PAPER Do centrally obese Chinese with normal BMI have ... - Nature

International Journal of Obesity (2005) 29, 818–825 & 2005 Nature Publishing Group All rights reserved 0307-0565/05 $30.00 www.nature.com/ijo

PAPER Do centrally obese Chinese with normal BMI have increased risk of metabolic disorders? W-T Yeh1, H-Y Chang2, C-J Yeh1,3, K-S Tsai4, H-J Chen1 and W-H Pan1,5* 1 Institute of Biomedical Sciences, Academia Sinica, Nan-Kong, Taipei, Taiwan; 2Division of Health Policy Research, National Health Research Institute, Taichung, Taiwan; 3Department of Public Health, College of Health Care and Management, Chung Shan Medical University, Taichung, Taiwan; 4Department of Laboratory Medicine, National Taiwan University Hospital, Taipei, Taiwan; and 5Institute of Microbiology and Biochemistry, National Taiwan University, Taipei, Taiwan

BACKGROUND: Body mass index (BMI) and waist circumference are highly correlated. One or the other predicts the metabolic syndromes better, depending on characteristic of the population studied, such as age, gender, and ethnicity. We examined the impact of isolated central obesity, isolated BMI elevation, and the combined type of obesity on metabolic disorders, in order to shed lights on the strategy of obesity screening. METHODS: The study subjects were Chinese aged 20 or above residing in Taiwan. Their data were derived from two large-scale studies: the Nutrition and Health Survey in Taiwan (NAHSIT 1993–1996) and the Cardiovascular Disease Risk Factor Twotownship Study (CVDFACTS, 1994–1997). In evaluating the relations between obesity and health risks, the cut-points of BMI (Z24 kg/m2 for overweight) and waist circumference (Z80 cm for women and Z90 cm for men) recommended by Department of Health in Taiwan for Taiwanese people were used to define various types of obesity. RESULTS: We found that there was a small but nontrivial proportion (1.7% for men and 4.0% for women) of Taiwanese people for whom BMI was in the normal range but their waist circumferences were above normal. These people were at a higher risk of developing metabolic syndromes than those with isolated BMI elevation. Their risks were close to that of the combined type. CONCLUSIONS: In order to screen out high-risk obese individuals, isolated centrally obese subjects should not be overlooked. Therefore, we recommend to assess waist circumference in parallel to, not just sequential to the measurement of BMI in Chinese. International Journal of Obesity (2005) 29, 818–825. doi:10.1038/sj.ijo.0802975 Published online 10 May 2005 Keywords: waist circumference; metabolic syndrome; screening

Introduction Issue on how to define obesity has been widely discussed and evaluated to deal with the worldwide emerging health problems associated with obesity, particularly cardiovascular related diseases and diabetes mellitus.1–4 Body mass index (BMI) and waist circumference are two most popular indicators for assessing obesity, which are relatively inexpensive and easy to use. BMI is strongly associated with body fat content, but has the limitation of overestimating the degree of fatness for very muscular man and underestimating those who have low muscle mass, as in the case of elderly people. In addition, there is gender, age, and ethnicity related variations in body fat for a given BMI probably due to differences in the composition of lean and fat tissue, sitting *Correspondence: Dr W-H Pan, Institute of Biomedical Sciences, Academia Sinica, Taipei 115–29, Taiwan. E-mail: [email protected] Received 30 November 2004; accepted 26 February 2005; published online 10 May 2005

height, hydration status,5 and probably genetic make-ups. On the other hand, waist circumference is highly correlated with the amount of abdominal fat, which is a predictor of metabolic syndrome independent of total body fat.6 In many populations, waist circumference is considered as a better indicator of metabolic risk than BMI.7–12 Different age trends and gender patterns have been observed for BMI and waist circumference in Taiwan as well as in other populations.13,14 The association patterns between obesity and metabolic syndrome/or CVD related disorders differ with gender and with obesity indices.15,16 It is likely that prediction power of BMI and waist circumference vary slightly in an age- and gender-specific manner. However, a simple definition of obesity with good predictive powers is preferred in the public health study to screen people at risk. Many have evaluated whether BMI is better than waist circumference or vice versa and whether one indicator is sufficient for screening, considering adults of all ages and both genders. Since BMI and waist circumference are highly

Isolated central obesity and CVD risks W-T Yeh et al

819 correlated and depending upon the age and gender structure of the studied sample, one or the other will perform better and the magnitudes of their association with related diseases often differ just by a minute amount; we therefore set out to compare the impacts of isolated central obesity and isolated BMI elevation on diseases in various age-gender groups, in order to shed light on the screening strategies for high-risk obese individuals.

for each participant. Two blood pressure measurements were made 30 s apart with the arm at the level of the heart. If two measurements were more than 10 mmHg apart, a third measurement was taken. The two nearest blood pressure values were averaged to obtain the mean blood pressure. In CVDFACTS, the same protocol was followed except that three measurements were obtained in the clinics set up in the neighborhood of the study sites. The last two readings were averaged.

Research methods and procedures Research design The study subjects were Chinese aged 20 or above residing in Taiwan. Data were derived from two large-scale studies. The first set of data was derived from the Nutrition and Health Survey in Taiwan (NAHSIT 1993–1996), a national survey with a complex sampling scheme.17,18 A standardized and structured questionnaire was administered to collect data on detailed demographic characteristics, past medical history, and history of medication along with other nutrition related information. A response rate of 74% was obtained for the household interviews. Around 62% of those interviewed had participated the physical examinations. Following an overnight fast of 8 h or more, subjects reported to a local clinical research station in the morning for physical examinations. The target population included individuals aged 4 and above. Excluding aboriginal people and people with missing data of any of the studied variables, there were 2143 participants (1020 men and 1123 women) aged 20 and above included in this study. The second set of data set is from the Cardiovascular Disease Risk Factor Two-township Study (CVDFACTS, 1994– 1997),19 which was a longitudinal study on risk factors, incidence, and subclinical disorders of the cardiovascular diseases, carried out in Chu-Dong and Pu-Tzu townships of Taiwan. Recruitment had been made from all residents of five selected villages in each township. Three invitation letters had been sent out consecutively to every household. In this study, baseline data on biomedical and blood pressure values, information on medication and anthropometric parameters were derived from 1784 men and 2168 women aged 20 and above.

Data collection NAHSIT and CVDFACTS used the same protocol for measuring anthropometric parameters, and similar ones for measuring biochemical indicators and blood pressure. All participants of both studies have signed informed consent forms.

Blood pressure measurement In NAHSIT, blood pressure was measured by standard sphygmomanometers at home after the participants were rested for at least 5 min. A cuff of appropriate size was used

Blood chemistry In NAHSIT, fasting whole-blood glucose (WBG) was measured with heparinized blood by the glucose oxidase method using a glucose analyzer (Model 23A, YSI, USA) immediately after blood drawing. In collaboration with the clinical chemistry laboratory of the affiliated hospital of National Taiwan University Medical School, a Hitachi 747 analyzer was used to measure serum uric acid, triglyceride, cholesterol, and HDL-C values of the fasting sera in both studies and to measure glucose of the NaF plasma for CVDFACTS within a month following the blood draw.

Anthropometric assessment Anthropometric measurements were carried out after subjects had removed their shoes and heavy clothes. Subjects were asked to put on an examining gown if their apparel was not appropriate for taking measurements. Body weight was measured to the nearest 0.1 kg, and body height to the nearest millimeter. BMI was calculated as weight divided by height squared (kg/m2). Waist circumference was measured horizontally at the level of the natural waist, which was identified as the level at the hollow molding of the trunk when the trunk was concaved laterally.

Definitions of obesity Taiwanese obesity definition was used in this study. Overweight was defined as BMIZ24 kg/m2 and obesity was defined as BMIZ27 kg/m2.20 These cut-points were set lower than that of WHO, since we found that metabolic syndromes were more severe in Chinese than in Caucasians.20 These cut-points are higher than those recommended for Asians by International Obesity Task Force (http://www.iotf.org/), an affiliated organization of WHO. Central obesity was defined as waist circumference Z80 cm for women and Z90 cm for men as IOTF recommended. The expected waist circumference values at BMI ¼ 27 kg/m2 for men and women approximate the above values by regressing waist circumference on BMI.

Definitions for selected conditions Hypertension was defined as using prescribed antihypertensive medication or having a systolic blood pressure International Journal of Obesity


820 Z140 mmHg or above or a diastolic blood pressure Z90 mmHg. Diabetes mellitus was defined as using insulin or hypoglycemic agents or having an abnormal fasting plasma glucose level. The abnormal level was defined if plasma level was Z126 mg/dl in CVDFACTS or WBG level was Z110 mg/dl in NAHSIT.21 Hypercholesterolemia was defined as using lipid-lowering medication or having a serum cholesterol level Z240 mg/dl. Hypertriglyceridemia was defined as using lipid-lowering medication or having a serum triglyceride level Z200 mg/dl. Low HDL-C was defined as serum HDL-C r35 mg/dl. Hyperuricemia was defined as a serum uric acid level Z6.6 mg/dl for women and Z7.7 mg/dl for men.

Statistical analysis In order to examine the age-specific phenomenon, the subjects were grouped into young (20–44), middle-aged (45–64), and old (Z65) age groups. In estimating prevalence rate ratio of various types of obesity, the 75th percentile value of both waist circumference and BMI in each age and gender group was used as cutpoint to define nonobese group and three obesity types to ensure adequate sample size in each subgroups. The nonobese group consisted of those with both waist circumference and BMI below their 75th percentiles. The isolated central obesity group was those with waist circumference above 75th percentile but BMI below 75th percentile. The isolated BMI elevation group was that with BMI above 75th percentile but waist circumference below 75th percentile. The combined type was that with both waist circumference and BMI above their 75th percentile values. Waist circumference, BMI, and age were expressed as mean7standard deviation (s.d.) and age was compared among nonobese group and those obesity types by Duncan analysis. The prevalence of several obesity-related diseases was also compared among these groups.

Table 1

Age-adjusted relative risk (RR) was estimated by Poisson regression to study the effect of obesity types on DM and CVD related conditions in each age and gender group. These RR estimates were compared using CONTRAST statement under Poisson regression in SAS. In evaluating the need to screen isolated central obese individuals in Taiwan, the cut-points of BMI (Z24 kg/m2 for overweight) and waist circumference (Z80 cm for women and Z90 cm for men) recommended by Department of Health in Taiwan for Taiwanese people were used to define various types of obesity. Statistical package SAS, version 8.0 was used for the above analyses.

Results Table 1 showed the quartile values of waist circumference and BMI, and prevalence of obesity by age and gender. The quartile values of waist circumference and the prevalence of central obesity increased with age in both genders, and subsequent to middle age it increased to a greater extent in women than in men. On the other hand, BMI values and the prevalence of general obesity increased with age and declined after middle age and this decrease was more so in men than in women. There were more centrally obese women than men in every age group. There were more obese men than women in young age, but vice versa in middle and old age groups. Table 2 showed the estimates of waist circumference, BMI values, and disease prevalence rates by sex, age, and obesity types. Varying with age and gender groups, nearly 66–69% of the people were grouped into the nonobese group. Around 6–9% people were categorized into the isolated central obesity group, 5.5–8% into the isolated BMI elevation group, and 17–19.5% into the combined type group. The mean age was usually younger in the nonobese group than in the three

Quartile values of waist circumference and BMI and prevalence of obesity by age and gender groups Men

Gender group Age group N Prevalence of central obesitya Waist circumference (cm) 25th 50th 75th Prevalence of general obesityb BMI (kg/m2) 25th 50th 75th

Women

20–44 1006

45–64 1151

65+ 647

20–44 1348

45–64 1391

65+ 552

14.8%

29.5%

33.2%

15.3%

41.9%

60.1%

73.5 80.1 87.0

79.0 85.0 91.0

79.9 85.8 91.0

66.6 70.6 76.0

72.3 78.0 84.0

76.0 82.6 88.6

14.1% 20.9 23.1 25.3

16.8% 22.1 24.1 26.0

13.1% 21.5 23.5 25.6

9.9% 20.3 22.1 24.4

25.0% 22.6 24.5 27.0

22.6% 22.1 24.4 26.8

a Central obesity was defined as waist circumference Z90 cm for men and Z80 cm for women. bGeneral obesity was defined as BMIZ27 kg/m2 for both men and women. (According to the obesity definition recommended by DOH in Taiwan for Taiwanese people).

International Journal of Obesity

Table 2 Mean age, waist circumference, and BMI and disease prevalence by sex, age, and obesity types

Gender Men

Obesity group

N

N%

Agea,b (y)

20–44

Control Iso. central ob Iso. elevated BMI Combined

692 62 59

68.8 6.2 5.9

33.177.1b 37.875.0a 34.576.3b

75.876.0 88.971.8 83.572.5

21.672.2 24.370.8 26.470.9

42.6 74.2 66.1

7.7 16.1 13.6

0.9 1.6 3.4

9 11.3 16.9

6.4 11.3 10.2

12 30.6 15.3

24.3 46.8 42.4

193

19.2

36.776.0a

93.976.0

28.372.3

81.3

25.4

4.7

39.9

13

34.2

47.2


787 76 65

68.4 6.6 5.7

54.975.9b 56.775.4a 54.676.5b

80.676.3 93.372.5 87.173.9

22.772.1 25.070.8 27.070.8

60.7 78.9 72.3

27.8 44.7 49.2

8.4 10.5 15.4

12.2 28.9 15.4

10.9 13.2 16.9

17.3 32.9 20

24.1 35.5 33.8

223

19.4

55.675.7a,b

97.275.8

28.672.2

81.2

41.7

17.5

28.3

14.3

30.9

43.5


427 58 38

66.0 9.0 5.9

71.375.2a 72.076.4a 69.374.2b

80.676.7 92.972.1 88.372.3

22.072.2 24.071.3 26.771.0

69.1 79.3 78.9

43.8 48.3 52.6

9.4 15.5 26.3

6.6 15.5 18.4

13.1 13.8 13.2

14.5 29.3 15.8

26.5 37.9 18.4

124

19.2

70.374.6a,b

98.074.5

27.971.8

88.7

65.3

24.2

22.6

17.7

33.1

46.8


924 87 75

68.6 6.5 5.6

32.776.9b 37.175.6a 37.175.5a

67.774.4 78.472.9 73.272.1

20.971.9 23.071.2 25.370.8

21.2 29.9 26.7

2.1 4.6 9.3

0.4 0 1.3

1.8 3.4 2.7

4.5 5.7 5.3

5.1 17.2 2.7

11.8 10.3 10.7

262

19.4

36.675.6a

83.376.0

27.672.5

56.5

13.7

5

12.2

8.4

16

29.4


926 117 101

66.6 8.4 7.3

54.175.7c 57.974.7a 51.775.7d

73.975.4 87.273.6 80.472.5

23.172.3 25.371.4 28.170.9

54 78.6 61.4

24.1 43.6 23.8

6.3 21.4 8.9

8.6 24.8 13.9

18.5 29.1 19.8

9.7 18.8 7.9

18.8 31.6 29.7

247

17.8

56.175.4b

90.575.0

29.972.2

84.2

47.4

23.5

21.9

20.6

17

42.5

Control Iso. central Ob Iso. elevated BMI Combined

370 44 43

67.0 8.0 7.8

71.075.4b 72.875.9a 70.473.9b

77.976.4 91.772.4 84.872.5

22.972.4 25.171.2 27.871.0

77.8 95.5 86

49.7 63.6 55.8

17.8 43.2 23.3

17.6 29.5 32.6

20.8 20.5 20.9

17 18.2 14

30.3 38.6 46.5

95

17.2

70.475.0b

95.174.7

30.172.6

90.5

63.2

30.5

24.2

27.4

17

45.3

65+

Women

20–44

45–64

65+


Age group

45–64

Waista (cm)

ZOne conditions Hypertension Hypertriglyceridemia Hypercholesterolemia Low HDL Hyperuricemia (%) (%) DM (%) (%) (%) (%) (%) BMIa (kg/m2)

821


Control: Both waist circumference and BMI values are below their 75th percentiles; isolated central obesity: waist circumference value is above 75th percentile but BMI value is below 75th percentile; isolated BMI elevation: BMI value is above 75th percentile but waist circumference value is below 75th percentile; combined type: both waist circumference and BMI values are above their 75th percentiles. a Mean7s.d. bMeans with the same letter superscript are not significantly different, P ¼ 0.05 (Duncan’s multiple-range test).


822 obesity types in each age and gender strata. In general, the disease prevalence rates were low in most nonobese groups and high in most of the combined type groups. For the nearly 6–9% of the population with isolated central obesity, their prevalence of many of the studied conditions were almost as high as that of the combined type. However, the isolated centrally obese group was usually older than the group with isolated BMI elevation. The age effect is adjusted in the following analysis. Age-adjusted RRs of obesity-associated conditions in three types of obesity were shown in Table 3. In terms of the risk of having one or more of the studied conditions, we found that isolated central obesity had higher RR than isolated BMI elevation in most age and gender groups, except in old age men, in whom no significant difference in RR estimates was found between these two types of isolated obesity. For men in all age groups and women in young and middle age groups, the combined type showed the highest RR compared to the other obesity groups. However in old age women, isolated central obesity had the highest RR. For every obesity type, the RRs for having one or more of the condition decreased with age in men. In women, only the RR of the combined type of obesity has a clear decreasing trend with age.

Table 3 group)

In terms of individual disorders, RR was higher for isolated central obesity than for isolated BMI elevation in middle age women for most of the studied conditions. However, this phenomenon holds for about half of the studied conditions in other gender and age group. The RRs of hypertension and diabetes were higher for isolated BMI elevation group than isolated centrally obese group in men and young women, but vice versa in middle and old age women. Isolated central obese men in all age groups had higher RR of hyperuricemia than men with isolated BMI elevation, whereas vice versa in women after 45 y old. However, in comparing RRs between isolated centrally obese groups and isolated BMI elevation groups, there were no consistent patterns of RRs for hypertriglycemia and hypercholesterolemia in all age and sex groups. For the condition of low HDL-C, isolated central obesity showed higher RR than isolated BMI elevation in all age and gender groups, and it was especially high in the young age group. In women, the RR of low HDL-C for isolated central obesity was apparent in its decrease with age. The prevalence of obesity defined by Taiwan definition and associated disorders by BMI and waist circumference categories are shown in Table 4. According to the definition for Taiwanese people, near 50% of both men and women

Age-adjusted relative risk on obesity-associated conditions for three types of obesity by each gender and by age group (all were compared to the nonobese

Men Types of obesity*

Women

Iso. central ob

Iso. elevated BMI

Combined

Iso. central ob

Iso. elevated BMI

Combined

Age ¼ 20–44 ZOne conditions Hypertension DM Hypertriglyceridemia Hypercholesterolemia Low HDL Hyperuricemia

N ¼ 62 1.68b 1.69b (1.21)b 1.13c 1.42b 2.40b 2.10a

N ¼ 59 1.53c 1.67b 3.62a 1.84b 1.51b 1.25c 1.79c

N ¼ 193 1.85a 2.79a 3.80a 4.11a 1.71a 2.72a 2.07b

N ¼ 87 1.32b 1.76c 0 1.84b (0.98)a 3.08a 0.92b

N ¼ 75 1.18c 3.59b 2.23b 1.42c (0.91)a 0.48c (0.95)b

N ¼ 262 2.51a 5.43a 8.65a 6.52a 1.48a 2.90b 2.60a

Age ¼ 45–64 ZOne conditions Hypertension DM Hypertriglyceridemia Hypercholesterolemia Low HDL Hyperuricemia

N ¼ 76 1.30b 1.54b 1.18c 2.49a 1.22c 1.91a 1.51b

N ¼ 65 1.19c 1.78a 1.84b 1.25c 1.55a 1.16c 1.40c

N ¼ 223 1.34a 1.48c 2.04a 2.36b 1.32b 1.79b 1.82a

N ¼ 117 1.34b 1.54b 2.76b 2.75a 1.32a 1.97a 1.55c

N ¼ 101 1.20c 1.10c 1.64c 1.65c 1.20b 0.81c 1.67b

N ¼ 247 1.49a 1.81a 3.36a 2.48b (1.02)c 1.76b 2.17a

Age ¼ 65+ ZOne conditions Hypertension DM Hypertriglyceridemia Hypercholesterolemia Low HDL Hyperuricemia

N ¼ 58 1.15b 1.09c 1.69c 2.42b (1.07)a 2.03a 1.43b

N ¼ 38 1.15b 1.24b 2.61a 2.58b (0.94)b (1.07)b 0.70c

N ¼ 124 1.29a 1.51a 2.49b 3.30a 1.31ab 2.26a 1.77a

N ¼ 44 1.21a 1.26b 2.41a 1.68b (1.05)a 1.09a 1.25c

N ¼ 43 1.11c 1.13c 1.31c 1.85a (0.99)a 0.81c 1.55a

N ¼ 95 1.17b 1.28a 1.71b 1.38c 1.29a (0.99)b 1.51b

*Isolated central obesity: Waist is above 75th percentile but BMI is below 75th percentile. Isolated BMI elevation: BMI is above 75th percentile but waist is below 75th percentile. Combined: Waist is above 75th percentile and BMI is below 75th percentile. a,b,cDifferent symbols represent significantly different (Po0.05) from each other in the descending order, tested using CONTRAST statement under Poisson regression in SAS. (): not significantly different from control group.



823 Table 4

Age group

The prevalence of obesity and associated disorders by BMI and waist circumference categories recommended by Department of Health

Obesity typea

ZOne conditions Hypertension (%) (%)

Hyperuricemia (%)

N%

Normal Iso. central ob Iso. elevated BMI Combined

1510 48 590 656

53.9 1.7 21.0 23.4

52 71 73 82

21 38 33 43

6 13 7 15

8 13 18 28

9 13 13 14

13 25 21 32

23 31 36 43

20–44


619 1 238 148

61.5 0.1 23.7 14.7

42 100 68 80

8 0 16 24

1 0 2 6

8 100 21 35

7 0 9 12

11 100 26 32

24 100 40 47

45–64


548 17 263 323

47.6 1.5 22.9 28.1

55 59 73 80

24 29 39 43

9 12 8 16

9 24 18 27

10 12 16 13

16 24 21 30

22 24 30 41

65+


343 30 89 185

53.0 4.6 13.8 28.6

65 77 87 85

40 43 61 60

8 13 19 21

7 3 7 22

13 13 13 17

14 23 8 35

22 33 42 42


1685 130 485 991

51.2 4.0 14.7 30.1

34 70 54 78

12 36 23 42

4 15 4 19

5 20 9 19

11 20 13 20

7 20 11 17

13 32 24 36

20–44


952 18 190 188

70.6 1.3 14.1 14.0

21 28 39 59

2 6 10 15

0 0 1 6

2 6 5 15

4 6 7 9

5 11 12 16

11 17 18 31

45–64


551 51 257 532

39.6 3.7 18.5 38.3

46 69 61 78

19 33 28 42

6 14 4 18

7 22 10 19

20 24 16 22

7 16 11 17

13 29 26 36

65+

Normal Iso. central Ob Iso. elevated BMI Combined

182 61 38 271

33.0 11.1 6.9 49.1

70 84 82 90

44 48 53 62

15 20 18 29

15 23 26 23

19 21 21 24

14 26 11 17

25 39 34 40

Men 20+

Women 20+

DM (%)

Hypertriglyceridemia Hypercholesterolemiaemia Low HDL (%) (%) (%)

N

a Cut-points for waist circumference: 90 cm for men and 80 cm for women; Cut points for BMI: 24 kg/m2. Normal: Both waist circumference and BMI are below their cut-ponts. Isolated central obesity: Waist circumference is above its cut-point but BMI is below its cut-point. Isolated BMI elevation: BMI is above its cut-point but waist circumference is below its cut-point. Combined type: both waist circumference and BMI are above their cut-points.

were classified in the normal BMI and normal waist circumference group, 21% of men and 15% of women were overweight (BMIZ24 kg/m2) but normal in their waist circumference (isolated BMI elevation). There were small but nontrivial proportions of men (1.7%) and women (4%) classified in normal BMI but centrally obese group (isolated central obesity). Another 23% of men and 28% of women fell within the category of both high BMI and waist circumference (the combined type). The proportions of those classified in the isolated central obesity group increased with age, and were greater in women than in men. In both adult men and women, the prevalence of having one or more of the studied conditions was much higher in

the large waist circumference group than in the normal waist circumference group at both normal (isolated centrally obese group vs normal) and elevated BMI groups (combined group vs isolated BMI elevation). For those isolated centrally obese people, their prevalence of having one or more of the studied conditions was around 70% in both genders, which was almost equivalent to that for men with isolated BMI elevation, but much higher than that for women with isolated BMI elevation. In terms of individual disorders, for women, the pattern was similar to that of having one or more conditions as described above. In men, prevalence of most conditions did not differ much between the isolated central obesity group International Journal of Obesity


824 and the isolated BMI elevation group, except for the prevalence of diabetes, which was particularly high in isolated centrally obese group. Considering age and genderspecific patterns, higher prevalence rates for those with isolated central obesity than those with isolated BMI elevation were seen for about half of the studied conditions in young and middle aged men, and for low HDL-C in old age men. This phenomenon was also observed for hypertriglyceridemia in young age women, for all of the studied conditions in middle age women and or about half of the studied conditions in old age women.

Discussion From this study we found that there was a small but nontrivial proportion of Taiwanese people for whom BMI was in the normal range, but their waist circumferences were above normal (Table 4). These people were at a high risk of developing metabolic syndromes than their counterpart with only BMI elevation. The phenomenon could only partially be explained by their older age. It is well known that for a given BMI, older persons often have more fat than younger persons, and women may have more body fat than men.5 Our age and gender-specific obesity prevalence data not only support this view, but also depict an even more apparent age and gender associated pattern of abdominal fat accumulation. This phenomenon is consistent with our previous study13 and others.22,23 In both men and women, the prevalence of central obesity increased with age to a greater degree than that of general obesity. The prevalence of central obesity differed between women and men also to a greater extent than that of general obesity. Thus, the predictability of anthropometric measurements on metabolic disorders may be a function of gender and agerelated differences in body fat distribution. Findings of many studies7,15,24 are consistent with the above hypothesis in that waist circumference is a better predictor of obesity-related diseases in older than in younger individuals, and in women than in men. Irrespective of the obesity types, the RRs of obesity for having one or more of the metabolic disorders tend to decrease with age, consistent with other studies.25,26 This demonstrates that the relative impact of obesity on disease risk was stronger in young age. Aging and its related factors (such as accumulative effect of environmental exposures) took a more important role on the development of disease in the elderly than in their younger counterpart. The largest RR observed for the middle aged women among all age groups in women suggested that the high prevalence of obesity, especially the sudden rise of the central obesity in them, had a strong effect on the development of metabolic disorders for middle age women in Taiwan. Among all the studied disorders, RRs of having low HDL-C were consistently higher for isolated central obesity groups than isolated BMI elevation group in all age and gender International Journal of Obesity

groups, suggests an intimate relationship between central obesity and lowered HDL-C. A possible mechanism, as suggested by Vajo,27 is that intra-abdominal fat may lower HDL-C levels through the increase of the fractional catabolic rate of Lp A-I (apolipoprotein A-II). The degrees of predictivity of waist circumference and BMI often varied for each studied disease in an age- and genderspecific manner. We also found that in women of young age, hypertension was related more to BMI perhaps because body fluid and peripheral resistance expansion is the primary mechanism,28 while in the middle aged, hypertension may largely emerge from metabolic disorder and thus related more to waist circumference. This age- and gender-specific phenomenon may explain why, in the literature, BMI or waist circumference takes turns to be the more significant predictor of obesity-related metabolic disorders,29 depending on the age and gender structure of the population studied. In order to screen out high-risk obese individuals including isolated centrally obese subjects, we recommend to measure waist circumference in parallel to, not just sequential to the measurement of BMI in Chinese. Although it is more challenging to measure waist circumference reliably than BMI, designing an easy-to-use reliable measuring tape should be able to circumvent the situation.

Acknowledgements The survey was supported by grants (DOH-83-FS-41, DOH84-FS-11, DOH-85-FS-11, DOH-86-FS-11, and DOH-83-TD015, DOH-84-TD-006) from Department of Health, Taiwan, Republic of China, and grant (NHRI-EX93-9225PP) from National Health Research Institute, Taipei, Taiwan.

References 1 Ford ES, Williamson DF, Liu S. Weight change and diabetes incidence: findings from a national cohort of US adults. Am J Epidemiol 1997; 146: 214–222. 2 Hubert HB, Feinleib M, McNamara PM, Castelli WP. Obesity as an independent risk factor for cardiovascular disease: a 26-year follow-up of participants in the Framingham Heart Study. Circulation 1983; 67: 968–977. 3 Lipton RB, Liao Y, Cao G, Cooper RS, McGee D. Determinants of incident non-insulin-dependent diabetes mellitus among blacks and whites in a national sample. The NHANES I Epidemiologic Follow-up Study. Am J Epidemiol 1993; 138: 826–839. 4 Paccaud F, Schluter-Fasmeyer V, Wietlisbach V, Bovet P. Dyslipidemia and abdominal obesity: an assessment in three general populations. J Clin Epidemiol 2000; 53: 393–400. 5 Gallagher D, Visser M, Sepulveda D, Pierson RN, Harris T, Heymsfield SB. How useful is body mass index for comparison of body fatness across age, sex, and ethnic groups? Am J Epidemiol 1996; 143: 228–239. 6 Okosun IS, Liao Y, Rotimi CN, Prewitt TE, Cooper RS. Abdominal adiposity and clustering of multiple metabolic syndrome in White, Black and Hispanic Americans. Ann Epidemiol 2000; 10: 263–270. 7 Fujimoto WY, Newell-Morris LL, Grote M, Bergstrom RW, Shuman WP. Visceral fat obesity and morbidity: NIDDM and


825 8

9

10

11

12

13

14

15

16

17

atherogenic risk in Japanese American men and women. Int J Obes Relat Metab Disord 1991; 15 (Suppl 2): 41–44. Visscher TL, Seidell JC, Molarius A, van der Kuip D, Hofman A, Witteman JC. A comparison of body mass index, waist–hip ratio and waist circumference as predictors of all-cause mortality among the elderly: the Rotterdam study. Int J Obes Relat Metab Disord 2001; 25: 1730–1735. Charlton KE, Schloss I, Visser M, Lambert EV, Kolbe T, Levitt NS, Temple N. Waist circumference predicts clustering of cardiovascular risk factors in older South Africans. Cardiovasc J S Africa 2001; 12: 142–150. Van Pelt RE, Evans EM, Schechtman KB, Ehsani AA, Kohrt WM. Waist circumference vs body mass index for prediction of disease risk in postmenopausal women. Int J Obes Relat Metab Disord 2001; 25: 1183–1188. Dobbelsteyn CJ, Joffres MR, MacLean DR, Flowerdew G. A comparative evaluation of waist circumference, waist-to-hip ratio and body mass index as indicators of cardiovascular risk factors. The Canadian Heart Health Surveys. Int J Obes Relat Metab Disord 2001; 25: 652–661. Turcato E, Bosello O, Francesco VD, Harris TB, Zoico E, Bissoli L, Fracassi E, Zamboni M. Waist circumference and abdominal sagittal diameter as surrogates of body fat distribution in the elderly: their relation with cardiovascular risk factors. Int J Obes Relat Metab Disord 2000; 24: 1005–1010. Teh BH, Pan WH, Chen CJ. The reallocation of body fat toward the abdomen persists to very old age, while body mass index declines after middle age in Chinese. Int J Obes Relat Metab Disord 1996; 20: 683–687. Ito H, Ohshima A, Ohto N, Ogasawara M, Tsuzuki M, Takao K, Hijii C, Tanaka H, Nishioka K. Relation between body composition and age in healthy Japanese subjects. Eur J Clin Nutr 2001; 55: 462–470. Ho SC, Chen YM, Woo JL, Leung SS, Lam TH, Janus ED. Association between simple anthropometric indices and cardiovascular risk factors. Int J Obes Relat Metab Disord 2001; 25: 1689–1697. Hu D, Hannah J, Gray RS, Jablonski KA, Henderson JA, Robbins DC, Lee ET, Welty TK, Howard BV. Effects of obesity and body fat distribution on lipids and lipoproteins in nondiabetic American Indians: The Strong Heart Study. Obes Res 2000; 8: 411–421. Pan WH, Kao MD, Tzeng MS, Yen LL, Hung YT, Li LA, Hsiao SY, Yeh WT, Huang PC. Nutrition and health survey in Taiwan

18

19

20

21

22

23

24

25

26

27

28 29

(NAHSIT) 1993–1996: design, contents, and operations. Nutr Sci J 1999; 24: 1–8. Chang HY, Pan WH, Yeh WT, Tsai KS. Hyperuricemia and gout in Taiwan: results from the Nutritional and Health Survey in Taiwan (1993–96). J Rheumatol 2001; 28: 1640–1646. Yeh CJ, Pan WH, Bai CH, You MS, Wang WC, Wang LY, Lee TK. Curvilinear relations between age and hemostatic parameters in Chinese. Thromb Haemost 1994; 72: 239–243. Pan WH, Flegal KM, Chang HY, Yeh WT, Yeh CJ, Lee WC. Body mass index and obesity-related metabolic disorders in Taiwanese and US whites and blacks: implications for definitions of overweight and obesity for Asians. Am J Clin Nutr 2004; 79: 31–39. Alberti KG, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med 1998; 15: 539–553. Stevens J, Knapp RG, Keil JE, Verdugo RR. Changes in body weight and girths in black and white adults studied over a 25 year interval. Int J Obes 1991; 15: 803–808. Perissinotto E, Pisent C, Sergi G, Grigoletto F. Anthropometric measurements in the elderly: age and gender differences. Br J Nutr 2002; 87: 177–186. Baik I, Ascherio A, Rimm EB, Giovannucci E, Spiegelman D, Stampfer MJ, Willett WC. Adiposity and mortality in men. Am J Epidemiol 2000; 152: 264–271. Rimm EB, Stampfer MJ, Giovannucci E, Ascherio A, Spiegelman D, Colditz GA, Willett WC. Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men. Am J Epidemiol 1995; 141: 1117–1127. Lahmann PH, Lissner L, Gullberg B, Berglund G. A prospective study of adiposity and all-cause mortality: the Malmo Diet and Cancer Study. Obes Res 2002; 10: 361–369. Vajo Z, Terry JG, Brinton EA. Increased intra-abdominal fat may lower HDL levels by increasing the fractional catabolic rate of Lp A-I in postmenopausal women. Atherosclerosis 2002; 160: 495–501. Ferrannini E. Physiological and metabolic consequences of obesity. Metabolism 1995; 9: 15–17. Baumgartner RN, Heymsfield SB, Roche AF. Human body composition and the epidemiology of chronic disease. Obes Res 1995; 3: 73–95.
