Fish oil capsules vs eating fish: what the heart trials found

Capsules and fish are not the same intervention. Across 86 trials in 162,796 people, taking long-chain omega-3 made little or no difference to deaths or heart events.¹ Yet people who eat fatty fish keep turning out healthier.⁶ The gap between those two facts is the whole story.

What the capsule trials found

The reference point is a Cochrane review covering 86 trials and 162,796 participants, running from 12 to 88 months, at doses from half a gram a day to more than five.¹ Most of them tested capsules.

The results come as risk ratios. A ratio of 0.97 means 97 events in the treated group for every 100 in the comparison group. Long-chain omega-3 came in at 0.97 for death from any cause and 0.96 for heart and blood-vessel events.¹ Stroke was flat. Irregular heartbeat was flat.

Two things moved a little, at low certainty. For every 334 people who took omega-3, one avoided a death from coronary heart disease; for every 167, one avoided a coronary event.¹ Triglycerides, a fat carried in the blood, fell about 15%, and more so at higher doses.¹

Omega-3 capsules and coronary events, in people one coronary event avoided no difference

The phrase worth pausing on is high certainty, which the review gave to the findings on deaths and heart events overall. This is not thin evidence waiting to be filled in. It is good evidence of a very small effect. The same review says plainly that there is little evidence about eating fish, because these trials tested a supplement rather than a meal.¹

The trials that did find something

Other analyses of overlapping trials report modest benefits. What separates them is which outcomes get counted, and which kind of oil dominates.

One brought together 15 trials and found fewer major heart events, at 0.95, fewer heart attacks at 0.90, and fewer cardiovascular deaths at 0.94. Heart failure, stroke and overall death were unchanged.² The real finding sits in the breakdown. The benefit came from prescription EPA ethyl ester, a single purified fatty acid, and not from mixed fish oil.²

A second analysis of 18 trials in 134,144 people pressed the same point. Omega-3 lowered artery-reopening procedures to 0.90, heart attacks to 0.89 and cardiovascular death to 0.92, and the effect held up in trials where most participants were already taking statins.³ Set against EPA and DHA together, EPA on its own cut those procedures further, to 0.76.³

So the honest reading is not that omega-3 does nothing. It is that purified EPA seems to do something measurable, and mixed-oil capsules do very little.

What eating fish looks like instead

The evidence from following people over years runs the other way, and it is not a thin literature. Across 40 such studies, more fish went with less coronary heart disease: 0.91 for new cases, 0.85 for deaths from it.⁵ The relationship was a slope rather than a cliff. Each extra 20 g of fish a day went with 4% less of each.⁵

Twenty grams a day is 140 g across a week, which is one ordinary serving.⁵ That is the scale being described here, not a fish-heavy diet.

A wider sweep of 298 such studies put moderate-quality evidence behind each extra 100 g a day: death from any cause at 0.92, cardiovascular death at 0.75, coronary heart disease at 0.88, heart attack at 0.75.⁷ For most cancers there was no association worth reporting.⁷

These are observations, not assignments. Fish eaters may differ from everyone else in many ways, and part of what a serving of fish does may be to push something else off the plate. That is why these numbers describe who eats fish, not what fish does.

Fatty fish, not fish

One analysis separated the two kinds, and it narrows the claim usefully. Across 19 reports, the fatty-fish studies covered 1,320,596 person-years of follow-up — one person tracked for one year — and the lean-fish studies 937,362.⁶ Fatty fish went with less coronary heart disease at 0.92, fewer coronary deaths at 0.83 and lower overall mortality at 0.97.⁶ Lean fish went with none of them.⁶

So the benefit tracks salmon, mackerel and sardines, not white fish like cod. That is the strongest hint here that the omega-3 itself does part of the work, which makes the failure of the capsules harder to explain rather than easier.

The irregular-heartbeat signal

The one consistent harm in the capsule trials is atrial fibrillation, a fast and irregular heartbeat. The 15-trial analysis put it at 1.25, sitting alongside that analysis's modest benefits.²

A recent review explains why food and capsules point opposite ways here. Across trials in 83,112 people, omega-3 raised the relative risk by 24%. In absolute terms that is a move from 3.3% of people to 4.0%.⁴ It tracked the dose closely: about 12% higher at roughly 1,000 mg a day, which is one gram, and about 50% higher at 1,800 to 4,000 mg a day.⁴

Observational data went the other way. Higher blood levels and higher dietary intake generally came with lower risk, the largest reduction being about 12% at roughly 650 mg a day of dietary omega-3.⁴

Change in atrial fibrillation risk with omega-3% change in risk

Dietary intake, about 650 mg a day-12%

Capsule trials, about 1,000 mg a day+12%

Capsule trials, all doses pooled+24%

Capsule trials, 1,800 to 4,000 mg a day+50%

The proposed mechanism is vagal tone, the brake the nervous system holds on heart rate. Omega-3 raises it, and more raises it more. Gentle vagal stimulation lowers the risk of atrial fibrillation in experiments, and strong stimulation raises it.⁴ Same molecule, opposite directions, separated by how much arrives at once.

What remains unknown

The comparison in the title has never actually been run. No trial here assigned people to eat fish against taking a matched capsule, and the Cochrane review says outright that evidence on eating fish is scarce.¹ The fish side of this argument is observational, and it will probably stay that way.

The prescription-EPA result rests on breakdowns inside combined trials rather than on a trial built to compare one oil against another. That is a weaker footing than the size of the effect suggests.^2,3

And nobody has established what makes a fish meal work. The split between fatty and lean fish points at the fatty acids, but a fillet also carries protein, selenium, iodine and vitamin D, and it displaces whatever would otherwise have been on the plate.⁶ Nothing here separates those strands.