LCA has been widely used in the alcohol and drug policy literature to classify regulatory approaches

There are viable arguments for and against modeling cannabis policies after alcohol. Those against coordination argue that existing approaches to regulating alcohol have flaws that should not be carried over to cannabis. For example, whereas successful approaches to tobacco control typically involve robust demand reduction policies, some publicly-funded education campaigns painted alcohol as an appealing “forbidden fruit”, thereby unintentionally increasing demand . Prior alcohol policies have relied on “responsible use” by adults, absolving governments of the responsibility to protect public health and failing to convey the health risks associated with even low levels of consumption . Past approaches to alcohol control policy making have also involved partnerships with the alcohol industry which created regulatory conflicts of interest and resulted in policies that prioritized business interests over public health . Thus, distinct, yet stronger, recreational cannabis controls may be desirable. There are also differences in the psychoactive properties of alcohol and cannabis that may warrant different policies. For instance, driving under the influence of alcohol appears to be riskier than driving under the influence of cannabis . The alcohol control template may also lead policymakers to overlook the many moderate alternatives between the extremes of complete prohibition and an alcohol-like commercial model, such as allowing adults to grow their own cannabis or government monopolies .

Those favorable to modeling cannabis policies after alcohol argue that cannabis is being commercialized in ways that are similar to other legal substances,drying rack cannabis and a number of public health experts have urged US states to model recreational cannabis policies on existing policies for alcohol or tobacco . Alcohol and cannabis share similar risks: both are intoxicating, addictive, and particularly harmful to youth . Decades of research documents that greater access to retailers, exposure to marketing, and lower prices increase alcohol use by youth and young adults . Specific alcohol control policies can prevent underage use, addiction, and related harms . These policies have obvious analogues in cannabis control. Policy regimes that are coordinated across substances may be more effective in achieving their goals, whereas inconsistencies in alcohol and cannabis control policies can lead to gaps in public health protection . Policy coordination across substances can also ease implementation through a common enforcement and compliance infrastructure. Thus, some U.S. states have made alcohol control agencies responsible for cannabis control . Further, because different social groups may preferentially use alcohol or cannabis, policy coordination can promote fairness and avoid stigmatization . Finally, policy coordination promotes political learning . Lessons learned from decades of passing and implementing alcohol control policies can inform cannabis controls, while reducing uncertainty both for policymakers and the general public .

Even when policy coordination is intended, in federated countries local governments may frustrate efforts to coordinate cannabis and alcohol control policies if given discretion by higher levels of government . In the US, for example, local governments are granted considerable discretion by states to regulate where retail cannabis outlets can be located, their hours of operation, product potency, packaging, marketing, and tax rates, resulting in extreme local-level heterogeneity in approaches to cannabis control . On the one hand, this local discretion can be used to regulate the co-location of alcohol and cannabis outlets in low income neighborhoods already overburdened by alcohol outlets, thus promoting public health and equity. On the other hand, constituents may pressure their local policymakers to keep both types of outlets out of wealthier neighborhoods—the so-called “not in my backyard” or NIMBY phenonmenon—resulting in the opposite . The concern that local governments may compromise the coordination of alcohol and cannabis policies applies not only to the US but also to the 25 other countries with federalist systems, representing 40% of the world’s population . This study seeks to fill a gap in empirical policy research which is largely limited to mapping local variation in alcohol and cannabis policies separately, as if they occurred in isolation . To our knowledge, no research has compared and contrasted the approaches local governments take to alcohol and cannabis control, yet such assessments are essential to guide governments in understanding the public health implications of legalizing recreational cannabis.

To fill this gap, we conducted two analyses: First, we empirically evaluated similarities and differences in the approaches local governments have taken to alcohol and cannabis control by collecting, coding, and analyzing policy data for 241 cities and counties in California. Second, to inform hypotheses for future work concerning why local governments adopt similar or different approaches, we examined policy variation in relation to population data on the demographic, socioeconomic, political, and retail market characteristics of cities and counties. California legalized recreational cannabis in 2016 and is home to the largest legal cannabis market worldwide. The state has a long tradition of granting alcohol control authority to local governments and has taken a similar approach to cannabis control. This study focused on policies regulating recreational cannabis, because recreational legalization effectively dissolved the medical cannabis system, and because parallels with alcohol control are most relevant to recreational cannabis. Based on political learning theory, recommendations from public health experts, and prior US state and federal initiatives to model cannabis control policies after alcohol, we hypothesized that local cannabis control policies would be similar to local alcohol control policies both in their overall stringency and specific provisions .We collected and coded detailed data on local alcohol and cannabis control policies for 12 of California’s 58 counties and all incorporated cities within them, collectively covering 59% of the California population . Counties were manually selected to capture a range of local drug policies, population sizes, sociodemographic compositions, and political orientations, and to be consistent with an existing study of local substance use policies . City policies apply within incorporated city borders, and county policies apply to county areas outside of incorporated cities. Because San Francisco is a consolidated city-county, the final study covered 241 unique jurisdictions . The cannabis policy data collection is described elsewhere , and the alcohol policy data collection process was identical . Complete protocols and data collection instruments are provided in Appendices 1-4. Briefly, using a legal epidemiological approach , we systematically identified relevant legal text. We used structured data collection instruments to code the presence or absence and content of pre-specified policy provisions. All localities were coded separately by two analysts until achieving >95% agreement. Policy data collection and coding were conducted from July 2020 to January 2021. Measured policies correspond to those applicable at the time of data collection. California state law specifies which alcohol and cannabis policies apply statewide and which policy areas can be controlled by city and county governments. These differed for alcohol and cannabis. For cannabis, the state dictates the minimum age of legal use, establishes impaired driving prohibitions, manages licensing of cannabis businesses at all stages of the production chain, and sets minimum standards for product safety, packaging, and labeling. Local governments retain considerable discretion to dictate the number, type, and location of commercial cannabis businesses, hours and days of sale, types of products that may be sold, additional requirements for packaging and labeling, tax rates, and clear air laws. For alcohol, the state controls most aspects of production, retail licensing, pricing and taxation, impaired driving,commercial greenhouse supplies and underage prohibitions. However, local governments have the authority to regulate land use to protect health and welfare and thereby regulate the locations, density, and operations of retail outlets . Localities can also require responsible beverage service training, hold individuals civilly or criminally liable for hosting underage drinking on their property, limit advertising, and place restrictions on alcohol availability at special events such as concerts. Within the bounds of state law, the local policy measures we collected were guided by an established comprehensive schema of local alcohol policies in California , and a taxonomy of all possible cannabis policies . We focused on alcohol and cannabis control policies that regulate availability because these policies are a primary modifier of population-level consumption and associated health outcomes . For each substance, we coded all major categories of policies that: could be regulated at the local level according to state law, varied across California localities, were more restrictive than state law , and were plausibly related to public health given prior evidence, public health best practices, and research involving interviews with experts . Because the policy areas with local discretion differed for alcohol and cannabis, there were 37 policies fulfilling these criteria for alcohol and 20 policies fulfilling these criteria for cannabis.We developed two measures of alcohol and cannabis control policies: 1) an overall score capturing stringency and comprehensiveness of the local policy regime , and 2) specific policy provisions that could be applied to both alcohol and cannabis.

For alcohol, the overall stringency score was computed as the weighted sum of all 37 binary policies we collected, which covered: zoning and land use restrictions on the density, locations, and operations of outlets ; responsible server training requirements; prohibitions on hosting underage drinking; limitations on advertising; restrictions on alcohol availability at special events; and public drinking prohibitions. The weighting scheme was developed by Thomas and colleagues based on a systematic review of evidence on the strength of each policy in reducing alcohol-related harms, with weaker policies assigned a weight of 1 and stronger policies assigned a weight of 2. For the one policy with the negative weight, we inverted it so the absence of the policy received a weight of 1. For cannabis, the overall stringency score was based on all 20 policies we collected, which covered: restrictions on the density, locations, and operations of outlets; on-site consumption bans; responsible server training requirements; prohibitions on hosting underage consumption; limitations on advertising; restrictions on cannabis availability at special events; limits on product types, potency, packaging, and labelling; price controls; retail taxes; and personal cultivation practices . Absent evidence to guide a weighting scheme, the stringency score was calculated as an unweighted sum of the 20 binary policies; this approach assumes that all cannabis policies are equally important. Cannabis policy stringency scores were computed for the subset of localities that do not ban retail cannabis sales because most cannabis controls were not relevant in localities without retail sales. To compare the prevalence of individual provisions, we identified all those policies that could be applied both to alcohol and cannabis. Of the 37 alcohol policies and 20 cannabis policies we collected, only 9 policies overlapped, because the policy areas with local discretion differed for alcohol and cannabis. These 9 provisions were: limits on hours of sale; limits on advertising; responsible server training requirements; prohibitions on outlet over concentration; minimum distances between outlets and sensitive locations ; prohibitions on hosting underage consumption ; restrictions on special outdoor events; local permitting for retail sales ; and outlet safety requirements such as night lighting. While these local provisions can generally be applied to both alcohol and cannabis, there were some small discrepancies due to differences in the state regulatory frameworks for alcohol and cannabis. These are detailed in Appendix 5.We conducted two sets of analyses focused on: 1) similarities and differences in local governments’ approaches to alcohol and cannabis control, and 2) how different combinations of alcohol-cannabis control policies mapped on to the demographic, political, and retail market characteristics of localities. To assess similarities and differences in local governments’ approaches to alcohol and cannabis control, we scaled the alcohol and cannabis policy stringency scores to range between 0 and 100 and then compared them using unpaired t-tests and linear regression. Then we compared each of the 9 specific provisions using Cohen’s Kappa . To evaluate how different combinations of alcohol-cannabis control policies mapped on to the characteristics of localities, we first categorized local governments’ overall approaches to alcohol-cannabis regulation using latent class analysis . Briefly, LCA uses a set of observed variables to identify subgroups with similar characteristics. Because of our small sample size of 241 localities, we could not use all 57 measured alcohol and cannabis policies in the LCA. Instead, we prioritized including policies with the most variability across localities and tested the sensitivity of the results to varying the choice of included policies. Following recommended practice, wetested models ranging from 1 to 5 classes and selected the best-fit models by jointly considering substantive interpretability, Akaike Information Criterion , and Bayesian Information Criterion . We conducted separate LCA models for alcohol and cannabis policy variables, because models incorporating alcohol and cannabis policy variables simultaneously were dominated by the distinction between localities permitting versus banning retail cannabis businesses and failed to retain further nuance.