Identifying promising themes and messages for youth vaping prevention: A national study

Abstract

Use of e-cigarettes and vapes among adolescents continues to be a major public health concern. Health communication efforts can discourage e-cigarette use among adolescents by influencing beliefs and behavior. However, to do so, studies need to identify the most promising themes and messages based on the latest evidence about the health harms of e-cigarettes and vaping. Participants were a nationally representative sample of 1,603 US adolescents aged 13–17 years, recruited in the summer of 2022. Adolescents were randomly assigned to view 7 vaping prevention statements (one from each theme: nicotine addiction, chemical harms, health symptoms, mental health, organ effects, cosmetic effects, and monetary cost) and 1 control statement (vape litter theme) from a pool of 46 statements that were developed through a systematic process. Participants rated each statement on perceived message effectiveness (PME), awareness, and believability. Results of linear mixed models indicated that all vaping prevention themes out-performed control messages on PME, with chemical harms and organ effects having the largest effects, followed by nicotine addiction and then other themes. For most message themes, PME effects were stronger for youth susceptible to vaping compared to non-susceptible youth and users. Both awareness and believability predicted higher levels of PME. In secondary analyses, we found that statements specifying the target (“you”) and longer statements were also rated higher on PME. Results suggests that the most potent vaping prevention messages for adolescents are those that focus on vape chemicals and the potential of vaping to damage organs and increase disease risk.

Introduction

Use of e-cigarettes or vapes among adolescents continues to be a major public health concern. Results from the 2022 National Youth Tobacco Survey (NYTS) show that more than 2.5 million adolescents – or about 1 in 10 adolescents – report using e-cigarettes or vapes in the past 30 days¹. Despite efforts by the U.S. Food and Drug Administration (FDA) to curb the availability of vaping products, scores of flavored and disposable e-cigarettes that are popular among adolescents remain on the US market^2,3.

E-cigarettes contain nicotine and are highly addictive, with studies showing adolescent and young adult use of e-cigarettes is linked to progression to combustible tobacco product use^4–6. In addition, e-cigarette use has been associated with a growing number of health harms. The aerosol emitted from e-cigarettes may expose users to toxic chemicals and metals—albeit at lower levels than those found in cigarettes—^7–9, including carcinogens such as formaldehyde and acrolein^7,10. Use of e-cigarettes has been linked to impaired cognitive function¹⁰, diminished respiratory function^11–13, potential heart and lung damage^14,15, and poor mental health^16–19. Increased coughing and wheezing and asthma exacerbations have been observed specifically among adolescents¹³.

Health communication efforts such as communication campaigns and preventive interventions can discourage vaping among adolescents by influencing beliefs about the consequences of behavior^20–22. Prominent theories of health behavior – including the reasoned action approach²³, social cognitive theory²⁴, and the health belief model²⁵ – posit that such behavioral beliefs (also called outcome expectations) are key determinants of behavior. Moreover, a large empirical literature supports the notion that beliefs about the consequences of behavior are key drivers of behavior,^26,27 including in the context of youth vaping^28–30. In order to design effective communication campaigns to change such beliefs, however, studies are needed to identify the most promising message themes based on the latest evidence about the negative consequences of using e-cigarettes and vapes^31,32. In the current work, we focus not only on the more traditional message themes of nicotine addiction, chemical exposures, vaping symptoms, and organ effects, but also more novel themes such as the impact of vaping on mental health, cosmetic effects of vaping, and the monetary cost of vapes.

Message Themes

Several potential message themes have been explored in prior vaping prevention studies with adolescents. In a comprehensive examination of potential vaping prevention themes, Sangalang et al. (2016) generated a set of campaign beliefs focused on preventing youth e-cigarette initiation, resulting in 23 potential themes. Their empirical study suggested that short-term health effects (e.g., headaches, sinus issues) may be among the most promising themes for vaping prevention messages for adolescents³². In a large study testing 220 vaping prevention print ads used by practitioners, results indicated that message themes about addiction, chemicals, negative health effects, and health-related symptoms were associated with higher perceived message effectiveness (PME) ratings and were among the most promising message themes for adolescents³³.

Other studies of adolescents have examined vaping prevention message themes using experiments. A national experiment with adolescents found that message themes about chemicals in e-cigarette vapor and lung harms of vaping elicited higher PME than nicotine addiction or a control theme (vaping litter)³⁴. An experiment with 2 ads from the FDA’s The Real Cost national vaping prevention campaign – 1 nicotine addiction and 1 health harms ad – found that the two together demonstrated promising results at increasing risk beliefs about vaping and reducing intentions to vape³⁵. Finally, a randomized clinical trial of The Real Cost vaping prevention ads demonstrated impact of both nicotine addiction and health harms advertisements (in separate experimental arms) on a series of outcomes over time, including reduced susceptibility to vaping³⁶.

There have also been several studies of young adults (and adults) that may inform message themes for discouraging vaping behavior. Several e-cigarette warning experiments have used nicotine warnings^37,38, with one study revealing that FDA addiction warnings were more effective in increasing health risk beliefs compared to reduced-risk messages³⁹. Other e-cigarette message experiments targeting nicotine addiction and health harms themes have shown promising results at increasing risk beliefs about vaping and reducing intentions to vape^40,41. Studies have demonstrated that targeting toxic chemicals may be promising in motivating increased intentions to quit vaping⁴⁰, while another study found that themes focused on brain development and chemical harms were more effective in discouraging young adults from vaping initiation compared to addiction themes⁴². In fact, burgeoning research indicates that health harms messages beyond nicotine addiction may be promising in discouraging vaping among young adult (and adult) populations^40,42,43.

In addition to themes found to be promising, studies have found themes that may diminish the effectiveness of vaping prevention messages. For instance, themes that do not appear promising include those that target flavors or social consequences in both adolescent^33,44 and young adult⁴⁵ populations. Other themes that may diminish effectiveness are those related to industry targeting and environmental impact³³.

Altogether, this body of work provides fairly robust evidence for the promise of vaping prevention messages that emphasize the negative consequences of using vapes, especially chemical exposures, health harms, and possibly nicotine addiction. However, the literature on the health consequences of vaping is both nuanced and rapidly evolving, and there is a need for national studies that develop and test message themes with accurate messages based on the latest scientific evidence. In addition, there is a need to test novel themes for vaping prevention such as the impact of vaping on mental health, cosmetic effects of vapes, and the monetary cost of vapes, which have seldom been tested in the literature to date.

In the current investigation, we tested seven vaping prevention message themes focused on the negative consequences of using vapes (nicotine addiction, chemical harms, health symptoms, mental health, organ effects, cosmetic effects, and monetary cost). To do so, we employed a mixed within-and-between experimental design with a nationally representative sample of adolescents aged 13–17. We tested a series of vaping prevention messages representing seven message themes, with perceived message effectiveness (PME) as our primary outcome.

We posed the following research questions:

• RQ1: Which themes (of the seven examined) are perceived as most effective for discouraging vaping among adolescents?

• RQ2: Are theme effects moderated by vaping status?

• RQ3: Are message characteristics that vary across the set of messages (referencing the target “you” in the message and message length) associated with greater perceived message effectiveness?

Materials and Methods

Participants and Procedures

Participants were a sample of US adolescents (aged 13–17 years) recruited in summer of 2022 from the KnowledgePanel maintained by Ipsos. KnowledgePanel is a probability-based web panel designed to be representative of the US population. For recruitment, Ipsos contacted English-speaking US adults in their KnowledgePanel with children in the household aged 13 through 17 years old. Parents received an email invitation to complete a screener and those with eligible children living in their household were asked to allow their child to take the survey. Parental consent and adolescent assent were obtained for all participants before enrollment in the study. Of 5649 parents invited to participate in the survey, 2852 completed an initial screener (a 50.5% completion rate). A total of 1249 parents were ineligible after completing the screener (e.g., outside the age range or parent did not provide consent for youth to complete), resulting in 1603 parents who completed the screener, had a child in the eligible age range, and provided consent for their child to participate. The qualification rate was 56.2% (qualified completes out of total completes). A total of n=1603 adolescents comprised the final sample and the survey took a median of 16 min to complete. All surveys were completed in August 2022. The incentive for participation was a US $5 cash-equivalent, which was deposited in the parent’s account on behalf of the adolescent. The weighting process accounted for unequal probability of selection and non-response, with final weights aligning with population benchmark demographic distributions for those aged 13–17 years in the USA.

The survey assessed adolescents’ perceptions of vaping prevention messages. In the first section of the survey, participants were randomly assigned to view one message from each of eight thematic categories, including a control theme. In total, there were 46 messages - 43 intervention messages and 3 control messages. Each participant viewed 7 randomly selected vaping prevention messages (1 from each theme) and 1 control message (see Table 1). Messages were presented to each participant as text statements. Adolescents rated each message on PME, awareness, and believability. Control messages were about vaping-related litter and were derived from previous studies^34,40. The study was approved by University’s Institutional Review Board.

Table 1.

Vaping Prevention Themes and Example Messages Tested in the Study

Theme	Description	Total # of Messages	Example Message
Nicotine addiction	The addictive nature of nicotine found in vapes	7	“Some vapes deliver as much addictive nicotine as a pack of 20 cigarettes.”
Chemical harms	Chemicals found in e-cigarette aerosol that the user inhales while vaping	7	“E-cigarette vapor can contain formaldehyde, a known cause of cancer.”
Health symptoms	Potential symptoms of vaping, such as coughing, wheezing, and headaches	5	“Vaping can make you cough and wheeze.”
Mental health	Potential effects of vaping in worsening mental health	6	“Vaping could worsen feelings of depression.”
Organ effects	Potential effects of vaping on the body, including lungs, heart, and brain development	6	“Vaping could lead to lung diseases.”
Cosmetic effects	Potential effects of vaping on physical appearance	6	“Vaping could make your tongue look gross.”
Monetary costs	Monetary cost of vapes	6	“Vaping is an expensive habit. It could cost you $50 every week.”
Control (Littering)	Vape litter	3	“Vaping litter requires cleanup. Discard vapes properly.”

Theme Development

Guided by formative research with adolescents, including a quantitative study³³, meta-analysis⁴⁶, 7-member youth advisory board, in-depth interviews with adolescents⁴⁷, and existing literature^32,34,35,48, our overall message strategy focused on how vaping negatively affects youth users or prospective users.

More specifically, we arrived at a set of seven message themes (nicotine addiction, chemical harms, health symptoms, mental health, organ effects, cosmetic effects, and monetary cost) as a result of several research activities. First, we tested 220 existing vaping prevention messages among youth, with results pointing to nicotine addiction and several health harms of vaping as promising themes³³. Second, we conducted a meta-analysis that synthesized experiments testing the impact of vaping prevention messages on adolescents and young adults, finding support for addiction and health harms themes⁴⁶. Third, we explored a series of message concepts with our youth advisory board who gave us feedback on novel message concepts, with results suggesting less resonance of themes that were more distal from the adolescent (e.g., industry tactics, environmental impact).⁴⁷ After a broader set of concepts was winnowed to those focused on proximal negative consequences to the adolescent, we garnered responses to those message concepts through in-depth interviews with 27 adolescents⁴⁷. Results indicated that messages about nicotine addiction, health harms of vaping, and the monetary cost of vaping resonated most with youth. Fourth, we examined the literature on adolescent beliefs about vaping, which is premised on health behavior theories²⁶ and empirical evidence that beliefs about the negative consequences of behavior influence behavior and behavior change²⁷. Studies have revealed that adolescents who vape or are susceptible to vaping are less likely to believe that they will get addicted, experience health symptoms from vaping, and experience other health effects compared to their lower risk counterparts^29,49–51. Emerging evidence also suggested that beliefs about the mental health effects of vaping^16,17,52 and beliefs about the costs of vapes^53–55 may also be promising areas for messages. Thus, all of this work supported the testing of the seven themes (nicotine addiction, chemical harms, health symptoms, mental health, organ effects, cosmetic effects, and monetary cost).

Message Development

After arriving at the seven themes, we conducted an extensive literature review to examine the latest scientific evidence about the vaping harms emanating from our seven themes. The process of writing messages within each message theme was iterative and included several rounds of feedback from our research team. Once we had a broad set of messages covering all domains, we garnered feedback from a panel of four experts in toxicology and addiction science, focusing on 1) scientific accuracy of the messages, 2) scientific sources supporting the messages, and 3) appropriate hedging language to use in each message (e.g., can, could, may, etc.). The panel provided written feedback on all messages, followed by a virtual meeting to discuss the messages. The feedback resulted in some messages being revised and others being discarded. The expert panel then reviewed our revised messages, after which the set of messages was finalized. Our final set of messages consisted of 43 unique messages across the seven thematic categories, as well as three messages serving as controls (see Table 1). Control messages were focused on vape litter and were adapted from previous studies^34,40.

In addition to varying in theme, the messages also varied in length from a minimum of 5 words to a maximum of 15 words (median = 9 words). Messages also varied to the extent that they referred to the target of the message, with 40% referencing the target (using the word “you”) and 60% not referencing the target (not using “you.”)

Measures

Perceived Message Effectiveness (PME).

Prior studies have demonstrated PME to be a mediator of the impact of messages on behavioral and psychosocial outcomes^56,57 and to align with the actual effectiveness of tobacco prevention and control messages longitudinally^57,58. We assessed PME using the UNC PME Scale for Youth (α=.95)⁵⁹. After each message, participants were asked “how much does this message…” 1) make you worry about what vaping will do to you? 2) make you think vaping is a bad idea? and 3) discourage you from vaping? The 5-point response scale was 1 = not at all to 5 = a great deal.

Awareness and Believability.

Awareness assessed the extent to which the messages provided new information, while believability assessed message credibility. Adapted from Rohde et al.⁴³, awareness of the message was assessed by asking “Before today, had you heard about the vaping-related effect described in this message?” to which they could answer yes, no, or not sure. To understand how believable participants felt the messages were, they were asked “How believable is this message to you?” The 5-point response scales ranged from (1 = not at all, 5 = a great deal)⁶⁰.

Tobacco Product Use.

Participants were asked if they had ever used e-cigarettes or vaped, as well as whether they had vaped in the past 30 days. They were also asked if they had ever used cigarettes, cigars, hookah, pipe tobacco, or smokeless tobacco, both ever and in the past 30 days. Finally, participants were asked if they had ever vaped non-nicotine products, with a list that included THC, CBD, and other substances.

Vaping Status.

Youth who used an e-cigarette or vaped in the past 30 days were classified as current users. Both youth who had never vaped before and youth who indicated that they had vaped before, but not in the past 30 days, were asked a series of 5 items about their curiosity about and openness to vaping, with responses on 4-point scales (e.g., definitely not to definitely yes). If they answered anything other than “not at all curious” and “definitely not” to all items, they were classified as susceptible vaping. All other adolescents were classified as not susceptible to vaping⁶¹.

Demographics.

The survey assessed participant age, race, Hispanic ethnicity, gender, sexual orientation, and year in school. Adolescents were also asked if anyone they live with now uses tobacco products. Finally, household income and parent education were asked of parents.

Data Analysis

Our primary objective was to quantify the difference in PME for each of the seven message themes versus the control message, and to determine whether vaping status moderated these effects. To do this, we used the lme4 package⁶² in RStudio to predict the N=1,603 message PME ratings from dummy-coded message theme (reference=control) and control variables. Control variables used in all analyses were: gender, race, Hispanic ethnicity, sexual orientation, age, parental education, family income, and tobacco use in the home. We used a random intercept to account for nonindependence due to nesting of ratings within participant. We also used sampling weights in all analyses.

There was very little missing data (all variables <.5%), and missing responses occurred only for endogenous variables. We therefore relied on the full information maximum likelihood estimator for the linear mixed models to handle missing responses, assuming that missing outcomes were missing at random after accounting for an individual’s observed responses and demographic information.

Model-building occurred in three steps: 1) only control variables predicting PME; 2) control variables and message theme predicting PME; and 3) interactions between message theme and vaping status (non-susceptible, susceptible, user). We tested the omnibus effect of message theme on PME using likelihood ratio tests (LRT) to compare the fit of nested models. If the likelihood ratio test was significant in steps 2 or 3, we inspected individual terms to determine whether average PME for each “active” (non-control) message theme differed significantly from the control message (in step 2) or whether the effect of message theme on PME depended on participant vaping status (in step 3). Because interaction effects tested in step 3 were not hypothesis-driven, we used the Benjamini-Hochberg correction to control the false discovery rate for interaction effects⁶³.

Secondary objectives included determining whether message believability and awareness predicted PME, and also determining whether two other factors that varied across the messages had an impact on PME (i.e., use of the word “you” and message length). We used the same linear mixed modeling strategy described above, replacing message theme with the secondary predictors. We used the same set of covariates for these models. Message awareness and believability were included together in one set of models; message characteristics (use of “you” and number of words in each message) were included together in another set of models. We applied the Benjamini-Hochberg correction before interpreting the significance of the interaction effects.

Results

Participant characteristics

The mean age of adolescents was 15.07 years. Weighted percentages of participant characteristics are reported in Supplementary Table 1 and described here. Participants reported being male (49%), White (71%), and 24% identified as Hispanic or Latino. Participants primarily identified as heterosexual (90%) and about half of adolescents (54%) had a parent with a bachelor’s degree or higher. Finally, household income levels varied with 45% of participants living in households with an income of $100,000 or greater, about 27% living in households with an income of less than $50,000, 15% between $50,000 and $74,900, and 13% between $75,000 and $99,999.

Seven percent of adolescents were current vapers (i.e., used in past 30 days), while 40% were susceptible to vaping and 53% were not susceptible to vaping. Past 30-day other tobacco product use among the sample was low, and included cigarettes (4%), little cigars and cigarillos (3%), hookah (3%), and traditional cigars (2%). Participants reported ever vaping non-nicotine products, such as THC (5%) and CBD (5%), with 5% not knowing what substances they had vaped. Finally, 24% of adolescents lived in households with someone else who used tobacco products.

Effects of Vaping Themes on PME

Model fit for each of the models is shown in Supplementary Table 2. Message theme explained within-person variance in PME over and above the set of control variables. The LRT was highly significant when message theme was added to the model with only control variables. Although adding interactions between message theme and susceptibility status did not explain any additional within-person variance in PME, the LRT was significant, indicating that the addition of the interaction terms improved model fit.

The weighted means for vaping prevention message categories by vaping status are shown in Table 2. The main effects of message theme (versus control messages) are shown in Table 3. In the main effects model (Table 3), every intervention message theme was rated as significantly more effective than the control message theme. Table 4 reports pairwise comparisons of the effects of each message theme on PME. Most message themes differed from one another on PME even after applying the Benjamini-Hochberg correction, with both chemical harms and organ effects out-performing all other themes. The nicotine addiction theme out-performed all themes except for chemical harms and organ effects.

Table 2.

Weighted Means for Vaping Prevention Message Themes by Vaping Status

	PME	Believability	Awareness
Message category	M (SD)	M (SD)	%
Overall sample (n=1603)
Nicotine addiction	4.00 (1.09)	4.05 (1.07)	65
Chemical harms	4.12 (1.07)	3.93 (1.09)	42
Health symptoms	3.86 (1.12)	3.82 (1.12)	39
Mental health	3.74 (1.17)	3.55 (1.17)	34
Organ effects	4.13 (1.01)	4.02 (1.04)	45
Cosmetic effects	3.75 (1.18)	3.58 (1.14)	28
Monetary cost	3.63 (1.23)	3.82 (1.16)	34
Control	3.08 (1.37)	3.55 (1.23)	21
Non-susceptible to vaping (n=840)
Nicotine addiction	4.24 (1.05)	4.25 (1.04)	59
Chemical harms	4.31 (1.05)	4.10 (1.11)	44
Health symptoms	4.10 (1.09)	4.03 (1.12)	39
Mental health	3.99 (1.15)	3.77 (1.17)	35
Organ effects	4.32 (.98)	4.23 (1.01)	48
Cosmetic effects	4.03 (1.15)	3.81 (1.14)	31
Monetary cost	3.91 (1.22)	4.00 (1.18)	33
Control	3.39 (1.39)	3.63 (1.29)	18
Susceptible to vaping (n=656)
Nicotine addiction	3.82 (1.03)	3.90 (1.01)	62
Chemical harms	4.02 (1.00)	3.84 (.99)	39
Health symptoms	3.72 (1.03)	3.67 (1.02)	40
Mental health	3.57 (1.08)	3.40 (1.09)	34
Organ effects	4.03 (.95)	3.90 (.96)	43
Cosmetic effects	3.53 (1.11)	3.42 (1.05)	25
Monetary cost	3.42 (1.14)	3.72 (1.08)	34
Control	2.80 (1.28)	3.54 (1.14)	23
Current users (n=107)
Nicotine addiction	3.20 (1.79)	3.35 (1.19)	55
Chemical harms	3.22 (1.79)	3.16 (1.13)	41
Health symptoms	2.83 (1.68)	3.03 (1.15)	37
Mental health	2.75 (1.66)	2.70 (1.19)	33
Organ effects	3.22 (1.79)	3.17 (1.12)	38
Cosmetic effects	2.88 (1.70)	2.83 (1.18)	28
Monetary cost	2.69 (1.64)	3.08 (1.17)	41
Control	2.39 (1.55)	3.01 (1.18)	28

Note. PME = Perceived message effectiveness; PME and believability were assessed on a 5-point scale where higher scores indicate a greater amount of the construct. Awareness indicates the percent who answered yes.

Table 3.

Effects of Message Themes on PME

	Main Effects
	B	95% LB	95% UB
Susceptible (vs. non-susceptible)	−.41	−.51	−.31
User (vs. non-susceptible)	−1.07	−1.25	−.89
Message themes
Nicotine addiction (vs. control)	.92	.86	.98
Chemical harms (vs. control)	1.04	.98	1.10
Health symptoms (vs. control)	.76	.70	.82
Mental health (vs. control)	.65	.59	.71
Organ effects (vs. control)	1.04	.98	1.10
Cosmetic (vs. control)	.66	.60	.72
Monetary costs (vs. control)	.55	.49	.61

Note. Model controlled for gender, race, Hispanic ethnicity, sexual orientation, age, parental education, family income, and tobacco use in the home. Table displays regression coefficients (B) from linear mixed models, along with lower and upper bounds of 95% confidence intervals (LB and UB, respectively) for the regression effects. Bolded coefficients are statistically significant at p<.05.

Table 4.

Pairwise Comparisons of Message Themes on PME

	Nicotine addiction	Chemical harms	Health symptoms	Mental health	Organ effects	Cosmetic effects
Chemical harms	.12* (.03)
Health symptoms	−.16* (.03)	−.28* (.03)
Mental health	−.27* (.03)	−.39* (.03)	−.11* (.03)
Organ effects	.12* (.03)	.00 (.03)	.28* (.03)	.39* (.03)
Cosmetic effects	−.26* (.03)	−.38* (.03)	−.10* (.03)	.01 (.03)	−.38* (.03)
Monetary costs	−.38* (.03)	−.50* (.03)	−.22* (.03)	−.11* (.03)	−.50* (.03)	−.12* .03)

Note. Cells contain the difference in the effects of each pair of message themes, with standard errors in parentheses. These estimates were obtained by iterating the reference group in the linear mixed model.

^*Pairwise comparison is significant at p<.05 after applying Benjamini-Hochberg correction. A positive value indicates that the theme in the column to the left is more effective than the theme it is being compared to. A negative value indicates the opposite.

Message effects on PME were stronger for susceptible participants than for non-susceptible participants for chemical harms, health symptoms, mental health, and organ effects (Supplementary Table 3). None of the effects of message themes were significantly different for users versus non-susceptible youth. Except for nicotine addiction-themed messages, all message themes exerted significantly larger effects for susceptible youth versus users (see Supplementary Table 3 and Figure 1).

Figure 1.

Effects for each message theme (versus control theme) on PME by vaping status. Bars represent 95% confidence intervals around each weighted covariate-adjusted regression coefficient.

Effects of Believability and Awareness on PME

Table 5 shows the relationships between awareness and believability on PME. Adding message awareness and believability to the model resulted in significantly better model fit when compared to a model with covariates only (p<.001), but inclusion of interaction terms with susceptibility status did not result in significantly improved fit (p=.07). Both awareness and believability predicted higher levels of PME in the main effects model. The effect of awareness on PME was diminished amongst users, and this interaction term remained significant after applying the Benjamini-Hochberg correction. None of the other interaction terms were significant. Weighted PME means, believability means, and awareness percentages by vaping status are reported in Table 2 and Supplementary Table 5.

Table 5.

Effects of Awareness and Believability on PME and Moderation by Susceptibility

	Main Effects			Moderation by Susceptibility
	B	95% LB	95% UB	B	95% LB	95% UB
Susceptible (vs. non-susceptible)	−.27	−.33	−.21	−.37	−.51	−.23
User (vs. non-susceptible)	−.68	−.82	−.54	−.72	−.96	−.48
Awareness	.17	.13	.21	.18	.14	.22
Believability	.44	.42	.46	.43	.41	.45
Awareness* susceptible				.00	−.08	.08
Believability* susceptible				.03	−.01	.07
Awareness* user				−.15	−.27	−.03
Believability* user				.03	−.03	.09

Note. Models controlled for gender, race, Hispanic ethnicity, sexual orientation, age, parental education, family income, and tobacco use in the home. Table displays regression coefficients from linear mixed models. All bolded effects are statistically significant at p<.05. Benjamini-Hochberg correction only applied to interaction tests.

Effects of Message Characteristics on PME

Supplementary Table 4 shows the effects of message characteristics on PME. The main effect model with message characteristics fit significantly better than a model with covariates only (p<.001), and the model with interaction terms fit significantly better than the main effects-only model (p=.03). Adolescents rated messages that included the word “you” higher on PME compared to messages without the word “you.” They also rated messages with more words higher on PME than messages with fewer words. Although the model with interaction terms had significantly better fit than the main effects-only model, none of the individual interaction terms were significant after applying the Benjamini-Hochberg correction.

Discussion

We found that all message themes tested in this study had some level of promise for vaping prevention, with chemical harms and organ effects emerging as the most promising themes, followed by nicotine addiction and then the other themes. For most of the message themes, PME effects were stronger for youth susceptible to vaping compared to non-susceptible youth and users. Both awareness and believability were positively associated with PME. Finally, messages referencing the target (i.e., “you”) and longer messages were rated higher on PME and appear more promising for vaping prevention. Results from this nationally representative study can inform the development of evidence-based campaigns and interventions for youth vaping prevention.

Our findings for the promise of chemical harms and organ effects themes – followed by nicotine addiction – is consistent with some prior research^33,36,45. One explanation for these findings is that there is more scientific evidence in these areas compared to areas such as mental health and cosmetic effects of vaping. The less studied themes have been seldom used in vaping prevention campaigns, which have mostly focused on addiction and health harms of vaping⁶⁴. It may be that participants were skeptical of the more novel messages on mental health and cosmetic effects because they were less likely to have heard them before. This hypothesis seems somewhat supported by the relatively low awareness of the more novel themes and the fact that awareness and believability were associated with PME.

It is also possible, however, that the consequences of chemical harms and organ effects were perhaps viewed as more severe than other themes and were thus rated as more effective. Prior work with smokers has found that messages about more severe outcomes, such as lung cancer and heart disease, tend to be rated as more effective in discouraging smoking than less severe outcomes such as high cholesterol and tooth decay⁶⁵. Thus, the less severe consequences of vaping such as feeling sick to stomach (health symptoms) or one’s tongue looking gross (cosmetic effects) may have been perceived as less impactful than the more severe consequences encompassed by organ effects, especially impact on heart and lungs, which is borne out by the higher PME ratings for the heart and lung messages (see Supplementary Table 5).

Another key finding from this study was that message effects on PME were stronger for susceptible youth compared to non-susceptible youth for the chemical harms, health symptoms, mental health, and organ effect themes. This is a promising finding because youth who are susceptible to vaping are an important at-risk group who are much more likely to initiate vaping compared to youth who are not susceptible to vaping^49,66. The fact that nicotine addiction was not among these interaction effects may indicate that it is not as strong a message for susceptible youth as compared to these other health harms, perhaps because of susceptible youth’s inexperience with vaping and potential misperceptions of addiction^44,67 It is important to note, however, that other evidence suggests that nicotine addiction messages are promising for susceptible youth^33,36. The fact that our findings differ from these studies could be due to several factors, including the use of imagery or video in those other studies or the fact that nicotine addiction messages in other studies may include content not present in our messages, such as the health impact of nicotine or the social consequences of addiction⁶⁸.

Both awareness and believability were positively associated with PME, echoing prior research on tobacco control messages^43,60. Awareness of harms varied greatly both within and across themes, and given the well-known connection between message exposure and impact⁶⁹, messages are likely to have greater effects with greater levels of exposure and thus familiarity. Those developing and implementing campaigns should carefully choose which messages themes and messages to emphasize and ensure adequate exposure and repetition to maximize the likelihood of impact.

We also found that messages that referenced the target of the message – “you” –were rated more highly than messages without “you” language. Past research suggests that you-based appeals are more broadly effective among younger audiences⁷⁰. Compared to messages from the third-person perspective (e.g., they, teens), messages from the second-person perspective (e.g., you) should encourage self-referencing and increase relevance^71,72. One other prior vaping prevention study also found that messages containing “you” language were associated with higher PME scores compared to messages where other perspectives (e.g., “I/we” or “teens”) were used³³. Therefore, using “you” in vaping prevention messages may help increase their effectiveness.

Longer messages were also rated as more discouraging of vaping. Longer messages are able to say more and may be perceived as more effective since they are able to both name a health hazard and explain the potential impact of hazard (e.g., “E-cigarette vapor can contain formaldehyde, a known cause of cancer.”). This can be compared to messages that only name a chemical or health effect, but provide no explanation of the “why.” In addition, the current study only focused on relatively short statements, all of which ranged from 5 to 14 words. Thus, we would caution against interpreting this finding to mean that adolescents prefer “long” messages given other studies that suggest that shorter messages are generally preferred and easier to comprehend⁴⁷.

Implications

Our results suggest that in order to have the greatest impact, youth vaping prevention efforts should prioritize messages about vape chemicals (chemical harms) and the potential of vaping to damage organs and increase disease risk (organ effects). Communication campaigns and preventive interventions can apply the theory and evidence-based insights from this study to maximize the potential of vaping prevention messages for youth, including selecting among most promising messages within each theme (see Supplementary Table 5). Further work is needed to examine which themes and messages will have the most impact on adolescents’ vaping beliefs and behaviors, while also identifying the most promising message channels in which to reach adolescents.

Strengths and Limitations

Strengths of this study include a nationally representative survey of US adolescents, the use of evidence-based messages developed through an iterative step-by-step process, the testing of novel themes for vaping prevention, an experimental design that allowed for testing of a large pool of messages across themes, and the use of robust measures including the UNC PME Scale for Youth⁵⁹.

One limitation of this study is the modest number of current users of vapes and other tobacco products, though we did have a large number of adolescents who were susceptible to vaping. A second limitation is that while we were able to identify promising vaping prevention themes, we did not test the impact of these themes on risk beliefs or behaviors. However, there is a growing body of literature suggesting that PME is a good proxy for message effectiveness, and is especially useful in the early stages of message development^57,58,73. Future work should test vaping prevention messages within the promising themes identified here for impact on key outcomes including vaping beliefs, susceptibility, and behavior. While this study advances the literature on vaping prevention messages for adolescent populations, a final limitation is that we are unable to inform vaping prevention strategies for key disparity subgroups, such as LGBTQ+ youth. Future work should oversample disparate subgroups and examine whether the themes studied here are more or less effective with LGBTQ+ youth.

Conclusion

Youth in the current study rated messages about chemical harms and organ effects most promising, followed by nicotine addiction, with all vaping prevention themes outperforming our control theme (vape litter). In addition, message effects on PME were stronger for susceptible youth compared to non-susceptible youth and users for the majority of message themes. Vaping prevention messages that used second-person (“you”) language and were longer were also rated as more effective than those that did not contain the word “you” and shorter messages. Communication campaigns and preventive interventions should prioritize these promising themes and message characteristics in attempts to increase the effectiveness of vaping prevention communication, and in turn reduce the harms of e-cigarettes and vaping among adolescent populations.

Highlights.

• Nationally representative sample of 1,603 US youth rated vaping prevention messages

• Chemical harms and organ effects message themes had the largest effects

• Effects were stronger for youth susceptible to vaping for most message themes

• Messages using the word “you” and longer messages were rated more effective