Comparison of different anthropometric measures as predictors of diabetes incidence in a Chinese population

diabetes research and clinical practice 92 (2011) 265–271

Contents lists available at ScienceDirect

Diabetes Research and Clinical Practice journ al h omepage: www .elsevier.co m/lo cate/diabres

Comparison of different anthropometric measures as predictors of diabetes incidence in a Chinese population Zhaoxia Jia a, Yong Zhou b, Xiurong Liu c, Yilong Wang b, Xingquan Zhao b, Yongjun Wang b, Wannian Liang a,d,*, Shouling Wu c,** a

Department of Epidemiology and Health Statistics, School of Public Health and Family Medicine, Capital Medical University, Beijing 100069, China b Beijing Tian Tan Hospital, Capital Medical University, Beijing 100050, China c Kailuan Hospital Affiliated to North China Coal Medical College, Tangshan 063000, China d Center for Public Health Emergency, The Ministry Health of P.R.C., Beijing 100040, China

article info

abstract

Article history:

Objective: We aimed to explore an optimal anthropometric indicator and optimal cut-off

Received 16 September 2010

points for incident diabetes in Chinese adults.

Received in revised form

Methods: 61,703 subjects were followed for a median duration of 2 years. Body mass index,

15 January 2011

waist circumference, waist-to-hip ratio and waist-to-height ratio were collected base on a

Accepted 24 January 2011

standard protocol. Receiver Operating Characteristic curve analyses were used to compare

Published on line 21 February 2011

the predictive power of baseline BMI, WC, WHpR and WHtR for development of type 2 diabetes.

Keywords:

Results: There were 2991 new cases of type 2 diabetes during follow-up. ROC curve analyses

Abdominal adiposity

indicated that WHtR was the best predictor of type 2 diabetes for male (AUC = 0.633). For

Type 2 diabetes

female, WHtR and WC had similar predictive ability (AUC = 0.701 and 0.695 respectively) and

Waist circumference

were superior to BMI. WHpR was the weakest predictor in both genders. The optimal WHtR

Waist to hip ratio

cut-off values for incidence of type 2 diabetes were similar in both genders (0.53 vs. 0.52). BMI

Body mass index

was higher in men (26 kg/m2) than women (24 kg/m2); and so did WC (91 cm in men vs. 85 cm

Waist to height ratio

in women). Conclusions: WHtR, and to some degree WC, are the best predictors of type 2 diabetes, followed by BMI then WHpR which is the weakest predictor in the tested adults. # 2011 Elsevier Ireland Ltd. All rights reserved.

1.

Introduction

The prevalence of diabetes is high and is increasing over the world, especially in China. A recent survey in China suggested that 92.4 million adults 20 years of age or older (9.7% of the adult population) have diabetes [1]. Obesity has been proved to be an important independent risk factor for type 2 diabetes [2].

Clinical trials have shown that lifestyle intervention, including weight reduction, can benefit individuals at increased risk for type 2 diabetes [3,4], so WHO recommends to develop simple strategies to identify those at risk of diabetes and provide them with early lifestyle interventions [5]. As easy to operate and noninvasive, obesity indicators such as body mass index (BMI), have been proposed and applied in diabetes prevention.

* Corresponding author at: Department of Epidemiology and Health Statistics, School of Public Health and Family Medicine, Capital Medical University, No. 10 Xitoutiao, You An Men, Beijing 100069, China. Tel.: +86 010 83911508; fax: +86 010 83911508. ** Corresponding author at: Kailuan Hospital Affiliated to North China Coal Medical College, No. 57 Xin Hua Dong Dao, Tangshan 063000, China. Tel.: +86 0315 3025655. E-mail address: [email protected] (W. Liang). 0168-8227/$ – see front matter # 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2011.01.021

266

diabetes research and clinical practice 92 (2011) 265–271

In the recent years, it has become clear that mainly visceral, rather than subcutaneous fat, is associated with noninfectious chronic disease [6–10]. This finding suggested that measures of central fat distribution such as waist circumference (WC), waist to hip ratio (WHpR), and waist to height ratio (WHtR) may be better than BMI in predicting type 2 diabetes. However, it is not yet established which specific measures of obesity might be most strongly associated with risk of type 2 diabetes [11–13]. From clinical and public prevention point of view, it is practically important to clarify the optimal or more reliable anthropometric indicator for diabetes incidence. To our knowledge, the number of prospective studies about this issue is limited in mainland China. We have therefore compared the predictive ability of BMI with WC, WHpR and WHtR for predicting diabetes incidence based on a prospective study in Chinese adults. On the other hand, the optimal cut-off values of anthropometric measures for definition of obesity were controversial. It often underestimates obesity in Asia using the same criteria as for whites [14,15]. We have also attempted to identify the optimal cut-off values to assess the risk of diabetes.

2.

Methods

Height was measured to the nearest 0.1 cm using a portable stadiometer and weight was measured to the nearest 0.1 kg using calibrated platform scales. WC was measured to the nearest 0.1 cm at the midpoint between the subcostal margin and the margin of the supracrestal plane. Hip circumference was measured to the nearest 0.1 cm at the point of maximum circumference over the buttocks. Body mass index (BMI) was calculated using the formula: weight/height2 (kg/m2). Waist to hip ratio (WHpR) was calculated as WC divided by hip circumference, and waist to height ratio (WHtR) as WC divided by height. Blood pressure (BP) was measured in the right arm with subjects in a sitting position using a regular mercury sphygmomanometer after resting for 15 min. The three consecutive blood pressure readings were used for mean value of the BP. Laboratory test: Blood samples were obtained from the antecubital vein and transfused into vacuum tubes containing EDTA in the morning after an overnight fasting period. Tubes were centrifuged at 3000  g for 10 min at 25 8C. An auto analyzer (Hitachi 747; Hitachi, Tokyo, Japan) was used to measure fasting plasma glucose, total cholesterol (TC), triglycerides (TG), HDL and LDL at the central laboratory of Kailuan hospital.

2.1.

Study population

2.3.

The data were obtained from health examinations of employees of the Kailuan Company in Tangshan city in the central north of China, which had been reported before [16]. From June 2006 to September 2007, a total of 101,510 employees and retirees underwent the baseline survey. After the baseline survey, all subjects were followed up every year by face-toface or telephone interview and were invited to a biennial health examination. Between 2007 and 2009, a total of 3802 subjects were lost for follow-up and 955 subjects died. In the present analyses, we selected subjects aged 18–85 years old and free of cardiovascular diseases or diabetes at baseline. We excluded subjects who died during follow-up, who were pregnant during baseline survey, or who had missing one of anthropometric measurements data. We also excluded subjects who had no plasma glucose tests at baseline. This left 48,015 men and 13,688 women for inclusion in this study. We followed standard protocols in all measurements. The protocol for this study was in accordance with the guidelines of the Helsinki Declaration, and was approved by the Ethics Committee of the Hospital. All participants have given their written informed consent.

2.2.

Baseline data collection

Questionnaire: All data collections were performed by specially trained doctors and nurses. A standardized questionnaire was used for collecting information on subjects’ demographic characteristics; family and personnel medical history; and lifestyle, including smoking status, alcohol consumption, physical activity, sleeping time and quality. Anthropometric measurements: Measurements included height, weight, WC, and hip circumference. All subjects were measured standing in light clothing without shoes and hats.

Follow-up data collection

After the baseline survey, all subjects were followed up for development of MI, stroke and type 2 diabetes. Information on the incidence of new events was obtained from the combination of regular 12 months face-to-face or telephone interview and reviews of patients’ hospital record and repeated health examinations. Diabetes was defined as the presence of any of the following at follow-up assessment: (1) fasting plasma glucose level  7.0 mmol/L on two occasions [17]; (2) current use of insulin or oral hypoglycemic agents; or (3) a positive response to the question, ‘‘Has a doctor ever told you that you have diabetes?’’

2.4.

Statistical analyses

Continuous variables were expressed as the mean  SD or median (interquartile range) as appropriate. Differences of baseline characteristics between participants with and without diabetes were tested by Student’s t test for normally distributed variables and by Mann–Whitney’s rank test for variables with a skewed distribution. Categorical data were expressed as frequencies. Differences between participants with and without diabetes were tested by Chi-square test. This study population was stratified into sex-specific quartiles of BMI, WC, WHpR and WHtR. Logistic regression analyses were performed to study the association of baseline anthropometric measures (BMI, WC, WHpR and WHtR) with the incidence of type 2 diabetes. The odds ratios (ORs) were computed for quartiles 2, 3, and 4 as compared with the lowest quartile in different Logistic regression models. Covariates including age, systolic blood pressure (SBP), lg triglyceride, lg HDL-cholesterol, and lg fasting glucose were fitted as continuous variables in the multivariate analyses and smoking, alcohol intake, regular physical exercise, family history of diabetes were fitted as

267

diabetes research and clinical practice 92 (2011) 265–271

3.2. Association of general and central obesity with diabetes analyzed by odds ratios

categorical variables. The linear trend in ORs was evaluated using the likelihood ratio test. Receiver Operating Characteristic (ROC) curve analyses and the respective area under the curves (AUC) were used to compare the predictive power of baseline BMI, WC, WHpR and WHtR on risk of type 2 diabetes in both genders. The significance of the differences between AUCs was assessed using the algorithm developed by DeLong [18]. The point on the ROC which represented the largest sum of sensitivity and specificity was chosen to obtain the optimal cut-off point for these four measurements in predicting type 2 diabetes. All analyses for men and women were performed separately. The Statistical Program for Social Sciences, Version 15.0 (SPSS Inc., Chicago, IL) was used for statistical analyses. All statistical tests were two-sided and p < 0.05 was considered to indicate statistical significance.

3.

Results

3.1.

Basic characteristics of subjects

The relationships between BMI, WC, WHpR and WHtR and risk of type 2 diabetes in men and women are displayed in Tables 2 and 3. All measurements were significantly associated with risk of type 2 diabetes in both men and women before and after adjustment for potential risk factors including age, alcohol intake, smoking, regular physical exercise and family history of diabetes. After further adjustment for SBP, plasma glucose, TG and HDL, the associations were reduced, but remained to be significant. However, it can be seen that WHpR showed the weakest association in both men and women (adjusted OR = 1.64 for men and 2.76 for women), and WHtR showed the strongest association (adjusted OR = 1.98 for men and 3.30 for women). In men BMI and WC showed similar associations (adjusted OR = 1.78 and 1.76 respectively), while in women WC showed higher association than BMI (adjusted OR = 3.24 and 3.08 respectively).

3.3. Optimal predictor and cut-off points for predicting diabetes by ROC curve analyses

For a median duration of 2 years follow-up, 2991 subjects developed type 2 diabetes among them 2514 were men and 477 were women. Baseline characteristics by gender are shown in Table 1. In men, those individuals with incident diabetes were older than those without diabetes ( p < 0.001). The same association was seen in women. Compared with participants who did not have diabetes at follow-up period, those who had incident diabetes had a higher BMI, WC, WHpR, WHtR in both genders (all p < 0.001). They also had a higher baseline fasting plasma glucose, triglyceride and systolic blood pressure (all p < 0.001). Participants with incident diabetes were also more likely than those without diabetes to have impaired fasting glucose at baseline in men and women (all p < 0.001).

In men, ROC analyses indicated that the AUC was 0.620 for BMI, 0.619 for WC, 0.597 for WHpR and 0.633 for WHtR. WHtR was the best predictor of type 2 diabetes ( p < 0.001 for differences in AUC for WHtR vs. BMI or WC) based on this analysis. BMI and WC had similar ability in predicting type 2 diabetes ( p = 0.34 for differences in AUC for WC vs. BMI). And WHpR was the weakest predictor ( p < 0.001 for differences in AUC for WHpR vs. BMI). While in women, the AUC was 0.672 for BMI, 0.695 for WC, 0.628 for WHpR and 0.701 for WHtR. WC and WHtR had similar predictive ability ( p = 0.16 for differences in AUC for WC vs. WHtR) and were significantly superior to BMI ( p < 0.001 for differences in AUC for WC or WHtR vs. BMI). And WHpR was still the weakest predictor in women

Table 1 – Baseline characteristics of study subjects stratified by gender. Males

n Age (years) Waist circumference (WC, cm) Body mass index (BMI, kg/m2) Waist to hip ratio (WHpR) Waist to height ratio (WHtR) Systolic blood pressure (SBP, mmHg) Total cholesterol (TC, mmol/L) Triglyceride (TG, mmol/L) HDL-cholesterol (mmol/L) Fasting glucose (mmol/L) FPG  5.6 mmol/L (n, %) Current smokers (n, %) Regular alcohol drinking (n, %) Regular physical exercise (n, %) Family history of diabetes (n, %) Compared with those not incident diabetes. p < 0.001. b p < 0.05. a

Females

With diabetes

Without diabetes

With diabetes

Without diabetes

2514 52.3  10.9a 90.8  9.7a 26.3  3.5a 0.91  0.06a 0.54  0.06a 138.1  20.4a 5.03(4.36–5.72)a 1.54(1.08–2.39)a 1.49(1.27–1.76) 5.88(5.21–6.38)a 1544(61.4%)a 980(39.0%) 639(25.4%)b 719(28.6%)a 399(15.9%)

45,501 49.0  12.5 87.4  9.5 25.1  3.7 0.89  0.08 0.52  0.06 130.4  19.9 4.89(4.25–5.53) 1.26(0.89–1.91) 1.49(1.27–1.75) 5.03(4.61–5.50) 9791(21.5%) 18,370(40.4%) 8675(19.1%) 8019(17.6%) 6598(14.5%)

477 53.8  9.4a 89.4  10.4a 26.9  4.0a 0.89  0.07a 0.56  0.07a 136.5  22.6a 5.09(4.52–5.86)a 1.61(1.18–2.45)a 1.56(1.33–1.82) 5.88(5.20–6.39)a 289(60.6%)a 15(3.3%)a 3(0.6%) 138(28.9%)a 101(21.2%)

13,211 46.4  11.5 82.1  12.7 24.4  3.8 0.86  0.12 0.51  0.08 122.1  20.0 4.88(4.26–5.53) 1.11(0.77–1.64) 1.56(1.34–1.81) 4.90(4.50–5.30) 1791(13.6%) 137(1.1%) 57(0.4%) 1888(14.3%) 2532(19.2%)

268

diabetes research and clinical practice 92 (2011) 265–271

Table 2 – Unadjusted and multivariate-adjusted odds ratios and 95% CI for type 2 diabetes according to quartiles of anthropometric measurements in men. Variable

Cases Unadjusted

Quartiles BMI (kg/m2) <22.7 375 1.00 22.7523 1.36(1.19–1.57) 24.9657 1.77(1.55–2.02) 27.3959 2.64(2.32–2.99) p for trends <0.001 WC (cm) <81 376 1.00 81469 1.27(1.11–1.47) 87605 1.53(1.34–1.75) 931064 2.48(2.19–2.81) p for trends <0.001 WHpR <0.86 392 1.00 0.86604 1.41(1.24–1.61) 0.90731 1.86(1.63–2.11) 0.94787 2.14(1.88–2.43) p for trends <0.001 WHtR <0.47 268 1.00 0.47538 1.45(1.25–1.68) 0.51684 1.79(1.55–2.07) 0.551024 2.92(2.53–3.36) p for trends <0.001 a Adjusted for exercise, family b Adjusted for exercise, family

Adjusteda

Adjustedb

1.00 1.35(1.17–1.57) 1.78(1.56–2.03) 2.72(2.39–3.08) <0.001

1.00 1.13(0.98–1.31) 1.32(1.15–1.52) 1.78(1.56–2.04) <0.001

1.00 1.24(1.07–1.42) 1.46(1.28–1.67) 2.26(2.00–2.55) <0.001

1.00 1.14(0.98–1.32) 1.25(1.09–1.43) 1.76(1.55–2.01) <0.001

1.00 1.37(1.20–1.57) 1.74(1.53–1.98) 1.96(1.73–2.23) <0.001

1.00 1.22(1.06–1.40) 1.44(1.26–1.65) 1.64(1.43–1.87) <0.001

1.00 1.38(1.18–1.61) 1.65(1.42–1.91) 2.55(2.21–2.94) <0.001

1.00 1.29(1.10–1.50) 1.39(1.19–1.62) 1.98(1.70–2.29) <0.001

age, smoking, alcohol intake, regular physical history of diabetes. age, smoking, alcohol intake, regular physical history of diabetes, SBP, lgHDL, lgTG and lgBS.

(Fig. 1). We subsequently assessed the improvement of prediction of these four anthropometric measures when adding to the model including age and family history of [()TD$FIG]diabetes. The AUC for a model of age and family history of

Table 3 – Unadjusted and multivariate-adjusted odds ratios and 95% CI for type 2 diabetes according to quartiles of anthropometric measurements in women. Variable

Cases Unadjusted

Quartiles BMI (kg/m2) <21.8 21.824.026.6p for trends WC (cm) <75 758289p for trends WHpR <0.80 0.800.850.90p for trends WHtR <0.47 0.470.510.55p for trends a

Adjusted for exercise, family b Adjusted for exercise, family

Adjusteda

Adjustedb

35 86 129 227

1.00 2.66(1.75–4.03) 3.92(2.62–5.83) 7.51(5.13–10.98) <0.001

1.00 1.00 2.26(1.49–3.45) 1.98(1.29–3.05) 2.97(1.99–4.45) 2.00(1.32–3.06) 5.45(3.72–8.01) 3.08(2.06–4.59) <0.001 <0.001

28 82 129 238

1.00 2.71(1.74–4.20) 4.47(2.93–6.81) 8.38(5.60–12.53) <0.001

1.00 1.00 2.08(1.33–3.24) 1.79(1.13–2.82) 2.97(1.93–4.56) 2.04(1.31–3.18) 5.06(3.35–7.66) 3.24(2.11–4.97) <0.001 <0.001

30 102 149 196

1.00 3.18(2.09–4.83) 5.14(3.43–7.68) 5.58(3.75–8.29) <0.001

1.00 1.00 2.37(1.56–3.61) 2.11(1.37–3.26) 3.45(2.29–5.19) 2.68(1.76–4.09) 3.79(2.54–5.67) 2.76(1.82–4.19) <0.001 <0.001

30 79 112 256

1.00 3.17(2.04–4.91) 4.88(3.21–7.43) 8.70(5.86–12.91) <0.001

1.00 1.00 2.48(1.60–3.86) 2.07(1.31–3.24) 3.33(2.17–5.11) 2.38(1.53–3.71) 5.20(3.45–7.83) 3.30(2.16–5.04) <0.001 <0.001

age, smoking, alcohol intake, regular physical history of diabetes. age, smoking, alcohol intake, regular physical history of diabetes, SBP, lgHDL, lgTG and lgBS.

diabetes was 0.587 for men and 0.676 for women. The AUC increased significantly to 0.693 with the addition of WHtR, 0.645 with BMI, 0.642 with WC and 0.601 with WHpR in men. In women, the AUC increased significantly from 0.676 to 0.751

Fig. 1 – ROC curves of WC, BMI, WHpR and WHtR for predicting type 2 diabetes in men and women.

269

diabetes research and clinical practice 92 (2011) 265–271

Table 4 – Areas under the ROC curve, cut-off points, sensitivity and specificity of anthropometric measurements to predict type 2 diabetes.

Men BMI (kg/m2) WC (cm) WHpR WHtR Women BMI (kg/m2) WC (cm) WHpR WHtR

AUC (95% CI)

p value

Cut-off points

Sen

Spe

Youden’s index

0.620(0.610–0.632) 0.619(0.608–0.630) 0.597(0.586–0.608) 0.633(0.623–0.644)

– 0.34 <0.001 <0.001

26 91 0.89 0.52

0.578 0.509 0.635 0.654

0.632 0.691 0.529 0.582

0.201 0.200 0.164 0.236

0.672(0.650–0.696) 0.695(0.673–0.718) 0.628(0.607–0.650) 0.701(0.677–0.722)

– <0.001 <0.001 <0.001

24 85 0.85 0.53

0.710 0.623 0.752 0.631

0.564 0.664 0.475 0.655

0.274 0.287 0.227 0.286

with the addition of WHtR, 0.735 with BMI, 0.733 with WC and 0.702 with WHpR. In addition, we assessed the optimal cut-off points for BMI, WC and WHtR for identifying diabetes in men and women. Table 4 indicated the sensitivity, specificity and Youden’s index for the optimal cut-off points. The optimal cut-off points of BMI and WC in men (26 kg/m2, 91 cm) were all higher than that in women (24 kg/m2, 85 cm). And in terms of WHtR, the optimal cut-off point in men was similar with that in women (0.52 vs. 0.53).

4.

Discussion

In this study, we found that among men and women, both general obesity (BMI) and central obesity (WC, WHpR, WHtR) had strong associations with the risk of diabetes after controlling for a variety of potential confounders, including age, plasma glucose, plasma lipids and blood pressure. By analyses of ROC curves we have shown that among men and women, WHtR appeared to be the best predictor of diabetes risk compared with other parameters, while WHpR was the weakest predictor. Among men, WC and BMI showed similar predicting effects for diabetes risk. Among women, WC was the stronger predictor of risk than BMI. Previous studies comparing BMI, WC, WHpR and WHtR for predicting the incidence of diabetes have been inconsistent [11,19–25]. The variations are independent of gender, ethnicity and methods used in analyses. For example, researches from Jamaican [19] and Pima Indians [11] showed that WHtR was a simple but better predictor compared with other anthropometric measures. In the EPIC-Potsdam Study [23] of adults aged 35–65 years, WHtR appeared to be similar to WC but was somewhat better compared with BMI or WHpR in predicting the incidence of diabetes. These conclusions are in close agreement with our outcomes. That WHtR is the best predictor for non-communicable diseases also confirmed in a Japanese mass epidemiological study [26] and Nurses’ Health Study [27]. In the MONICA/KORA Augsburg Cohort Study [22], WC emerged as the stronger predictor of type 2 diabetes in women but in men BMI seemed to be the stronger predictor. However, evidence from Iraq [24,25] and Hoorn [28] studies showed that WHpR was a better predictor of future type 2 diabetes than WC and WHpR. In these studies, however, WHtR was not taken into account. This study was carried out in an employee population, in which ‘‘healthy worker effect’’ might exist.

Therefore, we can not extend our findings to general population. Further study would be required. Taken together, the results suggest that the best predictors of future diabetes may depend on sex and ethnic, and that central obesity is better associated with diabetes than general obesity. In men, the cut-off points to predict incident diabetes for BMI, WC, WHpR, and WHtR were 26 kg/m2, 91 cm, 0.89 and 0.52 respectively. These results were almost equal to outcomes of Iraq men [25]. In women, the cut-off points for BMI, WC, WHpR, and WHtR were 24 kg/m2, 85 cm, 0.85 and 0.53 respectively while in Iraq women those were 26.1 kg/m2, 91 cm, 0.91 and 0.56. The optimal cut-off points of BMI in our study is lower compared with the current definitions of obesity recommended by WHO (BMI  30) [29]. This result suggests that the threshold of BMI could be decreased in Chinese. The cut-off point for WC recommended for clinical practice remains controversial. The Asian-specific WC cut-off points for central obesity which were adopted in IDF definition [30] and ATPIII definition [31] were 90 cm for men and 80 cm for women. However, Chinese definition [32] for metabolic syndrome (MS) set the cut-off points to 90 cm for men and 85 cm for women based on the fact that 90 cm for men and 85 cm for women were the corresponding waist circumference for a BMI value of 25 kg/m2. It is interesting to note that the optimal cut-off points of WC in our study are in line with the Chinese definition of MS. In this study, most of the unadjusted AUCs were between 0.6 and 0.7, which means the overall performance of the test was low. Age might play a major role in this issue. Study from Iraq [25], in which the mean age of participants was 45.6 y, has shown that the AUCs of these four indicators ranged from 0.59 to 0.74. In an older population of British [21], in which the mean age of subjects was about 68 y, the AUCs were between 0.65 and 0.78. It would be better to test the reliability of cut-off points from our study in other population before using the findings of our study. After adjusted for age and family history of diabetes, the AUCs increased to 0.64–0.69 in men and 0.73–0.75 in women. In practice, it can combine one of these four measures with age and family history of diabetes to identify people in high risk of diabetes. One of the strengths of our study is the prospective identification of new cases of diabetes in a relatively large sample of Chinese adults. Another strengths is that a standardized protocol was used in all centers, and all anthropometric variables were collected using direct measurement by trained doctors and nurses rather than self-

270

diabetes research and clinical practice 92 (2011) 265–271

report. In contrast to these strengths, there are some limitations of our study. The first is that this study had relatively short duration of follow-up. This may be underestimated the association between anthropometric measures with diabetes risk. Although, the similar short-term follow-up was seen in other studies [20,33]. The second is that the sampled number of male is larger than women, but we did all analyses classified by gender. We think this limitation would not bring much bias in our study. The third is that we excluded participants with cardiovascular diseases from the baseline which would possibly bias the results. Previous study had shown that obese people were associated with an increased risk of cardiovascular diseases [34]. Thus, people with cardiovascular diseases may have a high prevalence of obesity and these people are at high risk for incident diabetes. This limitation may cause to underestimate the association between anthropometric measures with diabetes risk.

[10]

[11]

[12]

[13]

[14]

Acknowledgements [15]

We thank all participants for their continuous dedication and commitment in this study. And we thank all staff in Kailuan Hospital Affiliated to North China Coal Medical College for their support.

[16]

[17]

Conflict of interest [18]

There are no conflicts of interest.

references

[19]

[20] [1] Yang WY, Lu JM, Weng JP, Jia WP, Ji LN, Xiao JZ, et al. Prevalence of diabetes among men and women in China. N Engl J Med 2010;362:1090–101. [2] DeFronzo RA. From the triumvirate to the ominous octet: a new paradigm for the treatment of type 2 diabetes. Diabetes 2009;58:773–95. [3] Pan XR, Li GW, Hu YH, Wang JX, Yang WY, An ZX, et al. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care 1997;20:537–44. [4] Tuomilehto J, Lindstrom J, Eriksson JG, Valle TT, Helena H, Ilanne-Parikka P, et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001;344:1343–50. [5] World Health Organization. 2008–2013 action plan for the global strategy for the prevention and control of noncommunicable disease. Geneva: WHO; 2008. [6] Bjorntorp P. Metabolic implications of body fat distribution. Diabetes Care 1991;14:1132–43. [7] Vague J. The degree of masculine differentiation of obesities: a factor determining predisposition to diabetes, atherosclerosis, gout, and uric calculus disease. Am J Clin Nutr 1956;4:20–34. [8] Folsom AR, Kushi LH, Anderson KE, Mink PJ, Olson JE, Hong CP, et al. Associations of general and abdominal obesity with multiple health outcomes in older women: the Iowa Women’s Health Study. Arch Intern Med 2000;160:2117–28. [9] Yusuf S, Hawken S, Ounpuu S, Bautista L, Franzosi MG, Commerford P, et al. Obesity and the risk of myocardial

[21]

[22]

[23]

[24] [25]

[26]

infarction in 27,000 participants from 52 countries: a case– control study. Lancet 2005;366:1640–9. Hu D, Xie J, Fu P, Zhou J, Yu D, Whelton PK, et al. Central rather than overall obesity is related to diabetes in the Chinese population: the InterASIA study. Obesity (Silver Spring) 2007;15:2809–16. Tulloch-Reid MK, Williams DE, Looker HC, Hanson RL, Knowler WC. Do measures of body fat distribution provide information on the risk of type 2 diabetes in addition to measures of general obesity? Comparison of anthropometric predictors of type 2 diabetes in Pima Indians. Diabetes Care 2003;26:2556–61. Esmaillzadeh A, Mirmiran P, Azizi F. Waist-to-hip ratio is a better screening measure for cardiovascular risk factors than other anthropometric indicators in Tehranian adult men. Int J Obes Relat Metab Disord 2004;28:1325–32. Rosenthal AD, Jin F, Shu XO, Yang G, Elasy TA, Chow WH, et al. Body fat distribution and risk of diabetes among Chinese women. Int J Obes Relat Metab Disord 2004;28:594–9. Deurenberg PM, Deurenberg-Yap SG. Asians are different from Caucasians and from each other in their body mass index/body fat percent relationship. Obes Rev 2002;3:141–6. Prentice AM. The emerging epidemic of obesity in developing countries. Int J Epidemiol 2006;35:93–9. Wang F, Wu SL, Song Y, Tang X, Marshall R, Liang M, et al. Waist circumference, body mass index and waist to hip ratio for prediction of the metabolic syndrome in Chinese. Nutr Metab Cardiovasc Dis 2009;19:542–7. American Diabetic Association. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes 2003;S5–20. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837–45. Sargeant LA, Bennett FI, Forrester TE, Cooper RS, Wilks RJ. Predicting incident diabetes in Jamaica: the role of anthropometry. Obes Res 2002;10:792–8. Hadaegh FA, Zabetian HH, Azizi F. Waist/height ratio as a better predictor of type 2 diabetes compared to body mass index in Tehranian adult men – a 3.6-year prospective study. Exp Clin Endocrinol Diabetes 2006;114:310–5. Wannamethee SG, Papacosta O, Whincup PH, Carson C, Thomas MC, Lawlor DA, et al. Assessing prediction of diabetes in older adults using different adiposity measures: a 7 year prospective study in 6,923 older men and women. Diabetologia 2010;53(5):890–8. Meisinger C, Do¨ring A, Thorand B, Heier M, Lo¨wel H. Body fat distribution and risk of type 2 diabetes in the general population: are there differences between men and women? The MONICA/KORA Augsburg Cohort Study. Am J Clin Nutr 2006;84:483–9. Schulze MB, Heidemann C, Schienkiewitz A, Bergmann MM, Hoffmann K, Boeing H. Comparison of anthropometric characteristics in predicting the incidence of type 2 diabetes in the EPIC-Potsdam Study. Diabetes Care 2006;29:1921–3. Mansour AA, Al-Jazairi MI. Predictors of incident diabetes mellitus in Basrah, Iraq. Ann Nutr Metab 2007;51:277–80. Mansour AA, Al-Jazairi MI. Cut-off values for anthropometric variables that confer increased risk of type 2 diabetes mellitus and hypertension in Iraq. Arch Med Res 2007;38:253–8. Hsieh SD, Muto T. The superiority of waist-to-height ratio as an anthropometric index to evaluate clustering of coronary risk factors among non-obese men and women. Prev Med 2005;40:216–20.

diabetes research and clinical practice 92 (2011) 265–271

[27] John HP, Kathryn MR, Frank Hu, Christine MA, Claudia UC, JoAnn EM. Waist-height ratio as a predictor of coronary heart disease among women. Epidemiology 2009;20:361–6. [28] Marieke BS, Jacqueline MD, Marjolein V, Lex MB, Coen DAS, Piet JK, et al. Associations of hip and thigh circumferences independent of waist circumference with the incidence of type 2 diabetes: the Hoorn Study. Am J Clin Nutr 2003;77(5):1192–7. [29] World Health Organization. Definition, diagnosis and classification of diabetes mellitus and its complications: report of a WHO consultation. Geneva, Switzerland: Department of Noncommunicable Disease Surveillance; 1999. p. 31–3. [30] Alberti KG, Zimmet P, Shaw J. The metabolic syndrome – a new worldwide definition. Lancet 2005;366:1059–62.

271

[31] Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, et al. Diagnosis and management of the metabolic syndrome: an American Heart Association/ National Heart Lung, and Blood Institute Scientific Statement. Circulation 2005;112:2735–52. [32] Expert Panel on Metabolic Syndrome of Chinese Diabetes Society. Recommendations on metabolic syndrome of Chinese diabetes society. Chin J Diabetes 2004;12:156–61 (Chinese). [33] Ko GT, Chan JC, Chow CC, Yeung VT, Chan WB, So WY, et al. Effects of obesity on the conversion from normal glucose tolerance to diabetes in Hong Kong Chinese. Obes Res 2004;12:889–95. [34] Song YM, Sung J, Davey SG, Ebrahim S. Body mass index and ischemic and hemorrhagic stroke: a prospective study in Korean men. Stroke 2004;35:831–6.