Breast is Best: Intriguing Study, Flawed Reporting

Source: Riskin A, Almog M, Peri R, et al. Changes in immunomodulatory constituents of human milk in response to active infection in the nursing infant. Pediatr Res. 2012;71(2);220-225; doi:10.1038/pr.2011.34.  See AAP Grand Rounds commentary by Drs. Khyati Mehta and Neal LeLeiko (subscription required).

PICO
Question: Among mothers of infected infants, does the immunological composition of their milk change in response to infant illness?
Question type: Descriptive
Study design: Cross-sectional

It's dangerous to criticize a study on breast feeding.  I don't want to come across as someone opposed to the concept; in fact, I enthusiastically endorse the AAP's breastfeeding guidance. 

However, the article by Riskin et al demonstrates pitfalls that happen all too frequently in the medical literature, namely that results are overintrepreted or misrepresented. This can lead to overblown claims in the lay press, similar to the issues I discussed last week. 

The current study posed an interesting question.  The authors wanted to "investigate whether some of the immunologic factors in breast milk change in response to infections in nursing infants."  The problem is, their study wasn't really set up to answer this question, but they proceed (almost) as if it did.  They collected milk samples from 31 lactating mothers of infants < 3 months of age who were hospitalized for sepsis evaluation, and compared these samples to those collected from 20 lactating mothers of healthy control infants.  They found that the breast milk from the mothers with ill infants had higher numbers of macrophages and TNF alpha levels, compared to the control group. 

The problem here is that there isn't any way to know if these differences were a maternal response to the infants' illness, or perhaps from the fact that the mothers had the same viral infection that was causing their infants' fevers.  In fact, 35% of the mothers of sick infants reported being sick as well, compared to none of the control mothers.  The authors recognized this issue in their results and discussion, as did the AAP Grand Rounds reviewers.  However, I think the authors went a little overboard in their conclusions:  "Our findings support the dynamic immunological connection between lactating mothers and their nursing infants, especially during active infection, further encouraging breastfeeding."  I might argue that the first part of this statement is correct, but that the connection might be that the mothers and babies are sharing the same infection, rather than assuming it results from as yet undescribed signals from an ill infant can change breast milk composition. 

It's one word change that has major implications.  If the authors ask whether some of the immunological factors in breast milk change in response to infections in nursing infants, I would say this study didn't answer the question.  If instead they ask whether some of the immunological factors in breast milk change in association with infections in nursing infants, then I would say their study has successfully supported this hypothesis.  It's the difference between cause and effect versus association, and it's an important distinction in medicine.  I learned a new Latin phrase for this: cum hoc ergo propter hoc. 

Writing this post also got me thinking along other lines.  I'm beginning work on a sci-fi short story on an infant who manipulates parental behavior through mind control.  I'm sure my own kids did it to me!