Fortified Milk, Growth and Micronutrient Status in Mexican Toddlers (MilkGrowth)

The study was conducted in Mexico. The investigators included children 12 to 30 months of age who attended day-care centers. These day- care centers are part of a national government program which main objective is to support working mothers and those who intend to work and do not have access to a health service and are at risk of poverty. The program also includes single parents in charge of their family with children between 12 and 36 months of age.

The study described herein is a randomized, parallel,double-blind, controlled trial by administration of growing up milk (GM) and fortified milk (FM).

Randomization and masking:

The Moses-Oakford method was used to randomly assign 189 children to 1 of 2 groups in this double-blind controlled trial by the administration of growth milk (Enfa-Grow® GM) (GM) and fortified milk (FM). There were 96 children in the red-gray group and 93 in the blue-green group, the study will be blinded until the statistical analysis is finished. One group received GM and the other group receive FM. Both groups received 480 mL/d (240 mL at 7:00 a.m. and 240 mL at 16:00 p.m.). Both supplements were indistinguishable, same color, (white milk powder) odor, and flavor except for the color-coding of the can. The color code was unknown for researchers, field workers, users and analysts. Milk was offered to the children 5d/wk for a period of 4mo under supervision of the study personnel who recorded whether the supplement was consumed as well as the amount.

Exclusion Criteria:

The investigators did not included children, who were breastfed at the time of the study, neither child receiving multiple micronutrient supplementations. If a child had capillary hemoglobin <9.0g/dl, at baseline or was clinically sick was not included.

Data Collection Anthropometric measures Anthropometric measures were taken at baseline, 8 and 16wks. Weight was measured to the nearest 10g with an electronic scale (TANITA Scale, Tanita Corp., Arlington Heights, IL capacity 14 kg for infants and 36 kg adults, Tokyo) and length and height were measured to the nearest millimeter with a length board (Schorr Industries, Glenn Burney, MD). Head circumference was taken with a flexible non-stretchable plasticized measuring tape to the nearest millimeter (model, Hoechstmass-Rollfix 150cm). Field workers were trained and standardized using standard techniques according to protocols. Weight, length and head circumference were transformed to Z scores by using the 2006 World Health Organization (WHO) reference standards. Prevalence of stunting, wasting and underweight were calculated using minus 2 z-scores for each indicator (length- or height-for-age, weight-for-length or height- and weight-for-age) at specific cutoff points for age and sex. Overweight /obesity were defined as weight for length/height above plus 2 z-Score.

Hemoglobin concentration was determined in capillary blood samples obtained by finger prick and measured in a Portable Photometer -HemoCue-, at baseline. According to selection criteria if a child had <9.0g/dl, the child was referred to the nearest health center. After overnight fasting, venous blood samples (10ml) were drawn from antecubital vein at baseline and at 8 and 16 weeks thereafter. Samples were obtained by a trained nurse who had had experience working with children and according to protocol procedures established by the Biosecurity Commission at the National Institute of Public Health (NIPH). Capillary and venous blood samples were taken at each day care-center. Capillary blood samples were taken only at baseline. Some mothers were withdrawing from the study because of the blood sampling. Samples were centrifuged using a portable centrifuge EBA8 (Hettich, Tuttlingen, Germany) at 280 X g; 20 min in situ and serum was stored in color-coded cryogenic vials, and preserved in liquid nitrogen protected from light until delivery to the Biochemical and Nutrition Laboratory at the NIPH.

Once in the laboratory, samples were stored at -70ºC until ready to be analyzed. Each micronutrient was measured as follows: Ferritin, zinc and folate: Commercial kits were used to assess the serum concentrations of ferritin (Dade Behring Inc.) The serum concentrations of zinc were measured by inductively coupled plasma atomic emission spectroscopy using a Varian Vista Pro CCD spectrometer. The serum folate was released from protein using sodium hydroxide, dithiothreitol, and potassium cyanide, and transformed into cyanocobalamin and stable folates. Their concentrations were measured by competitive enzymatic immunoassays in a TOSOH automated immune-analysis. Vitamin D: serum 25-OH-D3 was analyzed by CMIA using an Abbott Architect module. Vitamin A: Serum retinol was extracted with 100% pure ethanol; the ethanol was evaporated under a nitrogen flux and dissolved in 100 ml of 10% ethanol. Determinations were performed by high-performance liquid chromatography (HPLC) in a Waters instrument (Waters Co., Milford, MA, USA), using a 3.9 x 150 mm column Nova-Pak C18 ODS (Waters Co.), with a mobile phase of CH3OH, 100% with a flux velocity of 1.5 ml/min, at a wavelength of 325 nm. Serum lipids: lipid serum profile was determined by a gas chromatographer (5890 Series II, Hewlett-Packard, United States) Morbidity data were collected daily at each day care center. A checklist was used to record symptoms of diarrhea and acute respiratory infections. Mothers or caretakers were asked to recall any symptoms regarding milk consumption.

Socioeconomic status Information about the participant's household's characteristics, the education level and parents' occupation was obtained by interviewing of the mother.

Intervention Milk powder was prepared and reconstituted every weekday at each day-care center according to WHO and Day Care guidelines. In the morning approximately at 7a.m., trained personnel prepared the milk according to the color-coding assigned to each child. Bottles were used to facilitate measuring in milliliters. Each bottle was weighted before and after preparing milk in a diet measuring scale (OHOUSE, model CS5000, capacity 5000gx2g).Trained cups were also used. Each child received an individual bottle or cup with its own milk (240ml). Once each bottle or cup was prepared it was given to the child. After the child finished drinking the milk, leftovers were weighted and recorded in the consumption form. Milk was prepared again (240ml) according to protocols before each child left approximately at 16:00 p.m. Leftover milk was weighted, measured and recorded. Mothers or caregivers were instructed on how to prepare powder milk at home according to WHO guidelines. This was done on weekends and for children who drank milk before arriving to the day care center. Mothers were given the amount of milk necessary to prepare 480 ml per day during the weekend. Mothers registered the amount of milk left by the child in a consumption form and were trained to estimate and register milk consumption.

Definition of micronutrient deficiencies:

1. Folate: Children with serum concentrations of folate below< 4ng/mL were categorized as deficient.
2. Ferritin: Children with serum concentrations below <12 μg/L (LIS).
3. Zinc: Children with serum concentrations below <65 μg/dL.
4. Vitamin A: Children with serum concentrations below < 20.02 ug/dL (<0.70 µmol/l).
5. Vitamin D: Children with serum concentrations below < 20nmol/L (<8ng/ml) were considered with severe deficiency.

Statistical analysis Baseline general characteristics were compared between study groups using the t statistic for independent samples. For categorical characteristics the chi-squared statistic was used.

For each continuous response variable -i.e. growth or micronutrients-, the differential effect of supplementation was estimated with a multiple regression model that included an indicator variable of study group, an indicator variable of the follow-up period and the corresponding interaction of these two terms. For binary outcomes logistic regression model was used. Baseline characteristics that significantly differed between study groups were included as covariates. All standard errors were adjusted for data dependencies at the subject level. Anthropometric continuous responses were analyzed assuming normality whereas micronutrient continuous responses were assumed to have a lognormal distribution. For the latter, the outcome variable was log transformed before estimating the model and adjusted geometric means were obtained as post estimation. Adjusted probabilities were obtained after fitting the logistic models and applying the inverse of the logit function to the linear prediction. Adjusted geometric means and probabilities were obtained through predictive margins and their standard errors through the Delta method; changes over time for these predictive margins were obtained as average marginal effects.

A socioeconomic status indicator was obtained abstracting the first principal component from household material characteristics and services. The first principal component explained 30% of total variance.