For years, advocates of medical cannabis have argued that expanding access to marijuana could help mitigate the opioid overdose crisis. This belief was largely fueled by a 2014 study showing that states with medical cannabis laws had lower rates of opioid overdose deaths between 1999 and 2010. That finding quickly gained traction in the media, in policy discussions, and in the public imagination. Some states even cited it as a justification for approving medical marijuana as a treatment for opioid use disorder.
“The observed association between these two phenomena is likely spurious rather than a reflection of medical cannabis saving lives 10 years ago and killing people today.”
Shover et al., 2019
But new research tells a very different story. 
In a 2019 study published in the Proceedings of the National Academy of Sciences, researchers revisited the original analysis with updated data extending through 2017. Their conclusion? The association between medical cannabis laws and reduced opioid deaths has not only disappeared—it has reversed.
A Closer Look at the Evidence
The original 2014 study, conducted by Bachhuber et al., found that opioid overdose mortality rose more slowly in states that enacted medical marijuana laws between 1999 and 2010. The implication was that people in those states may have substituted cannabis for opioids, thereby reducing their risk of overdose.
To test whether this relationship held over time, Shover and colleagues replicated the original analysis using the same methods but extended the dataset by seven more years. Their updated results revealed that the earlier negative association had vanished. In fact, by 2017, states with medical cannabis laws showed a 22.7% increase in opioid overdose mortality, a sharp reversal from the 21% decrease observed in the earlier time period.
Notably, this shift remained even after accounting for the effects of recreational cannabis laws and low-THC-only medical programs.
Interpreting the Reversal
The authors offer a compelling explanation for the change. They argue that the earlier association may have been spurious—driven not by a causal relationship between cannabis access and overdose prevention, but by other unmeasured factors. For example, states that adopted medical cannabis laws early on may have differed systematically in other ways: in public health infrastructure, criminal justice policy, or patterns of drug use unrelated to marijuana.
This phenomenon, known as the ecological fallacy, warns against making individual-level conclusions based on population-level data. Just because two trends move together at the state level doesn’t mean one causes the other.
In fact, newer individual-level studies have shown that people who use medical cannabis are more likely—not less—to use and misuse prescription opioids. This casts further doubt on the notion that access to marijuana reduces opioid harm on a large scale.
Beyond Correlation: What This Means for Policy
Despite the cautions expressed in the original research, the 2014 findings were widely interpreted as evidence that medical cannabis could be a tool in the fight against opioid addiction. This interpretation influenced both policy and public opinion, but the latest evidence suggests it may have been premature.
Medical cannabis has demonstrated therapeutic benefits for some conditions. However, the idea that it can significantly curb opioid overdose mortality at the population level is not currently supported by robust evidence. Policymakers should be cautious about promoting cannabis legalization on these grounds without further study.
The Need for Nuanced Research
The Shover et al. study underscores the challenges of studying complex public health issues through aggregate data. As cannabis laws evolve, so do many other policies and social trends that influence opioid use and overdose risk. Parsing out these overlapping effects requires careful modeling, long-term data, and ideally, individual-level research designs.
It also highlights the risk of allowing initial, attention-grabbing findings to harden into policy orthodoxy without adequate replication. Scientific claims that become political slogans are difficult to revisit—but this study does exactly that, and in doing so, calls for a higher standard of evidence in cannabis policy debates.
