Monday 21 October 2013

Expecting Better

My review of Emily Oster's excellent Expecting Better appeared in this weekend's Christchurch Press; it's copied below. I've added in a few relevant links.
I had a lot of questions for our midwife when we expected our first a few years ago. Questions like, "Which shared-care obstetrician has the best delivery outcomes adjusting for the risk of the patient base?" My wife later warned me that our excellent midwife seemed to think me some kind of pod-person for asking questions about the evidentiary base around the standard pregnancy recommendations. Had we then had a copy of Professor Emily Oster's new book, "Expecting Better", I wouldn't have needed to torture her. I should send her a copy now. And if you're expecting, or thinking about getting pregnant, you should pick up a copy too.

Emily Oster is a University of Chicago economist. When she and her husband, economist Jessie Shapiro, decided to start a family, she started into the data. The pregnancy books and doctors provide a lot of recommendations. Some of these benefit the foetus at next to no cost to the mother. Others provide only a small benefit to the foetus while imposing some cost on the mothers. And, sadly, some impose reasonable cost on the mother while doing nothing to help.

Oster provides us the story of her pregnancy, in an accessible and conversational style, while walking us through the research she conducted along the way. While trying to conceive, she worked out the conception probabilities at each date of the cycle. She also found that while obesity makes it harder to conceive, merely being overweight isn't much different from being normal weight. When the pregnancy test came back positive, she wanted to know, and quickly, whether she could stick with her standard caffeine addiction. She charted the probability of miscarriage at every week of pregnancy. All the “no, you can't eat that” rules? She shows us which of them make sense. Later in the book, she walks through the difficult cost-benefit analysis around amniocentesis - the exact same calculation we ran ourselves.

So what's an economist doing writing a book on pregnancy? Economists are good at framing tradeoffs between risks and benefits to help others decide what's right for them. Further, economists' statistical toolkits are built for solving the kinds of statistical inference problems that plague the population health literature. If a study finds mothers who drink more coffee during pregnancy have worse outcomes, is it because of the coffee, or is it because more nauseous women both have better outcomes and find coffee unappealing? Economists' training helps us sort out which studies have done a reasonable job and which ones really haven't. Oster concludes that a couple cups of coffee a day are just fine.

If costs to the mother never mattered, and if statistical inference were easy, then life would be simpler. But in most cases, we need to weigh how much real risk is imposed by, say, smoking a cigarette during pregnancy, and then decide whether the benefit to the mother could outweigh the cost to the foetus. Smoking during pregnancy is very risky and even light smoking is associated with worse outcomes, so the calculus for most mothers on that one should be easy. But what about drinking? Oster's review of the literature here coincided perfectly with what I found when I looked at the same studies: light drinking during pregnancy, on the order of a small glass of wine every other night or so, does absolutely no harm to the foetus. Heavy drinking is very very bad indeed. But the bulk of the well-designed studies show no risk, and in some cases some benefits, from light drinking. If a small glass of wine with dinner helps you relax and you've been abstaining because you're scared that one glass will do harm, you're making pregnancy less pleasant than it could be and achieving little for it.

Oster concluded, as I did, that most recommendations in this area seem motivated by the fear of encouraging binge and heavy drinkers to continue such very harmful practices during pregnancy. But is it really ethical to lie to pregnant women because we're scared they can't handle the truth?

Similarly, and at least in North America and the UK, there's no good reason to avoid sushi during pregnancy as the dangerous forms of salmonella there are very rare. Were we ever to contemplate a third child, I'd be investigating whether those strains are at all common around New Zealand. On the other side, gardening is riskier than I had expected because of toxiplasmosis gondi.

I worry that the standard full set of dos and don'ts make pregnancy sufficiently costly and nerve-wracking that some families are smaller than they otherwise would be. A lot of current recommendations seem to be public rituals designed to allow the mother to display how much she cares about the infant, and for others to display similar amounts of care for the infant by tut-tutting the mother-to-be. This does harm. We make pregnancy a worse experience, and too-often needlessly so, for the mothers who care and worry most about following all the rules. I worry further that by promoting rules that most people know make little sense, we encourage some mothers to discount even the rules that make a lot of sense. Oster provides a sound assessment of the actual risks to help families make their own decisions about the path through pregnancy that is right for them.
Related posts:


  1. She notes the last major listeria outbreaks in the US were celery and cantaloupe. And sprouts. Deli meats is way back in the list. Why do we recommend avoiding deli meats rather than celery then? But you're right - recommendations here would have to be country-specific. For her, it was raw-milk cheeses for listeria and deli turkey for listeria. It would be useful to have the local list of "top sources of listeria in the past decade, weighted by prevalence in eating." So if hardly anybody eats mussels but there are lots of cases of listeria with mussels, that's way worse than an even greater absolute number of listeria cases for, say, carrots if everybody eats carrots all the time.

  2. These sorts of things, if identified, are definitely notifiable. The problem with Listeria is that in normal healthy adults it tends to present as nothing worse than a flu-like illness and maybe a bit of an upset tummy, so the data on listeriosis is likely to be pretty incomplete.
    Celery and cantaloupe-related outbreaks come as no surprise as L. monocytogenes is very common in soil, and eating raw unwashed vegetables or fruit is kind of asking for trouble if you are immunocompromised. As Homer J Simpson famously said, you can't make friends with salad :)