<<

Pack Modifications Influence Perceptions of E-

Amy M. Cohn, PhD Amanda L. Johnson, MHS Haneen Abudayyeh, MPH Bonnie King, MHS Jess Wilhelm, PhD

Objectives: package colors and descriptors influence attitudes and intentions to use. This study examined the impact of flavor, color, and descriptors on electronic (e-cigarette) packages young adults’ perceptions of e-cigarettes. Methods: We recruited 2872 US participants ages 18-24 from Amazon Mechanical Turk (2018-2019) and randomized them to view one of 7 e-cigarette package images that varied by flavor (menthol vs tobacco), color (green or brown vs black and white), and descriptor (present vs absent). Models examined main and interactive effects of flavor, color, and descriptor on perceptions of appeal, harm, and addictiveness, and the moderating effects of product appeal. Results: Menthol e-cigarette packages were rated as more “attention grabbing,” “appetizing,” and “fun to use.” Perceptions of harm and addictiveness did not vary across package conditions. Interactions of menthol pack conditions with appeal emerged. Specifically, participants exposed to the green package with the menthol descriptor reported low e-cigarette harm perceptions across all levels of “attention grabbing” and “discour- ages use,” while those exposed to the green package without the menthol descriptor or the brown package with the tobacco descriptor reported lower harm perceptions as ratings of prod- uct appeal increased. Conclusions: Colors and descriptors on e-cigarette packaging influence product appeal and harm perceptions.

Key words: ; menthol; e-cigarettes; young adults; perceived harm; perceived addictiveness; appeal; packaging; marketing; tobacco companies Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1

urveillance data show that the prevalence of bacco smoke or vapor, making it easier for new us- current e-cigarette use has increased signifi- ers to initiate tobacco use. Flavors also have been cantly among young adults ages 18-24,1 and posited to enhance the appeal and attractiveness Sflavors are a prominent reason for use.2-6 E-ciga- of tobacco products because of their high reward rettes are electronic or battery-powered devices that value and history of associations with and heat liquid, usually containing , into aero- other food items.7 Experimentation with flavored solized content that can be inhaled. Findings from tobacco products has been linked to progression the PATH Study (2014-2015) show that 66.7% to regular tobacco use and nicotine dependence in of past 30-day young adult e-cigarette users report youth and young adults.8-13 use of a flavored e-cigarette.5 Flavors have been The US Food and Drug Administration (FDA) hypothesized to mask the harshness of inhaled to- issued regulatory action in January 2020 ban-

Amy M. Cohn, Associate Professor, Department of Pediatrics, TSET Health Promotion Research Center, Stephenson Cancer Center, University of Oklahoma Health Sciences Center, Oklahoma City, OK, . Amanda L. Johnson, Senior Research Biostatistician, TSET Health Promo- tion Research Center, University of Oklahoma Health Sciences Center, Oklahoma City, OK, United States. Haneen Abudayyeh, Researcher, Public Health Center for Substance Use Research, Battelle Memorial Institute, Baltimore, MD, United States. Bonnie King, Research Project Manager, Pub- lic Health Center for Substance Use Research, Battelle Memorial Institute, Baltimore, MD, United States. Jess Wilhelm, Principal Research Scientist and Biostatistician, Public Health Center for Substance Use Research, Battelle Memorial Institute, Baltimore, MD, United States. Correspondence Dr Cohn; [email protected]

Tob Regul Sci.™ 2021;7(2):87-102 87 Pack Modifications Influence Perceptions of Menthol E-cigarettes ning cartridge-based flavored e-cigarettes, includ- tally examined whether modifying different aspects ing fruit and mint flavors. However, menthol and of e-cigarette packaging (flavor, descriptor, color) tobacco flavored e-cigarettes, both cartridge and influences perceptions of appeal, harm, and ad- non-cartridge-based, remain available to consum- dictiveness, using a large sample of over 2800 US ers. This is concerning because menthol is one of young adults. The first objective examined the the top 3 most common e-cigarette flavors used by main and interactive effects of flavor type (men- young adults.5 Published data suggest that sales of thol vs tobacco), package color (color vs black and menthol e-cigarettes could increase following the white/B&W), and flavor descriptor (present or ab- removal of fruit and candy-flavored e-cigarettes sent) on perceptions of e-cigarette appeal, harm, from the market.14 Furthermore, many new dispos- and addictiveness. The second objective examined able e-cigarette devices have entered the market to the associations between ratings of appeal with per- provide flavors that have been banned in cartridge- ceptions of e-cigarette harm and addictiveness, and based e-cigarettes. the moderating effect of appeal ratings on the as- Packaging and marketing also influence the al- sociations of menthol packages with perceptions of lure and perceived attractiveness of flavored to- e-cigarette harm and addictiveness. We specifically bacco products.7,15,16 Tobacco companies have long recruited a general sample of young adults rather manipulated the characteristics of tobacco prod- than a selected group defined by their tobacco use ucts through packaging and marketing, targeted to status (eg, susceptible e-cigarette users, never e-cig- specific vulnerable and at-risk sub-groups, includ- arette users) so that findings could be generalizable ing young adult smokers and non-smokers who are to a larger audience of young adults, who may be susceptible to use.17-20 documents consumers of tobacco products in the future. highlight that youth and young adult smokers are especially curious about trying flavored products.16 METHODS Younger individuals who have never used a tobacco Participants and Procedure product are familiar with candy and fruit flavors Participants were recruited via Amazon Mechani- and less likely to have experience with “tobacco” as cal Turk (AMT) between 2018 and 2019. Eligible a flavor. Thus, images and descriptors on tobacco individuals were ages of 18 and 24, resided in the product packaging that mimic or portray these ap- US, and reported a 95% approval rating on AMT. pealing flavors (eg, cherry, ) may increase AMT is a crowdsourcing platform that allows for desire to use a tobacco product. Young people also rapid and cost-effective study recruitment.23,24 may be more likely to associate the characteriz- AMT workers are compensated for completing ing flavor of “tobacco” with greater perceptions of small, discrete tasks, called “Human Intelligence harm and addiction, given the widespread knowl- Tasks,” or HITs. Workers search the AMT website edge about the deleterious health effects of ciga- 21 for a HIT, which can be pre-viewed before starting rette . This may make tobacco-flavored a task.25,26 This study was completed online. products less appealing. After completing a brief eligibility screen and Certain descriptors on tobacco packaging and providing consent, 2879 participants were ran- advertisements, such as “light,” “mild,” and “low,” domized to view one of 7 different images of an e- have been found to mislead consumers into think- cigarette package in which flavor type (menthol vs ing that the product is less harmful than other to- 22 tobacco), presence or absence of a flavor descriptor, bacco products. As a result, several descriptors and color were manipulated. One of the conditions were banned under the 2009 Family Smoking was a “control condition” that had no flavor de- Prevention and Act (FSPTCA). scriptor and black and white color on the package. However, flavor descriptors on cigarettes and other Twenty participants completed the survey in less tobacco products have not been banned. Few data than 120 seconds and were removed from all analy- exist on the ways in which young adult appeal for ses in an effort to control for inattentiveness to the flavored e-cigarettes is influenced by packaging, task, leaving a total sample of 2859 participants in a product characteristic that can be regulated by the analysis. FDA. To address this gap, this study experimen- The 7 package conditions are described as follows

88 Cohn et al

Figure 1 Experimental Stimuli by Study Condition

No Descriptor “Classic Tobacco” “Magnificent Menthol” Black and Package C White Package A Package F (B&W)

Brown Color Package G Package E

Green Color Package D Package B

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 89 Pack Modifications Influence Perceptions of Menthol E-cigarettes

(Figure 1): (1) control condition (black and white, ever (not current), and past 30-day cigarette and no descriptor); (2) green e-cigarette package with e-cigarette use, as well as ever and past 30-day use menthol descriptor (“Magnificent Menthol”); (3) of menthol flavored e-cigarettes. black and white (B&W) e-cigarette package with Perceived absolute addictiveness. After viewing menthol descriptor; (4) green e-cigarette package the randomized package image, participants were without menthol descriptor; (5) brown e-cigarette asked: “How addictive do you think e-cigarettes are package with tobacco descriptor (“Classic Tobac- to health?” (1 = not at all addictive to 5 = extremely co”); (6) B&W e-cigarette package with tobacco addictive). descriptor; and (7) brown e-cigarette package with- Perceived absolute harmfulness. Participants out tobacco descriptor. We used the Blu e-cigarette were asked: “How harmful do you think e-ciga- brand given its high level of familiarity and pop- rettes are to health?” (1 = not at all harmful to 5= ularity at the time the study was conceptualized 28 extremely harmful). and funded (2017). Menthol was selected given Package appeal. After viewing the randomized its high policy relevance and popularity in young 29-37 package image, participants were asked about 8 dif- adults. For the color conditions, the brown ferent dimensions of package appeal, with the fol- packaging was designed to mimic the brown color lowing item stem: “Now, thinking about the image of a tobacco leaf and the green packaging was de- you just saw of an e-cigarette, how much do you signed to mimic the green color of a mint leaf. Fla- agree or disagree with the following”… with the vor descriptor conditions included text related to following responses (1) the packaging grasped my flavor as well as an associated image. Images that attention (“attention grabbing”); (2) the packag- were shown were from photographs taken of the ing was appealing (“appealing”); (3) the packaging product by project staff and modified by the mar- made the product appear appetizing, like it would keting team at the awarding institution. taste good (“taste good/appetizing”); (4) the packag- Prior to viewing package images, participants ing suggested this product would be fun to vape/ provided information on demographics and tobac- use(“fun to vape”); (5) the packaging put thoughts in co use behavior. After viewing the assigned pack- my mind about not wanting to vape/use the product age image, participants answered questions about (“discourages use”); (6) the packaging gave me good perceived addictiveness, harm, and 8 dimensions of reason to vape/use the product(“encourages use”); appeal of the product image they had just viewed. (7) after seeing this packaging, I would consider try- Surveys took approximately 7 minutes to complete ing one(“intention to try”); and (8) after this pack- and participants were compensated $1. age, my friends would be interested in trying one (“friends would try”). Response options ranged from Measures 1 = strongly disagree to 7 = strongly agree. Higher The measures used in this study were adapted scores indicate greater appeal for all items except for from items in the PATH study.53 These measures “discourages use,” where higher scores indicate lower are also similar to those used in a study published appeal (eg, lower intentions to use). by our team.49 Demographic information. We collected infor- Data Analysis mation on sex, race (white, black, other), ethnic- We first examined demographic and tobacco use ity (Hispanic, non-Hispanic), education (< high factors of the sample, and equivalence across study school, high school/GED/vocational training, conditions on these factors to ensure randomiza- some college or greater), and income (< $20,000; tion was complete. Next, we examined main effects 20,000-$35,000; $35,001-$50,000; $50,001- of ever e-cigarette use and ever use of menthol e- $75,000; > $75,000). cigarettes on ratings of appeal, harmfulness, and Tobacco use. Participants were asked about ever, addictiveness. Ever e-cigarette users were comprised past year, and past 30-day use of cigarettes and e- of current users as well as those who had ever tried cigarettes, as well as ever and past 30-day menthol an e-cigarette in their lifetime. flavored use of e-cigarettes. Variables were created For objective one, analysis of covariance (AN- reflecting mutually exclusive categories of never, COVA) tests were used to compare main effects

90 Cohn et al of e-cigarette flavor (any menthol package versus tiveness, at different levels of appeal (moderator). any tobacco package), package color (any color Comparison 2 examined differences between pack- package versus any B&W package), and descriptor age B versus package D (the green package without (any package with a descriptor versus any package the menthol) on perceived harm and addictiveness, without a descriptor) on perceived harm, addictive- at different levels of appeal (moderator). Com- ness, and the 8 indices of appeal. To do this, 3 new parison 3 examined differences between package variables were created. A new variable capturing B versus package E (the brown package with the menthol packaging was created by coding menthol tobacco descriptor) on perceived harm and addic- packages B, C, and D (Figure 1) as “1” and coding tiveness, at different levels of appeal (moderator). tobacco packages E, F, G as “0”. A new variable cap- In all comparisons, package B was coded as 1 and turing color packaging was created by coding pack- the comparison package was coded as 0. Following ages B, D, E, and G (Figure 1) as “1”, and all other recommendations of Aiken and West,38 using the packages were coded as “0.” Lastly, a new variable relevant package comparison as the independent capturing packages with a descriptor was created variable (eg, comparison 1) and the relevant item by coding packages B, D, D, and F as “1” and all of appeal as the moderator (eg, “fun to use/vape”), other package images as “0.” The control condition separate regression equations were computed by (package A) with no color and no descriptor was entering ever e-cigarette use in the first step, main categorized as “0” for both the color condition vari- effects of package comparison and appeal in the able and the descriptor condition variable. To take second step, and the interaction of package com- a conservative analytic approach, models controlled parison X appeal in the third step. For equations for “ever use” of e-cigarettes (yes = 1; 0 = no), be- with significant interactions, regression coefficients cause prior e-cigarette use may impact product per- for simple effects were examined.38 Unstandardized ceptions. This also aligned with published research betas are reported. All analyses were conducted using a nearly identical experimental manipula- with SPSS 26. For Bonferroni adjustments, SPSS tion,49 but with little /. multiplies the p-value of the least significant differ- To determine interactive effects, ANCOVA tests ences by the number of tests, and produces a new were conducted to examine pairwise comparisons p-value. Thus, the p-value presented in the tables is across each of the 7 conditions on the outcomes an adjusted p-value. Analyses used listwise deletion. of interest, with Bonferroni adjusted pairwise com- parisons. All models controlled for “ever use” of RESULTS e-cigarettes. Sample Characteristics For objective 2, linear regression models were The sample was primarily male (55.8%) and used to examine main and interactive effects of white (72.2%); 16.7% reported Hispanic ethnic- specific menthol packaging and ratings of appeal ity (Table 1). The mean age of respondents was on the outcomes of perceived harm and addictive- 21.37 years (SD = 1.98). For tobacco use histo- ness. All models controlled for “ever use” of e-ciga- ry, 57.7% reported past 30-day cigarette use and rettes. To reduce the number of regression models 31.6% reported past 30-day e-cigarette use. Nearly that were conducted, and thus, the Type 1 error three-fourths of e-cigarette users had ever used a rate, appeal items that were statistically significant menthol flavored e-cigarette. The study conditions in the previous Bonferroni adjusted simple effects were about equally distributed in terms of sample analyses were examined as moderators; these were size (see Table 2 for sample size per condition), and “attention grabbing,” “taste good/appetizing,” “fun there were no statistically significant differences to use/vape,” and “discourages use.” In models test- across study conditions on demographic or tobacco ing for interactions, 3 specific menthol package use factors. comparisons were examined in separate regression models. Comparison 1 examined differences be- tween package B (green package with menthol de- Main Effects of Ever E-cigarette Use and Ever scriptor) versus package C (B&W package with the Use of Menthol E-cigarettes menthol descriptor) on perceived harm and addic- There were statistically significant main effects of ever use of e-cigarettes on perceptions of harm, ad-

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 91 Pack Modifications Influence Perceptions of Menthol E-cigarettes

Table 1 Demographic and Tobacco Use Characteristics of the Sample of Young Adults (N = 2859) Mean Standard Deviation DEMOGRAPHICS Age 21.4 1.98 % N Sex Male 55.8 1594 Female 44.2 1265 Race White 72.2 2060 Black/African-American 11.4 327 Asian 8.0 230 Other 8.4 241 Ethnicity Hispanic 16.7 478 Non-Hispanic 83.3 2381 Income < $20,000 17.8 509 $20,000-$35,000 23.8 680 $35,001-$50,000 19.2 549 $50,001-$75,000 19.7 562 > $75,000 16.2 462 Prefer not to say 3.3 95 Education < High school degree 2.2 62 High school 30.8 881 degree/GED/vocational training Some college education or higher 67.0 1916 TOBACCO USE BEHAVIOR Cigarette use Never 21.1 604 Ever (not past 30-day) 21.1 604 Past 30-day 57.7 1651 E-cigarette use Never 45.0 1287 Ever (not past 30-day) 23.3 668 Past 30-day 31.6 904 Menthol e-cigarette use a Never 26.1 411 Ever 73.9 1161

Note. a Among ever e-cigarette users. Never, ever, and past 30-day use are mutually exclusive categories.

92 Cohn et al

Table 2 Main Effects of E-cigarette Menthol Flavoring, Package Color, and Presence of a Flavor Descriptor on Perceptions of E-cigarette Product Harm, Addictiveness, and Appeala Flavor Color Descriptor Black & No Menthol Tobacco Color Descriptor White Descriptor N = 1229 N = 1221 N = 1634 N = 1634 N = 1225 N = 1224 M (SD) M (SD) p M (SD) M (SD) p M (SD) M (SD) p

Harm 3.21 (1.0) 3.21 (1.1) .67 3.22 (1.0) 3.18 (1.05) .44 3.19 (1.0) 3.22 (1.0) .53 Addictiveness 3.53 (1.1) 3.47 (1.3) .19 3.50 (1.1) 3.50 (1.1) .99 3.51 (1.1) 3.45 (1.1) .18 Appeal

Attention grabbing 4.39 (1.6) 4.24 (1.7) .03 4.39 (1.7) 4.18 (1.7) .001 4.30 (1.7) 4.30 (1.7) .91 Appealing 4.52 (1.6) 4.39 (1.7) .07 4.49 (1.7) 4.42 (1.7) .25 4.48 (1.6) 4.43 (1.7) .46 Taste good 4.31 (1.7) 4.10 (1.8) .005 4.25 (1.8) 4.08 (1.7) .01 4.24 (1.7) 4.09 (1.8) .04 Fun to vape 4.22 (1.7) 4.03 (1.7) .007 4.16 (1.7) 4.06 (1.8) .12 4.14 (1.7) 4.08 (1.8) .39 Discourages Useb 3.28 (1.8) 3.40 (1.8) .10 3.31 (1.8) 3.40 (1.8) .14 3.36 (1.8) 3.34 (1.8) .74 Encourages Use 3.60 (1.7) 3.51 (1.7) .22 3.56 (1.7) 3.55 (1.7) .88 3.58 (1.7) 3.52 (1.7) .36 Intention to try 3.65 (1.9) 3.55 (1.9) .27 3.59 (1.9) 3.62 (1.9) .82 3.62 (1.9) 3.59 (1.9) .84 Friends would try 4.04 (1.8) 3.91 (1.8) .10 3.99 (1.8) 3.99 (1.8) .96 4.00 (1.8) 3.97 (1.8) .69

Note. All models control for ever use of e-cigarettes. Unadjusted means and standard deviations are reported. The p-values from adjusted models are reported. F-values and degrees of freedom are presented in the text. Items in bold are signifi- cantly different.

a Menthol flavor condition was coded as “1” for the following images: the green menthol package with the menthol descriptor (package B), the green menthol package without the menthol descriptor (package D), and the black and white (B&W) menthol package with the menthol descriptor (package C); menthol flavor condition was coded as “0” for the following images: the brown tobacco package with the tobacco descriptor (package E), the brown tobacco package without the tobacco descriptor (package G), and the B&W tobacco package with the tobacco descriptor (package F). Color condition was coded as “1” for the following images: the green menthol package with the menthol descriptor (package B), the green menthol package without the menthol descriptor (package B), the brown tobacco package with the tobacco descriptor (package E), and the brown tobacco package without the tobacco descriptor (package G); all other packs were coded as “0”. Descriptor condition was coded as “1” for the following images: the green menthol package with the menthol descriptor (package B), the B&W menthol package with the menthol descriptor (package C), the brown tobacco package with the tobacco descriptor (package E), or the B&W tobacco package with the tobacco descriptor (package F); all other images were coded as “0”. The control package with no flavor and no descriptor (package A) was coded as “0” for both the color and descriptor conditions. b Higher scores indicate lower appeal (eg, lower intentions to use); for all other indices of appeal, higher scores indicate greater appeal. dictiveness, and all indices of appeal (all ps < .05). fects of ever menthol e-cigarette use on perceptions Specifically, compared to never e-cigarette users, of harm, and several indices of appeal (“taste good/ ever e-cigarette users reported lower perceptions of appetizing,” “fun to use,” “encourages use,” “inten- e-cigarette harm and addictiveness, and rated e-cig- tions to try,” and “friends would try”) (all ps < .05). arette packages as more appealing (Supplemental Supplemental Table 1 (S1) shows the mean differ- Table S1). ences between ever (vs never) e-cigarette users and There were also statistically significant main -ef menthol (vs non-menthol) e-cigarette users.

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 93 Pack Modifications Influence Perceptions of Menthol E-cigarettes

Main Effects of Flavor Type, Package Color, tobacco descriptor; p = .032). Lastly, package B and Package Descriptor received lower ratings on “discourages use” com- Table 2 shows ANCOVA results of the main pared to both package C (the B&W package with effects of flavor condition (menthol vs tobacco), the menthol descriptor; p = .013) and package E package color (color vs B&W), and flavor descrip- (the brown package with the tobacco descriptor; p tor (present vs absent) on perceptions of e-cigarette = .015). No other significant differences between harm, addictiveness, and package appeal, control- conditions were found. ling for “ever use” of e-cigarettes. There were sta- Interaction of menthol package conditions and tistically significant main effects of menthol (vs ratings of appeal on perceptions of harm and tobacco) packages on several indices of appeal: addictiveness. After controlling for “ever use” of “attention-grabbing” [F(1, 2433) = 4.61, p = .03], e-cigarettes, regression analyses revealed signifi- “taste good/appetizing” [F(1, 2433 = 7.99, p = cant differences in e-cigarette harm perceptions .005], and “fun to use” [F(1, 2433) = 7.20, p = between those who viewed package B (the green .007), where menthol packages were rated as sig- package with the menthol descriptor) versus those nificantly more appealing than tobacco packages who viewed package C (the green package without on these indices. the menthol descriptor), as a function of different There were also statistically significant main -ef ratings of “attention-grabbing” (interaction effect: fects of package color (versus B&W) on “attention b = .09, p = .04). Specifically, Figure 2 shows that grabbing” (F(1, 2841) = 10.73, p = .001) and “taste participants exposed to package B showed low e- good/appetizing” [F(1, 2841 = 6.31, p = .012], cigarette harm perceptions across low and high rat- where packages in color were rated as significantly ings of “attention grabbing” (b = .000, p =.992), more appealing than packages in B&W (see Table while those exposed to the green package without 2 for mean differences). the menthol descriptor reported lower harm per- There was a statistically significant main effect of ceptions at higher ratings of “attention-grabbing” descriptor (versus no descriptor) on “taste good/ (b = -.09, p = .004). Thus, ratings of e-cigarette appetizing” [F(1, 2841 = 4.34, p = .037], where harm remained the same across low versus high packages with a flavor descriptor were perceived as ratings of product appeal (“attention grabbing”) significantly more appetizing than packages with- among those exposed to the green package with out a descriptor. No other statistically significant the menthol descriptor; while ratings of package main effects emerged. appeal appeared to have an impact at reducing risk perceptions when individuals view the green e-cigarette package where the menthol descriptor Simple Effects Comparisons of Flavor Type, was missing. Package Color, and Package Descriptor Similar to the analyses above, after controlling Table 3 shows mean differences between pack for “ever use” of e-cigarettes, regression analyses conditions on perceptions of e-cigarette harm, ad- revealed significant differences in e-cigarette harm dictiveness, and package appeal, and results of sim- perceptions between those who viewed package B ple effects comparisons of flavor type, color, and and those who viewed package E (the brown pack- descriptor controlling for “ever use” of e-cigarettes. age with the tobacco descriptor), as a function of Bonferroni-adjusted ANCOVA results revealed different ratings of “discourages use” (interaction significant differences between package B (the effect: b = -.08, p = .04). Figure 3 shows that partic- green package with the menthol descriptor) and ipants exposed to package B endorsed consistently package F (the B&W package with the tobacco low ratings of e-cigarette harm perceptions across descriptor), where package B was rated as signifi- low and high ratings of product appeal (b = .01, p cantly more appealing on “attention-grabbing” (p = .980), while participants exposed to the brown = .006), “taste good/appetizing” (p = .001), and package with the tobacco descriptor reported lower “fun to use/vape” (p = .005). Further, package B harm perceptions at lower ratings of “discourages was rated as significantly more appealing on “taste use” (b = .08, p = .004). Thus, ratings of e-cigarette good/appetizing” compared to the control package harm remained the same across ratings of low ver- (p = .001) and package G (brown package with no

94 Cohn et al

G – Brown, M (SD) 4.12 (1.8) (N = 405) Descriptor No Tobacco No Tobacco

F – B&W, B&W, M (SD) Tobacco Tobacco 4.08 (1.7) 3.99 (1.8) 3.91 (1.7) (N = 408) Descriptor

E – Brown, Brown, M (SD) Tobacco E-cigarette Package Images E-cigarette Tobacco Tobacco Tobacco 3.52 (1.8) (N = 408) Descriptor

D – Green, M (SD) (N =410) Descriptor No Menthol Condition

C – B&W, B&W, M (SD) Menthol 3.51 (1.8) (N = 408) Descriptor

B – Menthol E-cigarette Package Images Menthol E-cigarette Green, Green, M (SD) Menthol (N =411) 4.50 (1.6) 4.50 (1.7) 4.36 (1.7) 3.10 (1.8) Table 3 Table Descriptor for all other indices of appeal, higher scores indicate greater appeal. indicate greater scores indices of appeal, higher all other for

No A – B&W, B&W, M (SD) Control 4.00 (1.8) (N = 409) Descriptor E-cigarette Product Harm, Addictiveness, and Appeal Addictiveness, Harm, Product E-cigarette

a Statistically significant pairwise comparisons (p-values) NS NS B vs. F (.006) NS (.001), A B vs. B vs. F (.001), B vs. G (.032) B vs. F (.005) B vs. C (.013), B vs. E (.015) NS NS NS Simple Effects Comparisons of Flavor, Package Color, and Package Descriptor on Perceptions of Perceptions Descriptor on Package and Package Color, Flavor, Effects Comparisons of Simple b Only statistically significant Bonferroni adjusted p-values are reported. Unadjusted means and standard deviations are reported. All other pairwise All other reported. reported. Unadjusted means and standard deviations are adjusted p-values are Only statistically significant Bonferroni Higher scores indicate lower appeal (eg. lower intention/desire to use); intention/desire appeal (eg. lower indicate lower scores Higher Perceived Harm Perceived Addictiveness Perceived Appeal Attention grabbing Appealing good/appetizing Taste Fun to vape Discourages Use Encourages Use Intention to try Friends would try Note. a not statistically significant. comparisons were b

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 95 Pack Modifications Influence Perceptions of Menthol E-cigarettes

Figure 2 Effects of Exposure to Menthol E-cigarette Package Images (with and without the menthol flavor descriptor) and Ratings of Product Appeal (“attention grabbing”) on E-cigarette Harm Perceptions

3.8

3.7 b = -.09, p = .004 3.6

3.5

3.4 b = .000, p = .992 3.3

3.2 cigarette harm perceptions cigarette - Menthol flavor, green color, no menthol descriptor E 3.1 Menthol flavor, green color, with menthol descriptor 3 Low Attention grabbing High Attention grabbing

sus product appeal (eg, “discouraged use) among less likely to discourage use) appeared to decrease those exposed to the green package with the men- risk perceptions among those who viewed a tradi- thol descriptor; while ratings of package appeal (eg, tional tobacco e-cigarette package.

Figure 3 Effects of Exposure to Menthol and Tobacco Flavored E-cigarette Package Images and Ratings of Product Appeal (“discourages use”) on E-cigarette Harm Perceptions

3.8

3.7

3.6 b = .08, p = .04 3.5

3.4

3.3 b = .01, p = .980 cigarette harm perceptions cigarette -

E 3.2 Tobacco flavor, brown color, with tobacco descriptor 3.1 Menthol flavor, green color, with menthol descriptor

3 Low Discourages Use High Discourages Use

96 Cohn et al

Other Main Effect and Interaction Models io, as certain menthol e-cigarettes are still available Regression models revealed significant main ef- on the market. Consistent with our expectations, fects of appeal ratings on perceptions of e-cigarette this “real world” menthol e-cigarette package was harm and addictiveness (all ps < .01), such that rated as more appealing than the “real world” to- greater product appeal was associated with lower bacco e-cigarette package (eg, in color, with the perceptions of e-cigarette harm and addictiveness. descriptor) on multiple dimensions of appeal – There was only one exception in which “discour- “attention grabbing,” “taste good/appetizing,” and ages use” was unrelated to perceptions of addictive- “fun to use/vape.” Whereas the FDA does have the ness (Supplemental Table 2 [S2]). authority to institute plain packaging or ban the All models were re-examined with ever e-cigarette use of flavor descriptors, as these actions fit under use and ever menthol e-cigarette use as modera- the domain of marketing and label regulation, the tors, but no significant interaction were found. We FDA would need to propose a rule, support this conducted ANCOVAs comparing green packages rule with scientific evidence, and likely defend its (packages B and C; coded as 1) to brown packages decision in the courts. Our study findings support only (packages E and G; coded as 0) on perceived the notion that some aspects of tobacco product harm, addictiveness, and ratings of appeal. There packaging (colors and descriptors) contribute to were no significant differences on any of the out- increased attraction to flavored products and pro- comes. Furthermore, there were no significant in- vide some empirical support to the FDA in this teractions of green versus brown packages with any respect. Individuals exposed to packages in color indices of appeal (moderators) on perceived harm (compared to those in B&W), and packages with and addictiveness. a descriptor (compared to those without a descrip- tor) did not report lower perceptions of harm and addictiveness, but did report greater product ap- Discussion peal on several indices (“attention grabbing,” “taste Our study revealed several important findings good/appetizing”). about the effects of e-cigarette package modifica- For objective 2, main effects showed that higher tions on perceptions of product appeal, e-cigarette product appeal overall was associated with lower harm, and addictiveness. Consistent with previous- 39,40 perceptions of harm and addictiveness, even after ly published literature, young adults who had controlling for e-cigarette use history. Furthermore, ever used an e-cigarette reported lower perceptions ratings of post-exposure product appeal influenced of e-cigarette harm and addictiveness, and higher (moderated) differences between certain menthol ratings of e-cigarette product appeal compared to packages with respect to post-exposure assessment never users, regardless of which package image they of harm perceptions and addictiveness. Overall, in- were randomized to view. For objective one, main teraction analyses showed participants randomized effect analyses showed that any package that was to view the green package with the menthol descrip- intended to depict a menthol e-cigarette (regard- tor (package B) indicated consistently low levels of less of the presence/absence of a flavor descriptor e-cigarette risk at any level of product appeal. Thus, or package color) was perceived as more attention- product appeal did little to influence these respon- grabbing, appetizing, and fun to use compared to dents’ perceptions of the e-cigarette product they any package that was intended to depict a tobacco had just viewed. This might be because this pack- flavored e-cigarette. Furthermore, simple effects age condition had higher appeal ratings, in general, analyses showed that the green package with the on “attention grabbing,” “taste good/appetizing,” menthol descriptor (package B) was rated higher “fun to use/vape,” and lower ratings on “discourages on “attention grabbing,” “taste good/appetizing,” use” compared to a number of other package condi- and “fun to use/vape” and lower on “discourages tions, as evidenced by the simple effects analyses. In use” compared to several other packages, even after contrast, perceptions of e-cigarette harm decreased controlling for differences in e-cigarette use history. significantly as ratings of appeal increased, specifi- It is worth noting that the green menthol package cally among participants who viewed the brown with the mint/menthol leaf descriptor aligns with tobacco package with the descriptor or the green the current “real world” e-cigarette market scenar- menthol package without the menthol descriptor.

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 97 Pack Modifications Influence Perceptions of Menthol E-cigarettes

Thus, higher ratings of package appeal appeared sible that a single exposure, rather than multiple to influence lower risk perceptions, but only when exposures over time (such as at the point of sale), participants viewed the e-cigarette package with- has little impact on changing behavior, but does out the menthol descriptor, or the brown tobacco change attitudes in the interim. package with the tobacco descriptor. This further There are several reasons to explain our results. highlights the potentially detrimental combination First, although exposure to different packs did not of menthol flavoring, package color, and the pres- appear to directly impact harm perceptions or in- ence of a menthol flavor descriptor young people’s tentions to use, they did appear to lead to greater perceptions of tobacco product risk. product appeal, and higher product appeal was Whereas the main effects results broadly compare found to be associated with perceptions that are e- differences across a variety of package conditions, cigarettes are less harmful and less addictive. This the Bonferroni-adjusted simple effects compari- is consistent with published work, showing that e- sons specifically examined differences comparing cigarette users are more receptive to marketing and across each individual study condition (rather than this receptivity is related to lower e-cigarette harm across combined conditions). Thus, results from perceptions, which in turn predict use behavior.39 the main effects of package flavor, for example, tell This is particularly alarming as recent data show us information, globally, about the potential im- that sales of menthol flavored e-cigarette pact of menthol flavoring on e-cigarette product products have increased after JUUL limited its sales perceptions, while results from the pairwise com- of fruit and candy flavors.14 Second, prior research parisons provide greater granularity about which also suggests that menthol cigarettes are popu- specific package characteristics (flavor, color, and lar among young adults because the cooling and descriptor) appear to influence these perceptions. “throat soothing” properties mask the harshness of Although we did not find differences between green inhaled tobacco smoke.41 This notion may extend and brown colored packs, we did find that color, to e-cigarettes and may explain why some menthol overall, did have some impact at increasing product packages were perceived as more appealing. Third, appeal. Such information provides supporting evi- given the widespread use of tobacco marketing tar- dence to the FDA if it moves forward with stricter geting young adults,16,42 it is possible this age group regulations of the marketing of tobacco products to associates the green color of an e-- reduce their appeal to younger individuals. age with a mint/menthol flavored tobacco product Overall, our study findings showed that menthol – much as they would with a menthol e-cigarette packages were rated higher on several cigarette.52 Young adults may also associate the im- domains of product appeal (3 out of 8). Discrete age of a mint leaf with a menthol flavored product, choice experiments (DCEs) testing the influence even without having had prior e-cigarette use expe- of e-cigarette products characteristics (eg, flavor, rience. It is possible that product appeal mediates warning labels) on choice behavior and intentions the association between package exposure and atti- to use have found similar results.54-56 In a DCE tudes about e-cigarette-related harm. Longitudinal that examined the impact of varying attributes of studies examining causal links between exposure, e-cigarettes, including flavors, on adolescents’ in- appeal, and perceptions of e-cigarette harm would tentions to use e-cigarettes, Shang et al54 found that need to be conducted to make this determination. flavors had the most pronounced impact at increas- According to the Tobacco Control Act, tobacco ing the probability of choosing e-cigarettes among packaging that is misleading to consumers in mate- both “never users” and “ever users.” In our study, rial ways directly violates the Tobacco Control Act even though some young adults may have rated [ Sec. 903(a)(1) and (7)], even if there is no intent certain e-cigarette packages as more appealing, by the manufacturer to do so. Consequently, results they did not seem to feel more motivated to try from this study could provide support for the FDA these products. This might be acceptable for public to take enforcement actions against some flavored health if lower product appeal reduces intentions tobacco product marketing techniques in the US to use in the future; however, this would need to in attempts to reduce product appeal. In our study, be measured in longitudinal studies. It is also pos- exposure to certain menthol e-cigarette package im-

98 Cohn et al ages did appear to lead to greater product appeal, investigation about attitudes about tobacco prod- which in turn was correlated with lower perceptions ucts. Other tobacco-related studies conducted on of e-cigarette harm and addictiveness. In addition, this platform report similarly high rates of tobacco whereas menthol in cigarettes and e-cigarettes has use and cigarette smoking.46,49,50 Third, we did not not been banned at the federal level, states and lo- examine effects of package modifications on per- calities have the ability to restrict the time, place, ceptions of JUUL, which has risen significantly in and manner of flavored tobacco product sales. By popularity in this age group. Fourth, we did not focusing on whether menthol flavored tobacco test a wide of variety of e-cigarette flavor profiles, product packaging misleads consumers into be- beyond menthol. Fifth, all models adjusted for dif- lieving that these products are more appealing and ferences between e-cigarette users and non-users. less harmful than traditional tobacco flavored ver- Thus, statistically significant associations between sions of these products, the findings from this study study conditions and the outcomes of interest add an important new dimension to the available were observed beyond these individual differences. research about the abuse liability associated with However, although the use of ANCOVA and re- menthol flavoring and inform policy changes to gression analyses provide some benefits in control- protect public health. Additionally, because there ling for confounding factors, such as e-cigarette were statistically significant main effects of package use, this technique has its limitations and cannot color and package descriptors on some indices of be assumed to provide complete adjustment. Last- appeal, the FDA or local jurisdictions may want to ly, we did not control for individual differences in consider further restrictions on these marketing fac- perceptions of e-cigarette harm and addictiveness tors, in an effort to reduce young consumer appeal prior to randomization. Given the relatively short to use tobacco products. Previous research shows duration of the experiment (5-7 minutes), we were that cigarette pack colors and imagery can impact concerned that pre-randomization assessment of consumer perceptions,54,57,58 and that plain packag- these factors could influence post-randomization ing can reduce product appeal.59 Lastly, our findings ratings. have implications of our findings for Pre-Market Tobacco Product Applications (PMTA), which Implications for Tobacco Regulation are now required for e-cigarettes to remain on the There is a need to understand the impact of to- market. Companies will have to show evidence that bacco product descriptors on tobacco-related harm the introduction of their e-cigarette product could perceptions and behaviors given the rapidly chang- “benefit public health,” and will have to include in- ing tobacco market and policy landscape. Findings formation about possible negative impacts to popu- from this study could inform the development of lation health, such as appeal to young people. Our new policies to reduce the public health impact results suggest that some characteristics, like flavor of tobacco use or the appeal of tobacco products and color on e-cigarette packaging impact measures through marketing.16,51 Stricter regulation of e- of appeal among young people. cigarette packaging may be one method to reduce Study results should be interpreted in light of e-cigarette appeal among this vulnerable age group several limitations. First, this study did not use and thus the prevalence and public health of to- a nationally-representative sample. One demo- bacco use. graphic analysis of AMT workers suggests that the population is predominantly non-Hispanic white and approximates the representativeness of the US Human Subjects Approval Statement population.27 We chose AMT as our platform for The study methods and procedures were ap- data collection because it allows for rapid and cost- proved by the Institutional Review Board of the effective data collection on a discrete task. AMT Battelle Memorial Institute. tasks are increasingly used in public health and addiction research, including for tobacco use, and Conflict of Interest Disclosure Statement 27,43-49 have been widely validated. Second, it is pos- All authors of this article declare they have no sible that we had high rates of cigarette smoking in conflicts of interest. our sample because the study was advertised as an

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 99 Pack Modifications Influence Perceptions of Menthol E-cigarettes

Acknowledgements uary 2015 through October 2019. Am J Public Health. 2020; 110(6):785-787. Research reported in this publication was sup- 15. Manning KC, Kelly KJ, Comello ML. Flavoured ciga- ported by the National Institute on Drug Abuse rettes, sensation seeking and adolescents’ perceptions of award number R03DA042010-01A1S, awarded to cigarette brands. Tob Control. 2009;18(6):459-465. the first author. The content is solely the responsi- 16. Carpenter CM, Wayne GF, Pauly JL, et al. New ciga- rette brands with flavors that appeal to youth: to- bility of the authors and does not necessarily repre- bacco marketing strategies. Health Affairs (Millwood). sent the official views of the National Institutes of 2005;24(6):1601-1610. Health. Portions of this manuscript were presented 17. Anderson SJ. Marketing of menthol cigarettes and con- at the annual Tobacco Regulatory Science Meet- sumer perceptions: a review of tobacco industry docu- ments. Tob Control. 2011;20(Suppl 2):ii20-ii28. ing in October 2019. We thank Lexie Perreras who 18. Cruz TB, Wright LT, Crawford G. The menthol mar- provided project management support, developed keting mix: targeted promotions for focus communities the surveys, and assisted with proofreading. in the United States. Nicotine Tob Res. 2010;12(Suppl 2):S147-S153. 19. Carpenter CM, Wayne GF, Connolly GN. The role of References sensory perception in the development and targeting of 1. Dai H, Leventhal AM. Prevalence of e-cigarette use tobacco products. Addiction. 2007;102(1):136-147. among adults in the United States, 2014-2018. JAMA. 20. Lewis MJ, Wackowski O. Dealing with an innovative in- 2019:322(18):1824-1827. dustry: a look at flavored cigarettes promoted by main- 2. Harrell MB, Loukas A, Jackson CD, et al. Flavored to- stream brands. Am J Public Health. 2006;96(2):244-251. bacco product use among youth and young adults: what 21. US Centers for Disease Control and Prevention (CDC), if flavors didn’t exist?Tob Regul Sci. 2017;3(2):168-173. US Department of Health and Human Services, US Sur- 3. Harrell M, Weaver S, Loukas A, et al. Flavored e-cigarette geon General. The Health Consequences of Smoking – 50 use: characterizing youth, young adult, and adult users. years of Progress: a Report of the Surgeon General. Atlanta, Prev Med Rep. 2017;5:33-40. GA: CDC; 2014. 4. Patel D, Davis KC, Cox S, et al. Reasons for current E-cig- 22. US Department of Health and Human Services. Prevent- arette use among US adults. Prev Med Rep. 2016;93:14- ing Tobacco Use Among Youth and Young Adults: A Report 20. of the Surgeon General. Atlanta, GA: US Department of 5. Soneji SS, Knutzen KE, Villanti AC. Use of flavored e-cig- Health and Human Services, US Centers for Disease arettes among adolescents, young adults, and older adults: Control and Prevention, National Center for Chronic findings from the Population Assessment for Tobacco and Disease Prevention and Health Promotion, Office on Health Study. Public Health Rep. 2019;134(3):282-292. Smoking and Health; 2012. 6. Villanti AC, Johnson AL, Ambrose BK, et al. Flavored 23. Buhrmester M, Kwang T, Gosling SD. Amazon’s Me- tobacco product use in youth and adults: findings from chanical Turk: a new source of inexpensive, yet high- the first wave of the PATH Study (2013–2014). Am J quality, data? Perspect Psychol Sci. 2011;6(1):3-5. Prev Med. 2017;53(2):39-151. 24. Crump MJ, McDonnell JV, Gureckis TM. Evaluating 7. Brown JE, Luo W, Isabelle LM, Pankow JF. Candy flavor- Amazon’s Mechanical Turk as a tool for experimental be- ings in tobacco. N Engl J Med. 2014;370(23):2250-2252. havioral research. PloS One. 2013;8(3):e57410. 8. Hersey JC, Ng SW, Nonnemaker JM, et al. Are menthol 25. Ilakkuvan V, Tacelosky M, Ivey KC, et al. Cameras for cigarettes a starter product for youth? Nicotine Tob Res. public health surveillance: a methods protocol for crowd- 2006;8(3):403-413. sourced annotation of point-of-sale photographs. JMIR 9. Hersey JC, Nonnemaker JM, Homsi G. Menthol ciga- Res Protoc. 2014;3(2):e22. rettes contribute to the appeal and addiction potential 26. Barrett LF, Barrett DJ. An introduction to computerized of smoking for youth. Nicotine Tob Res. 2010;12(Suppl experience sampling in psychology. Soc Sci Comput Rev. 2):S136-S146. 2001;19(2):175-185. 10. Nonnemaker J, Hersey J, Homsi G, et al. Initiation with 27. Kraemer JD, Strasser AA, Lindblom EN, et al. Crowd- menthol cigarettes and youth smoking uptake. Addiction. sourced data collection for public health: a comparison 2013;108(1):171-178. with nationally representative, population tobacco use 11. Wackowski O, Delnevo CD. Menthol cigarettes and indi- data. Prev Med. 2017;102:93-99. cators of tobacco dependence among adolescents. Addict 28. King BA, Gammon DG, Marynak KL, Rogers TJJ. Elec- Behav. 2007;32(9):1964-1969. tronic cigarette sales in the United States, 2013-2017. 12. Collins CC, Moolchan ET. Shorter time to first cigarette JAMA. 2018;320(13):1379-1380. of the day in menthol adolescent cigarette smokers. Ad- 29. Abrams DB. Promise and peril of e-cigarettes: can dis- dict Behav. 2006;31(8):1460-1464. ruptive technology make cigarettes obsolete? JAMA. 13. Villanti AC, Johnson AL, Glasser AM, et al. Association 2014;311(2):135-136. of flavored tobacco use with tobacco initiation and sub- 30. Giovino GA, Villanti AC, Mowery PD, et al. Differential sequent use among US youth and adults, 2013-2015. trends in cigarette smoking in the USA: is menthol slow- JAMA Netw Open. 2019;2(10):e1913804-e1913804. ing progress? Tob Control. 2013;24:28-37. 14. Liber A, Cahn Z, Larsen A, Drope J. Flavored e-cigarette 31. King BA, Alam S, Promoff G, et al. Awareness and ever- sales in the United States under self-regulation from Jan- use of electronic cigarettes among US adults, 2010–2011.

100 Cohn et al

Nicotine Tob Res. 2013;15(9):1623-1627. 2016;18(8):1749-1756. 32. Chen I-L. FDA summary of adverse events on electronic 47. Strickland JC, Stoops WW. The use of crowdsourcing in cigarettes. Nicotine Tob Res. 2013;15(2):615-616. addiction science research: Amazon Mechanical Turk. 33. Regan AK, Promoff G, Dube SR, Arrazola R. Electronic Exp Clin Psychopharmacol. 2019;27(1):1-18. nicotine delivery systems: adult use and awareness of the 48. Moran MB, Brown J, Lindblom E, et al. Beyond ‘natural’: ‘e-cigarette’in the USA. Tob Control. 2013;22(1):19-23. cigarette ad tactics that mislead about relative risk. Tob 34. Pepper JK, Brewer NT. Electronic nicotine delivery sys- Regul Sci. 2018;4(5):3-19. tem () awareness, use, reactions and 49. Evans AT, Wilhelm J, Abudayyeh H, et al. Impact of beliefs: a systematic review. Tob Control. 2014;23(5):375- package descriptors on young adults’ perceptions of ciga- 384. rillos. Tob Regul Sci. 2020;6(2):118-135. 35. Pearson JL, Abrams DB, Niaura RS, et al. A ban on men- 50. Mays D, Villanti A, Niaura RS, et al. The effects of vary- thol cigarettes: impact on public opinion and smokers’ in- ing electronic cigarette warning label design features on tention to quit. Am J Public Health. 2012;102(11):e107- attention, recall, and product perceptions among young e114. adults. Health Commun. 2019;34(3):317-324. 36. Benowitz NL, Goniewicz ML. The regulatory challenge 51. Wackowski OA, Evans KR, Harrell MB, et al. In their of electronic cigarettes. JAMA. 2013;310(7):685-686. own words: young adults’ initiation, 37. US Food Drug Administration (FDA). Tobacco Products perceptions, experiences and regulation perspectives. Nic- Scientific Advisory Committee (TPSAC). Menthol ciga- otine Tob Res. 2017;20(9):1076-1084. rettes and public health: review of the scientific evidence 52. Richardson A, Ganz O, Pearson J, et al. How the industry and recommendations. Rockville, MD: FDA, Center for is marketing menthol cigarettes: the audience, the mes- Tobacco Products; 2011. sage and the medium. Tob Control. 2014; 24(6):594-600. 38. Aiken LS, West SG, Reno RR. Multiple Regression: Testing 53. Hyland A, Ambrose BK, Conway KP, et al. Design and and Interpreting Interactions. Thousand Oaks, CA: Sage; methods of the Population Assessment of Tobacco and 1991. Health (PATH) Study. Tob Control. 2016;26:371–378. 39. Pokhrel P, Fagan P, Kehl L, Herzog TA. Receptivity to 54. Shang C, Huang J, Chaloupka FJ, Emery SL. The impact e-cigarette marketing, harm perceptions, and e-cigarette of flavour, device type and warning messages on youth use. Am J Health Behav. 2015;39(1):121-131. preferences for electronic nicotine delivery systems: evi- 40. Ambrose BK, Rostron BL, Johnson SE, et al. Perceptions dence from an online discrete choice experiment. Tob of the relative harm of cigarettes and e-cigarettes among Control. 2018;27(e2):e152-e159. US youth. Am J Prev Med. 2014;47(2):S53-S60. 55. Buckell J, Marti J, Sindelar JL. Should flavours be banned 41. Ahijevych K, Garrett BE. The role of menthol in ciga- in cigarettes and e-cigarettes? Evidence on adult smokers rettes as a reinforcer of smoking behavior. Nicotine Tob and recent quitters from a discrete choice experiment. Tob Res. 2010;12(Suppl 2):S110-S116. Control. 2019;28(2):168-175. 42. Carpenter CM, Wayne GF, Connolly GN. Designing 56. Lee H-Y, Lin H-C, Seo D-C, Lohrmann DK. The effect cigarettes for women: new findings from the tobacco in- of e-cigarette warning labels on college students’ percep- dustry documents. Addiction. 2005;100(6):837-851. tion of e-cigarettes and intention to use e-cigarettes. Ad- 43. Lipkus IM, Mays DP, Tercyak K. Characterizing young dict Behav. 2018;76:106-112 adults’ susceptibility to waterpipe tobacco use and their 57. Lempert LK, Glantz SA. Implications of tobacco industry reactions to messages about product harms and addictive- research on packaging colors for designing health warn- ness. Nicotine Tob Res. 2016;19(10):1216-1223. ing labels. Nicotine Tob Res. 2016;18(9):1910-1914. 44. Mays D, Moran MB, Levy DT, Niaura RS. The impact of 58. Lempert LK, Glantz S. Packaging colour research by to- health warning labels for Swedish advertisements on bacco companies: the pack as a product characteristic. young adults’ snus perceptions and behavioral intentions. Tob Control. 2017;26(3):307-315. Nicotine Tob Res. 2015;18(5):1371-1375. 59. Moodie C, Stead M, Bauld L, et al. Plain Tobacco Pack- 45. Mays D, Smith C, Johnson AC, Tercyak KP, Niaura RS. aging: A Systematic Review. London UK: Public Health An experimental study of the effects of electronic ciga- Research Consortium, University of London, Institute of rette warnings on young adult nonsmokers’ perceptions Education, Social Science Research Unit, EPPI-Centre; and behavioral intentions. Tob Induc Dis. 2016;14(1):17. 2012. https://discovery.ucl.ac.uk/id/eprint/1565951/2/ doi:10.1186/s12971-016-0083-x Moodie_et_al._2012._Plain_Tobacco_Packaging._A_ 46. Pearson JL, Richardson A, Feirman SP, et al. American Systematic_Review.pdf. Published 2012. Accessed Febru- Spirit pack descriptors and perceptions of harm: a crowd- ary 16, 2021. sourced comparison of modified packs. Nicotine Tob Res.

Tob Regul Sci.™ 2021;7(2):87-102 DOI: doi.org/10.18001/TRS.7.2.1 101 Pack Modifications Influence Perceptions of Menthol E-cigarettes

Supplemental Table 1 (Table S1) Main Effects of Ever E-cigarette Use and Menthol E-cigarette Use on Perceptions of E-cigarette Product Harm, Addictiveness, and Appeal E-cigarette Use (Ever) Menthol E-cigarette Use (Ever) a Never Use Ever Use Never Use Ever Use M (SD) M (SD) p M (SD) M (SD) p Harm 3.46 (1.06) 2.98 (0.99) .000 3.12 (1.00) 2.93 (0.98) .001 Addictiveness 3.55 (1.13) 3.43 (1.13) .009 3.44 (1.14) 3.43 (1.13) NS Appeal Attention grabbing 4.23 (1.70) 4.36 (1.68) .041 4.25 (1.70) 4.40 (1.67) NS Appealing 4.36 (1.71) 4.54 (1.63) .003 4.53 (1.64) 4.55 (1.62) NS Taste good/appetizing 4.05 (1.82) 4.28 (1.74) .001 4.05 (1.81) 4.25 (1.70) .002 Fun to vape 3.98 (1.78) 4.23 (1.71) .000 4.04 (1.77) 4.30 (1.68) .008 Discourages Useb 3.58 (1.78) 3.16 (1.75) .001 3.06 (1.79) 3.19 (1.74) NS Encourages Use 3.27 (1.75) 3.79 (1.65) .000 3.44 (1.65) 3.91 (1.63) .000 Intention to try 3.00 (1.88) 4.10 (1.77) .000 3.66 (1.86) 4.25 (1.70) .000 Friends would try 3.70 (1.85) 4.23 (1.68) .000 4.07 (1.75) 4.28 (1.65) .024 Note. Items in bod are significant at p < .05. M = mean value; SD = standard deviation. NS = non-significant a Among ever e-cigarette users b Higher scores indicate lower appeal (eg, intentions to use); for all other indices of appeal, higher scores indicate greater appeal

Supplement Table 2 (Table S2) Main Effects of Indices of E-cigarette Package Appeal on Perceptions of Harm and Addictiveness of E-cigarettes Perceived Harm Perceived Addictiveness B (SE) p B (SE) p Appeal Item Attention grabbing -0.05 (.01) .000 -0.03 (.01) .018 Appealing -0.07 (.01) .000 -0.03 (.01) .009 Taste good/appetizing -0.06 (.01) .000 -0.04 (.01) .000 Fun to use/vape -0.07 (.01) .000 0.05 (.01) .000 Discourages Usea 0.05 (.01) .000 0.01 (.01) .394 Encourages Use -0.12 (.01) .000 -0.09 (.01) .000 Intention to try -0.16 (.01) .000 -0.11 (.01) .000 Friends would try -0.11 (.01) .000 -0.07 (.01) .000

Note. B = unstandardized regression beta weight; SE = standard error a Higher scores indicate lower appeal (eg, lower intention/desire to use); for all other indices of appeal, higher scores indicate greater appeal. All models adjusted for ever e-cigarette use.

102