European Journal of Clinical Nutrition (2003) 57, 956–963 & 2003 Nature Publishing Group All rights reserved 0954-3007/03 $25.00 www.nature.com/ejcn

ORIGINAL COMMUNICATION Relation of adiposity and body fat distribution to body mass index in of Aboriginal and European ancestry

LS Piers1*, KG Rowley2, MJ Soares3 and K O’Dea1

1Menzies School of Health Research, Casuarina, NT, ; 2Department of Medicine, The University of Melbourne, St Vincent’s Hospital, Fitzroy, Vic., Australia; and 3Department of Nutrition, Dietetics & Food Science, Curtin University of Technology, School of Public Health, , WA, Australia

Objective: To compare the relations of adiposity and body fat distribution to body mass index (BMI) in Australians of Aboriginal and European ancestry. Design: Cross-sectional volunteer samples. Setting: Australian Aboriginal communities in remote central and northern Australia, urban European Australians resident in Melbourne, Australia. Subjects: Healthy Aboriginal (130 women, 120 men) and European Australians (100 women, 47 men) with a BMIo30 kg/m2, aged 18–35 y; all women were nonpregnant. Interventions: Anthropometric variables and resistanceFusing a four-terminal impedance plethysmographFwere measured. Results: Aboriginal women and men were significantly shorter and weighed less than European Australians (Po0.05). Aboriginal women had a significantly larger waist circumference and waist-to-hip ratio (WHR, Po0.0005) compared to European Australian women. The sum of four skinfold thicknesses (SFT) (S4) and trunk SFT was higher in Aboriginals as compared to European Australian women (Po0.0005); however, limb SFT tended to be lower (P ¼ 0.06). On the other hand, BMI was significantly lower in Aboriginals compared to European Australian men (P ¼ 0.011), as was hip circumference (P ¼ 0.001); however, WHR was significantly (P ¼ 0.007) higher. On regression analysis, Aboriginal women and men were significantly heavier than European Australians for the same height2/resistance (surrogate for fat-free mass) and S4 (surrogate for subcutaneous fat); and that Aboriginal men had a significantly higher BMI (by 1.2 kg/m2; Po0.0005) for any given S4 and height2/resistance values, compared to European Australian men. Conclusion: Aboriginal and European Australians have a significantly different body fat distribution and fat mass for a given body weight or BMI. Use of the World Health Organization recommended BMI ranges to determine weight status may be inappropriate in Australian Aboriginal people. Sponsorship: This study was supported, in part, by the National Health and Medical Research Council (NHMRC) of Australia grants (Nos. 934502, 954605). LSP was supported by an NHMRC grant (No. 981019). KGR was supported by an NH&MRC Postdoctoral Research Fellowship (No. 974302) and is currently a VicHealth Public Health Research Fellow. European Journal of Clinical Nutrition (2003) 57, 956–963. doi:10.1038/sj.ejcn.1601630

Keywords: aboriginal people; anthropometry; body composition; body mass index; bioelectrical impedance analysis; skinfold thickness

*Correspondence: LS Piers, Division of Population Health and Chronic Disease, Menzies School of Health Research, PO Box 41096, Casuarina NT 0811, Australia. E-mail: [email protected], [email protected] Introduction Guarantor: LS Piers. Contributors: KGR and KOD conceived the study. LSP and KGR A World Health Organization (WHO) sponsored expert coordinated the study and undertook the analyses, supervised by K consultation on obesity recently advocated the use of the O’Dea. Measurements were carried out by LSP, KGR and MJS. LSP body mass index (BMI) to determine the weight status of wrote the paper and all authors were involved in interpreting the individuals (WHO, 1997). It was suggested that this would results and critical revision of the paper. Received 18 July 2002; revised 26 August 2002; permit meaningful comparisons of weight status within and accepted 2 September 2002 between populations, the identification of individuals and Body composition of Australian Aboriginal people LS Piers et al 957 groups at increased risk of morbidity and mortality, identi- 1997). There is also evidence that WHR is a better single fication of priorities for intervention at the individual or screening measure for Type II diabetes risk than is BMI community level and a firm basis for evaluating interven- (Haffner et al, 1992) and that waist circumference identifies tions (WHO, 1997). BMI and the waist/hip circumference normal weight and overweight individuals with excess ratio (WHR) are now used conventionally as indices of visceral adipose tissue that increases their risk of Type II obesity and fat distribution in epidemiological studies. diabetes (Lemieux et al, 1996; Lean et al, 1998). Body fat Although some general limitations of these indices are distribution may also vary between ethnic groups. (Baum- recognised, others that affect their use in relative risks for gartner et al, 1995; Gasperino, 1996; Lovejoy et al, 2001). We disease are not well recognised. These include effects of sex, were, therefore, interested in determining if body weight, ethnicity, and especially age on the relations between these BMI and body fat distribution were related in the same indices and body composition, which can result in sub- manner to body composition in young (18–35 y of age), stantial misclassification of obesity and fat distribution healthy (nondiabetic) Australians of Aboriginal and Eur- (Baumgartner et al, 1995). opean ancestry. Cutoff points for obesity as defined by the WHO are based on BMI-values, but these cutoff points are based on studies on the relation between BMI and morbidity and mortality in Participants Western populations (Lew & Garfinkel, 1979; WHO, 1990). It Aboriginal people were recruited from among residents of may be questionable whether these cutoff points are remote communities in central and north-eastern Australia universally valid to identify those at increased risk of chronic in the course of community-based risk factor screening disease. We have recently shown that BMI has poor initiatives (Rowley et al, 2000). The residents were mainly of sensitivity and positive predictive value in identifying over- Aboriginal descent, the remoteness of the communities weight/obese individuals when classified by body fat percen- limiting admixture with other groups. All fulfilled the tage (BF%), measured using deuterium dilution, in definition of ‘Aboriginal’ defined by Australian law, that is, Australians of European ancestry (Piers et al, 2000). In recent ‘(1) is of Aboriginaly. descent; (2) identifies as an Australian years, several studies have shown a different relation Aboriginal (person)y; and (3) is accepted as such by the between BMI and BF% among ethnic groups. For example, community in which he or she lives or has lived’. The Wang et al (1994) in a study in New York found that Asians European Australians, of mainly Anglo-Celtic descent, were had a lower mean BMI, but a higher BF% than Caucasians of recruited from among the staff and students of Deakin the same age and sex. Gurrici et al (1998) reported that University and by advertisement in the local media in the Indonesians had, for the same BF%, a BMI about 3 units general surroundings of the Toorak campus of Deakin lower than Dutch Caucasians. In a meta-analysis of available University, in Melbourne, Australia, as part of other studies data from the literature, it was shown that differences in the (Diffey et al, 1997; Piers et al, 1997, 1998, 2000; Soares et al, relations of BMI to BF% exist among ethnic groups 1998). Only nondiabetic men and women, between 18 and (Deurenberg et al, 1998). 35 y of age, who were nonpregnant, and with a BMIo30 kg/ Australian Aboriginal people have extremely high and m2 were included in the analysis. Written informed consent increasing prevalence of obesity, based on BMI, and its was obtained from all participants and the study was attendant health problems (Cunningham & Mackerras, approved by the Deakin University Ethics Committee and 1998). Among adults aged 18 y or more, about 25% of The Alice Springs Institutional Ethics Committee. Aboriginal men and 28% of Aboriginal women may be classified as obese based on the BMI classification of weight status (McLennan & Madden, 1999). Rutishauser and McKay observed that for a given sum of skinfold thickness’, Methods Aboriginal women from north- had a BMI Anthropometry 2 units lower than age-matched Caucasians (Rutishauser & Standing height was measured using a portable stadiometer McKay, 1986) implying a greater proportion of body fat. This (Holtain, Crymych, UK) in the Aboriginal people and a suggests that the currently advocated classification of weight stadiometer fixed to the laboratory wall in the European status, based on BMI, may be inappropriate for use in Australian subjects (SECA, model 708, Germany) and Aboriginal people. Use of this classification may result in recorded to the nearest 0.1 cm. The portable stadiometer individuals who have similar amounts of body fat when was checked for accuracy against a metal measuring tape compared to non-, who are over- measure prior to each field trip. Body weight was measured, weight or obese, being classified as normal weight because with subjects wearing light indoor clothing and without of their ‘normal’ BMI. Furthermore, the anatomical distribu- shoes, immediately after voiding, using a digital weighing tion of body fat is also of importance in defining risk of ill scale (UC-300, AND, Tokyo, Japan), and recorded to the health. There is evidence that BMI, WHR and waist nearest 100 g. The weighing scale was checked for accuracy circumference are powerful independent predictors of Type using standard weights. Regional fat distribution was II diabetes (Chan et al, 1994; Carey et al, 1997; Ledoux et al, estimated from the ratio of waist-to-hip circumferences.

European Journal of Clinical Nutrition Body composition of Australian Aboriginal people LS Piers et al 958 The waist was measured at the smallest abdominal circum- one electrode was in line with the proximal edge of the ulnar ference. The hip measurement was made at the level of the tubercle at the wrist and the proximal edge of the other was in greater trochanter and the most prominent point over the line with the medial malleolus of the ankle (Lukaski, 1987). buttocks, the subject standing erect with their feet together. Each participant lay still and supine with arms by their sides Circumference measurements were all made to the nearest and all four limbs abducted. An excitation current of 800 mAat millimetre, without compressing the skin and underlying 50 kHz was introduced at the distal electrodes of the hand and tissues, using a metal measuring tape (Callaway et al, 1988). foot, and the voltage drop was detected by the proximal electrodes. The lowest resistance values were recorded. Measurements were always performed immediately after Skinfold thickness voiding. Participants removed all jewellery and metallic Skinfold thickness (SFT) was measured, using Holtain objects from their person. An alcohol swab was used to clean skinfold callipers (Crymych, UK) calibrated to exert a the skin before electrode placement. The plethysmograph was constant pressure of 10 g/mm2. SFT at four anatomical sites checked using a 500 O resistor, supplied by the manufacturer, (biceps, triceps, subscapular, and suprailiac;) were measured every morning prior to making any measurements. on the right side of the body and recorded to the nearest 0.2 mm (Harrison et al, 1988). The thumb and index finger of the left hand elevated a double fold of skin and subcuta- Statistical analysis neous adipose tissue about 1 cm proximal to the site of All data were analysed using SPSS version 10.0 (SPSS Inc, IL, measurement to ensure that the fingers did not compress the USA). Statistical significance was accepted at Po0.05. All measurement site. Each skinfold was measured in triplicate variables were tested for normality using the one-sample and the mean of the three measurements was used in further Kolmogorov–SmirnovFgoodness-of-fit test. All variables analyses. The sum of the four SFT (S4) was used as a surrogate with distributions significantly different (Po0.05) from the

measure of total body FM, while the sum of the biceps and normal distribution were log10 transformed and retested. triceps SFT (limb SFT) was used as a surrogate measure of Between-group comparisons were made using independent t peripheral FM. The sum of subscapular and suprailiac SFT tests. The contribution of SFT, bioelectrical impedance, (trunk SFT) was used as a surrogate measure of central FM. height and ethnicity to body weight and BMI was assessed using multiple linear regression. Slopes and intercepts of regression lines were compared as described by Kleinbaum Resistance measurements et al (1988). This approach uses a single multiple regression Measurement of resistance was made using a four-terminal model that contains a (one or more) dummy variable(s) to impedance plethysmograph (RJL Systems, model 101, De- distinguish between groups being compared (in this study troit, USA). Disposable electrodes (Nikotabs-E, Medical 0 ¼ European Australian, 1 ¼ Aboriginal people). Equipment Services, Australia) cut in half were used, as instructed by the manufacturer. Each half was positioned in the midline of the dorsal surfaces of the hands and feet, Results proximal to the metacarpal–phalangeal and metatarsal– A total of 397 participants were included in this analysis; phalangeal joints, between the distal prominences of the these were 250 Aboriginal people (120 and 130 men) and 147 radius and ulna and between the medial and lateral malleoli European Australians (100 women and 47 men). Participant at the ankle, respectively. Specifically, the proximal edge of characteristics are presented in Table 1. Aboriginal people

Table 1 Participant characteristics

Women Men

European Australians (n=100) Aboriginal Australians (n=120) European Australians (n=47) Aboriginal Australians (n=130)

Mean s.d. Mean s.d. Mean s.d. Mean s.d.

Age (y) 24.5 3.5 25.3 4.7 25.8 4.2 24.8 5.3 Height (cm) 165.8 6.9 160.2* 5.5 178.8 7.2 172.1* 6.3 Weight (kg) 60.2 6.9 57.3* 10.8 76.3 10.5 66.8* 11.1 BMI (kg/m2) 21.9 2.2 22.3 4.1 23.8 2.6 22.5* 3.2 Waist (cm) 68.5 5.0 77.7* 11.3 80.7 8.2 79.7 9.2 Hip (cm) 94.0 10.7 92.7 8.5 97.0 5.9 93.2* 6.6 WHR 0.72 0.04 0.84* 0.07 0.83 0.05 0.85* 0.06

*Significantly different (Po0.05) from European Australians of same sex.

European Journal of Clinical Nutrition Body composition of Australian Aboriginal people LS Piers et al 959 Table 2 SFT and resistance data

Women Men

European Australians (n=100) Aboriginal Australians (n=120) European Australians (n=47) Aboriginal Australians (n=130)

Geometric mean 95% CI Geometric mean 95% CI Geometric mean 95% CI Geometric mean 95% CI

SFT data: Sum of four SFT (mm) 54.5 51.3, 57.9 68.7* 63.5, 74.3 42.0 37.4, 47.1 42.2 39.7, 46.0 Limb SFT (mm) 25.8 24.3, 27.5 23.6** 21.9, 25.3 13.7 12.2, 15.5 13.5 12.6, 14.5 Trunk SFT (mm) 28.3 26.4, 30.2 44.6* 41.0, 48.7 27.9 24.6, 31.5 28.9 26.6, 31.4

Mean s.d. Mean s.d. Mean s.d. Mean s.d. Bioelectrical impedance data: Resistance (O) 616.1 58.0 720.1* 98.1 477.0 44.9 572.1* 71.5 Height2/resistance (cm2/O) 45.1 5.4 36.3* 5.4 67.6 7.5 52.7* 8.4

*Po0.0005 significantly different from European Australians of same sex. **Tends toward being significantly different (P=0.06) from European Australians of same sex.

2 Table 3 Regression analyses of body weight and BMI on surrogate measures of body composition (log10 S4 and height /resistance), height and ethnicity

Co-efficients for

2 * 2 Dependent variable (y) log10 S4 (x1) Height /Resistance (x2) Height (x3) Ethnicity (x4) Interactions Constant (c) R

w w z z y y w Women Weight 32.22 0.79 19.10 2.0 x1 Â x4:NS, x2 Â x4:NS À63.2 0.80 z y BMI 11.52 0.14 X À7.7 x1 Â x4:4.5, x2 Â x4:NS À4.4 0.70

w w w w y y w Men Weight 30.01 0.73 30.13 3.8 x1 Â x4:NS, x2 Â x4:NS À75.9 0.90 w w w y y w BMI 10.84 0.16 X 1.2 x1 Â x4:NS, x2 Â x4:NS À4.8 0.78

XFvariable not included in model. Import of coefficients in italics could not be interpreted because of significant interaction between independent variables. *0=European Australians, 1=Aboriginal Australians. w Po0.0005, zPr0.01 coefficient significantly different from zero. y interactions nonsignificant, therefore, excluded from model. were significantly shorter and weighed less than their had significantly higher (Po0.0005) resistance values but European Australian counterparts. BMI was similar in women significantly lower (Po0.0005) height2/resistance ratios (P ¼ 0.285); however, Aboriginal women had significantly when compared to their European Australian counterparts. greater waist circumference and WHR (Po0.0005), but not Table 3 shows results of regression analyses of body weight 2 hip circumference (P ¼ 0.327), compared to European Aus- and BMI as functions of log10 S4, height /resistance, tralian women. BMI was significantly lower in Aboriginal men height and ethnicity. Relations are illustrated graphically in as compared to European Australian men (P ¼ 0.011). Waist Figures 1 and 2. On modelling body weight in women as a 2 circumference was not significantly different (P ¼ 0.614), function of ethnicity, log10 S4, height /resistance, height and however, hip circumference was significantly lower their interactions, there was no significant interaction (P ¼ 0.001) and WHR was significantly (P ¼ 0.007) higher in between ethnicity and any of the other independent Aboriginal men as compared to European Australian men. variables (P40.05). However, the coefficient for ethnicity SFT and resistance data are presented in Table 2. The sum was significant (P ¼ 0.01), with Aboriginal women being 2 of the four SFT (S4) was higher in Aboriginal women as 2.0 kg heavier, for the same height, log10 S4, height / compared to European Australian women (Po0.0005). resistance values, than their European Australian counter- While their trunk SFT was significantly higher (Po0.0005), parts. On modelling body weight in men as a function of 2 the limb SFT of Aboriginal women tended to be lower than ethnicity, log10 S4, height /resistance, height and their that of European Australian women (P ¼ 0.06) (Table 2). interaction terms, there was no significant interaction There was no significant difference in S4 between Aboriginal between ethnicity and any of the other independent and European Australian men (P ¼ 0.485), nor were there variables (P40.05). However, the coefficient for ethnicity significant differences between limb and trunk SFT between was significant (Po0.00005), with Aboriginal men being 2 the two groups of men. Both Aboriginal men and women 3.8 kg heavier, for the same height, log10 S4, height /

European Journal of Clinical Nutrition Body composition of Australian Aboriginal people LS Piers et al 960

Figure 1 Relationship of body weight to surrogate measures of fat mass (sum of four SFT) and fat-free mass (height2/resistance) in Australians of Aboriginal (filled squares, solid line) and European (open circles, broken line) ancestry.

resistance values, than their European Australian counter- quantified. Aboriginal men had WHR that was 0.036 units parts (Table 3, Figure 1). higher for any given BMI compared to European Australian On modelling BMI in women as a function of ethnicity, men (Po0.0005). 2 log10 S4, height /resistance, and their interactions, there was a significant interaction between log10 S4 and ethnicity (Po0.01). Because of this interaction the coefficient of the Discussion ethnicity variable could not be interpreted, except to say that Our results indicated that this group of Aboriginal women 2 the relation of log10 S4 and height /resistance to BMI was was shorter and proportionately less heavy than European different in the two ethnic groups (Table 3, Figure 2). On Australian women and consequently had a similar mean modelling BMI in men as a function of ethnicity, log10 S4, BMI. However, while Aboriginal men were also shorter, they height2/resistance and their interactions, the interaction weighed considerably less than European Australian men terms were not significant (P40.05). However, the coeffi- and therefore had a significantly lower mean BMI. Never- cient for ethnicity was significant (Po0.05). Hence, for any theless, for any given combination of S4 (subcutaneous fat) given combination of S4 and height2/resitance Aboriginal and height2/resistance value (fat-free mass), Aboriginal men men had a BMI that was B1.2 kg/m2 higher than European had a BMI that was approximately 1.2 kg/m2 greater than Australian men (Table 3, Figure 2). As the analysis was European Australian men. This would indicate that for any restricted to those who were 18–35 y of age, ‘age’ was not given BMI, the ratio of FM to FFM was greater in Aboriginal included in any of the analyses. men. The significant interaction of ethnicity and body Figure 3 shows the relation pf WHR to BMI in Aboriginal composition variables in women indicated that the relation and European Australian men and women. On modelling of these surrogate measures of body composition to BMI was WHR as a function of BMI and ethnicity in women, significantly different in the two ethnic groups; however, the Aboriginal women had significantly greater WHR magnitude of the difference could not be assessed (Table 3). (Po0.0005), but as there was significant interaction between SFT measurements are said to provide an estimate of the ethnicity and BMI (P ¼ 0.01) this difference could not be size of the subcutaneous fat depot, which in turn provides an

European Journal of Clinical Nutrition Body composition of Australian Aboriginal people LS Piers et al 961

Figure 2 Relationship of BMI (kg/m2) to surrogate measures of fat mass (sum of four SFT) and fat-free mass (height2/resistance) in Australians of Aboriginal (filled squares, solid line) and European (open circles, broken line) ancestry.

Figure 3 Relationship of WHR to BMI (kg/m2) in Australians of Aboriginal (filled squares, solid line) and European (open circles, broken line) ancestry.

estimate of total body fat (Durnin & Rahaman, 1967). Such Unfortunately, variations in the distribution of subcutaneous estimates are based on two assumptions: (a) the thickness of fat occur with sex, ethnicity and age (Robson et al, 1971; the subcutaneous fat reflects a constant proportion of the Durnin & Womersley, 1974). In addition, it is still unclear if total fat and (b) the skinfold sites selected for measurements, ethnic-specific equations are required to estimate FFM and either singly or in combination, represent the average FM using bioelectrical impedance analysis (BIA) (Ellis et al, thickness of the entire subcutaneous tissue (Lukaski, 1987). 1999) as there is evidence of population specificity in the

European Journal of Clinical Nutrition Body composition of Australian Aboriginal people LS Piers et al 962 validity of BIA (Chertow et al, 1997). As such, we chose to assumed that the present observations arise from genetic compare directly measured variables (SFT and height2/ characteristics. resistance) rather than derived estimates of FM and FFM In conclusion, this study indicates that Aboriginal people using regression equations generated from data that did not have a very different body fat distribution, consistent with a include Aboriginal people. We also restricted the analysis to greater amount of abdominal fat as compared to their healthy young people aged between 18 and 35 y, analysed European Australian counterparts. In addition, the relation data from women and men separately and excluded of surrogate estimates of body composition, S4 for subcuta- pregnant women altogether. neous fat and height2/resistance for FFM, to body weight and If one accepts that S4 is a surrogate marker of subcuta- BMI is significantly different in Aboriginal people and neous FM, and height2/resistance representative of FFM, European Australians. If the relation of body composition then for the same S4, height2/resistance and height both to body weight in young healthy Aboriginal people is Aboriginal men and women had a higher body weight, by significantly different from European Australians, then the 3.8 and 2.0 kg, respectively, as compared to their European currently recommended classification of weight status, based Australian counterparts (Table 3). We assume that part of this on BMI, may be inappropriate for use in this population. The difference in body weight was because of nonsubcutaneous present data support the results of the WHO consultation on fat, as this component of FM was not accounted for in the obesity which, while recommending that the BMI be used to regression model. Given the significantly higher WHR, in classify obesity, recognise that Aboriginal Australians ‘ytend both Aboriginal men and women when compared to the to have a deceptively low BMI; a healthy BMI for this range corresponding European Australian group (see below), we appears to be between 17 and 22, y’ (WHO, 1997). Further speculate that this unaccounted body weight was probably studies, using criterion methods for measuring body compo- intra-abdominal visceral fat. sition, are required to confirm the observations of this study Aboriginal men had significantly narrower hips, an in this and older age groups, where the difference is likely to observation that has been made before (Abbie, 1957), but be greater. similar mean waist circumference compared to European Australian men. A difference in hip circumference in the two groups of women was not evident, but Aboriginal women Acknowledgements had a significantly greater waist circumference as compared We thank the volunteers who made this study possible. This to European Australian women. Consequently, both men study was supported, in part, by the National Health and and women had significantly greater waist-to-hip circumfer- Medical Research Council (NHMRC) of Australia grants (Nos. ence ratios, as compared to their European Australian 934502, 954605). LSP was supported by an NHMRC grant counterparts, suggesting greater abdominal fat even at this (No. 981019). KGR was supported by an NH&MRC young age. This difference was evident at any level of BMI Postdoctoral Research Fellowship (No. 974302) and is 2 (up to 30 kg/m ). The health risks associated with centrally currently a VicHealth Public Health Research Fellow. We deposited fat are well documented (Pi-Sunyer, 1993). Our wish to thank Ms Dympna Leonard and Ms Simone Lowson observations on waist circumference and WHR are in for help with this study. keeping with the observations made in an earlier study involving Aboriginal Australians (Rowley et al, 1997). They are also consistent with the significantly higher References prevalence and incidence, in this population, of risk factors Abbie AA (1957): Metrical characters of a central Australian tribe. that are components of the insulin resistance syndrome, as 27, 220–243. compared to the general Australian population (Daniel et al, Baumgartner RN, Heymsfield SB & Roche AF (1995): Human body Obes. Res. 1999). Future studies using computerised tomography scans composition and the epidemiology of chronic disease. 3, 73–95. of the abdominal region are required to confirm these Bjorntorp P (1999): Neuroendocrine perturbations as a cause of results. insulin resistance. Diabetes Metab. Res. Rev. 15, 427–441. are a diverse group of populations Callaway CW, Chumlea WC, Bouchard C, Himes JH, Lohman TG, and the results identified here are not necessarily generali- Martin AD, Mitchell CD, Mueller WH, Roche AF & Seefeldt VD (1988): Circumferences. In Anthropometric Standardisation Reference sable to all other areas of Australia. The Aboriginal people Manual, eds TG Lohman, AF Roche & R Martorell, pp 55–70. included in this study were from central Australia and north Champaign IL: Human Kinetics Books. . Similar characteristics have been observed in Carey VJ, Walters EE, Colditz GA, Solomon CG, Willett WC, Rosner the Kimberley region of Western Australia (Rutishauser & BA, Speizer FE & Manson JE (1997): Body fat distribution and risk of non-insulin-dependent diabetes mellitus in women. The McKay, 1986). Comparable data from south-eastern Australia Nurses’ Health Study. Am. J. Epidemiol. 145, 614–619. are not readily available, although greater WHR in Abori- Chan JM, Rimm EB, Colditz GA, Stampfer MJ & Willett WC (1994): ginal people in the ‘healthy’ BMI range was reported in at Obesity, fat distribution, and weight gain as risk factors for clinical diabetes in men. Diabetes Care 17, 961–969. least one study (Guest et al, 1993). Body fat distribution may Chertow GM, Lazarus JM, Lew NL, Ma L & Lowrie EG (1997): be modulated by a number of factors, including neuroendo- Bioimpedance norms for the hemodialysis population. Kidney Int. crine responses (Bjorntorp, 1999) and it should not be 52, 1617–1621.

European Journal of Clinical Nutrition Body composition of Australian Aboriginal people LS Piers et al 963 Cunningham J & Mackerras D (1998): Overweight and Obesity, Lemieux S, Prud’homme D, Bouchard C, Tremblay A & Despre´sJP Indigenous Australians. Canberra: Australian Bureau of Statistics. (1996): A single threshold value of waist girth identifies normal- Daniel M, Rowley KG, McDermott R, Mylvaganam A & O’Dea K weight and overweight subjects with excess visceral adipose tissue. (1999): Diabetes incidence in an Australian aboriginal population. Am. J. Clin. Nutr. 64, 685–693. An 8-year follow-up study. Diabetes Care 22, 1993–1998. Lew EA & Garfinkel L (1979): Variations in mortality by weight Deurenberg P, Yap M & van Staveren WA (1998): Body mass index among 750,000 men and women. J. Chronic Dis. 32, 563–576. and percent body fat: a meta analysis among different ethnic Lovejoy JC, Smith SR & Rood JC (2001): Comparison of regional fat groups. Int. J. Obes. Relat. Metab. Disord. 22, 1164–1171. distribution and health risk factors in middle-aged white and Diffey B, Piers LS, Soares MJ & O’Dea K (1997): The effect of oral African American women: The Healthy Transitions Study. Obes. contraceptive agents on the basal metabolic rate of young women. Res. 9, 10–16. Br. J. Nutr. 77, 853–862. Lukaski HC (1987): Methods for the assessment of human body Durnin JV & Rahaman MM (1967): The assessment of the amount of composition: traditional and new. Am. J. Clin. Nutr. 46, 537–556. fat in the human body from measurements of skinfold thickness. McLennan W & Madden R (1999): The Health and Welfare of Br. J. Nutr. 21, 681–689. Australia’s Aboriginal and Torres Strait Islander Peoples. Canberra: Durnin JV & Womersley J (1974): Body fat assessed from total body Australian Bureau of Statistics. density and its estimation from skinfold thickness: measurements Piers LS, Diffey B, Soares MJ, Frandsen SL, McCormack LM, Lutschini on 481 men and women aged from 16 to 72 years. Br. J. Nutr. 32, MJ & O’Dea K (1997): The validity of predicting the basal 77–97. metabolic rate of young Australian men and women. Eur. J. Clin. Ellis KJ, Bell SJ, Chertow GM, Chumlea WC, Knox TA, Kotler DP, Nutr. 51, 333–337. Lukaski HC & Schoeller DA (1999): Bioelectrical impedance Piers LS, Soares MJ, McCormack LM & O’Dea K (1998): Is there methods in clinical research: a follow-up to the NIH Technology evidence for an age-related reduction in metabolic rate? J. Appl. Assessment Conference. Nutrition 15, 874–880. Physiol. 85, 2196–2204. Gasperino J (1996): Ethnic differences in body composition and their Piers LS, Soares MJ, Frandsen SL & O’Dea K (2000): Indirect estimates relation to health and disease in women. Ethn. Health 1, 337–347. of body composition are useful for groups but unreliable in Guest CS, O’Dea K, Hopper JL & Larkins RG (1993): Hyperinsulinae- individuals. Int. J. Obesity 24, 1145–1152. mia and obesity in aborigines of south-eastern Australia, with Pi-Sunyer FX (1993): Medical hazards of obesity. Ann. Intern. Med. comparisons from rural and urban Europid populations. Diabetes 119, 655–660. Res. Clin. Pract. 20, 155–164. Robson JR, Bazin M & Soderstrom R (1971): Ethnic differences in Gurrici S, Hartriyanti Y, Hautvast JG & Deurenberg P (1998): skin-fold thickness. Am. J. Clin. Nutr. 24, 864–868. Relationship between body fat and body mass index: differences Rowley KG, Best JD, McDermott R, Green EA, Piers LS & O’Dea K between Indonesians and Dutch Caucasians. Eur. J. Clin. Nutr. 52, (1997): Insulin resistance syndrome in Australian Aboriginal 779–783. people. Clin. Exp. Pharmacol. Physiol. 24, 776–781. Haffner SM, Mitchell BD, Stern MP, Hazuda HP & Patterson JK (1992): Rowley KG, Iser DM, Best JD, O’Dea K, Leonard D & McDermott R Public health significance of upper body adiposity for non-insulin (2000): Albuminuria in Australian Aboriginal people: prevalence dependent diabetes mellitus in Mexican Americans. Int. J. Obes. and associations with components of the metabolic syndrome. Relat. Metab. Disord. 16, 177–184. Diabetologia 43, 1397–1403. Harrison GG, Buskirk ER, Lindsay Carter JE, Johnston FE, Lohman Rutishauser IH & McKay H (1986): Anthropometric status and body TG, Pollock ML, Roche AF & Wilmore J (1988): Skinfold composition in aboriginal women of the Kimberley region. Med. J. thicknesses and measurement technique. In Anthropometric Stan- Aust. 144(Suppl), S8–S10. dardisation Reference Manual, eds TG Lohman, AF Roche & R Soares MJ, Piers LS, O’Dea K & Shetty PS (1998): No evidence for an Martorell, pp 55–70. Champaign, IL: Human Kinetics Books. ethnic influence on basal metabolism: an examination of data Kleinbaum DG, Kupper LL & Muller KE (1988): Applied Regression from India and Australia. Br. J. Nutr. 79, 333–341. Analysis and Other Multivariable Methods. Boston: PWS-Kent Wang J, Thornton JC, Russell M, Burastero S, Heymsfield S & Pierson Publishing Company. RN (1994): Asians have lower body mass index (BMI) but higher Lean ME, Han TS & Seidell JC (1998): Impairment of health and percent body fat than do whites: comparisons of anthropometric quality of life in people with large waist circumference. Lancet 351, measurements. Am. J. Clin. Nutr. 60, 23–28. 853–856. WHO (1990): Diet Nutrition and the Prevention of Chronic Diseases. Ledoux M, Lambert J, Reeder BA & Despre´s JP (1997): Correlation Report of a World Health Organization study group Technical between cardiovascular disease risk factors and simple anthropo- Report Series 797. Geneva: WHO. metric measures. Canadian Heart Health Surveys Research Group. WHO (1997): Obesity: Preventing and Managing the Global Epidemic. CMAJ 157(Suppl 1), S46–S53. Geneva: WHO.

European Journal of Clinical Nutrition