known_biases: The same worldwide prospective cohorts are repeatedly re-pooled across egg-type-2-diabetes dose-response meta-analyses (e.g. Wallin 2016’s 12-cohort set and Drouin-Chartier 2020’s 16-cohort set), so region-stratified pooled estimates reported by different papers are NOT independent confirmations of one another - they share cohort membership, single/repeated-FFQ measurement error, and residual dietary-pattern (healthy/unhealthy-eater) confounding. The US stratum is dominated by the Harvard NHS/NHS II/HPFS cohorts; the non-US stratum by European and Japanese cohorts. D-25