This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Original Article

Upper levels of nutrients in infant formulas: Comparison of analytical datawith the revised Codex infant formula standard

William C. MacLean Jr.a,*, Peter Van Dael b, Roger Clemens c, Jayne Davies d, Elaine Underwood d,Linda O’Risky b, Deborah Rooney e, Jaap Schrijver f

a School of Medicine, The Ohio State University, Columbus, OH, USAb Mead Johnson Nutrition, Evansville, IN, USAc School of Pharmacy, University of Southern California, Los Angeles, CA, USAd Wyeth Nutrition, Collegeville, PA, USAe Abbott Nutrition, Columbus, OH, USAf Danone Baby Nutrition, Schiphol Airport, The Netherlands

1. Introduction

Codex Alimentarius (‘‘Codex’’) is a global food standardsprogram sponsored jointly by the Food and Agricultural Organiza-tion (FAO) and World Health Organization (WHO). Its goal is toestablish standards for various foods that will assure consumersafety and facilitate trade. Codex periodically revises foodstandards as new scientific data become available. In 1995, thedecision was taken to revise the Codex standard for infant formula,which had last been amended in 1987. After 11 years of discussionin the Codex Committee on Foods for Special Dietary Uses(CCNFSDU), a revised standard (Codex STAN 72-1981; revision2007) was adopted in 2007 (Codex Alimentarius Commission,2007a). A key part of that standard, the essential nutrient

composition of infant formula, had been discussed by theCommittee at annual meetings for several years without reachingconsensus. In 2004, CCNFSDU requested additional advice from aninternational expert group of scientific experts in the area of infantnutrition (referred to as the ‘‘IEG’’). The IEG presented itsrecommendations (Koletzko et al., 2005) at the 27th session inNovember 2005, and those recommendations formed the basis forfurther discussions of compositional criteria.

During those discussions, several issues became apparent:

� For several nutrients the maximum levels recommended by theIEG were close to or lower than levels in current infant formulas.� The scientific basis for setting maximum levels was variable and

inconsistent. This problem resulted in part because for somenutrients the data related to adverse effects from excessiveintakes were relatively well documented, which enabledestablishing science-based upper nutrient levels, while for othernutrients the database was quite meager, which createduncertainties about how best to set upper levels.

Journal of Food Composition and Analysis 23 (2010) 44–53

A R T I C L E I N F O

Article history:

Received 23 December 2008

Received in revised form 28 July 2009

Accepted 29 July 2009

Keywords:

Codex Alimentarius

Infant formula composition

Infant nutrition

Nutrient levels

Vitamins

Minerals

Regulatory requirements

Food analysis

Food composition

A B S T R A C T

The Codex Alimentarius Commission adopted a revised standard for infant formula in 2007. This

standard provides a regulatory framework for infant formula, including provisions for its essential

composition. The recommendations for the essential composition specify minimum levels and either

maximum values (MVs) or guidance upper levels (GULs) for 31 nutrients. As part of the revision process,

the first cooperative survey of levels of nutrients in infant formulas was conducted by several global

manufacturers. Whereas formulas met proposed minimum levels of all nutrients, 15 nutrients were

identified whose levels were likely to exceed the proposed MV or GUL: vitamins A and K, thiamine,

riboflavin, niacin, vitamin B6, folic acid, vitamin B12, vitamin C, iron, copper, manganese, potassium and

iodine. Analytical data were collected for those nutrients from 21,385 batches of milk-based infant

formula and 9070 batches of soy-based infant formula, whose total volumes were sufficient to feed more

than 33 million infants for periods of three months. The number of batches analyzed ranged from 440

(vitamin K) to 27,920 (vitamin C). Of nutrients with an MV, only levels of vitamin A in some batches

exceeded the maximum; no batch contained levels previously reported in the literature to be associated

with adverse effects. There were several nutrients with GULs for which there were batches that exceeded

the suggested upper limit. Data for some nutrients showed considerable variability, which related to

form (liquid vs. powder), inherent levels of nutrients in formula ingredients, protein source, nutrient

stability, analytical variability and effects of process, package and container size.

� 2009 Elsevier Inc. All rights reserved.

* Corresponding author at: 1800 Upper Chelsea Road, Columbus, OH 43212, USA.

Tel.: +1 614 486 6170; fax: +1 614 486 6170.

E-mail address: william.maclean@earthlink.net (W.C. MacLean Jr.).

Contents lists available at ScienceDirect

Journal of Food Composition and Analysis

journa l homepage: www.e lsev ier .com/ locate / j fca

0889-1575/$ – see front matter � 2009 Elsevier Inc. All rights reserved.

doi:10.1016/j.jfca.2009.07.008

Author's personal copy

� It was recognized that information on actual nutrient levels ininfant formulas historically fed without apparent adverse effectscould be used to support the safety of use of the actual nutrientlevels.

With the above in mind, the concept of specifying maximumvalues (MVs) for some nutrients and guidance upper levels (GULs)for others was adopted. For those nutrients for which there wereknown levels above which adverse effects were a possibility, e.g.vitamin D, an MV would be set. For those nutrients for whichelevated intakes had not been associated with adverse effects, e.g.many of the B vitamins, a GUL would be suggested based onknowledge of nutrient requirements of infants, technological andmanufacturing considerations, known variability in current for-mulas and a history of apparent safe use in infant formulas. GULswould not be absolute maximums.

To help in establishing MVs or GULs based on a history ofapparent safe use, the International Special Dietary FoodsIndustries (ISDI), the organization representing the infant formulamanufacturers at Codex, proposed submitting global data on thevariability of nutrient levels in current products. Those data wouldreflect the nutrient levels that formula-fed infants were actuallyconsuming. They would also provide perspective on the techno-logical issues that constrain manufacturing infant formulas withlevels of certain nutrients within narrow limits. This paperpresents data from the survey of the levels of selected nutrientsin infant formula manufactured during the period 2000–2005 andcompares those levels with the essential composition in therevised Codex standard for infant formula.

2. Materials and methods

At the time the survey was conducted, the draft revisedstandard specified minimum values and MVs or GULs for 31nutrients in infant formulas—3 macronutrients (including speci-fications for the essential fatty acids), 13 vitamins, 12 minerals and3 other nutrients. Participating infant formula manufacturerscompared levels of nutrients in their infant formulas marketedglobally with the proposed minimum and upper nutrient levels.This initial assessment determined that the levels of all nutrientsmet the proposed minimum values. However, in the case of 15

nutrients the data suggested that levels in some batches wereexpected to be at or above the proposed upper nutrient levels.These included two fat soluble vitamins (vitamins A and K), eightwater soluble vitamins (thiamine, riboflavin, niacin, vitamin B6,folic acid, vitamin B12, vitamin C and biotin) and five minerals/traceelements (iron, copper, manganese, potassium and iodine).

Formal data collection and analysis focusing on those 15nutrients were completed in a 4-month period between December2005 and March 2006 in order to meet the Codex agenda. Thepresent paper includes data provided by four global infant formulamanufacturers, namely Abbott Laboratories, Mead Johnson Nutri-tion, Numico (currently Danone Baby Foods), and Wyeth Nutrition.

2.1. Nutrient data collection

The data covered infant formulas as defined by CodexAlimentarius, i.e. ‘‘a breast milk substitute specifically manufacturedto satisfy, by itself, the nutritional requirements of infants during thefirst months of life up to the introduction of appropriatecomplementary feeding.’’ Analytical data were collected for liquidand powder forms of milk-based and soy-based infant formulasfrom the quality assurance databases for batches produced andreleased into the market during the period 2000–2005. The nutrientcompositions of formulas reflected by these data are representativeof formulations that have been manufactured and sold for more thanthe 5 years specifically covered by the survey and that in most casescontinue to be sold today. The analytical data covered globallymarketed infant formulas produced in Asia, Europe and theAmericas according to Good Manufacturing Practices.

Analytical values were determined shortly after manufacturing.Table 1 summarizes the analytical methods used and shows thecomparable standard reference methods, where applicable. Allanalytical procedures were formally validated and performedunder stringent quality control according to Good LaboratoryPractices.1 Because these were surveillance data, not all batches

Table 1Summary of analytical methods used.

Nutrient Analytical principle used Comparable standard reference methods where applicable

Alpha-linoleic acids All methods converted fatty acids in oils

to FAMES that were then measured by GC with FID.

AOAC 992.25, AOAC 991.39, 996.06

Vitamin A Saponification/extraction/HPLC measurement of retinol (and isomers). AOAC 992.04, AOAC 992.06, EN 12823-1

Vitamin K HPLC with fluorescence or UV detection. AOAC 999.15, AOAC 992.27, EN 14148

Thiamin All methods relied on conversion of thiamine

to thiochrome and then HPLC using fluorescence detection.

AOAC 986.27, EN 14122

Riboflavin All methods relied on HPLC using fluorescence detection. EN 14152

Niacin Two methods were used, either the standard microbiological

reference method for infant formula or HPLC.

AOAC 985.34

Vitamin B12 Two methods were used, either microbiological or HPLC. AOAC 986.23

Folic acid Three basic methodologies were employed: (1) microbiological,

(2) HPLC, (3) BIA.

AOAC 992.05

Vitamin C Vitamin C analysis was based upon its reduction potential and

measured in one of three ways, potentiometrically, coulometrically,

or colorimetrically.

AOAC 985.33

Biotin Three basic methodologies that were employed: (1) microbiological,

(2) HPLC, (3) BIA.

Proposed EN 15607

Iron Atomic absorption spectrometry or ICP AOAC 984.27, AOAC 985.35

Potassium Atomic absorption spectrometry or ICP. AOAC 985.35, AOAC 984.27

Manganese Atomic absorption spectrometry or ICP. AOAC 984.27, AOAC 985.35

Iodine Ion-selective electrode or inverse colorimetric method. AOAC 992.24

Copper Atomic absorption spectrometry or ICP. AOAC 984.27, AOAC 985.35

Abbreviations: GC: gas chromatography; HPLC: high pressure liquid chromatography; UV: ultraviolet; FAMES: fatty acid methyl esters; FID: flame ionization detection; BIA:

Biomolecular interaction analysis; AA: atomic absorption; ICP-AES: Inductively-coupled plasma atomic emission spectroscopy; AOAC: AOAC International (Official Methods,

2006); CEN: European Committee for Standarization (Comite Europeen de Normalisation); EN: European Norm.

1 The recommended analytical methods for infant formulas in Codex standards

are still under discussion (Codex Alimentarius Commission, 2008). The methods

used to obtain the data reported here generally were equivalent to the currently

proposed Type II and Type III methods as defined by the Codex Committee on

Methods of Analysis and Sampling (Codex Alimentarius Commission, 2007b).

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–53 45

Author's personal copy

were analyzed for each nutrient. Quality assurance proceduressystematically monitor so-called marker nutrients that can be usedto determine the overall composition of all batches of formula, butall nutrients in a batch are directly analyzed only at specificintervals. All nutrient data were expressed per 100 kcal andpresented as means and standard deviations (SD) for each of thefour types of formulas. These data along with the number ofbatches and batch sizes, when available, were submitted by eachcompany on a confidential basis to a data manager contracted forthe study. The means and means + 2SD were calculated for eachcompany’s data and the data expressed as the range of means andthe range of means + 2SD, which were subsequently converted tovalues per 100 kJ. Because different companies use different typesof databases, the data submitted could not be merged into a singledatabase, which would have allowed more extensive statisticalanalysis.

2.2. Other data collected

In addition to the analytical data, food scientists in the companiesprovided information on technological issues that would result inactual nutrient levels exceeding the proposed upper levels for somespecific nutrients. These data included the following areas: form(liquid vs. powder product); inherent levels of vitamins and mineralsin formula ingredients used to supply macronutrients (e.g. riboflavinin ingredients supplying milk protein and lactose); protein source;significant nutrient instability (defined as�25% loss over shelf life);problematical analytical variability based on company experience(defined as� 10% relative standard deviation in the same laboratoryand/or � 20% variability between different laboratories); othertechnical considerations, such as effects of packaging, containermaterial or container size and effects of process, all of which mayaffect stability of some nutrients.

Data on nutrient losses over shelf life were based on analyticalvalues for nutrient levels at the time of release and at the end ofshelf life and were expressed as a percentage of the initial releasevalues. The typical maximum losses were tabulated, rather thanthe average losses, as these values are the ones used by productdevelopment and quality assurance scientists to determinefortification rates during manufacturing. Although processinglosses were not specifically examined (because they occur priorto the time of release) they do account for some of the variabilityof values for some nutrients at the time of release into themarket.

2.3. Defined variables and calculations

In addition to the means and SDs, four additional variables weredefined and, where required, calculated. These were batch, batchsize in metric tons, and infant feeding days and infants fed for three

months from specific batches. Definitions and calculations were asfollows.

2.3.1. Batch

A volume of product manufactured in a single run on whichanalytical data were provided.

2.3.2. Batch size in metric tons

The volumes based on individual batch sizes or total productionvolumes, depending on the company, were converted to metrictons as follows: Data on powder products were presented inpounds or kilograms depending on the manufacturer and thesedata were converted to metric tons (dry weight). Pounds wereconverted by dividing by 2204. Kilograms were divided by 1000.Liquid product was converted using the assumption that 1 Lweighed 1 kg and was expressed as wet weight.

2.3.3. Infant feeding days

The theoretical number of days an infant could be fed formulafrom batches of specific sizes was calculated using the followingassumptions: (1) the average energy intake of the reference infantused for Codex calculations is 2.1 MJ/day (500 kcal/day), (2) a literof formula contains 2.8 MJ (670 kcal) and thus will feed an averageinfant for 1.34 days, (3) one kilogram of powder formula containsapproximately 21.6 MJ (5170 kcal) and thus is sufficient to feed anaverage infant for 10.3 days. The estimate of infant feeding days is atheoretical maximum; it assumes that all formula produced wasconsumed, i.e. that there was no wastage.

2.3.4. Infants fed for three months

The theoretical number of infants who could have been fed aspecific batch of formula was calculated by dividing the number ofinfant feeding days by 90. As with infant feeding days, this numberis a maximum theoretical number. Three months (90 days) werechosen because this duration was considered to be the longestperiod of time that any infant is likely to be fed infant formula froma single batch.

Both ‘‘infant feeding days’’ and ‘‘infants fed for three months’’were thought to be useful in giving perspective on the amount offormula represented by the data in the survey, especially forindividuals not familiar with batch sizes.

3. Results and discussion of specific nutrient-related issues

Table 2 summarizes the number of batches and total metrictons of product in each of the four formula categories: milk-basedliquid and powder and soy-based liquid and powder. The datasetwas derived from 21,385 batches of milk-based and 9070 batchesof soy-based infant formula. This volume of formula was sufficientto feed more than 33,000,000 infants for 3 months. Data for North

Table 2Numbers of batches tested, volume of product and theoretical number of infants fed for three months.a.

Liquid Powder

Batches Metric tons,

wet weight

Infants fed 3 months Batches Metric tons,

dry weight

Infants fed 3 months

Milk-based formula

N. America 2107 235,895 3,518,599 16,235 108,675 12,694,533

Europe/other 32 799 11,911 3,011 118,218 13,809,334

Total 2139 236,693 3,530,510 19,246 226,893 26,503,867

Soy-based formula

N. America 556 62,362 930,194 8,226 17,361 2,085,586

Europe/other – - - 288 4,168 486,826

Total 556 62,362 930,194 8,514 21,529 2,572,412

a See text for calculations of metric tons and ‘‘infants fed for 3 months.’’

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–5346

Author's personal copy

America represented 57% of the total volume of product based oninfant feeding days; data for South America, Africa, Asia, Europe,the Middle East and Oceania covered 43% of the volume of productrepresented by the dataset. Data on vitamin K and iodine werederived only from U.S.-manufactured products, although some ofthese products were sold outside the U.S. There was a wide range inthe percentage of the total volume of product represented by eachcompany’s data. However, the data from the companies were notmerged and/or averaged, which could have given disproportionateweight to the data of one company or another. The means andmeans +2SD from each company for each nutrient were givenequal weight in assembling the results and no company’s data hada disproportionate effect on the data presented (see Table 4).

Table 3 shows the recently adopted Codex minimum valuesand MVs or GULs for nutrients in infant formulas. The shaded rowsindicate nutrients that on initial inspection of manufacturing datawere felt likely to be at odds with the then-proposed upper limits(either MV or GUL). Those nutrients are the subject of this report.

Table 4 compares the analytical data on the nutrients inquestion with the Codex MV, where specified, or GUL. The numbersof batches actually analyzed for each nutrient and the ranges of themeans and of the means + 2SD are shown. The median number ofbatches analyzed for each nutrient was 3080 with a range from 440batches (vitamin K) to 27,920 batches (vitamin C). Although themanufacturer must assure the nutrient levels of all nutrients in aformula prior to release, as mentioned above, not all nutrients aredirectly analyzed during the manufacturing process or at the time

of release. For example, when premixes are used to add watersoluble or fat soluble vitamins, the analysis of a marker nutrient,whose chemical form is unique and stable, in the premix allows theaccurate projection in final product of the concentrations of allother nutrients in the premix. Only data directly analyzed, and notthose calculated from marker nutrients, were included in thesurvey. Table 4 also tabulates the technological issues that affectedvariability.

Table 5 summarizes analytical variability of the methods used.Table 6 shows the maximum loss over shelf life for individualnutrients as a percentage of the value at the time of release. Asdiscussed later and shown in Table 4, these losses are highlyvariable nutrient to nutrient and depend heavily on a variety offactors such as ingredient matrix, effects of process and packageand analytical variability.

3.1. Linoleic acid

As shown in Table 4, the range of means for linoleic acid was193–286 mg/100 kJ (810–1197 mg/100 kcal), the range ofmeans + 2SD, 229–379 mg/100 kJ (958–1585 mg/100 kcal). Themaximum mean + 2SD for linoleic acid of 379 mg/100 kJ(1585 mg/100 kcal) exceeded the finally adopted GUL for linoleicacid of 330 mg/100 kJ (1400 mg/100 kcal).

Note that for this nutrient and a number of others in Table 4 thehighest mean value exceeds the lowest mean + 2SD. This apparentdiscrepancy reflects the fact that the means and means + 2SD came

Table 3Summary of Codex minimum values, maximum values and guidance upper levels for macronutrients, vitamins and minerals and other substancesa,b (nutrients in shaded cells

are the subject of this report).

Nutrient Unit Codex

Minimum value per 100 kJ

(per 100 kcal)

Maximum value (MV) per

100 kJ (per 100 kcal)

Guidance upper level (GUL)

per 100 kJ (per 100 kcal)

Protein g 0.45 (1.8) 0.7 (3.0) –

Lipid g 1.05 (4.4) 1.4 (6.0) –

Linoleic acid mg 70 (300) – 330 (1400)

a-Linolenic acid mg 12 (50) NS –

Carbohydrate 2.2 (9) 3.3 (14) –

Vitamin A-RE mg 14 (60) 43 (180) –

Vitamin D mg 0.25 (1) 0.6 (2.5) –

Vitamin E-a TE mg 0.12 (0.5) – 1.2 (5)

Vitamin K mg 1 (4) – 6.5 (27)

Thiamin mg 14 (60) – 72 (300)

Riboflavin mg 19 (80) – 119 (500)

Niacin mg 70 (300) – 360 (1500)

Vitamin B6 mg 8.5 (35) – 45 (175)

Vitamin B12 mg 0.025 (0.1) – 0.36 (1.5)

Pantothenic acid mg 96 (400) – 478 (2000)

Folic acid mg 2.5 (10) – 12 (50)

Vitamin C mg 2.5 (10) – 17 (70)

Biotin mg 0.4 (1.5) – 2.4 (10)

Iron mg 0.1 (0.45) NS NS

Calcium mg 12 (50) – 35 (140)

Phosphorus mg 6 (25) – 24 (100)

Magnesium mg 1.2 (5) – 3.6 (15)

Sodium mg 5 (200) 14 (60) –

Chloride mg 12 (50) 38 (160) –

Potassium mg 14 (60) 43 (180) –

Manganese mg 0.25 (1) – 24 (100)

Iodine mg 2.5 (10) – 14 (60)

Selenium mg 0.24 (1) – 2.2 (9)

Copper mg 8.5 (35) – 29 (120)

Zinc mg 0.12 (0.5) – 0.36 (1.5)

Choline mg 1.7 (7) – 12 (50)

Myo-inositol mg 1 (4) – 9.5 (40)

L-Carnitine mg 0.3 (1.2) NS NS

a Codex STAN 72-1981; revision 2007 (Codex Alimentarius Commission, 2007a). See text for distinction between maximum and upper guidance levels. NS = not specified.b In addition to the nutrients above there are three optional nutrients, which have no minimum level: taurine (maximum = 3 mg/100 kJ (12 mg/100 kcal), nucleotides

(upper levels at discretion of national authorities) and docosahexaenoic acid (DHA; GUL of 0.5% of fatty acids, with proviso that national authorities may chose to deviate from

the GUL).

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–53 47

Author's personal copy

from individual companies. The manufacturers with the lowestmean value for any nutrient generally had the lowest value for themean + 2SD. As was the case here, one of the other manufacturerscould have reported a mean value that was higher than themean + 2SD of the first company.

The analytical variability for linoleic acid determination wassmall (<10%) and was not a major factor in the reported variability.Because of the difficulty of stabilizing polyunsaturated fatty acidsin powders, both linoleic acid and a-linolenic acid levels weregenerally lower in powder products (not shown separately inTable 4), which contributed to variability.

Linoleic acid is added to infant formula as part of a blend of twoto four different oils. The considerable variability of both linoleicand a-linolenic acids in different oils presents a challenge to the

manufacturer. In soy oil, for example, linoleic acid accounts onaverage for�53% of fatty acids, but the level can vary between 49%and 57%. Corn oil contains similar levels, whereas high oleicsafflower oil contains only 15–18% linoleic acid. Ingredients otherthan the oil blend, soy lecithin, for example, can add up to 7.2 mg oflinoleic acid per 100 kJ (30 mg/100 kcal) of finished product. Therewas no loss of linoleic acid over shelf life.

There is a long history of feeding formulas with levels of linoleicacid above the maximum values reported here without adverseeffects. Prior to the late 1980s, when the importance of assuringadequate intakes a-linolenic acid and maintaining an appropriateratio of linoleic to a-linolenic acids was recognized, formulas withsubstantially higher levels of linoleic acid were fed. Oil blendsbased on mixtures of corn oil and soy oil were often used and

Table 4Comparison of maximum values and upper guidance levels for selected nutrients in the new Codex regulations for starter infant formulas and survey data.

Nutrient Unit Codex maximums:

GUL or MV1

Current survey data Key factors affecting

variability4

Number of

batches tested

Range of means Range of means + 2SD

Linoleic acid (mg) /100 kJ 330 715 193–286 229–379 A, C

/100 kcal 1400 810–1197 958–1585

Vitamin A (mg RE) /100 kJ 43 3,680 27–36 29–59 A, D, F1

/100 kcal 180 112–151 121–245

Vitamin K3 (mg) /100 kJ 6.5 440 3.8–6.0 5.0–7.4 C

/100 kcal 27 16–25 21–31

Thiamin (mg) /100 kJ 72 3,080 31–67 34–75 A, D

/100 kcal 300 129–280 143–315

Riboflavin (mg) /100 kJ 119 3,317 29–94 33–124 A, B, C, D, F1

/100 kcal 500 121–392 138–519

Niacin (mg) /100 kJ 360 965 214–457 237–606 A, D

/100 kcal 1500 898–1915 992–2538

Vitamin B12 (mg) /100 kJ 0.36 1,200 0.11–0.24 0.12–0.34 B, D, E

/100 kcal 1.5 0.45–1.0 0.51–1.44

Folic acid (mg) /100 kJ 12 1,002 3.8–9.6 4.3–13.4 C, E

/100 kcal 50 16–40 18–56

Vitamin C (mg) /100 kJ 17 27,920 3.6–9.3 4.3–17.2 A, D, E, F1, F2

/100 kcal 70 15–39 18–72

Biotin (mg) /100 kJ 2.4 962 1.0–2.4 1.2–2.7 D, E

/100 kcal 10 4.0–10.1 4.9–11.4

Iron (mg) /100 kJ ** 24,394 0.3–0.5 0.3–0.6 G

/100 kcal ** 1.3–2.2 1.4–2.4

Potassium (mg) /100 kJ 43 5,132 24–35 27–39 B, C, F2

/100 kcal 180 99–146 111–164

Manganese (mg) /100 kJ 24 4,613 2.4–6.7 (9.3–14.1)5 3.3–14.6 (13.6–23.6)5 B, C

/100 kcal 100 10–28 (39–59)5 14–61 (57–99)5

Iodine3 (mg) /100 kJ 14 2,499 5.0–6.9 7.2–10.0 B, C

/100 kcal 60 21–29 30–42

Copper (mg) /100 kJ 29 23,878 21–39 22–45 B

/100 kcal 120 87–165 94–189

1See text for definitions of MV (maximum value) and GUL (guidance upper limit). MVs are shown in bold font.2Codex STAN 72-1981; revision 2007 (Codex Alimentarius Commission, 2007a).3Only U.S. companies supplied data for vitamin K and iodine.4Codes for this column on key factors affecting need for higher limits are as follows:

(A) Form (liquid vs. powder).

(B) Protein source (inherent levels of the nutrient are significant).

(C) Inherent levels and variability of the nutrient in question in other ingredients are a problem.

(D) Nutrient stability (�25% loss over shelf life). This may interact with form and package.

(E) Analytical variability (� 10% relative standard deviation in the same laboratory and/or � 20% variability between different laboratories).

(F) Other technical considerations:

(F1) Effects of packaging, container material or container size.

(F2) Effects of process.

(G) Views of nutrition experts in different countries have produced different views on appropriate levels for formulas and hence different regulations; differences in use of

starter and follow-up formulas.5Values for milk-based formulas are shown without parentheses; values for soy-based formulas within parentheses.

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–5348

Author's personal copy

provided linoleic acid in amounts up to an estimated 654 mg/100 kJ (2740 mg/100 kcal) of infant formula (Fomon et al., 1971).

3.2. Vitamin A

The highest mean + 2SD for vitamin A of 59 mg RE/100 kJ(245 mg RE/100 kcal) was higher than the newly set Codexmaximum of 43 mg RE/100 kJ (180 mg RE/100 kcal). The maximumanalytical variability for vitamin A was 16%. Vitamin A in infantformula is derived principally from its specific addition duringmanufacture. Only a small percentage (<10%) of preformedvitamin A comes from inherent levels in formula ingredients.Losses over shelf life were up to 30% and depended on the form ofthe product, the packaging and the assigned shelf life. A wide rangeof mean + 2SD in liquid products (not shown separately in theTable 4) reflected the susceptibility to variable degradation duringprocessing.

There were large numbers of infants (>6000) estimated to havebeen fed at intakes of vitamin A above the new MV. This substantialhistory of apparent safe use with levels of vitamin A at or above theMV may be interpreted in light of other literature on the safety ofvitamin A. The lowest chronic intake associated with vitamin Atoxicity cited by the Scientific Committee on Food (SCF, 2003) intheir review of the literature equated to levels in infant formula of344 mg RE/100 kJ (1440 mg RE/100 kcal) (assuming an intake of

2.1 MJ [500 kcal]/day from formula). The lowest chronic intakesassociated with toxicity reported by Olsen (1989) in his reviewwere approximately 145–217 mg RE/100 kJ (606–909 mg/100kcal). Based on the survey data the intakes by infants fed currentlymarketed products even at the high end of the range were wellbelow those levels.

3.3. Vitamin K

The survey data, based on 440 batches of U.S.-manufacturedproducts, showed that a small percentage of batches would beexpected to be above the new Codex GUL of 6.5 mg RE/100 kJ(27 mg/100 kcal). The highest mean + 2SD was reported to be7.4 mg RE/100 kJ (31 mg/100 kcal). Analytical variability was low(<10%) and there was no loss over shelf life. Vitamin K is added toinfant formula in the form of phylloquinone (phytomenadione)–vitamin K1. In addition there are substantial, but variablecontributions of vitamin K from soy oil, which can contributefrom 3% to 25% of total contents of vitamin K in finished productand affected the reported variability. Vitamin K is known to be safeat these intakes (Olsen, 1989). The SCF observed in its report: ‘‘Noadverse effects of enteral vitamin K supply in term infants havebeen observed.’’ (SCF, 2003).

3.4. Thiamine

The highest value for the mean + 2SD was 75 mg/100 kJ(315 mg/100 kcal), which was above the new Codex GUL of72 mg/100 kJ (300 mg/100 kcal). The analytical variability forthiamine was low (<10%). Variability related to product formand nutrient stability. The majority of thiamine is added duringmanufacturing as part of a water soluble vitamin premix. A smallcontribution (<10%) to total formula thiamine levels comes frominfant formula ingredients. Thiamine levels can be difficult tocontrol because thiamine is heat labile during processing (reflectedin variability at the time of release) and there are losses of up to40% during shelf life. Thiamine is considered non-toxic at thereported levels (McCormick, 1989; SCF, 2003).

3.5. Riboflavin

Riboflavin levels were highly variable; values for the mean + 2SDranged from 33 mg/100 kJ to 124 mg/100 kJ (138 mg/100 kcal to519 mg/100 kcal), the highest being above the Codex GUL of 119 mg/100 kJ (500 mg/100 kcal). The analytical variability for riboflavinwas low (<5%).

Riboflavin is one of the more variable nutrients in infantformulas because of the differences in inherent levels in variousingredients and the effects of form and package on stability (seekey factors in Table 4). The variability in riboflavin levels in milk-based infant formulas is considerable due to the fact that lactoseand milk protein ingredients are important and highly variablesources of riboflavin. The riboflavin content of lactose can be as lowas 200 mg/100 g or as high as 2600 mg/100 g. Thus, in a standardmilk-based formula, lactose alone may contribute as little as 5 mg/100 kJ (22 mg/100 kcal) or as much as 67 mg/100 kJ (279 mg/100 kcal) in the finished product. Riboflavin in milk proteiningredients contributes �8–10 mg/100 kJ (�35–40 mg/100 kcal)and can vary by 25%. When the highest values for both ingredientsare used, the amount contributed by them alone is >55% of theGUL. Soy protein is not a significant source of riboflavin in soyformulas, nor are the carbohydrates, other than lactose, used ininfant formulas.

Both milk-based and soy-based formulas showed losses overshelf life that was quite significant—up to 60%. Thus, the inherentlevels in ingredients cannot be relied upon alone to provide

Table 5Analytical variability of the methods based on manufacturers’ experience.

Nutrient Intralaboratory

variability, RSD (%)

Interlaboratory

variability, RSD (%)

Linoleic acid 4.2 n/a

Vitamin A 7.0 16.0

Thiamine 8.2 4

Riboflavin 4.7 2

Niacin 4.5 12.3

Vitamin B12 15 20

Folic acid 10 18

Biotin 11 16

Vitamin C 10 6.8

Iron 2.5 2.7

Copper 13.2 4.8

Manganese 3.2 4.5

Potassium 2.7 2

Vitamin K n/a 7.3

Iodine n/a 3.0

n/a = not available.

Table 6Maximum degradation of selected nutrients in infant formulas over shelf life.a.

Nutrient Maximum % loss at end of shelf life

Linoleic acid 0

Vitamin A 30

Thiamine 40

Riboflavin 60

Niacin 27

Vitamin B12 55

Folic acid 15

Biotin 30

Vitamin C 75

Iron 0

Copper 0

Manganese 0

Potassium 0

Vitamin K 0

Iodine 0

a Data are based on analytical data provided by manufacturers. Values shown are

calculated as value at the end of shelf life as a percentage of the initial value,

rounded to the nearest 5%.

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–53 49

Author's personal copy

adequate riboflavin in finished product and riboflavin must beadded as part of the water soluble vitamin premix. In view of theabove it is apparent that managing riboflavin levels, especially inmilk-based infant formulas, is a challenging task. There is noclearly defined adverse effect related to riboflavin levels at thelevels reported in currently marketed infant formulas (SCF, 2003).

3.6. Niacin

The survey showed wide variability in niacin content with themaximum mean + 2SD of 606 mg/100 kJ (2538 mg/100 kcal).There were a large number of batches above the new CodexGUL of 360 mg/100 kJ (1500 mg/100 kcal). Analytical variabilitywas low (<10%) and levels were not related to product type orform. There is little preformed niacin in the ingredients used tomake either milk-based or soy-based infant formulas. (Onlypreformed niacin and not niacin equivalents [NE] was consideredin the analysis.) Nearly all of the niacin in infant formula is addedas nicotinamide as part of a premix of water soluble vitamins.There is little degradation during processing, but loss over shelflife may reach up to 27%. This loss depends on product form – lessin powder than in liquid – and the type of package. There is aconsiderable margin of safety of niacin in the form of nicotina-mide, which has been fed in much larger amounts in themanagement of inherited metabolic disorders (McCormick,1989), and the current history of apparent safe use in consistentwith that conclusion.

3.7. Vitamin B12

The highest mean + 2SD for vitamin B12 levels in currentlymarketed infant formulas was reported to be 0.34 mg/100 kJ(1.44 mg/100 kcal), which was below the new Codex GUL of0.36 mg/100 kJ (1.5 mg/100 kcal). The analytical method forvitamin B12, which most often has been microbiological, hasconsiderable variability with a relative standard deviation of up to15% in most laboratories and up to 20% between laboratories. Morerecently, a high pressure liquid chromatography (HPLC) methodhas been used. Whichever method is used, when measuring anynutrient in mg/L amounts, the analyst is asked to detect levels inthe parts per billion (ppb) range. Industry experience shows thatthe AOAC microbiological assay is increasingly more variable andless precise at levels below 5 ppb, which equates to a level of0.18 mg/100 kJ (0.75 mg/100 kcal). HPLC methodology is morerobust, but suffers the same decrease in precision at about 3.5 ppbto 0.12 mg/100 kJ (0.52 mg/100 kcal). Vitamin B12 levels of milkprotein sources are low (<0.03 mg/100 kJ [<0.14 mg/100 kcal] offinished product). Thus, most vitamin B12 in infant formulas isadded as part of a vitamin premix. Shelf life losses may reach up to55% and are highly variable depending on product, form, andassigned shelf life.

3.8. Folic acid

The highest mean + 2SD for folic acid of currently marketedinfant formulas was 13.4 mg/100 kJ (56 mg/100 kcal), slightlyabove the Codex GUL of 12 mg/100 kJ (50 mg/100 kcal). Theanalytical method, which most often is microbiological, has arelative standard deviation of about 10% in the same laboratorywith variability between laboratories of up to 25%. Up to 40% of thefolate in finished product is inherent in the ingredients used toproduce infant formula. Because folate is one of the most labilevitamins, formulas are fortified with folic acid, which exhibitssubstantially greater stability than folate. Losses over shelf lifewere reported to be up to 15%. Folate is acknowledged to be safe inthe ranges found in products currently in the marketplace

(McCormick, 1989). The principal concern with very high folateintakes is that they will mask vitamin B12 deficiency, which is not aconcern in formula-fed infants.

3.9. Vitamin C

The survey data showed a wide range of means (3.6–9.3 mg/100 kJ, 15–39 mg/100 kcal) and of means + 2SD (4.3–17.2 mg/100 kJ, 18–72 mg/100 kcal). The highest means + 2SD indicatedthat some batches exceeded the GUL of 17 mg/100 kJ (70 mg/100 kcal). The analytical variability for vitamin C was low (<10%).

Vitamin C is one of the most challenging nutrients for the infantformula manufacturer due to its considerable and variable lability,which was reflected in the high variability reported (see ‘‘Keyfactors’’ in Table 4). Virtually all the vitamin C in infant formula isadded; there is little or no contribution from other ingredients.Vitamin C degrades rapidly in water and when exposed to air. Lossover shelf life is variable, depending on product form and packageand is considerably greater in liquids than in powders. Powderproducts are generally packed under nitrogen and the availableoxygen that remains in the powder after packaging quickly dropsto close to nothing during the first week as the antioxidants in theproduct react with it. Liquid products generally do not have thisstability after the first week and, depending on package and shelflife, typical losses of 30–50% are not out of the ordinary; losses maygo as high as 75%. Loss of vitamin C also occurs after the producthas been opened. One manufacturer reported there was a loss of35% in 72 h after ready-to-use product had been opened. Bycontrast, ‘‘open can’’ studies of powder (cans left open at roomtemperature for a period of up to 4 weeks) by the samemanufacturer showed losses of 8–10%.

3.10. Biotin

The data showed that the highest mean + 2SD for biotin was2.7 mg/100 kJ (11.4 mg/100 kcal) whereas the new Codex GUL wasset at 2.4 mg/100 kJ (10 mg/100 kcal). The intralaboratory analy-tical variability of biotin was up to 11%, the interlaboratoryvariability up to 16%. Only a small percentage (<20%) of the biotinin finished product is inherent in the formula ingredients. Most ofthe biotin in infant formulas is added as part of a vitamin premix.There is some loss during processing. Losses over shelf life mayapproach 30%, depending on the form of the product. There is along history of apparent safe use at currently manufactured levels;the literature supports the safety of such levels as well (Fomon andMcCormick, 1993; SCF, 2003).

3.11. Iron

The ranges of means + 2SD of iron in the formulas analyzed was0.3–0.6 mg/100 kJ (1.4–2.4 mg/100 kcal). The analytical methodhas low variability (<10%). The iron level in infant formula isdetermined primarily by the amount of iron specifically added bythe manufacturer. Without the addition, most infant formulaswould contain no more than �0.1 mg/100 kJ (�0.4 mg/100 kcal).

The iron content of infant formulas globally covers a widerange, primarily for two reasons. First, there are long establisheddifferences between countries and regions of the world in the levelof iron recommended or required in infant formulas. Second, inmost countries, the use of infant formula is limited to the first fourto six months of life and the introduction thereafter of follow-onformulas, with their higher allowable iron levels, is wellestablished. In some countries, such as the United States, despitethe availability of formulas for older infants, infant formulas arethe ones principally fed for the entire first year of life. In the lattercountries, the maximum levels of iron in iron-fortified infant

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–5350

Author's personal copy

formulas have generally been higher. Because of such differences,Codex chose not to set either an MV or GUL for iron. Instead, upperlevels were left to the discretion of individual regulatoryauthorities.

3.12. Potassium

The survey found that the highest mean + 2SD was 39 mg/100 kJ (164 mg/100 kcal), whereas Codex has set an MV of 43 mg/100 kJ (180 mg/100 kcal). The survey data suggested that althoughthere was considerable variability in potassium levels in formulasthe new maximum would encompass currently manufacturedproducts. Analytical variability for potassium levels was low (<3%).

There are many sources of potassium in infant formulas andthere are several major technical constraints to controllingpotassium levels. The levels in ingredients are variable. Thepotassium level in milk protein ingredients varies by up to 75%;milk protein ingredients alone, non-fat dry milk, for example, cancontribute>24 mg/100 kJ (>100 mg/100 kcal) of finished product.The potassium content of soy protein isolates varies by up to 100%due to the fact that potassium hydroxide is used as a neutralizingagent in the purification process of soy protein isolates. Ingredientssupplying maltodextrins and glucose polymers also contributepotassium. Potassium hydroxide and citrate are used as processingaids to control pH in the manufacturing of the formula itself and aspart of the antioxidant system that protects polyunsaturated lipidsfrom oxidation.

3.13. Manganese

The survey data demonstrated a clear difference between milk-based and soy-based infant formulas: the highest means + 2SDbeing 14.1 mg/100 kJ (61 mg/100 kcal), and 23.6 mg/100 kJ (99 mg/100 kcal), respectively. Analytical variability was low <5%.

Codex set a GUL of 24 mg/100 kJ (100 mg/100 kcal), a level thatincluded nearly all batches of formula analyzed and that is inharmony with the level recommended by the Life Science ResearchOffice (LSRO, 1998), the SCF (2003), and now incorporated into therecast European Directive for Infant Formula (Commission of theEuropean Communities, 2006). The manganese in milk-based andsoy-based infant formulas derives from different formula ingre-dients. In milk-based products, manganese may be specificallyadded during manufacture, whereas in soy-based products,because of the significant and variable quantities of manganesein soy protein isolate, it generally is not. The content of manganesein soy protein isolate can vary from 500 mg/100 g to 3000 mg/100 g. In both types of formulas there are also small amounts ofmanganese contributed by some of the calcium salts and by ferroussulfate.

3.14. Iodine

The analytical data showed a range of means + 2SD of 7.2–10.1 mg/100 kJ (30–42 mg/100 kcal), below the Codex GUL of14 mg/100 kJ (60 mg/100 kcal). There were occasional batches,however, that reached as high as 18 mg/100 kJ (74 mg/100 kcal).Inspection of the data showed that they were not normallydistributed. There was a significant tail to the distribution curvethat was skewed to the right (towards higher values) in milk-basedproducts. Analytical variability was very low.

There is considerable variability of iodine content in milk-derived ingredients depending on the milk source from which theingredient comes, dairying practices (fortification of feed withiodine, iodine containing antiseptics) and seasonality. For example,iodine in non-fat dry milk, one potential source of milk protein ininfant formulas, can range from about 150 mg/100 g to 800 mg/

100 g of ingredient. In addition to the variable iodine content ofmilk ingredients, there is an effect of the protein content of milkingredients, which also is variable, on the final content of iodine inthe formula. Based on the current dataset, the Codex GUL willencompass almost all of the batches, though unexpected spikes ofiodine in cows’ milk ingredients will results in occasional batchesabove this level. In his review of upper limits of iodine in infantformulas, Fisher (1989) concluded that a maximum concentrationof approximately 18 mg/100 kcal (75 mg/100 kcal) and an intake of100 mg/kg body weight per day were reasonable.

3.15. Copper

The range of means (21–39 mg/100 kcal, 87–165 mg/100 kcal)and mean + 2SD (22–45 mg/100 kcal, 94–189 mg/100 kcal)showed that many batches analyzed exceeded the new CodexGUL 29 mg/100 kcal (120 mg/100 kcal). Analytical variabilityreached 13%.

The majority of the copper in infant formula is added throughthe trace element premix during the manufacturing process.Additionally, the protein sources (e.g. non-fat milk protein, wheyprotein concentrates, soy protein isolates) variably contribute tothe copper level of infant formulas. Non-fat dry milk, for example,may contribute up to about 10 mg/100 kcal (43 mg/100 kcal), soyprotein isolate up to �18 mg/100 kcal (�76 mg/100 kcal). As aconsequence of the higher inherent copper level of soy proteinisolates, the soy-based infant formulas showed on average �25–30% higher copper levels than milk-based infant formulas. Therewas no change in copper content in finished product over shelf life.In their review of upper limits of copper, Hambidge and Krebs(1989) concluded that an upper limit of 48 mg/100 kcal (200 mg/100 kcal) in formulas for full term infants would be safe andreasonable.

4. General discussion

There has been no previous attempt by global manufacturers tocollect and review actual nutrient levels in infant formula on aworld-wide basis. The current survey covering the years from 2000to 2005 provides the most comprehensive review to date of infantformula nutrient levels presently in the global marketplace. Thestrengths of the dataset include the number of batches available,the numbers of different manufacturing facilities involved, thebroad geographical origin and ultimate distribution of the productsand the fact that analytical methods validated for the differentproduct matrices were used to determine nutrient levels. Thevolume of product analyzed was sufficient to feed more than 33million infants for periods of 3 months. Thus, the data provideuseful information about the nutrient composition of infantformulas and by extension nutrient intakes of many exclusivelyformula-fed infants around the world. As such, the data provide abasis to help establish a history of apparent safe use of thenutrients studied.

It is also important to note the limitations of the datapresented. First the survey did not include data from all globalmanufacturers nor from regional and/or national infant formulaproducers. As a consequence the survey cannot be consideredrepresentative of nutrient levels in all infant formulas. Second,within the timeframe available to meet the deadlines forproviding data for the CCNFSDU meeting of 2006, the numbersof batches and types of products that could be surveyed were notall inclusive and not every nutrient was directly analyzed in everybatch of formula. Incompatibility of the databases of theindividual companies and time constraints precluded reenteringthe individual data into a single database, which would haveallowed more sophisticated analysis. A decision was made to

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–53 51

Author's personal copy

focus only on milk-based and soy-based infant formulas thatwould be used for exclusive formula feeding for the first four to sixmonths of life. Differences between liquids and powders could beexamined to some extent, but the data could not be broken downextensively by process and package. However, the inability todissect some of these explanatory factors more fully does notlessen the overall value of the data.

The data for all nutrients showed wide ranges of means + 2SD(Table 4). In the cases of nutrients for which MVs were specified byCodex, the data showed that only in the case of vitamin A werethere levels in some batches of current formulas that exceeded theMV, and current vitamin A levels will need to be reduced to meetthe revised Codex standard. There was no indication that any of thereported vitamin A levels would produce adverse effects, as thehighest reported levels were still well below those reported in theliterature as being associated with adverse outcomes (Olsen,1989). In the case of the 9 nutrients2 for which GULs were specified,the data indicated that there were some batches of formula inwhich one or more of the nine exceeded the recommendations. Itshould be remembered, however, that GULs, rather than MVs, wereassigned to these nutrients specifically because there is an absenceof data indicating problems from consumption of these higherlevels (Fomon et al., 1971; Hambidge and Krebs, 1989; McCormick,1989; Olsen, 1989; SCF, 2003).

The infant formula manufacturers who participated in thissurvey use analytical surveillance data in two principal ways. Thefirst is to monitor production of individual batches and thus toassure that no batch that does not meet all regulatory require-ments is introduced into commerce. The second is to monitorspecific nutrients against narrowly specified ranges. This allowsthe manufacturer to track changes in the inherent nutrient levels ofdifferent ingredients (e.g. the linoleic acid content of soy oil) andpossible trends in manufacturing processes over time.

Managing the levels of more than 30 individual nutrients withintight specifications in each batch of formula is a major challenge forinfant formula manufacturers. This challenge is reflected in thewide ranges of means and means + 2SD reported, which reflected anumber of factors: (1) differences in manufacturing processesbetween companies and between plants within the samecompany; (2) the form of the product: liquids and powders ofthe same formula may require different ingredients to accom-modate the processes used, and the reactivity of certain nutrientsin the same product will vary considerably depending, for example,on whether a ready-to-use liquid, a concentrated liquid or apowder is being manufactured; (3) variability of inherentingredient levels of some nutrients in raw ingredients (e.g.riboflavin in lactose) and the differences in those ingredientsprovided by different suppliers in different regions of the world;(4) variable processing losses, which may be minimal for somenutrients or large and highly variable for others; (5) the variableeffects of the type of container, the assigned shelf life andvariability of the analytical methods used.

For all the above reasons there will invariably be batches withone nutrient at the higher or lower end of the range, even thoughthe others are nearer the mean, and the data do not implyprolonged consumption of the highest or lowest levels byindividual infants. Since any one batch of formula is widelydistributed even within the same country, infants are not fedformula from a single batch for prolonged periods of time. One-kilogram can of powder is sufficient for feeding an average infantfor about 10 days. Even parents who purchase formula by the casewill feed their infants formulas from a variety of different batchesduring the first six months of life.

The revised Codex standard incorporated for the first timeguidance on the interpretation of upper levels of all essentialnutrients in infant formulas. When establishing acceptableminimum and maximum nutrient levels for infant formulas itis important to set levels that assure that the nutritional needs ofall healthy infants will be met while simultaneously avoidingupper levels leading to intakes that could produce adverse effects.While it was deemed important to provide guidance in the newstandard on upper limits for all nutrients, it became clear in thecourse of the discussions that the science base on which to makejudgments about upper levels was highly variable. Far moreinformation is available on nutrients known to cause adverseeffects when consumed in excess than for those for which adverseeffects have not been reported or at least not in the rangescurrently consumed. The differences in the extent of the availablescience base required differentiation between MVs and GULs. Insetting guidance upper limits, where ‘‘hard data’’ were insuffi-cient, a history of apparent safe use was relied upon. Thedifference between MVs, which are ‘‘absolute’’, and GUL, whichare not, was clearly stated in the revised standard: ‘‘The purpose ofthe GULs [guidance upper levels] is to provide guidance tomanufacturers and they should not be interpreted as goal values.Nutrient contents in infant formulas should usually not exceedthe GULs unless higher nutrient levels cannot be avoided due tohigh or variable contents in constituents of infant formulas or dueto technological reasons. When a product type or form hasordinarily contained lower levels than the GULs, manufacturersshould not increase levels of nutrients to approach the GULs’’(Codex Alimentarius Commission, 2007a).

In conclusion, the present survey, carried out to provideguidance to CCNFSDU in its deliberations concerning MVs andGULs for nutrients in infant formula, was the first effort toreview the nutrient levels in infant formulas produced byseveral of the global manufacturers. The survey provided insightinto actual nutrient levels and their variability, demonstratedthat most nutrients levels in the surveyed products currentlymeet the newly established Codex upper levels, and providedinformation to help establish a history of apparent safe use. Thisinformation will hopefully prove useful to both regulatoryauthorities and industry as they work together in the future toimplement the new concept of MVs and GULs in the revisedCodex standard.

Acknowledgements

The authors acknowledge the financial support from theInternational Formula Council. The authors also acknowledgethe expert assistance and advice from member companyrepresentatives and staff of the International Special DietaryFoods Industry and the International Formula Council including,Andree Bronner, Margaret Creedon, Amandine Devergies, MardiMountford, and Dan Schmitz.

References

Codex Alimentarius Commission, 2007a. Standards for Infant Formulas and For-mulas for Special Medical Purposes Intended for Infants (Codex Stan 72-1981)(Revised 2007). Retrieved October 31, 2008 from www.codexalimentarius.net/download/standards/288/CXS_072e.pdf.

Codex Alimentarius Commission, 2008. Report of the Electronic Working Group onMethods of Analysis For Infant Formulas and Formulas for Special MedicalPurposes Intended for Infants (Codex Stan 72-1981). Retrieved October 31,2008 from http://www.ccnfsdu.de/fileadmin/user_upload/PDF/nf3002ae.pdf.

Codex Alimentarius Commission, 2007b. CAC Procedural Manual. 17th Edition Foodand Agricultural Organization/World Health Organization, Rome 2007.Retrieved 10-31-08 ftp://ftp.fao.org/docrep/fao/010/a1472e/a1472e03.pdf.

Commission of the European Communities, 2006. Commission Directive 2006/141/EC of 22 December 2006 on infant formulae and follow-on formulae andamending Directive 1999/21/EC. Official Journal of the European Union.

2 Linoleic acid, vitamin K, thiamin, riboflavin, niacin, folic acid, vitamin C, biotin,

and copper.

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–5352

Author's personal copy

L401/1-L401/33. Retrieved November 4, 2008 from http://eur-lex.europa.eu/LexUriServ/site/en/oj/2006/l_401/l_40120061230en00010033.pdf.

Fisher, D.A., 1989. Upper limits of iodine in infant formulas. Journal of Nutrition 119(Suppl. 12S), 1865–1868.

Fomon, S.J., Thomas, L.N., Filer, L.J., Ziegler, E.E., Leonard, M.T., 1971. Food con-sumption and growth of normal infants fed milk-based formulas. Acta Paedia-trica Scandinavica Supplement 223, 1–24.

Fomon, S.J., McCormick, D.B., 1993. B vitamins and choline. In: Fomon, S.J. (Ed.),Nutrition of Normal Infants. Mosby-Year Book, Inc., St. Louis, MO, USA, pp. 366–394.

Hambidge, K.M., Krebs, N., 1989. Upper limits of zinc, copper and manganese ininfant formulas. Journal of Nutrition 119 (Suppl. 12S), 1861–1864.

Koletzko, B., Baker, S., Cleghorn, G., Neto, U.F., Gopalan, S., Hernell, O., Hock, Q.S.,Jirapinyo, P., Lonnerdal, B., Pencharz, P., Pzyrembel, H., Ramirez-Mayans, J.,Shamir, R., Turck, D., Yamashiro, Y., Zong-Yi, D., 2005. Global standard for thecomposition of infant formula: recommendations of an ESPGHAN coordinated

international expert group. Journal of Pediatric Gastroenterology and Nutrition41, 584–599.

LSRO (Life Sciences Research Office), 1998. In: Raiten, D.J., Talbot, J.M., Waters, J.H.(Eds.), Assessment of Nutrient Requirements for Infant Formulas. Life ScienceResearch Office, Bethesda, MD (also published (1999) Journal of Nutrition 128,Suppl. 11S, 2057S–2298S).

McCormick, D.B., 1989. Water-soluble vitamins: bases for suggested upper limits ininfant formulas. Journal of Nutrition 119 (Suppl. 12S), 1818–1819.

Official Methods of Analysis of AOAC International, 2006. 18th ed., Revision 1.Gaithersburg, MD, USA.

Olsen, J.A., 1989. Upper limits of vitamin A in infant formulas, with some commentson vitamin K. Journal of Nutrition 119 (Suppl. 12S), 1820–1824.

Scientific Committee on Food, 2003. Report on the Revision of Essential Require-ments of Infant Formulae and Follow-on Formulae SCF/CS/NUT/IF/65 2003.Retrieved on 10-31-08 from http://ec.europa.eu/food/fs/sc/scf/out199_-en.pdf.

W.C. MacLean Jr. et al. / Journal of Food Composition and Analysis 23 (2010) 44–53 53