Individual-participant-data pooling of 6 US prospective cohorts (ARIC, CARDIA, CHS, Framingham Offspring Study, Jackson Heart Study, MESA — the NHLBI-funded Lifetime Risk Pooling Project), 29,615 participants (mean age 51.6, 44.9% men, 31.1% Black), data collected 1985-2016, median follow-up 17.5 years (max 31.3), 5,400 incident CVD events and 6,132 deaths. Each +300mg/day dietary cholesterol was associated with higher incident CVD (adjusted HR 1.17, 1.09-1.26) and all-cause mortality (1.18, 1.10-1.26) in a dose-response, monotonic way; each +0.5 egg/day was likewise associated with higher CVD (1.06, 1.03-1.10) and mortality (1.08, 1.04-1.11) — but the egg association became non-significant once adjusted for dietary cholesterol, i.e. cholesterol content, not egg per se, is the paper’s proposed driver. relevance_note: The largest, most-cited harm-finding Western pooled cohort on hard CVD/mortality endpoints, and the main empirical anchor for the “dietary cholesterol is the operative exposure” mechanistic claim.
Methodology
Individual-participant-data pooling of 6 US prospective cohorts (ARIC, CARDIA, CHS, Framingham Offspring, Jackson Heart Study, MESA — the NHLBI Lifetime Risk Pooling Project; D-1), 29,615 participants (mean age 51.6; 44.9% men; 31.1% Black), data 1985-2016, median follow-up 17.5 years (max 31.3); 5,400 incident CVD events, 6,132 deaths. Self-reported diet (mostly a single baseline FFQ/diet-history per cohort) was harmonized across instruments via a standardized protocol. Exposures: dietary cholesterol (mg/day) and egg consumption (number/day). Outcomes: incident CVD (composite of fatal/nonfatal CHD, stroke, heart failure, other CVD death) and all-cause mortality, reported as HR and absolute risk difference over follow-up, adjusted for demographic, socioeconomic, and behavioral factors; a further model adds dietary cholesterol to the egg model. Observational design with mostly single-baseline diet (regression-dilution and residual-confounding vulnerability).
Results
O-61 - Eggs are a major dietary source of cholesterol (~185-200 mg per large egg)
An established food-composition fact (not resting on this cohort). Load-bearing for the cholesterol-mediation reasoning: because eggs are a dominant swing component of dietary cholesterol, an egg-CVD association is expected under the hypothesis that dietary cholesterol is the operative exposure.
Link to original
O-58 - Each +300 mg-day dietary cholesterol associated with higher CVD (HR 1.17) and mortality (1.18) across 6 US cohorts
The dietary-cholesterol exposure result. Associations were monotonic with no significant nonlinear terms (P for nonlinear 0.19-0.83), i.e. no apparent threshold.
Link to originalMethodology
Individual-participant-data pooling of 6 US prospective cohorts (ARIC, CARDIA, CHS, Framingham Offspring, Jackson Heart Study, MESA — the NHLBI Lifetime Risk Pooling Project; D-1), 29,615 participants (mean age 51.6; 44.9% men; 31.1% Black), data 1985-2016, median follow-up 17.5 years (max 31.3); 5,400 incident CVD events, 6,132 deaths. Self-reported diet (mostly a single baseline FFQ/diet-history per cohort) was harmonized across instruments via a standardized protocol. Exposures: dietary cholesterol (mg/day) and egg consumption (number/day). Outcomes: incident CVD (composite of fatal/nonfatal CHD, stroke, heart failure, other CVD death) and all-cause mortality, reported as HR and absolute risk difference over follow-up, adjusted for demographic, socioeconomic, and behavioral factors; a further model adds dietary cholesterol to the egg model. Observational design with mostly single-baseline diet (regression-dilution and residual-confounding vulnerability).
Link to original
O-59 - Each +half-egg-day associated with higher CVD (HR 1.06) and mortality (1.08) before cholesterol adjustment (6 US cohorts)
The raw (cholesterol-unadjusted) egg-consumption association — a positive, dose-responsive harm signal on hard CVD and mortality endpoints.
Link to original
O-60 - Egg-CVD and egg-mortality associations become non-significant after adjusting for dietary cholesterol (6 US cohorts)
The key attenuation result: once dietary cholesterol is in the model, the egg association vanishes — distinct from the raw egg association (the same data, different adjustment set).
Link to original
A-21 - Egg association vanishing on cholesterol adjustment implies dietary cholesterol is the operative exposure
Eggs are a major contributor to dietary cholesterol (generally-known fact), so egg intake and dietary-cholesterol intake are strongly positively correlated within these cohorts. The raw egg association (HR 1.06 CVD, 1.08 mortality; O-59) is about the magnitude expected if egg’s effect operated through its cholesterol load. When dietary cholesterol is entered into the model, the egg coefficient attenuates to null (0.99 CVD, 1.03 mortality; O-60) while dietary cholesterol retains a significant, monotonic dose-response (1.17 CVD, 1.18 mortality; O-58). In regression terms, conditioning on the proposed mediator/common driver (dietary cholesterol) removes the egg association — the signature of dietary cholesterol accounting for the egg-outcome link rather than of an independent egg effect, supporting H-26. Caveat: because eggs are the dominant swing component of dietary cholesterol in this population, egg intake and dietary cholesterol are near-collinear, so the model cannot cleanly separate ‘egg-via-cholesterol’ from cholesterol contributed by other foods; residual confounding by overall diet/lifestyle (single-baseline diet in most cohorts) could also inflate the dietary-cholesterol estimate. The attenuation is thus consistent with, but does not prove, dietary cholesterol as the sole operative pathway.
Validity (step 6): approved / checked. Reconstructed step: in a regression, conditioning on X removes the coefficient of a correlated Y when X screens off Y from the outcome; here entering dietary cholesterol nulls the egg coefficient (1.06→0.99, 1.08→1.03) while dietary cholesterol keeps a monotonic dose-response (1.17, 1.18), which is the statistical signature of dietary cholesterol accounting for the egg–outcome link. The decisive undercutting defeater is collinearity-symmetry: with two near-collinear exposures, adjusting for either can null the other regardless of which is causal, so the vanishing alone cannot orient the arrow — and this is exactly the caveat the statement carries. What keeps the step alive as an evidential (not proof) claim is the asymmetry that dietary cholesterol has independent variance from non-egg sources and retains significance while egg does not; that mild asymmetry supports (does not establish) cholesterol-content as operative. Because the conclusion is already stated in the weakened form (“indicates / supporting… does not prove,” with the collinearity and residual-confounding caveats explicit), the inferential step holds as written — no correction needed. Had the statement claimed the adjustment “proves” cholesterol is the exposure, it would have been corrected down to this evidential version; it already is that version.
Link to original
Discussion
H-26 - Dietary cholesterol content, not egg per se, is the operative exposure behind the egg-CVD association
The paper’s proposed mechanism, motivated by the egg association vanishing once dietary cholesterol is adjusted for while dietary cholesterol retains a monotonic dose-response.
Link to original
H-27 - Higher dietary cholesterol - egg intake causally raises CVD and mortality risk (dose-response)
The harm-direction candidate answer on hard CVD and mortality endpoints — the Western harm anchor contrasting with null-finding repeated-measure cohorts.
Link to original