Parâmetros antropométricos e de composição corporal na predição do percentual de gordura e perfil lipídico em escolares

OBJETIVO: Avaliar a eficacia de indicadores antropometricos e da composicao corporal na predicao do percentual de gordura e perfil lipidico em escolares. METODOS: Estudo transversal envolvendo 209 escolares entre sete e nove anos. Foram consideradas as seguintes variaveis: peso, estatura, indice de massa corporal, percentual de gordura corporal, circunferencias do braco e cintura, indice de conicidade, razao cintura/estatura, razao cintura/quadril, colesterol total, triglicerideos e lipoproteinas de alta e baixa densidades. O tratamento estatistico incluiu a utilizacao dos testes de Kolmogorov-Smirnov, t de Student, Mann-Whitney e a correlacao de Pearson e Spearman. Para identificacao dos preditores do elevado percentual de gordura corporal e das alteracoes lipidicas, adotou-se a analise da Receiver Operating Characteristic Curve. RESULTADOS: O percentual de gordura corporal foi a variavel que apresentou maior numero de correlacoes, correlacionando-se fortemente com peso, indice de massa corporal e circunferencia do braco em ambos os generos, alem de apresentar fortes correlacoes com a circunferencia da cintura e a razao cintura/estatura nos meninos. O indice de massa corporal, as circunferencias do braco e da cintura em ambos os generos e a razao cintura/quadril para o sexo masculino apresentaram poder discriminatorio satisfatorio para predicao do percentual de gordura corporal elevado. Os parâmetros antropometricos e de composicao corporal nao foram capazes de predizer alteracoes no perfil lipidico, com excecao do percentual de gordura corporal, da circunferencia do braco e cintura e da razao cintura/quadril, que se mostraram bons preditores de alteracoes de triglicerideos no genero masculino. CONCLUSOES: A dislipidemia nao pode ser predita por medidas antropometricas e de composicao corporal na faixa etaria pediatrica, especialmente no genero feminino, sugerindo a necessidade da investigacao do perfil lipidico por meio de exames laboratoriais.


Introduction
As cardiovascular diseases have progressively advanced among children (1) , greater attention should be paid to the identification of risk factors in this age group.In the 5-to 9-year-old population, about half of the children were overweight according to the Brazilian Expenditure and Income Survey (Pesquisa de Orçamentos Familiares, POF, 2008-2009).Overweight was found in 33.5% of the children, and obesity already affected 14.3% (2) .
Among adults, the role of obesity and abdominal fat have been clearly established as factors that determine cardiovascular risk (3) .Among children, however, further studies should be conducted, because of the relevance of this issue and the fact that several authors have reported contradictory results (4)(5)(6)(7) .Dyslipidemias are the major risk factor for coronary heart disease, the main cause of death in Brazil (8) .Although signs and symptoms usually appear only in adulthood, their silent onset occurs in childhood, and the disease progresses as individuals grow older (9) .
Laboratory tests to assess lipid profile are not routinely requested for children, but easy-to-obtain data, such as anthropometric measurements, may prove to be useful as screening tools in clinical practice.This study evaluated the efficacy of anthropometric and body composition parameters in the prediction of body fat and lipid profile in schoolchildren.

Methods
This cross-sectional study enrolled children attending public and private schools in the urban area of the city of Diamantina (13 schools: 10 public and three private), in the Vale do Jequitinhonha, state of Minas Gerais, Brazil.A total of 209 children (126 girls) aged seven to nine years were included.This study was part of a larger investigation, Gestation and birth characteristics and feeding in the first year of life and their association with dyslipidemia and overweight in schoolchildren (10) , approved by the Committee on Ethics in Research with Human Beings of Universidade de Viçosa, under number 040/2009.This larger study was designed to investigate the association between breastfeeding and obesity and dyslipidemia in schoolchildren.
To calculate sample size, the maximum expected frequency of overweight in the age group under study was assumed to be 10%, based on municipal data and previous studies (11,12) , and the confidence interval was set at 95%.The minimum sample size was 150 children.
To select participants, all classrooms with students in the age group of interest in all schools were enrolled for random drawing.The number of students randomly drawn in each school was proportional to the total number of students enrolled in each school.Children included in the study lived in the urban area, did not have any chronic diseases nor made use of medications for chronic diseases, had not had diarrhea (defined as more than three episodes of watery stools per day on at least three days) one week before the interview and had no medical condition that might affect growth (diabetes, juvenile rheumatoid arthritis, inflammatory bowel disease, cerebral palsy, Down syndrome, leukemia, orthopedic conditions, congenital anomalies).
Data were collected from August 2008 to February 2009 in classes that meet in the morning and in the afternoon by a properly trained dietician, who was also responsible for the study.The anthropometric variables under study were: weight, height, triceps and subscapular skin folds, and waist, hip and arm circumferences.Blood samples were collected to measure total cholesterol (TC), triglycerides (TG) and low and high density lipoproteins (LDL and HDL).
Weight was measured using a 136-kg weighing scale to the nearest 0.1kg, and children were asked to take off shoes and accessories.At the same time, height was measured using a 200-cm stadiometer to the nearest 0.5cm while the child was standing without shoes.After weight and height were measured, body mass index (BMI) was calculated to classify children according to nutritional status using the 2007 World Health Organization (WHO) criteria (13) .The cut-off point for overweight was a Z score equal to or greater than +1.
Triceps and subscapular skin folds were measured in the right side of the body using a Lange skinfold caliper.Each measurement was made three times, and the final result was the mean of the two closest values.The triceps skinfold was measured in the posterior area of the right arm, over the triceps, at the midpoint between the acromion and olecranon, and the subscapular fold was measured immediately below and away from the angle of the right scapula (14) .The percentage of body fat in children was estimated using the equations described by Slaughter et al (15) , derived from a multicomponent model and adjusted for sex, maturation level and ethnicity.These equations used the values of the triceps and subscapular skinfolds.Excessive body fat was diagnosed according to the body fat thresholds defined by Lohman (16) : more than 20% for boys and 25% for girls.
Waist, hip and arm circumferences were measured using a flexible and inelastic measuring tape.Waist circumference was measured at the midpoint between the costal margin and the iliac crest and the hip, around the maximum circumference of the buttocks.The arm circumference was measured while the child was standing.First the right forearm was raised to form a 90 degree angle with the upper arm, and the measurement was made at the midpoint between the lateral region of the acromion and olecranon.Waist circumference (WC) was classified according to the cut-off points suggested by Taylor et al (17) , which define that WC is elevated when ≥62 and 62.9cm at 7 years, 64.7 and 65.3cm at 8 years and 67.3 and 67.7cm at 9 years for boys and girls (17) .
The waist-to-hip ratio (WHR) was calculated by dividing WC by hip circumference (HC).The waist-to-height ratio (WHeR) was calculated by dividing WC (cm) by height (cm).The conicity index (CI) was calculated using WC, weight and height, according to the equation below: Blood samples to assess lipid profiles were collected in the morning after 24-hour fasting by venous puncture and using disposable materials in a municipal reference laboratory.Plasma triglycerides (TG), total cholesterol (TC) and HDL were measured using colorimetric assays.LDL was calculated using the Friedwald equation (18) .Lipid profile was classified according to the First Brazilian Guidelines for the Prevention of Atherosclerosis in Childhood and Adolescence (1) .
The SPSS 11.0 was used for data analysis.The level of significance was set at 5% (p<0.05).The Kolmogorov-Smirnov test was used to define variable distribution; if distribution was normal, the Student t test and the Pearson correlation coefficient were used; otherwise, the Mann Whitney test and the Spearman correlation coefficient were calculated.Correlations were interpreted according to the classification described by Callegari-Jacques (19) .Receiver operating characteristics (ROC) curves were built, and the area under the curve was calculated, at a 95% confidence interval.
Children were only examined after the informed consent term was signed by their parents or guardians.
The characteristics of the schoolchildren included in the study are described in Table 1.Mean BMI and median WC, CI, WHeR and WHR were higher for boys, whereas girls had higher %BF.
The variable with the greatest number of correlations with the other anthropometric and body composition parameters was %BF, which was also strongly correlated with weight, BMI, AC for both boys and girls and with AC and WHeR for boys.TC was positively correlated only with %BF, whereas HDL was negatively correlated with WHR in both boys and girls.The correlation between lipid profile (TC, TG and HDL) and the anthropometric and body composition parameters was weak.The plasma levels of LDL in both boys and girls were not correlated with any of the variables under analysis (Table 2).
Figures 1 and 2 show the areas under the ROC curves for WHeR, WHR, BMI, CI, AC and WC as predictors of high %BF, TC, TG and LDL and low HDL in boys and girls.
BMI, AC and WC in both sexes and WHR in boys were good predictors of elevated %BF.In general, anthropometric and body composition parameters did not predict lipid profile alterations, except %BF, AC, WC and WHR, which were good predictors of alterations in triglycerides in boys (Table 3).overweight (20) .Our results suggest that interventions should be conducted, as overweight in childhood and adolescence is correlated with risk factors of diseases at this age (21,22) and later in life (23,24) .Moreover, when onset occurs in the beginning of life, the disease tends to be permanent or to worsen as age advances (25) .The prevalence of lipid profile changes was high among the children in this study, as well as in studies conducted in other Brazilian regions, such as in the cities of Florianópolis (4) and Belém (7) .
These results indicate a high risk of coronary heart disease in a group of still very young individuals.One of the possible explanations for the growing prevalence of dyslipidemia among children may be the changes in eating habits and in the level of physical activities that the Brazilian population has experienced in the last years (26) .Current life styles favor the consumption of industrialized foods, often greasier and nutritionally poorer, eating snacks and restaurant meals instead of eating at home, and consuming foods that have more calories and less fiber.In addition, modern life amenities, such as elevators and escalators, remote controls and cars, as well as the new forms of entertainment enjoyed by children, such as computers and videogames, promote a sedentary life style (6) .
Classically, dyslipidemia is one of the major risk factors for several cerebrovascular events, such as atherosclerosis, myocardial infarction and stroke.For that reason, the Brazilian Association of Cardiology (Sociedade Brasileira de Cardiologia -SBC) recommends that prevention and treatment should begin as early as in childhood (1) , when atherosclerotic events, although silent, begin to develop (9) .
The diagnosis of dyslipidemia in childhood has received special attention due to the importance of early detection and treatment of this disease.In routine pediatric practice, however, blood tests to assess lipid profiles are not requested, although recommended by the SBC whenever the child has any risk factor (1) .In our study, however, classical risk factors of dyslipidemia in adults, such as high percentage of body fat, overweight or obesity and elevated waist circumference, were not good predictors of dyslipidemia in children.That is, children that did not have risk factors for this disease may have serum lipid and lipoprotein changes and yet go undiagnosed.
In this study, the level of triglycerides was the lipid variable correlated with most anthropometric parameters.In agreement with the correlation analyses and as demonstrated by the ROC curves, only serum levels of triglycerides were predicted by anthropometric parameters for boys, as the percentage of body fat, the waist and arm circumferences and the waist-to-hip ratio had the largest areas under the ROC curve for triglycerides.
Therefore, in addition to evaluating fat mass, fat distribution should also be analyzed.Abdominal fat, in addition to being an important predictor of an increase in the level of triglycerides, is associated with a decrease in HDL levels and an increase in left ventricular mass in children and adolescents (27) .
The cut-off point of 18.99% for body fat suggested in this study for predicting hypertriglyceridemia in boys is below the reference value, that is, even boys with a percentage of body fat classified as adequate by Lohman (16) may already have alterations in their levels of serum triglycerides.The cut-off point for waist circumference seems to be very close to that suggested for 7-year-old children.However, waist circumference increases gradually with age, and it is clearly necessary to adjust the cut-off to age for a better prediction of triglyceride levels in boys.
The anthropometric parameters used in this study were also evaluated as predictors of level of serum lipids by other authors, who found similar results (28) .Among older children and adolescents, the associations between anthropometric parameters and lipid profile are more evident (5,7,29) , although some results are contradictory (6) .
Because of the lack of agreement between results, other investigations should evaluate the efficacy of anthropometric parameters in predicting lipid profiles in children, so that the diagnosis of this disorder may be established as early as possible, as about 50% of the children with elevated TC and LDL levels remain like that for 10 to 15 years (tracking) (30,31) .
Nevertheless, this study demonstrated that, in the age group under study, anthropometric and body composition parameters did not predict dyslipidemia, particularly not among girls, except for hypertriglyceridemia alone, which was accurately predicted by percentage of body fat in both sexes and waist circumference in boys.The assessment of lipid profiles using laboratory tests should be routine for this age group because of the high prevalence of serum lipid alterations found in our study and their serious consequences to public health.

Figure 1 -Figure 2 -
Figure 1 -ROC curve for anthropometric and body composition parameters in the evaluation of percentage of body fat and lipid profiles in female schoolchildren.Diamantina, Brazil, 2009

Table 2 -
Correlations between anthropometric and body composition parameters with lipid profile variables among schoolchildren.