Background The UK (UK) became the third country in the European Union to require pictorial warnings on the back of cigarette packs, in October 2008.
Methods A repeat cross-sectional survey was conducted with 11–16-year-olds in the UK between August and September 2008 (N=1401) and August and September 2011 (N=1373). At both waves the same text warnings appeared on the front and back of packs, with the only difference being the inclusion of images on the back of packs to support the text warnings in 2011. Warning related measures assessed were salience (noticing, looking closely at warnings), depth of processing (thinking about warnings, discussing them with others), comprehension and credibility (warning comprehensibility, believability and truthfulness), unaided recall, persuasiveness (warnings as a deterrent to smoking), avoidance techniques (eg, hiding packs) and a behavioural indicator (forgoing cigarettes due to warnings).
Results For never smokers, warning persuasiveness and thinking about what warnings are telling them when the pack is in sight significantly increased from 2008 to 2011, but warning comprehensibility significantly decreased. For experimental smokers, there was a significant increase from 2008 to 2011 for warning persuasiveness, believing warnings and considering them truthful. For regular smokers, there were no significant changes from 2008 to 2011, except for an increase in hiding packs to avoid warnings and a decrease in warning salience.
Conclusions Including pictorial images on the back of cigarette packaging improved warning persuasiveness for never and experimental smokers, but had a negligible impact on regular smokers. The findings have implications for warning design.
- Packaging and Labelling
- Public Policy
Statistics from Altmetric.com
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.
The WHO explains that consumers have a ‘fundamental right to accurate information about the risks of smoking,’1 with warnings on tobacco packaging a simple means of communicating these risks. More than 60 countries now require pictorial health warnings on packs and in five countries they cover 75% or more of the principal surfaces (Canada, Brunei, Uruguay, Sri Lanka, Australia).2 European Union (EU) countries trail behind this move to larger pictorial warnings. The Tobacco Products Directive 2001/37/EC3 stipulates written warnings: two general warnings covering 30–35% of the pack front, and 14 specific warnings covering 40–50% of the reverse. In 2005 the European Commission adopted a library of 42 images (three per specific warning) which member states can use on the reverse of packs. Less than half of member states have done so however, and none has warnings covering 50% of the main pack surfaces; the WHO ‘standard’ for strong warning design.1
Surprisingly, only one EU-based study with adolescents has compared the actual, rather than perceived, impacts of combined text/pictorial warnings with text-only warnings.4 In the UK a repeat cross-sectional study with 13–17-year-olds, conducted before and shortly after pictorial warnings were introduced, found that they increased awareness of the warnings and smoking harms.4 However, no significant changes emerged for avoidant (eg, hiding packs) or smoking behaviour (eg, forgoing cigarettes, reducing consumption or contemplating quitting).
A review of the warnings literature5 identified only one other study exploring the actual impacts of pictorial health warnings with a sample under the age of 18 years.6 White et al found that adolescents were significantly more likely to read, attend to, think and talk about pictorial warnings in Australia shortly after being introduced, and experimental and regular smokers were significantly more likely to think about quitting and forego cigarettes.6
Given the dearth of research with adolescents, we employ a repeat cross-sectional design to assess the medium-term (3 years postimplementation) impact of EU-style warnings.
Data comes from the Youth Tobacco Policy Survey. At wave 5, conducted between August and September 2008, text warnings featured on 43% of the front and 53% of the back of packs. At wave 6, conducted between August and September 2011, the warnings were identical to those used in 2008 except that images were used on the back of packs to support the text warning; the images were implemented from October 2008. The fieldwork comprised face-to-face inhome interviews and a self-completion questionnaire. Showcards were used for the interviews to maximise privacy, allowing participants to read response options and give a number corresponding to their answer. Prior consent was obtained from parents and participants.
Cross-sectional samples of 11–16-year-olds were drawn from UK households by random location quota sampling. Samples were randomly selected from 92 electoral wards stratified by Government Office Region and using ACORN (A Classification of Residential Neighbourhoods); a geodemographic system providing demographic and lifestyle profiles of small geographical areas. A quota sample, balanced across sex and age, was obtained in each ward. One thousand four hundred and one adolescents were recruited at wave 5 (2008) and 1373 at wave 6 (2011), see table 1. Ethics approval was obtained from the Marketing Department Ethics Committee at the University of Stirling.
Two items assessed smoking status. Never smokers had never smoked; experimental smokers had tried smoking, used to smoke or smoked less than one cigarette a week; regular smokers smoked at least one cigarette a week.
Information was obtained on gender, age, social grade (based on occupation of the chief income earner and coded using the National Readership Survey Social Grading System) and smoking by mother, father, siblings (if any) and close friends.
Two questions asked how often, in the last month, participants had noticed warnings on cigarette packs, and read or looked closely at these warnings. Response options were: Never (1), Rarely (2), Sometimes (3), Often (4), Very Often (5).
Depth of processing
Three questions asked how often, in the last month, participants had: thought about what warnings are telling them (with the pack in sight/out of sight), and talked with anyone about warnings. Response options were: Never (1), Rarely (2), Sometimes (3), Often (4), Very Often (5).
Comprehension and credibility
Understanding and credibility (believability and truthfulness) were assessed via three items measured on 5-point semantic scales: Easy (1)/difficult to understand (5); Believable (1)/Not believable (5); Truthful (1)/Not truthful about the health risks of smoking (5).
Participants were asked to recall, unaided, information they could remember from warnings. Responses were positively coded if they matched warning themes.
Warning persuasiveness was assessed with two items: Does (1)/does not put me off smoking (5); Makes me less (1)/more likely to smoke (5).
Avoidant behaviour/behavioural compliance
Regular smokers were asked whether they avoided looking at warnings by: Avoiding buying cigarette packets with particular warnings on them; Covering the warnings up; Putting the packet away; Using a cigarette case. All had dichotomous responses (Yes/No). They were also asked, in the last month, if warnings had stopped them from having a cigarette when they were about to smoke. Response options were: Never (1), Once (2), A few times (3), Many times (4).
Data were analysed using SPSS V.19. Data from 5-point scales were recoded into binary variables to facilitate analyses. Items measured on the 5-point ‘never’ to ‘very often’ scale were recoded to ‘often or very often’ (1) versus ‘never, rarely or sometimes’ (0). Items measured on 5-point semantic scales had codes ‘1’ and ‘2’ recoded to ‘1’ and codes ‘3’, ‘4’ and ‘5’ recoded to ‘0’. To examine differences in warning response by year, while controlling for potentially confounding variables, logistic regressions were conducted on each of the binary variables. For each dependent variable (salience; depth of processing; comprehension and credibility; persuasiveness; unaided recall) logistic regressions were run for the total sample and each smoking group (regular, experimental and never smokers). Each analysis compared responses in 2011 with 2008, controlling for age, gender and social grade, and parental, sibling and close friend smoking. Logistic regressions on the total sample also controlled for smoking status.
In the previous month, half the respondents (51% in 2008, 50% in 2011) had ‘often’ or ‘very often’ noticed warnings on packs and a fifth (22% in 2008, 20% in 2011) had ‘often’ or ‘very often’ read or looked closely at warnings (table 2). There was no significant overall change, from 2008 to 2011, in noticing warnings. However, regular smokers were less likely to report noticing warnings in 2011 (77% in 2008, 66% in 2011, adjusted ORs (AORs)=0.526, p=0.032). There was no between-wave change in reporting ‘often’ or ‘very often’ reading or looking closely at warnings.
Depth of processing
Thinking about warnings ‘often’ or ‘very often’ when a pack is in sight rose from 25% in 2008 to 31% in 2011 (AOR=1.335, p=0.001), although only significantly for never smokers (26% in 2008, 32% in 2011, AOR=1.305, p=0.009). One in 10 (9% in 2008, 10% in 2011) think about warnings when a pack is not in sight, with no significant between-wave difference (table 2). Five per cent at each wave reported talking about warnings ‘often’ or ‘very often’.
Comprehension and credibility
At each wave >85% of respondents considered warnings easy to understand, believable and truthful about the health risks (table 3). There was a between-wave increase in considering warnings truthful (86% in 2008, 90% in 2011, AOR=1.430, p=0.005) but no overall change for ease of understanding and believability. In 2011 never smokers were less likely to consider warnings easy to understand (92% in 2008, 89% in 2011, AOR=0.695, p=0.024) while experimental smokers were more likely to consider the warnings believable (84% in 2008, 93% in 2011, AOR=2.523, p=0.002) and truthful (85% in 2008, 94% in 2011, AOR=2.969, p=0.001).
Most respondents (>80%) perceived warnings as capable of putting them off smoking and making them less likely to smoke at both waves (table 3). ‘Putting me off smoking’ increased from 81% in 2008 to 85% in 2011 (AOR=1.380, p=0.014), and ‘makes me less likely to smoke’ from 81% in 2008 to 87% in 2011 (AOR=1.527, p=0.001). These increases were most pronounced for never and experimental smokers, with no significant change observed among regular smokers (table 3).
In 2008, the most commonly recalled messages were on the pack front: ‘Smoking Kills’ (58%) and ‘Smoking seriously harms you and others around you’ (41%). Recall of all other warnings, except ‘Smoking when pregnant can harm the baby’ (11%) and ‘Smoking causes fatal lung cancer’ (11%), was below 10%. In 2011 the most commonly recalled warning was again ‘Smoking Kills’ (47%), although recall decreased from 2008 (AOR=0.633, p<0.001), as did recall of ‘Smoking seriously harms you and others around you’ (25%) (AOR=0.462, p<0.001). There was an increase in recall for three fear appeal pictorial warnings: ‘Smoking causes lung cancer’ (healthy/diseased lungs image) from 11% to 33% (AOR=4.063, p<0.001); ‘Smoke contains benzene, nitrosamines, formaldehyde and hydrogen cyanide’ (rotten teeth image) from <1% to 13% (AOR=54.81, p<0.001), and ‘Smoking can cause a slow and painful death’ (neck tumour image), from <1% to 16% (AOR=64.08, p<0.001). For all other warnings recall remained below 10%. The three warnings on the pack reverse with no supporting image (‘You can do it, we can help’, ‘Choose freedom, we'll help you’ and ‘The risk of coronary heart disease is reduced by 50% after 1 year of smoking abstinence’) were recalled by less than 1% at either wave.
Avoidant behaviour/behavioural compliance
Between 2008 and 2011, there was a significant increase, from 12% to 23%, in hiding the pack among regular smokers (AOR=2.378, p=0.025), but not for other avoidant behaviours. There was no between-wave difference for regular smokers reporting that, in the last month, warnings had stopped them from having a cigarette (32% in 2008, 23% in 2011, AOR=0.116, p=0.611).
In keeping with another repeat cross-sectional study in the EU with adolescents, we found that although text/pictorial warnings improved some measures, most remained unchanged.4 The findings differ, however, from research with adolescents in Australia,6 where pictorial warnings feature on both sides of the pack. White et al6 found that salience, depth of processing and behavioural compliance (foregoing cigarettes) all significantly increased following the change from text-only to pictorial warnings.
Multiple factors likely contribute to these findings. As warnings need to be salient to be effective,7 positioning pictorial warnings only on the less visible reverse panel8 limits their impact. While recall was high at both waves for pack-front warnings, it was low (<10%) for the pictorial warnings on the pack reverse, fear-appeal pictures aside.5 Tellingly, the three warnings on the reverse panel that include only text were recalled by <1% at both waves, matching previous research.4 Using text-only messages within a set of pictorial warnings is inconsistent with best international practice; no country outside Europe does likewise.2 Lack of rotation also likely influenced the findings, with the same warnings used in the UK since 2008. This has likely increased wear-out9 and may help explain the decrease in salience from 2008 to 2011 for regular smokers.
In terms of potential limitations, the repeat cross-sectional design prohibits causal associations, and we relied on self-report measures. Where parents were present during the interview this may have influenced how adolescents responded. The use of random location quota sampling may question sample representativeness, and the relatively small number of regular smokers makes it difficult to detect between-wave statistical differences.
The findings have implications for warning design. Positioning pictorial warnings only on the back of packs may have had a deterrent effect on never and experimental smokers, but for most measures no significant differences were observed. The impact on regular smokers was negligible. As such, the large pictorial warnings (75%) on both principal display areas of packs in the EU, which was endorsed by the European Parliament's Committee on the Environment, Public Health and Food Safety in July 2013,10 will likely increase the impact warnings have for all adolescents, irrespective of smoking status.
What this paper adds
This is the first study in Europe to assess 11–16-year-olds’ response to the actual impacts of text/pictorial health warnings on tobacco packaging.
The inclusion of picture warnings on the reverse panel of packaging resulted in never and experimental smokers reporting that they make them less likely to smoke; the warnings had little impact on regular smokers.
The findings suggest that positioning picture warnings only on the reverse panel of packaging, and failing to rotate them, lessens the impact that they have for young people.
We would like to thank our late colleague Dr Lynn MacFadyen for her contribution to the development and design of this study, and FACTS International for undertaking the field work.
Contributors All authors were actively and substantially involved in drafting the article and final approval of the version to be published.
Funding This work was supported by a grant from Cancer Research UK (C312/A8721). GH is part funded by UK Centre for Tobacco Control Studies.
Competing interests None.
Ethics approval Marketing Department Ethics Committee at the University of Stirling.
Provenance and peer review Not commissioned; externally peer reviewed.