Iron deficiency and anemia in infants and young children impair neurodevelopment. Efficacious interventions for reducing the prevalences of iron deficiency and anemia, under controlled conditions, are available. However, little information is published about the effectiveness of large-scale programs.

Objective. Assess the effectiveness on iron deficiency and anemia in young children of a large-scale program that provides iron fortified milk at a subsidized price to low income households. Design, Setting, and Sampling. A randomized effectiveness study in 12 milk distribution clusters randomly selected from 542 clusters in 4 States in Mexico. Selected clusters were randomly assigned to receive iron-fortified milk FM, N=7) or non-fortified milk (NFM, N=5). Over 95% program beneficiaries with children 12-30 mo of age in the 12 clusters (490 in FM and 275 in NFM) were enrolled in the FM and NFM interventions and participated in a baseline survey. 635 children (405 in FM and 230 in NFM) and 584 children (371 in FM and 213 in NFM) completed the 6 and 12-mo follow-up surveys, respectively. Intervention: A daily portion of FM contained 5.28 mg of iron, 48 mg of sodium ascorbate and other added micronutrients. The FM and NFM were distributed to beneficiaries at a subsidized price in Milk Distribution Centers in the 12 clusters.

Main outcome measures: Mild-to-moderate anemia (Hb: 90-109 g/L), mild anemia (Hb: 100-109 g/L), moderate anemia (Hb: 90-99 g/L),iron deficiency: Serum Ferritin (SF) <12 ug/L and soluble transferring receptors (sTfR) >6 mg/L.Serum Zinc deficiency (< 65ug/dL), and stunting (<2SD Height/Age.

Intervention effects at 6 and 12 moths were assessed using General Linear Mixed Models with three repeated measures.

Introduction. Iron deficiency anemia is the most common nutritional deficiency worldwide (1-3).Children less than 5 years are the most affected(4). Almost 24% of Mexican children 1-4 years of age are anemic and 52% iron deficient (5). The peak prevalence of anemia (38%) occurs in children 12-23 months of age (6). The prevalence is higher in low socioeconomic households (29 %) but remains high even in the highest quintile (17%). Anemia and iron deficiency in infants and young children are associated with adverse effects on neurodevelopment (7-12).

During several decades, the Mexican government has sold whole milk at subsidized prices to low income households with children 1-11 y of age through a federal program (Liconsa). In 2000 a decision was made to fortify the subsidized milk with iron, vitamin C and other micronutrients to contribute to the reduction of IDA and other micronutrient deficiencies. At that time, 4.2 million children 1-11 y old from low income families were beneficiaries of the program. There are very few examples of large scale public health interventions using iron fortified whole milk.

The effectiveness of interventions such as food fortification for reducing anemia and iron deficiency is often questioned (14) and little information is available on the effectiveness of these strategies implemented through large scale programs. Reliable evidence of the effectiveness of these strategies applied in large-scale programs is needed for planning interventions aimed at reducing the prevalence of iron and zinc deficiencies and anemia in children.This study aims to document the effectiveness on iron and zinc deficiencies,anemia and stunting, in young children of a large-scale program that distributes iron and zinc fortified milk to low income households at subsidized prices.

Methods The subsidized milk distribution program. The program was created in 1944 and is operated by a public organization (Liconsa)(15). It distributes whole milk at subsidized prices to families living in poverty. The milk distributed is intended mostly for children 1-11 y.

It is distributed in milk distribution centers throughout the Country as dry powder in 240 g sachets which yield two liters of fluid milk after hydration or as a 1 L container of fluid milk. Children receive the equivalent of 400 mL of fluid milk per day either in dry or fluid form for reducing the prevalence of iron deficiency anemia and other micronutrient deficiencies, through the fortification of the product with iron, zinc, vitamin C and micronutrients that were deficient in the diet of low income children in Mexico (5, 16). The program serves approximately 5 million individuals of which about 4.2 million are children <11 years and about 650,000 were children 12-30 months of age.

Design and sample. This randomized effectiveness trial on children 12-30 mo of age was conducted in 4 States of Central and Eastern Mexico. The randomized effectiveness trial was conducted in 12 milk distribution clusters (MDC) comprising periurban areas serving around 5000 beneficiaries with one or more milk distribution centers. These clusters were randomly selected from a universe of all MDC (n=542) in 4 States in central and eastern Mexico. The twelve clusters were randomly assigned to receive either fortified (n=7) or non-fortified (n=5) milk using a random numbers list. The clusters assigned to the non-fortified milk started receiving fortified milk at the end of the trial. The milk distributed to the participants in the trial was all in the dry form.

The duration of the trial was 12 months with a baseline survey and 2 follow-up surveys at approximately 6 and 12 months. The baseline study was conducted between May 28 and July 30, 2003. The 6 mo and 12 mo follow-up surveys were conducted between December 2003 and February 2004 and between June and July 2004, respectively. The mean time between the baseline and the 6 mo follow up was 5.4 mo and from the 6 to the 12 mo follow was 6.8 months.

The packages of fortified and non-fortified milk were undistinguishable, except for a color-coded band in the upper corner of the sachet. The color code was unknown to researchers, field workers, and program beneficiaries, and was not disclosed before data analysis. The iron and zinc content of the fortified milk was analyzed by our laboratory on samples of milk powder obtained from LICONSA at the beginning of the intervention.

Mothers or caretakers of children participating in the study received the same information and instructions provided regularly to all beneficiaries of the milk distribution program at the MDC. They were instructed how to reconstitute the milk powder and were encouraged to feed the selected children 400 mL, preferably at two separate times during the day (200 mL in the morning and 200 mL in the afternoon). Milk was bought and collected by each family on a weekly or bimonthly basis at the MDC. The portion of 400 mL of rehydrated fortified milk (FM) contained 5.28 mg of iron as ferrous gluconate, 48 mg of sodium ascorbate, 5.28 mg of zinc as zinc oxide and 36.2 mg of folic acid. The non-fortified milk (NFM) contained per 400 mL: 0.16 mg of iron, 1.6 mg of zinc and 6.8 mg of vitamin C. The micronutrient premix,purchased from DSM Nutritional Products (Herleen, The Nederland) was added to the powdered milk in the plant by the manufacturer (LICONSA, SA,Queretaro, Mexico) and 220 g units of the products were packed in metallic foil sachets.

A list of households with children 12-30 months of age (n=798) who were beneficiaries of the program in the 12 distribution clusters was provided by program officers at the beginning of the study. 510 of these children corresponded to the distribution clusters assigned to FM and 288 to the clusters assigned to NFM. The number of children in the FM clusters was larger for two reasons: more clusters were assigned to receive FM as a request from program officials who wished to minimize the number of children who would not receive the fortified milk immediately. In addition, some of the clusters randomly assigned to the fortified milk intervention served larger populations of beneficiaries. The parents or legal guardians of the 12-30 month old children were invited to participate and those who accepted signed an informed consent letter after a detailed explanation of the objectives, nature, and risks posed by the study. Prior to baseline measurements 33 children were excluded from the study (20 in the FM and 13 in the NFM groups) because they were either severely anemic (Hb concentrations < 90 g/L, n= 26) or refused to participate (n=7. Therefore, 765 children participated in the baseline measurements (490 in the FM and 275 in the NFM clusters). Between the baseline and the 6-mo follow-up measurements, 130 out of these children were lost to follow up for several reasons. Therefore 635 children completed 6 mo of follow up (n= 405 in the FM and 230 in the NFM groups). Between the 6-mo and the 12-mo follow-up measurements, 51 additional children were lost to follow up. Therefore 584 children completed 12 mo of follow up (n= 371 in the FM and 213 in the NFM groups).

For the 635 children who completed the 6 mo follow-up and the 584 who completed the 12 mo follow-up, hemoglobin concentrations (the main outcome of this study) as well as basic socio-demographic variables were available. iron status. The protocol was reviewed and approved by the Human Subjects and Ethics Committee of the National Public Health Institute, Mexico.

Data collection and processing. At the beginning of the study a questionnaire about the housing characteristics and possession of household goods was applied to the mother or child caretaker in all households. Hemoglobin concentrations were determined at baseline and 6 and 12 months after in capillary blood samples obtained by finger prick using a Portable Photometer Hemocue (HemoCue, Angelholm, Sweden)17, 18. Reliability of the photometers was assessed during fieldwork at the beginning and at the end of each working day. A three-level liquid quality control check (4C-ESControl, Beckman-Coulter, Miami Fla, USA), and the readings of a precalibrated reference cuvette included with the equipment were used for such purpose. The mean difference between duplicates was 0.3 ± 9.9 g/L, (P=0.36 for liquid quality control material and -0.24 ± 3.6 g/L, P= 0.27 for the reference cuvette). Venous blood samples were drawn from children at baseline and 6 and 12 mo after. Samples were centrifuged and serum was stored in color coded cryovials kept in liquid nitrogen until delivery to a central laboratory. Commercial kits were used to measure the SF concentrations, (Dade Behring Inc, Newark, DE 19714, U.S.A.) and sTfR (Dade Behring, Maburg, Germany) by ELISA. C reactive protein (CPR) was determined by nephelometry, using monoclonal antibodies (Behring Nephelometer 100 Analyzer, Behring Laboratories, Messer Grisheim Gmbh,Frankfurt, Germany. Serun zinc concentrations were mesured by Atomic Absorption spectrometry by a Analyst 300 model with graphite furnace(Perkin Elmer).

Length and weight were measured at baseline and at 6 and 12 mo after enrollment. Weight was measured to the nearest 10 g using an electronic scale (Tanita, Model 1583, Tokyo, Japan). Length was measured to the nearest millimeter using a locally made measuring board in children <24 months of age and standing height, using a stadimeter with precision of 1 mm (Dyna-Top, model E-1,Mexico City, Mexico) in children 24 months or older. The measurements were obtained using standard techniques (19, 20). The birth date was reported by the mother and corroborated in a large proportion of children using birth certificates. Length and weight data were transformed to z-scores by using the WHO/NCHS/CDC reference data (21).

To verify compliance a field worker visited the household every month, at the usual times of milk intake by the child, to verify the correct reconstitution of milk and to register the amount of milk consumed. During the same visit a questionnaire was applied regarding the intake of milk during the previous week.

A morbidity questionnaire was applied during the same visit. Feeding practices were assessed every 2 months through a food frequency questionnaire applied to the mother or caretaker.

Anemia was defined according to CDC recommendations as hemoglobin concentrations <110 g/L at sea level (18). Hemoglobin concentrations were adjusted for altitude following the equation proposed by Cohen and Haas (22). Iron deficiency was defined as SF <12 ug/L and as sTfR > 6 mg/L. SF indicates the level of storage iron and sTfR reflects tissue iron deficiency. Zinc deficiency was defined as by the IZnCG (<65ug/dL).

Using the information about housing characteristics and possession of household goods, an indicator of socioeconomic status (SES) was derived by the first component obtained by Principal Components Analysis (23) Only variables with factor loadings > 0.5 were maintained in the model. The variables included in the factor were: flooring material, ceiling material, availability of piped water,sanitary service, possession of refrigerator and washing machine as well as the number of electric appliances in the household: radio, TV, video player,telephone, and computer. The resulting standardized factor scores were divided into tertiles which were further used to construct SES status categories.

Statistical methods. Characteristics of study children at baseline and at 5 and 12 months and the socioeconomic status of their families were compared between intervention groups using t-test on continuous variables and Ji2 for categorical variables (24). Distributions which were not normal were transformed and were compared using t-test. When adequate transformations were not achieved, the Wilcoxon Rank-sum test was employed (24. Intervention effects at 6 and 12 moths were assessed using General Linear Mixed Models with three repeated measures (25). Dependent variables were the prevalences of anemia or iron deficiency by either SF or sTfR. The interaction term between intervention and indicator variables for 6 and 12 months were used as the measure of effects at 6 and 12 months adjusting for baseline measurements. All models were adjusted for cluster effects (Milk Distribution Centers) and covariates. Adjusted probabilities were obtained to illustrate effects in graphic form26. Main effects were considered statistically significant at p values < 0.05 and interactions at p values <0.10. Statistical analyses were carried out using STATA (version 9.0)

Inclusion Criteria:

• Healthy children 12-30 months of age who were beneficiaries of the program in the 12 milk distribution clusters

Exclusion Criteria:

• Children with severe anemia (Hb concentrations < 90 g/L)