SEARCH

SEARCH BY CITATION

Keywords:

  • cardiovascular risk factors;
  • anthropometric indices;
  • receiver operating characteristics analysis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Objective: To derive the optimal BMI and waist circumference (WC) cut-off values to predict clustering of cardiovascular risk factors in Hong Kong Chinese adolescents.

Research Methods and Procedures: A total of 2102 Hong Kong Chinese 12 to 19 years of age were recruited. Participants were considered to have clustering of risk factors if at least three of the following risk factors were present: 1) high-density lipoprotein cholesterol (HDL-C) ≤1.03 mM, 2) low-density lipoprotein cholesterol (LDL-C) ≥2.6 mM, 3) triglyceride (TG) ≥1.24 mM, 4) fasting plasma glucose (FPG) ≥6.1 mM, and 5) age-, sex-, and height-adjusted systolic or diastolic blood pressure (BP) ≥ 90th percentile. Receiver operating characteristics (ROC) curves were generated to identify the optimal age-adjusted BMI and WC cut-off values to predict clustering of risk factors in boys and girls separately. These age-adjusted BMI and WC cut-offs were transformed to percentile values. Cole's lambda-mu-sigma (LMS) method was used to obtain smoothed age-specific BMI and WC at these percentile values.

Results: The areas under ROC curves for BMI in girls and boys were 0.85 [95% confidence interval (CI), 0.77 to 0.92] and 0.76 (95% CI, 0.66 to 0.85), respectively. The respective areas under ROC curves for WC in girls and boys were 0.82 (95% CI, 0.74 to 0.91) and 0.78 (95% CI, 0.68 to 0.87). The optimal BMI thresholds were at the 78th percentile for girls and the 72nd percentile for boys. The respective values for WC were at the 77th percentile for girls and the 76th percentile for boys. The sensitivities and specificities of these cut-off values ranged from 72% to 80%.

Discussion: Age- and sex-specific BMI and WC cut-off values can be used to identify adolescents with clustering of cardiovascular risk factors.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Obesity is a global health problem (1, 2). It is closely associated with risk factors including hypertension, dyslipidemia, diabetes mellitus, and insulin resistance, which independently and collectively increase the risk of cardiovascular morbidity and mortality (3, 4). Apart from adult obesity, the prevalence of childhood and adolescent obesity has also increased worldwide in recent years (5).

BMI and waist circumference (WC)1 are simple clinical measures for obesity in adult populations. In white adults, BMI cut-off values ≥25 and ≥30 kg/m2 are used to define overweight and obesity, respectively, whereas WC ≥102 cm in men and ≥88 cm in women are used to define central obesity (1, 3). There is substantial tracking of body weight and obesity from adolescence into adulthood (6, 7, 8, 9, 10), which may increase future cardiovascular risk. However, it remains uncertain whether the concept of using BMI and WC cut-off values to identify high-risk adolescents is valid or useful.

In this cross-sectional study involving more than 2000 adolescents, we examined the associations between anthropometric indices and conventional cardiovascular risk factors using factor analysis. Then we used receiver operating characteristics (ROC) analysis to determine the optimal BMI and WC cut-off values for predicting clustering of three or more cardiovascular risk factors.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Study Population

This study was conducted from February 2003 to December 2003. A full list of all secondary schools in Hong Kong was obtained from the Education Department of Hong Kong. From a total of 477 secondary schools, 53 were randomly selected using a computer-generated coding system. Of the selected 53 schools, 14 schools participated in the study with approval from their principals. Six classes, one from each form (Form 1 to Form 6), were randomly selected to obtain a proportional number of subjects between 12 and 20 years of age. Only adolescents of Chinese ethnicity were recruited for assessment. This study was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong. Informed consents were obtained from all participants together with their parents’ informed and written consents before the study started.

Clinical Measurements and Blood Sample Collections

Anthropometric indices were measured, and blood samples were obtained from each participant during the field study by the same team of trained nurses and research assistants and using the same equipment. Body weight (measured to the nearest 0.1 kg by a Tanita physician digital scale; Tanita Corp., Tokyo, Japan) and body height (measured to the nearest 0.1 cm using a portable stadiometer) were measured with the adolescents wearing light clothing and without shoes. Waist circumference was taken midway between the lowest rib and the superior border of the iliac crest in the midaxillary line, and the measurements were taken to the nearest 0.1 cm. Blood pressure (BP) was taken from the nondominant arm after at least 5 minutes of rest using the Omron blood pressure device (Omron Healthcare Inc., Tokyo, Japan). The average of two readings was used for the analysis.

After an overnight fast of at least 8 hours, blood samples were collected for measurement of fasting plasma glucose (FPG), total cholesterol (TC), triglyceride (TG), and high-density lipoprotein cholesterol (HDL-C) levels.

Laboratory Analysis

All blood samples were kept in ice at 0 °C before they were transported back to our laboratory for further processing. All assays were performed within 6 hours after collection, and additional aliquots of serum were stored at −70 °C. Plasma glucose was measured by the hexokinase method (Hitachi 911 automated analyzer; Boehringer Mannheim, Mannheim, Germany). Both the intraassay and interassay coefficients of variation for glucose were 2% at 6.6 mM. TC (enzymatic method), TG (enzymatic method without glycerol blanking), and HDL-C (dextran sulfate-MgCl2 precipitation) were measured on a Hitachi 911 automated analyzer (Boehringer Mannheim) using reagent kits supplied by the manufacturer of the analyzer. Low-density lipoprotein cholesterol (LDL-C) level was calculated using the Friedewald's formula for TG <4.5 mM (11).

Statistical Analysis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

Factor Analysis

Exploratory factor analysis was used to uncover the main components among cardiovascular risk factors, including body weight, WC, BMI, systolic and diastolic BP, TG, HDL-C, LDL-C, and FPG. Fasting plasma TG was log-transformed to correct for its skewness. Age-adjusted variables were used for factor analysis. Each of these variables was regressed with up to a full cubic polynomial with respect to age (age, age2, and age3) for boys and girls separately using stepwise regression procedures. The standardized residuals were retained to represent the age-adjusted value. Factor extraction was conducted using principal component method. Those components with eigenvalues ≥1 were retained in the analysis. The principal components (factors) retained were rotated using orthogonal varimax method to facilitate their interpretation. Variables that had factor loadings >0.4 (or less than −0.4) on a factor extracted were considered as major constituents of this factor.

ROC Analysis

Because there were no standardized international or local guidelines to define cardiovascular risk factors clustering in adolescents, we adapted the modified definitions of various cardiovascular risk factors that had been used in the recent National Health and Nutrition Education survey (NHANES) (12). A participant having any of the following was considered to have a cardiovascular risk factor: 1) HDL-C ≤1.03 mM, 2) LDL-C ≥2.6 mM, 3) TG ≥1.24 mM, 4) FPG ≥6.1 mM, and 5) age-, sex-, and height-adjusted systolic BP or diastolic BP ≥90th percentile (13). They were considered to have clustering of risk factors if three or more risk factors were present. The ROC curves were generated to identify the optimal age-adjusted BMI and WC cut-off values for predicting clustering of cardiovascular risk factors. These optimal thresholds corresponded to the point closest to the top left-handed corner in each of the corresponding ROC curves with optimal sensitivity and specificity, both of which were given equal weighting. The lambda-mu-sigma (LMS) method first described by Cole and Green (14) was used to generate the smoothed sex- and age-specific BMI and WC cut-off values corresponding to the percentile values represented by the cut-off values chosen by the ROC analysis. The area under ROC curves (AUC) was used to give a measure of the global performance of using BMI and WC as indicators for diagnosis of cardiovascular risk factors clustering.

For comparison purposes, another set of ROC analysis based on the National Cholesterol Education Program-Adult Treatment Panel III (NCEP-ATP III) definition (except for the definition of hypertension) was performed. These risk factors include 1) HDL-C ≤1.03 mM for boys and ≤1.29 mM for girls, 2) LDL-C ≥2.6 mM, 3) TG ≥1.69 mM, 4) FPG ≥6.1 mM, and 5) age-, sex-, and height-adjusted systolic BP or diastolic BP ≥90th percentile (3, 13). Participants were considered to have clustering of risk factors if three or more were present.

Statistical analyses were performed with SPSS 13.0 (SPSS, Chicago, IL) and LMS program Version 1.16 (Tim Cole and Huiqi Pan, Institute of Child Health, London, UK). All statistical tests were two-sided, and p < 0.05 was considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

A total of 2115 students randomly selected from the 14 recruited schools consented and were enrolled into the study. Thirteen participants, 20 years of age, were excluded from analysis because of the small number in this age group. The remaining 2102 adolescents, 12 to 19 years of age, were used in these analyses.

There were 1144 girls (54.4%) and 958 boys (45.6%). Table 1 showed the characteristics and risk factors in boys and girls, respectively. High LDL-C and high BP were the two most common risk factors. From the nine selected variables, four factors were identified using factor analysis that accounted for 76.3% and 75.9% of the total variance for boys and girls, respectively (Table 2). Except for the factors reflecting body size (BMI, WC, and body weight), there were minor differences in other factors between boys and girls. In boys, there were three additional factors representing BP, dyslipidemia, and glycemia. In girls, the BP factor included FPG, whereas LDL-C was selected as a separate factor, with TG and HDL-C loaded on another factor.

Table 1.  Demographic characteristics and pattern of cardiovascular risk factors in 2102 Hong Kong Chinese adolescents 12 to 19 years of age
 Female (n = 1144)Male (n = 958)
  1. SBP/DBP, systolic/diastolic blood pressure.

  2. Mean ± standard deviation or frequencies (%).

Characteristics  
 Age (years)15.8 ± 2.015.4 ± 2.0
 Body weight (kg)49.0 ± 8.856.3 ± 13.1
 Waist circumference (cm)65.8 ± 6.971.3 ± 8.6
 Body mass index (kg/m2)19.7 ± 3.220.3 ± 3.7
 Systolic BP (mm Hg)113.5 ± 11.5120.9 ± 13.0
 Diastolic BP (mm Hg)73.1 ± 9.071.8 ± 9.7
 Triglyceride (mM)0.82 ± 0.380.83 ± 0.41
 HDL cholesterol (mM)1.64 ± 0.321.55 ± 0.30
 LDL cholesterol (mM)2.24 ± 0.612.15 ± 0.62
 Fasting plasma glucose (mM)4.67 ± 0.354.75 ± 0.35
Risk factors  
 High BP (≥90th percentile in either SBP or DBP adjusted for age, sex, and height)340 (29.7%)340 (35.5%)
 High triglyceride level (≥1.24 mM)112 (9.8%)117 (12.2%)
 Low HDL cholesterol (≤1.03 mM)25 (2.2%)25 (2.6%)
 High LDL cholesterol (≥2.6 mM)290 (25.3%)215 (22.4%)
 High fasting plasma glucose (≥6.1 mM)2 (0.2%)4 (0.4%)
 Clustering of three or more risk factors25 (2.2%)33 (3.4%)
Pattern of risk factors clustering  
 None561 (49.0%)445 (46.5%)
 One risk factor only425 (37.2%)360 (37.6%)
 Two risk factors133 (11.6%)120 (12.5%)
 Three risk factors22 (1.9%)31 (3.2%)
 Four risk factors3 (0.3%)2 (0.2%)
 Five risk factors00
Table 2.  Factor loadings for age-adjusted cardiovascular risk variables after varimax rotation of principal components
 BoysGirls
 Body sizeBlood pressureDyslipidemiaGlycemiaBody sizeBlood pressureNon-LDL- related dyslipidemiaLDL-related dyslipidemia
  • *

    Variables that had factor loadings <0.4 (or <−0.4) on a factor were considered as major constituents of this factor.

Body weight0.95*0.150.100.000.95*0.130.070.04
Waist circumference0.92*0.120.200.020.91*0.110.110.04
BMI0.95*0.130.160.030.95*0.100.110.07
Systolic BP0.310.79*0.060.150.170.86*0.040.07
Diastolic BP0.030.90*0.08−0.050.040.84*0.100.12
Triglycerides0.100.190.77*−0.170.010.050.84*0.26
HDL cholesterol−0.270.09−0.63*0.07−0.270.01−0.73*0.31
LDL cholesterol0.040.060.64*0.210.110.020.030.93*
Fasting plasma glucose0.020.06−0.010.96*0.060.51*−0.05−0.10
% variance explained31.417.016.511.430.519.314.311.8

Based on the adapted definitions of risk factors used in the NHANES, 3.4% of boys and 2.2% of girls had three or more risk factors. The AUC for both BMI and WC to predict clustering of risk factors ranged from 0.76 to 0.85 with 70% to 80% sensitivity and specificity (Table 3; Figure 1A). The age-adjusted cut-off values corresponded approximately to the 70th percentile values. Table 3 and Figure 1B show the corresponding AUCs and ROC curves using the NCEP-ATP III to define clustering of risk factors. The AUCs of the latter ROC curves were higher than that based on the adapted definition in the NHANES. Table 3B shows the smoothed, sex- and age-specific cut-off values for BMI and WC using the LMS method, above which there was increased probability of clustering of risk factors as defined by the NHANES.

Table 3A.  Results of Receiver Operating Characteristics (ROC) curve analyses to identify the optimal waist circumference and body mass index cutoff values to predict clustering of risk factors in 2102 Hong Kong Chinese adolescents (12–19 years of age) using different definitions.
 AUC (95% CI)Percentile valuesSensitivity (95% CI)Specificity (95% CI)
  • *

    HDL-C ≤1.03 mM, LDL-C ≥2.6 mM, TG ≥1.24 mM, FPG ≥6.1 mM and age-, sex- and height-adjusted SBP or DBP ≥90th percentile (12).

  • HDL-C ≤1.03 mM for male and ≤1.29 mM for female, LDL-C ≥2.6 mM, TG ≥1.69 mM, FPG ≥6.1 mM and age-, sex- and height-adjusted SBP or DBP ≥90th percentile (3, 13).

Waist circumference    
 Female*0.82 (0.74–0.91)76.8%0.72 (0.51–0.88)0.78 (0.75–0.80)
 Female0.86 (0.78–0.94)76.5%0.82 (0.60–0.95)0.78 (0.75–0.80)
 Male*0.78 (0.68–0.87)76.1%0.76 (0.58–0.89)0.78 (0.75–0.81)
 Male0.87 (0.82–0.92)77.2%0.89 (0.65–0.99)0.78 (0.76–0.81)
Body mass index    
 Female*0.85 (0.77–0.92)78.2%0.80 (0.59–0.93)0.79 (0.77–0.82)
 Female0.89 (0.82–0.95)85.7%0.77 (0.55–0.92)0.87 (0.85–0.89)
 Male*0.76 (0.66–0.85)71.8%0.76 (0.58–0.89)0.73 (0.71–0.76)
 Male0.86 (0.80–0.92)73.4%0.89 (0.65–0.99)0.75 (0.72–0.77)
image

Figure 1. ROC curves for age-adjusted WC and BMI (dashed line) to predict clustering of risk factors in 2102 Hong Kong Chinese adolescents 12 to 19 years of age using different definitions. (A) HDL-C ≤1.03 mM, LDL-C ≥2.6 mM, TG ≥1.24 mM, FPG ≥6.1 mM, and age-, sex-, and height-adjusted SBP or DBP ≥90th percentile (12). (B) HDL-C ≤1.03 mM for boys and ≤1.29 mM for girls, LDL-C ≥2.6 mM, TG ≥1.69 mM, FPG ≥6.1 mM, and age-, sex-, and height-adjusted SBP or DBP ≥90th percentile (3, 13).

Download figure to PowerPoint

Table 3B.  Age and sex-specific optimal waist circumference and body mass index cutoff values derived from the percentile values in the ROC analysis using the LMS method in 2102 Hong Kong Chinese adolescents 12 to 19 years of age
 Waist circumference (cm)BMI (kg/m2)
AgeFemaleMaleFemaleMale
     
1267.871.620.220.5
1368.973.420.820.9
1469.474.521.221.2
1569.675.421.521.5
     
1669.776.421.621.8
1769.777.421.622.1
1869.778.421.722.4
1969.879.321.722.7

Table 4 compared the 50th percentile BMI values between this cohort recruited in 2003 and that in the 1993 Hong Kong Growth Survey (15). For the same percentile value in boys, there was an increase of 0.40 to 1.33 kg/m2 in BMI values across all age groups during the 10-year period. Compared with girls, there was an increase in BMI (0.26 to 0.71 kg/m2) from 12 to 15 years of age. However, for girls 16 to 18 years of age, there was a decrease in BMI values of 0.06 to 0.9 kg/m2 compared with BMI values 10 years ago. Table 5 lists the odds ratio for clustering of risk factors in adolescents within the upper three deciles of BMI or WC compared with those in the lower seven deciles.

Table 4.  Comparison of the BMI values at the 50th percentile in our study cohort recruited in 2000 with the corresponding values in the 1993 Hong Kong Growth Survey
 BMI (kg/m2) of girlsBMI (kg/m2) of boys
AgeOur study (2003)1993 Hong Kong Growth SurveyOur study (2003)1993 Hong Kong Growth Survey
1217.8717.1618.5217.19
1318.3917.7618.9117.70
1418.8218.3419.2018.21
1519.1418.8819.4818.70
1619.3319.3919.7819.19
1719.3719.8720.1119.64
1819.4220.3220.4820.08
Table 5.  Comparisons in prevalence of clustering of three or more cardiovascular risk factors between adolescents in the lower seven deciles and those in the upper three deciles as well as those with values above or below the optimal cutoff values derived from the ROC analysis and expressed as odds ratio (95% confidence intervals)
 No clustering n (row %)Clustering n (row %)Odds ratio
Waist circumference   
 Girls   
  Lower 7 deciles795 (99.3%)6 (0.7%)1
  Upper 3 deciles324 (94.5%)19 (5.5%)7.8 (3.1–19.6)
  < optimal cutoff (76.8%)871 (99.2%)7 (0.8%)1
  ≥ optimal cutoff (76.8%)248 (93.2%)18 (6.8%)9.0 (3.7–21.9)
 Boys   
  Lower 7 deciles664 (99.0%)7 (1.0%)1
  Upper 3 deciles261 (90.9%)26 (9.1%)9.4 (4.1–22.0)
  < optimal cutoff (76.1%)731 (98.9%)8 (1.1%)1
  ≥ optimal cutoff (76.1%)194 (88.6%)25 (11.4%)11.8 (5.2–26.5)
BMI   
 Girls   
  Lower 7 deciles796 (99.4%)5 (0.6%)1
  Upper 3 deciles323 (94.2%)20 (5.8%)9.9 (3.7–26.5)
  < optimal cutoff (78.2%)889 (99.4%)5 (0.6%)1
  ≥ optimal cutoff (78.2%)230 (92.0%)20 (8.0%)15.5 (5.7–41.6)
 Boys   
  Lower 7 deciles662 (98.8%)8 (1.2%)1
  Upper 3 deciles263 (91.3%)25 (8.7%)7.9 (3.5–17.7)
  < optimal cutoff (71.8%)679 (98.8%)8 (1.2%)1
  ≥ optimal cutoff (71.8%)246 (90.8%)25 (9.2%)8.6 (3.8–19.4)

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

There is now a wealth of data showing the associations between anthropometric indices and cardiovascular risk factors in the Chinese adult population (16, 17). However, similar data in adolescents and children are relatively sparse. The latest BMI reference curves for Hong Kong Chinese children from birth to 18 years of age had been constructed using data from the 1993 Hong Kong Growth Survey (15). This data set has also been used in comparative studies to establish the international definition of obesity for children using BMI values (14). However, the predictive power of these BMI values for cardiovascular risks has not been fully evaluated. In this study, we used factor analysis to confirm the clustering of lipid, blood pressure, and glycemia and their close associations with anthropometric indices, a pattern similar to that observed in adults (18, 19). These risk factors have all been shown to predict future cardiovascular diseases, at least in adults (20).

In this analysis, we established references of simple anthropometric indices to identify risk factors clustering in adolescents, rather than to diagnose metabolic syndrome. There are ongoing debates regarding the concepts and clinical use of metabolic syndrome (21). The many versions of definitions of metabolic syndrome further add to its controversies (20). Although central obesity, frequently indicated by WC, is a major correlate with insulin resistance, its superiority over BMI remains debatable (20). Compared with BMI, measurements of WC are less reproducible. Nevertheless, using factor analysis, BMI, WC, and body weight were loaded on the same factor in our adolescents. Given the close association between this factor and other cardiovascular risk factors, our data support the usefulness of using both BMI and WC to indicate the presence of risk factors clustering.

In the absence of prospective data on clinical outcomes, the selection of cut-off values and number of risk factors remains arbitrary in defining clustering of risk factors. Indeed, the use of different cut-off values may give rise to different results and conclusions regarding the evaluation of cardiovascular risk factors in pediatric populations (22). In this study, we selected risk factors that have been used by other workers in pediatric populations (12, 23, 24, 25, 26). In 1999, the American Diabetes Association convened a consensus development conference that has provided the definition of type 2 diabetes in children and adolescents (27). The Fourth Report on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents from the National Blood Pressure Education Program Working Group on High Blood Pressure in Children and Adolescents (13) has also provided clear definition of hypertension for children and adolescents of different ages. The NCEP Pediatric Panel in 1992 (28) classified TC <170 mg/dL (4.4 mM) and LDL-C <110 mg/dL (2.8 mM) as acceptable levels in children and adolescents. In the NHANES (1988 to 1994), TG threshold value of <110 mg/dL (1.24 mM) was used to define metabolic syndrome in adolescents (12). Based on these cut-off values of various risk factors (BP, FPG, TG, HDL-C, and LDL-C), we established two sets of definitions of clustering of risk factors (three or more) in this cohort. Based on the ROC analysis, the cut-off values of BMI and WC approximate to the 70th percentile values, with sensitivity and specificity ranging from 72% to 80% in boys and girls.

Ethnicity is an important consideration in defining obesity (29). Our group was among the first to point out the need to use a lower BMI cut-off value to define obesity in adult Hong Kong Chinese (30). This ethnic difference may reflect genetic differences in our tendency to accumulate fat and its distribution (26, 31). Our analysis in these adolescents also seems to support this notion. Using the age of 18 years as the beginning of early adulthood, the BMI values of 21.7 kg/m2 for girls and 22.4 kg/m2 for boys were selected as the optimal cut-off values. These values are similar to 23 kg/m2 now used to define overweight in Asian adults compared with 25 kg/m2 in whites (31). These relatively low cut-off values are further supported by the 8- to 16-fold difference in odd ratios for clustering of risk factors between subjects in the upper three deciles and those in the lower seven deciles of BMI or WC.

In line with the rising prevalence of obesity and young onset diabetes in our population (32), the 50th BMI percentile values of our cohort recruited in 2003 were higher for boys across all age groups and for girls from 12 to 15 years of age than those reported in the 1993 Hong Kong Growth Survey (15). These findings suggest that Hong Kong adolescents have become more obese over the last decade, in association with rapid changes in lifestyle and socioeconomic developments. Several large-scale studies have now confirmed the effects of excessive television watching, consumption of energy dense food, and physical inactivity as major causes of obesity and diabetes in both adults and children (33, 34, 35).

Despite an increasing trend of BMI over time in our adolescents, the age-related trends are different between boys and girls. In girls, the 50th percentile BMI values increased from the age of 12 years, reaching a plateau around 16 years of age, which was followed by declining values. Similar trends have also been reported by the Student Health Program of the Hong Kong Department of Health (36). On the other hand, the BMI values increased linearly with age in boys. These differences might be caused by the earlier onset of puberty in girls and differences in timing and peak of growth spurts between the two sexes. The plateau in BMI and WC among girls 16 to 19 years of age was also observed as in the 1993 Growth Survey in Hong Kong. However, the 50th BMI percentile of our girls 16 to 19 years of age were lower than that reported in 1993. The reasons for these differences are not immediately obvious. However, it is plausible that increasing awareness of obesity and societal and peer influences, including mass media and advertisements of slimming products, might have influenced these young girls to lose weight as they approached womanhood (36).

Despite the growing burden of obesity, definitions for overweight and obesity in adolescents remain controversial. Cole et al. (37) first defined international cut-off points for overweight/obesity in children by tying the BMI values to the adult overweight (25 kg/m2) and obesity thresholds (30 kg/m2) and averaging the results obtained from six international regions. Based on these BMI values and their corresponding percentile values in the 18-year age group in the 1993 Hong Kong Growth Survey, these cut-off values of 25 and 30 k/gm (2) will correspond to the 90.2 and 98.2 percentiles in girls and the 98.2 and 96.9 percentiles in boys, respectively. However, in this survey conducted 10 years later when there is an overall increase in obesity trend, these values corresponded to the 90 and 99.4 percentiles in girls and 87.6 and 97.8 percentiles in boys. On the other hand, if we use clustering of risk factors as the referent point, the optimal BMI cut-off values were set at the 78.2 percentile for girls and the 71.8 percentile for boys. These findings show that the optimal cut-off values in predicting the high cardiovascular risk group can be markedly different when different methods or studies are used. Longitudinal studies are clearly needed to confirm the validity of these different cut-off values using clinical events as endpoints.

Our survey has one major limitation. Because of the practical difficulty, we were not able to assess the pubertal state of all participants during the field study. Puberty and associated hormonal changes can influence insulin sensitivity and lipoprotein profiles. However, the mean age of our study population was 15 years, with a slight female preponderance. Because girls have earlier puberty than boys, the majority of the participants were likely to be postpubertal, thus upholding the validity of our analysis, except for the younger boys who might be prepubertal at the time of the survey.

In conclusion, we identified age- and sex-specific BMI and WC cut-off values with high sensitivity and specificity to predict clustering of risk factors in Chinese adolescents. Longitudinal studies are needed to confirm whether they are useful in predicting future cardiovascular events.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References

We thank all school personnel, parents, and participants for making the study possible. Special thanks are extended to Delanda Wong, Yee-Mui Lee, and Stanley Wong for conducting the survey. This study was supported by the Research Grants Council of the Hong Kong Special Administrative Region, China (CUHK 4055/01M) and Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong.

Footnotes
  • 1

    Nonstandard abbreviations: WC, waist circumference; ROC, receiver operating characteristic; BP, blood pressure; FPG, fasting plasma glucose; TC, total cholesterol; TG, triglyceride; HDL-C, high-density lipoprotein-cholesterol; LDL-C, low-density lipoprotein-cholesterol; NHANES, National Health and Nutrition Education Survey; LMS, lambda-mu-sigma; AUC, area under the ROC curve; NCEP-ATP III, National Cholesterol Education Program-Adult Treatment Panel III.

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgments
  9. References
  • 1
    Kuczmarski, R. J., Flegal, K. M., Campbell, S. M., Johnson, CL (1994) Increasing prevalence of obesity among US adults. The National Health and Nutrition Examination Surveys, 1960 to 1991. J Am Med Assoc. 272: 205211.
  • 2
    Deurenberg, P., Yap, M. (1999) The assessment of obesity: methods for measuring body fat and global prevalence of obesity. Baillieres Best Pract Res Clin Endocrinol Metab. 13: 111.
  • 3
    Expert Panel on Detection Evaluation and Treatment of High Blood Cholesterol in Adults (2001) Executive summary of the Third Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation and Treatment of high blood cholesterol in adults (Adult Treatment Panel III). JAMA 285: 24862497.
  • 4
    Isomaa, B., Almgren, P., Tuomi, T., et al (2001) Cardiovascular morbidity and mortality associated with the metabolic syndrome. Diabetes Care 24: 683689.
  • 5
    World Health Organization (1998) Obesity: Preventing and Managing the Global Epidemic World Health Organization Geneva, Switzerland. (WHO Technical Report Series, No. 894).
  • 6
    Kvaavik, E., Tell, G. S., Klepp, KI (2003) Predictors and tracking of body mass index from adolescence into adulthood: follow-up of 18 to 20 years in the Oslo Youth Study. Arch Pediatr Adolesc Med. 157: 12121218.
  • 7
    Srinivasan, S. R., Bao, W., Wattigney, W. A., Berenson, GS (1996) Adolescent overweight is associated with adult overweight and related multiple cardiovascular risk factors: the Bogalusa Heart Study. Metabolism 45: 235240.
  • 8
    Barnekow-Bergkvist, M., Hedberg, G., Janlert, U., Jansson, E. (2001) Adolescent determinants of cardiovascular risk factors in adult men and women. Scand J Public Health 29: 208217.
  • 9
    Guo, S. S., Huang, C., Maynard, L. M., et al (2000) Body mass index during childhood, adolescence and young adulthood in relation to adult overweight and adiposity: the Fels Longitudinal Study. Int J Obes Relat Metab Disord. 24: 16281635.
  • 10
    Myers, L., Coughlin, S. S., Webber, L. S., Srinivasan, S. R., Berenson, GS (1995) Prediction of adult cardiovascular multifactorial risk status from childhood risk factor levels. The Bogalusa Heart Study. Am J Epidemiol. 142: 918924.
  • 11
    Friedewald, W. T., Levy, R. I., Fredrickson, DS (1972) Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 18: 499502.
  • 12
    Cook, S., Weitzman, M., Auinger, P., Nguyen, M., Dietz, W. (2003) Prevalence of a metabolic syndrome phenotype in adolescents. Findings from the Third National Health and Nutrition Examination Survey, 1988–1994. Arch Pediatr Adolesc Med. 157: 821827.
  • 13
    The National High Blood Pressure Education Program Working Group on High Blood Pressure in Children and Adolescents (2004) The Fourth Report on the Diagnosis, Evaluation and Treatment of High Blood Pressure in Children and Adolescents. Pediatrics 114: 555576.
  • 14
    Cole, T. J., Green, PJ (1992) Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med. 11: 13051319.
  • 15
    Leung, S. S., Cole, T. J., Tse, L. Y., Lau, JT (1998) Body mass index reference curves for Chinese children. Ann Hum Biol. 25: 169174.
  • 16
    Ko, G. T. C., Chan, J. C. N., Woo, J., Cockram, CS (1999) Prediction of hypertension, diabetes, dyslipidaemia or albuminuria using simple arthropometric indexes in Hong Kong Chinese. Int J Obes Relat Metab Disord. 23: 11361142.
  • 17
    Deurenberg-Yap, M., Chew, S. K., Lin, V. F., Tan, B. Y., van Staveren, W. A., Deurenberg, P. (2001) Relationships between indices of obesity and its co-morbidities in multi-ethnic Singapore. Int J Obes Relat Metab Disord. 25: 15541562.
  • 18
    Alberti, K. G., Zimmet, P., Shaw, J. (2005) The metabolic syndrome—a new worldwide definition. Lancet 366: 10591062.
  • 19
    Anderson, P. J., Critchley, Joby, Chan, J. C. N., et al (2001) Factor analysis of the metabolic syndrome: obesity vs insulin resistance as the central abnormality. Int J Obes Relat Metab Disord. 25: 17821788.
  • 20
    Reaven, GM (2005) The metabolic syndrome: requiescat in pace. Clin Chem. 51: 931938.
  • 21
    Kahn, R., Buse, J., Ferrannini, E., Stern, M. (2005) The metabolic syndrome: time for a critical appraisal: joint statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 28: 22892304.
  • 22
    Goodman, E. (2006) Pediatric metabolic syndrome: smoke and mirrors or true magic? J Pediatr. 148: 149151.
  • 23
    de Ferranti, S. D., Gauvreau, K., Ludwig, D. S., Neufeld, E. J., Newburger, J. W., Rifai, N. (2004) Prevalence of the metabolic syndrome in American adolescents: findings from the Third National Health and Nutrition Examination Survey. Circulation 110: 24942497.
  • 24
    Goodman, E., Daniels, S. R., Morrison, J. A., Huang, B., Dolan, LM (2004) Contrasting prevalence of and demographic disparities in the World Health Organization and National Cholesterol Education Program Adult Treatment Panel III definitions of metabolic syndrome among adolescents. J Pediatr. 145: 445451.
  • 25
    Weiss, R., Dziura, J., Burgert, T. S., et al (2004) Obesity and the metabolic syndrome in children and adolescents. N Engl J Med. 350: 23622374.
  • 26
    Katzmarzyk, P. T., Srinivasan, S. R., Chen, W., Malina, R. M., Bouchard, C., Berenson, GS (2004) Body mass index, waist circumference, and clustering of cardiovascular disease risk factors in a biracial sample of children and adolescents. Pediatrics 114: e198e205.
  • 27
    American Diabetes Association (2000) Type 2 diabetes in children and adolescents. American Diabetes Association. Diabetes Care 23: 381389.
  • 28
    National Cholesterol Education Program (NCEP) (1992) Highlights of the report of the Expert Panel on Blood Cholesterol Levels in Children and Adolescents. Pediatrics 89: 495501.
  • 29
    WHO Expert Consultation (2004) Appropriate body mass index for Asian populations and its implications for policy and intervention strategies. Lancet 363: 157163.
  • 30
    Ko, G. T. C., Tang, J., Chan, J. C. N., et al (2001) Lower BMI cut off value to define BMI obesity in Hong Kong Chinese: an analysis based on body fat assessment by bioelectrical impedence. Brit J Nutr. 85: 239242.
  • 31
    Deurenberg, P., Deurenbery-Yap, M., Guricci, S. (2002) Asians are different from Caucasians and from each other in their body mass index/body fat per cent relationship. Obes Rev. 3: 141146.
  • 32
    Chan, J. C. N., Ng, MCY (2003) Lessons learned from young onset diabetes in China. Curr Diabetes Rep. 3: 101107.
  • 33
    Pereira, M. A., Kartashov, A. I., Ebbeling, C. B., et al (2005) Fast-food habits, weight gain, and insulin resistance (the CARDIA study): 15-year prospective analysis. Lancet 365: 3642.
  • 34
    Hancox, R. J., Milne, B. J., Poulton, R. (2004) Association between child and adolescent television viewing and adult health: a longitudinal birth cohort study. Lancet 364: 257262.
  • 35
    Ebbeling, C. B., Pawlak, D. B., Ludwig, DS (2002) Childhood obesity: public-health crisis, common sense cure. Lancet 360: 473482.
  • 36
    Kwok, P., Tse, LY (2004) Overweight and obesity in Hong Kong- what do we know? Public Health Epidemiol Bull. 13: 5460.
  • 37
    Cole, T. J., Bellizzi, M. C., Flegal, K. M., Dietz, WH (2000) Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 320: 12401243.