Sleep Position and Location

Cutting through the noise#

This is the topic where online discussion is most likely to stop being about evidence and start being about whether you're a good person. One side says co-sleeping is how humans have always slept and the risk claims are confused. The other says there is no safe way to do it and your baby could die.

Both sides are doing the same thing: deciding the answer first, then treating the risk data accordingly. If you want to co-sleep, there's a strong pull to find the one study showing no effect and declare the matter settled. That's not reasoning.

The honest framing: there is a real, small increase in risk. The question is not whether it exists but how big it is, how much you can shrink it, and whether the remaining risk is one you'll accept given what you get in return. That's an uncomfortable way to think, and it's the only way that works.

One anchor before the numbers. We already accept risks constantly — we put children in cars. The first-year death rate from car accidents is around 0.2 per 1,000. Hold that in mind; it makes the sleep numbers legible instead of just frightening.

And the second anchor: sleep choices have real consequences for functioning adults. If co-sleeping is the only way anyone in the house sleeps, then the alternative isn't "safer" — it's a sleep-deprived parent driving a car and managing their mental health badly. Taking care of yourself is part of the responsibility, not a competing interest.

The 60-second recap#

• Back sleeping is the strongest finding in the chapter, and the campaign built on it demonstrably worked.
• A bare, firm, flat surface removes bedding, entrapment, and overlay hazards.
• Sofa sleeping with an infant is dramatically dangerous — the single most alarming number here.
• Bed sharing adds a small risk in low-risk families and a huge one when smoking or alcohol are involved.
• Room sharing has much weaker evidence than the other three, and any benefit appears concentrated in the first few months.
• 90% of SIDS deaths happen in the first four months. This single fact reframes most location decisions.
• Wearable blankets give warmth without loose bedding.
• Back sleeping causes flat head, which is cosmetic and manageable.

What to actually do#

Why back sleeping is the strong one#

Until the early 1990s most babies slept on their stomachs, because they sleep better that way — they wake less. The 1980s edition of Dr. Spock that Oster's parents used says it's "preferable to accustom babies to sleep on the stomach from the start."

SIDS is too rare to study with a normal trial, so the evidence is case-control: find babies who died, find matched babies who didn't, compare. That design is weak in general — but here, four independent lines of evidence point the same way.

1. The case-control finding was extreme. A 1990 BMJ study in Avon, UK, examined 67 SIDS deaths with matched controls:

• 92% of the babies who died had been sleeping on their stomachs
• 56% of the surviving control babies were

That's roughly an eightfold difference — far too large to be explained away by confounding.

2. There's a mechanism. Babies sleep more deeply on their stomachs, and deeper sleep raises SIDS risk. The thing that makes stomach sleeping attractive is the thing that makes it dangerous.

3. A natural experiment. The Netherlands ran a campaign in the 1970s encouraging stomach sleeping, then reversed it in 1988. SIDS rates rose after the first campaign and fell after the second.

4. The US campaign replicated it. "Back to Sleep" launched in 1992, when about 70% of babies were put down on their stomachs. By 1996 it was 20% — and SIDS rates fell alongside.

On side sleeping: the actual danger is the stomach. Side sleeping is discouraged mainly because babies roll from side to stomach. Back sleeping is the recommendation because it's the furthest from the risk.

Once your baby rolls over on their own, don't flip them back. By then the highest-risk window has passed, and a baby who can roll has the head control to move and breathe.

The one real side effect: flat head#

Deformational plagiocephaly has become more common since "Back to Sleep." It's more likely if the baby's head is always turned the same way, and more common in twins and premature babies.

It's cosmetic. No effect on brain growth or function. Tummy time while awake and varying head position help prevent it. The standard treatment is a helmet worn most of the day and night, though there's genuine debate about whether helmets outperform doing nothing.

Bed sharing: the actual numbers#

The best evidence is a 2013 BMJ meta-analysis pooling case-control studies from Scotland, New Zealand, Germany and elsewhere. What makes it useful is that the authors estimated absolute risk for specific combinations of circumstances rather than one average number.

"Low-risk" here means full-term, normal birth weight, breastfed, neither parent smokes, and the mother isn't drinking heavily.

Making the 0.14 mean something: about 7,100 low-risk families would have to avoid bed sharing to prevent one death. For comparison, overall US infant mortality is around 5 per 1,000, and first-year car-accident deaths are around 0.2 per 1,000. So the added bed-sharing risk in a low-risk family is real, is not zero, and is smaller than a risk you already accept every time you drive somewhere.

The dominant factor is smoking. This shows up across the whole literature. The mechanism isn't fully understood but seems to involve secondhand-smoke chemicals interfering with infant breathing — and it gets worse the closer the baby is to the smoker, even when nobody is actively smoking.

Timing matters: the risk is concentrated early. After about three months, there doesn't appear to be elevated risk from co-sleeping if neither parent smokes or drinks.

What you get in return#

Sleep. The most commonly cited benefit, and the honest one. Some babies wake when moved after falling asleep feeding. For Oster's friend Sophie — a doctor with two other children and a working partner — the real choice wasn't between safer and less safe. It was between co-sleeping and nobody sleeping. She knew the guidelines, didn't disagree with them, removed every blanket from the bed, ensured neither adult smoked or drank, and accepted a small residual risk. That's the process working correctly.

Easier breastfeeding? Probably not causal. Bed sharing and breastfeeding are correlated, but women who strongly intend to breastfeed before giving birth are more likely to bed share — so the intent likely drives both. The one randomized trial comparing an attached side cot with a separate bed found no breastfeeding link.

Room sharing: where Oster pushes back#

The AAP recommends the baby sleep in the parents' room for at least six months and ideally a year. Oster's argument is that this specific recommendation isn't supported.

The evidence is thin. A 1999 BMJ study (about 320 deaths, 1,300 controls) found higher risk for babies sleeping alone — but the result flipped depending on whether researchers analyzed the baby's usual sleep location or the most recent one. Usual location showed no risk; last night's showed some. There's no obvious reason for that, which raises the worry that something unusual happened on the final night. The AAP cites this and three similar studies, all small, all sensitive to which variables get adjusted for, and most not actually designed to examine room sharing.

The timing argument is stronger than the study critique. Up to 90% of SIDS deaths occur in the first four months. Sleep location after four months is very unlikely to affect SIDS, and the data agrees — no measurable effect of room or bed sharing after three or four months in non-smoking families. So the second half of that year-long recommendation has essentially no mechanism to work through.

And there's a measured cost. A 2017 study looked at whether sharing a room affects sleep:

That last one is striking — a difference still visible two years later. It may not be causal; perhaps parents move children out once they sleep well. But it's suggestive, and sleep matters for brain development, not just parental convenience.

Also worth knowing: sleep training is very unlikely to succeed while the child is in your room. See Chapter 11.

Oster's conclusion: share a room if you want to, and perhaps the data supports a mild recommendation for the early months. But telling parents to do it for a year — sacrificing short- and long-term sleep for both child and parents, with no clear benefit after month four — may not be good policy.

Bumpers, in proportion#

Bumpers are banned by the AAP and by some cities. But it's worth seeing the scale.

A 2016 Journal of Pediatrics paper counted every US death attributed to bumpers from 1985 to 2012 and found 48. Over that period there were roughly 108 million births and around 650,000 infant deaths. Eliminating bumpers entirely would prevent about 1 in 13,500 deaths.

For comparison, "Back to Sleep" is estimated to have reduced death risk by about 8% — roughly 1 in 13.

So should you use them? Still no. There's a reason they exist (arms and legs getting stuck between slats, which hurts but isn't life-threatening), but older children can use bumpers to climb out, which is its own hazard. The point isn't that bumpers are fine — it's that if you used one before knowing, the actual risk was very small.

Making the choice#

Sorting by how confident you can be:

Settled — just do these:

• Back sleeping
• Bare, firm, flat surface
• No sofa or armchair sleeping
• Don't smoke; secondhand smoke raises risk with or without bed sharing

The real tradeoff, and it's concentrated in the first four months:

• Own bed in your room (safest of the location options)
• Your bed (small added risk, larger if any risk factor is present)
• Own room (possible small early increase, possibly better sleep later)

If you're going to bed share, shrink the risk first:

• No smoking, no alcohol, no sedating medication — either parent
• Strip the bed of covers, pillows, and soft bedding
• Account for your baby: prematurity or low birth weight raises the baseline, so the absolute added risk is larger too
• Recognize the risk is concentrated in the early months and drops after that

If you're not going to room share — Oster's own choice, since it wasn't going to work for her family — do the compensating things: bare crib, monitor, and an explicit plan for exhausted night feeds that doesn't end on a sofa.

Practical playbook#

• Set up the crib or bassinet before the baby arrives. Fitted sheet, nothing else
• Put a bassinet next to the adult bed if night feeding convenience matters
• Decide in advance what you do when you're too tired to stay awake feeding. This is the plan that prevents sofa sleeping, and nobody makes it in the moment
• If the baby falls asleep in a car seat, swing, or carrier outside of actual travel, move them to a flat surface when practical
• Supervised tummy time while awake; vary which way the head faces

Important

Safe-sleep guidance and product safety standards have changed since 2019. Use this chapter to understand how the risks compare, then follow current pediatric and product-safety guidance for your actual setup.

Where the evidence runs out#

• How large the room-sharing benefit is, and how long it lasts.
• How much specific precautions actually reduce residual bed-sharing risk — the studies aren't granular enough.
• Whether consumer breathing monitors reduce SIDS risk. The book doesn't establish that they do.

Our family's notes#

• Sleep surface and room plan for the first four months:
• Our plan for night feeds when we're too exhausted to stay awake:
• Substances or medications to raise with the clinician:
• When we'll revisit room location:
• Our pediatrician's current safe-sleep guidance: