Correlation, causation, and breastfeeding

An interesting piece in the Atlantic on breastfeeding. Here’s a snippet:

One day, while nursing my baby in my pediatrician’s office, I noticed a 2001 issue of the Journal of the American Medical Association open to an article about breast-feeding: “Conclusions: There are inconsistent associations among breastfeeding, its duration, and the risk of being overweight in young children.” Inconsistent? There I was, sitting half-naked in public for the tenth time that day, the hundredth time that month, the millionth time in my life—and the associations were inconsistent? The seed was planted. That night, I did what any sleep-deprived, slightly paranoid mother of a newborn would do. I called my doctor friend for her password to an online medical library, and then sat up and read dozens of studies examining breast-feeding’s association with allergies, obesity, leukemia, mother-infant bonding, intelligence, and all the Dr. Sears highlights.

After a couple of hours, the basic pattern became obvious: the medical literature looks nothing like the popular literature. It shows that breast-feeding is probably, maybe, a little better; but it is far from the stampede of evidence that Sears describes. More like tiny, unsure baby steps: two forward, two back, with much meandering and bumping into walls. A couple of studies will show fewer allergies, and then the next one will turn up no difference. Same with mother-infant bonding, IQ, leukemia, cholesterol, diabetes. Even where consensus is mounting, the meta studies—reviews of existing studies—consistently complain about biases, missing evidence, and other major flaws in study design. “The studies do not demonstrate a universal phenomenon, in which one method is superior to another in all instances,” concluded one of the first, and still one of the broadest, meta studies, in a 1984 issue of Pediatrics, “and they do not support making a mother feel that she is doing psychological harm to her child if she is unable or unwilling to breastfeed.” Twenty-five years later, the picture hasn’t changed all that much.

And here’s a snippet of econ-evidence, reaching a similar conclusion.

This entry was posted in Health economics. Bookmark the permalink.

3 Responses to Correlation, causation, and breastfeeding

  1. derrida derider says:

    Oh dear, Andrew – expect to be descended on by the thought police.

    This is one of those topics where people have tremendous emotion invested – some really mix it up with their identity as mothers. A cold, hard assessment of the evidence is just not gonna make anyone happy.

  2. Sinclair Davidson says:

    You’re too kind DD to describe these people as the ‘thought’ police – they never had a thought in their whole lives. Andrew the PC police are going to take you away. You’ll be an unperson soon. 🙂

    On a serious note, congratulations. I see you’re the 75th ranked economist less than 10 years out.

  3. Peter Fyfe says:

    Thought police, statistics, and “lactivists” not withstanding, breast milk is still the only 100% purpose-suitable food for babies: 100% effective, cheap, and extremely low risk, but really bad for western economies and professional cartels hell bent on medicalising or commodifying every aspect of human existence even as they forget they are mammals, and what that means by their own definition.

    [turns with paranoia from the computer]

    What’s that at the door?

    Put down that syringe, put down that spreadsheet…


Comments are closed.