Proper risk analysis

2010 February 6
by Daniel Lakeland

As some of you know, my wife is pregnant with our first child (a boy!). As we eagerly await his zeroth birthday, we spend our time between working on our "day jobs" and working on figuring out what to do when the baby comes.

An interesting question is how to care for a child at night. A certain number of people advocate keeping the child very nearby, either in a "co-sleeper" type crib directly next to your bed, or in the adult bed, with proper precautions. If you go to hospital classes on child care, you will receive a pamphlet in which they will absolutely and unequivocally state that an adult bed is not a safe place for a child due to the risk of suffocation. The truth is somewhat less straightforward. Children do die tragically via suffocation in adult beds, but children also die tragically due to suffocation in cribs. The proper analysis is to determine the relative risk of the two situations. In other words, how much more likely is the child to die tragically in one situation than the other.

Tina Kimmel, a PhD student in social welfare at UC Berkeley, published an article in Mothering magazine where she does a simple analysis of the relative risk by combining the infant death rate data with the prevalence of co-sleeping vs crib sleeping data. Her analysis is not perfect because the data is not perfect, and she makes some assumptions about what the proper rate of prevalence is for non-quantitative survey responses like "sometimes" or "always". But the results of her analysis show that crib sleeping is 2.4 times more dangerous than bed sleeping, not, as the pamphlet says, the other way around.

Unfortunately, although she's done a good job of collecting the data, she doesn't give any estimates of uncertainty on these numbers, and I haven't tried to reproduce her analysis. But I imagine with such a strong contrast that other analyses are unlikely to find a strong reversal of this information. We may find that they are similarly risky (ie. 1:1 risk rate) or that perhaps bed sleeping is somewhat more risky (ie. 1.25:1 against beds) or that crib sleeping is even worse than her prediction (ie. greater than 2.4:1 against cribs)  but the point is that dire warnings on specially printed pamphlets are certainly not even close to born out by data. They are, however, funded by the manufacturers of cribs.

http://www.mothering.com/how-stats-really-stack-cosleeping-twice-safe

No comments yet

Leave a Reply

Note: You can use basic XHTML in your comments. Your email address will never be published.

Subscribe to this comment feed via RSS