Optimal Cutoff Points of Waist Circumference for the Criteria of Abdominal Obesity

ORIGINAL ARTICLE Epidemiology Circ J 2009; 73: 2068 2075 Optimal Cutoff Points of Waist Circumference for the Criteria of Abdominal Obesity Comparison With the Criteria of the International Diabetes Federation Inkyung Baik, PhD Background: There are few studies that compare measures of diagnostic accuracy for selected waist cutoff points and the criteria of abdominal obesity given by the International Diabetes Federation (IDF) and the National Cholesterol Education Program (NCEP). Methods and Results: The present study utilized data from the Third Korean National Health and Nutrition Examination Survey. Analyses for receiver-operating characteristic (ROC) curve were performed with data for 4,677 men and women aged 20 80 years who reported being free of a physician-diagnosed cardiovascular disease or cancer. On the basis of measures of diagnostic accuracy, including minimum distance to ROC curve and Youden index, waist circumference of 84 86 cm for men and of 78 80 cm for women were found to be optimal cutoff points. The study also demonstrated that the use of smaller waist circumference for diagnosis of the metabolic syndrome decreases discrepancies between the prevalence of the IDF-defined metabolic syndrome and that of the NCEP-defined metabolic syndrome. Conclusions: Waist circumference of 90 cm, the cutoff point given by the IDF for Asian men, may not be an appropriate value for Korean men, while its cutoff point for Asian women is appropriate for Korean women. (Circ J 2009; 73: 2068 2075) Key Words: KNHANES; Korean adults; Metabolic syndrome; Optimal cutoff point; Waist circumference Metabolic syndrome is reportedly associated with elevated risks of mortality and morbidity due to coronary heart disease (CHD). 1,2 Focusing on reducing CHD risk, the National Cholesterol Education Program (NCEP) Adult Treatment Panel III and the International Diabetes Federation (IDF) have characterized metabolic syndrome as a confluence of components, including abdominal obesity, dyslipidemia, raised blood pressure and impaired fasting glucose, and established diagnostic criteria for each component. 3,4 Compared with the NCEP criteria for the metabolic syndrome, however, the IDF has proposed the requirement of abdominal obesity and emphasized ethnic-specific cutoff points of waist circumference. 4 Cutoff points for waist circumference suggested by the IDF are 94 cm for men with Europoid origin, 90 cm for South Asian, Chinese and Japanese men, and 80 cm for women of all ethnicities. The revised NCEP upholds the criteria of larger waist circumference, 102 cm for men and 88 cm for women, for North Americans of non-asian origin and the same criteria given by the IDF for those with Asian-origin. 4,5 Although the IDF and the NCEP seem to have recently agreed on the cutoff points of waist circumference for Asians, a number of studies for Korean, Chinese and Japanese adults have produced conflicting findings. 6 16 Studies for Korean adults have suggested that 89 90 cm for men and 85 86 cm for women may be appropriate cutoff points of waist circumference for the definition of abdominal obesity. 7,8 Chinese data have demonstrated that minimum waist cutoff points associated with the metabolic syndrome are 90 cm for men and 85 cm for women, while nationally representative data have revealed that it is 80 cm for both men and women. 9,10 Showing more inconsistent results, investigations among Japanese adults have provided varied optimal cutoff points of waist circumference. 11 16 On the basis of these findings for some Asian ethnicities, more reliable cutoff points of waist circumference need to be proposed than those from the IDF and the NCEP criteria. In spite of several reports on cutoff points of waist circumference among Korean adults, data are limited in comparing measures of diagnostic accuracy, which are obtained from receiver-operating characteristic (ROC) curves, for selected optimal cutoff points with those for waist cutoff points given by the IDF. The present study analyzed nationally representative data for Korean adults to select optimal cutoff points of waist circumference and to evaluate which waist cutoff points are appropriate among varied cutoff points, including those suggested by the IDF and the NCEP, for the criteria of abdominal obesity as a component of the metabolic syndrome. Received May 7, 2009; revised manuscript received June 24, 2009; accepted June 25, 2009; released online August 28, 2009 Department of Food and Nutrition, College of Natural Sciences, Kookmin University, Seoul, Republic of Korea Mailing address: Inkyung Baik, PhD, Department of Food and Nutrition, College of Natural Sciences, Kookmin University, Jeongneung-gil 77, Jeongneung-dong 861-1, Seongbuk-gu, Seoul 136-702, Republic of Korea. E-mail: ibaik@kookmin.ac.kr All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: cj@j-circ.or.jp

Optimal Cutoff Point of Waist Circumference Methods Study Population Male and female adults aged 20 80 years were identified from data of the Third Korean National Health and Nutrition Examination Survey (KNHANES III), which was conducted in 2005 by the Korea Centers for Disease Control and Prevention. Detailed information on survey design and characteristics was reported elsewhere. 17,18 Briefly, KNHANES has been periodically performed to collect data related to health and nutritional status from noninstitutionalized civilians in Korea. To select a representative sample, a multistage cluster sampling technique was used, based on information of governing districts and types of dwelling. The survey procedures consist of health interviews and examinations, a health behavior survey and a nutrition survey. These are conducted after written consent has been obtained from each participant. The health examinations include anthropometric measurements, measurements of blood pressure and blood sampling for biochemical assays. The health interviews obtain demographic and socioeconomic information, medical history and health conditions, behaviors and lifestyle patterns related to health status, and dietary intake. The KNHANES III has been performed for a total of 33,848 individuals and approximately 15% completed health examinations. The present study has initially identified 5,152 participants aged 20 80 years who had completed health examinations after the exclusion of 38 pregnant women. To minimize the influence of anthropometric changes after a diagnosis of disease, 475 patients who had reported physician-diagnosed cardiovascular disease or cancer, or who had reported the use of hypoglycemic agents, were further excluded. After this exclusion, the remaining 4,677 adults were considered to have self-reported good health and thus comprised study subjects for statistical analyses. 19 Definition of the Metabolic Syndrome To obtain the prevalence of the metabolic syndrome, the diagnostic criteria given by the IDF and the NCEP were used. According to the IDF criteria, a diagnosis for the metabolic syndrome requires abdominal obesity as a prerequisite component, with any 2 of the other 4 components; (1) high concentrations of serum triglycerides ( 150 mg/dl), (2) low concentrations of serum high-density lipoprotein (HDL)-cholesterol (<40 mg/dl for men and <50 mg/dl for women), (3) high blood pressure (systolic 130 mmhg, diastolic 85 mmhg, or the use of antihypertensive medications), and (4) high concentrations of fasting plasma glucose ( 100 mg/dl or the use of hypoglycemic agents). Abdominal obesity is defined as having 90 cm for Asian men and 80 cm for Asian women on the basis of the IDF. 4 The present study has also made a diagnosis of abdominal obesity using the selected optimal cutoff points of waist circumference. The revised NCEP criteria have established a diagnosis of the metabolic syndrome as the presence of 3 or more of 5 components, including abdominal obesity and 4 nonadipose components, of which diagnostic criteria are same as suggested by the IDF. 5 Measurements of the Metabolic Syndrome Components Trained researchers conducted measurements of waist circumference at the end of a normal expiration, with the subject standing upright with arms by their side. Waist 2069 circumference was measured at the midpoint between the lower rib and the iliac crest, and recorded to the nearest 0.1 cm. Blood pressure was measured in a sitting position with mercury sphygmomanometers after a 10-min period of rest. Repeated measurements of blood pressure were performed after an approximately 5-min interval and recorded to the nearest 2 mmhg. The systolic and diastolic blood pressures used in this report were based on the average of the 2 readings. Individuals participating in the health examinations had fasted for at least 8-h before the beginning of blood collection. Assays were conducted for blood concentrations of total cholesterol, HDL-cholesterol, triglycerides and glucose. CHD Prediction Scores The CHD prediction scores were calculated using Framingham risk equations, as suggested by D Agostino et al, which estimate risks for CHD in the next 10 years. 20 The equations include risk variables such as age, diagnosis of diabetes, current smoking, categories of systolic/diastolic blood pressure (<120/<80, <130/<85, <140/<90, <160/<100, 160/ 100 mmhg), total cholesterol (<160, 160 199, 200 239, 240 279, 280 mg/dl), and HDL-cholesterol (<35, 35 44, 45 49, 50 59, 60 mg/dl). The CHD prediction scores are evaluated as high risk ranges for greater than 20%, as intermediate risk ranges for 10 20%, and as low risk ranges for less than 10%. As seen above, the CHD scoring system includes components, except smoking, of the metabolic syndrome. Because the metabolic syndrome is considered a secondary target for the prevention of CHD risk, the association of waist circumference with the CHD prediction scores is able to provide additional data along with the association with the metabolic syndrome. Statistical Analysis Unweighted and weighted values were calculated for 5 components of the metabolic syndrome. To assess differences in the means or proportions of unweighted values between men and women, Student s t-test and a Kruskal- Wallis test were used. To obtain optimal cutoff points of waist circumference, ROC curve analyses were performed according to 2 outcomes; (1) whether subjects had 2 or more nonadipose components such as triglycerides, HDL-cholesterol, blood pressure and glucose; and (2) whether subjects were in the high risk ranges on the basis of the CHD prediction scores. Optimal cutoff points were determined mainly on the basis of the maximum values of the Youden index, calculated by [sensitivity + specificity 1] and the minimum values of the square root of [(1-sensitivity) 2 + (1-specificity) 2 ], which indicates minimum distance from the upperleft corner to the point on the ROC curve. 21,22 In addition, other measures of diagnostic accuracy, such as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and total accuracy, were next considered. Briefly explaining these measures of diagnostic accuracy, sensitivity is an index with the ability to evaluate to identify true cases with a certain condition (or disease) correctly as outcome; specificity is an index with the ability to evaluate to identify true non-cases that do not have the condition correctly; PPV is the proportion of true cases with the condition among all cases classified with the condition; NPV is the proportion of true non-cases without the condition among all non-cases classified without the condition; total accuracy is the proportion of cases whose tests accurately predict the true cases or non-cases. 21 Because the measure-

2070 BAIK I Table 1. Characteristics of 4,677 Study Subjects Aged 20 80 Years From the Third Korea National Health and Nutrition Examination Survey Data by Sex Men Unweighted values Weighted values Women Unweighted values Weighted values P values* Total n of subjects (%) 1,995 (42.7) (50.1) 2,682 (57.3) (49.9) Age, mean ± SE, years 45.3±0.3 41.1±0.4 45.2±0.3 42.4±0.4 0.68 Body mass index, mean ± SE, kg/m 2 24.0±0.1 24.0±0.1 23.4±0.1 23.2±0.1 <0.001 Waist circumference, mean ± SE, cm 84.1±0.2 83.5±0.3 77.9±0.2 77.3±0.3 <0.001 Nonadipose MS components Triglycerides, mean ± SE, mg/dl 161.7±3.6 156.2±3.8 109.9±1.3 106.7±1.6 <0.001 HDL-cholesterol, mean ± SE, mg/dl 42.6±0.2 42.7±0.3 47.4±0.2 47.7±0.3 <0.001 Systolic blood pressure, mean ± SE, mmhg 121.9±0.3 120.3±0.4 115.1±0.3 113.4±0.5 <0.001 Diastolic blood pressure, mean ± SE, mmhg 80.8±0.2 80.2±0.4 74.5±0.2 73.8±0.4 <0.001 Fasting glucose, mean ± SE, mg/dl 95.3±0.4 93.7±0.5 91.0±0.3 90.3±0.4 <0.001 n of subjects having 2 or more MS components (%) 920 (46.1) (41.9) 889 (33.2) (29.4) <0.001 n of subjects having HTG (%) 751 (37.6) (35.6) 496 (18.5) (17.0) <0.001 n of subjects having LHDL (%) 883 (44.3) (42.5) 1,684 (62.8) (61.8) <0.001 n of subjects having HBP (%) 848 (42.5) (38.3) 702 (26.2) (22.9) <0.001 n of subjects having HFG (%) 471 (23.6) (22.4) 374 (13.9) (12.4) <0.001 n of subjects in the high risk ranges of CHD (%) 309 (15.5) (11.8) 345 (12.9) (10.8) <0.05 MS, metabolic syndrome; HDL, high-density lipoprotein; HTG, high triglycerides; LHDL, low HDL-cholesterol; HBP, high blood pressure; HFG, high fasting glucose; CHD, coronary heart disease. *To assess differences in the means or proportions of unweighted values between men and women, Student s t-test and a Kruskal-Wallis test were used. Nonadipose MS components include high levels of blood triglycerides ( 150 mg/dl) and fasting glucose ( 100 mg/dl), low levels of blood HDL-cholesterol (<50 mg/dl for women or <40 mg/dl for men), and HBP (systolic/diastolic 130/85 mmhg). The high risk ranges of CHD are defined as CHD prediction scores greater than 20%, which are calculated using Framingham risk equations. Table 2. Measures of Diagnostic Accuracy From the ROC Curve and Cutoff Points of Waist Circumference for the Presence of 2 or More Nonadipose Components* by Sex Cutoff Sensitivity Specificity PPV NPV Accuracy Distance to Youden point (%) (%) (%) (%) (%) ROC curve index Men ROC curve area = 0.746 Optimal cutoff point 1 85 69.2 68.0 64.9 72.1 68.6 0.20 0.37 Optimal cutoff point 2 86 64.9 71.8 66.3 70.5 68.6 0.20 0.37 NCEP cutoff point for North Americans (non-asian origin) 102 3.6 99.1 76.7 54.6 55.0 0.93 0.03 IDF cutoff point for Europoids 94 20.5 94.2 75.3 58.1 60.3 0.63 0.15 IDF cutoff point for South Asians, Chinese, and Japanese 90 41.0 85.0 70.1 62.7 64.7 0.37 0.26 Women ROC curve area = 0.771 Optimal cutoff point 1 78 76.8 64.9 52.0 85.0 68.8 0.18 0.42 Optimal cutoff point 2 79 72.1 69.1 53.6 83.3 70.1 0.17 0.41 NCEP cutoff point for North Americans (non-asian origin) 88 33.9 92.0 67.8 73.7 72.7 0.44 0.26 IDF cutoff point for Europoids, South Asians, Chinese, and Japanese 80 67.5 73.1 55.4 81.9 71.2 0.18 0.41 Suggestion by other studies 85 57.3 79.6 63.9 74.7 70.9 0.22 0.37 PPV, positive predictive value; NPV, negative predictive value; ROC, receiver-operating characteristic; NCEP, the National Cholesterol Education Program Adult Treatment Panel III; IDF, the International Diabetes Federation. Other abbreviations see in Table 1. *Nonadipose MS components include high levels of blood triglycerides ( 150 mg/dl) and fasting glucose ( 100 mg/dl), low levels of blood HDL-cholesterol (<50 mg/dl for women or <40 mg/dl for men), and HBP (systolic/diastolic 130/85 mmhg). References 8 and 9. ment of waist circumference is a simple, easy and inexpensive method, in the present study, a measure with higher sensitivity and NPV has been preferred to another with higher specificity when both measures have same total accuracy. The measures of diagnostic accuracy from ROC curve analyses were compared for waist circumference suggested by the IDF, the NCEP and previous studies. 4,5,8,9 Using the optimal cutoff points of waist circumference, finally, weighted proportions of individuals with abdominal obesity and weighted prevalence of the metabolic syndrome were calculated among adults aged 20 years or older included in the KNHANES III data. All testing was based on a twosided level of significance. Results Among the study subjects with self-reported good health, 38.7% (n=1,809) were identified to have 2 or more nonadipose components of the metabolic syndrome. In Table 1, prevalence of the metabolic syndrome components were compared between men and women. Men are more likely to have greater waist circumference, higher blood pressure, higher levels of triglycerides and glucose, and lower levels of HDL-cholesterol compared with women (p value <0.001). Thus, more men were identified to have 2 or more components of the metabolic syndrome (p value <0.001) as well as a higher risk for CHD (p value <0.05) (Table 1). Table 2 presents results of ROC curve analyses for the

Optimal Cutoff Point of Waist Circumference 2071 Table 3. Measures of Diagnostic Accuracy From the ROC Curve and Cutoff Points of Waist Circumference for the Presence of 2 or More Nonadipose Components* by Sex Among Subjects Aged Younger Than 65 Years Cutoff Sensitivity Specificity PPV NPV Accuracy Distance to Youden point (%) (%) (%) (%) (%) ROC curve index Men aged <65 years ROC curve area = 0.751 Optimal cutoff point 1 85 69.1 68.5 64.4 72.9 68.8 0.19 0.38 Optimal cutoff point 2 86 64.7 72.5 66.0 71.4 69.0 0.20 0.37 NCEP cutoff point for North Americans (non-asian origin) 102 3.6 99.4 82.9 55.6 56.1 0.93 0.03 IDF cutoff point for Europoids 94 20.1 94.5 75.1 58.9 60.9 0.64 0.15 IDF cutoff point for South Asians, Chinese and Japanese 90 40.5 85.6 69.8 63.5 65.2 0.38 0.26 Women aged <65 years ROC curve area = 0.772 Optimal cutoff point 1 78 76.6 65.6 45.8 88.0 68.6 0.17 0.42 Optimal cutoff point 2 79 71.0 69.8 47.2 86.3 70.1 0.18 0.41 NCEP cutoff point for North Americans (non-asian origin) 88 30.7 92.5 60.9 77.8 75.5 0.49 0.23 IDF cutoff point for Europoids, South Asians, Chinese, and Japanese 80 65.8 73.8 48.9 85.0 71.6 0.19 0.40 Suggestion by other studies 85 44.9 86.5 62.2 76.0 72.7 0.32 0.31 Abbreviations see in Tables 1,2. *Nonadipose MS components include high levels of blood triglycerides ( 150 mg/dl) and fasting glucose ( 100 mg/dl), low levels of blood HDL-cholesterol (<50 mg/dl for women or <40 mg/dl for men), and HBP (systolic/diastolic 130/85 mmhg). References 8 and 9. Table 4. Measures of Diagnostic Accuracy From the ROC Curve and Cutoff Points of Waist Circumference for the High Risk Ranges of CHD Prediction Scores* by Sex Among Subjects Aged Younger Than 65 Years Cutoff Sensitivity Specificity PPV NPV Accuracy Distance to Youden point (%) (%) (%) (%) (%) ROC curve index Men aged <65 years ROC curve area = 0.700 Optimal cutoff point 1 85 77.5 54.8 16.5 95.5 57.2 0.25 0.32 Optimal cutoff point 2 84 87.4 49.3 16.6 97.1 53.2 0.27 0.37 NCEP cutoff point for North Americans (non-asian origin) 102 5.5 98.4 28.6 90.0 88.8 0.89 0.04 IDF cutoff point for Europoids 94 20.9 88.9 17.8 90.7 81.9 0.64 0.10 IDF cutoff point for South Asians, Chinese, and Japanese 90 44.0 75.9 17.4 92.2 72.6 0.37 0.20 Women aged <65 years ROC curve area = 0.783 Optimal cutoff point and IDF cutoff point for Europoids, South Asians, Chinese, and Japanese 80 79.1 65.6 12.7 98.0 66.4 0.16 0.45 NCEP cutoff point for North Americans (non-asian origin) 88 42.4 87.9 18.2 96.0 85.2 0.35 0.30 Suggestion by other studies 85 49.6 79.9 26.7 91.5 76.0 0.29 0.29 Abbreviations see in Tables 1,2. *The high risk ranges of CHD are defined as CHD prediction scores greater than 20%, which are calculated using Framingham risk equations. References 8 and 9. prevalence of having 2 or more nonadipose components of the metabolic syndrome. ROC curve areas are 0.75 for men and 0.77 for women, indicating that detection accuracy is quite fair according to the use of waist circumference measurements. Minimum distance to ROC curve, a maximum value of Youden index and high total accuracy were observed among men with a waist circumference of 85 cm and 86 cm. Women with a waist circumference of 78 cm and 79 cm showed a maximum value of Youden index and minimum distance to ROC curve, respectively. Although high total accuracy was observed among women with a waist circumference of 88 cm due to a high value of specificity, distance to ROC curve was longer and values of Youden index and sensitivity were lower than among those with other values of waist circumference. Because those with 80 cm showed fairly reliable results in terms of distance to ROC curve, Youden index, total accuracy and sensitivity, this cutoff point of waist circumference may be more appropriate than 88 cm for women (Table 2). Because the establishment of optimal cutoff points for waist circumference is the goal for preventing the metabolic syndrome and risks of the metabolic syndrome are increased in old age, analyses limited to data for adults younger than 65 years are needed. Thus, Table 3 demonstrates optimal cutoff points for waist circumference and ROC curve results among adults younger than 65 years. As shown similarly in Table 2, waist circumference values of 85 cm and 86 cm for men and of 78 cm, 79 cm and 80 cm for women appear more appropriate compared with other values given by the IDF, the NECP or previous studies (Table 3). 4,5,8,9 I attempted to find optimal cutoff points for waist circumference to identify persons who are in the high risk ranges of the CHD prediction risk scores. Table 4 shows that waist cutoff points of 84 cm and 85 cm for men and of 80 cm for women are optimal values for screening persons with high CHD risk prediction scores. Because of long distance to ROC curve and low values of Youden index and sensitivity, larger waist cutoff points, such as 102 cm and 94 cm for men and 88 cm and 85 cm for women, showed less ability to identify persons who are in the high risk

2072 BAIK I Figure 1. Comparison of receiver-operating characteristic (ROC) curves for waist circumference (continuous line) and body mass index (dashed line) by sex among subjects aged younger than 65 years. The curves have been plotted with sensitivity (y value) and [1-specificity] (x value). A illustration denotes ROC curves for the presence of 2 or more nonadipose components of the metabolic syndrome and B illustration for the high risk ranges (greater than 20%) of coronary heart disease prediction scores. Differences between the area under the ROC curve for waist circumference and that of body mass index were significant (p<0.001). Table 5. Weighted Prevalence Percent of Abdominal Obesity and the MS Using Waist Cutoff Points of 85 cm (Men) and of 80 cm (Women) vs of 90 cm (Men) and of 85 cm (Women) by Sex and Age Groups Among Adults Aged 20 Years or Older in the KNHANES III Data (n=5,255) Sex Age groups Prevalence using waist cutoff points of 85 cm (men) and of 80 cm (women) Prevalence using waist cutoff points of 90 cm (men) and of 85 cm (women) Abdominal obesity IDF-MS* NCEP-MS Abdominal obesity IDF-MS* NCEP-MS Men 20 39 years 32.6 16.8 19.8 17.0 10.2 15.4 40 59 years 56.1 40.3 46.5 30.9 23.7 38.2 60 79 years 54.3 41.2 46.6 31.9 25.0 37.6 80+ years 38.7 19.4 32.6 36.1 19.4 32.6 All age groups 44.7 29.2 33.8 24.5 17.5 27.4 Women 20 39 years 21.1 5.5 6.4 11.4 3.4 4.4 40 59 years 45.5 25.8 29.3 26.5 17.2 23.6 60 79 years 66.2 53.7 60.9 45.9 38.5 54.4 80+ years 59.3 47.6 59.0 32.7 28.5 48.7 All age groups 38.6 22.1 25.3 23.4 15.1 21.0 All 41.6 25.6 29.5 24.0 16.3 24.2 KNHANES III, the Third Korea National Health and Nutrition Examination Survey; IDF-MS, the metabolic syndrome defined by the International Diabetes Federation; NCEP-MS, the metabolic syndrome defined by the National Cholesterol Education Program Adult Treatment Panel III. Other abbreviations see in Tables 1,2. *Prevalence of the MS, which is defined as the presence of abdominal obesity plus 2 or more nonadipose components on the basis of the IDF criteria. Prevalence of the MS, which is defined as the presence of 3 or more components on the basis of the NCEP criteria. ranges of the CHD prediction scores (Table 4). Figure 1 shows optimal cutoff points of waist circumference, indicating a circle on the ROC curves. As presented in the A illustration showing the ROC curves for having 2 or more nonadipose components of the metabolic syndrome and in the B illustration showing those included in the high risk ranges of the CHD prediction scores, waist circumference values of 84 86 cm for men and of 78 80 cm for women were found to be optimal cutoff points of waist circumference for preventing metabolic syndrome and

Optimal Cutoff Point of Waist Circumference CHD. In addition, the presented ROC curves also show that waist circumference (continuous line) is a better index of adiposity than body mass index (dashed lines) for screening individuals at risk of metabolic syndrome and CHD (p values <0.001) (Figure 1). According to the waist circumference of 85 cm (men) and of 80 cm (women) obtained from the present study vs that of 90 cm (men) and of 85 cm (women) suggested by a previous study for Korean adults, Table 5 presents weighted proportions of adults with abdominal obesity and weighted prevalence of the metabolic syndrome among adults aged 20 years or older included in the KNHANES III data. 8 In categories by sex and age groups, greater proportions of abdominal obesity were calculated through the use of a waist circumference of 85 cm (men) and of 80 cm (women) than by the use of 90 cm (men) and 85 cm (women). Almost 45% of male adults were classified as having abdominal obesity through the use of 85 cm as a cutoff point of waist circumference, while 25% were men with greater than 90 cm. Among women, 39% and 23% were classified as having abdominal obesity through the use of 80 cm and 85 cm, respectively. The highest proportions of abdominal obesity were observed among men aged 40 59 years and women aged 60 79 years according to a waist circumference of 85 cm (men) and of 80 cm or 85 cm (women). With the use of a waist circumference of 90 cm, however, the highest proportion of abdominal obesity was observed among men aged 80 years or older (Table 5). These observations indicate that a waist circumference of 90 cm for men may be too large to screen individuals with abdominal obesity before the age ranges of high risk of CHD. Greater prevalence of the metabolic syndrome was calculated through the use of a waist circumference of 85 cm (men) and 80 cm (women) than by the use of 90 cm (men) and 85 cm (women). Among men, almost 29% (IDF-MS) and 34% (NCEP-MS) were classified to have the metabolic syndrome according to the use of waist circumference of 85 cm, while 18% (IDF-MS) and 27% (NCEP-MS) were according to the use of 90 cm. Among women, almost 22% (IDF-MS) and 25% (NCEP-MS) were diagnosed as having metabolic syndrome with the use of waist circumference of 80 cm, while 15% (IDF-MS) and 21% (NCEP-MS) were diagnosed with the use of 85 cm (Table 5). These observations reflect that there is a greater discrepancy, in particular among men, between the prevalence of the IDF-defined metabolic syndrome and that of the NCEP-defined metabolic syndrome when a waist circumference of 90 cm is used for the criteria of abdominal obesity rather than when the cutoff point of 85 cm is used. As shown in Figure 2, discrepancies between the prevalence of the IDF-defined metabolic syndrome and that of the NCEP-defined metabolic syndrome were greater when larger values of waist circumference were used. Discussion The nationally representative data for Korean adults suggest that a waist circumference of 84 86 cm for men and of 78 80 cm for women may be optimal cutoff points of waist circumference for the criteria of abdominal obesity, which is a component of metabolic syndrome. As proposed by the IDF and the revised NCEP Adult Treatment Panel III for Asian women, 80 cm is an appropriate cutoff point for Korean women. However, a waist circumference of 90 cm, which is the cutoff point for Asian men according to the 2073 Figure 2. Comparison of prevalence of the IDF-defined (continuous line) and the NCEP-defined metabolic syndrome (dashed line) according to the varied criteria of waist circumference using the KNHANES III data including male and female adults aged 20 years or older. Abbreviations: IDF, the International Diabetes Federation; NCEP, the National Cholesterol Education Program Adult Treatment Panel III; KNHANES III, the Third Korean National Health and Nutrition Examination Survey. IDF-defined metabolic syndrome: the metabolic syndrome is defined as the presence of abdominal obesity plus 2 or more nonadipose components. NCEP-defined metabolic syndrome: the metabolic syndrome is defined as the presence of 3 or more of 5 components including abdominal obesity and 4 nonadipose components. IDF criteria and the NCEP criteria, may not be appropriate for Korean men. Smaller waist circumference values, such as 85 cm or 86 cm, may be better cutoff points rather than 90 cm for screening male adults with abdominal obesity as well as for preventing the metabolic syndrome before the onset of old age. There is a clear discrepancy between the prevalence of the IDF-defined metabolic syndrome and that of the NCEP-defined metabolic syndrome when a waist circumference of 90 cm, rather than of 85 cm or 86 cm, is used as the cutoff point of adipose component for men. There have been accumulated data on optimal cutoff points of waist circumference for Asians since the NCEP published the criteria of abdominal obesity and metabolic syndrome. 3 The early proposal of the NCEP for the criteria of abdominal obesity was waist circumference of 102 cm for men and of 88 cm for women. 3 On the basis of these criteria, only 2% (weighted prevalence) of men and 16% (weighted prevalence) of women are classified with abdominal obesity among adults aged 20 years or older in the

2074 BAIK I KNHANES III data. Because of such an underestimation of the prevalence of abdominal obesity, specific criteria to define adiposity for Asians have been called for. Thus, the IDF has ambitiously proposed the ethnic-specific criteria for abdominal obesity and the revision of the NCEP criteria including waist cutoff points for people with Asian-origin has followed. 4,5 Nevertheless, because a number of studies for Asians have suggested lower cutoff points for waist circumference, controversies over this criteria have not yet been resolved. 10 12,14 16 To define abdominal obesity for Korean adults, some studies, but not all, have suggested that a waist circumference of 90 cm for men and 85 cm or 86 cm for women may be appropriate cutoff points. 6 8 However, there are few studies that compare varied measures of diagnostic accuracy, beyond sensitivity and specificity, for selected cutoff points with other cutoff points of waist circumference, such as those proposed by the IDF and the NCEP. 13 To obtain optimal cutoff points, the present study has utilized several measures of diagnostic accuracy, such as minimum distance to ROC curve, maximum Youden index and total accuracy, along with sensitivity and specificity, and compared these measures for selected cutoff points and other waist cutoff points. Such comparison seemed to provide additional information to justify the suggestion on the optimal cutoff points of waist circumference. The findings have shown data supportive of the waist cutoff point of 80 cm for women as proposed by the IDF and the NCEP, but conflicting with their criterion of waist circumference for men. Thus, a new criterion of waist circumference, in particular of smaller waist circumference such as 84 cm, 85 cm or 86 cm, is proposed for Asian men, at least for Korean men, rather than a waist circumference of 90 cm. Oizumi et al has proposed that waist cutoff points of 85 cm and 80 cm for Japanese men and women, respectively, are appropriate for diagnosis of the metabolic syndrome. 23 For Chinese men and women, recently, the same criteria have been suggested as the optimal waist circumference in the ROC analyses for diabetes, hypertension and dyslipidemia. 24 These criteria may be also preferred among Korean men and women for the practical use of abdominal obesity criteria. Recent studies have pointed out that the IDF criteria identify less cases of the metabolic syndrome among men than the revised NCEP criteria. 25,26 Similarly, the present study has observed a lower prevalence of the metabolic syndrome with the use of the IDF criteria compared with that of the NCEP criteria. Furthermore, because discrepancies in the prevalence estimated using the 2 criteria are remarkably larger among men than among women and increase with larger waist circumference, it would be essential to determine a waist cutoff point that is proper for Korean men if both the IDF and NCEP criteria continue to be utilized. Strengths of the present study include the use of recent and nationally representative data, assessment of a number of measures of diagnostic accuracy to select optimal cutoff points for waist circumference, and comparison of measures of diagnostic accuracy for the waist cutoff points selected in the present study and other cutoff points established previously. In addition, because of the use of national data, which had been collected periodically, the present study has additional strengths such as the large sample size and anthropometric measurements based on standardized methods. A potential limitation is the use of cross-sectional survey data. Further studies may need to investigate for optimal cutoff points of waist circumference in relation to the incidence of the metabolic syndrome. The present study included subjects who met the criteria of health level 1, which was suggested by Flegal. 19 If more stringent definitions of health level, such as meeting the criteria of health level 1 plus the absence of hypertension, hyperlipidemia or endocrine/metabolic dysfunctions, had been used, a smaller waist circumference might have been observed to be optimal. Thus, studies for individuals with different health status may be needed. Waist circumference is often measured at the umbilicus. 27 Although it is less likely that different measurement procedures for waist circumference could have significantly altered the study results, further investigations are warranted to clarify the effect of measurement procedures on determining optimal waist circumference. 28 In conclusion, the nationally representative data for Korean adults suggest that a waist circumference of 84 86 cm for men and of 78 80 cm for women may be optimal cutoff points of waist circumference for the criteria of abdominal obesity. To prevent metabolic syndrome, smaller waist circumference values for Asian men may be more desirable rather than a waist circumference of 90 cm proposed by the IDF. In addition, the use of smaller waist circumference may resolve discrepancies between the prevalence of the IDF-defined metabolic syndrome and that of the NCEP-defined metabolic syndrome. Further investigation is warranted to confirm the criteria of waist circumference given by the present study. Acknowledgment This work was supported by the research program 2009 of Kookmin University in Korea. References 1. Isomaa B, Almgren P, Tuomi T, Forsén B, Lahti K, Nissén M, et al. Cardiovascular morbidity and mortality associated with the metabolic syndrome. Diabetes Care 2001; 24: 683 689. 2. Lakka HM, Laaksonen DE, Lakka TA, Niskanen LK, Kumpusalo E, Tuomilehto J, et al. The metabolic syndrome and total and cardiovascular disease mortality in middle-aged men. JAMA 2002; 288: 2709 2716. 3. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. 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 2001; 285: 2486 2497. 4. International Diabetes Federation. IDF worldwide definition of the metabolic syndrome. Available at http://www.idf.org/webdata/docs/ IDF_Meta_def_final.pdf (accessed April 28, 2009). 5. Grundy SM, Cleeman JI, Daniels SR, Donato KA, Eckel RH, Franklin BA, et al; American Heart Association; National Heart, Lung, and Blood Institute. Diagnosis and management of the metabolic syndrome: An American Heart Association/National Heart, Lung, and Blood Institute Scientific Statement. Circulation 2005; 112: 2735 2752. 6. Lee K. Waist circumference as a screening tool for cardiovascular risk factors in Korea: Evaluations of receiver operating characteristics (ROC). J Korean Acd Fam Med 2000; 21: 395 405. 7. Kim JA, Choi CJ, Yum KS. Cut-off values of visceral fat area and waist circumference: Diagnostic criteria for abdominal obesity in a Korean population. J Korean Med Sci 2006; 21: 1048 1053. 8. Lee SY, Park HS, Kim DJ, Han JH, Kim SM, Cho GJ, et al. Appropriate waist circumference cutoff points for central obesity in Korean adults. Diabetes Res Clin Pract 2007; 75: 72 80. 9. Bao Y, Lu J, Wang C, Yang M, Li H, Zhang X, et al. Optimal waist circumference cutoffs for abdominal obesity in Chinese. Atherosclerosis 2008; 201: 378 384. 10. Wildman RP, Gu D, Reynolds K, Duan X, He J. Appropriate body mass index and waist circumference cutoffs for categorization of over-

Optimal Cutoff Point of Waist Circumference weight and central adiposity among Chinese adults. Am J Clin Nutr 2004; 80: 1129 1136. 11. Nishimura R, Nakagami T, Tominaga M, Yoshiike N, Tajima N. Prevalence of metabolic syndrome and optimal waist circumference cut-off values in Japan. Diabetes Res Clin Pract 2007; 78: 77 84. 12. Hara K, Matsushita Y, Horikoshi M, Yoshiike N, Yokoyama T, Tanaka H, et al. A proposal for the cutoff point of waist circumference for the diagnosis of metabolic syndrome in the Japanese population. Diabetes Care 2006; 29: 1123 1124. 13. Hayashi T, Boyko EJ, McNeely MJ, Leonetti DL, Kahn SE, Fujimoto WY. Minimum waist and visceral fat values for identifying Japanese Americans at risk for the metabolic syndrome. Diabetes Care 2007; 30: 120 127. 14. Narisawa S, Nakamura K, Kato K, Yamada K, Sasaki J, Yamamoto M. Appropriate waist circumference cutoff values for persons with multiple cardiovascular risk factors in Japan: A large cross-sectional study. J Epidemiol 2008; 18: 37 42. 15. Sato A, Asayama K, Ohkubo T, Kikuya M, Obara T, Metoki H, et al. Optimal cutoff point of waist circumference and use of home blood pressure as a definition of metabolic syndrome: The Ohasama study. Am J Hypertens 2008; 21: 514 520. 16. Shiwaku K, Anuurad E, Enkhmaa B, Nogi A, Kitajima K, Yamasaki M, et al. Predictive values of anthropometric measurements for multiple metabolic disorders in Asian populations. Diabetes Res Clin Pract 2005; 69: 52 62. 17. Kim HM, Kim DJ, Jung IH, Park C, Park J. Prevalence of the metabolic syndrome among Korean adults using the new International Diabetes Federation definition and the new abdominal obesity criteria for the Korean people. Diabetes Res Clin Pract 2007; 77: 99 106. 18. Park SH, Lee KS, Park HY. Dietary carbohydrate intake is associated with cardiovascular disease risk in Korean: Analysis of the third Korea National Health and Nutrition Examination Survey (KNHANES III). Int J Cardiol 2008 Nov 13 [Epub ahead of print]. 19. Flegal KM. Waist circumference of healthy men and women in the 2075 United States. Int J Obes (Lond) 2007; 31: 1134 1139. 20. D Agostino RB Sr, Grundy S, Sullivan LM, Wilson P; CHD Risk Prediction Group. Validation of the Framingham coronary heart disease prediction scores: Results of a multiple ethnic groups investigation. JAMA 2001; 286: 180 187. 21. Szklo M, Nieto FJ. Epidemiology beyond the basics. Gaithersburg: ASPEN Publishers, Inc, 2000. 22. Perkins NJ, Schisterman EF. The inconsistency of optimal cutpoints obtained using two criteria based on the receiver operating characteristic curve. Am J Epidemiol 2006; 163: 670 675. 23. Oizumi T, Daimon M, Wada K, Jimbu Y, Kameda W, Susa S, et al. A proposal for the cutoff point of waist circumference for the diagnosis of metabolic syndrome in the Japanese population. Circ J 2006; 70: 1663. 24. Li R, Lu W, Jia J, Zhang S, Shi L, Li Y, et al. Relationship between indices of obesity and its cardiovascular comorbidities in a Chinese population. Circ J 2008; 72: 973 978. 25. Yoon YS, Lee ES, Park C, Lee S, Oh SW. The new definition of metabolic syndrome by the international diabetes federation is less likely to identify metabolically abnormal but non-obese individuals than the definition by the revised national cholesterol education program: The Korea NHANES study. Int J Obes (Lond) 2007; 31: 528 534. 26. Oda E, Abe M, Veeraveedu PT, Watanabe K. Considerable disagreement among definitions of metabolic syndrome for Japanese. Circ J 2007; 71: 1239 1243. 27. Nakamura Y, Turin T, Kita Y, Tamaki S, Tsujita Y, Kadowaki T, et al. Associations of obesity measures with metabolic risk factors in a community-based population in Japan. Circ J 2007; 71: 776 781. 28. Ross R, Berentzen T, Bradshaw AJ, Janssen I, Kahn HS, Katzmarzyk PT, et al. Does the relationship between waist circumference, morbidity and mortality depend on measurement protocol for waist circumference? Obes Rev 2008; 9: 312 325.