Effects of Omega-3 Fatty Acid Treatment on Risk for Atrial Fibrillation: An Updated Meta-Analysis of 35 Trials including 114 592 Individuals

Abstract

BACKGROUND:

Recent meta-analyses of randomized controlled trials have raised concerns that treatment with omega-3 fatty acids may increase the risk of atrial fibrillation (AF). However, these meta-analyses included at most 8 trials. The aim of this current meta-analysis was to expand the search by including other eligible omega-3 randomized controlled trials with AF incidence data, incorporating both published and unpublished data.

METHODS:

Eligible studies were randomized controlled trials investigating daily doses of ≥500 mg/d of docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA). Additional inclusion criteria included ≥12 months of treatment with EPA/DHA, participants ≥50 years of age, and, where possible, the absence of known AF/atrial flutter at baseline. The primary outcome was the occurrence of new-onset AF. Our primary hypothesis was that risk for AF would simultaneously depend on both omega-3 dose (above or below 1500 mg/d) and background cardiovascular disease risk status, and that their combined impact on AF risk would be synergistic.

RESULTS:

A total of 35 randomized controlled trials (37 data sets; n=114 592) were included in this meta-analysis. Only studies including patients at high-risk for cardiovascular disease who were treated with high-doses of EPA/DHA (>1500 mg/d) showed a statistically significant increase in AF risk with a pooled odds ratio (OR) of 1.43 (95% CI, 1.14–1.79) and an absolute risk difference of 0.8% (0.40%–1.1%). None of the other 3 groups showed statistically significant levels of AF risk (odds ratios, 1.07 [high risk–low dose], 1.06 [low risk–low dose], and 1.03 [low risk–high dose]).

CONCLUSIONS:

This meta-analysis suggests that high-dose EPA/DHA treatment is associated with an increased risk of AF in patients at high cardiovascular disease risk, whereas low-dose EPA/DHA does not appear to increase AF risk, even in high-risk populations. Further prospective studies are needed to evaluate any potential increased risk of higher doses balanced against potential benefits.

What is Known?

• Previous meta-analyses of omega-3 fatty acid randomized controlled trials have reported a dose-response relationship with risk for atrial fibrillation (AF), but these have included no more than 8 trials.

• Emerging evidence has challenged these findings, with observational biomarker studies suggesting higher circulating marine omega-3 fatty acid levels may be associated with lower AF risk (particularly at low-to-intermediate doses), while concerns have also been raised regarding inconsistent AF ascertainment and selective outcome reporting in prior analyses.

What the Study Adds

• This meta-analysis incorporated both published and underutilized data sources, including adverse event records, trial registry data, and investigator-provided unpublished AF data, thereby substantially expanding the evidence base on omega-3 fatty acids and AF risk.

• Stratification by omega-3 dose and baseline cardiovascular disease risk provided a more clinically relevant understanding of how lower-dose omega-3 supplementation (including doses commonly achievable through diet or standard nutritional supplementation) may affect AF risk across different patient populations, as distinct from the effects observed with high-dose pharmaceutical omega-3 products.

• Increased AF risk was observed primarily in high-risk cardiovascular disease populations receiving high-dose omega-3s (>1500 mg/d eicosapentaenoic acid+docosahexaenoic acid), whereas lower doses (≤1500 mg/d) were not associated with increased AF risk, even in high-risk populations.

Recent meta-analyses of randomized controlled trials (RCTs) have raised concerns that omega-3 fatty acid (FA) supplementation may increase the risk of incident atrial fibrillation (AF), potentially in a dose-dependent manner. Gencer et al¹ reviewed 7 RCTs focused on cardiovascular outcomes and found that the risk for AF was increased by 25% with omega-3 treatment. Bae et al² reported similar associations, and Jia et al³ found a lower risk for AF among the 5 low-dose omega-3 FA studies compared with the risk in 3 higher risk studies.

Emerging evidence, however, has called these findings into question. For example, a 2023 biomarker-based meta-analysis of 17 prospective cohorts found that higher circulating levels of marine omega-3 FAs were inversely associated with incident AF, suggesting a potential protective impact of low to intermediate doses known to produce such blood levels.⁴ Furthermore, methodological concerns have been raised regarding prior interventional analyses, including inconsistent reporting of AF outcomes, potential informative censoring due to longer survival in omega-3 treated groups compared with placebo groups,⁵ and discrepancies in AF case ascertainment.⁶ These factors may have contributed to an overestimation of AF risk in prior analyses.

According to PubMed (August 16, 2025), over the last 40 years, there have been over 3900 published RCTs involving omega-3 FAs, there is a significant opportunity to reexamine the omega-3/AF issue using a broader evidence base by including supplementation studies where AF was not a prespecified outcome but would have been recorded as an adverse event had it developed. The goal of this analysis was to conduct a more comprehensive review of the literature and to perform an updated meta-analysis of RCTs to better understand the relationship between omega-3 supplementation and AF risk.

In contrast to earlier meta-analyses,^1–3 we integrated both published and underutilized data sources, including trial registries and unpublished data sets, to reduce selective reporting bias. We further ensured that trials reporting no AF events contributed to the pooled evidence base. Finally, we applied a cardiovascular disease (CVD) risk/dose stratification framework to distinguish between doses largely achievable through diet or typical supplementation versus approved doses of pharmacological formulations.

Methods

The protocol for this study is registered at the International Prospective Register of Systematic Reviews (PROSPERO, CRD420251233813). The conduct and reporting of this systematic review and meta-analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines.⁷

This meta-analysis included published data and unpublished summary data provided by individual study investigators. Restrictions apply to the availability of unpublished data, which were obtained with permission for the purposes of this study and are not publicly available. Studies contributing unpublished data are identified in Table S1. All analytic methods are described within the Methods section, with additional procedural details provided in the Supplemental Methods. Requests for access to underlying data should be directed to the original study investigators or relevant data custodians.

Institutional review board approval and written informed consent were obtained in the original studies, as reported by the respective investigators.

Eligibility Criteria

Eligible studies were parallel-group RCTs investigating omega-3 FA supplementation containing eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), administered in capsule or liquid oil form, at a dosage meeting or exceeding ≥500 mg/d of EPA+DHA, consistent with international intake recommendations.^8–10 This followed the precedent of all prior meta-analyses, which only included studies where EPA+DHA intakes exceeded this threshold. There was no upper limit on dosage. Trials were required to have a minimum treatment duration of 12 months and to enroll participants aged ≥50 years, reflecting the age at which AF risk begins to increase.¹¹ There were no restrictions on EPA:DHA ratio or chemical form (eg, ethyl esters, triglycerides, phospholipids). Multicomponent intervention studies were eligible if data from an appropriate nonomega-3 control group were available. We excluded trials that: administered omega-3s solely through dietary sources or food-based delivery formats; relied on dietary or lifestyle counseling as the sole intervention; used omega-3s in the comparator arm; or lacked a placebo or control group. Studies were also excluded if intervention and control groups were not treated in parallel for at least 12 months. Studies were included regardless of participants’ baseline health status. All individuals with known AF (paroxysmal, persistent, or permanent) at baseline were excluded. For studies enrolling participants both with and without prior AF, data from participants with documented baseline AF were excluded where possible. Additional eligibility details are provided in the Supplemental Methods.

Search Strategy

A systematic search was conducted across multiple bibliographic databases, clinical trial registries, and data repositories, including PubMed, ClinicalTrials.gov, International Standard Randomised Controlled Trial Number Registry, Clinical Drug Experience Knowledgebase, medRxiv, and other relevant sources. Searches were conducted through November 2025 and were updated approximately monthly. Reference lists of prior systematic reviews and key primary studies were also screened to ensure comprehensive coverage.

Titles and abstracts were screened by 1 reviewer (N.A.), with all exclusions independently checked by a second reviewer (W.H.). Full-text eligibility was assessed independently by both reviewers, with disagreements resolved through discussion or consultation with a third reviewer (F.Q.). Screening was performed in an unblinded manner. Full search terms, registry-specific strategies, and targeted searches for unpublished or registry-only trials are detailed in the Supplemental Methods.

Outcome

The primary outcome was the occurrence of new-onset AF/flutter as determined through central adjudication, ECG, electronic medical records, insurance claims, and patient-reported adverse events.

Data Collection and Extraction

One reviewer (N.A.) extracted all relevant data, which were independently verified by a second reviewer (W.H.). Extracted variables included trial name, registration number, publication year, study design, comparator, omega-3 dose and formulation, supplementation duration, sample size, participant characteristics, target population, and outcomes of interest. Definitions of end points were not standardized across studies. Trial quality was assessed using the modified Jadad scale (0–8), with scores ≥5 indicating high quality.

Published AF data were defined as new-onset AF events reported in peer-reviewed publications, Supplemental Materials, publicly accessible safety reports, clinical trial registries, or other secondary sources. Adverse and serious adverse events, reported using standardized adverse event classification systems,^12–16 were systematically reviewed to identify AF events. Broader event categories (eg, cardiac events or palpitations) were not classified as AF unless clarified by principal investigators (PIs).

For studies that did not prespecify AF and did not report AF-related events, zero events were not assumed unless outcome ascertainment procedures and investigator confirmation justified this classification. Detailed procedures for adverse event review, AF ascertainment, zero-event classification, and investigator verification are described in the Supplemental Methods.

Acquiring Unpublished Data

To minimize reporting bias, we proactively contacted PIs of eligible trials to determine whether AF outcomes had been collected or could be verified. For the purposes of this review, unpublished data were defined as AF/flutter outcome data not reported in peer-reviewed publications or publicly accessible sources, irrespective of whether the primary trial results had already been published. This included investigator-provided data, nonpublic data sets obtained through formal access requests, and data from completed registered trials for which primary results were not yet published at the time of the meta-analysis. Studies were included only when AF outcome status could be confirmed or when monitoring procedures justified the absence of reported events.

A maximum of 3 follow-up attempts were made to contact coauthors in cases of nonresponding lead authors over a period of ≈6 months. All direct investigator outreach efforts were concluded in December 2025. All responses from PIs, including confirmations, clarifications, and statements of unavailable or unreliable AF data, are documented in Tables S1 and S2. Expanded descriptions of investigator contact procedures, follow-up strategy, and interpretation of unavailable AF data are provided in the Supplemental Methods.

Data Synthesis

Since CVD risk status and the EPA/DHA dose may each plausibly influence the development of AF, and this influence may be synergistic, we prespecified analyses that simultaneously stratified by both variables to explore their potential modifying effects. Trials were classified into 4 strata: high risk-high dose (HR-HD), high risk-low dose (HR-LD), low risk-high dose (LR-HD), and low risk-low dose (LR-LD).

CVD risk status was defined based on baseline atherosclerotic cardiovascular disease or the presence of multiple CVD risk factors. Low CVD risk was defined as the absence of diagnosed atherosclerotic cardiovascular disease, including trials enrolling participants for non-cardiovascular indications or general wellness.

Omega-3 dose was categorized as ≤1500 mg/d or >1500 mg/d to distinguish intakes achievable through diet or dietary supplements from those requiring pharmacological formulations. This threshold was selected based on several factors: it is below the lowest approved daily dose of pharmaceutical-grade omega-3 products (ie, Vascepa 1800 mg/d) and it has been shown to produce upper quintile levels of the omega-3 index (ie, erythrocyte EPA+DHA, a validated biomarker of omega-3 intake^17,18) which have been associated with the lowest risk of AF,^4,19 stroke,²⁰ and all-cause mortality²¹ in large long-term epidemiological studies. Further details on the rationale used to distinguish EPA/DHA intakes from dietary supplements versus pharmacological agents can be found in Supplemental Appendix 1.

AF event counts and denominators were extracted separately for intervention and control arms. Where multiple omega-3 dose arms were present, arms were pooled or stratified as appropriate to avoid double-counting comparators. Trials with unresolved AF outcome status after investigator contact were excluded from the meta-analysis (Table S2). Additional details on trial stratification, CVD risk classification, and arm pooling procedures are provided in the Supplemental Methods.

Data Analysis

Absolute risk for AF was computed as the number of reported AF events divided by the number of subjects in the treatment group. The relative risk for AF was computed by dividing the absolute risk in the omega-3 group by that of the control group.

Unadjusted logistic regression models predicting AF status (yes/no) by treatment group (omega-3/control), with Firth correction (brglm2 package, version 1.0.1) to adjust for small sample sizes, were fit on each cohort when predicting odds ratios (ORs). Cohort-specific, unadjusted ORs were then pooled by inverse-variance weighted, random effects meta-analysis (restricted maximum likelihood) using the metafor package (v. 4.4.0) in R (v. 4.3.2). As a sensitivity analysis, we also conducted meta-analysis on the absolute risk difference (Figure S1) using the meta (v. 8.1.0) package in R. A second sensitivity analysis excluded studies which did not assess prevalent AF status at baseline (Figure S2). In all cases, heterogeneity was assessed by I² and Cochran Q statistics and further explored by meta-analyzing prespecified subgroups as created by classifying cohorts based on participants baseline CVD risk (high versus low), and the omega-3 dose (≤1500 mg/d or >1500 mg/d). A significance level of 0.05 was used for all analyses.

Results

Results of the Search

Figure 1 presents the study selection process and the acquisition of unpublished data, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines.⁷ The electronic search yielded 3688 records, of which 1962 were duplicates and removed. Title and abstract screening of the remaining 1726 records led to the exclusion of 1646 studies. The primary reasons for exclusion were trial duration <12 months and mean participant age <50 years (n=1273). Additional reasons for exclusion are shown in Figure 1.

Figure 1.

PRISMA 2020 flow diagram of study selection and acquisition of unpublished data. Records were identified through bibliographic databases, clinical trial registries, and additional sources through November 2025. After duplicate removal, records underwent title/abstract screening and full-text assessment according to predefined eligibility criteria. The diagram also summarizes investigator contact procedures and acquisition of unpublished atrial fibrillation (AF) outcome data from principal investigators (PIs), registry sources, and shared data sets. Detailed investigator responses and study-level data availability are provided in Tables S1 and S2. ANZCTR indicates Australian New Zealand Clinical Trials Registry; CDEK, indicates Clinical Drug Experience Knowledgebase; GOED, Global Organization of Omega-3 EPA and DHA; ISRCTN, International Standard Randomised Controlled Trial Number Registry; NIH DASH, National Institutes of Health Data and Specimen Hub; NIHR, National Institute for Health and Care Research; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; and SAE, serious adverse event. Adapted from the PRISMA 2020 statement.⁷

After full-text assessment of the remaining 80 RCTs that met the inclusion criteria, 20 had available published data on incident AF, while 60 were identified as studies potentially containing unpublished data. Formal data access requests were submitted where applicable (n=1), or PIs were contacted directly (n=58). One data set was obtained via internal collaboration from a coauthor (F.Q.) who had previously received the data set.

Among the 58 PIs contacted, 33 responded. Of these, 13 provided usable AF data, 8 confirmed that AF data was not collected in their study,^22–30 2 confirmed relevant data collected but not shareable- either because the data is pending first presentation³¹ (n=1), or the article is under review³² (n=1). Ten investigators could not confirm the status of AF data. This was due to the inability to access or review raw data sets or uncertainty regarding whether AF events were captured within broader AE categories^33–37 (n=5), ongoing trials with no available safety data³⁸ (n=1), or responses indicating they were unable to assist without clarifying whether AF data existed³⁹ (n=1). In addition, some investigators indicated that institutional approvals would be required to determine data availability, but did not confirm whether such data had been collected or respond to further correspondence⁴⁰ (n=1), while others suggested that relevant data (eg, ECG abnormalities or AE data sets) could be reviewed, but did not provide confirmation despite follow-up requests^41,42 (n=2). 25 PIs did not respond after 3 follow-up attempts.^43–67 Detailed study-level characteristics, investigator responses, and reasons for exclusion of eligible RCTs not included in the meta-analysis are provided in Table S2.

In total, 35 RCTs were included in the meta-analysis. This included 20 RCTs with published AF data^68–87 and 15 with unpublished AF data^88–102 (13 obtained directly from PIs,^{88–93,95–100,102} 1 via a granted database request,¹⁰³ and 1 through internal collaboration¹⁰⁴). Published data sources included original research articles^{69–73,77,81,82} (n=8), research letter⁶ (n=1), systematic review¹⁰⁵ (n=3), supplemental tables or figures^{76,78,84,85,87} (n=5), ClinicalTrials.gov^106,107(n=2), and Clinical Drug Experience Knowledgebase¹⁰⁸ (n=1) adverse event records. For 2 studies, additional publicly accessible sources^109,110 were used to clarify or supplement AF outcome reporting from the original reports.^6,69 Detailed study-level information on publication status, AF data sources, and how AF outcome data were identified or obtained for each trial is provided in Table S1.

Characteristics of Trials

Among the 35 RCTs there were 37 data sets comprising 114 592 participants (Table). In the pooled data set, the mean baseline age was 66 years (range, 53–78) with 61% men. Daily omega-3 doses ranged from 600 to 4000 mg/d, and EPA/DHA was most commonly administered in either an ethyl ester^{69,71,73–80,85,87–92,95,97,101} (57%) or triglyceride-based formulation^{72,82–84,86,93,94,96,98–100,102} (34%). The duration of interventions ranged from 12 to 88.8 months. Most studies employed a double-blind, placebo-controlled design. Control groups most commonly received oil-based placebos. A subset of trials adopted factorial or multi-factorial designs (45%), combining omega-3 supplementation with other active interventions. These included pharmacological therapies,^{71,73,76,79,82,83,85,92} nutritional supplements,^{80,86,87,91,100–102} or structured lifestyle interventions.^74,99,102 Trials were conducted across 46 countries and included 5 multinational studies.^{69,70,92,100,102} Detailed study characteristics are presented in Table S3.

Table.

Summary of 35 RCTs (37 Data Sets) Including Data From 114 592 Subjects and the Observed AF Rates* Segregated by High Versus Low Dose EPA+DHA (≤ Versus >1500 mg/d) and by High Versus Low Risk for CVD

Rates of New-Onset AF by Dose/Risk Group

After stratifying the studies by dose and CVD risk status (Table), there were 12 studies in the HR-HD stratum^{68–76,88–90} including 25 603 patients given an average of 2693 mg EPA+DHA/d. The 6 HR-LD studies^{77–80,91,92} included 53 676 patients with an average dose of 823 mg/d. The LR-HD stratum^{81–85,93–99} included 12 studies with 3264 subjects given a mean dose of 2439 mg/d, and the LR-LD group^{86,87,95,96,100–102} included 7 studies, 32 049 individuals with a mean dose of 1029 mg/d. A summary of all included trials, including AF event counts and stratified rates, is presented in the Table. Detailed study-level information on AF ascertainment and outcome definitions is provided in Table S1.

Meta-Analysis

Looking at each of the 4 groups separately, only the HR-HD group^{68–76,88–90} (ie, individuals at elevated cardiovascular risk receiving ≥1500 mg/d of EPA/DHA) showed a statistically significant increase in AF risk with a pooled OR of 1.43 (Figure 2). The ORs for the other 3 categories were not statistically significant (1.07 for HR-LD^{77–80,91,92}; 1.03 for LR-HD^{81–85,93–99}; and 1.06 for LR-LD^{86,87,95,96,100–102}). We also note that only 2 studies^70,76 (both in the HR-HD group) found a statistically significant increase in risk of AF. In all 4 strata, residual heterogeneity estimates were <16% and not statistically significant (all 4; P>0.41). Across all studies, risk for AF was significantly increased by 14%.

Figure 2.

Meta-analysis of incident atrial fibrillation (AF) according to omega-3 fatty acid dose and cardiovascular disease (CVD) risk status across 35 randomized controlled trials (RCTs). Forest plot demonstrates pooled odds ratios (ORs) for incident AF from 35 RCTs (37 data sets), including 114 592 participants. Cohort-specific unadjusted ORs were estimated using logistic regression with Firth correction and pooled using inverse-variance weighted random-effects meta-analysis with restricted maximum likelihood (REML) estimation. Studies were stratified according to baseline CVD risk (high vs low) and omega-3 fatty acid dose (high dose >1500 mg/d eicosapentaenoic acid [EPA] + docosahexaenoic acid [DHA] vs low dose ≤1500 mg/d EPA+DHA). Squares represent individual study effect estimates, with square size proportional to study weighting; horizontal lines indicate 95% CIs; diamonds represent subgroup and overall pooled estimates. The vertical dashed line denotes the null effect (OR=1). Statistical heterogeneity was assessed using the Cochran Q and I² statistics. ADCS-DHA indicates Alzheimer’s Disease Cooperative Study–Docosahexaenoic Acid; AREDS2, Age-Related Eye Disease Study 2; ASCEND, A Study of Cardiovascular Events in Diabetes; GISSI-Prevenzione, Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico–Prevenzione; BRAVE-EPA, Brain Amyloid and Vascular Effects of Eicosapentaenoic Acid; CANN, Cognitive Ageing Nutrition and Neurogenesis Trial; CAPFISH-3, CAncer Prostate FISH Oil Trial Phase 3; DO-HEALTH, Vitamin D3–Omega-3–Home Exercise–Healthy Ageing and Longevity Trial; DREAM, Dry Eye Assessment and Management Study; ENRGISE, Enabling Reduction of Low-Grade Inflammation in Seniors; EPOCH, Older People, Omega-3, and Cognitive Health Trial; EVAPORATE, Effect of Vascepa on Improving Coronary Atherosclerosis in People With High Triglycerides Taking Statin Therapy; FAAT, Fatty Acid Antiarrhythmia Trial; GISSI-HF, Gruppo Italiano per lo Studio della Sopravvivenza nell’Insufficienza Cardiaca; HEARTS, Slowing HEART Disease With Lifestyle and Omega-3 Fatty Acids; MARINA, Modulation of Atherosclerosis Risk by Increasing Doses of n-3 Fatty Acids; OFAMI, Omega-3 Fatty Acids in Myocardial Infarction; OMEMI, OMega-3 Fatty Acids in Elderly Patients With Myocardial Infarction; ORIGIN, Outcome Reduction With an Initial Glargine Intervention; PEACH, Pitavastatin and Combined Omega-3 Health Study; RCT-EPA, Effects of EPA on Prostate Cancer Cell Proliferation and Quality of Life; REDUCE-IT, Reduction of Cardiovascular Events With Icosapent Ethyl–Intervention Trial; RESPECT-EPA, Randomized Trial for Evaluation in Secondary Prevention Efficacy of Combination Therapy–Statin and Eicosapentaenoic Acid; SAREPITASO, Sarpogrelate-EPA-Pitavastatin-ASO Study; seAFOod Polyp Prevention, Systematic Evaluation of Aspirin and Fish Oil Polyp Prevention Trial; STRENGTH, Outcomes Study to Assess STatin Residual Risk Reduction With EpaNova in HiGh Cardiovascular Risk PatienTs With Hypertriglyceridemia; SU.FOL.OM3, SUpplementation with FOlate, Vitamin B6 and B12 and/or OMega-3 Fatty Acids; and VITAL Rhythm, VITamin D and Omega-3 Trial–Rhythm Study.

Posthoc comparisons, using the HR-HD group as the reference, demonstrated the following (Figure 2). The HR-LD group exhibited a significantly lower AF risk (OR=0.73 [95% CI, 0.60–0.88]; P=0.0012). Second, the LR-LD group had a significantly lower AF risk (OR=0.72 [95% CI, 0.59–0.88]; P=0.0015). Finally, the LR-HD group showed the largest reduction in AF risk (OR=0.71 [95% CI, 0.30–1.64]) as compared with the HR-HD group, although this was not statistically significant (P=0.49).

Sensitivity Analysis: Risk Difference

Figure S1 illustrates a similar pattern of results when considering the risk difference instead of ORs. Namely, only the HR-HD group showed a statistically significant increased risk (risk difference, 0.8% [95% CI, 0.40–1.1]; P<0.0001).

Figure S2 illustrates a similar pattern of results in sensitivity analyses restricted to studies with a confirmed absence of prevalent AF at baseline.^{6,68,69,72,75–77,82–84,87–90,96,97,101,102} In 3 of the 4 strata, ORs were nearly identical to those in the primary analysis (Figure 2; HR-HD: 1.46 versus 1.43; HR-LD: 1.07 versus 1.07; LR-LD: 1.06 versus 1.06), and the HR-HD group remained the only statistically significant stratum. Although the LR-HD OR differed numerically (0.68 versus 1.03), CIs were wide, and neither estimate was statistically significant.

Discussion

Recent meta-analyses reporting that DHA and EPA increase the risk of AF,^1–3 have raised doubts among physicians and the public about the safety of omega-3 products, whether dietary supplements or pharmaceutical preparations. As noted earlier, these previous meta-analyses included no more than 8 RCTs, whereas the present analysis now includes 35 RCTs. By examining how both background CVD risk and omega-3 dose influence AF outcomes, our findings provide a more nuanced perspective on the relationship between omega-3 FAs and risk for AF.

We found a statistically significant, 43% increase in the relative risk of AF (Figure 2) and an absolute risk increase of <1% in the HR-HD group (Figure S1). In contrast, the HR-LD, LR-HD, and LR-LD strata (Figure 2) showed pooled risk estimates that were substantially lower than those of the HR-HD group (ORs ≈1.0), and none were statistically significant. Importantly, post hoc comparisons using HR-HD as the reference, HR-LD, and LR-LD groups each showed approximately null and significantly lower AF risk (OR ≈0.70). Although not formally assessed, the increased AF risk observed in the HR-HD group was evident across trials employing different omega-3 formulations (Table), including EPA ethyl esters,⁶⁹ carboxylic acids,^70,111 and reesterified triglycerides.⁷² This consistency across formulations suggests that the observed AF signal in high-risk patients is unlikely to be attributable to a single omega-3 formulation, but rather appears to be linked to dose and background CVD risk.

Among LR individuals, the mean intake in the HD group was ≈2400 mg/d (Table), and this intake was associated with a reduced AF risk (OR=0.71) as compared with the HR-HD group, with a pooled OR of 1.03 (Figure 1). This suggests that in the general population, EPA/DHA intakes at this level are safe to consume (vis-à-vis risk for AF). This finding, however, was based on relatively few events; therefore should be considered tentative. Nevertheless, this is consistent with the Food and Drug Administration’s designation of EPA/DHA intakes of up to 3000 mg/d as generally recognized as safe.¹¹²

There are a variety of other considerations that would support the safety vis-à-vis AF in the general population. For example, on average, North American adults eat <1 serving per week of fish/seafood; thus, the mean intake of DHA+EPA in the United States is only ≈100 mg/d.¹¹³ This is associated with a mean DHA+EPA level in red blood cells (ie, the omega-3 index^17,18) of ≈5.5% in the United States.¹¹⁴ Observational studies consistently show that an omega-3 index of ≥8% is ideal for reducing risk of major adverse cardiovascular events,¹¹⁵ stroke,²⁰ and all-cause mortality.¹¹⁶ For typical Americans, achieving this would require an intake of ≈1000 mg/d of DHA+EPA.¹⁸ In Japan, where average omega-3 intake approximates this level,¹¹⁷ age-adjusted AF incidence is substantially lower than in the United States.^118,119 These ecological considerations are at least consistent with the view that in the general population, even relatively high doses (1–1.5 g/d) do not increase risk, and indeed may decrease risk, for AF.

There are several potential mechanisms that might explain the increased risk for AF in patients with HR taking high-dose omega-3. Omega-3 treatment has been shown to reduce resting heart rate¹²⁰ through increasing vagal tone even at relatively low doses of about 500 mg/d.^120,121 These dose-dependent vagotonic actions of omega-3 suggest a mechanism that may in part explain the divergent effects on risk of developing AF, whereby low-dose omega-3 does not significantly increase AF risk and may actually reduce risk, but high-doses can increase AF, especially in high-risk individuals.¹²² These patients frequently have underlying structural heart disease and are commonly treated with medications such as beta blockers, which may further increase vagal tone through sympathetic inhibition. The dose-dependent vagal-stimulating effects are likely to be especially problematic for individuals at high risk for bradycardia-dependent AF.¹²³ This includes individuals who perform high volumes of strenuous endurance exercise, patients with untreated obstructive sleep apnea, sick sinus syndrome, and chronically slow heart rates.¹²³ Reanalysis of high-dose omega-3 RCTs examining interactions between baseline heart rate, omega-3 intake, and AF incidence may help clarify this potential relationship.

Considering that stroke is the major complication of AF, it is paradoxical that higher omega-3 intake and blood levels have been repeatedly associated with a lower risk of stroke. Specifically, a comprehensive meta-analysis (n>180 000 subjects) found patients in the top quintile of marine omega-3 blood levels had an 18% lower risk of ischemic stroke, and no effect on hemorrhagic stroke compared with patients in the lowest quintile of omega-3.²⁰ Similarly, in the REDUCE-IT (Reduction of Cardiovascular Events With Icosapent Ethyl–Intervention Trial),^69,124 which was comprised of high-CVD risk patients, high-dose EPA (4000 mg/d) reduced relative risk of stroke by 28% despite increasing relative risk of AF by 35%.¹²⁵ Even those in the EPA arm of the REDUCE-IT study who developed AF experienced lower risks of stroke and major adverse cardiovascular events compared with those in the control group who developed AF.¹²⁵ This presumed protective effect against stroke may relate to the mild anti-thrombotic effects of EPA and DHA.^126,127 Still, for high-risk individuals with a history of AF, Olshansky et al reported a nonsignificantly higher risk for developing AF on EPA than similar patients in REDUCE-IT without a history of AF.⁶⁹ Accordingly, a lower omega-3 target dose of <1500 mg/d of DHA/EPA may be a safer level of intake for patients with CVD with a history of AF.

The potential risk of AF must be considered within the broader risk-benefit profile of omega-3 therapy, particularly in high-risk populations where higher doses are typically used. Across several major RCTs, such as REDUCE-IT,^69,124 RESPECT-EPA (Randomized Trial for Evaluation in Secondary Prevention Efficacy of Combination Therapy–Statin and Eicosapentaenoic Acid),⁷⁶ GISSI-HF (Gruppo Italiano per lo Studio della Sopravvivenza nell’Insufficienza Cardiaca),⁷⁷ and ASCEND (A Study of Cardiovascular Events in Diabetes)^6,79; the magnitude of cardiovascular benefit equaled or exceeded the observed increase in AF risk. For example, in REDUCE-IT, the 1.0% absolute increase in AF hospitalization with 4 g/d EPA was offset by a 4.8% absolute reduction in major adverse cardiovascular events,⁶⁹ a trade-off that strongly favors treatment given that events, such as myocardial infarction, ischemic stroke, or cardiovascular death, carry far greater clinical consequences than increased risk for new-onset AF. This interpretation aligns with recent cost-effectiveness analyses demonstrating that the reduction in major adverse cardiovascular events more than compensates for increased AF risk, supporting an overall favorable benefit-risk balance for high-dose omega-3 therapy.¹²⁸

Meta-analyses of prospective cohort studies focusing on omega-3 blood levels as the exposure have reported lower risk for mortality from all causes and from CVD, cancer, and other causes.²¹ Other biomarker-based studies have found higher levels of omega-3s to be associated with lower risk for atherosclerotic cardiovascular disease,¹¹⁵ Alzheimer’s disease¹²⁹ and total dementia,¹²⁹ heart failure,¹³⁰ diabetes,¹³¹ suicidal ideation/self-harm,¹³² poor lung function,¹³³ colorectal cancer,¹³⁴ and several other malignancies,¹³⁵ atopic dermatitis,¹³⁶ hospitalization for sepsis,¹³⁷ liver cirrhosis,¹³⁸ chronic kidney disease,¹³⁹ frailty¹⁴⁰ and hip fractures.¹⁴¹ These associations, in addition to a lower risk for AF,^4,19 support a generalized overall benefit of omega-3 in human health.

Strengths and Limitations

This meta-analysis represents a substantial advance in our understanding of the relationship between omega-3 treatment and risk for AF. Unlike earlier work that relied almost exclusively on published reports, we systematically integrated underutilized data sources, including trial registry adverse event records and unpublished data sets (Table S1); thereby reducing selective reporting bias. Importantly, we also incorporated trials in which no AF events were observed (where, had they occurred, they would have been noted), ensuring that such data contributed to the overall evidence base rather than being excluded. A further novel contribution is our risk-dose stratification framework, which moves 1 step beyond the high-dose versus low-dose paradigm used in past studies.¹ This approach provides a clearer understanding of how simple dietary supplementation might affect risk for AF as opposed to the use of high-dose phar major adverse cardiovascular events utical products. It also illuminates the importance of patient substrate, as we found limited evidence of an increase in AF risk even in high-risk CVD patients who are taking relatively low omega-3 doses (<1500 mg/d). Finally, by including populations outside of high-risk CVD cohorts, our analysis broadens generalizability and offers more nuanced insight into how AF risk may differ across clinical contexts. There are also limitations to this work, chief among them being the inability to ensure that all relevant studies were included in the analysis^43–67 (Table S2). In addition, for many RCTs, the status of AF outcomes was unclear, as PIs no longer had access to the original raw datasets^33–37 and were, therefore, unable to confirm whether AF events were systematically collected or adjudicated (Table S2).

Conclusions

This meta-analysis demonstrates a clear dose-dependent association between omega-3 FA supplementation and AF risk. Importantly, low-dose omega-3 supplementation does not appear to increase AF risk, even among patients at high CVD risk, providing reassurance regarding the safety of commonly used nutritional doses. In contrast, high doses of omega-3 FAs (typically from pharmacological products) administered to patients at high CVD risk were associated with a ≈40% to 45% increase in AF risk, although the absolute risk increase is modest (0.8%). These findings suggest that omega-3 supplementation providing <1500 mg of EPA+DHA per day (and possibly up to 2400 mg) can be used safely with respect to AF risk, whereas doses exceeding 3000 mg/d in high-risk CVD populations may warrant caution. Given that elevated blood omega-3 levels (achieved through diet and supplementation) are associated with reduced atherosclerotic cardiovascular disease risk and other favorable health outcomes, the small AF signal in HR-HD populations should be weighed against the substantial cardiovascular benefits when making treatment decisions.

Future prospective RCTs designed to evaluate the risk-benefit ratio of high-dose omega-3 supplementation in high-risk populations should incorporate systematic AF detection, including wearable-based monitoring, and objective biomarkers of omega-3 exposure, such as the omega-3 index.

Article Information

Acknowledgments

The authors thank Hertzel Gerstein, MD, for providing access to data from the ORIGIN trial (NCT00069784). Data from the ADCS-DHA trial (NCT00440050) used in the preparation of this article were obtained from the University of California, San Diego Alzheimer’s Disease Cooperative Study (ADCS) legacy database (https://www.adcs.org/). Data collection and sharing for this project were funded by the ADCS (National Institutes of Health, grant U19 AG010483). Data from the AREDS2 trial (NCT00345176) were obtained from the National Eye Institute, National Institutes of Health, and are available through dbGaP (accession phs002015.v2. p1).

Disclosures

Dr Abuknesha reports receiving the GOED Clinical Study Database Research Award (2023), which supported participation in clinical trial database research activities related to omega-3 RCTs. Dr O’Keefe is the Chief Medical Officer of CardioTabs, a company that sells omega-3 products. Dr Harris holds stock in OmegaQuant Analytics, LLC, a laboratory that offers blood fatty acid testing (including the Omega-3 Index) for researchers, clinicians, and consumers. The other authors report no conflicts.

Supplemental Material

Supplemental Methods

Tables S1–S3

Figures S1 and S2

Supplemental Appendix

References ^142–156

Supplementary Material

hae-19-e014785-s001.pdf ^{(1.4MB, pdf)}

Funding Statement

The authors declare that financial support was received for the research and publication of this article. The project was supported in part by The Global Organization of Omega-3 EPA and DHA (GOED), which had no role in the research design or conduct; data analysis or interpretation; preparation, review, or approval of the article; or the decision to submit the article for publication. Dr Qian is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under award number T32HL125232 (Multidisciplinary Training Program in Cardiovascular Epidemiology).