The impact of the point of sale tobacco display ban on young people in Scotland: before and after study

E-cigarettes

E-cigarettes, also known as nicotine vapour products, are battery-operated devices that produce a nicotine vapour that is inhaled. The vapour contains far fewer toxins than cigarette smoke and those that are present are far less concentrated. Evidence about the impact of the long-term use of e-cigarettes on health is not yet available; in the UK, vaping is generally regarded as much less harmful than smoking tobacco25 in the short term but could be potentially damaging to health after long-term use. 26 In the UK, e-cigarettes are now more popular as a quit aid than as nicotine replacement therapy and it is estimated that there are about 2.9 million adult users, the vast majority of whom are smokers or ex-smokers. 25 Evidence is still limited and is subject to varying interpretations, but some argue that e-cigarettes show promise both as a quit aid and as a harm-reduction aid27 and, with their considerable reach into the smoking population, should be made widely available and be heavily promoted to reduce morbidity and mortality among smokers who cannot quit smoking. 28

Although e-cigarettes may benefit smokers, they are unlikely to be beneficial for young people. In Scotland, e-cigarette use among young people is mainly experimental or a ‘once- or twice-off’ event29 and is most common among young people who already smoke. Nevertheless, the number of never-smokers experimenting with e-cigarettes is increasing and, by 2015, 8% of 13-year-old and 14% of 15-year-old never-smokers in Scotland had tried an e-cigarette. 29 Direct comparisons with data collected from elsewhere in the UK are difficult to make because of the different age groups surveyed; UK-level surveys conducted in 2016 found that only 4–5% of 11- to 16-year-old never-smokers had tried e-cigarettes, whereas in Wales in 2015, 10% of 11- to 16-year-old never-smokers had. 30 However, vaping is a relatively new phenomenon and it is not possible to predict with confidence how patterns of use will develop. The market is also rapidly evolving, with new products, such as JUUL (Juul Labs, Inc., San Francisco, CA, USA), entering the marketplace, and there is some concern that vaping could become a lifestyle choice for young never-smokers. 31

There is also concern that e-cigarettes could act as a catalyst for cigarette-smoking in young never-smokers. 32 The majority of published data on this are cross-sectional, which does not allow causation to be inferred. However, a growing number of longitudinal studies have found a positive association between e-cigarette use and smoking initiation in young never-smokers at 6- or 12-month follow-up. 33–40 Common liability, where there are common risk factors for both smoking and e-cigarette use, is a plausible explanation of these findings. 41 However, some of the US studies did control for smoking susceptibility. 33,34 Most of the longitudinal studies were conducted in the USA, where policies on tobacco and e-cigarette marketing restrictions are much less stringent than in the UK. However, three studies from the UK have now reported similar findings. 42–44

In 2016, a revised EU Tobacco Products Directive45 came into force that included measures to regulate the sale, promotion and marketing of e-cigarettes. Cross-national marketing of e-cigarettes and refill containers on television and radio, in newspapers and in most magazines is prohibited. These controls came into force in the UK in May 2016, with non-compliant products and refill containers allowed to be sold until May 2017. 46 The new e-cigarette regulations are much less stringent than the regulations for tobacco, and individual jurisdictions may decide whether or not to introduce further restrictions on the domestic marketing of e-cigarettes, for example at POS or on billboards. In Scotland, the government is considering whether or not further restrictions are required to ‘reduce the visibility and attraction of e-cigarettes to children and young people under 18 and adult non-smokers’ (© Crown copyright. Contains public sector information licensed under the Open Government Licence v3.0). 47 Unlike in England, external advertising (e.g. billboards and bus shelters) may not be permitted in Scotland in the future.

Standardised packaging

In the face of ever-stricter controls over tobacco advertising, cigarette packs have become a powerful marketing tool. Pack shape and colour increase attractiveness and aid brand differentiation. Recent innovation in pack design, including novel closing mechanisms, textures or coatings to create tactile effects and olfactory and auditory cues, can also reduce risk perceptions and increase brand appeal, purchase interest, sales and market share. 48,49 In December 2012 Australia was the first country to introduce standardised cigarette packs (standard size and colour, no branding and only the brand name in a standard small font) with large graphic health warnings. 50 Early research suggests that, since these packs were introduced, there has been a shift in the attitudes and beliefs of adult smokers, who reported disliking their packs more and that packs had less appeal. The enhanced graphic health warnings also increased motivation to quit. 51 This was accompanied by a decline in expenditure on cigarette products between 2012 and 201552 and a decline in adult smoking prevalence from 16.1% to 14.7% in a similar period. 53 In 12- to 17-year-olds, smoking prevalence also fell, from 7% in both 2008 and 2011 to 5% in 2014 following the introduction of standardised tobacco packs. 54 Survey data suggest that young people find cigarette packs less appealing,55 but, to date, there has been no evidence of increased cognitive processing of the larger graphic health warnings. 56

Following an extensive public consultation, underpinned by a major evidence review,57 standardised cigarette packs were introduced in Scotland and the rest of the UK in May 2016. The new regulations58,59 are compliant with, and go beyond, the requirements of the revised EU Tobacco Products Directive. 45 As in Australia, the pack shape, colour, opening mechanism and font are regulated, together with the size and the position of health warnings and the number of cigarettes in a pack. Misleading descriptors such as ‘lite’ are also prohibited. After May 2016 UK retailers were able to continue to sell branded products to use up old stock, but from May 2017 only the sale of cigarettes and tobacco in standardised packs was allowed.

Outcome measures

A logic model was developed to provide a framework for the evaluation (Figure 1). It proposed pathways that linked together the implementation of the POS legislation with a set of short-term, intermediate and longer-term outcomes, which were to be assessed by the four study components. Rather than defining primary and secondary outcomes, we set out a time frame within which we believed the outcomes would occur. We classified outcomes as short term if they were likely to occur within 3 months of implementation of the legislation, as intermediate if they were likely to occur up to 1 year post implementation and as longer term if they were likely to occur more than 1 year post implementation. Short-term outcomes of interest were prevalence of POS advertising in tobacco retail outlets (assessed by components 1 and 2) and exposure to POS advertising (assessed by components 1, 2 and 3). Intermediate outcomes of most interest were awareness of POS advertising, tobacco brand awareness, perceived accessibility of cigarettes, perceived youth smoking prevalence, pro-smoking norms and pro-tobacco attitudes. Long-term outcomes of most interest were smoking susceptibility, smoking initiation and prevalence of smoking.

FIGURE 1.

Logic model of activities, outputs and outcomes associated with POS legislation.

Data obtained from the retailer panel interviews (component 2) and focus group discussions with young people (component 4) provided contextual qualitative data facilitating the interpretation of findings about outcomes. In addition to the outcomes outlined above, the study design enabled us to identify, to a limited extent, any unintended or adverse consequences associated with the legislation. We hypothesised that if the ban on tobacco displays reduced perceived accessibility, then smokers (including young smokers) might turn to other sources of tobacco, including the black market. Retailers or the tobacco industry might also increase tobacco promotions before the new regulations were implemented or might develop strategies to circumvent them. Indeed, this was observed in Ireland, where, following implementation of POS legislation, images of tobacco-related paraphernalia such as cigarette lighters regularly appear on the blank covers of the cigarette gantries. 20

Mapping and spatial analysis of the location and density of tobacco retail outlets

For each year of the study, geographical data on the tobacco retailers were collected, geocoded and integrated into a Geographical Information System. Data (including address and full postcode) for all tobacco retailers in the study communities were extracted from the Register of Tobacco and Nicotine Vapour Product Retailers65 and mapped at baseline (January 2013) and then annually to 2017. These data were then verified through field visits during which every street in the four communities was inspected. The community-level data enabled the monitoring of the number and rate (per population) of tobacco outlets in each wave. In addition, it provided a verified list of outlets to be visited and observed as part of the observational audit of all retailers in each community (see Tobacco advertising and marketing audits of tobacco retail outlets). This in turn ensured the accuracy of the number and rate per head of population of outlets selling tobacco in each community during the study period.

Baseline tobacco outlet data were geocoded using Code-Point® [Ordnance Survey Limited, Southampton, UK; www.ordnancesurvey.co.uk/business-government/products/code-point (accessed 16 December 2019)] to provide geographical co-ordinates and then integrated into the Geographical Information System. These data were combined with data from the marketing audits (see Tobacco advertising and marketing audits of tobacco retail outlets) and analysed to provide an assessment of changes in tobacco retailing and advertising over the study period. In the analysis, the focus was restricted to large supermarkets and small retailers most likely to sell cigarettes to young people, including off-licences; confectioners, tobacconists and newsagents (CTN); small grocers (including licensed); petrol stations; and fish and chip shops. In the final stage, the outlet data were integrated with the audit data to provide information on exposure to tobacco retailing (combining availability with visibility of tobacco products).

National-level data were also extracted from the Register of Tobacco and Nicotine Vapour Product Retailers65 and used to assess changes in the availability of tobacco products across Scotland. This information provided broader insights into how the availability changed during the study period in response to new tobacco control policies, sector-wide interventions and/or the economic climate. The national availability of tobacco products was compared across local authorities and across neighbourhoods, stratified by a measure of area-level deprivation and by urban/rural status.

Full details of all methods are provided in Chapter 5.

Tobacco advertising and marketing audits of tobacco retail outlets

Some material in this section has been adapted from Stead et al. 10 with permission from The Royal Society for Public Health. © 2016 The Authors. Published by Elsevier Ltd on behalf of The Royal Society for Public Health. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

As with the mapping and spatial analysis component, the tobacco marketing audit focused on large supermarkets and small shops, including off-licences, CTN, small grocers (including licensed), petrol stations, and fish and chip shops.

There were two parts to this study component.

Cross-sectional school survey of school children with embedded cohorts

Focus group interviews with purposive samples of pupils

Sixteen focus groups were conducted every year with S2 and S4 pupils in each study community. All were single-sex groups of between 3 and 9 participants, and they lasted between 30 and 50 minutes. Focus group participants were recruited with the help of teachers in the study schools so that they included young people who were smokers or had regular contact with smoking, such as having friends who smoke or living in a home with smoker(s). The aim was to include young people who were most at risk of becoming adult smokers. These recruitment methods have been used successfully in a recent study by one of the research team (AA) on young people’s sources of cigarettes. 68 The focus groups were repeated annually until 2017 and provided more detailed and nuanced contextual information and insights into young people’s experiences and perceptions.

In order that the pupils’ discussions did not influence their questionnaire responses, the focus groups were conducted 1–2 weeks after the school survey and were audio-recorded with the permission of group participants. The topic guide included general discussion about the community; leisure time activities; local smoking behaviours and cultures; access to tobacco products including direct, indirect/proxy and black-market; awareness of and views on tobacco promotion, including POS, other direct marketing methods, packaging, branding; awareness and perceptions of the impact of the legislation; and views about preventing youth smoking.

Table 5 shows the number of focus groups achieved and the number of participants involved.

School survey data analysis

As previously described, the school survey sample was mixed longitudinal and cross-sectional in design (see Cross-sectional school survey of school children with embedded cohorts, Survey design). That is, some young people took part in the survey only once, whereas others were followed up annually for 3 or 4 years consecutively. The DISPLAY study examined a range of outcomes. The analysis approach by outcome type is described below.

Missing data

In common with most longitudinal samples, subject attrition differentially affected male young people, young people from lower socioeconomic groups and young people with more smokers in their social circle. The data were tested for missing completely at random by conducting multiple logistic regressions on the ‘missingness’ of each variable using all of the other variables in the model. The proportion of significant results suggested that the hypothesis of missing completely at random should be rejected. ‘Missingness’ in the recording of smoking status was associated with the observed variables ‘age’ and ‘low family affluence’ and therefore likely to be missing at random. As there was information in the observed data that predicted ‘missingness’, the data were assumed to be missing at random. 69

Multiple imputation by chained equations was used where there was concern about numbers of missing data (e.g. for e-cigarette longitudinal analyses and the smoking prevalence models) so that the effect of missing data on the parameter estimates could be assessed. All dependent and independent variables in the estimation models were included in the imputation model with the following auxiliary variables: mother in employment, father in employment, entitlement to free school meals and whether or not lives with both parents. One hundred imputed data sets were produced with a burn-in of 10. Trace plots were inspected to confirm the stationarity of each chain by the end of the specified burn-in period. This results in valid statistical inferences that properly reflect the uncertainty due to missing data. 70 Few differences were found between the imputed and the complete-case analysis results.

Analyses were conducted in IBM SPSS Statistics versions 21 and 23 (IBM Corporation, Armonk, NY, USA) or in Stata^® version 15 (StataCorp LP, College Station, TX, USA).

Mapping and spatial analysis

Ethics approval for the mapping study was confirmed by the School of GeoSciences Ethics Committee, University of Edinburgh. No ethical issues were identified.

Marketing audit

The protocol for the marketing audit and retailer panel interviews was approved by the University of Stirling School of Management Research Ethics Committee. In relation to the discreet audit, we argued that conducting observational research without consent was in the public interest (the legislation was passed to protect the health of young people) and that disclosing the observational research would have risked jeopardising the whole study. The ethics application made it clear that all observations would be restricted to spaces readily accessible to members of the general public, that no retailer or premises could be identified, that no retailer or customer behaviour would be recorded and that researchers would use memory aids to enable them to record information after leaving the premises (i.e. no photographic records would be made). The approach taken in this study is consistent with accepted norms within the retail sector, where it is standard practice to collect information in store from rival chains. Similar precedents have also been set with regard to research studies involving test purchasing of age-restricted goods, where pre-notification would have confounding effects. As previously indicated, the study was not designed to assess brand-level or pricing effects and photographic data were not collected.

In line with ethics requirements, all sites and participants were assigned non-identifiable codes to maintain their anonymity. Identifiable data (e.g. participant names and premises names and addresses) were held in a separate database and were linked to electronic data files using these non-identifiable codes. Technical reports, presentations and publications ensured that no participants or retail premises could be identified. Appropriate measures were in place to ensure the safety of fieldworkers. These procedures have been developed and tested as part of other studies undertaken by the research team.

School survey

The survey research protocol was approved by the University of St Andrews School of Medicine Ethics Committee. In addition, research approval was obtained from all local authorities (for both first- and second-choice schools). The research team also made contact with all of relevant health boards to obtain ‘unofficial’ support and backing for the study. For each validation and pilot study, ethics approval was gained from the St Andrews School of Medicine Ethics Committee.

In all survey waves, parental opt-out consent was gained before survey completion. This is the standard consent procedure used in the majority of national school surveys and is believed to result in less biased samples in terms of pupil demographic characteristics and health behaviours. 71 Parents were given 2 weeks to decide whether or not to provide consent. Pupils were also given the opportunity to withdraw from the survey on the day of the survey.

Data were collected using the anonymous self-completed pupil questionnaire, which was administered by class teachers under exam conditions. Each pupil was given an envelope in which to place and seal the completed questionnaire. Therefore, teachers did not have access to the data provided by the pupils. In Scotland, all school pupils have a unique Scottish Candidate Number (SCN). This was linked to the questionnaire serial number. The SCN was used to link questionnaires completed in different years using a deterministic linkage process. The SCN and questionnaire serial numbers were stored on a secure server, separate from the questionnaire data. Data sets were shared with the research team without identifiers (date of birth and postcode); date of birth was converted to age, and postcode was converted to the SIMD quintile.

Focus group study

Ethics approval for the focus group interviews was obtained from the National Centre for Social Research (NatCen) Research Ethics Committee. This committee reviews research projects conducted by NatCen and ScotCen that are not subject to the committee’s approval, has a membership comprising senior NatCen staff, external research experts and external professional experts and is consistent with the requirements of the Economic and Social Research Council and Government Social Research Professional Guidance. Opt-out consent was sought from participants before they took part in the focus groups using the same approach as with the school surveys. At the start of each focus group, participants were assured that their anonymity would be maintained, and ground-rules were established about disclosure.

Methods

In examining POS tobacco display characteristics we focused on seven key features: internal and external visibility of storage units and products (i.e., whether they could be seen both inside the shop and from the street outside); any internal and external tobacco advertising (e.g., on the window or door); the style of storage unit; any brand advertising on the storage unit itself; the visibility of on-pack health warnings when packs were stocked in the unit; storage unit size; and proximity of tobacco products to products with potential appeal to children. For most items, observers simply noted their presence or absence, while overall visibility inside the outlet was rated on a scale of 1–5 and visibility from outside the shop was rated as ‘not very visible’, ‘fairly visible’ or ‘very visible’. Retail outlets were linked to the SIMD63 by their postcode and grouped by SIMD quintile for the purpose of analysis by area deprivation. Data analysis was conducted in IBM SPSS Statistics version 21.

The methods for the school survey are described in Chapter 2, Cross-sectional school survey of school children with embedded cohorts. The findings in this paper focus on young people’s recall of tobacco POS displays and their frequency of visiting small shops and supermarkets. The data presented are from February 2013, wave 1 (see Table 2), before the legislation was implemented in large supermarkets.

Results

Compliance in large supermarkets

Nine large supermarkets (i.e. > 280 m²) were in the sample at wave 1a; these represented five supermarket chains. An additional supermarket opened at a later wave but was not in operation at wave 1a. All nine large supermarkets audited were 100% compliant for all measures relating to displaying tobacco products and price information. Three out of the nine stores also had petrol station shops that were classed as small shops (i.e. ≤ 280 m²) and these, therefore, were not assessed for compliance at wave 1a.

Compliance in small shops

Findings on small retailers’ compliance with the legislation have been published in full in an article in PLOS ONE by Eadie et al. 72 © 2016 Eadie et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Small retailers’ initial expectations about implementation

Retailers’ initial attitudes towards and expectations of the implementation of the legislation were predominantly negative, and focused on three main themes: practical concerns, perceived government interference and perceived effects of the legislation.

Retailers raised concerns at wave 1 (prior to the implementation of the POS display ban in April 2013) about the logistical implications of covering their tobacco storage units (typically referred to as gantries), perceiving that this would involve ‘hassle’ and potential expense. All but one of the retailers had a contract with a tobacco company for the provision of a gantry, and views differed about the amount of help that retailers could expect from a tobacco company: some assumed that company representatives would arrange the gantry adaptation as part of the contract, whereas others worried that they would bear the responsibility themselves. One retailer whose storage unit was not provided by a tobacco company was particularly uncertain about what would happen in terms of covering it up.

Several retailers were concerned that customer transactions would take longer as staff struggled to locate the right product in the gantry, and that this would lead to customer frustration and potential loss of custom, with customers choosing to go elsewhere if the process was too slow:

Yes, definitely . . . there’s no doubt about it. Where at the moment we know exactly where everything is, fair enough if the display behind the curtain or what not, it’s going to be the same, but it’s still going to be the hassle of opening the curtain or shutters and adding a few more seconds, if not longer to serving . . . onto the time it takes to serve a customer, and you know yourself nowadays people are in more of a hurry.

Retailer S, Wave 1 (W1)

Protracted transaction times were felt to be particularly likely for price-conscious customers, who might ask to see all of the lower-price brands on the gantry, requiring the retailer to open each flap in turn or to consult a price list. Concerns were also expressed in some shops that the covered gantries would require retailers to turn their backs on customers for longer to retrieve products, and that this could lead to increased thefts:

We can’t turn our body around for like 10 seconds just to lift a shutter and find the fags, ‘cause there’s folk trying to steal and that . . . we have to have our eagle eyes in the shop all the time.

Retailer C, W1

Many of the retailers stocked price-marked tobacco products (packs with the price printed on the wrapper or box, commonly used for value and mid-range brands); these were perceived as attractive to price-sensitive customers, and the transparent pricing protected the retailer from accusations of overcharging or non-competitive prices. Retailers noted that, once the gantries were covered, customers would no longer be able to scan the shelves to see pack prices at a glance. They realised that the onus would be on counter staff to familiarise themselves with the prices and pass this information on to customers verbally, and some expressed concerns about how this would work in practice and whether or not customers would complain that they were being ‘ripped off’ once they could no longer see price-marked packs. At wave 1, some retailers were unaware that they would be permitted to display a price list, assuming that price lists would also be banned under the legislation. As a result of customers not being able to scan the gantry to see the full range of products available, some retailers anticipated an impact on the relative availability and sales of different brands. It was suggested that there would be little point in stocking more obscure brands because sales would dry up if customers could not see that the product was in stock, leading to a concentration on best-selling brands.

Some retailers regarded the display ban as yet another example of government interference. The legislation was framed as a ‘big brother’ (retailer Q, W1) response, and several contended that the government was hypocritical, simultaneously benefiting from tobacco profits and wanting to be seen to be implementing populist health measures. Arguments were not always coherent; for example, some framed the legislation as a way of generating revenue. Some resented what they perceived to be an implication that retailers themselves had some responsibility for encouraging people to smoke, with several observing that they were simply selling tobacco, not promoting it, and that if the product was so harmful then it should not be allowed to be sold. Linked to these views was something of a siege mentality evident among some retailers, with the POS legislation seen as part of a wider set of forces making life difficult for small retailers, tying them up in ever-increasing regulations, undermining their already perceived precarious existence:

I mean your margins are getting closed all the time and then you’re losing money to, you know silly government schemes and, it just doesn’t make any sense, you know.

Retailer F, W1

Several retailers disputed that the legislation would have its intended effects and dismissed it as ‘pointless’ on this basis. A common view was that existing smokers would not be deterred because they would carry on buying tobacco, and that young people were primarily influenced by peers and family, and therefore reducing exposure in shops would have little effect. To back up their contention that the legislation would be ineffective, several referred to articles they had read in the trade press or to conversations with tobacco representatives:

I talked to my rep[resentative] and . . . the Irish model, and they reckon that cigarette sales have gone up there, it’s not really affected it.

Retailer M, W3

In addition, some claimed that illicit trade would increase following the ban and would even threaten the viability of small shops. The basis for this argument was not always clear, but again it seemed to have been partly informed by stories read in the trade press:

I mean, [see] that trade magazine . . . There is your biggest danger because they can’t monitor, they can’t monitor out the back of the van, they can’t monitor the illegal trade . . . It’s driving it to the illegal market, which is completely unscrupulous.

Retailer Q, W3

Although very much a minority view among our sample, one or two retailers did acknowledge the possibility that the ban could discourage young people from smoking by reducing their ability to see tobacco products:

I mean you want folk tae stop smoking . . . Kids, like I’ve got kids, three kids, 4, 5 and 11; I wouldnae like tae see my 11-year-old smoking, you know . . . so it’s a good thing tae shut it off so the next generation doesn’t know what’s [behind the shutter].

Retailer L, W1

It was also suggested, however, that the shutters could create a ‘forbidden fruit’ effect, with the barrier giving tobacco products extra appeal.

Other comments retailers made implicitly, and in some cases explicitly, acknowledged the role of the gantry in drawing attention to the availability of tobacco products and in shaping customers’ preferences. For example, a few were worried that, without a visual reminder that tobacco products were on sale, some customers might go elsewhere. It was acknowledged that seeing a product on display acted as a visual cue that could reinforce or trigger purchases, and that the absence of such a cue could, at least in the short term, have an impact on smokers:

I think personally that it’s going to drop sales as well then because . . . people can’t visualise what they’re buying, and a lot of people won’t remember . . . some people actually just remember by looking at a packet what it is . . . so I think it will have some sort of knock-on effect in terms of sales and stuff as well, yeah . . .

Retailer A, W1

I think it will cut sales. Maybe impulse because you don’t really see . . . I don’t know how visual it is . . . I just don’t really know, you know, it’s uncharted territory.

Retailer O, W3

The implementation process

The majority of retailers found that implementation was straightforward. At baseline (wave 1), all but one of the retailers in our sample had a contract with a major tobacco company to provide a tobacco storage unit. As part of this contract, the tobacco company supplied, maintained and periodically updated the storage unit, in return for which the retailer was required to stock and display the manufacturers’ products, usually in accordance with a detailed planogram or layout diagram. Financial payments or reward scheme points that could be converted subsequently to cash or gifts were sometimes offered as part of these contracts: ‘they [Imperial] pay us for having a gantry, they pay us £250 a year’ (retailer, grocery/convenience store, wave 1). In the months leading up to the 2015 implementation deadline in small shops, 17 of the retailers were contacted by tobacco companies and informed that the company would continue their contract and would arrange for covers to be fitted to their existing storage units or for the storage unit to be replaced or adapted without charge. Despite the anxieties and concerns expressed (see Small retailers’ initial expectations about implementation), these retailers generally described the process of storage unit conversion as straightforward and incurring little disruption and no cost. Gantry re-fits by tobacco companies began several months before May 2015, meaning that many retailers potentially were compliant well in advance of the deadline. However, the retail audit found that none of the study retailers was fully compliant at baseline.

For the remaining retailers with contracts, the tobacco manufacturer chose to terminate the existing contract and offered either to remove the storage unit or to leave it in situ, free of charge, with the retailer then bearing responsibility for making the storage unit compliant. The termination of contracts was generally perceived by retailers to be a commercial decision based on the level of tobacco sales their business was able to generate, although some were unsure of the rationale:

It’s an Imperial one, but . . . they came back and said we will either take it out or you can sign it away, or we will sign it away to you which is what we have done . . . They will take nothing to do with this. They will do for selected retailers, for the ones I presume that have very, very high volume and it’s worth their while, but for the likes of us they have said no, so it’s up to us to cover it up at our cost.

Retailer Q, W3

One retailer opted to move tobacco stock to another part of the store, out of public view, rather than go to the effort of trying to modify an existing unit; the remaining retailers whose contracts had been terminated decided to continue to store tobacco in the existing unit. In some cases, retailers had been given information by tobacco representatives about third-party fitters who offered deals for gantry adaptation through wholesalers, and a few had taken up these offers. Others felt that they had received little advice or help, and some had become anxious and frustrated in the months leading up to the implementation deadline:

But what they did is, instead of helping us, they said . . . ‘Oh no, you just keep this one an’ we don’t have to worry about it,’ an’ then, I was really disappointed . . . They were just giving us an alternative ways to, do yourself . . . in the end I said, ‘Are you guys helping or what?’ . . . in my point of view, they need to help because, end of the day, we selling their Imperial tobaccos [sic].

Retailer L, W4

In the end, four retailers were offered, and agreed to, new storage unit covers supplied by the Philip Morris International Inc. subsidiary e-cigarette brand Vivid. This was sometimes facilitated by a Philip Morris representative offering to put a Vivid representative in touch with the retailer to discuss their storage unit requirements; retailers varied in their awareness of the links between the brand and the company. The Vivid covers supplied and fitted were printed with large, colourful images of Vivid products.

Impact on the retail environment

We report findings under two headings: transactions and customers, and changes in tobacco marketing practices.

Methods

As there was no comprehensive tool for assessing tobacco visibility, we developed a dedicated tool to assess the visibility of tobacco products and tobacco storage units. Data were collected as part of observational audits of all fixed retail outlets selling tobacco in our four study communities over a 5-year period carried out annually from February 2013 to February 2017. The data were used to compute annual visibility scores for the two dimensions of visibility: tobacco products and tobacco storage units. Full details of the methods for the observational audit and retailer panel are in Chapter 2, Tobacco advertising and marketing audits of tobacco retail outlets.

Measurement of tobacco product visibility was assessed using a 5-point categorical scale on which ‘4’ represented full visibility and ‘0’ represented no visibility. Assessment of storage unit visibility involved measuring five key storage characteristics: visibility of storage units at POS, proximity of storage units to traffic flows through the store, storage unit size, storage unit conspicuousness and visibility of storage units from outside the store. Storage unit visibility scores for each outlet were computed by aggregating the individual scores for all five measures (range 0–22).

In the analysis we examined changes in visibility scores by SIMD quintile and study community. These were represented graphically using plots of means and associated standard errors. We carried out multivariable regression analyses for continuous measures and used alternative non-parametric methods for ordinal outcomes to assess whether or not observed differences remained significant after adjustment was made for other factors, such as type of store and interactions between ‘time’ and ‘study community’.

Results

The scale was assessed for inter-rater reliability. There was considerable inter-rater agreement, with all measurement items exceeding the suggested 70% threshold for acceptable inter-rater agreement using the percentage approach (range 92–100%).

Product visibility

Mean product visibility scores for all retail outlets (n = 96) fell from 3.96 [standard deviation (SD) 0.41, range 0–4] before the ban to 3.16 (SD 1.33, range 0–4) after the partial ban was implemented. It then fell further to 1.13 (SD 0.57, range 0–3) after the ban was extended to include small shops (n = 93). Table 6 shows the results of two cross-sectional Mann–Whitney U-tests that indicate that retail outlets in one of the study communities (C4) had lower product visibility than those in the other three areas and that this difference was not affected by the display ban.

TABLE 6 - Comparison of tobacco product visibility in C4 with that in the rest of the study communities

Wave	Community	n	Rank-sum	Expected	z	p-value
2013	C4	19	883	921.5	2.013	0.0441
2013	C1, C2, C3	77	3773	3734.5
2017	C4	17	646	799	1.985	0.0471
2017	Rest	76	3725	3572

Figure 2 shows the mean product visibility over time across the four study areas. There were no differences in product visibility between outlets in high-deprivation areas and those in low-deprivation areas before the display ban and no differences in product visibility scores between outlets in high-deprivation areas and those in low-deprivation areas after implementation of the comprehensive display ban.

FIGURE 2.

Change in average tobacco product visibility over time by study area.

An examination of change in product visibility scores over time by outlet type revealed that product visibility in supermarkets reduced markedly after 2013 (after the display ban in large shops came into force). However, for other retail outlet types there was a more gradual change, with grocers and petrol stations, in particular, having a wide range of product visibility scores in early 2015, before the ban was extended to include smaller outlets. This indicates that some small grocery stores and garage shops had modified their tobacco product displays before the ban was extended to include small shops.

Storage unit visibility

The mean storage unit visibility score for all outlets (n = 96) before the display ban was 15.91 (SD 1.82, range 5–19). After the ban had been implemented in supermarkets, the mean storage unit visibility score for all outlets (n = 94) dropped slightly, to 15.52 (SD 2.52, range 5–20), and then dropped further, to 14.20 (SD 3.20, range 5–19), after the ban had been implemented fully in all shop types (n = 93). The change in storage unit visibility can be attributed largely to a reduction in the conspicuousness of storage units. Storage units that are used to display tobacco products stand out more in the shop environment when they are uncovered than do storage units in which products are concealed by shutters or covers.

Figure 3 shows the change in mean storage visibility over time across the four communities. C4 (small town, medium/low deprivation) had lower mean storage unit visibility scores than the other areas before and after the ban. Interacting time and area as a predictor had no effect, indicating that the display ban did not lessen the difference between C4 and the other areas in mean storage unit visibility. This suggests that the display bans did not eliminate shop-level urban/rural differences in tobacco storage unit visibility.

FIGURE 3.

Change in average tobacco storage visibility over time by study area.

Finally, there was a marginal decline in mean storage visibility scores for most outlet types over time, with the exception of large supermarkets.

Discussion

The visibility tool was a robust and reliable means of measuring tobacco visibility and proved practical to use in a range of retail environments and shop types without causing any recognisable disruption or being detected by shopkeepers or their customers. Findings show that the Scottish POS display ban has been effective in removing products from public view, thereby reducing young people’s exposure to tobacco brands, but it has had very little influence on where tobacco products are stored or on the visibility of tobacco storage units.

The visibility tool has the potential to be used elsewhere to measure product and storage unit visibility where tobacco POS display bans are planned or have been implemented. It could also be modified to assess the visibility of tobacco displays in jurisdictions where tobacco products are sold outdoors.

Methods

Results

National-level changes in availability by deprivation and urban/rural category

The number of tobacco retailers declined following the introduction of the POS legislation in 2013. However, large spatial variations were evident. For example, during the study period, in around one-fifth of local authorities in Scotland the provision of tobacco retail outlets increased from 2015 onwards (see Appendix 3).

The national-level findings demonstrated that tobacco retailer density was highest in the most deprived quintile throughout the study period (Figure 4). The drop in density found at the national level was apparent in all of the deprivation quintiles during the study period; however, the reduction was smaller in the most deprived quintile. The rise in density towards the latter part of the study period (2015–17) was observed in all deprivation quintiles but was significantly greater in the two highest quintiles. The ratio between the most deprived and least deprived quintiles increased from 2.01 : 1 at the start of the study period to 2.17 : 1 by 2017.

FIGURE 4.

Tobacco retailer density estimates for Scottish data zones by SIMD income deprivation quintile (2012–17). (a) Density by SIMD quintile; and (b) density by SIMD quintile (scaled to individual SIMD quintile).

There were also significant differences in the change in tobacco retailer density by urban/rural category. Large urban areas had twice the retailer density of other urban areas, accessible small towns and remote small towns. In the first part of the study period (2012–15) there was a notable drop in retailer density across all urban/rural categories. Post 2015, the downwards trends in retailer density deviated, with substantial increases in both large urban areas and other urban areas. In accessible small towns, remote small towns and accessible rural areas, retailer density changed very little in this latter period; in remote rural areas, it dropped slightly.

Changes in visibility in the four DISPLAY study communities

During the study period, tobacco retail provision in the four study communities was fairly stable. The total number of retailers across the four communities was 95 during the period 2013–16. This fell slightly to 93 in 2017 (Table 7). The number of outlets in each community ranged from 17 in C4 (2014) to 40 in C1 (2013).

TABLE 7 - Number of tobacco retailers in each study community

Year	Number of retailers
	C1	C2	C3	C4	Total
2013	40	19	18	18	95
2014	39	21	18	17	95
2015	38	21	18	18	95
2016	39	20	18	18	95
2017	38	19	18	18	93

Between each year, there was a small number of store openings and closures in each of the communities, but this figure did not exceed two retailers in any communities for any year.

Figure 5 shows the change in the availability and visibility of tobacco products for the study participants in the four study communities. Unsurprisingly, given the stability in the number of tobacco retailers during the study period, before weighting by the visibility score there was little change in the exposure estimates (Figures 5a and 5c). However, and as anticipated, once the estimates were weighted by the total product visibility scores, there was a large reduction in total product visibility among young people from our study communities following the introduction of the POS legislation. Weighting the density measures by the visibility score showed that there was a significant reduction in the participants’ neighbourhood exposure after the phased implementation of the POS legislation in 2013 (large supermarkets) and 2015 (small retailers) (Figures 5b and 5d). The drop was particularly notable between 2014 and 2017. By the end of the study period, there was a markedly smaller difference in the overall visibility scores between the four communities.

FIGURE 5.

Tobacco retailer density estimates for participants’ neighbourhood and activity space, for the four display communities, by visibility weighting and FAS score tertile (2013–17). (a) Neighbourhood; (b) neighbourhood weighted by visibility scores; (c) activity space; and (d) activity space weighted by visibility scores. Adapted from Pearce et al. 83 with permission of the BMJ Publishing Group Ltd. © Author(s) (or their employer(s)) 2019. No commercial re-use. See rights and permissions. Published by BMJ.

Adapted from Pearce et al. 83 with permission of the BMJ Publishing Group Ltd. © Author(s) (or their employer(s)) 2019. No commercial re-use. See rights and permissions. Published by BMJ.

The results based on the measures of the participants’ neighbourhoods and activity spaces (see Figures 5a and 5c vs. Figures 5b and 5d) were broadly consistent. However, socioeconomic inequalities also increased in terms of the visibility of tobacco storage units during the study period (see Figures 5b and 5d). When the results were stratified by participants’ socioeconomic status (FAS score), the declines in retailer density, weighted by total product visibility, were found to be restricted to the least disadvantaged tertile of participants. This finding suggests that inequalities between the most deprived and the least deprived participants increased during the study period.

Conclusions

Previous work has shown that there are strong geographical differences in the availability and visibility of tobacco products between communities;84,87 tobacco retailing tends to be disproportionately located in more socially disadvantaged places. It is also apparent that higher neighbourhood availability of tobacco products is associated with higher levels of smoking prevalence and smoking initiation, and lower levels of smoking cessation. However, little was known about the implications of the introduction of POS legislation on the provision of tobacco retailing, or about the impact this has on inequalities.

This DISPLAY study has shown that the introduction of POS legislation in Scotland has been successful in reducing the exposure to and visibility of tobacco products in study communities. However, at a national level, although there was modest reduction in the number of tobacco retailers around the introduction of the POS legislation, after 2015 there was a small increase. There is also strong evidence of growing socioeconomic disparities in both the availability and the visibility of tobacco. Future research should focus on assessing the longitudinal relationships between tobacco outlet availability and visibility and intermediary outcomes such as attitudes to smoking, brand awareness and perceived ease of access to tobacco products. These findings are important from policy perspectives, as, despite the considerable progress being made in tobacco control in Scotland, tobacco products remain widely available in communities across Scotland.

Brand awareness prior to the introduction of point-of-sale legislation

Change in brand awareness after the implementation of point-of-sale legislation

Methods

The DISPLAY school survey formed the basis for this analysis. We selected data for all S2 and S4 young people, from all years of the school survey (i.e. 2013–17: observations, n = 7168; individuals, n = 5376), which allowed for a comparable group across all survey years. Including S5 and S6 pupils would have led to the selective inclusion of students who had better academic performance and may have been less likely to start smoking.

The outcome variables were self-reported perceived tobacco accessibility and smoking acceptability. A measure of perceived tobacco accessibility was derived from the following question: ‘If you, or someone your age, tried to buy cigarettes or tobacco in a shop, do you think you would be successful?’ Response categories were ‘yes’, ‘no’ and ‘don’t know’. ‘Don’t know’ was categorised together with ‘no’, meaning that we distinguished those who were confident that people their age would be able to buy tobacco.

Smoking acceptability was measured in two ways. The first focused on the injunctive social norm (i.e. whether or not the behaviour would be approved of, hereafter called pro-smoking norm) and the second focused on the attitude of the respondent (hereafter called pro-smoking attitude). The smoking norm was measured with the question ‘Do your friends think it is OK for people your age to smoke cigarettes or hand-rolled cigarettes (roll-ups)?’ The response options were ‘they think it’s OK’, ‘they do not think it’s OK’ and ‘don’t know’. The young person’s attitude towards smoking was measured with the question ‘Do you think it is OK for someone your age to smoke cigarettes or hand-rolled cigarettes (roll-ups) once a week?’, to which students responded ‘it’s OK’, ‘it’s not OK’ or ‘don’t know’. As for accessibility and smoking norm, ‘don’t know’ responses were merged with ‘it’s not OK’ responses.

To quantify the changes in our outcome variables with the introduction of the partial and comprehensive POS tobacco display bans, we segmented the five survey years into three time periods that matched the implementation of the POS legislation. Time was segmented into the year before the partial ban in supermarkets (2013), the years between the partial ban and the implementation of the comprehensive ban in small shops (2014–15), and years after the comprehensive ban (2016–17).

In this study, the variables for frequency of visits to large supermarkets and small shops were each computed using three categories: ‘often’ (every day, most days), ‘sometimes’ (two or two times per week, about once per week) and ‘rarely’ (less than once per week, never).

Smoking status was measured with the question ‘Have you ever smoked cigarettes or hand-rolled cigarettes (roll-ups), even if it is just one or two puffs?’; among those answering ‘yes’, those who indicated that they were current smokers were considered smokers. This does not include those who had smoked only once or twice. E-cigarette use was categorised as ‘not having tried’, ‘having tried once or twice’ or ‘occasional or regular use (currently or in the past)’. In 2013, the survey did not include questions on e-cigarettes, and all 2013 responses were therefore coded as missing.

To study the association between time segments and both the accessibility and the acceptability outcomes, we applied generalised estimating equations analyses. Nested models were fitted first to include sociodemographics and then to control for smoking-related variables. Fully adjusted models for each outcome variable revealed large confounding effects when our e-cigarette variable was included. As a result, we added e-cigarettes to separate models.

We also considered whether or not potential reductions in accessibility and acceptability (pro-smoking norms and pro-smoking attitudes) after the POS display ban showed any association with frequency of visits to tobacco outlets. Differential associations were tested by assessing interactions between time segments and visits to shops (supermarkets and small shops separately) in the fully adjusted model.

Results

Unexpectedly, the analysis of the school survey data showed that crude trends in smoking accessibility and both measures of smoking acceptability (pro-smoking attitudes and pro-smoking norms) increased over time. Table 11 presents three models for each of the three outcome variables.

Our analysis shows that, after adjusting for sociodemographics and smoking status (models 1 and 2), there was no significant change in perceived tobacco accessibility and pro-smoking attitudes following the introduction of either the partial or the comprehensive POS display ban. However, pro-smoking norms increased following the implementation of the comprehensive ban. Only after e-cigarette use was included in the model (model 3) did we find a significant decline in perceived access [odds ratio (OR) 0.72, 95% CI 0.57 to 0.90] and in pro-smoking attitudes (OR 0.67, 95% CI 0.49 to 0.91). Including e-cigarette use in the model also eliminated the significant increase in pro-smoking norms (OR 1.00, 95% CI 0.78 to 1.29).

When testing for interactions between time segment and shop visit frequency (not shown), we found that young people who visited small shops frequently showed the greatest fall in perceived accessibility. Other interactions were not significant.

The focus group discussions provided insights into how young smokers accessed tobacco. This was achieved most commonly through proxy purchasing, which often involved approaching someone outside the shop (often described as a ‘junkie’ or ‘jakey’) to act as the purchaser (‘jump-in’) for them. At times, adult purchasers were allowed to keep the change or were given cigarettes by the young smoker. Family members were also a source of tobacco, although at times they did not providing this directly and the young smokers appeared to take the tobacco without permission:

Get a junkie-looking guy man, an old junkie-looking guy . . .

For fags it’s pure easy because see my bit, I have junkie flats a’roond the corner fae me . . . You’ve literally got to walk past and ask ‘will you go in for me?’

They always want one fag but that’s the deal . . .

I tend to ask junkies they’ll jump in a shop for you or young lassies, but mainly junkies, ‘do you want to go into the shop for us and I’ll give you a fag?’

Do they no’ normally keep the change or something?

I’ll just say ‘gonna go in the shop and get fags for me, and ah’ll gie ye wan’, then I pretend tae tear off the wrapper an’ run aff . . .

School C1, group S2, male

The focus group discussions conducted after 2015 suggested that access to tobacco was becoming more difficult for young people, hindering their ability to buy cigarettes directly from shops, which was possibly linked to increased test purchasing, Challenge 25, and changes in retailer practice in relation to underage sales. However, some young people also linked this to the effect of covering up of tobacco displays on young people’s ability or inclination to ask for cigarettes:

It’s so you can’t like see them as clearly so you’re not like, ‘Oh I’m going to turn around and get some cigarettes!’

It’s like um . . . sort of not as eye drawing that you’re wanting to go and get some.

So you don’t just ask ‘can I get some?’, so you won’t see them. If you need them you’d ask for them.

School C2, group S2, female

Nevertheless, in every wave of qualitative interviews, young people reported that a few shops would sell tobacco to young people, and that on occasion the retailer waited for other customers to leave before doing so. These shops tended to be smaller and independent:

I think if you don’t look old enough they know and I think they just do it for the money and plus they see you outside the shops all the time, and they probably get fed up of standing outside the shop to stop you. So when you walk in I think they just know and they turn a blind eye and get you ‘oot as fast as they can do, you know what I mean?

But some shops do it on the sly like they actual mean it, they get you to stand up the back and they walk down and give you them like down the back. But that’s when you know it’s sly . . .

School C1, group S4, female

Well, I was helping a pal. I went in for a pal. That’s the only way I see it: I just went in to help a friend. Like she’s smoked for like a year or two now, so it’s like she went in and I think she didn’t get sold, so she came back out, and I went in and got sold. And I basically . . . I just went in. I was like, ‘Can I have 20 big Superkings yeah?’ But he just gave me it, and he was like, ‘Put it in your pocket before you leave’. I just gave him the money and put it in my pocket and took them, and I got out.

Yeah. I don’t know how shopkeepers can be like stupid, like . . .

‘Cause it wouldn’t happen in like big shops like Asdas [Asda Stores Ltd, Leeds, UK] or that. That just wouldn’t . . .

School C3, group S4, female

The focus group discussions also demonstrated that both smokers and non-smokers held predominantly negative views about smoking. The young people described smoking as a disgusting, addictive and costly habit, with adverse longer-term health effects. Those who pursued active sports spoke of smoking affecting their performance, which in some instances had led them to quit tobacco. Smokers were also thought to be stigmatised by other young people in school, and were said to be ‘dodgy’, ‘lower class’ and unpopular. There was little evidence that these views had hardened after the POS display ban came into force:

It’s not good for your lungs and it’s just . . . it’s also not good for like you as a person because then you could get addicted to it and then not be able to stop.

School C3, group S2, female

It’s pretty much why I have like nae mates, ‘cause like everyone actually shuns upon smoking, and I only have like two mates ‘cause of smoking. It’s quite bad in that sense.

School C4, group S4, male

However, some focus group participants did express positive views about smoking:

. . . it just gave you a different feel like . . . when you started . . . you would get a nicotine rush and stuff. And then when the nicotine rush would go away you’d feel kind of calm and stuff.

School C3, group S2, female

Well, it is a disgusting habit, but it’s like just the nicotine relieves stress . . .

. . . if we’re out, it’s more just because it’s like the social factor. Everyone else around you’s smoking.

School C4, group S4, female

Method

In the school survey, we asked young people ‘In the area where you live, do you know of people selling cigarettes on the black market (e.g. illegal, fake and dodgy cigarettes)?’ The response options were ‘yes’, ‘no’ and ‘don’t know’. Using data from all school survey waves, we examined the perceived availability of black-market tobacco products among S2 and S4 pupils between 2013 and 2017.

To study the change in black-market cigarette availability, we applied a generalised estimating equations analysis, fitting an unadjusted model and a model adjusted for sociodemographic (age, gender, family affluence, school attended) and smoking-related (their own smoking status and the smoking status of their wider social circle) characteristics.

Results

Table 12 presents the prevalence of black-market cigarette availability by survey year, and the ORs (and 95% CIs) from the unadjusted model (model 1) and the adjusted model (model 2). The prevalence of black-market cigarette availability did not change much between years, ranging from 16.0% in 2014 to 18.6% in 2017. The unadjusted and adjusted models gave similar findings; the odds of black-market tobacco availability were not significantly different from those at baseline (2013) in any subsequent year of the study. We therefore found no evidence of a change in the availability of black-market tobacco after the legislation was implemented either in large supermarkets in 2013 or in small shops in 2015.

The focus group discussions indicated that young people interpreted ‘black-market tobacco’ in a number of ways, including tobacco brought back from abroad, often in bulk, and sold cheaply in the local community; counterfeit or ‘fake’ tobacco sold in markets or in some shops, often containing impurities and perceived to be of low quality; and cheap, imported tobacco from abroad:

Some people get like counterfeit and it means like total fake, like it’s got . . . like I got one from someone and it had like a big stick in it, it’s total minging!

Aye they put like shit and everything in them eh?

School C4, group S2, male

. . . somebody on the street that sells like them in huge packs. Yeah. They sell it for 50p, and it’s like . . . it’s got Arabic on it, and that’s . . . it’s like it’s in a pack, and it’s like an Amber Leaf pack, but it’s in Arabic, and it’s . . . they sell it for 50p cause they can get it for . . . so cheap.

It must be pretty dodgy.

School C4, group S2, male

At waves 1 and 2, in particular, a small number of houses where tobacco and other goods were known to be sold unlawfully were mentioned:

If they are cheap and they’re JPS [John Player Special, John Player & Sons] they’re fake . . . If they’re £2.50 they’re fake.

Per 20?

No that’s for 10 . . . They’re still fake if they’re £2.50 . . .

It’s like his own hoose but it’s like his little store as well if you know what I mean? It’s like a little shop.

School C2, group S4, female

From the focus group discussions, there was no evidence that the young people were more aware of black-market tobacco after the complete ban on POS promotions came into force – indeed, more accounts of black-market tobacco were provided in the earlier focus group waves – but, owing to the different views about what constitutes black-market tobacco, it is not possible to draw firm conclusions.

Methods

In this analysis we used data from the DISPLAY school surveys conducted between 2013 and 2017. We restricted our analysis of cross-sectional means to S2 and S4 pupils in all survey years, which allowed a consistent comparison of perception over our time segments.

The outcome measure for our models was young people’s perception of smoking prevalence. At each survey wave young people were asked two questions: (1) ‘Out of 100 people your age, how many do you think smoke cigarettes or hand-rolled cigarettes at least once a week?’ and (2) ‘Out of 100 adults, how many of them do you think smoke cigarettes or hand-rolled cigarettes at least once a week?’ A small number of responses (19 and 15, respectively) were outside the range 0–100 and were excluded from the analyses.

In the manner described in Chapter 6, Perception of tobacco accessibility and smoking acceptability, Methods, in the analyses of perceptions of tobacco accessibility and smoking acceptability, time was segmented into the year before the partial POS ban (2013), the years between the partial and the comprehensive POS bans (2014–15) and the years after the comprehensive POS ban (2016–17). Frequency of visits to both large supermarkets and small shops was recoded into binary variables: visit often (every day or most days) and visit less than two or three times per week.

We used mixed-effects regression models with robust standard errors to establish whether or not reductions in adolescents’ perceptions of adult or youth smoking prevalence were associated with the implementation of either the partial or the comprehensive POS display ban. In the first instance, we included only the sociodemographic variables age, gender and FAS score, and the pupils’ school attributes. We then included in our models smoking-related factors: smoking status, e-cigarette use, family and friends’ smoking status, and susceptibility to smoking.

Further models were tested that included variables for frequency of visits to supermarkets and small shops. In these models, we tested the interaction effect between the frequencies of shop visits and the introduction of the partial and comprehensive bans (i.e. with our time segments) on the pupils’ perceptions of smoking prevalence.

Results

Table 13 shows that pupils’ estimates of peer smoking and adult smoking fell from 38.1% and 58.5%, respectively, at baseline (2013) to 36.7% and 56.1%, respectively, in the time segment 2016–17. The pupils’ estimates of smoking prevalence are considerably higher than the prevalence estimates in 2016 national data for Scotland, which show that in 2016 smoking prevalence among 13-year-olds was only 2% and among 15-year-olds was 9%,106 and that adult smoking prevalence was 18% in 2017. 107

The results of the regression analyses are shown in Table 14. We developed two models each for both perceived smoking prevalence among peers and perceived smoking prevalence among adults. The first model adjusted for sociodemographics and individual and family and friends’ smoking status. In the second model, e-cigarette use was also included as a covariate. Table 14 shows that the decline in smoking prevalence was significant in both models, but was greater when e-cigarettes were included. Thus, in the fully adjusted model (e-cigarettes included), peer smoking prevalence fell by 3.4% (95% CI –4.8% to –2.1%) following the partial POS ban (supermarkets) and by 2.9% (95% CI –4.3% to –1.5%) following the comprehensive ban (small shops), compared with the pre-ban period. However, it can be seen that the fall in perception of the prevalence of smoking among adults after the partial ban became significant only after controlling for e-cigarette use (coefficient –1.7, 95% CI –3.0 to –0.4).

The perception of smoking prevalence among peers varied significantly by frequency of visits to tobacco outlets. It was significantly lower than pre-ban only among those adolescents who visited supermarkets (coefficient –2.4, 95% CI –3.9 to –0.8) or small shops (coefficient –3.1, 95% CI –4.9 to –1.2) less often. The perception of smoking prevalence among both peers and adults showed no significant reduction among young people who visited outlets frequently.

In the focus group discussions, young people were asked about the prevalence of smoking among their peers, family members and local communities. As those who were recruited to the focus groups were smokers or those perceived to be at risk of smoking, it might be argued that they might overestimate smoking prevalence. The young people identified the school years in which smoking appeared to be most common. These tended to be the upper school years, from S4 to S6 (aged 15–18 years). Throughout the project, young people also reported communities in which ‘everybody smokes’, as well as areas in which smoking was very uncommon. On the whole, the young people did seem to overestimate smoking prevalence, particularly in the wider community, but there was little evidence that this changed over time.

Discussion

We found that young people greatly overestimated the prevalence of smoking both among their peers and among adults. Perception of peer smoking at each school varied from 28.1% to 44.4% in the 2014–15 time period. In the same period, perception of adult smoking varied from 49.5% to 65.2%. At ward level, the highest prevalence at any ward in the whole of Scotland among adults was 37.3% for the period 2012–15, which is lower than any of the perceived adult smoking average by any school. Data on smoking prevalence at a ward level are not available for young people, but, given that the national prevalence of smoking in 2016 was 2% among 13-year-olds and 9% among 15-year-olds, young people’s estimates of youth smoking prevalence far exceed local youth smoking prevalence.

The observed overestimation of smoking prevalence is consistent with the ‘normalisation’ of smoking behaviour, which has been attributed in part to the widespread availability of tobacco retailers and the promotion of tobacco through POS displays. 20 Perceived smoking prevalence among young people fell significantly both after the partial ban and after the comprehensive ban. However, perceived smoking prevalence among adults fell only after the implementation of the comprehensive ban and only after controlling for use of e-cigarettes. Our results mirror those of other POS studies that found a decline in perception of smoking prevalence. 101,102 However, our results again suggest that e-cigarette use may have disrupted the impact of the POS legislation on the de-normalisation of smoking behaviour.

Nevertheless, in our study we also found that there was no significant reduction among young people who regularly frequented small shops. In Chapter 5, we saw that, whereas the POS display ban had led to a reduction in the visibility of tobacco products, exposure to the storage units had decreased only slightly. The results reported here suggest that frequent exposure to tobacco storage units, which promote tobacco as a generic product, may contribute to the continued normalisation of smoking.

Methods

In this analysis, we used data from the DISPLAY schools surveys conducted between 2014 and 2017. The primary outcome measure in our models was the risk of the young person becoming susceptible to smoking in the year after they entered the study. Consequently, the analysis is restricted to young people who were non-susceptible never-smokers on entry to the study (n = 2409). Data from the 2013 baseline survey were not included in the analysis as none of the participants in that wave was sampled in the 2014 wave (see Chapter 2, Survey design, and Table 3).

Our outcome variable, susceptibility to smoking, was measured using the answers to two questions, ‘If one of your friends offered you a cigarette or hand-rolled cigarettes (roll-ups), would you smoke it?’ and ‘Do you think you will smoke a cigarette or hand-rolled cigarettes (roll-ups) at any time during the next year?’, to which the young person could respond ‘definitely yes’, ‘probably yes’, ‘probably not’ or ‘definitely not’. Any answer other than ‘definitely not’ for either of the questions indicated smoking susceptibility. If one of the responses was negative for susceptibility but the other response was missing, then the respondent was coded as missing for susceptibility.

The analysis was conducted using discrete time survival methods. We elected to use survival analysis as it enabled us to make use of information on cases that are censored (i.e. we included in the analysis participants who had not experienced the event before the end of the study) as long as the censoring was non-informative (i.e. unrelated to the outcome). We first examined the life tables to determine the proportion of young people who remained non-susceptible never-smokers for each year under observation. We then examined the effect of shop visit frequency on susceptibility using Kaplan–Meier survival estimates. Complementary log-log models were then used to examine the effect of year of entry to the study on susceptibility, with full adjustment made for other predictors of smoking susceptibility. Complementary log-log models were selected as this approach is deemed most suitable for discrete observations in an underlying continuous process. 109

The event for which participants were ‘at risk’ was becoming susceptible to smoking. The sample consisted of all participants who were non-susceptible never-smokers on entry to the study. When a participant reported at the next wave that they either were susceptible to smoking (as defined above) or had smoked since the previous survey wave, then they were coded as the event having occurred. If a participant reported ever smoking, it was assumed that they had passed through a stage of susceptibility before doing so.

The independent variable was the year the young person entered the study. We hypothesised that young people who entered the study before the full ban, that is, up to 2015, would have higher odds of becoming susceptible to smoking than young people who entered the study after the full ban.

We also included an interaction effect between shop visit frequency and the year of entry to study, as we hypothesised that shop visit frequency would have a greater effect on probability of susceptibility before the ban than after it.

Results

From the life tables we found that 73% of young people remained non-susceptible after 1 year in the study, 51% remained non-susceptible after 2 years and 36% remained non-susceptible after 3 years.

The results of the discrete survival analysis are presented in Table 15. Model 1 shows that the effect of year of entry to the study on the risk of becoming susceptible to smoking is highly significant. Young people who entered the study in 2016 after the introduction of the comprehensive ban were much less likely to become susceptible to smoking in the following year than young people who entered the study in 2014 (adjusted hazard ratio 0.42, 95% CI 0.32 to 0.54) when only the partial ban was in place. In model 2, we assessed whether or not there was an interaction between year of entry to the study and shop visit frequency. The interaction effect is not significant as measured by the hazard ratio; however, when examining the marginal probabilities, a significant interaction effect can be seen.

Figure 6 compares the predictive margins for frequent and infrequent small-shop visitors. It suggests significant differences in the probability of becoming susceptible among young people entering the study in 2014 (pre ban), and the effect of shop visit frequency diminishes after the implementation of the POS ban first in supermarkets and then in small shops (χ² for contrasts of predictive margins = 4.12; p = 0.047).

FIGURE 6.

Comparison of predictive margins of risk of becoming susceptible to smoking by year of entry into study and frequency of visiting small shops.

As in the previous two analyses, we assessed a third model that included the use of e-cigarettes as a covariate. However, this made only a marginal difference to smoking susceptibility and was therefore rejected.

We also assessed an interaction effect between FAS score and the year of entry to the study (not shown) to assess whether or not there were any differential impacts of socioeconomic status on probability of susceptibility. However, this interaction was also found not to be significant.

Discussion

To our knowledge, this is the first study to look at the effect of a comprehensive ban on POS tobacco marketing on the risk of young people becoming susceptible to smoking. We found that young people who entered the study in 2016 after the comprehensive ban was in place were less likely to become susceptible to smoking in the following year than participants who entered the study in 2014 when there was only a partial ban. Participants who entered the study in 2015 before the comprehensive ban were also less likely to become susceptible to smoking in the following year than those who entered in 2014. We hypothesised that, if the POS display ban causes a decrease in smoking susceptibility, the effect of shop visit frequency on smoking susceptibility should diminish after the comprehensive ban. This was indeed the case, with the difference in probability that young people would become susceptible to smoking between those who visited shops frequently and those who did not being significantly reduced in 2016 compared with 2014. This effect was significant only in the predicted marginal probabilities analysis and not in the hazard ratios. In addition, we found no evidence that the POS legislation increased inequalities in smoking susceptibility.

Methods

In this analysis we used the DISPLAY schools survey data for the years 2014–17. The analysis was restricted to young people who had never smoked on entry to the study (n = 8214). As in the susceptibility study, analysis used discrete time survival methods. We first examined the life tables to determine the proportion of young people who remained never-smokers for each year under observation. We then examined the effect of small-shop visit frequency on susceptibility using Kaplan–Meier survival estimates.

Complementary log-log models were then used to examine the effect of year of entry to the study on smoking initiation, with full adjustment made for other predictors of smoking susceptibility. As noted above, this was regarded as the most suitable approach for analysing discrete observations on an underlying continuous process. 109 The event for which participants were ‘at risk’ was smoking initiation. When a participant reported at a survey wave that they had smoked since the previous survey wave, they were coded as the event having occurred.

The independent variable in the analysis was the year of entry into the study. We hypothesised that young people who entered the study before the comprehensive ban was in place (up to 2015) would have higher odds of smoking initiation than young people who entered the study after the comprehensive ban was in place.

We also included an interaction effect between small-shop visit frequency and the year of entry to study, as we hypothesised that shop visit frequency would have a greater effect on probability of smoking initiation before the ban than after the ban.

Results

From the life tables we found that 87.8% of young people remained never-smokers after 1 year in the study, 72.4% after 2 years, and 59.3% were still never-smokers after 3 years in the study.

The results of the discrete survival analysis are presented in Table 16. Model 1 shows the effect of year of entry to the study on risk of smoking initiation without interaction effects. The year of entry to the study is highly significant. Both young people who entered the study in 2015 and those who entered in 2016 were much less likely to initiate smoking in the following year than young people who entered the study in 2014 (adjusted hazard ratio 0.72, 95% CI 0.59 to 0.88, p < 0.01; and adjusted hazard ratio 0.72, 95% CI 0.59 to 0.88, p < 0.01). Model 2 shows the effect of adding an interaction term between year of entry to study and shop visit frequency. The year of entry into the study remains highly significant but the interaction effect is not significant when measured by the hazard ratio; however, when examining the marginal probabilities, a significant interaction effect can be seen.

Figure 7 compares the predictive margins for frequent and infrequent small-shop visitors. It suggests a significant difference in probabilities of smoking initiation by frequency of shop visits among young people when they entered the study in 2014 (pre ban). This significant difference in probabilities of smoking initiation by shop visit frequency remained in 2015 and 2016 but was attenuated slightly by 2016 after the implementation of the comprehensive ban (χ² for contrasts of predictive margins = 11.99; p = 0.0025).

FIGURE 7.

Comparison of predictive margins of risk of becoming an ever-smoker by year of entry to the study and frequency of visiting small shops.

As in the analysis of smoking susceptibility, the inclusion of e-cigarette use as a covariate resulted in a marginal reduction in effect of year of entry on smoking initiation, but as the effect was very small the model was rejected. We then assessed whether or not there was an interaction between year of entry into the study and FAS score (not included), but again the interaction effect was found not to be significant.

Discussion

The introduction of the comprehensive POS display ban was associated with a significant reduction in the proportion of young people transitioning from never having smoked to smoking initiation. We found that young people who entered the study in 2016, after the comprehensive ban was in place, were less likely to initiate smoking in the following year than young people who entered the study in 2014, prior to implementation. We also found that frequent visits to small shops were associated with an increased likelihood of smoking initiation but there was a significant reduction in the strength of this association after the comprehensive ban was implemented.

As noted in Chapter 1, Original research question, one of our research questions asked whether or not a reduction in exposure to POS advertising was associated with changes in smoking prevalence among young people aged 12–17 years. At the time the study was designed in 2011, the available data used in the power calculation suggested that the sample size we selected was sufficient to detect differences in smoking prevalence over time. However, we underestimated the rate of decline in youth smoking prevalence that occurred over the study period 2012–17, and therefore the study was not powered to detect changes in smoking in prevalence associated with implementation of the POS legislation.

Introduction

A number of studies have found a significant positive relationship between exposure to tobacco retailing surrounding the home and smoking outcomes in adolescents,110 but the mechanisms behind this association are poorly understood. One study showed direct effects, whereby greater tobacco retailer density was associated with higher brand awareness in never-smokers, and indirect effects through higher smoking among family and friends. 111 However, this study was at a single time point and therefore unable to explore differences in tobacco marketing practices either temporally or spatially. In this analysis, we used a repeated cross-sectional design to understand whether or not the tobacco visibility weighted outlet density (as explained in Chapter 5, Point-of-sale tobacco visibility tool, Results) was associated with the proximal smoking outcomes of brand awareness and perceived youth smoking prevalence for each year of the study (2013–2017). The hypothesised structural equation model is shown in Figure 8. It describes the relationship between tobacco outlet density, individual characteristics and proximal smoking outcomes.

FIGURE 8.

Structural equation model for pathways from tobacco retailer density and proximal smoking. Rectangles indicate measured variables, ellipses indicate latent variables, bold arrows indicate direct effects, bold dashed lines indicate indirect effects and curved lines indicate covariances.

Research questions

In this section, we answer the following research questions:

• What was the direct relationship between tobacco retailer density and brand awareness and perceived youth smoking prevalence before, during and after the POS ban?

• What was the indirect relationship between tobacco retailer density, and brand awareness and perceived youth smoking prevalence before, during and after the POS ban?

• How were these associations modified by frequency of small-shop visits?

Methods

We used repeated cross-sectional samples of participants in S2 and S4 for 2013 (n = 1482), 2014 (n = 1403), 2015 (n = 3807), 2016 (n = 3790) and 2017 (n = 3861). We excluded those with no density values (i.e. where the postcode was missing or incorrect).

We developed a structural equation model (see Figure 8) to examine the association between tobacco retailer density, weighted by total product visibility, and two proximal smoking outcomes. We hypothesised that declines in tobacco visibility as a result of the POS display ban would reduce the significance of both direct and indirect pathways between outlet density and the proximal smoking outcomes – brand awareness (as defined in Chapter 6, Brand awareness prior to introduction of point-of-sale legislation, Method) and perceived smoking prevalence (as defined in Perceived smoking prevalence, Methods) – among young people.

These outcomes were chosen as they show awareness of tobacco as a product and of smoking as a behaviour. We defined two mediators: whether or not the participant’s best friend smoked and whether or not someone in their family smoked (a latent variable from mother, father, brother and sister smoking status). We used age, sex, ethnicity, urban/rural status of the school, current smoking status, SIMD deprivation quintile, FAS score and family structure (whether or not the participant lived with two parents) as covariates. We defined small-shop visitation (whether they visited more than once a week) as an effect modifier and ran separate stratified models based on this variable. The models were run in a complete-case analysis.

Results

Analysis of crude trends in our data showed that brand awareness declined during the study period, whereas perceived youth smoking prevalence declined from 2013 to 2016, and rose in 2017. The prevalence of family members smoking also declined over the study period, but smoking among young people’s best friends declined to 2015 and then rose year on year to 2017.

The results from the structural equation modelling indicate that, for most years, there were no significant associations between tobacco retailer density and either brand awareness or perceived youth smoking prevalence (Tables 17 and 18). However, we found a direct effect of tobacco retailer density on perceived youth smoking prevalence in 2013. For every SD increase in density, the perceived prevalence of youth smoking would be expected to increase by 0.17 (95% CI 0.03 to 0.31) of its own SD (24.29) from its own mean (38.67), holding all other paths constant (see Table 17). We also found evidence of an indirect effect of tobacco retailer density increasing brand awareness by 0.10 (95% CI 0.02 to 0.18; SD 0.057; mean 0.052) through a family smoking pathway in 2014 (see Table 18). Importantly, these effects were found only in young people who reported a high frequency of visiting small shops.

Discussion

We found support in the data for a direct association between tobacco retailer density and perceived smoking prevalence before the implementation of the partial POS tobacco display ban in early 2013, and an indirect association through family smoking in 2014 before the implementation of the complete POS ban. However, this was among only those young people who visited small shops frequently and were therefore most exposed to tobacco marketing at POS. No significant effects were found after the comprehensive ban in 2016 or 2017. Our results suggest that the impact of tobacco retailer density on brand awareness is largely mediated by family smoking. This was also found in a study using another neighbourhood characteristic, deprivation. 112 Taken together, these findings suggest that, although a direct effect of legislation can be immediate, indirect effects through social networks can continue the impact on adolescent tobacco and smoking awareness for years after, highlighting the need to continue to emphasise reducing parental smoking in tobacco control policy. In conclusion, the POS ban may have disrupted the link between tobacco retailer density and brand awareness and perceived smoking prevalence. Further research is required to determine whether or not the POS ban has disrupted the link between tobacco retailer density and smoking susceptibility and smoking initiation.

Patterns and trends in e-cigarette use

Following spontaneous discussions about e-cigarettes in 3 of the 16 focus groups conducted in 2013, questions on e-cigarettes were added to the school survey in 2014. Regarding e-cigarette awareness, young people were given the following statement and question: ‘An e-cigarette/vapouriser/shisha pen is a tube that may look like a normal cigarette. An e-cigarette or vapouriser may have a glowing tip and puffs a vapour that looks like smoke but, unlike normal cigarettes, they don’t burn tobacco. Have you ever heard of e-cigarettes/vapourisers/shisha pens?’ To this they could respond ‘yes I have’, ‘no I have not’ or ‘don’t know’. The question related to e-cigarette use was ‘Which one of the following is closest to describing your experience of e-cigarettes/vapourisers/shisha pens?’, for which the response options were ‘I have never used them’, ‘ I have tried them once or twice’, ‘I use them sometimes (more than once a month)’ or ‘I use them often (more than once a week)’. During the DISPLAY study period, there was a sharp increase in the awareness and use of e-cigarettes among young people. Awareness of e-cigarettes increased from 74.7% (1028/1376) in 2014 to 85.1% (3226/3702) in 2016. Young people who responded that they had not heard of e-cigarettes or did not know whether they had heard of them were recoded as never having used them. Table 19 shows the change in frequency of use between 2014 and 2016.

It can be seen that the proportion of young people who had used e-cigarettes steadily increased over time from 14% to 28%. However, it should be noted that the majority of young people had tried e-cigarettes only once or twice. Nevertheless, the proportion of young people using e-cigarettes more than once per month increased from 3.9% in 2014 to 8.3% 2 years later.

In the 2017 survey, the question about awareness was dropped from the questionnaire and the response categories about frequency of use were changed to allow us to capture previous as well as current e-cigarette use. Table 20 shows the frequencies of e-cigarette use responses in 2017.

It can be seen that, by 2017, the proportion of young people who had ever used e-cigarettes increased further, to 33.7%. Thus, by 2017, a greater proportion of young people had tried an e-cigarette than had tried a conventional cigarette (21.1%).

Socioeconomic patterning of e-cigarette use and smoking

There was some evidence of socioeconomic patterning in e-cigarette use. Young people with high FAS scores were less likely on average to have tried e-cigarettes across all survey waves. This was ascertained using a logistic regression with FAS score, survey year, FAS score/survey wave interaction and robust standard errors for repeated measures on some participants (adjusted odds ratio for high FAS score vs. low FAS score 0.57, 95% CI 0.39 to 0.83).

Figure 9 shows the proportion of young people in S2 and S4 (i.e. aged 13 and 15 years) who had tried an e-cigarette by survey wave and by FAS score tertile.

FIGURE 9.

E-cigarette use by FAS score tertile.

The increase over time in the proportion of young people who had tried an e-cigarette can be seen clearly in Figure 9. There is some socioeconomic patterning in that those with a high FAS score have the lowest probability of trying e-cigarettes, but there is no significant interaction with time, suggesting that this pattern is fairly constant over the survey waves.

The proportion of young people in S2 and S4 (i.e. aged 13 and 15 years) who had tried smoking fell from 23.9% in 2013 to 15.3% in 2015 but then increased to 21.7% in 2017. The interaction between smoking prevalence by survey wave and FAS score is also shown below for comparison Figure 10. A logistic regression analysis found that there was a significant difference in the probability of having tried a cigarette between 2013 and 2015, and between 2013 and 2016, but no difference between 2013 and 2017. Figure 10 shows that there was a slight drop in prevalence and then a recovery. The patterning by FAS score in probability of having tried smoking is not statistically significant.

FIGURE 10.

Smoking prevalence by FAS score tertile.

Exposure to e-cigarettes

During the DISPLAY study, the e-cigarette market underwent a period of substantial growth internationally113 and the number of convenience-type stores stocking and displaying e-cigarettes markedly increased. 114,115 The retail audit element of the DISPLAY study found that around 77% of tobacco retailers sold e-cigarettes in 2014 and around 49% had e-cigarette POS displays. 116 The increase in awareness and use of e-cigarettes is mirrored in the focus group discussions with S2 and S4 pupils. During the first wave of focus groups in 2013, e-cigarettes were mentioned spontaneously by individuals in only three of the eight focus groups, but in the following year they were discussed in nearly every group. Therefore, from 2015 onwards, the topic guide was modified and addressed e-cigarettes directly. From the group discussions it was also evident that e-cigarettes and vapes were being actively promoted at the till point or adjacent to (or within) shuttered tobacco cabinets. These displays were described as striking and appealing, and caused the shuttered tobacco displays to be seen as very dull in comparison:

Like you see you walk into a shop and you don’t see it [tobacco], so you’re like, ‘Oh well. Fair enough. There’s no point buying it then,’ ‘cause you can’t see what you’re actually buying.

Whereas e-cigs, it’s all like decorated and everything.

Sitting right there – like all the colours.

And it influenced . . .

. . . Like if you seen the colours in the corner of your eye, you’d look.

School C4, S4, female

In addition, the young people spoke of other ways in which they were exposed to e-cigarettes. Important sources included family members and friends who used the devices:

I don’t really know . . . but like when my uncle got it, like he was just . . . well, he was passing it round to everyone ‘cause he thought it was cool! But . . . I don’t know. Wasn’t really . . . Don’t know.

School C2, S4, female

Well, I’ve tried a vape like quite a lot. I’ve just tried my brother’s. Like just when he’s got it out, I try it.

School C4, S2, male

. . . ‘cause my mum bought it for me, and it just never worked.

School C4, S4, female

Billboards, bus shelter adverts and social media websites, including YouTube (YouTube, LLC, San Bruno, CA, USA) and Facebook (Facebook, Inc., Menlo Park, CA, USA), were also frequently mentioned:

Like sometimes on buses and bus stops, and there’s TV adverts as well. Now there’s that ‘Vaping Rocks’ thing as well.

Like you just see them because they . . .

. . . they’re not like . . . I dunno. They seem to advertise them now like they used to advertise cigarettes ages ago . . .

I don’t know what it is, but I’ve seen it on bus stop – like the posters that roll round. I’ve seen it there, and on TV, and the radio as well.

School C2, group S4, male

Relationship between exposure to point-of-sale e-cigarette displays and subsequent e-cigarette use

It is well established that being exposed to POS cigarette displays influences smoking behaviour and intentions in young people. However, there is little evidence regarding the relationship between exposure to POS e-cigarette displays and e-cigarette use in young people.

In a cross-sectional study117 using data from the 2015 DISPLAY school surveys, we investigated the impact of the increasing retail availability and in-store promotions of e-cigarettes on the likelihood of young people trying e-cigarettes. Text in the following section has been adapted from Best et al. 117 This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

The analysis was conducted by logistic regression adjusted for recall of other e-cigarette adverts, smoking status and demographic variables as covariates.

Young people who recalled seeing e-cigarettes in small shops were more likely to have tried an e-cigarette (OR 1.92, 99% CI 1.61 to 2.29). Adolescents who recalled seeing e-cigarettes for sale in small shops (OR 1.80, 99% CI 1.08 to 2.99) or supermarkets (OR 1.70, 99% CI 1.22 to 2.36) were more likely to intend to try them in the next 6 months.

We also examined the same relationships longitudinally following young people across 1 year from 2015 to 2016. This confirmed that young never-users of e-cigarette who recalled seeing e-cigarette displays in shops and also on the internet were more likely to have tried e-cigarettes within 1 year than those who did not (Table 21).

Developing social norms around e-cigarette use

In the focus groups, young people’s views about e-cigarettes became increasingly voluble and complex over the course of the study. By 2017, it was clear that the young people were giving detailed accounts of emerging social norms among young people in relation to e-cigarettes.

From the 2017 focus group discussions, we identified four types of social exposure that were having an impact on social norms around e-cigarettes and their use.

Use of e-cigarettes and subsequent smoking initiation

At the point in the DISPLAY study when we had two waves of data (2015 and 2016) with information on e-cigarette use and conventional tobacco use, eight longitudinal studies, all conducted in the USA, had identified an association between e-cigarette use and subsequent smoking initiation in young never-smokers. 33–40 All of these studies found that young people who were never-smokers at baseline but had used an e-cigarette were more likely to have tried a cigarette at follow-up. However, all of these studies were conducted in the regulatory framework and social context of the USA, which is very different from those of the UK. Therefore, we sought to investigate these relationships in the DISPLAY study sample.

The summary below has been adapted from Best et al. 43 © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted. This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

We found that, of the young people who had never smoked but had tried e-cigarettes in 2015 (n = 183), 74 (40.4%) went on to initiate smoking conventional cigarettes at follow-up by 2016. This compares with 249 (12.8%) of those who had never smoked or tried an e-cigarette in 2015 (n = 1942) who went on to initiate smoking cigarettes by 2016.

In the logistic regression model, those who had tried an e-cigarette were found to be more likely (adjusted odds ratio 2.42, 95% CI 1.63 to 3.60) to subsequently try smoking a cigarette than never-smokers who had never used an e-cigarette. The analysis was adjusted for smoking susceptibility, smoking status of friends and family, age, sex, FAS score, ethnic group and school.

The interaction between e-cigarette use and smoking susceptibility was also significant, with e-cigarette use having a greater impact on the odds of a non-susceptible never-smoker trying smoking than on the odds of a susceptible smoker (who was already at high risk of smoking) doing so. In addition, an interaction was found between e-cigarette use and smoking within the friendship group, with e-cigarette use having a greater impact on the odds of trying smoking among young people who had no friends who smoked.

These results seem to indicate that young people who traditionally would not have been regarded as at high risk of initiating smoking (i.e. those who were non-susceptible to smoking and those with no friends who smoked) were at higher risk of smoking initiation if they had used an e-cigarette. However, there is still difficulty attributing causality as it is possible that these young people were already at high risk in some other way that was not captured by the measures we used to assess susceptibility.

Relationship between use of e-cigarettes and change in psychosocial risk factors and smoking initiation

More detailed analysis was possible once we had three waves of survey data with e-cigarette use, tobacco use and the other variables to explore the temporal relationships between the use of e-cigarettes and changes in psychosocial risk factors and smoking behaviour. Thus, it became possible to examine whether or not changes in psychosocial risk factors mediated the effects of e-cigarettes on tobacco use.

In the analysis we assessed whether or not e-cigarette use in 2015 in never-smoking young people predicted subsequent smoking by the 2-year follow-up in 2017. Table 22 shows that at the 1-year follow-up discussed above there was a significant relationship between e-cigarette use at baseline and subsequent initiation of smoking.

In a separate mediation analysis using the Stata package ‘ldecomp’ (Table 23), we found that friends’ smoking status in 2016 mediated 48.8% of the total effect of e-cigarette experimentation on subsequent smoking and that attitude to smoking mediated 49.1% of the total effect. The combined effect of the two mediators accounted for 66.7% of the total effect of e-cigarette use on subsequent smoking initiation.

These results suggest that the effect of e-cigarette experience on subsequent smoking experimentation is mediated by psychosocial mechanisms, including changes both in whom young people associate with and in their own attitudes to smoking. These findings are at variance with the common liability hypothesis,41 which suggests that the association between e-cigarettes and subsequent tobacco use is due to an underlying predisposition in young people to try such products. Instead, we suggest that there are changes in peer associations and attitudes following the use of e-cigarettes that make smoking initiation and tobacco use more likely.

Standardised packs

Standardised cigarette packs were introduced in Scotland and the rest of the UK in May 2016. Retailers were permitted to continue to sell the branded products to use up old stock; however, from May 2017, only the sale of cigarettes and tobacco in standardised packs was allowed. Australia was the first country to implement standardised packaging in 2012, and the UK followed the Australian lead by regulating the pack shape, colour, opening mechanism and font, along with the size and position of health warnings and the number of cigarettes in a pack.

In the 2017 school survey, young people were asked ‘In the past 30 days, have you seen any of these new greenish brown plain packs of cigarettes or tobacco (similar to above)?’, and a colour picture of a standardised pack was shown. In total, 2812 (76.7%) said that they had not seen one in the past 30 days. Of those who had, the most common place that they had seen it was on the ground or in rubbish (13.7% of all respondents had seen a pack there). In the 2017 focus groups, the facilitator first sought unprompted views about standardised packs. After this, she distributed three or four examples of actual standardised packs to the young people in the group and asked them what they thought of these examples (see Report Supplementary Material 1). The focus groups took place 2 or 3 months before full implementation of standardised packaging of tobacco. In spite of this, there was a high level of awareness of standardised packs before May 2017. Smokers in the focus groups had purchased them, and they and other pupils had seen them in the possession of friends and family members, as well as in litter. Pupils’ views of the new packaging were generally negative. It was described as unattractive and depressing, and was compared unfavourably with branded tobacco packaging and current e-cigarette marketing:

They’re very depressing, you just want to sit there and slit your wrists.

School C1, group S2, male

It’s like sludge green . . .

. . . army green.

The Golden Virginia [Imperial Brands plc] ones are like scaffie green . . .

School C4, group S4, female

The images on the packs elicited particularly negative responses:

See my niece? – she’d . . . she’d cry if she saw that, that’s not something you should expose children to . . .

School C1, group S4, male

I hate that one with the hole [in throat] . . .

Like that’s so disgusting!

I don’t really want to see that . . .

Is it on the other side? Can we turn it over?!

It is on the other side. Oh!

It’s so disgusting.

Makes me feel sick.

School C2, group S4, female

However, there was no consensus about the standardised packs’ likely impact. Some argued that they would have a widespread impact, whereas others stated that any impact would be restricted to young people who were non-smokers/occasional smokers and that established smokers would not change their practice:

Just the fact that it’s . . . you only see a bad image. You don’t see like fancy colours or bright colours that make you look . . . well, ‘Oh, that looks cool’.

But do like some people actually care about that?

People that smoke, I would say don’t care about it because they just want to smoke, but for . . . for someone who’s not tried it before . . . it will not tempt them so much I’d say. It is helpful to stop young people getting in to it in the first place.

School C2, group S4, male

Like so they’ve changed it, but anyone that smokes is gonna smoke regardless of what the box looks like.

They’re not gonna give up just because of a fag packet.

School C4, group S4, female

Point-of-sale displays and compliance with the legislation

The DISPLAY study found that, before the implementation of the POS tobacco display ban in Scotland, tobacco displays were highly visible in both large supermarkets and small shops. These displays were usually placed at customer eye-level, behind the POS, and often in proximity to goods, such as confectionery, that were likely to be of interest to children. Some POS displays were also visible from the public footpath outside small shops. The great majority of young people participating in the school surveys recalled seeing POS tobacco displays prior to the implementation of the POS legislation, with young people from the least affluent backgrounds most likely to recall seeing them. This seems to be consistent with the finding that CTN and convenience stores in areas of higher deprivation had larger average display sizes than those in less deprived areas. In focus group discussions, young people reported finding POS tobacco displays very eye-catching and colourful, and in one group they were likened to an ‘adult sweetie shop’.

Small retailers had largely negative expectations about the potential impact of the POS legislation prior to its implementation. Concerns were expressed about the cost of converting tobacco gantries and the impact of shutters and covers on transaction times and shoplifting (while servers’ backs were turned to access tobacco products). However, in practice, implementation was straightforward. Among the retailer panel members, gantries were most often supplied or converted by tobacco companies at no extra cost. Contrary to their expectations, retailers and staff adapted easily to the covered displays, and the anticipated increases in transaction times and shoplifting did not occur. Views about the impact on customer behaviour were mixed, with some retailers reporting no change and others reporting a drop in tobacco sales.

The observational audit demonstrated that compliance with the legislation was high after implementation both in large supermarkets and in small shops. Our findings are consistent with other evaluations of POS legislation. For example, an evaluation of legislation in Victoria, Australia,119 found that 94% of retail outlets complied with POS legislation 3–4 months after implementation, and in Norway and Ireland compliance rates as high as 97% post implementation were recorded. 21 Historically, when only partial restrictions have been introduced, compliance has been lower, suggesting that complete bans are likely to be both more effective and easier to implement. 9

The high compliance found in this study demonstrates that it is possible to implement comprehensive bans across a wide range of retail outlets. However, although the POS legislation was very successful in concealing tobacco products from public view, tobacco products continue to be stored at POS, with the covered storage units carrying large signs indicating that tobacco is for sale. Thus, tobacco as a generic product continues to have a very strong presence in the retail environment. Furthermore, the financial incentives retailers were offered by tobacco company representatives persisted even after gantries had been covered up, with retailers continuing to be rewarded for product placement, availability and sales. There was also evidence that retailers were incentivised to promote products verbally to customers.

Changing retailer density, tobacco product visibility and storage unit visibility

Retailer density not only increases the availability of tobacco products by increasing opportunities to purchase them, but also enhances their visibility,120–124 and it is a likely driver of smoking inequalities. 84 Other research in Scotland125 has found that higher neighbourhood density of tobacco retailing is associated with higher odds of having ever smoked or reporting current smoking among adolescents. Similarly, adults living in areas with the higher densities of tobacco retailing also have higher odds of being a current smoker and lower odds of being an ex-smoker. 126

In the four DISPLAY study communities, the overall availability of tobacco products as measured by retailer density remained largely unchanged; however, when the retailer density measure was weighted by total product visibility scores, we found that a significant reduction in young people’s neighbourhood exposure to tobacco products, indicating that the POS legislation has been successful in reducing the visibility of branded tobacco products.

The study team was also able to extract data for the whole of Scotland from the Register of Tobacco and Nicotine Vapour Product Retailers. 65 Analysis of these national data showed that between 2013 and 2015 the number of retailers in Scotland reduced by 13% following the implementation of the partial POS ban. Retailer density fell across all deprivation neighbourhood quintiles, but less so in the most deprived ones. This was followed by a modest increase in 2016 and 2017, and by the end of the study period the overall number of retailers was similar to baseline (2013). The overall increase in tobacco retailer density was driven largely by an increase in the number of tobacco retailers in about one-fifth of local authorities, which occurred in the most disadvantaged neighbourhoods, resulting in increased inequality in tobacco availability.

The findings emphasise that tobacco remains ubiquitous across Scotland; this has important implications for public health policy.

Smoking susceptibility and smoking experimentation

A key objective of the POS display legislation was to reduce smoking susceptibility and the risk of smoking initiation and regular smoking among young people. Previous research88,96 has found that high levels of exposure to POS tobacco displays (as measured by shop visit frequency) increases smoking susceptibility. However, there have been no studies of the impact of a POS display ban on smoking susceptibility.

Consistent with the earlier research, we found that young people who visited shops more frequently were at greater risk of smoking susceptibility than those who visited shops less often. Using survival analyses techniques, we found also that young people’s risk of becoming susceptible to smoking after 1 year in the study was much lower among those who entered the study after the comprehensive ban than among those who entered the study after the partial ban. To our knowledge, this is the first study to demonstrate that legislation banning POS tobacco displays can reduce young people’s susceptibility to smoking. Moreover, the finding that, compared with the partial ban, the comprehensive ban has reduced the impact of shop visit frequency on smoking susceptibility is suggestive of a dose–response relationship.

In our original study protocol, we intended to assess the impact of the POS display ban on smoking prevalence. However, in line with other national data,29,97 smoking prevalence in our samples of young people was low. Therefore, the study was underpowered to assess the impact of the legislation on smoking prevalence. Previous research has demonstrated an association between exposure to POS displays and smoking initiation,127 and so, instead, we examined the impact of POS legislation on smoking initiation.

Using survival analysis techniques, we also found that the risk of smoking initiation 1 year after entry into the study was lower among those who entered the study after the comprehensive ban than among those who entered after the partial ban. As in the analysis of smoking susceptibility, the implementation of a comprehensive ban of POS tobacco displays reduced the impact of shop visit frequency on smoking experimentation compared with only a partial ban, suggesting that there could be evidence of a dose–response relationship. However, it should be noted that the young people in our sample who initiated smoking are actually a subgroup of the young people who became susceptible to smoking. This is because, in the analysis of smoking susceptibility, young people who initiated smoking were defined as smoking susceptible, as they were assumed to have passed through a period of smoking susceptibility prior to smoking initiation. The two analyses are therefore not independent, and it is important to assess the impact of POS legislation on independent samples.

Policy

The high level of compliance found in this study demonstrates that it is possible to implement comprehensive legislation that prohibits POS displays across a wide range of retail outlets without disrupting the retail environment. In addition, as found with other tobacco control legislation, such as the smoke-free legislation, comprehensive bans are likely to be easier to understand and implement than partial bans.

However, before the legislation had been implemented, small retailers had some concerns about its impact on their businesses, and many looked to, and received advice and support from, tobacco industry representatives. Additional advice from trading standards support would have been welcomed, and would most likely have exposed retailers to a wider range of views regarding tobacco control measures. Some minor instances of non-compliance were observed, which further underline the need for advice and monitoring. Given that most of the contraventions observed were accidental, it is likely that only ‘light touch’ enforcement will be necessary in the future.

Small retailers are important stakeholders in tobacco control, and it is important to take into account their perspectives and to learn from their experiences. They can provide feedback on aspects of legislation that might be problematic to implement or might incur resistance. Future studies to evaluate or inform the development of tobacco control policies might include exploration of retailer perspectives.

The POS legislation has been successful in reducing exposure to tobacco products at both a community and an individual level. Although the research design precludes the conclusion that the relationships are causal, implementation of the legislation in the four study communities was associated with a reduction in the anticipated youth outcomes. However, although tobacco storage units have been covered up, the units themselves continue to be prominent in retail outlets. Thus, the POS display ban has not fully eliminated tobacco as a generic product from the retail environment in Scotland. Cues that tobacco is for sale are still highly visible in many shops. The continuing existence of these cues in the retail environment was observed, particularly in areas of deprivation, where inequalities in storage visibility have widened since the POS ban was implemented. Furthermore, the availability and density of tobacco retailers has remained high across Scotland, with greater availability in more disadvantaged neighbourhoods. This research suggests that regulating the number of tobacco retail outlets and the size, design and position of storage units, and using generic signage to indicate tobacco availability, might contribute to reducing these inequalities.

Tobacco companies are already prohibited from using financial incentives with retailers for product placement, availability and sales and the verbal promotion of products to customers, and this should continue to be emphasised to retailers.

With the reduction in product visibility in shops that young people visit most frequently, the relative influence of parental smoking on young people’s smoking susceptibility and tobacco access will increase. This underlines the importance of continuing to focus tobacco control policy on reducing parental smoking, particularly in front of children and in the home, and encouraging parents to keep tobacco products out of sight to minimise the influence of their behaviour on the uptake of smoking by their children.

Although the implementation of the POS legislation was associated with a reduction in perceived smoking prevalence, the young people in our study continued to greatly overestimate smoking prevalence, especially among their peers. Targeting young people with accurate information about the decline in youth smoking prevalence might help to counter the continuing influence of the generic presence of tobacco in the retail environment on the normalisation of smoking behaviour.

It is estimated that about 80% of the world’s population continues to be exposed to POS advertising and displays of tobacco products. This study has provided robust evidence about implementing POS legislation and protecting young people from exposure to tobacco products and the subsequent improvement in youth attitudes and reduction in smoking susceptibility and smoking initiation. The methods reported here, including the visibility tool, provide a baseline measure for evaluating the legislation, and an approach that can be replicated in other settings and jurisdictions where similar legislation is planned or has been implemented. The methods reported here could also be modified to assess visibility of tobacco displays in jurisdictions where tobacco products are sold outdoors.