The Impact of Smokeless Tobacco Risk Information on Smokers’ Risk Perceptions and Use Intentions: A News Media Experiment

ABSTRACT

Little research exists on the impact of risk information comparing smokeless tobacco (SLT) use, particularly snus, to cigarette smoking. This study explored this topic using a communication channel where smokers may be exposed to such information—the news media. We randomly assigned 1008 current smokers to read one of three constructed news stories or to a control group (no article). The “favorable” story framed snus as a “safer” smoking alternative while the “cautious” story described snus risks. The “mixed” version described potential risks and harm-reduction benefits. Participants completed a post-article survey with snus risk and harm perception and use intention measures. Article condition was significantly associated with perceived harm of daily snus use relative to smoking (1 = a lot less harmful – 5 = a lot more harmful; p < .0001), and mean ratings of snus harm in the favorable (2.46) and mixed conditions (2.66) were significantly lower than those of the cautious (2.96) and control conditions (2.98). Mean interest in trying snus in the next 6 months was low, but significantly higher for those in the favorable (1.55) and mixed conditions (1.32) versus those in the cautious (1.17) and control conditions (1.16)(1 = not at all – 5 = extremely interested, p < .0001). There were no significant differences by group in terms of the story’s perceived interestingness, importance, or relevance. Exposure to reduced-risk news messages about SLT and snus relative to cigarettes may impact smokers’ SLT harm perceptions and use intentions. Tobacco control professionals and FDA officials should consider the potential impact of the news media when communicating about tobacco risks.

Introduction

The promotion of smokeless tobacco (SLT) for harm reduction among smokers has been a source of debate in the tobacco control community. SLT contains nicotine, is addictive and some forms have been associated with oral and pancreatic cancer (Boffetta, Hecht, Gray, Gupta, & Straif, 2008) and heart disease (Boffetta & Straif, 2009). However, SLT has lower risks than smoking when used exclusively, particularly low nitrosamine forms such as snus, a Swedish style of moist snuff (Levy et al., 2004; Zeller, 2013). Among males in Sweden, snus use is more prevalent than smoking and is believed to have contributed to the country’s decline of male smoking and lung cancer rates (Foulds, Ramstrom, Burke, & Fagerstrom, 2003).

However, concerns exist that communicating about SLT as having reduced-risks might introduce population-level harms by deterring smokers from quitting completely, encouraging dual use of both products, and facilitating new users among those who misperceive reduced-risk messages as meaning SLT is safe (Tomar, Fox, & Severson, 2009). Yet, others have argued that individuals have a right to relative harm information and that without it, smokers might not switch to lower harm products because they think all tobacco products are equally harmful (Biener & Bogen, 2009; Borland, Cooper, McNeill, O’Connor, & Cummings, 2011). Indeed, previous research has found that many perceive SLT to be as or more harmful than cigarettes and to generally have low interest in SLT (Borland et al., 2011; Timberlake, 2009).

In an effort to change this trend, one SLT company, Swedish Match, submitted the first modified-risk tobacco product (MRTP) application to the FDA in 2014, a pathway developed by the 2009 Tobacco Control Act that allows tobacco companies to apply to market their products as having “modified” or reduced risks. In its application, Swedish Match included a request to stop using the current warning that snus is “not a safe alternative to cigarettes” and to market its General Snus brand as presenting “substantially lower risks to health than cigarettes”. Although the application was denied in December 2016, the FDA responded that the application could be amended and resubmitted for consideration (Food & Administration, 2016). Since then another tobacco company has submitted a MRTP application for Camel Snus (Craver, 2017), the current leader in the US snus market (Miller Lo et al., 2017).

However, research on whether reduced-risk (RR) information about SLT and/or snus can change smokers’ SLT risk and harm perceptions and use intentions is limited and mixed. Three studies examining RR messages presented in SLT warnings found that participants experienced reduced SLT harm perceptions, but only two found associations with increased use intentions (Callery, Hammond, O’Connor, & Fong, 2011; Mays, Moran, Levy, & Niaura, 2016; Rodu, Plurphanswat, Hughes, & Fagerstrom, 2016). Two studies have examined the potential impact of a RR message presented as a benefit claim in SLT ads, with only one finding an effect on smokers’ RR perceptions and neither finding an effect on use intentions (Fix et al., 2017; Capella, Taylor, & Kees, 2012). Only two studies have examined the impact of RR information in more in-depth educational formats. One found that while exposure to a brochure with anti-smoking information increased smokers’ SLT demand, exposure to pro-SLT information did not (Rousu et al., 2014). Another found reduced harm perceptions of SLT and greater interest in SLT among smokers following exposure to an educational fact sheet about the relative harms of SLT and cigarettes. However, exposure effects were found to be modest (Borland et al., 2012).

In the absence of any modified-risk educational campaigns or marketing orders, another potential source of information about the relative risks and harms of SLT and cigarettes may be the news media, which has played a role in informing the public about tobacco risks since the 1950s (National Cancer Institute, 2008). The news is a unique communication channel that can shape the public’s perception about the importance of tobacco issues by its coverage volume and agenda setting effects (McCombs & Shaw, 1972). According to Agenda Setting Theory, media attention given to issues can be influential in terms of communicating that by receiving news coverage, they are newsworthy and salient. In addition, news coverage not only suggests what the public should think about, but can also influence attitudes by shaping how we think about issues through the way in which they are portrayed and framed (National Cancer Institute, 2008). Framing is a way of selecting, organizing and “packaging” information about an issue to influence the way it is perceived and thought about (Menashe & Siegel, 1998). Frames can construct how information is presented, define problems and suggest solutions. Important to framing is the issue of selection—frames may influence our perceptions by selecting and making certain aspects of an issue more salient, while omitting others (Entman, 1993).

Framing has been studied in news about tobacco policy debates such as smoking bans (National Cancer Institute, 2008) and in coverage of tobacco control more generally (Menashe & Siegel, 1998). Yet, little news research has focused on the characterization of non-cigarette products. An analysis of cigar news found that most stories presented one-sided positive portrayals of cigars, tending to frame cigars as less harmful than cigarettes and as more of a “trendy habit or lucrative business” than as a health risk (Wenger, Malone, & Bero, 2001). A UK study found that early e-cigarette news coverage was more likely to frame e-cigarettes as a “healthier choice” compared to cigarettes than to describe them as potentially harmful to health (Rooke & Amos, 2014).

In a content analysis of SLT news between 2006 and 2010, we previously found that 16.5% of news articles referred to SLT as possibly being less harmful than smoking, references that were more frequent in articles discussing new SLT products Camel and Marlboro Snus, which were introduced in the US during this time period (Wackowski, Lewis, Delnevo, & Ling, 2013). Moreover, these “less risky” messages were attributed as frequently to public health professionals or researchers as to tobacco company representatives, potentially adding to their salience. On the other hand, the study also found that 37% of news articles referred to some SLT risk (e.g., oral cancer), and that articles included messages indicating that SLT is not a safe smoking alternative (Wackowski, Lewis, Delnevo, & Ling, 2014). These messages may inform readers about SLT risks and/or shape or reinforce beliefs that SLT is just as harmful as smoking.

This study aimed to contribute to the literature on the potential impact of SLT risk information with a SLT news experiment. Although the news has been a traditional source of health information about tobacco, its impact on tobacco perceptions and behaviors has been understudied (National Cancer Institute, 2008). Also, we are not aware of any studies that have experimentally examined the effects of different tobacco risk frames on readers. The calls for providing more accurate SLT/cigarette relative risk communication presume that doing so might change SLT risk perceptions and use. However, studies on this topic have been limited and with mixed results. In this current study, we hypothesized that smokers exposed to news stories with messaging that snus is less harmful than smoking would rate snus as being less harmful (H1) and have higher intentions to use snus (H2) than smokers exposed to a news story without such harm-reduction messages and compared to a control group of smokers not exposed to any story.

Research about the effects of two-sided perspectives in news stories is also limited, despite their common use (Chang, 2013). In our SLT news analysis we also found that articles included mixed SLT information (i.e., suggestions that SLT is safer than smoking and warnings against its use), likely in an attempt to provide “balanced” news coverage. Indeed, providing two sides or perspectives on an issue is a traditional journalistic value to provide unbiased accurate information and is associated with news credibility (Nelson, Hesse, & Croyle, 2009). However, communication research has also suggested that two-sided news stories might create confusion, uncertainty or feelings of ambiguity. For example, Chang (2013) found that men exposed to two versus one-sided news stories about food and nutrition topics experienced greater ambivalence towards the health topics and in turn, lower intentions to adopt the health behaviors. With respect to the current topic, stories with mixed SLT information might present readers with a balanced view of SLT risks and benefits but potentially an unclear overall “take away” message (Tomar et al., 2009). As such, we also hypothesized that smokers exposed to a SLT news story with mixed information about its harmfulness would be rated by smokers as more difficult to understand than stories framing SLT as either harmful or less harmful than cigarettes (H3).

Finally, some research has also suggested that consumers may respond to SLT harm-reduction messages with skepticism because they contradict what they have previously heard - that “SLT is not a safe alternative to cigarettes”. One study found that respondents were less trusting of snus ads including modified-risk claims (Fix et al., 2017) and in work leading up to this study, we found that some smokers were skeptical of “less harmful” messages in SLT news stories they read, particularly one-sided “less harmful” stories that described no SLT risks, perceived by some as feeling like “advertisements” (Wackowski, Lewis, & Delnevo, 2016). Given this, our final hypothesis was that a one-sided story framing SLT as being less harmful than smoking without discussing any risks would be rated as the least credible news story (H4).

Methods

Experimental conditions

We conducted a between-subjects experiment, randomly assigning 1008 current smokers to read one of three news stories about SLT and snus (i.e., a story with “cautious,” “favorable,” or “mixed” SLT/snus risk information) or to a control group which did not read any news story. The stimulus articles were short news stories (approximately 460 words) constructed by our research team, as informed by our previous content analysis of SLT news (Wackowski et al., 2013). They included content adapted from previous news articles (Koch, 2011; Smith, 2008) about SLT and snus (driven by the introduction of Camel and Marlboro Snus in the US). They were designed to look like online news articles (including a large bolded headline, date and picture of Camel Snus), but were presented as stand-alone articles with no surrounding ads or other news content. All three articles began the same, i.e., by discussing the introduction of Camel and Marlboro Snus in the United States, what snus is, the growth and decline of the SLT and cigarette markets, respectively, and the industry’s hopes for snus to be seen as a modern SLT product. However, each article version was manipulated to include or exclude certain SLT/snus risk information presented in the second half of the article (see Table 1). The “cautious” article provided a one-sided view of the risks of SLT and snus and included the traditional message that SLT is “not a safe alternative to cigarettes.” In contrast, the “favorable” news story presented one-sided risk information that snus may be a less harmful alternative to smoking. The “mixed” news story presented both risk and harm-reduction information and arguments found in the other two conditions, which were presented as part of a debate amongst tobacco control professionals.

Table 1. Description of news story conditions.

All conditions	Introduced and described Camel and Marlboro Snus Described new snus products as modern, more attractive, acceptable forms of smokeless tobacco (SLT) Described sales of SLT and snus as growing
“Cautious” condition	Article title: “New smokeless tobacco products – same old story?” Risk information presented under subheading, “Not a safe alternative” Referred to SLT risks, including links to heart disease, oral and pancreatic cancer Referred to study finding Camel Snus has at least two carcinogens Included warnings from quoted researchers/health professionals that snus is addictive, “not a safe alternative to cigarettes”, and that there is much about it still unknown
“Favorable” condition	Article title: “New smokeless tobacco products gaining traction – some say they’re safer” Risk information presented under subheading, “A safe alternative to smoking?” Made no references to SLT health risks/consequences Included references from health professionals, research studies describing snus as seemingly safer than cigarettes. Brief description of Swedish construction worker study finding snus users had lower cancer risks than smokers and no increased oral cancer risk Researcher quote that smokers should be encouraged to quit tobacco entirely, but also be told that SLT products have fewer risks than cigarettes
“Mixed” condition	Article title: “New tobacco product debate – safer alternative or same old story?” Risk information under subheading, “A debate over the potential risks versus benefits” Presented risk and harm reduction messages found in “cautious” & “favorable” versions

Study sample and procedures

Participants were obtained from a commercially available national research panel (GFK’s KnowledgePanel) assembled through probability-based sampling of addresses from the US Postal Service’s Delivery Sequence File. Eligible participants had to be adult (at least 18 years old) current smokers (i.e., have smoked 100 cigarettes in their lifetime and now smoke “everyday” or “some days”) (Agaku, King, Husten et al., 2014) who were not current SLT users (i.e., no use in the past 30 days)(Center for Behavioral Health Statistics and Quality, 2015) and had used SLT fewer than 20 times in their lifetime (Agaku et al., 2014). We focused on smokers given the potential role of snus to serve as a harm reduction alternative to cigarettes and the presumption that risk information might influence smokers’ SLT risk and harm beliefs and use intentions. GFK sampled 2,654 participants and 1288 (48.5%) completed eligibility questions. Of these, 1008 qualified for and completed the study online. Participants were told the study was about tobacco use and perceptions and were instructed to begin by reading the news article on the next screen page, which would be followed by questions on their thoughts about their own tobacco use. The control group did not read any story (survey only). After the survey, participants were shown a debriefing page which explained the study design and, in the case of the favorable and cautious conditions, described the SLT risk information that was left out of the article that they read. It also included a message that quitting smoking is extremely important for health and provided cessation resource information. Data were collected in June 2016. This study was approved by the Rutgers Biomedical and Health Sciences Institutional Review Board.

Measures

Main outcome measures were adapted from previous SLT studies. Relative harm was measured on a 5-point response scale (1 = a lot less harmful – 5 = a lot more harmful) by asking respondents, “compared with daily cigarette smoking, how harmful to health do you think daily use of snus is?” (Lund, 2012). Absolute harm was measured by asking respondents how harmful to health they think regular use of snus is (1 = not at all harmful – 5 = extremely harmful)(Popova & Ling, 2013). Respondents were also asked how likely or unlikely they believed that regularly using snus would cause a snus user to develop heart disease, lung cancer, oral cancer or “other cancer” in their lifetime (from 1 = Not at all likely—5 = Extremely likely)(Lund & Scheffels, 2014). Use intentions were measured by asking respondents how interested (1 = not at all interested– 5 = extremely interested) they were in trying snus in the next 6 months and how likely or unlikely they were to buy snus in the next 6 months (1 = Not at all likely—5 = Extremely likely). We also asked respondents if they would ever use any form of SLT to quit smoking and if they would ever completely replace their cigarettes with any form of SLT to reduce their health risk (1 = definitely would not—4 = definitely would)(Popova, Neilands, & Ling, 2014).

Participants were also asked about their perceptions of the news story they read, including how interesting, relevant and important they perceived the story to be (e.g., “how interesting, if at all, is this story?,” 1 = not at all interesting – 5 = extremely interesting) (Coleman, Thorson, & Wilkins, 2011). Perceived difficulty of story was rated on a 4-point scale (1 = very difficult—4 = very easy), as was perceived learning (1 = Nothing at all—4 = A great deal) (Borland et al., 2012; Popova et al., 2014). Likelihood of seeking more information about the content in the article was also measured (1 = Not at all likely—5 = Extremely likely). Perceived credibility of the news story was measured using Meyer’s news credibility scale (5 point semantic differential items: e.g., unfair/fair, biased/unbiased) (α = 0.89) (Meyer, 1988). Standard tobacco surveillance measures (e.g., cigarettes per day, past year quit attempt, number of years smoked, quit intentions in the next 30 days and six months) and demographic measures (e.g., gender, race/ethnicity, age, education, income levels, geographic region) were also included.

Analysis

Mean responses were calculated by article condition based on the underlying numeric scale. Overall bivariate comparisons were assessed using One-way Analysis of Variance tests for normally distributed variables, or Kruskal-Wallis tests for non-normally distributed data. Eta-squared statistic (SS_effect/SS_total) were calculated to estimate effect size for both parametric and nonparametric (i.e., using ranks) analyses. When overall associations were statistically significant (p < 0.05), post hoc pair-wise comparisons were assessed by Tukey’s HSD or Dwass-Steel-Critchlow-Fligner (DSCF) multiple comparison methods, respectively. For categorical analyses, differences by condition were assessed with Cochran-Mantel-Haenzel (CMH) tests.

Results

The study sample (n = 1008) was approximately equally distributed by gender (50.5% male), most (70.5%) were white (14.2% black, 10.3% Hispanic, 5.0% other), 47.3% had at least some college education, and 54.8% were employed. The average age was 48 (range 19–65)(11% ages 18–29, 21.9% ages 30–44, 46.8% ages 45–59 and 20.2% ages 60 and up). Most (82.6%) participants were daily smokers, and 18.4% had ever tried or used SLT before, 54.8% had ever heard of snus before and 8.7% had ever tried snus before. There were no significant associations between study conditions and demographics or smoker characteristics (cigarettes per day, past year quit attempt, number of years smoked, quit intentions in the next 30 days and six months, and ever use of SLT, snus, and e-cigarettes), indicating that randomization was successful.

Product risk and harm perceptions

Article condition was significantly associated with harm perception of snus relative to smoking (Kruskal-Wallis X² = 65.8, df = 3, p < .0001). The perceived harm of daily use of snus versus daily cigarette smoking was lowest among smokers in the favorable condition (x̄ = 2.46), followed by those in the mixed (x̄ = 2.66), cautious (x̄ = 2.96) and control (x̄ = 2.98) conditions, consistent with Hypothesis 1 (see Table 2). Post-hoc analysis indicated that ratings in the favorable and mixed conditions were not significantly different from each other but were significantly lower than those of the cautious and control conditions. However, approximately half of smokers in the favorable condition and 57.3% of those in the mixed condition still rated snus as being as or more harmful than smoking (see Figure 1). Ratings between those in the cautious and control conditions were not significantly different from each other.

Table 2. Mean scores of snus risk and harm perceptions and use intentions, by experimental condition (n = 1008).

	Experimental condition
	Favorable	Mixed	Cautious	Control	Omnibus*
	mean (SD)	mean (SD)	mean (SD)	mean (SD)	η^2	value
Risk and Harm Perceptions
Harmfulness of daily snus use relative to smoking ^1	2.46 (0.91) c,d	2.66 (0.84) c,d	2.96 (0.86) a,b	2.98 (0.90) a,b	0.07	<.0001
Perceived harmfulness of regular snus use ^2	3.21 (0.94) b,c,d	3.49 (0.79) a	3.63 (0.92) a	3.62 (0.99) a	0.04	<.0001
Harmfulness of daily snus use relative to traditional SLT ^1	2.70 (0.84) c,d	2.83 (0.77) c,d	2.99 (0.67) a,b	3.06 (0.77) a,b	0.04	<.0001
Likelihood of lung cancer from regular snus use ^3	2.72 (1.00)	2.70 (1.07)	2.82 (1.09)	2.94 (1.21)	0.01	0.1508
Likelihood of heart disease from regular snus use ^3	3.03 (0.91) b,c,d	3.23 (0.93) a	3.39 (0.91) a	3.34 (1.07) a	0.02	<.0001
Likelihood of oral cancer from regular snus use ^3	3.55 (0.94) c,d	3.71 (0.90) d	3.89 (0.89) a	3.93 (0.94) a,b	0.03	<.0001
Likelihood of other cancer from regular snus use ^3	3.15 (0.90) c,d	3.43 (0.88) d	3.54 (0.67) a	3.52 (1.09) a,b	0.03	<.0001
Use Intentions
Interest in trying snus in next six months ^4	1.55 (0.94) c,d	1.32 (0.66) c,d	1.17 (0.47) a,b	1.16 (0.47) a,b	0.05	<.0001
Likelihood of buying snus in next six months ^3	1.42 (0.76) b,c,d	1.26 (0.60) a,d	1.15 (0.48) a	1.13 (0.49) a,b	0.04	<.0001
Likelihood of using SLT to quit smoking ^5	1.61 (0.80) c,d	1.50 (0.69) d	1.39 (0.60) a	1.32 (0.61) a,b	0.02	<.0001
Likelihood of completely replacing cigs with SLT ^5	1.59 (0.74) c,d	1.48 (0.67) c,d	1.32 (0.53) a,b	1.33 (0.61) a,b	0.03	<.0001

* η² = SS_between/SS_total (ranks); p-value is from Kruskal-Wallis test

1 = A lot less harmful, 2 = Somewhat less harmful, 3 = About as harmful, 4 = Somewhat more harmful, 5 = A lot more harmful

2 = Not at all harmful, 2 = Not too harmful, 3 = Somewhat harmful, 4 = Very harmful, 5 = Extremely harmful

3 = Not at all likely, 2 = Not too likely, 3 = Somewhat likely, 4 = Very likely, 5 = Extremely likely

4 = Not at all interested, 2 = Not too interested, 3 = Somewhat interested, 4 = Very interested, 5 = Extremely interested

5 = Definitely would not, 2 = Probably would not, 3 = Probably would, 4 = Definitely would

Post hoc pairwise comparisons (Dwass-Steel-Critchlow-Fligner method):

^a Significantly different (p < .05) from Favorable; ^b Significantly different (p < .05) from Mixed; ^c Significantly different (p < .05) from Cautious; ^d Significantly different (p < .05) from Control

Figure 1. Prevalence of harm perceptions of daily snus use relative to daily cigarette smoking, by experimental condition.

A significant effect of experimental condition on absolute perceived harmfulness of regular snus use was also observed (Kruskal-Wallis X² = 39.65, df = 3, p < .0001). Significant mean differences were observed between the favorable condition and all other groups, but there were no significant mean differences between the mixed, cautious and control groups (see Table 2). Those in the cautious condition had the highest prevalence of rating regular use of snus to be very or extremely harmful (53.6%), followed by those in the control condition (46.4%), those in the mixed condition (40.7%) and favorable condition (27.2%) (data not in table).

No significant differences by experimental group were found for the perceived likelihood of getting lung cancer from regular snus use, but were found for heart disease, oral cancer and other cancer types (p < .0001, Table 2). Perceived risk of heart disease differed between the favorable group and all other conditions. Significant differences for perceived risk of oral cancer existed between control and favorable groups, control and mixed groups, and cautious and favorable groups (Table 2).

In addition, there was a significant relationship between experimental condition and participants’ self-reported perceived article impact (CMH X² = 79.9, df = 2, p < .0001). Although most smokers across conditions indicated that the article did not change their existing perceptions about the relative harm of SLT versus cigarettes (74%), 30.4% of those in the favorable condition and 17.6% of those in the mixed condition indicated that the article made them think that SLT products may be less harmful (relative to smoking) compared to what they previously thought versus only 4% of those in the cautious condition who thought so. In contrast, 17.4% of those in the cautious condition indicated that the article made them think that SLT products may be more harmful relative to cigarettes than they previously thought (not in table).

Intentions to use snus/slt

Mean interest in trying snus in the next 6 months (x̄ = 1.3), buying snus in the next 6 months (x̄ = 1.24), likelihood of using SLT to quit smoking (x̄ = 1.45) and likelihood of completely replacing cigarettes with SLT (x̄ = 1.43) was low but significant experimental group differences were detected (p < .0001), with higher ratings for those in the favorable and mixed conditions compared to those in the cautious and control conditions (Table 2), consistent with Hypothesis 2.

Article perceptions

The majority (90.2%) of participants in the news article conditions found their article to be very or somewhat easy to read. The mixed news story received the lowest difficulty rating and thus Hypothesis 3 was not supported. About 39% indicated that they had learned quite a bit or a great deal of new information from the article and no significant mean group differences were observed. About 29% indicated that they were very or extremely likely to try to learn more about information presented in the article. There were also no significant mean differences by group in terms of the story’s perceived interestingness, importance, relevance to their life, or likelihood of searching for more information (see Table 3). However, the favorable news story received a significantly lower credibility rating than the mixed and cautious stories, consistent with Hypothesis 4. The credibility rating of the mixed story did not differ from the cautious story.

Table 3. Mean scores of news article perceptions, by experimental condition (n = 1008).

	Experimental Conditions
	Favorable	Mixed	Cautious	Omnibus
	mean(SD)	mean(SD)	mean(SD)	η^2	p-value
Article Perceptions*
Difficulty of news story ^1	3.46 (0.66) b	3.29 (0.70) a	3.40 (0.66)	0.01	0.0111
Interest of news story ^2	2.68 (1.08)	2.74 (0.93)	2.66 (0.97)	0.00	0.6438
Importance of news story ^2	2.82 (1.06)	3.01 (0.94)	3.01 (0.97)	0.01	0.0783
Relevance of news story ^2	2.32 (1.16)	2.78 (1.07(	2.16 (0.99)	0.00	0.4252
Amount of new information learned ^3	2.31 (0.83)	2.37 (0.73)	2.23 (0.74)	0.01	0.0994
Likelihood of trying to learn more ^2	2.06 (1.08)	2.00 (0.97)	1.90 (0.92)	0.00	0.3955
Credibility**
Can’t be trusted … Can be trusted	3.00 (1.00) b,c	3.29 (1.09) a	3.26 (1.13) a	0.01	0.0039
Is inaccurate … Is accurate	2.94 (1.04)	3.13 (1.15)	3.17 (1.12)	0.01	0.0456
Is unfair … Is fair	2.98 (1.07)	3.16 (1.17)	3.19 (1.20)	0.01	0.0886
Doesn’t tell the whole story … Tells the whole story	2.73 (1.14) b,c	3.05 (1.14) a	3.02 (1.17) a	0.02	0.0025
Is biased … Is unbiased	2.80 (1.17) c	3.05 (1.12)	2.99 (1.19) a	0.01	0.0395
Index (mean)	2.95 (0.80) b,c	3.21 (0.86) a	3.20 (0.85) a	0.02	0.0007

*Kruskal-Wallis (omnibus) and Dwass-Steel-Critchlow-Fligner (post hoc pairwise multiple comparisons); η² = SS_between/SS_total (of ranks)

**ANOVA (omnibus) and Tukey’s HSD (post hoc pairwise multiple comparisons); η² = r²= SS_between/SS_total

1 = Very difficult, 2 = Somewhat difficult, 3 = Somewhat easy, 4 = Very easy

2 = Not at all [interesting/important/relevant/likely], 2 = Not too, 3 = Somewhat, 4 = Very, 5 = Extremely

3 = Nothing at all, 2 = Very little, 3 = Quite a bit, 4 = A great deal

Post hoc pairwise comparisons:

^a Significantly different (p < .05) from Favorable; ^b Significantly different (p < .05) from Mixed;

^c Significantly different (p < .05) from Cautious; ^d Significantly different (p < .05) from Control

Discussion

The news has been a traditional source of information about tobacco and previous studies have documented associations between tobacco news exposure and tobacco attitudes and beliefs (National Cancer Institute, 2008). Previous studies have also described ways in which tobacco issues and risks are framed in news stories, but experimental research on the effects of different messages and frames has been lacking. This study contributes to both the tobacco news and SLT communication literature with the first experimental study of snus risk messages in news articles on smokers’ snus risk perceptions and use intentions. We observed that news articles with messages about the harm-reduction potential of snus relative to cigarettes had an impact on lowered snus risk and harm perceptions and increased use intentions. This is important given that tobacco news articles with harm-reduction frames have increased over the last decade (Eversman, 2015) and may play a role in public thinking about this topic, and interest in use of these products.

This study also has important implications for tobacco regulatory science. Although scientists and the FDA cannot regulate the news media nor control its content, they should be aware of the potential impact of tobacco news coverage on public perceptions of tobacco products, given that such perceptions may influence product use as well as public and lawmakers’ opinions about tobacco control policies. Given the reach of the news media, the FDA and tobacco control professionals can also leverage it as a source to communicate with the public about tobacco products (e.g., through press releases, editorials, opinion letters), particularly for those products which the public may have less knowledge about. However, care should be taken in communicating messages about potentially reduced-risk tobacco products. Messages that focus only on their potential risks without contextualizing their risks relative to cigarettes may contribute to misperceptions about the risks of these products (Majeed et al., 2017).

Results from this study may also suggest some practical implications for messaging about modified-risk products such as snus that may not be limited to news articles. We found that the snus risk and harm perceptions and use intentions of smokers in the “cautious” condition did not significantly differ from those in the control group who read no article. This may suggest that the framing of SLT as being “not a safe alternative” to smoking used in the cautious article is consistent with the types of messages that control participants have been previously exposed to, and that their risk perceptions about these products may not change in the absence of different messaging about them. Previous research about SLT has also found that smokers misinterpret the “not safe alternative to cigarettes” message as meaning that the products are not safer and are just as harmful as cigarettes (Wackowski et al., 2016). If a goal of tobacco regulatory science is to accurately communicate with the public about tobacco products and their risks based on available science, then the FDA and tobacco control professionals should consider discontinuing use of the “not safe alternative” message for products such as snus that may in fact be less harmful than cigarettes on a continuum of risk (Zeller, 2013) and invest in research about relative tobacco risk communication.

On the other hand, we observed that two-sided news articles which referenced SLT risks and the “not safe alternative” message but which also included messages about snus and SLT’s risk-reduction potential did have an impact on lowered relative risk perceptions and increased use intentions. Furthermore these two-sided news stories were perceived as being as credible as the one-sided cautious news story, and were not rated as being more difficult to understand. This may suggest that communication materials about modified risk products can successfully communicate about both product risks and potential harm-reduction benefits and may lend support to previous calls for more “nuanced” risk communication about such products (Biener & Bogen, 2009). However, we did not directly measure how such stories may have contributed to feelings of ambiguity or uncertainty, a potential effect of two-sided “balanced” coverages of news issues (Chang, 2013). Future research should continue to explore the impact of such message combinations in the news as well as in other media formats (e.g., fact sheets, pamphlets, websites, warning labels, and advertisements).

It is also important to note that although we observed significant effects of the harm-reduction messages, these were modest as approximately half of smokers in the favorable condition and 57% of those in the mixed condition still rated snus as being as or more harmful than smoking, and snus use intentions remained very low, even among these groups. This trend is consistent with previous studies finding that smokers’ generally have low interest in SLT and snus (Biener & Bogen, 2009; Rodu et al., 2016;) and could be a limitation of a single exposure to such messaging, a message which may be inconsistent from what many smokers have always heard about SLT and seemingly intuitive assumptions (i.e., that SLT causes oral cancer and that all tobacco is harmful)(Borland et al., 2012; Mays et al., 2016). As such, these results may suggest that consistent repeated messages about snus’ harm reduction potential may be needed to substantially move the needle on snus risk and harm perceptions and use intentions.

But risk perceptions are not the only important factor in smokers’ SLT use interest, with many simply finding SLT to be gross, unacceptable and unsatisfying, regardless of their reduced-risks (Bahreinifar, Sheon, & Ling, 2013; Sami et al., 2012; Wackowski et al., 2016). This suggests that even if reduced-risk messages about SLT and snus were widely disseminated, their uptake may still be limited, particularly given the recent growth in popularity of e-cigarettes, which also appear to have reduced risk but much more appeal among smokers than SLT.

Although this study is limited by use of only one news story for each condition, the stimuli were adapted from real news articles to improve their external validity. However, the presentation of the articles in a stand-alone format, without any surrounding ads or other content as might be observed in an online news site, might detract from the finding’s external validity. Study strengths include use of a large national sample of smokers, randomized experimental design, and multiple risk/harm measures. However, other priority audiences for FDA tobacco communications including youth and non-tobacco users were not included. Future research should continue to study the impact of tobacco news message framing with various audiences and products such as e-cigarettes. Indeed, the news has been named as one public source of e-cigarette information (Wackowski, Bover Manderski, & Delnevo, 2015) and may be contributing to growing perceptions that e-cigarettes are as harmful as cigarettes (Majeed et al., 2017).

In conclusion, we found that exposure to reduced-risk messages about SLT and snus relative to cigarettes, such as those found in news stories, may impact smokers’ SLT risk and harm perceptions and use intentions. Although SLT reduced-risk messages may be contrary to what smokers have traditionally heard, two-sided informational materials that present both the risks and harm-reduction potential of SLT may be considered just as credible as materials that only warn about SLT risks. Future research should continue to explore the potential impact of reduced-risk SLT information in various communication formats, and for other non-cigarette products such as e-cigarettes. Tobacco control professionals and FDA officials should consider the potential impact of the news media when communicating about tobacco product risks.

Acknowledgments

Research reported in this publication was supported by the National Cancer Institute and FDA. Center for Tobacco Products (CTP) (R03CA175901). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the FDA. Thank you to Drs. Richard O’Connor, David Timberlake, Lucy Popova and Ganna Kostygina for early feedback on drafts of our stimulus materials.