You are at a conference. You had 1 glass of wine too many last night, so you had to add an extra coffee to get moving this morning. Then you feel a slight flutter in your chest. Caffeine is a stimulant, right? Could it be causing atrial fibrillation?
The paper
Wong CX, Cheung CC, Montenegro G, Oo HH, Peña IJ, Tang JJ, Tu SJ, Wall G, Dewland TA, Moss JD, Gerstenfeld EP, Tseng ZH, Hsia HH, Lee RJ, Olgin JE, Vedantham V, Scheinman MM, Lee C, Sanders P, Marcus GM. Caffeinated Coffee Consumption or Abstinence to Reduce Atrial Fibrillation: The DECAF Randomized Clinical Trial. JAMA. 2025 Nov 9:e2521056. doi: 10.1001/jama.2025.21056. Epub ahead of print. PMID: 41206802 NCT05121519
The question
Does abstinence from coffee reduce atrial fibrillation recurrence rates?
The methods
The DECAF (Does Eliminating Coffee Avoid Fibrillation?) trial is a multicenter RCT from 5 hospitals in the US, Australia, and Canada.
Credit where credit is due: they came up with an excellent study acronym.
Patients
Adult patients with sustained atrial fibrillation, planned electrical cardioversion, and coffee consumption of 1 cup per day or greater sometime in the past 5 years, willingness and ability to adhere to coffee abstinence or continuation, and life expectancy of at least 6 months. They included patients who were not currently drinking coffee, which seems like a potential mistake. Randomization occurred after successful cardioversion.
Intervention
Caffeinated coffee consumption: patients were encouraged to drink at least 1 cup of caffeinated coffee (or at least 1 espresso shot) and other caffeine-containing products every day as per their usual lifestyle. It was recommended that patients in the coffee consumption group not intentionally increase or decrease consumption of coffee or other caffeine-containing products.
Comparison
Abstinence from coffee and caffeine: patients were encouraged to completely abstain from coffee, including decaffeinated coffee, and other caffeine containing products.
Outcome
The primary outcome was “clinically detected recurrence of AF or atrial flutter lasting 30 seconds or longer assessed in a time-to-event analysis.”
The results
Of 1965 adults screened, they ended up randomizing 200. Overall coffee consumption is low in both groups (discussed below), and goes down in both groups during the trial period, but there is a statistical difference of 7 cups per week between the groups.
In the 6 month follow-up period, atrial fibrillation recurrence was 47% in the caffeine group and 64% in the abstinence group. In their time to recurrence analysis, caffeine intake resulted in a statistically improved time to recurrence (HR, 0.61 95% CI 0.42-0.89 P =0.01).
My thoughts
Given that caffeine is my primary drug of choice, my bias is just to accept these results and move on. It would be lovely if coffee was providing me with benefits aside from managing my sleep deprivation. I love that they are asking this question, but there are a number of issues with this trial that should significantly decrease your confidence.
Generalizability is going to be a major issue. The majority of patients approached simply said no to participating in this trial, mostly because they wouldn’t change their coffee habits. There were more than 3 times as many patients who declined to participate than who actually participated. Chances are, these results won’t apply to your patient.
I understand the choice of using “willingness to adhere” as an inclusion criteria, but it really impacts the generalizability issue. Patients willing to stop drinking coffee are probably quite different from those who are unwilling to change their habits. Perhaps the patients who are incapable or unwilling to stop caffeine consumption are exactly those who are consuming enough caffeine to cause recurrent atrial fibrillation. The “at least 1 cup per day” seems like an almost trivial amount of coffee to me. I wouldn’t expect atrial fibrillation to be affected by such a small intake, but look at the caffeine intake of your average emergency doctor, and the story might be different. They also excluded a large number of people who apparently drink coffee, but couldn’t or were unwilling to drink 1 cup a day, which again is a very strange population to me. Who counts as a regular coffee drinker, but isn’t having a daily coffee?
As I mentioned above, they counted coffee consumption going back 5 years, which meant that they included patients as “coffee drinkers” who were not actually drinking any coffee. That seems like a mistake to me. One sure fire way to get bad data out of this trial is to have a bunch of people in the coffee group not drinking coffee. Considering my personal mean coffee consumption, I thought this might be more of a theoretical concern, but according to table 1, a full 46% of the coffee consumption group was drinking absolutely no coffee at enrollment, and another 29% were drinking 3 or less cups a week! Moving from 0 coffee to “abstinence” is not so much of a change.
The manuscript lacks some important details about the patients’ ongoing cardiac care that I would really like to see. At enrollment, 70% of these patients were on a beta-blocker and 50% were on an antiarrhythmic. All of these patients were cardioverted at enrollment, but they don’t tell us if these medications were continued or not. The use of these medications seems like it would influence risk of atrial fibrillation recurrence, and is therefore pretty important information if you are going to try to counsel your patients using this data.
There are also problems that arise from the fact that caffeine consumption is based on patient self-report. As lovely as all my patients are, they are not always completely honest in research settings. I could imagine that spikes in caffeine use, or irregular caffeine use, might result in a higher risk for arrhythmias than daily use. The data on coffee consumption is entirely self-reported. The coffee consumption group had little reason to lie (although if they were asked to consume daily, they might increase their numbers to make the researchers happy). However, I think the abstinence group is more likely to lie about their consumption. You can imagine a patient who had a completely sleepless night and just needed a couple cups of coffee to get them through the day. If they go into atrial fibrillation that day, they might be embarrassed about ‘ruining’ the study, and opt not to tell the researchers about the coffee consumption.
I think it is a small issue, but I will note there is a change in their primary outcome from the registry. They changed from looking at 2 primary outcomes – recurrent AF and time to AF – to “confirmed clinical recurrence of AF or atrial flutter, or device-detected recurrence lasting longer than 30 seconds, analyzed as a time-to-event outcome”. Any adjustment raises concerns about research degrees of freedom after data has been collected. The change to include devices might be a simple clarification, as these devices became rampant, but there is some debate about the clinical importance of asymptomatic runs of atrial fibrillation caught only by device. For example, in a paper I had already included in the next Research Roundup, Svendsen 2021 found no benefit in screening elderly patients with a loop recorder for atrial fibrillation, despite finding and treating a lot of atrial fibrillation. Therefore, this change might have resulted in a primary outcome that is actually less clinically important.
In fact, it is pretty clear to me that the outcome here is not clinically significant. There were exactly 5 ED visits in both groups for atrial fibrillation. On other words, almost their entire primary outcome is made up of clinically irrelevant atrial fibrillation events caught on smart watches.
When you combine concerns about self-reporting, the fact that abstinence was not perfect, and coffee intake in the caffeine group was quite low, my sense is that we are comparing 2 very similar groups. I would expect most biases to push the trial towards a null finding, which makes the statistical reduction in recurrence interesting. However, considering that the theory when starting this trial was that caffeine would cause recurrence, not protect against it, the pretest probability for this finding has to be low, and therefore it is unlikely that this finding is true. Conversely, I think this does provide some degree of evidence of the opposite: that caffeine is unlikely to be harmful (when consumed in the very low doses discussed here).
I don’t think we should be encouraging caffeine intake based on these results, and I am not sure we can even conclude that it is perfectly safe. However, there is a bunch of prior data on this topic, and the results seem to be consistent. The CRAVE study was a prospective, randomized, case-crossover trial which showed no difference in the rate of PACs, although PVCs did increase with coffee intake. (Marcus 2023) PACs are known to be a risk factor for developing atrial fibrillation. (Dewland 2013) Similarly, there are numerous observational trials that did not find an association between caffeine and atrial fibrillation risk. (Kim 2021; Chieng 2022; Larson 2015
Therefore, I think it is reasonable to counsel patients that coffee and caffeine consumption are not going to put them at a higher risk of atrial fibrillation recurrence, and that they should therefore live their lives as they please.
Bottom line
This RCT showed a reduced risk of atrial fibrillation recurrence with daily coffee consumption as compared to abstinence. There are numerous reasons to think these results are not true, but this probably is evidence that caffeine is not going to cause increased atrial fibrillation.
Other FOAMed
Evidence based medicine is easy
Evidence based medicine resources
References
Chieng D, Canovas R, Segan L, Sugumar H, Voskoboinik A, Prabhu S, Ling LH, Lee G, Morton JB, Kaye DM, Kalman JM, Kistler PM. The impact of coffee subtypes on incident cardiovascular disease, arrhythmias, and mortality: long-term outcomes from the UK Biobank. Eur J Prev Cardiol. 2022 Dec 7;29(17):2240-2249. doi: 10.1093/eurjpc/zwac189. PMID: 36162818
Dewland TA, Vittinghoff E, Mandyam MC, Heckbert SR, Siscovick DS, Stein PK, Psaty BM, Sotoodehnia N, Gottdiener JS, Marcus GM. Atrial ectopy as a predictor of incident atrial fibrillation: a cohort study. Ann Intern Med. 2013 Dec 3;159(11):721-8. doi: 10.7326/0003-4819-159-11-201312030-00004. PMID: 24297188
Larsson SC, Drca N, Jensen-Urstad M, Wolk A. Coffee consumption is not associated with increased risk of atrial fibrillation: results from two prospective cohorts and a meta-analysis. BMC Med. 2015 Sep 23;13:207. doi: 10.1186/s12916-015-0447-8. PMID: 26394673
Kim EJ, Hoffmann TJ, Nah G, Vittinghoff E, Delling F, Marcus GM. Coffee Consumption and Incident Tachyarrhythmias: Reported Behavior, Mendelian Randomization, and Their Interactions. JAMA Intern Med. 2021 Sep 1;181(9):1185-1193. doi: 10.1001/jamainternmed.2021.3616. Erratum in: JAMA Intern Med. 2023 Apr 1;183(4):394. doi: 10.1001/jamainternmed.2022.6962. PMID: 34279564
Marcus GM, Rosenthal DG, Nah G, Vittinghoff E, Fang C, Ogomori K, Joyce S, Yilmaz D, Yang V, Kessedjian T, Wilson E, Yang M, Chang K, Wall G, Olgin JE. Acute Effects of Coffee Consumption on Health among Ambulatory Adults. N Engl J Med. 2023 Mar 23;388(12):1092-1100. doi: 10.1056/NEJMoa2204737. PMID: 36947466
Svendsen JH, Diederichsen SZ, Højberg S, Krieger DW, Graff C, Kronborg C, Olesen MS, Nielsen JB, Holst AG, Brandes A, Haugan KJ, Køber L. Implantable loop recorder detection of atrial fibrillation to prevent stroke (The LOOP Study): a randomised controlled trial. Lancet. 2021 Oct 23;398(10310):1507-1516. doi: 10.1016/S0140-6736(21)01698-6. Epub 2021 Aug 29. Erratum in: Lancet. 2021 Oct 23;398(10310):1486. doi: 10.1016/S0140-6736(21)02090-0. PMID: 34469766
Wong CX, Cheung CC, Montenegro G, Oo HH, Peña IJ, Tang JJ, Tu SJ, Wall G, Dewland TA, Moss JD, Gerstenfeld EP, Tseng ZH, Hsia HH, Lee RJ, Olgin JE, Vedantham V, Scheinman MM, Lee C, Sanders P, Marcus GM. Caffeinated Coffee Consumption or Abstinence to Reduce Atrial Fibrillation: The DECAF Randomized Clinical Trial. JAMA. 2025 Nov 9:e2521056. doi: 10.1001/jama.2025.21056. Epub ahead of print. PMID: 41206802
