Effects of ‘Ice’ flavoured e-cigarettes with synthetic cooling agent WS-23 or menthol on user-reported appeal and sensory attributes

Abstract

Background:

This clinical experiment tested the effects of exposure to e-cigarettes with WS-23 or menthol cooling additives on user appeal and sensory attributes, and, secondarily, whether WS-23 effects generalized across base characterizing flavor, nicotine concentration, nicotine/tobacco product use status, and mint/menthol nicotine/tobacco flavor preference.

Methods:

In this within-participant double-blind experiment, adult tobacco/nicotine users administered standardized puffs of 18 different e-cigarette solutions in randomized sequences using a pod-style device. Each of three base characterizing flavor solutions (“bold tobacco”, “mango,” “wintergreen”) in both 2% and 4% concentrations of nicotine benzoate salt were manipulated by adding either: (i) menthol (0.5%), (ii) WS-23 (0.75%), or (ii) no cooling agent. After each administration, participants rated 3 appeal and 5 sensory attributes (0–100 scales).

Results:

Participants (n=84; M[SD]=38.6[13.6] years-old) were either exclusive e-cigarette (25.0%), cigarette (36.9%), or dual (38.1%) users. WS-23 vs. no coolant products produced higher liking, willingness to use again, smoothness, and coolness and lower disliking, bitterness, and harshness ratings (|B|_difference range: 4.8 to 20.1; ps<.005). Menthol (vs. no coolant) increased willingness to use again and reduced harshness and coolness (ps<.05). Flavors with WS-23 (vs. menthol) were rated as smoother, cooler, and less harsh (ps<.05). Coolant effects did not differ by base flavor, nicotine concentration, tobacco use, or preferred flavor status.

Conclusions:

Adding synthetic coolant WS-23 to e-cigarettes appears to make the vaping user experience more appealing, regardless of characterizing base flavor. Regulatory agencies should be aware that the manufacturing process of adding synthetic coolants may increase the attractiveness of various e-cigarette products.

Introduction

“Ice” flavored electronic cigarettes (e-cigarettes) are products with flavor variants that are composed of a base characterizing flavor that is enhanced with a cooling agent additive.[1] Ice-flavored e-cigarettes are increasingly marketed internationally, including in products that are widely sold by major e-cigarette manufacturers (e.g., Puff Bar), using names that connote both characterizing and cooling features (e.g., “blueberry ice,” “frosted mint,” “tobacco chill”).[1–4] In the USA, total unit sales of e-cigarettes marketed as “ice” rose from 20,300 in January 2017 to 2,700,000 in November 2021, occupying 12.1% of the market share of all flavored e-cigarettes in late 2021.[5] In recent surveys of adolescent and young adult e-cigarette users, ice and fruit combinations were the top or second-most commonly used flavors.[6, 7] Identifying why ice flavored e-cigarettes may be appealing is important to inform regulation of this flavor category.

Whether the appeal of ice flavored e-cigarettes is driven by the addition of cooling agents is unclear. This distinction is important because the European Tobacco Products Directive Article 7.6 prescribes a ban on additives that enhance a product’s addictiveness and attractiveness or facilitates inhalation and uptake.[8] On social media platforms, e-cigarette users anecdotally report that ice-flavored e-cigarettes are appealing because they possess cooling properties and supplement the base flavor.[1, 9] However, there has not been a controlled double-blind study that examines if coolant additives used in ice-flavored e-cigarettes makes the user experience of vaping more attractive. Such evidence is needed to inform whether regulatory policies limiting the inclusion of cooling agent additives in e-cigarettes would be impactful. [3]

Ice flavored e-cigarettes can include menthol, non-menthol synthetic coolant additives, or a combination of both menthol and synthetic coolant additives.[10] A common synthetic coolant detected in widely-sold ice-flavored e-cigarette brands, such as Puff Bar, was developed by the Wilkinson Sword (WS) company—a compound called WS-23 (2-isopropyl-N,2,3-trimethylbutyramide).[3, 11] [10] These substances are commonly used for topical use (e.g., shaving cream) or ingestion (e.g., chewing gum).[1, 12] Unlike menthol, which can produce a strong minty aroma that can be unappealing for some individuals, WS-23 and its synthetic coolant derivatives (WS-3, WS-5) were designed to be odorless but also produce cooling effects that are similar to menthol.[12] Further, potential regulatory implications of restricting menthol as a chemical additive or characterizing flavor could be circumvented by replacing menthol with a non-menthol synthetic cooling agent by manufacturers of tobacco products, as these non-menthol additives provide a cooling sensation but no characterizing flavor.[1, 2] Together, this could suggest that ice-flavored e-cigarettes made with WS-23 may be more appealing than menthol-based ice-flavored e-cigarettes and serve as a potential menthol replacement in other tobacco, as observed recently in non-menthol cigarettes marketed in California,[13] and e-cigarette products.

In addition to investigating the overall effect of cooling agent additives in flavored e-cigarettes, it is important to determine the generalizability of such effects across other factors. First, it is critical to determine whether coolant additives improve e-cigarette appeal for all types of base characterizing flavors, including tobacco-ice variants which use tobacco as the base flavor. Such evidence can inform whether flavored e-cigarette sales ban policies exempting only tobacco-flavored products should be amended to expressly prohibit all cooling flavors, including tobacco-ice. Second, it is useful to understand whether the appeal-enhancing effects of coolants generalize across different nicotine concentrations. Standard menthol-only flavored e-cigarettes suppress nicotine’s harsh and bitter properties,[14–16] which may make nicotine-rich aerosol easier to inhale and more appealing. Hence, appeal enhancement by menthol or WS-23 additives in ice-flavored e-cigarettes might be amplified in products with higher nicotine concentrations. Finally, it is important to identify whether the appeal-altering effects of coolant additives in ice-flavored e-cigarettes generalizes across different nicotine/tobacco product user groups. Vaping may adversely impact non-smokers but may also reduce harm for smokers who adopt e-cigarettes to quit smoking, switch to vaping, and avoid dual use of both e-cigarettes and cigarettes.[17] Determining whether the appeal of ice-flavored e-cigarettes differ between exclusive vapers, exclusive smokers, and dual-users can inform the total public health impact of regulating this flavor category, including for distinct subpopulations with divergent vaping-associated health implications.

In this double-blind, within-subject randomized clinical experiment in adult nicotine/tobacco users, we tested the effects of exposure to flavored e-cigarettes with WS-23 or menthol (vs. no) coolant additives and a direct comparison of WS-23 vs. Menthol on user-reported appeal and sensory attributes. Secondarily, we tested whether cooling agent effects generalized across: i) fruit, mint, and tobacco base characterizing flavors; ii) low (2%) and high (4%) nicotine concentrations, and (iii) individuals who exclusively vape e-cigarettes, smoke combustible cigarettes, or use both e-cigarettes and smoke cigarettes.

Methods

Participants

USA residents were recruited in 2022 via nationwide website and social media advertisements, announcing research on perceptions of different e-cigarette types. Participation was completed remotely via video conferencing. Inclusion criteria required current e-cigarette use (≥3 days a week for ≥2 months) and/or cigarette smoking (≥4 cigarettes/day for ≥2 years, ≥100 cigarettes lifetime) similar to prior research.[18, 19] Exclusion criteria were planning to quit using nicotine or tobacco products, currently or planning to become pregnant/breastfeeding, current daily use of tobacco other than combustible cigarettes or e-cigarettes, hospitalization for COVID-19, and/or lack of access to a wireless-enabled digital device with videoconferencing or private space for remote study visits. Participants provided written and verbal informed consent during the video conference. The University of Southern California Institutional Review Board approved the study.

Design and materials

This study used a within-subject 3×3×2 factorial design with a primary experimental factor—Coolant Additive (WS-23, menthol, no coolant) and two secondary factors—Base Characterizing Flavor (Tobacco, Fruit, Mint) and Nicotine Concentration (2%, 4%). This was achieved by administering each participant a randomized ordering of 18 different custom-manufactured e-cigarette solutions containing ingredients representative of publicly-marketed products[3] (Molecule Labs; Concord, CA). Each of three base characterizing flavor solutions (“bold tobacco”, “mango,” “wintergreen”) in both 2% and 4% concentrations of nicotine benzoate salt (1:1 benzoic acid to nicotine molar ratios) were manipulated by adding either: (i) menthol (0.5%), (ii) WS-23 (0.75%), or (ii) no cooling agent. Nicotine concentrations were selected based on previous work examining the appeal of e-cigarette products, with a range of 2% to 4%, given the number of trials participants were administered in the current study.[16, 18–20] The Roswell Park Comprehensive Cancer Center Nicotine and Tobacco Product Assessment Core conducted independent constituent tests of the solutions to determine concentrations of menthol, WS-23 concentration, nicotine concentration, density, propylene glycol/vegetable glycerin (PG/VG) vehicle, benzoic acid, and pH (see constituent testing methods in supplemental materials). Solutions were administered by an Avatar Go (114 mm height; 19 mm width; 10.5 mm thickness; 3. 7V DC lithium polymer battery input; 350 mAh battery) pod-style device with refillable cartridge inserts as in prior work.[19] Participants used the Avatar Go device at the highest setting level (Red; 14 watt).

Procedure

We used a remote data collection protocol for product appeal testing, which was successfully used previously.[19] After preliminary telephone screens, participants completed an initial orientation remote videoconference visit involving eligibility confirmation, informed consent, and study overview. Those eligible were scheduled for a subsequent experimental remote study visit for which they were instructed to abstain from using nicotine/tobacco for 2 hours prior to visit, as in prior work.[18, 19] The two-hour abstinence instruction prevents participants from entering the e-cigarette administration procedure with high baseline nicotine levels, which reduces risk of unpleasant nicotine saturation effects during the procedure, but is too brief to elicit severe nicotine withdrawal symptoms. Prior to the experimental visit, participants were shipped study materials, including urine specimen collection kits for pregnancy and nicotine exposure test kits that assessed presence of nicotine, cotinine, and other nicotine metabolites (NicCheck; Mossman Associates, Milford, MA, USA), e-cigarette device, and 18 pre-filled cartridges. To maintain experimental blind, staff who did not interact with participants prepared the cartridges with coded labels #1 – #18, reflecting the participants’ personalized randomly-assigned sequence for administering the 18 solutions.

After urine collection, conducting nicotine exposure tests, and confirmation of negative pregnancy test at the experimental visit’s outset, staff guided participants through the product appeal and sensory rating protocol. During each administration, participants viewed a tutorial video with instructions directing them through the controlled guided puffing procedure, which involved a 2-puff cycle (10-second preparation, 4-second inhalation, 1-second hold, and 2-second exhale interval) for each product immediately followed by appeal and sensory attribute ratings, per previous research.[19] This standardization helps equate the puff “dose” of each study condition. Consistent with previous work,[19] participants were monitored with video conferencing software throughout the entire study to ensure compliance to administration procedures. After ratings, participants were instructed to drink water to prevent sensory effects from carrying over to the next product administration. The procedure was separated into trial blocks spaced at least 10 minutes apart, during which participants completed sociodemographic and nicotine use history questionnaires. For each trial block, the appeal/sensory items had the same number of questions. Each of the questionnaires that occurred in between the trial blocks had a similar number of questionnaire items. If a participant completed all non-appeal/sensory questionnaire items prior the end of the last trial block, they were allowed to use the restroom, refill their water cup, or read a book/magazine. All participants were asked to stay on camera with video/audio active. All administration were conducted in one virtual session which lasted 4 hours and participants were compensated $190.

Measures

Appeal and Sensory Attributes.[19]

After each two-puff trial, participants rated appeal and sensory attribute outcomes on visual analog scales (range, 0–100) using items previously demonstrated to be sensitive to effects of e-cigarette flavor and menthol variation.[18, 20] Appeal measures were: (1) Liking (“How much did you like it [the e-cigarette]?”); (2) Disliking (“How much did you dislike it?”); (3) Willingness to use again (Would you use it again?”). Sensory attributes were: (1) Sweetness (“How sweet was it?)”; (2) Smoothness (“How smooth was it?”); (3) Bitterness (“How bitter was it?”); (4) Harshness (“How harsh was it?”); and (5) Coolness (“How cool or cooling was it?”). Rating anchors were not at all and extremely for each measure, except use again (not at all and definitely). Each rating was analyzed individually.

Current Nicotine/Tobacco Use Status and Pre-existing Menthol/Mint Flavor Preference.

Participants completed sociodemographic and nicotine/tobacco product use history questions described in Table 1. Responses to combustible cigarette and e-cigarette use classified current e-cigarette use vs. non-use (≥3 days a week for ≥2 months) and cigarette smoking vs. non-smoking (≥4 cigarettes/day for ≥2 years, ≥100 cigarettes lifetime). These were re-classified to create three mutually exclusive groups: (a) exclusive current e-cigarette use (current vaping with no past-30-day smoking), (b) exclusive current smoking (current smoking without past-30-day vaping), (c) dual use (current vaping and current smoking). Pre-existing menthol/mint flavor preference was defined as usual use of menthol cigarettes among cigarette smokers and most often use of menthol or mint solutions among e-cigarette users.

Table 1.

Participant Characteristics (N = 84)

Variables	N (%) or M (SD)
Sociodemographics
Age, M (SD)	38.6 (13.6)
Gender identity, N (%)
Cisgender Man	38 (45.2)
Cisgender Woman	41 (48.8)
Another gendera	5 (6.0)
Sexual identity, N (%)
Straight/heterosexual	64 (76.2)
Another sexual identityb	20 (23.8)
Race/ethnicity, N (%)
Non-Hispanic White	47 (56.0)
Non-Hispanic Black	11 (13.1)
Another race/ethnicityc	26 (31.0)
Educational attainment, N (%)
High school diploma/GED or less	13 (15.5)
Some college completed or currently enrolled	30 (35.7)
College degree or higher	41 (48.8)
Employment status, N (%)
Full-time	26 (31.0)
Part-time	24 (28.6)
Unemployed	12 (14.3)
Another employment statusd	22 (26.2)
Current tobacco use status e
Dual use of cigarettes and e-cigarettes, N (%)	32 (38.1)
Exclusive cigarette smoker, N (%)	31 (36.9)
Exclusive e-cigarette user, N (%)	21 (25.0)
Combustible cigarette use characteristics f
No. of days smoked in past 30 days, M (SD)	25.1 (9.0)
No. of cigarettes smoked per day, M (SD)	11.1 (7.8)
Usually smokes menthol cigarettes, N (%)	27 (42.2)
E-cigarette use characteristics g
No. of days vaped in past 30 days, M (SD)	24.4 (8.5)
No. of times vaped per day, M (SD)	13.6 (7.0)
No. of puffs per vaping episode, M (SD)	4.5 (3.9)
Nicotine concentration typically used in e-cigarette
1–5 mg/ml	9 (17.0)
6–10 mg/ml	5 (9.4)
11–25 mg/ml	3 (5.7)
26–50 mg/ml	11 (20.8)
≥51 mg/ml or ≥5%	20 (37.8)
Do not know
E-cigarette flavor used most often, N (%)	5 (9.4)
Tobacco	7 (13.2)
Menthol/mint	5 (9.4)
Fruit	25 (47.2)
Iceh	6 (11.3)
Another flavori	10 (18.9)
E-cigarette device type usedj, N (%)
Refillable, rechargeable devicek	21 (25.0)
Disposable device	30 (35.7)
Juul	11 (13.1)
Other pod device	13 (15.5)
Another device	1 (1.2)

Note. Frequencies may not sum to the total due to different patterns of missing data across variables.

^aTransgender man/woman, non-binary/genderqueer, or something else.

^bAsexual, bisexual, gay, lesbian, pansexual, queer, questioning/unsure, or something else.

^cAlaska Native, American Indian, Asian or Pacific Islander, Hispanic, Latino, or Spanish origin, or multi-racial.

^dRetired/disability, summer only, student only, or student with full or part-time job.

^eExclusive e-cigarette users = vape ≥ 3 days a week for ≥ 2 months; exclusive cigarette smokers = ≥ 4 cigarettes/day for ≥2 years; dual users = dual users of e-cigarettes and cigarettes.

^fSubsample of current exclusive smokers (n=31) and dual users (n=32).

^gSubsample of current exclusive e-cigarette users (n=21) and dual users (n=32).

^hCombination of characterizing flavor with cooling properties (e.g., blueberry ice).

ⁱUnfavored, alcohol, clove, spice, dessert, or other sweets.

^jEach device type was measured as binary variable (yes/no). Multiple response was allowed.

^kTank system, canister system, vape pen or pen-like rechargeable device, mod or mech-mod rechargeable device, box mod.

Data analysis

After examining descriptive statistics of the sample and e-cigarette solutions, the primary analysis used multilevel linear models (MLMs) that tested the fixed effects of synthetic cooling agent condition on appeal and sensory ratings. Separate models were tested for each rating outcome construct, which accounted for nesting of data within participants (18 data points per participant: 3 cooling agent conditions × 3 base flavors × 2 nicotine concentrations). The study’s primary aim was indicated by the main effect of WS-23 and menthol (vs. no coolant [reference category]) collapsed across the 2 nicotine concentrations and 3 base flavor conditions. These were followed by pairwise comparisons of WS-23 and menthol. The secondary aim examined effect generalizability and was tested in subsequent MLMs modeling cooling agent × base flavor, cooling agent × nicotine concentration, cooling agent × nicotine/tobacco product use, and cooling agent × pre-existing menthol/mint flavor preference omnibus interaction tests. Because of the regulatory importance of determining effect generalizability to each base flavor, we examined cooling agent effects in supplementary MLMs stratified by tobacco, fruit, and mint base flavors. To determine if the primary results were affected by sensory habituation or sensitization caused by repeated exposures, sensitivity analyses tested if cooling agent effects generalized across the testing protocol via cooling agent × trial number interactions.

MLMs produced unstandardized effect estimates (B) with 95% Confidence Intervals (CIs), reflecting difference in mean ratings between experimental conditions. MLMs use all available data; 6 individuals had trial-level missing data (range: 5–17 trials). Statistical significance was 0.05 (two tailed) with Benjamini-Hochberg corrections for multiple testing to control false-discovery rate at 0.05.[21] Analyses were conducted in R version 4.2.0.[22]

RESULTS

Participant characteristics

Of 529 individuals completing phone eligibility screens, 122 attended an orientation visit, of whom 84 enrolled and participated in the product appeal testing procedure, constituting the analytic sample (participant accrual detailed in Figure S1 in the supplement). Depicted in Table 1, the sample (age, M(SD)=38.6(13.6) years) was demographically diverse and included 38.1% dual use of cigarettes and e-cigarettes, 36.9% exclusively used cigarettes, and 25.0% exclusively used e-cigarettes. Urine nicotine exposure tests produced positive results (i.e., equivalent to ≥200 ng/ml cotinine) for 83 (98.8%) participants. Among current smokers (exclusive or dual use), past-30-day number of days smoked and cigarettes smoked per smoking day, on average, were 25.1 (SD=9.0) and 11.1 (SD=7.8), respectively, with 42% usually smoking menthol-flavored cigarettes. Among current e-cigarette users (exclusive or dual use), there was an average of 24.4 (SD=8.5) days vaped in the past 30 days, 13.6 (SD=7.0) times per vaping day, and 4.5 puffs (SD=3.8) per vaping episode. Fruit (47.2%) and disposables (35.7%) were the most used e-cigarette flavor and device, respectively, among e-cigarette users; six participants (11.3%) used ice-flavored e-cigarettes.

Constituent analyses

Results verified that study solutions aligned with specifications as detailed in supplemental Table S1. Across all 18 solutions, benzoic acid was detected and the mean(SD) pH was 5.8(0.6), density was 1.1(0.0), and PG/VG was 60.6/39.4(3.4). Mean (SD) nicotine concentration was 23.9(0.8) mg/ml in the 2% and 44.5(2.5) mg/ml in 4% solutions, respectively. The mean(SD) menthol concentration was 5.3 (0.4) mg/ml in the 6 menthol-ice flavor solutions; menthol was not detected in the other solutions. The mean(SD) WS-23 concentration was 7.3 (0.3) mg/ml in the 6 WS-ice flavor solutions. For the wintergreen flavor conditions that were not ice or menthol variants, levels of menthol and WS-23 were below limit of quantitation.

Primary results: Cooling agent main effects on appeal and sensory experience

Depicted as overall main effects in Table 2, MLMs found that products with menthol (vs. no coolant) elicited higher willingness to use again, lower harshness, and higher coolness ratings, but did not differ in other outcomes. Products with synthetic coolant WS-23 (vs. no coolant) elicited significantly favorable appeal ratings, yielding 5.8(95% CI=2.7, 8.9) points higher liking, lower disliking (B= −6.8[95% CI=−10.2, −3.4]), and higher willingness to use again: B=6.5[95% CI=2.9, 10.1]). WS-23 vs. no coolant also improved sensory attributes of smoothness (B=6.1[95% CI=2.9, 9.3), bitterness (B=−4.8[95% CI=−7.9, −1.8]), harshness (B=−7.6 [95% CI=−10.9, −4.3]), and coolness (b=20.1[95% CI=17.1, 23.1]). Omnibus F-tests also revealed the significant effects of cooling agent condition on appeal and sensory outcomes (ps<.01) except sweetness. Pairwise comparisons displayed in Figure 1 show that ice-hybrid flavors produced by WS-23 (vs. menthol) additives elicited significantly higher smoothness and coolness and lower harshness but did not differ for other outcomes. Neither cooling agents affected sweetness. Means (SD) appeal and sensory attribute ratings are reported in Supplemental Table S2.

Table 2.

Estimates (95% CIs) of Main Effects of Cooling Agent Condition on Appeal and Sensory Attribute Outcomes

	Appeal			Sensory Attributes
	Liking	Disliking	Use Again	Sweetness	Smoothness	Bitterness	Harshness	Coolness
Synthetic coolant WS-23 vs. no coolant
Overall	5.8 (2.7, 8.9)	−6.8 (−10.2, −3.4)	6.5 (2.9, 10.1)	2.3 (−0.6, 5.2)	6.1 (2.9, 9.3)	−4.8 (−7.9, −1.8)	−7.6 (−10.9, −4.3)	20.1 (17.1, 23.1)
By base characterizing flavor
Tobacco	5.8 (1.0, 10.6)	−6.7 (−12.1, −1.3)	6.1 (0.3, 11.9)	2.8 (−1.5, 7.1)	6.9 (1.7, 12.1)	−7.3 (−12.3, −2.4)	−12.5 (−18.1, −6.9)	19.4 (14.8, 24.1)
Mint	6.2 (1.3, 11.2)	−7.2 (−12.9, −1.6)	5.1 (−0.7, 10.8)	2.3 (−2.1,6.8)	4.2 (−1.4, 9.7)	−1.3 (−6.6, 4.1)	−1.3 (−7.2, 4.5)	19.6 (14.5, 24.7)
Fruit	5.4 (0.4, 10.5)	−6.5 (−12.1, −1.0)	8.3 (2.4, 14.2)	2.2 (−2.4, 6.8)	7.2 (1.6, 12.7)	−6.0 (−10.3, −1.7)	−8.8 (−14.3, −3.4)	21.3 (16.8, 25.8)
Menthol vs. no coolant ^
Overall	3.0 (−0.1,6.1)	−3.1 (−6.5, 0.3)	4.0 (0.4, 7.6)	1.2 (−1.7, 4.1)	0.6 (−2.6, 3.8)	−1.6 (−4.5, 1.5)	−4.3 (−7.6, −1.0)	12.7 (9.7, 15.7)
By base characterizing flavor
Tobacco	3.8 (−1.0, 8.6)	−3.7 (−9.0, 1.7)	4.6 (−1.2, 10.4)	2.4 (−2.0, 6.7)	3.4 (−1.8, 8.6)	−2.8 (−7.8, 2.2)	−9.2 (−14.8, −3.5)	11.1 (6.4, 15.7)
Mint	2.3 (−2.7, 7.2)	−2.4 (−8.1,3.2)	1.3 (−4.5, 7.1)	0.6 (−3.9, 5.1)	−2.0 (−7.5, 3.6)	0.5 (−4.9, 5.8)	−0.0 (−5.9, 5.8)	16.3 (11.2, 21.4)
Fruit	3.2 (−1.9, 8.2)	−3.1 (−8.7, 2.4)	5.9 (0.0, 11.8)	1.2 (−3.4, 5.8)	0.4 (−5.2, 5.9)	−2.7 (−7.0, 1.7)	−3.4 (−8.9, 2.1)	10.9 (6.4, 15.5)

Note. Estimate = difference in means by cooling agent condition, adjusted by trial sequence (range: 1–18). CI = Confidence Interval.

Bold values are statistically significant after Benjamini-Hochberg corrections for multiple testing to control false-discovery rate at 0.05.

Results of omnibus F tests of non-equivalence of study outcomes across the three cooling agent conditions were as follows: liking (p=.002), disliking (p=.001), use again (p=.002), sweetness (p=.306), smoothness (p=.001), bitterness (p=.007), harshness (p<.001), and coolness (p<.001).

Figure 1.

Study conditions not sharing letters are significantly different in pairwise contrasts after correcting p values for multiple tests to maintain a false discovery rate <0.05. Benjamini-Hochberg corrections were used for all pairwise comparisons.

Secondary results: Cooling agent effect generalizability

Presented in Table 3, separate two-way interaction effects of cooling agent with base flavor, nicotine concentration, nicotine/tobacco use status, and pre-existing menthol/mint flavor preference on study outcomes were non-significant (ps≥.062), indicating that the magnitude of coolant effects did not differ across these factors. Effects stratified by base flavor largely show the same pattern of effects for tobacco, mint, and fruit ice variants (Table 2). Menthol (vs. no coolant) increased coolness for all base flavors and reduced harshness for tobacco base flavored products but did not affect other outcomes. WS-23 (vs. no coolant) improved nearly every appeal and sensory attribute except sweetness for tobacco and fruit base flavors. WS-23 (vs. no coolant) improved only liking, disliking, and coolness for mint base flavor.

Table 3.

Interaction Effects Cooling Agent with Base Flavor, Nicotine Concentration, and Nicotine/Tobacco Use on Appeal and Sensory Attributes

	Appeal						Sensory attributes
	Liking		Disliking		Use again		Sweetness		Smoothness		Bitterness		Harshness		Coolness
	F	P	F	P	F	P	F	P	F	P	F	P	F	P	F	P
Cooling agenta × Flavorb	0.1	0.993	0.1	0.993	0.3	0.976	0.1	0.993	0.5	0.771	0.9	0.688	2.5	0.067	1.1	0.470
Cooling agenta × Nicotine Concentrationc	0.1	0.996	0.0	0.996	0.0	0.996	0.5	0.791	0.4	0.905	0.0	0.996	0.0	0.996	1.7	0.273
Cooling agenta × Nicotine/tobacco used	0.2	0.968	0.9	0.795	1.0	0.760	0.2	0.968	0.1	0.968	0.3	0.968	0.5	0.968	1.7	0.383
Cooling agenta × Pre-existing menthol/mint flavor preferencee	1.2	0.484	0.5	0.778	1.3	0.484	0.3	0.854	1.6	0.484	0.3	0.854	1.4	0.484	4.0	0.062

Note. p = p-value corrected for multiple testing to control false-discovery rate <.05 from omnibus F tests.

^aNo cooling agent vs. menthol vs. synthetic cooling agent WS-23.

^bTobacco vs. mint vs. fruit.

^c2% vs. 4%.

^dCurrent exclusive combustible cigarette use vs. exclusive e-cigarette use vs. dual use of cigarettes and e-cigarettes.

^eMenthol as usually used flavor among cigarette smokers and menthol or mint as most often used flavors among e-cigarette users.

Sensitivity analysis

Interactions between cooling agent and trial number were non-significant (ps≥.88), providing no evidence of habituation or sensitization to cooling agent effects during the procedure (Supplement Table S3).

Discussion

This double-blind within-subject randomized experiment found that enhancing e-cigarettes by adding cooling agents improved the user experience of vaping, regardless of the product’s base characterizing flavor and nicotine concentration or the user’s nicotine/tobacco use status. Adding coolants—the manufacturing practice for creating ice-hybrid flavors—was found to have statistically significant (albeit subtle) effects on appeal measures, such as whether participants would use the product again. Coolant additives also improved various sensory attributes, including countering aerosol harshness, which is a property that could make ice flavored e-cigarettes easier to inhale than standard flavors.

The experimental control of this study’s paradigm isolates the effects of the cooling constituent, per se, from other sources of inter-product variation in appeal (e.g., packaging, advertising, cultural trends). Hence, it is reasonable to infer that ice-flavored e-cigarettes’ increasing sales[5] and use prevalence[6] cannot be solely attributed to their marketing. Rather, this experiment provides the first evidence directly implicating coolant additives, particularly WS-23, in the attractiveness of ice-flavored products. This result could be applied in regulatory policies, such as the European Tobacco Products Directive, which prescribes restrictions that prohibit manufacturers from including additives that increase a product’s attractiveness or facilitate inhalation.[8] Canadian regulatory policies, which prohibit most flavoring additives,[23] could also apply the current findings.

In this study, the transformation of standard flavors into ice flavors by adding synthetic coolant WS-23 more consistently improved product appeal and sensory attributes than ice flavors created with menthol. The current findings are consistent with anecdotal reports from users who mix their own ice flavors on social media describing WS-23 as having strong cooling effects without altering the taste of the base flavor and being preferable to menthol.[1] WS-23 has similar activity on TRPM8 receptors that mediate cooling, analgesic, and anti-irritant effects, but less activity on the sensory irritant receptor, TRPA1 (transient receptor potential cation channel subfamily A member 1).[24] While WS-23 and menthol may not have the same cooling power at the same concentration levels, this study provides support that consumers of e-cigarette products may prefer the sensory profile of WS-23 of ~7.3 mg/mL in comparison to products with menthol at a comparatively lower concentration (~5.3 mg/ml). WS-23’s favorable pharmaco-sensory profile may explain why WS-23 and other synthetic coolants are more commonly found in marketed ice-flavored products than menthol.[3]

While this study found that the addition of synthetic coolant WS-23 increased sensory attributes and product appeal, it is not known how these results translate to actual user behavior (e.g., inhale more deeply, use the product more often) or impact the toxicant profile of e-cigarette products with WS-23. Previous cross-sectional research has shown that among e-cigarette users who vape ice flavors report vaping on more days[7] and use their e-cigarette more times throughout the day.[7] Additionally, little is known about the pulmonary toxicity of WS-23. Synthetic cooling agents have been tested for dermal exposure or ingestion.[12] Current published literature and in vitro studies have shown cytotoxicity[10] and ROS generation[25] by lung epithelial cells when exposed to WS-23. Additionally, alterations in airway physiology and respiratory disease susceptibility have been observed in human airway epithelial cells exposed to WS-23 aerosols.[26] If e-cigarette manufacturers are adding compounds that may enhance the toxicant profile or the pulmonary toxicity of e-cigarette aerosol, regulatory agencies should be aware and potential restrictions on the constituents included in e-cigarettes may warrant consideration.

The secondary aim of this study examined if sensory improvement by synthetic coolants is amplified at higher nicotine concentrations. Yet, there was no evidence of differences in effects of WS-23 and menthol across 2% and 4% nicotine concentrations in this study. Previous pre-clinical research examining the sensory evaluation of WS-23 found that WS-23 was a less effective oral cooling agent than menthol,[24], which may explain why no differences were observed at higher nicotine concentrations. Whether coolant effect amplification is observed at higher nicotine concentrations (≥5%) or free-base formulations merits further study to inform whether regulating coolant additives would impact appeal of an array of e-cigarette products, regardless of nicotine level or formulation.

The secondary aim found that the sensory effects of menthol and WS-23 generalized across exclusive e-cigarette, cigarette, or individuals who dual use both products, even when examining user preference for and use of menthol cigarettes and/or menthol/mint e-cigarette flavors. This could be considered in the context of regulatory policies, such as the US FDA,[27] which consider health impacts across different subpopulations. For example, permitting marketing of e-cigarettes that appeal to those who vape to quit smoking and avoid dual use patterns could be beneficial, especially if those products do not also appeal to non-smokers and young people. This study did not find evidence of any such association as illustrated by null cooling agent × nicotine/tobacco product use interactions. Coolant-induced appeal and sensory enhancement from menthol or WS-23 was not limited to individuals who already vape, including those with pre-existing preferences for ice or other cooling flavors. This finding is particularly relevant as many regulatory agencies consider a potential menthol restriction in tobacco and e-cigarette products, as WS-23 may serve as an effective replacement for menthol in tobacco and e-cigarette products. Recent research has highlighted the use of another synthetic cooling chemical WS-3 in California cigarettes marketed as non-menthol following regulation restricting the retail sale of flavored tobacco products which included menthol.[13]

The appeal- and sensory-enhancing effects of including coolants, especially WS-23, largely generalized across fruit, mint, and tobacco base characterizing flavors. In a prior study of young adults, ice-fruit flavor combinations were the most prevalent flavor used and their use was associated with more frequent vaping, more vaping episodes per vaping day, and greater nicotine dependence in comparison to use of some non-hybrid flavors.[7, 28] The higher quantity and frequency of e-cigarette use observed in use of ice-fruit flavors in that study could be attributable to their appealing user experience. Any heightened risk of dependence associated with ice-fruit flavors could be explained by WS-23’s improvement of airway sensations (increased smoothness and reduced harshness) that make ice-flavored aerosol easier to inhale, which could amplify nicotine absorption. The current finding that adding WS-23 to tobacco base flavors enhanced appeal is important for the interpretation and implementation of flavored e-cigarette sales bans that prohibit any characterizing flavor besides tobacco. Some of these rules could be ambiguous as to whether tobacco-ice flavored e-cigarettes that contain WS-23 (or other odorless, tasteless synthetic coolants) are considered within the family of standard tobacco flavors and therefore exempted from flavor restriction regulations. Given the enhanced appeal of tobacco-ice vs. standard tobacco flavors demonstrated in this study, policy makers should consider whether such regulations need to be amended or clarified to address tobacco-ice flavored products.

This study paradigm assessed immediate sensory reactions to aerosol of various study products under controlled settings in a single visit, which does not resemble certain real-world conditions and has other limitations. First, some vape shops may offer customers samples of various products during a visit,[28] but many users self-purchase a single product, which differs from the current procedure. Second, this study was conducted using a remote, virtual laboratory paradigm and did not utilize an independent verification of adherence to the puffing procedure. While another study utilized a similar methodology,[19] future studies incorporating validity checks or remote user topography are warranted. Second, repeated exposure to coolants and other constituents in this study could cause sensitization or habituation effects that modify users’ responses later into the procedure. However, sensitivity analyses found no evidence of such effects. Third, participants were not exposed to product marketing, which does not resemble real world use. Fourth, the current appeal paradigm cannot address neuropharmacological-mediated effects of coolant additives nor ad-libitum use patterns, which are best examined using addiction liability test paradigms. Fifth, although we used study products with constituents similar to marketed ice-flavored products and the coolant did not differ across base flavors or nicotine concentrations, generalization to the wide spectrum of products on the marketplace is unclear. Additionally, this study did not examine the relative effective of menthol and other cooling agents when delivered in equal concentrations in e-cigarettes. Menthol and WS-23 may not have equal cooling power at the same concentration,[24] and, as such, it is possible that ice-flavored products on the market with WS-23 may have higher comparative coolant concentrations than those with menthol. Sixth, although recent studies of youth and young adults find high prevalence of ice-flavored e-cigarette use,[6, 7] it is unknown if the findings from this adult sample would generalize those under age 21.

In sum, this study of current nicotine/tobacco product users found that “ice” enhancement of tobacco, fruit, and mint flavored e-cigarettes from synthetic coolant WS-23 additives improved product appeal and the sensory experience of vaping. Further, the data indicated that ice-flavored products with WS-23 provided more desirable sensory attributes than those with menthol. These findings motivate further research on cooling agent effects in ice-flavored e-cigarettes to provide a comprehensive picture of their public health impacts. Until then, regulators should be aware that the manufacturing process of adding synthetic coolants may increase the attractiveness of various e-cigarette products.

Key Messages.

What is already known on this topic

E-cigarettes composed of a base characterizing flavor that is enhanced with a synthetic (i.e., WS-23) and/or menthol cooling agent additive are increasingly marketed in names such as “blueberry ice,” “frosted mint,” and “tobacco chill.” It is unclear whether the inclusion of non-menthol additives like WS-23 increase the attractiveness of e-cigarettes, leaving uncertain whether regulating synthetic cooling additives in e-cigarettes would be impactful.

What this study adds

This double-blind within-subject randomized experiment provides the first evidence that enhancing flavored e-cigarettes by adding synthetic cooling agent WS-23 (or menthol to some extent) heightens the perceived appeal and sensory experience of vaping—an effect that did not vary across the product’s base characterizing flavor and nicotine concentration or the users’ nicotine/tobacco use status.

How this study might affect research, practice or policy

As many regulatory agencies consider a menthol restriction in tobacco and e-cigarette products, non-menthol or synthetic cooling agents may become a menthol replacement to enhance consumer appeal.