Prospective cohort studies show positive egg–cardiovascular disease associations in US populations but null associations in European and Asian populations.

The claim bundles a regional-heterogeneity premise with a specific three-part directional pattern (US positive; Europe null; Asia null). The heterogeneity premise is captured by the subclaim "the association between egg consumption and cardiovascular disease risk varies significantly by geographic region," which is supported: both principal meta-analyses find a region-dependent gradient (Circulation 2022 P-for-interaction significant; BMJ 2020 P=0.07, marginal), with US estimates most adverse and Asian estimates least adverse.

Where the claim founders is the specific labels. Two credible meta-analyses give materially different regional pictures on two of the three components:

1) Drouin-Chartier et al., BMJ 2020 (m513): the most direct evidence on CVD *incidence*, ~1.72M participants, 33 risk estimates. Per one egg/day: US cohorts RR 1.01 (0.96–1.06, null), European cohorts 1.05 (0.92–1.19, null), Asian cohorts 0.92 (0.85–0.99, significant inverse/protective). This contradicts "US positive" (US is null here) and "Asia null" (Asia is significantly protective, not null). Only "Europe null" is affirmed.

2) Zhuang/Zhao et al., Circulation 2022 (dietary cholesterol, serum cholesterol, egg consumption and mortality): per 50 g/day, US cohorts RR 1.08 (1.02–1.14, positive), European cohorts borderline positive, Asian cohorts no association. This affirms "US positive" and "Asia null" but not "Europe null" (Europe leans borderline positive).

Weighing: neither leading analysis affirms the full three-part wording. The BMJ analysis, which measures CVD incidence directly and is the more relevant to a claim phrased about "cardiovascular disease," affirms only one of the three elements; the Circulation analysis, focused on mortality, affirms two but not the European element. The differences are not mere noise: the analyses use different exposure increments (per egg vs per 50 g), different endpoints (incidence vs mortality), and partially different cohort sets, so they represent genuine analytic disagreement rather than a single settled result.

Status: contested rather than supported, because credible evidence stands on multiple sides of the claim's specific directional pattern, and the single most direct CVD-incidence meta-analysis contradicts two of its three regional characterizations. The status is not "contradicted" because the load-bearing heterogeneity premise is real and one major analysis does affirm the US-positive/Asia-null pattern. A related graph claim, "positive egg–CVD associations are concentrated among people with diabetes" (supported), offers a partial mechanism for why some US cohorts show adverse estimates, but does not establish the region-specific significance labels.

No single claim_credence is given: the proposition is a composite whose three regional parts are supported to different degrees and pull in different directions, so one number would be false precision. Confidence 0.7 reflects fairly firm judgment that the specific pattern is disputed between credible sources, tempered by the defensible alternative reading under which the agreed regional gradient makes the claim "supported with caveats." The verdict would move toward supported if newer harmonized pooled analyses replicate a positive-US / null-elsewhere pattern for CVD incidence, and toward contradicted if the inverse Asian and null US findings are further confirmed.