Walk into most cardiac units and the coffee on the tray is decaffeinated. The swap is usually habit rather than written protocol: caffeine nudges blood pressure up and prompts a release of catecholamines, so regular coffee has long been treated as one more thing to take off a heart patient’s tray.
The stakes are not small. The CDC estimates that 12.1 million people in the United States will have atrial fibrillation by 2050, and the condition is associated with about a fivefold increase in the risk of ischemic stroke — roughly one stroke in seven.
A review published on 24 October in Cardiology Research argues that the habit has outlived its rationale. Its authors, Shereif Rezkalla of the Marshfield Clinic Research Institute and Robert Kloner of the Huntington Medical Research Institutes and the University of Southern California, conclude that regular coffee in small to moderate amounts looks safe in cardiac patients — particularly in people who already drink it — and that serving only decaffeinated coffee in hospitals and cardiac rehabilitation units “should be revisited.”
The review is a narrative synthesis, not a meta-analysis. The authors searched PubMed for 2000 through 2026, turned up 135 articles and read the relevant ones critically; nothing was pooled into a single estimate. That matters for how much weight to give its conclusion. But the individual studies it assembles are unusually consistent, and one of them was randomized.
The trial that changed the terms
The DECAF trial, published in JAMA, enrolled 200 current or recent coffee drinkers who had persistent atrial fibrillation, or atrial flutter with a history of AF, and were scheduled for electrical cardioversion at five hospitals in the United States, Canada and Australia. Half were asked to drink at least one cup of caffeinated coffee a day for six months. The other half were asked to abstain from coffee and from all caffeine, decaffeinated coffee included.
Baseline intake was a median of seven cups a week in both groups. During follow-up, the median was seven cups a week in the coffee arm and zero in the abstinence arm. Clinically detected recurrence of atrial fibrillation or flutter came in at 47% in the coffee group and 64% in the abstinence group — a 39% lower hazard of recurrence (HR 0.61, 95% CI 0.42–0.89, P = .01). The two groups did not differ significantly in adverse events.
Where that trial stops
The authors are candid about the boundaries, and they are wide enough to matter. The trial was open-label: nobody could be blinded to whether they were drinking coffee. It was modest in size. There was no protocol-mandated schedule for detecting AF; follow-up was per usual care, with most participants using continuous recording devices, and wearable ECG monitors were used somewhat more often in the coffee group — a difference that would, if anything, have biased the results against coffee.
A chance finding remains possible, the authors write, and baseline characteristics were not perfectly balanced, though sensitivity analyses adjusting for that produced similar results. Only a minority of screened patients agreed to enrol: many did not want to give up coffee, and others believed coffee worsened their AF, so people who genuinely react to caffeine may be underrepresented. Paroxysmal AF was not studied at all, and adherence in the abstinence group was imperfect, with 69% reporting no coffee.
A 495,000-person cohort lands on the same side
The second recent dataset is observational and far larger. Published on 31 October in Food & Nutrition Research, it followed 495,269 UK Biobank participants who had no atrial fibrillation at baseline, with a mean age of 57, for a median of about 13 years. During that time, 37,261 developed AF.
Compared with non-coffee drinkers, the adjusted hazard ratios were 0.93 (95% CI 0.90–0.96) for half a cup to one cup a day, 0.89 (0.86–0.92) for two to three cups, and 0.91 (0.89–0.94) for four or more. The relationship was non-linear, with the lowest risk in the middle of the range rather than at the top. Tea was a much weaker signal: a 5% lower risk only at two to three cups a day (HR 0.95, 0.92–0.98), and no significant association at the other intake levels.
The authors are careful with their own result. Coffee intake was asked once, at baseline, so a decade of changed habits is invisible. Cup size, brewing method and strength were not measured. The coffee question included decaffeinated coffee and caffeine was never quantified, so the exposure is the beverage, not caffeine. The cohort is 94.4% white and was recruited between the ages of 40 and 69. Residual confounding cannot be excluded. And the finding is an association, not a cause.
They are more cautious still about tea, noting that Mendelian randomization studies have not supported a causal effect of tea on cardiovascular disease — a reminder that a cohort’s inverse association and a causal claim are different objects.
Why the older caution existed
There is a plausible mechanism in both directions, which is why the question stayed open for so long. Caffeine blocks adenosine receptors, and the review points to that blockade prolonging the atrial refractory period — an environment that would make AF harder to sustain. Set against it is the older observation that caffeine raises blood pressure acutely and releases norepinephrine, most noticeably in people who do not drink coffee regularly.
That second effect fades with habit. The review notes that in regular coffee drinkers there is complete tolerance to the catecholamine rise, so that within a few days coffee no longer produces a rise in adrenaline or noradrenaline. It also notes that older ACC/AHA guidance for ST-elevation myocardial infarction allowed patients who were regular coffee drinkers to keep drinking coffee, limited to one cup a day.
The distinction the review insists on is between coffee and concentrated caffeine products. Energy drinks, which the FDA puts at 54 to 328 milligrams of caffeine per 16 fluid ounces, are treated separately: the review says they “may indeed be harmful and should be avoided.”
What the cup actually contains
For most adults, the FDA has cited 400 milligrams of caffeine a day — roughly two to three 12-fluid-ounce cups of coffee — as an amount not generally associated with negative effects. The agency is explicit that sensitivity and the speed of elimination vary widely between people, and that around 1,200 milligrams consumed rapidly can produce toxic effects such as seizures.
Decaffeinated coffee is not caffeine-free. The FDA gives a typical range of 2 to 15 milligrams per 8-fluid-ounce cup, which is one reason the review’s point about decaf is awkward: the substitute that cardiac units default to has not itself been studied for safety in cardiac patients.
Consumer guidance has not moved as fast as the review. MedlinePlus, the National Library of Medicine’s encyclopedia, still lists “problems with fast or irregular heart rhythms” among the reasons a person may want to limit caffeine — a list that also includes stress or anxiety, sleep problems, and acid reflux or stomach ulcers. That entry also carries the long-standing AMA Council on Scientific Affairs position that moderate tea or coffee drinking is not likely to be harmful “as long as you have other good health habits.” Both ideas are in circulation, which is roughly why the hospital practice persists.
What the numbers do not settle
Three gaps are worth naming plainly.
The first is acute coronary syndrome. The review says there are not enough data for patients with ST-elevation or non-ST-elevation myocardial infarction, and that until there are, those patients should avoid caffeinated coffee. Its reassurance is aimed at stable cardiac patients, not at the coronary care unit in the first days after an event.
The second is hard outcomes. The DECAF trial’s endpoint was recurrence of an arrhythmia detected clinically over six months, not stroke, hospitalisation or death. No trial has been powered to answer whether coffee changes those.
The third is individual response. Population averages say nothing about the person who reliably gets palpitations after a cup. The review’s own wording concedes this: patients who are not habitual drinkers, or who have a history of palpitations after coffee, may consider avoiding regular coffee.
What we did not find
No randomized trial of coffee in paroxysmal atrial fibrillation. No trial with stroke or mortality endpoints. No study of decaffeinated coffee in cardiac patients, despite decades of hospitals serving it by default. And no basis for the stronger claim that coffee protects the heart — the cohort differences here are 7% to 11% in relative terms, from self-reported intake, and the randomised evidence covers one narrow group of patients over six months.
What the two papers do support is narrower and more useful: the fear that a cup of coffee will trigger atrial fibrillation is not well supported in people who already drink coffee, and the decaf default in cardiac units is a practice waiting for evidence rather than resting on it.
If you have atrial fibrillation, palpitations or a recent cardiac event, how much caffeine you drink is a conversation to have with your own clinician — the population data above cannot answer it for you. Our medical disclaimer explains the limits of what we publish. For the wider picture on how cold months shift cardiac risk, see our piece on why heart attacks cluster in winter; for caffeine timing rather than quantity, the AASM’s shift-work guideline covers the night-shift case.