Medical marijuana dispensaries decrease opiate harm, but only if they're not too heavily regulated

In a new paper by David Powell (RAND), Rosalie Liccardo Pacula (RAND), and Mireille Jacobson (University of California, Irvine) published in the Journal of Health Economics (ungated earlier version here) caught my attention this week, not least because it continues the theme of this week's posts (see here and here and here). In the paper, the authors investigate the impact of medical marijuana laws on opioid death rates, treatment admissions, the volume of legally distributed opioids, and non-medical use of pain relievers. In case you're unaware, opioid use has become a serious (and deadly) epidemic in the U.S. over the last decade. Figure 1 from the Powell et al. paper paints the picture, with opioid distribution and harms (treatment admissions and mortality) all climbing substantially since the late 1990s:

In their analysis, Powell et al. use data from the U.S. over the period from 1999 to 2013. Over this period, some states introduced medical marijuana laws, while others did not. The key difference between this study and earlier studies that have looked at similar questions is that this study correctly recognises that medical marijuana laws by themselves are unlikely to have any effect, so they look at changes when medical marijuana dispensaries became operative as well (rather than just when the State medical marijuana legislation passed).

They find:

...fairly strong and consistent evidence using difference-in-differences and event study methods that states providing legal access to marijuana through dispensaries reduce deaths due to opioid overdoses, particularly prior to the October 2009 Ogden memo when dispensary systems were not tightly regulated by the states. We provide complementary evidence that dispensary provisions lower treatment admissions for addiction to pain medications. We find in all cases that the effectiveness of having any medical marijuana law completely disappears when data after 2010 are included. Furthermore, while we show that legally protected and active dispensaries remain an important factor in reduced opioid harm, the magnitude of even this component of the policy has changed since 2010, when states more actively and tightly regulated marijuana dispensaries and as the opioid epidemic has shifted toward heroin consumption.

The final key point that makes this of relevance to my earlier posts on substitutes and complements is this:

In short, our findings that legally protected and operating medical marijuana dispensaries reduce opioid-related harms suggests that some individuals maybe substituting towards marijuana, reducing the quantity of opioids they consume or forgoing initiation of opiates altogether. 

In other words, the paper provides evidence that opioids and marijuana are substitutes, and making marijuana more easily available (and therefore, less costly) reduces demand for and use of opioids, and consequently reduces opioid-related harm. However, the most negative part of these results is that the effect disappears after dispensaries became more tightly regulated in 2010. Presumably, after that point it may have become too difficult for those at risk of opioid harm to obtain marijuana as a substitute. Maintaining looser regulation may have been a better option.