Digoxin in Atrial Fibrillation with HFrEF: An Old Drug's Case for a Second Look
A single patient, eight months into a rate-control strategy that is quietly failing. The disagreement isn't about whether the next drug works — it's about whether a drug's decades-old bad reputation is really about the drug itself, or about how it used to be dosed.
H.M. is a 79-year-old man who still spends two mornings a week driving neighbors who no longer trust themselves behind the wheel to their own cardiology and dialysis appointments, an irony he mentions himself now that he is the one struggling to finish the route. His hypertension has been managed for two decades, and a mildly reduced kidney function turned up incidentally on routine labs about two years ago, stable since. Eight months ago, a routine visit picked up an irregular pulse he hadn't noticed himself; the echocardiogram that followed found an ejection fraction of 32%, read at the time as a mix of two decades of hypertensive strain on the heart and a tachycardia-driven component from atrial fibrillation that had likely been running fast, unrecognized, for some stretch beforehand.
Quadruple GDMT went in over the following months as his blood pressure allowed, along with apixaban for stroke prevention. A single attempt at electrical cardioversion four months ago restored sinus rhythm for less than two weeks before the atrial fibrillation returned, and the team settled on rate control as the working strategy rather than chasing a second attempt. For a while, carvedilol held his resting heart rate in the 70s to 80s. Over the past six to eight weeks it has drifted back up to 104–112 at rest, and pushing the carvedilol dose further has meant symptomatic drops into the low 90s systolic — dizziness real enough that his last two attempts at uptitration were both reversed within days. He has started cutting his volunteer route short.
The obvious next rate-control agent, a non-dihydropyridine calcium channel blocker, isn't actually available to him: diltiazem and verapamil are negative inotropes, and adding one to a heart already ejecting less than a third of its volume with each beat risks trading a rate problem for a pump problem. That leaves the room debating a drug with a considerably worse reputation than either of the two agents already tried — digoxin, which slows AV-nodal conduction through vagal potentiation and direct effects without the negative inotropic penalty a calcium channel blocker would carry, but which has accumulated two decades of observational data linking it to excess mortality in atrial fibrillation. Whether that reputation reflects the drug itself, or the way it used to be dosed and monitored, is exactly what the room disagrees about.
Eight months into rate control, and the numbers are drifting back
Add digoxin. He's had two uptitration attempts on carvedilol and two reversals, both for a blood pressure that is simply his baseline now — pushing further doesn't buy more rate control, it buys another dizzy spell and another missed volunteer shift. A non-dihydropyridine calcium channel blocker isn't a real option at an ejection fraction of 32% — negative inotropy on a ventricle already working this hard isn't a theoretical concern, it's a well-documented way to precipitate decompensation. Digoxin slows the ventricular response through vagal potentiation and direct effects on the AV node, with no negative inotropic penalty, and it isn't just decades-old data anymore — RATE-AF randomized digoxin against a beta-blocker head to head for exactly this presentation, and digoxin came out ahead on quality of life at twelve months, with fewer adverse events.
I know what's coming next. AFFIRM's post-hoc mortality signal is the reason most of my own training pushed digoxin toward last-resort status, and I'm not dismissing it — I'm saying the newest, most directly relevant trial we have doesn't reproduce it.
RATE-AF is a genuinely useful trial, and I don't want to talk him out of digoxin — I want to be precise about what it actually supports. It enrolled permanent AF with heart failure symptoms, but look at the ejection fractions: roughly four in five of those patients had an EF of 50% or better. That's a preserved-EF population wearing an HF label. His EF is 32. Extending RATE-AF's result to him is a reasonable extrapolation, not a direct read of the trial's own population, and I'd want that said out loud before it becomes the headline argument.
The mortality question is the other half of this, and it isn't settled just because AFFIRM is observational. What AFFIRM couldn't do, and what DIG's own post-hoc reanalysis could, was tie the mortality signal to serum level — in that analysis the men who sat between 0.5 and 0.8 ng/mL had lower mortality, those between 0.9 and 1.1 showed no excess, and only the ones at 1.2 and above carried higher mortality. His creatinine clearance is already reduced. If we start this drug and don't control the level tightly, we're not testing whether digoxin is safe in HFrEF — we're testing whether we can keep a narrow-therapeutic-index drug narrow in a 79-year-old with reduced clearance, and that's a monitoring commitment, not a prescribing decision.
Both of those things are true, and neither one is a reason not to do this — they're the actual operating instructions. Start low: 0.125 mg daily is a reasonable entry dose at his renal function, not the 0.25 mg median dose used in the original DIG trial, which predates the current recommended target range. Check a trough level in about a week — long enough to approach steady state given his reduced clearance — and the number we're aiming for is 0.5 to 0.9 ng/mL, not the older 0.8 to 2.0 range some references still list. Potassium matters here in a way it wouldn't with any of his other drugs: digoxin and potassium compete for the same binding site on the Na⁺/K⁺-ATPase pump, so a low potassium doesn't just coexist with digoxin toxicity, it potentiates it directly. He's on nothing that lowers potassium today, so his 4.1 is fine. Chart that this particular potassium needs protecting now, before a future decompensation adds a diuretic and nobody remembers why this drug makes that number matter more than it used to.
The calcium channel blocker was never really in competition here — it's contraindicated outright, not just second-choice. The actual comparison is digoxin against doing nothing further, and doing nothing further leaves a 79-year-old with a resting heart rate over 100 and a shrinking volunteer route.
Agreed by the end of the visit: start digoxin at 0.125 mg daily, hold carvedilol at its current dose rather than push it again, and recheck a basic metabolic panel with a digoxin trough in about a week. Diltiazem and verapamil are off the table outright, not held in reserve.
Not agreed, and left as an explicit branch point rather than a shared assumption:
The plan holds as written; his volunteer schedule and functional status get reassessed at the next visit, not the rate number alone.
The cardiologist would add amiodarone next. The electrophysiologist would rather refer for catheter ablation now that the current AF guideline classifies it as a Class 1 recommendation in appropriate patients with reduced ejection fraction — arguing that a rate this stubborn is itself a reason to stop optimizing around it.
Both sides agreed the digoxin trial is worth running either way; what happens if it doesn't fully work is where they diverge, and neither treated the other's position as unreasonable.