Recreational and Medical Cannabis Legalization and Opioid Prescriptions and Mortality
School of Pharmacy, Memorial University of Newfoundland, St John’s, Newfoundland & Labrador, Canada
Division of Health Policy and Economics, Weill Cornell Medicine, New York, New York
College of Pharmacy, University of Manitoba, Winnipeg, Manitoba, Canada
Center for Drug Safety and Effectiveness, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland
Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland
Division of General Internal Medicine, Johns Hopkins Medicine, Baltimore, Maryland
Key Points
Question
What is the association of recreational and medical cannabis legalization with opioid prescriptions and fatal overdoses in the US?
Findings
In this cohort study using state-level data and a generalized difference-in-differences method that accounted for possible contamination from multiple laws, there was no discernible association found between cannabis laws and opioid prescriptions nor fatal opioid overdose, although the results suggested a potential reduction in synthetic opioid deaths associated with recreational cannabis laws. These results were robust to excluding state economic indicators, accounting for additional opioid laws and using alternative ways to code treatment dates.
Meaning
The study results suggest that recreational and medical cannabis legalization were not associated with significant increases or decreases in opioid prescriptions and fatal overdose with the exception of a possible reduction in synthetic opioid deaths that was associated with recreational cannabis law implementation.
Abstract
Importance
While some have argued that cannabis legalization has helped to reduce opioid-related morbidity and mortality in the US, evidence has been mixed. Moreover, existing studies did not account for biases that could arise when policy effects vary over time or across states or when multiple policies are assessed at the same time, as in the case of recreational and medical cannabis legalization.
Objective
To quantify changes in opioid prescriptions and opioid overdose deaths associated with recreational and medical cannabis legalization in the US.
Design, Setting, and Participants
This quasiexperimental, generalized difference-in-differences analysis used annual state-level data between January 2006 and December 2020 to compare states that legalized recreational or medical cannabis vs those that did not.
Intervention
Recreational and medical cannabis law implementation (proxied by recreational and medical cannabis dispensary openings) between 2006 and 2020 across US states.
Main Outcomes and Measures
Opioid prescription rates per 100 persons and opioid overdose deaths per 100 000 population based on data from the US Centers for Disease Control and Prevention.
Results
Between 2006 and 2020, 13 states legalized recreational cannabis and 23 states legalized medical cannabis. There was no statistically significant association of recreational or medical cannabis laws with opioid prescriptions or overall opioid overdose mortality across the 15-year study period, although the results also suggested a potential reduction in synthetic opioid deaths associated with recreational cannabis laws (4.9 fewer deaths per 100 000 population; 95% CI, −9.49 to −0.30; P = .04). Sensitivity analyses excluding state economic indicators, accounting for additional opioid laws and using alternative ways to code treatment dates yielded substantively similar results, suggesting the absence of statistically significant associations between cannabis laws and the outcomes of interest during the full study period.
Conclusions and Relevance
The results of this study suggest that, after accounting for biases due to possible heterogeneous effects and simultaneous assessment of recreational and medical cannabis legalization, the implementation of recreational or medical cannabis laws was not associated with opioid prescriptions or opioid mortality, with the exception of a possible reduction in synthetic opioid deaths associated with recreational cannabis law implementation.
Teaser
This cohort study examines changes in opioid prescriptions and opioid overdose deaths associated with recreational and medical cannabis legalization in the US.
Article notes
Untitled section
Received 2023 Jul 11; Accepted 2023 Nov 18; Collection date 2024 Jan.
Introduction
Despite the efforts of many parties, the opioid epidemic persists in the US, with more individuals experiencing fatal overdoses during 2021 than any year on record.1 Continued opioid-related harms have generated interest in potential mechanisms to ameliorate the crisis, including the legalization of cannabis for recreational or medical use. For example, it is theoretically possible that individuals with acute or chronic pain may substitute cannabis for prescription opioids, thereby reducing opioid use and opioid-related harms.2,3 Greater availability of cannabis, a substance with a much lower risk profile of overdose than opioids, may also be associated with reduced initiation of prescription or illicit opioids. However, many clinical and economic barriers may prevent the substitution of cannabis for prescription opioids or the translation of such substitution into reduction in downstream harms.4,5,6,7 There are also concerns that cannabis may serve as a gateway to illegal drug use.8
Empirical work examining the association between cannabis legalization and opioid-related outcomes in the US has yielded mixed conclusions. Some evidence suggests that medical cannabis legalization has been associated with reductions in opioid prescribing,9,10,11 and in some cases, opioid-related mortality,12 whereas other evidence suggests that medical cannabis legalization was not associated with changes in opioid prescriptions13 or has been associated with increased, rather than decreased, opioid-related mortality.14,15,16,17 Studies examining the effect of recreational cannabis laws have found reductions in opioid prescriptions associated with recreational cannabis legalization,18,19,20,21,22 and evidence on the effects of recreational cannabis laws is mixed, with studies indicating either no association with or an increase in opioid mortality.14,15,16
Apart from finding mixed results on opioid overdose deaths, the current literature has several limitations. Most studies use traditional difference-in-differences (DD) analyses to estimate the effects of legalization. Recent advances in the DD literature indicate that estimates of policy effects in these analyses may be contaminated when policy effects vary over time or across states.23 Moreover, the bias can be compounded when multiple policies adopted in a staggered fashion are assessed at the same time, as in the case of medical and recreational cannabis laws. While some studies have sought to address the issue of heterogeneous policy effects by conducting stacked regressions14 or considering a subset of states or time periods,15 to our knowledge, no study has rigorously accounted for biases that could arise from evaluating medical and recreational cannabis law implementation at the same time. Also, most studies examine the associations with opioid prescriptions in specific populations (eg, Medicaid enrollees or individuals with employer insurance). In this study, we quantified the association of recreational and medical cannabis law implementation with opioid prescriptions and mortality in the US using national data and a novel analytic approach that overcame the limitations of previous studies including possible biases due to heterogeneous policy effects and simultaneous evaluation of medical and recreational cannabis laws.23
Methods
Study Design
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. We used publicly available, state-level aggregate data. Thus, ethics approval was not required based on Newfoundland and Labrador’s Health Research Ethics Board guidelines.
Estimating the effects of state cannabis law implementation entails 2 challenges. First, different states adopted these laws at varied points (eTable 1 in Supplement 1). Recent literature, including studies by Callaway and Sant’Anna24 and de Chaisemartin and d’Haultfoeuille,25 has shown that treatment effects (overall and dynamic effects over time) estimated using traditional DD methods may be biased if the laws are implemented in a staggered fashion and the effects of the laws vary across states or over time (for example, as implementation ramps up).24,25,26,27,28 Second, states implemented multiple laws that could be associated with opioid outcomes, including not only recreational and medical cannabis laws but also other laws designed to curb opioid prescriptions and overdose. As a result, estimates of 1 law’s effect could be contaminated by that of others.
To address these challenges, we used a generalized DD design recently developed by de Chaisemartin and d’Haultfoeuille23 to test the association between state cannabis laws and opioid-related outcomes. This method was suitable for estimating the effects of a staggered intervention while accounting for other interventions that could be associated with the outcomes of interest. Like other methods, this method involves comparing outcome changes in treated states (states that implemented the law) with changes in outcomes in control states, but it carefully defines and selects valid control states for these comparisons. Specifically, it uses only control states that had not yet implemented the law but had the same other laws as treated states during the baseline (defined as the year before the cannabis law was implemented). Further details on this method, its underlying assumptions, and its advantages vs alternative methods to estimate effects of multiple treatments are provided in the eMethods in Supplement 1.
Treatment and Control Assignment
A state was considered treated in a period if it had cannabis dispensaries operating during that period (eFigure 1 and eTable 1 in Supplement 1). Dates of dispensary openings were obtained from Mathur and Ruhm.14 As outcome data were available at the yearly level, we constructed yearly indicators for recreational and medical cannabis dispensary openings. If the dispensaries were opened during the first half of a year, we coded the indicator as 1 for that year (and subsequent years) and 0 otherwise. We tested the sensitivity of our results to this coding (as described later).
For each treated state with a cannabis law, we identified control states as those that had not yet implemented the cannabis law and had the same confounding opioid laws as the treated state during the year before cannabis law implementation. We considered 3 potentially confounding opioid laws, namely mandatory prescription drug monitoring program laws (that require prescribers to check the prescription drug monitoring program database before prescribing an opioid), Good Samaritan laws (that provide legal protection to individuals calling for help in the event of an overdose), and naloxone access laws (including those allowing standing orders in which prescribers may authorize pharmacists to dispense naloxone without an outside prescription or permitting first responders to carry naloxone).
As an example, Colorado implemented a medical cannabis law in 2010, a Good Samaritan law in 2012, a naloxone access law in 2013, and a recreational cannabis law in 2014. To estimate changes in outcomes associated with recreational cannabis laws in Colorado, this method used the period after 2013 during which there was only 1 legislative change, the recreational cannabis law in 2014, and compared outcome changes in Colorado with outcome changes in control states (eg, Washington) that had not yet implemented recreational cannabis laws but, up to 2013, already had implemented medical cannabis laws and adopted a Good Samaritan law and a naloxone access law like Colorado. The list of control states for each treated state is provided in eTable 2 in the Supplement 1.
Data Sources, Study Period, and Outcomes
We used publicly available data from the US Centers for Disease Control and Prevention (CDC). Opioid prescription rates were based on the IQVIA Xponent database, which collects data from more than 50 000 retail pharmacies, accounting for more than 90% of all retail prescriptions across the US.30 Data on the number of opioid overdose deaths were derived from the National Vital Statistics Multiple Cause of Death files (International Statistical Classification of Diseases and Related Health Problem [ICD], Tenth Revision underlying cause-of-death codes X40–X44, X60–X64, X85, and Y10–Y14 and multiple cause of death codes T40.1-T40.4) and available from the public-access CDC Wide-Ranging Online Data for Epidemiological Research (CDC-WONDER) online database.29 We focused on the period from January 2006 through December 2020 due to the availability of data on opioid prescriptions during that window.
The outcomes of interest were the number of opioid prescriptions (per 100 persons) and number of opioid overdose deaths (per 100 000 population), calculated at state-year level for all 50 US states and Washington, DC. Prescription opioids included buprenorphine (except products to treat opioid use disorder), codeine, fentanyl, hydrocodone, hydromorphone, methadone, morphine, oxycodone, oxymorphone, propoxyphene, tapentadol, and tramadol.30
Statistical Analysis
We implemented the approach of de Chaisemartin and d’Haultfoeuille23 using regression analysis. All regressions controlled for 2 state-level economic indicators (ie, state poverty rates and real gross domestic product). All regressions also included state indicators to control for time-invariant state-level characteristics and year indicators to control for secular changes or shocks in outcomes that are common to all states. Detailed descriptions of the variables in the regression are provided in eTable 3 in Supplement 1.
We modeled all outcomes using linear regressions. As the method of de Chaisemartin and d’Haultfoeuille23 evaluates only 1 policy at a time, the effects of medical and recreational cannabis laws were estimated in separate regressions. Also, the opioid policies were used only to identify valid control states for the analysis; thus, their effects were not estimated by the regressions. For comparison, we also used traditional DD analyses to examine the association of cannabis law implementation with opioid outcomes. In addition to examining the association of cannabis law implementation with overall opioid overdose mortality, we conducted subgroup analyses in which we assessed these associations with opioid mortality by the type of opioid involved in overdose.
We also conducted additional analyses to examine the robustness of the results. First, we examined the sensitivity of our results to the exclusion of time-varying state-level economic indicators. Second, we coded the treatment exposure indicators as 1 for the whole year if the legalization came into effect anytime during the first 3 quarters of that year (instead of anytime during the first half of the year in the main analysis) and 0 otherwise. Third, we accounted for additional opioid laws, namely the prescription limit laws that restricted the number of days that clinicians dispensed opioids for acute pain and pill mill laws. Fourth, we restricted the analysis period to 2011 to 2020 as the nature of the opioid crisis changed around 2011 to 2012 when prescription opioids started to decline rapidly and as opioid overdose deaths started growing at an accelerated pace around 2014 to 2015 (which was driven by heroin and synthetic opioids). Finally, to assess the potential validity of the parallel trends assumption, we examined the prepolicy trends in treated vs control groups using event study based on the approach of de Chaisemartin and d’Haultfoeuille23 and the traditional DD analyses. All analyses were performed with Stata, version 17 (StataCorp). We implemented the method of de Chaisemartin and d’Haultfoeuille23 using Stata command did_multiplegt. All tests were 2-sided, and a 5% significance level was used.
Results
A plot of unadjusted differences in outcomes between states with and without recreational or medical cannabis laws against the time since opening of cannabis dispensaries is shown in eFigure 2 in Supplement 1. This plot suggested no association of cannabis law implementation with opioid outcomes in most states.
Table 1 presents the adjusted associations of interest. Each entry in this panel is from a separate regression model. The changes in prescribed opioids associated with recreational cannabis law implementation were not statistically significant (3.08 fewer prescriptions per 100 persons; P = .17), with the 95% CI ranging from a decrease of 7.43 prescriptions (or a 10% decrease compared with the annual average of 73.4 prescriptions across all states during the study period) to an increase of 1.27 prescriptions (or a 2% increase). Neither were the changes in opioid overdose mortality (3.05 fewer deaths per 100 000 population; P = .24) statistically significant, ranging from a decrease of −8.18 deaths (or a 78% decline compared with the annual average of 10.5 deaths across all states during the study period) to an increase of 2.07 deaths (or a 20% increase). The changes in the outcomes associated with medical cannabis law implementation, while larger in magnitude than those for recreational cannabis law implementation, were also not statistically significant at 3.54 additional prescriptions per 100 persons (95% CI, −1.49 to 8.57; P = .17) and 3.09 additional deaths per 100 000 population (95% CI, −0.26 to 6.44; P = .07).
| Laws | Change in opioid prescriptions per 100 persons (95% CI) | P value | Change in opioid overdose deaths per 100 000 population (95% CI) | P value | No. | |
|---|---|---|---|---|---|---|
| Treated states | Control statesb | |||||
| Recreational cannabis | −3.08 (−7.43 to 1.27) | .17 | −3.05 (−8.18 to 2.07) | .24 | 7 | 22 |
| Medical cannabis | 3.54 (−1.49 to 8.57) | .17 | 3.09 (−0.26 to 6.44) | .07 | 15 | 47 |
The results for opioid prescriptions using the method of de Chaisemartin and d’Haultfoueille23 were broadly consistent with those from traditional DD analyses (eTable 4 in Supplement 1). For opioid mortality, the results were also consistent for medical cannabis laws. However, for recreational cannabis laws, the estimates were large, negative, and statistically significant using the traditional DD method, but smaller and no longer statistically significant using the method of de Chaisemartin and d’Haultfoeuille.23
The sensitivity analyses showed that the results for recreational and medical cannabis legalization were robust to the exclusion of time-varying state economic indicators (eTable 5 in Supplement 1). The results were also similar to the main analysis when we coded the treatment exposure variable as 1 for the whole year if the laws were implemented anytime during the first 3 quarters of that year and accounted for additional opioid laws. When we restricted the analysis period to 2011 to 2020, the results were similar for recreational cannabis law (ie, recreational cannabis law implementation was not associated with changes in opioid prescriptions or opioid mortality), but there was a statistically significant increase in opioid prescriptions associated with medical cannabis laws. However, among states that legalized medical cannabis and opened medical cannabis dispensary after 2011, many did so shortly after 2011, making the prepolicy period for these states fairly short for the DD analysis. The event study plots using the approach of de Chaisemartin and d’Haultfoueille23 (Figure 1 and Figure 2) and traditional DD analyses (eFigure 3 in Supplement 1) indicated that, except for a slight difference in prepolicy trends between treated and control states for the association between recreational cannabis laws and opioid mortality, the parallel trend assumption was satisfied in our DD analyses.
Table 2 presents the association of law implementation with opioid mortality by the type of opioid overdose. We observed a statistically significant decline in deaths due to synthetic opioids (4.9 fewer deaths; 95% CI, −9.49 to −0.30; P = .04) associated with recreational cannabis law implementation. Meanwhile, there were no statistically significant changes in mortality due to heroin or prescriptions opioids associated with cannabis laws. Figure 1 and Figure 2 also show the dynamic changes associated with cannabis law implementation. The null associations of cannabis laws with opioid prescriptions and opioid mortality did not change over time. However, implementation of recreational cannabis laws was associated with non–statistically significant gradual decreases in mortality due to synthetic opioids during the first 5 years of implementation. Meanwhile, there were no clear trends in changes in mortality due to heroin or prescription opioids associated with cannabis laws.
| Drug | Change in opioid overdose deaths per 100 000 population (95% CI) | P value |
|---|---|---|
| Heroin | ||
| Recreational cannabis laws | 0.18 (−2.63 to 2.98) | .90 |
| Medical cannabis laws | 0.01 (−0.91 to 0.93) | .98 |
| Prescription opioids | ||
| Recreational cannabis laws | 0.02 (−1.65 to 1.68) | .98 |
| Medical cannabis laws | 1.07 (−0.08 to 2.22) | .07 |
| Synthetic opioids | ||
| Recreational cannabis laws | −4.90 (−9.49 to −0.30) | .04 |
| Medical cannabis laws | 2.05 (−2.11 to 6.22) | .33 |
Discussion
While some have argued that cannabis legalization has helped to reduce opioid-related morbidity and mortality, evidence is mixed, and several studies have been subject to potential bias. We used a recently developed generalized DD approach to quantify the association between recreational and medical cannabis laws and opioid prescriptions and mortality in the US between 2006 and 2020. After accounting for time-varying state-level economic indicators and state opioid laws, we found no evidence that implementation of state recreational or medical cannabis laws was associated with changes in opioid prescriptions or overdose deaths. However, there was a statistically significant reduction in overdose deaths due to synthetic opioids associated with recreational cannabis laws, and this reduction appeared to increase gradually during the first 5 years after law implementation.
These results contrast with recent studies that suggested that recreational and medical cannabis legalization are associated with reductions in opioid prescriptions21,22 and medical cannabis legalization is associated with an increase in opioid mortality.14,15,16,17 These conflicting findings may arise from many factors, including differences in the study cohorts and periods examined, methods of defining the exposures and outcomes of interest, and empirical approaches used to evaluate these laws.
We did not observe statistically significant reductions in opioid prescriptions or overall opioid mortality associated with cannabis laws in any of the analyses we performed during the entire study period. This result was consistent with reports of physicians’ reluctance to recommend cannabis4,5 due to the lack of clinical guidelines and insufficient information on efficacy and adverse effects of cannabis6 and high economic costs of switching from opioids to cannabis.7 However, there was an increase in overdose deaths involving prescription opioids between years 2 and 4 after medical cannabis law implementation, suggesting potential co-use of cannabis and opioids that was not always under clinical supervision. Meanwhile, our finding indicating a possible association between recreational cannabis laws and reduced synthetic opioid deaths may suggest that users of fentanyl (and other synthetic opioids) switched to recreational cannabis or reduced use of other sedating substances, such as benzodiazepines, that can increase overdose risk when mixed with fentanyl.17,31 It is also possible that recreational cannabis laws make cannabis more available and thus reduce initiation of opioid use. However, this finding should be treated cautiously given that the quality of data on specific types of opioid deaths varies across states and over time32 and there were fewer treated states in years 2 to 5 in the analysis.
Limitations
Our study had several limitations. First, our analysis used aggregated data at the state-year level, rendering us unable to observe changes in opioid use within individual patients over time or to examine the estimated effects of laws on subgroups of patients. We were also unable to code legalization dates at a more granular month level. However, our sensitivity analyses indicated that the results were robust to different ways of coding the treatment dates. Second, our intervention date captured only the opening of the first (recreational or medical) cannabis dispensary. We were unable to account for number of dispensaries in a specific state owing to lack of availability of longitudinal dispensary data spanning the study period. Lastly, the study data captured only opioids prescribed in outpatient setting; thus, we were unable to shed light on changes in opioid use in hospital and emergency department settings after cannabis legalization.
Conclusions
More than 2 decades after the opioid epidemic began in the US, more individuals are dying of opioid overdose than ever before. This ongoing morbidity and mortality has heightened interest in how policy interventions, including those governing recreational and medical cannabis, may intersect with the opioid epidemic. Despite this, evidence to date regarding these matters has been highly mixed. In our analyses accounting for the staggered implementation of cannabis laws in a dynamic clinical and policy environment, we found no evidence that the implementation of recreational or medical cannabis laws was associated with opioid prescriptions or opioid mortality, with the exception of a possible reduction in synthetic opioid deaths associated with recreational cannabis law implementation.
Untitled section
References
Untitled section
References
- 1.National Institute on Drug Abuse . Overdose death rates. Accessed June 17, 2022. https://nida.nih.gov/research-topics/trends-statistics/overdose-death-rates
- 2.Pritchett CE, Flynn H, Wang Y, Polston JE. Medical cannabis patients report improvements in health functioning and reductions in opiate use. Subst Use Misuse. 2022;57(13):1883-1892. doi: 10.1080/10826084.2022.2107673
- 3.Lucas P, Baron EP, Jikomes N. Medical cannabis patterns of use and substitution for opioids & other pharmaceutical drugs, alcohol, tobacco, and illicit substances; results from a cross-sectional survey of authorized patients. Harm Reduct J. 2019;16(1):9. doi: 10.1186/s12954-019-0278-6
- 4.Healio . Medical cannabis data remain limited, leaving physicians reluctant to prescribe it. Accessed October 3, 2022. https://www.healio.com/news/primary-care/20210623/medical-cannabis-data-remain-limited-leaving-physicians-reluctant-to-prescribe-it
- 5.Evanoff AB, Quan T, Dufault C, Awad M, Bierut LJ. Physicians-in-training are not prepared to prescribe medical marijuana. Drug Alcohol Depend. 2017;180:151-155. doi: 10.1016/j.drugalcdep.2017.08.010
- 6.Cooke AC, Knight KR, Miaskowski C. Patients’ and clinicians’ perspectives of co-use of cannabis and opioids for chronic non-cancer pain management in primary care. Int J Drug Policy. 2019;63:23-28. doi: 10.1016/j.drugpo.2018.09.002
- 7.NBC . “Ridiculous” price of medical marijuana leaves patients scrambling. Accessed June 21, 2022. https://www.nbcnews.com/news/us-news/ridiculous-price-medical-marijuana-leaves-patients-scrambling-n1274085
- 8.BBC News . Cannabis legalisation: does it lead to harder drug use? Accessed October 4, 2023. https://www.bbc.com/news/world-us-canada-50494610
- 9.Bradford AC, Bradford WD. Medical marijuana laws reduce prescription medication use in Medicare Part D. Health Aff (Millwood). 2016;35(7):1230-1236. doi: 10.1377/hlthaff.2015.1661
- 10.Bradford AC, Bradford WD. Medical marijuana laws may be associated with a decline in the number of prescriptions for Medicaid enrollees. Health Aff (Millwood). 2017;36(5):945-951. doi: 10.1377/hlthaff.2016.1135
- 11.Bradford AC, Bradford WD. The impact of medical cannabis legalization on prescription medication use and costs under Medicare Part D. J Law Econ. 2018;61(3):461-487. doi: 10.1086/699620
- 12.Bachhuber MA, Saloner B, Cunningham CO, Barry CL. Medical cannabis laws and opioid analgesic overdose mortality in the United States, 1999-2010. JAMA Intern Med. 2014;174(10):1668-1673. doi: 10.1001/jamainternmed.2014.4005
- 13.McGinty EE, Tormohlen KN, Seewald NJ, et al. Effects of U.S. state medical cannabis laws on treatment of chronic noncancer pain. Ann Intern Med. 2023;176(7):904-912. doi: 10.7326/M23-0053
- 14.Mathur NK, Ruhm CJ. Marijuana legalization and opioid deaths. J Health Econ. 2023;88:102728. doi: 10.1016/j.jhealeco.2023.102728
- 15.Chan NW, Burkhardt J, Flyr M. The effects of recreational marijuana legalization and dispensing on opioid mortality. Econ Inq. 2020;58(2):589-606. doi: 10.1111/ecin.12819
- 16.Shover CL, Davis CS, Gordon SC, Humphreys K. Association between medical cannabis laws and opioid overdose mortality has reversed over time. Proc Natl Acad Sci U S A. 2019;116(26):12624-12626. doi: 10.1073/pnas.1903434116
- 17.Castillo-Carniglia A, Rivera-Aguirre A, Santaella-Tenorio J, et al. Changes in opioid and benzodiazepine poisoning deaths after cannabis legalization in the US: a county-level analysis, 2002-2020. Epidemiology. 2023;34(4):467-475. doi: 10.1097/EDE.0000000000001609
- 18.Shi Y, Liang D, Bao Y, An R, Wallace MS, Grant I. Recreational marijuana legalization and prescription opioids received by Medicaid enrollees. Drug Alcohol Depend. 2019;194:13-19. doi: 10.1016/j.drugalcdep.2018.09.016
- 19.Raman S, Bradford AC. Recreational cannabis legalizations associated with reductions in prescription drug utilization among Medicaid enrollees. Health Econ. 2022;31(7):1513-1521. Published online 2022. doi: 10.1002/hec.4519
- 20.Wen H, Hockenberry JM. Association of medical and adult-use marijuana laws with opioid prescribing for Medicaid enrollees. JAMA Intern Med. 2018;178(5):673-679. doi: 10.1001/jamainternmed.2018.1007
- 21.McMichael BJ, Van Horn RL, Viscusi WK. The impact of cannabis access laws on opioid prescribing. J Health Econ. 2020;69:102273. doi: 10.1016/j.jhealeco.2019.102273
- 22.Wen J, Wen H, Butler JS, Talbert JC. The impact of medical and recreational marijuana laws on opioid prescribing in employer-sponsored health insurance. Health Econ. 2021;30(5):989-1000. doi: 10.1002/hec.4237
- 23.de Chaisemartin C, D’Haultfoeuille X. Two-way fixed effects and differences-in-differences estimators with several treatments. J Econom. 2023;236(2). doi: 10.1016/j.jeconom.2023.105480
- 24.Callaway B, Sant’Anna PH. Difference-in-differences with multiple time periods. J Econom. 2021;225(2):200-230. doi: 10.1016/j.jeconom.2020.12.001
- 25.De Chaisemartin C, d’Haultfoeuille X. Two-way fixed effects estimators with heterogeneous treatment effects. Am Econ Rev. 2020;110(9):2964-2996. doi: 10.1257/aer.20181169
- 26.Roth J, Sant’Anna PH, Bilinski A, Poe J. What’s trending in difference-in-differences? a synthesis of the recent econometrics literature. ArXiv. Published online January 4, 2022. doi: 10.48550/arXiv.2201.01194
- 27.Sun L, Abraham S. Estimating dynamic treatment effects in event studies with heterogeneous treatment effects. J Econom. 2021;225(2):175-199. doi: 10.1016/j.jeconom.2020.09.006
- 28.Goodman-Bacon A. Difference-in-differences with variation in treatment timing. J Econom. 2021;225(2):254-277. doi: 10.1016/j.jeconom.2021.03.014
- 29.US Centers for Disease Control and Prevention . CDC WONDER. Accessed July 16, 2022. https://wonder.cdc.gov/
- 30.US Centers for Disease Control and Prevention . U.S. opioid dispensing rate maps. Accessed December 30, 2022. https://www.cdc.gov/drugoverdose/rxrate-maps/index.html
- 31.Pardo B. Insights into mixing fentanyl and benzodiazepines from Canadian drug seizures. JAMA Psychiatry. 2022;79(1):81-83. doi: 10.1001/jamapsychiatry.2021.3292
- 32.National Center for Health Statistics . Provisional drug overdose data. Accessed January 5, 2023. https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm