Applying a mixed-methods approach to evaluate and tailor electronic cigarette pictorial health warning labels for young adults in the United States Authors: Taghrid Asfar, Ian Abrams, Shannon Varroney, Michael Schmidt, Olusanya Joshua Oluwole, Seth M. Noar, Laura Gonzalez, Maria Fernanda Garcia, Wasim Maziak Published: 2026 ABSTRACT BackgroundThe use of electronic nicotine delivery systems (ENDS) has risen significantly among US young adults, mainly driven by marketing, product design, and addictive nicotine. Pictorial warning labels (PWLs) effectively inform consumers about the risks associated with ENDS use, yet young users were rarely involved in their development. A set of 24 PWLs, categorized into three themes, was developed by a panel of international tobacco control experts via a Delphi study: ENDS toxicity, health risks, and specific harm. In this study, we involved young ENDS users through a mixed-methods approach, including a rating survey to identify the most effective warnings and focus group discussions to gather their feedback on the PWLs design, content, and needed improvements.MethodsWe conducted 22 HIPAA-compliant Zoom focus groups with 61 ENDS users (57% females, aged 21–29). Guided by the Message Impact Framework, participants completed an anonymous rating survey of the PWLs to assess attention, reactions, and perceived effectiveness, followed by in-depth discussions of each PWL. Descriptive statistical analysis was conducted for the rating survey. Focus group sessions were audio-recorded, transcribed, and thematically analyzed.ResultsSurvey results showed toxicity-focused PWLs outperformed those on health effects and specific harm across most outcomes. Focus group results showed warnings depicting human suffering were most impactful. Relevance to participants (e.g., pregnancy, health experiences) emerged as influential factors shaping perceptions of warning effectiveness. Metaphorical images were often perceived as confusing or less believable, thereby diminishing their persuasive potential. Participants recommended applying clear, simple visual and verbal explanations.ConclusionsThis study identified promising PWLs for young adult ENDS users and provides practical guidance to strengthen warning design and regulatory efforts. Warnings featuring clear, literal imagery and human health consequences elicited greater fear, perceived severity, and motivation to reconsider ENDS use, whereas metaphor-based images were often viewed as less credible or confusing. The convergence of quantitative and qualitative findings underscores the importance of targeting emotional and cognitive message responses to enhance warning effectiveness. Since personal relevance varies per individual, using a rotating, diverse portfolio of ENDS warnings targeting different health risks likely improves subgroup resonance and maximizes population impact.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-26775-w. Background Recent progress in tobacco control has reduced cigarette smoking, but use of new products like electronic nicotine delivery systems (ENDS) has surged. ENDS has become the most popular product among young adults in the US over the past decade [1–3]. The increase is fueled by appealing flavors, ease of use, efficient nicotine delivery, targeted marketing, and the belief that they are safe [4]. This misconception about safety is concerning, as ENDS pose health risks such as toxicity, nicotine addiction, and cardiopulmonary problems that many young users are unaware of [4, 5]. Health warning labels effectively educate tobacco users about risks and encourage them to quit [6]. The Food and Drug Administration (FDA) now regulates ENDS under the Tobacco Control Act, requiring packaging to state: “WARNING: This product contains nicotine. Nicotine is an addictive chemical.” [7]. However, research demonstrates that pictorial warning labels (PWLs) are more effective than text-only labels, and that ENDS’ risks extend beyond nicotine addiction [5, 8]. In 2022, our team began developing and testing ENDS-specific PWLs to inform young adults about the risks associated with ENDS. We started with an umbrella review to evaluate the evidence on the ENDS risk profile, including ninety systematic reviews [5]. This review identified risks such as nicotine dependence, respiratory diseases, exposure to heavy metals and toxins, poisoning, and explosions. Supported by strong evidence, we created 24 PWLs. We then used a Delphi study with tobacco control and health communication experts to validate and refine the PWLs [9] (Fig 1).Fig. 1The 24 pictorial warning labels used in the focus groups The 24 pictorial warning labels used in the focus groups Current guidelines for designing and evaluating PWLs emphasize the importance of involving the target population during development, including the use of both qualitative and quantitative assessments to gather their feedback before implementing policies [10]. Several prior studies have evaluated ENDS warnings using experimental designs. For example, Mays et al. tested variations of warning design features, including size, placement, and pictorial versus text-only elements, among young adults, demonstrating that visual prominence and imagery increased attention and recall [11]. Similarly, Brewer et al. and Sontag et al. examined ENDS PWLs focused on nicotine addiction and health harms in national samples of adolescents and young adults, including smokers and non-users [12–14]. While these studies show that PWLs can enhance attention and perceived harm compared to text-only messages, most relied on researcher-generated stimuli and experimental-based evaluations without iterative qualitative input from the target population [11–16]. The present study addresses these gaps by directly engaging young adult ENDS users in the evaluation and refinement of expert-developed PWLs, using a mixed-methods design guided by the Message Impact Framework [17]. Participants completed a brief survey followed by a focus group discussion to identify the most promising warnings in terms of overall perceived effectiveness and to gather feedback on the PWLs design and content, as well as any needed improvements. This project aims to develop evidence-based ENDS PWLs for health promotion, campaigns, or warnings to reduce ENDS use among young adults. These PWLs can also be tailored and tested across diverse US populations (e.g., adolescents). Methods Design The study employed a convergent parallel mixed-methods design with simultaneous quantitative (survey) and qualitative (focus group) data collection and analysis, allowing for triangulation and cross-verification of findings [18]. While the survey offered a snapshot of PWLs’ effectiveness and prioritized top PWLs for development, focus groups provided deeper insights into participants’ reactions and key suggestions for improvements. Between August 2022 and June 2023, twenty-two mixed-gender focus groups, combined with a brief survey, were conducted online via HIPAA-authorized Zoom among young ENDS users (N = 61; 57% female; age 21–29 years). The PWLs used in this study were developed by a panel of international experts in tobacco control through a Delphi study and categorized into three themes (Ts): T1) ENDS toxicity, T2) health risks, and T3) specific harm [19]. While nicotine addiction is a known ENDS risk included in current FDA warnings, this study examined a broader range of ENDS-specific health risks beyond addiction. This approach enabled comparison of addiction messages with warnings on toxicity and disease, identified via our review and Delphi process, to assess which themes prompted stronger responses among young adults. The group discussion was guided by a semi-structured script based on the Message Impact Framework, which suggests that PWL features lead to behavioral change through psychological stages. These stages involve paying attention to the label that triggers reactions and emotions, influencing risk perception, which then prompts an intention to change and ultimately results in actual behavioral change [20]. The Institutional Review Board at the University of Miami approved the study. Participants received $75, paid via gift card or cash. Participants and recruitment We targeted 21–29-year-olds who are current ENDS users (past month use). Although adolescence is a critical period for ENDS initiation in the US, we focused on young adults because this group remains at elevated risk for ENDS uptake and progression, [21] and messages developed for young adults may apply to younger populations. Importantly, the products evaluated are legally available only to individuals aged ≥ 21 years, necessitating this age restriction to address ethical and regulatory considerations. Additionally, this age range facilitated recruitment through university-based sampling. Collectively, these considerations support our focus on young adults as both methodologically appropriate and practically justified. Individuals who used other tobacco products were not excluded because national surveillance data (e.g., the Population Assessment of Tobacco and Health—PATH) indicated that dual and poly-tobacco use is common among young adult ENDS users [22]. This approach was intended to improve external validity by reflecting the heterogeneity of tobacco use patterns in this population. Recruitment used online and offline methods, including posters, flyers at universities and ENDS stores, and student listservs, Facebook, and social media [23]. To enhance recruitment efforts, we implemented snowballing methods. Materials were IRB-approved and included contact information for initial phone screening and scheduling focus groups. Procedures The focus group met via HIPAA-authorized Zoom for about 2 h, moderated by an investigator and 2 public health students trained in qualitative research. Each focus group viewed all 24 PWLs presented as high-resolution images. Each session began with a discussion of the group's nature, confidentiality, and interaction mode. After explaining the study and obtaining written informed consent, participants completed a baseline assessment focusing on demographics and tobacco-use history, followed by a survey rating each PWL on attention, emotional reaction (e.g., fear, believability), harm perception, and perceived effectiveness on intention to stop ENDS use, using a 5-point Likert scale (0 = not at all to 5 = a lot) [24]. The individual rating was completed before the group discussion to avoid contamination of individual responses by peer influence [25, 26]. Following the survey, the focus group discussion took place in three 15–20-min segments, each dedicated to a specific theme. Each segment began with a PowerPoint (PPT) presentation on theme-specific PWLs, with each image displayed sequentially to facilitate an in-depth conversation about the same PWLs that participants had already viewed and rated individually. Participants were instructed to review each PWL (both image and text) carefully, reflect on their immediate reactions, clarity, perceived effectiveness, and relevance, and provide a recommendation for improvement. During the discussion, images were displayed for approximately 1–2 min or longer as needed to support the conversation. PWLs in both the focus group discussion and surveys were rotated using Latin-square randomization to mitigate order and carryover effects [27]. Analysis Descriptive statistical analysis using means and standard deviations (mean [SD]) for continuous variables and frequencies and proportions (N) for categorical variables was conducted for the demographics, tobacco use history, and the PWLs’ rating. Inferential statistical tests were conducted only for demographic characteristics (Table 1); ratings of pictorial warning labels (Table 2) were analyzed descriptively using means and standard deviations to examine relative patterns across warnings, and no inferential comparisons were performed on these outcomes. Accordingly, continuous variables were compared using independent-samples t-tests, and categorical variables were compared using chi-square tests only in Table 1. All analyses were performed using SAS 9.4, with alpha levels < 0.05 deemed significant for pairwise comparisons. The survey component was designed to complement qualitative findings rather than test formal hypotheses.Table 1Demographics and tobacco use behaviorAll N = 61 (100%)Male N = 26 (42.6%)Female N = 35 (57.4%)P valueAge, mean (SD)23.61 (2.00)24.23 (2.03)23.14 (1.88)0.035Ethnicity, n (%)0.088 Non-Hispanic White20 (33.3)6 (24.0)14 (40.0) Hispanic21 (35.0)13 (52.0)8 (22.9) Non-Hispanic Black2 (3.3)1 (4.0)1 (2.9) Others17 (28.3)5 (20.0)12 (34.3)Level of education, n (%)0.041 High school or less7 (11.7)4 (16.0)3 (8.6) More than high school53 (88.3)21 (84.0)32 (91.4)Sexual orientation, n (%)0.409 Heterosexual45 (75.0)22 (88.0)23 (65.7) Homosexual15 (25.0)3 (12.0)12 (34.3)Income per year, n (%)0.329 $30,000 or less34 (56.7)15 (60.0)19 (54.3) $30,000 to $50,00015 (25.0)4 (16.0)11 (31.4) More the $50,00011 (18.3)6 (24.0)5 (14.3)Cigarette smoking in the past 30 days – Yes, n (%)9 (18.0)5 (22.7)4 (14.3)0.441Ever tried quitting cigarette smoking—Yes, n (%)12 (75.0)6 (85.7)6 (66.7)0.585Other tobacco use in the past 30 days—Yes, n (%)28 (45.9)17 (65.4)16 (45.7)0.127Seeing HWLs on a package in the past 30 days—Yes, n (%) ENDS package or device45 (76.3)19 (76.0)26 (76.5)0.967 Cigarette37 (61.7)17 (65.4)20 (58.8)0.6045 Cigar, little cigar, cigarillo, or flavored cigar28 (49.1)14 (53.9)14 (45.2)0.514 Chewing tobacco, snuff, or dip19 (33.9)8 (32.0)11 (35.5)0.784 Hookah or waterpipe19 (32.8)9 (36.0)10 (30.3)0.647ENDS age of initiation, mean (SD)19.44 (2.86)19.62 (3.06)19.31 (2.74)0.895Cigarette age of initiation, mean (SD)18.5 (2.37)18.86 (2.67)18.22 (2.22)0.714Time since starting vaping, n (%)0.523 More than a year53 (86.9)21 (80.8)32 (91.4) Within past 6 months5 (8.2)3 (11.5)2 (5.7) More than 6 months but less than a year3 (4.9)2 (7.7)1 (2.9)ENDS frequency use in the past 6 months, n (%)0.379 Every day41 (68.3)17 (68.0)24 (68.6) At least once a week, but not every day15 (25.0)5 (20.0)10 (28.6) Less than once a week, or a couple of times per month4 (6.7)3 (12.0)1 (2.9)Number of ENDS used per day1.85 (2.53)2.25 (2.59)1.57 (2.50)0.092ENDS type used in the past 30 days Disposable (Yes), n (%)51 (85.0)19 (73.1)32 (94.1)0.032 Vape pod (Yes), n (%)36 (60.0)15 (60.0)21 (60.0)1.000 Vape-pen (Yes), n (%)24 (40.7)13 (52.0)11 (32.4)0.129 Vuse (Yes), n (%)17 (28.3)6 (23.1)11 (32.4)0.429 Juul (Yes), n (%)14 (23.3)7 (26.9)7 (20.6)0.565 Vape mods (Yes), n (%)13 (21.7)5 (19.2)8 (23.5)0.689 Cig-a-like (Yes), n (%)4 (6.6)2 (7.7)2 (5.7)1.000 Others (Yes), n (%)15 (27.3)6 (26.1)9 (28.1)0.867Ever tried to quit ENDS—Yes, n (%)47 (77.0)20 (76.9)27 (77.1)0.984Number of serious quit attemptsa—One or more, n (%)37 (88.1)16 (94.1)21 (84.0)0.632Intention to quit ENDSb, mean (SD)3.53 (1.37)3.64 (1.60)3.46 (1.20)0.247Motivation to quit ENDS in the next monthb, mean (SD)2.85 (1.54)2.88 (1.86)2.83 (1.29)0.837aIn the past year, how many times have you tried to quit and succeeded in going without an e-cigarette for at least 24 hbOn a scale of 0 to 5Bold font values indicate significance (P < .05)Table 2Participants’ rating of the warnings using a 5-point scale in the brief survey (1: strongly disagree to 5: strongly agree;n=61)Bold values indicate the top-rated warnings discussed in the qualitative results section Table 1Demographics and tobacco use behaviorAll N = 61 (100%)Male N = 26 (42.6%)Female N = 35 (57.4%)P valueAge, mean (SD)23.61 (2.00)24.23 (2.03)23.14 (1.88)0.035Ethnicity, n (%)0.088 Non-Hispanic White20 (33.3)6 (24.0)14 (40.0) Hispanic21 (35.0)13 (52.0)8 (22.9) Non-Hispanic Black2 (3.3)1 (4.0)1 (2.9) Others17 (28.3)5 (20.0)12 (34.3)Level of education, n (%)0.041 High school or less7 (11.7)4 (16.0)3 (8.6) More than high school53 (88.3)21 (84.0)32 (91.4)Sexual orientation, n (%)0.409 Heterosexual45 (75.0)22 (88.0)23 (65.7) Homosexual15 (25.0)3 (12.0)12 (34.3)Income per year, n (%)0.329 $30,000 or less34 (56.7)15 (60.0)19 (54.3) $30,000 to $50,00015 (25.0)4 (16.0)11 (31.4) More the $50,00011 (18.3)6 (24.0)5 (14.3)Cigarette smoking in the past 30 days – Yes, n (%)9 (18.0)5 (22.7)4 (14.3)0.441Ever tried quitting cigarette smoking—Yes, n (%)12 (75.0)6 (85.7)6 (66.7)0.585Other tobacco use in the past 30 days—Yes, n (%)28 (45.9)17 (65.4)16 (45.7)0.127Seeing HWLs on a package in the past 30 days—Yes, n (%) ENDS package or device45 (76.3)19 (76.0)26 (76.5)0.967 Cigarette37 (61.7)17 (65.4)20 (58.8)0.6045 Cigar, little cigar, cigarillo, or flavored cigar28 (49.1)14 (53.9)14 (45.2)0.514 Chewing tobacco, snuff, or dip19 (33.9)8 (32.0)11 (35.5)0.784 Hookah or waterpipe19 (32.8)9 (36.0)10 (30.3)0.647ENDS age of initiation, mean (SD)19.44 (2.86)19.62 (3.06)19.31 (2.74)0.895Cigarette age of initiation, mean (SD)18.5 (2.37)18.86 (2.67)18.22 (2.22)0.714Time since starting vaping, n (%)0.523 More than a year53 (86.9)21 (80.8)32 (91.4) Within past 6 months5 (8.2)3 (11.5)2 (5.7) More than 6 months but less than a year3 (4.9)2 (7.7)1 (2.9)ENDS frequency use in the past 6 months, n (%)0.379 Every day41 (68.3)17 (68.0)24 (68.6) At least once a week, but not every day15 (25.0)5 (20.0)10 (28.6) Less than once a week, or a couple of times per month4 (6.7)3 (12.0)1 (2.9)Number of ENDS used per day1.85 (2.53)2.25 (2.59)1.57 (2.50)0.092ENDS type used in the past 30 days Disposable (Yes), n (%)51 (85.0)19 (73.1)32 (94.1)0.032 Vape pod (Yes), n (%)36 (60.0)15 (60.0)21 (60.0)1.000 Vape-pen (Yes), n (%)24 (40.7)13 (52.0)11 (32.4)0.129 Vuse (Yes), n (%)17 (28.3)6 (23.1)11 (32.4)0.429 Juul (Yes), n (%)14 (23.3)7 (26.9)7 (20.6)0.565 Vape mods (Yes), n (%)13 (21.7)5 (19.2)8 (23.5)0.689 Cig-a-like (Yes), n (%)4 (6.6)2 (7.7)2 (5.7)1.000 Others (Yes), n (%)15 (27.3)6 (26.1)9 (28.1)0.867Ever tried to quit ENDS—Yes, n (%)47 (77.0)20 (76.9)27 (77.1)0.984Number of serious quit attemptsa—One or more, n (%)37 (88.1)16 (94.1)21 (84.0)0.632Intention to quit ENDSb, mean (SD)3.53 (1.37)3.64 (1.60)3.46 (1.20)0.247Motivation to quit ENDS in the next monthb, mean (SD)2.85 (1.54)2.88 (1.86)2.83 (1.29)0.837aIn the past year, how many times have you tried to quit and succeeded in going without an e-cigarette for at least 24 hbOn a scale of 0 to 5Bold font values indicate significance (P < .05) Demographics and tobacco use behavior aIn the past year, how many times have you tried to quit and succeeded in going without an e-cigarette for at least 24 h bOn a scale of 0 to 5 Bold font values indicate significance (P < .05) Table 2Participants’ rating of the warnings using a 5-point scale in the brief survey (1: strongly disagree to 5: strongly agree;n=61)Bold values indicate the top-rated warnings discussed in the qualitative results section Participants’ rating of the warnings using a 5-point scale in the brief survey (1: strongly disagree to 5: strongly agree;n=61) Bold values indicate the top-rated warnings discussed in the qualitative results section Focus groups were audio recorded, transcribed verbatim, and analyzed using NVivo. Two researchers reviewed transcripts and reached consensus to identify recurring themes based on pre-defined codes within qualitative content analysis [28]. A third coder reviewed summaries against data, resolving differences through peer review debriefing [29]. Coding was stratified by sex and deductively aligned with the message impact framework's five constructs: attention, emotional reactions, cognitive reactions, harm perception, and intentions to change behavior. Quotes were categorized under these constructs, and summaries drafted. Preliminary codes were revised using the constant comparative method, enabling inductively derived codes to emerge and new main categories to be added [30]. Finally, given well-documented sex differences in tobacco use patterns, risk perceptions, and responses to warnings, we stratified qualitative coding by sex. This approach allowed us to systematically explore potential variation in attention, emotional reactions, perceived relevance, and behavioral intentions in response to the PWLs. Results Quantitative results Overall, 61 young adults participated in 22 online focus groups, averaging two to five participants per group (57% female; mean age 23.6 years) (Table 1). Over half of participants (56.7%) had household incomes ≤ $30,000. Most used disposable ENDs daily (85.0%) for over a year (86.9%). Other tobacco use was common (45.9%), with 18.0% being current cigarette smokers. Most top-rated PWLs for overall effectiveness were in theme 1, including T1-PWL5 (lung damage by acrolein), T2-PWL11 (popcorn lung disease), T1-PWL3 (exposure to glycidol, a possible human carcinogen), and T1-PWL1 (exposure to formaldehyde, a highly toxic systemic poison) (Table 2). Qualitative results PWLs in T1 (toxicity) outperformed those in T2 (health effect) and T3 (specific harm), receiving the highest ratings across all health communication outcomes (e.g., reaction, perceived effectiveness). However, we selected the top four PWLs from each theme (totaling 12) for this study to ensure broad representation of important themes. Below is a qualitative analysis of the top 12 PWLs, accompanied by a supplementary table for the remaining 12 (Supplement, Table 1). T1-PWL1 (formaldehyde toxicity) Strong emotional reactions to the warning were noted: a man remarked, “Strong words: highly toxic and poison,” while a woman said ‘‘highly toxic, systemic poison’ is very scary.” Many found the warning effective because of its imagery. One woman noted, “The black liquid stands out.” Some saw a weak link between the picture and text. A woman said, “I don’t understand the connection; I’d ignore it,” and another added, “The picture doesn’t effectively match the sentence. The frog in water is a bit confusing.” Opinions on effectiveness mainly stem from triggering fear: one individual stated, “I find it effective because it creeps me out.” Both males and females clearly voiced these opinions. T1-PWL2 (formaldehyde toxicity) The warning drew significant attention but also generated confusion. A woman remarked, “The mention of DNA is compelling because of its crucial role in life.” A man echoed this enthusiasm, stating, “The picture and words caught my eye, especially the DNA aspect; it makes me want to learn how it affects my DNA.” Reactions included “shock,” “regret,” “believable,” and “scary.” One said, “The image is disturbing, and words like glycidol and respiratory issues—absolutely raise alarms.” Another added, “It’s shocking; it suggests a broader impact than just lungs.” Some perceived it as effective, while others remained skeptical; one said, “It doesn’t give enough reason to worry since I don’t know formaldehyde. Saying it affects DNA is vague.” Another added, “It piques my curiosity, but lacks clarity. What does damaging my DNA mean?” Someone else said, “It feels neutral. If ‘damage’ isn’t clear, how does it impact me in the short term? Am I at risk of cancer if I vape?”. T1-PWL3 (glycidol toxicity) Although the survey showed that this warning received the second-highest rating, it did not provoke strong emotional reactions from most participants. Participants found the message ineffective, with one stating, “I don’t think it’s really strong. It’s too wordy and refers to a possible carcinogen.” Another said, “I don’t fully understand it, so I wouldn’t be particularly concerned.” One observed, “To me, it looks like healthy lungs, so it doesn’t scare me as much.” When asked if the label made them think about quitting vaping, participants shook their heads and said, “No,” with one adding, “I don’t think this label is very effective.” These negative opinions were consistent among both men and women. T1-PWL5 (acrolein toxicity) Both male and female participants agree that this message effectively captures attention. One said, “Intubated sounds serious and will attract attention.” Another noted, “It caught my attention and felt frightening.” Overall, participants described their reactions as “real,” “intimidating,” “shocking,” “scary,” “empathy,” and “urgency.” One woman asserted, “It feels more authentic because nobody wants tubes on their throat.” Another added, “It is frightening to consider being on a ventilator.” Some participants were in denial. A skeptic remarked, “It’s just a random person.” Another commented, “It looks frightening with tubes, but I interpret it as a person in a coma, not a lung issue.” Despite different interpretations, most participants found the message effective in motivating quitting. A male said, “If I saw that warning, I wouldn’t buy.” T2-PWL10 (asthma risk) Participants generally felt that attention, reactions, and perceived effectiveness of this warning depended on their knowledge of the condition “asthma” and doubted its effectiveness for the broader population. One woman said, “This is good. I struggle with asthma, so I needed the reminder,” while another commented, “I will ignore it because I don’t have asthma.” Male participants shared similar opinions, with one stating, “It’s too specific and targets only a certain group,” and another added, “It’s effective, but only for a targeted group like those with asthma.” T2-PWL11 (lung disease risk) This message grabbed attention, but some felt its emphasis on THC limited appeal to non-THC users. One participant said, “It was good except for THC; it might alienate non-THC users.” Another questioned the focus, noting, “It only mentions vaping THC causing popcorn lung, which raises doubts for dual users.” Reactions included feeling “grossed out,” “scared,” and “worried.” A woman said, “I think the picture is scary,” and another said, “The word ‘fatal’ worried me, especially linked to something I thought was safe… The picture is effective.” A man shared, “… ‘potentially fatal lung disease’ doesn’t sound good; I don’t want that.” Some thought lack of knowledge could lessen emotional impact; one male said, “I have no idea what popcorn lung is, so it doesn’t bother me.” Overall, opinions mainly varied by relevance: one female regarded it as impactful, suggesting a serious lung disease associated with a hospital setting, whereas another female stated, “This is ineffective for me because I don’t smoke THC.” T2-PWL14 (explosions risk) Participants found the eye injury image attention-grabbing but sometimes distracting from the text. One said, “The picture definitely grabs attention. I feel like people would pay less attention to the words and just look at the picture.” Many shared vaping explosion experiences, with one stating, “I actually have had a vape with a battery explode. So, it is kind of a scary thing.” A participant suggested showing other injuries, like burns, instead of only eye injuries: “It’s not only the eyes being harmed… Maybe your hand is getting burned.” The label evoked strong emotions, with words like “crazy,” “dramatic,” and “scary.” A female participant said, “It is kind of like a fear of mine… But I know it doesn’t happen always.” Some questioned the message; one noted, “I’ve gone through a thousand vapes, and nothing like that has ever happened.” Overall, the label was seen as effective, especially because of personal experiences: “I’ve had friends whose vapes exploded… I think this one is very effective, and I love the picture,” and “Yes, definitely heard of vapes exploding, so I think this one is pretty effective.” T2-PWL16 (children’s risk) Participants found the message was unengaging, mainly because it targets adolescents and children. The image was deemed inappropriate, causing confusion among older viewers. One participant noted, “The child in the picture makes it clear this is meant for kids.” Another participant added, “I might relate more if I were younger, but I still do not perceive it as a major concern.” One person stated, “The image doesn’t connect to concentration and learning; it looks like a tutoring session.” Questions regarding the relevance of the image were prevalent, with one person saying, “If we’re talking about middle school, it makes sense, but the kid seems out of place.” Overall, the message was informative but failed to attract attention, evoke emotion, or effectively promote prevention or cessation. One participant commented, “While I appreciate the perspective, it seems more focused on cognitive learning than health effects.” T3-PWL17 (pregnancy risk) Participants noted that the message primarily targeted women, especially those who are pregnant or considering pregnancy. A female participant remarked, “It feels like it’s more targeted to women,” while a male participant added, “I wouldn’t smoke around pregnant women.” Many described the message using terms like “disturbing,” “creepy,” “scary,” and “concerning.” One participant said, “The image is disturbing but irrelevant if I’m not pregnant.” Another shared concerns about discussions on smoking with their pregnant partner, despite not having personal issues with smoking. Overall, many found the warning effective for communicating risks to pregnant women. A male participant stated, “I think people have a soft spot for babies and pregnancies. The picture is effective,” while another noted, “I think this one is effective. It’s good to see a message showing that prenatal care includes non-smoking. The picture, I believe, effectively shows a deformity and a baby.” T3-PWL18 (heart disease risk) Participants showed varied reactions—some curious, some indifferent, some prompted to rethink vaping, others unbothered or encouraged. One said, “This is a good one. I like the picture. It makes you feel grossed out,” while another noted, “It looks cool. It’ll make you want to hit a vape right now. Thank goodness I don’t want to because of the information sinking in my head.” Despite differences, most agreed that the message was effective, often due to the new information. One remarked, “I didn’t know that was a heart, but the caption clarifies it. It makes you think, ‘Oh, my God!’ I didn’t know you could get heart disease from that.” T3-PWL20 (cigarette initiation risk) The message did not capture the attention of most participants, as one remarked, “I’m kind of intermediate, but this one I don’t really feel like it’s grabbing my attention, but it’s not the worst.” Reactions varied significantly. Some participants expressed fear, stating, “It makes sense, and that is like a fear—you don’t want people to turn to cigarettes.” Others found it confusing: “It’s not direct enough.” Opinions on its effectiveness were also mixed. One said it was ineffective: “Both are bad, and nicotine makes you think of smoking. This isn’t new or impressive.” In contrast, another participant argued, “I think this one’s effective: It’s relevant now, since people switch to cigarettes due to restrictions, and the picture works well.” T3-PWL21 (industry risk) Many participants found the label’s image appealing. One noted, “It attracts me, but I don’t view the flavors as a negative aspect.” Reactions to the label were clearly divided. A female participant stated firmly, “Even though this isn’t a health message, I encounter it daily, whether it’s me or those around me. It’s undeniably relatable.” Conversely, another male participant asserted, “I would discard it because it’s common knowledge; people already understand this.” Opinions on the label’s effectiveness varied. One woman said, “This is not effective at all. We’re here for the flavors. They know how to market it, and the picture looks like a cartoon.” In contrast, another participant argued that the message is somewhat effective, explaining, “The picture conveys the overall idea effectively, but it risks being too distracting.” Suggestions for improvements Participants emphasized the need for clear, well-defined, and understandable terms to describe depicted risks or diseases in the text (Table 3). One noted, “I have never heard of DNA damage. Explaining how it affects DNA would help.” They also urged assertive language, as in response to T1L3 (glycidol as a possible carcinogen): “Avoid words like ‘possible’ or ‘likelihood’—if not direct, I dismiss it.” For images, participants suggested showing danger in a more relatable way to humans rather than animals. In response to T1L3 (frog in the jar), it was noted, “The picture does not really relate because of the frog; it looks more like a lab experiment.” One participant added: “Maybe a human brain would be more effective than the frog.” Participants recommended using more impactful graphic images of human suffering rather than illustrations of diseased organs. One suggested, “The picture should be more graphic, showing more damage” (T1L3; diseased lung). They also advised using visuals and text to illustrate dramatic scenarios for better risk communication, as in T2L10 (asthma risk), where a participant said, “Depict a more dramatic scenario, she just looks like she’s coughing with her inhaler.”Table 3Proposed enhancements for the Top 12 selected pictorial warning labels as suggested by focus group participants Table 3Proposed enhancements for the Top 12 selected pictorial warning labels as suggested by focus group participants Proposed enhancements for the Top 12 selected pictorial warning labels as suggested by focus group participants Discussion This study advances the development of ENDS PWLs using a theory-driven mixed-methods approach guided by the Message Impact Framework. Consistent with the framework, we examined warning effectiveness as a staged process involving attention, emotional reactions (e.g., fear), cognitive responses (e.g., believability and harm perception), and behavioral intentions. Survey findings demonstrated that toxicity-themed warnings outperformed other themes not only in intention to quit but also across upstream processing indicators, including attention, fear, and perceived harm. These quantitative patterns were reinforced by focus group discussions, where participants described toxicity warnings as “scary,” “real,” and “urgent,” suggesting that strong emotional activation and perceived severity contributed to their higher behavioral impact. Focus group discussions also showed that PWLs resonated more when highlighting personal health suffering, rather than images of damaged organs. Similarly, PWLs that use direct imagery were perceived as more effective than figurative PWLs using visual metaphors. Fear and personal relevance emerged as key factors influencing participants' perceptions of the PWLs’ effectiveness. Because personal relevance varied by individual circumstances (e.g., asthma status, pregnancy, prior health experiences), warnings targeting different risk domains are likely to resonate with different subgroups of users. Accordingly, our findings underscore the importance of implementing a diverse portfolio of ENDS warning labels. A regulatory strategy that rotates multiple evidence-based warnings reflecting distinct health risks may therefore maximize population-level relevance and overall impact. Similar to earlier studies, our study demonstrated that toxicity-focused warnings elicited higher levels of attention, fear, harm perception, and perceived effectiveness ratings than PWLs in the health effect and specific theme [31, 32]. Messages highlighting the presence of harmful chemicals, such as acrolein and formaldehyde, and their detrimental health impacts resonated strongly with our survey participants, and this finding was reinforced during the focus group discussions. This may be because ENDS are relatively new products mainly used by healthy young adults who are curious about their composition and the scientific evidence surrounding potentially harmful exposures associated with their use [33]. Importantly, the survey ratings provided diagnostic insight into why certain warnings were perceived as more effective. For example, some warnings generated high attention but elicited skepticism or confusion, resulting in lower believability and weaker quit intentions (e.g., T2-PWL14: Vape explosion). Others elicited strong fear responses but were perceived as less personally relevant, which tempered their motivational impact (e.g., T2-PWL10: Vaping worsens asthma). In contrast, warnings that combined vivid, literal imagery with clear and assertive language tended to score consistently high across attention, fear, harm perception, and intention measures (e.g., T1-PWL5: Lung damage from acrolein). These findings align with the message impact framework proposition that behavioral intentions are strengthened when emotional and cognitive processing operate synergistically. Focus group results indicated that PWLs were more attention-grabbing when they depicted human suffering associated with serious ENDS-related health consequences, rather than isolated images of damaged organs. For instance, the image of an intubated patient attracted most participants’ attention, while the bladder cancer image was too ambiguous for some to notice, despite the serious nature of a cancer outcome. This suggests in line of previouse research that focusing on personal narratives and the human experience may resonate more with viewers [34, 35]. Additionally, fear and relevance reactions emerged as the most influential factors in determining participants’ perceptions of the warning’s effectiveness. Strong emotional responses, particularly fear, were predominant in the ENDS health risk PWLs theme. For instance, depiction of a disfigured face resulting from an ENDS explosion significantly heightened participants’ concerns for their safety. Furthermore, PWLs associated with severe health outcomes, such as irreversible lung disease and heart disease, also elicited strong feelings of alarm and fear in both men and women. On the other hand, the depiction of scenarios that participants have encountered or expect to encounter resonated with participants as relevant. For instance, PWLs that show a disease participants have experienced, like asthma, were seen as effective because they serve as a reminder to avoid vaping. These reactions stem from vulnerability. Warnings about vaping risk to children were seen as targeting demographics like married women, men, or parents, not well represented in our sample. PWLs focused on nicotine addiction and tobacco industry tactics drew less concern and engagement from participants. This aligns with a US study of 1,629 young adults, which found that messages about nicotine addiction caused little harm consideration and minimal discouragement from ENDS use [13]. Similar to findings in Rohed et al., these results stress the need to include diverse themes aligned with ENDS risks, beyond just the FDA’s warning about nicotine addiction, to engage various segments of the population [36]. Stratification by sex provided additional insight into how young adults responded to the ENDS PWLs. Differences were most evident for pregnancy-related warnings, which were perceived as highly relevant and emotionally impactful among female participants. In contrast, male participants often framed these messages as important but less personally relevant. Personal relevance also influenced responses to condition-specific warnings (e.g., asthma), with stronger engagement among those who identified with the risk. In contrast, reactions to toxicity- and severe disease-focused warnings (e.g., intubation, lung disease) were largely consistent across males and females, particularly in eliciting fear and perceived severity. These findings suggest that while graphic depictions of serious health consequences may resonate broadly, certain themes may differentially engage subgroups, underscoring the value of incorporating diverse warning approaches to maximize relevance and impact. For improvements, participants preferred clear and easy-to-understand visual representations and simple verbal explanations. This highlights what is already known: that comprehension relies on explicit, literal representations of health outcomes rather than figurative or implicit ones, as reported in Thrasher et al. [37]. Additionally, in line with earlier health communication studies, providing direct verbal descriptions of disease severity is crucial because it helps improve understanding and promotes careful evaluation of health risks [38]. This level of clarity is particularly critical when presenting new information, such as the connection between heart disease and dual use of ENDS and cigarettes. Similar findings were reported by Asfar et al. regarding the communication of unknown risk associated with waterpipe use to smokers [5]. On the other hand, PWLs lacking clear imagery or text were rejected as confusing, uninformative, and unbelievable. Such reactions arose specifically from visual depictions that relied on figurative representations—using connotations or metaphors—rather than clear, literal imagery (e.g., a girl with a USB port in her mouse as a symbol of nicotine addiction). Research consistently shows that clear messaging is not only helpful for understanding but also crucial for motivating effective communication action [39, 40]. Evidence suggests that utilizing clear messages not only makes warnings more accessible to a wider audience but also improves the perception of the PWLs’ source [17, 39]. This approach fosters greater trust and credibility, which ultimately enhances the effectiveness of the communication. This study is the first to engage the target population in early development of ENDS-specific PWLs, using a mixed-methods approach guided by a theoretical health communication framework. However, as with other focus group studies, the findings are based on a relatively small and geographically limited group of ENDS users, which may not capture the full diversity of ENDS users across regions, cultures, or regulatory contexts. Nonetheless, participants reflect the primary target population for whom these PWLs are being developed, and thus provide directly relevant insights to guide optimization of warning content and design. Additionally, the quantitative ratings were derived from a small, non-random convenience sample and were intended to complement the qualitative findings rather than provide population-level estimates. Therefore, results should be interpreted cautiously and viewed as exploratory. Conclusions This study identified promising PWL for young adult ENDS users and offered practical guidance for improving warnings. These refinements aim to enhance health communication and support regulatory efforts to reduce ENDS use. Collectively, the convergence between quantitative component ratings and qualitative feedback strengthens confidence in our findings and clarifies the mechanisms underlying warning effectiveness. Warnings that elicited higher fear and perceived severity were consistently described as “frightening,” “authentic,” and motivating reconsideration of use, whereas figurative or metaphor-based images were often viewed as confusing or less believable. These results underscore the value of assessing intermediate message responses, such as attention, emotional reactions, and perceived harm, alongside behavioral intentions to better understand how PWLs influence behavior. In addition, PWLs emphasizing clear, literal imagery and human health consequences enhance resonance by increasing relevance and emotional impact. Therefore, incorporating diverse, evidence-based ENDS risk themes can further broaden engagement across subgroups. Supplementary Information Supplementary Material 1. Supplementary Material 1.