Perceptions of pack inserts promoting cessation and dissuasive among young adult smokers in the

Crawford Moodie1 PhD Rosemary Hiscock2 PhD Garth Reid3 PhD

1 Centre for Control Research, Institute for Social Marketing, University of Stirling

2 Department for Health, University of Bath

3 Policy Evaluation, NHS Health Scotland, Edinburgh

Suggested Citation Moodie C, Hiscock R, and Reid G (2018). Perceptions of inserts promoting cessation and dissuasive cigarettes among young adult smokers in the United Kingdom. Stirling, Scotland: Centre for Research, University of Stirling.

September 2018

ISM Institute for Social Marketing

1

Summary Combustible cigarettes continue to dominate the global market (Eriksen et al., 2015), with 5.5 trillion cigarettes sold in 2016 (CTFK, 2017). Cigarettes, unsurprisingly, are responsible for most of the seven million tobacco related deaths per year. While remains the leading cause of avoidable morbidity and mortality in Scotland and across the UK, prevalence and consumption is lower than it has been for a number of decades. This trend is expected to continue, with the Scottish Government ambitiously pledging to reduce smoking prevalence to five percent or less by 2034. If this target is to be achieved, however, it is likely that a raft of novel tobacco control measures will be required between now and then. Tobacco control is already strong in Scotland, and across the UK, with prices for tobacco products high, tobacco advertising, promotion and sponsorship banned (including the open display of tobacco products in retailers), the continued use of anti-smoking mass media campaigns, and smoking prohibited in public places and in cars with minors. In addition, as of May 2017, the Standardised Packaging of Tobacco Products Regulations and Tobacco Products Directive significantly altered the appearance of cigarette packaging. The legislation requires cigarettes and rolling tobacco to come in packs with a drab brown base colour, large pictorial health warnings on the front and back of packs and two additional text warnings on the secondary surfaces (see Figure 1). These changes drastically reduce the ability of the pack to create favourable perceptions of the brand and of smoking (Moodie, 2013). However, there is clearly scope for using the packaging to dissuade consumers. Regulators and academics alike have tended to focus on the exterior of the cigarette pack, with very little consideration of how the pack interior, for instance pack inserts or indeed the cigarettes themselves, could potentially be used to encourage smokers to think about their smoking behaviour.

Figure 1: Standardised cigarette packaging in the UK

We conducted an online survey with 1766 young adult smokers (aged 16-34 years) to explore their perceptions of cigarette design (a standard and two dissuasive cigarettes, displaying either a warning or unattractively coloured) and pack inserts with messaging promoting cessation. The two dissuasive cigarettes were perceived as significantly less appealing, and more harmful, than the standard cigarette. Participants were also significantly less likely to indicate that they (or a never smoker) would try one of the dissuasive cigarettes, compared to the standard cigarette. Approximately half the sample indicated that the two dissuasive cigarettes would put people off starting to smoke and make smokers want to quit. Half the sample thought that they would read the pack inserts, with three-fifths considering them a

2 good way to provide information to smokers about quitting. Just over half thought that the inserts would make them think more about quitting, might help them if they decided to quit, and were an effective way of encouraging smokers to quit. Those who had made a quit attempt in the last six months or were intending to quit within the next six months were most likely to find the inserts helpful for cessation. Approximately half supported having inserts promoting cessation inside packs, with a slightly lower proportion supporting all cigarettes having a warning on them or being an unattractive colour.

3

1. Introduction: The rise of the cigarette

While Southwest Indians, Aztecs and Mayans used hollow reeds, cane or maize to fashion cylindrical holders for smoking tobacco as far back as the 9th century, and Sevillians had smoked discarded scraps wrapped in (papeletas) since the 17th century, the cigarette as we know it did not emerge until the 19th century (Elliot, 2009). Cigarettes gained traction in Europe soon after, although being hand-rolled limited how many could be manufactured. This changed with the introduction of mechanised cigarette-rolling, and the mass-production of cheap cigarettes (Harvey, 2014). In the UK, by the early 20th century the cigarette was becoming more popular than snuff, plug, pipe tobacco and , which had heretofore dominated the market (Leonard, 2015). The move to cigarettes was helped not only by advancements in cigarette making machinery but also changing social mores, lower taxation than for other tobacco products, a widespread distribution network and concentrated advertising campaigns (Proctor, 2011). The cigarette also represented a break from the past and symbolised a dynamism associated with modernity (Schechter, 2003). War too played an important role, with the cigarette a better fit with the cramped intimacy and frequent interruptions of front-line service for smokers than pipe or (Proctor, 2011). By 1911 cigarettes accounted for 39% of the British tobacco market (Hannah, 2006), with market share steadily increasing thereafter. Globally, cigarette sales grew from 1.6 trillion in 1950 to 5.7 trillion by 2000 and, despite the 21st century being marked by significant advances in tobacco control in many countries, 5.5 trillion in 2016 (CFTK, 2017). Even with novel nicotine-containing products such as electronic cigarettes experiencing impressive growth, analysts predict that combustible cigarettes will remain the most popular of consumption for some time yet (Hedley, 2015).

1.1 The cigarette as the primary package

Packaging has a number of distinct levels (primary, secondary and tertiary) (Oostendrop et al., 2006), although there is debate about how best to define each of these (Hellström & Saghir, 2007). There is consensus that the primary packaging is in direct contact with the product (Ampuero & Vila, 2006), although there is some disagreement about whether the primary packaging is also the sales packaging (Prendergast & Pitt, 1996). For Simms and Trott (2010) the primary pack is in contact with the product and protects it (e.g. the plastic that contains cereal), while the secondary pack functions as the sales packaging, identifying it and communicating with the consumer (e.g. the cereal ) (Simms & Trott, 2010). If following the definition used by Simms and Trott (2010) then the primary pack would be the cigarette, as it is in direct contact with and protects the product (tobacco), with the cigarette pack being the secondary pack given that it identifies the brand, communicates product attributes and is on sale to consumers. This definition may not be accepted by those that maintain that the primary packaging must also be able to function as the sales packaging (Prendergast & Pitt, 1996), although single cigarettes have been on direct sale to consumers, at one time or another, the world over. Indeed, even though the Framework Convention on Tobacco Control (WHO, 2005) recommends a ban on the sale of single cigarettes, with the practice prohibited in the UK since 1992, it is still widespread in some parts of the world. For instance, single cigarettes are available for sale in vending machines in China, and in markets such as India and the Philippines more cigarettes are sold by the stick rather than by the pack (Lal et al., 2015). Single cigarettes are also available for sale across much of Africa, including in countries where their sale is banned (Wherry et al., 2014; African Tobacco Control Alliance, 2018). As such, defining the cigarette as the primary package would appear appropriate.

4

Irrespective of how the cigarette is best defined, tobacco industry documents support the notion of the cigarette being another level of packaging. A Philip Morris report, for instance, argues that “The cigarette should be conceived not as a product but as a package. The product is nicotine. The cigarette is but one of many package layers” (Dunn, 1972).

1.2 Dissuasive cigarettes

While there are a number of tried and tested tobacco control measures, it is argued that novel interventions are needed to stop children taking up smoking and help smokers to quit (Beaglehole et al., 2015). Myriad ideas have been proposed, focusing on the user (e.g. restricting sales by year born, requiring a smoker’s licence or prescription-only sales), the market (e.g. minimum pricing, advantaging cleaner nicotine products such as e-cigarettes over combustibles, quotas on tobacco manufacture and imports that are regularly reduced under a ‘sinking ’) and institutional structures (e.g. a regulated market model or state takeover of tobacco companies to be managed with a health mandate) (McDaniel et al., 2016). There have also been a number of product-focused proposals, including banning flavours, reducing nicotine levels, and increasing the pH level of cigarettes to make inhalation more unpleasant (McDaniel et al., 2016). Another product-related proposal that has recently emerged concerns altering the appearance of cigarettes to make them more unappealing (Hoek & Robertson, 2015; Moodie et al., 2015). Given that the cigarette has dominated the UK tobacco market for approximately a century, is responsible for most tobacco related mortality and morbidity, can be considered the primary packaging, and is an important promotional tool, the paucity of research exploring the potential impact of standardising the appearance of cigarettes to make them dissuasive is surprising. While research exploring the idea of dissuasive (standardised) cigarettes is at an embryonic stage, three concepts have emerged: 1) unattractively coloured cigarettes (Ford et al., 2014; Hoek & Robertson, 2015; Gallopel-Morvan et al., in press); 2) cigarettes with the health warning ‘Smoking kills’ on the cigarette paper (Moodie et al., 2015; Moodie, 2016; Moodie et al., 2017; Moodie et al., in press); and 3) cigarettes displaying the minutes of life lost due to smoking on the cigarette paper (Hassan & Shui, 2015; Hoek et al., 2016).

1.3 Research exploring dissuasive cigarettes

Two focus group studies in Glasgow explored young peoples’ perceptions of cigarettes which differed in diameter, colour, length and smell; the first in 2011 with 15 year-old smokers and non-smokers (N=48), the second in 2013 with 12-24 year-old female non-smokers and occasional smokers (N=75). A range of cigarettes were shown to participants in both studies, with appearance having a strong influence on perceptions of appeal and harm (Ford et al., 2014; Moodie et al., 2015). An unattractively coloured (brown) cigarette was viewed very negatively, thought to be unpleasant to smoke, disgusting and harmful (Ford et al., 2014; Moodie et al., 2015). Hoek and Robertson (2015) also used qualitative research to explore young women smokers’ (N=22) perceptions of cigarettes, including cigarettes with dissuasive colours. Dark green and brown cigarettes were perceived very negatively, exposing smoking as dirty, reducing social acceptability and thought to make the smoking experience less satisfying. They created an unsettling dissonance as participants struggled to reconcile these unappealing cues with the experience and identity they sought (Hoek & Robertson, 2015). An online survey in in 2013, with 15-30 year old smokers and non-smokers (N=998), explored their perceptions of a plain cigarette (grey with no brand name) and three branded cigarettes (regular, slim, pink). Participants were randomly assigned to view the plain

5 cigarette and either the regular, slim or pink cigarette and asked to rate the cigarettes on appeal, harm and perceived behavioural impact. Each branded cigarette was viewed as more appealing, less dangerous, more likely to motivate teenagers to start smoking, and less likely to motivate smokers to reduce consumption or quit than the plain grey cigarette (Gallopel- Morvan et al., in press). Two qualitative studies have explored the potential impact of including a health warning, ‘Smoking kills’, on cigarettes. The first, conducted in 2012, involved focus groups with young women smokers aged 16-24 years (N=49) who were shown a range of novel packaging measures, including cigarette packs with QR barcodes that could be scanned with a smartphone (taking the user to a stop-smoking website), packs that played audio health messages when the lid was opened, and cigarettes displaying the warning ‘Smoking kills’; this warning was chosen as it is concise, credible, has been used on cigarette packs in countries within each continent, and is supported by evidence (e.g. Wardle et al., 2010; GfK Bluemoon, 2011). The cigarette warning was viewed as unappealing for some participants, being considered a constant reminder of the associated health risks and off-putting, primarily because of the perceived discomfort of being observed by others smoking a cigarette displaying this message (Moodie, Purves et al., 2015). The second study, in 2014, involved interviews with packaging, branding and design experts (N=12) who were shown the novel packaging measures used in the previous study. The on-cigarette warning was described as ‘fantastic’, ‘powerful’ and ‘brilliant’ and suggested to have multiple benefits: it would confront the smoker, deter the non-smoker, signal to youth that it is neither cool nor intelligent to smoke, prolong the health message, serve as a continual reminder of the associated risks, and undermine the use of an alternative carrier for those decanting cigarettes to avoid exposure to on-pack warnings or plain packaging (Moodie, 2016). Some experts suggested an unpleasant colour for the cigarette paper instead of a warning, e.g. brown or dark green - as tested in earlier research (Ford et al., 2014; Hoek & Robertson, 2015; Moodie et al., 2015) - given the importance of colour in eliciting meaning and associations. In addition to these qualitative studies, an in-home survey with 11-16 year olds from across the UK in 2014 involved participants (N=1205) being shown a visual prompt, an image of a cigarette with ‘Smoking kills’, and asked whether warnings on cigarettes would put people off starting to smoke, make people want to give up smoking, and whether all cigarettes should have a health warning on them (Moodie et al., 2017). Almost three-quarters (71%) of the sample indicated that warnings on cigarettes would put people off starting to smoke, more so never smokers than current smokers. Approximately half (53%) thought that they would make people want to give up smoking, with never smokers and experimenters/past smokers more likely than current smokers to think that cigarette warnings would make people want to give up smoking. The vast majority (85%) thought that all cigarettes should have warnings on them, with support greater among never smokers and experimenters/past smokers than current smokers (Moodie et al., 2017). An online survey in 2015 with 16-24 year old smokers and non-smokers (N=997) explored perceptions of a standard cigarette and two dissuasive cigarettes (a cigarette with the warning ‘Smoking kills’ and a green cigarette). Participants were asked to rate each cigarette on appeal, harm, strength, and taste, which were combined to provide a ‘favourability’ score, and also perceived trial (Moodie et al., in press). Compared with the standard cigarette, both smokers and non-smokers rated the warning cigarette and green cigarette as more unfavourable, and indicated that they would be less likely to try them (Moodie et al., in press). Two additional studies have explored smokers’ perceptions of a cigarette which displayed ‘minutes of life lost’ on the cigarette paper. Hassan and Shui (2015) conducted two studies with adult smokers (N=208) and found that quit intentions, assessed before and after participants were shown either an image of the cigarette (study 1) or an actual cigarette (study

6

2), significantly increased post-exposure. Hoek et al (2016) conducted an online survey of 313 smokers in 2014 using a Best–Worst Choice experiment and rating task and explored dissuasive cigarettes (with the warning ‘Smoking kills’, an unpleasant colour, or displaying minutes of life lost). Each dissuasive cigarette was rated as less appealing than a standard cigarette, with the cigarette with ‘minutes of life lost’ rated most negatively. However, should policy makers decide to pursue dissuasive cigarettes in the future, if existing evidence suggests that they may help deter smoking, there may be a reluctance to opt for the ‘minutes of life lost’ message as it would be impossible to prove that each cigarette reduces life expectancy by a certain number of minutes due to differences in cigarette geometry (length and diameter) and smoking/cessation-related behaviours (e.g. smoking intensity, level of consumption, smoking duration, age of initiation and cessation). Calculating the average minutes of life lost is reliant on averages and assumes that the health effects of smoking are evenly spread throughout a smoker's lifetime (Shaw et al., 2000). Collectively, these studies suggest that the cigarette is an important communications tool, consistent with how it is viewed within tobacco industry journals (Mapother, 2012), and altering the appearance of cigarettes, for instance with the addition of a health warning or an unpleasant colour, can influence how cigarettes, and indeed smoking, are perceived.

2. Cigarette cards, and inserts

The use of a piece of card inside a cigarette pack to communicate with consumers has been a long-term strategy of tobacco companies. In the 1870s the fragile paper cigarette packs of the time had a piece of thick card inside to stiffen the box and protect the cigarettes from being crushed or bent. While the card initially had a protective rather than communicative function, it was not long before tobacco companies in the US decided to employ it as a marketing tool (Kluger, 1996). Just as tradesmen began giving away colourful business cards in the mid-19th century, taking advantage of advances in colour technology (Blum, 1995), tobacco companies started to print colourful pictures and advertisements on the card (The London Company, 1982). The president of the is reported to have helped pioneer this approach, transforming these simple work-horses into a powerful marketing tool by printing the brand name of the cigarettes along with a picture (Pritcher, undated). The first cards were ‘singles’, unrelated in subject matter, but manufacturers soon began to produce sequences of cards on particular themes (The London Cigarette Card Company, 1982). With most smokers at the time male the most dominant themes were actresses, sports, pioneering and politics. The cards were typically organised in sets of between 25 and 50 and designed to be collected in full (The London Cigarette Card Company, 1982). Thousands of different sets were issued by cigarette manufacturers, all competing to sell their products and establish or strengthen brand loyalty. The novelty of these cards, allied to the fact that many in the US did not carry pictures at the time, and a significant proportion of the population was illiterate, helps explain why they captured the attention of the public. They were also appealing to young people, with it apparently a common sight to see children outside shops asking smokers for the cigarette card (Johnson, undated). These cards were introduced in Britain in the 1880s, and by the early 20th century they had become a global phenomenon (The London Cigarette Card Company, 1982). Indeed, by the early 1900s cigarette cards had established an almost fanatical following, becoming a ‘national craze’ in Britain (Mullen, 1979). The outbreak of the first World War generated many patriotic cigarette cards, expressing propaganda and national pride (Pringle, 1991), but it also saw the use of cigarette cards temporarily suspended due to a shortage of materials and paper rationing (Evans, 2013). Nevertheless, they quickly regained momentum from the early

7

1920s and 1930s before production was largely discontinued in the 1940s (The London Cigarette Card Company, 1982), with the British wartime government citing them as ‘a waste of vital raw materials’ (Johnson, undated). Use of these cards in the UK did continue sporadically for decades thereafter, for instance some manufacturers reintroduced tobacco cards in the 1950s, but their decline was paralleled by the rise of the moving image (Pringle, 1991), the growing popularity of the more sturdy flip-top pack, and the fact that chewing gum, tea and sweet manufacturers had entered the field and became the biggest publishers of cards (Blum, 1995). Nevertheless, they had not fallen completely out of fashion, with RJ Reynolds printing cigarette cards in 2000 in the US for the brand, with themes including US states, American festivals, cars and national parks (Simpson, 2000). In the UK they were used up until 2003, when Imperial Tobacco included cigarette cards depicting former Superkings advertising campaigns (The Grocer, 2003). This was in breach of the Tobacco Advertising and Promotion Act (TAPA) however, and widely publicised, essentially putting an end to cigarette cards in the UK. Like cigarette cards, coupons have been used in cigarette packs for some time. As early as the 1880s tobacco companies in the US included coupons that could be exchanged for gifts (Goodrum & Dalrymple, 1990). Coupons, such as cash certificates and money back guarantees, were included in packs in Britain since at least the early 1900s (Congdon-Martin, 1992). Used to promote sales, coupons continued in the US and elsewhere for decades (Yan, 2010), with over twenty schemes in the UK by the early 1930s (Mullen, 1979). In a difficult economic climate Mullen (1979) points out that the gifts from these schemes were not merely novelties but an important source of household necessities, such as clothing and razors – of course money not spent on cigarettes could have been put to this use. These coupon schemes, for instance offering money off future purchases (Off Licence News, 2002), continued in the UK until the 21st century, when they were no longer permitted by the TAPA. The demise of the cigarette card and coupon has not stopped tobacco companies from taking advantage of the opportunities that the pack interior offers. They continue to use inserts as a cost-effective way of using the package to convey additional information to the consumer (Selin & Sweanor, 1998), as they have done for over a century. For instance, early in the 20th century tobacco companies included inserts printed or woven on various types of cloth inside packs, often referred to as 'silk inserts', possibly to attract more women to smoking or, for those that were already smokers, increase the amount they smoked (Lee, 2006). Nowadays, inserts are used in packs to exude brand qualities and communicate the brand message, inform of brand extensions, pack redesigns and new product development, and as the basis of promotions (Simpson, 2004; Goswami, 2006; Moodie & Hastings, 2010; Solmajer, 2010). It has also been suggested that they have been used to appeal to youth. In the US, for example, a pack contained an insert resembling a Pokemon card, which were popular with pre- teens (Connolly, 2004). Another common use is to undermine tobacco control. Inserts were used in the UK, prior to the ban on pack descriptors such as Light and Mild, to explain how colour coding would be subsequently used to signify product strength (Moodie et al., 2012). When pictorial health warnings became mandatory in Brazil some tobacco companies started including inserts ( ) inside packs which were the same size as the health warnings, thus allowing smokers to use these to conceal the warning if they chose to do so (Cavalcante, 2003). Tobacco companies also used inserts in packs in the UK to encourage consumers to challenge tobacco regulation (Gyekye, 2013), see Figure 2.

2.1 Pack inserts to communicate health messages

That pack inserts are viewed within the marketing literature as an effective and inexpensive means of communication (Rowse & Fish, 2005), and offer almost endless opportunities for

8 communicating with smokers (Mahood, 1995), helps to explain why they are so frequently used by tobacco companies. Regulators, in contrast, have overlooked the potential of inserts to communicate health messaging to smokers. The Canadian government is the exception, having required tobacco companies to include inserts in packs since the turn of the century to augment the information appearing on the pack exterior (Health Canada, 1999). Sixteen messages were required on the inside of packs via inserts or on the interior of the pack itself. Nine of these 16 messages encouraged cessation (beginning “You CAN quit smoking!”), with the remaining seven providing detailed health information to complement the exterior warnings (Mahood, 2004), see Figure 3.

Figure 2: Insert inside packs of Marlboro

The failure of the Canadian government to commit to the messages on the inside of the pack early in the development process affected their quality, being limited to text highlighted in yellow but without full colour, pictures or graphics (Mahood, 2004). However, in 2012 eight new inserts were introduced, this time with coloured graphics and behavioural recommendations for quitting and the benefits of doing so (Figure 3). The inserts provide messages that are consistent with the health communication literature and the idea that fear- arousing messages should be followed by behavioural recommendations to help escape the source of the fear (Thrasher et al., 2015).

Figure 3: Example of early and later Canadian inserts

9

2.3 Research exploring pack inserts

Few studies have explored perceptions of pack inserts. Two early studies were conducted in Canada on behalf of Health Canada. The first, interviews with smokers aged 12 to 50 years (N=228), tested messages on warnings and inserts (Tandemar Research Inc, 1992). Eight inserts were used, featuring messages about addiction and health risks to smokers and others. Inserts were seen to allow for significant amounts of information to be presented, and were generally liked, with the information considered believable, important and relevant. Participants tended to focus on the health risks rather than the information about available help (Tandemar Research Inc, 1992). A second study, interviews with smokers aged 15 years and older (N=670), tested health messages on the outside and inside of cigarette packs among smokers (Environics, 2000). Almost half indicated that the messaging on the inserts, describing the health risks of smoking and help available to quit, was a good way to provide information to smokers (48%), with a similar proportion (47%) thinking that they would not be a good way to provide information to smokers. For those indicating that inserts were a good way to provide information, 46% said that people would read the message, and 39% that inserts may convince/help smokers to quit. For those indicating that inserts were not a good way to provide information, 63% said that the message would be better on the outside of the pack or that the inserts would be thrown away. Sixty percent thought that the messages would be a very effective/somewhat effective way to provide information to smokers who want to quit, with approximately two-fifths thinking that they would discourage young people from starting to smoke (42%) and encourage young smokers to quit (42%). Those who noticed the insert about getting help to quit said they were more likely to notice quitting information than those who did not notice this insert (Environics, 2000). Face-to-face interviews with smokers and recent quitters (N=73) in New Zealand assessed perceptions of health warnings and pack inserts. Participants were randomly exposed to six inserts, although all were shown an insert used in Canada (‘Can second hand smoke harm my family’). They were asked whether they agreed or disagreed that having these inserts inside cigarette packs would be an effective way of encouraging smokers to reduce the amount they smoke or quit. A third (34%) agreed/strongly agreed that inserts would be an effective way of encouraging reduced consumption or quitting, while a similar proportion (33%) disagreed/strongly disagreed. While some were positive about the informative, educational tone of the inserts, the general view was that inserts would inevitably be thrown away and would receive only a cursory glance from smokers or be ignored completely (BRC Marketing and Social Research, 2004). A study with adult smokers and non-smokers (N=836) in France explored perceptions of different ways of displaying emission labelling, either on the side of the pack (as the numerical display of tar, nicotine and carbon monoxide levels, or as descriptive information: ‘tar includes hundred chemical substances that cause cancer’) or as a pack insert. The insert read: ‘Tar includes hundreds of chemical substances that cause cancer. They hurt smokers’ lungs and bronchial tubes. They cause emphysema and chronic bronchitis. Ask your doctor for help to quit smoking. Call the quit line 0825309310 or consult the website: www.tabac- infos-sercice.fr’. The insert was considered easier to understand and more helpful in informing of the health risks than the numerical display of tobacco constituents printed on the side of packs (Gallopel-Morvan et al., 2010). In the UK, interviews with packaging, branding and design experts (N=12) explored a number of novel ways of using tobacco packaging to communicate with smokers, including pack inserts with either loss- or gain-framed messages (Moodie, 2016). The inserts, with coloured graphics, were viewed as a suitable platform for communicating with consumers and a cheap and simple method of capturing attention. As warnings on the exterior of packs

10 typically feature loss-framed messages, having a similar style of message on the interior was not considered the best option. Instead, inserts with gain-framed messages (which were referred to as “help cards”) were thought to be more effective in engaging smokers and more likely to have an impact on those wanting to quit. A focus group study in Scotland, with smokers aged 16 and over (N=120), found that pack inserts featuring messages highlighting the benefits of quitting or providing tips on how to do so (the same inserts as used in Canada) were consistently well received (Moodie, 2018). The consensus was that they would be read as a result of their novelty, visibility before reaching the cigarettes, and as they can be taken out of the pack. While it was commented that inserts may be ignored or discarded, and that they would need to be rotated, they were generally thought to prolong the health message. The positive style of messaging was considered encouraging, reassuring, sympathetic, motivational, inspirational, refreshing, educational and informative, and considered to increase message engagement. The inserts were considered a useful supplement to the on- pack warnings, although several groups felt that both would be needed. The inserts were generally viewed as having the potential to alter others smoking behaviour, particularly younger people and those wanting to quit, and could help smokers think about or question their smoking behaviour and encourage them to stop (Moodie, 2018). Eight focus groups were conducted with 47 young adult smokers in Turkey in December 2016 (Mucan & Moodie, in press). Participants were shown three pack inserts with messages encouraging quitting or offering tips on how to do so, adapted from Canada, and asked their perceptions of these. The inserts were not considered as eye-catching, memorable or effective as the on-pack warnings (Mucan & Moodie, in press). However, consistent with qualitative research in Scotland (Moodie, 2018), within several female groups they were considered an appropriate supplement to the on-pack warnings. With respect to discouraging smoking, it was felt that inserts would be useful for young people, those who have just started smoking and those intending to quit, as found in past research (Environics, 2000; Thrasher et al., 2015; Moodie, 2018). Only a single study from Canada, a longitudinal online survey, has explored the response of smokers to the pack inserts introduced in 2012. It found that approximately a quarter of smokers at each wave (between 26% and 31%) reported having read the inserts at least once in the prior month, with those intending to quit or having recently tried to quit significantly more likely to have read them. In comparison to smokers that had not read the inserts, smokers who read the inserts a few times or more in the past month were more likely to make a quit attempt at the subsequent wave (Thrasher et al., 2015). It was also found that while reading on-pack health warnings significantly decreased across waves, reading inserts significantly increased, with more frequent reading of inserts associated with self-efficacy to quit, quit attempts, and sustained quitting at follow-up (Thrasher et al., 2016).

2.4 Research questions

As tobacco companies exploit the entire cigarette pack as a communications tool (Mawditt, 2006), then supplementing plain packaging with inserts featuring coloured graphics and promoting cessation, and also cigarettes displaying a health message or having an unpleasant colour, would be consistent with this approach. However, research exploring dissuasive cigarettes and pack inserts is at a nascent stage. In this study we explored young adult smokers’ perceptions of a standard and two dissuasive cigarettes (standard cigarette displaying a health warning, and an unattractively coloured cigarette), see Figure 4. We also explored perceptions of inserts highlighting the benefits of quitting and promoting self-efficacy, as used in Canada, with four inserts used (see Figure 5). The study was guided by three main research questions:

11

1) Will the two dissuasive cigarettes be rated as more undesirable than the standard cigarette in terms of appeal, harm and product trial? 2) Will pack inserts be perceived as useful for helping to communicate the benefits of quitting and encouraging smokers to do so? 3) Will responses to the cigarettes and inserts differ by sociodemographic or smoking related characteristics?

Figure 4: Standard cigarette, cigarette with warning, and dark green cigarette

Figure 5: Pack inserts used in this study

3. Methods

3.1 Design and sample

An online survey was conducted with young adult smokers (aged 16-34 years), between January and February 2016, to explore their perceptions of cigarette design and pack inserts. The online panel provider was Research Now, a commercial market research company with a panel of over 400,000 participants in the UK. Online surveys have been frequently employed in research on cigarette packaging (Doxey & Hammond, 2011; Moodie, et al., 2012; White et

12 al., 2012; Kotnowski et al., 2016), and are particularly appropriate for young adults given high rates of internet use among this population, with 99% of 16-34 year olds having used the internet within the last three months (ONS, 2015). We focused on young adult smokers for two key reasons: 1) Smoking prevalence in the UK was higher among 25-34 year olds and 16- 24 year olds than in any other age group (HSCIC, 2014), and 2) tobacco industry documents reveal young adults to be a key target group (Leading Edge, 1997; M&C Saatchi, 1998). The inclusion criteria were that participants were current smokers of factory-made cigarettes and aged 16-34 years. During fieldwork, the online panel provider eliminated 193 participants for completing the survey too quickly (speeders) and 11 for comments made within an open-ended question at the end of the survey that suggested that they had not taken it seriously. The final sample was 1766, with 50.3% male, 53.9% aged 25-34 years and 71.6% white British. Most participants smoked 10 or less cigarettes per day, with 46.0% exclusive factory-made cigarette smokers (see Table 1 and Appendix section 3 for sample and smoking- related characteristics).

Table 1: Sample and smoking-related characteristics

Characteristic N % Total 1766 100.0

Age group 16-19 413 23.4 20-24 401 22.7 25-34 952 53.9

Gender Male 888 50.3 Female 878 49.7

Educational qualifications Other qualifications 1357 76.8 None or GCSE 409 23.2

Economic status Other status 1350 76.4 Routine or manual occupation, unemployed or long term sick 416 23.6

Socioeconomic status (SES) No indicators of low SES 1114 63.1 Low education and/or low SES 652 36.9

Ethnicity White British 1264 71.6 White non-British 162 9.2 Black (including mixed black and white) 79 4.5 Asian (including mixed Asian and white) 196 11.1 Other or not declared 65 3.7

Location England 1550 87.8 Scotland 109 6.2 Wales 73 4.1 Northern Ireland 34 1.9

13

Characteristic N %

Tobacco products used Only factory-made () cigarettes 813 46.0 Factory-made and roll-your-own cigarettes 681 38.6 Factory-made cigarettes and other products (e.g. cigars, shisha) 272 15.4

Cigarettes per day 10 or less 1272 72.0 11-20 433 24.5 21-30 46 2.6 31 or more 15 0.8

Time to first cigarette Within 5 minutes 263 14.9 6 to 30 minutes 570 32.3 31 to 60 minutes 315 17.8 After 60 minutes 618 35.0

Heaviness of Smoking Index (HSI) 0 little dependence 601 34.0 1 257 14.6 2 418 23.7 3 293 16.6 4 156 8.8 5 28 1.6 6 high dependence 13 0.7

Dependence (Tertiles of HSI) Low-dependence 601 34.0 Mid-dependence 675 38.2 High-dependence 490 27.7

Made an attempt to quit smoking that lasted at least 24 hours? Yes, within the last six months 788 44.6 Yes, more than six months ago 552 31.3 No, I have never tried to quit smoking for more than 24 hours 426 24.1

How likely are you to try to quit smoking within the next six months? Not at all 198 11.2 A little 382 21.6 Moderately 508 28.8 Very 308 17.4 Extremely 272 15.4 Don't know 98 5.5

If you decided to quit smoking in the next six months, how sure are you that you would succeed? Not at all 147 8.3 A little 346 19.6 Moderately 612 34.7 Very 297 16.8 Extremely 241 13.6 Don't know 123 7.0

14

Characteristic N %

Likelihood of quitting Moderately or less likely to make quit attempt in next six months 1186 67.2 (unlikely to make a quit attempt in the next six months) Very or extremely likely to attempt but moderately or less likely to succeed 304 17.2 (unlikely to make a successful quit attempt in the next six months) Very or extremely likely to attempt and very or extremely likely to succeed 276 15.6 (likely to make a successful quit attempt in the next six months)

3.2 Measures

Sociodemographic characteristics Age, gender, ethnicity, educational attainment and economic status (based on chief income earner) were obtained. Preliminary analysis showed that education was associated with how pack inserts were perceived, whereas both education and economic status were associated with how cigarettes were perceived. As such, for the analysis of the cigarettes a count procedure was used to create a variable for low socioeconomic status (SES): low education (General Certificate of Secondary Education: GCSE or below) and/or low economic status (routine or manual occupation, long-term unemployed or long-term sick or disabled.

Smoking behaviour Smoking status was assessed with ‘Which of these best describes you?’ with response options: ‘I have never smoked’, ‘I used to smoke, but don’t now’, ‘I smoke, but not every day’, and ‘I smoke every day’. Only current smokers were recruited. One question was used to assess tobacco products used, ‘What type(s) of tobacco products do you smoke?’ with response options: ‘Only factory-made (packet) cigarettes’, ‘Factory-made and roll-your-own cigarettes’, ‘Factory-made cigarettes and other tobacco products (e.g. cigars, shisha, etc)’, ‘Only roll-your-own cigarettes’ and ‘Only smoked other tobacco products (e.g. cigars, shisha, etc)’. Given the focus on cigarettes and cigarette pack inserts, smokers who did not use factory-made cigarettes were excluded. The Heaviness of Smoking Index (HSI) (Kozlowski et al., 1994) was used to measure dependence, based on daily consumption and time to first cigarette.

Quitting and self-efficacy Participants were asked ‘Have you ever made an attempt to quit smoking that lasted at least 24 hours?’ (‘Yes, within the last six months’, ‘Yes, more than six months ago’, ‘I have never tried to quit for more than 24 hours’), and ‘How likely are you to try to quit smoking within the next six months?’ (‘Not at all’, ‘A little’, ‘Moderately’, ‘Very’, ‘Extremely’, ‘Don’t know’). To measure self-efficacy, participants were asked ‘If you decided to quit smoking in the next six months, how sure are you that you would succeed?’ (‘Not at all’, ‘A little’, ‘Moderately’, ‘Very’, ‘Extremely’, ‘Don’t know’). Those who responded to the likelihood of quitting question with ‘Very or ‘Extremely’ and to the quitting efficacy question with ‘Not at all’, ‘A little’, ‘Moderately’ or ‘Don’t know’ were classified as ‘unlikely to make a successful quit attempt in the next six months’. Those who responded ‘Very’ or ‘Extremely’ to both questions were classified as ‘likely to make a successful quit attempt in the next six months’.

Cigarette design: Appeal, harm and trial Seven-point semantic scales assessed appeal, harm and trial. Appeal was assessed via four scales, with anchors ‘Attractive-Unattractive’, ‘Stylish-Not stylish’, ‘Not nice to be seen with-

15

Nice to be seen with’ and ‘Not appealing to people my age-Appealing to people my age’. Harm was assessed via two scales, with anchors ‘Looks harmful to health-Does not look harmful to health’ and ‘Makes me think about the dangers of smoking-Does not make me think about the dangers of smoking’. Likely trial was assessed via two scales, ‘If a friend offered you each of these cigarettes, how likely would you be to try them?’ and ‘If someone your age who had never smoked before was going to try a cigarette, how likely do you think they would be to try each of these cigarettes?’ Both scales assessing trial ranged from ‘Not at all likely’ to ‘Very likely’.

Dissuasive cigarettes: Initiation, cessation and support For the two dissuasive cigarettes, five-point semantic scales were used to assess initiation, cessation and support (Moodie et al., 2017). Initiation was assessed with ‘Would put people off starting to smoke - Would not put people off starting to smoke’, and for cessation ‘Would not make people want to give up smoking - Would make people want to give up smoking’. For support, the anchor points were ‘All cigarettes should have a health warning on them - No cigarettes should have a health warning on them’ for the cigarette warning, and ‘All cigarettes should be an unattractive colour – No cigarettes should be an unattractive colour’ for the green cigarette.

Inserts: Salience and information provision Questions exploring pack inserts were adapted from, or informed by, past research (Environics, 2000; BRC Marketing and Social Research, 2004; Thrasher et al., 2015; Moodie, 2016). For all questions exploring inserts, participants were shown an image of one of four inserts (Figure 5). To capture salience, participants were asked ‘If this type of insert was in your cigarette pack, do you think that you would read it?’ and separately ‘If this type of insert was in your cigarette pack, do you think that you would read it if you were interested in quitting?’ They were also asked ‘Do you think that inserts would be a good way to provide information to smokers about quitting?’ All questions had response options ‘Yes’, ‘No’ and ‘Not sure’.

Inserts: Cessation and support Three questions were used to assess the extent to which participants agreed or disagreed that inserts would make them think about quitting, and help them quit: ‘Do you agree or disagree that having these types of inserts in every cigarette pack would make you think more about quitting?’ and ‘Do you agree or disagree that having these types of inserts in every cigarette pack might help you if you decided to quit?’ They were also asked ‘Do you agree or disagree that having these types of inserts inside every cigarette pack would be an effective way of helping smokers who want to quit?’ For each question response options were ‘Strongly disagree’, ‘Disagree’, ‘Neither agree nor disagree’, ‘Agree’, ‘Strongly agree’ and ‘Don’t know’. A five-point semantic scale was used to assess support: ‘All cigarette packs should have inserts like this in them - No cigarette packs should have inserts like this in them’.

3.3 Procedure

Participants were recruited from an online panel provider (Research Now). An email was sent inviting panel members to participate and for those that responded, and eligible for inclusion, they were presented with an information page explaining the details of the study: who it was being run by, the aims, participants sought, what they would be asked to do, the option to withdraw at any time, how the data would be stored and used, and who participants could

16 contact if they had any questions or ethical concerns. Following the information page was a consent page, with consent required prior to survey completion. Questions were presented in the same order for all participants, although for the questions relating to perceptions of cigarettes, the ordering of the presentation of the three cigarettes (standard cigarette, cigarette with warning, green cigarette) was randomised. For each question about inserts, participants were shown the question and an image of one insert. Four different inserts were used in total, as shown in Figure 5, with these chosen from the eight used in Canada as they were considered most relevant to our sample. The words ‘Health Canada’ were removed from the bottom of each insert to make them more relevant for participants in the UK. Participants received a nominal incentive (50 pence) for participation. The median time taken for survey completion was 9 minutes 28 seconds (inter quartile range: 7 minutes 23 seconds to 12 minutes 32 seconds). The study received ethical approval from the School of Health Sciences ethics committee (Ref: SREC 15/16), at the University of Stirling, in December 2015.

3.4 Analysis

The software packages used for analyses were SPSS v22 and v23 and MlWin v2.33 (Rasbash et al., 2015). The research questions were analysed as follows:

1) Will the two dissuasive cigarettes be rated as more undesirable than the standard cigarette in terms of appeal, harm and perceived product trial?

Perceptions of the cigarettes were assessed with seven-point semantic scales, ranging from desirable and undesirable. Percentages were calculated for those indicating one of the three points nearest the desirable (e.g. stylish) and undesirable (e.g. not nice to be seen with) anchors.

2) Will pack inserts be perceived as useful for helping to communicate the benefits of quitting and encouraging smokers to do so?

Percentages indicating ‘yes’ or ‘no’/’not sure’ were tabulated for relevant questions. For the five-point semantic scales indicating the extent to which participants agreed or disagreed with the question, the percentages agreeing (indicating ‘strongly agree’ or ‘agree’), disagreeing (indicating ‘strongly disagree’ or ‘disagree’), and who were ‘neutral’ or ‘unknown’, were calculated.

3) Will responses to the cigarettes and inserts differ by sociodemographic or smoking related characteristics?

Cigarette type

Factor analysis A data file was created where each row represented one of the three cigarettes (standard, with warning, green), giving three times as many rows as the original file (n=5298). The eight questions assessing harm, appeal and trial were collated for each cigarette, giving eight variables instead of the 24 in the original file (see Figure 6). A factor analysis of the eight variables on appeal, harm and trial, collated for the three cigarette types, was undertaken. Checks indicated that the data was suitable for factor analysis

17

(Kaiser Meyer Olkin=0.845, Bartlett’s test of Sphericity (approx. chi-square 18062.84, df=276, p<0.001), and no correlations between the variables >0.9). The factor extraction method was Principal Axis Factoring and the criteria for extraction was eigenvalues >1. No rotation was undertaken because only one factor was extracted. All eight variables loaded on the factor >0.5. High scores on the factor indicated that the cigarette was desirable and low scores that the cigarette was undesirable. Inspection of histograms (see Figures A1 and A2 in the Appendix) and the Kolmogorov-Smirnov test indicated that the factor was non-normal. Attempts to normalise it using normit rankit methods failed, thus the factor was divided into tertiles and the tertile with the ‘most undesirable’ factor scores was compared with the other two tertiles. This was the outcome variable in logistic regression analysis.

Figure 6: Manipulation of data file from wide to long format

Original dataset extract–wide format Harmful to Harmful to Harmful to Think of Think of Think of health- health- health- dangers- dangers- dangers- standard warning green standard warning green respondent id cigarette cigarette cigarette cigarette cigarette cigarette respondent 1 1 3 3 1 3 4

respondent 2 2 2 2 2 2 2

respondent 3 6 5 5 5 4 4

Manipulated dataset extract -long format

Harmful Think of respondent & cigarette to health dangers respondent 1 standard 1 1 respondent 1 warning 3 3 respondent 1 green 3 4 respondent 2 standard 2 2 respondent 2 warning 2 2 respondent 2 green 2 2 respondent 3 standard 6 5 respondent 3 warning 5 4 respondent 3 green 5 4

Multivariable regression with the factor desirability as the outcome Multilevel logistic regression modelling was undertaken with cigarette types (at level one) clustered with individual participants (at level two). Second order PQL linearization was undertaken. All models included cigarette type as a fixed effect where the standard cigarette was compared with the cigarette with warning and green cigarette. Other fixed effects were at the individual (participant) level and also sociodemographic characteristics and smoking related characteristics (see Appendix A2.1 and A3.3 for distributions of these characteristics). The following sociodemographic characteristics were included. Gender was entered with male as the reference category. Age was divided into three categories: 16 to 19 year olds were compared with 20-24 year olds and 25 to 34 year olds. The derived SES variable compared low SES respondents with other respondents. For ethnicity, white British participants were compared to white but not British, Black, Asian and other or unknown. The

18

Black and Asian categories included mixed Black and white or mixed Asian and white respectively to boost numbers. A location variable was used to contrast participants living in England with participants living in Scotland, Wales or Northern Ireland. To assess smoking characteristics, dependence was measured using the HSI and the tobacco products smoked variable contrasted participants who smoked factory-made only with those who also smoked roll-your-own or other tobacco products. Those who had never attempted to quit were compared with those who had made a quit attempt within the last six months and those who had made a quit attempt longer ago than six months. Those likely to quit in the next six months were compared with those likely to make an unsuccessful attempt and those unlikely to make a quit attempt. The analysis proceeded in the following stages. First, cigarette type was entered into the model. Second, each of the other fixed effects was entered individually. Third, if the fixed effect was significant (p<0.05) an interaction between the fixed effect and cigarette type was entered. Fourth, significant fixed effects were entered together in a multivariable model and multi-collinearity was checked by comparing standard errors in the multivariable model with the models in the second stage. This main effects model provides information about how participants generally answered questions about cigarettes. Fifth, significant interactions revealed in stage three were entered into the multivariable model in order to identify differences between cigarette types. Sixth, a cross classified variable was derived from cigarette type and SES in order to study the interaction. The cross classified variable included the following categories: ‘not low SES view of standard cigarette’, ‘not low SES view of cigarette with warning’, ‘not low SES view of green cigarette’, ‘low SES view of standard cigarette’, ‘low SES view of cigarette with warning’ and ‘low SES view of green cigarette’. Main effects of cigarette type and SES and the interaction term were removed from the model and replaced with the cross classified variable. In the final step the reference category of the cross classified variable was rotated so that where there were significant differences between cigarette types could be understood.

Pack inserts The seven insert variables were dichotomised into yes/agreement and no/disagreement/ neutral/not sure/don’t know. The dichotomised insert variables were the outcomes of the logistic regression models. The independent variables were gender, age, education, ethnicity, dependence (tertiles of HSI), tobacco product(s) smoked, previous quit attempt lasting at least 24 hours, and likely efficacy of a quit attempt in the next six months. Percentages in agreement were calculated. Age, gender and education (as a measure of SES) were entered into all models to account for any sampling inadequacies. Other variables were entered where p<0.10 in chi square tests. The models were assessed for multi-collinearity via comparison of standard errors and none was found.

4. Results

4.1 Cigarette design

For each item assessing harm, appeal and product trial, the cigarette with the warning and the green cigarette were rated as more undesirable than the standard cigarette. With respect to harm, participants were less likely to think that the standard cigarette looked harmful than the warning cigarette or green cigarette (p<0.001), and less likely to think that the standard cigarette made them think more about the dangers of smoking than the warning cigarette or green cigarette (p<0.001), see Table 2. Participants were also less likely to indicate that the green cigarette would make them think of the dangers of smoking than the warning cigarette

19

(p=0.01). In terms of appeal, participants were more likely to consider the standard cigarette attractive, and stylish, than the warning cigarette or green cigarette (both p<0.001). The standard cigarette was also considered to be nicer to be seen with, and more appealing to people their age, than the warning cigarette or green cigarette (both p<0.001). In terms of trial, whereas only 8.9% indicated that they would be unlikely to try a standard cigarette if offered by a friend, this was 45.4% for the warning cigarette and 66.5% for the green cigarette (both p<0.001). Similarly, while only 14.8% indicated that a never smoker their age would be unlikely to try a standard cigarette, this was 63.3% for the warning cigarette and 71.6% for the green cigarette (both p<0.001) - response frequencies presented in Appendix Table a3.1.

Table 2: Perceptions of cigarette design (harm, appeal, trial)

Standard Cigarette with Green cigarette warning cigarette %1 %1 %1 Harmful to health 38.8 69.1* 70.2* Think of dangers 20.9 58.1* # 53.5* Unattractive 25.2 61.7* 68.7* Unstylish 37.4 66.0* 69.4* Not nice to be seen with 19.8 55.2* 60.2* Not appealing to people my age 17.8 51.5* 57.4* Unlikely to try (personally) 8.9 45.4* 66.5* Unlikely to try (for never smokers) 14.8 63.3* 71.6* 1 Percentages shown indicate participants choosing one of the three points nearest the undesirable anchor on a seven-point semantic scale. * Significant difference in comparison to the standard cigarette (p<0.001) # Significant difference in comparison to the green cigarette (p<0.05)

For the two dissuasive cigarettes, approximately half the sample indicated that they would put people off starting to smoke (53.1% for cigarette with warning and 64.9% for the green cigarette) and make smokers want to stop (46.7% for cigarette with warning and 54.2% for the green cigarette). Just under half supported having a warning on all cigarettes (46.3%) with a similar proportion supporting all cigarettes being an unattractive colour (44.1%), see Table 2.

4.3 Characteristics associated with perceiving cigarettes as undesirable

Main effects multivariable multilevel logistic regression modelling suggested that in comparison to the standard cigarette, the warning cigarette (aOR=17.71; 95%CI 13.75-22.80) and green cigarette (aOR=30.88; 95%CI 23.98-39.76) were much more likely to be perceived as undesirable (i.e. less appealing, more harmful, less likely to be tried), see Table 3 - Appendix Tables A1.1 to A1.4 show the full results of the modelling. The model also indicated which smokers were more likely to rate the cigarettes as undesirable: women were more likely than men (aOR=1.30; 95%CI 1.10-1.54), and low SES more likely than those not low SES (aOR=1.26; 95%CI 1.06-1.50), to consider all three cigarettes undesirable. Compared to exclusive factory-made cigarette smokers, those who also smoked roll-your-own cigarettes (aOR=0.78; 95%CI 0.65-0.93) or other tobacco products (aOR=0.73; 95%CI 0.56- 0.93) were less likely to consider all three cigarettes undesirable. Those not likely to make a quit attempt in the next six months were less likely than those likely to make a quit attempt in the next six months (aOR=0.62; 95%CI 0.49-0.78) to consider all three cigarettes undesirable.

20

Only one significant interaction, between cigarette type and SES, was found (p<0.05). Both SES groups perceived the warning cigarette significantly more undesirable than the standard cigarette, and the green cigarette significantly more undesirable than the warning cigarette. Low SES participants were significantly more likely than those not low SES to perceive the standard cigarette (aOR=1.89; 95%CI 1.18-3.03) and green cigarette (aOR=1.43; 95%CI 1.13-1.80) as undesirable; there was no difference for the warning cigarette (aOR=0.99; 95%CI 0.78-1.25).

21

Table 3: Multilevel and multivariable modelling of perceiving cigarettes as undesirable (n=5298 cigarette evaluations, u=1766 participants)

Multivariable model Multivariable model + cigarette*SES interaction Odds ratio (95%CI) Odds ratio (95%CI) cons 0.05 (0.04 to 0.07) 0.04 (0.03 to 0.06) Cigarette type warning on cigarette (WC) 17.71 (13.75 to 22.80) 23.29 (16.40 to 33.08) 1 (ref = standard cigarette (SC) ) green cigarette (GC) 30.88 (23.98 to 39.76) 35.41 (24.93 to 50.29) Gender (ref = male) Female 1.30 (1.10 to 1.54) 1.30 (1.10 to 1.55) SES (ref = higher SES) low education AND/OR low economic status 1.26 (1.06 to 1.50) 1.89 (1.18 to 3.04) Ethnicity (ref = White British) White but not British 0.96 (0.72 to 1.30) 0.96 (0.72 to 1.30) Black (inc mixed black & white) 0.94 (0.62 to 1.42) 0.94 (0.62 to 1.42) Asian (inc mixed Asian & white) 0.79 (0.60 to 1.05) 0.79 (0.60 to 1.05) other or not declared 0.90 (0.58 to 1.42) 0.90 (0.57 to 1.42) Product category Factory-made and roll-your-own cigarettes 0.78 (0.65 to 0.90) 0.77 (0.64 to 0.93) (ref = Factory–made only) Factory-made cigarettes and other tobacco products (e.g. cigars, shisha, etc) 0.73 (0.56 to 0.93) 0.72 (0.56 to 0.93) Efficacy (ref = likely to quit) Not likely to make a quit attempt in next six months 0.62 (0.49 to 0.78) 0.61 (0.49 to 0.78) Likely to make unsuccessful attempt 1.05 (0.78 to 1.41) 1.05 (0.78 to 1.41) Interaction Cigarette type * SES WC*low SES 0.52 (0.31 to 0.87) (ref = SC*higher SES) GC*low SES 0.76 (0.46 to 1.26) Variation between participants (U(std err)) 1.14(0.11) 1.14(0.11)

Models varying reference category of cross classified variable1 Reference Category: SC not low SES SC low SES WC not low SES WC low SES GC not low SES Cigarette type & SES SC: not low SES 1 0.53 (0.33 to 0.85) 0.04 (0.03 to 0.06) 0.04 (0.03 to 0.06) 0.03 (0.02 to 0.04) SC: low SES 1.89 (1.18 to 3.03) 1 0.08 (0.06 to 0.12) 0.08 (0.06 to 0.12) 0.05 (0.04 to 0.08) WC: not low SES 23.13 (16.28 to 32.85) 12.21 (8.48 to 17.58) 1 1.01 (0.80 to 1.28) 0.66 (0.55 to 0.79) WC: low SES 22.83 (15.58 to 33.46) 12.05 (8.37 to 17.35) 0.99 (0.78 to 1.25) 1 0.65 (0.52 to 0.82) GC: not low SES 35.09 (24.71 to 49.84) 18.52 (12.86 to 26.67) 1.52 (1.27 to 1.81) 1.54 (1.22 to 1.94) 1 GC: low SES 50.15 (34.29 to 73.35) 26.47 (18.35 to 38.19) 2.17 (1.72 to 2.74) 2.20 (1.74 to 2.77) 1.43 (1.13 to 1.80) 1Variables included are those in the above models with the exception that cigarette type and SES are replaced with the cross classified variable

22

4.3 Pack inserts

Half the sample indicated that they would read the pack inserts and approximately three-fifths indicated that they would read them if interested in quitting (60%) and that inserts would be a good way to provide information about quitting (61%). A quarter thought that they would not read the inserts if interested in quitting, and that they were not a good way to provide information about quitting, see Table 4 - response frequencies are presented in Appendix Table A3.4. Just over half the sample strongly agreed/agreed that the inserts may make them think more about quitting (53%), help them if they decided to quit (52%), and that they are an effective way of encouraging smokers to quit (53%); less than a fifth of the sample strongly disagreed/disagreed with each (17-19%). Over half the sample strongly agreed/agreed that all packs should have inserts (55%), whereas a fifth strongly disagreed/disagreed.

Table 4: Perceptions of whether inserts would be read, are a good way to provide information, whether they would help smokers to think about quitting or quit, and support

Yes No Not sure % % % Would they be read 50 37 13 Would they be read if interested in quitting 60 25 15 Good way to provide information about 61 25 14 quitting

Agree Disagree Neither / % % Don’t know % Make you think more about quitting 53 18 29 Might help you if you decided to quit 52 19 29 Effective way of encouraging smokers to quit 53 17 30 All packs should have inserts 55 20 25

4.4 Perceptions of inserts: Sociodemographic and smoking related differences

4.4.1 Sociodemographic differences

Women were more likely than men to indicate that they would read inserts (aOR=1.24; 95%CI 1.02-1.50), and 25-34 year olds less likely than 16-19 year olds to think that they were a good way of providing information about quitting (aOR=0.76; 95%CI 0.60-0.98), see Table 5 and tables A2.1 to 2.5 in the Appendix. While education differences never reached significance in multivariable analysis they were significant in bivariable analysis (Appendix Table A2.1), with participants with GCSEs or higher more likely than those with less than GCSEs or no qualifications to indicate that they would read inserts (p<0.05), make them think about quitting (p<0.05), help them if they decided to quit (p<0.05) and are effective way of encouraging smokers to quit (p<0.05). In terms of ethnicity, compared with white British participants, white non-British (aOR=0.70; 95%CI 0.50-0.98) and Asian (aOR=0.67; 95%CI 0.49-0.92) participants were less likely to suggest that they would read inserts if trying to quit, white non-British (aOR=0.58; 95%CI 0.41-0.81) and Black (aOR=0.61; 95%CI 0.38-0.98) participants were less likely to indicate that inserts would make them think about quitting, and white non-

23

British (aOR=0.62; 95%CI 0.44-0.87) and Asian (aOR=0.70; 95%CI 0.51-0.96) participants were less likely to support having inserts in all packs.

4.4.2 Smoking related differences

Compared to exclusive factory-made cigarette smokers, those who also smoked roll-your-own cigarettes were more likely to indicate they would read inserts (aOR=1.35; 95%CI 1.09-1.66), read them if trying to quit (aOR=1.61; 95%CI 1.30-2.00), that they would make them think about quitting (aOR=1.31; 95%CI 1.06-1.62), help them if they decided to quit (aOR=1.31; 95%CI 1.06-1.61), and that they would be an effective way of encouraging smokers to quit (aOR=1.27; 95%CI 1.03-1.56), see Table 5. Compared to exclusive factory-made cigarette smokers, those who also smoked other tobacco products (e.g. cigars, shisha) were more likely to indicate they would read inserts if trying to quit (aOR=1.39; 95%CI 1.04-1.86) and that inserts might help them if they decided to quit (aOR=1.34; 95%CI 1.01-1.78). Participants who had made a quit attempt more than six months ago (aOR=1.30; 95%CI 1.00-1.69), or within the last six months (aOR=1.67; 95%CI 1.29-2.15), were more likely to indicate that they would read inserts than those who had never made a quit attempt. Those who had made a quit attempt in the last six months were also more likely than those who had never made a quit attempt to indicate that inserts were a good way to provide information about quitting (aOR=1.54; 95%CI 1.20-1.98), that they would read them if trying to quit (aOR=1.51; 95%CI 1.17-1.94), make them think about quitting (aOR=1.46; 95%CI 1.14-1.88), help them if they decided to quit (aOR=1.35; 95%CI 1.05-1.73), and that they would be an effective way of encouraging smokers to quit (aOR=1.33; 95%CI 1.04-1.71). Compared to those likely to make a successful quit attempt in the next six months (who responded that they were likely to quit within the next six months and that they would succeed), those unlikely to make a quit attempt in the next six months were less likely to indicate that they would read inserts (aOR=0.58; 95%CI 0.44-0.75), read them if trying to quit (aOR=0.74; 95%CI 0.55-0.99), that they would make them think about quitting (aOR 0.59; 95%CI 0.45-0.78), help them if they decided to quit (aOR=0.51; 95%CI 0.38-0.67), that they would be effective for smokers if they decided to quit (aOR=0.55; 95%CI 0.41-0.73), or support them (aOR=0.56; 95%CI 0.42-0.74). Compared to those likely to make a successful quit attempt in the next six months, those unlikely to make a successful quit attempt in the next six months were more likely to read inserts if trying to quit (aOR=1.43; 95%CI 1.00- 2.06), thought that they were a good way to provide information to smokers about quitting (aOR=1.46; 95%CI 1.02-2.08), and support them (aOR=1.43; 95%CI 1.00-2.04).

24

Table 5: Logistic regression models exploring perceptions of inserts by sociodemographic and smoking related characteristics1

(n=1766) Would read insert Would read insert Inserts make you Inserts might Inserts a good way Inserts are an All packs should if trying to quit think about help you quit of providing effective way of have inserts quitting information about encouraging quitting smokers to quit 2 Gender (ref = male) Female 1.24 (1.02 to 1.50) 1.11 (0.91 to 1.35) 0.98 (0.81 to 1.19) 0.95 (0.79 to 1.15) 1.13 (0.93 to 1.37) 0.88 (0.73 to 1.07) 1.20 (0.99 to 1.46)

Age (ref = 16-19) 20-24 1.16 (0.87 to 1.54) 0.88 (0.66 to 1.18) 1.18 (0.89 to 1.56) 1.19 (0.89 to 1.58) 0.87 (0.65 to 1.16) 0.97 (0.73 to 1.28) 0.96 (0.72 to 1.29) 25-34 1.25 (0.97 to 1.60) 0.83 (0.65 to 1.07) 0.99 (0.78 to 1.26) 1.18 (0.92 to 1.50) 0.76 (0.60 to 0.98) 0.88 (0.69 to 1.12) 0.84 (0.65 to 1.07)

Education (ref = GCSEs (or equivalent) or none) More than GCSEs (or equivalent) 1.25 (0.99 to 1.58) 1.12 (0.89 to 1.42) 1.22 (0.97 to 1.54) 1.21 (0.97 to 1.52) 1.12 (0.89 to 1.40) 1.19 (0.95 to 1.50) 1.10 (0.87 to 1.40)

Ethnicity (ref = White British) White but not British 0.70 (0.50 to 0.98) 0.58 (0.41 to 0.81) 0.62 (0.44 to 0.87) Black (inc mixed black & white) 0.92 (0.57 to 1.49) 0.61 (0.38 to 0.98) 0.99 (0.62 to 1.59) Asian (inc mixed Asian & white) 0.67 (0.49 to 0.92) 1.19 (0.87 to 1.63) 0.70 (0.51 to 0.96) other or not declared 0.84 (0.50 to 1.42) 1.06 (0.64 to 1.78) 1.08 (0.64 to 1.81)

Dependence (tertiles of HSI) (ref = lower dependence) Mid dependence 1.39 (1.11 to 1.76) 1.02 (0.80 to 1.29) Higher dependence 1.22 (0.94 to 1.59) 0.86 (0.66 to 1.12)

Tobacco products smoked (ref = only factory-made cigarettes) Factory-made and roll-your-own 1.35 (1.09 to 1.66) 1.61 (1.30 to 2.00) 1.31 (1.06 to 1.62) 1.31 (1.06 to 1.61) 1.27 (1.03 to 1.56) Factory-made and other 1.20 (0.90 to 1.59) 1.39 (1.04 to 1.86) 1.22 (0.92 to 1.63) 1.34 (1.01 to 1.78) 1.20 (0.91 to 1.60)

Quit attempt lasting at least 24 hours (ref = no) Yes, more than six months ago 1.30 (1.00 to 1.69) 1.12 (0.86 to 1.45) 1.20 (0.93 to 1.56) 1.05 (0.81 to 1.36) 1.16 (0.90 to 1.50) 1.07 (0.82 to 1.38) 0.78 (0.60 to 1.01) Yes, within the last six months 1.67 (1.29 to 2.15) 1.51 (1.17 to 1.94) 1.46 (1.14 to 1.88) 1.35 (1.05 to 1.73) 1.54 (1.20 to 1.98) 1.33 (1.04 to 1.71) 1.06 (0.82 to 1.37)

Efficacy of quit attempt in next 6 months (ref = likely to quit) Likely to make unsuccessful attempt 1.01 (0.72 to 1.40) 1.43 (1.00 to 2.06) 0.97 (0.69 to 1.37) 0.92 (0.65 to 1.29) 1.46 (1.02 to 2.08) 1.10 (0.78 to 1.55) 1.43 (1.00 to 2.04) Unlikely to make attempt 0.58 (0.44 to 0.75) 0.74 (0.55 to 0.99) 0.59 (0.45 to 0.78) 0.51 (0.38 to 0.67) 0.76 (0.57 to 1.01) 0.55 (0.41 to 0.73) 0.56 (0.42 to 0.74) 1 Blank cells indicate no significant relationship in bivariate analysis; 2 Odds ratios for reference categories are always 1.

25

5. Discussion

Few studies have explored the response of smokers to dissuasive cigarettes or pack inserts. We found that altering the appearance of the cigarette, either with the addition of a warning or by using a different colour, changed how they were perceived, and pack inserts with messages highlighting the benefits of quitting or promoting self-efficacy to do so were considered potentially beneficial for smokers.

5.1 Dissuasive cigarettes

Tobacco industry journals describe the cigarette as an increasingly important advertising medium for tobacco companies (Mapother, 2012). However, until recently, the focus has been on the potential of regulating the contents of cigarettes to reduce palatability or addictiveness (Warner, 2015), with the possibility of regulating the appearance of cigarettes and reducing its importance as a promotional tool completely overlooked. We found that young adult smokers perceived the two dissuasive cigarettes as significantly more harmful and less appealing than the standard cigarette. The dissuasive cigarettes were also thought to reduce likely product trial for themselves and never smokers. The harm, appeal and product trial items loaded onto a single ‘undesirability’ factor. Multivariable logistic regression modelling, with the undesirability factor as the outcome, found that for all participants the two dissuasive cigarettes were much more undesirable than the standard cigarette. However the cigarette with a warning had less impact on smokers with low SES. These findings are consistent with earlier research, where cigarettes with the warning ‘Smoking kills’ were considered a constant reminder of the associated harms and, partly due to the perceived discomfort of being observed by others smoking a cigarette displaying this message, unappealing for smokers (Moodie, Purves et al., 2015; Moodie, 2016; Moodie et al., 2016). Previous studies have also found unattractively coloured cigarettes to be perceived as more harmful than other cigarettes and also repellent, being a cigarette that young people did not think that others their age would use (Ford et al., 2014; Hoek & Robertson, 2015; Moodie et al., 2015; Gallopel-Morvan et al., in press). The findings are also very similar to an online survey with 16-24 year old smokers and non-smokers, where both smokers and non-smokers rated the warning cigarette and green cigarette as more unfavourable than the standard cigarette, and indicated that they would be less likely to try them (Moodie et al., in press). More than half of our sample thought that the two dissuasive cigarettes would put people off starting to smoke, with the green cigarette considered more likely than the cigarette with the warning ‘Smoking kills’ to do so (65% vs 53%). A cross-sectional in-home survey with 11-16 year olds from across the UK, which also explored perceptions of a cigarette with the warning ‘Smoking kills’, found that an even greater proportion (71%) indicated that this would put people off starting to smoke (Moodie et al., 2017). Approximately half our sample (53% for green cigarette vs 47% for cigarette with warning), and a similar figure for the in- home survey with children exploring the cigarette with warning (53%), indicated that these cigarettes would make smokers want to stop. These findings suggest that both of these dissuasive cigarettes are viewed by young people (children and young adults) as having a potentially stronger impact on non-smokers than smokers. Finally, we found a similar level of support for both the dissuasive cigarettes (46% for the cigarette with warning and 44% for the green cigarette). That almost half of smokers supported both (and only a quarter opposed them) requires further exploration, although this may, at least in part, reflect the fact that the dissuasive cigarettes were viewed as having the potential to encourage cessation and discourage initiation. At a population level, support for dissuasive cigarettes would almost certainly be higher, given that support for tobacco control

26 measures is generally higher for non-smokers than smokers (Hoek et al., 2012; Rosenberg et al., 2012). For instance, the in-home survey with 11-16 year olds found that support for the cigarette warning, which was 85% overall, increased from 51% for regular smokers to 77% for experimenters/past smokers and 89% for never smokers (Moodie et al., in press).

5.2 Pack inserts

For messaging on both the inside and outside of the pack, salience is a pre-requisite for effectiveness, as messages that are not noticed or read cannot be effective (Wogalter et al., 2002). Half our sample indicated that they would read the inserts, with that proportion increasing to three-fifths if they were interested in quitting. A longitudinal survey in Canada found that approximately a quarter of smokers (between 26-31%) at each wave reported reading them within the last month (Thrasher et al., 2015). In Canada, smokers who read the inserts were more likely to be female, younger, have a higher income, intend to quit or had recently tried to quit. In this study smokers indicating that they would read inserts were more likely to be female, white-British, smokers of factory-made cigarettes and also RYO, have moderate dependence according to the HSI, and intend to quit or had recently tried to quit. We found that approximately three-fifths (61%) thought that inserts were a good way to provide information about quitting to smokers, with 25% thinking that inserts were not a good way to provide information about quitting to smokers. An earlier study in Canada with smokers found that just under half indicated that the messaging on the inserts was a good way to provide information to smokers (48%), with a similar proportion (47%) thinking that they would not be a good way to provide information to smokers. Just over half our sample agreed/ strongly agreed that inserts may make them think more about quitting, help them if they decided to quit, and that they are an effective way of encouraging smokers to quit. Previous research in Canada found that 42% of smokers thought that inserts would encourage young smokers to quit (Environics, 2000), and in New Zealand 34% of smokers and recent quitters agreed/strongly agreed that inserts would be an effective way of encouraging reduced consumption or quitting (BRC Marketing and Social Research, 2004). There may be a number of reasons for the differences between our findings and earlier research, but design is very likely a relevant factor. Whereas the inserts used in earlier research were limited to text, the inserts used in this study (which have been used in Canada since 2012) included coloured graphics, which is very likely to have helped to enhance the impact. Indeed, a discrete choice experiment with adult smokers in the US, which explored insert design, found that inserts were rated as most helpful and motivating to quit when they included an image compared to when they did not include an image (Thrasher et al., 2018). These findings are consistent with the health communications and warnings literature, which demonstrates the importance of supporting text with pictorials (Wogalter et al., 2002; Houts et al., 2006; Hammond, 2011). Health communication theories, such as the Extended Parallel Process Model, predict that messages that combine threatening information with information that increases self- efficacy for behaviour change are most likely to result in positive behaviour change (Mead et al., 2015). If applied to smoking, the on-pack warnings, which often feature fear-appeal messages, can help to create high threat, but the challenge is building the self-efficacy required for smokers to take steps towards quitting (Ruiters et al., 2014). Pack inserts, with messages promoting self-efficacy, may help to achieve this, and would appear a meaningful complement to the on-pack warnings, as suggested in previous qualitative research (Moodie, 2016; Moodie, 2018). More than half our sample supported the inclusion of inserts promoting cessation inside every cigarette pack, with only a fifth opposing them. Neither the Standardised Packaging of Tobacco Products Regulations nor the Tobacco Products Directive (TPD)

27 requires tobacco companies to include health communication inserts, although member states are permitted to go beyond the TPD. In response to the consultation on the possible revision of the TPD a number of responses were received from governmental representatives, either presenting the view of the government in question or a specific ministry. There was high support among those participants for improving consumer information via mandatory pictorial health warnings, with those supportive arguing that additional information, such as the inclusion of pack inserts, would deliver more accurate information (European Commission, 2011). In January 2018 Philip Morris International placed adverts in newspapers in the UK explaining that their New Year’s resolution was ‘to stop selling cigarettes in the UK’, and that one of their commitments for 2018 was to ‘seek Government approval to insert, directly into our cigarette packs, information on quitting and on switching’ (Moodie, Hammond & Bauld, 2018). If there is support for inserts among governmental representatives, little opposition to them among smokers (the group most likely to be resistant), and a willingness to use them by some tobacco companies, then they are clearly a viable option for regulators.

5.3 Limitations

The cross-sectional design did not allow us to assess causality; that inserts and dissuasive cigarettes are not available on the UK market prevents more robust study designs such as longitudinal studies. Another potential limitation concerns the novelty of the stimuli, which may have influenced responses. While online surveys have been used for previous research exploring cigarette packaging, inserts and dissuasive cigarettes (Doxey & Hammond, 2011; White et al., 2012; Thrasher et al., 2015; Kotnowski et al., 2016; Hoek et al., 2016) and are a suitable survey mode for young adults, the use of an online panel limits the representativeness of our sample, particularly given literacy requirements and disparities in internet access. In addition, we only used four inserts, rather than the full set of eight used in Canada, which includes inserts less relevant to our sample. The use of semantic differential scales can also be criticised because answers can be subject to various response biases, although we attempted to diminish these through varying scale item direction and through our multivariate modelling methodology.

5.4 Further research

For governments legislation is typically driven by bodies of research rather than individual studies, and for tobacco companies multiple waves of research are commonly used to explore new concepts (Pollay & Dewhirst, 2003), so while this and other studies suggest that dissuasive cigarettes and pack inserts may have a role to play in encouraging , further research is warranted. We focused on young adults as they are a key target group, with standardised packaging and bans on tobacco marketing and the open display of tobacco at the point-of-sale within the retail environment generally thought to have the greatest impact among younger people. Research with even younger samples is critical. One possibility would be a survey with children exploring perceptions of a standard, novel (e.g. coloured, slimmer or capsule cigarettes) and dissuasive cigarettes. A previous study with children explored perceptions of a standard, novel (a slim pack, a brightly coloured pack, a pack with a novel opening) and dissuasive (plain) cigarette packs (Ford et al., 2013), with the plain pack viewed more negatively than the standard pack, which in turn was viewed more negatively than the novel packs. Quantitative research exploring the impact of inserts and dissuasive cigarettes with all adult smokers is needed. Previous qualitative research with younger, middle and older aged

28 adult smokers found that the pack inserts used in Canada were thought to capture attention, prolong the health message, supplement the on-pack warnings, and potentially alter the behaviour of some smokers, particularly those wanting to quit (Moodie, 2018). The perception that the inserts, and also the dissuasive cigarettes, could promote cessation was generally stronger among younger adults but also evident in middle aged and older adults (Moodie et al., 2016; Moodie, 2018), who are more likely to quit than younger groups (Dobbie et al., 2015). Aside from age, research exploring the role of ethnicity in response to inserts and dissuasive cigarettes would be fruitful, given that white British smokers in this study were, in general, more likely to view inserts as helpful for cessation and dissuasive cigarettes as undesirable. Marketing to, and research concerning, ethnic minorities requires an understanding of socio-demographic differences, language difficulties, culture and acculturation (Nwanko & Lindridge, 1998), and further research exploring the understanding of, and response to, dissuasive cigarettes and inserts among ethnic minorities would be of value.

5.5 Conclusions

It was argued, over two decades ago, that to offer greater protection to consumers cigarettes should come in plain packs with messaging on both the pack exterior and interior (Mahood, 1995). With the Standardised Packaging of Tobacco Products Regulations and Tobacco Products Directive being fully implemented in May 2017, this is a step closer, although there will still be no messaging on the pack interior. That more than half of the participants in this study suggested that inserts may help to promote cessation suggests that their inclusion in packs would be a meaningful supplement to the on-pack warnings. Our findings suggest however that to offer the greatest protection to consumers, it may be beneficial to supplement plain packaging and inserts with cigarettes designed to be dissuasive. Unattractively coloured cigarettes would complement the unattractively coloured packs, just as warnings on the cigarette (the primary pack) would extend the warnings on the cigarette pack (secondary packaging). Both options are clearly viable.

6. Acknowledgements

We would like to thank Research Now for undertaking data collection, Professor Kelvyn Jones for statistical advice and Professors James Thrasher and Linda Bauld for feedback on the survey design.

29

7. References

African Tobacco Control Alliance (2018). Sale of single sticks of cigarettes in Africa. Survey report from 10 capital cities. https://atca-africa.org/images/pdf/Atca-single-sticks/Report- Sale-of-Single-Sticks-in-Africa.pdf

Ampuero O, Vila N (2006). Consumer perceptions of product packaging. Journal of Consumer Marketing, 23, 100-12.

Banfield S (2015). Tasting tobacco. Tobacco Reporter, 11, 52-4

Beaglehole R, Bonita R, Yach F, Mackay J, Reddy KS (2015). A tobacco-free world: a call to action to phase out the sale of tobacco products by 2040. The Lancet, 385, 1011-8.

Blum A (1995). Irony in propaganda. Tobacco Control, 4, 117-8.

BRC Marketing and Social Research (2004). Smoking health warnings study. Wellington: Ministry of Health. http://www.tobaccolabels.ca/wp/wp-content/uploads/2013/12/NZ-2004- Effectiveness-of-Different-Health-Warnings-in-Helping-People-Consider-Their-Smoking- Related-Behaviour-Government-Report.pdf

CTFK (2017). The global cigarette industry. www.tobaccofreekids.org/assets/global/pdfs/en/Global_Cigarette_Industry_pdf.pdf

Cavalcante TM (2003). Labelling and packaging in Brazil. National Cancer Institute Health Ministry of Brazil. Geneva: World Health Organization.

Connolly G (2004). Sweet and spicy flavours: new brands for minorities and youth. Tobacco Control, 13, 211-2.

Congdon-Martin D (1992). Tobacco tins. A collector’s guide. Atglen, US: Shiffer Publishing Ltd.

Dobbie F, Hiscock R, Bee JL, Murray S, Shahab L, Aveyard P, Coleman T, McEwen T, McRobbie H, Purves R, Bauld L (2015). Evaluating long term outcomes of NHS Stop Smoking Services (ELONS): a prospective cohort study. Health Technology Assessment, 19, 1-156.

Doxey J, Hammond D (2011). Deadly in pink: The impact of female-oriented cigarette packaging among young women. Tobacco Control, 20, 353-60.

Dunn WL [Philip Morris] (1972). Motives and incentives in cigarette smoking. July, 1972. Bates No. 1003291922/1939.

Elliott R (2009). The early history of cigarettes in America. Brandstand, 34. http://cigarhistory.info/Cigarette_items/Cigarette-History.html

Environics (2000). Health warning messages on the flip/side and inserts of cigarette packaging. A survey of smokers. A report prepared for Health Canada.

30

Eriksen MP, Mackay J, Ross H (2012). The tobacco atlas (4th ed). Atlanta: American Cancer Society.

European Commission (2011). Report on the public consultation on the possible revision of the Tobacco Products Directive (2001/37/EC). Health and Consumers Directorate-General – Directorate D – Health systems and products D4 – Substances of human origin and Tobacco control. http://ec.europa.eu/health/tobacco/docs/consultation_report_en.pdf

Evans IO (2013). Cigarette cards and how to collect them. Worcestershire, England: Read Ltd.

Ford A, MacKintosh AM, Moodie C, Richardson S, Hastings G (2013). Cigarette pack design and adolescent smoking susceptibility: a cross-sectional survey. British Medical Journal Open, 3, e003282.

Ford A, Moodie C, Mackintosh AM, Hastings G (2014). Adolescent perceptions of cigarette appearance. European Journal of Public Health, 24, 464-8.

Gallopel-Morvan K, Moodie C, Eker F, Beguinot E, Martinet Y (2015). Perceptions of plain packaging among young adult roll-your-own smokers in France: A naturalistic approach. Tobacco Control, 24, e39-44.

Gallopel-Morvan K, Moodie C, Guignard R, Eker F, Beguinot E (in press). Consumer perceptions of cigarette design in France: A comparison of standard, innovative (slim, pink) and plain cigarettes. Nicotine & Tobacco Research.

Gfk Bluemoon (2011). Market testing of new health warnings and information messages for tobacco product packaging: Phase 3 refinement of health warnings. Sydney: Gfk Bluemoon.

Goswami H (2006). India: promoting tobacco via “research”. Tobacco Control, 15, 274-5.

Gyekye L (2013). Tobacco giant launches marketing campaign against plain packs. Packaging news. www.packagingnews.co.uk/news/tobacco-giantlaunches-marketing- campaignagainstplain-packs/

Hammond D (2011). Health warning messages on tobacco products: a review. Tobacco Control, 20, 327–37.

Hannah L (2006). Competition and productivity in tobacco: a qualification. The Journal of Industrial Economics, LIII, 1-10.

Harvey AD (2014). The cigarette as style accessory. Critical Quarterly, 56, 83-7.

Hassan L, Shui E (2015). No place to hide: two pilot studies assessing the effectiveness of adding a health warning to the cigarette cigarette. Tobacco Control, 24, e3-5.

Hedley D (2015). Tobacco wars: A new – the changing future. Tobacco Journal International, 3, 31-4.

31

Hellström D, Saghir M (2007). Packaging and logistics interactions in retail supply chains. Packaging Technology and Science, 20, 197–216.

Hoek J, Gendall P, Maubach N, Edwards R (2012). Strong public support for plain packaging of tobacco products. Australian and New Zealand Journal of Public Health, 36, 405-7.

Hoek J, Robertson C (2015). How do young adult female smokers interpret dissuasive cigarettes? Journal of Social Marketing, 5, 21–39.

Hoek J, Gendall P, Eckert C, Louviere J (2016). Dissuasive cigarette cigarettes: the next step in standardised (‘plain’) packaging? Tobacco Control, 25, 699-705.

Houts PS, Doak CC, Doak LG, Loscalzo MJ (2006). The role of pictures in improving health communication: A review of research on attention, comprehension, recall, and adherence. Patient Education and Counseling, 61, 173–190.

HSCIC (2014). Statistics on smoking: England 2014. Health and Social Care Information Centre.

Johnson B (undated). Cigarette cards and cartophily. www.historic- uk.com/CultureUK/Cigarette-Cards-Cartophily/

Kluger R (1996). Ashes to ashes: America’s hundred-year cigarette war, the public health, and the unabashed triumph of Philip Morris. New York: Alfred A. Knopf.

Kotnowksi K, Fong GT, Gallopel-Morvan K, Islam T, Hammond D (2016). The impact of cigarette packaging design among young females in Canada: Findings from a discrete choice experiment. Nicotine & Tobacco Research, 18, 1348-56.

Kozlowski LT, Porter CQ, Orleans T, Pope MA, Heatherton T (1994). Predicting smoking cessation with self-reported measures of nicotine dependence: FTQ, FTND, and HSI. and Dependence, 34, 211-6.

Lal P, Kumar R, Ray S, Sharma N, Bhattarcharya B, Mishra D, Sinha MK, Christian A, Rathinam A, Singh G (2015). The single cigarette economy in India - a back of the survey to estimate its magnitude. Asian Pacific Journal of Cancer Prevention, 16, 5597-82.

Lambat I (2015). Tobacco 2015: Could it be a watershed year. Tobacco International, 2, 16- 21.

Leading Edge (1997). American blends NPD. Qualitative debrief presentation 1997. www.tobaccopapers.com/PDFs/0400-0499/0479.pdf

Lee YJ (2006). The study and conservation of woven cigarette inserts. Journal of the Korean Society of Clothing and Textiles, 30:447-57.

Leonard B (2015). The small roots of big tobacco. Tobacco Reporter, 9, 104-14.

London Cigarette Company (1982). The catalogue of international cigarette cards. Exeter, England: Webb & Bower.

32

M&C Saatchi (1998). Sovereign and the cheap cigarette market qualitative research debrief (Version 1 of 2). www.tobaccopapers.com/PDFs/0500-0599/0564.pdf

Mahood G (1995). Canada tobacco package warning system. Tobacco Control, 4, 10-4.

Mahood G (2004). Canada’s tobacco package or warning system: “telling the truth” about tobacco product risks. Geneva: WHO.

Mapother J (2012). Putting a shine on tipping. Tobacco Journal International, 2, 77-83.

Mawditt N (2006). Putting pack opportunities into the frame. World Tobacco, 212, 36–7.

McDaniel PA, Smith EA, Malone RE (2016). The tobacco endgame: a qualitative review and synthesis. Tobacco Control, 25, 594-604.

Mead EL, Cohen JE, Kennedy CE, Gallo J, Latkin CA (2015). The role of theory-driven graphic warning in motivation to quit: a qualitative study on perceptions from low- income, urban smokers. BMC Public Health, 15, 92.

Moodie C (2013). Commentary on Kotnowski & Hammond (2013): The importance of cigarette packaging innovation. Addiction, 108, 1669-70.

Moodie C (2016). Novel ways of using tobacco packaging to communicate health messages: Interviews with packaging and marketing experts. Addiction Research & Theory, 24, 54-61.

Moodie C (2018). Adult smokers’ perceptions of cigarette pack inserts promoting cessation: A focus group study. Tobacco Control, 27, 72-77.

Moodie C, Ford A, Mackintosh AM, Hastings GB (2012). Young people’s perceptions of cigarette packaging and plain packaging: An online survey. Nicotine & Tobacco Research, 14, 98-105.

Moodie C, Stead M, Bauld L, McNeill A, Angus K, Hinds K, Kwan I, Thomas J, Hastings G, O’Meara A (2012). : A systematic review. Report prepared for the Department of Health. Stirling: Centre for Tobacco Control Research, University of Stirling.

Moodie C, Ford A, Mackintosh AM, Purves R (2015). Are all cigarettes just the same? Female’s perceptions of slim, coloured, aromatised and capsule cigarettes. Health Education Research, 30, 1-12.

Moodie C, Purves R, McKell J, de Andrade M (2015). Novel means of using cigarette packaging and cigarettes to communicate health risk and cessation messages: A qualitative study. International Journal of Mental Health and Addiction, 13, 333-44.

Moodie C, Mackintosh AM, Gallopel-Morvan K, Hastings G, Ford A (2017). Adolescents’ perceptions of health warnings on cigarettes. Nicotine & Tobacco Research, 29: 1232-1237.

Moodie C, Hammond D, Bauld L (2018). Philip Morris International: A New Year’s resolution. Tobacco Control, 27, e79-80.

33

Moodie C, Gendall P, Hoek J, Mackintosh AM, Best C, Murray S (in press). The response of young adult smokers and non-smokers in the United Kingdom to dissuasive cigarettes: An online survey. Nicotine & Tobacco Research.

Mucan B, Moodie C (in press). Young adult’s perceptions of plain packs, numbered packs and pack inserts in Turkey: A focus group study. Tobacco Control.

Mullen C (1979). Cigarette pack art. London: Hamlyn.

Nwankwo S, Lindridge A (1998). Marketing to ethnic minorities in Britain. Journal of Marketing Practice: Applied Marketing Science, 4, 200-16.

Off Licence News (2002). Advert for Hamlet miniatures. Off Licence News, 29 March 2002.

Office for National Statistics (2015). Statistical bulletin: Internet users, 2015. London, UK: Office for National Statistics.

Oostendorp JA, Bode JM, Lutters E, van Houten FJAM (2006). The (development) life cycle for packaging and the relation to product design. 13th CIRP International Conference on Life cycle Engineering 2006. http://portfolio.io.utwente.nl/student/bodejm/packaging%20paper.pdf

Pollay R, Dewhirst T (2003). A Premiere example of the illusion of cigarettes in the 1990s. Tobacco Control, 12, 322-32.

Prendergast G, Pitt L (1996). Packaging, marketing, logistics and the environment: are there trade‐offs? International Journal of Physical Distribution & Logistics Management, 26, 60-72.

Pringle TR (1991). Cold comfort: the polar landscape in English & American popular culture 1845–1990. Landscape Research, 16, 43-8.

Pritcher L (undated). More about tobacco advertising and the tobacco collections. http://library.duke.edu/digitalcollections/eaa/guide/tobacco/

Proctor R (2011). Golden holocaust. California: University of California Press.

Rasbash J, Browne W, Healy M, Cameron B, Charlton C (2015). MLwiN Version 2.33 Centre for Multilevel Modelling, University of Bristol.

Rosenberg M, Pettigrew S, Wood L, Ferguson R, Houghton S (2012). Public support for tobacco control policy extensions in Western Australia: a cross-sectional study. British Medical Journal Open, 2, e000784.

Rowse EJ, Fish LJ (2005). Fundamentals of advertising. Whitefish, MT: Kessinger Publishing.

Ruiters RAC, Kessels LTE, Peters G-YJ, Kok G (2014). Sixty years of fear appeal research: Current state of the evidence. International Journal of Psychology, 49, 63-70.

34

Shechter R (2003). Selling luxury: The rise of the Egyptian cigarette and the transformation of the Egyptian tobacco market, 1850-1914. International Journal of Middle East Studies, 35, 51–75.

Selin H, Sweanor D (1998). Tobacco product packaging and labelling. Health Science Analysis Project, Policy Analysis No. 12. Washington, DC: Advocacy Institute.

Shaw M, Mitchell R, Dorling D (2000). Time for a smoke? One cigarette reduces your life by 11 minutes. BMJ, 320, 53.

Simms C, Trott P (2010). Packaging development: a conceptual framework for identifying new product development. Marketing Theory, 10, 397-415.

Simpson A (2000). Doral revives collector cards with purchase. Tobacco Control, 9.

Simpson D (2004). USA/Brazil: the flavour of things to come? Tobacco Control, 13, 105-6.

Solmajer L (2010). Slovenia: New brand claims to be "no brand". Tobacco Control, 19, 263.

Tandemar Research Inc (1992). Tobacco health warning messages, inserts and toxic constituents information study - Final Report. Toronto: Tandemar Research Inc.

Thrasher JT, Osman A, Abad EN, Hammond D, Bansal-Travers M, Cummings KM, Hardin J, Moodie C (2015). The innovative use of cigarette package inserts to supplement pictorial health warnings: An evaluation of the Canadian policy. Nicotine & Tobacco Research, 17, 870-5.

Thrasher JF, Swayampakala K, Cummings KM, Hammond D, Anshari D, Krugman DM, Hardin JW (2016). Cigarette package inserts can promote efficacy beliefs and sustained smoking cessation attempts: A longitudinal assessment of an innovative policy in Canada. Preventive Medicine, 88, 59–65.

Thrasher JF, Anshari D, Lambert-Jessup V, Islam F, Mead E, Popova L, Salloum R, Moodie C, Louviere J, Lindblom EN (2018). Assessing smoking cessation messages with a discrete choice experiment. Tobacco Regulatory Science, 4, 73-87.

Wardle H, Pickup D, Lee L, Hall J, Pickering K, Grieg K, Moodie C, Mackintosh AM (2010). Evaluating the impact of picture health warnings on cigarette packets. London: Public Health Research Consortium.

Warner K (2015). The national and international regulatory environment in tobacco control. Public Health Research & Practice, 25, e2531527.

Wherry AE, McCray CA, Adedeji-Fajobi TI, Sibiya X, Ucko P, Lebina L, Golub JE, Cohen JE, Martinson NA (2014). A comparative assessment of the price, brands and pack characteristics of illicitly traded cigarettes in five cities and towns in South Africa. British Medical Journal Open, 4, e004562.

35

White CM, Hammond D, Thrasher JF, Fong GT (2012). The potential impact of plain packaging of cigarette products among Brazilian young women: an experimental study. BMC Public Health, 12, 737.

Wogalter MS, Conzola VC, Smith-Jackson TL (2002). Research-based guidelines for warning design and evaluation. Applied Ergonomics, 33, 219–230.

World Health Organisation. (2005). WHO framework convention on tobacco control. Geneva, Switzerland: World Health Organisation.

Yan Q (2010). Storytelling, branding, and debranding in a localized context: Revisiting cigarette advertising strategies by BAT (British and American Tobacco Company) in China (1902-1949). Society for Marketing Advances Proceedings, 128-32. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.411.8423&rep=rep1&type=pdf#pa ge=163

36

Appendix

Contents Section 1 Multivariate models of perceptions of cigarette type ...... 38 Section 2 Pack inserts analysis...... 47 Section 3 Distributions of variables used in the analysis ...... 54 Histograms of cigarette opinions factor ...... 56

37

Section 1 Multivariate models of perceptions of cigarette type

Table A1.1: Main effects models

Odds ratio (95%CI) SOCIODEMOGRAPHIC CHARACTERISTICS BASE MODEL cons 0.038 (0.030 to 0.049)

Cigarette type standard cigarette 1 health warning on cigarette 17.219 (13.398 to 22.129) green cigarette 29.874 (23.246 to 38.393) Variation between individuals (U(std err)) 1.187(0.109)

G ENDER cons 0.033 (0.025 to 0.042)

Cigarette type standard cigarette 1 health warning on cigarette 17.340 (13.492 to 22.284) green cigarette 30.144 (23.410 to 38.816)

Gender Male 1 Female 1.346 (1.137 to 1.593)1

Variation between individuals (U(std err)) 1.188(0.110)

A GE cons 0.037 (0.028 to 0.049)

Cigarette type standard cigarette 1 health warning on cigarette 17.374 (13.493 to 22.373) green cigarette 30.235 (23.480 to 38.933)

Age 16-19 1 20-24 1.127 (0.879 to 1.446) 25-34 0.994 (0.806 to 1.226)

Variation between individuals (U(std err)) 1.213(0.111)

38

Odds ratio (95%CI) S OCIOECONOMIC STATUS cons 0.035 (0.027 to 0.045)

Cigarette type standard cigarette 1 health warning on cigarette 17.357 (13.479 to 22.350) green cigarette 30.205 (23.457 to 38.894)

Socioeconomic status No indication of low socioeconomic status 1 (GCSE or no qualifications) &/or (routine or manual 1.260 (1.058 to 1.500)1 work or long term unemployed, sick or disabled)

Variation between individuals (U(std err)) 1.207(0.111)

ETHNICITY cons 0.040 (0.031 to 0.051)

Cigarette type standard cigarette 1 health warning on cigarette 17.409 (13.546 to 22.374) green cigarette 30.265 (23.550 to 38.896)

Ethnicity White British 1 White but not British 0.946 (0.702 to 1.274) Black (inc mixed black & white) 0.843 (0.556 to 1.277) Asian (inc mixed Asian & white) 0.705 (0.533 to 0.931)1 other or not declared 0.854 (0.543 to 1.343)

Variation between individuals (U(std err)) 1.203(0.111)

L OCATION cons 0.037 (0.029 to 0.047)

Cigarette type standard cigarette 1 health warning on cigarette 17.322 (13.452 to 22.306) green cigarette 30.114 (23.387 to 38.777)

Location England 1 Scotland 1.182 (0.832 to 1.678) Wales 1.055 (0.690 to 1.615) Northern Ireland 1.508 (0.820 to 2.775) 39

Odds ratio (95%CI) Variation between individuals (U(std err)) 1.207(0.111)

SMOKING RELATED CHARACTERISTICS DEPENDENCE cons 0.038 (0.030 to 0.048)

Cigarette type standard cigarette 1 health warning on cigarette 17.340 (13.492 to 22.284) green cigarette 30.144 (23.410 to 38.816)

HSI (heaviness of smoking index) higher dependence 0.844 (0.312 to 2.279)

Variation between individuals (U(std err)) 1.207(0.111)

T OBACCO PRODUCTS SMOKED cons 0.045 (0.035 to 0.057)

Cigarette type standard cigarette 1 health warning on cigarette 17.392 (13.506 to 22.395) green coloured cigarette 30.235 (23.480 to 38.933)

Tobacco products smoked Factory-made only 1 Factory-made and roll-your-own cigarettes 0.759 (0.631 to 0.912)1 Factory-made cigarettes and other tobacco products 0.678 (0.528 to 0.872)1 (e.g. cigars, shisha, etc)

Variation between individuals (U(std err)) 1.190(0.110)

P AST QUIT ATTEMPT HISTORY cons 0.038 (0.029 to 0.050)

Cigarette type standard cigarette 1 health warning on cigarette 17.236 (13.412 to 22.151) green cigarette 29.904 (23.269 to 38.432)

Ever made a quit attempt lasting 24 hours No 1 Yes, within the last six months 1.049 (0.847 to 1.299) Yes, more than six months ago 0.975 (0.775 to 1.227)

40

Odds ratio (95%CI) Variation between individuals (U(std err)) 1.189(0.109)

EFFICACY OF FUTURE QUIT ATTEMPT cons 0.050 (0.037 to 0.068)

Cigarette type standard cigarette 1 health warning on cigarette 17.462 (13.587 to 22.441) green cigarette 30.356 (23.574 to 39.089)

Efficacy of future quit attempt in next 6 months Likely to quit 1 Likely to make an unsuccessful attempt 1.111 (0.831 to 1.485) Unlikely to make attempt 0.639 (0.505 to 0.808)1

Variation between individuals (U(std err)) 1.162(0.109) 1Sociodemographic or smoking related characteristic significantly associated with dissuasiveness of opinion about cigarettes and thus interaction explored (see Table A1.2)

41

Table A1.2: Interactions between cigarette type and sociodemographic and smoking related characteristics with significant main effects

Odds ratio (95%CI) SOCIODEMOGRAPHIC CHARACTERISTICS GENDER cons 0.038 (0.027 to 0.053)

Cigarette type standard cigarette 1 health warning on cigarette 14.556 (10.169 to 20.836) green coloured cigarette 26.390 (18.472 to 37.702)

Gender Male 1 Female 1.034 (0.643 to 1.661) health warning on cigarette.Female 1.405 (0.849 to 2.325) green cigarette.Female 1.294 (0.782 to 2.142)

Variation between individuals (U(std err)) 1.195(0.110)

S OCIOECONOMIC STATUS cons 0.029 (0.021 to 0.041)

Cigarette type standard cigarette 1 health warning on cigarette 22.714 (15.993 to 32.260) green cigarette 34.398 (24.220 to 48.854)

SES No indication of low socioeconomic status 1 (GCSE or no qualifications) &/or (routine or manual work or long term unemployed, sick or disabled) 1.870 (1.164 to 3.005) health warning on cigarette. low SES 0.529 (0.319 to 0.880)1 green cigarette. low SES 0.767 (0.462 to 1.275) Variation between individuals (U(std err)) 1.202(0.111)

E THNICITY cons 0.035 (0.026 to 0.047)

Cigarette type standard cigarette 1 health warning on cigarette 20.005 (14.735 to 27.160) green cigarette 35.304 (25.953 to 48.025)

42

Odds ratio (95%CI)

Ethnicity White British 1 White but not British 2.022 (1.028 to 3.976) Black (inc mixed black & white) 1.777 (0.678 to 4.661) Asian (inc mixed Asian & white) 0.944 (0.422 to 2.108) other or not declared 0.271 (0.026 to 2.791) health warning on cigarette. White but not British 0.472 (0.225 to 0.990)1 green cigarette. White but not British 0.366 (0.175 to 0.769)1 health warning on cigarette. Black (inc mixed black & white) 0.407 (0.142 to 1.167) green cigarette. Black (inc mixed black & white) 0.443 (0.155 to 1.266) health warning on cigarette. Asian (inc mixed Asian & white) 0.725 (0.309 to 1.702) green cigarette. Asian (inc mixed Asian & white) 0.719 (0.307 to 1.683) health warning on cigarette. Other or not declared 3.155 (0.294 to 33.871) green cigarette. Other or not declared 3.615 (0.336 to 38.882) Variation between individuals (U(std err)) 1.199(0.110)

P RODUCT cons 0.050 (0.037 to 0.068)

Cigarette type standard cigarette 1 health warning on cigarette 15.816 (11.290 to 22.156) green coloured cigarette 25.662 (18.318 to 35.950)

Tobacco Products smoked Factory-made only 1 Factory-made and roll-your-own cigarettes 0.599 (0.354 to 1.012) Factory-made cigarettes and other tobacco products (e.g. cigars, shisha, etc) 0.529 (0.244 to 1.148) health warning on cigarette. Factory-made and roll-your-own cigarettes 1.234 (0.706 to 2.157) green cigarette. Factory-made and roll-your-own cigarettes 1.383 (0.791 to 2.417) health warning on cigarette. Factory-made cigarettes and other tobacco products (e.g. cigars, shisha, etc) 1.218 (0.539 to 2.752) green cigarette. Factory-made cigarettes and other tobacco products (e.g. cigars, shisha, etc) 1.426 (0.631 to 3.223) Variation between individuals (U(std err)) 1.178(0.110)

43

Odds ratio (95%CI) EFFICACY OF FUTURE QUIT ATTEMPT cons 0.065 (0.040 to 0.105) Cigarette type standard cigarette 1 health warning on cigarette 15.518 (9.123 to 26.395) green cigarette 18.634 (10.934 to 31.757)

Efficacy of quit attempt Likely to quit 1 Likely to make an unsuccessful attempt 0.462 (0.258 to 0.825) Not likely to make an attempt 0.794 (0.393 to 1.604) health warning on cigarette. no attempt 1.142 (0.609 to 2.143) green coloured cigarette. no attempt 1.855 (0.987 to 3.487) health warning on cigarette. failed attempt 1.298 (0.604 to 2.788) green cigarette. failed attempt 1.694 (0.786 to 3.652)

Variation between individuals (U(std err)) 1.167(0.110) 1Sociodemographic or smoking related characteristic significantly associated with dissuasiveness of opinion about cigarettes and thus interaction explored in multivariable modelling if main effect significant in multivariable model (see table 3))

44

Table A1.3: Multivariate model with cross classified variable cigarette type* low SES

Odds ratio (95%CI) cons 0.041 (0.028 to 0.062) Gender male 1 Female 1.302 (1.098 to 1.544)

Ethnicity White British 1 White but not British 0.965 (0.719 to 1.294) Black (inc mixed black & white) 0.935 (0.617 to 1.417) Asian (inc mixed Asian & white) 0.791 (0.597 to 1.046) other or not declared 0.902 (0.576 to 1.413)

Tobacco products smoked Factory-made only 1 Factory-made and roll-your-own cigarettes 0.775 (0.645 to 0.932) Factory-made cigarettes and other tobacco products (e.g. cigars, 0.725 (0.564 to 0.931) shisha, etc)

Efficacy of quit attempt Likely to quit 1 Likely to make an unsuccessful attempt 0.615 (0.486 to 0.778) Unlikely to make attempt 1.049 (0.783 to 1.405)

Interaction between cigarette type & low SES reg not low 1 warn not low 23.127 (16.284 to 32.846) green not low 35.093 (24.709 to 49.841) reg low 1.891 (1.179 to 3.032) warn low 22.828 (15.577 to 33.455) green low 50.149 (34.287 to 73.350) Variation between individuals (U(std err)) 1.117(0.108)

45

Table A1.4: Odds ratios of cross classified variable cigarette type* SES varying the reference category Reference Category: Standard not low SES Standard low SES Warning not low SES Warning low SES Green not low SES

Cig/SES Standard not low SES 1 0.529 (0.330 to 0.848) 0.043 (0.030 to 0.061) 0.044 (0.030 to 0.064) 0.029 (0.020 to 0.041) low SES 1.891 (1.179 to 3.032) 1 0.082 (0.057 to 0.118) 0.083 (0.058 to 0.119) 0.054 (0.037 to 0.078)

Warning not low SES 23.127 (16.284 to 32.846) 12.207 (8.478 to 17.576) 1 1.013 (0.802 to 1.279) 0.659 (0.552 to 0.786) low SES 22.828 (15.577 to 33.455) 12.049 (8.368 to 17.349) 0.990 (0.780 to 1.250) 1 0.651 (0.515 to 0.821)

Green not low SES 35.093 (24.709 to 49.841) 18.523 (12.864 to 26.671) 1.517 (1.272 to 1.810) 1.537 (1.217 to 1.941) 1 low SES 50.149 (34.287 to 73.350) 26.470 (18.347 to 38.188) 2.171 (1.719 to 2.741) 2.197 (1.740 to 2.774) 1.430 (1.130 to 1.800) All variables included in models (see table A1.3).

46

Section 2 Pack inserts analysis

Table A2.1: Sociodemographic and smoking related characteristic assocations1 with opinions on pack inserts – bivariable analysis

would read Inserts would Inserts Inserts a good inserts are an all packs would insert if make you would help if way to provide effective way of should read interested in think about you decided information about encouraging have N insert2 quitting2 quitting3 to quit3 quitting2 smokers to quit3 inserts4 Gender Male 888 47.7% 59.2% 52.7% 53.6% 61.8% 54.8% 52.5% Female 878 52.1% 61.7% 51.7% 51.8% 59.9% 51.4% 57.2% p .071 .306 .711 .483 .438 .157 .053

Age 16-19 413 46.2% 64.9% 52.5% 50.8% 57.6% 55.9% 58.8% 20-24 401 50.9% 61.6% 56.1% 54.4% 66.1% 54.6% 56.9% 25-34 952 51.1% 58.1% 50.4% 52.8% 60.1% 51.3% 52.2% p .239 .054 .158 .600 .036 .224 .050

SES Not low SES 1114 50.4% 61.8% 53.6% 53.6% 61.3% 54.3% 55.7% Low SES 652 48.9% 58.3% 49.8% 51.2% 60.1% 51.1% 53.2% p .570 .164 .141 .362 .658 .206 .328

Education More than GCSEs (or equivalent) 1357 51.3% 61.2% 53.6% 54.2% 61.5% 54.5% 55.6% GCSEs (or equivalent) or none 409 45.2% 57.9% 47.4% 47.7% 58.7% 48.7% 52.1% p .037 .256 .032 .023 .328 .045 .226

Economic status Not low status 1350 49.2% 60.9% 52.7% 52.4% 61.0% 52.7% 55.4% Routine or manual occupation, long term 52.2% 59.1% 50.5% 53.8% 60.6% 54.6% 52.9% unemployed or permanently sick or disabled 416 p .314 .560 .453 .638 .933 .533 .397

47

would read Inserts would Inserts Inserts a good inserts are an all packs would insert if make you would help if way to provide effective way of should read interested in think about you decided information about encouraging have N insert2 quitting2 quitting3 to quit3 quitting2 smokers to quit3 inserts4 Ethnicity White British 1264 49.5% 62.7% 53.4% 53.9% 61.4% 54.3% 56.9% White but not British 162 49.4% 53.7% 40.1% 46.3% 57.4% 43.8% 45.1% Black (inc mixed black & white) 79 53.2% 62.0% 43.0% 48.1% 64.6% 50.6% 57.0% Asian (inc mixed Asian & white) 196 52.0% 51.5% 57.1% 53.1% 61.2% 54.1% 47.4% other or not declared 65 47.7% 58.5% 55.4% 50.8% 53.8% 53.8% 58.5% p .923 .014 .005 .387 .597 .160 .010

Location England 1550 49.5% 59.9% 51.7% 52.6% 60.6% 53.2% 54.1% Scotland 109 53.2% 61.5% 54.1% 49.5% 55.0% 49.5% 52.3% Wales 73 54.8% 68.5% 56.2% 57.5% 69.9% 54.8% 65.8% Northern Ireland 34 47.1% 64.7% 61.8% 58.8% 73.5% 58.8% 70.6% p .707 .484 .564 .650 .095 .780 .057

Dependence (tertiles of HSI) lower dependence 601 46.4% 60.7% 50.2% 51.2% 58.4% 52.4% 57.2% mid dependence 675 53.8% 61.3% 53.3% 55.3% 63.3% 54.8% 55.6% higher dependence 490 48.8% 59.0% 53.1% 51.0% 60.6% 51.6% 50.8% p .027 .711 .494 .242 .205 .513 .094

Tobacco products smoked Only factory-made (packet) cigarettes 813 46.5% 54.9% 48.8% 49.2% 58.4% 49.9% 54.0% Factory-made and roll-your-own cigarettes 681 53.5% 66.5% 55.2% 55.2% 62.7% 55.9% 57.1% Factory-made cigarettes and other tobacco 51.1% 62.1% 54.8% 57.0% 63.6% 55.5% 51.5% products (e.g. cigars, shisha, etc) 272 p .025 <.001 .032 .021 .146 .047 .233

Quit attempt lasting at least 24 hours No 426 40.1% 53.1% 44.4% 46.2% 52.1% 46.7% 52.6% Yes, more than six months ago 552 48.7% 56.9% 49.8% 49.1% 60.5% 49.3% 48.0% Yes within the last six months 788 56.0% 67.0% 58.1% 58.8% 65.9% 59.3% 60.8% p <.001 <.001 <.001 <.001 <.001 <.001 <.001

48

would read Inserts would Inserts Inserts a good inserts are an all packs would insert if make you would help if way to provide effective way of should read interested in think about you decided information about encouraging have N insert2 quitting2 quitting3 to quit3 quitting2 smokers to quit3 inserts4 Efficacy of quit attempt in next six months Likely to quit 276 59.1% 65.9% 62.7% 65.9% 65.6% 64.5% 64.9% Likely to make an unsuccessful attempt 304 59.2% 72.7% 61.2% 63.2% 70.7% 65.5% 70.7% Unlikely to make attempt 1186 45.4% 56.1% 47.5% 47.0% 57.3% 47.3% 48.4% p <.001 <.001 <.001 <.001 <.001 <.001 <.001

Total 49.9 60.5 52.2 52.7 60.9 53.1 54.8 1Chi square tests with continuity corrections for 2 by 2 tables 2Responded ‘yes’ compared with ‘no’ and ‘unsure’ 3Responded ‘agree’ or ‘strongly agree’ compared with ‘neither’, ‘disagree’, ‘strongly disagree’ or ‘don’t know’ 4Responded 1 or 2 on a 5 point Likert scale indicating that they were closer to the opinion that ‘All packs should have inserts’ rather than ‘NO packs should have inserts’

49

Table A2.2: Logistic regression modelling of associations between sociodemographic and smoking related characteristics and reading pack inserts

Would read insert Would read if trying to quit bivariable multivariable bivariable multivariable

Gender

Male 1 1 1 1 Female 1.19 (0.99 to 1.43) 1.24 (1.02 to 1.50) 1.11 (0.92 to 1.34) 1.11 (0.91 to 1.35)

Age

16-19 1 1 1 1 20-24 1.20 (0.91 to 1.58) 1.16 (0.87 to 1.54) 0.87 (0.65 to 1.15) 0.88 (0.66 to 1.18)

25-34 1.21 (0.96 to 1.53) 1.25 (0.97 to 1.60) 0.75 (0.59 to 0.95) 0.83 (0.65 to 1.07)

Education

GCSEs (or equivalent) or none 1 1 1 1

More than GCSEs (or equivalent) 1.27 (1.02 to 1.59) 1.25 (0.99 to 1.58) 1.15 (0.92 to 1.44) 1.12 (0.89 to 1.42)

Ethnicity White British 1 1

White but not British 0.69 (0.50 to 0.96) 0.70 (0.50 to 0.98)

Black (inc mixed black & white) 0.97 (0.61 to 1.55) 0.92 (0.57 to 1.49)

Asian (inc mixed Asian & white) 0.63 (0.47 to 0.85) 0.67 (0.49 to 0.92)

other or not declared 0.84 (0.50 to 1.39) 0.84 (0.50 to 1.42)

Dependence (tertiles of HSI)

lower dependence 1 1

mid dependence 1.34 (1.08 to 1.67) 1.39 (1.11 to 1.76)

higher dependence 1.10 (0.87 to 1.40) 1.22 (0.94 to 1.59)

Tobacco products smoked

Only factory-made 1 1 1 1

Factory-made and roll-your-own 1.32 (1.08 to 1.62) 1.35 (1.09 to 1.66) 1.63 (1.32 to 2.02) 1.61 (1.30 to 2.00) Factory-made cigarettes and other 1.20 (0.91 to 1.58) 1.20 (0.90 to 1.59) 1.35 (1.02 to 1.79) 1.39 (1.04 to 1.86)

Quit attempt lasting at least 24 hrs

No 1 1 1 1 Yes, more than six months ago 1.42 (1.10 to 1.83) 1.30 (1.00 to 1.69) 1.17 (0.91 to 1.51) 1.12 (0.86 to 1.45)

Yes within the last six months 1.90 (1.49 to 2.41) 1.67 (1.29 to 2.15) 1.80 (1.41 to 2.29) 1.51 (1.17 to 1.94)

Efficacy of quit attempt in next 6 months Likely to quit 1 1 1 1

Likely to make unsuccessful attempt 1.01 (0.72 to 1.40) 1.01 (0.72 to 1.40) 1.38 (0.96 to 1.96) 1.43 (1.00 to 2.06)

Unlikely to make attempt 0.58 (0.44 to 0.75) 0.58 (0.44 to 0.75) 0.66 (0.50 to 0.87) 0.74 (0.55 to 0.99)

50

Table A2.3: Logistic regression modelling of associations between sociodemographic and smoking related characteristics and pack inserts prompting

Inserts make you think about quitting Inserts might help you quit

bivariable multivariable bivariable multivariable

Gender Male 1 1 1 1

Female 0.96 (0.80 to 1.16) 0.98 (0.81 to 1.19) 0.93 (0.77 to 1.12) 0.95 (0.79 to 1.15)

Age

16-19 1 1 1 1

20-24 1.15 (0.88 to 1.52) 1.18 (0.89 to 1.56) 1.15 (0.87 to 1.52) 1.19 (0.89 to 1.58)

25-34 0.92 (0.73 to 1.16) 0.99 (0.78 to 1.26) 1.08 (0.86 to 1.36) 1.18 (0.92 to 1.50)

Education

GCSEs (or equivalent) or none 1 1 1 1

More than GCSEs (or equivalent) 1.28 (1.03 to 1.60) 1.22 (0.97 to 1.54) 1.30 (1.04 to 1.62) 1.21 (0.97 to 1.52)

Ethnicity

White British 1 1

White but not British 0.58 (0.42 to 0.82) 0.58 (0.41 to 0.81)

Black (inc mixed black & white) 0.66 (0.42 to 1.04) 0.61 (0.38 to 0.98)

Asian (inc mixed Asian & white) 1.16 (0.86 to 1.58) 1.19 (0.87 to 1.63)

other or not declared 1.08 (0.66 to 1.79) 1.06 (0.64 to 1.78)

Dependence (tertiles of HSI)

lower dependence

mid dependence

higher dependence

Tobacco products smoked

Only factory-made 1 1 1 1 Factory-made and roll-your-own 1.29 (1.05 to 1.58) 1.31 (1.06 to 1.62) 1.27 (1.04 to 1.56) 1.31 (1.06 to 1.61)

Factory-made cigarettes and other 1.27 (0.96 to 1.67) 1.22 (0.92 to 1.63) 1.37 (1.04 to 1.80) 1.34 (1.01 to 1.78)

Quit attempt lasting at least 24 hours No 1 1 1 1

Yes, more than six months ago 1.24 (0.97 to 1.60) 1.20 (0.93 to 1.56) 1.12 (0.87 to 1.44) 1.05 (0.81 to 1.36)

Yes within the last six months 1.74 (1.37 to 2.21) 1.46 (1.14 to 1.88) 1.66 (1.31 to 2.10) 1.35 (1.05 to 1.73)

Efficacy of quit attempt in next six months

Likely to quit 1 1 1 1

Likely to make unsuccessful attempt 0.94 (0.67 to 1.31) 0.97 (0.69 to 1.37) 0.89 (0.63 to 1.25) 0.92 (0.65 to 1.29) Unlikely to make attempt 0.54 (0.41 to 0.70) 0.59 (0.45 to 0.78) 0.46 (0.35 to 0.60) 0.51 (0.38 to 0.67)

51

Table A2.4: Logistic regression modelling of associations between sociodemographic and smoking related characteristics and pack inserts efficacy

Inserts a good way of providing quit information Inserts effective way of encouraging quitting bivariable multivariable bivariable multivariable

Gender

Male 1 1 1 1 Female 0.92 (0.76 to 1.12) 1.13 (0.93 to 1.37) 0.87 (0.72 to 1.05) 0.88 (0.73 to 1.07)

Age

16-19 1 1 1 1 20-24 1.43 (1.08 to 1.90) 0.87 (0.65 to 1.16) 0.95 (0.72 to 1.25) 0.97 (0.73 to 1.28)

25-34 1.11 (0.88 to 1.40) 0.76 (0.60 to 0.98) 0.83 (0.66 to 1.05) 0.88 (0.69 to 1.12)

Education GCSEs (or equivalent) or none 1 1 1 1

More than GCSEs (or equivalent) 1.13 (0.90 to 1.41) 1.12 (0.89 to 1.40) 1.26 (1.01 to 1.57) 1.19 (0.95 to 1.50)

Ethnicity White British

White but not British

Black (inc mixed black & white)

Asian (inc mixed Asian & white)

other or not declared

Dependence (tertiles of HSI)

Lower dependence

Mid dependence

Higher dependence

Tobacco products smoked

Only factory-made 1 1

Factory-made and roll-your-own 1.27 (1.04 to 1.56) 1.27 (1.03 to 1.56) Factory-made cigarettes and other 1.25 (0.95 to 1.65) 1.20 (0.91 to 1.60)

Quit attempt lasting at least 24 hours

No 1 1 1 1 Yes, more than six months ago 1.41 (1.09 to 1.82) 1.16 (0.90 to 1.50) 1.11 (0.86 to 1.43) 1.07 (0.82 to 1.38)

Yes within the last six months 1.77 (1.39 to 2.25) 1.54 (1.20 to 1.98) 1.66 (1.31 to 2.10) 1.33 (1.04 to 1.71)

Efficacy of quit attempt in next 6 months Likely to quit 1 1 1 1

Likely to make unsuccessful attempt 1.27 (0.89 to 1.80) 1.46 (1.02 to 2.08) 1.04 (0.74 to 1.47) 1.10 (0.78 to 1.55)

Unlikely to make attempt 0.70 (0.53 to 0.92) 0.76 (0.57 to 1.01) 0.49 (0.38 to 0.65) 0.55 (0.41 to 0.73)

52

Table A2.5: Logistic regression modelling of associations between sociodemographic and smoking related characteristics and pack inserts opinion

All packs should have inserts bivariable multivariable

Gender

Male 1 1 Female 1.21 (1.00 to 1.46) 1.20 (0.99 to 1.46)

Age

16-19 1 1 20-24 0.92 (0.70 to 1.22) 0.96 (0.72 to 1.29)

25-34 0.76 (0.60 to 0.97) 0.84 (0.65 to 1.07)

Education More than GCSEs (or equivalent) 1.15 (0.92 to 1.44) 1.10 (0.87 to 1.40)

GCSEs (or equivalent) or none 1 1

Ethnicity White British 1 1

White but not British 0.62 (0.45 to 0.86) 0.62 (0.44 to 0.87)

Black (inc mixed black & white) 1.00 (0.63 to 1.59) 0.99 (0.62 to 1.59) Asian (inc mixed Asian & white) 0.68 (0.51 to 0.93) 0.70 (0.51 to 0.96) other or not declared 1.07 (0.64 to 1.77) 1.08 (0.64 to 1.81)

Dependence (tertiles of HSI) Lower dependence 1 1

Mid dependence 0.93 (0.75 to 1.17) 1.02 (0.80 to 1.29)

Higher dependence 0.77 (0.61 to 0.98) 0.86 (0.66 to 1.12)

Tobacco products smoked

Only factory-made

Factory-made and roll-your-own Factory-made cigarettes and other

Quit attempt lasting at least 24 hours

No 1 1 Yes, more than six months ago 0.83 (0.65 to 1.07) 0.78 (0.60 to 1.01)

Yes within the last six months 1.40 (1.10 to 1.77) 1.06 (0.82 to 1.37)

Efficacy of quit attempt in next six months Likely to quit 1 1

Likely to make unsuccessful attempt 1.31 (0.92 to 1.86) 1.43 (1.00 to 2.04)

Unlikely to make attempt 0.51 (0.39 to 0.67) 0.56 (0.42 to 0.74)

53

Section 3 Distributions of variables used in the analysis Table A3.1 Distribution of cigarette type opinions variables (n=1766) Doesn't vs Makes me think about the Unattractive vs Stylish vs not Not vs appealing to Not vs something Not vs very Not vs very likely Scale Looks harmful dangers of smoking vs attractive people of my age I'd be happy to be likely to try to try (non response to health Doesn't seen with (self) smoker) Standard N % N % N % N % N % N % N % N % 1 210 11.9 91 5.2 144 8.2 134 7.6 89 5.0 88 5.0 49 2.8 69 3.9 2 157 8.9 90 5.1 110 6.2 157 8.9 90 5.1 92 5.2 29 1.6 65 3.7 3 268 15.2 188 10.6 191 10.8 293 16.6 135 7.6 170 9.6 80 4.5 128 7.2 4 446 25.3 383 21.7 493 27.9 522 29.6 527 29.8 507 28.7 205 11.6 266 15.1 5 324 18.3 363 20.6 422 23.9 296 16.8 409 23.2 416 23.6 282 16.0 296 16.8 6 165 9.3 272 15.4 204 11.6 170 9.6 244 13.8 236 13.4 332 18.8 299 16.9 7 196 11.1 379 21.5 202 11.4 194 11.0 272 15.4 257 14.6 789 44.7 643 36.4 Warning 1 53 3.0 433 24.5 539 30.5 60 3.4 441 25.0 508 28.8 715 40.5 800 45.3 2 58 3.3 303 17.2 284 16.1 84 4.8 226 12.8 250 14.2 271 15.3 302 17.1 3 136 7.7 290 16.4 267 15.1 149 8.4 243 13.8 217 12.3 188 10.6 162 9.2 4 299 16.9 249 14.1 345 19.5 308 17.4 414 23.4 386 21.9 213 12.1 211 11.9 5 360 20.4 210 11.9 180 10.2 286 16.2 229 13.0 213 12.1 146 8.3 134 7.6 6 309 17.5 142 8.0 84 4.8 249 14.1 116 6.6 109 6.2 88 5.0 72 4.1 7 551 31.2 139 7.9 67 3.8 630 35.7 97 5.5 83 4.7 145 8.2 85 4.8 Green 1 74 4.2 426 24.1 800 45.3 60 3.4 585 33.1 653 37.0 316 17.9 494 28.0 2 67 3.8 258 14.6 233 13.2 92 5.2 230 13.0 229 13.0 218 12.3 343 19.4 3 110 6.2 260 14.7 181 10.2 132 7.5 198 11.2 181 10.2 268 15.2 281 15.9 4 276 15.6 303 17.2 253 14.3 257 14.6 341 19.3 349 19.8 333 18.9 276 15.6 5 301 17.0 225 12.7 156 8.8 237 13.4 208 11.8 177 10.0 237 13.4 185 10.5 6 294 16.6 124 7.0 84 4.8 237 13.4 114 6.5 98 5.5 198 11.2 97 5.5 7 644 36.5 170 9.6 59 3.3 751 42.5 90 5.1 79 4.5 196 11.1 90 5.1

54

Table A3.2: Distribution of cigarette type opinions variables from cigarette* person data file (n=5298)

Makes me think about the Unattractive vs Stylish vs not Not vs appealing to Not vs Not vs very likely Not vs very likely Doesn't vs dangers of smoking vs attractive people of my age something I'd to try (self) to try Looks harmful Scale Doesn't be happy to be (never smoker) to health response seen with N % N % N % N % N % N % N % N % 1 337 6.4 950 17.9 1483 28.0 254 4.8 1115 21.0 1249 23.6 1080 20.4 1363 25.7 2 282 5.3 651 12.3 627 11.8 333 6.3 546 10.3 571 10.8 518 9.8 710 13.4 3 514 9.7 738 13.9 639 12.1 574 10.8 576 10.9 568 10.7 536 10.1 571 10.8 4 1021 19.3 935 17.6 1091 20.6 1087 20.5 1282 24.2 1242 23.4 751 14.2 753 14.2 5 985 18.6 798 15.1 758 14.3 819 15.5 846 16.0 806 15.2 665 12.6 615 11.6 6 768 14.5 538 10.2 372 7.0 656 12.4 474 8.9 443 8.4 618 11.7 468 8.8 7 1391 26.3 688 13.0 328 6.2 1575 29.7 459 8.7 419 7.9 1130 21.3 818 15.4

55

Histograms of cigarette opinions factor High values indicates more desirable and low values indicate more dissuasive

Figure A1 Histogram of factor Figure A2 Histograms of factor by cigarette type

56

Table A3.3: Distribution of other variables from the cigarette*participant file (n=5298)

N % Factor tertiles Less dissuaded tertiles 3533 66.7 Most dissuaded tertile 1765 33.3 Total 5298 100.0

Interaction between cigarette type and socioeconomic status standard not low 1114 21.0 warn not low 1114 21.0 green not low 1114 21.0 standard low 652 12.3 warn low 652 12.3 green low 652 12.3 Total 5298 100.0

57

Table A3.4: Distribution of pack insert opinion variables (n=1766)

Would read if Good way to Would make you Might help you quit Effective way of Support for pack Would read interested in provide information think about quitting helping smokers to inserts insert quitting about quitting quit Scale response N % N % N % N % N % N % N %

Yes 881 49.9 1068 60.5 1075 60.9 No 648 36.7 439 24.9 444 25.1 unsure 237 13.4 259 14.7 247 14.0

Strongly disagree 114 6.5 101 5.7 73 4.1 Disagree 215 12.2 236 13.4 221 12.5 Neither agree nor disagree 441 25.0 412 23.3 448 25.4 Agree 683 38.7 697 39.5 701 39.7 Strongly agree 239 13.5 234 13.3 237 13.4 Don't know 74 4.2 86 4.9 86 4.9

1 ALL cig packs should have 597 33.8 inserts like this in them 2 371 21.0 3 436 24.7 4 230 13.0 5 NO cig packs should have 132 7.5 inserts like this in them

58