The evidence base for not drinking while pregnant

2019/02/18

In my experience, US culture is militant about not drinking while pregnant. The first lines on the CDC’s web page on alcohol use in pregnancy are: “There is no known safe amount of alcohol use during pregnancy or while trying to get pregnant. There is also no safe time during pregnancy to drink.” Their punchline is: “FASDs [fetal alcohol spectrum disorders] are completely preventable if a woman does not drink alcohol during pregnancy. Why take the risk?”

The reason to take a risk is obvious: that the benefits of a choice outweigh its harms, weighted by the probabilities of the benefits and the harms. The CDC presents its evidence as if the harms will certainly outweight the benefits. I’m not saying they’re wrong, but this logic does not seem universal. For example, my experience living in Britain is that mothers there are more less relaxed about drinking small amounts of alcohol while pregnant. And people seem less aggressive in socially policing mothers’ behavior.

The benefits in question are the pleasures of drinking alcohol. This is an entirely personal experience that does not require further discussion. (There are also mild health benefits from moderate alcohol use, but I’ll leave those aside.) The harms are the complex part.

The most-touted harm are the fetal alcohol spectrum disorders. This term includes the more familiar fetal alcohol syndrome (FAS) as well as other alcohol-related neurodevelopmental problems and birth defects. Symptoms of FAS include physical problems like abnormal facial features (e.g., a smooth upper lip without the typical dent) and developmental disabilities like learning disabilities and behavioral problems. In extreme cases, the baby can die from FAS.

FAS is clearly a bad outcome, and it is clearly related to alcohol: mothers who do not drink any alcohol never give birth to babies with FAS, while around one-third of heavy drinking mothers (4+ drinks per day) give birth to babies with “complete” FAS (meaning that they have many of the suite of possible symptoms).

The data become less conclusive for moderate and light levels of drinking. Not surprisingly, the effects of drinking become harder to measure as drinking levels become lighter. For a concrete example, let’s look at the results of a 1984 study in the Journal of the American Medical Association (JAMA) that became critical to policies like the CDC’s measured babies’ birth weight as a funtion of mothers’ drinking levels. Among mothers who did not drink, babies’ weights were distributed with around a mean of 3,433 grams, with standard deviation 540 grams. Among heavy-drinking mothers (3-5 drinks per day), mean birth weight was 3,183 (± 496 s.d.) grams, a difference of 165 grams (95% confidence interval 74 to 256 grams, $p = 0.0004$). The average birth weight among babies born to heavy-drinking mothers was therefore 5% smaller than the average weight among babies both to non-drinking mothers. A 5% smaller baby is not on its own a cause for concern, but decreased birth weight is here used as an easy-to-measure proxy for the later, more important effects of FAS. It is much easier to do a study weighing babies and asking mothers about their drinking habits than it is to follow children to see if their mothers’ behavior before birth correlates with behavioral patterns a decade later.

In terms of light drinking, the study found that mothers who drank less than 1 drink per day on average (but who did drink) had babies with mean weight 3,442 grams (± 534 s.d.), or 14 grams lighter on average (95% confidence interval 2.8 to 28 grams lighter, $p = 0.01$). That even very light drinking was associated with some decrease in average birth weight is part of the CDC’s logic in that there is no “safe” amount of alcohol for pregnant mothers. In this reasoning, infinitesimal drinking is associated with infinitesimal but non-zero risk, making it “unsafe”.

This use of the word “safe” is problematic. On the one hand, I sympathize with the CDC’s position: it seems like it would be better, from a public health point of view, if all pregnant and potentially-pregnant women drank zero alcohol. On the other hand, almost no activity is without risk. There is no amount of riding in a car that is “safe” because you are always at risk of an accident, but we always balance that risk against the benefits of transportation. In the socially-conservative and still-Puritan-influenced US, a mother’s pleasure is valued very, very low in comparison with a child’s health, even to the point that the pleasure of a single alcoholic drink in nine month’s time is considered insufficient to justify the immeasurably-small risk of a slight change in a baby’s biology.

Mothers should also take these old adages, derived from studies like the one in the 1984 JAMA paper, with a grain of salt. A 2007 meta-analysis found that mothers’ mild drinking is associated with slightly higher baby birthweights and slightly lower risks of stillbirth. It was unclear that this association was causal: it could be that healthier mothers feel more confident about their babies’ future health and are therefore more willing to drink small amounts of alcohol rather than abstaining completely. In any case, my main point holds: small amounts of drinking are almost certainly unlikely to lead to serious negative outcomes for infants’ health.

I expect that social pressures against mothers’ drinking are typically stronger influences on a mother’s behavior than her own rational weighing of potential medical harms. If I’m right, then this is good from a public health point of view, as it curtails even mild drinking, but this is, more importantly, an example of mommy shaming.