Effects of Modified Risk Tobacco Product Claims on Consumer Responses

Abstract

Introduction

US tobacco manufacturers can seek authorization from the US Food and Drug Administration (FDA) to market products using modified risk tobacco product (MRTP) claims. To inform regulatory decisions, we examined the impact of MRTP claim specificity and content, including whether the claims produced halo effects (ie, inferring health benefits beyond what is stated).

Aims and Methods

Participants were 3161 US adult cigarette smokers. Using a two (general vs. specific) × 2 (risk vs. exposure) plus independent control design, we randomized participants to view one message from these conditions: general risk claim (eg, “smoking-related diseases”), general exposure claim (eg, “chemicals in smoke”), specific risk claim (eg, “lung cancer”), specific exposure claim (eg, “arsenic”), or control. Claims described the benefits of completely switching from cigarettes to the heated tobacco product IQOS.

Results

MRTP claims of any sort elicited a higher willingness to try IQOS relative to control (d = 0.09, p = .043). Claims also elicited lower perceived risk of disease and exposure to harmful chemicals for completely switching from cigarettes to IQOS (d = –0.32 and –0.31) and partially switching (d = –0.25 and d = –0.26; all p < .05). Relative to specific MRTP claims, general MRTP claims led to lower perceived risk and exposure for complete switching (d = –0.13 and d = –0.16) and partial switching (d = –0.14 and d = −0.12; all p < .05). Risk and exposure MRTP claims had similar effects (all p > .05).

Discussion

MRTP claims led to lower perceived risk and exposure, and higher willingness to try IQOS. General claims elicited larger effects than specific claims. MRTP claims also promoted unintended halo effects (eg, lower perceived risk of disease and chemical exposure for partial switching).

Implications

We found evidence that MRTP claims promoted health halo effects. In light of these findings, the FDA should require research on halo effects prior to authorization. Further, if an MRTP claim is authorized, FDA should require tobacco manufacturers to conduct post-market surveillance of how the claim affects consumer understanding, including partial switching perceived risk and exposure beliefs, as well as monitoring of dual-use behaviors.

Introduction

The 2009 Family Smoking Prevention and Tobacco Control Act (TCA) empowered the US Food and Drug Administration (FDA) to regulate tobacco products.¹ FDA’s regulatory authority includes oversight of advertising, marketing, and promotion of tobacco products, and public education about the harms of tobacco products.² Through these and other actions, the TCA aims to prevent youth tobacco initiation, encourage tobacco cessation, and decrease the harm and addictiveness of tobacco products.²

FDA has acknowledged that individual harm associated with tobacco product use exists along a continuum.³ While combusted tobacco products are believed to be most detrimental to human health, non-combusted products (although still dangerous) are believed to cause comparatively less harm. Consequently, some have proposed tobacco harm-reduction as a strategy to help smokers unable or unwilling to give up nicotine by encouraging the transition to a less harmful nicotine source.⁴

Tobacco companies have a well-documented history of using false and misleading claims in marketing campaigns, such as implying there are fewer health risks from certain tobacco products (eg, “low-tar” or filtered cigarettes).^5,6 The TCA prohibits tobacco manufacturers from making any unauthorized implicit or explicit health claims about their products¹; however, to help educate the public about tobacco product risks and encourage transitions away from combustible products, the TCA created a scientifically grounded process for tobacco manufacturers to seek FDA authorization to make certain health claims. The TCA refers to potentially less harmful tobacco products as modified risk tobacco products (MRTP).¹

The TCA describes two types of MRTP claims that can be authorized: (1) Claims describing reduced disease risk (risk modification) and (2) Claims describing reduced exposure to harmful substances (exposure modification).¹ In addition to type (ie, risk vs. exposure), MRTP claims can vary in their specificity.⁷ For instance, claims may be general and broadly describe reduced disease risk or exposure (eg, “tobacco-related diseases,” “harmful chemicals”), or claims can be specific and identify precise disease risks and exposures that are reduced (eg, “heart disease,” “carbon monoxide”). Claim type and specificity are important messaging decisions that manufacturers will make in designing and proposing MRTP claims for FDA review. These message features may play an important role in attracting and persuading consumers, and may affect consumer risk and exposure beliefs, as well as MRTP use behaviors.

Although MRTP claims have the potential to encourage smokers to transition to a less harmful tobacco product, such claims may also mislead consumers, which could result in unintended negative consequences. Of particular concern to the FDA is whether MRTP claims will elicit “halo effects,” leading consumers to infer health benefits beyond those in a stated claim.⁸ For instance, a tobacco user may generalize claims describing reduced exposure to a harmful chemical (eg, carbon monoxide) to mean that the product also conveys reduced disease risk (eg, lung cancer).⁹ According to both the TCA and industry guidance released by FDA, exposure claims should not be interpreted as reduced risk claims.^1,10 Further, MRTP claims that describe benefits (eg, reduced disease risk) from completely switching from cigarettes to the MRTP may be generalized to mean that partially switching to the MRTP (ie, using the MRTP and continuing to smoke) provides similar health benefits.⁹ In some situations, these generalizations may be inaccurate. For instance, partial switching may do little to reduce disease risk and could result in increased exposure to harmful chemicals compared to cigarette use alone.¹¹ Thus, if MRTP claims create halo effects, consumer decision-making may be based on inaccurate risk beliefs, which could result in negative unintended consequences such as dual tobacco product use.⁹

Understanding how MRTP claims affect consumer responses may aid in the development of claims that are maximally effective at educating the public about tobacco product relative risk and exposure while minimizing unintended consequences associated with misleading consumers. Moreover, such research is needed to guide FDA regulatory decisions for MRTP applications.

In the current experiment, we evaluated the impact of potential MRTP claims about IQOS, a heated tobacco product sold by Philip Morris International.¹² Based on previous empirical results and theory,^13–17 we hypothesized that MRTP claims (1) reduce perceived risk of disease, (2) reduce perceived exposure to harmful chemicals, and (3) increase willingness to try IQOS, compared to control. We also hypothesized that specific MRTP claims (1) reduce perceived risk of disease, (2) reduce perceived exposure to harmful chemicals, and (3) increase willingness to try IQOS, compared to general MRTP claims. Finally, given previous research findings,¹⁴ we hypothesized that MRTP claims describing reduced risk and reduced exposure have similar effects on (1) perceived risk of disease, (2) perceived exposure to harmful chemicals, and (3) willingness to try IQOS.

Methods

Sample

Eligible participants were ≥18 years of age, US residents, and current combustible cigarette smokers (≥100 cigarettes in lifetime and currently smoking every day or some days). All were enrolled in August 2019. We initially recruited Amazon Mechanical Turk (MTurk) workers (n = 1116), limiting recruitment to those meeting the smoker premium qualification (self-reported current cigarette smokers) and those with an MTurk rating of ≥90%.¹⁸ Because of slow enrollment (likely because of the premium qualification), we recruited additional participants with Prime Panels (n = 2045) using the same eligibility criteria.¹⁹

Procedures

Experimental Design

The between-subjects experiment used a 2 × 2 factorial design with an independent control (five total conditions). The first experimental factor was the MRTP claim type, which manipulated whether the claim described reduced risk (eg, “lung cancer”) or reduced exposure (eg, “carbon monoxide) benefits of completely switching from cigarettes to IQOS. The second experimental factor was claim specificity, which manipulated whether the claim described reductions in specific (eg, “carbon monoxide) or general (eg, “smoking-caused diseases”) exposures and risks. Control messages discouraged the littering of IQOS waste. We chose this type of control message because it was relevant to IQOS but did not communicate any modified risk or exposure information.

The experiment was conducted online. Following consent and screening, eligible participants first viewed two web pages that contained information about and photographs of IQOS (Figure S1). All-participants could advance to the experiment after viewing each page for a minimum of 10 seconds. These web pages explained what IQOS is, how IQOS is used, and how IQOS differs from e-cigarettes.

Next, participants were randomized to view one message, either an MRTP claim or a control message. Above the message was the following instruction: “Please read the below message carefully and imagine you saw the message in an advertisement for IQOS.” Participants then responded to a variety of outcome measures, followed by demographic and tobacco use questions. To allow for repeated exposure to the randomly assigned claim or control message, the claim or message appeared at the top of each page for all-outcome measures. Participants were debriefed at the end of the study and informed that the information they saw was created by our research team, may not be true, and was used for research purposes only. Participants also received web site links to evidence-based quit-smoking resources.

MTurk participants received $2.15 for completing the online questionnaire (median time: 8 minutes, 13 seconds). Prime Panels participants (median time: 9 minutes, 16 seconds) were compensated in the amount that they agreed to with the platform through which they entered this survey. This study was approved by the institutional review board at the University of North Carolina.

Stimuli

The experiment evaluated MRTP claims for IQOS. Because IQOS was not sold in the US at the time of data collection, we anticipated participants would be IQOS naïve and less likely to have existing attitudes or beliefs about IQOS; therefore, we anticipated that most responses would be fully shaped by the experimental stimuli. In 2017, only 5% of US adults were aware of heated tobacco products and fewer than 1% reported ever use.²⁰

To increase the generalizability of study findings, we created three messages per condition. Thus, we created 12 MRTP claims and three control messages. All MRTP claims had the same sentence structure to control for potential confounding by other claim features (eg, framing, length). The sentence structure used was modeled off of MRTP claims included in the IQOS application submitted to FDA.²¹ The following sentence structure was used for all-experimental MRTP claims: “Switching completely from cigarettes to IQOS will reduce [exposure to or risk of] [exposure or risk]”.

In selecting the diseases and exposures for the MRTP claims, we developed claims that would discourage smoking and increase interest in IQOS, while also considering word length and readability. For the three specific exposure MRTP claims, we selected arsenic, carbon monoxide, and lead because these chemicals were among those that scored highest (out of 24 tested) for awareness and discouragement from smoking in research by Brewer et al.²² Similarly, we selected lung cancer, emphysema, and cardiovascular disease because these outcomes were among those that scored highest on awareness and discouragement (out of 25 messages tested) in research by Kelley et al.²³

In choosing content for the general MRTP claims we created claims with similar word counts and readability as the specific MRTP claims, while including broad language to describe relative risk and exposure modifications. General risk MRTP claims included the phrases “smoking-related diseases,” “smoking-caused diseases,” and “diseases caused by smoking.” General exposure MRTP claims included the phrases “smoke chemicals,” “chemicals in smoke,” and “chemicals in cigarette smoke.” We did not include words such as carcinogens or toxins in the general exposure claims because cognitive interviews revealed these terms are often interpreted as communicating reduced risk. The wording for all-MRTP claims and control messages appears in Table 1. Mean word counts for each experimental condition and control were well matched, ranging from 11.3 to 13.7 words. Readability was also similar, with mean Flesch–Kincaid grade level below 12 for all-groups (range: 10.6–11.8). Here forward we refer to general risk and general exposure claims as “general claims,” and specific risk and specific exposure claims are referred to as “specific claims.”

Table 1.

Experimental claims and control messages

1. General exposure (mean: 13.0 words; 11.5 Flesch–Kincaid grade level)

Switching completely from cigarettes to IQOS will reduce exposure to smoke chemicals.

Switching completely from cigarettes to IQOS will reduce exposure to chemicals in smoke.

Switching completely from cigarettes to IQOS will reduce exposure to chemicals in cigarette smoke.

2. Specific exposure (mean: 11.3 words; 11.6 Flesch–Kincaid grade level)

Switching completely from cigarettes to IQOS will reduce exposure to arsenic.

Switching completely from cigarettes to IQOS will reduce exposure to lead.

Switching completely from cigarettes to IQOS will reduce exposure to carbon monoxide.

3. General risk (mean: 13.7 words; 11.8 Flesch–Kincaid grade level)

Switching completely from cigarettes to IQOS will reduce the risk of smoking-related diseases.

Switching completely from cigarettes to IQOS will reduce the risk of smoking-caused diseases.

Switching completely from cigarettes to IQOS will reduce the risk of diseases caused by smoking.

4. Specific risk (mean: 12.7 words; 10.6 Flesch–Kincaid grade level)

Switching completely from cigarettes to IQOS will reduce the risk of lung cancer.

Switching completely from cigarettes to IQOS will reduce the risk of cardiovascular disease.

Switching completely from cigarettes to IQOS will reduce the risk of emphysema.

5. Control (mean: 12.7 words; 11.8 Flesch–Kincaid grade level)

Protect the environment by discarding used IQOS tobacco sticks in a trash receptacle.

Protect the environment by discarding used IQOS tobacco sticks in a trashcan.

Protect the environment by discarding used IQOS tobacco sticks in a garbage can.

Measures

The primary outcome was a willingness to try IQOS, which was measured using four items adapted from measures used by Nodora et al.²⁴ and Duke et al.²⁵ (eg, “willing to try…,” “are you curious . . .”) The four-point response scale for the willingness items ranged from definitely yes (coded as one) to definitely not (four). To create a willingness score, we reverse-coded and averaged the items (α = .93). Secondary outcomes included perceived risk and exposure from complete switching (from cigarettes to IQOS) and perceived risk and exposure from partial switching (from cigarettes to IQOS). Single items were used to measure perceived risk and exposure (for partially and completely switching from cigarettes to IQOS) with a five-point response scale ranging from much less (risk or exposure) to much more (risk or exposure). These items were accompanied by a figure depicting the type of switching assessed in the item (complete [100%] switching or partial [50%] switching). For instance, “20 cigarettes per day” → “10 cigarettes and 10 IQOS sticks per day,” and “Partial Switch” appearing above the arrow was used in the item measuring perceived risk and exposure from partial switching. A questionnaire coding error caused a typo in the instructions for perceived risk and exposure from partial switching items: “Imagine you smoke 20 cigarettes a day, and then partially switch from cigarettes to using 20 IQOS sticks a day.” The instructions should have stated, “. . . switch to using 10 cigarettes and 10 IQOS sticks a day.” However, the accompanying figure and item wording were correct (Figure S2).

Additional secondary outcomes included message believability, ease of understanding, relevance, and perceived cognitive elaboration, all of which had five-point response scales. The survey measured perceived cognitive elaboration using three items, which we averaged to create a scale (α = .91). Single-item measures were used for believability, ease of understanding, and relevance (Table S1).

The survey included three attention check items: (1) instructed participants to not respond to one item, (2) instructed participants to respond “no” to one item, regardless of their true answer, (3) asked participants about how much effort they gave in completing the questionnaire (responses of “somewhat” and “a lot” were deemed acceptable, while responses of “not at all” and “a little” were unacceptable). Missing responses for attention check items that required a response were deemed unacceptable. Three-quarters of participants (75.9%) answered all-three attention check items acceptably, and 20.6% gave two acceptable answers. Only 3.3% and 0.25% of participants gave one or no acceptable answers to the attention check items, respectively. To minimize demographic bias, we retained all-respondents in our analyses and reported results based on the full sample. Sensitivity analyses excluding participants who gave unacceptable answers to two or more attention check items yielded the same pattern of results.

Statistical Analysis

We computed mean standardized differences (ie, Cohen’s d) comparing responses for (1) intervention (all-MRTP claims) versus control, (2) claim specificity (general vs. specific), and (3) claim type (risk vs. exposure). For all outcomes, we used linear regression models to (1) compare intervention (all-MRTP claims) versus control, (2) examine the main effects of claim specificity and type, and (3) examine the interaction of specificity and type, for all primary and secondary outcomes. For any statistically significant interaction (p < .05), we created interaction plots and calculated standardized mean differences between specific risk and specific exposure claims and between general risk and exposure claims. Given the experimental design, regression models did not control for any covariates. All-tests were two-tailed with a critical alpha of 0.05.

Results

A total of 3584 participants attempted to complete the questionnaire (Figure 1). Analyses excluded participants who were non-smokers (n = 323), did not report smoking status (n = 12), responded from a duplicate IP address (n = 69), used an invalid MTurk ID (n = 8), or had a duplicate MTurk ID (n = 11). For participants with duplicate MTurk ID and IP address responses, we included data from the participants with the earliest questionnaire start time. Thus, analyses included data from a total of 3161 eligible participants.

Figure 1.

CONSORT flow diagram.

Overall, 63.9% of participants were female, the participant mean age was 41.7 (SD = 12.9) years, and 82.9% and 9.2% of the sample were white and Hispanic, respectively (Table 2). Awareness of IQOS was low among all-participants (8.8%). Missingness across all primary and secondary outcomes was low, ranging from 0.13% to 0.70% per outcome. Analyses comparing the five conditions did not identify differences with respect to recruitment source, demographic or tobacco use characteristics. Mean willingness to try IQOS (range: 1–4) across the five conditions were as follows: general risk: 3.36 (SD = 0.74); general exposure: 3.40 (SD = 0.70); specific risk: 3.40 (SD = 0.70); specific exposure: 3.36 (SD = 0.66); control: 3.32 (SD = 0.73). Means for the other outcomes appear in Table S2.

Table 2.

Demographic characteristics and tobacco use behaviors among 3161 smokers randomized to condition (%, unless indicated)

	General claim		Specific claim
	Risk	Exposure	Risk	Exposure	Control	p
Total, n	630	623	643	642	623	–
MTurk, n	229	223	220	217	227	0.79
Prime Panels, n	401	400	423	425	396
Sex
Male	35.4	38.5	33.4	35.8	37.6	0.36
Age, mean (SD)	42.2 (13.1)	41.5 (12.7)	41.5 (13.2)	42.1 (12.7)	41.3 (12.7)	0.69
Hispanic	7.7	9.1	10.9	9.5	8.7	0.41
Race
Am Indian/Alaska Native	1.0	1.8	1.3	0.8	1.0	0.65
Asian	1.8	3.2	2.1	2.8	1.9
Black/African American	9.2	8.3	9.8	9.5	8.6
Native Hawaiian/Pacific Islander	0.2	0.0	0.3	0.6	0.2
White	83.2	83.0	82.3	81.1	84.7
Other/multiracial	4.7	3.7	4.3	5.2	3.7
Sexual orientation
Heterosexual	86.2	88.7	85.5	87.3	89.0	0.44
Gay or Lesbian	4.2	3.2	3.5	4.4	3.2
Bisexual	9.7	8.1	11.0	8.3	7.7
Education
High school graduate or less	27.4	24.6	27.4	26.9	27.1	0.47
Some college/Associate degree	44.9	45.8	42.1	41.6	40.3
Bachelor’s degree	18.0	19.1	20.6	21.7	19.5
Graduate or professional degree	9.7	10.5	9.9	9.8	13.1
Smoking frequency
Everyday	83.3	81.5	82.7	83.3	84.8	0.66
Some days	16.7	18.5	17.3	16.7	15.3
Current e-cig use
Yes	37.8	42.3	41.7	40.2	43.9	0.24
Want to quit smoking w/in 6 months
Yes	66.8	65.9	69.7	66.4	66.3	0.62
IQOS awareness
Yes	7.6	8.9	6.5	8.9	8.7	0.41
First cigarette after waking
Within 5 minutes	30.7	30.1	32.4	31.3	33.2	0.19
6–30 minutes	40.1	43.2	38.4	43.4	36.8
31–60 minutes	14.9	11.3	15.7	12.0	13.6
After 60 minutes	14.3	15.4	13.5	13.3	16.5

MRTP Claims Versus Control

Participants viewing MRTP claims reported a slightly higher willingness to try IQOS (d = 0.09), relative to control (litter message). Additionally, participants viewing any MRTP claim had lower scores for perceived risk (d = –0.32) and exposure from complete switching from cigarettes to IQOS (d = –0.31; Table 3). We also found evidence of the “exclusive-to-dual use halo effect,”⁹ with MRTP claims resulting in lower scores for perceived risk (d = –0.25) and exposure from partial switching (d = –0.26;), compared to control. Thus, claims describing the benefits of complete switching (exclusive IQOS use) were interpreted as also reducing risk and exposure when IQOS was used to partially replace cigarette use (dual use).

Table 3.

Impact of the presence, specificity, and type of MRTP claims.

			Experimental conditions
	All-MRTP claims vs. control		Specificity		Type		Type × specificity
Outcomes	d	p	d	p	d	p	p
Complete switching perceived risk	–0.32	<.001	–0.13	.001	0.05	.235	.961
Complete switching perceived exposure	–0.31	<.001	–0.16	<.001	–0.07	.073	.972
Partial switching perceived risk	–0.25	<.001	–0.14	<.001	0.05	.186	.542
Partial switching perceived exposure	–0.26	<.001	–0.12	.003	–0.07	.095	.368
Willingness to try	0.09	.043	0.00	.972	0.00	.891	.174
Believability	–0.37	<.001	–0.00	.929	0.15	<.001	.305
Relevance	0.32	<.001	0.11	.008	–0.03	.479	.019
Ease of understanding	–0.13	.003	0.05	.193	–0.03	.412	.995
Elaboration	0.66	<.001	0.03	.400	–0.01	.919	.127

Positive effect size d indicates higher scores for intervention compared to control, higher scores for specific claims compared to general claims, and higher scores for risk claims compared to exposure claims.

Participants found MRTP claims to be more personally relevant (d = 0.32) and MRTP claims led to greater cognitive elaboration (d = 0.66), compared to control messages. In contrast, MRTP claims were less believable (d = –0.37) and less easy to understand (d = –0.13), compared to control messages.

Claim Specificity

When examining the effect of claim specificity on responses (ie, examining responses from general claims group vs. specific claims group), specificity had no effect on willingness to try IQOS (d = 0.00). In contrast, general claims led to lower perceived risk (d = –0.13) and exposure (d = –0.16) from complete switching, as compared to specific claims. Similarly, general claims resulted in lower perceived risk (d = –0.14) and exposure (d = –0.12) from partial switching, compared to specific claims. General MRTP claims were perceived as more personally relevant, compared to specific MRTP claims (d = 0.11). Specificity did not influence claim believability (d = –0.00), ease of understanding (d = 0.05), or elaboration (d = 0.03).

Claim Type

Claim type was not associated with willingness to try IQOS (d = 0.00), perceived risk (d = 0.05) or exposure (d = -0.07) from complete switching, or perceived risk (d = 0.05) or exposure (d = -0.07) from partial switching. However, exposure MRTP claims were rated as more believable than risk MRTP claims (d = 0.15). Claim type did not influence relevance (d = –0.03), ease of understanding (d = –0.03), or perceived cognitive elaboration (d = –0.01).

Claim Specificity × Type Interaction

The interaction of specificity and type was present for message relevance (p = .019). Post hoc analysis showed that the difference between specific risk and specific exposure claims (d = 0.12; p = .028) was greater than the difference between general risk and general exposure claims (d = 0.07; p = .223). None of the eight other interactions tested were statistically significant. Given the number of tests conducted for the interaction testing and the small effects observed, these findings should be interpreted with caution.

Discussion

MRTP claims are intended to communicate relative risk and exposure information to consumers, but must not be misleading and encourage unsafe use. We observed that MRTP claims for IQOS led to lower perceived risk and exposure from complete switching from cigarettes to IQOS, and higher willingness to try IQOS, compared to control messages about littering. Viewing MRTP claims also reduced perceived risk and exposure from partial switching from cigarettes to IQOS, a potentially problematic halo effect. We found that general MRTP claims reduced perceived risk and exposure more than specific claims, though specificity was not associated with willingness to try IQOS. There was no association between claim type with perceived risk and exposure, or willingness to try IQOS, suggesting that risk and exposure claims are interpreted similarly.

Our experiment found that MRTP claims reduced perceived risk and exposure from complete switching, suggesting MRTP claims may help correct tobacco product relative-risk misperceptions that have been previously identified. For instance, Fong et al. analyzed data from the Population Assessment of Tobacco and Health Study and found that a majority of adult Americans perceived e-cigarettes and smokeless tobacco products as being more harmful or about the same harmfulness as cigarettes.²⁶ Similarly, Kiviniemi and Kozlowski analyzed data from the Health Information National Trends Survey and found that <10% of adult Americans correctly believed smokeless tobacco products to be less harmful than cigarettes.²⁷ Using the same data source, researchers found an increase between 2019 and 2020 in the percentage of US adults believing that e-cigarettes are more harmful than cigarettes.²⁸

Previous research has reported mixed results on how MRTP claims affect perceived risk. For instance, studies by Callery et al. (smokeless tobacco)²⁹ and Katz et al. (e-cigarettes)³⁰ found that MRTP claims reduced perceived risk, whereas research by Wackowski et al. (e-cigarettes)³¹ and Mumford et al. (e-cigarettes)³² observed no association between viewing a claim and perceived risk. Because perceived risk has been found to be an important theoretical antecedent to behavior change,^13,17 it is possible that by reducing perceived risk (and likely perceived exposure as well), MRTP claims could increase MRTP initiation rates. Further research is needed to understand these mixed findings and elucidate how MRTP claims affect product use.

The finding that MRTP claims reduced perceived risk and exposure from partial switching underscores how MRTP claims could potentially mislead consumers. Moreover, the magnitude of these effects was only slightly smaller than the effect sizes for perceived risk and exposure from complete switching. Thus, the effect size for the intended effect of MRTP claims (reducing perceived risk or exposure from complete switching) was similar to the effect size of unintended effects (reducing perceived risk or exposure from partial switching).

Additional research is needed to explore whether perceived risk and exposure from partial switching are important predictors of dual use behaviors, and if these beliefs are a barrier to complete switching or quitting tobacco use entirely. Further research is also needed to develop and test MRTP claims designed to minimize halo effects and misperceptions about partial switching. For instance, MRTP claims could include language to explicitly communicate that partial switching does not reduce risk and exposure. In light of these findings, if the FDA authorizes an MRTP claim, it should require tobacco manufacturers to conduct post-market surveillance of how the claim affects partial switching risk and exposure beliefs, as well as dual-use behaviors. Research may also be needed to develop and evaluate new measures of perceptions related to complete versus partial switching.

To our knowledge, this is the first study to examine how MRTP claim specificity affects consumer responses. While claim specificity was not associated with willingness to try IQOS, we found general claims produced lower perceived risk and exposure, compared to specific claims. Such findings are inconsistent with previous research from other domains and theories. For instance, the advertising literature suggests that specific messages are more accepted and produce fewer counterarguments compared to general messages.³³ Additionally, specific advertising messages are more credible and attract greater attention, compared to more abstract messages.^34,35 Previous research has also found that specific messages are perceived as being more persuasive, compared to general messages.^36,37 However, our current study found no significant difference in believability between general and specific claims. Further research is needed to identify putative factors driving these unexplained null effects.

We found that consumer responses to risk versus exposure claims were similar (eg, no main effect of claim type on perceived risk and exposure). These findings are consistent with a previous experiment by El-Toukhy and colleagues,¹⁴ who found that risk and exposure claims reduced both perceived risk and exposure. Additionally, qualitative research by Wackowski et al. found that MRTP claims describing reduced exposure were interpreted as reducing disease risk.³⁸ Fearing consumer misunderstanding, the TCA explicitly states that exposure modifications must not be interpreted as risk modifications.¹ Given this mandated provision, some researchers have argued that the exposure modification pathway is not legally viable for MRTP claims.¹⁴ The findings reported here provide additional evidence that exposure to MRTP claims reduces perceived risk. However, after the FDA authorized an exposure MRTP claim for IQOS (eg, “Scientific studies have shown that switching completely from conventional cigarettes to the IQOS system significantly reduces your body’s exposure to harmful or potentially harmful chemicals”) the agency acknowledged that an exposure claim can be authorized even if the claim reduces perceived risk. In a Technical Project Lead document, the FDA expounded, “. . . the totality of the evidence supports that risk reduction is reasonably likely to be demonstrated in subsequent studies” and therefore, “. . . consumer understanding is in line with the relative health risks of the product that are reasonably likely.”³⁹ It is unclear how future research in this domain will inform tobacco regulatory science; this example highlights the complexities involved with integrating scientific knowledge with regulatory concerns.

Study strengths include the use of a large sample of smokers, testing of two important MRTP design features (while controlling for other features), testing multiple claims per condition, and the experimental design. Participants were also repeatedly exposed to claim and control messages, maximizing exposure to the stimuli.

The study’s major limitations include the absence of behavioral outcomes (eg, IQOS use, dual use with cigarettes) and the use of a convenience sample. However, at the time of data collection, IQOS was not available for sale in the US and assessment of IQOS use post-message exposure was not possible. Although large and diverse, the study sample was non-probability-based. Research by Jeong et al. found that using convenience samples for online tobacco product perception experiments consistently yields similar results as national probability samples.⁴⁰ Furthermore, MRTP claims were presented to participants as text only to maximize claim visibility. However, such claims in the real world typically appear in marketing materials with color images. This study also assessed responses to MRTP claims for a single product; therefore, findings may not generalize to other potential MRTPs. Additionally, as previously described, incorrect instructions were displayed for the perceived risk and exposure from partial switching items because of a coding error. It is unknown whether this typo affected responses; however, our sample size was large and the accompanying figure and item wording were correct. We therefore speculate that the typo likely added some modest noise to the results but did not qualitatively affect the overall size or direction of the effect. Finally, contemporaneous with data collection (August, 2019), an outbreak of severe lung injury associated with e-cigarette use in the US occurred and was widely reported in the media (e-cigarette, or Vaping Product, Use Associated Lung Injury; EVALI), which could have affect perceived risk of IQOS use.⁴¹ However, we believe that the use of informative text and pictures describing IQOS at the start of the experiment, which included text stating that IQOS was not an e-cigarette product, mitigated potential confusion on this point.

Conclusions

MRTPs represent an important harm-reduction strategy for the FDA’s regulatory framework. Research evaluating MRTP claims is critical to elucidating the potential public health impact of authorizing MRTP claims and informing FDA regulatory decisions. We experimentally tested how MRTP claims and claim features affect consumer responses. As hypothesized, risk claims reduced perceived risk from complete switching and exposure claims reduced perceived exposure from complete switching. Such findings suggest that MRTP claims may be effective at educating consumers about tobacco product relative risk. Additionally, we found that viewing an MRTP claim reduced perceived risk and exposure from partial switching. Such unintended misperceptions could promote dual-use behaviors and potentially weaken the harm-reduction potential of MRTPs. These findings suggest that consumers may be inferring health benefits beyond what is stated in an MRTP claim. Additional research is needed to understand how to reduce these generalizations in response to MRTP claims and how MRTP claims affect tobacco use behaviors.

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.

Funding

This research was supposed by grant P50CA180907 from the National Cancer Institute and Food and Drug Administration’s Center for Tobacco Products, and grant F31DA045424 from the National Institute on Drug Abuse. 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 Food and Drug Administration.

Declaration of Interests

None declared.

Author Contributions

Andrew Seidenberg (Conceptualization [Lead], Formal analysis [Lead], Funding acquisition [Equal], Methodology [Lead], Project administration [Lead], Writing—original draft [Lead], Writing—review & editing [Lead]), Marcella Boynton (Conceptualization [Supporting], Methodology [Supporting], Writing—review & editing [Equal]), Noel Brewer (Conceptualization [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), Allison Lazard (Conceptualization [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), Paschal Sheeran (Conceptualization [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), and Kurt Ribisl (Conceptualization [Supporting], Funding acquisition [Equal], Methodology [Supporting], Supervision [Lead], Writing—review & editing [Supporting])

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.