Short answer: yes, and you're right to ask for proof. This page rebuilds the case from scratch, corrects three things the older version got wrong, and puts the weak spots in the open instead of the footnotes. Read it and decide for yourself.
You don't have to take a stranger's word for it. On October 1, 2025, the US diagnostic coding system added a specific code for CHS. It has an official number that a hospital anywhere in the country can bill under, today.
Codes don't get invented for hoaxes.
A condition earns one after enough hospitals, enough billing, and enough peer reviewed papers force the system to name it. The first medical description came in 2004.[1] It took twenty years and a wave of ER visits to get here.
One honest correction belongs right here, because it's the kind of thing a skeptical reader checks first. In 2025 several outlets reported that the World Health Organization gave CHS its own code too. That overstates it. ICD-11 code DD90.4 is titled "Functional nausea or vomiting," a broader entity, and CHS is listed under it as a synonym, not as a distinct titled condition in its own right.[6] The code that is unambiguously CHS's own is the American one above. Two things can be true: the WHO record is real and worth citing accurately, and the CDC code is the stronger claim.
The recognition is not only a billing code. By 2024 the American Gastroenterological Association had issued a formal clinical practice update on diagnosing and managing CHS,[3] and a multi society emergency medicine guideline, GRACE-4, set out ED management steps on the strength of the evidence available, most of it rated low or very low certainty, but a guideline nonetheless.[4] Two children's hospitals, Johns Hopkins All Children's and Children's Minnesota, built their own clinical pathways around the same recognition, and a family can hand either one to an ER doctor who hasn't seen it before.[7][8]
If CHS is new, something must have changed. It did, and the honest version of that story is more interesting than the simple one, because part of it plateaued and part of it didn't.
That plateau is the surprising finding this rebuild has to lead with honestly, because it cuts against the simple "potency just keeps climbing" story. Washington's own market regulator found flower holding roughly steady near 22% for six straight years.[15] Colorado's independent measurements land close by: mean 20.8% in one 2025 sample, median 18.1% in a separate 2026 sample.[13][14] Flower stopped climbing years ago. If flower were the whole story, CHS trend lines should have plateaued with it. They haven't, which is itself a clue about where to look next.
Where the climb didn't stop is concentrates, and here the site can now cite measured lab data instead of a DEA factsheet. In Colorado, ninety nine concentrate products tested across fifty two dispensaries in nineteen counties averaged 70.7% THC, with a range up to 83.6%.[13] A separate, independent Colorado sample found a median of 72.6%.[14] Washington's inhalable extracts averaged almost 85% in 2023, up from roughly 70% in 2017.[15] That is a potency with no natural analog anywhere in the plant's history.
Colorado flower measured 2.2 to 4.25 percentage points below its own label. One sample's measured mean was 20.8% against a labeled 22.5%. A separate sample measured a median of 18.1% against a labeled 23.1%, a difference of 4.25 points that was not chance (95% CI 2.83 to 5.96, p<0.001). Only 56.7% of flower products landed within 15% of their label at all, and 30.3% were actually over labeled.[13][14]
Concentrates, by contrast, were accurate: 96% fell within 15% of the label.[13] If you've ever bought flower that felt weaker than the number on the jar, the lab data says you were probably right.
The product mix moved even where potency held flat. In Washington's 2023 sales, flower was 35% of the market, vape pens and cartridges 26%, pre rolls 16%, and everything else 23%. Total THC sold through flower actually rose 70% from 2017 to 2023, simply because more flower moved, while THC sold through extracts and edibles roughly tripled.[15] People are buying more of everything, and a larger share of it is concentrated.
At the same time, CBD, the part of the plant that buffers THC's effects, was bred out. In 1995 the ratio of THC to CBD sat around 14 to 1. By 2014 it was roughly 80 to 1.[11]
A global meta analysis of 12 studies from 1970 to 2017 found herbal cannabis THC rising at a rate of 0.29 percentage points per year, and cannabis resin at 0.57 points per year, with CBD unchanged across the same period.[12] This site cites that rate and nothing more. The paper's absolute regional starting and ending percentages could not be independently verified across five separate access attempts, so no specific "from X% to Y%" claim is made from it here. When a source is behind a wall, the honest move is to cite what was actually read, not to round up.
A lot of people reach for "they must be genetically modifying it." That instinct is pointing at something true, but the word is wrong, and it's worth being precise, because being precise is what makes the rest of this page trustworthy.
There is no genetically modified cannabis for sale, anywhere. No transgenic strains, no CRISPR flower on dispensary shelves. What actually happened is selective breeding, the same kind of thing that turned wild mustard into broccoli and cauliflower, plus sinsemilla growing, feminized seeds, and solvent extraction. Real gene editing cannabis science exists, but it lives in yeast vats and university labs, not the stuff people smoke.[16][17]
So if you've said "that's not GMO," you're right. But here's the part that matters: your deeper instinct, that we've pushed this plant far past anything nature or history ever made, is dead on. Bred for maximum THC, stripped of CBD, extracted to eighty five percent or more, and used all day long. Not genetically modified. Genuinely re-engineered.
Cannabis is one of the oldest medicines on earth, and for most of that history it was used to stop vomiting, not cause it. The full three thousand year treatment, every era, every source, now lives on the history page, so this chapter carries only what a skeptic needs mid argument, plus one correction this rebuild insists on making carefully.
The older version of this page said the Commission recorded no cyclic vomiting among heavy users. That claim came from a summary, not the source, and it overreached. Read directly, Volume I discusses digestive harm from habitual heavy use at length, including this from Dr Prain: "digestion is impaired, the daily consumption of such a dose of narcotic overburdens the alimentary and eliminating systems, the man's appetite goes, and the food he takes is not properly absorbed." One passage names an acute effect, not a cyclic one: "the man becomes peevish, stupefied, sees double; and occasionally it may cause vomiting." And a clinical account the Commission itself quotes, O'Shaughnessy's, reports the opposite of a vomiting illness: "no nausea or sickness of stomach succeeds, nor are the bowels at all affected next day."[9]
What is actually true: the words "bath," "bathing," "hot water," "cyclic," and "periodic vomiting" do not occur anywhere in Volume I. That is a real and searchable fact. It is not the same claim as "no cyclic vomiting was recorded," because the Commission was never asked about episodic vomiting or bathing behavior, nineteenth century Indian bathing customs would have made hot bathing unlikely to surface even if it happened, the OCR is machine read from a century old scan and could hide a garbled mention, and Volumes II through VII, the raw witness testimony, were not sampled. The honest claim is narrower than the old one: we looked, here is exactly what we searched, and here is why that search has real limits.[9]
One more correction belongs here. The 2021 trial usually cited as proof haloperidol beats standard ondansetron for CHS is real, but the older version of this page misquoted its numbers. The actual finding: 31% of patients on haloperidol needed rescue medication against 59% on ondansetron in the first treatment period, a real difference, but with a confidence interval on that difference that crosses zero.[52] Promising, not settled. That is the more useful sentence.
In the department
Haloperidol beat ondansetron in the only randomised trial. Thirty patients, first period of a crossover, and the interval on the difference crosses zero. Electrolytes and an ECG come before any of it.
A chapter that leads with a thousand percent headline while the newest numbers show a plateau is selecting for alarm, and this rebuild exists partly to stop doing that. So this chapter leads with the newest and strongest evidence available, the whole curve, including the fall.
Canada's national data says the same thing more directly, because the researchers ran the statistics on the flattening itself. Hospitalizations rose from 1.5 to 3.8 per 100,000, a 2.5 fold increase, from fiscal year 2016-17 to 2024-25, 7,533 hospitalizations in total. Growth ran at 23.5% a year through 2020-21. After that it flattened to 3.6% a year, and that slower growth was not statistically significant.[19]
Canada added its own diagnostic code, ICD-10-CA K92.81, in April 2022, which is itself a sign of how seriously the trend was taken while it was still climbing.[19]
Ontario's data also breaks by sex, and the split is worth carrying. Women showed a real increase, relative rate 1.49 (95% CI 1.16 to 1.92). Men showed no statistically significant change, relative rate 1.08 (95% CI 0.85 to 1.37). The female share of CHS related ED visits rose from 47.5% to 53.5% over the study period.[20] Nobody has a settled explanation for why.
Two more regional data points, presented as regional headlines rather than the whole picture: a Massachusetts analysis of more than 15 million ED visits from 2012 to 2021 found CHS coded cases up more than 1,350%,[21] and Colorado's prevalence ratio for cyclic vomiting ED visits nearly doubled after cannabis liberalization.[22] Both are real. Neither is the newest or the most complete picture, which is why this chapter put the national curve, rise and fall together, first.
This is the part that makes the case personal. Because so few doctors recognize CHS, patients get bounced through the system for years, running up scans, scopes, and even surgeries that find nothing, while the real cause goes unnamed.[23]
An earlier version of this page also cited a "4.1 years to diagnosis" figure attributed to the GRACE-4 guideline. It isn't there. Two independent reads of GRACE-4 and a full read of the Zimmer paper above found no such number anywhere, so it has been removed rather than left in with a shaky footnote.
Part of the gap is stigma, not just unfamiliarity. In a 2024 survey of 249 people, 27.8% said cannabis use had never been discussed with a health care provider at all, and anticipated stigma, not the drug itself, was the factor most strongly associated with staying silent.[25]
The stakes are not abstract. The only rigorous forensic case series found three deaths where CHS was the determined cause in two and a contributing factor in the third, ages 27, 27, and 31.[26] No published study reports a case fatality rate, so no percentage claim about CHS being fatal in some share of cases is supportable, and none is made here. Short of death, low potassium during severe vomiting can prolong the heart's QT interval into dangerous territory, documented now in separate pediatric case reports.[27][28] One related Cureus case is sometimes cited alongside these and shouldn't be: a daily cannabis user with a potassium of 1.8 mmol/L and a QTc of 628 ms, both figures confirmed exact, but the patient explicitly denied vomiting or diarrhea and the authors themselves distinguish the case by that absence. It's a real cannabis and electrolyte danger. It is not a CHS case, and this page won't describe it as one.[55]
That is the entire reason a page like this exists. The condition is common, brutal, and expensive, and it is still mostly invisible to the system meant to catch it.
Nearly every person with CHS discovers the same bizarre thing on their own, before they've read a word about it: a scalding hot shower stops the vomiting, at least for a while. In the original 98 patient Mayo Clinic series, 91% got relief from hot water.[2] A pooled systematic review across the published literature puts it at 92.3%.[29] When thousands of strangers independently land on the exact same strange fix, something real is going on.
The older version of this page called the hot shower "so specific that doctors now use it as a diagnostic tell." That statement does not survive contact with the largest cyclic vomiting syndrome survey ever run, and it needed to come out of this page entirely, not get quietly softened.
In that survey, hot showers were found "not pathognomonic of marijuana use." Sixty seven percent of all cyclic vomiting syndrome respondents, most of whom did not use cannabis regularly, used hot showers or baths for relief. The association with marijuana use was real but modest, odds ratio 2.54 (95% CI 1.50 to 4.31, p=0.0006), not the near certain marker the older framing implied.[30] A separate 111 patient cyclic vomiting cohort found 73% used hot water bathing, with more than 80% getting marked improvement, in a group where only about 17% were regular cannabis users at all. Regular users did lean toward hotter water, choosing "very hot" 50% of the time versus 16% for others, a real but partial signal.[31]
And in the 2026 US national survey, only 47.2% of people whose symptoms fit CHS reported that a hot bath or shower improved them at all, a lower number than the older clinical case series because it asks a broader, less selected population.[24]
Hot water relief raises the probability of CHS. It does not confirm it, and its absence does not rule it out. Roughly half of cyclic vomiting patients who don't use cannabis get the same relief.
Why hot water works at all is still unsettled, for anyone. The leading idea is that heat activates a receptor called TRPV1 that chronic cannabis use dulls, which is also why capsaicin cream helps some people. But there are other live theories, and the most thorough 2024 review states plainly that the mechanism "has not been empirically validated."[32] Anyone who tells you the mechanism is settled is overselling. The relief is real and reliable. The explanation is still a working hypothesis, and now so is the diagnosis it was once mistaken for confirming.
Almost nobody in CHS communities doubts CHS is real. They doubt the cause, and pesticides, dirty grow operations, and adulterated supply are the explanation offered most often. It's an understandable place to look, and it deserves a real answer, including the part that's actually true.
The unregulated hemp derived THCA market, sold under the 2018 Farm Bill's hemp definition, is genuinely less tested than the licensed cannabis market. Independent spot testing has found THCA flower and vape products with illegal levels of THC relative to their labels, microbial contamination, and myclobutanil, a pesticide that converts to hydrogen cyanide when heated.[56] Congress banned Delta-8 and THCA under a new hemp law in December 2025, which is itself an admission that the gap in testing was real. No study yet links that specific supply to CHS case counts, so this site says the market is worse tested without claiming it causes CHS. That distinction matters, and it isn't a dodge.
Then the evidence on the theory itself. Pesticide test failure rates in the legal, regulated markets are low: Washington 0.01 to 0.03% from 2022 to 2025, Oregon 1 to 4% from 2016 to 2025, California about 5% overall in 2022, with extracts failing more than four times as often as flower.[33] Azadirachtin, the neem derived compound blamed most specifically, has an acute oral LD50 in rats above 3,540 to 5,000 mg per kg and sits in EPA toxicity category IV, "practically nontoxic." The residue level a person would actually be exposed to is nowhere near a vomiting dose.[35]
Fentanyl gets raised often enough to name the investigations. Across roughly 12 million seized drug samples analyzed from 2013 to 2023, cannabis had the lowest fentanyl co-occurrence of any drug class. Named investigations that chased specific suspected clusters and came back negative or nearly negative: Vermont, DEA and Brattleboro Police, November 2021; Connecticut, state health officials, February 2022, 40 claimed cases and 1 positive attributed to cross contamination; Michigan, state poison control, December 2021, 8 cases, none confirmed; Massachusetts, Hampden County District Attorney, December 2021; New York, 2019, lab negative.
The decisive shape of the evidence is this: CHS occurs where no plant contaminant can possibly reach. There is a published case caused by synthetic cannabinoids, no plant material involved at all.[36] A majority of the 2026 survey's respondents, 61.9%, sourced from licensed, tested dispensaries, not the dirty end of the market.[37] And by the survey team's own count, roughly a quarter of episodes were triggered by stopping cannabis rather than using it, a pattern no contaminant can explain, since nothing sprayed on a plant can poison you once it's out of your system. That particular figure comes from the team's conference presentation of this dataset rather than the peer reviewed paper itself, and has not been independently confirmed, so treat it as a real but unconfirmed data point, not a settled one.[37]
In a Washington state survey, 76% of self reported CHS cases involved any illegal source cannabis, against 24% legal source only.[34] That looks like it supports the contamination theory, and it deserves an honest explanation rather than a disappearing act. It's self reported, single state, and almost certainly confounded: people relying on illicit supply tend to use more heavily and for longer, and frequency and duration are the two strongest known risk factors for CHS regardless of where the product came from. The number is real. The explanation for it probably isn't contamination.
None of this means cannabis is always clean, or that testing shouldn't improve. It means the syndrome itself is driven by heavy, high potency THC exposure, not by what's sprayed on it.
This is the honest counterweight, and it belongs high in this chapter, not buried. In a nationally sampled 2026 US survey of 7,034 people, 17.8% of daily or near daily cannabis users screened positive for CHS symptoms, meaning roughly 82% did not.[24] No prospective cohort with a clean denominator, tracking the same group of heavy users forward through time, has ever been run. That study is the single most useful thing that doesn't exist yet in this field.
The genetic story you may have heard rests on one small, unreplicated study, and this page says so plainly. Russo and colleagues genotyped 28 CHS patients and 12 controls, not the 205 people who met full survey criteria in the same project, and reported five candidate variants: COMT (odds ratio 12.0, p=0.012), TRPV1 (5.8, p=0.015), CYP2C9 (7.8, p=0.043), DRD2 (6.2, p=0.031), and ABCA1 (8.4, p=0.012).[39] No correction for multiple comparisons was applied. Run five tests at the standard 0.05 threshold and the chance of at least one false positive by pure chance is roughly 23%. A stricter Bonferroni threshold would clear none of the five. Four years on, there is no independent replication and no genome wide study of CHS exists. A 2024 review by the same lead author restates the same five variants without new validation.[40] The study itself is a fact. The mechanism it points to is still speculation.
One candidate is more mechanistically concrete than the rest: CYP2C9, an enzyme that helps clear THC from the body. People homozygous for the CYP2C9*3 variant show roughly threefold higher THC exposure from the same dose, with about 70% less of the inactive breakdown product THC-COOH, in a study of 43 healthy volunteers.[41] But this variant is rare, present in 3 to 4% of Southern European and Eastern Mediterranean populations and under 1% in Asian and African populations.[42] Too rare to explain most cases. Worth watching. Not an answer by itself.
In the same 2026 published survey, 96.5% of respondents used cannabis at least daily, 45% used six or more times a day around the time symptoms began, and 65.4% had used for more than three years before onset. Sixty three percent had morning prodromal symptoms, mostly nausea and stomach pain, and women reported more frequent and prolonged symptoms than men. Vape cartridge use was associated with a shorter time to onset than flower, an association, not proof of a separate mechanism.[37] This is a self selected survey of an already affected online population, recruited from CHS focused social media communities. It can never be used to estimate a population rate, an incidence, or a prevalence, and it is labeled that way every time it appears on this page for a reason.
How common CHS actually is depends entirely on how you ask. Estimates in the literature run from about 6 to 8% of past year Washington cannabis users who self identify with the label,[34] to 17.8% of daily users in the national survey above,[24] to 32.9% in a single urban ED where the definition was loose and the authors themselves called it "a phenomenon similar to CHS."[38] That spread is a definitions problem, not three measurements disagreeing about the same thing, and no population representative prevalence estimate exists for any of it.
Youth follow a pattern worth naming separately. Pooled across 21 studies and 211 pediatric and young adult cases, the median age of first cannabis use was 16, median age of symptom onset was 24, and median age of diagnosis was 28, a full decade after first use.[43]
What is still unknown
Nobody has shown why heat helps. The genetics rest on twenty eight cases and have not been replicated in four years. A page that hides that is not worth trusting.
A page that only argued one side wouldn't deserve your trust. So here's what's genuinely unresolved, stated as plainly as the rest.
Potency rose and CHS appeared. That timing is striking, and it's the best explanation available, but it is not proof. One of the researchers behind the Colorado prevalence work said outright that the data cannot distinguish "greater use, improved reporting, or both." No study has ever expressed CHS incidence over a per heavy user denominator tracked across time, and no natural experiment comparing CHS rates across a THC potency policy boundary exists, despite real attempts to find one.
The closest thing to that ideal study would be Quebec, which caps THC at 30% and bans most concentrates, compared against the rest of Canada. That comparison currently cannot be built, because Quebec is entirely excluded from the national hospitalization database used for the Canadian trend data above, "due to differences in collection and reporting systems," in the researchers' own words. That is why the study you'd actually want doesn't exist yet, and saying so plainly is more honest than leaving the gap unexplained. The closest available per user estimate, roughly 0.5 to 0.6% of frequent cannabis users aged 15 to 44 needing an ED visit for CHS in Alberta after legalization, comes from a secondary source that could not be independently verified, so treat it as a rough signal, not a rate.
The same recognition that finally names CHS correctly in some patients can also get slapped on too fast in others, before ruling out something else. In one documented case, a 17 year old using cannabis five to six times a day and meeting full criteria for CHS also had superior mesenteric artery syndrome, a real surgical problem, and her acute presentation "could have simply been attributed to exacerbation of her chronic CHS symptom pattern, leading to less comprehensive evaluation."[45] The next chapter goes deeper on exactly how far this argument can and can't be pushed.
Some of what's repeated as fact rests on thin ground. The widely quoted capsaicin cream recommendation traces to one underpowered 30 patient pilot trial.[53] The genetic findings covered in the last chapter come from a single small unreplicated study. The brain chemistry side is thin too: chronic daily smokers showed roughly 20% lower cortical CB1 receptor density than healthy controls, and after about four weeks of continuously monitored, locked unit abstinence, most of that downregulation moved back toward normal in cortical regions, though not in subcortical ones, and sources differ on whether the recovery was full or partial.[54] That's a finding about brain chemistry on a research ward, not a promise about when anyone's vomiting actually stops. Twenty years in, this field is still early, and this page would rather say that than dress the evidence up.
A fair question, and the only place on this whole site where the honest answer gets more nuanced than "never again." A minority clinical view holds that some people, after a long break and a lot of self awareness, can go back to a small amount of low potency flower, never concentrates, and stay well by watching closely for their own early warning signs. That path is real for some people. It is also unproven at scale, it needs a doctor who actually knows this syndrome, and the relapse data cuts hard against it: 79.4% of people in one self selected survey returned to cannabis after diagnosis,[39] and roughly 70% of those who tried resuming in a separate, not yet peer reviewed count from the same research team had it return.[37] So the message everywhere else on this site doesn't change: total abstinence is the only cure anyone can count on. If you're standing in that shower tonight, that's still the plan to trust. This nuance exists here, on the evidence page, and nowhere else on the site, on purpose.
Naming your strongest critic and answering him is the cheapest credibility a site like this can buy, so this is the chapter that does it.
In February 2026, an economist named D. Kief posted a working paper on SSRN making the sharpest argument against high CHS prevalence estimates that exists in print.[48] The logic: cessation is the accepted cure, so if prevalence is genuinely high, quitting should show up as a visible dent in cannabis market volume. Using 319 weeks of Florida medical marijuana seed to sale data and consumption baselines near 1.0 gram per day across seven jurisdictions, the paper models that even a 1% annual CHS incidence should remove 15.8 to 31.6% of market flower volume over six years. It argues that at a 32.9% prevalence estimate, "the entire market would need to exist more than once over within the first year." No such signal shows up. Per participant consumption in the data is stable or rising. The paper separately attacks the hot shower screen as non specific, using very similar reasoning to chapter seven above.
Status, stated plainly: an SSRN working paper, single author, not yet peer reviewed. It should still be answered on the merits rather than dismissed on the venue.
The honest answer: the inference only works if people with CHS quit permanently and leave the retail market for good. Most do not. Relapse runs 79.4% in one published survey and roughly 70% of temporary stoppers in another.[39][37] Flat or rising per user consumption is equally consistent with high prevalence and low, unsuccessful cessation, which is exactly what a real syndrome with an unpopular cure would produce in market data. Where Kief is right, and this page now concedes it without qualification: the hot shower screen is not specific, and chapter seven of this page says so in full rather than in a footnote.
Jordan Tishler, an emergency trained physician and president of the Association of Cannabinoid Specialists, wrote in STAT First Opinion in July 2025 that ED physicians have "developed a certain antipathy toward patients they perceive as 'drug users'" and are "attributing any vomiting in a cannabis user to CHS." His diagnosis of the diagnosis is anchoring: "a clinician fixates on an initial piece of information, in this case cannabis use, and fails to consider alternative or concurrent diagnoses."[49] A retired ED physician replied in the letters section a week and a half later: "SMA syndrome is very, very rare," and "cannabinoid hyperemesis syndrome is not."[50]
Both physicians are pointing at something real. Documented cases exist of cannabis users given, or nearly given, a CHS explanation who had something else going on as well: superior mesenteric artery syndrome in a teenager whose weight loss was itself caused by her CHS,[45] a series of nine adolescents with both chronic cannabis use and confirmed SMAS,[46] and a 69 year old daily user whose presumed cannabinoid hyperemesis turned out to be an eight centimeter adrenal pheochromocytoma driving the retching through hypertensive urgency.[47] An audit of 184 adult ED presentations for suspected CHS concluded the proposed criteria have "not been externally validated and thus its utility, particularly to the clinician at the bedside in the ED, has not been established."[44]
Here is where this page has to be as careful with the skeptical case as it is with the believing one. There is no published series of patients harmed by a wrong CHS diagnosis. Nobody has measured how often a CHS diagnosis gets revised after the fact. Nobody has validated the diagnostic criteria against a gold standard. And in the same 2026 national survey cited throughout this page, only 11.5% of people whose symptoms actually fit CHS had ever been told what they had.[24] Right now, the dominant error in the system is still that nobody names it, not that everyone jumps to it.
Both are true at once, and holding both is a credibility asset, not a weakness. One separate figure this site deliberately does not use: a widely repeated "2.75 million Americans" extrapolation from a single 2018 paper, publicly and fairly criticized as resting on "a single 2018 paper with a poorly sampled study group and an outrageously broad definition of CHS."[51] This page doesn't use that number, and it isn't starting now.
This isn't about whether cannabis is good or bad, or a political argument in either direction. Most people who use cannabis will never get this. CHS is a specific syndrome that shows up in a subset of heavy, chronic, high potency users, roughly one in five even among daily users, and it's typically young adults, though clinicians have seen it from teenagers to people in their eighties.
If any of this sounded like your 3 a.m., you already know which part was real. The good news is the same as the whole point: it ends when you stop.
If the evidence landed, the guide is where the practical help is, what to do tonight, what to say at the ER, and what quitting actually feels like.
Every number above is numbered below with a working link. How each source was graded, and the specific papers this project could not get past a paywall or a login wall, are written up in full on how this was made.