Association for Public Policy Analysis & Management

Over the past two decades, there has been a dramatic rise in opioid overdose deaths associated with prescription opioids and heroin in the United States. In 2016, opioids were involved in a record 42,249deaths. In addition, rates of opioid use disorder and admissions for opioid use disorder treatment are rising. Although not often discussed in the context of the opioid epidemic, the Medicare population is experiencing high rates of opioid use disorder and opioid prescribing. In 2016, one in three Medicare Part D beneficiaries received at least one opioid prescription and approximately 500,000 beneficiaries received high amounts of opioids.

Concurrent to the opioid epidemic, 29 states and the District of Columbia have implemented medical cannabis legislation (MCL) allowing physicians to recommend marijuana to patients if they present with a set of defined illnesses. Recent evidence suggests both a substitution away from opioid pain medications and a reduction in opioid-overdose deaths in states with MCLs in place. However, little is know about the potential influence of MCL implementation on the availability of treatment for opioid use disorder.

We use data from Medicare Part D (2010-2015) to estimate a series of difference-in-differences regression models to examine the relationship of MML implementation to availability of buprenorphine and naltrexone at the county-level. Availability of opioid use disorder medications is measured by the number of physicians prescribing buprenorphine and naltrexone, number of daily doses per physician, and the total number of daily doses of each medication in the county. We estimate ordinary least squares regression models with county fixed effects separately for each outcome variable. Multivariate models control for county-level demographics, county-level opioid overdose death rate, and physician specialty.

Results indicated a significantly greater number of physicians prescribing buprenorphine and oral naltrexone in counties with an MCL in effect. There were also significantly more total daily doses of both medications filled in counties with an MCL effect, compared to counties with no MCL. There was no relationship between MCL implementation and the number of daily doses per physician.

This study finds increased capacity to treat opioid use disorder in counties with an MCL in effect. Results add to the growing body of evidence suggesting that MCLs may be a valuable policy tool to address the opioid epidemic, particularly among Medicare Part D beneficiaries who experience high rates of opioid prescribing and opioid use disorder.