ASSESSMENT OF THE PREVALENCE OF OBESITY AND RELATED RISK FACTORS IN VIETNAMESE ADULTS LIVING IN URBAN AREAS OF HO CHI MINH CITY, VIETNAM.

Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (4.49 MB, 155 trang )

Trang 1<div class="page_container" data-page="1">

ASSESSMENT OF THE PREVALENCE OF OBESITY AND RELATED RISK FACTORS IN VIETNAMESE ADULTS LIVING

IN URBAN AREAS OF HO CHI MINH CITY, VIETNAM.

Cuong Quoc TRAN, M.D.

A thesis submitted in fulfilment of Requirement for the degree of

Master of Medical Science Faculty of Health University of Newcastle

Australia

December 2004

</div>Trang 2<div class="page_container" data-page="2">

DECLARATION

I hereby certify that the work embodied in this thesis is the result of original research and has not been submitted for a higher degree to any other University or Institution.

(Signed) ……….……….Cuong Quoc TRAN

</div>Trang 3<div class="page_container" data-page="3">

ACKNOWLEDGMENTS

I am grateful to the one thousand four hundred and eighty eight Vietnamese adults living in 13 urban districts of Ho Chi Minh City (HCMC) for their kindness and enthusiasm to spend their valuable time and effort to join the survey. My thanks also go to the local health workers, anthropometrists, interviewers, and medical lab technicians for their excellent cooperation in this survey.

For their invaluable contribution to the study as supervisors, I am grateful to Dr. Michael J. Dibley, Mr. Steve J. Bowe from the Centre for Clinical Epidemiology and Biostatistics, Faculty of Health, University of Newcastle, Australia; Dr. Hung T.K. Nguyen, Dr. Loan T.H. Tran, Dr. Hanh T.M. Tran from the Nutrition Centre HCMC Vietnam.

I also extend my thanks to Health Consequences for Population Change Program of The Wellcome Trust, United Kingdom for financial support for this study.

My thanks go to the staff of the Centre for Clinical Epidemiology and Biostatistics- University of Newcastle, Australia, the Nutrition Centre HCMC Vietnam for their ongoing support and assistance with my research.

Finally, I would like to thank my family for their never-ending support especially my mother Mai T.H. Nguyen, my wife Ha T.M. Tran and my sister Loan T.H. Tran for their love, assistance and encouragement.

</div>Trang 4<div class="page_container" data-page="4">

TABLE OF CONTENTS

Chapter 1 : Introduction... 15

Chapter 2 : Literature review... 17

2.1 Nutrition transition in Asia... 17

2.2 Levels and trends of obesity in East Asia and Southeast Asia... 19

2.2.1 East Asia and Southeast Asia:... 19

2.2.2 Vietnam... 21

2.3 Measurement of obesity and adiposity... 22

2.3.1 BMI cut-off values for underweight and obesity ... 22

2.3.2 Waist circumference cut-off values for abdominal obesity ... 26

2.3.3 Assessment of adiposity... 27

2.4 Obesity related risk factors... 32

2.4.1 Socio-economic risk factors... 32

2.4.2 Cardio-vascular risk factors ... 33

2.4.3 Assessment of physical activity: ... 34

2.4.4 Assessment of diet in adults... 36

2.5 Weight perception, weight concern and weight control behaviour... 41

Chapter 3 : Methodology ... 43

3.1 Study design... 43

</div>Trang 5<div class="page_container" data-page="5">

3.2 Study population and setting... 43

3.5.2 Pilot test of instruments... 51

3.5.3 Data collection schedule ... 51

3.6 Data analysis ... 52

3.7 Ethical considerations: ... 55

</div>Trang 6<div class="page_container" data-page="6">

Chapter 4 : Receiver operating characteristic (ROC) analysis to determine optimal cut-off values for anthropometric indices to identify populations with increased risk of diabetes or

5.3.1 Socio-demographic and clinical characteristics of the sample ... 80

5.3.2 Prevalence of Overweight and Obesity... 84

</div>Trang 7<div class="page_container" data-page="7">

6.3 Results... 96

6.4 Discussion ... 100

Chapter 7 : Conclusions and recommendations... 103

Annex 1: Information statement, script and consent form for participants...113

Annex 2: 95% CI of d0, lower and upper level of intermediate range of individual risk factors and pooled risk factors for BMI, waist circumference, WHtR and WHpR ...118

Annex 3: Questionnaires ...122

</div>Trang 8<div class="page_container" data-page="8">

LIST OF TABLES

Table 2-1 Prevalence of overweight and obesity among Asian populations ... 20

Table 2-2 Traditional cut-off values (36) for BMI... 24

Table 2-3 Suggested BMI cut-off values for defining obesity in Asian populations from different countries and the World Health Organisation ... 25

Table 2-4 Recommended BMI cut-off values for Asian populations from WHO (7) ... 26

Table 2-5 Traditional waist circumference cut off values for defining abdominal obesity . 27 Table 2-6 Suggested waist circumference cut off values for defining abdominal obesity in Asian populations... 27

Table 2-7 Body fat range for adults ... 31

Table 2-8 Socio-economic risk factors of obesity in men in different countries ... 32

Table 2-9 Obesity socio-economic risk factors in women in different countries ... 33

Table 2-10 Anthropometry and CVD risk factors for men from different populations in Asia ... 34

Table 2-11 Anthropometry and CVD risk factors for women from different populations in Asia ... 34

Table 2-12 Physical activity level from different countries using IPAQ... 36

Table 2-13 Reproducibility of the FFQ: Mean daily intake and standard deviation (SD) of energy and nutrients (n=118) ... 40

Table 2-14 Mean daily intakes and standard deviation (SD) of energy and nutrients, assessed by 24HRs and FFQ2, and Pearson correlation coefficients between the two methods (n=118). ... 41

</div>Trang 9<div class="page_container" data-page="9">

Table 2-15 Weight concern and weight control behaviour among overweight (BMI>25 Table 4-1 Gender-specific anthropometric indices and cardiovascular risk factors (mean ± SD) among Vietnamese adults [n=1488] ... 64 Table 4-2 Area under the ROC curve (AUC) of individual risk factors and pooled risk

factor for BMI, waist circumference, waist to height ratio and waist to hip ratio ... 66 Table 4-3 Optimal BMI, waist circumference cut-off values to detect increased

cardiovascular disease and diabetes risk calculated using TG-ROC method among Vietnamese adults living in HCMC. ... 67 Table 4-4 Optimal waist to height ratio, waist to hip ratio cut-off values to detect increased cardiovascular disease and diabetes risk calculated using TG-ROC method among Vietnamese adults living in HCMC. ... 68 Table 4-5 Age-specific prevalence (%) of overweight defined by BMI (n=1488), waist

circumference (n=1488) using cut-off values from ROC-analysis+ among Vietnamese adults ... 73 Table 4-6 Table comparison results of studies assessing BMI cut-off values for

cardiovascular risk factors using sensitivity and specificity in Asian populations... 76 Table 5-1 Sociodemographic of 1488 Vietnamese adults aged 20-60 years in HCMC,

Vietnam 2004... 82 Table 5-2 Clinical characteristics of 1488 Vietnamese adults aged 20-60 years in HCMC, Vietnam 2004... 83

</div>Trang 10<div class="page_container" data-page="10">

Table 5-3 Age-specific prevalence (%) of overweight defined by traditional BMI and waist circumference cut-off values+ among Vietnamese adults (n=1488). ... 85 Table 5-4 Age-specific prevalence (%) of overweight defined by BMI and waist

circumference using cut-off values for Asian populations+ among Vietnamese adults (n=1488)... 86 Table 5-5 : Age-specific prevalence (%) of underweight defined by BMI among

Vietnamese adults (n=1488) ... 87 Table 5-6 Gender-specific prevalence (%) of overweight [BMI ≥ 23 kg/m2] according to

the household wealth status among Vietnamese adults (n=1488) ... 88 Table 5-7 : Gender-specific prevalence (%) of abdominal overweight* according to the

household wealth status among Vietnamese adults (n=1488) ... 88 Table 5-8 Gender-specific weight perception, weight concern and weight control behaviour by BMI. ... 89 Table 6-1 Association between overweight (measured using BMI+) and potential risk

factors among Vietnamese male adults [n=717]... 98 Table 6-2 Association between overweight (BMI+) and potential risk factors among

Vietnamese female adults [n=771]. ... 99

</div>Trang 11<div class="page_container" data-page="11">

LIST OF FIGURES

Figure 4-1 Relationships between BMI and individual and pooled risk factors... 65 Figure 4-2 ROC curves for BMI, waist circumference, waist to height and waist to hip ratio to detect pooled cardiovascular risk factors ... 69 Figure 4-3 Combined ROC curves for BMI, waist circumference, waist to height and waist to hip ratio to detect pooled cardiovascular risk factors ... 70 Figure 4-4 Sensitivity and specificity curves for BMI, waist circumference to detect pooled cardiovascular risk factors ... 71 Figure 4-5 Sensitivity and specificity curves for waist to height and waist to hip ratio to

detect pooled cardiovascular risk factors ... 72

</div>Trang 12<div class="page_container" data-page="12">

SYNOPSIS

Objective: To determine optimal cut offs for anthropometric indicators to identify

increased cardiovascular disease and diabetes risks, and to assess the prevalence of overweight and obesity and obesity related risk factors among Vietnamese adults living in urban areas of Hochiminh City (HCMC), Vietnam.

Design: This cross sectional survey was conducted in the local health stations of 30

randomly selected wards, which represent all 13 urban districts of HCMC, over a period of two months from March to April 2004.

Participants: 1488 participants aged 20 to 60 years completed the interview, physical

examination, and venous blood collection.

Main outcome measures: Anthropometric indicators of body mass index (BMI), waist

circumference, waist to height and waist to hip ratio and percent body fat (derived from measurements of bioelectrical impedance analysis and skinfold thicknesses), systolic and diastolic blood pressure, dietary intake measured with a validated food frequency questionnaire, physical activity measured with the IPAQ, weight perception, weight concern, weight control behaviour, and biochemical indicators of cardiovascular disease and type 2 diabetes risk (lipid profile and fasting blood glucose).

Results: ROC analysis based on increased cardiovascular and diabetes risk revealed

optimal BMI cut-off values to define overweight of 23 kg/m2 for men and women, and obesity of 26 kg/m2 in men and 27 kg/m2 in women. The optimal waist circumference cut-off values to define abdominal overweight were 79 cm for men and 77 cm for women, and for abdominal obesity were 86 cm for men and women.

The age and sex standardized prevalence of overweight and obesity using the WHO recommended BMI cut offs of 23.0 and 27.5 kg/m2 was 26.2% and 6.4% respectively. The prevalence of overweight and obesity was slightly higher in females (33.6%) than males (31.6%), and progressively increased with age. Vietnamese men of middle age, of high economic status, having sedentary occupations, were at greatest risk of overweight and

</div>Trang 13<div class="page_container" data-page="13">

obesity. While, women of middle age, having low education level were at greatest risk of overweight and obesity.

The age and sex standardized prevalence of underweight (BMI <18.5 kg/m2) among Vietnamese adults living in HCMC was 20.4%. The prevalence was slightly higher in males (22.0%) than females (18.9%), and there was a much higher prevalence in all underweight categories in younger women than men but this was reversed for older men.

Conclusions: The adult population in HCMC Vietnam is in an early “nutrition transition”

with approximately equal prevalence of low and high BMI. The prevalence of overweight and obesity of Vietnamese urban adults was lower than that reported other East and Southeast Asian countries. The optimal cut-off values for BMI from ROC analysis were similar to those recommended by WHO for Asian populations. However, the optimal cut-off values for waist circumference were lower than those recommended by WHO for Asian populations and lower than those reported for others countries in the region.

</div>Trang 14<div class="page_container" data-page="14">

ABBREVIATION

WHtR : Waist to height ratio WHpR : Waist to hip ratio

TG-ROC-analysis : Two-graph receiver operating characteristic analysis BIA : Bioelectrical impedance analysis

CVD : Cardiovascular risk factors. FFQ : Food frequency questionnaire AUC : Area under the curve

WHO : World Health Organization

IPAQ : International physical activity questionnaire IOTF : International Obesity Task Force

IASO : International Association for the Study of Obesity

</div>Trang 15<div class="page_container" data-page="15">

Chapter

urrently many countries in Asia, including developing countries, are undergoing a “nutrition transition” (1, 2), which is due to rapid social and economic development and urbanization. These changes include demographic, nutritional and epidemiological transitions. In a typical nutrition transition society both under and over nutrition co-exist.

1 : Introduction

C

Vietnam is no exception and exhibits evidence of an early “nutrition transition”. About ten years ago, malnutrition was the main public health nutrition problem of Vietnam and over nutrition almost did not exist. But in recent years, the economic status of Vietnam has improved significantly following the introduction of the “Doi moi”, social and economic policy reforms, and the lifting of the embargo on Vietnam by the US government. Malnutrition has now partially been solved; however in the meantime a new problem of over nutrition has emerged, especially in urban areas. Preliminary evidence of an emerging problem with over nutrition has been found in 2000 nutrition surveillance data where the prevalence of overweight or a body mass index (BMI) >25 kg/m2 of Vietnamese adults (> 15 years) living in HCMC was 12.9%, and the prevalence of overweight of women (15 to 49 years) was 9.7% (3). Furthermore a survey of 300 Vietnamese aged 40-60 years living in HCMC reported the prevalence of overweight (BMI ≥ 25kg/m2) adults in urban, suburban and rural areas of the city was 18%, 13% and 6% respectively (4). However to date there has been no detailed assessment of obesity and overweight of adults in HCMC and an identification of the associated risk factors. This information is needed to formulate appropriate public health policies and to plan interventions to prevent the development of an obesity epidemic with continued economic development.

In 2002 Vietnam formulated its first national program for prevention of chronic non-communicable diseases. The findings from this study provide an evidence basis for the development of the national program for prevention of chronic non-communicable diseases especially in the area of prevention of obesity and appropriate dietary recommendations for healthy living.

</div>Trang 16<div class="page_container" data-page="16">

The findings from the study should also be of use to the Nutrition Centre as the key Health Department organization responsible for planning and developing local policies and programs for obesity prevention in HCMC. In particular, the results will help the Nutrition Centre plan future research and evaluations of interventions to prevent overweight, and other related chronic non-communicable diseases, and to plan communication strategies about nutrition and prevention of chronic non-communicable diseases in HCMC.

This thesis aims to assess the prevalence of obesity and related risk factors in Vietnamese adults aged 20-60 years living in urban areas of HCMC Vietnam, using newly defined optimal cut-off values of some important anthropometric indices for Vietnamese population. Then prevalence of obesity based on these cut-off values will be presented following by factors associated with high BMI in Vietnamese adults living in HCMC. The thesis has 7 chapters and is organized as follows: Chapter 2 provides a review of the relevant literature including an assessment of the levels and trends in overweight and obesity in adults in East and Southeast Asia. Chapter 3 describes the main survey methods including sampling, measurements, data collection and data analysis. Chapter 4 uses ROC analysis to determine optimal cut-off values for anthropometric indices to identify populations with increased risk of diabetes or CVD risk factors. Chapter 5 examines the prevalence of overweight and obesity in this urban adult population in Vietnam using traditional recently recommended cut offs for Asian populations. Chapter 6 uses a multi-variable logistic model to assessment factors associated with high BMI. Chapter 7 summarizes the key findings and proposes conclusion and recommendation that can be drawn from the results.

</div>Trang 17<div class="page_container" data-page="17">

Chapter 2 : Literature review

2.1 Nutrition transition in Asia

“Nutrition transition” is a term coined to describe the changes in diet and nutritional status that take place as communities in developing countries develop economically and urbanize. Nutrition transition has been reported in Asia and the Pacific as well as in the developing world. Evidences of nutrition transition can be found in several countries including China, Sri Lanka, Thailand, Malaysia, South Korea (1, 5, 6). It begins with dietary change in the population from traditional diets rich in fruits and vegetables to a diet full of pre-processed foods, foods of animal origin, and foods that contain more sugar, fat and alcohol. In addition, urban populations tend to do less physical activity in work and leisure due to the many convenient domestic and work related labour saving devices and mechanized transportation. During last half century, there have been large shifts of populations from rural to urban areas throughout the developing world. (1)

A further factor contributing to the increase in obesity and chronic diseases is the possibility of “foetal programming”. Firstly, under-nutrition of the foetus during pregnancy and of young children, which are both common in developing countries, can cause metabolic changes that are required to adapt to the nutritional stress during these periods of life. These metabolic changes help the foetus or young child to survive in the low energy environment. Such processes themselves do not directly lead to increasing morbidity and mortality, but rather, leave these children (and later adults) susceptible to obesity, adult-onset diabetes and cardiovascular diseases when faced with a richer, more energy dense diet and reduced physical activity (1).

As a result of changes in diet and physical activity as well as foetal programming, the number of overweight and obese people has increased especially in urban environments in developing countries. In these ‘nutrition transition’ countries there are new patterns of morbidity with increasing diet-related chronic diseases (coronary heart disease, diabetes, hypertension, stroke and certain cancer such as gallbladder cancer, breast cancer, colon and

</div>Trang 18<div class="page_container" data-page="18">

rectal cancer) existing concurrently with long-established problems (nutritional deficiencies and infectious diseases). As the nutrition transition develops obesity becomes a problem of both urban poor as well as rich, rather than urban rich alone. A key characteristic of the nutrition transition is the existence of both under and overweight in the same population (7). This may also be true even at a family level. There are approximately 3-15% Asian households that include both under and over weight individuals, typically, an underweight child and overweight non-elderly adults. In other words, there are 30-60% of households where a household member is underweight but another is overweight (1).

The changes in morbidity and mortality patterns that occur with a nutrition transition put greater burdens on society in terms of both human and economic costs. The human costs have been quantified in terms of disability and death from five diet-related chronic diseases including cardiovascular diseases, diabetes, hypertension, stroke and cancer. The economic costs include the cost of hospital resources to treatment adults with these conditions, the premature mortality from these conditions, the costs of lost work output whether from lost workdays or from lowered productivity by those with these chronic diseases who continue to work. To date, however, there are almost no estimates of the economic costs of this expanding chronic disease burden in developing countries (1).

The establishment of appropriate programs to control epidemic diet-related chronic diseases in rapidly developing countries are major challenges for governments and policies makers in these countries. A number of different strategies are worth considering when formulating policies in response to a nutrition transition. These include promoting agriculture development to encourage vegetable and fruit production; developing price mechanisms to solve the micronutrient deficiencies problems by increasing fruit and vegetable intake and other food sources of micronutrients; supporting the preservation of traditional diets for example by proving training on traditional cooking methods for newly married women; using mass media and school-based programs to promote good nutrition, local foods and regular physical activity (1).

</div>Trang 19<div class="page_container" data-page="19">

2.2 Levels and trends of obesity in East Asia and Southeast Asia

2.2.1 East Asia and Southeast Asia:

In this section, the levels and trends of obesity are presented with a focus on studies from East Asia and Southeast Asia, where the populations are similar to Vietnam. There is strong evidence of problems with overweight and obesity in many countries in East Asia and Southeast Asia, and some examples are presented in Table 2-1. As can be seen from Table 2-1, currently there are many countries where over 20% of the adults are overweight (BMI ≥ 25kg/m2). These numbers would be even higher if the prevalence was based on the recently proposed lower BMI cut-off values (BMI ≥ 23 and 27.5 kg/m2 for overweight and obesity respectively) for Asian populations (8). In addition to current data, the table also presents the results of earlier surveys of overweight and obesity in these selected countries, and in many reveals trends of increasing overweight and obesity. Over the 10 year period, the prevalence of overweight increased by over 5% in China, Thailand and Malaysia. In China, for instance, the level of overweight in adults increased from 14.9% in 1992 to 18.6% in 2001. However, the prevalence of overweight and obesity seems to be settled at a relatively high level (over 20%) in some countries like Japan, Hong Kong, Singapore.

</div>Trang 20<div class="page_container" data-page="20">

Table 2-1 Prevalence of overweight and obesity among Asian populations

Nation Previous data Most recent data

</div>Trang 21<div class="page_container" data-page="21">

2.2.2 Vietnam

Evidence of obesity in Vietnam has been found mainly in Hanoi and Hochiminh City which are the two largest cities in the north and the south of Vietnam respectively.

2.2.2.1 Hanoi (Capital), north of Vietnam

There is only one study reporting the prevalence of obesity in adults in Hanoi. This study assessed the prevalence of overweight and related risk factors in retired women and housewives aged 20-59 years living in a rural district of Hanoi (22). The author found that the prevalence of overweight (BMI ≥ 25 kg/m2) and obesity (BMI ≥ 30 kg/m2) was 15.5% and 1.1% respectively. Overweight increased with age and reached a peak at 50-59 years (19.9%). This study also indicated that central adiposity was a problem with 33.6% having a high waist to hip ratio (WHpR) (define as WHpR>0.85 for women) and 35.3% women had high waist circumference (define as ≥ 80cm).

2.2.2.2 Hochiminh City, south of Vietnam

A study about the nutritional status of children under 5 years and women of reproductive age (15-49 years) in HCMC found that the prevalence of obesity (defined as BMI ≥ 25 kg/m2) in women of reproductive age was 9.7%, and that it increased with age from 6.6% in women at 20-29 years, to 7% at 30-39 years, and to 16.9%, at 40-49 years (3).

An epidemiological survey of diabetes in the population over 15 years in HCMC in 2001(23) found the overall prevalence of overweight and obesity (BMI>25 kg/m2) was 12.9%. This included 11.3% with a BMI between 25-30 kg/m2, 1.3% with BMI 30-35 kg/m2 and 0.2% with BMI >35 kg/m2. In this study, the prevalence of overweight was relatively low in the sub-population aged 20-29 years (5.3%) compared with 16.3% for 30-39 years, 21.2% for 40-49 years, and 22.1% for 50-59 years. The survey also found that the overweight rate was significantly higher in urban areas (18.3% for urban versus 7.6% for rural areas, p<0.05). The prevalence of high waist to hip (defined as WHpR>0.85 in female and >0.95 in male) was significantly higher in the high BMI group (73.5%) in comparison with the normal BMI group (34.4%).

</div>Trang 22<div class="page_container" data-page="22">

Also this survey (23) found the overall prevalence of under-weight (BMI <18.5) was 27%. Under-weight was predominant in age groups 15-19 years and 20-29 years (51.6% and 36.2% respectively). Under-weight decreased with age and was 18.5%, 14.8% and 13.6% in age groups 30-39, 40-49 and 50-60 years respectively. Under-weight in rural areas was higher than in urban areas (38.3% versus 24.9%).

A study assessing the nutritional status of middle aged (40-59 years) Vietnamese adults in HCMC, reported a prevalence of overweight (BMI ≥ 25 kg/m2) in urban areas of 18% (4).

2.3 Measurement of obesity and adiposity

A number of anthropometric indices and indicators have been proposed to assess obesity in epidemiological studies. These indices are based on the anthropometric measurements of height, weight and waist circumference. The indices include body mass index (weight/(height)2), waist to hip ratio, and waist to height ratio. The most widely used indices are body mass index (BMI) and waist circumference. Although BMI can misclassify some subjects as having a high total body fat, such as athletes, very short and very tall persons, it is easy to calculate and generally correlates highly with adiposity. Thus high BMI has been widely used to measure obesity in adults in population studies (16). Waist circumference, on the other hand, has been considered a good tool for measuring abdominal visceral fat (24),(25-27). Finally, BMI and waist circumference have been shown to be directly related to health risk and death rates in many population in different cross sectional and longitudinal studies (8, 24, 28-30),(31).

2.3.1 BMI cut-off values for overweight and obesity

There are well-established BMI cut-off values of >25 kg/m2 and >30 kg/m2 for defining overweight and obesity in Western populations (Table 2-2). However, these cut-off values may not be appropriate for Asian populations. Many studies in Asia have found a higher risk of diabetes mellitus, cardiovascular diseases, percent body fat compared with Western populations at a given level of BMI (8, 32). This implies lower cut-off values for BMI are

</div>Trang 23<div class="page_container" data-page="23">

needed to define overweight and obesity for Asian populations in comparison to Western populations.

Many studies have attempted to find optimal BMI cut-off values to define overweight and obesity in Asia. The two most common approaches to identify BMI cut offs have been comparisons of BMI with the level of body fat (8, 32), (33) and with indicators of cardiovascular diseases (34, 35). In the body fat approach, optimal BMI cut-off values have been calculated based on the assumption that the percentage of body fat in Asians at the cut-off values for overweight and obesity should be the same as the percentage of body fat in Caucasians with a BMI of 25 and 30 kg/m2. Initially the percent body fat of the studied population has been measured using a “gold standard” body composition method. Then optimal cut-off values for BMI have been calculated based on a body fat prediction equation for Caucasians (BF% =1.2 x BMI + 0.23 x age – 10.8x sex - 5.4)(32). In the cardiovascular disease approach, optimal BMI cut off values for a given population have been calculated based on individual and pooled cardiovascular risk factors including hypertension, blood glucose, and lipid profiles using different methods including using odds ratio and logistic regression(36), sensitivity and specificity analysis and population attributable risk percentage (34) and receiver operating characteristic (ROC) analysis (35)(Table 2-3).

Different optimal BMI cut-off values for defining overweight and obesity have been reported in different countries Table 2-3. In general, the optimal BMI cut-off values to define overweight are lower than those observed for Western populations. They vary from 21 to 25 kg/cm2 in different countries. Similarly, optimal cut-off values for defining obesity vary from 25 to 28 kg/cm2. These differences are explained by different body build, ethnicity, degrees of urbanisation, socio-economic status and stage of nutrition transition among Asian countries.

Table 2-2 presents the traditional (Western) cut-off values for BMI recommended by WHO for comparisons between studies. However, WHO has also proposed a modified BMI classification for public health and clinical use for defining overweight and obesity in Asian populations (Table 2-4). In this definition, the cut-off values for underweight and

</div>Trang 24<div class="page_container" data-page="24">

normal weight are similar to those recommended for Western populations. The differences are with the cut-off values for defining overweight (> 23kg/m2) and obesity (> 27.5kg/m2), which are lower than those for Western populations. Whenever possible, Asian countries should use both BMI categories in reporting prevalence of overweight or obesity to facilitate international comparisons.

Table 2-2 Traditional cut-off values (37) for BMI

Classification BMI (kg/m2) Risk of co-morbidities

</div>Trang 25<div class="page_container" data-page="25">

Table 2-3 Suggested BMI cut-off values for defining obesity in Asian populations from different countries and the World Health Organization

Population Overweight Obesity Reference Year Analytical method Thai 22.1 27.0 (33) 1998 Comparison with Caucasian populations using prediction equation for percentage

body fat from BMI, age and sex.

Comparison with Caucasian populations using prediction equation for percentage body fat from BMI, age and sex. Singaporean 21.0 27.0 (32) 2000 Comparison with Caucasian populations using prediction equation for percentage

body fat from BMI, age and sex. Hong Kong

Comparison with Caucasian populations using prediction equation for percentage body fat from BMI, age and sex. Hong Kong

Chinese 23.0-24.3

Not

reported (39) 1999

ROC analysis to examine the relationship between anthropometric indices and cardiovascular risk factors.

Examined relationship between anthropometric indices and

cardiovascular risk factors using OR and logistic regression

Hong Kong

Examined relationship between anthropometric indices and

cardiovascular risk factors using partial correlation and covariance analysis, and OR and logistic regression

Mainland

Analysis based on sensitivity, specificity for cardiovascular risk factors and population-attributable risk percentage.

ROC* analysis to examine the relationship between anthropometric indices and cardiovascular risk factors with optimal cut-off value determined by equal sensitivity and specificity

Hong Kong

Chinese 23.4

Not

reported (41) 2003

ROC* analysis to examine the relationship between anthropometric indices and cardiovascular risk factors with optimal cut-off value determined by largest sum of sensitivity and specificity

ROC* analysis to examine the relationship between anthropometric indices and cardiovascular risk factors but exact optimal cut-off value not determined

* ROC: receiver operating characteristic analysis.

</div>Trang 26<div class="page_container" data-page="26">

Table 2-4 Recommended BMI cut-off values for Asian populations from WHO (8)

Classification BMI (kg/m2) Risk of co-morbidities

2.3.2 Waist circumference cut-off values for abdominal obesity

It is not only the amount of fat but also its distribution that determines the health risks associated with obesity. Abdominal or viscera fat (android obesity) is strongly associated with the cardiovascular risk factors, which include type 2 diabetes, impaired glucose tolerance, hypertension and dyslipidemia (high triglyceride, low HDL cholesterol) (16). Waist circumference is reported as a simple clinical measure that can reflect visceral fat, and thus is also an important anthropometric index in terms of body fat distribution. Similar to the situation with BMI, there are well-established cut-off values of waist circumference for defining overweight and obesity for Western populations (see Table 2-5). However, these values do not seem to be appropriate for Asian populations, which have higher risks for diabetes mellitus and CVD at lower waist circumference cut-off values (16). As with BMI there have been many studies that aimed to determine appropriate waist circumference cut-off values for Asian populations. Some examples of optimal waist circumference cut-off values reported in different studies are presented in Table 2-6. The optimal waist circumference cut-off values varied from 80.5 to 90.0 cm for men and from 71.5 -85 cm for women.

</div>Trang 27<div class="page_container" data-page="27">

Table 2-5 Traditional waist circumference cut off values for defining abdominal obesity

Classification Waist circumference (cm)

There are many methods of assessment of adiposity in humans ranging from the most commonly used methods like the combination of weight and height (BMI), skinfold thickness, and body circumferences to newer methods based on electrical resistance or impedance, densitometry, dual energy X-ray Absorptionometry (DEXA), magnetic resonance imaging and computer-assisted tomography. Densitometry is regarded as the gold standard method for measuring body fat (45). However, due to its cost and complexity of instrumentation, densitometry is limited to use in laboratory rather than large-scale population-based studies.

Combinations of weight and height (BMI), skinfold thickness and bioelectrical impedance analysis (BIA) are the most commonly used methods of measurement body composition in epidemiological studies (45-47). These methods are popular because they are simple to implement, require less time than other methods, are relatively low cost and

</div>Trang 28<div class="page_container" data-page="28">

use portable instruments. BMI has been discussed above. In this section, skinfold thickness and bioelectrical impedance analysis methods will be examined in detail.

2.3.3.1 Skinfold thickness measurements

Skinfold thickness is the second most widely used method to determine body composition since it provides a direct measure of body fat. Source of variation in skinfold thickness measurements include differences in the site of measuring of each skinfold, the manner in which the skinfold is picked up and the depth of calliper bite (45). The correlation of skinfold thickness with densitometry (which is consider as the gold standard method for body composition(45)) is similar to the correlation of BMI with densitometry, and in some reports even higher (48-50).

Skinfold thickness is actually the thickness of two skinfolds and the adipose tissue under the skin at each measurement site. Usually, skinfold-callipers, which have two pincers, are used to measure the skinfold thickness. During measurement, the two pincers of the skinfold callipers are applied at a pre-determined force to the skinfold and the thickness of the two skinfolds under the callipers are obtained from the meter on the body of the calliper. There are numerous types of skinfold callipers, but Harpenden callipers are considered to be the most accurate.

Skinfold measurements taken at several different sites can be used in equations to estimate total body fat. These equations are based on studies in which skinfold thickness measurements have been validated against a “gold standard” of body fat measurement, such as densitometry. Measuring skinfold thickness at several sites also helps to evaluate the pattern of distribution of body fat. The Durnin-Womersley and Jackson-Pollock-Ward equations are amongst the most commonly used to estimate body fat but a number of other equations have been reported in Norton et al 1996 (51). However, these prediction equations have only been validated for Western populations and so have limited application for Asian populations due to the difference in body composition and body build. Several studies have pointed out that these prediction equations may not be appropriate for Asian populations and that specific equations may be needed for different ethnic groups in Asian

</div>Trang 29<div class="page_container" data-page="29">

(46, 52, 53). To date, there are no published body fat prediction equations that have been validated for Asian populations in general. However, there are reports of body fat prediction equations that have been developed in specific populations and countries (54-57).

The body fat prediction equations developed in Japanese populations (55, 56) seem to be the most appropriate for use in Vietnam and have been used in the analyses because they were developed in a population of similar age (19-60 years) and using the skinfold sites (triceps, subscapular and abdominal) used in our study:

For women: body density = 1.07931-0.00059 x sum of three skinfolds (mm)-0.00015 x age For men: body density = 1.09556-0.00062 x sum of three skinfolds (mm)-0.00028 x age

The usual sites of skinfold thickness measurements are: subscapular, midaxillary, pectoral or chest, abdominal, suprailiac, front thigh, suprapatellar, medial calf, triceps, biceps and forearm. There are some specific sites for men and women due to differences in body fat distribution. The specific sites for women are triceps, suprailiac and front thigh; and for men are: chest, abdominal and thigh.

The main advantages of the skinfold method to assess body fat are the ease with which measurements can be taken, the non-invasive nature of the measurements, the speed and low cost of the measurements. The disadvantages of the method include the measurement error, the focus of the measurements being only on subcutaneous adiposity, and the limitations of the method for the extremely under or overweight.

The following measurement guidelines (51, 58, 59) are recommended:

- Skinfold thickness should be measured on the right hand site of the body at all sites.

- At each site, two repeated measurements are needed, if the difference between these measurements is greater than one millimetre or 10% in obese subjects, then a third measurement is needed.

</div>Trang 30<div class="page_container" data-page="30">

- After applying the calliper on the skin, the results should be read after 3 seconds. If the calliper is left applied for more than three seconds, the value of the skinfold thickness will decrease because the fluid in the skinfold under the calliper will begin to escape.

- The calliper should be held in the right hand of the measurer, and the left hand should grasp the skinfold about one centimetre away from the measuring site throughout the measurement.

- Care should be taken to avoid pinching the muscle layer under the skinfold when measuring the skinfold thickness.

2.3.3.2 Bioelectrical Impedance Analysis

Bioelectrical impedance analysis is a method in which four electrodes are attached to the body (in arms or legs). Undetectable low voltage electric current (500 micro Amps, 50 kHz) is sent through these electrodes to the body and passes through the water compartment of the body. The electric current leaves two electrodes and returns to the instrument via the other two electrodes. Since fat is a very poor conductor of electricity, a high level of body fat will impede the current more than a high level of lean tissue. By measuring the resistance to the current, the machine estimates the percent body fat.

The accuracy of bioelectrical impedance analysis is approximately ± 3% if done correctly on properly operating equipment. However, accuracy is dependant upon several subject-based variables especially hydration status of subject. Dehydration of the subject is a potential problem of this method, and to avoid it the following measurement guidelines are recommended: subject should abstain from eating and drinking for four hours prior to the test; avoid exercising for 12 hours prior to the test, not drink alcohol for 48 hours prior to the test; void (urinate) completely prior to testing; and avoid taking diuretics prior to testing unless instructed to do so by a physician. In large-scale population based studies it would be difficult to adhere to these measurement guidelines.

</div>Trang 31<div class="page_container" data-page="31">

BIA method has several advantages including: a requirement for little or no technical knowledge of the operator or the client; the short time for testing which is usually less than a minute; the portability of the unit which can be easily moved from place to place; the simple requirement of only an electrical outlet and the machine itself. On the other hand, it has several disadvantages: it has a higher standard error range; it tends to consistently overestimate lean people and underestimate obese people; and the accuracy of BIA is very dependant on multiple factors which may be hard to control for some people.

Usually the BIA instrument has two arm-electrodes and two leg-electrodes which are attached to one arm and one leg on the same body side. This method of attachment of electrodes yields good electrical impedance results for the whole body. However, for convenience some manufacturers have been incorporated BIA instruments in body scales. In this type of BIA instrument, all four electrodes are placed on the standing plate of the scale, and are attached to subject’s feet. This type of instrument is less accuracy compared with the usual four-electrode instrument, because it tends to measure electrical impedance of the low body part in stead of the whole body.

Percent body fat cut-off values for defining low body fat, healthy body fat, high body fat and obesity differ by gender and age group. Women normally have a higher percentage of body fat compared to men. Similarly, a high percentage of body fat is normally expected as age increases (details can be found in Table 2-7).

Table 2-7 Body fat range for adults (Caucasian)

20-39 40-59 60-79 20-39 40-59 60-79

Low body fat 0 – 7.9 0 – 10.9 0 – 12.9 0 - 20.9 0 – 22.9 0 – 23.9 Healthy body fat 8 – 18.9 11 – 21.9 13 – 24.9 21 – 32.9 23 – 34.9 24 – 35.9

High body fat 19 – 24.9 22 – 26.9 25 – 29.9 33 – 38.9 35 – 39.9 36 – 41.9

</div>Trang 32<div class="page_container" data-page="32">

2.4 Obesity related risk factors

2.4.1 Socio-economic risk factors

A variety of socio-economic risk factors have been reported to be associated with obesity in different countries, and examples from a selection of countries are presented in Table 2-8 and Table 2-9. Smoking status, occupation, education, level of dietary energy intake, level of physical activity, wealth status, television viewing, residence (urban versus rural) and marital status have all commonly been associated with obesity in adults. For some factors, the patterns of association are reversed between Western and Asian countries. For example, obesity is associated with people who have low levels of education, low economic status in Western countries, but in Asian countries, the association is reversed with more obesity in people who are successful, have high-education and high socio-economic status.

Table 2-8 Socio-economic risk factors of obesity in men in different countries

Australia Low education, high-skill occupation, frequent television viewing (60) France Low education level, ex-smoker, marital status (married), high energy intake, low carbohydrate intake. (61) Taiwan Residing location (mountainous), metabolic equivalent (MET) score, high alcohol consumption (62)

China

Marital status (married), higher education, non-smoker, low active occupation, high energy intake, high carbohydrate intake, high protein

intake, high fat intake, low commuting physical activity, higher socio-economic class

(63)

</div>Trang 33<div class="page_container" data-page="33">

Table 2-9 Obesity socio-economic risk factors in women in different countries

Australia education level, country of birth (Australia and New Zealand), high Non-smoker, low physical activity, frequent television viewing, low income,

(60)

France

Low education level, use of oral contraceptives, menopausal status (yes), young age at menarche (9-11 year), high energy intake, low carbohydrate

Taiwan Location of residence (mountainous), low education level (62)

China energy intake, low carbohydrate intake, high commuting physical activity, Marital status (married), higher education, low active occupation, high higher socio-economic class

(63)

2.4.2 Cardio-vascular risk factors

Again, the diversity of Asian populations mentioned earlier is reflected in the anthropometry, lipid profile and blood pressure of populations from different Asian countries (some examples can be found in Table 2-10 and Table 2-11). Mean height varied from 167.0 to 169.8 cm for men and from 154.3 to 157.5 cm for women. Similarly, mean weight varied from 67.5 to 69.2 kg for men and from 53.8 to 62.5 kg for women. Gender difference was also seen for height, weight, waist circumference and blood pressure.

</div>Trang 34<div class="page_container" data-page="34">

Table 2-10 Anthropometry and CVD risk factors for men from different populations in Asia

Countries Age Height weight BMI Waist TC HDL LDL TG SBP DBP Ref.

Table 2-11 Anthropometry and CVD risk factors for women from different populations in Asia

Countries Age Height weight BMI Waist TC HDL LDL TG SBP DBP Reference TC: total cholesterol (mmol/L), HDL: high density lipoprotein cholesterol (mmol/L), LDL: low density lipoprotein cholesterol (mmol/L), TG: triglyceride (mmol/L), SBP: systolic blood pressure (mmHg), DBP: diastolic blood pressure (mmHg).

2.4.3 Assessment of physical activity:

Physical activity is defined as “any bodily movement produced by skeletal muscles that results in energy expenditure”(65). However, assessment and measurement of physical activity is much more difficult than its definition. Physical activity can occur in multiple contexts for different purposes like transportation, occupation (paid or unpaid), household maintenance or child care tasks, and recreation (or leisure time) (66). This makes physical activity as difficult to measure as dietary intake.

There are many ways of measuring physical activity. These methods can be grouped into 5 categories: behavioural observation, physiological markers (like heart rate), calorimetry, motion sensors, and questionnaires (including diaries, recall questionnaires and interview) (67). Each method has its advantages and disadvantages. Only the use of

</div>Trang 35<div class="page_container" data-page="35">

questionnaires for assessment of physical activity is reviewed below because this was the method used in the thesis research.

Self-report is the most widely used type of physical activity measure. Self-reports are defined as self-administered or interviewer-administered recall questionnaires, activity logs or diaries, and proxy reports (typically to assess young children) (68). Self-report has many advantages. It can be conducted on a large number of participants with low cost. Recall questionnaires do not alter the behaviour of the participants being studied. It can also be used to assess all dimensions of physical activity and the patterns of behaviour. Finally, self-report can be used in a wide range of ages and measurement can be adapted to fit the needs of a particular population or research question (68). However, self-report also has some disadvantages. Social desirability can lead to over reporting of physical activity. Recalling physical activity is a highly complex cognitive task. Children and very young adults are likely to have particular memory and recall skill limitations. Some terms used in the questionnaire such as “vigorous physical activity”, “moderate physical activity”, “leisure time” are sometimes ambiguous to participants (68).

Several different questionnaires have been developed to assess levels of physical activity and some of the most commonly used questionnaires include: Active Australia, US CDC Behavioural Risk Factor Surveillance System (BRFSS), and the European physical activity surveillance system (EUPASS) (69, 70).

Physical activity is a global health concern, but the diverse physical activity measures in use often prevent valid international comparisons of studies of this important health related behaviour. The international physical activity questionnaire (IPAQ) was developed as an instrument for cross-national monitoring of physical activity and inactivity (71).

“The development of an international measure for physical activity commenced in Geneva in 1998 and was followed by extensive reliability and validity testing undertaken in 12 countries (14 sites) across 6 continents during 2000 (71). The final results suggest that these measures have acceptable measurement properties for use in many settings and in different

</div>Trang 36<div class="page_container" data-page="36">

languages”. ”Worldwide use of the IPAQ instruments for monitoring and research purposes is encouraged. It is strongly recommended that no changes be made to the order or wording of the questions as this will affect the psychometric properties of the instruments” (70). There are two forms of IPAQ: the long form (5 parts with 27 questions) and the short form (7 questions). In each questionnaire, there are telephone and self-administered versions. The IPAQ is also available in several languages other than English. All IPAQ questionnaires are available at the website www.ipaq.ki.se (70).

Table 2-12 presents physical activity level using IPAQ questionnaire. It demonstrates that the median of MET-hour/week dose vary across countries. For example, there is a higher median MET-hour/week in the Netherlands, which is consistent with the higher level of active local transport in that country.

Table 2-12 Physical activity level from different countries using IPAQ

Country Median of MET-hour/week Reference:

2.4.4 Assessment of diet in adults

There are three major methods used to assess dietary intake in epidemiological research: food records, food recalls and food frequency questionnaires. Each of them has both advantages and disadvantages.

In the food record method, subjects are asked to weigh and write down or tape record all foods they eat in a day for several days (usually 3 to 7 days). The advantages of this method are that it yields precise and detailed information about the amount and kinds of foods consumed. Because the foods have been recorded immediately when preparing or eating, this method does not depend on the memory of the subjects which may be a

</div>Trang 37<div class="page_container" data-page="37">

problem in memory-dependent methods like food frequency questionnaires or food recalls. The disadvantages of the methods are that it requires motivated, trained and literate respondents, and it may be impossible for some age groups such as teenagers who are likely to be too shy to weigh their foods in front of their friends. Furthermore, it has been established that subjects often modify their eating habits to make the task easier and/or represent their diet in a more positive/healthy way. For example, respondents might under-eat or try to avoid mixed dishes in order to avoid complications in weighting all their foods. In addition, the quality of the data may decrease with increasing number of days due to respondent fatigue. Finally, this method may not provide enough information about the dietary intake of those nutrients with large day to day variations in the diet (for example vitamin A).

In the twenty-four hour dietary recall method, subjects are asked by a trained interviewer to remember (to recall) all the foods and drinks they consumed in the past 24 hours (or in the previous day). A common problem with 24h-recalls is the number of days needed. The greater the number of days of dietary intake data collected the greater the precision of the nutrient intakes measured. However, increasing the number of days of measurement of dietary intake is costly and sometimes impossible for those nutrients that require a very large number of days to yield precise estimates of their intake. The decision about the number of days to measure is also related to the purpose of the study. If the study only aims to assess group mean intakes in the population, then one day is appropriate. Otherwise, several days of measured intakes will be needed. Finally, the 24-hour recall method has other disadvantages such as poor memory of food intakes (for example with the elderly), and difficulties in reporting portion sizes (for example with young children).

A food frequency questionnaire (FFQ) is an instrument in which respondents are asked to answer their usual frequency (and sometimes the portion size) of intake of a list of foods items, which represent the usual eating habits of that population. The number of food items on the list depends on the purpose of the study. In general it varies between 75 and 150 food items. FFQs can capture the long-term eating habits of respondents including food items that vary in consumption from day to day or from season to season. They are less costly than the others methods because they can be self-administered in literate populations.

</div>Trang 38<div class="page_container" data-page="38">

This is the method of choice for large-scale epidemiological studies. The disadvantages of FFQs include the number of food-items, for which respondents have to answer, which may leave them distracted and bored by the end of the interview. In addition, it is difficult to develop the instrument to be sure that sufficient number appropriate foods are included. Also the FFQ lacks the detail and specificity of record or recall methods. The FFQ method can be cognitively difficult, requiring good memory and estimation skills from respondents (74).

In order to yield precise information about long-term dietary intake, a food frequency questionnaire should be validated for each ethnic, regional and age group for which it will be used. A well-designed and evaluated FFQ should have good reproducibility and validity. Reproducibility (or reliability) refers to the ability of the dietary instrument to yield similar results when used at two different times on the same person (under the same circumstances). Validity, on the other hand, refers to the degree to which the dietary instrument accurately measures the person’s dietary consumption.

There are several commonly used, and highly developed FFQs such as the NCI Diet history questionnaire, and FFQs developed by Block at the University of California and Willet at Harvard University (74). However, they were developed for use in Western populations rather than populations in the Asia. Thus FFQs modified for the food culture and food habits of specific populations are needed. For example it would not be appropriate to use the Willet FFQ in Vietnam without considerable adjustment and an evaluation of the performance of the modified FFQ.

In the south of Vietnam, a food frequency questionnaire has been validated for adult populations by Kusama et al (75). This is the only validated FFQ at present for Vietnamese adults living in the south of Vietnam. A limitation of this FFQ is that it was validated with a relatively short reference period of 3 months. Also the results of the validation study for this FFQ may be too optimistic because the respondents did not adequately represent the general Vietnamese adult population (for example some of respondents were academic staff of the Nutrition Centre in Ho Chi Minh City and would have been experienced and knowledgeable about answering questions about food). Despite these limitations it remains

</div>Trang 39<div class="page_container" data-page="39">

a valuable tool for assessing dietary intakes in adult population from the south of Vietnam. The validation study of this FFQ reported that the Pearson and intra class correlation coefficients, adjusted for energy intake, ranged from 0.56 to 0.89 (median=0.77) (See Table 2-13). And in terms of validity, the adjusted correlation coefficients between the nutrients measured by the FFQ versus the 24-hour recall ranged from 0.16 for calcium to 0.48 for fibre (median=0.31) (see Table 2-14). The proportion of subjects classified by the FFQ into the same or adjacent quartiles defined by the 24-hour recall was between 69% and 92% (median=79%). As pointed out by the authors, this newly developed FFQ was an adequate tool for the assessment of dietary intakes of energy, protein, fat, carbohydrate, fibre, retinol, carotene, vitamin B1, Vitamin B2, Vitamin C, iron, potassium among Vietnamese adults (75). A careful examination of the results of this validation study revealed that for the majority of nutrients, respondents tended to over-estimate intakes in the FFQ in comparison to the 24-hour recall (Table 2-13). The level of energy intake, for example, increased from 1880 kcal in the 24-hour recall to 2350 kcal in FFQ. Similar patterns of overestimation were observed for most of the macronutrients and micronutrients, but this is often observed in validation studies of FFQs (45).

</div>Trang 40<div class="page_container" data-page="40">

Table 2-13 Reproducibility of the FFQ: Mean daily intake and standard deviation (SD) of energy and nutrients (n=118)

a Nutrient values were transformed (loge) to improve normality.

b Energy intake was adjusted by the residual model. * Food frequency questionnaire 1st time

** Food frequency questionnaire 2nd time Adapted from Kusama et all (75)

</div>