Cribsheet

by Emily Oster

  This changed in the early 1990s with the release of a series of studies showing more directly that stomach sleeping was associated with a dramatically elevated risk of SIDS.

  Studying this problem with data is challenging. SIDS deaths are thankfully rare, so some of the more standard research techniques are difficult to implement. Even a very large randomized trial or observational study is not likely to have enough observations to draw statistically meaningful conclusions.4 Instead, researchers typically look at SIDS using case-control studies.

  In 1990, the British Medical Journal published one of these studies, based on data drawn from the UK.5 The researchers focused on a particular area (Avon) and identified sixty-seven infants in that area who had died of SIDS. They then searched for two control infants for each of the cases—those of similar age, or similar age and birth weight—and surveyed both sets of parents.

  Their most striking findings related to stomach sleeping. Nearly all the infants who died of SIDS were sleeping on their stomachs (62 of the 67 infants, or 92 percent). However, among the surviving infants, only 56 percent were sleeping on their stomachs. Based on this comparison, the authors argued that babies who sleep on their stomachs are eight times as likely to die of SIDS. This paper also cited overheating as a risk factor—the babies who died were more likely to be wearing heavy clothing to bed, sleeping under a lot of bedding, or sleeping in a hot room.

  Other research with similar approaches shows the same results.6 This is not the only type of evidence we have. There is a biological mechanism for the link: babies tend to sleep more deeply on their stomachs, and SIDS risk is increased with deeper sleep. In addition, we have evidence from the Netherlands based on variation in sleep position over time.

  In the 1970s, a Dutch campaign encouraged parents to put their children to sleep on their stomachs. In 1988, the recommendation changed, and parents were told to put children to sleep on their backs. With these changes in sleep position came changes in incidences of SIDS. SIDS rates rose after the stomach-sleep recommendation, and fell after parents were told to put their children to sleep on their backs.7 Alone, this type of variation over time wouldn’t prove a causal relationship between SIDS and sleep position. But combined with the other evidence, it begins to paint a causal picture.

  By the early 1990s, it seemed clear that stomach sleeping was risky. A review article in the Journal of the American Medical Association at this time discussed all the evidence, and concluded that despite having no randomized trial, the data warranted a serious effort to prevent parents from putting their babies in this position to sleep.8

  This effort came in the form of the aforementioned “Back to Sleep” campaign, which began in the US in 1992, and was remarkably successful. In surveys done in 1992, researchers found that around 70 percent of babies were put to sleep on their stomach.9 By 1996, this figure was only 20 percent. This large change in sleeping position was also accompanied by a decrease in the SIDS rate, further suggesting that sleep position plays a role in SIDS.

  The “Back to Sleep” campaign emphasizes the importance of putting an infant on their back, not on either their side or their stomach. The evidence, however, largely points to stomach sleeping as high risk, rather than side sleeping. The concern about side sleeping is mainly that infants can inadvertently roll onto their stomach. The back sleeping recommendations are, therefore, really designed to avoid the risk of stomach sleeping as fully as possible.

  One note: If your infant does roll over, there is no need to go rolling them back. Once they can do this on their own, the highest risk of SIDS has also passed, probably because the baby now has enough head strength to move their head to breathe more easily.

  Side Effects: Deformational Plagiocephaly

  There is one substantial side effect to back sleeping: deformational plagiocephaly, or, colloquially, flat head. Infants who sleep on their back are at higher risk for head flattening. The frequency of this issue has been rising over time since the implementation of “Back to Sleep.”10

  Deformational plagiocephaly is more likely to occur if the infant always has their head turned to one particular side when they sleep. And at least some literature suggests it is exacerbated by having some degree of head flattening at birth.11 It is also more common in twins and premature babies. It doesn’t have any effect on brain growth or function, so this is purely an aesthetic concern. Making sure your baby has tummy time during the day or, generally, does not spend all day lying on their back can help avoid this condition.

  Flat head is at least somewhat fixable. The standard treatment is a helmet, which is worn for most of the day and night, but there is some debate over whether the helmet actually fixes the problem more successfully than doing nothing. If you face this issue, discuss your treatment options with your pediatrician.12

  RECOMMENDATION 2: “ALONE IN THE CRIB”

  The second piece of advice from the AAP is to have your infant alone in their crib. In other words, no co-sleeping.

  This recommendation is extremely controversial among parents.

  Some people strongly support co-sleeping. A common argument from this group is that this is how infants have slept for millennia. This is true: there was no crib in the cave, and even now it is common for infants and children in many cultures to sleep in bed with their parents for many years. This is, however, not a reliable argument for safety. There are plenty of ways we have changed infant practices to improve survival.

  A common argument in the other direction is that there have been infant deaths from suffocation under a sleeping parent. This is also true. But the fact that this is a possibility doesn’t mean the risk is large, and the risk may be mitigated by how you co-sleep.

  The real question, then, is whether the risk of SIDS is significantly higher when co-sleeping, and if so, how large the increase is. Evidence on this comes, again, from case-control studies similar to those used to study the role of sleep position. In this case, researchers collect information about a set of infant deaths, focusing on the usual sleep location of the infant, where they were sleeping when they died, and whether they were breastfed or bottle-fed, as well as on characteristics of the parents, including their typical alcohol consumption and smoking habits. The researchers then find a set of controls—infants similar in terms of age and other characteristics, but who survived. They ask the parents the same questions and compare their answers.

  Many of the individual studies of this are small, so it is helpful to have “meta-analyses,” which combine data from many similar studies. One excellent example was published in the British Medical Journal in 2013.13 This paper combines data from studies run in Scotland, New Zealand, Germany, and elsewhere (although notably not the US). What is helpful about this analysis is that the authors explicitly tried to estimate the excess risk in groups with varying behaviors. They focused on whether the parents smoked or used alcohol (more than two drinks a day), and whether the infant was breastfed.

  The following graph—based on results from their paper—shows differences in death rates for infants who do and do not bed share. The absolute risks here are constructed based on a normal-weight, nonpremature infant. The various bars show different combinations of risk factors.

  The first thing this graph makes clear is that both overall SIDS rates and the increased risks from co-sleeping are much larger in the presence of other risk factors—parental smoking and drinking, in particular. In the most extreme example, the predicted mortality for a bottle-fed infant with parents who both smoke and where the mother drinks more than two drinks a day is 27 deaths per 1,000 births, fully 16 times higher than the comparable infant who doesn’t share the parents’ bed.

  The observation that smoking, in particular, increases the risks associated with bed sharing is widely shared in other literature.14 The mechanisms for links between SIDS and smoking are not fully understood but seem to relate to the role of chemicals in secondhand sm
oke and their interference with infant breathing. This problem becomes more acute if the baby is closer to the smoker (even if the parent is not actively smoking).15

  This graph also speaks to perhaps the more central question for many families, which is, are there still risks to co-sleeping if you do it as safely as possible—that is, if neither parent smokes or drinks a lot, and if the baby is breastfed?

  The data here says yes. The risk of death for infants who do not bed share in the lowest risk group is 0.08 SIDS deaths per 1,000 births. For those who bed share, it is 0.22 deaths per 1,000 births. Again, we want to put these risks into a broader context. In the US, the overall infant mortality rate is around 5 deaths per 1,000 births. This therefore represents a very small increase relative to the overall mortality rate. A perhaps more useful way to say this is that among families with no other risk factors, roughly 7,100 of them would have to avoid co-sleeping to prevent one death.

  The finding that co-sleeping carries a small risk even if done as safely as possible is largely consistent across studies, and although the exact size of the increased risk varies from report to report, they are in a similar range.16 These risks are concentrated early in life. Notably, there does not seem to be any elevated risk from co-sleeping after three months if both parents are not drinking or smoking.

  Putting these risk analyses together, a main takeaway here is that if you are going to co-sleep, you should definitely not drink a lot or smoke, and neither should your partner. Limiting these behaviors will let you co-sleep in the safest way possible, although it will not completely eliminate the risks. On the other side, though, there may be some benefits.

  The main benefit—the one I see cited by moms most often—is that bed sharing is convenient, and if you try to move an infant who has fallen asleep, they tend to wake up. This is certainly true, at least for some babies, and probably something you can evaluate yourself. If the baby wakes up less, parents may also sleep more.

  Indeed, for my friend Sophie—and other friends, many of them doctors, who told me they co-slept—more sleep was the main reason to do it. For Sophie, whose family comprised two working parents and two other children, it didn’t seem feasible for her to be up all night going back and forth to a crib. Never mind that her son also slept much better in her bed than out of it. It came down to co-sleep or no sleep, and Sophie and her husband ultimately decided that having the baby in their bed was the best thing for their whole family.

  A second possible benefit, one that we can evaluate with data, is the possibility of improved success with breastfeeding. Certainly, there is a correlation: moms who bed share are also more likely to be breastfeeding and to persist until the child is older.17 But this doesn’t necessarily point to causality. We know from data that women who have a strong desire to breastfeed before they give birth are more likely to bed share.18 It could be that the desire to breastfeed prompts bed sharing, not the other way around. And indeed, the one randomized trial that evaluated the relationship between breastfeeding and having an infant in an attached cot rather than a separate bed fails to find any link between bed sharing and breastfeeding.19

  This doesn’t mean there are no benefits for your family to bed sharing, just that it probably isn’t a panacea to improve your breastfeeding success.

  RECOMMENDATION 3: “IN THE PARENTS’ ROOM”

  In the spate of recommendations, bed sharing is forbidden, but room sharing is encouraged. The American Academy of Pediatrics recommends that infants be in their parents’ room through at least the first six months, and ideally the first year, of life as a guard against SIDS. The theory is that parents can be more attentive to the baby if they are in the same room.

  The evidence on room sharing and SIDS is substantially less complete than the evidence on bed sharing. The studies have the same basic structure, but they are smaller and there are fewer of them. Less attention is paid to other factors that might influence the relationship. For example, what if you have a video monitor in the baby’s room? Is that enough? You will not find evidence for that here.

  With that caveat, we can review the studies we do have.

  To take one concrete example, consider a study published in the British Medical Journal in 1999. The authors, using a sample of about 320 infant deaths and 1,300 control infants, argue that sleeping in a room alone is associated with a higher risk of death.20 However, the results in the paper are inconsistent. For example, it matters a lot whether they analyze the usual sleeping location or the most recent sleeping location; there seems to be no risk when they analyze usual sleeping location, but a higher risk when they analyze the most recent one. It’s not clear why this would be, and leads to concerns that something else unusual happened on the last night of life.

  In forming their recommendations on room sharing, the AAP cited this study and three others.21 These show similarly small increases in SIDS rates for babies who sleep in their own room, but the results are not overwhelming. They all tend to be very sensitive to which variables researchers adjust for, and, important, most of these studies were not actually designed to look at room sharing. Although these studies are too small to really analyze mitigating factors, the benefits of room sharing seem to be larger if the infant also sleeps on their stomach22 and depend on whether parents also sometimes bed share.23

  While I think one can debate the merits of room sharing at all, given the data, in my view, the AAP’s recommendation that room sharing extend through the baby’s first year is problematic.

  Why do I say this?

  The vast majority—up to 90 percent—of SIDS deaths occur in the first four months of life, so sleeping choices after four months are very unlikely to matter for SIDS. This also shows up in the data. The choice of sharing a room, or even sharing a bed, does not seem to affect SIDS risk after three or four months, at least for parents who are nonsmokers.24

  This means there is seemingly no benefit to extending room sharing for so long. There is, however, a real cost: child sleep. In a 2017 study, researchers evaluated whether a child’s sleeping in a room with a parent made for worse sleep. They found that it did. At four months old, total sleep time was similar for babies sleeping in a parents’ room and those sleeping in their own room, but sleep was more consolidated (i.e., in longer stretches) for those in the latter group. This makes sense: their own room will be quieter.

  At nine months, infants who slept alone slept longer; this effect was largest for those who slept alone by four months, but also appears for babies who moved to their own room between four and nine months. Most notably, these differences were still present when the child was two and a half years old: children who slept alone by nine months slept forty-five minutes more during the night than those who were room sharing at nine months. Sleep is crucial for child brain development; it is not just a selfish parental indulgence. Of course, this may not be causal—maybe parents move their kids to their own room when they start sleeping well—but it is suggestive.

  Related to this, it should be said that if you plan to sleep train your child, success is very unlikely while the child is sleeping in your room. And finally, most people sleep better without a child in the room, and parents being well rested is important, too.

  Pulling all this together, I believe the AAP recommendations go too far. If you want to share a room with your child, by all means do. And perhaps—perhaps—the data warrants a mild recommendation in favor of very early room sharing. But to tell people they need to keep their child in their room for a year, sacrificing both short- and long-term sleep success with no clear benefit in the process, may not be a good policy.

  The Sofa

  Across virtually all studies of sleep location, the one thing that jumps out as really, really risky is babies sharing a sofa with an adult. Death rates as a result of this behavior are twenty to sixty times higher than the baseline risk. It is not difficult to see why: an exhausted adult falls asleep holding an infant o
n a cushiony sofa, and it is easy for the infant to be smothered by a pillow. The unfortunate thing is that in at least some of these sofa deaths, the parent involved is trying to avoid the risks associated with bed sharing. They hope that if they sit up, they will stay awake, and then they fall asleep by accident. Even with the small risks of bed sharing, you’d be much better off sharing a bed than accidentally co-sleeping on a sofa.

  RECOMMENDATION 4: “NO SOFT STUFF”

  The final AAP guideline for sleep is that (aside from the baby) your child’s crib should be empty, with no toys, no bumpers, no blankets or pillows. Nothing.

  This is probably the easiest recommendation to follow. Other than adorableness, there is no reason to have toys or pillows in the baby’s crib (bumpers may be a different story). There are also some advantages to this if you ever travel with your child. No parent wants to be carting along Lamby and Special Bear and Stinky Dino and Captain Poodlepants when they travel to Grandma’s house. If you can limit the number of things your child absolutely needs to fall asleep, your luggage will thank you.

  In terms of risks, there are two central parts of the no-stuff-in-the-crib recommendation. One is that infants should not have blankets. This conclusion is based on the results of a number of the studies discussed previously. Infants who die of SIDS are more likely to be found with blankets over their heads than control infants. The infant-clothing industry has come up with a solution to this, which is the “wearable blanket”—basically, a zipped-up bag you put your child in. Since there is no real reason to have another kind of blanket, this recommendation seems like a reasonable one to follow.