Promotion of tobacco contributes to initiation of smoking by adolescents1 2 yet tobacco continues to be promoted at the point of sale (POS) in most jurisdictions. Tobacco companies invest significantly in the promotion of tobacco at the POS3 4 using in-store advertising, prominent displays of tobacco products, or both. The volume and type of POS tobacco promotion permitted varies considerably between jurisdictions but most have few or no restrictions. However, implementation of restrictions on POS promotion is increasing. For example some Canadian provinces and Iceland do not permit any advertising or display of tobacco products at the POS. Ireland will introduce a similar ban on 1 July 2009, and consultations on a POS ban are underway in England and New Zealand.

Some jurisdictions have limited restrictions on POS displays. In New Zealand, restrictions include that “Smoking kills” signs must be displayed, tobacco product displays are limited to 100 packets per point of sale, tobacco products may not be within 1 m of children’s products, and tobacco product displays should not be visible from outside the store (http://www.moh.govt.nz/moh.nsf/wpg_Index/About-smokefreelaw-retailers#1). However, compliance with the regulations is often poor.5 In a compliance study in New Zealand, the most common breaches of the regulations were: failure to appropriately display a “Smoking kills” sign (30% of all stores), POS display visible from outside the store (25%), tobacco products displayed within 1 m of children’s products (24%) and displaying more than 100 packets per point of sale (16%).

In a recent systematic review, uptake of smoking or increased susceptibility to smoking was significantly associated with tobacco promotion at the POS in seven of eight observational studies and two experimental studies of teenagers.2 3 6^–13 However, most studies that have investigated the association between POS tobacco promotion and smoking or susceptibility to smoking were conducted in the USA.6 10^–12 In the USA there are few restrictions on tobacco promotion (including on advertising and display of tobacco at POS), and in-store tobacco promotion is usually extensive.14 15 Therefore, findings may not apply to different settings such as those with greater restrictions on tobacco promotion. In addition, the large amount of in-store advertising in the US means it is not possible to separate the impact of POS tobacco displays on smoking uptake from other types of in-store advertising in these studies.

We investigated the association between exposure to POS displays of tobacco products and smoking, experimenting with smoking and susceptibility to smoking uptake among teenagers in a jurisdiction where most tobacco advertising (including POS advertising) is banned, and where tobacco promotion is restricted to a limited, but still prominent, display of tobacco products at the POS.

METHODS

All schools in New Zealand with Year 10 (age 14–15) students were identified from the Ministry of Education database and invited to participate in this study. Schools that agreed to participate were provided with an information sheet, brief instructions, questionnaires and a reply-paid envelope for returning the questionnaires. Schools were asked to administer the questionnaire to all their Year 10 students. Students were informed that the survey was anonymous, and that they could refuse to participate at any time. Teachers remained in class while the students completed the pen and paper questionnaires, but were instructed not to examine the surveys for completeness or response. The Auckland Ethics Committee of the Ministry of Health granted a waiver of the formal review and consenting processes.

Smoking behaviour and susceptibility variables

Smoking status was measured using two questions: (i) “have you ever smoked (even just a few puffs)?” with yes/no response categories; and (ii) “how often do you smoke now?” with the response categories, “I have never smoked/I am not a smoker now”, “less often than once a month”, “at least once a month”, “at least once a week” and “at least once a day”. If students said they smoked at least daily, weekly or monthly they were defined as “current smokers”. “Experimental smokers” were defined as students who had smoked a cigarette but were not a smoker at the time of the survey or smoked less than once a month. “Never smokers” were those who had never smoked.

Susceptibility for smoking uptake among students who had never smoked was assessed using two questions developed and validated by Pierce and colleagues,16^–18 and employed by the Global Youth Tobacco Surveillance team.19 Students were defined as being non-susceptible if they reported that they definitely will not smoke in the next year and definitely would not smoke if offered a cigarette by their best friend. Students who answered “probably not”, “probably” or “definitely yes” to either of these questions were defined as being susceptible.

The outcomes of interest in this study were: (1) being a student who has never smoked but is susceptible to smoking initiation compared to being a non-susceptible never smoker, (2) being an experimenter compared to having never smoked and (3) being a current smoker compared to all other students.

Exposure and perception of tobacco displays at the point of sale (POS)

Two measures were used to assess exposure to tobacco displays. Firstly, students were asked how often they visited three store types: corner shop/convenience stores, petrol/service stations and supermarkets. They could answer on a response scale from “more than once a day” to “less than once a week” and “never”. These store types are the main retail settings where tobacco products are sold in New Zealand, and nearly all prominently display POS tobacco products.20 A variable “store visit frequency” was derived by using the response from the most frequently visited of the three store types for each participant to divide students into the categories “at least daily”, “2–3 times/week”, “weekly” and “<1 time/week”. Previous work examining student exposure to POS advertising collapsed POS exposure to different store types into two categories, high exposure or low exposure.10

Secondly, based on questions used in the US National Youth Tobacco Survey (http://www.cdc.gov/tobacco/data_statistics/surveys/NYTS), students were asked separately for each store type visit, how often they noticed tobacco. Response options were “every time”, “most times”, “sometimes”, “hardly ever”, “never” and “I never go to this type of store”. Subjects’ exposure status was assessed from their response for the store type they most frequently visited.

Covariates

Questions about age, gender, ethnicity, peer/parental smoking and smoking in the home were also included. Smoking in the home is assessed by asking, “Do people smoke inside your home?” The response options are yes, no or sometimes. As students were free to choose more than one ethnicity, ethnic classification here used a priority ordering of Māori, Pasifika, Asian, New Zealand European and other for multiple specifications. This order of prioritisation is consistent with the method used in other New Zealand surveys including the New Zealand Tobacco Use Survey.21

Socioeconomic status (SES) was measured using school decile. This is a measure of the characteristics of the local community, whereby “low” decile usually indicates “low” socioeconomic status. Every school in New Zealand is assigned a decile by the Ministry of Education and participating schools were grouped according to their decile. In line with New Zealand reporting conventions, we collapsed deciles into three groups: low SES (deciles 1–3), medium SES (deciles 4–7) and high SES (deciles 8–10). In New Zealand there are 21 geographically defined District Health Boards (DHBs) which provide health and disability services for their local population.

Statistical analysis

Data checks and descriptive statistics were undertaken for all variables. The χ² goodness of fit test was used to compare the distribution of response across store type and visit frequency. Three-level mixed-effect logistic regression models were used to investigate the association between the three outcomes (ie, being susceptible to initiation among all those who have never smoked, an experimenter among those who do not currently smoke or a current smoker among all students) and the two measures of POS tobacco exposure: “store visit frequency” and self-reported perception of tobacco at the POS entered simultaneously into the models. The first level was defined using DHB, and school defined the second level of the regression models. Students defined the third level. Adjusted odds ratios controlled for age, sex, ethnicity, peer and parental smoking, school socioeconomic status and whether or not smoking occurs in the home. Analyses were conducted using Stata version 10 (StataCorp, College Station, Texas, USA). Interactions between the two exposure variables and between the exposure variables and school socioeconomic status (decile) were examined.

RESULTS

Every secondary school in New Zealand with Year 10 students was invited to participate in this survey (n = 511). Almost half (47.4%) of all eligible schools agreed to participate (n = 242). In total, 27 757 (44.8%) student questionnaires were returned from the national population of 62 018 Year 10 students. Most (98.2%) Year 10 students were 14 or 15 years old. Younger and older Year 10 students were excluded due to the potential confounding effect of age on smoking behaviour, and the difficulty of adjusting for this given the small number of Year 10 students who are younger than 14 years and older than 15 years of age.

Demographic characteristics and smoking behaviour of the participants are shown in table 1. The distribution of age, ethnicity and gender of respondents matched that of the national Year 10 population. However, the lowest decile schools were underrepresented in the sample. A total of 30% of schools with Year 10 students are deciles 1 and 2 compared to 20% of the sample schools. The lowest socioeconomic category comprised 46 schools, most schools were in the medium decile category (113) and 80 schools were in the high socioeconomic category.

Corner shop/convenience stores were the most frequently visited store type with 74% of students visiting them weekly or more often. Petrol/service stations and supermarkets were visited at least weekly by 57% of students. Descriptive statistics for store visit frequency and noticing tobacco in the most frequently visited stores for each subject are described in table 2.

The frequency of noticing tobacco varied by store type and frequency of visiting stores (χ² goodness of fit test, p<0.001). For example, students visiting corner shop/convenience stores were significantly more likely to report noticing cigarettes “every time” or “most times” whereas students visiting petrol/service stations were less likely to report noticing cigarettes “every time” or “most times”. Moreover, students were significantly more likely to visit corner shop/convenience stores at least 2–3 times weekly compared with petrol/service stations or supermarket (p<0.001). Most never smokers (88%) reported visiting stores weekly or more often.

Table 3 shows the crude and adjusted associations between the exposure variables and smoking-behaviours. Among never smokers, the adjusted odds of being susceptible to smoking uptake among those visiting stores at least 2–3 times per week were approximately double compared to students who visited stores less than weekly. The odds of a student being susceptible to initiation were also about double if the student reported noticing tobacco “most” or “every” time they visited their most frequently visited store type compared to those who never noticed tobacco products. Odds of being susceptible to smoking increased with increasing exposure or reported noticing of tobacco.

Analysis of the association between exposure variables and experimenting with smoking or current smoking yielded very similar results (table 3). The adjusted odds of being a current smoker were over three times greater among subjects who reported noticing tobacco products at every visit compared to those who report never noticing tobacco.

When other variables associated with smoking are adjusted for exposure to tobacco displays and noticing tobacco, parental smoking significantly increased the odds that a teenager had experimented with smoking (odds ratio (OR) 1.7 one parent smoking, OR 1.8 both smoke) or was a current smoker (OR 1.3 one smoker, OR 1.4) but was not significantly associated with susceptibility to initiation of smoking. Smoking in the home was still significantly associated with all three outcomes (OR ranging from 1.2 to 1.4) but the effect was slightly less than frequent exposure to displays and noticing tobacco. Females were slightly more likely to be susceptible to initiation and current smoking but gender was not significantly associated with experimenting. There were no significant interactions between ethnicity and exposure to displays for all three outcomes. Māori, Pasifika and Asians were slightly but significantly less likely to be susceptible to initiation (OR 0.84) but Māori were more likely to have experimented or be current smokers compared to New Zealand Europeans. Students at low or medium SES schools had slightly lower odds of being susceptible to initiation (OR 0.74 medium, OR 0.63 low) but school SES had no association with experimenting. Peer smoking remains significantly and strongly associated with all three outcomes (one friend OR 2.4, two or more 3.0 for being susceptible to initiation, OR one friend 3.3, two or more 6.0 for experimented and one friend OR 5.2, two or more 28.3 for being a current smoker).

An interaction between socioeconomic status and store visit frequency was present when examining the odds of being a current smoker, so these results are not displayed in table 3. Table 4 shows the results of the analysis of the association between store visit frequency and current smoking stratified by SES. The association between current smoking and store visit frequency was strongest in students from high socioeconomic status (SES) schools, intermediate in medium school SES and minimal in low school SES groups (where all subjects had increased odds of being a current smoker regardless of their store visit frequency).

DISCUSSION

We investigated the association between exposure to tobacco displays in shops and three different measures of smoking susceptibility and behaviour in teenagers. Two-thirds of teenagers visited stores where tobacco products are prominently displayed at least 2–3 times per week, and most reported that they always or most times noticed cigarettes in these settings. Strong statistically significant associations were found between both measures of exposure and all three outcome measures after adjustment for a range of known determinants of smoking susceptibility and initiation. Both measures of exposure were independently associated with susceptibility to smoking uptake, experimenting or current smoking. The association between exposure to point of sale tobacco displays and susceptibility to smoking uptake or experimenting is comparable to the associations between these outcomes and parental smoking, smoking in the home and school SES.

A dose response (ie, increasing odds of being susceptible, or an experimenter or currently smoking when exposure increased) was present. The only exception to this was elevated odds of being a current smoker for all frequencies of store visits among students from low SES schools. One explanation for this could be that low SES students have other significant determinants, which have greater impact and were not included in this analysis, such as more smoking among non-parental adult role models and siblings. Another explanation may be a greater reliance by low SES current smokers on less expensive social supplies of tobacco rather than purchasing from retail sources. However this needs to be verified by further investigation.

Strengths and limitations

This is the first study that the authors are aware of to explore the association between exposure to POS tobacco displays and smoking uptake in a jurisdiction where tobacco advertising is banned, and where retail displays of tobacco products are the only form of tobacco promotion. The study had a very large sample size, and used three different outcome measures and two different exposure measures. Store visit frequency is a neutral measure of exposure, which may be less susceptible to biased recall because it is not an overtly tobacco-related enquiry. The analysis adjusted for a range of established influences on smoking behaviours in this age group, including ethnicity, school socioeconomic status and parental and peer smoking.

A possible weakness is that the school response rate to this census survey (47%) was modest and was lower in 2007 compared to previous levels for the same survey (57% to 67%). Decreasing survey response rates is a global phenomenon.22 However, all deciles and DHBs are represented and the sample includes almost half of all Year 10 students in New Zealand. Another possible criticism is that this study does not investigate the impact of POS displays stipulated by the current New Zealand regulations due to the high level of non-compliance that has been demonstrated.5 However; our study investigates the association of POS exposure which actually occurs in New Zealand with smoking-related outcomes. This is a more relevant investigation, since this is how POS displays are experienced in practice in New Zealand, and we suspect non-compliance will be a common feature in jurisdictions with such policies. Finally, the descriptive data about the frequency of visiting stores and noticing tobacco products focused only on the most frequently visited store type so will be an underestimate of total adolescent exposure to tobacco displays.

Comparison with literature and implications

Seven out of eight cross-sectional studies which have investigated this issue have found statistically significant associations between exposure to POS tobacco promotion and smoking.3 One of these studies also found a statistically significant association between exposure to POS tobacco promotion and susceptibility to smoking.10 Tobacco promotion at the POS was a significant predictor of smoking initiation in a longitudinal study conducted in California, USA.12 A statistically significant association between student smoking and the density of stores selling tobacco has also been described.23 This New Zealand study is therefore consistent with previous studies conducted in other countries.

The discovery of an independent association of the two different measures of exposure, frequency of visits and reported noticing of tobacco (ie, perceived exposure) is also consistent with findings of Feighery et al10 who suggested that self-reported perceived exposure measures a different construct compared to a frequency of store visit measure and may reflect a bias in attention to tobacco.

The use of cross-sectional data prevents us from assigning cause and effect definitively to visiting stores or noticing tobacco displays and smoking behaviour. In particular an association between current smoking and increased store visiting or noticing of cigarettes in stores may be due to smokers buying their cigarettes from those stores, or becoming more aware of cigarettes as a result of being a smoker. Youths contemplating or curious about smoking may also be drawn to the displays. However industry documents and general marketing principles indicate that displays and other types of marketing create and enhance curiosity about products, so the direction of effect is highly unlikely to be solely from outcome to exposure.

Tobacco company investment in promotion at the POS has grown as other modes of promotion are banned.4 24 25 Information about the purpose of POS displays as described in internal tobacco industry documents is consistent with an association between smoking and exposure to tobacco displays. For example, Philip Morris sales representatives are advised that one of their most important jobs in to gain optimal product exposure and effective in-store visibility because this attracts new consumers.24 The purposes of POS merchandising are described clearly by British American Tobacco: (1) to inform the consumer of the presence of the brand, (2) to promote recognition of the brand, (3) to generate interest and excitement about the brand and (4) to stimulate trial purchase and repurchase.26

The reason for the impact of POS promotion of tobacco have been analysed by several authors using tobacco industry documents. The widespread, conspicuous presence of tobacco displays at every corner store creates an impression of perceived popularity and normalises the use of tobacco within society.4 Young people often overestimate the prevalence of smoking and this overestimation is associated with an increased risk of smoking.27 School students seeing photos with POS tobacco advertising were significantly more likely to overestimate smoking prevalence than students seeing photos of POS without tobacco promotion.28 Also, in this study, students exposed to a photo of a POS where cigarettes were displayed without tobacco advertising were more likely to perceive it would be easier to buy tobacco in these stores and more likely to recall displayed brands popular among adults. The proximity of tobacco to staple consumer items and attractive sweets creates a “friendly familiarity” impression of tobacco.29 Any notion that a child may have that tobacco is addictive and dangerous is undermined because tobacco is sold in the same way as treats, bread and milk, and it is not sold or restricted in the same way as other dangerous legal products such as medicines.

Conclusions

This study provides evidence that young people’s exposure to tobacco displays at the POS is significantly associated with being susceptible to smoking, experimenting with smoking and current smoking. The use of cross-sectional data means that cause and effect cannot be attributed to the exposure and outcome variable. Further prospective studies and research to investigate the impact of POS bans on smoking uptake should be undertaken.