Associations with measures of maternal alcohol consumption might reflect teratogenic effects related to drinking during pregnancy rather than to the maximum drinks phenotype per se. To address this possibility, we entered a dummy variable coding for weekly alcohol use during pregnancy and the log of the maximum drinks consumed during pregnancy in separate regression analyses. We first examined effects of each measure of drinking during pregnancy by itself to assess the sensitivity of the measures. The (log of) the maximum number of drinks consumed during pregnancy was associated with virtually all outcomes in the younger cohort. We subsequently examined the replicated associations with maternal maximum consumption from our primary analyses – conduct disorder, ODD in females, any disruptive disorder, and all substance-related phenotypes – with each measure of drinking during pregnancy as a covariate. If drinking during pregnancy accounts for associations between lifetime maximum consumption and these outcome measures, maximum consumption's effects should become nonsignificant when adjusted for effects of drinking during pregnancy. This was not the case; all of the significant effects of maximum consumption remained significant when adjusted for drinking during pregnancy. Indeed, adjusted regression coefficients were virtually identical to unadjusted coefficients.
