A blind cluster-randomized field trial was conducted with fifth grade school children. The sample was representative of the population attending fifth grade elementary schools in the State of Mexico. The study was carried out during the 2010-2011 school year in public elementary schools in the 125 municipalities of the State of Mexico. Subjects were beneficiaries of a school breakfast program in both federal and state educational systems with morning and evening shifts. The baseline assessment was conducted in early November 2010; the strategy was implemented between November 2010 and the first half of May 2011 and the final evaluation was conducted between late May and early June 2011. Two measurements--baseline and final--were conducted during the study. Children with disabilities for whom anthropometric measurements could not be performed were excluded.
Design and sample
To determine the sample size, age-adjusted BMI was used as the variable of interest, based on the variances and design effects of a previous study with the same population . For BMI, a standard deviation of 2.79 and a design effect of 1.68 were estimated. The minimum detectable difference was considered to be 0.45a BMI units, with a 95% confidence interval (CI) and a power of 80%. A sample size of 1,000 children was calculated . Sixty schools were selected at random, of a total of 2,969 public schools in the State of Mexico that receive school breakfasts. Thirty were randomly assigned to the intervention group (IG) and 30 to the control group (CG). The average group sizes in schools were 40 students. Within each school, 17 fifth grade children were also randomly selected, resulting in a total of 510 children per intervention group in order to have a sufficient sample size at follow-up. The non-response rate expected in this study was ≤ 5%.
"Nutrition on the go" strategy
The strategy "Nutrition on the Go" consisted of 4 components:
1. A gradual decrease of the energy content of school breakfasts by reducing the fat content in milk, not increasing carbohydrates, decreasing the sugar content of the cereals provided and including fruit.
2. The gradual regulation of food offered within the school, through the technical council of the State of Mexico.
3. Gradual adherence to the physical activity program, according to the requirements of the Ministry of Public Education (SEP, Spanish acronym) .
4. Implementation of an educational campaign, called "Healthy Break," for healthy eating and physical activity. The objectives of this program are to promote consuming one fruit and one vegetable, drinking pure water and performing physical activity (organized games and calisthenics) during break. The campaign was developed by identifying the audience and the interests and needs of the school population regarding basic concepts related to nutrition and physical activity, and defining the materials and methods to convey the information.
Based on these factors, an intervention to be implemented in schools was developed according to the perceived needs of the target population and formative research; the resulting intervention was tested on a small scale using an efficacy study.
The educational materials produced for the "Nutrition on the Go" strategy for healthy eating and physical activity included: student booklets and a facilitator's guide; a school guide; a calendar for parents, as well as videos (or printed handouts for schools with no DVD players) and audio spots.
This study used two types of validation: a technical validation which consisted of a review by specialists and approval of the educational content of the material, and a validation at the population level to test, with a representative group from the target population, whether the contents and form of the materials actually work as intended.
The stages of the validation process  were: 1) Identify the target population and the content of the campaign; 2) First draft of the campaign and the media guide; 3) Select and train staff to validate the material; 3) Prepare material; 4) Convene a group to validate and validation through a workshop; 5) Analyze results and 6) Change the material according to the validated results. A second validation was performed and, lastly, the process to produce the materials was initiated.
The messages and materials were evaluated by a committee of 38 experts that included academic representatives from the Ministries of Education and Health, NGOs, and food industry representatives, among others. Comments were subsequently requested using a questionnaire and modifications were applied according to the suggestions from the experts.
Eight elementary schools were then selected (4 urban and 4 rural) in which a pilot test with the material was conducted in order to identify its strengths and limitations and make the necessary adjustments. An additional study of the materials was subsequently performed using both a quantitative approach of frequencies and a qualitative study based on the categories analyzed. The criteria explored were: attraction, understanding, identification, acceptance and induction into action, as rated by Ziememdorff .
The efficacy study was conducted in 6 schools in the State of Mexico (3 urban and 3 rural) randomly assigned to intervention or control groups. The study's primary audience was the school population between 10 and 12 years of age who were in the fifth and sixth grades of elementary school. Also included were students between first and fourth grade, teachers and parents; a total of 2,762 students participated.
The principal objective was to maintain the BMI after the intervention. The results of this study show that 4 months following the intervention, age-adjusted BMI remained stable (18.6 vs. 18.55). The components of the intervention are described below.
Stages of the intervention
Stage 1. Training. Forty-five promoters were standardized and trained during 3 weeks in the activities that the schools would perform in order to implement the strategy. Anthropometric measurements were obtained by field personnel who were trained and standardized using conventional techniques [34, 35]. Agreement and consistency for the process of standardization was obtained using the Cohen's Kappa test , with a coefficient of 0.76 (95%CI 0.54, 0.83) for height and for weight of 0.87 (95%CI 0.78, 0.90), which indicated acceptance of the measurements. For the standardization in obtaining information using the questionnaires, field workers underwent repeated practice and were evaluated on an ongoing basis.
Stage 2. Strategy implementation in schools. The 60 schools included in the study were visited and baseline evaluations were performed. The intervention was implemented in the 30 schools in the IG for a period of 3 weeks in each school; implementation of the strategy was conducted for 6 months. At the end of the period a baseline assessment both in IG and CG schools was conducted and a final evaluation was carried out 6 months later. Additionally, questionnaires were administered to the students regarding knowledge and self-efficacy in the areas of nutrition and physical activity during both stages.
During implementation in IG schools, a trained and standardized staff member supervised and supported the implementation of the strategy and also evaluated the performance of the activities by monitoring them.
The intervention took place while school was in session. The ongoing activities in schools in the IG were:
a) Nutrition and physical activity workshops. These were divided into 6 sessions which included participatory recreational activities for children to gain knowledge and skills to properly select healthy foods and promoting physical activity. Teaching resources used included the school guide, facilitator's guide, student booklets, videos and printed material.
The workshop was intended to reinforce and expand knowledge and foster self-assessment, proper food choices and physical activity. The dynamics during each session allowed students to actively participate.
b) Puppet Theatre, based on the theory of peer learning . The fifth grade students participating in the study presented a puppet show to students from first to third grades after they studied the script and rehearsed for the performance. Teaching resources provided to each school included the script of the play, puppets and the backdrop.
These activities were conducted once per week for 4 weeks and the puppet show was presented once per month.
Activities with elementary school teachers included:
c) Two-day workshops in each school to raise awareness about healthy eating and physical activity. The workshops sought to convey to teachers the importance of healthy eating and physical activity through dynamic and playful activities to promote participation. Teaching resources included PowerPoint presentations, games, group dynamics, handouts and descriptive cards.
The workshops were taught by nutritionists and health professionals (nurses and social workers) who were previously trained by nutritionists, psychologists and educators and physical trainers with bachelor degrees.
Activities to change the school environment included:
d) Sale of fruits, vegetables and pure water in the school's store cooperative. A session was held for store personnel to convey information about healthy eating, make suggestions about types of food to sell in schools and recommend the daily sale of vegetables, fruit and pure water. The importance of the responsibility of the cooperative (the food store inside the school) for preserving the health of the school community was addressed. The duration of each session was 1 hour.
e) To promote the consumption of pure water, spots were broadcast using the schools' PA systems, and water bottles were delivered to children and teachers to encourage water consumption. A journal was kept on a sporadic basis to verify that the children were carrying their water bottles with them and that they only contained water.
Activities with the educational community included:
f) Physical activation. Organized activities involving motion were conducted twice per week. Activities performed each day before the start of classes included warm-ups, activation and relaxation. Recommendations to support physical activation were provided through the school guide and a CD with music for established activities. Weekly activation sessions gradually increased from 2 to 5 days.
g) Broadcasting of audio spots on the schools' PA systems. Spots were broadcast 3 times per week during the break. The central messages were aimed at promoting the consumption of fruits, vegetables and pure water during break and to promote physical activity in children, with an average length of 1 min and 15 seconds per spot.
h) Organized games during break (once per week). Active and safe participation of teachers and children was promoted during break. Educational materials were provided for these activities, including posters with suggestions for team games and activities that involved moving during 30 minute breaks. To this end, the schools were provided with balls, ropes and hoops, as well as a guidebook, which also were useful for physical activation sessions and physical education classes.
i) Placement of banners at the entrance of the school. In order to highlight the campaign in the school community, a banner was hung that read, "This school promotes healthy breaks."
Activities with parents included:
j) Delivery of recipe calendars. Calendars were distributed that included healthy recipes for school lunches in order to disseminate information to parents about healthy eating and physical activity.
A baseline test was conducted in all schools to establish the initial characteristics of intervention and control groups. Information was also obtained related to anthropometrics, socioeconomic level, food, physical activity, self-efficacy and knowledge. The same information was obtained for the final evaluation
Body mass index
Weight and height were measured. The promoter, who was previously standardized, asked the children to remove their shoes, wear minimal clothing, and unbraid hair that could interfere with the height measurement. The weight was determined with Tanita electronic scales with an accuracy of 100 g and the height was measured using Dynatop stadiometers with a capacity of 2 m and an accuracy of 1 mm. Weight and height of the children who participated in the survey were measured by a trained and standardized field team [34, 35].
BMI (BMI = kg/m2) was determined for all students to classify them as adequate BMI, overweight or obese, considering the distribution and cutoff points proposed by the International Obesity Task Force (IOTF) .
A Food Frequency Questionnaire (FFQ) was administered to the parents of every child. The FFQ employed was used in the 2006 National Health and Nutrition Survey (ENSANUT-2006) .
A semi-quantitative questionnaire was used to record the physical activity of students, based on the Youth Activity Questionnaire developed and validated by Hernández et al. 1999 .
We designed a multiple choice questionnaire with two sections and answers presented graphically, to assess children's knowledge about diet and physical activity. The questionnaire consisted of 13 items, of which 7 corresponded to diet and 6 to physical activity.
To evaluate the self-efficacy of children with respect to physical activity, a dichotomous scale with 12 items was used, which was designed and validated for school populations by Aedo A. and Avila H. in 2009 . The scale consists of 3 dimensions for self-efficacy: the search for positive alternatives, the ability to face potential barriers and skill or competence related to expectations.
To assess self-efficacy on the topic of healthy eating, certain questionnaires were used as a reference, and the items were adapted to tie them with the physical activity scale, resulting in a dichotomous choice questionnaire consisting of 13 items. The values "1" and "0" were assigned to the dichotomous scale (meaning yes and no, respectively). Both scales have been tested and validated with children of similar ages in the United States [42, 43] and were adapted for the characteristics of Mexican children.
The results were obtained from the sum of positive responses (value 1), where the minimum value obtained was "0" and the maximum was "12" or "13," according to the number of items in each scale. The percentage of positive responses was then estimated for each questionnaire.
We previously categorized the level of self-efficacy for each child into 3 categories:
1) Low self-efficacy, rated between 0 to 33.3%, when the child had little confidence in himself or herself in terms of modifying eating or physical activity behavior.
2) Medium self-efficacy, rated between 33.4 to 66.6%, when the child believed he/she could perform various activities (related to his/her nutrition or physical activity) but was not sure about his/her ability to successfully complete them.
3) High self-efficacy, ranked between 66.7 to 100%, when the child was convinced he/she would succeed in performing a certain behavior and was willing to modify his/her actions and behaviors.
To measure self-efficacy, the Eating Self-Efficacy Scale was used, adapted into Spanish .
Socioeconomic status was calculated by obtaining the children's age, sex, housing characteristics and possession of goods. A socioeconomic index (SEI) was calculated using the principal components method, with 7 variables, where the first principal component explained 40.2% of the total variance. This in turn was divided into tertiles to obtain socioeconomic levels.
Ethical aspects of the study
First, a written authorization for the school's participation in the study was requested from school principals and the teachers of the groups. During school meetings, parents were explained the purpose and procedure of the study, the lack of risks, the time needed to administer the questionnaire and the process to measure weight and height, as well as the importance of the participation of the child. They were also informed that the children's participation would be voluntary and that there would be no consequences or limitations to their right to receive school breakfasts if the child withdrew from the study, which he/she could do at any moment. After the explanation, mothers were asked for written informed consent to interview their children.
The same procedure was repeated with the children selected to participate in the study through an agreement letter.
The protocol for this study was submitted at every stage to the ethics, biosafety and research committees at the Mexico National Institute of Public Health.
For the baseline stage, a descriptive analysis was conducted of frequencies and means, with their respective confidence intervals, in order to observe whether there were differences between intervention and control groups at the beginning of the study.
For the variables related to knowledge, results were compared and analyzed using the average scores obtained by the children from each assessment, based on a scale of 1 to 10 for both topics and the treatment groups.
The following values were used to determine the percentage of students who passed the knowledge questionnaires: > 6 = passed; ≤ 6 = failed (Secretarial Agreement, SEP 2009).
Subsequently, a generalized ordinal logistic regression model was developed in order to observe changes in BMI ordinal categories (normal, overweight and obesity) as a result of the implementation of the strategy (intervention effect TR = 1 as a dummy variable), the stage (baseline or final also using a dummy variable with stage = 1 for final) and its interaction as an estimator of the average effect of the time-corrected effect of treatment (difference-in-differences estimate). The regression variables used were physical activity and the difference-in-differences effect. The following variables were included: physical activity, consumption of food energy, self-efficacy scale and knowledge about eating and physical activity for both cases, adjusted for age, sex and SEI.
As part of the analysis, a validation was conducted in which the possible co-linearity between variables and the corresponding statistical assumptions were verified, as well as the adjustment with the use of the cluster and panel data design structure to correct the standard errors in the estimation of coefficients for models . A significance of p ≤ 0.05 was used for main effects and p ≤ 0.1 for factor interactions. All analyses were conducted using STATA 11, svy and gologit2 free usage routine .