The first objective of this report was to determine whether recreational cannabis passage in November 2012 [28] and implementation in 2014 in Colorado, relative to two comparison states that had not yet endorsed recreational marijuana, was associated with increased rates of decline in prescription opioid use. The rationale for this hypothesis was a diverse evidence base resulting from basic [35] and epidemiological data [36] including patient reports [18], drug expenditures [16,17], and overdoses [12]. Table 1 provides some support for this hypothesis with greater reductions in six of the nine studied opioids relative to Utah or Maryland. However, the general temporal pattern in Figure 1, including parallel rates of decline in pain medications in Colorado with Maryland, does not support the hypothesis. Overall, this preliminary report based on only two comparison states provides inconclusive evidence. A recent report that examined the entire country with a more granular dataset of outpatient pharmacies discovered that recreational marijuana law implementation was associated with a 12% reduction in the MME and total days’ supply of prescription opioids [37]. This finding is congruent with a prior study limited to Medicaid [15] indicating that recreational and medical cannabis laws may be, at a population level, beneficial for reversing the earlier excesses in prescription opioids used for pain.

Perhaps a greater strength of this dataset was its detailed description of the changes in prescription opioid use in these three states. There were two key findings. Methadone by MME was far and away the predominant prescription opioid in Maryland. In 2012, methadone accounted for over half of the MME attributed to the eleven studied opioids. In contrast, methadone use was declining in Utah. The number of overdoses involving methadone increased fourteen-fold from 1991 to 2003 in Utah [38]. There was one NTP in 2017 in Maryland for every 74K population, compared with one per 239K in Utah and 244K in Colorado, a three-fold difference. These findings are consistent with and extend upon past OUD treatment research [3,5]. Earlier investigations have examined prescription opioid use patterns in Hawaii, Florida, and Texas [8,38]. Over three-quarters of the total MME in Puerto Rico was distributed by NTPs [8]. Unfortunately, NTPs are not included in most pharmacoepidemiological reports [15,37]. Second, there were appreciable variations in the use of hydrocodone. Hydrocodone and fentanyl in morphine equivalents in 2017 were equivalent in Utah. However, hydrocodone use was only one-third that of fentanyl in Maryland. Utah was ranked second in the US for per capita fentanyl use (1,650 ug) which was almost double that of Maryland (845 ug) or Colorado (894 ug) [4]. Unless there are regional differences in chronic or acute pain conditions, these sizable regional disparities may be inconsistent with evidence-based medicine and warrant further study.

There are some important limitations to this report and future directions. The US DEA’s Automation of Report and Consolidated Orders System (ARCOS) provides an inclusive dataset available for studying opioid distribution [29]. However, ARCOS is unable to account for opioids obtained through illegal or undetected means such as the Dark Web or opioids that cross state or national borders. Additionally, the choice of comparative states may limit generalizability. Colorado and Maryland share similar demographics in terms of population size, home ownership, education, and uninsured rates. However, there is an appreciable difference in the percentage of non-white people living in Colorado (12.7%) versus Maryland (41.0%) (Table 2). In the past two decades (2000–2015), non-Hispanic black people and Hispanics have displayed low mortality rates involving opioid overdose [39]. Future studies may need to take these factors into consideration to account for any differences. Comparison of other states that legalized the recreational use of marijuana during the same time period would also be beneficial. Washington (Initiative 502, 2012) and Oregon (Measure 91, 2014) may serve as comparative states with their similar legalization timelines and demographics [40,41]. Finally, there have been several national and state policy changes which has resulted in reductions in prescription opioid use since the national peak in 2011 [2,3]. Naturalistic studies like this are extremely challenging to interpret as there are multiple demographic and socioeconomic factors which contribute to both prescription opioid and marijuana use [6,7,9]. There may be no single state, or states, that perfectly matches the characteristics of Colorado (Table 2). Future research with more states could employ a multivariate approach in an attempt to statistically account for these differences. A detailed characterization of the temporal changes in these three states, which together have a population of 14.9 million, may have more value than any inference, however tentative, about public policy that can be made with cross-state comparisons. It would also be important and influential to consider the overall health status of Colorado and the comparison states. This would allow for improved comparison of the influence of marijuana legalization.