Can the odd glass of wine help the unborn baby?

Occasional drinking in pregnancy does no harm, report the Daily Mail, the Herald, the Guardian and the Daily Telegraph today.
Some claim - but not in the intro, or the headline - that occasional drinking is actually good for babies. The Daily Express goes for broke, headlining its story “Why a glass of wine while pregnant is good for baby”.
But Sarah Bosely in The Guardian does not even mention this possibility, while Jenny Hope in the Mail touches on it, but only in paragraph nine, and Rebecca Smith in the Daily Telegraph casts doubt on it by pointing out that the apparent link becomes weaker when confounding factors such as a mother’s educational level are taken into account.
Many websites on local newspapers take the opposite line, running an agency story that leads on official advice to avoid alcohol entirely in pregnancy. The Belfast Telegraph’s story, for example, begins: “Women were advised that official guidance to avoid alcohol in pregnancy remains in place after experts said drinking one or two units a week does no harm a child’s development”. That’s what’s called backing into a story.
The health writers have mostly done their best with a study that appeared in the Journal of Epidemiology and Community Health (1). The only reasonable conclusion you can draw from it is that no level of drinking – abstaining in pregnancy, light, moderate, or even (on most measures) binge, makes any statistically significant difference to children’s behaviour at the age of five.  
The study is large but heavily confounded. The group of women who claim to have drunk “lightly” in pregnancy (not more than 1-2 units per week or per occasion) is in a different class, literally, from all the rest.
These women are four times more likely to earn over £52,000 a year than the non-drinkers, and twice as likely to do so as binge drinkers. Only 3.2 per cent have no educational qualifications at all, compared to 14.0 per cent of non-drinkers and 9.3 per cent of binge drinkers. A third of them are managerial or professional, against 12 per cent of non-drinkers and 20 per cent of binge drinkers.
Of course the authors, led by Dr Yvonne Kelly of University College London, have attempted to correct for the class, income, and educational differences. After correction for these factors and for birth weight, the difference in problem behaviour between the children of light drinkers and those who did not drink at all in pregnancy fell from 33 per cent to 9 per cent, and was no longer statistically significant.
When corrected for all identified confounding factors, the links between drinking in mothers and problem behaviour in their children do not reach statistical significance even for binge drinkers, save in one single category in boys (emotional symptoms). The best the authors can argue is that there is a trend of increasing behavioural difficulties with increasing alcohol intake, though it’s hardly compelling, and the children of women who have never drunk in their lives buck this trend.
Boys of teetotal women, for example, are almost exactly as likely to have conduct problems as boys of binge drinkers. There is some biological basis for believing that drinking and heart disease may follow a U-shaped curve but in this instance it is very hard to think of one. 
Only if we take it on trust that the corrections fully allow for the differences between the mothers do the conclusions have any meaning at all, and it is more likely that any remaining links are to do with incomplete correction for confounding factors than they are to do with alcohol.
The study also assumes that women told the truth about their drinking patterns, and not just what they were expected to say. When not pregnant, fully 94 per cent of the mothers were drinkers, so how plausible is it to claim that during their pregnancies only 34 per cent drank anything at all? Drinking in the first trimester is supposed to be the riskiest, yet most women do not even discover they are pregnant until well into the first trimester.
So if we can draw any conclusion, it is that the study showed no evidence of harm from drinking occasionally or regularly, and scant evidence of harm from drinking hard.
The claim that the occasional drink can be good for babies came from measurements of cognitive ability that show quite large differences uncorrected, but attenuate sharply when confounding factors such as class, education and wealth are taken into account.
In boys, those from mothers who drank at all levels up to and including binge do better on all tests than those whose mothers didn’t drink at all while pregnant – an odd finding. But after correction for all confounding factors only two remained (just) statistically significant – tests of vocabulary and recognising the similarities between pictures showed higher scores in boys from light-drinking mothers than those in other categories. In girls, there were no statistically significant differences in any test, for any level of drinking.
These are wafer-thin findings, as the authors seem to acknowledge. “It is likely that social circumstances to a large extent are responsible for the relatively low rates of subsequent behavioural differences and the cognitive advantage in children whose mothers were light drinkers” they say. That’s honest.
So it’s not drink, but social class, that makes the difference. None of the reports quite captured this, though most did better than the Express or the Daily Record, whose intro ran: “Mums-to-be who have a glass or two of wine a week have brighter, better-behaved babies than pregnant women who stay teetotal.”


(1) Kelly YJ, Sacker A, Gray R et al, J Epidemiol Community Health (2010) (online|)