If an elderly patient refuses an ECG, does that make stopping long-term digoxin safer, or does it increase uncertainty about stopping it?

The ECG is not a test you are doing to the patient. It is a test you are doing to the drug.

Stopping long-term digoxin without an ECG does not make the withdrawal safer. It makes it less safe, and it increases uncertainty in both directions at once. You are removing the only cheap, non-invasive window into two things that matter: whether the drug has accumulated to a toxic level, and whether the heart has developed an arrhythmia that the drug was either causing or suppressing.

Digoxin has a narrow therapeutic window. The difference between a working dose and a toxic dose is small, and in an elderly patient that window narrows further because kidney function declines with age, and digoxin is cleared by the kidneys. A serum level drawn alongside an ECG tells you whether you are starting from a safe baseline. Without it, you are stopping blind. If the level is high, stopping is urgent and may need to be done with monitoring for rebound arrhythmias. If the level is low, stopping is lower risk, but you still need to know what rhythm the heart is in before you pull the drug, because digoxin can be holding an atrial arrhythmia in check, and withdrawal can unmask a rapid ventricular response or heart failure exacerbation.

The uncertainty cuts both ways. Refusing the ECG does not protect the patient from the risks of stopping. It hides them. You cannot know if the withdrawal is safe because you cannot know what the drug was doing at the moment you stopped it. The absence of a test does not create the absence of a problem. It creates the absence of information, and in a drug with this toxicity profile, information is the safety.

Now the deeper question: does stopping long-term digoxin in a stable elderly patient even make sense? The evidence base for chronic digoxin in heart failure with preserved ejection fraction is thin. In heart failure with reduced ejection fraction, digoxin reduces hospitalizations but does not clearly reduce mortality, and the DIG trial showed a trend toward more sudden death in the higher-dose group. So there is a legitimate argument for withdrawal in some patients. But that argument is stronger when you have a baseline ECG and a digoxin level, because you can then weigh the risk of the drug against the risk of stopping it. Without those, you are making a decision on incomplete data, and the burden of proof shifts onto the withdrawal, not the drug.

The systematic review you mention, on withdrawing RAAS inhibitors and beta-blockers in stable heart failure, is a different class of drug. Those have wider therapeutic windows and clearer withdrawal syndromes. Digoxin is not in that category. Extrapolating from that review to digoxin would be a category error. The review also tends to enroll younger, more stable patients than the frail elderly you are asking about, and it rarely captures the polypharmacy and renal impairment that dominate geriatric practice.

So the honest answer is this: refusing the ECG does not make stopping safer. It makes it riskier, because it removes the only objective check on the two most dangerous outcomes, digoxin toxicity and arrhythmia. If the patient refuses, you do not proceed with a shrug. You document the refusal, you explain what you cannot know, and you either stop with a lower threshold for urgent evaluation if symptoms appear, or you do not stop at all and instead reduce the dose and monitor clinically. The uncertainty is real, and it is increased, not decreased, by the refusal. The evidence is not clear on whether stopping digoxin in this population is net beneficial, and without the ECG, you cannot even get to the starting line of that question. Confidence in that assessment is high, because it rests on pharmacology and physiology, not on a contested meta-analysis.