A study published in PLoS ONE suggested drinking one or two glasses of wine a week during pregnancy can have a negative impact on a child’s IQ. This Before the Headlines analysis accompanied a Roundup which can be viewed here.
Title, Date of Publication & Journal
Fetal alcohol exposure and IQ at age 8: evidence from a population based birth-cohort study. Published online 10pm UK time Wednesday 14 November 2012 in PLOS One.
Claim supported by evidence?
The paper supports the claim that moderate maternal alcohol use during pregnancy may have an adverse effect on offspring cognition.
Uses genetic variation in alcohol metabolism to look at in-utero exposure.
It is reasonable to infer a relationship between alcohol consumption and IQ.
Relationship found between alcohol metabolism and IQ in moderate drinkers.
As would be expected no relationship found amongst abstainers.
Genetic variation in alcohol metabolism were related to lower IQ at age 8
The study looks to be well conducted and reliable.
The size of the effect is -1.8 IQ points for moderate drinking in mothers who are poor alcohol metabolisers. The normal IQ range spans 20 points (scores of 90-110). So the effect is limited in size, but large enough to have noticeable effects at a population level.
The paper does not comment on whether it is a small effect on all children, or an occasional large effect on a few children.
It is unclear what level of drinking in normal metabolisers corresponds to the moderate drinking in poor alcohol metabolisers.
The study excludes heavy drinking, which is already known to have harmful effects.
The study is a large (4167 children in the analysis) prospective population based study. The paper acknowledges that this is not a randomized controlled trial, which could be unethical and difficult to conduct.
The method relies on good principles, and balances for other (confounding) factors – smoking, diet, other lifestyle factors.
Alcohol consumption was self-reported and it is possible some mothers consumed larger amounts of alcohol than they were prepared to report. But because only the lowest reporting groups were used, this is unlikely to have caused gross misrepresentations in the data from misreporting by heavier drinkers.
No adjustment has been made for selection of the four genes from a pool of 10 genes (backwards selection stage). After such adjustment these results will most likely still hold good, but with less extreme p-values. This means the size of the effect could be overestimated by selection. Given that there are a large number of subjects and up to 10 genes, however, the result will probably still be significant. These considerations are unlikely to affect the overall conclusions of the paper.
Prospective population based study: a study where the subjects are recruited at the start and followed up later (in this case, mothers recruited to the surviving at pregnancy and the children followed up 8 years later). This is a powerful study design.