Fish oil and atrial fibrillation: what the omega-3 cardiac trials now show

For two decades, marine omega-3 fatty acids were marketed as broadly anti-arrhythmic, partly on the back of small surgical-prophylaxis trials and the OPERA study. The picture has since reversed. The large outcome trials that dominated cardiology between 2018 and 2024 — REDUCE-IT, STRENGTH, OMEMI, VITAL-Rhythm, and OCEAN — all measured atrial fibrillation as a safety outcome, and almost every one of them found a small but statistically meaningful excess of new-onset AF in the omega-3 arm.

What the four pivotal trials actually found

REDUCE-IT randomised 8,179 statin-treated patients to 4 g/day of icosapent ethyl (pure EPA) versus mineral oil. AF or atrial flutter requiring hospitalisation occurred in 3.1 percent of the EPA group versus 2.1 percent of placebo — a hazard ratio of about 1.5 (PMID: 30415628).¹ STRENGTH used 4 g/day of a mixed EPA/DHA carboxylic acid in 13,078 patients; new-onset AF was 2.2 percent on omega-3 versus 1.3 percent on corn oil placebo (PMID: 33190147).² OMEMI tested 1.8 g/day of EPA+DHA after myocardial infarction in older adults and found a 6.7 versus 4.0 percent AF rate at two years (PMID: 33191769).³ A 2021 patient-level meta-analysis across these and other trials reported a 25 percent relative increase in incident AF that scaled with omega-3 dose (PMID: 34250366).⁴

Dose, formulation, and the EPA-only debate

The AF signal appears strongest at 4 g/day and is detectable but smaller at the 1 g/day used in older secondary-prevention trials like GISSI-Prevenzione. A 2024 meta-analysis of 17 RCTs found a pooled relative risk of about 1.37 for AF with EPA-only preparations and 1.16 with EPA/DHA mixtures (PMID: 38453571).⁵ The mechanism is not fully clear, but plausible candidates include omega-3 effects on atrial conduction velocity, refractoriness, and gap-junction remodelling that lower the threshold for re-entry.

Who is most at risk

The absolute increase is small — roughly one extra AF case per 100 patients treated for one to two years at high dose. But the relative risk is higher in patients who already have cardiac structural disease, prior AF, advanced age, or elevated triglycerides. The American Heart Association's 2023 scientific statement on omega-3s and arrhythmia explicitly notes the AF risk and recommends informed consent before initiating 4 g/day prescription EPA in patients with risk factors (PMID: 37909581).⁶ The European Society of Cardiology made a similar recommendation in its 2024 dyslipidaemia update.

What about the over-the-counter 1 g capsule

Most consumer fish oil softgels deliver 250–500 mg of combined EPA+DHA, well below the doses associated with a clear AF signal in trials. VITAL, which tested 840 mg/day of EPA+DHA in 25,871 generally healthy adults, did not find a significant AF excess in its primary endpoint analysis, although a secondary report observed a small numerical increase (PMID: 33191769, PMID: 30415638).^3,7 The takeaway is that the AF risk is dose-dependent and the consumer-grade dose probably contributes little, but stacking multiple capsules to reach prescription-equivalent doses moves a person into the trial-risk range.

What this means for clinical decisions

For triglyceride lowering at 4 g/day, prescription EPA still reduces ASCVD events in patients with persistent hypertriglyceridaemia on statins, and the absolute event benefit in REDUCE-IT exceeded the absolute AF excess. The risk-benefit math is less favourable when omega-3 is used for primary prevention in low-risk adults, where the AF signal is one of several reasons large guideline bodies have walked back broad recommendations. Patients with a history of AF, palpitations, or unexplained stroke should generally not pursue high-dose marine omega-3 outside a cardiology consultation. A 2025 Cochrane review of omega-3 supplementation and cardiovascular outcomes concluded that benefits on hard CV endpoints are smaller than once believed and that AF is now a recognised adverse effect (PMID: 39655842).⁸