Being exposed to alcohol during pregnancy negatively affects children''s test scores later in life, according to research by Stephanie von Hinke Kessler Scholder and colleagues, published in the May 2014 issue of the Economic Journal.

Their study notes that a major problem in estimating the effects of prenatal alcohol consumption on children is that mothers who choose to drink may be systematically different from mothers who choose not to drink. Indeed, several studies show positive ''social gradients'' in alcohol consumption: higher-educated women are more likely to drink in general and to drink (moderately) during pregnancy. As higher maternal education is associated with better children''s test scores, this can mask any negative effect that alcohol exposure may have.

This clearly shows the difficulty in simply comparing the outcomes of children whose mothers drink with the outcomes of children whose mothers do not drink. In addition, as these mothers may differ in other ways unobserved by the researcher, it is not possible simply to control for a set of background factors and assume that this accounts for all relevant confounders.

Furthermore, as it is obviously unethical to design a randomised controlled trial, it is necessary to rely on alternative ways to estimate and identify both the short- and longer-term effects of prenatal alcohol exposure.

The new study exploits the fact that genetic markers are randomly allocated, and uses a carefully validated (maternal) genetic variant that is associated with decreased alcohol intake to estimate the effect on children''s test scores later in life. This approach is also known as ''Mendelian randomisation'', essentially comparing test scores of children whose mothers carry the specific genetic variant with those whose mothers do not. With this method, any differences in test scores are due to foetal alcohol exposure, not to how affluent or educated their mothers are.

The researchers confirm their previous findings (Zuccolo et al, 2009) showing that a genetic variant in a maternal alcohol-metabolising gene (ADH1B) is negatively related to prenatal alcohol exposure, and unrelated to any of the background characteristics that the new study shows to be associated with prenatal drinking. Using this Mendelian randomisation approach, the authors find negative effects of prenatal alcohol exposure on children''s educational achievement, based on data from up to 4,000 mothers and their children in the ''Children of the 90s'' study at the University of Bristol.

The US Surgeon General first published a report on drinking during pregnancy in 1981, drawing attention to the link between prenatal alcohol consumption and birth defects. The detrimental effects of excessive drinking during pregnancy are now well-known: it potentially leads to a pattern of mental and physical defects known as ''foetal alcohol spectrum disorder''. The effects of low-to-moderate drinking, however, are less conclusive, and there is no consensus as to what level of exposure is toxic to the foetus.

Most countries, including, for example, the United States, advise women to abstain from alcohol during pregnancy. Other countries, such as the UK, recommend women not to drink in the first three months of pregnancy, and if they choose to drink, not to exceed one or two units once or twice a week: according to the National Institute for Health and Care Excellence, ''at this low level, there is no evidence of any harm to the unborn baby''.

These conflicting recommendations arise from inconsistent findings in observational studies of the correlation between alcohol exposure and child development, including physical and mental health, cognitive and non-cognitive outcomes.

Studies that attempt to account for the fact that mothers who drink during pregnancy may be systematically different compared with mothers who do not drink (in both observable and unobservable ways) tend to find that prenatal alcohol exposure is harmful to children. As randomised experiments are obviously unethical and infeasible, these studies are crucial in contributing to knowledge and understanding of the effects of prenatal alcohol exposure on children''s later life development.

Alcohol Exposure in Utero and Child Academic Achievement by Stephanie von Hinke Kessler Scholder, George Wehby, Sarah Lewis and Luisa Zuccolo is published in the Conference issue of the Economic Journal. This paper is available free.

A non-synonymous variant in ADH1B is strongly associated with prenatal alcohol use in a European sample of pregnant women by Luisa Zuccolo et al was published in 2009 in Human Molecular Genetics.

Stephanie von Hinke Kessler Scholder is at the University of York. George Wehby is at the University of Iowa. Sarah Lewis and Luisa Zuccolo are at the University of Bristol.