ORIGINAL COMMUNICATION

Transcription

1 (2004) 58, & 2004 Nature Publishing Group All rights reserved /04 $ ORIGINAL COMMUNICATION Waist circumference as a measurement of obesity in the Netherlands Antilles; associations with hypertension and diabetes mellitus L Grievink 1,2, JF Alberts 2 *, J O Niel 2 and I Gerstenbluth 2 1 Northern Centre for Health Care Research, University of Groningen, the Netherlands; and 2 Epidemiology and Research Unit, Medical and Public Health Service of Curaçao, Netherlands Antilles Objectives: To evaluate waist circumference (WC) as a screening tool for obesity in a Caribbean population. To identify risk groups with a high prevalence of (central) obesity in a Caribbean population, and to evaluate associations between (central) obesity and self-reported hypertension and diabetes mellitus. Design: Cross-sectional. Setting: Population-based study. Subjects: A random sample of adults (18 y or older) was selected from the Population Registries of three islands of the Netherlands Antilles. Response was over 80%. Complete data were available for 2025 subjects. Intervention: A questionnaire and measurements of weight, height, waist and hip. Main outcome measurement: Central obesity indicator (WC Z102 cm men, Z88 cm women). Results: WC was positively associated with age (65 74 y vs y) in men (OR ¼ 7.7, 95% CI ) and women (OR ¼ 6.4, 95% CI ). Women with a low education had a higher prevalence of central obesity than women with a high education (OR ¼ 0.5, 95% CI ). However, men with a high income had a higher prevalence of a central obesity than men with a low income (OR ¼ 1.7, 95% CI ¼ ). WC was the strongest independent obesity indicator associated with self-reported hypertension (OR ¼ 1.7, 95% CI ) and diabetes mellitus (OR ¼ 1.6, 95% CI ). Conclusions: The identified risk groups were women aged y, women with a low educational level and men with a high income. WC appears to be the major obesity indicator associated with hypertension and diabetes mellitus. Sponsorship: Island Governments of Saba, St Eustatius and Bonaire, the Federal Government of the Netherlands Antilles, Dutch Directorate for Kingdom relationships. (2004) 58, doi: /sj.ejcn Published online 31 March 2004 Keywords: BMI; waist-to-hip ratio; waist circumference; Caribbean Introduction In the Caribbean island of Curaçao (Netherlands Antilles), a very high prevalence of obesity (BMI Z30 kg/m 2 ) was observed, that is: 36% in women and 19% in men (Grol *Correspondence: JF Alberts, Epidemiology and Research Unit, Medical and Public Health Service (GGD), Piscaderaweg 49, Curaçao, Netherlands Antilles. ggdepi.cur@attglobal.net Guarantor: JF Alberts. Contributors: L Grievink was involved in the study design and analysed the data. JF Alberts coordinated the study, J O Niel and I Gerstenbluth were involved in the study design. All authors contributed to the writing of the article. Received 29 April 2003; revised 10 November 2003; accepted 22 December 2003; published online 31 March 2004 et al, 1997). In addition, hypertension (20 30%) and diabetes mellitus (10%) were estimated to be highly prevalent in Curaçao (Gerstenbluth et al, 1995; Alberts et al, 1996). From a number of epidemiological studies it is clear that obesity, and in particular central obesity, is related to an increased risk of these and other chronic diseases (Seidell et al, 1997; Okosun et al, 2001; Visscher & Seidell, 2001). Some reports have suggested that waist circumference (WC) might be a better predictor of cardiovascular disease risk and diabetes mellitus (NIDDM) than waist-to-hip (WHR) ratio (Han et al, 1995; Wei et al, 1997). WC was introduced as a simple measurement of (central) obesity (Lean et al, 1995). So far, the applicability of the cut-offs of WC have mainly been demonstrated among Caucasian populations (Molarius & Seidell, 1998). Grol et al (1997) were the first to explore

2 1160 WC cut-off points in an ethnically mixed population (predominantly of west African/European descent), through the Curaçao Health Study. During 1999 and 2000 data were collected on other islands of the Netherlands Antilles (Bonaire, St Eustatius and Saba). One objective was to identify risk groups with a high prevalence of obesity. We also evaluated whether the associations between obesity indicators (WC, BMI and WHR ratio), and hypertension and diabetes mellitus hold true for the population in this study. Finally, we examined whether WC as indicator of central obesity was associated with these disorders independent of overall obesity (BMI). Methods and procedures Subjects Data on three islands of the Netherlands Antilles (Bonaire, St Eustatius, Saba) are included in this paper; the data were collected during 1999/2000. Geographically, Bonaire and Curaçao pertain to the Leeward islands close to Venezuela; Saba and St Eustatius pertain to the Windward islands. In Bonaire and St Eustatius, study samples were randomly drawn from the Population Registries and consisted of registered inhabitants of 18 y or older, not living in institutions. In Saba, we included all permanent adult inhabitants to obtain representative study results because of the small overall number of residents. The response rates for the three islands were above 80% (82.1% in Bonaire, 87.9% in St Eustatius and 80.6% in Saba) with the following number of participants: 1003 for Bonaire, 572 for St Eustatius, and 562 for Saba. On all islands, the age distribution of the study sample was representative of the age distribution of the entire population. In Bonaire and Saba, women were slightly over-represented in the study sample. Weight and height measurements were missing for 110 individuals and data on waist and hip circumferences were not available for 101 individuals. Overall 2025 persons (1019 men and 1006 women) with complete anthropometric measures were included in the analyses. Questionnaire The study design consisted of an interview survey with internationally validated instruments and some additional instruments that had been adapted to local culture (Alberts et al, 1996). The four main topics of the questionnaire were health status, use of health care, lifestyle and knowledge/ attitudes towards health and health care. The questionnaire included a list of 33 chronic conditions or diseases and the respondent was asked whether he/she had these conditions currently or had them in the 12 months preceding the interview. The list was an adapted version of the continuous household survey of the Statistics Netherlands (Mootz & van den Berg, 1989). For evaluating possible associations between obesity and obesity-related diseases in this Dutch Caribbean population, we selected hypertension and diabetes mellitus because these are known to be highly prevalent among the Caribbean population, in particular among individuals of African or Indian subcontinent origin (Gulliford, 1994). The following variables were used among others for defining risk groups: level of education, occupational prestige and net household income; these were chosen as indicators of socio-economic status (SES). The highest level of education completed was coded according to the International Standard Classifications of Education (ISCED), developed by UNESCO in Paris (1976). For the analyses, the scores were reduced to three categories of about equal size, defined as low, middle and high educational level. Respondents indicated their current or past occupation and, if applicable, that of their (deceased) spouses. The highest occupation of either the respondent or partner was used for calculation of occupational prestige. Occupations were classified according to the International Standard Classification of Occupations (ISCO-88). The ISCO-88 was coded into Treiman s International Prestige Scale (Ganzeboom et al, 1992). Finally, the scale was divided into three categories varying from low to high occupational prestige, each containing approximately equal percentages of respondents. Respondents indicated their net household income category out of 11 possibilities, which were reduced into three categories of about equal sizes, defined as low, intermediate and high income with the cut-off points at 825 USD and 1650 USD per month. Anthropometry In addition to the interview, some anthropometric measurements were taken. The interviewers were trained to measure weight and height according to the World Health Organisation (WHO, 1987) standards. The participants were asked beforehand to wear light clothing and to remove jewellery, shoes and socks. The scales used were calibrated by the Government Calibration Office, allowing for a margin of error of 0.5%. To determine the obese proportion of the study population, the body mass index (BMI) was calculated (weight [kg]/height [m] squared). Obesity was classified according to the WHO definition (WHO, 1998) as a BMI Z30 kg/m 2, and we will refer to this in the text as overall obesity. With the subject standing, hip circumference was measured with a flexible tape to the nearest 0.5 cm at the widest point of the hip area. WC was measured to the nearest 0.5 cm at the mid-point between the lower rib and the iliac crest at the end of a normal expiration. As a measure of both overall obesity and central obesity, we determined the optimal cut-off points for WC. These cut-off points were in agreement with internationally used cut-off points as determined by Lean et al (1995): 102 cm for men and 88 cm for women.

3 As an additional measure of central obesity, we calculated the WHR. For international comparison, we used the cut-off values proposed by Lean et al (1995), that is, a WHR equal to or greater than 0.95 for men and 0.80 for women. Statistical methods To test the appropriateness of using WC cut-off points as a screening method for obesity, we determined the sensitivity and specificity of these cut-off points. Sensitivity and specificity were calculated according to the method used by Molarius et al (1999). To identify possible risk groups in the population, we studied the associations between the three obesity indicators (BMI, WC and WHR) and some socio-demographic variables (age, sex, educational level, occupational prestige and net household income). To calculate the associations, we used logistic regression models; the estimate was given as an odds ratio (OR) with a 95% confidence interval (95% CI). The OR and 95% CI for the three SESindicators were adjusted for age. All analyses were performed separately by gender. Furthermore, using logistic regression we analysed associations between each of the obesity indicators and hypertension and diabetes mellitus. In these analyses, each obesity indicator was entered into the regression model as a continuous variable. Odds ratios are presented for an increase of one standard deviation in the indicators, so the magnitude of the odds ratios for each of the obesity indicators can be compared. The odds ratios were adjusted for sex, age, educational level (three levels), smoking cigarettes (yes/no), physical exercise (yes/no), alcohol consumption (yes/no) and island of residence. Finally, to evaluate whether the obesity indicators were independently associated with having hypertension and diabetes mellitus, we adjusted each obesity indicator for the other two. SPSS software (version 9.0) was used for all statistical analyses (SPSS Inc., 1999). Results The study sample consisted of about equal proportions of men (50.3%) and women (49.7%). Most of the respondents (about 71%) were born in the Netherlands Antilles or Aruba (formerly part of the Netherlands Antilles). The mean age was 44.4 y in general (men: 44.0, and women: 44.7 y) (Table 1). The mean BMI was higher in women, whereas the mean WC and WHR were higher in men. The prevalence of obesity (BMI Z30 kg/m 2 ) was about 34% in women and 22% in men. About 54 and 66% of the women, respectively, had a high WHR and a high WC. In men these were 24 and 25%, respectively (Table 1). The overall sensitivity of the cut-offs points of WC was 58.1% in men and 73.5% in women. The overall specificity was 95.6% in men and 95.4% in women. The highest correlation between obesity indicators was observed for BMI and WC, with a Pearson correlation coefficient of WC and WHR were second highest correlated (Pearson r ¼ 0.65) and the correlation coefficient between BMI and WHR was lowest (Pearson r ¼ 0.27). Table 2 presents the association of each of the obesity indicators (BMI, WC and WHR), with age and socioeconomic variables in men. A high BMI was associated with age; the prevalence of obesity started to increase at 25 y and decreased after 75 y and older. A similar age-related pattern was observed for WC and WHR (Table 2). Men with an intermediate or high income (OR and 95% CI were 1.5 and ; and 1.7 and , respectively) had a higher prevalence of central obesity defined by WC than men with a low income. WC was not associated with educational level or occupational prestige. BMI and WHR were not associated with either a high education, occupational prestige or income in men (Table 2). Table 3 presents the association between the three obesity indicators, age, and socio-economic status in women. Although BMI as obesity indicator was not clearly associated with age when evaluating the odds ratios (OR of one was included in the 95% CI), the prevalence of obesity (determined by BMI) increased from 29.7% in the age group of y to 40.8% in the age group of y and decreased 1161 Table 1 Mean age, BMI, waist circumference and waist-to-hip ratio and the prevalence estimates of the three obesity indicators, overall and by sex Men (N ¼ 1019) Women (N ¼ 1006) Overall (N ¼ 2025) Mean s.d. Mean s.d. Mean s.d. Mean age (y) Mean BMI (kg/m 2 ) Mean waist circumference (cm) Mean waist-to-hip ratio Prevalence of obesity (%) BMI Z30.0 kg/m % 33.8% 27.8% WC Z102 cm for men, Z88 cm for women 23.7% 54.1% 38.8% WHR Z0.95 for men, Z0.80 for women 24.6% 65.9% 45.1% s.d. ¼ standard deviation; WHR ¼ waist-to-hip ratio; WC ¼ waist circumference; BMI ¼ body mass index.

4 1162 Table 2 Prevalence (%) of obesity indicators by age and SES among men BMI Z30 kg/m 2 WC Z102 cm WHR Z0.95 Socio-demographic variables Categories Prevalence (%) OR (95% CI) Prevalence (%) OR (95% CI) Prevalence (%) OR (95% CI) Age groups (y) w w w ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 75 þ ( ) ( ) ( ) Educational level* Low w w w Intermediate ( ) ( ) ( ) High ( ) ( ) ( ) Occupational prestige* Low w w w Intermediate ( ) ( ) ( ) High ( ) ( ) ( ) Net household income* o825 USD w w w USD ( ) ( ) ( ) Z1650 USD ( ) ( ) ( ) *Odds ratios of SES were adjusted for age; w Reference group. BMI¼ body mass index; WC¼ waist circumference; OR ¼ odds ratio; 95% CI¼ 95% confidence interval. Table 3 Prevalence of obesity indicators by age and SES among women BMI Z30 kg/m 2 WC Z88 cm WHR Z0.80 Socio-demographic variables Categories Prevalence (%) OR (95% CI) Prevalence (%) OR (95% CI) Prevalence (%) OR (95% CI) Age groups (y) w w w ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 75þ ( ) ( ) ( ) Educational level* Low w w w Intermediate ( ) ( ) ( ) High ( ) ( ) ( ) Occupational prestige* Low w w w Intermediate ( ) ( ) ( ) High ( ) ( ) ( ) Net household income* o825 USD w w w USD ( ) ( ) ( ) Z1650 USD ( ) ( ) ( ) *Odds ratio of SES were adjusted for age; w Reference group. BMI¼ body mass index; WC¼ waist circumference; OR ¼ odds ratio; 95% CI¼ 95% confidence interval. after 75 y of age to 17.2%. The central obesity indicators WC and WHR showed the same age-related pattern as in men, be it that women in the age groups between 55 and 74 y had a particularly high prevalence of central obesity compared to women below the age of 34 y. Women with an intermediate or high education (OR and 95% CI were 0.7 and ; and 0.5 and , respectively) had a lower prevalence of central obesity (WC) than women with a low education. In addition, women with a high education had a lower prevalence of overall obesity and central obesity (WHR). Women with a higher occupational prestige had a lower prevalence of overall obesity (BMI Z30 kg/m 2 ) than women with a lower occupational prestige. The prevalence of central obesity (WC and WHR) tended to decrease with a higher occupational prestige and income compared to a low status but this was not statistically significant after adjustment for age. Overall, the prevalence of self-reported hypertension was 16.1% and that of diabetes mellitus 7.9%. In Table 4, we present the odds ratios (OR) of hypertension and diabetes

5 Table 4 Association between hypertension, diabetes mellitus and each of the continuous obesity indicators (body mass index, waist circumference, waist-to-hip ratio), presented as odds ratios (95% confidence intervals) for a s.d. difference of each obesity indicator 1163 Crude Multi*-adjusted Multi þ BMI adjusted Multi þ WC adjusted Multi þ WHR adjusted Hypertension BMI (s.d. ¼ 5.7 kg/m 2 ) 1.6 ( ) 1.6 ( ) 0.9 ( ) 1.4 ( ) WC (s.d. ¼ 13.5 cm) 2.0 ( ) 1.9 ( ) 2.0 ( ) 1.7 ( ) WHR (s.d. ¼ 0.08) 1.7 ( ) 1.8 ( ) 1.5 ( ) 1.2 ( ) Diabetes mellitus BMI (s.d. ¼ 5.7 kg/m 2 ) 1.6 ( ) 1.6 ( ) 1.0 ( ) 1.5 ( ) WC (s.d. ¼ 13.5 cm) 2.0 ( ) 1.9 ( ) 1.8 ( ) 1.6 ( ) WHR (s.d. ¼ 0.08) 1.9 ( ) 1.9 ( ) 1.7 ( ) 1.4 ( ) s.d. ¼ standard deviation. *Multi-adjusted: age, sex, education, smoking, use of alcohol, sports activities and island of domicile. mellitus for the obesity indicators as continuous variables. We analysed the unique contributions of the obesity indicators to evaluate whether they were independently associated with hypertension or diabetes mellitus. Since the associations (odds ratios) of hypertension and diabetes mellitus with the obesity indicators did not deviate between men and women after adjustment for possible confounding factors (data not shown), we did not present these associations separately for each sex. All three obesity indicators were positively associated with hypertension and diabetes mellitus after adjustment for all possible confounders. However, the positive association between BMI and hypertension did not remain after adjustment for WC, whereas the association between WC and hypertension did remain after adjustment for BMI. The association between WHR and hypertension decreased slightly but remained statistically significant after adjustment for either BMI or WC. The OR of WC for hypertension decreased slightly after adjustment for WHR but was higher than the OR of WHR for hypertension after adjustment for WC. A similar pattern was observed for the associations between the obesity indicators and diabetes mellitus (Table 4). Discussion The cut-off points of WC had very high specificity (495%) in this population, for both men and women. This indicates that very few subjects would be recommended weight management unnecessarily if this cut-off point was applied in this study population. The sensitivity was low for men (58%) and somewhat higher for women (74%). Molarius et al (1999) demonstrated that a high prevalence of obesity was related to a high sensitivity. Men in our study had a prevalence of obesity comparable to what was found in Australia and the Czech Republic; the sensitivity of the WC cut-offs was also comparable. The women in our study had a higher prevalence of obesity and accordingly the corresponding sensitivity was also higher. The low sensitivity observed among men in our study, reflects relatively large proportions of men who had a low BMI and a high waist-tohip ratio. These men would be missed if the cut-off for WC were to be adopted as screening method for necessary weight management. However, it is debatable whether subjects with only a high WHR would really need weight reduction and would benefit from it (Molarius et al, 1999). In the present study, the prevalence of obesity determined by BMI was high (22% among men and 34% among women). This high prevalence was similar to the prevalence observed in Curaçao in 1993/1994, where 19% of the men and 36% of the women were obese (Grol et al, 1997). The prevalence of obesity among women in this ethnically mixed population of which the majority are of west African descent was also similar to that observed among black women in the USA (37%). Men in the current study had a higher prevalence than black men in the USA (15%) (Flegal et al, 1998). In Jamaica, women had a similarly high prevalence of obesity (32%), but again men had a lower prevalence (7%) than the men in the current study (Wilks et al, 1998). It appears that compared to some other (predominantly) black populations the prevalence of obesity in the Netherlands Antilles is high, in particular, among men. In the present study, an age of 25 y or older was associated with higher prevalence of overall and central obesity among men and with central obesity (WC and WHR) among women. Among women, the prevalence of central obesity further increased after the age of 55 y. The prevalence of all obesity indicators was highest in the y age group among both men and women, while overall obesity (as determined by BMI) decreased again after the age of 75 y. Similar results were observed in the Curaçao Health Study, except that among men the decrease of mean BMI, WHR and WC already set in after the age of 65 y (Grol et al, 1997). Studies among westernised populations show that the prevalence of both overall and central obesity increases with age to about y and then declines (Lackland et al, 1992;

6 1164 Seidell & Visscher, 2000). Compared to Western populations, men as well as women in these three islands of the Netherlands Antilles remain obese until an older age. It is alarming that young women (below the age of 25 y) in this study had a prevalence of 30% for overall obesity. We observed a higher prevalence of (central) obesity among women with a lower socio-economic status (SES) as was also found in Curaçao (Grol et al, 1997). Similar inverse associations between SES and obesitas have been demonstrated for women in most developed societies (Sobal & Stunkard, 1989; Gutierrez-Fisac et al, 1995; Visscher & Seidell, 2001). In our study, the association was most clear for educational level and not as clear for the other SESindicators occupational prestige and income. Higher educated women may have a more favourable attitude towards being slim. This hypothesis is supported by the fact that almost half of the women in the highest educational group participated in regular physical exercise vs 9% of the lower educational group (data not shown). For men, we found that those with a high income had higher prevalence estimates of central obesity (WC). A positive association between overall obesity and income was observed in Brazil (Monteiro, 2000). Positive associations between SES and obesity have been mostly observed in developing countries (Sobal & Stunkard, 1989). A possible explanation might be that a high central obesity (ie a big belly) among men with high prestige is generally accepted in society, thus showing their affluence. Further research is needed to identify possible factors, such as social and cultural perception and the acceptance of obesity, that explain the observed associations between socio-economic status and obesity among men and women. A better understanding of these underlying factors on these islands is needed to prevent or treat obesity. In the present study, WC was the strongest independent obesity indicator associated with self-reported hypertension and diabetes mellitus. WHR was also positively associated with these conditions but the strength of the association with hypertension decreased after adjustment for WC. BMI was not independently associated with hypertension and diabetes mellitus. A study by Han et al (1995) showed clear associations between WC and hypertension and other cardiovascular risk factors but did not adjust for the overall body fatness. WC was a better predictor of diabetes mellitus (NIDDM) compared to BMI and WHR in Mexican Americans in a prospective study (Wei et al, 1997). A high WC was associated with a higher prevalence of hypertension among populations of West African descent in Jamaica, Barbados and the United States but not in St Lucia. After adjustment for BMI, the association only remained significant among women (Okosun et al, 2000). In a large survey (NHANES III) the same author found that, independently of BMI, a high WC was associated with hypertension among black Americans (Okosun et al, 2001). The use of prevalence estimates for hypertension and diabetes mellitus based on self-report can introduce a bias. However, these two conditions showed good correlations when comparing self-report and doctors lists in the Curaçao Health Study (Gerstenbluth et al, 1996). The true prevalence is probably up to twice as high for diabetes mellitus and hypertension because a large proportion of these individuals have not (yet) been diagnosed with this disorder (Gerstenbluth et al, 1995). If differential misclassification may exist, that is, if obese individuals would be more likely to be screened for diabetes or hypertension because the GP or specialist is aware of the existing association with obesity, the true estimate of the association (odds ratios) would be slightly weaker. It is clear that from a practical point of view in large epidemiological settings, WC compared to BMI and WHR is the most simple measurement to use as obesity indicator. In large-scale health studies where limited resources are involved, the WC appears to reveal sufficient information about overall and central obesity in both this population and European populations (Lean et al, 1995). Traditionally, WHR was used as an indicator for abdominal obesity. Although this indicator reflects central fat distribution, it is not as strongly correlated to visceral fat accumulation among Caucasians and African Americans as is WC (Conway et al, 1997; Hill et al, 1999; Rankinen et al, 1999). As a ratio, waist to hip is difficult to interpret biologically because a change in body adiposity or weight does not necessarily result in a change of WHR. In addition, ratios have limitations in statistical modelling since their use can introduce spurious correlations between the ratios and other variables (Allison et al, 1995). WC might be used for screening the population for necessary weight management, but additional information is needed on the relation between WC and (total and visceral) body fat in the Caribbean population under study (Luke et al, 1997; Deurenberg et al, 1998). Also, further knowledge is needed about the distribution of risk factors and health outcomes in subjects who were classified in the false negative group, that is, those with a high WHR but normal to low BMI or WC. In conclusion, the risk groups identified in the population were women aged y, women with a low educational level and men with an intermediate or high income. Independent of BMI, a high WC was the strongest obesity indicator associated with a higher prevalence of hypertension and diabetes mellitus. Although further knowledge is needed before WC can be used as single screening method for weight management, weight management is definitely recommended in a population with such high prevalence estimates of obesity (WHO, 1998). Acknowledgements This study is a joint project of the Epidemiology & Research Unit of the Medical and Public Health Service of Curaçao, the Northern Centre for Health Care Research of the University of Groningen in the Netherlands, the Foundation

7 for Promotion of Research and International Cooperation in Health Care (ISOG) in Curaçao, the Island Governments of Saba, St Eustatius and Bonaire and the Federal Government of the Netherlands Antilles. The study was co-financed by the Dutch Directorate for Kingdom relationships. The authors wish to thank Gabi Fuchs, Wim van den Heuvel and Eric van Sonderen and for their valuable contributions. References Alberts JF, Gerstenbluth I, Halabi YT, Koopmans PC, O Niel J & Heuvel van den WJA (1996): The Curacao Health Study, Methodology and Main Results. pp 119. Assen: Van Gorcum. Allison DB, Paultre F, Goran MI, Poehlman ET & Heymsfield SB (1995): Statistical considerations regarding the use of ratios to adjust data. Int. J. Obes. Relat. Metab. Disord. 19, Conway JM, Chanetsa FF & Wang P (1997): Intraabdominal adipose tissue and anthropometric surrogates in African American women with upper- and lower-body obesity. Am. J. Clin. Nutr. 66, Deurenberg P, Yap M & van Staveren WA (1998): Body mass index and percent body fat: a meta analysis among different ethnic groups. Int. J. Obes. Relat. Metab. Disord. 22, Flegal KM, Carroll MD, Kuczmarski RJ & Johnson CL (1998): Overweight and obesity in the United States: prevalence and trends, Int. J. Obes. Relat. Metab. Disord. 22, Ganzeboom HBG, de Graaf PM & Treiman DJ (1992): A standard international socioeconomic index of occupational status. Soc. Sci. Res. 21, Gerstenbluth I, Alberts JF, te Velde B & Leerink CB (1995): De prevalentie van chronische aandoeningen op Curaçao (in Dutch), In Results of the Curacao Health Study and Policy Implications. pp Netherlands Antilles: ISOG 2000 Curacao. Gerstenbluth I, Alberts JF, Huirne JAF & Smits IMH (1996): Prevalentieschattingen van chronische aandoeningen op Curacao: zelfrapportage versus huisartsenregistratie (in Dutch). Tijdschrift voor Sociale Gezondheidszorg 74, Grol MEC, Eimers JM, Alberts JF, Bouter LM, Gerstenbluth I, Halabi Y, van Sonderen E & van den Heuvel WJA (1997): Alarmingly high prevalence of obesity in Curaçao: data from an interview survey stratified for socioeconomic status. Int. J. Obes. Relat. Metab. Disord. 21, Gulliford MC (1994): Health and health care in the English-speaking Caribbean: a British public health physician s view of the Caribbean. J. Pub. Health Med. 16, Gutierrez-Fisac JL, Regidor E & Rodriguez C (1995): Economic and social factors associated with body mass index and obesity in the Spanish population aged years. Eur. J. Pub. Health 5, Han TS, Leer EM van, Seidell JC & Lean MEJ (1995): Waist circumference action levels in the identification of cardiovascular risk factors: prevalence study in a random sample. Br. Med. J. 311, Hill JO, Sidney S, Lewis CE, Tolan K, Scherzinger AL & Stamm ER (1999): Racial differences in amounts of visceral adipose tissue in young adults: the CARDIA (Coronary Artery Risk Development in Young Adults) Study. Am. J. Clin. Nutr. 69, Lackland DT, Orchard TJ, Keil JE, Saunders Jr DE, Wheeler FC, Adams-Campbell LL, McDonald RH & Knapp RG (1992): Are race differences in the prevalence of hypertension explained by body mass and fat distribution? A survey in a biracial population. Int. J. Epidemiol. 21, Lean MEJ, Han TS & Morrison CE (1995): Waist circumference as a measure for indicating need for weight management. Br. Med. J. 311, Luke A, Durazo-Arvizu R, Rotimi C, Prewitt TE, Forrester T, Wilks R, Ogunbiyi OJ, Schoeller DA, McGee D & Cooper RS (1997): Relation between body mass index and body fat in black population samples from Nigeria, Jamaica, and the United States. Am. J. Epidemiol. 145, Molarius A & Seidell JC (1998): Selection of anthropometric indicators for classification of abdominal fatness a critical review. Int. J. Obes. Relat. Metab. Disord. 22, Molarius A, Seidell JC, Sans S, Tuomilehto J & Kuulasmaa K (1999): Varying sensitivity of waist action levels to identify subjects with overweight or obesity in 19 populations of the WHO MONICA Project. J. Clin. Epidemiol. 52, Monteiro C (2000): The epidemiological transition in Brazil, In WHO Scientific publications No. 576 (Obesity and Poverty: a New Public Health Challenge). pp Washington: PAHO. Mootz M & van den Berg J (1989): Indicatoren voor gezondheidstoestand in de CBS-gezondheidsenquête (in Dutch). Maandbericht Gezondheid (CBS) 89, Okosun IS, Liao Y, Rotimi CN, Choi S & Cooper RS (2000): Predictive values of waist circumference for dyslipidemia, type 2 diabetes and hypertension in overweight white, black, and Hispanic American adults. J. Clin. Epidemiol. 53, Okosun IS, Choi S, Dent MM, Jobin T & Dever GE (2001): Abdominal obesity defined as a larger than expected waist girth is associated with racial/ethnic differences in risk of hypertension. J. Hum. Hypertens. 15, Rankinen T, Kim S-Y, Pérusse L, Després J-P & Bouchard C (1999): The prediction of abdominal visceral fat level form body composition and anthropometry: ROC analysis. Int. J. Obes. Relat. Metab. Disord. 23, Seidell JC, Han TS, Feskens EJM & Lean MEJ (1997): Narrow hips and broad waist circumferences independently contribute to increased risk of non-insulin-dependent diabetes mellitus. J. Int. Med. 242, Seidell JC & Visscher TLS (2000): Body weight and weight change and their health implications for the elderly. Eur. J. Clin. Nutr. 54(Suppl 3), S33 S39. Sobal J & Stunkard AJ (1989): Socioeconomic status and obesity: a review of the literature. Psychol. Bull. 105, SPSS Inc (1999): Guide to Data Analyses. 9th edition. Chicago: SPSS. Inc. Visscher TLS & Seidell JC (2001): The public health impact of obesity. Annual Rev. Public Health 22, Wei M, Gaskill SP, Haffner SM & Stern MP (1997): Waist circumference as the best predictor of noninsulin dependent diabetes mellitus (NIDDM) compared to body mass index, waist/hip ratio and other anthropometric measurements in Mexican Americans, a 7-year prospective study. Obes. Res. 5, Wilks R, Bennett F, Forrester T & McFarlane-Anderson N (1998): Chronic diseases: the new epidemic. West Indian Med. J. 47(Suppl 4), S40 S44. World Health Organisation (WHO, 1987): Obesity classification and description of anthropometric data, In Report on WHO Consultation on the Epidemiology of Obesity. Warsaw: WHO. World Health Organisation (WHO, 1998): Obesity: Prevention and managing the global epidemic, In Report of a WHO Consultation on Obesity. WHO/NUT/NCD/98.1. Geneva: WHO. 1165