We performed three analyses: Meta-analysis of single SNP associations with birth weight: we performed fixed-effects inverse variance meta-analyses of available studies, as described above, for each of the 7 loci.Ethnicity-specific regional analysis: we performed fixed effects inverse variance meta-analyses for SNPs within the 500 kb surrounding the 7 index SNPs in an ethnicity-specific manner for n = 2,135 East/Southeast Asian and n = 6,315 African-American samples. We plotted the association results against chromosomal position using LocusZoom (http://csg.sph.umich.edu/locuszoom/).Combined genotype risk score analysis: to assess the associations between birth weight and the 7 confirmed loci in combination, we created a risk allele count (RAC) by summing the birth-weight lowering alleles at each SNP. We performed this analysis in 7 non-European studies in which 6 to 7 SNPs were available (combined n = 11,014) and one representative European Stage 2 study (NFBC1986, n = 4647). If a SNP was missing, all individuals were assigned a value of 2*frequency (HapMap, ethnicity-specific) of the birth weight-lowering allele. We performed linear regression of birth weight z score against RAC (additive model), with sex and gestational age
