Procedure and participants
From February 2008 to December 2010, a case–control study was conducted in nine middle-class elementary schools in Mexico City, which did not represent all socio-economic groups. The study was approved by the Research, Ethic and Biosecurity Committees of the Hospital Infantil de Mexico Federico Gómez; written informed assent and consent were obtained from participants and their parents, respectively.
Because this report is part of another comprehensive study in which the genes associated with obesity were examined [9], the sample was designed to contain two groups that contrasted in their nutritional status, normal weight or obese children, and overweight children were excluded. The participants in this study were school children between 6 and 12 years of age, including obese (cases) and normal-weight children (controls). The children were considered to be obese when their body mass index (BMI) for age and gender were in the ≥95th percentile, whereas the children were considered to be a normal weight when their BMI was between the 25th and 75th percentiles, as defined by the Centers for Disease Control and Prevention 2000 [10]. Individuals who were overweight (BMI >75th but <95th percentile) or malnourished children (BMI < 25th percentile) were excluded.
Instruments
The procedures for measuring anthropometric data and gathering information on food habits, food intake, and physical activity/sedentary lifestyles were conducted by two nutritionists and two nurses from our staff, who were trained and standardized in all procedures. Anthropometric measurements were conducted following international protocols that have been described elsewhere [11], and the Frequency of Food Questionnaires (FFQ) were administered and followed a reference manual for dietetic survey [12]. Information was gathered on children’s physical activity/sedentary lifestyles using a previously validated questionnaire that details activities inside and outside of school [13].
Anthropometric measures
Briefly, the weights and the heights of children were recorded while they wore lightweight clothes and no shoes. Their weights were measured to the nearest 0.1 kg using a digital scale (SECA model-882, SECA Corp., Hamburg, Germany); their heights were measured to the nearest 0.1 cm using a stadiometer (SECA model-225, SECA Corp., Hamburg, Germany). Waist circumference was measured at the end of an exhalation with non-elastic flexible tape (Seca 200) in a standing position at the midpoint between the lower costal border and the iliac crest. In order to decrease any parallax error, participants were asked to climb on an anthropometric box designed for this purpose.
Questionnaire on dietary habits during the school day
The children and their parents were asked the following three main questions: 1) whether the child had eaten breakfast at home, 2) whether the child had brought his/her lunch to eat at school, and 3) whether the child had brought money to buy food at school. The children’s responses required them to only recall information from the day of the study; however, the parents’ responses involved a recall of information from the previous week. Eating breakfast at home, bringing lunch to school, and not bringing money to purchase food at school were considered to be healthy habits. We did not estimate the amount or the quality of the foods that children brought for lunch.
Frequency of food consumed
This measure was collected using an adapted version of the semi-quantitative FFQ for information from the previous month. The course of fieldwork in the schools lasted from March 2008 to May 2009, and questionnaires were applied from Monday to Friday; a meeting with the parents was arranged in advance before school activities began. The questionnaire was sent to parents who could not attend, and another appointment was made to administer the questionnaire to their child. As support material, the interviewer used food replicas in order to standardize the types and amounts of the main food groups consumed by the participants.
The questionnaire contained 110 food items classified in 13 groups. Participants’ food intake per day was estimated, and the amount of food consumed was calculated in terms of the units of measurement (e.g., piece, cup, plate, or spoon) and the size of the unit (i.e., small, medium, or large). For the analysis, the frequencies of consumption were calculated in grams or milliliters ingested per day for each of 110 food items. The participants’ daily intake levels of energy, macronutrients, and fiber were calculated using a food composition database from the USDA-SR23 [14] and Mexican food tables that include data regarding traditional Mexican food [15,16]. The percentage of adequate energy and fiber intake were calculated using the recommendations for Mexican populations [17]. The foods were categorized as solid or liquid and then re-categorized according to their cardiovascular risk. The category of foods without cardiovascular risk included fruits, vegetables, cereals and legumes (e.g., corn tortillas, rice, and oats). On the other hand, foods with cardiovascular risk included the following: those food made from refined flour (e.g., white bread, bread rolls, and wheat flour tortillas), dishes with added fat such as “tamales” (steamed or fried), “sopes” (fried stuffed tortillas), natural fruit juices and industrialized sweetened beverages (e.g., soft drinks, energy drinks, and commercial juices). Finally, the foods were analyzed for their cardiovascular risk according to tertiles of consumption.
Physical activity and sedentary lifestyle
Inside the school, physical activity was estimated using the number of days and hours that children participated in physical education classes and expressed as the number of hours per week. To estimate the amount of exercise performed outside of school, each child was asked to provide the number of days and times per week that he/she played sports such as soccer, football, basketball, cycling, and swimming or engaged in active free playing. Based on the number of days and hours per day of exercise, the metabolic equivalents (METs), which are multiples of the resting metabolic rates, were estimated. This value corresponds to 3.5 mL O2/kg−1/min−1[18], and the children were grouped into tertiles for the analysis.
Sedentary behavior was assessed based on the following information: 1) the number of television sets at home, 2) the time spent in front of a screen (television, computer, or video games; categorized as ≤2 h/d or >2 h/d), and 3) the type of transportation used to go from home to school on an ordinary weekday. In addition, each participant’s cumulative (daytime and nighttime) sleeping hours were categorized as <9 h/d or ≥9 h/d.
Data analysis
Tests were performed to identify the differences between the cases and the controls regarding the following variables: socio-demographic and anthropometric data, dietary habits, physical activity and sedentary lifestyles. The groups were compared using an unpaired Student’s t-test for the continuous variables and Pearson’s χ2 test for the categorical variables. The differences in the medians of energy and macronutrient intake were evaluated using Mann-Whitney’s U test. To evaluate the association between dietary habits, exercise, and sedentary lifestyle and obesity, odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using multivariate logistic regression models. The first model was built to evaluate the association between obesity (dependent variable) and each one of the independent variables (dietary habits, carbohydrates with or without cardiovascular risk, exercise, sedentary behavior, and number of hours of sleep), adjusting by age, gender, and energy intake. Using this information, another model that best fit the data was used, which included different variables (fruits, vegetables, cereals, fried foods from wheat or corn, natural fruit juice, sweetened commercial beverages, physical activity at school, television sets at home and number of hours of sleep), adjusted by age, gender, energy intake and METs. In both models, the correlations within the schools were considered. Additionally, in both regression models, the first tertile of consuming the different types of foods was considered to be the reference category; the assumptions of the logistic regression models were also examined. A p value <0.05 was accepted to be statistically significant. The analysis was performed using the STATA/SE v.11.0 statistical software package (Stata Corporation, College Station, TX, USA).