Article Text
PDF
Abstract
Background Associations between retail tobacco availability and tobacco use have been mixed. This study examined associations between school-based retail environment exposures and current use of cigarettes, cigar products and e-cigarettes among middle school youth in Cleveland, OH.
Methods Retailers selling tobacco products were identified using the 2015 Cleveland Food Retail Database (n=639 stores). Youth survey data were drawn from the 2016 Cleveland Youth Risk Behavior Survey, administered to all 7th/8th graders across the Cleveland Metropolitan School District (n=3778, response rate=83.0%). Past 30-day cigarette, cigar product and e-cigarette use were assessed. Student demographics, number of days walking to/from school each week and number of times youth stopped at a retailer to/from school each week were included. For each school (n=63), tobacco retail density (TRD) and proximity (TRP) to nearest retailer were calculated for each product. Multiple regression analysis assessed associations between retail exposures and youth tobacco use.
Results Across all schools, 3.9%, 10.2% and 8.6% of students currently use cigarettes, cigar products and e-cigarettes, respectively, and 15.2% currently use at least one tobacco product. TRD and TRP were not associated with current use; frequency of walking to school and stopping at retailers were strongly associated with current use.
Conclusions Although TRD and TRP were not significantly associated with tobacco product use, youth who reported regularly walking to/from school or who reported stopping at a retail store before/after school were significantly more likely to be a current tobacco product user. This may be due to increased exposure to exterior and point-of-sale marketing.
- non-cigarette tobacco products
- priority/special populations
- prevention
- public policy
Data availability statement
Data are available upon reasonable request. All of the individual participant survey data is available upon request via data requests or research proposals submitted through the Cleveland Metropolitan School District Office of Research and Evaluation. Individual retailer data from the Cleveland Food Retail Database as well as the data collection tools and protocols can be made available upon request. Data requests and research proposals should be directed to the lead author, erika.trapl@case.edu.
Statistics from Altmetric.com
Introduction
The US tobacco burden starts with youth1; 1 out of every 5 (19.5%) high school students currently use a tobacco product.2 If current rates persist, 5.6 million of today’s youth will eventually die from a tobacco-related illness.3 Since the 1998 Master Settlement Agreement (MSA), tobacco companies depend more on the retail environment to market their products.4 5 Currently, an estimated 375 000 retailers sell tobacco in the USA, nearly 7 for every 1000 school-aged (ages 5–17) youth.6
Youth are repeatedly exposed to cigarette marketing in the retail environment.7–9 Despite efforts to limit youth exposure to tobacco advertising in the MSA, tobacco company marketing expenditures actually increased and there was an increase in exterior ads following the MSA.10–13 Henriksen et al found that stores where youth frequently shopped displayed more cigarette advertisement than other stores in the same community that youth did not shop8; other research has found similar results.14–16 Further, research has shown that tobacco companies have targeted lower-income and minority communities with increased exterior retail cigarette advertising.9 15 17–19 Taken together, low-income and minority youth are likely to experience significant exposure to tobacco advertising and may therefore be more likely to initiate or use tobacco products compared with wealthier or white peers.
Adolescent cigarette smoking rates are associated with greater density of retailers in general.7 20–24 The tobacco retail density (TRD) within a youth’s environment greatly influences their overall availability to tobacco products and exposure to tobacco marketing. However, studies examining whether youth who live or attend schools in areas with higher TRD are more likely to smoke have resulted in mixed findings.7 25–30
A growing body of research indicates that frequent exposure to tobacco advertising is linked to adolescent smoking, and it is thought that a greater density of tobacco retailers could contribute to greater exposures to tobacco retail advertisements.31–35 In addition, cigarette advertisements have been found to be associated with adolescent perceptions that smoking is prevalent among peers36 37 and that cigarettes are easy to obtain,36 37 as well as intentions to smoke cigarettes in the future.38
Despite this significant body of work linking cigarette advertisements and youth cigarette use and attitudes, there are few comparable bodies of work to inform adolescent cigar product use or electronic cigarette (e-cigarette) use.29 39 Trapl and colleagues previously found the adolescent self-report of stopping at a retail shop, such as a corner store, before or after school was associated with an increased odds of cigarillo use among middle school students; however, this study did not account for the objective retail environment.40 Giovenco et al examined the association between retail density and e-cigarette use among high school youth, finding that density around schools was positively associated with ever and current e-cigarette use.29
By pairing comprehensive, ground-truthed retail data with school-level adolescent surveillance data, this study examined associations between school-based retail environment exposures and current use of cigarettes, cigar products, and e-cigarettes among 7th and 8th grade youth in Cleveland, OH.
Methods
Retail data
Retail data were drawn from the Cleveland Food Retail Database established by the authors; methodology can be found elsewhere.41 Briefly, the database was established in summer 2012 by using a systematic ground-truthing approach in each neighbourhood, documenting the location of all food retail locations, and completing a brief store audit in each to ascertain types of food sold as well as presence of tobacco products.
A team of 10 research assistants were trained in the data collection tool and audit protocol. The 12 hours training consisted of an extensive review of the data collection form and protocols, followed by three in-store practice assessments. Research assistants completed the training assessments, and answers were compared immediately following the assessment to check for agreement; high agreement (>90%) was reached after three stores.
Data were collected at each store from June through August 2015, between 10:00 and 14:00. Research assistants conducted audits in pairs to increase data quality and to ensure safety. On entering a store, data collectors approached the store manager to explain the procedures of the data collection and obtain permission to assess the store. Data collectors documented the presence of cigars, cigarillos or little cigars (CCLCs), cigarettes and e-cigarettes available in each retail location. A total of 1828 retailers were audited; an additional five vape shops were identified using validated online search techniques.42 This included 978 restaurants and fast food establishments. Data were not collected from stores that were permanently closed (n=189), or stores where the manager did not provide permission to enter the store (n=2).
Youth survey data
Youth data for this study were drawn from the 2016 Cleveland Youth Risk Behavior Survey (CLE-YRBS), which follows procedures similar to the national YRBS conducted by the Centers for Disease Control and Prevention but is tailored to fit local needs.43 The CLE-YRBS is conducted as a census administration; that is, the survey is administered to all 7th and 8th grade students within the Cleveland Metropolitan School District (CMSD) who are in school on the day of survey administration. No weighting procedures were required because the survey was administered to the full population.
Of the 63 CMSD K-8 schools, 100% agreed to participate in the CLE-YRBS in the Spring of 2016. A total of 4553 students were eligible to complete the survey; 3778 completed the survey (response rate=83.0%). Student non-participation was due to student refusal, absence on the day of survey administration or parental refusal.
Survey measures
Student self-reported grade level (ie, 7th or 8th), gender (ie, men or women), and race/ethnicity were assessed as individual-level demographic variables. Students were asked two questions to determine race/ethnicity. The first asked whether a student was Hispanic or Latino, and the second asked ‘What is your race?’; students were instructed to select one or more responses to the race question. Response options were: American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, and White. Students who reported that they were Hispanic or Latino were coded as such regardless of their response to the second question. Students who reported that they were not Hispanic or Latino were separated into one of three categories: non-Hispanic White, non-Hispanic Black and other/multiple races.
The Family Affluence Scale (FAS) was used as a proxy for family socioeconomic status (SES).44 45 FAS is calculated based on responses from four questions assessing family car ownership, family computer ownership, having a bedroom for oneself and number of family vacations in a year. The scale ranges from 0 to 9 and is categorised as low (0–4), medium (5–6) and high (7–9).
Walking to or from school was assessed with a single item asking students to report how many days they walk or ride a bike to or from school on an average week. Responses were organised as 0 days, 1–2 days, 3–4 days or 5 days.
Retail exposure was assessed through adolescent report of recent visits to corner stores or other retail locations and is similar to items used in other studies.46 47 These visits were assessed with one item: ‘In an average school week, how often do you stop at a corner store, convenience store, drug store, grocery store or other store that sells food on your way to or from school?’. Responses included never, one or two times a week, 3–4 times per week and every day.
Current cigarette use was assessed by asking, ‘During the past 30 days, on how many days did you smoke cigarettes?’. Responses were dichotomised as none (0) versus 1 day or more in the past 30 days.1 48
CCLC product use was determined by asking, ‘During the past 30 days, on how many days did you smoke cigars, cigarillos, little cigars or flavoured cigars such as Black & Mild’s, Swisher Sweets or Phillies?’. A survey item that included cigar brand names has been shown to yield greater endorsement among minority youth.49–52 Responses were dichotomised as none (0) versus 1 day or more in the past 30 days (1).
Current e-cigarette use was assessed using a single item asking, ‘During the past 30 days, on how many days did you smoke an electronic vapour product?’.48 A brief description of electronic vapour products is provided prior to the question. Responses were dichotomised as none (0) versus 1 day or more in the past 30 days (1).
Any current tobacco use was based on reporting current use of cigarettes, CCLC or e-cigarettes and was dichotomised as no tobacco (0) versus any tobacco (1).
Age at first tobacco use was assessed using a single item asking, ‘How old were you when you used your first tobacco product?’. Responses included 8 or younger, 9-10, 11-12, or 13 or older.
Tobacco retail measures
The relative and proximal availability of cigarettes, CCLCs and e-cigarettes within the retail environment surrounding each of the 63 CMSD K-8 schools that participated in the CLE-YRBS was assessed. Each school was geocoded to the rooftop centroid. To assess the availability for each of these tobacco products, three (n=3) product-specific retail outlet datasets were created based on whether the store sold: (1) cigarettes (n=639), (2) CCLC (n=525) and (3) e-cigarettes (n=383).
Relative availability was assessed for each tobacco product by generating a TRD surface from each product-specific retail outlet dataset using static-bandwidth kernel density estimation (KDE). KDE is a non-parametric method that extrapolates point location data over a study area by calculating the density of the point locations using kernel functions.53 Here, a kernel with a specified bandwidth (ie, a circle of a given radius centred at the focal location) was moved across Cleveland, and the density of retail outlets selling each specific tobacco product within the kernel was computed. At the point where density was being estimated (ie, focal location), retail outlets selling each specific product within the kernel were weighted according to their distance from the centre of the kernel. This resulted in a smooth continuous TRD surface where every location in Cleveland had an assigned density value. Gaussian kernels with a fixed quarter-mile bandwidth were used to generate each product-specific TRD surface across Cleveland, from which product-specific density estimates were extracted from the underlying density surface based on the rooftop location of each school with a resolution of 0.16 miles (250 m). We calculated TRD using a 6.5-mile bandwidth per the 2013 National Household Transportation Survey, the Default Silverman’s Rule of Thumb, a half-mile and a quarter mile bandwidth. We chose a quarter-mile search radius since it has been used in numerous school-based studies7 29 54 TRD values (retailers per square mile) were extracted from the underlying final product-specific density surface in ArcGIS for every public K-8 school.
Proximal availability was assessed by linking the location of every retail tobacco outlet and public K-8 school to a national street network dataset, allowing us to calculate network routes extending from each setting to the nearest tobacco retailer. Because network routes were based on the existing street network, they identified the walkable access to the nearest tobacco retail outlet. To properly measure walking distance to each product-specific tobacco retailer, all street network restrictions were removed (ie, one-way streets and non-routable roads). For each facility, proximity to the closest product-specific tobacco retailer (ie. tobacco retailer proximity, TRP) was calculated in miles in ArcGIS.55
Analysis
Univariate analyses were conducted to describe the characteristics of the sample and explore exposure variables and behavioural outcomes. Bivariate analyses, including χ2 tests, were used to test for bivariate associations between tobacco product use and demographics, walking to or from school and corner store visits. Variables found to be significantly associated with tobacco product use along with outcome-specific TRD and TRP were included in the final models. A generalised linear model, or GLIMMIX procedure, was used to run a multilevel model, using school-level estimates of TRD and TRP and individual-level estimates of youth behaviour including all specified variables yielding adjusted estimates. Data were analysed using SAS V9.4.56
Results
Descriptive statistics of the study sample are provided in table 1. Overall, 52.3% of the sample self-identified as female, with roughly an equal split between 7th and 8th grade. Over 54% of the sample identified as non-Hispanic Blacks or African American, 11.5% identified as non-Hispanic White, 22.1% identified as Hispanic and 11.7% identified as another, non-Hispanic other or multiple races. Nearly half of students (47.3%) reported walking to/from school at least ne time per week while 26.8% of students reported walking to/from school every day. Over 70% of all students reported stopping at a retail store before or after school at least one time a week.
Descriptive characteristics and prevalence of current tobacco product use among youth
Just under 4% of students reported current cigarette use, 10.2% reported current CCLC use and 8.6% reported current e-cigarette use; 15.2% of students reported using at least one tobacco and 1.6% reported using all three types of tobacco in the previous 30 days. Prevalence of current smoking was similar across gender except for CCLC, which was significantly higher among females. Racial and ethnic differences in prevalence of tobacco use existed for cigarettes and e-cigarettes, with non-Hispanic Black students reporting lower rates. Prevalence of all smoking was higher among those who regularly walked to or from school and those who stopped at a retail store before or after school.
Students attended one of 63 participating schools. Mean number of participants per school was 58.3; 100% of students were eligible to participate in the free and reduced lunch programme. Across the 63 schools, TRD was highly variable (table 2). There were 11 schools (17.5%) that had no tobacco retailers available within ½-mile. Mean TRD was 9.70 stores per square mile (SD=11.88) and ranged from 0 to 58.78; TRD for ‘any tobacco product’ was identical to cigarette TRD. That is, among the 639 stores selling ‘any tobacco product’, all 639 sold cigarettes; 60% (n=383) of stores selling any tobacco products sold e-cigarettes, and 82% (n=525) sold CCLC. Mean CCLC TRD was slightly lower than cigarette TRD at 8.52 (SD=10.31); mean e-cigarette TRD was lowest at 7.09 stores per square mile (SD=10.01). TRD ranges for CCLC and e-cigarettes were similar. Mean TRP was similar across all products.
Descriptive characteristics of student population and tobacco retailer near schools (n=63)
Variables found to be significantly associated with tobacco use at the individual-level were included in a multilevel model with TRD and TRP for the corresponding tobacco product; results are shown in table 3. Overall, TRD and retail TRP were not associated with current cigarette, CCLC, e-cigarette or any tobacco use. For current cigarette use, frequency of walking to or from school and the frequency of stopping at a retail store before or after school exhibited the strongest significant association. Patterns of association were similar for current CCLC use, e-cigarette use and any tobacco use. Results did not vary when models were run with TRD calculated at larger bandwidths.
Predictors of current cigarette, cigar, cigarillo, little cigar (CCLC) and e-cigarette use among young adolescents across all Cleveland Metropolitan School District K-8 schools
Discussion
This study is the first to examine the role of the objective tobacco retail environment on multiple tobacco product use (cigarettes, CCLCs and e-cigarettes) among a young, adolescent population. We used a comprehensive, ground-truthed retail database to calculate TRD using KDE and integrated it with a well-established adolescent behaviour surveillance system. While there was no association between current tobacco use behaviours and presence of tobacco retailers (ie, TRD and TRP), there was a strong, consistent and independent association between current tobacco use and self-reported walking to or from school as well as corner store visits before or after school.
Our results show no association between TRD and TRP with current youth tobacco product use. This finding contributes to a body of literature that has mixed results regarding tobacco retail outlets and tobacco use among youth. There is supporting research that has similarly found no association between TRD or TRP with current youth tobacco product use overall,57 58 and others that have identified associations between tobacco retail factors but only among specific youth. Among those who already use tobacco products, there may be a greater likelihood that they will use more of the tobacco product58 59 and among those who have never used tobacco products there, may be a greater likelihood of initiating tobacco product use.59 60
In this research, we observed high rates of non-cigarette tobacco product use with regard to CCLC and e-cigarette products among youth. Perceived risk associated with these products is low.61 62 Perception of harm from e-cigarette products is lowest among specific groups such as men, non-Hispanic Black populations, urban area residents and families with low SES.63 Future research may benefit from understanding the role of tobacco retail in conjunction with adolescents’ perceived risk and harm associated with non-cigarette tobacco products, particularly among these populations. Use of e-cigarettes is likely even greater now given recent national trends which further highlights the need for timely research on e-cigarette initiation and use behaviours among youth.64
While no associations were observed with TRD and TRP, our study found a strong association with the number of visits to retailers before and after school. This study replicates similar results published by Trapl et al which examined use of CCLC products among 7th and 8th grade youth in Cleveland.65 Importantly, in this study, TRD and TRP are explicitly measures of availability of tobacco products and are not meant to serve as indicators of advertising exposure. Thus, while objective, environmental measures of tobacco retail availability (ie, TRD and TRP) were not significant in this study, the importance of tobacco retail exposure (ie, number of retail visits) indicates that the retail environment is still very strongly associated with young adolescent tobacco product use. The relationship between the retail exposure and visits may be related to point-of-sale and in-store marketing. Middle and high school aged youth are more likely to use tobacco products if they recalled tobacco marketing signage suggesting that the exposure was substantive enough to become memorable.66 Measurement of the retail exposure therefore is not fully encapsulated by measures of TRD and TRP and by combining adolescent tobacco retail use measures with advertising and product exposure may provide an opportunity to expand on our understanding of the retail environment as an exposure.67
This research provided a unique opportunity to examine tobacco use behaviours among young adolescents in an environment with a high density of tobacco retail. In this sample, more than two-thirds of students (71.0%) who had smoked any of the three products in the last 30 days tried their first tobacco product before the age of 13 and nearly a third (30.5%) had tried their first tobacco product at age 10 or younger. Initiating tobacco product use before the age of 13 puts adolescents at greater risk for nicotine dependence compared with those who initiate at a later age.68 The substantial proportion of students trying tobacco at or before the age of 10 highlights a further need to understand environmental influences on youth tobacco initiation behaviours.
Despite the strengths of this study, there are limitations in our measurement of the tobacco retail environment. Our objective retail measures were specific to schools and did not capture measures of retailers in the adolescents’ activity space (eg, space between home and school). Many more adolescents reported stopping at a retail store before or after school than reported walking to/from school; thus, they may have been engaging in retail environment around their homes. This is consistent with prior findings that suggest that the tobacco retail environment surrounding homes has a greater impact on youth tobacco usage than the tobacco retail environment surrounding schools.69 70 Adolescent use of corner stores is substantial71 and, from a young age, youth visit corner stores unaccompanied by a parent or guardian.72 Understanding how and where retailers are used by adolescent consumers could provide novel insight into appropriately quantifying the tobacco retail exposure.
This study was limited to Cleveland, Ohio where there may not have been sufficient heterogeneity in the environment to be sensitive to the effects of TRD or TRP. In fact, Cleveland has a large number of retailers, and it is possible that there was insufficient variability in the lower end of density and higher end of proximity, which would illicit an effect. Maximum distance from school to any tobacco retailer was 0.63 miles, a very walkable distance, indicating that all students likely had reasonable access within a 10 min walk from their school building. This may dampen the generalisability of our results, but it may also highlight the importance of context. While there is limited variability in Cleveland’s objective retail environment, adolescent populations in different areas may perceive, experience, and use their environment in different ways which again, points to the need to understand how different populations make use of the objective environment. Finally, this study was not designed to be able to assess the potential mediating or moderating effects of tobacco retail availability, retail exposures and young adolescent tobacco use behaviours. This, however, highlights a unique opportunity for future research to explore how characteristics of the built tobacco environment may shape tobacco use behaviours among young adolescents.
What this paper adds
Adolescent cigarette smoking rates are associated with greater density of retailers in general; however, studies examining whether youth who live or attend schools in areas with higher tobacco retail density (TRD) are more likely to smoke have resulted in mixed findings.
These studies have almost exclusively examined TRD and cigarette smoking, ignoring the growing trend of youth use of non-cigarette products.
By pairing comprehensive, ground-truthed retail data with school-level adolescent surveillance data, this study examined associations between school-based retail environment exposures and current use of cigarettes, cigar products and e-cigarettes among young adolescents.
Students who self-reported walking to/from school or visiting retailers before/after school had significantly greater odds of current cigarette, cigar, e-cigarette and any tobacco use.
Measures of the objective retail environment were not associated with young adolescent use of tobacco products.
Data availability statement
Data are available upon reasonable request. All of the individual participant survey data is available upon request via data requests or research proposals submitted through the Cleveland Metropolitan School District Office of Research and Evaluation. Individual retailer data from the Cleveland Food Retail Database as well as the data collection tools and protocols can be made available upon request. Data requests and research proposals should be directed to the lead author, erika.trapl@case.edu.
Ethics statements
Patient consent for publication
Ethics approval
The CLE-YRBS protocol was approved by the Institutional Review Board at Case Western Reserve University.
References
Footnotes
Twitter @erikatrapl
Contributors All authors contributed the writing of this study. The study was conceptualised by ET and AA-R. Analysis was conducted by HG and SPM. ET was the lead author of the manuscript. All authors have reviewed the manuscript.
Funding This journal article is a product of a Health Promotion and Disease Prevention Research Centre supported by Cooperative Agreement Number 1U48DP005030 from the Centres for Disease Control and Prevention. The findings and conclusions in this journal article are those of the author(s) and do not necessarily represent the official position of the Centres for Disease Control and Prevention. Although author AAR is an FDA/CTP employee, this work was not done as part of his official duties. This publication reflects the views of the author and should not be construed to reflect the FDA/CTP’s views or policies.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.