Prevalence and significance of generalized and central body obesity in an urban Asian Indian population in Chennai, India (CURES: 47)

Transcription

1 (2009) 63, & 2009 Macmillan Publishers Limited All rights reserved /09 $ ORIGINAL ARTICLE Prevalence and significance of generalized and central body obesity in an urban Asian Indian population in Chennai, India (CURES: 47) M Deepa 1, S Farooq 1, R Deepa 1, D Manjula 2 and V Mohan 1 1 Madras Diabetes Research Foundation, Dr Mohan s Diabetes Specialities Centre, Chennai, India and 2 Department of Epidemiology, The Tamil Nadu Dr MGR Medical University, Chennai, India Objectives: To determine the prevalence of generalized and abdominal obesity in urban Asian Indians and compare the association of body mass index (BMI) and waist circumference (WC) with metabolic risk variables. Methods: Subjects were recruited from the Chennai Urban Rural Epidemiology Study (CURES) carried out between 2001 and 2004 and involved 2350/2600 eligible subjects (response rate 90.4%). Anthropometric measurements, lipids and oral glucose tolerance tests were carried out. Generalized obesity (BMIX23 kg m 2 ) and abdominal obesity (WCX90 cm in men and X80 cm in women) were defined using WHO Asia Pacific guidelines. Results: The age standardized prevalence of generalized obesity was 45.9% (95% CI: %), (women: 47.4%; men: 43.2%, P ¼ 0.210), while that of abdominal obesity was 46.6% (95% CI: %), (women: 56.2%4 men: 35.1%, Po0.001). Area under the curve for identifying subjects with any three metabolic risk factors using BMI was 0.66, 95% CI: , Po0.001, while, for WC, it was 0.70, 95% CI: , Po0.001 for men, and 0.69, 95% CI: , Po0.001 for women. Isolated generalized obesity (normal WC, increased BMI) was present in 12.7% of men and 6.1% of women. Isolated abdominal obesity (increased WC, normal BMI) was present in 4.7% of men and 14% of women. Combined obesity was present in 32.6% of men and 43.3% of women. Conclusions: In Asian Indians, the prevalence of combined obesity is high among both sexes, while isolated generalized obesity is more common in men and isolated abdominal obesity more common in women. However, these prevalence rates vary markedly depending on cut points used. WC is a better marker of obesity-related metabolic risk than BMI in women compared to men in this population. (2009) 63, ; doi: /sj.ejcn ; published online 10 October 2007 Keywords: generalized obesity; central body obesity; BMI; waist circumference; cardiovascular risk factors; Asian Indians Introduction Most available studies on the prevalence of obesity are from developed countries, and there is paucity of data from developing countries. Epidemiological surveys use body Correspondence: Dr V Mohan, Madras Diabetes Research Foundation, Dr Mohan s Diabetes Specialities Centre, 4, Conran Smith Road, Gopalapuram, Chennai, TN , India. mvdsc@vsnl.com Contributors: VM designed the study. MD and SF conducted the study and wrote the first draft of the manuscript and VM rewrote the subsequent drafts. RD assisted in doing statistical analysis. VM, DM and RD contributed to the interpretation of the data and all contributors participated in the revisions and final draft of the manuscript. Received 8 November 2006; revised 12 August 2007; accepted 28 August 2007; published online 10 October 2007 mass index (BMI) as an indicator of generalized obesity and waist circumference (WC) or waist-to-hip ratio (WHR) as a measure of central or abdominal obesity. Both generalized and abdominal obesity have been associated with a number of metabolic abnormalities. However, there appear to be ethnic differences in the representative significance (International Diabetes Federation, 2005). This study reports on the prevalence and significance of generalized and abdominal obesity in urban south Indians. Study design Chennai Urban Rural Epidemiology Study (CURES) is a large cross-sectional study done in Chennai in southern India

2 260 with a population of 5 million people. The study design is described elsewhere (Deepa et al., 2003), and the sampling frame is available at mdrf/cures.pdf and summarized in Figure 1. Phase 1 Phase 1 of CURES was conducted from June 2001 to August 2002, involving a representative sample of individuals X20 years of age. Blood pressure and basic anthropometric measures were available in all and fasting capillary blood glucose measurements in 99.3%. Phase 2 Phase 2 of CURES deals with studies of the prevalence of complications of diabetes. Phase 3 In Phase 3 of CURES, to maintain the representativeness of the sample as well as considering the logistics involved, every tenth subject recruited in phase 1 (n ¼ 2600) was invited to our center between September 2001 and September 2004 for detailed anthropometric measurements and biochemical tests. Repeated contacts (atleast five) were made by door-to-door visits to ensure better response. Eventually, 2350 participated in the present study (response rate ¼ 90.4%). There were no significant differences in the clinical characteristics of the responders (n ¼ 2,350) and the non-responders (n ¼ 150). All study subjects underwent an oral glucose tolerance test using 75-gm glucose load, except self-reported diabetic subjects, for whom fasting venous plasma glucose was measured. The fasting sample was taken, after ensuring 8 h of overnight fast, for estimation of plasma glucose and serum lipids using a Hitachi 912 Autoanalyser (Roche Diagnostics GmbH, Mannheim, Germany), utilizing kits supplied by Roche Diagnostics GmbH. Glycated hemoglobin was measured using the Variant machine (Bio-Rad, Hercules, CA, USA). Anthropometric measurements including weight, height, waist and hip measurements were obtained using standardized techniques by two well-trained interviewers (Deepa et al., 2003). Inter- and intra-observer variability coefficients of variation were o5%. Height Height was measured with a tape to the nearest cm. Subjects were requested to stand upright without shoes with their back against the wall, heels together and eyes directed forward. Weight Weight was measured with a spring balance that was kept on a firm horizontal surface. Subjects wore light clothing, stood upright without shoes and weight was recorded to the nearest 0.5 kg. The scale was calibrated every day with standard weights. Chennai city population (corporation limits) (approx 5 million) Chennai corporation: 10 zones & 155 wards 46 wards were selected Systematic random sampling method 26,001 individuals [age 20 years] screened PHASE 1 PHASE 2 PHASE 3* Door-to-door survey in the community (n = 26,001) [June 2001 August 2002] [Studies of prevalence of diabetes] All diabetic subjects from Phase 1 [Studies of micro and macro vascular complications of diabetes] Every tenth subject from Phase 1 (n = 2,350) [September ] [Studies of prevalence of cardio-vascular risk factors] Figure 1 The sampling protocol of Chennai Urban Rural Epidemiology Study (CURES). *This study deals with phase 3.

3 Body mass index Body mass index was calculated using the formula: weight (kg)/height (m 2 ). Waist circumference Waist circumference was measured using a non-stretchable measuring tape. Subjects were asked to stand erect in a relaxed position with both feet together on a flat surface; one layer of clothing was accepted. WC was measured as the smallest horizontal girth between the costal margins and the iliac crests at minimal respiration. Mean of two readings was taken as WC. Blood pressure Blood pressure was recorded in the sitting position in the right arm to the nearest 2 mm Hg using mercury sphygmomanometer (Diamond Deluxe, Pune, Maharashtra, India). Two readings were taken 5 min apart and mean calculated. If the difference between the first and the second reading was 46 mm Hg for systolic and/or 44 mm Hg for diastolic pressure, then a third reading was taken. Institutional Ethical Committee approval, and informed consent (from all study subjects) were obtained. Definitions Diabetes Diagnosis of diabetes was based on WHO Consulting group criteria, that is, 2-h post load (75 g glucose) plasma glucose (2-h PG) X200 mg dl 1 (X11.1 mmol l 1 ). Impaired glucose tolerance (IGT) was diagnosed if the 2-h PG was X140 mg dl 1 (X7.8 mmol l 1 ) and o200 mg dl 1 (o11.1 mmol l 1 ), and normal glucose tolerance if 2-h PG was o140 mg dl 1 (o7.8 mmol l 1 ) (Alberti and Zimmet, 1998). Hypertension Hypertension was diagnosed based on drug treatment for hypertension or BP4130/85 mm Hg (Grundy, 2005). Hypercholesterolemia Hypercholesterolemia was diagnosed if serum cholesterol levels were X200 mg dl 1 (X5.2 mmol l 1 ) or were under drug treatment for hypercholesterolemia (National Cholesterol Education Programme (NCEP), 2001). Hypertriglyceridemia Hypertriglyceridemia was diagnosed if serum triglyceride levels were X150 mg dl 1 (X1.7 mmol l 1 ) or were under drug treatment for hypertriglyceridemia (Grundy, 2005). Low high-density lipoprotein cholesterol Low high-density lipoprotein cholesterol was diagnosed if levels were o40 mg dl 1 (o1.04 mmol l 1 ) for men and o50 mg dl 1 (o1.3 mmol l 1 ) for women (Grundy, 2005). Obesity Generalized obesity was defined using the WHO Expert Consultation guidelines (WHO Expert Consultation, 2004) as BMIX23 kg m 2 and abdominal obesity as WCX90 cm for men and X80 cm for women (WHO, 2000). Metabolic risk factors Metabolic risk factors were defined based on revised ATP III definition for Asian Indians (Grundy, 2005) as shown in Table 5. Statistics Statistical analyses were performed using SPSS for windows version 10.0 software (SPSS Inc., Chicago, IL, USA). Student s t-tests were used for continuous variable and w 2 test for proportions, and all comparisons were only between two groups. The prevalence rate obtained in the present study was age-standardized to the 2001 Census of India. Correlation between the obesity indices and the metabolic risk variables was done using bivariate correlation analysis. Sensitivity and specificity of BMI and WC cut points were estimated for determining the presence or absence of various metabolic risk factors. The area under the receiver operating characteristic curves and the 95% confidence intervals (CIs) were computed. Association of obesity indices with metabolic risk factors were computed using multiple logistic regression analysis after adjusting for age. P-value o0.05 was considered significant. Results Characteristics of the study subjects are shown in Table 1. The mean age of the study population was years, and 47% of subjects were men. Men were older, had higher WC, WHR, systolic blood pressure, diastolic blood pressure, fasting plasma glucose, serum triglycerides and glycated hemoglobin (Po0.001 for all except fasting plasma glucose Po0.05) compared to women. Women had a higher BMI (Po0.05), hip circumference (Po0.001) and high-density lipoprotein cholesterol compared to men. Men had higher prevalence of diabetes compared to women; however, there is no significant difference in the prevalence of IGT between men and women. The socioeconomic status of the population was categorized as monthly income level less than Rs (72.6%), Rs (22.3%) and above Rs (5.1%) (1 US$ ¼ Rs. 43 approximately). 261

4 262 Table 1 General characteristics of the study population Variables Men (n ¼ 1096) Women (n ¼ 1254) P-value Age (years) o0.001 Education Illiterate Some schooling Graduate and above Body mass index (kg m 2 ) o0.05 Waist circumference (cm) o0.001 Hip circumference (cm) o0.001 Waist-to-hip ratio o0.001 Systolic blood pressure (mm Hg) o0.001 Diastolic blood pressure (mm Hg) o0.001 Fasting plasma glucose (mg dl 1 ) o0.05 Total cholesterol (mg dl 1 ) Triglycerides (mg dl 1 ) o0.001 HDL cholesterol (mg dl 1 ) o0.001 HbA1C (min: 3.6, max: 14.6) (min: 3.1, max: 16.6) o0.001 Diabetes o0.05 Impaired glucose tolerance Abbreviations: HbA1c, glycated hemoglobin; HDL, high-density lipoprotein. Prevalence of obesity (generalized and abdominal) Table 2 shows the crude- and age-standardized prevalence of generalized obesity based on different BMI cut points. The overall age-standardized prevalence of obesity using the international (western) definition of BMIX30 was 4.0%, CI: %, (women: 4.6%4 men: 3.3%, P ¼ 0.135). However, using the BMIX27.5, it was 9.9%, CI: %, (women: 10.7%4 men: 8.6%, P ¼ 0.085), while using the Asia Pacific definition of obesity BMIX25, it was 26.5%, CI: %, (women: 29.4%4 men: 22.8%, P ¼ 0.004), and finally, using the new definition of BMIX23, it was 45.9% CI: % (women: 47.4%4 men: 43.2%, P ¼ 0.210). The age-standardized prevalence of abdominal obesity using Asia Pacific definition (WC: men X90 cm and women X80 cm) was 46.6%, CI: % (women: 56.2%4 men: 35.1%, Po0.001). The prevalence of generalized obesity (Figure 2a) and abdominal obesity (Figure 2b) increased until the age of 50 years in both men and women and decreased thereafter. The prevalence of generalized obesity (BMIX23) did not vary significantly between the sexes, except for the age group of years, where women had significantly higher prevalence compared to men (Po0.05). The prevalence of abdominal obesity was significantly higher in women till the age of 60 years (Po0.05) after which the significance disappeared. Prevalence of isolated abdominal and isolated generalized obesity Isolated abdominal obesity (increased WC, BMIo23) was present in 4.7% of men and 14% of women (Table 3). The prevalence of isolated generalized obesity (normal WC, BMIX23) was 12.7% in men and 6.1% in women. Combined obesity (that is generalized plus abdominal) was seen in Table 2 points Prevalence of generalized obesity based on different BMI cut Prevalence rates Different cut points of BMI (kg m 2 ) BMIX23 BMIX25 BMIX27.5 BMIX30 Crude prevalence Overall Men Women Age-standardized prevalence Overall Men Women Abbreviation: BMI, body mass index. 32.6% of men and 43.3% of women. Men with isolated abdominal obesity had higher prevalence of all cardiovascular risk factors (diabetes, hypertriglyceridemia, hypercholesterolemia and low high-density lipoprotein cholesterol) except hypertension, compared with those with combined obesity. Women with combined obesity had higher prevalence of all cardiovascular risk factors (except hypercholesterolemia) compared with those with isolated abdominal and generalized obesity. Prevalence of obesity among subjects with different degrees of glucose intolerance In women, the prevalence of generalized obesity was highest in subjects with diabetes (72.6%) compared to those with IGT (66.2%) and normal glucose tolerance (43.9%) (trend w 2 : 62.8, Po0.001). In men, the prevalence was highest in subjects with IGT (normal glucose tolerance: 39.8%,

5 Prevalence of generalized obesity Prevalence of abdominal obesity * * * * * Men Women Age group (years) 22.5 Men Women Age group (years) Figure 2 The age- and gender-specific prevalence of generalized and abdominal obesity among urban south Indian population. (a) Generalized obesity. BMIX23 kg m 2 for both men and women. (b) Abdominal obesity (WCX90 cm for men and X80 cm for women). *Po0.05 compared to men. 20 diabetes: 61.4% and IGT: 62.8% (trend w 2 : 43.8, Po0.001)). Diabetic subjects had higher prevalence of abdominal obesity (men: 54.4% and women: 80.5%, Po0.001) compared to those with IGT (men: 53.6% and women: 72.9%, Po0.001) and normal glucose tolerance (men: 32.2% and women: 52.4%, Po0.001) (Figure 3). Correlation between obesity indices and metabolic risk factors Bivariate correlation between BMI, WC and metabolic risk factors are shown in Table 4. The r values of BMI with other metabolic risk factors ranged from (for fasting blood glucose) to (for diastolic blood pressure). The r values of WC with the metabolic risk factors ranged from (for fasting blood glucose) to (for diastolic blood pressure). Sensitivity and specificity of the obesity cut points A BMI cut point of 23 kg m 2 identified those with cardiometabolic risk factors with a sensitivity of and specificity of Using the WC cut point of X90 cm for men and X80 cm for women, the sensitivity ranged from to in men and from to in women and specificity from to in men and from to in women for identifying metabolic risk factors. The area under the curve of the obesity indices are summarized in Table 5. The area under the curve of WC was greater than that of BMI in all combinations of metabolic risk factors in women when compared to men. For identifying subjects with any three metabolic risk factors, the receiver operating characteristic of BMI was area under the curve 0.66, 95% CI: , Po0.001, while that of WC was 0.70, 95% CI: , Po0.001 for men and 0.69, 95% CI: , Po0.001 for women. 263 Table 3 Prevalence of cardiovascular risk factors by BMI and waist circumference categories Obesity type n Diabetes Hypertension Hypertriglyceridemia Hypercholesterolemia Low HDL cholesterol Men (using BMIX23 kg m 2 ) Normal 511 (46.6) Isolated abdominal obesity 52 (4.7) 32.7* 50.0* 48.1* 40.4* 65.9 # Isolated generalized obesity 139 (12.7) 21.6 # * 26.6 # 69.1* Combined 357 (32.6) 24.9* 50.7* 43.1* 35.9* 61.3* Women (using BMIX23 kg m 2 ) Normal 437 (34.8) Isolated abdominal obesity 175 (14.0) 12.0 # 28.6 # 24.5* 36.6* 69.0 # Isolated generalized obesity 76 (6.1) 11.8 # # Combined 543 (43.3) 20.4* 34.3* 28.4* 32.8 * 78.8* Abbreviations: BMI, body mass index; HDL, high-density lipoprotein. Waist circumference cut point: X90 cm for men and X80 cm for women; BMI cut point X23. Normal: both waist circumference and BMI are below their cut points; isolated abdominal obesity: waist circumference above cut point but BMI below cut point; isolated generalized obesity: BMI above cut point but waist circumference below cut point; combined: both waist circumference and BMI above their cut point. Diabetes 2-h post glucose X200 mg dl 1 (X11.1 mmol l 1 ) or self-reported diabetes; hypertension systolic BP/diastolic BPX130/85 mm Hg or self-reported hypertension; hypertriglyceridemia serum triglycerides X150 mg dl 1 (X1.7 mmol l 1 ) or on drug treatment; hypercholesterolemia: serum cholesterol X200 mg dl 1 (X5.2 mmol l 1 ) or on drug treatment; low HDL cholesterol HDL cholesterol o40 mg dl 1 (o1.04 mmol l 1 ) for men and o50 mg dl 1 (o1.3 mmol l 1 ) for women or on drug treatment. *Po0.001 and # Po0.05 compared to subjects with normal waist and normal BMI.

6 264 The age-adjusted odds ratios for predicting metabolic risk factors using BMIX23 ranged from 0.74 for any one risk factor to 3.23 for identifying any three risk factors (Table 5). Using the WC cut point, in men it ranged from 0.69 for any one risk factor to 2.81 for any three risk factors. In women, it ranged from 0.72 for any one risk factor to 4.09 for any three risk factors. Discussion This study shows the following results in urban Asian Indians studied at Chennai in southern India, using WHO Asia Pacific guidelines to define generalized and abdominal obesity: (i) the age-standardized prevalence of generalized obesity (BMIX23 kg m 2 ) was 45.9%, while that of abdominal obesity was 46.6%; (ii) isolated generalized obesity was more common in men and isolated abdominal obesity more common in women; and (iii) abdominal obesity showed greater correlation with cardiometabolic abnormalities than generalized obesity. The age-standardized prevalence rate of generalized obesity in urban South Indians in this study was 4% using BMIX30; 9.9% using BMIX27.5; 26.5% using BMIX25; and 45.9% using BMIX23 kg m 2. This is quite similar to figures reported from other Asian countries. The Chinese National Nutrition Survey showed that the prevalence of obesity (BMIX25) was 17.2% in Shanghai, 26.5% in Tianjin and 32.8% in Beijing (Ge, 1997). Using (BMIX30), the prevalence of obesity (BMIX30) among Hong Kong Chinese population was 2.2% in men and 4.8% in women, while using a cut point of BMIX27 in men and X25 in women, the prevalence was 10% in men and 27.9% in women (Ko et al., 1997). In Malaysia, the prevalence of obesity using BMI (25 30) was 24% in men and 18.1% in women, while using BMI430, it was 4.7% in men and 7.7% in women (Ismail et al., 1995). In Thailand, the prevalence of obesity was 16.7% using the cut point of BMI and 4% using BMI430 (Chuprapavarn, 1996). The age-standardized prevalence of abdominal obesity in the present study was 46.6% (men: 35.1%, women: 56.2%, Po0.001). Prevalence rates of abdominal obesity (WC: men X102 cm and women X88 cm) were reported to be 27.1, 20.2 and 21.4% for white, black and hispanic men, and corresponding figures were 43.2, 56 and 55.4% for white, black and hispanic women, respectively, in the United States (Okosun et al., 1999). Hong Kong Chinese women had much Table 4 Bivariate correlation between obesity indices and metabolic risk variables Variables Body mass index, r(p-value) Waist circumference, r (P-value) Body mass index (kg m 2 ) (Po0.001) Waist circumference (cm) (Po0.001) 1 Fasting blood glucose (Po0.001) (Po0.001) (mg dl 1 ) Serum cholesterol (Po0.001) (Po0.001) (mg dl 1 ) Serum triglycerides (Po0.001) (Po0.001) (mg dl 1 ) HDL cholesterol (mg dl 1 ) (Po0.001) (Po0.001) Systolic blood pressure (Po0.001) (Po0.001) (mm Hg) Diastolic blood pressure (mm Hg) (Po0.001) (Po0.001) Abbreviation: HDL, high-density lipoprotein. All variables are adjusted for age and gender. r is correlation coefficient Prevalence of obesity Abdominal obesity (waist circumference -WC) Generalized obesity (BMI) Normal glucose tolerance Impaired glucose tolerance Diabetes Normal glucose tolerance Impaired glucose tolerance Diabetes Figure 3 Men The prevalence of obesity based on diabetes status among south Indian adults. Women

7 Table 5 Sensitivity, specificity, area under curve of obesity index cut points and the age-adjusted odds ratio to predict metabolic risk factors 265 Obesity measure No. of metabolic risk factors Sensitivity Specificity Area under the curve (AUC) (95% CI) Age- and gender-adjusted odds ratios (95% CI) P-value BMIX23 (Men and Women) Any one ( ) 0.74 ( ) * Any two ( ) 1.82 ( ) * Any three ( ) 3.23 ( ) * Waist: men X90 cm and women X80 cm Any one Men ( ) 0.69 ( ) # Women ( ) 0.72 ( )# Any two Men ( ) 1.61 ( )# Women ( ) 1.96 ( ) * Any three Men ( ) 2.81 ( ) * Women ( ) 4.09 ( ) * ; Abbreviations: BMI, body mass index; BP, blood pressure; CI, confidence interval; HDL, high-density lipoprotein. Metabolic risk factors are based on revised ATP III definition (Grundy, 2005), for defining metabolic syndrome, which includes elevated fasting blood glucose (fasting blood glucose X100 mg dl 1 (X5.6 mmol l 1 ) or on treatment for elevated fasting glucose), elevated blood pressure (systolic BP/diastolic BPX130/85 mm Hg or on antihypertensive drug treatment), elevated triglycerides (serum triglycerides X150 mg dl 1 (X1.7 mmol l 1 ) or on drug treatment), reduced HDL cholesterol (HDL cholesterol o40 mg dl 1 (o1.04 mmol l 1 ) for men and o50 mg dl 1 (o1.3 mmol l 1 ) for women or on drug treatment). *Po0.001 and # Po0.05. higher prevalence of abdominal obesity (22.5%) (WC: men X94 cm and women X80 cm) compared to men (4.5%). A case control study in Nagpur, Maharashtra, India showed that abdominal obesity using WC is a better predictor of the risk of diabetes compared to WHR, abdominal volume index and conicity index (Mamtani and Kulkarni, 2005). The Heart Outcomes Prevention Evaluation (HOPE) study reported that abdominal adiposity worsens the prognosis of patients with cardiovascular disease (CVD) (Dagenais et al., 2005). Our study shows that in men, at all age groups, generalized obesity is higher compared to abdominal obesity (except at years). Among women, abdominal obesity is higher than generalized obesity. However, when we raised the cut point to BMIX25 kg m 2, isolated generalized obesity was found in only 2.9% of men and 1.6% of women; whereas, isolated abdominal obesity was found in 16.6% of men and 28.7% of women (data not shown). It is worth emphasizing that the prevalence of generalized obesity ranged from 4.3 to 48.3% in the same population studied depending on the cut points used. It is therefore clear that defining obesity either by BMI or WC is purely arbitrary, and the figures can vary widely based on the cut points used. Thus, data from different ethnic groups using different obesity index cut points should be interpreted with caution. Studies by Kumar et al. (2006) reported that there is a large difference in generalized and central obesity between immigrant groups from developing countries, especially in south Asians. In this study also, we find that WC yields higher correlation coefficients for metabolic risk factors compared to BMI, suggesting that WC may be a better index than BMI to identify metabolic risk factors in Asian Indians. However, there is need for prospective studies to confirm these findings. It is of interest that 10% of subjects without generalized obesity and 10.4% of subjects without abdominal obesity had diabetes in this study. This shows that the risk of developing diabetes starts even at normal BMI and WC in our population. Indeed, four of the five countries with the most cases of diabetes are in Asia, and the risk of diabetes has been shown to start at a much lower BMI (Nakagami et al., 2003; Yajnik, 2006) and WC (Vikram et al., 2003) in Asians compared to Europeans. High cardiovascular-related morbidities at upper normal BMI range have been reported in several Asian countries including Korea (Moon et al., 2002), Singapore (Deurenberg-Yap et al., 2001), Japan (Ito et al., 2003) and Taiwan (Lin et al., 2002). Waist-to-hip ratio, which is another useful measure of abdominal obesity, was not included in this analysis, because this article aims to study the impact of generalized obesity (measured by BMI) and abdominal obesity (measured by WC) in predicting metabolic risk factors. However, it would be of interest to study the association of WHR with metabolic risk factors in future studies, as the INTERHEART study based on data from 52 countries has shown the importance of WHR, and indeed, the protective effect of hip circumference (Yusuf et al., 2005). Studies by Yajnik (2004, 2006) have reported the association between WHR and hyperglycemia and have demonstrated that the higher WHR represents an exaggerated thrifty phenotype due to nutritional transition: small hips represent fetal deprivation and a large waist the subsequent abundance. Pua and Ong (2005) showed that obesity indices like BMI, WC, WHR and waist-to-stature ratio complemented one another, and the latter proved to be the best anthropometric index for screening cardiovascular disease risk factors among Singaporean women. A meta-analysis of

8 266 prospective studies emphasized the importance of incorporating WHR or WC into cardiovascular disease risk assessment (de Koning et al., 2007). Metabolic and vascular risks for obesity are manifested at a lower BMI in developing countries compared with those in developed countries (Deurenberg-Yap et al., 2002). The strengths of the present study are that the subjects studied are representative of the urban population of Chennai, the sample size is large (n ¼ 2350) and the response rate is good (90.4%). However, there are several limitations. First, while representative of urban India, this data cannot be applied to rural India where 70% of India s population lives. Second, the cross-sectional nature of the design does not allow for cause effect relationships to be made. Finally, the conclusions were based on metabolic risk factors rather than clinical end points or mortality data. Prospective studies are needed to address these issues and to determine the relationship of obesity- to disease-related end points in South Asians. The following three important findings from the public health perspective emerge from the present study: (a) the prevalence of obesity, both generalized and abdominal is alarmingly high in this population; (b) isolated generalized obesity is more common in men, whereas isolated abdominal obesity is more common in women in this population; (c) abdominal obesity shows a greater correlation with cardiometabolic abnormalities. These findings suggest that in Asian Indians, measuring WC is a better method of estimating obesity-related cardiovascular disease risk than BMI. These findings lend further support for the inclusion of WC as a component of Indian Diabetes Risk Score (IDRS), which has been shown to be a very effective tool for predicting undiagnosed diabetes (Mohan et al., 2005) and also for identifying the metabolic syndrome (Mohan et al., 2006) in this population. We intend to follow up this cohort prospectively for outcome events, which could provide more definite information regarding the relationship of obesity indices with cardiovascular mortality and/or morbidity. Acknowledgements We are grateful to the Chennai Willingdon Corporate Foundation, Chennai for the financial support provided for the study. We thank the Epidemiology team members for conducting the CURES field studies. This is the 47th publication from CURES. References Alberti KG, Zimmet PZ (1998). Definition diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus, provisional report of a WHO consultation. Diabet Med 15, Chuprapavarn J (1996). First report on survey of health status in Thai populations by questionnaires and physical examination, Institute of Public Health: Bangkok. Dagenais GR, Yi Q, Mann JFE, Bosch J, Pogue J, Yusuf S (2005). On behalf of the Heart Outcomes Prevention Evaluation (HOPE) study investigators. Prognostic impact of body weight and abdominal obesity in women and men with cardiovascular disease. Am Heart J 149, de Koning L, Merchant AT, Pogue J, Anand S (2007). Waist circumference and waist-to-hip ratio as predictors of cardiovascular events: meta-regression analysis of prospective studies. Eur Heart J 28, Deepa M, Pradeepa R, Rema M, Anjana M, Deepa R, Shanthirani S et al. (2003). The Chennai Urban Rural Epidemiology Study (CURES) study design and methodology (Urban Component) (CURES 1). J Assoc Physicians India 51, Deurenberg-Yap M, Chew SK, Deurenberg P (2002). Elevated body fat percentage and cardiovascular risks at low body mass index levels among Singaporean Chinese, Malays and Indians. Obes Rev 3, Deurenberg-Yap M, Chew SK, Lin VF, Tan BY, van Staveren WA, Deurenberg P (2001). Relationships between indices of obesity and its co-morbidities in multi-ethnic Singapore. Int J Obes Relat Metab Disord 25, Ge L (1997). Body mass index in young Chinese adults. Asia Pac J Clin Nutr 6, Grundy SM (2005). Metabolic Syndrome Scientific Statement by the American Heart Association and the National Heart, Lung, and Blood Institute. Arterioscler Thromb Vasc Biol 25, International Diabetes Federation (2005). New IDF worldwide definition of the metabolic syndrome. Press Conference, 1st International Congress on Pre-diabetes and the Metabolic Syndrome, Berlin, Germany, 14 April,( Ismail MN, Zawiah H, Chee SS, Ng KK (1995). Prevalence of obesity and chronic energy deficiency (CED) in adult Malaysians. Malays J Nutr 1, 1 9. Ito H, Nakasuga K, Ohshima A, Maruyama T, Kaji Y, Harada M et al. (2003). Detection of cardiovascular risk factors by indices of obesity obtained from anthropometry and dual-energy X-ray absorptiometry in Japanese individuals. Int J Obes Relat Metab Disord 27, Ko GT, Chan JC, Woo J, Lau E, Yeung VT, Chow CC et al. (1997). Simple anthropometric indexes and cardiovascular risk factors in Chinese. Int J Obes 21, Kumar BN, Meyer HE, Wandel M, Dalen I, Holmboe-Ottesen G (2006). Ethnic differences in obesity among immigrants from developing countries, in Oslo, Norway. Int J Obes 30, Lin WY, Lee LT, Chen CY, Lo H, Hsia HH, Liu IL et al. (2002). Optimal cut-off values for obesity: using simple anthropometric indices to predict cardiovascular risk factors in Taiwan. Int J Obes Relat Metab Disord 26, Mamtani MR, Kulkarni HR (2005). Predictive performance of anthropometric indexes of central obesity for the risk of type 2 diabetes. Arch Med Res 36, Mohan V, Deepa R, Deepa M, Somannavar S, Datta M (2005). A simplified Indian Diabetes Risk Score for screening for undiagnosed diabetic subjects. J Assoc Physicians India 53, Mohan V, Sandeep S, Deepa M, Gokulakrishnan K, Datta M, Deepa R (2006). A diabetes risk score helps identify metabolic syndrome and cardiovascular risk in Indians the Chennai Urban Rural Epidemiology Study (CURES 38). Diabetes Obes Metab 9, Moon OR, Kim NS, Jang SM, Yoon TH, Kim SO (2002). The relationship between body mass index and the prevalence of obesity-related diseases based on 1995 National Health Interview Survey in Korea. Obes Rev 3, Nakagami T, Qiao Q, Carstensen B, Nhr-Hansen C, Hu G, Tuomilehto J et al. (2003). The DECODE-DECODA Study Group. Age, body

9 mass index and type 2 diabetes-associations modified by ethnicity. Diabetologia 46, National cholesterol education program (NCEP) (2001). Executive summary of the third report of the national cholesterol education program (NCEP) Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 285, Okosun IS, Prewitt TE, Cooper RS (1999). Abdominal obesity in United States: prevalence and attributable risk of hypertension. J Hum Hypertens 13, 425. Pua YH, Ong PH (2005). Anthropometric indices as screening tools for cardiovascular risk factors in Singaporean women. Asia Pac J Clin Nutr 14, Vikram NK, Pandey RM, Misra A, Sharma R, Rama Devi J, Khanna N (2003). Non-obese (BMIo25 kg/m 2 ) Asian Indians with normal waist circumference have high cardiovascular risk. Nutrition 19, World Health Organization (WHO) (2000). The Asia Pacific Perspective. Redefining Obesity and Its Treatment. International Associaton for the Study of Obesity and International Obesity Task Force. International Diabetes Institute: Melbourne. World Health Organization (WHO) Expert Consultation (2004). Appropriate body mass index for Asian populations and its implications for policy and intervention strategies. Lancet 363, Yajnik CS (2004). Symposium on Adipose tissue development and the programming of adult obesity. Obesity epidemic in India: intrauterine origins? Proc Nutr Soc 63, Yajnik CS (2006). Waist hip ratio: a thrifty phenotype? Natl Med J India 19, Yusuf S, Hawken S, Ounpuu S, Bautista L, Franzosi MG, Commerford P et al. (2005). Obesity and the risk of myocardial infarction in participants from 52 countries: a case control study. Lancet 366,