Why this poison is interesting

Phenytoin earns its place for two reasons, and they point in opposite directions. The first is that it is the library's clearest antiepileptic example of saturation kinetics — the same phenomenon that makes salicylate frightening, arrived at by a different enzyme. The second is more surprising: for such a notorious drug, the swallowed overdose is remarkably benign. It produces a florid but self-limiting cerebellar and vestibular syndrome, and death from oral phenytoin alone is rare.¹² The cardiotoxicity everyone associates with phenytoin is real — but it is a property of the intravenous route, not of the poison in the stomach.³

Clearance saturates at therapeutic concentrations. Phenytoin is hydroxylated by a capacity-limited hepatic enzyme, so once the enzyme is near-saturated small incremental doses produce very substantial increases in serum levels.² Dose and concentration stop being proportional in the therapeutic range, let alone in overdose. Established
The concentration itself is only half the story, because phenytoin is ~90% protein-bound and it is the free fraction that acts.² Hypoalbuminaemia raises the free fraction at any total concentration, so the number on the report can badly understate the effect.
The toxic syndrome is orderly and mostly reversible. Signs appear in a concentration-ranked sequence and resolve as the concentration falls; the rare permanent injury is cerebellar atrophy after prolonged severe exposure.²

The same molecule can be near-harmless swallowed and lethal infused — the poison is the route as much as the drug.

The toxic principle

The poison is phenytoin, unchanged — a hydantoin anticonvulsant that stabilises neuronal membranes by blocking voltage-gated sodium channels in their inactivated state. That mechanism is what makes it an antiepileptic; it is also, at high concentration and given fast intravenously, what makes it cardiotoxic, because the same use-dependent sodium-channel block acts on the myocardium. Established

There is no toxic metabolite. Phenytoin is hydroxylated in the liver to inactive products;² the parent drug is responsible for everything, and its clearance is the whole story.

Toxicokinetics

Phenytoin — a saturable enzyme and a protein-binding caveat
Parameter	Therapeutic	In overdose	Why it changes
Absorption	Absorbed from the small intestine; non-linear techniques estimate absorption as essentially complete, though formulation affects bioavailability²	Slow and sometimes prolonged after a large ingestion; peak may be delayed	Absorption is not the interesting variable here — unlike theophylline or salicylate, there is no modified-release trap that dominates the timeline. The variability that matters is in clearance, not uptake.
Protein binding	High — usually ~90% bound to albumin in adults²	Free fraction rises in hypoalbuminaemia, renal or hepatic disease, or hyperbilirubinaemia²	This is the row that makes the total concentration treacherous. Two patients with the same total phenytoin are not equally toxic if one is hypoalbuminaemic — the free, active fraction is higher in them. It is also the reason phenytoin is only moderately dialysable: most of it is bound and unavailable to the machine.
Volume of distribution	Moderate — estimated 0.52–1.19 L/kg; distributes into CSF²	Unchanged	A moderate volume of distribution, combined with high protein binding, is why extracorporeal removal is limited: there is a large bound and tissue reservoir the dialyser cannot reach in a single pass.
Metabolism	Hepatic hydroxylation by a saturable enzyme (principally CYP2C9 and CYP2C19)²	Saturation is the defining feature — clearance becomes concentration-dependent and the half-life lengthens as the level rises	This is the archetype phenytoin shares with salicylate, phenobarbital, ethanol and theophylline. A half-life that depends on the concentration means you cannot extrapolate a fall from two points, and it is why phenytoin toxicity can take days to resolve after a large ingestion.
Half-life	Averages ~22 h, range 7–42 h, at therapeutic concentrations²	Prolonged, and lengthens further with the dose	The wide therapeutic range is itself a clue to the saturable clearance and the inter-patient variability in the enzyme. In overdose the effective half-life can be much longer, which is why the cerebellar signs persist for days rather than hours.
Dialysability	—	*Moderately dialysable, EXTRIP level of evidence C, despite* high protein binding**¹; recommended only in very select severe cases	The high protein binding would suggest phenytoin is a poor dialysis candidate, and largely it is — but the workgroup graded it moderately dialysable because binding is not complete and saturates somewhat at high concentrations. The more important point is that dialysis is rarely needed, because the oral poisoning is seldom life-threatening.¹

Metabolism and the metabolites

Once again — as with salicylate and theophylline — this is the section where nothing toxic happens, and that absence shapes the treatment.

Phenytoin — hydroxylation to inactive products, and a saturable step

Phenytoin (ingested)This is the poison. A sodium-channel-stabilising hydantoin; no activation required
Hepatic hydroxylation by a saturable enzyme (CYP2C9, CYP2C19)The capacity-limited step — this is where dose and level decouple²
Para-hydroxylation and glucuronidation5-(p-hydroxyphenyl)-5-phenylhydantoin glucuronideThe major inactive metabolite; renally excreted²
MinimalPhenytoin excreted unchangedA small fraction; renal manipulation offers nothing useful²

What changes the answer

CYP2C9 and CYP2C19 genotype and interacting drugs shift the clearance, which is part of why the therapeutic level is so hard to hit and why some patients accumulate toward toxicity on an unchanged dose. Inferred
Hypoalbuminaemia, renal and hepatic disease all raise the free fraction, so the same total level means more in the sick, the elderly and the malnourished.² Established
Chronic toxicity is the under-recognised presentation — an ataxic, dysarthric patient on long-term phenytoin whose dose or interacting medication has nudged the saturable enzyme over its ceiling, presenting not with an overdose but with a gait disorder.

Elimination and accumulation

Phenytoin is eliminated almost entirely by hepatic metabolism, and that metabolism is saturable — so unlike salicylate there is no renal exit to enlarge, and unlike theophylline there is no strong enterocapillary recirculation to interrupt. Elimination is mostly a matter of waiting for a slow, concentration-dependent enzyme to work through the load.

The EXTRIP position — and why it is nearly a refusal

51 articles met the inclusion criteria — only case reports, case series and pharmacokinetic studies, yielding a very low quality of evidence. Clinical data came from 31 patients.¹ The workgroup concluded that phenytoin is moderately dialysable (level of evidence C) despite its high protein binding — and then, unusually, recommended against using that fact in most cases.

ECTR would be reasonable only in select cases of severe phenytoin poisoning (neutral, 3D).¹
Suggested if prolonged coma is present or expected (2D); reasonable if prolonged incapacitating ataxia is present or expected (3D).¹
Recommended against performing ECTR on the basis of the suspected dose ingested (1D), and against it on the basis of the serum concentration alone (1D) — two of the workgroup's stronger statements, and both are prohibitions.¹
If used, stop when clinical improvement is apparent (1D); intermittent haemodialysis preferred (1D), haemoperfusion an alternative (1D).¹

Target organs — and why those

Cerebellum and vestibular system

TargetPurkinje cells and central vestibular pathways, where phenytoin's sodium-channel action is most conspicuous in excess

Why hereThe cerebellar and vestibular systems are exquisitely sensitive to phenytoin, which is why the earliest and most reliable toxic signs are nystagmus and ataxia — and why they appear in a concentration-ordered sequence.² Prolonged severe exposure can cause irreversible cerebellar atrophy, the one lasting injury oral phenytoin reliably produces.² Established

At the bedsideNystagmus on lateral gaze at around 20 mg/L, ataxia around 30, dysarthria and lethargy above 40² — a bedside titration curve. The reassuring corollary is that a florid cerebellar syndrome, in isolation, is a benign and reversible finding.

Brainstem and consciousness

TargetReticular and brainstem centres at high concentration

Why hereAt high levels the sodium-channel stabilisation extends to depress consciousness and, eventually, respiration — the mechanism of the rare oral death, which the SmPC attributes to respiratory and circulatory depression and which usually involves co-ingestants.² Inferred

At the bedsideComa and respiratory depression after oral phenytoin should prompt a search for a co-ingested sedative; phenytoin alone rarely takes a patient that far, and the possibility of other CNS depressants, including alcohol, should be borne in mind.²

Heart — but chiefly by the intravenous route

TargetMyocardial sodium channels and the conducting system, when the drug arrives fast and at high concentration

Why hereThe same use-dependent sodium-channel block that stabilises neurons destabilises the myocardium when phenytoin is delivered rapidly intravenously, producing bradycardia, ventricular depression, VF and asystole.³ The propylene-glycol vehicle of the injection contributes to hypotension. Crucially, this is a rate-and-route phenomenon, largely absent after ingestion. Established

At the bedsideCardiac monitoring is mandatory for intravenous loading, the rate is capped at 25 mg/min and lower in the frail,³ and it links phenytoin conceptually to sodium-channel blockade — but a swallowed phenytoin overdose does not usually need a cardiac monitor for the phenytoin itself.

Timeline of effects

Phenytoin has no latent phase and no metabolic surprise. Its timeline is simply the concentration curve read as a sequence of neurological signs — appearing as the level climbs and receding, slowly, as the saturable enzyme grinds it down.

Phenytoin — a concentration-ordered march, then a slow recovery

Time

What you seeWhat is happening

0–a few hoursOnset
What you seeNystagmus first, then a broad-based ataxic gait, then slurred speech and drowsiness. Nausea and vomiting, tremor and hyperreflexia. The patient is unsteady and dysarthric but usually rousable and protecting their airway.
What is happeningThe concentration is rising through the cerebellar-sensitivity range: nystagmus at ~20 mg/L, ataxia at ~30, dysarthria and lethargy above 40.² Because clearance is saturable, the level may keep climbing after a large ingestion for longer than first-order intuition suggests.
PeakPlateau
What you seeMaximal cerebellar and vestibular signs, sometimes with agitation or a confusional state. In pure oral overdose this is usually as bad as it gets. Coma and respiratory depression, if present, should raise the question of a co-ingestant.
What is happeningThe free fraction — not the total — determines how deep the syndrome goes, so a hypoalbuminaemic patient reaches a given severity at a lower total number.² There is no acid–base disturbance and no toxic metabolite adding a second illness underneath.
DaysSlow resolution
What you seeThe signs recede in reverse order over days rather than hours. Prolonged severe exposure carries a small risk of permanent cerebellar dysfunction and atrophy.
What is happeningRecovery is governed by the saturable enzyme working back down the concentration curve — first-order once the level falls below the enzyme's ceiling, so the last part of the decline accelerates. The length of the tail is why phenytoin toxicity is measured in days. Established

What the mechanism predicts at the bedside

Why the total concentration must be read against the albumin

Because phenytoin is ~90% protein-bound and only the free fraction acts, a total level in a hypoalbuminaemic, uraemic or critically ill patient understates the effective exposure.² A patient who looks toxic with a total level in range is not a diagnostic puzzle — their free fraction is high. Correcting for albumin, or measuring free phenytoin, aligns the number with the patient. Established

Why the level may keep climbing after a large ingestion

Saturable clearance means elimination is near-maximal and cannot speed up to match a large load, so the concentration can rise for longer and fall more slowly than first-order reasoning predicts.² A single early level is a floor; serial measurement until it is clearly declining is the safe course, exactly as for the other saturation poisons. Inferred

Why oral overdose rarely needs more than observation

EXTRIP's evidence review found a low incidence of irreversible injury or death from phenytoin poisoning.¹ The cerebellar syndrome is reversible, there is no toxic metabolite and no acid–base emergency, and the main risks are falls while ataxic and aspiration if consciousness is depressed. The management that follows is protective and supportive, not antidotal.

Why dialysis is almost never the answer for the tablet

High protein binding limits removal, and — more to the point — the poisoning is rarely dangerous enough to warrant it. EXTRIP recommends against dialysis on the basis of dose or concentration alone,¹ reserving it for prolonged coma or incapacitating ataxia. A high phenytoin number is not, by itself, a reason to dialyse. Established

Why the intravenous route is the one to respect

The cardiotoxicity is real, rate-related and occasionally fatal even at recommended doses,³ so it is intravenous loading — not the ingested overdose — that demands cardiac monitoring and a controlled infusion rate. The clinical skill is holding both facts at once: relaxed about the swallowed tablet, careful with the syringe driver.

The antidote, from the poison's side

There is no antidote, and — unusually for this library — that is genuinely fine. Phenytoin has no toxic metabolite to block, no cofactor to replace, and an oral overdose that resolves on its own. The treatment is time, safety and supportive care.

The supportive priorities follow from the organ map: prevent falls and aspiration while the patient is ataxic and drowsy, and look for a co-ingested sedative if consciousness is depressed beyond what cerebellar signs alone explain.² Inferred
Charcoal has a limited role — phenytoin does not recirculate enterocapillarily the way theophylline does, so multi-dose charcoal is not the disease-modifying step it is there. A single early dose after a large ingestion is the usual extent of decontamination, on TOXBASE advice.
Dialysis is reserved, not routine — for prolonged coma or incapacitating ataxia, not for a number.¹ It is the rare phenytoin patient who needs it, and the workgroup's strong recommendations are the ones warning against reaching for it.
The most useful 'antidote' is a corrected concentration — reading the total level against the albumin, so the number matches the patient and neither over- nor under-treats.²

Critical appraisal

The saturation-kinetics account is secure and quoted from the label itself.² Michaelis–Menten elimination explains the decoupling of dose and concentration, the difficulty of therapeutic dosing, and the long tail of recovery, and nothing about it is controversial. It is the cleanest antiepileptic illustration of the principle salicylate demonstrates metabolically.
The concentration thresholds for the neurological signs are approximate and vary widely between patients. The label gives ~20/30/40 mg/L for nystagmus/ataxia/dysarthria but immediately records 50 mg/L without toxicity and 100 mg/L with full recovery.² These are useful bedside anchors, not sharp cut-offs, and the inter-individual variation is large.
The benignity of oral overdose is well supported but not absolute.¹² Deaths do occur, almost always with co-ingestants or in the gravely ill, and prolonged severe exposure can cause irreversible cerebellar atrophy. 'Rarely fatal' is not 'never harmful', and the ataxic patient still needs protecting from falls and aspiration.
The route distinction is the page's central claim and it is strongly evidenced Established — the cardiotoxicity data come from the intravenous label and are explicitly rate- and route-related.³ The one caveat is that a genuinely massive oral ingestion producing an extreme concentration could in principle exert some myocardial sodium-channel effect; the literature simply does not show this to be a practical danger from the tablet.
EXTRIP's phenytoin recommendations rest on 31 patients and very low quality evidence,¹ yet two of them are graded strong — both prohibitions, against dialysing on dose or level. That strength reflects consensus that the poisoning does not warrant dialysis, not strong data that dialysis fails; the distinction matters when reading the grade.
The under-recognised presentation is chronic, not acute. An ataxic, dysarthric patient on long-term phenytoin whose saturable enzyme has been tipped over its ceiling by an interacting drug or a dose change is easily mistaken for a cerebellar or vestibular disorder. The mechanism predicts it, and a phenytoin level, corrected for albumin, settles it.

References

1
Anseeuw K, Mowry JB, Burdmann EA, Ghannoum M, Hoffman RS, Gosselin S, Lavergne V, Nolin TD; EXTRIP Workgroup. Extracorporeal Treatment in Phenytoin Poisoning: Systematic Review and Recommendations from the EXTRIP (Extracorporeal Treatments in Poisoning) Workgroup. Am J Kidney Dis 2016;67(2):187–97. PubMed 26578149 · Recommendation set also at extrip-workgroup.org/phenytoin. Source of the moderately-dialysable (level C) grading despite high protein binding, the neutral 3D general recommendation, the 2D/3D indications for prolonged coma and incapacitating ataxia, the two strong recommendations against dialysing on the basis of dose or concentration alone, and the 51-article / 31-patient evidence base with its conclusion that irreversible injury or death is uncommon. Verified 12 Sep 2026 from the abstract and the workgroup's published recommendation page.
2
Phenytoin 100 mg film-coated tablets — Summary of Product Characteristics. emc product 4225. §4.9 and §5.2 fetched and read in full. Source of the concentration-ordered sign sequence (nystagmus ~20, ataxia ~30, dysarthria/lethargy >40 mg/L, 50 mg/L without toxicity, 25× dose and 100 mg/L with complete recovery), the saturable-hydroxylation statement and its 'small incremental doses… substantial increases' consequence, the ~90% protein binding, the 0.52–1.19 L/kg volume of distribution, the 22 h (range 7–42 h) half-life, the raised free fraction in hypoalbuminaemia and renal or hepatic disease, the 'no known antidote' and 'consider haemodialysis' statements, the cerebellar-atrophy note, and the caution about co-ingested CNS depressants. The label's lethal-dose estimate was read and is deliberately not reproduced here. Verified 12 Sep 2026.
3
Phenytoin Hospira 50 mg/ml Injection BP — Summary of Product Characteristics, Pfizer. emc product 3794. §4.4 and §4.9 fetched and read in full. Source of the intravenous cardiotoxicity statement (bradycardia, atrial and ventricular depression, ventricular fibrillation, asystole/cardiac arrest and death), the observation that severe reactions are most common in elderly or gravely ill patients but occur at recommended doses and rates, and the 25 mg/min (5–10 mg/min in the compromised) infusion-rate limits. Verified 12 Sep 2026.
4
TOXBASE — phenytoin. National Poisons Information Service. toxbase.org (NHS login required. NPIS: 0344 892 0111. The authoritative UK source for monitoring, decontamination and the rare indications for dialysis. Login-gated, so not quoted here.)

Phenytoin

At a glance