By Michael H. Crawford, MD, Editor

SYNOPSIS: An analysis demonstrated an inverse association of coffee consumption and cardiac arrhythmias, which was not altered by genetic variations in caffeine metabolism, age, or sex.

SOURCE: Kim EJ, Hoffmann TJ, Nah G, et al. Coffee consumption and incident tachyarrhythmias: Reported behavior, mendelian randomization, and their interactions. JAMA Intern Med 2021; Jul 19. doi: 10.1001/jamainternmed.2021.3616. [Online ahead of print].

There are conflicting data about the potential risks and benefits of coffee consumption. Kim et al studied the association of coffee intake with the risk of tachyarrhythmias in the UK Biobank, a population of more than 500,000 subjects recruited between 2006 and 2010.

Coffee consumption was measured from enrollment questionnaires and grouped into eight categories based on cups consumed per day (< 1 cup, one cup, two cups, three cups, four cups, five cups, and six or more cups). The primary outcome was incident tachyarrhythmias during follow up (2006-2018). Those with previous diagnoses of tachyarrhythmia before enrollment were excluded from the primary analysis. Potential mediators and confounders were obtained from the entry questionnaires or subsequent in- or outpatient diagnoses (ICD codes). Genetic analyses for variants in the major metabolic pathway for caffeine were used to construct a polygenic score wherein a higher score reflected slower caffeine metabolism.

The authors performed a Mendelian randomization analysis on only those who self-identified as white British and were not closely related. Kim et al performed sensitivity analyses for such variables as decaffeinated coffee consumption only, sex, and age. Smoking status was used as a positive control. They excluded those who were pregnant on enrollment, those who dropped out, and those for whom data were missing

UK Biobank participants consumed a median of two cups per day; 22% did not consume coffee. After exclusion criteria were applied, 386,258 subjects were included in the primary analyses. Mean age was 56 years, 52% were women, and almost all were white. During the mean follow-up of 4.5 years, 12,881 had atrial fibrillation (AF), 1,920 had other supraventricular tachyarrhythmias (SVT), 909 had ventricular tachycardias (VT), 97 had premature atrial complexes (PAC), 632 had premature ventricular complexes, and 610 had unspecified arrhythmias. After adjustment for potential mediators and confounders, each one-cup increase in coffee consumption was associated with a 3% lower risk of an incident arrhythmia (HR, 0.97; 95% CI, 0.96-0.98; P < 0.001), which was of a similar magnitude and statistical significance when considering AF and SVT individually. Although point estimates for VT and PAC were similar, they did not achieve statistical significance.

The positive control of smoking showed a higher incidence of arrhythmias (HR, 1.09; 95% CI, 0.03-1.15; P = 0.002). Those with genetic variants associated with slower caffeine metabolism consumed less coffee but did not exhibit any association with arrhythmia risk. There were no differences in these results in the Mendelian randomization sensitivity analyses. Also, the authors did not observe any interactions by age or sex. The authors concluded the habitual consumption of greater amounts of coffee was associated with progressively lower rates of arrhythmias and was not altered by genetic variants in caffeine metabolism.


Since caffeine is a stimulant that increases serum catecholamine levels and promotes wakefulness, it could cause or exacerbate cardiac arrhythmias. I have advised patients with arrhythmias to cut back or eliminate caffeine intake. Coffee is the main source of caffeine for most people, so this meant reducing or eliminating coffee intake or switching to decaffeinated coffee. However, many have touted the antioxidant and anti-inflammatory properties of coffee. Also, recent studies have either not supported an arrhythmogenic role for coffee or actually have shown a decrease in arrhythmias.1 This study supports the latter conclusion. However, observational research cannot explore the reason people drink coffee and the amount they drink. Such decisions may be related to unmeasured confounders, which always makes this type of research problematic compared to a randomized, controlled trial.

There were several strengths of the Kim et al study. It was a large, prospective, community-based study of unprecedented sample size. The results were adjusted for a comprehensive array of confounders. A Mendelian randomization sub-study strongly supported the results by showing a genetic propensity to coffee consumption also was not related to arrhythmias. Also, consideration of the variants in the genes that regulate caffeine metabolism did not support an arrhythmogenic effect of coffee.

In addition to its observational nature, there were other limitations to this work, with the potential for unmeasured confounders. Coffee consumption was self-reported at enrollment; investigators assumed what was reported did not change over the follow-up period. There was no information on the type of coffee (i.e., drip, espresso) or the intake of other caffeinated beverages — except for tea, whose consumption averaged about three cups a day in the study population. Also, the diagnosis of arrhythmias was based on ICD codes, which can be inaccurate. In addition, the follow-up period was relatively short for a group whose mean age was in the mid-50s. AF in particular is more prevalent in older individuals.

The authors believe advising against coffee consumption or reducing its intake to reduce the risk of arrhythmias is not warranted. Considering there is no compelling research to refute this advice, I am inclined to agree with them. However, they excluded patients with known arrhythmias at enrollment, so an exacerbating role of caffeine intake in these patients has not been disproven.


  1. Mostofsky E, Johansen MB, Lundbye-Christesen S, et al. Risk of atrial fibrillation associated with coffee intake: Findings from the Danish Diet, Cancer, and Health study. Eur J Prev Cardiol 2016;23:922-930.