Since the early 90's several authors have noted the obesity epidemic affecting developing countries as this change in nutritional status is occurring much faster than in developed countries [3,4,6,7]. The obesity epidemic quickly spread to teenagers, school age children and recently to preschool children, mainly three years old [7]. As a result of early obesity onset several other health problems may arise in childhood, including hypertension, dyslipidemia, type 2 diabetes and cardiovascular problems, which may impair quality of life and decrease life expectancy [8,9].

The literature has already demonstrated the relationship between childhood poverty and abdominal obesity in adulthood [10]. A rise in inactivity is also a risk factor for obesity [11,12]. As mentioned above, obesity is a risk factor for many diseases such as certain cancers, hypertension, type II diabetes mellitus, dyslipidemia, metabolic syndrome and coronary heart disease [13,14]. The primary cause of malnutrition in developed countries is disease. Disease related malnutrition is associated with adverse effects on clinical outcome. These adverse effects vary from impaired wound healing and postoperative complications to mortality. Poor nutritional status has not only been associated with in-hospital adverse effects, but also with adverse effects at both pre-admission and post-discharge [15,16].

The study population included 1639 children aged between 2 and 6 years old, all from low income families. We excluded 95 (5.8%) because they were not of preschool age, hence, it remained 840 (54.4%) boys and 704 (45.6%) girls from a total of 1544 children (Table ​(Table1).1). We also excluded children who presented congenital diseases, history of nutritional and metabolic diseases, chronic diseases that could influence their growth and those who did not agree to collect data. This information was collected during the interview. According to the government (http://www2.santoandre.sp.gov.br/), this population represents a proportion of approximately 40% of all children from 2 to 6 years old from Santo Andre. We observed average monthly income per capita (AMPC) of 0.55 minimum wages (MW - ~U$175.00), the average number of persons per house was 4.6, the median years of mother schooling was 6 years and the median years of father schooling was 5 years. Half of the parents worked in trades such as unskilled and 15% were unemployed (Table ​(Table22).

Weight (W) and height (H) were measured by researchers using internationally accepted techniques [19] under supervision of pediatricians from the daycare centers and all data were collected from records of child care using a standardized form. Anthropometric data were collected in the daycare center, according to Lohman et al [20] method. The child's weight was measured by using the Seca^® electronic scale with a split of 0.1 g, and height was measured by stadiometer wall (Seca^®) with two meters and centimeters and millimeters subdivision [21]. In addition to weight and height we included the following variables: age, gender, birth weight, duration of breastfeeding, maternal age at child birth, level of maternal education, AMPC and how long they had attended the daycare center. Scales and stadiometers were calibrated regularly and reliably.

The values of W, H and body mass index (BMI) were transformed into z scores and percentiles based on the National Center for Health Statistics - 2000 (CDC/NCHS) framework and it was statistically analyzed in relation to the distribution and the association with socio-demographic and health data.

The cutoff points for nutritional disorders analysis were: -2 z scores and 2.5 and 10 percentiles for risk of malnutrition, 85 to 95 percentile for overweight and above the 95 percentile for obesity [22,23].

Data were stored in an Excel^® worksheet. In order to calculate anthropometric indices we used the software EpiInfo 3.3.2 Nutrition February 2005, which described the frequency distribution and central tendency measures (mean, standard deviation and median). Pearson's coefficient was used to evaluate the correlation between age and anthropometric indices.

The analysis was performed by the Stepwise Backward Regression method including the following variables: age, gender, birth weight, duration of breastfeeding, age of mother at birth and how long they had attended the daycare center which in univariate analysis showed a p < 0.15 in the evaluation of its association with excess weight (overweight and obesity, evaluated by z score of BMI). In order to verify if the proposed regression model fits the data well, we applied the F-test.

In Table ​Table44 the z scores for W and BMI tended to reach higher values than those of H in older children. Moreover, higher' values can also be observed in younger children: the z-score of BMI (or BMI/BMI/age) is higher than the z-score of H (or H/A - height/age) in children aged 2 years old as well. However, those differences did not reach statistical significance. The average values of z scores for W and H were higher in females, however, BMI were identical between the both genders. The slope of the three lines was statistically significant with F of 11.53 (p = 0.0426), 20.81 (p = 0.0197) and 27.70 (p = 0.0134), respectively, for W, H and BMI. The average H was the only variable in those younger than 3 years old below the median of the reference. The correlation coefficient for each of the 3 variables was quite high: 0.89, 0.93 and 0.95, respectively for W, H and BMI. Our findings indicate high prevalence of overweight and obesity affecting approximately one in four children.

Child growth evaluation values in Figure ​Figure11 represented the distribution of z scores with respect to W, H, and BMI, which were shifted to the right of the standard, towards higher values, especially BMI. The average values of z scores of W and H were higher for girls, however, BMI were identical, suggesting no significant gender differences in the prevalence of obesity.

In Figure ​Figure2,2, noting nutritional risks below 2.5 p or between 2.5 p and 10 p, for W and BMI, the proportion of children was decreased or equal compared to what was expected according to the CDC/NCHS reference, while the other spectrum extremity showed a higher than expected percentile proportion of obese or overweight values.

Daycare centers in Santo Andre showed growth in stature was similar and even higher, from four years old on, compared to the U.S. reference, even though they were children from low income families. We may propose some reasons for the prevalence of being overweight. First, this fact associated with low prevalence (or even absence) of malnutrition indicate improvement of living conditions of this population, particularly with regard to access to food and health care [4,26]. Second, our results indicated high prevalence of overweight and obesity, affecting approximately one in four children. The high prevalence to be overweight and obesity in children at preschool age was described in several countries, including Chile [27]. The anthropometric profile of preschool children found in Santo Andre is compatible with the rapid nutritional transition shown in developing countries, including Latin America [6]. Third, in these stages of epidemiological nutritional transition, daycare and preschools, which are traditionally regarded as a protective factor for malnutrition, might become a risk for overweight and obesity, since it is where children stay on average 8 hours per day, 5 days per week, and where about 75% of daily calories [28-30] provided. It is clear that this finding cannot be attributed directly and exclusively to daycare centers. Possibly low income families cannot access proper nutrition, which would account for the rest of their daily caloric intake. Moreover, it is possible that these pre-school children perform less physical activity at home, due to the use of television [3,29]. This situation may be considered as a risk factor for children born with normal or high weight [30]. Fourth, the interpretation of our findings indicates that overweight and obesity compromise growth at this stage of epidemiological and nutritional transition in low income children from an urban area of a developing country. This nutritional deficiency due to increased access to inadequate food quality tends to continue throughout life [31-33]. It results in elevated social costs to the health care system, which will financially affect these. All factors cited above may affect our population.

Our data are particularly alarming because obesity was already observed in 3 year old children with a tendency to increase as they grow. This trend may result in high rates of obesity in young adults, with a very early onset of morbidity [9,34]. The metabolic changes that occur as a result of rapid and intense growth tend to continue into itself until adulthood, hence compromising health [8].