.9 (0.72, 1.1)	2.0 (1.7, 2.3)
Race/ethnicity
White, Non-Hispanic	0.7 (0.6, 0.89)	1.7 (1.5, 2.0)
Black, Non-Hispanic	1.1 (0.92, 1.4)	1.8 (1.4, 2.3)
Hispanic or Latino	0.9 (0.72, 1.3)	1.6 (1.3, 2.1)
Other, Non-Hispanica	0.4 (0.2, 0.68)	1.1 (0.73, 1.6)
Sexual orientation
Lesbian, gay, bisexual, something else	2.1 (1.4, 3.2)	2.6 (1.83, 3.7)
Straight	0.7 (0.63, 0.90)	1.6 (1.5, 1.86)
Education
Less than HS	1.2 (0.95, 1.6)	1.7 (1.2, 2.4)
GED	1.1 (0.76, 1.6)	2.1 (1.6, 2.8)
HS/diploma	1.0 (0.73, 1.3)	1.8 (1.5, 2.5)
Some college, technical, associate degree	0.9 (0.69, 1.3)	2.1 (1.8, 2.5)
Bachelor’s or more	0.3 (0.2, 0.38)	0.9 (0.73, 1.2)
Income
Less than $25,000	1.4 (1.1, 1.7)	2.5 (2.2, 3.0)
$25,000–49,999	0.8 (0.53, 1.2)	1.5 (1.2, 1.9)
$50,000-$74,999	0.9 (0.58, 1.4)	1.7 (1.2, 2.2)
≥$75,000	0.3 (0.21, 0.43)	1.0 (0.7, 1.4)
Receipt of coupon or promotion in past 12 months
Yes	4.8 (3.9, 5.8)	6.7 (5.9, 7.6)
No	0.2 (0.17, 0.29)	0.8 (0.65, 0.94)

Note. The Wave 3 all-wave weights were used for the analyses presented in this table. Therefore, the individuals represented in this table had to be present at all three waves – Wave 1, Wave 2, and Wave 3. Bolded values indicate statistical significance (p < 0.05) based on a chi-square test.

^aOther race/ethnicity includes American Indian or Alaskan Native, Non-Hispanic; Asian, Non-Hispanic; Native Hawaiian or Other Pacific Islander, Non-Hispanic; or more than one race, Non-Hispanic.

^bEstimate has been flagged because it is statistically unreliable. It is based on a sample size of less than 50, or the coefficient of variation of the estimate is larger than 30%.

3.2. Free sample receipt and tobacco use status

Free sample receipt at Wave 2 was associated with ever, past 30-day, and new tobacco product use among youth and adults (Table 2a). Youth receiving a free tobacco product sample at Wave 2 had greater odds of ever use (AOR: 12.0; 95%CI: 5.9, 24.3), past 30-day use (AOR: 18.8; 95%CI: 9.8, 35.8), and new product use (AOR: 6.3, 95%CI: 3.5, 11.2) compared to youth who had not received a sample. Adults receiving a free tobacco product sample at Wave 2 had greater odds of ever use (AOR: 4.9; 95%CI: 1.7, 14.0), past 30-day use (AOR: 7.5; 95%CI: 3.2, 17.1), and new product use at Wave 2 (AOR: 1.6; 95%CI: 1.2, 2.2) compared to adults who had not received a free product.

Table 2a.

Associations of free tobacco product sample receipt at Wave 2 with tobacco use status at Wave 2 (2014–2015).

Youth		Wave 2 ever tobacco use			Wave 2 past 30-day tobacco use			New product use between Wave 1 and Wave 2
Count (weighted %) [SE]		2489 (26.9) [0.51]			814 (9.16) [0.37]			1306 (14.51) [0.45]
	No. of observations	Weighted %	OR (95%CI)	aORa (95%CI)	Weighted %	OR (95%CI)	aORb (95%CI)	Weighted %	OR (95%CI)	aORb (95%CI)
Wave 2 free tobacco product sample receipt
No	9231	26.5	1 [ref]	1 [ref]	8.8	1 [ref]	1 [ref]	14.2	1 [ref]	1 [ref]
Yes	71	82.3	13.12 (6.65, 25.86)	12.01 (5.93, 24.3)	60.2	18.4 (11.6, 29.2)	18.8 (9.8, 35.8)	55.6	7.9 (5.0, 12.4)	6.3 (3.5, 11.2)
Adults		Wave 2 ever tobacco use			Wave 2 past 30-day tobacco use			New product use between Wave 1 and Wave 2
Count (weighted %) [SE]		21,402 (74.2) [0.58]			13,164 (29.3) [0.34]			5250 (11.8) [0.23]
	No. of observations	Weighted %	OR (95%CI)	aORc (95%CI)	Weighted %	OR (95%CI)	aORd (95%CI)	Weighted %	OR (95%CI)	aORd (95%CI)
Wave 2 free tobacco* product sample receipt
No	25,026	72.8	1 [ref]	1 [ref]	28.7	1 [ref]	1 [ref]	11.6	1 [ref]	1 [ref]
Yes	323	96.1	8.6 (3.2, 23.4)	4.9 (1.7, 14.0)	81.0	12.7 (8.6, 18.9)	7.5 (3.2, 17.1)	30.1	3.3 (2.5, 4.3)	1.6 (1.2, 2.2)

Notes. The Wave 3 all-wave weights were used for the analyses presented in this table. Therefore, the individuals represented in this table had to be present at all three waves – Wave 1, Wave 2, and Wave 3. Bolded values indicate statistical significance (p < 0.05).

^aThe logistic regression model included the following covariates: age, sex, race/ethnicity, and Wave 2 coupon or promotions receipt.

^bThe logistic regression model included the following covariates: age, sex, race/ethnicity, Wave 2 coupon or promotions receipt, and Wave 1 and Wave 2 past 30-day tobacco use.

^cThe logistic regression model included the following covariates: age, sex, race/ethnicity, sexual orientation, income, education, and Wave 2 coupon or promotions receipt.

^dThe logistic regression model included the following covariates: age, sex, race/ethnicity, sexual orientation, income, education, Wave 2 coupon or promotions receipt, and Wave 1 and Wave 2 past 30-day tobacco use.

Youth receipt of a free ENDS or e-liquid sample at Wave 3 was associated with ever use, past 30-day use, and new tobacco product use (Table 2b). Youth who received a free ENDS or e-liquid sample in Wave 3 had greater odds of ever use (AOR: 3.3, 95%CI: 1.8; 6.1), past 30-day use (AOR: 4.8; 95%CI: 2.3, 10.1), and new product use (AOR: 3.6; 95%CI: 1.9; 6.5) compared to youth who had not received a sample. Adults receiving an ENDS or e-liquid sample in Wave 3 had greater odds of past 30-day use (AOR: 2.5; 95%CI: 1.3, 4.9) and new product use at Wave 3 (AOR: 1.7; 95%CI: 1.1, 2.6) compared to adults who did not receive a sample.

Table 2b.

Associations of electronic nicotine products (ENDS) or E-liquid sample receipt at Wave 3 with tobacco use status at Wave 3 (2015–2016).

Youth		Wave 3 ever tobacco use			Wave 3 past 30-day tobacco use			New product use between Wave 2 and Wave 3
Count (weighted %) [SE]		3172 (36.1) [0.59]			1307 (14.69) [0.46]			1358 (15.3) [0.44]
	No. of observations	Weighted %	OR (95%CI)	aORa (95%CI)	Weighted %	OR (95%CI)	aORb (95%CI)	Weighted %	OR (95%CI)	aORb (95%CI)
Wave 3 free ENDS or E-liquid sample receipt
No	9206	35.6	1 [ref]	1 [ref]	14.2	1 [ref]	1 [ref]	14.9	1 [ref]	1 [ref]
Yes	93	74.9	5.39 (3.06, 9.52)	3.3 (1.8, 6.1)	60.8	9.4 (5.9, 14.9)	4.8 (2.3, 10.1)	51.9	6.20 (3.80, 10.12)	3.56 (1.9, 6.48)
Adults		Wave 3 ever tobacco use			Wave 3 past 30-day tobacco use			New product use between Wave 2 and Wave 3
Count (weighted %) [SE]		21,668 (72.7) [0.59]			12,767 (28.1) [0.32]			2292 (5.2) [0.16]
	No. of observations	Weighted %	OR (95%CI)	aORc (95%CI)	Weighted %	OR (95%CI)	aORd (95%CI)	Weighted %	OR (95%CI)	aORd (95%CI)
Wave 3 free ENDS or E-liquid sample receipt
No	24,704	73.7	1 [ref]	1 [ref]	27.9	1 [ref]	1 [ref]	5.2	1 [ref]	1 [ref]
Yes	624	80.5	1.4 (0.98, 2.1)	0.86 (0.5, 1.4)	66.7	5.2 (4.0, 6.7)	2.5 (1.3, 4.9)	14.1	3.0 (2.2, 4.2)	1.7 (1.1, 2.6)

Notes. The Wave 3 all-wave weights were used for the analyses presented in this table. Therefore, the individuals represented in this table had to be present at all three waves – Wave 1, Wave 2, and Wave 3. Bolded values indicate statistical significance (p < 0.05).

^aThe logistic regression model included the following covariates: age, sex, race/ethnicity, Wave 2 receipt of a free tobacco product sample, and Wave 3 coupon or promotions receipt.

^bThe logistic regression model included the following covariates: age, sex, race/ethnicity, Wave 2 receipt of a free tobacco product sample, Wave 3 coupon or promotions receipt, and Wave 1 and Wave 2 past 30-day tobacco use.

^cThe logistic regression model included the following covariates: age, sex, race/ethnicity, sexual orientation, income, education, Wave 2 receipt of a free tobacco product sample, and Wave 3 coupon or promotions receipt.

^dThe logistic regression model included the following covariates: age, sex, race/ethnicity, sexual orientation, income, and education, Wave 2 receipt of a free tobacco product sample, Wave 3 coupon or promotions receipt, and Wave 1 and Wave 2 past 30-day tobacco use.

Receipt of a free ENDS or e-liquid sample at Wave 3 was associated with ever use, past 30-day use, and new ENDS use in the past 12 months among youth and adults at Wave 3 (Supplemental Table 1).

3.3. Free sample receipt and subsequent tobacco use

Among adults, free sample receipt at Wave 2 was associated with greater odds of ever tobacco use at Wave 3 (AOR: 5.2; 95%CI: 1.6, 16.7) (Table 3). Free sample receipt in adults was not associated with past 30-day use or new tobacco product use one year later.

Table 3.

Associations of free tobacco product sample receipt in Wave 2 (2014–2015) with subsequent tobacco use in Wave 3 (2015–2016).

Youth		Wave 3 ever tobacco use			Wave 3 past 30-day tobacco use			New product use between Wave 2 and Wave 3
Count (weighted %) [SE]		3172 (36.1) [0.59]			1307 (14.69) [0.46]			1358 (15.3) [0.44]
	No. of observations	Weighted %	OR (95%CI)	aORa (95%CI)	Weighted %	OR (95%CI)	aORb (95%CI)	Weighted %	OR (95%CI)	aORb (95%CI)
Wave 2 free tobacco product sample receipt
No	9231	34.0	1 [ref]	1 [ref]	14.2	1 [ref]	1 [ref]	14.9	1 [ref]	1 [ref]
Yes	71	87.4	14.0 (6.4, 31.0)	12.2 (5.4, 27.6)	57.2	9.2 (5.4, 15.8)	2.7 (1.1, 6.5)	38.8	3.9 (2.1, 7.1)	2.2 (1.05, 4.4)
Adults		Wave 3 ever tobacco use			Wave 3 past 30-day tobacco use			New product use between wave 2 and wave 3
Count (weighted %) [SE]		21,668 (72.7) [0.59]			12,767 (28.1) [0.32]			2292 (5.2) [0.16]
	No. of observations	Weighted %	OR (95%CI)	aORc (95%CI)	Weighted %	OR (95%CI)	aORd (95%CI)	Weighted %	OR (95%CI)	aORd (95%CI)
Wave 2 free tobacco product sample receipt
No	25,026	73.7	1 [ref]	1 [ref]	28.2	1 [ref]	1 [ref]	5.2	1 [ref]	1 [ref]
Yes	323	96.5	9.1 (2.9, 27.8)	5.2 (1.6, 16.7)	72.1	6.7 (4.1, 10.9)	1.0 (0.5, 2.1)	10.5	2.1 (1.4, 3.1)	1.1 (0.7, 1.6)

Notes. The Wave 3 all-wave weights were used for the analyses presented in this table. Therefore, the individuals represented in this table had to be present at all three waves – Wave 1, Wave 2, and Wave 3. Bolded values indicate statistical significance (p < 0.05).

^aThe logistic regression model included the following covariates: age, sex, race/ethnicity, Wave 3 receipt of a free sample of e-cigarettes or e-liquid, Wave 3 receipt of a free sample of a tobacco product, and Wave 2 coupon or promotions receipt.

^bThe logistic regression model included the following covariates: age, sex, race/ethnicity, Wave 3 receipt of a free sample of e-cigarettes or e-liquid, Wave 3 receipt of a free sample of a tobacco product, Wave 2 coupon or promotions receipt, and Wave 1 and Wave 2 past 30-day tobacco use.

^cThe logistic regression model included the following covariates: age, sex, race/ethnicity, sexual orientation, income, education, and Wave 2 coupon or promotions receipt.

^dThe logistic regression model included the following covariates: age, sex, race/ethnicity, sexual orientation, income, education, Wave 2 coupon or promotions receipt, and Wave 1 and Wave 2 past 30-day tobacco use.

Among youth, receipt of a free tobacco product sample at Wave 2 was associated with greater odds of ever tobacco product use (AOR: 12.2; 95%CI: 5.4, 27.6), past 30-day use (AOR: 2.7; 95%CI 1.1, 6.5), and new product use (AOR: 2.2; 95%CI: 1.0, 4.4) one year later at Wave 3.

4. Discussion

This study estimates that over 1.9 million adults and over 140,000 youth received a free tobacco product sample in 2014–2015 and that over 3.9 million adults and 215,000 youth received a free sample of an ENDS or e-liquid in 2015–2016. In addition, our findings indicate that a greater proportion of vulnerable subgroups including Black, lesbian/gay/bisexual, lower income respondents, and those with less than a bachelor’s degree reported receipt of free samples compared to other groups, aligning with studies showing disparities in tobacco marketing and promotions (Moran et al., 2019; Lee et al., 2015). Receipt of a free tobacco product sample as well as an ENDS product or e-liquid was associated with ever and past 30-day tobacco use, and ENDS use status among youth and adults. Free tobacco product sample receipt was associated with new product use among youth one year later. Therefore, we confirm cross-sectional findings that free tobacco samples are associated with new use with longitudinal analyses in youth.

The association between free sample receipt and ever and past 30-day tobacco and ENDS use could indicate that current users were more likely to receive a free sample or that free sample receipt in the prior 30 days led to further product use. In examining the temporal order of receipt and subsequent use, we observed a prospective association between receipt and subsequent use of a new tobacco product among youth but not adults. These findings suggest that free sample receipt could contribute to trial among youth.

Our study contributes to the evidence on tobacco marketing (Cruz et al., 2018; Pierce et al., 1998; Henriksen et al., 2010; Lovato et al., 2011; Pierce et al., 2018; Soneji et al., 2014). Similar to our findings, Rose et al. (2018) observed an increase in new product use and past 30-day product use one year later among youth who received a tobacco coupon at Wave 1 in the PATH Study. Soneji et al. (2018) examined a related measure of online engagement that included coupon receipt but not free samples and found that among youth in the PATH Study, coupon receipt was associated with an increase in trial, frequency of use, progression to poly-use, and decrease in cessation one year later.

Although we observed a cross-sectional association between receipt of a free sample in adults and tobacco product use, including trial of a new product, the relationship was not sustained prospectively. One possible explanation is that adults try new products at much lower rates than youth, and therefore, free samples serve more to encourage switching between brands but not to take up new products. However, since we did not look at the specific tobacco brands used, or the brands of the free samples received, we are unable to assess whether brand switching within a product category occurred. Another possible explanation is that adults are less sensitive to price than youth, and therefore, receiving a free product does not engage adults in the same way that it may encourage additional trial in youth.

Free tobacco product samples are associated with product trial and subsequent use among youth, something that was hypothesized but not confirmed with data. Numerous studies have found that marketing practices, including exposure to and receipt of price promotions such as coupons, contribute to susceptibility to tobacco use (Choi et al., 2018a; Pierce et al., 2018) and use behavior in adults (El-Toukhy et al., 2018). While our study did not find a relationship between free sample receipt and subsequent use in adults, understanding how distribution of free samples of smokeless products to adults shift in the absence of other free tobacco product samples would be informative. Choi et al. (2018a) found that receipt of snus coupons was associated with susceptibility to snus use. Local US jurisdictions, including New York City and Providence, Rhode Island enacted restrictions on price promotions including coupons (McLaughlin et al., 2014). Therefore, it may be useful to examine whether coupons or discounts can lower the price of a tobacco product in a way that would function similarly to free samples.

4.1. Study limitations

The current analysis has several limitations. First, we are unable to assess the extent to which free sample receipt resulted in regular product use versus one-time product trial. We also do not know whether free sample receipt and ever or current use is a result of users seeking out free samples or being likely recipients of a sample due to their use status. We addressed the issue of directionality in part by extending our analyses to evaluate a possible association between Wave 2 tobacco product sample receipt and Wave 3 tobacco use outcomes including new use of a tobacco product which looked at transitions from never use of a given product to new use in the 12 months after free sample receipt. Second, the question used to assess free tobacco product sample receipt at Wave 2 did not specify the type of product received. Therefore, we do not know if the subsequent tobacco products used are the same types of products received as free samples. Third, we did not examine changes in use behavior, such as increased tobacco use frequency or use of multiple products. An analysis of more waves of PATH Study data would allow examination of the relationship between free ENDS sample receipt and use patterns including sustained ENDS use, as well as dual use. Fourth, we do not know the extent of question misinterpretation which may lead to an overestimation of free sample receipt; specifically, we do not know if adults and youth are interpreting the free sample question as a coupon from a tobacco manufacturer or a friend handing them a product to try. While there is the potential for misclassified reporting of free sample receipt, the free samples questions follow other questions on tobacco marketing, advertising and promotions, which may have reduced misclassification. Despite these limitations, this is the first analysis to establish the temporal order of free sample receipt and youth new product use.

4.2. Conclusions

To our knowledge, this is the first analysis to describe the prevalence of free sample receipt in a nationally representative sample of US youth and adults. The data used in the current analyses were collected in 2014–2016 when free samples of some tobacco products including e-cigarettes were still available. During this time, receiving a product for free meant more youth trying tobacco products at the time of receipt and one year later. Since 2016, only free samples of smokeless tobacco products are allowed for adults in the US in qualified adult-only facilities. The study findings reinforce that banning free samples in youth is a way to reduce youth access and may help reduce trial and subsequent use.