Diphenhydramine

Diphenhydramine is a first-generation over-the-counter H1 antihistamine whose sought-after effect is not an effect of its own but the shadow of a central muscarinic blockade. It is the molecule that psychedelic research uses as an active placebo, and it is also the substance most frequently mentioned in overdose deaths classified as suicide in the United States in 2023. These two facts do not contradict each other: at the doses where it does something, it damages, and its dose-effect curve has a shape that almost no other substance here shares. Beyond about one and a half grams, the frequency of delirium stops rising, while coma and convulsions carry on. One cannot take more of it to get more of it.

Neurochemical action

Three actions are superimposed, with three different dose thresholds, and it is the offset between them that does everything.

H1, sedation. An inverse agonist of the H1 receptor, which crosses the blood-brain barrier. The cleanest measurement on this page is an 11C-doxepin PET study in eight healthy volunteers: 30 mg by the oral route occupy on average 56.4% of the cortical H1 receptors, with a correlation to plasma concentration. The lowest dose of a standard tablet already saturates more than half of the H1 receptors of the cortex.

Muscarinic, delirium. A competitive antagonist of the muscarinic acetylcholine receptors. This is the mechanism of the anticholinergic syndrome and of the delirious state, and it appears several hundred milligrams higher up. A caveat not to be hidden: the affinities of diphenhydramine at the M1 to M5 subtypes could not be obtained (the Kubo 1987 paper accessible only as an abstract, the IUPHAR database behind a login, a dedicated PubMed search with no result). The declared class of the molecule therefore announces a mechanism that this page cannot put numbers on, and no figure found elsewhere will come to fill that hole.

Sodium and potassium channels, cardiotoxicity. Blockade of the voltage-gated sodium channels, hence QRS widening, identical in form to that of the tricyclic antidepressants. Blockade of the repolarising potassium channels, hence QT prolongation and the risk of torsade de pointes. This third curve overlaps the zone of the delirium. It is the central fact of all the harm reduction here.

To this are added the affinities measured in the rat in 2008, at H1 and at the three monoamine transporters: H1 96 nM, DAT 851 nM, NET 2,070 nM, SERT 9,050 nM. Two things come out of it. Intravenously in the awake rat, diphenhydramine raises extracellular dopamine in the shell of the nucleus accumbens to as much as 340% of baseline, that is, the pattern cocaine produces, and this effect is not mediated by H1: triprolidine, a more potent H1 antagonist, produces nothing. The authors maintain DAT as the site principally responsible, while noting that in vivo potencies do not follow the affinities. And it has about 94 times lower affinity for the serotonin transporter than for H1, the lowest affinity of the four antihistamines in their series. Three caveats inseparable from these figures: animal species, intravenous route, 1 to 3 mg/kg. Nobody has verified that this transposes to a tablet swallowed by a human.

Metabolism proceeds by hepatic N-demethylation, presented everywhere as principally carried by CYP2D6. One measurement contradicts that presentation, and it must be kept: in poor metabolisers, that is, when CYP2D6 is abolished, the disposition of diphenhydramine is only slightly affected. There are therefore sufficient back-up pathways. The practical consequence is asymmetrical, and it is the most poorly understood point about the molecule: diphenhydramine increases exposure to CYP2D6 substrates, and it undergoes almost none of it itself.

What the delirium is, and what it is not

The constant picture is a delirium in the strict psychiatric sense: confusion, disorientation, and no awareness of the unreal character of the perceptions. The clinical descriptions are strikingly consistent: scratching at insects that are not there, talking to absent people, smoking non-existent cigarettes, without ever recognising anything as false. This is a difference in kind, and not in degree, from the serotonergic hallucinogens of the rest of this wiki, where the subject usually keeps track of what is happening to them.

The only case report that really describes the content dates from 1988: a 46-year-old woman, 1,600 mg, a series of highly plastic and rapidly changing sensations of insects, spiders, beetles and ticks, with no equivalent in the real, and which seemed absolutely real to her throughout the episode. Eight hours, complete resolution without treatment.

The closest thing to a collective phenomenology is a peer-reviewed thematic analysis of 32 first-person accounts published on video: alterations in most sensory modalities, a tendency toward strongly unpleasant experiences, positive effects rare and most often eclipsed by the negative aspects, frequent warnings from the video authors themselves, post-acute adverse effects sometimes reported, and, in spite of all that, patterns of repeated misuse. The motives invoked, when they are, are availability, price, legality. The authors open their abstract by writing that deliriants are the least studied class of hallucinogens.

Two honest clarifications. The hat man motif circulates massively in the online folklore attached to this molecule: no peer-reviewed source documents it, and it is an object of social transmission, not a clinical fact. And above all, there is no controlled human study of the deliriant effect of diphenhydramine. Nobody has ever administered a deliriant dose in the laboratory to describe what happens. The only experimental human use of this molecule is that of active placebo, in a psilocybin trial and in a metergoline trial.

Duration, dose and tolerance

At therapeutic doses, an adult takes 25 to 50 mg every 4 to 6 hours, for a daily maximum of 300 mg by the oral route. In children aged 2 to 5 years, 6.25 mg per dose, 37.5 mg per day at most; from 6 to 11 years, 12.5 to 25 mg per dose, 150 mg per day.

Toxicity thresholds, and they overlap. A Swiss series of 282 mono-intoxications places moderate symptoms, among them hallucinations and ECG abnormalities, above 0.3 g, and severe symptoms, delirium, psychosis, convulsions, coma, above 1.0 g. An American dose-response analysis of 509 cases gives an optimal cut-off point at 8.2 mg/kg (95% CI 5.6 to 10.5) and concludes that its results support the poison centres' threshold of 7.5 mg/kg; each tenfold increase of the dose in mg/kg multiplies the odds of a clinically significant response by 47, with a very wide interval, from 17 to 154. The American consensus retains 7.5 mg/kg or 300 mg, whichever is the smaller.

The fact to remember fits in one sentence: the maximum authorised daily dose of the medicine, 300 mg, is also the threshold above which one must go to the emergency department if it is taken all at once. There is no comfortable window between the dose of the medicine and the dose that disturbs.

On lethality, a recent review estimates the adult lethal dose at 20 to 40 mg/kg, and reports paediatric deaths below 500 mg, with convulsions reported from 150 mg, that is, three tablets. These are review estimates, not measurements. A bibliographic warning in passing: one of the two reference clinical syntheses carries, on this lethal threshold, a published unit error by a factor of ten. The usable value is 20 mg/kg, and a figure in grams per kilo, on this molecule, should raise the suspicion of a typo.

Blood concentrations are of no use. Five infants of 6 to 12 weeks died with 1.1 to 1.6 mg/L in post mortem blood and a normal anatomy. An adolescent girl survived 2.24 mg/L after 1,800 mg, an adult 2.5 mg/L. The clinical syntheses say it plainly: plasma levels provide no clinically relevant information, and they do not compare from one age to another.

On duration, no source gives a validated range for the deliriant effect. What is available: onset of symptoms 1 to 2 hours after ingestion, an 8-hour hallucinatory episode after 1,600 mg, a full picture established in under 5 hours after 1,800 mg with resolution only on the fourth day. The Tmax itself is not stable from one source to another, 1.5 hours for one, 2 to 3 hours for the other. The half-life depends on age: about 5 hours in the child (4 to 7), 9 hours in the adult (7 to 12), 13.5 hours in the elderly person (9 to 18). And the anticholinergic effect slows gut motility, which delays and spreads out absorption: the peak can arrive later than expected.

Tolerance, and the only clean study does not measure what one would want. Fifteen healthy men, 50 mg twice a day against placebo, double-blind. By the fourth day, drowsiness is indistinguishable from placebo and the impairment of performance is completely reversed. That is a tolerance to sedation, not to delirium. Nobody has measured whether tolerance also develops to the central anticholinergic effect. The question remains open in both directions, and it matters, because patterns of repeated misuse are described.

Dependence exists, it is rare, it is documented: a review finds five case reports, six patients, published between 1986 and 2001, all meeting the DSM-IV criteria, with an apparent link to antipsychotic use. The index case is a 56-year-old woman taking up to thirty 50 mg tablets a day.

Harm reduction

The most banal danger lies in the fixed combinations. A large part of the diphenhydramine sold to the public is combined with paracetamol or ibuprofen, in the products called PM or night-time. The arithmetic is enough: a dose of the order of a gram taken in the form of tablets combined with 500 mg of paracetamol amounts to about 10 grams of paracetamol, that is, a potentially fatal dose by way of the liver, delayed, and with no symptom at all in the first 24 hours. In the United States in 2023, diphenhydramine is the first substance mentioned in overdose deaths classified as suicide, 14.7%, and paracetamol the fifth, 8.0%. The two often travel in the same tablet. Rule: read the list of active ingredients at every purchase, including for a product already used, and treat any ingestion of a PM product beyond the labelled dosage as a double poisoning.

Cardiotoxicity, for its part, rises when the rest plateaus. A prospective registry from more than 50 American hospitals, 6,473 patients, finds severe QTc prolongation in 13% of them, and classes diphenhydramine among the molecules at increased risk, alongside amitriptyline, doxepin, imipramine and nortriptyline. In overdose, it behaves like a tricyclic. Signals that count on the ECG: QRS above 100 ms, QRS above 120 ms, which is a strong predictor of poor outcome, convulsions included, QTc above 500 ms. First-line treatment is hypertonic sodium bicarbonate, and it can fail: a published case describes a QRS at 162 ms that goes to 174 ms after eleven boluses and 550 mEq of bicarbonate, then to 104 ms after 95 mg of lidocaine. Haemodialysis does not clear this molecule, too protein-bound and too widely distributed.

Rhabdomyolysis can occur with nothing else. One documented case with no preceding convulsion, coma or hypothermia, with a toxic concentration confirmed in the cerebrospinal fluid. Muscle pain or dark urine after an episode warrant a CPK assay, even if everything else went well.

Then come the other substances of this wiki. Diphenhydramine is one of the three sedative substances of the set, along with nitrous oxide and dextromethorphan, and it is by that route that it enters the other pages. Three of them name it, for two distinct reasons. DPT ranks it among the things that worsen the aspiration chain that killed its only published fatality. AMT and 2C-I cite it twice each: in their convulsive list, and for the false positive it produces on urine screening, which muddles their management as much as its own. None of these three links is a measured pharmacological interaction, and they must not be read as one: they are a mechanics of inhalation, a shared threshold and an analytical problem.

• Dextromethorphan. The worry is real and has never been measured. DXM is a CYP2D6 substrate, diphenhydramine is an inhibitor of it measured in humans, and this interaction runs in that direction only. The order of magnitude that this type of inhibition reaches is known from elsewhere: under bupropion, DXM exposure is increased 20 to 30 fold. And in a series of five deaths of adolescents who had ingested powdered DXM, three carried diphenhydramine. What is missing, and it must be said: neither an interaction trial nor a dedicated case report exists for this precise pair, and a query made expressly to find one turned up nothing. The pair is co-consumed and has never been studied as a pair. Let us add that diphenhydramine is a weak inhibitor in intrinsic potency, the weakest of the six classic antihistamines tested in vitro: what makes it important is not its strength, it is its availability.
• Syrian rue, with DXM. The only human case of a serious combination inside this set of fourteen contains diphenhydramine. A 46-year-old man, after three days of ayahuasca on a retreat, took a DXM cough syrup and diphenhydramine: severe serotonin syndrome, 38.9 °C, CK at 1,432 IU/L, cyproheptadine, midazolam, emergency intubation, discharge on the third day. Two caveats that are part of the case: three products are present there together and the urine toxicology was in addition positive for benzodiazepines, which makes four. Only the conference abstract was read, not the article.
• 5-MeO-DiPT. CYP2D6 is the only enzyme, among six tested, capable of its O-demethylation, with no back-up pathway. A CYP2D6 inhibitor therefore closes that door. But the direction of the clinical consequence is not established, because it is not known whether the metabolite is active, and it must not be asserted that closing this pathway makes things worse. Inference, not fact, and no human case exists.
• Nutmeg. Co-ingestion is documented: in an Illinois series, five of the fifteen intentional exposures were polyintoxications, all in 15 to 20 year olds, including one patient who required ventilation, and diphenhydramine figures among the co-ingestants recorded. A caveat inseparable from this point: the abstract of this series confirms only the five polyintoxications out of fifteen and the ventilated patient, while the list of co-ingestants and the duration of 24 hours come from a full text that could not be re-verified. This is not an anticholinergic interaction, and one must not build one out of it: no muscarinic binding data has ever been published for any nutmeg compound, and one of them inhibits acetylcholinesterase, which is the pharmacological opposite. The documented risk is that of polyintoxication in adolescents, not that of an addition of mechanisms.
• Vomiting under a sedative. It is the best established chain of the whole set, and diphenhydramine sits on the wrong side of it. Vomiting is the rule with nutmeg, morning glory, Hawaiian baby woodrose and Syrian rue, and it figures in the 2C-T-7 deaths. What establishes it is one precise case, and it should be given rather than asserted: the only published DPT death is a deep aspiration of gastric contents, in a 20-year-old man, in an order that is not the one expected, collapse, convulsive seizure, then vomiting, and so in a person already unconscious. Nobody has documented this chain with diphenhydramine, but it is diphenhydramine that supplies what was missing there, the blunted consciousness, and its own toxicity supplies the two other pieces, the convulsion and the coma, above the gram. The practical consequence is that supervision must not wait for the first vomit before it begins. And there is nothing to counter it with on the spot: the only antidote for this molecule is a hospital one, under ECG conditions, and what is left is someone sober to roll a person who will not ask for help onto their side.
• The other anticholinergics. The consensus applies the same thresholds of 7.5 mg/kg and of 300 mg to dimenhydrinate, which indicates that it treats them as equivalent. By mechanistic extension, combining diphenhydramine with scopolamine, with datura, with tricyclics, with antipsychotics carrying a high anticholinergic load or with bladder antispasmodics adds the effects together. No quantified interaction study exists for these combinations.
• The seizure threshold, which its own curve already lowers. The convulsion is, along with coma, what carries on rising in it when the delirium plateaus, and its signal is on the best supported side of the corpus, that of counted human cases. Ten of the fourteen substances of this wiki carry a convulsive signal, six resting on counted human cases, it among them, and four on animal material, isolated or unquantified; the remaining four carry none. Combining one of those ten with it therefore amounts to stacking two lowerings of the same threshold, on a molecule that already supplies the blunted consciousness that will prevent calling for help.
• Alcohol. In the Iowa driving simulator, overall performance was worst under diphenhydramine 50 mg, worse than at a blood alcohol level of 0.1%, on car following and lane keeping. A nuance that comes from the same trial: reaction time to a vehicle blocking the lane was slowest under alcohol. The result to hammer home is elsewhere: felt drowsiness does not predict the impairment, and drivers cannot rely on it to decide not to drive.

One can take it without knowing. Diphenhydramine has been identified among the substances of the powders called tusi or pink cocaine in a Miami-Dade forensic series. An American alert from 2023 on its presence as an adulterant in the street drug supply exists, but it has not been read, it exists only as a PDF, and its content cannot be reported here. A practical consequence all the same: an unexpected anticholinergic picture in a user of stimulants or opioids should bring it to mind.

In the emergency department, it muddles things in both directions. It figures among the medicines associated with false positives on urine screening, notably for phencyclidine. This is the mirror of the problem that all the rest of this wiki meets, where a negative screen rules nothing out: here, a positive screen may be false and must be confirmed by an analytical method. The problem does not stay within this page, because it circulates through the other pictures: tryptamines such as AMT and DPT figure in no routine assay, and 2C-I gives the direct proof of it, a patient who was convulsing, with 2C-I in serum and in urine, and an immunoassay urine screen that came back negative. A person who has taken one of these products with diphenhydramine therefore leaves with a screen that shows neither one nor the other, and sometimes a false positive in their place. In both cases, the same conclusion: saying what has been taken is the only way to make it known, and keeping the packet materially changes the management.

Certain actions make things worse, and the antidote the hospital has is not a simple one. Do not induce vomiting, and no activated charcoal during transport, because of the risk of loss of consciousness and of convulsions. Benzodiazepines for the agitation and the convulsions, but without dose escalation, which perpetuates the confusion and increases the risk of intubation. Avoid phenytoin, itself a sodium channel blocker. The cholinergic antidote is physostigmine, whose reputation for dangerousness, inherited from two cases of bradyasystolic arrest published in 1980 in patients poisoned with tricyclics, has been revised by modern series. That rehabilitation does not transpose here as it stands, however, because diphenhydramine is itself a sodium channel blocker: the contraindications listed are a QRS above 120 ms, a PR above 200 ms or a higher-degree block, a heart rate below 80 per minute. The published dosages do not agree from one source to another, and the discrepancy is shown rather than averaged. The product is moreover in short supply, the American manufacturer having ceased production, which has led to looking toward rivastigmine, with for the moment a single retrospective series of twelve patients with no control group.

Across populations, the most serious and the most avoidable risk is that of the infant: five fatal poisonings between 6 and 12 weeks, in one case because a father had given some to put the child to sleep, in another because a childminder had put some in a baby's bottle. Another series describes ten deaths of infants under twelve months in eight months with over-the-counter cold medicines, and concludes that carers wrongly believe these products to be harmless at that age. In the child, the paradoxical reaction, excitation, confusion, agitation, convulsions, is a documented and expected effect, not an individual oddity. In the elderly person, a lengthened half-life and a geriatric recommendation generally to avoid it, because of the risk of falls. During pregnancy, it crosses the placenta; during breastfeeding, it sedates the infant and can reduce lactation.

Two reputations stick to it, and they are to be taken separately. As a sleeping aid, the American Academy of Sleep Medicine explicitly recommends against using it in sleep-onset and sleep-maintenance insomnia in the adult, a recommendation of weak strength, and tolerance to the sedation is complete in three days. A 2025 expert consensus concludes on the contrary that 50 mg is effective on short-term insomnia in the young healthy adult; that work was funded by a manufacturer of the products concerned, and it acknowledges itself the tolerance by the third day, the next-day residual effects, the absence of any recommendation beyond three months and the absence of consensus in the elderly person. Even the most favourable source reduces the indication to a few nights. On dementia, the reassuring formula does not hold: the association between cumulative anticholinergic load and dementia is solid and dose-dependent, and the American cohort in which the first-generation antihistamines weigh the most finds it, relative risk 1.54, 95% CI 1.21 to 1.96. The largest European study does not find it for the antihistamines, but it rests on British general practice prescriptions, where an over-the-counter product is largely invisible. Neither of these studies establishes a causality, and both of them say so.

There remains the sober third party, and it is not optional here. No source explicitly deals with supervision, and this recommendation is a deduction, not a citation. But it follows from the best established clinical fact of this page: in a true delirium, the person is not aware that their perceptions are false. They can therefore neither self-assess, nor judge the seriousness of what is happening to them, nor ask for help in a reliable way. It is exactly the reverse of the situation where someone goes through a difficult experience knowing that they brought it on.

What nobody has measured

It should be said rather than filled in. Six measurements are missing, and they are all missing on the same side, that of the effect people come looking for: nobody has administered a deliriant dose in the laboratory, put numbers on the muscarinic affinities by subtype although that is the very mechanism by which this molecule is classed, measured whether tolerance also develops to the delirium, characterised the cognitive sequelae of repeated episodes although witnesses describe post-acute effects, verified that the dopaminergic effect in the rat transposes to the human, or tested the interaction with dextromethorphan. To this is added a hole for which we are responsible: the French data have not been sought, neither poison centres, nor ANSM, nor addictovigilance, whereas a 2015 Prescrire note on misuse, not read and with no abstract available, is indexed with the keyword France, the only sign that a French-language literature exists.

Sources

• M. Tashiro et al. (2008), occupancy of brain H1 receptors by antihistamines, 11C-doxepin PET, Br J Clin Pharmacol (56.4% at 30 mg).
• G. Tanda, T.A. Kopajtic, J.L. Katz (2008), cocaine-like neurochemical effects of antihistamines, J Neurochem (Ki at H1, DAT, NET, SERT on rat tissue; microdialysis; also the source of the mention of cheese and of T's and blues, cited second-hand).
• N. Kubo, O. Shirakawa, T. Kuno, C. Tanaka (1987), antimuscarinic effects of antihistamines, Jpn J Pharmacol 43(3):277-282. Abstract read, table not read (403 at the publisher): diphenhydramine appears in neither of the two groups cited in the abstract, and its value must not be invented.
• D. Radovanovic et al. (2000), dose-dependent toxicity of diphenhydramine overdose, Hum Exp Toxicol (0.3 g and 1.0 g; plateauing of the delirium beyond 1.5 g).
• B.E. Benson et al. (2010), dose-response of diphenhydramine and triage thresholds, Clin Toxicol (509 cases; 8.2 mg/kg; factor of 47, 95% CI 17 to 154).
• E.J. Scharman et al. (2006), consensus recommendation on out-of-hospital management, Clin Toxicol (7.5 mg/kg or 300 mg, whichever is the smaller; charcoal; in-hospital physostigmine).
• S.L. Campleman et al. (2020), risk of severe QT prolongation by molecule, ToxIC registry, Clin Toxicol.
• K. Makki et al. (2025), lidocaine in diphenhydramine sodium channel toxicity, Clin Pract Cases Emerg Med 9(2):223-227, PMC12097242 (failure of bicarbonate; estimated lethal dose 20 to 40 mg/kg; convulsions from 150 mg in the child, review estimates).
• B.A. Hamelin et al. (1998), inhibition of CYP2D6 by the classic antihistamines in vitro, Drug Metab Dispos (Ki of about 11 µM; diphenhydramine is the weakest inhibitor of the six).
• B.A. Hamelin et al. (2000), interaction with metoprolol, Clin Pharmacol Ther and E. Lessard et al. (2001), interaction with venlafaxine, J Clin Psychopharmacol. It is Lessard who measures that the disposition of diphenhydramine is only slightly affected by an abolished CYP2D6 activity: the interaction runs in one direction only.
• H.M. Dourron et al. (2024), qualitative analysis of 32 misuse accounts published on YouTube, J Psychoactive Drugs.
• W. Schreiber, A.M. Pauls, J.C. Krieg (1988), toxic psychosis from diphenhydramine, Dtsch Med Wochenschr. English abstract read, German text not read.
• A. Thomas, D.G. Nallur, N. Jones, P.N. Deslandes (2009), misuse and withdrawal of diphenhydramine, J Psychopharmacol (five reports, six patients, link with antipsychotics).
• G.S. Richardson et al. (2002), tolerance to the daytime sedative effects of the H1 antihistamines, J Clin Psychopharmacol (complete tolerance in three days, on sedation alone).
• A. Nemanich, E. Liebelt, A.K. Sabbatini (2021), rise in overdoses and misuse in the United States 2005-2016, Clin Toxicol and M.A. Darracq, S.L. Thornton (2022), ingestions in adolescents 2007-2020, Clin Toxicol.
• M.F. Garnett, J.A. Cisewski, F.B. Ahmad (2026), substances most involved in overdose deaths, United States 2017-2023, National Vital Statistics Reports (14.7%; paracetamol 8.0%).
• A.M. Baker et al. (2003), fatal poisonings in infants, J Forensic Sci and L. Marinetti et al. (2005), over-the-counter cold medicines and infant deaths, J Anal Toxicol.
• S.M. Emadian, E.M. Caravati, R.D. Herr (1996), rhabdomyolysis, Am J Emerg Med (without convulsion, coma or hypothermia); U. Khosla, K.S. Ruel, D.P. Hunt (2003), antihistamine rhabdomyolysis, South Med J.
• J.M. Weiler et al. (2000), fexofenadine, diphenhydramine and alcohol in the Iowa driving simulator, Ann Intern Med and J.G. Ramaekers, J.F. O'Hanlon (1994), Eur J Clin Pharmacol.
• M.J. Sateia et al. (2017), recommendation of the American Academy of Sleep Medicine on chronic insomnia, J Clin Sleep Med against D.F. Ariza-Salamanca et al. (2025), expert consensus, J Clin Med, PMC12112657, funded by Procter and Gamble.
• S.L. Gray et al. (2015), cumulative anticholinergic use and incident dementia, JAMA Intern Med and C.A.C. Coupland et al. (2019), nested case-control study, JAMA Intern Med. The second rests on British prescriptions and can barely see an over-the-counter product.
• N.C. Brahm et al. (2010), false positives of urine drug screens, Am J Health Syst Pharm (including phencyclidine).
• D.M. Moore et al. (2025), Miami-Dade forensic series on the coloured powders called tusi, J Forensic Sci.
• N.G. Blackstone, A. Olson, B. Ainapurapu (2020), physostigmine in anticholinergic poisoning, Cureus and M. Berg et al. (2025), rivastigmine as an alternative, series of 12 cases, Am J Emerg Med. The work of Burns 2000, Pentel and Peterson 1980, Arens 2017 and Arens and Kearney 2019 is cited second-hand, not read.
• StatPearls, chapters Diphenhydramine (V. Sicari, P. Patel, C.P. Zabbo) and Diphenhydramine Toxicity (D.A. Huynh, M. Abbas, A. Dabaja); emDocs, ToxCard Diphenhydramine Toxicity (R. Hamm, K. Kopec); EMCrit IBCC, Anticholinergic intoxication chapter. This last page carries a published unit error on the lethal threshold, a factor of ten.
• S. Pack, T. Ellett, J. Pham, Y. Ahmad (2026), Am J Respir Crit Care Med 212(Suppl 1), aamag162.4811. Conference abstract read, full article not read. This is the Syrian rue plus DXM plus diphenhydramine case.
• B.K. Logan et al. (2009), series of five deaths from powdered dextromethorphan, three of them with diphenhydramine, known here through the DXM file of this wiki.
• M.A. Neukamm et al. (2024), fatal aspiration case after consumption of DPT, J Pharm Biomed Anal, known here through the DPT file of this wiki. It is the mechanical anchor of the chain of vomiting under a sedative, and the only published death of the set that documents it.
• T. Tanaka et al. (2011), diphenhydramine overdose mimicking a serotonin syndrome, Psychiatry Clin Neurosci 65(5):534. Full text not read, known through two concordant extractions: contrary to what its title suggests, this letter defends a serotonergic hypothesis. Its published reply, H. Sanaei-Zadeh (2012), Psychiatry Clin Neurosci 66(3):244-245, PMID 22443251, has not been read and has no abstract available. The debate is not settled.
• A.E. Rosenberg, R. Wang, M.H. Rosenberg (2025), Plast Reconstr Surg Glob Open 13(3):e6634: the only case where diphenhydramine is presented as precipitating a serotonin syndrome simultaneously involves duloxetine, asenapine, trazodone and fentanyl, which makes the attribution indeterminable.
• R.N. Serio (2004), acute delirium on diphenhydramine and linezolid, Ann Pharmacother 38(1):62-65, PMID 14742796. Naranjo causality possible, and the mechanism proposed by the author is dopaminergic, by MAO blockade, not anticholinergic.
• Center for Forensic Science Research and Education, alert of 28 April 2023 on diphenhydramine as an adulterant. PDF not opened, content not verified.
• FDA, safety communication of 24 September 2020 on high doses of diphenhydramine. Primary page gone, HTTP 404 at the two addresses tested: its existence is attested by the peer-reviewed literature, its content is not.
• Erowid, Diphenhydramine Vault and its experience report index (a corpus of users, to be read on site; its licence forbids reproduction and ingestion by an automated system).
• S. Fourniquet et al. (2023), Seeing Spiders, PsyArXiv. Preprint not read, not peer-reviewed: only the title is known, and it grounds nothing here.