Pre-conception and prenatal alcohol exposure from mothers and fathers drinking and head circumference: results from the Norwegian Mother-Child Study (MoBa)

Although microcephaly is a feature of Fetal Alcohol Syndrome, it is currently unknown whether low-to-moderate prenatal alcohol exposure affects head circumference. Small magnitude associations reported in observational studies are likely to be misleading due to confounding and misclassification biases. Alternative analytical approaches such as the use of family negative controls (e.g. comparing the effects of maternal and paternal exposure) could help disentangle causal effects. We investigated the association of maternal and paternal alcohol drinking before and early in pregnancy with infant head circumference, using data from 68,244 mother-father-offspring trios from the Norwegian Mother and Child Cohort Study (MoBa) (1999–2009). In analyses adjusted for potential confounders, we found no consistent pattern of association between maternal or paternal alcohol intake before or during pregnancy and offspring head circumference modelled as a continuous outcome. However, we found higher odds of microcephaly at birth for higher paternal, but not maternal, alcohol consumption before pregnancy, and similar but weaker effect estimates for first trimester drinking. Associations with paternal drinking before pregnancy were unexpected and should be regarded as hypothesis generating, until independently replicated, although potentially important given the absence of guidelines on safe drinking levels for men in couples trying for a pregnancy.


Description of confounders-alcohol associations
Confounders-alcoho l associations were similar or very similar for mothers and fathers (with a few exceptions), with maternal characteristics more strongly associated with maternal drinking and vice versa, and directions of association as expected (Supplementary Information, Tables S7-S10, Figures S1 and S2).
For maternal drinking before pregnancy, the shape of association was often curvilinea r (U, J or inverse U and J), so often non-drinkers and heavier drinkers were more like each other, and differed from light-to-moderate drinkers (drinking 1-2 drinks per occasion), in particular in terms of socioeconomic factors (Supplementary Information, Table S7). The latter was the category most likely to be taking folic acid supplements, to have planned the pregnancy, to be recruited earlier, to have taller partners with lower BMI, and both themselves and their partners were more likely to be older, better educated, and earning more. Where confoundersalcohol relationship was monotonic, increasing alcohol use was associated with no previous pregnancies, Norwegian ethnic origin, no financia l strains, increased smoking and drug taking in pregnancy (maternal and paternal).
Paternal drinking before pregnancy was less obviously associated with maternal prepregnancy BMI and gross income, or paternal height, compared to maternal drinking (Supplementary Information, Tables S7 and S8, Figure S1).
Most women quit drinking in the first trimester of pregnancy or markedly reduced their alcohol intake (Supplementary Information,Tables S4 and S9). The biggest differences compared to before pregnancy were in the patterning of maternal drinking in association with planned pregnancy, which went from U-shaped to monotonic), and with maternal smoking in pregnancy, folic acid and financia l strain, which went from showing some to no association (Supplementary Information,Tables S7 and S9,Figures S1 and S2).
Paternal drinking patterns of association were very stable before-during pregnancy (Supplementary Information, Tables S8 Vs S10). The patterning of alcohol drinking duringpregnancy in association with most confounders was very similar for mothers and fathers, with the exception of maternal ethnic background and education, and maternal and paternal age and smoking during pregnancy, and paternal age (Supplementary Information,F igure S2).

Description of confounders-outcomes associations
There was evidence of association with offspring head circumference at birth for most of the potential confounders identified, the only exceptions being financia l strain and planned pregnancy (Supplementary Information, Tab les S11 Vs S12). The strength and direction of associations were similar for the outcome at 3 months, except for maternal and pate rnal education and gross income, which went from an inverse U shape, with babies of parents at both ends of the education and income spectrum having smaller head circumference s compared to babies born to parents in the middle of the spectrum, to a positive trend, with increasing head size for increasing education levels and wealth (Supplementary Information, Tab les S11 Vs S12), possibly suggesting catch-up growth for children of richer, better educated parents. The other difference was for folic acid use, which went from a negative association with head circumference at birth to a borderline positive association at 3 months. Breech fetal presentation predicts larger head circumference at birth but the correlation is less strong for head circumference measured 3 months, which was the motivation for repeating all analyses with the outcome measured at 3 months, to overcome at least some of the bias potentially caused by this association.   Te s t for tre n d maternal an d p aternal alcoh ol in take in full mod el: p =0.343 a n d p =0.910