To assess whether adjustment for gestational age or measurement/recall bias of birth weight influenced the associations between each of the 21 SNPs and birth weight, we repeated the fixed-effects inverse-variance meta-analyses of the European results in three different sub-sets of studies: (i) studies with birth weight collected by any method that adjusted for gestational age (n = 35); (ii) studies with measured or medical record of birth weight that adjusted for gestational age only where available (n = 26); (iii) studies with measured or medical record of birth weight which also adjusted for gestational age (n = 24). We compared the effect size estimates between each of these three meta-analyses and the overall meta-analysis result (Supplementary Table 4).
