Hyperinsulinaemia and obesity in Aborigines of south-eastern Australia, with comparisons from rural and urban Europid populations

Hyperinsulinaemia and obesity in Aborigines of south-eastern Australia, with comparisons from rural and urban Europid populations

Diabetes Research and Clinical Practice, 20 (1993) 155-164 0 1993 Elsevier Scientific Publishers Ireland Ltd. All rights reserved. DIABET 155 0168-8...

868KB Sizes 0 Downloads 15 Views

Diabetes Research and Clinical Practice, 20 (1993) 155-164 0 1993 Elsevier Scientific Publishers Ireland Ltd. All rights reserved.

DIABET

155 0168-8227/93/$06.00

0750

Hyperinsulinaemia and obesity in Aborigines of south-eastern Australia, with comparisons from rural and urban Europid populations C.S. Guestavb, K. O’Dea”, J.L. Hopper”

and R.G. Larkins”

‘Department of Medicine (Royal Melbourne Hospital), University oj’ Melbourne, hDepartment of Communit!~ Medicine, Universit!~of Melbourne and “Department of Humon Nutrition, Deakin University. Australia (Received

12 October

1993; revision

accepted

I8 January

1993)

Summary

Diabetes is more common in Aborigines than in other Australian populations, even in groups that have lived in contact with Europids for 150 years. Prevalence data on hyperinsulinaemia and obesity from urbanized south eastern Australian Aborigines are presented with Europid comparisons. Aborigines had higher mean insulin levels than Europids. In females, mean fasting insulin was 15.5 mu/l in Aborigines, compared with 9.5 mU/l in Europids (P < 0.001). The means for males were 15.1 mU/l (Aborigines) and 8.3 (Europids) (P < 0.005). Obesity was more prevalent in Aborigines. In Aboriginal females aged 25-64 years, 41/108 (38%) had BMI >30.0, compared with 37/208 (18%) Europids (P < 0.001). In males, the difference in the prevalence of obesity in Aborigines (17169, 25%) and Europids (341195, 17”/0)was not statistically significant. Waist-hip ratio was significantly greater among Aboriginal females (mean 0.87 in persons aged 25-64 years) than among Europids (mean 0.81, P < 0.001). In males, the mean ratio in Aborigines and Europids was the same (0.94). Abdominal obesity was most prevalent among Aboriginal females. For females aged 20-49 years, 83/l 10 (75%) Aborigines had a waist-hip ratio >0.80, compared with 711165 (43%) Europids (P < 0.001). Being overweight or obese is perceived with least accuracy by Aboriginal males of the four ethnicityjgender groups. Comparisons with national data suggest a gradient in the prevalence of obesity, lowest in urban groups, more in the country, and higher still among Aborigines, which is in reciprocal order to socio-economic status. In multivariate analyses, the association of BMI with insulin was highly significant. Hyperinsulinaemia in an Aboriginal group after many years of contact with Europids may result from environmental as well as genetic influences. Relative hyperinsulinaemia is not found among those Aborigines who have developed glucose intolerance, which could be explained by earlier pancreatic exhaustion in this group.

Key words: Hyperinsulinaemia; Rural; Urban

Corrwondence

to: C.S. Guest,

Obesity; Aborigine; Europid; Australia; Body mass index; Waist-hip ratio:

Department

of Community

Medicine,

University

of Adelaide.

South

Australia

5005. Australia.

156

Introduction Hyperinsulinaemia is associated with diabetes and cardiovascular disease [l-3] but the role of insulin in the pathogenesis of these disorders is poorly understood. Hyperinsulinaemia could be the primary defect in non-insulin-dependent diabetes mellitus (NIDDM), or secondary to insulin resistance [4]. The importance of hyperinsulinaemia as a marker of cardiovascular disease risk in different populations may vary. Obesity is strongly associated with diabetes in many populations, but variations in diabetes prevalence are not fully explained by differences in obesity [5]. A comparison of blacks and whites in the United States, for example, showed that racial differences in diabetes prevalence were not simply related to obesity [6]. Central fat distribution has been recognised as a risk factor for diabetes, independent of the total body mass [7]. Bjorntorp has suggested that intraabdominal fat has a specific role in the development of NIDDM (81. In most Australian settings, diabetes is more prevalent in Aborigines than in Europids [9]. We recently demonstrated the higher diabetes prevalence in Aborigines in south-eastern Australia, where there has been more cultural and genetic mix between Aborigines and Europids than in the more rural central, northern and western parts of the continent [lo]. Diabetes is prevalent at earlier ages in Aborigines than in other Australians, Pooled data from south-eastern and central Australian showed a 12% prevalence in Aborigines Aborigines aged 20-49 years, compared with 1% in Europids restricted to the same age range [l I]. The marked difference in diabetes prevalence at early ages suggests that genetic influences on glucose intolerence are important. This report details the insulin levels and the prevalence of obesity in the subjects in our recent study. The null hypothesis was that insulin levels, at least in the non-obese normoglycaemic sub-populations of Aborigines and Europids, would be the same. Methods Biochemistry Insulin levels were determined by radioimmunoassay based on a double antibody solid

phase technique (Phadeseph Insulin RIA, Pharmacia Diagnostics, Uppsala, Sweden). Glucose levels were determined by a glucose oxidase method, described previously, with details of the oral glucose tolerance test [lo]. Anthropometry Height and weight were measured without shoes according to standard protocols [12]. Waist and hip measurements were taken at the smallest abdominal circumference (between the ribs and iliac crests) and at the level of the greater trochanters, respectively. Standard definitions were used for Body Mass Index (BMI): (weight(kg)/height2(m2)) and waist-hip ratio (cm/cm). For categorical analysis, definitions used by the National Heart Foundation [13] were followed: Body Mass Index (kg/m’) Descriptive Term

Men and Women

Underweight Acceptable weight Overweight Obesity

c20.0 20.0-25.0 25. I-30.0 >30.0

Bjorntorp has defined ‘abdominal obesity’ (sometimes also called central, truncal or android obesity) as a waist-hip ratio of > 1.0 in men and > 0.8 in women [ 141, levels which are applied for categorical analysis of body fat distribution in this study. Further analyses were undertaken on the data from males, applying a lower waist-hip ratio of >0.9 as the definition of abdominal obesity. All subjects were asked for their perception of their current weight, for cross-tabulation with the measurement of BMI restricted to subjects aged 25-64 years. Statistics Student’s t-test was performed for the comparison of means (15,161 and chi-square tests for the comparison of proportions [16]. Regression analyses were performed with statistical software packages (SPSS and GLIM) [ 16,171, following standard assumptions [ 181. The regression analyses examine the relationship between insulin and other biochemical and physi-

157

each model, R-square is also given, showing the proportion of the variance of the dependent variable explained by the model.

cal variables associated with diabetes and cardiovascular disease. Logarithmic transformations of insulin were included in some models. With the identification of participants as Aboriginal or Europid, the inclusion of the variable ‘ethnicity’ in regression models tested the hypothesis that insulin was associated with this proxy for unmeasured genetic and environmental influences, in addition to its known physical and biochemical associations. Non-standardized coefficients (B) indicate the amount of change in the dependent variable for each change in a unit of the predictor under consideration, all other predictors held constant. The magnitude of the coefficient thus depends on the size of units of both independent and dependent variables. The sign of the coefficients (positive or negative) is needed for biological interpretation. The codes used were, for ethnicity, 1, Aborigine; 2, Europid; for gender, 1, Male, 2, Female. In the tables following, a negative coefficient for ethnicity or gender implies that the dependent variable was higher among Aborigines or males, respectively. To compare the different components of each model, ‘standardized parameter estimates’ were calculated by dividing each regression coefficient by its standard error. Only the statistical significance of these standardized estimates, analogous to ‘z scores’ in a normal distribution [19], are shown in the tabulated regression analyses. For

TABLE Fasting

Ethics

Both communities were consulted during the design of the study, which was approved by the Royal Melbourne Hospital Human Ethics Committee.

Results Study population

The sample size and characteristics have been described previously [lo]. Briefly, 306 Aborigines and 553 Europids aged 13 years or older were included, participation rates of 90% and 94%, respectively. As expected from the national age profile of the Aboriginal population, the mean age for Aborigines in this sample was considerably younger [lo]. The mean age ( lS.E.M.), in years, for the population sub-groups were: males, Aborigines 29.9 (1.2); Europids 43.0 (1 .O); females, Aborigines 34.1 (1.2); Europids 43.2 (1 .O). The target populations in the study were the Aborigines and Europids of south-eastern Australia. The extremes of highly urban or ‘outback’ existence were avoided in the choice of defined areas in population centres of the Victorian countryside, where Aborigines and

I and 2-h glucose

and insulin

levels: comparison

of means,

stratified

by race and gender

Mean (S.E.) P”

Fasting glucose

P”

2-h glucose (mmol/l)

(mmol/l)

P”

Fasting insulin (mu/l)

P”

2-h insulin (mu/l)

Males 0.06

ns Europid (n = 272) Aboriginal (n = 125)

5.5 (0.1) 5.6 (0.2)

5.4 (0.2) 6.1 (0.3)

1

0.005 x.3 (0.4) 15.1 (2.3)

1

0.04 28.2 (2.2) 38.2 (4.2)

I

I

Females

0.05

ns Europid (n = 281) Aboriginal (n = 179) a Significance

5.3 (0.1) 5.6 (0.2)

of the difference

1

of means between

5.7 (0.2) 6.4 (0.3) Aborigines

1 and Europids:



1

40.6 (2.7) 59.1 (3.7)

r-test. ns. not significant.

1

158

5.0-6.39

C5.0

6.4-1.79

Fasting glucose (mmol/U Fig. 1. Fasting insulin by grouped fasting glucose, Aborigines (shaded bars) and Europids (unshaded), excluding diabetics. (Means with standard errors. Comparisons, Student’s r-test: all strata, P c 0.01).

Europids began genetic and cultural contact before 1850 [IO]. Genetic admixture was not quantified, but is most probably greater than in Aboriginal populations in the central and northwestern parts of Australia [3]. Biochemistry

Intra- and inter-assay coefficients of variation

for the insulin assays were 2% and 4%, respectively. Fasting and 2-h post-load insulin levels were significantly higher in Aborigines (Table 1). For glucose, fasting levels were similar in Aborigines and Europids, and the statistical significance of the difference for 2-h levels was marginal. With stratification by level of glucose, and restriction to normoglycaemic subjects, the differences between Aborigines and Europids in insulin levels remained statistically significant (Fig. 1). The 2-h data indicate relative hyperinsulinaemia among Aborigines with normal glucose tolerance, again statistically significant (Fig. 2). The apparent cross-over - the higher insulin levels among glucose-intolerant Europids, compared with Aborigines - did not achieve statistical significance, however. When comparisons are matched on the degree of glucose intolerance, Aborigines with impaired glucose tolerance or diabetes did not have higher insulin levels than Europids. In non-obese young subjects, higher insulin levels were found in Aborigines than in Europids, consistent with insulin resistance (Table 2, with mean insulin values 27% and 11% higher in Aboriginal males and females, respectively). For the female comparison, the mean fasting glucose was 6% lower among Aborigines, of doubtful biological significance, although the small standard errors of these measurements and large sample

160

140 1 120g 1

loo-

SO-

.”

2-hr glucose (mmolil) Fig. 2. Two-hour insulin by grouped 2-h glucose, Aborigines (shaded bars) and Europids (unshaded). (Means with standard errors. Comparisons, Student’s f-test: glucose strata < 6.4 mmol/l, P < 0.01; glucose 6.4-7.79, P = 0.04: others ns.).

159

TABLE

2

Fasting

glucose

and insulin:

comparison

of means in subjects

aged 13-34

years, with BMI < 25 kg/m2

Means (S.E.) Europids

P”

Males (54 Aborigines, 48 Europids) Glucose (mmolfl) 5.1 (0.1) Insulin (mu/l) 8.9 (0.7)

5.2 (0.1) 7.0 (0.5)

ns 0.03

Females (52 Aborigines, 67 Europids) Glucose (mmohl) 4.7 (0.1) Insulin (mu/l) 8.8 (0.5)

5.0 (0.1) 7.9 (0.4)

0.001 ns

Aborigines

a Student’s

r-test, 2-tail.

sizes produced a difference that was statistically significant.

Both male and female Aborigines are shorter than their Europid counterparts. These countrytown Europids approximated the national sample in height, but they had a 5-kg greater mean body weight than the national sample. The modal category for Aboriginal and Europid males was ‘overweight’ and for Aboriginal females, ‘obese’. Only Europid females had ‘acceptable weight’ as their mode, conforming to

Anthropometry The greatest differences in anthropometric measures between Aboriginal and Europid subjects occurred in females, where BMI and waist-hip ratio were both significantly higher in Aborigines (Table 3).

TABLE Height.

3 weight,

BMI and waist-hip

ratio in Aborigines,

Europids

and a national

sample.”

Ages 25-64

years

Mean (SE.)

Males Height (cm) Weight (kg) BMI (kg/m*) Waist-hip ratio

Females Height (cm) Weight (kg) BMI (kg/m2) Waist-hip ratio a Adapted from National Heart b Student’s r-test for comparison

Aborigines (n = 69)

Europids (n = 195)

Pb

National sample (n = 4552)

170.9 (0.7) 77.3 (1.9) 26.4 (0.6) 0.94 (0.01)

175.1 (0.5) 83.1 (0.8) 27.1 (0.2) 0.94 (0.01)


175 79.4 25.6 0.90

Aborigines (n = 108)

Europids (n = 208)

Pb

National (n=4631)

157.4 (0.5) 71.3 (1.5) 28.8 (0.6) 0.87 (0.01)

163.0 (0.4) 69.4 (0.9) 26.1 (0.3) 0.81 (0.01)


162 65.4 24.6 0.76

Foundation [13]. of Aborigines and Europids

of the present

study.

sample

160 TABLE 4 Categories of BMI in persons aged 25-64 years: Aborigines and Europids compared with a national sample” Category b

Aborigines (%)

Europids (‘XI)

National sample ‘X

Males Underweight Acceptable weight Overweight Obese Total

5 (7.2) 21 (30.4) 26 (37.7) 17 (24.6) 69 (100)

2 (1.0) 50 (25.6) 109 (55.9) 34 (17.4) 155 (100)

100

Females Underweight Acceptable weight Overweight Obese Total

3 (2.8) 29 (26.9) 35 (32.4) 41 (38.0) 108 (100)

8 (3.8) 87 (41.8) 76 (36.5) 37 (17.8) 208 ( 100)

12.5 50.3 24.6 12.5 100

3.3 42.1 43.3 11.3

s Adapted from National Heart Foundation [l3]. b Defined in the text.

the national samples (Table 4). Obesity was commoner in Aborigines than Europids in the present study (collapsing underweight with acceptable weight to form a single category, for males, chi-square, 2 d.f. = 6.7, P < 0.05; for females chisquare, 2 d.f. = 16.6, P < 0.001.) In turn, the Europids in this Victorian setting were much more frequently overweight or obese than reported in the national sample [ 131 (frequencies compared as above, with 2 d.f: in males, chi-square = 27.2,

P < 0.001; in females, chi-square = 24.9, P < 0.001). The perception of being overweight among Aborigines and Europids differed by gender. Among males, 1341272 (46%) Europids compared with 41/123 (33%) Aborigines considered themselves overweight (other categories were ‘OK or average’ and ‘underweight’; chi-square, 2 d.f. = 7.4, P < 0.05). The reverse held for females, where a higher proportion of Aborigines (114/l 79,

TABLE 5 Waist-hip ratios in females, stratified by age, ethnicity and BMI mean (SE.) Age group (years)

Aborigines

tl

Europids

II

BMI ~25.0kg/m' 13-34 35-54 >54

0.80(0.01) 0.85(0.02) 0.87(0.04)

51 12 3

0.77(0.01) 0.79(0.01) 0.81(0.01)

67 43 32

BMI >25.0kg/m2 13-34 35-54 >54

0.87(0.01) 0.91(0.01) 0.92(0.02)

54 37 21

0.81(0.02) 0.81(0.01) 0.86(0.01)

28 73 38

a Student’s r-test, 2-tail.

P"

0.03 0.02 0.11

<
161

I

13-19 20-29 30-39 40-49 50-59 Age(yead

64%) judged themselves overweight (Europids: 155/281, 55%; chi-square, 2 d.f. = 7.4 (again), P < 0.05). Both subjective and objective assessments of body weight were available on 61 Aboriginal males aged 25-64 years. In nine cases (lS”/o), the subjective assessment was ‘acceptable’, while the objective finding was ‘overweight or obese’ (chi-square (McNemar) = 4.9, P < 0.05). Thus, a significant proportion of Aboriginal males assessed their weights as acceptable when objective measurement suggested otherwise. For Aboriginal females and Europids, however, the proportions of discordant pairs were not significant, i.e. in each of these groups, there was a close correspondence between subjective and objective assessments. The differences in waist-hip ratios persisted with stratification by age, race and BMI, with the fat distribution of Aborigines significantly more central or ‘android’ than in Europids (Table 5). With application of Bjorntorp’s (1985) definitions of central obesity to these samples, in males,

I’: 260

Fig. 3. Central obesity in females (waist:hip ratio > 0.80). by age-group in Aborigines (shaded bars) and Europids (unshaded) (Proportions with 95”/~ confidence intervals. Comparisons. chi-square test: 30-39 years stratum, P < 0.05; 260 years stratum P = 0.07: other strata, P < 0.01).

TABLE

6

Associations

of plasma

Dependent variable

insulin in Aborigines

and Europids:

Predictor variable

multiple

linear regression

Regression coefficient

analyses

SE. of B

P”

0.086 0.409 0.597

<0.0001
(B) Fasting Insulin (mu/l)

BMI (kg/m2) Ethnicityb Fasting glucose

Other variables R-square:

initially.

but not achieving

statistical

significance:

age. gender,

waist-hip

ratio

0.17

2-h Insulin (mu/l)

2-h glucose (mmolll) BMI (kg/m*) GenderC Ethnicityb

Other variables R-square:

entered

(mmohl)

0.708 -2.821 2.741

entered

initially,

but not achieving

11.448 1.985 11.070 -4.493 statistical

0.688 0.244 2.403 I.271 significance:

age, fasting

<0.0001 0.0001 < 0.0001 0.0004 glucose,

waist-hip

ratio

0.36

a Statistical significance of the difference between the regression coefficient and zero. b Codes: Aborigines = 1, Europids = 2; the negative coefftcients imply the dependent variables were higher among ’ Codes: males = 1, females = 2; the positive coefficient implies the dependent variable was higher among females.

Aborigines

162

the prevalence of central obesity in Aborigines and Europids was similar (considering ages 20-49 years, 14/80 (18%) Aborigines had a waist-hip ratio > 1.00, compared with 22067 (13%) Europids, n.s.). Pooling the data on Aboriginal and Europid males, in age group 20-29 years, the prevalence of central obesity was 6/70 (8%); for those aged 60 years or over, the proportion was 1l/51 (22%). The prevalence of central obesity is higher if a waist-hip ratio >0.90 is applied (in the agegroup 20-49 years, 30% of Aborigines, compared with 32% of Europids, difference again not significant). In females, however, the differences between populations were highly significant: for ages 20-49 years, 831110 (75%) Aborigines had a waisthip ratio >0.80, compared with 71/165 (43%) Europids (chi-square = 28.2, P < 0.001). The great difference in the prevalence of central obesity between Aboriginal and Europid females was statistically significant in each age group (Fig. 3). In the multiple linear regression models (Table 6), 24 Aboriginal and 19 Europid diabetics were excluded, leaving 816 subjects. Only the statistical significance of variables remaining in the equation after the final iteration are shown, but for each model, the variables that did not achieve statistical significance are also listed, so that the range of variables initially entered can be seen. The strong positive association between insulin and BMI, well established in Aboriginal and other populations [3], was confirmed. The variable ‘ethnicity’ (differences between Aborigines and Europids, independent of other variables in the equation) was a highly significant association of both fasting and 2-h insulin. BMI was a stronger association of fasting insulin within each population (when Aborigines and Europids were considered in separate models, details not shown), than in the model with all subjects (the merged data on both Aborigines and Europids, Table 6, upper model). When considered simultaneously with BMI, waist-hip ratio was not a statistically significant predictor of insulin in any of these models.

Discussion The study clearly shows relative hyperinsulinaemia among Aborigines with normal glucose

tolerance, compared with Europids. Relative hyperinsulinaemia is not found among Aborigines with glucose intolerance, however, suggesting that diabetes in Aborigines is accompanied by ‘earlier’ pancreatic failure, when insulin levels may be lower than in Europids. Obesity, frequently associated with central fat distribution, is also highly prevalent in this urbanized Aboriginal population. The greater BMI among Aboriginal females compared with Europid females is confirmed in this study. The comparison with the national data shows further that BMI (in Europids) is higher in the country than in the city. This finding, one of the clearest distinctions in this study, shows that the distribution of cardiovascular risk factors may differ between urban and rural populations, at least in the highly urbanized state of Victoria. Mean BMI in this country-town Europid sample is higher than for the urban population reported by the NHF [13] in which the all-ages average was 25.3 kg/m* (males) and 24.3 (females). Further investigation is indicated, as this finding contradicts the expectation of higher BMI in the supposedly more sedentary urban population. In future studies, precise consideration of socio-economic status and physical activity in urban-rural comparisons is needed. The waist-hip ratio and BMI of Australian Aborigines vary according to geographic location, an index of environmental differences and unknown pre-European genetic effects and more recent genetic admixture (especially in southern groups). From Aborigines in northern Australia, O’Dea reported on subjects aged 1 35 years [20]: in males - mean BMI of 23.5 kg/m2 and mean waist-hip ratio of 0.95 (comparable with 26.4 kg/m* and 0.94 from Table 3); in females - mean BMI of 22.7 kg/m* and waist-hip ratio of 0.91 (comparison from Table 3, 28.8 kg/m’ and 0.87). Thus, males in the south-east have higher mean BMI, but no greater centrality of fat distribution. The contrast was even more marked for females, with much greater BMI, but somewhat lower centrality. Similar or even lower degrees of central fat distribution may explain why diabetes prevalence is not much higher in the south-eastern Aborigines, although the regression analyses suggested that BMI is a more important association of insulin, and thus glucose intolerance, than waisthip ratio. Centralized body fat distribution in

163

Aborigines may have prevailed before the diabetes epidemic: evidence for this is at least as old as the photographs taken on the ethnographic expeditions of Baldwin Spencer in the early years of this century [21]. The methods of these studies of Aboriginal groups were the same, but inter-observer variation has not been quantified in the comparisons above. Aboriginal-Europid comparisons should be even more tentative, as Rutishauser and McKay [22] have suggested that conventional, western standards are inappropriate for the interpretation of the BMI of Aboriginal women. For a given BMI. Aboriginal women have more subcutaneous fat and thus a higher percentage of body fat than Europid women [22]. The marked differences of BMI as measured (Table 3) therefore may still underestimate the physiological differences due to the greater adiposity of Aboriginal women compared with Europids. The perception of being overweight was greatest among Aboriginal females. Although the questionnaire was not validated, the perception of this group agreed with the objective assessments, as it did for Europids, male and female, while Aboriginal males’ subjective assessment tended to be more favourable than the objective measures. This may indicate a greater tolerance for obesity among Aboriginal men than women, consistent with a recent study of obese men and women in Sydney which showed greater social acceptability of being overweight in men than in women

individuals as well as between the population groups, remains unexplained. The results are consistent with the thrifty genotype hypothesis [3,9], but the observation of hyperinsulinaemia in an Aboriginal group thought to have a large proportion of Europid genetic admixture raises the possibility that environmental influences on insulin levels and glucose tolerance are also important. The high prevalence of obesity, at least from the time of adolescence in these Aborigines, casts doubt on the hypothesis that foetal malnutrition leads to diabetes [24], unless a ‘rebound’ to obesity during early childhood occurred, and went undetected in this study. Hyperinsulinaemia is widespread in Aborigines, but comparisons between Aboriginal groups are limited by uncertainty about the heterogeneity of Aborigines before European contact, as well as after it. Abbie’s conclusion [25] that the Australian Aborigines were a homogeneous population has been cast in doubt by re-analysis of old data [26]. Climatic influence on the development of Aborigines, including the putative thrifty genotype, may have been underestimated. Elsewhere, we have reported an association between smoking and waist-hip ratio found in these population groups [27]. In the present paper, however, waist-hip ratio appears to be a less important association of insulin than BMI. Thus, there is no evidence here that smoking cessation would improve glucose tolerance, although it would have many other health benefits.

~231.

With the present cross-sectional data, genetic effects cannot be separated from the environmental differences by any statistical method. All of the biochemical data are likely to be partly determined by genetic differences not only between Aborigines and Europids, but also within each of these population groups. The variable ‘ethnicity’ in these analyses should therefore be understood only as differences between Aborigines and Europids not explained by other variables in the regression models. It is theoretically possible for all of these differences to be environmentally determined; they are assumed, however, to have at least some genetic basis, with some genetic-environmental interaction. The low R-squares in the regression models presented, while typical of analyses of this type, indicate that much of the variation between

Conclusion

Higher levels of insulin were found in urbanized Aborigines of south-eastern Australia compared with Europids. Obesity was more common among Aborigines than Europids in this study, with the main disparity in females. The comparison with National Heart Foundation data suggested a gradient in the prevalence of obesity, with the highest among Aborigines, the lowest among urban Europids. and rural Europids in between. The major difference in BMI between Aboriginal males and females was not reflected by a gender difference in diabetes prevalence [IO]. Obesity is a cardiovascular risk factor [28], but its relationship with overall mortality, independent of

164

other risk factors, is not well established. Prospective studies of obesity in Aborigines that include measurement of the rate of change of BMI and waist-hip ratio are needed.

II

I2 I3

Acknowledgements The authors thank the participating communities and Kathy Traianedes for assistance with biochemical tests in the laboratory. During field work, CSG was supported as a National Health and Medical Research Council (Australia) Scholar.

14 I5

I6 I7

References I

2

3

4

5

6

7

8

9

10

I8

Welborn, T.A., Breckenridge, A., Rubenstein, A.H., Dollery, C.T. and Fraser T.R. (1966) Serum-insulin in essential hypertension and in peripheral vascular disease. Lancet i, 1336-1337. Alberti, K.G.M.M. and Mazze, R.S. (1989) (Eds.), Frontiers of Diabetes Research: Current trends in non-insulindependent diabetics. Elsevier, Amsterdam. O’Dea, K. (1991) Westernisation. insulin resistance diabetes in Australian Aborigines. Med. J. Aust. 258-264.

and 155.

Zimmet, P., Finch, C. and Dowse G. (1990) The epidemiology of diabetes mellitus. In: Alberti, K.G.M.M. and Krall, L.P. (Eds.), The Diabetes Annual, 5. Elsevier, Amsterdam.

19

20

21

22

Ekoe, J.M. (1988) Diabetes Mellitus. Aspects of the world-wide epidemiology of diabetes mellitus and its longterm complications. Elsevier, Amsterdam. O’Brien, T.R., Flanders, W.D.. Decoufle. P., Boule, C.A.. DeStefano, F. and Teutsch, S. (1989) Are racial differences in the prevalence of diabetes in adults explained by differences in obesity? J. Am. Med. Assoc. 262, 1485-1488.

23

Kissebah, A.H., Vydelingum, N., Murry, R. et al. (1982) Relation of body fat distribution to metabolic complications of obesity. J. Clin. Endocrinol. 54, 254-260. Bjorntorp, P. (1988) The associations between obesity, adipose tissue distribution and disease. Acad. Med. Stand. (Suppl.) 723. 121-134. Guest, C.S. and O’Dea, K. (1992) Diabetes in Aborigines and other Australian populations. Aust. J. Public Health 16, 340-349. Guest, C.S., O’Dea. K., Hopper, J.L., Nankervis. A.J. and Larkins, R.G. (1992) The prevalence of glucose intolerance in Aborigines and Europids of south-eastern Australia. Diab. Res. Clin. Pratt. 15. 227-235.

26

24

25

27

28

Guest, C.S. (1991) Glucose intolerance and risk factors for cardiovascular disease in Aborigines and Europids of south-eastern Australia. PhD Thesis, University of Melbourne, Chapter 6. Truswell, A.S. (1981) Anthropometric assessment with heights and weights. Trans Menzies Foundation 3. 59-69. Risk Factor Prevalence Study Management Committee. Risk Factor Prevalence Study: Survey no 3 1989. Canberra: National Heart Foundation of Australia and Australian Institute of Health, 1990. Bjorntorp, P. (1985) Obesity and the risk of cardiovascular disease. Ann. Clin. Res. 17, 3-9. Armitage, P. and Berry, G. (1987) Statistical Methods in Medical Research, 2nd edn. Blackwell, Oxford, pp. 107-109. Nie, N.H. SPSS Users Guide. McGraw Hill, New York, 1983. Baker, R.J. and Nelder, J.A. (1978) The GLIM System, Release 3. Numerical Algorithm Group, Oxford. Kleinbaum. D.G.. Kupper. L.L. and Muller. K.E. (1988) Applied Regression Analysis and other Multivariable Methods. PWS-Kent Publishing Co, Boston, p. 107. Matthews, D.E. and Farewell. V.T. (1988) Using and Understanding Medical Statistics, 2nd edn. Karger. Base]. p. 128. O’Dea. K., Lion, R.J., Lee, A., Hopper, J.L. and Rae. C. (1990) Diabetes, hyperinsulinemia and hyperlipidemia in a small Aboriginal community in Northern Australia. Diabetes Care 13, 830-835. Mulvaney, J., Walker, G. and Vanderwal R. (1982) The Aboriginal Photographs of Baldwin Spencer. John Currey, O’Neil, Melbourne. Rutishauser, I.H.E. and McKay. H. (1986) Anthropometric status and body composition in Aboriginal women of the Kimberley region. Med. J. Aust. 144. S8-SIO. Craig, P.L. and Caterson, I.D. (1990) Weight and perceptions of body image in women and men in a Sydney sample. Comm. Health Stud. XIV. 373-383. Hales, C.N. and Barker, D.J.P. (1993) Type 2 (noninsulin-dependent) diabetes mellitus: the thrifty phenotype hypothesis. Diabetologia, in press. Abbie, A.A. (1966) The anthropological status of Australian Aborigines. Homo. 17, 73-88. Macho, G. and Freedman, L. (1987) A re-analysis of the Andrew A Abbie morphometric data on Australian Aborigines. Occasional papers in human biology, no. 4. Australian Institute of Aboriginal Studies, Canberra. Guest. C.S.. O’Dea. K.. Carlin, J.B. and Larkins. R.G. (1992) Smoking in Aborigines and persons of European descent in south-eastern Australia: prevalence and associations with food habits, body fat distribution and other cardiovascular risk factors. Aust. J. Public Health 16, 397-402. Dowse, G.. Zimmet, P., Collins, V. and Finch, C. (1990) Obesity in PaciIic Populations. In: Bjorntorp, P. and Brodoff. B. (Eds.), Obesity. JB Lippincott. Philadelphia.