The egg has been convicted and acquitted several times in nutrition science. The largest and most careful pass so far is the 2020 BMJ paper combining three US cohorts with an updated meta-analysis, and its answer is a null result — which, in this field, is itself a hard-won thing.
The cohorts
Three cohorts, all American, all long: the Nurses’ Health Study (1980–2012), NHS II (1991–2013) and the Health Professionals’ Follow-Up Study (1986–2012). The analyses covered 83,349 women in the first, 90,214 in the second and 42,055 men in the third, all free of cardiovascular disease, type 2 diabetes and cancer at baseline. That is 215,618 people followed for up to 32 years — 5.54 million person-years — and 14,806 incident cardiovascular events, defined as non-fatal myocardial infarction, fatal coronary heart disease and stroke.
That last number is what separates a cohort from a survey: 14,806 events is enough to detect a difference that matters.
The result, and the confounding that came with it
Eating at least one egg a day compared with less than one a month gave a hazard ratio of 0.93 (95% CI 0.82 to 1.05) after adjusting for the lifestyle and dietary factors associated with egg eating.
The interval is the story. It runs from a 7% lower risk to a 5% higher risk, which means the study cannot tell you which of those is closer to the truth. It also told you something useful about the confounders: people who ate the most eggs had a higher body mass index, were less likely to be on statins and ate more red meat — so an unadjusted comparison would have been measuring diet, not eggs.
The updated meta-analysis pooled 33 risk estimates, 1,720,108 participants and 139,195 cardiovascular events: one additional egg a day, relative risk 0.98 (0.93 to 1.03), I²=62.3%. Coronary heart disease read 0.96 (0.91 to 1.03, I²=38.2%), stroke 0.99 (0.91 to 1.07, I²=71.5%). One place the association went the other way: Asian cohorts, 0.92 (0.85 to 0.99, I²=44.8%) — with the interaction test sitting at P=0.07, which is a hint, not a verdict.
And the meta-analysis that did find something
A separate 2021 meta-analysis, pooling 39 studies and nearly 2 million people with 85,053 coronary events, 25,103 strokes, 7,536 heart failures and 147,124 cardiovascular events in total, read slightly lower for cardiovascular disease: four eggs a week gave a summary relative risk of 0.95 (0.90 to 1.00) against no consumption, and up to one egg a day 0.94 (0.89 to 0.99).
It also produced the one signal in this literature that points the wrong way: for heart failure, one egg a day came with a summary relative risk of 1.15 (1.02 to 1.30) against no consumption, higher still at higher intakes, on six datasets from four studies. The authors rated the strength of evidence low for every outcome except stroke, where it was moderate and described no risk. Their conclusion is the careful one: no conclusive evidence on the role of the egg in cardiovascular risk, in either direction.
What these numbers do not show
They do not show that eggs are protective, because nothing here points that way reliably. They do not show eggs are harmful either: a hazard ratio of 0.93 with that interval is what “no measurable effect at this scale” looks like, not a finding of safety in every population. And they cannot say anything about the person who already has high cholesterol, because the cohorts enrolled people without diagnosed cardiovascular disease or diabetes at baseline — the question most people asking about eggs are actually asking was not the question asked here.
One more limit worth stating plainly: an egg in a cohort is a food someone ate often, not an intervention someone was assigned. Free-range versus cage, size of breakfast alongside it, how the person who eats eggs differs from the person who doesn’t — all of that travels with the variable.
