Adolescents’ marijuana use is associated with adverse neurobehavioral and health effects from adolescence through adulthood. According to the Monitoring the Future study, secondary- and high-school students reported increasing trends in past-year and past-month marijuana use in the past decade . These trends coincided with the increasing recreational marijuana legalization and commercialization in the US since 2012. Emerging evidence suggested that statewide recreational marijuana legalization or commercialization might be associated with a reduction in perceived harmfulness and increases in marijuana use and marijuana use disorder among adolescents. There is scant research on the influences of local variations in marijuana regulatory and retail environments on adolescents’ marijuana use outcomes. Within a state with recreational marijuana legalization and commercialization, local jurisdictions may opt to prohibit recreational marijuana dispensaries or restrict their density and locations. RMDs may select their locations for cost and demand considerations. These local variations have resulted in considerable differences in the availability, proximity, and density of RMDs at neighborhood level. Although adolescents are prohibited from purchasing marijuana in RMDs, grow racks the presence of RMDs near where they congregate such as schools may promote the visibility and awareness of marijuana and shape favorable perceptions, attitudes, and behaviors towards marijuana use.
Indirect access to RMDs through third party adults, such as older friends, relatives, and street dealers, could also increase the availability of and accessibility to marijuana among adolescents. To date, only a few studies have evaluated the association between RMDs’ availability and crime rates using aggregate data at neighborhood level . No studies have focused on RMDs’ proximity and density and marijuana use outcomes in adolescent population. Another major concern on RMDs is their point-of-sale marketing activities, particularly those targeting adolescents. Dispensary storefronts have become the most commonly reported source of advertising among adolescents and adults after recreational marijuana commercialization. In a study conducted by our team in 2018, we observed that nearly three quarters of RMDs in school neighborhoods had marketing activities that were appealing to children. Informed by strong scientific evidence from tobacco literature that marketing activities promote initiation and use of tobacco among adolescents, the American Academy of Pediatrics policy statement “strongly recommends strict enforcement of rules and regulations that limit access and marketing and advertising to youth”. However, empirical evidence regarding the associations between RMDs’ marketing activities and adolescents’ marijuana use outcomes is still lacking. The goal of this study was to provide the first data point on the relationships of proximity and child-appealing marketing activities of RMDs with adolescents’ marijuana use.
We had three hypotheses. First, the proximity of RMDs is positively associated with the likelihood of adolescents’ marijuana use . Second, the presence of child-appealing marketing activities in RMDs is positively associated with the likelihood of adolescents’ marijuana use . Third, the associations between RMDs’ child-appealing marketing activities and adolescents’ marijuana use depend upon the proximity of RMDs . The study findings are expected to inform prevention and regulatory strategies that aim to protect adolescents from the negative influences of RMDs in school neighborhoods.The California Student Tobacco Survey is a repeated cross-sectional and state representative school survey among California middle and high school students . It has been administered biennially since 2001-2 with the purpose of monitoring adolescents’ tobacco use behaviors and perceptions. This study used CSTS 2017-8, which was administered between September 2017 and June 2018. CSTS 2017-8 used a two-stage cluster random sampling approach, where public and nonsectarian schools were first randomly selected from regions and then classrooms in selected schools were invited to participate. All the participating schools were invited to survey all the students in a particular grade. If a school chose not to survey all the students in a grade, five classrooms were randomly selected to participate. Of the 623 schools invited, 359 schools fielded the survey. The final data excluded 26 schools with response rates below certain thresholds, resulting in 151,404 adolescents in 333 schools in CSTS 2017-8.
The survey was administered online in both English and Spanish and completed between 15 and 25 minutes. Details about CSTS 2017-8 sampling approaches and survey implementation have been reported elsewhere. Because California started legal sales of marijuana in RMDs on January 1st, 2018, RMDs presumably had no impacts on adolescents’ reporting on past-month marijuana use before February 2018. Therefore, we restricted our analysis to the 244 school that completed the survey on or after February 1st, 2018. There were 103,914 adolescents surveyed in these 244 schools. After removing adolescents with incomplete information on demographics and marijuana use, the final study sample included 96,011 adolescents. It accounted for 73% of the schools and 67% of the adolescents in the complete CSTS 2017-8 sample.Between June and September 2018, our team collected data on locations and marketing activities of RMDs that had the closest proximity to the CSTS 2017-8 participating schools. Six trained fieldworkers obtained a list of marijuana dispensaries closest to the participating schools from crowd sourced online websites, including Weedmaps, Wheresweed, Leafly, and Yelp. Fieldworkers then made calls to verify dispensaries’ street addresses, operation status, and dispensary classification . We excluded dispensaries that were delivery only, inactive, or classified as medical marijuana dispensaries . When the dispensary closest to a school was excluded for the above reasons, the second closest dispensary was verified until an active RMD with a storefront was identified. We developed and pilot-tested a Web/smartphone-based surveillance tool for dispensary auditing, namely “Standardized Marijuana Dispensary Assessment-Children Focused ”. SMDA-CF assessed physical, economic, and marketing characteristics of RMDs, with special attention to marketing activities appealing to children. SMDA-CF items had moderate to high reliability overall, with a median kappa score of 0.8. After receiving training, fieldworkers audited the call-verified RMDs in teams of two to improve the reliability of data collection. Each visit to an RMD took 10-15 minutes on average.The primary predictors of interest included 1) the proximity of a school to the nearest RMD and 2) the presence of child-appealing marketing activities in the nearest RMD. We also control for the density of RMDs in school neighborhoods. The proximity between a school and its nearest RMD was computed with straight-line method using ArcGIS Version 10.5. In the main analysis, three proximity indicators were assessed: ≤1 mile, 1-3 miles, and > 3 miles. The 1- and 3-mile cutoffs were chosen based on common practices in tobacco and alcohol literature and geographic distribution of the RMDs around schools. The 3-mile buffer represents a distance that is easily reachable by bicycles, vehicles, and public transportation. In addition, the 1- and 3-mile cutoffs provided sufficient sample sizes of schools and adolescents in each cell for statistical analysis. In the sensitivity analysis, we also tested the sensitivity of results to different cutoffs, including 2-, 4-, and 5-mile. Child-appealing marketing activities in SMDA-CF were defined as products, packages, paraphernalia, vertical grow and advertisements that are “characterized by promotional characters , shaped like commercially sold products usually consumed by children , or using bright colors or bubble-like fonts ”. Separate binary indicators were created to indicate the presence of child-appealing products/packages, paraphernalia, and advertisements in RMDs.
To account for the influences of other RMDs that were more distantly located yet close enough to schools, we also considered the density of RMDs in school neighborhoods. It was measured as the weighted number of RMDs within a 3-mile buffer of a school, with weight .7 assigned to the number of RMDs within the 1-mile buffer and weight .3 assigned to the number of RMDs within the 1-3 mile buffer. To avoid double counting and collinearity, the computation excluded the audited RMD itself, which had the closet proximity to the school.Descriptive statistics were computed for individual, RMD, school, and census tract variables. To examine the associations between the proximity and marketing activities of RMDs with adolescents’ marijuana use, a series of multilevel logistic regressions were conducted with the first level of students nested in the second level of schools. All the statistical analyses were conducted using R packages. The three marijuana use outcomes were analyzed separately. Current use was analyzed among all the adolescents included in this study, heavy use was analyzed among current users who used marijuana in the past 30 days, and curiosity was analyzed among never users who never used marijuana before. We took a stepwise approach to test the three hypotheses. We first included proximity indicators in the model , then added indicators of marketing activities simultaneously with proximity indicators , and finally added interaction terms between the proximity and marketing activities along with indicators of proximity and marketing activities . All the regressions controlled for RMD density and individual, school, and census tract characteristics. Sampling weights were applied to all the analyses on adolescents.This study was the first attempt to assess the relationships between objectively measured recreational marijuana retail environments and adolescents’ marijuana use. We audited the locations and point-of-sale marketing activities of RMDs in school neighborhoods and merged auditing data with school survey data on a large sample of adolescents in California. We paid particular attention to child-appealing marketing activities, which were presumably more influential to adolescents than general marketing activities. Instead of aggregating data at zip code or census tract level, we examined individual-level outcomes and simultaneously accounted for between- and within-school variations. Our first hypothesis that a closer proximity of RMDs is associated with a greater likelihood of adolescents’ marijuana use was not supported by the findings. In fact, a closer proximity was found to be associated with lower likelihoods of some outcomes in some model specifications in sensitivity analysis. Although no similar studies on RMDs can be used to compare to our findings, existing evidence on medical marijuana dispensaries did show mixed relationships between dispensaries’ proximity and marijuana use among adolescents. Whether and how the proximity of RMDs in school neighborhoods is associated with adolescents’ marijuana use outcomes deserve further research. Our second hypothesis that the presence of child-appealing marketing activities in RMDs is associated with a greater likelihood of adolescents’ marijuana use was not supported by the findings, either. However, when we examined the third hypothesis , we did find some evidence that child-appealing products, packages, and paraphernalia in RMDs in very close proximity to schools might be associated with a greater odds of current use or heavy use. It is likely that these items were resold or freely distributed to adolescents by third party adults, such as older friends,relatives, street dealers, who resided or worked in school neighborhoods. The interaction effects of RMDs’ proximity and marketing activities were not found on child-appealing advertisements. One plausible explanation is that nearly all RMDs we audited complied with age restrictions by ID check. Adolescents therefore had little chance to see advertisements inside of the RMDs, which could not be taken out by third party adults. It should be noted that the findings on interaction effects were sensitive to the selection of proximity cutoffs and model specifications. This is why we considered the strength of the evidence on interaction effects to be only moderate. Future research is strongly encouraged to add more data points. The findings have policy implications. If the impacts of point-of-sale child-appealing marketing activities depend upon the proximity of RMDs to schools, stronger surveillance may be needed to monitor marijuana-related perceptions and behaviors in schools that have RMDs located near to them. Even though almost all states with legal sales of recreational marijuana prohibit products and advertisements specifically targeting children, our dispensary auditing data demonstrated a wide presence of these prohibited items in school neighborhoods. Actions should be taken to reduce child-appealing marketing activities and prevent adolescents from potential exposure. This study has limitations. First, the cross-sectional examination of associations should not be interpreted as causality. Second, the study sample was restricted to 73% of the CSTS 2017-8 schools that completed the survey on or after February 1st, 2018. The generalizability of the findings to the entire California may be a concern. Third, we audited RMDs after the CSTS 2017-8 was completed in order to have an accurate and complete list of surveyed schools and conduct auditing in a cost-efficient manner. To what extent our observations on RMDs applied to the time when the schools were actually surveyed was unknown. Fourth, the marketing activity predictors were indicators of presence instead of continuous quantity measures due to feasibility considerations in fieldwork.