Association for Public Policy Analysis & Management

In the past 20 years the drive to legalize medical marijuana has gained national attention with the public and policy makers. Beginning in 1996 states started implementing medical marijuana laws (MMLs), and by the summer of 2014, 23 states and the District of Columbia had adopted laws legalizing marijuana use for medical purposes.  Marijuana is used clinically to treat a variety of conditions, including chronic pain, nausea, wasting syndrome, glaucoma, acquired immune deficiency syndrome, and side effects from cancer treatment. There is significant variation between state policies surrounding medical marijuana; for example, states that have adopted some form of a MML more recently have been relatively more restrictive in the range of conditions for which physicians may legally prescribe the substance.  While MMLs have attracted a great deal of popular and academic attention, there is relatively little agreement about the impact of the laws. To date, most studies have focused on whether there are negative externalities, in terms of increases in other illegal drug use, traffic accidents and fatalities, or adolescent educational outcomes.

One issue that has received surprisingly little attention is the question of whether medical marijuana is actually being used clinically to any significant degree.  To the extent that marijuana is used by physicians to manage the conditions for which it has clinical evidence, then one would expect it to be used primarily as a substitute for existing prescription medications - for patients who did not respond to prior therapy, or who respond better to marijuana.[1] Nonetheless, there are no published studies that investigate whether the introduction of medical marijuana changes the prescribing patterns of FDA-approved prescription drugs used to treat the conditions listed above.

In this paper, we use data on all prescriptions filled by Medicare Part D enrollees in the U.S. from 2010 to 2012.  The data is compiled at the physician level, and is merged to detailed geographic health care related data.  In addition, we use a unique data source to identify which prescription drugs are used on- and off-label for the conditions in question.  We use a series of difference-in-difference models to determine whether physicians operating in states with an MML have different overall, on-label, and off-label prescribing patterns for Medicare Part D enrollees than their counterparts in non-MML states.  Finally, we discuss the implications for patient welfare and the social impact of medical marijuana policies.