Where both birth weight and birth length were available, we used linear regression to assess the association between birth weight z score and the 7 confirmed birth weight SNPs, with sex, gestational age (where available) and birth length as covariables. In the same set of samples, we again performed linear regression to assess the association between birth weight z score and SNP, with only sex and gestational age (where available) as covariables to allow direct comparison of analyses with and without adjustment for birth length. Meta-analysis was performed as above.
