Novel Nicotine Concentration Labels Improve Adolescents’ and Young Adults’ Understanding of the Nicotine Strength of Electronic Nicotine Delivery System Products

Abstract

Introduction

E-cigarette liquid nicotine concentrations typically are labeled as mg/mL or percent, which poorly convey nicotine strength to users. We evaluated whether four novel nicotine concentration labels better convey information about nicotine strength and addictiveness.

Methods

Adolescents and young adults (N = 826) completed an online survey in 2020. Participants rated nicotine concentrations (3, 6, 18, 30, 40, and 50 mg/mL) from “no nicotine” to “very high nicotine” communicated using current market labels (mg/mL, percent) and four new labels (text-based, caution sign-shaped, horizonal stoplight, vertical thermometer) which used color, symbols, and verbal strength descriptors. Participants reported on perceived addictiveness for all labels viewed and rank-ordered labels on perceived ability to convey information accurately.

Results

Participants ranked the vertical (77%) and horizontal (70%) labels in first or second place and mg/mL (59.1%) and percent (47.2%) in last or second-to-last place. All new labels conveyed nicotine strength more accurately than did market labels (M[SD] correct of 6: percent = 1.50[1.08]; mg/mL = 2.14[1.52]; caution = 5.23[1.37]; vertical thermometer = 5.28[1.51]; text = 5.33[1.36]; horizontal stoplight = 5.47[1.14]), with the horizontal label also outperforming the thermometer and caution labels. Underestimating nicotine strength was uncommon among all new labels (4.7–6.8%). The new labels also were associated with increased perceived addictiveness at higher concentrations (30, 40, and 50 mg/mL), although the thermometer label underperformed the others. When considering perceived nicotine strength, rates of strength underestimates, perceived addictiveness, and rank ordering, the horizontal stoplight label performed best.

Conclusions

Novel labeling could improve understanding of nicotine strength and e-cigarette-related risk.

Implications

Extending prior research showing that adolescents and young adults who use Electronic Nicotine Delivery Systems (ENDS) have difficulty understanding nicotine concentrations labeled using mg/mL and percent nicotine, the current study demonstrates that novel nicotine concentration labels can improve understanding of nicotine strength and influence perceptions of addictiveness among young ENDS users. While four novel labels were tested, each outperforming current market labeling, the novel label that resembles a horizontal stoplight performed best. The study provides proof-of-concept that creating an easy-to-understand nicotine concentration label is possible and that new labeling better conveys information about nicotine strength and addictiveness.

Introduction

Electronic Nicotine Delivery Systems (ENDS), also known as e-cigarettes or “vapes,” continue to be the most prevalent form of nicotine use among youth; as of 2020, 19.6% of high school students reported past-month ENDS use.¹ Unfortunately, ENDS use may be especially harmful for adolescents and young adults (AYAs); among other negative consequences like risk for using combustible tobacco^2–5 and exposure to toxicants,⁶ ENDS use typically exposes AYAs to nicotine,⁷ which has negative effects on the maturing brain and rapidly can lead to dependence.^8,9 Given the negative effects associated with youth nicotine use, conveying information about the nicotine in ENDS products in an easy-to-understand way is important. However, there remains uncertainty among some AYAs about whether ENDS even contain nicotine, and there appears to be widespread misunderstanding about the strength of nicotine in ENDS products among both youth and adults.^10–12

Misunderstanding about nicotine in ENDS products may be due, at least in part, to ineffective nicotine labeling. While effective labeling can inform the public about product contents that may not already be known, well-understood, or readily observable to consumers, ineffective labeling can result in confusion and/or disregard.^13,14 Currently, ENDS liquids typically carry two nicotine-related labels: the FDA nicotine warning (“This product contains nicotine. Nicotine is an addictive chemical.” ¹⁵) and nicotine concentration (typically presented in mg/mL or percent). Both pieces of information must be understandable to inform the public effectively about nicotine in ENDS. The FDA warning may increase awareness that the product contains nicotine,^16,17 but it does not address the wide range of commercially available nicotine concentrations in ENDS liquids.¹⁸ Thus, ENDS users must rely on the separately labeled nicotine concentration to obtain information about nicotine strength.

Typically, nicotine concentration labels on ENDS liquids contain only text and are plain and small—features of ineffective labels.¹⁹ Importantly, prior work has shown that both older adults and AYAs have difficulty understanding nicotine concentrations presented using the two most common labeling metrics (i.e., mg/mL and percent nicotine), translating them into a clear conceptualization of nicotine strength, and converting between the two metrics.^10,20 In a prior study,¹⁰ AYAs rated commercially available nicotine concentrations presented as mg/mL and percent (equivalent to 0to 60 mg/mL) from “no nicotine” to “very high nicotine” and completed head-to-head comparisons of concentrations presented using these two market labels (eg, 18 mg/mL nicotine is stronger than, weaker than, or equivalent to 5% nicotine).¹⁰ Note that to covert percent nicotine to mg/mL, one must multiple by 10 (eg, 3% nicotine = 30 mg/mL). AYAs incorrectly identified nicotine strength more than 60% of the time when concentrations were presented as mg/mL and more than 70% of the time when presented as percent. Participants also struggled when making direct comparisons, answering incorrectly more than 60% of the time. These findings suggest current market labels are not sufficiently informative and cannot be readily compared to one another. These misunderstandings may contribute to inadvertent exposure to high nicotine levels and increased risk for negative consequences like dependence.

In the current study, we expanded upon prior research^10,20 by developing and testing four novel nicotine concentration labels to determine if they were superior to current market labels in terms of conveying information about nicotine concentration strength. As a secondary aim, we also examined the impact of label type on perceived addictive potential.

Methods

Label Stimuli Development

Given that no evidence-based labels for nicotine concentration existed, the study team sought guidance on label development from the broader labeling literature (eg, for nutrition, eco-friendliness, fuel economy, tobacco products, hazardous chemicals). We designed the labels to be informational at their core with features to better convey nicotine strength/addictiveness (eg, color coding and text descriptors).²¹ While the informational nicotine concentration labels are not warning labels, we used recommendations from the literature on tobacco product warning labels as well as best practices for general product labeling to guide basic labeling features. Specifically, we designed the informational labels to be direct, prominent, feature color, utilize graphics/symbols, and take up considerable space on the front of product packaging.^21–31

Initially, we developed 6 label variants, and, during an earlier phase of the project (results not presented), we sought feedback on them via a survey of 40 subject matter experts (ie, research experts on ENDS who had published at least 3 peer-reviewed articles in the past 5 years). Expert feedback led to the retention of four label formats for testing in the current study (see Figure 1, “Example of New Labels”): (1) a text-based label which paired numeric concentration levels with descriptors such as “high nicotine,” (2) a safety sign label based on research demonstrating the triangle symbol’s ability to convey risk information,²¹ including for ENDS,¹⁶ (3) a vertical thermometer label based on its ability to display the full range of nicotine concentrations in one image,³² and (4) a horizontal stoplight label based on its ability to convey information about nutrition,^33,34 health outcomes (eg, blood pressure³⁵), and the strength of THC/CBD products to cannabis users.^36,37 We used mg/mL for displaying the numerical value on all labels based on our work showing that mg/mL is better understood than is percent among youth and adults.^10,20 We also included color-coding on all labels consistent with research indicating that red implies danger (high nicotine), orange implies warning (medium nicotine), and yellow implies caution (low nicotine).²¹ Color coding also may improve label interpretability for individuals with limited reading ability and/or health literacy.^35,38 Descriptive text (eg, low, medium, high) also was included on all labels to help convey information about strength because it is easy to understand, adds context to the number denoting nicotine concentration, and is consistent with effective labeling of other products (eg, THC in cannabis³⁹). Finally, we considered that the new labels could be used as supplements to the approved FDA nicotine addiction warning, not as replacements. Thus, we designed study stimuli to contain the new concentration labels in addition to the FDA warning label (Figure 1, “Sample Study Stimuli”).

Figure 1.

Examples of novel nicotine concentration labels and study stimuli. Note. Within the “Examples of New Labels” section, three nicotine concentrations are labeled on the thermometer for illustrative purposes. However, for the actual study stimuli, only one concentration was displayed on each stimuli.

Participants and Procedures

Study procedures were approved by the Oberlin College Institutional Review Board. In 2020, 826 AYAs (ages 13–24 years) completed an anonymous online survey. Of note, despite underage use of ENDS being prohibited, high school-aged participants were included in the study because rates of ENDS use are highest among this age group.¹ Participants were recruited by Qualtrics Online Sample, a secure market research service. Basic eligibility criteria included being ages 13–24 years and endorsing ever using an ENDS product on a screening question assessing lifetime ENDS use. Additional quotas were set to ensure diversity in terms of region of the country (ie, Northeast, Midwest, West, South), sex (ie, an approximately equal number of males and females), race/ethnicity (ie, at least 30% racial/ethnic minorities), and ENDS use experience (ie, a minimum of 50% had to report using ENDS at least weekly to ensure familiarity with ENDS use). Upon survey completion, Qualtrics compensated participants based on the terms of pre-established agreements between Qualtrics and their panelists. The specific amount of compensation was not revealed to the research team, but it was equal to or less than $8 (which was the cost paid to Qualtrics per participant).

Ensuring Data Quality

Qualtrics Online Sample guarantees “good completes,” meaning high-quality, usable data. Qualtrics automatically includes several features to eliminate poor-quality responders. First, a participant must finish the survey (but they may skip individual questions per IRB requirements). Second, Qualtrics prohibits participants, who have a unique ID, from taking the survey multiple times. Third, Qualtrics excludes participants who complete the survey unreasonably quickly. Finally, Qualtrics allows researchers to embed attention check questions to identify inattentive responders. We embedded several attention check items throughout the survey that had question stems that mirrored the real questions, making them difficult to identify. Qualtrics replaced all participants who failed the attention checks with “good completes” free of charge.

Demographics

Participants reported on the region of the country in which they reside, biological sex, race (which was dichotomized to non-Hispanic white vs other due to insufficient sample sizes for racial/ethnic minority groups), and age (which was dichotomized to reflect legal status for purchasing ENDS products [ie, not legal (under 21 years) vs. legal (21 years and older)]).

ENDS-related Constructs

Participants reported on their age at ENDS use onset. They also reported on ENDS devices used in the past month (ie, disposable vape, vape-pen, JUUL, pod system other than JUUL, Mod/APV) from which a variable was calculated reflecting the total number of devices used. Next, they reported on where they typically acquire their END liquids/pods (someone gives it to me, I buy it at a store like a gas station, I buy it at a vape shop, I buy it online, other [write-in option]) from which a variable was created that reflected personal purchasing of liquids (no/yes). Finally, a variable reflecting weekly ENDS use status (no/yes) was created from past-month vaping frequency.

Awareness of the FDA Nicotine Warning Label and Nicotine Concentration Labels

Participants responded “true” or “false” to the following statements: “There is a nicotine warning printed on the packaging of e-liquids or vape pods that contain nicotine” and “The nicotine concentration (or strength) of e-liquids and/or vape pods is printed on the packaging of e-liquids/pods.”

Perceptions of Nicotine Strength and Addictive Potential

All participants viewed three blocks of stimuli that depicted nicotine concentrations on product packaging using mg/mL, percent, and one of the four new labels; participants were randomized to view one of the four new label conditions to minimize participant burden. Within the new label conditions, the numerical values for the nicotine concentrations presented on the label were depicted using mg/mL based on prior research suggesting that youth and adults better understand concentrations presented as mg/mL than percent.¹⁰ Within each block, participants viewed in a randomly presented order nicotine concentrations that were equivalent to 3, 6, 18, 30, 40, and 50 mg/mL. These six nicotine concentrations were chosen for inclusion based on their popularity in the market and the fact that each was the best exemplar of a strength category as determined by subject matter expert opinion in prior research (ie, 3 mg/mL = very low; 6 mg/mL = low, 18 mg/mL = medium, 30 mg/mL = high, 40 mg/mL = very high, 50 mg/ml = very high; see Morean et al.¹⁰ for a thorough description). After viewing each stimuli, participants reported on perceived nicotine strength of that product (nicotine-free [contains no nicotine], very low, low, medium, high, very high, I don’t know). For each stimuli, participants also reported on perceived addictive potential associated with using that concentration (not at all, slightly, moderately, very, extremely).

Rank Ordering of Nicotine Concentration Labels

After completing the blocks of nicotine strength and addictiveness ratings, participants viewed examples of all six nicotine concentration label types (ie, mg/mL, percent, and all four new labels) and were asked to rank ordered them from best (1) to worst (6) in terms of “how well each [showed] nicotine strength and the amount of risk associated with using the product.” Note that nicotine strength and risk were combined within the question stem.

Analytic Plan

Descriptive Statistics

We calculated means and standard deviations for continuous variables and percentages for categorical variables.

Correct Nicotine Concentration Classifications and Perceptions of Nicotine Strength and Addiction Potential

First, we calculated the mean number of “correct” strength classifications for each label type. As was done in prior research,^10,20 correct classifications were based on the correspondence between a participant’s response and the opinions of subject matter experts that we obtained during our earlier work.¹⁰ For example, the majority of subject matter experts rated a nicotine concentration of 50mg/mL as “very high,” so a correct response for the nicotine concentration of 50mg/mL was “very high.” ¹⁰ All other responses were coded as incorrect. We then examined both the raw mean number of correct responses by label type and the percentage of responses for each label that reflected underestimates and overestimates of nicotine strength. Of note, underestimates of nicotine strength are of greatest concern given that this may lead to unintended exposure to higher nicotine levels and increase related risk. Rank ordering of labels also was considered.

Next, we ran a related-samples Wilcoxon signed-rank test to evaluate whether mg/mL or percent was associated with more correct responses. Whichever market label outperformed the other would serve as the comparison group for all further relevant analyses.

We then ran a Generalized Estimating Equation model (GEE) with label type (5-levels) as a within-subjects factor predicting the total number of correct nicotine strength responses. A Poisson distribution with log link was specified with an unstructured correlation matrix. The following covariates were included based on our expectations that each may relate to an individual’s ability to correctly classify the strength of ENDS liquids: legal purchasing status, personal purchase of ENDS liquids, awareness of the FDA nicotine warning on ENDS product packaging, awareness that nicotine concentrations are printed on ENDS product packaging, familiarity with market labels (mg/mL only, percent only, both, neither), age at ENDS use onset, total number of ENDS devices used in the past month, and weekly ENDS use status. Basic demographics (sex, race, region of the country) also were included as covariates.

While the number of correct strength categorizations was the primary outcome, we also examined the impact of label type, nicotine concentration, and their interactions on perceived nicotine strength and addiction potential, respectively. Two sets of linear mixed models were fit for each outcome: one comparing the four new labels to mg/mL and one comparing the four new labels only to each other. Each model contained the same covariates as were included in the GEE model described above. An unstructured variance-covariance matrix was used to account for the correlation among repeated observations on an individual. Significance was set at 0.05. Post-hoc tests were used to explain significant main and interactive effects.

Results

Descriptive Statistics

Descriptive statistics are presented in Table 1.

Table 1.

Descriptive Statistics

	Percent or M(SD)
Region of the U.S.
Northeast	20.0%
Midwest	24.3%
South	38.1%
West	17.6%
Sex
Male	47.3%
Female	52.7%
Legal age to purchase ENDS products
No (14–20 years old)	67.9%
Yes (21–24 years old)	32.1%
Race
Race other than non-Hispanic White	35.5%
White	64.5%
Purchases own ENDS liquids/pods
No	69.1%
Yes	30.9%
Aware of FDA nicotine warning
No	20.5%
Yes	79.5%
Aware of nicotine concentration on packaging
No	30.1%
Yes	69.9%
Familiarity with market labels (mg/mL, %)
Neither	30.3%
mg/mL only	16.0%
Percent only	30.5%
Both	23.2%
ENDS use at least weekly
No	43.0%
Yes	57.0%
Age at ENDS use onset	16.57 (2.42)
Total # ENDS devices used (past month)	2.02 (1.67)

Descriptive Information about the Nicotine Concentration Labels and Rank Ordering

Raw data indicated the new nicotine concentration labels were associated with more correct strength responses (out of 6) than was either mg/mL or percent (Table 2). Rates of underestimating nicotine strength, which raises concerns about inadvertent exposure to high nicotine levels, were considerable for mg/mL and especially high for percent. In contrast, rates of underestimating strength were low for all new labels. Finally, current market labels were most commonly ranked last or second-to-last while the new vertical thermometer and horizontal stoplight labels were ranked in first or second place by the majority of participants (Table 2).

Table 2.

Nicotine Strength Ratings and Rank Ordering of Labels

Market labels	Nicotine strength ratings					Rank ordering
	Total # correct (of 6)	Breakdown of responses (%)				Rankings (%)
	(M [SD])	Correct	Underestimate	Overestimate	Don’t know	1st	1st or 2nd
mg/mL	2.14 (1.52)	35.7	36.8	15.3	12.2	4.1	14.2
Percent nicotine	1.50 (1.08)	24.9	63.5	7.4	4.1	6.1	11.9
New labels
Text only	5.33 (1.36)	88.9	4.8	6.0	0.4	6.1	11.2
Safety symbol	5.23 (1.37)	87.2	6.8	5.4	0.6	8.3	16.3
Vertical thermometer	5.28 (1.51)	88.0	6.7	4.5	0.8	45.2	76.9
Horizontal stoplight	5.47 (1.14)	91.2	4.9	3.3	0.6	30.2	69.6

Comparing the Total Number of Correct Responses Associated with Current Market Labels

The related-samples Wilcoxon signed-rank test indicated that viewing nicotine concentrations in mg/mL resulted in more correct strength responses than did viewing concentrations presented in percent (49 375.50 [4186.82], p < .001). Thus, the new labels were tested only against mg/mL and not percent in further models.

GEE Model Predicting Total Number of Correct Strength Responses

All new labels outperformed the mg/mL label (Table 3). Although variables were treated as count data for the analysis, we present the estimated marginal means and standard errors here for context (mg/mL = 2.02[0.44], vertical thermometer = 4.89[0.12], caution label = 4.91[0.11], text label = 5.07[0.11], horizontal stoplight = 5.18[0.11], p <.001). Pairwise comparisons showed that the horizontal stoplight label also outperformed both the vertical thermometer label and the caution label in terms of number of correct responses (p < .05).

Table 3.

Parameter Estimates for the GEE Model with Label Type and Covariates as Predictors of the Total Number of Correct Nicotine Strength Responses

	B	Std. Error	95% CI		Wald
Intercept	1.13	0.09	0.95	1.30	167.22***
Region of the U.S.
Northeast	–0.07	0.03	–0.13	–0.02	8.02**
Midwest	–0.01	0.02	–0.06	0.03	0.35
South	–0.01	0.02	–0.05	0.03	0.21
West (ref)	--	--	--	--	--
Label Type
Text Only	0.92	0.03	0.87	0.98	1032.52***
Safety Signal	0.89	0.03	0.83	0.95	824.04***
Horizontal stoplight	0.94	0.03	0.88	1.00	957.57***
Vertical thermometer	0.89	0.03	0.82	0.95	730.94***
mg/mL (ref)	--	--	--	--	--
Sex
Male	0.20	0.02	0.17	0.23	143.53***
Female (ref)	--	--	--	--	--
Legal age to purchase ENDS products
No	0.20	0.02	0.16	0.25	75.89***
Yes (ref)	--	--	--	--	--
Race
Race other than non-Hispanic White	–0.05	0.02	–0.09	–0.01	5.66*
White (ref)	--	--	--	--	--
Purchases own END liquids/pods
No	0.00	0.02	-0.04	0.04	0.00
Yes (ref)	--	--	--	--	--
Aware of FDA nicotine warning
No	–0.02	0.21	–0.06	0.02	0.83
Yes (ref)	--	--	--	--	--
Aware of nicotine concentration on packaging
No	0.01	0.02	–0.03	0.05	0.33
Yes (ref)	--	--	--	--	--
Familiarity with market labels (mg/mL,%)
Neither	0.04	0.02	–0.01	0.09	2.73
mg/mL only	0.04	0.02	–0.01	0.09	2.68
Percent only	–0.01	0.02	–0.05	0.04	0.13
Both	--	--	--	--	--
Age of onset of ENDS use	–0.03	0.00	–0.04	–0.02	53.83***
Total # ENDS devices used (past month)	–0.03	0.01	–0.04	–0.01	14.76***
ENDS use at least weekly
No	0.01	0.02	–0.03	0.06	0.31
Yes (ref)	--	--	--	--	--

*p < .05

^**p < .01

^***p < .001

With regard to continuous model covariates, more correct responses were associated with an earlier age of onset of ENDS use and with using a greater total number of ENDS devices in the past month (p < .001; Table 3). With regard to categorical covariates, male participants (4.61 [0.07]) provided more correct responses than did female participants (3.78 [0.05], p < .001). Interestingly, underage individuals (ie, <21 years; 4.62 [0.06]) provided more correct strength responses than did individuals who could legally purchase ENDS products (3.77 [0.08] p <.001). Non-Hispanic white participants (4.28 [0.06]) provided more correct responses than did participants of other races/ethnicities (4.08 [0.07], p = .017). Finally, participants who resided in the Northeast (3.97 [0.09]) provided fewer correct responses than did participants who resided in the other three quadrants of the US (Midwest = 4.22 [0.70]; South = 4.24 [0.07]; West = 4.28[0.08], p-values < .05).

Linear Mixed Models

With regard to nicotine strength ratings, there was a significant interaction of label and concentration (F[20 902] = 52.29, p < .0001) accompanied by significant main effects of both label (F[4867] = 36.56, p < .0001) and concentration (F[5823)] = 2621.83, p < .0001). Strength ratings for the new labels were significantly lower compared to the mg/mL label at the 3 mg/mL concentration, were not significantly different at 6 mg/mL, and were significantly higher at 18, 30, 40, and 50 mg/mL concentrations (see Figure 2, “Nicotine Strength Ratings”). The main effect for concentration indicated that higher strength ratings were assigned as nicotine concentration increased, and the main effect of label type indicated that strength ratings for mg/mL were lower than were strength ratings for the new labels overall. The only significant model covariate, an older age of onset was associated with providing higher strength ratings overall (B = .01, SE = .005, p = .04).

Figure 2.

Least squares means for ratings of nicotine strength and addictiveness by label type overall and by label type across increasing nicotine concentrations.

In the model comparing the new labels amongst themselves, there was a significant main effect of concentration (F[5811] = 2729.95, p < .001), again indicating that higher strength ratings were associated with increasing nicotine concentrations. Neither the label by concentration interaction (F[15 811] = 1.47, p = .11) nor the main effect of label type (F[3794] = .92, p = .43) were statistically significant, indicating no significant differences among the new label types on nicotine strength ratings. Again, an older age of onset was the only covariate associated with strength ratings (B = .01, SE = .006, p = .04).

With regard to perceived addictiveness, in the model comparing the new labels to the mg/mL label there was a significant interaction of label and concentration (F[20 837] = 13.18, p < .0001), accompanied by significant main effects of label (F[4,823] = 4.51, p = 0.001) and concentration (F[5821] = 400.74, p < .0001; Figure 2, “Addictiveness Ratings”). Addictiveness ratings were significantly lower for the new labels compared to the mg/mL label at the 3 mg concentration and were significantly higher for the new labels compared to the mg/mL label at 30 mg/mL, 40 mg/mL, and 50 mg/mL concentrations. At 18mg, the vertical thermometer label was associated with significantly lower addictive ratings than was the mg/mL label (–0.24 [0.07], p < .001). The remaining new labels did not differ from the mg/mL label at 18mg. The main effects for concentration and label type mirrored those observed for strength ratings, with higher addictiveness ratings associated with increasing nicotine concentrations and the mg/mL label being associated with lower addictiveness ratings overall. In terms of model covariates, not knowing that nicotine concentrations are printed on ENDS product packaging was associated with higher perceived addictiveness (B = .11, SE = .05, p = .02).

In the model comparing the new labels amongst themselves, there were significant main effects of concentration (F[5821] = 372.36, p < .0001) and label (F[3799] = 3.26, p = .02). Higher concentrations were perceived as more addictive, and the vertical label was associated with significantly lower addictiveness ratings than each of the other new labels. As was the case in the prior model, not knowing that nicotine concentrations are printed on ENDS product packaging was associated with higher perceived addictiveness (B = 0.11, SE = 0.05, p = .03).

Discussion

The current study replicates and extends prior research indicating that young users of ENDS products generally have a poor understanding of the nicotine concentration labels that are most commonly employed for commercially available ENDS liquids/pods (ie, mg/mL, percent).^10,20 Importantly, we demonstrated that AYA’s understanding of the nicotine strength of ENDS liquids can be enhanced through improved labeling. Indeed, relative to the best performing market label (ie, mg/mL), all four novel labels produced an improved understanding of nicotine strength as defined by more “correct” strength classifications (ie, strength perceptions that were more in line with those of nicotine/tobacco researchers). Product strength ratings also were significantly higher for the new labels than the mg/mL label at higher concentrations (ie, 18, 30, 40, and 50 mg/mL). Although the new labels were designed primarily to be informational, they also were associated with increased perceptions of addiction potential at higher nicotine concentrations (ie, 30, 40, and 50 mg/mL). Importantly, both strength and addictiveness ratings for the new labels were lower than were ratings for the mg/mL label at the 3mg/mL concentration. These results likely are linked to the inclusion of the descriptor “very low nicotine” which corresponded to the 3mg/mL concentration on the new labels. Thus, it seems that the use of “very low nicotine” may have unintended consequences, and future research should consider labeling 3 mg/mL as “low nicotine,” “lower nicotine,” or simply reserving strength descriptors for higher nicotine concentrations. Although some alterations may be needed, this study indicates that novel labeling could improve the way we impart information about nicotine strength and addiction-related risk to people who use ENDS products. That said, further research is needed to establish an accurate dose-response relationship between the nicotine concentration of ENDS liquids and likelihood of addiction.

Ultimately, our results provided the greatest support for the utility of the horizontal stoplight design for labeling nicotine concentration. Of note, this format also has been shown to communicate effectively the strength of THC/CBD in cannabis products to users.^36,37 While all new labels outperformed mg/mL in terms of correct strength responses, the horizontal stoplight label also outperformed the vertical thermometer and caution sign-shaped labels. In addition, the horizontal thermometer label outperformed the vertical thermometer label in terms of conveying the addiction potential of e-liquids at higher nicotine concentrations (eg, 30, 40, and 50 mg/mL). Although the horizontal label did not explicitly outperform the text-based label in any model, when considered in concert with participants’ raw number of correct responses; raw percentages of underestimates and overestimates of nicotine strength; and label rank ordering (Table 2), there was the greatest support for the horizontal stoplight label.

Although the horizontal stoplight outperformed the vertical thermometer label where differences were observed, the performance of the vertical thermometer deserves consideration given that it was rank ordered in first and second place by the greatest percentage of participants. We suspect that the vertical thermometer’s comparatively worse performance related to its scaling; the vertical thermometer uniquely visually depicts nicotine concentrations from 0 to 60 mg/mL (Figure 1). In contrast, the other labels do not actually show numerical values across the range of commercially available nicotine concentrations. In the case of the vertical thermometer, for example, 30 mg/mL visually appears situated in the moderate space of the scale that ranges of 0 to 60 mg/mL despite its label as “high nicotine.” Thus, it may be interpreted visually as being “medium” nicotine despite its descriptor. Even for 50 mg/mL (“very high”) which is near the top of the scale, participants could see the scale’s upper limit of 60 mg/mL which appears to have indicated that any value less than 60 mg/mL is neither the strongest nor most addictive concentration. Given participants’ greater preference for the vertical thermometer, future research may consider testing a label version in which the low, medium, high, and very high concentrations are clustered more closely together within their respective strength category as opposed to being evenly spaced as was the case here.

The study findings must be interpreted alongside several limitations. First, we relied on data from a convenience sample of online panelists ages 13–24 years, which is associated with several potential limitations including poor generalizability to the population as a whole and biased results based on who chose to complete the survey. Additional research is needed to evaluate the extent to which the current findings apply to other age groups, to those who have never used ENDS, and to those who are very experienced ENDS users. Second, the nicotine strength categories ranging from “very low” to “very high” were based on correspondence to the opinions of ENDS researchers. Although a more objective metric for the nicotine strength of ENDS liquids is needed, the current study provides strong evidence for proof-of-concept. In other words, the cut points for the strength categories may shift based on future research, but participants generally reported back the “correct” answer based on the information that was provided to them. This observation provides promising evidence that the labels can work as effective information delivery tools for nicotine concentration. Third, nicotine concentration and addictive potential do not necessary track in a linear fashion. Nicotine delivery can be related to yet distinct from nicotine concentration. A variety of additional factors including vaping frequency and device power, which varies considerably across device types, impact nicotine delivery.^40–45 While we focused on nicotine concentration alone here, other metrics such as nicotine flux, the rate at which nicotine is emitted by an ENDS device per second, may be a closer proxy to nicotine delivery⁴¹ and merit future investigation. Fourth, all nicotine concentration labels (ie, market labeling, the new nicotine concentration labels) were included on the upper right corner of ENDS product packaging. Alternate positioning of the labels should be tested to determine the optimal placement for nicotine concentration labels. Finally, research is needed to evaluate the extent to which employing novel labeling impacts real-world behavior. Although new labeling may improve understanding of nicotine strength and influence perceptions of addiction potential, knowledge and perceptions do not always translate to behavior. For example, more informative labeling may serve to deter uptake of ENDS and/or the use of high-nicotine ENDS among young nonusers and also could facilitate successful switching from smoking to exclusive ENDS use in this population as a means of harm reduction. However, this needs to be evaluated directly, and there is a need to explore unintended consequences of the proposed labeling. For instance, increased perceptions of addictive potential, specifically, may contribute to deterring ENDS uptake or the use of high nicotine concentrations among young users. However, increased perceptions of addictiveness (especially if they are found to be higher than risk perceptions for the addictiveness of cigarettes) also could deter people who currently smoke cigarettes from attempting to switch from smoking to exclusive ENDS use.

In sum, the current study indicates that labeling the nicotine concentration of ENDS liquids is one avenue through which AYAs could be better informed about the constituents of ENDS products and related risk. While additional modifications may be needed to fully optimize the new nicotine concentration labeling (eg, replacing the term “very low” with “low”), the horizontal stoplight label was the most promising for accurately conveying information about nicotine concentration and impacting perceptions of addictive potential.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Acknowledgments

The authors thank Izzy Lederman and Liam Mai for their help with data collection and all participants for contributing to this work.

Declaration of Interests

Dr. Eissenberg is a paid consultant in litigation against the tobacco industry and also the electronic cigarette industry and is named on one patent for a device that measures the puffing behavior of electronic cigarette users, on another patent application for a smartphone app that determines electronic cigarette device and liquid characteristics, and a third patent application for a smoking cessation intervention. Dr. Morean holds a restricted stock agreement with Gofire, Inc. although this is not directly relevant to the current study. Dr. Gueorguieva discloses the following interests unrelated to this work: royalties from book “Statistical Methods in Psychiatry and Related Fields” published by CRC Press and a United States patent application 20200143922 by Yale University: Chekroud, A., Krystal, J., Gueorguieva, R. and Chandra, A. “Methods and Apparatus for Predicting Depression Treatment Outcomes”. Drs. Wackoski, Delnevo, and Krishnan-Sarin have no conflicts of interest to declare related to the current study.

Funding

Dr. Morean’s effort was supported by grant number U54CA180908 from the National Cancer Institute (NCI) and the US Food and Drug Administration (FDA) Center for Tobacco Products (CTP) and by grant number U54DA036151 from the National Institute on Drug Abuse of the National Institutes of Health and the Center for Tobacco Products of the U.S. Food and Drug Administration. Dr. Eissenberg’s effort was supported by grant number U54DA036105 from the National Institute on Drug Abuse of the National Institutes of Health and the Center for Tobacco Products of the U.S. Food and Drug Administration. Efforts by Drs. Wackowski and Delnevo were supported in part by grant R37CA222002 from the National Cancer Institute and U54CA229973 from the National Cancer Institute and the U.S. Food and Drug Administration. Efforts by Drs. Krishnan-Sarin and Gueorguieva were supported by grant number U54DA036151 from the National Institute on Drug Abuse of the National Institutes of Health and the Center for Tobacco Products of the U.S. Food and Drug Administration. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the U.S. Food and Drug Administration.

Data Availability

Data will be made available upon reasonable request to the corresponding author, Meghan Morean, PhD ([email protected] ).