known_biases: Repeated FFQ in health-professional cohorts with strong healthy-user correlations and shared covariate-updating pipeline; correlated residual confounding that tends to attenuate single-food associations toward null in these Western analyses.