For many people with atrial fibrillation, coffee has lived in the same mental drawer as alcohol, missed sleep and other suspected triggers. The logic felt plausible. Caffeine can make the heart feel faster or more noticeable, so taking it away might help keep an irregular rhythm away.
That advice has been handed out often enough to become part of the condition’s folklore. Yet the evidence beneath it was never as tidy as the warning. Observational studies generally failed to show that ordinary coffee consumption caused more atrial fibrillation, and people’s perception of caffeine as a trigger did not consistently match rhythm recordings.
Now a randomized trial has made the old story harder to repeat. In DECAF, 200 adults who had persistent atrial fibrillation, or atrial flutter with a history of AF, were assigned after electrical cardioversion either to consume caffeinated coffee or to avoid coffee and caffeine. Over six months, AF or flutter returned in 47 people in the coffee group and 64 in the abstinence group.
The result is counterintuitive, clinically interesting and stronger than a simple association. It is also narrower than a general claim that coffee protects the heart. Understanding both sides is the difference between useful evidence and a new piece of medical folklore.
The blanket advice had already begun to change
Atrial fibrillation occurs when electrical activity in the heart’s upper chambers becomes disorganized. The atria quiver instead of contracting in a coordinated way, producing an irregular rhythm that can bring fatigue, breathlessness, dizziness or palpitations. AF also raises stroke risk, which is why treatment can involve blood-thinning medication as well as efforts to control heart rate or restore rhythm.
Electrical cardioversion delivers a controlled shock to reset the rhythm. It can work immediately without guaranteeing that normal rhythm will last. Recurrence is common, making the months after cardioversion a meaningful period in which to test a suspected trigger.
Caffeine’s stimulating reputation made it an easy suspect. But by the time DECAF began reporting results, formal guidance had already softened. The 2023 US guideline for atrial fibrillation says that recommending caffeine abstinence to prevent AF episodes is of no benefit. It leaves room for an important exception: avoidance may reduce symptoms in people who report that caffeine triggers or worsens them.
That is not a contradiction. Feeling a forceful or rapid heartbeat and having a documented AF episode are related but distinct outcomes. Caffeine may heighten palpitations, anxiety or awareness of the heart without creating the electrical pattern of AF. A person may reasonably dislike those symptoms even if the drink is not provoking recorded arrhythmia.
What the DECAF trial randomized
The DECAF trial, whose name stands for Does Eliminating Coffee Avoid Fibrillation?, enrolled participants at five hospitals in the United States, Canada and Australia. Recruitment ran from November 2021 through December 2024, with follow-up completed in June 2025. The results were published online by JAMA in November 2025.
All participants had persistent AF, or atrial flutter plus a history of AF, and were scheduled for electrical cardioversion. They also had to be current coffee drinkers or to have consumed coffee within the previous five years. Their mean age was 69, 71% were men, and both trial groups reported a median of seven cups of coffee a week at the start.
After successful cardioversion, 100 people were encouraged to drink at least one cup of caffeinated coffee each day. Another 100 were encouraged to abstain not only from caffeinated coffee but also from decaf and other caffeine-containing products.
The headline shorthand is “one cup a day,” but the protocol said at least one. In practice, the coffee group reported a median of seven cups a week during follow-up, with an interquartile range of six to 11. The abstinence group reported a median of zero, with an interquartile range of zero to two. This was an experiment in ordinary behavior, not a laboratory dose of purified caffeine.
What 47%, 64% and “39% lower” each mean
By six months, 47 of the 100 people assigned to coffee had a clinically detected recurrence of AF or atrial flutter. In the abstinence group, 64 of 100 did. The raw difference was therefore 17 percentage points.
The widely quoted 39% reduction comes from a different calculation. Researchers used a time-to-event model adjusted for study site and obtained a hazard ratio of 0.61, with a 95% confidence interval from 0.42 to 0.89. In plain language, the rate at which recurrences appeared through follow-up was estimated to be 39% lower in the coffee group.
A hazard ratio is not a promise that any particular patient will reduce their personal probability by 39 percentage points. It also is not the same as subtracting 47 from 64. One number describes events accumulated over time; the other compares the final proportions observed in this particular six-month trial.
As ScienceBlog reported when DECAF first appeared, random assignment is what makes the result unusually valuable. It reduces the chance that the difference merely reflects the healthier habits, income, diet or other characteristics of people who choose to drink coffee. Still, a 200-person trial remains modest, and the confidence interval shows that the exact size of the effect is uncertain.
How the study decided that AF had returned
The primary outcome was clinically detected AF or atrial flutter lasting at least 30 seconds. A physician had to confirm it from an electrocardiogram, wearable ECG recording or electrograms stored by an implanted cardiac device. The investigators also reviewed ordinary clinical records and rhythm documentation from participants’ physicians.
Just over half of the participants had a device capable of continuous rhythm recording. The rest depended on the rhythm checks and clinical encounters that occurred during normal care. That makes the outcome pragmatic: it resembles how recurrence is discovered outside a tightly instrumented laboratory.
It also means the study did not continuously measure every participant in the same way. Brief or silent episodes could have gone undetected more often in people without continuous monitoring. The trial addressed this concern in additional analyses, but it was designed around clinically detected recurrence rather than a complete second-by-second measure of arrhythmia burden.
The people judging rhythm evidence could work from objective tracings, but the study itself was open-label. Participants knew whether they had been asked to drink or avoid coffee. Blinding would have been difficult because coffee has an obvious taste, routine and physiological feel, yet knowledge of assignment can still change behavior and reporting.
The experiment changed more than caffeine
It is tempting to point to caffeine and conclude that the stimulant protected the heart. DECAF cannot isolate that mechanism. One group continued caffeinated coffee; the other was asked to stop coffee, including decaf, and avoid other sources of caffeine. The comparison bundled a beverage, a chemical and a familiar daily ritual.
Coffee contains chlorogenic acids and many other compounds with antioxidant or anti-inflammatory activity. Caffeine blocks adenosine receptors and may have electrical effects that are more complicated than simply “speeding up” the heart. The researchers also noted that coffee consumption has been linked to more physical activity in another randomized experiment, and regular activity can support AF management.
There are less glamorous possibilities. Removing a familiar source of caffeine can cause headaches, fatigue or changes in routine. Participants assigned to abstinence may have replaced coffee with different drinks or altered sleep and activity. Conversely, people continuing coffee could experience worse sleep or higher blood pressure, depending on dose and individual sensitivity. DECAF was not built to decide which pathway mattered.
Adherence adds another layer. Coffee intake was self-reported, and only 69% of the abstinence group reported consuming no coffee during follow-up. A pragmatic trial captures the messiness of real behavior, but it cannot turn a cup into a precisely measured drug.
The population boundary is unusually important here
DECAF did not recruit people at random from the population. It studied previous or current coffee drinkers who had persistent AF, or related flutter with an AF history, and who had just undergone successful cardioversion. It therefore answers a practical question about continuing a familiar habit in a specific clinical setting.
It does not show that a non-coffee drinker should start drinking coffee. It does not establish that coffee prevents a first episode of AF, and it does not settle the question for people with paroxysmal AF whose episodes begin and end on their own. It says nothing reliable about high-dose caffeine tablets or multi-ingredient energy drinks.
Selection may matter in another way. People who strongly believed coffee provoked their arrhythmia could have stopped drinking it long before recruitment or declined to enter a study that might assign them coffee. The sample may therefore underrepresent the people most sensitive to the drink.
An earlier personalized-trigger experiment, covered by ScienceBlog in 2021, found no overall near-term link between caffeine and AF episodes even though caffeine was the trigger participants most often chose to test. Taken together, the studies weaken a blanket rule. They do not prove that no individual has a repeatable caffeine-related response.
Six months can reassure without settling long-term safety
Adverse events were broadly similar. The coffee and abstinence groups had 13 versus 16 emergency-department visits, 23 versus 21 hospitalizations, and 17 cardiovascular hospitalizations in each group. There were no deaths, strokes or transient ischemic attacks in either group during follow-up.
Those numbers are reassuring within the trial. They are not proof that coffee is risk-free for every person with heart disease. With only 200 participants and six months of observation, DECAF did not have enough statistical power to rule out differences in uncommon harms.
AF management also extends far beyond recurrence. Stroke prevention, blood pressure, sleep apnea, alcohol use, weight, physical activity, other heart conditions and medication effects can all matter. A coffee decision does not replace anticoagulation when it is indicated, rhythm-control treatment or follow-up with a clinician.
A careful takeaway for the next cup
The strongest defensible conclusion is narrow and useful. Among adults who already had a coffee habit and underwent successful cardioversion for persistent AF or a closely related rhythm history, continuing roughly one cup of caffeinated coffee a day did not increase clinically detected recurrence. In this trial, recurrence was lower than among participants asked to avoid coffee and caffeine.
That is enough to challenge automatic advice that every patient with AF must give up coffee. It is not enough to tell every patient to begin drinking it, increase the dose or ignore symptoms that appear reliably after a cup.
Someone who notices palpitations, sleep disruption or a rise in blood pressure after caffeine has a different decision from someone who comfortably drinks one morning coffee. The useful conversation is not “Is coffee universally good or bad?” It is whether continuing a modest, familiar habit makes sense for this patient, after this treatment, alongside the rest of their care.
DECAF replaces a simple prohibition with a better kind of uncertainty. For regular coffee drinkers after cardioversion, the mug may not be the enemy clinicians once assumed. The trial gives those patients and their doctors stronger grounds to discuss keeping it, while leaving room for individual triggers and the limits of a small, open-label study.