Metabolic syndrome (MetS) can be considered as a clustering of cardiovascular risk factors and type 2 diabetes. In adults, diagnosis of MetS requires the presence of at least three of the following factors: central obesity, dyslipidaemia (low HDL-cholesterol (HDL-C) and hypertriacylglycerolaemia), hypertension and hyperglycaemia. However, there is no consensus regarding the diagnosis of MetS in children(Reference Lee, Bacha and Gungor1). Some of the criteria used have been based on the recommendations of the National Cholesterol Education Program – Adult Treatment Panel III, with appropriate adjustments and specific cut-off points for age and gender(Reference de Ferranti, Gauvreau and Ludwig2–Reference Cook, Weitzman and Auinger4).
For the diagnosis of metabolic risk and the components of MetS, biochemical evaluation has been more commonly used in clinical practice; however, this is invasive, requires trained personnel and is costly. In contrast, anthropometry has been reported as an important tool for both clinical practice and population studies, including the identification and prediction of metabolic risk(Reference Freedman, Katzmarzyk and Dietz5–Reference Agirbasli, Agaoglu and Ergonul9).
In the context of epidemiological studies, several have reported gender differences for percentage body fat(Reference Raman, Fitch and Hudes10), waist circumference (WC)(Reference Moreno, Fleta and Mur11), insulin resistance (by homeostatic model assessment, HOMA-IR)(Reference Raman, Fitch and Hudes10) and MetS(Reference Friend, Craig and Turner12) in children. Moreover, gender differences have also been reported following an intervention programme designed to control insulin resistance(Reference Raman, Ritchie and Lustig13) and after a follow-up of 6·6 years for trends of MetS in children(Reference Barzin, Hosseinpanah and Saber14). The biochemical and anthropometric changes and differential response to intervention programmes between boys and girls suggest the need for separate analyses.
It is widely acknowledged that lifestyle factors (hours of screen viewing, level of physical activity) influence predictors of the MetS(Reference Hong, Kim and Kang15, Reference Casazza, Dulin-Keita and Gower16). For example, regular physical activity plays a fundamental role in the prevention and management of obesity, hypertension, dyslipidaemia and insulin resistance(Reference Steinberger, Daniels and Eckel17); however, data regarding the relationship between physical activity level (PAL) and the prediction of MetS in children remain scarce. The present study evaluated the effectiveness of anthropometry and PAL in the prediction of MetS in children.
Materials and methods
Study design and participants
The present cross-sectional study was conducted in a sample comprising 187 children (106 girls, 56·7 %) of mean age 9·90 (sd 0·7) years. To calculate the minimum sample size we employed the equation proposed by Lwanga and Lemeshow(Reference Lwanga and Lemeshow18). Considering the total number of 1049 students in the age group studied, the population total of 72 220, a standard error of 5 % with a 95 % confidence interval, eight schools (25 % of the thirty-two schools with 5th graders) were randomly selected. All 5th-grade students in each school were invited to participate in the study, and 74·8 % consented with authorization from their parents.
The sample comprised schoolchildren from the urban area of Viçosa, a university town in south-eastern Brazil with a population of ∼72 220 inhabitants and an area of 299 km2. The Human Development Index is 0·809, and the index of development of basic education is 5·7 (a score of 6·0 corresponds to a quality of education system comparable to that in developed countries). The service sector is the main component of the local economy and the Gross Domestic Product is $US 4320 per capita.
Procedures
Children underwent a comprehensive anthropometric protocol and blood pressure evaluation after the study was approved by the Ethics in Human Research Committee of the Federal University of Viçosa and parental consent was given. Children were provided instructions regarding the completion of a physical activity recall questionnaire and survey questions regarding socio-economic status and lifestyle were completed by parents. Participants also received instructions regarding the necessity to fast for 12 h in preparation for the collection of blood. All children received instruction on wearing a pedometer prior to wearing it for seven consecutive days.
Anthropometry
Body weight was assessed with an electronic digital scale (Lumina-02550; Plenna, São Paulo, Brazil) with a precision of 100 g and height was measured using a portable 2-m anthropometer (Rigor e Técnica, Viçosa, Brazil) with gradations of 0·1 cm. Children were measured wearing light clothes and without shoes following standardized procedures(19). BMI was calculated as weight (in kilograms) divided by the square of height (in metres) and children were classified as overweight or obese, according to criteria established by Cole et al.(Reference Cole, Bellizzi and Flegal20).
WC was measured after normal expiration using a flexible and non-elastic 2-m tape (Sanny, São Paulo, Brazil). All measurements were carried out in triplicate and the average recorded. Measurements were taken at the following anatomic landmarks: (i) at the narrowest point of the waist between the iliac crest and the ribs (WC1) and (ii) at midpoint between the iliac crest and the lower rib (WC2), as recommended by the WHO(Reference Hong, Kim and Kang15); and (iii) at the level of the umbilicus (WC3), as recommended by the Anthropometric Standardization Reference Manual (Reference Lohman, Roche and Martorell21). The conicity index was calculated from measurements of WC, weight and height, according to a formula proposed by Valdez(Reference Valdez22). The purpose of this index was to assess body fat distribution for which cut-off points are not available in children.
Triceps (TST), subscapular (SST), suprailiac (SuST) and biceps (BST) skinfold thicknesses were measured according to standardized techniques using a Lange skinfold calliper (Cambridge Scientific Instruments, Cambridge, MA, USA) which exerts a constant pressure of 10 g/m2. All measurements were taken on the right side of the body with three non-consecutive replicates of each measurement(Reference Guedes and Guedes23) which were subsequently averaged. The sum of two skinfolds (TST + SST; Σ2ST) and the sum of four skinfolds (TST + SST + SuST + BST; Σ4ST) were also calculated.
The equations proposed by Slaughter et al.(Reference Slaughter, Lohman and Boileau24) were used to calculate percentage body fat (%BF) and children were classified according to the categories of adiposity proposed by Lohman(Reference Lohman25).
Blood pressure
Blood pressure was measured using a sphygmomanometer according to standardized approaches(26). Only children whose systolic or diastolic blood pressure values exceeded the 90th percentile (according to height, gender and age) in three different measurements were diagnosed with high blood pressure.
Socio-economic status and lifestyle survey
Parents received a survey developed specifically for the present study, with eighteen closed questions regarding socio-economic status, lifestyle factors and sedentary behaviour.
Physical activity level
A 3 d physical activity diary (two weekdays and one weekend day) was completed(Reference Bouchard, Tremblay and Leblanc27). Activities were classified according to metabolic equivalents of task (MET) into light (≤3·9 MET), moderate (>4 and ≤6·9 MET), vigorous (≥7 MET) and moderate-to-vigorous (>4 MET) intensity. Children who accumulated at least 60 min of moderate-to vigorous-intensity physical activity (MVPA) per day achieved the recommendation for a healthy lifestyle(28).
All children wore a pedometer (DX 8897; Pulse Rate Pedometer, Hong Kong, China) and the average number of steps per day was derived across seven consecutive days.
Blood collection and analyses
After 12 h of fasting, blood samples were collected and immediately processed. Serum was analysed using the enzymatic colorimetric method to determine HDL-C, TAG and glucose on an automated haematology analyser (Cobas Mira Plus; Roche, Rotkreuz, Switzerland).
Metabolic syndrome
In the current study, MetS was diagnosed according to the presence of three or more of the following criteria: (i) TAG ≥100 mg/dl; (ii) HDL-C <50 mg/dl; (iii) glucose ≥110 mg/dl; (iv) WC ≥75th percentile for age and sex; and (v) blood pressure (diastolic or systolic) >90th percentile adjusted for age, height and sex. The cut-off points adapted to the age of the population studied were chosen according the criteria of de Ferranti et al.(Reference de Ferranti, Gauvreau and Ludwig2).
Statistical analysis
Analyses were performed using the statistical software package SPSS version 17·0 for Windows. The continuous variables were described by means of measurements with central and dispersion tendency, including mean, standard deviation, median, and minimum and maximum values. The categorical variables were described in percentage values.
All variables were tested for normality by the Kolmogorov–Smirnov test. Student's t test was used to compare the means of independent variables with parametric distribution and the Mann–Whitney test was used for the variables having non-parametric distribution. P < 0·05 was considered for statistical significance. The Kruskall–Wallis test with Dunn's post hoc test was selected for comparison among the groups on non-parametric distribution.
Measures of association between MetS, lifestyle and biochemical variables were obtained by the χ 2 test and odds ratios with their 95 % confidence intervals are presented.
Cut-off points for the variables analysed were suggested through the use of receiver-operating characteristic (ROC) curves. The optimal cut-offs were determined by finding the values that allowed the best balance between sensitivity and specificity. The statistical significance of each analysis was assessed by the area under the ROC curve (AUC) and by the inferior limit of the 95 % confidence interval being higher than 0·5(Reference Schisterman, Faraggi and Reiser29).
Results
Anthropometry
Table 1 shows the anthropometric characteristics of the children. Significant differences (P < 0·05) were observed between boys and girls for all four skinfolds, the sums of skinfolds (Σ2ST, Σ4ST) and %BF.
Table 1 Anthropometric, biochemical and blood pressure parameters for 5th-grade girls and boys (n 187; mean age 9·90 (sd 0·7) years), Viçosa, south-eastern Brazil
WC1, waist circumference taken at the narrowest point of the waist, between the iliac crest and the ribs; WC2, waist circumference taken at the midpoint between the iliac crest and the lower rib; WC3, waist circumference taken at the level of the umbilicus; TST, triceps skinfold thickness; BST, biceps skinfold thickness; SuST, suprailiac skinfold thickness; SST, subscapular skinfold thickness; Σ2ST, sum of two skinfolds (TST + SST); Σ4ST, sum of four skinfolds (TST + SST + SuST + BST); %BF, percentage body fat; HDL-C, HDL-cholesterol; SBP, systolic blood pressure; DBP, diastolic blood pressure.
*Significant differences between genders, Mann–Whitney test (P < 0·05).
Physical activity level
The number of steps per day did not differ significantly (P > 0·05) between genders, with boys recording a median of 5952 steps/d (range 807–31 459 steps/d) and girls a median of 4906 steps/d (range 387–20 803 steps/d). On the other hand, time spent in MVPA differed significantly (P < 0·05) between genders: median (range) of 105 min/d (0–390 min/d) and 15 min/d (0–360 min/d) for boys and girls, respectively.
Only 13·6 % of girls and 14·5 % of boys achieved the American recommendation of 11 000 and 13 000 steps/d, respectively(Reference Tudor-Locke and Bassett30), and 36·7 % of girls and 68·1 % of boys met the recommended 60 min of daily MVPA(28).
Biochemical variables and blood pressure
None of the participants had impaired fasting glucose, 24·1 % of the children had HDL-C levels below the recommended level, 10·6 % had hypertriacylglycerolaemia and 14·4 % recorded high blood pressure. For all biochemical variables and blood pressure, no significant between-gender differences were found (P > 0·05; see Table 1).
Metabolic syndrome
MetS was found in 8·5 % of the children. In respect of the five components of MetS (TAG, HDL-C, glucose, WC and blood pressure), 53·2 % of children showed normal values in all measures while 26·2 % had at least one component, 12·1 % had two components, 6·4 % had three, and 2·1 % had four or more components of MetS.
The prevalence of overweight and obesity was 15·0 % and 5·9 %, respectively, with boys showing a higher obesity prevalence (8·6 %) than girls (3·8 %).
For girls, hypertriacylglycerolaemia and high %BF were significantly associated with MetS (OR = 31·3; 95 % CI 5·43, 179) and (OR = 6·23; 95 % CI 1·03, 48·3), respectively. For boys, significant associations were found for MetS and hypertriacylglycerolaemia (OR = 47·0; 95 % CI 3·68, 599) and MetS and low HDL-C (OR = 34·5; 95 % CI 2·80, 413). There was no significant association between screen time (hours in front of a television and computer) and MetS.
Predictors of the metabolic syndrome
Anthropometric indices presented in Table 1 were tested as predictors of MetS in both genders. For boys, the AUC was significant for conicity index (AUC = 0·737; 95 % CI 0·520, 0·955) and Σ4ST (AUC = 0·897; 95 % CI 0·785, 0·963). Weight, height, BMI, WC, the four skinfolds assessed separately, Σ2ST and %BF were ruled out as predictors of MetS in boys. Cut-off points for anthropometric measures were suggested for boys that showed the best balance between sensitivity and specificity (Table 2).
Table 2 Cut-off points for predictors of MetS in 5th-grade boys (n 81), Viçosa, south-eastern Brazil
MetS, metabolic syndrome; Σ4ST, sum of four skinfolds (triceps + subscapular + suprailiac + biceps); SENS, sensitivity; SPEC, specificity; PPV, positive predictive value; NPV, negative predictive value.
For girls, anthropometric measures including BMI, WC1, WC3, four skinfolds assessed separately, Σ2ST, Σ4ST (higher AUC = 0·908) and %BF (Table 3) were useful for the prediction of MetS, and cut-off points were proposed for these measures (Table 4).
Table 3 Area under the ROC curve of anthropometric predictors of MetS in 5th-grade girls (n 106), Viçosa, south-eastern Brazil
ROC, receiver-operating characteristic; MetS, metabolic syndrome; AUC, area under the curve; WC1, waist circumference taken at the narrowest point of the waist, between the iliac crest and the ribs; WC3, waist circumference taken at the level of the umbilicus; TST, triceps skinfold thickness; BST, biceps skinfold thickness; SuST, suprailiac skinfold thickness; SST, subscapular skinfold thickness; Σ2ST, sum of two skinfolds (TST + SST); Σ4ST, sum of four skinfolds (TST + SST + SuST + BST); %BF, percentage body fat.
*Significant AUC, inferior limit to the 95 % CI greater than 0·5.
Table 4 Cut-off points for anthropometric predictors of MetS in 5th-grade girls (n 106), Viçosa, south-eastern Brazil
MetS, metabolic syndrome; WC1, waist circumference taken at the narrowest point of the waist, between the iliac crest and the ribs; WC3, waist circumference taken at the level of the umbilicus; TST, triceps skinfold thickness; BST, biceps skinfold thickness; SuST, suprailiac skinfold thickness; SST, subscapular skinfold thickness; Σ2ST, sum of two skinfolds (TST + SST); Σ4ST, sum of four skinfolds (TST + SST + SuST + BST); %BF, percentage body fat; SENS, sensitivity; SPEC, specificity; PPV, positive predictive value; NPV, negative predictive value.
The number of steps per day was an accurate predictor of MetS, acting as a protective factor in boys (AUC = 0·891; 95 % CI 0·736, 0·971). Cut-off points with the best balance between sensitivity and specificity for protection against MetS are shown in Table 2. However, the number of steps per day by girls and the time spent in MVPA for both genders were not accurate in predicting MetS.
Discussion
The present study proposes a simple, easily applied, reproducible and low-cost alternative for the prediction of MetS in children. Currently, there is no consensus regarding the most appropriate anthropometric, haemodynamic and biochemical parameters to diagnose MetS in children. Accordingly, the present study sought to elucidate the accuracy of anthropometric parameters to diagnose the MetS.
Σ4ST was the main anthropometric predictor of MetS in boys and girls (AUC = 0·897 and AUC = 0·908, respectively) and showed cut-off point values with sensitivity and specificity above 80 %. It is worth noting that the closer to 1·0 is the AUC, the more accurate and the higher the power of a diagnostic test.
Σ2ST was accurate in only girls (AUC = 0·733). Freedman et al.(Reference Freedman, Katzmarzyk and Dietz5) reported that Σ2ST had lower accuracy than BMI in predicting cardiovascular risk factor components of MetS, in agreement with data from the present study (Table 4).
Several studies have emphasized the relationship between TST and %BF(Reference Freedman, Wang and Ogden31–Reference Pecoraro, Guida and Caroli35); however, studies investigating the power of TST to predict MetS remain scarce. Our results demonstrated that all individual skinfold thickness measures predict MetS in girls. Among these, TST is more accurate (AUC = 0·737; 95 % CI 0·633, 0·826; sensitivity = 100 %, specificity = 53·75 %) with a cut-off point of 14·8 mm. Sardinha et al.(Reference Sardinha, Going and Teixeira32) reported that TST is also efficient in predicting %BF in 10–11-year-old girls (against the gold standard of dual energy X-ray absorptiometry) and obtained an AUC of 0·96 (95 % CI 0·87, 0·99; sensitivity = 79 %, specificity = 100 %) with a cut-off point of 21 mm. Moreover, this cut-off point predicts increases in total cholesterol and LDL-cholesterol (sensitivity = 83·3 %, specificity = 78·7 %) for 11-year-old Brazilian boys(Reference Lunardi and Petroski36). Given this evidence, measurement of TST is an important indicator of cardiovascular risk factors and should be included in both scientific research and the clinical monitoring of children.
In the present study, the cut-off point suggested for %BF in the prediction of MetS in girls was 25·7 % (sensitivity = 75 %, specificity = 71·25 %), a value similar to the one proposed by Lohman(Reference Freedman, Wang and Ogden31) to classify girls with moderately high %BF (%BF ≥25 %). Thus, %BF >25 % for girls seems to be confirmed as a key value in increasing the risk for developing MetS.
To our knowledge, the present study is the first to suggest a conicity index cut-off point as a predictor of MetS in Brazilian children. The conicity index was accurate in the prediction of MetS for boys only, with a cut-off point of 1·16 (AUC = 0·737; 95 % CI 0·520, 0·955). The conicity index predominantly assesses central fat that is highly correlated with components of the MetS(Reference Lee, Ma and Heng37). However, as the conicity index was created for adults and with sex-specific cut-off points, its use in children has remained somewhat controversial. Taylor et al.(Reference Taylor, Jones and Williams38) tested the accuracy of three anthropometric measures in predicting the percentage of central fat in children and adolescents (3–19 years old). The conicity index had lower accuracy (AUC = 0·81; 95 % CI 0·74, 0·88) compared with WC (AUC = 0·97; 95 % CI 0·95, 0·99) and greater accuracy in relation to the waist-to-hip ratio (AUC = 0·71; 95 % CI 0·62, 0·80). The median value for the conicity index was 1·16 and did not show the value of the cut-off point(Reference Taylor, Jones and Williams38). Alternatively, the conicity index is a simple and, in the absence of sophisticated equipment, can be used as a measure of body fat distribution in children(Reference Perez, Landaeta-Jimenez and Vasquez39).
The BMI is widely reported as a diagnostic tool for obesity. As an indicator of MetS in children, numerous cut-off points have been suggested for BMI: >97th percentile for age and sex(Reference Cali and Caprio40–Reference Calcaterra, De Amici and Klersy42); ≥85th percentile(Reference Ice, Murphy and Minor43); and the cut-off point proposed by Cole et al.(Reference Cole, Bellizzi and Flegal20) to define overweight and obesity(Reference Agirbasli, Agaoglu and Orak44). We suggest, for girls, a BMI cut-off point of 19·2 kg/m2 (sensitivity = 50 %, specificity = 78·25 %), approaching the values of Cole et al. (BMI = 19·86 kg/m2) and Conde and Monteiro (BMI = 18·63 kg/m2)(Reference Conde and Monteiro45), a national reference for the diagnosis of overweight in 10-year-old girls. In a case–control study(Reference Wee, Poh and Bulgiba46), only 12 % of the overweight/obese group were free of MetS risk factors in contrast to 83·9 % of normal-weight children. These findings underline the importance of BMI as a predictor of MetS risk.
As an indicator of MetS, WC is assessed with the largest number of methodological variations. Different points of measurement were detected and cut-offs points have varied: ≥85th percentile of the population studied(Reference Stewart, Christian and Schulze47) and ≥90th percentile for age and sex(Reference Calcaterra, De Amici and Klersy42, Reference Ice, Murphy and Minor43). In the present study, two measurement sites were accurate in predicting MetS in girls: WC1 and WC3, with cut-off points of 56·0 and 63·8 cm, respectively, in the prediction of MetS. WC3 was more accurate (AUC = 0·709) with a better balance between sensitivity and specificity compared with WC1 (AUC = 0·683).
Hirschler et al.(Reference Hirschler, Aranda and Calcagno Mde6) reported that WC3 correlated significantly with all components of MetS in children aged 6–13 years. Furthermore, they highlighted that children with abdominal obesity (>90th percentile of the sample) had increased risk for CVD and type 2 diabetes. In the current study, we observed that 10·1 % of children were >90th percentile for abdominal obesity (WC3 > 78·68 cm). On the other hand, when using the cut-off point suggested (WC3 > 63·8 cm) for the prediction of MetS, an estimated 38·7 % of children were at risk of MetS.
The current study is the first to suggest a cut-off point for the number of steps per day as a predictor of MetS in Brazilian boys. Use of this indicator suggested that 40 % of boys studied were protected from MetS based on attainment of the recommended cut-off point of 7872 steps/d (AUC = 0·891; 95 % CI 0·736, 0·971; sensitivity = 100 %; specificity = 78·12 %). The ROC curve analysis demonstrated better sensitivity and specificity with much lower values of steps/d compared with the American recommendation of 13 000 steps/d to achieve a minimum standard of health(Reference Tudor-Locke and Bassett30). All children classified as having MetS achieved less than 7872 steps/d, indicating that this number of steps may be a critical marker of poor metabolic health.
There is an urgent need to establish criteria to facilitate the diagnosis of MetS in children and recommended cut-off points should be similar to those of adults to facilitate comparisons(Reference Cruz and Goran3). Current recommendations and cut-off points need to be developed and reviewed, particularly for WC, dyslipidaemia and hyperglycaemia.
Our data provide evidence for the incorporation of a comprehensive tool to be used by health workers and educators involved in health programmes for schoolchildren. As Brazil has a government-endorsed School Health Program target to evaluate students, the findings of the present study provide justification for the addition of a physical assessment to assist in the prediction of schoolchildren at risk of MetS.
The power of the results related to the anthropometric measures presented herein support their use in screening children at risk of poor metabolic health. Most importantly, a large number of children could be assessed inexpensively, those at risk diagnosed early, and referred to the appropriate professionals for assistance.
Conclusion
We conclude that, for both genders, Σ4ST was the most accurate method to predict MetS in children. Moreover, BMI and WC, already widely used in the diagnosis of MetS, were accurate in the prediction of MetS only in girls. PAL and the number of steps per day, however, were effective in the prediction of MetS in boys. Thus, monitoring of anthropometric measures and PAL is essential in the control and prevention of MetS in children.
Acknowledgements
Sources of funding: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Conflicts of interest: All authors declare that there are no conflicts of interest. Ethical approval: The protocol for the research project was approved by the Federal University of Viçosa Human Research Ethics Committee and conformed to the provisions of the Declaration of Helsinki in 1995 (as revised in Edinburgh 2000). All participants gave informed consent and participant anonymity has been preserved. Authors’ contributions: A.C.R.A. conceived the manuscript, performed the statistical analyses, interpreted the data and drafted the manuscript. A.L.A.T. partly contributed to the conception and design of the current study, provided statistical expertise and critical revision. E.L.M. partly contributed to the conception and design of the current study, provided statistical expertise and critical revision. R.A.J. performed the statistical analyses, interpreted the data and drafted the manuscript. A.P.H. revised English, contributed to the analyses and interpretation of the data and provided critical revision of the manuscript. P.R.S.A. revised English and partly contributed to the conception and design of the current study, contributed to the analyses and interpretation of the data and provided critical revision of the manuscript. All authors provided critical feedback on drafts and read and approved the final manuscript. Acknowledgements: The authors wish to thank the children for their participation in the study, as well as their parents for their cooperation.
