Jackson Heart Study, the largest single-site African American cohort in the US (4,568 participants at baseline; 3,564 free of prevalent diabetes for the prospective analysis), egg intake by population-specific FFQ, mean 7.3y follow-up, 531 incident T2D cases. Cross-sectionally, egg consumption was positively associated with prevalent diabetes (prevalence ratio 1.52, 95% CI 1.17-1.97, for >=5 eggs/week vs reference, p-trend=0.0024). But in the prospective analysis the authors’ main target endpoint, the association with incident diabetes was null: HR 1.17 (95% CI 0.81-1.70) for >=5/week, p-trend=0.22. Useful complication for the “US cohorts uniformly show elevated risk” narrative - here the prospective (least confounded by reverse causation) result is null even within a US cohort. relevance_note: adds a demographically distinct US cohort (African American) and shows the US-elevated pattern is not perfectly uniform even within the US once restricted to prospective incident-disease analysis.