How Cannabis Laws and Mental Health Prescription Drugs Are Linked

In this video interview, Ashley Bradford, PhD, of the Georgia Institute of Technology School of Public Policy in Atlanta, discusses her new paper in JAMA Network Open, which assessed how cannabis laws were associated with prescription fill rates of other mental health medications.

The following is a transcript of her remarks:

My name is Ashley Bradford. I’m an assistant professor at Georgia Tech, and I’m here today to talk about a new paper coming out in JAMA Network Open that’s titled “Cannabis Laws and Utilization of Medications for the Treatment of Mental Health Disorders.”

So, the goal of this project was to understand the impact of cannabis policy changes on the dispensing of psychotropic medications that are used to treat mental health disorders.

Now, we focused specifically on the privately insured population, and we investigated changes in dispensing for five categories of medications — those are benzodiazepines, antidepressants, antipsychotics, barbiturates, and sleep medications. We explored how different types of cannabis policies may have different impacts on our outcomes. We included medical cannabis laws, medical cannabis dispensaries, recreational cannabis laws, and recreational cannabis dispensaries.

Finally, because of the method that we employed, which is a synthetic control model, we were able to estimate how individual states might be responding differently to these policy changes rather than estimating a single overall treatment effect.

The strongest finding that we have — and by that I mean the most consistent and the most statistically significant finding — is for benzodiazepines, and we find that for each of our four policy levers. So again, that’s medical and recreational laws, and medical and recreational dispensaries. Each of those four policy levers were associated with reductions in benzodiazepine prescription fills. And importantly, we didn’t find any disagreement at the individual state level.

So, as I said previously, we ran a synthetic control study, and this means that we were able to focus on each state individually. None of the individual case studies for which we found statistically significant results showed an increase in benzodiazepine prescribing. So all of the states for which we found significant results saw decreases in benzodiazepine dispensing.

This isn’t the case for the other classes of medications that we included. We actually find suggestive evidence that cannabis laws are associated with increases in antidepressant and antipsychotic medication dispensing.

Now, we didn’t find much for sleep medication and barbiturates, but that being said we did find that some of the individual states saw significant changes in their sleep medication and barbiturate dispensing rates. But this effect was completely washed out when we aggregated all of these individual state results into a single estimate.

I think there are several important takeaways from our findings. First of all, our results add to a growing literature on the relationship between cannabis laws and pharmaceutical dispensing. Much of this work so far is focused on opioid dispensing, reasonably so, and most of this work has focused on the Medicare and Medicaid populations. We show that these laws are also impacting psychotropic dispensing within the privately insured population.

Additionally, the fact that individual states are seeing different effects from these cannabis policy changes is important on its own, not only from a statistical point of view. What I mean by that is that the common methods that are used in the field to estimate the overall impact of cannabis policy changes might be masking some important and socially meaningful impacts at the individual state level. Although that is itself important, this also means that we might be able to identify the aspects of individual cannabis laws that are leading to socially optimal or socially beneficial outcomes.

There is a lot of variation in the design of these policies and in the implementation of these policies, and that makes them a bit tricky to study. But that also means that states might be able to learn from each other which policy details matter.

One thing that I do want to emphasize is that our findings cannot speak to how these policy changes are impacting patient health outcomes. There are a lot of questions we can’t answer. So we find consistent decreases in benzodiazepine prescription fills, but does that reflect improvements in mental health outcomes? Are patients actually treating their mental health disorders with cannabis? Are they switching between these psychotropics and cannabis? Is that a better treatment option for them?

We know that benzodiazepine use can lead to significant harm, with patients developing tolerance, potentially experiencing withdrawal symptoms when treatment is discontinued, or potentially overdosing on benzodiazepines. So do the results in our paper point to potential decreases in these negative outcomes? We don’t know that yet.

The results for antipsychotics and for antidepressants are potentially quite concerning. Does that mean that these laws are leading to increases in cannabis use, which is itself leading to worsening mental health outcomes? If we follow these patients for whom we are seeing changes in these dispensing patterns of psychotropic medications following these cannabis policy changes, do we see changes in their health outcomes? Again, our results can’t speak to that, but that does speak to the need for additional research in this area.

This is a really exciting policy area to study, but it is also somewhat limited by the lack of high-quality data. Currently, we cannot observe patients’ cannabis use in the medical claims data, and the survey data that we have often does not allow us to differentiate between cannabis use medically or recreationally.

So to this point, no one can examine how individual patients clinically use cannabis and psychotropic drugs together or substitute between psychotropic drugs and cannabis use. And that’s an important next step, but the data don’t currently exist.

Please enable JavaScript to view the comments powered by Disqus.