Emily Oster

3 minute read Emily Oster

Emily Oster

New Research on Back Sleeping and SIDS

Strong support for putting babies to sleep on their backs

Emily Oster

3 minute read

As befitting a book on babies, Cribsheet has a sizable chunk of content on sleeping. How long kids sleep, when they nap, sleep training and so on. I devote a chapter to safe sleep — notably, sleep position, sleep location and whether you can have stuff in the crib.

In that segment, I talk through the “Back to Sleep” recommendation which says, simply, that babies should be put to sleep on their back to reduce the risk of SIDS. This recommendation is widely credited with big reductions in infant mortality since its implementation in the early 1990s. In Cribsheet I argue that the constellation of evidence strongly supports the back sleeping recommendation. But I also note that there is no single smoking gun data source. The conclusion is based on a combination of small scale studies of SIDS deaths and evidence on variation in SIDS rates over time.

I get questions on this, often from parents who want to know if this evidence is really convincing. Their own parents remember advice to put infants to sleep on their stomachs and they often note that babies sometimes sleep better that way. What I’ve generally been able to say in response is that the circumstantial evidence is very compelling but not completely ironclad.

Then, a couple of weeks ago, a new working paper was released that (in my view) adds substantially to this evidence base, and strongly supports the back sleeping recommendation.

You can find the paper here if you want to look at the whole thing. The authors (Onur Altindag, Jane Greve and Erdal Tekin) do a simple but powerful analysis using data from Denmark. Prior to December 1991, official governmental guidelines suggested infants should be put to sleep on their stomach. After December, 1991 these guidelines abruptly switched to suggest infants be put to sleep on their back or side.

This change in recommendation was implemented in many countries. And, broadly, there have been reductions in SIDS deaths beginning in 1992, in line with this timing. However, they are slow and it’s been difficult to isolate the back sleeping effect from overall time trends.

These authors are able to make use of the fact that, in Denmark, there is a robust home visiting program from nurses. The Home Visiting Program (as it is called) began in 1937 and by 1962 reached most families. There were (are!) multiple home nurse visits in the first year and this is a key way many pediatric guidelines are conveyed. Notably, in this case, the changed guidelines on sleep position were immediately shared by these visiting nurses. This sharp timing change lets the researchers look at whether there is a resulting sharp change in SIDS deaths.

The paper uses a technique economists call “regression discontinuity” but which in lay terms may be better understood (in this case) as “graphing over time.” That is: the authors look at SIDS mortality rates before and after December 1991. The key graph is below.


What you see here is the sharp drop in SIDS mortality rate after December 1991. The drop amounts to, on average, a decrease of 11 SIDS deaths per 10,000 births. The authors also see a drop in overall infant mortality, which is about 13 deaths per 10,000 births (about 17% of the pre-period mean). (This may be slightly larger than the SIDS number because not all SIDS deaths are classified correctly).

Notably — and part of what makes this paper convincing — the graph below shows the no impact on non-SIDS infant deaths. If there was something else driving these improvements over time, we’d expect to see it show up here, and we do not.

The effects of back sleeping are larger for more vulnerable infants (low birth weight, preterm) and for families who likely have fewer resources (e.g. mothers who are less educated). This is consistent with the overall higher risk for these groups, and possibly with the fact that this was a more important source of information for them.

This stands out for me as an important step in really nailing this relationship and in making clear the magnitude. This is not just significant, it’s big. Based on the existing evidence, you should already have been putting infants on their back to sleep. This new data should further convince you that was the right thing to do.

(Final note: this work is a big advertisement for home nurse visiting. I’ve talked about this before, and even written a paper with some suggestive discussion of a possible role for this in explaining differences in mortality risks across the US versus Europe. I wish we had more of these!).

Community Guidelines