What is dangerous for pregnant women?

Not that much, stop worrying

Remember to participate in the 2025 readers survey!

There's a very large literature on the outcomes of children to mothers who did X. The problem with most of these is that mothers who do and don't do X are not necessarily otherwise comparable to mothers who didn't do X. In other words, the primary problem of social science applies, that of self-selection. So what can we do about this problem? Well, nature has been kind enough to make siblings. They are very useful because they don't differ in certain ways. They have the same parents, and thus the same expectation of genetic scores for every trait. They don't actually have the same actual genetic scores (genetic potential), of course, but on average they do. Their genetic differences are due to the randomness inherent in the making of gametes (recombination). This is very important because this means the genetic variation among siblings cannot in general be caused by anything the parents do (except genetic methods!). As such, siblings gives us a near-perfect control for certain kinds of parental behaviors, including prenatal. These cannot otherwise be easily studied without the self-selection issue.

Now I've been asked repeatedly about many of these dangers, so I've reviewed a lot of studies. These are not all the relevant studies, but they include most of the things people worry about researchers have studied. Most research is from the Nordic countries because they have the best data. Fortunately, because interactions are mostly not real, the effects are likely to be the same in your country and ethnic group (this is also generally why Aella's big self-selected samples are fine for many kinds of analyses, but of course not all).

Barclay & Kolk 2018 studied whether birth spacing impacted long-term health outcomes in Sweden. Much prior literature has found that shorter intervals between births are associated with all sorts of negative outcomes. The health recommendation based on this would be to tell women to space out their births, thus lowering fertility (as it takes more years to produce the same number of offspring). However, a moment's reflection should tell us that women who have more children and thus more likely to have short intervals are also likely to have other problems in line with the typical dysgenic fertility findings. So, the association may just reflect the difference in which women have more children faster or not, and nothing to do with birth spacing itself. Let's begin with height. This is often used as an overall health proxy and is easy to measure:

Light blue shows the between family associations. There aren't that many to see, but at least we see that if you get a sibling shortly after you are born, you are likely to be a bit shorter (light blue, right figure, 9-12 months). We could speculate that this is because you competed for resources with them such as food or parental attention, or maybe they bring an infectious disease home from daycare. Also, if you are born long after a prior sibling, you are also likely to be somewhat shorter (I don't have a remotely plausible speculation except for parental age). However, among siblings, there is nothing to see for either of these putative causes. What about physical fitness as measured by the Swedish army?

Here we see a relatively clear pattern: between families if you are born long after a prior sibling, you are likely somewhat less fit. However, there's not much to see when looking only between siblings.

Giving birth is a painful affair as has been long noted (genesis: in sorrow thou shalt bring forth children). So what if we give women effective painkillers (in the spinal cord) to reduce this pain? Well, some research suggested this might damage the children. This puts women in a tough situation: pain or risk their children? Hegvik et al 2023 decided to look at this using siblings from Finland, Norway, and Sweden (there's a Danish replication as well). Replicating earlier findings, they found that epidural use was associated with 12-20% higher odds of offspring getting an ADHD or autism diagnoses. However, when they looked at siblings only, they found nothing.

This is rather fortunate because the use of epidurals have been growing over the decades.

Conclusion: women who are more likely to have children with ADHD and autism are more likely to elect to have epidurals for whatever reason. It's not something to worry about.

What about maternal smoking? Clearly smoking is bad for you, but does it also harm your children even before they are born and set them up for failure? 1000s of studies have found that children to smoking parents or mothers have an assortment of worse outcomes. By now you probably get the idea: maybe women who choose to smoke during pregnancy (and in general) aren't quite the same as those who don't, and maybe this accounts for whatever differences we see. Brand et al 2019 decided to look at this question using Swedish register data and their outcome of choice was fractures up to age 32 (I swear researchers choose the weirdest outcomes in their salami publishing). They found the usual pattern, namely that maternal smoking during pregnancy was associated with worse outcomes, about 27% higher odds for fractures. And you might also have guessed that this pattern disappeared among siblings. Skoglund et al 2013 found that smoking during pregnancy predicted offspring ADHD, but not within families. You probably won't be surprised if you think ADHD in mothers predicts smoking, the genetic scores for which they pass on to their children. Obel et al 2015 confirmed this conclusion with Danish data. Wei et al 2022 found that smoking during pregnancy might even be protective against type 1 diabetes (hard to believe!). To be fair, Juarez and Merlo 2013 found that smoking during pregnancy does slightly reduce birth weights in sibling analyses. Kuja-Halkola et al 2014 did the most comprehensive analysis of Swedish data, and found that smoking during pregnancy predicts perinatal outcomes among siblings, but not long-term outcomes (academic achievement, intelligence, criminality, drug abuse). Thus, overall, it seems that aside from some minor risk increase around birth, smoking during pregnancy doesn't cause much of the associations it shows, so there is little to worry about from a public baby health perspective (smoking remains bad for yourself, however).

So let's up the ante. What about alcohol during pregnancy? Most people know about fetal alcohol syndrome. May et al 2018 estimated this might affected 1-5% of all Americans, which sounds hard to believe. Eilertsen et al 2017 examined this question for ADHD diagnoses and symptoms using Norwegian data. They found that maybe there's some causality among siblings, but it was perhaps 20% of the crude effect size. Lund et al 2019 repeated this study with more outcomes but generally found no association between drinking during pregnancy and children's mental health or behavior at various ages. It would seen fetal alcohol syndrome worries have been overestimated, at least, we cannot seem to find them using regular methods. My guess is that to cause this syndrome, quite a lot of drinking is needed and very few mothers are stupid enough to do this.

OK, what about Cesarean birth (c-section)? Again, there are many, many studies finding associations with this or that bad outcome. There are a few sibling control studies too. Curran et al 2015 studied Swedish data. They found that elective/emergency c-section was associated with 15-20% higher risk of offspring autism diagnosis. However, among siblings, there was no effect to be seen. What about maternal antibiotic use during pregnancy? Nakitanda et al 2023 studied Swedish data and found that in the total population, the association was about 15% extra risk of offspring infections, but only 5% in the sibling design, so it appears about 30% of the effect is causal, perhaps because mothers inadvertently transmitted infections to their offspring early on, which weakened the immune system (but who knows).

Phew! Maybe you are now thinking, "maybe I shouldn't worry so much and just have more children" and I agree! If you need more reasons, give Bryan Caplan's book Selfish reasons to have more kids a read (or just click the link and read Scott Alexander's review). Unless you do something very crazy, you are unlikely to fuck up your child during pregnancy, or at some later point during upbringing. This is because the shared environment variance in explaining outcomes is usually small or undetectable, and this includes parenting efforts. So relax, have fun, go forth and multiply. And if you are wondering why I suspiciously write this post about (the general lack of) pregnancy dangers, well, you will see in August.

Now I know there are some nerds among the readers. They will object that one can get false negatives from lower power in sibling studies both due to sample size reductions and larger measurement errors. I agree, of course, so that's why I was careful when reviewing the above studies to not include obvious low power studies as plausible negatives. Still, a better review could be done. I saw a few academic reviews and meta-analyses, but they didn't do things properly to my satisfaction so I didn't include them here.