New research

If a picture is worth a 1000 words, what is a video worth? Milk with Altitude (Summer 2013) in video form, by my awesome collaborator, Geoff Childs.


Friday, November 1, 2013

What can evolution tell us about iron fortification of infant formula?



I usually avoid blogging about my own work here. Mostly, this is a space for me to explore new topics, or share my excitement over shiny new and cool breastfeeding science, or force my students to show off their own work. However, I recently wrote a paper I think is worth discussing. The paper was this: “Too much of a good thing: evolutionary perspectives on infant formula fortification in the United States and its effects on infant health” soon be published in the American Journal of Human Biology and currently available in Early View. 

Unlike most of my work which is centered on human milk, this paper focused instead on infant formula, specifically iron fortification of infant formula. I applied concepts from evolutionary medicine to fortification practices, and suggested that the current practice of fortifying infant formula with 12 mg/L of iron was excessive. I stand by this, even as I know many clinicians may challenge this, and even last year the Section on Nutrition at the American Academy of Pediatrics recommended universal fortification of breastfed infants out of concern that infants may be at risk of developing iron deficiency anemia. This viewpoint was immediately challenged internally by the American Academy of Pediatrics Section on Breastfeeding and scholars who study infant nutrition. You can read the responses here, and here.

Figure 1: Me, and a wall of infant formula in Cebu, Philippines. I'm five feet one inch tall if you need a scale. Photo by Chris Kuzawa.

Iron deficiency anemia (IDA) is global problem, with approximately 2 billion (yes, with a b) suffering from some form of anemia based on estimates from the World Health Organization. IDA during development is associated with increased infection, mortality, delayed cognitive development, and impairments with growth in weight and length.  It is a terrible nutritional deficiency, and it makes total sense that we would want to prevent and treat IDA as much as possible.  What I suggest in this paper is that in that noble goal, we may have gone too far, and commercial infant formula may contain an excess of iron. 

For most of us living in the United States, we live in high resource, low pathogen environments. Iron depleting infections, especially those caused by intestinal helminthes, are rare.  And iron fortification is quite plentiful for formula fed infants – formulas are typically fortified with 10-12 mg/L of iron, and low iron formulas (4 mg/L) are actually quite hard to find.  Breastfed babies receive milk with much lower levels of iron – about 0.2-0.5 mg/L. While differences seem huge on pixels, all the iron in milk and formula is not bioavailable – about 15-50% of human milk iron is bioavailable and about 7-14% of infant formula iron. The differences actually look like this across infancy, as shown here for an “average” female infant. I have defined average as growing on the 50th percentile of weight for age, consuming the standard recommended amount of formula (ounces per pound) or equilivent amount of human milk. As you can see, the differences in intake are striking. Recommended daily intakes (FDA) are 0.27 mg/day for infants less than 6 months; breastfed infants are meeting these requirements while formula fed intakes are consuming vastly more. 
Figure 2: Iron intake of a typical female infant (assume she is 50th percentile of weight for age and drinks the volume of formula or human milk recommended for that weight. The amount of dietary iron she ingests assumes a 7% absorption rate from formula and a 50% absorption rate from human milk. BOTH may be underestimates. The reference line is recommended infant intake. Note, this graph does not contain information on the leftover iron - the one that may be an all you can eat buffet for gut bacteria.

I hypothesized that this increase dietary iron would be mismatched to infant needs, and may result in an excess of iron. While adults can down regulate iron intake when they are iron replete, infants do not have the same capacity and will continue to absorb dietary iron. This excess iron may increase the concentrations of free radicals, lead to oxidative damage in cells, and most importantly, serve as an iron source for pathogens, increasing the risk of infection. The iron that is not absorbed by the infant (that other 86-93%) will spend some time in the infant’s digestive system before being excreted in feces, and may provide an iron source for pathogenic, iron requiring bacteria such as E. coli. By comparison, the common intestinal microflora of breastfed babies, Lactobacillus and Bifidobacterium, are either iron independent (Lactobacillus) or require minimal iron (Bifidobacterium). These bacteria even contribute to immune responses in breastfed infants AND competitively inhibit E. coli. Everything may shift with too much iron, allowing for increased amounts of iron requiring bacteria, including pathogenic bacteria and even altering the pH of the intestines to support additional pathogenic bacteria, increasing the risk of GI infections and diarrhea. Too much iron – absorbed or not – can have consequences for infant health. 

Elsewhere, it has been argued that maintaining lower levels of bodily iron – not anemic – may be protective against the risk of infection and may an evolved response to minimize infection risk. This actually makes a lot of sense – limiting iron puts the breaks on pathogenic growth and replication and may reduce infection risk. 
In infants, transplacental iron, especially from delayed cord clamping, is sufficient to meet iron requirements for the first several months of life.  Iron levels in unsupplemented infants are quite low at 6 months of life, although few will develop full blown anemia. I argue that these low levels at 6 months may be adaptive – this is the time period when infants will be introduced to foods besides breast milk. Consequently, their exposure to pathogens will increase greatly (it is also the time when they become more mobile, which may also contribute). Having low levels of bodily iron may, as suggested for adults in 1976 (Bullen et al., 1976), be protective against infection.  Infants with lower levels of bodily iron may have been less likely to contract infections or die from them, leading to gradual evolutionary change in how human infants handled iron – and possibly on the iron content of human milk. 

Commercial infant formula with the really high concentrations of iron undermines this normal biological rhythm, and in our important attempts to prevent IDA in infants, we may have overshot the mark. In Europe, the ESPGHAN Global Standards recommend fortification at 4-8 mg/L (Koletzko et al., 2005), and guess what – the incidence of IDA in infants is not higher than in the United States. Several randomized control trials, the gold standard of clinical investigation, have found the same thing – infants receiving formula with 4-8 mg/L of iron do not have increased risks of IDA compared to infants receiving 12 mg/L.  This has been interpreted as evidence that higher fortification levels are safe but it also demonstrates that lower levels of iron fortification are appropriate to meet infant needs.  Too much iron, I suggest may promote the growth of pathogenic bacteria, alter the composition of the microbiome, and may even increase long term risks of Parkinson’s disease. 

Infant formula clearly needs iron fortification. But the current levels of fortification used in the United States may be a case of too much of a good thing. And as suggested below in the comments - the needs of premature babies will be very different, and the model above is for full term infants of appropriate for gestational age (not premature or small for gestational age). 

Author's note: The Alpha Parent has recently discussed a similar topic , and I learned that the Science of Mom had made similar points in 2011 - after the paper had been published.  This project was originally presented as a conference talk in April, 2007 at the American Association of Physical Anthropologists.

Tuesday, October 15, 2013

Milky mismatch: Vitamin D levels in human milk and legacies of past behaviors



I have been thinking a lot about Vitamin D lately. Wrapping up field work at high altitude, coupled with my love of outdoor running means I have spent a near fortune on sunscreen as of late. We also had a baby with early stage jaundice in our study. His parents were understandably concerned, and treated the jaundice with lots of breastfeeding (see ABM treatment protocol here: http://www.ncbi.nlm.nih.gov/pubmed/20387269 ) and sunshine. In fact, he was put in his bassinet outside under a mosquito net every time the monsoon eased up. 

It was a stark contrast to my skin cancer concerns, where twice daily I coated myself in any number of sunblocking chemicals. Reading the labels, the products were safe for infants older than six months . . . below that, ask a physician.  The general recommendation is to keep infants out of the sun and reduce the risk of sunburn and UV exposure. The source of Vitamin D for infants is breast milk (or formula). The Vitamin D in human milk comes from maternal synthesis. 

Vitamin D synthesis by the body requires a UV wavelength of 290-300nm; this is only available when the UV index is above 3. The UV rays absorbed by the skin convert the prohormone 7-dehydrocholesterol into cholecalciferol. This travels via the bloodstream to the liver, where it is metabolized into 25-hydroxyvitamin D. Synthesis is then almost done: the hydroxyvitamin D travels to the kidneys, where it is converted to the metabolically active dihydroxyvitamin D (Vitamin D). Vitamin D aids the body in calcium absorption, and appears to play a major role in regulating insulin, calcium, and phosphorus levels in the body. 

And as most people know, skin color is directly associated with UV absorption and Vitamin D production. Skin pigmentation is determined largely by the amount of melanin – more melanin = darker skin. More melanin results in increased UV deflection which means decreased risk of harmful UV rays being absorbed (and a decreased risk of skin cancer) but increased risk of Vitamin D deficiency at higher latitudes. UV light, and Vitamin D synthesis, is thought to have played a major role in the evolution of skin color, with darker skin colors found around the equator, where there is plenty of sunlight and opportunity to make Vitamin D and UV damage is a bigger risk. Lighter skin colors are found at higher latitudes as the amount of daily and direct sunlight decreases: less melanin increases UV absorption (Antoniou et al., 2009). This may be beneficial in preventing Vitamin D deficiency, including rickets. Some populations, like Inuit, may also supplement through dietary sources of Vitamin D (whale liver anyone?). Sunscreen is incredibly effective at blocking UV rays: a SPF of 8 blocks 95% of the UV; SPF 15 99%.  Other factors influencing vitamin D levels are body size, specifically the amount of body fat individuals may have. Vitamin D is fat soluble. Extra Vitamin D is stored in fat cells, and may not be accessible unless the fat is metabolized.

So how do you get enough Vitamin D without exposure to too much sun? The good news is for most of us, especially during the summer, we get enough in short bursts that our Vitamin D levels are pretty good. On a sunny day, walking to and from your parking space at work or the grocery store or similar is probably enough. The best estimates are 5-30 minutes of exposure, from 10am to 3pm, 2-3 times a week are sufficient to meet most individual’s Vitamin D needs, provided the face, arms, and neck are uncovered. Darker skin tones will need more exposure. There is a handy online calculator where you can put in your data (including latitude) and it will generate an estimate. The human body is remarkably efficient at making Vitamin D: 10,000-20,000 IU can be synthesized in 30 minutes. 

Figure 1: Capacity for Vitamin D synthesis in light skinned (low melanin) individuals by latitude during winter. Image is from: Tavera-Mendoza and White, Scientific American, Nov. 2007, by way of http://www.medicine.mcgill.ca/physio/whitelab/research.htm

However, nursing mothers will need more Vitamin D, as will individuals with limited sun exposure, heavy use of sunscreen, darker skin colors, living at higher latitudes (especially during the winter), higher body fat, and vegetarians. Most milk sold in the United States is fortified with Vitamin D, as are many breakfast foods. Between sunshine and food fortification, most women are likely meeting their own needs. 

But the real question you came for is about babies. Should breastfed babies, especially exclusively breastfed babies, receive Vitamin D supplementation? Or is supplementing mothers with extra Vitamin D an alternative treatment strategy?

Vitamin D deficiency is probably fairly common: Choi et al., (2013) reported a prevalence of 48.7% in Korean infants, with breastfed infants more likely to be vitamin D deficient than formula fed infants, likely reflecting fortification of infant formula with supplemental vitamin D. Similarly high rates of Vitamin D deficiency were reported in Turkish infants (Halicioglu et al., 2012). In the United States, the incidence rate is approximately 25-40% for unsupplemented exclusively breastfed infant. Infants need approximately 400 IU of Vitamin D per day, and based on current estimates for human milk, infants are unlikely to get sufficient Vitamin D from human milk alone. 

 “Despite the association between sunlight exposure and human milk vitamin D concentration, there are no reports of the effect of long-term sunlight exposure of the mother on her milk vitamin D concentration.”  Dawodu A, Tsang RC. 2012 Adv Nutr 3: 353-361. 

However, we do have some evidence: a few studies do exist looking at the relationship between maternal and milk Vitamin D levels, often called antirachitic activity, as the measure includes both the biological activity of Vitamin D and its metabolites. Most of these studies are supplementation studies – providing mothers with additional vitamin D, rather than relying on maternal synthesis. 

One of the first major supplementation studies is that of Hollis and Wagner (2004). Eighteen mothers at one month postpartum were enrolled into one of two treatment groups: 1600 IU D2 + 400 IU D3 or 3600 IU D2 + 400 IU D3. Mothers continued in the study for 3 months when milk antirachitic activity was tested. Both groups showed an increase in milk antirachitic activity: group one had a milk mean of 34.2 IU/L and group 2 a milk mean of 94.2 IU/L. However, neither increase was sufficient to meet infant metabolic requirements. 

This was followed by a study by Saadi et al., (2009).  Working with a sample of Middle Eastern women, Saadi et al., used two treatment groups: one receiving 2000 IU/day of Vitamin D and the other receiving 60,000 IU/month. Mothers reported seven minutes per week of sun exposure, low dietary intakes of Vitamin D rich fish, and had undetectable antirachitic activity in their milk prior to entering the study. Supplementation increased milk antirachitic levels in these women to 50 IU/L (10-63 IU/L), within the range of US women relying only on incidental sun exposure for synthesis. The 50 IU/L levels are considered low, and well below the recommended intake for infants.  

In a large meta-analaysis of available studies on Vitamin D supplementation of mothers as a way of managing infant Vitamin D needs, Dawodu and Tsang (2012) conclude that based on the evidence currently available, it is unlikely that maternal supplementation could increase the antirachitic activity of milk enough to meet infant requirements. 

While human milk is almost always the ideal first food for human infants, that does not mean it meets 100% of needs 100% of the time. Specifically, given that human babies likely had plenty of sun exposure for the majority of human evolutionary history (including as recently as our grandparents and still in many parts of the world) there would have been minimal selective pressure on increasing Vitamin D transfer into milk. Babies, especially in tropical climates and during certain seasons of the year, may have received plenty of sunlight, certainly enough for individual synthesis of Vitamin D. Long term exposure to damaging UVs would have a byproduct, but probably not as important as synthesizing enough Vitamin D to prevent rickets, seizures, and other factors associated with low Vitamin D synthesis. Mothers also, likely had plenty of exposure to sunlight, probably had much higher levels of circulating Vitamin D, and greater amounts of it in milk.  Vitamin D requirements were probably meet by the mutual sun exposure of mothers and infants, and Vitamin D requirements during infancy and childhood may have contributed to selection against melanin at high latitudes and a reduction in skin pigmentation to maximize synthesis.

Figure 1: A mother and baby from Nurbi, Nepal. Babies are typically worn on the back or carried in baskets and receive plenty of daily sun exposure. Photo: Geoff Childs, used with permission.

In evolutionary medicine, we use the term mismatch to describe situations where current behaviors have changed dramatically from similar behaviors throughout human evolutionary history. That is not to suggest some sort of fictionalized single environment that humans are perfectly adapted to, but a general observation about how we likely cared for babies during most of our evolutionary history and even today in many parts of the world, including my field sites in the Philippines and the Himalayas.  Babies and mothers were outside in the sun, and had plenty of opportunities for Vitamin D synthesis . . . and also exposure to harmful UV rates and sunburn. Further, with the continued degradation of the ozone layer, the potential for sunburn and skin damage is high. And Vitamin D supplementation of moms and babies is great solution. 

Mismatch does not have to mean pathology, and this is one of those great situations where understanding why something isn’t present in milk can help us better understand current clinical practice.

References
Antoniou C, Lademann J, Schanzer S, Richter H, Sterry W, Zastrow L, Koch S. 2009. Do different ethnic groups need different sun protection? Skin Res Technol. 15(3):323-9. doi: 10.1111/j.1600-0846.2009.00366.x.

Choi YJ, Kim MK, Jeong SJ. 2013. Vitamin D deficiency in infants aged 1 to 6 months. Korean J Pediatr. 56(5):205-10. doi: 10.3345/kjp.2013.56.5.205. 

Dawodu A, Tsang RC. 2012. Maternal vitamin D status: effect on milk vitamin D content and vitamin D status of breastfeeding infants. Adv Nutr. May 1;3(3):353-61. doi: 10.3945/an.111.000950.

Halicioglu O, Aksit S, Koc F, Akman SA, Albudak E, Yaprak I, Coker I, Colak A, Ozturk C, Gulec ES. 2012. Vitamin D deficiency in pregnant women and their neonates in spring time in western Turkey. Paediatr Perinat Epidemiol. 26(1):53-60. doi: 10.1111/j.1365-3016.2011.01238.x.

Hollis BW, Wagner CL. 2004. Vitamin D requirements during lactation: high-dose maternal supplementation as therapy to prevent hypovitaminosis D for both the mother and the nursing infant. Am J Clin Nutr. 80(6 Suppl):1752S-8S.

Saadi HF, Dawodu A, Afandi B, Zayed R, Benedict S, Nagelkerke N, Hollis BW. 2009. Effect of combined maternal and infant vitamin D supplementation on vitamin D status of exclusively breastfed infants. Matern Child Nutr. 5(1):25-32. doi: 10.1111/j.1740-8709.2008.00145.x.

Wagner CL, Hulsey TC, Fanning D, Ebeling M, Hollis BW. 2006. High-dose vitamin D3 supplementation in a cohort of breastfeeding mothers and their infants: a 6-month follow-up pilot study. Breastfeed Med. 1(2):59-70.
 

Wednesday, September 25, 2013

The importance of thinking about human milk and infant needs



It is very interesting teaching a course on the importance of an evolutionary perspective on health, because as we often see in our in class examples or in news reports, this perspective is often lost in clinical research and in scientific publications. One big one has come to my attention this week, and is worth discussing in light of its limitations and general lack of evolutionary thinking. 

The paper, “Can we define an infant’s need from the composition of human milk?” by  Stam et al., (2013), was published this month in the American Journal of Clinical Nutrition. The authors immediately set up a straw man: milk should meet a baby’s needs, and thus, given the variation in human milk composition, this cannot possibly be true. As a follow-up, they go on to suggest that human milk therefore should not be used as the standard reference for infant formula, stating “The composition of infant formula is presently based on mean values for human milk. It might be better to base the composition on actual requirements of the newborn infant” (Stam: 526S). 

Let’s start with acknowledging that they do in fact have a point here. It would be optimal to feed an infant to their metabolic requirements, and not over- or under-feed. However, what this obscures is the considerable variation in infant metabolic requirements, as evidence by prior doubly labeled water studies (Butte et al., 1990, 1996; de Bruin et al., 1998).  It also assumes that the newborn period alone is sufficient to capture the energy requirements of an infant. It seems quite logical to think that a 7.5 pound newborn will have very different metabolic requirements than a 16 pound 6 month infant (Butte et al., 2000). For requirements to be used as a substitute reference value, you would have to have repeated measures on the same infant – and would likely need to do every infant to come up with an individualized nutritional recommendation. Measuring the total energy expenditure of an individual – including an infant – is quite costly, and requires very specific protocols. 
 
Figure 1: Seven pound, five ounce baby Jesus with your Baby Einstein tapes . . .

The second big issue here is the idea of means as somehow representative of an ideal of human milk, and that deviations from these means are problematic only if you assume uniformity is the goal. And on some levels, we like the idea of nutritional uniformity. That’s essentially what the nutritional information on a package is: the measure of how much fat, energy, protein, etc. is in the food item. And the printed information assumes uniformity per gram. 

Except the error of calculation for the nutritional information on food packaging is in the range of 8-20% (yup, the FDA allows for underestimates of up to 20%! (Urban et al., 2010)). So even products that we might think of as uniform, such as the Fig Newtons on my desk or formula in a bottle, are not uniform.  Means are just that – a value indicating that half the sample distribution falls above and half below that value. An individual with that mean may not even be present in the sample! So in basing reference values for infant formula nutritional composition off a hypothesized “mean milk” value, what is happening is that you are trying to balance over- and under- nutrition. Ideally, we’d use a product tailor made to individual infant needs. Like you for example . . . human milk. But for women who cannot breastfeed or for families where breastfeeding is not the best option, formula is necessary and its composition should be referenced to the natural first food an infant receives: human milk. Matching to TEE, unless collected on each infant at highly frequent intervals, is not going to be any better than matching to mean values derived from human milk. And we know from numerous studies that the composition of human milk changes over the course of lactation (Mitoulas et al., 2001). Perhaps you could make an argument that formula fed infants need more options: a sequence of formulas that are changed as the infant ages. Similar to toddler or follow-up milks, but more specific. 

Stam et al., also question the validity of measures of milk intake in breastfeeding studies. Yes, getting an accurate measure of infant milk intake is hard. Like the physics in the background of The Big Bang Theory hard. And the gold standard method – doubly labeled water is expensive. And if you’re seeing double (doubly labeled water that is) it is because the same method used to measure TEE is used to measure breast milk intake. In a TEE measure, the infant is given the doubly labeled water, in measuring human milk intake, the mother drinks the water. You then collect infant urine over a minimum of 3 days (12 is optimal). In an exclusively breastfed infant, TEE and milk energy intake should be fairly close – and the difference should be in energy allocated to fat deposition. Formula fed infants have higher TEEs compared to breastfeed infants (Butte et al., 1990).

Figure 2: Sheldon attempts to measure infant milk intake and decides to go back to physics. Image: Big Bang Theory, by way of npr.com

And here’s the final thing. Babies have agency. I work with breastfeeding infants primarily, so my exposure is largely limited to them. And in collecting milk samples, we need the infant to nurse. And as any mother can tell you – you really can’t force a baby to nurse. Maybe a newborn. But for most infants, you cannot force a baby to nurse – successful transfer of milk requires anatomical coordination between the infant and the mother. And infants can stop when they get full, or not eat if not hungry – or eat for longer or nurse more frequently if hungry. There is a large body of literature supporting the role of self-regulation in infant intake, and in particular the likely importance of this self-regulation for the development of appetite control and satiety. These factors have been suggested to play a role in the protective effects of breastfeeding on later risk of obesity and related metabolic disorders. 

Finally, and this is something that I think anthropology really brings to the study of human milk in particular and lactation in general, is an appreciation for the fact that human milk has evolved. There have been distinctive selective pressures on human milk – well on all milk (Milligan and Hinde, 2011). This likely includes meeting the TEE of an infant, with some energy left over for storage as fat. If you don’t meet the TEE needs of the infant, the infant is not going to do well – chronic malnutrition is associated with growth faltering, wasting, reduced immune function leading to increased risk of infection, and if prolonged – death. 

And providing too much energy is wasteful – in models of fitness, we often talk about reproductive fitness, and the allocation of energy between competing functions such as reproduction and maintenance. In reproduction, energy needs to be balanced between current and future reproduction. Investing too much energy in a single reproduction – by making milk that is far in excess of infant nutritional needs would be wasteful and may have long term consequences on reproductive fitness. Ergo, there should be relative balance between the TEE of the infant and the energy intake from the milk.  Some aspects of milk will not be always be perfectly matched to infant needs – especially those aspects that may vary based on maternal diet or activity, such as DHA and Vitamin D. For example, milk from mothers in the United States is low in DHA compared to habitually fish eating populations. Many lactating mothers take DHA supplements, and the recent move to fortify formula with DHA reflects knowledge on the importance of DHA for infant development. Too often, we are thinking about human milk nutritional composition as homogeneous, and balking uneasily at the variation found within and between populations in composition. But this variation in composition is actually incredibly important – and only problematic if we think that all infants have a uniform nutritional need and the variation in human milk around some fictional mean is starving some infants and overfeeding others. No; variation in infant TEE is perfectly normal, variation in milk composition is perfectly normal, and infants are not passive consumers of milk but active participants in getting their needs met.*

*Obviously, mothers can limit and control access to the breast, but infants can often compensate by altering intake during suckling. 

References
Agostoni C.  2005. Ghrelin, leptin and the neurometabolic axis of breastfed and formula-fed infants. Acta Paediatr. 94(5):523-5.

Butte NF, Wong WW, Hopkinson JM, Heinz CJ, Mehta NR, Smith EO. 2000. Energy requirements derived from total energy expenditure and energy deposition during the first 2 y of life. Am J Clin Nutr. 72(6):1558-69.

Butte NF, Wong WW, Ferlic L, Smith EO, Klein PD, Garza C. 1990. Energy expenditure and deposition of breast-fed and formula-fed infants during early infancy. Pediatr Res. 28(6):631-40.

Butte NF. 1996. Energy requirements of infants. Eur J Clin Nutr. 50 Suppl 1:S24-36.

de Bruin NC, Degenhart HJ, Gàl S, Westerterp KR, Stijnen T, Visser HK. 1998. Energy utilization and growth in breast-fed and formula-fed infants measured prospectively during the first year of life. Am J Clin Nutr. 67(5):885-96.

Hinde K, Milligan LA. 2011. Primate milk: proximate mechanisms and ultimate perspectives. Evol Anthropol. 20(1):9-23. doi: 10.1002/evan.20289.

Mitoulas LR, Kent JC, Cox DB, Owens RA, Sherriff JL, Hartmann PE. 2002. Variation in fat, lactose and protein in human milk over 24 h and throughout the first year of lactation. Br J Nutr. 88(1):29-37.

Stam J, Sauer PJ, Boehm G. 2013. Can we define an infant's need from the composition of human milk? Am J Clin Nutr. 98(2):521S-8S. doi: 10.3945/ajcn.112.044370.

Urban LE, Dallal GE, Robinson, LM, Ausman, LM, Saltzman E, Roberts SB. 2010. The accuracy of stated energy contents of reduced-energy, commercially prepared foods. J Am Diet Assoc. 110(1):116-123.