Nutritional Status in Adults in the Pluri-Ethnic Population of New Caledonia

JM TassieÂ1, L Papoz1, S Barny1, D Simon2 and The CALDIA Study Group3

1INSERM CJF 93-06, HoÃpital Saint-Charles 34295, Montpellier, Cedex 5, France, 2INSERM UniteÂ 21, 16 av Paul Vaillant-Couturier 94807, Villejuif, Cedex. France and 3Health and Social Services in New Caledonia

OBJECTIVE: To describe the nutritional status (body mass index (BMI) and waist-hip ratio (WHR)) of the population of New Caledonia in relation to ethnicity and urban-rural environment. DESIGN: Diabetes screening survey in two rural provinces of New Caledonia and in the suburbs of Noumea. SUBJECTS: 8875 subjects aged 30±59 y, Europeans, Melanesians and Polynesians. MEASUREMENTS: BMI, WHR. RESULTS: Obesity (BMI 27 kg/m2 in men, 25 kg/m2 in women) was highly prevalent in all groups, but varied according to ethnicity: respectively, 43% and 52% in Europeans, 46% and 72% in Melanesians, 72% and 83% in Polynesians. In the urban area, mean WHR values, adjusted for age and BMI, were signi®cantly higher than in rural areas, especially in Melanesians. CONCLUSION: Both ethnicity and urban-rural environment are linked to the amount and distribution of adiposity, which appeared worsened in the urban area in Europeans, and even more in Melanesians.

Keywords: body mass index; waist-hip ratio; prevalence; ethnicity; environment

Introduction constitutes a simple and inexpensive index for epidemiologic purposes. A recent study has shown the overall high preva- Obesity is associated with a signi®cant morbidity and lence rate of diabetes mellitus in New Caledonia, and mortality.1,2 A major impact has been found on its relationship to ethnicity and environmental fac- mortality from cardiovascular disease,2,3 because tors.18 This report presents the analysis of the data weight excess induces a propensity to hypertension, collected on the nutritional status (obesity and fat hypertriglyceridemia, glucose intolerance and dia- distribution) in the pluri-ethnic population of New betes mellitus,1±6 which are risk factors for ischaemic Caledonia (Melanesians, Europeans and Polynesians) disease. However, obesity includes different pheno- and the rural-urban differences. types, the distribution of body fat appearing to act speci®cally.7 Upperbody obesity, or android type, de®ned as excess sub-cutaneous fat on the truncal- abdominal area, is often associated with metabolic Population and methods complications.5±14 This phenotype is one of the factors likely to explain the sex differences in propensity for Population these abnormalities and excess cardiovascular risk of 9 New Caledonia is a French South Paci®c Island, about women with android fat distribution. Different 2000 km east of Australia. The total population indices of fat distribution have been proposed in the (164 000 inhabitants) is made of Melanesians (46%), literature, some based on anthropometric measure- Europeans (34%), Polynesians (12%), and some other ments, and others on sophisticated methods, such as migrant ethnic groups (8%). The island is divided into dual-energy X-ray absorptiometry or bioelectrical 15 three provinces. The capital, Noumea (65 000) inha- impedance measurement. In ®eld studies, anthropo- bitants), is situated in the South Province. The North metric measurements are the most practicable and Province and the Loyalty Islands Province (three based on skinfold thickness and/or circumferences 16 small islands) are mainly rural. Melanesians live in measured in various parts of the body. During the 357 small traditional villages (tribes). They have free last ten years, the ratio of waist to hip circumference access to health centres, on average one in about 4000 (WHR) has been shown to be highly correlated with 17 inhabitants. The lifestyle is very different between the occurrence of chronic disease. This parameter urban and rural areas, and cultural and socio-eco- nomic differences are also observed among the main Correspondence: Dr L Papoz, INSERM CJF 93-06, HoÃ pital Saint- Charles 34295, Montpellier, Cedex 5, France. ethnic groups, especially in rural areas. In peri-urban Received 8 March 1996, revised 27 September 1996; accepted 30 districts, the difference is much less, since the various September 1996 ethnic groups are living in the same block of ¯ats. BMI and WHR in New Caledonia J-M TassieÂ et al 62 The present study was included in a large diabetes index (RWI) of 120% of ideal weight.22 In standing screening programme, the CALDIA survey, con- position, waist and hip circumferences were measured ducted from 1992±1994. The rationale, design and with a special tape to the nearest cm, the ®rst at the methods of the survey have been described in detail umbilicus, the second at the iliac spines. The waist to elsewhere.18 hip ratio (WHR) was calculated as an index of fat The target population was de®ned as residents for distribution. Subjects were classi®ed as having an more than ten years in New Caledonia, aged 30±59 y. android fat distribution if WHR was 0.95 in men Subjects with life threatening diseases and pregnant and 0.85 in women. In the 239 subjects who were women were excluded (the latter were invited to enter measured twice within the same month, the coef®- the study three months after delivery). cients of variation for height, weight, BMI, waist and Subjects were recruited in selected villages of the hip circumferences and WHR were respectively 0.5, rural areas and suburbs of Noumea. In each location, 2.1, 2.3, 3.5, 3.2 and 3.8%. all subjects corresponding to the selection criteria Ethnicity was reported by the subjects during the were invited to participate. In the North province, interview. In case of doubt about the origin, ethnicity 101 villages out of 199 were randomly selected. Only was quoted as `other' (n 186) and excluded from the one refused. In the Loyalty Islands, all 85 villages present analysis. participated. Finally, of the 9130 subjects aged 30±59 who lived in rural areas, 7514 (82%) were screened. Among them, 653 were Europeans and 6503 Melane- Statistical analysis sians. Only a minority (90 subjects) were Polynesians. In the urban area of Noumea, six suburbs with a pluri-ethnic population were selected to avoid districts Analyses were performed separately for men and where temporary residents lived: of 2800 subjects women. One-way and two-way analysis of variance, corresponding to the selection criteria, 1875 (67%) covariance analysis and Student's t±test were used to were screened, including 523 Europeans, 641 Mela- compare means. Comparisons between sexes were nesians and 465 Polynesians. adjusted for age and environment (urban/rural) in Melanesians and Europeans, and adjusted for age in Polynesians (as 84% of the Polynesians are living in Methods the urban area). All statistical analyses were computed on CSS `Statistica' Software (Statsoft, Tulsa, Okla- homa, USA). After intensive information campaigns, data were collected at home by specially trained medical staffs. Weight was measured in underclothes to the nearest kg using a clinical scale. Height was recorded Results to the nearest cm. Body mass index (BMI weight/ height2 (kg/m2)) was calculated as an index of obesity Morphotype in ethnic groups independent of height,19±21 and appropriate for differ- Average anthropometric measurements and indices ent populations and races.19 Obesity was de®ned by are presented in Table 1, by ethnic groups, according a BMI value of 27 kg/m2 or greater for men and 25 to sex and urban/rural environment, except for Poly- kg/m2 for women (corresponding to a relative weight nesians, which are presented as a whole.

Table 1 Mean ( Æ s.d.) values of weight, height, body mass index (BMI), prevalence rates of obesity, waist circumference (WC), hip circumference (HC) and wait-hip ratio (WHR), by ethnic group in subjects aged 30±59 y, according to sex and urban-rural environment (CALDIA Study, 1992±1994, New-Caledonia)

Ethnic group Sample Weight Height BMI Obesitya WC HC WHR size (kg) (cm) (kg/m2 (%) (cm) (cm)

Europeans urban men 224 79.9 Æ 16.0 173 Æ 8 26.6 Æ 4.8 39.7 Æ 0.03 96.6 Æ 13.6 100.5 Æ 9.0 0.96 Æ 0.07 women 299 67.5 Æ 17.5 161 Æ 6 26.1 Æ 6.3 46.5 Æ 0.03 93.4 Æ 16.6 102.9 Æ 12.6 0.90 Æ 0.08 rural men 336 81.5 Æ 16.8 173 Æ 7 27.2 Æ 5.2 45.4 Æ 0.03 93.8 Æ 13.8 100.1 Æ 10.3 0.94 Æ 0.07 women 317 70.8 Æ 18.1 161 Æ 6 27.3 Æ 6.7 56.9 Æ 0.03 86.3 Æ 16.4 103.3 Æ 15.3 0.83 Æ 0.08 Melanesians urban men 213 82.3 Æ 14.0 170 Æ 6 28.5 Æ 4.6 59.1 Æ 0.03 97.3 Æ 11.5 101.7 Æ 8.1 0.96 Æ 0.06 women 428 75.4 Æ 14.9 159 Æ 5 29.7 Æ 5.7 79.6 Æ 0.02 103.0 Æ 19.1 106.2 Æ 10.1 0.97 Æ 0.08 rural men 3010 77.5 Æ 13.3 170 Æ 6 26.8 Æ 4.4 44.6 Æ 0.01 89.5 Æ 11.4 97.6 Æ 9.5 0.92 Æ 0.06 women 3493 72.1 Æ 14.6 159 Æ 6 28.5 Æ 5.5 71.4 Æ 0.01 92.3 Æ 13.2 102.1 Æ 11.4 0.90 Æ 0.07 Polynesians (all) men 222 90.7 Æ 18.0 174 Æ 7 29.9 Æ 5.4 72.1 Æ 0.04 101.9 Æ 13.6 104.5 Æ 9.7 0.98 Æ 0.06 women 333 85.9 Æ 19.6 164 Æ 6 32.0 Æ 6.8 82.9 Æ 0.02 105.4 Æ 16.3 110.9 Æ 11.9 0.96 Æ 0.08

aBMI 27 kg/m2 in men, 25 kg/m2 in women BMI and WHR in New Caledonia J-M TassieÂ et al 63

Figure 1 Distribution of body mass index (BMI) by sex and ethnic group in New Caledonia, 1992±1994. u obesity (BMI 27 kg/m2 in men, 25 kg/m2 in women, relative weight index 120%). In the upper right square: n number of subjects examined, and mean BMI Æ standard deviation.

Figure 1 presents the BMI distributions by sex and sians, 0.952 in Europeans, 0.961 in Polynesians ethnic group. In men, the overall prevalence rates of (P < 0.0001); and in females, respectively, 0.913, obesity were 43% in Melanesians, 46% in Europeans 0.877, 0.937 (P > 0.0.0001). and 72% in Polynesians. In women, the corresponding Sex differences in the body fat distribution were ®gures were respectively 52, 72 and 83%. Two found in Europeans (95% CI for WHR differences: features appear. Firstly, excess body fat is particularly 0.078 Æ 0.009) and, to a lesser degree, in Polynesians high in the Polynesian group. Secondly, Polynesian (0.029 Æ 0.012) with an android morphotype being women had a higher mean BMI than men more marked in men than in women. However, this (P 0.0004); this was also observed in Melanesians sexual dimorphism was greatly reduced in the Mela- (P 0.0001), but not in Europeans (P 0.90). The nesian sample in which both sexes showed the same comparison of BMI according to age using covariance fat distribution (95% CI for WHR; 0.009 Æ 0.004). analysis showed no signi®cant variation in the male sample (P 0.60), while mean BMI increased with age in women (P 0.005) (Figure 2, upper part). Morphotype and environment In addition, morphotype differences were found Since Polynesians live mainly in Noumea, this analy- among the three ethnic groups, in men as well as in sis was restricted to Melanesians and Europeans. On women (Figure 2, lower part). The android pattern of average, BMI levels (Figure 3) were signi®cantly obesity was marked in the Polynesian sample. In all higher in urban Melanesians than in rural Melanesians ethnic groups, WHR was positively correlated to age, (P 0.0001 in each sex). However, in European as well in men as in women. WHR also increased with women, higher BMI values were present in the rural BMI (regression coef®cient 0.0068 for 1 kg/m2 in sample (P 0.02), while no systematic difference was men, 0.0039 for 1 kg/m2 in women, (P < 0.0001 in seen in European men. both sexes). After adjustment for age and BMI, Urban-rural differences regarding WHR presented differences in mean WHR remained signi®cant the same trend whatever the ethnic group and sex between ethnic groups: in males, 0.929 in Melane- (Figure 4). Mean WHR values were systematically BMI and WHR in New Caledonia J-M TassieÂ et al 64

Figure 4 Environmental differences in body fat distribution in Melanesians and Europeans: variation of mean waist-hip ratio (WHR) according to age by sex and urban-rural environment in New Caledonia, 1992±1994.

Discussion

As in any epidemiologic approach in population samples, the results observed in this survey may be Figure 2 Variation of mean body index (BMI) and mean waist- hip ratio (WHR) according to age by sex and ethnic group in New affected by some biases. Thus, all efforts were direc- Caledonia, 1992±1994. ---- usual limits of BMI for obesity and of ted at the standardization of the methods, through WHR for central adiposity. training sessions of the medical staff involved in the survey, and regular checking by the coordination group. In spite of many efforts, the participation was higher in women than in men. This is observed in most epidemiologic surveys, because men are more often employees than women, and are less motivated by programmes regarding health problems. In New Caledonia, work time begins early in the morning, and therefore most of the male non-responders were not at home when the medical staff came for examination. This is one of the reasons explaining the excess of females in our sample, as already observed in other published material from the South Paci®c.23,24 We took this into account by presenting the results separately for each sex. Regarding the reliability of the anthropometric measurements, we assessed on a sub- sample the intraindividual coef®cient of variation of the parameters presented in this report. All were lower than 4%. In any case, in this region important varia- Figure 3 Environmental differences in fatness in Melanesians tions occur in the health status of native populations, and Europeans: variation of mean body mass index (BMI) according to age by sex and urban-rural environment in New so that attention should be focused on trends, rather Caledonia, 1992±1994. than on absolute values which are obviously changing rapidly.24 First of all, our study con®rmed the excessive higher in the urban sample in Melanesians (P < 10 7 fatness in the pluri-ethnic population of New Caledo- in each sex) and in Europeans (P 0.01 in men, nia, as previously reported in Melanesians23 and P < 0.0001 in women). In addition, the two-way Polynesians.25 In Europeans, prevalence rates of obe- analysis of variance has shown a signi®cant interac- sity were about twice those described by Laurier et tion between ethnicity and environment in men, the al26 in a sample of 1941 French subjects aged 16±50 urban-rural increase for WHR being two-fold in y. In New Caledonia, Europeans present early with Melanesians compared with Europeans (P < 0.02). It overweight, so that the classical increase with age was is also noteworthy that mean WHR values were nearly not observed.26,27 identical in Melanesian men and women, whereas In addition, central adiposity was present in the they were quite different in Europeans. three ethnic groups, as early of thirty years, and BMI and WHR in New Caledonia J-M TassieÂ et al 65 increased steadily with age. Similar relationships, Neel's hypothesis about fatness, suggesting a particu- whatever the sex, have been found in other Caucasoid lar genotype in this population, `the thrifty genotype', populations6,7,14,28 as well as in Australian Aborigines, modulated by environmental conditions.34 Our results a population resembling the Melanesians.29 The com- also suggest an effect on body fat distribution. parison of mean values of WHR with the population of mainland France28 shows that Europeans from the urban area of New Caledonia had higher WHR values (0.96 vs 0.92 in men, 0.90 vs 0.83 in women). Conclusions Important inter-ethnic differences in the fat distribution persisted after adjustment for obesity. Polyne- sians of both sexes have the highest rate of obesity, The fact that ethnic differences are increased in the including a majority with android type obesity. In urban area, in comparison with the traditional rural Melanesians, obesity was less prevalent, in particular environment, is in favour a deleterious impact regard- in men, who were similar to Europeans in this respect. ing the exposure to obesity, in particular central In both sexes, they had systematically higher WHR obesity, which may lead to diabetes and other chronic values in comparison with Europeans. disease. Our ®ndings are at variance with the results of 24 Doswe et al, who observed in the Melanesians of Acknowledgements Papua New Guinea relatively low values of WHR, This work was supported by Grant 492015 from with sex differences, mean values of 0.89 in men and INSERM, Grant 607006 from the Ministry of Coop- 0.83 in women over 25 y. A difference between sexes eration, by the Medical Commission and the Pro- was also observed for WHR in an ethnic group having vinces of New Caledonia, the Rotary Club and the common characteristics with Melanesians, an urba- Lions Club of Noumea. We thank the medical staff of nized group of Aborigines of south-eastern Australia. the 19 health centers that participated in the survey In the age group 25±64 y, mean WHR was 0.94 in and the population of New Caledonia. mean and 0.87 in women.30 The absolute differences of mean values may be due to differences in the method of measurement between studies. However, this cannot explain the inter-sex differences which References show that fat distribution remained a sexual charac- 1 Negri E, Pagano R, Decarli A, La Vecchia C. Body weight teristic in these populations, while this was not and prevalence of chronic diseases. J Epidemiol Community observed in Melanesians of New Caledonia. No Health 1988; 42: 24±29. 2 Gar®nkel L. Overweight and mortality. Cancer 1986; 58: clear explanation for this phenomenon can be pro- 1826±1829. posed from the present data. We know that body fat 3 Larsson B. Obesity, fat distribution and cardiovascular dis- distribution is controlled by a complex neuroendo- ease. Int J Obes 1991; 15: 53±57. crine regulation, involving genetic and nongenetic 4 Colditz GA, Willett WC, Rotnitzky A, Manson JE. Weight factors.7 Our hypotheses are directed at the role of gain as a risk factor for clinical diabetes mellitus in women. Ann Intern Med 1995; 122: 481±486. hormonal status, which may differ between groups of 5 Chan JM et al. Obesity, fat distribution and weight gain as different genetic origins, as it has been suggested for risk factors for clinical diabetes in men. Diabetes Care 1994; glucose tolerance.31 Other nongenetic factors linked to 17: 961±969. environmental and social conditions should be taken 6 Rimm EB et al. Body size and fat distribution as predictors of into account (for example, physical exercise, diet, coronary heart disease among middle-aged and older US men. Am J Epidemiol 1995; 141: 1117±1127. availability of food), as previously studied in other 7 Bouchard C, Despres JP, Mauriege P. Genetic and nonge- 32 Paci®c Islands. Further studies involving genetic netic determinants of regional fat distribution. Endocr Rev and hormonal factors in Melanesians should contri- 1993; 14: 72±93. bute to an understanding of the particular pattern of 8 Vague J. The degree of masculine differentiation of obesities, body fat distribution in this population. a factor determining predisposition to diabetes, atherosclero- sis, gout and uric calculous disease. Am J Clin Nutr 1956; 4: In favour of the environmental hypothesis, the 20±34. degree of abdominal obesity appeared signi®cantly 9 Krotiewski M, BjoÈrntorp P, SjoÈstroÈm L, Smith U. Impact of higher in Noumea than in the rural area. This impact obesity on metabolism in men and women. J Clin Invest of urbanization on fatness has already been described 1983; 72: 1150±1162. in Melanesian women32,33 and in Polynesians of both 10 Kissebah AH et al. Relation of body fat distribution to 32 metabolic complications of obesity. J Clin Endocrinol sexes. These inter-ethnic and urban-rural differences Metab 1982; 54: 254±260. in morphotype parallel those recently reported for 11 Lapidus L et al. Distribution of adipose tissue and risk of diabetes mellitus prevalence rates 18 Diabetes preva- cardiovascular disease and death: a 12 year follow up of lence was higher in Polynesians of both sexes and, in participants in the population study of women in Gothenburg, Melanesians, the prevalence was higher in Noumea Sweden. BMJ 1984; 289: 1257±1261. 12 Ohlson LO et al. The in¯uence of body fat distribution on the than in rural area. The urban-rural difference in incidence of diabetes mellitus ± 13.5 years of follow-up of diabetes prevalence persisted after adjustment on the participants in the study of men born in 1913. Diabetes body mass index.18 These results are consistent with 1985; 34: 1055±1058. BMI and WHR in New Caledonia J-M TassieÂ et al 66 13 DucimetieÁre P, Richard J, Cambien F. The pattern of sub- sians in rural New Caledonia and Ouvea (Loyalty Islands). cutaneous fat distribution in middle-aged men and the risk of Diabetologia 1982; 23: 393±398. coronary heart disease: The Paris Prospective Study. Int J 24 Dowse GK et al. Extraordinary prevalence of non-insulin Obesity 1986; 10: 229±240. dependent diabetes mellitus and bimodal plasma glucose 14 Gillum RF. The association of body fat distribution with distribution in the Wanigela people of Papua New Guinea. hypertension, hypertensive heart disease, coronary heart Med J Aust 1994; 160: 767±774. disease, diabetes and cardiovascular risk factors in men and 25 Taylor R et al. Diabetes in Wallis Polynesians: a comparison women aged 18±79 years. J Chron Dis 1987; 40: 421±428. of residents of Wallis Island and ®rst generation migrants to 15 Svendsen OL et al. Measurement of body fat in elderly Noumea, New Caledonia. Diabetes Res Clin Pract 1985; 1: subjects by dual-energy x-ray absorptiometry, bioelectrical 169±178. impedance and anthropometry. Am J Clin Nutr 1991; 53: 26 Laurier D et al. Prevalence of obesity: a comparative survey 117±1123. in France, the United Kingdom and the United Studies. Int J 16 Durnin JVGA, Womersley J. Body fat assessed from total Obes 1992; 16: 565±572. body density and its estimation from skinfold thickness: 27 Rolland-Cachera MF et al. Body mass index variations: measurements on 481 men and women aged from 16 to 72 centiles from birth to 87 years. Eur J Clin Nutr 1991; 45: years. Br J Nutr 1974; 32: 77±97. 13±21. 17 Larsson B et al. Abdominal adipose tissue distribution, 28 Tichet J et al. Android fat distribution by age and sex. The obesity, and risk of cardiovascular disease and death: 13- waist hip ratio. Diabete Metab 1993; 19: 273±276. year follow up of participants in the study of men born in 29 O'Dea K et al. Obesity, diabetes, and hyperlipidemia in a 1913. BMJ 1984; 288: 1401±1404. Central Australian Aboriginal community. Diabetes Care 18 Papoz L, Barny S, Simon D and the CALDIA Study Group. 1993; 16: 1004±1010. Prevalence of diabetes mellitus in New Caledonia: ethnic and 30 Guest CS, O'Dea K, Hoper JL, Larkins RG. Hyperinsulinae- urban-rural differences. Am J Epidemiol 1996; 143: 1018± mia and obesity in Aborigines of south-eastern Australia. 1024. Diabetes Res Clin Pract 1993; 20: 155±164. 19 Grimley EJ, Prior IAM. Indices of obesity derived from 31 King H et al. Glucose tolerance and ancestral genetic height and weight in two Polynesian populations. Brit J admixture in six semitraditional Paci®c populations. Genet Prev Soc Med 1969; 23: 56±59. Epidemiol 1984; 1: 315±328. 20 Keys A et al. Indices of relative weight and obesity. J Chron 32 Taylor R et al. Dietary intake, exercise, obesity and non- Dis 1972; 25: 329±343. communicable disease is rural and urban populations of three 21 Garrow JS, Webster J. Quetelet's index (Wt/H2) as a measure Paci®c island countries. J Am Coll Nutr 1992; 11: 283±293. of fatness. Int J Obes 1985; 9: 147±153. 33 Zimmet P et al. Prevalence of diabetes and impaired glucose 22 Bennet PH. Recommendations on the standardization of tolerance in the biracial (Melanesian and Indian) population methods and reporting of tests for diabetes and its micro- of Fiji: a rural-urban comparison. Am J Epidemiol 1993; 118: vascular complications in epidemiologic studies. Diabetes 673±688. Care 1979; 2: 98±104. 34 Neel JV. Diabetes mellitus: a thrifty genotype rendered 23 Zimmet P et al. The prevalence of diabetes mellitus and detrimental by progress?. Am J Hum Genet 1962; 14: 353± impaired glucose tolerance in Melanesians and Part-Polyne- 362.