The link between medical cannabis and lower levels of opioid overdose deaths — identified previously in several studies — is more complex than previously labelled and appears to be changing as both medical marijuana laws and the opioid crisis evolve.

Authorizing medical cannabis was associated with lower levels of opioid deaths only in states that had provisions for dispensaries that made medical cannabis easily available to patients. Opioid death rates were not lower in states that just provided legal protections to patients and caregivers, allowing them to grow their own cannabis.

In addition, the association between medical cannabis dispensaries and less opioid deaths appears to have declined sharply after 2010, when states began to constrict requirements on sales by dispensaries.

Researchers examined information about treatment admissions for addiction to pain medications from 1999 to 2012 and state-level overdose deaths from opioids from 1999 to 2013. They also recognized state laws legalizing medical cannabis, examining provisions such as whether the regulations made cannabis easily accessible to patients by allowing dispensaries.

The scientists deliver two explanations for the weakening in the association between medical cannabis dispensaries and opioid harm. First, states that more recently adopted laws with medical cannabis dispensaries more tightly regulated them, in response to a U.S. Justice Department memo saying it would not challenge state-level medical marijuana laws so long as dispensary sales were in full compliance with state regulations. Second, beginning in 2010, the primary driver of the opioid crisis and related deaths became illicit opioids, mainly heroin and then fentanyl, not prescription opioids.

The study also found no indication that states with medical cannabis laws experience reductions in the volume of legally distributed opioid analgesics used to treat pain. Even if medical marijuana patients were substituting medical cannabis for opioids in medical cannabis states, these patients did not represent a measurable part of the medical opioid analgesic market.