Occurrence of overweight in schoolchildren and analysis of agreement between anthropometric methods

Abstract The child population is strongly affected by obesity. Accessible and reliable strategies for the obesity diagnosis are of utmost importance.. The aim of this study was to identify childhood obesity according the WHO (World Health Organization): malnourished, healthy weight, overweight and obese. It was collected measures of height, Body Mass Index (BMI), Waist Circumference (WC) and Triceps Skinfold Thickness (TSF) of 449 children from Municipal School of Araras/SP, from 7 to 10 years old. It was performed a Spearman correlation test between BMI, WC and TSF variables. Also, was realized cross tabulation between the found results by the different methods, constructing a contingency table 2x2, with absolute frequency of boys and girls classified as “without overweight” and “with overweight”. The concordance between methods was analyzed by kappa index. In the results, 28.3% of children presented overweight according to BMI, with higher prevalence in boys. Generally, the found results through TSF showed strong correlation with both BMI and WC (rs=0.7994 e rs=0.7519, respectively). The same was observed when data was analyzed separately by sex. When crossed the TSF data with BMI and WC, the kappa index demonstrated a satisfactory concordance (0.4419 e 0.5161, respectively). The TSF can be suggested a method to body composition assessment and cardiometabolic risk in children.


INTRODUTION
Obesity is defined as the excessive accumulation of body fat and can be determined by the percentage of adipose tissue presented by the individual. It is a chronic disease considered a worldwide problem, reaching epidemic proportions in both developing and developed countries, regardless of gender and age 1 .
The child population is, also, strongly affected by obesity, which is considered a concerned scenario. Currently in Brazil, 34.8% of girls and 25.9% of boys are overweight and have been increasing over the decades 2,3 . Obesity phenotype is high associated with lifestyle changes achieved by contemporary society, where children consume large amounts of foods with high levels of fat and carbohydrate. As well, the reduction of physical activity levels 4,5 .
The overweight at this stage of life is a risk factor for a number of health problems, which may manifest on early age or late age. Among these, we can highlight physiological and metabolic complications such as elevation in blood pressure, dyslipidemia, cardiovascular disease and type 2 diabetes; postural changes, accompanied by joint pain, as well as psychological problems related to low self-esteem and self-concept [6][7][8][9] .
There are evidences that a great number of obese children and adolescents will remain obese when they reach adulthood 9 . Therefore, the older you are and the greater overweighed you are, the most difficult it will be to reverse, due to the incorporated eating habits and installed metabolic changes. For that reason, strategies to combat this condition have become important. Not only in pursing the weight lost, but also in diagnostic overweight in an accessible and reliable way. Thus, both Body Mass Index (BMI) 10 and skinfold method 11 are widely used for diagnosis. However, the studies which uses skinfold measures the thickness of more than one skinfold, such as triceps, subscapular and suprailiac 12 . Thus, it was proposed the use of triceps skinfold (TSF) as a respectable method to identify overweight in school-age children.
The present study aimed to identify the nutritional status and body fat content of children aged 7 to 10 years in a municipal school of the city of Araras, state of São Paulo, Brazil. Therefore, BMI, TSF and WC methods were used to correlate the results obtained and to compare the agreement between the methods, besides providing subsidies for the best choice of methods for the analysis of body composition of school age children.

METHOD
The present study was classified as a cross-sectional, observational and quantitative field research. Anthropometric measurements of height, body mass, WC and TSF were collected from 449 children (209 boys and 240 girls from a municipal school in the city of Araras/SP, aged 7 to 10 years. All children were previously instructed to wear shorts and t-shirt, and to remain barefoot to measure body mass and height. Participants' body mass was measured using a Filizola ® scale with 100g divisions and maximum load of 150 kg, as well as a portable Sanny® stadiometer for height measurement. Those two data were used to calculate BMI, where the body mass value in kilograms (Kg) is divided by the square of the height in meters (m 2 ). The BMI value was applied to classify the nutritional state of the subjects as malnourished, healthy, overweight and obesity. For this, we used the percentile table proposed by the World Health Organization, considering gender and age 13 , and the cutoff values proposed by Division of Nutrition, Physical Activity and Obesity of Centers for Disease Control and Prevention 14 .
To obtain the WC, all children were positioned in anatomical position, with a measuring tape in a horizontal plane over the umbilical scar, and being analyzed in WC percentiles according gender and age 15 .
To obtain the TSF, a Sanny ® adipometer was used. The anatomical point of the TSF was determined parallel to the longitudinal axis of the arm on the posterior face, and the point was the average distance between the superolateral edge of the acromion and the oclecranon. For the analysis of the Percentiles, the gender and age of the children were considered 16 . The same Physical Education teacher performed all evaluations.
The Spearmen correlation test was performed between the variables BMI, WC and TSF, with a statistical significance as p < 0,05. The test was executed on BioEstat 5.3. ® program.
A cross tabulation was also performed between the results found by the different methods and constructing a 2x2 contingency table, with the absolute frequency of boys and girls classified as "non-overweight" and "overweight". Then, the kappa index analysis was performed, thus allowing evaluate of the agreement between the methods os diagnosis of obesity. The kappa index was classified as proposed by Landis and Koch 17 and, also, performed by the BioEstat 5.3. ® program.
All procedures followed the principles of Ethics Committee for Human Research, under protocol CAE number 64158617.3.0000.5385. For involving minors, parents and/or guardians signed the Informed Consent agreeing to data collection. Table 1 shows the results regarding the occurrence of overweight and obesity in the children participating in the study, totally and divided by gender. Overall, 28.3% of 449 children were overweight (127 children) according to BMI, 31% male (n = 65) and 25.9% female (n = 62). On the other hand, girls had a higher occurrence of overweight, but the number of obese was higher among boys ( Table 1).

RESULTS
The Lillieforns test showed that it was a non-normal sample. Thus, the Spearman coefficient was performed, so that when correlated BMI and TSF, we observed an r s = 0.7994 (p<0.05), revealing a strong correlation ( Figure 1A). Strong correlations between the methods were also observed when analyzes were made separately by gender, with r s = 0.7901 (p<0.05) for boys and r s = 0.8220 (p<0.05) for girls ( Figures 1C and 1E respectively). The same pattern was noticed when we executed the correlations of WC with TSF, with r s = 0.7519 (p<0.05) for the whole sample ( Figure 1B), with r s = 0.7653 (p<0.05) for the boys ( Figure 1D) and r s = 0.7550 (p<0.05) for girls ( Figure 1F). The cross tabulation of the results obtained by BMI, WC and TSF are presented in 2x2 contingency tables, with Table 2 being the cross tabulation of BMI TSF results and Table 3 the cross tabulation of WC and TSF results. When the BMI and TSF data were crossed, the kappa index revealed a satisfactory agreement, both for the whole sample (0.454) and for boys (0.4260) and girls (0.4527) ( Table 2). Satisfactory agreement was also found when WC and TSF results were crossed, with a kappa index of 0.5161 for the entire sample, 0.5381 for boys and 0.4970 for girls (Table 3).

DISCUSSÃO
In our study, we found overweight in 28.3% of children (15.6% overweight and 12.7% obesity) ( Table 1). These numbers may be related to the fact that currently a number of social factors contribute to increased physical inactivity and decreased energy expenditure of children, among them the advancement of technology that makes electronic games more attractive and policy-related factors, such as decreased safety for leisure activities on the streets and reduction of free space for such activities 18 . Corroborating our finding, studies report an alarming number of overweight children 19,20 .
Among genders, boys had a higher occurrence of overweight than girls, accompanied by a higher occurrence of obesity (Table 1). Opposite data were observed in the literature, since there are studies showing the highest occurrence in girls 21 , or even found no difference between the genders 19 . Therefore, we can suggest that obesity at school age is not correlated with gender.
According to literature data [19][20][21] , BMI is widely used for the diagnosis of overweight and obesity when dealing with large populations, supported by a several studies that have shown that it is a reliable method for this purpose. A study of children in a public school with a mean age of 9.2 ± 1 years showed good agreement between BMI and skinfold method 22 . Similar results were compared with those obtained by X-ray double emission densitometry (DEXA) method, considered the gold standard for body composition evaluation 23 . In the present study, we observed that only the measurement of TSF thickness is a method that shows good agreement with BMI and may be a new strategy for investigating the nutritional status of school-age children. Moreover, it is also considered a fast, practical and relatively low cost method, with easy interpret results. Others studies were performed comparing the results obtained by BMI and skinfold method 11,24 . Januário et al. 11 investigated the occurrence of obesity in 200 children from 8 to 10 years old in public schools in Londrina -PR. The authors observed a kappa index of 0.43 for boys and 0.50 for girls. These results are similar to those observed in our study, 0.43 for boys and 0.45 for girls (Table 2). Similar results were observed  by Landis and Koch 17 that evidenced a moderate agreement between the methods. It should be taken into consideration that in our study only the TSF was evaluated, while the study by Januário et al. 11 used both TSF and subscapular fold. Thus, this study proposes that only TSF measurement can be a new method for diagnosis of obesity for adolescents, given the strong correlation with BMI ( Figure 1) and the moderate agreement by the kappa index (Table 3), revealing the TSF may also be a good method for the diagnosis of cardiometabolic risk in this population. Since WC has been proposed as an important tool for this purpose 25,26 . However, it is important to highlight that the present study used more than one doubly indirect method to obtain body fat 12 . Although the literature describes the gold standard technique 23 as a more accurate technique, the results obtained by measuring the TSF corroborate the other methods used to analyze body composition. In view of this, it would be interesting to perform a comparative analysis between the method used in this study with a gold standard technique, in an attempt to obtain consistent and homogeneous information among them, to elect a new method with potential of easy applicability and cost benefit.

CONCLUSION
In the present study, it was possible to observe that approximately one in four children from 7 to 10 years old are overweight, and this occurrence is higher among boys. The TSF method showed a strong correlation with both BMI and WC, with a satisfactory agreement between the methods. Thus, we can conclude that TSF can be a good method for investigating body composition and cardiometabolic risk for schoolchildren of both sexes.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. This study was funded by the authors. ments: RDL. Analyzed the data: RDL, RMP, VRM, RSC, PHC. Contributed reagents/materials/analysis tools: PHC. Wrote the paper: RDL, RMP, VRM, RSC, PHC. All authors read and approved the final version of the manuscript.