From Foxglove to Digoxin: Legacy and Later Research

Table of Contents

1. From Herbal Leaf to Standardised Drug
2. Measuring the Margin: Serum Digoxin Assays
3. An Antidote from Antibodies
4. The DIG Trial (1997)
5. Blood Levels and Outcomes
6. Observational Studies Versus Trials
7. A Withering Drug? The Decline Debate
8. RATE-AF: Digoxin in Atrial Fibrillation
9. DIGIT-HF: Digitoxin Returns
10. The Body’s Own Digitalis and Withering’s Methods Today
11. Key Research Papers
12. Connections
13. Featured Videos

1. From Herbal Leaf to Standardised Drug

Withering worked with the purple foxglove, Digitalis purpurea. He learned quickly that the plant was not a uniform medicine. He gave up the root as too uncertain, chose the leaves gathered when the plant was in flower and carefully dried, and moved from a long-boiled decoction to an infusion and then to a powder so that doses could be more nearly the same from one patient to the next. Even so, the strength of a leaf depended on where and when it grew, how it was dried and how long it was kept. A spoonful of powder was a rough measure of an unknown quantity of the active substances.

The pharmacologist Sir Alasdair Breckenridge, writing in 2006 on Withering’s legacy, placed this problem first among the lessons of the foxglove story: a non-standard preparation gives unpredictable effects. Withering’s careful rules for gathering and drying were an eighteenth-century attempt to tame that variation by hand.

Pure compounds from the plant

The decisive change came with chemistry. The active principles of foxglove turned out to be a family of cardiac glycosides: steroid-like molecules joined to sugars. Once a single glycoside could be separated and crystallised, a dose could be stated as an exact weight of one known substance, rather than as a weight of dried leaf. In 1930 the chemist Sydney Smith reported to the Chemical Society in London the isolation of digoxin, “a new digitalis glucoside”, from a foxglove species; historians attribute it to the woolly foxglove, Digitalis lanata. Digoxin became the most widely used of the foxglove medicines in the twentieth century. Its older relative digitoxin, also a foxglove glycoside, remained in use in some countries and returns later in this story.

The move from leaf to pure compound did not remove the danger Withering had described. It made it measurable. With a fixed dose of a pure compound, the remaining differences in response came from the patient — the kidneys that clear digoxin, the body’s potassium level, age, and other medicines — and that set the stage for the next step, measuring the drug in the blood. The chemistry of the glycosides and how they act on the heart is told on Foxglove: The Plant and the Science of Digitalis.

Back to Table of Contents

2. Measuring the Margin: Serum Digoxin Assays

In his chapter of “Effects, Rules, and Cautions” Withering listed what too much foxglove did: sickness, vomiting, purging, giddiness, confused vision with objects appearing green or yellow, a pulse as slow as 35 a minute, cold sweats, convulsions, fainting and death. For nearly two hundred years doctors had little more to go on than those signs. Toxicity could only be recognised once it had begun.

In 1969 Thomas W. Smith, Vincent P. Butler Jr and Edgar Haber published in the New England Journal of Medicine a method for the “determination of therapeutic and toxic serum digoxin concentrations by radioimmunoassay”. A radioimmunoassay uses an antibody that binds the drug, together with a radioactively labelled version of the drug, to measure very small amounts of it in a blood sample. For the first time a doctor could put a number on the amount of digoxin circulating in a patient and compare it with the concentrations seen in people who were doing well and in people who were poisoned.

This was, in a sense, Withering’s narrow margin made visible. The same blood level could mean different things in different patients, because a low potassium level, for example, makes the heart more sensitive to glycosides; but the assay turned a judgement made from symptoms into one that could be checked against a laboratory value. It also opened the way to research that asked how blood levels related to survival, described in section 5.

The antibodies raised for the assay had a second use. An antibody that can grab digoxin in a test tube can, in principle, grab it in the bloodstream as well.

Back to Table of Contents

3. An Antidote from Antibodies

For most of its history digitalis poisoning had no specific antidote. Doctors could stop the drug, correct potassium and treat dangerous heart rhythms, but they could not take the glycoside back out of the heart muscle.

In 1976 the same group — Thomas W. Smith, Edgar Haber, L. Yeatman and Vincent P. Butler Jr — reported the “reversal of advanced digoxin intoxication with Fab fragments of digoxin-specific antibodies”. The antibodies were raised in sheep and cut into smaller pieces called Fab fragments, the part of an antibody that does the binding. Their patient had taken 22.5 mg of digoxin, many times the amounts used in treatment, and was severely poisoned. According to the published report, the heart returned to a normal (sinus) rhythm about ten minutes after the infusion of the fragments, and the patient’s dangerously high blood potassium fell from 8.7 to 4.0 milliequivalents per litre.

Why fragments rather than whole antibodies

A Fab fragment is smaller than a complete antibody. The fragment keeps the part that binds digoxin while leaving out the rest of the antibody molecule. Digoxin-specific antibody fragments became the standard antidote for serious digoxin and digitoxin poisoning, including poisoning from foxglove and some other plants that contain cardiac glycosides.

The antidote closed a circle that began in Birmingham in 1785. Withering had described the poisoning in detail but could only wait for it to pass; two centuries later, the knowledge of the molecule’s exact shape allowed it to be neutralised.

Back to Table of Contents

4. The DIG Trial (1997)

For two hundred years after Withering, foxglove medicines were given to people with heart failure on the strength of clinical experience. By the late twentieth century the question had become sharper: did digoxin help people with heart failure live longer, or did it only make them feel better? Only a large randomised trial could answer it.

The Digitalis Investigation Group answered with a randomised, placebo-controlled trial published in the New England Journal of Medicine in 1997. It enrolled 6,800 patients with heart failure and a left-ventricular ejection fraction of 0.45 or less — that is, a heart pumping out less than the usual share of its blood with each beat. Half received digoxin and half a placebo, on top of their other treatment.

What the trial found

• Death from any cause: 34.8% in the digoxin group and 35.1% in the placebo group, a risk ratio of 0.99. Digoxin made no difference to overall mortality.
• Hospital admission for worsening heart failure: 26.8% with digoxin and 34.7% with placebo, a risk ratio of 0.72. Fewer people taking digoxin were admitted to hospital because their heart failure had worsened.

The result was neither a vindication nor a rejection. Digoxin did not extend life in this population, but it reduced one of the burdens of heart failure that patients feel most directly, the repeated hospital stays. The DIG trial became the reference point for nearly every later discussion of the drug, and its large data set was later re-analysed to ask a question Withering would have recognised: did it matter how much digoxin was in the blood?

Back to Table of Contents

5. Blood Levels and Outcomes

In 2003 Saif S. Rathore, Harlan M. Krumholz and colleagues published in JAMA a post hoc analysis of the DIG trial: an analysis carried out after the trial, looking at the men in the trial whose serum digoxin concentration had been measured, and comparing them with the men given placebo.

They reported that men with serum digoxin concentrations of 0.5 to 0.8 ng/mL had an absolute mortality rate 6.3 percentage points lower than men taking placebo, while men with concentrations of 1.2 ng/mL or higher had an absolute mortality rate 11.8 points higher. In other words, the possible benefit appeared to sit in a low band of blood levels, and higher levels went with worse outcomes.

Reading the result carefully

Because this was a post hoc, observational analysis within a trial, the authors could report only an association. Patients were not randomised to a particular blood level, and the people who end up with high levels may differ in other ways — for instance in kidney function or in how ill they are — from those with low levels. The analysis also included men only. Its findings have been widely discussed as evidence that, with digoxin, the dose–response curve has a narrow useful band.

That conclusion echoes Withering’s own experience. He admitted that in his early cases he had given foxglove “in doses very much too large” and continued it too long, and he came to believe that the medicine worked best below the point of sickness. The modern analysis speaks the language of nanograms per millilitre, but the theme is the one he set down in 1785.

Back to Table of Contents

6. Observational Studies Versus Trials

In the years after the DIG trial, a number of large observational studies — studies that look at what happens to patients who were or were not prescribed a drug in everyday care — reported that people taking digoxin died more often than people who were not. This seemed to conflict with the neutral mortality result of the randomised trial.

In 2015 Oliver J. Ziff, Dipak Kotecha and colleagues published in the BMJ a systematic review and meta-analysis that put both kinds of evidence side by side. It covered 52 studies with a total of 621,845 patients.

What the meta-analysis found

• In the randomised controlled trials, digoxin had a neutral effect on death from any cause (risk ratio 0.99).
• In the observational studies, digoxin appeared to be linked with more deaths; the authors attributed this apparent harm largely to prescription bias — the tendency for doctors to give digoxin to sicker patients, so that the drug becomes a marker of illness rather than a cause of death.
• Across the studies, digoxin was associated with a small reduction in hospital admissions from any cause (risk ratio 0.92).

The Ziff analysis is often cited as a general lesson in how to read medical evidence: when a drug is given selectively to people who are already at higher risk, comparisons of treated and untreated patients can mislead, even with very large numbers. Randomisation exists to remove that bias.

Withering met an early form of the same problem. In the preface to his book he wrote that he had reported every case in which he had prescribed foxglove, “proper or improper, successful or otherwise”, because selecting only the good cases would mislead. The historian Ulrich Tröhler has described that preface as an early appeal for caution in reporting on a new medicine. The details are on An Account of the Foxglove (1785).

Back to Table of Contents

7. A Withering Drug? The Decline Debate

From 1785 until well into the twentieth century, digitalis was one of the mainstays of treatment for heart failure and, later, for slowing the heart rate in atrial fibrillation. In the second half of the century that position changed. Potent diuretics took over much of the work of removing excess fluid — the “dropsy” that Withering had treated — and newer classes of heart medicine, tested in large trials that showed longer survival, moved to the front of treatment guidelines.

In 2010 Allen B. Weisse wrote an essay in the Journal of Cardiac Failure titled “A fond farewell to the foxglove? The decline in the use of digitalis”. He traced the drug’s long run since 1785 and its de-emphasis after the arrival of strong diuretics and newer agents, and argued that it still had a useful place. His piece is an opinion article, and it records a view held by some clinicians rather than the result of a trial.

The two sides of the debate

The decline debate turned on the findings already described. On one side were the neutral mortality result of the DIG trial, the narrow margin between useful and toxic levels and the observational studies linking digoxin with deaths. On the other were the reduction in hospital admissions in the DIG trial, the evidence that the observational signal reflected prescription bias, the low cost of a very old drug and the possibility that lower blood levels might be both safer and more useful. The question was not settled by argument; it led to new randomised trials, two of which are described in the next sections.

Back to Table of Contents

8. RATE-AF: Digoxin in Atrial Fibrillation

Atrial fibrillation is an irregular, often fast heart rhythm in which the upper chambers of the heart quiver instead of contracting in an orderly way. One common approach to permanent atrial fibrillation is to slow the rate at which the lower chambers beat. Foxglove medicines slow conduction through the junction between the upper and lower chambers, and Withering himself recorded that foxglove slowed the pulse.

The RATE-AF trial, led by Dipak Kotecha at the University of Birmingham — the city where Withering practised and published — compared low-dose digoxin with bisoprolol, a beta-blocker, for rate control in older people with permanent atrial fibrillation and symptoms. The results were published in JAMA in 2020.

What RATE-AF found

• The trial enrolled 160 patients aged 60 or over, with a mean age of 76; 46% were women.
• The primary outcome, patient-reported physical quality of life measured by the SF-36 physical component summary at six months, did not differ significantly between the groups.
• Resting heart rate at six months was similar in both groups (a mean of 76.9 beats per minute with digoxin and 74.8 with bisoprolol).
• On a symptom scale (the modified EHRA class), 53% of patients taking digoxin reported a two-class improvement at six months, compared with 9% taking bisoprolol.
• Adverse events were less common with digoxin: 25% of patients in the digoxin group had at least one adverse event, against 64% in the bisoprolol group.

RATE-AF was a small trial and was designed around quality of life, not survival. Its findings were reported as showing that, in this group of older patients, low-dose digoxin and bisoprolol gave similar quality of life, with fewer adverse events in the digoxin group. The emphasis on a low dose connects it with the blood-level analysis in section 5.

Back to Table of Contents

9. DIGIT-HF: Digitoxin Returns

Digitoxin is the older cousin of digoxin, another cardiac glycoside from the foxglove. It differs in how the body handles it: digoxin is cleared mainly by the kidneys, digitoxin mainly by the liver, and it stays in the body much longer. These differences led a German research group to test digitoxin in a large modern trial.

The DIGIT-HF trial, reported by Udo Bavendiek and colleagues in the New England Journal of Medicine in 2025, randomised patients with heart failure and reduced ejection fraction who were already receiving guideline-based treatment to digitoxin or placebo. The modified intention-to-treat population included 1,212 patients (613 digitoxin, 599 placebo), followed for a median of 36 months.

What DIGIT-HF found

• Primary outcome (death from any cause or first hospital admission for worsening heart failure): 39.5% in the digitoxin group and 44.1% in the placebo group, a hazard ratio of 0.82 (95% confidence interval 0.69 to 0.98).
• Death from any cause alone: 27.2% with digitoxin and 29.5% with placebo, a hazard ratio of 0.86 (0.69 to 1.07) — not statistically significant on its own.
• First hospital admission for worsening heart failure alone: 28.1% and 30.4%, a hazard ratio of 0.85 (0.69 to 1.05), also not significant on its own.
• At least one serious adverse event occurred in 4.7% of the digitoxin group and 2.8% of the placebo group.

DIGIT-HF thus found a lower combined risk of death or heart-failure admission with digitoxin, while neither part of that combination reached significance by itself. It was the first large placebo-controlled trial of a foxglove glycoside in heart failure since the DIG trial, almost three decades earlier, and it brought digitoxin — a substance Withering’s dried leaves contained without anyone knowing its name — back into the research literature on heart failure.

Back to Table of Contents

10. The Body’s Own Digitalis and Withering’s Methods Today

The sodium pump and a puzzle

In 1957 the Danish scientist Jens Christian Skou described an enzyme in nerve membranes that used energy from ATP and depended on sodium and potassium: the sodium–potassium pump, or Na⁺/K⁺-ATPase. This pump turned out to be the molecular target of the foxglove glycosides, which block it. Skou later shared the 1997 Nobel Prize in Chemistry for the discovery. How blocking the pump strengthens the heartbeat is explained on the sodium–potassium pump animation and on this wing’s science page.

The discovery raised a puzzle. The pump carries a highly conserved binding site for plant glycosides — preserved across species through evolution. Why would animals keep a receptor for a compound made by a foxglove? As Vardaman M. Buckalew recounted in a 2015 history in Frontiers in Endocrinology, this question drove a long search for endogenous digitalis-like factors: substances made by the body itself that bind the same site. That search means that a line of inquiry beginning with a Shropshire herbal recipe now reaches into the body’s own chemistry.

Breckenridge’s four lessons

Breckenridge’s 2006 essay, “William Withering’s legacy — for the good of the patient”, drew four lessons from the foxglove story that still frame how new medicines are studied:

1. Standardised preparation. A preparation of uncertain strength gives unpredictable effects; Withering’s rules on gathering and drying the leaf, and later the pure compounds, answered this.
2. Individual differences. Patients differ in their response, both in how sensitive their tissues are and in how their bodies absorb and clear the drug.
3. Dose and response. The effect depends on the dose, and the dose has to be found for each patient; Withering’s careful titration was an early example.
4. Adverse effects. Recording harm is part of studying a medicine; Withering’s list of toxic effects is still recognisable.

Each of the later developments on this page fits one of those lessons: the pure glycosides the first, the blood test and the antidote the second and fourth, the DIG blood-level analysis and the low-dose trials the third. H. A. Bessen, writing in 1986 in the Journal of Emergency Medicine, noted that Withering was largely unaware of the drug’s direct action on the heart — he thought it worked mainly on the kidneys — yet treated many patients who had what is now called congestive heart failure, and recorded striking examples of toxicity. The scientific explanation came long after him; the habits of careful observation and honest reporting were his.

Back to Table of Contents

Key Research Papers

1. Smith S. Digoxin, a new digitalis glucoside. J Chem Soc. 1930:508-510. DOI: 10.1039/JR9300000508
2. Smith TW, Butler VP Jr, Haber E. Determination of therapeutic and toxic serum digoxin concentrations by radioimmunoassay. N Engl J Med. 1969;281(22):1212-6. PubMed PMID: 5388455
3. Smith TW, Haber E, Yeatman L, Butler VP Jr. Reversal of advanced digoxin intoxication with Fab fragments of digoxin-specific antibodies. N Engl J Med. 1976;294(15):797-800. PubMed PMID: 943040
4. Digitalis Investigation Group. The effect of digoxin on mortality and morbidity in patients with heart failure. N Engl J Med. 1997;336(8):525-33. PubMed PMID: 9036306
5. Rathore SS, Curtis JP, Wang Y, Bristow MR, Krumholz HM. Association of serum digoxin concentration and outcomes in patients with heart failure. JAMA. 2003;289(7):871-8. PubMed PMID: 12588271
6. Ziff OJ, Lane DA, Samra M, Griffith M, Kirchhof P, Lip GY, Steeds RP, Townend J, Kotecha D. Safety and efficacy of digoxin: systematic review and meta-analysis of observational and controlled trial data. BMJ. 2015;351:h4451. PubMed PMID: 26321114
7. Weisse AB. A fond farewell to the foxglove? The decline in the use of digitalis. J Card Fail. 2010;16(1):45-8. PubMed PMID: 20123317
8. Kotecha D, Bunting KV, Gill SK, Mehta S, Stanbury M, Jones JC, Haynes S, Calvert MJ, Deeks JJ, Steeds RP, Strauss VY, Rahimi K, Camm AJ, Griffith M, Lip GYH, Townend JN, Kirchhof P; RATE-AF Team. Effect of Digoxin vs Bisoprolol for Heart Rate Control in Atrial Fibrillation on Patient-Reported Quality of Life: The RATE-AF Randomized Clinical Trial. JAMA. 2020;324(24):2497-2508. PubMed PMID: 33351042
9. Bavendiek U, Großhennig A, Schwab J, Berliner D, Rieth A, Maier LS, Gaspar T, Thomas NH, Liu X, Schallhorn S, et al. Digitoxin in Patients with Heart Failure and Reduced Ejection Fraction. N Engl J Med. 2025;393(12):1155-1165. PubMed PMID: 40879434
10. Skou JC. The influence of some cations on an adenosine triphosphatase from peripheral nerves. Biochim Biophys Acta. 1957;23(2):394-401. PubMed PMID: 13412736
11. Buckalew VM. Endogenous digitalis-like factors: an overview of the history. Front Endocrinol (Lausanne). 2015;6:49. PubMed PMID: 25918512
12. Breckenridge A. William Withering’s legacy--for the good of the patient. Clin Med (Lond). 2006;6(4):393-7. PubMed PMID: 16956147
13. Bessen HA. Therapeutic and toxic effects of digitalis: William Withering, 1785. J Emerg Med. 1986;4(3):243-8. PubMed PMID: 3543113
14. Tröhler U. Withering’s 1785 appeal for caution when reporting on a new medicine. J R Soc Med. 2007;100(3):155-6. PubMed PMID: 17339312

PubMed Topic Searches

1. https://pubmed.ncbi.nlm.nih.gov/?term=digoxin+heart+failure+randomized+trial
2. https://pubmed.ncbi.nlm.nih.gov/?term=serum+digoxin+concentration+outcomes
3. https://pubmed.ncbi.nlm.nih.gov/?term=digoxin+atrial+fibrillation+rate+control
4. https://pubmed.ncbi.nlm.nih.gov/?term=digoxin-specific+Fab+fragments
5. https://pubmed.ncbi.nlm.nih.gov/?term=endogenous+digitalis-like+factors

Further Reading

• Withering W. An Account of the Foxglove, and some of its Medical Uses: with Practical Remarks on Dropsy, and Other Diseases. Birmingham; 1785. Project Gutenberg eBook #24886. https://www.gutenberg.org/ebooks/24886

Back to Table of Contents

Connections

• William Withering: Foxglove and the Birth of Digitalis
• William Withering: Life and Career
• An Account of the Foxglove (1785)
• Foxglove: The Plant and the Science of Digitalis
• Pharmacology
• Heart Failure
• History of Heart Failure
• Atrial Fibrillation
• History of Atrial Fibrillation
• Arrhythmia
• The Sodium–Potassium Pump
• Potassium

Back to Table of Contents

Featured Videos