“Tell Your Children” by Alex Berenson: reviewing the controversy
“This book could have been longer,” writes Alex Berenson in the epilogue of Tell Your Children: The Truth About Marijuana, Mental Illness and Violence. He’s right. He should have also made it easier for us to check his claims.
For the most part, however, Berenson clearly and vividly outlines the foundations of his controversial argument. In studies ranging in time and place from colonial India to mid-20th century New Zealand to 21st century Western Europe — but not the United States, where research has been scant — cannabis use has been strongly associated with the development of psychosis. Users are at a greater risk of developing schizophrenia later in life; the more they smoke, the higher the risk. Although marijuana advocates have deflected the blame from marijuana by theorizing that a single genetic disorder is responsible for cannabis use and schizophrenia, subsequent genetic studies (in the book, Berenson did not cite authors or titles here) have demonstrated that THC increases the chances of schizophrenia in people regardless of their genetic susceptibility to that disorder.
In an email, Berenson referred me to the review “Schizophrenia and substance use comorbidity: a genome-wide perspective” by Polimanti et al., which referenced the paper “Assessing causality in associations between cannabis use and schizophrenia risk: a two-sample Mendelian randomization study” by Gage et al.
“Note that [Polimanti et al. citing Gage et al.] reports an odds ratio of 1.1 for genetic liability to schizophrenia on cannabis use, and an odds ratio of 1.07 for genetic liability to cannabis use on the risk of schizophrenia,” Berenson wrote (bolding added). “In other words, even the most sophisticated genetic tools can find little genetic overlap, even though schizophrenia is a disease with a very powerful genetic component.”
Berenson misrepresents the research here. The actual odds ratio Gage et al. reported for genetic liability to cannabis use on schizophrenia risk — basically, the strength of the association between genetic susceptibility to marijuana use and developing schizophrenia — is 1.04, with a 95% confidence interval from 1.01 to 1.07. Both Gage et al. and Polimanti et al. emphasized that the 1.1 odds ratio (with a confidence interval from 1.05 to 1.14) made for a notably stronger association.
If this was unintentional, it does not bolster one’s confidence in Berenson’s scientific judgment. A cynic, however, might wonder if Berenson played up the link between a genetic vulnerability to marijuana use and schizophrenia to make the association between the genetic predisposition to schizophrenia and cannabis use seem less strong. Fortunately, I am not a cynic.
Berenson added, “In comparison, the odds ratio for actual regular to heavy cannabis use in adolescence for schizophrenia is far higher, even after accounting for confounders — approximately 2x to 4x, depending on age of initiation and degree of use. (See Andreasson’s seminal 1987 study and many others since.)”
This is at least in line with Andreasson et al., a massive Swedish study of more than 50,000 military conscripts that Berenson describes in detail in his book. After controlling for a number of possible confounders, they reported a relative risk of 2.3 (confidence interval: 1.0–5.3) of schizophrenia in those who had smoked more than 10 times at conscription. Those who had not smoked at conscription (41,280 subjects) had a schizophrenia rate of roughly 0.48%. This supports an odds ratio of roughly 2.3.
Where he evidently found the science lacking, Berenson ran his own analyses. Working with Sanford Gordon, a professor in the Department of Politics at New York University, Berenson crunched some numbers and showed that the passage of decreasingly stringent marijuana laws in the United States has been followed by a rise in emergency room admissions of patients diagnosed with both psychosis and marijuana use disorder.
Berenson then argues that because marijuana causes or intensifies paranoia and psychosis, it also leads to violence. He describes studies and government surveys from around the world linking cannabis with fighting, homicide, aggression, and the like. Surprisingly, several apparently showed that marijuana has a stronger association with violence than alcohol. Next, Berenson establishes that schizophrenia and other types of psychosis are linked with violence. He then shows that marijuana use is particularly associated with violence among people diagnosed with schizophrenia.
Even as this evidence has mounted, however, cannabis activists in the United States have led an increasingly successful public relations campaign on behalf of the drug. With funding from financier George Soros, Peter Lewis of Progressive Insurance Company, and Men’s Wearhouse founder George “You’re gonna like the way you look — I guarantee it” Zimmer, Ethan Nadelmann advocated for the successful passage of Proposition 215, a ballot initiative approving medical cannabis for use in the state of California. On Berenson’s reading, this was the first push down a slippery slope to legalization at the state level and, more recently, bipartisan marijuana legalization bills introduced in the House and Senate.
A former New York Times reporter, Berenson takes pains to tailor his rhetoric to that audience’s sensibilities. After describing his original skepticism when his psychiatrist wife told him that her criminal clients all smoked marijuana, Berenson reprimands his past self for mansplaining; in addition to signaling his present self’s virtue, this subtly, if unconvincingly, aligns his current anti-marijuana stance with feminism. Berenson is likewise a bit too visibly clever in his coverage of a famous 1894 report from the Indian Hemp Drugs Commission. Lauded to this day by marijuana advocates, the report declared that the cannabis products bhang, charas, and ganja should not be outlawed despite concerns that cannabis use was causing insanity and criminal brutality. Berenson points out that two objections came from the native Indians on the commission, who were outnumbered by British commissioners. One of them argued that ganja and charas, which are now understood to have high levels of THC, were more dangerous than bhang, a low-THC preparation traditionally consumed during the Holi festival. Berenson is telling us that those who agree with the 1894 report are on the side of culturally insensitive colonialists — and who wants that? Berenson does not hesitate to point out that cannabis prohibitionist Harry Anslinger was a “racist jerk.” In the end, though, he concludes that Ansligner was “probably right about marijuana.”
Berenson’s middlebrow rhetorical flair extends to the structure of the book. He starts the book by appealing to the reader’s head using the arguments and evidence outlined above. Switching gears in chapters twelve and thirteen, he touches the heart and turns the stomach by describing several recent incidents at the center of the Venn diagram of marijuana, psychosis, and violence. Though anecdotal, these gruesome, tragic events are more convincingly damning than they would have been absent Berenson’s previous exposition.
He repeatedly confesses that he liked Nadelmann. And in the epilogue, he shocks us by admitting that, yes, he opposes legalization.
Tell Your Children has met with criticism. Much of it has focused on Berenson’s assertion that murders and aggravated assaults have risen in four recreational cannabis states — Washington, Colorado, Alaska, and Oregon — at a rate far outpacing the national average. He claims that a statistical analysis demonstrates that it is very unlikely that the increases in murder and aggravated assault were due to chance — 6% and “almost no probability,” respectively. The analysis was unavailable in the book or on the book’s website; in an email, Berenson stated that he relied on a Fisher’s exact test.
At Vox.com, German Lopez has called this argument into question. Correlation doesn’t equal causation, after all. Lopez also referred to a blog post from University of Oregon economist Benjamin Hansen. Hansen applied synthetic control methods to project recent homicide rates in Colorado and Washington based on data from 2000 to 2012. He found that the actual rise in homicides was lower than that predicted by his model. Hansen does not seem to have performed similar analyses for homicide rates in Oregon or Alaska or for aggravated assault rates in any of the four states. I hesitate to echo Lopez on Berenson and accuse Hansen of cherry-picking data, but the rebuttal does seem incomplete.
A similar critique came from Aaron E. Carroll, one of Berenson’s competitors in the niche of New York Times-associated pot explainers (Carroll concludes his thread by linking to his own recent article in the Gray Lady describing the hazards of marijuana). Like Hansen, Carroll, a professor of pediatrics with the Indiana University School of Medicine, has credentials on his side. Notably, Carroll was the blogger who solicited Hansen’s above-mentioned analysis for The Incidental Economist. To his credit, Carroll does not seem excessively Polyannaish about marijuana either. Yet while his thread critiques the pot-violent crime link, it ignores Berenson’s data on E.R. hospitalizations. Assuming that the connection between psychosis and violence is valid — and that link appears the strongest in Berenson’s logical chain — this lends support to the notion that marijuana-related violence may be on the rise. So do the studies Berenson cited linking violence with marijuana intake, which Carroll also ignored.
Berenson has been accused of overstating the case that cannabis causes psychosis. In the book, he claims that a 2017 report from the National Academy of Sciences, Engineering and Medicine is “arguably the most important finding” linking marijuana with psychosis. Yet as quoted in Rolling Stone by the pot journalist Amanda Chicago Lewis, Ziva Cooper, a member of the committee behind the report, stated, “To say that we concluded cannabis causes schizophrenia, it’s just wrong, and it’s meant to precipitate fear.”
In asserting that “the committee found strong evidence that marijuana causes schizophrenia,” Berenson has gone too far. Even Malcolm Gladwell, whose review of Tell Your Children is otherwise fairly cautious in emphasizing our lack of definitive knowledge about marijuana, errs in describing this as an “unequivocal” declaration (Gladwell, an iconically poofy-haired non-scientist who was famously taken to task by iconically poofy-haired neuroscientist Steven Pinker for his misunderstanding of the term “Eigenvalue,” seems an odd choice for the New Yorker’s review of another non-scientist’s heavily scientific book).
“The conclusion in the report is: ‘There is substantial evidence of a statistical association between cannabis use and the development of schizophrenia or other psychoses, with the highest risk among the most frequent users.’ However, the direction of any causal relationship is not clear,” wrote Marie McCormick, a Harvard professor emerita who chaired the committee that released the report, in an email. “[Berenson] is not the only one to over-interpret the statement, but the methods section makes clear that association and causality are not the same. Gladwell is more correct in that he uses the work ‘risk’ but his phrasing is more causal than the report.”
In email communications, Berenson stuck to his guns: “If the National Academy of Medicine decides to revisit this issue in the future, I will happily review its findings. Until then, I will stick with the 2017 report, whose plain language went uncontested for two years until I highlighted it in Tell Your Children.”
Still, the National Academy’s uncertainty about the causal relationship between cannabis and schizophrenia does not necessarily make one sanguine about the prospect of full legalization. McCormick, for example, emphasized that researchers are not as good at precisely measuring cannabis exposure as they are at measuring tobacco and alcohol exposure. For his part, Carroll acknowledges that we know virtually nothing about edibles or other non-smoking routes of marijuana administration; even e-cigarettes, which have drawn ire because their dangers are unclear, seem to be better understood. More and better research must be done.
In a Twitter thread, Beau Kilmer of the Rand Corporation referred to his own 2013 report, which concluded that marijuana use does not cause violent crime. Interestingly, the relevant portion of that report, “Marijuana and Violent and Property Crime (8 papers),” includes a study by Louise Arseneault, “Mental Disorders and Violence in a Total Birth Cohort,” in which Arseneault et al. concluded that marijuana didn’t induce violence directly but rather through marijuana users’ involvement in the violent underground economy. In Tell Your Children, Berenson cited some of Arseneault’s other research, which linked cannabis use during adolescence with adult schizophrenia. It’s strange that her work on marijuana and violence didn’t make the cut.
Does marijuana sometimes cause psychosis? The two are clearly associated with each other. Anecdotally, I can think of a few cases where cannabis seemed to precipitate something very like psychosis in people I’ve known; even Lewis mentions a friend who has had two psychotic episodes and must therefore steer clear of weed. Does psychosis lead to violence relatively often? Undeniably.
Berenson should have been far more cautious in interpreting research. He should have also made more of his data public. Still, some of his arguments deserve consideration in the ongoing battle over whether and how to legalize marijuana throughout the United States. If demon alcohol is the devil we know, weed may be the one we least suspect.
[Note: this story will be republished in the first issue of 420windycity.]