Article Text

Download PDFPDF

Long-term impact of plain packaging of cigarettes with larger graphic health warnings: findings from cross-sectional surveys of Australian adolescents between 2011 and 2017
  1. Victoria M White1,2,
  2. Nicola Guerin2,
  3. Tahlia Williams2,
  4. Melanie A Wakefield2
  1. 1 Psychology, Deakin University, Burwood, Victoria, Australia
  2. 2 Centre for Behavioural Research in Cancer, Cancer Council Victoria, Melbourne, Victoria, Australia
  1. Correspondence to Dr Victoria M White, Psychology, Deakin University, Burwood 3125, Victoria, Australia; vicki.white{at}deakin.edu.au

Abstract

Objective To assess the long-term impact of plain packaging (PP) of cigarettes with larger graphic health warnings (HW) introduced in December 2012 on adolescents’ relevant tobacco-related perceptions.

Methods Cross-sectional school-based surveys of 12 to 17 year olds in 2011 (n=4413), 2013 (n=4423), 2014 (n=4576) and 2017 (n=4266). Students rated the character of four popular cigarette brands, indicated their agreement regarding brand differences in smoking ease, quitting, addictiveness, harmfulness and pack attractiveness and positive/negative perceptions of pack image. The frequency of students reading, attending to, thinking and talking about HW was assessed. Responses of students seeing cigarette packs in the previous 6 months (2011: 63%; 2013: 67%, 2014: 56%, 2017: 56%) were examined.

Results Smoking prevalence declined from 2011 to 2017. Among students who had recently seen packs, cigarette packs were rated less positively and more negatively in 2017 than in 2011 (p<0.001) with ratings similar between 2013 and 2017. Positive character ratings for each brand reduced between 2011 and 2013 (ps<0.05) with further reductions between 2013 and 2017 (ps<0.05). Fewer students agreed, and more were uncertain, that brands differed in their smoking ease, addictiveness, harmfulness and pack attractiveness in 2017 than 2011. The frequency of students reading, attending, talking or thinking about HW did not change between 2011 and 2017.

Conclusions PP’s initial impact in reducing adolescent’s positive perceptions of cigarette packs and brand differences continued in the following years with tobacco packaging less appealing to young people in 2017 than 2011 and students more uncertain about brand differences.

  • Smoking adolescents
  • plain packaging
  • tobacco products
  • standardised packaging
  • graphic warning labels

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

By December 2018, six countries had followed Australia’s lead in implementing legislation to standardise tobacco packaging. In these countries, plain packaging (PP) has mandated a standard pack colour, text size, font and position.1 All countries require text and pictorial health warnings (HW) to be displayed on the front and back of packs, although the size, type and positioning of HW differ between countries. The introduction of PP in Australia in late 2012 saw text and associated photographic warnings increase in size from 30% to 75% of the front of packs with no change to HW size (90%) on the back.2 3

Underlying the introduction of PP was evidence suggesting branded imagery on cigarette packaging promoted the appeal and use of tobacco and reduced health concerns.4 5 Tobacco industry documents show how branding is designed to communicate brand attributes, including taste, tobacco quality and image to consumers.5–7 Tobacco packaging shape, colour and livery influences the attractiveness of brands and perceptions of brand users.8–11 Pack colours and brand descriptors also influence harm perceptions, with light-coloured packaging and the descriptors ‘light’, ‘mild’ and ‘smooth’ associated with perceiving less harm.12 13 These packaging effects have also been reported for young people.14–17 Stripping explicit branding from cigarette packs and standardising their appearance should reduce the ability of packaging to promote smoking10 16 18 19 including to young people.

Due to the novelty of the policy, few studies have examined adolescents’ responses to real-world PP introduction. An Australian study found greater support for PP 6 months after implementation than before.20 Another study found less positive ratings of both pack and smokers’ image, for adolescents surveyed 6–12 months after PP implementation than before.21 A recent French study showed adolescent smokers were less attached to their cigarette brand and less likely to think their own brand was less harmful than other brands after PP implementation.22 However, an Australian study found the implementation of PP with larger HW had no immediate impact on adolescents’ attention and processing of HWs.23

The longer term impact of PP with larger HWs on adolescents’ perceptions of packaging or specific brands is unknown given policy implementation is so recent. Adolescents and adults habituate to pack changes.24 25 While positive pack image ratings declined and cognitive processing of HWs increased soon after text warnings changed to pictorial HWs covering 30% of the front of fully branded packs in Australia in 2006, both returned to prepictorial HW levels after 5 years.24 Changes to pack designs made by tobacco companies after the introduction of these HWs, including matching pack colours to minimise the visual impact of HW,26 are likely to have influenced these findings. As tobacco companies can no longer revise their pack designs, the initial effect of PP with larger HW may be maintained.

The current study used data from cross-sectional surveys of adolescents conducted before PP introduction and up to 5 years later to investigate the long-term impact of PP on adolescents’ perceptions of cigarette packs and brands, differences in characteristics of brands and cognitive processing of HW. To examine whether smokers and non-smokers responded similarly to PP with large HW, we stratified analyses by adolescents’ smoking status.

Methods

Study overview and design

We used data from cross-sectional school-based surveys of adolescents conducted in 2011 (before PP introduction), 2013 (approximately 6–12 months post-PP introduction), 2014 (18–24 months postintroduction) and 2017 (54–60 months postintroduction). All surveys had institutional ethics approval and relevant school authorities’ approvals and followed the same procedure as described previously21 23 and briefly described here.

School samples

Data from 2011, 2014 and 2017 were collected in the state of Victoria as part of the triennial Australian School Students Alcohol and Drug survey from a representative sample of students aged 12 to 17 years. Proportionally representative school samples were randomly selected from government, Catholic and independent education sectors. After obtaining appropriate approvals, we approached schools by mail and invited their participation. If a school declined, we approached the geographically closest school within the same education sector. We worked with a contact teacher in participating schools to identify classes of students from lower (years 7–10) or upper (years 11–12) year levels for surveying. To avoid selection bias, elective or specialist classes were excluded. School response rates were around 30%. We surveyed students between June and December, aiming to survey 80–100 students per school (students per school, M2011=73; M2014=97; M2017=89).

As described previously,23 in 2013 we aimed to survey students from schools participating in the 2011 Victorian survey. When a school declined, we recruited a replacement from the list identified for the 2011 survey. In all, 82 schools participated in 2013. Data were collected from June to November 2013 with an average of 65 students participating per school.

Survey administration

Research assistants attended schools to administer a pencil-and-paper questionnaire to students. Research assistants worked through a practice survey with students, before participants anonymously completed the main survey.

Measures

Questions were included in a larger survey assessing use of tobacco, alcohol and illicit substances (2011, 2014, 2017) or tobacco and alcohol (2013).

Pack awareness

Students indicated when they last saw a pack of cigarettes (response options: ‘within the past 6 months’, ‘more than 6 months ago’ or ‘never’).

Pack image ratings

As previously described,24 27 students who had seen a cigarette pack in the previous 6 months indicated their agreement to four positive (‘cool’, ‘good’, ‘interesting’, ‘exciting’), and four negative (‘ugly’, ‘daggy (uncool)’, ‘gross’, ‘disgusting’) pack descriptors using a 5-point scale (1, ‘strongly disagree’ to 5, ‘strongly agree’) or indicated ‘cannot comment’ (coded as missing). Positive and negative subscales were created by averaging ratings for the four corresponding items.27 Internal reliability for the current study remained high (eg, positive pack image: 2014 α=0.85, 2017 α=0.85; negative pack image: 2014 α=0.76; 2017 α=0.77)

Brand character ratings

Students were shown photographic images of four Australian cigarette brands and indicated their agreement with six statements for each: three about the brand and pack (eg, ‘this brand appeals to me’) and three relating to smokers of the brand (eg, ‘are cool’). Responses were made on a 5-point scale (1. ‘strongly disagree’, 3. ‘not sure’ 5. ‘strongly agree’). The three mainstream brands most commonly smoked by Australian adolescents in 2011 were assessed (Winfield: 44%; Peter Jackson: 25%; Longbeach 10%) along with a common premium brand (Benson & Hedges: 8% of adolescent smokers).28

Within each survey year, pack images included the same graphic HW from those mandated at that time (eg, 2011: ‘Smoking causes mouth and throat cancer’; 2013, 2014, 2017: ‘Smoking causes mouth cancer’). Responses were coded so higher scores indicated positive views and summed (range: 6–30). Cronbach’s alpha for each brand in 2014 and 2017 were similar to those in 2011 and 201321 (2014: 0.77–0.79; 2017: 0.75–0.77).

Brand differences

Students indicated whether they agreed (strongly agree, agree) or disagreed (strongly disagree, disagree) or did not know whether some brands of cigarettes are: ‘easier to smoke’, ‘more addictive’, ‘easier to quit’, ‘have more harmful substances in them’ and ‘have better looking packs’ than others.21

Cognitive processing of HWs

Those who had seen a cigarette pack in the previous 6 months indicated how frequently they: read; paid close attention to; thought about and talked about, the warning labels, using a 5-point scale: (1) ‘never’; (2) ‘once or twice’; (3) ‘sometimes’; (4) ‘often’ and, (5) ‘every time I see them’. Mean scores for each item are reported.

Student level variables

Smoking status: Following previous work, we classified students into four levels of smoking status24 27: non-susceptible never smokers (NSNS) (had never smoked a cigarette, not even a puff and certain not to smoke in the next year); susceptible never smokers (SNS) (never smokers, not even a puff, who were not certain they would not smoke in the next year); experimental smokers (ES) (had smoked at least a puff of a cigarette but not in the previous month) and current smokers (had smoked in the past month).

Demographic variables: students indicated their sex and age and provided their residential postcode which was used to determine socioeconomic status (SES) through a postcode-related index of relative SES disadvantage determined by the Australian Bureau of Statistics.29 Scores were recoded to represent three groups (0%–40%—most disadvantage; 41%–80%—moderate disadvantage; 81%–100%—least disadvantage). Education sector of the school was recorded.

Statistical analyses

Population weights were used to bring the sample distribution into line with school enrolment data for Victoria.30 Univariate χ2 tests examined sample similarity across years.

Linear regression models tested differences in mean scores on pack image and cognitive processing variables between survey periods. Logistic regression analyses compared the proportion of adolescents in each smoking status level across time periods, with sex, age and education sector controlled and survey year entered as a categorical variable in analyses. Logistic regression also compared proportions indicating agree or strongly agree to the five questions assessing brand differences across survey periods. In regression models, the predictor was a categorical indicator of survey period, and age, sex, school sector and smoking status were controlled. We included the survey period×smoking status interaction term as a fixed effect to evaluate differences in the survey period effect by smoking. To examine whether responses after PP implementation differed from those before PP, 2011 was the comparison year.

Multilevel mixed effects regression examined differences in brand characteristic ratings between brands and across years and tested the interaction between year and brand. Models adjusted for age, sex, education sector and smoking status. If the main effects of year and brand and their interaction were significant, post-hoc regression analyses were conducted for each brand to examine pattern of change over time. All models adjusted for clustering of students within school, and SEs robust to potential non-independence of students were obtained. Analyses were conducted using the statistical package STATA V.15.31

Results

The numbers of students surveyed and their characteristics are shown in table 1. When data were weighted, the proportions of males, those aged 12–15 years and those attending government schools were similar across survey years. However, the proportion of students surveyed from the lowest SES groups was slightly lower in 2014 and 2017 than in the earlier years.

Table 1

Characteristics of students participating in each survey in unweighted and weighted data sets

Over the study period, the proportion of students classified as NSNS increased and was greater in 2013 (p<0.001), 2014 (p<0.001) and 2017 (p<0.001) than in 2011 (table 2). Additionally the proportion of experimental smokers (<0.001) and past-month smokers (p=0.014) declined between 2011 and 2017 (table 2).

Table 2

Proportion of students in each survey year in each of the smoking status levels

The proportion of students who had seen a cigarette pack in the previous 6 months was lower in 2017 (56%) than 2011 (63%) or 2013 (67%, p<0.001) (table 1). All remaining analyses focused on responses from students who had seen a cigarette pack in the previous 6 months.

Positive pack image ratings were lower in 2013, 2014 and 2017 than in 2011 (p<0.001, table 3). However, mean ratings in 2017 were similar to 2013 and 2014 (table 3). Positive pack image ratings varied by smoking status (p<0.001) with current smokers having more positive perceptions than NSNS. A significant interaction between survey year and smoking status (p=0.04) was found, reflecting a lack of significant change in image ratings for SNS between 2011 and 2014 compared with the decreases in other smoking status groups. Negative pack image ratings increased over the study period (p=0.04), with ratings significantly more negative in 2017 than in 2011 (p=0.02). However, negative image ratings for 2013 and 2017 did not differ (table 3). In all years, students who had smoked reported less negative perceptions of cigarette packs than NSNS (p<0.001).

Table 3

Mean* pack image and brand characteristic ratings for leading cigarette brands by smoking status and survey year among students who had seen a cigarette pack in the past 6 months

Perceptions regarding brand characteristics became less positive over the study period (table 3). Combining data across brands and years, multilevel analyses showed significant main effects of brand (p<0.001), year (p<0.001) and their interaction (p=0.018). In separate analyses for each brand (table 3) mean scores were significantly lower in 2013, 2014 and 2017 than in 2011, with the greatest effect found for Benson & Hedges (premium brand) and the smallest effect for Longbeach (a mainstream brand). For all brands, 2017 means were significantly lower than 2013 and 2014 (table 3). Except for Longbeach interactions between smoking status and year were of borderline significance (table 3) and were consistent with smaller decreases for SNS than other groups.

Table 4 shows the proportions of students agreeing, disagreeing or indicating ‘do not know’ to the five brand difference statements. The distribution of responses changed significantly between 2011 and 2017 with a general pattern of decreasing agreement and increasing uncertainty regarding the statements over the study period. For all statements, the proportion of students agreeing in 2017 was significantly lower than in 2011. For ‘some brands have better looking packs than others’, the proportion agreeing in 2013 and 2014 were also significantly lower than in 2011. Additionally, post-hoc analyses found significantly higher proportions of students reporting ‘do not know’ in 2013 and 2017 than in 2011 (all ps <0.001). This pattern was similar across smoking status groups. However, as shown in figure 1, the proportions agreeing differed between non-smokers (NSNS and SNS) and smokers (experimenters, current), with a significantly lower proportion of non-smokers agreeing in 2017 than in 2014, while no significant difference in agreement was observed for smokers (year×smoking status interactions; table 4).

Figure 1

Per cent of students agreeing to the brand differences items by smoking status and year (only those items with significant year×smoking status interaction shown), among students who had seen a cigarette pack in the past 6 months. NSNS, non-susceptible non-smoker; SNS, suceptible non-smoker; experimenter and current smoker—smoked in past month.

Table 4

Response proportions (agreeing*, disagreeing or do not know) for statements regarding cigarette brand differences in ease of smoking, ease of quitting, addictiveness, harmfulness and have pack attractiveness by year in students who had seen cigarette packs in the past 6 months.

There were no significant differences between survey years in average levels of attention to, reading or talking about HWs (table 5). However, mean scores for thinking about HWs decreased between 2011 and 2017 (p=0.04). NSNS were significantly more likely to report paying close attention to packs or thinking about packs than current smokers (table 5), but there was no other effect of smoking status on cognitive processing.

Table 5

Means and standard errors for frequency of reading, attending to, thinking about, and talking about health warning labels in students who had seen a cigarette pack in the previous 6 months in each survey year by smoking status (means adjusted for age, sex and school system)

Discussion

Australia’s Tobacco Plain Packaging Act aimed to reduce the appeal of tobacco products, increase the effectiveness of health warnings and reduce the ability of packaging to mislead consumers about the harms of smoking.32 Introduced as part of a comprehensive tobacco control programme, the legislation was intended to contribute to reduced smoking uptake in the long term.32 This is the first study to examine adolescent’s responses to PP during the 5 years following its implementation in any nation. We focused on responses from students who had seen cigarette packs in the previous 6 months, as questions about pack images and cognitive processing of HWs were most relevant to this group. The frequency of adolescents attending to, reading or talking about HWs on cigarette packs did not change significantly between 2011 and 2017. However, negativity towards cigarette packs increased and characteristics associated with different cigarette brands were less positive in 2017 than in 2011. Fewer adolescents agreed that brands differed in addictiveness, harmfulness or ease of smoking in 2017 compared with 2011. However, rather than disagreeing, more adolescents became uncertain, with over 50% reporting ‘do not know’ to our five statements in 2017. Our results suggest that 5 years after its introduction, the effect of PP in reducing the appeal of cigarette branding and perceptions of brand differences in addictiveness and harm to Australian adolescents has been maintained.

Our study showed that the increase in negative and reduction in positive pack image ratings observed soon after the introduction of PP were maintained over the longer term. This is encouraging as our previous work found that when branding could still appear on cigarette packs, immediate declines in adolescents’ positive pack image ratings associated with the change from text to small pictorial HWs in 2006 were not maintained in the long term.24 These contrasting results underscore the importance of restricting the tobacco industry’s ability to redesign packs to camouflage regulated pictorial HWs. Our finding that cigarette packaging appeal remained low 5 years after the introduction of PP with larger pictorial HW extends experimental evidence that PP reduces the appeal of cigarettes to adolescents.16–18 33

Studies show that brief exposure to cigarette packaging colours, imagery and descriptors influence immediate perceptions of the harm associated with smoking different brands.12 18 33 Findings from a recent French study found that within a year of PP introduction, adolescent smokers were less likely to perceive their brand as less harmful than other brands.22 Our earlier study found Australian adolescents expressed more uncertainty about the harmfulness and addictiveness of cigarettes after the introduction of PP than before, although this change was not significant.21 By 2017, agreement levels had decreased significantly. However, rather than greater belief that cigarette brands did not differ on different characteristics, our study suggests that adolescents are more uncertain about whether, and how, brands differ. One potential reason for this uncertainty is that the increasingly large number of brands and variants on the market34 implicitly suggests variation or difference between brands. However, without the visual cues once provided by pack designs, adolescents may now feel they have insufficient information to form an opinion about brand differences. The diminishing number of smokers among adolescents may also contribute to uncertainty, with adolescents increasingly less likely to receive cues about brand differences from their smoking peers.

Overall, our results suggest that 5 years after its introduction, PP is still working to reduce the misleading effect of tobacco branding on adolescents’ perceptions of brand differences, including relative harm and addictiveness. However, some caution is needed as there was no difference in the perceptions of smokers between 2014 and 2017. While our findings need confirmation, these results might indicate that experiences with smoking are now influencing perceptions of brand differences. Novel cigarette features (eg, different filters, menthol flavour capsules and new menthol variants) introduced by tobacco companies may be influencing smokers’ perceived brand differences.

Similar to our previous study,23 we found no substantial change in adolescents’ cognitive processing of HW over the study period. These findings confirm and extend our earlier conclusion that increasing HW size on cigarette pack fronts does not elicit greater adolescent cognitive engagement. This result is also in line with findings from a 2017 cross-sectional study of Australian adolescents which reported that HWs on cigarette packs no longer had cut through with this population.35 However, studies of adult smokers found changes in packaging and HW size were associated with greater worry,36 negative affect37 and cognitive processing of HWs3 38 in the year following their implementation. As adult smokers smoke a higher average number of cigarettes a week (M=91)39 than adolescent smokers (M=15),40 adults have substantially more interactions with their cigarette packs and potential exposure to HWs than adolescents. With over 50% of Australian adult smokers indicating a desire to quit,41 greater exposure to cigarette packs may stimulate their interest in and processing of HW. However, to ensure their continued effectiveness, regular refreshing of HW may be needed.37

While adolescents’ exposure to cigarette packs decreased between 2011 and 2017, over half of Australian adolescents in 2017 had still seen cigarette packs in the past 6 months. Our finding that PP has reduced aspects of cigarette pack appeal may mean that pack exposure now contributes to smoking reduction rather than promotion.

Reductions in smoking uptake were proposed as long-term outcomes of the introduction of PP in Australia.32 Encouragingly, we found an increase in the proportion of NSNS since the introduction of PP and a decrease in experimental and past month smoking between 2011 and 2017. While a more comprehensive analysis is needed to tease out the relative contribution of PP and other tobacco policies on changes in adolescent smoking, the findings are in line with the long-term goals of PP legislation.

Several study limitations need to be noted. Although differences in brand image and perception ratings between 2011 and 2017 were significant, their size was modest. However, given the real-world nature of our study and the consistency in findings, our results suggest a meaningful reduction in the appeal of cigarette brands to Australian adolescents. Data from cross-sectional surveys of students cannot assess whether individuals’ responses to cigarette packs changed over time. However, our study shows important differences in population-level responses to cigarette packaging since the introduction of PP. While school response rates were relatively low, they were consistent across survey years, and there was no evidence that schools with more or less smoking refused participation. Rather, participation seemed to be influenced by school scheduling, staff availability and participation in other research.

Despite these limitations, our study provides new information on the longer term impact of PP on adolescent perceptions of cigarette brands and packs. In an environment where adolescents have been exposed to pictorial HW on cigarettes for most of their lives (17-year olds surveyed as part of the current study in 2017 were aged 6 when pictorial HW were first introduced), habituation seems to have occurred. However, as previous work has found that larger pictorial HWs contribute to negative perceptions of cigarette brands,18 the reduction we found in positive pack image ratings and brand characteristics may be partially attributable to larger HWs with PP. Less positive ratings of pack image and brand perceptions in 2017 compared with 2011 and greater uncertainty about the harmfulness and addictiveness of different brands suggests 4 to 5 years after its introduction, PP continues to meet its aim of reducing the attractiveness and appeal of tobacco products to young people.42

What this paper adds

  • The introduction of plain packaging (PP) of tobacco products with large graphic health warnings in Australia in 2012 was associated with an immediate reduction in the appeal of cigarette packs and brand image to adolescents, but no change in cognitive processing of warning labels.

  • The long-term impact of standardised packaging and larger graphic health warning (HW) on adolescents’ pack perceptions is unknown; although studies of graphic HW on fully branded packaging suggests initial effects may wear out over time.

  • This study found that the initial positive effects of PP on adolescents’ perceptions of cigarette packs persisted or were further enhanced 5 years after its introduction.

Acknowledgments

The authors thank and acknowledge the government and non-government education authorities, the school principals, teachers and students who cooperated to make this study possible. We thank the research staff for assistance with data collection.

References

Footnotes

  • Contributors VMW oversaw data collection for all survey years and took the lead in writing the paper and in data analyses. NG led data collection in 2017 and contributed to data analyses and study write up. TW managed data collection for 2013 and 2014 and contributed to the design of data analyses and paper writing. MAW contributed to the design of the study and contributed to paper writing. All authors reviewed the final version of the paper.

  • Funding Australian Government Department of Health, Victorian Government Department of Health and Human Services, Cancer Council Victoria.

  • Competing interests No, there are no competing interests.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No data are available.