Alcohol, pregnancy and the precautionary principle

As of the 23rd May 2022 this website is archived and will receive no further updates.

was produced by the Winton programme for the public understanding of risk based in the Statistical Laboratory in the University of Cambridge. The aim was to help improve the way that uncertainty and risk are discussed in society, and show how probability and statistics can be both useful and entertaining.

Many of the animations were produced using Flash and will no longer work.

A recent study showed no evidence of an adverse effect of moderate consumption of alcohol during pregnancy and subsequent child development. But the Department of Health has commented "After assessing the available evidence, we cannot say with confidence that drinking during pregnancy is safe and will not harm your baby. Therefore, as a precautionary measure, our advice to pregnant women and women trying to conceive is to avoid alcohol."
Is this still a tenable position?

The precautionary principle is intended to cover situations where there is a possibility of harm but little evidence exists, and so to be on the safe side a recommendation is made to avoid exposure to the possible hazard. It is based on the popular idea that ‘no evidence of harm’ is not the same as ‘evidence of no harm’. But this old adage does not hold when there has been substantial research done and no indication of harm found.

The Department of Health is reporting as saying that that it wants to offer consistent advice that will not confuse people. We know that the constantly changing stories in the media about is safe and what is not safe can lead to bafflement. But the attitude of the Department could also be interpreted as being patronising and arrogant, and takes no account of the possible cost, in terms of reputation and trust, of providing advice that contradicts the available evidence. Let alone the increased anxiety and guilt of women who have a drink while they're pregnant.

We have covered this issue before, when in 2007 the media jumped on an apparent discrepancy between Department of Health advice to avoid all alcohol and draft NICE guidelines which stated:

Pregnant women should limit their alcohol intake to less than one standard drink (1.5 UK units or 12g of alcohol) per day and if possible avoid alcohol in the first 3 months of pregnancy.
Women should be informed that binge drinking (defined as more than 5 standard drinks on a single occasion) may be particularly harmful during pregnancy.

(NICE 2007 p.36)

In spite of this pressure, NICE only slightly changed its view, and the final version of the guidelines which appeared in March 2008 (pdf) say:

If you are pregnant or planning to become pregnant, you should try to
avoid alcohol completely in the first 3 months of pregnancy because
there may be an increased risk of miscarriage.
If you choose to drink while you are pregnant, you should drink
no more than 1 or 2 UK units of alcohol once or twice a week.
There is uncertainty about how much alcohol is safe to drink in
pregnancy, but at this low level there is no evidence of any harm to
the unborn baby.
You should not get drunk or binge drink (drinking more than 7.5 UK
units of alcohol on a single occasion) while you are pregnant because
this can harm your unborn baby.

It is reassuring to see NICE showing its independence from the Department.

In fact, as covered by Straight Statistics, the study even points to a possible protective effect of alcohol. This is likely to be a statistical artefact, but even considering this possibility is clearly as unthinkable as in the Million Women study, where teetotallers were simply eliminated from the analysis in spite of them displaying higher cancer rates than moderate drinkers.



We all act like Bayesians, don't we? We all have our prior beliefs, and weigh new evidence in light of how it changes those beliefs. You (appropriately, I think) take issue with the Department of Health's safety advice about alcohol and pregnancy. In your last paragraph, you mention a study that points to a possible protective effect of alcohol. Your prior belief shows when you remark "This is likely to be a statistical artifact..."

A fair cop, guv. I was revealing my prior belief that alcohol was unlikely to be protective.

The Million Women authors eliminated teetotallers from the analysis because the study design was flawed: they used current drinking habits as a proxy for lifelong drinking habits. Thus, they didn't discriminate between never drinkers and current non-drinkers, even though many current non-drinkers are likely to have been heavy drinkers in the past who stopped because of illness. On the other hand, I suppose it's fairly reasonable to assume most current drinkers are likely to have been drinking with similar frequency in the past, so the proxy assumption might be OK. I.e. the correlation between current and past drinking habits is probably much higher for current drinkers than current non-drinkers, something which wasn't taken into account. Given the flaw in the study, excluding the teetotallers post-hoc seems fairly reasonable, but in your post you made it sound like they were dishonest!

The possibility that people may not be drinking now because they have given up through illness is a case of 'reverse causation'. There is a standard way of dealing with this: ignore all events in the first few years of follow-up. It is a shame that the Million Women article did not contain such an analysis. But I am not suggesting this was dishonest.

I think that the main issue is the profit the goverment gets from the sales of alcohol. If the goverment would be interested purely in public health then all the girls from the very young age would know that "There is no estimated "safe dose" of alcohol for pregnant women. Since there is no known safe amount of alcohol consumption diring pregnancy, the Academy (American Academy of Pediatrics) recommends abstinence from alcohol for women who are pregnant or who are planning for pregnancy." (American Academy of Pediatrics, 1993). If there is the risk of birth defects, low birth weight, brain damage, etc., why do take that risk? And in this context statistics become an art - it means you're supposed to know when you should discard your expectancy. The main research field in this matter is financed by the goverment and their expectancy is "alcohol is safe during pregnancy". We know that there is always the risk of making Type 1 error and Type 2 error in the research. The results are almost always interpreted in connection with the hypothesis and expectations. And the question is how likely is the goverment to launch a compain "No alcohol = No risk" if they are going to lose money?