Before Philip Seymour Hoffman died, he revealed that he had recently slipped back into a heroin habit after having been clean for more than two decades. The gateway drug for Hoffman, as it is for many thousands of Americans, was prescription painkillers.
When a patient has chronic pain or major surgery, a doctor might prescribe an extremely potent painkiller, such as Vicodin or OxyContin. The patient’s relatives or friends might then raid their medicine cabinet and get their hands on the drugs, too. The pills are highly addictive, and when combined with alcohol or other substances, they can be lethal.
There’s been a threefold increase in the prescription of strong painkillers since 1999, and the drugs now kill more people than car crashes. The problem has escalated so rapidly that last week the DEA tightened restrictions on when and how patients could obtain the prescriptions.
There’s another substance that may help with chronic pain, though: Medical marijuana. According to a study published Monday in JAMA Internal Medicine, the 13 states that had legalized medical marijuana prior to 2010 had a 25 percent lower rate of opioid mortality than those that didn’t. This equates to roughly 1,729 fewer painkiller deaths, just in 2010. The results suggest, in other words, that people were choosing pot over Percocet.
There are a few limitations to keep in mind. The rate of opioid deaths increased in all the states, it just increased less in the states that allowed medical marijuana. It’s not as though everyone with a backache bought a bong and lived stoned and pain-free ever after.
Marijuana is also not a perfect replacement for painkillers, though it does have some analgesic effects. And there might be something else about those states entirely that was responsible for the drop-off in opioid deaths. Finally, say it with me: Correlation does not equal causation.
Still, the debate about marijuana legalization revolves around whether pot will replace alcohol. If it’s a substitute, people would smoke instead of drink. Some public health experts think that would be preferable to the status quo, since marijuana is much less deadly than alcohol is.
Others, though, worry that the two might be complements: People will drink, and that will make them want to smoke weed. And that would be even worse, since marijuana seems to amplify the intoxicating effects of alcohol, especially for drivers.
We still need more research, but this study suggests that marijuana is, in fact, a substitute—but for prescription painkillers, not booze. And since drug overdose death rates in the U.S. have more than tripled since 1990, this might turn out to be an unexpected public-health benefit of legalization.
This post originally appeared on The Atlantic.