Gronroos · PloS one 2012 · prospective cohort study · n=14222

Fish, fish-derived n-3 fatty acids, and risk of incident atrial fibrillation in the Atherosclerosis Risk in Communities (ARIC) study.

Cited 37 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective observational cohort study

PubMed 22570739 · doi:10.1371/journal.pone.0036686 · record verified 2026-08-30

What was done

The authors evaluated whether dietary intake of fish, eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) or plasma phospholipid levels of EPA and DHA were associated with incident atrial fibrillation (AF). The study followed 14,222 participants aged 45-64 (27% African American) in the Atherosclerosis Risk in Communities (ARIC) study for an average of 17.6 years. Diet was assessed via a food frequency questionnaire, plasma phospholipid fatty acids were measured in a subgroup (n = 3,757), and AF was ascertained through ECGs, hospital discharge codes, and death certificates. Hazard ratios were estimated using multivariable Cox proportional hazards models.

What was found

Over follow-up, 1,604 incident AF events occurred. Total fish intake, dietary DHA, and dietary EPA were not significantly associated with AF risk. Higher intake of oily fish and canned tuna showed a non-significant lower risk (p for trend = 0.09). Total phospholipid DHA+EPA was not associated with AF. When analyzed separately, higher plasma phospholipid DHA was associated with a lower risk of AF, whereas EPA showed no association (exact hazard ratios and confidence intervals not reported in the abstract).

Why it matters

Whether long-chain n-3 polyunsaturated fatty acids prevent atrial fibrillation remains controversial. These findings from a large, biracial cohort indicate that general dietary fish consumption and total n-3 PUFAs do not confer meaningful protection against incident AF, although individual fatty acids like DHA warrant further mechanistic investigation.

Limits

Dietary intake was assessed by self-reported food frequency questionnaire, which is susceptible to measurement error and recall bias. Plasma biomarkers were measured in only a subset (n = 3,757) and at a single baseline time point. As an observational study, residual confounding cannot be excluded, and asymptomatic AF episodes not presenting to medical care may have been missed.

Cited by