To sell or not to sell: cigarette sales in alcohol ...

Downloaded from http://tobaccocontrol.bmj.com/ on November 28, 2017 - Published by group.bmj.com

TC Online First, published on November 27, 2017 as 10.1136/tobaccocontrol-2017-053944 Research paper

To sell or not to sell: cigarette sales in alcohollicenced premises Suzan Burton,1 Mark Ludbrooke,2 Kelly Williams,2 Scott C Walsberger,2 Sam Egger2 1

Abstract Objective To obtain insight into tobacco retailing by alcohol-licenced premises, in order to understand the financial importance of tobacco sales for such retailers. Method Data were collected by a telephone survey of 1042 clubs, hotels and packaged liquor outlets in New Correspondence to South Wales, Australia. The response rate was 86.1%. Professor Suzan Burton, School Qualitative and quantitative data were obtained. Logistic of Business, Western Sydney and linear regression were used to determine factors University, Sydney, NSW 2751, associated with the probability of selling and stopping Australia; s .b urton@w esternsydney.e du.a u selling and the importance of cigarette sales. Results More than a third (36.4%) of premises Received 12 July 2017 contacted did not sell cigarettes. 147 (an estimated Revised 26 October 2017 18.1% of those who had ever sold) had stopped selling. Accepted 27 October 2017 There were significant differences in the probability of selling, in the reported importance of cigarette sales and in the probability of stopping selling, between different outlet types and other outlet characteristics (number of gaming machines, proximity of nearest alternative tobacco retailer and remoteness). Outlets where alcohol can be consumed were more likely to rate cigarette sales as ’not important’ than ’important’. Conclusions Despite claims by tobacco companies that tobacco sales are important for many Australian retailers, tobacco sales appear to be of limited importance for alcohol-licenced premises. This means that opposition to stopping tobacco sales where alcohol is consumed and/ or sold may be less than expected. School of Business, Western Sydney University, Sydney, NSW, Australia 2 Cancer Council New South Wales, Kings Cross, New South Wales, Australia

To cite: Burton S, Ludbrooke M, Williams K, et al. Tob Control Published Online First: [please include Day Month Year]. doi:10.1136/ tobaccocontrol-2017-053944

In response to evidence that tobacco retailers contribute to smoking by creating environmental cues to smoke and by making cigarettes accessible,1–5 multiple commentators have called for a review of tobacco retailing and a decrease in the number of retailers.4 6–8 Reductions in tobacco retailer numbers can be achieved by large increases in an annual licencing fee, as was observed in South Australia.9 Reductions can also be achieved by other legislative actions: a growing number of jurisdictions (at the time of writing, in Hungary and in the USA) have implemented, or are implementing, policies with the aim of decreasing the density of tobacco retailers, by limiting the types and/or locations of stores that can sell tobacco and/or by imposing caps on the number of retailers.10–13 With the exception of changes in Hungary,11 such legislation appears to grandfather existing retailers but should result in a decrease in the number of retailers over time. However, the possibility of using policy to restrict the number of tobacco retailers was not even mentioned in a survey of tobacco control policies in 34 European countries,14 suggesting that retailer reduction policies are not high on the legislative agenda in most jurisdictions.

In the absence of specific policies, retailers voluntarily stopping selling would appear to be the only mechanism for reducing the number of tobacco retailers. Until recently, research examining why retailers stop selling tobacco has focused on pharmacies,15 a retail group that does not sell tobacco in countries such as Australia. Research with other types of retailers found the main motivators to stop selling tobacco were regulatory pressure, low profit from tobacco sales and/or health or ethics-related concerns.16–18 However, a limitation of only studying retailers who stop selling is that research cannot determine to what extent the same factors may be present for retailers who do not stop selling. For example, the only systematic analysis to date of retailers who have stopped selling tobacco found that declining profits from tobacco sales were generally a necessary but insufficient condition for stopping selling.16 This leaves open the question of whether retailers who still sell tobacco are also experiencing low profits from tobacco sales, thus increasing the likelihood of other retailers leaving the market. Yet studies of retailers’ attitudes towards selling tobacco are limited to small samples in England19 (n=62) and New Zealand20 (n=21). Both studies found that many retailers believe that tobacco sales are important, despite low profit margins, because they are believed to result in sales of other items to customers who purchase tobacco. One study in Pennsylvania found, however, that only a small percentage of purchases from urban corner stores (13%) included tobacco, and the majority of those sales (61%) did not include any other products.21 In the face of this limited understanding of why many retailers continue to sell tobacco while others stop, we planned a large and systematic study of tobacco retailers in New South Wales (NSW), the largest state in Australia. NSW has what is sometimes called a ‘negative licensing scheme’.22 This means that a retailer can lose their right to sell tobacco if they are repeatedly convicted of violating retail regulations, but there are no other limitations on who can sell and no licence fee.23 We decided to focus on tobacco retailers who are licenced to sell alcohol for three reasons. First, tobacco retailers who sell alcohol constitute the largest category of tobacco retailers in NSW (34% of all tobacco retailers) but, together with newsagents, comprise only 1.4% of all tobacco sales.22 24 This discrepancy between availability and consumption is consistent with the notion that alcohol-licenced outlets are being disproportionately used by lighter smokers for impulse purchases while consuming alcohol.25 Second, consumption of alcohol has repeatedly been associated with

Burton S, et al. Tob Control 2017;0:1–8. doi:10.1136/tobaccocontrol-2017-053944

Copyright Article author (or their employer) 2017. Produced by BMJ Publishing Group Ltd under licence.

1


Research paper Table 1 Operational definitions for possible predictors of selling, stopping selling and importance of selling cigarette sales Variable

Notes

Licence type Club

Refers to a member based organisation (a ‘registered club’), where alcohol (for consumption on the premises) and other services are available to members, their guests and visitors. May have gaming machines.

Hotel

Refers to a licence available to accommodation hotels, ‘pubs’ and ‘bars’. Alcohol may be sold for consumption on the premises and also to take away. May have gaming machines.

Packaged liquor outlets

This licence allows alcohol to be sold through a bottle-shop (liquor store), supermarket, grocery store, home delivery, mail order or the internet, for consumption off the premises only. Bottle-shops and supermarkets may be independent/franchisee outlets or part of a corporate owned (not franchised) chain.

Extended hours licence

For clubs and hotels, refers to the ability to sell alcohol outside 05:00 to midnight Monday to Saturday, and 10:00–22:00 on Sundays (with some exceptions relating to public holidays and particular licence types).32 For packaged liquor outlets, this licence allows sales on a Sunday morning (beyond the standard closing time of 23:00 Monday–Sat and 22:00pm Sunday).

Number of gaming machines

Relevant for club and hotel licences only. Number obtained from licence details, with categories for models chosen to provide reasonably balanced groups subject to the constraint that most outlets had no gaming machines.

Remoteness

Calculated using the Australian Bureau of Statistics Postcode to Remoteness Areas correspondence.33

Number of staff*

Proxy for turnover. Categorical groupings shown in models were based on survey categories (chosen to reflect expected variations in size).

Distance to nearest cigarette retailer*

Proxy for availability of substitute retailers. Categorical groupings shown in models were based on survey categories (chosen to reflect expected variations in distance).

*These questions only asked to current sellers, so these are not included in models predicting stopping selling and importance of cigarette sales.

increased rates of smoking, and alcohol-licenced premises where people can smoke (usually in outdoor or unenclosed areas) are the most common place for attempting quitters to lapse.26 27 Third, the only systematic study of retailers who have stopped selling tobacco found that alcohol-licenced premises are disproportionately likely to stop selling tobacco, with managers citing low profits and the ‘hassle’ of selling tobacco as major contributing factors.16 We therefore decided to survey alcohol-licenced premises to (1) identify the factors associated with the probability of selling and stopping selling tobacco, (2) determine if selling tobacco is the norm for alcohol-licenced premises, (3) provide insight into the reported importance of tobacco sales for different types of licenced premises and (4) identify retailers’ attitudes to stopping tobacco sales.

Method

A target quota of 500 alcohol-licenced outlets selling cigarettes was predetermined to ensure sufficient precision for the estimates of dichotomous proportions among sellers (approximately ±4.0%). Target interviewees were randomly chosen from a list of 5869 alcohol-licenced ‘registered clubs’ (‘clubs’), ‘hotels’ and ‘packaged liquor’ outlets obtained from the NSW Office of Liquor & Gaming in March 2016 (see table 1 for explanation of these categories). Other alcohol-licence categories such as restaurants, small bars and alcohol producers were excluded from the study, because they do not usually sell tobacco. Randomly selected premises were called by a market research company, and a request was made to speak to the licencee or general manager. After obtaining informed consent, participants were asked if they sold cigarettes (including in any associated bottle-shop for hotels, clubs, supermarkets and grocery stores), and if not, if they had ever sold cigarettes. (The survey used the term ‘cigarettes’ to avoid confusion with sales of loose tobacco, but we found no evidence, and do not anticipate, that any retailers sold any tobacco products if they did not sell cigarettes.) Ex-sellers (n=147) were asked when they had stopped, and the interview was terminated. Those still selling cigarettes (n=663) were asked to rate the importance of cigarette sales on a seven-point scale (from 1=‘not at all important’ to 7=‘very important’, presented in randomised order), with the scale 2

midpoint of 4 unlabelled. Other questions assessed the number of employees, the distance of the nearest cigarette retailer, selling by vending machine (no/yes). Respondents’ attitudes to cigarette sales and their predictions of clients’ responses if they did not sell cigarettes were probed with open-ended questions. Information about whether the outlet had an ‘extended hours’ licence and for clubs and hotels, the number of gaming machines, if any, was available from the list of outlets. The target quota of 500 outlets selling cigarettes was achieved after the selling status of 927 outlets had been determined. However, analysis of completed responses indicated that packaged liquor outlets without phone numbers listed on the database (39.1% of such outlets) had inadvertently been excluded from the list for randomised selection. In order to remove potential bias due to non-selection of such outlets, phone numbers were obtained for these outlets, and randomly selected packaged liquor outlets without phone numbers on the database were contacted until they comprised 39% of packaged liquor outlets in the sample (thus reflecting their representation among packaged liquor outlets listed on the database). Of 1230 outlets surveyed, 1062 (86.3%) agreed to answer the first screening question on whether they sold cigarettes, with the remainder either refusing (n=122, 10.1%) or unavailable (n=46, 3.8%) (figure 1). After removing 20 responses that were unusable due to communication difficulties, this resulted in a final sample of 1042 outlets where the selling status was established. Of these 1042 respondents, 954 (91.5%) answered the additional questions applicable to their selling status. Table 2 shows the sample composition, relative to the total number of licences of those types in NSW.

Respondent details

In order to understand any obvious variability within licence types, premises that reported that they sold cigarettes were asked to describe their business. All those with club licences described their business as a ‘club’, but outlets with hotel and packaged liquor licences used a variety of names to describe their business. Hotel licences (n=252) most commonly called themselves a ‘pub’ (72.6%, n=183), ‘hotel’ (7.1%, n=18) or ‘bar’ (12.3%, n=31), but also a ‘bottle-shop’ (7.1%, n=18), ‘club’ (0.4%, n=1) and ‘supermarket’ (0.4%, n=1). Packaged liquor licences Burton S, et al. Tob Control 2017;0:1–8. doi:10.1136/tobaccocontrol-2017-053944


Research paper

Figure 1 Sample details, relative to population of clubs, hotels and packaged liquor outlets. NSW, New South Wales. (n=192) most commonly described themselves as a ‘bottle-shop’ (68.2%, n=131), but also used various descriptions, such as a ‘supermarket’ (16.1%, n=31), ‘general store’ (9.4%, n=18), ‘convenience store’ (2.6%, n=5), ‘café’, ‘marina’, ‘tobacco retailer’ or ‘caravan park shop’ (each 0.5%, n=1).

Analysis

Logistic regression was used to examine associations between (1) selling cigarettes (vs not selling) and (2) stopping selling cigarettes (vs not stopping) and characteristics of the outlet (with OR as effect measure). Linear regression was used to examine associations between rated importance of cigarette sales (on a scale of 1–7) and outlet characteristics

Table 2 Alcohol-licenced retailers in New South Wales: survey sample versus population Sample* Licence type

Count

All licences in NSW Per cent

Count

Per cent

Club licence

309

29.7

1325

22.6

Hotel licence

434

41.7

2133

36.3

Packaged liquor

299

28.7

2411

41.1

All

1042

100

5869

100

*Includes sellers, ex-sellers and non-sellers.

Burton S, et al. Tob Control 2017;0:1–8. doi:10.1136/tobaccocontrol-2017-053944

(with differences in mean score as effect measure). Independent variables included outlet type, extended trading hours (no/yes), number of gaming machines and remoteness. The linear regression models also included number of staff, distance to the nearest alternative cigarette seller, and in exploratory analysis, selling via a vending machine (vs not selling by vending). After allowing for other variables, selling by vending (48.6% of the 599 sellers who answered all questions) was not significant nor a confounder of other effects in any model (with change-in-estimate criterion of 20%), 28 so it was omitted from the final models. Tests for linear trends were performed by substituting the categorical versions of the three relevant independent variables with an ordinal or continuous version of the variable, as appropriate. Ordinal versions were used for ‘number of staff ’ and ‘distance to the nearest cigarette retailer’, where respondents had chosen an answer from a range. The ordinal versions for those variables were coded with the left-hand cutpoints (the minimum value from the selected range of values) and with ‘Don’t know/can’t say’ responses coded as missing values. For ‘number of gaming machines’, the exact number of gaming machines was available from each outlet’s licence details, so that number was used to provide a continuous measure. Analyses were carried out using StataV.13. 3


Research paper Table 3 Associations between retailer characteristics and selling cigarettes (n=1042) OR for selling (vs not selling) Characteristic

Sellers n/N (%)

All retailers with known sell status

663/1042 (63.6)

Unadjusted OR (95% CI)

Adjusted OR* (95% CI)

Ref.

Ref.

Retailer type Hotel

283/434 (65.2)

Club

175/309 (56.6)

Bottle-shop – corporate owned chain† Bottle-shop – independent/franchisee owned‡ Supermarket/general store – corporate owned chain§ Supermarket/general store – independent/franchisee owned¶

0.70 (0.52 to 0.94)

0.37 (0.24 to 0.57)

86/90 (95.6)

11.47 (4.13 to 31.89)

62/118 (52.5)

0.59 (0.39 to 0.89)

1/30 (3.3)

0.02 (0.00 to 0.14)

0.07 (0.01 to 0.56)

54/59 (91.5)

5.76 (2.26 to 14.72)

21.48 (7.74 to 59.56)

Other**

2/2 (100.0)

P value‡‡

44.52 (14.81 to 133.86) 2.25 (1.25 to 4.03)

n/a††

n/a††