Habitual fish oil supplementation and the risk of incident atrial fibrillation: findings from a large prospective longitudinal cohort study 

Abstract

Aims

Mixed effects of fish oil supplementation on the risks of atrial fibrillation (AF) were observed in several large-scale randomized controlled trials. Whether this relationship would be modified by genetic AF risk, baseline cardiovascular disease (CVD) status and background oily fish consumption are unknown.

Methods and results

We included 468 665 participants without AF at baseline from the UK Biobank cohort. The association between fish oil supplementation and the AF risk was assessed in the study cohort and in several subgroups, including genetic AF predisposition, baseline CVD status, and background oily fish consumption. During a median follow-up of 11.1 years, fish oil users had a higher rate of incident AF (6.2% vs. 5.2%, adjusted hazard ratio of 1.10, and 95% confidence interval of 1.07, 1.13). Compared with non-users, fish oil users had a higher rate of incident AF in the low (3.7% vs. 3.0%, P= 0.02), intermediate (5.8% vs. 4.8%, P < 0.0001), and high (9.8% vs. 8.1%, P < 0.0001) genetic AF risk groups. In participants without CVD at baseline, fish oil users had a higher rate of incident AF (5.3% vs. 4.1%, P < 0.0001), which was not observed in participants with CVD at baseline (11.6% vs. 11.1%, P = 0.56), with significant interaction (P-interaction < 0.0001). The association between fish oil supplementation and the AF risk was not modified by background oily fish consumption (P-interaction = 0.62).

Conclusion

Habitual fish oil supplementation was associated with the risk of incident AF, regardless of genetic AF predisposition and background oily fish consumption. This association was observed only in individuals without CVD at baseline.

Graphical Abstract

Atrial fibrillation, Fish oil, Genetic risk score, Cardiovascular disease, Oily fish

What’s new?

• In this large prospective cohort study including approximately half million general populations, habitual fish oil supplementation was significantly associated with the risk of incident atrial fibrillation (AF).

• The positive association between habitual fish oil supplementation and the risk of incident AF was not modified by genetic predisposition for AF or background oily fish consumption.

• The association between fish oil supplementation and the risk of incident AF was observed only in people without cardiovascular disease at baseline.

What is clinical implication?

• The general population should be informed of the potential AF risks with fish oil supplementation; and the physicians should balance the risk–benefit ratio with fish oil supplementation, given that the cardiovascular benefit of fish oil supplementation is indetermined in prior clinical trials.

• Future clinical trials examining the effects of fish oil supplementation on the risk of incident AF should consider a longer duration of follow-up.

Introduction

Cardiovascular benefits of fish oil supplementation have been examined in several clinical trials, and the results are mixed.^1–6 Exploratory analyses of these trials suggest that there might be a trend towards increasing atrial fibrillation (AF) risk with fish oil supplementation.^1–5,7 However, the VITAL Rhythm study, the largest AF primary prevention randomized trial, did not show an increased AF risk with fish oil supplementation.⁸

A recent meta-analysis, which included aforementioned clinical trials, suggested that fish oil supplementation was associated with a significantly higher risk of AF.⁹ These mixed findings raise concerns regarding the off-target adverse effects of fish oil supplementation. However, the participants from seven large clinical trials included in the meta-analysis might not represent populations that are seen in daily clinical practice.⁹

In addition, these findings are subject to unmeasured confounding biases such as background oily fish consumption and genetic AF risk. Notably, genetic risk plays an important role in AF development,¹⁰ and a recent genome-wide association study identified ~140 loci associated with AF.¹¹ Therefore, studies accounting for genetic AF risk would provide more convincing evidence regarding the association between fish oil supplementation and the AF risk, if any.

Based on results from prior studies,^6,12,13 the Science Advisory from the American Heart Association recommends fish oil supplementation for individuals with prevalent coronary heart disease or heart failure with reduced ejection fraction.¹⁴ However, results of two recent clinical trials are again conflicting in terms of cardiovascular benefit with fish oil supplementation for individuals with prevalent cardiovascular disease (CVD).^1,2 In those without CVD, there is no indication for fish oil supplementation.^3,4,15,16 Given the differential recommendations, it is clinically relevant to evaluate the association between fish oil supplementation and the AF risk in individuals with and without prevalent CVD, respectively.

Herein, leveraging data from the large population-based UK Biobank cohort study, the aims of this study were to examine whether habitual fish oil supplementation at baseline was associated with the AF risk in the real-world setting. Second, we evaluated whether genetic AF predisposition, CVD status at baseline, and background oily fish consumption would modify the relationship between habitual fish oil supplementation and the AF risk.

Methods

Study design and population

Data from the UK Biobank (application # 69550) were used for the current analysis. In brief, UK Biobank is a large-scale prospective study including over 500 000 participants aged 40–69 years when recruited in 2006–2010.¹⁷ Demographics, including age, gender, and ethnicity, were recorded through the touch screen questionnaire and face-to-face interview. Blood sampling was used for genotyping. Information on clinical characteristics were collected through record linkage to Health Episode Statistics in England and Wales and the Scottish Morbidity Records in Scotland.¹⁷ UK Biobank received ethical approval from the research ethics committee (REC reference for UK Biobank 11/NW/0382), and informed consent was obtained during participants’ enrolment.

A total of 502 461 participants were examined for the eligibility of the present study (Figure 1). The reasons for exclusion included the following: withdrawal from the study (n = 1298), missing quality-controlled genotyping data or high relatedness (n = 15 139), presence of AF at baseline (n = 8092), and had missing or outlier data (n = 9867).

Figure 1

Study flowchart

Exposure assessment

In UK Biobank, information on habitual fish oil supplementation was recorded via an electronic questionnaire and a verbal interview. For the questionnaire, each participant was asked: ‘Do you regularly take any of the following?’ and participants could select the answer from a list of supplements which include fish oil. In addition, trained staff conducted a verbal interview and asked whether the participants were currently receiving any treatment/medication including omega-3 or fish oil supplementation. In order to validate and reinforce this self-reported information, it was linked with the participants’ past medical records. We summarized the above information and defined habitual fish oil supplementation as 0 = no and 1 = yes.

Outcome ascertainment

The outcome of the present study was incident AF. The information on incident AF [field ID 131351 and The international classification of diseases (ICD) code I48] was extracted from the first occurrence of health outcomes which was defined by a three-character International Statistical Classification of Diseases and Related Health Problems 10th Revision code (category ID in UKB 1712). The diagnosis of AF was obtained using linkage with death register, primary care, and hospital inpatient records. Follow-up duration for each participant was calculated from the date of first assessment and censored at the date associated with the development of AF, date of death or last known follow-up until 31 March 2021, whichever occurred earliest.

Genetic risk score

A total of 165 single nucleotide polymorphisms (SNPs) associated with AF were used to calculate the genetic risk score (GRS) for AF (see Supplementary material online, TableS1), as reported in a meta-analysis of genome-wide association studies with over 1 000 000 people that did not include UK Biobank participants.¹¹ Briefly, most of genome-wide significant risk variants for AF fall in gene which cause serious heart defects in humans (e.g. PITX2, TBX5) or near genes important for striated muscle function and integrity (e.g. CFL2, MYH7). Via cardiac structural remodelling, they are crucial for cardiac ion channels and calcium signalling, along with cardiac transcription factors. Each SNP was recoded as 0, 1, or 2 according to the number of risk alleles and then multiplied by the regression coefficient obtained from the reported meta-analysis to calculate the GRS: GRS = (β₁ × SNP₁ + β₂ × SNP₂ + … + β₁₆₅ × SNP₁₆₅). We classified participants into three groups as previously reported:¹⁸ low (lowest quartile), intermediate (mid two quartiles), and high (highest quartile) genetic AF risk groups.

Covariates

A series of covariates were considered in this study, including age, gender, ethnicity, Townsend deprivation index, oily and non-oily fish consumption, current smoking, and drinking. In brief, information on oily fish consumption was collected by asking ‘How often do you eat oily fish? (e.g. sardines, salmon, mackerel, herring)’ and the response was fallen into two categories (that is < 2 and ≥ 2 times per week). Information on comorbidities [obesity (field ID 130793, ICD code E66), hypertension (field ID 131287/131295, ICD code I10/I15), diabetes mellitus (field ID 130707/130709/130711/130713/130715, ICD code E10–E14), chronic obstructive pulmonary disease (field ID 131493, ICD code J44), chronic renal failure (field ID 131355, ICD code I50), coronary heart disease, and heart failure (field ID 131297/131299/131301/131303/131305/131307, ICD code I20–I25)] was extracted from the ‘first occurrence’ (category ID in UKB 1712). Having CVD at baseline were coded according to ICD code I00–I99. Medication use including antihypertensives, statins, and antidiabetics were extracted through treatment/medication records (field ID 20003). Physical activity (Field 22032) was classified into low [< 600 metabolic equivalent (MET)-min/week], moderate (600–3000 MET-min/week), and high (≥ 3000 MET-min/week). Binge drinking was defined as consumption ≥6 standard drinks/day for women or ≥8 standard drinks/day for men. Detailed information on alcohol consumption and binge drinking in the UK Biobank was reported previously.¹⁹

Statistical analysis

Baseline characteristics were presented as number (proportion) for categorical variables and mean ± standard deviation for continuous variables. The association between habitual fish oil supplementation and incident AF was evaluated using Cox proportional hazard models. Hazard ratio (HR) and 95% confidence interval (CI) was reported. The proportional hazard assumption was tested using Schoenfeld residuals. The models were adjusted for age, gender, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, GRS, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics. Several subgroup analyses were conducted to assess potential modification effects by the following factors: GRS, baseline CVD status, age, gender, ethnicity, background oily fish consumption, drinking, obesity, and hypertension. Modification effect was evaluated by modelling the additive scales of the stratifying variable with fish oil supplementation. Cumulative rate of AF was evaluated using Kaplan–Meier method with log-rank test. To test the robustness of the main results, three sensitivity analyses were performed: (i) using a competing risk analysis that defined all-cause mortality as a competing event for AF; (ii) excluding participants who developed AF events within the first two years’ follow-up; (iii) performing propensity matching score analysis to achieve exchangeability between fish oil user and non-user groups; performing Cox proportional hazard models; (iv) adjusted by all covariates including physical activity and alcohol consumption; and (v) adjusted by all covariates including physical activity and binge drinking. All the statistical analyses were implemented in R software 4.1.1 (R Development Core Team, Vienna, Austria). All statistical tests were two-tailed, and a P-value < 0.05 was considered statistically significant.

Patient and public involvement

No patients or the public were involved in raising the question and the outcome of interest nor were they involved in developing plans for recruitment, design, or conduction of the study. No patients or the public were asked to provide comment or advise to interpret and write up the results.

Results

Baseline characteristics

A total of 468 665 participants were included in the present study, and 31.6% (n = 148 192) of participants reported to have habitual fish oil supplementation at baseline. The mean age of the study cohort was 56.5 years, with 45.2% (n = 212 020) males and 5.0% (n = 23 575) with CVD at baseline. Differences in baseline characteristics between fish oil users and non-users are presented in Table 1. Fish oil users were older and more likely to be White, female, and current drinkers. In addition, they had lower mean Townsend deprivation index and ate oily and non-oily fish more frequently. Fish oil users had a lower prevalence of obesity, diabetes mellitus and chronic obstructive pulmonary disease, and a higher prevalence of hypertension and coronary heart disease, and use of antihypertensives and statins.

Table 1

UK Biobank participants’ characteristics according to fish oil supplementation

	Overall (n = 468 665)	Fish oil non-users (n = 320 473)	Fish oil users (n = 148 192)	P-value
Age (years)	56.5 ± 8.1	55.5 ± 8.2	58.6 ± 7.4	<0.0001
Age ≥ 65 years, n (%)	87 649 (18.7)	50 978 (15.9)	36 671 (24.8)	<0.0001
White, n (%)	444 687 (94.9)	303 200 (94.6)	141 487 (95.5)	<0.0001
Male, n (%)	212 020 (45.2)	148 561 (46.4)	63 459 (42.8)	<0.0001
Townsend deprivation index	−1.34 ± 3.07	−1.24 ± 3.12	−1.55 ± 2.95	<0.0001
Oily fish consumption, n (%)				<0.0001
< 2 times/week	207 382 (44.3)	153 791 (48.0)	53 591 (36.2)
≥ 2 times/week	261 283 (55.8)	166 682 (52.0)	94 601 (63.8)
Non-oily fish consumption, n (%)				<0.0001
< 2 times/week	158 114 (33.7)	114 545 (35.7)	43 569 (29.4)
≥ 2 times/week	310 551 (66.3)	205 928 (64.3)	104 623 (70.6)
Current smoking, n (%)	49 304 (10.5)	37 212 (11.6)	12 092 (8.2)	<0.0001
Current drinking, n (%)	432 058 (92.2)	294 288 (91.8)	137 770 (93.0)	<0.0001
Obesity, n (%)	12 276 (2.6)	8772 (2.7)	3504 (2.4)	<0.0001
Hypertension, n (%)	45 531 (9.7)	30 454 (9.5)	15 077 (10.2)	<0.0001
Diabetes mellitus, n (%)	23 417 (5.0)	16 614 (5.2)	6803 (4.6)	<0.0001
COPD, n (%)	4256 (0.9)	3070 (1.0)	1186 (0.8)	<0.0001
CRF, n (%)	5311 (1.1)	3644 (1.1)	1667 (1.1)	0.71
CHD, n (%)	23 135 (4.9)	15 505 (4.8)	7630 (5.2)	<0.0001
Heart failure, n (%)	1692 (0.4)	1191 (0.4)	501 (0.3)	0.07
Antihypertensives, n (%)	102 068 (21.8)	67 547 (21.1)	34 521 (23.3)	<0.0001
Statins, n (%)	74 496 (15.9)	48 733 (15.2)	25 763 (17.4)	<0.0001
Antidiabetics, n (%)	17 020 (3.6)	12 324 (3.9)	4696 (3.2)	<0.0001

COPD, chronic obstructive pulmonary disease; CRF, chronic renal failure; CHD, coronary heart disease

Fish oil supplementation and the risk of incident atrial fibrillation

During a median follow-up of 11.1 (interquartile range 10.3–11.8) years, 9241 and 16 507 cases of incident AF occurred in fish oil users and non-users, and the cumulative rate was 6.2% vs. 5.2%, with an unadjusted HR of 1.22 (95% CI: 1.19, 1.25; Table 2). After adjusted for covariates including GRS, fish oil supplementation remained associated with an increased risk of incident AF, with adjusted HR of 1.10 (95% CI: 1.07, 1.13; Figure 2A).

Figure 2

Cumulative rate of atrial fibrillation with fish oil supplementation. (A) In the overall cohort, fish oil supplementation was associated with an increased risk of incident atrial fibrillation, with an adjusted HR of 1.10 (95% confidence interval: 1.07, 1.13). (B) Fish oil supplementation was associated with an increased risk of incident atrial fibrillation in low genetic risk group, with an adjusted hazard ratio of 1.08 (95% confidence interval: 1.01, 1.16). (C) Fish oil supplementation was associated with an increased risk of incident atrial fibrillation in intermediate genetic risk group, with an adjusted hazard ratio of 1.10 (95% confidence interval: 1.06, 1.14). (D) Fish oil supplementation was associated with an increased risk of incident atrial fibrillation in high genetic risk group, with an adjusted hazard ratio of 1.11 (95% confidence interval: 1.06, 1.15). (E) In participants with cardiovascular disease at baseline, fish oil supplementation was not associated with an increased risk of incident atrial fibrillation, with an adjusted hazard ratio of 1.01 (95% confidence interval: 0.97, 1.06). (F) In those without cardiovascular disease at baseline, fish oil supplementation was associated with a high risk of incident atrial fibrillation, with an adjusted hazard ratio of 1.13 (95% confidence interval: 1.10, 1.17). Covariates included age, gender, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, genetic risk score, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

Table 2

Risk of incident atrial fibrillation with fish oil supplementation

	Events n (%)	HR (95% CI)	P-value	Adjusted HR (95% CI)	P-value
Fish oil non-users (n = 320 473)	16 507 (5.2)	Reference		Reference
Fish oil users (n = 148 192)	9241 (6.2)	1.22 (1.19, 1.25)	<0.0001	1.10 (1.07, 1.13)	<0.0001

HR, hazard ratio; CI, confidence interval.

Adjusted for age, sex, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, genetic risk score, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

Incident atrial fibrillation according to genetic risk score, baseline cardiovascular disease status, and background oily fish consumption

In low, intermediate, and high genetic risk groups, fish oil users had a significantly higher rate of incident AF compared with non-users, with adjusted HRs of 1.08 (95% CI: 1.01, 1.16), 1.10 (95% CI: 1.06, 1.14), and 1.11 (95% CI: 1.06, 1.15), respectively (Table 3; and Figure 2B–D). The magnitude of the associations between fish oil supplementation and the AF risk were greater in the high vs. low genetic risk group (P-interaction = 0.008).

Table 3

Risk of incident atrial fibrillation with fish oil supplementation according to genetic risk score

		Events n (%)	Adjusted HR (95% CI)	P-value	P-value for interactiona
Low	Fish oil non-users (n = 80 740)	2421 (3.0)	Reference
	Fish oil users (n = 37 523)	1387 (3.7)	1.08 (1.01, 1.16)	0.02
Intermediate	Fish oil non-users (n = 160 767)	7677 (4.8)	Reference		0.10
	Fish oil users (n = 74 279)	4289 (5.8)	1.10 (1.06, 1.14)	<0.0001
High	Fish oil non-users (n = 78 966)	6409 (8.1)	Reference		0.008
	Fish oil users (n = 36 390)	3565 (9.8)	1.11 (1.06, 1.15)	<0.0001

Adjusted for age, sex, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

HR, hazard ratio; CI, confidence interval.

a

The P-value for additive interaction was adjusted by FDR (false discovery rate).

In participants with CVD at baseline, fish oil supplementation was not associated with an increased risk of incident AF, with an adjusted HR of 1.01 (95% CI: 0.97, 1.06; Table 4 and Figure 2E). Although in those without CVD at baseline, fish oil supplementation was associated with an increased risk of incident AF, with an adjusted HR of 1.13 (95% CI: 1.10, 1.17; Figure 2F). The magnitude of the association between fish oil supplementation and the AF risk was greater in participants without CVD vs. those with CVD (P-interaction < 0.0001).

Table 4

Risk of incident atrial fibrillation with fish oil supplementation according to baseline cardiovascular disease status

		Events n (%)	Adjusted HR (95% CI)	P-value	P-value for interactiona
Yes	Fish oil non-users (n = 46 984)	5232 (11.1)	Reference		<0.0001
	Fish oil users (n = 23 067)	2670 (11.6)	1.01 (0.97,1.06)	0.56
No	Fish oil non-users (n = 273 489)	11 275 (4.1)	Reference
	Fish oil users (n = 125 125)	6571 (5.3)	1.13 (1.10,1.17)	<0.0001

Adjusted for age, sex, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, genetic risk score, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

HR, hazard ratio; CI, confidence interval.

a

The P-value for additive interaction.

In participants with oily fish consumption < 2 times/week, fish oil supplementation was associated with 9% increased risk of incident AF (95% CI: 1.04, 1.14; Figure 3), and in those with oily fish consumption ≥ 2 times/week, fish oil supplementation was associated with 11% increased risk of incident AF (95% CI: 1.05, 1.15; P-interaction = 0.62).

Figure 3

Subgroup analysis. The magnitude of the association between fish oil supplementation and the atrial fibrillation risk was greater in White (P-interaction = 0.005), in non-obese participants (P-interaction = 0.005), in normotensive participants (P-interaction < 0.001), and in participants with current drinking (P-interaction = 0.02). Covariates included age, gender, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, genetic risk score, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

Subgroup and sensitivity analysis

Several subgroup analyses were conducted and as presented in Figure 3, and the magnitude of the associations between fish oil supplementation and the AF risk were greater in White ethnicity (P-interaction = 0.005), non-obese participants (P-interaction = 0.005), normotensive participants (P-interaction < 0.001), and in participants with current drinking (P-interaction = 0.02). Sensitivity analyses showed that there was no substantial change when we excluded participants who died before developing AF or who developed AF during the first 2 years’ follow-up and performed propensity matching score analysis or Cox proportional hazard models adjusted by different covariates (Table 5; and see Supplementary material online, TableS2).

Table 5

Sensitivity analysis

		Events n (%)	Adjusted HR (95% CI)	P-value for interactiona
Performing competing risk analysis	Fish oil non-users (n = 320 473)	16 454 (5.1)	Reference
	Fish oil users (n = 148 192)	9241 (6.2)	1.11 (1.08, 1.13)	<0.0001
Excluding AF event within first 2 years’ follow-up	Fish oil non-users (n = 318 822)	14 856 (4.7)	Reference
	Fish oil users (n = 147 256)	8305 (5.6)	1.10 (1.07, 1.13)	<0.0001
Adjusting for propensity scoreb	Fish oil non-users (n = 230 488)	11 191 (4.9)	Reference
	Fish oil users (n = 115 244)	6219 (5.4)	1.11 (1.08, 1.15)	<0.0001
Adjusting for all variables with physical activity and alcohol consumption	Fish oil non-users (n = 207 055)	10 210 (4.9)	Reference
	Fish oil users (n = 97 547)	6024 (6.18)	1.22 (1.18, 1.26)	<0.0001
Adjusting for all variables with physical activity and binge drinking	Fish oil non-users (n = 226 894)	11 376 (5.0)	Reference
	Fish oil users (n = 105 593)	6553 (6.2)	1.20 (1.17, 1.24)	<0.0001

Adjusted for age, sex, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, genetic risk score, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

HR, hazard ratio; CI, confidence interval.

a

The P-value for additive interaction was adjusted by FDR (false discovery rate).

b

Propensity score was calculated based on logistic regression with independent variables including age, sex, ethnicity, Townsend deprivation index, oily fish consumption, non-oily fish consumption, genetic risk score, smoking, drinking, obesity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, chronic renal failure, coronary heart disease, heart failure, antihypertensives, statins, and antidiabetics.

Discussion

Leveraging data from approximately half million community populations with a median follow-up of 11.1 years, the present study has great statistical power to evaluate the associations between fish oil supplementation and the AF risk in the real-world setting. There are four potentially clinically important findings. First, fish oil supplementation was associated with a 10% increased risk of AF even after adjusting for multiple covariates including GRS and background oily fish consumption; second, the positive association between fish oil supplementation and the AF risk was consistent in the low, intermediate, and high genetic AF risk groups; third, the AF risk associated with fish oil supplementation was observed only in individuals without CVD at baseline; last, the AF risk associated with fish oil supplementation was not modified by background oily fish consumption.

Two large clinical trials, which used 4 g/daily of different formulation of fish oil supplementation, showed a significantly higher rate of incident AF with fish oil supplementation.^1,2 The other four trials showed an insignificantly higher rate of incident AF with fish oil supplementation,^3–5,7 which might be due to small number of AF cases. The VITAL Rhythm study, the largest AF primary prevention randomized trial, also did not show a significantly higher rate of AF with fish oil supplementation,⁸ which might be partly explained by a low dosage used (460 mg/d of EPA [Eicosapentaenoic Acid] and 380 mg/d of DHA [Docosahexaenoic Acid]). Indeed, investigators of the VITAL Rhythm study conducted a meta-analysis that included the aforementioned clinical trials, and the results demonstrated that fish oil supplementation was associated with a 25% higher risk of incident AF.⁹ A dose-dependent effect of fish oil supplementation on the AF risk was reported.⁹

Our study was consistent with prior trials^1,2 and the meta-analysis.⁹ However, we were unable to confirm if there was a dose-dependent effect due to lack of dosage information. Change in fatty acid content of the cellular membrane with fish oil supplementation over time might be a potential mechanism to explain this observation. Indeed, one recent study suggested that when human endothelial cells were exposed to different concentration (0–20 µM) of fish oil, high concentrations of fish oil altered cell membrane properties, and inhibited Na-K-ATPase pump activity, whereas a low concentration (3.75 µM/mL) of fish oil minimized peroxidation potential and optimized activity.²⁰ In addition, another study further showed that compared with the healthy control, the percentage of total polyunsaturated fatty acid (PUFA), PUFA n-3 and n-6 on the red blood cell membrane was higher in individuals with AF or flutter.²¹ In addition, among populations with heart failure, the incidence of ventricular arrhythmia was higher in those with higher PUFA content on the red blood cell membrane.²² Elevated PUFA level on cell membrane increases its susceptibility to oxidative damage.²¹ These observations together suggest that altered PUFA level on cell membrane might be one of the mechanisms associated with the AF risk with fish oil supplementation. Further studies are needed to further elucidate the precise mechanisms of arrhythmogenesis related to incident AF with fish oil supplementation.

Extending prior reports, our present study also examined whether genetic AF predisposition would modify the relationship between fish oil supplementation and the AF risk. We found that fish oil supplementation remained associated with the AF risk even after adjusting for GRS and other covariates, suggesting that the AF risk associated with fish oil supplementation was independent of genetic AF predisposition. In addition, the AF risk associated with fish oil supplementation was consistent in the low, intermediate, and high genetic risk groups, suggesting that the genetic background might not modify the relationship between fish oil supplementation and the AF risk. Furthermore, it is noted that the absolute rate of incident AF was increased with increasing GRS, and the additive interaction analysis was positive for the high vs. low genetic risk groups, confirming the importance of genetic predisposition for AF development.^11,23

Extending prior trials^1,2 and the meta-analysis,⁹ our study was able to evaluate whether background oily fish consumption would modify the relationship between fish oil supplementation and the AF risk. On the one hand, the overall result suggests that after adjusting for oily fish consumption and other covariates, fish oil supplementation remained associated with the AF risk. On the other hand, in the subgroup analysis, fish oil supplementation was associated with a comparable AF risk in both subgroups with < 2 times/week and ≥ 2 times/week oily fish consumption (P-interaction = 0.62). These findings confirm that the association between fish oil supplementation and the AF risk was not modified by background oily fish consumption.

In participants without CVD at baseline, fish oil supplementation was associated with 10% increased risk of AF, which was similar to that in the VITAL Rhythm trial (HR: 1.09).⁸ These observations suggest that it is feasible that such a small adverse effect might have been present in the VITAL Rhythm trial but was not statistically significant due to the relatively small sample size and relatively short follow-up duration. Indeed, the Kaplan–Meier curve continued to separate at the end of 5-year follow-up in the trial, which was also observed in the present study (Figure 2E). These observations have implications for clinical trial design in the future.

Importantly, the AF risk associated with fish oil supplementation was not found in those with CVD at baseline, which was contrast to the observations from the clinical trials that recruited participants with prevalent CVD. For example, in the REDUCE-IT and STRENGTH trials,^1,2 fish oil supplementation was associated with the AF risk in participants with CVD at baseline. One potential explanation is that participants with CVD at baseline only comprised 5.0% of the present study population, which might not have sufficient statistical power to detect the association. Different dosages used between these studies (that is, high dosage used in the clinical trial) might be also related to the divergent findings. Furthermore, unmeasured confounding by indication in individuals without CVD at baseline could also account for these differing findings. Finally, this may represent a ceiling effect for some individuals who already are at a heightened AF risk. Notably, CVD per se is a strong risk factor of AF, and individuals with prevalent CVD might not experience substantial elevation in the AF risk with fish oil supplementation. Indeed, fish oil supplementation was associated with a 10% increased risk of AF in the present study. This postulation was further supported by subgroup analysis. In non-obese and normotensive participants, fish oil supplementation was associated with an increased AF risk, whereas in those with obesity and hypertension, both well-known risk factors of AF, fish oil supplementation was not associated with an increased AF risk. Further studies are needed to corroborate these findings.

Strengths and limitations

The present study has two strengths. First, this study has a large sample size with a long-term follow-up, which enabled us to observe sufficient incident AF cases and perform several clinically relevant subgroup analyses. Second, genetic risk information and background oily fish consumption was included and accounted for, which provided more convincing evidence regarding the association of fish oil supplementation and the AF risk.

There are also some limitations. First, this is an observational study, and no causal relationship can be drawn from the current findings. Second, although we have adjusted for multiple covariates including GRS and background oily fish consumption, residual confounding by imperfectly measured or unmeasured confounders might account for the modest association observed. Third, information on dosages and formulations of fish oil supplementation (e.g. eicosapentaenoic acid or docosahexaenoic acid) used were unavailable, which could not allow us to evaluate the dose-response effects and to differentiate the effects according to the fish oil formulation. In addition, the present study also lacks measures of PUFA level on the membrane of red blood cell, which could not allow us to explore whether the AF risk associated with fish oil supplementation was related to the alteration of FUFA on cell membrane. Fourth, only information on fish oil supplementation at baseline was captured, and this information might change over time during the follow-up, which might influence the relationship. Fifth, participants of the present study were relatively young, and these findings might not be extrapolated to the elderly populations, in whom the AF risk is elevated with age. Finally, most of the participants in the present study were White European and whether the findings can be generalized into other racial/ethnic groups were unknown.

Conclusion

In the large community populations from the UK Biobank cohort, habitual fish oil supplementation was associated with the AF risk, regardless of genetic AF predisposition and background oily fish consumption, and this association was observed only in individuals without CVD at baseline.

Authors’ contributions

J.Z., A.C., and HL.L. designed research and generated study plan; J.Z., G.C., X.W., and M.C. conducted the analysis and drafted the manuscript; J.Z., A.C., G.C., and HL.L. analyzed the data; A.C., HT.L., S.E.N., G.Y.H.L., and HL.L. revised the manuscript. HL.L. had primary responsibility for final content. All gave final approval and agree to be accountable for all aspects of work ensuring integrity and accuracy.

Supplementary material

Supplementary material is available at European Journal of Preventive Cardiology

Acknowledgements

We wish to acknowledge the UK Biobank Resource under Application Number 69550 and also gratefully acknowledge many participants who provided medical data to the UK Biobank.

Funding

None.

Data availability

Data are available upon reasonable request to the UK Biobank management team (https://www.ukbiobank.ac.uk/media/px5gbq4q/access_019-access-management-system-user-guide-v4-1.pdf).

Ethics approval and consent to participate

The project has approval from the North West Multi-centre Research Ethics Committee (MREC) (REC reference: 21/NW/0157), and informed written consent was obtained from each participant.

Author notes

Conflict of interest: None declared.