CHS ER Visits Tripled Overnight. The Cause Was a Diagnosis Code.

You are going to see a headline this month claiming cannabis-related vomiting illness exploded across America. It will cite the CDC. It will use the word “surge.” And you will be left with two bad options: dismiss it as reefer-madness noise, or panic about a condition you had barely heard of last year.

Both are wrong, because of a small administrative fact almost nobody covering this story will mention. The numbers are real — the CDC published them. But the biggest single-month jump in the dataset lands in exactly the month American hospitals were given a brand new diagnosis code for the condition being counted.

This is a piece about how to read a health statistic. Cannabinoid hyperemesis syndrome is the example, and a genuinely serious one, so we will take it seriously while refusing to overstate it.

First: CHS is real, and it is miserable

Cannabinoid hyperemesis syndrome is a pattern of severe, cyclical nausea and vomiting in people who use cannabis heavily and chronically. It arrives in waves lasting a day or more, often with intense abdominal pain, and is frequently accompanied by compulsive hot bathing — people spend hours in showers hot enough to redden the skin because it is the only thing that touches the nausea.

The cruel irony is familiar to anyone who has lived it. Cannabis is one of the more reliable antiemetic tools we have, so the instinctive response to vomiting is to use more of it. That makes it worse. People cycle through this for months with normal scans and normal bloodwork, being told nothing is wrong. For the clinical picture in full, see our dedicated explainer on CHS.

CHS is not greening out, which is an acute reaction that passes in hours. It is chronic and recurring, and the mechanism is still unsettled — competing hypotheses point at CB1 receptor changes in the gut, vagus nerve signalling, and broader disruption of endocannabinoid tone in the digestive system.

Real condition, real suffering, real uncertainty about mechanism. Hold all of that while we look at the chart.

What the CDC data actually says

The specifics matter, so here they are precisely.

CDC researchers led by Vivolo-Kantor analysed the National Syndromic Surveillance Program and counted CHS-involved emergency department visits between January 2023 and May 2026 — 199,565 of them. Rather than raw counts, they reported a proportion: CHS-involved visits per 10,000 all-cause ED visits, which controls for overall emergency department traffic.

Period	CHS-involved ED visits per 10,000 all-cause visits
Jan 2023 – Sep 2025	“remained mostly steady”
September 2025	3.35
October 1, 2025	CHS-specific ICD-10-CM code takes effect
October 2025	11.26
Oct 2025 – May 2026 (8-month average)	3.7x the Jan 2023 – Sep 2025 average

Read that table top to bottom. For 33 consecutive months the line is flat. Then it more than triples between one month and the next. Then it stays up.

That shape is the whole story.

Why the timing is decisive

Diseases do not work like that.

Real changes in incidence across a population of 340 million people produce curves. Something has to drive them — a change in supply, in product potency, in who is using and how much. Those drivers take months to move a national number, and they leave a slope behind them.

The CHS data does not show a curve. It shows a step: flat, then high, with the transition across a single month boundary.

Now put the administrative fact next to it. On October 1, 2025, a CHS-specific ICD-10-CM diagnosis code took effect in the United States — October 1 is when new ICD-10-CM codes go live each year. The jump is September to October. The step and the code change are the same event.

To their credit, the CDC authors say so themselves: the abrupt, sustained increase “might partly reflect improved recognition and coding of CHS rather than a true rise in incidence.” That sentence is in the paper. It is the sentence that will be missing from most of the coverage.

What “no code” actually meant

This explains why the before-numbers were low, rather than the after-numbers being inflated.

ICD-10-CM codes are how American clinical encounters get recorded, billed, and counted. A condition with no code of its own does not disappear — it gets recorded as something adjacent. Before October 2025, a patient arriving with textbook CHS would typically leave coded as some combination of cyclic vomiting, unspecified nausea and vomiting, abdominal pain, or a cannabis use disorder. The clinical reality was captured. The specific syndrome was not.

You can see the workaround in the research itself. When Swartz and Franceschini studied CHS in the Nationwide Emergency Department Sample from 2016 to 2022, they could not simply count cases, because there was nothing to count. They built a proxy: visits coded with both a cannabis-related condition and cyclic vomiting syndrome. Best available approach — but an inference, and one that systematically misses cases where only one label got applied.

So the pre-October 2025 numbers never measured how much CHS there was. They measured how often CHS left a recognisable two-part fingerprint in the record. The new code did not create patients. It made a blurry group of patients countable.

There is a subtler clue in the Swartz data. Over the same window in which their CHS proxy climbed from 4.4 to a peak of 33.1 per 100,000 visits, cyclic-vomiting-only visits fell from 300 to 186 per 100,000. One category rising while an adjacent category falls is what you would expect if cases were being reclassified rather than newly appearing. Not proof — both can move at once for real reasons — but a signal worth noticing.

The principle, and how to use it elsewhere

Epidemiologists have a name for this: a surveillance artifact. The thing that changed was the instrument, not the phenomenon. Once you know its shape, you will see it constantly.

Here is the checklist. Apply it to any “cases of X have skyrocketed” headline:

1. Did the definition change? A new diagnostic code, clinical criterion, screening guideline, or reporting rule can move a number without a single additional person getting sick.
2. Is the jump a step or a slope? Biology produces slopes. Paperwork produces steps. A near-vertical change at a clean calendar boundary — January 1, October 1, the start of a fiscal year — should make you suspicious immediately.
3. Does the timing match an administrative date rather than a plausible cause? If the inflection point is a policy’s effective date, the policy is the leading suspect.
4. Did the denominator change? A rate rises if the numerator grows or the denominator shrinks. During the pandemic, all-cause ED visits collapsed, inflating the proportion of everything else.
5. Does an independent dataset, measured a different way, agree? The strongest test available — and we will run it in a moment.

This is not a cannabis-specific skill. We applied the same lens to a 300,000-patient lung cancer study and to what the Lancet Psychiatry meta-analysis actually found, and it is the discipline that dismantles the gateway drug argument and most stoner stereotypes claiming data backing.

What the data does legitimately tell us

Dismissing the whole dataset would be its own error. Three things survive the artifact problem intact.

First: CHS was badly undercounted, and probably still is. This is the CDC’s own framing — the findings point to “a likely underestimated impact of CHS.” If a code change alone can more than triple the visible caseload, the invisible caseload before it was large. Every patient in that October jump was already having this experience in September. They just were not in the statistics.

Second: the age signal is consistent everywhere. The MMWR data shows higher proportions among people aged 15–24. Swartz found the highest relative risk at 18–25. Canadian data puts the peak at 20–24. Survey data finds younger users more likely to report it. Four datasets, four methods, one answer — that is what a real finding looks like, and it is worth weighing alongside what we know about cannabis and the developing brain and what parents ought to understand.

Third: the sex signal is genuinely unsettled, and you should be told that. Here is where most coverage will flatten something that deserves nuance:

Dataset	What it measures	Period	Sex finding
Vivolo-Kantor (CDC MMWR)	ED visits per 10,000 all-cause ED visits	2023–2026	Higher proportions among females
Swartz (Nationwide ED Sample)	CHS proxy among ED visits	2016–2022	Females at lower relative risk (RRR 0.92, 95% CI 0.88–0.95)
Glodosky (survey, n=10,255)	Self-reported past-year CHS	2023	Males more likely
Plebon-Huff (Canada)	Hospitalisations per 100,000 population	2016–17 to 2024–25	52.1% male

Three of four point toward men; the CDC surveillance data points toward women. These are not identical quantities — a proportion of ED visits is not a risk ratio, and women use emergency departments more than men overall, which can push a proportion upward without any difference in underlying risk. That is a plausible explanation, but it is a hypothesis, not a resolution. The honest answer is that we do not yet know whether CHS falls more heavily on men or women, and anyone confident either way is reading one dataset.

An independent estimate: about 6% of users

The best antidote to a surveillance artifact is a completely different measuring instrument, and one arrived in 2026.

Glodosky and colleagues used the International Cannabis Policy Study’s 2023 US survey data — 10,255 cannabis users, 45.8% female, mean age 39.5 — and simply asked people. Roughly 6% reported past-year CHS.

That estimate has an entirely different failure mode from the ED data: it does not depend on a diagnosis code, a clinician’s recognition, or whether anyone went to hospital. It depends on self-report and question framing instead, which brings its own problems. That is the point — two flawed instruments pointing the same direction beat one instrument pointing hard.

The risk factors they identified are the more actionable output:

• Younger age, and male sex
• Hispanic or two-or-more-races identity
• Older age of first cannabis use, which runs against intuition
• More frequent use of edibles and concentrates
• Riskier use patterns generally, and past-year alcohol use
• Growing your own
• Using cannabis specifically for headaches/migraines or nausea/vomiting
• Bipolar disorder, or psychosis/dissociative disorder

Two deserve emphasis. The edibles-and-concentrates association fits a dose story — 11-hydroxy-THC from edibles and dabbed concentrates both deliver far more THC than flower, and the potency arms race raises the ceiling every year. And using cannabis for nausea is the trap from the top of this article: treatment and cause look identical from the inside. Notably, living in a state-legal market had no significant effect.

The Canadian comparison: what a real trend looks like

Canada is the control condition: its hospital data was not disrupted by an American coding change.

Plebon-Huff and colleagues at Health Canada tracked CHS hospitalisations from fiscal year 2016–17 through 2024–25 — 7,533 records. The rate rose from 1.5 to 3.8 per 100,000 population, an average annual increase of 13.1%, climbing significantly through 2020–21 and then largely stabilising. Mean patient age was 32.4; the 20–24 group was generally highest.

Compare the shapes. Canada: a multi-year climb of roughly 2.5x, then a plateau. The United States: flat, then a single-month tripling, then flat again at a higher level.

The Canadian curve is what a genuine epidemiological trend looks like — gradual, driven by something, eventually levelling off. Read alongside Canada’s legalization timeline, it suggests CHS presentations really did increase over the past decade in a legal market, measured in years rather than in one October.

Both things are true: CHS is more common than it was in 2016, and the American 3.7x figure is mostly the sound of a filing cabinet being reorganised.

What to do with this if it might be you

The practical guidance does not depend on how the counting works.

If you use cannabis frequently and you are having recurring episodes of severe nausea and vomiting — especially if hot showers are the only relief you have found — CHS belongs on the list of things to raise with a clinician. Bring up your cannabis use directly: how much, how often, in what form. The biggest historical obstacle to diagnosis has been that nobody asked and nobody volunteered.

The consistent clinical picture is that symptoms resolve with sustained cessation, and that resolution can take weeks rather than days. For this condition, cutting down is not the same as stopping. If a full stop feels out of reach right now, our guide to tolerance breaks and the structured reset plan are reasonable starting points, though neither substitutes for medical advice. Severe dehydration from repeated vomiting is a legitimate emergency — go in.

If you are not experiencing this, the takeaway is dose awareness rather than alarm. Know when to increase a dose and when not to, and know that dispensary labels are frequently wrong about how much THC you are taking — which is why reading a COA is worth learning. Cannabis remains genuinely useful for some digestive conditions, and compounds like caryophyllene and limonene are why people reaching for relief keep coming back. None of that is undone by CHS existing.

The thesis, one more time

Population statistics tell you what happened to a country. They cannot tell you what happens to you.

The 6% prevalence figure is a useful national number and a useless personal one. What predicts your experience is your chemistry, your frequency, your format, and your own response — not a headline, and not a strain name on a jar. That is the whole argument of this site: the variables that matter are measurable, and most of them are yours rather than the plant’s.

So the only dataset that reliably answers “is this working for me, or against me” is the one you keep yourself. The High IQ app exists for that — logging what you took and how you actually felt, so patterns like escalating nausea show up as a trend line rather than a surprise.

Frequently asked questions

Did CHS emergency visits really triple? Recorded visits did. The best reading is that most of the jump reflects a CHS-specific ICD-10-CM diagnosis code effective October 1, 2025 — the exact month of the increase. The CDC authors note the rise “might partly reflect improved recognition and coding of CHS rather than a true rise in incidence.”

So CHS is not actually getting more common? That is not the conclusion either. Canadian hospitalisation data, unaffected by the American code change, shows CHS admissions rising roughly 2.5-fold between 2016–17 and 2024–25 before stabilising. A real increase over a decade is well supported. A tripling in one month is not.

How common is CHS? The best current estimate comes from a 2026 survey of 10,255 US cannabis users, in which about 6% reported past-year CHS. ED and hospitalisation data cannot answer this directly — they only capture people who sought care.

Who is most at risk? Age 15–24 shows up consistently across every dataset. Frequent concentrate and edible use, riskier use patterns, alcohol co-use, and using cannabis to treat nausea or migraines all carry higher risk. The evidence on sex is genuinely conflicting — three datasets point toward men, the CDC surveillance data toward women.

Does stopping cannabis fix it? Sustained cessation has the most consistent clinical support, and resolution typically takes weeks rather than days. Reducing intake is generally not sufficient. Talk to a clinician, and be specific about your use.

Sources

1. Vivolo-Kantor AM, Liu S, Tanz LJ, Mattson CL, Schier J. Trends in Emergency Department Visits Involving Cannabis Hyperemesis Syndrome Identified Using a New Diagnosis Code — United States, January 2023–May 2026. MMWR Morb Mortal Wkly Rep. 2026;75(30):391–396. DOI: 10.15585/mmwr.mm7530a2 · PubMed 42560855

2. Swartz JA, Franceschini D. Cannabinoid Hyperemesis Syndrome, 2016 to 2022. JAMA Network Open. 2025;8(11):e2545310. DOI: 10.1001/jamanetworkopen.2025.45310 · PubMed 41284293

3. Glodosky NC, Okey SA, Watson TD, Hammond D. Cannabinoid hyperemesis syndrome prevalence and risk factors in the U.S. International Journal of Drug Policy. 2026;153:105307. DOI: 10.1016/j.drugpo.2026.105307 · PubMed 42066407

4. Plebon-Huff S, Goodman S, Abramovici H. Trends in hospitalizations for cannabinoid hyperemesis syndrome in Canada, 2016–17 to 2024–25. Frontiers in Public Health. 2026;14:1740300. DOI: 10.3389/fpubh.2026.1740300 · PubMed 41908777

This article is educational and is not medical advice. If you are experiencing recurrent severe vomiting, speak with a healthcare professional. For background on how the body’s own signalling system fits into all of this, start with our endocannabinoid system guide, and if you have overdone it acutely rather than chronically, see how to sober up from being too high.