To be sure, Dr. Chelsea Shover’s widely circulated paper interpreting the relationship between medical cannabis laws and the ongoing opioid crisis as non-causal was published in the more prestigious journal Proceedings of the National Academy of Sciences, and not, as NORML’s Paul Armentano mistakenly wrote, in PLoS ONE. For those of us not mired in the petty vanities of academia the slight, found in Armentano’s public response to Dr. Shover’s findings may seem minor if it bothers to register at all.
But the “Trading Places” journal swap — of which there’s no evidence it was anything but a misprint — was a serendipitous mistake. At a minimum, it played towards the prejudices of the ‘publish or perish’ set at a time when, in Armentano’s own words, “the belief that expanding cannabis access plays a role in mitigating opioid use and abuse came under fire.”
Dr. Shover, a postdoctoral research fellow at Stanford University worked with fellow researchers Corey Davis, Sanford Gordon, and Keith Humphreys to buck a cannabis industry pearl of conventional wisdom; the oft-cited, often-sloganeered study by Marcus Bachhuber that found a positive correlation between medical cannabis states and a decrease in opioid overdose deaths from 1999-2010. Using the same methods and data, the team replicated those original findings — contrary to Armentano’s assertion — but found the trend reversed when extending the analysis through 2017.
Logically speaking, Dr. Shover explained to me over the phone, that meant that cannabis went from saving lives to killing people.
“I think that’s absurd on its face,” she said, being the rational scientist that she is. Having read some of the criticisms lobbed at her work by NORML and other advocates, she was eager to discuss her findings.
What the data clearly showed is the relationship between medical cannabis and opioid use disorder is non-causal and never was, she told me.
“In doing this analysis I’m saying is this a good strategy,” Dr. Shover asked. “Is passing medical cannabis legalization a good strategy to prevent opioid overdose death, and I think our study gives pretty good evidence, that it’s not.”
“And that doesn’t mean we shouldn’t have medical cannabis,” she continued, describing the plant’s other medical applications.
However, an either/or proposition has always been crucial to the success of the cannabis movement. It was advocates like Armentano, Steve Fox, and Mason Tvert — once dubbed “the Don Draper of Pot” by Politico — that convinced a generation of college students that marijuana is safer than alcohol, and then taught them to lobby on that talking point.
More than 15 years ago, as legalization languished, and arguments for increased tax revenue and protecting the children held little to no sway with politicians, the message of moral relativism turned the tide. Fueled by polling data gathered by Tvert, advocates rode the notion to a string of successes, from Denver’s effort to decriminalize pot in 2005 through to Amendment 64 in Colorado and up to the modern legalization movement we have in the country today.
Now, with a growing body count advocates and the industry that make up ‘Big Cannabis’ have a new foil in the opioid epidemic. Anchored by Bachhuber’s original study, the either/or proposition is now framed as a question of life-or-death, with many, if not most, claiming the science already settled.
“[Bachhuber’s study] was really latched onto by a lot of people who wanted it to be true,” Dr. Shover told me, recalling how, at the time, even she had high hopes for the findings.
“It is an association, correlation is not causation,” she said, pointing out the overall issue in making causal claims with extensive population studies. Later in our conversation, she hammered home the point, noting that “you can’t use them to make inferences about individual behavior or outcomes.”
Bachhuber noted this too, in his findings. Now, companies like WeedMaps, marketing gurus that they are have turned his research into this generation’s “Where’s the Beef?” It’s now slapped on billboards across the U.S.
“My favorite application of correlation is not causation is the WeedMaps article about [Bachhuber’s] study,” recalled Dr. Shover. “The end talks about how starting cannabis use is associated with starting other drugs, and that’s when they say ‘and correlation is not causation.’ I’m like, okay. Well, if correlation is not causation there, it’s not causation at the top either.”
But that doesn’t matter to WeedMaps, NORML, MedMen, the Marijuana Policy Project, or the myriad of other groups or businesses that continue to parrot the findings. The cannabis industry affords Bachhuber’s study a level of gravitas that has allowed it to reach decision-makers and become policy in several states. In Pennsylvania, New Jersey, and New York, for example, opioid use disorder is now a qualifying condition for medical cannabis.
“If patients are given, are encouraged, or driven to try medical cannabis, instead of therapy for opioid use disorder that we do have evidence for, my concern is just that it won’t be as effective and that will be worse for patients and their families,” said Dr. Shover.
Or to put it more bluntly, she continued: “The issue is medical marijuana, in general, is when you put the word ‘medical’ on it, that implies that it’s regulated in a way that it really isn’t. And so it’s not that cannabis is dangerous. It’s that we’re treating this thing like a medicine, but we’re not regulating it like a medicine.”
“Why the disparate results?”
“…it remains premature at best, and sensational at worst to claim that a causal relationship exists between marijuana use and the onset of psychiatric disorders, particularly among those not predisposed to the condition,” wrote Paul Armentano in an op-ed for NORML titled “Will smoking cannabis make you psychotic? Not likely.”
His response was to a widely circulated paper published in The Lancet earlier this year that found a causal relationship between daily cannabis use and psychosis was, well, likely — more likely than we’ve been led to believe, anyway. Researchers from Kings College London compared 901 daily cannabis users to 1237 non-users, finding that the former group had anywhere from a three to five times higher likelihood of developing psychosis, with THC potency being the deciding factor.
Like a southern Christian school board elected to “teach the controversy,” Armentano’s quote was inserted in several media reports about the findings, used as a counterfactual to balance out the claim.
But NORML’s bully pulpit, which has the ear of both the public and the press tends to shut down science as much as it tries to embrace it.
“I knew that doing anything on the topic of cannabis would be controversial,” Dr. Shover told me. Still, she was surprised to see her team’s paper receive the response from NORML and others that it did.
“Saying that cannabis doesn’t solve every problem doesn’t mean we shouldn’t regulate it and we shouldn’t have reforms that should correct the harms of prohibition,” she continued.
NORML, however, has always had a winner-take-all approach when it comes to legalization. As far back as 1996, the organization claimed that anecdotal evidence showed cannabis was an effective treatment for opiate addiction, along with other substance use disorders. This, despite scientific research at the time proving it wasn’t.
“Why the disparate results,” wrote Armentano, questioning how Dr. Shover’s study failed to replicate Marcus Bachhuber’s findings even though it did. His answer: data variables.
According to Armentano, Dr. Shover’s study broadened the definition of what it means to have a medical cannabis program, including “states with non-traditional medical marijuana laws” in her findings. He then claimed she further skewed her results by including states with medical cannabis laws that had “yet to establish regulated cannabis production or sales.” In other words, medical marijuana was legal, but no one had set up shop.
It was deception in plain sight, and cannabis advocates ate it up.
“That’s why we did the follow-up analysis that we reported in column four of our table because we expected that sort of criticism,” she told me. She’s way too nice of a person to say that Armentano was full of shit.
“We did consider those states to have medical cannabis laws,” she continued, in no way conceding Armentano’s point. “Then we did the subsequent analysis to say, okay, well what if you treat recreational as most access, comprehensive medical without recreational as second most, and then what is low THC-high CBD as a third category.”
It’s all right there in the study results, where the authors found a 22.7 percent increase in opioid overdose deaths using the more comprehensive data set, with a 95 percent confidence interval, and a 28.2 percent increase using the more restrictive definitions as preferred by NORML.
“About them saying that we categorized states as having medical cannabis laws as like we counted from the wrong date,” Dr. Shover continued, addressing the criticisms. “They claimed that the Bachhuber paper counted from the date dispensaries opened. That’s not true. You can run the data; you can read the papers.”
Again, she’s right. Both studies use the National Council of State Legislature data counting from the days the laws went into effect. It was a criticism of Bachhuber’s paper initially, so much so that another follow-up study from the RAND Corporation in 2018 argued that it wasn’t medical cannabis laws but fully-functioning dispensaries that were contributing to the decrease in opioid mortality.
That study also found the trend abating after 2010.
But jumping to conclusions based on a single study is ridiculous, and, as Armentano noted, dozens of peer-reviewed studies do support the cannabis substitution theory.
“They did cite a number of studies to support their point and the ones using state-level data they have the same issue as the Bachhuber study and our study,” Dr. Shover said, reminding me once again how they really can’t be used to make inferences about individual behavior outcomes.
“There are some with individual outcomes that they cite, and they’re fair game,” she continued. “There are two notable studies they don’t cite though, and that’s concerning — if their goal is to present the evidence.”
Last year, a large Australian cohort study on pain and cannabis found that cannabis wasn’t associated with improvements. The study, published in The Lancet, followed 1514 participants over four years and found that “people who used cannabis had greater pain and lower self-efficacy in managing pain, and there was no evidence that cannabis use reduced pain severity or interference or exerted an opioid-sparing effect.”
Moreover, another study, by Theodore Caputi and Keith Humphries, looked at the 2015 National Survey on Drug Use and Health and found that medical cannabis users were significantly more likely to have used prescription drugs in the past 12 months.
All in all, by the numbers, Caputi put to rest some of the more significant mixed messages surrounding cannabis and opioids — surrounding cannabis and a lot of things, to be precise.
“Studies have found that medical marijuana laws (MMLs) are responsible for a 25% decrease in opioid overdose rates,” wrote Caputi in a letter to the journal Addiction. He noted that similar studies had found medical cannabis to be the cause of, or associated with similar reductions in obesity, Medicaid and Medicare prescriptions, and alcohol sales as well.
“Unfortunately, these effects are probably too grand to even possibly be ascribed to medical marijuana use,” he continued. “Simply put, medical marijuana use is probably not prevalent enough to account for any one of these effects, let alone all of them.”
For example, as Caputi wrote, medical cannabis users make up roughly 2.5 percent of any state, yet researchers ascribed to them a reduction of 15 percent of all alcohol sales. That’s $4800 per user, per year, or 3000 drinks — a lot of unsafe booze, as Armentano would say.
“Those are two pieces of evidence that are counter to their goals that they did not include,” Dr. Shover noted, circling back to NORML. “They’re two of the larger studies on this topic, so it’s weird to not include them.”
“…the plural of anecdote isn’t data.”
The internet is awash in stories of those that suffer from opioid use disorder that use cannabis to stay clean, heroin addicts that moved to Colorado to save their lives, and dope fiends that have foregone the 12 steps for a daily dose of Bubba Kush.
“My first thought is that’s awesome,” Dr. Shover told me. After all, she reminded me, she’s first and foremost trying to reduce morbidity and mortality. “I’m really glad that someone has a good outcome like that.”
“My second thought is, the plural of anecdote isn’t data,” she said.
And she’s right. As we spoke, Dr. Shover stated what should be obvious; we don’t have enough evidence to recommend cannabis as an opioid substitution on a population level. But we do for other medications, like buprenorphine, methadone, and naltrexone.
“We should not be making policy based on that,” Dr. Shover stressed to me.
Worse yet, we’re stigmatizing treatments that work in pursuit of such policies. Dangerous arguments are now made promoting the idea that life-saving medications like naloxone encourage drug use, while the debate over cannabis has become an all-out culture war.
“There’s no shame in using medication,” said Dr. Shover. “But there shouldn’t be right like medically there’s not. These things work.”
She continued: “I still have a conversation with people fairly often where I’m like, we have treatment for opioid use disorder, and they’re hard to access, and there’s a lot of stigma, but they do exist. And if we could get over that thing that would really help a lot of people have a better time, in a lot of ways save lives, but also just enable people to make their own life where they want it to be.”
The truth goes a long way towards ending those stigmas.
For example, to be sure, Dr. Marcus Bachhuber’s widely circulated study which found an association between medical cannabis laws and a decrease in opioid overdose deaths was published in the slightly less prestigious JAMA Internal Medicine, and not, as NORML’s Paul Armentano mistakenly wrote, in the “Journal of the American Medical Association.”