Ages 2-5BabiesFoodHealth

“Is Breast Truly Best?” What this new breastfeeding study does (and doesn’t) say.

You might have seen in some news outlets reports about a study which apparently shows ‘no benefits’ of breastfeeding, and you’re even more likely to have seen the flurry of responses this study has produced. As is often the case, studies which run counter to current thinking (i.e., that breastfeeding has significant and long-term health effects) are subjected to far more scrutiny and criticism than studies reinforcing the mainstream thinking.

Unfortunately the original study is a) written in quite dense technical language and b) behind a paywall, so what follows is this scientist/med student’s attempt to lay out what the study says, and doesn’t say.

Who performed this study?

Cynthia Colen and David Ramey, from the sociology department at Ohio State University.

Why was it important to do this study?

Studies which directly compare breastfed children with formula fed children tend to find that the breastfed children have better outcomes than formula fed children in some aspects of physical and mental health, intelligence and behaviour. However, it’s well known that children who are breastfed tend to be born into families which are better off in many ways than children who are formula fed. This is for a number of reasons. Firstly, breastfeeding is recommended by health professionals and public health organisations, and better-educated mothers are more likely to be able to seek out (and to trust) this information and these recommendations. Better-educated mothers tend to have children who do better on IQ tests, and following health guidelines about breastfeeding also means you are more likely to follow health guidelines in other areas, such as not smoking around your child and providing a safe sleep environment. Better-off mothers are more likely to have the resources to breastfeed, like being able to stay home for longer, paying for independent breastfeeding support and accessories and choosing childcare which is compatible with continuing breastfeeding, and having money means you can afford high-quality housing, healthy food and regular medical appointments. And both mum and baby need to be in reasonable health for breastfeeding to work – meaning sicker babies are more likely to be formula fed in the first place.

What all this means is that if you compare groups of breastfed and formula fed babies, the breastfed ones are likely to be better off in many ways that are nothing to do with them being breastfed. This gives public health scientists a problem, because it means it’s hard to know how much effort we should be making (and how much of our limited pot of money we should be spending) to get women breastfeeding. It might be that breastfeeding will save us $10.5 billion dollars a year, or it might be that switching to breastfeeding doesn’t address all the other factors that are associated with formula use so doesn’t improve health at all.

So how do we separate the effect of breastmilk from the effect of ‘having the kind of mum who breastfeeds’?

People have tried three things so far. The first is to try and collect all the information about families that might be linked to both infant wellbeing and breastfeeding, to account for them in the stats. This is called ‘controlling for’ something. So, for instance, you might find that breastfed children, on average, own more books, but this effect disappears if you take into account parents’ education and income, meaning that you can say that wealth and education cause both book-ownership and breastfeeding, rather than breastfeeding directly stimulating a love of literature. This method is not perfect because some of these effects are hard to capture with blunt categories of income bracket and educational level, which are the only kinds of things you can examine on large surveys, and when the things you are interested in are very closely related it’s impossible to get rid of the effects altogether.

The other method is ‘randomised controlled trials’, where you take two similar groups of babies and expose only one to an intervention. This would give more reliable results than the first method, but very hard to do for breastfeeding as we know enough about the beneficial short-term effects to make it unethical to tell women not to breastfeed (and it’s unlikely enough would agree to take part to make a study worthwhile). However, a study called the PROBIT trial in Belarus gave half of a group of women an education course about breastfeeding and breastfeeding support, and saw substantially higher rates of breastfeeding in that half. They then looked at all the babies in one group versus the other, thus getting round the fact that the better-off families in the supported group were more likely to be influenced by the education and advice. It’s worth noting that the PROBIT trial also found no effect of breastfeeding on obesity, asthma or child behaviour.

This study tries a third approach – looking at families in which one child was breastfed and the other wasn’t. This means that they can account for the effects of individual family circumstances on health and separate them from the effects of breastfeeding versus bottle-feeding.

Who did the study look at?

This study looked at children of mothers who formed part of the National Longitudinal Survey of Youth, a group of people (‘cohort’), who were representative of the US population in terms of race, income, education and location. The children studied are now between 4 and 14 years old, but the data was collected from them over the course of their lives (i.e., mothers of 14-year-olds weren’t being asked to remember how long they breastfed and how their child did in nursery school).

7,319 siblings, including 1,773 siblings who were fed differently, were included in the analysis. The researchers looked both at whether the children were ever breastfed, and at how long, in weeks, they were breastfed for.

The researchers looked at BMI, obesity, asthma, hyperactivity, behavioural compliance, attachment to parents, and five measures of academic capability in the breastfed and bottlefed siblings.

What were the results?

When looking at all the siblings in the study, the researchers found significant effects of both ever having breastfed and duration of breastfeeding on almost all the things they looked out. When comparing breastfed children with their bottle-fed siblings, all these differences disappeared.

How did the authors interpret their results?

The authors say “Our results suggest that much of the beneficial long-term effects typically attributed to breastfeeding, per se, may primarily be due to selection pressures into infant feeding practices along key demographic characteristics such as race and socioeconomic status.”

What they are saying is that when we think we are seeing long-term beneficial effects of breastfeeding, we are actually observing that better-educated, more well-off people tend to breastfeed, and also tend to have healthier children.

So there’s no point in breastfeeding then.

This is definitely not what the study says! This study specifically looks at children aged 4 and over to study long term effects – the short term benefits of breastfeeding, such as reductions in eczema and stomach infections in breastfed babies are very well known.

But that’s what the Daily Mail said.

Why are you reading the Daily Mail for science or health advice??

The Daily Mail chose the most inflammatory headline they could have done. Nowhere do the text of the paper or the authors suggest that breastfeeding has no benefit. In fact, the lead author’s quote in the article itself directly contradicts the headline! Professor Colen specifically says: ‘I’m not saying breastfeeding is not beneficial, especially for boosting nutrition and immunity in newborns.’

What are the limitations of this study?

It’s possible that the women in this study exaggerated how long they breastfed for, but that’s unlikely as they were asked before their children were 2 years old and we know from other research that those answers are likely to be accurate. There might also be health differences between the siblings which also affected breastfeeding, for example, if one child was born premature and couldn’t nurse. However, as healthier babies are more likely to breastfeed this will generally lead to an overestimate of the benefits rather than an underestimate. This reinforces the main conclusion – that breastfeeding is less beneficial than previously thought – because even given the fact that the breastfed babies might have started off healthier, no long-term health benefits were seen.

The sample size of the differently-fed siblings group was also much smaller than the wider study group, leading to the suggestion that the size of the group just wasn’t big enough to spot these differences. However, the ‘power’ of the sibling group is actually higher, because each mother in effect ‘controls’ for everything for both her children, meaning it’s much easier to account for the effects of factors other than breastfeeding on health.

The ways of describing breastfeeding (yes/no and duration) lack nuance, missing whether the mothers were exclusively breastfeeding, mix feeding or pumping. However, this is not a flaw that is unique to this study – most cohort studies that show beneficial long-term effects of breastfeeding use similarly crude measures.

The study authors also don’t provide the number of children who were breastfed, and the distribution of breastfeeding durations. It’s possible that the lack of beneficial effect is due to the ‘breastfed’ children actually experiencing only minimal breastfeeding. However, this would affect both the full sample (where effects are seen) and the sample of siblings fed differently. Additionally, the authors do state that “prevalence of breastfeeding and our outcomes are similar to those found in nationally representative surveys conducted during the same time period” (Chase-Lansdale et al., 1991; Der et al., 2006).

The study also couldn’t look at a lot of other diseases that had been linked to formula feeding such as high cholesterol, Crohn’s disease and type-2 diabetes, because the oldest children in this study were 14 and these tend to be diseases of late adulthood. However, it’s important to note that formula has changed a lot over the last 50 years so looking at a more recent (and therefore younger) study sample is also a strength of this research.

So what is the take-home message?

This study suggests that as a public health measure, investing in breastfeeding promotion and support might not be as cost-effective as we thought, in terms of decreasing rates of diseases such as obesity and asthma. If we are worried about child health and wellbeing, we should maybe look at addressing some of the factors associated with formula use, such as poverty, poor-quality schooling, lack of affordable daycare and paid parental leave. These innovations are likely to be welcomed by all mothers, whereas advice to breastfeed exclusively for six months is unrealistic and demoralising for many mothers.

It also means that if your formula-fed kid is 4 and fine, you’re probably in the clear.

Previous post

On the Recent Homebirth Statistics Fracas

Next post

Besotted Big Sister Breaks Baby (... not really)



I'm currently trying to finish off my PhD in neuroscience and my medical training. I hope to get a proper job sometime in my 30s. I have a toddler who is, naturally, the bestest, most gifted, prettiest and nicest child to walk this earth, at least until I can persuade my partner to have another. I plan to use this platform to rant semi-coherently about people willfully misunderstanding neuroscience in the service of their favorite parenting soapbox, as well as trawling Medline for interesting stuff, so you don't have to.


  1. March 5, 2014 at 10:14 am —

    In 1976, the Billings (MT) Gazette wrote an article about my mother breastfeeding me called “Burp for Burp, the Breast has the Bottle Beat”:

    Are you saying that an alliterative puff piece might have been overstating the case?

  2. March 5, 2014 at 1:10 pm —

    There are a lot of other issues with breastfeeding vs formula outside the US, especially among lower-income women in poorer countries. Water safety is key with formula use, as is ability to buy sufficient formula (and not have to dilute it). Not easy problems for a woman to address when she’s also worried about transmitting HIV through mixed feeding and has to decide on exclusive breastfeeding or only formula. In such situations breastfeeding promotion is now the public health message of choice for reasons less to do with “is my kid brain-ier on the bottle or breast” and more about survival. That’s not saying I disagree with your article by the way – because I totally agree, especially on the unrealism of saying “breastfeed!!” without any consideration of a woman’s constraints – just supporting the point that the breastfeeding/formula debate is really, really complex…

    • March 5, 2014 at 1:15 pm —

      neopare those are really important issues, and as the ‘mommy wars’ facet of infant feeding seems to be mostly fought by women who have at least the privilege of reliable internet access, they definitely get overlooked. And I think that in situations where HIV transmission or water quality are going to be important considerations in infant feeding, the kind of information and support needed is going to be very specific, and possibly warrant entirely separate consideration from national efforts to increase breastfeeding skills and uptake.

      • March 5, 2014 at 1:24 pm —

        Totally agree. I guess my comment is also because the internet’s reach to every far corner of the globe (and the moon? Mars? Are there readers on Pluto?) means you never know who will read what. Newly pregnant women or new mothers with internet access – a growing population with mobile phones – are apt to google the most diverse things, and the US-focused version of debates tend to dominate English-language hits.

  3. March 5, 2014 at 5:04 pm —

    This was a very cool study with a neat way to get around confounding variables between families!

    I agree that pushing breastfeeding without making it easier for women (especially women with fewer resources) to do so isn’t helpful and I’d imagine potentially harmful. Focusing efforts on reducing the systemic inequalities in education, childcare, and paid leave and other work policies that affect parents of young children would result not only in improvement in measures of health, but make it easier for more women to choose breastfeeding in the first place (if they wish).

  4. March 5, 2014 at 6:05 pm —

    Great article!

    For those with access to clean water and who can afford it, infant formula is an adequate alternative for women who can’t or don’t want to breastfeed. Breastfeeding for me was too demanding and painful and was drastically affecting my emotional well being. For women like me, infant formula wasn’t only adequate, it was actually better than breastfeeding and definitely better than any of the alternatives outside those two choices! I was a better mom when I could finally relax and rid myself of pressures and expectations I could not meet.

  5. March 5, 2014 at 7:18 pm —

    Yes, yes, YES! Especially this:

    “If we are worried about child health and wellbeing, we should maybe look at addressing some of the factors associated with formula use, such as poverty, poor-quality schooling, lack of affordable daycare and paid parental leave.”

  6. March 6, 2014 at 10:16 am —

    Good piece. Far more clearly written than the original paper, too.

    One aspect you’ve omitted is the weird selection of ‘outcomes’ of infant feeding.

    The three health outcomes – obesity, asthma, BMI (actually 2, ‘cos obesity and BMI could be conflatable in this context) – should not be linked to breastfeeding/formula feeding, not in a study which does not( cannot) specify length or exclusivity of breastfeeding. Asthma has long been known as only very weakly (if that…) reduced by breastfeeding, and it would be crazy to think that a very short period of breastfeeding would overcome the known family/environmental influences on obesity (most of the breastfeeding of the ‘breastfeeders’ in the study will have been short and ‘any breastfeeding’ was the definition of breastfeeding) . Breastfeeding advocates should probably stop listing these phenomena as ‘benefits’.

    What struck me even more though was the use of ‘behavioural compliance’, ‘number recognition’ and ‘relationship with parents’ in the list of outcomes. Behavioural compliance??? Really? And lo and behold, ‘any breastfeeding’ makes no difference to it. Who’d a thunk it? Same with the other stuff. Why would we expect ‘any breastfeeding’ to make an isolatable, significant, difference at age 4-14?

    None of that has stopped headlines claiming the ‘experts’ have found breastfeeding is ‘not beneficial’, though.

    • March 6, 2014 at 12:45 pm —

      I think their outcomes were probably partly determined by what’s available in the cohort – IQ might not have been measured but the five measures of educational achievement were; attachment security wasn’t measured but this parental relationship outcome was a good proxy etc.

      Although they don’t take account of the exclusivity they do ask for the length (in weeks), and found a correlation when sibling relationships weren’t taken into account but no correlation when they were, as with the yes/no answers. And the point of doing a very large study like this is that even if the effect is diluted by the duration of breastfeeding being quite short on average, you should still pick it up. So although this way of measuring the extent of breastfeeding is not ideal, it’s not a ‘fatal flaw’ and it’s worth noticing that most cohort studies of infant feeding have the same problems.

      • March 11, 2014 at 9:28 am —

        Yeah, I realise their outcomes had to be selected from what they had available….and it’s another weakness of the study. You end up using someone else’s list, which may or may not be well-defined. I still don’t know why they bothered with behavioural compliance, and the educational achievement measures were unconvincing (though I recognise the massive probs with IQ measurement as well – aaaargh). Many of the outcomes are not touted as ‘benefits’ of breastfeeding, at least not by any source you’d want to take seriously. Parental relationship at ages 4-14 for example – who would posit that relatively short periods of breastfeeding would make a difference over and above all the other bazillion influences on that one? It’s not an especially large study, either.

        It’s not an especially large study.

Leave a reply