Article Text

A systematic review of store audit methods for assessing tobacco marketing and products at the point of sale
  1. Joseph G L Lee1,
  2. Lisa Henriksen2,
  3. Allison E Myers1,
  4. Amanda L Dauphinee2,
  5. Kurt M Ribisl1,3
  1. 1Department of Health Behavior, Gillings School of Global Public Health, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  2. 2Department of Medicine, Stanford Prevention Research Center, Stanford University School of Medicine, Palo Alto, California, USA
  3. 3Cancer Prevention and Control Program, Lineberger Comprehensive Cancer Center, School of Medicine, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  1. Correspondence to Dr Kurt M Ribisl, Department of Health Behavior, The University of North Carolina at Chapel Hill, CB 7440, Chapel Hill, NC 27599 USA; kurt_ribisl{at}unc.edu

Abstract

Objective Over four-fifths of reported expenditures for marketing tobacco products occur at the retail point of sale (POS). To date, no systematic review has synthesised the methods used for surveillance of POS marketing. This review sought to describe the audit objectives, methods and measures used to study retail tobacco environments.

Methods We systematically searched 11 academic databases for papers indexed on or before 14 March 2012, identifying 2906 papers. Two coders independently reviewed each abstract or full text to identify papers with the following criteria: (1) data collectors visited and assessed (2) retail environments using (3) a data collection instrument for (4) tobacco products or marketing. We excluded papers where limited measures of products and/or marketing were incidental. Two abstractors independently coded included papers for research aims, locale, methods, measures used and measurement properties. We calculated descriptive statistics regarding the use of four P's of marketing (product, price, placement, promotion) and for measures of study design, sampling strategy and sample size.

Results We identified 88 store audit studies. Most studies focus on enumerating the number of signs or other promotions. Several strengths, particularly in sampling, are noted, but substantial improvements are indicated in the reporting of reliability, validity and audit procedures.

Conclusions Audits of POS tobacco marketing have made important contributions to understanding industry behaviour, the uses of marketing and resulting health behaviours. Increased emphasis on standardisation and the use of theory are needed in the field. We propose key components of audit methodology that should be routinely reported.

  • Advertising and Promotion
  • Price
  • Surveillance and monitoring
  • Tobacco industry

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Over the last 30 years, tobacco companies have transitioned their marketing from mass media to the point of sale (POS),1 which provides a visible marketing environment for products that cut short the life of one person every 6  s around the globe.2 POS marketing includes ‘powerwalls’ of branded products, extensive price discounting, calculated placement of tobacco products and advertisements for maximum visibility. The ubiquity of tobacco marketing at the POS is such that tobacco advertisements are the only indication of a grocer's presence in some neighbourhoods.3 Industry documents show careful attention to improving profits by leveraging retail environments,4 with an emphasis on initiating or retaining smokers5 and exploiting vulnerable markets.6 POS tobacco marketing expenditures are significant. In the USA alone, the tobacco industry reported spending US$7.20 billion in 2010 on POS marketing, which constituted 85% of total reported marketing and is the equivalent of $13 694 per min.7 ,8 i POS expenditures accounted for 87% and 52% of 2010 marketing expenditures on cigarettes and smokeless tobacco, respectively.

The role of exposure to POS tobacco marketing in smoking initiation and uptake is well documented.9–11 For example, one longitudinal study observed higher odds of trying smoking among never smokers who reported at least weekly exposure to POS marketing at baseline.12 Other studies report adults’ exposure to POS marketing is associated with stymied quit attempts.9 ,13

Measurement of tobacco marketing at the POS is essential to understanding the POS as an environmental determinant of tobacco use. However, no previous research has systematically examined methods for assessment of tobacco products and marketing in retail environments. This is in stark contrast to environmental assessment of food products and marketing, for which multiple reviews have examined measurement, reliability and validity.14–19

Following previous work examining the properties and domains of retail food environment instruments,16 ,19 we conducted a systematic review of the peer-reviewed literature for protocols, approaches and measures to study POS tobacco marketing. Two questions guided this review: (1) What are key uses of POS tobacco marketing audits? and (2) What are the measures and methods employed in implementing such audits?

Methods

Search strategy

We developed keywords to identify articles about POS tobacco marketing until new keywords generated no relevant new articles.ii We searched 11 databases for peer-reviewed literature: Business Source Complete, Cumulative Index to Nursing and Allied Health Literature, Communication and Mass Media Complete, Education Resources Information Center, Health And Psychosocial Instruments, Public Administration Abstracts and PsycINFO via EBSCO; Embase; ISI Web of Knowledge; PubMed; and Public Affairs Information Service via ProQuest. We used no language, date or geographical limitations in the search, but limited it to peer-reviewed journals in the seven EBSCO-based sources. Exact search strings and settings for each database are available from the first author. The search was first conducted on 9 September 2011, and we updated and expanded our search with additional keywords on 14 March 2012. We included advance access publications; thus, some papers show publication dates after the search. We did not search the grey literature or review actual instruments, which were often unavailable. We did not collect data on the findings of identified papers, as the purpose of this review was to examine measures and methodologies.

Inclusion coding

Two coders (JGLL, AEM) independently coded each abstract or full-text, when necessary, for the following inclusion criteria: Studies reporting results where (1) data collectors visited and assessed (2) retail environments using (3) a data collection instrument for (4) tobacco products or marketing. Articles not meeting these criteria were coded either as background articles or as excluded. Background articles addressed published research related to POS tobacco marketing without the use of an audit instrument (eg, classroom surveys of youth exposure to convenience store tobacco marketing). We excluded articles not written in English, French or Spanish. We excluded studies that reported youth purchase attempts where environmental data characteristics (eg, sign posting minimum age of sale) were incidental to the attempted buy. We also excluded studies where product availability or advertising was coded simply as present or absent (eg, nutrition environment studies that solely documented the presence of tobacco products). Finally, we excluded the couple of content analysis studies that obtained data from coding store photographs because our focus was on studies that conducted real-time and in person store audits. The two coders reviewed articles with divergent coding and agreed upon a code. In the event that agreement could not be reached, a third author (KMR) decided.

Data abstraction

We trained a team of eight graduate student and staff research assistants and associates who independently reviewed the papers and entered information about the audit instrument, methodology and findings into a web-based abstraction form. One author reviewed and reconciled divergent information (JGLL) from each pair of reviews. When the reviewing author was also one of the data abstractors, a third author (KMR) provided additional review.

Measures

The aims of each paper were coded under more than one category to fully capture study objectives. First, we coded for papers describing the environment, under which we also coded for (a) assessing compliance with regulations, (b) investigating disparities in exposure to POS marketing or attendant health risks and (c) looking at the relationship between POS marketing and underage sales of tobacco products. Second, we coded for quasi-experimental tests of the relationship between POS marketing and health-related behaviours, under which we also coded for (a) youth behaviour and (b) adult behaviour. Third, we coded for interventions to reduce POS tobacco marketing. We noted if a theory or framework was explicitly mentioned in guiding the research question and/or measurement. We coded multiple aspects of the audit methodology including sampling, training of auditors, data collection modality and study design (cross-sectional or longitudinal). To characterise the types of measures, we used the four P's marketing mix,20 which has previously been adapted for tobacco marketing.21 Specifically, we coded information about product (availability of cigarettes, smokeless tobacco, cigars, pipes/pipe tobacco, emerging products, e-cigarettes and aggregated/undefined tobacco products), price (including multipack or other discounts), promotion (signs, displays, shelving, ads inside, ads outside, ads by brand, ad size, functional items, gifts with purchase) and placement (product or marketing materials near youth products or at a child's height/line of vision). For promotion, we provided additional detail on measurement strategies relating to the link of POS exposure to behaviours. Each paper was also coded for its year of data collection and study setting. A final question specifically asked for any measures or other information that did not fit into the data abstraction form.

Results

The systematic search identified 2906 de-duplicated articles, including 9 through citations (figure 1). Of these, 88 met inclusion criteria for the systematic review, and 106 articles were coded as background articles.

Figure 1

Inclusion of studies, 14 March 2012.

The first POS audit was a study of San Francisco, California, USA, pharmacies published in 1978 by Schroeder and Showtack.22 No other audits were published until 1989, when Braverman et al published a study of smokeless tobacco marketing in California.23 Over the past two decades, annual publications increased from approximately 1 to 10 per year (figure 2). The first audit outside of the USA was an examination of compliance with a UK voluntary advertising agreement published in 1993.24 Overall, 64% of audits were conducted in the USA, followed by Canada (n=8), the UK (n=7), Australia (n=5), India (n=3), Guatemala (n=2) and Mexico (n=2). Audits have also been conducted in Argentina (reported with Guatemala data), China, Greece, Ireland, New Zealand and Thailand. The leading publisher of POS tobacco marketing audit studies is Tobacco Control (n=23), followed by Nicotine & Tobacco Research (n=8), American Journal of Public Health (n=6), Preventive Medicine (n=4), Australian & New Zealand Journal of Public Health (n=3), Canadian Journal of Public Health (n=3), Health & Place (n=3) and Journal of Health Communication (n=3).

Figure 2

US and non-US publications by year, 14 March 2012, N=88. *Note: data for 2012 include only through March 14.

Research questions

The most common objectives for store audits were to describe the content and prevalence of POS tobacco marketing (75%), to investigate disparities (26%) and to assess compliance with regulations (28%). Sixteen papers examined compliance with national regulations or voluntary marketing agreements, seven with provincial regulations and two with local regulations. Fewer studies correlated audit data with smoking behaviour among youth or adults (13%), or with illegal sales to minors (6%). Only four studies evaluated interventions to reduce tobacco marketing at the POS.

The few studies (10%) that explicitly mentioned an underlying theoretical framework referenced the Bernard framework of neighbourhood health disparities,25 community organising, diffusion of innovation, ecosocial theory, the social ecological framework, structural violence and youth empowerment. Apart from work by Shareck et al 26 using the Bernard framework,25 the literature was largely atheoretical.

Research methods

Among the 88 studies, 64 (73%) used a single cross-sectional design,12 ,22 ,23 ,27–87 four (5%) used a repeated cross-sectional design88–91 and 20 (23%) used a panel design.24 ,26 ,67 ,92–108 Sixty-eight studies (77%) reported a sampling frame, most commonly using licensing, tax or inspection lists (32%) or ground truthing (26%) (ie, canvassing an area to identify each retail location) to develop a list of stores. A few studies used business lists (9%) or phone books (7%). Seventy-eight studies (89%) reported how stores were selected, and among these random sampling (45%) and census (41%) were more common than convenience sampling (8%). Fourteen of these studies (18%) reported stratified sampling, generally using area demographics or store type. Nineteen of these studies (24%) sampled based on proximity to schools. One study sampled 60-meter street sections instead of stores as the population element.26 Among the 65 studies reporting the sample size, excluding one study with a sample size of 17 476,106 the average sample size was 569 (SD=858, median=206).

Audit protocol

Store audits require careful and systematic observation of the retail environment and sometimes demand interaction with store clerks. Surprisingly, fewer than half (44%) of studies mentioned any characteristics of their data collectors and fewer (19%) specified any details about how they were trained. Of these, 47% reported the length of training provided, and 65% reported field training by visiting actual stores. Forty-one per cent of all studies reported how data collectors approached stores. Of these, 42% covertly collected data, 8% explained their presence only as needed and 39% notified store staff as part of protocol. One intervention study reported using passive consent through mailings to store owners.108 Nine per cent of papers reported the mode of the audit instrument, and all reported using paper. Among the 13% of papers that reported time spent per store, the estimates ranged from 5–30 min. One 2006 study estimated data collection costs at approximately US$60 per store within a single community.39

Measuring the ‘marketing mix’

Twenty-seven per cent of papers used or adapted existing tobacco audit instruments (table 1). The most frequently cited audit instrument or protocol was from the ImpacTeen research group followed by Operation Storefront. Authors made only one comment on particular audit instruments: a report of challenges in using the Store Alert form, noting challenges in calculating the prevalence of particular items.75

Table 1

Use of existing tobacco audit instruments, 14 March 2012, n=24

Twice as many papers examined promotion (86%) as price (43%), placement (42%) or products (41%).

Product

Cigarettes were the most commonly examined product (table 2). Little research examined any form of smokeless tobacco, cigars, cigarillos or other tobacco products. For example, the only examination of e-cigarette availability we identified was conducted in Guatemalan pharmacies.81 In an effort to document the availability and variety of culturally-specific tobacco products, Longman et al audited stores in South Asian immigrant communities in England, identifying 98 different smokeless tobacco products,59 and Schensul et al identified 68 smokeless tobacco products in low-income areas of Mumbai.72

Table 2

Measures of product availability, 14 March 2012, n=36

Price

Although 43% of studies examined the price of tobacco products, it was surprisingly difficult to confirm whether price was determined from actual purchase (eg, store receipt), observed from advertisement, or verbally requested from the merchant. Sixty-three per cent of these studies reported measures of price collected by purchase or observation. Sixty-three per cent of these also recorded the presence of any type of price discounts, and 39% noted the presence of multipack discounts. Two suggestions emerged for future research: One study noted that the collection of product weight would have allowed for better comparison of prices for non-standard smokeless tobacco products.59 Another study noted that data from single-pack purchases could not address the impact of multipack discounts, which have been a popular marketing tool in the USA.79

Promotion

Eighty-six per cent of papers collected and/or reported on promotional methods for communicating with customers such as informational signs; images indicating virility, independence or other characteristics; cues to purchase; gifts with purchase; etc. (table 3). Unique measures included counts of store-made tobacco signs,55 exclusively Marlboro-branded stores,80 the ethnicity of models in tobacco advertisements,103 the presence of youthful models in tobacco advertisements,83 and the per cent of product fronts visibly displayed.37 Two papers used a four-point visual impact scale developed by Wakefield et al that ranged from ‘no advertising’ to ‘advertising covering most of the store/storefront’85 or ‘in your face’ advertising.49

Table 3

Presence, count or scaled measure of promotion, 14 March 2012, n=76

Placement

Thirty-seven papers measured the placement of tobacco products or marketing. However, the most frequently examined placement issue was self-service sales of tobacco products, identified in 49% of these papers, that is, tobacco products that can be handled prior to purchase by the consumer. Forty-one per cent of placement papers measured proximity to youth products and 38% measured advertisements at a child's height or line of vision. Child's height was universally defined as ≤91 cm (3 ft.); however, measures of what constitutes proximity and what constitutes a youth product showed greater variability. Proximity measures used range from ‘adjacent to’ (the first audit to use this term defines it as within 61 cm (2 ft.)),23 to 15.25 cm (6 inches),40 ,111 to 30.5 cm (1 ft.),34 ,86 to 50 cm (20 inches),29 and finally to 1 m (3.3 ft.).64 Others leave ‘adjacent to’ undefined. Youth products have been defined to include candy, toys, ice cream and sometimes snack foods.

Operationalisation of the four P's in correlational analysis

To better express the diversity of measures used for relating POS tobacco marketing and products to health behaviours, we describe measures used among studies correlating measures of tobacco use behaviours and POS marketing (n=7). Feighery et al39 and Henriksen et al12 summed counts of advertisements and estimated exposure per week based on reported visits to stores. In a separate study, Henriksen et al48 used the total number of advertisements observed in the school neighbourhood. Lovato et al60 used three separate variables: the proportion of stores with any promotions, any advertisements and any powerwalls. Pabayo et al62 similarly used the proportion of stores with a higher ‘smoking tolerance’ score based on the presence of three indicators: sales, marketing and allowed smoking. Slater et al106 measured whether or not five types of marketing were present in stores (ie, parking lot, exterior, interior, functional object or low-height advertisements) and used the proportion of stores with each of these measures to create a summary score for each research site (Cronbach's α=0.70). Wakefield et al,84 in a study on brand preference, used five key measures: presence of a gift with purchase, proportion of exterior brand-share, proportion of interior brand-share, total number of exterior ads and total number of interior ads.

Reliability and validity

Surprisingly few studies (11%) reported the reliability of audit measures. Although reliability of measures was generally acceptable to high, some items involving physical measurement (eg, products within a certain distance from candy) show poor or reduced reliability (see online appendix).

Few studies (6%) explicitly reported any type of validity assessment. Cohen et al reported that two experts provided feedback on the audit instrument and previous audit instruments were consulted for additional relevant measures, signifying face and content validity.34 Feighery et al compared audit measures with school survey measures (eg, self-reported exposure to marketing), and found both yielded similar findings, signifying convergent validity.39 Freedman et al, using self-reported surveys and store audits, found that ‘participants’ perceptions regarding the availability of tobacco and alcohol products do appear to differ from the reality of their local food environments’,43(p. 834) signifying a failed test of concurrent validity relevant to survey research assessing the perceived presence of tobacco products. Quedley et al used compliance officer reviews to assess the face and content validity of their audit instrument.64 Lastly, Toomey et al  noted that phone surveys of cigarette prices show poor agreement with audit measures and suggested that audit measures are likely more precise.79

Discussion

We sought to describe the objectives and methodology for audits of POS tobacco marketing in order to synthesise a growing literature and to make recommendations to improve future research. Strengths of the existing literature are numerous. Longitudinal designs have been used to track trends in marketing and to correlate change in exposure with change in behaviour. Random sampling or census approaches are frequently used, thereby improving generalisability. The reliability of key measures is generally good, albeit with some concerns about measures requiring spatial estimation or physical measurement. However, too many studies leave critical details of measurement and methodology unreported.

In addition, we noted the atheoretical nature of most audit-based work, a challenge that has also faced the food environment literature17 and neighbourhood environment more generally.113 Researchers frequently use the Social Ecological Framework114 to structure thinking about multiple levels of social influence on human development and health.115 More recent work takes a systems-based approach that examines the relationships between individuals’ behaviour and how those behaviours in turn reproduce and change social and environmental systems.116 Examining the influence of POS tobacco marketing on behaviour can provide feedback for refining existing theories and frameworks on neighbourhood health; indeed, such applied work testing theories is critical to scientific advancement.117 Moreover, the measurement of retail tobacco marketing can itself be informed by work proposing domains of influence on health behaviours across physical, social, economic, regulatory and community influence.25 That is, theory can help inform the selection of critical constructs of exposure to marketing at the POS that are likely linked to resulting behaviour. There has been only limited work to integrate theory into the measurement of the retail environment's influence on tobacco use behaviour,26 ,45 primarily using a framework of neighbourhood health disparities developed by Bernard et al.25

The Bernard framework suggests the importance of institutional (eg, government) and community intervention for health promotion. Much of the research identified in this review is designed to be relevant for policymakers, regulators and/or community members rather than to directly examine health behaviours. The role of theory in this research is also important, but the emphasis shifts away from testing hypotheses of behavioural outcomes to understanding the role of research in promoting healthful environments. Multiple theories and frameworks for policy change posit a role for research findings in policy change.118 The Advocacy Coalition Framework, for example, is explicitly designed to account for the use of technical information by policymakers, advocates and stakeholders in policy development.119 The content, format and products of research findings matter in diffusion theories.120 ,121 For measurement, this calls our attention to the importance of carefully linking POS tobacco marketing measures to current and future policy developments, to questions of interest to regulators and to provide actionable information to community members. Strategic use of measurement can provide information of use to city planners,122 note levels of marketing near schools, identify tobacco industry evasions of regulations and monitor compliance with regulations. Policy-making theories provide a guide to how different measures and research products regarding POS tobacco marketing are more or less relevant for different stakeholders.

Strengths and weaknesses of this review

This study used a systematic approach with independent inclusion coding and data abstraction. However, the search may have missed some peer-reviewed papers, and it did not incorporate any grey literature, such as reports of retail marketing surveillance sponsored by state health departments or other government agencies.123 ,124 Such reports may contain additional measures, more detailed information about reliability and descriptive results about changes in the quantity and nature of retail tobacco marketing over time. Such information would be particularly informative if the instruments were readily available to interested researchers.

Our coding of included articles did not address information on the availability of flavoured tobacco products, which was an oversight given their importance to the tobacco industry, nor did the coding address the availability of single cigarettes, which are of interest because they reduce the unit price of tobacco products. Future work should synthesise the results of the studies we identified, which was beyond the scope of our aims.

Implications for measurement and reporting

Our findings suggest potential improvements in the reporting of audit-based papers (box 1), and we propose that editors and authors strongly consider presenting key information about the audit protocol that was frequently not reported in the papers we reviewed.

Box 1

Recommendations for reporting

Key information to report in audit-based papers

  1. The sampling frame

  2. The covert or disclosed nature of data collection, as this can substantially change what data can be collected around counters and clerks

  3. Who is collecting the data, eg, students, community members or professional auditors

  4. The reliability of measures, particularly novel survey items and items involving spatial estimation or physical measurement

  5. If data collectors are blinded to tests of reliability or are aware they are participating in a reliability test

  6. The wording and/or definition of items used as key variables in analyses, as elegantly done by Cohen et al, Feighery et al and Lovato et al;34 ,40 ,60 these can be operationalised in multiple ways (eg, is snus measured as an emerging product or a smokeless product?)

  7. The mode of the survey, as iPads, iPhones or other electronic data collection devices can influence interactions with retailers and may have instrumentation effects

In comparison with the field of nutrition environment measurement, where researchers have identified problematic measurement and reporting issues, less attention appears to have been paid to assessments of the retail tobacco environment. Tobacco marketing instruments may be lagging behind the science of nutrition environment measurement, as Ohri-Vachaspati and Leviton found 39% of 48 nutrition environment instruments had been assessed for reliability and validity (albeit including weaker face and content validity).16 In a different assessment of papers measuring the nutrition environment, however, the reporting of measurement quality is much lower as noted by McKinnon et al who found that only 13% of 137 nutrition environment papers reported measures of reliability and/or validity of any type.15 Concurrent development of audit methodologies raises the question of future efforts exploring most practical ways of linking food, alcohol, physical activity and tobacco audits.

Future work might examine lessons from store audits for other health-relevant products and consider the merging of key constructs into modular standardised instruments. We note that substantial work on the consumer nutrition environment has coalesced around the Nutrition Environment Measures Survey,19 providing standardised protocols that are used widely, albeit not universally. Such standardisation allows for easier comparisons of findings between studies and streamlines measurement. The development of these instrument modules will require ongoing research to determine the most salient measures and their validity. Some guidance is available from the field of nutrition environment assessment: Lytle has published an excellent discussion of validity and reliability in environmental measures and notes additional areas of measurement concern including variance and utility for generalising across populations and health behaviours.14

Unanswered questions and future research

In spite of its importance to tobacco control, the unit price of tobacco products has not received as much attention as the availability of promotions and other advertising. Price is relevant to the study of target marketing patterns,50 to assess compliance within minimum price laws42 and to indicate the presence of illicitly traded tobacco products. Research is needed to examine the concordance of price measured from multiple sources, including advertised price, reports of store clerks (by phone or in person), purchase receipts and scanner data, especially given the dramatic difference in cost and effort that these sources represent.

The growing presence of antismoking cues at the POS also suggests a new focus for retail audits. Only one of the studies we examined reported on the visibility of graphic warnings on products, advertisements or elsewhere at POS.53 Future efforts may need to consider how product shelving or displays may be used to obscure warning labels from a customer's view.

In addition, the utility of audit methods to evaluate compliance with regulations on advertising, labelling and product sales may require additional research: What, for example, is the best way to measure the availability of single cigarettes for sale? Various measurement approaches evident in the literature: advertised or visible sales of single cigarettes,44 attempted purchase by youth and adults,56 observations of youth,51 inspection of the counter area107 and clerk interviews.107 In a 1990–1991 study, Wildey et al  noted that observed availability of single cigarettes among a sample of stores was much lower (24.8%) than clerk-reported availability (40.6%).107 Given heightened penalties for single cigarette sales in some countries, audits that do not evaluate inaccessible areas behind the counter could underestimate the availability of single cigarettes for sale.

More consistent measurement and repeated surveillance of retail tobacco marketing could provide valuable information for policymakers and regulators.1 ,125 However, as audits are conducted more frequently, retailers may be more reluctant to participate, yielding higher refusal rates. This will require new innovations in observing the retail tobacco environment. Although we excluded from this review two studies that coded photographs rather than conducting real-time audits,126 ,127 we recognise that photographs may provide a valuable data source for more reliable measurement. Unfortunately, retailers are reluctant to permit photographs (eg, a UK study obtained permission from only 56% of tobacco retailers, none of which were chain or franchises with a corporate headquarters).127 Satellite and digitised storefronts collected for other purposes (eg, Google StreetView) offer opportunities to develop new measures for studying exterior marketing materials and raise the option of outsourced image coding. Lastly, there may be opportunities for combining work on the role of tobacco retailer density and proximity128 with audit-based information on tobacco marketing.

Conclusion

The POS tobacco marketing audits that we examined highlight long overdue—and increasingly global—attention to the regulation of tobacco products and their POS marketing.129 The global tobacco epidemic has vectors that transmit the use of addictive products, a transnational tobacco industry and its lawyers,130 ,131 and this paper shows the growing emphasis on measurement of an important component of the epidemic, the retail environment. Marketing at the POS is used by the tobacco industry in carefully calibrated ways, and addressing the tobacco epidemic will require continued and improved surveillance of marketing strategies to promote this addictive agent.

Key messages

  • Audit-based measurement of retail tobacco marketing is increasingly used to monitor tobacco industry strategies and their influence on tobacco use by multiple population subgroups.

  • The reliability of audit measures is generally good, albeit with some concerns when measuring spatial distances or heights.

  • Given a variety of unstandardised measures reported in the literature, we make recommendations to improve the conduct and reporting of retail marketing audits.

Acknowledgments

Thanks to our amazing data abstraction team: Amber Anderson, Catherine Jo, Yuna Kim, Ashley Leighton and Sarah Moreland-Russell. Thanks also to the UNC Health Sciences Library's consultation services on database searching.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors JGLL conducted the search, managed the review process, conducted all analyses and drafted the manuscript. LH and KMR provided expert guidance on the topic area. JGLL and AEM coded all abstracts. JGLL, AEM and ALD abstracted data. All authors developed the data abstraction form, provided critical feedback on drafts of the manuscript and approved the final manuscript. KMR originated the study and guided the research process.

  • Funding This research was partially funded by the National Institutes of Health/National Cancer Institute (U01 CA154281). This paper was partially supported by doctoral training support from the University of North Carolina Lineberger Comprehensive Cancer Center's University Cancer Research Fund to JGLL.

  • Disclaimer Opinions are those of the authors and do not represent those of the funders or institutions.

  • Competing interests KMR is the executive director and AEM is the deputy director of Counter Tools (http://countertools.org), a non-profit that distributes store mapping and store audit tools from which they receive compensation. JGLL, KMR, and AEM also have a pending royalty interest in a store audit and mapping system owned by the University of North Carolina at Chapel Hill. The tools and audit mapping system were not used in this study.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • i The most recent year for which data are available. This underestimates total POS marketing expenditure by excluding outdoor expenditures, some of which are for retail locations. Based upon 2010 US Federal Trade Commission lines we calculated for (1) point-of-sale, (2) price discounts, (3) promotional allowances—retailers, (4) retail-value-added—bonus tobacco products, (5) retail-value-added –non-tobacco bonus.

  • ii Example of final PubMed search string: (store[tw] OR stores[tw] OR “point of sale”[tw] OR “points of sale”[tw] OR “point-of-sale”[tw] OR “points-of-sale”[tw] OR retail[tw] OR retailers[tw] OR retailer[tw] OR shop[tw] OR “gas station”[tw] OR “gas stations”[tw] OR “point of purchase”[tw] OR “points of purchase”[tw] OR “points-of-purchase”[tw] OR “point-of-purchase”[tw] OR outlet[tw] OR outlets[tw] OR milkbar*[tw] OR petrol[tw] OR garage*[tw] OR pharmacy[tw] OR pharmacies[tw] OR druggist*[tw] OR supermarket*[tw]) AND (“tobacco”[MeSH Terms] OR “tobacco”[All Fields] OR “smoking”[MeSH Terms] OR “smoking”[All Fields] OR cigarette[All Fields] OR cigarettes[All Fields] OR cigar[All Fields] OR cigars[All Fields] OR cigarillo[All Fields] OR cigarillos[All Fields]).