Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Body fat distribution and risk of diabetes among Chinese women


OBJECTIVE: To assess the relationship between measures of central and overall obesity and risk of diabetes.

DESIGN: Nested case–control study.

SETTING: Shanghai, China.

PARTICIPANTS: A total of 57 130 women were screened for diabetes at enrollment for the Shanghai Women's Health Study (SWHS), a population-based cohort study of Chinese women aged 40–70 y. In this study, 345 women diagnosed with diabetes and 2760 age-matched controls (eight controls per case), randomly selected from women who tested negative for urine glucose, were included.

RESULTS: Risk of diabetes increased significantly with increasing levels of obesity, particularly with measures of central obesity. Compared to those in the lowest quartile, women in the highest quartile of body mass index (BMI) (≥26.57) and waist to hip ratio (WHR) (≥0.855) had a 2.57-fold (95% CI 1.75–3.77) and a 6.05-fold (95% CI 4.05–9.04) increased risk of diabetes, respectively. The risk of diabetes was elevated with increasing WHR at all levels of BMI, while the positive association between BMI and diabetes was observed primarily among women with a low WHR. However, test for multiplicative interaction was not statistically significant.

CONCLUSIONS: Our data indicated that central obesity is a stronger risk factor for diabetes than overall obesity, suggesting that WHR may be a better indicator of risk of diabetes than BMI among Chinese women.


Estimates from the World Health Organization predict that by the year 2025, 300 million people worldwide will be diagnosed with diabetes.1 The Asian/Pacific region, accounting for 46 percent of the global burden of diabetes, includes the largest population of people diagnosed with diabetes in the world.2 Studies among the Chinese have demonstrated an increase in the prevalence of diabetes during the past 10 years, particularly in urban areas such as Shanghai.3,4,5,6,7,8,9 The increased prevalence of type II diabetes in these and other Asian populations can be attributed, at least partially, to increases in obesity.2,5,6,7,10

Body mass index (BMI) and waist to hip ratio (WHR) are the two anthropometric measurements most frequently used to assess obesity and central obesity. Although epidemiological studies have demonstrated that both BMI and WHR are powerful predictors of type II diabetes, the relative contribution of each to an individual's risk remains unclear.11,12,13,14,15,16,17,18,19 Further complicating this issue, quantitative definitions used to indicate obesity differ among studies and among gender and ethnic groups. The World Health Organization currently defines overweight as a BMI of 25–29.9 kg/m2 and obese as a BMI ≥30 kg/m2.20 However, compared to Caucasians, Chinese people appear to have a higher body fat percentage given the same BMI.5 Specifically, 32 percent body fat, which is considered obese by the WHO, corresponded to a BMI of only 21.2 kg/m2 in Chinese women, which is considered nearly optimal by US standards.21 In addition, there is growing evidence, particularly among Asian and female populations, that central obesity may be a more consistent predictor of glucose intolerance and type II diabetes than overall obesity.11,12,13,14,17,18,22,23,24,25 Thus, it is imperative to investigate the utility of BMI and WHR in predicting risk of diabetes in Chinese and other Asian women.

Furthermore, given the strong correlation between central and overall obesity, and the high prevalence of obesity among the US population, it is difficult to examine the effect of central fat distribution on risk of diabetes among normal-weight women in the US26 Traditionally, Chinese women have a low rate of obesity, providing a unique opportunity to examine the effects of central adiposity on risk of diabetes among nonobese women.5 We examined the association between anthropometric measurements and risk of diabetes among a subset of participants who were screened for diabetes at enrollment for the Shanghai Women's Health Study (SWHS).


The SWHS is a population-based prospective cohort study conducted in seven urban communities in Shanghai, China. All eligible women (n=81 170), who were aged 40–70 y and resided in these communities between March 1997 and May 2000, were approached for the study and 75 221 women were enrolled, yielding a participation rate of 92.7%. After further exclusion of 278 women who were later found to be younger than 40 or older than 70 y at the time of interview, 74 443 women remained for the SWHS. The major reasons for nonresponse were refusal to participate (3.0%), absent during enrollment period (2.6%), and other miscellaneous reasons (ie, health, hearing, and speaking problems; 1.6%). For 56 832 (75.7%) women who donated a spot urine sample at enrollment, a semiquantitative urinalysis dipstick assay was performed to screen for diabetes. Of the 1254 women who tested positive for urine glucose (>trace), 566 had no prior history of diabetes. Among the later, 345 women were subsequently diagnosed with diabetes using a fasting blood glucose test, an oral glucose tolerance test, or both. The WHO guidelines for fasting blood glucose testing (≥7 mg/dl) and oral glucose tolerance testing (≥11.1 mg/dl) were used to confirm cases of diabetes. A total of 228 of these women were tested at the study's designated testing facilities, while the remainder of subjects had their tests performed at the their primary care hospital. The 345 subjects with confirmed diabetes comprised the case group for the study. Controls were randomly selected from study participants, who tested negative for urine glucose, had no prior history of diabetes, and were individually matched to the index cases by age (within 1 y) at a ratio of eight controls per case.

Information on usual dietary intake during the past year, personal habit, physical development, occupation history, and medical history were elicited by trained interviewers using a structured questionnaire during in-person interviews. In order to enhance the quality of interview data, interviews were tape recorded and selectively monitored by quality control staff.

Study participants were measured for weight, standing and sitting height, and waist and hip circumferences by trained interviewers according to a standard protocol at the baseline survey. Waist circumferences were measured at 2.5 cm above the navel and hip circumferences at the level of maximum width of the buttocks. All measurements were taken twice with a tolerance limit of 1 kg for weight and 1 cm for heights and circumferences. A third measurement was taken if the difference of the two measurements was greater than the tolerance limit. The average of the two closest measurements were used in the current analysis. BMI was calculated as the subject's weight in kilograms divided by the square of height in meters. WHR was calculated by dividing the subject's waist circumference in centimeters by hip circumference in centimeters. Study variables were grouped into quartiles based on distributions of the controls.

Odds ratios (OR) were used to measure the association of diabetes with BMI and WHR. Conditional logistic regression models were used to obtain maximum likelihood estimates of the odds ratios and their 95% confidence intervals (CI), after adjusting for potential confounders.27 Tests for trend were performed by entering the categorical variables as continuous parameters in the models. Tests for interaction were performed by introducing a multiplicative interaction term into the logistic model. All analyses were performed using SAS 8.10 and all tests of statistical significance were based on two-sided probability.


Demographics and suggested risk factors for diabetes are presented in Table 1. The average age was 56.4 y for cases and 55.9 y for controls. Compared to controls, cases tended to have less education and lower income, but higher caloric intake, BMI, WHR, and parity, and were more likely to have a history of hypertension. We found no significant differences between diabetes cases and women with glucosuria. All of these variables were adjusted for in the multiple regression analysis to control for potential confounding effects. There were no significant differences between cases and controls with regard to age, alcohol intake, smoking, regular exercise, oral contraceptive use, history of chronic pancreatitis, and age at diagnosis of hypertension.

Table 1 Comparison of cases and controls, by sample population, regarding demographics and selected diabetes risk factorsa

Risk of diabetes increased significantly with all obesity measures, especially with measures of central obesity (Table 2). Compared to women weighing less than 54.0 kg, women weighing at least 65.5 kg had nearly twice the risk of diabetes (OR 1.79, 95% CI 1.25–2.59), after adjustment for nonanthropometric variables. Similarly, compared to women with a BMI of less than 22.06, women with a BMI of at least 26.57 had roughly 2.5 times the risk of diabetes (OR 2.57, 95% CI 1.75–3.77). An increased risk of diabetes was observed among overweight (OR 1.71, 95% CI 1.33–2.19) and obese (OR 2.19, 95% CI 1.47–3.27) women, as defined by standard World Health Organization (WHO) cut-points for BMI. Comparison of the highest quartile with the lowest quartile for waist circumference, hip circumference, and WHR yielded odds ratios of 5.15 (95% CI 3.40–7.79), 1.67 (95% CI 1.17–2.39), and 6.05 (95% CI 4.05–9.04), respectively. Although recent studies have implied that waist circumference alone can be an adequate predictor of central adiposity, WHR was used for further analysis in order to provide some adjustment for the frame size of Chinese women and to make our results comparable to earlier studies, which frequently used WHR as a measure of central adiposity.28 Increasing height was inversely associated with risk of diabetes (P=0.003). However, after additional adjustment for weight (data not shown), there was no association between height and diabetes (P=0.93). All analyses were conducted among women with glucosuria (n=566) and the subset of women tested at the designated study facilities (n=288) with no significant differences in results (data not shown).

Table 2 Associations between anthropometric measurements and risk of diabetes. Shanghai Women's Health Study, 1997–2000a

Further analyses were conducted to evaluate the joint and independent effects of BMI and WHR on the risk of diabetes (Table 3). WHR was positively associated with risk of diabetes at all levels of BMI. However, the positive relationship between BMI and the risk of diabetes weakened with increasing WHR. This pattern was observed using quartiles of BMI from our population as well as the standard WHO cut-points. After adjustment for BMI, risk of diabetes increased with increasing WHR regardless of BMI categorization (quartiles P<0.0001, WHO P<0.0001). However, following adjustment for WHR, BMI no longer predicted risk of diabetes (quartiles P=0.15, WHO P=0.05). Tests for multiplicative interaction between BMI and WHR were not statistically significant (quartiles P=0.09, WHO P=0.05). The analysis presented in Table 3 was also performed using the International Obesity Task Force's newly proposed BMI cut-point for adult Asians, with similar results (data not shown).29

Table 3 Odds ratios and 95% confidence intervals for joint effects of WHR and BMI on the risk of diabetes. Shanghai Women's Health Study, 1997–2000a


Our study, one of the largest population-based case–control studies of diabetes, indicated that central obesity, as measured by WHR, is an important predictor of diabetes risk among Chinese women. A clear dose–response relationship was observed between central obesity and risk of diabetes, particularly among nonobese and underweight women. Furthermore, we found that BMI conferred an increased risk of diabetes primarily among women with a relatively low WHR.

Our findings are consistent with those from previous epidemiological studies, including two earlier prospective studies of women in the US.12,17 In the Nurses’ Health Study, risk of type II diabetes was more strongly related to waist circumference than with BMI.12 In the Iowa Women's Study, the dose–response relationship between risk of diabetes and WHR was much stronger than that with BMI.17 Findings in support of waist measurements as an important indicator of risk for diabetes have also been reported in a number of prospective studies of men11,13,14 These results may be explained, in part, by the fact that BMI does not accurately reflect percent of total body fat among individuals with significant visceral adiposity.26 These cohorts, and individuals at risk for diabetes, are often older individuals. Consequently, age-related variation in lean body mass reduces the validity of BMI as a measure of adiposity.26 In contrast, estimation of total fatness via abdominal circumference improves with age, implying that WHR may be a better predictor of risk of diabetes than BMI in aging populations.26

Analysis of the joint effects of BMI and WHR on risk of diabetes demonstrated the predictive strength of WHR across all levels of BMI. Although one would predict that individuals in the highest quartiles of both BMI and WHR should have a compounded risk of diabetes, the data suggest that there may be a limit to the risk conferred by increased adiposity. Persons with low overall adiposity increase their risk of diabetes most drastically by gaining fat centrally. Even persons considered obese can significantly increase their risk of diabetes by gaining fat in this region. However, increasing overall fat content conferred little, if any, additional risk among individuals in the higher quartiles of WHR. Biologically, these data may represent the fact the central and overall adiposity have both independent and common endocrine-based mechanisms of diabetes pathogenesis. Both contribute significantly to overall risk, yet central adiposity appears to have a more profound effect.

It is generally accepted that obesity, particularly central obesity, can have deleterious metabolic effects, thereby increasing the risk of developing various chronic diseases.21 Expanded fat stores, a hallmark of obesity, results in increased lipolysis, causes a rise in circulating free fatty acids, and promotes peripheral and hepatic insulin resistance. In order to compensate for increased glucose production and insulin resistance, stimulation of insulin secretion occurs. While normal obese individuals can adjust to elevated plasma free fatty acids in such a manner, a subpopulation of obese individuals lack the ability to hypersecrete insulin.30 Finally, recent data has suggested that progression to diabetes may also be associated with two hormones produced in adipocytes, leptin and resistin31,32

Central adiposity is believed to differ in physiology from fat stores elsewhere in the body. It was found that lipolysis occurs more frequently in visceral fat and is less sensitive to the antilipolytic effects of insulin than other fat stores.33 In addition, it appears that the proximity of abdominal fat to the liver may deliver excess amounts of free fatty acids directly to the portal vein, thus increasing the hepatic burden of free fatty acids.34 If true, it would follow that elevated levels of leptin and resistin may also be delivered into portal circulation in a similar fashion. These new biological links between obesity, glucose impairment, and diabetes further support the distribution and amount of fat content as risk factors for diabetes.

This study has several limitations. First, cases were identified from women who initially tested positive using nonfasting urine glucose tests. Although the specificity of such tests is near 100%, the sensitivity may range from 37–75%.35 Therefore, it is possible that controls used in this study may actually have subclinical diabetes, thus biasing the risk estimate towards the null. Furthermore, there is the potential for selection bias since not all subjects with glucosuria were tested for diabetes. However, we found no differences between confirmed diabetes cases and all women with glucosuria regarding demographic and suggested risk factors for diabetes. Additionally, further analyses were conducted among all women with glucosuria, with no significant differences in results between them and confirmed diabetes cases. Second, because anthropometric measurements such as BMI and WHR are not single past exposures, but cumulative ones, the temporal sequence of body fat distribution and disease progression cannot be firmly established in this study. However, we excluded all subjects with a previous history of diabetes from this study, thus minimizing the potential effect of diabetes on the level of body fat and its distribution. Furthermore, since diabetes is a wasting disease, any potential bias resulting from prevalent diabetes would be conservative and tend to underestimate the risks. In addition, when analyses were performed using self-reported data for weight from age 50 y (7 y earlier than the mean age of diagnosis), the results were comparable to those reported in this paper (data not shown). As with any epidemiological investigation, unobserved confounders cannot be excluded and may skew our results.

Several strengths of this study should be noted as well. First, the population-based case–control design and high participation rate (92.7%) reduces potential selection biases. Second, subjects were interviewed before incidence cases of diabetes were diagnosed, thus eliminating the possibility of recall bias based on disease status. Third, the use of standardized measurements ensures that the anthropometric variables used in this study are comparable among study participants. Finally, our study population consists of a large number of subjects with normal and below average BMI, thus increasing the statistical power for studying the effect of central adiposity among nonobese subjects.

In summary, our study suggests that among Chinese women and aging populations, measures of central obesity are better predictors of risk of diabetes than measures of overall obesity. Importantly, this association holds true for nonobese and underweight women, as well. Our findings suggest that in a population with a low prevalence of obesity, WHR is an important and significant predictor of risk of diabetes.


  1. 1

    King H, Aubert RE, Herman WH . Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections. Diabetes Care; 1998; 21: 1414–1431.

    CAS  Article  Google Scholar 

  2. 2

    Coughlan A, McCarty DJ, Jorgensen LN, Zimmet P . The epidemic of NIDDM in Asian and Pacific Island populations: prevalence and risk factors. Horm Metab Res; 1997; 29: 323–331.

    CAS  Article  Google Scholar 

  3. 3

    Gu D, Reynolds K, Duan X, Xin X, Chen J, Wu X, Mo J, Whelton PK, He J, InterASIA Collaborative Group. Prevelance of diabetes and impaired fasting glucose in the Chinese adult population: International Collaborative Study of Cardiovascular Disease in Asia (Inter ASIA). Diabetologia; 2003; 46: 1190–1198.

    CAS  Article  Google Scholar 

  4. 4

    Cockram CS, Woo J, Lau E, Chan JC, Chan AY, Lau J, Swaminathan R, Donnan SP . The prevalence of diabetes mellitus and impaired glucose tolerance among Hong Kong Chinese adults of working age. Diabetes Res Clin Pract; 1993; 21: 67–73.

    CAS  Article  Google Scholar 

  5. 5

    Cockram CS . The epidemiology of diabetes mellitus in the Asia-Pacific region. Hong Kong Med J; 2000; 6: 43–52.

    CAS  PubMed  Google Scholar 

  6. 6

    Hu YH, Li GW, Pan XR . Incidence of NIDDM in Daqing and forecasting of NIDDM in China in 21st century. Zhonghua Nei Ke Za Zhi; 1993; 32: 173–175.

    CAS  PubMed  Google Scholar 

  7. 7

    Pan XR, Yang WY, Li GW, Liu J . Prevalence of diabetes and its risk factors in China, 1994. National Diabetes Prevention and Control Cooperative Group. Diabetes Care; 1997; 20: 1664–1669.

    CAS  Article  Google Scholar 

  8. 8

    Shi HL, Fang JC, Zhu XX . Prevalence of diabetes mellitus and associated risk factors in an adult urban population in Shanghai. Diabetes Metab; 1998; 24: 539–542.

    CAS  PubMed  Google Scholar 

  9. 9

    Zhong XL . Diabetes mellitus survey in China. Chin Med J (Engl); 1982; 95: 423–430.

    CAS  Google Scholar 

  10. 10

    Pan CY, Lu JM, Tian H, Kong XT, Lu XP, Yao C, Jiang CE, Deng XX, Wang SY, Zhang XL, Wang ZS, Cui L . Study of the prevalence of diabetes mellitus in adults in the Shougang Corporation in Beijing. Diabet Med; 1996; 13: 663–668.

    CAS  Article  Google Scholar 

  11. 11

    Boyko EJ, Fujimoto WY, Leonetti DL, Newell-Morris L . Visceral adiposity and risk of type 2 diabetes: a prospective study among Japanese Americans. Diabetes Care; 2000; 23: 465–471.

    CAS  Article  Google Scholar 

  12. 12

    Carey VJ, Walters EE, Colditz GA, Solomon CG, Willett WC, Rosner BA, Speizer FE, Manson JE . Body fat distribution and risk of non-insulin-dependent diabetes mellitus in women. The Nurses’ Health Study. Am J Epidemiol; 1997; 145: 614–619.

    CAS  Article  Google Scholar 

  13. 13

    Cassano PA, Rosner B, Vokonas PS, Weiss ST . Obesity and body fat distribution in relation to the incidence of non-insulin-dependent diabetes mellitus. A prospective cohort study of men in the normative aging study. Am J Epidemiol; 1992; 136: 1474–1486.

    CAS  Article  Google Scholar 

  14. 14

    Chan JM, Rimm EB, Colditz GA, Stampfer MJ, Willett WC . Obesity, fat distribution, and weight gain as risk factors for clinical diabetes in men. Diabetes Care; 1994; 17: 961–969.

    CAS  Article  Google Scholar 

  15. 15

    Colditz GA, Willett WC, Rotnitzky A, Manson JE . Weight gain as a risk factor for clinical diabetes mellitus in women. Ann Intern Med; 1995; 122: 481–486.

    CAS  Article  Google Scholar 

  16. 16

    Haffner SM, Stern MP, Mitchell BD, Hazuda HP, Patterson JK . Incidence of type II diabetes in Mexican Americans predicted by fasting insulin and glucose levels, obesity, and body-fat distribution. Diabetes; 1990; 39: 283–288.

    CAS  Article  Google Scholar 

  17. 17

    Kaye SA, Folsom AR, Sprafka JM, Princas RJ, Wallace RB . Increased incidence of diabetes mellitus in relation to abdominal adiposity in older women. J Clin Epidemiol; 1991; 44: 329–334.

    CAS  Article  Google Scholar 

  18. 18

    Lundgren H, Bengtsson C, Blohme G, Lapidus L, Sjöström L . Adiposity and adipose tissue distribution in relation to incidence of diabetes in women: results from a prospective population study in Gothenburg, Sweden. Int J Obes Relat Metab Disord; 1989; 13: 413–423.

    CAS  Google Scholar 

  19. 19

    Warne DK, Charles MA, Hanson RL, Jacobsson LT, McCance DR, Knowler WC, Pettitt DJ . Comparison of body size measurements as predictors of NIDDM in Pima Indians. Diabetes Care; 1995; 18: 435–439.

    CAS  Article  Google Scholar 

  20. 20

    WHO Expert Committee. Report of a WHO Expert Committee. World Health Organ Tech Rep Ser; 1995; 854: 1–452.

  21. 21

    NULBI Obesity Education Initiative Expert Panel. National Institutes of Health. Obes Res; 1998; 6 (Suppl 2): 51S–209S.

  22. 22

    Anderson PJ, Chan JC, Chan YL, Tomlinson B, Young RP, Lee ZS, Lee KK, Metreweli C, Cockram CS, Critchley JA . Visceral fat and cardiovascular risk factors in Chinese NIDDM patients. Diabetes Care; 1997; 20: 1854–1858.

    CAS  Article  Google Scholar 

  23. 23

    Hodge AM, Dowse GK, Zimmet PZ, Collins VR . Prevalence and secular trends in obesity in Pacific and Indian Ocean island populations. Obes Res; 1995; 3 (Suppl 2): 77s–87s.

    Article  Google Scholar 

  24. 24

    Shelgikar KM, Hockaday TD, Yajnik CS . Central rather than generalized obesity is related to hyperglycaemia in Asian Indian subjects. Diabetes Med; 1991; 8: 712–717.

    CAS  Article  Google Scholar 

  25. 25

    Unwin N, Harland J, White M, Bhopal R, Winocour P, Stephenson P, Watson W, Turner C, Alberti KG . Body mass index, waist circumference, waist-hip ratio, and glucose intolerance in Chinese and Europid adults in Newcastle, UK. J Epidemiol Community Health; 1997; 51: 160–166.

    CAS  Article  Google Scholar 

  26. 26

    Willett W . Nutritional epidemiology, 2nd edn. Oxford University Press: New York; 1998.

    Book  Google Scholar 

  27. 27

    Hosmer DW, Lemeshow S . Applied logistic regression, 2nd edn. Wiley: New York; 2000.

    Book  Google Scholar 

  28. 28

    Ruderman N, Chisholm D, Pi-Sunyer X, Schneider S . The metabolically obese, normal-weight individual revisited. Diabetes; 1998; 47: 699–713.

    CAS  Article  Google Scholar 

  29. 29

    WHO/IASO/IOTF. The Asia-Pacific perspective: redefining obesity and its treatment. 2000. Sydney, Australia, Health Communications Australia.

  30. 30

    Kopelman PG, Albon L . Obesity, non-insulin-dependent diabetes mellitus and the metabolic syndrome. Br Med Bull; 1997; 53: 322–340.

    CAS  Article  Google Scholar 

  31. 31

    Nonogaki K . New insights into sympathetic regulation of glucose and fat metabolism. Diabetologia; 2000; 43: 533–549.

    CAS  Article  Google Scholar 

  32. 32

    Steppan CM, Bailey ST, Bhat S, Brown EJ, Banerjee RR, Wright CM, Patel HR, Ahima RS, Lazar MA . The hormone resistin links obesity to diabetes. Nature; 2001; 409: 307–312.

    CAS  Article  Google Scholar 

  33. 33

    Bolinder J, Kager L, Ostman J, Arner P . Differences at the receptor and postreceptor levels between human omental and subcutaneous adipose tissue in the action of insulin on lipolysis. Diabetes; 1983; 32: 117–123.

    CAS  Article  Google Scholar 

  34. 34

    Harris MI, Hadden WC, Knowler WC, Bennett PH . International criteria for the diagnosis of diabetes and impaired glucose tolerance. Diabetes Care; 1985; 8: 562–567.

    CAS  Article  Google Scholar 

  35. 35

    Hanson RL, Nelson RG, McCance DR, Beart JA, Charles MA, Pettitt DJ, Knowler WC . Comparison of screening tests for non-insulin-dependent diabetes mellitus. Arch Intern Med; 1993; 153: 2133–2140.

    CAS  Article  Google Scholar 

Download references


This study was supported by USPHS Grant R01CA70867, RFP No. N02-CP11010-66, RFP No. N02-CP-81040-50. We thank the doctors and health workers in the study communities for their important contributions, and Dr Wanqing Wen for statistical consultation in the data analysis. This study would not have been possible without the support of all of the study participants and research staff of the Shanghai Women's Health Study.

Author information



Corresponding author

Correspondence to X-O Shu.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Rosenthal, A., Jin, F., Shu, XO. et al. Body fat distribution and risk of diabetes among Chinese women. Int J Obes 28, 594–599 (2004).

Download citation


  • diabetes
  • BMI

Further reading


Quick links