Development and evaluation of a method to compile a national list of cannabis dispensaries.

Abstract

Understanding how retail availability impacts cannabis attitudes and use requires data on the geographic distribution of cannabis dispensaries. However, with no single comprehensive source for dispensary locations, accurately mapping dispensaries presents unique challenges. This study's purpose was to: 1) describe a systematic process for identifying dispensaries that can be applied nationwide using two unexplored resources-RefUSA and Google Maps; 2) compare the accuracy of these resources with three commonly used sources of location data - state regulatory lists, Weedmaps, and Potguide; and 3) examine whether accuracy varies by urbanicity. We obtained dispensary lists from Google Maps, Weedmaps, Potguide, RefUSA, and state regulatory data in three zip codes in every state that had legal recreational cannabis marketplaces, calling the dispensaries to confirm that they were operational. We present sensitivity and positive predictive value (PPV) by list and by Rural-Urban Commuting Area Codes (RUCAs), using the call-verified dispensary list as the gold standard. Google Maps identified the greatest number of dispensaries (n = 264), had the best sensitivity of all lists (96.0 %; 95 % CI: 92.6, 99.2), and yielded the highest proportion of potentially unlicensed dispensaries (29 %). There were no clear trends in sensitivity or PPV by RUCA. Removing listings that Google Maps marked as closed or that lacked phone numbers increased PPV from 82.2 % (95 % CI: 77.0, 86.6) to 96.4 % (95 % CI: 93.1, 98.5) without decreasing sensitivity. Google Maps currently offers the most accurate source of dispensary data for cross sectional analyses and performs equally well in urban and rural areas.