WHAT IS ALREADY KNOWN ON THIS TOPIC

WHAT THIS STUDY ADDS

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Our primary research questions were:

We also explored the following:

We had no specific a priori hypotheses for research question 3, so the analysis of potential moderators should be considered exploratory.

This study was a randomised experiment whereby participants were randomly allocated (at a 1:1 ratio) to either the branded or standardised condition, stratified by and balanced within each country (ie, randomised without replacement, with randomisation sequence calculated via the survey programming software). This experiment was embedded within the 2023 wave of the online ITC Youth and Youth Adult Tobacco & Vaping Survey. Thus, the sample size was determined based on the number of participants recruited in this survey, rather than planned specifically for this experiment. Technical reports are available online (https://davidhammond.ca/projects/tobacco-vaping/itc-youth-tobacco-ecig/).

Participants were aged between 16 and 29 years old, residing in England, Canada or the US. Participants were recruited via the Nielsen Consumer Insights Global Panel and partner panels. Participants received remuneration in accordance with their panel’s usual incentive structure, which could include points-based or monetary rewards (redeemed for catalogue items, as cash or donated) and/or opportunities to win monthly prizes.

Participants were randomly allocated to either a standardised (experimental) condition or branded (control) condition. Those randomised to the standardised condition viewed a set of four images of vaping devices with standardised colour (white) and text (brand name and flavour in uniform typeface (eg, upper images in figure 1). Those in the branded condition viewed a set of four images of vaping devices with branding and colours as they are currently sold in England (eg, lower images in figure 1). Then, they were shown one of the four vapes within their condition at random and asked to rate its harmfulness (see specific wording in ‘Outcome measures’ section).

Products from popular brands were chosen for the experiment,1 with a range of colours to partially reflect the array of disposable products currently available on the market. Fruit flavours were chosen as they are the most popular among youth.26

Participants were assigned to either only view standardised devices or only view branded devices for two reasons. First, this prevents contamination effects that could occur if participants were exposed to both standardised and branded vapes simultaneously. Second, this meant the experiment more closely captured the choice that an individual would make if regulations were introduced, where all vapes on display would be standardised.

A standardised colour of white was selected because it is more neutral than most branding but distinct from the green/brown (Pantone 448 C) required for tobacco packaging in many countries including England and Canada, thereby intending to reflect risk differences in the use of cigarettes and vapes.

Interest in trying. Respondents were asked “Which of the following vapes would you be interested in trying?” with options to select any of the four vapes displayed within each condition (including multiple) with response options “I have no interest in trying any of these products”, “Don’t know” or “Refused”. For the primary analyses, this was coded as no interest in trying versus otherwise (ie, selected at least one of the vapes displayed or “Don’t know”). In a sensitivity analysis, we used the following ordinal scale, where higher scores indicate greater interest: 1=No interest in trying, 2=“Don’t know”, 3=Selected one device, 4=Selected two devices, 5=Selected three devices, 6=Selected all four devices. In both analyses, we excluded individuals who selected “Refused”.

Harm perceptions. Participants were shown one of the products within their condition at random and were asked “How harmful do you think it is to vape this product?” with response options “Not at all harmful”, “Harmful, but less harmful than smoking cigarettes”, “As harmful as smoking cigarettes”, “More harmful than smoking cigarettes”, “Don’t know” or “Refused”. Current evidence suggests that vaping is less harmful than smoking,27 28 so in the primary analysis responses were dichotomised as less harmful than smoking (ie, “Not at all harmful” or “Harmful, but less harmful than smoking cigarettes”) versus otherwise (including “Don’t know” but excluding “Refused”). In a sensitivity analysis, we dichotomised responses as those who perceived e-cigarettes as “Not at all harmful” versus otherwise (excluding “Refused”) to examine whether standardisation affected the proportion of participants who inaccurately viewed vaping as entirely harmless.

The full questionnaire is available online.29

Age group. Current age in years was categorised into three groups (16–17, 18–19 and 20–29 years).

Sex at birth. Sex at birth (male/female) was reported by participants, or inferred from reported gender in the small number of cases where male or female sex at birth was not specified.

Race/ethnicity. Participants were asked about their race/ethnicity using country-specific measures based on Census questions in England, Canada and the US. For a common measure, we collapsed country-specific options to examine differences between White (the majority group) and other groups (other and mixed ethnicities, “Don’t know” and “Refused”).30 Sample sizes did not permit detailed analysis of specific minority ethnic groups.

Past 30-day vaping/smoking status. Participants were asked separate sets of questions about smoking cigarettes and using e-cigarettes/vaping, beginning with “Have you ever tried [an e-cigarette/vaped OR cigarette smoking], even one or two puffs?”. Those who responded “Yes” were then asked “When was the last time you [used an e-cigarette/vaped OR smoked a cigarette], even one or two puffs?”. Responses were categorised into five mutually exclusive use subgroups, aligning with previous studies21: (1) past 30-day exclusive vaping (ie, vaped but did not smoke in the past 30 days); (2) past 30-day exclusive smoking (ie, smoked but did not vape in the past 30 days); (3) past 30-day use (ie, both vaped and smoked in the past 30 days); (4) ever but not past 30-day vaping and/or smoking (ie, ever vaped and/or smoked, but not in the past 30 days) and (5) never vaped or smoked (never vaped and never smoked).

Perceived family income adequacy. Participants were asked “How would you describe your family’s financial situation?”. Responses were ordered on a four-point scale: 0=“Not meeting basic expenses”, 1=“Just meeting basic expenses”, 2=“Meeting needs with a little left over” and 3=“Living comfortably”. “Don’t know” and “Refused” responses were included as a separate category.

Country. England, Canada, US.

The following analysis was pre-registered prior to data analysis, and analysis code is openly available (https://osf.io/yfbj3/). Participants with missing outcome data (ie, “Refused” to respond) were excluded from the analytic sample. All analyses were conducted in R Version 4.2.3. We present sample characteristics overall and stratified by study condition and country.

For research question 1, we ran a logistic regression predicting no interest in trying, with study condition included (standardised vs branded) as an independent variable. We reported the percentage of participants in each condition with the outcome, and risk ratios (RRs) with 95% confidence intervals (CIs) showing how much more or less likely individuals were to have the outcome in the standardised versus branded condition. Primary models were adjusted for vaping/smoking status, demographics (age group, sex, race/ethnicity, perceived income adequacy) and country. For research question 2, we repeated these analyses with vaping harm perceptions as the outcome.

For research question 3, we repeated the above adjusted (for the same covariates listed previously) models for each outcome, but included interaction terms between condition and age (eg, age*condition), sex, vaping/smoking status, perceived income adequacy and country. Each of these potential interactions with the study conditions was tested in a separate model. We reported the interaction as a ratio of risk ratios (RRRs) alongside 95% CIs. We also reported the RRs and 95% CIs for the intervention effect in each subgroup.

Sensitivity analyses We repeated the analysis for research question 1, using the outcome of interest in trying vapes coded as ordinal, as described earlier. Ordinal logistic regression was used to predict interest in trying (one- to six-point ordinal scale) from study condition (standardised vs branded). We reported odds ratios (ORs) with 95% CIs for the standardised versus branded condition. We repeated analyses for research question 2, with the outcome coded as perceiving the vape displayed as not harmful at all (vs otherwise).

Several effects (eg, adjusted models for interest in vaping) could not be estimated using log-binomial regression because the models did not converge. Therefore, we instead calculated marginal risk ratios using logistic regression alongside the marginal effects package.

Of the 15 479 young people surveyed, 15 259 (98.6%) had complete data on both outcomes, with 188 refusals for interest in trying (75 in branded and 113 in standardised condition) and 50 refusals for harm perceptions (27 in branded and 23 in standardised condition). There were no missing covariate data. Of participants who provided complete outcome data, a total of 7638 were randomised to the branded condition and 7621 to the standardised condition. The CONSORT flow diagram is available in online supplemental figure 1. Sample characteristics are shown in table 1 (by country in online supplemental table 1).

Standardisation increased the percentage of participants reporting no interest in trying at least one of the vapes displayed: 67.1% (5112/7621) of participants randomised to view the white standardised devices reported no interest in trying any of the devices shown, compared with 62.8% (4797/ 7638) of those in the branded condition. In the primary covariate-adjusted analysis, standardisation increased the likelihood of reporting no interest in any of the vaping products by 12.7% (adjusted risk ratio (ARR) 1.127, 95% CI 1.085 to 1.169).

Figure 2 shows the effect of device standardisation within subgroups (more details of these moderation effects are provided in table 2). Standardisation had an effect on all vaping and smoking subgroups, but it was smallest in those who had never vaped or smoked: 92.9% (3182/3424) of those randomised to view standardised white devices reported no interest in trying the vapes compared with 91.2% (3121/3422) of those in the branded condition (ARR 1.020, 95% CI 1.005 to 1.035). Standardisation also had a smaller effect in younger participants compared with older participants (as shown in figure 2 and table 2).

In sensitivity analyses using an ordinal outcome, interest in vaping was lower in the standardised condition compared with the branded condition (adjusted odds ratio (AOR) 0.747, 95% CI 0.693 to 0.804) as shown in online supplemental figure 2.

Standardisation did not have a large effect on participants’ harm perceptions of the vaping product displayed: 31.2% (2377/7621) of those randomised to the standardised condition viewed the vape as less harmful than smoking compared with 32.7% (2496/7638) in the branded condition. In primary covariate-adjusted models, the effect of standardisation on harm perceptions was non-significant, but we could not rule out the possibility of small effects (ARR 0.960, 95% CI 0.919 to 1.001). Table 3 shows that there were no clear moderators of the effect of standardisation on harm perceptions.

In sensitivity analyses, similarly low proportions of participants in both conditions perceived the displayed vaping device as not at all harmful: 2.57% (196/7621) in the standardised versus 2.19% (167/7638) in the branded condition (ARR 1.176, 95% CI 0.946 to 1.407). The full distribution of harm perceptions is provided in online supplemental table 2.

Sensitivity analyses were also conducted examining moderation effects when all two-way interactions were included in a single model; results using this approach were analogous to those when examining each moderator in a separate model (online supplemental tables 3 and 4).

As hypothesised, in this randomised experiment of 15 259 16–29-year-olds in England, Canada and the US, we found that standardising the colour and branding of disposable vaping devices increased the percentage who were not interested in trying any of the vaping products from 63% to 67%, with minimal adverse impact on harm perceptions. Standardisation increased those uninterested in all vaping and smoking subgroups, but the greatest impact was among those who had smoked and/or vaped in the past 30 days. For example, standardisation increased the percentage with no interest in trying any vape from 38% to 48% among those who had smoked but not vaped in the past 30 days, individuals who would gain from switching entirely from smoking to vaping. Among those who had never smoked or vaped, standardisation also increased those not interested in trying, although to a lesser extent. The vast majority of those who never smoked or vaped reported no interest in trying the vapes, whether they were branded (91.2%) or standardised (92.9%). Standardisation also increased those uninterested in those who exclusively vaped in the past 30 days (16.4% reported no interest in trying the branded devices vs 19.9% for the standardised devices), a group who would benefit from stopping vaping. The effect of standardisation on reducing interest in vaping was also smallest in the youngest age group, among whom most had never vaped or smoked. There was no clear evidence of differences in the effect of standardisation by sex, country (Canada, England, US) or perceived income adequacy.

Our findings are broadly consistent with prior literature on e-cigarette product packaging that suggests that standardisation reduces appeal among young people,21–23 31 and similar to literature on tobacco cigarettes showing that standardised cigarette sticks are less appealing to youth.25 However, inconsistent with prior work,22 we found greater effects of standardisation on interest among young adults compared with youth and among participants who vaped and/or smoked compared with those who had never done so. These differences may reflect differing samples; prior work has compared youth aged 11–18 years with all adults aged 18+ years while we compared 16–17-year-olds with young adults aged 18–29 years. Finally, we only examined effects of standardisation among youth and young adults. It is possible that the impact of standardisation would not generalise to other groups; it will be important for subsequent studies to examine effects in older age groups, especially those who smoke.

The smaller effect identified among people who have never vaped or smoked does not necessarily mean that standardisation will fail to discourage uptake; there may have been a ceiling effect associated with the outcome measure, given that over 90% of never-users reported no interest in any of the vaping products presented. Future studies could usefully look at the impact on later uptake or use measures that capture varying levels of interest.

Importantly, our findings suggest little to no effect of standardising devices on relative harm perceptions. Relative harm perceptions of vapes are often inaccurate among youth and adults, and this can influence the likelihood of someone using a vape to quit.13 32 33 Therefore, it is promising that standardising the appearance of devices to be white and unbranded would likely reduce appeal without unintentionally exacerbating misperceptions (although there may have been greater impact on harm perceptions if the colour of standardisation was brown not white).34

In January 2024, the UK government announced plans to introduce new measures to ensure that e-cigarette manufacturers implement standardised packaging, but it remains unclear exactly what standardised packaging would look like.20 Our findings suggest that the integration of regulation on device design into new policy further reduces the appeal of vapes to young people. However, compared with people who have never smoked or vaped, the reduction in interest was more pronounced among smokers, who might benefit from using vapes to quit smoking. There is a risk that the public health benefits of preventing youth uptake of vaping could be offset by a decline in the number of young people transitioning from smoking to exclusive vaping, or an increase in relapse from exclusive vaping to smoking, including dual use.10 35 Future research should explore how standardisation impacts these transition pathways between exclusive smoking, exclusive vaping and dual use. Nonetheless, if legislation on standardised vaping devices is introduced, any unintended consequences could be mitigated through complementary policies aimed at promoting smoking cessation and preventing relapse among former smokers.

Strengths of this study include the large sample size, to precisely estimate the effect of standardisation overall and across important subgroups (eg, vaping/smoking status), use of a randomised controlled experiment to avoid confounding, and consistency of results across three countries with differing vaping markets and regulations.

There were also limitations. First, as mentioned earlier, our binary measure of interest (no interest vs other) in vaping may have been subject to a ceiling effect, making it insensitive to detecting differences in appeal of vapes to people who have never vaped or smoked. Future studies should replicate results using a wider array of measures, including — if possible — measures that capture people’s automatic cognitive processing of packaging (eg, eye tracking) as well as conscious reporting of interest and appeal. Second, interest in vaping was self-reported, yet people do not have perfect insight into the factors of products that drive them to use them.36 Third, an online experimental paradigm lacks ecological validity1 37; it may not reflect an individual's interest if they encountered these products in a more naturalistic setting, such as when offered by a friend or viewing a product in a store among a wider range of competing products. However, at least for cigarettes, experimental studies and real-world evaluations of the impact of standardised packaging have found highly consistent results.16 38 Fourth, the low rates of reporting no interest among people who both smoke and vape may have limited our power to detect any genuine difference between this group and people who have never smoked or vaped. Fifth, different provinces and countries have different ages of sale for vaping products. Future research could examine what effect this has on interest in vaping. Finally, the wording of the interest in trying vaping measure may have influenced responses by prompting participants to report interest in trying at least one product. However, the low interest rates (<10%) among people who have never smoked or vaped suggest that participants were comfortable reporting no interest. As this potential demand effect would have applied equally to both study conditions, it is unlikely to have biased the observed differences between standardised and branded packaging.

Consent obtained directly from patient(s).

This study involves human participants and received ethical approval from the University of Waterloo Research Ethics Board (ORE#31017) and the King’s College London Psychiatry, Nursing & Midwifery Research Ethics Subcommittee. All participants were provided with study information and gave informed consent to participate in the study before taking part.