Table 1.

Key points of methodology used in all three cohorts

	KORAFF4	LLS	TwinsUK
Dietary data	Usual dietary intake in g/day (methodology described as in [20]. Repeated 24 h food lists (246 items) and FFQ as adjusting variable	FFQ (218 items), calculated in g/day	FFQ (131 items), calculated in g/day
Methylation data	Infinium MethylationEPIC BeadCHip (~ 850 k loci); preprocessing with the package minfi	Illumina HumanMethylation450 array (~ 450 k loci); quality control using MethylAid	Illumina HumanMethylation450 array (~ 450 k loci); preprocessing with ENmix and minfi
Statistical model	Linear multivariable regression with technical covariates and food intake residual as exposure and methylation beta values as outcome	Linear multivariable regression with technical covariates and food intake residual as exposure and methylation beta values as outcome	Mixed-model with cohort-specific random-effects and food intake residual as exposure and methylation beta values as outcome