body composition and its relationship to …eprints.qut.edu.au/47432/1/ailing_liu_thesis.pdf ·...

Body Composition and Its Relationship to Metabolic Risk Factors in Asian Children Ailing Liu

1

BODY COMPOSITION AND ITS RELATIONSHIP TO METABOLIC RISK

FACTORS IN ASIAN CHILDREN

A thesis submitted in fulfillment of the requirements for the award of:

Doctor of Philosophy

Submitted By:

Ailing Liu

MD, MSc

Institute of Health and Biomedical Innovation

School of Human Movements Studies

Faculty of Health

Queensland University of Technology

2011


2

KEYWORDS

Obesity, anthropometry, body composition, body fat, total body water, fat‐free mass,

body mass index, body fat distribution, skinfold thickness, bioelectric impedance

analysis, deuterium dilution technique, metabolic syndrome, cardiovascular risk

factors, Asians, children, ethnicity


3

ABSTRACT

Obesity is a major public health problem in both developed and developing

countries. The body mass index (BMI) is the most common index used to define

obesity. The universal application of the same BMI classification across different

ethnic groups is being challenged due to the inability of the index to differentiate fat

mass (FM) and fat‐free mass (FFM) and the recognized ethnic differences in body

composition. A better understanding of the body composition of Asian children

from different backgrounds would help to better understand the obesity‐related

health risks of people in this region. Moreover, the limitations of the BMI

underscore the necessity to use where possible, more accurate measures of body fat

assessment in research and clinical settings in addition to BMI, particularly in

relation to the monitoring of prevention and treatment efforts.

The aim of the first study was to determine the ethnic difference in the relationship

between BMI and percent body fat (%BF) in pre‐pubertal Asian children from China,

Lebanon, Malaysia, the Philippines, and Thailand. A total of 1039 children aged 8‐10

y were recruited using a non‐random purposive sampling approach aiming to

encompass a wide BMI range from the five countries. Percent body fat (%BF) was

determined using the deuterium dilution technique to quantify total body water

(TBW) and subsequently derive proportions of FM and FFM. The study highlighted

the sex and ethnic differences between BMI and %BF in Asian children from

different countries. Girls had approximately 4.0% higher %BF compared with boys at

a given BMI. Filipino boys tended to have a lower %BF than their Chinese, Lebanese,

Malay and Thai counterparts at the same age and BMI level (corrected mean %BF

was 25.7±0.8%, 27.4±0.4%, 27.1±0.6%, 27.7±0.5%, 28.1±0.5% for Filipino, Chinese,

Lebanese, Malay and Thai boys, respectively), although they differed significantly

from Thai and Malay boys. Thai girls had approximately 2.0% higher %BF values than

Chinese, Lebanese, Filipino and Malay counterparts (however no significant

difference was seen among the four ethnic groups) at a given BMI (corrected mean

%BF was 31.1±0.5%, 28.6±0.4%, 29.2±0.6%, 29.5±0.6%, 29.5±0.5% for Thai, Chinese,

Lebanese, Malay and Filipino girls, respectively). However, the ethnic difference in


4

BMI‐%BF relationship varied by BMI. Compared with Caucasians, Asian children had

a BMI 3‐6 units lower for a given %BF. More than one third of obese Asian children

in the study were not identified using the WHO classification and more than half

were not identified using the International Obesity Task Force (IOTF) classification.

However, use of the Chinese classification increased the sensitivity by 19.7%, 18.1%,

2.3%, 2.3%, and 11.3% for Chinese, Lebanese, Malay, Filipino and Thai girls,

respectively. A further aim of the first study was to determine the ethnic difference

in body fat distribution in pre‐pubertal Asian children from China, Lebanon,

Malaysia, and Thailand. The skin fold thicknesses, height, weight, waist

circumference (WC) and total adiposity (as determined by deuterium dilution

technique) of 922 children from the four countries was assessed. Chinese boys and

girls had a similar trunk‐to‐extremity skin fold thickness ratio to Thai counterparts

and both groups had higher ratios than the Malays and Lebanese at a given total FM.

At a given BMI, both Chinese and Thai boys and girls had a higher WC than Malays

and Lebanese (corrected mean WC was 68.1±0.2 cm, 67.8±0.3 cm, 65.8±0.4 cm,

64.1±0.3 cm for Chinese, Thai, Lebanese and Malay boys, respectively; 64.2±0.2 cm,

65.0±0.3 cm, 62.9±0.4 cm, 60.6±0.3 cm for Chinese, Thai, Lebanese and Malay girls,

respectively). Chinese boys and girls had lower trunk fat adjusted

subscapular/suprailiac skinfold ratio compared with Lebanese and Malay

counterparts.

The second study aimed to develop and cross‐validate bioelectrical impedance

analysis (BIA) prediction equations of TBW and FFM for Asian pre‐pubertal children

from China, Lebanon, Malaysia, the Philippines, and Thailand. Data on height,

weight, age, gender, resistance and reactance measured by BIA were collected from

948 Asian children (492 boys and 456 girls) aged 8‐10 y from the five countries. The

deuterium dilution technique was used as the criterion method for the estimation of

TBW and FFM. The BIA equations were developed from the validation group (630

children randomly selected from the total sample) using stepwise multiple

regression analysis and cross‐validated in a separate group (318 children) using the

Bland‐Altman approach. Age, gender and ethnicity influenced the relationship

between the resistance index (RI = height2/resistance), TBW and FFM. The BIA


5

prediction equation for the estimation of TBW was: TBW (kg) = 0.231×Height2

(cm)/resistance (Ω) + 0.066×Height (cm) + 0.188×Weight (kg) + 0.128×Age (yr) +

0.500×Sex (male=1, female=0) – 0.316×Ethnicity (Thai ethnicity=1, others=0) ‐ 4.574,

and for the estimation of FFM: FFM (kg) = 0.299×Height2 (cm)/resistance (Ω) +

0.086×Height (cm) + 0.245×Weight (kg) + 0.260×Age (yr) + 0.901×Sex (male=1,

female=0) ‐ 0.415×Ethnicity (Thai ethnicity=1, others=0) ‐ 6.952. The R2 was 88.0%

(root mean square error, RSME = 1.3 kg), 88.3% (RSME = 1.7 kg) for TBW and FFM

equation, respectively. No significant difference between measured and predicted

TBW and between measured and predicted FFM for the whole cross‐validation

sample was found (bias = ‐0.1±1.4 kg, pure error = 1.4±2.0 kg for TBW and bias =

‐0.2±1.9 kg, pure error = 1.8±2.6 kg for FFM). However, the prediction equation for

estimation of TBW/FFM tended to overestimate TBW/FFM at lower levels while

underestimate at higher levels of TBW/FFM. Accuracy of the general equation for

TBW and FFM compared favorably with both BMI‐specific and ethnic‐specific

equations. There were significant differences between predicted TBW and FFM from

external BIA equations derived from Caucasian populations and measured values in

Asian children.

There were three specific aims of the third study. The first was to explore the

relationship between obesity and metabolic syndrome and abnormalities in Chinese

children. A total of 608 boys and 800 girls aged 6‐12 y were recruited from four

cities in China. Three definitions of pediatric metabolic syndrome and abnormalities

were used, including the International Diabetes Federation (IDF) and National

Cholesterol Education Program (NCEP) definition for adults modified by Cook et al.

and de Ferranti et al. The prevalence of metabolic syndrome varied with different

definitions, was highest using the de Ferranti definition (5.4%, 24.6% and 42.0%,

respectively for normal‐weight, overweight and obese children), followed by the

Cook definition (1.5%, 8.1%, and 25.1%, respectively), and the IDF definition (0.5%,

1.8% and 8.3%, respectively). Overweight and obese children had a higher risk of

developing the metabolic syndrome compared to normal‐weight children (odds

ratio varied with different definitions from 3.958 to 6.866 for overweight children,

and 12.640‐26.007 for obese children). Overweight and obesity also increased the


6

risk of developing metabolic abnormalities. Central obesity and high triglycerides

(TG) were the most common while hyperglycemia was the least frequent in Chinese

children regardless of different definitions. The second purpose was to determine

the best obesity index for the prediction of cardiovascular (CV) risk factor clustering

across a 2‐y follow‐up among BMI, %BF, WC and waist‐to‐height ratio (WHtR) in

Chinese children. Height, weight, WC, %BF as determined by BIA, blood pressure,

TG, high‐density lipoprotein cholesterol (HDL‐C), and fasting glucose were collected

at baseline and 2 years later in 292 boys and 277 girls aged 8‐10 y. The results

showed the percentage of children who remained overweight/obese defined on the

basis of BMI, WC, WHtR and %BF was 89.7%, 93.5%, 84.5%, and 80.4%, respectively

after 2 years. Obesity indices at baseline significantly correlated with TG, HDL‐C, and

blood pressure at both baseline and 2 years later with a similar strength of

correlations. BMI at baseline explained the greatest variance of later blood pressure.

WC at baseline explained the greatest variance of later HDL‐C and glucose, while

WHtR at baseline was the main predictor of later TG. Receiver‐operating

characteristic (ROC) analysis explored the ability of the four indices to identify the

later presence of CV risk. The overweight/obese children defined on the basis of

BMI, WC, WHtR or %BF were more likely to develop CV risk 2 years later with

relative risk (RR) scores of 3.670, 3.762, 2.767, and 2.804, respectively. The final

purpose of the third study was to develop age‐ and gender‐specific percentiles of

WC and WHtR and cut‐off points of WC and WHtR for the prediction of CV risk in

Chinese children. Smoothed percentile curves of WC and WHtR were produced in

2830 boys and 2699 girls aged 6‐12 y randomly selected from southern and

northern China using the LMS method. The optimal age‐ and gender‐specific

thresholds of WC and WHtR for the prediction of cardiovascular risk factors

clustering were derived in a sub‐sample (n=1845) by ROC analysis. Age‐ and

gender‐specific WC and WHtR percentiles were constructed. The WC thresholds

were at the 90th and 84th percentiles for Chinese boys and girls, respectively, with

sensitivity and specificity ranging from 67.2% to 83.3%. The WHtR thresholds were

at the 91st and 94th percentiles for Chinese boys and girls, respectively, with

sensitivity and specificity ranging from 78.6% to 88.9%. The cut‐offs of both WC and


7

WHtR were age‐ and gender‐dependent.

In conclusion, the current thesis quantifies the ethnic differences in the BMI‐%BF

relationship and body fat distribution between Asian children from different origins

and confirms the necessity to consider ethnic differences in body composition when

developing BMI and other obesity index criteria for obesity in Asian children.

Moreover, ethnicity is also important in BIA prediction equations. In addition, WC

and WHtR percentiles and thresholds for the prediction of CV risk in Chinese

children differ from other populations. Although there was no advantage of WC or

WHtR over BMI or %BF in the prediction of CV risk, obese children had a higher risk

of developing the metabolic syndrome and abnormalities than normal‐weight

children regardless of the obesity index used.


8

TABLE OF CONTENTS

KEYWORDS… .......................................................................................................... 2

ABSTRACT…. ........................................................................................................... 3

TABLE OF CONTENTS ............................................................................................... 8

LIST OF PUBLICATIONS .......................................................................................... 12

LIST OF TABLES...................................................................................................... 14

LIST OF FIGURES.................................................................................................... 17

LIST OF ABBREVIATIONS........................................................................................ 22

STATEMENT OF ORIGINAL AUTHORSHIP................................................................ 24

ACKNOWLEDGEMENTS ......................................................................................... 25

CHAPTER 1 INTRODUCTION................................................................................. 27

CHAPTER 2 LITERATURE REVIEW ......................................................................... 35

2.1 CHILDHOOD OBESITY WORLDWIDE ......................................................................................... 35

2.2 DEFINITION OF OVERWEIGHT AND OBESITY IN CHILDREN...................................................... 38

2.3 AGE, SEX AND ETHNIC DIFFERENCES IN BODY COMPOSITION IN CHILDREN........................... 44

2.3.1 Age differences in body composition in children ........................................................... 44

2.3.1.1 Total body fat (TBF) .................................................................................................. 44 2.3.1.2 Fat‐free mass (FFM) ................................................................................................. 48

2.3.2 Sex differences in body composition in children ............................................................ 53

2.3.2.1 Total body fat ........................................................................................................... 53 2.3.2.2 Fat‐free mass............................................................................................................ 55

2.3.3 Ethnic differences in body composition in children ....................................................... 57

2.3.3.1 Total body fat ........................................................................................................... 58 2.3.3.2 Fat‐free mass............................................................................................................ 69

2.3.4 Ethnic differences in the relationship between BMI and %BF........................................ 73

2.4 BODY COMPOSITION ASSESSMENT IN CHILDREN.................................................................... 85

2.4.1 Bioelectrical impedance analysis (BIA) ........................................................................... 85

2.4.1.1 Assumptions and principles...................................................................................... 86 2.4.1.2 BIA methods ............................................................................................................. 86 2.4.1.3 Sources of measurement error ................................................................................. 87 2.4.1.4 BIA prediction equations .......................................................................................... 89 2.4.1.5 Choice of BIA prediction equations .......................................................................... 91

2.4.2 Reference methods......................................................................................................... 94

2.4.2.1 Principle and assumptions of TBW method ............................................................. 94


9

2.4.2.2 Equipment used in the TBW method........................................................................ 94 2.4.2.3 Measurement procedures of TBW method.............................................................. 95 2.4.2.4 Precision and accuracy of TBW method................................................................... 97

2.5 OBESITY AND THE METABOLIC SYNDROME IN CHILDREN........................................................ 98

CHAPTER 3 ETHNIC DIFFERENCE IN BODY COMPOSITION AMONG ASIAN CHILDREN FROM DIFFERENT ORIGINS..............................................107

3.1 ETHNIC DIFFERENCES IN THE RELATIONSHIP BETWEEN BMI AND %BF AMONG ASIAN CHILDREN FROM DIFFERENT BACKGROUNDS...................................................................... 107

3.1.1 Introduction.................................................................................................................. 107

3.1.2 Methodology ................................................................................................................ 108

3.1.2.1 Participants ............................................................................................................ 108 3.1.2.2 Anthropometric measurements ............................................................................. 109 3.1.2.3 Body composition measurement............................................................................ 110 3.1.2.4 Statistical analysis .................................................................................................. 112

3.1.3 Results .......................................................................................................................... 115

3.1.3.1 BMI‐age relationship.............................................................................................. 117 3.1.3.2 BMI‐sex relationships ............................................................................................. 119 3.1.3.3 BMI‐ethnicity relationship...................................................................................... 121 3.1.3.4 Validation of WHO, IOTF and China obesity classification ..................................... 124 3.1.3.5 Equivalent BMI cut‐offs for obesity ........................................................................ 126

3.1.4 Discussion..................................................................................................................... 126

3.2 ETHNIC DIFFERENCES IN SUBCUTANEOUS ADIPOSITY AND WAIST CIRCUMFERENCE IN ASIAN PRE‐PUBERTAL CHILDREN..................................................................................... 133

3.2.1 Introduction.................................................................................................................. 133

3.2.2 Methodology ................................................................................................................ 134

3.2.2.1 Participants ............................................................................................................ 134 3.2.2.2 Anthropometric measurements ............................................................................. 135 3.2.2.3 Skin fold measurements ......................................................................................... 136 3.2.2.4 Statistical analysis .................................................................................................. 138

3.2.3 Results .......................................................................................................................... 139

3.2.3.1 Ethnic differences in body fat distribution ............................................................. 140 3.2.3.2 Sex difference in body fat distribution.................................................................... 159 3.2.3.3 Age difference in body fat contribution ................................................................. 160

3.2.4 Discussion..................................................................................................................... 167

CHAPTER 4 VALIDATION OF BIA FOR TBW ANALYSIS AGAINST THE DEUTERIUM DILUTION TECHNIQUE IN ASIAN CHILDREN .................171

4.1 INTRODUCTION ...................................................................................................................... 171

4.2 METHODOLOGY...................................................................................................................... 173

4.2.1 Participants................................................................................................................... 173

4.2.2 Anthropometric measurements ................................................................................... 174

4.2.3 Bioelectrical impedance analysis measurement .......................................................... 174


10

4.2.4 Deuterium oxide dilution technique ............................................................................ 175

4.2.5 Statistical analysis ......................................................................................................... 176

4.3 RESULTS .................................................................................................................................. 179

4.3.1 Development of BIA equations..................................................................................... 179

4.3.1.1 Prediction equation for FFM................................................................................... 179 4.3.1.2 Prediction equation for TBW .................................................................................. 180 4.3.1.3 Ethnic‐specific equations........................................................................................ 182 4.3.1.4 BMI‐specific prediction equations .......................................................................... 184

4.3.2 Cross‐validation of the equations................................................................................. 184

4.3.2.1 Cross‐validation of equations for FFM ................................................................... 184 4.3.2.2 Cross‐validation of equations for TBW ................................................................... 195

4.3.3 Cross‐validation of external BIA prediction equation................................................... 205

4.4 DISCUSSION............................................................................................................................ 207

CHAPTER 5 OBESITY AND THE METABOLIC SYNDROME IN CHILDREN ................ 213

5.1 THE ASSOCIATION OF OBESITY WITH THE METABOLIC SYNDROME IN CHINESE CHILDREN ............................................................................................................................. 213

5.1.1 Introduction.................................................................................................................. 213

5.1.2 Methodology ................................................................................................................ 214

5.1.2.1 Participants ............................................................................................................ 214 5.1.2.2 Anthropometric measurements ............................................................................. 214 5.1.2.3 Metabolic variables .............................................................................................. 214 5.1.2.4 Definition of paediatric metabolic syndrome ......................................................... 215 5.1.2.5 Statistical analysis .................................................................................................. 216

5.1.3 Results .......................................................................................................................... 216

5.1.4 Discussion ..................................................................................................................... 220

5.2 ASSOCIATIONS OF OBESITY INDICES WITH METABOLIC RISK FACTORS AMONG CHINESE CHILDREN .............................................................................................................. 223

5.2.1 Introduction.................................................................................................................. 223

5.2.2 Methodology ................................................................................................................ 224

5.2.2.1 Participants ............................................................................................................ 224 5.2.2.2 Anthropometric measurements ............................................................................. 224 5.2.2.3 Body composition assessmen................................................................................. 225 5.2.2.4 Pubertal stage assessment..................................................................................... 225 5.2.2.5 Metabolic variables .............................................................................................. 225 5.2.2.6 Definition of overweight, obesity and central adiposity......................................... 226 5.2.2.7 Definition of metabolic risk clustering.................................................................... 226 5.2.2.8 Statistical analysis .................................................................................................. 226

5.2.3 Results .......................................................................................................................... 227

5.2.4 Discussion ..................................................................................................................... 236

5.3 WAIST CIRCUMFERENCE AND WAIST‐TO‐HEIGHT RATIO CUT‐OFF VALUES FOR THE PREDICTION OF CARDIOVASCULAR RISK FACTORS CLUSTERING IN CHINESE SCHOOL‐AGED CHILDREN..................................................................................................... 240


11

5.3.1 Introduction.................................................................................................................. 240

5.3.2 Methodology ................................................................................................................ 242

5.3.2.1 Participants ............................................................................................................ 242 5.3.2.2 Anthropometric measurements ............................................................................. 242 5.3.2.3 Cardiovascular risk factors measurements ............................................................ 242 5.3.2.4 Definition of high CV risk factors clustering ........................................................... 243 5.3.2.5 Statistical analysis .................................................................................................. 243

5.3.3 Results .......................................................................................................................... 244

5.3.4 Discussion..................................................................................................................... 252

CHAPTER 6 GENERAL DISCUSSION AND CONCLUSIONS........................................259

6.1 GENERAL DISCUSSION............................................................................................................ 259

6.2 CONCLUSIONS AND IMPLICATIONS……………………………………………………………………………………266

REFERENCES.. ......................................................................................................269

APPENDICES.. ......................................................................................................303

Appendix 1: WHO growth reference for school‐aged children and adolescents ......................... 303

Appendix 2: Standard operating procedures ‐ deuterium dilution technique............................. 304

Appendix 3: Human research ethics approval certificate............................................................. 308

Appendix 4: Contribution of Candidate to the whole project...................................................... 310


12

LIST OF PUBLICATIONS

Peer‐reviewed International Journal articles

Liu A, Hills AP, Hu X, Li Y, Du L, Xu Y, Byrne NM, Ma G (2010). Waist circumference

cut‐off points for the prediction of cardiovascular risk factors clustering in Chinese

school‐aged children: a cross‐sectional study. BMC Public Health, 10: 82.

Liu A, Byrne NM, Kagawa M, Ma G, Poh BK, Ismail MN, Kijboonchoo K, Nasreddine L,

Trinidad TP, Hills AP (2011). Ethnic differences in the relationship between body

mass index and percentage body fat among Asian children from different

backgrounds. Br J Nutr, DOI: 10.1017/S0007114511001681.

Liu A, Byrne NM, Kagawa M, Ma G, Kijboonchoo K, Nasreddine L, Poh BK, Ismail MN,

Hills AP (2011). Ethnic differences in body fat distribution among Asian pre‐pubertal

children: a cross‐sectional multicenter study. BMC Public Health, 11:500.

Liu A, Byrne NM, Ma G, Nasreddine L, Trinidad TP, Kijboonchoo K, Ismail MN,

Kagawa M, Poh BK, Hills AP (2011). Validation of bioelectrical impedance analysis for

total body water assessment against the deuterium dilution technique in Asian

children. Accepted by Eur J Clin Nutr.

Conference Presentations: Oral

Liu A, Byrne NM, Ma G, Nasreddine L, Trinidad TP, Kijboonchoo K, Ismail MN,

Kagawa M, Poh BK, Hills AP. Validation of bioelectrical impedance analysis for total

body water assessment against the deuterium dilution technique in Asian children.

The Ninth International Symposium on In Vivo Body Composition Studies, Hangzhou,

China, 21‐24 May, 2011.

Conference Presentations: Poster

Liu A, Hills AP, Byrne NM, Hu X, Ma G. Which obesity index is the best predictor of

metabolic risk in Chinese school children? The 19th International Congress on

Nutrition, Bangkok, 4‐9 October, 2009.


13

Liu A, Ma G, Kijboonchoo K, Ismail MN, Nasreddine L, Trinidad TP, Byrne NM, Hills AP.

Ethnic differences in the relationship between body mass index and percent body

fat among Asian children. The 4th Scandinavian Pediatric Obesity Conference,

Stockholm, 9‐10th July, 2010.


14

LIST OF TABLES

Table 2.1. Indices and cut‐off points used in different countries to define obesity ...................................................................................................................... 39

Table 2.2. Body composition of reference children ................................................. 45

Table 2.3. Total body water volume (L) for age and gender..................................... 49

Table 2.4. Total body bone mineral content and density for age and gender (Mean±SD)................................................................................................................ 51

Table 2.5. Muscle mass as a percentage of body weight and width for age and sex............................................................................................................................. 53

Table 2.6. The 50th percentile of WC in different ethnicities/countries.................. 68

Table 2.7. Body composition for age, sex and ethnicity: NHANES III ....................... 70

Table 2.8. Total body bone mineral content and bone mineral density for children and adolescents from different ethnicities ................................................ 72

Table 2.9 Comparison of BMI in whites with other ethnic groups reflecting the same %BF ................................................................................................................. 84

Table 2.10. Standards for evaluating prediction errors............................................ 90

Table 2.11. Some BIA prediction equations for children and adolescents............... 93

Table 2.12. Metabolic indicators and cut‐off points used to classify paediatric metabolic syndrome................................................................................................. 99

Table 2.13. Proposed waist circumference cut‐off values for predicting cardiovascular disease risk factors......................................................................... 106

Table 3.1. BMI cut‐off points for overweight and obesity by sex between 8 and 10 y proposed by IOTF..................................................................................... 110

Table 3.2. BMI cut‐off points for overweight and obesity by sex between 8 and 10 y proposed by the Chinese classification ................................................... 110

Table 3.3. Age‐ and gender‐specific hydration constants of the FFM in children .. 112

Table 3.4. Characteristics of the participants (mean±SD) ...................................... 116

Table 3.5. Pearson correlation coefficients between BMI and %BF by sex and ethnicity.................................................................................................................. 117

Table 3.6. The coefficients for the multiple regression between %BF as dependent variables and LnBMI and age as independent variables by sex and


15

ethnic groups.......................................................................................................... 118

Table 3.7. The coefficients for the stepwise multiple regression between %BF as dependent variables and LnBMI, age, and sex as independent variables by ethnic groups.......................................................................................................... 119

Table 3.8. The sensitivity, specificity and agreement rate of WHO, IOTF and Chinese BMI classifications for obesity .................................................................. 125

Table 3.9. Comparison of BMI cut‐off points for obesity proposed by WHO using Caucasian data with calculated BMI equivalents for Chinese, Lebanese, Malay, Filipino and Thai boys and girls derived from regression equations for predicting %BF from BMI. ...................................................................................... 126

Table 3.10. Number of participants for each variable by sex................................. 135

Table 3.11. Descriptive characteristics of the participants (mean±SD).................. 140

Table 3.12. Comparison of subcutaneous adiposity and fat distribution variables among four ethnic groups by sex (adjusted means±SE) .......................... 141

Table 4.1. Sample size of validation and cross‐validation group in each country by sex....................................................................................................................... 174

Table 4.2. Characteristics of participants ................................................................ 179

Table 4.3. Stepwise multiple regression models for FFM and TBW........................ 181

Table 4.4. Standardized coefficients of the final model for estimation of FFM and TBW.................................................................................................................. 182

Table 4.5. BIA prediction equations for FFM and TBW for each ethnic group........ 183

Table 4.6. BIA prediction equations for FFM and TBW for each BMI category....... 185

Table 4.7. Comparison of measured FFM with predicted FFM by ethnicity ........... 187

Table 4.8. Pearson correlation coefficients between measured and predicted FFM and between bias and mean FFM by ethnicity ............................................... 188

Table 4.9. Comparison of measured FFM with predicted FFM by BMI category.... 192

Table 4.10. Pearson correlation coefficients between measured and predicted FFM by ethnicity and BMI category ........................................................................ 193

Table 4.11. Comparison of measured TBW with predicted TBW by ethnicity ........ 197

Table 4.12. Pearson correlation coefficients between measured and predicted TBW and between bias and mean TBW by ethnicity .............................................. 198


16

Table 4.13. Comparison of measured TBW with predicted TBW by BMI category................................................................................................................... 202

Table 4.14. Pearson correlation coefficients between measured and predicted TBW by BMI category .............................................................................................. 203

Table 4.15. Bias of external BIA equations derived from Caucasian children for the assessment of FFM and TBW in Asian children ................................................ 206

Table 5.1. Characteristics of participants ............................................................... 217

Table 5.2. Prevalence of the metabolic syndrome and abnormalities among normal‐weight, overweight and obese children. ................................................... 219

Table 5.3. Changes in anthropometric and metabolic variables (292 boys, 277 girls) after 2‐year follow‐up (mean±SD)................................................................. 228

Table 5.4. Prevalence of total and central adiposity at baseline and 2 years later ........................................................................................................................ 229

Table 5.5. Partial Pearson correlations between BMI, WC, WHtR and %BF and metabolic risk factors at baseline and 2 years later .............................................. 230

Table 5.6. Stepwise multiple regression analysis model to explain the variance in metabolic risk factors using BMI, WC, %BF and WHtR age, sex, and pubertal stage as independent variables at baseline. ........................................... 231

Table 5.7. Stepwise multiple regression analysis model to explain the variance in metabolic risk factors at 2‐yr later using BMI, WC, %BF and WHtR at baseline, change of each indices, age, sex, and pubertal stage as independent variables. ................................................................................................................ 232

Table 5.8. The relative risk (95% CI) of developing metabolic abnormalities 2 years later among total or central adiposity based on level of BMI, WC, WHtR and %BF at baseline adjusted for age, gender and pubertal stage. ...................... 235

Table 5.9. Sample size and mean and SD for height, weight, BMI, WC, WHtR for Chinese children ............................................................................................... 245

Table 5.10. Waist circumference percentiles (cm) by age and gender .................. 246

Table 5.11. Waist‐to‐height ratio percentiles by age and gender .......................... 248

Table 5.12. Optimal waist circumference and waist‐to‐height ratio thresholds for CV risk factors clustering in boys (n=982) and girls (n=863)............................. 251

Table 5.13. Optimal age‐ and gender‐specific WC and WHtR cut‐off values for Chinese children ..................................................................................................... 252


17

LIST OF FIGURES

Figure 2.1. Prevalence of overweight in young people (aged 10‐16 y) in 2001‐2 in a range of countries ................................................................................. 36

Figure 2.2. Changes in body composition (a: total body fat; b: percent body fat; c: fat‐free mass; d: body mass index) between 8‐20 years................................ 46

Figure 2.3. Adjustments to be made in BMI to reflect equal levels of body fat compared to Caucasians of the same age and gender (mean, 95%CI).................... 74

Figure 2.4. The predicted levels of %BF from sex‐specific equations using age and BMI‐for‐age as independent variables expressed relative to levels among white children. ......................................................................................................... 80

Figure 2.5. The effect of relative leg length and frame size on the BMI‐%BF relationship. ............................................................................................................. 82

Figure 3.1. Scatter plots of %BF against BMI and %BF against LnBMI................... 114

Figure 3.2. Relationship between %BF and LnBMI of boys and girls for Chinese, Lebanese, Malay, Filipino and Thai population...................................................... 120

Figure 3.3. Relationship between %BF by deuterium dilution and LnBMI of Chinese, Lebanese, Malay, Filipino, and Thai boys and girls.................................. 123

Figure 3.4. Ethnic differences in age‐ and FM‐corrected biceps skinfold among Chinese, Lebanese, Malays and Thais by sex. ........................................................ 142

Figure 3.5. Ethnic differences in age‐ and FM‐corrected triceps skinfold among Chinese, Lebanese, Malays and Thais by sex. ............................................ 143

Figure 3.6. Ethnic differences in age‐ and FM‐corrected subscapular skinfold among Chinese, Lebanese and Malays by sex. ...................................................... 144

Figure 3.7. Ethnic differences in age‐ and FM‐corrected suprailiac skinfold among Chinese, Lebanese, Malays and Thais by sex. ............................................ 145

Figure 3.8 Ethnic differences in age‐ and FM‐corrected abdominal skinfold among Chinese and Malays by sex......................................................................... 145

Figure 3.9. Ethnic differences in age‐ and FM‐corrected abdominal skinfold among Lebanese and Malays by sex ...................................................................... 146

Figure 3.10. Ethnic differences in age‐ and FM‐corrected medial calf skinfold among Lebanese, Malays and Thais by sex............................................................ 147

Figure 3.11. Ethnic differences in age‐ and FM‐corrected trunk skinfold among Chinese, Lebanese, and Malays by sex. ..................................................... 148


18

Figure 3.12. Relationship between trunk skinfold thickness and total body fat of Chinese, Lebanese and Malays in each sex........................................................ 149

Figure 3.13. Ethnic differences in age‐ and FM‐corrected trunk/upper extremity ratio among Chinese, Lebanese, and Malays by sex.............................. 150

Figure 3.14. Relationship between trunk/upper extremity ratio and FM of Chinese, Lebanese, Malays and Thais in each sex.................................................. 151

Figure 3.15. Ethnic differences in age‐ and FM‐corrected suprailiac/upper extremity ratio among Chinese, Lebanese, Malays and Thais by sex. ................... 152

Figure 3.16. Relationships between suprailiac/upper extremity ratio and FM of Chinese, Lebanese, Malays and Thais in each sex. ............................................ 153

Figure 3.17. Ethnic differences in age‐ and FM‐corrected trunk/extremity ratio among Lebanese and Malays by sex.............................................................. 154

Figure 3.18. Relationships between trunk/extremity ratio and FM of Lebanese and Malays in each sex........................................................................................... 155

Figure 3.19. Ethnic differences in age‐ and trunk SFT‐corrected subscapular/suprailiac ratio among Chinese, Lebanese and Malays by sex. ......... 156

Figure 3.20. Relationship between subscapular/suprailiac ratio and FM of Chinese, Lebanese and Malays in each sex. ........................................................... 157

Figure 3.21. Ethnic differences in age‐ and BMI‐corrected waist circumference among Chinese, Lebanese, Malays and Thais by sex. ............................................ 158

Figure 3.22. Relationships between WC and BMI of Chinese, Lebanese, Malays and Thais in each sex. ................................................................................ 159

Figure 3.23. Age differences in sex‐ and FM‐corrected biceps skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. .........161

Figure 3.24. Age differences in sex‐ and FM‐corrected triceps skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 161

Figure 3.25. Age differences in sex‐ and FM‐corrected subscapular skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 162

Figure 3.26. Age differences in sex‐ and FM‐corrected suprailiac skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 162

Figure 3.27. Age differences in sex‐ and FM‐corrected abdominal skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 163

Figure 3.28. Age differences in sex‐ and FM‐corrected front thigh skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 163


19

Figure 3.29. Age differences in sex‐ and FM‐corrected medial calf skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 164

Figure 3.30. Age differences in sex‐ and FM‐corrected trunk skinfold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........ 164

Figure 3.31. Age differences in sex‐ and FM‐corrected trunk/upper extremity ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. .............................................................................................................. 165

Figure 3.32. Age differences in sex‐ and FM‐corrected suprailiac/upper extremity ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ........................................................................................ 165

Figure 3.33. Age differences in sex‐ and FM‐corrected trunk/extremity ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. ... 166

Figure 3.34. Age differences in sex‐ and trunk SFT‐corrected subscapular/suprailiac ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. .......................................................................... 166

Figure 3.35. Age differences waist circumference adjusted for BMI and sex using ANCOVA with Bonferroni multiple comparison procedure by ethnicity. ..... 167

Figure 4.1. Scatter plot of FFM measured by D2O against estimated FFM by BIA in Asian children. ............................................................................................. 186

Figure 4.2. Difference in FFM measured by D2O and estimated by BIA in Asian children. ................................................................................................................. 186

Figure 4.3. Difference in FFM measured by D2O and estimated by BIA in Chinese children..................................................................................................... 189

Figure 4.4. Difference in FFM measured by D2O and estimated by BIA in Lebanese children. ................................................................................................. 189

Figure 4.5. Difference in FFM measured by D2O and estimated by BIA in Malay children. ................................................................................................................. 190

Figure 4.6. Difference in FFM measured by D2O and estimated by BIA in Filipino children...................................................................................................... 190

Figure 4.7. Difference in FFM measured by D2O and estimated by BIA in Thai children. ................................................................................................................. 191

Figure 4.8. Difference in FFM measured by D2O and estimated by BIA in children with BMI z score <1 SD............................................................................. 194

Figure 4.9. Difference in FFM measured by D2O and estimated by BIA in children with BMI z score 1‐2 SD. .......................................................................... 194


20

Figure 4.10. Difference in FFM measured by D2O and estimated by BIA in children with BMI z score ≥2 SD............................................................................. 195

Figure 4.11. Scatter plot of measured TBW by D2O against predicted FFM by BIA in Asian children............................................................................................... 196

Figure 4.12. Difference in TBW measured by D2O and estimated by BIA in Asian children......................................................................................................... 196

Figure 4.13. Difference in TBW measured by D2O and estimated by BIA in Chinese children. .................................................................................................... 199

Figure 4.14. Difference in TBW measured by D2O and estimated by BIA in Lebanese children. ................................................................................................. 199

Figure 4.15. Difference in TBW measured by D2O and estimated by BIA in Malay children........................................................................................................ 200

Figure 4.16. Difference in TBW measured by D2O and estimated by BIA in Filipino children...................................................................................................... 200

Figure 4.17. Difference in TBW measured by D2O and estimated by BIA in Thai children................................................................................................................... 201

Figure 4.18. Difference in TBW measured by D2O and estimated by BIA in children with BMI z score <1 SD............................................................................. 204

Figure 4.19. Difference in TBW measured by D2O and estimated by BIA in children with BMI z score 1‐2 SD............................................................................ 204

Figure 4.20. Difference in TBW measured by D2O and estimated by BIA in children with BMI z score ≥2 SD. ............................................................................. 205

Figure 5.1. ROC curves for BMI, WC, WHtR and %BF at baseline for the prediction of metabolic risk clustering at baseline (a) and 2 years later (b) in boys and girls........................................................................................................... 233

Figure 5.2. Smoothed percentile curves for waist circumference in boys (n=2830) and girls (n=2699). ................................................................................... 247

Figure 5.3. Smoothed percentile curves for waist‐to‐height ratio in boys (n=2830) and girls (n=2699). ................................................................................... 249

Figure 5.4. ROC curves for waist circumference with higher CV risk factor clustering (≥3 of 5 CV risk factors) in Boys (n = 982) and girls (n = 863). ................ 250

Figure 5.5 ROC curves for waist‐to‐height ratio with high CV risk factors clustering ( ≥3 of 5 CV risk factors) in boys (n =982) and girls (n=863). .................. 250

Figure 5.6 The 50th percentiles for waist circumference in seven studies for


21

boys and girls........................................................................................................... 254

Figure 5.7 Mean of waist‐to‐height ratio for children in three studies. ................. 257


22

LIST OF ABBREVIATIONS

%BF Percent body fat

ANCOVA Analysis of covariance

ANOVA Analysis of variance

AUC Area under the curve

BIA Bioelectrical impedance analysis

BMC Bone mineral content

BMD Bone mineral density

BMI Body mass index

CI Confidence interval

CT Computed tomography

CV Cardiovascular

DBP Diastolic blood pressure

D2O Deuterium oxide

DXA Dual‐energy X‐ray absorptiometry

ECW Extracelluar water

FFM Fat‐free mass

FM Fat mass

HDL‐C High‐density lipoprotein cholesterol

IAAT Intra‐abdominal adipose tissue

ICW Intracelluar water

IDF International Diabetes Federation

IOTF International Obesity Task Force

IRMS Isotope ratio mass spectrometry

LDL‐C Low‐density lipoprotein cholesterol

MRI Magnetic resonance imaging

NCEP National Cholesterol Education Program

NCHS National Center for Health Statistics

NHANES National Health and Nutrition Examination Survey

OR Odds ratio


23

PE Pure error

R Resistance

RI Resistance index

RMSE Root mean square error

ROC Receiver‐operating characteristic

RR Relative risk

SAAT Subcutaneous abdominal adipose tissue

SAT Subcutaneous adipose tissue

SBP Systolic blood pressure

SD Standard deviation

SEE Standard error of the estimate

SFT Skin fold thickness

TBF Total body fat

TBW Total body water

TC Total cholesterol

TG Triglyceride

VAAT Visceral abdominal adipose tissue

VAT Visceral adipose tissue

WC Waist circumference

WHT Waist‐to‐hip ratio

WHtR Waist‐to‐height ratio

Xc Reactance

Z Impedance


24

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis has not been previously submitted for a degree or

diploma at any other higher education institution. To the best of my knowledge and

belief, the thesis contains no material previously published or written by another

except where due reference in made.

Signed: (Ailing Liu)

Date: July 5, 2011


25

ACKNOWLEDGEMENTS

I would like to take this opportunity to express my appreciation to all who supported

and assisted me to complete my PhD program. First of all, I wish to thank my

principal supervisor, Professor Andrew P. Hills, for his support, expert guidance and

encouragement in all occasions. I am a very lucky person and could not have asked

for a better PhD mentor during my study in Australia. I would like to also thank my

associate supervisor, Associate Professor Nuala M. Byrne, and external supervisor,

Professor Guansheng Ma, for their valuable and wise advice and support.

I wish to acknowledge the support from the project teams in five countries:

Professor Xiaoqi Hu and her team at National Institute for Nutrition and Food Safety,

Chinese Center for Disease Control and Prevention; Professor Lara Nasreddine and

her team at American University of Beirut, Lebanon; Professor Bin Mohamed Noor

Mohamed Ismail and his team at University Kebangsaan Malaysia; Professor Trinidad

Palad Trinidad and her team at Food and Nutrition Research Institute, Philippines;

Professor Kallaya Kijboonchoo and her team at Institute of Nutrition, Mahidol

University at Salaya, Thailand. Without their hard work on data collection in their

own country and their agreement for use of the data, the thesis would not have

been possible. Equally, I would like to thank all participants involved in the study for

their cooperation.

Special mention goes out to Dr Masaharu Kagawa and Ms Connie Wishart for their

great help to analyze the samples and sharing their knowledge as wonderful

teachers. Dr Kagawa also gave valuable comments on the thesis. I also wish to thank

the EMG group for sharing their knowledge with me, and all my friends in China and

Australia for their encouragement and help.

It is very important for me to take this opportunity to express my deepest

appreciation to my husband Yunchang, my daughter Ruitong and my parents.

Without their ongoing unconditional love, understanding and encouragement, I

would never have been to QUT for my PhD study and made it through.

Finally, I wish also to acknowledge the financial support for instruments and training


26

courses from the International Atomic Energy Agency.


27

CHAPTER 1 INTRODUCTION

1.1 Background

The increasing prevalence of obesity is a major public health problem in both the

developed and the developing world (Ashby, Ward, Mughal, & Adams, 2003; Li,

Schouten et al., 2008; Lobstein, Baur, & Uauy, 2004; Magarey, Daniels, & Boulton,

2001) and related to the increased risk of chronic diseases such as cardiovascular

disease and Type‐2 diabetes. In Asia, there is an alarming increase in the proportion

of overweight and obese children and adolescents with a parallel increase in the

incidence of chronic disease, especially in countries undergoing nutritional and

lifestyle transition (Li, Schouten et al., 2008; Li et al., 2005). Therefore, accurate and

accessible assessment of body composition is increasingly important for research

and clinical practice, especially as it relates to the monitoring of prevention and

treatment efforts. The most common definition of paediatric obesity utilises the

body mass index (BMI). However, there are a number of limitations regarding the

use of the BMI in children and adolescents.

One of the major limitations of the index is its inability to differentiate levels of

fatness and leanness among individuals (Freedman et al., 2005; Gallagher et al.,

1996; Roubenoff, Dallal, & Wilson, 1995). BMI measures the degree of excess weight

rather than total adiposity. As a measure of relative fatness it is inappropriate to use

universal BMI cut‐offs to define obesity across ethnic groups as there are ethnic

differences in the relationship between BMI and percent body fat (%BF)

(Deurenberg, Deurenberg‐Yap, & Guricci, 2002; Deurenberg, Deurenberg‐Yap, & van

Staveren, 1998; Freedman et al., 2008; Navder et al., 2009). The primary purpose of

defining obesity is to predict potential health risks. Using BMI cut‐offs developed

from Caucasian populations underestimates Asians who are at a high health risk.

Asian adults have 2‐4 units lower BMI at a given %BF compared with Caucasians

(Deurenberg‐Yap, Chew, & Deurenberg, 2002; Deurenberg et al., 1998; Gurrici,

Hartriyanti, Hautvast, & Deurenberg, 1998; Kagawa, Kerr, Uchida, & Binns, 2006).

Therefore, WHO (2004) proposed separate BMI cut‐offs of 23 kg/m2, 27.5 kg/m2,


28

32.5 kg/m2, and 37.5 kg/m2 for the Asian adult population. In Asian children and

adolescents, the ethnic difference in the relationship between BMI and %BF is

further complicated due to variations in body composition with age, sex, and

maturational level (Heymsfield, Lohman, Wang, & Going, 2005). Despite the recent

acceptance of some international classification systems of paediatric obesity based

on BMI including WHO (de Onis et al., 2007) and International Obesity Task Force

(IOTF) definitions (Cole, Bellizzi, Flegal, & Dietz, 2000), national variants still exist

(Group of China Obesity Task Force, 2004; Guillaume, 1999; Matsushita, Yoshiike,

Kaneda, Yoshita, & Takimoto, 2004). Some studies have reported the low sensitivity

with WHO and IOTF definitions to identify obesity in Asian children (Duncan, Duncan,

& Schofield, 2009; Wickramasinghe, Lamabadusuriya, Cleghom, & Davies, 2009).

Controversy regarding the use of BMI in children and adolescents makes it difficult

to monitor global and national trends, make comparisons between studies, stratify

for public health measures, and screen in clinical practice. To date, there is no

international consensus on a definition of obesity for Asian children as there is for

Asian adults due to the lack of evidence for an ethnic difference in the BMI‐%BF

relationship in this age group. Despite the existence of some studies on the

difference in BMI‐%BF relationship between Caucasian and Asian children and

adolescents (Deurenberg, Bhaskaran et al., 2003; Deurenberg, Deurenberg‐Yap, Foo,

Schmidt, & Wang, 2003; Duncan et al., 2009; Ehtisham, Crabtree, Clark, Shaw, &

Barrett, 2005; Freedman et al., 2008; Mehta, Mahajan, Steinbeck, & Bermingham,

2002; Navder et al., 2009), there are limitations to cross‐comparisons of these data

due to the use of different assessment methods. In short, despite confirmed

differences in the body composition of Asian adults from different origins

(Deurenberg‐Yap, Schmidt, van Staveren, & Deurenberg, 2000; Deurenberg et al.,

1998; Gurrici, Hartriyanti, Hautvast, & Deurenberg, 1999), limited data are available

in Asian children. Accordingly, the current thesis explored the ethnic difference in

BMI‐%BF relationship in Asian pre‐pubertal children from China, Lebanon, Malaysia,

The Philippines and Thailand using the same approaches and instruments for data

collection.

In addition, the inability of BMI to differentiate levels of fatness and leanness means


29

the index only provides a poor to fair identification of those who are truly

overweight and obese as determined from objective measures of body fat (Sampei,

Novo, Yuliano, & Sigulem, 2001; Wickramasinghe et al., 2005). This implies that

many children who are at risk of the health‐related problems accompanying

overweight and obesity may not be identified and therefore targeted for

intervention to treat or prevent the progression of obesity if based solely on BMI.

Moreover, BMI does not predict change in body fatness in intervention programs

which may result in misinterpretation of the body composition outcomes of these

programs (Dao et al., 2004). Therefore, more accurate measures of body fat

assessment should be used in addition to BMI and other anthropometric measures

(Wickramasinghe et al., 2005).

Numerous field and laboratory measurements of body composition have been used

in children and adolescents (Jürimäe & Hills, 2001). Of these, bioelectrical

impedance analysis (BIA) has been identified as one of the most appropriate options

for use in the field with this population as the measurement is fast, non‐invasive,

inexpensive, painless, requires minimal participant burden, does not require a high

level of technical skill, and may be used to estimate body composition across a

range of body composition. Numerous studies have contributed to the development

of equations to predict body composition from impedance in children (Nielsen et al.,

2007), and some BIA equations have been recommended (Heyward & Stolarczyk,

1996), including those of Houtkooper et al. (1992) for boys and girls 10‐19 y, and

Lohman (1992) or Kushner et al. (1992) for children younger than 10 y. However,

most equations have been generated from Caucasians and relatively few have been

developed for or cross‐validated with Asian children (Heyward & Stolarczyk, 1996;

Nielsen et al., 2007).

BIA measures may have the potential to more accurately predict %BF in children

than BMI provided that appropriate prediction equations are utilized (Deurenberg,

Deurenberg‐Yap, & Schouten, 2002). Population‐specific equations are likely to

systematically over‐ or underestimate body composition if they are applied to

individuals who do not belong to the same population subgroup due to the age, sex


30

and ethnic differences in body composition (Heymsfield et al., 2005). Some studies

have demonstrated the age, sex and ethnic effect on the relationship between

resistance and FFM or TBW (Deurenberg, Kusters, & Smit, 1990; Going et al., 2006;

Haroun et al., 2010; Lohman et al., 2000; Sluyter, Schaaf, Scragg, & Plank, 2010).

The current thesis aimed to develop and cross‐validate population‐specific BIA

equations for the estimation of TBW and FFM using the deuterium dilution

technique in Asian pre‐pubertal children from China, Lebanon, Malaysia, The

Philippines and Thailand to address the major need to characterise the increasing

prevalence of obesity in this region. The goal was to overcome the limitations of BMI

and anthropometry to discriminate between differences in body composition and to

be better placed to evaluate diet and physical activity intervention efforts.

There is a well known association between obesity and an increased risk of

co‐morbidities and all‐cause mortality (Kopelman, 2007; Lobstein et al., 2004).

However, the health risks of obesity can not be fully explained by BMI and total

adiposity (McAuley, Williams, Goulding, & Murphy, 2002). In contrast, body fat

distribution, especially central fat depots, is more closely associated with health risk

than overall adiposity (Berman et al., 2001; Ehtisham et al., 2005; Freedman,

Srinivasan, Harsha, Webber, & Berenson, 1989; Hara, Saitou, Iwata, Okada, & Harada,

2002; Misra et al., 2006; Okosun, 2000; Okosun, Liao, Rotimi, Prewitt, & Cooper,

2000; Savva et al., 2000).

Similar to the BMI‐%BF relationship, body fat distribution is also ethnic‐dependent

(Chandalia, Abate, Garg, Stray‐Gundersen, & Grundy, 1999; He et al., 2002; Lear et

al., 2007; McKeigue, Shah, & Marmot, 1991; Novotny, Daida, Grove, Marchand, &

Vijayadeva, 2006; Park, Allison, Heymsfield, & Gallagher, 2001; Raji, Seely, Arky, &

Simmons, 2001; Wu et al., 2007). Moreover, the ethnic difference in body fat

distribution is a contributor to the ethnic difference in prevalence of obesity‐related

health risk among different ethnic groups (Berman et al., 2001; Daniels, Khoury, &

Morrison, 1997; Ehtisham et al., 2005; Freedman et al., 1989; He et al., 2002;

Okosun, 2000; Okosun et al., 2000). Some studies have reported that Asian children

and adolescents have greater trunk fat depots than Caucasians (Ehtisham et al.,


31

2005; He et al., 2002; Malina, Huang, & Brown, 1995; Novotny et al., 2006) however

a limited number of studies have explored the difference in body fat distribution

among Asian children from different origins. Therefore, to enable a better

estimation of the obesity‐related health risks in the Asian region, a better

understanding of the body fat distribution of Asians is required. Accordingly, another

of the aims of the current thesis was to determine the ethnic difference in body fat

distribution of Asian pre‐pubertal children from China, Lebanon, Malaysia and

Thailand.

In addition, the use of different methods of body composition assessment

influences studies on obesity‐related health problems, including the paediatric

metabolic syndrome. Despite obesity being a central feature of the metabolic

syndrome in both children and adults (Alberti, Zimmet, & Shaw, 2005; Cook,

Weitzman, Auinger, Nguyen, & Dietz, 2003; de Ferranti et al., 2004; Expert Panel on

Detection, 2001; Weiss, Dziura, & Burgert, 2004; Zimmet et al., 2007), the various

obesity indices such as the BMI (Lambert et al, 2004; Weiss et al., 2004) and WC

(Cook et al., 2003; de Ferranti et al., 2004; Zimmet et al., 2007) used to predict

metabolic risk, are contradictory. This has contributed to different prevalence data

(Golley, Magarey, Steinbecks, Baur, & Daniels, 2006; Goodman et al, 2004; Reinehr,

de Sousa, Toschke, & Andler, 2007; Weiss et al., 2004) which makes it impossible to

estimate the global prevalence of the metabolic syndrome and to make valid

comparisons between countries. While the choice between BMI and WC as simple

markers remains a matter of ongoing debate, some studies have indicated that

other obesity indices may be better predictors of cardiovascular disease risk, for

example the waist‐to‐height ratio (WHtR) (Hara et al., 2002; Savva et al., 2000), %BF

(Moussa, Skaik, Selwanes, Yaghy, & Bin‐Othman, 1994), and skinfold thickness (SFT)

(Maffeis, Pietrobelli, Grezzani, Provera, & Tato, 2001; Misra et al., 2006; Teixeira,

Sardinha, Going, & Lohman, 2001). Ethnic differences in body composition

(Heymsfield et al., 2005) and the susceptibility to cardiovascular risk and insulin

resistance (Misra, Khurana, Vikram, Goel, & Wasir, 2007) might explain, in part, the

inconsistency in obesity index recommended by different studies. However, given

that Asians differ in body composition from Caucasians, there is limited published


32

data regarding associations between obesity indices and metabolic risk factors in

Asian children. Moreover, body composition in children varies with age, sex and

pubertal status (Heymsfield et al., 2005). Accordingly, it is more challenging to

explore relationships between obesity indices and metabolic risk factors in children

than adults. However, although previous studies have reported a relationship

between obesity indices and metabolic risk factors in different age groups (Hara et

al., 2002; Hsieh & Muto, 2005; Janssen, Katzmarzyk, & Ross, 2004; Lee, Huxley,

Wildman, & Woodward, 2008; Misra et al., 2006; Savva et al., 2000; Watts, Bell,

Byrne, Jones, & Davis, 2008), most have been cross‐sectional rather than

longitudinal and therefore not considered the influence of growth. Therefore, the

International Diabetes Federation (IDF) has recommended that further investigation

of how obesity is defined in children should be undertaken, for example considering

the respective value of measures such as WHtR, WC, etc (Zimmet et al., 2007).

Accordingly, the current thesis aimed to determine which obesity index (among BMI,

%BF, WC, WHtR) was the best predictor of metabolic risk factors clustering across a

2‐y period among Chinese children.

Besides the selection of an appropriate obesity index for use in the definition of the

metabolic syndrome, ethnic‐specific criteria of obesity is another important issue

which should be considered in the development of a universal definition in children

and adolescents (Misra et al., 2007). This is due mainly to the ethnic differences in

body composition (Heymsfield et al., 2005) and susceptibility to develop various

cardiovascular risk factors and the metabolic syndrome (Misra et al., 2007). In order

to compensate for the variation in child development and ethnic origin, percentiles,

rather than absolute values of WC have been used in the IDF definitions ( Zimmet et

al., 2007). However, compared with the BMI, relatively few studies have referenced

a threshold cut‐off for WC and WHtR to define obesity in different ethnic

populations. A small number of studies have referenced WC cut‐off points for

predicting cardiovascular risk factors in different ethnic populations (Maffeis et al.,

2001; Sung et al., 2007), but even in the same population, the thresholds are

inconsistent (Meng et al., 2007; Ng et al., 2007; Sung et al., 2007). Accordingly, the

percentile used as a cut‐off for WC should be reassessed when more data are


33

available and the development of ethnic‐specific age and sex normal ranges for WC

based on healthy values (Zimmet et al., 2007). In addition, WHtR is regarded as

more practical to employ in field settings than WC, due to the cut‐off (WHtR = 0.5)

for the prediction of cardiovascular risk being age‐ and sex‐independent (Ashwell,

LeJeune, & McPherson, 1996). However, some studies have challenged the universal

cut‐off for children of different age and gender (Kelishadi et al., 2007). To date, two

studies have reported age‐ and gender‐specific WC percentiles and cut‐offs in

Chinese children and adolescents, one in Hong Kong Chinese children and

adolescents (Sung et al., 2007), and the second in children from Xinjiang province

(Yan et al., 2008). No study has reported age‐ and gender‐specific cut‐offs for WHtR

in Chinese children. As there are well documented regional differences in the body

composition of Chinese, for example those living in the North and South of the

country (China's National Group on Student's Constitution and Health Survey, 2007),

the development of WC and WHtR percentiles and cut‐offs for different groups

would be particularly valuable. Therefore, we aimed to develop percentiles of WC

and WHtR for Chinese children aged 6‐12 y in a representative sample from three

cities in China and proposed WC and WHtR cut‐offs for the prediction of

cardiovascular risk.

1.2 Research questions

Accordingly, the thesis consists of three studies to meet the gaps in the literature

outlined mentioned above. The research questions for each study are listed below.

Chapter 3

A‐ Do ethnic differences exist in the BMI‐%BF relationship in children from different

Asian countries?

B‐ Are there ethnic differences in body fat distribution among Asian children from

different backgrounds?

C‐ Which BMI cut‐off for screening the obese child is the more appropriate for

Asian children: the IOTF definition, WHO or population‐specific values?

Chapter 4


34

A‐ How useful is bioelectrical impedance analysis (BIA) in predicting TBW and FFM

in Asian children using the deuterium dilution technique as the criterion

measure?

Chapter 5

A‐ Do obese Chinese children have a higher risk of the indicators of metabolic

syndrome than normal‐weight children?

B‐ Which obesity index is the best predictor of metabolic risk in Chinese children,

%BF, BMI, WC or WHtR?

C‐ Which threshold as a cut‐off for WC and WHtR, best predicts the clustering of

metabolic risk factors in Chinese children?


35

CHAPTER 2 LITERATURE REVIEW

2.1 CHILDHOOD OBESITY WORLDWIDE

The prevalence of obesity in all age groups poses such a serious problem that the

World Health Organization (WHO) has described it as a ‘global epidemic’ (WHO,

2000). Worldwide, one in 10 children aged 5‐17 years of age is overweight, a total of

155 million, of which an estimated 30‐45 million are obese. There are an estimated

14 million overweight school‐age children in the European Union and of these, three

million are obese (Lobstein et al., 2004). The number of overweight children has

been predicted to be rising by around 400,000 per annum, of whom 85,000 are

obese (The BMA Board of Science, 2005). In the US, overweight is also a serious

health concern for children and adolescents. Data from two National Health and

Nutrition Examination Surveys (NHANES) (1976‐1980 and 2007‐2008) show that the

prevalence of overweight is increasing: for children aged 2‐5 years, prevalence

increased from 5.0% to 10.4%; for those aged 6‐11 years, the increase was from

6.5% to 19.6%; and for those aged 12‐19 years, the increase was from 5.0% to 18.1%

(Ogden & Carroll, 2010). In Australia, the prevalence of overweight in children aged

7‐15 years almost doubled between 1985 and 1995 (from 10.7% to 19.5% for boys

and from 11.8% to 21.1% for girls), with an additional 1% of all children becoming

overweight every year during the last two decades (Magarey et al., 2001). In Japan,

data from the National Nutrition Survey shows that the prevalence of obese boys

and girls increased from 6.1% and 7.1%, respectively, in the time period 1976‐1980

to 11.1% and 10.2% in 1996‐2000 (Matsushita et al., 2004).

The increase in obesity is not confined to industrialized countries but is also

prevalent in many developing countries, especially in Asia, which are undergoing a

major nutritional transition (WHO, 2000). In China, data from two national Nutrition

and Health Surveys (1982, 2002) show that the prevalence of overweight defined by

the IOTF criteria increased from 1.4% to 5.3% for children aged 7‐17 years. The

number of overweight youngsters was 12 million in China in 2002 (Li, Schouten et al.,

2008). In Brazil between 1974 and 1997, the prevalence of overweight and obesity


36

(also using IOTF definitions) among young people aged 6‐17 years more than tripled

(increasing from 4.1% to 13.9%) (Lobstein et al., 2004).

4.4

5.3

7.0

7.4

7.8

9.8

9.7

10.0

10.3

11.3

11.8

13.1

14.8

14.8

13.3

16.3

15.2

16.7

18.3

1.7

1.6

2.2

2.4

2.5

3.0

2.2

2.5

5.1

2.5

4.1

4.8

6.8

1.4

1.3

1.1

0.8

0.8

0.9

0.0 5.0 10.0 15.0 20.0 25.0

China(mainland)*

Russia

Netherlands

Poland

Switzerland

Sweden

Germany

France

Norway

Ireland

Fenlind

Scotland

Greece

Italy

England

Spain

Canada

Wales

United States

Overweight prevalence (%)

Pre-obese

Obese

Figure 2.1. Prevalence of overweight and obese in young people (aged 10‐16 y) in 2001‐2 in a range of countries (* for China, data in young people aged 7‐17 years in 2002)

Childhood obesity is a risk factor for a number of chronic diseases in adult life

including heart disease, some cancers and osteoarthritis. Some diseases however,

can become manifest during childhood, particularly type 2 diabetes (The BMA Board

of Science, 2005). The dramatic increase in the prevalence of childhood overweight

and obesity and its resultant co‐morbidities is associated with significant health and

financial burden and warrants strong and comprehensive efforts at prevention.

Meanwhile, the limited success of adult obesity intervention programs has led to


37

increased interest in prevention programs aimed at children. However, the limited

availability of sound and convenient screening tools, plus effective evaluation

approaches has been a problem in devising programs for obesity intervention and

management in children.


38

2.2 DEFINITION OF OVERWEIGHT AND OBESITY IN CHILDREN

The primary purpose for defining overweight and obesity is to predict health risks

and to provide comparisons between populations and to develop intervention

programs. Obesity is characterized by an increased amount of body fat. Therefore,

strictly speaking, obesity should be diagnosed based on the amount of body fat, for

example %BF instead of BMI. Alternatively, weight‐for‐height Z scores or other

anthropometric measures should be used as it is not the degree of excess weight (as

measured by BMI), but the degree of body fatness that is important as a risk factor.

However, many of the methods available for in vivo assessment of %BF are either

expensive or time consuming therefore various anthropometric measures have been

widely used in epidemiological studies as indirect measures of obesity, including BMI

(Ogden, Flegal, Carroll, & Johnson, 2002; Rolland‐Cachera et al., 2002),

weight‐for‐height ratios (Leung, Lau, Tse, & Oppenheimer, 1996; World Health

Organization, 1995), and others (Table 2.1 ).


39

Table 2.1. Indices and cut‐off points used in different countries to define obesity

Country/ Organization BMI (Percentile)

BMI (Z scores)

Weight/height(Percentile)

Weight‐for‐ Height (%)

Weight/ideal weight (%)

Skinfold thickness (Percentile)

Europe Belgium > 97th > 120 Finland > 90th > 90th France > 90th Greece > 90th Hungary > 90th > 90th Italy NA > 120 Netherlands > 97th United Kingdom > 90th North and South America United States (whites) > 95th > 85th >120 Canada > 95th Argentina > 120 Chile > 120 Venezuela > 120 Asia‐Pacific region Australia > 85th

> 120

China ‐ Mainland1 > 95th

‐ Hong Kong > 120 ‐ Taiwan > 120 Japan > 90th > 120 Singapore > 120 Saudi Arabia > 95th > 120 Thailand > 120 WHO2 > 95th (HANES）

WHO3 (5‐19 years ) > +2SD ( ≥ 30 kg/m2 at age 19)

IOTF4 (2‐18 years) ≥ 30 kg/m2 at age 18

Adapted from Guillaume (1999) except 1 from Group of China Obesity Task Force (2004), 2 from Must (1991), 3 from de Onis (2007), and 4 from Cole (2000).


40

BMI has frequently been used for evaluation of the nutritional status of individuals,

including those focusing on obesity because it is a convenient measure to perform in

both field and clinical settings. BMI can be used as a crude reference of relative

fatness. A cut‐off point of 25 kg/m2 is recognized internationally as the definition of

adult overweight, and ≥30 kg/m2 for adult obesity by WHO (World Health

Organization, 1995). Despite its widespread use in adults, the ability of BMI to

identify obesity in children and adolescents is somewhat limited due to the

differences in the rates of maturation and the effects of maturation on body

composition (Cole, Freeman, & Preece, 1995). More appropriate cut‐off points

related to age and gender are needed to define child and adolescent obesity.

The reference population is one of the most important issues when establishing a

definition of childhood obesity and a representative sample of the whole population

should be used. Accordingly, some countries have developed their own BMI‐for‐age

reference charts based on nationally representative survey data, including the well

known US National Center for Health Statistics reference (Must et al., 1991), the UK

(Cole et al., 1995), French (Rolland‐Cachera et al., 1991), and Chinese references

(Group of China Obesity Task Force, 2004). Moreover, choice of the appropriate

index suitable to define obesity is another important issue. Most of the existing

definitions are based on the same principle at different ages by using reference

centiles. For example, in the United States, the 85th and 95th centiles of BMI‐for‐age

and sex based on nationally representative survey data have been recommended as

cut‐off points to identify overweight and obesity (Barlow & Dietz, 1998). This was

recommended by a WHO Expert Committee for international use to define at risk of

overweight for adolescents aged 10‐19 years (World Health Organization, 1995).

However, for wider international use, the cut‐off points have been criticized as

arbitrary considering centiles and the reference population. A centile cut‐off point

could in theory be identified as the point on the distribution of BMI where the

health risk of obesity starts to rise steeply. Unfortunately, such a point cannot be

identified with any precision: children have less disease related to obesity than

adults, and the association between child obesity and adult health risk may be


41

mediated through adult obesity, which is associated both with child obesity and

adult disease. Therefore, it is proposed that the adult cut‐off points should be linked

to BMI centiles for children to provide cut‐off points for children (Cole et al., 2000).

According to these points, some international standards to define childhood

overweight/obesity which link BMI centiles for children to adult cut‐off points, have

been developed.

The IOTF, using a worldwide representative sample of children (pooling data from

Brazil, Great Britain, Hong Kong, the Netherlands, Singapore, and the United States,

and representing 97,876 males and 94,851 females from birth to 25 years age) as

the reference population, derived age‐ and gender‐specific overweight and obesity

cut‐off points for young people aged 2‐18 years, which are equivalent to a BMI of 25

kg/m2 and 30 kg/m,2 respectively at 18 years of age (Cole et al., 2000). Although the

cut‐off points were developed using several data sets, the authors acknowledged

that the reference data set may not adequately represent non‐Western populations.

The WHO recommended the US NCHS BMI‐for‐age centiles for international use to

define at risk of overweight for adolescents aged 10‐19 years in 1995 (World Health

Organization, 1995). In 2007, WHO published a growth reference for school‐aged

children and adolescents (5‐19 years) (de Onis, 2007). The new curves are closely

aligned with the WHO Child Growth Standards at 5 years. For BMI‐for‐age across all

centiles, the magnitude of the difference between the two curves at age 5 years is

mostly 0.0 kg/m2 to 0.1 kg/m2. At 19 years of age, the new BMI values at +1

standard deviation (SD) are 25.4 kg/m2 for boys and 25.0 kg/m2 for girls. These

values are equivalent to the overweight cut‐off for adults (≥ 25.0 kg/m2). Similarly,

the +2SD values (29.7 kg/m2 for both sexes) compare closely with the cut‐off for

obesity (≥ 30.0 kg/m2). But the BMI‐for‐age curves were developed from only three

data sets, the Health Examination Survey Cycle II (6‐11 years) and Cycle III (12‐17

years), and the NHANES Cycle (birth to 74 years). Further validation is needed for

Asian and other populations.

Although these standard definitions were intended to overcome arbitrary


42

definitions of overweight and obesity and to provide international standards of

overweight and obesity prevalence across countries, the validity studies using these

BMI cut‐off points to identify children with excess adiposity have generally

documented low to fair identification of those who are truly overweight, as

determined from %BF. Li et al. (2003) validated the IOTF and NCHS definition of

overweight among Chinese children against %BF measured by BODPOD. The

sensitivity was 52.8% for the IOTF definition and 50% for the NCHS definition.

Wickramasinghe et al. (2005) also showed that none of the BMI cut‐offs from IOTF,

CDC/NCHS percentile charts and BMI‐Z was sensitive enough to detect cases of

obesity in either Australian white Caucasian or Australian Sri Lankan children.

Further, Gaskin et al. (2003) indicated that although having high specificity for

determining overweight among black Jamaican children, the IOTF BMI cut‐off points

had low sensitivity (2‐86%) to identify overweight defined by NHANES reference for

SSF, TSF and SF (> 85th percentile), extremely low at age 7‐8 years compared with at

age 11‐12 years (Gaskin & Walker, 2003). Moreover, even in Caucasian children and

adolescents, the sensitivity of the IOTF definition is still low, about 48% and 62% in

Swiss boys and girls aged 6‐12 years (Zimmermann, Gubeli, Puntener, & Molinari,

2004) and only 22‐25% for girls and 72‐84% for boys aged 17 y (Neovius, Linne,

Barkeling, & Rossner, 2004). These findings imply that many children who are at risk

for the health‐related problems accompanying overweight and obesity may not be

detected and therefore effectively targeted for intervention programs designed to

treat or prevent the progression of obesity when selection for such programs is

based solely on BMI (Maynard et al., 2001).

There are two potential reasons to explain the low sensitivity of the BMI

classification system to detect cases of childhood obesity. One is the inability of BMI

to differentiate levels of fatness and leanness among individuals (Freedman et al.,

2005; Gallagher et al., 1996; Roubenoff et al., 1995). Despite high correlations

between BMI and body fat and %BF, BMI is also correlated with FFM. In children,

these relationships between BMI and the fat and fat‐free components of the body

are further complicated by varying growth rates and maturity levels (Guo, Chumlea,

Roche, & Siervogel, 1997; He et al, 2004). BMI can explain 41‐88% and 14‐81% of


43

the variance in %BF or total body fat (TBF) for boys and girls aged 8‐18 years,

respectively (Maynard et al., 2001). Deurenberg et al. (1991) indicated a rather low

(0.38‐0.42) variance of %BF explained by BMI in children compared with that of

adults (0.79‐0.80). Therefore, it is better to use other methods to assess body fat to

identify overweight and obese children in field work in addition to BMI. The second

reason relates to the significant ethnic differences in body composition and the

relationship between BMI and %BF (Chung et al., 2005; Deurenberg‐Yap et al., 2000;

Deurenberg et al., 1998; Duncan et al., 2009; Freedman et al., 2008; Jackson et al.,

2002; Malina, 2005; Navder et al., 2009; Rush, Freitas, & Plank, 2009). Body weight

is the sum of fat, muscle, visceral organs, and bone. Therefore, individuals with long

trunks and short legs for height have a higher BMI regardless of their body fat

content and it is well known that the length of the trunk and limbs relative to height

varies by race (Craig, Halavatau, Comino, & Caterson, 2001; Deurenberg,

Deurenberg‐Yap, Wang, Lin, & Schmidt, 1999; Gurrici et al., 1999). These ethnic

differences in body composition indicate the necessity to consider different BMI

levels to define obesity in different ethnic or population groups. The definitions of

obesity based primarily on the relationship of weight with adiposity in white

populations may not adequately reflect the body composition of other racial or

ethnic populations. Therefore, WHO proposed additional BMI cut‐off points of 23

kg/m2, 27.5 kg/m2, 32.5 kg/m2 and 37.5 kg/m2 for the Asian adult population in

2004, which are lower than for the Caucasian or Europid population (WHO, 2004).

However for children, a clear ethnic difference in the relationship between BMI and

%BF is yet to be established. This makes it difficult to develop a universal obesity

classification based on BMI and explains why significant national variance exists

(Table 2.1).


44

2.3 AGE, SEX AND ETHNIC DIFFERENCES IN BODY COMPOSITION IN CHILDREN

Many factors can influence body composition, including age, gender, ethnicity, and

the environment. In addition to the variability in body composition value derived

from different assessment techniques, there are distinct age, gender and ethnic

differences in body composition in children and adolescents (Heymsfield et al.,

2005).

2.3.1 Age differences in body composition in children

Changes in body composition with age begin at the moment of conception and

continue throughout life. Growth and development during the formative years of

childhood is one of the key phases of life for substantial changes in body

composition.

2.3.1.1 Total body fat

Fat mass (FM) FM is the most variable component of body composition.

Between‐individual variability ranges from about 6% to more than 60% of total body

weight. Within‐individual variability can also be considerable but changes with age

in individuals tend to “track” and follow distinct trajectories over time.

Both cross‐sectional and longitudinal studies (Chumlea et al., 2002; Fomon, Haschke,

Zeigler, & Nelson, 1982; Goulding, Taylor, Jones, Lewis‐Barned, & Williams, 2003;

Guo et al., 1997) have indicated that FM increased with age during the growth and

development period. For example, Fomon et al. (1982) described the body

composition of reference children from birth to age 10 years. At birth, TBF was 486 g

and 495 g for boys and girls, respectively. This increased to 2656 g and 2757 g for

boys and girls at the age of 4 years, and continued to increase to 4318 g and 6318 g

for boys and girls by 10 year of age (Table 2.2). Goulding et al. (2003) indicated in a

longitudinal study that between 4‐5 y and 8‐9 y, FM increased by 124% in 23 New

Zealand girls. Guo et al. (1997) also followed changes in body composition in 130

Caucasian males and 114 Caucasian females. The results showed that FM continued

to increase between 8 and 20 y in both males and females. The average increase in


45

FM for females was from 6.4 kg at the age of 8 y to 16.3 kg at 20 y and from 4.7 kg

to 9.7 kg for males (Figure 2.2 a).

Table 2.2. Body composition of reference children

FM %BF FFM TBW Hydration Protein Protein/FFMAge

(g) (%) (g) (g) (%) (g) (%)

Boys

Birth 486 13.7 3059 2466 80.6 457 15

1 mon 671 15.1 3781 3044 80.5 574 15.1

2 mon 1095 19.9 4414 3544 80.3 678 15.4

3 mon 1495 23.2 4940 3952 80 772 15.6

4 mon 1743 24.7 5317 4248 79.9 840 15.8

5 mon 1913 25.3 5662 4513 79.7 901 15.9

6 mon 2037 25.4 5993 4770 79.6 964 16

9 mon 2199 24 6981 5536 79.3 1138 16.4

12 mon 2287 22.5 7863 6212 79 1309 16.6

18 mon 2382 20.8 9088 7134 78.5 1548 17.1

24 mon 2456 19.5 10134 7915 78.1 1763 17.4

3 yr 2576 17.5 12099 9377 77.5 2157 17.8

4 yr 2656 15.9 14034 10806 77 2554 18.2

5 yr 2720 14.6 15950 12218 76.6 2950 18.5

6 yr 2795 13.5 17895 13654 76.3 3352 18.7

7 yr 2931 12.8 19919 15119 75.9 3770 18.9

8 yr 3293 13 22007 16659 75.7 4200 19.1

9 yr 3724 13.2 24406 18402 75.4 4726 19.3

10 yr 4318 13.7 27122 20369 75.1 5282 19.5

Girls

Birth 495 14.9 2830 2281 80.6 426 15

1 mon 668 16.2 3463 2788 80.5 525 15.2

2 mon 1053 21.1 3936 3157 80.2 609 15.5

3 mon 1366 23.8 4377 3497 79.9 689 15.8

4 mon 1585 25.2 4715 3758 79.7 750 15.9

5 mon 1769 26 5031 4000 79.5 809 16.1

6 mon 1915 26.4 5335 4236 79.4 870 16.3

9 mon 2066 25 6204 4901 79 1034 16.6

12 mon 2175 23.7 7005 5520 78.8 1184 16.9

18 mon 2346 21.8 8434 6612 78.4 1455 17.2

24 mon 2433 20.4 9477 7411 78.2 1655 17.4

3 yr 2606 18.5 11494 8954 77.9 2030 17.7

4 yr 2757 17.3 13203 10259 77.7 2362 17.9

5 yr 2949 16.7 14711 11416 77.6 2649 18

6 yr 3208 16.4 16312 12642 77.5 2967 18.1

7 yr 3662 16.8 18178 14052 77.3 3320 18.3

8 yr 4319 17.4 20521 15842 77.2 3776 18.4

9 yr 5207 18.3 23253 17928 77.1 4297 18.5

10 yr 6318 19.4 26232 20172 76.9 4883 18.7

From Fomon et al. (1982)


46

Figure 2.2. Changes in body composition (a: total body fat; b: percent body fat; c: fat‐free mass; d: body mass index) between 8‐20 years. (Guo et al., 1997)

Percent body fat Infants have on average about 10 to 17% fat at birth (Ma et al.,

2004). This increases to about 30% by 6 months of age and then begins to gradually

decline during early childhood (Butte, Hopkinson, Wong, Smith, & Ellis, 2000).

During mid‐childhood, between about 5 and 8 years of age, a pre‐adolescent “fat

wave”, or “adiposity rebound”, occurs (Fomon et al., 1982; Rolland‐Cachera et al.,

1991) and about a 31% increase was found in girls between 4‐5 y and 8‐9 y

(Goulding et al., 2003). %BF then continues to increase in girls, as shown by Guo et

al. (1997) in a longitudinal study, from an average of 20 to 26% between 9 y and 20 y.

In contrast for boys, %BF increased from 15 to 17% between 9 and 13 y then

decreased to 13% at the age of 20 years as FFM rapidly increased (Figure 2.2 b).

Changes with age in body fatness suggest that early infancy, the mid‐childhood

“adiposity rebound”, and adolescence were “critical periods” for the development of

obesity (Dietz, 1994).

0

10

20

30

40

50

60

70

8-10 y 10-12 y 12-14 y 14-16 y 16-18 y 18-20 y

FFM

(kg)

Males Females

10

12

14

16

18

20

22

24

8-10 y 10-12 y 12-14 y 14-16 y 16-18 y 18-20 y

BM

I(kg

/m2)

Males Females

0

5

10

15

20

25

30

8-10 y 10-12 y 12-14 y 14-16 y 16-18 y 18-20 y

PBF(%)

Males Females

d

0

2

4

6

8

10

12

14

16

18

8-10 y 10-12 y 12-14 y 14-16 y 16-18 y 18-20 y

TB

F(kg

)

Males Females

b

c

a


47

Fat distribution It has been recognized that fat distribution, in addition to total body

fat, is a risk factor for cardiovascular disease in both children and adults. The term

distribution refers to the absolute or relative amount of adipose tissue in different

regions or compartments of the body. Total adipose tissue comprises subcutaneous

adipose tissue (SAT) and visceral adipose tissue (VAT).

Skin fold thickness at different body sites is one of the most common measures of

SAT. The sum of skin fold thickness at triceps, subscapular, abdominal, suprailiac,

chest, thigh and mid‐axilliary sites has been reported as 67 mm for boys aged 11

years, declining to 57 mm by the age of 14 years and subsequently increasing to 90

mm by 17 years of age. The sum of the seven skin folds increases gradually with age

in girls, from 77 mm at 11 years of age to 127 mm at 17 years of age (Tahara et al,

2002). However, changes in skin fold thickness at different sites show variable

patterns. For example, triceps skin fold thickness (extremity) and subscapular (trunk)

thickness both increase with age in girls, as well as subscapular skinfold thickness in

boys. Skin fold thicknesses on the extremities decrease in boys between 7 y and 13

y and then increase after 13 years (China's National Group on Student's Constitution

and Health Survey, 2000).

VAT or intra‐abdominal adipose tissue (IAAT), and subcutaneous abdominal adipose

tissue (SAAT), are two discrete compartments of abdominal fat which are often

measured using MRI or computed tomography (CT). Currently, more emphasis is

placed on the distribution of visceral and subcutaneous adipose tissue in the

abdominal region because many studies have suggested that abdominal fat,

especially VAT, is highly related to a wide range of health indicators in both adults

(Bjorntorp, 1992) and children (Huang, Johnson, Figueroa‐Colon, Dwyer, & Goran,

2001). Huang et al. (2001) followed changes in IAAT and SAAT in 138 children aged

8.1±1.6 years over a 3‐5 year period. The increase in IAAT was 5.2±2.2 cm2/yr, but

the growth of SAAT was not significant. Fox et al. (2000) indicated that VAT and

SAAT increased by 69.1% (from 17.8 to 30.1 cm2) and 19.1% (from 78.1 to 93.0 cm2)

for boys, and 48.4% (from 25.8 to 38.3 cm2) and 78.1% (from 75.2 to 133.9 cm2) for

girls between 11 and 13 years of age. WC also increased by 9.7 mm for boys and 7.5


48

mm for girls.

2.3.1.2 Fat‐free mass

FFM increases with age regardless of sex and ethnicity. At birth, the average FFM in

boys and girls is 3059 g and 2830 g, respectively, increasing to 5993 g and 5335 g by

6‐months of age. In boys and girls aged 10 years, FFM is 27.1 kg and 26.2 kg,

respectively (Fomon et al., 1982) (Table 2.2). Longitudinal studies have confirmed

the changes in FFM with age, for example Goulding et al. (2003) reported that FFM

increased by 55% between 4‐5 y and 8‐9 y in girls, while Guo et al. (1997) indicated

that the average amount of FFM increased from 24.5 kg at 8 y to 60.2 kg at 20 y of

age in males and from 23.1 kg to 44.0 kg in females (Figure 2.2 c).

There are also important age‐related changes in the composition of FFM. The main

molecular components of the FFM are water, protein, osseous and non‐osseous

mineral, and glycogen. At the tissue‐system level, the FFM is comprised of skeletal

muscle and non‐skeletal muscle, organ, connective tissue and bone. The proportions

of these components are known to vary systematically with age (Heymsfield et al.,

2005). This variation is important because it affects assumptions underlying some in

vivo measurement methods, but also because it has health and functional

significance.

Total body water The absolute amount of TBW increases with age during growth

and development (Chumlea et al., 2002; Chumlea, Schubert, Reo, Sun, & Siervogel,

2005; Fomon et al., 1982). The average TBW is 2466 g in boys and 2281 g in girls at

birth, 6212 g in boys and 5520 g in girls at 1 year of age, and 12.2 kg in boys and

11.4 kg in girls at 5 years (Fomon et al., 1982) (Table 2.2). Chumlea et al. (2005)

followed changes in TBW in 124 white boys and 116 white girls and indicated that

mean TBW volumes in boys were 16 L at 8 years of age, and this amount doubled by

age 14 years, and was +40 L by 16 years of age. In girls, mean volumes are about 15

L at 8 years increasing up to about 29 L by 17 years of age (Table 2.3). The mean

percentages for TBW/FFM, namely hydration of the FFM, is higher in infants and

young children, (as much as 80% on average), and decreases to about 73% by about


49

10 to 15 y of age (Ellis, 1990; Fomon et al., 1982; Wang et al., 1999). This change is

accompanied by corresponding increases in the proportions of protein and mineral

in the FFM, and a consequent increase in its density.

Table 2.3. Total body water volume (L) for age and gender

Age (yr) Males Females

8 16.2±2.0 14.8±1.9

9 18.5±2.4 16.8±3.4

10 19.5±2.8 19.7±3.5

11 21.5±3.2 21.2±4.2

12 24.2±5.4 24.9±4.3

13 29.2±6.6 25.8±3.9

14 32.6±6.8 27.5±3.7

15 36.7±7.0 27.6±3.8

16 40.6±7.7 28.4±4.1

17 40.7±6.8 29.0±3.5

18 41.3±6.5 29.1±4.6

19 43.1±8.5 29.4±3.2

20 42.0±5.0 29.0±3.4

From Chumlea et al. (2005)

Bone mineral content (BMC) and density (BMD) Total body bone content increases

with age during growth and development to maturity, reaching “peak bone mass”

between 20 and 30 years of age in most individuals (Mora & Gilsanz, 2003; Sala,

Webber, Morrison, Beaumont, & Barr, 2007). In a cross‐sectional study, Møllgard et

al. (1997) indicated that the BMC in boys and girls at the age of 6.5 years was 611 g

and 616 g and it increased to 2598 g and 2218 g at the age of 18 years. Goulding et

al. (2003) reported in a longitudinal study that BMC had increased by 77% between

4‐5 y and 8‐9 y in girls. The density of mineral (mainly calcium and phosphorus) in

bone, which is a significant determinant of bone strength, also increases with age

from birth to maturity. Zhang et al. (Zhang, Liu, Zhai, Cao, & Duan, 2003) developed


50

normal reference values of BMD in 1025 Chinese children aged 6‐18 years. BMD was

0.66 g/cm2 and 0.62 g/cm2 for Chinese boys and girls at 7 years of age, 0.83 g/cm2

and 0.84 g/cm2 at 13 years of aged and increased to 1.03 g/cm2 and 0.93 g/cm2 at

17 years of age (Table 2.4).


51

Table 2.4. Total body bone mineral content and density for age and gender（Mean±SD）

BMC (g) BMD (g/cm2)

Age (yr)

Chinese males

Chinese females

White males

White females

Chinese males

Chinese females

White males

White females

8 1133±138 1122±181 1059±101 1059±145 0.69±0.06 0.66±0.05 0.856±0.020 0.858±0.068

9 1309±287 1251±196 1215±108 1167±122 0.69±0.06 0.67±0.05 0.900±0.047 0.890±0.046

10 1426±282 1416±252 1297±126 1244±147 0.72±0.07 0.72±0.07 0.916±0.039 0.904±0.056

11 1534±238 1617±282 1447±222 1438±235 0.74±0.07 0.76±0.07 0.944±0.058 0.932±0.074

12 1772±308 1785±301 1514±251 1673±325 0.79±0.08 0.79±0.08 0.943±0.052 0.985±0.093

13 2035±438 2036±316 1805±413 1835±331 0.83±0.10 0.84±0.09 0.990±0.076 1.019±0.100

14 2202±377 2225±330 2151±480 2004±335 0.84±0.09 0.88±0.10 1.039±0.090 1.063±0.079

15 2439±422 2329±316 2455±521 2123±351 0.90±0.11 0.93±0.09 1.098±0.108 1.104±0.082

16 2724±325 2304±284 2747±392 2142±314 0.96±0.10 0.92±0.09 1.168±0.094 1.103±0.077

17 2998±408 2364±290 2904±464 2249±278 1.03±0.13 0.93±0.08 1.198±0.091 1.138±0.061

18 2949±292 2410±305 3057±396 2258±243 ‐ 0.94±0.06 1.236±0.090 1.138±0.070

From Maynard et al. (1998); Zhai et al. (2004); Zhang et al. (2003).


52

Skeletal Muscle Skeletal muscle is the next most variable component, within and

between individuals, after fat mass. About 60% of FFM is composed of muscle, and

approximately 75% of total skeletal muscle mass is appendicular. At the molecular

level, skeletal muscle is composed of about 80% water and minerals, 19% protein,

and ~1% glycogen and lipid. Data representing changes with age in skeletal muscle

mass children and adolescents are relatively sparse compared to those for body fat,

fat distribution, or bone. In general, growth and development is a period for rapid

accretion of skeletal muscle with marked sexual dimorphism developing during

adolescence. Total skeletal muscle mass increased from 8.9 kg to 27.4 kg between

8.2±2.0 y and 14.9±2.0 y in boys and from 10.5 kg to 18.2 kg in girls between

8.0±1.1 y and 15.1±1.3 y (Kim et al., 2006). Skeletal muscle mass as a proportion of

body weight accounts for 20‐22% at infancy (Heymsfield, Gallagher, visser, Nunez, &

Wang, 1995), 41% at 5 years of age, 48% at adolescence and 45% in adulthood

(Malina, Bouchard, & Bar‐Or, 2004). Tanner et al. (1981) followed change in arm and

calf muscle widths in boys and girls aged 3 years over a 14‐year period. The results

indicated that the arm muscle width increased from 35.50 mm to 65.21 mm in boys

and from 30.93 mm to 41.53 mm in girls between the ages of 3 years and 18 years;

while the corresponding values of calf muscle width increased from 43.13 mm to

77.55 mm in boys and 41.53 mm to 70.54 mm in girls. The skeletal muscle mass

appears to increase in distinct phases, one between 1.5 y and 5 y, and the other is in

adolescence which is more apparent, especially in boys (Malina et al., 2004) (Table

2.5).


53

Table 2.5. Muscle mass as a percentage of body weight and width for age and sex

Boys Girls Age (yr) TMS

(%) AMW (mm)

AMWV (mm/yr)

CMW (mm)

CMWV (mm/yr)

TMS(%)

AMW (mm)

AMWV (mm/yr)

CMW (mm)

CMWV (mm/yr)

4 ‐ 36.56 2.21 45.31 2.99 ‐ 34.19 3.61 44.68 3.14

5 42 37.28 0.79 47.6 2.57 40.2 36.41 0.17 47.5 2.76

7 42.5 39.41 1.84 51.13 1.68 46.6 38.01 1.30 50.19 1.66

9 45.9 42.28 2.01 54.48 1.74 42.2 40.28 1.59 53.29 2.01

11 45.9 45.65 1.55 57.60 1.51 44.2 43.57 1.89 57.24 2.77

13 46.2 48.90 3.29 61.25 3.02 43.1 47.28 1.81 64.73 3.56

13.5 50.2 50.13 4.24 62.72 3.37 45.5 49.37 2.76 66.62 3.73

15 50.3 56.84 5.44 69.55 4.80 43.2 50.81 0.84 69.51 2.72

15.5 50.6 58.45 4.26 71.92 3.94 44.2 50.63 ‐0.38 70.52 0.26

17 52.6 63.74 ‐ 76.04 ‐ 42 52.81 ‐ 71.89 ‐

17.5 53.6 ‐ 1.31 ‐ 1.84 42.5 ‐ ‐1.32 ‐ 0.38

TMS: total muscle mass as a percentage of body weight; AMW: arm muscle width; AMWV: velocity of arm muscle width; CMW: calf muscle width; CMWV: velocity of calf muscle width. Data from Malina (2004) and Tanner (1981).

2.3.2 Sex differences in body composition in children

Sex is also one of the major sources of variation in body composition. Sex

differences in body composition are apparent early in life, are magnified during the

adolescent growth spurt and sexual maturation, and persist through adulthood.


Fat mass The sex difference in FM is negligible before 5 or 6 years of age (Fomon

et al., 1982) but subsequently, FM increases in both girls and boys, but more rapidly

in girls than in boys (Table 2.2). Moreover, FM increases continuously across the

adolescent period in girls with only modest changes in boys during their pubertal

growth spurt. These differences in accretion velocity and duration of FM change

result in higher FM in girls than in boys at each age group (Fomon et al., 1982; Guo

et al., 1997). In a longitudinal study, FM increased by 0.86 kg/y in females, higher

than that in males (0.42 kg/y) between 8 y and 20 y. In late adolescence and young


54

adulthood, females have, on average, about 1.5 times the FM of males (Guo et al.,

1997) (Figure 2.2 a). When FM is expressed as fat mass index (FM/height2, kg/m2),

the mean values are also higher in girls (4.0, 5.5, 7.0, and 7.3 kg/m2 at the age of 5‐8

y, 9‐11 y, 12‐14 y and 15‐18 y, respectively) than boys (3.0, 5.0, 4.9, and 3.7 kg/m2 at

the ages of 5‐8 y, 9‐11 y, 12‐14 y and 15‐18 y, respectively) (Nakao & Komiya, 2003).

Percentage of body fat %BF is only slightly greater in girls than boys during infancy

and early childhood (Fomon et al., 1982; Song & Zhang, 1999; Wang & Wang, 2003)

(Table 2.2). An approximate 2% difference in percent body fat between boys and

girls is evident by about 5 years of age (Fomon et al., 1982) and from 5‐6 years

through adolescence, girls consistently have a greater %BF than boys. Girls have

3.8% higher %BF at 5 years of age increasing to 12.9% at 18 years of age (Shaw,

Crabtree, Kibirige, & Fordham, 2007). The pattern of change in %BF is also different

between boys and girls. The relative fatness of females increases gradually through

adolescence in the same manner as FM, reaching 20 to 26% from the age 8‐20 y.

%BF also increases gradually in males until just before the adolescent spurt (about

11‐12 years) and then gradually declines. It reaches its lowest point at about 16‐17

years in males and then gradually rises into young adulthood (China's National

Group on Student's Constitution and Health Survey, 2000; Guo et al., 1997) (Figure

2.2 b). %BF tends to decline in males during adolescence due to the rapid growth of

FFM and slower accumulation of FM at this time.

Fat distribution Sex differences exist in both absolute and relative amounts of

adipose tissue. Before 5‐6 years of age, sex differences are not apparent in both

trunk and limb skin fold thickness (Mast, Kortzinger, Konig, & Muller, 1998). Girls

then have a greater trunk and extremity adipose tissue than boys by 7‐18 years of

age (China's National Group on Student's Constitution and Health Survey, 2000; He

et al., 2002; Tahara et al., 2002). Sex differences in relative fat distribution are

negligible from infancy through childhood into early adolescence. The T‐E skin fold

ratio (sum of subscapular and suprailiac/sum of triceps and biceps) is 0.97 for boys

and 1.00 for girls at the age of 5‐7 years (Mast et al., 1998). The T‐E ratio

(subscapular/triceps) is 0.84 for boys and 0.86 for girls aged 12 years (China's


55

National Group on Student's Constitution and Health Survey, 2000). Subsequently,

males accumulate proportionally more SAT on the trunk compared with extremities.

The T‐E skin fold ratio in males increases to 1.20 by the age of 18 years. The

disproportional accumulation of trunk SAT in males is marked during adolescence

and then continues more slowly through the fifth decade. The adolescent increase

in the T‐E ratio in males is, in part, a function of a reduction in the absolute thickness

of extremity skin folds during the growth spurt (China's National Group on Student's

Constitution and Health Survey, 2000). Females, in contrast, gain proportionally

similar amounts of SAT at trunk and extremity sites so that the T‐E ratio is

reasonably stable through the fourth decade (Moreno et al., 2007).

There are relative few studies on the IAAT and SAAT in children. Therefore,

consistent outcomes have not been seen. Goran et al. (1997) indicated that both

IAAT and SAAT were lower in pre‐pubertal boys than girls however, Huang et al.

(2001) found no significant sex difference in SAAT in American black and white

children aged 8.1±1.6 years, while IAAT was higher in boys. Moreover, accumulation

of IAAT and SAAT was similar between boys and girls over the 3‐5 year period. Fox et

al. (2000) followed the changes in IAAT and SAAT in children by 2 years. The results,

however, indicated that the VAT increased by 69.1% for boys and 48.4% for girls,

while SAAT increased by 78.1% for girls and 19.1% for boys.


Although FFM increases in both boys and girls during childhood and adolescence,

the sex difference in FFM is magnified gradually with age and is clearly established

during the adolescent spurt. During adolescence, the accretion velocity of FFM is

higher in boys with longer duration, while girls have slower accumulation of FFM

with shorter duration. Average amounts of FFM increased about 10 kg between 8 y

and 14 y of age in both males and females. After the age of 14 years, the mean

increase in FFM was 9.0 kg for females by 18 years of age, but for males, the average

amount of FFM increased about 25.0 kg. These changes in FFM are reflected by

corresponding changes in height and weight (Guo et al., 1997) (Figure 2.2 c).


56

When FFM is expressed per unit height, sex differences are small in childhood and

early adolescence, but after 14 years of age, males have considerably more FFM for

the same height as females (Malina et al., 2004). The same sex difference is found

when FFM is expressed as fat‐free mass index (FFM/height2, kg/m2). The FFM index

is 14.0, 14.9, 16.9 and 18.9 kg/m2 for boys aged 5‐8, 9‐11, 12‐14, and 15‐18 years,

respectively; the corresponding index in girls is 13.2, 14.2, 15.5, and 16.1 kg/m2

(Freedman et al., 2005).

Sex differences in the relative components of FFM are negligible during infancy but

apparent in early childhood. After about 3 years of age, the estimated relative

component of FFM indicates lower water and more protein and mineral in boys

(Fomon et al., 1982) (Table 2.2). The relative mineral content of FFM in boys

increased from 5.4% at about 10 years of age to 6.6% at 17‐20 years and the gain in

relative mineral content of FFM from early through late adolescence was about 22%

of the initial value at age of 10 years. The corresponding increase in mineral content

of FFM in girls was less, 5.2% to 6.1%, a relative increase of about 16% (Lohman,

1986).

Total body water During infancy and early childhood, TBW in boys is slightly greater

than girls (Fomon et al., 1982; Wang & Wang, 2003). Only 9.5% greater %BF in boys

is found at the age of 8 years (16.2 L vs 14.8 L). However, the sex difference in TBW

magnifies during adolescence being significantly higher in boys than girls. The TBW

in boys at the age of 16 y is about 40 L, 60% greater than girls at the same age

(Chumlea et al., 2005). Sex difference in percentages for TBW/FFM% is negligible

during infancy but apparent in early childhood. After about 3 years of age, water

comprised a slightly greater percentage of FFM in girls, which continued through

childhood and adolescence (Fomon et al., 1982; Lohman, 1986) (Table 2.2).

BMC and BMD Sex differences in whole body bone mineral content and bone

mineral density is negligible during infancy and childhood, with the average of BMC

being 472.5 g and 451.2 g in boys and girls aged 4‐6 years, respectively. The sex

difference in BMC and BMD becomes apparent during adolescence (Maynard et al.,

1998). BMC and BMD are slightly higher in girls during the adolescent spurt than


57

boys at the same age. But for boys, the accumulation of BMC and BMD is

significantly greater during their growth spurt than girls at the same age. Between

10 and 18 y, both absolute and relative accretion of BMC and BMD is higher in boys

than girls. This results in about a 20% higher BMC and 11% higher BMD in boys than

girls at late adolescence (Mølgaard & Michaelsen, 1998; Maynard et al., 1998; Zhai

et al., 2004; Zhang et al., 2003) (Table 2.4).

Skeletal muscle Skeletal muscle mass is slightly higher in boys than girls during

childhood. Appendicular skeletal muscle in Asian pre‐pubertal boys and girls is 10.3

kg and 8.5 kg, 10.1 kg and 9.5 kg for Caucasian pre‐pubertal boys and girls, in

contrast to 11.5 kg and 10.2 kg for African‐American pre‐pubertal boys and girls

(Song et al., 2002). The sex difference becomes more apparent during the growth

spurt with males gaining considerably more muscle mass than females during

adolescence with the sex difference persisting across the life span (Malina et al.,

2004). The width of limb skeletal muscle in boys is slightly higher than girls while the

width of the calf muscle is higher in girls during their growth spurt (about 11 y) than

boys at the same age (1.86 mm higher in girls at the age of 11 years and 3.38 mm

higher at 13 years of age), however the sex difference in width of arm muscle is less.

This sex difference is temporary with the muscle of boys being wider during their

growth spurt than girls of the same age (7.01 mm higher in calf muscle width and

13.69 mm higher in arm muscle width in boys). Compared with females, males show

well‐defined growth spurts in both arms and calf musculature (Tanner et al., 1981).

When skeletal muscle mass is expressed as a percentage of body weight, boys have

higher values than girls (Table 2.5).

2.3.3 Ethnic differences in body composition in children

In addition to the variability in body composition values derived from different

assessment techniques, there are distinct ethnic differences in body composition in

children and adolescents (Heymsfield et al., 2005). Moreover, some studies have

shown that ethnic differences in total body composition vary according to age

(pubertal status) and sex. A number of racial differences in body composition seem

to become apparent only after the onset of puberty.


58


Fat mass

Asian children have a higher FM compared with their black and white counterparts

for a given BMI level or the same height and weight level (Deurenberg, Bhaskaran et

al., 2003; Navder et al., 2009). Navder et al. (2009) conducted a study among

Caucasians, African Americans and Asians living in New York and Chinese living in

mainland of China. The study showed that Chinese pre‐pubertal children living in

mainland China had about 3.6 kg and 3.0 kg higher BMI‐, age‐ and sex‐adjusted

mean FM (as determined by dural‐energy X‐ray absorptiometry (DXA)) than African

Americans and Caucasians living in New York City, respectively, at the same level of

BMI, age and sex. An ethnic difference in FM between blacks and Asians (Chinese

and Korean) living in New York City was also evident. However, no difference was

seen between Caucasians and Asians living in New York City, indicating that

environmental factors might influence the relationship between FM and BMI.

Deurenberg et al. (2003) also reported a different FM measured via the sum of four

skin folds (biceps, triceps, subscapular and suprailiac) in 101 Singaporean Chinese

and 89 Dutch Caucasian adolescents aged 16‐18 years. The Chinese girls had a

higher sum of skin folds (69.1±15.4 mm) than Caucasian girls (52.4±17.8 mm)

although they had a lower BMI. Similarly, the sum of skin folds in Chinese boys was

significantly higher than Caucasians (48.8±17.0 mm vs 31.1±10.2 mm) although they

had similar BMI scores.

Typically, black children have a lower FM than white children. For example, the

DXA‐derived FM in African‐American pre‐pubertal children living in New York City

was significantly lower than their white counterparts for the same BMI, age and sex

(Navder et al., 2009). Sisson et al. (2009) compared the FM via the sum of

subscapular and triceps skin folds between black and white children and

adolescents aged 4‐18 y by BMI categories. In normal‐weight, overweight and obese

groups, white girls had higher SKF compared to their black counterparts adjusted for

BMI and age (25.0 mm, 45.0 mm, and 68.6 mm for white girls who were

normal‐weight, overweight, and obese, respectively; and 22.5 mm, 43.0 mm, and


59

64.9 mm for black girls, respectively). Results were similar for normal‐weight and

overweight boys (18.4 mm and 33.9 mm for white boys with normal‐weight and

overweight, respectively; 15.7 mm and 28.2 mm for black boys with normal‐weight

and overweight, respectively).

Further, Hispanic children have a higher FM than white counterparts. Ellis et al.

reported that Mexican‐American young boys and girls aged 3‐18 y had higher

DXA‐derived FM than European Americans and the difference remained significant

after adjustment for height, weight and age (Ellis, 1997; Ellis, Abrams, & Wong,

1997).

However, there are some inconsistent results regarding ethnic differences in FM. For

example, Novotny et al. (2007) reported that both white and Hispanic pubertal girls

had greater FM compared with their Asian counterparts at a given weight, height

and Tanner pubertal stage (14.41 kg, 14.52 kg, and 13.24 kg, respectively). Morrison

et al. (2001) also showed that black girls had a greater mean sum of triceps and

subscapular skinfolds at every age than whites. The data from the NHANES III in the

US also indicated that estimated mean FM from BIA was generally larger for

non‐Hispanic blacks than whites (Chumlea et al., 2002).

One of the reasons for the inconsistent results might be the impact of

environmental factors such as dietary patterns and physical activity levels on body

composition. Navder et al. (2009) reported that the Asian children living in New York

City had significantly higher FM than those living in mainland China and similar to

whites living in New York City. Moreover, sexual maturation is another contributing

factor for the divergent ethnic differences in FM. In a prospective study (Kimm et al.,

2001) in 1213 black and 1166 white girls aged 9‐10 y at baseline, no significant racial

difference in adiposity via skin fold thickness was seen among black and white

pre‐pubertal girls. However, significant racial divergence in adiposity was found

during early adolescence with the critical age for divergence being 12 y. The median

for the sum of skin fold thickness at the triceps, subscapular and suprailiac sites for

black girls was similar to that for white girls at 9 y of age but became greater for

black girls at 12 y (36 mm vs 32.5 mm). At 19 y the difference was 6 mm (49.5 mm vs


60

43.5 mm). The 85th percentile was substantially higher in black girls compared with

white girls at all ages and this racial difference widened with age, from a difference

of 9 mm at baseline to 20 mm by age 19 years. Sampei et al. (2003) also showed the

effect of sexual maturation on the ethnic difference in body fat mass. Japanese

pre‐menarcheal girls presented with less body fat than pre‐menarcheal Caucasian

girls using near‐infrared interactance. Conversely, the Japanese post‐menarcheal

girls accumulated more fat than their Caucasian counterparts, which led to no

significant difference in fat mass between the groups. Furthermore, some results

suffer from lack of consideration of differences in BMI or height and weight among

groups. Misinterpretation of the results can occur when comparing the FM between

groups if groups are not selected randomly and/or differ in height, weight or BMI.

For example, Ellis et al. (1997) reported that the difference in FM between black and

white in each sex was removed when adjustment was made for height and weight.

Therefore, it is critical to determine whether ethnic differences in FM are

independent of body size.

Fat distribution

A number of models have been used to explore the nature of whole body fat

distribution among different ethnicities.

Trunk and extremity fat model (or central and peripheral model)

The pattern of body fat on the trunk has been reported as a greater predictor of

obesity‐related health risk such as cardiovascular disease and type 2 diabetes than

overall adiposity (Berman et al., 2001; Ehtisham et al., 2005; Freedman et al., 1989;

Okosun, 2000; Okosun et al., 2000). The absolute amount of trunk and extremity fat

can be measured using skin fold thickness measurements or DXA measurements.

The relative distribution of fat mass in both the trunk and extremities is commonly

addressed with ratios of trunk to extremity. For skin fold thickness measurement,

the sum of all or a number of skin folds from the subscapular, suprailiac and

abdominal sites are often used as a surrogate index of trunk fat while the sum of all

or a number of the biceps, triceps, thigh, and medial calf skin fold sites are used for


61

extremity fat. Accordingly, studies vary in reported ratios of trunk‐to‐extremity fat.

Furthermore, waist‐to‐thigh ratio has also been used as an index for the central and

peripheral model (Ehtisham et al., 2005).

Asian children and adolescents appear to have higher trunk fat than white children

and adolescents while whites tend to have higher fat on the extremities than Asians,

in both absolute and relative terms. In a study of 143 Asian‐American and 120

Caucasian pre‐pubertal children aged 5‐12 y, skin fold thickness derived extremity

(sum of calf, thigh, biceps and triceps sites) and trunk fat (sum of subscapular,

suprailiac, and abdominal sites) plus DXA‐derived extremity (sum of legs and arms)

and trunk fat (sum of ribs, spine, and pelvis) were used to explore body fat

distribution. A main effect for ethnicity was found. Among girls, Caucasians had

higher skin fold and DXA‐derived extremity fat than Asians while in boys, Asians had

less DXA‐derived extremity fat than Caucasians (He et al., 2002). In another study in

American adolescent girls aged 11‐18 y (Malina et al., 1995), Asian girls had more

trunk subcutaneous fat than Caucasian girls. Moreover, Asians also had

proportionally more trunk SAT in terms of T (sum of subscapular and suprailiac

skinfolds) / E (sum of triceps and medial calf skinfolds) skinfold ratio, T/SUM (sum of

skinfold thickness at the four sites), and S (subscapular skinfold) / T (triceps skinfold)

ratio compared with their Mexican, white and black counterparts. These differences

remained after adjustment for overall adiposity and age. Novotny et al. (2006)

confirmed that Asian adolescents had a higher DXA‐derived trunk / peripheral fat

ratio than whites. This difference remained significant after adjustment for Tanner

stage, physical activity, energy intake, biacromial breadth and height. When using

waist‐to‐thigh ratio as a surrogate of central and peripheral model, South Asian

adolescents had significantly more central fat (waist‐thigh‐ratio in girls = 1.36 vs

1.25; boys = 1.52 vs 1.42) than white European adolescents at a given BMI SD score

(Ehtisham et al., 2005). The greater central fat depot in Asian children persists into

adulthood (Potts & Simmons, 1994; Rush et al., 2004; Rush, Freitas et al., 2009;

Wang et al., 1994).

White children have less trunk fat depots than blacks and Hispanics and the


62

difference is more apparent in adolescents. Harsha et al. (1980) found a higher

subscapular skin fold while lower triceps skin fold in black boys and girls aged 7‐15 y

compared with white children using a multivariate regression analysis across

maturation levels. Okosun et al. (2000) also reported that black and Hispanic boys

and girls aged 5‐11 y had less triceps and thigh skin fold thickness but a higher value

for subscapular/triceps skin fold ratio and sum of subscapular and suprailiac skin

fold / sum of triceps and thigh skin fold ratio than whites. However, some studies

showed higher values for both trunk and extremity fat in black children than white

children. For example, in a study conducted in 20 black girls and 20 white girls aged

from 7 to 10 y, no significant difference in absolute triceps, subscapular, or suprailiac

skin fold thicknesses was found except for higher ticeps skin fold thickness in white

children. DXA analyses also revealed there was greater fat mass in the arms and

trunk of whites (Yanovski et al., 1996). However, the sample size was small and

these differences in absolute values were not adjusted for difference in total body

fat or body size. When the trunk/extremity ratio is calculated as the sum of mean

subscapular and suprailiac/sum of mean triceps and biceps presented in the

reference, the ratio is similar between blacks and whites. Goran et al. (1997) and He

et al. (2002) also reported no significant differences in limb fat mass and trunk fat

mass between white and African‐American pre‐pubertal children. However, the

trunk fat depots in blacks and Hispanics were more pronounced during adolescence

because blacks and Mexicans accumulate proportionally more SAT on the trunk than

the extremities compared to European Caucasians (Bray, Delany, Harsha, Volaufova,

& Champagne, 2001). Malina et al. (1995) reported that Mexican adolescent girls

had higher T (sum of subscapular and suprailiac skinfold)/E (sum of triceps and

medial calf skin fold) ratio, subscapular skin fold/triceps skin fold ratio, and T/sum of

T and E ratio compared with their white counterparts, while blacks had lower ratios

than whites and Hispanics. The inconsistent results between blacks and whites

might be due to the much smaller sample size of blacks than whites and Mexicans

(27, 81 and 327, respectively). The proportionally greater body fat on the trunk of

blacks and Hispanics persists into adulthood (Nindl et al., 1998; Robson, Bazin, &

Soderstrom, 1971; Thomas, Keller, & Holbert, 1997; Zillikens & Conway, 1990).


63

Polynesian children also have more trunk fat than whites. Rush and Freitas et al.

(2009) showed that among 643 New Zealand European, Maori and Pacific Islander

children, subscapular skinfolds and subscapular‐to‐triceps ratio were highest in

Pacific boys and Maori girls. The similar ethnic difference as determined by DXA was

also reported in adults.

Gynoid fat (lower body fat) and android fat (upper body fat) model

An android or male fat pattern, with relatively greater fat in the trunk and upper

body, is associated with negative metabolic predictors (Daniels, Morrison, Sprecher,

Khoury, & Kimball, 1999; He et al., 2002). In contrast, a gynoid or female fat pattern,

with relatively greater fat in the hip and thigh areas, is associated with a lower

metabolic risk (Ashwell, 1994).

White children have less upper body fat than Asian, black and Hispanic children.

Morrison et al. (2001) compared the difference in upper and lower body fat

distribution using skin fold measurements between black girls and white girls

between 9‐19 y. Black girls had significantly higher upper body fat (sum of triceps

and subscapular skin folds) and greater increases of upper body fat (increases of

87.3% and 60.2%, respectively) than white girls. Moreover, an increase in upper

body fat was associated with a larger increase in total fat in black girls than in white

girls, which indicates that subcutaneous fat accounts for a larger proportion of total

fat in black girls than in white girls. Malina et al. (1995) also showed that greater SAT

on the upper body in Asian and black adolescent girls was evident compared to

white adolescent girls. Similar results were apparent in principal components

analyses of Asian and white adults (Wang et al., 1994). Zillikens et al. (1990) used

the suprailiac‐subcapular skin fold thickness ratio to explore the relative distribution

of fat mass in the upper and lower body. Black boys and girls had a lower

suprailiac‐subcapular ratio than whites (1.21±0.42 and 1.66±0.49) for black and

white boys, respectively; (1.01±0.32 and 1.59±0.46) for black and white girls,

respectively.

Beyond skin fold thickness measurements, the ethnic differences in gynoid and


64

android fat distribution have been found using more advanced body composition

techniques including DXA. Novotny et al. (2007) reported that at a given height,

weight, and Tanner stage, Asian pubertal girls had less DXA‐derived gynoid fat mass

than whites and Hispanics, and less android fat mass than Hispanics. Both Asian and

Hispanic pubertal girls had greater android/gynoid fat ratio than whites (0.369,

0.364, and 0.318, respectively). However, in another study conducted in 95

African‐Americans, 143 Asians and 120 Caucasians aged 5‐12 y (He et al., 2002),

white boys and girls had similar DXA‐derived gynoid (sum of pelvis and legs) and

android fat (sum of ribs and spine) to Asian and blacks although Asian girls had less

skin fold‐derived gynoid fat than white girls. However, He et al. (2004) also reported

that fat pattern varied according to puberty. Among girls, race differences were

more evident in the pre‐ and late pubertal groups compared with early puberty.

Asians had statistically less gynoid fat than whites using both skin folds and DXA at

late puberty and by skin folds at pre‐puberty. Asians had less adjusted gynoid fat

mass by DXA than African‐Americans in all pubertal groups, although the difference

was not statistically significant in early puberty. Among boys, the differences

between Asians and the other two races were not consistent. Asians had statistically

less gynoid fat than whites in early puberty, as measured by skin folds and DXA, and

in late puberty, as measured by DXA. Compared with African‐Americans, Asians had

more gynoid fat measured by skin folds in pre‐puberty and less gynoid fat by DXA in

late puberty.

Abdominal fat

The abdominal adipose tissue consists of SAAT and VAAT or intra‐abdominal adipose

tissue (IAAT). There are significant racial differences in abdominal fat distribution in

both children and adults which can explain the variation in the prevalence of some

obesity‐related diseases in different ethnic groups. The amount of abdominal fat,

SAAT, VAAT can be determined by CT and MRI. Abdominal fat can also be measured

by DXA. Moreover, some anthropometric variables are often used as an index of

abdominal fat, such as the suprailiac and abdominal skin folds and waist

circumference.


65

Black children have less absolute SAAT, VAAT and total abdominal adipose tissue

than white children. In 62 African‐American (8.3±1.4 y) and 39 Caucasian (8.6±1.2 y)

pre‐pubertal children, Herd et al. (2001) showed that abdominal fat, SAAT and IAAT

were greater in Caucasian children than in their African‐American counterparts after

adjusting for total body fat mass. Yanovski et al. (1996) also indicated that SAAT, IAAT,

and total abdominal adipose tissue were significantly smaller in black girls (n = 20)

than white girls (n = 20) matched for age, socioeconomic status, pubertal stage, BMI,

and weight. Owens et al. (1999) predicted VAAT from simple anthropometric

measurement in black and white youths aged 4‐16 y. Ethnicity entered the model

and blacks had lower VAAT than whites. A similar result was found between black

and white obese adolescents despite similar BMI and %BF (Bacha, Saad, Gungor,

Janosky, & Arslanian, 2003).

There is also a racial difference in the relative distribution of abdominal fat and

change in abdominal fat in children. Goran et al. (1997) studied the relation

between IAAT and SAAT in 101 white and African‐American pre‐pubertal children.

The relative distribution of adipose tissue in the intra‐abdominal compared with the

subcutaneous abdominal region is significantly lower in African‐American children

than in white children. White children had higher IAAT (adjusted mean value were

40.2±3.1 cm2 and 43.2±2.7 cm2 in boys and girls, respectively) compared with their

black counterparts (adjusted mean value were 26.4±1.9 cm2 and 25.2±1.6 cm2 in

boys and girls, respectively) at a given SAAT. Although limited to a cross‐sectional

analysis, the findings of this study imply that the rate of accumulation of IAAT

relative to SAAT is 26% lower in African‐American compared to Caucasian children.

Huang et al. (2001) followed the change in abdominal adipose tissue of 138 children

(8.1±1.6 years at baseline) using CT over a 3‐ to 5‐y period. Whites showed a higher

visceral fat growth than African‐Americans (difference = 1.9±0.8 cm2/yr), but there

was no ethnic difference in growth of the subcutaneous abdominal fat. The less

VAAT at an early stage in life and the lower growth in African‐American children

compared with white children may explain the lower VAAT in African‐American

adults (Conway, Yanovski, Avila, & Hubbard, 1995).


66

Few studies have compared the difference in VAT between Asian and children from

other ethnic groups. However, Asian adults appear to have more abdominal fat mass

compared with their European, Maori and Pacific Island counterparts after

adjustment for age, height and weight (Rush, Freitas et al., 2009), and have a higher

VAAT compared with European Americans after adjusting for age and TBF/SAAT

(Park et al., 2001; Tanaka, Horimai, & Katsukawa, 2003).

Waist circumference

WC, an index of central fat, has been widely used as a predictor of cardiovascular

risk factors in both children and adults (Bacha, Saad, Gungor, & Arslanian, 2006;

Janssen et al., 2004; Lee, Bacha, Gungor, & Arslanian, 2006; Lofren et al., 2004; Ng

et al., 2007). WC is also recognized as a key component of the metabolic syndrome

in both children and adults (Alberti, Zimmet, & Shaw, 2006; Zimmet et al., 2007).

Therefore, many countries have developed WC percentiles for children and

adolescents. Table 2.6 shows the 50th percentile for WC in different

ethnicities/countries.

The 50th percentile values for WC of Asian boys and girls (Sung et al., 2008; Yan et

al., 2008) are lower than those of Caucasian (Eisenmann, 2005; Fernández, Redden,

Pietrobelli, & Allison, 2004; Katzmarzyk, 2004; McCarthy, Jarrett, & Crawley, 2001),

black (Fernández et al., 2004), Mexican (Fernández et al., 2004) and Turkish children

(Hatipoglu et al., 2008). The main reason for the difference is body size. A strong

positive relationship between WC and height exists throughout childhood and into

adulthood therefore one study indicated that after controlling for height, weight or

BMI, Asians appear to have higher WC than whites (Ehtisham et al., 2005). It is

interesting to compare the data from the two studies in Chinese children. Children

in Xinjiang province, which lies in North‐West China, had higher WC than those in

Hong Kong, in South China, indicating body composition is influenced by the

environment in addition to age, gender and ethnicity.

The 50th percentile values for WC of black boys are lower than that of white boys,

while higher than white girls (Fernández et al., 2004) with the difference between


67

black and white girls more evident during adolescence. In a 9‐year longitudinal

study conducted in 1213 black and 1166 white girls (aged 9‐10 y at baseline), the

mean annual increase in WC was significantly higher for black females compared to

white females. However, after controlling for BMI, the mean annual increase in WC

for white females was significantly higher than for black females (0.08 cm/yr vs ‐0.07

cm/yr, respectively) (Tybor, Lichtenstein, Dallal, Daniels, & Must, 2010). Another

cross‐sectional study also reported the similar difference between the two groups

adjusted for BMI and age. WC was significantly higher in white boys across the BMI

range compared to black boys adjusted for BMI and age (63.3 cm, 74.9 cm, and 88.3

cm for white boys with normal weight overweight, obesity, respectively; 62.0 cm,

71.7 cm, and 83.1 cm for black boys with normal‐weight overweight, obesity,

respectively). In the normal‐weight group, white girls had lower WC compared to

black girls (60.3 cm vs 61.1 cm) but had higher WC in obese group (84.0 cm vs 82.1

cm). No significant difference in WC was found between white and black overweight

girls (70.8 cm vs 69.8 cm) (Sisson et al., 2009).

Turkish children (Hatipoglu et al., 2008) had lower 50th percentile values for WC

than Caucasians, blacks and Mexicans (Fernández et al., 2004) while higher than

Chinese (Sung et al., 2008; Yan et al., 2008) across the age group. A comparison of

50th percentile values with other ethnic groups showed that Mexican children had

the highest WC (Fernández et al., 2004).

The ethnic difference in WC, in addition to age and sex differences, makes the WC

percentile more appropriate as a threshold to predict cardiovascular risk factors and

metabolic syndrome in children and adolescents rather than using fixed point values

(de Ferranti et al., 2004; Zimmet et al., 2007).


68

Table 2.6. The 50th percentile of WC in different ethnicities/countries

Age (yr)

European American

African American

Mexican American

Australian British Canada Turkish Xinjiang Hong Kong

Boys

5 53.3 52.0 54.2 50.3

6 55.4 53.9 56.3 51.5 50.3

7 57.5 55.7 58.5 54.9 52.7 53.3 53.4 51.7

8 59.6 57.6 60.7 56.3 54.1 54.7 54.2 53.2

9 61.7 59.4 62.9 57.6 55.3 56.4 55.4 54.7

10 63.7 61.3 65.1 59.0 56.7 58.2 57.1 56.2

11 65.8 63.2 67.2 60.0 58.2 62.3 59.9 59.1 57.8

12 67.9 65.0 69.4 62.7 60.0 64.0 61.7 61.7 59.2

13 70.0 66.9 71.6 64.9 61.7 66.1 63.4 64.4 60.3

14 72.1 68.7 73.8 66.8 63.2 68.8 64.7 66.5 61.1

15 74.1 70.6 76.0 68.2 64.4 71.4 65.7 67.8 61.7

16 76.2 72.5 78.1 65.3 73.0 66.2 68.0 62.2

17 78.3 74.3 80.3 74.3 66.5 67.5 62.6

18 80.4 76.2 82.5 75.5 66.8 62.9

Girls

5 53.1 52.3 54.2 51.3

6 55.0 54.5 56.3 52.6 52.2 52.5

7 56.9 56.6 58.4 55.0 53.3 54.2 55.4 53.9

8 58.8 58.7 60.4 57.4 54.7 56.4 57.1 55.3

9 60.7 60.9 62.5 60.5 56.4 58.5 58.6 57.0

10 62.5 63.0 64.6 63.2 58.2 60.5 60.1 58.8

11 64.4 65.1 66.6 65.4 60.2 60.2 62.5 61.8 60.4

12 66.3 67.3 68.7 69.3 62.3 61.6 64.6 63.8 61.8

13 68.2 69.4 70.8 64.6 63.8 66.8 66.6 63.0

14 70.1 71.5 72.9 67.0 65.7 68.9 69.4 64.3

15 72.0 73.6 74.9 69.3 67.0 70.6 71.6 65.7

16 73.9 75.8 77.0 71.6 67.4 71.8 66.9

17 75.8 77.9 79.1 67.8 73.0 68.0

18 77.7 80.0 81.1 68.4 68.8

Data for European Americans, African‐American and Mexican Americans are from the Third National Health and Nutrition Examination Survey (NHANES III) in the US (Fernández et al., 2004). Data for Australians are from the 1985 Australian Health and Fitness Survey (Eisenmann, 2005). Data for Canadians are from 1981 Canada Fitness Survey (Katzmarzyk, 2004). Data for British children are from a representative sample (1990) (McCarthy et al., 2001). Data for Hong Kong Chinese children are from a largely representative sample (2005/6) (Sung et al., 2008). Data for Turkish children are from the study of the Determination of Anthropometric Measures of Turkish Children and Adolescents in 2005 (Hatipoglu et al., 2008). Data for Xinjang Chinese children (Xinjiang province) are collected in 2005 (Yan et al., 2008).


69


The data from the NHANES III in the US indicated that estimated FFM means from

BIA were generally larger for non‐Hispanic whites than non‐Hispanic blacks and then

Mexican‐Americans among males. For females, estimated mean FFM values were

larger for non‐Hispanic blacks than non‐Hispanic whites and then

Mexican‐Americans (Chumlea et al., 2002) (Table 2.7). Going et al. (2006) also

showed that black girls had significantly greater FFM (10‐13%) than Hispanic and

non‐Hispanic white girls (38.3 kg vs 34.3 kg vs 34.9 kg). However, some studies have

indicated that there was no significant difference in FFM from DXA between black

and white girls aged 6‐17 y (Morrison, Guo et al., 2001).

For the Asian population, Sampei et al. (2003) indicated that both Japanese pre‐ and

post‐menarcheal adolescents have lower FFM (29.9 kg vs 32.0 kg, and 37.2 kg vs

40.7 kg, respectively) and FFMI (14.5 kg/m2 vs 14.7 kg/cm2; and 14.9 kg/m2 vs 15.3

kg/m,2 respectively) than their Caucasian counterparts. Results from a number of

studies show that Asian boys and girls have lower FFM than black, white, and

Hispanic children matched for age (Eston, Cruz, Fu, & Fung, 1993; Group of China

Obesity Task Force, 2004).

For New Zealand children at the same height and weight (BMI), Pacific and Māori

girls had more FFM (as determined by BIA) than European girls while in boys no

differences are seen between these ethnic groups (Rush, Scraqq et al., 2009).


70

Table 2.7. Body composition for age, sex and ethnicity: NHANES III

Non‐Hispanic white Non‐Hispanic black Mexican‐American Age (y)

FM (kg)

FFM

(kg) TBW (L)

FM (kg)

FFM (kg)

TBW (L)

FM (kg)

FFM (kg)

TBW (L)

Males

12‐13.9 10.0 41.8 31.3 11.2 40.9 30.7 12.3 40.3 30.2

14‐15.9 14.0 54.3 40.6 12.2 52.2 38.9 12.6 49.8 37.2

16‐17.9 13.1 57.8 43.1 13.4 55.3 41.2 14.9 53.0 39.6

Females

12‐13.9 14.0 38.1 28.5 15.8 39.3 29.3 16.0 37.3 27.9

14‐15.9 17.4 40.4 29.9 20.2 41.8 30.9 18.4 37.8 28.1

16‐17.9 19.5 41.6 30.7 22.0 42.0 31.0 21.6 40.5 30.2

Data from Chumlea et al. (2002).

Total body water

The data from NHANES III in the US indicated that estimated TBW means from BIA

were generally larger for non‐Hispanic whites than non‐Hispanic blacks and then

Mexican‐Americans among males. For females, estimated mean TBW values were

higher for non‐Hispanic blacks than non‐Hispanic whites and then

Mexican‐Americans (Chumlea et al., 2002) (Table 2.7). Some studies with small

sample sizes have also indicated that black children and adolescents have higher

TBW than whites (Wong, Stuff, Butte, Smith, & Ellis, 2000). However, Slaughter et al.

(1990) showed that no significant difference in TBW (32.6 L vs 29.7 L for boys and

28.5 L vs 26.7 for girls) or TBW/FFM (71.6% vs 72.4 % for boys and 72.2% vs 72.0%

for girls) between blacks and whites although there was a trend of higher TBW

values in blacks. Bray et al. (2002) also drew similar conclusions to those from

Slaughter et al. (1990).

Bone mineral content and density

Estimated total body BMC and BMD are greater in American black children and

adolescents of both sexes compared with American whites (Bray et al., 2001; Ellis et

al., 1997; Horlick et al., 2000; Malina et al., 2004; Nelson, Simpson, Johnson,


71

Barondess & Kleerekoper, 1997; Slaughter et al., 1990) (Table 2.8). However, Afghani

et al. (2006) indicated that there was no significant difference in mean values of

total body BMC and total body bone area in Caucasian and African‐American

children aged 5‐10 y (1170 g vs 1154 g; 1297 cm2 vs 1315 cm2). Bray et al. (2001)

also showed that the changes in BMC and BMD were not different between black

and white males aged 12‐14 y during the 2‐y follow‐up period (0.47 g vs 0.44 g;

0.079 g/cm2 vs 0.073 g/cm2). For the Asian population, Song et al. (2002) showed

that Asian girls had lower TBBMC than Caucasian and African‐American girls, but no

significant difference was found among boys from the three ethnicities. Horlick et al.

(2000) also indicated that Asian children had a lower total body BMC than blacks,

but no significant difference between Asians and whites. Ellis measured the BMC

using DXA in white, black and Mexican‐American males aged 3‐18 y and results

showed no difference in BMC between whites and Mexican‐American (Hispanic) but

greater BMC in blacks than Hispanics (Ellis, 1997). Nelson et al. (1997) compared the

difference in BMC between white and Middle Eastern (Chaldeans) boys and girls

(8.9±0.6 y) and showed that whites had lower BMC than Chaldeans.

Data from specific bone sites (e.g., vertebrae, femur, pelvis, and distal radius) also

indicated greater BMD in American black than in American white infants, children,

and adolescents (Gilsanz et al., 1998; Rupich, Specker, Lieuw‐A‐Fa, & Ho, 1996;

Slaughter et al., 1990; Yanovski et al., 1996). Corresponding data for Hispanic

children and adolescents indicate similar lumbar BMD to American whites but lower

BMD compared with American blacks (Malina et al., 2004).


72

Table 2.8. Total body bone mineral content and bone mass area for children and adolescents from different ethnicities

Total body BMC (kg) Total body bone mass area (cm2)

Asian Black White Hispanic Chaldean Asian Black White

Horlick et al. (2000)

Boys 1.131±0.258 1.160±0.261 1.100±0.285 1272±215 1273±205 1251±236

Girls 0.973±0.193 1.136±0.239 1.083±0.272 1156±170 1258±186 1240±227

Song et al. (2002)

Boys 1.15±0.28 1.20±0.32 1.09±0.29

Girls 0.97±0.19 1.12±0.23 1.08±0.26

Ellis (1997)

Boys

3‐5 yr 0.456±0.106 0.423±0.094 0.403±0.063

5‐9 yr 0.900±0.195 0.793±0.232 0.827±0.192

10‐14 yr 2.038±0.633 1.655±0.496 1.724±0.475

15‐18 yr 3.181±0.440 2.545±0.430 2.545±0.334

Afghani et al. (2006)

Boys and girls (5‐10 y) 1.154±0.298 1.170±0.346

Nelson et al. (1997)

Boys 0.961±0.259 0.855±0.191 0.972±0.280

Girls 0.966±0.266 0.865±0.257 0.926±0.300 1315.1±238.1 1297.7±280.5

Data from Horlick et al. (2000), Song et al. (2002), Ellis (1997), Nelson et al. (1997), and Afghani et al. (2006).


73

Skeletal muscle

A few studies have considered ethnic differences in skeletal muscle. Song et al.

(2002) measured appendicular skeletal muscle mass in 170 girls and 166 boys by

DXA. The results indicated that Asian pre‐pubertal girls have less appendicular

muscle mass than Caucasian girls and then African‐American girls after adjustment

for age, height and weight (8.5 kg vs 9.5 kg vs 10.2 kg). Caucasian boys had lower

amounts of appendicular muscle mass than Asian and African‐American boys (10.1

kg vs 10.3 kg vs 11.5 kg).

2.3.4 Ethnic differences in the relationship between BMI and %BF

It is well established that the relationship between BMI and body fat is age‐ and

gender‐dependent. Among adults, for an equivalent BMI, women have a

significantly greater amount of total body fat than men across the entire adult

lifespan. Compared to younger individuals, older persons have a higher %BF at a

comparable BMI, a difference which holds in both men and women (Deurenberg et

al., 1991, 1998; Gallagher et al., 1996; Jackson et al., 2002). In children and

adolescents, despite the results of some studies showing that girls have a higher

%BF than boys at the same age and BMI, and older children typically have a lower

%BF than younger children at the same BMI and sex (Daniels et al., 1997;

Deurenberg et al., 1991), the age‐ and gender‐relationship is complicated by varying

growth rates and maturity levels (Wang & Bachrach, 1996). This finding may help to

explain the relatively weak correlations in children and adolescents (Deurenberg et

al., 1991) compared with adults (Deurenberg et al., 1998; Gallagher et al., 1996).

There is also increasing evidence of the ethnic variation in the relationship between

%BF and BMI in adults (Chung et al., 2005; Deurenberg‐Yap et al., 2000; Deurenberg

et al., 1998; Jackson et al., 2002). Deurenberg (1998) concluded from a

meta‐analysis that for the same level of fatness, age and sex, American blacks (+1.3

kg/m2) and Polynesians (+4.5 kg/m2) had higher BMI values than Caucasians. Further,

black women had a lower %BF than white women at the same BMI (Jackson et al.,

2002). Asian individuals, including Japanese, Chinese, Thais and Malays have greater


74

body fat deposition than Caucasians (Chang et al., 2003; Chung et al., 2005;

Deurenberg et al., 1998, 2002; Gurrici et al., 1998; Kagawa et al., 2006; Wang et al.,

1994) and blacks (Chang et al., 2003) at the same BMI value. For the same %BF, age

and sex, BMI scores were 2‐5 units lower compared to Caucasians (‐2.9 kg/m2 for

Indonesians (Gurrici, 1998), ‐1.5 kg/m2 for Japanese men (Kagawa et al., 2006), ‐1.9,

‐3.2 and ‐2.9 kg/m2 in Chinese, Indonesians and Thais (Deurenberg et al., 1998), ‐3.0

kg/m2 for Singaporeans (Deurenberg‐Yap et al., 2002).

Figure 2.3. Adjustments to be made in BMI to reflect equal levels of body fat compared to Caucasians of the same age and gender (mean, 95%CI). Differences in BMI: differences from BMI cut‐off points as suggested by the WHO. * P <0.05. Adapted from Deurenberg et al. (1998).

Ethnic differences in the relationship between BMI and %BF seen in adulthood have

their origins in childhood. Some studies have also shown similar ethnic differences

in the relationship between BMI and %BF among children and adolescents. For

example, white children and adolescents had a higher %BF compared with blacks


75

(Daniels et al., 1997; Freedman et al., 2008; Going et al., 2006; Navder et al., 2009),

Hispanics (Freedman et al., 2008) and Polynesians (Rush, Puniani, Valencia, Davies,

& Plank, 2003; Rush, Scraqq et al., 2009) for a given BMI. Freedman et al. (2008)

reported that white boys aged 5‐18 y had 3.5% more body fat (estimated by DXA)

than black boys and 0.5% more body fat than Hispanic boys at an equivalent age

and BMI‐for‐age. The mean body fatness of white girls was 2.6% higher than black

girls and 0.5% higher than Hispanic girls adjusted for age and BMI. Going et al. (2006)

also showed that black adolescent girls aged 10‐15 y had a 3.3% lower %BF (as

estimated by DXA) than white and Hispanic girls at a given weight,

height2/resistance and age. Black children and adolescents ranging in age from 7‐17

years had a 1.5% lower %BF compared with whites for a given BMI and waist‐to hip

ratio after controlling for gender and maturation stage (Daniels et al., 1997). Navder

et al. (2009) found that black pre‐pubertal children living in New York City had a

2.7% lower %BF compared with their white counterparts. Rush et al. (2003; 2009)

reported that European girls aged 5‐14 y had a 3.7% higher %BF than Maori and

Pacific Island girls at the same BMI, but for boys, no ethnic difference was found.

Finally, Duncan et al. (2009) also found that Pacific Islander girls aged 5‐16 y had an

average of 1.8% less %BF than European girls living in Auckland, New Zealand.

Asian children and adolescents have a higher %BF than Caucasians (Deurenberg,

Bhaskaran et al., 2003; Deurenberg, Deurenberg‐Yap et al., 2003; Duncan et al.,

2009; Ehtisham et al., 2005; Freedman et al., 2008; Mehta et al., 2002; Navder et al.,

2009), blacks (Freedman et al., 2008; Navder et al., 2009) and Pacific islanders

(Duncan et al., 2009) at a given BMI. In a study conducted among 1676 girls aged

5‐16 y from five ethnic groups living in Auckland, New Zealand, including European,

Pacific Island, Maori, East Asian (Chinese, Korean, Filipino, Thai, and other East

Asian), South Asian (Indian, Sri Lankan, and others). South Asian girls had the highest

%BF (as determined by BIA) at a given BMI and age, followed by East Asians,

European, Maori, and Pacific Island girls (31.0±0.6%, 28.1±0.5%, 26.8±0.3%,

26.2±0.5%, and 25.0±0.4% for South Asian, East Asian, European, Maori, and Pacific

Island girls, respectively) (Duncan et al., 2009). Mehta et al. (2002) also found that


76

South Asian boys (India, Bangladesh and Sri Lanka origin) aged 15‐16 y had

approximately 4.5% more body fat for a given BMI than their European‐Caucasian

counterparts. East Asian boys (China, Taiwan, Korea, Indochina origin) had 2.8% less

body fat for a given BMI than European‐Caucasian boys but this mean difference did

not reach statistical significance. In other words, at the same %BF, Caucasians had

2.0 kg/m2 higher BMI than South Asian boys. Ehtisham et al. (2005) confirmed that

the South Asian adolescents living in the UK (Indian, Sri Lankan, Pakistani, and

Bangladeshi) had significantly higher %BF predicted from skin fold thicknesses across

the BMI SDS range compared to their European counterparts after adjusting for BMI

SDS. Freedman et al. (2008) reported that in 1104 healthy 5‐18 year‐old white, black,

Hispanic and Asian children, Asian boys had similar body fat to white boys while

3.5% higher %BF compared to black boys. Asian girls had 0.7% and 3.3% lower %BF

compared with white and black girls, respectively, at the equivalent age and

BMI‐for‐age. Deurenberg et al. (2003) found that after controlling for differences in

age and %BF, Singaporean Chinese children had a lower BMI (15.6±0.3 kg/m2) than

the Dutch (16.9±0.3 kg/m2) children. For the same BMI, age and sex, Singaporean

Chinese children had a significantly higher %BF (24.6±0.7%) than the Dutch

(20.3±0.7%) children. In another study conducted in adolescents aged 16‐18 y,

Deurenberg et al. (2003) also reported that predicted %BF was 5.8% higher in

Singaporean Chinese girls and 6.0% higher in Singaporean Chinese boys compared

to their Caucasian counterparts of the same age and BMI. Among pre‐pubertal

children, Asians living in Jinan city (China) and in New York City have significantly

higher %BF (as determined by DXA), 8.6% and 2.1%, respectively, compared with

Caucasians, and 10.5% and 4.0%, respectively, higher %BF compared with African

Americans for the same BMI, age and sex (Navder et al., 2009).

Asians are also different from each other in their BMI‐%BF relationship in both

children (Duncan et al., 2009; Mehta et al., 2002) and adults (Deurenberg‐Yap et al.,

2000; Deurenberg et al., 1998; Gurrici et al., 1999). South Asian girls (India, Sri Lanka

and others) had a 2.9% higher %BF than East Asian girls (China, Korea, The

Philippines, Thailand and others) (Duncan et al., 2009) and South Asian adolescent


77

boys (India, Bangladesh, and Sri Lanka) had a 2.8% higher %BF than East Asian

adolescent boys at the same BMI and age (Mehta et al., 2002). Deurenberg‐Yap et al.

(2000) also reported that Indian Singaporean adults had a 1.5% higher %BF than

Malay Singaporeans and 2.0% higher than Chinese Singaporeans for the same

gender, age and BMI. There are also differences in the BMI‐%BF relationship

between East Asians. For example, Guricci et al. (1999) compared the BMI‐%BF

relationship between Chinese Indonesians and Malay Indonesians. Results showed

that %BF predicted from BMI using a Caucasian prediction formula was

underestimated by 5.8±4.8% and 7.7±3.8% in the male and female Malay

Indonesians and by 1.3±3.0% and 1.7±3.7% in the male and female Chinese

Indonesians. After correction for differences in age, sex and %BF the Chinese

Indonesians had a 1.7±0.3 kg/m2 (P<0.0001) higher BMI than the Malay Indonesians.

After correcting for body build and relative sitting height the difference was reduced

to 0.9±0.4 kg/m2 but was still significant. In another study conducted in Chinese,

Malays, and Indian Singaporeans, Malays had a slightly higher %BF (0.5%) than

Chinese for the same age, gender, and BMI (Deurenberg‐Yap et al., 2000). In a

meta‐analysis conducted by Deurenberg et al. (1998), although Asian populations

had lower BMIs compared with Caucasians at a given %BF, age and gender, the

difference in BMI was greater between Indonesian (3.2 kg/m2) and Thai (2.9 kg/m2)

and Caucasians than between Chinese and Caucasians (1.9 kg/m2). A similar

difference in the BMI‐%BF relationship has also been reported in Caucasians of

different origin. For example, Deurenberg et al. (1998) found that the American

white Caucasians had a 3.8% lower %BF compared with their European counterparts

(mainly UK and The Netherlands) at the same BMI level after correction for age and

gender.

Importantly, the geographic location and socio‐economic conditions in which people

live also influences body composition in both children and adults of the same

ethnicity. Deurenberg et al. (2003) found that after controlling for differences in age

and %BF, Singaporean Chinese children had a lower BMI (15.6±0.3 kg/m2) than

Beijing Chinese (17.6±0.3 kg/m2) children. For the same BMI, age and sex the


78

Singaporean Chinese children had a significantly higher %BF (24.6±0.7%) than the

Beijing Chinese (19.2±0.8%) children. Navder et al. (2009) recently compared the

BMI‐%BF relationship in Chinese pre‐pubertal children living in Jinan city, China and

Chinese and Korean pre‐pubertal children living in New York city, USA. Chinese

children living in mainland China had a 6.5% lower %BF (as determined by DXA) than

those living in New York at the same BMI, age, and sex. Luke et al. (1997) also

showed the difference in BMI‐%BF relationship in three black populations of West

African heritage living in different environments, including Nigeria, Jamaica and the

United States. Blacks in the US had the highest %BF, followed by Jamaicans and

Nigerians with the lowest %BF (Luke et al., 1997).

Some studies have shown an interaction of BMI and ethnicity. For example, the

relationship between BMI and %BF varies by BMI category (Fernández et al., 2003;

Freedman et al., 2008; Swinburn, Ley, Carmichael, & Plank, 1999; Wang & Bachrach,

1996). Fernández et al. (2003) found that at a BMI <30 kg/m2, Hispanic Americans

aged 18‐110 y tended to have more body fat (as determined by DXA) than did

European Americans and Africa Americans, and at a BMI >35kg/m2, European

Americans tended to have more body fat than Hispanic Americans and African

Americans. Wang et al. (1996) also reported that Whites youths aged 9‐15 y had

more body fat (as determined by DXA) than Hispanics at a BMI <20 kg/m2, but less

body fat at a BMI >20 kg/m2. Wang et al. (1994) also found variation in the

relationship between fat and BMI at different BMIs among 445 whites and 242 Asian

adults aged 18‐94 y. For individuals who were lean or normal (BMI = 15 kg/m2 for

lean, BMI = 25 kg/m2 for normal, and BMI = 35 kg/m2 for obese), Asians were fatter

than whites in both sexes, but the differences in %BF between whites and Asians

varied by BMI in different directions for males and females: %BF increased with BMI

for males but decreased with BMI for females. Freedman et al. (2008) reported that

the ethnic differences in BMI‐%BF between white and Asian children and

adolescents varied by BMI‐for‐age. Asian boys had a 1.5‐1.7% higher %BF than white

boys between the 50th and 95th percentiles of BMI‐for‐age, but overweight Asian

boys (≥95th percentile of BMI‐for‐age) had a 2.2% lower %BF than overweight white


79

boys. Fairly similar results were seen among girls, with relatively thin (BMI‐for‐age

<85th percentile) Asians having 1.2‐1.8% higher %BF than whites, while overweight

Asian girls having 3.0% lower %BF than white girls (Figure 2.4). The interaction

between BMI and ethnicity in the estimation of body fatness suggests that it may be

difficult to identify equivalent levels of body fatness by simply adjusting BMI for the

average difference in body fatness across ethnicity groups (Freedman, & Sherry,

2009).


80

Figure 2.4. The predicted levels of %BF from sex‐specific equations using age and BMI‐for‐age as independent variables expressed relative to levels among white children (represented by the horizontal line at y = 0): (a) Boys; (b) girls. A positive value indicates that the mean value of %BF is higher than that of white children and a negative value indicates a lower %BF than white children. The arrow on the x‐axis represents a BMI‐for‐age percentile of 97.5 (about the mean value among overweight children), and the vertical lines represent ±1 standard error (se) around the estimated difference (relative to whites) for each race/ethnicity (Freedman et al., 2008).

Several factors might contribute to the ethnic differences in the BMI‐%BF

relationship. Firstly, differences in physical activity level might be a contributing

factor (Gurrici et al., 1999; Luke et al., 1997). Groups with a higher level of physical

activity might have a relatively higher muscle mass thus less fat mass at the same

body weight, which would result in an overestimation of %BF from BMI, using

prediction equations developed in a less active population.

Secondly, a difference in the trunk‐to‐leg‐length ratio, namely relative leg length or

relative sitting height, might also be a contributing factor to ethnic differences in


81

the BMI‐%BF relationship. Individuals with relatively long legs or relative low sitting

height have less mass per unit length, thus have a lower BMI (Norgan, 1994a,

1994b). A lower relative sitting height of 0.01 coincided with a lower BMI of 0.88

kg/m2, with Caucasians having a lower relative sitting height than Asians, indicative

of a higher relative leg length (Gurrici et al., 1999; Norgan, 1994a). Blacks have

longer appendicular bone lengths relative to height compared with whites

(Gallagher et al., 1996) and similarly, Asians differ in terms of body build. For

example, Chinese Indonesians have a higher relative sitting height than Malay

Indonesians (Gurrici et al., 1999).

Thirdly, a difference in frame size can also contribute to differences in the BMI‐%BF

relationship. Individuals of slender build are likely to have less bone, muscle and

connective tissue, indicative of a higher fat mass than a stocky person for a given

height and weight, thus having a higher %BF at the same BMI (Craig et al., 2001;

Deurenberg et al., 1999; Gurrici et al., 1999). Compared with Caucasians, Asians

have a smaller frame (higher slender index (height/sum of wrist and knee widths)).

Deurenberg et al. (1999) indicated that both Singaporean and Beijing Chinese had a

higher slender index than Dutch Caucasians, indicating a smaller frame (Deurenberg

et al., 1999). Even within Asian populations there is variability in frame size. For

example, Guricci et al. (1999) found that Malay Indonesians had a more slender

body build in terms of skeletal widths compared to the Chinese Indonesians, that is,

they had a higher slenderness index, and Singaporean Chinese had a more slender

frame than Beijing Chinese (Deurenberg et al., 1999). Polynesians also have a

different frame size to whites (Craig et al., 2001) with Tongans having significantly

greater elbow width, mid‐arm muscle area, and significantly lower %BF than

Australians.

In summary, the impact of differences in body build and relative leg length on

ethnic difference in the BMI‐%BF relationship can be explained as follows.

Deurenberg et al. (1999) indicated that after correction for differences in %BF,

slenderness and relative sitting height, the differences between measured and

predicted %BF compared to the Dutch group decreased from 1.4±0.8% (not


82

statistically significant) to 0.2±0.5% (not statistically significant ) in Beijing and from

4.0±0.8% (P <0.05) to 0.3±0.5% (not statistically significant) in Singapore. Gucci et al.

(1999) also showed that after correction for differences in body build and relative

sitting height, the difference in BMI between Malay Indonesians and Chinese

Indonesians at the same BMI, age and sex, decreased from 1.7±0.3 kg/m2 to 0.9±0.4

kg/m2.

Figure 2.5. The effect of relative leg length and frame size on the BMI‐%BF

relationship. Subject A has the same %BF as subject B, but because he has shorter legs his BMI will be higher (more mass per cm length in the trunk). Subject C has the same BMI as subject D, but because his frame is bigger (stockier) he will have more skeletal mass, more muscle mass and more connective tissue. Therefore, for the same BMI he will have a lower %BF. Adapted from Deurenberg et al. (2002).

A number of studies have reported no differences in the BMI‐%BF relationship

across ethnic groups in both children and adults. For example, Gallagher et al. (1996)

found no significant difference in the BMI‐%BF relationship between Caucasians and

Afro‐Americans living in New York after controlling for age and sex. Similarly,

Deurenberg et al. (1997) found no difference between Dutch Caucasians and Beijing

Chinese. Similarly, Wichramasinghe et al. (2005) found no significant ethnic


83

difference in the relationship in comparisons between Australian white Caucasians

and Australian Sri Lankan children. There are some logical explanations for the

conflicting findings. Firstly, results may be biased owing to violations of assumptions

that could differ for each study group (Deurenberg & Deurenberg‐Yap, 2001). For

example, the basic principle of densitometry is the assumption that the body

consists of two components, FM and FFM, each with a constant density of 0.9 kg/L

and 1.1kg/L, respectively. However, the density of FFM differs across ethnic groups,

therefore the Siri formula normally used to calculate the %BF from density may

underestimate the %BF significantly (Côté & Adams, 1993; Deurenberg‐Yap,

Schmidt, van Staveren, Hautvast, & Deurenberg, 2001). Secondly, environmental

factors may have an impact on ethnic differences in the BMI‐%BF relationship.

People of the same ethnic origin living in different countries tend to display a

different BMI‐%BF relationship (Deurenberg, Deurenberg‐Yap et al., 2003; Luke et

al., 1997; Navder et al., 2009). Therefore, when comparing the %BF among different

ethnic groups living in the same place, the ethnic difference might be attenuated.

Ethnic differences in the BMI‐%BF relationship result in differences in the ability of

BMI to identify children with excess body fatness. Freedman et al. (2008) showed

that the ability of the CDC 95th percentile to identify girls with excess body fat

varied by ethnicity. Of the girls with excess body fat, 89% of blacks, 88% of Hispanics,

72% of whites, but only 50% of Asians, were overweight. More black children (50%)

who had a BMI for age between the 85th and 95th percentiles had normal body

fatness compared with whites (27%), Hispanics (33%) and Asians (23%). Duncan et al.

(2009) also indicated that the sensitivity and specificity of both the CDC and IOTF

criteria were dependent on ethnic group. East Asian girls showed the lowest

sensitivity (65.7%) and Pacific Island girls showed the lowest specificity (42.6%),

indicating that more than a third of the East Asian girls with excess body fatness

were not identified as overweight, whereas over half of the Pacific Island girls with

normal levels of body fat were incorrectly categorized as overweight. A significant

increase in the prevalence of overweight in East and South Asian girls, and a

significant decrease in Polynesian girls was found when ethnic‐specific percentiles


84

were used. Table 2.9 shows the BMI values across other ethnic groups that

correspond to the same %BF in Caucasians.

Table 2.9. Comparison of BMI in whites with other ethnic groups reflecting the same %BF

Adults Children

Polynesians Taiwanese White

Male Female

Male FemaleSingaporeans Polynesians

15 17.0

20 19.3 19.8 22.6

25 25.7 25.7 23.6 22.7 21 28.3

30 32.1 31.6 25.3 24.8 27 34.0

35 38.4 37.4 39.6

40 44.8 43.4 45.3

Polynesia adults (Swinburn et al., 1999); Taiwanese (Chang et al., 2003); Singaporeans (Deurenberg‐Yap et al., 2000); Polynesian children (Rush et al., 2003).

On the basis of these findings, it has been suggested that BMI cut‐points should be

lowered among Asian adults (Misra, 2003). This would result in BMI cut‐points

identifying similar levels of body fatness and possibly disease risk, across ethnic

groups. Accordingly, the WHO proposed additional BMI cut‐off points of 23 kg/m2,

27.5 kg/m2, 32.5 kg/m2 and 37.5 kg/m2 for overweight, obesity Class I, obesity Class

II and obesity Class III for the Asian adult population in 2004, which are lower than

for the Caucasian population (corresponding to 25 kg/m2, 30 kg/m2, 35 kg/m2, and

40 kg/m2) (WHO, 2004). However, a comprehensive overview of the ethnic

differences between BMI and %BF has not been established among children and

adolescents and this makes it difficult to establish a universal obesity classification

based on BMI during these stages of growth and development.


85

2.4 BODY COMPOSITION ASSESSMENT IN CHILDREN

Body composition refers to the characteristic size and distribution of the component

parts of the total body weight. Body composition analysis involves subdividing body

weight into two or more compartments according to elemental, chemical,

anatomical, or fluid components. The two‐component model in which body weight

is divided into FM and FFM is the most widely used model in adults. Although the

two‐component model can not give information regarding the nutritionally

important total body protein and mineral component, it is useful in older children in

whom the focus of interest is FM and FFM.

Technological advances in recent decades have increased the range of opportunities

for the assessment of the human body. Assessment methods range from simple and

inexpensive field methods to highly complex and expensive laboratory procedures.

Ideally, evaluation of obesity should be made using such indirect measures as

hydrostatic weighing, DXA, air‐displacement plethysmography (BOD PODTM), TBW by

deuterium dilution, CT, and magnetic resonance imaging (MRI) (Lonzer et al.), but

these techniques require sophisticated laboratory settings and/or may be unsuitable

for some participants. Further, it is impractical to use precise and sophisticated

laboratory methods on large samples. Accordingly, the most popular methods of

body composition assessment in the field include BIA and anthropometric measures,

including BMI and skin folds.

2.4.1 Bioelectrical impedance analysis

BIA has been identified as one of the most appropriate methods for measuring

paediatric body composition in the field because the measurement is fast,

non‐invasive, inexpensive, painless, requires minimal participant burden, does not

require a high level of technical skill, and can be used to estimate body composition

in obese individuals (Ellis, 2000; Jürimäe & Hills, 2001; Mattsson & Thomas, 2006;

Norgan, 2005). Moreover, compared with skin fold measures, BIA has a far better

intra‐observer and inter‐observer reliability (Schaefer, Georgi, Zieger, & Scharer,


86

1992).

2.4.1.1 Assumptions and principles

In brief, the underlying principle of this method is that when an electrical current

passes through the body it will mainly passes through water‐containing tissues since

bone and fat have a large impedance (Z) and do not conduct significant current.

When the volume of TBW is large, the current flows more easily through the body

with less resistance (R). The resistance to current flow is greater in individuals with

large amounts of body fat. BIA measures the impedance and resistance of the small

alternating current through the body and then assesses total body water. Certain

basic assumptions about the geometric shape of the body and the relationship of

impedance to the length and volume of the conductor are made (Heyward, 1998;

Kyle et al., 2004; Mattsson & Thomas, 2006):

(1) The human body is a simple cylinder of a known length and cross‐sectional

area, that water and electrolytes are uniformly distributed and that body

temperature is constant.

(2) At a fixed signal frequency, the impedance (Z = (R2 + Xc2)1/2) to current flow

through the body is directly related to the length (L) of the conductor

(height) and inversely related to its cross‐sectional area (A): Z = p(L/A) (p is

the specific resistivity of the body’s tissues and is assumed to be constant)

and volume of the body then be calculate as V = pL2/Z. Because the water

content of the FFM is relatively large and constant (73% water for adults),

FFM can be predicted from TBW estimates (FFM = TBW/hydration

constant).

2.4.1.2 BIA methods

By the 1970s, with the establishment of the foundations of BIA, a variety of single

frequency BIA analyzers became commercially available, and by the 1990s, several

multi‐frequency analyzers were available. Since then other methods of BIA have

been developed as detailed by Kyle et al. (2004), such as segmental‐BIA,


87

bioelectrical spectroscopy (Shaikh, Mahalanabis, Kurpad, & Khaled, 2002), localized

BIA, bioelectrical impedance vector analysis (BICA or vector BIA).

BIA at a single frequency (usually 50 kHz) or at multiple frequencies is one of the

more common techniques applied to the measurement of body composition. Single

frequency BIA, involves a current being passed between surface electrodes located

on hand and foot. Some use other locations, such as foot‐to‐foot (Jebb, Cole, Doman,

Murgatroyd, & Prentice, 2000; Utter, Nieman, Ward, & Butterworth, 1999) or

hand‐to‐hand electrodes. The resistance or impedance to the flow from two source

electrodes is inversely proportional to the total body weight and the RI is directly

related to the volume of TBW (Simpson et al., 2001). This enables TBW to be

predicted, and subsequently FFM and FM derived by difference. Prediction of TBW

from single frequency BIA measures is reasonably accurate (Sun et al., 2003)

however single frequency BIA is limited in its ability to differentiate body water into

intra‐ and extracellular compartments. In contrast, multi‐frequency BIA is able to

differentiate between compartments and uses empirical linear regression models as

with single frequency BIA, but includes impedances at multiple frequencies. Such

devices can evaluate FFM, TBW, intracellular water (ICW) and extracellular water

(ECW) by using different frequencies based on the theory that the proportion of the

alternating current which flows through the intracellular and extracellular pathways

is frequency dependent (low frequency to measure ECW and high for ICW). The

ability of MF‐BIA to distinguish TBW into ECW and ICW is potentially important to

describe fluid shift and balance and to explore the variations in levels of hydration

(Chumlea & Guo, 1994). These two BIA approaches have been widely used both in

children and adults.

2.4.1.3 Sources of measurement error

BIA has been shown to be highly reliable over repeated trials and for repeated

measurements within a day and over several days or weeks for inter‐observer and

intra‐observer comparisons with standardized measurement techniques. As

reviewed by Kyle et al. (2004), reported mean coefficients of variation for within‐day


88

R measurements are ≈1‐2%; daily or weekly intra‐individual variability is slightly

larger ranging from ≈2% to 3.5%. Day‐to‐day coefficients of variation increases for

frequencies lower than 50 kHz. Overall reproducibility/precision is 2.7‐4.0%.

Prediction errors were estimated to be 3‐8% for TBW and 3.5‐6.0% for FFM,

respectively. However, the accuracy and precision of the BIA measurements are

affected by several factors, including instrumentation, body position, electrode

placement, subject factors (eating, being dehydrated, exercise) and environmental

temperature (Heyward, 1998; Hills, Lyell, & Byrne, 2001; Huoutkooper, Lohman,

Going, & Howell, 1996).

Resistance differs within different impedance instruments from different

manufacturers or different types from the same company. Deurenberg et al. (1989)

observed differences in readings while measuring the same subjects of 7‐16 Ω.

Therefore, Baumgartner et al. (1989) recommended that impedance measurements

be made with the same type of analyzer used in developing the prediction equation.

Moreover, left side measurement is generally higher with mean differences of 7 Ω

(Lukaski, Johnson, Bolunchuk, & Lykken, 1985) and 12 Ω (Houtkoper, Lohman, Going,

& Hall, 1989). Furthermore, a slightly different electrode placement on the foot and

hand could cause important differences in measured body impedance. Deurenberg

et al. (1991) found that replacement of the electrodes on the hand with only 2 cm

difference proximal to the trunk resulted in a mean 26 Ω decrease in body

impedance. In addition, factors altering the individual’s hydration state can affect

the total body resistance. For example, moderate/vigorous exercise (Deurenberg,

Weststrate, Paymans, & van der Kooy, 1988; Khaled et al., 1988) will decrease the

resistance. Performing measurements after food intake results in an expansion of

the TBW pool by metabolic water produced during the postprandial state (Lohman

et al., 2000) and decreases the resistance by 13‐17 Ω (Deurenberg et al., 1988).

Room temperature can also influence skin temperature of participants (Caton, Molé,

Adams, & Heustis, 1988; Gudivaka, Schoeller, & Kushner, 1996). Gudivaka et al.

(1996) indicated that the change in proximal impedance per degree centigrade

change in skin surface temperature was approximately 60% of distal impedance and


89

the change in measured impedance at 50 kHz erroneously over‐predicted TBW by

2.6±0.9 L (P<0.001) and underestimated FM by 3.3±1.3 kg (P<0.0001). The most

appropriate room temperature is 20‐22℃. Therefore, BIA measurements must be

standardized in order to obtain reproducible results.

2.4.1.4 BIA prediction equations

During the past decade, numerous studies have contributed to the development of

equations to predict body composition from impedance in children. As summarized

by Nielsen et al. (2007), from 1988 to 2004, 45 studies were published on the

development of BIA predictive equations among healthy children. Of these studies,

22 were conducted in the United States, two in New Zealand, three in Asia (Eston et

al., 1993; Kim, Tanaka, Nakadomo, & Watanabe, 1994; Masuda & KomIya, 2004),

two in South America, one in Nigeria and 15 in Europe. Recently, more BIA

equations have been published (Going et al., 2006; Haroun et al., 2010; Kriemler et

al., 2009; Nielsen et al., 2007; Sluyter et al., 2010; Wickramasinghe, Lamabadusuriya,

Cleghorn, & Davies, 2008).

However, no consistent methodology for the development of predictive equations

can be found in the literature with different reference methods identified. As listed

by Nielsen et al. (2007), twenty‐eight studies predicted FFM, one predicted FM, six

predicted %BF, sixteen predicted TBW and two predicted extracellular water. The

predictive equations were derived by different criterion methods; 16 studies used

deuterium dilution, 16 studies used hydrostatic weighing, 12 used DXA, seven used

isotopic dilution (18O), two used bromide dilution, one used 40K spectrometry and

one used skin fold thickness. The use of different reference methods is due to the

absence of a gold standard method for obtaining in vivo reference measures of body

composition. Although the methods mentioned above are often used as reference

methods, they provide only indirect measurement of body composition and,

therefore, are subject to measurement error. In light of this limitation, Lohman

(1992) developed standards for evaluating prediction errors for body composition

prediction equations estimating %BF and FFM (Table 2.10).


90

Table 2.10. Standards for evaluating prediction errors

SEE %BF SEE FFM(kg) Subjective

Male and female Male Female rating

2.0 2.0‐2.5 1.5‐1.8 Ideal

2.5 2.5 1.8 Excellent

3.0 3.0 2.3 Very good

3.5 3.5 2.8 Good

4.0 4.0 2.8 Fairy good

4.5 4.5 3.6 Fair

5.0 > 4.5 > 4.0 Poor

Further, from a statistical perspective, equations have been based on different

variables (including R, Xc, Z, height, weight, age, puberty, gender, race, skin fold

thickness, arm length and shoulder height). Among these independent variables,

the resistance index (RI = Ht2/R), instead of Ht2/Z, is considered to be the best

predictor of FFM and often used in many BIA prediction equations (Houtkooper et

al., 1996; Lukaski et al., 1985; Segal, Gutin, Presta, Wang, & Van Itallie, 1985). The

reason is that typically, R (a measure of pure opposition to current flow through the

body) is more than 10 times larger than reactance (Xc, the opposition to current

flow caused by capacitance produced by the cell membrane) (at a 50 kHz frequency)

when whole‐body impedance is measured. Therefore, R alone provides an accurate

approximation of Z. As shown in studies among adults, resistance or

height2/resistance have higher correlation coefficients than impedance, reactance,

age, sex, body mass and BMI. Typically, the accuracy of predicting TBW, FFM or %BF

is improved when body weight is included as a predictor in the equation, and also

when age and sex are included (Houtkooper et al., 1996). Body weight is correlated

with FFM so that the inclusion of body weight with a positive regression coefficient

is to be expected (Bunc, 2001).


91

2.4.1.5 Choice of BIA prediction equations

As mentioned above, BIA depends on several assumptions regarding the

composition of the fat and the fat‐free body. These include constancy in the

relationships among these body compartments. However, the level of hydration

and the density of the FFM are not constant among individuals and vary according

to age, gender and ethnicity (Deurenberg et al., 2002). Therefore, BIA equations

should be age‐, gender‐, and ethnicity‐specific and BIA accurately predicts %BF in

children and adults provided that an appropriate prediction equation is utilized. The

problems inherent in the application of existing prediction equations to other

samples are that the individuals should have the same physical characteristics as

the sample used to derive the equation.

As reported by Reilly (1998), the FFM of children has lower density, lower

mineralisation, higher water content and lower potassium levels than that of adults

and therefore lead to an overestimation of body fat if adult values were utilised in

prediction equations. Similarly, failing to adjust for differences in FFM density in

ethnic groups may result in systematic biases of up to 3% (Deurenberg &

Deurenberg‐Yap, 2003). Nielsen et al. (2007) cross‐validated all the previously

published equations and revealed significant differences between FFM (TBW)

predicted by almost all equations and DXA in his study sample (9‐ to 11‐year‐old

Swedish children). Wickramasinghe et al. (2005) used TBW as a reference method

determined by deuterium dilution to validate six BIA equations and the mean

differences between FM predicted by each equation and assessed by isotope

dilution technique ranged from 0.3‐4.4 kg.

As shown in Table 2.11, most existing equations have been derived from Caucasian

individuals. Therefore, it is easy to find an equation for white children and a number

have been recommended (Heyward & Stolarczyk, 1996), including Houtkooper et al.

(1992) for boys and girls 10‐19 y, and Lohman (1992) or Kushner et al. (1992) for

children younger than 10 y. These equations were developed using a

three‐compartment model and have prediction errors of 2.1 kg or less. In contrast,


92

to date only several studies have developed BIA equations specifically for Asian

children. Kim et al. (1994) developed BIA equations from 141 Japanese boys aged

9‐14 y using hydrostatic weighing. Eston et al. (1993) developed six BIA equations

from 94 Chinese children (46 boys and 48 girls) aged 11‐17 y using skin fold

methods as the reference. Masuda et al. (2004) derived a BIA equation from 20

boys and 26 girls aged 3 to 8 y using deuterium oxide (D2O) as the criterion.

However, in each case the sample size was relatively small and no equation has

been derived from multiple Asian countries. It is well known that Asian children

have a different body build to Caucasians and other ethnic groups (Deurenberg,

Deurenber‐Yap, & Schouten, 2002). Therefore, ethnic‐, age‐ and gender‐specific

impedance‐based equations for body composition assessment in Asian children are

urgently required, particularly given the burgeoning paediatric obesity epidemic.


93

Table 2.11. Some BIA prediction equations for children and adolescents

Reference Ethnicity/gender Equation

White

Kushner et al. (1992) Boys and girls (6‐10 y, USA) TBW = 0.593×HT2/R + 0.065×BW + 0.04

Davies et al. (1988) Boys and girls (5‐17 y, UK) TBW = ‐0.50 + 0.60×HT2/Z

Houtkooper et al. (1992) Boys and girls (10‐19 y, USA) FFM = 0.61×HT2 /R + 0.25×BW + 1.31

Lohman (1992) Boys and girls (8‐15 y, USA) FFM = 0.62×HT2 /R + 0.21×BW + 0.10×Χc + 4.2

Deurenberg et al. (1990) Boys and girls (7‐9 y, The Netherlands) FFM = 0.64×HT2 /Z + 4.83

Boys (10‐15 y) and girls (10‐12 y) FFM = 0.488×HT2 /Z + 0.221×BW + 0.1277×HT – 14.7

Schaefer et al. (1994) Boys and girls (3‐19 y, Germany) FFM = 0.65×HT2 / Z + 0.68×Age + 0.15

Nielsen et al. (2007) Boys and girls (9‐11 y, Swedish) FFM = 0.54×HT2/R + 0.05×Xc + 0.06×HT + 0.09×BW + 0.97×Sex ‐ 5.11

Blacks

Lewy et al. (1999) Boys and girls (10.9±1.1 y) FFM = 1.10+0.84×HT2/R

Leman et al. (2003) Boys and girls (5‐18 y) TBW = 1.00 + 0.42×HT2/R + 0.18×BW

Asians

Kim et al. (1994) Boys (9‐14 yr, Japanese) FFM = 0.56×HT2/Z + 0.20×BW + 1.66

Masuda et al. (2004) Boys and girls (3‐6 yr, Japanese) TBW = 0.149×HT2 /R + 0.244×BW + 0.460×Age + 0.501×Sex + 1.628

Eston et al. (1993) Boys and girls (11.2‐17.1 yr, Chinese) FFM = 0.52×HT2/R + 0.28×BW + 3.25

TBW (kg) = total body water, FFM = fat‐mass free (kg), HT = height (cm), BW = body weight (kg), R = resistance (Ω), Χc = reactance (Ω), Z = impedance (Ω), Sex: male = 1, female = 0, Age: y. Adapted from Applied Body Composition Assessment (Heyward & Stolarczyk, 1996), Eston et al. (1993), Kim et al. (1994), Horlick (2002) and Nielsen et al. (2007).


94

2.4.2 Reference methods

According to Heyward (1998), good prediction equations have several

characteristics, the most important being the use of acceptable reference methods

to obtain criterion measures. The more sophisticated methods of hydrodensitomery,

DXA, TBW and total body potassium are so expensive for large scale studies that

they are generally considered as reference methods for the assessment of body

composition against which other assessment methods are validated (Hills et al.,

2001). Among these reference methods, TBW by isotope dilution is the gold

standard for measuring TBW and has an accuracy of 1‐2%. Further, the technique is

widely used as a reference method to validate other methods including BIA in both

children and adults (Bell, McClure, Hill, & Davies, 1998; Deurenberg, Tagliabue, &

Schouten, 1995; Lohman et al., 2000).

2.4.2.1 Principle and assumptions of TBW method

As reviewed by Ellis (2000), the basic principle of this method is that the volume of a

compartment can be defined as the ratio of the dose of 2H2O, administered orally,

to its concentration in that body compartment within a short time after the dose is

administered. Inherent in any tracer dilution technique are four basic assumptions

as follows (Schoeller, 2005):

1) The tracer is distributed only in body water;

2) The tracer is equally distributed in all anatomical water compartments (e.g.,

saliva, urine, plasma, sweat, human milk);

3) Tracer equilibration is achieved relatively quickly;

4) Neither the tracer nor body water is metabolized during the time of tracer

equilibration.

2.4.2.2 Equipment of TBW method

Three isotope tracers, deuterium (2H) (Lanham, Stead, Tsang, & Davies, 2001;

Masuda & Komiya, 2004), tritium (3H) and 18O (Phillips, Bandini, Compton, Naumova,

& Must, 2003), have been used to measure TBW. Tritium is a radioactive tracer and


95

is contraindicated in research with children. 2H is the method of choice today

because of the radiation hazards associated with the use of 3H and the relative

expense of using 18O.

2.4.2.3 Measurement procedures of TBW method

Technical errors associated with estimating TBW from isotope dilution include

variations in the physiological fluid measured, equilibration time of the isotope,

water changes during equilibration, correction for isotopic dilution space, and the

method for measuring the isotopic enrichment following equilibration (Lohman,

1992). Therefore, each aspect of the measurement, including subject preparation,

dosing, sample collection, and isotope analysis requires careful attention.

Subject preparation is somewhat dependent on the goals of the investigation. To

avoid over‐hydration and dehydration, subjects should: 1) have normal fluid and

food intake the day before measurement; 2) avoid vigorous exercise after the final

meal of the previous day; 3) avoid sweating excessively after the final meal of the

previous day; 4) have the final meal between 12 and 15 h before the dose; 5) not

drink for several hours before the test (Schoeller, 2005).

The dose should be: 1) weighed by a balance with a high precision and accuracy; 2)

stored in a screw‐capped container to minimize evaporation; 3) given with the

subjects fasting to maximize the rate of absorption; 4) given orally as this is less

invasive although the dose can be given intravenously. Moreover, the subjects

should be prevented from eating or drinking 1 h after the dose is consumed under

the assumption that the dose will have emptied from the stomach and yet there is

still time for the water in the meal to mix with the body water pool during the

equilibration period.

Physiological samples, typically including blood, saliva or urine, must be obtained

before the dose is consumed to determine the natural background level of the

isotope which occurs naturally in the body. Meanwhile, the physiological samples

must be collected over a period long enough to ensure equilibration. Typically, a


96

sufficient amount of time for equilibration is 3‐4 h for blood and saliva, and 5‐6 h for

urine (International Atomic Energy Agency, 2009).

Isotope ratio mass spectroscopy (IRMS) is often used to analyze the samples (Bell et

al., 1998; Thomas, Van Der Velde, & Schloeb, 1991; Wong et al., 1988) because

measurement with a precision of <1.0% can be achieved. The results are expressed

in delta units (%) relative to an international standard (standard mean ocean water).

The deuterium dilution space is determined from the equation as follows:

V (kg) = EpEs

EtEax

a

TA

)(

Where T is the amount of tap water in which the dose was diluted in grams, A is the

amount of dose taken by the participant in grams, a is the amount of the dose in

grams retained for mass spectrometer analysis, and Ea, Et, Ep and Es are the

isotopic enrichment in delta units relative to standard mean ocean water of the

dilute dose, the tap water used, the pre‐dose urine sample and the post‐dose urine

sample.

In fact, TBW will be overestimated by ~4% because of deuterium exchange with the

non‐aqueous hydrogen in the body (Racette et al., 1994). Therefore, the equation

needs to be corrected as follows:

TBW =V/1.04

Using TBW to estimate body composition, FFM is then calculated as:

FFM (kg) = TBW /hydration constant

Deriving FFM from TBW requires consideration of the hydration constant of FFM. In

healthy adults, TBW constitutes 73.2% of the FFM however the hydration of FFM is

not constant across the life span (Fomon et al., 1982) and varies according to health

(disease) status (Kotler et al., 1999). Children have a higher level of water content

than adults which will result in an underestimation of percent body fat if the

hydration constant of adults is used. Therefore, Lohman’s age‐ and gender‐specific

constants for hydration of the FFM have commonly been used in research with

children (Heyward & Stolarczyk, 1996). Subsequently, FM and %BF can be calculated


97

based on the two‐compartment model of body composition from the FFM:

FM (kg) = Body weight ‐ FFM

%BF = (Body weight ‐ FFM)/Body weight×100

2.4.2.4 Precision and accuracy of TBW method

As mentioned above, there are four basic assumptions in the measurement of TBW

by dilution. If any of these requirements is violated, then the ratio of administered

dose to fluid concentration must be adjusted. For the measurement of TBW,

corrections for overexpansion, non‐equilibrium, and excretion of the tracers are

needed (Ellis, 2000; Mattsson & Thomas, 2006). With careful attention to detail,

TBW can be measured from either repeat measurements or simultaneous

measurements using two isotopes with a precision of 1‐2% (Racette et al., 1994;

Speakman, Nair & Goran, 1993). The accuracy of TBW estimation is of 1‐2% and the

estimated error is typically <1 kg (Wang et al., 1999).

In general, the deuterium dilution technique is the most accurate of methods using

the 2‐C model for the assessment of body composition and it is relatively simple to

perform in the field. Therefore, TBW values obtained using the dilution technique

are considered the reference or criterion values for comparison with alternate

measurement techniques including BIA in both children (Masuda & Komiya, 2004;

Phillips et al., 2003; Shaikh et al., 2002) and adults (Deurenberg, Deurenber‐Yap &

Schouten, 2002; Lanham et al., 2001). In the studies predicting TBW by BIA, values

for R2 ranged from 0.92‐0.99, SEEs ranged from 0.3 to 1.8 kg when the criterion

TBW values were measured by the isotope dilution method (Huoutkooper et al.,

1996).


98

2.5 OBESITY AND THE METABOLIC SYNDROME IN CHILDREN

The metabolic syndrome describes the clustering of central obesity, dyslipidaemia

(raised triglycerides and/or low‐ or high‐density lipoprotein), hyperinsulinaemia,

impaired glucose tolerance and elevated blood pressure (Alberti et al., 2005). People

with the metabolic syndrome are two to three times as likely to have a heart attack

or stroke and five times as likely to develop type 2 diabetes compared with people

without the syndrome, therefore it is a clear indicator of adult morbidity and

all‐cause mortality (Alberti et al., 2005; Haffner et al., 1992; Isomaa et al., 2001;

Trevisan, Liu, Bahsas & Menotti, 1998). Paediatric metabolic syndrome can also

increase cardiovascular risk (Ronnemaa et al., 1991) and can track from childhood to

adulthood (Duncan, Li & Zhou, 2004).

It has been estimated that a quarter of the world’s adult population has the

metabolic syndrome (International Diabetes Federation, 2010). Unfortunately, the

condition is increasingly common in children and adolescents, mainly due to the

growing obesity epidemic within this population (Cook et al., 2003; Weiss et al.,

2004). Data from NHANES III indicate that nearly one in ten adolescents in the US

have three or more of the risk factors involved in the metabolic syndrome (de

Ferranti et al., 2004). The prevalence of the metabolic syndrome in Chinese

adolescents was estimated to be 3.3% in 2002 (Li, Yang et al., 2008). Given the

increasing prevalence of the syndrome in children and adolescents, greater

attention is being paid to research in the area. However, the metabolic syndrome

has been described in many ways including using different obesity indices, which in

large part is due to the lack of a “gold standard” diagnostic test (Table 2.12). This

situation has contributed to differences in the reported prevalence and incidence of

the metabolic syndrome in several studies (Golley et al., 2006; Goodman et al., 2004;

Reinehr et al., 2007; Weiss et al., 2004). Accordingly, it is impossible to estimate the

global prevalence of the metabolic syndrome and also to make valid comparisons

between countries.


99

Table 2.12. Metabolic indicators and cut‐off points used to classify paediatric metabolic syndrome

Indicator criteria

Definition of metabolic syndrome Glucose

(mmol/L)

Triglycerides

(mmol/L)

HDL‐C

(mmol/L)

SBP/DBP

(mmHg)

Insulin

(pmol/L)

BMI WC

1 IDF for children aged 10 to <16 years (Zimmet et al., 2007)

5.6 1.7 1.03 130/85 ‐ ‐ P‐90%

2 IDF for children aged 16+ years (Zimmet et al., 2007)

5.6 1.7 M: 1.03 F: 1.29 130/85 ‐ ‐ Europids and Arab: M 94 cm; F 80 cm Asians: M 90 cm; F 80 cm

3 Lambert for children (Lambert et

al., 2004)

6.1 M: 0.85 (9 yr) 1.05 (13 yr) 1.08 (16 yr) F: 0.96 (9 yr) 1.07 (13 yr) 1.18 (16 yr)

M: 1.22 (9 yr) 1.11 (13 yr) 1.00 (16 yr) F: 1.20 (9 yr) 1.13 (13 yr) 1.13 (16 yr)

P‐75% height M: 35.01 (9 yr) 60.04 (13 yr) 50.70 (16 yr)F: 40.64 (9 yr) 69.92 (13 yr) 62.76 (16 yr)

P‐85% ‐

4 Adult NCEP ATP III modified for children aged 12‐19 years by Cook et al. (2003)

6.1 1.24 1.03 P‐90% height ‐ ‐ P‐90%

5 Adult NCEP ATP III modified for children aged 12‐19 years by de Ferranti et al. ( 2004)

6.1 1.1 1.3 P‐90% height ‐ ‐ P‐75%

6 Weiss for children and adolescents (4‐20 y) (Weiss et al., 2004)

IGT: 7.8‐11.7 P‐95% age, sex and ethnicity

P‐5% age, sex and ethnicity

P‐95% height ‐ BMI‐Z score ≥ 2

7 WHO 6.1 or IGT: 7.8

1.75 0.9 P‐95% P‐95%

IDF: International Diabetes Federation; IGT: Impaired glucose tolerance; NCEP ATP III: National Cholesterol Education Program‐Third Adult Treatment Panel.


100

Obese children have a higher risk of the paediatric metabolic syndrome compared

with normal‐weight children. For example, Goodman et al. (2004) indicated that the

prevalence of the NCEP‐defined metabolic syndrome was 4.2% while the prevalence

of the WHO‐defined metabolic syndrome was 4.8% among 1513 black, white and

Hispanic teens, but among obese teens, the prevalence was 19.5% and 38.9%,

respectively. De Ferranti et al. (2004) also showed that nearly one third of

overweight/obese adolescents had the metabolic syndrome, much higher than

when compared to the total US adolescent population aged 12‐18 years (9.2%).

Obesity alone can increase the risk of cardiovascular disease, including hypertension

and dyslipidaemia. Perichart‐Perera et al. (2007) indicated significantly higher WC,

systolic blood pressure (SBP), insulin resistance indices, and triglyceride (TG) levels

among the obese when compared with normal‐weight Mexican children aged 9‐12

years. Similarly, Burke et al. (2005) studied 741 boys and 689 girls aged 8 years and

reported that overweight and obese children had 6 mm higher SBP and 2 mm

higher diastolic blood pressure (DBP), 8% lower high‐density lipoprotein cholesterol

(HDL‐C), and 27% higher TG than normal‐weight children. Moreover, childhood

obesity is not only associated with the earlier development of the syndrome and

metabolic abnormalities in children but also appears to increase the likelihood of

developing obesity, the syndrome and abnormal lipids in adults (Freedman, Khan,

Dietz, Srinivasan & Berenson, 2001; Sinaiko, Donahue, Jacobs & Prineas, 1999;

Srinivasan, Bao, Wattigney & Berenson, 1996; Steinberger, Moran, Hong, Jacobs &

Sinaiko, 2001). The risk of the metabolic syndrome was 2.9 (95% confidence interval

(CI) 1.1 to 7.6) for adults who had been obese as children compared with those who

had not been obese as children (Vanhala, Vanhala, Keinänen‐Kiukaanniemi,

Kumpusalo & Takala, 1999; Vanhala, Vanhala, Kumpusalo, Halonen & Takala, 1998).

In addition, a number of prospective studies have also identified obesity as the

central feature of the metabolic syndrome among adults (Miaison, Byrne, Hales, Day

& Wareham, 2001; Palaniappan et al., 2004). In summary, each of the definitions of

the metabolic syndrome in both children and adults include obesity (Cook et al.,

2003; de Ferranti et al., 2004; Expert Panel on Detection, 2001; Weiss et al., 2004)

and similarly, in the IDF metabolic syndrome definitions for both children and adults,


101

obesity is the central component (Alberti et al., 2005; Zimmet et al., 2007).

However, despite years of research there is still much uncertainty regarding the

best cut‐points of obesity indices, and which measure or combination of measures

of excess weight has the greatest discriminatory capability for cardiovascular risk

factors. BMI, %BF, WC, WHtR, and waist‐to‐hip ratio (WHR) have been used to

define the metabolic syndrome in children and adolescents (Lobstein et al., 2004).

For example, the Quebec Family Cohort Study (Katzmarzyk et al., 2001) used skin

fold measurements, WHO recommends BMI (Alberti, Zimmet, & WHO Consultation,

1998), whereas the WC is included in NCEP (de Ferranti et al., 2004) and IDF

(Zimmet et al., 2007) definitions of the metabolic syndrome. This situation has

contributed to differences in the reported prevalence and incidence of metabolic

syndrome in several studies (Golley et al., 2006; Goodman et al., 2004; Reinehr et al.,

2007; Weiss et al., 2004).

Central fat distribution is reported to constitute a greater metabolic risk than

peripheral distribution, and WHR and WC are the two commonly used

anthropometric indices of abdominal visceral adipose tissue mass in children and

adults. WC in children, as is the case in adults, is an independent predictor of insulin

resistance, lipid levels, and blood pressure (Bacha et al., 2006; Lee et al., 2006; Ng et

al., 2007). Moreover, in young people who are obese and have a similar BMI, insulin

sensitivity is lower in those with high visceral adipose tissue and WHR than in those

with low values (Bacha et al., 2006). In addition, because a high WC can persist to

adulthood this measure is included in many definitions (Cook et al., 2003; de

Ferranti et al., 2004; Zimmet et al., 2007).

Despite the acceptance of WC as a predictor of cardiovascular disease risk in

children and adolescents, many studies have indicated that WHtR is a better

predictor than WC and BMI. Hara et al. (2002) compared BMI, WC, WHR, %BF, and

WHtR as indices to evaluate clustering of cardiovascular disease risk factors in

Japanese schoolchildren aged 9‐13 years. The results showed that among the

anthropometric indices, WHtR was the most significant predictor of cardiovascular

disease. Savva et al. (2000) also reported that WC and WHtR were better predictors


102

of cardiovascular disease risk than BMI in Greek‐Cypriot children aged 10‐14 years.

Similar results have been observed in adults (Hsieh, Yoshinaga & Muto, 2003).

Moreover, in adults, there is a known advantage to the establishment of a unisex

cut‐off point (0.5) when using the WHtR as a predicator of CV risk. For children and

adolescents, height continues to increase throughout the growing years and into

young adulthood and height influences the magnitude of WC. Therefore, a single

cut‐off point independent of age and gender could be set for children and

adolescents since the WHtR takes into account height. It is much simpler and more

practical for epidemiological and clinical setting use compared with BMI and WC.

Previous studies have also indicated there was a marginal difference between age

groups and sex (Hara et al., 2002; McCarthy & Ashwell, 2006; Sung et al., 2008).

Some studies have indicated that BMI is a better predictor of the metabolic

syndrome while others consider that WC or WHtR are superior (Yeh et al., 2005).

Children with a higher BMI have impaired glucose tolerance (Sinaiko, Steinberger,

Moran, Prineas & Jacobs, 2002; Sinha et al., 2002), and plasma leptin and BMI in

combination seem to be significant predictive markers of the metabolic syndrome

among children (Chu, Wang, Shieh & Rimm, 2000). These results suggest that BMI

as an index of total body fatness is related to insulin resistance and that the index

may be useful for screening children and adolescents for diabetes risk (Heymsfield

et al., 2005).

While the choice between WC or WHtR and BMI remains a matter of ongoing

debate, direct assessment of fat mass may be a better means of assessing

obesity‐related health risk. However, studies that have compared values for %BF and

BMI in the prediction of metabolic risk are contradictory in adults with some

indicating that when compared to %BF, BMI was similar or even more closely

associated with cardiovascular risk factors (Bosy‐Westphal et al., 2006; Tulloch‐Reid,

Williams, Looker, Hanson & Knowler, 2003). In contrast, the value of %BF was

deemed to be greater than BMI in other studies (Lahmann, Lissner, Gullberg &

Berglund, 2002; Nagaya, Yoshida, Takahashi, Matsuda & Kawai, 1991). One

explanation for these discrepant results may be that the data suffer from the


103

methodological drawbacks of field methods used for body composition assessment

(Bosy‐Westphal et al., 2006). Relatively few studies have compared %BF with

anthropometric variables to predict metabolic risk in children and adolescents and

the findings are contradictory (Daniels et al., 1999; Moussa et al., 1994). However, a

number of studies have reported that %BF or total body fat is associated with CV risk

factors in children (Johnson, Figueroa‐Colon, Huang, Dwyer & Goran, 2001; Ondrak,

McMurray, Bangdiwala & Harrell, 2007; Ribeiro et al., 2004).

In addition, a number of studies have indicated that trunk skin folds may be a

particularly sensitive tool to detect metabolic abnormalities in children and

adolescents (Maffeis et al., 2001; Misra et al., 2006; Teixeira et al., 2001). In these

studies, individual trunk skin folds or the sum of several skin folds were stronger

predictors of metabolic variables than WC.

However, results in a number of studies indicate a similar association of obesity

indices with CV risks (Garnett, Baur, Srinivasan, Lee, & Cowell, 2007; Huxley et al.,

2008; Lee, Song, & Sung, 2008; Plachta‐Danielzik, Landsberg, Johannsen, Lange, &

Müller, 2008). The correlations between each obesity index and metabolic risk

factors are similar and the area under the curves (AUCs) from receiver operating

characteristic (ROC) analysis for BMI, WC, WHtR and %BF showed marginal

differences in each measure’s capability for the prediction of metabolic risks.

Therefore, Plachta‐Danielzik et al. (2008) recommended the use of a second obesity

index in children with normal BMI or normal WC as well as in adolescents with

elevated WC.

Due to the inconsistency of obesity indices included in the definitions of the

metabolic syndrome in children and adolescents, the IDF has recommended that

further investigation of how obesity is defined in children should occur, for example

considering the respective value of measures such as WHtR, WC, etc (Zimmet et al.,

2007).

Besides the selection of the most appropriate obesity index in the metabolic

syndrome definition, ethnic‐specific criteria of obesity are another important issue


104

which should be considered (Misra et al., 2007). There are potential ethnic

differences in the strength of the associations between obesity and cardiovascular

risk factors (Arslanian, Suprasongsin & Janosky, 1997; Ehtisham et al., 2005; Huxley

et al., 2008; Ke, Brick, Cant, Li & Morrell, 2009; Whincup et al., 2002, 2005; Zhu et al.,

2005). The odds of prevalent hypertension associated with the same standard

increment in BMI and WC were stronger in Asians compared with Caucasians

(Huxley et al., 2008). Zhu et al. (2005) also found that the odds ratio for having one

or more metabolic risk factors was highest in obese white women, following by

Hispanic and black at a given BMI and WC. Hispanic obese men had higher risk for

having one or more metabolic risk factors than white and black men. Moreover,

there is an ethnic difference in the susceptibility in metabolic variables such that

South Asian children are less insulin sensitive than white Europeans, and are more

likely than white European children to show impaired fasting glucose and develop

type 2 diabetes. The differences persisted even after adjustment for adiposity and

pubertal status (Ehtisham et al., 2005; Whincup et al., 2005). Ethnic‐specific cut‐offs

of metabolic variables, not only WC but also lipid profiles should be proposed to

diagnose metabolic abnormalities.

In order to compensate for the variation in child development and ethnic origin,

percentiles, rather than absolute values of WC have been used in the IDF definitions.

Some studies have indicated that children who have a WC higher than the 90th

percentile were more likely to have multiple risk factors for cardiovascular disease

than were those with lower WC (Maffeis et al., 2001; Ng et al., 2007). Therefore, the

IDF and National Cholesterol Education Program‐Third Adult Treatment Panel (NCEP

ATP III) definitions for children and adolescents has chosen to use the 90th

percentile as a cut‐off for WC. In contrast, Sung et al. (2007) recently indicated that

among 1055 Chinese children from Hong Kong aged 6‐12 years, the optimal WC risk

threshold for predicting cardiovascular disease risk factors was the 85th percentile

and the age‐specific cut‐off values were smaller than for American children.

Similarly, Meng et al. (2007) found that for North Chinese children and adolescents,

the threshold was the 80th percentile (Table 2.13). Therefore, the IDF has further

recommended that the use of the percentile as a cut‐off for WC should be


105

reassessed when more data are available and ethnic‐specific age and sex normal

ranges for WC based on healthy values should be developed (Zimmet et al., 2007).


106

Table 2.13. Proposed waist circumference cut‐off values for predicting cardiovascular disease risk factors

Boys Girls Age (yr)

European‐ American

African‐ American

Mexican‐American

South Chinese

North Chinese

European‐ American

African‐ American

Mexican‐ American

South Chinese

North Chinese

2 50.6 50.0 53.2 ‐ ‐ 52.5 50.1 53.5 ‐ ‐

3 54.0 53.2 56.7 ‐ 52.2 55.4 53.8 56.7 ‐ 51.9

4 57.4 56.4 60.2 ‐ 53.6 58.2 57.5 59.9 ‐ 52.9

5 60.8 59.6 63.6 ‐ 55.1 61.1 61.1 63.0 ‐ 54.9

6 64.2 62.8 67.1 58.4 57.2 64.0 64.8 66.2 57.2 55.0

7 67.6 66.1 70.6 60.4 60.3 66.8 68.5 69.4 59.2 56.2

8 71.0 69.3 74.1 62.6 61.9 69.7 72.2 72.6 61.1 59.1

9 74.3 72.5 77.6 65.6 68.4 72.6 75.8 75.8 63.0 62.6

10 77.7 75.7 81.0 69.0 72.5 75.5 79.5 78.9 65.1 65.1

11 81.1 78.9 84.5 71.9 76.4 78.3 83.2 82.1 67.8 68.9

12 84.5 82.1 88.0 74.5 77.3 81.2 86.9 85.3 70.0 71.1

13 87.9 85.3 91.5 ‐ 78.3 84.1 90.5 88.5 ‐ 72.6

14 91.3 88.5 95.0 ‐ 78.4 86.9 94.2 91.7 ‐ 73.0

15 94.7 91.7 98.4 ‐ 81.5 89.8 97.9 94.8 ‐ 72.2

16 98.1 94.9 101.9 ‐ 81.5 92.7 101.6 98.0 ‐ 73.0

17 101.5 98.2 105.4 ‐ 81.7 95.5 105.2 101.2 ‐ 73.8

18 104.9 101.4 108.9 ‐ ‐ 98.4 108.9 104.4 ‐ ‐

Data for European‐American, African‐American and Mexican‐American population from Fernandez et al. (2004) (90th percentile); for South Chinese population from Sung et al. (2007) (85th percentile); for North Chinese population from Meng et al. (2007) (80th percentile).


107

CHAPTER 3 ETHNIC DIFFERENCE IN BODY COMPOSITION AMONG ASIAN CHILDREN FROM DIFFERENT ORIGINS

3.1 ETHNIC DIFFERENCES IN THE RELATIONSHIP BETWEEN BMI AND %BF AMONG ASIAN CHILDREN FROM DIFFERENT BACKGROUNDS

Modified from: Liu A, Byrne NM, Kagawa M, Ma G,Poh BK, Ismail MN, Kijboonchoo K, Nasreddine L, Trinidad TP, Hills AP (2011). Ethnic differences in the relationship between body mass index and percentage body fat among Asian children from different backgrounds. British Journal of Nutrition, 31 May 2011. doi:10.1017/S0007114511001681

3.1.1 Introduction

Obesity, commonly defined as an excess of body fat, is a global problem with rapid

increases seen in both developed and developing countries. BMI has achieved

international acceptance as a standard approach to define obesity however there

are a number of limitations in the use of this index as a measure of relative fatness,

including the inability to distinguish body fat and FFM. These limitations may be

even more important when attempting to compare individuals from different ethnic

groups.

Body composition seems to be ethnicity‐dependent (WHO, 2004). Many studies

have identified an ethnic variation in the relationship between %BF and BMI among

Caucasian and Asian adults (Deurenberg et al., 1998; Gurrici et al., 1998; Wang et al.,

1994). Some studies have also shown similar ethnic differences in the relationship

between BMI and %BF among white, black and Asian children (Daniels et al., 1997;

Deurenberg, Deurenberg‐Yap et al., 2003; Freedman et al., 2008; Going et al., 2006;

Mehta et al., 2002; Navder et al., 2009; Rush, Scraqq et al., 2009). White children

have a higher %BF than blacks for a given BMI after controlling for gender and

maturation stage, but a lower %BF than Asian children at the same BMI. However,

the relationship between BMI and %BF in children is further complicated by

variations in growth rates and maturity levels (Guo et al., 1997; He et al., 2004). In

addition, the BMI‐%BF relationship also differs among Asian adults from different


108

origins with Indians having the highest %BF and Chinese the lowest for the same

BMI when considering Chinese, Malays and Indians living in Singapore

(Deurenberg‐Yap et al., 2000). However, there is comparatively less data to indicate

differences in the BMI‐%BF relationship among Asian children from different origins.

To date, there has been no comprehensive overview of the ethnic differences

between BMI and %BF among children which makes it difficult to formulate a global

obesity classification system based on BMI with relevance for Asian children.

Therefore, despite the availability of a number of international classification systems

for paediatric obesity based on BMI (Cole et al., 2000; de Onis, 2007), national

variants still exist (Group of China Obesity Task Force, 2004; Matsushita et al., 2004).

In summary, a lack of data and also controversy around the optimal classification

system makes it difficult to monitor global and national trends, make comparisons

between studies, and to stratify for public health measures.

The objective of this study was to generate new knowledge on the body

composition characteristics of children from different Asian backgrounds.

3.1.2 Methodology

3.1.2.1 Participants

Five countries were involved in this study, including one East Asian country (China),

one West Asian country (Lebanon), and three South‐East Asian countries (Malaysia,

The Philippines and Thailand). In each country, a non‐random purposive sampling

approach was used which aimed to enrol children encompassing a wide BMI range

for each year of age between 8‐10 y and each sex. A total of 1039 participants (533

boys and 506 girls) aged 8‐10 y was involved in this study, including 352 Chinese

children (202 boys and 150 girls) living in Beijing, China, 155 Lebanese children (74

boys and 81 girls) living in Beirut, Lebanon, 197 Malay children (105 boys and 92

girls) living in Kuala Lumpur, Malaysia, 112 Filipino children (51 boys and 61 girls)

living in Manila, The Philippines, and 223 Thai children (101 boys and 122 girls) living

in Bangkok, Thailand. Ethnicity was determined by self‐identification and those


109

whose parents were identified as having the same origin were included. Additional

inclusion criteria required that participants be at Tanner stage 1 of puberty and free

from any diagnosed medical condition that might potentially interfere with body

composition measurement. Each participating country obtained ethical clearance

from appropriate ethics committees. The study protocol was explained to the

parent(s) and the children and written consent obtained from each child and/or

their parent(s).

3.1.2.2 Anthropometric measurements

Height was measured using a portable Holtain Stadiometer to the nearest 0.1 cm.

The technique for measuring height was as follows: the participant stood barefoot

with feet and heels together. The head was placed in the Frankfort plane which was

achieved when the Orbitale (lower edge of the eye socket) was in the same

horizontal plane as the Tragion (the notch superior to the tragus of the ear). When

the two landmarks were aligned, the Vertex was measured as the highest point on

the skull. The participant was instructed to take a deep breath while the measurer

applied gentle traction along the mastoid processes and the head piece of the

stadiometer was brought down firmly on the vertex at the same time and the

reading taken at this point.

Body weight was measured using a SECATM electronic scale (Hamburg, Germany) to

the nearest 0.1 kg. Participants were measured wearing only underwear after

urinating in the morning.

BMI was calculated as body weight (kg) divided by the square of height (m).

Overweight and obesity was defined on the basis of BMI for age and sex, as

recommended by the IOTF (Cole et al., 2000) (Table 3.1), WHO (de Onis, 2007)(see

Appendix 2.1) and Group of China Obesity Task Force (2004)(Table 3.2).

The pubertal status of each participant was assessed according to the criteria of

Tanner by trained investigators (Tanner & Whitehouse, 1976).


110

Table 3.1. BMI cut‐off points for overweight and obesity by sex between 8 and 10 y proposed by IOTF

Overweight Obesity Age (y)

Males Females Males Females

8 18.44 18.35 21.60 21.57

8.5 18.76 18.69 22.17 22.18

9 19.10 19.07 22.77 22.81

9.5 19.46 19.45 23.39 23.46

10 19.84 19.86 24.00 24.11

10.5 20.20 20.29 24.57 24.77

Table 3.2. BMI cut‐off points for overweight and obesity by sex between 8 and 10 y

proposed by the Chinese classification

Boys Girls Age(yr)

Overweight Obesity Overweight Obesity

8‐ 18.1 20.3 18.1 19.9

9‐ 18.9 21.4 19.0 21.0

10‐ 19.6 22.5 20.0 22.1

3.1.2.3 Body composition measurement

TBW was assessed using the deuterium dilution technique. A 5 mL sample of urine

was collected before consuming a dose of the isotope and this was used to

determine the basal deuterium level in the body. A 10% D2O dose of 0.5 g per kg

body weight was given orally and the dose measured using a weighing scale (Model

EG0620‐3NM, Laboratory Supply Company GmbH & Co. KG, Germany) to the

nearest 0.001 g. A second urine sample was collected about 5 h later thus allowing

complete equilibration within the body water compartments. Research groups from

each country administered the deuterium dose and collected samples using the

same standard operating procedure (see Appendix 2). The enrichment of the


111

pre‐dose urine sample, post‐dose urine sample, the dose given and the local tap

water were measured by isotope ratio mass spectrometry (IRMS, 20:20 Hydra

Model, PDZ Europa, Crewe, UK) in the Queensland University of Technology

laboratories.

Measurement of deuterium dilution space

A 0.5 mL urine sample was placed into a 10 mL labelled vacutainer tube. A

chromacol vial containing a small amount of platinum on alumina powder was

placed upright in the sample. Subsequently, the samples were evacuated for 5 min

and then 99% hydrogen gas was introduced into the vacutainer. The samples were

then left at room temperature for three days to allow the deuterium in the sample

to equilibrate with the hydrogen gas above the sample, with the platinum on

alumina powder acting as a catalyst. Reference waters of known abundance were

prepared at the same time and in the same way as the urine samples and included

in the analysis to allow for correction of results for electronic drift in the mass

spectrometer. All analyses were completed in triplicate to obtain a reliable result

for each sample. Results were expressed in delta units (%) relative to an

international standard (standard mean ocean water). The deuterium dilution space

was determined from the equation as follows:

TBW (kg) = 041.1

1)(x

EpEs

EtEax

a

TA

Where T is the amount of tap water in which the dose was diluted in grams, A is the

amount of dose taken by the participant in grams, a is the amount of the dose in

grams retained for mass spectrometer analysis, and Ea, Et, Ep and Es are the

isotopic enrichment in delta units relative to standard mean ocean water of the

dilute dose, the tap water used, the pre‐dose urine sample and the post‐dose urine

sample. The constant (1.041) was used to adjust for the non‐aqueous exchange of

hydrogen atoms in the body (Racette et al., 1994).


112

Derivation of FFM and FM mass from TBW

FFM was derived from TBW using a hydration coefficient, that is, the fraction of

FFM comprised of water. Lohman’s age‐ and gender‐specific constants for hydration

of the FFM for children were used to calculate FFM (Lohman, 1986) (see Table 3.3).

The absolute FM was derived by subtracting FFM from weight, based on the

two‐compartment body composition model and %BF was then calculated as follows:

FFM (kg) = TBW/Hydration constant

FM (kg) = Body weight ‐ FFM

%BF = 100 × FM/Body weight

Table 3.3. Age‐ and gender‐specific hydration constants of the FFM (%) in children

Age (yr) Male Female

7‐8 76.8 77.6

9‐10 76.2 77.0

3.1.2.4 Statistical analysis

Single regression model in which %BF was regressed on BMI was used to identify

outliers. Participants with standardized residual values above 3.3 or less than ‐3.3

were excluded from subsequent analyses (Tabachnick & Fidell, 2007).

Body composition results were expressed as mean ± standard deviation. Differences

in body composition among age, sex and ethnic groups were tested by analysis of

(co)variance (AN(C)OVA). Pearson’s correlation was used to assess the correlation

coefficient between BMI and %BF. To test the significance of differences between

correlation coefficients, all coefficients were transformed by using Fisher’s z’s

transformation and t‐test was used to test for the equality between the transformed

coefficients.

The relationship between %BF and BMI throughout the entire biological range is


113

curvilinear (Garrow & Webster, 1985). Therefore, natural logarithmically

transforming BMI (LnBMI) was performed to linearize the curvilinear relationship

between %BF and BMI (Kagawa et al., 2006; Mehta et al., 2002; Rush, Scraqq et al.,

2009) (see Figure 3.1). Stepwise multiple linear regression analysis was conducted

using %BF as the dependent variable, with BMI, age, gender (males=1 and

females=0), ethnicity as independent variables to determine the BMI‐%BF

relationship. The dummy variables for ethnicity were E1, E2, E3, and E4. For Chinese

E1 = 1, E2 = 0, E3 = 0, and E4 = 0; for Lebanese E1 = 0, E2 = 1, E3 = 0, and E4 = 0; for

Malays E1 = 0, E2 = 0, E3 = 1, and E4 = 0; for Thais E1 = 0, E2 = 0, E3 = 0, and E4 = 1; for

Filipinos E1 = 0, E2 = 0, E3 = 0, and E4 = 0. Homogeneity of regression slopes among

the groups was examined by the significance of the interaction between the

covariate and the group variables which was tested by general linear model.

ANCOVA was used to compare differences in %BF between different ethnic groups,

taking difference in gender, age and BMI into account.


114

Figure 3.1. Scatter plots of %BF against BMI and %BF against LnBMI

Validity and accuracy of BMI indicators in the diagnosis of obesity were evaluated

by calculating sensitivity, specificity and agreement rate, relative to true obesity

diagnosed by absolute %BF (%BF above 25% for boys and above 30% for girls

(Williams et al., 1992), using cross‐tabulation. Sensitivity is the proportion of truly


115

obese subjects identified correctly as obese by BMI (true positives). Specificity is the

proportion of truly non‐obese subjects identified correctly as non‐obese by BMI

(true negatives). The agreement rate is the proportion of all subjects that are being

diagnosed correctly as obese and non‐obese by BMI.

The Caucasian prediction equation derived from Dutch children (Deurenberg et al.,

1991) was used to calculate the %BF if their BMI was calculated using the cut‐offs for

obesity as proposed by WHO.

%BF = 1.51×BMI ‐ 0.7×Age ‐ 3.6×Sex + 1.4

The predicted %BF level was then used to recalculate the BMI using the

ethnic‐specific prediction equation to obtain a BMI level for Asian children that is

equivalent with the WHO cut‐offs for obesity developed from the Caucasian

population.

All statistical analyses were performed with SAS 8.02, and a two‐sided P value of

<0.05 was regarded as statistically significant.

3.1.3 Results

The characteristics of 1039 pre‐pubertal children by sex and ethnicity are given in

Table 3.4. All participants were in BMI range from 12.2 to 34.9 kg/m2 and in %BF

range from 5.5% to 54.5%. There was a significant difference in age among ethnic

groups for boys but no significant difference for girls. Significant differences in

height, weight, BMI, and %BF of boys was found among ethnic groups and

significant difference in height and %BF of girls was found among ethnic groups after

adjustment for age. The predicted %BF from BMI using a Caucasian equation was

significantly lower than measured %BF in each sex‐ethnicity group, meaning that

Asian children had a higher %BF compared with Caucasians at a given age and BMI.


116

Table 3.4. Characteristics of the participants (mean±SD)

Chinese Lebanese Malays Filipinos Thais P value

Boys

n 203 74 105 51 101

Age (yr) 9.4±0.8 9.4±0.7 9.4±0.8 9.0±0.8 c 9.4±0.7 0.007

Height (cm) 138.5±7.2 c 134.8±8.0 131.4±7.0 131.3±6.2 134.7±6.6 <0.0001

Weight (kg) 39.3±10.9 d 34.7±9.7 33.1±12.0 32.6±7.5 33.8±10.1 <0.0001

BMI (kg/m2) 20.2±4.4 c 18.8±3.7 18.8±5.3 18.8±3.3 18.4±4.2 0.003

%BF 29.2±10.0 26.3±9.6 26.9±11.3 24.9±7.3 26.5±8.9 0.026

%BF BMIa 22.0±6.3 19.9 ±5.4 19.9±7.9 20.1±5.0 19.3±6.3 0.0043

Girls

n 150 81 92 61 122

Age (yr) 9.4±0.8 9.2±0.7 9.4±0.8 9.3±0.7 9.3±0.8 0.132

Height (cm) 136.2±7.6 133.4±8.0 133.0±9.5 132.8±7.5 134.5±8.0 0.0004

Weight (kg) 34.2±9.1 31.0±7.4 33.6±9.5 33.3±7.6 34.3±10.4 0.152

BMI (kg/m2) 18.2±3.5 17.2±2.7 18.6±4.7 18.7±3.1 18.6±4.1 0.061

%BF 28.5±7.9 26.9±7.6 30.0±10.4 30.2±6.5 31.8±8.6 <0.0001

%BF BMIa 22.6±5.3 21.2 ±3.9 23.1±7.1 23.4±4.8 23.3±6.0 0.0763

All comparisons among ethnic groups adjusted for age except for the comparison of age. %BFBMI: predicted %BF from BMI using a Caucasian equation. a significant difference between measured %BF and %BFBMI in each sex‐ethnicity group with P<0.001.

The correlations of BMI with body composition variables by sex and ethnicity are

shown in Table 3.5. BMI was significantly and positively correlated with all of the

variables in each sex‐ethnicity group. The correlation tended to be stronger among

boys than girls, however, did not reach significance. There was also no significant

difference in the strength of correlation across ethnic groups.


117

Table 3.5. Pearson correlation coefficients between BMI and %BF by sex and ethnicity

Chinese Lebanese Malays Filipinos Thais

Boys

Height 0.45 0.49 0.56 0.34 0.56

Weight 0.94 0.94 0.97 0.92 0.96

FM 0.94 0.96 0.94 0.88 0.93

FFM 0.73 0.73 0.84 0.78 0.86

%BF 0.86 0.89 0.81 0.67 0.84

Girls

Height 0.45 0.46 0.52 0.29 0.55

Weight 0.93 0.89 0.94 0.89 0.95

FM 0.92 0.91 0.95 0.86 0.94

FFM 0.79 0.67 0.77 0.71 0.78

%BF 0.77 0.80 0.77 0.63 0.80

All correlations significant (P<0.0001) except for the correlation of BMI with height in Malay boys and in Filipino girls.

3.1.3.1 BMI‐age relationship

To characterize whether the BMI‐%BF relationship was independent of age, multiple

regression analysis was performed with %BF as the dependent variable and LnBMI,

age and their interaction as independent variables for each sex‐ethnicity group

(Table 3.6). There was no significant difference between the slopes and intercepts of

the regressions for each age group in each sex‐ethnic groups, indicating no

significant difference in BMI‐%BF relationship across age groups.


118

Table 3.6. The coefficients for the multiple regressions between %BF as dependent variables and LnBMI and age as independent variables by sex and ethnic groups

Chinese Lebanese Malays Filipinos Thais

Coefficient±se P value Coefficient±se P value Coefficient±se P value Coefficient±se P value Coefficient±se P value

Boys

Intercept ‐90.36±6.02 <0.0001 ‐113.26±8.46 <0.0001 ‐82.78±9.10 <0.0001 ‐61.83±16.59 0.0005 ‐76.55±8.51 <0.0001

LnBMI 43.48±1.61 <0.0001 47.61±2.80 <0.0001 35.95±2.61 <0.0001 30.16±4.73 <0.0001 35.55±2.30 <0.0001

Age ‐1.13±0.42 0.097 0.08±0.72 0.9166 0.60±0.79 0.4464 ‐0.15±1.07 0.890 0.05±0.63 0.9419

R2 0.785 0.824 0.669 0.459 0.710

SEE (%) 4.8 4.1 6.6 5.5 4.9

Girls

Intercept ‐71.28±7.67 <0.0001 ‐84.00±10.06 <0.0001 ‐56.06±10.54 <0.0001 ‐40.99±14.52 0.0065 ‐59.27±7.09 <0.0001

LnBMI 33.73±2.20 <0.0001 42.09±3.65 <0.0001 34.30±2.89 <0.0001 25.07±4.05 <0.0001 34.66±2.22 <0.0001

Age 0.28±0.49 0.5784 ‐0.96±0.73 0.1961 ‐1.45±0.84 0.0846 ‐0.22±0.95 0.8205 ‐1.07±0.57 0.0615

R2 0.617 0.646 0.613 0.398 0.676

SEE (%) 4.9 4.6 6.5 5.1 4.9


119

3.1.3.2 BMI‐sex relationships

The multiple regression model with %BF as the dependent variable and LnBMI, age,

sex and LnBMI*sex as independent variables was performed separately for each

ethnic group to test whether the BMI‐%BF relationship was independent of sex.

Similarity of regression slopes among the groups was examined by the significance

of the interaction term between LnBMI and sex. No significant difference in the

slopes of Ln BMI‐%BF relationship between boys and girls across ethnic groups was

found except for among Chinese children. As can be seen from Figure 3.2, the thin

Chinese boys had a significantly lower %BF than thin Chinese girls at the same BMI,

the difference in %BF between Chinese boys and girls decreased with an increase of

BMI, and there was no significant difference in %BF between severely obese Chinese

boys and girls. However, the interaction between LnBMI and sex increased the

explained variance by only 0.004, suggesting that the association between BMI and

%BF is generally consistent between Chinese boys and girls. Therefore, the

interaction between LnBMI and sex was removed from the final regression model. In

Table 3.7, the coefficients for the stepwise multiple regressions between %BF as

dependent variables and BMI, age, and sex as independent variables are given for

the Chinese, Lebanese, Malay, Filipino and Thai groups. The intercept of LnBMI‐%BF

relationship for boys and girls within each ethnic group was significantly different (P

<0.0001). The significant negative regression coefficients indicated that boys had a

3.16‐5.35% lower %BF than girls at a given BMI. The R2 value for %BF model with

LnBMI and sex as independent variables was 0.719 (standard error of the estimate,

SEE = 5.0%) for Chinese children, 0.744 (SEE = 4.4%) for Lebanese children, 0.645

(SEE = 6.6%) for Malay children, 0.502 (SEE = 5.2%) for Filipino children, and 0.713

(SEE = 4.9%) for Thai children.


120

.

Figure 3.2. Relationship between %BF and

LnBMI of boys and girls for Chinese, Lebanese,

Malay, Filipino and Thai population. Boys;

Girls; Boys; Girls.


121

Table 3.7. The coefficients for the stepwise multiple regression between %BF as dependent variables and LnBMI, age, and sex as independent variables by ethnic

groups

Coefficient Chinese Lebanese Malays Filipinos Thais

Intercept ‐86.57±3.87 ‐99.94±6.04 ‐71.83±5.56 ‐49.72±8.92 ‐68.26±4.58

Ln BMI 39.89±1.33 44.60±2.13 35.22±1.91 27.40±3.05 34.48±1.57

Sex ‐3.16±0.55 ‐4.27±0.72 ‐3.40±0.94 ‐5.35±0.99 ‐4.81±0.66

R2 0.719 0.744 0.645 0.502 0.713

SEE (%) 5.0 4.4 6.6 5.2 4.9

3.1.3.3 BMI‐ethnicity relationship

There was a significant ethnic difference in the BMI‐%BF relationship among Asian

children in the current study by stepwise multiple regression analysis.

In boys, the model using Filipinos as a reference population was %BF = 39.60×LnBMI

+ 1.31×Malays + 1.42×Thais – 89.25 (R2 = 0.717, SEE = 5.3%). After correcting for age

and BMI, %BF in Malays (27.7±0.5%) and Thai boys (28.1±0.5%) was 2.0% and 2.4%

higher than Filipino boys (25.7±0.8%). Moreover, the regression slopes were

different between Malays and Filipinos (F = 6.52, P = 0.012) and between Thais and

Filipinos (F = 9.82, P = 0.002). The difference of %BF between Malays and Filipinos

and between Thais and Filipinos was more apparent as BMI increased. Despite no

significant difference between Filipinos and Chinese and between Filipinos and

Lebanese in corrected %BF, the regression slopes differed (P<0.05). Chinese and

Lebanese boys tended to have lower %BF than Filipinos at a lower BMI level, while

have higher %BF at a higher BMI level. A similar corrected %BF was found among

Chinese (27.4±0.4%), Lebanese (27.1±0.6%), Malay and Thai boys and no interaction

between LnBMI and ethnicity was found in each pair of Chinese, Lebanese and

Malays and Thais (Figure 3.3).

In girls, the model using Filipinos as a reference population was %BF = 33.70×LnBMI

+ 2.31×Thais – 68.31 (R2 = 0.618, SEE = 5.3%). After correcting for age and BMI, %BF

in Thais (31.1±0.5%) was 1.6% higher than their Filipino counterparts (29.5±0.7%).


122

However, the regression slopes were different between Thais and Filipinos (F = 7.50,

P = 0.007) and the difference in %BF between Thais and Filipinos was more apparent

in higher BMI level. Thai girls had 1.9% higher corrected %BF compared with their

Lebanese counterparts (29.2±0.6%) and the difference decreased with the increase

of BMI (F = 12.30, P = 0.0006 for the regression slopes comparison). Thai girls had

1.6% and 2.5% higher corrected %BF than their Malays (29.5±0.7%) and Chinese

(28.6±0.4%) counterparts, respectively, and there was no significant interaction term

of LnBMI *ethnicity between Thais and Malays (F = 3.39, P = 0.067 for the regression

slopes comparison) and between Thais and Chinese (F = ‐0.141, P = 0.888 for the

regression slopes comparison). A similar corrected %BF was found among Chinese,

Lebanese, Malay and Filipino girls and no interaction between LnBMI and ethnicity

was found in each pair of ethnic groups of Chinese, Lebanese, Malay and Filipino

except for significant difference in the regression slopes between Lebanese and

Filipinos (F = 2.567, P = 0.011). Lebanese girls tended to have lower %BF than

Filipinos at a lower BMI level, while have higher %BF at a higher BMI level (Figure

3.3).


123

Figure 3.3. Relationship between %BF by deuterium dilution and LnBMI of Chinese, Lebanese, Malay, Filipino, and Thai boys and girls.


124

3.1.3.4 Validation of WHO, IOTF and China obesity classification

Table 3.8 shows the sensitivity and specificity of BMI cut‐off points for obesity, as

defined by the WHO, IOTF and China classification in identifying boys with a %BF

>25 and girls with a %BF >30, by ethnic group. The results indicated the sensitivity

and specificity with the three criteria was dependent on ethnicity. Filipino boys

showed the highest sensitivity with WHO and China classification and Malay boys

showed the highest sensitivity with IOTF classification. Lebanese boys showed the

lowest sensitivity with all three classifications. Malay girls showed the highest

sensitivity with all three classifications while Lebanese girls showed the lowest

sensitivity with all three classifications. When the WHO classification was applied to

the five Asian countries, 52.6 to 61.5% of boys who had a %BF of above 25% were

identified as obese and only 11.5 to 57.4% girls who had a %BF of above 30% were

identified. When the IOTF classification was applied, less boys and girls with excess

body fat were identified as obese. In contrast, using the China classification the

sensitivity increased to 29.6 to 59.6% for girls, while decreased by 1.1 to 7.9% for

boys with excess body fat identified by BMI to WHO classification was found,

indicating that the China classification is more appropriate for Asian girls.


125

Table 3.8. The sensitivity, specificity and agreement rate with WHO, IOTF and Chinese BMI classifications for obesity

WHO IOTF China

% Sensitivity Specificity Agreement % Sensitivity Specificity Agreement % Sensitivity Specificity Agreement

Boys

China 38.1 57.8 97.3 72.3 21.9 33.6 98.6 57.4 33.2 51.6 98.6 68.8

Lebanon 27.0 52.6 100.0 75.7 13.5 26.3 100.0 62.2 23.0 44.7 100.0 71.6

Malay 31.4 56.6 94.2 75.2 21.0 39.6 98.1 68.6 29.5 52.8 94.2 73.3

Philippines 31.4 61.5 100.0 80.4 15.7 30.8 100.0 64.7 27.5 53.8 100.0 76.5

Thailand 24.8 54.3 100.0 79.2 14.9 32.6 100.0 69.3 23.8 52.2 100.0 78.2

Girls

China 12.7 25.4 100.0 64.7 8.7 18.3 100.0 61.3 21.3 45.1 100.0 74.0

Lebanon 3.7 11.5 100.0 71.6 2.5 7.7 100.0 70.4 9.9 29.6 98.2 75.6

Malay 30.4 57.4 97.8 77.2 26.1 48.9 97.8 72.8 31.5 59.6 97.8 67.2

Philippines 19.7 36.7 96.8 67.2 14.8 26.7 96.8 62.3 23.0 40.0 93.5 67.2

Thailand 25.4 46.8 96.7 71.3 14.8 27.4 98.3 62.3 31.2 58.1 96.7 77.0


126

3.1.3.5 Equivalent BMI cut‐offs for obesity

The Caucasian prediction equation was used to calculate how high participants’ %BF

would be if their BMI had been according to the cut‐offs for obesity proposed by

WHO. The predicted %BF level was then used to recalculate their BMI using each

ethnic prediction equation to obtain a BMI level that was equivalent to the WHO

cut‐offs for obesity developed from the Caucasian population. Table 3.9 shows that

the BMIs of Asian children who had the same level of %BF as Caucasians were 2.9 to

6.1 kg/m2 lower than Caucasians. For example, Thai boys aged 8 y with a BMI of

15.7 kg/m2 had the same level of %BF as Caucasians with a BMI of 19.7 kg/m2.

Table 3.9. Comparison of BMI cut‐off points for obesity proposed by WHO using Caucasian data with calculated BMI equivalents for Chinese, Lebanese, Malay, Filipino and Thai boys and girls derived from regression equations for predicting

%BF from BMI.

Caucasians Chinese Lebanese Malays Filipinos Thais Age

(year) BMI (kg/m2)*

%BF# Predicted BMI equivalent (kg/m2)

Boys

8 19.7 21.9 16.4 16.8 15.8 16.6 15.7

9 20.5 22.5 16.6 17.0 16.0 16.9 16.0

10 21.4 23.1 16.9 17.2 16.3 17.3 16.3

Girls

8 20.6 26.9 17.2 17.0 16.5 16.4 15.8

9 21.5 27.6 17.5 17.3 16.8 16.8 16.1

10 22.6 28.5 17.9 17.7 17.3 17.4 16.5

* BMI cut‐offs for obesity proposed by WHO for school‐aged children and adolescents (de Onis M, 2007). #Corresponding value of %BF predicted from the Caucasian equation from BMI (P Deurenberg et al., 1991).

3.1.4 Discussion

The present study indicates that the relationship between %BF as determined by the

deuterium dilution method and BMI is dependent of sex and ethnicity but


127

independent of age among the cohort of Asian pre‐pubertal children from different

origins. Our results indicate that at an equivalent level of BMI, Filipino boys had a

lower %BF compared with their Malay and Thai counterparts. Thai girls had a

significantly higher %BF compared with their Chinese, Lebanese, Malay and Filipino

counterparts. Although a limited number of studies have determined the difference

in the BMI‐%BF relationship among Asians from different backgrounds, evidence

suggests that Asians differ in this relationship among both children and adults.

South Asians had a higher %BF than East Asians (Deurenberg‐Yap et al., 2000;

Duncan et al., 2009) and East Asians had a lower %BF than South‐East Asians

(Deurenberg‐Yap et al., 2000; Gurrici et al., 1999) at a given BMI, sex and age. In a

study conducted in Chinese and Malay Indonesian adults, Chinese Indonesians had a

1.7±0.3 kg/m2 higher BMI than the Malay Indonesians after correcting for

differences in age, sex and %BF (Gurrici et al., 1999). In another study,

Deurenberg‐Yap et al. (2000) found that Malay Singaporeans had a 0.5% higher %BF

than Chinese Singaporeans. Although our study didn’t find the significant difference

in the BMI‐%BF relationship between Chinese and Malay children, it is interesting to

compare the ethnic differences in the BMI‐%BF relationship among South‐East Asian

populations. Malay and Thai boys had a significantly higher %BF than Filipino boys

and Thai girls had a significantly higher %BF than Malay and Filipino girls at a fixed

BMI level, despite the fact that both groups live in the similar climatic conditions

and having similar food supply. To date, there is little published information on

difference in the BMI‐%BF relationship among these ethnic groups. Lebanon lies in

the far West of Asia and we hypothesized that the body composition of children in

this cohort would be similar to the European population with a lower %BF than the

Asian population at a given BMI. However, no significant difference in BMI‐%BF

relationship was found among Chinese, Lebanese and Malay children. It should be

mentioned that the precision of the TBW technique (1‐2%) may influence the

estimation of %BF, and accordingly, influence the ethnic difference in BMI‐%BF

relationship. However, the estimated error of TBW is typically <1 kg (Wang et al.,

1999). According to the finding from Wang et al. (1999), the mean estimated error

of %BF in our study was 0.04% (SD=0.01%). Therefore, the difference in %BF


128

between Filipino boys and Thai and Malaysian boys (2.0% and 2.4%, respectively),

and between Thai girls and other ethnicities (1.6%‐2.5%) at a given age and BMI

should still be significant even considering the precision of the body composition

measurement.

However, the ethnic difference in the BMI‐%BF relationship among the five Asian

countries studied varied with BMI. In boys, the higher %BF in Malays and Thais than

Filipinos was most evident at a higher BMI level. Chinese and Lebanese tended to

have lower %BF than Filipinos at a lower BMI level, while have higher %BF at a

higher BMI level. In girls, the higher %BF in Thais than Filipinos was most evident at

a higher BMI level while than Lebanese was most evident at a lower BMI level.

Lebanese girls tended to have lower %BF than Filipinos at a lower BMI level, while

have higher %BF at a higher BMI level. Some studies conducted among white,

Hispanic and Asian populations have also reported that the relationship between

BMI and %BF varies by BMI category (Fernández et al., 2003; Freedman et al., 2008;

Swinburn et al., 1999; Wang & Bachrach, 1996). Relatively thin Asian children tend

to have a higher %BF, while overweight Asian children tend to have lower %BF

compared with their white counterparts at a given BMI and age (Freedman et al.,

2008). Hispanic youth have less body fat than whites at a BMI <20 kg/m2, but more

body fat at a BMI >20 kg/m2 (Wang & Bachrach, 1996). The interaction between

BMI and ethnicity in the estimation of body fatness indicates that it may be difficult

in identifying equivalent levels of body fatness by simply adjusting BMI for the

average difference in body fatness across ethnic groups (Freedman & Sherry, 2009).

Our results also indicate that Asian children had higher %BF than Caucasian children

at an equivalent level of BMI, age and sex. When a Caucasian prediction formula

developed in Dutch children aged 7‐15 y (Deurenberg et al., 1991) was used to

predict the %BF from BMI, age and sex, the predicted %BF was lower than the

measured %BF, 6.6% for Chinese, 6.0% for Lebanese, 6.9% for Malays, 5.9% for

Filipinos, and 7.9% for Thais. Caucasian equations generally show a remarkable

under‐estimation when used in the Asian population. For example, Gurrici et al.

(1999) indicated that the prediction formula developed in the Dutch population


129

underestimated %BF by 5.8±4.8% and 7.7±3.8% in the male and female Malay

Indonesians, and 1.3±3.0% and 1.7±3.7% in the male and female Chinese

Indonesians. These results are in agreement with the studies in which the ethnic

difference in measured %BF and BMI relationship was determined between

Caucasians and Asians. Caucasians had a lower %BF (as determined by DXA, 4‐C

model, deuterium dilution technique or skin fold measurement) than Asians at a

given BMI level among children (Deurenberg, Deurenberg‐Yap et al., 2003; Duncan

et al., 2009; Mehta et al., 2002; Navder et al., 2009) and adults (Chang et al., 2003;

Chung et al., 2005; Deurenberg et al., 1998, 2002; Gurrici et al., 1998; Kagawa et al.,

2006; Wang et al., 1994).

As discussed in the literature review, several factors might be responsible for a

dependency of the relationship between %BF and BMI on ethnicity, including

differences in relative leg length or relative sitting height, frame size, and physical

activity level. The groups with a higher relative leg length and bigger frame have a

lower %BF at the same BMI. Caucasians have a higher relative leg length and bigger

frame than Asians (Deurenberg et al., 1999; Gurrici et al., 1999; Norgan, 1994a).

Studies on the difference in body build among Asian populations are limited. In a

study conducted in Malay and Chinese Indonesians, Malay Indonesians had a higher

slenderness index (height/sum of wrists and knee widths) compared to the Chinese

Indonesians (Gurrici et al., 1999). The groups with a higher activity level might have

a higher proportion of muscle mass in the body, meaning less body fat at the same

body weight (Gurrici et al., 1999; Luke et al., 1997). However, no physical activity

information was available for the current study. Moreover, those with the same

ethnic background living in different places differ in BMI‐%BF relationship,

indicating that environmental and socioeconomic factors also contribute to the

BMI‐%BF relationship such as family income. For example, both Chinese

Singaporeans and Chinese Americans had a higher %BF than Chinese living in

mainland China at a given BMI (Deurenberg, Deurenberg‐Yap et al., 2003; Navder et

al., 2009). Blacks living in America also showed a higher %BF than those living in

Africa (Luke et al., 1997).


130

In the present study, a Caucasian equation was employed to predict the

corresponding %BF from BMI thresholds for obesity proposed by WHO then the

equivalent BMIs recalculated for our subjects using the ethnic‐specific equation. Our

results suggest that the BMI thresholds proposed by WHO need to be lowered by

3‐6 units for Asian children. Previous studies also employed Caucasian equations to

derive the corresponding BMI for obesity for other ethnic groups among children

and adults (Chang et al., 2003; Deurenberg‐Yap et al., 2000; Rush et al., 2003;

Swinburn et al., 1999). Similar to our study, BMI cut‐offs for obesity for Taiwanese,

Indians, Malays and Chinese adults should be lowered compared with that for

Caucasians.

Low sensitivity with WHO and IOTF BMI classifications was found in our study. More

than one third of boys with excess body fat failed to be identified, and more than

two thirds of Chinese, Lebanese and Filipino girls and about half of the Malay and

Thai girls with excess body fat failed to be identified when the WHO BMI

classification was applied. The WHO proposed BMI cut‐off points ranging from 18.3

to 29.7 kg/m2 for children and adolescents aged 5‐19 y in 2007, which correspond to

the adult obesity threshold of 30 kg/m2. These cut‐offs were developed from data

from the United States population. The IOTF has also developed international BMI

cut‐off points for obesity by sex for children and adolescents aged 2‐18 y, defined to

pass through the BMI of 30 kg/m2 at the age of 18 y. Although the IOTF BMI

classification was derived using data from Brazil, Great Britain, Hong Kong, the

Netherlands, Singapore, and the United States, the sensitivity is even lower than the

WHO classification. There are no country‐specific BMI cut‐off points for overweight

and obesity in Lebanon, Malaysia, and The Philippines and the WHO criteria are

employed in these countries. Weight for height is used to classify obesity in Thailand.

China has its own age‐ and gender‐specific BMI cut‐offs. The BMI cut‐off values are

slightly higher than the WHO cut‐offs in boys while lower than the WHO cut‐offs in

girls. These lower BMI cut‐offs increase the sensitivity by 19.7%, 18.1%, 2.2%, 3.3%,

and 11.3% for Chinese, Lebanese, Malay, Filipino, and Thai girls, respectively. Results

from our study, combined with the predicted equivalent BMI cut‐off points for the


131

five countries (Table 3.9), suggest that, as lower BMI cut‐offs were proposed for

Asian adults, BMI cut‐offs for Asian children should be lowered when the BMI is

used as a screening tool for overweight and obesity.

Results of the current study confirm the significantly higher %BF in pre‐pubertal girls

compared with boys across all five Asian countries at a given BMI, which is in

agreement with the previous studies in this maturational category (Daniels et al.,

1997; Navder et al., 2009) and other age groups (Deurenberg et al., 1991, 1998;

Gallagher et al., 1996; Gurrici et al., 1998). For both children and adults, males had a

lower %BF than females at an equivalent level of BMI. This difference was

substantial and persists across the lifespan. Moreover, age was independent of the

BMI‐%BF relationship in the current study, which is not in agreement with other

studies. Previous studies have shown that older children have a lower %BF than

younger children at an equivalent level of BMI (Deurenberg et al., 1991; Navder et

al., 2009). However, Daniels et al. (1997) indicated that the stage of sexual

maturation was a more important determinant of %BF than age. This may explain

the results in our study, in which all the participants were within the narrow age

range of 8‐10 y, and all were pre‐pubertal.

There are some limitations in our study. Firstly, despite the large sample (n = 1039),

it was not randomly selected and may not be representative of the general

population. However, we gained a better understanding of the BMI‐%BF relationship

because, participants were representative of a wide BMI range which provides us

with a better understanding of the BMI‐%BF relationship across the BMI range.

Secondly, although the validity of the deuterium dilution technique for the

assessment of TBW and subsequent prediction of %BF using the 2‐compartment

model, a bias may have occurred due to the violation of the assumptions of the

technique. The primary assumption is FFM has a constant hydration fraction. There

may be a difference in the hydration of the FFM among different ethnic groups due

to a combination of genetic and environmental (for example, climatic conditions)

factors. However to date, there is no published information on the ethnic‐specific

hydration of the FFM. Moreover, the deuterium dilution technique is the only option


132

for the remote measurement of body composition with high precision. Thirdly, we

know very little about the ethnic variation in health risk across %BF ranges and there

is no widely accepted classification of excess body fat among children. It is clear that

there is an urgent need to describe the dose‐response relationship between %BF

and health risk in children and adolescents in different ethnic groups. However, the

delayed onset of numerous health conditions related to obesity during childhood

and adolescence makes this work very difficult. In addition, habitual physical activity

and dietary intake was not collected in the current study, which may have the

potential to influence body fat distribution.

In conclusion, the current study demonstrated that the relationship between BMI

and %BF differs by ethnicity among Asian children. However, these ethnic

differences vary according to BMI suggesting that the application of average ethnic

differences in body fatness across the BMI range may not be appropriate. Asian

children had a greater %BF than Caucasian children at any given BMI, indicating that

Asian children may be at higher risk of developing obesity‐related health conditions

at lower BMI values than the Caucasian population. These ethnic differences in the

BMI‐%BF relationship should be considered when BMI is used as a screening tool for

obesity in epidemiologic studies and in clinical settings.


133

3.2 ETHNIC DIFFERENCES IN BODY FAT DISTRIBUTION AMONG ASIAN PRE‐PUBERTAL CHILDREN

Modified from: Liu A, Byrne NM, Kagawa M, Ma G, Kijboonchoo K, Nasreddine L, Poh BK, Ismail MN, Hills AP (2011). Ethnic differences in body fat distribution among Asian pre‐pubertal children: a cross‐sectional multicenter study. BMC Public Health, 11:500.

3.2.1 Introduction

Given the high and rapidly increasing prevalence rates of pediatric obesity and

obesity‐related health risks such as hypertension, metabolic syndrome, and type 2

diabetes in both developed and developing countries (Booth, Dobbins, Okely,

Denney‐Wilson, & Hardy, 2007; Li, Schouten et al., 2008; Li, Yang et al., 2008;

Lobstein et al., 2004), body fat distribution, in addition to total body fat, is an

important issue. BMI and %BF do not fully explain the ethnic differences in

cardiovascular morbidities and diabetes (McAuley et al., 2002), although the ethnic

differences in cardiovascular morbidities and diabetes have been well documented

(Cossrow & Falkner, 2004; Ehtisham, Hattersley, Dunger, & Barrett, 2004; Ford, 2005;

Whincup et al., 2002). An android fat pattern and a central fat deposition is a

stronger predictor of cardiovascular disease, type 2 diabetes and metabolic risk

factors than overall adiposity in both adults and children (Berman et al., 2001;

Daniels et al., 1997; Ehtisham et al., 2005; Freedman et al., 1989; He et al., 2002;

Okosun, 2000).

Some previous studies investigated the ethnic difference in fat distribution in terms

of trunk and extremity model (Ehtisham et al., 2005; Harsha et al., 1980; He et al.,

2002; Malina et al., 1995; Nindl et al., 1998; Okosun et al., 2000; Potts & Simmons,

1994; Rush, Freitas et al., 2009; Thomas et al., 1997), upper and lower body fat

model (Malina et al., 1995; Morrison, Barton et al., 2001; Novotny et al., 2007;

Zillikens & Conway, 1990), and subcutaneous and visceral abdominal adipose tissue

model (Goran et al., 1997; Herd et al., 2001; Rush, Freitas et al., 2009; Tanaka et al.,


134

2003), among Caucasians, blacks and Asian adults and children. Asians and blacks

appear to have more central fat, upper body fat, SAAT and VAAT than Caucasians.

Some studies have also found that Asians and blacks have higher upper body fat

than whites. However the ethnic difference in fat distribution is less well understood,

especially in children and adolescents due to the effect of sexual maturation on

body composition (He et al., 2004). Moreover, some studies also report different fat

patterns among Asian adults from different origins (Lear et al., 2007). However, to

our knowledge, few data are available on the difference in fat distribution across

Asian children with different origins. A lack of understanding of ethnic differences in

body fat distribution might cause misuse or misinterpretation of results obtained

from anthropometric indices. Data on different body fat distribution of individuals

from different ethnicities are necessary to better assess the risk of cardiovascular

risk factors clustering in children. Accordingly, the purpose of the current study was

to investigate the influence of age, sex and ethnicity on the body fat patterns in

Asian pre‐pubertal children from four distinctly different origins.

3.2.2 Methodology


Four countries were involved in the current study, including one East Asian country

(China), one West Asian country (Lebanon), and two South‐East Asian countries

(Malaysia and Thailand). In each country, a non‐random purposive sampling

approach was used which aimed to enrol children encompassing a wide BMI range

for each year of age between 8‐10 y and each sex. A total of 922 participants were

recruited from the four countries. Table 3.10 shows the sample size of each country

for each anthropometric and body composition variables. Ethnicity was determined

by self‐identification and those whose parents were identified as having the same

origin were included. Additional inclusion criteria required that participants be

healthy, at Tanner stage 1 of puberty, and free from any diagnosed medical

condition that might potentially interfere with body composition measurement. The

study protocol was explained to the parent(s) and the children and written consent


135

obtained from each child and/or their parent(s).

Table 3.10. Numbers of participants for each variable by sex

Boys Girls Variables

Chinese Lebanese Malays Thais Chinese Lebanese Malays Thais

Height 198 74 105 101 150 80 92 122

Weight 198 74 105 101 150 80 92 122

WC 198 74 105 101 150 80 92 122

%BF (D2O) 198 74 105 101 150 80 92 122

Skin fold

Biceps 127 74 105 83 93 80 92 104

Triceps 127 74 105 83 93 80 92 104

Subscapular 127 74 105 ‐ 93 80 92 ‐

Suprailiac 127 74 104 83 93 80 92 105

Abdominal 127 ‐ 104 ‐ 93 ‐ 92 ‐

Front thigh ‐ 71 105 ‐ ‐ 72 92 ‐

Medial calf 71 105 83 ‐ 72 92 105


Height was measured using a portable Holtain Stadiometer to the nearest 0.1 cm.

Body weight was measured using a SECATM electronic scale (Hamburg, Germany) to

the nearest 0.1 kg with participants wearing only underwear after urinating in the

morning. The detailed measurement procedure was described in section 3.1.2.2.

BMI (kg/m2) was then calculated as weight (kg) divided by the square of height (m).

WC was measured using a tape to the nearest 0.1 cm midpoint between the lower

costal border and the top of the iliac crest, and the measurement was taken at the

end of normal expiration. The technique for measuring WC was as follows:

1) Hold the tape at right angles to the long axis of the body segment being

measured;

2) Use the cross‐hand technique;


136

3) Use constant tension throughout the measurement to make sure there is no

indentation on the skin but the tape holds the place at the designated landmark;

4) Take the reading at eye level to avoid a parallax error.

3.2.2.3 Skin fold measurements

Skin fold thickness (SFT) was measured using a Holtain T/W skinfold caliper (Holtain

Ltd., Crosswell, Crymych, Pembs, SA41 3UF, UK) to the nearest of 0.1 mm according

to the International Society for the Advancement of Kinanthropometry (ISAK)

protocol (Marfell‐Jones, Olds, Stewart, & Carter, 2006). Measurement sites included

seven skin folds: biceps, triceps, subscapular, suprailiac, abdominal, front thigh and

medial calf.

Biceps skin fold site is the point on the most anterior surface of the arm in the

mid‐line at the level of the mid‐acromiale‐radiale landmark. A vertical fold is raised

with the left thumb and index finger at the marked site and the caliper is applied 1

cm distantly. The participant’s arm should be relaxed with both arms hanging by

their side.

Triceps skin fold site is the point on the most posterior surface of the arm over the

triceps muscle, at the level of the mid‐acromiale‐radiale landmark when viewed

from the side. A vertical fold is raised with the left thumb and index finger at the

marked site and the caliper is applied 1 cm distally. The participant’s arm should be

relaxed with both arms hanging by their side.

Subscapular skin fold site is at 2 cm along a line running laterally and obliquely

downward from the subscapular landmark at a 45⁰ angle. The fold is raised with the

left thumb and index finger at the marked site and the caliper is applied 1 cm away.

The participants should stand erect with their arm by their side.

Suprailiac skin fold site is the point where the line from the marked iliospinale to

the anterior axillary border intersects with the horizontal line at the superior border

of the ilium. The caliper is applied 1 cm anteriorly from the left thumb and index


137

finger, raising a fold at the point. The fold runs medially downward at about a 45⁰

angle.

Abdominal skin fold site is the point 5 cm horizontally to the right hand side of the

midpoint of the navel. The distance of 5 cm assumes an adult height of about 170

cm. Hence the distance should be scaled for height by calculated 5 cm*height of

participant/170 if the height differs markedly from 170 cm. The vertical fold is raised

with the left thumb and index fingers and the caliper is applied 1 cm inferiorly to the

point.

Front thigh skin fold site is the point parallel to the long axis of the femur at the

mid‐point of the distance between the inguinal fold and the superior border of the

patella. The participant’s knee is bent at right angles by placing the right foot on a

box or by being seated. A vertical fold is raised with the left thumb and index finger

at the marked site and the caliper is applied 1 cm distantly with the knee straight

and resting on a box.

Medial calf skin fold site is the point on the most medial aspect of the calf at the

level where it has maximal circumference. A vertical fold is raised with the left

thumb and index finger at the marked site and the caliper is applied 1 cm distally

with the participant standing with his right foot on a box with the knee at 90⁰ and

the calf relaxed.

The technique for skin fold measurement was:

1) Always measure the right side of the body regardless of the dominant side;

2) Raise the skin fold at the marked site and apply the caliper 1 cm away from the

controlling thumb and index finger;

3) Face the back of the left hand to the measurer all the time;

4) Place the caliper at a depth of about the level of the mid‐fingernail;

5) Hold the fold at 90⁰ to the surface of the skin fold throughout the measurement

and take the reading at two seconds after the full pressure of the caliper is

applied;

6) Measure the skin fold sites in the same order as on the data sheet and obtain a


138

complete set first before repeating a second or third set of measurement;

7) Take the reading at eye level to avoid a parallax error.

All measurements were taken twice and the mean was calculated. If the difference

between the two measurements was more than 0.5 mm, then a third measurement

was taken and the median was regarded as the skin fold thickness.

Trunk SFT (sum of subscapular and suprailiac) and upper extremity (sum of biceps

and triceps) was used as the index for the absolute amount of trunk fat and upper

extremity fat, respectively. WC was used as the index for the absolute amount of

abdominal fat. Trunk /upper extremity ratio and suprailiac/upper extremity ratio

were used as the indices for the relative distribution of fat mass in the trunk and

extremities. Higher trunk SFT, WC, Trunk /upper extremity ratio or suprailiac/upper

extremity ratio indicates more trunk fat or central depots. Suprailiac/subscapular

ratio was used to describe the upper to lower trunk fat distribution pattern.

Body fat assessment

The isotope deuterium dilution technique was used to measure FM. The detailed

procedure was described in section 3.1.2.3.


All variables were transformed to achieve the normal distribution before analysis

wherever necessary using natural logarithms. One‐way analysis of variance (ANOVA)

was used to test differences in base characteristics and anthropometric parameters

between sex and among ethnic groups. General linear model of ANCOVA was

employed to test the age, sex, and ethnic differences in body fat variables with

Bonferroni multiple comparisons. The comparison of WC among ethnic groups was

tested by ANCOVA adjusted for height and weight. The comparison of SFT and

trunk/extremity ratios among different groups was tested by ANCOVA adjusted for

total body fat. The comparisons of upper/lower trunk fat ratio among different

groups were tested by ANCOVA adjusted for trunk fat. Stepwise multiple regression

analysis was used to determine the ethnic difference in the relationship between


139

trunk body fat and selected anthropometric variables. A separate model was used

for each dependent variable (WC, trunk SFT, trunk/extremity ratios and

subscapular/suprailiac skin fold ratio). For the WC model, BMI, age, sex, ethnicity

and their interaction terms as the independent variables. For trunk SFT and

trunk/extremity ratios models, FM, age, sex, ethnicity and their interaction terms as

the independent variables. For the subscapular/suprailiac ratio model, trunk SFT,

age, sex, ethnicity and their interaction terms were used as the independent

variables. The dummy variables for ethnicity were E1, E2, E3. For Chinese E1 = 1, E2 = 0

and E3 = 0; for Lebanese E1 = 0, E2 = 1 and E3 = 0; for Thai E1 = 0, E2 = 0 and E3 = 1.

Malays were regarded as a reference because all anthropometric and body fat

variables were measured in the Malay population. The dummy variable for sex was

female = 0 and male = 1. Homogeneity of regression slopes among the groups was

examined by the significance of the interaction between the covariate and the group

variables. These analyses were repeated for boys and girls separately. All statistical

analyses were performed using SPSS version 13.0 (SPSS, Inc., Chicago, IL, USA). A

two‐tailed P<0.05 was considered significant.

3.2.3 Results

Descriptive characteristics of the participants are presented in Table 3.11 by

sex‐ethnic subgroups. All participants were in the BMI range 12.2 to 34.9 kg/m2.

There was no difference in age among Chinese, Lebanese, Malays and Thais in each

sex. In boys, Chinese were the tallest while Malays were the shortest, and there was

no significant difference in height between Lebanese and Thais. Chinese were also

heaviest and there were no significant differences in weight among Lebanese,

Malays and Thais. There were no significant differences in BMI and FM among the

four groups except for between Chinese and Thais. In girls, there was no significant

difference in height among the four groups except for Chinese with greater height

than Malays. No significant differences in weight and BMI were found among the

four ethnic groups. There was no significant difference in FM between each ethnic

group except for between Lebanese and Thais.


140

Table 3.11. Descriptive characteristics of the participants (mean±SD)

Chinese Lebanese Malay Thai P value*

Boys

Age (yr) 9.0±0.8 9.0±0.7 9.0±0.8 8.8±0.8 0.451

Height (cm) 138.5±7.3 134.8±8.0 a 131.4±7.0 134.7±6.6 a <0.0001

Weight (kg) 39.0±10.5 34.7±9.7 a 33.1±12.0 a 33.8±10.1 a <0.0001

BMI (kg/m2) 20.0±4.1 a 18.8±3.7 a,b 18.8±5.3 a,b 18.4±4.2 b 0.005

FM (kg) 12.1±6.7 a 9.9±6.3 a,b 10.0±7.8 a,b 9.7±6.1 b 0.0049

Girls

Age (yr) 9.0±0.8 8.8±0.7 9.0±0.8 8.9±0.8 0.386

Height (cm) 136.2±7.6a 133.4±7.9 a,b 133.0±9.5b 134.5±8.2 a,b 0.012

Weight (kg) 34.2±9.1 31.0±7.4 33.6±12.0 34.3±10.4 0.083

BMI (kg/m2) 18.2±3.5 17.2±2.7 18.6±4.7 18.6±4.1 0.056

FM (kg) 10.3±5.2 a,b 8.7±4.3a 10.9±7.2 a,b 11.5±6.2b 0.0065

*One‐way ANOVA. Sharing the same letter means no significant difference between groups by Bonferroni multiple comparisons in each sex.

3.2.3.1 Ethnic differences in body fat distribution

Biceps skin fold

There was a significant ethnic difference in biceps SFT among Chinese, Lebanese,

Malays, and Thais in each sex (P<0.0001) after adjustment for age and FM (Table

3.12). In boys, Lebanese had similar biceps SFT to Malays and both had a higher

biceps SFT than Chinese, who in turn were higher than Thais. In girls, biceps SFT was

thicker in Malays than in Lebanese and Chinese (between whom no significant

difference was found) and was thinnest in Thais (Figure 3.4).


141

Table 3.12. Comparison of subcutaneous adiposity and fat distribution variables among four ethnic groups by sex (adjusted means±SE)

Chinese Lebanese Malays Thais P value‡

Boys

Skin fold

Biceps (mm)# 7.9±0.2 9.1±0.3§ 9.5±0.2 6.6±0.3 <0.0001

Triceps (mm)# 15.5±0.3 11.9±0.4 13.0±0.3 12.3±0.4* <0.0001

Subscapular (mm)# 12.1±0.4 11.4±0.5 12.7±0.4 ‐ 0.115

Suprailiac (mm)# 15.1±0.4 9.4±0.5 12.1±0.4 12.0±0.5 <0.0001

Abdominal (mm)# 15.8±0.5 ‐ 16.6±0.6 ‐ 0.307

Front thigh (mm)# ‐ 21.3±0.7 18.4±0.5 ‐ 0.001

Medial calf (mm)# ‐ 14.1±0.4 11.8±0.3 10.1±0.4* <0.0001

Trunk (supr+sub, mm) # 28.0±0.7 21.6±0.9 25.6±0.8 ‐ <0.0001

Trunk/U extremity ratio # 1.15±0.02 0.92±0.03 0.99±0.02 ‐ <0.0001

Suprailiac/ upper extremity ratio # 0.62±0.01 0.42±0.02* 0.48±0.0

* 0.60±0.02 <0.0001

Trunk/extremity ratio # ‐ 0.34±0.01 0.42±0.01 ‐ <0.0001

Subscapular/suprailiac ratio† 0.88±0.03 1.17±0.04

§ 1.09±0.04§ ‐ <0.0001

WC (cm) ※ 68.1±0.2§ 65.8±0.4 64.1±0.3§ 67.8±0.3 <0.0001

Girls

Skin fold

Biceps (mm)# 7.7±0.2 8.2±0.2 9.7±0.2 6.7±0.2 <0.0001

Triceps (mm)# 15.2±0.3 11.7±0.3 13.1±0.3 13.2±0.3 <0.0001;

Subscapular (mm)# 11.4±0.3 9.8±0.3 11.3±0.3 ‐ 0.003

Suprailiac (mm)# 14.9±0.4 10.0±0.4 12.6±0.4 13.1±0.4 <0.0001

Abdominal (mm)# 14.9±0.6 ‐ 16.4±0.6 ‐ 0.046

Front thigh (mm)# ‐ 21.0±0.6 19.0±0.5 ‐ 0.012

Medial calf (mm)# ‐ 14.5±0.4 13.5±0.3 11.9±0.3 0.009

Trunk ( mm) # 26.2±0.6 19.7±0.7 23.9±0.6 ‐ <0.0001

Trunk/U extremity ratio # 1.13±0.02 0.94±0.02 0.99±0.02 ‐ <0.0001

Suprailiac/extremity ratio # 0.63±0.01 0.47±0.01 0.52±0.01 0.64±0.01 <0.0001

Trunk/extremity ratio # ‐ 0.34±0.01 0.41±0.01 ‐ <0.0001

Subscapular/suprailiac ratio† 0.84±0.02 1.06±0.03 0.98±0.03 ‐ <0.0001

WC (cm) ※ 64.2±0.2 62.9±0.4 60.6±0.3 65.0±0.3 <0.0001※ adjusted for BMI and age; # adjusted for FM and age;† adjusted for trunk fat; ‡ ANCOVA analysis. § higher adjusted mean value in boys and * for lower adjusted mean value in boys within each ethnic group using ANCOVA with Bonferroni multiple comparisons procedure.


142

0

2

4

6

8

10

12

Boys Girls

Bic

eps

skin

fold

(m

m)

Chinese Lebanese Malays Thais

Figure 3.4. Ethnic differences in age‐ and FM‐corrected biceps skin fold among Chinese, Lebanese, Malays and Thais by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANCOVA with Bonferroni multiple comparisons procedure.

Triceps skin fold

There was also a significant ethnic difference in triceps SFT among the four groups in

each sex (P<0.0001) after adjustment for age and FM (Table 3.12). In boys, Chinese

had the thickest triceps SFT and there were no significant differences among the

other three groups. In girls, Malays had a similar triceps SFT to Thais with both

having a higher mean value than Lebanese and lower mean value than Chinese at a

given FM (Figure 3.5).

a a

a

aa


143

0

2

4

6

8

10

12

14

16

18

Boys Girls

Tri

cep

s sk

info

ld (

mm

)


Figure 3.5. Ethnic differences in age‐ and FM‐corrected triceps skin fold among Chinese, Lebanese, Malays and Thais by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANCOVA with Bonferroni multiple comparisons procedure.

Subscapular skinfold

There is no significant difference in corrected subscapular SFT among Chinese,

Lebanese and Malay boys (P = 0.115), however there was a significant difference in

girls (P = 0.003) after adjustment for age and FM (Table 3.12). Chinese and Malay

girls had similar subscapular SFTs but higher values than Lebanese (Figure 3.6).

a a a

aa


144

0

2

4

6

8

10

12

14

Boys Girls

Su

bsc

apu

lar

skin

fold

(m

m)

Chinese Lebanese Malays

Figure 3.6. Ethnic differences in age‐ and FM‐corrected subscapular skinfold among Chinese, Lebanese and Malays by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANCOVA with Bonferroni multiple comparisons procedure.

Suprailiac skin fold

There was a significant difference in suprailiac skin fold among Chinese, Lebanese,

Malays, and Thais in both boys (P <0.0001) and girls (P <0.0001) after adjustment for

age and FM (Table 3.12). In both boys and girls, Chinese had the highest suprailiac

SFT, Malays and Thais had a similar value which was higher than for the Lebanese

(Figure 3.7).

a

a

a

a a


145

0

2

4

6

8

10

12

14

16

18

Boys Girls

Su

pra

ilia

c sk

info

ld (

mm

)


Figure 3.7. Ethnic differences in age‐ and FM‐corrected suprailiac skin fold among Chinese, Lebanese, Malays and Thais by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANCOVA with Bonferroni multiple comparisons procedure.

Abdominal skin fold

There were no significant differences in abdominal SFT between Malay and Chinese

boys (P = 0.307), while Malay girls had a significantly higher value than Chinese girls

(P = 0.046) after adjustment for age and FM (Figure 3.8).

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

Boys Girls

Ab

do

min

al s

kin

fold

(m

m)

Chinese Malays

Figure 3.8 Ethnic differences in age‐ and FM‐corrected abdominal skin fold among Chinese and Malays by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

a a aa

aa


146

Front thigh skin fold

Both Lebanese boys and girls had a higher front thigh SFT than their Malay

counterparts after adjustment for age and FM (P = 0.001 and P = 0.012 for boys and

girls, respectively) (Figure 3.9).

0

2

4

6

8

10

12

14

16

18

20

22

24

Boys Girls

Fro

nt

thig

h s

kin

fold

(m

m)

Malays Lebanese

Figure 3.9. Ethnic differences in age‐ and FM‐corrected abdominal skin fold among Lebanese and Malays by sex (P = 0.001 and P = 0.012 for boys and girls, respectively).

Medial calf skin fold

There was a significant ethnic difference in medial calf SFT among Lebanese, Malays

and Thais in each sex after adjustment for age and FM (P <0.0001 for both boys and

girls) (Table 3.12). In boys, Lebanese had a higher medial calf SFT than Malays, who

in turn had a higher value than Thais. In girls, Malays had a similar triceps SFT to

Lebanese with both having a higher mean value than Thais (Figure 3.10).


147

Figure 3.10. Ethnic differences in age‐ and FM‐corrected medial calf skin fold among Lebanese, Malays and Thais by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

Trunk skin fold

There was a significant ethnic difference in trunk SFT (sum of subscapular and

suprailiac sites) among Chinese, Lebanese and Malays in each sex after adjustment

for age and FM (P <0.0001) (Table 3.12). In boys, Chinese and Malays had a similar

trunk SFT and both had a higher value than the Lebanese. Chinese girls had a higher

trunk SFT than Malays, who in turn were higher than the Lebanese (Figure 3.11).


148

0

5

10

15

20

25

30

35

Boys Girls

Tru

nk

skin

fold

(m

m)


Figure 3.11. Ethnic differences in age‐ and FM‐corrected trunk skin fold among Chinese, Lebanese, and Malays by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

No significant differences in slopes of regression for Chinese, Lebanese and Malays

by general line model analysis were found in each sex (Figure 3.12). The interaction

terms were excluded in the final model. The stepwise multiple regression equation

for trunk SFT was:

LnTrunk SFT = 0.84 × LnFM + 0.12 ×E1 ‐ 0.21 × E2 + 1.23 (R2=0.79, SEE=0.28)

a

a


149

Figure 3.12. Relationship between trunk skin fold thickness and total body fat of Chinese,

Lebanese and Malays in each sex.

Trunk/extremity ratios

Three indices were used to describe the trunk/extremity ratio, including trunk (sum

Chinese × LnFM, P=0.580 Lebanese × LnFM, P=0.231



150

of subscapular and suprailiac)/upper extremity (sum of biceps and triceps) ratio,

suprailiac/upper extremity (sum of biceps and triceps) ratio, trunk (sum of

subscapular and suprailiac)/extremity (sum of biceps, triceps, front thigh, and

medial calf) ratio.

There was a significant ethnic difference in trunk/upper extremity ratio among

Chinese, Lebanese and Malays in each sex after adjustment for age and FM (P

<0.0001) (Table 3.12). In boys, Chinese had a higher trunk/upper extremity ratio

than Malays, who in turn was higher than the Lebanese. In girls, the Chinese had a

higher ratio than Lebanese and Malays with no significant differences in the latter

two groups (Figure 3.13). The differences between Chinese and Malay were more

apparent at lower FM levels (Figure 3.14). However, the interaction terms only

increased explained variance of the regression by 0.036 for boys and 0.025 for girls.

Accordingly, the interaction terms were excluded in the final model. The stepwise

multiple regression equation for trunk/upper extremity ratio was:

Ln (Trunk/upper extremity ratio) = 0.19×LnFM + 0.13×E1 ‐ 0.09×E2 – 0.43 (R2 = 0.38, SEE = 0.19)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Boys Girls

Tru

nk/

Up

per

ext

rem

ity

rati

o


Figure 3.13. Ethnic differences in age‐ and FM‐corrected trunk/upper extremity ratio among Chinese, Lebanese, and Malays by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

aa


151

Figure 3.14. Relationship between trunk/upper extremity ratio and FM of Chinese, Lebanese,

Malays and Thais in each sex.

A significant ethnic difference in suprailiac/upper extremity ratio was also found

among Chinese, Lebanese, Malays and Thais in each sex after adjustment for age




152

and FM (P<0.0001) (Table 3.12). In boys, Chinese and Thais had a similar

suprailiac/upper extremity ratio and both had a higher value than Malays, who in

turn were higher than the Lebanese. In girls, Chinese and Thais also had a similar

suprailiac/upper extremity ratio and both had a higher value than Malays and

Lebanese. There was no significant difference between the latter two groups (Figure

3.15). However, the difference between Chinese and Malays was most evident at a

lower FM level in each sex by testing the similarity of slopes. There was also a

significant difference in the slope of regression for Chinese and Thais with a higher

ratio at a lower FM level while a lower ratio at the higher FM level for a given FM in

Chinese in each sex (Figure 3.16). However, the interaction terms only increased the

explained variance of regression by 0.013 for boys and 0.015 for girls. Again, the

interaction terms were excluded in the final model. The stepwise multiple regression

equation for the suprailiac/upper extremity ratio was:

Ln (suprailiac/upper extremity ratio) = 0.28×LnFM + 0.22×E1 ‐ 0.14×E2 + 0.20×E3 ‐ 1.31

(R2 = 0.52, SEE = 0.22)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Boys Girls

Su

pra

ilia

c/U

pp

er e

xtre

mit

y ra

tio


Figure 3.15. Ethnic differences in age‐ and FM‐corrected suprailiac/upper extremity ratio

among Chinese, Lebanese, Malays and Thais by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

a aa a

b

b


153

Figure 3.16. Relationships between suprailiac/upper extremity ratio and FM of Chinese,

Lebanese, Malays and Thais in each sex.

Malays had a higher trunk/extremity ratio than Lebanese in each sex after

adjustment for age and FM (P <0.0001) (Figure 3.17) with the difference more

apparent at a high level of FM in girls (Figure 3.18). However, the interaction terms

Chinese × LnFM, P=0.003 Lebanese × LnFM, P=0.901 Thai × LnFM, P=0.106

Chinese × LnFM, P=0.001 Lebanese × LnFM, P=0.528 Thai × LnFM, P=1.000


154

only increased the explained variance of regression by 0.015 for girls so the

interaction terms were excluded in the final model. The stepwise multiple regression

equation for the trunk/ extremity ratio was:

Ln (Trunk/extremity ratio) = 0.22×LnFM ‐ 0.22× E2 + 1.23 (R2 = 0.47, SEE = 0.18)

0.0

0.1

0.2

0.3

0.4

0.5

Boys Girls

Tru

nk/

Ext

rem

ity

rati

o

Malays Lebanese

Figure 3.17. Ethnic differences in age‐ and FM‐corrected trunk/extremity ratio among Lebanese and Malays by sex.


155

Figure 3.18. Relationships between trunk/extremity ratio and FM of Lebanese and Malays in

each sex.

Upper/lower trunk fat

Subscapular skin fold/suprailiac skin fold ratio was used as the index of upper and

Lebanese × LnFM, P=0.679

Lebanese × LnFM, P=0.005


156

lower trunk fat model. Significant differences in subscapular/suprailiac ratio were

found among Chinese, Lebanese and Malays after adjustment for the trunk SFT and

age (P<0.0001, Table 3.12). Both Chinese boys and girls had a lower trunk

fat‐adjusted ratio compared with their Lebanese and Malay counterparts (Figure

3.19). No significant interactions between ethnicity and LnTrunk were found (Figure

3.20). So the interaction terms were excluded in the final model. The stepwise

multiple regression equation for the subscapular/suprailiac ratio was:

Ln (Subscapular/suprailiac ratio) = ‐ 0.17×Ln Trunk – 0.20 ×E1 + 0.07×E2 + 0.08 ×Sex + 0.46

(R2 = 0.34, SEE = 0.25)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Boys Girls

Su

bsc

apu

lar/

Su

pra

ilia

c ra

tio


Figure 3.19. Ethnic differences in age‐ and trunk SFT‐corrected subscapular/suprailiac ratio among Chinese, Lebanese and Malays by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

a

a a

a


157

Figure 3.20. Relationship between subscapular/suprailiac ratio and FM of Chinese, Lebanese and Malays in each sex.

Chinese × LnFM, P=0.559Lebanese × LnFM, P=0.269

Chinese × LnFM, P=0.848Lebanese × LnFM, P=0.680


158

Waist circumference

Significant differences in WC among ethnic groups was found in both boys and girls

for a given BMI and age (P<0.0001) (Table 3.12). Chinese boys had a similar WC to

Thai boys and both had a higher WC compared with their Lebanese counterparts,

who in turn were higher than Malays for a given BMI and age. A similar difference

was found among girls (Figure 3.21). No significant difference in slopes of regression

were found for Chinese, Lebanese, Malays and Thais by general line model analysis

(Figure 3.22), so the interaction terms were excluded in the final model. The

stepwise multiple regression equation for WC was:

LnWC = 0.73×LnBMI + 0.01×age + 0.01×sex + 0.05×E1 + 0.03×E2 + 0.06×E3 + 1.85

(R2 = 0.92, SEE = 0.05)

50

52

54

56

58

60

62

64

66

68

70

Boys Girls

Wai

st c

ircu

mfe

ren

ce (

cm)


Figure 3.21. Ethnic differences in age‐ and BMI‐corrected waist circumference among Chinese, Lebanese, Malays and Thais by sex. Sharing the same letter indicates no significant difference between groups in each sex using ANOCVA with Bonferroni multiple comparisons procedure.

aa

aa


159

Figure 3.22. Relationships between WC and BMI of Chinese, Lebanese, Malays and Thais in each sex.

3.2.3.2 Sex difference in body fat distribution

There was no significant difference between boys and girls in all skin fold variables in

Chinese × LnBMI, P=0.673 Lebanese × LnBMI, P=0.071 Thai × LnBMI, P=0.240

Chinese × LnBMI, P=0.635 Lebanese × LnBMI, P=0.053 Thai × LnBMI, P=0.402


160

each ethnic group except for higher biceps SFT in Lebanese boys and lower triceps

and medial calf SFT in Thai boys after adjustment for FM and age. No significant sex

differences in trunk/extremity ratios were found except for lower suprailiac/upper

extremity ratio in Lebanese boys and Malay boys. Both Lebanese and Malay boys

had a higher subscapular/suprailiac ratio than their counterparts at a given trunk SFT.

Boys tended to have a higher WC than girls which was only significant in the Chinese

and Malays after adjustment for BMI and age (Table 3.12).

3.2.3.3 Age difference in body fat distribution

For a given total body fat, there was no significant difference in SFT at different sites

among age groups in each ethnic group except for the biceps SFT in Lebanese,

triceps, subscapular, abdominal, and trunk SFT in Malays after adjustment for sex

and FM (Figures 3.23‐3.30). Lebanese aged 8 y had a higher biceps SFT than their

counterparts aged 9 y. Malays aged 10 y had significantly lower triceps, subscapular,

abdominal, and trunk SFT than their counterparts aged 9 y.

No significant difference in each trunk/extremity ratio among age groups in each

country was found at a given of FM (Figures 3.31‐3.33).

No significant difference in subscapular/suprailiac ratio among age groups in each

ethnicity was found except for Malays aged 10 y who had a lower ratio than the

other two age groups at a given trunk fat and sex (Figure 3.34). In Chinese, Malays

and Thais, children aged 8 y had a significantly lower in WC than those aged 9 and

10 y (P <0.05). In Lebanese, children aged 10 y had a significant higher WC than the

other two age groups (P <0.05) (Figure 3.35).


161

0

2

4

6

8

10

12


Bic

eps

skin

fold

(m

m)

8 yr 9 yr 10 yr

Figure 3.23. Age differences in sex‐ and FM‐corrected biceps skin fold by ethnicity using

ANCOVA with Bonferroni multiple comparisons procedure. *Significant difference between 8 yr and 9 yr groups in Lebanese (P <0.05).

0

2

4

6

8

10

12

14

16

18


Tri

cep

s sk

info

ld (

mm

)

8 yr 9 yr 10 yr

Figure 3.24. Age differences in sex‐ and FM‐corrected triceps skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. *Malay children aged 10 y had significantly lower triceps than the other two age groups (P <0.05).

*

*


162

0

2

4

6

8

10

12

14

16


Su

bsc

apu

lar

skin

fold

(m

m)

8 yr 9 yr 10 yr

Figure 3.25. Age differences in sex‐ and FM‐corrected subscapular skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. *Malay children aged 10 y had significantly lower subscapular SFT than the other two age groups (P <0.05).

0

2

4

6

8

10

12

14

16

18


Su

pra

ilia

c sk

info

ld (

mm

)

8 yr 9 yr 10 yr

Figure 3.26. Age differences in sex‐ and FM‐corrected suprailiac skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. No significant difference among age groups was found in each ethnic group.

*


163

0

2

4

6

8

10

12

14

16

18

20

Chinese Malays

Ab

do

min

al s

kin

fold

(m

m)

8 yr 9 yr 10 yr

Figure 3.27. Age differences in sex‐ and FM‐corrected abdominal skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. * Significant difference between 9 and 10 y group in Malays (P <0.05).

0

2

4

6

8

10

12

14

16

18

20

22

Lebanese Malays

Fro

nt

thig

h s

kin

fold

(m

m)

8 yr 9 yr 10 yr

Figure 3.28. Age differences in sex‐ and FM‐corrected front thigh skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. No significant difference among age groups was found in each ethnic group.

*


164

0

2

4

6

8

10

12

14

16

Lebanese Malays Thais

Med

ical

cal

f sk

info

ld (

mm

)

8 yr 9 yr 10 yr

Figure 3.29. Age differences in sex‐ and FM‐corrected medial calf skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. No significant difference among age groups was found in each ethnic group.

0

5

10

15

20

25

30

35


Tru

nk

skin

fold

(m

m)

8 yr 9 yr 10 yr

Figure 3.30. Age differences in sex‐ and FM‐corrected trunk skin fold by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. *Significant difference between 9 and 10 y group in Malays (P <0.05).

*


165

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4


Tru

nk/

Up

per

ext

rem

ity

rati

0

8 yr 9 yr 10 yr

Figure 3.31. Age differences in sex‐ and FM‐corrected trunk/upper extremity ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. No significant difference among age groups in each ethnic group.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8


Su

pra

ilia

c/U

pp

er e

xtre

mit

y ra

tio

8 yr 9 yr 10 yr

Figure 3.32. Age differences in sex‐ and FM‐corrected suprailiac/upper extremity ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. No significant difference among age groups in each ethnic group.


166

0

0.1

0.2

0.3

0.4

0.5

Lebanese Malays

Tru

nk/

Ext

rem

ity

rati

o

8 yr 9 yr 10 yr

Figure 3.33. Age differences in sex‐ and FM‐corrected trunk/extremity ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. No significant difference among age groups in each ethnic group.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4


Sb

uca

pu

lar/

Su

pra

ilia

c ra

tio

8 yr 9 yr 10 yr

Figure 3.34. Age differences in sex‐ and trunk SFT‐corrected subscapular/suprailiac ratio by ethnicity using ANCOVA with Bonferroni multiple comparisons procedure. * Significant difference between Malay children aged 10 y than the other two age groups (P <0.001).

*


167

50

55

60

65

70


Wai

st c

ircu

mfe

ren

ce (

cm)

8 yr 9 yr 10 yr

Figure 3.35. Age differences in WC adjusted for BMI and sex using ANCOVA with Bonferroni multiple comparison procedure by ethnicity. *Children aged 8 y had a significantly lower WC than the other two age groups (P<0.05). # Children aged 10 y had a significantly higher WC than the other two age groups (P<0.05).

3.2.4 Discussion

The current study explored ethnic differences in body fat distribution among the

four ethnic groups in terms of absolute skin fold thickness at different sites, WC and

the trunk to extremity ratios. Chinese boys and girls had a similar fat distribution

pattern with Thai boys and girls and both groups had more trunk fat or central fat

depots than Malays and Lebanese. Malays tended to have higher trunk/extremity

ratios than Lebanese whereas lower WC than Lebanese at a given BMI.

Fat distribution patterns are more closely associated with cardiovascular disease,

type 2 diabetes and metabolic risk factors than overall adiposity in both adults and

children and contribute to explanations of ethnic differences in cardiovascular

morbidities and diabetes (Berman et al., 2001; Daniels et al., 1997; Ehtisham et al.,

2005; Freedman et al., 1989; He et al., 2002; Okosun, 2000; Okosun et al., 2000). A

number of previous studies have compared the body fat distribution patterns in

Asians with those in Caucasians and blacks. In general, Asians have more central fat

depots and visceral fat accumulation than whites in both children (He et al., 2002;

Malina et al., 1995) and adults (Chandalia et al., 1999; Lear et al., 2007; McKeigue et

*#

*

*


168

al., 1991; Park et al., 2001; Raji et al., 2001; Wu et al., 2007) regardless of body

composition measurement method used. Asian girls enter puberty earlier than

white girls which results in shorter legs due to endplate closure in the long bones.

However, trunk length continues to grow indicating a long period of pubescent

growth. This apparent longer period of pubescence might contribute to the greater

trunk FM in Asian girls (Novotny, Daida, Grove, Achsrya, & Vogt, 2003). It is difficult

to make comparisons regarding differences in fat patterns between Asian and other

ethnic groups because of the limited data among Asian children and adolescents

from different backgrounds. Several studies in adults have reported on differences

among Asians from different origins. For example, Lear et al. (2007) reported that

Chinese (China, Hong Kong, and Taiwan) had less DXA‐derived FM while South

Asians (Bangladesh, India, Nepal, Pakistan, and Sri Lanka) but had more FM than

their European counterparts at the same BMI. South Asians had higher FM and total

abdominal adipose tissue (as determined by CT) than Chinese at the same BMI. In

another study conducted in Indian, Malay and Chinese Singaporeans (Hughes, Aw,

Kuperan & Choo, 1997), Indians had more central fat via WHR than Malays and

Chinese, and both Malays and Indians had higher abdominal diameter (as measured

at the level of the iliac crests with a ruler and tape) than Chinese. Although different

adiposity indices were used, our study confirms the differences in fat distribution

among Asian children from different origins.

WC, a simple and practical index of visceral adipose tissue (Daniels, Khoury &

Morrison, 2000; Pouliot et al., 1994), has been proposed as effective to predict

metabolic risk factors in both children and adults (Bacha et al., 2006; Janssen et al.,

2004; Lee et al., 2006; Lofren et al., 2004; Ng et al., 2007). WC is also recognized as a

key component of the metabolic syndrome in both children and adults (Alberti et al.,

2006; Zimmet et al., 2007). Therefore, some countries have developed their own

WC percentiles for children and adolescents. Sung et al. (2008) and Yan et al. (2008)

developed WC percentiles for Hong Kong children aged 6‐18 y and children from

Xinjiang province, China aged 7‐18 y, respectively. Comparisons of the 50th

percentiles for WC of Chinese children with those of white, black and Mexican


169

children, shows a lower WC in Chinese children across ages in each sex. The lower

WC in Asians can be explained, in part, by the shorter height compared to other

ethnic groups. Ehtisham et al. (2005) indicated that after controlling for height,

weight or BMI, both Asian boys and girls aged 14‐17 y appear to have higher WC

than their white counterparts. But Novotny et al. (2007) found that Asian girls

(11.8±0.05 y) had lower WC than white and Hispanic children even after controlling

for height and weight. The inconsistency in results may be, in part, due to the

difference in age or stage of maturation and WC measurement site used (narrowest

part of the waist, midway between the 10th rib and the iliac crest for the former

study, at the level of umbilicus for the latter study). Ethnic differences in body

composition vary according to age or pubertal status (Kimm et al., 2001) and

significant differences in WC values measured at different sites have been reported

(Wang et al., 2003). In addition, the difference in the Asian population in the two

studies might be a major contributing factor to the inconsistent findings. The Asians

in the former study were South Asians including Indian, Pakistani, Bangladeshi, and

Sri Lankan, while those in the latter study are general Asian population. The

difference in other body composition variables among Asians from different

backgrounds has been reported (Lear et al., 2007) so it may be hypothesized that

WC will differ among Asians from different origins. Our study confirms this

hypothesis indicating that Chinese and Thai children had higher WC values than the

Lebanese, who in turn were higher than Malays.

In addition, our results showed that Lebanese children had higher leg SFT than

Malays at a given total body fat. Leg fat is negatively related to CV risk factors and

metabolic complications (Tatsukawa, Kurokawa, Yoshimatsu & Sakata, 2000;

Williams, Hunter, Kekes‐Szabo, Snyder & Treuth, 1997).

Our study also aimed to determine the effect of age and sex on body fat distribution.

There were no apparent age and sex differences in fat distribution variables except

for WC in the current study, a finding which is not consistent with previous studies.

One of the main reasons is that our participants were all in the pre‐pubertal stage

and within quite a narrow age range. Sex differences in body composition are more


170

apparent after the onset of the growth spurt (Fomon et al., 1982; Mast et al., 1998).

An interesting finding in our study relates to the coefficients of regression equations

for trunk/extremity ratios. All were positive which indicates a shift in a fat patterning

from the extremities to the trunk with increasing body fatness. Ramirez & Mueller

(1980) reported a similar shift in Tokelau children.

One of the main limitations of the present study was the field‐type measures of

body fat distribution used rather than more advanced and specific measures such as

DXA, MRI or CT which may provide additional insight into the racial influence on fat

distribution. However, compared to these expensive, advanced techniques which

also require highly trained technicians, the anthropometric measures used are

relatively simple and inexpensive. These advantages are not insignificant for

epidemiological research and clinical practice as they provide an excellent

opportunity to screen people with high risk. Moreover, measurement of

subcutaneous fat with a skin fold caliper has been shown to correlate well with that

measured with CT or MRI (Hayes, Sowood, Belyavin, Cohen & Simth, 1988;

Orphanidou, McCargar, Birmingham, Mathieson & Goldner, 1994) and WC

correlates well with fat distribution measured by DXA (Daniels et al., 2000). The

second limitation to our study was the non‐random population sample in each

ethnic group. However, our study sample was purposively recruited to encompass a

number of overweight and obese children. Given that fat distribution may differ

with BMI status, data based on this sampling method are more convincible. In

addition, habitual physical activity and dietary intake was not collected in the

current study, which may have the potential to influence body fat distribution.

In conclusion, the current study indicates that Asian pre‐pubertal children from

different origins vary in body fat distribution with greater trunk fat depots in

Chinese and Thai children compared with Malays who in turn have higher values

than Lebanese. These results indicate the importance of population‐specific WC

cut‐off points or other fat distribution indices to identify the population at risk of

obesity‐related health problems.


171

CHAPTER 4 VALIDATION OF BIA FOR TBW ANALYSIS AGAINST THE DEUTERIUM DILUTION TECHNIQUE IN ASIAN CHILDREN

Modified from: Liu A, Byrne NM, Ma G, Nasreddine L, Trinidad TP, Kijboonchoo K, Ismail MN, Kagawa M, Poh BK, Hills AP (2011). Validation of bioelectrical impedance analysis for total body water assessment against the deuterium dilution technique in Asian children. Accepted by European Journal of Clinical Nutrition.

4.1 INTRODUCTION

The increasing prevalence of obesity is a major public health problem in both the

developed and the developing world and is related to the increased risk of chronic

diseases such as cardiovascular disease and type‐2 diabetes. In Asia, there is an

alarming increase in the proportion of overweight and obese children and

adolescents with a parallel increase in the incidence of associated chronic disease,

especially in countries undergoing nutritional and lifestyle transition.

Obesity is characterised by an excess of body fat. The accurate assessment of body

composition, including body fat, is increasingly important for research and clinical

practice, especially as it relates to the monitoring of obesity prevention and

treatment efforts. The most common means of defining paediatric obesity is to

utilize the BMI. Whilst BMI‐for‐age is a useful indication of size and shape of a child

relative to normative data, BMI is not a measure of body composition. The major

limitation of the use of BMI in children and adolescents is its inability to differentiate

levels of fatness and leanness among individuals (Freedman et al., 2005; Gallagher

et al., 1996; Roubenoff et al., 1995). Some studies have indicated that a BMI score

only enables a poor to fair identification of those individuals who are truly

overweight and obese as determined from objective measures of %BF (Sampei et al.,

2001; Wickramasinghe et al., 2005). Moreover, as BMI does not predict the change

in body fatness which is a more appropriate approach to evaluate the effect of

intervention programs on obesity prevention and treatment (Dao et al., 2004) it is

appropriate to use more accurate and objective measures of body fat in addition to


172

BMI (Wickramasinghe et al., 2005).

BIA is one of the most appropriate body composition assessment techniques

suitable for use in the field with children (Jürimäe & Hills, 2001), however the

approach depends on several assumptions regarding the composition of the FM and

the FFM. One of the assumptions relates to the constancy of the relationship

between these body compartments however the level of hydration and the density

of the FFM are not constant among individuals and vary according to age, gender

and ethnicity (Deurenberg, Deurenberg‐Yap & Schouten, 2002). Therefore, there are

inherent problems in the application of existing prediction equations to samples

other than individuals who have the same physical characteristics as the sample

used to derive the equation. To date, most of the existing BIA prediction equations

have been derived from Caucasian individuals with a few equations developed

specifically for Japanese children (Heyward & Stolarczyk, 1996; Nielsen et al., 2007).

Some studies have assessed whether ethnicity contributes to the relationship

between bioimpedance and body composition in adolescents (Going et al., 2006;

Haroun et al., 2010; Sluyter et al., 2010) rather in children. Therefore, the validity of

BIA in a multi‐ethnic sample of Asian children is unknown. Accordingly, the objective

of the current study was to assess the validity of BIA for the estimation of TBW and

FFM using the deuterium dilution technique as a reference.


173

4.2 METHODOLOGY

4.2.1 Participants

A total of 948 participants (492 boys and 456 girls) were recruited from five

Asian countries, including China, Lebanon, Malaysia, The Philippines and

Thailand. In each country, a non‐random purposive sampling approach was

used which aimed to enrol children encompassing a wide BMI range for each

year of age between 8‐10 yr and each sex. The sample size of each country is

shown in Table 4.1. Ethnicity was determined by self‐identification and those

whose parents were identified as having the same origin were included.

Additional inclusion criteria required that participants be healthy, at Tanner

stage 1 of puberty, and free from any diagnosed medical condition that might

potentially interfere with body composition measurement. The study protocol

was explained to the parent(s) and the children and written consent obtained

from each child and/or their parent(s).

In each country, participants in each gender and age group were randomly

divided into two groups, a validation group (328 boys and 302 girls) and a

cross‐validation group (164 boys and 154 girls) (Table 4.1). There were no

significant differences in age, height and weight between validation and

cross‐validation groups.


174

Table 4.1. Sample size of validation and cross‐validation group in each country by sex

Validation Group

Cross‐validation group

Boys Girls All Boys Girls All

China 133 98 231 66 49 115

Lebanon 25 24 49 13 13 26

Malaysia 69 60 129 34 31 65

The Philippines 34 39 73 17 20 37

Thailand 67 81 148 34 41 75

All 328 302 630 164 154 318

4.2.2 Anthropometric measurements

Body height was measured using a portable Holtain stadiometer to the nearest 0.1

cm. The detailed technique for measuring height was described in the section

3.2.2.2. Body weight was measured using a SECATM electronic scale (Hamburg,

Germany) to the nearest 0.1 kg with participants wearing only underwear after

urinating in the morning. BMI was calculated as body weight (kg) divided by the

square of height (m).

Pubertal status of each participant was assessed according to the criteria of Tanner

(Tanner & Whitehouse, 1976) by trained investigators.

4.2.3 Bioelectrical impedance analysis measurement

A tetra‐polar single frequency (200μA at 50 kHz) electrical bioimpedance analyzer

(Imp DF50, ImpediMed Limited, Australia) was used to measure impedance (Z, Ω),

resistance (R, Ω) and reactance (Xc, Ω). In order to reduce the measurement error,

all participants were asked to follow pre‐test guidelines (Heyward, 1998):

1) Do not eat or drink within 4 h of the test.

2) Do not exercise within 12 h of the test.

3) Urinate within 30 min preceding the test.


175

BIA measurement was taken on the right side of the body and standardized testing

procedures followed (Heyward, 1998).

1) Participant removed all jewellery and shoes and socks;

2) Participant lay in a fully supine position on a non‐conductive surface in a room

with normal ambient temperature (~22°C), with legs and arms abducted to

each other and palms flat against the surface.

3) Skin contact areas were cleaned with alcohol pads.

4) The sensor (proximal) electrodes (a) placed on the dorsal surfaces of the right

wrist so that the upper border of the electrode bisects the head of the ulna

and (b) on the dorsal surface of the ankle so that the upper border of the

electrode bisects the medial and lateral malleoli. A measuring tape and

surgical marking pen was used to mark these points for electrode placement.

5) The source (distal) electrodes were placed at the base of the second or third

metacarpal‐phalangeal joints of the hand and foot ensuring at least 5 cm

between the proximal and distal electrodes.

6) Lead wires attached to the appropriate electrodes.

7) Participants’ legs and arms abducted to approximately 45°with no contact

between the thighs and between the arms and the trunk.

8) Measurement completed approximately 5 minutes after the participant lies

down and relaxes.

Resistance, reactance, and impedance were read directly from the device. RI was

calculated as the square of height divided by resistance (RI = height (cm)2/R (Ω).

Before each testing session, the device was checked using the test cell provided by

the manufacturer.

4.2.4 Deuterium dilution technique

TBW was assessed by deuterium dilution technique and used as the criterion

method for the estimation of TBW and FFM. A detailed overview of this technique

was provided in section 3.1.2.3. In brief, a 10% D2O dose of 0.5 g per kg body weight


176

was given orally to the participant after collection of the baseline urine sample. The

post‐dose sample was collected about 5 h after the administration of dose. IRMS

was used to analyse the urine samples and TBW determined from the equation as

follows:

TBW (kg) = 041.1

1)(x

EpEs

EtEax

a

TA

Subsequently, FFM was derived from TBW using a hydration coefficient. Lohman’s

age‐ and gender‐specific constants for hydration of the FFM for children were used

to calculate FFM (Lohman, 1986).

4.2.5 Statistical analysis

BIA measures, TBW, and FFM ≥3.3 internal SD scores were classified as outliers and

excluded from the final analysis. Descriptive statistics was reported as mean±SD.

The total sample from these five study sites were randomly separated into a

validation and cross‐validation group. The validation group was used to develop the

prediction equations for TBW and FFM and to evaluate the precision of the

equations. The cross‐validation group was used to evaluate the accuracy of the

equations. Precision is the ability to explain the variation of the dependent variables

within the sample from which it was derived. Accuracy evaluates the performance

of a prediction equation when it is applied to an independent sample.

In the validation group, stepwise multiple regression analysis and “all‐possible

subsets” regression procedure were employed to develop BIA equations. TBW and

FFM derived from deuterium dilution methods were used as dependent variables

for the development of prediction equations separately. The potential predictor

variables included age, sex (male=1, female=0), height, weight, RI, XC, and ethnicity

(coded as four dummy variables for Lebanese, Malays, Filipinos and Thais with

Chinese as the reference category). Potential interactions between ethnicity and RI,

between sex and RI, and between age and RI were examined. Mallow’s Cp statistic

(Mallows, 1973) was used as a measure of the appropriate number of predictors.


177

High R2 values, small root mean square error (RMSE) and small Cp values indicated

optimal models. These equations were examined for the significance of the

regression coefficients. Standardized regression coefficients were calculated to

quantify the independent contribution of predictors.

The multiple regression equations developed from the validation group were

cross‐validated on the cross‐validation group. The pure error (PE) calculated as the

square root of the mean of squares of differences between measured and predicted

values was used to assess the performance of the prediction equations on

cross‐validation. The smaller the PE, the greater the accuracy of the equation. There

is no criterion value for PE that indicted successful cross‐validation, but the PE

should be similar to the value of the RMSE of the same equation from its validation

(Sun et al., 2003). Moreover, correlation between measured and predicted values

(Pearson correlation coefficient), the bias (difference between measured and

predicted values tested against zero using paired t‐test) was used to assess the

performance of the equations. Furthermore, the approach of Bland and Altman was

used to assess the agreement between BIA and deuterium dilution methods. This

statistical approach is recognised as the most appropriate way to compare the

ability of different methods to measure the same parameter (Bland & Altman, 1986).

Limits of agreement (expressed as 2 SD above and below the bias) and the

dependence of the bias on the mean of measured and predicted values was

analysed.

Moreover, external BIA equations were chosen for cross‐validation in our Asian

children. The criteria used to choose external BIA equations were: 1) use single

frequency tetra‐polar BIA as the indirect method; 2) use of FFM or TBW as the

outcome measure; 3) developed from Caucasian children; 4) age comparable to our

participants; 5) developed from both genders; 6) no use of predictors other than age,

gender, height, weight, and BIA‐based indices. Three FFM equations (Deurenberg et

al., 1991; Lohman, 1992; Rush et al., 2003) and two TBW equations (Davies & Preece,

1988; Kushner et al., 1992) were obtained. The Bland and Altman approach was

used to assess the agreement between the external BIA equations and deuterium


178

dilution methods.

All statistical analyses were performed with the SAS 8.02 and a two‐tailed P value of

<0.05 was regarded as statistically significant.


179

4.3 RESULTS

The characteristics of participants are shown in Table 4.2. There were wide ranges

for height, weight, BMI, TBW, FFM and %BF. The mean BMI was 18.9 kg/m2 with a

range of 12.2‐34.9 kg/m2. The mean FFM was estimated at 24.4 kg with a range of

12.5‐48.2 kg and the estimated mean %BF was 27.9% with a range of 5.5‐54.5%. The

resistance and reactance values ranged from 464.1 to 943.4 Ω and 32.0‐106.1 Ω,

respectively.

Table 4.2. Characteristics of participants

Mean±SD Range

Age (y) 9.4±0.8 8‐10

Height (cm) 135.0±7.8 112.1‐163.1

Weight (kg) 34.9±10.3 16.7‐ 77.7

BMI (kg/m2) 18.9±4.1 12.2 ‐ 34.9

TBW (kg) 18.5±3.8 9.2 ‐ 36.8

FFM (kg) 24.4±5.1 12.5 ‐ 48.2

%BF 27.9±9.5 5.5 ‐ 54.5

Resistance (Ω) 675.1±84.8 464.1‐943.4

Reactance (Ω) 61.1±7.8 32.0‐106.1

4.3.1 Development of BIA equations

No significant interactions between ethnicity and RI, between sex and RI and

between ethnicity and sex were found. Therefore, a single equation was developed

for the whole validation sample.

4.3.1.1 Prediction equation for FFM

The results of the regression analysis with FFM as dependent variable are given in

Table 3.3. Weight, height, RI, age, sex and ethnicity were all significant predictors of

FFM. Of these predictors, weight entered the model first and the standardized


180

estimate for weight was the largest in the model predicting FFM. RI entered the

model second and explained the second largest variance of the model (Table 3.4).

The equation for the estimation of FFM accounted for 88.3% of the variation in FFM

with RMSE 1.7 kg. The final prediction equation of BIA for estimation of FFM was as

follows:

FFM (kg) = 0.299×Height2 (cm)/resistance (Ω) + 0.086×Height (cm) + 0.245×Weight (kg) +

0.260×Age (yr) + 0.901×Sex (male=1, female=0) ‐ 0.415×Ethnicity (Thai

ethnicity=1, others=0) ‐ 6.952

4.3.1.2 Prediction equation for TBW

RI, weight, height, age, sex and ethnicity were all significant predictors of TBW

(Table 4.3). Of these predictors, the standardized estimate for weight was the largest

in the model predicting TBW, following by RI (Table 4.4). The equation for the

estimation of TBW accounted for 88.0% of the variation in TBW with RMSE 1.3 kg.

The final prediction equation of BIA for estimation of TBW was as follows:

TBW (kg) = 0.231×Height2 (cm)/resistance (Ω) + 0.066×Height (cm) + 0.188×Weight (kg) +

0.128×Age (yr) + 0.500×Sex (male=1, female=0) – 0.316×Ethnicity (Thai

ethnicity=1, others=0) ‐ 4.574


181

Table 4.3. Stepwise multiple regression models for FFM and TBW

Weight (kg) RI (cm2/Ω) Height (cm) Sex Thais Age (y) Intercepts R2 RMSE (kg) C(P)

FFM

0.442±0.009 8.751±0.324 0.793 2.2 446.5

0.247±0.013 0.425±0.022 3.750±0.374 0.867 1.8 80.7

0.252±0.012 0.406±0.023 0.765±0.145 3.726±0.367 0.873 1.8 52.5

0.239±0.012 0.309±0.027 0.099±0.016 0.905±0.142 ‐6.582±1.687 0.880 1.8 14.8

0.243±0.012 0.311±0.027 0.081±0.017 0.913±0.142 0.267±0.101 ‐6.724±1.680 0.881 1.8 9.7

0.245±0.012 0.299±0.027 0.086±0.017 0.901±0.141 ‐0.415±0.164 0.260±0.100 ‐6.952±1.675 0.883 1.7 5.3

TBW

0.337±0.007 6.811±0.244 0.797 1.7 421.9

0.191±0.010 0.318±0.017 3.066±0.282 0.869 1.4 55.5

0.182±0.010 0.260±0.020 0.062±0.012 ‐3.393±1.297 0.874 1.3 30.4

0.184±0.009 0.239±0.021 0.071±0.012 0.505±0.109 ‐4.330±1.292 0.878 10.7

0.185±0.009 0.230±0.021 0.074±0.012 0.496±0.108 ‐0.322±0.126 ‐4.509±1.288 0.879 1.3 6.2

0.188±0.009 0.231±0.021 0.066±0.013 0.500±0.108 ‐0.316±0.126 0.128±0.077 ‐4.574±1.287 0.880 1.3 5.4

RI: resistance index=height*height/resistance; RMSE: root mean square error. FFM: fat‐free mass; TBW: total body water; Sex: male=1, female=0.


182

Table 4.4. Standardized coefficients of the final model for estimation of FFM and TBW

FFM TBW

Standardized estimate

VIF Standardized estimate

VIF

Weight 0.493 3.3 0.497 3.3

RI 0.335 4.9 0.340 4.9

Height 0.136 3.9 0.136 3.9

Sex 0.091 1.1 0.066 1.1

Thais ‐0.035 1.0 ‐0.036 1.0

Age 0.041 1.3 0.027 1.3

4.3.1.3 Ethnic‐specific equations

Ethnic‐specific prediction equations for FFM and TBW were also developed for the

five ethnic groups (Table 4.5). Weight and RI were included into each equation and

similar to prediction equation for FFM and TBW derived from total sample, weight

explained the highest variance of both FFM and TBW, followed by RI in each

ethnic‐specific prediction equation. The explained variance of the equation for

estimation of FFM by the predictor variables ranged from 83.8‐92.5% with RMSE

ranged from 1.2‐2.1 kg. The explained variance of the equation for estimation of

TBW by the predictor variables ranged from 83.8‐92.3% with RMSE ranged from

0.9‐1.6 kg.


183

Table 4.5. BIA prediction equations for FFM and TBW for each ethnic group

R2 RMSE (kg)

FFM

Chinese FFM (kg) = 0.276×RI (cm2/Ω) + 0.077×Height (cm) + 0.274×Weight (kg) + 0.471×Age (yr) + 0.718×Sex ‐ 7.854 0.883 1.7

Lebanese FFM (kg) = 0.192×RI (cm2/Ω) + 0.227×Height (cm) + 0.187×Weight (kg) + 1.510×Sex ‐ 19.072 0.912 1.2

Malays FFM (kg) = 0.353×RI (cm2/Ω) + 0.119×Height (cm) + 0.221×Weight (kg) + 0.572×Sex + 0.081×Xc ‐ 14.261 0.841 2.1

Filipinos FFM (kg) = 0.296×RI(cm2/Ω) + 0.363×Weight (kg) + 1.356×Sex + 2.611 0.879 1.5

Thais FFM (kg) = 0.598×RI (cm2/Ω) + 0.143×Weight (kg) + 0.322×Age (yr) + 0.566×Sex + 0.051×Xc ‐ 3.849 0.925 1.3

TBW

Chinese TBW (kg) = 0.211×RI (cm2/Ω) + 0.059×Height (cm) + 0.210×Weight (kg) + 0.283×Age (yr) + 0.350×Sex ‐ 5.182 0.880 1.3

Lebanese TBW (kg) = 0.154×RI (cm2/Ω) + 0.169×Height (cm) + 0.141×Weight (kg) + 0.953×Sex ‐ 13.944 0.910 0.9

Malays TBW (kg) = 0.281×RI (cm2/Ω) + 0.081×Height (cm) + 0.170×Weight (kg) + 0.063×Xc – 9.685 0.838 1.6

Filipinos TBW (kg) = 0.222×RI (cm2/Ω) + 0.282×Weight (kg) + 0.898×Sex + 2.152 0.879 1.2

Thais TBW (kg) = 0.480×RI (cm2/Ω) + 0.102×Weight (kg) + 0.165×Age (yr) + 0.045×Xc – 2.595 0.923 1.0

FFM; fat‐free mass; TBW: total body water; RI: resistance index=height×height/resistance; Xc: reactance; Sex: male=1, female=0; RMSE: root mean square error.


184

4.3.1.4 BMI‐specific prediction equations

BMI‐specific prediction equations for FFM and TBW were also developed for each

BMI category (Table 4.6). The explained variance of the equation for estimation of

FFM by the predictor variables ranged from 77.9‐82.3% with RMSE ranged from

1.5‐2.2 kg. The explained variance of the equation for estimation of TBW by the

predictor variables ranged from 77.6‐81.9% with RMSE ranged from 1.1‐1.7 kg. The

R2 of the equation for obese children were slightly lower than the other two groups

while RMSE was higher.

4.3.2 Cross‐validation of the equations

The developed equations were applied to the cross‐validation group to evaluate

their accuracy.

4.3.2.1 Cross‐validation of equations for FFM

No significant difference between measured and predicted FFM for the whole

cross‐validation sample was found (Bias = ‐0.2±1.9 kg, PE = 1.8±2.6 kg). Measured

FFM correlated highly with predicted FFM (r = 0.94, P<0.0001) (Figure 4.1). The

difference between measured and predicted FFM plotted against the mean of the

predicted and measured FFM is shown in Figure 4.2. The graph identified

approximately 17 participants (5.3%) whose residuals exceeded the 95% confidence

limits: 4 over‐predicted and 13 under‐predicted FFM. A marginal correlation

between difference of measured and predicted FFM and mean of measured and

predicted FFM (r = 0.110, P = 0.051) was observed. The plots showed that the

prediction equation for estimation of FFM tended to overestimate FFM at lower

level of FFM while underestimate FFM at higher level of FFM.


185

Table 4.6. BIA prediction equations for FFM and TBW for each BMI category

R2 RMSE

(kg)

FFM

BMI z score < 1SD FFM (kg) = 0.321×RI(cm2/Ω) + 0.387×Weight (kg) + 1.016×Sex +0.317×Age + 0.562 ×Malays – 0.850 0.823 1.5

BMI z score =1‐2SD FFM (kg) = 0.196×RI(cm2/Ω) + 0.099×Height (cm) + 0.283×Weight (kg) + 1.333×Sex – 4.962 0.791 1.5

BMI z score ≥2SD FFM (kg) = 0.430×RI(cm2/Ω) + 0.277×Weight (kg) ‐ 0.994 ×Thais + 2.162 0.779 2.2

TBW

BMI z score < 1SD TBW (kg) = 0.233×RI (cm2/Ω) + 0.040×Height (cm) + 0.266×Weight (kg) + 0.607×Sex + 0.452×Malays ‐ 2.463 0.819 1.1

BMI z score =1‐2SD TBW (kg) = 0.152×RI (cm2/Ω) + 0.080×Height (cm) + 0.204×Weight (kg) + 0.794×Sex – 3.751 0.781 1.1

BMI z score ≥2SD TBW (kg) = 0.323×RI (cm2/Ω)+ 0.212×Weight (kg) ‐ 0.720 ×Thais + 1.872 0.776 1.7

FFM; fat‐free mass; TBW: total body water; RI: resistance index=height×height/resistance; Sex: male=1, female=0; RMSE: root mean square error.


186

Figure 4.1. Scatter plot of FFM measured by D2O against estimated FFM by BIA in Asian

children.

Figure 4.2. Difference in FFM measured by D2O and estimated by BIA in Asian children.

In order to ensure that the prediction equation derived from the total validation

sample is valid at each ethnic group, the predicted FFM was compared with the

measured FFM at each ethnic group. No difference between measured and

Mean + 2SD

Mean ‐ 2SD

Mean


187

predicted FFM in each ethnic group was found. The bias was ‐0.5 kg, ‐0.3 kg, ‐0.2 kg,

‐0.1 kg and 0.4 kg for Chinese, Lebanese, Malays, Filipinos and Thais, respectively,

and the pure error was 1.8 kg, 1.4 kg, 2.1 kg, 1.5 kg, 1.9 kg, respectively (Table 4.7).

Predicted FFM correlated highly with measured FFM in each ethnic group with r of

0.92‐0.94 (Table 4.8). Figure 4.3‐4.7 shows the difference between measured and

predicted FFM plotted against the mean of the predicted and measured FFM for

each ethnic group. The graphs identified 3.8‐10.8% participants whose residuals

exceeded the 95% CI. The prediction equation for estimation of FFM tended to

overestimate FFM at lower level of FFM while underestimate FFM at higher level of

FFM for Filipinos and Thais. Moreover, the predicted FFM from ethnic‐specific

prediction equation was also compared with the measured FFM at each ethnic

group and no significant differences were found. The bias and pure errors were

similar to those using equation derived from total validation sample. For example,

the bias calculated from equation derived from total validation sample and

ethnic‐specific equation was ‐0.5 kg and ‐0.4 kg, respectively, and the PE was 1.8 kg

and 1.9 kg, respectively, for Chinese children. The concordance correlation

coefficients for measured FFM against predicted FFM were also similar to those

using equation derived from total validation sample in each ethnic group (Table 4.8).

Table 4.7. Comparison of measured FFM with predicted FFM by ethnicity

Total validation sample equation Ethnic‐specific equation FFMm (kg)

FFMp (kg) Bias (kg) PE (kg) FFMp(kg) Bias (kg) PE (kg)

Chinese 25.5±4.9 26.0±4.9 ‐0.5±1.8 1.8±2.5 25.9±5.0 ‐0.4±1.8 1.9±2.5

Lebanese 24.2±4.0 24.5±4.1 ‐0.3±1.4 1.4±2.0 24.5±4.1 ‐0.2±1.7 1.7±2.5

Malays 23.6±6.0 23.8±5.8 ‐0.2±2.1 2.1±2.7 23.8±5.5 ‐0.2±2.1 2.0±2.8

Filipinos 24.0±3.9 24.1±3.3 ‐0.1±1.5 1.5±1.9 24.0±3.7 0.0±1.5 1.5±1.9

Thais 23.8±5.5 23.5±4.9 0.4±1.8 1.9±3.1 23.2±4.9 0.7±2.0 2.1±3.7

FFMm: measured FFM by D2O; FFMp: predicted FFM by equation. No significant differences between measured and predicted FFM at each ethnic groups by paired t‐test (P >0.05)


188

Table 4.8. Pearson correlation coefficients between measured and predicted FFM and between bias and mean FFM by ethnicity

Total validation sample equation Ethnic‐specific equation

FFMm vs FFMp FFMd vs FFMmean FFMm vs FFMp FFMd vs FFMmean

Coefficient P value Coefficient P value Coefficient P value Coefficient P value

Chinese 0.94 <0.0001 0.004 0.963 0.93 <0.0001 ‐0.067 0.476

Lebanese 0.94 <0.0001 ‐0.071 0.729 0.92 <0.0001 ‐0.091 0.657

Malays 0.94 <0.0001 0.126 0.319 0.94 <0.0001 0.247 0.047

Filipinos 0.92 <0.0001 0.382 0.020 0.92 <0.0001 0.090 0.587

Thais 0.95 <0.0001 0.333 0.004 0.93 <0.0001 0.280 0.015

FFMm: measured FFM by D2O; FFMp: predicted FFM by equation; FFMd: measured FFM minus predicted FFM. FFMmean: mean of measured FFM and predicted FFM.


189

Figure 4.3. Difference in FFM measured by D2O and estimated by BIA in Chinese children.

Figure 4.4. Difference in FFM measured by D2O and estimated by BIA in Lebanese children.

Mean ‐ 2SD

Mean

Mean + 2SD

Mean ‐ 2SD

Mean + 2SD

Mean


190

Figure 4.5. Difference in FFM measured by D2O and estimated by BIA in Malay children.

Figure 4.6. Difference in FFM measured by D2O and estimated by BIA in Filipino children.

Mean ‐ 2SD

Mean

Mean + 2SD

Mean + 2SD

Mean

Mean ‐ 2SD


191

Figure 4.7. Difference in FFM measured by D2O and estimated by BIA in Thai children.

The predicted FFM using prediction equations derived from total validation sample

was compared with the measured FFM at each BMI category to ensure this

prediction equation was valid for each BMI category. No difference between

measured and predicted FFM in each BMI category was found. The PE in lower BMI z

score was slightly lower than that in higher BMI z score, indicating slight higher

accuracy of the equations for lean children (Table 4.9). Figures 4.8‐4.10 show the

difference between measured and predicted FFM plotted against the mean of the

predicted and measured FFM for each BMI category. The graphs identified 6.0‐9.5%

participants whose residuals exceeded the 95% confidence limits. The prediction

equation for estimation of FFM tended to overestimate FFM at lower level of FFM

while underestimate FFM at higher level of FFM for higher BMI categories.

Moreover, the predicted FFM from BMI‐specific prediction equation was also

compared with the measured FFM at each BMI category and no significant

differences were found. The bias and PEs were similar to those using equation

derived from total validation sample. For example, the bias calculated from equation

derived from total validation sample and ethnic‐specific equation was ‐0.1 kg and

0.0 kg, respectively, and the PE was 1.4 kg and 1.4 kg, respectively, for participants

Mean ‐ 2SD

Mean+ 2SD

Mean


192

with BMI z score less than 1 SD. The concordance correlation coefficients for

measured FFM against predicted FFM were high in each BMI category (r = 0.88‐0.91)

and similar to those between measured FFM and predicted FFM using equation

derived from total validation sample(r = 0.86‐0.89) (Table 4.10).

Table 4.9. Comparison of measured FFM with predicted FFM by BMI category

Total sample equation BMI‐specific equation FFMm

FFMp (kg) Bias (kg) PE (kg) FFMp (kg) Bias (kg) PE (kg)

<1SD (n=167) 21.2±3.0 21.3±2.9 ‐0.1±1.4 1.4±1.9 21.2±2.9 0.0±1.5 1.4±1.9

1‐2SD (n=56) 25.7±4.0 25.9±3.4 ‐0.2±1.8 1.8±2.5 26.0±3.1 ‐0.3±1.9 1.9±2.6

≥2SD (n=95) 29.3±4.7 29.6±4.1 ‐0.3±2.3 2.4±3.2 29.4±4.4 ‐0.1±2.5 2.5±3.6

FFMm: measured FFM by D2O; FFMp: predicted FFM by equation. No significant differences between measured and predicted FFM at each BMI category by paired t‐test (P >0.05)


193

Table 4.10. Pearson correlation coefficients between measured and predicted FFM by ethnicity and BMI category

Total validation sample equation BMI‐specific equation

FFMm vs FFMp FFMd vs FFMmean FFMm vs FFMp FFMd vs FFMmean


<1SD 0.88 <0.0001 0.125 0.107 0.86 <0.0001 0.097 0.212

1‐2SD 0.90 <0.0001 0.333 0.012 0.87 <0.0001 0.461 0.000

≥2SD 0.86 <0.0001 0.256 0.012 0.82 <0.0001 0.096 0.355

FFMm: measured FFM by D2O; FFMp: predicted FFM by equation; FFMd: measured FFM minus predicted FFM. FFMmean: mean of measured FFM and predicted FFM.


194

Figure 4.8. Difference in FFM measured by D2O and estimated by BIA in children with BMI z score <1 SD.

Figure 4.9. Difference in FFM measured by D2O and estimated by BIA in children with BMI z score 1‐2 SD.

Mean ‐ 2SD

Mean

Mean + 2SD

Mean + 2SD

Mean

Mean ‐ 2SD


195

Figure 4.10. Difference in FFM measured by D2O and estimated by BIA in children with BMI z score ≥2 SD.

4.3.2.2 Cross‐validation of equations for TBW

No significant difference between measured and predicted TBW for the whole

cross‐validation sample was found (Bias = ‐0.1±1.4 kg, PE = 1.4±2.0 kg). Measured

TBW correlated highly with predicted TBW (r = 0.94, P<0.0001) (Figure 4.11). The

difference between measured and predicted TBW plotted against the mean of the

predicted and measured TBW is shown in Figure 4.12. The graph identified

approximately 19 participants (6.0%) whose residuals exceeded the 95% confidence

limits: 6 over‐predicted and 13 under‐predicted FFM. Significant correlation

between difference of measured and predicted TBW and mean of measured and

predicted TBW (r = 0.115, P = 0.041) was observed. Similar to prediction equation

for estimation of FFM, the prediction equation for estimation of TBW tended to

overestimate TBW at lower level of TBW while underestimate TBW at higher level of

TBW.

Mean ‐ 2SD

Mean

Mean + 2SD


196

Figure 4.11. Scatter plot of measured TBW by D2O against predicted TBW by BIA in Asian children

Figure 4.12. Difference in TBW measured by D2O and estimated by BIA in Asian children.

Mean + 2SD

Mean ‐ 2SD

Mean


197

The predicted TBW from the prediction equation derived from the total validation

sample was compared with the measured TBW at each ethnic group to ensure this

equation is valid for each ethnic group. Similar to the results of TBW, no difference

between measured and predicted TBW in each ethnic group was found (Table 4.11).

The difference between measured and predicted TBW plotted against the mean of

the predicted and measured TBW for each ethnic group is shown in Figure 4.13‐4.17.

The graph identified 3.8‐8.1% participants whose residuals exceeded the 95%

confidence limits. The prediction equation for estimation of TBW tended to

overestimate TBW at lower level of TBW while underestimate TBW at higher level of

TBW for both Filipinos and Thais. Moreover, there was no significant difference

between the predicted TBW from ethnic‐specific prediction equation and the

measured TBW at each ethnic group (Table 4.11). The bias and PE using equation

derived from total validation sample were similar to those using ethnic‐specific

equations. The concordance correlation coefficients for measured TBW against

predicted FFM using equation derived from total validation sample in each ethnic

group were high (r = 0.92‐0.94) and similar to those between measured TBW and

predicted FFM using ethnic‐specific equation (r = 0.91‐0.94) (Table 4.12).

Table 4.11. Comparison of measured TBW with predicted TBW by ethnicity

Total validation sample equation

Ethnic‐specific equation

TBWm

(kg) TBWp(kg) Bias(kg) PE(kg) TBWp(kg) Bias (kg) PE (kg)

Chinese 19.5±3.7 19.9±3.7 ‐0.4±1.4 1.4±1.9 19.8±3.8 ‐0.3±1.4 1.4±1.9

Lebanese 18.6±3.0 18.8±3.1 ‐0.2±1.1 1.1±1.6 18.7±3.1 ‐0.1±1.3 1.3±1.9

Malays 18.1±4.6 18.2±4.3 ‐0.1±1.6 1.5±2.1 18.3±4.2 ‐0.1±1.6 1.6±2.2

Filipinos 18.4±3.0 18.5±2.5 ‐0.1±1.2 1.2±1.5 18.4±2.9 ‐0.0±1.2 1.2±1.5

Thais 18.3±4.2 18.0±3.8 0.3±1.4 1.4±2.3 17.9±3.7 0.5±1.5 1.6±2.9

TBWm: measured TBW by D2O; TBWp: predicted TBW by equation. No significant differences between measured and predicted TBW at each ethnic groups by paired t‐test (P >0.05)


198

Table 4.12. Pearson correlation coefficients between measured and predicted TBW and between bias and mean TBW by ethnicity

Total validation sample equation Ethnic‐specific equation

TBWm vs TBWp TBWd vs TBWmean TBWm vs TBWp TBWd vs TBWmean


Chinese 0.93 <0.0001 0.015 0.871 0.93 <0.0001 ‐0.051 0.589

Lebanese 0.93 <0.0001 ‐0.076 0.711 0.91 <0.0001 ‐0.080 0.699

Malays 0.94 <0.0001 0.127 0.314 0.94 <0.0001 0.231 0.063

Filipinos 0.92 <0.0001 0.368 0.025 0.92 <0.0001 0.077 0.651

Thais 0.94 <0.0001 0.339 0.003 0.93 <0.0001 0.347 0.002

TBWm: measured TBW by D2O; TBWp: predicted TBW by equation; TBWd: measured TBW minus predicted TBW; TBEmean: mean of measured TBW and predicted TBW.


199

Figure 4.13. Difference in TBW measured by D2O and estimated by BIA in Chinese children.

Figure 4.14. Difference in TBW measured by D2O and estimated by BIA in Lebanese children.

Mean ‐ 2SD

Mean

Mean + 2SD

Mean ‐ 2SD

Mean

Mean + 2SD


200

Figure 4.15. Difference in TBW measured by D2O and estimated by BIA in Malay children.

Figure 4.16. Difference in TBW measured by D2O and estimated by BIA in Filipino children.

Mean ‐ 2SD

Mean

Mean +2SD

Mean ‐ 2SD

Mean

Mean + 2SD


201

Figure 4.17. Difference in TBW measured by D2O and estimated by BIA in Thai children.

The predicted TBW using prediction equations derived from total validation sample

was also compared with the measured TBW at each BMI category to ensure this

prediction equation are valid for each BMI category. No difference between

measured and predicted TBW in each BMI category was found. The PE in lower BMI

z score was slightly lower than that in higher BMI z score (1.1 kg, 1.4 kg, 1.8 kg for

BMI z score <1 SD, 1‐2 SD, ≥2 SD, respectively), indicating slightly higher accuracy of

the equation for lean children. The difference between measured and predicted

TBW plotted against the mean of the predicted and measured TBW for each ethnic

group is shown in Figure 4.18‐4.20. The graph identified 6.0‐9.5% participants

whose residuals exceeded the 95% confidence limits. The prediction equation for

estimation of TBW tended to overestimate TBW at lower level of TBW while

underestimate TBW at higher level of TBW for higher BMI categories. Moreover,

there were also no significant differences between the predicted values from

BMI‐specific prediction equation and the measured values at each BMI category.

The bias and PEs using equation derived from total validation sample were similar to

those using BMI‐specific equations (Table 4.13). The concordance correlation

Mean ‐ 2SD

Mean

Mean + 2SD


202

coefficients for measured TBW against predicted TBW using equation derived from

total validation sample in each BMI category were high (r = 0.87‐0.89) and similar to

those between measured TBW and predicted TBW using BMI‐specific equation (r =

0.87‐0.91) (Table 4.14).

Table 4.13. Comparison of measured TBW with predicted TBW by BMI category

Total validation sample equation BMI‐specific equation TBWm

TBWp (kg) Bias (kg) PE (kg) TBWp (kg) Bias (kg) PE (kg)

< 1SD 16.3±2.3 16.3±2.1 0.0±1.1 1.1±1.5 16.3±2.2 ‐0.0±1.1 1.1±1.4

1‐2SD 19.7±3.0 19.9±2.5 ‐0.2±1.4 1.4±2.0 20.0±2.3 ‐0.5±1.5 1.5±2.0

≥2SD 22.5±3.5 22.7±3.1 ‐0.2±1.8 1.8±2.4 22.5±3.4 0.1±1.9 1.9±2.7

TBWm: measured TBW by D2O; TBWp: predicted TBW by equation. No significant differences between measured and predicted TBW at each BMI category by paired t‐test (P >0.05).


203

Table 4.14. Pearson correlation coefficients between measured and predicted TBW by BMI category

Total validation sample equation BMI‐specific equation

TBWm and TBWp TBWd and TBWmean TBWm and TBWp TBWd and TBWmean


< 1SD 0.88 <0.0001 0.133 0.086 0.88 <0.0001 0.105 0.178

1‐2SD 0.89 <0.0001 0.360 0.007 0.91 <0.0001 0.485 0.000

≥2SD 0.87 <0.0001 0.264 0.010 0.87 <0.0001 0.108 0.286

TBWm: measured TBW by D2O; TBWp: predicted TBW by equation; TBWd: measured TBW‐ predicted TBW; TBWmean: (measured TBW + predicted TBW)/2.


204

Figure 4.18. Difference in TBW measured by D2O and estimated by BIA in children with BMI z score <1 SD.

Figure 4.19. Difference in TBW measured by D2O and estimated by BIA in children with BMI z score 1‐2 SD.

Mean + 2SD

Mean

Mean ‐ 2SD

Mean + 2SD

Mean

Mean ‐ 2SD


205

Figure 4.20. Difference in TBW measured by D2O and estimated by BIA in children with BMI z score ≥2 SD.

4.3.3 Cross‐validation of external BIA prediction equations

The cross‐validation of selected three FFM equations and 2 TBW equations is shown

in Table 4.15. The predicted FFM was significant higher than measured FFM for the

three equations developed by Deurenberg et al. (1991), Rush et al. (2003) and

Lohman (1992), respectively and significant correlation between bias and mean FFM

was found for the two former equations with r of ‐0.503 and ‐0.311, respectively.

The TBW equation developed by Davies & Preece (1988) under‐predicted TBW for

our Asian children with a bias of 2.3 kg and 1.7 kg, respectively and significant

correlation between bias and mean TBW was found for the two equations with r of

‐0.486 and 0.343, respectively. The equation developed by Kushner et al. (1992) for

the estimation of TBW performed well in Asian children with a bias of 0.2 kg and

without systematic error.

Mean + 2SD

Mean ‐ 2SD

Mean


206

Table 4.15. Bias of external BIA equations derived from Caucasian children for the assessment of FFM and TBW in Asian children

Age (yr)

Equation Bias (kg)

Limits of agreement (kg)

r

FFM

Deurenberg et al (1991) 7‐15 FFM = ‐6.48 + 0.406×RI + 0.360×Weight + 5.58×Height (m) + 0.56×Gender ‐1.3±2.3* ‐5.9 to 3.3 ‐0.503#

Lohman et al (1992) 8‐15 FFM = 0.62×RI + 0.21×Weight + 0.10×Xc + 4.2 ‐10.6±2.0* ‐14.6 to ‐6.6 ‐0.061

Rush et al (2003) 5‐14 FFM = 1.166 + 0.622×RI + 0.234×Weight ‐2.4±2.1* ‐6.6 to 1.8 ‐0.311#

TBW

Kushner et al (1992) 6‐10 TBW = 0.593×RI + 0.065×Weight + 0.04 ‐0.2±1.6 ‐3.8 to 3.0 ‐0.061

Davies et al (1988) 5‐17 TBW = 0.5 + 0.6×RI 2.3±5.4* ‐8.5 to 13.1 ‐0.486#

* Significant difference between measured values and predicted values from BIA equations by paired t‐test (P <0.001) # Significant correlation between bias (measured‐predicted) and mean of measured value and predicted value by Pearson’s correlation (P <0.001).


207

4.4 DISCUSSION

The main purpose of the current study was to develop a BIA prediction equation for

the estimation of body composition in Asian children aged 8 to 10 y. To our

knowledge, this is the first study to develop a BIA equation across Western, Eastern

and South‐East Asian groups. A FFM and TBW equation was developed for our

participants, respectively. FFM and TBW were used as our outcome variables rather

than FM or %BF because of the functional relationship between resistance and the

hydrated lean tissue of the body. The developed FFM equation in the current study

was with R2 of 0.883 and RMSE of 1.7 kg and the TBW equation was with R2 of 0.88

and RMSE of 1.3 kg. These results compare favourably with those reported from

other studies using the deuterium dilution technique as the criterion method.

Previous BIA equations for estimating TBW in children and adolescents showed

multiple regression coefficients (R2) ranging from 0.65 to 0.99 and RMSE ranging

from 0.41 to 3.81 kg for TBW as reviewed by Nielsen et al. (2007). The R2 ranged

from 0.87 to 0.96 and RMSE ranging from 2.4 to 2.7 kg for FFM in previous BIA

equations for FFM (Rush et al., 2003; Wickramasinghe et al., 2008). Our prediction

equation showed ideal predictive accuracy when Lohman’s classification system for

prediction errors (1992) was used (Table 2.10).

Our study demonstrates the ethnic effect on the relationship between RI and TBW

and FFM. The inclusion of ethnicity into the equation improved the estimation of

FFM and TBW in Thai children. Thai children had less ‐0.4 kg for FFM and ‐0.3 kg for

TBW than Chinese, Lebanese, Malays and Filipinos at a given RI, height, weight, age

and sex. Some studies have shown the advantage of including race in BIA prediction

equations (Going et al., 2006; Haroun et al., 2010; Sluyter et al., 2010). Ignoring the

race effect would result in under‐estimation of FFM by 1.6 kg in black girls (Going et

al., 2006), 1.8 kg in Maori and Pacific boys and 1.6 kg in Pacific girls (Sluyter et al.,

2010) and 1.5 kg in South Asian girls and 1.3 kg in South Asian boys (Haroun et al.,

2010). Several factors can explain the ethnic difference in the relationship between

BIA impedance or resistance and FFM or TBW. Firstly, differences in relative limb

length might have an effect on the estimation of FFM and TBW by BIA. It has been


208

shown that the contribution of the limbs to total body impedance is not

proportional to the amount of body water in these body segments (Baumgartner,

Chumlea & Roche, 1989; Fuller & Elia 1989). The total body impedance is largely

determined by limb impedance and the longer the limbs are, the higher the total

impedance (Bracco et al., 1996; Deurenberg, Deurenberg‐Yap & Schouten, 2002).

Those who have a higher relative leg/arm length will have a higher

resistance/impedance, and thus have a lower RI at a given FFM. This will result in

underestimation of FFM and TBW (Snijder, Kuyf & Deurenberg, 1999). The ethnic

difference in relative length has been shown by some studies (Deurenberg et al.,

1999; Gallagher et al., 1996; Gurrici et al., 1999; Norgan, 1994a). Blacks have a

higher relative leg length than whites, who in turn are higher than Asian. Data on

differences in relative leg length among Asian children from different origins are

limited, however, Deurenberg et al. (2002) showed that Indian adults had a higher

relative leg length than Chinese and Malays (0.47, 0.45, 0.46, respectively, for

females; 0.48, 0.47 and 0.46 for males, respectively). Gurrici et al. (1999) also found

that Chinese Indonesians have a higher relative sitting height than Malay

Indonesians. Furthermore, the ethnic difference in TBW distribution can also

influence the estimation of FFM and TBW by single frequency BIA. Single frequency

BIA at 50 kHz primarily reflects the ECW space (Ellis et al., 1999). Those who have

higher relative extracellular water (lower ratio of intra‐ to extra‐cellular water) will

have a lower resistance when measured by single frequency BIA, thus the term RI

will be higher for a given FFM/TBW and body weight. This will result in

overestimation of TBW/FFM. Deurenberg et al. (2002) found Indian females had

significant lower ECW/TBW than Chinese and Malays (0.44, 0.46, 0.46, respectively)

and no significant difference was found between Chinese and Malays. After

controlling for the difference in body water distribution and body build, the bias of

estimation FFM disappears (Deurenberg et al., 2002). Hence, the

population‐specific BIA prediction equations were commonly recommended when

applying BIA to predict body composition due to these ethnic differences in body

composition and body build (Kyle et al., 2004; Lohman et al., 2000; Nielsen et al.,

2007). Our results confirmed the necessity for population‐specific BIA equations.


209

We cross‐validated the external equations derived from Caucasian children in our

participants and most of these equations overestimated FFM for Asian children

(Table 4.15). Lohman et al. (2000) also showed that published BIA equations

developed in white children under‐predicted %BF in American Indian children.

Weight was included into both FFM and TBW prediction equations. As determined

by standardized regression coefficients, weight contributed more than RI to the

prediction of FFM and TBW, which is in line with another study (Kriemler et al.,

2009). The reason for the inclusion of body weight is that water is the most

abundant compartment in the body, making up approximately 65% of body weight

for 8 to 10‐year‐old children (Fomon et al., 1982).

Apart from RI and weight, sex and age was also a predictor of TBW and FFM in our

study. This can be explained by differences in body water distribution between

males and females, and between ages as well. In general females have a higher ratio

of extracellular water to total body water (Fomon et al., 1982), thus the body

resistance measured by single‐frequency BIA at 50 kHz will be lower and the term RI

will be higher for a given FFM and body weight in females than males. The

distribution of body water changes toward a lower relative amount of extracellular

water with the increase of age during childhood (Fomon et al., 1982). Hence, older

children will have a higher resistance (lower RI) than younger children at the same

FFM level (Deurenberg et al., 1990).

The accuracy of an equation usually reduces when it is applied to other samples.

Therefore, it is necessary to cross‐validate to confirm the validity of the prediction

equation. Our cross‐validation results indicated that the predicted values correlated

with the measured values very highly (r = 0.94 for FFM and TBW) and the pure error

was low (1.8 kg for FFM and 1.4 kg for TBW) with no significant difference between

predicted and measured values. Moreover, the limits of agreement assessed by

Bland‐Altman approach were very small and comparable to those from previous

studies (Deurenberg et al., 1991; McClanahan et al., 2009; Sluyter et al, 2010).

However, it must be noted that application of Bland‐Altman analysis showed that


210

there may be a potential systematic bias in applying the equation. More specifically,

there is a trend for over‐prediction in participants of low FFM and under‐prediction

in participants of high FFM. Previous studies have also found that the bias was

dependent on the level of body fat (Deurenberg et al., 1991; McClanahan et al.,

2009). The lower accuracy of BIA in obese subjects might be explained by the body

water distribution changing with the increase of body fat. The obese subjects have a

higher relative amount of extracellular water than the lean (Ritz et al., 2008;

Sartorio et al., 2005; Wang & Pierson Jr, 1976). BIA impedance will therefore be

lower in the obese at a given FFM and body weight and the general BIA equation

will overestimate the FFM and TBW in the obese (Deurenberg et al., 1991).

Moreover, Baumgartner et al. (1998) found that adipose tissue could affect

measured resistance when the volume of adipose tissue is greater than muscle

volume which resulted in a slight overestimation of FFM by 3 kg when a BIA

equation derived for non‐obese female subjects was applied. So we assessed the

accuracy of the FFM and TBW general equation derived from the total validation

group for each BMI category. The predicted values did not differ from the measured

values at each BMI category although the predicted values were slightly higher than

measured values and the standard deviation of bias was larger in higher BMI

category (difference between measured and predicted FFM was ‐0.1±1.5 kg,

‐0.3±1.8 kg and ‐0.3±2.4 kg for BMI z score <1 SD, 1‐2 SD, ≥2 SD, respectively;

difference between measured and predicted TBW was 0.0±1.1 kg, ‐0.1±1.4 kg and

‐0.2±1.8 kg for BMI z score <1 SD, 1‐2 SD and ≥2 SD, respectively). Meanwhile, high

correlation coefficients between predicted and measured values, and low pure

errors were also found in each BMI category. Furthermore, BMI‐specific equations

were also developed in our study. However, the bias between measured and

predicted values from BMI‐specific equations showed similar SD of bias, the

correlation coefficients between predicted and measured values and pure errors.

Therefore, the general equation gave the same and valid result for children across a

wide BMI range. Moreover, the accuracy of the FFM and TBW equation derived from

the total validation groups was assessed for each ethnic group in the current study.

No significant difference between predicted values and measured values was found


211

in each ethnic group. We also observed high correlation coefficients between

predicted and measured values and low PEs in each ethnic group, which are

comparable with those from cross‐validation of the derived ethnic‐specific

equations. These results indicated that the general equation derived from the total

validation groups performed as well as the ethnic‐specific equations and can be

used as universal equation for the five ethnic groups.

It is essential to mention that technical errors might contribute to the bias. Errors

associated with BIA measurement include instrumentation, body position, electrode

placement, subject factors (eating, being dehydrated, exercise) and environmental

temperature (Heyward, 1998; Hills et al., 2001; Houtkooper et al., 1996). Errors for

the isotope dilution technique include variations in the physiological fluid measured,

equilibration time of the isotope, water changes during equilibration, correction for

isotopic dilution space, and the method for measuring the isotopic enrichment

following equilibration (Lohman, 1992). Therefore in the current study, the same

instruments were used and standard operating procedures followed by research

groups in each country.

There are several limitations to our study. Firstly, our participants were purposively

selected to obtain a wide range of BMI and were not a representative sample for

each ethnicity. However, for a validation study, it is better to have a wide range in

body fat in the study sample. Secondly, the deuterium dilution technique was used

as the criterion method in our study and FFM was derived on the basis of a

two‐component model considering the ethnic difference in hydration of FFM.

However, the deuterium dilution technique is the gold standard for estimation of

TBW and is also the only option for the remote measurement of body composition

with high precision. Finally, the same constants for hydration were used across the

five groups in our study although previous studies have shown an ethnic difference

in body water distribution among adults. However, no data is available on hydration

constants for Asian children.

In conclusion, the BIA prediction equations for FFM and TBW developed in our study

and are valid for use in Chinese, Lebanese, Malay, Filipino and Thai children aged


212

8‐10 y with a wide range of BMI from 12.2‐34.9 kg/m2. Ethnicity entered the

equations and our equations performed better than those developed in white

children, indicating that ethnicity is very important in the generation of BIA

prediction equations.


213

CHAPTER 5 OBESITY AND THE METABOLIC SYNDROME IN CHILDREN

5.1 THE ASSOCIATION OF OBESITY WITH THE METABOLIC SYNDROME IN CHINESE CHILDREN

5.1.1 Introduction

The metabolic syndrome describes the clustering of central obesity, dyslipidaemia

(raised triglycerides and/or low‐ or high‐density lipoprotein), hyperinsulinaemia,

impaired glucose tolerance and elevated blood pressure (Alberti et al., 2005). People

with metabolic syndrome are two to three times as likely to have a heart attack or

stroke and five times as likely to develop type 2 diabetes compared with people

without the syndrome, therefore it is a clear indicator of adult morbidity and

all‐cause mortality (Alberti et al., 2005; Haffner et al., 1992; Isomaa et al., 2001;

Trevisan et al., 1998). Paediatric metabolic syndrome can also increase

cardiovascular risk (Ronnemaa et al., 1991) and can track from childhood to

adulthood (Duncan et al., 2004).

Some studies have indicated that obese children have a higher risk for paediatric

metabolic syndrome compared with children who are normal‐weight (Cook et al.,

2003; de Ferranti et al., 2004; Reinehr et al., 2007; Weiss et al., 2004; Yoshinaga,

Tanaka, & Snimago, 2005). With the rapid increase of obesity, more and more

children are at risk of developing the metabolic syndrome. In China, the prevalence

of paediatric overweight and obesity increased three times from 1982 to 2002 and

about 12 million Chinese children were overweight and obese in 2002 (Li, Schouten

et al., 2008). However, a limited number of studies have explored the prevalence of

the metabolic syndrome in Chinese overweight or obese children and the

association of obesity with the metabolic syndrome. Moreover, there are ethnic

differences in the susceptibility to metabolic risk factors such as hypertension and

insulin resistance (Arslanian et al., 1997; Huxley et al., 2008; Ke et al., 2009;

Whincup et al., 2002, 2005; Zhu et al., 2005). Therefore, the specific purposes of this

study were to determine the prevalence of the metabolic syndrome among Chinese

overweight and obese children and to explore the association of obesity with the


214

metabolic syndrome.

5.1.2 Methodology


A total of 1408 elementary school children (608 boys and 800 girls) aged 6‐12 years

were recruited from four cities in China, including Liaoyang in the Northeast, Beijing

and Tianjin in the North, and Guangzhou in the South in 2007. Of these, 867

children were normal‐weight, 333 children were overweight and 508 children were

obese. Written consent was obtained from both children and their parents.


Height was measured using a stadiometer to the nearest 0.1 cm in bare feet.

Weight was measured to the nearest 0.1 kg with a balance‐beam scale with

participants wearing lightweight clothing. The BMI (kg/m2) was calculated as weight

(kg) divided by the square of height (m). The criterion developed by the Group of

China Obesity Task Force (2004) was used to define overweight and obesity based

on age‐ and gender‐specific BMI. WC was measured to the nearest 0.1 cm at the

mid‐point between the lower costal border and the top of the iliac crest with the

measurement taken at the end of a normal expiration.

5.1.2.3 Metabolic variables measurements

Blood pressure was measured on the study morning using a random‐zero

sphygmomanometer after the participant rested for 5‐min in a seated position. Two

resting blood pressure measurements were taken to the nearest 2 mmHg, and the

first and fifth Korotkoff sounds were used to represent the SBP and DBP,

respectively.

A venous blood sample of 5 mL was collected from each participant in the morning

after an overnight fast. Serum glucose concentration was measured enzymatically

using an automated analyzer (Cobas Mira; Roche Diagnostic systems, Indianapolis,


215

IN). Serum TG, total cholesterol (TC), and HDL‐C were determined enzymatically with

a bichromatic analyzer (Abbott Diagnostics Spectrum CCX, Abbott Laboratory, North

Chicago, IL).

5.1.2.4 Definition of paediatric metabolic syndrome

To enable comparisons with other studies, three definitions were used to diagnose

the presence of the metabolic syndrome.

Definition 1. The IDF criteria for the metabolic syndrome in children aged 10‐16

years. The metabolic syndrome is diagnosed as the presence of abdominal obesity

and two or more of the following clinical features:

(1) Central obesity (WC) ≥90th percentile for age and gender;

(2) TG ≥1.7 mmol/L;

(3) HDL‐C <1.03 mmol/L;

(4) Blood pressure ≥130 mm Hg systolic or ≥85 mm Hg diastolic;

(5) Fasting glucose ≥5.6 mmol/L.

Definition 2. NCEP definition for adults modified by Cook et al. (2003). Metabolic

syndrome is diagnosed as three or more of the following variables:

(1) Central obesity (WC) ≥90th percentile for age and gender;

(2) TG ≥1.24 mmol/L;

(3) HDL‐C <1.03 mmol/L;

(4) Blood pressure ≥90th percentile for age, gender, and height;


Defintion 3. NCEP definition for adults modified by de Ferranti et al. (2004).

Metabolic syndrome is diagnosed as three or more of the following variables:

(1) Central obesity (WC) >75th percentile for age and gender;

(2) TG ≥1.1 mmol/L;

(3) HDL‐C <1.3 mmol/L;

(4) Blood pressure >90th percentile for age gender, and height;



216


Mean±SD was used to describe the distribution of continuous variables. ANOVA was

employed to compare the difference in each variable among normal‐weight,

overweight and obese children. Odds ratios (OR) and 95% CI were calculated to

explore the risk of the presence of metabolic syndrome and abnormalities among

overweight and obese children compared to normal‐weight children.

5.1.3 Results

Table 5.1 shows the differences in anthropometric variables, blood pressure and

blood biochemical parameters among children in each category. There was no

significant difference in age among the three groups of boys and girls. Overweight

and obese children had higher values in all variables than children who were normal

weight.


217

Table 5.1. Characteristics of participants

Normal‐weight Overweight Obese P value

Boys (n=608)

n 390 199 319

Age (years) 9.3±1.0 9.4±1.1 9.3±1.0 0.4931

Height (cm) 138.3±8.1 141.8±8.8 145.3±8.2 ＜0.0001

Weight (kg) 31.7±5.9 41.6±6.9 53.7±11.1 ＜0.0001

BMI (kg/m2) 16.4±1.7 20.5±1.3 25.2±3.2 ＜0.0001

WC (cm) 58.7±6.0 70.3±6.0 81.5±8.6 ＜0.0001

TG (mmol/L) 0.79±0.33 1.08±0.86 1.37±0.79 ＜0.0001

TC (mmol/L) 3.99±0.64 4.21±0.70 4.44±0.78 ＜0.0001

HDL‐C (mmol/L) 1.66±0.34 1.54±0.35 1.41±0.31 ＜0.0001

LDL‐C (mmol/L) 2.24±0.68 2.60±0.72 2.86±0.79 ＜0.0001

SBP (mmHg) 102±9 107±10 111±12 ＜0.0001

DBP (mmHg) 65±8 68±8 71±8 ＜0.0001

Glucose (mmol/L) 4.70±0.42 4.80±0.43 4.76±0.42 0.0269

Girls (n=800)

n 477 134 189

Age (years) 9.2±1.0 9.2±1.1 9.1±1.0 0.3467

Height (cm) 138.0±8.4 140.8±8.8 142.9±8.5 ＜0.0001

Weight (kg) 31.2±6.2 40.6±6.9 49.2±9.1 ＜0.0001

BMI (kg/m2) 16.2±1.8 20.3±1.3 23.9±2.5 ＜0.0001

WC (cm) 57.0±5.6 67.9±5.6 75.8±7.2 ＜0.0001

TG (mmol/L) 0.95±0.42 1.07±0.47 1.31±0.73 ＜0.0001

TC (mmol/L) 4.06±0.69 4.18±0.80 4.38±0.80 ＜0.0001

HDL‐C (mmol/L) 1.56±0.39 1.44±0.33 1.39±0.31 ＜0.0001

LDL‐C (mmol/L) 2.35±0.71 2.66±0.80 2.85±0.77 ＜0.0001

SBP (mmHg) 100±10 105±10 109±9 ＜0.0001

DBP (mmHg) 65±8 68±8 69±9 ＜0.0001

Glucose (mmol/L) 4.62±0.39 4.74±0.37 4.66±0.41 0.0094

Table 5.2 identifies the prevalence of the metabolic syndrome and abnormalities

among children in each of the three categories. The prevalence of the metabolic


218

syndrome varied with different definitions, was highest using the de Ferranti

definition (5.4%, 24.6% and 42.0%, respectively for normal‐weight, overweight and

obese children), followed by the Cook definition (1.5%, 8.1%, and 25.1%,

respectively), and the IDF definition (0.5%, 1.8% and 8.3%, respectively). In summary,

overweight children had a higher risk of developing the metabolic syndrome

compared to normal‐weight children (OR = 3.958, 95% CI: 1.110‐14.118 using the

IDF definition; OR = 6.866, 95% CI: 3.365‐14.010 using the Cook definition; and OR =

5.700, 95% CI: 3.877‐8.380 for the de Ferranti definition). However, obese children

had a much higher risk of developing the metabolic syndrome (OR = 19.487, 95% CI:

6.945‐54.681 using the IDF definition; OR = 26.007, 95% CI: 13.883‐48.722 using the

Cook definition; and OR = 12.640, 95% CI: 8.971‐17.808 using the de Ferranti

definition) compared with normal‐weight children. Moreover, overweight and obese

children also had a higher risk of developing abdominal obesity, high TG, low HDL‐C,

elevated blood pressure and hyperglycemia, than normal‐weight children,

regardless of the definition used.


219

Table 5.2. Prevalence of the metabolic syndrome and abnormalities among normal‐weight, overweight and obese children.

Normal‐weight (n=867) Overweight (n=333) Obese (n=507)

% OR % OR (95% CI) % OR (95% CI)

IDF definition

Metabolic syndrome 0.5 1.000 1.8 3.958 (1.110‐14.118) 8.3 19.487 (6.945‐54.681)

Central obesity 8.0 1.000 69.1 25.825 (18.412‐36.224) 96.9 355.607 (204.080‐619.639)

High TG 4.5 1.000 10.2 2.414 (1.496‐3.896) 23.7 6.583 (4.498‐9.635)

Low HDL‐C 2.9 1.000 4.2 1.478 (0.759‐2.879) 9.9 3.686 (2.250‐6.036)

Elevated BP 0.8 1.000 1.8 2.254 (0.752‐6.757) 7.5 9.954 (4.411‐22.464)

Hyperglycaemia 1.5 1.000 3.6 2.456 (1.109‐5.438) 3.4 2.279 (1.098‐4.732)

Cook definition

Metabolic syndrome 1.3 1.000 8.1 6.866 (3.365‐14.010) 25.1 26.007(13.883‐48.722)

Central obesity 8.0 1.000 69.1 25.825 (18.412‐36.224) 96.9 355.607 (204.080‐619.639)

High TG 14.3 1.000 27.9 2.322 (1.710‐3.152) 46.6 5.218 (4.031‐6.754)

Low HDL‐C 2.9 1.000 4.4 1.478 (0.759‐2.879) 9.9 3.686 (2.250‐6.036)

Elevated BP 11.9 1.000 27.0 2.747 (2.000‐3.774) 38.5 4.636 (3.530‐6.089)

Hyperglycaemia 0.4 1.000 0.3 0.867 (0.090‐8.369) 0.0 ‐

De Ferranti definition

Metabolic syndrome 5.4 1.000 24.6 5.700 (3.877‐8.380) 42.0 12.640 (8.971‐17.808)

Central obesity 29.5 1.000 96.4 63.845 (35.233‐115.690) 99.4 401.763 (127.942‐>999.999)

High TG 20.9 1.000 35.4 2.080 (1.575‐2.747) 56.4 4.905 (3.858‐6.236)

Low HDL‐C 16.8 1.000 28.2 1.942 (1.442‐2.616) 35.9 2.765 (2.145‐3.566)

Elevated BP 11.9 1.000 27.0 2.747 (2.000‐3.774) 38.5 4.636 (3.530‐6.089)

Hyperglycaemia 0.4 1.000 0.3 0.867 (0.090‐8.369) 0.0 ‐


220

5.1.4 Discussion

This is the first study to describe the prevalence of the metabolic syndrome in

Chinese overweight/obese children aged 6‐12 years recruited from four cities. The

prevalence of the metabolic syndrome in overweight/obese Chinese children was

5.7%, 18.3%, and 35.1% using the IDF, Cook, and de Ferranti definitions, respectively.

The prevalence rate in this Chinese cohort is similar to that found in western

populations. For example, Cook et al. (2003) showed that the metabolic syndrome

was present in 28.7% of obese US adolescents (BMI ≥95th percentile), 6.8% in

overweight adolescents (BMI 85th percentile to <95th percentile), and 0.1% in

normal‐weight adolescents. De Ferranti et al. (2004) also showed that 31.2% of

overweight/obese adolescents had the metabolic syndrome in the US. Reinehr et al.

(2007) conducted a study in 1205 German overweight children aged 4‐16 years and

reported the prevalence of metabolic syndrome using the Cook and de Ferranti

definitions as 21.0% and 39.0%, respectively. However, the distribution each

element of the metabolic syndrome is different. In the current study, abdominal

obesity, high TG and elevated blood pressure were most frequent in both

overweight and obese Chinese children using the definitions proposed by Cook et al.

(2003) and de Ferranti et al. (2004). This result is similar to studies in Chinese adults

(Zuoa et al., 2009) and Japanese children (Yoshinaga et al., 2005) in which the

prevalence of elevated blood pressure was higher than low HDL. In contrast, in the

studies of Cook et al. (2003) and de Ferranti et al. (2004), high TG, low HDL‐C, and

abdominal obesity were the most common components in US overweight and obese

adolescents. Moreover, even using the same cut‐off points to define metabolic

abnormalities, the prevalence of high TG, low HDL‐C, and hyperglycaemia in US

overweight (33.5%, 32.3% and 4.5%, respectively) and obese US adolescents (51.8%,

50.0%, and 2.6%, respectively) were higher than those in Chinese overweight (27.9%,

4.4%, and 0.3%, respectively) and obese children (46.6%, 9.9%, and 0.0%,

respectively) using Cook definition. The prevalence of abdominal obesity and low

HDL‐C (11.5% and 8.6%, respectively in overweight and 74.5% and 11.2%,

respectively in obese US adolescents) were lower than Chinese children (69.1% and


221

27.0%, respectively in the overweight, and 96.9% and 38.5% respectively, in the

obese. Ethnic differences in the prevalence of the metabolic syndrome and

abnormalities have also been reported in some earlier studies among children (Cook

et al., 2003; Duncan et al., 2004) and adults (Ford, 2005). For example, Cook et al.

(2003) compared the prevalence of the metabolic syndrome and abnormalities in

whites, Mexican Americans and blacks. White adolescents had the highest

proportion of high TG and low HDL‐C while Mexican American adolescents had the

highest rate of abdominal obesity. Black adolescents had the highest proportion of

elevated blood pressure. One of the main reasons for the difference in the

prevalence of metabolic abnormalities may be an ethnic difference in the

susceptibility to metabolic variables (Arslanian et al., 1997; Huxley et al., 2008; Ke et

al., 2009; Whincup et al., 2002, 2005; Zhu et al., 2005). For example, it has been

reported that the absolute risk of hypertension is higher in Asians than Caucasians

among both men and women at any given level of BMI and WC. The odds of

prevalent hypertension associated with the same standard increment in BMI and

WC were stronger in Asians compared with Caucasians (Huxley et al., 2008).

Whincup et al. (2005) reported that South Asian children 13‐16 years of age had

higher mean fasting insulin concentration and fasting blood glucose and a higher

prevalence of impaired fasting glucose compared with European Caucasians. The

differences persisted even after adjustment for adiposity and pubertal status. Zhu et

al. (2005) also found that the odds ratio for having one or more metabolic risk

factors was highest in obese white women, followed by Hispanic and black at a given

level of BMI and WC. Hispanic obese men had higher risk for having one or more

metabolic risk factors than white and black. In a 3‐year longitudinal study, SBP

increased 1.7 mmHg for each unit BMI increase in South‐east Asian children, which

was higher than Australian children (0.8 mmHg) and similar changes of SBP with WC

increase was found (Ke et al., 2009). Ethnic‐specific cut‐offs of metabolic variables,

not only WC but also lipid profiles should be proposed to diagnose metabolic

abnormalities. In addition to ethnicity, different age range, degree of

overweight/obesity, dietary patterns and lifestyle might also contribute to the

differences in findings between studies.


222

The present study also found that overweight and obese Chinese children had a

higher risk of the metabolic syndrome and abnormalities compared with children

who were normal‐weight, regardless of the definition. This is consistent with

previous studies in white, Hispanic, and black participants despite a variation in

prevalence rates depending on diverse criteria (Cook et al., 2003; de Ferranti et al.,

2004; Reinehr et al., 2007; Weiss et al., 2004; Yoshinaga et al., 2005). Cook et al.

(2003) found that the prevalence of the metabolic syndrome was much higher in

overweight compared with normal‐weight adolescents (28.7% vs 0.1%). de Ferranti

(2004) also explored the higher prevalence in metabolic syndrome in overweight

adolescents compared to the total population (31.2% vs 9.2%), and Goodman et al.

(2004) also indicated that the prevalence of the NCEP‐defined metabolic syndrome

was 4.2% among 1513 black, white and Hispanic teens, but 19.5% among obese

teens. Moreover, the risk for the presence of the metabolic syndrome increases with

the severity of obesity. Reinehr et al. (2007) explored that the odds of prevalent

metabolic syndrome associated with a unit increase of standard BMI were 1.72 (95%

CI 1.23‐2.41) to 3.81 (95% CI 2.72‐5.32) using different definitions in 1205 German

overweight children aged 4‐16 years. Yoshinaga et al. (2005) showed that 17.7% of

obese Japanese children had the metabolic syndrome, 2 times more than in

overweight children (8.7%).

In conclusion, although the overall prevalence of the metabolic syndrome in

Chinese children is lower than in American counterparts, the prevalence rates in

overweight and obese children is similar in both countries. This means that with the

rapid increase in obesity in China, the prevalence of the metabolic syndrome is

becoming more common. Due to the rapid nutritional and lifestyle transition being

experienced in China, it is a critical period for the prevention of obesity and the

metabolic syndrome.


223

5.2 ASSOCIATIONS OF OBESITY INDICES WITH METABOLIC RISK FACTORS AMONG CHINESE CHILDREN

5.2.1 Introduction

It has been reported that a quarter of the world’s adult population has the

metabolic syndrome and the condition is appearing more commonly in children and

adolescents due to the growing obesity epidemic within this population (Cook et al.,

2003; Weiss et al., 2004). Data from NHANES III indicate that nearly one in ten

adolescents in the US have three or more of the risk factors involved in the

metabolic syndrome (de Ferranti et al., 2004). The prevalence of the metabolic

syndrome in Chinese adolescents was estimated to be 3.3% in 2002 (Li, Yang et al.,

2008). Metabolic syndrome in adulthood is a clear indicator of morbidity including

greater risk for cardiovascular disease and diabetes mellitus (Alberti et al., 2005;

Haffner et al., 1992; Isomaa et al., 2001; Trevisan et al., 1998). Paediatric metabolic

syndrome can also increase cardiovascular risks (Ronnemaa et al., 1991) and track

from childhood to adulthood (Duncan et al., 2004). Given the increasing prevalence

of the syndrome in children and adolescents, greater attention is being paid to

research in the area.

Obesity is regarded as a central feature of the metabolic syndrome and all

definitions of the syndrome, for children and adults alike, incorporate obesity

(Alberti et al., 2005; Cook et al., 2003; de Ferranti et al., 2004; Expert Panel on

Detection, 2001; Weiss et al., 2004; Zimmet et al., 2007). However, the inconsistent

use of various obesity indices, including BMI (Lambert et al., 2004; Weiss et al., 2004)

and WC (Cook et al., 2003; de Ferranti et al., 2004; Zimmet et al., 2007) has resulted

in different prevalence data. While the choice between these two parameters

remains a matter of ongoing debate, some studies contend that other obesity

indices may be better predictors of cardiovascular disease risk than either BMI or

WC, for example WHtR (Hara et al., 2002; Savva et al., 2000) and skin fold thickness

(Maffeis et al., 2001; Misra et al., 2006; Teixeira et al., 2001). Moreover, an

increasing number of studies found that obesity indices predicted metabolic risk

factors equally well (Garnett et al., 2007; Lee, Song, et al., 2008; Plachta‐Danielzik et


224

al., 2008). Therefore, the IDF has recommended further investigation regarding the

definition of obesity in children, including the potential use of measures such as

WHtR and WC (Zimmet et al., 2007).

Body composition in children varies with age, sex and pubertal status (Heymsfield et

al., 2005). Accordingly, it is more challenging to explore relationships between

obesity indices and metabolic risk factors in children than adults. Although previous

studies have reported a relationship between obesity indices and metabolic risk

factors in different age groups, most have been cross‐sectional rather than

longitudinal and therefore have not considered the influence of growth. Moreover,

associations between obesity indices and metabolic risk factors are also likely to

vary between ethnic groups due to differences in both body composition and

metabolic variables (Heymsfield et al., 2005; Whincup et al., 2002). To our

knowledge, there is minimal published data regarding associations between obesity

indices and metabolic risk factors in Chinese children. The purpose of this study was

to determine which obesity index was the best predictor of metabolic risk clustering

across a 2‐year period among Chinese children.

5.2.2 Methodology


A total of 742 children aged between 8 and 10 years from ten elementary schools in

Beijing, China were recruited for the study in 2005. All participants were

recontacted and examined 2 years later. A signed consent form was obtained from

the participants and their parent(s). All data were collected at baseline and 2 years

later and on each occasion, measurements were performed on the same day in the

morning after an overnight fast.


Height was measured to the nearest 0.1 cm in bare feet. Weight was measured to

the nearest 0.1 kg using a balance‐beam scale with participants wearing lightweight


225

clothing. BMI (kg/m2) was calculated as weight divided by the square of height (m).

WC was measured to the nearest 0.1 cm at the mid‐point between the lower costal

border and the top of the iliac crest with the measurement taken at the end of

normal expiration. WHtR was calculated as WC (cm) divided by height (cm).

5.2.2.3 Body composition assessment

Body composition was assessed using a single frequency 50 kHz BIA device (RJL2

System 101, USA). Total body resistance and reactance were measured with

electrodes placed on landmarks on the right side of each participant lying supine.

Body impedance (Ω) was calculated as the square root of (resistance2+reactance2).

FFM was calculated using the prediction equations developed by Deurenberg et al.

(1991). FM and %BF were then calculated based on the 2‐component model of body

composition.

5.2.2.4 Pubertal stage assessment

Pubertal stage of each participant (genitalia development and pubic hair in boys,

breast development and pubic hair in girls) was assessed according to the criteria of

Tanner (Tanner & Whitehouse, 1976) by trained investigators.

5.2.2.5 Metabolic variables measurements

Blood pressure was measured using a random‐zero sphygmomanometer after the

participant rested for 5 min in a seated position. Two resting blood pressure

measurements were taken to the nearest 4 mmHg, and the first and fifth Korotkoff

sounds were used to represent SBP and DBP, respectively.

A venous blood sample of 5 mL was collected after an overnight fast. Serum glucose

concentration was measured enzymatically using an automated analyzer (Cobas

Mira; Roche Diagnostic systems, Indianapolis, IN). Serum TG and HDL‐C were

determined enzymatically with a bichromatic analyzer (Abbott Diagnostics Spectrum

CCX, Abbott Laboratory, North Chicago, IL).


226

5.2.2.6 Definition of overweight, obesity and central adiposity

Criteria proposed by the Working Group on Obesity in China was used to define

overweight and obesity on the basis of BMI (Group of China Obesity Task Force,

2004). Overweight and obesity was also defined as ≥25 %BF for boys and ≥30 %BF

for girls (Williams et al., 1992). Central adiposity was defined both by ≥90th

percentile of WC suggested by the IDF (Zimmet et al., 2007) with cut‐off values from

Hong Kong Chinese children (Sung et al., 2008) and by ≥0.5 for the WHtR for both

genders (Ashwell et al., 1996).

5.2.2.7 Definition of metabolic risk clustering

Metabolic risk clustering was defined as two or more of the following characteristics:

1) TG ≥1.7 mmol/L;

2) HDL‐C <1.03 mmol/L;

3) SBP and/or DBP ≥90th percentiles for age, gender and height (National High

Blood Pressure Education Program Working Group on Hypertension Control in

Children and Adolescents, 1996);

4) Fasting glucose ≥5.6 mmol/L or known diabetes mellitus.


Continuous variables were presented as mean±SD. Paired t‐test was employed to

compare means between baseline and 2‐year follow‐up. OR was calculated for the

risk of children who were overweight and obese or with central adiposity continuing

to be overweight or obese 2 years later. Pearson’s correlation coefficients were used

to characterize the relationships between metabolic risk factors and obesity indices

adjusted for age, sex and pubertal stage. To test the significance of differences

between correlation coefficients, all coefficients were transformed by using Fisher’s

z’s transformation and t‐test was used to test for the equality between the

transformed coefficients. Stepwise multiple linear regression analysis was employed

to test the influence of independent variables on the variance of dependent


227

variables. The dependent variables for this study were the metabolic risk factors

including TG, HDL‐C, SBP, DBP and fasting glucose. Independent variables were BMI,

WC, WHtR, %BF and pubertal stage, as well as age and sex. ROC analysis was used to

calculate AUC to explore the diagnostic ability of BMI, WC, WHtR and %BF to

identify the presence or absence of high metabolic risk clustering. Relative risk (RR)

and 95% CI was calculated to explore the likelihood of developing metabolic risk

clustering 2 years later among children categorized by BMI, WC, WHtR and %BF at

baseline. Only children who were free of metabolic risk clustering at baseline were

included in this analysis. Statistical analyses were performed with SAS 8.02 (SAS,

Cary, NC). All statistical tests were two‐sided and statistical significant level was set

at a P‐value <0.05.

5.2.3 Results

A total of 569 children (292 boys and 277 girls) were re‐examined 2 years later and

173 (23.3%) dropped out. However, there was no significant difference in the

characteristics between the participants who remained and those who were lost. The

changes in body composition and metabolic risk factors over the 2‐year follow‐up

period are presented in Table 5.3. Both boys and girls had significantly higher values

on all anthropometric variables and metabolic risk factors at follow‐up except for

WHtR, %BF and DBP in boys, and WHtR and %BF in girls.


228

Table 5.3. Changes in anthropometric and metabolic variables (292 boys, 277 girls) after 2‐year follow‐up (mean±SD)

Baseline 2 years later 2‐yr changes

Boys (n=292)

Age (yr) 9.4±0.7 11.4±0.7 2.0±0.0*

Height (cm) 141.9±7.0 154.0±8.6 12.0±2.9*

Weight (kg) 39.8±10.6 51.0±13.5 11.2±4.4*

FM (kg) 11.0±4.7 13.7±5.9 2.7±2.6*

FFM (kg) 28.8±6.2 37.3±8.4 8.5±3.4*

BMI (kg/m2) 19.5±4.0 21.2±4.2 1.7±1.4*

WC (cm) 67.2±11.5 73.8±12.3 6.6±5.1*

WHtR 0.47±0.07 0.48±0.07 0.01±0.04

%BF 26.6±5.2 25.9±5.9 ‐0.7±4.0

TG (mmol/L) 1.04±0.81 1.85±0.70 0.81±0.98*

HDL‐C (mmol/L) 1.43±0.28 2.00±0.42 0.57±0.36*

Glucose (mmol/L) 4.64±0.36 5.11±0.44 0.47±0.45*

SBP (mmHg) 106±9 109±10 3±10*

DBP (mmHg) 69±7 69±7 1±8

Girls (n=277)

Age (yr) 9.1±0.7 11.1±0.7 2.0±0.0*

Height (cm) 139.4±7.1 151.6±7.1 12.2±2.4*

Weight (kg) 35.6±8.7 46.2±10.9 10.6±5.2*

FM (kg) 10.1±3.8 13.1±5.2 3.0±3.1*

FFM (kg) 25.5±5.3 33.1±6.2 7.6±2.7*

BMI (kg/m2) 18.2±3.4 20.0±3.8 1.8±2.0*

WC (cm) 61.7±9.4 67.8±10.1 6.1±5.0*

WHtR 0.44±0.06 0.45±0.06 0.01±0.03

%BF 27.6±4.6 27.3±5.7 ‐0.3±4.6

TG (mmol/L) 1.06±0.51 1.88±0.75 0.83±0.79*

HDL‐C (mmol/L) 1.40±0.28 2.01±0.40 0.62±0.28*

Glucose (mmol/L) 4.56±0.33 5.06±0.35 0.50±0.39*

SBP (mmHg) 103±9 107±9 4±9*

DBP (mmHg) 67±7 70±7 3±8*

Paired t‐test was used to compare the changes in variables between baseline and 2‐year follow‐up. * Significant difference in the changes, P <0.001.


229

Table 5.4 shows the prevalence of overweight and obesity on the basis of different

indices at baseline and 2 years later. After two years, the percentage of children

who remained overweight and obese defined on the basis of BMI and %BF was

89.7% (OR = 88.823, 95% CI: 50.492‐156.254) and 80.4% (OR = 22.519, 95% CI:

14.582, 34.777), respectively; the percentage of children who continued to be

centrally obese, defined on the basis of WC and WHtR, was 93.5% (OR = 75.539,

95% CI: 41.507, 137.474) and 84.5% (OR = 47.094, 95% CI: 27.508, 80.624).

Table 5.4. Prevalence of total and central adiposity at baseline and 2 years later

Baseline

n (%)

2 years later

n (%)

Boys

Overweight and obesity by BMI1 138 (47.3) 150 (51.4)

Central adiposity by WC2 122 (41.8) 139 (47.6)

Central adiposity by WHtR3 99 (33.9) 115 (39.4)

Overweight and obesity by %BF4 178 (61.0) 174 (59.6)

Girls

Overweight and obesity by BMI1 95 (34.3) 89 (32.1)

Central adiposity by WC2 107 (38.6) 129 (46.6)

Central adiposity by WHtR3 56 (20.2) 59 (21.3)

Overweight and obesity by %BF4 92 (33.2) 89 (32.1)

1 Defined on the basis of BMI according to Group of China Obesity Task Force criteria (2004). 2 Defined as ≥90th percentile of Chinese children (Sung et al., 2008; Zimmet et al., 2007). 3 Defined as a WHtR ≥0.5 (Ashwell et al., 1996). 4 Defined as a %BF ≥25 for boys and ≥30 for girls (Williams et al., 1992).

Partial Pearson’s correlations between each obesity index at baseline and metabolic

risk factors at baseline and 2 years later are presented in Table 5.5. Obesity indices

at baseline correlated significantly with TG, HDL‐C, SBP and DBP at both baseline

and 2 years later. There were differences in the strength of correlations between

different obesity indices within one metabolic risk factor however these did not

reach significance.


230

Table 5.5. Partial Pearson correlations between BMI, WC, WHtR and %BF and metabolic risk factors at baseline and 2 years later

(after adjusting for age and pubertal status)

Risk factors at baseline Risk factors 2 years later Obesity indices at baseline TG HDL‐C SBP DBP Glucose TG HDL‐C SBP DBP Glucose

Boys

BMI* 0.285 ‐0.314 0.481 0.394 0.132 0.320 ‐0.312 0.438 0.192 0.128

WC* 0.304 ‐0.332 0.480 0.388 0.138 0.347 ‐0.338 0.421 0.167 0.149

WHtR* 0.299 ‐0.327 0.465 0.373 0.137 0.370 ‐0.311 0.393 0.152 0.137

%BF* 0.291 ‐0.283 0.382 0.297 0.108** 0.278 ‐0.246 0.335 0.085** 0.085**

Girls

BMI* 0.298 ‐0.289 0.367 0.301 0.066** 0.288 ‐0.328 0.310 0.149 0.071**

WC* 0.308 ‐0.319 0.311 0.294 0.070** 0.280 ‐0.341 0.246 0.117 0.054**

WHtR* 0.315 ‐0.302 0.291 0.263 0.068** 0.325 ‐0.328 0.229 0.101** 0.082**

%BF* 0.285 ‐0.187 0.242 0.236 0.004** 0.337 ‐0.189 0.179 0.057** 0.110**

* No significant difference in the strength of correlations between different obesity indices within one metabolic risk factor.

** No significant relationship between obesity index and metabolic risk factor.


231

Results of the multiple regression analysis for each dependent variable are shown in

Table 5.6 and Table 5.7. At baseline, WC explained the greatest proportion of the

variance of TG and HDL‐C; BMI explained the greatest variance of blood pressure;

and none of the obesity indices contributed to the significant variance of glucose.

After 2 years, BMI at baseline still explained the greatest variance of later blood

pressure. WC at baseline explained the greatest variance of later HDL‐C and glucose,

while WHtR at baseline contributed in the greatest variance of later TG.

Table 5.6. Stepwise multiple regression analysis model to explain the variance in metabolic risk factors using BMI, WC, %BF and WHtR age, sex, and pubertal stage

as independent variables at baseline.

Variables Coefficient se Standardized estimate

R2 change

P value

TG Constant ‐0.433 0.176 0.000 ‐ 0.0143

WC 0.013 0.004 0.210 0.098 0.0005

%BF 0.020 0.008 0.140 0.013 0.0151

Pubertal

stage 0.084 0.049

0.072 0.005 0.0872

HDL‐C Constant 1.855 0.172 0.000 ‐

WC ‐0.009 0.001 ‐0.360 0.097 < 0.0001

Sex ‐0.075 0.023 ‐0.135 0.020 0.0005

Age 0.029 0.016 0.068 0.005 0.0778

SBP Constant 86.164 2.202 0.000 ‐ <0.0001

BMI 1.083 0.095 0.446 0.235 <0.0001

Sex ‐3.099 0.937 ‐0.257 0.007 0.0208

Pubertal

stage 1.926 0.817

0.200 0.007 0.0188

DBP Constant 44.167 3.722 0.000 ‐ <0.0001

BMI 0.696 0.082 0.382 0.158 <0.0001

Age 1.138 0.384 0.115 0.007 0.0313

Glucose Constant 4.585 0.088 0.000 ‐ <0.0001

%BF 0.005 0.003 0.074 0.005 0.0789

Sex ‐0.082 0.029 ‐0.120 0.012 0.0081


232

Table 5.7. Stepwise multiple regression analysis model to explain the variance in metabolic risk factors at 2‐yr later using BMI, WC, %BF and WHtR at baseline, change of each indices, age, sex, and pubertal stage as independent variables.

Variables Coefficient s.e. Standardized estimate

R2 change

P value

TG Constant ‐0.327 0.241 0.000 ‐ 0.1762

WHtR at baseline 8.366 1.418 0.774 0.101 <0.0001

ΔWHtR 9.408 3.742 0.443 0.030 0.0122

WC ‐0.025 0.009 ‐0.374 0.018 0.0058

Sex 0.139 0.058 0.094 0.008 0.0201

ΔWC ‐0.043 0.024 ‐0.301 0.007 0.073

HDL‐C Constant 3.015 0.112 0.000 ‐ <0.0001

WC at baseline ‐0.017 0.002 ‐0.440 0.123 <0.0001

Pubertal stage ‐0.050 0.017 ‐0.119 0.007 0.0043

ΔWC ‐0.008 0.003 ‐0.097 0.003 0.0136

%BF at baseline 0.008 0.005 0.098 0.003 0.1026

SBP Constant 81.150 6.939 0.000 ‐ <0.0001

BMI at baseline 1.892 0.289 0.738 0.178 <0.0001

WHtR at baseline ‐59.026 16.465 ‐0.409 0.021 0.0004

ΔWC 1.070 0.297 0.556 0.014 0.0003

ΔWHtR ‐128.923 44.808 ‐0.454 0.013 0.0042

Age 1.070 0.530 0.077 0.006 0.0438

DBP Constant 61.211 1.739 0.000 ‐ <0.0001

BMI at baseline 0.652 0.118 0.350 0.034 <0.0001

Pubertal stage 1.464 0.300 0.206 0.026 <0.0001

ΔWHtR 24.236 8.521 0.118 0.014 0.0046

%BF at baseline ‐0.254 0.091 ‐0.179 0.013 0.0052

Glucose Constant 4.799 0.101 0.000 ‐ <0.0001

WC at baseline 0.004 0.002 0.122 0.015 0.0037


233

The ROC curves of the BMI, WC, WHtR, and %BF at baseline for diagnosing

metabolic risk clustering at baseline and at 2 years later are shown in Figure 5.1.

The AUC values for different obesity indices were similar among both boys and girls.

Figure 5.1. ROC curves for BMI, WC, WHtR and %BF at baseline for the prediction of metabolic risk clustering at baseline (a) and 2 years later (b) in boys and girls.

Table 5.8 presents the effect of BMI, WC, WHtR and %BF at baseline on developing

metabolic abnormalities and metabolic risk clustering 2 years later. Children who

were overweight or obese, or had increased central adiposity defined on the basis

b

a


234

of BMI, WC, WHtR or %BF at baseline, were more likely to develop metabolic

abnormalities and metabolic risk clustering 2 years later than those who were

normal‐weight. After adjustment for age, gender and pubertal status, the RR for

BMI (3.670, 95% CI: 2.285, 5.893) and WC (3.762, 95% CI: 2.355, 6.009) was similar

and slightly higher than for WHtR and %BF.


235

Table 5.8. The relative risk (95% CI) of developing metabolic abnormalities 2 years later among total or central adiposity based on level of BMI, WC, WHtR and %BF at baseline adjusted for age, gender and pubertal stage.

2‐yr later Weight status at baseline

Increased TG Decreased HDL‐C Elevated blood pressure Increased glucose CV risk clustering

Overweight and obesity by BMI 2.729 (1.873‐3.977) ‐ 3.062 (1.928‐4.863) 2.153 (1.117‐4.150) 3.670 (2.285‐5.893)

Central adiposity by WC 2.858 (1.962‐4.163) ‐ 2.500 (1.595‐3.918) 2.964 (1.523‐5.770) 3.762 (2.355‐6.009)

Central adiposity by WHtR 2.680 (1.750‐4.104) ‐ 2.981 (1.878‐4.740) 1.838 (0.940‐3.595) 2.767 (1.743‐4.392)

Overweight and obesity by %BF 2.976 (2.037‐4.379) ‐ 2.545 (1.580‐4.101) 2.285 (1.144‐4.566) 2.804 (1.727‐4.552)

For each metabolic abnormality, only children without the condition at baseline were included in the logistic regression analysis. RRs for decreased HDL could not calculated because only one child was diagnosed as having decreased HDL‐C 2 years later.


236

5.2.4 Discussion

The objective of the present study was to assess which obesity index was superior

in predicting metabolic risk clustering. However, there was no significant difference

in the ability of BMI, WC, WHtR or %BF to predict an increased metabolic risk

clustering in Chinese children across a 2‐year period.

In the current study, baseline data were used for cross‐sectional analyses. The AUCs

calculated by ROC analysis for BMI, WC, WHtR, and %BF with overlapping

confidence interval revealed a similar accuracy for the four indices in the prediction

of the prevalence of metabolic risk clustering (Figure 5.1). Meanwhile, longitudinal

analyses showed that both total and central adiposity at baseline remained as

predictors of later metabolic risk clustering. At baseline, the four indices had a

similar ability to distinguish later metabolic risk clustering among both boys and

girls using ROC analysis. Importantly, children who were overweight or obese, or

had central adiposity at baseline had an increased risk for the development of later

metabolic risk clustering than those who were normal‐weight at baseline. The

relative risks for BMI and WC were similar and slightly higher than for %BF and

WHtR however the 95% CIs were overlapping and the differences were not

significant. Our results are in agreement with some cross‐sectional studies in which

obesity indices predicted metabolic risk factors equally well (Lee, Song et al., 2008;

Plachta‐Danielzik et al., 2008) and a longitudinal study in which WC in

mid‐childhood was no better for identifying those at increased risk of metabolic risk

clustering in adolescence than BMI (Garnett et al., 2007).

However, the findings of the present study contrast with some adult and pediatric

cross‐sectional studies which indicated that WC or WHtR was a better predictor of

metabolic risk clustering than BMI or %BF (Hara et al., 2002; Hsieh & Muto, 2005;

Janssen et al., 2004; Lee, Huxley, et al., 2008; Savva et al., 2000; Watts et al., 2008).

Although WC or WHtR appeared to be a better predictor of TG and HDL‐C, the

indices did not show an advantage over BMI in predicting metabolic risk clustering

in the current study. Fat distribution appears to be a more important influence on


237

metabolic risk factors than overall adiposity. Central adiposity, particularly excess

visceral fat accumulation, has been shown to be more closely related to

obesity‐related health risk (Garan & Gower, 1999). Therefore, the measurement of

visceral adipose tissue may surpass the value of BMI or %BF. WC, as a simple and

practical index of visceral adipose tissue (Daniels et al., 2000; Pouliot et al., 1994)

has been proposed as an effective index to predict metabolic risk factors for both

children and adults. However, the association of WC with metabolic risk may be

confounded by body height due to a strong positive relationship between WC and

height throughout childhood and into adulthood, and the independent negative

effect of height on the metabolic risk factors (Henriksson, Lindblad, Agren,

Nilsson‐Ehle, & Rastam, 2001). Accordingly, WHtR has been proposed as a better

predictor than BMI or WC in both children (Hara et al., 2002; Savva et al., 2000) and

adults (Hsieh & Muto, 2005). However, it should be noted that many of the

differences in the strength of correlations and AUCs among BMI, WC, WHtR and

%BF were relatively small in these studies. Further, a range of factors may have

contributed to the contradictory findings in previous studies. First of all, the best

predictive index for metabolic risk factors varied according to age and gender

(Kelishadi et al., 2007) which makes it difficult to make cross‐sectional comparisons.

Secondly, measurement sites used for WC vary between studies. Significant

differences in WC values measured at different sites have been reported (Wang et

al., 2003), whereas studies to compare the difference in the associations between

different landmarks and obesity‐related health risk are limited. Finally, various

subsets of risk factors have been included in the definition of metabolic risk

clustering in each study. The various obesity indices showed different associations

with various metabolic risk factors. For example, BMI seemed to be a better

predictor of blood pressure than WC and %BF among adults and children (Savva et

al., 2000), despite a lack of evidence for a relationship between WC and blood

pressure (McCarthy, 2006). However, WC has been well documented as a predictor

of blood lipid profiles in children (McCarthy, 2006). Our findings were consistent

with these studies. Accordingly, different definitions of metabolic risk clustering

may result in different predictive obesity indices. The present study also defined


238

metabolic risk clustering using different definitions (one or more, two or more,

three or more, and all four of the characteristics as shown in the methods section)

and also showed that neither WC or WHtR were superior in the prediction of

metabolic risk clustering among children (data not shown).

%BF is a more direct measure of adiposity and is assumed to have an advantage

over BMI, WC or WHtR in predicting metabolic risk factors. However our results, in

agreement with previous studies among children (Plachta‐Danielzik et al., 2008;

Steinberger et al., 2006) and adults (Bosy‐Westphal et al., 2006; Shen et al., 2006),

did not support this hypothesis. Despite inaccuracies in the use of BIA to predict

%BF in the present study, previous studies have indicated that %BF measured by

densitometry and under‐water weighing was also not able to confirm any

advantage of %BF as an index of metabolic risk factors (Bosy‐Westphal et al., 2006;

Tanaka et al., 2002).

WC and WHtR have been preferred by some investigators, mainly because they are

more convenient than %BF. Moreover, self‐reported WC is deemed to be more

accurate than self‐reported BMI (Han & Lean, 1998; Spencer, Appleby, Davey, & Key,

2002). However, a number of issues need to be considered. Besides the limitations

of WC as mentioned above, the tracking of BMI from childhood to adulthood has

been well documented (Freedman et al., 2005; Srinivasan, Myers, & Berenson, 2002)

as have the associations of BMI in childhood with metabolic risk factors in

adulthood (Freedman et al., 2001; Vanhala et al., 1998) despite some limitations in

the use of BMI to define overweight and obesity. However, longitudinal studies

tracking WC, WHtR or %BF from childhood to adulthood have been very limited,

and similarly, their relationships with metabolic risk factors in adulthood. In the

current study, overweight, obese or centrally obese children defined by BMI and

WC, were more persistent than those defined by WHtR and %BF.

There are a number of limitations in the present study. Firstly, WC cut‐off values to

define central adiposity were derived from Hong Kong children. Differences may

exist in WC distribution between the southern and northern areas of China because


239

of differences in height and weight (China's National Group on Student's

Constitution and Health Survey, 2007). However, presently there are no

nationally‐representative WC distributions in China and one could contend that

cut‐off values for Hong Kong Chinese children would be more suitable than those

from other countries. Furthermore, the WHtR value of 0.5 used in the current study

was proposed in adults however WHtR is weakly associated with age. Therefore, the

same cut‐off could possibly be used among both children and adults (Ashwell et al.,

1996; McCarthy & Ashwell, 2006). One of the main strengths of the current study

was its longitudinal design and exploration of the ability of obesity indices to predict

long‐term health risk.

In conclusion, the present study indicated that BMI, WC, WHtR and %BF performed

similarly in the prediction of metabolic risk clustering among Chinese children.

However, BMI appeared to be more closely related to blood pressure while WC

tended to be more closely related to TG and HDL‐C.


240

5.3 WAIST CIRCUMFERENCE AND WAIST‐TO‐HEIGHT RATIO CUT‐OFF VALUES FOR THE PREDICTION OF CARDIOVASCULAR RISK FACTORS CLUSTERING IN CHINESE SCHOOL‐AGED CHILDREN

Modified from: Liu A, Hills AP, Hu X, Li Y, Du L, Xu Y, Byrne NM, Ma G (2010). Waist circumference cut‐off points for the prediction of cardiovascular risk factors clustering in Chinese school‐aged children: a cross‐sectional study. BMC Public Health, 10:82.

5.3.1 Introduction

The global prevalence of overweight and obesity has increased dramatically in

developed countries in recent decades (Booth et al., 2007; Lobstein et al., 2004;

Magarey et al., 2001; National Center for Health Statistics, 2004), however evidence

suggests that a greater potential problem exists for developing nations such as China

(Li, Schouten et al., 2008; Lobstein et al., 2004).

Higher than desirable levels of body fat pose an increased risk of ill‐health, however

the location of excess fat appears to have particular implications (Després et al.,

1990, 2007; Janssen et al., 2004). For example, a greater concentration of adipose

tissue in the abdomen, specifically in the visceral area, is directly related to

metabolic and cardiovascular risk in adults (Leenen, van der Kooy, Seidell, &

Deurenberg, 1992). Visceral adiposity is best quantified using sophisticated imaging

techniques however such approaches are not feasible at the population level

(Heymsfield et al., 2005).

Recent attention has been paid to the applicability of anthropometric markers to

measure abdominal obesity and the WC and WHtR have consistently been identified

as better measures of CV risk than BMI (Hara et al., 2002; Janssen et al., 2004;

Lofren et al., 2004; Savva et al., 2000). WC is also recognized as a key component of

the metabolic syndrome in both children and adults (Alberti et al., 2006; Zimmet et

al., 2007). WC and WHtR cut‐off points associated with increased risk have been

developed for adult men and women, however relatively less work has been

undertaken in children and adolescents. A further shortcoming of research to date is


241

that reference standards have more commonly been developed on Caucasian

populations and may have limited usefulness to people from different ethnic and

racial backgrounds (Maffeis et al., 2001; Ng et al., 2007).

An increasing body of research has explored ethnic differences in body composition

in both children and adults (Heymsfield et al., 2005), but considerably more work is

needed. For example, the IDF uses the 90th percentile as a cut‐off for WC to define

the pediatric metabolic syndrome but has recommended the development of

ethnic‐, age‐ and gender‐specific normal ranges for WC based on healthy values. In

short, the percentiles used as cut‐offs for WC should be reassessed when more data

are available (Zimmet et al., 2007).

A number of studies have reported on reference WC percentiles (Eisenmann, 2005;

GómezDíaz et al., 2005; Hatipoglu et al., 2008; Katzmarzyk, 2004; McCarthy et al.,

2001; Moreno et al., 1999) and WHtR (McCarthy & Ashwell, 2006; Sung et al., 2008)

values developed for children and adolescents in different countries. To date, two

studies have reported age‐ and gender‐specific WC cut‐offs in Chinese children and

adolescents, one in Hong Kong Chinese children and adolescents (Sung et al., 2007),

and the other in children from Xinjiang province (Yan et al., 2008). No study has

reported age‐ and gender‐specific cut‐offs for WHtR in Chinese children. As there

are well documented regional differences in the body composition of Chinese, for

example those living in the North and South of the country (China's National Group

on Student's Constitution and Health Survey, 2007), the development of WC and

WHtR percentiles and cut‐offs for different groups would be particularly valuable.

Therefore, the purpose of the present study was to develop WC and WHtR

percentiles for Chinese children from both the North and South of the country, and

secondly, to explore the optimal WC and WHtR cut‐off values for predicting CV risk

factors clustering in this population.


242

5.3.2 Methodology


Participants were randomly recruited from three cities, Liaoyang in the North‐east of

China, Tianjin in the North of China, and Guangzhou in the South. Height, weight,

and WC were measured for all children at each school (aged 6‐12 years) and a

sub‐set of 40% of the participants at each school selected for the collection of blood

samples. Written consent was obtained from both children and their parents.


Height was measured to the nearest 0.1 cm in bare feet. Body weight was measured

to the nearest 0.1 kg with a balance‐beam scale with participants wearing

lightweight clothing. The BMI (kg/m2) was calculated as weight (kg) divided by the

square of height (m). WC was measured to the nearest 0.1 cm at the mid‐point

between the lower costal border and the top of the iliac crest with the

measurement taken at the end of a normal expiration. WHtR was then calculated as

WC (cm) divided by height (cm).

5.3.2.3 Cardiovascular risk factors measurements

Blood pressure was measured on the study morning using a random‐zero

sphygmomanometer after the participant rested for 5‐min in a seated position. Two

resting blood pressure measurements were taken to the nearest 4 mmHg, and the

first and fifth Korotkoff sounds were used to represent SBP and DBP, respectively.

A venous blood sample of 5 mL was collected from each participant after an

overnight fast. Serum glucose concentration was measured enzymatically using an

automated analyzer (Cobas Mira; Roche Diagnostic systems, Indianapolis, IN). Serum

TG, TC, and HDL‐C were determined enzymatically with a bichromatic analyzer

(Abbott Diagnostics Spectrum CCX, Abbott Laboratory, North Chicago, IL). LDL‐C was

determined using the Friedewald formula.


243

5.3.2.4 Definition of high CV risk factors clustering

Each participant was classified as having a high CV risk factors clustering with ≥3 of

the following risk factors (Ng et al., 2007):

(1) SBP and/or DBP ≥90th percentiles for age, gender and height recommended by

the National Heart, Lung, and Blood Institute (US) (National High Blood Pressure

Education Program Working Group on Hypertension Control in Children and

Adolescents, 1996);

(2) TG ≥1.7 mmol/L;

(3) HDL‐C <1.03 mmol/L;

(4) LDL‐C ≥3.4 mmol/L;

(5) fasting glucose ≥5.6 mmol/L (NCEP Expert Panel on Blood Cholesterol Levels in

Children and Adolescents, 1992; Zimmet et al., 2007)


Continuous data were described as means±SD. The age and gender differences

were assessed by t‐test. Smoothed age‐ and gender‐specific percentiles were

constructed using the LMS ChartMaker Pro software package (The Institute of Child

Health, London) for the whole cohort. ROC analysis was used to explore the

diagnostic ability of WC and WHtR to identify the presence or absence of high CV

risk factors clustering for those children who provided blood. The value which

maximized both sensitivity and specificity was regarded as the optimal threshold for

predicting high CV risk factors clustering. Then the age‐ and sex‐specific WC and

WHtR cut‐off values were read directly from the corresponding smoothed

percentiles constructed from the whole sample by the LMS method. In addition, OR

was calculated using logistic regression analysis adjusted for age to explore the risk

of having high CV risk factors clustering among boys and girls who were at the

optimal threshold of WC and WHtR and higher compared with their counterparts.

The SAS 8.02 software package was used for analysis. All statistical analyses were

two‐sided and a P value of <0.05 was considered statistically significant.


244

5.3.3 Results

A total of 5529 children (2830 boys and 2699 girls) aged 6‐12 years participated in

the study. The descriptive characteristics of the sample by age and gender are

shown in Table 5.9. Boys were taller than girls at 7, 10, 11, and 12 year old. Boys

were heavier than girls at each age group except for at 12 years. Boys had higher

values in BMI, WC, and WHtR than girls except for WC at 6 years of age.


245

Table 5.9. Sample size and mean and SD for height, weight, BMI, WC, WHtR for Chinese children aged 6 to 12 years by age and gender

Age n Height (cm) Weight (kg) BMI (kg/m2) WC (cm) WHtR

Boys 2830

6 152 120.3±5.0 24.1±4.9 * 16.5±2.5* 54.9±7.2 0.46±0.05**

7 469 124.5±5.3** 25.8±5.7** 16.5±2.9** 56.5±7.0** 0.45±0.05**

8 506 129.0±6.3 28.6±8.1** 17.0±3.6** 59.1±8.4** 0.46±0.06**

9 421 134.6±6.8 33.5±10.8 ** 18.3±5.4** 62.2±10.1** 0.46±0.06**

10 550 138.8±6.8* 36.4±10.2** 18.6±3.9** 65.5±10.5** 0.47±0.06**

11 499 143.5±7.3* 40.6±11.4 ** 19.5±4.4** 67.6±11.0** 0.47±0.07**

12 233 147.0±7.2* 42.7 ±12.5 19.5±4.4** 68.0±11.8** 0.46±0.07**

Girls 2699

6 165 119.6±5.0 22.9±4.2 15.9±2.1 53.8±5.5 0.45±0.04

7 470 123.1±5.7 24.0±4.4 15.8±2.1 54.0±5.4 0.44±0.04

8 522 128.7±5.9 27.0±5.6 16.2±2.7 56.6±6.6 0.44±0.05

9 416 133.5±7.1 30.6±7.5 17.0±3.0 59.3±7.6 0.44±0.05

10 546 138.0±7.1 33.9±9.1 17.6±3.6 61.3±8.9 0.44±0.06

11 440 144.3±7.3 38.0±9.1 18.1±3.3 62.8±9.0 0.43±0.06

12 140 148.5±7.3 41.6±10.1 18.7±3.8 64.3±9.0 0.43±0.06

* P <0.05, **P <0.001 for the difference between boys and girls.


246

The age‐ and gender‐specific 3rd, 10th, 25th, 50th, 75th, 90th, and 97th percentiles

for WC are shown in Table 5.10 and smoothed WC percentile curves are presented

in Figure 5.2. In general, WC increased with age in both boys and girls. Boys had a

higher WC value than girls at every age and percentile level except for the 3rd, 10th,

and 25th percentiles at 6 years.

Table 5.10. Waist circumference percentiles (cm) by age and gender

Age (years) n 3rd 10th 25th 50th 75th 90th 97th

Boys 2830

6 152 44.3 46.5 49.3 52.6 57.0 62.7 71.2

7 469 46.4 48.7 51.5 55.0 59.6 66.1 76.6

8 506 48.2 50.6 53.6 57.4 62.5 70.0 83.1

9 421 49.5 52.4 55.9 60.5 66.6 75.5 90.3

10 550 49.9 53.5 57.8 63.2 70.3 79.9 94.2

11 499 50.3 54.4 59.4 65.4 73.1 83.1 96.8

12 233 51.7 56.6 62.4 69.3 77.7 88.0 101.0

Girls 2699

6 165 45.3 47.3 49.6 52.3 55.6 59.9 65.6

7 470 45.9 48.0 50.4 53.4 57.1 61.9 68.6

8 522 47.1 49.4 52.2 55.6 60.0 65.9 74.4

9 416 47.9 50.5 53.6 57.5 62.4 69.2 79.3

10 546 48.5 51.4 55.0 59.3 64.9 72.4 83.4

11 440 49.3 52.7 56.8 61.9 68.2 76.6 88.2

12 140 50.4 54.5 59.4 65.4 72.9 82.6 95.5


247

Figure 5.2. Smoothed percentile curves for waist circumference in boys (n=2830)

and girls (n=2699).

Table 5.11 shows the age‐ and gender‐specific 3rd, 10th, 25th, 50th, 75th, 90th, and

97th percentiles for WHtR and Figure 5.3 presents the smoothed WHtR percentile

curves. In general, the mean value of WHtR increased with age in boys up to 8 years

old and remained stable until 12 years of age. For girls, the mean value of WHtR was

Boys

Girls


248

similar at 6 to 10 years of age then decreased significantly. Boys had a higher mean

WHtR value than girls at each age, however, differences in WHtR among different

age groups and gender was marginal.

Table 5.11. Waist‐to‐height ratio percentiles by age and gender

Age (years) n 3rd 10th 25th 50th 75th 90th 97th

Boys 2830

6 152 0.39 0.40 0.42 0.44 0.47 0.51 0.56

7 469 0.39 0.40 0.42 0.45 0.48 0.52 0.57

8 506 0.38 0.40 0.42 0.45 0.48 0.53 0.59

9 421 0.38 0.40 0.42 0.45 0.49 0.54 0.60

10 550 0.37 0.39 0.42 0.46 0.50 0.55 0.61

11 499 0.36 0.39 0.42 0.46 0.50 0.55 0.61

12 233 0.36 0.39 0.42 0.46 0.51 0.56 0.63

Girls 2699

6 165 0.39 0.41 0.43 0.44 0.47 0.50 0.53

7 470 0.38 0.39 0.41 0.44 0.46 0.50 0.54

8 522 0.37 0.39 0.41 0.43 0.46 0.50 0.56

9 416 0.36 0.38 0.40 0.43 0.47 0.51 0.57

10 546 0.36 0.38 0.40 0.43 0.46 0.51 0.58

11 440 0.35 0.37 0.39 0.42 0.46 0.51 0.58

12 140 0.34 0.37 0.39 0.43 0.47 0.52 0.58


249

Figure 5.3. Smoothed percentile curves for waist‐to‐height ratio in boys (n=2830) and girls (n=2699).

ROC curves for WC and WHtR with CV risk factors clustering (≥3 of 5 CV risk factors)

in boys and girls is shown in Figures 5.4 and 5.5, repectively. Table 5.12 summarizes

the optimal threshold of WC and WHtR for boys and girls. The 90th and 84th

percentiles for WC represent the cut‐offs for boys and girls, respectively. The OR of a

higher CV risk factors clustering among boys at the 90th percentile of WC and higher,

and 84th and higher percentiles of WC in girls, is 10.349 (95% CI 4.466 to 23.979)

and 8.084 (95% CI 3.147 to 20.767) compared with their counterparts. The 91th and

94th percentiles for WHtR represent the cut‐offs for boys and girls, respectively. The

OR of a higher CV risk factors clustering among boys at the 91th percentile of WHtR

Boys

Girls


250

and higher, and 94th and higher percentiles of WHtR in girls is 36.819 (95% CI 8.273

to 163.856) and 26.189 (95% CI 7.173 to 95.613) compared with their counterparts.

Figure 5.4. ROC curves for waist circumference with higher CV risk factor clustering (≥3 of 5 CV risk factors) in Boys (n = 982) and girls (n = 863).

Figure 5.5. ROC curves for waist‐to‐height ratio with high CV risk factors clustering

( ≥3 of 5 CV risk factors) in boys (n =982) and girls (n=863).


251

Table 5.12. Optimal waist circumference and waist‐to‐height ratio thresholds for CV risk factors clustering in boys (n=982) and girls (n=863)

AUC (95% CI) Sensitivity (%) Specificity (%) Threshold (percentiles) OR (95% CI)

WC Boys 0.814 (0.761‐0.866) 83.3 73.7 90th 10.349 (4.466‐23.979)

Girls 0.810 (0.720‐0.899) 75.0 67.2 84th 8.084 (3.147‐20.767)

WHtR Boys 0.901 (0.865‐0.938) 88.9 79.3 91st 36.819 (8.273‐163.856)

Girls 0.857 (0.743‐0.970) 78.6 87.6 94th 26.189 (7.173‐95.613)


252

Table 5.13 shows the age‐specific WC and WHtR thresholds according to gender.

These cut‐off points are those above which there is an increased likelihood of being

at high risk of CV.

Table 5.13. Optimal age‐ and gender‐specific WC and WHtR cut‐off values for Chinese children

WC WHtR Age

(years) Girls Boys Girls Boys

6 57.2 62.7 0.51 0.51

7 59.3 66.1 0.51 0.52

8 62.6 70.0 0.52 0.53

9 65.7 75.5 0.53 0.54

10 68.5 79.9 0.53 0.55

11 71.8 83.1 0.53 0.56

12 76.3 88.0 0.54 0.57

5.3.4 Discussion

This study provides the age‐ and gender‐specific WC and WHtR reference

percentiles for Chinese children aged 6‐12 years living in the North and South of the

country. Consistent with findings in previous studies (Eisenmann, 2005; Hatipoglu et

al., 2008; Katzmarzyk, 2004; McCarthy et al., 2001), WC increases with age and boys

have a higher value than girls at each age. The age‐ and gender‐related variation of

WC shows a similarity to other body dimensions.

It is interesting to compare our data with those previously reported in British

(McCarthy et al., 2001), Turkish (Hatipoglu et al., 2008), Australian (Eisenmann,

2005), Mexican (GómezDíaz et al., 2005), and Chinese children in Hong Kong (Sung

et al., 2007) and the Xinjiang Province (Yan et al., 2008) (see Fig 5.6). Boys in the

present study had higher WC values than Australian, Turkish, and British boys, and

lower values than Mexican boys aged 6‐12 years, however were similar to British

boys aged 6‐7 years. Girls in the present study had higher WC values than British,


253

Australian and Turkish girls, and lower values than Mexican girls in all age groups

except lower values than Australian girls at 7‐9 years. Compared with Caucasians,

Asians are generally smaller (Deurenberg et al., 1999), however WC results from the

present study contradict this finding. A number of factors may contribute to this

inconsistency. Firstly, the site of WC measurement has varied between studies and

previous research has reported significant differences in WC at different sites (Wang

et al., 2003). International agreement regarding the measurement site for WC is

required if meaningful comparisons regarding central obesity are to be made

between children from different countries and regions. Secondly, considerable

changes in body size and shape, including WC, have occurred in recent decades in

both children and adults (China's National Group on Student's Constitution and

Health Survey, 2007; Utter, Scraqq, Denny, & Schaaf, 2009; Waedle & Boniface, 2008)

and this may be a consideration when comparing data collected in different periods.

For example, Utter et al. (2009) indicated that the mean WC in New Zealand

adolescents increased from 76.2 cm in 1997/1998 to 89.4 cm in 2005, with increases

in WC measurements at all points in the distribution. Data from Britain and Australia

were collected in 1985 and 1990, respectively, a gap of approximately twenty years

from the present study. Compared with other studies of Chinese children, both boys

and girls in the present study have higher WC values in all age groups however had

similar values to 7‐year‐old children from Xinjiang Province. It is widely accepted

that body composition is influenced by the environment in addition to age, gender

and ethnicity (Heymsfield et al., 2005). Compared with the population in Northern

China, both children and adults in Southern China have a smaller body size (China's

National Group on Student's Constitution and Health Survey, 2007). This study also

demonstrated differences in body size among children in Guangzhou (Southern

China), and Liaoning and Tianjin (Northern China) (data not shown). Accordingly,

further study is required in more districts of China to generate more representative

WC percentiles.


254

Figure 5.6. The 50th percentiles for waist circumference in seven studies for boys and girls.

In a sub‐sample, the present study also evaluated the threshold value of WC to

predict a higher CV risk factors clustering using ROC analysis. The 90th and 84th

percentiles were identified as the thresholds for diagnosing a higher CV risk factors

clustering in Chinese boys and girls, respectively. The threshold of WC for boys is

consistent with the IDF recommendation (Zimmet et al., 2007) and the findings of

Maffeis et al. (2001) and Ng et al. (2007), but higher than the two previous studies in

Chinese boys. The threshold of WC for girls is lower than the IDF recommendation

and similar to the two studies in Chinese girls (Sung et al., 2007; Yan et al., 2008).

The thresholds of WC in the present study for both boys and girls are higher than

those for Caucasian children (Katzmarzyk et al., 2004).

The threshold of WC in previous studies varies according to the definition of CV risk


255

factors clustering. The two studies from Hong Kong (Sung et al., 2007) and the study

from Xinjiang province (Yan et al., 2008), proposed the 85th WC percentile as the

appropriate threshold for WC in predicting high CV risk factor clustering. However,

Katzmarzyk et al. (2004) proposed the 50‐57th percentiles for white and black boys

and girls. A number of factors may explain these differences. Firstly, WC is

ethnic‐independent (Misra, Wasir, & Vikram, 2005; Zhu et al., 2005), as well as the

susceptibility to CV risk factors (Jafar et al., 2005; Ke et al., 2009). Secondly, the

definition of CV risk factors clustering varied between studies with different

groupings of CV risk factors used. Sung et al. reported the 85th percentile was the

optimal threshold for diagnosing the presence of high CV risk clustering which was

defined as ≥4 of 6 risk factors, including elevated SBP and/or DBP, high TG, low

HDL‐C, high LDL‐C, glucose and insulin. However, if high CV risk factors clustering

was defined as ≥3 of the 6 risk factors, the cut‐offs were the 74th and 69th

percentiles for boys and girls, respectively. Furthermore, the level of each CV risk

factor to define an abnormal level of a marker also varies between studies. The 75th

percentile (Moreno et al., 2002), outer quintiles (Katzmarzyk et al., 2004), and 85th

percentiles (Sung et al., 2007) of CV risk factors of its own sample were used. In the

study by Sung et al. (2007) for example, the lower the level to define each individual

CV risk factors, the lower the WC percentile. Risk factors tend to cluster together for

individuals among both children and adults. The cluster of risk factors and

thresholds with the strongest predictive relationship to cardiovascular disease

should be identified for both clinical practice and prevention‐oriented research and

practice for the whole population, especially when developing universal cut‐off

points to predict it.

The present study also provides the age‐ and gender‐specific WHtR reference

percentiles for Chinese children. One of the main reasons for WHtR being

recommended as a good index to classify obesity in epidemiologic and clinic settings

for both children and adults is that WHtR is a relatively age‐ and

gender‐independent measure that could obviate the need for age‐and

gender‐related reference standards in children. However, our study challenges the


256

proposal and indicated that boys have a higher WHtR than girls at all ages. The

previous studies in 8135 British children aged 5 to 16 years (data collected from

boys in 1977 and girls in 1987) (McCarthy & Ashwell, 2006) , 14842 Hong Kong

Chinese children aged 6‐18 years (data collected in 2005‐2006) (Sung et al., 2008)

and 4811 Iran children aged 6‐18 years (data collected in 2003‐2004) found similar

results. WHtR in girls at all ages compared with boys reflects the gender differences

in both body shape and proportions. In addition to the gender difference, WHtR was

also not independent of age in our study. WHtR increased with age in boys and

decreased with age in girls. The previous studies also showed that WHtR during

childhood was influenced by age and growth (McCarthy & Ashwell, 2006; Sung et al.,

2008). However, previous studies indicated WHtR decreased with age in both boys

and girls. The age difference in mean WHtR reflected the divergence in the velocities

of growth in height and WC with age. Moreover, both boys and girls in the current

study had higher mean values for WHtR than British and Hong Kong Chinese

children at each age group except for the similar values to British girls at 6 and 7

years old. The difference in physical development and body composition may be due

to ethnicity and environmental factors.


257

Boys

0.41

0.42

0.43

0.44

0.45

0.46

0.47

0.48

6 7 8 9 10 11 12

Age (yr)

WHtR

Present study (2008)

Hong Kong (2005‐2006)

British (1977)

Girls

0.39

0.40

0.41

0.42

0.43

0.44

0.45

0.46

6 7 8 9 10 11 12

Age (yr)

WHtR

Present study (2008)

Hong Kong (2005‐2006)

British (1987)

Figure 5.7. Mean of waist‐to‐height ratio for children in three studies.

The current study also provided 91th and 94th percentiles of WHtR as the optimal

cut‐offs to predict the clustering of CV risk factors for boys and girl, respectively. The

values are higher than the 0.5 value proposed as the cut‐off point for adults by

Ashwell (1996). To our knowledge, only one study has reported WHtR cut‐offs for

predicting high CV risk factors in children, in which values ranged from 0.3 to 0.4 in

Iranian boys, and 0.40 to 0.46 in Iranian girls (Kelishadi et al., 2007). Both the

current study and that of Kelishadi (2007) challenge the universal WHtR cut‐offs of


258

0.5 for children of different age and gender. Further confirmation is required from

analyses involving different ethnic groups.

In conclusion, the current study provides reference values and percentiles for WC

and WHtR percentiles of Chinese children aged 6‐12 years living in the North and

South of the country. Optimal age‐ and gender‐specific cut‐off points for WC and

WHtR to predict CV risk factors clustering are also proposed. WHtR is not superior to

WC as a public health tool in Chinese children.


259

CHAPTER 6 GENERAL DISCUSSION AND CONCLUSIONS

6.1 GENERAL DISCUSSION

The rapid increasing prevalence of obesity and obesity‐related health risk, such as

cardiovascular disease, type 2 diabetes, place a heavy burden on individuals,

families and societies. Therefore, accurate and accessible assessment of body

composition is increasingly important for clinical and field settings to screen the

population at risk and monitor prevention and treatment efforts.

Obesity is characterized by an increased amount of body fat. BMI is the most

common index to define obesity because it is a convenient measure to perform in

both field and clinical settings. However, BMI is an index of weight and height and

provides no indication of body composition or adiposity per se. The inability of BMI

to differentiate levels of fatness and leanness among individuals makes it an

inappropriate index when using universal cut‐offs to define obesity across ethnic

groups as there are well established ethnic differences in the BMI‐%BF relationship

(Deurenberg, Deurenberg‐Yap & Guricci, 2002; Deurenberg et al., 1998; Freedman

et al., 2008; Navder et al., 2009).

The ethnic differences in body composition support the need to consider different

BMI levels to define obesity in different ethnic groups. For example, compared with

Caucasians, Asians have a 2‐4 units lower BMI at a given %BF (Deurenberg‐Yap et al.,

2002; Deurenberg et al., 1998; Gurrici et al., 1998; Kagawa et al., 2006). Therefore,

in 2004 WHO (2004) proposed new BMI cut‐offs of 23 kg/m2, 27.5 kg/m2, 32.5

kg/m2, and 37.5 kg/m2 for the Asian adult population, lower than the cut‐offs for the

Caucasian population. However, for children, a clear ethnic difference in the

BMI‐%BF relationship is yet to be established as varying growth rates and maturity

levels during the childhood and adolescence periods make work with the population

challenging. Some studies have shown similar ethnic differences in the BMI‐%BF

relationship between Asian and Caucasian children (Deurenberg, Bhaskaran et al.,

2003; Deurenberg, Deurenberg‐Yap et al., 2003; Duncan et al., 2009; Ehtisham et al.,


260

2005; Freedman et al., 2008; Mehta et al., 2002; Navder et al., 2009) while the small

number of studies to explore differences among Asian children and adolescents

from various origins (Duncan et al., 2009; Mehta et al., 2002). To our knowledge, the

current study is to explore the BMI‐%BF relationship among Asian children from

different origins and utilizing a large sample from West, South‐East and East Asian

countries. We found that Thai and Malay boys had significantly higher %BF than

Filipinos, 2.0% and 2.3% respectively, at the same BMI level and age. Thai girls had

approximately 2.0% higher %BF than their Chinese, Lebanese, Filipino and Malay

counterparts. These ethnic differences in the BMI‐%BF relationship among Asian

children should be considered when developing universal BMI cut‐offs to define

obesity in this population. Meanwhile, similar to other studies (Fernández et al.,

2003; Freedman et al., 2008; Swinburn et al., 1999; Wang & Bachrach, 1996), we

also found that the ethnic difference in the BMI‐%BF relationship varied by BMI. The

interaction between BMI and ethnicity in the estimation of body fatness indicates

the difficulty in identifying equivalent levels of fatness by simply adjusting BMI for

the average difference in fatness across ethnic groups.

The low sensitivity with the WHO and IOTF BMI classifications to detect cases of

obesity was evidenced in our first study. Two reasons may be advanced to explain

this result. On the one hand, BMI is not able to differentiate levels of fatness and

leanness among individuals (Freedman et al., 2005; Gallagher et al., 1996;

Roubenoff et al., 1995) and only explains 41‐88% and 14‐81% of the variance in %BF

or FM for boys and girls aged 8‐18 y, respectively (Maynard et al., 2001). Thus, BMI

only provides a poor to fair identification of those who are truly overweight and

obese as determined from objective measures of body fat (Sampei et al., 2001;

Wickramasinghe et al., 2005). Our results also indicate that BMI only explained

39.8‐78.5% of the variance in %BF in Asian children. However, on the other hand,

ethnic differences in the BMI‐%BF relationship also contributes to the low sensitivity.

Our first study confirmed the difference in the BMI‐%BF relationship between Asian

and Caucasians originates from childhood. For a given %BF, Asian children have a

3‐6 units lower BMI compared with Caucasians. Therefore, more than one‐third of


261

obese children in our study were not identified using WHO classification and more

than one half were not identified using the IOTF classification. The use of the

Chinese BMI classification increased the sensitivity by 19.7%, 18.1%, 2.2%, 3.3% and

11.3% for Chinese, Lebanese, Malay, Filipino and Thai girls. In summary, the results

of our first study indicate the necessity to use different BMI levels to define obesity

in Asian children.

To overcome the limitation of BMI to identify obesity, more accurate measures of

body fat assessment should be used in addition to BMI. A number of field and

laboratory measurements of body composition have been used in children and

adolescents, such as BIA, DXA, MRI, CT, densitometry, and the deuterium dilution

technique (Hills et al., 2001; Sopher, Shen & Pietrobelli, 2005). Among these

methods, BIA is one of the more appropriate for use in a range of settings including

community centers, field settings, clinics, and schools, because the measurement is

fast, non‐invasive, inexpensive, painless, requires minimal participant burden, does

not require a high level of technical skill. Despite the development of numerous BIA

prediction equations for children, most have been based on Caucasian subjects

(Heyward & Stolarczyk, 1996; Nielsen et al., 2007). As ethnic differences exist in

body composition (Malina, 2005), population‐specific BIA equations are essential. To

our knowledge, our second study is the first to develop and cross‐validate BIA

prediction equations in Asian children from different origins. The general equations

for the prediction of TBW and FFM developed on our complete sample indicates the

ethnic effect on the estimation of TBW and FFM from BIA resistance, a finding

consistent with other studies (Going et al., 2006; Haroun et al., 2010; Sluyter et al.,

2010) and confirmation of the need for population‐specific BIA equations. In

addition to ethnicity, RI, weight, age and sex explained approximately 88% of the

variance in equations with RMSE less than 2.0 kg. The R2 and RMSE values compare

favorably with those reported from other studies using the deuterium dilution

technique as the criterion method (Nielsen et al., 2007; Rush et al., 2003;

Wickramasinghe et al., 2008) and indicate the performance of developed equations

for Asian children is ideal predictive accuracy according to Lohman’s classification


262

system for prediction errors (Lohman, 1992). The equations should be

cross‐validated in a separate group to confirm the validity of the prediction equation.

The cross‐validation of our equations showed a good agreement of BIA and the

deuterium dilution technique for the estimation of TBW and FFM using the

Bland‐Altman method. However, in agreement with previous studies (Deurenberg et

al., 1991; McClanahan et al., 2009), there may be a potential systematic bias in

applying our equations with a trend for over‐prediction in participants of low FFM

and for under‐prediction in participants of high FFM. However, the accuracy of the

equations developed from all ethnic groups tested compared favorably with

BMI‐specific equations and ethnic‐specific equations, indicating the general

equations can be used as universal equations for the five ethnic groups.

Obesity can increase the risk of chronic diseases such as cardiovascular disease and

type 2 diabetes. However, the ethnic difference in cardiovascular morbidities and

diabetes can not be fully explained by BMI and total body fat (McAuley et al., 2002).

Some studies have indicated that central obesity is more closely related to health

risk in both children and adults than overall adiposity and the ethnic difference in

body fat distribution is a contributor to the ethnic difference in prevalence of

obesity‐related health risk among different ethnic groups (Berman et al., 2001;

Daniels et al., 1997; Ehtisham et al., 2005; Freedman et al., 1989; He et al., 2002;

Okosun, 2000). Some studies have demonstrated that Asian children and

adolescents have greater trunk depots than Caucasians (Ehtisham et al., 2005; He et

al., 2002; Malina et al., 1995; Novotny et al., 2006). However limited studies have

explored the differences in body fat distribution among Asian children from

different origins. Our study expands knowledge and understanding of the

differences in body fat distribution in children from different Asian origins. For

example, for a given fat mass, Chinese and Thai children had greater central fat

depots as determined by trunk/extremity SFT ratios and WC, than their Lebanese,

Malay and Filipino counterparts. These results support the importance of the

inclusion of a variety of screening measurements for risk stratification including WC

and skin fold thickness.


263

WC has been used as a simple and practical index for central obesity (Daniels et al.,

2000; Pouliot et al., 1994) and is recognized as a key component of the metabolic

syndrome in both children and adults (Alberti et al., 2006; Zimmet et al., 2007). A

strong positive correlation between WC and height exists throughout childhood and

into adulthood therefore, many countries have developed their own WC percentiles

for children and adolescents (Eisenmann, 2005; Fernández et al., 2004; Katzmarzyk,

2004; McCarthy et al., 2001). Furthermore, ethnic differences in WC remain even

after adjustment for height, weight or BMI (Ehtisham et al., 2005). Our study also

verified that Asian children from different origins also differ in terms of WC. Chinese

children had approximately 2 cm and 4 cm higher WC than their Lebanese and

Malay counterparts, respectively, at the same BMI. Due to the documented ethnic

difference in WC between Asians and Caucasians and among individuals from

different Asian origins, plus ethnic differences in susceptibility to the cardiovascular

morbidities (Jafar et al., 2005; Ke et al., 2009), age‐ and gender‐specific WC

percentiles were developed for Chinese children in our third study. Further, the 90th

percentile of WC for Chinese boys and the 84th percentile for Chinese girls were

proposed as the thresholds for the prediction of CV risk factors clustering. In

addition, some studies have showed that WHtR is a better predictor for

cardiovascular risk factors than WC because it is adjusted for the effect of height on

WC (Hara et al., 2002; Savva et al., 2000). Therefore, we also developed age‐ and

gender‐specific WHtR percentiles in our third study and the 91st and 94th

percentiles for WHtR were the thresholds for the prediction of cardiovascular risk

factors clustering for Chinese boys and girls, respectively. The thresholds of WHtR, in

agreement with Kelishadi’s study (2007) challenge the universal WHtR cut‐offs of 0.5

for children of different age and gender. Compared with the other two studies

conducted in Hong Kong (Sung et al., 2008) and Xinjiang province, China (Yan et al.,

2008), our study involved a more representative sample with children from South,

North‐east and North China. However, data are also required from a more extensive

nationally representative sample.

The metabolic syndrome describes the clustering of central obesity, dyslipidemia


264

(raised triglycerides and/or low‐ or high‐density lipoprotein), hyperinsulinemia,

impaired glucose tolerance and elevated blood pressure (Alberti et al., 2005) and is

a clear indicator of adult morbidity and all‐cause mortality (Alberti et al., 2005;

Haffner et al., 1992; Isomaa et al., 2001; Trevisan et al., 1998). Pediatric metabolic

syndrome can also increase cardiovascular risk (Ronnemaa et al., 1991) and can

track from childhood to adulthood (Duncan et al., 2004). Obese children have a

higher risk for metabolic syndrome and abnormalities (Cook et al., 2003; de Ferranti

et al., 2004; Reinehr et al., 2007; Weiss et al., 2004; Yoshinaga et al., 2005). Our

cross‐sectional analysis confirmed the previous findings in Chinese children by

indicating that the risk of metabolic syndrome in overweight and obese Chinese

children increased to 4‐6 times and 13‐26 times, respectively, varying by different

definitions of metabolic syndrome. Moreover, our longitudinal analysis also showed

that obese children without cardiovascular risk factors clustering had approximately

3 times greater risk to develop cardiovascular risk factors clustering after a 2‐y

follow‐up than normal‐weight children regardless of which obesity index was used

to define obesity. However, abdominal obesity, high TG and elevated blood pressure

were most frequent in both overweight and obese Chinese children, similar to other

studies in Asian populations (Yoshinaga et al., 2005; Zuoa et al., 2009) while

inconsistent with studies in other ethnic populations (Cook et al., 2003; de Ferranti

et al., 2004). These results imply the possibility of ethnic differences in susceptibility

to metabolic variables.

Although WC is used as an index to define central obesity in the definition of

metabolic syndrome (Alberti et al., 2006; Zimmet et al., 2007), decisions regarding

the best obesity index to predict cardiovascular risk factors clustering in children is

still subject to an ongoing debate (Garnett et al., 2007; Hara et al., 2002; Misra et al.,

2006; Savva et al., 2000). A number of factors may contribute to the inconsistency.

Firstly, ethnic differences in both body composition and the susceptibility to

cardiovascular risk factors might explain the inconsistent findings. However, few

studies have been conducted in Chinese children despite the knowledge that the

Chinese differ from Caucasian and other Asian children in BMI‐%BF relationship,


265

body fat distribution and prevalence of metabolic abnormities. This finding is

consistent with the present and previous studies (Deurenberg‐Yap et al., 2000;

Gurrici et al., 1999; Navder et al., 2009). It is important to stress that most of the

previous studies have been cross‐sectional. Our third study explored the ability of

four obesity indices, including BMI, %BF, WC, and WHtR, to predict the development

of cardiovascular risk factors clustering 2‐y later in Chinese children. We found

equivalent performance of the four indices in prediction of the development of

cardiovascular risk factors clustering in Chinese children aged 7‐12 y. The failure of

WC and WHtR to be superior predictors of cardiovascular risk factors clustering in

our study is similar to another longitudinal (Garnett et al., 2007) and cross‐sectional

study in Caucasian children and adolescents (Plachta‐Danielzik et al., 2008).

Secondly, various subsets of risk factors included in the definition of metabolic risk

clustering in each study is another contributor to the inconsistency because the

various obesity indices showed different associations with various metabolic risk

factors (McCarthy, 2006; Savva et al., 2000). Our study also showed that the BMI

appeared to be more closely related to blood pressure while WC tended to be more

closely related to TG and HDL‐C in Chinese children. Therefore, subset of metabolic

risk factors which is more predictive for the later incidence of cardiovascular disease

or diabetes best is required for determining the best obesity index to screen the

population at high risk.

To our best knowledge, the current studies are the first to explore the ethnic

difference in total body adiposity and body fat distribution across Asian children

from 5 countries with a large study sample, and also the first study to cross‐validate

BIA across Asian children from different origins with the deuterium dilution

technique as the criterion. Moreover, our studies proposed the threshold of WC

and WHtR for the prediction of the metabolic risk factors clustering among Chinese

children in a large representative sample and explored the ability of different

obesity indices to predict future health risk in a follow‐up study. However, there are

some limitations in our studies. Firstly, the cohort in study 1 and study 2 was

purposively selected in each ethnic group and was a non‐random population sample.


266

However, participants purposively recruited from a wide range of BMI can provide

us with a better understanding of the ethnic differences inBMI‐%BF relationship and

body fat distribution. Moreover, the BIA prediction equations developed from the

population with wide range of BMI in the current thesis can be used not only in

normal‐weight children but also in severely obese children. Secondly, the cut‐offs

for defining the metabolic abnorlities were not population ‐specific which might not

be appropriate due to the ethnic differences in metabolic profiles. However, no

population‐specific cut‐offs of metabolic profiles for Asian children are available.

Therefore, futher research should be conducted to develop these cut‐offs for Asian

children.

6.2 CONCLUSIONS ANS IMPLICATIONS

Based on the key research questions, the following findings were concluded:

1. There were ethnic differences in the relationship between BMI‐%BF among

Asian pre‐pubertal children aged 8‐10 y from China, Lebanon, Malaysia, The

Philippines and Thailand.

2. BMI cut‐offs for defining obesity proposed by WHO and IOTF were not

appropriate for Asian children.

3. Ethnic differences in body fat distribution among Asian pre‐pubertal children

aged 8‐10 y from China, Lebanon, Malaysia and Thailand were found.

4. BIA was a valid method for the assessment of body composition across Asian

children from China, Lebanon, Malaysia, The Philippines and Thailand.

5. Obese Chinese children had a higher risk of developing the metabolic syndrome

and abnormities than normal‐weight children.

6. No superiority of either WC or WHtR to predict cardiovascular risk factors

clustering in the 2‐y follow‐up study in Chinese children was found.

7. WC and WHtR percentiles and optimal age‐ and gender‐specific thresholds for


267

the prediction of cardiovascular risk factors clustering were developed for

Chinese children aged 6‐12 y.

Findings from the current thesis have the potential to contribute to theory and

practice in the following ways:

1. It is necessary to develop BMI cut‐offs to define obesity for Asian children. For

Asian children, the BMI cut‐off points should be lower compared with those for

Caucasian population. Otherwise, many children who are truly overweight and

obese and at the risk of health problems accompanying overweight and obesity

may not be identified. Moreover, the reference population for developing BMI

cut‐offs for Asian children should be recruited from various countries across the

Asian region.

2. It is import to include a variety of screening measurements to classify subjects at

risk of obesity and obesity‐related health problems in both community and

clinical settings in addition to BMI, such as WC and skin fold thickness. Moreover,

population‐specifice thresholds for other obesity indices including WC should be

developed.

3. Our population‐specific BIA prediction equation provides an appropriate tool for

the accurate assessment of body composition among Asian children in a

multi‐country study. It also provides the opportunity to compare the prevalence

of obesity based on %BF rather than BMI between Asian countries using the BIA

method which is relatively inexpensive, simple to use and not time consuming.

Moreover, these equations can be employed by manufactures whose products

are designed for the Asian population.

4. More efforts should be made for pediatric obesity control and prevention in

China. Although the overall prevalence of the metabolic syndrome in Chinese

children is lower than in American counterparts, the prevalence rates in

overweight and obese children is similar in both countries. This means that with

the rapid increase in obesity in China, the prevalence of the metabolic syndrome

is becoming more common. Due to the rapid nutritional and lifestyle transition


268

being experienced in China, it is a critical period for the prevention of pediatric

obesity.

5. Although BMI has limitations to define obesity, it still has the similar ability to

predict metabolic risk factors along with WC and WHtR. The developed cut‐off

points of WC and WHtR for Chinese children, combined with BMI, will help to

better screen this population for health risk in both community and clinical

practice.


269

REFERENCES

Afghani, A, & Goran, MI (2006). Racial differences in the association of subcutaneous

and visceral fat on bone mineral content in prepubertal children. Calcified

Tissue Int, 79, 383‐388.

Alberti, K, Zimmet, P, & Shaw, J (2006). Metabolic syndrome ‐ a new world‐wide

definition. A consensus statement from the international diabetes federation.

Diabetic Med, 23(5), 469‐480.

Alberti, K, Zimmet, P, & WHO Consultation. (1998). Definition, diagnosis and

classification of diabetes mellitus and its complications. Part 1: diagnosis and

classification of diabetes mellitus. Provisional report of a WHO Consultation.

Diabetic Med, 15, 539‐553.

Alberti, KGMM, Zimmet, P, & Shaw, J (2005). The metabolic syndrome ‐ a new

worldwide definition. Lancet, 366(9491), 1059‐1062.

Arslanian, S, Suprasongsin, C, & Janosky, JE (1997). Insulin secretion and sensitivity

in black verus white prepubertal healthy children. J Clin Endocrinol Metab,

82, 1923‐1927.

Ashby, RL, Ward, KA, Mughal, MZ, & Adams, JE (2003). Age‐related changes in

metacarpal morphometry and areal bone mineral density in children

assessed by DXR. Osteoporosis Int, 14, S25‐S25.

Ashwell, M (1994). Obesity in men and women. Int J Obes Relat Metab Disord, 18,

S1‐7.

Ashwell, MA, LeJeune, SRE, & McPherson, B (1996). Ratio of waist circumference to

height may be better indicator of need for weight management. BMJ, 312,

377.

Bacha, B, Saad, R, Gungor, N, & Arslanian, SA (2006). Are Obesity‐Related Metabolic

Risk Factors Modulated by the Degree of Insulin Resistance in Adolescents?.

Diabetes Care, 29, 1599‐1604.

Bacha, F, Saad, R, Gungor, N, Janosky, J, & Arslanian, SA (2003). Obesity, regional fat

distribution, and syndrome X in obese black versus white adolescents: Race

differential in diabetogenic and atherogenic risk factors. J Clin Endocrinol


270

Metab, 88(6), 2534‐2540.

Barlow, SE, & Dietz, WH (1998). Obesity evaluation and treatment: expert

committee recommendations. The Maternal and Children Health Bureau,

Health Resources and Services Adminstration, and the Department of Health

and Human Services. Pediatrics, 102, E29.

Baumgartner, RN, Chumlea, WC, & Roche, AF (1989). Estimation of body

composition from bioelectrical impedance of body segments. Am J Clin Nutr,

50, 221‐226.

Baumgartner, RN, Ross, R, & Heymsfield, SB (1998). Does adipose tissue infuence

bioelectric impedance in obese men and women? J Appl Physiol, 84,

257‐262.

Bell, N, McClure, P, Hill, R, & Davies, P (1998). Assessment of foot‐to‐foot

bioelectrical inpedance analysis for the prediction of total body water. Eur J

Clin Nutr, 52, 856‐859.

Berman, DM, Rodrigues, LM, Nicklas, BJ, Ryan, AS, Dennis, KE, & Goldberg, AP

(2001). Racial disparities in metabolism, central obesity, and sex

hormone‐binding globulin in postmenopausal women. J Clin Endocrinol

Metab, 86, 97‐103.

Bjorntorp, P (1992). Abdominal fat distribution and disease: an overriew of

epidemiological data. Ann Med, 24, 15‐18.

Bland, J, & Altman, D (1986). Statistical methods for assessing agreement between

two methods of clinical measurement. Lancet, 8, 307‐310.

Booth, ML, Dobbins, T, Okely, AD, Denney‐Wilson, E, & Hardy, LL (2007). Trends in

the prevalence of overweight and obesity among young Australians, 1985,

1997, and 2004. Obesity, 15, 1089‐1095.

Bosy‐Westphal, A, Geisler, C, Onur, S, Korth, O, Selberg, O, Schrezenmeir, J, et al.

(2006). Value of body fat mass vs anthropometric obesity indices in the

assessment of metabolic risk factors. Int J Obes, 30, 475‐483.

Bracco, D, Thiébaud, D, Chioléro, RL, Landry, M, Burckhardt, P, & Schutz, Y (1996).

Segmental body composition assessed by bioelectrical impedance analysis

and DEXA in humans. J Appl Physiol, 81, 2580‐2587.


271

Bray, GA, Delany, JP, Harsha, DW, Volaufova, J, & Champagne, CM (2001). Body

composition of African American and white children: a 2‐year follow‐up of

the BAROC study. Obes Res, 9, 605‐621.

Bray, GA, DeLany, JP, Volaufova, J, Harsha, DW, & Champagne, C (2002). Prediction of

body fat in 12‐y‐old African American and white children: evaluation of

methods. Am J Clin Nutr, 76, 980‐990.

Bunc, V (2001). Predicton equations for the determination of body composition in

children useing bioimpedence anyalis. (Vol. 44). Basel, Karger: Med Sport Sci.

Burke, Y, Beilin, LJ, Simmer, K, Oddy, WH, Blake, KV, Doherty, D, et al. (2005).

Predictors of body mass index and associations with cardivascular risk factors

in Australian children: a prospective cohort study. Int J Obes, 29, 15‐23.

Butte, NF, Hopkinson, JM, Wong, W.W, Smith, EO, & Ellis, KJ (2000). Body

composition during the first 2 years of life: An updated reference. Pediatr Res,

47, 578‐585.

Côté, KD, & Adams, WC (1993). Effect of bone density on body composition

estimates in young adult black and white women. Med Sci Sports Exerc, 23,

290‐296.

Caton, JR, Molé, PA, Adams, WC, & Heustis, DC (1988). Body composition analysis by

bioelectrical impedance: effect of skin temperature. Med Sci Sports Exerc, 20,

489‐491.

Chandalia, M, Abate, N, Garg, A, Stray‐Gundersen, J, & Grundy, SM (1999).

Relationship between generalized and upper body obesity to insulin

resistance in Asian Indian men. J Clin Endocrinol Metab, 84, 2329‐2335.

Chang, CJ, Wu, CH, Chang, CS, Yao, WJ, Yang, YC, Wu, JS, et al. (2003). Low body

mass index but high percent body fat in Taiwanese subjects: implications of

obesity cutoffs. Int J Obes, 27, 253‐259.

China's National Group on Student's Constitution and Health Survey. (2000). Study

on the measurement of skinfold thickness and estimated of body

composition in Chinese primary and secondary school students. Chin J Prev

Med, 34, 212‐214.

China's National Group on Student's Constitution and Health Survey. (2007). Report


272

on the physical fitness and health surveillance of Chinese school students.

Beijing: Higher Education Publishing House.

Chu, NF, Wang, DJ, Shieh, SM, & Rimm, EB (2000). Plasma leptin concentrations and

obesity in relation to insulin resistance syndrome components among school

children in Taiwan‐ the Taipei children heart study. Int J Obes Relat Metab

Disord, 24 1265‐1271.

Chumlea, W, Schubert, C, Reo, N, Sun, S, & Siervogel, R (2005). Total body water

volume for white children and adolescents and anthropometric prediction

equations: the Fels Longitudinal Study. Kidney Int, 68, 2317.

Chumlea, WC, Guo, S, Kuczmarski, R, Flegal, KM, Johnson, CL, Heymsfield, SB, et al.

(2002). Body composition estimates from NHANES III bioelectrical

impedance data. Int J Obes, 26, 1596‐1609.

Chumlea, WC, & Guo, SS (1994). Bioelectrical impedance and body composition:

present status and future directions. Nutr Rev, 52, 123‐131.

Chung, SC, Song, MY, Shin, HD, Kim, DY, He, Q, Heshka, S, et al. (2005). Korean and

Caucasian overweight premenopausal women have different relationship of

body mass index to percent body fat with age. J Appl Physiol, 99, 103‐107.

Cole, T, Bellizzi, M, Flegal, K, & Dietz, WH (2000). Establishing a standard definition

for child overweight and obesity worldwide: international survey. BMJ, 320,

1‐6.

Cole, TJ, Freeman, JV, & Preece, MA (1995). Body mass index reference curves for

the UK, 1990. Arch Dis Child, 73, 25‐29.

Conway, J, Yanovski, S, Avila, N, & Hubbard, V (1995). Visceral adipose tissue

differences in black and white women. Am J Clin Nutr, 61, 765‐771.

Cook, S, Weitzman, M, Auinger, P, Nguyen, M, & Dietz, WH (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, 821‐827.

Cossrow, N, & Falkner, B (2004). Race/ethnic issues in obesity and obesity‐related

comorbidities. J Clin Endocrinol Metab, 89, 2590‐2594.

Craig, P, Halavatau, V, Comino, E, & Caterson, I (2001). Differences in body


273

composition between Tongans and Australians: time to rethink the healthy

weight ranges? Int J Obes, 25, 1806‐1814.

Daniels, S, Morrison, J, Sprecher, D, Khoury, P, & Kimball, T (1999). Association of

body fat distribution and cardiovascular risk factors in children and

adolescents. Circulation, 99, 541‐545.

Daniels, SR, Khoury, PR, & Morrison, JA (1997). The utility of body mass index as a

measure of body fatness in children and adolescents: Differences by race and

gender. Pediatrics, 99, 804‐807.

Daniels, SR, Khoury, PR, & Morrison, JA (2000). Utility of different measures of body

fat distribution in children and adolescents. Am J Epidemiol, 152, 1179‐1184.

Dao, HH, Frelut, ML, Oberlin, F, Peres, G, Bourgeois, P, & Navarro, J (2004). Effects of

a multidisciplinary weight loss intervention on body composition in obese

adolescents. Int J Obes, 28, 290‐299.

Davies, PS, & Preece, MA (1988). The prediction of total body water using

bioelectrical impedance in children and adolescents. Ann Hum Biol, 15,

237‐240.

de Ferranti, S, Gauvreau, K, Ludwig, D, Neufeld, E, Newburger, J, & Rifai, N (2004).

Prevalence of the metabolic syndrome in American adolescents: findings

from the Third National Health and Nutrition Examination Survey. Circulation,

110, 2494‐2497.

de Onis, MOA, Borghi, E, Siyam, A, Nishida, C, & Siekmann, J (2007). Development of

a WHO growth reference for school‐aged children and adolescents. Accessed

15 March, 2008, from

http://www.who.int/growthref/growthref_who_bull/en/index.html.

Després, JP. Moorjani, S, Lupien, PJ, Tremblay, A, Nadeau, A, & Bouchard, C (1990).

Regional distribution of body fat, plasma lipoproteins, and cardiovascular

disease. Arteriosclerosis, 10, 495‐551.

Després, JP (2007). Cardiovascular disease under the influence of excess visceral fat.

Crit Pathw Cardiol, 6, 51‐59.

Deurenberg‐Yap, M, Chew, SK, & Deurenberg, P (2002). Elevated body fat

percentage and cariovascular risks at low body mass index levels among


274

Singaporean Chinese, Malays and Indians. Obes Rev, 3, 209‐215.

Deurenberg‐Yap, M, Schmidt G, van Staveren WA, & Deurenberg, P (2000). The

paradox of low body mass index and high body fat percentage among

Chinese, Malays and Indians in Singapore. Int J Obes Relat Metab Disord, 24,

1011‐1017.

Deurenberg‐Yap, M, Schmidt, G, van Staveren, WA, Hautvast, J, & Deurenberg, P

(2001). Body fat measurement among Singaporean Chinese, Malays and

Indians: a comparative study using a four‐compartment model and different

two‐compartment models. Brit J Nutr, 85, 491‐498.

Deurenberg, P, Bhaskaran, K, & Lian, PL (2003). Singaporean Chinese adolescents

have more subcutaneous adipose tissue than Dutch Caucasians of the same

age and body mass index. Asia Pac J Clin Nutr., 12, 261‐265.

Deurenberg, P, & Deurenberg‐Yap, M (2001). Differences in body‐composition

assumptions across ethnic groups: practical consequences. Curr Opin Clin

Nutr Metab Care, 4, 377‐383.

Deurenberg, P, & Deurenberg‐Yap, M (2003). Validity of body composition methods

across ethnic population groups. Acta Diabetol, 40(S1), S246‐S249.

Deurenberg, P, Deurenberg‐Yap, M, Foo, LF, Schmidt, G, & Wang, J (2003).

Differences in body composition between Singapore Chinese, Beijing Chinese

and Dutch children. Eur J Clin Nutr, 57, 405‐409.

Deurenberg, P, Deurenberg‐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 Res, 3, 141‐146.

Deurenberg, P, Deurenberg‐Yap, M, & Schouten, F (2002). Validity of total and

segmental impedance measurements for prediction of body composition

across ethnic groups. Eur J Clin Nutr, 56, 214‐220.

Deurenberg, P, Ge, K, Hautvast, J, & Wang, J (1997). BMI as predictor for body fat:

comparison between Chinese and Dutch adult subjects. Asia Pac J Clin Nutr,

6, 102‐105.

Deurenberg, P, Kusters, CSL, & Smit, HE (1990). Assessment of body composition by

bioelectrical impedance of children and young adults is strongly age


275

dependent. Eur J Clin Nutr, 44, 261‐268.

Deurenberg, P, Tagliabue, A, & Schouten, . (1995). Multi‐frequency impedance for

the prediction of extracellular water and total body water. Brit J Nutr, 73,

349‐358.

Deurenberg, P, van der Kooy, K, & Leenen, R (1989). Differences in body impedance

when measured with different instrumetns (letter). Eur J Clin Nutr, 43,

885‐886.

Deurenberg, P, van der Kooy, K, Leenen, R, Weststrate, JA, & Seidell, JC (1991). Sex

and age specific prediction formulas for estimating body composition from

bioelectrical impedance: a cross‐validation study. Int J Obes, 15, 17‐25.

Deurenberg, P, Weststrate, J, & Seidell, J (1991). Body mass index as a measure of

body fatness: age‐ and sex‐specific prediction formulas. Brit J Nutr, 65,

105‐114.

Deurenberg, P, Weststrate, JA, Paymans, I, & van der Kooy, K (1988). Factors

affecting bioelectrical impedance measurements in humans. Eur J Clin Nutr,

42, 1017‐1022.

Deurenberg, P, Deurenberg‐Yap, M, & van Staveren, WA (1998). Body mass index

and percent body fat: a meta analysis among different ethnic groups. Int J

Obes, 22, 1164‐1171.

Deurenberg, P, Deurenberg‐Yap, M, Wang, J, Lin, FP, & Schmidt, G (1999). The impact

of body build on the relationship between body mass index and percent

body fat. Int J Obes, 23, 537‐542.

Dietz, WW (1994). Critical periods in childhood for the development of obesity. Am J

Clin Nutr, 59, 955‐959.

Duncan, GE, Li, SM, & Zhou, XH (2004). Prevalence and trends of a metabolic

syndrome phenotype among U.S. adolescents, 1999‐2000. Diabetes Care, 27,

2438‐2443.

Duncan, JS, Duncan, EK, & Schofield, G (2009). Accuracy of body mass index (BMI)

thresholds for predicting excess body fat in girls from five ethnicities. Asia

Pac J Clin Nutr, 18, 404‐411.

Ehtisham, S, Crabtree, N, Clark, P, Shaw, N, & Barrett, T (2005). Ethnic differences in


276

insulin resistance and body composition in United Kingdom adolescents. J

Clin Endocrinol Metab, 90, 3963‐3969.

Ehtisham, S, Hattersley, AT, Dunger, DB, & Barrett, TG (2004). First UK survey of

paediatric type 2 diabetes and MODY. Arch Dis Child, 89, 526‐529.

Eisenmann, JC (2005). Waist circumference percentiles for 7‐ to 15‐year‐old

Australian children. Acta Paediatrica, 94, 1182‐1185.

Ellis, KJ (1990). Reference man and woman more fully characterized: variations of

the basis of body size, age, sex, and race. Biol Trace Elem Res, 26, 385‐400.

Ellis, KJ (1997). Body composition of a young, multiethnic, male population. Am J

Clin Nutr, 66, 1323‐1331.

Ellis, KJ (2000). Human Body Composition: In Vivo Methods. Physiol Rev 80,

649‐680.

Ellis, KJ, Abrams, SA, & Wong, GW (1997). Body composition of a young, multiethnic

female population. Am J Clin Nutr, 65, 724‐731.

Ellis, KJ, Bell, SJ, Chertow, GM, Chumlea, WC, Knox, DP, Lukaski, HC, et al. (1999).

Bioelectrical impedance methods in clinical research: a follow‐up to the NIH

Technology Assessment Conference. Nutrition, 15, 874‐880.

Eston, RG, Cruz, A, Fu, F, & Fung, LM (1993). Fat‐free mass estimation by

bioelectrical impedance and anthropometrc techniques in Chinese children. J

Sports Sci, 11, 241‐247.

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 (NCEP) Expert Panel on Detection, Evaluation,

and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III).

JAMA 285, 2486‐2497.

Fernández, JR, Heo, MS, Heymsfield, SB, Pierson, RN, Pi‐Sunyer, FX, Wang, ZM, et al.

(2003). Is percentage body fat differentially related to body mass index in

Hispanic Americans, African Americans, and European Americans? Am J Clin

Nutr, 77, 71‐75.

Fernández, JR, Redden, DT, Pietrobelli, A, & Allison, DB (2004). Waist circumference

percentiles in nationally representative samples of African‐American,


277

European‐American, and Mexican‐American children and adolescents. J

Pediatr, 145, 439‐444.

Fomon, SJ, Haschke, F, Zeigler, EE, & Nelson, SE (1982). Body composition of

reference children from birth to age 10 years. Am J Clin Nutr, 35, 1169‐1175.

Ford, ES (2005). Prevalence of the metabolic syndrome defined by the International

Diabetes Federation among adults in the U.S. Diabetes Care, 28, 2745‐2749.

Fox, KR, Peters, DM, Sharpe, P, & Bell, M (2000). Assessment of abdominal fat

development in young adolescents using magnetic resonance imaging. Int J

Obes, 24, 1653‐1659.

Freedman, DS, Srinivasan, S, Harsha, D, Webber, L, & Berenson, G (1989). Relation of

body fat patterning to lipid and lipoprotein concentrations in children and

adolescents: the Bogalusa Heart Study. Am J Clin Nutr, 50, 930‐939.

Freedman, DS, Wang, J, Thornton, J, Mei, Z, Pierson Jr, RN, Dietz, W, et al. (2008).

Racial/ethnic differences in body fatness among children and adolescents.

Obesity (Silver Spring), 16, 1105‐1111.

Freedman, DS, & Sherry, B (2009). The validity of BMI as an indicator of body fatness

and risk among children. Pediatrics, 124, S23‐S34.

Freedman, DS, Khan, LK, Dietz, WH, Srinivasan, SR, & Berenson, GS (2001).

Relationship of Childhood Obesity to Coronary Heart Disease Risk Factors in

Adulthood: The Bogalusa Heart Study. Pediatrics, 108, 712‐718.

Freedman, DS, Khan, L.K, Serdula, MK, Dietz, WH, Srinivasan, SR, & Berenson, GS

(2005). The relation of childhood BMI to adult adiposity: the Bogalusa Heart

Study. Pediatrics, 115, 22‐27.

Freedman, DS, Wang, J, Maynard, LM, Thornton, JC, Mei, Z, Pierson, RN, et al. (2005).

Relation of BMI to fat and fat‐free mass among children and adolescents. Int

J Obes, 29, 1‐8.

Fuller, NJ, & Elia, M (1989). Potential use of bioelectrical impedance of the 'whole

body' and of body segments for the assessment of body compostion:

comparison with densitometry and antropometry. Eur J Clin Nutr, 43,

779‐791.

GómezDíaz, RA, Martínez‐Hernández, AJ, Aguilar‐Salinas, CA, Villante, R,


278

López‐Alarcón, M, Jiménez‐Villarruel, M, et al. (2005). Percentiles

distribution of the waist circumference among Mexican pre‐adolescents of a

primary school in Mexico city. Diabetes Obes Metab, 7, 716‐721.

Gallagher, D, Visser, M, Sepúlveda, D, Pierson, RN, Harris, T, & Heymsfield, SB (1996).

How useful is body mass index for comparison of body fatness across age,

sex, and ethnic groups? Am J Epidemiol, 143, 228‐239.

Garan, MI, & Gower, BA (1999). Relation between visceral fat and disease risk in

children and adolescents. Am J Clin Nutr, 70, 149S‐156S.

Garnett, SP, Baur, LA, Srinivasan, S, Lee, JW, & Cowell, CT (2007). Body mass index

and waist circumference in midchildhood and adverse cardiovascular disease

risk clustering in adolescence. Am J Clin Nutr, 86, 549‐555.

Garrow, JS, & Webster, JD (1985). Quetelets's index (W/H2) as a measure of fatness.

Int J Obes, 9, 147‐153.

Gaskin, PS, & Walker, SP (2003). Obesity in a cohort of black Jamaican children as

estimated by BMI and other indices of adiposity. Eur J Clin Nutr, 57, 420‐426.

Gilsanz, V, Skaggs, DL, Kovanlikaya, A, Sayre, J, Loro, ML, Kaufman, F, et al. (1998).

Differential Effect of Race on the Axial and Appendicular Skeletons of

Children. J Clin Endocrinol Metab, 83, 1420‐1427.

Going, S, Nichols, J, Loftin, M, Stewart, D, Lohman, T, Tuuri, G, et al. (2006).

Validation of bioelectrical impedance analysis (BIA) for estimation of body

composition in Black, White and Hispanic adolescent girls. Int J Body Compos

Res, 4, 161‐167.

Golley, R, Magarey, A, Steinbecks, K, Baur, L, & Daniels, L (2006). Comparison of

metabolic syndrome prevalence using six different definitions in overweight

pre‐pubertal children enrolled in a weight management study. Int J Obes, 30,

853‐860.

Goodman, E, Daniels, SR, Morrison, JA, 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,

445‐451.


279

Goran, MI, Nagy, TR, Treuth, MS, Trowbridge, C, Dezenberg, C, McGloin, A, et al.

(1997). Visceral fat in white and African American prepubertal children. Am J

Clin Nutr, 65, 1703‐1708.

Goulding, A, Taylor, RW, Jones, IE, Lewis‐Barned, NJ, & Williams, SM (2003). Body

composition of 4‐ and 5‐year‐old New Zealand girls: a DXA study of initial

adiposity and subsequent 4‐year fat change. Int J Obes, 27, 410‐415.

Group of China Obesity Task Force (2004). Body mass index reference norm for

screening overeweight and obesity in Chinese children and adolescents. Chin

J Epidemiol, 25, 97‐102.

Gudivaka, R, Schoeller, D, & Kushner, RF (1996). Effect of skin temperature on

multifrequency bioelectrical impedance analysis. J Appl Physiol, 81, 838‐845.

Guillaume, M (1999). Defining obesity in childhood: current practice. Am J Clin Nutr,

70, S126‐S130.

Guo, S, Chumlea, W, Roche, A, & Siervogel, R (1997). Age‐ and maturity‐related

changes in body composition during adolescence into adulthood: The Fels

Longitudinal Study. Int J Obes, 21, 1167‐1175.

Gurrici, S, Hartriyanti, Y, Hautvast, J, & Deurenberg, P (1998). Relationship between

body fat and body mass index: differences between Indonesians and Dutch

Caucasians. Eur J Clin Nutr, 52, 779‐783.

Gurrici, S, Hartriyanti, Y, Hautvast, J, & Deurenberg, P (1999). Differences in the

relationship between body fat and body mass index between two different

Indonesian ethnic groups: The effect of body build. Eur J Clin Nutr, 53,

468‐472.

Haffner, SM, Valdez, RA, Hazuda, HP, Mitchell, BD, Morales, PA, & Stern, MP (1992).

Prospective analysis of the insulin‐resistance syndrome (syndrome X).

Diabetes, 41, 715‐722.

Han, TS, & Lean, ME (1998). Self‐reported waist circumference compared with the

'Waist Watcher' tape‐measure to identify individuals at increased health risk

through intra‐abdominal fat accumulation. Br J Nutr, 80, 81‐88.

Hara, M, Saitou, E, Iwata, F, Okada, T, & Harada, K (2002). Waist‐to‐height ratio is the

best predictor of cardiovascular disease risk factors in Japanese


280

schoolchildren. J Atheroscler Thromb, 9, 127‐132.

Haroun, D, Taylor, SJC, Viner, RM, Hayward, RS, Darch, TS, Eaton, S, et al. (2010).

Validation of bioelectrical impedance analysis in adolescents across different

ethnic groups. Obesity (Sliver Spring), 18, 1252‐1259.

Harsha, DW, Voors, AW, & Berenson, GS (1980). Racial differences in subcutaneous

fat patterns in children aged 7‐15 years. Am J Phys Anthropol, 53, 333‐337.

Hatipoglu, N, Ozturk, A, Mazicioglu, MM, Kurtoglu, S, Seyhan, S, & Lokoglu, F (2008).

Waist circumference percentiles for 7‐ to 17‐year‐old Turkish children and

adolescents. Eur J Pediatr, 167, 383‐389.

Hayes, PA, Sowood, PJ, Belyavin, A, Cohen, J, & Simth, FW (1988). Sub‐cutaneous fat

thickness measured by magnetic resonance imaging, ultrasound, and calipers.

Med Sci Sports Exerc, 20, 303‐309.

He, Q, Horlick, M, Thornton, J, Wang, J, Pierson Jr, RN, Heshka, S et al. (2004).

Sex‐specific fat distribution is not linear across pubertal groups in a

multiethnic study. Obes Res, 12, 725‐733.

He, Q, Horlick, M, Thornton, J, Wang, J, Pierson Jr, RN, Heshka, S, et al. (2002). Sex

and race differences in fat distribution among Asian, African‐American, and

Caucasian prepubertal children. J Clin Endocrinol Metab, 87, 2164‐2170.

Henriksson, KM, Lindblad, U, Agren, B, Nilsson‐Ehle, P, & Rastam, L (2001).

Associations between body height, body composition and cholesterol levels

in middle‐aged men. The coronary risk factor study in southern Sweden

(CRISS). Eur J Epidemiol, 17, 521‐526.

Herd, SL, Gower, BA, Dashti, N, & Goran, M (2001). Body fat, fat distribution and

serum lipids, lipoproteins and apolipoproteins in African‐American and

Caucasian‐American prepubertal children. Int J Obes, 25, 198‐204.

Heymsfield, SB, Gallagher, D, Visser, M, Nunez, C, & Wang, Z (1995). Measurement

of skeletal muscle: laboratory and epidemiological methods. J Gerontol, 50,

23‐29.

Heymsfield, SB, Lohman, TG, Wang, Z, & Going, SB (2005). Human body composition

(Second ed.). Champaign, IL: Human Kinetics.

Heyward, V, & Stolarczyk, L (1996). Applied body composition assessment:


281

Champaign, IL: Human Kinetics.

Heyward, VH (1998). Practical body composition assessment for children, adults,

and older adults. Int J Sports Nutr 8, 285‐307.

Hills, AP, Lyell, L, & Byrne, NM (2001). An evaluation of the methodology for the

assessment of body composition in children and adolescents. Jürimäe T &

Hills AP(Eds.), Body composition assessment in children and adolescents.

Basel: S. Karger AG.

Horlick, M, Arpadi, S, Bethel, J, Wang, J, Moye Jr, J, Cuff, P et al. (2002). Bioelectrical

impedance analysis models for prediction of total body water and fat‐free

mass in healthy and HIV‐infected children and adolescents. Am J Clin Nutr, 76,

991‐999.

Horlick, M, Thornton, J, Wang, J, Levine, LS, Fedun, B, & Pierson Jr, RN (2000). Bone

mineral in prepubertal children: gender and ethnicity. J Bone Miner Res, 15,

1393‐1397.

Houtkooper, LB, Going, S, Lohman, TG, Roche, AF, & Van Loan, M (1992).

Bioelectrical impedance estimation of fat‐free body mass in children and

youth: a cross‐validation study. J Appl Physiol, 72, 366‐373.

Houtkoper, LB, Lohman, TG, Going, S, & Hall, MC (1989). Validity of bioelectrical

impedance for body composition assessment in children. J Appl Physiol, 66,

814‐821.

Hsieh, SD, & Muto, TM (2005). The superiority of waist‐to‐height ratio as an

anthropometric index to evaluate clustering of coronary risk factors among

non‐obese men and women. Prev Med, 40, 216‐220.

Hsieh, SD, Yoshinaga, H, & Muto, T (2003). Waist‐to‐height ratio, a simple and

practical index for assessing central fat distribution ans metabolic risk in

Japanese men and women. Int J Obes Relat Metab Disord, 27, 610‐616.

Huang, TTK, Johnson, MS, Figueroa‐Colon, R, Dwyer, JH, & Goran, MI (2001). Growth

of visceral fat, subcutaneous abdominal fat, and total body fat in children.

Obes Res, 9, 283‐289.

Hughes, K, Aw, T, Kuperan, P, & Choo, M (1997). Central obesity, insulin resistance,

syndrome X, lipoprotein (a), and cardiovascular risk in Indians, Malays, and


282

Chinese in Singapore. J Epidemiol Commun Health, 51, 394‐399.

Huoutkooper, L, Lohman, T, Going, S, & Howell, W (1996). Why bioelectrical

impedance analysis should be used for estimating adiposity. Am J Clin Nutr,

64, 436S‐448S.

Huxley, R, James, WPT, Barzi, F, Patel, JV, Lear, SA, Suriyawomgpaisal, P, et al. (2008).

Ethnic comparisons to the cross‐sectional relationships between measures of

body size with diabetes and hypertension. Obes Rev, 9(Suppl1), 53‐61.

International Atomic Energy Agency (2009). Assessment of body composition and

total energy expenditure in humans using stable isotope techniques. Vienna:

International Atomic Energy Agency.

Isomaa, B, Almgren, P, Tuomi, T, Forsén, B, Lahti, K, Nissén, M, et al. (2001).

Cardiovascular Morbidity and Mortality Associated With the Metabolic

Syndrome. Diabetes Care, 24, 683‐689.

Jürimäe, T, & Hills, AP (2001). Body composition assessment in children and

adolescents. (Vol. 44): Medicine and Sport Science: S. Karger AG, Basel.

Jackson, AS, Stanforth, PR, Gagnon, J, Rankinen, T, Leon, AS, Rao, DC, et al. (2002).

The effect of sex, age and race on estimating percentage body fat from body

mass index: The Heritage Family Study. Int J Obes, 26, 789‐796.

Jafar, TH, Islam, M, Poulter, N, Hatcher, J, Schmid, CH, Levey, AS, et al. (2005).

Children in South Asia have higher body mass‐adjusted blood pressure levels

than white children in the United States: a comparative study. Circulation

111, 1291‐1297.

Janssen, I, Katzmarzyk, P, & Ross, R (2004). Waist circumference and not body mass

index explains obesity‐related health risk. Am J Clin Nutr, 79, 379‐384.

Jebb, SA, Cole, TJ, Doman, D, Murgatroyd, PR, & Prentice, AM (2000). Evaluation of

the novel Tanita body‐fat analyser to measure body composition by

comparison with a four‐compartment model. Brit J Nutr, 83, 115‐122.

Johnson, M, Figueroa‐Colon, R, Huang, T, Dwyer, J, & Goran, M (2001). Longitudinal

changes in body fat in African American and Caucasian children: influence of

fasting insulin and insulin sensitivity. J Clin Endocrinol Metab, 86, 3182‐3187.

Kagawa, M, Kerr, D, Uchida, H, & Binns, CW (2006). Differences in the relationship


283

between BMI and percentage body fat between Japanese and

Australian‐Caucasian young men. Br J Nutr, 95, 1002‐1007.

Katzmarzyk, P, Pérusse, L, Malina, R, Bergeron, J, Després, JP, & Bouchard C (2001).

Stability of indicators of the metabolic syndrome from childhood and

adolescence to young adulthood: the Quebec Family Study. J Clin Epidemiol,

54, 190‐195.

Katzmarzyk, PT(2004). Waist circumference percentiles for Canadian youth 11‐18 y

age. Eur J Clin Nutr, 58, 1011‐1015.

Katzmarzyk, PT, Srinivasan, SR, Chen, W, Malina, RM, 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, e198‐e205.

Ke, L, Brick, KE, Cant, RV, Li, Y, & Morrell, SL (2009). The relationship between

obesity and blood pressure differs by ethnicity in Sydney school children. Am

J Hypertens, 22, 52‐58.

Kelishadi, R, Gheiratmand, R, Ardalan, G, Adeli, K, Gouya, MM, Razaghi, EM, et al.

(2007). Association of anthropometric indices with cardiovascular disease

risk factors among children and adolescents: CASPIAN Study. Int J Cardiol,

117, 340‐348.

Khaled, MA, McCutcheon, MJ, Reddy, S, Pearman, PL, Hunter, GR, & Weinsier, RL

(1988). Electrical impedance in assessing human body composition: The BIA

method. Am J Clin Nutr, 47, 789‐792.

Kim, HK, Tanaka, K, Nakadomo, F, & Watanabe, K (1994). Fat‐free mass in Japanese

boys predicted from bioelectrical impedance and anthropometric variables.

Eur J Clin Nutr, 48, 482‐489.

Kim, J, Shen, W, Gallagher, D, Jones Jr, A, Wang, Z., Wang, J, et al. (2006). Total‐body

skeletal muscle mass: estimation by dual‐energy X‐ray absorptiometry in

children and adolescemts. Am J Clin Nutr, 84, 1014‐1020.

Kimm, S, Barton, B, Obarzanek, E, McMahon, R, Sabry, Z, Waclawiw, M, et al. (2001).

Racial divergence in adiposity during adolescence: The NHLBI growth and

health study. Pediatrics, 107, 30‐36.


284

Kopelman, P (2007). Health risks associated with overweight and obesity. Obes Rev,

8(Suppl 1), S13‐S17.

Kotler, DP, Thea, DM, Heo, M, Allison, DB, Engelson, ES, Wang, J, et al. (1999).

Relative influences of sex, race, environment, and HIV infection on body

composition in adultes. Am J Clin Nutr, 69, 432‐439.

Kriemler, S, Puder, J, Zahner, L, Roth, R, Braun‐Fahrländer, C, & Bedogni, G (2009).

Cross‐validation of bioelectrical impedance analysis for the assessment of

body composition in a representative sample of 6‐ to 13‐year‐old children.

Eur J Clin Nutr, 63, 619‐626.

Kushner, RF, Schoeller, DA, Fjeld, CR, & Danford, L (1992). Is the impedance index

(ht2/R) significant in predicting total body water? Am J Clin Nutr, 56, 835‐839.

Kyle, UG, Bosaeus, I, De Lorenzo, AD, Deurenberg, P, Elia, M, Gomez, JM, et al.

(2004). Bioelectrical impedance analysis‐‐part I: review of principles and

methods. Clin Nutr, 23, 1226‐1243.

Lahmann, PH, Lissner, L, Gullberg, B, & Berglund, G (2002). A prospective study of

adiposity and all‐cause mortality: the Malmo Diet and Cancer Study. Obes

Res, 10, 361‐369.

Lambert, M, Paradis, G, O'Loughlin, J, Delvin, EE, Hanley, JA, & Levy, E (2004). Insulin

resistance syndrome in a representative sample of children and adolescents

from Quebec, Canada. Int J Obes, 28, 833‐841.

Lanham, D. A., Stead, M. A., Tsang, K., & Davies, P. S. W. (2001). The prediction of

body composition in Chinese australian females. Int J Obesity, 25, 286‐291.

Lear, SA, Humphries, KH, Kohli, S, Chockalingam, A, Frohlich, JJ, & Birmingham, CL

(2007). Visceral adipose tissue accumulation differs according to ethnic

background: results of the Multicultural Community Health Assessment Trial

(M‐CHAT). Am J Clin Nutr, 86, 353‐359.

Lee, CMY., Huxley, RR, Wildman, RP, & Woodward, M (2008). Indices of abdominal

obesity are better discriminators of cardiovascular risk factors than BMI: a

meta‐analysis. J Clinl Epidemiol, 61, 646‐653.

Lee, K, Song, YM, & Sung, J (2008). Which obesity indicators are better predictors of

metabolic risk? Healthy twin study. Obesity, 16, 834‐840.


285

Lee, S, Bacha, F, Gungor, N, & Arslanian, SA (2006). Waist circumference is an

independent predictor of insulin resistance in black and white youths. J

Pediatr, 148, 188‐194.

Leenen, R, van der Kooy, K, Seidell, JC, & Deurenberg, P (1992). Visceral fat

accumulation measured by magnetic resonance imaging in relation to serum

lipids in obese men and women. Atherosclerosis, 94, 171‐181.

Leung, SS, Lau, JT, Tse, LY, & Oppenheimer, SJ (1996). Weight‐for‐age and

weight‐for‐height references for Hong Kong children from birth to 18 years.

J Paediatr Child Health, 32, 103‐109.

Li, Y, Ma, G, Pan, H, Cui, Z, Zhang, C, Song, J, et al. (2003). Comparison of prevalcenc

of overweight using different definitions in adolescents. Chin J Prevention

Control Chronic Non‐Communicable Diseases, 11, 281‐282.

Li, Y, Schouten, EG, Hu, X, Cui, Z, Luan, D, & Ma, G (2008). Obesity prevalence and

time trend among youngsters in China, 1982‐2002. Asia Pac J Clin Nutr, 17,

131‐137.

Li, Y, Yang, X, Zhai, F, Kok, FJ, Zhao, W, Piao, J, et al. (2008). Prevalence of the

metabolic syndrome in Chinese adolescents. Brit J Nutr, 99, 565‐570.

Li, Y, Yang, X, Zhai, F, Piao, J, Zhao, W, Zhang, J, et al. (2005). Disease risks of

childhood obesity in China. Biomed Environ Sci, 18, 401‐410.

Lobstein, T, Baur, L, & Uauy, R (2004). Obesity in children and young people: a crisis

in public health. Obes Rev, 5(Suppl 1), S4‐85.

Lofren, I, Herron, K, Zern, T, West, K, Patalay, M, & Shachter, NS (2004). Waist

circumference is a better predictor than body mass index for coronary heart

disease risk in overweight premenopausal women. J Nutr, 134, 1071‐1076.

Lohman, T, Caballero, B, Himes, J, Davis, CE, Stewart, D, Houtkooper, L et al. (2000).

Estimation of body fat from anthropometry and bioelctrical impedance in

Native American children. Int J Obes, 24, 982‐988.

Lohman, TG (1986). Applicability of body composition techniques and constants for

children and youths. Exerc Sport Sci Rev, 14, 325‐357.

Lohman, TG (1992). Advances in Body Composition Assessement: Champaign, IL:

Human Kinetics.


286

Lonzer, MD, Imrie, R, Rogers, D, Worley, D, Licata, A, & Secic, M (1996). Effects of

heredity, age, weight, puberty, activity, and calcium intake on bone mineral

density in children. Clin Pediatr, 35, 185‐189.

Lukaski, HC, Johnson, PE, Bolunchuk, WE, & Lykken, GI (1985). Assessment of

fat‐free mass using bioelectrical impedance measurements of the human

body. Am J Clin Nutr, 41, 810‐817.

Luke, A, DurazoArvizu, R, Rotimi, C, Prewitt, TE, Forrester, T, Wilks, R, et al. (1997).

Relation between body mass index and body fat in black population samples

from Nigeria, Jamaica, and the United States. Am J Epidemiol, 145, 620‐628.

Mølgaard, C, & Michaelsen, KF (1998). Changes in body composition during growth

in healthy school‐age children. Appl Radiat Isotopes, 49, 577‐579.

Mølgaard, C, Tohomsen, B, Prentice, A, Cole, TJ, & Michaelsen, KF (1997). Whole

body bone mineral content in healthy children and adolescents. Arch Dis

Child, 76, 9‐15.

Ma, G, Yao, M, Liu, Y, Lin, A, Zou, H, Urlando, A, et al. (2004). Validation of a new

pediatric air‐displacement plethysmograph for assessing body composition in

infants. Am J Clin Nutr, 79, 653‐660.

Maffeis, C, Pietrobelli, A, Grezzani, A, Provera, S, & Tato, L (2001). Waist

circumference and cardiovascular risk factors in prepubertal children. Obes

Res, 9, 179‐187.

Magarey, AM, Daniels, L, & Boulton, T (2001). Prevalence of overweight and obesity

in Australian children and adolescents: reassessment of 1985 and 1995 data

against new standard international definitions. Med J Aust, 174, 561‐564.

Malina, RM (2005). Variation in body composition associated with sex and ethnicity.

In Heymsfield, SB, Lohman,TG, Wang, Z, & Going, S (Eds.), Human body

composition (2nd ed., pp. 271‐298). Chanpaign IL: Huamn Kinetics.

Malina, R.M, Bouchard, C, & Bar‐Or, O (2004). Growth, Maturation, and Physical

Activity. (2nd ed.): Champaign, IL: Human Kinetics.

Malina, RM, Huang, YC, & Brown, KH (1995). Subcutaneous adipose tissue

distribution in adolescent girls of 4 ethnic groups. Int J Obes Relat Metab

Disord, 19, 793‐797.


287

Mallows, CL (1973). Some comments on Cp. Technometrics, 15, 661‐675.

Marfell‐Jones M, Olds T, Stewart A, & Carter JEL (2006). International standards for

anthropometric assessment (revised 2006). 2ed: The International Society

for the Advancement of Kinanthropometry.

Mast, M, Kortzinger, I, Konig, E, & Muller, M (1998). Gender differences in fat mass

of 5‐7‐year old children. Int J Obes, 22, 878‐884.

Masuda, T, & Komiya, S (2004). A prediction equation for total body water from

bioelectrical impedance in Japanese children. J Physiol Anthropol Appl

Human Sci, 23, 35‐39.

Matsushita, Y, Yoshiike, N, Kaneda, F, Yoshita, K, & Takimoto, H (2004). Trends in

childhood obesity in Japan over the last 25 years from the nutrition survey.

Obes Res, 12, 205‐214.

Mattsson, S, & Thomas, B (2006). Development of methods for body composition

studies. Phys Med Biol, 51, 203‐228.

Maynard, L, Wisemandle, W, Roche, A, Chumlea, W, Guo, S, & Siervogel, R (2001).

Childhood body composition in relation to body mass index. Pediatrics, 107,

344.

Maynard, LM, Guo, SS, Chumlea, WC, Roche, AF, Wisemandle, WA, & Zeller, C (1998).

Total‐body and regional bone mineral content and areal bone mineral

density in children aged 8‐18 y: the Fels longitudinal study. Am J Clin Nutr, 68,

1111‐1117.

McAuley, KA, Williams, SM, Mann, JI, Goulding, A, & Murphy, E (2002). Increased

risk of type 2 diabetes despite same degree of adiposity in different racial

groups. Diabetes Care, 25, 2360‐2361.

McCarthy, H (2006). Body fat measurements in children as predictors for the

metabolic syndrome: focus on waist circumference. Proc Nutr Soc, 65,

385‐392.

McCarthy, H, & Ashwell, M (2006). A study of central fatness using waist‐to‐height

ratios in UK children and adolescents over two decades supports the simple

message ‐'keeping your waist circumference to less than half your height'. Int

J Obes, 30, 988‐992.


288

McCarthy, H, Jarrett, H, & Crawley, H (2001). The development of waist

circumference percentiles in British children aged 5.0‐16.9 y. Eur J Clin Nutr,

55, 902‐907.

McClanahan, BS, Stockton, MB, Lanctot, JQ, Relyea, G, Klesges, RC, Slawson, DL, et al.

(2009). Measurement of body composition in 8‐10‐year‐old

African‐American girls: A comparison of dual‐energy X‐ray absorptiometry

and foot‐to‐foot bioimpedance methods. Int J Pediatric Obes, 4, 389‐396.

McKeigue, PM, Shah, B, & Marmot, MG (1991). Relation of central obesity and

insulin resistance with high diabetes prevalence and cardiovascular risk in

South Asians. Lancet, 337, 382‐386.

Mehta, S, Mahajan, D, Steinbeck, KS, & Bermingham, MA (2002). Relationship

between measures of fatness, lipids and ethnicity in a cohort of adolescent

boys. Ann Nutr Metab, 46, 192‐199.

Meng, L, Mi, J, Cheng, H, Hou, D, Zhao, X, & Ding, X (2007). Using waist

circumference and waist‐to‐height ration to access central obesity in children

and adolescents. Chin J Evid Based Pediatr 12, 245‐252.

Miaison, P, Byrne, CD, Hales, CN, Day, NE, & Wareham, NJ (2001). Do different

dimensions of the metabolic syndrome change together? Evidenc supporting

obesity as the central feature. Diabetes Care, 24, 1758‐1763.

Misra, A (2003). Revisions of cutoffs of body mass index to define overweight and

obesity are needed for the Asian‐ethnic groups. Int J Obes Relat Metab

Disord, 27, 1294‐1296.

Misra, A, Khurana, L, Vikram, NK, Goel, A, & Wasir, JS (2007). Metabolic syndromme

in children: current issues and South Asian perspective. Nutrition 23,

895‐910.

Misra, A, Madhavan, M, Vikram, NK, Pandey, RM, Dhingra, V, & Luthra, K (2006).

Simple anthropometric measures identify fasting hyperindulinemia and

clustering of cardiovascular risk factors in Asian Indian adolescents.

Metabolism, 55, 1569‐1573.

Misra, A, Wasir, JS, & Vikram, NK (2005). Waist circumference criteria for the

diagnosis of abdominal obesity are not applicable uniformly to all


289

populations and ethnic groups. Nutrition, 21, 969‐976.

Mora, S, & Gilsanz, V (2003). Establishment of peak bone mass. Endocrin Metab Clin

North America, 32, 39‐40.

Moreno, L, Fleta, J, Mur, L, Rodríguez, G, Sarría, A, & Bueno, M (1999). Waist

circumference values in Spanish children‐gender related differences. Eur J

Clin Nutr, 53, 429‐433.

Moreno, L, Mesana, M, González‐Gross, M, Gil, CM, Ortega, FB, Fleta, J, et al. (2007).

Body fat distribution reference standards in Spanish adolescents the AVENA

Study. Int J Obes, 31, 1‐8.

Moreno, L, Pineda, I, Rodríguez, G, Fleta, J. Sarría, A, & Bueno, M (2002). Waist

circumference for the screening of the metabolic syndrome in children. Acta

Paediatr, 91, 1307‐1312.

Morrison, JA, Barton, BA, Obarzanek, E, Crawford, PB, Guo, SS, & Schreiber, GB

(2001). Racial differences in the sums of skinfolds and percentage of body fat

estimated from impedance in black and white girls, 9 to 19 years of age: The

National Heart, Lung, and Blood Institute Growth and Health Study. Obes Res,

9, 297‐305.

Morrison, JA, Guo, SS, Specker, B, Chumlea, WC, Yanovski, SZ, & Yanovski, JA (2001).

Assessing the body composition of 6‐17‐year‐old black and white girls in field

studies. Am J Hum Biol, 13, 249‐254.

Moussa, M, Skaik, M, Selwanes, S, Yaghy, O, & Bin‐Othman, S (1994). Contribution of

body fat and fat pattern to blood pressure level in school children. Eur J Clin

Nutr, 48, 587‐590.

Must, A, Dallal, G, & Dietz, W (1991). Reference data for obesity: 85th and 95th

percentiles of body mass index (wt/ht2) and triceps skinfold thickness. Am J

Clin Nutr, 53, 839‐846.

Nagaya, T, Yoshida, H, Takahashi, H, Matsuda, Y, & Kawai, M (1991). Body mass index

(weight/height2) or percentage body fat by bioelectrical impedance analysis:

which variable better reflects serum lipid profile? Int J Obes Relat Metab

Disord, 23, 771‐774.

Nakao, T, & Komiya, S (2003). Reference norms for a fat‐free mass index and fat


290

mass index in the Japanese child population. J Physiol Anthropol Appl Human

Sci, 22, 293‐298.

National Center for Health Statistics (2004). Health, United States, 2004‐ with

chartbook on trends in the health of Americans. Maryland: Hyattsville.

National High Blood Pressure Education Program Working Group on Hypertension

Control in Children and Adolescents (1996). Update on the 1987 task force

report on high blood pressure in children and adolescents: A working group

report from the National High Blood Pressure Education Program. Pediatrics,

98, 649‐658.

Navder, KP, He, Q, Zhang, X, He, S, Gong, L, Sun, Y, et al. (2009). Relationship

between body mass index and adiposity in prepubertal children: ethnic and

geographic comparisons between New York City and Jinan City (China). J

Appl Physiol, 107, 488‐493.

NCEP Expert Panel on Blood Cholesterol Levels in Children and Adolescents (1992).

National cholesterol education program (NCEP): Highlights of the report of

the expert panel on blood cholesterol levels in children and adolescents.

Pediatrics, 89, 495‐501.

Nelson, DA, & Barondess, DA (1997). Whole body bone, fat and lean mass in

children: Comparison of three ethnic groups. Am J Phys Anthropol, 103,

157‐162.

Nelson, DA, Simpson, PM, Johnson, CC, Barondess, DA, & Kleerekoper, M (1997).

The accumulation of whole body skeletal mass in third‐ and fourth‐grade

children: Effects of age, gender, ethnicity, and body composition. Bone, 20,

73‐78.

Neovius, MG, Linne, YM, Barkeling, BS, & Rossner, SO (2004). Sensitivity ans

specificity of classification systems for fatness in adolescents. Am J Clin Nutr,

80, 597‐603.

Ng, VW, Kong, AP, Choi, KC, Ozaki, R, Wong, GW, So, WY, et al. (2007). BMI and waist

circumference in predicting cardiovascular risk factor clustering in Chinese

adolescents. Obesity, 15, 494‐503.

Nielsen, BM, Dencker, M, Ward, L, Linden, C, Thorsson, O, Karlsson, MK, et al. (2007).


291

Prediction of fat‐free body mass from bioelectrical impedance among 9‐to

11‐year‐old Swedish children. Diabetes Obes Metab, 9, 521‐539.

Nindl, BC, Kraemer, WJ, Emmert, WH, Mazzetti, SA, Gotshalk, LA, Putukian, M, et al.

(1998). Comparison of body composition assessment among lean black and

white male collegiate athletes. Med Sci Sports Exerc, 30, 769‐776.

Norgan, NG (1994a). Interpretation of low body mass indices: Australian Aborigines.

Am J Phys Anthropol, 94, 229‐237.

Norgan, NG (1994b). Population differences in body composition in relation to BMI.

Eur J Clin Nutr, 48(suppl 3), S10‐S27.

Norgan, NG (2005). Laboratory and field measurements of body composition. Public

Health Nutr, 8, 1108‐1122.

Novotny, R, Daida, Y, Grove, J, Marchand, L, & Vijayadeva, V (2006). Asian

Adolescents Have a Higher Trunk: Peripheral Fat Ratio than Whites. J Nutr,

136, 642‐647.

Novotny, R, Daida, YG, Grove, JS, Achsrya, S, & Vogt, TM (2003). Formula feeding in

infancy is associated with adolescent body fat and earlier menarche. Cell Mol

Biol, 49, 1289‐1293.

Novotny, R, Going, S, Teegarden, D, Loan, MV, McCabe, G, McCabe, L, et al. (2007).

Hispanic and Asian pubertal girls have higher android/gynoid fat ratio than

whites. Obeisty, 15, 1565‐1570.

Ogden, C, & Carroll, M (2010). Prevalence of obesity among children and

adolescents: United States, trends 1963‐1965 through 2007‐2008.

Accessed 26 August, 2010, from

http://www.cdc.gov/nchs/data/hestat/obesity_child_07_08/obesity_child_0

7_08.htm

Ogden, CL, Flegal, KM, Carroll, M.D, & Johnson, CL (2002). Prevalence and trends in

overweight among US children and adolescents 1999‐2000. JAMA, 288,

1728‐1732.

Okosun, IS (2000). Ethnic differences in the risk of type 2 diabetes attributable to

differences in abdominal adiposity in American women. J Cardiovasc Risk, 7,

425‐430.


292

Okosun, IS, Liao, Y, Rotimi, CN, Dever, GEA, & Cooper, RS (2000). Impact of birth

weight on ethnic variations in subcutaneous and central adiposity in

American children aged 5‐11 years. A study from the Third National Health

and Nutrition Examination Survey. Int J Obes Relat Metab Disord, 24,

479‐484.

Okosun, IS, Liao, Y, Rotimi, CN, Prewitt, TE, & Cooper, RS (2000). Abdominal adiposity

and clustering of multiple metabolic syndrome in White, Black and Hispanic

Americans. Ann Epidemiol, 10, 263‐270.

Ondrak, K, McMurray, R, Bangdiwala, S, & Harrell, J (2007). Influence of aerobic

power and percent body fat on cardiovascular disease risk in youth. J Adolesc

Health, 42, 146‐152.

Orphanidou, C, McCargar, L, Birmingham, CL, Mathieson, J, & Goldner, E (1994).

Accuracy of subcutaneous fat measurement: comparison of skinfold calipers,

ultrasound, and computed tomography. J Am Diet Assoc, 94, 855‐858.

Owens, S, Litaker, M, Allison, J, Riggs, S, Ferguson, M, & Gutin, B (1999). Prediction

of visceral adipose tissue from simple anthropometric measurements in

youths with obesity. Obes Res, 7, 16‐22.

Palaniappan, L, Carnethon, M, Wang, Y, Hanley, AJG, Fortmann, SP, Hafner, SM, et al.

(2004). Predictors of the incident metabolic syndrome in adults. Diabetes

Care, 27, 788‐793.

Park, YW, Allison, DB, Heymsfield, SB, & Gallagher, D (2001). Larger amounts of

visceral adipose tissue in Asian Americans. Obes Res, 9, 381‐387.

Perichart‐Perera, O, Balas‐Nakash, M, Schiffman‐Selechnik, E, & Barbato‐Dosal, A

(2007). Obesity Increases Metabolic Syndrome Risk Factors in School‐Aged

Children from an Urban School in Mexico City. J Am Diet Assoc, 107, 81‐91.

Phillips, SM, Bandini, LG, Compton, DV, Naumova, EN, & Must, A (2003). A

Longitudinal Comparison of Body Composition by Total Body Water and

Bioelectrical Impedance in Adolescent Girls. J Nutr, 133, 1419‐1425.

Plachta‐Danielzik, S, Landsberg, B, Johannsen, M, Lange, D, & Müller, MJ (2008).

Association of different obesity indices with blood pressure and blood lipids

in children and adolescents. Brit J Nutr 100, 208‐218.


293

Potts, J, & Simmons, D (1994). Sex and ethnic groups differences in fat distribution in

young United Kingdom South Asians and Europids. J Clin Eepidemiol, 47,

837‐841.

Pouliot, MC, Despres, JP, Lemieux, S, Moorjani, S, Bouchard, C, Tremblay, A, et al.

(1994). Waist circumference and abdominal sagittal diamerer: best simple

anthropometric indexes of abdominal visceral adipose tissue accumulation

and related cardiovascular risk in men and women. Am J Cardiol, 73,

460‐468.

Racette, SB, Schoeller, DA., Luke, AH, Shay, K, Hnilicka, JH, & Kusher, RF (1994).

Relative dilution spaces of 2H‐ and 18O‐labeld water in humans. Am J

Physiol‐Endoc Metab, 267, e585‐e590.

Raji, A, Seely, EW, Arky, RA, & Simmons, DC (2001). Body fat distribution and insulin

resistance in healthy Asian Indians and Caucasians. J Clin Endocrinol Metab,

86, 5366‐5371.

Ramirez, ME, & Mueller, WH (1980). The development of obesity and fat patterns on

Tokelau children. Hum Biol, 52, 675‐687.

Reilly, JJ (1998). Assessment of body compostion in infants and children. Nutrition,

14, 821‐825.

Reinehr, T, de Sousa, G, Toschke, A, & Andler, W (2007). Comparison of metabolic

syndrome prevalence using eight different definitions: A critical approach.

Arch Dis Child, 92, 1053‐1054.

Ribeiro, J, Guerra, S, Oliveira, J, Andersen, L, Duarte, J, & Mota, J (2004). Body

fatness and clustering of cardiovascular disease risk factors in Portuguese

children and adolescents. Am J Hum Biol, 16, 556‐562.

Ritz, P, Vol, S, Berrut, G, Tack, I, Arnaud, MJ, & Tichet, J (2008). Influence of gender

and body composition on hydration and body water spaces. Clin Nutr, 27,

740‐746.

Robson, JRK, Bazin, M, & Soderstrom, R (1971). Ethnic differences in skinfold

thickness. Am J Clin Nutr, 24, 864‐868.

Rolland‐Cachera, MF, Castetbon, K, Arnault, N, Bellisle, F, Romano, MC, Lehingue, Y,

et al. (2002). Body mass index in 7–9‐y‐old French children: frequency of


294

obesity, overweight and thinness. Int J Obes, 26, 1610‐1616.

Rolland‐Cachera, MF, Cole, TJ, Sempe, M, Tichet, J, Rossignol, C, & Charraud, A

(1991). Body Mass Index variations: centiles from birth to 87 years. Eur J Clin

Nutr, 45, 13‐21.

Rönnemaa, T, Knip, M, Lautala, P, Viikari, J, Uhari, M, Leino, A, et al. (1991). Serum

insulin and other cardiovascular risk indicators in children, adolescents and

young adults. Ann Med, 23, 67‐72.

Roubenoff, R, Dallal, G, & Wilson, P (1995). Predicting body fatness: the body mass

index vs estimation by bioelectrical impedance. Am J Public Health, 85,

726‐728.

Rupich, RC, Specker, BL, Lieuw‐A‐Fa, M, & Ho, M (1996). Gender and race differences

in bone mass during infancy. Calcified Tissue Int, 58, 395‐397.

Rush, E, Plank, L, Chandu, V, Laulu, M, Simmons, D, Swinburn, B, et al. (2004). Body

size, body composition, and fat distribution: a comparison of young New

Zealand men of European, Pacific Island, and Asian Indian ethnicities. N Z

Med J, 117, U1203.

Rush, EC, Freitas, I, & Plank, LD (2009). Body size, body composition and fat

distribution: comparative analysis of European, Maori, Pacific Island and

Asian Indian adults. Brit J Nutr, 102, 632‐641.

Rush, EC, Puniani, K, Valencia, ME, Davies, PSW, & Plank, LD (2003). Estimation of

body fatness from body mass index and bioelectrical impedance:

Comparison of New Zealand European, Maori and Pacific Island children. Eur

J Clin Nutr, 57, 1394.

Rush, EC, Scraqq, R, Schaaf, D, Juranovich, G, & Plank, LD (2009). Indices of fatness

and relationships with age, ethnicity and lipids in New Zealand European,

Māori and Pacific children. Eur J Clin Nutr, 63, 627‐633.

Sala, M, Webber, CE, Morrison, J, Beaumont, LF, & Barr, RD (2007). Whole‐body

bone mineral content, lean body mass, and fat mass measured by

Dual‐energy X‐ray aborptiometry in a population of normal Canadian

children and adolescents. Can Assoc Radiol J, 58, 46‐52.

Sampei, MA, Novo, NF, Juliano, Y, Colugnati, FAB, & Sigulem, DM (2003).


295

Anthropometry and body composition in ethnic Japanese and Caucasian

adolescent girls: Considerations on ethnicity and menarche. Int J Obes, 27,

1114‐1120.

Sampei, MA, Novo, NF, Yuliano, Y, & Sigulem, DM (2001). Comparison of the body

mass index to other methods of body fat evaluation in ethnic Japanese and

Caucasian adolescent girls. Int J Obes, 25, 400‐408.

Sartorio, A, Malavolti, M, Agosti, F, Marinone, PG, Caiti, O, Battistini, N, et al. (2005).

Body water distribution in severe obesity and its assessment from

eight‐polar bioelectrical impedance analysis. Eur J Clin Nutr, 59, 155‐160.

Savva, SC, Tornaritis, M, Savva, ME, kourides, Y, Panagi, A, Silikiotou, N, et al. (2000).

Waist circumference and waist‐height ration are better predictors of

cardivascular disease risk factors in children than body mass index. Int J Obes,

24, 1453‐1458.

Schaefer, F, Georgi, M, Zieger, A, & Scharer, K (1992). Usefulness of bioelectrical

impedance estimation of fat‐free body mass in children and youth: a

cross‐validation study. J Appl Physiol, 72, 366‐373.

Schoeller, DA (2005). Hydrometry. Human body composition (2nd ed.). Champaign, IL:

Human Kinetics, pp35‐46.

Segal, KR, Gutin, B, Presta, E, Wang, J, & Van Itallie, TB (1985). Estimation of human

body composition by electrical impedance methods: A comparative study. J

Appl Physiol, 58, 1565‐1571.

Shaikh, S, Mahalanabis, D, Kurpad, AV, & Khaled, MA (2002). Validation of an

anthropometric equation and bioelectrical impedance analysis (BIA)

technique to measure body composition of children in India using D2O

dilution method. Nutr Res, 22, 685‐694.

Shaw, NJ, Crabtree, NJ, Kibirige, MS, & Fordham, JN (2007). Ethnic and gender

differences in body fat in British schoolchildren as measured by DXA. Arch Dis

Child, 92, 872‐875.

Shen, W, Punyanitya, M, Chen, J, Gallagher, D, Albu, J, Pi‐Sunyer, X, et al. (2006).

Waist circumference correlates with metabolic syndrome indicators better

than percentage fat. Obesity, 14, 727‐736.


296

Simpson, JA, Lobo, DN, Anderson, JA, Macdonald, IA, Perkins, AC, Neal, KR, et al.

(2001). Body water compartment measurements: a comparison of

bioelectrical impedance analysis with tritium and sodium bromide dilution

techniques. Clin Nutr, 20, 339‐343.

Sinaiko, AR, Donahue, RP, Jacobs, DR, & Prineas, RJ (1999). Relation of weight and

rate of increase in weight during childhood and adolescence to body size,

blood pressure, fasting insulin, and lipids in young adults: the Minneapolis

children's Blood Pressure Study. Circulation, 99, 1471‐1476.

Sinaiko, AR, Steinberger, J, Moran, A, Prineas, RJ, & Jacobs, D (2002). Relation of

insulin resistance to blood pressure in childhood. J Hypertens, 20, 509‐517.

Sinha, R, Fisch, G, Teague, B, Tamborlane, WV, Banyas, B, Allen, K, et al. (2002).

Prevalence of impaired glucose tolerance among children and adolescents

with marked obesity. N Engl J Med 346, 802‐810.

Sisson, SB, Katzmarzyk, PT, Srinivasan, SR, Chen, W, Freedman, DS, Bouchard, C, et al.

(2009). Ethnic differences in subcutaneous adiposity and waist girth in

children and adolescents. Obesity, 17, 2075‐2081.

Slaughter, MH, Lohman, TG, Boileau, RA, Christ, CB, & Stillman, RJ (1990).

Differences in the subcomponents of fat‐free body in relation to height

between black and white children. Am J Hum Biol, 2, 209‐217.

Sluyter, JD, Schaaf, D, Scragg, RKR, & Plank, LD (2010). Prediction of fatness by

standing 8‐electrode bioimpedance: A multiethnic adolescent population.

Obesity, 18, 183‐189.

Snijder, MB, Kuyf, BEM, & Deurenberg, P (1999). Effect of body build on the validity

of predicted body fat from body mass index and bioelectrical impedance.

Ann Nutr Metab, 43, 277‐285.

Song, J, & Zhang, B (1999). Study of body composition of new‐born infants. Chin J

Perinat Med, 2, 161‐164.

Song, MY, Kim, J, Horlick, M, Wang, J, Pierson, RN, Heo, M, et al. (2002). Prepubertal

Asians have less limb sletetal muscle. J Appl Physiol, 92, 2285‐2291.

Sopher, A, Shen, W, & Pietrobelli, A (2005). Pediatric body composition methods. In

Heymsfield, SB, Lohman, TG, Wang, Z, & Going, SB (Eds.), Human body


297

composition (pp. 129‐149). Champaign IL: Human Kinetics.

Speakman, JR, Nair, KS, & Goran, MI (1993). Revised equations for calculating CO2

production from doubly labeled water in humans. Am J Phyiol, 264,

e912‐e917.

Spencer, EA, Appleby, PN, Davey, GK, & Key, TJ (2002). Validity of self‐reported

weight and height in 4808 EPIC‐Oxford participants. Public Health Nutr, 5,

561‐565.

Srinivasan, SR, Bao, W, Wattigney, WA, & Berenson, GS (1996). Adolescents

overweight is associated with adult overweight and related multiple

cardiovascular risk factors: the Bogalusa Heart Study. Metabolism, 45,

235‐240.

Srinivasan, SR, Myers, L, & Berenson, GS (2002). Predictability of childhood adiposity

and insulin for developing insulin resistance syndrome (syndrome X) in young

adulthood: the Bogalusa Heart Study. Diabetes, 51, 204‐209.

Steinberger, J, Jacobs, DR, Raatz, S, Moran, A, Hong, CP, & Sinaiko, AR (2006).

Comparison of body fatness measurements by BMI and skinfolds vs dual

energy X‐ray absorptiometry and their relation to cardiovascular risk factors

in adolescents . Int J Obes, 30, 1170.

Steinberger, J, Moran, A, Hong, CP, Jacobs, DRJ, & Sinaiko, AR (2001). Adiposity in

childhood predicts obesity and insulin resistance in young adulthood. J

Pediatr, 138, 469‐473.

Sun, SM, Chumlea, WC, Heymsfield, SB, Lukaski, HC, Schoeller, D, FriedL, K, et al.

(2003). Development of bioelectrical impedance analysis prediction

equations for body composition with the use of a multicomponent model for

use in epidemiologic surveys. Am J Clin Nutr, 77, 331‐340.

Sung, RY, So, H, Choi, K, Nelson, EA, Li, AM, Yin, JA, et al. (2008). Waist circumference

and waist‐to‐height ratio of Hong Kong Chinese children. BMC Public Health,

8, 324‐333.

Sung, RYT, Yu, CCW, Choi, KC, McManus, A, Li, AMC, Xu, SLY, et al. (2007). Waist

circumference and body mass index in Chinese children: cutoff values for

predicting cardiovascular risk factors. Int J Obes, 31, 550‐558.


298

Swinburn, BA, Ley, SJ, Carmichael, HE, & Plank, LD (1999). Body size and

composition in Polynesians. Int J Obes Relat Metab Disord, 23, 1178‐1183.

Tabachnick, BG, & Fidell, LS (2007). Using multivariate statistics. (5th ed.). Boston:

Pearson/Allyn & Bacon. p73.

Tahara, Y, Moji, K, Aoyagi, K, Tsunawake, N, Muraki, S, & Mascie‐Taylor, CGN (2002).

Age‐related pattern of body density and body composition of Japanese men

and women 18‐59 years of age. Am J Hum Biol, 14, 743‐752.

Tanaka, S, Horimai, C, & Katsukawa, F (2003). Ethnic difference in abdominal visceral

fat accumulation between Japanese, African‐Americans, and Caucasians: A

meta‐analysis. Acta Diabetol, 40, S302‐S304.

Tanaka, S, Togasbi, K, Rankinen, T, Persse, L, Leon, AS, Rao, DC, et al. (2002). Is

adiposity at normal body weight relevant for cardiovascular disease risk? Int

J Obes Relat Metab Disord, 26, 176‐183.

Tanner, J, & Whitehouse, R (1976). Clinical longitudinal standards for height, weight,

height velocity, weight velocity, and stages of puberty. Arch Dis Child, 51,

170‐179.

Tanner, JM, Hughes, PCR, & Whitehouse, RH (1981). Radiographically determinentd

widths of bone muscles and fat in the upper arm and calf from age 3‐18

years. Ann Hum Biol, 8, 495‐517.

Tatsukawa, M, Kurokawa, MTY, Yoshimatsu, H, & Sakata, T (2000). Regional fat

deposition in the legs is useful as a presumptive marker of

antiatherogenesity in Japanese. Proc Soc Exp Biol Med, 223, 156‐162.

Teixeira, PJ, Sardinha, LB, Going, SB, & Lohman, TG (2001). Total and regional fat and

serum cardiovascular disease risk factors in lean and obese children and

adolescents. Obes Res, 9, 432‐442.

The BMA Board of Science (2005). Preventing childhood obesity. Accessed 10 May,

2008, from http://www.bma.org.uk/ap.nsf/content/childhoodobesityo.

Thomas, K, Keller, CS, & Holbert, KE (1997). Ethnic and age trends for body

composition in women residing in the U.S. Southwest: regional fat. Med Sci

Sports Exerc, 29, 82‐89.

Thomas, LD, Van Der Velde, D, & Schloeb, PR (1991). Optimum doses of deuterium


299

oxide and sodium bromide for the determination of total body water and

extracellular fluid. J Pharm Biomed Analysis, 9, 581‐584.

Trevisan, M, Liu, J, Bahsas, FB, & Menotti, A (1998). Syndrome X and Mortality: A

Population‐based Study. Am J Epidemiol, 148, 958‐966.

Tulloch‐Reid, MK, Williams, DE, Looker, HC, Hanson, RL, & Knowler, WC (2003). Do

measures of body fat distribution provide information on the risk of type 2

diabetes in addition to measures of general obesity?. Diabetes Care, 26,

2556‐2561.

Tybor, DJ, Lichtenstein, AH, Dallal, GE, Daniels, SR., & Must, A (2010). Racial

differences in central adiposity in a longitudinal cohort of black and white

adolescent females. BMC Pediatrics, 10, 2.

Utter, AC, Nieman, DC, Ward, AN, & Butterworth, DE (1999). Use of the leg‐to‐leg

bioelectrical impedance method in assessing body‐composition change in

obese women. Am J Clin Nutr 69, 603‐607.

Utter, J, Scraqq, R, Denny, S, & Schaaf, D (2009). Trends in body mass index and waist

circumference among New Zealand adolescents, 1997/1998‐2005. Obes Rev,

Epub ahead of print.

Vanhala, MJ, Vanhala, PT, Keinänen‐Kiukaanniemi, SM, Kumpusalo, EA, & Takala JK

(1999). Relative weight gain and obesity as a child predict metabolic

syndromes as an adult. Int J Obes Relat Metab Disord, 23, 656‐659.

Vanhala, M, Vanhala, P, Kumpusalo, E, Halonen, P, & Takala, J (1998). Relation

between obesity from childhood to adulthood and the metabolic syndrome:

population based study. BMJ, 317, 319‐320.

Waedle, J, & Boniface, D (2008). Changes in the distributions of body mass index

and waist circumference in English adults, 1993/1994 to 2002/2003. Int J Obes,

32, 527‐532.

Wang, J, & Pierson Jr, RN (1976). Disparate hydration of adipose and lean tissue

requires a new model for body water distribution in man. J Nutr, 106,

1687‐1693.

Wang, J, Thornton, JC, Bari, S, Williamson, B, Gallagher, D, Heymsfield, SB, et al.

(2003). Comparisons of waist circumferences measured at 4 sites. Am J Clin


300

Nutr, 77, 379‐384.

Wang, J, Thornton, JC, Russell, M, Burastero, S, Heymsfield, SB, & Pierson, RN (1994).

Asians Have Lower‐Body Mass Index (Bmi) but Higher Percent Body‐Fat Than

Do Whites ‐ Comparisons of Anthropometric Measurements. Am J Clin Nutr,

60, 23‐28.

Wang, J, & Wang, X (2003). Body composition of Chinese infants aged 4 months and

6 mongths by double labeled water. J Hygiene Res, 32, 487‐489.

Wang, WC, & Bachrach, LK (1996). Validity of the body mass index as an indicator of

adiposity in an ethnically diverse population of youths. Am J Hum Biol, 8,

641‐651.

Wang, ZM, Deurenberg, P, Wang, W, Pietrobelli, A, Baumgartner, RN, & Heymsfield,

SB (1999). Hydration of fat‐free mass: New physiological modeling approach.

Am J Physiol‐Endoc Metab, 276, e995‐e1003.

Watts, K, Bell, LM, Byrne, SM, Jones, TW, & Davis, EA (2008). Waist circumference

predicts cardiovascular risk in young Australian children. J Paediatr Child

Health, 44, 709‐715.

Weiss, R, Dziura, J, & Burgert, T (2004). Obesity and the metabolic syndrome in

children and adolescents. N Engl J Med, 350, 2362‐2374.

Whincup, PH, Gilg, JA, Owen, CG, Odoki, K, Alberti, KG, & Cook, DG (2005). British

South Asians aged 13‐16 years have higher fasting glucose and insulin levels

than Europeans. Diabet Med, 22, 1275‐1277.

Whincup, PH, Gilg, JA, Papacosta, O, Seymour, C, Miller, GJ, Alberti, KGMM, et al.

(2002). Early evidence of ethnic differences in cardiovascular risk: cross

sectional comparison of British South Asian and white children. BMJ (Clinical

Research Ed.), 324, 635‐639.

WHO (2004). Appropriate body‐mass index for Asian populations and its

implications for policy and intervention strtegies. Lancet, 363, 157‐163.

Wickramasinghe, VP, Cleghorn, GJ, Edmiston, KA, Murphy, AJ, Abbott, RA, & Davies,

PSW (2005). Validity of BMI as a measure of obesity in Australian white

Caucasian and Australian Sri Lankan children. Ann Hum Biol, 32, 60‐71.

Wickramasinghe, VP, Lamabadusuriya, SP, Cleghom, GJ, & Davies, PS (2009). Validity


301

of currently used cutoff values of body mass index as a measure of obesity in

Sri Lankan children. Ceylon Medical J, 54, 114‐119.

Wickramasinghe, VP, Lamabadusuriya, SP, Cleghorn, GJ, & Davies, PS (2008).

Assessment of body composition in Sri Lankan children: validation of a

bioelectrical impedance prediction equation. Eur J Clin Nutr, 62, 1170‐1177.

Williams, DP, Going, SB, Lohman, TG, Harsha, DW, Srinivasan, SR, Webber, LS, et al.

(1992). Body fatness and risk for elevated blood pressure, total cholesterol,

and serum lipoprotein ratios in children and adolescents. Am J Public Health,

82, 358‐363.

Williams, MJ, Hunter, GR, Kekes‐Szabo, T, Snyder, S, & Treuth, MS (1997). Regional

fat distribution in women and risk of cardiovascular disease. Am J Clin Nutr,

65, 855‐860.

Wong, WW, Cochran, WJ, Klish, WJ, Smith, EO, Lee, LS, & Klein, PD (1988). In vivo

isotope‐fractionation of factors and the measurement of deuterium‐ and

oxygen‐18‐dilution spaces from plasma, urine, saliva, respiratory water vapor,

and carbon dioxide. Am J Clin Nutr, 47, 1‐6.

Wong, WW, Stuff, JE, Butte, NF, Smith, EO, & Ellis, KJ (2000). Estimation of body fat in

Caucasian and African‐American girls: total‐body electrical conductivity

methodology versus a four‐component model. Int J Obes Relat Disord, 24,

1200.

World Health Organization (1995). Report of a WHO Expert Committee. Physical

status: the use and interpretation of anthropometry. Technical Report Series

no. 854. Geneva.

Wu, CH, Heshka, S, Wang, J, Pierson Jr, RN, Heymsfield, SB, Laferrère, B, et al. (2007).

Truncal fat in relation to total body fat: influences of age, sex and ethnicity

and fatness. Int J Obes, 31, 1384‐1391.

Yan, W, Yao, H, Dai, J, Cui, J, Chen, Y, Yang, X, et al. (2008). Waist circumference

cutoff points in school‐aged Chinese Han and Uygur children. Obeisty, 16,

1687‐1692.

Yanovski, JA, Yanovski, SZ, Filmer, KM, Hubbard, VS, Avila, N, Lewis, B, et al. (1996).

Differences in body composition of black and white girls. Am J Clin Nutr, 64,


302

833‐839.

Yeh, WT, Chang, HY, Yeh, CJ, Tsai, KS, Chen, HJ, & Pan, WJ (2005). Do centrally obese

Chinese with normal BMI have increased risk of metabolic disorders? Int J

Obes, 29, 818‐825.

Yoshinaga, M, Tanaka, S, & Snimago, A (2005). Metabolic syndrome in overweight

and obese Japanese children. Obes Res, 13, 1135‐1140.

Zhai, F, Zhang, L, Wang, C, Pan, H, Yu, W, & Fu, D (2004). Study of normal reference

values for bone mineral contents in children and adolescents in Beijing. J

Hygiene Res, 33, 172‐175.

Zhang, L, Liu, J, Zhai, F, Cao, R, & Duan, J (2003). Normal reference values for bone

mineral density in children and adolescents aged 6‐18 years, Beijing, China.

Chin J Osteoporos, 9, 134‐136.

Zhu, SK, Heymsfield, SB, Toyoshima, H, Wang, ZM, Pietrobelli, A, & Heshka, S (2005).

Race‐ethnicity‐specific waist circumference cutoffs for identifying

cardiovascular disease risk factors. Am J Clin Nutr, 81, 409‐415.

Zillikens, M, & Conway, J (1990). Anthropometry in blacks: applicability of

generalized skinfold equations and differences in fat patterning between

blacks and whites. Am J Clin Nutr, 52, 45‐51.

Zimmermann, MB, Gubeli, C, Puntener, C, & Molinari, L (2004). Detection of

overweight and obesity in a national sample of 6‐12‐y‐old Swiss children:

accuracy and validity of reference values for body mass index from the US

Centers for Disease Control and Prevention and the International Obesity

Task Force. Am J Clin Nutr, 79, 838‐843.

Zimmet, P, Alberti, G, Kaufman, F, Tajima, N, Silink, M, Arslanian, S, et al. (2007). The

metabolic syndrome in children and adolescents. Lancet, 369, 2059‐2061.

Zuoa, H, Shi, ZM, Hu, XS, Wu, N, Guo, ZR, & Hussain, A (2009). Prevalence of

metabolic syndrome and factors associated with its components in Chinese

adults. Metab‐Clin Exp, 58, 1102‐1108.


303

APPENDICES

Appendix 1: WHO growth reference for school‐aged children and adolescents


304

Appendix 2: Standard operating procedures ‐ deuterium dilution technique

Preparation of 10% solution

Before data collection, deuterium oxide (D2O) 99.9% concentration is diluted to a

10% solution using tap water using the following steps:

Use a volumetric flask (100 mL) to measure 100 mL 99.9% D2O;

Pour into another volumetric flask (1000 mL) and use tap water to wash the

volumetric flask (100 mL) 2‐3 times;

Use tap water for diluting to 1000 mL;

Pour the 10% solution into Schott bottles and place into autoclave and sterilize

at 120℃ for 10 mins;

Cool the solution and label (Deuterium oxide, 10% solution, date made and

dose number);

Store in fridge or at room temperature.

B. Administration of deuterium oxide

• Ask the participant to provide 10 mL pre‐dose urine sample (5 mL for

analysis and 5 mL for back‐up);

• Weigh the participant;

• Calculate the required dose as 0.5 g/kg body weight × body weight (kg);

• Weigh the cup and straw;

• Pour the 10% solution into cup and weigh the sum of the cup + straw + dose;

• Re‐weigh the cup and straw after participant drinks the dose to calculate the

actual dose given;

• Collect 10 mL post‐dose urine sample 5 h later.


305

The administration steps will be as follows:

Records for administration of deuterium oxide

Subject ID

Name Weight (kg)

Required dose

(0.5g/kg×kg)

Cup + Straw (g)

Cup + Straw + Dose (g)

Cup +

Straw (post‐drink)

(g)

Actual dose given (g)

Admini‐ stration time

Time for post‐dose urine

collection

Dose number

1126 Tan 37.2 18.6 4.921

23.766 5.016

18.750 8:19 13:19

20080515

1) Preparation of samples

Urine samples should be stored in airtight vessels and frozen until they are prepared

for analysis.

• Place 0.5 mL urine in a 10 mL labelled vacutainer tube, add a chromacol vial

with a small amount of platinum on Alumina upright to the vacutainer tube;

• Recap the tube and attach them to the sample preparation line;

5 h later

Pre‐dose urine sample

Measure weight

Weigh paper cup, straw and 10% D2O

Drink the D2O

Re‐weigh the paper cup and t

Post‐dose urine sample

Weigh paper cup and straw

Calculate required dosage

Record the time of administration

Calculate the actual drinking dosage

Pour the 10% solution into cup


306

• Close the gas inlet valve, open the vacuum tap and switch the pump on at the

wall plug;

• Open the tube valves and evacuate for 5 min (timed on the stopwatch);

• Switch off the pump and close the vacuum tap;

• Open the 2H valve at the wall, open gas inlet valve on preparation line and open

the last valve on the preparation line to allow the gas to bubble into water

filled tubes;

• Check the flow of bubbles to ensure a steady stream;

• Flush the line for 30 sec prior to use;

• Fill the tube for 10 sec and then close the sample valves, remove from the

preparation line;

• Repeat the above procedure with the remainder of the batch;

• Leave the samples at room temperature for a minimum of 72 hrs and a

maximum of 2 weeks.

D Analysis of Samples by IRMS

The 2H enrichment of the prepared samples and references are measured by isotope

ratio mass spectrometry (IRMS). Results are expressed in delta units (%) relative to

an international standard (standard mean ocean water, SMOW).

E. Total body water calculation

a) Abundance is measured in delta units relative to Standard Marine Ocean

Water (SMOW)

b) 2. Dilution space (N) is calculated as follows:

N = EpEs

xa

A

)Et(EaT

Where A is the amount of isotope given in grams, a is the portion of the dose in

grams retained for mass spectrometer analysis, T is the amount of tap water in

which the portion a is diluted before analysis, and Ea, Et, Ep and Es are the

isotopic abundances in delta units relative to SMOW of the portion of dose, the

tap water used, the pre‐dose urine sample and the post‐dose urine sample.


307

c) The TBW will be overestimated by ~4% because of deuterium exchange with

non‐aqueous hydrogen in the body (Racette et al., 1994). Therefore, the

equation needs to be modified as follows:

TBW (kg) = 1.041

N

d) FFM is then calculated as follows:

FFM (kg) = constantHydration

TBW

The hydration constant of the FFM is 0.732 for healthy adults. Lohman’s age‐ and

gender‐specific hydration constants of the FFM have commonly been used in

research with children.

e) Subsequently, FM and %BF can be calculated based on the two‐compartment

model of body composition from the FFM:

FM (kg) = Weight – FFM

%BF = 100tBody weigh

FM


308

Appendix 3: Human ethics approval certificate


309


310

Appendix 4: Contribution of the candidate to the project

The project titled “Control and prevention of childhood malnutrition in Asia

(RAS/6/050)” was funded by the International Atomic and Energy Agency (IAEA) from

2007‐2010. This project is a regional technical cooperation project and five Asian

countries including China, Lebanon, Malaysia, the Philippines and Thailand were

involved as the participant countries and two countries including Asutralia and

Janpan were involved as technical support countries. The specific roles of the

candidate in the project are outlined below.

1. From 17‐20 April 2007, the IAEA held the planning and coordination meeting

in Beijing and 18 participants from the IAEA, Australia, Japan, China, Lebanon,

Malaysia, The Philippines, Thailand and Vietnam attended the meeting. The

candidate organized and participated in the meeting on behalf of the

responsible institute in China ‐ National Institute for Nutrition and Food

Safety. During this meeting, the protocols and materials of the project was

discussed and finalized.

2. As the main investigator of the sub‐project in China, the candidate was

responsible for leading the sub‐project in China in cooperation with the

Institute for Nutrition and Food Safety, China CDC. The specific roles of the

candidate in the sub‐project included:

Attending the training course held by IAEA in Australia to learn the

techniques on body composition and total energy expenditute

assessment;

Organize and run training workshops involving all project team

members before data collection in China;

Contacting primary schools and recruitment of participants;

Organizing the field work and participating in data collection;

Analysing urine samples;

Data entry, cleaning and analysis;

Preparation of progress reports;


311

Attending progress meeting and reporting on the progress of the

sub‐project to the IAEA;

Collaborating with IAEA and other participating countries.

3. For the whole project, the candidate helped to analyse urine samples from

other participating countries. Moreover, the candidate was responsible for

the all data management associated with the project. On behalf of the

project team, the candidate reported the progress of the whole data

management and preliminary results to the IAEA and all participating

countries. Furthermore, the candidate drafted the articles and submitted to

international journals on behalf of the whole team.

body composition and its relationship to …eprints.qut.edu.au/47432/1/ailing_liu_thesis.pdf ·...

Documents