This site gives medical instruction and shows no medical degree. That is a fair thing to be suspicious of. This page is the answer, not a defense of it, a plain account of who wrote it, how a claim gets checked, what changed and why, what could not be verified, and what would prove the whole thing wrong.
My name is KP. I had cannabinoid hyperemesis syndrome for about ten years before I understood what it was, quit cannabis, and got better. I built this site afterward, alone, because the page I needed at 3 a.m. did not exist. I have no medical training, no research degree, and no clinical license. Nothing on this site is medical advice, and nothing in this sentence is false modesty. It is the fact you need before you read anything else here.
So what is the value of a site built by someone with no letters after their name. Not authority. A clinician's authority comes from years of training and from having seen hundreds of cases in a room, and I have neither. What I have instead is time, the kind a fifteen minute appointment does not allow, to read a paper cover to cover instead of its abstract, to sit with a table of numbers until they make sense, and to publish the parts that are still unknown instead of smoothing over them because an unanswered question looks unfinished. That trade, credentials for time and honesty about the gaps, is the whole premise of this site. You should weigh it accordingly.
Some plain facts about how this project runs, stated once here so they never need restating. It sells nothing. It runs no advertising. It takes no affiliate commission from any product, test, or service it mentions. It has no financial relationship, no consulting arrangement, and no partnership with any cannabis business, dispensary, treatment center, rehabilitation program, or advocacy group, on either side of the legalization argument. If that ever changes, this paragraph changes with it, in public, with a date.
You will not find a "reviewed by Dr. So and So" line anywhere on this site. That slot is empty because it should be empty rather than filled with something vague, a friend of a friend who is technically an MD but has never treated a case of this, or a name used for the credential and not the review. It is waiting for an actual clinician, ideally one who has managed CHS in an emergency department or a gastroenterology clinic, willing to read the guide, the ER card and this page and put their name on what holds up. If that is you, the correction line below reaches me directly. Until that happens, the honest state of this site is: written by a patient, checked against primary sources, not yet checked by a clinician. Read it that way.
The method
Nothing here came from a search summary. Every claim on this site traces to something somebody wrote down and published, and the ones that did not survive checking are listed further down with the reason why.
Every number on this site went through the same six steps before it was allowed on a page. None of them are complicated. Most health content on the internet skips at least two of them.
One rule that matters more than it sounds like it should: ten news outlets carrying the same wire story is one source, not ten. Volume of coverage is not independent confirmation, it is one fact traveling. The correction on the WHO's ICD-11 code in chapter four exists because that exact pattern happened, several outlets, one overstated framing, repeated until it read like consensus.
Every claim of consequence on this site carries one of four labels. The labels are load bearing. Nothing gets upgraded because it would read better, or because the truth is inconvenient at a lower label.
Established. Read directly at the primary source, not at a summary of it. A trial result, a measured lab value, a clinical pathway's own wording, a coroner's finding.
Well supported by the evidence available, but not proven the way a fact is proven. Usually because the source is secondary, the sample is small, or the finding has not been independently replicated.
Plausible and worth naming, but thin. A single small study, a mechanism nobody has tested directly, a pattern that could have another explanation nobody has ruled out.
Genuinely unresolved. Nobody has run the study that would answer it. Stating this plainly, instead of picking a side, is itself the honest answer.
The rule underneath all four labels is the one that does the actual work: a self selected survey stays a self selected survey no matter how large it gets, a preprint stays a preprint until it clears peer review, and a conference presentation is not a peer reviewed paper, even when the same team publishes the paper a year later on the same dataset. Size and repetition do not change a source's kind.
This site's single largest source is a 1,134 person survey of cannabis users with CHS symptoms, published in a peer reviewed journal in 2026.[1] It is genuinely the biggest dataset on this population anywhere. It is also recruited entirely from CHS focused social media communities, which means every respondent already suspected they had the condition before they answered a single question.
That detail is not a footnote, it is the whole ballgame. A survey of people who found their way to a CHS support group cannot tell you what share of cannabis users get CHS, because the people who never got sick, or who got sick and never connected it to cannabis, are not in the room. So this site uses that 1,134 person survey constantly, for what people with confirmed CHS experience, how they source their cannabis, how they describe onset, and never once for a prevalence or incidence rate. No sample size fixes a selection problem. A bigger wrong number is still wrong, just louder.
That is also why the guide states prevalence as a range instead of a headline. Explaining why a big, dramatic number is the wrong number to lead with is probably the most persuasive thing on this page, more persuasive than any single fact, because it shows the standard applied even against the site's own best story.
This is the center of this page. Everything below was live on this site, wrong in some way, and fixed on 31 July 2026. The reasons given are the actual reasons, not the tidy version. A correction log that only ever says "we updated our sources for clarity" is not a correction log, it is a press release. This one names the specific failure each time.
| Claim | Was | Now | Why |
|---|---|---|---|
| Shower temperature | No established number exists. Go as hot as feels soothing, not as hot as you can stand. Never above 120 F. | The 109 to 112 figure had no source anywhere on the old site. It sits almost exactly on the TRPV1 heat activation threshold of 43 C, 109.4 F, which was measured by patch clamp on a receptor grown in a dish of cells, not in a shower on a person.[2] The only published clinical pathway that names a temperature caps it lower, not higher, at 100 F.[3] Burns in this population are documented, including a case requiring ICU care for sepsis after compulsive hot bathing.[4] | |
| HaVOC trial numbers | 31% (4 of 13) versus 59% (10 of 17), first period of a crossover trial, confidence interval crosses zero. | The published paper says 31% versus 59%.[5] The old "61" is almost certainly the trial's own confidence interval bound, negative 61%, on the difference between the two arms, misread at some point as a proportion and copied forward from there. | |
| Time to diagnosis | Removed entirely. | That figure is not in GRACE-4. GRACE-4 is an emergency department treatment guideline, not a study of diagnostic delay, and would not naturally produce that statistic. Two independent reads of the guideline found nothing resembling it, and no source for it could be found anywhere. It is replaced with the verified Zimmer figures, 17.9 emergency room visits and $76,920.92 in combined imaging and testing costs per patient before diagnosis.[6] | |
| The death anecdote | Removed and replaced with the forensic case series documenting three deaths. | No published case, news report, or named patient account matching that story could be found anywhere, after checking the peer reviewed fatal case literature and every named CHS death in the press. An uncited specific death claim is the single most damaging thing a hostile reader could find on a site like this, so it is gone. What replaces it is real and citable: a forensic series where CHS was the determined cause of death in two cases and a contributing factor in a third, ages 27, 27 and 31.[7] | |
| Prevalence | A range of roughly 8% to a third, with each estimate's population and screen stated next to it. | Three estimates exist and they measure different things. 32.9% comes from 155 smokers screened in one urban emergency department, using hot shower relief and no vomiting frequency threshold at all.[8] 17.8% comes from a national sample of daily or near daily users.[9] Roughly 8% lifetime and 6% past year comes from a state survey where people self identified the diagnosis.[10] Publishing only the largest of the three was number shopping. See chapter three for why the size of a number is not the same as its accuracy. | |
| The hot shower as a diagnostic sign | Presented as raising the probability of CHS, never as confirming it, with the specificity evidence stated plainly. | Hot water relief is not specific to cannabis users. In the largest cyclic vomiting syndrome survey ever run, hot water relief was explicitly "not pathognomonic of marijuana use," and roughly half of cyclic vomiting patients who do not use cannabis report the same relief.[11] In a 2026 national survey, fewer than half, 47.2%, of people whose symptoms fit CHS said hot water helped them at all.[9] This was the site's best story, the one that made people feel truly seen. It was also too strong, and a claim that feels the best is exactly the one that most needs checking. | |
| Recovery timeline | A distribution with honest tails, from days to years, and a clear statement of how thin the underlying data is. | One to two weeks described roughly the fastest group and nothing else. In the only case series with meaningful follow up, only 10 of 98 patients were tracked at all, and of the 7 who stopped cannabis, 6 fully recovered over 1 to 3 months.[12] Families reading the old number and still vomiting in week three were losing hope over a statistic that was never representative in the first place. | |
| The 1894 Commission | The primary 1893 to 1894 report was read directly, and the claim reframed with its own search limits published alongside it. | A historical claim carrying this much weight should not rest on a summary of a summary. The full report was read at the source.[13] A load bearing claim earns the extra work. | |
| Concentrate potency | Measured laboratory values from three primary sources, stated by state and year. | Primary lab data existed and was reachable the entire time. Colorado 2025 dispensary testing measured a mean of 70.7%,[14] an independent 2026 Colorado sample measured a median of 72.6%,[15] and Washington state testing put inhalable extracts near 85%.[16] A factsheet citation was laziness when the actual measurements were one search away. | |
| ICD-11 | Stated accurately: ICD-11 code DD90.4 is titled "Functional nausea or vomiting" and lists CHS as a synonym under that broader entity, not as its own titled code. | The earlier framing overstated what the registry actually shows, and it traveled through several outlets carrying the same overstatement, which is one source repeated, not several confirming each other. This project had already made one WHO versus CDC attribution error before this correction, which is exactly why it got a second, careful look.[17] | |
| The Cureus hypokalemia case | Cited accurately as a cannabis and electrolyte danger case, not a CHS case. | The potassium of 1.8 mmol per liter and the QT of 628 milliseconds in that report are both real and confirmed. But the patient in that specific case explicitly denied vomiting, and the authors of the paper distinguish their case by that exact absence. It is a real warning about cannabis and electrolytes. It is not evidence about CHS.[18] |
| Claim | Was | Now | Why |
|---|---|---|---|
| How long cannabis withdrawal lasts | Starts 24 to 48 hours after last use, peaks day 2 to 6, and commonly runs 2 to 3 weeks or longer. Sleep problems typically peak after the second week rather than settling. | The old wording rested on a 2003 study and was not wrong so much as too short. The 2022 clinical management review in Addiction, which this site had listed as unreachable, was sent in by a reader with library access and read in full. It puts the general tail at two to three weeks or longer and puts the peak of sleep disturbance after day 14. That matters more than accuracy alone: this sentence exists so that somebody still feeling terrible weeks after quitting does not conclude the CHS has come back. Telling them it eases by day 14 produced exactly the wrong conclusion on day 16. |
If you find a claim on this site that is wrong, out of date, or misread its own source, I want to know. Send the page, the claim, and what you think the source actually says to corrections@thehotshower.org. Corrections that hold up get added to this table with a date, whether or not they are flattering.
Almost no health site publishes what it failed to read. That silence is easy to mistake for completeness. It usually is not. Below is every source this project tried to reach for the current version of the site and could not, exactly what stopped it, and exactly what would get past it. Nothing on the site rests on these sources as if they had been read. Where one of them would settle an open question, that question is marked open, not quietly resolved with a citation to a paper nobody here actually read.
| What | The wall | What would get past it |
|---|---|---|
| Peterson 2026 conference results, full section | SAGE journal paywall. Unpaywall shows closed access. Two separate attempts through an institutional repository returned a 403. | Institutional journal access, an author's own copy, or asking a co-author directly. |
| The 2024 Gastroenterology paper on CHS versus cyclic vomiting recovery | The publisher's site returned a 403 on the full text. | Institutional access. This is the single highest value unread paper for the recovery page, since its title implies some non resolving cases are misdiagnosed cyclic vomiting syndrome rather than unresolved CHS. |
| Collins 2023, methods section | PMC returned a CAPTCHA loop, Europe PMC returned a 429 rate limit response. | Retrying outside the rate limit window, or pulling the PMC record directly by its identifier. |
| A German language dronabinol and hyperemesis case report | No metadata could be retrieved through any route tried, including PubMed directly. | A working PubMed session, or access through a German medical library. |
| Habboushe 2018, methods detail beyond the published abstract | Wiley 403, PMC CAPTCHA loop. | Institutional access. |
| The WHO ICD-11 registry record itself | icd.who.int is a JavaScript rendered application, and its API requires OAuth client registration this project did not have. | Registering an ICD-11 API client, or opening the browser interface manually and reading the rendered page. |
| Rubin and Comitas, Ganja in Jamaica (1975), and Carter, Cannabis in Costa Rica (1980) | Neither monograph is digitized anywhere this project could find. | A university library with the physical volumes or an interlibrary loan. |
| Indian Hemp Drugs Commission, Volumes II through VII, the witness testimony | Not a hard wall, a time and budget one. The text is available on archive.org and simply was not sampled for this rebuild. | More reading time. This one is honestly just unfinished, not blocked. |
If you have institutional library access and want to help, the most valuable send is now the 2024 Gastroenterology recovery paper. Forwarded to the correction address above, it would let this site answer a question it currently has to leave open.
A reader with library access read this table and sent the papers. The Connor 2022 withdrawal review is now read in full, and it corrected a claim that was live on this site, which is logged above. The Freeman regional potency figures were closed a different way, by a later open access paper that tabulates the same ground for the United States by region (ElSohly MA, Majumdar CG, Chandra S, Radwan MM, Frontiers in Public Health 2024;12:1442522).
This is the whole reason the table is published. It is kept up to date rather than quietly trimmed, so what is still missing stays visible.
Separate from anything this project could not personally reach, these are questions nobody in the published literature has answered yet. No amount of better searching fixes these. The studies that would answer them have not been run.
Nobody has separated a true rise in cases from more diagnosing, more heavy users, or a coding artifact. The closest thing to the ideal natural experiment, comparing CHS rates in Quebec, which caps THC potency and bans most concentrates, against the rest of Canada, cannot currently be built, because Quebec is entirely excluded from the national hospitalization database used for that kind of analysis. That absence is itself worth stating plainly rather than leaving unexplained.[19]
TRPV1 receptor activation is the leading hypothesis, with blood flow redistribution and a hypothalamic thermostat effect as alternates. The most thorough recent review states plainly that the mechanism has not been empirically validated.[20] The relief is documented. The reason for it is not.
In one national survey, roughly 82% of daily or near daily users did not screen positive for CHS.[9] No prospective cohort study, following a group of heavy users forward in time to see who develops CHS and who does not, has ever been run.
No longitudinal study has tracked CHS patients past the first three months of abstinence. Community reports of the long tail run from six days to four years, which is either a wide range of biology or a wide range of confounding, and nobody has designed the study that would tell you which.
A single small study reported five candidate genetic variants in 2022. Four years on, there has been no independent replication and no genome wide study of CHS at all.[21] An unreplicated finding does not become more true with age. It just gets older.
This has never been measured. Case reports exist of superior mesenteric artery syndrome, pheochromocytoma, and diabetic gastroparesis being missed or delayed under a CHS assumption, but no study has ever counted how often that happens across a population.[22]
CHS has no confirmatory test and no validated biomarker. An audit of 184 emergency department cases concluded the proposed diagnostic criteria have "not been externally validated," and their sensitivity and specificity have never been measured.[23] That is a finding, not an oversight, and it belongs on this page rather than buried in a methods appendix somewhere.
A site that will not say what would prove it wrong is not being careful, it is being unfalsifiable, and unfalsifiable is not the same thing as honest. Here is what would force a real rewrite of this site, stated in advance rather than after the fact.
A properly powered randomized trial showing topical capsaicin clearly works, or clearly does not, for CHS abdominal pain. The current recommendation rests on one underpowered 30 patient pilot.
A prospective cohort study with a real denominator, following a defined population of cannabis users forward in time to measure who actually develops CHS. No such study exists today.
An independent replication, or a clear refutation, of the 2022 candidate gene findings. Either result would move that chapter from speculation to something firmer.
A longitudinal study following CHS patients past one year of abstinence, with a real sample size. This would let the recovery page state an actual expected timeline instead of a distribution with an open ended tail.
A demonstration that the association between rising potency and rising CHS cases is substantially confounded, for instance by a study showing the rise tracks something else entirely, like screening behavior or a coding change, once potency is properly controlled for.
If any of these five land, this page and the pages they touch get rewritten, with the same dated correction format used in chapter four. Naming that in advance is not a formality. It is the strongest signal available that the confidence expressed elsewhere on this site is actually earned and not just asserted.
Every medical claim on this site, numbered and linked to its source, lives on the evidence page. What follows here is the short version of how this project treats different kinds of sources, so a reader can judge any claim on sight without clicking through every time.
A peer reviewed original study, read in full, not through its abstract. Used for the strongest claims on the site, always with sample size and design stated.
Posted but not yet peer reviewed. Used, but always labeled as a preprint next to the number, every time it appears, not just the first time.
Data shown at a conference but not published in a journal. Treated as one step below a preprint. Used sparingly, and never for a number the site leans on hard.
A hospital's own published protocol, such as the Johns Hopkins All Children's or Children's Minnesota pathways. Treated as a fact about current practice, not as a research finding, and named directly because a link to a real hospital's own pathway is worth more to a reader in an ER than anything this site could write on its own.
State cannabis regulator data, CDC coding records, consumer safety data. Treated as reliable for what it measures directly, and flagged when a figure is an extrapolation the agency itself did not make.
Used only when it reports something independently verifiable, like named lab testing results or a named source's own quote, and never as the sole source for a clinical or statistical claim.
Forums, support groups, and self selected surveys of people who already suspect they have CHS. Genuinely useful for what a lived pattern looks like. Never used for a rate, a percentage of the general population, or a prevalence claim, for the exact reason explained in chapter three.
This page will need another correction someday, and when it does, it goes in the table in chapter four with a real date and a real reason, the same as everything else. That is not a weakness in the plan. It is the plan.
The full cited case for CHS, built to be read by a skeptic, lives on the evidence page. The practical guide, what to do tonight and what to say at the ER, is where the help actually is.
These are the sources cited on this page specifically. The complete cited case for CHS, with every source graded, lives on the evidence page.