In this light, our initial validation of a candidate genetic instrument that predicts lower risk of binge drinking and higher likelihood of giving up alcohol early in gestation, and allows causal inference, takes on a significant public health importance. Implications of the present work are that several phenotypic outcomes could be studied in association with prenatal alcohol exposure, provided maternal ADH1B genotype is available. For example, we have estimated through simulations that in an instrumental-variable analysis using rs1229984 as the genetic instrument, approximately 6000 mother–child pairs would be required to find evidence of a causal association between alcohol consumption and a school-based performance score at age 11 (Key Stage 2—http://en.wikipedia.org/wiki/Key_Stage_2) at a 5% significance level, assuming a small change of as little as 0.20 standard deviations in the log-transformed outcome comparing A-allele carriers to non-carriers.
