There’s a new study out finding that breastfeeding reduces the risk of childhood leukemia. It’s a meta-analysis, which means it pools together the findings from a dozen and a half different previously published studies. I’ve gone through the details on the numbers below, but I’ve written about the more important issues tied up with this study over at Forbes. Why write about it in two places? It was far too lengthy to cover the problematic aspects of a study like this in social terms as well as the problems in the study itself. So I go over the findings and problems in the study itself here, and I discuss the implications of this study at Forbes, which I hope you’ll read.
But I will give you the spoiler summary: studies like these, especially when they do not really provide any new information (as this one does not), are most commonly used as weapons, even if not intentionally, on women who do not meet the breastfeeding recommendations. It’s yet another study telling us “breast is best” but without providing the social support and the resources that are absolutely essential to successful breastfeeding. And for those moms who cannot breastfeed – regardless of the reason – studies like this make them feel even more inadequate than their physicians, neighbors, fellow moms and the rest of society often already have, an issue Suzie Barston, author of Bottled Up and blogger at Fearless Formula Feeder, regularly addresses.
So, what about this study? It’s honestly nothing new. And as Barston points out on my Forbes post, it’s riddled with the same problems as past meta-analyses (and plenty of past breastfeeding studies in general). The study is a meta-analysis and systematic review, which means the authors, a couple of researchers from the University of Haifa School of Public Health in Israel, searched for all the studies on breastfeeding and leukemia risk that they could find across three large research databases (PubMed, the Cochrane Library and Scopus) and then analyzed the findings together. They only include case-control studies. In this study type, researchers identify a certain number of leukemia cases and then match these to controls, children with similar characteristics who did not develop leukemia. Then they compare the breastfeeding histories of all the children. For this study, the authors found 18 studies published between 1960 and 2014, involving more than 10,000 children with leukemia cases and more than 17,500 matched children without leukemia. (Three previous meta-analyses, all of which contained studies that overlapped with those in this study, found similar results as this one did.)
When they combined the results of all 18, they determined that children breastfed for at least six months had a 19 percent lower risk of developing childhood leukemia than children not breastfed or breastfed for a shorter period. Then they analyzed only the 15 studies with data on children who had never been breastfed (though the meaning of “never breastfed” varied) and found an 11 percent lower risk of leukemia among kids ever breastfed. Analyzing various other combinations (only the highest quality studies, only the high-income Western studies, only the largest studies) found reduced risks falling roughly between that 11 to 19 percent spread. Separating the types of leukemia showed that only the risk of acute lymphoblastic leukemia (ALL) dropped (by 18 percent) from breastfeeding; breastfeeding didn’t change risk for acute myeloid leukemia (AML).
These numbers aren’t surprising. In fact, they’re not new – literally. It’s an analysis of previous studies. “These results are quite consistent with earlier meta-analyses of the association between breastfeeding and leukemia,” said Dr. Alison Stuebe, an OBGYN and assistant professor of maternal and child health at the University of North Carolina School of Medicine who has studied and written extensively about breastfeeding. In fact, “the effect estimates and confidence intervals are essentially unchanged” from the last meta-analysis done in 2007, she said. (Confidence intervals are similar to margins of errors.)
Those numbers are the relative risk, but the absolute risk puts them into even clearer perspective. As I describe at Forbes, those numbers mean a drop from 4 cases per 100,000 children (under age 20) a year to 3.2 cases per 100,000 children per year. For comparison, approximately 2.7 of 100,000 children between ages 1 and 4 die in car crashes each year.
So those are the numbers. How much can we rely on them? If there is anything we can say about the breadth of research on breastfeeding, it is that it is incredibly difficult to find definitive answers about benefits of breastfeeding or effects of formal feeding. A meta-analysis is only as good as its individual studies. The authors note that not all the studies clearly differentiated between exclusive and partial breastfeeding, the data came from what moms reported (sometimes years after their children were born), and response rates varied greatly. But the biggest issue is that so many different factors determine whether moms breastfeed and so many different factors may or may not contribute to leukemia risk – we still know very little about what causes it – that it’s incredibly difficult for researchers to think of and control for those (possibly interacting or overlapping) factors. Each of these individual studies accounted for some of these factors, such as socioeconomic status or factors related to mothers’ health, but none accounted for enough factors. Even if a couple had, this meta-analysis only used the raw results from each study – the calculations before adjustment for other factors. In the researchers’ assessment of the studies’ quality, they rated the “Comparability of Cases and Controls,” which implies that cases and controls are similar, but there are tons of differences between women who breastfeed and those who do not.
In reality, the number and variety of these possible underlying differences – confounders – between women who breastfeed and those who didn’t is so great that it’s darn near impossible to parse out what could be contributing to what. Even when researchers account for the biggest of these confounders – factors such as the family’s socioeconomic status, whether mom smokes, mom’s weight, a family history of cancer and so on – they may not be able to completely remove the effect of those factors (something called “residual confounding”). Math can only do so much to make up for all the ways being poor, for example, can influence a person’s risk of certain illnesses. Then there are the unknowns, such as details about diet and environmental exposures, that are too complex to catalogue and calculate.
That doesn’t mean that breastfeeding doesn’t reduce leukemia risk. It very well might. If it does, there needs to be a way that breastfeeding can biologically affect leukemia risk, and the authors offer a couple possibilities: antibodies in the breastmilk might reduce inflammation or positively affect the bacteria makeup in a child’s gut. Or stem cells in the breastmilk might help. Or changes to gut bacteria from eating formula might affect a child’s risk. The authors discuss these and related possibilities a bit more in detail in the study, but there are many other possibilities they did not address at all, including whether breastmilk exclusivity matters. Since the study did not use exclusive breastfeeding to six months, infants may gain the benefit of reduced leukemia risk if they receive both breastmilk and formula, “a type of breastfeeding more women are finding it possible to do,” Barston said. She suggests more research into the possible biological mechanisms involved: “Is it something in formula or simply something that formula is lacking? Could introduction of solid foods have something to do with it? What is so important about the 6-month mark? Why not 3 or 4 or 5 months?”
So, more research is needed, as usual, but not just because we want to understand it for science’s sake. We need to understand it for the babies who cannot be breastfed. “I actually think it would be amazing if they did discover some factor in breastmilk that could reduce cancer. It’s certainly plausible,” Barston told me, and I agree with her. But she added – and I agree – “But our goal should be isolating that factor in a way that we might recreate it, or at least mitigate the risks to babies who cannot receive their mothers’ milk.”
And in the meantime, we don’t need more meta-analyses, with crude-number results of questionable clinical value, concluding that breastfeeding reduces risk of XYZ and offering no new information about how or why. We certainly don’t need studies claiming that breastfeeding is a “highly accessible, low-cost public health measure,” the conclusion I primarily addressed in my Forbes piece. What we mothers need is support. Lots and lots and lots of it. Immense amounts of it. And so far, we’re not getting it.