The effect of messaging about reductions in exposure to harmful chemicals on perceived risk of JUUL e-cigarettes

doi:10.21203/rs.3.rs-3870431/v1

Adults who smoke (AWS) who switch completely to e-cigarettes reduce their exposures to many harmful chemicals, suggesting that switching could reduce risk. AWS who believe e-cigarettes are less harmful than cigarettes are more likely to use e-cigarettes to switch away from smoking. Yet most AWS believe e-cigarettes are at least as harmful as cigarettes. This study assessed the effect on risk perceptions of communicating that switching can reduce exposure to harmful chemicals. In a randomized experiment, AWS (n=3485), dual users of cigarettes and e-cigarettes (DU; n=1756), and nonusers of tobacco (n=7316) viewed an ad for JUUL with or without a reduced-exposure message, and completed assessments of perceived risk (PR). Message exposure decreased PR of JUUL among AWS, while increasing PR of smoking, thus increasing the perceived risk differential between the two. Among DU, the message increased PR of smoking, but did not change PR of JUUL. Among nonusers, who rated the PR of all assessed tobacco products higher than did AWS and DU, the message decreased PR of JUUL, and did not affect PR of smoking. Accurate modified exposure communications have potential to shift the perceived risk differential of smoking and e-cigarettes in a way that could benefit public health.

Health Policy

e-cigarettes

modified risk tobacco product

electronic nicotine delivery system (ENDS)

risk messages

Cigarette smoking is responsible for 480,000 premature deaths per year in the US and more than 7 million deaths worldwide [1, 2]. While nicotine and nicotine dependence motivate persistent smoking, it is the products of combustion that account for the terrible health toll of smoking [1, 3]. This has led scientists and regulators, including the US FDA, to the concept of a “risk continuum” for tobacco and nicotine products [4–8]. The key principle is that it is the by-products of combustion that are responsible for almost all the harm of smoking [3, 8], and that products that deliver nicotine without combustion will have much lower risk than smoking.

Evidence suggests that e-cigarettes, referred to as electronic nicotine delivery systems (ENDS), which produce nicotine aerosols without combustion, lie nearer to the low-risk end of the continuum [5, 6, 9, 10], and thus hold promise for harm reduction among adults who smoke (AWS) who would not otherwise stop smoking. For example, studies of biomarkers assessing exposure to harmful chemicals in tobacco smoke show that switching to ENDS results in reductions of 85% or more – a degree of reduction similar to that seen when AWS quit all tobacco and nicotine completely [11–13]. A recent study showed that, compared to those adults who continue to smoke, AWS who had switched to JUUL^® (a brand of closed-system, pod-based ENDS using nicotine-salt e-liquids sold in the US) not only had biomarker levels indicating lower exposures to harmful chemicals, but also lower levels of biomarkers of inflammation and oxidative stress, and lesser respiratory symptoms[14, 15]; see also [16–18]. While ENDS have not yet been in use long enough to document favorable preventive health benefits of such switching, these observations make such benefits likely.

Theories of health behavior [19–22] posit that AWS would be motivated to switch away from smoking to ENDS if they understood that ENDS were associated with lower toxic exposures and thus likely lower risk. Studies have shown that more favorable risk perceptions of ENDS are associated with subsequent adoption of ENDS [23–26] among AWS and switching away from cigarettes with ENDS [26–29]. However, most AWS currently believe that using ENDS is at least as harmful as smoking, and such misperceptions have been increasing over time [26, 30, 31], possibly due to conflicting messages about the risks of ENDS [30–32]. This suggests the need for communication of factual information about reduced exposure to harmful chemicals from ENDS compared to cigarettes. Current evidence indicates considerable potential for this approach: understanding ENDS deliver fewer harmful chemicals than cigarettes was closely related to the perception that ENDS are less harmful than cigarettes, the interest in using ENDS among current non-ENDS-using AWS, and actual ENDS use among AWS but not among non-smokers [33].

The present study assesses the impact of a communication about an ENDS product’s reduced exposure to harmful chemicals on risk perceptions. A subsequent paper will address message effects on behavioral intentions, including the effect of risk perceptions on behavioral intentions.

The reduced-exposure message was part of an advertisement for JUUL designed for this research study. The message stated, in part, “Switching completely from cigarettes to JUUL reduces exposure to harmful chemicals in cigarette smoke to you and to those around you.” This message is factually supported by several studies showing drastic reductions in toxicants in the aerosol from US JUUL ENDS,[34] in biomarkers of exposure among AWS switching completely to that product,[11, 13] and in biomarkers of potential harm, such as systemic inflammation, oxidative stress, and respiratory symptoms, among long-term switchers. [15] Evaluation of comprehension of the message found that participants understood the message well, and gross misperceptions (e.g., JUUL being harmless) were rare.[35] Here, we analyze the effects of the message on risk perceptions of JUUL with samples differing in tobacco-use histories. To put these in context, we evaluated risk perceptions of other tobacco-use practices on the risk continuum, from using no tobacco product at all to smoking cigarettes. As the literature shows that the effects of health messages are influenced by perception of message credibility,[36, 37] the analyses also consider credibility as a potential moderator of message effects.

Design and procedures. This study was a randomized experiment, with US adults (18+) randomized to Message or Control conditions, after which risk perceptions were assessed. Participants were recruited by quota into five study groups: AWS planning and not planning to quit smoking, adults currently smoking and using ENDS (Dual Users), and adults who formerly used and never used tobacco.

Data were collected online. Participants viewed a video advertisement for JUUL that either included (Message condition) or did not include (Control) an exposure-reduction message, and then completed assessments. The study was determined to be exempt by an independent IRB, and participants were compensated by their respective consumer panels (for those recruited online) or $35 (for those recruited offline) for participation.

Participants. Inclusion criteria included age 18 or older, ability to read English, and fit into predefined demographic and tobacco status targets. Participants who had completed another tobacco survey within 30 days were excluded, as were employees of tobacco and vaping companies and litigants in suits against such companies or their relatives.

Recruitment was primarily online (via research panels, 91.9% of the sample), supplemented by offline recruitment (in-person, through street or mall intercept and from consumer-research-site databases) to reduce sole reliance on a single recruitment source. Demographic target quotas for each study group were set for age nested within gender, race/ethnicity, geographic region, and educational attainment. However, due to an initial survey programming error, recruitment quotas were not initially applied, resulting in over-recruitment of certain groups.

Altogether, 15,192 US-resident adults completed the survey and qualified for the study. As described in McCaffrey et al,[35] respondents who failed to complete the survey within 4 hours, who showed evidence of invalid/careless responding by speeding (≤ 1/3 the median completion time), who failed an attention check,[38] who failed a manipulation check (Message condition respondents who could not identify the content of the message in a multiple choice item[39, 40]), or who gave logically invalid responses, were excluded. The analytic sample included 12,557 participants; their demographics are shown in Table 1.

Analytic groups. For purpose of analysis, participants were classified into three groups based on their current tobacco use status, and weighted accordingly (see Table 2): AWS (combination of the two AWS groups), Dual Users, and Nonusers (combination of those who formerly used and never used tobacco).

Video stimuli: All participants were shown a 35-second online video advertisement that described JUUL, explaining that JUUL is an alternative to cigarettes for adult smokers, comes in tobacco and menthol flavors, provides smokers with a familiar experience, and does not create ash or smoke. The ad included a legally-mandated nicotine addiction warning (Supplement 1). Participants could view the video as many times as they wished, and it was available for re-view via a link during parts of the survey as well. For the Message condition only, the ad concluded with the following text, which was both displayed and voiced over (an additional 16 seconds):

“JUUL does not burn tobacco or produce smoke.

EVIDENCE TO DATE SHOWS:

Switching completely from cigarettes to JUUL reduces exposure to harmful chemicals in cigarette smoke to you and to those around you.”

This message, which is consistent with the reduced-exposure message authorized by FDA for IQOS, a reduced-exposure cigarette that electronically heats tobacco, is supported by analyses of JUUL aerosol chemical toxicology [34, 41, 42], and biomarker studies [11, 13, 14], as well as an exhaled-breath study [43, 44] relevant to bystanders’ exposures. The advertisement and message language were honed in preliminary qualitative and mixed-methods work with 144 AWS and nonusers of tobacco products.[35] As the proposed message is not yet authorized by FDA for dissemination, after completing the survey all participants were shown a debriefing statement (Supplement 1).

Assessment

All assessment was performed online. Except for demographics and tobacco history, assessments were administered after the video exposure. Assessments were largely drawn from existing literature[45, 46] and national surveys [47], and modified based on recent FDA guidance[40] or were developed for purposes of the study through cognitive testing. Survey items relevant to the current study (risk perceptions, message credibility) are presented in Table 3. Survey items relating to message comprehension, susceptibility to smoking, and intention to use JUUL are reported elsewhere.

Risk perceptions

Relative risk perceptions. To assess participants’ perceptions of the ‘risk continuum’ [4–8], participants were asked “In your opinion, how harmful are these behaviors to your health?” (from 0% harmful to health to 100%, 10% increments) followed by a list of five behaviors, presented in random order. This rating task was designed to encourage participants to consider the range of risks and their relative ordering and risk magnitude, reflecting the concept of a risk continuum.

Besides yielding perceived risks for each behavior, the data were used to compute the perceived risk differential (PRD) between smoking and JUUL, defined as the perceived risk of smoking minus the perceived risk of JUUL.

Absolute risk perceptions of JUUL.

Specific diseases. Using a 0-100% scale (with response options at 10-point intervals) participants rated the likelihood that a person using JUUL daily would experience mouth cancer, lung cancer, heart disease/heart attack, and chronic lung diseases. These disease-specific ratings were highly intercorrelated and formed a unidimensional scale (Cronbach’s α = 0.96, first factor eigenvalue = 3.5,), and were averaged to form a Perceived Risk of Diseases scale.

Harm to others. Participants used the same 0%-100% scale to rate the likelihood of daily JUUL use harming others nearby.

Risk of addiction. Participants used the 0%-100% scale to rate the likelihood of daily JUUL use leading to nicotine addiction.

Message credibility

Message condition participants rated the credibility of the message using two items assessing truthfulness and skepticism [46] which were averaged (with skepticism reverse-scored) to form a message credibility score.

Analysis

Continuous risk perception measures were analyzed using weighted independent samples t-tests comparing test and Control responses. As the distributions showed considerable skew (notably, many participants assigned 100% risk to smoking and dual use), robustness-checks were additionally performed on dichotomized measures (100% vs all others), using weighted likelihood ratio chi-square tests, where the distribution indicated. These largely mirrored parametric analyses, and are reported in Supplement 2. Weighted likelihood ratio chi-square tests were also used for categorical outcomes, and weighted Pearson correlations for associations between message credibility and risk perceptions. All analyses were conducted using R version 4.0.2 with the Type I error set to 0.05.

Table 1 shows the demographic characteristics of the analytic sample. Most respondents were female, 35 or older, married or cohabiting, had some college education, and earned $50,000 dollars or less. A mix of races and ethnicities were reported, with most identifying as non-Hispanic White (80.5%). Demographics differed between groups in expected ways. For example, AWS were less educated, and were more likely to be unmarried [48–50]. Even when statistically significant differences were observed between the Message and Control groups, the differences were too small to be material (Uncertainty Coefficients[51, 52] < 0.3%), indicating that randomization resulted in balanced assignment to conditions.

Table 1

Participant demographic characteristics by group and condition
N (Weighted %)		Adults who smoke			Dual users			Nonusers			Total
N (Weighted %)		Message	Control	p-value /UC¹	Message	Control	p-value /UC¹	Message	Control	p-value /UC¹	Total
Age	18–24	138 (9.0%)	117 (7.5%)	NS	130 (16.0%)	164 (17.1%)	NS	307 (7.6%)	355 (8.3%)	NS	1211 (9.2%)
	25–34	284 (14.8%)	334 (17.2%)		250 (30.3%)	270 (30.2%)		282 (7.4%)	301 (7.5%)		1721 (13.0%)
	35–49	515 (29.9%)	587 (31.5%)		304 (36.6%)	337 (35%)		610 (16.5%)	601 (14.8%)		2954 (22.6%)
	50–64	525 (31.7%)	544 (29.4%)		122 (14.6%)	139 (15.1%)		1087 (31.1%)	1133 (29.8%)		3550 (28.3%)
	65+	216 (14.5%)	225 (14.3%)		18 (2.6%)	22 (2.6%)		1234 (37.4%)	1406 (39.6%)		3121 (26.8%)
Sex	Male	763 (45.4%)	812 (46%)	NS	397 (45.8%)	455 (46.3%)	NS	1020 (28.9%)	1079 (28.4%)	NS	4526 (35.8%)
Sex	Female	913 (54.4%)	987 (53.6%)	NS	423 (53.7%)	472 (53%)	NS	2487 (70.8%)	2706 (71.3%)	NS	7988 (63.8%)
Race/ Ethnicity	White, non-Hispanic	1261 (79.4%)	1332 (77.1%)	NS	643 (80.6%)	718 (77.9%)	NS	2847 (82.0%)	3082 (82.0%)	NS	9883 (80.5%)
	Black, non-Hispanic	183 (9.0%)	182 (8.9%)		41 (4.2%)	63 (5.8%)		193 (5.3%)	216 (5.6%)		878 (6.4%)
	Hispanic	147 (7.2%)	193 (8.5%)		80 (8.5%)	85 (9.2%)		267 (7.5%)	277 (7.4%)		1049 (7.8%)
	Others, non-Hispanic	78 (4.4%)	92 (5.5%)		51 (6.7%)	61 (7.0%)		186 (5.2%)	202 (5.1%)		670 (5.3%)
Marital Status	Married	552 (33.5%)	562 (29.9%)	p = 0.01/ UC = 0.2%	253 (29.1%)	274 (27.2%)	NS	1871 (53.2%)	1953 (51.4%)	NS	5465 (43.1%)
	Living with partner	282 (15.6%)	318 (18.8%)		169 (21.8%)	197 (22.0%)		222 (6.4%)	240 (6.4%)		1428 (11.6%)
	Widowed	97 (6.1%)	88 (4.7%)		18 (2.6%)	25 (2.9%)		261 (8.0%)	295 (8.4%)		784 (6.7%)
	Divorced/ Separated	303 (18.7%)	376 (20.4%)		124 (16.4%)	147 (17.1%)		426 (13.1%)	520 (14.4%)		1896 (15.8%)
	Never married	444 (26.1%)	463 (26.2%)		260 (30.2%)	289 (30.8%)		740 (19.3%)	788 (19.4%)		2984 (22.8%)
Household income in past 12 months	Less than $25,000	515 (31.9%)	555 (32.8%)	NS	238 (31.8%)	277 (31.6%)	NS	502 (14.4%)	631 (16.9%)	p = 0.01/ UC = 0.1%	2718 (22.5%)
	$25,000 - $49,999	567 (33.6%)	621 (34.8%)		246 (29.9%)	290 (32.1%)		880 (25.4%)	908 (24.2%)		3512 (28.3%)
	$50,000 - $74,999	298 (18.2%)	319 (16.2%)		156 (18.1%)	165 (17.1%)		775 (22.3%)	843 (22.2%)		2556 (20.2%)
	$75,000 - $99,999	148 (7.8%)	166 (8.6%)		75 (8.3%)	95 (9.7%)		512 (14.4%)	591 (15.4%)		1587 (12.2%)
	$100,000 or more	150 (8.5%)	146 (7.6%)		109 (11.9%)	105 (9.5%)		848 (23.4%)	820 (21.3%)		2178 (16.7%)
Highest level of education completed	High school or less	672 (38.4%)	760 (40.9%)	NS	330 (40.8%)	361 (40.5%)	NS	690 (20.3%)	811 (22.1%)	p = 0.04/ UC = 0.1%	3624 (29.1%)
	Some college	727 (43.0%)	773 (42.8%)		366 (45.9%)	423 (46.6%)		990 (29.4%)	1108 (30.4%)		4387 (35.8%)
	College or more	279 (18.6%)	274 (16.3%)		128 (13.3%)	148 (12.9%)		1840 (50.3%)	1877 (47.4%)		4546 (35.1%)

¹ p-values were calculated from the likelihood chi-square tests between Message and Control groups. NS stands for “not significant,” based on the results of the likelihood chi-square tests. For statically significant associations, uncertainty coefficients (UC) were calculated to assess the effect sizes of the contingency tables.

Note Statistics represent unweighted N (weighted %).

Table 2

Analytic Group Definitions and Weighting
	Criteria for cigarettes	Criteria for ENDS	Criteria for other tobacco products	Weighting
Adults who Smoke (AWS)	• Smoked 100 + cigarettes in their life • Currently smoke every day or some days	• Have not used ENDS in the past 30 days	-	AWS planning to quit within 30 days and those not planning to quit in 30 days, originally recruited in roughly equal numbers, were weighted to their proportions in the AWS population (per PATH Wave 5: 14.4% and 86.6%, respectively).
Dual Users		• Have used ENDS in the past 30 days	-	Current JUUL users (50.0% of the Dual Users) were weighted to represent the JUUL ENDS market share of 30%, estimated from tracked sales data and commercial shipments.
Nonusers¹	• Former Users o Met lifetime criteria for ≥ 1 tobacco product(s) o Met lifetime criteria for all products tried* o Not used any tobacco product in the past 6 months • Never Users o Never met lifetime criteria for any tobacco product o Not currently using any tobacco product o Not used any tobacco in the past 6 months†			Former Users and Never Users were weighted to their proportions in the population of adults who do not smoke currently (per PATH Wave 5: 37.4% and 62.6%, respectively).
¹ See Supplementary Material 4 on Former and Never Users * This criterion was imposed due to a programming error that included only respondents who qualified as established users of all tobacco products they reported ever using. † Most (68.1%) had never used any tobacco product.

Risk perceptions

Participants’ perceptions of the risk continuum were anchored on the low end by the negligible perceived risk of No Tobacco, which was assigned a risk rating of “0%” by 91.9% of all respondents, with means of < 3.0% (out of 100%) in all groups.

Figure 1 shows perceived risk continuum, expressed by the mean risk ratings of Dual Use, Smoking, JUUL, and NRT, by condition and group. In all groups and both conditions, the ordering of risk was the same: Dual Use > Smoking > JUUL > NRT > No Tobacco, consistent with risk continuum concepts [4–8]. Results for Perceived Risk of Diseases, Harm to Others, and Nicotine Addiction are presented in Fig. 2.

AWS

Relative risk perceptions. Among AWS, smoking was perceived to carry substantial risk, rated at around 85% (out of 100%) in both conditions. Exposure to the message significantly increased the perceived risk of smoking, from 84.5–86.5% (p = 0.002).

Use of JUUL was perceived to carry sizable risk, averaging > 60% in both conditions. As expected, the message significantly reduced perceived risk of JUUL, from 67.9–62.7%, p < .001. As a result of the combination of these effects, the message resulted in a greater PRD for JUUL relative to smoking, i.e., a risk difference of 23.7 vs. 16.7 (p < .001) (Table 4).

In both the Message and Control conditions, Dual Use (that was additive in nature, i.e., smoking 10 cigarettes per day and using JUUL 10 times per day for the rest of your life [see Table 3]) was perceived as the most harmful behavior, and its risk was not affected by the message. In both conditions, the risk of Dual Use was perceived as significantly greater than the risk of smoking alone (p < 0.001). The fact that the message increased the perceived risk of Smoking (above) narrowed the difference between Smoking and Dual Use.

NRT was perceived as carrying non-negligible amount of risk, with risk scores > 45%, with no effect of message. JUUL use was seen as carrying more risk than NRT use in both the Message and Control conditions, though more so in the Control (21.2 vs 17.4 risk difference, p < .001).

Specific diseases. Among AWS, the message resulted in a significant decline in the perceived likelihood of smoking-associated diseases for a daily user of JUUL (Perceived Risk of Diseases means: 58.6% vs. 51.9%, p < .001).

Harm to others. The message significantly reduced the perceived likelihood of harm to others (from 40.2–32.3%, p < .001).

Risk of addiction. AWS perceived a substantial likelihood (at least 66%) of addiction in both conditions. Exposure to the message reduced the perceived risk of addiction to JUUL (from 70.4–66.3%, p < .001).

Dual Users

Relative risk perceptions. Dual Users’ perceived risk of JUUL use was lower than that reported by AWS, regardless of condition (Fig. 1). The message had no effect on perceived risk of JUUL or of Dual Use (ps > .05). However, the message significantly increased Dual Users’ perceived risk of Smoking (from 84.1–87.2%, p < .001), which also resulted in a significant increase in the PRD for JUUL versus Smoking (from 26.4 to 32.1, p < .001) (Table 4), and a narrowing of the difference between Smoking and Dual Use (from 4.6 to 2.5, p = .026). The message increased Dual Users’ perceived risk of NRT (from 43.8–46.3%, p = .046).

Specific diseases. The message had no effect on Dual Users’ perceived risk of JUUL for specific diseases.

Harm to others. Exposure to the message reduced Dual Users’ perception of risk of JUUL use to others (from 33.3–28.2%, p = .001).

Risk of addiction. The message did not affect Dual Users’ perceived risk of addiction, which was rated > 68% in both conditions.

Nonusers

Relative risk perceptions. Nonusers of tobacco perceived all assessed products as more harmful than did AWS and Dual Users (Fig. 1). Nonusers in both conditions gave Smoking near-maximum ratings of > 93% harm with no effects from the message. However, exposure to the message significantly decreased their perceived risk of JUUL (from 85.6–75.4%, p < .001), resulting in a significantly greater PRD between Smoking and JUUL (18.8 vs. 11.3, p < .001) (Table 4). Nonusers in both conditions perceived JUUL as having about 80–87% the harm of smoking. The message reduced Nonusers’ perceived risk of NRT (from 55.1–50.4%, p < .001), and dual use (from 96.1–95.5%, p = .031), although the magnitude of the change was negligible (Cohen’s d = 0.05).

Specific diseases. Compared to those in the Control condition, Nonusers in the Message condition perceived a lower likelihood of specific diseases (72.1% vs. 64.9%, p < .001).

Harm to others. Exposure to the claim significantly reduced Nonusers’ perception of harm to others (53.1–43.4%, p < .001).

Risk of addiction. Compared to the Control, participants in the Message condition reported significantly lower perceived risk of nicotine addiction (82.0–76.9%, p < .001) from using JUUL daily.

Message credibility and message rejection

There was a wide range in the degree to which participants in the Message condition found the message credible (range 1–5, median = 3.5, SD = 1.1). Across all participants, those who found the message more credible perceived JUUL use (including specific risk of diseases from JUUL use) and dual use as less harmful (Supplement 3). Similarly, higher credibility scores were associated with a higher PRD between cigarettes and JUUL (r = 0.48, p < 0.001). Credibility also correlated negatively with the perceived risk of NRT (r=-0.18, p < 0.001).

Based on credibility scores, participants were classified as either “Message Rejectors” (below the median of 3.5 on credibility) or “Message Acceptors” (≥ the median). On each of these perceived risk measures, “Message Rejectors” not only had significantly higher risk perception ratings than “Message Acceptors” but also significantly higher risk perceptions than Control respondents, who had not been exposed to the message.

This study assessed US adults’ risk perceptions of tobacco and nicotine products along the FDA-defined risk continuum [4-8], and examined the effect of providing information about reduced exposure to harmful chemicals when using a particular ENDS product – JUUL – compared to smoking. Respondents of various tobacco use status showed a rudimentary understanding of the risk continuum, ordering products in a manner consistent with expert opinion, i.e., Dual Use (that is additive in nature) > Smoking > JUUL > NRT > no use of any tobacco or nicotine product[4-8] . Seeing the message that complete switching to JUUL use results in less exposure to harmful chemicals than does smoking modestly lowered participants’ perceived risk of JUUL, consistent with the previous literature suggesting that accurate information can mitigate risk misperceptions.[33]

Although respondents ordered behaviors appropriately on the risk continuum, their quantified estimates of the relative risks were substantially at variance with the science. For example, respondents assigned NRT medications over 50% the risk they attributed to smoking. This likely reflects the well-documented misperceptions about the risk of NRT [53, 54], which in turn reflects misperceptions about the risks of nicotine and its role in smoking-related harm [55, 56].

Similarly, respondents who did not see the message assigned JUUL use a risk corresponding roughly to 83% the risk of smoking, despite the fact that data has led scientists and medical authorities to conclude that the risk of ENDS use is likely closer to 5% to 15% that of smoking [6, 9, 10, 57]. Seeing the reduced-exposure message lowered respondents’ risk perceptions of JUUL, but even those who saw the message still estimated its risk at 75% that of smoking. Thus, the effect of the message on risk perception of JUUL was modest, well below what scientists regard as likely [58]. Reasons for this may include the fact that some respondents did not believe the message. Also, the message did not specifically claim a reduction in risk, but only reduction in exposure to harmful chemicals, from which many respondents, but likely not all, reasonably inferred reduced risk (referred as “exposure-to-risk halo” by Seidenberg et al.[59]). This inference is recognized in the FDA’s regulation of tobacco products [60], which allows messages of reduced exposure when evidence of reduced risk is not yet available, but where it is “reasonably likely” based on reductions in exposure to toxicants.

The effect of the reduced-exposure message on perceived risk of JUUL was evident in AWS, who are the primary intended audience for this communication. As more favorable relative-risk perceptions are associated with adoption of ENDS [23-26], with switching away from smoking [26-28], and with avoiding resumption of smoking after having switched [26, 29], communicating accurately about reduced exposure to harmful chemicals may help some AWS move towards use of less harmful products to the benefit of public health, consistent with FDA’s strategic plan for reducing the disease burden of smoking [7]. The effect of the message, and of these risk perceptions, on study participants’ intention to use JUUL will be evaluated in a subsequent paper.

The message did not significantly change risk perceptions of JUUL in current dual users. This is likely because dual users already had lower risk perceptions for JUUL, even without being exposed to the message, which may be conveying information they already know and believe. Dual users may have lower perceived risk of JUUL use both because perceiving reduced risk from ENDS may have been what led them to adopt ENDS in the first place [23-26], and because of the tendency for people to see their own behaviors as less risky [61, 62].

Notably, the message increased perceived risks of smoking in AWS and dual users, likely by making salient the crucial role of combustion and smoke inhalation in tobacco-related harm to AWS, as well to others. This may encourage AWS and dual users to stop smoking, which would reduce their risk.

Nonusers of tobacco attributed greater risk to all products and practices than did current users. This may be both a cause and an effect of their non-use status. Nonusers also shifted their risk perceptions of JUUL lower when exposed to the message. It would be unfavorable to public health if this shift in risk perception led nonusers to use ENDS, but this may be unlikely, as even after seeing the message, these groups attributed to JUUL 80% the risk overall of smoking and a 65% risk of developing smoking-related diseases (Figure 2), rating the risk of JUUL use to be about 75 out of 100 (Figure 1).

Consistent with prior research on the effects of message credibility [36, 37], those who doubted or rejected the message assigned higher risk ratings to JUUL. Surprisingly, Message Rejectors reported risk perceptions higher than those of Control respondents who did not see the message at all. Message Rejectors may have been reacting against the message, perhaps because of their perception of the message source (as coming from the company that makes JUUL). It is well established that “source credibility” is low for advertisements [63] and when the source is a tobacco company [36, 64, 65]. JUUL in particular has been highly scrutinized in the media, which may further undermine the message’s credibility.

Message Rejectors may also have come into the study with greater risk misperceptions. This hypothesis is consistent with the finding that Rejectors also assigned higher risk to NRT, whose risk was not addressed by the message. In any case, message acceptance is an important moderator of message effects on risk perceptions, and possibly other outcomes.

Another factor limiting the effect of the message is the single exposure in an artificial study setting. Several studies have suggested that the brief exposure to the message in an experimental setting may have been a limiting factor in the modest effects of a message [66, 67]. Repeat exposure over time in more naturalistic settings, and from more credible sources, may enhance the effect of the message on risk perceptions.

While the Control group was not exposed to the message, they did view an advertisement that explained the concept of ENDS and JUUL; they do not represent the naïve state, but their responses do control for the context in which the message was communicated. The sample was diverse, but it was not fully representative of the US population. Importantly, representativeness is less crucial in a randomized experiment, where the focus is on the effects of the manipulation, rather than on estimating a population parameter [68]. Finally, the study evaluated a particular message expression about a particular product, so some conclusions may not be generalizable to risk messaging or risk perceptions in general, or to other products, even other ENDS.

The study’s strengths include the randomized experimental design, with a large and demographically diverse sample with varied tobacco-use histories. Risk perceptions were evaluated in several ways, and produced convergent results, making the findings robust. The survey items were largely drawn from prior research and were also subject to cognitive testing. Importantly, the message itself is supported by data from multiple studies, is consistent with a message authorized by the FDA under its provisions for reduced-exposure communications, and its wording was pre-tested with the target audience.

In this study, information about reduced toxicant exposure with JUUL, compared to cigarette smoking, shifted risk perceptions of JUUL in a more realistic direction, despite lingering misperceptions. The impact of the message was limited by participants’ skepticism of the information. Conveying accurate information about exposures to harmful chemicals may be one important step towards helping AWS move away from smoking and shift towards less harmful products. This is consistent with FDA’s Center for Tobacco Products’ identification of risk misperceptions as a significant issue that the agency is planning to address through a public education campaign [69].

This study was approved by the Advarra® Institutional Review Board.

Declaration of Interest Statement

Author S.A.M. is an employees of Juul Labs, Inc. Through PinneyAssociates, Inc, authors S.S. and S.K. provide consulting to Juul Labs, Inc. on tobacco harm reduction, on an exclusive basis.

Funding Statement

This work was supported by Juul Labs, Inc., which reviewed a final draft of the manuscript.

Data Availability Statement

Due to the nature of the research, due to legal and commercial reasons, supporting data are not available.

National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. Reports of the Surgeon General. The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Atlanta (GA): Centers for Disease Control and Prevention (US), 2014.
World Health Organization. WHO report on the global tobacco epidemic, 2017: monitoring tobacco use and prevention policies. Geneva: World Health Organization,2017.
Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. Publications and Reports of the Surgeon General. How Tobacco Smoke Causes Disease: The Biology and Behavioral Basis for Smoking-Attributable Disease: A Report of the Surgeon General. Atlanta (GA): Centers for Disease Control and Prevention (US), 2010.
Gottlieb S, Zeller M. A Nicotine-Focused Framework for Public Health N Engl J Med 2017;377:1111-1114. First published on 2017/08/17, 10.1056/NEJMp1707409.
Abrams DB, Glasser AM, Pearson JL et al. Harm Minimization and Tobacco Control: Reframing Societal Views of Nicotine Use to Rapidly Save Lives Annu Rev Public Health 2018;39:193-213. First published on 2018/01/13, 10.1146/annurev-publhealth-040617-013849.
Nutt DJ, Phillips LD, Balfour D et al. Estimating the harms of nicotine-containing products using the MCDA approach Eur Addict Res 2014;20:218-225. First published on 2014/04/10, 10.1159/000360220.
US Food and Drug Administration. FDA announces comprehensive regulatory plan to shift trajectory of tobacco-related disease, death.
Niaura R. Re-thinking nicotine and its effects Truth Initiative 2016.
Royal College of Physicians. Nicotine without smoke: Tobacco harm reduction. London: RCP, 2016.
McNeill A, Brose LS, Calder R et al. Evidence review of e-cigarettes and heated tobacco products 2018 A report commissioned by public health England London: Public Health England 2018;6.
Cohen G, Goldenson NI, Bailey PC et al. Changes in Biomarkers of Cigarette Smoke Exposure After 6 Days of Switching Exclusively or Partially to Use of the JUUL System with Two Nicotine Concentrations: A Randomized Controlled Confinement Study in Adult Smokers Nicotine Tob Res 2021;23:2153-2161. First published on 2021/06/24, 10.1093/ntr/ntab134.
D'Ruiz CD, Graff DW, Robinson E. Reductions in biomarkers of exposure, impacts on smoking urge and assessment of product use and tolerability in adult smokers following partial or complete substitution of cigarettes with electronic cigarettes BMC Public Health 2016;16:543. First published on 2016/07/13, 10.1186/s12889-016-3236-1.
Jay J, Pfaunmiller EL, Huang NJ et al. Five-Day Changes in Biomarkers of Exposure Among Adult Smokers After Completely Switching From Combustible Cigarettes to a Nicotine-Salt Pod System Nicotine Tob Res 2020;22:1285-1293. First published on 2019/11/07, 10.1093/ntr/ntz206.
Pulvers K, Nollen NL, Rice M et al. Effect of Pod e-Cigarettes vs Cigarettes on Carcinogen Exposure Among African American and Latinx Smokers: A Randomized Clinical Trial JAMA Netw Open 2020;3:e2026324. First published on 2020/11/19, 10.1001/jamanetworkopen.2020.26324.
Shiffman S, Oliveri DR, Goldenson NI et al. Comparing Adult Smokers who Switched to JUUL vs Continuing Smokers: Biomarkers of Exposure and of Potential Harm and Respiratory Symptoms Nicotine & Tobacco Research 2023. 10.1093/ntr/ntad197.
D'Ruiz CD, O'Connell G, Graff DW, Yan XS. Measurement of cardiovascular and pulmonary function endpoints and other physiological effects following partial or complete substitution of cigarettes with electronic cigarettes in adult smokers Regul Toxicol Pharmacol 2017;87:36-53. First published on 2017/05/10, 10.1016/j.yrtph.2017.05.002.
Christensen CH, Chang JT, Rostron BL et al. Biomarkers of Inflammation and Oxidative Stress among Adult Former Smoker, Current E-Cigarette Users-Results from Wave 1 PATH Study Cancer Epidemiol Biomarkers Prev 2021;30:1947-1955. First published on 2021/07/23, 10.1158/1055-9965.EPI-21-0140.
Oliveri D, Liang Q, Sarkar M. Real-World Evidence of Differences in Biomarkers of Exposure to Select Harmful and Potentially Harmful Constituents and Biomarkers of Potential Harm Between Adult E-Vapor Users and Adult Cigarette Smokers Nicotine Tob Res 2020;22:1114-1122. First published on 2019/09/30, 10.1093/ntr/ntz185.
Sheeran P, Maki A, Montanaro E et al. The impact of changing attitudes, norms, and self-efficacy on health-related intentions and behavior: A meta-analysis Health Psychol 2016;35:1178-1188. First published on 2016/10/18, 10.1037/hea0000387.
Ajzen I. The theory of planned behavior Organizational behavior and human decision processes 1991;50:179-211.
Rosenstock IM. The health belief model and preventive health behavior Health education monographs 1974;2:354-386.
Fishbein M. A theory of reasoned action: some applications and implications Nebr Symp Motiv 1980;27:65-116. First published on 1980/01/01.
Brose LS, Brown J, Hitchman SC, McNeill A. Perceived relative harm of electronic cigarettes over time and impact on subsequent use. A survey with 1-year and 2-year follow-ups Drug Alcohol Depend 2015;157:106-111. First published on 2015/10/29, 10.1016/j.drugalcdep.2015.10.014.
Elton-Marshall T, Driezen P, Fong GT et al. Adult perceptions of the relative harm of tobacco products and subsequent tobacco product use: Longitudinal findings from waves 1 and 2 of the population assessment of tobacco and health (PATH) study Addict Behav 2020;106:106337. First published on 2020/03/08, 10.1016/j.addbeh.2020.106337.
Wackowski OA, Delnevo CD. Young Adults' Risk Perceptions of Various Tobacco Products Relative to Cigarettes: Results From the National Young Adult Health Survey Health Educ Behav 2016;43:328-336. First published on 2015/08/26, 10.1177/1090198115599988.
Kim S, Shiffman S, Sembower MA. US adult smokers' perceived relative risk on ENDS and its effects on their transitions between cigarettes and ENDS BMC Public Health 2022;22:1771. First published on 2022/09/20, 10.1186/s12889-022-14168-8.
Evans AT, Henderson KC, Geier A et al. What Motivates Smokers to Switch to ENDS? A Qualitative Study of Perceptions and Use Int J Environ Res Public Health 2020;17. First published on 2020/12/03, 10.3390/ijerph17238865.
Persoskie A, O'Brien EK, Poonai K. Perceived relative harm of using e-cigarettes predicts future product switching among US adult cigarette and e-cigarette dual users Addiction 2019;114:2197-2205. First published on 2019/07/07, 10.1111/add.14730.
Farsalinos KE, Romagna G, Voudris V. Factors associated with dual use of tobacco and electronic cigarettes: A case control study Int J Drug Policy 2015;26:595-600. First published on 2015/02/18, 10.1016/j.drugpo.2015.01.006.
Huang J, Feng B, Weaver SR et al. Changing Perceptions of Harm of e-Cigarette vs Cigarette Use Among Adults in 2 US National Surveys From 2012 to 2017 JAMA Netw Open 2019;2:e191047. First published on 2019/03/30, 10.1001/jamanetworkopen.2019.1047.
Malt L, Verron T, Cahours X et al. Perception of the relative harm of electronic cigarettes compared to cigarettes amongst US adults from 2013 to 2016: analysis of the Population Assessment of Tobacco and Health (PATH) study data Harm Reduct J 2020;17:65. First published on 2020/09/20, 10.1186/s12954-020-00410-2.
Wilson S, Partos T, McNeill A, Brose LS. Harm perceptions of e-cigarettes and other nicotine products in a UK sample Addiction 2019;114:879-888. First published on 2019/01/05, 10.1111/add.14502.
Wackowski OA, Bover Manderski MT, Gratale SK et al. Perceptions about levels of harmful chemicals in e-cigarettes relative to cigarettes, and associations with relative e-cigarette harm perceptions, e-cigarette use and interest Addiction 2023;118:1881-1891. https://doi.org/10.1111/add.16258.
Chen X, Bailey PC, Yang C et al. Targeted Characterization of the Chemical Composition of JUUL Systems Aerosol and Comparison with 3R4F Reference Cigarettes and IQOS Heat Sticks Separations 2021;8:168.
McCaffrey S, Shiffman S, Sembower M, Black R. Randomized Experimental Test of a Reduced-Exposure Message for an E-cigarette: Comprehension and Related Misperceptions Qeios. 10.32388/ZCGESA.
Hammond SL. Health advertising: The credibility of organizational sources Annals of the International Communication Association 1987;10:673-628.
Schmidt AM, Ranney LM, Pepper JK, Goldstein AO. Source Credibility in Tobacco Control Messaging Tob Regul Sci 2016;2:31-37. First published on 2016/08/16, 10.18001/TRS.2.1.3.
Kung FY, Kwok N, Brown DJ. Are attention check questions a threat to scale validity? Applied Psychology 2018;67:264-283.
O’Brien EK, Baig SA, Persoskie A. Developing and Validating Measures of Absolute and Relative E-Cigarette Product Risk Perceptions: Single Items Can Be Surprisingly Comprehensive Nicotine & Tobacco Research 2021;24:316-323. 10.1093/ntr/ntab156.
U.S. Food and Drug Administration. Tobacco products: Principles for designing and conducting tobacco product perception and intention studies. In: U.S. Department of Health and Human Services (ed). Silver Spring, MD Aug 2022.
Crosswhite MR, Jeong LN, Bailey PC et al. Non-Targeted Chemical Characterization of JUUL-Menthol-Flavored Aerosols Using Liquid and Gas Chromatography Separations 2022;9:367.
Crosswhite MR, Bailey PC, Jeong LN et al. Non-Targeted Chemical Characterization of JUUL Virginia Tobacco Flavored Aerosols Using Liquid and Gas Chromatography Separations 2021;8:130.
Oldham MJ, Bailey PC, Castro N et al. Prediction of potential passive exposure from commercial electronic nicotine delivery systems using exhaled breath analysis and computational fluid dynamic techniques J Breath Res 2021;15. First published on 2021/09/21, 10.1088/1752-7163/ac2884.
Oldham MJ, Sehgal A, Cohen G et al. Room air constituent concentrations from use of electronic nicotine delivery systems and cigarettes using different ventilation conditions Sci Rep 2021;11:1736. First published on 2021/01/20, 10.1038/s41598-021-80963-9.
McCaffrey SA, Zdinak JP, Plunkett S et al. Development and validation of behavioral intention measures of an E-vapor product: intention to try, use, dual use, and switch Health and Quality of Life Outcomes 2021;19:123. 10.1186/s12955-021-01764-2.
Fix BV, Adkison SE, O’Connor RJ et al. Evaluation of modified risk claim advertising formats for Camel Snus Health Education Journal 2017;76:971-985. 10.1177/0017896917729723.
United States Department of Health and Human Services, National Institutes of Health, National Institute on Drug Abuse et al. Population Assessment of Tobacco and Health (PATH) Study [United States] Public-Use Files: Inter-university Consortium for Political and Social Research [distributor], 2023.
Hoover DS, Vidrine JI, Shete S et al. Health Literacy, Smoking, and Health Indicators in African American Adults J Health Commun 2015;20 Suppl 2:24-33. First published on 2015/10/30, 10.1080/10810730.2015.1066465.
Casetta B, Videla AJ, Bardach A et al. Association Between Cigarette Smoking Prevalence and Income Level: A Systematic Review and Meta-Analysis Nicotine Tob Res 2017;19:1401-1407. First published on 2016/11/01, 10.1093/ntr/ntw266.
Ramsey MW, Jr., Chen-Sankey JC, Reese-Smith J, Choi K. Association between marital status and cigarette smoking: Variation by race and ethnicity Prev Med 2019;119:48-51. First published on 2018/12/24, 10.1016/j.ypmed.2018.12.010.
Theil H. On the estimation of relationships involving qualitative variables American Journal of Sociology 1970;76:103-154.
Cohen J. Statistical power analysis for the behavioral sciences. Routledge,2013.
Czoli CD, Fong GT, Mays D, Hammond D. How do consumers perceive differences in risk across nicotine products? A review of relative risk perceptions across smokeless tobacco, e-cigarettes, nicotine replacement therapy and combustible cigarettes Tob Control 2017;26:e49-e58. First published on 2016/09/15, 10.1136/tobaccocontrol-2016-053060.
Shiffman S, Ferguson SG, Rohay J, Gitchell JG. Perceived safety and efficacy of nicotine replacement therapies among US smokers and ex-smokers: relationship with use and compliance Addiction 2008;103:1371-1378. First published on 2008/10/16, 10.1111/j.1360-0443.2008.02268.x.
Byron MJ, Jeong M, Abrams DB, Brewer NT. Public misperception that very low nicotine cigarettes are less carcinogenic Tob Control 2018;27:712-714. First published on 2018/01/25, 10.1136/tobaccocontrol-2017-054124.
O'Brien EK, Nguyen AB, Persoskie A, Hoffman AC. U.S. adults' addiction and harm beliefs about nicotine and low nicotine cigarettes Prev Med 2017;96:94-100. First published on 2016/12/31, 10.1016/j.ypmed.2016.12.048.
Levy DT, Borland R, Lindblom EN et al. Potential deaths averted in USA by replacing cigarettes with e-cigarettes Tob Control 2018;27:18-25. First published on 2017/10/04, 10.1136/tobaccocontrol-2017-053759.
U.S. Food and Drug Administration. Scientific Review of Modified Risk Tobacco Product Application (MRTPA) Under Section 911(d) of the FD&C Act -Technical Project Lead.
Seidenberg AB, Popova L, Ashley DL, Wackowski OA. Inferences beyond a claim: a typology of potential halo effects related to modified risk tobacco product claims Tob Control 2020. First published on 2020/10/14, 10.1136/tobaccocontrol-2019-055560.
U.S. Food and Drug Administration, Center for Tobacco Products. Modified Risk Tobacco Product Applications Draft Guidance. In: U.S. Department of Health and Human Services (ed), Apr 2012.
Fotuhi O, Fong GT, Zanna MP et al. Patterns of cognitive dissonance-reducing beliefs among smokers: a longitudinal analysis from the International Tobacco Control (ITC) Four Country Survey Tob Control 2013;22:52-58. First published on 2012/01/06, 10.1136/tobaccocontrol-2011-050139.
McMaster C, Lee C. Cognitive dissonance in tobacco smokers Addict Behav 1991;16:349-353. First published on 1991/01/01, 10.1016/0306-4603(91)90028-g.
O’Neil J, Eisenmann M. An examination of how source classification impacts credibility and consumer behavior Public relations review 2017;43:278-292.
Zagona SV, Harter MR. Credibility of source and recipient's attitude: factors in the perception and retention of information on smoking behavior Percept Mot Skills 1966;23:155-168. First published on 1966/08/01, 10.2466/pms.1966.23.1.155.
Case KR, Lazard AJ, Mackert MS, Perry CL. Source Credibility and E-Cigarette Attitudes: Implications for Tobacco Communication Health Commun 2018;33:1059-1067. First published on 2017/06/18, 10.1080/10410236.2017.1331190.
Chen-Sankey JC, Kechter A, Barrington-Trimis J et al. Effect of a hypothetical modified risk tobacco product claim on heated tobacco product use intention and perceptions in young adults Tob Control 2023;32:42-50. First published on 2021/06/02, 10.1136/tobaccocontrol-2021-056479.
Pepper JK, Squiers LB, Peinado SC et al. Impact of messages about scientific uncertainty on risk perceptions and intentions to use electronic vaping products Addict Behav 2019;91:136-140. First published on 2018/11/06, 10.1016/j.addbeh.2018.10.025.
Stephenson J, Imrie J. Why do we need randomised controlled trials to assess behavioural interventions? BMJ 1998;316:611-613. First published on 1998/03/31, 10.1136/bmj.316.7131.611.
King BA, Toll BA. Commentary on Wackowski et al.: Opportunities and Considerations for Addressing Misperceptions About the Relative Risks of Tobacco Products among Adult Smokers Addiction 2023;118:1892-1894. https://doi.org/10.1111/add.16296.

The authors declare potential competing interests as follows: This work was supported by Juul Labs, Inc., which reviewed a final draft of the manuscript. Through PinneyAssociates, Inc, authors S.S. and S.K. provide consulting to Juul Labs, Inc. on tobacco harm reduction, on an exclusive basis. Author S.A.M. is an employee of Juul Labs, Inc.

SupplementaryMaterial1.docx
Supplementary Material 1
SupplementaryMaterial2Nonparametricanalyses.docx
Supplementary Material 2: Nonparametric analyses
SupplementaryMaterial3Messagecredibility.docx
Supplementary Material 3: Message credibility
SupplementaryMaterial4DifferencesbetweenFormerUsersandNeverUsers.docx
Supplementary Material 4: Differences between former users and never users

The effect of messaging about reductions in exposure to harmful chemicals on perceived risk of JUUL e-cigarettes

Status:

Version 1

Abstract

Figures

Introduction

Methods

EVIDENCE TO DATE SHOWS:

Assessment

Analysis

Results

Risk perceptions

AWS

Dual Users

Nonusers

Message credibility and message rejection

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1