Robin Murray is right. The American marijuana industry is increasingly powerful and well-financed. As USA Today wrote in April 2018, cannabis investors have “built—largely unseen—a powerful network of businesses poised to take advantage of a more favorable federal climate. That industry already has woven itself into the fabric of states where pot is legal.”
But the industry here also has a huge advantage in spreading its dubious messages about marijuana safety. The link between cannabis and psychosis isn’t nearly as well known outside Britain—at least in part because the United States has no research center like the Institute of Psychiatry.
Of course, the United States has had hundreds, if not thousands, of scientists devoted to studying psychosis. Many are interested in marijuana. But they are scattered across the country. And some of the places best positioned to do marijuana-related work do none at all.
The most glaring example is at the University of Colorado, which has a schizophrenia research center at its Denver campus. Its chair is Dr. Robert Freedman, a respected psychiatrist who edited the American Journal of Psychiatry from 2006 through 2018. Considering that Colorado has among the highest rates of cannabis use anywhere, the center would seem to be in a unique position to study the drug’s effects.
It has indeed focused on the relationship between smoking and schizophrenia.
Smoking cigarettes, that is.
For decades, Freedman has centered his research on whether nicotine-like drugs can help the cognitive symptoms of schizophrenia. The federal government has given him millions of dollars for studies.
Yes, treating the cognitive symptoms of schizophrenia is important. Yes, scientific research is about failure as well as success. Researchers can’t know if their theories will work until they test them. But the fact that a schizophrenia research center in Colorado hasn’t even looked at marijuana seems like an incredible missed opportunity.
The federal National Institute on Drug Abuse has a $1 billion annual budget. But for the last decade it has understandably focused on the opiate crisis. And because NIDA is a government agency, research it conducts or pays for—no matter how objective—can always be criticized as biased against marijuana.
In contrast, the Institute of Psychiatry can’t be accused of having a structural bias against marijuana. It doesn’t just work for the British government. Its researchers collaborate with pharmaceutical companies to research CBD and other cannabinoids for medical purposes. (A side note: Studies have shown CBD to be weakly effective in reducing some psychotic symptoms, leading to marijuana is good for schizophrenia stories from news outlets that should know better. In August 2017, a Forbes headline claimed that “Cannabis Shows Promise in Treating Schizophrenia.” No, no, no. Once again, CBD isn’t cannabis. It’s one chemical in marijuana out of many. And unlike THC, CBD doesn’t activate the CB1 brain receptor. Whether or not CBD helps psychosis has nothing to do with whether THC worsens it. Finally, it bears repeating that nearly all marijuana sold today contains high levels of THC and almost no CBD.)
Yet the trans-Atlantic knowledge gap has also exposed a structural difference between Europe and the United States. American researchers often focus on the ways that mental illness affects the brain on the cellular level. They make careers of running neuroimaging studies examining the brains of people with psychosis. But decades of that work have so far produced no useful new drug treatments or reliable ways to predict who may develop schizophrenia. Research that links the brain changes that cannabinoids produce to psychotic symptoms is even more primitive.
“Brain research around cannabis is at a relatively novel stage,” says Valentina Lorenzetti, a neuroscientist at Australian Catholic University who studies the way the drug affects brain cells and structure. “There isn’t much strong evidence. We need so much more work.”
British and European researchers are often more interested in the epidemiology behind mental illness—teasing out patterns among users. That work is less likely to lead to cures but often more useful for prevention. Epidemiologists realized that cigarette smoking caused lung cancer long before cancer researchers figured out exactly how the tars in tobacco damaged lung cells.
But new treatments are far sexier—and more profitable—than public health advice. Robert Freedman co-owns patents related to screening the nicotine receptor gene for schizophrenia-related mutations. He is listed on another patent for one of the nicotine-like drugs his center has studied. Did those financial interests encourage Freedman’s obsession with nicotine at a time when marijuana use was exploding? Only he knows for sure.
So, while British psychiatrists pounded away on the cannabis-psychosis connection, American doctors ceded the field to legalizers. In Britain, Robin Murray was probably the most important voice on cannabis. In the United States, Ethan Nadelmann and Rob Kampia played that role, aided by journalists.
Americans heard little about marijuana’s mental health risks and plenty about its benefits. And they listened. An annual survey of American high school students showed that in 2005, about 60 percent of seniors saw great harm in smoking marijuana regularly. By 2017, fewer than 30 percent of seniors felt that way. Adult surveys showed similar trends.
No surprise, after remaining roughly flat from 2000 through 2006, American marijuana use began rising in 2007. It hasn’t stopped since.
In 2005, Americans and people in Britain used marijuana at roughly the same annual rates. By 2016, American adult rates had risen 50 percent, while British rates fell more than a third. That year, about 14 percent of American adults reported having used marijuana, more than double the British rate. American adults were more likely to have used cannabis in the last month than British adults were in the last year.
The legalization community likes to call the United States exceptional in its attitudes toward drugs, implying that Europe has a more civilized attitude toward marijuana. They’re right. The United States is exceptional. But not because it’s strict. The United States has the loosest laws and highest rates of cannabis use among any major countries. It also has the noisiest, most aggressive community of users.
The same Twitter search for “cannabis psychosis” that turns up British teenagers writing about psych wards finds Americans complaining about Big Pharma conspiracies to demonize their favorite weed. In Colorado, which has supplanted Northern California as the center of the modern American marijuana industry, the proselytizing is especially aggressive.
I saw and heard their attitudes for myself last April, when I visited Colorado to see the “420” festivities. (Since the 1970s, smokers have used the term “420” to refer to marijuana. Many now treat April twentieth as a quasi-holiday.) In Colorado Springs, I stopped at a dispensary called Epic Remedy. “For heads by heads,” its logo proclaimed.
It’s worth noting that tolerance of THC rises far more quickly than alcohol tolerance does. Usually, 2.5 milligrams of THC are considered the equivalent of one drink. But regular marijuana users can use 100 milligrams or more daily, the equivalent of 40-plus drinks for a novice user, and still function.
The difference in tolerance probably occurs in part because cannabis is less physically toxic than alcohol, so in that way it is a sign of marijuana’s safety. Still, it has a paradoxically negative effect. It means that heavy marijuana users make up a huge portion of overall cannabis use. In Marijuana Legalization: What Everyone Needs to Know, a 2016 book, three nonpartisan researchers explain that “normal marijuana use more closely resembles binge drinking than it does mere drinking.” Only 7 percent of drinkers use alcohol every day. But about one-fifth of marijuana users smoke every day. Further, because use is so concentrated, heavy consumers of THC are even more important to cannabis companies than heavy drinkers are to alcohol companies. Thus, dispensaries and users wind up with a shared interest in keeping cannabis taxes low and playing down the effects of heavy use.
As I pulled out of the Epic Remedy parking lot that day in Colorado Springs, the afternoon hosts on KOA—a popular Denver news-talk station—were discussing a survey that showed 15 percent of people in the state had reported using marijuana on the job. Callers explained they smoked vapes, which can be odorless, for their entire shifts. I couldn’t imagine a liquor store marketing itself as “For drunks by drunks,” or callers to a radio show proudly explaining that they were nipping vodka all day at the office.
On KOA, the hosts pushed back against one particularly cannabis-happy caller, asking if he would be comfortable having a surgeon operate on him after smoking. Yes, the man responded, after coming up with a bizarre hypothetical scenario where the surgeon had operated all night on another patient and needed to sleep for a few hours before returning to the operating room. I’d rather have him smoke and get to sleep naturally than take a sleeping pill, he said.
Hearing the caller contort himself to justify a surgeon getting high was almost funny. It would have been, anyway, if it hadn’t been so reminiscent of the newest strategy that advocates have used to promote their drug—calling marijuana a way to reduce opiate use, a theory breathtaking in its counterfactual audacity.
• • •
The belief that marijuana might somehow be a solution to the opiate epidemic took hold after October 2014. That month JAMA Internal Medicine published a paper claiming that states that legalized medical marijuana had a 25 percent reduction in opiate overdose deaths between 1999 and 2010 compared to those that didn’t. The reduction was greatest immediately after legalization but persisted afterward, the paper found.
“Medical cannabis laws are associated with significantly lower state-level opioid overdose mortality rates,” the paper’s author, Dr. Marcus Bachhuber, reported. In 2010, the laws had resulted in 1,729 fewer deaths than expected, he wrote.
The finding has since become accepted wisdom. Other researchers have cited Bachhuber’s paper more than 250 times. Marijuana companies advertise its findings as fact on electronic billboards in cannabis-legal states. In February 2018, Dr. Richard A. Friedman, a psychiatrist at Weill Cornell Medical College, wrote an opinion piece in the New York Times called “Marijuana Can Save Lives.” Friedman pointed to the JAMA paper’s findings to criticize Attorney General Sessions for enforcing federal laws against marijuana.
“Marijuana isn’t a gateway drug to opioid addiction; it’s a safer alternative to pain medicines. Mr. Sessions’s vow to crack down on marijuana will only make the opioid epidemic worse,” Friedman wrote. “If cannabis were actually a dangerous gateway drug, as the attorney general suggested, it would be very easy to see in the data. We would find that medical-marijuana laws increased opiate drug use and overdose deaths, when in fact just the opposite has happened.”
Just the opposite?
Even at the time JAMA published Bachhuber’s paper, its findings should have been viewed as dubious for a half-dozen reasons.
Many previous studies had shown marijuana use was connected to later heroin use. Researchers had found opiate addicts in methadone treatment relapsed more frequently if they used marijuana. Other studies showed adolescents who used marijuana were more likely to use heroin later, the so-called gateway theory Friedman mentioned.
Two of the most interesting studies looked at twins. A 2003 study that examined twins in Australia found that in cases where one twin used cannabis before 17 and the other did not, the twin who used was almost four times as likely to develop an opiate or cocaine use disorder. A 2006 study of twins in the Netherlands found even bigger risks from early use.
The twin studies naturally controlled for both genetics and upbringing—twins lived in the same house at the same time, and identical twins had the same genes. The gap “could not be explained by common familial risk factors, either genetic or environmental,” the Netherlands researchers wrote.
Other researchers argued against the gateway theory, noting that plenty of people used marijuana without moving on to opiates, cocaine, or other drugs. But until Bachhuber, no serious researcher had suggested that marijuana use—whether called recreational or medicinal—might actually discourage other drug use.
Okay, but the JAMA paper focused on medical marijuana. Maybe medical marijuana was different than recreational. Only it wasn’t. The drug was the same, of course. And as you’ve already learned, most people who used medical marijuana had been recreational users and received authorizations from doctors who specialized in writing them.
Bachhuber’s paper also showed a very strange time effect. Early medical marijuana programs needed years to accumulate significant numbers of patients. Smokers didn’t want to join state registries before they were sure authorities wouldn’t use the participation against them. Colorado’s registry had only 94 patients during the first year of its program, in 2000. Until 2009, it never had more than five thousand patients. Other early states had similar statistics. Yet Bachhuber found the greatest impact of the programs came in their first year. How could that be, when almost no one used them?
Another problem with the theory was that even after they took off, medical marijuana programs covered only a small fraction of the adults in any state, even ones like Colorado. At the same time, overall marijuana use began to soar across the United States in 2006. The increase happened both in states that had medical laws and those that didn’t. In other words, medical marijuana laws were something of a sideshow. The marketing of marijuana as medicine encouraged use across the country, but the new laws themselves made only a modest difference to use.
Yet another problem with the theory was that marijuana simply wasn’t a strong enough painkiller to be effective for most people who truly needed opiates. The studies that showed marijuana’s effectiveness as a painkiller usually tested it against placebos rather than other painkillers. Like alcohol, cannabis works as an intoxicant first and a painkiller second. And even the modest improvements that cannabis shows may decrease over time. In July 2018, Australian researchers published a study of 1,514 people with chronic pain and found that long-term cannabis use was associated with greater pain over time, and “no evidence that cannabis use reduced prescribed opioid use or increased rates of opioid discontinuation.”
A simple look at a map makes obvious that geography more than anything else drove the first stages of the opioid epidemic. From the mid-1990s through 2005, painkiller prescriptions and overdoses spread from Appalachia through the Midwest, with a jump to Florida, which had loose prescribing laws. Addicts and reporters called OxyContin “hillbilly heroin.”
The medical marijuana wave was also geographically driven, but on the other side of the country. Until 2008, fewer than a dozen states had legalized medical marijuana. They were nearly all west of the Mississippi. That geographic coincidence—and nothing else—is the most likely explanation for the link that the JAMA Internal Medicine paper found.
After 2010, the medical marijuana movement moved east, and the opiate epidemic spread nationally. Suddenly many states had both severe opiate problems and medical marijuana laws. If medical marijuana really offered a solution to the overdose crisis, the evidence after 2010 should have been incredibly strong.
But it isn’t. In fact, the post-2010 data shows that medical marijuana laws are correlated with an increase in prescription deaths. Increase, not decrease.
How can I be so sure?
Because I’ve run the numbers.
Or, to be more accurate, Sandy Gordon has. Dr. Sanford Gordon is a professor at New York University and an expert in data analysis. He’s also an old friend. Running correlations on a multiyear dataset with thousands of data points was beyond me, but I knew he would have no problem with it. By his standards, the question I had was simple: Do medical marijuana laws, and changes in state use rates more generally, increase or decrease opiate overdose deaths? And what about cocaine, marijuana’s friend from the 1970s?
Finding state-level data on marijuana use, cocaine use, and overdose death rates is easy enough. The Centers for Disease Control tracks overdose deaths. A federal agency called the Substance Abuse and Mental Health Services Administration tracks drug, alcohol, and tobacco use at the state and national level. All the data is public and available for free.
We spent a sunny spring morning downloading it. Then Gordon plugged it into an analysis program called Stata and created a statistical model to run the numbers. Once Gordon had set up the model, we didn’t have to wait long for the answer.
As I looked at the numbers I had a sense of how Sven Andréasson or Mary Cannon must have felt when they found a cannabis-psychosis link: people need to see this.
From 1999 through 2016, the most recent year for which state-level data was available, no link between medical marijuana laws and opiate deaths existed. In other words, the effect in Bachhuber’s paper disappeared after 2010.
Further, on a state-by-state basis, overall marijuana use showed a moderately positive link with overall opiate deaths. In other words, states where more people used cannabis tended to have more overdoses.
The analysis also revealed a connection between marijuana and cocaine. Cocaine use fell during most of the aughts and bottomed around 2010. Since then it has risen nationally along with marijuana use—especially among young adults. On a state-level basis the results are even more striking. Marijuana use is strongly correlated with cocaine use at the state level, and changes in marijuana use are correlated with changes in cocaine use.
Other studies since the JAMA paper have also suggested that trying to use cannabis to stem the opiate epidemic is a dangerous mistake. A February 2018 paper in the International Journal of Drug Policy that studied 245 poor women in San Francisco showed that those who used marijuana were more than twice as likely to use opioids as those who didn’t. No other factors—not even homelessness or exposure to violence—increased opiate use nearly as much.
And a July 2017 paper in the Journal of Opioid Management found that medical cannabis laws were associated with a 22 percent increase in age-adjusted opioid-related mortality between 2011 and 2014. Worse, mortality increased as time passed.
“It was surprising for me too, when I ran the numbers and got the results,” said Elyse Phillips, the study’s author. “When you just look at yes or no having a medical marijuana law, there was a correlation with those states having much higher deaths.”
But practically no one has noticed Phillips’s paper. It has been cited only once in the year since it was published.
An even more worrisome result came from a 2017 study that traced drug use in individuals over time rather than depending on state-level data. Trying to tease out all the factors driving marijuana or opiate use in an entire state is next to impossible. Looking at changes in individual behavior over a period of years is a far better way to determine cause and effect.
So what scientists really needed was a big national survey that asked people about their drug use and then returned to the same people years later. But no one seemed to have conducted that research.
Then Dr. Mark Olfson, a psychiatrist at Columbia University who specializes in addiction, realized that he could find the data in a survey initially designed to measure alcohol use. In 2001–2002 and again three years later, the National Institute on Alcohol Abuse and Alcoholism surveyed 34,000 Americans on their substance use and psychiatric problems.
The study was called the National Epidemiologic Survey on Alcohol and Related Conditions. It focused on alcohol—but it covered other drugs, too. It also had plenty of questions on other risk factors, those tricky confounders.
Olfson and his coauthors examined whether people who used cannabis but not opiates in 2001 were more likely than people who didn’t use cannabis to start using opiates over the next three years. The answer was yes. Cannabis users were almost three times as likely to be using opiates in 2004, even after adjusting for other potential risks. The risk was even greater for people who used cannabis heavily. Those people had a 4 percent chance of becoming addicted to opiates three years later, while other people had a 0.5 percent chance—a risk of 1 in 25 versus 1 in 200.
Olfson and his coauthors published their study in the January 2018 issue of the American Journal of Psychiatry. “Adults who use cannabis, three years later, controlling for a wide range of things, were found to be at greater risk of opioid use and opioid use disorders,” Olfson told me.
Yet Olfson’s paper has gone practically unnoticed. In the first eight months after it was posted online, it was cited by other scientific researchers only six times and received little media attention. Olfson says the lack of interest puzzles him. Brain studies of cannabis and real-world experience offer evidence that marijuana use might lead to use of other drugs, he said.
“Initial experiences with marijuana are often pleasurable, they may encourage continued use, they may encourage use of other drugs,” he said. “There’s evidence that there’s a shared underlying biology.”
I asked Bachhuber if he would talk about his JAMA Internal Medicine paper. He readily agreed. He seemed like a nice guy, genuinely concerned about the welfare of his patients. He’s a specialist in internal medicine—an old-school front-line doctor. He said he decided to study the issue after hearing from patients who had used cannabis for their pain.
“I just became really interested in the idea that medical cannabis has been emerging as a treatment option,” he said. “What might this increased access to cannabis have on opioid-related issues?”
Bachhuber said he wasn’t a strong advocate for marijuana at the time he carried out his study, though he leaned toward legalization. “I had a general liberal attitude about it, but also tempered by medical training, which is extremely negative about cannabis.” He was pleased with his findings and doubly pleased the study had received so much attention.
I asked Bachhuber about the fact that the later data doesn’t support his pre-2010 findings. He didn’t disagree, but he said that finding didn’t change the truth of his paper. “When you’re looking at the effect of a policy, it may have different impacts over different periods of time.”
The updated results have not changed his point of view on medical cannabis, he said. In fact, he believes more strongly than ever that cannabis should be legal. He treats patients who are trying to use cannabis to wean themselves from opioids, sometimes with success. “The story that people substitute cannabis for prescription opioids is really common,” he said. “I’ve never seen it harm anyone yet.”
I’ve never seen it harm anyone yet. The language struck me. It was so similar to the insistence that marijuana has never killed anyone, despite the death certificates that tell a different story. Of course, marijuana is safer than fentanyl. Hang gliding without a glider is probably safer than fentanyl.
But never?
Never is an advocate’s word.
Bachhuber has every right to advocate for cannabis legalization, and not to go out of his way to discuss the more recent findings. But his study, which is deeply flawed, has been taken as gospel. That misimpression is not his fault. It’s the fault of everyone who is looking for a quick and easy solution to the opiate crisis.
In 2017, Dr. Chinazo Cunningham, one of Bachhuber’s colleagues, criticized a federal commission that had noted the findings from Mark Olfson’s study showing that cannabis users were much more likely to become opioid addicts.
“People are dying every day from opioid overdoses,” Cunningham told CNN. “We must act now.”
Do something! Act now! But sometimes doing something is worse than doing nothing. The same urge to do something—anything—about patients’ pain fueled the opiate mess.
A few days before I spoke to Bachhuber, I talked to Dr. Michael Lynskey. Lynskey was the lead author on the Australian and Dutch twin studies that showed that teenagers who used cannabis were more likely to use other drugs later. He’s a professor of addiction at—inevitably—the Institute of Psychiatry. I asked him what he thought of the theory that medical cannabis can somehow substitute for opiates.
Lynskey chose his words carefully. “It’s something that I think everyone really hopes is true,” he said.
But Lynskey came back to a point that other researchers had made. Studies that try to tease out effects by looking at large groups of people provide far weaker evidence than research examining how individuals change over time. And even tracking individuals doesn’t always work. For example, observational studies show that older people who exercise are less likely to get dementia. But when researchers tested exercise in a clinical trial of 500 older people with mild dementia, they found that the group that exercised wound up with more symptoms by the end of the trial.
The reason scientists view clinical trials as the gold standard of research is that those trials take two groups of people who otherwise are almost exactly alike and give one group a drug that the other doesn’t get. If the groups are different afterward, it is reasonable to assume that the drug is responsible.
But anything less than a randomized trial is guesswork, to a greater or lesser degree. And ecological studies like Bachhuber’s are barely better than nothing. In a February 2018 editorial in the journal Addiction, leading experts on drug use called the evidence for the cannabis-stops-opiate use hypothesis “very weak.”
Lynskey has spent his career studying addiction. But unlike Bachhuber, he is careful about making any claims about cannabis, positive or negative. Lynskey even cautioned against reading too much into the twin studies, his own work.
“I’m not a clinician—I’ve got a PhD, and I mainly do statistics,” he said. “With my training, we tend to be cautious.” He said he understood the pressure that physicians like Bachhuber felt. “Doctors are presented every day with people who are in pain and need relief . . . people will become advocates for something based on their personal experience.”
Lynskey tries very hard to avoid becoming an advocate. Drug use and addiction are complex, he said. Being wrong means pushing for policies that can lead to more addiction, more ruined lives.
“My approach, and I’ve been criticized for it, is to try not to be a particularly strong advocate for anything, because a lot of my work is trying to delve into uncertainty.”
So, is the gateway effect true? Does cannabis use make people more likely to use harder drugs? I asked. Lynskey didn’t exactly say yes, but he didn’t say no, either.
“There’s been a number of studies that have used different ways to address this question and have been unable to disprove this association.” The social effects of cannabis use might be more responsible than brain changes the drug causes, he said. People who use drugs tend to spend time with other drug users, and if one person in a group tries opiates or cocaine, the others may follow.
As with groups of friends, so with entire nations. What’s gone unnoticed in the discussion over state-by-state changes is the striking correlation between the opiate epidemic and cannabis use at the national level. The United States and Canada are outliers among Western countries for marijuana use. Adult and young adult marijuana use have risen sharply since 2005. Meanwhile, Britain has gone the other way.
In many other ways, the United States, Britain, and Canada have a lot in common. They are all wealthy, English-speaking countries with sophisticated health care systems and a long history of opiate use.
While the United States and Canada are suffering an epidemic of overdose deaths, Britain isn’t. In 2000, the United Kingdom and the United States had similar drug death rates. That year, about 17,000 Americans and 3,000 people in England and Wales died of overdoses—a death rate of about 6 people per 100,000. On both sides of the Atlantic, about half of those died from opiates.
In 2016, about 65,000 Americans died from overdoses, including almost 45,000 from opiates. In England and Wales, the number was 3,700, including 2,000 opiate deaths. Americans now die from drugs at three times the rate of people in the United Kingdom. And the overdose epidemic in Canada is nearly as bad as that in the United States.
Richard Friedman was more right than he knew in his New York Times piece:
If cannabis were actually a dangerous gateway drug, as the attorney general suggested, it would be very easy to see in the data. We would find that medical-marijuana laws increased opiate drug use and overdose deaths.
So they have.
Of course, drawing conclusions based on national-level changes is even more dangerous than drawing them based on changes by state. Many factors—including prescription painkiller advertising, the rise of fentanyl, and possibly even increased Medicaid access—have driven the American opiate epidemic. Attributing all of it to rising cannabis use is clearly wrong. Opiate deaths began rising in the late 1990s, years before American and Canadian marijuana use spiked.
But if marijuana use prevents opioid use, why are overdose deaths centered in the two industrial countries with the highest rates of cannabis use?
Why have they doubled in the decade since marijuana use took off in the United States and Canada?
I don’t expect cannabis advocates to have a convincing answer to that question anytime soon.