Evaluation of meat as a first complementary food for breastfed infants: impact on iron intake


KM Hambidge, University of Colorado Denver, 12700 E 19th Avenue, Box C225, Aurora, CO 80045, USA. E-mail: Michael.Hambidge@ucdenver.edu, Phone: +1-303-724-3261, Fax: +1-303-724-3206.


The rationale for promoting the availability of local, affordable, non-fortified food sources of bioavailable iron in developing countries is considered in this review. Intake of iron from the regular consumption of meat from the age of 6 months is evaluated with respect to physiological requirements. Two major randomized controlled trials evaluating meat as a first and regular complementary food are described in this article. These trials are presently in progress in poor communities in Guatemala, Pakistan, Zambia, Democratic Republic of the Congo, and China.


The principal objective of the present article is to provide an overview description of two major efficacy trials currently in progress that are evaluating the use of meat as a first and major complementary food for older infants and toddlers. Though the studies were designed with providing adequate dietary zinc as a principal objective and with linear growth velocity as the primary outcome measure, the magnitude of the effects of early meat consumption on iron status in populations at high risk of iron deficiency render the effects of meat intake on iron status a major secondary outcome measure. The extent of the theoretical benefits of meat consumption on the quantity of iron absorbed and the extent to which this is sufficient to meet iron requirements is considered here.


Much has been gained, though not without some risk and unresolved challenges, from the judicious use of iron supplements, iron-containing Sprinkles™, and iron-fortified food staples in both low-income and more affluent populations. However, even at a time in which these programs are being rapidly scaled up in low-income countries, recognition has been given to the fundamental importance of promoting the availability of locally available, affordable food sources of all essential nutrients, including iron.1 In times of severe food insecurity, emphasis on dietary diversity may need to give ground to the immediate threat of energyinadequacy, but recognition that dietary diversity is essential for successful reliance on non-fortified/supplemental foods should be maintained. The outstanding challenge to achieving adequate food diversity in young children is the provision of animal products, especially meat. Iron deficiency itself remains the single most prevalent nutrient deficiency globally. Biofortification of local food staples may eventually offer a partial alternative to meat as a local source of iron, especially in vegetarian communities, but this is not a current reality.

With respect to nutrient composition of the diet, the value of including meat as a complementary food from the age of 6 months is indisputable. Among the micronutrients of key public health importance, this applies especially to zinc, iron and, frequently, vitamin B12. The concentrations of zinc in meats are especially favorable. One ounce of beef provides approximately 2 mg of zinc with favorable bioavailability.2 If provided daily, this leaves only 0.5 mg of bioavailable zinc/day to be obtained from all other dietary sources in order to meet the estimated average requirement (EAR), or an extra 1 mg zinc/day to meet the recommended dietary allowance.3 In the two international studies described in this article, the selection of linear growth as the primary outcome measure was influenced heavily by the strong association between linear growth and morbidity/mortality globally.4 Linear growth may benefit from the multiple nutritional improvements that result from adding meat, on a regular basis, to the complementary feeding provided to breastfed infants from the age of 6 months. However, the association between zinc nutrition and linear growth5 and the inadequate amounts of zinc in the diets of the populations being studied were outstanding factors. Of relevance to the four-country study described below, pilot growth and diet studies in toddlers at the four study sites revealed a statistically significant positive association between linear growth, Z-scores, and consumption of meat or/and eggs at least three times per week.6

Though anemia and indices of iron status are only secondary outcomes in the two studies described below, this should not detract from the interest and importance attached to them. Iron concentrations in beef are no more than half those of zinc,2 and one ounce of beef per day provides approximately 1 mg iron/day. To place this figure in perspective, 70% of that iron needs to be absorbed to meet the physiologic requirement of 0.69 mg iron/day determined by the Institute of Medicine3 for infants 6–11 months of age. This appears to be an unlikely achievement with the EAR set at 7 mg iron/day for older infants.3 However, the EAR was based on the assumption that the diet of older infants did not include meat; it was, therefore, based on the absorption of inorganic iron, for which a figure of 10% absorption was applied. The iron derived from meat is primarily heme iron, the absorption of which, in adults, is very favorable3,7,8 in contrast to that of inorganic iron. For example, in non-pregnant adult women of child-bearing age with moderately low ferritin, heme iron absorption has been estimated to be 34 ± 2% SE, with individual values potentially in excess of 60% (JA Hunt, personal communication). Data on absorption of heme iron, specifically, in later infancy would be invaluable.

Important physiologic gains may be achieved even if absorption levels do not reach estimated requirements. For example, even 1 mg of inorganic iron/day given in a wheat product has been shown to increase the amount of hemoglobin in young children.9 This is consistent with the hypothesis that even smaller quantities of beef will have beneficial physiologic effects in iron-deficient older infants/toddlers even if optimal iron stores are not achieved. This could be especially relevant when other sources of meat are used in complementary feeding. Lean pork and chicken leg have only about half of the iron concentration of beef, and white meat from poultry has even less.2

The importance of incorporating animal source foods (ASF), including meat, into complementary feeding has been emphasized by investigators,10–14 by international organizations, including the World Health Organization,15–17 by national ministries of health, as in Guatemala18; and by national committees, including those in the United States.19–21 In the United Kingdom, meat consumption was found to be positively associated with psychomotor outcome in children up to 24 months of age22 and with iron status in late infancy.23 Beef has been shown to improve growth and cognitive function in Kenyan schoolchildren.24,25 Though this concept appears foreign to many people today, there is little doubt that ASF including meats, perhaps premasticated,26 were fed to older infants and toddlers at the time the human genome developed (i.e., when the lifestyle was that of the hunter-gatherer), and the change to an agrarian lifestyle has been an important contributory factor to iron deficiency in young children today, especially in resource-poor countries.

An earlier study performed in the US city of Denver27,28 demonstrated that the acceptability of beef as a first complementary food by exclusively breastfed infants at age 5–6 months is identical to that for infant rice cereal. By age 7 months, the average intake of pureed beef was 2 oz/day. Informal acceptability studies in the Western Highlands of Guatemala indicate the ready acceptance of cooked, minced liver by infants aged 6–12 months (Krebs et al. unpublished data). Acceptance of meat and liver by infants has also been confirmed by formal testing in Peru,9 despite the mothers' preferences for non-meat-containing porridges.

Though there is broad-based recognition of the importance of ASF, including meats, the evidence base to justify major programmatic efforts to promote meat as a major complementary food remains inadequate. This has essentially provided the rationale for two major trials by our group. It has also dictated a focus on efficacy trials, fully recognizing that effectiveness in different populations will demand further study in different populations. These trials have been accompanied by a more intensive metabolic study in Denver, which is also nearing completion at this time (Krebs et al. unpublished data).


The field studies for our two international trials of meat as a complementary food are nearing completion, but results are not yet available. The primary purpose of this article, therefore, is to provide an overview description of the two trials. One trial utilizes a common protocol that is shared by four sites in the NICHD Global Network for Women's and Children's Health Research, with organizational, data management, and statistical support from RTI International. The four sites are located in Guatemala (small town, Western Highlands), Democratic Republic of the Congo (DRC) (rural, Equateur); Zambia (rural), and Pakistan (peri-urban, Karachi), with leadership provided by the University of Colorado Denver in collaboration with the NICHD.

The other trial, supported by the Thrasher Research Fund, covers virtually the entire county of Xichou (population approximately 250,000), Wen-Shan Canton, Yunnan Province, China, and is a mix of rural and small-town rural communities with leadership provided by a collaboration between the University of Colorado, Jiao Tong University, Shanghai, and the Head of Women's and Children's Health Program for Xichou County.

Both studies are testing the primary hypothesis that daily feeding of meat between the ages of 6 and 18 months will result in significantly greater linear growth velocity compared to that achieved by daily feeding of an equi-caloric micronutrient-fortified, cereal-based supplement (rice-soy for the Global Network, rice for Xichou). The Xichou trial has a third arm of non-fortified rice supplement. Among the secondary hypotheses, and pertinent to this article, is that the meat group will have adequate indices of iron status.

Both trials were designed as cluster-randomized, non-masked, controlled efficacy trials. For unavoidable logistical reasons the Xichou study randomization had to be modified and additional data analyses are planned to exclude or determine the effect of possible confounding factors.

Staff training

The Global Network study utilizes a “train the trainer” model with senior members of each of the participating sites receiving comprehensive training, both at Network Steering Committee meetings and in Colorado, and are provided with the necessary materials, including a Manual of Operations to train the country and community coordinators and others at their sites. The Colorado team and RTI provide follow-up training/advice in subsequent site visits.

University of Colorado researchers traveled to Shanghai for planning and training sessions with the senior investigator at Jiao Tong University (Professor Sheng) prior to on-site training in Xichou: activities included the development of a Manual of Operations, and on-site visits in Xichou County. Prof. Sheng, with limited assistance from Shanghai colleagues, is responsible for training and supervising the leaders of the Xichou Women's and Children's Health Services who then train the directors of the seven county hospitals. The latter, with limited staff, have a vital role in distribution of the interventions and data collection.

Pilot studies

The Global Network undertook a four-country pilot study to attain supervised experience in accurate anthropometry measurements under field conditions. The pilot studies included 1,685 toddlers with equal gender distribution, approximately equal site distribution and a mean age of 17 months. Major dietary staples were maize, potato, and rice in DRC; maize, followed by rice in Zambia; maize, rice, and potato in Guatemala; and rice and potato in Pakistan. Stunting (length-for-age Z scores < 2 SD) rates were approximately 60% in DRC and Guatemala and 40% in Zambia and Pakistan in convenience samples unselected for socio-economic status (SES), diet, or previously known growth. Relevant to our planned trial, a general estimating equation model revealed a significant inverse relationship between stunting and meat consumption, a relationship that was significant even after controlling for all covariates of potential interest (P < 0.01). Eating meat significantly reduced the likelihood of stunting (odds ratio -0.61, confidence interval 0.43–0.85).6

In Xichou, a detailed zinc homeostasis study was undertaken recently.29 Observations pertinent to the study were a stunting rate of 30% for toddlers, with no toddlers being underweight for length and no significant differences in zinc intake between the rural and small-town study groups. Intake of flesh foods was very low in both groups. Results of homeostasis studies indicated that dietary zinc did not meet requirements.

Study subjects and enrollment

Breastfed infants were enrolled at 3–5 months of age in each trial, allowing time for nutrition education, including support of exclusive breastfeeding, before intervention commencement at age 6 months. Subjects at all sites have low incomes. Comparison data on SES between sites is not yet available. Important exclusion criteria included feeding with or the intention to feed infant formula and/or micronutrient-fortified commercial complementary foods; this was a special challenge in Guatemala. All participants were encouraged to continue exclusive breastfeeding until 6 months. Both studies received ethical approval by the Colorado Multiple Institutional Review Board and from the local in-country review boards. Participants gave written informed consent before enrolling in each study.


In each of the four Global Network sites, 10 clusters (communities) were randomized by RTI to participate in either the test or the control group (5 clusters in each group with 30 subjects per cluster) and pair-matched on stunting rates determined from pilot data. The invitation to participate within clusters involved random sampling from cluster-specific lists of births of eligible participants.

In China, randomization initially took place at the Administrative Village level in the domains of six of the eight rural town areas (the remaining two were logistically inaccessible for this study). There are 70 Administrative Villages in Xichou County, each with several village-rural communities. A total of 60 villages that had ≥ 30 births per year were selected randomly for participation. For logistical/budgetary reasons, this planned randomization had to be modified to pre-define three large domains in which the provision of meat with the available resources (including transport) was feasible. All six domains were randomized to participate in one of the two cereal control groups.


Potential participants in both studies were identified at birth using various strategies and were recruited at 3 months of age. In Xichou, a convenience sample of mother-infant dyads was recruited by the community doctors who are an integral part of each of these small communities. Early recruitment in all sites provided the opportunity to provide additional encouragement for mothers to continue exclusive breastfeeding, though this was achieved in only a minority of dyads. Mothers in both studies and in both test and control groups received nutrition education messages.30 These were more formalized and frequent in the Global Network study, focusing on frequency of feeding, thickened gruels, and using the widest possible variety of locally available foods. Hygienic preparation, feeding, and temporary storage of food were also reinforced at every contact.


The test intervention in each study was a daily supply of meat. For the Global Network study, meat is provided as a cooked, diced, lyophilized beef product marketed by Mountain House, Inc. (Albany, Oregon, USA) in 17 oz cans that are stable for >15 years at <100° F and for 10 days after cans are opened. Halal meat is provided for Pakistan. Typically, several cans are distributed to cluster community coordinators weekly. Daily portions are weighed into small zip-lock bags and labeled; seven of these are delivered weekly to participants' homes or to the health facility. It is expected that the infant will consume the initial target quantity of 15 g (equivalent to 30 g cooked meat). The daily portion is increased to 22.5 g/day starting at 12 months of age. For the first few days/weeks, the meat is fed with a minimum of others foods to maximize consumption; remaining meat may be covered and re-fed within 2 hours. Initially, the lyophilized meat can be provided as a puree by crumbling the product into a powder and mixing it with a little boiled water. Later, it can be mixed with other foods and/or fed as a finger food.

In China, pork is a favored meat and has been selected for use in the test intervention. Fresh cuts of pork from Chinese Department of Agriculture-certified animals are purchased weekly in the local market by participating hospital directors and subsequently minced. Two ounces of the minced meat is weighed into plastic bags, stored frozen, and collected the following day by community doctors assigned to participants in the meat group. The community doctors store the packets in their freezers and distribute them daily to the homes of the participants. The quantity of pork offered is twice that of the beef in the Global Network trial because of the lower zinc (and iron) content of pork compared to beef. The amount of the daily serving was not increased at 12 months.

Comparison groups

For the Global Network four-country study, there is one comparison group, a micronutrient-fortified pre-cooked rice-soy blend; the blend is deliberately designed to require additional cooking in boiling water for final hygienic preparation. The supplement was formulated specifically for this study by Nutrica, Inc. (Guatemala City, Guatemala). The cereal supplement is provided in a 20 g package for infants <12 months of age and then 30 g until 18 months of age. These quantities provide an equi-caloric supplement to that provided by the beef, i.e., 70 kcal/day for the infant in addition to 2.2 mg zinc/day and 5.5 mg iron/day. The 30 g packet of cereal provides 3.3 mg zinc/day and 8.3 mg iron/day. Participants in both arms are supplied with a container to store the weekly supplies together with a small cooking and serving pan and a plastic infant spoon. Verbal and pictorial instructions (and initial observation) for maternal-infant hand-washing, dissolving cereal in 120 mL water, and gently mixing/heating the cereal for 2 min are provided.

Two comparison groups are included in the Xichou study. One comparison arm receives a commercial micronutrient-fortified infant rice cereal product. The quantity of cereal is designed to be equi-caloric to the daily supply of pork, i.e., approximately 150 kcal/day with 2 mg zinc and 4 mg iron. The second comparison group is a non-micronutrient-fortified rice biscuit that is slightly sweetened with sugar.


Observation of each meal is not only impossible but undesirable. In the Global Network study, home visits by community coordinators are initially conducted five times per week, decreasing to a weekly visit after 3 months. The number of unused food packets is counted.

For Xichou, home visits are undertaken daily for the meat group to deliver the daily supply. For cereal groups, the visits are weekly. This study represents an additional duty for the community doctors whose commitment to the study quality likely varies.


Initial demographic data, longitudinal anthropometric measurements, dietary data, neurocognitive development measures, and blood collections are obtained by highly trained teams of community workers (Global Network) or graduate students (from Jiao Tong University). The latter are supported and supervised by the head of maternity and child services and her deputy and staff in Xichou County. Morbidity data are collected in the homes by community health workers or community doctors on their regularly scheduled weekly visits.

Sample size estimations

Different assumptions were made for the two trials resulting in minor differences in projected final group size.

Data processing

For both trials, data are entered and checked locally on a daily basis. Global Network data are transferred to RTI International for storage, check edits, and subsequent analyses. Xichou data are entered each evening by the graduate students, checked, and stored on a memory stick that is collected monthly by Prof. Sheng for transfer to Jiao Tong. Further checking and processing are undertaken there with eventual transfer of data to the University of Colorado Denver for analysis.


The two trials described here have many similarities. Nevertheless, the rate of stunting in Xichou toddlers is lower than in the Global Network sites and, apart from very limited meat intake, their diets appear adequate. The study in Xichou County has relatively limited funds, leading to limitations in the number of frontline workers and the ability to perform close monitoring. It is notable that two control groups are included.

The type of meat used in the two intervention studies is different. Each provides different levels of nutrients, but both studies were designed to ensure an adequate intake of bioavailable zinc. Though iron was also a micronutrient of interest, estimates of the quantities of meat to provide were not based on calculations for iron or on iron requirements for infants and toddlers.

Results are not yet available from the two ongoing studies described here, which will be completed in 2011. However, very early and incomplete hemoglobin data from the current Denver study, which is also in progress, are encouraging for the beef arm.

Liver was not provided as a complementary food in either of the described studies, but the ease of preparation and infant acceptability of liver were observed in informal acceptability testing in one of the participating Global Network sites. As a complementary food, liver has attractive features, apart from those above, including lower cost and higher iron content than other forms of meat. For example, the iron content of liver is twice as high as that of beef2; however, much of this iron will be the non-heme form. The high and potentially toxic vitamin A content of liver2,31 is a reason to be cautious about its use, but when provided in judicious quantities along with other organs (especially lung and heart, both of which have relatively high heme iron content) and meats, its advantages are likely to outweigh its disadvantages, especially in eliminating the risk of vitamin A deficiency. It is noted that the use of liver has been promoted30 with no report of adverse effect, and is currently used widely, including in ongoing research projects (one of which is in a Global Network site).


At the time the human genome developed, our ancestors were hunter gatherers. There is now evidence that meat from large mammals was consumed by the hominid species Australopithecus afarensis at least 3.4 million years ago.32 It seems likely that meat was provided by hunter gatherers as a first and frequent complementary food33–35 following maternal premastication.26 This practice remains common in some populations, including in China, though it is now discouraged because of the perceived risk of vertical transfer of HIV. Foman's classic text36 draws attention to the use of meat broth as a complementary food in the United States 200 years ago, but more recent advice commonly provided to mothers, which focuses on plant-based foods during the first months of complementary feeding, does not appear to be based on scientific validation. Without maternal pre-mastication, the feeding of meat at 6–9 months can be facilitated by the use of a simple grinder in addition to cooking.

When considering the provision of meat as a complementary food to infants and toddlers, issues of feasibility related to product availability and affordability must also be considered. The commercial production of meat for mass consumption has been discouraged by many because of the large amounts of agricultural land it requires and because of the carbon emissions it produces; the purchase costs for such meat are also beyond the means of much of the global poor. Though data are not currently available that readily allow estimates of costs in terms of disability-adjusted life years and net present value, estimates of the high costs of meat production have not taken into account the costs of alternative means of meat accrual; this is especially pertinent for the rural poor who cannot afford, or may not have access to, meat products or micronutrient-fortified products that would provide the same nutrients. Where grass for animal grazing is available at the community level, good animal husbandry results in very favorable carbon capture, in contrast with maize-fed animals.37

Finally, a frequent barrier to the provision of meat as an early complementary food, which is not limited to low-income populations, is that current feeding practices typically do not give priority to the older infant and young child when meat is available for familial consumption. Communication and education regarding behavioral modification and its positive effects has the potential to help change current practices, except when they are based on religious practices.

The challenges to providing meat to infants and toddlers are sufficiently great that a strong evidence base for its efficacy is required. Trials such as those described here are helping to build that base.



The work reported on here was funded by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (HD040657 [UCD], HD043464 [UAB], HD040607 [Drexel], HD043475 [UNC], HD040636 [RTI]), the Office of Dietary Supplements, NIH NIDDK K24 DK083772, and the Thrasher Research Fund (02827-4). The National Cattlemen's Beef Association partially supported the analyses of biomarkers for this project.

Declaration of interest.

The authors have no relevant interests to declare.