Socioeconomic Disparities in Vape Shop Density and Proximity to Public Schools in the Conterminous United States, 2018

Abstract

We conducted an environmental justice assessment examining the distribution of specialty vape shops in relation to where minority and low-income youth live and attend school. We collated and examined the density of vape shops in public school districts in 2018 throughout the conterminous United States using geographic information systems. We calculated the proximity of vape shops to public middle and high schools through nearest neighbor analysis in QGIS software. We examined the statistical relationships between the density of vape shops in school districts, and proximity to schools, with the proportion of racial/ethnic minorities and those living in poverty. We found that vape shops are more densely distributed, and are in closer proximity to schools, in school districts with higher proportions of Asian and Black or African American populations. However, vape shops were further away from schools in school districts with higher proportions of the population in poverty. The proximity and higher density of vape shops in relationship to schools in Asian and Black or African American communities may result in disproportionate health impacts due to greater access and exposure to vape products and advertisements. Our results may help school district administrators prioritize and target efforts to curb youth vaping (e.g., health education curricula) in these school districts with high density and closer proximity of vape shops to schools. Policy efforts, such as local ordinances restricting the promotion and sale of vaping products close to schools, could help prevent disproportionate human and environmental health impacts to minorities.

INTRODUCTION

The disproportionate presence of traditional tobacco retailers in racial/ethnic minority and low-income communities in the United States is well documented (Lee et al., 2017; Rodriguez, Carlos, Adachi-Mejia, Berke, & Sargent, 2013). Scientific evidence suggests that tobacco retailer proximity and density may contribute to environments that facilitate youth tobacco use both where youth live and where they go to school, thereby potentially contributing to health disparities (Berg, 2018; McCarthy et al., 2009; Mennis & Mason, 2016).

This raises concerns about environmental justice (EJ); EJ is achieved when all people have equal protection from environmental and health hazards and live, work, and go to school in healthy environments (https://www.epa.gov/environmentaljustice). Eliminating differential exposures to environmental health hazards among minority and low-income populations is a strategy for reducing health disparities and promoting health equity (Nweke & Lee, 2011). Therefore, EJ assessments examining tobacco retailer density and proximity in minority and low-income communities aid in tackling health disparities due to tobacco use among youth.

As with traditional tobacco retailers, the distribution and proximity of retailers specializing in electronic nicotine delivery systems (ENDS) may pose environmental health risks (Bostean, Crespi, Vorapharuek, & McCarthy, 2016), raising EJ and health equity concerns (Lee, Orlan, Sewell, & Ribisl, 2018), especially in terms of youth exposure to such retailers. National Youth Tobacco Survey data from 2018 show a marked increase in ENDS use among U.S. youth (Gentzke et al., 2019). ENDS are now the most commonly used tobacco product among middle and high school students. As nicotine use may result in addiction and may harm the developing brain (Gentzke et al., 2019), ENDS use among youth may result in health harms.

Recent evidence suggests that retail environments may play a role in this uptick in youth use. Nationally representative data show that, after social sources, retailers specializing in tobacco products are the second most common source for ENDS products among youth (Tanski et al., 2018). Both exposure to advertising at ENDS specialty retailers and the density of ENDS specialty retailers around high schools are positively associated with ENDS use among students (Giovenco, Casseus, et al., 2016). A recent study found that close proximity of ENDS specialty retailers to middle schools is associated with ENDS use among middle school students, but there was no significant relationship with use among high school students (Bostean et al., 2016). A recent systematic review found that ENDS specialty retailers use promotional strategies different from those used by traditional tobacco retailers, including description of ENDS as having health benefits relative to combustible tobacco products. This raises concerns that the disproportionate presence of these retailers may pose risks for youth initiation (Lee et al., 2018). Given this body of evidence, the question arises whether ENDS specialty retailers are concentrated in areas where minority and low-income youth live and attend school and does this contribute to health disparities related to tobacco use.

Several studies examine ENDS specialty retailer density and its socioeconomic correlates. One study of a county in California found increased ENDS retailer density in areas with lower socioeconomic status and a higher proportion of Hispanic population in the adjusted model (Bostean, Sanchez, & Lippert, 2018). The only national study of ENDS retail density of which we are aware yielded similar findings and established that ENDS retailers are primarily an urban phenomenon (Dai, Hao, & Catley, 2017). Conversely, studies of ENDS retailers in New Jersey (Giovenco, Duncan, Coups, Lewis, & Delnevo, 2016) and New York (Giovenco, 2018) found that increased ENDS retail density was significantly more likely in middle-income census tracts with relatively high proportion of non-Hispanic White residents.

Interpreting the divergent results of these studies requires consideration of potential regional differences, as well as how ENDS retailers, that is, vape shops, are defined (Giovenco, 2018; Lee et al., 2018). Lee et al. (2018) observe that previous studies have defined vape shops in two ways: (1) retailers selling only ENDS products and no other types of tobacco products and (2) retailers who specialize in selling not only ENDS but also other types of tobacco products. As noted by Giovenco (2018), a more restricted definition of vape shop is needed to accurately assess the socioeconomic correlates of specialty ENDS retailers. This helps reduce the potentially confounding influence of tobacco retailers who carry a variety of product types, as they are far more numerous and the association of traditional tobacco retailers with minority and low-income communities is well-established. In our study, we follow the more restrictive definition, excluding establishments that sell tobacco products other than ENDS.

To our knowledge, no study has employed this more restrictive definition to examine the distribution of vape shops on a national scale, including their density and proximity with respect to where minority and low-income youth both live and go to school. This information would help characterize the EJ aspects of vape shops (Berg, 2018). Executive Order 12898 and the associated Council on Environmental Quality guidance direct federal agencies to identify and address EJ aspects of their activities. Following this, the Federal Interagency Working Group on EJ recommends that federal agencies consider the potential for environmental health hazards to disproportionately affect minority and low-income populations through the National Environmental Policy Act processes (U.S. Environmental Protection Agency, 2016).

In our present research, we therefore attempt to provide scientific evidence to help inform this research gap and address the EJ implications of vape shop density and proximity. We examined the density of vape shops in public school districts, and their proximity to public middle and high schools, throughout the conterminous United States. We then examined the statistical relationships between the density of vape shops in school districts, and proximity to schools, with the proportion of racial/ethnic minorities and those living in poverty.

METHODS

We used geographic information systems (GIS) to collate and examine the density of vape shops in public school districts and their proximity to schools throughout the conterminous United States. We analyzed data from four sources—vape shop listings and addresses from online directories, school districts geospatial data for school year 2017-2018 (https://nces.ed.gov/programs/edge/Geographic/DistrictBoundaries), race/ethnicity and federally defined poverty data (American Community Survey 2013-2017 averages) from the National Center for Education Statistics (https://nces.ed.gov/programs/edge/Demographic/ACS), and public schools location data for 2014-2015 school year provided by the National Geospatial-Intelligence Agency in collaboration with Oak Ridge National Laboratory (https://hifld-geoplatform.opendata.arcgis.com/datasets/public-schools). We included five mutually exclusive race/ethnicity categories: Hispanic or Latino, American Indian and Alaska Native (AIAN), Asian, Black or African American, and White.

For compiling a database of vape shop listings, we utilized Internet sources Yelp, Google Maps, and Google searches as well as a business database provided by ReferenceUSA, following recommendations by Lee et al. (2018). Yelp and ReferenceUSA were the primary sources of compilation performed July through December 2018, with relatively few entries not found in either added from Google Maps through supplemental searches (see Supplemental Figure S1, available with the online version of this article). Online search protocols for Yelp and ReferenceUSA largely followed procedures as laid out in the supplementary material provided by Lee, D’Angelo, Kuteh, & Martin (2016). We used the search term vape within Yelp, with vape shops for category, and vape shops in ReferenceUSA (Standard Industry Classification Code for E-cigarette shops, 599306) and Google Maps. We compiled a total of 17,361 listings including 12,072 from Yelp; 5,231 from ReferenceUSA; and 58 from Google Maps. We used Yelp reviews and Google reviews for confirmation of the business being open (Kong, Unger, Baezconde-Garbanati, & Sussman, 2017).

After removal of duplicates and closed listings, 12,969 listings remained. Following the descriptions of “vape shops” provided by Giovenco (2018), we excluded establishments that sold any other types of tobacco products, cannabidiols, or related paraphernalia. To identify establishments that sold any other types of tobacco products, we used the descriptions and pictures provided by the owners of the establishments in their websites and Yelp profiles and also perused Google and Yelp reviews and reviewer-submitted pictures. Vape shop data searches and reviews of business for vape shop classification were manually conducted by the first author. We identified 7,479 unique vape shops (Yelp 6,152; ReferenceUSA 1,294; Google Maps 33) and georeferenced them using the U.S. Census Bureau’s Geocoder (https://geocoding.geo.census.gov/). For those with no exact matches (~10%), we manually geocoded them using Google Maps.

From the overall public schools GIS data file we selected all the middle and high schools (37,270) for the conterminous United States. We spatially overlaid the school location and vape shops data on the 13,205 school districts in the conterminous United States with resident population and selected the 10,989 school districts that contained at least one middle or high school. Unified school districts typically constitute both elementary and secondary (middle and high) schools. Given the study focus on secondary schools, we removed the 2,216 school districts that did not contain a high or middle school. We then calculated the density of vape shops and schools at each school district with the following respective formulae:

$$\mathit{Vape}\mathit{shop}\mathit{density}=\frac{\mathit{number}\mathit{of}\mathit{vape}\mathit{shops}\mathit{at}\mathit{each}\mathit{school}\mathit{district}}{\mathit{total}\mathit{population}\mathit{of}\mathit{school}\mathit{district}}\times \mathrm{100,000}$$

$$\mathit{Schools}\mathit{density}=\frac{\mathit{number}\mathit{of}\mathit{public}\mathit{middle}\mathit{and}\mathit{high}\mathit{schools}\mathit{at}\mathit{each}\mathit{school}\mathit{district}}{\mathit{total}\mathit{population}\mathit{of}\mathit{school}\mathit{district}}\times \mathrm{100,000}$$

We further derived the ratio of vape shops to school densities. We also measured the Euclidean distance (straight-line distance) between each school and its closest vape shop through NNJoin plugin in QGIS software (QGIS Development Team, 2018) using USA Contiguous Albers Equal Area Conic projection (EPSG: 102003). We used the school location GIS data as the “input layer” and vape shops GIS data as “join layer” and measured distance of each school to the nearest vape shop. We then calculated the median distance of schools to vape shops for 10,687 of the 10,989 school districts selected.

We examined the statistical relationships of the density of vape shops in school districts, and proximity to schools (measured in distance), with the proportion of each racial/ethnic category and those living in poverty. We ran separate quasi-Poisson generalized linear models to associate vape shop density, ratio of vape shops to school densities, and median vape shop distance from schools to each race/ethnicity category in school districts (%) and to those living in poverty (%). We used F tests to determine the statistical significance of the fixed effects. We used QGIS software for GIS data manipulations and visualizations and R program for the statistical analyses (R Core Team, 2018). We extracted and plotted the generalized linear models results with R packages effects (Fox, 2003) and ggplot2 (Wickham, 2016), respectively.

RESULTS

A total of 10,989 school districts and 7,479 vape shop addresses were included in the final analysis. Of these school districts, 73.3% (8,053) had no vape shops, 13.7% (1,502) had one vape shop, 7.4% (818) had two, and 5.6% (616) had three or more vape shops within them.

Vape shop density in the school districts was significantly positively associated with percentage Asian and Black or African American populations, with density increasing as a function of both races based on unadjusted results. Vape shop density in the school districts was significantly negatively associated with percentage AIAN and percentage White population, with density decreasing as a function of both races based on unadjusted results. Hispanic population and poverty were not significantly associated with the density of vape shops in school districts (Figure 1A–F, Supplemental Table S1).

FIGURE 1. Relationship Between Vape Shop Density (per 100,000 Population) and Socioeconomic Variables in School Districts of the Conterminous United States During 2018.

NOTE: Depicted are the unadjusted model estimates (dark line) and confidence limits (95% confidence interval; shaded area) from separate quasi-Poisson generalized linear models analyzing vape shop density associations with races/ethnicities (A-E) and poverty (F). Values presented here were back-transformed from the original log-link function estimated model coefficients. Results except those in D and F were statistically significant (α = .05). Note that the x and y axes differ for each separate plot to allow for better visual understanding of the trend in each separate model.

Ratio of vape shops to schools in the school districts was significantly positively associated with percentage Asian and Black or African American populations, with ratio increasing as a function of both races based on unadjusted results. Ratio of vape shops to schools was significantly negatively associated with percentage AIAN, percentage White populations, and poverty, with the ratio decreasing as a function of these factors based on unadjusted results. Hispanic population was not significantly associated with the ratio of vape shops to schools in school districts (Figure 2A–F, Supplemental Table S1).

FIGURE 2. Relationship Between the Ratio of Vape Shops to Schools (per 100,000 Population) and Socioeconomic Variables in School Districts of the Conterminous United States During 2018.

NOTE: Depicted are the unadjusted model estimates (dark line) and confidence limits (95% confidence interval; shaded area) from separate quasi-Poisson generalized linear models analyzing ratio of vape shop to school associations with races/ethnicities (A-E) and poverty (F). Values presented here were back transformed from the original log-link function estimated model coefficients. Results except those in D were statistically significant (α = .05). Note that the x and y axes differ for each separate plot to allow for better visual understanding of the trend in each separate model.

Median distance of schools to vape shops in the school districts was significantly negatively associated with percentage Asian, Black or African American, and Hispanic or Latino populations, with vape shops in closer proximity to schools in districts with higher proportions of these populations based on unadjusted results. Median distance of schools to vape shops was significantly positively associated with percentage White population and poverty. Distance to vape shops increased with percentage White population and population in poverty (Figure 3A–F, Supplemental Table S1).

FIGURE 3. Relationship Between Distance From Schools to Vape Shops (km) and Socioeconomic Variables in School Districts of the Conterminous United States During 2018.

NOTE: Depicted are the unadjusted model estimates (dark line) and confidence limits (95% confidence interval; shaded area) from separate quasi-Poisson generalized linear models analyzing median distance between schools and vape shops and associations with race/ethnicity (A-E) and poverty (F). Values presented here were back transformed from the original log-link function estimated model coefficients. All results were statistically significant (α = .05). Note that the x and y axes differ for each separate plot to allow for better visual understanding of the trend in each separate model.

DISCUSSION

The rapid recent growth of the e-cigarette market corresponds with a rapid increase in e-cigarette use among teenagers and young adults, raising public health concerns (Berg, 2018; Bostean et al., 2016; Dai et al., 2017; Gentzke et al., 2019; Giovenco, Casseus, et al., 2016). We examined the density of vape shops in 10,989 public school districts and their proximity to public middle and high schools in the conterminous United States. Our results show that vape shops are more densely distributed in school districts with higher proportions of Asian and Black or African American populations. Vape shops are in closer proximity to schools in school districts with higher proportions of Asian, Black or African American, and Hispanic or Latino populations. This is consistent with reports of the disproportionate concentration of traditional tobacco retailers in racial/ethnic minority communities (Berg, 2018; Rodriguez et al., 2013). However, vape shops are further away from schools in school districts with higher proportions of the population in poverty. In this respect, the distribution of vape shops differs from that of traditional tobacco retailers, which are disproportionately present in low-income neighborhoods (Rodriguez et al., 2013).

A key strength of our study is that we quantify the association of vape shop density and proximity to schools with socioeconomic factors at the national scale, using strict criteria for inclusion of vape shops, thereby controlling for the potentially confounding effects of retailers who sell different types of tobacco products (Giovenco, 2018). Our study is the first to provide the socioeconomic correlates of three separate measures of vape shop distribution around schools nationally—density, ratio of vape shops to schools, and proximity to schools. Broadly, these separate measures show that vape shops are more densely distributed and are in closer proximity to schools in districts with higher proportions of certain racial/ethnic minority populations. Our results do not show significant associations between vape shop density, ratio of vape shops to schools, and the proportion of Hispanic or Latino population, in contrast to the findings of Dai et al. (2017) and adjusted model results of Bostean et al. (2018). Our results also indicate a negative association between vape shop distribution and the proportion of a non-Hispanic White population, in contrast to findings from New York and New Jersey (Giovenco, 2018; Giovenco, Duncan, et al., 2016). The positive association of vape shop density and proximity with higher minority populations, in general, could be attributed to the higher racial diversity and minority populations in urban school districts than rural ones and vape shops being predominantly an urban phenomenon (Dai et al., 2017).

The ratio of vape shops to school densities and the proximity of vape shops to schools were both negatively associated with poverty, with a higher ratio of vape shops and a closer proximity to schools in school districts with a lower proportion of population in poverty. We did not find a significant association between vape shop density and poverty levels, consistent with reported national trends of a lack of significant statistical relationship (Dai et al., 2017). Our results showing a lack of statistically significant association between vape shop density and poverty differ from findings in New Jersey where vape shops density was associated with median to higher income communities, than low-income communities (Giovenco, Duncan, et al., 2016). These studies examined vape shop distribution at the census tract level, while we examined it in school districts, which might also explain potential differences in results. The negative associations of the ratio of vape shops to schools and proximity to schools with poverty could be due to the higher relative costs associated with ENDS usage compared with traditional tobacco products and could also help explain the higher usage of e-cigarettes among White, non-Hispanic youth than in racial/ethnic minorities (Gentzke et al., 2019).

We examined the density and proximity of vape shops around schools and associations with socioeconomic attributes; our results suggest inequitable exposures to environmental risks among youth in Asian and Black or African American communities. The close proximity and higher density of vape shops in relationship to schools in Asian and Black or African American communities may provide youth with greater access and exposure to ENDS products and advertising, which may potentially result in greater use (Giovenco, Casseus, et al., 2016) and disproportionate health impacts on these communities.

The study results should be interpreted with several limitations in mind. First, this study employed a cross-sectional design examining a highly dynamic retail sector that is likely to change over time. Second, the school GIS layer, which represents the latest available national compilation of school locations, was from a different school year (2014-2015) than the school district boundaries data (2018). We suspect this is a relatively minor factor that is unlikely to have affected study results because of relatively few secondary school closings compared with total number of schools (294 during 2015-2016; https://nces.ed.gov/fastfacts/display.asp?id=619). Third, our analytical procedures did not address spatial autocorrelation (similarity in attributes at locations closer to one another than locations farther apart). Finally, network distances, providing on-road distances of vape shops to schools, may provide better accuracy for proximity calculations than Euclidian distances. Nevertheless, we provide an overall robust assessment of the socioeconomic correlates of vape shop density and proximity to schools.

Our results highlight the relevance of conducting and using EJ assessments to inform tobacco control policies, including for ENDS. While ENDS use among youth is currently highest among non-Hispanic Whites (Gentzke et al., 2019), our findings show a disproportionate distribution of vape shops where Asian and Black or African American youth live and go to school, raising EJ and health equity concerns. By documenting that Asian and Black or African American youth may experience disproportionate exposure to vape shops, potentially including ENDS labeling and advertising, our results can help inform regulatory efforts and postmarketing surveillance of ENDS products. Our results may help school district administrators prioritize and target efforts to curb youth vaping (e.g., health education curricula) in these school districts with high density and close proximity of vape shops to schools.

Additionally, retailers in racial/ethnic minority areas are significantly more likely to sell tobacco products to minors (Lee, Landrine, Torres, & Gregory, 2016). Therefore, our results may also help prioritize targeted enforcement actions in vape shops near schools in moderate to higher income areas with a high proportion of Asian and Black or African American youth. Our study adds to growing evidence (Bostean et al., 2016; Giovenco, Casseus, et al., 2016) that informs policy efforts, such as local ordinances restricting the promotion and sale of vaping products close to schools, to help prevent disproportionate human and environmental health impacts to minority youth. Overall, we provide a baseline view of the national ENDS retail landscape, disentangled from the effects of retailers who also sell other tobacco products, which future research efforts can build on.