Why this poison is interesting

Digoxin opens Band B because it breaks the assumption that quietly underlies most of clinical toxicology: that the concentration you measure bears a fixed relationship to the amount of poison in the patient. For digoxin it does not, and the reason is structural rather than technical. The drug distributes into a volume of about 510 litres — several times total body water — and reaches a concentration in cardiac tissue roughly thirty times that in plasma.¹ The blood is the one compartment where the poison mostly is not.

Three consequences follow, and between them they organise this page. A level drawn during the six-to-eight-hour distribution phase is not a low reading of a stable system but a high reading of a falling one.¹ A dialyser offered the plasma compartment removes about 3% of the body burden in five hours, which is why the EXTRIP workgroup recommends against extracorporeal treatment in any form, at any severity, in any clinical context.¹³ And an antidote that works only in plasma nonetheless empties the tissue — because binding free drug creates the gradient that drags the rest back.

The second reason digoxin is interesting is that its principal laboratory abnormality is not a complication. In acute poisoning the serum potassium rises because the pump that is being poisoned is the pump that holds potassium inside cells. The hyperkalaemia is therefore a direct read-out of the fraction of Na⁺/K⁺-ATPase inhibited — the closest thing in this library to a bedside assay of receptor occupancy. That is why it carried prognostic weight in the era before any antidote existed.⁴

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

The toxic principle

Digoxin binds the extracellular face of Na⁺/K⁺-ATPase and holds it in a conformation that cannot complete its cycle. The pump normally exports three sodium ions and imports two potassium ions per molecule of ATP, and it does this continuously in every cell as the price of maintaining the resting membrane potential. Inhibit it and two things happen at once, in every tissue that expresses it.

The therapeutic and the toxic effects are therefore the same effect at different degrees of the same receptor occupancy. There is no separate toxic mechanism to look for, and no metabolite to blame. This is unusual: most of the poisons in this library kill by a mechanism the therapeutic dose never invokes.

Digoxin has a second action, unrelated to the pump and important at the bedside. It is vagotonic, increasing parasympathetic tone at the sinoatrial and atrioventricular nodes. That produces the sinus bradycardia and the AV block, while the calcium overload produces the ventricular ectopy — which is why the classic digoxin rhythms combine suppressed conduction with increased automaticity. A rhythm that is simultaneously slow at the node and irritable in the ventricle is close to a signature.

Paroxysmal atrial tachycardia with block: An atrial focus firing rapidly (automaticity, from calcium overload) while the AV node refuses to conduct it (vagotonia). Both halves are digoxin.¹ Established
Bidirectional ventricular tachycardia: Alternating QRS axis beat to beat. Rare, and rare enough outside digitalis and catecholaminergic polymorphic VT that it is close to pathognomonic. Listed explicitly in the label as a manifestation of toxicity.¹ Established
Regularised atrial fibrillation: Atrial fibrillation with a strikingly regular ventricular rate — the label describes slow atrial fibrillation with very little variation in the ventricular rate.¹ The atria are still fibrillating; complete AV block has been superimposed and a junctional escape is now driving the ventricle. An irregularly irregular rhythm becoming regular is a deterioration, not an improvement. Established
Any arrhythmia at all: The label's own position is that digoxin toxicity may result in almost any type of arrhythmia, and that multiple rhythm disturbances in the same patient are common.¹ The pattern-recognition above helps when it is present and proves nothing when it is absent.

Toxicokinetics

Digoxin — a drug that is almost entirely somewhere other than the blood
Parameter	Therapeutic	In overdose	Why it changes
Bioavailability	63% from tablets, 75% from oral solution¹	Unchanged in kind; large ingestions of tablets may absorb more slowly	The formulation difference matters when a dose is estimated from tablets found at the scene, and it is one reason a dose-based estimate is weaker than a concentration-based one.
Time to peak	Oral Tmax 2 to 6 h; intravenous Tmax 1 to 5 h¹	Similar, but the clinically relevant peak is the cardiac one — peak cardiac effects at 3 to 6 h, persisting 24 h or longer¹	The plasma peak and the effect peak are not the same event, because the effect follows tissue concentration, which lags plasma. This is the whole reason for the six-hour rule below.
Volume of distribution	Vdss 510 L in healthy volunteers — extensively tissue-bound. Highest concentrations in heart, liver and kidney; cardiac tissue averages 30-fold plasma¹	Unchanged — and this is precisely the problem	The single most important number on this page. A 510-litre volume means the plasma holds a negligible fraction of the body burden. Skeletal muscle alone is about 40% of body weight and, though its concentration is far lower than the heart's, the label warns this store cannot be overlooked.¹ Everything downstream — the uninterpretable early level, the failure of dialysis, the size of the Fab dose — follows from this one figure.
Distribution phase	Initial distribution from central to peripheral compartment lasts 6 to 8 h¹	Unchanged in duration; the concentrations traversed are higher	A concentration drawn inside this window is not a steady-state measurement. It is a snapshot of a compartment that is still emptying into tissue, and it overstates the eventual tissue-equilibrated concentration. The DigiFab label therefore specifies measurement 6 hours after overdose, unless an arrhythmia forces the issue earlier.²
Protein binding	Approximately 25% of the small plasma fraction¹	Unchanged	Low binding would ordinarily favour extracorporeal removal. It is comprehensively outweighed by the volume of distribution — a reminder that dialysability is a property of both numbers and that protein binding alone predicts nothing.
Metabolism	Minimal. A small fraction to dihydrodigoxin and digoxygenin, both largely inactive¹	Unchanged	There is no bioactivation and no toxic metabolite. This is a rarity in this library — most pages here are about what the liver makes. Hepatic impairment has little effect on digoxin clearance.¹
Elimination	Renal excretion of unchanged drug; 60–75% recovered unchanged in urine. Total clearance 193 ± 25 mL/min, renal clearance 152 ± 24 mL/min¹	Unchanged mechanism, but clearance is directly proportional to renal function	Digoxin is a P-glycoprotein substrate, and P-gp in the proximal tubule is an important part of its renal handling.¹ Every P-gp inhibitor — amiodarone, verapamil, clarithromycin, ciclosporin — is therefore a route into chronic toxicity in a patient whose dose has not changed.
Half-life	30 to 40 h with normal renal function¹	Of the order of 100 h in anuric patients¹	A long half-life in a drug with a narrow therapeutic index is the arithmetic of chronic toxicity: several days of any new impairment are enough to move a stable patient into the toxic range without a single change of dose.
Order of kinetics	First order	First order	Nothing saturates. Digoxin poisoning is not a kinetic escape — it is a distribution problem and a target-ubiquity problem.
Dialysability	—	No. Only about 3% of a digoxin dose is removed during 5 h of haemodialysis¹	EXTRIP is unusually absolute: extracorporeal treatment in any form is not indicated for suspected or proven digoxin toxicity regardless of the clinical context, and is not indicated for removal of the digoxin-Fab complex either.³ Most EXTRIP recommendations carve out an exception for the severe case; this one does not, because the 510-litre volume of distribution does not change with severity.

Metabolism and the metabolites

This section is short, and its shortness is the point. Digoxin is not bioactivated. The molecule that is swallowed is the molecule that inhibits the pump, and the liver plays almost no part in the story.

Digoxin — the pathway that mostly does not happen

Digoxin (oral)Already the toxic species. Bioavailability 63% tablet, 75% solution¹
P-glycoprotein efflux in the enterocyteLimits absorption; inhibited by amiodarone, verapamil, clarithromycin and others¹
Renal excretion of unchanged drugDigoxin in urineThe major route — 60–75% of an IV dose over six days¹
Colonic bacterial reduction (a minority of people)Digoxin reduction productsCardio-inactive. In these individuals over 40% of the dose may be excreted as reduction products¹
Minor hepatic metabolismDihydrodigoxin · digoxygeninSmall fraction; renal clearances 79 ± 13 and 100 ± 26 mL/min respectively¹

The practical residue of all this is that hepatic impairment is not a route into digoxin toxicity, and renal impairment is. That asymmetry is the opposite of most of the pharmaceutical poisons in this library, and it is worth carrying deliberately: the patient who becomes digoxin-toxic without a dose change has almost always lost renal function, gained a P-glycoprotein inhibitor, or become dehydrated.

Elimination and accumulation

Digoxin's elimination is simple and its accumulation is not. Clearance tracks creatinine clearance almost linearly, the half-life runs from 30–40 h to about 100 h in anuria, and there is no metabolic step to saturate.¹ What makes accumulation dangerous is that the therapeutic index is narrow and the reservoir is enormous: by the time the plasma concentration has risen, a large tissue burden has already been laid down.

The label places the threshold at which signs and symptoms become more frequent above 2.0 nanograms/mL (2.56 nanomol/L), while noting considerable inter-individual variation.¹ That variation is not a measurement problem. It reflects the fact that the same plasma concentration sits on top of different tissue burdens depending on how it got there, and that the pump's sensitivity is modified by potassium, magnesium, calcium and thyroid status.

Target organs — and why those

The honest answer to why that organ for digoxin is all of them — Na⁺/K⁺-ATPase is ubiquitous. What differs between tissues is how much a partially inhibited pump matters there, and how visible the consequence is.

Myocardium and conducting system

TargetNa⁺/K⁺-ATPase in cardiomyocytes; vagal afferents to the SA and AV nodes

Why hereNot because the pump is more sensitive here, but because cardiac tissue concentrates the drug about 30-fold over plasma¹, and because the heart is the one organ in which a modest change in intracellular calcium is immediately converted into a rhythm. Calcium overload generates delayed afterdepolarisations and therefore automaticity, while the vagotonic action suppresses nodal conduction — the two acting together produce the characteristic combination of a slow node and an irritable ventricle. Established

At the bedsidePeak cardiac effects 3 to 6 h after overdose, persisting 24 h or longer.¹ Premature ventricular contractions are often the earliest and commonest arrhythmia; PR prolongation may be the only early sign.¹

Skeletal muscle

TargetThe same pump, in the largest tissue mass in the body

Why hereSkeletal muscle is about 40% of body weight, so even at a much lower concentration than myocardium it holds a substantial share of the body burden — the label warns explicitly that this store cannot be overlooked.¹ It is also the principal source of the potassium that appears in serum when the pump is inhibited: the hyperkalaemia of acute digoxin poisoning is largely skeletal muscle letting go. Inferred

At the bedsideExplains why the potassium tracks severity in acute poisoning⁴, and why it is not a marker of renal failure in that setting.

Gut and area postrema

TargetPump inhibition in enteric tissue, plus direct stimulation of the chemoreceptor trigger zone

Why hereThe area postrema sits outside the blood–brain barrier and samples plasma directly, which is why a drug that penetrates the CNS poorly can still cause profound vomiting. Nausea and vomiting are reported in up to 80% of cases and precede cardiac manifestations in roughly half.¹ Inferred

At the bedsideVomiting is usually the first symptom of acute poisoning, and in the chronic patient it is easily attributed to the gastroenteritis that in fact precipitated the toxicity by causing dehydration.

Retina

TargetNa⁺/K⁺-ATPase in photoreceptors and retinal pigment epithelium

Why herePhototransduction depends on steep transmembrane ion gradients that the pump maintains; partial inhibition degrades colour discrimination before it degrades acuity. The label records that the most frequent visual disturbance is an aberration of colour vision with a predominance of yellow-green.¹ The receptor-level account of why yellow-green specifically is a reasonable inference from cone physiology rather than something demonstrated in poisoned humans. Inferred

At the bedsideXanthopsia is a chronic-toxicity sign far more often than an acute one, and the label notes visual and neurological symptoms may persist after other signs have resolved.¹

Kidney

TargetNot a target of injury — a determinant of exposure

Why hereIncluded deliberately as the organ digoxin does not damage. The kidney is where digoxin leaves, and P-glycoprotein in the proximal tubule is an important part of that exit.¹ Renal function is therefore the variable that converts a stable prescription into a poisoning. Established

At the bedsideAnuria extends the half-life to roughly 100 h.¹ Any new AKI, diuretic, NSAID or P-gp inhibitor is a plausible precipitant in a chronically toxic patient.

Timeline of effects

Acute digoxin overdose — the gap between the plasma peak and the effect peak

Time

What you seeWhat is happening

0–2 hAbsorption
What you seeOften nothing, or early nausea. The patient may look entirely well.
What is happeningDigoxin absorbed from stomach and upper small intestine; plasma concentration climbing towards a Tmax of 2–6 h.¹ Tissue uptake has barely begun.
1–6 hGastrointestinal phase
What you seeNausea and vomiting, reported in up to 80% of cases and preceding cardiac features in about half.¹
What is happeningArea postrema stimulation plus enteric pump inhibition. This is the phase in which a concentration is most misleading — it is high, and the tissue compartment is still filling.
3–6 hCardiac peak
What you seePR prolongation, ventricular ectopy, bradyarrhythmias, AV block; in severe poisoning the characteristic combined patterns.
What is happeningPeak cardiac effects occur 3 to 6 hours following overdose.¹ Tissue concentration is now approaching its maximum as the 6–8 h distribution phase completes.¹ Intracellular calcium overload produces afterdepolarisations.
≥6 hThe interpretable window
What you seeClinical picture established; potassium informative.
What is happeningDistribution complete, so a concentration now reflects the tissue burden. The DigiFab label specifies measurement at 6 hours unless arrhythmia forces it earlier.²
6–24 h+Plateau
What you seeArrhythmias may persist for 24 hours or longer.¹ Visual and neurological symptoms may outlast the cardiac ones.¹
What is happeningElimination half-life 30–40 h with normal kidneys — so the body burden falls slowly, and there is no metabolic route to accelerate it.
DaysAfter Fab, in renal impairment
What you seeNothing, usually — but this is where rebound lives. A recovered patient with poor renal function is not yet finished.
What is happeningFab–digoxin complexes are cleared renally; if that is impaired they are retained, and digoxin may be released after some days from retained complexes.² Free digoxin rebound was observed in 25 of 36 patients in ATOM-1, though none deteriorated clinically.⁷

What the mechanism predicts at the bedside

Everything in this section is a consequence of two facts already established: the drug lives in tissue rather than plasma, and its target is a pump present on every cell.

A digoxin concentration before six hours cannot be used the way a six-hour concentration can. It is not that the assay is unreliable; it is that the compartment is still emptying. The label's distribution phase is 6–8 h and the DigiFab label's sampling instruction is 6 h.¹²
Potassium is the most informative single number in acute poisoning, because it estimates the fraction of pump inhibited rather than the concentration in a compartment the poison has mostly left.⁴ In chronic toxicity it loses that meaning and may be low.
Dialysis is futile, and this is not a close call. About 3% removed in five hours, against a 510-litre volume of distribution.¹ EXTRIP recommends against it regardless of clinical context — including for the Fab complex.³
An irregularly irregular rhythm that becomes regular has usually got worse. Regularisation of atrial fibrillation means complete AV block with a junctional escape, not restored sinus rhythm.¹
Vomiting in a patient on long-term digoxin is a double signal — it may be the toxicity, and the dehydration it causes is also a mechanism by which toxicity develops. Both readings are correct at once.
A new prescription is a plausible cause of toxicity without any dose change, because digoxin is a P-glycoprotein substrate.¹ So is a course of antibiotics, by a different route entirely — inactivating the gut flora that were destroying part of the dose.⁶
Total digoxin concentrations become uninterpretable after Fab. The immunoassay measures bound drug too, so the total rises sharply while the free — pharmacologically active — fraction falls to nothing. The label warns this may persist several days or more than a week in renal impairment.² A rising level after antidote is expected and is not treatment failure.

The antidote, from the poison's side

Digoxin-specific antibody fragments are the most mechanistically satisfying antidote in Band B, and the one with the least secure outcome evidence in its commonest indication. Both halves of that sentence deserve their space.

This also explains the two peculiarities of Fab that most often confuse. Total measured digoxin rises steeply after administration, because the immunoassay counts bound drug; it is the free fraction that matters and that has collapsed.² And the complex is cleared renally, so in renal impairment the complexes are retained and digoxin may be released after some days — the label says so explicitly.² Rebound is a predictable consequence of the mechanism, not an anomaly.

The UK label's dosing is deliberately practical rather than pharmacokinetic. Where a concentration is available — measured 6 hours after overdose, unless arrhythmia forces it earlier — the scheme is:

The formula is a body-burden calculation wearing ordinary clothes. Concentration multiplied by weight is a proxy for total body load given a roughly fixed volume of distribution per kilogram; dividing by a constant converts milligrams of digoxin into vials of Fab. A dosing rule of this shape is only possible because the volume of distribution is large and consistent — the same property that makes the poison undialysable makes the antidote calculable.

And then the outcome data

ATOM-1 recruited 36 patients with chronic digoxin poisoning (median age 78, median heart rate 49/min, median digoxin 4.7 nmol/L, median potassium 5.3 mmol/L). Fab bound the digoxin exactly as predicted — free digoxin fell from 3.8 to 0 nmol/L — but the median heart rate change was 8 beats/min and the median potassium fall 0.3 mmol/L. The authors concluded that the bradyarrhythmia and hyperkalaemia may be arising from other co-morbidities.⁷ ATOM-4 then compared 78 patients given Fab with 50 managed supportively: mortality 9 (12%) versus 7 (14%), risk difference −2.5% (95% CI −14 to 9%, p = 0.68), median length of stay six days in both groups, and no difference in potassium or heart-rate change. The conclusion was that the study did not appear to show any benefit from the routine use of digoxin-Fab in patients thought to have chronic digoxin poisoning.⁸ Established

Read carefully, this is not evidence that Fab does not work. It is evidence that in chronic toxicity the thing being treated may not be the thing causing the presentation. These were patients aged around 80 with renal impairment and multiple cardioactive co-prescriptions — 18 of 36 in ATOM-1 were also on beta-blockers, six on calcium antagonists.⁷ Removing the digoxin from such a patient removes one of several causes of their bradycardia. The acute, large, single-ingestion presentation — the one with the hyperkalaemia and the rising concentration — is a different disease, and neither ATOM study was designed to test Fab in it.

Critical appraisal

The 'calcium causes stone heart in digoxin toxicity' teaching is not supported by the evidence usually invoked for it. The concern — that giving calcium to a calcium-overloaded myocardium precipitates irreversible contraction — is mechanistically plausible and is widely taught as established. Levine and colleagues reviewed 159 records of digoxin toxicity, of which 23 patients received intravenous calcium, and found no association with increased mortality (22% versus 20%; odds ratio 0.76, 95% CI 0.24–2.5) or with dysrhythmias.⁵ That is a retrospective study with 23 exposed patients and a confidence interval wide enough to contain a substantial effect in either direction, and it does not license giving calcium; it does mean the prohibition rests on animal data and case reports rather than on human evidence, and it should be taught as such. Traditional teaching
The retinal mechanism of xanthopsia is inference. That colour vision is disturbed, and that yellow-green predominates, is recorded in the label.¹ That this specifically reflects Na⁺/K⁺-ATPase inhibition in cone photoreceptors is a reasonable reconstruction from retinal physiology, not a demonstrated causal chain in poisoned humans. Inferred
Bidirectional VT is quoted more often than it is seen. It is listed in the label among the arrhythmias digoxin may cause¹, and the label's own framing — almost any type of arrhythmia, with multiple rhythm disturbances in the same patient being common — is a better guide to practice than the pattern-recognition list. Its absence excludes nothing.
Hyperkalaemia's prognostic weight comes from a pre-antidote cohort. Bismuth's 1973 series established the association at a time when no specific treatment existed, which is what makes it clean as prognosis — and also what makes it a poor guide to the modern treated patient, whose potassium is being actively managed.⁴
The DigiFab dosing formulae are estimation tools, not pharmacokinetic truth. Both routes in the UK label are approximations built on an assumed volume of distribution and an assumed neutralisation capacity per vial, and both instruct rounding up.² Where the label gives fixed weight-banded doses for the arrest and severe-arrhythmia indications, that is an acknowledgement that in those situations there is no time for an estimate.
The ATOM findings are about chronic toxicity and should not be generalised to acute overdose. Both studies recruited patients with elevated concentrations and symptoms attributed to digoxin, predominantly elderly with renal impairment and co-morbidities.⁷⁸ Extending 'no benefit' to the young patient who has swallowed a bottle would be an unsupported inference in the other direction.
EXTRIP's digoxin recommendation is unusually unhedged, and the reason is arithmetic. Most EXTRIP statements carve out the severe case; this one does not, because 3% removal in five hours against a 510-litre volume does not improve with severity.¹³ It is one of the few places where a guideline can be re-derived from two numbers on the label.

How this page connects to the rest of Band B

Digoxin is the band's outlier: it is the only member whose toxicity is not a channel blockade. Its neighbours poison calcium channels, beta-adrenoceptors and sodium channels; digoxin poisons a pump, and the difference shows up in the potassium. Lithium is its mirror image on every kinetic parameter that matters — an ion with a 0.7–0.9 L/kg volume of distribution, no protein binding and no metabolism, and therefore the band's most dialysable poison against the band's least. The two pages share a second structure as well: lithium is dialysable at level of evidence A while every EXTRIP recommendation about when to dialyse it is graded D, which is the same shape as Fab binding digoxin exactly as designed while ATOM-4 could not show a mortality benefit. An intervention can provably do what it claims and still not be shown to change the outcome.

References

1
Digoxin 0.25 mg Tablets — Summary of Product Characteristics. electronic medicines compendium, product 5465. Sections 4.9 (Overdose) and 5.2 (Pharmacokinetic properties). medicines.org.uk/emc/product/5465
2
DigiFab 40 mg/vial digoxin immune Fab (ovine) — Summary of Product Characteristics. electronic medicines compendium, product 102213, updated 8 June 2026. Sections 4.2 (Posology), 4.4 (Warnings — immunoassay interference) and 5.2. medicines.org.uk/emc/product/102213
3
EXTRIP Workgroup. Digoxin — recommendations. Extracorporeal Treatments in Poisoning Workgroup. extrip-workgroup.org/digoxin
4
Bismuth C, Gaultier M, Conso F, Efthymiou ML. Hyperkalemia in acute digitalis poisoning: prognostic significance and therapeutic implications. Clinical Toxicology 1973;6(2):153–62. PMID 4715199.
5
Levine M, Nikkanen H, Pallin DJ. The effects of intravenous calcium in patients with digoxin toxicity. The Journal of Emergency Medicine 2011 Jan;40(1):41–6. PMID 19201134.
6
Lindenbaum J, Rund DG, Butler VP Jr, et al. Inactivation of digoxin by the gut flora: reversal by antibiotic therapy. New England Journal of Medicine 1981 Oct 1;305(14):789–94. PMID 7266632.
7
Chan BS, Isbister GK, O'Leary M, Chiew A, Buckley NA. Efficacy and effectiveness of anti-digoxin antibodies in chronic digoxin poisonings from the DORA study (ATOM-1). Clinical Toxicology 2016 Jul;54(6):488–94. PMID 27118413.
8
Chan BS, Isbister GK, Page CB, Isoardi KZ, Chiew AL, Kirby KA, Buckley NA. Clinical outcomes from early use of digoxin-specific antibodies versus observation in chronic digoxin poisoning (ATOM-4). Clinical Toxicology 2019 Jul;57(7):638–43. PMID 30585517.

Digoxin

At a glance