Authoritative sources keep using a tiny, barely relevant study to raise cannabis panic

An op-ed published yesterday in The Washington Post, "Beware of this 'miserable' complication of long-term cannabis use," suggested that millions of Americans are suffering from Cannabinoid Hyperemesis Syndrome, a condition where "cyclical vomiting" is the signature symptom.

CHS is real and it's awful. I'm not saying otherwise. But do millions of Americans have it? Probably not.

The recent spate of alarmist articles about CHS is less about the true prevalence of the condition, and more about a tried-and-true American practice—raising unnecessary panic about cannabis. Measured statements about low effect sizes for increased risk don't make good headlines.

In WaPo op-ed, author Leana S. Wen invokes some of worst manifestations of the condition. Patients "scromiting" or "screaming while vomiting." Patients tearing their esophagus.

I first wrote about this last year, when The New York Times estimated that a whopping 6 million Americans have CHS.

The Times authors failed to disclose that Dr. Ethan Russo, the expert they interviewed, had a financial stake in raising panic about CHS. His company, CReDO Science, sells a genetic susceptibility test for the condition.

The Times came to the 6 million number by extrapolating from a single 2018 hospital survey of 155 regular cannabis users,—1/3 of them had CHS, according to the study. The authors of the study write "If this is extractable to the general population, approximately 2.75 million (2.13–3.38 million) Americans may suffer annually from a phenomenon similar to CHS." That's an important "if," because to extrapolate these numbers to the general population of cannabis users is to completely ignore basic epidemiological principals. You can't gather data in West Virginia and assume it applies to California. And you certainly can't survey 155 people in a single emergency department and assume the results apply to the entire country. Because, and this is key... people in the ER are more likely than the general population to be experiencing an emergency.

Note that the authors further hedge the 2.75 million number by calling the diagnosis a "phenomenon similar to CHS" not simply "CHS."

While the 2018 hospital survey estimated 2.75 million Americans have CHS—the Times authors came up with their own estimate by applying the 1/3 prevalence to the much higher number of US cannabis users in 2024. This, again, is not a valid way to estimate population prevalence, but they at least did some math.

In the more recent WaPo op-ed, Wen simply copies the 2018 hospital survey's 2.75 million estimate (and rounds it down, incorrectly) without updating it to account for the current population of cannabis users. She also fails to note that the survey's criteria for establishing CHS are milder than the excruciating, worst-case scenario symptoms she describes. The linked source for the estimate is a JAMA information sheet on CHS, which in turn cites the 2018 hospital survey. JAMA states that an estimated 2.75 million Americans suffer from CHS, without including any of the source authors' hedging around that number. JAMA is JAMA. They should know better. Wen is an emergency physician, professor, and fellow at the Brookings Institution. She should know better. But it's common for very credentialed people and institutions to spread bad science if it's related to cannabis.

In January, medRxiv published a preprint of a study that attempts to make a more nationally representative estimate of CHS prevalence. But these authors, too, designed their study in a way that would inevitably inflate the numbers. Their final estimate of CHS prevalence (or what they call a "CHS-like symptoms") among regular cannabis users, 17.8%, is based on just two criteria: near-daily cannabis use in the past five years; and periods of severe nausea, vomiting or abdominal pain that coincide with that use. They use this number to estimate that 7.2 million Americans have CHS (similar to the 2018 survey's authors, they hedge and call it a "CHS-like" condition—a rather meaningless diagnosis.)

But you can use cannabis daily, and have periods of nausea and abdominal pain for other reasons (reasons that have been super common in recent years, too.) Food poisoning. Norovirus. Covid. That doesn't mean you have CHS. Physician reports suggest that CHS is characterized not just by nausea and vomiting, but "intractable/cyclical vomiting," and that it's distinct from other nausea-inducing conditions based on whether hot showers or quitting cannabis provide relief, among other criteria.

The pre-print seems an especially bad-faith reading, because the study authors actually did include more CHS specific criteria in their study design. Things like symptom frequency (CHS is often characterized by multiple bouts of vomiting in an hour) and whether taking hot showers or stopping cannabis provided relief. If they had used these more detailed criteria for their prevalence estimate, it would have been lower. It's possible they chose to omit these variables in their final estimate because the survey design was poor. The authors admit in the discussion section that the survey wording doesn't clarify whether stopping cannabis or hot showers provided relief to patients—just that they tried these things.

They also write:

This survey included items intended to map onto the description of the CHS that exists in clinical literature, but have not been validated or compared to consensus clinical diagnosis.

At the end of her op-ed piece, Wen writes, in a familiar finger-wagging tone, "there is a major disconnect between public perceptions of the drug and scientific reality. The Gallup poll found that nearly half of respondents believed marijuana had a positive effect on most users, even though most purported medical uses are not supported by science." Well, the prevalence of this particular purported harm isn't, really, either.

Coming soon: Improvising in an emergency—doctors respond to the medetomidine withdrawal crisis without a roadmap

and

Why "But the children!" is a hypocritical justification for prohibition