Why this poison is interesting

Ethanol is the poison every emergency clinician sees most often and thinks about least. It is worth a mechanism page for three reasons, none of which is the intoxication itself.

It is the reference case for saturation. Alcohol dehydrogenase has a low Michaelis constant — around 1 mM — so it is saturated at low blood alcohol concentrations and works at maximal velocity thereafter.¹ Every other saturation poison on this site is a variation on this.
Its harm comes from a consumed cofactor, not from a metabolite. Oxidising ethanol reduces NAD⁺ to NADH twice over, and the resulting fall in the NAD⁺/NADH ratio shuts down gluconeogenesis, the Krebs cycle, pyruvate dehydrogenase and fatty acid oxidation.¹ The hypoglycaemia, the ketoacidosis and the raised lactate are all the same lesion seen from different angles.
Its most-quoted number is wrong. The figure everybody carries for how many milligrams per decilitre fall per hour cannot be a constant, and the biochemistry says why. See the appraisal.

There is a fourth reason, which is clinical rather than mechanistic. Ethanol is the commonest thing to be co-ingested with everything else on this site, and it changes those poisonings — protectively in the toxic alcohols, dangerously almost everywhere else. A page about ethanol is partly a page about what ethanol is concealing.

A poison is a drug whose kinetics have escaped its pharmacology.

The toxic principle

The poison is ethanol. That is unusual on this site and worth stating explicitly: there is no NAPQI, no glycolate, no formate. The parent molecule is the CNS depressant, and it is cleared to two successively less toxic products.

The one situation in which a metabolite matters is when the second step is blocked. Acetaldehyde accumulates if aldehyde dehydrogenase is inhibited pharmacologically (disulfiram, and incidentally metronidazole and some cephalosporins) or genetically. The ALDH2 polymorphism is instructive here: a large meta-analysis found that carriers of the less active ADH alleles and the highly active ALDH2*1 allele had an increased risk of alcoholism, which the review attributes to low accumulation of acetaldehyde in those individuals.¹ The flush reaction is the same mechanism running the other way. Established

Toxicokinetics

Ethanol — why there is no single elimination rate
Parameter	Therapeutic	In overdose	Why it changes
Absorption	Rapid from the duodenum. Some ingested alcohol is oxidised in the stomach by gastric ADH isoforms before reaching the systemic circulation — first-pass metabolism¹	The same, but the modifiers dominate: ethanol passes rapidly into the duodenum in the fasted state, minimising first-pass metabolism and producing higher blood concentrations than the same dose taken with food¹	The stomach is a metabolic organ here, not just a reservoir. First-pass metabolism is reported to be low in people with alcohol dependence, and especially low in dependent women, which the review links to the higher blood concentrations seen in women after an equivalent oral dose.¹ H₂-receptor blockers such as cimetidine reduce it further.
Protein binding	None of consequence	Unchanged	There is no binding to saturate, so unlike salicylate the free fraction cannot rise. Every dose–response oddity in ethanol comes from metabolism, not distribution.
Volume of distribution	Total body water — tissue concentration tracks tissue water content¹	Unchanged	This is why body size and composition, rather than weight alone, determine the concentration produced by a given drink; and why a small, dehydrated or elderly patient reaches a higher concentration from the same amount.
Half-life	The concept does not apply cleanly. A half-life presumes first-order elimination; ethanol's is saturated over most of the clinically relevant range¹	Elimination proceeds at close to maximal velocity, so the concentration falls roughly linearly rather than exponentially — but the slope is not a fixed constant¹	Two separate departures from linearity, in opposite directions. At low concentrations ADH is no longer saturated and Michaelis–Menten kinetics reappear.¹ At high concentrations, CYP2E1 and the high-Km ADH4 isoform contribute increasingly, so higher rates of elimination are observed at higher blood alcohol concentrations.¹ The review's own conclusion: "it is not possible to estimate one single rate of alcohol metabolism."
Elimination route	Hepatic oxidation. ADH in the cytosol is the main route; CYP2E1 accounts for about 10% of the liver's alcohol-oxidising capacity at low concentrations; catalase is capped by hydrogen-peroxide availability — peroxide is generated at only about 2% of the rate of alcohol oxidation — and it is insignificant in the liver¹	The CYP2E1 share rises with concentration because of its higher Km, and rises again with chronic use because CYP2E1 is induced¹	Induction is the mechanistically important part, and it is not classical enzyme induction: CYP2E1 levels are increased by chronic ethanol largely by protecting the enzyme from proteasomal degradation, and CYP2E1 is also induced in diabetes and in the fasted state.¹ That same induced enzyme is the one that makes NAPQI from paracetamol — which is the honest mechanistic basis for the interaction everyone half-remembers.
Order of kinetics	First order only at low concentrations¹	Effectively zero order across most of the clinical range, with a concentration-dependent rate at the top end¹	The archetype for this whole family. A saturated pathway means a linear fall in concentration and a disproportionate rise in concentration for a small rise in dose — the same shape as salicylate, phenytoin and theophylline.
Dialysability	—	Physicochemically highly dialysable — small, water-soluble, unbound, distributed in total body water¹. EXTRIP has published no recommendation for ethanol²	The absence is itself informative. Endogenous clearance is already fast, the poisoning is self-limiting with airway support, and there is no metabolite to chase. Dialysis would work and is essentially never worth doing — the opposite situation to methanol, a molecule one carbon smaller.

Metabolism and the metabolites

Two oxidations, each consuming a molecule of NAD⁺. That arithmetic is the entire metabolic story of alcohol poisoning.

Ethanol — two steps, and the cofactor they both spend

EthanolThe CNS depressant. Distributes into total body water¹
Alcohol dehydrogenase (cytosol)AcetaldehydeThe main route. NAD⁺ → NADH. Km ≈ 1 mM, so saturated at low concentrations¹
CYP2E1 (microsomal, MEOS)Acetaldehyde~10% at low concentrations; more at higher concentrations and after chronic use, when the enzyme is induced¹
Catalase (peroxisomal)AcetaldehydeCapped by H₂O₂ supply, generated at only ~2% of the rate of alcohol oxidation — insignificant in liver, but present in brain peroxisomes¹
AcetaldehydeReactive; normally cleared so fast that circulating concentrations are very low¹
Aldehyde dehydrogenase, ALDH2 (mitochondria) — the step disulfiram blocksNAD⁺ → NADH again¹
AcetateNon-toxic; leaves the liver and is oxidised to CO₂ and water in peripheral tissue

The consequence: a shifted redox ratio

Both oxidations reduce NAD⁺ to NADH, so the cellular NAD⁺/NADH ratio falls — in the cytosol because ADH works there, and in the mitochondrion because ALDH2 works there.¹ Those two compartments announce themselves through two different laboratory couples:

Cytosolic redox: Reported by the pyruvate/lactate ratio, through lactate dehydrogenase: pyruvate + NADH ⇌ lactate + NAD⁺.¹ Push NADH up and pyruvate is dragged to lactate. Established
Mitochondrial redox: Reported by the β-hydroxybutyrate/acetoacetate ratio: acetoacetate + NADH ⇌ β-hydroxybutyrate + NAD⁺.¹ Push NADH up and ketone bodies shift towards β-hydroxybutyrate. Established

The review lists the reactions inhibited by that fallen ratio, and the list is the clinical picture written as biochemistry: glycolysis, the citric acid cycle (with ketogenesis favoured), pyruvate dehydrogenase, fatty acid oxidation, and gluconeogenesis.¹

Three bedside findings, one equation

Hypoglycaemia — gluconeogenesis is inhibited, so once glycogen is gone there is no second line. This is why it is the malnourished, the fasted, the chronically dependent and the small child who become hypoglycaemic, and not the well-fed adult. Established
Alcoholic ketoacidosis — the citric acid cycle is inhibited and ketogenesis is favoured.¹ Established
A raised lactate that is not a perfusion lactate — pyruvate is pulled to lactate by mass action on a reduced cytosolic couple.¹ Treating it as evidence of shock leads to fluid and antibiotics rather than to glucose and thiamine. Inferred

And a fourth, which is a laboratory artefact of the same shift: because the mitochondrial couple favours β-hydroxybutyrate, and the classical urine nitroprusside reaction detects acetoacetate rather than β-hydroxybutyrate, urine ketone testing can substantially understate ketosis in alcoholic ketoacidosis. The redox direction is established;¹ the diagnostic consequence follows from the chemistry of the assay rather than from a clinical trial. Inferred

What changes the answer

Food. Alcohol metabolism is higher in the fed state — ADH levels are higher, substrate shuttles move reducing equivalents into mitochondria more effectively, and liver blood flow may increase. The effect was similar for carbohydrate, fat and protein meals.¹ Established
Chronic use, through CYP2E1 induction — which raises both alcohol clearance and, separately, susceptibility to anything else that CYP2E1 activates.¹
Fasting and diabetes also induce CYP2E1.¹ The same three conditions that most predispose to alcohol-related hypoglycaemia also upregulate the enzyme that makes reactive metabolites.

Elimination and accumulation

Ethanol is the anti-accumulation poison. There is no ion trapping (it is not ionisable), no protein binding to come undone, no enterohepatic recirculation, no active metabolite to build up behind a bottleneck, and no compartment it hides in — it goes where the water is.¹ A small amount is excreted unchanged in breath, urine and sweat, which is the basis of breath testing.

What that leaves is a poison whose entire pharmacokinetic interest lies in the rate of a single saturated step, and whose entire clinical management is therefore time and supportive care. That is worth stating positively rather than as an absence: there is nothing to do to ethanol, only things to do for the patient.

Target organs — and why those

Brain — acutely

TargetLigand-gated ion channels, principally GABA-A potentiation and NMDA-receptor inhibition

Why hereEthanol is not a channel blocker or a receptor agonist in the clean sense; it is a low-potency modulator that reaches the millimolar concentrations needed because people drink grams of it. The clinical progression — disinhibition, ataxia, nystagmus, dysarthria, then obtundation and loss of airway reflexes — tracks increasing depression of the same systems. Established

At the bedsideThe ceiling that benzodiazepines have, ethanol does not: respiratory depression and airway loss are dose-related and real. The commonest mechanism of death is a protected airway that was not.

Liver — the redox economy

TargetThe cytosolic and mitochondrial NAD⁺/NADH couples

Why hereBecause both oxidising steps happen here and both spend NAD⁺.¹ The liver is not injured acutely by a single episode; it is functionally reprogrammed for a few hours — gluconeogenesis off, ketogenesis on, lactate up. The organ selectivity is simply where the enzymes are. Established

At the bedsideHypoglycaemia, ketoacidosis and lactataemia in the same patient are one lesion, not three. Glucose and thiamine address it; fluid alone does not.

Brain — the thiamine problem

TargetThiamine-dependent enzymes, principally pyruvate dehydrogenase and transketolase

Why hereThis is chronic rather than acute, and it compounds with the redox shift: pyruvate dehydrogenase is already inhibited by the fallen NAD⁺/NADH ratio,¹ and thiamine deficiency removes its cofactor as well. Two independent hits on the same enzyme in the same patient. Inferred

At the bedsideThe practical rule that thiamine is given before or with glucose in a malnourished drinker follows from the same enzymology — a glucose load increases demand on a thiamine-dependent step that is already failing.

Timeline of effects

Ethanol is the Band A poison with no latent phase, no antidote and no toxic metabolite — benzodiazepines and tricyclics also lack a latent phase, but only ethanol lacks all three — and including it here is deliberate: the contrast is what makes the other timelines legible. There is no gap between the two tracks — and that turns out to be its own kind of danger, because a patient who looks their worst on arrival invites the assumption that they can only improve.

Ethanol — the two tracks that do not come apart

Time

What you seeWhat is happening

0–1 hAbsorption
What you seeRising intoxication: disinhibition, then ataxia and dysarthria. On an empty stomach this is faster and reaches further.
What is happeningRapid duodenal absorption, with first-pass gastric metabolism largely bypassed in the fasted state.¹ ADH is already saturated. NAD⁺ is already being spent.
1–4 hPeak
What you seeMaximal depression: nystagmus, vomiting, obtundation, loss of airway reflexes at the top end. This is usually the presentation.
What is happeningConcentration is at its highest and the redox shift is fully established: gluconeogenesis inhibited, ketogenesis favoured, lactate raised by mass action.¹ In a patient with no glycogen reserve, this is when the glucose runs out.
4 h onwardsElimination
What you seeSteady, roughly linear improvement. Failure to improve on schedule is the finding that matters — it means something else is present.
What is happeningSaturated ADH clears alcohol at close to maximal velocity, so concentration falls almost linearly — faster at the top of the range where CYP2E1 contributes, and dropping off exponentially at the bottom as ADH desaturates.¹ The slope is not a constant and should not be used as one.
6–72 hWithdrawal, in the dependent
What you seeTremor, sweating, tachycardia and anxiety from around 6–12 h; seizures; then delirium tremens at 48–72 h in a minority. This can begin while the patient is still measurably intoxicated.
What is happeningChronic exposure produces compensatory downregulation of GABA-A responsiveness and upregulation of NMDA signalling. Removing the alcohol unmasks that compensation, and it is unmasked at whatever concentration the individual is adapted to — not at zero. Inferred

What the mechanism predicts at the bedside

Why a blood alcohol concentration answers almost nothing

Three mechanistic reasons. Tolerance means the same concentration produces very different conscious levels in different people, so the number does not grade the patient. The elimination rate is not a constant, so a concentration cannot be used to predict when they will be sober.¹ And a concentration that is high enough to explain the presentation does not exclude anything else — it merely provides a reason to stop looking. The number's one genuine use is the opposite of the usual one: a concentration too low to explain the conscious level is a positive finding.

Why not to expect a fixed number of hours

Because at high concentrations CYP2E1 and ADH4 add capacity, so the fall is faster than the familiar figure; at low concentrations ADH desaturates and the fall becomes exponential, so the last part takes longer than a straight line predicts; and induction, nutritional state, sex and dependence all move the slope.¹ The direction of every one of these is known and none of them is measurable at the bedside.

Why glucose and thiamine are given on the picture, not the number

Because the mechanism predicts a normal glucose right up until it collapses. Glycogenolysis is unaffected by the redox shift, so blood sugar is maintained normally while stores last; it is gluconeogenesis — the backup — that is inhibited.¹ So there is no gradual downward drift to catch. In a fasted, dependent or small patient, the reading is normal and then it is not. Inferred

Why ethanol is protective in toxic-alcohol poisoning

Because it competes for the same enzyme. Ethanol is alcohol dehydrogenase's preferred substrate, so an ethanol-containing co-ingestion slows the conversion of methanol to formate and ethylene glycol to glycolate — the mechanism the EXTRIP ethylene glycol review lists explicitly as a modulator of toxicity, and the same mechanism that makes therapeutic ethanol an antidote where fomepizole is unavailable.³ A patient who drank vodka with the antifreeze has accidentally started their own treatment. See ethylene glycol and methanol. Established

Why the osmolar gap belongs to ethanol first

Ethanol is small and present in molar quantities, so it contributes fully to measured osmolality. In practice the commonest cause of a raised osmolar gap is ethanol, which means the gap must be corrected for the measured ethanol before it says anything about methanol or ethylene glycol. An uncorrected gap in a drinker is not evidence of a toxic alcohol. Established

The antidote, from the poison's side

What exists instead is treatment aimed at the consequences of the redox shift, and it is worth reading in that light rather than as a generic bundle:

Glucose replaces what inhibited gluconeogenesis cannot make.¹ It is a substrate, not an antagonist.
Thiamine restores the cofactor for enzymes that the redox shift has already inhibited — the two insults converge on pyruvate dehydrogenase.¹ Given before or with a glucose load in anyone malnourished. Inferred
Airway protection is the definitive treatment for the acute poisoning, because loss of airway reflexes is what kills people at the top of the dose range.
Benzodiazepines, in withdrawal rather than in intoxication, substitute at the GABA-A site the brain has adapted to.

Critical appraisal

The fixed hourly elimination rate is traditional teaching, and its own biochemistry refutes it. Traditional teaching The claim that alcohol falls by a set number of milligrams per decilitre per hour is taught universally and used to predict sobriety. Cederbaum's review sets out why it cannot hold: elimination was "originally believed to be a zero-order process", but linearity fails at low concentrations as ADH desaturates, and at high concentrations the high-Km systems (CYP2E1, ADH4) add capacity so that higher rates are seen at higher concentrations (observed, the review notes, in some but not all studies) — concluding directly that "it is not possible to estimate one single rate of alcohol metabolism."¹ A single figure is a useful order-of-magnitude prior and a poor basis for a discharge time.
"Alcohol is a zero-order drug" is a simplification of the same kind, and a defensible one. Over the concentration range that brings people to hospital, the approximation is good. The failure is not the model; it is treating the model's slope as a personal constant.
The redox mechanism is exceptionally well established; the specific clinical attributions are inferred to varying degrees. Inferred That ADH and ALDH2 lower the NAD⁺/NADH ratio and that this inhibits gluconeogenesis, the citric acid cycle, pyruvate dehydrogenase and fatty acid oxidation is stated flatly in the review.¹ That a given patient's lactate is therefore redox rather than perfusion in origin is a clinical inference about that patient, and treating it as certain has its own failure mode — alcohol and sepsis coexist routinely.
The urine-ketone limitation is chemistry, not evidence. Inferred The redox shift towards β-hydroxybutyrate is documented;¹ the inference that a nitroprusside-based urine ketone test therefore understates ketosis follows from what that reaction detects, not from a study of alcoholic ketoacidosis patients. The practical consequence — measure β-hydroxybutyrate if the question matters — is sound either way.
Ethanol's exclusion from the EXTRIP programme is worth noticing rather than assuming. Established The workgroup has published on 22 poisons and not this one.² The reasoning is not stated by them, but the physicochemistry says ethanol is eminently dialysable, so the absence must reflect a judgement about benefit, not about removability.
The most consequential error in alcohol presentations is not pharmacological. Attributing every finding to intoxication is the standard failure, and the mechanism supports the corrective: ethanol has no latent phase and a predictable, roughly linear recovery, so a patient who is not improving as expected has a second diagnosis until proven otherwise.

References

1
Cederbaum AI. Alcohol metabolism. Clin Liver Dis 2012;16(4):667–85. PMC3484320 Open access. Source of the ADH Km of about 1 mM, the statement that no single rate of alcohol metabolism can be estimated, the CYP2E1 (~10%) and catalase (~2%) contributions and the mechanism of CYP2E1 induction, the first-pass gastric metabolism findings including those in women and in dependence, the tissue-water distribution statement, the food effect, and the full account of the NAD⁺/NADH redox shift with the pyruvate/lactate and acetoacetate/β-hydroxybutyrate couples and the list of inhibited reactions. Verified 31 Aug 2026 from the full text.
2
EXTRIP workgroup — published recommendations index. extrip-workgroup.org/recommendations Cited for an absence: the workgroup's 22 published poison recommendations do not include ethanol. Verified 31 Aug 2026.
3
Ghannoum M, Gosselin S, Hoffman RS, et al.; EXTRIP workgroup. Extracorporeal treatment for ethylene glycol poisoning: systematic review and recommendations from the EXTRIP workgroup. Crit Care 2023;27:56. PMC9921105 Cited here only for the statement that ethylene glycol toxicity is modulated by co-ingested ethanol because this decreases its metabolism. Verified 31 Aug 2026.
4
Kraut JA, Mullins ME. Toxic alcohols. N Engl J Med 2018;378(3):270–80. PubMed 29342392 Cited as the standard review of therapeutic ethanol as an alcohol dehydrogenase blocker. Paywalled; not used as the source of any figure on this page. Citation verified 31 Aug 2026.
5
TOXBASE — ethanol. National Poisons Information Service. toxbase.org (NHS login required. NPIS: 0344 892 0111. The authoritative UK source for assessment and management of the intoxicated patient. Login-gated, so not quoted here.)

Ethanol

At a glance