Monday, March 17, 2014

Mother’s milk/Baby’s Water: Selective pressures of infant hydration on milk composition and infant feeding structure in human populations

Human milk is, on average, approximately 85-90% water (Hinde and Milligan, 2011), reflecting our long evolutionary history as primates with dilute, high sugar milks. The reference composition data for milk describes milk as approximately 4.0% fat, 1.2% protein, and 7.2% sugar, although both individuals and populations will vary in the distribution of these macronutrients.

Human milk is considered dilute when compared to that of other mammals. The best interpretation of this comes from the work of Ben Shaul (1962) who described humans as a high contact, high frequency nursing species. Other primates, with a few exceptions, also have dilute milks (Hinde and Milligan, 2011).  As with humans, the dilute composition of primate milk is thought to be driven by distinct patterns of infant care – high levels of contact, high nursing frequency, and low volume transfer per bout. Environmental factors, such as dry and arid environments and elevated rates of whole body water turnover, also predict dilute milks (an example would be camels).

However, within human populations, much less is known about the ways in which the environment may influence the composition of milk, particularly the amount of water in the milk. For many species, geographical ranges are quite limited, and there is little variation in ecological stressors such as temperature or aridity.  Humans are among a small number of primates with a wide geographical distribution, and occupy a diversity of environments. These environments present any number of ecological stressors (temperature, aridity, altitude, humidity, pathogens) that may influence breastfeeding behaviors or milk composition.

Human infants are at increased risk of dehydration compared to older children and adults. Infants have higher whole body water turnover rates and immature kidney function, limiting their capacity to reserve water should intake decrease or water loss (often associated with diarrhea) increase.  However, the majority of discussions on milk composition and breastfeeding behaviors have largely ignored maintaining infant hydration as a possible selective pressure (Bentley 1998). Is it possible that hydration has been an important factor in shaping breastfeeding behaviors, and maybe even some aspects of milk variation, in humans?

“We may conjecture that at different climatic extremes different nutritional priorities are placed upon breastfeeding. So in tropical or arid conditions, one might presume that high evaporative water loss would place the emphasis on maximizing water throughput to prevent dehydration.” – Michael Woodridge (1995).

Looking at population diversity in nursing frequency among traditional foraging societies, Woolridge (1995) reports increased nursing frequency among mothers living in arid or semi-arid environments. One of the best examples of this are the !Kung, a population of foragers living in the Kalahari desert (Figure 1). The !Kung have very high nursing frequencies – on the order of 4 times an hour! Other foraging populations living in arid or semi-arid conditions also show increased nursing frequency, although there appears to be variation within these climate extremes.

Infants who feed with a high frequency but short duration may be biologically altering the composition of the milk. Daly et al., (1993) and more recently Kent et al., (2006) have shown that the amount of fat in milk (they did not measure milk lactose) changes based on the frequency of feeding. More frequent feeds predict lower overall change from the end to the beginning of the next feed, and closely spaced feeds may have an overall effect of maximizing water transfer and reducing the amount of fat in the milk. The infant must nurse more frequently to meet energy and fat needs and may, over the course of the day, take a large quantity of milk. 

Biologically, this might be incredibly important in thinking about the ways in which differences in milk composition are often interpreted as differences in milk quality, especially in older published studies. Discussions of human milk quality are often synonymous with energy and fat, completely ignoring the fact that in certain environments where dehydration is a risk and total body water turnover rates high, low fat milk with a high water content might be adaptive. This milk will keep the infant drinking regularly, and may be important for maintaining physiological well-being and preventing dehydration. As discussed above, total daily fat intake by the infant should be equivalent to that reported in other populations because of the overall increase in milk volume consumed by the infant. However, if we were to simply look at the milk composition without the added context of nursing frequency or total volume of milk consumed over the course of the day (and these are separate issues), it might appear that the infant was not getting enough to eat. 

Figure 1: A Bedouin mother struggles with a tent and a baby. Image:AFP, by way of

One other interesting piece of support for this dehydration hypothesis comes from studies looking at milk composition across seasons, comparing summer and winter, or rainy and dry seasons for example. Few studies are available, and observations are confounded by the fact that in many tropical and sub-tropical populations, seasonal workloads and pathogen risk will vary greatly. Yagil (1986), in a sample of Bedouin mothers, reported a modest increase (+4%) in the percentage of water in milk during the summer compared to the winter (and a decrease in milk fat), although the collection methodology was somewhat problematic. 

While the case is far from solved – and really, the population level data on aridity and breastfeeding frequency may not exist – it remains an interesting hypothesis for some of the possible variation in milk composition and breastfeeding behaviors. Moreover, it may be important to think about changes to milk composition across seasons – is milk more dilute during hot or arid months, promoting increased nursing to maintain hydration? Is some of the regional variation in human milk composition and nursing frequency driven not by hunger but by thirst?

Bentley, G.R. (1998), Hydration as a limiting factor in lactation. Am. J. Hum. Biol., 10: 151–161. doi: 10.1002/(SICI)1520-6300(1998)10:2<151::AID-AJHB2>3.0.CO;2-O

Daly SE, Di Rosso A, Owens RA, Hartmann PE. (1993) Degree of breast emptying explains changes in the fat content, but not fatty acid composition, of human milk. Exp Physiol. 78(6):741-55.

Hinde, K. and Milligan, L. A. (2011), Primate milk: Proximate mechanisms and ultimate perspectives. Evol. Anthropol., 20: 9–23. doi: 10.1002/evan.20289

Kent JC. (2007) How breastfeeding works.J Midwifery Womens Health. 52(6):564-70.

WoolridgeM(1995) Baby-controlled breastfeeding: Biocultural implications. In P Stuart-Macadam and KA Dettwyler (eds.): Breastfeeding: Biocultural Perspectives. New York: de Gruyter, pp. 217–242.

Yagil R, Amir H, Abu-Rabiya Y, Etzion Z (1986) Dilution of milk: A physiological adaptation of 
mammals to water stress. J. Arid Environ. 11:243–247.

Sunday, March 2, 2014

Signifying Nothing: A Response to “Is Breast Truly Best?

As most of you know, there has been a lot of public attention paid to a recent piece published in Social Science and Medicine, titled “Is Breast Truly Best”, by Cynthia Colen and David Ramey. The popular press has (predictably) seized on the story, with outlets from the Daily Mail to Yahoo News to Slate running stories.  The public (through the press) is getting its information from a press release issued by the author and university. But what does the paper actually say? A few blogs have looked at the accepted manuscript and have some extremely intelligent responses (inbabyattachmode: here, Evolutionary Parenting: here).  But you are here, and therefore you get my take: for most of the main points being promoted in the media there is actually nothing new in this paper. Further, the word choice is actually sabotaging legitimate dialogue in the area by using false dichotomies and inflammatory language rather than appropriate professional means of sharing their findings. For more on this, check out Melanie Martin’s excellent guest post on MammalsSuck . . .Milk

Figure 1: Time magazine cover. From Time's website.

First, the substance of the paper. The authors investigated 11 different outcomes (obesity, BMI, asthma, hyperactivity, paternal attachment, behavioral compliance, reading comprehension, vocabulary recognition, math ability, memory based intelligence, and scholastic competence) in a large sample of children from the National Longitudinal Study of Youth, using the 1979 Cohort. The sample was limited to only those children from ages 4 to 14 between 1986 and 2010, and excluded multiples. The final sample size was 8237 children from 4071 families; sample size of the discordant-sibling (i.e. one was breastfed and the other bottle-fed) subset was quite a bit smaller at 1773. As an important statistical point, the tables report in per person years rather than individuals, so the full group has a per person years of 35,572 and the discordant sibling group has 7663 person-years. The statistical analysis used nested regression, which allows for longitudinal measurements and multiple measurements to be nested within individuals and then these individuals (the kids) to be nested within their mothers. 

Reviewing the full paper could take pages and pages, and for the purposes of this blog, I am going to focus on only the physical health outcomes- which have attracted by far the most attention- of BMI, overweight/obesity, and asthma. Very simply, there is nothing in the findings that has not been part of the scientific dialogue, although perhaps not the public conversation, about breastfeeding for years.
Although the paper contains a literature review of current studies on breastfeeding and long term (long term here meaning only ages 4-14!) on child health, what is the most striking is what is omitted.  In particular, the authors fail to mention several large studies previously reporting the same findings- modest or no protective effect of breastfeeding on BMI or obesity. 

The largest of these is Fall et al., (2011), reporting in a sample of 10,912 individuals (ages 20-45) pooled from 5 low and middle income countries (Brazil, Philippines, Guatemala, India, South Africa). They found  no clear evidence for a protective effect of breastfeeding on BMI or risk of overweight in adulthood in this adult group using both an ever breastfed and a duration of breastfeeding measure collected while the adults were still breastfeeding in infancy and early childhood.  While the international study was not without its weaknesses -- the number of participants who were never breastfed was quite small, as was the number of individuals who were obese – the finding still stands: they reported no protective effect from breastfeeding.  And unlike the majority of studies reviewed and critiqued in the Colen et al., piece, Fall et al., (2011) is not hampered by the association between socio-economic status and breastfeeding rates found in the United States-- in these other populations, breastfeeding is most common among individuals with lower SES. In other words, the findings for BMI and overweight which have caused such a stir have been part of the literature for several years but the Fall article and many others have not received the same kind of media attention as the Colen and Ramey piece.

Additionally, the Colen and Ramey piece is missing essential data, undermining what new information it might otherwise contribute to the knowledge base. The authors conclude that increased breastfeeding duration as not protective, based on maternal recall collected within a few years of the cessation of breastfeeding. What we are not told are the mean duration of breastfeeding for any group (all, siblings only, discordant siblings) or the number of individuals breastfeeding or using formula or using both (yes/no). These values do not exist in the publication, and for interpretation, these values are essential. They should be at the bottom of Table 2. Without them, we have no way of knowing if we are looking at a sample with a large or small number of breastfeeding women and how long these mothers breastfed for. Breastfeeding for two weeks may not be that biologically different from never breastfeeding in regards to long term outcomes of overweight or obesity, and without these data, we have no way to know how these samples compare to one another or to the general population. 

Another issue with the Colen and Ramey paper is that it conflates these measurements with health outcomes (this conflation is common in research). One of the known limitations of using BMI or obesity as outcome measures is that they are only an approximation of actual physiology. For example, with the so-called “thin-fat” phenotype, you have thin (by BMI and weight standards) individuals with many risk factors for metabolic diseases normally associated with overweight, such as high blood pressure, insulin resistance, or other measures. Most infant-feeding studies do not look at these more direct measures of health outcomes, as it is practically problematic. Several studies which do delve deeper have reported protective effects of breastfeeding on metabolic function with dose dependent effects. For example, Singhal et al., (2002) in a study of several hundred pre-term infants actually randomized infants to receive breast milk or formula (usually impossible).  Individuals who received breast milk had lower leptin levels (a hormone produced by fat cells) and less insulin resistance at age 8 than individuals who received formula despite similar BMIs.  Infants who received breast milk (these infants were fed expressed donor milk) were metabolically healthier without having lower BMIs.
Figure 2: Both men have the same BMI, but not the same health risks. Image:

Finally, when it comes to asthma, this paper simply restates a conclusion which has been known in the community for more than eleven years. Sears et al., (2002) publishing in the Lancet, reported no protective effect of asthma on breastfeeding.  Subsequent studies (Grabenhenrich et al., 2014; Nwaru et al., 2013) have provided additional support for this finding- in fact, the majority of published evidence finds no protection against asthma by breastfeeding. Some of this may reflect genetic contributions to asthma risk (Ober and Yao 2011); these risks and related household risk factors which would likely be shared by siblings. The bottom line is that no one versed in the literature would claim that breastfeeding clearly had any protective effect against asthma, so Colen and Ramey’s breathless revelation is not a revelation at all – it is another piece of the story. And this story is not a game of Tetris where their magic brick will make the line collapse. It is simply another piece supporting a robust foundation that breastfeeding is not protective against asthma.
Figure 3: Tetris screen capture illustrating what this study was not.

In some ways, the study essentially asserts for BMI, obesity, and asthma that if breastfeeding is not a cure for this, then the health benefits must be overblown.  Further, in this paper, the author’s real point – that by focusing on breastfeeding as a protective measure against these factors we are ignoring the importance of other early life factors, such as quality day care, paid maternity leave, and health insurance is lost in the furor of “ is breast truly best” without ever actually measuring if breast milk is best for infants. The paper associates those who study the advantages (and limits) of breastfeeding with promoting a “cult of total motherhood”—a phrase sufficiently insulting and inflammatory as to derail any attempt to rationally discuss the author’s findings.        

Ironically, there is nothing in Colen and Ramey’s scientific conclusions which merit their provocative title question of whether “breast is best”. There is no question that regardless of whether there are or are not significant long-term benefits, the short-term benefits of breastfeeding (such as reduced risk of infection, especially diarrheal illness) are still tremendous. Recent publications put the cost savings of these reduced infections at more than a billion dollars (Bartick and Reinhold, 2010) with a reduction of nearly a thousand infant deaths in the United States alone – and this is for the 21st century!  

As the paper itself concludes, there are larger structural issues that limit access to paid maternity leave, quality day care, and even structured and protected time and space during work to pump can have huge influences on the ease or difficulty of breastfeeding, and therefore whether it is practical for a given woman to breastfeed. No informed participant in the current discussion would argue against a position that effective improvement in breastfeeding rates depends on both informing parents of the advantages of breastfeeding and making the choice to do so a practical option for more families. However, the authors appear to think that ‘breastfeeding advocacy’ is a zero-sum game: that anyone who touts the advantages of breastfeeding must inherently be shortchanging the social and economic factors which determine whether a woman will do so. As Martin points out over on MammalsSuck . . . good breastfeeding science is important to all mothers, not just those breastfeeding. Good research into the health benefits of breast milk and on the composition of that milk are important for promoting the health of all babies, as knowing what is in milk is necessary for making high quality formulas. This is not a debate of either or, and in doing so, it does a disservice to the scientific community and to mothers. 


Bartick M, Reinhold A. (2010) The burden of suboptimal breastfeeding in the United States: a pediatric cost analysis. Pediatrics. 125(5):e1048-56. doi: 10.1542/peds.2009-1616. 

Colen CG, Ramey DM. (nd) Is Breast Truly Best? Estimating the Effects of Breastfeeding on Long-term Child Health and Wellbeing in the United States Using Sibling Comparisons. Social Science & Medicine, 2014; DOI: 10.1016/j.socscimed.2014.01.027

Fall CH, Borja JB, Osmond C, Richter L, Bhargava SK, Martorell R, Stein AD, Barros FC, Victora CG; COHORTS group. (2011) Infant-feeding patterns and cardiovascular risk factors in young adulthood: data from five cohorts in low- and middle-income countries. Int J Epidemiol. 40(1):47-62. doi: 10.1093/ije/dyq155. 

Grabenhenrich LB, Gough H, Reich A, Eckers N, Zepp F, Nitsche O, Forster J, Schuster A, Schramm D, Bauer CP, Hoffmann U, Beschorner J, Wagner P, Bergmann R, Bergmann K, Matricardi PM, Wahn U, Lau S, Keil T. (2014) Early-life determinants of asthma from birth to age 20 years: A German birth cohort study. J Allergy Clin Immunol. pii: S0091-6749(13)01860-5. doi: 10.1016/j.jaci.2013.11.035. [Epub ahead of print]

Metzger MW, McDade TW. (2010) Breastfeeding as obesity prevention in the United States: a sibling difference model. Am J Hum Biol. 22(3):291-6. doi: 10.1002/ajhb.20982.

Nwaru BI, Craig LC, Allan K, Prabhu N, Turner SW, McNeill G, Erkkola M, Seaton A, Devereux G. (2013) Breastfeeding and introduction of complementary foods during infancy in relation to the risk of asthma and atopic diseases up to 10 years. Clin Exp Allergy 43(11):1263-73. doi: 10.1111/cea.12180.

Ober C, Yao TC. (2011) The genetics of asthma and allergic disease: a 21st century perspective. Immunol Rev.  242(1):10-30. doi: 10.1111/j.1600-065X.2011.01029.x.

Sears MR, Greene JM, Willan AR, Taylor DR, Flannery EM, Cowan JO, Herbison GP, Poulton R. (2002) Long-term relation between breastfeeding and development of atopy and asthma in children and young adults: a longitudinal study. Lancet 360(9337):901-7.

Singhal A, Farooqi IS, O'Rahilly S, Cole TJ, Fewtrell M, Lucas A. (2002) Early nutrition and leptin concentrations in later life. Am J Clin Nutr. 75(6):993-9.