Introduction. IS Okosun 1, CN Rotimi 2, TE Forrester 3, H Fraser 4, B Osotimehin 5, WF Muna 6 and RS Cooper 1

Transcription

1 (2000) 24, 180±186 ß 2000 Macmillan Publisher Ltd All rights reserved 0307±0565/00 $ Predictive value of abdominal obesity cut-off points for hypertension in Blacks from West African and Caribbean island nations IS Okosun 1, CN Rotimi 2, TE Forrester 3, H Fraser 4, B Osotimehin 5, WF Muna 6 and RS Cooper 1 1 Department of Preventive Medicine and Epidemiology, Loyola University Medical Center, Stritch School of Medicine, Maywood, Illinois, USA; 2 National Human Genome Center, Howard University College of Medicine, Washington, DC, USA; 3 Tropical Metabolism Research Unit, University of the West Indies, Mona: 4 Queen Elizabeth Hospital, University of the West Indies, Bridgetown, Barbados; 5 College of Medicine, University of Ibadan, Ibadan, Nigeria; and 6 General Hospital of Yaounde, Faculty of Medicine and Biomedical Sciences, University of Yaounde, Cameroon BACKGROUND: Waist circumferences (WC) 94 cm for men and 80 cm for women (action level I) and 102 cm for men and 88 cm for women (action level II) have been suggested as limits for health promotion purposes to alert the general public to the need for weight loss. In this analysis we examined the ability of the above cut-off points to correctly identify subjects with or without hypertension in Nigeria, Cameroon, Jamaica, St Lucia and Barbados. We also determined population- and gender-speci c abdominal adiposity cut-off points for epidemiological identi cation of risk of hypertension. METHODS: Waist measurement was made at the narrowest part of the torso as seen from the front or at midpoint between the bottom of the rib cage and 2 cm above the top of the iliac crest. Sensitivity and speci city of the established WC cut-off points for hypertension were compared across sites. With receiver operating characteristics (ROC), population- and gender-speci c cut-off points associated with risk of hypertension were determined over the entire range of WC values. RESULTS: Predictive abilities of the established WC cut-off points for hypertension were poor compared to the speci c cut-off points estimated for each population. Different values of WC were associated with increased risk of hypertension in these populations. In men, WC cut-off points of 76, 81, 80, 83 and 87 cm provided the highest sensitivity for identifying hypertensives in Nigeria, Cameroon, Jamaica, St Lucia and Barbados, respectively. The analogous cut-off points in women were 72, 82, 85, 86 and 88 cm. CONCLUSIONS: The waist cut-off points from this study represent values for epidemiological identi cation of risk of hypertension. For the purpose of health promotion, the decision on what cut ± off points to use must be made by considering other additional factors including overall impact on health due to intervention (e.g. weight reduction) and potential burden on health services if a low cut-off point is employed. There is a need to develop abdominal adiposity cut-off points associated with increased risks for cardiovascular diseases in different societies, especially for those populations where the distribution of obesity and associated risk factors tends to be very different from those of the technologically advanced nations. (2000) 24, 180±186 Keywords: abdominal obesity; central adiposity; waist circumference; hypertension; receiver operating characteristics Introduction It is increasingly being recognized that aberrant fat localization, such as abdominal adiposity, and not total body fat mass, is the most crucial determinant of cardiovascular diseases (CVDs). 1±4 First described as gynoid and android morphotypes 5 some 40 y ago, abdominal obesity is associated with increased risk of hypertension, 6±10 type 2 diabetes mellitus, 11,12 and stroke. 13,14 Abdominal adiposity has also been shown *Correspondence: IS Okosun, Department of Preventive Medicine and Epidemiology, Loyola University Medical Center, Stritch School of Medicine, 2160 S. First Ave, Maywood IL 60153, USA. iokosun@luc.edu. Received 13 April 1999; revised 20 July 1999; accepted 27 July 1999 to be associated with increased risk of overall mortality in some populations. 15,16 While the `gold standard' for assessing abdominal adiposity is the use of imaging techniques, these methods are impractical in large epidemiological studies because they are arduous and expensive. Hence, waist circumference (WC) is considered as the best anthropometric alternative for assessing abdominal adiposity. 2±4 Waist circumference is an aggregate measurement of the actual amount of total and abdominal fat accumulation and is a crucial correlate of abnormal syndromes found among obese and overweight patients. 17,18 Indeed, many are now advocating WC as a valid alternative to body mass index (BMI) for health promotion and the basis for alert values for those at risk of CVDs. 19 ± 21 In epidemiological research, waist=hip ratio (WHR) used to be the traditional anthropometric indicator for abdominal adiposity. However, because of the

2 inherent weakness of WHR as a ratio, 22,23 WC is becoming the most commonly used anthropometric surrogate of abdominal adiposity. 24 In contrast to WHR, WC has the advantage of being simpler to interpret and is better correlated with levels of visceral adipose tissue accumulation. 25 ± 29 Visceral adiposity is highly correlated with many metabolic abnormalities that are generally regarded as part of insulin resistance syndrome (IRS). 29 Components of these multiple metabolic aberrations include hypertension, glucose intolerance, hyperinsulinemia, hypercholesterolemia, hypertriglyceridemia, and high levels of low-density lipoprotein cholesterol. 30 There is no consensus on the WC cut-off points for abdominal adiposity. The most commonly cited cutoff points for abdominal adiposity are: (1) WC 94 cm for men and 80 cm for women for action level I; and (2) WC 102 cm for men and 88 cm for women for action level II. 10,21 These cut-off points were developed for health promotion purposes to give optimal enlightenment of individuals in need of weight management because of overweight=obesity or because of central fat distribution. Proposed originally by Lean et al, 21 and subsequently by Han et al, 10 at level I lifestyle modi cations were recommended, while level II required the use of professional help. Recently, the US National Heart, Lung, and Blood Institute (NHLBI), in cooperation with the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), af rmed action level II for Americans. 31 Action level II cut-off points were recommended for identifying increased relative risk of the development of obesity-associated risk factors for most adults with BMI of 25 ± 34.9 kg=m 2. The NHLBI=NIDDK panel made the recommendations based on review of published scienti c literature in MEDL1NE from 1980 to 1997 of topics identi ed as germane to the obesity evidence model. Although it has been suggested that the relationship between abdominal adiposity and CVDs may differ among populations and ethnic groups, 32,33 only very few studies have speci cally investigated abdominal adiposity cut-off points associated with different diseases in non-white populations. 34,35 In this study we take advantage of data collected from the International Collaborative Study of Hypertension in Blacks (ICSHIB) to (1) examine the ability of the recommended abdominal obesity cut-off points to correctly identify subjects with hypertension, and (2) determine gender-speci c values of abdominal adiposity predictive of hypertension risk in these African-origin nations. Methods and procedures ICSHIB was a multi-site study using random samples of residential communities in Africa, the Caribbean and the US. In this analysis, only data from West Africa and the Caribbean nations were utilized. In Nigeria, the sampling method was designed to cover rural (Idere) and urban (Ibadan) communities located in Oyo State in the southwest of Nigeria. In Cameroon, civil servants in the capital city of Yaounde and rural environs were sampled. The Caribbean samples were drawn from peri-urban areas of Spanish Town, Jamaica, Bridgetown, Barbados, and Vieux Forte, St Lucia. Overall, the ICSHIB survey and data collections were carried out between 1991 and 1995 in men and women 25 ± 74 y of age. Survey measurements Measurement protocols were standardized across sites using rigorous training techniques. Anthropometric measurements were obtained from participants without shoes and in light attire. Weight was measured using electronic digital scales and recorded in pounds to the nearest 0.5 lb. Height was measured in meters to the nearest 0.1 cm against a vertical wall with a rigid headboard using an inelastic tape measure. Two measurements of waist and hip circumference were made with a exible tape for each subject to the nearest 0.1 cm. When there was a discrepancy of more than 0.5 cm between two readings, a third measurement was taken and all three measurements were recorded. Waist measurement was made at the natural waist (narrowest part of the torso as seen from the front) or at midpoint between the bottom of the rib cage and 2 cm above the top of the iliac crest. Hip measurement was made at the point of maximum extension of the buttocks. The average of the last two measurements of waist and hip were employed for this investigation. Three blood pressure measurements were obtained from each subject using a standard mercury sphygmomanometer with 60 s intervals between cuff in ation. The average of the last two readings was utilized for this analysis. BMI was calculated as the measured weight in kilograms divided by height in meters squared (kg=m 2 ). WHR was calculated as waist divided by hip circumference. De nition of hypertension Hypertension was de ned in accordance with the de nition of the Joint National Committee on Detection, Evaluation and Treatment of High Blood Pressure as (a) systolic blood pressure (SBP) 140 mmhg, (b) diastolic blood pressure (DBP) 90 mmhg, or (c) current treatment for hypertension with medication Survey methods Detailed accounts of the sampling procedures for ICSHIB have been published elsewhere. 36 ± 38 Brie y, Statistical methods Statistical programs available in SPSS version 8.0 and SIMSTAT release 1.24 for Windows were utilized for

3 182 these analyses. 40,41 Gender-speci c means and standard deviations of WC and other anthropometric variables were computed and compared across sites. Pearson's correlation analysis was utilized to quantify the univariate association of WC with weight, height, hip circumference, BMI and WHR. To determine the ability of the established abdominal adiposity cut-off points to correctly identify subjects with or without hypertension we compared sensitivity and speci city across sites. Population- and gender-speci c numbers of subjects with or without hypertension at different levels of WC cut-off points were determined by cross-tabulation. These numbers were used to determine the receiver operating characteristics (ROC) of WC in identifying subjects with and without hypertension. With the ROC technique, comparison of sensitivity with the speci city rate was made over the entire range of WC. The WC cut-off point was determined by interpolation from the point of intersection of the lines of speci city and sensitivity (where sensitivity equaled speci city). The point of intersection between lines of speci city and sensitivity identi ed the highest numbers of subjects with and without hypertension. 1 We also calculated positive and negative predictive values of WC cut-off points for hypertension. Sensitivity was de ned as the percentage of the total number of subjects with hypertension who were correctly identi ed by the WC cut-off point, and speci city as the percentage of total number of subjects without hypertension who were correctly identi- ed by WC cut-off point. Positive predictive value was de ned as the percentage of subjects above or equal to the WC cut-off point who had hypertension, and negative predictive value was the percentage of those below the WC cut-off point who did not have hypertension. The overall performance of the ROC test was quanti ed by computing area under the curve (AUC). An AUC of 1 indicated perfect performance, while 0.5 indicated a performance that was not different from chance. 42 Ninety- ve percent con dence intervals were computed. Results Age and anthropometric characteristics of the 8746 eligible subjects are shown in Table 1. West African subjects tended to have lower height, weight, waist and hip circumference than the Caribbeans. With the exception of Jamaicans, men were heavier (weight) than women (P < 0.05). However, women had greater BMI than men in all populations. With the exception of Nigerians, women had larger waist girth than men. The largest relative difference between men and women in mean WC was observed in St Lucia, with a difference of 2.9 cm. West African subjects tended to have lower prevalence of hypertension than the Caribbeans. Site- and sex-speci c frequency distribution of waist girth is presented in Figure 1. Overall WC distribution differed between men and women. The most frequent waist size among Jamaica women, and Barbadian men and women was recorded at 90 cm, while in Nigeria, Cameroon, Jamaica (men), and St Lucia it was 80 cm. Frequencies of WC 120 cm for men in Nigeria, Cameroon, Jamaica, St Lucia and Barbados were 2.3%, 2.6%, 6.5%, 4.5% and 9.7%, respectively. The analogous values for women were 2.2%, 5%, 9.3%, 11.2% and 14.5%. The result of univariate analysis of the correlation of WC with weight, height, hip circumference, BMI and WHR is presented in Table 2. Waist circumference was positively correlated with weight, hip circumference, BMI and WHR among men and women across the populations (P < 0.01). With the exception of men from St Lucia and Barbados and women from Barbados, WC was also signi cantly correlated with height (P < 0.05), although the degree of correlation was weaker compared to weight, BMI and WHR. Overall, the correlation of WC with weight (except St Lucia), height, BMI and WHR was stronger in the Caribbeans compared to the West Africans. To evaluate the performance of the established abdominal adiposity cut-off points in identifying subjects with and without hypertension, we compared Table 1 Anthropometric pro le of subjects Site n Age (y) Height (cm)] Weight (kg) Waist (cm) Hip (circumference) WHR BMI (kg=m 2 ) Hypertension a (%) Nigeria Cameroon Jamaica St Lucia Barbados Nigeria Cameroon Jamaica St Lucia Barbados a Age-adjusted prevalence of hypertension, de ned as systolic blood pressure 140 mmhg or diastolic blood pressure 90 mmhg or using antihypertensive medications.

4 Figure 1 Frequency distribution of waist circumference in men and women from West African and Caribbean island nations. sensitivity and speci city associated with WC 94 cm and 102 cm for men and 80 cm and 88 cm for women across sites (Table 3). Overall, sensitivity was higher in women, while speci city was higher in men at any level of comparison at each site. Among men and women, sensitivity was signi cantly higher using WC cut-off points of 94 and 80 cm compared to 102 and 88 cm, respectively. In both men and women, sensitivity tended to be tightly linked with increasing values of WC and BMI. In men, a WC cut-point of 94 cm was associated with sensitivity for hypertension that ranged from 10% in Nigeria to 38% in Barbados, and with positive predictive values of 28% and 56%, respectively. The range of sensitivity for 102 cm varied from 3% in Nigeria to 12% in Barbados, with positive predictive values of 38% and 52%, respectively. In women a WC cut-off point of 80 cm resulted in sensitivity that ranged from 37% in Nigeria to 88% in Barbados, with positive predictive values of 26% and 45%, respectively. The corresponding values using a WC cut-off point of 88 cm in women ranged from 17% in Nigeria to 63% in Barbados, with positive predictive values of 27% and 50%, respectively. Table 4 summarizes the ndings of ROC analyses of WC values associated with the risk of hypertension that jointly provided optimal sensitivity without undue loss of speci city within sites. The results of the AUC analysis were signi cantly different from what was expected by chance (P < 0.05). The values of WC that correctly identi ed the highest numbers of subjects with or without hypertension were different across sites, ranging from 75.6 cm in Nigerian men to 87.3 cm in Barbadian men, with sensitivities of 63.1% and 65.4%, respectively. Among women, WC cut-off points that correctly identi ed the greatest number of hypertensives ranged from 71.5 cm in Nigeria to 87.5 cm in Barbados, with sensitivities of 59.6% and 62.8%, respectively. To estimate population- and sex-speci c trends in abdominal adiposity cut-off points related to age, WC cut-off points that correctly identi ed the highest numbers of subjects with or without hypertension in age groups 25 ± 34, 35 ± 44, 45 ± 54 and 55 and above were calculated (Table 5). In general, with the exception of men from Cameroon and women from Jamaica, WC cut-off points increased with increasing age, although the pattern was somewhat uneven between age group 45 ± 54 and 55 years and above. We also determined the abdominal adiposity cutpoints for epidemiological identi cation of hypertension risk in these Black populations of West African origin by analyzing the entire survey populations as a group. WC cut-off points of 80.6 and 81.0 cm correctly identi ed the greatest numbers of subjects with or without hypertension in men and women, respectively. Among men, the WC cut-off points that correctly identi ed the highest numbers of subjects with or without hypertension in age groups 25 ± 34, 35 ± 44, 45 ± 54 and 55 above were 77.5, 82.0, 82.8 and 81.6 cm respectively. The analogous values in women were 78.0, 81.9, 84.6 and 82.9 cm. Discussion Although it has been over 40 y since Vague rst described the importance of body habitus, and thus created awareness of the role of central or abdominal fat distribution in cardiovascular disease, 5 no consensus has yet emerged on the best cut-off points of 183 Table 2 (WHR) Correlation of waist circumference with weight, height, hip circumference, body mass index (BMI) and waist-to-hip ratio Site Weight Height Hip circumference BMI WHR Weight Height Hip circumference BMI WHR Nigeria 0.752** 0.070* 0.857** 0.633** 0.390** 0.809** 0.087** 0.783** 0.692** 0.436** Cameroon 0.785** 0.233** 0.784** 0.768** 0.518** 0.771** 0.161** 0.719** 0.778** 0.373** Jamaica 0.886** 0.120** 0.868** 0.888** 0.845** 0.895** 0.083* 0.866** 0.900** 0.580** St Lucia 0.768** ** 0.780** 0.756** 0.852** 0.149* 0.840** 0.839** 0.589** Barbados 0.809** ** 0.847** 0.831** 0.877** ** 0.889** 0.567** **P < 0.01; *P < 0.05.

5 184 Table 3 Country-speci c sensitivity, speci city, and positive and negative preventive values for hypertension using the NHLBI=NIDDK recommended waist circumference cut-off points Waist Sensitivity Specificity Positive predictive value Negative predictive value circumerence Site (cm) % [95% CI] % [95% CI] % [95% CI] % [95% CI] Nigeria [6.1 ± 14.1] 94.4 [92.9 ± 95.8] 28.6 [18.5 ± 38.7] 82.5 [80.3 ± 84.7] Cameroon [13.7 ± 22.2] 93.4 [91.9 ± 94.9] 44.8 [36.1 ± 53.5] 79.3 [77.0 ± 81.5] Jamaica [23.2 ± 38.3] 88.5 [85.6 ± 91.4] 45.4 [35.5 ± 55.3] 80.5 [77.0 ± 83.9] St Lucia [19.8 ± 34.5] 90.4 [87.3 ± 93.5] 52.8 [41.2 ± 64.3] 75.8 [71.7 ± 79.9] Barbados [26.9 ± 44.4] 85.1 [80.4 ± 89.9] 56.2 [44.8 ± 67.5] 71.2 [65.7 ± 76.7] Nigeria [0.6 ± 4.9] 98.9 [98.3 ± 99.6] 37.5 [13.8 ± 61.2] 82.1 [79.9 ± 84.2] Cameroon [2.7 ± 7.6] 98.8 [98.1 ± 99.4] 55.2 [37.1 ± 73.3] 77.8 [75.5 ± 80.0] Jamaica [5.5 ± 15.5] 95.7 [93.8 ± 97.5] 42.9 [26.5 ± 59.3] 77.5 [74.1 ± 80.9] St Lucia [2.4 ± 10.5] 98.3 [97.0 ± 99.7] 60.0 [32.3 ± 84.8] 72.7 [68.7 ± 76.6] Barbados [6.2 ± 18.2] 94.0 [90.8 ± 97.1] 51.9 [33.0 ± 70.0] 66.7 [61.4 ± 72.0] Nigeria [30.4 ± 41.9] 74.8 [72.3 ± 77.4] 25.8 [21.4 ± 30.2] 82.9 [80.6 ± 85.3] Cameroon [62.2 ± 72.3] 48.9 [46.0 ± 51.9] 28.0 [21.9 ± 31.1] 83.5 [80.7 ± 86.3] Jamaica [65.4 ± 75.4] 48.7 [44.4 ± 53.1] 58.6 [55.3 ± 62.0] 72.5 [67.8 ± 77.2] St Lucia [69.3 ± 81.4] 44.2 [39.3 ± 49.0] 39.5 [34.5 ± 44.5] 78.8 [73.4 ± 84.6] Barbados [82.1 ± 92.8] 39.6 [34.1 ± 44.9] 44.9 [39.6 ± 50.1] 85.4 [79.7 ± 91.2] Nigeria [12.3 ± 21.3] 88.9 [87.1 ± 90.8] 26.8 [20.1 ± 33.5] 81.6 [79.4 ± 83.8] Cameroon [32.9 ± 43.3] 82.5 [80.3 ± 84.7] 39.1 [33.9 ± 44.4] 81.8 [79.6 ± 84.1] Jamaica [41.6 ± 52.5] 74.2 [70.4 ± 78.0] 53.2 [47.4 ± 59.0] 69.2 [65.4 ± 73.1] St Lucia [44.8 ± 58.8] 64.5 [59.8 ± 69.2] 40.8 [36.7 ± 44.7] 73.4 [68.8 ± 78.0] Barbados [55.5 ± 69.8] 65.3 [60.0 ± 70.4] 50.2 [43.6 ± 56.9] 75.7 [70.6 ± 80.9] CI: con dence interval. Table 4 Estimated sensitivity, speci city, and positive and negative predictive values associated with hypertension using populationspeci c waist circumferences from West African and Carribean nations. Sensitivity Specificity Positive predictive value Negative predictive value Area under curve Site WC % [95% CI] % [95% CI] % [95% CI] % [95% CI] % [95% CI] Nigeria [57.4 ± 68.8] 63.1 [48.1 ± 54.5] 27.5 [24.0 ± 31.1] 82.6 [79.5 ± 82.7] 0.60 [0.56 ± 0.64] Cameroon [53.0 ± 63.9] 58.4 [51.6 ± 57.6] 27.8 [24.3 ± 31.2] 81.5 [78.6 ± 84.3] 0.60 [0.57 ± 0.63] Jamaica [53.8 ± 69.7] 61.8 [57.1 ± 66.0] 33.5 [27.8 ± 39.1] 83.7 [79.8 ± 87.7] 0.66 [0.61 ± 0.71] St Lucia [60.4 ± 75.8] 68.1 [61.6 ± 71.5] 44.9 [38.9 ± 47.7] 83.9 [79.6 ± 88.2] 0.72 [0.67 ± 0.78] Barbados [57.4 ± 73.4] 65.4 [57.6 ± 70.4] 53.6 [46.0 ± 61.2] 53.6 [46.0 ± 61.2] 0.65 [0.59 ± 0.72] Nigeria [54.3 ± 64.9] 59.6 [40.4 ± 64.5] 24.3 [21.3 ± 27.3] 77.9 [74.5 ± 81.2] 0.54 [0.52 ± 0.57] Cameroon [53.1 ± 63.6] 58.3 [55.7 ± 61.4] 29.4 [25.1 ± 32.8] 82.6 [80.0 ± 85.2] 0.54 [0.52 ± 0.58] Jamaica [53.4 ± 63.2] 57.7 [54.1 ± 66.0] 50.5 [45.4 ± 55.7] 70.9 [66.8 ± 75.1] 0.62 [0.58 ± 0.66] St Lucia [51.1 ± 65.1] 59.7 [54.9 ± 64.9] 40.4 [34.6 ± 46.2] 75.2 [70.5 ± 79.9] 0.63 [0.59 ± 0.68] Barbados [56.7 ± 70.0] 62.8 [57.7 ± 68.3] 48.8 [42.4 ± 55.4] 75.0 [69.7 ± 80.3] 0.68 [0.63 ± 0.73] WC, waist circumference (cm) cut-off point where sensitivity equaled speci city. abdominal adiposity associated with diseases. Although the NIH expert panel 29 has recommended the WC cut-off points at action level II for determining increased risks of CVD, the utility of these cut-off points for epidemiological identi cation of risk of CVD has been little studied. It is becoming increasingly apparent that WC values may be disease- and population-speci c. Indeed, Dowling and Pi-Sunyer observed upper body adiposity to be less strongly associated with risk factors of CVD and type 2 diabetes in Black women than White women from the US. 43 This is the rst study to investigate the predictive adequacy for hypertension of the action level abdominal obesity cut-off points recommended by the NHLBI=NIDDK expert panel using genetically similar populations from vastly different geographical regions. This is also the rst study to report speci c abdominal cut-off points, estimated with waist girth, that are associated with increased risk of hypertension in West African and Caribbean Blacks. Our analysis shows that in these West Africanorigin populations, WC cut-off point action level I had a low sensitivity for hypertension compared with site-speci c cut-off points. A WC cut-off point of 102 cm or more (action level II) was also found to have very low sensitivity for hypertension compared to site-speci c cut-off points. Our investigation further shows that WC cut-off points associated with hypertension in these nations vary with gender. The

6 Table 5 Sex, age and population-speci c waist circumference values associated with hypertension in West African and caribbean nations Age-group 25 ^ ^ ö Nigeria Cameroon Jamaica St Lucia Barbados Nigeria Cameroon Jamaica St Lucia Barbados Values are in cm. highest relative difference in WC cut-off point between men and women was observed in Nigeria with a value of 3.6 cm. Consistent with previous ndings, the higher WC cut-off points observed in men and women 55 y old at some sites are in line with the notion that high overall adiposity is not more associated with CVD in older than younger subjects. 1,44 In these populations and in West African men in particular, WC values were generally lower than those that have been described for Western populations. 10,21,45,46 The reason for this may due to the mean BMI values in these subjects, which were also somewhat low and may also have contributed to the predictive differences for hypertension. Indeed, the prevalences of hypertension varied enormously between these populations, and are related to BMI and WC. Four caveats of this investigation are noteworthy. First, due to the high multicollinearity between WC, BMI and HIP, no statistical adjustments were made for BMI and HIP. Second, no direct validation of abdominal adiposity estimated with WC was carried out. Third, unlike the study by Lean et al 21 that employed measurement of WC using bony landmarks mid-point between the iliac crest and the lowest ribs, this analysis also included subjects whose WC were measured at the narrowest part of the torso as seen from the front. It is unlikely however, that the different values of abdominal adiposity cut-off points identi ed in these populations compared to the established cut-off points are as a result of the different methods used for the measurement of WC. Fourth, the WC cut-off points reported in this study represent the cut-off points that provided the greatest statistical discrimination in terms of sensitivity and speci city for the prediction of hypertension. They do not necessarily represent the best cut-off points for health promotion. The WC cut-off points for Nigeria, for example, are at a level where most of the individuals would not be overweight, let alone obese and therefore these gures could not sensibly be applied for health promotion. Conclusion In summary, the waist cut-off points from this study represent values for epidemiological identi cation of risk for hypertension. For the purpose of health promotion, the decision on what cut-off points to use must be made by considering other additional factors including overall impact on health due to intervention (e.g. weight reduction) and potential burden on health services if a low cut-off point is employed. The ndings of this investigation provide evidence of differential abdominal adiposity cut-off points associated with risks of CVD across populations with different levels of technological development and environmental exposures. For the purpose of screening for CVD risk factors, we recommend that groups with similar levels of technological development, especially those populations whose distribution of obesity and associated risk factors tend to be very different from those of the technologically advanced, nations develop waist circumference cut-off points for CVD risks. Acknowledgement This work was supported in part by a grant from the National Heart, Lung, and Blood Institute (HL 45508). References 1 Han TS, van Leer EM, Seidell JC, Lean ME. Waist circumference as a screening tool for cardiovascular risk factors: Evaluation of receiver operating characteristics (ROC). Obes Res 1996; 6: 533 ± Han TS, Bijnen FC, Lean ME, Seidell JC. Separate associations of waist and hip circumference with lifestyle factors. Int J Epidemiol 1998; 3: 422 ± Bouchard C, Bray GA, Hubbard VS. Basic and clinical aspect of regional fat distribution. Am J Clin Nutr 1990; 52: 946 ± Han TS, Kelly LE, Walsh K, Greene RM, Lean ME. Relationship between volumes and areas from single transverse scans of intra-abdominal fat measured by magnetic resonance imaging. Int J Obes Relat Metab Disord 1997; l2: 1161 ± Vague J. The degree of masculine differentiation of obesities: a factor determining predisposition to diabetes, atherosclerosis, gout, and uric calculous disease. J Clin Nutr 1956; 4: 20 ± Cox BD, Whichelow MJ, Prevost AT. The development of cardiovascular disease in relation to anthropometric indices and hypertension in British adults. Int J Obes Relat Metab Disord 1998; 22: 966 ± Okosun IS, Cooper RS, Rotimi CN, Osotimehin B, Forrester T. Association of waist circumference with risk of hypertension and type 2 diabetes in Nigerians, Jamaicans, and African- Americans. Diabetes Care 1998; 21: 1836 ±

7 186 8 Lurbe B, Alvarez V, Liao Y, Tacons J, Cooper R, Cremades B, Torro I, Redon J. The impact of obesity and body fat distribution on ambulatory blood pressure in children and adolescents. Am J Hypertens 1998; 11(4 Pt 1): 418 ± Spangler JG, Bell RA, Summerson JH, Konen JC. Hyperinsulinemia in hypertension: Associations with race, abdominal obesity, and hyperlipidemia. Arch Fam Med 1998; 7: 53 ± Han TS, van Leer EM, Seidell JC, Lean ME. Waist circumference action levels in the identi cation of cardiovascular risk factors: prevalence study in a random sample. Br Med J 1995; 311: 1401 ± Han TS, Feskens EJ, Lean ME, Seidell JC. Associations of body composition with type 2 diabetes mellitus. Diabetes Med 1998; 15: 129 ± Seidell JC, Han TS, Feskens EJ, Lean ME. Narrow hips and broad waist circumferences independently contribute to increased risk of non-insulin-dependent diabetes mellitus. J lntern Med 1997; 242 : 401 ± Folsom AR, Prineas RJ, Kaye SA, Munger R. Incidence of hypertension and stroke in relation to body fat distribution and other risk factors in older women. Stroke 1990; 21: 701 ± Larsson B. Regional obesity as a health hazard in men ± prospective studies. Acta Med Scand 1998; 723 (Suppl) 45 ± Rexrode KM, Carey VJ, Hennekens CH, Walters EE, Colditz GA, Stampfer MJ, Willett WC, Manson JE. Abdominal adiposity and coronary heart disease in women. JAMA 1998; 280: 1843 ± Prineas RJ, Folsom AR, Kaye SA. Central adiposity and increased risk of coronary artery disease mortality in older women. Ann Epidemiol 1993, 3: 35 ± Ko GT, Chan JC, Woo J, Lau E, Yeung VT, Chow CC, Wai HP, Li JK, So WY, Cockram S. Simple anthropometric indexes and cardiovascular risk factors in Chinese. Int J Obes Relat Metab Disord 1997; 1l: 995 ± Kissebah AH, Vydelingum N, Murray R, Evans DJ, Hartz AJ, Kalkhoff RK. Relation of body fat distribution to metabolic complications of obesity. J Clin Endocrinol Metab 1982; 54: 254 ± Han TS, Tijhuis MA, Lean ME, Seidell JC. Quality of life in relation to overweight and body fat distribution. Am J Public Health 1998; 88: 1814 ± Han TS, Lean ME. Self-reported waist circumference compared with the `Waist Watcher' tape-measure to identify individuals at increased health risk through intra-abdominal fat accumulation. Br J Nutr 1998; 80: 81 ± Lean ME, Han TS, Morrison CE. Waist circumference as a measure for indicating need for weight management. Br Med J 1995; 311: 158 ± Kronmal RA. Spurious correlation and the fallacy of the ratio standard revisited. JR Statist Socl A 1993; 156: 379 ± Allison DB, Paultre F, Goran MI, Poehlman ET, Heyms eld SB. Statistical considerations regarding the use of ratios to adjust data. Int J Obes Relat Metab Disord 1995; 19: 644 ± Taylor RW, Keil D, Gold EJ, Williams SM, Goulding A. Body mass index, waist girth, and waist-to-hip ratio as indexes of total and regional adiposity in women: Evaluation using receiver operating characteristic curves. Am J Clin Nutr 1998; 67: 44 ± Lemieux S, Prud'homme D, Bouchard C, Tremblay A, Despres JP. A single threshold value of waist girth identi es normal-weight and overweight subjects with excess visceral adipose tissue. Am J Clin Nutr 1996; 5: 685 ± Pouliot MC, Despres JP, Lemieux S, Moorjani S, Bouchard C, Tremblay A, Nadeau A, Lupien PJ. Waist circumference and abdominal sagittal diameter: Best simple anthropometric indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol 1994; 7: 460 ± Lemieux S, Prud'homme D, Tremblay A, Bouchard C, Despres JP. Anthropometric correlates to changes in visceral adipose tissue over 7 years in women. Int J Obes Relat Metab Disord 1996; 7: 618 ± Haffner SM, Valdez RA, Hazed HP, Mitchell BD, Morales PA, Stem NIP. Prospective analysis of the insulin-resistance syndrome (syndrome X). Diabetes 1992; 41: 715 ± Gus M, Moreira LB, Pimentel M, Gleisener AL, Moraes RS, Fuchs FD. Association of various measurements of obesity and the prevalence of hypertension. Arq Bras Cardiol 1998; 2: 111 ± Reaven GM. Banting lecture 1988: Role of insulin resistance in human disease. Diabetes 1988; 37: 1595 ± National Institute of Health=National Heart Lungs and Blood. Clinical guidelines on the identi cation, evaluation, and treatment of overweight and obesity in adults. The evidence report. Obes Res 1998; 6: 51S ± 209S. 32 Molarius A, Seidell JC. Selection of anthropometric indicators for classi cation of abdominal fatness Ð a critical review. Int J Obes Relat Metab Disord 1998; 22: 719 ± World Health Organization. Report of a WHO consultation on obesity. Preventing and managing the global epidemic. WHO: Geneva, Folsom AR, Stevens J, Schreiner PJ, McGovern PG. Body mass index, waist=hip ratio, and coronary heart disease incidence in African Americans and whites. Atherosclerosis Risk in Communities Study Investigators. Am J Epidemiol 1998; : 1187 ± Haffner SM, Mitchell BD, Stern MP, Hazuda HP, Patterson JK. Public health signi cance of upper body adiposity for noninsulin dependent diabetes mellitus in Mexican Americans. Int J Obes Relat Metab Disord 1992; 16: l77 ± Cooper RS, Rotimi CN, Kaufhian JS, Owoaje EE, Fraser H, Forrester T, Wilks R, Riste LK, Cruickshank JK. Prevalence of NIDDM among populations of the African diaspora. Diabetes Care 1997; 20: 343 ± Cooper R, Rotimi C, Ataman S, McGee D, Osotimehin B, Kadiri S, Muna W, Kingue S, Fraser H, Forrester T, Bennett F, Wilks R. The prevalence of hypertension in seven populations of African origin. Am J Public Health 1997; 87: 160 ± Kaufman JS, Owoaje EE, James SA, Rotimi CN, Cooper RS. Determinants of hypertension in West Africa: Contribution of anthropometric and dietary factors to urban ± rural and socioeconomic gradients. Am J Epidemiol 1996; 143: 1203 ± Joint National Committee on Detection, Evaluation and Treatment of High Blood Pressure. Arch Intern Med 1988; 148: 1023 ± Norusis MJ. SPSS 8.0 for Windows, Release 8.0. SPSS Inc.: Chicago, IL, SIMSTAT for Windows, Version Provalis Research Corp: Montreal, Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristics (ROC) curve. Radiology 1982; 143: 29 ± Dowling HJ, Pi-Sunyer FX. Race-dependent health risks of upper body obesity. Diabetes 1993; 42: 537 ± Tate RB, Manfreda J, Cuddy TE. The effect of age on risk factors for ischemic art disease: The Manitoba Follow-Up Study, 194 ± Ann Epidemiol 1998; 8: 415 ± Molarius A, Seidell JC, Sans S, Tuomilehto J, Kuulasmaa K. Waist and hip circumferences, and waist-hip ratio in 19 populations of the WHO MONICA Project. Obes Relat Metab Disord 1999; 23: 116 ± Moreno LA, Fleta J, Mur L Rodriquez G, Sarria A, Bueno M. Waist circumference values in Spanish children Ð gender related differences. Eur J Clin Nutr 1999; 53 : 429 ± 433.