Defining Obesity Cut Points in a Multiethnic Population
Background— Body mass index (BMI) is widely used to assess risk for cardiovascular disease and type 2 diabetes. Cut points for the classification of obesity (BMI >30 kg/m2) have been developed and validated among people of European descent. It is unknown whether these cut points are appropriate for non-European populations. We assessed the metabolic risk associated with BMI among South Asians, Chinese, Aboriginals, and Europeans.
Methods and Results— We randomly sampled 1078 subjects from 4 ethnic groups (289 South Asians, 281 Chinese, 207 Aboriginals, and 301 Europeans) from 4 regions in Canada. Principal components factor analysis was used to derive underlying latent or “hidden” factors associated with 14 clinical and biochemical cardiometabolic markers. Ethnic-specific BMI cut points were derived for 3 cardiometabolic factors. Three primary latent factors emerged that accounted for 56% of the variation in markers of glucose metabolism, lipid metabolism, and blood pressure. For a given BMI, elevated levels of glucose- and lipid-related factors were more likely to be present in South Asians, Chinese, and Aboriginals compared with Europeans, and elevated levels of the blood pressure–related factor were more likely to be present among Chinese compared with Europeans. The cut point to define obesity, as defined by distribution of glucose and lipid factors, is lower by ≈6 kg/m2 among non-European groups compared with Europeans.
Conclusions— Revisions may be warranted for BMI cut points to define obesity among South Asians, Chinese, and Aboriginals. Using these revised cut points would greatly increase the estimated burden of obesity-related metabolic disorders among non-European populations.
Received June 19, 2006; accepted February 9, 2007.
Excess body weight is an important independent risk factor for cardiovascular disease (CVD) and other related chronic diseases such as type 2 diabetes.1 The most common measure of excess body weight in clinical practice and population surveys is the body mass index (BMI), defined as weight in kilograms divided by the square of height in meters. Conventional BMI classifications are overweight (25.0 kg/m2≤BMI<30.0 kg/m2) and obese (BMI ≥30.0 kg/m2). These cut points were derived primarily in European populations to correspond to risk thresholds for a wide range of chronic diseases and mortality.1 However, there is ongoing debate as to whether these criteria for obesity and overweight are appropriate for non-European populations.
Emerging evidence suggests that South Asians (people who originate from the Indian subcontinent) and Chinese suffer from an elevated risk of type 2 diabetes, hypertension, and dyslipidemia even if their BMI is low (ie, <25.0 kg/m2).2–6 Possible explanations include that non-Europeans have a relative excess of adipose tissue or deficit of lean body mass compared with Europeans for a given BMI.4,7,8
Editorial p 2089
Clinical Perspective p 2118
To date, 2 major consensus statements have attempted to define appropriate BMI ranges for Asian populations. A 2000 statement from the World Health Organization suggested that in Asian populations, being overweight should be defined as a BMI >23.0 kg/m2 and obesity as a BMI >25.0 kg/m2.9 In contrast, the 2004 World Health Organization statement indicated that a range of plausible BMI cut points for being overweight and obese existed for Asians (depending on the outcome assessed and the population examined) and suggested that it would not be possible to define a single set of new cut points in these populations.10
The aims of the present analysis are to derive appropriate BMI cut points to define obesity among individuals of South Asian, Chinese, Aboriginal, and European descent using a novel application of statistical methods and incorporating multiple metabolic abnormalities into a single construct.
Description of Study Sample
Individuals included in this analysis were participants in the Study of Health Assessment and Risk in Ethnic Groups (SHARE) and Risk Evaluation in Aboriginal Peoples (SHARE-AP). Subjects were randomly recruited from 3 Canadian cities (Toronto, Hamilton, and Edmonton) and the Six Nations Reserve, Ohsweken, Ontario.11,12 Targeted ethnic sampling of South Asians and Chinese was achieved by using surname databases merged with local telephone directories.13,14 To verify ethnicity, the subject had to report that both parents and all 4 grandparents had the same ethnicity, which was confirmed over the telephone and at the time of the clinic visit. Aboriginal subjects were required to be Six Nations Band Members and living on the Six Nations Reserve. Among eligible individuals, the response rate was 62% among Europeans, 59% among South Asians, 69% among Chinese, and 79% among Aboriginal people. All participants were between 35 and 75 years of age and had resided in Canada for a minimum of 5 years at their time of entry into the study. SHARE and SHARE-AP were approved by the McMaster University Hamilton Health Sciences Research Ethics Board and the Six Nations Band Council, respectively. More detailed descriptions of the study samples have been previously reported.11,12
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Anthropometric and Clinical Measures
All physical measures were gathered using a standardized protocol across the 4 centers. A detailed description was reported previously.15
All subjects underwent between an 8- and 12-hour fast before blood was drawn, and nondiabetic participants underwent an oral glucose tolerance test with measurement of glucose, insulin, triglyceride, and free fatty acid levels at baseline and at 2 hours after glucose load. HbA1c, measured with a 765 Glycomat analyzer using low-pressure cation exchange chromatography in conjunction with gradient elution to separate hemoglobin subtypes, was associated with a coefficient of variation <4%. Total cholesterol and glucose were measured with enzymatic methods16,17; low-density lipoprotein (LDL) cholesterol was calculated18; and high-density lipoprotein (HDL) cholesterol was measured after precipitation of very LDL and LDL fractions.19 Triglycerides were measured with the enzymatic, colorimetric method with cholesterol esterase, cholesterol oxidase, and 4 aminoantipyrine (ROCHE Cobas Integra, Basel, Switzerland). Insulin was measured by manual radioimmunoassay assay (Diagnostic Products Corp, Los Angeles, Calif). Free fatty acids were measured on the Chiron 560 Express using reagent kit NEFA-C (WAKO Chemical USA Inc, Richmond, Va). Homeostasis model assessment–insulin resistance (HOMA-IR) was used as a validated measure of estimated insulin resistance.20
First, principal components factor analysis was used to extract underlying factors related to 14 commonly used markers of glucose metabolism (fasting and 2-hour glucose, fasting and 2-hour insulin, HbA1c, HOMA-IR), lipid metabolism (LDL, HDL, fasting and 2-hour triglycerides, fasting and 2-hour free fatty acids), and blood pressure (systolic and diastolic blood pressures; Table 1). These variables were chosen to represent commonly used markers reported in the literature. Free fatty acid and triglyceride levels were included in this analysis because these markers are intricately related to glucose and lipid metabolism and to insulin resistance.21 Second, each underlying factor that emerged was used as the dependent variable in a saturated linear regression model with the independent variables BMI, ethnicity, and their interaction. Sex and age were not included as covariates because each marker was adjusted for these 2 variables before being entered into the principal components analysis (Appendix I in the online Data Supplement). Third, the predicted level of each factor (P) among Europeans was calculated at a BMI of 30.0 kg/m2 (the established cut point to classify obesity) using the models derived in step 2. Fourth, using the same models, we determined the value of BMI that predicted a factor level P among South Asians, Chinese, and Aboriginals (BS, BC, and BA, respectively). These values of B were taken as the candidate cut points for obesity in South Asians, Chinese, and Aboriginals because they were associated with a factor level of P (Appendix II in the Data Supplement contains a schematic outlining this process).
To approximate a 95% CI, we used a method similar to the fiducial approach.22 For example, in South Asians, we took the 95% confidence band on the regression line and identified for the upper and lower limits of the band the BMI values predicting a risk factor level P. These values were taken as the lower and upper limits for BS. A similar approach was used for the other ethnic groups.
All variable standardization was performed with a linear regression model. Skewed variables were transformed by taking the natural logarithm.
On average, European and Aboriginal participants had resided in Canada for twice as long as South Asian and Chinese participants (44.4 and 41.7 years compared with 19.8 and 20.6 years, respectively). All South Asians and 95% of Chinese participants reported being born outside of Canada.
Clinical and Biochemical Profiles
Of the initial 1247 participants, 169 had established diabetes and were excluded from this analysis because they did not have 2-hour post–oral glucose tolerance test levels of glucose, insulin, triglycerides, and free fatty acids. Of the remaining 1078 subjects, there were 301 European, 289 South Asians, 281 Chinese, and 207 Aboriginals. General demographic characteristics and mean marker levels are presented in Table 2. Aboriginals were older and a greater proportion were women compared with the other ethnic groups. There was marked ethnic variation in BMI in that the mean BMI among Aboriginals was >7 kg/m2 higher than in Chinese, with South Asians and Europeans intermediate to these groups. Fasting glucose and insulin levels were highest among Aboriginals, whereas the 2-hour post–oral glucose tolerance test levels of these parameters were highest among South Asians. South Asians also had the worst lipid profiles, with the highest levels of LDL and the lowest levels of HDL. Systolic blood pressure was similar between the 4 ethnic groups, whereas diastolic blood pressure was markedly lower among Aboriginals.
Principal Components Factor Analysis
Three factors emerged that accounted for 56% of the total variation of the 14 variables initially entered into the model (Table 1). Variables loading onto the first rotated factor are related to glucose metabolism (fasting and 2-hour glucose, HbA1c), insulin/estimated insulin resistance (HOMA-IR), and free fatty acids; those loading onto the second rotated factor are related to lipid metabolism (LDL, HDL, fasting and 2-hour triglycerides) and insulin/estimated insulin resistance; the third rotated factor is related uniquely to blood pressure (Figure 1). The 3 variables related to insulin/estimated insulin resistance—fasting and 2-hour insulin and HOMA-IR—are shared between the glucose and lipid factors. Free fatty acids also are shared to some extent between these 2 factors but loaded primarily on the glucose factor. Given that the variable loading patterns in the 3-factor model have face validity, agree with previous research, and have physiological underpinnings,23 they were used for comparison of risk level between ethnic groups in the next section. A visual summary of the relationship between observed variables and underlying factors is provided in Figure 1.
BMI Cut Points for Obesity
Linear regression models fit for each latent factor (glucose, lipid, and blood pressure) identified BMI cut points that corresponded to a BMI=30.0 kg/m2 among Europeans. The main effect of ethnicity was highly significant (P<0.001) for each factor. The levels of glucose factor (Figure 2) and lipid factor (Figure 3) were higher in non-Europeans compared with Europeans across the range of BMI values. The level of the blood pressure factor was higher in Chinese compared with Europeans, similar in South Asians and Europeans, and much lower in Aboriginals (Figure 4).
Elevated levels of the glucose factor appeared at very low BMI ranges in non-Europeans. Compared with the BMI cut point of 30.0 kg/m2 among Europeans, a similar glucose factor distribution was observed at corresponding BMI cut points of 21.0 kg/m2 in South Asians, 20.6 kg/m2 in Chinese, and 21.8 kg/m2 in Aboriginals (Table 3). Thus, the predicted cut point for the glucose factor in all 3 non-European ethnic groups was markedly lower than a BMI of 30.0 kg/m2 among Europeans. Cut points for the lipid factor comparing non-European ethnic groups with Europeans occurred at a BMI of 22.5 kg/m2 in South Asians, 25.9 kg/m2 in Chinese, and 26.1 kg/m2 in Aboriginals—moderately higher values than the cut point for the glucose factor but still substantially lower than a BMI of 30.0 kg/m2 in Europeans. Finally, cut points for the elevation in the blood pressure factor occurred in South Asians at a BMI of 28.8 kg/m2 and in Chinese at a BMI of 25.3 kg/m2 compared with a BMI of 30.0 kg/m2 in Europeans. Therefore, the BMI cut points for the blood pressure factor were significantly lower in Chinese but not in South Asians compared with Europeans. Across the range of BMI values, levels of the blood pressure factor were substantially lower in Aboriginals compared with all 3 ethnic groups, and no cut point was derived.
In this study, we demonstrate that South Asians, Chinese, and Aboriginal people have similar distributions of glucose and lipid factors at significantly lower BMI values compared with Europeans. Furthermore, Chinese have elevated blood pressure levels at a much lower BMI than Europeans, whereas blood pressure levels in Aboriginals are lower than the other 3 ethnic groups across the range of BMI. Our data support the belief that “normal ranges” for obesity using BMI cut points derived in European populations may be misleading when applied to populations such as South Asians, Chinese, and Aboriginals.2–8
Use of BMI cut points derived among Europeans understates the cardiometabolic risk associated with weight gain in other ethnic groups. The pathway linking obesity to clinical events is mediated partially through its strong association with the development of diabetes, hypertension, and dyslipidemia.1 Therefore, our focus on these important risk factors lends credence to the argument for varying BMI cut points for different ethnic groups. These findings have important implications for clinicians, who need to be aware that the definition of a “normal” or average BMI may not apply to non-European ethnic groups and that elevations in markers of glucose metabolism may occur at low BMI values (ie, 21.0 kg/m2). This suggests that to minimize the development of cardiometabolic risk factors, lower BMI targets should be used by healthcare professionals in some non-European populations.
Although a cut point to define obesity of a BMI of 21.0 kg/m2 may seem unrealistic, it should be noted that the majority of the populations of India and China still inhabit rural regions and have BMI values that are markedly lower than those observed in Western countries. For example, the 90th percentile of the BMI distribution in the rural Anqing region of China is 23.5 kg/m2 in men and 24.5 kg/m2 in women.24 In 1999, the proportion of the population living in urban areas was only 31.6% in China and 28.1% in India.25 It is only after the influence of an urban environment (abundant with calorie-dense foods and limited physical activity) that BMI ranges increase to a level that would be considered “normal” in Western countries. In countries such as India and China, there has historically been an emphasis on the prevention and treatment of underweight individuals (BMI <18.5 kg/m2). Future studies should examine the BMI values at which the transition occurs from risk of malnutrition-related illnesses to risk of chronic disease. However, it must be noted that in most developing countries, the prevalence of overweight (BMI >25.0 kg/m2) now exceeds the prevalence of underweight (BMI <18.5 kg/m2), and country-specific proportions of overweight to underweight should inform public health policies.26
An intriguing finding of our analysis was the similar cut points for obesity using either the glucose or lipid factor among South Asians (BMI, 21.0 and 22.0 kg/m2, respectively), whereas the cut point using the lipid factor was significantly higher among Aboriginals and Chinese compared with their cut points using the glucose factor (BMI, 25.9 versus 20.6 kg/m2 in Chinese and 26.1 versus 21.8 kg/m2 in Aboriginals). This suggests that the risk of CVD may be greater among overweight South Asians compared with overweight Chinese because South Asians face the double burden of dysglycemia and dyslipidemia. Further investigations are needed to elucidate why South Asians are prone to glucose and lipid changes with relatively minimal weight gain.9
The elevation in blood pressure levels among South Asians with weight gain does not appear to be greater than in Europeans. Similar results were found in a previous comparison of blood pressure values in South Asian and European Canadians.3 On the other hand, the elevation in blood pressure with weight gain observed among Chinese may explain in part their predilection to develop hypertension and thus their greater risk for stroke compared with ischemic heart disease.27 The low blood pressure levels in Aboriginals were unexpected and require further study. However, our observation is consistent with that of Weyer and colleagues,28 who compared normotensive Pima Indians and whites and found that despite higher levels of obesity among the Pima Indians (BMI, 32.2 versus 27.8 kg/m2), levels of systolic and diastolic blood pressures were similar between the 2 groups. The authors suggest that this finding may be explained by the fact that the Pima groups demonstrated a much weaker relationship between levels of insulin and adipose tissue to muscle sympathetic nervous activity compared with Europeans.
Comparison With Established Methods of Deriving BMI Cut Points in Non-European Populations
The elevated risk in South Asians and Chinese of metabolic abnormalities in normal weight ranges (BMI <25.0 kg/m2) has led to a number of attempts to redefine the cut point for overweight and obese in Asian populations.10 These approaches have used 2 classes of outcomes to compare risk levels between ethnic groups: elevated percent body fat or the prevalence of conditions such as diabetes, hypertension, or dyslipidemia. They also have used 2 approaches to derive ethnic-specific cut points: receiver-operating characteristic curves and logistic regression, both of which have shortcomings. The receiver-operating curve characteristic approach is dependent on the BMI distribution of the population under study and is restricted to studying 1 dichotomous outcome at a time.29,30 In general, a more rightward distribution will produce a higher optimal BMI cut point using the receiver-operating curve characteristic method.29 Logistics regression is limited because the interval at which the odds ratio attains significance is based on the choice of interval boundaries, and this impacts the sample size within each interval and the strength of the relationship with the outcome variable. Therefore, on the same data set, different BMI cut points could be obtained, depending on how BMI is categorized. Second, the point at which statistical significance is reached depends in part on sample size, and lack of a significant odds ratio could lead to the erroneous conclusion that an increase in risk level is not present.31 Finally, like the receiver-operating curve characteristic approach, the logistic regression approach examines only 1 outcome at a time.
Advantages of Using Factor Analysis
Prior analyses have used arbitrarily dichotomized cardiometabolic outcomes such as diabetes, hypertension, and dyslipidemia to define obesity. Therefore, all of the difficulties described above for the categorization of BMI could apply equally well to the problem, for example, of defining what levels of plasma glucose constitute diabetes. Furthermore, most of these risk markers share a continuous relationship with CVD, and large meta-analyses have not shown a clear “normal” level for plasma glucose, cholesterol, or blood pressure.32–34 Therefore, our analytic strategy uses an approach that simultaneously incorporates the varying markers used to measure a risk factor, is less dependent on the population distribution of the variable, and can be used to characterize continuous risk factors.
We derived BMI cut points at which similar risk factor distributions are present in non-Europeans compared with Europeans at a BMI of 30.0 kg/m2. However, because of the cross-sectional design and modest sample size of the present study, we cannot be certain whether these markers show ethnic variation in their relation to clinical events such as CVD. This would require a much larger prospective study involving several thousand CVD events in each of several ethnic groups. However, to the best of our knowledge, no evidence has emerged to support the hypothesis that certain ethnic groups are protected against the harmful effects of high glucose, cholesterol, or blood pressure. Second, although HOMA-IR is a validated measure of insulin resistance, we acknowledge that the use of more accurate measures from clamp studies would have improved our analysis. Third, our analytic approach could not identify cut points corresponding to overweight (BMI, 25.0 kg/m2) because of an insufficient sample of individuals in the SHARE study at very low BMIs (<20.0 kg/m2). Consequently, our regression models used to predict BMI cut points for overweight in non-Europeans produced variable results that depended greatly on the choice of model (data not shown). Similarly, the methods presented in the present study would plausibly work equally well to derive waist circumference and waist-to-hip ratio cut points but require a larger data set with greater variability in the range of anthropometric indexes. Fourth, our approach to derive a confidence region around predicted BMI cut points does not give an exact α level of 0.05. However, models suggest that the approach we used in the present study is conservative and gives an α level substantially below 0.05.35 Fifth, our 3 latent factors explained only 56% of the variation in observed markers, and although it would have been ideal if a greater proportion of variation had been explained, our analysis is consistent with the results of previous factor analyses of this type, and it may be possible that the residual amount of unexplained variation is an inherent limitation of the set of variables examined or their measurement error.23 Sixth, it is possible that some proportion of the ethnic variability in metabolic markers observed in the present study is due to differences in lifestyle factors (such as diet, physical activity, or tobacco), genetic factors, and lifestyle-gene interactions. Consideration of these factors may help explain the results of our analysis. It is important for future studies to examine the relative contributions of these factors to the development of metabolic abnormalities and clinical events. Finally, we cannot be certain of the extent to which results in non-European populations in Canada can be generalized to these populations in their country of origin.
Cut points for obesity in relation to the distribution of glucose-, lipid-, and blood pressure–related risk factors for CVD are much lower for South Asians, Chinese, and Aboriginals than the conventional value used for Europeans. Further research is required to examine whether individuals from these ethnic groups who exceed these lower BMI cut points have an increased risk of CVD.
Sources of Funding
Dr Anand is a recipient of a Canadian Institutes of Health Research Clinician Scientist Award (phase 2). Dr Yusuf is a recipient of a Canadian Institutes of Health Research Career Scientist Award and holds a Heart and Stroke Foundation of Ontario Research Chair. Grant support was received from the Heart and Stroke Foundation of Canada.
World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Report on a WHO consultation on obesity; June 3–5, 1997; Geneva, Switzerland. WHO/NUT/NCD/98.1; Geneva; 1998.
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.
Jafar TH, Chaturvedi N, Pappas G. Prevalence of overweight and obesity and their association with hypertension and diabetes mellitus in an Indo-Asian population. Can Med Assoc J. 2006; 175: 1071–1077.
WHO/IASO/IOTF. The Asia-Pacific Perspective: Redefining Obesity and Its Treatment. Melbourne, Australia: Health Communications Australia; 2000. ISBN 0–9577082–1–1.
Anand SS, Yusuf S, Vuksan V, Devanesen S, Teo KK, Montague PA, Kelemen L, Yi C, Lonn E, Gerstein H, Hegele RA, McQueen M. Differences in risk factors, atherosclerosis, and cardiovascular disease between ethnic groups in Canada: the Study of Health Assessment and Risk in Ethnic Groups (SHARE) Lancet. 2000; 356: 279–284.
Anand SS, Yusuf S, Jacobs R, Davis AD, Yi Q, Gerstein H, Montague PA, Lonn E. Risk factors, atherosclerosis, and cardiovascular disease among Aboriginal people in Canada: the Study of Health Assessment and Risk Evaluation in Aboriginal Peoples (SHARE-AP). Lancet. 2001; 358: 1147–1153.
Sheth T, Nargundkar M, Chagani K, Anand S, Nair C, Yusuf S. Classifying ethnicity utilizing the Canadian Mortality Data Base. Ethnic Health. 1997; 2: 287–295.
Allain CC, Poon LS, Chan CSG, Richmond W, Fu PC. Enzymatic determination of total serum cholesterol. Clin Chem. 1974; 20: 470–475.
Neeley WE. Simple automated determination of serum or plasma glucose by a hexokinase/glucose-6-phsophate dehydrogenase method. Clin Chem. 1972; 18: 509–542.
Friedwald WT, Levy RI, Fredickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972; 18: 499–502.
Warnick GR, Nguyen T, Albers AA. Comparison of improved precipitation methods for quantification of high-density lipoprotein cholesterol. Clin Chem. 1985; 31: 217–222.
Reaven GM. The metabolic syndrome: is this diagnosis necessary? Am J Clin Nutr. 2006; 83: 1237–1247.
Kendal MG, Stuart A. The Advanced Theory of Statistics, Volume 2: Inference and Relationships. 3rd ed. London, UK: Charles Griffin & Co Ltd; 1973.
Meigs JB. Invited commentary: insulin resistance syndrome? syndrome X? multiple metabolic syndrome? a syndrome at all? factor analysis reveals patterns in the fabric of correlated metabolic risk factors. Am J Epidemiol. 2000; 152: 908–911.
Hu FB, Wang B, Chen C, Jin Y, Yang J, Stampfer MJ, Xu X. Body mass index and cardiovascular risk factors in a rural Chinese population. Am J Epidemiol. 2000; 151: 88–97.
Human Development Report 2001. New York, NY: Oxford University Press; 2001. Published for the United Nations Development Programme (UNDP). Available at: http://hdr.undp.org/reports/global/2001/en/pdf/completenew.pdf. Accessed March 24, 2007.
Mendez MA, Monteiro CA, Popkin BM. Overweight exceeds underweight among women in most developing countries. Am J Clin Nutr. 2005; 81: 714–721.
Weyer C, Pratley RE, Snitker S, Spraul M, Ravussin E, Tataranni PA. Ethnic differences in insulinemia and sympathetic tone as links between obesity and blood pressure. Hypertension. 2000; 36: 531–537.
Stevens J, Juhaeri, Cai J, Jones DW. The effect of decision rules on the choice of a body mass index cutoff for obesity: examples from African American and white women. Am J Clin Nutr. 2002; 75: 986–992.
Coutinho M, Gerstein HC, Wang Y, Yusuf S. The relationship between glucose and incident cardiovascular events: a metaregression analysis of published data from 20 studies of 95,783 individuals followed for 12.4 years. Diabetes Care. 1999; 22: 233–240.
Chen Z, Peto R, Collins R, MacMahon S, Lu J, Li W. Serum cholesterol concentration and coronary heart disease in population with low cholesterol concentrations. BMJ. 1991; 303: 276–282.
Lewington S, Clarke R, Qizilbash N, Peto R, Collins R, for the Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies [published correction in Lancet. 2003;361:1060]. Lancet. 2002; 360: 1903–1913.
The prevalence of obesity has been increasing rapidly in both developing and developed countries. Overweight and obesity are primary determinants of cardiometabolic risk factors, including dysglycemia and dyslipidemia. There has been considerable debate in recent years over the appropriateness of using uniform cut points to define obesity in different ethnic groups. The most recent World Health Organization consensus statement on this topic suggested that, although lower body mass index cut points for the classification of obesity in Asian populations were likely, there were insufficient data to define these cut points clearly. In the present study, involving a multiethnic cohort from North America, ethnic-specific body mass index cut points are derived through the use of key cardiometabolic risk factors as outcomes. The results illustrate that current body mass index cutoffs derived in European populations may underestimate the prevalence of obesity and the associated detrimental cardiometabolic risk factors that occur with obesity among nonwhite populations. These results have implications for clinicians assessing the presence of obesity and cardiometabolic risk factors in their patients and have global relevance given the emerging obesity epidemic in the Indian subcontinent, China, and the Asia-Pacific region.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/cgi/content/full/CIRCULATIONAHA.106.635011/DC1.