Cribsheet

by Emily Oster

  Given these findings, why do we continue to see the “evidence-based” claim that breastfeeding reduces colds and ear infections? The main reason is there are many observational studies—which compare kids who are breastfed with those who are not, but not where breastfeeding is randomly varied—that do show that breastfeeding affects these illnesses. An especially large set of studies argues for an effect of breastfeeding on ear infections.10

  Should we give any weight to this evidence once we have a randomized trial?

  This is a complicated question. On one hand, all things being equal, randomized evidence is clearly better. We know that breastfeeding is not something people do on a whim, and we know that women who nurse have different circumstances from those who do not. This leads us to favor the randomized evidence.

  On the other hand, the randomized trial is only one study. And it is not infinitely large. If there are small benefits from breastfeeding, they might not show up as significant effects in the randomized trial, but we would still like to know about them. I think it is reasonable, therefore, to look at the non-randomized data, especially when it comes to ear infections, which are widely studied, and where some of the evidence comes from very large and high-quality datasets.

  For example, a study of 70,000 Danish women published in 2016 found that breastfeeding through six months reduced the risk of an ear infection from 7 percent to 5 percent over those months.11 This study was very careful and complete, with excellent data that allowed the authors to adjust for a lot of differences across mothers and children.

  This effect isn’t replicated everywhere. A similar study in the UK shows no impact on ear infections.12 But in my view, the weight of overall evidence puts this in the plausible category.

  In contrast, there isn’t any study as compelling as this Danish ear infection study on colds and coughs. The studies on these symptoms are smaller and less statistically convincing, and the results are fragile. There seems to be less to learn here.

  Where does this leave us? Certainly, it seems reasonable to conclude that breastfeeding lowers infant eczema and gastrointestinal infections. For the other illness outcomes, the most compelling evidence is in favor of a small reduction in ear infections in breastfed children.

  BREASTFEEDING AND SIDS

  I would be remiss to leave the discussion of breast milk and early-life health without discussing the relationship between breastfeeding and SIDS, the tragic cases in which an infant dies unexpectedly in the crib. The relationship of SIDS to breastfeeding, while frequently posited, is difficult to untangle.

  The death of a child is among the worst things you can imagine as a parent. In this book, we will look at many questions that feel weighted, but nothing will compare to this horrific circumstance. This gives added emotional valence to even the suggested possibility of a relationship between breastfeeding and infant mortality.

  SIDS is rare; ear infections and colds are common. Your kids will get colds for sure, whether you breastfeed or not. SIDS deaths, in contrast, occur in about 1 of every 1,800 births; among babies with no other risk factors (not premature, not sleeping on their stomachs), this is perhaps 1 in 10,000.13

  This should reassure anxious parents to some extent, but it also makes the SIDS–breastfeeding relationship hard to study, since you need an enormously large sample of babies to learn anything that can benefit other children.

  To get around this, studies of this relationship use the case-control method: They identify a number of infants who have died of SIDS, interview the parents, then interview a set of control parents with living children. The characteristics of the parents and children are compared.

  There are many of these studies.14 And, on average, they do find that the living children are more likely to be breastfed. This causes them to conclude that not breastfeeding increases the risk of SIDS. The most recent analyses suggest that these effects are most pronounced for breastfeeding longer than two months.15

  In my opinion, however, from a careful read of the data, this conclusion is not obvious. There are basic differences between the children who die and those who do not, differences that likely have nothing to do with breastfeeding but are driving many of the results. When the studies take into account things like a parent’s smoking, whether the baby was premature, and other risk factors—all of which are correlated with breastfeeding and linked to SIDS—their effects are much smaller or disappear altogether.

  Beyond this, some of the research papers with the largest effects also have a serious problem with their selection of the control group. A key component of designing these studies is to pick a control group that is as comparable as possible, and these studies are not always successful in this goal.

  For example, it is common to select all infants who die of SIDS in an area as the treatment group, and then recruit parents of living children with letters or phone calls. But this means the people in the control group are chosen differently, and we know that people who want to participate in a study are fundamentally different—in ways we can see and ways we cannot—from people who do not choose to be involved.16

  Reinforcing this concern, studies with a better selection of control babies—for example, one where the comparison group comprises babies who were visited by the same home-visiting nurse in England—do not show an elevated risk of SIDS from not breastfeeding.17

  SIDS deaths are thankfully rare. Because they are so rare, it is impossible to fully rule out the possibility that breastfeeding decreases the risk of SIDS by a small amount. However, I do not believe the best data supports a significant link.

  BREASTFEEDING AND LATER HEALTH

  Most of the academic research on breastfeeding focuses on early-life outcomes—infections, for example, in the time period in which you might actually be breastfeeding. In the popular discourse, however, the focus seems to be much more on the long-term benefits. This is where the guilt stacks up.

  You rarely hear people say, “It’s great to breastfeed since it lowers the chances of diarrhea in the next six months!” Rather, they say things like, “It’s great to breastfeed since that gives your kid the best start; they’ll be smarter, taller, thinner!” This problem isn’t limited to random people on the street: one woman told me her doctor had told her that by quitting breastfeeding, she was costing her child three IQ points.

  The idea that choosing not to breastfeed might be something your child would suffer from for their whole life is far worse as a parent than simply thinking they might get one more ear infection.

  The good news for guilt-ridden moms is that, even more than in the case of early-life health issues, I have not seen any convincing evidence for these long-term impacts.

  We can begin with the set of outcomes studied in PROBIT. These researchers have continued to follow the children in the trial through the age of seven. They find no evidence of any long-term health impacts: no change in allergies or asthma, cavities, height, blood pressure, weight, or indicators for being overweight or obese.18

  The results on obesity are worth pausing on, as this benefit of breastfeeding gets a lot of attention. (When I was pregnant with Finn, there was a very large poster in my midwife’s office claiming that breastfeeding lowered obesity, a message underscored by the image of two ice cream scoops, each topped with a cherry so they looked like breasts. It was a neat visual, although the point it was illustrating remains unclear to me. I suppose the idea was that you could eat more ice cream if you were breastfed.)

  It is certainly true that obesity and breastfeeding are correlated, as kids who are breastfed are less likely to be obese later in life. But this correlation doesn’t show causation—it doesn’t prove that those kids who go on to become obese do so because they weren’t breastfed. The randomized data from PROBIT shows no impact of breastfeeding on whether the child is obese at the age of seven or, in the latest follow-up, at close to eleven.19 Bolstering this, studies that compare sibli
ngs who are breastfed to those who are not show no differences in obesity. These studies often demonstrate that breastfeeding seems to matter when you compare across families, but not within a family. This suggests that something about the family, not the breastfeeding, is impacting the likelihood of a child becoming obese.20 In fact, when researchers look at many studies of obesity and breastfeeding together to get a fuller picture, they find that studies that carefully adjust for maternal socioeconomic status, maternal smoking, and maternal weight—even if they cannot compare siblings—also show no association.21

  All these results come with some statistical error. Can we say for sure that breastfeeding does not impact obesity? No. But we can say that nothing compelling in the data supports a significant link.

  A few long-term outcomes—for example, juvenile arthritis and urinary tract infections—could not be studied in PROBIT, but at least one or two studies have shown some link between these conditions and breastfeeding. The evidence on most of these links is simply very limited.22 A significant relationship shows up in only one of many studies, or the research design is poor, or the population is very unusual—basically, we cannot learn anything from the data about whether there is a relationship.

  More has been written on two more serious illnesses—type 1 diabetes and childhood cancer—but, again, given the limitations of the data, I do not think we learn much. More on these two in the endnote.23

  In many of these cases—like others in the breastfeeding arena—even very limited and poorly done studies get a lot of attention. Media attention tends to miss the nuance of published literature, even when the literature itself is good, which is often not the case. We see, again and again, aggressive headlines that often overstate the claims of the articles they report on.

  Why is this?

  One reason is that people seem to love a scary or shocking narrative. “Report: Formula-Fed Children More Likely to Drop Out of High School” is a more clickable headline than “Large, Well-Designed Study Shows Small Impacts of Breastfeeding on Diarrheal Diseases.” This desire for shock and awe interacts poorly with most people’s lack of statistical knowledge. There is no pressure on the media to focus on reporting the “best” studies, since people have a hard time separating the good studies from the less-good ones. Media reports can get away with saying “A new study shows . . .” without saying “A new study, with very likely biased results, shows . . .” And other than the few of us who get our dander up on Twitter, people are mostly none the wiser.

  It is hard to sort out study quality from this initial media coverage, although it’s probably easier in the age of the internet. Many media reports will now link to the original study. If the “Formula-Fed Children More Likely to Drop Out of High School” article is based on a study of forty-five people surveyed about their breastfeeding behavior when their now twenty-year-old children were infants, you can probably let it go.

  SMARTY-BOOBS: BREASTFEEDING AND IQ

  Breast milk is optimal for brain development, right? Nurse your way to a successful child! So they say. But is this true? Will breast milk make your kid smarter?

  Let’s start by returning from the land of magical breast milk to reality. Even in the most optimistic view about breastfeeding, the impact on IQ is small. Breastfeeding isn’t going to increase your child’s IQ by twenty points. How do we know? Because if it did, it would be really obvious in the data and in your everyday experience.

  The question is, really, whether breastfeeding gives children some small leg up in intelligence. If you believe studies that just compare kids who are breastfed to those who are not, you find that it does. I talked about one example of these studies on this page, and there are others. There is a clear correlation here—breastfed kids do seem to have higher IQs.

  But this isn’t the same as saying that breastfeeding causes the higher IQ. In reality, the causal link is much more tenuous. We can see this by looking carefully at a number of studies that compare children who were breastfed to their siblings who were not. These studies tend to find no relationship between breastfeeding and IQ. The children who were nursed did no better on IQ tests than their siblings who were not.

  This conclusion differs fundamentally from the studies without sibling comparisons. One very nice study gives us an answer to why.24 The key to this study is that the authors analyze the same sample of kids in a bunch of different ways. First, they compare children who are breastfed with those who are not with a few simple controls. When they do this, they find large differences in child IQ between the breastfed kids and those who are not. In the second phase, they add an adjustment for the mother’s IQ, and find that the effect of breastfeeding is much smaller—much of the effect attributed to breastfeeding in the first analysis was due to differences in the mothers’ IQs—but does still persist.

  But then the authors do a third analysis where they compare siblings—children born to the same mother—one of whom was breastfed and one who was not. This is valuable because it takes into account all the differences between the moms, not just their performance on one IQ test. In this analysis, researchers see that breastfeeding doesn’t have a significant impact on IQ. This suggests that it is something about the mother (or the parents in general), not anything about breast milk, that is driving the breastfeeding effect in the first analysis.

  PROBIT also looked at the relationship between breastfeeding and IQ. For this sample, the measurement of IQ was done by researchers who knew whether a child was in the breastfeeding-encouraged treatment group. There were no significant effects of breastfeeding on overall IQ or on teachers’ evaluations of the children’s performance in school. The researchers did see small impacts of breastfeeding on verbal IQ in some of their tests, but further analysis suggested that this may have been driven by the people doing the measurement—knowing which children were breastfed might have influenced their evaluation.25 Overall, therefore, this study doesn’t provide especially strong support for the claim that breastfeeding increases IQ.26

  In conclusion, there is no compelling evidence for smarty-boobs.

  BENEFITS FOR MOM

  For some women, breastfeeding makes them feel empowered and happy. It’s convenient to have a ready food source anywhere they go, and they find nursing their baby to be a peaceful and relaxing time. That’s great!

  For others, breastfeeding makes them feel like a cow. They hate lugging the breast pump around if they have to pump. It’s hard to tell if the baby even likes to nurse or is getting enough food. Their nipples hurt, and the experience basically sucks.

  All this is to say that many of the purported benefits of breastfeeding for moms are really subjective. I have been on both sides of this, as have most of my friends. There were definitely moments—especially with Finn—when I thought it was a superconvenient and awesome option. And then there were others—I am thinking in particular of an experience pumping in the bathroom at LaGuardia Airport—when the whole thing seemed like a farce.

  One of the things on every pro-breastfeeding list is “saves money.” This really depends. Yes, formula is expensive, but so are nursing tops, nipple creams, nursing pads, and the fourteen different breastfeeding pillows you need to make it work. And, more important, there is your time, which is valuable.

  Another claimed benefit is “stress resistance.” Does breastfeeding make you more resistant to stress? Again, pretty subjective. Stress is very often linked with sleep disturbance. Will you get more sleep if you nurse your baby? This depends on more than just breastfeeding.

  As mentioned earlier, “better friendships” has also been touted as a benefit. You’ll need to decide for yourself if your friendships will be enhanced by breastfeeding. (It probably depends on your friends.)

  These are just a few of the “benefits” of breastfeeding for which there is just no evidence. A few claimed benefits, however, do potentially have some basis in fact. The first is the claim that breastfeeding
is “free birth control.” Here is the truth: you are less likely to get pregnant if you breastfeed, but it is not—I repeat, NOT—a reliable birth control method, especially as your child ages and if you ever go more than a few hours without feeding or pumping. I do not have enough space in this book to list all the people I know who got pregnant while breastfeeding (shout-out here to my medical editor, Adam, his wife, and his second child). If you definitely do not want to get pregnant, you need to use some real birth control.

  A second claimed benefit with some evidence is “weight loss.” I’m sorry to report that, at best, any weight loss effects are small. One large study from North Carolina showed that at three months postpartum, weight loss was similar in moms who breastfed and those who did not. At six months postpartum, the breastfeeding moms had lost about 1.4 pounds more.27 Issues with this paper mean this is likely an overestimate of the effect of breastfeeding on weight loss, but at any rate, it is still very small.

  You may be wondering, Doesn’t breastfeeding burn calories? Didn’t I hear something about how you use five hundred calories a day nursing? This is true, but women who are nursing tend to eat more. Burning more calories is effective as a weight-loss strategy only if you do not make those calories up in what you eat. When I was nursing, I had a policy of eating an egg and cheese bagel sandwich at ten thirty every morning. This type of behavior pretty much guarantees you will replace the calories you burn.

  The evidence of the effect of breastfeeding on postpartum depression is similarly noncompelling. Studies of this relationship show mixed results, and it’s a hard question to evaluate since the causality goes both ways. Mothers suffering from postpartum depression are more likely to quit breastfeeding, which makes it look like breastfeeding relieves postpartum depression, when actually, the causality is the other way around.28 And the claim of lowered risk of developing osteoporosis and improved bone health is also not apparent in large datasets.29 Evidence on diabetes is also mixed, and likely confounded with differences across women.