Analyses were restricted to women of white ethnic origin. Associations between the SNPs and alcohol intake variables were tested in univariable models assuming additive effects of rare alleles. For SNPs with MAF < 0.05, rare homozygotes and heterozygotes were analyzed together and a dominant effect was assumed. Logistic regression was used for binary outcomes and ordinal logistic regression for ordered categorical variables (e.g. average weekly drinking before pregnancy—<1 unit/week, 1–6 units/week, ≥7 units/week).
