known_biases: Cohort of ~22,000 US male physicians with egg intake captured by a single crude abbreviated food-frequency question at baseline — every finding built on it shares a highly selected healthy, high-SES, male-only population (limited generalizability), coarse one-time exposure measurement (large non-differential misclassification / regression dilution), and residual confounding by baseline health behaviors, so all PHS egg papers rise and fall together.