Article Text
PDF
Abstract
Objective Tobacco retailers are an important source of tobacco products for minors. Previous research shows racial discrimination in sales to minors, but no national study has examined neighbourhood correlates of retailer under-age sales.
Methods We accessed publicly available results of 2015 US Food and Drug Administration (FDA) inspections of tobacco retailers (n=108 614). In this cross-sectional study, we used multilevel logistic regression to predict the likelihood of retailer sale to a minor based on tract characteristics. We assessed the proportion of residents identifying as American Indian, Asian, Black, Latino and White; Isolation Index scores for each racial/ethnic group; the proportion of people less than age 65 living in poverty; and the proportion of residents age 10–17 in relation to retailer inspection results.
Results The proportion of American Indian residents, Black residents, Latino residents and residents less than age 65 under the poverty line in a neighbourhood are independently, positively associated with the likelihood that a retailer in that neighbourhood will fail an under-age buy inspection. The proportion of White residents and residents age 10–17 are independently, negatively associated with the likelihood of sale of tobacco products to a minor. Isolation Index scores show a similar pattern. In multivariable models holding neighbourhood characteristics constant, higher proportions of Black (+), Latino (+) and age 10–17 (−) residents remained significant predictors of the likelihood of under-age sale.
Discussion Regulatory agencies should consider oversampling retailers in areas with higher likelihood of sales to minors for inspection. Interventions with tobacco retailers to reduce inequities in youth access should be implemented.
- Disparities
- Prevention
- Advertising and Promotion
Statistics from Altmetric.com
Introduction
Tobacco retailers remain an important source of tobacco products to minors, especially older minors.1 Given that addiction to tobacco products remains a leading cause of morbidity and mortality,2 curtailing access to tobacco products has a long history in public health policy efforts.3–5 The Family Smoking Prevention and Tobacco Control Act (FSPTCA) of 2009 authorised the US Food and Drug Administration (FDA) to restart a national tobacco retailer compliance and enforcement programme to compliment existing state-level enforcement required by the Synar Amendment. Between 2010 and 31 March 2016, FDA's state subcontractors had conducted over 365 000 under-age buy inspections.6
Where these inspections are conducted is decided by state subcontractors.7 The FDA's inspection programme requires participating states to develop sampling plans for inspections that consider ‘areas that are considered at higher risk for regulatory violations, such as: … regions with lower socioeconomic populations (historically associated with market targeting)’ (ref. 7, p. 15). The FDA also requires states to ensure coverage in ‘racial and ethnic minority communities’ (ref. 7, p. 15).
A 2016 systematic review of youth access to tobacco documented that the likelihood of a sale to a minor varies by the race, ethnicity and gender of the minor.8–10 However, that review also noted limited and conflicting research on the role of community or neighbourhood characteristics (eg, neighbourhood deprivation or racial/ethnic composition) in the likelihood of a retailer's sale to a minor.8 Thus, to inform our understanding of the aetiology of inequities in tobacco use and guide states' development of future sampling strategies, this paper had three aims: to assess the relationship between (1) neighbourhood poverty, (2) neighbourhood racial/ethnic composition and (3) racial/ethnic isolation with retailer non-compliance with youth access regulations.
Methods
We accessed publicly available inspection results posted by the FDA for the 2015 calendar year on 22 January 2016, for the US states, the District of Columbia and the US territories. The FDA provides retailer name, address, participation of minor in the inspection and if a violation was identified (no violation, warning letter for violation, or civil money penalty for violation). We used inspections where a minor was involved and where the date of FDA's decision about the outcome of the inspection was within the 2015 calendar year. We excluded inspections where the inspection result was listed as ‘N/A’ (n=2). We geocoded the resulting 137 578 inspection results using Texas A&M Geocoding Service based on the address of the retail location.11 All but 797 (0.6%) were successfully geocoded to latitude and longitude; 82% of geocode failures were located in US territories (eg, Guam). As estimates for tract composition can be unstable, we removed 143 inspections in tracts with 100 or fewer residents, leaving 136 638 inspections. There were 28 024 repeat inspections; we conducted analyses only on the first inspection reported (n=108 614). No poverty data were available for 14 inspections, and we used pairwise deletion for this variable.
We used census tracts to approximate neighbourhoods. There were 28 236 unique census tracts across 51 states, territories and the District of Columbia. The number of inspections in states ranged from 1 (Delaware) to 9855 (Florida) (mean=2129.7, SD=2036.0) and in census tracts from 1 to 68 (mean=3.4, SD=3.0). We used ArcGIS 10.2 to spatially join inspection locations to American Community Survey 5-year census tract estimates (2010–2014) obtained from Social Explorer.12 Tract data included (1) race/ethnicity (the proportion of residents self-identifying as American Indian/Alaska Native, Asian American/Pacific Islander, Black or African American race, as White race alone and as Hispanic or Latino of any race); (2) poverty status (the proportion of residents through age 64 living under the poverty line); and (3) residential segregation of American Indians, Asians, Blacks, Latinos and Whites measured by census tract Isolation Index for each group.13–15 Proportions of residents were multiplied by a constant for model convergence and interpretation (eg, 12%=1.2). The Isolation Index ranges from 0 to 1, and is the average probability that racial/ethnic minorities will encounter only racially/ethnically similar others (ie, no one of other races or ethnicities) in their census tract. An Isolation Index ≥0.60 indicates high residential segregation.16–18 We rescaled Isolation Index to a range of 0–10 (eg, a one-unit increase in Isolation Index is coded as the equivalent of going from 0.12 to 0.22).
The FDA contracts with states and private contractors to implement inspections.7 States are required to propose a sampling strategy;7 however, sampling strategies are not publicly available. Previous research has found substantial differences in outcomes, likely due to differences in how states implement FDA inspections.19 We identified similarity in inspection results within states (intraclass correlation=0.14) and used multilevel modelling with a random state intercept to address this violation of the assumption of independence between observations. We first modelled unadjusted relationships. We then report an adjusted model omitting White racial composition, which correlates strongly with Black racial composition (rs(n=28, 236)=−0.82, p<0.001), and isolation indices, which correlate strongly with racial/ethnic composition (rs's>0.90, p's<0.001). We conducted analyses in SAS V.9.4 (Cary, North Carolina, USA) using PROC GLIMMIX. Sensitivity analyses confirmed that the inclusion or exclusion of US territories did not change the pattern of results.
Results
Of the 108 614 inspections using minors under age 18 to assess under-age sales of a regulated tobacco product, violations were observed in 16.8% of all inspections. Some neighbourhood characteristics were associated with the likelihood of a sale to a minor (table 1). When adjusting for neighbourhood characteristics, the proportion of residents identifying as Black race and Latino ethnicity remained significant, positive predictors of the likelihood of sales to minors and the proportion of youth age 10–17 remained a significant, negative predictor.
ORs of retailer illegal sales to minors predicted by neighbourhood characteristics, USA, 2015
Discussion
Certain racial/ethnic composition and poverty characteristics of census tracts are associated with the likelihood that a retailer in a given tract will fail an under-age buy inspection for tobacco products. While the associations identified are not large (eg, for every 10-percentage point increase in the proportion of Black residents the odds of a retailer selling to a minor increased by 7%), such effects matter at a population level.20 A similar pattern exists for segregation as measured by the Isolation Index; however, the isolation index provides a more conceptually meaningful measure (ie, the average probability of encountering only other residents of the same demographic group) than neighbourhood composition. Given a long history of residential racial segregation in the production of health inequities,21 use of the Isolation Index to target retailer compliance efforts should be considered.
Our findings are consistent with data on existing health and neighbourhood inequities. The mix of retailers and retailers' behaviours in neighbourhoods with a greater proportion of Black residents may be different than in neighbourhoods with a greater proportion of White residents.22 Retailers are more likely to sell tobacco products to Black and Latino minors;9 ,23 Black and Latino minors are more likely to live in neighbourhoods with disproportionately higher levels of tobacco industry marketing;24 ,25 and Black and Latino youth are more likely than White youth to reside in low-income communities.17 Our findings add to this literature by showing that living in a neighbourhood with a greater proportion of American Indian, Black or Latino residents is associated with higher likelihood of tobacco sales to minors.
These inequities are amenable to regulatory and community intervention. Tobacco retailer licensing can help enforce youth access laws,26 community efforts can reduce the amount of tobacco marketing,27 and retailer density reduction has potential to work better in more urban, denser areas.28 FDA's inspections are not designed to be a statistically valid random sample. Instead, they are based on a plan developed by the states in conjunction with the FDA, which may consider several factors, including areas that are considered at higher risk for regulatory violations such as minority communities.7 FDA's Office of Compliance and Enforcement should work to ensure state subcontractors are fully identifying and covering retailers in communities with greater poverty and higher proportions of American Indian, African American or Hispanic residents.29 Use of the proportion of youth may not target areas with greater likelihood of violations. FDA's power over approving sampling plans7 should be leveraged in ways that reduce neighbourhood inequities.
Limitations
One source of concern about this study is that the sampling strategies used are unknown. Thus, although we had a large sample obtained from a government data source, the data are not generalisable outside of FDA inspections, and we cannot assess the prevalence of violations. Likewise, we cannot assess the race/ethnicity, age and gender of the minors involved, all of which play an important role in the identification of a violation.8–10 ,23 In addition, we did not have access to store type or chain versus non-chain status, and these are likely to be important covariates of non-compliance.
Conclusion
The availability of tobacco products to minors from tobacco retailers varies by neighbourhood characteristics including racial/ethnic composition, poverty and segregation. The relationships identified here are consistent with well-documented inequities in tobacco retailer marketing.23 State FDA subcontractors should design sampling strategies that ensure resources are directed at maximising retailer compliance. The FDA should adapt retailer education programmes to be based on these findings and leverage its power over state sampling strategies to reduce these inequities.
What this paper adds
Community characteristics including poverty, racial/ethnic composition and segregation are associated with the likelihood of a retailer violation of youth access provisions in the US Food and Drug Administration (FDA) inspections.
States should consider these inequities in designing sampling plans for FDA-funded retailer inspection programmes.
References
Footnotes
Twitter Follow Joseph Lee at @Joseph_GL_Lee
Contributors JGLL conceptualised the study, conducted the analyses and wrote the first draft. All authors provided critical feedback on the conceptualisation of the study, interpreted the results, edited the manuscript and approved the final draft.
Funding Research reported in this publication was supported with trainee support by grant number P50DA036128 from the US NIH/National Institute on Drug Abuse and Food and Drug Administration (FDA) Center for Tobacco Products (KRG). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the FDA. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests JGLL has a royalty interest in a store audit/compliance and mapping system, Counter Tools (http://countertools.org), 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.
Data sharing statement Data used in this study are available from the first author.