known_biases: Aggregate of 33 prospective-cohort risk estimates (1,720,108 participants, 139,195 CVD events) screened and pooled by Drouin-Chartier 2020, spanning US, European, and Asian cohorts with heterogeneous FFQ instruments, adjustment sets, and one-time-versus-repeated diet assessment (overall I2=62.3%, region the main heterogeneity source). Findings built on it share the pooled literature’s residual dietary-pattern/healthy-user confounding and cross-cohort heterogeneity, and partially overlap the Harvard cohorts (D-7) and the Lifetime Risk Pooling Project (D-1), which are among the pooled estimates. Minted at step 3 as an external base step 2 did not list (step 2 recorded only D-7 for this source, treating the meta-analysis as a discovery hub).